Poster Abstracts

2 downloads 0 Views 5MB Size Report
K.J.C. van Bommel1, A. Friggeri1, M. Stuart2, B.L. Feringa2, J. van Esch2 ..... TPGS. Journal of Controlled Release 86 (1) (2003) 33–48. [5] H. Jeffery, S.S. Davis, ...... 750. 200. WSE (%). 2.3. 1.4. 2.3. Tm (8C). 62.4. 62.4. 54.9. Crystallinity (%).

Journal of Controlled Release 101 (2005) 287 – 408

Poster Abstracts


K.J.C. van Bommel1, A. Friggeri1, M. Stuart2, B.L. Feringa2, J. van Esch2 1 BiOMaDe Technology Foundation, Nijenborgh 4, 9747 AG Groningen, The Netherlands 2 Department of Organic and Molecular Inorganic Chemistry, Stratingh Institute, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands


An enzymatically cleavable low molecular weight gelator (LMWG)–drug conjugate is described that is capable of gelating water at concentrations as low as 0.45 mM (=0.03 wt.%). By comparing the enzymatic cleavage kinetics of the LMWG–drug conjugate with those of a nongelating substrate, it was shown that although the enzyme (a-chymotrypsin) is still functional in the gel, molecules present within the gel fibers are protected from enzymatic cleavage.


Hydrogels of low molecular weight gelators (LMWGs) [1] are an attractive complement or even alternative for pharmaceutically interesting polymeric gel systems [2] as they possess properties unattainable by polymeric gelators [3]. The most important properties being a very rapid response to external stimuli, an inherent thermoreversibility owing to the noncovalent nature of the aggregation process, and a low molecular weight of the gelator, facilitating a fast clearance from the body after triggering the gel-to-sol transition. As we want to use LMWG systems in pharmaceutical applications, we require responsive and biocompatible systems, especially when envisioning their use for the triggered release of a pharmaceutical. Enzymatically cleavable systems are especially attractive triggerable systems, as they allow release of a pharmaceutical in very specific locations (e.g., tumors or designated areas of the GI tract). Enzymatically cleavable LMWG–drug conjugates (Fig. 1) could result in gels of which the molecules are protected against enzymatic cleavage as a result of their unavailability to the enzyme because they are incorporated in the fibrous aggregates. Upon triggering the gel-to-sol transition (e.g., via a pH or temperature change: D) the fibrous aggregates fall apart and the individual molecules can be cleaved by the enzyme. Here, we present the first enzymatically cleavable LMWG–drug conjugate system and prove that incorporation of a substrate into the gel fibers does indeed protect it from enzymatic cleavage.



Poster Abstracts

Fig. 1. Schematical representation of the enzymatic cleavage of a gelator–drug conjugate and the influence of aggregation on enzymatic cleavage.

Results and discussion We designed and synthesized compound 1, containing like many of our other LMWGs a cyclohexane core to which different moieties are linked via amide bonds [4]. The l-phenylalanylamidoquinoline [5] moiety was introduced to allow enzymatic cleavage to take place, using a-chymotrypsin capable of cleaving amide bonds at the CjO terminus of aromatic amino acids. The thus released fluorogenic Qmodel drugQ 6-aminoquinoline (6-AQ) allows easy monitoring of the cleavage kinetics [5]. Although gels of LMWG 1 can be easily switched into solutions and back into gels, either by temperature or pH changes, neither of these methods is very suitable for the inclusion of a-chymotrypsin in gels of 1, as a-chymotrypsin does not respond well to large changes in either temperature or pH. A solution was found by using a mixed solvent system. Gelator 1 was dissolved in a small amount of DMSO, whereas a-chymotrypsin was dissolved in a buffer solution (Tris–HCl, 0.1 M, pH 7.75). The rapid addition of the aqueous a-chymotrypsin solution to the DMSO solution of 1 resulted in the instantaneous formation of a clear, homogeneous gel that could be used for fluorescence experiments (cryoTEM measurements showed no differences in the gel fiber structure in this mixed solvent system when compared to the pure aqueous system). Compound 2 was synthesized in order to have a nongelating substrate, allowing us to evaluate the effect of gelation on enzymatic cleavage.

Chart 1. Molecular structures of LMWG 1 and nongelator 2.

In order to investigate the kinetics of enzymatically catalyzed cleavage of LMWG 1 as well as model substrate 2, experiments were carried out using different concentrations of substrates (i.e., 1 or 2). The initial velocity of hydrolysis of each experiment was then plotted as a function of the substrate concentration (S), resulting in the points plotted in graphic 1. From the points belonging to each substrate the V max (limiting or maximum enzyme velocity) and K m (Michaelis constant) can be determined, using either an Eadie–Hofstee or Lineweaver–Burke plot, both plots giving identical results for V max and K m with R 2 values of N0.995. Using the values thus obtained for substrate 2 (V max=22.3 Amol/min, K m=4.9 mM), it was possible to plot the theoretical curve, which accurately followed the experimentally determined points. For LMWG 1, however, the experimentally determined points (circles in graphic 1) clearly did not follow the trend common for enzyme substrates. Although the values for V max did increase with increasing substrate concentrations, they abruptly leveled

Poster Abstracts


off at a concentration of ca. 1.5 mM (see also inset of Graphic 1). Using only the experimentally obtained points at concentrations below 1.5 mM, values were calculated for V max and K m (4.1 Amol/min and 1.8 mM, respectively), which again allowed a theoretical curve to be drawn. It is clear that the experimentally determined points for 1 deviate from the theoretical curve at concentrations above 1.5 mM. As a result of its gelating properties LMWG 1 has an imposed V max of 1.8 Amol/min, rather than the expected 4.1 Amol/min. Interestingly, the concentration of 1.5 mM corresponds to the critical gelation concentration (CGC) of 1 in this particular solvent mixture. In gels of LMWGs, the concentration of gelator in solution no longer increases once the total concentration of gelator is above the CGC (i.e., all gelator in excess of the CGC will be in the gel fibers). As the value for V max no longer increases once the concentration of LMWG 1 has increased above the CGC, this suggests that only the molecules in solution and not those in the gel fibers can be cleaved by the enzyme. By performing an enzymatic cleavage experiment in which both LMWG 1 as well as nongelating substrate 2 were present, it was shown that the incorporation into a gel did not influence the functioning of the enzyme a-chymotrypsin.

Graphic 1. Initial velocity of hydrolysis of model substrate 2 (squares) and LMWG 1 (circles). Conditions: 25 8C, a-chymotrypsin: 40 AM, buffer solution: Tris–HCl, 0.1 M, pH 7.75. Solid lines: theoretical curves based on the calculated V max and K m values. Dashed line: experimental curve. Inset: enlarged view of data points and theoretical curve for LMWG 1.

Conclusions In conclusion, we designed and synthesized the first example of an enzymatically cleavable LMWG–(model) drug conjugate capable of gelating water at very low concentrations. Furthermore, we proved that although the LMWG–drug conjugate can be cleaved by achymotrypsin, the molecules that are incorporated into the gel fibers are protected from enzymatic cleavage. We plan to use similar systems for the controlled release of pharmaceuticals.

References [1] a) P. Terech, R.G. Weiss, Chem. Rev. 97 (1997) 3133–3159; b) D.J. Abdallah, R.G. Weiss, Adv. Mater. 12 (2000) 1237–1247; c) J.H. van Esch, B.L. Feringa, Angew. Chem. 112 (2000) 2351–2354; Angew. Chem. Int. Ed. 39 2000 2263–2266; d) O. Gronwald, S. Shinkai, Chem. Eur. J. 7 (2001) 4328–4334; e) G. Mieden–Gundert, L. Klein, M. Fischer, F. Vfgtle, K. Heuze´, J.L. Pozzo, M. Vallier, F. Fages, Angew. Chem. 113 (2001) 3266–3267; Angew. Chem., Int. Ed. 40 2001 3164–3166. [2] a) S. Deo, E. Moschou, S. Peteu, P. Eisenhardt, L. Bachas, M. Madou, S. Daunert, Anal. Chem. 75 (2003) 207A–213A; b) K.T. Nguyen, J.L. West, Biomaterials 23 (2002) 4307–4314; c) P. Gupta, K. Vermani, S. Garg, Drug Discov. Today 7 (2002) 569–579; d) N.A. Peppas, P. Bures, W. Leobandung, H. Ichikawa, Eur. J. Pharm. Biopharm. 50 (2000) 27–46.


Poster Abstracts

[3] J.C. Tiller, Angew. Chem. 115 (2003) 3180–3183; Angew. Chem., Int. Ed. 42 2003 3072–3075. [4] a) A. Heeres, C. van der Pol, M. Stuart, A. Friggeri, B.L. Feringa, J. van Esch, J. Am. Chem. Soc. 125 (2003) 14252–14253. b) K.J.C. van Bommel, C. van der Pol, I. Muizebelt, A. Friggeri, A. Heeres, A. Meetsma, B.L. Feringa, J. van Esch, Angew. Chem., Int. Ed. accepted. [5] P.J. Brynes, P. Bevilacqua, A. Green, Anal. Biochem. 116 (1981) 408–413.

STUDY OF THE SURFACE MOBILITY OF PLGA MICROSPHERES USING HIGH-RESOLUTION TOPOGRAPHY MEASUREMENTS WITH THE ATOMIC FORCE MICROSCOPE C. Bouissou1, P. Begat1, C. van der Walle2, R. Price1 1 Department of Pharmacy and Pharmacology, University of Bath, Bath, BA2 7AY, United Kingdom 2 Department of Physiology and Pharmacology, University of Strathclyde, Glasgow, G4 0NR, United Kingdom

Summary An atomic force microscope (AFM) has been used to obtain high-resolution topography measurements of surface properties of poly(lactideco-glycolide) (PLGA) microspheres. The effect of the tapping AFM mode on a low Tg polymer (Tg=37.5F0.2) was investigated and was shown to have no discernible effect on the surface stability. Real-time AFM measurements provides a novel means of observing moisture induced transformations to the porous structures of microspheres under controlled environmental conditions.

Introduction Biodegradable polymeric microspheres have received much attention for the sustained release of bioactive macromolecules, such as peptides or proteins. The various microencapsulation techniques lead to a wide variety of microcapsule behaviours [1,2]. The understanding of the different features is necessary in order to control and to manufacture specific pattern releases. The scanning electron microscope (SEM) is commonly used to observe the external morphology of the microspheres. However, conventional SEM requires coating of the sample with a conductive layer and operates in vacuo, thus limiting the mesoscopic resolution and study of the dynamic stabilities of microspheres particles. In contrast, the AFM offers the possibility of assessing accurately the conformational aspect and stability of any microspheres without the need for any significant sample preparation. Surprisingly, only very limited amount of work has been undertaken to study the surface conformational arrangement [3,4]. The aim of this study was to use high-resolution topographical AFM to investigate the surface properties of PLGA microspheres and to measure in situ moisture-induced transformations of porous microspheres.

Experimental methods Preparation of microspheres The encapsulation procedure used in this study was the water-in-oil-in-water double emulsion evaporation technique [5]. The inner aqueous phase contained 90 Al of phosphate-buffered saline (PBS, pH 7.5) with 5 Al of poly-vinyl alcohol (PVA, 5% w/v) in water, which was injected into 950 Al of oil phase consisting of dichloromethane (DCM) with 5% w/v PLGA. This primary emulsion was homogenised at 22 000 rpm for 15 s (with an Ultra TURRAX IKAT 18 basic), and transferred into 40 ml of water containing poly-vinyl alcohol (PVA, 0.5% w/v), then stirred at 500 rpm for 2 h at room temperature (lab-egg IKA RW11 basic). The solvent was allowed to evaporate for the capsides to harden. The microspheres were then harvested by centrifugation at 41 000 rpm for 1 min, washed, snap-frozen in liquid nitrogen and finally lyophilised overnight. Scanning Electron Microscope (SEM) Lyophilised microspheres were sprinkled onto a carbon adhesive disc, which mounted on an aluminium stub. Samples were coated with a thin layer of gold (Edwards Sputter Coater S150B). The SEM was operated at 10 kV (JEOL JSM6310). Topography measurement with the Atomic Force Microscope (AFM) All AFM surface topography images were recorded in TappingModek operation (TM-AFM). The effect of RH on surface topography was investigated using a custom-built perfusion unit. The partial vapour pressure of water within the imaging chamber of AFM was controlled by varying a mixture of dry nitrogen gas with the same gas humidified to 100% under constant temperature 25 8C (F0.2 8C).

Poster Abstracts


Results and discussion The polymeric capsules observed here were made with low molecular weight PLGA. Its characteristic glass transition temperature (Tg) was measured as 34.3 8CF0.8 8C. A representative SEM image of the prepared PLGA microspheres is shown in Fig. 1a. A corresponding highresolution topographical AFM image of a PLGA microsphere is shown in Fig. 1b. Under controlled environmental conditions (temperature=25F0.2 8C; relative humidity=28F3%), the surface properties of the porous microspheres remained stable during continuous imaging. Thus, the high-resolution AFM imaging was able to explore the surface structure and the stability of the porous structures, which may play a role in the surface erosion and drug release behaviour.

Fig. 1. SEM picture (a) and high-resolution AFM image (b) of the external view of a PLGA microsphere.

Furthermore, possible modifications to the surface properties of low Tg microspheres may be investigated under environmentally controlled AFM. Fig. 2a and b show topographical AFM images of a PLGA microsphere upon increasing the relative humidity from 28% RH (F3%) to 70% RH (F3%) at constant temperature of 25 8C (F0.2 8C). The real-time measurements enabled direct observation of dynamic morphological changes to a specific area of a microsphere. Upon increasing the relative humidity, the dimensions of the pores notably diminished with a decrease in surface roughness.

Fig. 2. AFM Topograpical images of a micropore and nanopores under normal condition 28% RHF3% (a) and under 70% RHF3% (b).

The modifications may be due to the sorption of condensed water, which has been proven to be dependent on the volume fraction of water present in the air surrounding [6]. This phenomenon is known to contribute to the plasticization of polymers and amorphous drugs [7,8]. The surface changes reported by the AFM may also be linked to the lowering of the Tg with increasing humidity [9]. This information may provide a novel means of preconditioning microspheres for altering drug diffusion processes, which has been correlated to polymeric chain mobility [10,11].


Poster Abstracts

Conclusion Atomic force microscopy investigations provided a valuable insight into the surface properties of porous microspheres and a novel means of determining the surface stability and the modifications to pores under controlled environmental conditions. The data obtained with the AFM will potentially lead to a step further into the work done with the SEM. References [1] T.H. Lee, J. Wang, C.-H. Wang, Double-walled microspheres for the sustained release of a highly water soluble drug: characterization and irradiation studies. Journal of Controlled Release 83 (3) (2002) 437–452. [2] P. Johansen et al., Revisiting PLA/PLGA microspheres: an analysis of their potential in parenteral vaccination. European Journal of Pharmaceutics and Biopharmaceutics 50 (1) (2000) 129–146. [3] L. Montanari et al., Poly(lactide-co-glycolide) microspheres containing bupivacaine: comparison between gamma and beta irradiation effects. Journal of Controlled Release 90 (3) (2003) 281–290. [4] L. Mu, S.S. Feng, A novel controlled release formulation for the anticancer drug paclitaxel (Taxol(R)): PLGA nanoparticles containing vitamin E TPGS. Journal of Controlled Release 86 (1) (2003) 33–48. [5] H. Jeffery, S.S. Davis, D.T. O’Hagan, The preparation and characterization of poly(lactide-co-glycolide) microparticles: II. The entrapment of a model protein using a (water-in-oil)-in-water emulsion solvent evaporation technique. Pharmacological Research 10 (3) (1993) 362–368. [6] J.S. Sharp, J.A. Forrest, R.A.L. Jones, Swelling of Poly(dl-lactide) and polylactide-co-glycolide in humid environments. Macromolecules 34 (25) (2001) 8752–8760. [7] M. Deng, K.E. U, Effects of in vitro degradation on properties of poly(dl-lactide-co-glycolide) pertinent to its biological performance. Journal of Materials Science. Materials in Medicine 13 (11) (2002) 1091–1096(6). [8] P.G. Royall, D.Q.M. Craig, C. Doherty, Characterisation of moisture uptake effects on the glass transitional behaviour of an amorphous drug using modulated temperature DSC. International Journal of Pharmaceutics 192 (1) (1999) 39–46. [9] L. Mackin et al., Quantification of low levels (b10%) of amorphous content in micronised active batches using dynamic vapour sorption and isothermal microcalorimetry. International Journal of Pharmaceutics 231 (2) (2002) 227–236. [10] Y. Yamaguchi, et al., Insulin-loaded biodegradable PLGA microcapsules: initial burst release controlled by hydrophilic additives. Journal of Controlled Release 81 (3) (2002) 235–249. [11] W. Friess, M. Schlapp, Release mechanisms from gentamicin loaded poly(lactic-co-glycolic acid) (PLGA) microparticles. Journal of Pharmaceutical Sciences 91 (3) (2002) 845–855.

Poster Abstracts


PHARMACEUTICAL APPLICATION OF THERMORESPONSIVE MATERIALS A. Gelencse´r, G. Cso´ka, I. Ra´cz, S. Marton, I. Antal Semmelweis University, Department of Pharmaceutics, H-1092 Budapest, Ho´gyes E. str. 7, Hungary Summary Drug delivery is optimal when controlled by symptoms. Smart systems provided pulsatile drug liberation according to different signs. Our aim was to formulate thermoresponsive systems applying gelatine, cellulose and cholesteryl (COC) derivatives in films, gels, patches, implants and TTS products. Properties were tested by static and dynamic methods. Metolose with excipients is suitable for the preparation of valve based delivery devices. COC embedded into the TTS membrane shows pulsatile drug delivery, similar to gelatine. Introduction Thermoresponsive drug delivery is controlled usually by a phase transition induced by a temperature change. These phase transitions can be the following: sol–gel transition, swelling–shrinking condition, glass-transition, phase transitions of liquid crystals, solved-precipitated form. The temperature, where these transitions occur is called the phase transition temperature (T f) which classifies thermoresponsive drug delivery systems into two different groups: Systems with negative temperature dependency (T f =LCST, Lower critical solution temperature): Above the phase transition temperature, the systems become insoluble or shrink in water and the drug liberation breaks off (off status). Below the phase transition temperature, they become soluble or swell in water, which makes drug delivery possible (on status). Systems with positive temperature dependency (T f =UCST, Upper critical solution temperature). Above the phase transition temperature, the systems become soluble or swell in water. In this status, the drug liberation increases (on status). Below the phase transition temperature, they become insoluble or shrink in water. In this status, the drug liberation breaks off (off status). Different polymer systems, liquid crystals, and colloid systems show thermoresponsive properties applied in films, gels, plasters, microspheres, microcapsules, micro and nano particles, liposomes. The aim of our study was the formulation and examination of thermoresponsive drug delivery systems (films, gels, patches, implants) using cellulose derivatives (metolose), chlolesteryl oleyl carbonate and gelatine as thermoresponsive auxiliary materials. Gelatine forms hydrogel with positive temperature dependency, which can be applied in thermoresponsive transdermal therapeutic systems (TTS). We prepared and examined gelatine TTS containing a NSAID (e.g., ibuprofen, diclofenac sodium). Thermoresponsive drug release and permeation, the effect of auxiliary materials on these processes were studied, respectively. Results and discussion Metolose is a polymer with negative temperature dependency and it demonstrated thermal gelation properties (Fig. 1). The change of phase transition temperature of Metolose was studied by viscometry.

Fig. 1. Thermal gelation of Metolose gel.


Poster Abstracts

Heating Metolose containing systems, the viscosity decreased, and below 80 8C, the viscosity started to increase. This can be assigned as reversible thermal gelation process which may serve as basis for thermoresponsive behaviour. The thermal gelation temperature depends on the substitution type of Metolose and using different additives the phase transition temperature can be shifted into the physiological range. Although Metolose is a negative temperature-dependent material, it may be converted into positive dependency by using valve-based technology (Fig. 2).

Fig. 2. Thermoresponsive drug delivery from a valve-based device containing Metolose gel.

In other parts of our experiements, cholesteryl oleyl carbonate (COC) was applied and studied as thermoresponsive excipient. Cholesteric– smectic–isotropic phase transitions of COC were examined by thermoanalytical (DSC) and spectroscopical (IR) methods. These examinations proved that one of the T f values is near to the body temperature and this transition is reversible. COC-embedded polymer membrane (Metolose as cellulose and Eudragit as polyacrylic derivative) with vacuum-filtration was prepared and the amount of adsorbed COC in function of COC concentration was examined. The hydrofobicity of embedded membrane was studied by the determination of the contact angle. The permeation of drug (NSAID) through different type embedded membranes (matrix and membrane controlled TTS) and the effect of adsorbed COC on the permeation rate was tested at static and dynamic conditions. Results show (Fig. 3) extremely high release from Metolose gel at body temperature corresponding to high fever (40 8C).

Fig. 3. Thermoresponsive drug delivery from different type embedded TTS (static method).

Gelatin gel has a thermosensitive property with positive dependency due to gel–sol phase transition at body temperature. This thermosensitivity can be decreased or ceased by the insertion of polyethylene glycol (PEG) as additive material. This detrimental effect becomes stronger when the molecular weight of the additive material increases (Fig. 4).

Poster Abstracts


Fig. 4. Liberation of NSAID from gelatin type TTS with different PEG.

Conclusion Gels and films prepared from Metolose, COC, and gelatine demonstrated thermoresponsive behaviour according to viscometry, DSC and drug release studies. Although Metolose has a negative temperature dependency, it can be converted to positive in the formulated transdermal therapeutic systems using additives and applying valve-based technology.

Acknowledgements We would like to thank Syntapharm, for giving us the possibility to use their material Metolose in our experiments.

References [1] J. Kost, R. Langer, Responsive polymeric delivery systems. Advanced Drug Delivery Reviews 46 (2001) 125–148. [2] S. Marton, G. Cso´ka, I. Ra´cz, Development and Application of Intelligent Drug Delivery Systems in Dermatology and Cosmetics, IFSCC Magazine 6 (1) (2003) 29–36. [3] E. Mathiowitz, Encyclopedia of controlled drug delivery, Volume 1, John Wiley and Sons, New York/Chichester/Weinheim/Brisbane/Singapore/ Toronto, 1999. [4] H. Ichikawa, Y. Fukumori, New applications of acrylic polymers for thermoresponsive drug release, STP Pharma Sciences 7 (6) (1997) 529–545.


Poster Abstracts

MOLECULAR MICROENCAPSULATION FOR IN VIVO TRANSPORT OF WATER-INSOLUBLE DRUGS I. Hernandez-Valdepen˜a, C. Braud, J. Coudane, D. Domurado, M. Vert CRBA, CNRS-UMR 5473, UM1-Faculte´ de Pharmacie, 15 avenue Charles Flahault, BP 14491, 34093 Montpellier Cedex 5, France Summary In studies carried out in our laboratory, a copolymer of maleic anhydride and alkyl vinyl ether (PMVE/Man) was modified; dodecyl side chains were introduced in different proportions (PMVE/AcM C12). With the aim of promoting the targeting of macrophage mannose-specific lectin and the internalisation of the modified polymer, copolymer C12 was complemented by mannosyl substituents. Molecularly microencapsulated conjugates were formed between the modified polymers and Yellow OB, a water-insoluble lipophilic dye. Only the presence of hydrophobic microdomains can account for the high solubility of the yellow OB independent of the degree of ionization of the COOH-units. With 38% of side lipophilic chains and an n=12 alkyl group, an efficient hypercoiled system was obtained. Introduction This work aimed to develop a hydrophobic drug delivery system based on molecular microencapsulation [1,2]. Molecular microencapsulation is a phenomenon where the drug is temporarily entrapped in hydrophobic pockets present in polymeric compounds based on water-dispersed macromolecules. These macromolecules can become made smart by combination with targeting species aimed at interacting specifically with selected receptors. The structure of these macromolecules consists of a hydrophilic polymer chain with hydrophobic side chains. Under appropriate conditions, these macromolecules may exhibit typical behavior of colloidal aggregates [3]. Hydrophilic polymer chains can be basic [4] or acidic [5] polyelectrolytes. Such modified polymers form polysoaps and are resistant to pH-induced conformational change. However, at low dissociation degrees, they adopt a compact coil conformation depending on the hydrophobic–hydrophilic balance. At high degrees of ionization, these compact states can be stable or disrupted to form an extended chain [6]. Experimental In studies carried out in our laboratory, a copolymer of maleic anhydride and alkyl vinyl ether (PMVE/Man) was modified. We took advantage of the high reactivity of the anhydride group to introduce dodecyl side chains in different proportions (PMVE/AcM C12). On dissolution in water, PMVE/Man hydrolyzed to form the poly-diacid (PMVE/AcM). With the aim of promoting the targeting of macrophage mannose-specific lectin and the internalisation of the modified polymer, copolymer C12 was complemented by mannosyl substituents. The mannosyl group was incorporated through a degradable glycolic spacer arm (PMVE/ AcM C12-Mannosyl). Partial esterification was assessed by various methods: Fourier transform infrared spectroscopy (FTIR), potentiometric titration and proton nuclear magnetic res onance (1H NMR). In model studies, molecularly microencapsulated conjugates were formed between the modified polymers and Yellow OB, a water-insoluble lipophilic dye. Results and discussion The apparent water solubility of Yellow OB due to the physical entrapment was determined by colorimetry [5]. Solubility data are given in Fig. 1.

Fig. 1. Solubility (S) of Yellow OB (mole/polymers’ mole motif ) in salt solutions ofPMVE/AcM ( ), PMVE/AcM C12 (n) and PMVE/AcM C12 Mannosyl ( ) as a function of the degree of ionization of –COOH.

Poster Abstracts


Evidence for the presence of microdomains in the three forms of copolymers was obtained by measuring the solubility of Yellow OB in salt solutions (NaCl, 9 g/l). Yellow OB was selected because it dissolves specifically in organic microdomains formed by compactly coiled polycarboxylic acids [5]. Solubility of Yellow OB in PMVE/AcM C12 (38% alkyl group with respect to the two carboxyl groups presents on the alternating repeating moieties) and in PMVE/AcM C12 Mannosyl was comparable. Approximately 8% of the polymer units were coupled to mannose. Insolubility of Yellow OB was observed in PMVE/cam. This finding agrees with the open-coil conformational structure of this copolymer. Only the presence of hydrophobic microdomains can account for the high solubility of the yellow OB whatever the degree of ionization of the units –COOH. The size of the side chains is important, since the compact conformation became increasingly stabilized with increasing alkyl group size. Conclusion Hydrophobic domains were associated with the hypercoiled conformation of this kind of copolymers. At high pH, the polyacids behaved as hydrophilic random coils, since the microdomains had disappeared. Large hydrophobic side chains (nN10 alkyl groups) led to hypercoiled conformations that were stable over the entire pH range. With 38% of lipophilic side chains and n=12 alkyl groups, an efficient hypercoiled system was obtained. As expected, Yellow OB solubility was high over the entire range of ionization for both copolymers, PMVE/AcM C12 and PMVE/AcM C12 Mannosyl. The solubility effect was thus retained despite the presence of the hydrophilic mannosyl group. The next step will be combinations with a real water-insoluble lipophilic drug. Acknowledgements I.Herna´ndez-Valdepen˜a acknowledges a grant from CONACYT and SEP, Mexico. References [1] M. Vert, Ionizable polymers for temporary use in Medicine, IUPAC 28th Macromolecular Symposium, 1982. [2] M. Vert, Polyvalent polymeric drug carriers, CRC Critical Reviews—Therapeutic Drug Carrier Systems, Editeur: S.D. Bruck, CRC Press, Boca Raton, 1986, pp. 291–327. [3] U.P. Strauss, Intramacromolecular micelles, Micellization, solubilization and microemulsions, Editeur: K.L. Mittal, Plenum Press, New York, 1977. [4] J. Huguet, M. Vert, (Acid–base)-dependent globular structures of partially N-alkylated poly(tertiary amines), Microdomains in Polymer solutions, Editeur: P.L. Dubin, Plenum Press, 1985, pp. 51–66. [5] C. Villiers, C. Braud, Influence de l’ionisation sur le comportement conformationnel de copolyme`res alterne´s acide male´ique/alkylvinyle´ther dans l’eau. Effet de la structure du groupe alkyle, Nouveau Journal de Chimie 2 (1978) 33–38. [6] S.R. Tonge, B.J. Tighe, Responsive hydrophobically associating polymers: a review of structure and properties, Advanced Drug Delivery Reviews 53 (2001) 109–122.


Poster Abstracts

MECHANISTIC ANALYSIS OF OSMOTICALLY PROMOTED RELEASE FROM HYDROPHOBIC ELASTOMERIC MONOLITHS—EFFECT OF SOLUTE LOAD D. Soulas, K.G. Papadokostaki, M. Sanopoulou, J.H. Petropoulos Institute of Physical Chemistry, National Center for Scientific Research bDemokritosQ, 15310 Ag. Paraskevi Attikis, Athens, Greece

Summary An experimental investigation of osmotically enhanced release from rubbery hydrophobic matrices was performed, using silicone rubber films filled with various amounts of NaCl particles. The release kinetic data were supplemented with data on the concurrent variation of the water content of the polymer matrix, as well as on the NaCl permeability and mechanical elasticity properties of salt loaded and/or depleted matrices. The combined results provide significant mechanistic information in the light of current theoretical models.

Introduction The controlled release of drugs from polymer monoliths is facilitated markedly by the presence of water-soluble excipients. If a watersoluble excipient is added to a hydrophobic polymer, the observed large increase in drug release is attributable to excess swelling and accompanying structural changes of the polymer, induced by the osmotic action of the excipient. Various models have been proposed to account for the observed kinetic behavior [1–4]. In order to gain further insight into these phenomena, we report a study of the kinetics of release of NaCl from silicon rubber matrices, as a function of NaCl load, supplemented by parallel measurements of the concurrent variation of the water content of the matrix and a study of the permeability and mechanical properties of the salt-depleted matrices.

Experimental methods Mixtures (10:1) of components A and B of silicone rubber (SR) type GE-615 (GE-Bayer Chemicals) were cured, at 100 8C for 1 h, to obtain the crosslinked matrices. Powdered analytical grade NaCl of particle size of approximately 6F1 Am was used as the osmotically active solute. Particle-loaded matrices, in the form of films (ca. 33 cm2, thickness L=180–300 Am), were obtained by dispersing the salt particles in the fluid SR mixture prior to curing. Salt release from these loaded films (mounted on stirring rods rotating at 37 rpm in frequently renewed, known volumes of distilled water, thermostated at 25F0.1 8C), was measured. The concurrent variation of the water content of the film was also monitored by weighing the blotted films at suitable intervals. The NaCl partition (K N) and diffusion (D N) coefficients in the hydrated salt-depleted films resulting from the above release experiments, were measured by equilibrating the said films with 10% w/v NaCl solutions and then monitoring the desorption of NaCl from the blotted equilibrated films into water, using the technique described above. The resulting permeability ( P N=K ND N) values were compared with that of the neat film measured in separate permeation experiments. Comparative values of the tensile modulus of elasticity E of dried neat, salt-loaded and salt-depleted, films was measured, in an atmosphere of 50% relative humidity, by means of a Tensilon UTM-II-20 instrument (Toyo Baldwin), at a strain rate of 0.8 min1.

Results and discussion Plots of Q Nt/Q Nl vs. t 1/2/L (where Q Nt, Q Nl represent the amount of NaCl released at time t and t=l, respectively) are shown in Fig. 1. Discounting a short initial steep portion (most probably due to exposed particles at the film surfaces), the NaCl release curves B–D exhibit marked deviation from Fickian kinetics in the form of an S shape, which is most pronounced in curves B and C, implies acceleration of NaCl transport through the film attributable to osmotically induced excess influx of water (see below). The shape of curve A is more complicated, possibly as a result of the difficulty of ensuring uniform particle distribution in the polymer matrix at very low salt loads. When replotted on a t scale (see Fig. 2), the said release curves exhibit linear portions, i.e., periods of constant rate of release (cf. similar observations in Refs. [1,2,5]), which are more extensive in curves B and C, while the magnitude of the release rate increases markedly with the salt load.

Poster Abstracts

Fig. 1. Normalized NaCl release kinetic curves on a t 1/2/L scale for 4–23 wt.% salt loads.

Fig. 2. NaCl release kinetic curves on t scale for 4–23 wt.% salt loads.

Fig. 3. Variation of osmotically induced water uptake during NaCl release for 4–23 wt.% salt load.



Poster Abstracts

Fig. 4. Variation in terms of fractional volume of peak (v WM) and final (v WF) values of osmotically induced water uptake with salt load (v N). As shown in Fig. 3, the concurrent variation of the water content of the film q Wt takes the form of an initial steep rise to a maximum value (corresponding to a fractional volume of imbibed water v WM), followed by a decline to a final value (corresponding to a fractional volume of imbibed water v WF) in the salt-depleted film. Fig. 4 shows that v WF increases regularly with the salt load (expressed as the volume fraction of NaCl in the SR film, v N) and hence with the variation of release rate noted above; v WM follows a similar trend, which is, however, reversed at the highest load, so that v WM approaches v WF. It is also noteworthy that v WFNv N in all cases. These results indicate that (a) the osmotically induced excess water uptake produces a large volume dilation of the SR film; (b) the said dilation consists of reversible (elastic) and irreversible components, represented (neglecting the very low normal degree of hydration of the neat SR matrix v WPb0.01) by (v WMv WF) and (v WFv N), respectively and (c) the elastic component exhibits a (prima facie) unexpected sharp drop at the highest salt load. According to current theoretical approaches to osmotically enhanced release of dispersed solutes [1–5], when the activity of invading water, at any given location across a loaded film, exceeds that in a saturated aqueous solution of the solute, osmotically driven penetration of imbibed water from the matrix into particle-containing cavities is initiated, leading to formation of saturated aqueous solution and buildup of hydrostatic pressure therein. The resulting tendency of these cavities to dilate is resisted by the cavity walls; which are thus subjected to increasing tangential tensile stresses, as long as the said water influx continues. Unless the wall of any particular cavity is able to withstand a rise in hydrostatic pressure sufficient to neutralize the osmotic driving force, it will rupture (cavity wall rupture or CWR model) [1], thus providing easy access to a neighboring cavity, and through it to a network of pores (consisting of previously ruptured cavities) extending to the film surface. In the case of relatively thick cavity walls (i.e., dilute particle dispersions), one may (more plausibly) envisage initiation of a crack, which can then propagate until it meets another cavity [2,3]. An alternative mechanism (which does not exclude CWR but can occur in parallel with it whenever CWR is present) is based on the development of zones of enhanced hydration (ZEH) of the polymer matrix surrounding each cavity (due to the osmotic action of solute absorbed from the intracavity solution), which expand and ultimately merge with one another to form continuous bsuperhydratedQ polymer matrix regions [4]. If polymer swelling in the ZEH has gone beyond elastic limits, these regions will survive (at least partly) in the salt-depleted part of the film; thus providing easy pathways for solute diffusing out of the film. Hence, the end result of the CWR and ZEH mechanisms is quite similar. The main difference lies in the location of the osmotically imbibed water. According to the former model, it is located exclusively in dilated cavities (implying as high an irreversible cavity volume dilation as 14-fold in depleted films of the lowest initial salt load); whereas in the latter model, it is also partly accommodated in the (superhydrated) polymer matrix. Table 1 Salt load (v N=fractional volume), NaCl permeability ( P N) of neat (v N=0) and salt-depleted films, and tensile modulus of dried neat, salt-loaded (E 1) and salt-depleted (E 2) films NaCl content of dry film wt.%


0 4.0 7.7 13.5 23.1

0 0.019 0.039 0.064 0.126

P N1011 (cm2/s)

E 1106 (N/m2)

E 2106 (N/m2)

2.0F0.2 2.2F0.3 2.5F0.5 4.5F0.5 23.2F2.4

6.8F0.2 8.1F0.9 11.9F0.1 13.9F0.1 3.1F0.4

6.2F0.1 4.8F0.2 4.6F1.0 3.2F0.5

Poster Abstracts


The observed substantial amounts of irreversibly imbibed excess water are consistent with the mechanical fracture or yielding processes invoked in the above CWR and ZEH models. Since the number of particle-containing cavities increases in line with the salt load, a parallel corresponding rise of the release rate, v WF and v WM is expected. The reason why this trend is reversed at the highest salt load (v N=0.126) for v WM, but not for v WF, can also be understood in this light, by invoking our previous findings relating to NaCl-loaded cellulose acetate matrices [4]; which strongly suggest that, at particle packing densities significantly in excess of v N=0.10, complete coverage of individual particles by the viscous fluid dispersion medium is not achieved, thus giving rise to an incipient pore network. This conclusion is borne out here too by the behavior of the NaCl permeability P N of salt-depleted films (Table 1). In particular, P N remains close to that for the neat film (consistent with the presence of largely isolated water-filled cavities or cavities+ZEH) at v Nb0.10 and then jumps to a much higher value (consistent with the appearance of a pore network) at v NN0.10. Further confirmatory evidence is provided by the continuous rise of the tensile modulus of salt-loaded films (E1; Table 1) with salt load, at v Nb0.10; followed by a sudden drop, at v NN0.10, to a value close to that for NaCl depleted films (indicating poor adhesion between solid particles and polymer matrix in the latter case, consistent with incomplete encapsulation of the said particles in the SR matrix).

Conclusions The analysis of solute release and concurrent water uptake data presented above has revealed some interesting kinetic features. These results, complemented with structural information about the solute-depleted matrix derived from a study of its mechanical and permeability properties, provide significant new insight into the mechanism of osmotically enhanced solute release in monolithic controlled release systems and related theoretical models.

References [1] [2] [3] [4] [5]

B.G. Amsden, Y.-L.Cheng, M.F.A. Goosen, J. Control. Release 30 (1994) 45. R. Schirrer, P. Thepin, G. Torres, J. Mater. Sci. 27 (1992) 3424. B. Amsden, J. Control. Release 93 (2003) 249. K.G. Papadokostaki, S.G. Amarantos, J.H. Petropoulos, J. Appl. Polym. Sci., 67 277; 69 (1998) 1275. D. Brown, Y.H. Bae, S.W. Kim, Macromolecules 27 (1994) 4952.


Poster Abstracts

CONTROLLED DRUG DELIVERY THROUGH TAILOR-MADE BLEND POLYMERIC MEMBRANES D.F. Stamatialis, H.H.M. Rolevink, J. Balster, G.H. Koops Institute for Biomedical Technology (BMTI), Membrane Technology Group—EMI Twente, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands Summary In this work, we prepare tailor-made membranes by blending sulfonated poly(ether ether ketone) (S-PEEK) and poly(ether sulfone) (PES) polymers, at various ratios. Timolol (TM) is used as a model drug for the investigation of the controlled delivery through these membranes and their application to a transdermal TM patch is discussed. Introduction For drugs with short half-lives, transdermal delivery provides a continuous noninvasive mode of administration. For charged drug molecules, the transdermal delivery can be assisted by the application of direct constant current, which drives the drug through the skin by electrostatic repulsion (method called iontophoresis) [1]. An important component of a drug delivery patch is the artificial membrane, which is in contact with the skin and provides controlled drug delivery [2]. In the present work, we study the passive (no current application) and iontophoretic drug transport through tailor-made membranes prepared by blending sulfonated poly(ether ether ketone) (S-PEEK) and poly(ether sulfone) (PES) polymers, at various ratios. The S-PEEK contains negatively charged sulfo-groups, which can be used for the transport of positively charged drugs. When the S-PEEK is blended with the nonconductive PES, its swelling is reduced and the drug delivery can be regulated [3]. Timolol (TM), a nonselective beta-adrenergic blocking agent is used as a model drug molecule. Its pKa is 9.21 and it is therefore positively charged at physiological pH (7.4). Experimental methods The poly(ether ether ketone) (PEEK 450PF, Victrex) was sulfonated following the procedure described elsewhere [3]. The polymer blends were prepared by mixing S-PEEK with PES at various weight ratios. Three polymer solutions of 20 wt.% polymer in N-Methyl2-pyrrolidinon (NMP) were prepared: (i) 100% S-PEEK (S-PEEK100), (ii) 80% S-PEEK/20% PES (S-PEEK80/PES20) and (iii) 60% SPEEK/40% PES (S-PEEK60/PES40). The membranes were prepared by solution casting and evaporation [3]. The membranes after preparation were kept in a phosphate buffer saline (PBS) 0.153 M solution at pH 7.4 and their thickness was measured in the swollen state (Table 1). Table 1 Membrane characteristics Membranes

Thickness (Am)

Swelling (%)

Electrical resistance (kV cm2)


50 46 35

89F20 41F12 26F8

1.1 40.5 66.2

The membrane swelling was calculated using the equation: [(W swolW dry)/W dry]100, where W swol, W dry is the weight of the membrane in swollen and dry state, respectively. Timolol maleate salt (MW=432.5, Sigma—The Netherlands) was dissolved in PBS. The concentration of TM was in the range of 10–15 mg cm3. The diffusion cell and the experimental procedure were described in detail elsewhere [2]. In the iontophoretic experiments, we applied a current density up to 0.5 mA cm2. The electrical resistance of the membranes was measured during the iontophoretic experiments [2] (Table 1). All the experiments were performed at least in triplicate for each membrane. The concentration of TM in the donor and acceptor chamber was determined by HPLC [2]. The steady state flux of TM ( J ss, in mg cm2 h1) through the membrane is expressed as:

Jss ¼ KP CTM


where K P is the TM permeability coefficient and C TM is the concentration of TM in the donor chamber. Results and discussion Fig. 1a shows a typical result of the amount of permeated TM through an S-PEEK80/PES20 membrane vs. time. From the slope, we calculate the J ss and by using Eq. (1), the TM permeability coefficient, K P. Fig. 1b presents the TM permeability through the blend membranes at various current densities. The error bars represent the average of at least three membrane samples.

Poster Abstracts


Fig. 1. (a) Typical result of TM transport through the S-PEEK80/PES20 membrane. C TM=13.9 mg cm3. (b) The TM permeability through the S-PEEK100 (full circles), S-PEEK80/PES20 (open circles) and S-PEEK60/PES40 (full triangles) blend membranes. For passive diffusion, the blending of S-PEEK with PES gives the opportunity to regulate the TM delivery in a broad range (TM permeability in the range: 0.4–13106 cm/s). For the S-PEEK100 and S-PEEK80/PES20 membranes, the transport of TM is high due to the high membrane swelling (Table 1). In addition, the application of electrical current does not have a significant effect on the TM delivery. For these membranes, the contribution of passive diffusion greatly outweighs the contribution of electrical current thereby making the TM transport with and without applied current indistinguishable. For the S-PEEK60/PES40 membrane, however, the transport of TM increases significantly due to the current application (Figs. 1b and 2). Interestingly, the effect of the electrical current on the TM transport through this membrane is comparable to that for the TM transport through pig stratum corneum (SC) [4] and human skin [5] (Fig. 2). The latter is a very interesting finding because the S-PEEK60/PES40 membrane can be considered for two applications:

(i) (ii)

As a membrane in a TM iontophoretic transdermal patch. In this case, it will act as a safe guard for the TM transport and control the delivery if for any reason the skin is compromised. As a possible substitute for the human or pig skin for the in vitro tests with TM.

It is finally important to note that during the iontophoretic experiments, the electrical resistance of the blend membranes increases significantly. Especially for the S-PEEK80/PES20 and S-PEEK60/PES40 membranes, it reaches rather high values (Table 1). This phenomenon, which is not observed for the PBS alone (blank solution, when no TM is used), can be attributed to the ion-pairing effect between the positively charged and bulky TM molecules and the sulfo-groups of the S-PEEK resulting in the decrease of the membrane conductivity [6].

Fig. 2. Effect of current density upon the TM permeability through the S-PEEK60/PES40 membrane (triangles), pig SC (squares, [4]) and human skin (circles, [5]).


Poster Abstracts

Conclusions The controlled delivery of TM through various S-PEEK/PES blend membranes was investigated. The presence of PES in the blend reduces the swelling of S-PEEK resulting in the regulation of the TM transport. The application of electrical current increases the TM delivery only for the S-PEEK60/PES40 membrane. For this membrane, the TM transport at various current densities is similar to that for the pig SC and human skin reported in the literature.

Acknowledgements The European Community is acknowledged for the financial support of this work, which is part of the ACTRADEL project within the 5th RTD Framework program.

References [1] A.K. Banga, in: M.H. Rubinstein, C.G. Wilson, J.P. Todd (Eds.), bElectrically assisted transdermal and topical drug deliveryQ, Taylor and Francis, London, 1998. [2] D.F. Stamatialis, H.H.M. Rolevink, G.H. Koops, bControlled transport of timolol maleate through membranes under passive and iontophoretic conditionsQ, J. Control. Release 81 (2002) 335. [3] F.G. Wilhelm, I.G.M. Punt, N.F.A. v.d. Vegt, H. Strathmann, M. Wessling, bCation permeable membranes from blends of sulfonated poly(ether ether ketone) and poly(ethersulfone)Q, J. Membr. Sci. 199 (2002) 167. [4] D.F. Stamatialis, H.H.M. Rolevink, G.H. Koops, bDelivery of timolol through artificial membranes and pig stratum corneumQ, J. Pharm. Sci. 92 (2003) 1037. [5] N. Kanikkannan, J. Singh and P. Ramarao, bIn vitro transdermal iontophoretic transport of timolol maleate: effect of age and speciesQ, J. Control. Release 71 (2001) 99. [6] S. Mafe, P. Ramirez, A. Tanioka, J. Pellicer, bModel for Counterion-Membrane-Fixed Ion Pairing and Donnan Equilibrium in Charged MembranesQ, J. Phys. Chem., B 101 (1997) 1851.

Poster Abstracts


SWELLING AND PARACETAMOL RELEASE STUDIES OF POLY(ACRYLAMIDE-CO-ITACONIC ACID) HYDROGELS M. Stanojevic´1, M. Kalagasidis Krusˇic´2, M. Stupar1, J. Filipovic´2 1 Department of Pharmaceutical Technology and Cosmetology, Faculty of Pharmacy, 11 000 Belgrade, Serbia and Montenegro 2 Department of Organic Chemical Technology, Faculty of Technology and Metallurgy, 11 000 Belgrade, Serbia and Montenegro Summary Hydrogels play a significant role in advanced drug delivery formulations. Copolymer hydrogels of acrylamide (AAm) and itaconic acid (IA) crosslinked with N,N-methylene bisacrylamide (MBA) were synthesized by radical polymerization, initiated with the redox couple potassium persulfate/potassium pyrosulfate (PPS/PyPS). Their swelling behavior and in vitro release of paracetamol as a model drug, were studied as a function of pH value of the external media and crosslinking agent concentration. Introduction Hydrogels have found widespread applications in biomedical and pharmaceutical fields. Hydrogels are hydrophilic, three-dimensional, polymeric networks, capable of imbibing significant amounts of water or biological fluids. The networks are usually composed of homoor copolymers which are insoluble due to the presence of chemical or physical crosslinks. Besides exhibiting swelling-controlled drug release, some hydrogels also show stimuli-responsive changes in their structural network and, hence, in their drug release. Because of the large variations in physiological pH values, as well as the pH in pathological conditions, pH-responsive polymeric networks have been extensively studied. The swelling of hydrogels is influenced by the chemical structure, the crosslinking ratio and in the case of stimuli-sensitive hydrogels by pH, temperature, ionic strength, etc. [1,2]. The aim of this study was to investigate the swelling properties of hydrogels based on AAm and IA as a response to the change of the crosslinking agent concentration and pH values of the media, as well as the potential use of these hydrogels as drug delivery systems. Experimental methods Preparation of hydrogels: The copolymer hydrogels of AAm and IA were obtained by radical crosslinking copolymerization at 60 8C during 24 h (N2 atmosphere). Ten weight percent of the monomers were dissolved in water with the redox couple PPS/PyPS (1.0 wt.%) and MBA (2.0, 2.5 and 3.0 wt.%) with respect to the monomers. The reaction mixture was placed between two glass plates sealed with a rubber spacer. After completion of the reaction, the gels were cut into discs and immersed in water to remove unreacted monomers. The discs were dried at room temperature to xerogels (1 mm thick and 5 mm in diameter). Equilibrium swelling studies: The xerogel discs were immersed in buffer solutions (pH 2.2, 4.5 and 6.8) to obtain equilibrium swelling at 30 8C. The progress of the swelling process was monitored gravimetrically and the degree of swelling was calculated using the equation: q=W t /Wo, where Wo is the weight of xerogel at time 0 and W t is the weight of swollen hydrogel at time t. Loading of drug: Dry hydrogel discs were loaded with paracetamol (pK a 9.5) by immersion in an aqueous solution of the drug (10 mg/ml) at room temperature for 2 days. The drug loaded hydrogels were removed from the solution and left to dry to constant weight. Release experiments: The release studies were performed using the dissolution testing apparatus 1 of USP 23 (Erweka DT 70), at 50 rpm at 30 8C [3]. Three buffer solutions (pH 2.2, 4.5 and 6.8) were used as dissolution media. At predetermined times, 3-ml samples were removed and assayed for paracetamol at 243 nm using a Varian Cary UV-VIS spectrophotometer. The cumulative amount of the paracetamol released from the hydrogel was determined from the calibration curve. Every data point is the average value of three independent experiments.

Fig. 1. The swelling degree of PAAm/IA hydrogels as a function of pH crosslinking agent concentration.


Poster Abstracts

Results and discussion The swelling degree is strongly dependent on the pH of the buffer: at low pH values, the itaconic acid groups are not ionized so the swelling ratio is low. Intermolecular complexation due to hydrogen bonding occurs, acting as physical crosslinks. Above the nominal pK a values of IA (3.85; 5.44) [4], the swelling ratio increases with increasing pH due to the progressive ionization of the COOH groups. It can be seen from Fig. 1 that the swelling decreases as the concentration of crosslinking agent increases. Water transport in polymer networks can be analyzed according to the Peppas equation: M t /M l=kd t n [1]. The obtained results (nz0.5), indicate that the water transport mechanism is non-Fickian in the investigated pH range. As gel ionization occurs the osmotic swelling pressure increases as well as macromolecular relaxation caused by electrostatic repulsion of the COO groups. The release profiles obtained under different experimental conditions are presented in Fig. 2. Distinct differences between the obtained release profiles were observed indicating the influence of the crosslinking agent concentration as a hydrogel property and pH as an environmental variable on the drug release. It can be seen that for all the investigated buffer solutions, increasing the concentration of crosslinking agent leads to an increase of the rate of drug release. This effect is more pronounced at pH 6.8 than at the pH values, which could be attributed to the swelling controlled mechanism of drug release. By controlling their swelling properties in the presence of a biological fluid, hydrogels can be useful devices for releasing drugs in a controlled manner. Further studies should determine the effect of the PAAm/IA hydrogel composition in order to achieve desired drug release profiles and to obtain a better understanding of the influence of the synthesis variables on the echanism of drug release.

Fig. 2. Effect of crosslinking agent concentration on the release of paracetamol from PAAm/IA hydrogels in different buffer solutions: pH 2.2 (a), pH 4.5 (b) and pH 6.8 (c).

Poster Abstracts


Conclusion The release studies indicate that the crosslinking ratio is one of the main factors affecting the swelling of hydrogels and, hence, drug release. Changing the feed composition, i.e., the crosslinking agent concentration, may lead to a favourable drug release, which is important for a potential application of the investigated hydrogels as controlled release drug delivery systems. References [1] N.A. Peppas, P. Bures, W. Leobandung, H. Ichikawa, Hydrogels for pharmaceutical formulations. European Journal of Pharmaceutics and Biopharmaceutics 50 (2000) 27–46. [2] P. Gupta, K. Vermani, S. Garg, Hydrogels: from controlled release to pH-responsive drug delivery. Drug Discovery Today 7 (10) (2002) 569–579. [3] P.M. De la Torre, Y. Enobakhare, G. Torrado, S. Torrado, Release of amoxicillin from polyionic complexes of chitosan and poly(acrylic acid), study of polymer/polymer and polymer/drug interactions within the networks structure. Biomaterials 24 (8) (2003) 1499–1506. [4] R.C. Wheast, editor. Handbook of Chemistry and Physics, 53 edn., Cleveland, Ohio, 1975.

SWELLING BEHAVIOR AND MECHANICAL PROPERTIES OF POLY(ACRYLAMIDE-CO-ITACONIC ACID) HYDROGELS AND THEIR SEMI-IPNS WITH POLY(ETHYLENE GLYCOL) M. Stanojevic´1, M. Stupar1, M. Kalagasidis Krusˇic´2, J. Filipovic´2 1 Department of Pharmaceutical Technology and Cosmetology, Faculty of Pharmacy, 11 000 Belgrade, Serbia and Montenegro 2 Department of Organic Chemical Technology, Faculty of Technology and Metallurgy, 11 000 Belgrade, Serbia and Montenegro Summary pH-responsive poly(acrylamide-co-itaconic acid) copolymer, and semi-IPNs with 5, 10 and 15 wt.% of poly(ethylene glycol), were synthesized. Their swelling behavior was studied as a function of pH in the range of 1.76 to 7.81. The shear storage (GV) and loss (GV) moduli as a function of frequency, obtained for the gels as formed and at equilibrium swelling, were higher for the semi-IPNs than for the copolymer hydrogel. The GV and GV values appeared to depend also on the polymer content.

Introduction A special class of hydrogels, called dintelligentT or stimuli-responsive hydrogels, exhibit significant volume changes in response to small changes in pH, temperature, light, electric field, etc., which enables their use for drug delivery systems [1]. For a given application, hydrogels must posses the desired mechanical properties, which depend on the hydrogel composition, the amount of crosslinking agent and the synthesis conditions. The mechanical properties of hydrogels may be improved by preparing semi-InterPenetrating Networks (IPNs), where the hydrogel network is obtained in the presence of a previously synthersized polymer. This paper reports on the influence of the interpenetrant on the equilibrium swelling, the oscillatory swelling, the dynamic swelling and the mechanical properties of copolymer hydrogels based on acrylamide (AAm) and itaconic acid (IA) and their semi-IPNs with poly(ethylene glycol) (PEG).

Experimental methods Preparation of samples The copolymer hydrogel was prepared by radical crosslinking copolymerization at 60 8C, in a nitrogen atmosphere, for 24 h. Ten parts by weight of the monomers (AAm/IA=90/10 wt. ratio) were dissolved in 90 parts by weight of water. The concentration of the initiator/ accelerator redox couple (potassium persulphate/potassium pyrosulphate) was 1 wt.% and the crosslinking agent (MBA) concentration was 2 wt.%, with respect to the monomers. The reaction mixture was poured between two glass plates (2050.4 cm), sealed by a rubber spacer. Once the reaction was completed, the gels were cut into discs and immersed in water to eliminate unreacted monomers. There was practically complete conversion of monomer to polymer. The gels dried at room temeprature were in the shape of discs, 10 mm in diameter and 1 mm thick. Equilibrium swelling studies Xerogel discs were immersed in an excess of buffer solution (pH from 1.76 to 7.81), to achieve equilibrium swelling at 30 8C. The progress of the swelling process was monitored gravimetrically. The swelling ratio ( q) was calculated from the folowing equation: q=W t /Wo where Wo and W t are the weights of xerogel at time 0 and of swollen hydrogel at time t, respectively [2].


Poster Abstracts

Oscillatory swelling studies For this study, a xerogel was placed in a buffer of pH=7.81 and allowed to swell to equilibrium. Then the gel was withdrawn from pH=7.81 buffer and placed in a buffer of pH=1.76, and allowed to attain equilibrium. The swelling–deswelling measurements were repeated three times. Dynamic mechanical analysis Strain-frequency sweeps were performed on hydrogel discs, as synthesized and swollen to equilibrium, using a Rheometrics 605 mechanical spectrometer, with parallel plates geometry (25 mm in diameter). The complex shear moduli were measured as a function of frequency (x), from 0.1 to 100 rad/s, at 28 8C. Results and discussion The most characteristic property of hydrogels is their ability to swell in the presence of water and to shrink in the absence of it. The hydrophilicity of polymer chains and the crosslink ratio are the main factors controlling the extent of swelling. In our study, small quantities of itaconic acid were used to impart pH-sensitive swelling behaviour to the gels. In contrast with the nonionic PAAm hydrogel, where the change of q e with pH is small, the PAAm/IA copolymer and semi-IPNs show pH sensitive swelling behaviour due to the fact that the extent of ionization of IA is strongly dependent on pH. There is a difference in the equilibrium mass swelling ( q e) between the copolymer (PAAm/IA) gel and the semi-IPNs. The change of q e with pH for the semi-IPNs is not as pronounced as in the case of the copolymer. This could be explained by the high flexibility of the PEG macromolecular chains, which are incorporated between the network chains and which lowers the electrostatic repulsion between the carboxylate groups (Fig. 1). The water sorption process for a PAAm gel is Fickian, with diffusion exponent (n) values of about 0.5. The n values of the IPNs vary between 0.5 and 1.0 as the pH of the external medium increases. Furthermore, n varies as the concentration of PEG in the semi-IPNs increases for the IPNs with 5 and 10 wt.% of PEG (PAAm/IA/PEG-5 and PAAm/IA/PEG-10) The water sorption process is non-Fickian over the whole pH range, but in the case of semi-IPNs with 15 wt.% PEG (PAAm/IA/PEG-15) this process converts from Fickian to non-Fickian, i.e., from a diffusion controlled to a combined diffusion and macromolecular relaxation controlled process, at pH values higher than pK a of the acid groups. Swelling is much faster than deswelling, which may be explained by the different mechanisms of these two processes. The deswelling process starts when the completely swollen gel is placed in the acidic medium where the H+ ions are abundant at its surface. The COO groups at the surface react with H+ ions producing COOH groups, which form hydrogen bonds in the surface layer which shrinks because additional physical crosslinks are formed. This surface layer acts as a barrier slowing down the release of water from the bulk of the gel. When the equilibrium is reached, H-bonds are formed throughout the gel volume and it is in the collapsed state. Semi-IPNs show good reversible characteristics in this experiment, which is very important for drug delivery systems. By placing the collapsed gel in a pH=7.81 buffer solution, the swelling process is induced. The pH value of the media is higher than both pK a values of itaconic acid (pK a1=3.85, pK a2=5.44 [3]), so ionization of carboxylic groups occurs, inducing a more pronounced swelling of the gel. The electrostatic repulsions between the COO groups is responsible for the fast advancement of the swelling from the gel surface layer to the center of the gel. The GVand GVvalues are independent of frequency over the whole experimental range (from 10 to 100 rad/s) for the semi-IPNs as formed, while for samples with equilibrium water content, both GVand GVincrease with increasing x. The higher GVvalues of the semi-IPNs compared to those of the copolymer are due to the presence of PEG chains incorporated in the network, which improves the shear resistance. GV decreases with increasing equilibrium degree of swelling values, which is to be expected. The PAAm/IA gel which exhibits the largest swelling has the lowest GV value. On the other hand, the GV values depend on the polymer content (Fig. 2). As can be seen, with increasing

Fig. 1. Equilibrium swelling degree as a function of pH and of PEG concentration.

Poster Abstracts


Fig. 2. GV as a function of the PEG wt.% of the hydrogels: (a) as formed; (b) at equilibrium swelling. percent of PEG over 5 wt.% in the semi-IPN (Fig. 2a) lower shear moduli were recorded, probably due to the plasticizing effect of the PEG. The semi-IPNs with 10 wt.% of PEG showed the best mechanical properties at equilibrium swelling. Conclusion PAAm/IA copolymer and semi-IPNs with poly(ethylene glycol) show pH-sensitive swelling behaviour due to the different extent of IA ionization. The equilibrium swelling ratio increases when the pH value of the solution is above the nominal pK a1 value of the first group in PIA and continues to increase above the second pK a2 value of PIA, due to the ionization of the carboxylic groups. Interpenetration of PEG in the network leads to lower degrees of swelling, with respect to the copolymer hydrogel, in the pH range from 3.83 to 7.81, due to a reduction of the electrostatic repulsions. The semi-IPNs show good reversibile characteristics in the swelling–deswelling experiments, which is very important for their use in drug delivery systems. The storage shear moduli of the copolymer hydrogel and semi-IPNs, as formed, were independent of the frequency over the whole experimental range, while GV at equilibrium decreases as the degree of swelling increases. The PAAm/IA hydrogel which exhibits the largest swelling, has the lowest GV value. The GV values of the semi-IPNs are higher compared to those of the copolymer and depend on the polymer content. The loss moduli, GV shows the same trend as GV. References [1] N.A. Peppas, P. Bures, W. Leobandun, H. Ichikawa, Hydrogels in pharmaceutical formulations, Eur. J. Pharm. Biopharm. 50 (2000) 27–46. [2] M.B. Hugglin, M.B. Zakaria, J. Appl. Polym. Sci. 18 (1986) 8–13. [3] R.C. Wheast, editor. Handbook of Chemistry and Physics, 53 edn., Cleveland, Ohio, 1975.


Poster Abstracts

EFFECT OF LOADING ON A SWELLING-CONTROLLED DRUG DELIVERY SYSTEM STUDIED BY OPTICAL TECHNIQUES A. Stavropoulou, K.G. Papadokostaki, M. Sanopoulou Institute of Physical Chemistry, National Center for Scientific Research bDemokritosQ, 153 10 Ag. Paraskevi, Athens, Greece Summary A combination of optical techniques is used to obtain kinetical data of water penetration in, and concurrent drug release from, PVA matrices loaded with a highly soluble drug, as well as information on water and drug distribution profiles. It is found that the degree of drug loading substantiously affects the kinetics and the rate of water penetration as well as the drug release rate. Introduction Solvent-activated controlled release devices consisting of a bioactive agent incorporated in a swellable polymeric matrix are cheap and easy to manufacture but suffer from the disadvantage that they cannot normally sustain a reasonably constant delivery rate (zero-order release kinetics). However, this disadvantage can be greatly alleviated by optimizing the design of such controlled release devices. In order to achieve this goal, the various parameters affecting the release rate and the mechanisms of solvent penetration and concurrent drug release kinetics need to be well understood. In previous work in our laboratory, polymer films have been clamped between two glass plates to determine the kinetics of solvent penetration fronts and the corresponding concentration profile [1]. A similar technique has been used in order to obtain the relevant front kinetics and drug concentration profile in tablets consisting of hydroxypropyl methylcellulose and a colored highly soluble drug, buflomedil pyridoxal phosphate (BPP) [2,3]. In this study, we apply various optical techniques for BPP loaded, as well as for pure unloaded, poly(vinyl alcohol) (PVA) matrices to obtain information about water and drug concentration profiles in parallel to the corresponding penetration and release kinetics. Experimental methods Rectangular, loaded matrices were prepared by casting 10% aqueous solutions of the polymer powder (Aldrich, Cat No 36313-8), containing 10% and 30% w/w BPP, on Petri dishes. Similarly, pure (unloaded) matrices were also prepared. The experimental data presented below was obtained at 23F1 8C, during unidimensional penetration of water in, and concurrent drug release from, PVA matrices clamped between two glass plates. In particular, optical microscopy and microinterferometry were used to obtain water penetration kinetics and information on the corresponding water concentration profile, respectively. The drug concentration profile was obtained by optical microdensitometry, and drug release kinetics was measured spectrophotometrically at 290 nm. Details on the optical techniques have been presented elsewhere [1]. Results and discussion Optical path difference (OPD) profiles (representing the corresponding concentration profile) during penetration of water in unloaded PVA matrices are presented in Fig. 1. The sharp drop of the OPD value near the dry film marks the water penetration frontadvancing into the polymer and the corresponding drop near the pure liquid marks the swelling front S (edge of the film) moving outwards. The distances x p and x s traveled by these fronts, respectively, are shown as a function of time in Fig. 2. In the same figure, water penetration kinetics for matrices loaded with 10% and 30% w/w BPP are also included. The enhanced water penetration rate in the presence of the drug is clearly evident, especially for the higher loading. Equally important is a change on the type of kinetics. The latter can be quantified on the basis of the power law x=kt n , where k and n are constants. Under semi-infinite conditions, Fickian kinetics are characterized by n=0.5, while increasing values of n denote increasing deviations from Fickian kinetics with n=1 for zero-order transport. On the basis of this analysis, the ln x P vs. ln t plots for water penetration data in pure PVA matrices indicate non-Fickian behavior (n=0.72F0.01), characteristic of slow structural relaxations of the glassy polymer matrix occurring on a time scale comparable to the diffusion time scale. On the other hand, for both loadings studied, the kinetics has shifted to Fickian (n=0.58F0.07 for 10% BPP and n=0.57F0.02 for 30% BPP) in agreement with similar observations found in analogous systems [4]. The observed shift to Fickian water penetration in loaded matrices indicates a change in the relative contributions of the water diffusion and polymer relaxation processes. This behavior can be attributed to a faster relaxation rate resulting either (i) from a drastic lowering of the polymer’s Tg upon loading or (ii) from the osmotic effect of the hydrophilic drug which results in increased water uptake and ensuing plasticization of the polymer. In the case studied here, DSC measurements have shown that the Tg of PVA is not materially affected by BPP loading, because the respective Tgs of the pure components are very similar. Thus, for the particular system under study, the observed change in the transport behavior of water in loaded matrices should be attributed to the osmotic effect of the hydrophilic BPP. As shown in Fig. 3, drug release and water penetration occur on comparable time scales. Moreover, the rate of drug release is distinctly higher for the higher loading, in line with the assumed enhanced water uptake in this case, which is expected to increase the diffusivity of the drug in the hydrated matrix. In addition, on the basis of the abovementioned power law, the kinetics of release is purely Fickian in both cases (n=0.52F0.03 and 0.55F0.03 for 10% and 30% BPP loading, respectively). The corresponding

Poster Abstracts

Fig. 1. Normalized optical path difference profiles obtained at different times t during water penetration in unloaded PVA matrix. The position of penetration and swelling S fronts is indicated.


Fig. 2. Kinetics of water penetration front and polymer swelling front S in unloaded (open points), loaded with 10% (full points) and 30% BPP (points with crosshair), PVA matrices.

Fig. 3. BPP release kinetics from PVA matrices loaded with 10% (open points) and 30% (full points) BPP. The arrows indicate the time at which water front reaches the center of the matrix in each case (estimated from the data of Fig. 2).

Fig. 4. Optical density profiles, representing BPP concentration profiles obtained during BPP release from PVA matrices loaded with 10% and 30% BPP. The position of water penetration (P) and polymer swelling (S) fronts are indicated.


Poster Abstracts

concentration of BPP in the swollen matrix (Fig. 4) drops relatively gradually from frontup to front S for the 10% loading, but for the 30% loading, there is a max in the optical density behind front P. Similar results have been found for high loadings of BPP in hydroxypropylmethyl cellulose tablets and attributed to an undissolved drug layer near the penetration front [2,3]. A similar explanation is plausible here, since the steep gradient of the OPD profile in Fig. 1 indicates that the degree of matrix hydration near frontis expected to be low. Conclusions The application of various optical techniques on PVA matrices loaded with two different amounts of a hydrophilic drug, has provided useful information on the effect of loading on water penetration and concurrent drug release. Relaxation-controlled, nonFickian water penetration kinetics in unloaded matrices was observed. In principle, under such conditions, an approach to the desired zero order release kinetics can be achieved. The data presented here indicate that the osmotic effect of a hydrophilic drug tends to counteract that of structural relaxation, in line with previous experimental findings [4] and model calculations [5]. Therefore, exploitation of structural relaxation properties of a glassy polymer matrix for the optimization of controlled release devices should always be considered in conjunction with the osmotic properties of the drug. Equally important is the observed increase in release rate with increasing loading, since drug delivery systems designed for prolonged use often contain high amounts of drug [6]. Acknowledgements This work was partially financed by the General Secretariat for Research and Technology, Greece and the European Union, in the framework of the Program bExcellence in the Research InstitutesQ. References [1] [2] [3] [4] [5] [6]

M. Sanopoulou, J.H. Petropoulos, J. Polym. Sci., B, Polym. Phys. 30 (1992) 971–982. P. Colombo, R. Bettini, N.A. Peppas, J. Control. Release 61 (1999) 83–91. P. Colombo, R. Bettini, P.L. Catellani, P. Santi, N.A. Peppas, Eur. J. Pharm. Sci. 9 (1999) 33–40. P.I. Lee, C.-J. Kim, J. Control. Release 16 (1991) 229–236. K.G. Papadokostaki, A. Ya. Polishchuk, J.K. Petrou, J. Polym. Sci., B, Polym. Phys. 40 (2002) 1171–1188. B. Gander, R. Gurny, E. Doelker, N.A. Peppas, Pharm. Res. 6 (1989) 578–584.

Poster Abstracts


EXPERIMENTAL AND THEORETICAL STUDY OF THE RELEASE KINETICS OF PROPRANOLOL HYDROCHLORIDE FROM PVA MATRICES A. Stavropoulou, K.G. Papadokostaki, M. Sanopoulou Institute of Physical Chemistry, National Center for Scientific Research bDemokritosQ, 153 10 Ag. Paraskevi, Athens, Greece Summary Experimental kinetic data of water uptake of unloaded, and propranolol hydrochloride release from loaded, crosslinked PVA matrices is presented. The behaviour of the system, under conditions of relaxation-controlled transport behaviour, is satisfactorily reproduced by model calculations mainly based on experimentally determined input parameters. Introduction Among the most widely used controlled delivery systems are the matrix devices, mainly due to the fact that they have a low manufacturing cost. The desired constant delivery rate over prolonged time periods can be achieved, in principle, by controlling the solvent-induced swelling of hydrophilic matrices [1]. Thus, (i) experimental studies on the effect of various parameters controlling the water uptake and drug release kinetics, and (ii) computer models which can simulate realistically the strong interacting fluxes of invading water and released solute in these systems (and hence predict the resulting rate and kinetics of release) can greatly help the optimization of controlled release devices. Here, we present experimental data on drug release and water uptake kinetics in poly (vinyl alcohol)-propranolol hydrochloride (PVA-PPH) matrix system. Experimental drug and water transport parameters are used as input data in a first attempt to simulate the behaviour of the system on the basis of a computer model previously developed in our lab [2,3]. Experimental methods Rectangular, crosslinked PVA matrices of thicknesses 2L=180–270 Am were used. Crosslinking was achieved in 10% aqueous solutions of PVA (Aldrich, Cat No 36313-8) using glutaraldehyde as a crosslinking agent, with a nominal molar crosslinking ratio of glutaraldehyde to VA monomer equal to 0.05. The crosslinked matrices were loaded with two different amounts of propranolol hydrochloride, namely, 0.49 and 0.33 g/g PVA by equilibration with saturated (76 mg/ml) and 45 mg/ml aqueous solutions of the drug, respectively. Water uptake kinetics from unloaded matrices was measured gravimetrically. Drug release rates from dry matrices as well as from fully hydrated ones were determined spectrophotometrically at 290 nm. All experimental data presented below were obtained at 23F1 8C. Results and discussion Experimental water uptake and drug release kinetics. As shown in Fig. 1, water uptake kinetics in unloaded PVA matrices was prominently non-Fickian, indicating slow structural polymer relaxations, which occur on a time scale comparable to the time scale of the water diffusion process. Similarly, non-Fickian behavior also characterizes the kinetics of drug release (Fig. 2). The practical coincidence of the release curves for the two drug loadings studied, points to the absence of any discernible effect due to osmotic action of the drug on the amount of imbibed water, in the particular loading range studied.

Fig. 1. Experimental water uptake kinetics from dry, unloaded PVA matrices, expressed in terms of mean imbibed water concentration C Wt, normalized with respect to the final equilibrium value C W0, vs. t 1/2/L.


Poster Abstracts

Fig. 2. Experimental drug release kinetics from dry PVA matrices loaded with 0.49 (open points) and 0.33 g PPH/g PVA (full points), expressed in terms of the fractional amount of drug released vs. t 1/2/L.

The prominent S-shape of the release curves in Fig. 2 indicates an approach to zero-order release kinetics. As expected, the corresponding release curves from fully hydrated, and thus relaxed PVA matrices, given in Fig. 3, show good conformity to Fickian kinetics, and thus enable the determination of a reliable value of the drug’s diffusion coefficient in the swollen matrix. Description, parameterization and results of the model. The basic equations describing water uptake, drug release and structural relaxation of the polymeric matrix respectively, are:   BCW B BaW DW SW ¼ Bx Bt Bx


  BCN B BaN DN SN ¼ Bx Bt Bx


BCW BCWI BaW ¼ þ bW ðCWF  CW Þ Bt BaW Bt


where the subscripts W and N refer to water and water-soluble drug respectively; C and a denote concentration and activity respectively of water or drug within the polymer; S and D are the relevant solubility and thermodynamic diffusion coefficients; and b W is the relaxation frequency, which governs the rate at which structural relaxation of the polymer occurs. The main parameters of the model are assumed here to behave as follows:


S W=C W/a W is an increasing function of C W and is enhanced by structural relaxation of the glassy polymer. For a given a W, structural relaxation causes C W to increase from C W=C WI (for unrelaxed polymer) to C W=C WF (for fully relaxed polymer).


S N=C NS/a N (where C NS is the concentration of drug dissolved in the water swollen polymer matrix and a N is equal to the fractional saturation of the aqueous drug solution at equilibrium with C NS).


D W and b W were here taken as constants. The important parameter is L 2b W/D W, which determines the relevant rates of relaxation and water uptake. D N is a strongly increasing function of C W, defined in Ref. [2].


The important parameter here is D NE/D W(D NE=D N in the fully hydrated matrix) which determines the relevant rates of drug release and water uptake. Eqs. (1)–(3) were solved numerically by an explicit finite difference method, described in detail elsewhere [2,3]. Values of the input model parameters D W and D NE were estimated from the experimental data of Figs. 1 and 3 (1107 and 1.5107 cm2/s, respectively). The functional dependence of D N on C W was obtained from the experimental D NE at C W0 (=4.0 g/g PVA), the calculated D NS value [5] in pure water (6106 cm2/s) and literature data on diffusion coefficients of solutes in PVA [4,6]. S N was measured at different a N values and found to be constant. Based on the observed S-shaped water uptake kinetics of Fig. 1 reasonable

Poster Abstracts


estimates of L 2b W/D W were used, together with adjustable C WI/C WF values, to fit the experimental water uptake data of Fig. 1. A fairly accurate fit was obtained with values of L 2b W/D W=4 and C WI/C WF=0.2, which were then used as input parameters for the calculation of the drug release kinetic curves. Representative experimental results of Figs. 1 and 2 are replotted in Fig. 4 on a t/L 2 scale, together with the corresponding model calculations. As it can be seen, both the shape as well as the relative rates of water uptake from unloaded matrices and of drug release from loaded ones are satisfactorily reproduced. The substantial linear parts of the drug release curve on this plot indicate an approach to constant release rate.

Fig. 3. Experimental drug release kinetics from fully hydrated PVA matrices loaded with 0.49 (open points) and 0.33 g PPH/g PVA(full points).

Fig. 4. Comparison of representative experimental water uptake and drug release kinetics with model calculations. For model input parameters, see text.

Conclusions The water uptake in, and PPH release from, PVA matrices under conditions of relaxation-controlled transport behaviour can be satisfactorily reproduced by model calculations mainly based on experimentally determined input parameters. Future work will focus on more extensive testing of the model over a broad range of experimental conditions, by detailed estimation of the requiredinput parameters from independent experimental studies on water and solute sorption and diffusion properties.

Acknowledgements This work was partially financed by the General Secretariat for Research and Technology, Greece and the European Union, in the framework of the Program bExcellence in the Research InstitutesQ.

References [1] [2] [3] [4] [5] [6]

C.S. Brazel, N.A. Peppas, Polymer 40 (1999) 3383–3398. J.H. Petropoulos, K.G. Papadokostaki, S.G. Amarantos, J. Polym. Sci., B, Polym. Phys. 30 (1992) 717–725. K.G. Papadokostaki , A. Ya. Polishchuk, J.K. Petrou, J. Polym. Sci., B, Polym. Phys. 40 (2002) 1171–1188. B. Amsden, Macromolecules 31 (1998) 8382–8395. Y.W. Chien, S.H. Yalkowsky, T.J. Roseman, J. Pharm. Sci. 63 (1974) 500. H. Matsuyama, M. Teramoto, H. Urano, J. Membr. Sci. 126 (1997) 151–160.


Poster Abstracts

SYNTHESIS OF NOVEL POLY(ETHYLENE GLYCOL)-BASED POLYURETHANES FOR DRUG DELIVERY SYSTEMS J. Tuominen, J.J.N. Lee, M. Livingstone, J.A. Halliday Controlled Therapeutics (Scotland), 1 Redwood Place, Peel Park Campus, East Kilbride, G74 5PB, United Kingdom Summary In this study, the effects of molar ratio of monomers, length and type of diol and poly(ethylene glycol) (PEG) on the properties of linear poly(ethylene glycol) based polyurethane were investigated. The swelling properties of these novel linear hydrogel polymers and the dissolution profiles of the model active agents indicated that these types of polymers could be loaded with pharmaceutically active agents to produce controlled release compositions such as pessaries, buccal inserts or films. Introduction Poly(ethylene glycol)-based crosslinked polyurethane (PU) hydrogels have been used successfully in different drug delivery applications [1– 3].These types of water-swellable polymers have been loaded with many different types of pharmaceutically active agents and the ability of releasing active agent over a prolonged period of time has opened up their commercial use. However, crosslinked polyurethane hydrogels have some limitations in their use as a polymer matrix for drug delivery systems. Only water-soluble active agents can be used and the molecular weight of the active agent is limited by the degree of polymer swelling. Wider range of swelling and more variety of polymer processing methods are desired properties for PEG-based polyurethanes. In this study, we investigated the effects of molar ratios of monomers, length and type of diol and PEG on the properties of linear poly(ethylene glycol)-based polyurethane. Experimental methods Linear PEG-based polyurethanes [4] were polymerised by using various molecular weight hydroxyl-terminated PEGs (PEG 4000, PEG 8000, PEG 12000 or PEG 35000), different type of diols (1,6-hexanediol [HD], 1,10-decanediol [DD], 1,12-dodecanediol [DDD] or 1,16hexadecanediol [HDD]) and dicyclohexylmethane-4,4-diisocyanate (DMDI) with different molar ratios of PEG:diol:diisocyanate (0.1:1:1.1 to 1.5:1:2.5). Generally, these polymers were made by melting the dried PEG together with the diol at 85 8C. The molten mixture was dried under vacuum at 95 8C to remove excess moisture. The catalyst, ferric chloride, was mixed with a small amount of molten mixture and fed to the reaction vessel together with the PEG–diol mixture before DMDI addition. The reaction mixture was agitated for 150 s at 427 rpm before the polymer was poured into billet moulds and reacted for 10 h at 95 8C. After cooling to ambient temperature, polymer was removed from the mould and polymer slabs were sliced. These pessaries were purified before loading model drugs and dissolution study. Crosslinked PEG based polyurethanes were polymerised as described elsewhere [1]. Results and discussion The molecular weight of PEG and the type of the diol had clear effects on the swelling of hydrogel pessaries as seen in Fig. 1. By increasing the molecular weight of PEG from 4000 to 12,000 g/mol, the swelling of pessaries increased almost linearly from 400% to 1000%, and by increasing the amount of CH2 in diol, the swelling of pessaries decreased almost linearly from 900% to 500%. Changing the molar ratios between PEG, diol and DMDI as shown in Table 1, could also control the swelling over a wide range (from 200% to 1000%).

Fig. 1. The effect of molecular weight of PEG and the number of CH2 on the swelling of pessaries.

Poster Abstracts


Table 1 Molar ratio between PEG 8000 and 1,10-decanediol was changed PEG 8 000 (molar ratio) DD (molar ratio) DMDI (molar ratio) Percentage swelling (%) WSE (%) Tm (8C) Crystallinity (%)

0.9 1 1.9 1050 2.3 62.4 48.6

0.7 1 1.7 750 1.4 62.4 49.3

0.1 1 1.1 200 2.3 54.9 33.1

In vitro dissolution studies (Fig. 2) of hydrogel pessaries showed that the dissolution profile of model drugs can be tailored by changing the structure and swelling of hydrogel polymer. High-swelling polymers may enhance the possibility of loading larger molecules and achieving higher doses than current crosslinked polymer.

Fig. 2. Mean dissolution profile (n=6) of misoprostol from crosslinked (CLP), linear high swelling (LHP) and linear low swelling (LLP) polymer pessaries.

Conclusion The swelling properties of novel linear poly(ethylene glycol)-based polyurethanes can be tailored to a wide range by changing the molar ratios of monomers and by independently changing the type and length of diol and PEG. Linear polymers behaved like crosslinked polymer in aqueous environments, but were thermoplastic and solvent soluble. Thus, a wide range of conventional polymer processing methods can be used in the manufacture of drug release devices. These linear polymers bring more variety to existing crosslinked PU hydrogels and the new properties can be utilized in the design of future drug delivery systems.

Acknowledgements Financial support via SPURPlus from the Scottish Executive is gratefully acknowledged.

References [1] [2] [3] [4]

M.P. Embrey, N.B. Graham (1983), GB2047093 B. M.P. Embrey, N.B. Graham (1983), GB2047094 B. J.A. Halliday, S. Robertson (2002), US6,488,953. J.A. Halliday et al. Pat. Appl. (2002), GB0222522.5.


Poster Abstracts

INJECTABLE AMPHIPHILIC HYDROGELS BASED ON DEGRADABLE PLA–PEO–PLA TRIBLOCK COPOLYMERS: THE NEED TO REVISIT PEG–PROTEIN INTERACTIONS D. Domurado1, M. Domurado1, J.-C. Gris2, M. Vert1 1 Centre de Recherche sur les Biopolyme`res Artificiels, CNRS-University Montpellier 1, UMR CNRS 5473, Faculty of Pharmacy, 15 avenue Charles Flahault, BP 14491, 34093 Montpellier cedex 5, France 2 Laboratoire d’He´matologie, UFR des Sciences Pharmaceutiques, University Montpellier 1, 15 avenue Charles Flahault, BP 14491, 34093 Montpellier cedex 5, France Introduction For many years, hydrogels have been regarded as systems of great interest to deliver proteins and other water-soluble or water-dispersed macromolecules of natural origin. However, entrapping temporarily a protein and delivering it slowly is not enough. One has to take into account the respect for living systems and the need to eliminate the delivery systems from the human body after healing or load exhaustion. This particular point has generated the search for hydrogels that can degrade in situ, i.e., under physiological conditions. Because of their sensitivity to hydrolysis, the presence of lactic acid-based (PLA) segments in a hydrogel polymeric network was regarded as a means to fulfil the degradation requirement. On the other hand, ethylene oxide-based segments (PEO) appeared as suitable complement to form devices that can swell more or less depending on their lengths and the relative lengths and distribution of PLA and PEO blocks when in contact with aqueous media and body fluids. One way to generate degradable hydrogel systems was based on chemically-cross-linked PLA–PEO–PLA dimethacrylates. Another possibility to cross-link PLA–PEO–PLA triblock copolymers is to take advantage of the hydrophobicity of PLA segments that can lead to micelle or hydrogel formation, depending on the relative lengths of PLA and PEO segments. In 1996, it was shown that highly swollen hydrogels could be obtained that can be injected in soft tissues using a syringe [1]. The process is based on the fact that the mixing of PLA– PEO–PLA triblock copolymer solution, dissolved in a water-miscible organic solvent such as acetone or tetraglycol, to water or to a salted medium, leads to a very soft hydrogel, provided the PLA/PEO ratio is in a suitable composition window. Such a hydrogel swells much more in a given aqueous medium than the same copolymer processed to dry tablets first. On the other hand, it shows an amphiphilic behavior as it is able to entrap hydrophilic macromolecules such as proteins and also hydrophobic small molecules, thanks to the presence of microphased-separated nanosized PLA microdomains that also act as reversible physical cross-links. Such hydrogel can thus accommodate and deliver a hydrophilic macromolecule and a lipophilic drug simultaneously. The possibility of temporarily entrapping proteins like albumin and fibrinogen was demonstrated. However, it was found that the release profiles obtained for albumin and fibrinogen were very different. In particular, albumin was released rather slowly, suggesting some kind of interactions between the protein and the hydrogel matrix. This finding appeared in conflict with the generally admitted protein-repulsive effect of adsorbed PEG-based amphiphilic molecules or PEO segments attached to a molecule or a surface. This discrepancy led us to revisit the present understanding of the behavior of PEG or PEO–protein mixtures in aqueous media. One particularity of the conditions under which albumin and PEO or PEG segments are in presence in blood or at a solid surface is the rather high local concentrations of the two species, namely, 40 g/L for albumin and about 100–300 g/L for PEO. Experimental methods


Equal volumes of mono-hydroxyl and di-hydroxyl PEG of different molar masses (600, 2000, 3000, 8000 and 20,000) were mixed at two different concentrations (200 and 400 g/L) and of different plasma proteins at twice their physiologic plasma concentration in order to obtain final concentrations of 200 and 400 g/L for PEG and physiologic plasma concentrations for proteins: fibrinogen (4 g/L), gammaglobulins (10 g/L) and albumin (40 g/L). The precipitate was weighed. The protein concentration in the supernatant was measured through UV absorbance at 280 nm. The PEG concentration in the supernatant was measured through colorimetry using I2/IK.


Equal volumes of whole lyophilized plasma redissolved to get twice the initial concentration and of di-hydroxyl PEG of different molar masses (600, 2000, 3000 and 8000) at two different concentrations (200 and 400 g/L) were also mixed to get final physiologic plasma concentration and final PEG concentrations of 100 and 200 g/L. The precipitate was weighed. The supernatant was electrophoresed on cellulose acetate. Concentrations of eight proteins (albumin, orosomucoid, haptoglobin, transferin, IgA, IgG, IgM and C3b complement factor) remaining in solution were specifically measured by immunonephelometry.

Results and discussion Mixing of pure proteins and PEG In order to mimic the rather high local concentrations of the two species, namely, 40 g/L for albumin and about 100–300 g/L for PEO, difficult to use under analytical conditions, we mixed concentrated solutions of PEG with concentrated solutions of albumin to get final concentrations in the above range. Surprisingly, it was found that albumin and PEG were miscible, despite the selected high concentrations, provided the PEG molecular weight was small enough. With PEG at 100 g/L, only PEG 20,000 induced a slight decrease of albumin concentration. With PEG at 200 g/L, PEG 3000 gave a 50% decrease in albumin concentration and PEG 8000 an almost complete precipitation of albumin.

Poster Abstracts


In contrast, fibrinogen phase-separated from PEG 2000 at 100 g/L and from PEG 600 at 200 g/L, and thus showed incompatibility at a much lower concentration (4 g/L) that mimicked the plasma concentration of this protein. In order to match the well-documented stealth effect generated by PEO segments and the compatibility of albumin, a bchameleonQ effect was introduced [2]. This effect is based on the fact that albumin is actually very compatible with PEO segments and thus prevents the recognition of the carrier as a foreign body by other blood proteins, especially opsonins, and thus cells. IgGs were then tested under similar conditions. It was found that their behavior is intermediate between those of fibrinogen and albumin. Fifty percent still remains in solution for PEG 2000 at 100 g/L, and IgGs disappeared completely from the solution for PEG 2000 at 200 g/L. Mixing of whole plasma and PEG Whole plasma was mixed with PEG under various conditions. Phase separation was always observed. The amount of separated phase increased with PEG molar mass and concentration. Also, supernatants were analyzed by electrophoresis, and the concentrations of proteins remaining in solution were specifically measured by immunonephelometry. These experiments showed that proteins having an opsonin function (IgA, IgG, IgM and C3b complement factor) are much more prone to PEG-precipitation than proteins with transport functions (albumin, orosomucoid, haptoglobin and transferin). These results agree well with the PEO-related stealth effect and the fact that some proteins may not be repulsed by PEO segments. They also agree with old data on PEG-induced protein precipitations [3]. The identification of the origin of the compatibility of polymers like PEO with some proteins at rather high concentrations in aqueous media requires further investigation.

References [1] S. Li, L. Espartero-Sanchez, I. Rashkov, M. Vert, Patent n8 WO97/19973, Appl. 11/29/1996. [2] M. Vert, D. Domurado, J. Biomater. Sci., Polym. Ed. 11 (2000) 1307–1317. [3] B. Chesebro, S.E. Svehag, Clin. Chim. Acta 20 (1968) 527–529.


Poster Abstracts

II. Tissue Engineering VIABILITY OF SMOOTH MUSCLE CELLS CULTURED ON COLLAGENOUS SCAFFOLDS FOR TISSUE ENGINEERING OF BLOOD VESSELS P. Buijtenhuijs1,2, L. Buttafoco1, A.A. Poot1, L.M.Th. Sterk3, R.A.I. de Vos3, R.H. Geelkerken2, I. Vermes1,2, J. Feijen1 1 Institute for Biomedical Technology (BMTI), Polymer Chemistry and Biomaterials, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands 2 Hospital Group Medisch Spectrum Twente, Department of Clinical Chemistry, P.O. Box 50.000, 7500 KA Enschede, The Netherlands 3 Laboratory of Pathology Oost-Nederland, P.O. Box 377, 7500 AJ Enschede, The Netherlands Summary To characterize cell cultures for tissue engineering purposes, a new method is developed to study cell viability. Markers of both proliferation and apoptosis were analyzed in one single assay with use of a semiquantitative real-time RT-PCR method. In this way no influence on growth behaviour of crosslinking of collagenous scaffolds was observed. These results resemble (qualitative) results of histology. We suggest that this test is suitable to standardize cell culturing for tissue engineering applications. Introduction The final aim of this study is to produce a tissue-engineered (TE) autologous vascular prosthesis, exclusively composed of biological materials and vascular cells, which is compatible with blood flow and can be used as a graft in vascular surgery especially for small-diameter applications (b6 mm). Our approach to make such an artificial blood vessel is that it will be characterised by the three-layered structure of a natural blood vessel: the intima, the media and the adventitia. The first step of this project includes culturing human vascular smooth muscle cells (SMCs) in a biodegradable tubular scaffold composed of collagen and elastin to obtain a media of a TE blood vessel. Here, static culturing of SMCs on top of flat porous scaffolds composed of collagen and elastin and the development of a new method to test viability of these cell cultures are described. To measure viability of the cell cultures, proliferation and programmed cell death (apoptosis) were analysed by measuring cyclin E [1] and tissue transglutaminase (tTG) [2] mRNA expression levels with use of a semiquantitative RT-PCR. This method can be used to test the phenotypic state of SMCs to compare growth behaviour of cells of different batches/sources and to compare growth behaviour of cells cultured on (and in) different collagenous scaffolds. Experimental methods SMCs were isolated from umbilical cord veins and identified with use of a specific monoclonal antibody against alpha-smooth muscle actin (alpha-SMA) and vimentin [3]. Cells were seeded and subsequently cultured on flat porous films composed of type I insoluble collagen (derived from bovine achilles tendons) and insoluble elastin (from equine ligamentum nuchae). Cross-linking of the films was performed either with a water-soluble carbodiimide or with a diamine, in presence of the carbodiimide. These scaffolds were optimised from a physical and chemical point of view. Cell attachment and growth were verified by histology using a standard Elastic Von Gieson staining procedure. Proliferation and apoptosis of the cells were determined by measuring cyclin E and tTG mRNA expression levels respectively. Cells were lysed; total RNA contents were isolated and cyclin E and tTG mRNA expression levels were determined by using a semiquantitative RT-PCR method on a realtime TaqMan analyser [4]. To set up this method for SMC cultures, cells were cultured on gelatin-coated tissue culture polystyrene (g-TCPS) for up to 36 h. Culture medium supplemented with 20% serum was used as a control and medium without serum (serum starvation) was used to induce apoptosis. mRNA expression levels of the two proteins were normalised to mRNA expression levels of porphobilinogen deaminase (PBGD) and compared to the mRNA expression levels at time point 0 (t=0). Ratios of cyclin E and tTG mRNA expression levels were calculated to make a quantified comparison of proliferation vs. apoptosis. Results are shown as meanFstandard errors of the mean of separate triplicate experiments. Cyclin E and tTG mRNA expression levels of cells cultured on native and crosslinked collagenous scaffolds were analysed in the same way. These levels were also normalised to PBGD mRNA expression levels, but for this purpose, expression levels of cells cultured on crosslinked scaffolds were compared with expression levels of cells cultured on native ones. Ratios of cyclin E and tTG mRNA expression levels were calculated and preliminary results are shown as meanFstandard errors of the mean of duplicate experiments. Results and discussion Human umbilical vein SMCs, were cultured on porous films composed of collagen and elastin. Cells adhered and grew in multilayers on top of the porous films and (to a lesser extent) in between the fibres of collagen and elastin after 14 days of static culturing. No influence of crosslinking of the scaffolds on growth behaviour of SMCs was observed by histology (see Fig. 1).

Poster Abstracts


Fig. 1. Histology (EG) of human umbilical vein SMCs cultured for 14 days on porous scaffolds composed of collagen and elastin. Scaffolds were noncrosslinked (A) or crosslinked with a carbodiimide (B) or crosslinked with a diamine in the presence of a carbodiimide (C). When SMCs were cultured on g-TCPS, tTG mRNA expression levels increased after 12 h in SMCs cultured in medium without serum as shown in Fig. 2A. Control cell cultures did not show this increase. Cyclin E mRNA expression levels were less influenced by serum starvation (Fig. 2B). mRNA expression levels of cyclin E and tTG were nearly constant in cells cultured on g-TCPS in medium with 20% serum. Ratios of cyclin E and tTG mRNA expression levels showed a significant reduction during growth of cells in medium without serum compared to medium with 20% serum as expected (Fig. 2C).

Fig. 2. mRNA expression levels of tTG (A) and cyclin E (B) and ratios of cyclin E and tTG mRNA expression levels (C) compared to levels at t=0 of umbilical vein SMCs plotted against time of culturing. Cells were cultured in medium with (control, black) and without serum (serum starvation, grey) on gelatin coated TCPS for up to 36 h. When SMCs were cultured on TE collagenous scaffolds, no influence of crosslinking of the scaffolds on growth behaviour was observed after 7 days of culturing (Fig. 3).

Fig. 3. mRNA expression levels of tTG (A) and cyclin E (B) and ratios of cyclin E and tTG mRNA expression levels (C) of SMCs cultured for 7 days on TE scaffolds. mRNA expression levels of cells cultured on native scaffolds were compared to expression levels of cells cultured on crosslinked scaffolds. Crosslinking was performed either with a carbodiimide or with a diamine.

Conclusions Human vascular SMCs of mesenchymal origin adhere and proliferate on flat scaffolds composed of insoluble collagen and elastin. A new method is developed to characterize and compare cell growth behaviour of SMCs cultured on and in TE scaffolds by measuring and comparing proliferation vs. apoptosis in one single assay. Serum starvation can be used in this test as a positive control for tTG mRNA expression levels. Crosslinking of TE scaffolds either with a water-soluble carbodiimide or with a diamine, in presence of the carbodiimide does not influence SMC cell behaviour. In addition this new method can be used to characterize and compare cell growth behaviour of different batches of cells. We suggest that with this approach, it will be possible to culture cells in a standardized way not only for obtaining an artificial media of a TE blood vessel using SMCs but also for all kinds of tissue-engineering purposes using other cell types.


Poster Abstracts

References [1] [2] [3] [4]

M. Ohtsubo, A.M. Theodoras, J. Schumacher, J.M. Roberts, M. Pagano, Mol. Cell. Biol. (1995) 2612–2624. E.B. Volokhina, R. Hulshof, C. Haanen, I. Vermes, Apoptosis (2003) 673–679. H. Heimli, H. Kahler, M.J. Endresen, T. Henriksen, T. Lyberg, Scand. J. Clin. Lab. Invest. (1997) 21–29. C.A. Heid, J. Stevens, K.J. Livak, P.M. Williams, Genome (1996) 986–994.

ELECTROSPINNING COLLAGEN AND ELASTIN FOR TISSUE ENGINEERING SMALL DIAMETER BLOOD VESSELS L. Buttafoco1, N.G. Kolkman1, A.A. Poot1, P.J. Dijkstra1, I. Vermes1,2, J. Feijen1 1 Institute for Biomedical Technology (BMTI), Polymer Chemistry and Biomaterials, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands 2 Laboratory of Medisch Spectrum Twente, P.O. Box 50000, 7500 KA Enschede, The Netherlands Summary Nonwoven fibrous matrices of collagen and elastin have been prepared by means of electrospinning. Variations in the morphology of the scaffolds have been evaluated as a function of the ratio of the two proteins. Scaffolds having an architecture resembling the media of natural blood vessels have been obtained. Crosslinking has been used as a valuable method to improve the stability of such matrices. The obtained nonwoven matrices can be used for tissue engineering applications. Introduction Tissue engineering of small diameter (b6 mm) blood vessels is regarded as an excellent opportunity to overcome the problems and poor performance of synthetic artificial blood vessels. Our approach comprises the design and development of a biodegradable porous threelayered tubular scaffold. Electrospinning is used to reach this aim. In this technique, a polymer solution is subjected to an electric field that permits the formation of a fibre from a charged jet. An advantage of this technique is that fibres having a diameter in the range of a few hundred nanometres can be produced. This permits to obtain scaffolds with a large surface area and a high porosity, two essential characteristics for cell culturing.

Experimental methods Solutions of collagen/elastin/PEO (C:E:P) are prepared in hydrochloric acid, containing NaCl and are then charged in a syringe connected to a syringe pump in a horizontal mount. The solution is directed to a capillary blunt needle tip through a silicone tube. An electronic potential is applied to the needle by an electrode connected to a high-voltage supply. Fibres are collected either on a rotating mandrel placed between the capillary tip and a grounded aluminium plate or on the grounded aluminium plate itself. A charged steel ring is placed perpendicular to the jet at the end of the capillary tip to stabilize the jet and to direct it downwards (Fig. 1).

Poster Abstracts


Results and discussion Solutions of C:E:P having different weight ratios of collagen and elastin have been prepared in order to study the influence of spinning conditions on the morphology of the resulting meshes. Fibres are formed when the repulsive force produced by the mutual charge repulsion in the drops at the tip of the spinneret, exceeds the surface tension [2]. Addition of PEO is necessary to produce fibres, while the presence of NaCl ensures a better conductivity of the solution. In this study, the effect of the concentration of elastin is evaluated since it is known that solution viscosity and surface tension of the solution are the main factors determining the formation of electrospun fibres [1]. Under otherwise similar conditions, a decrease in elastin concentration causes an increase in surface tension. As a consequence, the voltage needed to produce a fibre increases as well (data not shown). A decrease in elastin also causes an increase in viscosity, as can be seen from the rheometry properties of different solutions (Fig. 2). Both phenomena can be explained by the higher hydrophobicity of elastin compared to collagen. Increasing the concentration of the former protein leads to an increase in hydrophobic elements both at the surface of the solution and in the bulk, thus causing the observed decrease in viscosity and surface tension.

Fig. 2. Rheometry properties of different solutions. As expected, variations in the composition of the spun solution influence the morphology of the obtained fibre. In particular, a higher viscosity favours the formation of flat fibres [3]. This is verified by spinning at exactly the same conditions three different solutions having a different collagen:elastin ratio; a higher ratio ensures the formation of completely circular fibres while a lower one is responsible for the production of completely flat fibres (Fig. 3a, b, c).

Fig. 3. SEM images of the fibres obtained from solutions having different compositions: (a) C:E=1:1; (b) C:E=1:2; (c) C:E=1:3. All the fibres have been spun at 22 kV and collected at 20 cm distance. Magnification is 10 kV. Scale bars are 300 nm. Moreover, fibres containing a higher percentage of elastin have bigger diameters (Fig. 4a, b, c). The obtained nonwoven meshes have been crosslinked with a soluble carbodiimide in order to prevent solubilisation in aqueous environment and to improve the mechanical performance. Crosslinking efficiency is determined by measuring the amount of free amino


Poster Abstracts

groups left after the reaction, since this process occurs by formation of peptide-like bonds between the carboxylic groups of aspartic or glutamic acid residues and the free amine residues of lysine and hydroxylysine groups. A higher thermal stability is also observed as a result of crosslinking. For example, after crosslinking, only 26% amino groups are left free in a C:E=1:1 sample, while its denaturation temperature increases from ~46 to 79 8C. Moreover, PEO is washed out from the sample during the crosslinking procedure, as verified by differential scanning calorimetry and solid state CP-MAS 13C NMR. Crosslinking does not affect the morphology of the obtained scaffolds.

Fig. 4. A higher content in elastin leads to fibres having bigger diameters: (a) C:E=1:3, diameter=600 nm; (b) C:E=1:2, diameter=330 nm; (c) C:E=2:1, diameter=220 nm. All the fibres have been spun at 22 kV and collected at 20 cm distance. SEM images are taken at 2 k magnification; scale bars are 1 Am. Conclusions Three-dimensional, porous, nanoscale fibre-based collagen/elastin matrices have been produced by means of electrospinning. Unwoven meshes have been successfully crosslinked by means of a water-soluble carbodiimide thus improving the stability of the scaffolds. By varying the collagen:elastin ratio, structures having a wide variety of diameters can be obtained. This will influence also their mechanical properties since they depend on the dimensions and assembly of the fibres. Thanks to these characteristics, such scaffolds could be used in different tissue engineering applications, like vascular grafts, as well as in drug delivery applications. Acknowledgements The authors are grateful to M. Smithers (University of Twente) for the SEM pictures. The project is financially supported by the IOP Senter (Den Haag, The Netherlands) (project IIE00003). References [1] C.J. Buchko et al., Processing and microstructural characterization of porous biocompatible protein polymer thin films, Polymer 40(26) (1999) 7397–7407. [2] J. Doshi, D.H. Reneker, Electrospinning Process and Applications of Electrospun Fibers, Journal of Electrostatics 35(2–3) (1995) 151–160. [3] H. Fong, I. Chun, D.H. Reneker, Beaded nanofibers formed during electrospinning, Polymer 40(16) (1999) 4585–4592.

Poster Abstracts


PROPERTIES OF GAMMA-IRRADIATED POLY(TRIMETHYLENE CARBONATE) M.A. Foks1, K.A.J. Dijkhuis1, D.W. Grijpma1, L.A. Brouwer2, M.J.A. van Luyn2, J. Feijen1 1 Institute for Biomedical Technology (BMTI), Polymer Chemistry and Biomaterials, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands 2 Department of Pathology and Laboratory Medicine, Medical Biology, Tissue Engineering, Faculty of Medical Sciences, University of Groningen, Hanzeplein 1, 9713 GZ Groningen, The Netherlands Summary In this study, poly(trimethylene carbonate) (PTMC) of various molecular weights is gamma-irradiated at doses ranging from 25 to 100 kGy. The influence of gamma-irradiation on creep resistance under static and dynamic conditions is investigated. Also, the influence on the in vivo degradation rate is studied. The results show that crosslinking is an attractive way to improve the properties of high molecular weight PTMC for applications as cardiovascular tissue engineering scaffolds. Introduction Cardiovascular tissue engineering requires the use of a porous scaffold in which cells are cultured. To provide mechanical stimuli to the cells, the scaffold is subjected to mechanical stresses in a bioreactor. As a result, the cells excrete highly organized extracellular matrix (ECM) [1,2]. The scaffold should be resistant to long-term cyclic loading; ideally it is constructed from a biodegradable elastomeric material. We have shown that high molecular weight PTMC shows rubber-like properties despite being totally amorphous and not cross-linked [3]. However, creep resistance of the material is limited. The introduction of chemical and/or physical crosslinks should overcome these limitations. Gamma-irradiation introduces chemical crosslinkages into the polymer and simultaneously sterilizes the specimen. Furthermore, this method does not require additional chemicals. Experimental methods PTMC of various molecular weights was synthesized and purified as reported before [3]. Compression molded (140 8C, Fontijne laboratory press THB008, The Netherlands) films and discs were vacuum-packaged and gamma-irradiated at 0 (unirradiated), 25, 50 or 100 kGy from a 60 Co source (Isotron Nederland, The Netherlands). Equilibrium swelling experiments in chloroform were performed at room temperature. Gel contents and swelling ratios were calculated from the initial weight, the swollen weight and the dry weight of the extracted specimens. The densities of chloroform and PTMC used in the calculations were 1.48 and 1.31 g/ml respectively. Mechanical properties were determined in triplicate on films (10050.6 mm3). Tensile tests were performed at a crosshead speed of 50 mm/min, with an initial grip-to-grip distance of 50 mm using a Zwick Z020 universal tensile testing machine (Germany) at room temperature. Static creep tests were performed at 50% of the yield stress, by loading a sample with the appropriate weight and periodically measuring the elongation between two marks on the specimen. The permanent deformation of specimens loaded for three days was determined after a 1-week recovery period. Dynamic creep tests were performed by cyclic loading (20 cycles) of films to 50% strain at a crosshead speed of 50 mm/min with an initial grip-to-grip distance of 50 mm. The 21st cycle was started after a 2-h recovery period, after which the permanent deformation was determined. Circular compression molded and gamma-irradiated (25 kGy) PTMC films (1 cm in diameter, 500 Am thick) were subcutaneously implanted in rats (male Wistar rats of approximately 3 months of age). At predetermined times the films were explanted and their mass and thickness were determined. Results and discussion Physical properties Gamma-irradiation of PTMC results in crosslinking. Fig. 1 shows the gel fraction of the resultant networks after irradiating PTMC of various molecular weights at 25 kGy. The figure shows that a minimum molecular weight is required to form a gel. The highest gel fractions are obtained for polymers with the highest molecular weights. The applied radiation dose also has an influence on the gel fraction: PTMC (M n =300 kg/mol) irradiated at 25, 50, and 100 kGy had a gel fraction of 58%, 59%, and 76%, respectively. The short-term mechanical properties of PTMC are affected to some extent by gamma-irradiation. Irradiation of PTMC (300 kg/mol) at 0, 25, 50 and 100 kGy results in a decrease in E-modulus (6.4, 6.6, 5.1 and 4.4 MPa, respectively), a decrease in stress at yield (2.1, 1.9, 1.5 and 1.1 MPa, respectively) and an increase in stress at break (500%, 800%, 850% and 900%, respectively).


Poster Abstracts

Fig. 1. Effect of PTMC molecular weight on the gel fraction after gamma-irradiation at 25 kGy. Gamma-irradiation has a more significant effect on long-term mechanical properties of PTMC. The creep resistance under static and dynamic conditions is much improved upon crosslinking, as can be seen from Table 1. The ability to precondition a scaffold mechanically in a bioreactor highly depends on these parameters. Table 1 Static and dynamic creep behavior of PTMC before and after gamma-irradiation Radiation dose (kGy)

0 25 50 100 d

M n =110 kg/mol

M n =300 kg/mol




Creep rate (105 s1)

Permanent deformation (%)

Permanent deformation (%)

Creep rate (105 s1)

Permanent deformation (%)

Permanent deformation (%)

146 134 127 45


11.6 2.1 1 0.3

16.6 2 0.5 0.3

350 57 14 3.4

0.6 0 0.5 0.05


27 9


Sample fractured during creep test.

In vivo degradation In a 2-year time period, noncrosslinked high molecular weight PTMC did not degrade in vitro [3], whereas in vivo (in subcutaneous implantations in rats) it degraded in 3–4 weeks [4]. Figs. 2 and 3 present the results of in vivo degradation experiments of high molecular weight PTMC after gamma-irradiation at 25 kGy. Fig. 2 shows the effect of gamma-irradiation on the change in mass and thickness during in vivo degradation. It can be seen that the degradation of PTMC is retarded upon irradiation, resulting in a degradation time of approximately 6–7 weeks. Irradiated and nonirradiated PTMC degrade by surface erosion, as follows from the linear decrease in mass and thickness observed. This can be beneficial for tissue engineering applications.

Fig. 2. In vivo degradation of PTMC. The M n of nonirradiated PTMC is 320 kg/mol, the M n of the irradiated PTMC was 480 kg/mol before irradiation.

Poster Abstracts


Fig. 3. Gamma-irradiated (25 kGy) high molecular weight PTMC specimens after explantation at 0, 5, 10, 21 and 37 days. Conclusion Gamma-irradiation induces crosslinking of PTMC. Increasing the molecular weight of the polymer and the radiation dose results in higher gel fractions. Short-term mechanical properties are affected to some extent by gamma-irradiation, creep resistance under static and dynamic conditions is much improved. Gamma-irradiation decreases the rate of in vivo degradation of PTMC; complete degradation is expected at 6–7 weeks. References [1] A. Kadner, S.P. Hoerstrup, G. Zund, K. Eid, C. Maurus, S. Melnitchouk, J. Grunenfelder, M. Turina, European Journal of Cardio-Thoracic Surgery 21 (2002) 1055–1060. [2] B.S. Kim, D.J. Mooney, Transactions of the American Society of Mechanical Engineers 122 (2000) 210–215. [3] A.P. Pego, D.W. Grijpma, J. Feijen, Polymer 44 (2003) 6495–6504. [4] A.P. Pego, M.J.A. van Luyn, L.A. Brouwers, P.B. van Wachem, A.A. Poot, D.W. Grijpma, J. Feijen, Journal of Biomedical Materials Research 67A (2003) 1044–1054.


Poster Abstracts

CHEMICAL AND MORPHOLOGICAL SURFACE MODIFICATION OF PLA50: OPTIMIZATION FOR HUMAN SKIN CELL CULTURES UNDER BIODEGRADABLE POLYMERS X. Garric1,2, J.-P. Mole`s2, H. Garreau1, J.-J. Guilhou2, M. Vert1 1 CRBA- UMR CNRS 5473- University Montpellier 1, Faculty of Pharmacy, 15 Avenue Charles Flahault, 34093 Montpellier Cedex 5, France 2 Laboratoire de Dermatologie Mole´culaire, Institut Universitaire de Recherche Clinique, 641 avenue du doyen G. Giraud, 34093 Montpellier Cedex 5, France Summary Poly(a-hydroxy-acid)s, including poly(dl-lactic acid) (PLA50), appear to be of great interest in tissue engineering. The aim of this work is to improve skin cell cultures on a PLA50 support. Two types of modifications (chemical and morphological) were achieved in order to optimize the culture. Chemically modified PLA50 supports lead to better skin cell adhesion and proliferation. This study suggests that PLA50 supports could be good candidates for tissue engineering. Poly(a-hydroxy-acid)s derived from lactic (LA) and glycolic acids (GA) are bioresorbable polymers which present a great variability of mechanical properties. These various properties combined with their rates of hydrolytic degradation and surface compositions [1] appear to be of great interest in tissue engineering, including skin reconstruction. In a previous study [2], we showed that (1) human keratinocytes (HK) were able to grow on a PLA50 film; (2) their proliferation was completely inhibited when a PLA37.5GA25 polymer was used as culture support. We further demonstrated that this toxicity was the result of GA release during polymer degradation. However, we noticed that the keratinocyte growth was delayed on PLA50 film when compared to cultures on a polystyrene support even though they reached the same plateau. In order to further evaluate whether the PLA50 film could be considered a proper skin culture support, we first produced various PLA50 films and subsequently determined the adhesion and the proliferation of human dermal fibroblasts (HDF). Second, we tested whether chemical (NaOH treatment and collagen coating) or morphological (porous) modification on the PLA50 films could improve skin cell cultures. PLA50 was synthesized by ring opening polymerization using Zn hemi-lactate as initiator. The characterization was achieved by Size  P Exclusion Chromatography Mn¼ 59; 000 and Ip ¼ 1:8 . The chemical modifications consisted of NaOH treatment for 1h30 or 3h and by collagen coating. Hydrophilicity was characterized by measuring contact angles of water on a Krqss drop shape analysis system. The morphological modification was done to create porous through the film using the blended PLA/PEG ratio and dissolution of PEG in water [4]. The porosity was observed by ESEM (Environmental Scanning Electron Microscopy) on a Philips Xl30 apparatus with a tungsten lamp. Cell adhesion and proliferation was evaluated using the MTT assay [4,5]. The chemical modifications of PLA50 did not influence significantly the HDF and J2 (a murine fibroblast cell line) adhesions when compared to polystyrene used as control. However, the adhesion of HKs (from two donors) was significantly higher on hydrophilized PLA50 films than on the control film. The influence of chemical modifications on skin cell proliferation was monitored for a period of 11 days. Three types of response depending on the cell type studied were observed: (1) The J2 proliferation was not affected by the chemical modifications on PLA50 films when compared to polystyrene. (2) The HDF grew on PLA50 hydrophilized for 1h30 as they did in the control whereas the others support (NaOH 3h, and collagen coating) did not permit to reach the same values. (3) The proliferation of HKs on the PLA50 was delayed as previously described. The same delay was observed when HK were cultured on PLA 50 hydrophilized for 3h. Interestingly, collagen-coated PLA50 and PLA50 hydrophilized for 1h30 films were used, the HKs do proliferate like on polystyrene support. Morphological modifications of the support did not modify the adhesion of HDF, J2 and HKs. However, the proliferation of HKs was significantly reduced by the presence of pore when fibroblast proliferations were not affected. In summary, we showed that treatments of the PLA surface could influence the cell culture. In the case of skin cells, the hydrophilization of PLA by NaOH treatment for 1h30 improved the adhesion and the proliferation of HKs up to be comparable to the data obtained on a polystyrene support. This optimised support is a good candidate for tissue engineering. References [1] M. Vert, G. Schwach, R. Engel, J. Coudane, Something new in the field of PLA/GA bioresorbable polymers? J. Control. Release Apr 30 53(1–3) (1998) 85–92. [2] X. Garric, J.P. Moles, H. Garreau, C. Braud, J.J. Guilhou, M. Vert, Growth of various cell types in the presence of lactic and glycolic acids: the adverse effect of glycolic acid released from PLAGA copolymer on keratinocyte proliferation. J. Biomater. Sci., Polym. Ed. 13(11) (2002) 1189–1201. [3] H. Tsuji, R. Smith, W. Bonfield, Y. Ikada, Porous Biodegradable Polyesters. Preparation of Porous Poly(l-lactide) films by extraction of Poly(ethylene oxide) from their blends. J. Appl. Polym. Sci. 75 (2000) 629–637. [4] T. Mosmann, Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. J. Immunol. Methods Dec 16 65(1–2) (1983) 55–63. [5] M.B. Hansen, S.E. Nielsen, K. Berg, Re-examination and further development of a precise and rapid dye method for measuring cell growth/ cell kill. J. Immunol. Methods May 12 119(2) (1989) 203–210.

Poster Abstracts


GRAFTING OF POLY(ETHYLENE OXIDE)/POLY(BUTYLENE TEREPHTHALATE) BLOCK COPOLYMERS ONTO HYDROXYAPATITE PARTICLES M.B. Claase, D.W. Grijpma, J. Feijen Institute for Biomedical Technology (BMTI), Polymer Chemistry and Biomaterials, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands

Summary Composite materials based on hydroxyapatite (HA) particles and poly(ethylene oxide)/poly(butylene terephthalate) copolymers, PEOT/PBT, were prepared by blending methods and by grafting. Grafting of 1000PEOT70PBT30 onto HA filler particles (38–53 Am) was accomplished by directly adding HA particles to the polycondensation reaction mixture.

Introduction Biodegradable HA composite systems based on PEOT/PBT have attracted much attention [1]. Making use of the reactivity of surface hydroxyl groups, PEOT/PBT [2] copolymers have been grafted to HA particles using diisocyanate coupling agents. This resulted in improved mechanical properties of composites. In this study, we set out to covalently bind 1000PEOT70PBT30 to HA by direct surface grafting via polycondensation, avoiding the use of potentially harmful coupling agents or additional modifications of the particle surface. We aim at preparing porous PEOT/ PBT scaffolds for bone tissue engineering that can be seeded and cultured with bone marrow stromal cells before implantation in the bone defect.

Materials and methods HA-composites prepared by blending A 1000PEOT70PBT30 copolymer was prepared as previously described [3] from PEG of molecular weight 1000, 1,4-butanediol, dimethyl terephtalate and vitamin E as antioxidant. The soft segment to hard segment weight ratio was 70 to 30. To 20% (w/v) copolymer solutions in chloroform, sintered HA with a particle size of 38–53 Am was added; the suspension was precipitated in ethanol and dried. HA-composites prepared by grafting The HA particles were immediately added to the reaction mixture, which was allowed to polymerize by polycondensation for 8 h. Films prepared by compression molding Films of HA-composites prepared by blending were prepared by compression molding at 140 8C. HA-composites prepared by grafting were compression molded at 180 8C, since compression molding at 140 8C did not result in homogeneous films. The film thickness for both the composites prepared by blending and by grafting was approximately 500 Am. Water uptake Composite films were swollen in milliQ water at 37 8C for 48 h.

Tensile testing Strips of 5100 mm (HA-composites prepared by blending) or 550 mm (HA-composites prepared by grafting) were cut from compressionmolded films and subjected to tensile testing at 50 mm/min using a Zwick Z020.

Extraction of the soluble fraction of HA-composites. Compression molded HA-composite films were extracted for 48 h with chloroform using a Soxhlet setup.


Poster Abstracts

Thermogravimetric analysis (TGA) TGA measurements were performed using a Perkin Elmer TGA 7 from 50 to 700 8C at a heating rate of 10 8C/min under a nitrogen flow. Infrared spectroscopy IR spectra were recorded using a Biorad FTS-60. Extracted HA-composites, HA and 1000PEOT70PBT30 were mixed with KBr and pressed into a pellet. Results and discussion IR-spectra of the residues after extraction of the HA-composites prepared by blending and grafting were compared to the spectra of 1000PEOT70PBT30 and HA. An example (residues from HA-composites prepared with approximately 17.5 vol.% HA) is shown in Fig. 1. After extraction, the residue of the HA-composite prepared by grafting clearly shows characteristics of the absorption spectrum of the copolymer, indicating a strong bond (most likely chemical) between the copolymer and the HA.

Fig. 1. IR-spectra of 1000PEOT70PBT30, HA and the extracted residues of HA-composites prepared by blending and grafting (17.5 vol.% HA). The residues of the extracted HA-composite films were analyzed using TGA. The results are summarized in Table 1.

Table 1 TGA data of extracted composites films, prepared by compression molding Composite

vol.% HA

Onset of mass loss

mass% 1000PEOT70PBT30 remaining after extraction

Pure HA 10 vol.% HA blended§ 17.5 vol.% HA blended§ 25 vol.% HA blended§ 10 vol.% HA graftedz 17.5 vol.% HA graftedz 25 vol.% HA graftedz

98.9 95.6 96.8 97.2 11.7F0.1 71.5F3.0 95.5F0.1

No onset 385 8C 378 8C 378 8C 411F2 8C 401F1 8C 397F1 8C

– 1.7 1.2 1.1 73.9F0.2 13.1F1.7 1.7F0.1

§: Single measurement. z: Average result of a duplicate measurement. The residues of the extracted composites prepared by blending show only small amounts of copolymer (up to 1.7 mass%), this is most likely due to physical adsorption. The table shows that the residues of the extracted composites with 10 and 17.5 vol.% HA, prepared by grafting, contain high amounts of copolymer.

Poster Abstracts


Model calculations, assuming spherical and non-porous HA particles with a diameter of 38–53 Am, covered with polymer chains with a cross-sectional area of 0.13 nm2 (minimal area that a single ethylene oxide unit occupies [4] and a molecular weight of 150,000 g/mol (which is the maximum chain length obtained during polymerization in the absence of HA), result in a maximum value of 6.4–8.7 mass% of copolymer that can be expected on the HA particles. The higher amounts of copolymer observed in the extracted residues of HA-composites prepared by grafting with 10 and 17.5 vol.% HA, could be the result of a subsequent cross-linking reaction of bound (to HA) and unbound copolymer chains during the polycondensation reaction upon evaporation of the vitamin E antioxidant. The HA-composite with 10 vol.% HA prepared by grafting only showed swelling in CHCl3 and in HFIP. Gel formation was not observed for the HA-composites prepared by grafting containing 17.5 and 25 vol.% HA, although the presence of cross-linked copolymer chains cannot be excluded. Composite films were subjected to tensile testing, both in the dry and water-swollen state. Data are presented in Table 2. Table 2 Tensile properties of 1000PEOT70PBT30-HA composites. Composite

E-modulusy (N/mm2)

1000/70/30 10 vol.% HA blended 17.5 vol.% HA blended 25 vol.% HA blended 10 vol.% HA grafted 17.5 vol.% HA grafted 25 vol.% HA grafted

27F3 45F4 54F5 78F1 37F2 47F2 61F3

r max (N/mm2)

e break (%)

r break(N/mm2)

Energy up to break (Nmm)

(23F1) 11.1F0.5 (6.5F0.2) 981F153 (144F33) 10.9F0.7 (6.3F0.2) 6380F1237 (1084F334) (20F2) 7.0F0.2 (3.1F0.3) 214F32 (32F4) 6.9F0.2 (3.0F0.3) 960F188 (96F22) (15F1) 6.2F0.4 (1.2F0.3) 62F15 (11F3) 5.9F0.3 (1.1F0.3) 264F79 (13.4F6.4) (12F2) 6.8F0.1 (0.67F0.08) 34F2 (6.5F0.7) 6.7F0.1 (0.66F0.09) 133F10 (3.8F0.7) (16F1) 6.9F1.5 (3.7F0.7) 87F39 (29F6) 6.9F1.4 (3.4F0.9) 396F258 (113F48) (10F1) 7.2F0.6§ (3.3F0.2)z 271F93§ (58F11)z 7.1F0.6§ (3.2F0.3)z 1467F587§ (205F56)z (12F1) 6.1F0.2§ (2.0F0.2)z 179F20§ (28F5)z 5.8F0.1§ (1.9F0.2)z 813F104§ (57F17)z

Results after 48 h of water uptake are shown in parentheses. Experiments in fourfold (FS.D.). y The sample deformation was derived from the grip-to-grip separation; therefore the presented E-moduli are only an indication of the stiffness. § Significantly different from the corresponding values of the blended composite in the dry state. z Significantly different from the corresponding values of the blended composite in the water-swollen state. In the water-swollen state, these hydrophilic composites show substantially deteriorated mechanical properties. Significant differences are observed between the HA-composites prepared by grafting and the HA-composites prepared by blending. The HA-composites with 17.5 and 25 vol.% HA, prepared by grafting, show a higher elongation at break and a higher energy up to break both in the dry and the water-swollen state than the corresponding HA-composites prepared by blending. In contrast to our expectations, HA-composites prepared by blending and those prepared by grafting show a decrease in E-modulus in the water-swollen state with an increase in HA vol.%. Conclusions TGA and IR-data of extracted composites prepared by grafting show the presence of relatively high amounts of copolymer compared to the extracted composites prepared by blending. Both in the dry and water-swollen state, the composites prepared by grafting (with 17.5 and 25 vol.% HA) were significantly stronger and tougher than those prepared by blending. The combined IR, TGA and tensile data suggest that the copolymer is grafted onto the surface of HA particles. The incorporation of HA (either by grafting or by blending) does not result in composite materials, however, with improved mechanical properties compared to the original 1000PEOT70PBT30 copolymer, especially in the wet state. Acknowledgements We would like to thank F.J. Monteiro (INEB, Porto, Portugal) for supplying the sintered hydroxyapatite. This study was financially supported by the European Community (Brite-Euram project BE97-4612).

References [1] Q. Liu, J.R. de Wijn, C.A. van Blitterswijk, Nano-apatite/polymer composites: mechanical and physicochemical characteristics, Biomaterials 18 (1997) 1263–1270. [2] Q. Liu, J.R. de Wijn, C.A. van Blitterswijk, Composite biomaterials with chemical bonding between hydroxyapatite filler particles and PEG/ PBT copolymer matrix, J. Biomed. Mater. Res. 40 (1998) 490–497.


Poster Abstracts

[3] A.A. Deschamps, D.W. Grijpma, J. Feijen, Poly(ethylene oxide)/poly(butylene terephthalate) segmented block copolymers: the effect of copolymer composition on physical properties and degradation, Polymer 42 (2001) 9335–9345. [4] R. Myrvold, F.K. Hansen, B. Balinour, Monolayers of some ABA block-copolymers at the air–water interface, Colloids Surf., A Physicochem. Eng. Asp. 117 (1996) 27–36.

PEG–PLA HYDROGELS BY STEREOCOMPLEXATION FOR TISSUE ENGINEERING OF CARTILAGE C. Hiemstra, Z. Zhong, P.J. Dijkstra, J. Feijen Institute for Biomedical Technology (BMTI), Polymer Chemistry and Biomaterials, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands Summary PEG–PLA hydrogels were prepared by stereocomplexation of poly(lactide) (PLA) blocks of opposite chirality. Gelation occurred upon mixing aqueous solutions of both enantiomers of PEG6500–(PLA)2, PEG12500–(PLA)2 and PEG21800–(PLA)8 block copolymers. Stereocomplexed hydrogels from PEG6500–(PLA)2 polymers were found to be stable up to ~37 8C, while those from PEG12500–(PLA)2 and PEG21800–(PLA)8 block copolymers were stable up to ~50 8C. Introduction Due to their similarity with the extracellular matrix and their high biocompatibility, hydrogels have found wide interest as matrices for tissue engineering, in particular for soft tissues. Thermosensitive hydrogels composed of block copolymers of poly(ethylene glycol) (PEG) and aliphatic polyesters are promising materials for drug delivery applications and tissue engineering. PEG is known to have excellent antifouling properties and biocompatibility and is excreted by the kidney at molecular weights up to ca. 30,000 [1]. Aliphatic polyesters, such as poly(lactide) (PLA) are known to be biocompatible and render the hydrogel biodegradable. More recently, Li et al. [2], Fujiwara et al. [3] and Lee et al. [4] have shown that hydrogels can be prepared from PEG–PLLA and PEG–PDLA triblock copolymers, where the crosslinks in the hydrogel are provided by stereocomplexation between d- and l-lactide blocks. Stereocomplex-based oligolactide grafted dextran hydrogels have been developed by de Jong et al. [5]. These hydrogels provide a full preservation of enzymatic activity combined with a quantitative release and full degradation in 1 to 7 days. In this paper, we describe the solubility of a series of PEG–PLA triblock copolymers and the effect of stereocomplexation on gelation behaviour. PEG–PLA star-block copolymers with opposite chirality of PLA blocks were investigated to study the effect of multiple interaction sites on gelation and gel strength. Experimental methods PEG–(PLA)2 block copolymers were prepared by the Sn(Oct)2 catalysed ring opening polymerisation of d- or l-lactide initiated by hydroxyl groups of PEG–(OH)2 (Fig. 1) at 105 8C in toluene for 4 h. PEG–(PLA)8 block copolymers were prepared similarly at room temperature in dichloromethane for 4 h using the single site Zn catalyst Zn(Et)[S(C6H6–CH(Me)NC5H10]-2.

Fig. 1. Molecular structures of PEG–(PLA)2 and PEG–(PLA)8. Aqueous polymer solutions were prepared by dissolving the appropriate amount of polymer in distilled water at room temperature. In case of hydrogels by stereocomplexation, polymer solutions of both d- and l-enantiomers were mixed and the sample was stirred vigorously for ~2 min. For both single enantiomer polymer solutions and for polymer solutions containing both d- and l-enantiomers, temperature-dependant gelation was tested at temperatures between 20 and 60 8C with intervals of 10 8C using a water bath. At each temperature, the samples were allowed to equilibrate for at least 20 min. No flow within 20 s while inverting the vial was regarded as a gel state.

Poster Abstracts


Results and discussion Well-defined PEG–(PLA)2 block copolymers of desired molecular weights and low polydispersities (M w/M n) were obtained (Table 1) by the Sn(Oct)2 catalysed ring-opening polymerisation of l- or d-lactide initiated with dihydroxy PEGs. In case of PEG–(PLA)8 block copolymers, an eight arm hydroxy functionalized PEG was used and a single-site Zn-complex was applied to prevent gelation of the reaction mixture. The polydispersities of these star copolymers varied from 2 and 3 as determined with GPC. Similar polydispersity was observed for the starting PEG.

Table 1 Composition, molecular weight and polydispersity of PEG–PLA block copolymers Polymer

Number of lactyl units per PLA blocka

M n 1H NMR


M w/M n


9 12 10 13 10 15 19 10 15 19 9 13 10 14

7800 8300 8000 9500 13 900 14 700 15 300 14 000 14 700 15 200 27 200 29 500 27 400 30 000

7200 7400 7400 7700 12 300 12 800 13 300 12 300 12 700 13 200 – – – –

1.02 1.02 1.02 1.02 1.01 1.01 1.01 1.01 1.01 1.02 – – – –

PEG6500-(PDLA)2 PEG12500-(PLLA)2


PEG21800–(PLLA)8 PEG21800–(PDLA)8 a

Determined from the ratio of the PLA methyl protons and the PEG methylene protons in the 1H NMR spectra.

The polymers should be able to dissolve in water to facilitate stereocomplexation and to have homogeneous hydrogels. The solubility of PEG–PLA block copolymers in water is highly dependent on the block length of PEG and PLA. For example, the maximum numbers of lactyl units per PLA block for PEG6500–(PLA)2 and PEG12500–(PLA)2 are 14 and 22, respectively, in order to render these copolymers water soluble. PEG21800–(PLA)8 block copolymers were found to be water soluble up to 14 lactyl units per PLA block. It should be noted that enantiomeric triblock copolymers can also afford hydrogels above critical gelation concentrations (CGCs) at room temperature (Table 2). However, for tissue engineering applications, it is desirable that cells may be suspended into these enantiomeric polymer solutions, which upon mixing and stereocomplexation form a hydrogel. To this end, these enantiomeric polymer solutions should be fluid to allow mixing. Polymer concentrations should be lower than CGCs of the enantiomeric polymer solutions, but should be above the CGC of the stereocomplex (Table 2). For all PEG–PLA block copolymers studied, gelation occurred upon mixing of aqueous solutions of PEG–PLA polymers of opposite chirality (Fig. 2). Hydrogels from PEG6500–(PLA)2 polymers were found to be stable up to ~37 8C, while hydrogels from PEG12500–(PLA)2 and PEG21800–(PLA)8 block copolymers were stable up to ~50 8C. At higher temperatures phase, separation occurred. Preliminary results from rheology measurements showed a storage modulus of ~1 kPa for stereocomplexed PEG21800–(PLA)8 hydrogels, which is about five times higher than the value found for stereocomplexed PEG6500–(PLA)2 hydrogels. Therefore,

Fig. 2. Left: PEG6500–(PLLA12)2 15 wt.% solution at 37 8C. Right: PEG6500–(PLLA12)2+PEG6500–(PDLA13)2 15 wt.% hydrogel at 37 8C.


Poster Abstracts

PEG21800–(PLA)8 hydrogels may be more promising for tissue engineering of cartilage. Further rheology measurements will be carried out to determine gelation kinetics and gel strength. Cell culture experiments will be performed to investigate cell growth on the hydrogels.

Table 2 Minimal polymer concentration to induce gelation at room temperature Polymer

Critical gel concentration [wt.%]

PEG–PLLA hydrogels


Critical gel concentration [wt.%]

Stereocomplexed hydrogels















PEG6500–(PLLA9)2 PEG6500–(PDLA10)2 PEG6500–(PLLA12)2 PEG6500–(PDLA13)2 PEG12500–(PLLA10)2 PEG12500–(PDLA10)2 PEG12500–(PLLA15)2 PEG12500–(PDLA15)2 PEG12500–(PLLA19)2 PEG12500–(PDLA19)2 PEG21800–(PLLA9)8 PEG21800–(PDLA10)8 PEG21800–(PLLA13)8 PEG21800–(PDLA14)8

25 15 30 10 7.5 25 5

Conclusions PEG–PLA hydrogels can be prepared by stereocomplexation of PLA blocks. Preliminary results indicate that stereocomplexed PEG21800– (PLA)8 hydrogels have potential for tissue engineering of cartilage.

Acknowledgements This study is financed by the Netherlands Organisation for Scientific Research (NWO).

References [1] T. Yamaoka, Y. Tabata, Y. Ikada, Distribution and Tissue Uptake of Poly(ethylene glycol) with Different Molecular Weights after Intravenous Administration to Mice. J. Pharm. Sci. 83(4) (1994) 601–606. [2] S. Li, M. Vert (2003), Synthesis, Characterization, and Stereocomplexation-Induced Gelation of block Copolymers Prepared by RingOpening Polymerization of l(d)-Lactide in the Presence of Poly(ethylene glycol), Macromolecules. [3] T. Fujiwara et al., Novel thermo-responsive formation of a hydrogel by stereo-complexation between PLLA–PEG–PLLA and PDLA–PEG– PDLA block copolymers, Macromol. Biosci., 1(5) (2001) 204–208. [4] C.W. Lee et al., Characterization and biocompatibility with dispersed solution of PLA–POE–PLA block copolymer, Polymer (Korea) 26(2) (2002) 174–178. [5] S.J. de Jong et al., Physically crosslinked dextran hydrogels by stereocomplex formation of lactic acid oligomers: degradation and protein release behavior, J. Control. Release, 71(3) (2001) 261–275.

Poster Abstracts


PREPARATION OF HYDROGELS BY PHOTO-CROSSLINKING OF FUMARIC ACID MONOETHYL ESTER (FAME) FUNCTIONALIZED OLIGOMERS Q. Hou, D.W. Grijpma, J. Feijen Institute for Biomedical Technology (BMTI), Polymer Chemistry and Biomaterials, Faculty of Science and Technology, University of Twente, PO Box 217, 7500 AE Enschede, The Netherlands Summary Linear and star-shaped, hydroxy-terminated polyethylene glycol (PEG) and 1,3-trimethylene carbonate (TMC)-based oligomers were functionalized by esterification with fumaric acid monoethyl ester (FAME) in the presence of N,N-dicyclohexyl carbodiimide (DCC) and 4dimethylamino pyridine (DMAP) at room temperature. Networks, designed to release only nontoxic degradation products, were obtained by UV irradiation of the macromers using 2,2-dimethoxy-2-phenylacetophenone (DMPA) as photo-initiator. The swelling properties of the hydrogels in water largely depend on the composition and architecture of the macromers, as well as on the temperature. Introduction Due to their excellent biocompatibility, good control of solute transport characteristics and tunable physical properties, hydrogels are widely being employed as drug delivery vehicles, cell carriers and tissue engineering matrices [1,2]. Especially hydrogels prepared from (meth)acrylated PEG and poly(lactide) macromers have been much studied [3]. However, degradation products and (meth)acrylate residues might be toxic. For this reason, the use of macromers based on fumaric acid derivatives is expected to be advantageous [4]. In block copolymers based on hydrophilic PEG and hydrophobic poly(TMC), the poly(TMC) segment degrades in vivo to nonacidic degradation products. This makes these systems interesting for the controlled release of sensitive biologically active molecules such as peptides and proteins. Furthermore these block copolymers display thermo-sensitive behaviour. This paper describes the synthesis and properties of hydrogels by UV photo-crosslinking of FAME functionalised oligomers based on PEG and TMC. Materials and methods Hydroxy-terminated oligomeric precursors Linear PEG (molecular weights 2000 and 4000) were obtained from Fluka, 3-arm star-shaped PEG (molecular weight 10 000) was obtained from Shearwater. Hydroxy-terminated block copolymeric precursors were prepared by ring opening polymerisation of TMC (Boehringer Ingelheim) in the presence of linear or star-shaped PEG and stannous octoate (Sigma). The TMC content was 50 mol%. Functionalization with FAME Macromers were prepared by reacting the hydroxy-terminated oligomeric precursors with FAME (Aldrich) in dichloromethane at room temperature, using DCC (Fluka) as a coupling agent and DMAP (Aldrich) as a catalyst [5]. The macromers were purified by filtration and precipitation in petroleum ether of chloroform solutions. Photo-crosslinking of the macromers Macromer and DMPA photo-initiator (Aldrich) (1 wt.% of macromer) were mixed in chloroform and cast on glass. Transparent films 0.2 to 0.5 mm thick were exposed to UV light (15 W, 360 nm, Philips) for 3 h, the distance between the UV source and the sample was 10 cm. After photo-crosslinking, the gel content of the resultant networks was determined. Characterizations The synthesized oligomers and macromers were characterized by nuclear magnetic resonance (NMR) spectroscopy. Thermal properties of the oligomers and the polymer networks were determined after quenching by differential scanning calorimetry (DSC) at a heating rate of 10 8C per min. Network gel content determinations using chloroform and water uptake experiments were performed at 25 8C. Mechanical properties were determined at room temperature at 50 mm/min according to ASTM 882-91.

Results and discussion After the purification step, FAME-functionalised oligomers (macromers) were obtained in yields of approximately 80%. NMR analysis of the purified products shows that the esterification reaction between FAME and the hydroxy-terminated oligomers under the applied mild reaction conditions is successful.


Poster Abstracts

Resonances in the NMR spectrum at 6.80–6.84 ppm (A, 2H), 4.20–4.24 ppm (B, 2H) and 1.25–1.29 ppm (C, 3H) can be assigned to the FAME component. By calculations based on the initial molecular weight of the hydroxy-terminated precursors (GPC data not shown) and the macromer composition determined by NMR, the degree of functionalization with FAME is estimated to be 90%. Preliminary experiments, in which the photo-initiator and the irradiation time were varied, showed that irradiation of macromer films at 360 nm in the presence of 1 wt.% DMPA for 3 h resulted in appropriate network formation. Even though functionalization with FAME is not complete, Table 1 shows that networks with high gel contents up to 96% can indeed be obtained from the different macromers. The table also shows that upon functionalization and photo-crosslinking of the TMC containing oligomers the thermal properties do not change significantly, and only the melting temperature in PEG networks decreases. Table 1 Characteristics of PEG-containing hydroxy-terminated oligomers (precursors), FAME-functionalised oligomers (macromers) and the resultant networks after UV photo-crosslinking

PEG 2000 PEG 4000 2000PEG50PTMC50 3-arm 10000PEG50PTMC50

T g, (T m) of precursor (8C)

T g, (T m) of macromer (8C)

T g, (T m) of network (8C)

Gel content (%)

n.o., (54.4) n.o., (53.4) 44.1, (38.6) 44.7 (44.0)

n.o., (43.7) n.o., (53.6) 43.3, (37.9) 43.4 (46.8)

n.o., (37.3) n.o., (47.6) 38.8, (40.5) 42.5 (44.9)

88.6 96.2 94.5 90.7

n.o.: Not observed. Due to the hydrophilic PEG component in the PEG50PTMC50 oligomers, the cross-linked macromers take up significant amounts of water. Table 2 shows the mechanical properties of the networks in the dry state and in the water-swollen state. Table 2 Mechanical properties of PEG/poly(TMC) networks

2000PEG50PTMC50 2000PEG50PTMC50 3-arm 10000PEG50PTMC 3-arm 10000PEG50PTMC


E-modulus (MPa)

Tensile strength (MPa)

Elongation at break (%)

Dry Wet Dry Wet

94 4 66 3

3.0 1.0 14.0 1.5

3 33 475 375

In the wet state, significantly lower values for the E-modulus and tensile strength are found when compared to networks in the dry state. In both cases, the networks prepared from 3-arm 10000PEG50PTMC50 macromers have significantly higher tensile strengths and elongations at break than networks prepared from linear macromers. Fig. 1 shows the water-uptake as a function of temperature. It can be seen that the swelling capacity of 2000PEG50PTMC50 networks is much lower than that of networks prepared from 3-arm 10000PEG50PTMC50 macromers, although the weight percentage of the hydrophilic component is comparable. The higher water uptake of the latter may be due to the different architecture and the higher molecular weight of the PEG segments, which will allow enhanced phase separation in the wet state. In both cases, water-uptake decreases with temperature. Apparently, in these thermosensitive hydrogels, hydrophobic interactions dominate upon increasing the temperature.

Poster Abstracts


Fig. 1. Swelling behaviour in water of PEG/poly(TMC) networks as a function of temperature. Networks were prepared from 2000PEG50PTMC50 (E) and 3-arm 10000PEG50PTMC50 ( ) macromers.


Conclusions Functionalization of PEG and PEG/poly(TMC)-based oligomers with FAME (fumaric acid monoethyl ester) allows the preparation of hydrogel networks by photo-cross-linking. In the presence of a photo-initiator, high gel contents can be obtained by irradiation at 360 nm. These networks, designed to release only nontoxic degradation products, display thermo-sensitive behaviour. References [1] [2] [3] [4] [5]

A.S. Hoffman, Adv. Drug Deliv. Rev. 54 (2002) 3. B. Jeong, S.W. Kim, Y.H. Bae, Adv. Drug Deliv. Rev. 54 (2002) 37. A.S. Sawhney, C.P. Pathak, J.A. Hubbell, Macromolecules 26 (1993) 581. S. Jo, H. Shin, A.K. Shung, J.P. Fisher, A.G. Mikos, Macromolecules 34 (2001) 2839. A. Hassner, V. Alexanian, Tetrahedron Lett. 46 (1978) 4475.


Poster Abstracts

COLLAGEN SCAFFOLDS FOR HEMATOPOIETIC PROGENITOR CELL EXPANSION AND CONTROLLED DIFFERENTIATION B. Siebum1, D.W. Grijpma1, A.A. Poot1, I. Vermes1,2, W. Kruijer1, J. Feijen1 1 Institute for Biomedical Technology, Polymer Chemistry and Biomaterials, Faculty of Science and Technology, University of Twente, P.O. Box 217 Enschede, The Netherlands 2 Laboratory of Medisch Spectrum Twente, P.O. Box 50000 Enschede, The Netherlands Summary Controlled expansion and differentiation of primitive hematopoietic progenitor cells (HPCs) could allow the transfusion of specific blood cells in large quantities in patients treated with high dose chemo- or radiotherapy. In this study, we introduce the development of a novel, tissue engineering inspired approach to HPC expansion. We present the development of porous crosslinked collagen structures that allow proliferation of HPC-like cells and that can be further modified to create an optimal environment for HPCs to expand and differentiate. Introduction In adults, the hematopoietic system is localized in the bone marrow. This system contains a small population of self-renewing cells that are able to differentiate into distinct blood cells. These HPCs are destroyed in people receiving a high dose chemo- or radiotherapy. In these cases, bone marrow (BM) or peripheral blood stem cell (PBSC) transplantation is needed. Expansion and controlled differentiation of the HPCs could reduce the number of BM or PBSC harvest procedures while allowing the transfusion of the required cells in large quantities. The bone marrow contains niches in which HPCs are localized in a specific microenvironment. In these niches, the cells interact with stromal cells, extracellular matrix components and cytokines [1]. In this project, the aim is to mimic these niches in porous collagen structures that provide a three-dimensional surrounding for the HPCs. In later stages of the project, these scaffolds will be surface modified with bioactive molecules found in the bone marrow [2,3]. In this paper, we describe the processing of collagen type I into porous sponges and beads that allow proliferation of HPC-like cells and can be modified with the bioactive molecule heparin. Experimental methods Collagen scaffolds were prepared with three methods. Porous sponges were prepared from insoluble type I collagen from bovine achilles tendon. Collagen (1% w/v) was dispersed after it was swollen overnight in 0.05 M acetic acid solution at 4 8C. The resulting slurry was filtered and air was removed under vacuum. The suspension was cast in polystyrene flasks and frozen at different temperatures (20, 35 and 196 8C). The frozen suspensions were subsequently lyophilised, resulting in porous collagen structures. The sponges were crosslinked using N-(3-dimethylaminopropyl)-NV-ethylcarbodiimide hydrochloride (EDC) and N-hydroxysuccinimide (NHS). In order to obtain an open pore structure, these procedures were carried out in a 95% ethanol solution. After crosslinking the sponges were washed and lyophilised. Porous beads were prepared by electrospraying of a collagen suspension into liquid nitrogen. The collagen suspension was prepared as described above. The frozen beads were collected from the liquid nitrogen and lyophilised, resulting in porous beads. The porous beads were crosslinked using EDC and NHS as described above. Solid collagen beads were prepared by emulsification. Acid-soluble collagen type I from calf skin was dissolved overnight in 0.5 M acetic acid solution at 4 8C. The cooled collagen solution was neutralized and added drop-wise to stirred paraffin oil at 37 8C. The emulsion was left stirring for 3 h in order to allow the collagen to reconstitute as beads. Then, water was added and the beads were removed from the aqueous phase. The beads were washed with a 70% ethanol solution and water before the beads were crosslinked using EDC and NHS as described above. Sieves were used to obtain a narrow range of bead sizes. For preliminary modification experiments, the collagen structures were modified using heparin sodium salt (Bufa Chemie, Castricum, the Netherlands). The crosslinked collagen structures were first equilibrated with 0.05 M MES buffer for 30 min. The carboxylic acid groups were activated by adding EDC and NHS to a 2% (w/v) solution of heparin in 0.05 M MES-buffer (pH 5.5–6.0) at a molar ratio of EDC:NHS:Hep-COOH of 0.4:0.24:1.0. After preactivation for 10 min, 1 g of crosslinked collagen was added to 188.3 ml of EDC/NHS-activated heparin solution. After 2 h of reaction, the heparinized collagen was washed with 0.1 M Na2HPO4 for 2 h, four times for 24 h with 4M NaCl and three times for 24 h with distilled water. The shrinkage temperature (Ts) of (crosslinked) collagen was determined using Differential Scanning Calorimetry (DSC). The onset of the endothermic peak was recorded as the temperature at which the collagen sample undergoes thermal denaturation. This value is used as a measure of the crosslink density. A Hitachi S800 scanning electron microscope (SEM) was used to examine the morphology and pore size of the scaffolds. Cell culture experiments with the KG1a leukaemia cell line were used as a model to evaluate HPC growth on the prepared materials. These cells were seeded at a density of 15 000 and 3750 cells per well in a 96-well roundbottom plate with 200 Al IMDM containing 20% FBS and were cultured for 1 week. Alcian blue staining was used for localisation of the heparin in crosslinked collagen samples. Sections of the samples were incubated in a 3% (v/v) acetic acid solution for 3 min under vacuum and then stained by incubation in a 2% (w/v) solution of Alcian blue in 3% (v/v) acetic acid for 30 min under vacuum. After washing four times with demineralised water for 15 min, the samples were evaluated visually.

Poster Abstracts


Results and discussion We are able to produce highly porous sponges and porous or solid beads using collagen type I. The characteristics of the structures used for the KG1a cell culture are shown in Table 1. Table 1 Collagen scaffold characterization Scaffold type

Size (Am)

T freeze (8C)

Pores (Am)

Ts (8C)

Solid beads Porous sponges Porous sponges Porous sponges Porous beads

500 – – – 870

– 196 35 18 196

72F1 68F2 67F1 67F1 74F1

15F5 43F6 149F4 6F2

These techniques allowed us to tune the properties of the sponges and beads. Examples of the obtained porous sponges can be seen in Figs. 1 and 2. Light microscopic analysis of the cell culture experiments with the KG1a cell line showed that these HPC-like cells proliferate when they are cultured on the collagenous sponges and beads. A representative image of the cells on a bead is shown in Fig. 3. The modification of the beads and sponges with heparin resulted in a lower amount of free amino groups, indicating that heparin was covalently bound (not shown here). Staining with Alcian blue confirmed that we are able to homogeneously modify the scaffolds with heparin when the porous structures are interconnected. This can be seen in Fig. 4, where sample A is a scaffold with poor interconnectivity and C a highly interconnected scaffold.

Fig. 1. Collagen sponge with large average pore size at 100 magnification.

Fig. 2. Collagen sponge with small average pore size at 100 magnification.

Fig. 3. KG1a cells on collagen beads, day 3 (100 magnification).


Poster Abstracts

Fig. 4. Alcian blue stained heparinized scaffold sections: A—small average pore size (Fig. 2), B—not heparinized control scaffold and C—large average pore size (Fig. 1).

Conclusion We are able to produce collagenous scaffold structures that support the proliferation of the hematopoietic cell line KG1a and that can be modified with bioactive molecules like heparin. These structures will now be used for cell culture with HPCs (CD34+) obtained from peripheral blood. The interaction of the heparinized structures with chemo- and cytokines will then be investigated with radiolabeled molecules. Acknowledgements P. Linssen is acknowledged for help with the KG1a cell culture, M. Smithers for the SEM micrographs and STW for financial support. This research is part of the STW project Growing Blood performed in cooperation with: C. Figdor, R. Torensma, T. van Kuppevelt en R. Raymakers from the University of Nijmegen. References [1] S.K. Nilsson et al., Blood 97 (2001) 2293–2299. [2] M.J.B. Wissink et al., Biomaterials 22 (2001) 151–163. [3] J.S. Pieper et al. (2000), Biomaterials 21, pp. 581–593.

Poster Abstracts


III. Delivery of Proteins and Synthetic Macromolecules CHARACTERISATION OF THE TRANSFERRIN RECEPTOR FOR DRUG DELIVERY TO THE BLOOD–BRAIN BARRIER C.C. Visser1, L.H. Voorwinden1, L. van Bloois1, D.J.A. Crommelin2, M. Danhof1, A.G. de Boer1 1 LACDR, Leiden University, Division of Pharmacology, Blood–Brain Barrier Research Group, P.O. Box 9502, 2300 RA Leiden, The Netherlands 2 UIPS, Utrecht University, Department of Pharmaceutics, P.O. Box 80082, 3508 TB Utrecht, The Netherlands Introduction Delivery of drugs to the CNS is restricted by the presence of the blood–brain barrier (BBB). Due to its specific features, such as tight junctions between endothelial cells, a continuous basal membrane, low pinocytotic activity and a lack of fenestrae, only small lipophilic drugs or drugs that are substrates for a specific transporter, can pass the BBB. Our research focuses on (i) the characterisation of the transferrin receptor (TfR) at the BBB and (ii) the use of this receptor for drug targeting to the brain. Specifically, the objective is to deliver drugs tagged with transferrin (Tf) to the BBB. These studies are conducted in an in vitro BBB model, consisting of primary cultured bovine brain capillary endothelial cells (BCEC) and astrocytes conditioned medium [1–3]. Results and discussion The level of TfR expression on BCEC was determined by radioligand binding studies. For endothelial cells cultured in the presence of astrocytic factors a B max of 0.18F0.06 fmol/mg protein and a K d of 3.2F1.0 Ag/ml was found. The addition of saponin, a cell permeabilising agent, increased the B max to 2.3F0.6 fmol/mg protein, while the K d was unaffected. Endocytosis studies at 37 8C revealed that Tf is actively endocytosed by the TfR on BCEC. Furthermore, uptake of Tf was inhibited by phenylarsineoxide, an inhibitor of the clathrin-mediated pathway associated with the TfR. In contrast, indomethacin, an inhibitor of caveolae-mediated endocytosis, did not change Tf uptake. In order to study the uptake of carge coupled to Tf, we have first chosen to couple horseradish-peroxidase (HRP) to Tf. This compound can be easily measured in cells following cell lysis. The results showed that the uptake of Tf-HRP by BBB endothelial cells at 37 8C increased linearly following application of 0.25–10.0 Ag/ml. In addition, binding at 4 8C increased linearly but was about four times less than at 37 8C. Subsequently, the rate of uptake was estimated following incubation with a concentration of 3 Ag Tf-HRP/ml. Equilibrium was attained following 1 h indicating that uptake was a fast process. In addition, binding at 4 8C was again about a factor 4 smaller than uptake at 37 8C. Selective/specific transport of Tf-HRP was investigated in the presence of excess Tf (which competes for transport by the TfR) and bovine– serum–albumin (BSA) that competes for a-selective binding sites. 500-Fold excess of unconjugated Tf decreased the uptake of Tf-HRP at 37 8C by a factor of about 5 while 500-fold excess of BSA did not had any effect upon uptake. This illustrates that the uptake occurred by a selective transport process, particularly via the TfR. Conclusion These results show that the TfR is present on BCEC and that it is actively endocytosing Tf via a clathrin-mediated pathway. In addition, Tfconjugated compounds (Tf-HRP) are selectively taken up via the Tf-R at the BBB. This indicates that targeting drugs to the Tf-R at the BBB is a promising way for drug delivery to the brain. References [1] Gaillard et al., Eur. J. Pharm. Sci. 12 (2001) 215. [2] Gaillard et al., Microvas. Res. 65 (2003) 24. [3] Fan et al., Surgery 128 (2000) 332.


Poster Abstracts

CONTROLLED DELIVERY OF METHOTREXATE FROM CHANNEL-PROTEIN CONTAINING LIPOSOMES M. van Deemter, J. Sˇmisterova´, G. van der Schaaf, W. Meijberg, G. Robillard Biomade Technology Foundation, Nijenborgh 4, 9747 AG Groningen, The Netherlands

Summary The goal of this study is to develop a controlled methotrexate delivery system that releases its content by a target specific stimulus. To accomplish this, both methotrexate and an engineered channel protein (MscL) were incorporated into DOPC/CH/DSPE-PEG liposomes, providing a controlled drug delivery system. Reconstitution and encapsulation methods were optimized in order to ensure optimal channel gating activity and high drug:lipid ratio. At the optimal conditions, a 95% release of MTX could be reached.

Introduction In tumor therapy, many potential drugs cannot be used because they are toxic to several vital organs. For these drugs, encapsulation in liposomes can be a solution. Further improvements in the design of liposomes resulting in specific release of drug at the tumor site will make the utilization of liposomes more feasible. To accomplish this, we are developing a liposomal system containing a pHsensitive proteinacious valve that opens as a response to lowered pH values, such as these reported, e.g., for solid tumors [1]. This proteinacious valve is the mechanosensitive channel of large conductance (MscL) from E.coli. The glycine at the 22nd position of the MscL is mutated to a cysteine, which is used to target either pH-sensitive chemical groups or [2-(trimethylamonium)ethyl] methanethiosulfonate bromide (MTSET) to the protein. Binding of MTSET introduces a positive charge into the hydrophobic channel pore leading to opening of the channel. MscL-G22C was reconstituted in therapeutically optimized liposomes composed of DOPC/ Cholesterol/DSPE-PEG2000 (70:20:10), which was shown to be compatible with the channel protein activity. The encapsulation method for MTX has been optimized.

Experimental methods The preparation of MscL containing liposomes with encapsulated MTX combines the detergent-mediated reconstitution of MscL in liposomes by using BioBeads and the encapsulation of MTX by one of the following methods: (a) lipid film rehydration in a drug-containing solution, (b) reverse phase evaporation and (c) addition of the drug in the presence of detergent during protein reconstitution. For the analysis of drug:lipid ratio, MTX-proteoliposomes were separated from the free drug on a Sephadex G50 column. Total lipid content was determined from a sensitive fluorescence assay, using diphenylhexatriene as the probe (excitation at 355 nm and emission at 440 nm). MTX content was measured by HPLC (detection at 294/356 nm [2]). Drug:lipid ratio was calculated as Ag drug per mg total lipid. Release of MTX after activation with MTSET (pH nonsensitive) was determined after separation of liposomes and free drug on a desalting PD10 column followed by the measurement of total lipid and drug concentrations.

Results and discussion In order to see which MTX encapsulation method gives the highest drug:lipid ratio in combination with high drug release after activation, three methods were compared. The results can be seen in Table 1. All three encapsulation methods tested gave similar drug:lipid ratios (between 5 and 7 Ag/mg) and also similar release (about 95%) upon the activation of the channel. For practical reasons, the method in which the drug was encapsulated during protein reconstitution was used in further experiments.

Table 1 Drug:lipid ratios and drug release after different MTX encapsulation methods Encapsulation method

Drug:lipid (Ag/mg)

Released drug after activation (%)

Reverse phase evaporation During protein reconstitution Lipid film rehydration

6.99 6.39 5.13

95 95 94

Poster Abstracts


Conclusion A proteoliposomal delivery system for MTX has been developed, which can release almost all encapsulated drug upon the activation of the channel protein by chemical labeling. We are currently busy with the development of pH-sensitive proteoliposomes. References [1] G.R. Martin, K.R. Jain, Nonevasive measurement of interstitial pH profiles in normal and neoplastic tissue using fluorescence ratio imaging microscopy, Cancer Res. 54 (1994) 5670–5674. [2] S. Vezmar, U. Bode, U. Jaehde, Monitoring of methotrexate and reduced folates in the cerebrospial fluid of cancer patients, Int. J. Clin. Pharmacol. Ther. 40:582–583.

TARGETED NORFLOXACIN IS ACTIVE IN VIVO AGAINST PERSISTENT MYCOBACTERIUM BOVIS BCG D. Domurado1, A.-M. Balazuc2, M. Lagranderie2, P. Pescher2, P. Chavarot2, E. Roseeuw3, V. Coessens3, S. Stern4, E. Schacht3, G. Marchal2 1 Groupe de Pharmacocine´tique des Prodrogues et Conjugue´s Macromole´culaires (INSERM), CRBA-UMR 5473 CNRS, Faculte´ de Pharmacie, 15 avenue Charles Flahault, BP 14491, 34093 Montpellier cedex 5, France 2 Laboratoire de Re´fe´rence des Mycobacte´ries, Institut Pasteur, 25 rue du Dr Roux, 75724 Paris cedex 15, France 3 Biomaterial and Polymer Research Group, Department of Organic Chemistry, University of Ghent, Krijgslaan 281, S4 bis, 9000 Ghent, Belgium 4 Laboratoire de Radiobiologie Cellulaire, Institut Gustave Roussy, 39 rue Camille Desmoulins, 94805 Villejuif cedex, France

Summary Tuberculosis is difficult to treat because Mycobacterium tuberculosis persistent forms resist the usual antituberculous agents. This results in 2–3 million deaths each year. In order to improve tuberculosis treatment, we established a short-term in vivo model of persistent mycobacteria. Also, we synthetized a macromolecular prodrug targeted toward macrophages and demonstrated its in vivo efficacy against persistent mycobacteria. To our knowledge, this is the first report of a drug active in vivo against persistent mycobacteria.

Introduction Tuberculosis infects 2 billion people worldwide. Even if most of them present no symptoms and will never develop the disease, the eventual development of an overt tuberculosis, up to decades after contamination [1,2], results in 2–3 million deaths each year. The possible late development of tuberculosis results from the induction of Mycobacterium tuberculosis persistence by different conditions, such as nutrient starvation [3] or hypoxia [4]. This means that M. tuberculosis keeps its infectious capacity intact while it no longer multiplies. Persistent bacilli resist antituberculous agents, such as isoniazid, although the same antibiotics rapidly eliminate actively multiplying M. tuberculosis. Consequently, tuberculosis is very difficult to treat: the usual regimen requires four antibiotics for two months, then two antibiotics for the next four months, watching out for the return of persistent bacteria to normal metabolism. In patients, persistent bacilli survive in macrophages after phagocytosis where they are rather well protected from immune response and antibiotic treatment. It is expected that antibiotics targeted toward macrophages improve treatment efficacy by concentrating drugs close to bacteria. The aim of this work was to design a conjugate able to deliver an antibiotic into macrophages phagosomal vacuole in close contact with the intracellular bacteria. We synthesized targeted macromolecular prodrugs composed of a macromolecular carrier, of a targeting device and of an antibiotic active in vitro against M. tuberculosis. The carrier, dextran, bears both mannose, the homing device intended to target the macrophage, and norfloxacin, to be delivered into the phagosomal vacuole [5]. Norfloxacin was linked to the macromolecular carrier through two different peptide arms. M. bovis BCG was used as a mycobacterial model because this bacterium has a physiology closely related to the physiology of M. tuberculosis, while being at the same time much less pathogenic than the latter, and because the immune response of the mice rapidly controlled extracellular infection and left only intracellular bacilli, i.e., the goal of the targeted norfloxacin. This model takes advantage of different pO2 levels in different mouse organs, hence of different persistence statuses of bacteria in these organs, and it therefore allowed us to test the in vivo antibiotic activity of our macromolecular prodrugs against persistent M. bovis BCG in mice. In this paper, we report the measurement of pO2 in liver and spleen of mice. We study the therapeutic activity of our macromolecular prodrugs versus isoniazid and ofloxacin used as controls against well oxygenated and hypoxic, hence persistent, bacilli by counting them in lungs, spleen and liver. These organs are rich in macrophages and possess different oxygenation statuses.


Poster Abstracts

Experimental methods Anti-mycobacterial activities of norfloxacin and of its macromolecular conjugates were tested in vitro on M. bovis (BCG strain 1173P2, Institut Pasteur). Oxygen partial pressure (pO2) was measured in spleen and liver of mice through a Pt electrode implanted in the mouse organ. Measures were carried out every 200 Am (pO2 Histograph Eppendorf). Fifteen to 30 measures were done on a needle path. Specific-pathogen-free C57BL/6 mice (age, 6 weeks), obtained from Iffa-Credo (Saint-Germain sur l’Arbesle, France) were infected by intravenous injection of 106 viable units. Starting from the 6th day postinfection, the mice (5 per group) were injected intraperitoneally twice a day for 5 days with the different test molecules. Phosphate-buffered saline (PBS) was used as negative control treatment, isoniazid (0.5 mg per mouse per day) and ofloxacin (0.25 mg per mouse per day) being used as positive control treatments. The macromolecular conjugates containing norfloxacin were diluted in PBS, and each mouse was injected with the equivalent of 0.25 mg of norfloxacin per day. They were compared to native norfloxacin (0.5 mg per day). M. bovis BCG growth was monitored by counting the number of viable units in the lungs, spleen and liver of mice 2 days after the end of treatment. The tissues were homogenized, suitable dilutions were plated on Middlebrook 7H11 medium, and the number of CFU was counted after 21 days at 37 8C. Results We synthetized norfloxacin conjugates targeted, or not, toward macrophages because they harbor the bacilli. Also, norfloxacin was linked to the C-terminal amino acid of the peptide arm either through a C-terminal amide bond or through an N–C bond, called a bond [6]. In the presence of cathepsin B, the C-terminal peptide bond was hydrolysed. Finally, depending upon the bond linking norfloxacin to the peptidic spacer arm, norfloxacin was released in its native form in the case of the a bond and as an amino acid derivative in the case of the amide bond. The latter had no antibacterial activity in vitro. Accordingly, only conjugates including an a bond were active in vitro against M. bovis BCG. pO2 is not uniform throughout human organism. Since a low pO2 value triggers persistence of M. tuberculosis and of M. bovis BCG, we measured pO2 in liver and spleen, organs which harbor mycobacteria during infection, to determine whether their pO2 influences M. bovis BCG development. Liver was hypoxic (86% pO2 measures are below 5 mm Hg) whereas spleen was much better oxygenated (91% pO2 measures are above 10 mm Hg). M. tuberculosis is known to be less sensitive or unsensitive to antibiotics under hypoxia, i.e., in a persistence state [7]. This observation was made in vitro but was never confirmed directly in vivo even if strongly suspected given the clinical observations. In order to assess the in vivo influence of hypoxia on antibiotic activity, we studied the anti-infectious activities against M. bovis BCG of PBS, of isoniazid, of ofloxacin, of native norfloxacin and of targeted norfloxacin in lungs, spleen and liver, since oxygenation statuses of these organs are different. In lungs, only PBS and native norfloxacin had no effect against M. bovis BCG. Despite its in vitro activity, native norfloxacin was not active in vivo because of its rapid renal filtration. In spleen, results were similar to those obtained in lungs, but ofloxacin was already less active than isoniazid and norfloxacin conjugate. In liver, the only active product was the mannosylated norfloxacin conjugate with an a bond, all other treatments, including isoniazid and ofloxacin, being inactive. It must be noticed that only the conjugate containing both an a bond and mannose was as active in lungs as isoniazid against M. bovis BCG. The other conjugates were all inactive. Discussion Persistence of M. tuberculosis prevents rapid and definitive curing from tuberculosis. Several factors, among them hypoxia, determine persistence of M. tuberculosis and M. bovis. In these conditions, we tried to determine: (i) whether there were in a whole mammal organism macroscopic areas sufficiently hypoxic so that persistent bacilli could be easily studied in vivo, (ii) whether a molecule could produce an in vivo antibacterial activity against bacilli present in these hypoxic zones, bacilli therefore persistent. After its intravenous introduction into the body, M. bovis is taken up by macrophages, which are mainly located in liver and spleen. Therefore, we measured pO2 in liver and spleen. Liver is hypoxic because most blood arrives to liver by portal vein. This blood already irrigated digestive tract, spleen and pancreas, and is therefore mostly deprived of oxygen. On the contrary, spleen, only irrigated by arterial blood, is well oxygenated. On the basis of these results, we established an in vivo model of infection by persistent mycobacteria. As a mycobacterial model, we chose M. bovis BCG because: (i) it is less pathogenic for man than M. tuberculosis, (ii) the murine immune response rapidly leaves only intracellular M. bovis BCG, (iii) it becomes persistent when pO2 decreases, (iv) it possesses the same mechanism of adaptation to hypoxia than M. tuberculosis, and finally, (v) it can stay silent in humans for decades before the eventual development of overt infection as well as M. tuberculosis. When persistent, M. tuberculosis is practically insensitive to usual antituberculous drugs [7]. In order to check if M. bovis BCG was able to reproduce this behavior, efficacies of isoniazid and ofloxacin were studied as a function of the organ, thus as a function of pO2. In lungs, the most oxygenated organ, isoniazid treatment eliminated 95% of bacilli as referred to PBS control. In spleen, which had an intermediate oxygenation status, isoniazid treatment still eliminated 80% of bacilli as refered to PBS control. In liver, which was hypoxic, this same treatment

Poster Abstracts


eliminated only 25% of M. bovis BCG as refered to PBS control. Results were parallel for ofloxacin. The clear correlation existing between sensitivity of M. bovis BCG to antibiotics and pO2 of the host organ demonstrates that persistent M. bovis BCG is resistant to usual antituberculous drugs. Therefore, in vivo infection with M. bovis BCG may be used as a model allowing to test antibiotics efficacy. As yet, no in vivo model was published allowing to test the efficacy of molecules against persistent M. tuberculosis. Among the four conjugates tested, only the conjugate containing both mannose as a homing device and an a bond was active in vivo against M. bovis BCG. Several conclusions can be drawn from these results. First, the absence of antibacterial activity of native norfloxacin in vivo demonstrates the importance and interest of drug targeting. Second, only the breaking of the a bond was able to release native and, thus, active norfloxacin. Third, targeting of the macrophage is necessary: only the mannosylated conjugate was active in vivo. Moreover, our results indicate that targeting allows norfloxacin to reach intraphagosomal bacteria. This is corroborated by in vitro results (not shown here). Fourth, fluoroquinolones are active against persistent bacteria, i.e., not currently dividing, at variance with what is commonly admitted by the literature. Finally, the carrier had no antimicrobial effect by itself. Conclusions (1) Our model of persistent mycobacteria consisting of M. bovis BCG studied in liver of mice allows to study in vivo therapeutic efficacy of drugs against persisting bacilli. (2) From a practical viewpoint, the most important conclusion of this work is that targeted norfloxacin conjugate was active in liver, an organ where isoniazid and ofloxacin were not. This is the first demonstration that it is possible to design a drug able to fight hypoxic, hence persistent, mycobacteria in vivo. (3) From a therapeutic point of view, one expects that the use of targeted norfloxacin could shorten the curative treatment due to its action on bacilli otherwise protected in locations, as the liver in our experiments, acting as a shrine. References [1] G. Marchal, Recently transmitted tuberculosis is more frequent than reactivation of latent infections, Int. J. Tuberc. Lung Dis. 1 (1997) 192. [2] T. Lillebaek et al., Stability of DNA patterns and evidence of Mycobacterium tuberculosis reactivation occurring decades after the initial infection, J. Infect. Dis. 188 (2003) 1032. [3] J.C. Betts et al., Evaluation of a nutrient starvation model of Mycobacterium tuberculosis persistence by gene and protein expression profiling, Mol. Microbiol. 43 (2002) 717. [4] L.G. Wayne, Dynamics of submerged growth of Mycobacterium tuberculosis under aerobic and microaerophilic conditions, Am. Rev. Respir. Dis. 114 (1976) 807. [5] E. Roseeuw et al., Synthesis, degradation and antimicrobial properties of targeted macromolecular prodrugs of norfloxacin, Antimicrob. Agents Chemother. 47 (2003) 3435. [6] M. Nichifor, E. Schacht, Synthesis of peptide derivatives of 5-fluorouracil, Tetrahedron 50 (1994) 3747. [7] L.G. Wayne, C.D. Sohaskey, Nonreplicating persistence of Mycobacterium tuberculosis, Annu. Rev. Microbiol. 55 (2001) 139.


Poster Abstracts

CELL-MEDIATED PROTEIN RELEASE FROM CALCIUM-PHOSPHATE-COATED TITANIUM IMPLANTS Y.M. Liu1,2, K. de Groot2, E.B. Hunziker1 1 ITI Research Institute for Dental and Skeletal Biology, Bern University, Murtenstrasse 35, 3010 Bern, Switzerland 2 Institute for Biomedical Technology (BMTI), University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands Summary Implants used in dental and orthopaedic surgery should become osseointegrated into native bone. In an endeavour to potentiate bone formation, prostheses, coated with a coprecipitated layer of calcium phosphate and the osteogenic agent BMP-2, were implanted at an ectopic site in rats. Five weeks later, the implants were surrounded by a substantial mass of bone. Our findings indicate that the drug was gradually released. We postulate that the release of BMP-2 from coatings was cell-mediated. Introduction The osteoconductivity of metallic implants used in dentistry and orthopaedic surgery can be enhanced by coating their surfaces with a layer of either calcium-phosphate-based or bone-matrix-like material [1,2]. These inorganic layers are of course three-dimensional lattice works, and, when viewed as such, the idea of exploiting them as a carrier scaffolding for the slow delivery of osteoinductive agents to the peri-implant site appears logical. In a previous study, we pursued this idea by coprecipitating the osteogenic growth factor, BMP-2, and calcium phosphate upon the surfaces of titanium-alloy implants. BMP-2 formed an integral part of the three-dimensional inorganic latticework and was not merely adsorbed upon its surface [3,4]. Furthermore, the osteogenicity of BMP-2 thus incorporated was not only retained but potentiated in an in vitro system comprised of cultured osteoprogenitor cells [5]. In the present investigation, we wished to evaluate the kinetics and histomorphometry of osteoinduction by BMP-2 thus incorporated in an in vitro system, namely, at an ectopic site in rats. Materials and methods One experimental and three control groups were set up: titanium-alloy discs coated with a co-precipitated layer of calcium phosphate and BMP2 [1.7 Ag per disc (experimental group)]; naked discs (control); discs coated with a layer of pure calcium phosphate (control); and discs coated with a layer of calcium phosphate upon which BMP-2 was superficially adsorbed [0.98 Ag per disc (control)]. Discs (n=6 per group) were implanted subcutaneously in rats and retrieved at 7-day intervals over a period of 5 weeks for kinetic and histomorphometric analyses. Results and discussion In the experimental group, osteogenic activity was first observed 2 weeks after implantation and thereafter continued unabated until the end of the monitoring period. The net daily rates of bone formation per disc were 0.83 and 0.52 mm3 at 2 and 5 weeks, respectively. The total volumes of bone formed per disc at these junctures were 5.8 and 10.3 mm3, respectively. Bone, which was formed by an intramembranous mechanism, was deposited at distances of up to 340 Am from the implants. Coatings were degraded gradually, initially by foreign-body giant cells and then, after the first week, also by osteoclasts. Forty percent of the coating material (and thus presumably of the incorporated BMP-2) remained at the end of the monitoring period, indicating that its osteoinductive potential was not exhausted. At this 5-week juncture, no bone tissue was associated with any of the control implants. We postulate that during the initial postoperative phase, foreign-body giant cells, in being drawn to the site of implantation as part of the inflammatory response mounted against foreign material, and in embarking on their destructive task by attacking the coating, may actually promote osteogenic activity by liberating BMP-2 from the inorganic matrix as they degrade it. They thus assume the role played by osteoclasts in physiological bone formation and remodelling signalling pathways [6–8] and in our model after the first week of implantation. Hence, a potentially destructive agency in the system is turned into a constructive one, with the inflammatory reaction being enlisted in the osteogenic response. It could of course be argued that BMP-2 is released spontaneously from the coatings. Indeed, a small amount of the growth factor is known to be liberated in this manner [3], but the concentration probably lies below the osteoinductive threshold. This surmise is supported by our findings relating to preformed calcium phosphate layers bearing superficially adsorbed BMP-2. These coatings induced only an abortive osteogenic response during the first postoperative week, and the islands of bone formed were so small and so rare as to be quantitatively nonmeasurable; by the second week, this osseous tissue had been completely resorbed. Conclusion The biomimetic coprecipitation of BMP-2 and calcium phosphate yields a coating which is highly biocompatible, osteoconductive and osteoinductive. Furthermore, BMP-2 is released not only at a level that suffices to induce osteogenesis, but gradually, perhaps in a cellmediated manner, such that osteogenic activity is sustained for a considerable period of time. In future experiments, this principle will be optimized for application at orthotopic sites.

Poster Abstracts


Acknowledgements The authors would like to thank Wyeth for providing them with human recombinant BMP-2, Dr. P. Layrolle for his advice and scientific input, and C. de Valk, S. Nqssli, V. Gaschen and F. Seifriz for their technical support. References [1] K.E. Healy, A. Rezania, R.A. Stile, Designing biomaterials to direct biological responses, Ann. N.Y. Acad. Sci. 875 (1999) 24–35. [2] W.R. Lacefield, Current status of ceramic coatings for dental implants, Implant Dent. 7(4) (1998) 315–322. [3] Y. Liu, E.B. Hunziker, N.X. Randall, K. de Groot, P. Layrolle, Proteins incorporated into biomimetically prepared calcium phosphate coatings modulate their mechanical strength and dissolution rate, Biomaterials 24(1) (2003) 65–70. [4] Y. Liu, P. Layrolle, J. de Bruijn, C.A. van Blitterswijk, K. de Groot, Biomimetic coprecipitation of calcium phosphate and bovine serum albumin on titanium alloy, J. Biomed. Mater. Res. 57(3) (2001) 327–335. [5] Y. Liu, E.B. Hunziker, P. Layrolle, J. de Bruijn, K. de Groot, BMP-2 incorporated into biomimetic coatings retains its biological activity, Tissue Eng. Volum 10 (1/2) (2003) 101–108, 2004. [6] E. Filvaroff, R. Derynck, Bone remodelling: a signalling system for osteoclast regulation, Curr. Biol. 8(19) (1998) R679–R682. [7] K. Mostov, Z. Werb, Journey across the osteoclast, Science 276(5310) (1997) 219–220. [8] T. Katagiri, N. Takahashi, Regulatory mechanisms of osteoblast and osteoclast differentiation, Oral Dis. 8(3) (2002) 147–159.

RESPONSES OF CULTURED MACROPHAGES TO MICROSPHERES A. Luzardo-Alvarez, H.P. Merkle, B. Gander Institute of Pharmaceutical Sciences. Swiss Federal Institute of Technology Zurich (ETH), Winterthurerstrasse 190, 8057 Zurich, Switzerland Summary ¨ ) to poly(lactide-co-glycolide) (PLGA) and chitosan microspheres This study investigated responses of cultured RAW 264.7 macrophages (MO (MS), specifically the uptake of the MS and the subsequent production of reactive oxygen intermediates (ROI), nitric oxide (NO), TNF-a´ and COX-2. All MS types were efficiently phagocytosed and induced significant ROI production, whereas no significant production of the markers of inflammatory response (NO, TNF-a´ and COX-2) was elicited by the microspheres. Introduction ¨ ) and dendritic cells (DC) in the immune system, i.e., in the defence against infection, or The multipotent functions of macrophages (MO suppression of neoplastic growth, sustain the interest for targeting drugs, antigens and genes to these cells. For this purpose, biodegradable ¨ and PLGA or chitosan MS have proven to be of high interest [1,2]. For efficient intracellular delivery, MS must be ingested efficiently by MO DC. This, in turn, mainly depends on particle size and surface characteristics (hydrophobicity and charge) [3]. The purpose of this study was to ¨ is correlated to the production of ROI, NO, TNF-a´ and COX-2. analyze whether the uptake of MS by MO Experimental methods Unloaded microspheres (1–10 Am in diameter) were prepared by spray-drying (Mini Spray-Dryer 191, Buchi, CH-Flawil) from end groupuncapped 14 and 35 kDa PLGA50:50 (Resomer RG502H, RG503H, Boehringer Ingelheim, D-Ingelheim) and chitosan (6 mPa viscosity grade, deacetylation degree of 83%, Pronova, N-Drammen). Chitosan MS were cross-linked with 25% formaldehyde during spray-drying; residual glutaraldehyde was subsequently removed by sodium metabisulphite. For comparison, poly(styrene) latex beads (4.5F0.2 Am, Polysciences, Basel) were included. RAW 264.7 murine cells (5105 cells per well) were cultured under standard conditions and coincubated with microspheres (25 Ag) during 24 h, in the presence or absence of the stimulating agent LPS (5 Ag/ml). Uptake of microspheres was assessed by microscopy. Production of NO was measured by the Griess method, H2O2 by oxidation of 2V,7V-dichlorofluoresceindiacetate, DCFH-DA, and TNF-a by ELISA (Kit, R&D Systems, Abingdon). Cell viability upon co-incubation with MS and LPS was also ascertained. Further, production of COX-2 was assayed by Western blot analysis. Results and discussion All microsphere types were efficiently phagocytosed by the RAW264.7 cells, with RG502H yielding the highest levels. No particle uptake was measured upon co-incubation at 4 8C. Consistent with observations in the literature [4], this particle uptake was reflected by increased ROI-production, which resembled that observed when the cells were incubated with LPS (Fig. 1).


Poster Abstracts

Fig. 1. Effect of MS with RAW cells after 4 h of co-incubation using DCFH-DA as fluorescent probe (positive controls: LPS and Aeromonas salmonicida bacterin suspension).

Conversely, NO-production was not significantly induced upon co-incubation with MS alone during 24 h (Fig. 2), while LPS elicited strong NO-production. Upon co-incubation with MS and LPS, the NO-production appeared to be slightly though significantly ( pb0.05) reduced as compared to the results with LPS alone. The reason for this is unclear and might be related, for example, to LPS adsorption on the MS, or a modulating effect of the MS at the cellular level.

Fig. 2. Influence of co-incubation of MA and MS (25 Ag) in the presence or absence of 5 Ag/ml of LPS (n=3FS.D.) on the NOproduction of the cells. Significant differences ( pb0.05) were found between the activation obtained with LPS alone and co-incubation of MS and LPS. When the incubated MA were assessed for TNF-a´ production (Fig. 3), LPS again stimulated very well the cells, whereas none of the MStypes alone did so. For TNF-a´ production, the MS did not alter the stimulating activity of LPS. Western blot analysis performed on whole cell lysates showed no significant effects on COX-2 protein levels between basal concentrations and concentrations observed after co-incubation with MS during 24 h. Finally, cell viability was not compromised after coincubation with MS.

Conclusion Biodegradable MS composed of PLGA or chitosan were efficiently ingested by murine macrophages and subsequently induced an increase ROI production. Conversely, the MS alone did not initiate an inflammatory cellular response in terms of NO, TNF-a´ and COX-2 production. The present results suggest the possibility of considering MS as suitable delivery system for bioactive

Poster Abstracts


¨ and MS (25 Ag) in the presence or absence of 5 Ag/ml of LPS (n=3FS.D.) on the TNF-a production of Fig. 3. Influence of co-incubation of MO the cells. No significant differences ( pb0.05) were found between the activation obtained with LPS alone and co-incubation of MS and LPS.

compounds, possibly even in inflammatory stages. Nonetheless, translation from these in vitro data to in vivo events requires more investigation. Currently, we plan to compare the responses observed in immortal RAW cells with those in primary murine cells.

Acknowledgements We would like to acknowledge Dictagene (Epalinges, Switzerland) and KTI (Bern, Switzerland) for financial support.

References [1] P. Johansen, Y. Men, H.P. Merkle, B. Gander, Revisiting PLA/PLGA microspheres: an analysis of their potential in parenteral vaccination, Eur. J. Pharm. Biopharm. 50 (2002) 129–146. [2] V. Dodane, V.D. Vilivalam, Pharmaceutical Applications of Chitosan, Pharm. Sci. Technol. Today 1 (1998) 246–253. [3] Y. Tabata, Y. Ikada, Phagocytosis of polymeric microspheres. In: Szycher, M. (Ed.), High performance Biomaterials, Technomic Publishing, Lancaster (1991) 321–646. [4] S. Prior, B. Gander, N. Blarer, H.P. Merkle, M.L. Subira´, J.M. Irache, C. Gamazo, In vitro phagocytosis and monocyte activation with poly(lactide) and poly(lactide-co-glycolide) microspheres, Eur. J. Pharm. Sci. 15 (2002) 197–207.


Poster Abstracts

RENAL TARGETING OF CAPTOPRIL USING SUBCUTANEOUS ADMINISTRATION OF CAPTOPRIL–LYSOZYME CONJUGATE J. Prakash1, R.J. Kok1, A. van Loenen-Weemaes1, M. Haas2, J.H. Proost1, D.K.F. Meijer1, F. Moolenaar1 1 Groningen University Institute for Drug Exploration (GUIDE), Department of Pharmacokinetics and Drug Delivery, University of Groningen, Antonius Deusinglaan 1, 9713 AV, Groningen, The Netherlands 2 BioMaDe Technology Foundation, Nijenborgh 4, 9747 AG, Groningen, The Netherlands Summary In the present study, the renal handling was investigated of a renal selective drug carrier, lysozyme, and its conjugate, captopril– lysozyme, after subcutaneous or intravenous administration in rats. It was found, that both lysozyme and captopril–lysozyme were absorbed intact but slowly from the site of subcutaneous injection. In addition, subcutaneous administration of captopril–lysozyme led to a prolonged residence time in kidneys compared to intravenous administration. Introduction Low molecular weight proteins such as lysozyme are freely filtered through glomeruli in kidneys and reabsorbed in the proximal tubular cells by receptor-mediated endocytosis and subsequently catabolized intralysosomally. Therefore, we used lysozyme as a carrier to deliver the drugs specifically to kidneys (Fig. 1).

Fig. 1. Schematic presentation of renal uptake of drug–lysozyme conjugate. Recently, an angiotensin converting enzyme inhibitor, captopril, was delivered selectively to the kidneys by conjugating with lysozyme. Selective renal accumulation and ACE inhibition after administering captopril–lysozyme conjugate intravenously was found [1,2]. Since the inherent properties of this route of administration make it unfeasible for long-term use, subcutaneous administration is proposed here as an alternative which has also been used for other Macromolecules such as insulin and heparins. Therefore, in the present study, the renal handling of the renal selective drug carrier, lysozyme, and its conjugate, captopril–lysozyme, was investigated after subcutaneous or intravenous administration in rats. Experimental methods The influences of route of administration, dose and adriamycin-induced proteinuria on the renal uptake of lysozyme were determined by gamma-scintigraphy. The pharmacokinetic profile and renal handling of captopril–lysozyme was studied by measuring the captopril concentrations using HPLC [3] in plasma and kidneys at 1, 2, 4, 6, 12, and 24 h and in urine at various time points after single subcutaneous or intravenous injection of captopril–lysozyme. Plasma and renal ACE activities were measured using an enzyme assay in order to determine the release of free drug.

Poster Abstracts


Results and discussions After subcutaneous administration, radiolabeled lysozyme was found to be absorbed completely and intact from the site of injection within 24 h. 88F2% of the subcutaneously administered dose of radiolabeled lysozyme was accumulated in the kidneys within 24 h. In addition, the renal accumulation was decreased dose-dependently to 75F5% and 63F4% upon co-administration of 100 and 500 mg kg1 of unlabeled lysozyme, respectively and was also lowered by adriamycin-induced proteinuria to 59F8%. The dose-dependent and proteinuria-dependent reductions might be due to the saturation of receptors and competition between lysozyme and plasma proteins, respectively. However, these reductions in renal uptake after subcutaneous administration were significantly lower than after intravenous administration. Moreover, after subcutaneous administration of captopril–lysozyme conjugate, captopril concentrations in plasma and kidneys showed a gradual absorption of the conjugate from the injection site and also a higher renal accumulation than after intravenous administration. Renal ACE activity was gradually decreased until 6 h and retrieved to normal at 24 h after subcutaneous administration but not changed at any time-point in plasma which demonstrates that free captopril was released only after tubular uptake but not in plasma. Conclusion The present study concludes that the subcutaneous administration of captopril–lysozyme can be a promising dosage regimen for renal targeting, since good stability after subcutaneous injection was found as well as a sustained delivery and a prolonged residence time in the kidneys. References [1] R.J. Kok, F. Grijpstra, R.B. Walthuis, F. Moolenaar, D. De Zeeuw, D.K.F. Meijer, Specific delivery of captopril to the kidney with the prodrug captopril–lysozyme, J. Pharmacol. Exp. Ther., 288 (1999) 281–285. [2] R.J. Kok, R.F.G. Haverdings, F. Grijpstra, J. Koiter, F. Moolenaar, D. De Zeeuw, D.K.F. Meijer, Targeting of captopril to the kidney reduces renal ACE activity without affecting systemic blood pressure, J. Pharmacol. Exp. Ther., 301 (2002) 1139–1143. [3] R.J. Kok, J. Visser, F. Moolenaar, D. De Zeeuw, D.K.F. Meijer, Bioanalysis of captopril: two sensitive high-performance liquid chromatographic methods with pre- or postcolumn fluorescent labeling, J. Chromatogr., B, Biomed. Appl., 693 (1997) 181–189.


Poster Abstracts

FLUORESCENCE ENERGY TRANSFER TO STUDY THE DEGRADATION OF DOUBLE-LABELED OLIGONUCLEOTIDES K. Remaut, B. Lucas, K. Braeckmans, N. Sanders, S. De Smedt, J. Demeester Laboratory of General Biochemistry and Physical Pharmacy, Ghent University, Harelbekestraat 72, 9000 Ghent, Belgium Summary The degradation of double-labeled oligonucleotides (ONs) was followed by the disappearance of Fluorescence Energy Transfer (FRET) between the rhodamine green and Cy5 fluorophores attached to the 3V and 5V end of the ONs. The fluorescence intensities after rhodamine green excitation were recorded using the ultra-sensitive avalanche photodiode detectors of a dual-color FCS setup. The ratio of the Cy5 to rhodamine green fluorescence showed to give accurate information on the integrity of the ONs, both in buffer and in living cells. Introduction Delivery of ONs for antisense gene therapy is a very promising approach to cure genetic diseases. For the therapy to succeed, the ONs have to be delivered to the cytoplasm or nucleus, and have to stay intact at the site of action. Regrettable, the intracellular fate of antisense ONs remains rather unknown, also because of the limited methods that are applicable in living cells. In this study we present the use of Fluorescence Correlation Spectroscopy (FCS) to follow the degradation of double-labeled ONs, giving FRET between the two labels. The ratio of the acceptor to donor fluorescence after donor excitation (R/G ratio) could be easily used to follow the degradation of these double-labeled ONs, both in buffer and in living cells. Experimental methods Oligonucleotides ONs bearing a rhodamine green label on the 3Vend and a Cy5 label on the 5Vend were purchased from Eurogentec (Seraing, Belgium). The excitation and emission wavelengths of rhodamine green are respectively 488 nm and 532 nm and those of Cy5 are 647 and 670 nm. The ONs used were a 20mer phosphorothioate, 20mer phosphorodiester and 40mer phosphorodiester ON. The FRET efficiencies determined with fluorescence emission scans were 82%, 44% and 96%, respectively. FRET-FCS measurements Dual-color FCS experiments were performed on a setup installed on a MRC1024 Bio-Rad CLSM as described elsewhere [1]. The R/G ratios of the ONs solutions were determined before and after addition of DNaseI and DNaseII to the ONs solution. For the measurements in cytosolic cell extract, 2 Al of the ONs solutions was incubated with an equal volume of cytosolic cell extract at 37 8C. Subsequently, they were diluted to the appropriate concentration for the FCS measurements. All ON concentrations ranged between 5 and 10 nM and the red and green fluorescence intensities were recorded with laser excitation set to 488 nm. For intracellular measurements, Vero cells were injected with intact ONs in the cytoplasm of the cells and the fluorescence intensities in the nucleus were followed. Therefore, the detection volume of the FCS instrument was positioned in the nucleus of the cell and the green and red fluorescence intensity was recorded during 2 s with laser excitation set to 488 nm. From these values, the R/G ratio was calculated. Results and discussion FRET-FCS on buffer solutions The ratio of the red to the green fluorescence (R/G ratio) as measured by both detectors of a dual-color FCS setup before and after addition of different enzymes to the ONs solutions is depicted in Fig. 1. Laser excitation was set to 488 nm. For all intact ONs, FRET occurred, as can be seen by the high R/G ratio (black bars). From this initial R/G ratios, a FRET efficiency order of ON40merNPS20merNON20mer occurs, which is in agreement with the FRET efficiency order as determined from the fluorescence emission scans. Upon degradation, the elimination of FRET results in an increase in the rhodamine green fluorescence and a decrease in the Cy5 fluorescence, which causes the calculated R/G ratio to decrease. From the R/G ratios, it can be seen that the 40mer ON degrades in the presence of DNaseI, DNaseII and cytosolic cell extract, the 20mer ON degrades in the presence of DNaseII and cytosolic cell extract and the 20mer phosphorothioate ON does not degrade in the circumstances studied. Visualisation of the ONs degradation with polyacrylamide gelelectrophoresis could confirm the conclusions made from the R/G ratios as a drop in the R/G ratio corresponds to the disappearance of the ONs on the gel (see graph insert).

Poster Abstracts


Fig. 2 shows the time-dependent degradation of the ON40mer after addition of DNaseI. When a lower enzyme concentration is used, the R/G ratio decreases more slowly. Also the R/G ratio of the ON20mer and PS20mer is shown. The R/G ratio remains around its initial value, as no degradation occurs upon addition of DNaseI. This demonstrates the potential of FRET-FCS to study the kinetics of enzymatic reactions that result in the splicing or coupling of an acceptor and donor fluorophore.

Fig. 1. Ratio of the red (~Cy5) to the green (~rhodamine green) emission as obtained from the fluorescence intensity fluctuations registered by the FCS setup. Laser excitation was 488 nm. Measurements were performed in buffer (black bars), after addition of 2.5 units DNaseI (gray bars), 20 units DNaseII (light gray bars) or an equal volume of cytosolic cell extract (white bars) to 200 Al of the ONs solutions. Values are the mean of four independent measurements. Graph Insert represents the polyacrylamide gelelectrophoresis on ON40mer (A), ON20mer (B) and PS20mer (C). The wells contain 100 ng ON (lane 1), 100 ng ON+DNaseI (lane 2), 100 ng ON+DNaseII (lane 3) or 100 ng ON+an equal volume of cytosolic cell extract (lane 4).

Fig. 2. Time-dependent degradation of the ON40mer upon addition of 0.1 (triangle), 0.02 (diamond) and 0.005 (cross) units DNaseI. Also, the R/G ratios of the ON20mer (circle) and PS20mer (square) upon addition of 0.1 units DNaseI are shown. To normalize the R/G ratios, the R/G ratio of the intact ONs was considered to be 1.

FRET-FCS on living cells The potential of FRET-FCS to be applied on living cells was evaluated by monitoring the green and red fluorescence intensities in the nucleus of a cell microinjected with the intact ONs, with laser excitation set to 488 nm. From these fluorescence intensities, the R/G ratio was calculated. Since only intact ONs display FRET, only the intact ONs can contribute to the measured red fluorescence. This in contrast to the green fluorescence, which can come from both intact and degraded ONs. For all ONs, nuclear accumulation was observed from the increase in the green and/or red fluorescence intensities measured in the nucleus. For the PS20mer, this increase was observed both in the green and red detector (Fig. 3A). This demonstrates that the PS20mer ONs that enter the nucleus are intact. Also, the calculated R/G ratios do not decrease, showing no degradation occurred during the measured time period (Fig. 3D, square). The decrease in the nuclear fluorescence intensities, 20 min after injection, probably results from a redistribution of the PS20mer in the cell. Also for the ON20mer, the increase in the nuclear fluorescence intensities can be observed in both the green and the red detector (Fig. 3B). Again, this demonstrates that intact ONs are entering the nucleus. However, a decrease in the R/G ratio is observed starting from 10 min after the injection (Fig. 3D, circle). This demonstrates that after this time, both intact and degraded ONs are entering the nucleus. While the intact ONs still contribute to the increase in the green and red fluorescence intensity, the degraded ONs only contribute to the increase in the green fluorescence and cause the R/G ratio to decrease. For the ON40mer, the increase in the nuclear fluorescence intensities is only observed from the green detector (Fig. 3C). Apparently, even within minutes after the injection, the ONs entering the nucleus are already degraded. This can also be seen by the rapid decrease of the R/G ratio (Fig. 3D, triangle). It should be noted that the nuclear red fluorescence intensity remains constant, showing that the intact ONs that reached the nucleus in time, do not degrade in the nucleus. About 25 min after the injection, the red nuclear fluorescence start to decrease, which can come from a redistribution of the ON40mer in the cell or from the presence of some nuclease activity in the nucleus.


Poster Abstracts

Fig. 3. Fluorescence intensities measured by the green (square) and red (circle) detector of the dual-color FCS setup in the nucleus of Vero cells injected with (A) PS20mer, (B) ON20mer and (C) ON40mer. Laser excitation was set at 488 nm. (D) Normalized R/G ratios for the PS20mer (square), ON20mer (circle) and ON40mer (triangle). Values are the mean of three independent injections.

Conclusion We showed that monitoring the fluorescence intensities with both detectors of a dual-color FCS setup can be used to follow the degradation of single-stranded, double-labeled ONs, giving FRET between the two labels, both in buffer and in living cells. Upon donor excitation, the donor and acceptor fluorescence gives information on the amount of ONs, where the ratio of the acceptor to the donor fluorescence provides us with information on the integrity of the ONs. Due to the short analysis time (~s), low sample concentration (~nM) and low sample volume (~Al) needed, FRET-FCS is an ideal tool for screening of enzyme activities. Also, on living cells, the accurate detection of the intracellular fluorescence values can be used to give additional insight in many interactions currently studied with FRET-imaging. Acknowledgements K. Remaut is a doctoral fellow of FWO-Flanders. The support of FWO-Flanders is acknowledged with gratitude. The Ghent University (UG-BOF) and FWO-Flanders (G.0310.02) supported this project through instrumentation credits and financial support. References [1] B. Lucas, E. Van Rompaey, S.C. De Smedt, J. Demeester, Macromolecules, 35 (2002) 8152–8160.

Poster Abstracts


ADSORPTION/RELEASE BEHAVIOUR OF OLIGONUCLEOTIDES ON POLYMERIC CORE-SHELL MICROSPHERES L. Tondelli1, M. Ballestri1, A. Cazzato1, F. Federici1, L. Magnani1, K. Sparnacci2, M. Laus2 1 Istituto ISOF, Consiglio Nazionale delle Ricerche, Bologna, Italy 2 Dip. di Scienze e Tecnologie Avanzate, Universita` degli Studi del Piemonte Orientale, Italy Summary Dispersion polymerisation is a process, which generates latex microspheres in the 0.5–20 micron diameter range [1,2]. The particle surface is determined by the choice of the steric stabilizer, which can supply functional sites [3]. Dispersion polymerisation of methylmethacrylate in the presence of commercial Eudragit E100 leads to the formation of core-shell microspheres [4] able to reversibly bind oligonucleotides on the shell. Introduction Single-stranded antisense oligonucleotides (ODNs) are able to specifically bind to complementary segments of mRNAs and inhibit the related protein synthesis. Accordingly, they are widely used as diagnostic agents, as research tools for inhibiting gene expression and are now being developed as therapeutic agents. However, phosphodiester linkages are susceptible to degradation by serum and intracellular nucleases, thus resulting in a short ODN half-life in biological fluids. In addition, their polyanionic nature restrains their cellular uptake through lipophylic cell membranes. To overcome these problems, a large variety of synthetic polymers have been used or proposed as carriers for antisense ODNs [5]. To this reference, we here describe the adsorption/release behaviour of oligonucleotides (ODNs) onto poly(methyl methacrylate) core-shell microspheres, prepared by dispersion polymerisation. An appropriate selection of the experimental parameters and in particular of the initiator and stabilizer amount and of the medium solvency power allows the preparation of monodisperse samples [4] (Fig. 1).

Fig. 1. SEM micrograph of a representative sample of core-shell microspheres (2b).

The microspheres present a core-shell structure in which the shell is constituted by the commercial steric stabilizer Eudragit E100 (Fig. 2) which enables amino groups to interact with oligonucleotides (ODN) and DNA at physiological pH.

Fig. 2. Chemical structure of Eudragit E100 statistical copolymer.

Results and discussion Core-shell microspheres were prepared as described elsewhere [4]. Three representative samples were studied for their ability to bind and release ODNs (Table 1). The ability of the microsphere to bind ODNs was measured in the presence of 20 mM sodium phosphate buffer solutions (pH 7.4) at different ODN concentrations. The amount of adsorbed ODN increases gradually as the ODN concentration increases and is pH dependent. In addition, the adsorbed ODN can be recovered upon incubation in the presence of high salt concentration (1.0 M NaCl),


Poster Abstracts

confirming the ion pair formation between negatively charged ODN phosphate groups and positively charged amino groups present on the microsphere surface at physiological pH. Table 1 Physicochemical characteristics of core-shell microspheres Sample

Diameter (Am)

Surface charge (Amol/m2)

Surface charge (Amol/g)

2b 2c 2d

0.79F0.09 1.12F0.10 0.99F0.16

1.66 1.95 4.29

12.6 10.5 26.0

Conclusion Core-shell microspheres can be obtained by dispersion polymerisation reaction. The presence of amino groups on the surface allows specific ODN adsorption and suggests the use of these microspheres also for DNA delivery. In addition, the hydrophilic shell deriving from the Eudragit E100 could influence their in vivo distribution.

Acknowledgements This work is partially supported by Italian National Institute of Health (ISS)—AIDS National Research Project (grants 45D/1.02 and 40D.48).

References [1] [2] [3] [4] [5]

K.E.J. Barrett, Dispersion polymerization in organic media, Wiley, London, 1975. C.K. Ober, Makromol. Chem., Macromol. Symp., 35/36 (1990) 87. A.J. Paine, Y. Deslandes, P. Gerroir, B. Henrissat, J. Colloid Interface Sci. 138 (1990) 170. K. Sparnacci , M. Laus, L. Tondelli, L. Magnani, C. Bernardi, Makromol. Chem. Phys. 203 (2002) 1364. G. Lambert, E. Fattal, P. Couvreur, Adv. Drug Deliv. Rev. 47 (2001) 99.

Poster Abstracts


CORRELATION OF IN VITRO–IN VIVO RELEASE RATES FOR SUSTAINED RELEASE NEVIRAPINE IMPLANTS IN RATS J. Chen, K. Walters, P. Ashton Control Delivery Systems, Inc., Watertown, MA 02472, USA Summary Sustained release nevirapine implants have been developed and are designed for the prevention of maternal transmission of human immunodeficiency virus (HIV). This study evaluated the in vitro–in vivo release rate correlation using a rat model. Introduction Nevirapine (NVP), a dipyridodiaqepinone, was the first drug of the non-nucleoside reverse transcriptase inhibitor class to be approved for treating HIV infection in humans [1–3]. NVP has been reported to prevent mother to neonate transmission of HIV and is markedly more effective than standard therapy with AZT in the third world [4]. A sustained NVP-delivery system (6 months) has been developed and is designed for the prevention of maternal transmission via subcutaneous implantation. The purpose of this study was to evaluate the in vitro–in vivo release rate correlation using a rat model. Experimental methods Nevirapine (NVP) was mixed with 5% polyvinyl alcohol (PVA) solution and granulated. The granulation was then compressed into rod shaped NVP pellets 2.0 mm or 4.5 mm in diameter. The pellets were dip coated in 5% PVA solution, air-dried and inserted into precut silicone tubes. Sterilization was by gamma-irradiation.

In vitro release testing was conducted using 0.1 M phosphate buffer (pH 7.4) at 37 8C as the release medium. The amount of NVP released was determined by HPLC. The sterilized implants (one 4.5 mm or six 2.0 mm implants per rat) were implanted subcutaneously in female Sprague–Dawley rats. Blood samples were taken periodically and the plasma concentration of NVP was determined. Results and discussion The in vitro release profiles of the implants are displayed in Figs. 1 and 3. Data presented show the amount of drug released per implant.

Fig. 1. NVP release profile (4.5 mm implants).

Fig. 2. NVP plasma concentration in rats with one 4.5 mm implant.


Poster Abstracts

For the 4.5-mm implants, the release rate was faster over the first 4 weeks and then stabilized at 169 Ag/day for the next 7 weeks. A linear release profile was obtained for the 2.0 mm implants that delivered at 52.9 Ag of NVP per day over the entire test period (14 weeks). Gammairradiation (25 kGy) had no effect on the release rate. Based on the in vitro release rate (k r), the body weight of the rats and known NVP PK-data (data from Boehringer Ingelheim: V d, 984 ml/kg and k el, 0.629 h1,) in rats and assuming the PK follows a one-compartment model; the NVP plasma concentration (C) in rats at the steady state can be calculated using the following equation: C ¼ kr ðkel Vd Þ Therefore, a NVP plasma concentration of 38 or 70 ng/ml was expected for rats with one 4.5 mm or six 2.0 mm implants, respectively. Steady-state plasma concentrations of NVP following subcutaneous implantation were 35~45 ng/ml (for rats with one 4.5 mm implant) and 60~80 ng/ml (for rats with six 2.0 mm implants). The results are shown in Figs. 2 and 4; the calculated NVP-concentration is given by the red dotted line.

Fig. 3. NVP release profile (2.0 mm implants).

Fig. 4. NVP plasma concentration in rat with six 2.0 mm implants.

Conclusions Sustained NVP delivery systems with different release rates were developed. The release rates were determined in vitro in buffer and in vivo in rats. The results indicated that the correlation between in vitro and in vivo release rates was excellent. References [1] [2] [3] [4]

V.J. Merluzzi, K.D. Hargrave et al., Science 250 (1990) 1411–1413. D.D. Richman, A.S. Rosenthal et al., Antimicrob. Agents Chemother 35 (1991) 305–308. R.L. Murphy, J. Montaner, Expert Opin. Investig. Drugs 5 (1996) 1183–1199. L.A. Guay, P. Musoke, T. Fleming et al., Lancet 354 (1999) 795–802.

Poster Abstracts


MORPHINE PHARMACOKINETICS FOLLOWING INTRA-ARTICULAR ADMINISTRATION OF A NOVEL SUSTAINED RELEASE OPIOID (CDS-PM-101) FOR THE RELIEF OF POST-OPERATIVE ORTHOPAEDIC PAIN J. Chen, T. Cynkowski, H. Guo, K. Qin, D. Cabral-Lilly, K. Walters, P. Ashton Control Delivery Systems, Inc., Watertown, MA 02472, USA Summary A novel compound (CDS-PM-101) has been developed using CDS CODRUGk technology to provide extended morphine release. CDS-PM101 has very low aqueous solubility. When formulated and administered intra-articularly, the drug dissolves slowly. Once in solution, however, it rapidly hydrolyzes to release morphine within the joint. Introduction Intra-articular (IA) morphine is commonly administered following orthopaedic surgery to provide localized pain relief. Its duration of action, however, is limited due to rapid clearance from the joint. A novel compound (CDS-PM-101) has been developed using CDS CODRUGk technology (Fig. 1) to provide extended morphine release. CDS-PM-101 consists of morphine covalently linked to diclofenac via a labile ester bond. The release of morphine from CDS-PM-101 was evaluated in vitro. In addition, in vivo toxicokinetics was determined after IA dosing in dogs.

Fig. 1. Schematic of drug delivery using CDS CODRUG technology.

Experimental methods CDS-PM-101 was synthesized as both a free base and salt form (see Fig. 2 for salt form). The half-life for hydrolysis to the parent compound was determined by incubating the drug substance in buffer with or without serum, and measuring the amount of CDS-PM-101 remaining over time. The drug substance was then formulated as pellets (0.92 mm, 1.6 mg) containing 0.87 mg CDS-PM-101 (FW: 680D, which hydrolyzes to ~0.4 mg morphine). The pellets soften to a gel consistency within 10 min after exposure to biological fluid. In vitro release of morphine from the pellets was determined in simulated synovial fluid.

Fig. 2. Structure of morphine diclofenac maleate.


Poster Abstracts

In vivo, drug release was studied in dogs after IA injection of 10 mg CDS-PM-101 in pellets. Blood samples were taken at selected intervals (15 min to 192 h), and synovial fluid collected at sacrifice (intervals from 24 h to 29 days). Levels of CDS-PM-101, and the parent compounds, morphine and diclofenac, were determined. Results and discussion Hydrolysis of the ester bond is base catalyzed: t 1/2 for disappearance of the CODRUGk compound was 170 h at pH 3, 150 h at pH 5 and 6.5 h at pH 7.4. Hydrolysis was faster still in the presence of 50% serum at pH 7.4 (t 1/2 = 16 min). In vitro testing demonstrated continuous release of morphine from the pellet formulation over ten days (Fig. 3). Eight percent (8%), 32%, 55% and 77% of the total morphine was released after 17, 65, 137 and 257 h, respectively. No CDS-PM-101 was detected in simulated synovial fluid.

Fig. 3. In vitro release of morphine from CDS-PM-101 in pellets (n=6).

Fig. 4. Mean concentrations after intra-articular administration of CDS-PM-101 formulated as pellets.

The levels of CDS-PM-101, morphine and diclofenac were determined in plasma and synovial fluid at various time points after intraarticular administration of pellets into the stifle joints of beagles (Fig. 4). CDS-PM-101 was not detected in plasma and was detected in synovial fluid (17.2 ng/ml) in only one animal at 72 h. Morphine was measurable in plasma (~4.0 ng/ml) of one animal, at early timepoints. Morphine (4.0–63.5 ng/ml) was detected in the synovial fluid of 3/4 animals 24-h postadministration. Diclofenac was detectable in both plasma and synovial fluid for at least 72 h postdosing. Conclusion Extended release of morphine under physiological conditions was achieved by synthesizing a CODRUGk compound of the analgesic, combined with formulation into a continuous release pellet. IA injection of CDS-PM-101 in the pellets resulted in sustained release of morphine within the joint, with no measurable systemic exposure to the CODRUGk compound.

Poster Abstracts


COMPLEX COACERVATION WITH HEPARIN RESULTS IN DECREASED LYSOZYME STABILITY M. van de Weert, M. Bendix Andersen, S. Frokjaer Department of Pharmaceutics, the Danish University of Pharmaceutical Sciences, Universitetsparken 2, 2100 Copenhagen O, Denmark Summary Complex coacervates of lysozyme and heparin were prepared. Structural analysis by infrared and fluorescence spectroscopy did not reveal significant changes upon complexation. In contrast, differential scanning calorimetry revealed a significantly reduced thermal stability of lysozyme upon complexation. In agreement, the storage stability of lysozyme was markedly less in the complexes. In conclusion, complex coacervation should be used with care in drug delivery or purification processes, as protein stability may be negatively affected. Introduction Complex coacervation is the complexation of at least two compounds, of which at least one is a macromolecule, through charge–charge interactions, resulting in phase separation. The process has been used in protein purification [1] and in the formation of drug delivery systems [2]. However, little attention has been paid to stability of proteins in such complex coacervates. This is surprising, since it is well known that proteins may be destabilized by adsorption to charged surfaces [3]. We have investigated the stability of hen egg-white lysozyme (lysozyme) in lysozyme–heparin complexes as a model system. Lysozyme is a very stable protein and is positively charged below pH 10. Heparin is a polysulfated glucosaminoglycan that is negatively charged above pH 2. Lysozyme and heparin are known to form insoluble complexes between pH values of 2 and 10, and this complexation has been used in the purification of lysozyme. Previous results in our laboratory (unpublished results) indicate that at low ionic strength up to 10 lysozyme molecules can bind to a single heparin chain of 16–18 kDa. Materials and methods Hen egg-white lysozyme, heparin (~17 kDa) and all buffer salts were purchased from Sigma. Lysozyme-heparin complexes were formed by co-addition of appropriate amounts of lysozyme and heparin at ~5:1 molar ratio in 10 mM Tris buffer pH 7.2. Thermal stability of the protein in the complexes and in a control lysozyme solution were determined by differential scanning calorimetry (DSC) using an ultrasensitive Microcal VP-DSC. The protein tertiary structure was monitored by intrinsic tryptophan fluorescence on a Spex Fluorolog 3–22 spectrometer. The protein secondary structure was determined by Fourier transform infrared spectroscopy (FTIR) using a Bomem MB100 series FTIR spectrometer. Long-term protein stability was evaluated by storing the complexes and a control lysozyme solution for 12 weeks at 37 8C. The complexes were subsequently dissociated by adding sodium chloride and the protein concentration in the supernatant was determined by using size-exclusion chromatography on a Waters LC Module I. Results and discussion Some proteins are known to bind to heparin specifically, and this binding usually results in an increased thermal stability. In contrast, lysozyme complexation by heparin results in a large decrease in the thermal stability, as shown by DSC (Fig. 1). The melting temperature (T m) of lysozyme in the complexes is approximately 61 8C, as opposed to 77 8C in the uncomplexed lysozyme. In agreement with this apparent destabilization observed in the DSC, we found approximately 40% protein loss upon incubation of the complexes at 37 8C. No loss was observed in the plain lysozyme solutions.

Fig. 1. Normalized DSC thermograms of plain lysozyme solution (solid line and lysozyme-heparin complexes (dotted line).


Poster Abstracts

Structural analysis of the protein in the complexes did not show any significant deviations from the native protein structure. The second derivative FTIR spectra in the Amide I region (Fig. 2) and the fluorescence spectra (Fig. 3) are very similar. These results indicate that the decreased thermal stability of lysozyme upon complexation may be related solely to a bigger affinity of the unfolded lysozyme for heparin.

Fig. 2. Inverted second derivative FTIR spectra of plain lysozyme solution (solid line and lysozyme-heparin complexes (dotted line).

Fig. 3. Normalized intrinsic fluorescence spectra of plain lysozyme solution (solid line) and lysozyme-heparin complexes (dotted line). Conclusion Complexation to heparin significantly reduces the thermal and long-term stability of lysozyme. Heparin-binding does not cause any major structural rearrangements in lysozyme, suggesting that the unfolded lysozyme has a higher affinity for heparin than the natively folded lysozyme. Overall, these studies indicate that complex coacervation may decrease the stability of proteins significantly. Complexatin should thus be handled with care as a technique in, e.g., protein drug delivery systems or protein purification. Acknowledgements MvdW would like to acknowledge an Investigator Fellowship from the Alfred Benzon Foundation. Funding for the FTIR and fluorescence specteometer by Apotekerfonden af 1991 and partial funding of the DSC by Novo Nordisk is gratefully acknowledged. References [1] Y.-F. Wang, J.Y. Gao, P.L. Dubin, Biotechnol. Prog. 12 (1996) 356–362. [2] D. Renard, P. Robert, L. Lavenant, D. Melcion, Y. Popineau, J. Gue´guen, C. Duclairoir, E. Nakache, C. Sanchez, C. Schmitt, Int. J. Pharm. 242 (2002) 163–166. [3] R.J. Green, I. Hopkinson, R.A.L. Jones, Langmuir 15 (1999) 5102–5110.

Poster Abstracts


IV. Nano-particulate Systems and Other Novel Concepts in Drug Delivery BUBBLE ASSISTED DRUG DELIVERY M. Arora, C.D. Ohl Institute for Biomedical Technology (BMTI), Physics of Fluids, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500AE Enschede, The Netherlands Summary Adherent epidermal cells are shown to take up fluorescent molecules (calcein and FITC-dextran) following the exposure to cavitation bubble activity created by a tensile acoustic pulse of 4 MPa peak amplitude. The geometrical uptake pattern of the fluorescent dye molecules allows to identify two mechanisms for drug uptake, both induced by the bubble dynamics close to cells: First, shear stress generated from non-spherically collapsing individual bubbles and second, the vivid flow field from the growth and collapse of a cavitation bubble cluster. Introduction In various studies, acoustic devices have been shown to stimulate the uptake of extracellular material into the cytoplasm without causing permanent damage to the cell [1–5], a process called sonoporation. Acoustic waves promise to be advantageous to certain drug delivery applications because they can be applied noninvasively, localized, and remotely controlled inside the patient’s body. However, the exact mechanisms causing the permeabilization of cell membranes are not yet clarified. In this in vitro study, the poration of HeLa cells was investigated following exposure to cavitation bubble activity generated by a shockwave device. These shockwave devices are typically used for extracorporal treatment of kidney stones. Cavitation bubbles generate and expand during the tensile part of the pressure pulse. When the pressure recovers to normal atmospheric pressure, cavitation bubbles shrink in an accelerated manner (also known as cavitation collapse). The collapse of these bubbles especially near boundaries or under influence of other bubbles can become aspherical leading to formation of liquid jets. Using fast imaging techniques and fluorescence microscopy, we show that these bubble collapse events not only cause detachment of the adhering cells but are also responsible for uptake of normally nonpermeable molecules into the cell’s cytoplasm. Experimental methods HeLa cells were grown at 37 8C and 5% CO2 in Iscove’s Modified Dulbecco’s medium (Invitrogen, Breda Netherlands) and plated in polystyrol culture flasks. The flasks were filled up with medium prior to being exposed to cavitation bubble activity. Following exposure, cells were tested for viability using ethidium bromide/acridine orange (Fluka, Zwijndrecht, Netherlands) staining at final concentrations of 5 and 1.5 mg/ml, respectively. Transient membrane permeabilization was checked by the uptake of fluorescein isothiocyanate dextran (FITC-dextran, 20 kDa) or calcein. Non-membrane-permeant FITC-dextran (1 mg/ml) or calcein (1 mg/ml) was added to the cell medium before shock wave exposure. Dye uptake of attached cells was determined by fluorescence microscopy after washing the cells four times with PBS and adding ethidium bromide at a final concentration of 5 mg/ml. Transiently porated cells, loaded with FITC-dextran or calcein appear green under fluorescence excitation. Permanently damaged cells appear orange due to ethidium bromide staining.

Fig. 1. A typical pressure recording of the shock wave passage at 6 kV discharge voltage.


Poster Abstracts

Acoustic pulses were generated by a focused piezoceramic source, which was adapted from the commercial lithotripter Piezolith 3000 (Richard Wolf, Knittlingen, Germany). The lithotripter contains two piezoceramic layers with a diameter of 300 mm and a focusing angle of 948. In our experiments, the operating voltage of the lithotripter was set between 3.5 and 7 kV and only the frontal piezoceramic layer was used. The pressure in the free field at the lithotripter focus was measured with a calibrated needle type hydrophone. A typical recording at 6 kV is presented in Fig. 1. The wave consists of a positive pressure pulse of about 1.5 As with a steep front and amplitude between 10 and 40 MPa (depending on the discharge voltage) followed by a tensile pressure pulse of amplitudes around 4 MPa, which lasts for several microseconds. Cells adhering on the flask surface are placed in the acoustic focus of the shockwave generator. For the optical observations, a long-distance microscope (K2 with CF4 objective, Infinity, USA) was used, which was attached to a dual frame CCD (charged coupled device) camera (Imager 3S, LaVision, Gfttingen, Germany).

Fig. 2. Cavitation bubble nucleation on the substrate layer: Picture sequence depicts exemplarily the detachment of cells following the shock wave passage. Time data in the upper left is given with respect to the shock front hitting the substrate. (the bar depicts 200 Am).

Fig. 3. The image shows cells after shock wave exposure and subsequent washing with PBS under a fluorescence microscope. Cells displaying green color have taken up FITC-dextran. Red color reveals dead cells because of the intercalation of ethidium bromide into the DNA (frame size 0.720.60 mm).

Results and conclusions A typical example of the interaction of cavitation bubbles with cells is depicted in Fig. 2. The time data is given with respect to the shock wave impact on the substrate. Shortly after that impact, we see two bubbles appearing due to the tensile pressure pulse following the shock wave. Between the second and third frame of Fig. 2, both bubbles will develop a jet flow directed towards the substrate (rigid boundary) [6], which subsequently results into an outward radial spreading flow. This flow pattern is sufficiently strong to detach cells located in the vicinity of the bubbles. Cells only become detached at the positions where the bubble dynamics was visible. The initial detachment of cells could be directly correlated to the positions of cavitation bubbles. No detachment was observed when the strength of the lithotripter generated tensile pressure was below the threshold for the nucleation of cavitation bubbles. Fig. 3 depicts attached cells after shock wave exposure with the lithotripter operating at 6 kV. Prior to shock wave treatment, the cell medium was supplemented with FITC-dextran. It can be seen that the substrate is partially cleared of cells caused by cavitation activity. Cells, which have not been washed away but line the border between occupied and vacated regions, emit a strong green fluorescent light upon excitation originated by the uptake of FITC-dextran. Only slight emission of green fluorescence is detected for cells further away from the region of cavitation-induced detachment. Dead cells appear red or brown in the image due to the intercalation of ethidium bromide into the DNA of permanently damaged cells. The uptake of the smaller molecule calcein can be detected much further from the cavitation activity site than the uptake of FITC-dextran. Fig. 4 depicts a large-scale calcein uptake pattern with three different regions. Region 1 marks an area where a lot of ring-like uptake patterns are found which are formed by the collapse of single bubbles close to cells. Region 2 does not show ring-like structures though the cells in this region also show uptake of the fluorescent dye molecules. This is attributed to the vivid flow field generated by the formation and collapse of cluster of cavitation bubbles. In region 3, again, some ring-like pattern is found which correlates with bubble activity caused from reflection due to the geometry of the setup. From our experiments, we can conclude that cell detachment is caused by the generation of cavitation. This statement is based on the fact, that the detachment on cells was only observed when cavitation bubble dynamics took place. Also, it is shown that the violent

Poster Abstracts


bubble activity leads to permeabilization of the cell membrane causing uptake of the molecules which are normally nonpermeable to the cell membrane. The geometrical pattern of uptake reveals that not only individual bubble activity leads to membrane permeabilization but that also the flow field generated by the multiple bubble formation can lead to similar scenarios.

Fig. 4. Fluorescence image showing large-scale uptake pattern of calcein.

Acknowledgements The experimental work is supported by FOM (The Netherlands) under grant 00PMT04. R. Ikink (University of Twente) and B. Wolfrum (Third Physical Institute, Germany) are acknowledged for their support during this work. References [1] [2] [3] [4] [5] [6]

M. Delius, P.H. Hofschneider, U. Lauer, K. Messmer, Lancet 345 (1995) 1377. L. Junge, C.D. Ohl, B. Wolfrum, M. Arora, R. Ikink, Ultrasound Med. Biol. 29,1769–1776. K.Y. Ng, Y. Liu, Med. Res. Rev. 22 (2002) 204–223. M.W. Miller, D.L. Miller, A.A. Brayman, Ultrasound Med. Biol. 46 (1996) 1131–1154. C.D. Ohl, B. Wolfrum, Biochim. Biophys. Acta—Gen. Sub. 1624 131–138. B. Wolfrum, R. Mettin, T. Kurz, W. Lauterborn, Appl. Phys. Lett. 81 (2002) 5060–5062.


Poster Abstracts

QUALITATIVE PREDICTION OF SOLUBILIZATION OF HIGHLY HYDROPHOBIC DRUGS IN BLOCK COPOLYMER MICELLES J. Latere Dwan’Isa1, M. Dinguizli2, V. Pre´at2, A. Arie¨n1, M. Brewster1 1 J&J-Pharmaceutical Research and Development—Drug Evaluation, Beerse, Belgium 2 Universite´ Catholique de Louvain, Unite´ de Pharmacie Gale´nique, Brussels, Belgium Summary Micelles obtained from block copolymers of polyethylene glycol and random copolyesters of q-caprolactone and trimethylene carbonate (50/50) can be used as carriers for hydrophobic drugs. We show in this study that the drug loading into the micelles depends strongly on the compatibility of both blocks with the drug considered. Using modeling, we developed a methodology that opens the way to qualitatively predict the drug solubility in polymeric micelles based on polymer–drug interaction parameters.

Introduction During the last two decades, block copolymers have been extensively evaluated as drug carriers [1]. Indeed, micelles can be formed in aqueous solutions of amphiphilic di- or tri-block copolymers that associate in water in such a way that the hydrophobic blocks form the core of the micelle and the hydrophilic blocks come into contact with the aqueous environment as a corona. Various studies have shown that these micelles can encapsulate hydrophobic drugs and release them in vivo [2]. In recent years, researchers at Johnson & Johnson have developed a new family of biocompatible and biodegradable di-block copolymers containing polyethylene glycol (PEG) and a random copolyester of q-caprolactone (CL) and trimethylene carbonate (TMC) [3]. The net advantage of this new family of liquid diblock copolymers is their ability to self-emulsify in the presence of water to form micelles of ca. 20 nm. This family of polymers proved to be efficient in encapsulating hydrophobic drugs and further releasing them in a controlled way [4]. This paper aims at presenting a qualitatively predictive approach to the solubilization of some common hydrophobic drugs (risperidone, ketoconazole, indomethacin and hydrocortisone) in polymers and more specifically in polymeric micelles formed by di-block copolymers in which PEG is the first block and the random copolyester of CL with TMC, i.e., P(CL-co-TMC), is the hydrophobic segment. The prediction is based on the polymer–drug compatibily that was determined using a model based on the Hansen’s approach to solubility [5].

Experimental methods Di-block copolymers were prepared and characterized as presented elsewhere [3,4]. The molecular weight of PEG is 750 g/mol while the P(CL-co-TMC) is about 1500 g/mol and is a 50/50 mixture of both monomers. Drugs were first mixed with the copolymers at room temperature for 24 h and then water was added. The freshly prepared solutions were stirred for 24 h at room temperature. Solubility data are an average of at least three measurements carried out at room temperature. Molecular modeling was used to estimate solubility parameters. Hansen solubility data were determined by the group contribution method with Molecular Modeling Pro software (ChemSW).

Results and discussion Solubilization of drugs into polymeric micelles is a complex mechanism that involves different parameters, e.g., hydrophobicity, molecular volume, crystallinity, flexibility, charge and the interfacial tension against water [1]. However, one of the key parameter is certainly the polymer– drug compatibility. An excellent way to assess the compatibility of the drug (=solubilizate) and the polymer (=solvent) is to evaluate the Flory–Huggins solubility parameter (v sp) [6]. The polymer is a thermodynamically good solvent if v sp is low (typically close to zero), meaning that when the compatibility is significant, solubilization occurs. We use a thermodynamic approach based on the extended Hildebrand solubility model developed by Hansen to determine the interaction parameter v sp [5]. In the Hildebrand approach, the solubility parameter (d), which is defined as the root square of the cohesive energy (i.e., the energy of vaporization per volume unit), is used to calculate v sp using Eq. (1):

vsp ¼ ðds  dp Þ2  V =R  T


Poster Abstracts


where s and p refer to solubilizate and polymer, V is the molar volume of the solubilizate (=drug), R the gas constant and T the temperature in Kelvin. Hansen modified the Hildebrand approach and divided d into three components that take into account force of dispersion (d d), polarity (d p), and hydrogen bonds (d h). Therefore, the solubility difference (D) between the solubilizate and the drug is defined by Eq. (2). Therefore, v sp is calculated using Eq. (3). D ¼ ½ðds  dp Þ2dispersion þ ðds  dp Þ2polarity

þ ðds  dp Þ2hydrogen 1=2


vsp ¼ D2  V R  T


With Molecular Modeling Pro, we estimated the three components of the Hansen solubility parameter for the drugs and the block polymers (Table 1). This software utilizes a group contribution method to approximate d values. Solubility differences (D) between the drugs and both segments of the di-block copolymer and the respective v sp are then calculated at 298 K (Table 2). In general, the lower the value of D, the better is the solubilization. Typically, D must be lower than 5 (J/cm3)1/2. In the case of the drugs studied, D values are systematically higher than 5 suggesting limited solubility. As a general observation, values of v sp range from 3 to 16. In order to have a good solubility, v sp should be as close as possible to zero. It is clear from Table 2 that solubility of these drugs in both blocks should be quite low to almost impossible because of the poor (or bad) compatibility. Table 1 Hansen solubility parameters of drugs and block of PEG750 and P(CL-co-TMC) (50/50) at 298 K determined by Molecular Modeling Pro software Compound

MW (g/mol)

Molar volume (cm3)

d (J/cm3)1/2

d d (J/cm3)1/2

d p (J/cm3)1/2

d h (J/cm3)1/2

Risperidone Indomethacin Ketoconazole Hydrocortisone P(CL-co-TMC) PEG

410.49 357.79 532.43 362.47 1545 750

316 275 410 279 – –

24.4 23.6 25.8 23.0 23.9 21.6

21.4 20.9 22.9 15.6 22.6 16.5

6.9 6.5 7.5 7.1 1.4 9.9

9.5 8.8 9.3 15.4 7.6 9.8

Table 2 Evaluation of the compatibility between hydrophobic drugs with PEG and P(CL-co-TMC) (50/50) at 298 K Compound



v sp (CL/TMC)

v sp (PEG)

Solubility (mole/ml)

Risperidone Indomethacin Ketoconazole Hydrocortisone

5.941 5.508 6.340 11.930

5.753 5.650 6.853 6.325

4.501 3.367 6.651 16.026

4.221 3.543 7.771 4.505

0.527105 1.034105 0.347105 0.395105

Based on v sp parameter, we tried to predict qualitatively the solubility of the drugs in the copolymers and subsequently in the micelles. Indeed, we assumed that solubilization in the micelles can occur in either the core or the corona or in both regions. We predicted based on v sp values that the compatibility with the core, i.e., the P(CL-co-TMC) block, will be as follows: indomethacinN risperidoneNketoconazoleNhydrocortisone. In the same manner, the compatibility with the PEG-corona is: indomethacinNrisperidone z hydrocortisoneNketoconazole. Assuming that solubilization of a drug can occur in both regions of the micelles, i.e., core and corona, we can predict the overall compatibility and thus the solubility. Therefore, the solubility in micelles can be ranked as follows: indomethacinNrisperidoneNhydrocortisonecketoconazole. At this point, since we cannot quantitatively predict the solubility extent, we believe that ketoconazole and hydrocortisone should show similar values. Indeed, hydrocortisone has a better compatibility than ketoconazole towards PEG and is highly incompatible with the polyester core, but the reverse is observed when it comes to ketoconazole and the P(CL-co-TMC) core. Experiments were carried out to check the validity of the qualitative prediction. Excess of drug was mixed with the di-block copolymers then water was added to prepare a 10% w/v solution of polymer in water. Solubility data measured are listed in Table 2. Results show that indomethacin


Poster Abstracts

is the most soluble of all the drugs considered followed by risperidone and then hydrocortisone and ketoconazole with close values (ca. 0.4 mole/ ml). These results are in line with the prediction made based on the polymer–drug compatibility. Conclusions This study demonstrates that qualitative prediction of solubilization of hydrophobic drugs in micelles is possible using a thermodynamic model that assesses the polymer–drug compatibility through the Hildebrand interaction parameter v. The approach we used is based on the Hansen solubility parameter of the drugs and the different polymers of the di-block copolymer. Results indicate that the prediction is in line with what is experimentally observed proving the validity of the approach. This methodology is a promising tool because it allows the screening of drugs in development in pharmaceutical labs and thus represents a gain of time and money. On the other hand, it is possible to choose a specific drug and screen between different polymers for the most suitable to use for drug solubilization (or dispersion) in a polymer. References [1] C. Allen, D. Maysinger, A. Eisenberg, Colloids Surf., B Biointerfaces 16 (1999) 3–27. [2] (a). G.S. Kwon, K. Kataoka, Adv. Drug Deliv. Rev. 16, 295–309; (b). M.-C. Jones, J.-C. Leroux, Eur. J. Pharm. Biopharm. 48 (1995) 101–111. [3] A.M.E. Arie¨n, M.E. Brewster, A. Nathan, J. Rosenblatt, L.M. Ould-Ouali, V. Pre´at, Polymeric Microemulsions, Patent WO03093344. [4] L. Ould-Ouali, A. Arie¨n, J. Rosenblatt, A. Nathan, P. Twaddle, T. Matalenas, M. Borgia, S. Arnold, D. Leroy, M. Dinguizli, L. Rouxhet, M. Brewster, V. Pre´at, Pharm. Res. (submitted). [5] (a) C.M. Hansen, Ind. Eng. Chem. Prod. Res. Dev. 8 (1969) 2; (b) Hansen Solubility Parameters. A User’s Handbook, CRC Press LLC, Boca Raton, FL, USA (2000). [6](a) R. Nagarayan, M Barry, E. Ruckenstein, Langmuir 2 (1986) 210–215; (b) F. Gadelle, W.J. Koros, R.S. Schechter, Macromolecules 28 (1995) 4883–4892.

Poster Abstracts


FACTORIAL DESIGN, PHYSICOCHEMICAL CHARACTERISATION AND ACTIVITY OF CIPROFLOXACIN-LOADED PLGA NANOPARTICLES FOR OCULAR USE K. Dillen, J. Vandervoort, A. Ludwig Laboratory of Pharma ceutical Technology and Biopharmacy, Department of Pharmaceutical Sciences, University of Antwerp, Universiteitsplein 1, 2610 Antwerp (Wilrijk), Belgium Summary Poly (lactide-co-glycolide) nanoparticles incorporating ciprofloxacin HCl were prepared by W/O/W emulsification solvent evaporation. A 24 full factorial design based on four independent variables was used to plan the experiments. The effects of various preparation parameters on the particle size, zeta potential and drug release kinetics were investigated. The influence of gamma irradiation on drug release was evaluated and the activity of the nanoparticles against two microorganisms was examined. Introduction The bioavailability of eye drops is low due to the defence mechanisms of the eye, which result in a rapid elimination of the drug solution after instillation and a limited drug permeation. Ophthalmic use of colloidal carriers can solve some of these problems, since biodegradable nanoparticles have the advantage of a controlled release of the drug incorporated, achieving the required tear levels and therapeutic effects [1,2]. Nanoparticles are frequently made of poly (lactic-co-glycolic acid), which is biodegradable and biocompatible [3]. The physicochemical characteristics and release kinetics can be controlled by modifying the preparation method or the choice of additives [4]. Experimental methods The poly (lactic-co-glycolic acid) or PLGA chosen was ResomerR RG 503 (Boehringer Ingelheim, Ingelheim am Rhein, Germany). Poly (vinylalcohol) was purchased from Sigma (St. Louis, USA). Ciprofloxacin HCl was supplied by Roig Pharma (Barcelona, Spain). The nanoparticles were prepared by W/O/W emulsification solvent evaporation followed by high-pressure homogenisation [5]. The resulting nanosuspension was subsequently freeze-dried. The four preparation factors investigated were the ciprofloxacin HCl concentration in the inner water phase, the addition of boric acid to this phase, the oil:outer water phase (O:W) ratio and the number of homogenisation cycles. The mean particle size Z ave and the zeta potential of the nanoparticles in Simulated Lachrymal Fluid were determined by Photon Correlation Spectroscopy and Electrophoretic Light Scattering, respectively (Zetasizer 3000, Malvern Instruments, Malvern, UK). Franz diffusion cells were used for the in vitro drug release experiments. The drug loading was determined by a validated HPLC method. The samples were gamma-irradiated at a dose of 25 kGy with 60Co as irradiation source. The antimicrobial effectiveness of the drug-loaded nanoparticles was assessed in comparison with a ciprofloxacin aqueous solution and nanoparticles without drug by measuring the minimal inhibitory concentrations (MIC) and minimal bactericidal concentrations (MBC) on Pseudomonas aeruginosa and Staphylococcus aureus. Results and discussion The effects of the different preparation parameters on the physicochemical and release properties of the nanoparticles are summarised in Table 1. Nanoparticles that were homogenised for more cycles possessed a significantly reduced particle size because smaller emulsion droplets were formed. The mean particle size of 234.68 nm was reduced with 46.02 nm when the particles were homogenised for two more cycles. The other preparation factors had no significant influence on the particle size. After gamma-irradiation, the mean particle size increased a little; this points probably to a slight aggregation. Table 1 Effects of the different factors on the properties of the nanoparticles investigated Parameter



Boric acid

Ciproflox. conc.

Ratio O:W

Particle size (nm) Zeta potential (mV) Release 24 h (%)

234.68 6.01

46.02* 1.03

4.31 0.12

3.48 0.16

5.13 0.23






* Statistically significant effect ( pb0.05).


Poster Abstracts

The zeta potential of the nanoparticles was examined in Simulated Lachrymal Fluid, which possessed an electrolyte composition similar to tear fluid. The relative high salt concentration of SLF caused the equalising of the surface charges. Only a small, but not significant difference in zeta potential was measured between the different kinds of nanoparticles prepared. As can be derived from Table 1, homogenisation caused a significant retardation of the drug release. Since smaller nanoparticles were prepared, one would expect higher drug release rates, due to the corresponding increase in total surface area, unless a denser polymer matrix structure and more tortuous diffusion path compensate for this effect, and no significant effect on the release kinetics is observed. Changing the O:W ratio from 1:5 to 2:5 resulted in a significantly higher drug fraction released. As the organic solvent content was higher, solvent evaporation and particle hardening were slower. The resulting porosity of the polymeric matrix led to an increase of the diffusion rate.

Fig. 1. Ciprofloxacin release profiles. D: ratio O:W 2:5, o: ratio O:W 1:5, open symbols: one homogenisation cycle, filled symbols: three homogenisation cycles.

Fig. 1 depicts the drug release profiles of the nanoparticles. The highest drug release was observed in the case of nanoparticles prepared with an O:W ratio 2:5 which have undergone one homogenisation cycle. The slowest release rate was measured from nanoparticles prepared with an O:W ratio 1:5 which have undergone three homogenisation cycles. The drug release can thus be controlled by varying the number of homogenisation cycles and/or the O:W ratio. Drug release after gamma sterilisation of the nanoparticles was faster, because during irradiation, polymer chains are split and form radicals. Recombination of these radicals leads to crosslinking or chain scission, which affects the polymer’s molecular weight and degradation and consequently drug release kinetics. An overview of the results of the microbiological assay is shown in Table 2. The MIC and MBC values were determined during 4 days, but did not change as a function of time. Only the MBC value for P. aeruginosa depended on the dosage form used. The MIC for P. aeruginosa and the MIC and MBC values for S. aureus were independent of the formulation. The blank nanoparticles showed no antimicrobial activity, although the increased acidity of the medium after degradation of PLGA into lactic and glycolic acid could have inhibited bacterial growth. Gamma irradiation had no influence on the MIC and MBC values of the ciprofloxacin-loaded nanoparticles. Therefore, although the drug has not been released for 100% after 24 h, the ciprofloxacin concentration in the medium is high enough to kill microorganisms to the same extent as the aqueous solution.

Table 2 MIC and MBC values of ciprofloxacin aqueous solution and ciprofloxacin-loaded PLGA nanoparticles, before and after gamma-irradiation Bacteria Ps. aer. St. aur.

Cipro aqueous sol.


Nanoparticles g-ster.

MIC (Ag/ml)

MBC (Ag/ml)

MIC (Ag/ml)

MBC (Ag/ml)

MIC (Ag/ml)

MBC (Ag/ml)

0.146–0.293 0.586–1.17

0.293–0.586 0.586–1.17

0.293 0.586–1.17

0.293–1.17 0.586–1.17

0.146–0.293 0.586–1.17

0.293–1.17 0.586–1.17

Conclusion Homogenisation of the W/O/W emulsion decreased the release rate of ciprofloxacin from PLGA nanoparticles. Drug release was faster after gamma sterilisation and when the volume of the organic phase—and the O:W ratio—was higher. However, the optimal parameters for this ocular drug delivery system cannot be stated, since in vivo studies should be performed and tear levels measured to determine which release kinetics are optimal to treat ocular infections.

Poster Abstracts


Acknowledgements K. Dillen is a Research Assistant of the Fund for Scientific Research-Flanders (Belgium) (F.W.O. Vlaanderen). The authors wish to thank Prof. D. Vanden Berghe, dr. P. Cos and Mrs. R. Vingerhoets, Laboratory of Microbiology, University of Antwerp, for the assistance with the microbiological tests and Dr. Pierre Dardenne, IBA Mediris S.A., Fleurus, Belgium for gamma sterilisation of the nanoparticles. References [1] A. zur Mqhlen, C. Schwarz, W. Mehnert, Solid lipid nanoparticles (SLN) for controlled drug delivery, Drug release and release mechanism, Eur. J. Pharm. Biopharm. 45 (1998) 149–155. [2] M.S. Romero-Cano, B. Vincent, Controlled release of 4-nitroanisole from poly (lactic acid) nanoparticles, J. Control. Release 82 (2002) 127–135. [3] J.M. Anderson, M.S. Shive, Biodegradation and biocompatibility of PLA and PLGA microspheres, Adv. Drug Deliv. Rev. 28 (1997) 5–24. [4] F. Gabor, B. Ertl, M. Wirth, R. Mallinger, Ketoprofen-poly (D,l-lactic-co-glycolic acid) microspheres: influence of manufacturing parameters and type of polymer on the release characteristics, J. Microencapsul. 16 (1999) 1–12. [5] J. Vandervoort, A. Ludwig, Biocompatible stabilizers in the preparation of PLGA nanoparticles: a factorial design study, Int. J. Pharm. 238 (2002) 77–92.

DEVELOPMENT AND EVALUATION OF SUSTAINED RELEASE OCULAR MINITABLETS W. Weyenberg1, A. Vermeire2, J-P. Remon2, A. Ludwig1 1 Laboratory of Pharmaceutical Technology and Biopharmacy, University of Antwerp, Universiteitsplein 1, B2610 Wilrijk—Antwerp, Belgium 2 Laboratory of Pharmaceutical Technology, University of Ghent, Harelbekestraat 72, B9000 Ghent, Belgium Summary Various sustained release minitablets (F 2 mm, 6 mg) were manufactured by means of a compression force at 1.25 kN, using sodium fluorescein as model drug and different bioadhesive powders, based on drum dried waxy maize starchR, AmiocaR starch and CarbopolR 974. Afterwards, the tablets were sterilised with gamma-irradiation. The friability remained below 1% for all the minitablets prepared. The crushing strength and the in vitro release properties were influenced by the amount of Carbopol. Introduction Eye drops are considered to be an ineffective dosage form, despite their widespread use due to their easy administration. The bioavailability of aqueous eye drops is low due to reflex blinking, rapid drainage and lacrimation. When applying conventional eye drops, frequent instillations are necessary to maintain a therapeutic drug level in the tearfilm or at the site of action. Therefore, numerous studies are conducted to increase the drug bioavailability by prolonging the contact time between drug and corneal/ conjunctival epithelium. The use of films or inserts was proposed to allow a drug release over a long period of time [1]. These ophthalmic dosage forms are effective, but not well accepted by patients. Recently an ocular minitablet with sustained release properties was developed and optimized. The bioadhesive polymers employed were drum dried waxy maize starch and CarbopolR 974 P (5%, w/ w; PM95). This minitablet is bioerodible, well accepted by humans in a preliminary investigation and showed no mucosal irritation potential. The gelling behaviour in the fornix is an advantage since it results in an extended residence time of 8 h at the absorption site [2,3]. In the present study, other nonirritating bioadhesive polymer mixtures based on CarbopolR 974 P were evaluated to manufacture ocular minitablets. The friability, the crushing strength and the in vitro release properties of the minitablets were evaluated to obtain a controlled drug release. Experimental methods Materials Drum dried waxy maize starchR (Cerestar, Vilvoorde, Belgium), AmiocaR (National Starch, Bridgewater, NJ, USA), Carbopol 974R P (Noveon, Cleveland, OH, USA), co-spray dried AmiocaR and CarbopolR 974 P (National Starch, Bridgewater, NJ, USA), sodium stearyl


Poster Abstracts

fumarate (Edward Mendell, NY, USA) and sodium fluorescein (Sigma, St. Louis, MO, USA) were employed to prepare 10 different batches of ocular minitablets (6 mg and diameter 2 mm). Afterwards, the minitablets were sterilised at 25 kGy (Sterigenics, IBA-Mediris, Fleurus, Belgium). An overview of the composition of the different batches is given in Table 1. Table 1 Composition of the different powders employed for the preparation of ocular minitablets Physical mixture based on DDWM

Physical mixture based Co-spray dried powder mixture based on on Amioca Amioca and Carbopol

PM 95/5 PM 90/10 PM 85/15 PM 75/25 PM 95a/5 PM 85a/15 CS 95/5 CS 90/10 CS 85/15 CS 75/25 Sodium fluorescein Sodium stearyl fumarate Carbopol 974 P DDWM Amioca

2 1 5 92 –

2 1 10 87 –

2 1 15 82 –

2 1 25 72 –

2 1 5 – 92

2 1 15 – 82

2 1 5 – 92

2 1 10 – 87

2 1 15 – 82

2 1 25 – 72

Methods The surface structure of the powder mixtures was evaluated using scanning electron microscopy (JSM 5600 LV-SEM, JEOL, Tokyo, Japan). The friability of the minitablets was determined by subjecting 10 tablets weighed together with 100 glass beads (average diameter of 4 mm) to falling shocks for 10 min in an Erweka friabilator (TA3, Offenbach/Main, Germany). After 10 min, the glass beads were removed, the minitablets were then weighed and the percentage friability was calculated. The radial hardness of the minitablets was determined, using an instrumented uniaxial press with a 20-N load cell (type L1000R, Lloyd Instruments, Segenworth, Fareham, UK). Data were obtained from 10 minitablets, prepared at 1.250 kN. A minitablet was transferred to a vial containing 1.00 ml Simulated Lacrimal Fluid (pH 7.4) in an oscillatory shaking bath to study the release of fluorescein [3]. The profiles obtained (n=3) from the dissolution test were subjected to the Higuchi mathematical model (Eq. 1). pffi Qt ¼ K H t


t is time K H is release rate constant Q t is total amount of drug dissolved in time t. Results and discussion SEM photographs of the co-spray dried powder mixture containing 15% CarbopolR 974 P (CS85/15) and the physical mixture with 15% CarbopolR 974 P, based on drum dried waxy maizeR (PM85/15) are presented in Fig. 1. Scanning electron microscopy pictures and solid state NMR spectroscopy revealed that by spray drying AmiocaR starch/CarbopolR 974 P, Carbopol films are formed around the starch granules [4].

Fig. 1. SEM photographs of CS85/15 (A) and PM85/15 (B).

The physical characteristics of the ocular minitablets are summarised in Table 2. The crushing strength of the minitablets is higher when the amount of CarbopolR 974 P in the physical powder mixtures is increased. Contrary, the amount of CarbopolR 974 P in the co-spray dried powders had no influence on the crushing strength of the minitablets, prepared at 1.250 kN. The friability remained below 1% for all the tablets

Poster Abstracts


prepared. The release rate constants (K H ) were calculated by using the Higuchi mathematical model and are also given in Table 2. The average rate values were smaller in the case of minitablets with 25% (w/w) CarbopolR 974 P than the minitablets containing 5% (w/w) CarbopolR 974 P. A strong gel is formed with PM75/25 and CS75/25 and herewith the release from these sterilised minitablets is slower.

Table 2 Physical properties of the different minitablets prepared [mean (S.D.)] PM 95/5 PM 90/10 PM 85/15 PM 75/25 PM 95a/5 PM 85a/15 CS 95/5 CS 90/10 CS 85/15 CS 75/25 Crushing strength (kN) 10.98 (1.36) Friability (%) 0.78 (0.11) In vitro release: K H 6.88 (0.31)

13.47 (3.67) 0.92 (0.24) 6.01 (0.32)

14.45 (2.43) 0.82 (0.16) 5.63 (0.18)

24.94 (3.00) 0.38 (0.19) 4.75 (0.12)

8.67 (1.74) 0.94 (0.25) 6.89 (0.45)

19.38 (2.94) 0.56 (0.19) 5.21 (0.42)

25.5 (1.65) 0.51 (0.45) 7.05 (0.73)

24.3 (0.89) 0.56 (0.35) 5.97 (0.23)

23.16 (2.79) 0.11 (0.04) 5.13 (0.20)

21.64 (3.73) 0.05 (0.09) 4.98 (0.17)

Minitablets prepared with co-spray dried AmiocaR with 25% (w/w) CarbopolR 974 P (CS75/25) and the physical mixture (PM75/25) exhibit the slowest in vitro sustained release properties (Fig. 2). However, Adriaens et al. (2003) reported that these two powders have mucosal irritating properties [5]. No significant differences were obtained for the K H values of PM85/15, PM85a/15 and CS85/15. But, as indicated in Fig. 2, a slower release of fluorescein from CS85/15 compared to PM85/15 and PM85a/15 is observed. CS85/15 with no mucosal irritating properties in slugs, will be further selected for its in vivo evaluation in healthy volunteers. In this clinical study, the tearfilm concentrations of fluorescein will be determined to evaluate the release properties in vivo.

Fig. 2. Fluorescein release profiles.

Conclusion By using co-spray dried AmiocaR with 15% (w/w) CarbopolR 974 P (CS85/15), a slower release can be achieved compared to the physical mixtures of AmiocaR starch or drum dried waxy maize starchR with CarbopolR. Moreover, CS85/15 is preferred, as it does not cause any mucosal irritating properties. References [1] F. Gurtler, R. Gurny, Patent literature review of ophthalmic inserts, Drug Dev. Ind. Pharm. 21(1) (1995) 1–18. [2] J. Ceulemans, A. Vermeire, E. Adriaens, J.P. Remon, A. Ludwig, Evaluation of a mucoadhesive tablet for ocular use, J. Control. Release 77 (2001) 333–344.


Poster Abstracts

[3] W. Weyenberg, A. Vermeire, J.P. Remon, A. Ludwig, Characterisation and in vivo evaluation of ocular bioadhesive minitablets compressed at different forces, J. Control. Release 89 (2001) 329–340. [4] D. Ameye, J.P. Remon, Scanning Electron Microscopy and solid state NMR analysis of spray-dried starch/CarbopolR 974 P mixtures. Pharm. Res. (submitted). [5] E. Adriaens, D. Ameye, M. Dhondt, P. Foreman, J.P. Remon, Evaluation of the mucosal irritation potency of co-spray dried AmiocaR/Poly (Acrylic Acid) and AmiocaR/CarbopolR 974 P mixtures. J. Control. Release 88 (2003) 393–399. Acknowledgements This study was supported by a grant from the FRO (Funds for Research in Ophthalmology, Belgium). The authors are grateful to Dr. P. Dardenne for the performance of gamma irradiation (Sterigenics, IBA-Mediris, Fleurus, Belgium).

A SENSORY VALVE IN LIPOSOMAL DRUG DELIVERY SYSTEMS A. Kocer-Sagiroglu1, E. Bulten1, M. Walko2, B. Feringa2, G. Robillard1, W. Meijberg1 Biomade Technology Foundation, Nijenborgh 4, 9747 AG, Groningen, The Netherlands 2 Department of Organic and Molecular Inorganic Chemistry, Stratingh Institute, University of Groningen, Nijenborgh 4, 9747 AG, Groningen, The Netherlands 1

Summary A mechano-gated membrane channel protein is used as a controllable, nonselective, large, aqueous pore in sterically stabilized liposomal drug delivery systems. The channel protein is engineered first to sense the light and/or the pH under iso-osmotic conditions, and then to convert these signals into conformational changes ultimately leading to pore formation in the drug loaded liposomes. This system offers not only the target specific drug delivery but also control over the timing of the delivery. Introduction Gated membrane ion channels are the elementary units for rapid signal detection, transduction and transmission in living cells [1]. Mechanogated channels function as mechano-electrical switches in diverse physiological processes [2]. Among those, the large conductance mechanosensitive channel from E. coli (MscL) is the best-studied one. It is composed of five identical 15 kDa (136 amino acid) subunits and gated by tension transmitted through the bilayer alone [3] to form a pore of about 40 2 [4]. Mutagenesis experiments showed that the hydrophilicity of a largely conserved amino acid at the 22nd position of the protein affects the channel gating according to the hydrophobicity of the substitution [5]. In addition, the introduction of charges into this position has shown to open the channel even without any tension on the membrane [6]. The aim of this work is to chemically modify the MscL channel into a pH and/or a light responsive device that can be used as a sensory valve in liposomal systems for the controlled drug delivery. Materials and methods His-tagged MscL protein was purified to near homogeneity via a single Ni-NTA column. pH and/or light responsive chemical compounds were synthesized and covalently linked to the cysteine residue on the 22nd position of the protein either through disulfide bond formation or an alkylation reaction. The free label was removed with a desalting column step. The labelled protein, then, was reconstituted in synthetic lipids in 1:120 (w:w) protein to lipid ratio. After the labelling was confirmed by electron spray mass spectrometry, the activity of the channel in liposomal vehicles was followed by a modified calcein fluorescence efflux assay.

Results and discussion Chemical molecules that protonate in response to a drop in pH, were synthesized and successfully linked to the 22nd position of the MscL channel protein. The labeled protein was reconstituted in DOPC:Cholesterol:DSPE-PEG (70:20:10 Molar ratio) liposomes at a 1:120 (wt.:wt.) protein to lipid ratio in the presence of a self quenching fluorescent dye. After separation from the free dye, liposomes were analyzed for efflux of the fluorescent dye under iso-osmotic conditions but at different pHs. A typical pH-responsive channel activity is shown in Fig. 1. Below the pK a of the label, the constriction zone of the pore is charged and consequently more fluorescent dye is released. By varying the pK a of the chemical labels, the pH interval for the channel gating transition can be tuned to the the desired range.

Poster Abstracts


When a light-sensitive label is connected to the MscL channel, there is no release of dye at any pH in the absence of illumination of the proteoliposomes. After activation, the photolysable part of the label is released and a protonatable group is left on the protein. This group, in turn, charges the pore resulting in channel opening and release of the fluorescent dye. None of the labels interacted with the wild type MscL, which has no cysteine for the coupling reaction.

Fig. 1. Liposomal drug release mediated by a pH-sensitive valve.

Fig. 2. Liposomal drug release mediated by a light-sensitive valve. Conclusions In this work, we have engineered a mechano-sensitive channel into a pH- or light-sensor under iso-osmotic conditions by using molecular biology, biophysics and organic synthesis techniques. By designing and synthesizing chemical compounds with different pK a values we can tune the sensitivity of the sensor to a specific pH interval. References [1] O.P. Hamill, D.W. McBride Jr, TINS, 17 (1994) 439–443. [2] S.I. Sukharev, P. Blount, B. Martinac, C. Kung, Annu. Rev. Physiol. 59 (1997) 633–657; I.R. Booth, P. Louis, Curr. Opin. Microbiol. 2 (1999) 166–169; J.M. Wood, Microbiol. Mol. Biol. Rev. 63 (1999) 230–262; O.P. Hamill, B. Martinac, Physiol. Rev. 81 (2001) 685–740; B. Martinac, Cell. Physiol. Biochem. 11 (2001) 61–76.

376 [3] [4] [5] [6]

Poster Abstracts

S.I. Sukharev, M.J. Schroeder, D.R. McCaslin, J. Membr. Biol. 171 (1999) 19–183. G. Chang, R.H. Spencer, A.T. Lee, M.T. Barclay, D.C. Rees, Science 282 (1998) 2220–2226. K. Yoshimura, A. Batiza, M. Schroeder, P. Blount, C. Kung, Biophys. J. 77 (1999) 1960–1972. K. Yoshimura, A. Batiza, C. Kung, Biophys. J. 80 (2001) 2198–2206.

OLIGOPEPTIDES AS TARGETING STRUCTURES IN CANCER THERAPY M. Pechar1, K. Ulbrich1, T. Etrych1, A. Fabra2, M. Stevenson3, L. Seymour3 1 Institute of Macromolecular Chemistry, Academy of Sciences of the Czech Republic, 162 06 Prague 6, Czech Republic 2 Institut de Recerca Oncolo`gica, Gran Vı´a s/n, km 2.7, 08907 L’Hospitalet, Barcelona, Spain 3 Department of Clinical Pharmacology, University of Oxford, Radcliffe Infirmary, Woodstock Road, Oxford OX2 6HE, United Kingdom

Summary A laminin derived targeting oligopeptide sequence YESIGVAVS [1] was covalently linked to N-(2-hydroxypropyl)methacrylamide (HPMA) based copolymers [2] via either C- or N-terminus and the uptake of the polymers to prostate cancer cells was studied. Two peptides derived from gonadotropin releasing hormone (GnRH) [3] were also conjugated to HPMA copolymers and used for modification of adenoviral gene delivery vectors. The influence of the polymer coating with and without the targeting peptide on the expression of reporter genes was investigated.

Introduction A number of oligopeptide sequences has been described in recent years for targeting of cytostatic drugs to tumour cells. These peptides are either derived from natural biologically active peptides or proteins [4] or iterated from phage display libraries [5]. In some cases, the chosen sequences are attached directly to the drug molecule [6], in others they are bound via a suitable spacer to a natural or synthetic macromolecular carrier together with the drug [7]. The second approach, i.e., use of the polymer carrier combines improved pharmacokinetics of the drug (by regulating its plasma circulation) and enhanced specificity (due to the targeting moiety) to maximise efficacy and diminish undesired side effects. Gene delivery vectors represent a sophisticated, though still not fully developed, class of specific anti-cancer drugs and can be based both on viral or non viral vector systems [8,9]. In this work, we describe peptide-targeted gene delivery vectors based on polymer-modified adenovirus designed for prostate cancer therapy. We have used a laminin derived sequence YESIGVAVS [4] and two GnRH analogues Pyr-HWSYKLRPG-amide (GnRH I) and PyrHWSHKWYPG-amide (GnRH II) [3].

Experimental methods The targeting peptides were synthesized by manual solid phase peptide synthesis using 9-fluorenylmethoxycarbonyl/tert-butyl (Fmoc/tBu) strategy on 2-chlorotrityl chloride resin (YESIGVAVS) or Rink amide resin (GnRH analogues). Pseudoproline dipeptide derivative Fmoc-Glu(OtBu)-Ser(cMe,Me-pro)-OH was used in penultimate condensation step of YESIGVAVS synthesis to avoid aggregation of the peptide chains on the resin. Protected peptide was cleaved from the 2-chlorotrityl resin with 20% TFE (trifluoroethanol) in CH2Cl2 and bound to copolymer of HPMA with 8 mol% of N-methacryloyl-Gly-Gly-4-nitrophenyl ester (HPMA-Ma-GG-ONp) via the N-terminus of the peptide. The unreacted ONp groups were aminolysed with 1-aminopropan-2-ol and completely deprotected with 95% trifluoroacetic acid (TFA). Alternatively, the peptide was bound via its C-terminus to a copolymer modified with ethylene diamine: HPMA-Ma-GFLG-NHCH2CH2NH2. The unprotected peptides (GnRH I and GnRH II) were cleaved from the Rink amide resin with 95% TFA and reacted with a heterobifunctional poly(ethylene glycol) succinimidyl ester (Fmoc-NHPEG-COOSu) via the lysine 3-amino group. The PEG-peptides were reprotected with trityl chloride. Fmoc group was removed with piperidine/DMF, the NH2-PEG-peptide reacted with the ONp-containing copolymers and the trityl groups were removed with acetic acid. The ability of polymer–peptide conjugates to bind and internalise into cells was determined using PC-3 cells plated out in 10-cm dishes. 100 Ag/mL of fully aminolysed conjugates EC127-130 were incubated with the cells in DMEM, 2% foetal calf serum and at 37 8C for 75 min. The medium was then removed, the cells rinsed with PBS. Then, the cells were hydrolysed with 10 M HCl at 100 8C for 2 h. Samples were neutralised with NaOH and the 1-aminopropan-2-ol content analysed by HPLC analysis. The extent of internalisation was determined by incubating cells with the conjugates at either 4 8C (permissive of cell binding only) or 37 8C (permissive of binding and internalisation) for 75 min. Following removal of the medium, selected cells were washed with 0.2 M acetic

Poster Abstracts


acid to remove any peptide from the surface of the cells. Any conjugate taken up into the cell would be protected from this mild acid wash. Hydrolysis, neutralisation and HPLC analysis were then performed as described above. EC127 EC128 EC129 EC130 EC136 EC140 EC146 EC147

poly(HPMA-co-Ma-GG-aminopropanol) poly(HPMA-co-Ma-GFLG-aminopropanol) poly(HPMA-co-Ma-GG-YESIGVAVS) poly(YESIGVAVS-NHCH2CH2NH-Ma-co-HPMA) poly(HPMA-co-Ma-GFLG-ONp-co-Ma-GFLG-Nle-GnRH I) poly(HPMA-co-Ma-GFLG-ONp-co-Ma-GFLG-Nle-GnRH II) poly(HPMA-co-Ma-GG-ONp-co-Ma-GFLG-Nle-GnRH I) poly(HPMA-co-Ma-GG-ONp-co-Ma-GFLG-Nle-GnRH II)

Modification of virus tropism was examined by infecting PC-3 cells with unmodified, polymer-coated or conjugate-retargeted virus. The unreacted ONp groups of the conjugates were used to bind to amino groups on the surface of adenovirus particles encoding the luciferase reporter gene. Transduction efficiency was determined by measuring luciferase activity after 48 h. As a control a subset of cells were preincubated with adenovirus fibre protein to block virus attachment via CAR, the natural virus receptor in order to determine whether virus tropism was CAR independent. Results and discussion The peptide sequence SIKVAV is located in the alpha-1 chain of the laminin molecule [1]. Laminin is a major component of the extracellular matrix that plays an important role in cell adhesion. Integrins a1h1, a2h1, a3h1, a6h1, a7h1 and a6h4 serve as laminin receptors. Prostate epithelial cells express a6h4, whilst a6h1 has also been detected in highly metastatic tumours. Initial experiments demonstrated that polymer-coated adenovirus retargeted with a PEGylated YESIKVAVS peptide could deliver specific transgene expression in various metastatic cell lines. However, the PEGylated peptide carries two amino groups (at the end of the PEG spacer and on the side chain of lysine) available to react with the ONp-containing copolymers leading to a poorly defined product. The peptide YESIGVAVS was designed in order provide a single target for the ONp esters and thus generate a well-defined conjugate. In order to determine whether the modified peptide could still bind to cells we have studied the internalization of the YESIGVAVSpolymer conjugates in PC-3 cells (a prostate bone metastasis cell line). Analysis revealed substantially higher binding of the YESIGVAVS-polymer to the cells when the peptide was attached to the polymer via its C terminus (EC130) compared with the peptide-free polymer controls EC127, EC128. However, it should be noted that EC130 contained a relatively long spacer GFLGNHCH2CH2NH2 while the N-terminus-modified peptide–polymer conjugate (EC129) had only a short GG spacer. The internalisation assay demonstrated that EC130 was taken up by cells in levels not achieved by polymer alone. We are currently investigating the role of the spacer length and structure to the binding and internalisation activity of various targeting peptides. It is possible that the GFLG spacer has intrinsic binding activity since the control polymer EC128 exhibits slightly higher level of binding in comparison with EC127 that contains only GG.

We were also interested to examine gonadotropin-releasing hormone (GnRH) as a potential targeting ligand. GnRH receptors are mainly expressed on pituitary gonadotropes but are also expressed on PBMCs, ovary, testis, breast and prostate. GnRH agonists/ antagonists have been shown to cause dose dependent growth inhibition of tumour cells in vitro and in and animal xenografts. Cytotoxic


Poster Abstracts

analogues (e.g., doxorubicin linked to a GnRH agonist) have been used6 successfully to target GnRH receptors on human prostate cancer xenografts in nude mice resulting in tumour reduction. We have sythesised two derivatives of GnRH, GNRH I and GnRH II, both of which contain a Lys residue instead of Gly. The side chain of Lys was further modified with heterobifunctional spacer Fmoc-NH-PEGCOOSu. At this point, the peptides reacted with trityl chloride to reprotect His residue. When this step was omitted significant crosslinking was observed in subsequent conjugation to the polymer. After the Fmoc removal, the terminal amino group of PEG spacer reacted with the ONp groups of the polymers (HPMA-Ma-GFLG-ONp and HPMA-Ma-GG-ONp). We believe that insertion of PEG (M w=3400) spacer between the targeting sequence and the polymer carrier should enable better interaction of the peptide with receptor. Adenovirus was treated with GnRH I-polymer (EC146) or GnRH II-polymer (EC147) conjugates and the effect on transduction efficiency assessed. Transgene expression was significantly reduced in cells transduced with HPMA-Ma-GG-ONp coated virus in comparison with unmodified virus. Reporter gene expression was much greater in cells infected with EC146 or EC147 retargeted virus however preincubation with adenovirus fibre suggested that the majority of virus entry was still mediated by CAR and not by GnRH receptors. Surprisingly, the transduction properties of virus coated with HPMA-Ma-GFLG-ONp were affected to a much smaller degree than the HPMA-Ma-GG-ONp. Adenovirus treated with conjugates EC136 and EC140 containing GnRH I and GnRH II linked to HPMA via GFLG spacers generated reporter gene expression that was slightly higher than for unmodified virus. However, again much of this uptake was CAR dependent. Possible explanations for these findings could be the affinity of the GFLG sequence for cell binding mentioned above. Alternatively, expression of GnRH receptors may be too low for efficient uptake if the virus is reliant on integrin-mediated internalisation.

Conclusion Treatment of adenovirus with HPMA-Ma-GG-ONp polymer led to a 1000-fold reduction in reporter gene expression whilst treatment with HPMA-Ma-GFLG-ONp polymer at the same concentration-reduced transgene expression by only 10-fold. The YESIGVAVS peptide– polymer conjugate in which the peptide is attached via the C-terminus exhibited higher levels of internalisation compared with polymer alone. Viruses coated with GnRH-containing polymer conjugates whilst mediating significantly higher transgene expression than polymer alone were internalised in a CAR dependent manner. This would confirm that the new targeting polymers failed to ablate virus/fibre interactions. The fact that transgene expression in cells transduced with virus treated with conjugates containing GFLG spacers was slightly higher than unmodified virus may suggest an improved virus uptake mechanism either via GnRH receptors or mediated by the GFLG spacer itself. Acknowledgements This work was supported by EU grant No QLK 6 CT-2000-00280. References [1] [2] [3] [4] [5]

K. Tashiro, G.C. Sephel et al., J. Biol. Chem. 264 (27) (1989) 16174–16182. K. Ulbrich, V. Sˇubr et al., Macromol. Symp. 152 (2000) 151–162. J.N. Hislop, M.T. Madziva et al., Endocrinology 141 (12) (2000) 4564–4575. Y. Yamazaki, M. Tsuruga et al., Anticancer Res. 20 (3A) (2000) 1381–1384. F. Nilsson, L. Tarli et al., Adv. Drug Deliv. Rev. 43 (2–3) (2000) 165–196.

Poster Abstracts [6] [7] [8] [9]


M. Langer, F. Kratz et al., J. Med. Chem. 44 (9) (2001) 1341–1348. A.P.C.A. Janssen, R.M. Schiffelers et al., Int. J. Pharm. 254 (1) (2003) 55–58. N.K. Green, L.W. Seymour, Cancer Gene Ther. 9 (12) (2002) 1036–1042. F. Maruta, A.L. Parker et al., Cancer Gene Ther. 9 (6) (2002) 543–552.

MICELLES FROM NEW BIODEGRADABLE AMPHIPHILIC BLOCK COPOLYMERS CONTAINING PEG AND PCL F. Signori, R. Solaro, E. Chiellini Department of Chemistry and Industrial Chemistry, University of Pisa, via Risorgimento 35, 56126 Pisa, Italy Summary Biodegradable amphiphilic block copolymers, showing AB and ABA linear block structure, constituted by poly(ethylene glycol) [PEG] and poyl(caprolactone) [PCL] segments were prepared and characterized. Copolymer thermal behavior suggested that the bulk materials are phase separated. ABA copolymers, containing PCL in the central domain and PEG lateral chains of different length were found to self assemble in water. Their micellization behavior was investigated, and CMC values were determined by a fluorescence method, using pyrene as fluorescent probe. Introduction Recently, much interest has been focused on the study of polymeric micelles formed in aqueous solutions. Examples of biodegradable amphiphilic block copolymers able to self-assemble in water are reported. In most cases, the hydrophilic segment is a PEG residue, whereas the hydrophobic block is a biodegradable polymer chain, usually a polyester [1–3].Our interest in biodegradable polymeric micelles stems from their small size, apparent thermodynamic stability, and ability to deliver drug selectively in vivo with limited interaction with other biomolecules. Accordingly, a new class of potentially self-assembling amphiphilic block copolymers consisting of PEG and PCL segments was prepared by a straightforward synthetic procedure. Polymer micelle formation was investigated in water by a fluorescent probe technique. Experimental methods Amphiphilic block copolymers were prepared by Sn(Oct)2 catalyzed ring opening polymerization (ROP) of q-caprolactone initiated by monomethoxy poly(ethylene glycol) [MeOPEGOH]. Monomer feed was adjusted to obtain 10 PCL units in the resulting copolymer (PCx). Two equivalents of PCx copolymer were used in the hexamethylene diisocyanate promoted homocoupling, to give ABA amphiphilic block copolymers (Cx). Purified materials were characterized by 1H NMR and FT-IR spectroscopy, and by size exclusion chromatography (SEC). The polymer thermal behavior was investigated by DSC and TGA analysis. Solubility was tested in MilliporeR water at 25 8C. Fluorescence measurements were performed on polymer solutions (0.5–5d 105 g/l) in MilliporeR water at 258C, by using pyrene (3 107 M) as fluorescent probe.

Results and discussion Amphiphilic block copolymers were prepared in 68–94% yield (Table 1) by Sn(Oct)2 catalyzed bulk polymerization of 10 equivalents qcaprolactone [q-CL], initiated by MeOPEGOH samples of different molecular weight (M n=350, 750, 2000, and 5000). Table 1 Preparation of AB amphiphilic block copolymersa Run


Yield (%)

PCL blockb (units)

M cn (KDa)

M w/M cn


350 750 2000 5000

68 94 91 92

11 10 10 10

2.0 1.5 3.5 5.7

1.3 1.2 1.2 1.1

a b c

At 120 8C for 24 h; q-CL/MeOPEGOH=10. By 1H NMR. By SEC, PEG standards.


Poster Abstracts

Hexamethylene diisocyanate promoted homocoupling of the previously prepared AB diblock copolymers (Scheme 1) gave ABA (formally ABBA) amphiphilic block copolymers in high yields (Table 2).

Scheme 1. Preparation of AB (PCx) and ABA (Cx) amphiphilic block copolymers. Table 2 Preparation of ABA amphiphilic block copolymersa Run

PEG M n (Da)

Yield (%)

M bn (KDa)

M cn (KDa)

M w/M cn (%)

PCxd (g/l)

Solubilitye (mg/l)


C1 C2 C3 C4

350 750 2000 5000

95 91 94 96

3.4 3.4 nd nd

4.5 3.4 7.0 12.2

1.4 1.5 1.3 1.1

0 16 20 80

3 8 62 92

nd 3.27 4.31 4.92


In anhydrous toluene at 80 8C for 48 h; PCx/HMDI=2; nd=not determined. By 1H NMR. c By SEC, PEG standards. d Unreacted PCx. e In water at 25 8C. b


H NMR data, solubility and thermal (DSC and TGA) properties of the polymer samples are coherent with a block distribution of monomeric units. In some cases, the partial overlapping of diagnostic peaks in 1H NMR spectra prevented accurate determination of Cx polymer structure. Moreover, SEC chromatograms highlighted the residual presence of unreacted PCx prepolymer whose amount increased with increasing the PEG chain length. This result was attributed to partial isocianate hydrolysis by residual moisture traces. Both AB and ABA copolymers showed thermal stability better than 300 8C. DSC analysis showed in all cases the presence of a glass transition (Tg) between— 70 and 50 8C, and one complex endothermal transition between 40 and 80 8C, that were attributed to both PEG and PCL domains.

Fig. 1. Normalized emission (k exc 339 nm, left) and excitation spectra (k em 390 nm, right) of pyrene (3107 M) in water in the presence of increasing concentration of C3 sample.

Poster Abstracts


Moreover, Tg values were found to be highly dependent on PEG chain length. These results suggest that the materials are very likely phase separated. Due to the strong amphiphilic character, the prepared copolymers were expected to self-assemble in water. Micelle formation in water was investigated using a simple fluorescence method, in which pyrene acts as fluorescent probe [4,5]. Sample C1 was not analyzed due its to low water solubility. Normalized emission and excitation spectra of 3107 M pyrene in the presence of different concentrations of C3 copolymer are shown in Fig. 1. The most prominent feature is a progressive shift of the maxima of the pyrene excitation spectra on increasing the polymer concentration. This effect is due to the progressive change of the polarity of the pyrene environment, i.e., from the aqueous to the hydrophobic environment of the micelle core. The degree of polymer micellization was evaluated from the dependence of the I338/ I332.5 fluorescence intensity ratio on the logarithm of polymer concentration. Critical micellar concentration (CMC) values can be easily obtained by sigmoid fitting of the experimental data (Fig. 2). As a typical example, the plots of I338/I332.5 intensity ratio for C3

Fig. 2. General curve fitting and calculation procedure of CMC (sample C3).

Fig. 3. Dependence of I338/I332.5 intensity ratio in pyrene excitation spectra on the logarithm of the concentration (g/l) of C3 copolymer and MeOPEGOH 2000.

copolymer and MeOPEGOH vs. the sample concentration (expressed in g/l) are reported in Fig. 3. The clear difference between the two curves demonstrate that micellar organization is a peculiar phenomenon for ABA block copolymer structure. The CMC values of the investigated block copolymers ranged between 3103 and 5103 g/l (Table 2). These data are very close to those previously reported for materials of similar structure [4].The CMC increase observed on increasing the length of PEG segments was attributed to the increased hydrophilic/hydrophobic balance of the block copolymers. Conclusions A straightforward synthesis of amphiphilic AB and ABA block copolymers constituted by PCL and PEG segments is proposed. However, improvement of the experimental procedure is needed to increase the conversion of AB prepolymers to ABA copolymers. Thermal analysis suggests that all copolymers tend to phase separate. The strong propensity of copolymers to form micelles in water was clearly demonstrated by a fluorescent probe technique. Acknowledgements The financial support by MIUR is gratefully acknowledged. The assistance by Dr. Tarita Biver (University of Pisa) in the fluorescence measurements is greatly appreciated. References [1] S. Cammas-Marion, M.M. Bear, A. Harada, P. Guerin, K. Kataoka, Macromolecular Chemistry and Physics 201 (2000) 355–364. [2] S.Y. Kim, I.G. Shin, Y.M. Lee, Journal of Controlled Release 56 (1998a) 197–208. [3] I.G. Shin, S.Y. Kim, Y.M. Lee, C.S. Cho, Y.K. Sung, Journal of Controlled Release 51 (1998) 1–11. [4] M. Wilhelm, C.L. Zhao, Y. Wang, R. Xu, M.A. Winnik, J.L. Mura, G. Riess, M.D. Croucher, Macromolecules 24 (1991) 1033–1040. [5] R. Zana, in: Lindman, B., Ed., Amphiphilic Block Copolymers. Self Assembly and Applications. Elsevier, Amsterdam, 2000, pp. 221–252.


Poster Abstracts

CHANNEL PROTEIN-CONTAINING LIPOSOMES AS DELIVERY VEHICLES FOR THE CONTROLLED RELEASE OF DRUGS-OPTIMIZATION OF THE LIPID COMPOSITION J. Sˇmisterova´, M. van Deemter, G. van der Schaaf, W. Meijberg, G. Robillard BioMaDe Technology Foundation, Nijenborgh 4, 9747 AG Groningen, The Netherlands Summary In the design of liposomal drug formulations containing a controllable channel protein (MscL), the lipid composition is dictated in part by this membrane protein. This work addresses the question whether therapeutically optimal lipid compositions (phospholipid with high T m/cholesterol/ PEG) are compatible with channel activity. As a result of optimization, MscL-proteoliposomes have been prepared that rapidly release encapsulated model drug (calcein), as well as the chosen cytostatic drug, cisplatin. Introduction The mechanosensitive protein of large conductance (MscL) from E.coli, normally gated by the application of tension to the lipid bilayer, stays fully functional when purified and reconstituted in liposomes [1]. The signal leading to activation of the channel can, however, be changed from tension to, e.g., a drop in pH by altering the MscL via mutation and chemical modification. We have shown, for example, that the mutation of glycine at the 22nd position into cystein (G22C mutant) enables targeting of charged and pH-sensitive chemical groups to MscL. A lower pH (e.g., in solid tumors) can be used as a trigger for activation of the MscL after its reconstitution in the drug-containing liposomes leading to induced release of the drug. In the present work, the charge-induced gating of G22C by MTSET has been used to demonstrate the activity of MscL after reconstitution in proteoliposomes. The attachment of this positively charged reagent to the cystein of G22C causes MscL to gate spontaneously. Previously, we were succesful in reconstitution of overexpressed and purified MscL in liposomes consisting of both, neutral unsaturated lipids, like DOPC, DOPE and mixtures of neutral and negatively charged lipids, like DOPC/DOPS, DOPC/DOPG. However, it remained to be seen whether we could functionally reconstitute MscL in the therapeutically optimal lipid composition, typically represented by mixtures of a high transition temperature lipid, cholesterol and polyethyleneglycol (PEG). In addition, important parameters, like the protein-to-lipid ratio, lipid concentration, size of proteoliposomes and storage conditions have been optimized. Experimental methods Reconstitution: The detergent-mediated reconstitution of G22C mutant of the MscL into liposomes of different lipid compositions included titration of liposomes with Triton X-100 (or octylglucoside), incubation with MscL, addition of either the fluorescent marker (calcein) or cytostatic (cisplatin) and the removal of detergent by BioBeads SM-2. Unencapsulated drug was separated on Sephadex G50 column. Release of drug: The in vitro release profile of a model drug (fluorescent calcein) from G22C-containing liposomes was estimated in the calcein efflux assay after activation with MTSET ([2-(trimethylammonium)ethyl] methanethiosulfonate). The percentage release of a fluorescent drug was calculated from the dequenching of calcein fluorescence. The release profile of cisplatin was determined after the separation of the released drug on PD 10 column (110 cm) and detection with ICP-AES. The size of MscL-liposomes after extrusion through the filters of different pore size was determined by using dynamic light scattering. Results Reconstitution: MscL could be successfully reconstituted in the phospholipids with low transition temperature (fluid lipid bilayer), with the best activity obtained in DOPC, DOPG and less in POPC and eggPC. MscL-liposomes consisting of lipids with high transition temperature, like DMPC, DPPC, SM and DSPC, failed in preserving the channel gating activity. About 20 to 30 mol% cholesterol and up to 10 mol% DSPE-PEG 2000 could be incorporated in MscL-proteoliposomes without decreasing the channel activity. Thus, the best working lipid composition was the one consisting of DOPC/CHOL/PEG 2000 at a molar ratio of 70:20:10. Cisplatin was entrapped in the optimized MscL-proteoliposomes at a drug-to-lipid ratio of about 4 to 8 (Ag/mg total lipid). Release of drug: The optimized DOPC/CHOL/PEG 2000 proteoliposomes released about 80% of encapsulated calcein and about 45% of cisplatin under the charge-inducing activation of MscL with MTSET. The limited release of cisplatin might be explained by the possible interaction of this cytostatic with the channel protein. The protein-to-lipid ratio: For immunogenicity reasons, the amount of the reconstituted MscL should be kept as low as posssible. We were able to reduce the MscL-to-lipid ratio from 1:30 to 1:120 by optimizing the lipid concentration used in the reconstitution while retaining full functionality. Size: By extrusion of reconstituted MscL-liposomes through 200 nm filter, a homogenous population of particles of about 150 nm size was prepared. The extrusion of MscL-liposomes through 50 nm filter resulted in decreased channel protein activity.

Poster Abstracts


Conclusions We have demonstrated the controlled release of a model drug, as well as a real drug from channel-containing liposomes. Liposomes with MscL labelled with a pH-sensitive group will provide a useful tool for the delivery of drugs at target sites with a lowered pH value, such as solid tumors or sites of inflammation. The preparation of such MscL-containing liposomes is now in progress. References [1] C.C. Hase, A.C. Le Dain, B. Martinac, Purification and functional reconstitution of the recombinant large conductance mechanosensitive ion channel (MscL) of Escherichia coli, J. Biol. Chem. 270 (32) (1995) 18329–18334.

THERMOSENSITIVE AND BIODEGRADABLE POLYMERIC MICELLES WITH TRANSIENT STABILITY O. Soga, C.F. van Nostrum, W.E. Hennink Department of Pharmaceutics, Utrecht Institute for Pharmaceutical Sciences (UIPS), Faculty of Pharmaceutical Sciences, Utrecht University, P. O. Box 80.082, 3508TB, Utrecht, The Netherlands Summary A novel class of thermosensitive and biodegradable polymers, poly(N-(2-hydroxypropyl) methacrylamide lactate) (poly(HPMAm-lactate)), was synthesized. The cloud points of the polymers in aqueous solution were between 10 and 65 8C, depending on the copolymer composition. Block copolymers of poly(HPMAm-dilactate) with poly(ethylene glycol) (pHPMAmDL-b-PEG) formed polymeric micelles with a size around 50 nm in aqueous solution and showed transient stability due to the hydrolysis of lactate side chains, demonstrating the potential applicability of these systems for controlled drug delivery. Introduction Thermosensitive polymers with a lower critical solution temperature (LCST) are presently under investigation for biomedical and pharmaceutical applications [1,2]. These polymers are soluble in aqueous solution below the cloud point (CP), but precipitate above this temperature due to the dehydration of the polymer chains. For biomedical and pharmaceutical applications of thermosensitive polymers, it is important to have possibilities to control the CP around body temperature. Furthermore, polymers whose CP increase from below to above body temperature in time are very attractive materials, because, e.g., the controlled release of drugs without thermal treatment is feasible using such polymers [3]. The aim of this study is to develop a novel class of thermosensitive and biodegradable polymeric micelles that show controlled destabilization based on the increase of their critical micelle temperature (CMT) in time. Experimental methods Poly(HPMAm-monolactate) (pHPMAmML), poly(HPMAm-dilactate) (pHPMAmDL) as well as their copolymers (p(HPMAmML-coHPMAmDL)) were synthesized by radical polymerisation from the corresponding monomers. p(HPMAm-dilactate)-b-PEG block copolymers (pHPMAmDL-b-PEG) were synthesized by radical polymerisation using HPMAm-dilactate as monomer and methoxy-PEG-ABCPA (4,4azobis(4-cyanopentanoic acid) as macroinitiator. The products were characterized by 1H NMR (solvent: CDCl3) and gel permeation chromatography (GPC). The CP and the CMT of the polymer solution were determined with static light scattering (SLS). Onsets on the x-axis, obtained by extrapolation of the intensity–temperature curves to intensity zero were considered as the CP and the CMT. The critical micelle concentration (CMC) of block copolymers was determined using pyrene as a fluorescence probe. These measurements were performed in isotonic 120 mM ammonium acetate buffer (pH = 5.0, to minimize hydrolysis of lactate ester side groups) unless mentioned. Micelles of block copolymers were formed by quickly heating an aqueous polymer solution from 0 8C (below CMT) to 50 8C (above CMT) [4]. The size of micelles was determined by dynamic light scattering (DLS) at 37 8C. The destabilization of polymeric micelles at 37 8C and at different pH’s was evaluated by DLS and SLS. Results and dscussion Five poly(HPMAm-lactate) polymers (Fig. 1) with different monomer compositions were synthesized and their LCST properties were evaluated. Interestingly, all these polymers showed LCST behaviour, namely, reversible turbidity changes by heating and cooling.


Poster Abstracts

pHPMAmML has a rather high CP (63.0 8C) whereas pHPMAmDL has a relatively low CP (10.5 8C) (Fig. 2). This can be explained by the greater hydrophobicity of the dilactate side group over the monolactate side group. The CP of the copolymers linearly increased with mol% of HPMA-monolactate monomer (Fig. 2), meaning that the CP of the copolymers can be tailored by the copolymer composition.

Fig. 1. Structure of pHPMAmML (n=0), pHPMAmDL (m=0), and p(HPMAmML-co-HPMAmDL) (m,np0).

Fig. 2. CP of 1 mg/mL solution of p(HPMAmML-co-HPMAmDL) as a function of the mol% HPMAmML in the copolymer.

pHPMAmDL-b-PEG block copolymers with three different pHPMAmDL molecular weights and a fixed PEG molecular weight were synthesized (Table 1). With increasing pHPMAmDL block lengths, the CMT and the CMC decreased. These results can be attributed to the greater hydrophobicity of the thermosensitive block with increasing pHPMAmDL block molecular weight. All block copolymers formed micelles with unimodal size distribution in aqueous solution above their CMT. pHPMAmDL(3000)-b-PEG formed rather large micelles. Probably, this length of pHPMAmDL block (corresponding to 10 monomer units per polymer) is not hydrophobic enough to create a highly packed core structure, as indicated by its rather high CMC value. On the other hand, the other two polymers had low CMCs and formed micelles with a size around 50 nm. From these properties, prolonged blood circulation of these micelles after i.v. administration is expected.

Table 1 Characteristics of pHPMAmDL-b-PEG block copolymers Polymers

M na

M wa

M w/M n

CMT (8C)b

CMC (mg/mL)c

Z ave (nm)d

pHPMAmDL(3000)-b-PEGe pHPMAmDL(6900)-b-PEGe pHPMAmDL(13600)-b-PEGe

7400 11,900 15,000

10,400 23,300 32,800

1.41 1.95 2.18

12.5 7.5 6.0

0.15 0.03 0.015

59.8F0.15 51.6F0.20 53.1F0.98

a b c d e

M n=number average molar weight; M w=weight average molar weight determined by GPC. Determined by SLS for 10 mg/mL polymer solution. Determined from pyrene excitation spectra at 37 oˆ¯C. Determined by DLS for 1 mg/mL polymer solution at 37 oˆ¯C. Number in brackets is Mn of HPMAmDL block determined by 1H NMR. Mn of PEG is 5000.

The destabilization of pHPMAmDL-b-PEG at 37 8C and at different pHs was evaluated (Fig. 3). At pH 5.0, where the hydrolysis of lactate side groups is minimized, the micelles were stable over 60 h. On the other hand, at pH 9.0, the size of micelles started to increase after 2 h of incubation and the micelles completely dissolved in 4 h. Obviously, this destabilization of micelles resulted from the increase of CMT above 37 8C due to the hydrolysis of lactate side groups. It can be calculated that at physiological pH (7.4), the destabilization of micelles will start at 80 h and will be completed in 160 h [5].

Poster Abstracts


Fig. 3. Destabilization of pHPMAmDL(6900)-b-PEG micelles at 37 8C and at pH 5.0 (left) and pH 9.0 (right).

Conclusion Poly(HPMAm-lactate) is a novel class of thermosensitive polymers with a CP in aqueous solution between 10 and 65 8C, depending on the copolymer composition. pHPMAmDL-b-PEG formed polymeric micelles with a size around 50 nm and a narrow size distribution in aqueous solution. At pH9.0 the micelles were stable for 2 h, which corresponds to 80 h under physiological condition (pH 7.4, 37 8C), and then started to destabilize. This result indicates that controlled drug release systems can be developed with these new polymers, by which drugs start to be released after micelles reach the target site, e.g., tumor tissues. Furthermore, a good biocompatibility of these polymers is expected because the degradation products of poly(HPMAm-lactate) are lactic acid, an endogenous compound, and the watersoluble pHPMAm, which is known to be nontoxic macromolecular carrier [6]. Thus, we anticipate that pHPMAmDL-b-PEG have a great potential for biomedical and pharmaceutical applications. Drug loading into micelles is now under investigation in our group. Acknowledgements The authors thank Mitsubishi-Pharma Corporation (Japan) for their financial support. References [1] B. Jeong, S.W. Kim, Y.H. Bae, Adv. Drug. Deliv. Rev. 54 (2002) 37–51. [2] A. Kikuchi, T. Okano, Adv Drug Deliv Rev 54 (2002) 53–77. [3] D. Neradovic, C.F. van Nostrum, W.E. Hennink, Macromolecules 34 (2001) 7589–7591. [4] D. Neradovic, O. Soga, C.F. van Nostrum, W.E. Hennink, Biomaterials 25 (2004) 2409–2418. [5] D. Neradovic, M.J. van Steenbergen, L. Vansteelant, Y.J. Meijer, C.F. van Nostrum, W.E. Hennink, Macromolecules 36 (2003) 7491–7498. [6] R. Duncan, Nat Rev, Drug Discov. 2 (2003) 347–360.


Poster Abstracts


R. Vakil and G.S. Kwon School of Pharmacy, University of Wisconsin, Madison, WI 53705, USA Summary In this study, amphotericin B (AmB) has been incorporated in PEG-phospholipid micelles using a solvent evaporation method and the properties of this novel formulation have been studies. The aggregation state of the encapsulated AmB, an indicator of its toxicity, depends on the molar ratio of AmB to PEG-phospholipid. PEG-phospholipid micelles elute intact and with minimal drug loss in saline at 37 8C during SEC-HPLC experiments, indicating kinetic stability. AmB incorporated in PEG-phospholipid micelles showed low levels of hemolysis, which may indicate gradual release of the drug from the polymeric micelles. Introduction Amphiphilic block copolymers may assemble into micelles that encapsulate poorly water-soluble drugs. Our interest is in the delivery of poorly soluble polyene antibiotics, such as AmB. Despite the development of new antifungal drugs, AmB is still the drug of choice for the treatment of systemic fungal diseases. Monomers of the drug interact with the ergosterol in fungal membranes, causing changes in membrane permeability and cell death. On the other hand, soluble aggregates of AmB induce leakage of both mammalian and fungal cells, loss of selectivity, and toxic side effects [1,2]. Thus, delivery strategies that prohibit self-aggregation of AmB may increase its therapeutic index. The critical aggregation concentration of AmB is 1.0 AM. AmB is highly aggregated in its standard formulation, FungizoneR. Previous work in our lab with micelles formed from PEG-b-poly(N-hexyl-l-aspartamide) acyl esters have indicated that stearoyl chains favorably interact with AmB resulting in deaggregation of the drug [3]. PEG-phospholipids are generally regarded as safe and have been used with long circulating liposomes, which are approved for use in humans. Here we present results on PEG-phospholipid micelles with stearoyl chains, which show they can be used to incorporate and deaggregate AmB. PEG-phospholipid micelles reduce hemolysis caused by AmB by changes in aggregation state and sustained drug release. Experimental methods Reagents AmB was obtained from Chem-Impex (Wood Dale, IL) and was stored at 20 8C until use. 1,2-Distearoyl-sn-glycero-3-phosphoethanolamineN-methoxy(polyethylene glycol) (PEG-DSPE) (M n=2800, 5800 g/mol) was obtained from Avanti Polar Lipids (Alabaster, AL). All reagents used were of analytical grade and were used without further purification. Blood cells were obtained from ICR Female mice as packed cells in 1% EDTA from Harlan Bioproducts (Indianapolis, IN). Incorporation of AmB by PEG-DSPE micelles Stock solutions of PEG-DSPE (6.0 mg/mL) and AmB (0.3 mg/mL) prepared in chloroform/methanol (1:2) were mixed in a round bottom flask. The mixed solvent was evaporated under vacuum to produce a thin film of co-precipitated drug and polymer. This film was dissolved in 10 mM PBS (pH=7.4) and was incubated at 25 8C for 5 min. Unincorporated AmB was removed by filtration (0.22 Am). Spectroscopic analysis of AmB in PEG-DSPE micelles An aliquot of solution containing PEG-DSPE micelles with incorporated AmB was diluted with an equal volume of N,Ndimethylformamide (DMF), and drug content was evaluated by monitoring the absorbance of the drug at 412 nm. The encapsulation efficiency was expressed as a ratio of solubilized AmB relative to the initial amount of drug. The aggregation state of AmB in PEGDSPE micelles was evaluated from its absorbance spectra acquired from 300 to 450 nm at a scan step of 0.1 nm. Hydrodynamic diameter of PEG-DSPE micelles The sizing of the drug loaded PEG-DSPE micelles was performed using dynamic light scattering on the NICOMP 380ZLS particle sizer, using a fixed angle of 908. The light scattering data was interpreted using NICOMP analysis, and the particle sizes were expressed as volume weighted average diameters.

Poster Abstracts


SEC-HPLC elution profiles of PEG-DSPE micelles Aqueous solutions of PEG5k-DSPE micelles containing AmB (40 Ag/mL) were incubated at 37 8C in 10 mM PBS. At 30-min intervals, 100-AL samples were injected into a Shodex PROTEIN-KW 804 SEC column. Samples were eluted using 50 mM phosphate buffer at 0.75 mL/min. The elution of AmB was monitored by its absorbance at 412 nm. Hemolytic activity of AmB against murine erythrocytes The hemolysis caused by the formulations in murine erythrocytes was determined using a procedure similar to that described by Lavasanifar [4]. The percent lysis was determined by the equation: % lysis =(AbsAbs0)/ (Abs100Abs0)100, where Abs, Abs0, Abs100 is the absorbance of the supernatant for sample, control (with no AmB), and control for complete lysis (with 20 AM AmB). Results and discussion The level of AmB incorporated in PEG-DSPE by a simple solvent evaporation method was influenced by the starting level of PEG-DSPE and somewhat by the molecular weight of PEG. The level of AmB solubilized in water was 0.32 to 0.4 mg/mL. The yield of AmB ranged from 54% to 78%. The size of PEG-DSPE micelles was largely determined by the molecular weight of PEG and largely independent of drug content.


Molecular weight (g/mol)

PEG-DSPE: AmB (mol/mol)

PEG-DSPE: AmB (w/w)

[AmB] (mg/mL)

Yield of AmB

Diameter (nm)



0.93F0.01 5.53F0.37 0.64F0.02 5.20F0.05

2.81 16.81 4.03 32.61

0.32F0.01 0.39F0.01 0.38F0.12 0.4F0.02

0.54F0.01 0.73F0.07 0.78F0.03 0.77F0.01

13.8F4.3 10.8F2.5 21.8F6.1 19.3F4.4


The aggregation state of the encapsulated drug could be controlled by varying the molar ratio of the PEG-DSPE to AmB. The broad band in the absorption spectrum at 331 nm is characteristic of aggregated species of AmB. With increasing PEG-DSPE content, absorption at the bands corresponding to 368, 388 and 418 nm increase in intensity. The ratio of the absorbance of the drug at 331 to 418 nm, i.e., I/IV ratio, is taken to be a measure of the degree of aggregation [3]. This ratio decreases from 2.66 to 0.23 as the PEG-DSPE content is increased from a molar ratio of 0.5 to 2. The I/IV ratio remains constant at higher PEG-DSPE contents.

Fig. 1. Effect of initial PEG-DSPE content on the I/IV ratio, an indicator of the degree of aggregation of AmB.

After drug loading, PEG5k-DSPE micelles containing AmB (PEG-DSPE:AmB=4) were incubated at 37 8C and subjected to analysis by SEC-HPLC. AmB detected at 412 nm elutes as a sharp peak at 7.70 mL, indicating that most of the drug is loaded in PEG-DSPE micelles.


Poster Abstracts

Empty PEG5k-DSPE micelles elute at 7.68 mL (data not shown). Only a slight decrease in the micelle peak area was observed as a function of incubation time, which may indicate that the encapsulated drug is being released slowly from PEG-DSPE micelles.

Fig. 2. AmB elutes with PEG5k-DSPE micelles, in SEC-HPLC experiments. Free AmB causes considerable hemolysis for levels beyond its critical aggregation concentration. AmB in PEG-DSPE micelles incubated with RBCs induces low (b10%) levels of hemolysis. This contrasts with studies on AmB where the PluronicsR, successfully solubilize the drug but are unable to prevent hemolysis [5]. These results might indicate that the drug is stably encapsulated in the PEGDSPE micelles and may be released slowly as monomers, which are nontoxic at the membrane level.

Fig. 3. Hemolytic activity of AmB formulations against murine erythrocytes. Conclusions We have encapsulated AmB in PEG-DSPE micelles using a solvent evaporation method. Solubilized solutions of the otherwise poorly soluble drug were prepared, which may permit evaluation of in vivo toxicity in animal models. PEG-DSPE micelles effectively deaggregate AmB, which might be important in reducing the in vivo acute toxicity and nephrotoxicity associated with the self-aggregated drug. The in vitro hemolysis experiments indicate that the PEG-DSPE formulations effectively encapsulate and may slowly release the drug, resulting in low hemolysis of mammalian cells.

References [1] J. Brajtburg, J. Bolard, Carrier effects on biological activity of amphotericin B, Clin. Microbiol. Rev. (1996) 512–531. [2] I. Gruda, N. Dussault, Effect of the aggregation state of amphotericin B on its interaction with ergosterol, Biochem. Cell. Biol. 66 (1988) 177–183.

Poster Abstracts


[3] M. Adams, D. Andes, G. Kwon, Amphotericin B encapsulated in micelles based on poly(ethylene oxide)-block-poly(l-amino acid) derivates exerts reduced in vitro hemolysis but maintains potent in vivo antifungal activity, Biomacromolecules 4 (2003) 750–757. [4] A. Lavasanifar, J. Samuel, G. Kwon, Micelles self-assembled from poly(ethylene oxide)-block-poly(N-hexyl stearate l-aspartamide) by a solvent evaporation method: effect on the solubilization and haemolytic activity of amphotericin B, J. Control. Release 77 (2001) 155–160. [5] D. Forster, C. Washington, S. Davis, Toxicity of solubilized and colloidal amphotericin B formulations to human erythrocytes, J. Pharm. Pharmacol. 40 (1988) 325–328.

CONTROLLED DRUG DELIVERY WITH ULTRASOUND AND GAS MICROBUBBLES A. van Wamel1,2, A. Bouakaz1,2, B. Bernard3, F. ten Cate1, N. de Jong1,2 1 Thoraxcentre, Erasmus MC, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands 2 Interuniversity Cardiology Institute of the Netherlands, P.O. Box 19258, 3501 DG Utrecht, The Netherlands 3 Nuclear Medicine, Erasmus MC, The Netherlands Summary Although the underlying mechanisms are unknown, ultrasound together with contrast microbubbles can alter the permeability of cell membranes. We performed in vitro studies to demonstrate the enhanced internalization of a therapeutic drug using ultrasound and contrast microbubbles. Further, we report optical observations at a microsecond scale of the physical interaction between ultrasound, microbubbles and cells, which may help explaining the mechanisms underlying enhanced internalization of drugs by ultrasound and contrast microbubbles. Introduction A significant problem of therapeutic agents is the compromised quality of life experienced by the patient due to side effects of the therapeutic agents. Delivery of molecular medicine to the diseased area is often inefficient. An approach toward alleviating this problem is to develop innovative methods of drug targeting and delivery, thus minimizing the toxic effects of the drugs on healthy cells and tissues. Ultrasound in combination with contrast microbubbles has been shown to locally alter the permeability of cell membranes without affecting cell viability. This permeabilization feature is used to design new drug delivery systems using ultrasound and contrast agent microbubbles [1,2]. The aim of the study was to establish more insight in the increase in cell membrane permeability induced by interaction between ultrasound, microbubbles and cells. The present in vitro feasibility study demonstrates the enhanced internalization of a therapeutic drug. Further, the physical interaction between ultrasound, microbubbles and cells is shown optically which may help explaining the enhanced internalization of drugs. Methods Enhanced internalization of a therapeutic drug: The normal internalization levels of the radiolabelled peptide (111)In-octreotate were compared to the levels in rat-pancreatic tumour cells treated with ultrasound and contrast microbubbles. SonovueTM, a commercially available contrast-imaging agent is used. It consists of SF6 gas microbubbles encapsulated in a phospholipid shell. The mean size of the microbubbles is 2.5 Am. The cell cultures were positioned at a distance of 75 mm from the transducer and the temperature of the medium was kept constant 37 8C. Ultrasound exposure consisted of a burst of 10 As length, 1 MHz frequency and 420 kPa amplitude repeated each 75 ms. The total experimental times varied from 0 to 60 min. Optical observations of the interaction between ultrasound, microbubbles and cells: Optical observations of microbubbles and cells is possible through the use of a standard BH-2 Olympus microscope. Brandaris-128, an ultrafast camera, was used [3]. The Brandaris 128 is capable of making, in a single shot, a sequence of 128 images with a frame rate up to 25 millions frames per second. Porcine aortic endothelial cells were grown inside an Opticellk container. Sonovuek microbubbles diluted 1/1000 were used. Ultrasound was applied at 1 MHz frequency and the optical recording was performed at a frame rate of 10 MHz.

Results Fig. 1 shows the enhanced internalization of (111)In-octreotate when microbubbles and ultrasound are used compared to standard internalisation. The ultrasound contrast microbubble treatment increased the radiolabeled-peptide internalisation with 200%, 260%, 150% and


Poster Abstracts

110%, respectively, for 5, 15, 30 and 60 min exposure times. At t=60 min, an intracellular saturation level is reached at which the internalization is equal to the secretion.

Fig. 1. Internalization expressed as a percentage of the initially added doses of radioactivity per mg cell protein with and without ultrasound and microbubbles for four different treatment times: 5, 15, 30, and 60 min.

Fig. 2. Recording of the first two bursts of ultrasound applied on a Sonovuek microbubble nearby a cell.

Fig. 2 shows a selection of 24 optical images out of the 128 acquired during the first two cycles of ultrasound. A Sonovuek microbubble can be clearly seen nearby an endothelial cell. The initial interaction between an oscillating microbubble and the cell results in cell deformation. It was

Poster Abstracts


observed that during ultrasound irradiation, cell membrane deforms in a smooth curved-shaped reaching a few micrometers in length. Even when there is no contact between the microbubble and the cell, cell membrane displacement occurs as a result of the expansions of the microbubble. The cell membrane deformation changes with the oscillation behavior of the microbubble. Bubble oscillations occur as complex motions and a relation between bubble expansion and cell membrane displacement can be plotted (Fig. 3). Fig. 3 shows the changes in microbubble radius and the cell membrane displacement as calculated from Fig. 2. It is clearly seen that an expanding microbubble (max of excursion 4.7 Am) results in a cell membrane displacement (max of 1.0 Am). Further, motion of the microbubble oscillation and cell resulted in mutual attraction.

Fig. 3. Related changes in bubble radius (expansion/compression) and cell membrane displacement.

Conclusion The interaction with ultrasound contrast microbubbles results in an increased uptake of the drug by the cells [4]. In this in vitro study, we demonstrate that treatment with ultrasound and contrast microbubbles may be feasible and efficient. The optical study shows the physical action of oscillating microbubbles on the cell membrane. Although this study did not reveal how living cells sense mechanical forces, there is no doubt that perturbation of the oscillating microbubbles results in profound alterations in the cellular content. Further studies, revealing the functional relationships that lie beyond the ultrasound microbubble induced permeabilization of the cell membrane, will be of great value in designing an optimal local drug delivery system. Acknowledgements We gratefully acknowledge the financial support from the Netherlands Technology Foundation STW (Grant RKG 5104). References [1] K.Y. Ng, Y. Liu, Therapeutic ultrasound: its application in drug delivery, Med. Res. Rev. 22 (2) (2002) 204–223. [2] R.J. Price, S. Kaul, Contrast ultrasound targeted drug and gene delivery: an update on a new therapeutic modality, J. Cardiovas. Pharmacol. Ther. 7 (3) (2002) 171–180. [3] C.T. Chin et al., Brandaris 128: a 25 million frames per second digital camera with 128 highly sensitive frames, Rev. Sci. Instrum. 2004. In press. [4] A. van Wamel et al., Radionuclide tumour therapy with ultrasound contrast microbubbles. Ultrasonics 2004. In press.


Poster Abstracts

PTMC AND MPEG-PTMC MICROPARTICLES FOR HYDROPHILIC DRUG DELIVERY Z. Zhang, M.A. Foks, D.W. Grijpma, J. Feijen Institute for Biomedical Technology (BMTI), Polymer Chemistry and Biomaterials, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands Summary In this study, the potential of PTMC and mPEG-PTMC microparticles for controlled delivery of hydrophilic drugs was investigated. PTMC and mPEG-PTMC microparticles, loaded with hydrophilic model compounds (BSA, lysozyme and CoomassieR Brilliant Blue G) were prepared by the double emulsion method. High loading efficiencies can be achieved, and first order release profiles during 60 days were observed. Introduction The controlled release of proteins, peptides and DNA fragments has attracted much attention because of developments in tissue- and protein engineering, peptide synthesis and recombinant DNA technology. Due to the short in vivo half-lives of these hydrophilic compounds, sustained release systems are of great importance. Nano- and microparticles based on (co)polymers of lactide and glycolide have been developed as carriers in drug delivery. However, the degradation characteristics of these polymers result in the liberation of acidic compounds that can lead to denaturation and deactivation of the active components. Poly(trimethylene carbonate) (PTMC) degrades in vivo by surface erosion without the formation of acidic compounds [1]. In this study, we investigate the preparation of PTMC and mPEG-PTMC microparticles and the loading and delivery characteristics of hydrophilic model compounds. Experimental methods Polymer grade 1,3-trimethylene carbonate (TMC) was purchased from Boehringer Ingelheim (Germany). Monomethoxy poly(ethylene glycol) (mPEG3, M n=3000 g/mol) was obtained from Shearwater Polymers (USA). Stannous octoate (SnOct2), bovine serum albumin (BSA) (minimum 98% pure) and lysozyme from chicken egg white (3 crystallised, dialysed and lyophilised) were purchased from Sigma (USA). CoomassieR Brilliant Blue G (blue dye) was purchased from Aldrich (USA). All materials were used as received. 1-Hexanol (Merck, Germany) was distilled over CaH2 (Acros, Belgium) before use. The PTMC50 homopolymer and the mPEG3-PTMC50 diblock copolymer were prepared by ring opening polymerization of TMC using 1hexanol or mPEG3 as initiator and SnOct2 as catalyst [2]. The polymer characteristics determined by 1H NMR and by GPC are shown in Table 1.

Table 1 Characteristics of PTMC50 homopolymer and mPEG3-PTMC50 diblock coplymer Polymer


Targeted M n (kg/mol) 50 53

Purified polymer Mn (kg/mol)a

Mn (kg/mol)b

M w/ M bn

[g] (dL/g)c

– 55.9

69.2 69.0

1.6 2.0

1.48 1.21

By 1H NMR, bby GPC, cIntrinsic viscosity, by GPC.

Microparticles were prepared by double emulsion using PVA (2 wt.%) as a stabilizer. In the first (o/w) emulsion, 25 mg protein or 20 mg dye was emulsified in 5 ml polymer solutions in dichloromethane. After the second (w/o/w) emulsion, particles were formed upon evaporation of dichloromethane; the particles were then rinsed with water and freeze-dried [3]. The protein loading efficiency was determined by elemental analysis. The loading efficiency and the release in water at 37 8C of the blue dye were determined by UV absorption at 616 nm. The release experiments were conducted in duplicate. Results and discussion The different microparticles could readily be loaded with the hydrophilic model compounds: BSA (66 kg/mol), lysozyme (14 kg/mol) and CoomassieR Brilliant Blue G (blue dye, 854 g/mol). Spherical microparticles of PTMC50 and mPEG3-PTMC50, 1–50 Am in diameter, were

Poster Abstracts


formed. Fig. 1 shows light microscopy images of the mPEG3-PTMC50 microparticles, the dye was evenly distributed in the microparticles. Also, the shape and size did not change after freeze-drying and redispersion.

Fig. 1. Light microscopy images of mPEG3-PTMC50 microparticles. (A) microparticles loaded with lysozyme; (B) microparticles loaded with lysozyme, after freeze-drying and redispersion; (C) a microparticle loaded with CoomassieR Brilliant Blue G.

The loading efficiency of BSA and lysozyme in the different microparticles is shown in Table 2. The concentration of polymer in the organic phase mainly determines the loading efficiency. When 50 mg/ml polymer solutions were used, protein loading was negligible. At 100 mg/ml, high loading efficiency was achieved for PTMC50 microparticles, while loading efficiency of mPEG3-PTMC50 microparticles was only 10–20%. Increasing the polymer concentration to 140 mg/ml resulted in high loading efficiency for both polymers. When the polymer concentrations are higher than 100 mg/ml, PTMC50 microparticles can be loaded with protein more efficiently than mPEG3-PTMC50 microparticles. This can be related to the difference in viscosity: at a same polymer concentration (100 mg/ml or 140 mg/ml), the PTMC50 solution has a much higher viscosity than the mPEG3-PTMC50 solution. There is no significant difference in loading efficiency between BSA and lysozyme.

Table 2 Loading efficiency of PTMC50 and mPEG3-PTMC50 microparticles Polymer

Polymer concentration (mg/ml)

Loading efficiency(%) BSA



50 100 140 50 100 140

0 78.8 95.3 0 11.3 71.2

0 77.9 90.2 0 20.1 77.4


Release profiles of the microparticles were studied using CoomassieR Brilliant Blue G, as it allows convenient visual observation and detection by UV. CoomassieR Brilliant Blue G cannot be loaded into the microparticles when 60 mg/ml polymer solutions in dichloromethane are used in the preparation process. When 140 mg/ml mPEG3-PTMC50 solution or 100 mg/ml PTMC50 solution is used, the loading efficiency is 80.0% for mPEG3-PTMC50 and 78.0% for PTMC50 microparticles, respectively. Fig. 2 shows the aqueous phases after the preparation of mPEG3-PTMC50 microparticles loaded with CoomassieR Brilliant Blue G. The polymer concentration in dichloromethane was 140 mg/ml. When fully released, the dye concentration is 200 Ag/ml. A series of dye solutions in water (200–25 Ag/ml) are also shown. The concentration of sample A, the supernatant directly after the second (w/o/w) emulsion step, is approximately 25 Ag/ml. Sample B, the supernatant after solvent evaporation, has only a slightly higher concentration. This indicates that during solvent evaporation only little dye is released. Fig. 3 shows release profiles of microparticles loaded with CoomassieR Brilliant Blue G. For microparticles prepared from both polymers, first-order release profiles were achieved. More than 90% of the dye was released in 60 days.


Poster Abstracts

Fig. 2. CoomassieR Brilliant Blue G water solutions (200–25 Ag/ ml) and the aqueous phases after the preparation of mPEG3PTMC50 microparticles (polymer concentration in dichloromethane is 140 mg/ml; dye concentration is 200 Ag/ml when fully released). (A) supernatant after second (w/o/w) emulsion; (B) supernatant after solvent evaporation.

Fig. 3. Release profiles of microparticles loaded with CoomassieR Brilliant Blue G.

Conclusions Microparticles of PTMC and mPEG-PTMC, 1–50 Am in diameter, were prepared by the double emulsion method. After freeze-drying and redispersion, the shape and size of the microparticles did not change. Hydrophilic model compounds (BSA, lysozyme and CoomassieR Brilliant Blue G) could be loaded efficiently. Microparticles loaded with CoomassieR Brilliant Blue G showed a first order release profile of the dye for 60 days, during which 90% of the dye was released. References [1] A.P. Peˆgo, M.J.A. Van Luyn, L.A. Brouwer, P.B. Van Wachem, A.A. Poot, D.W. Grijpma, J. Feijen, J. Biomed. Mater. Res. 67A (2003) 1044. [2] Z. Zhang, D.W. Grijpma, J. Feijen (2004), Macromol. Chem. Phys., in press. [3] J. Bezemer, Protein Release Systems based on Biodegradable Amphiphilic Multiblock Copolymers, PhD thesis, University of Twente, 1999.

Poster Abstracts


V. Gene Delivery COVALENT ATTACHMENT OF AN NLS-PEPTIDE TO LINEAR DNA DOES NOT ENHANCE TRANSFECTION EFFICIENCY OF CATIONIC POLYMER BASED GENE DELIVERY SYSTEMS M. van der Aa1, G. Koning1,2, J. van der Gugten1, C. d’Oliveira3, R. Oosting4, W.E. Hennink1, D.J.A. Crommelin1 1 Department of Pharmaceutics, Utrecht Institute for Pharmaceutical Sciences (UIPS), Utrecht University, P.O Box 80082, 3508 TB Utrecht, The Netherlands 2 Department of Radiochemistry, Interfaculty Reactor Institute, Delft University, Mekelweg 15, 2629 JB Delft, The Netherlands 3 OctoPlus Technologies BV, Zernikedreef 12, 2333 CL Leiden, The Netherlands 4 Department of Psychopharmacology, Utrecht Institute for Pharmaceutical Sciences (UIPS), Utrecht University, P.O Box 80082, 3508 TB Utrecht, The Netherlands Summary To improve the transfection activity of cationic polymer-based gene delivery complexes, linearized plasmid DNA with a covalently attached NLS peptide was used to transfect both dividing and nondividing cells. Despite the functionality of the NLS peptide the transfection efficiency of the polymers used was not significantly increased. Introduction Gene therapy is based on delivery of genes into the nucleus and their subsequent translation into the therapeutic protein. Nonviral gene delivery systems, such as cationic polymers, are safer and more cost-effective than viral vectors, but are less efficient. Especially delivery into the nucleus is one of the major barriers for efficient nonviral gene delivery [1]. Overcoming this barrier would thus greatly improve the efficiency of nonviral gene delivery systems. For active transport into the nucleus, a nuclear localisation signal (NLS), which can interact with the nuclear transport system and thereby initialise nuclear import, is required. Attachment of an NLS peptide to DNA should, therefore, improve the efficiency of nonviral gene delivery systems [2]. In this study, an NLS peptide derived from Simian Virus 40 (SV40) was first coupled to BSA to test its nuclear import capability. Subsequently, it was coupled to linear DNA to test its ability to enhance transfection efficiency [3]. Experiments with these constructs complexed with cationic polymers were conducted in several cell types.

Fig. 1. The SV40-derived NLS peptide facilitated nuclear import of BSA-Texas Red.


Poster Abstracts

Experimental methods The nuclear import capacity of SV40 derived NLS peptide (PKKKRKVEDPYC), coupled to BSA-Texas Red (Molecular probes) with MBS (Pierce), was determined using digitonin-permeabilized COS-7 cells [4]. Plasmid DNA, linear DNA and linear DNA with an NLS peptide were obtained from Mologen (Berlin, Germany). COS-7, NIH/3T3 and MDCK-42 cells were incubated with pDMAEMA/DNA polyplexes (N/P 5) for 1 h and were subsequently incubated at 37 8C for 24 h. Luciferase expression was measured and normalized for total protein content. Results and discussion The SV40-derived NLS peptide facilitated the transport of BSA-Texas Red into the nucleus (Fig. 1). This nuclear localisation signal was, therefore, coupled to DNA. The transfection efficiency of the linear DNA construct in comparison with constructs without NLS and plasmid DNA was determined using COS-7, NIH/3T3 (proliferating) and MDCK-42 cells (nonproliferating). In COS-7 and NIH/3T3 cells the NLS peptide had no influence on transfection efficiency (Fig. 2).

Fig. 2. Transfection of Cos-7 (A) and NIH-3T3 (B) cells using various DNA constructs complexed with pDMAEMA.

In nondividing MDCK-42 cells, the NLS peptide slightly increased luciferase expression when compared to linear DNA (Fig. 3).

Fig. 3. Transfection of proliferating and nonproliferating MDCK-42 cell using various DNA constructs complexed with pDMAEMA.

Poster Abstracts


Conclusion Attachment of a nuclear localisation signal to DNA did not enhance transfection efficiency of cationic polymers. Whether this is due to poor binding of NLS to its receptor, the large size of the DNA or the delivery of the DNA at the NPC is still under investigation. References [1] C.W. Pouton, L.W. Seymour, Key issues in non-viral gene delivery, Adv. Drug Deliv. Rev. 46 (2001) 187–203. [2] R. Cartier, R. Reszka, Utilization of synthetic peptides containing nuclear localization signals for nonviral gene transfer systems, Gene Ther. 9 (2002) 157–167. [3] M.A. Zanta, P. Belguise-Valladier, J.P. Behr, Gene delivery: a single nuclear localization signal peptide is sufficient to carry DNA to the cell nucleus, Proc. Natl. Acad. Sci. U. S. A. 96 (1999) 91–96. [4] S.A. Adam, R. Sterne-Marr, L. Gerace, In vitro nuclear protein import using permeabilized mammalian cells, Methods Cell Biol. 35 (1991) 469-482.

ANTISENSE PEPTIDE NUCLEIC ACID DELIVERED BY CORE-SHELL MICROSPHERES L. Chiarantini1, A. Cerasi1, E. Millo2, U. Benatti3, M. Laus4, K. Sparnacci4, L. Magnani5, L. Tondelli5 1 Istituto di Chimica Biologica bG. FornainiQ, Universita` degli Studi di Urbino bCarlo BoQ, Italy 2 Istituto bGiannina GasliniQ, Genova, Italy 3 Dipartimento di Medicina Sperimentale, Universita` degli Studi di Genova, Italy 4 Dip. di Scienze e Tecnologie Avanzate, Universita` degli Studi del Piemonte Orie¨ntale, Italy 5 Istituto ISOF, Consiglio Nazionale delle Ricerche Bologna, Italy Summary Peptide Nucleic Acids (PNAs) are synthetic oligonucleotide analogues with a pseudopeptide backbone suitable for both antisense and antigene technology. However, the in vivo application has been hampered by their low permeability across cellular membranes. We here describe how specifically designed polymeric core-shell microspheres can be employed to increase PNA cellular uptake in order to be used as antisense drugs in murine macrophages. Introduction Peptide nucleic acids (PNAs) are DNA analogues with a pseudopeptide backbone formed by N-(2-amino-ethyl)glycine units to which nucleobases are linked through methylene carbonyl linkers. Despite the radical difference in the backbone chemical composition, PNAs improve the hybridisation characteristics of both DNA and RNA, are not degraded by nucleases, peptidases and other cellular enzymes, are easy to synthesize and to modify without impairing their binding characteristic to nucleic acids [1]. All these favourable properties make them suitable for both antisense and antigene technology. Beside their enzymatic stability, PNAs display low permeability across the cell membrane [2] .Thus cell targeting and the delivery of PNA to tissues need to be improved, i.e., by means of polymeric microspheres. When dispersion polymerisation of methyl methacrylate is run in the presence of the polymeric stabilizer Eudragit L100-55, core-shell microspheres can be obtained [3], whose core is mainly constituted of poly(methylmethacrylate) and with a shell of hydrophilic chains bearing carboxylic groups. Adsorption and release of native (ODN) and modified oligonucleotides (PNA) were run in physiological relevant buffers, leading to the formation of oligonucleotide-microsphere complexes (Fig. 1).

Fig. 1. Schematic representation of PNA-microsphere complex formation.


Poster Abstracts

Cytotoxicity, cellular uptake and biological activity of antisense PNA adsorbed onto a sample of core-shell microspheres were investigated on a murine peritoneal macrophage model. Results and discussion Core-shell microspheres H1D were prepared as already described [3] (Fig. 2). A 16-mer PNA (NH2-TTTTCTTCACGTTGTT-Lys-CONH2), complementary to the region 260–276 and a 14-mer PNA (NH2-CCTTTTCCTCTTTC-Lys-CONH2) complementary to the region 238–251 of iNOS mouse cDNA (GenBank accession number M84373) were synthesized, together with the unmodified 16-mer antisense sequence (AS ODN) and its W&C complementary sense sequence (SN ODN). For adsorption experiments, increasing amounts of PNA and ODN were added to a microsphere suspension at pH 7.4 (20 mM phosphate buffer) and the amount of unbound ODN was determined by UV absorbance at 260 nm. H1D microspheres were able to immobilize high amounts of PNAs. Moreover, when equimolar concentrations of SN ODN were added to the preformed PNA-microsphere complexes, hybrid PNA–DNA formation occurred with good efficiency, thus demonstrating that the binding to the microsphere surface does not affect the physicochemical properties of the PNA itself. Instead, when given in the same concentration range, H1D microspheres failed to adsorb native ODN, thus indicating that the adsorption mechanism is specific for PNAs. Extensive PNA release occurred in the presence of high salt concentration at physiological pH. In in vitro experiments, H1D microspheres showed no toxicity and were easily phagocytosed by murine macrophages. The antisense efficacy of PNAs was studied on inducible nitric oxide synthase (iNOS) in a murine peritoneal macrophage model activated by lipopolysaccharide (LPS).

Fig. 2. SEM micrograph of H1D core-shell microspheres. Conclusion Anti-iNOS PNA complexes with core-shell microspheres can be obtained easily in physiological buffer, with different degrees of surface coverage. The absence of cellular toxicity and the extent of macrophage uptake indicate that polymeric core-shell microspheres are able to express the antisense activity of PNA within the macrophages inhibiting the iNOS expression. Therefore, core-shell microspheres represent a potentially effective delivery system for therapeutic PNA treatments. Acknowledgements CIB 2002 project and MIUR PRIN 2003. References [1] P.E. Nielsen et al., Science 254 (1991) 1497–1500. [2] J. Wittung et al., FEBS Lett. 365 (1995) 27–29. [3] United Kingdom Patent application N.0325624.5, 2003.

Poster Abstracts


GENERATION OF LTB-BASED MUCOSAL VACCINES AND PRODUCTION IN PLANTS D.E.A. Florack, C.I. Matos, G.J.A. Rouwendal, D. Bosch Plant Research International, P.O. Box 16, 6700 AA Wageningen, The Netherlands Summary A number of recombinant B subunit-fusion proteins comprising either small and linear B and/or T cell epitope sequences, a large structural Nglycosylated viral surface antigen and two reporter enzymatic entities were expressed in potato tubers. All could be functionally expressed as determined by a GM1-ELISA but actual levels varied significantly and decreased strongly upon increased complexity. Introduction Escherichia coli heat-labile LT enterotoxin elicits severe diarrhoea in humans and a number of animals. LT is a protein-based toxin and composed out of a pentameric ring of identical B subunits (LTB) and one toxic A subunit (LTA) causing ADP-ribosylation (see, e.g., Ref. [10]). The five B subunits (LTB5) are nontoxic and are only responsible for binding of the holotoxin to ganglioside receptors such as GM1 found ubiquitously on cell membranes of eukaryotic cells. Binding to GM1 turns LTB into a powerful carrier molecule for delivery of proteins and especially antigens to mucosal surfaces and a number of recombinant LTB-antigen fusion proteins have been successfully employed as such [2,6]. Edible plants or plant parts might be suitable for delivery of LTB-based oral vaccines (for review, see Refs. [7,9]). By now, LTB has been successfully produced in various plants [4,8,11]. Most appeared to be immunogenic upon oral delivery amongst others to humans (Tacket et al., 1998). We have successfully produced LTB in potato tubers by expressing a synthetic gene optimised for expression in plants and under control of a tuber-specific promoter [8]. Here, we report on the expression of five LTB-based fusion proteins in potato to evaluate the potency of using potato and LTB-antigen fusion proteins for oral vaccination. We constructed genes coding for fusion proteins comprising LTB and a small 20 amino acid canine parvo virus B cell epitope [1] LTB and two of these combined with two 10 amino acid linear influenza T cell epitopes [12], and LTB and a large structural classical swine fever virus-derived glycoprotein [5]. In addition, we expressed fusion proteins of LTB and the reporters beta-glucoronidase (Gus) and green fluorescent protein (GFP) in potato which might be useful as so-called reporter vaccines. Experimental methods Design and construction of a synthetic gene for expression of LTB in potato (pLANTIGEN4), transformation using Agrobacterium tumefaciens, protein extraction and GM1 ELISA have been described previously [8]. A schematic overview of the five new gene constructs used for expression of various LTB-based proteins is given in Fig. 1. All gene constructs were placed under control of the tuber-specific patatin promoter and at least 20 independent transgenic potato plants were regenerated. Tubers were analysed for accumulation of GM1-binding entities as described.

Fig. 1. Schematic drawing LTB-based proteins.

Results and discussion Most of the tubers analysed contained GM1-binding recLTB or recLTB-fusion proteins as apparent from GM1-ELISA. However, accumulation levels greatly varied. The mean accumulation level in plants harbouring pLANTIGEN4 is 102 nM; pLANTIGEN12, 60 nM; pLANTIGEN15, 25 nM; pLANTIGEN20, 13 nM; pLANTIGEN13, 1.5 nM and pLANTIGEN19, 0.15 nM. The order in which decreased


Poster Abstracts

levels are observed reflects a combination of size and complexity. For example, highest levels are observed for pLANTIGEN4 which codes for LTB only, as expected. Addition of a small and linear epitope as in pLANTIGEN12 or four linear epitopes as in pLANTIGEN15 affect accumulation levels significantly (40% and 75% reduction, respectively) as does addition of GFP (pLANTIGEN20) which is much larger protein having relatively simple structure and is active as a monomer (85% reduction). However, addition of a large structural glycoprotein as in pLANTIGEN13, which has complex dimeric structure, a number of disulphide bonds and several N-linked glycans, affects accumulation levels much more (nearly 100-fold). Largest decrease is observed for pLANTIGEN19 which is a fusion protein of LTB and Gus. The latter is biologically active as a tetramer and the expected calculated molecalar weight for the complex exceeds 400 kDa compared to only 60 kDa for pLANTIGEN4. Conclusion LTB and LTB-based fusion proteins can be successfully produced in potato tubers. This opens novel ways for oral immunization. Size and complexity of the added fusion protein however greatly affect expression levels which drastically decrease upon increased complexity. This most likely is due to sterical hindrance of the fusion protein partner that has to compete with pentamerization of the LTB part of the molecule. Pentamerization is essential for binding to its natural receptor GM1. Hence, the use of LTB-based vaccines following this strategy is limited to smaller and less complex fusion proteins. Future experiments include analysis of immunogenicity upon oral delivery as such (potato material) or otherwise. Currently, we are developing a methodology that enables the use of LTB-based fusion proteins to much larger molecules further extending the applicability of this technology for targeted delivery. References [1] J.I. Casal, J.P.M. Langeveld, E. Corte´s, W.W.M. Schaaper, E. van Dijk, C. Vela, S. Kamstrup, R.H. Meloen, Peptide vaccine against canine parvovirus: identification of two neutralization subsites in the N terminus of VP2 and optimization of the amino acid sequence, Journal of Virology 69 (1995) 7274–7277. [2] M.T. Dertzbaugh, C. Elson, Comparative effectiveness of the cholera toxin B subunit and alkaline phosphatase as carriers for oral vaccines, Infection and Immunity 61 (1993) 48–55. [3] C.J. Hackett, J.L. Hurwitz, B. Dietzschold, W. Gerhard, Synthetic decapeptide of influenza virus hemagglutinin elicits helper T cells with the same fine recognition specificities as occur in response to whole virus, Journal of Immunology 135 (1985) 1391–1394. [4] T.A. Haq, H.S. Mason, J.D. Clements, C.J. Arntzen, Oral immunization with a recombinant antigen produced in transgenic plants, Science 268 (1995) 714–716. [5] M.M. Hulst, D.F. Westra, G. Wensvoort, R.J.M. Moormann, Glycoprotein E1 of hog cholera virus expressed in insect cells protects swine from hog cholera, Journal of Virology 67 (1993) 3542–5435. [6] E.K. Jagusztyn-Krynicka, J.E. Clark-Curtiss, R. Curtiss III, Escherichia coli heat-labile toxin subunit B fusions with Streptococcus sobrinus antigens expressed by Salmonella typhimurium oral vaccine strains: importance of the linker for antigenicity and biological activities of the hybrid protein, Infection and Immunity 61 (1993) 1004–1015. [7] W.H.R. Langridge, Edible vaccines, Scientific American Septembe (2000) 48–53. [8] T.G.M. Lauterslager, D.E.A. Florack, T.J. Van der Wal, J.W. Molthoff, J.P.M. Langeveld, D. Bosch, W.J.A. Boersma, L.A.Th. Hilgers, Oral immunisation of naRve and primed animals with transgenic potato tubers expressing LT-B, Vaccine 19 (2001) 2749–2755. [9] H.S. Mason, C.J. Arntzen, Transgenic plants as vaccine production systems, TIBTECH 13 (1995) 388–392. [10] B.D. Spangler, Struture and function of cholera toxin and the related Escherichia coli heat-labile enterotoxin, Microbiological Reviews 56 (1992) 622–647. [11] S.J. Streatfield, J.M. Jilka, E.E. Hood, D.D. Turner, M.R. Bailey, J.M. Mayor, S.L. Woodard, K.K. Beifuss, M.E. Horn, D.E. Delaney, I.R. Tizard, J.A. Howard, Plant-based vaccines: unique advantages, Vaccine 19 (2001) 2742–2748. [12] C.O. Tacket, H.S. Mason, A review of oral vaccination with transgenic vegetables, Microbes and Infection 1 (1999) 777–783. Acknowledgements All experiments with genetically modified plants were performed under the auspices of the Dutch Committee for Genetically Modified Organisms according to Dutch law and European guidelines 90/219/EC and 90/220/EC. This work was financed by grants from the DLO Foundation, the Dutch Ministry of Agriculture, Nature and Food Quality and European Union Framework 5 grant, QLRT-2000-01288 (FISHOV;

Poster Abstracts


IN-VIVO DELIVERY OF DNA AND PROTEIN USING CONCEPTUALLY NEW CATIONIC, SUNFISH’, AMPHIPHILES M. Haas, S. Audouy, I. Muizebelt, J. Sˇmisterova´, J.B.F.N. Engberts, D. Hoekstra, G. Storm, R. Hulst BioMaDe, Nijenborgh 4, 9747 AG Groningen, The Netherlands Summary Sunfish’ are suitable cationic amphiphiles for the in vivo delivery of genes and proteins. With PEGylated Sunfish included, the circulation time of the protein complex is prolonged making the complex applicable for extra-hepatic targeting delivery. In principle, further improvement can be obtained by optimising the Sunfish structure. Introduction Because viruses have shown to be risky, the interest to search for efficient nonviral DNA-carriers, like cationic amphiphiles, for in vivo application is high. Proteins and peptides have inappropriate pharmacokinetic profiles for therapeutic application and may therefore also benefit from complexation with cationic amphiphiles. The conceptually new cationic, Sunfish’, amphiphiles that appeared highly efficient for in vitro DNA delivery, may be suitable. PEGylation of liposomes appeared a great success to prolong the circulation time of the liposomes making these drug vehicles suitable for site-specific delivery. PEGylation of Sunfish’ amphiphiles may be beneficial for in-vivo delivery of DNA and protein. In the present study, the application of the cationic Sunfish amphiphiles for in vivo delivery of DNA and protein is studied. Besides, the effect of PEGylation on the circulation time of protein/Sunfish complexes is determined. Experimental methods DNA study: BALB/c mice were administered intravenously with luciferase-DNA complexed with different Sunfish’ (SF) or the commercial available DOTAP (450 nmol SF or DOTAP +30 Ag DNA/mouse). Luciferase activity in the organs was determined after 24 h. Protein study: Balb/c mice were administered intravenously with 125I-TC and 131I labelled myoglobulin (MW 17.8 kDa, pI 7.3) complexed with Sunfish SF-26 or SF-30 mixed with co-lipid cholesterol or DOPE (75 nmol SF:co-lipid 1:1 +3.5 Ag myoglobulin). Body distribution of radioactivity was determined after different time-intervals.

Results DNA study: Intravenous administration of DNA-Sunfish complexes in mice resulted in gene expression in the lung. The efficiency of geneexpression varied between the different Sunfish’. With the most efficient Sunfish (SF-46), a five times higher transfection was obtained than with the commercially available DOTAP. Protein study: Intravenous administration of the model-protein myoglobin complexed with the Sunfish/neutral phospholipid preparations resulted in a shift from myoglobin uptake in the kidney to uptake in the liver of the mouse. The liver uptake was higher with Sunfish SF-26 than with SF-30. Compared to DOPE, co-lipid cholesterol resulted in a higher liver uptake and a slower protein degradation in the liver. Tentatively, it might suggest that SF-26 in combination with co-lipid cholesterol is favourable for protein delivery since it is a better carrier and the lifetime of the protein in the cell is prolonged. A prolongation of the circulation time of myoglobin was obtained by including 8% but especially 20% PEGylated Sunfish’ in the complex.


Poster Abstracts

STUDIES ON THE INTRACELLULAR RELEASE OF GENETIC DRUGS FROM PHARMACEUTICAL CARRIERS B. Lucas*, K. Remaut*, S.C. De Smedt, J. Demeester Laboratory for General Biochemistry and Physical Pharmacy, Ghent University, Harelbekestraat 72, 9000 Ghent, Belgium (* both first authors) Introduction A critical step in the delivery of genetic drugs is the dissociation of the DNA from its pharmaceutical carrier at the right place: a critical balance between dbeing associated extracellularlyT and dbeing dissociated intracellularlyT needs to be maintained. To reveal how DNA/cationic carrier complexes physicochemically behave in cells, dual color fluorescence fluctuation spectroscopy (dual color FFS) is a promising tool. Our group has recently proved the usefulness of dual color FFS in studying the association/dissociation behavior of oligonucleotides and cationic polymers in buffer solution [1,2]. As can be seen in Fig. 1, FFS monitors the fluorescence fluctuations caused by the diffusion of fluorescently labeled molecules through the excitation volume of a microscope. In dual color FFS both the DNA and the cationic carrier are labeled with, spectrally different, fluorescent markers. Dual color FFS measures the emission light of both dyes by two separate detectors. These detectors monitor the fluorescence fluctuations caused by the diffusion of the DNA and the cationic carrier through the same excitation volume. Dual color FFS answers the question whether the DNA and the cationic carrier move together (i.e., when they are associated) or separately (i.e., when they are dissociated) through the excitation volume. In addition, diffusion characteristics and concentration of the fluorescent species can be derived from the registered fluorescence fluctuation data.

Fig. 1. Schematic setup of the FFS instrument. The excitation light from the laser is reflected (by a triple chroic mirror) into the objective lens of the microscope. The light emitted by the fluorescent molecules in the excitation volume passes through the objective lens and the triple chroic mirror, to be split by the subsequent dichroic mirror into a red and a green component. To study the dissociation of DNA/cationic carrier complexes in the cytoplasm, microinjection of the complexes is a helpful tool as it bypasses extracellular effects, endocytosis and endosomal escape. It allows focusing directly on the fate of these DNA-complexes in the cytoplasm. In the present study, the dissociation of (green) rhodamine green-labeled oligonucleotides (RhGr-ONs) from (red) Cy5 labeled poly-lLysine (Cy5-pLL) is considered. Dissociation of the pLL/ON complexes is induced by adding dextran sulfate, an anionic polymer that competes with the ONs for binding to the pLL chains. In this abstract, we describe and interpret our observations of both the microinjection and FFS measurements on the pLL/ON complexes. Experimental methods Rhodamine green (RhGr) was used as the fluorescent marker of the (20-mer) ONs. The fluorescent labeling occurred at the 5V end of the ONs. Each ON contained one label. Poly-l-Lysine (30 300 g/mol) was labeled with FluoroLinkTMCy5 monofunctional dye and purified on a G25Sephadex column (10100 mm). It was calculated that, on the average, each pLL chain bore one Cy5-label.

Poster Abstracts


The ONs were complexed with pLL at a charge ratio (+/) of 20. A 20 mM HEPES buffer at pH 7.4 was used. The ONs concentration equaled 10 Ag/mL for CLSM- and microinjection experiments and 0.2 Ag/mL for FFS experiments. Dextran sulfate (DS, 500,000 g/mol) was used to destabilize the complexes. Results and discussion Injecting free RhGr-ONs in the cytoplasm of cells results in an instant brightening of the nucleus (Fig. 2A), suggesting the accumulation of RhGr-ONs in the nucleus. Upon injection of pLL/RhGr-ON complexes in the cytoplasm, the fluorescence remains located at the site of injection (Fig. 2B), indicating that the RhGr-ONs remain trapped in the complexes. Upon incubating these cells for a longer time, the fluorescent spot in the cytoplasm dexpandedT while the nucleus turned homogeneously fluorescent, indicating that (at least part of) the RhGr-ONs were released from the complexes (Fig. 2C).

Fig. 2. Fluorescence images of Vero cells injected in the cytoplasm with free RhGr-ON (A) and pLL/RhGr-ON complexes directly after microinjection (B) and after 15V of incubation (C). Injecting Cy5-pLL/RhGr-ONs complexes in the cytoplasm resulted in co-localized fluorescent spots in the cytoplasm. After 10 min, however, also co-localization could be observed in the nuclei, indicating that both RhGr-ONs and Cy5-pLL strands were able to enter the nucleus (Fig. 3). Although both red and green fluorescence were more homogeneously spread in the nucleus compared to the cytoplasm, CLSM images could not reveal whether the nuclear fluorescence is due to intact Cy5-pLL/RhGr-ONs complexes or not, as co-localization of the fluorescent markers cannot rule out the possibility of the fluorescent labeled molecules being co-localized without being associated. To answer this question, we performed dual color FFS in the cytoplasm and in the nucleus of cells transfected with Cy-pLL/RhGr-ON complexes.

Fig. 3. Vero cells 10 min after cytoplasmatic microinjection of Cy5-pLL/RhGr-ONs complexes. (A) Transmission image, (B) green detector image, (C) red detector image.

Fig. 4A shows the fluorescence fluctuations as measured by dual color FFS on Cy-pLL/RhGr-ON complexes in buffer solution. It is especially clear that highly intense fluorecence peaks are registered by the two detectors at the same time. This means that, at certain times, a complex existing of several RhGr-ON strands and several Cy5-pLL chains moves through the excitation volume. As Fig. 4B shows, if these complexes were destabilized with dextran sulfate, only peaks of high fluorescence intensity were observed by the red detector, suggesting the binding of many Cy5-pLL strands to dextran sulfate. From the fluctuations in the green detection channel (Fig. 4B), a diffusion coefficient corresponding to the diffusion coefficient of free RhGr-ONs was calculated confirming that the RhGr-ONs were released from their carrier.


Poster Abstracts

Fig. 4. Fluorescence fluctuation profiles as simultaneously registered by the dredT and dgreenT detector in a Cy5-pLL/RhGr-ON dispersion, respectively, before (A) and after (B) adding dextran sulfate to the dispersion.

Fig. 5 shows FFS measurements in the cytoplasm of a cell after 30 min of incubation with Cy5-pLL/RhGr-ON complexes. Highly fluorescence peaks were simultaneously registered in both detectors, indicating complexes existing of many RhGr-ONs bound to many Cy5-pLL chains were present. However, after a longer incubation time (3 h), simultaneously occurring peaks were no longer observed. Instead, free moving Cy5-pLL and RhGr-ONs were observed in the nucleus, while peaks of high fluorescence intensity were only registered by the red detector in the cytoplasm (data not shown). The absence of simultaneous occurring peaks of high fluorescence intensity in the nucleus indicates that both Cy5-pLL and RhGr-ONs are present in the nucleus (as shown by CLSM measurements, Fig. 3), without being associated.

Fig. 5. Fluorescence fluctuation profiles as simultaneously registered by the dredT and dgreenT detector in the cytoplasm of a cell transfected with Cy5-pLL/RhGr-ON complexes. Injecting a mixture of RhGr-ONs and Cy5-pLL/dextran sulfate complexes in the cytoplasm immediately resulted in green fluorescent nuclei (nuclear accumulation of RhGr-ONs). However, red fluorescence could only be observed in the cytoplasm (Fig. 6). This proves that all the Cy5-pLL is bound to dextran sulfate and cannot enter the nucleus because of the high molecular weight of dextran sulfate (500 kDa). This also rules out the possibility of the red nuclear fluorescence in Fig. 3C being due to free Cy5-label accumulating in the nucleus.

Poster Abstracts


Fig. 6. Vero cells 10 min after cytoplasmatic microinjection of a mixture of RhGr-ONs and Cy5-pLL/dextran sulfate complexes. (A) transmission image, (B and C) CLSM images with simultaneous 488 nm and 647 nm excitation (B: green detector, C: red detector).

Conclusion Where the nuclear co-localization of red and green fluorescence in the CLSM images of Fig. 3 demonstrated the presence of both Cy5pLL and RhGr-ONs in the nucleus, dual color FFS measurements revealed that no intact Cy5-pLL/RhGr-ONs complexes were present in the nucleus. Compared with confocal microscopy, in which the spatial analysis of fluorescently labeled molecules is considered, FFS, which studies the movement of these molecules, may be a more robust method to obtain information on the cellular pathway of DNA complexes. We have shown that dual color FFS will be able to answer the question whether ONs will be released from their carrier intracellularly. References [1] B. Lucas, E. Van Rompaey, S.C. De Smedt, P. Van Oostveldt, J. Demeester, Macromolecules 35 (2002), in press. [2] B. Lucas, K. Remaut, K. Braeckmans, J. Haustraete, S.C. De Smedt, J. Demeester Submitted. Macromolecules; manuscript nr: MA035780L.


Poster Abstracts

STRUCTURALLY WELL-DEFINED COPOLYMERS OF POLY(ETHYLENE GLYCOL) AND LOW MOLECULAR WEIGHT LINEAR POLYETHYLENIMINE AS VECTORS FOR GENE DELIVERY Z. Zhong1, M.C. Lok2, P.J. Dijkstra1, W.E. Hennink2, J. Feijen1 1 Institute for Biomedical Technology (BMTI), Polymer Chemistry and Biomaterials, Faculty of Science and Technology, University of Twente, P.O.Box 217, 7500 AE Enschede, The Netherlands 2 Department of Pharmaceutics, Utrecht Institute for Pharmaceutical Science (UIPS), Utrecht University, P.O. Box 80082, 3508 TB, Utrecht, The Netherlands Summary Structurally well-defined ABA triblock copolymers with low molecular weight linear polyethylenimine (PEI) as A block and poly(ethylene glycol) (PEG) as B block have been prepared and used for gene delivery. PEI–PEG–PEI 2100–3400–2100 can effectively condense DNA, giving small-sized polyplexes (b100 nm) with reduced zeta-potentials (b10 mV). Initial experiments show that these polyplexes have a better transfection efficiency and a decreased cytotoxicity compared to polyplexes based on low molecular weight linear PEI polymers.

Introduction Gene therapy holds great promise for treating life-threatening diseases like cancer [1,2]. In the past decade, nonviral vectors especially cationic polymeric systems have attracted growing interest since they offer many advantages over the viral counterparts such as ease of production, low immunogenicity, and great flexibility with regard to vector modification and DNA incorporation. Poly(l-lysine) (PLL), polyethylenimine (PEI), poly(2-(dimethylamino)ethyl methacrylate) (PDMAEMA), and polyamidoamine (PAMAM) dendrimers are among the most widely studied polymers for gene delivery [3]. All these polymers have nevertheless shown in vivo toxicity, which has become one of the major limiting factors for their advance to the clinical uses. For example, PEI polymers were regarded as promising nonviral gene carriers due to their superior transfection efficiency, the in vivo administration of PEI/DNA complexes into mice was, however, reported to give severe toxicity [4]. The aim of this study was to design safer polymeric gene delivery systems, for which copolymers based on poly(ethylene glycol) (PEG) and low molecular weight linear PEI were synthesized. The preliminary in vitro experiments show that the polyplexes of such copolymer have a low toxicity and are able to transfect COS-7 cells. Experimental methods Polymer Synthesis PEI-PEG-PEI triblock copolymers were prepared from cationic polymerization of 2-methyl-2-oxazoline (MeOZO) using PEG-bis(tosylate) as a macroinitiator followed by hydrolysis. The polymerization was conducted in CH3CN at 70 8C for 3 days, which led to a quantitative yield of PMeOZO–PEG–PMeOZO copolymers with prescribed compositions. The hydrolysis of PMeOZO–PEG–PMeOZO was carried out in 10 wt.% HCl solutions overnight at 100 8C. A low molecular weight linear PEI polymer has been synthesized in an analogous way except using methyl tosylate instead of PEG-bis(tosylate). In vitro transfection and cell viability assay Transfection experiments were performed with COS-7 cells by using the plasmid pCMV-LacZ as reporter gene as reported previously for PDMAEMA [5]. Two parallel transfection series, one for the determination of reporter gene expression (h-galactosidase) using ONPG assay and the other for the evaluation of cell viability by XTT assay, were carried out in separate 96-well plates. PDMAEMA/DNA complex with a ratio of 3:1 (w/w) was used as a reference for gene transfection efficiency.

Results and discussion ABA triblock copolymers with low molecular weight linear polyethylenimine (PEI) as A block and poly(ethylene glycol) (PEG) as B block were synthesized by cationic polymerization of 2-methyl-2-oxazoline using PEG-bis(tosylate) macroinitiator followed by acid hydrolysis (scheme 1). NMR and GPC results showed that all copolymers have controlled compositions and molecular weights. An initial in vitro transfection experiment revealed that low molecular weight linear PEI polymer with an Mn=100 (denoted as PEI 2100) has low toxicity and is not able to transfer DNA into COS-7 cells, despite the fact that PEI 2100 can complex DNA into small-sized

Poster Abstracts


Scheme 1. Synthesis of polyethylenimine-b-poly(ethylene glycol)-b-polyethylenimine (PEI–PEG–PEI) triblock copolymers. nanoparticles (80~100 nm) when PEI 2100/DNA ratio is above 1:1 (w/w). In contrast, the triblock copolymer, PEI–PEG–PEI with Mn (PEI)=2100 and Mn (PEG)=3400 (denoted as PEI–PEG–PEI 2100–3400–2100), displayed marked transfection activity. The transfection efficiency increased with increasing polymer/DNA ratios, in which a transfection efficiency of 25% relative to PDMAEMA was achieved at a polymer/DNA ratio of 96:1 (w/w) (Fig. 1). The addition of endosomal escape peptide INF-7 did not improve the transfection efficiency, indicating that the PEI–PEG–PEI triblock copolymers are well capable of disrupting endosomes. Importantly, this triblock copolymer has a substantial lower cytotoxicity than low molecular weight PEI polymer (Fig. 2). Biophysical characterizations revealed that PEI–PEG–PEI 2100–3400–2100 effectively condenses DNA at a polymer/DNA ratio of 3:1 (w/w), giving polyplexes with sizes less than 100 nm and reduced zeta-potentials (b10 mV). The low zeta-potential of these polyplexes is likely due to the shielding of the surface charge of the polyplexes by PEG chains and this may account for the low toxicity of the triblock copolymer. It should be noted that in the presence of serum, both PEI 2100 and PEI–PEG–PEI 2100–3400–2100 are essentially nontoxic at polymer/DNA ratios up to 96:1 (w/w). Although the transfection efficiency of PEI–PEG–PEI 2100–3400–2100 is low compared to high molecular weight PEI and PDMAEMA, this triblock copolymer has the apparent advantage of its low toxicity. These first results have demonstrated that it is possible to markedly enhance the transfection efficiency and meanwhile decrease cytotoxicity of low molecular weight linear PEI by conjugating to PEG. Currently, copolymers with different macromolecular architectures like star-shaped and multi-block copolymers of PEG and low molecular weight linear PEI are under investigation. The relationship between polymer structures and their biological properties, particularly transfection activity and cytotoxicity, will be studied.

Fig. 1. Transfection activity of PEI–PEG–PEI 2100–3400–2100/ plasmid DNA Complexes in COS-7 cells as a function of polymer/ DNA ratios. Transfection data are relative to PDMAEMA/plasmid DNA at 3:1 ratio (w/w).

Fig. 2. The viability of COS-7 cells incubated with polyplexes prepared at different polymer/plasmid ratios measured by the XTT assay.

Conclusion We have demonstrated that copolymers based on PEG and low molecular weight linear PEI can efficiently condense DNA into nanoparticles. These polyplexes have markedly improved transfection efficiency and reduced cytotoxicity as compared to the polyplexes of low molecular weight linear PEI polymers. In the future, the macromolecular structure and the molecular weights of the different blocks of the copolymers will be varied to achieve optimal transfection efficiency, which may eventually lead to promising vectors for gene delivery.


Poster Abstracts

References [1] J.D. Hood, M. Bednarski, R. Frausto, S. Guccione, R.A. Reisfeld, R. Xiang, D.A. Cheresh, Tumor regression by targeted gene delivery to the neovasculature, Science 296 (2002) 2404–2407. [2] T. Merdan, J. Kopecek, T. Kissel, Prospects for cationic polymers in gene and oligonucleotide therapy against cancer, Adv. Drug Deliv. Rev. 54 (2002) 715–758. [3] S. Han, R.I. Mahato, Y.K. Sung, S.W. Kim, Development of biomaterials for gene therapy, Molec. Ther. 2 (2000) 302–317. [4] P. Chollet, M.C. Favrot, A. Hurbin, J.L. Coll, Side-effects of a systemic injection of linear polyethylenimine–DNA complexes, J. Gene Med. 4 (2002) 84–91. [5] J.Y. Cherng, P. van de Wetering, H. Talsma, D.J.A. Crommelin, W.E. Hennink, Effect of size and serum proteins on transfection efficiency of poly((2-dimethylamino)ethyl methacrylate)-plasmid nanoparticles, Pharm. Res. 13 (1996) 1038–1042.