Intracellular siRNA delivery dynamics of integrin-targeted, PEGylated

Journal of Controlled Release 211 (2015) 1–9

Contents lists available at ScienceDirect

Journal of Controlled Release journal homepage: www.elsevier.com/locate/jconrel

Intracellular siRNA delivery dynamics of integrin-targeted, PEGylated chitosan–poly(ethylene imine) hybrid nanoparticles: A mechanistic insight Héloïse Ragelle a,1, Stefano Colombo b, Vincent Pourcelle c, Kevin Vanvarenberg a, Gaëlle Vandermeulen a, Caroline Bouzin d, Jacqueline Marchand-Brynaert c, Olivier Feron d, Camilla Foged b,⁎, Véronique Préat a,⁎⁎ a

Université Catholique de Louvain, Louvain Drug Research Institute, Advanced Drug Delivery and Biomaterials, 1200 Brussels, Belgium University of Copenhagen, Faculty of Health and Medical Sciences, Department of Pharmacy, Section for Biologics, Universitetsparken 2, 2100 Copenhagen O, Denmark c Université Catholique de Louvain, Molecules, Solids and Reactivity, Institute of Condensed Matter and Nanosciences, 1348 Louvain-la-Neuve, Belgium d Université Catholique de Louvain, Institut de Recherche Clinique et Expérimentale, Pole of Pharmacology and Therapeutics, 1200 Brussels, Belgium b

a r t i c l e

i n f o

Article history: Received 10 April 2015 Received in revised form 14 May 2015 Accepted 15 May 2015 Available online 16 May 2015 Keywords: αvβ3 Integrin Nanoparticles siRNA Stem-loop RT-qPCR RGD peptidomimetic Chitosan

a b s t r a c t Integrin-targeted nanoparticles are promising for the delivery of small interfering RNA (siRNA) to tumor cells or tumor endothelium in cancer therapy aiming at silencing genes essential for tumor growth. However, during the process of optimizing and realizing their full potential, it is pertinent to gain a basic mechanistic understanding of the bottlenecks existing for nanoparticle-mediated intracellular delivery. We designed αvβ3 integrin-targeted nanoparticles by coupling arginine–glycine–aspartate (RGD) or RGD peptidomimetic (RGDp) ligands to the surface of poly(ethylene glycol) (PEG) grafted chitosan–poly(ethylene imine) hybrid nanoparticles. The amount of intracellular siRNA delivered by αvβ3-targeted versus non-targeted nanoparticles was quantified in the human non-small cell lung carcinoma cell line H1299 expressing enhanced green fluorescent protein (EGFP) using a stem-loop reverse transcription quantitative polymerase chain reaction (RT-qPCR) approach. Data demonstrated that the internalization of αvβ3-targeted nanoparticles was highly dependent on the surface concentration of the ligand. Above a certain threshold concentration, the use of targeted nanoparticles provided a two-fold increase in the number of siRNA copies/cell, subsequently resulting in as much as 90% silencing of EGFP at well-tolerated carrier concentrations. In contrast, non-targeted nanoparticles mediated low levels of gene silencing, despite relatively high intracellular siRNA concentrations, indicating that these nanoparticles might end up in late endosomes or lysosomes without releasing their cargo to the cell cytoplasm. Thus, the silencing efficiency of the chitosan-based nanoparticles is strongly dependent on the uptake and the intracellular trafficking in H1299 EGFP cells, which is critical information towards a more complete understanding of the delivery mechanism that can facilitate the future design of efficient siRNA delivery systems. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The therapeutic potential of RNA interference (RNAi)-based medicine is fully dependent on the development of delivery systems that can efficiently overcome the numerous barriers existing for the intracellular delivery of the mediators of the RNAi process, e.g., the small interfering RNAs (siRNAs) [1]. A major direction in the translation of RNAi-based therapies towards clinical trials is the design of

⁎ Correspondence to: C. Foged, Department of Pharmacy, University of Copenhagen, Universitetsparken 2, DK-2100 Copenhagen O, Denmark. ⁎⁎ Correspondence to: V. Préat, Université Catholique de Louvain, Faculté de Pharmacie et Sciences Biomedicales, Louvain Drug Research Institute, Pharmaceutics and Drug Delivery, Avenue Mounier, 73 UCL B1 73.12, 1200 Brussels, Belgium. 1 Present address: David H. Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA.

http://dx.doi.org/10.1016/j.jconrel.2015.05.274 0168-3659/© 2015 Elsevier B.V. All rights reserved.

polysaccharide-based nanocarriers that can transport siRNA across these barriers and to the site of action in the cell cytoplasm [2]. Among the polysaccharides described in the literature for siRNA delivery, chitosan offers a number of advantages; (i) its polycationic nature allows for complexation and condensation of siRNA into nanoparticles via a fast, easy and gentle process [3,4]; (ii) chitosan is biodegradable and biocompatible, which is a crucial factor for in vivo administration [5,6]; and (iii) its versatile chemical structure enables facile chemical modification to equip the polymer with new properties and improve its performance [2,7]. However, the use of chitosan as a carrier for intracellular siRNA delivery is limited by low water solubility at pH values above 6.5 and poor colloidal stability in physiologically relevant media [8]. Moreover, the transfection efficiency mediated by chitosan is limited due to its poor buffering capacity and inability to mediate endosomal escape [9,10]. A number of chitosan derivatives and formulation improvements have recently been described to overcome

2

H. Ragelle et al. / Journal of Controlled Release 211 (2015) 1–9

these limitations [11]. Here, we used poly(ethylene glycol) (PEG) grafted chitosan (CPEG) to enhance the polymer solubility as well as the nanoparticle colloidal stability. Moreover, poly(ethylene imine) (PEI) was included in the formulation to improve the transfection efficiency of the nanoparticles, as previously described [8]. Finally, the nanoparticles were targeted to αvβ3 integrin receptors by conjugating the arginine–glycine–aspartate (RGD) peptide, or an RGD peptidomimetic (RGDp) that mimics the RGD motif, to the distal ends of the PEG chains. The αvβ3 integrins are upregulated in tumor cells and in angiogenic endothelial cells, as compared to healthy cells [12]. They therefore represent attractive targets for the delivery of siRNA in cancer therapy. Grafting of delivery systems with RGD or RGDp has been shown to increase the treatment efficacy on account of the tumor targeting properties of the ligands, as compared to their nontargeted counterparts [13,14]. Moreover, RGDp-grafted systems have a longer half-life and a higher bioavailability than RGD-grafted systems because of the improved chemical stability of the peptidomimetics [15]. However, RGDp-grafted systems have yet to be explored for siRNA delivery, and a direct comparison of their delivery efficiency to the RGD-grafted systems is of interest for potential clinical applications. Ultimately, the efficacy of nucleic acid-based therapies critically depends on the ability of the delivery system to carry the active compounds to their site of action, which for siRNA is the cell cytoplasm. Nevertheless, little knowledge is currently available about the intracellular delivery mechanisms of nucleic acid delivery systems [16]. Generally, the cellular uptake process is monitored by measuring the delivery of nucleic acids labeled with fluorophores by fluorescence microscopy or flow cytometry. These techniques are not fully quantitative, as the fluorophore is detected and not the active compound itself. Fully quantitative approaches based on reverse transcription quantitative polymerase chain reaction (RT-qPCR) have recently been developed to improve the understanding of the intracellular delivery dynamics of nucleic acid–based drugs. Among them, stem-loop qPCR was successfully applied to accurately quantify the amount of the full-length active small RNAs delivered and maintained in the cells [17,18]. In the presented work, we used a stem-loop primer optimized for the detection of a 27-mer duplex dicer substrate siRNA (DsiRNA) targeting enhanced green fluorescent protein (EGFP) for highly accurate and sensitive quantification of the amount of intracellular active DsiRNA molecules delivered by the integrin-targeted nanoparticles [19,20]. In the cell, DsiRNAs are first processed by the enzyme Dicer by nipping at the sequence ends, which subsequently facilitates incorporation of the diced antisense strand into the RNA-induced silencing complex (RISC) complex, increasing their potency, as compared to traditional siRNA, which are directly loaded into RISC [21]. The nanoparticles were investigated with regard to their delivery efficiency and dynamics using the H1299 EGFP cell model expressing the αvβ3 integrin receptor [20]. Furthermore, the parallel evaluation of EGFP silencing allowed for an elegant and direct correlation between the delivered intracellular DsiRNA dose and the effect. 2. Materials and methods 2.1. Materials Branched PEI (25 kDa) and sodium tripolyphosphate (TPP) were purchased from Sigma-Aldrich (Diegem, Belgium) and medical grade

chitosan (Kiomedine, 90 kDa, deacetylation degree 79.7%) was from Kitozyme (Belgium). The peptide GRGDS (named RGD, Fig. 1) was purchased from NeoMPS (Strasbourg, France). HO-PEG-OMe (5000 g/mol) was purchased from Sigma-Aldrich. (2S)-3-[3-{6-aminohexanamido}4-{3-(5,6,7,8-tetrahydro-[1,8]-naphthyridin-2-yl)propoxy}phenyl]-2[3(trifluoromethyl)phenylsulfonamido] propanoïc acid named RGD peptidomimetic (RGDp, Fig. 1) was synthesized and purified by previously reported pathways [22,23]. O-succinimidyl-4-(1-azi-2,2,2trifluoroethyl)benzoate (NHS-TPD clip) was prepared as described in [24]. The siRNA duplex directed against green fluorescent protein (GFP) and a control (CTRL) siRNA duplex, as well as the Dicer substrate 25/27-mer siRNA targeting GFP (DsiRNA GFP) were supplied by Eurogentec (Seraing, Belgium). The siRNA sequences and modification patterns are reported in Supplementary data Table S1. SiRNA labeled with Alexa647 at the 3′ end of the sense strand was purchased from Qiagen (Venlo, The Netherlands). DsiRNA GFP labeled with Alexa647 at the 5′ end of the antisense strand was purchased from Eurogentec. INTERFERin® was purchased from Polyplus transfection (Illkirch, France). Unless otherwise stated, the solutions were prepared using RNase-free water (Gibco, Invitrogen, Merelbeke, Belgium) and filtered through 0.22 μm filters (VWR, Leuven, Belgium). RNase free materials and conditions were carefully applied throughout all experiments.

2.2. Synthesis and characterization of PEG–RGD and PEG–RGDp PEG–RGD and PEG–RGDp (Fig. 1) were synthesized by using the clip photochemistry methodology, and purified and characterized as previously described [24] with slight modifications (see Supplementary data). Using this technique, RGD or RGDp were grafted to the PEG (Mn ~ 5000 Da) backbone allowing subsequent functionalization of the hydroxyl terminus.

2.3. Synthesis and characterization of the CPEG conjugates To synthesize chitosan–PEG (CPEG) conjugates (i.e., CPEG, CPEG RGD and CPEG RGDp), the hydroxyl groups of PEG, PEG–RGD and PEG–RGDp were first activated into imidazole carbonate groups by using N,N′carbonyldiimidazole (CDI). The CDI-activated PEG was subsequently reacted with the primary amines of chitosan. The characteristics of the CPEG conjugates are summarized in Table 1 (see Supplemetary data and Fig. S1 for more details about the methods and calculations of coupling efficiencies). The obtained CPEG conjugates were composed of PEGylated chitosan and free PEG conjugated with the targeting ligands. This is a result of the use of the acidic pH required for solubilizing chitosan, in which only small amounts of PEG (around 1–2 mol%) reacted with the chitosan amines: the majority of PEG–CDI underwent degradation into nonconjugated OH–PEG–OMe that remained closely associated with chitosan. It has been previously described that chitosan and PEG form strong non-covalent complexes that cannot be dissociated even by extensive dialysis [25], and that these sheddable PEGs can enhance the internalization of nanoparticles [26]. The obtained PEG grafting densities were similar to those found in other studies and are considered appropriate to confer stealth properties to the nanoparticles [27].

Fig. 1. Structure of (A) PEG–RGD and (B) PEG–RGDp.

H. Ragelle et al. / Journal of Controlled Release 211 (2015) 1–9 Table 1 Characteristics of the CPEG conjugates. Polymers

Ligand/PEG chain (mol%)

PEG/chitosan (mol%)c

Ligand/CPEG (mol%)d

CPEG CPEG RGD 2% CPEG RGDp 2% CPEG RGDp 4%

/ 10a 8b 21b

20 20 20 16

/ 2.0 1.6 3.4

a

Estimated from ratios of reagents used. b Calculated by 19F qNMR. c Molar ratio of the free deacetylated amines estimated by 1H NMR. d Molar ratio of the free deacetylated amines. For an example of calculation, see Supplementary data.

2.4. Formulation of siRNA-loaded nanoparticles The nanoparticles were prepared by using the ionic gelation method [3,12]. In brief, the negatively charged components, i.e. the DsiRNA or siRNA (50 μM) and TPP (1 mg/ml) were added to the positively charged components, i.e., CPEG, CPEG RGD or CPEG RGDp (1 mg/ml in 0.2 M sodium acetate buffer, pH 5.7) and PEI (1 mg/ml in RNase-free water, pH 7.4) and vortexed for 30 s. The mixtures were incubated for 1 h at room temperature. The ratio between CPEG, CPEG RGD or CPEG RGDp and PEI was 5/1 (w/w) for all formulations. The concentration of siRNA or DsiRNA in the formulations was kept constant and was 11.35 μg/ml. The amount of TPP was optimized towards monodisperse nanoparticle size distributions (results not shown). The copolymer/TPP ratio was between 2.7/1 and 3.3/1 (w/w). The size and the zeta potential of the nanoparticles were determined using a Nanosizer NanoZS (Malvern Instruments, Malvern, UK) by photon correlation spectroscopy and electrophoretic mobility, respectively. All samples were measured in triplicate in RNase-free water. 2.5. H1299 EGFP cell culture The human non-small cell lung carcinoma cell line H1299 expressing the destabilized enhanced GFP (H1299 EGFP, protein halflife ~ 2.5 h) was used for the in vitro tests as previously reported [28]. The cells were grown in a RPMI 1640 medium (Gibco, Belgium) supplemented with 10% (v/v) fetal bovine serum (FBS, Gibco). H1299 EGFP cells were characterized as αvβ3 integrin positive (see Supplementary data). 2.6. EGFP silencing of H1299 EGFP cells The H1299 EGFP cells were seeded in 12-well plates at a density of 10 cells per well 24 h prior to the experiments. On the following day, the cells were transfected with the nanoparticles loaded with DsiRNA GFP, siRNA GFP and siRNA CTRL, respectively, in a RPMI medium containing 10% (v/v) FBS for 4 h. After 24 and 48 h, respectively, the cells were washed with PBS, trypsinized and resuspended in 300 μl PBS prior to flow cytometric analysis using a Gallios flow cytometer (Beckman Coulter, Brea, CA, USA). Data was analyzed using the FlowJo 7.6.5 software (Three Star, Ashland, OR, USA). The percentage of gene silencing was calculated by using the FL-1 median (median of the GFP fluorescence histogram). The cells transfected with nanoparticles loaded with the siRNA CTRL were used as negative control. For each condition, three separate samples were analyzed. 5

3

immediately frozen on dry ice and stored at − 80 °C. RNA isolation and purification were performed as previously described [20]. The concentration and the purity of the samples were measured using the Nanodrop 2000c Spectrophotometer (Thermo Scientific, USA). If the RNA quality was not acceptable (A260/A280 b 1.80 and A260/ A230 b 1.75), the samples were diluted to a final volume of 200 μl and re-precipitated in 400 μl of ethanol and 60 μl of 3 M sodium acetate. 2.8. Stem-loop RT-qPCR The sequences of the stem-loop reverse transcription primers and of the PCR primers are described in [20]. Briefly, a purified RNA amount of 700 ng was reverse transcribed in a total reaction volume of 20 μl that also included 500 μM deoxynucleotide mix, 20 U Protector RNase Inhibitor, 10 U Transcriptor Reverse Transcriptase, 1× Transcriptor Reverse Transcriptase buffer (all from Roche, Basel, Switzerland). The RNA template was first denatured at 75 °C for 10 min and immediately cooled on ice for 2 min. The reaction mix and the stem-loop primers (11 nM of each, final concentration) were then added during the cooling phase. The pulsed RT program consisted of 15 min at 14 °C, 10 min at 42 °C, followed by 25 cycles (15 s at 14 °C, 10 s at 42 °C and 15 s at 65 °C). Subsequently, the mixture was incubated for 5 min at 85 °C for transcriptase inactivation and cooled at 4 °C. The qPCR was performed with a LightCycler 480® (Roche) using the SYBR Green Master mix according to the manufacturer's instructions. An amount of 5 ng DNA was analyzed. The housekeeping gene snoRNA U109 (GenBank ID: AM055742. 1) was used for normalization. The PCR program was 95 °C for 5 min, 37 cycles (95 °C for 15 s, 62 °C for 15 s, 72 °C for 1 s) followed by cooling at 4 °C. The PCR data analyses were performed as described in [20]. 2.9. Uptake of the siRNA-loaded nanoparticles by H1299 EGFP cells The H1299 EGFP cells were seeded in 12-well plates (105 cells/well) containing a cover slip. After 24 h, the cells were incubated for 1 h with the nanoparticles loaded with Alexa647-labeled siRNA (100 nM), fixed with fresh paraformaldehyde [2% (w/v) in PBS] and rinsed three times with PBS. Cells were incubated with 4′,6-diamidino-2-phenylindole (DAPI, 0.1 μg/ml, 5 min) to stain nuclei, rinsed three times with PBS and incubated for 1 h with AlexaFluor488-conjugated concanavalin A (5 μg/ml in PBS, Molecular Probes) to stain the cell membranes. After washing, the cover slips were placed on a slide using the Vectashield® mounting medium. The slides were imaged using a structured illumination AxioImager microscope equipped with an Apotome module (Zeiss, Jena, Germany, 40× magnification). Cellular uptake was quantified by flow cytometry using the same conditions as above. 2.10. Effect of siRNA-loaded nanoparticles on cell viability An MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) test (Sigma, Belgium) was performed to quantify the H1299 EGFP cell viability 48 h after incubation with the several formulations (NP, RGD NP 2% and RGDp NP 2%). Briefly, 10 μl MTT (5 mg/ml filtered solution in medium) was added to the cells, and the cells were incubated for 3 h at 37 °C. After incubation, the medium was removed, and 100 μl of dimethylsulfoxide was added to each well to dissolve the formazan crystals. Cells incubated with Triton X-100 (1%, v/v) were used as a positive control, and untreated cells were used a negative control, respectively. The cell viability was determined by measuring the absorbance of enzymatically formed formazan at 560 nm using a microplate photometer (Thermo Scientific, Belgium).

2.7. RNA isolation and purification

2.11. Statistics

At 24 or 48 h post transfection, the cells were washed with PBS and trypsinized with TrypLE™ Express (Invitrogen, UK). Subsequently, the detached cells were washed twice with 1 ml PBS. The pellets were

The experiments were performed in triplicate, unless otherwise stated. Values are given as means ± standard deviations (SD). For the statistics and plotting, PRISM (GraphPad, La Jolla, CA, USA) was used.

4


Statistically significant differences were assessed by an analysis of variance (ANOVA) at a 0.05 significance level and followed by the Tukey's post test or by a 2-way analysis of variance followed by a Bonferroni's post test.

but in fact RNAi dependent. Nanoparticles loaded with DsiRNA showed comparable results (data not shown). 3.3. Evaluation of the effect of αvβ3 targeting on intracellular DsiRNA delivery by using RT-qPCR

3. Results 3.1. Formulation and physicochemical characteristics of the nanoparticles Four different batches of chitosan–PEG (grafting of 20 mol%) were prepared: (i) chitosan reacted with non-conjugated PEG (CPEG), (ii) chitosan reacted with PEG conjugated with RGD at 10 mol% (CPEG RGD 2%), and (iii) chitosan reacted with PEG conjugated with RGDp at 10 mol% (CPEG RGDp 2%) or (iv) 20 mol% (CPEG RGDp 4%). Nanoparticles were formed by using the ionic gelation method with each batch of the CPEG conjugates, and both traditional siRNA and DsiRNA were used to compare their gene silencing efficiencies. Regardless of their composition, the nanoparticles had a mean diameter between 165 and 211 nm and a polydispersity index (PDI) below 0.20, indicating monodispersity (Table 2). All formulations presented a positive zeta potential between 5.6 and 11.0 mV (data not shown). Neither the presence of the ligand nor the ligand grafting ratio significantly influenced the particle characteristics. 3.2. Targeting nanoparticles to αvβ3 integrins enhances their transfection efficiency of H1299 EGFP cells without affecting cell viability First, non-targeted nanoparticles (NP), RGDp-grafted nanoparticles (RGDp NP 2%) and the RGD-grafted nanoparticles (RGD NP 2%) loaded with DsiRNA were compared in terms of their ability to silence EGFP protein expression on H1299 cells which overexpress αvβ3 integrins (Fig. 2A and Supplementary data Fig. S2). EGFP silencing was measured by flow cytometry and the specificity of the EGFP silencing was ensured for each formulation by comparison to similar formulations encapsulating the negative control siRNA. At low DsiRNA concentrations (12.5– 50 nM), no statistically significant differences were observed in the EGFP silencing between the different types of nanoparticles. However, at a DsiRNA concentration of 100 nM and above, the gene silencing increased dramatically after transfection with the targeted nanoparticles, as compared to the non-targeted nanoparticles: as much as approximately 90% EGFP inhibition was achieved at 150 nM DsiRNA with both the RGDp NP 2% and the RGD NP 2%. On the other hand, increasing the initial DsiRNA concentration did not enhance the gene silencing efficiency of the non-targeted nanoparticles. Similar results were obtained 24 h after transfection (data not shown) as well as with siRNA (Fig. 2B). Subsequently, the effect of the ligand density was investigated (Fig. 2B). At 50 nM, approximately 76% EGFP silencing was obtained with the RGDp NP 4% while only approximately 35% knockdown was achieved with the RGDp NP 2%. At higher siRNA concentrations (100 to 200 nM), no difference was observed. Cell viability results (Fig. 2C) demonstrate that the nanoparticles were well tolerated by the cells, irrespective of the surface grafting, and that the observed EGFP inhibition was not a result of cell death

Table 2 Characteristics of the nanoparticles. Values represent mean ± SD (n = 3–6). Type of siRNA

Formulation

Copolymer

Size (nm)

PDI

DsiRNA

Non-targeted nanoparticle, NP RGDp NP 2% RGD NP 2% NP RGDp NP 2% RGDp NP 4%

CPEG

164 ± 15

0.10 ± 0.03

CPEG RGDp 2% CPEG RGD 2% CPEG CPEG RGDp 2% CPEG RGDp 4%

166 ± 16 177 ± 30 207 ± 36 191 ± 32 211 ± 18

0.12 ± 0.04 0.15 ± 0.02 0.11 ± 0.06 0.10 ± 0.04 0.16 ± 0.02

siRNA

The amount of DsiRNA delivered intracellularly after transfection was quantified by using the stem-loop RT-qPCR method [20]. First, a relative quantification approach was used, where the amount of DsiRNA was normalized using a housekeeping gene. These results are thus expressed as arbitrary units (Fig. 3A). Then, absolute quantitative RTPCR was performed by using a calibration curve and used to determine the copy number of DsiRNA molecules per cell (Fig. 3B and Supplementary data Fig. S3). Similar trends in the results were obtained by using both quantification methods, which underlines their robustness. At low concentrations (from 12.5 to 50 nM), the intracellular amount of DsiRNA was similar for all formulations, regardless of the type of surface modification (p N 0.05). However, at 100 nM and above, the targeted nanoparticles were able to deliver a significantly higher amount of DsiRNA to the cells than the non-targeted nanoparticles (p b 0.001). The relative quantities delivered by RGDp NP 2% and RGD NP 2% were 1.5 and 2 times higher, respectively, than the ones delivered using the non-targeted nanoparticles (Fig. 3A), which is in accordance with the two-fold increase in the number of copies/cell (Figs. 3B and S3). At an initial concentration of 100 nM, a number of approximately 2.7 × 105 DsiRNA copies/cell resulted in approximately 80% EGFP silencing for both RGD and RGDp targeted nanoparticles (Figs. S3 and 2A). Given that the full-length active sequence of DsiRNA, i.e. the biologically active molecule, was quantified by using the stem-loop RT-qPCR method, it is possible to correlate the quantitative results with the EGFP expression to clarify the relationship between the intracellular DsiRNA dose and the effect on gene silencing (Fig. 4A–C). Similar profiles were observed for both types of targeted nanoparticles: the EGFP expression was reduced when the amount of DsiRNA in the cells was increased. With the non-targeted nanoparticles, little EGFP silencing was observed, despite relatively high levels of DsiRNA in the cells (Fig. 4A). At a comparable intracellular DsiRNA concentration, estimated in the range of 2.5 × 105 copies per cell, RGD and RGDp targeted nanoparticles mediated extensive EGFP silencing. There appeared to be a threshold concentration of intracellular siRNA above which the targeted NPs silenced the EGFP while the non-targeted NPs failed to silence EGFP. Therefore, we hypothesized that there might be differences in the endocytic uptake pathways between the two types of nanoparticles. 3.4. The RGD and RGDp ligands trigger enhanced nanoparticle endocytosis To investigate the nanoparticle cell uptake mechanism(s) further, supporting microscopy experiments and flow cytometric analyses were conducted. After 1 h incubation, a very low uptake was observed for the non-targeted nanoparticles (Fig. 5A and E) while siRNA contained in the targeted RGDp NP was readily detectable in the cells (Fig. 5B and E). After 2 h incubation, increased regions of fluorescence were observed in the cells for both types of formulations (Fig. 5C and D). The uptake process was also evaluated using flow cytometry. After 1 h with the RGDp NP and RGD NP, the fluorescence peaks presented a shoulder, corresponding to a fraction of cells, which internalized a higher amount of siRNA (Fig. 6A, upper panel). This was not observed for the non-targeted nanoparticles, suggesting a slower uptake of these nanoparticles. After 4 h of incubation, the fluorescence peaks shifted to the right (Fig. 6B, lower panel), indicating similar internalization of the siRNA for all the formulations. αvβ3 positive H1299 EGFP and αvβ3 negative HeLa cells, respectively, were used to confirm the receptor specific uptake. The cellular internalization of the targeted nanoparticles was increased in the αvβ+ 3 H1299 EGFP cells, as compared to αvβ− 3 HeLa cells (Fig. 6B). For the non-targeted nanoparticles, similar uptake was observed in both cell


A

5

B **** **

% EGFP silencing

100 80

****

**** ***

******

NP RGD Dp NP P 2%

***

RGD Dp NP P 4%

60 40 20 0 12.55

25 5 50 1000 150 In nitial siRNA NA con ncentrration n (nM M)

2000

C NP

100 % cell viability

RGDp NP 2% 80

RGD NP 2%

60 40 20

Tr

ito n

1%

0 100

150

200

siRNA (nM/well)

30

45

60

copolymer (µg/well)

Fig. 2. Percentage of EGFP silencing quantified by flow cytometry in H1299 EGFP cells 48 h after transfection with (A) NP, RGD NP 2% and RGDp NP 2% loaded with DsiRNA and (B) NP or RGDp NP presenting different ligand grafting densities (2% and 4%) loaded with siRNA. Results were normalized to the EGFP expression levels in cells transfected with the same formulation containing the negative control siRNA. (C) Cell viability 48 h after transfection with the nanoparticles loaded with siRNA measured using the MTT assay. Results denote mean values ± SD (n = 3). Results significantly different from NP are indicated: *p b 0.05, **p b 0.01 and ***p b 0.001.

lines. This supported that the presence of RGD and RGDp on the nanoparticle surface increases the nanoparticle uptake via the αvβ3 integrin receptors and that both non-specific endocytosis and αvβ3-mediated endocytosis are implicated in the cellular internalization of the targeted nanoparticles. 4. Discussion Despite the prevalence in the literature of the formulation and optimization of siRNA delivery systems as well as their silencing activity in preclinical models, little is known about the specific mechanisms of nanocarrier-mediated intracellular siRNA delivery. Although analytical techniques, such as stem-loop RT-qPCR, have been described for the quantification of small RNAs, they have been employed primarily to

quantify endogenous microRNA [18]. Only few studies have reported the use of RT-qPCR for the quantification of exogenous siRNA and to date, the only vehicles tested have been commercial transfection reagents and lipid nanoparticles [17,29–32]. Here, we report the use of stem-loop RT-qPCR to quantify and compare the amount of DsiRNA delivered by using αvβ3-targeted versus non-targeted chitosan-based nanoparticles. In addition, we evaluated the delivery efficiency mediated by two different types of targeting ligands; the conventional RGD peptide versus its synthetic counterpart, the peptidomimetic RGDp that possesses similar affinity for the receptor while being chemically more stable. Interestingly, the selectivity of the αvβ3 targeting appeared to be dependent on the initially applied concentration of nanoparticles (Fig. 3A). At low concentrations, the small amounts of intracellular DsiRNA

B 14000

NP

600000

12000

RGDp NP 2%

500000

300000

150

200

N

100

p

50

R

DsiRNA initial concentration (nM)

G

D

0

P

N

0

P

0 N

2000

2%

100000

P

4000

2%

200000

D

6000

400000

G

8000

** *

R

RGD NP 2%

10000

copies /cell

relative DsiRNA quantification A.U.

A

Fig. 3. (A) Relative quantification of intracellular DsiRNA 24 h after transfection with NP, RGDp NP 2% and RGD NP 2%. (B) Absolute quantification of DsiRNA 24 h after transfection with NP, RGDp NP 2% or RGD NP 2% at 200 nM DsiRNA. Results denote mean values ± SD (n = 3). Results that are significantly different are indicated: *p b 0.05, **p b 0.01.

6


A

B % EGFP expression

NP 48H

RGDp NP 2% 48H

% EGFP expression

RQ

RQ

15000

60

10000

40

5000

20 0

0 12.5

25

50

100

150

20000

% EGFP expression

% EGFP expression

80

100 80 15000

60

10000

40

5000

20 0

0

12.5

200

Initial DsiRNA concentration (nM)

25

50

100

150

relative DsiRNA quantification

20000


100

200

Initial DsiRNA concentration (nM)

C % EGFP expression

RGD NP 2% 48H

RQ

% EGFP expression

20000 80 15000

60 40

10000

20

5000

0

0 12.5

25

50

100

150


100

200

Initial DsiRNA concentration (nM) Fig. 4. Correlation between the EGFP expression 48 h after transfection of H1299 EGFP cells with (A) NP, (B) RGD NP 2% and (C) RGDp NP 2%. Results denote mean values ± SD (n = 3).

delivered by using the αvβ3-targeted nanoparticles and the nontargeted nanoparticles were comparable. However, at higher concentrations, the amount of DsiRNA delivered by αvβ3-targeted nanoparticles

increased dramatically. The absolute quantity delivered at an applied DsiRNA concentration of 100 nM was 5.9-fold and 4.4-fold enhanced for RGD nanoparticles and RGDp nanoparticles, respectively, as

E

2.0

% pixel/cell

1.5

1.0

0.5

R G

D p

N P

2%

N P

0.0

Fig. 5. Representative images of H1299 EGFP cells (blue: nuclei, green: membrane) transfected with nanoparticles loaded with 100 nM Alexa647-siRNA (red). (A): 1 h incubation with NP, (B): 1 h incubation with RGDp NP 2%, (C): 2 h incubation with NP and (D): 2 h incubation with RGDp NP 2%. The pictures are representative of three experiments. (E): red pixel quantification after 1 h of incubation. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)


7

Fig. 6. Cellular uptake of NP, RGDp NP 2% and RGD NP 2% loaded with Alexa647-siRNA at 100 nM. (A): fluorescence histograms in the FL4 channel of flow cytometric analysis after 1 h (upper panel) and 4 h incubation (bottom panel) of H1299 EGFP cells with the nanoparticles, (B): median of fluorescence intensity after 1 h incubation of H1299 EGFP (αvβ+ 3 ) and HeLa (αvβ− 3 ) cells with the nanoparticles. Results denote mean values ± SD (n = 3). Results of H1299 EGFP uptake significantly different from the results of HeLa uptake are indicated: ***p b 0.001.

compared to the quantity at 50 nM (Fig. 4A). Moreover, RGD nanoparticles delivered a significantly higher quantity than RGDp nanoparticles at the applied concentrations of 150 and 200 nM. For the non-targeted nanoparticles, the level of intracellular DsiRNA increased as well, but to a lower extent. The higher amount of DsiRNA delivered by using the targeted nanoparticles can be explained by the very quick recycling (within minutes) of the integrins to the cell surface, which provides the cells with a high endocytic capacity upon integrin receptor targeting [33]. In comparison, the non-specific uptake appeared less efficient and is probably caused by saturation of non-specific binding sites. The values obtained by relative quantification using RT-qPCR were then correlated to the EGFP silencing, evaluated by using flow cytometry. At low concentrations, poor EGFP silencing (b 40%) was generally observed, regardless of the type of nanoparticles. However, at 100 nM and above, the EGFP inhibition increased dramatically after transfection with the αvβ3-targeted nanoparticles, reaching up to approximately 90% gene silencing. This increased EGFP silencing efficiency is in agreement with the enhancement of the number of active DsiRNA molecules in the cells. Although RGD nanoparticles delivered a significantly higher amount than RGDp nanoparticles (p b 0.001), they did not mediate improved EGFP silencing. In contrast, the EGFP inhibition mediated by the non-targeted nanoparticles remained low, independent of the applied initial concentration. Yet, according to the RT-qPCR results, the amount of DsiRNA delivered by using the non-targeted nanoparticles at 150 or 200 nM should be sufficient to induce EGFP silencing (Figs. 3A and 4A). Indeed, 2.5 × 105 copies were delivered per cell, and this amount mediated approximately 80% EGFP silencing in the case of the αvβ3targeted nanoparticles. Given that the quantification values correspond to the total amount of full-length DsiRNA antisense strand present intracellularly, i.e. in the endocytic vesicles and in the cytoplasm, data suggest that the non-targeted nanoparticles are taken up by the cells, but do not escape from the endosomes and/or release their DsiRNA cargo in the cytoplasm. To further investigate the uptake mechanisms (integrin-mediated and non-specific endocytosis), we performed microscopy and flow cytometric analyses. Both showed slower uptake kinetics of the nontargeted nanoparticles, as compared to the integrin-targeted nanoparticles (Figs. 5 and 6A). These findings are in line with previous reports, stating that the internalization rate of RGD-grafted nanocarriers is higher than the rate of the non-targeted carriers [34,35]. Additionally, the local ligand concentration must be sufficiently high to activate integrin-mediated endocytosis (Fig. 3A). Below a minimum threshold concentration of the targeting ligands, only non-specific interactions between the αvβ3-targeted NP and the cells occurred, resulting in non-specific endocytosis. It has been described that a sufficient concentration of RGD is required to achieve optimal binding to the

integrins [36]. In addition, Mickler et al. demonstrated that the local RGD concentration had an influence on the internalization of RGDgrafted micelles, when the micelles presented an incomplete PEG shielding [34]. In that work, the insufficient PEG shielding induced non-specific interactions of the delivery system with the cell membrane, hindering the binding of the RGD peptide to integrins at low doses. These findings are in accordance with our results, as the PEG shielding of our nanoparticles might be incomplete, as suggested by the slightly positive zeta potential. The hypothesis of the requirement for a minimal ligand concentration to trigger integrin-mediated endocytosis was confirmed by the EGFP silencing results obtained with the RGDp nanoparticles grafted with 4 mol% (Fig. 2B). At 50 nM, these nanoparticles mediated a two-fold higher EGFP inhibition than the same formulation containing 2 mol% RGDp. At 100 nM however, this difference was no longer observed. The specific internalization pathway used to engulf the delivery system has been shown to strongly influence the final efficiency of the delivery systems [37]. It was observed that chitosan–alginate DNA complexes taken up via the caveloae pathway remained entrapped in caveosomes, whereas complexes taken up via the clathrin-mediated pathway in another cell line mediated endosomal escape eventually resulting in high transfection efficiency. However, αvβ3 integrinmediated endocytosis has been described to be mainly clathrindependent and this was verified for the targeted nanoparticles as well as for the non-targeted nanoparticles investigated in the current study (data not shown). For clathrin-mediated uptake, there are some indications in the literature towards the existence of two distinct populations of early endosomes, depending on the nature and destination of their cargos [38]. Cargos destined for the degradation pathway are predominantly transported in highly mobile endosomes, which mature rapidly into late endosomes. On the other hand, cargos destined for recycling such as the recycling ligand transferrin are sorted in both types of endosome populations, the second population being largely static and maturing more slowly. Given that the static endosomes are more important than the rapidly maturing endosomes, the transferrin ligand was found in the second population to a larger extent [38]. As shown in Fig. 6B, both non-specific endocytosis and αvβ3-mediated endocytosis were implicated in the cellular internalization of the targeted nanoparticles. We hypothesize that the previous model is applicable to αvβ3 integrins, suggesting that a large fraction of the targeted nanoparticles may be taken up into the static endosomes and succeed in subsequent endosomal escape. On the other hand, the non-targeted nanoparticles may be predominantly localized in rapidly-maturing endosomes that quickly become late endosomes, thereby being degraded before releasing into the cytoplasm. Moreover, it has been suggested that the presence of adhesive ligands can affect the intracellular processing as

8


RGD possesses membrane destabilizing properties that can help the endosomal escape of the delivery system [39,40]. This phenomenon may contribute to enhancing the escape of the targeted nanoparticles from the endosomes, relative to the non-targeted nanoparticles. Further studies are needed to elucidate this, preferably in a cell line with a higher expression level of αvβ3 integrin receptors, which would facilitate data interpretation. From the results obtained by RT-qPCR, we could explain the lack of gene silencing of the non-targeted nanoparticles in H1299 EGFP cells. Moreover, our results suggested that the concentration of nucleic acid is not the only parameter to optimize in obtaining efficient gene silencing. A minimal local concentration of ligand is required to achieve integrin-mediated uptake and efficient gene silencing in H1299 EGFP cells. The broad application of these quantification methods will certainly help in the understanding of the small RNA delivery mechanisms and dynamics for specific delivery carriers and cell lines.

5. Conclusions The stem loop RT-qPCR technique allows for quantification and comparison of the amount of the full-length DsiRNA antisense strand delivered into the cells by using αvβ3-targeted chitosan-based nanoparticles versus non-targeted nanoparticles. Our results demonstrated that the internalization behavior of the αvβ3-targeted nanoparticles was dependent on the initial concentration, suggesting that a threshold ligand concentration is required to induce αvβ3 integrin-mediated uptake. When this minimum concentration was reached, the αvβ3-targeted nanoparticles were internalized rapidly and at higher amounts than the non-targeted nanoparticles and silenced EGFP expression up to 90% in H1299 EGFP cells without affecting cell viability. On the other hand, the non-targeted nanoparticles showed very low gene silencing efficiency, despite relatively high intracellular DsiRNA levels, suggesting that these nanoparticles might end up in late endosomes or lysosomes without releasing their cargo to the cell cytoplasm. Thus, the silencing efficiency of the chitosan-based nanoparticles is strongly dependent on the uptake and the intracellular trafficking in H1299 EGFP cells. This study provides useful information about the siRNA delivery mechanisms of non-targeted and αvβ3-targeted chitosan-based nanoparticles. A more complete understanding of these mechanisms is critical for the future design of efficient siRNA delivery systems and the translation of RNAi-based drugs towards clinical trials. Conflicts of interest The authors declare that they have no competing interests.

Acknowledgments We gratefully acknowledge M.L. Pedersen, T. Hussein and F. Rose for excellent technical assistance, and H.M. Nielsen for valuable scientific discussions. The authors gratefully acknowledge the Walloon Region (BioWin project Targetum), Wallonie Bruxelles International (bourse d'excellence WBI world courte durée, Héloïse Ragelle), the Fonds De La Recherche Scientifique (mobility grant, Héloïse Ragelle), the Danish Agency for Science, Technology and Innovation, the Danish Council for Independent Research|Medical Sciences (mobility grant, Stefano Colombo), the Drug Research Academy (University of Copenhagen) and the Carlsberg Foundation (Grant No. 2010_01_0294) for financial support. Gaëlle Vandermeulen is a postdoctoral researcher of the Fonds de la Recherche Scientifique — FNRS (Belgium). The funding sources had no involvement in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; nor in the decision to submit the paper for publication.

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jconrel.2015.05.274.

References [1] R. Kanasty, J.R. Dorkin, A. Vegas, D. Anderson, Delivery materials for siRNA therapeutics, Nat. Mater. 12 (2013) 967–977. [2] K. Raemdonck, T.F. Martens, K. Braeckmans, J. Demeester, S.C. De Smedt, Polysaccharide-based nucleic acid nanoformulations, Adv. Drug Deliv. Rev. 65 (2013) 1123–1147. [3] H. Katas, H.O. Alpar, Development and characterisation of chitosan nanoparticles for siRNA delivery, J. Control. Release 115 (2006) 216–225. [4] S. Mao, W. Sun, T. Kissel, Chitosan-based formulations for delivery of DNA and siRNA, Adv. Drug Deliv. Rev. 62 (2010) 12–27. [5] P. Baldrick, The safety of chitosan as a pharmaceutical excipient, Regul. Toxicol. Pharmacol. 56 (2010) 290–299. [6] T. Kean, M. Thanou, Biodegradation, biodistribution and toxicity of chitosan, Adv. Drug Deliv. Rev. 62 (2010) 3–11. [7] R. Riva, H. Ragelle, A. des Rieux, N. Duhem, C. Jerome, V. Preat, Chitosan and chitosan derivatives in drug delivery and tissue engineering, Adv. Polym. Sci. 244 (2011) 19–44. [8] H. Ragelle, R. Riva, G. Vandermeulen, B. Naeye, V. Pourcelle, C.S. Le Duff, et al., Chitosan nanoparticles for siRNA delivery: Optimizing formulation to increase stability and efficiency, J. Control. Release 176C (2014) 54–63. [9] T.H. Kim, S. Il Kim, T. Akaike, C.S. Cho, Synergistic effect of poly(ethylenimine) on the transfection efficiency of galactosylated chitosan/DNA complexes, J. Control. Release 105 (2005) 354–366. [10] W.-F. Lai, M.C.-M. Lin, Nucleic acid delivery with chitosan and its derivatives, J. Control. Release 134 (2009) 158–168. [11] H. Ragelle, G. Vandermeulen, V. Préat, Chitosan-based siRNA delivery systems, J. Control. Release 172 (2013) 207–218. [12] G.C. Tucker, Integrins: molecular targets in cancer therapy, Expert Opin. Drug Deliv. 8 (2006) 96–103. [13] H.D. Han, L.S. Mangala, J.W. Lee, M.M.K. Shahzad, H.S. Kim, D. Shen, et al., Targeted gene silencing using RGD-labeled chitosan nanoparticles, Clin. Cancer Res. 16 (2010) 3910–3922. [14] Y. Sakurai, H. Hatakeyama, Y. Sato, M. Hyodo, H. Akita, N. Ohga, et al., RNAimediated gene knockdown and anti-angiogenic therapy of RCCs using a cyclic RGD-modified liposomal–siRNA system, J. Control. Release 173 (2014) 110–118. [15] F. Danhier, A. Le Breton, V. Préat, RGD-based strategies to target alpha(v) beta(3) integrin in cancer therapy and diagnosis, Mol. Pharm. 9 (2012) 2961–2973. [16] S. Colombo, X. Zeng, H. Ragelle, C. Foged, Complexity in the therapeutic delivery of RNAi medicines: an analytical challenge, Expert Opin. Drug Deliv. (2014) 1–15. [17] A. Cheng, M. Li, Y. Liang, Y. Wang, L. Wong, C. Chen, et al., Stem-loop RT-PCR quantification of siRNAs in vitro and in vivo, Oligonucleotides 19 (2009) 203–208. [18] C. Chen, R. Tan, L. Wong, R. Fekete, J. Halsey, Quantitation of microRNAs by real-time RT-qPCR, Methods Mol. Biol. 687 (2011) 113–134. [19] S. Colombo, D. Cun, K. Remaut, M. Bunker, J. Zhang, B. Martin-Bertelsen, et al., Mechanistic profiling of the siRNA delivery dynamics of lipid–polymer hybrid nanoparticles, J. Control. Release 201 (2015) 22–31. [20] S. Colombo, H.M. Nielsen, C. Foged, Evaluation of carrier-mediated siRNA delivery: lessons for the design of a stem-loop qPCR-based approach for quantification of intracellular full-length siRNA, J. Control. Release 166 (2013) 220–226. [21] S.D. Rose, D.-H. Kim, M. Amarzguioui, J.D. Heidel, M.A. Collingwood, M.E. Davis, et al., Functional polarity is introduced by Dicer processing of short substrate RNAs, Nucleic Acids Res. 33 (2005) 4140–4156. [22] V. Rerat, G. Dive, A. a Cordi, G.C. Tucker, R. Bareille, J. Amédée, et al., Alphavbeta3 integrin-targeting Arg-Gly-Asp (RGD) peptidomimetics containing oligoethylene glycol (OEG) spacers, J. Med. Chem. 52 (2009) 7029–7043. [23] V. Rerat, S. Laurent, C. Burtéa, B. Driesschaert, V. Pourcelle, L. Vander Elst, et al., Ultrasmall particle of iron oxide–RGD peptidomimetic conjugate: synthesis and characterisation, Bioorg. Med. Chem. Lett. 20 (2010) 1861–1865. [24] V. Pourcelle, C.S. Le Duff, H. Freichels, C. Jérôme, J. Marchand-Brynaert, Clickable PEG conjugate obtained by “clip” photochemistry: synthesis and characterization by quantitative 19F NMR, J. Fluor. Chem. 140 (2012) 62–69. [25] R. Kulbokaite, G. Ciuta, M. Netopilik, R. Makuska, N-PEG'ylation of chitosan via “click chemistry” reactions, React. Funct. Polym. 69 (2009) 771–778. [26] S.-D. Li, L. Huang, Stealth nanoparticles: high density but sheddable PEG is a key for tumor targeting, J. Control. Release 145 (2010) 178–181. [27] M. Raviña, E. Cubillo, D. Olmeda, R. Novoa-Carballal, E. Fernandez-Megia, R. Riguera, et al., Hyaluronic acid/chitosan-g-poly(ethylene glycol) nanoparticles for gene therapy: an application for pDNA and siRNA delivery, Pharm. Res. 27 (2010) 2544–2555. [28] D.M.K. Jensen, D. Cun, M.J. Maltesen, S. Frokjaer, H.M. Nielsen, C. Foged, Spray drying of siRNA-containing PLGA nanoparticles intended for inhalation, J. Control. Release 142 (2010) 138–145. [29] A. Cheng, A.V. Vlassov, S. Magdaleno, Quantification of siRNAs in vitro and in vivo, Methods Mol. Biol. 764 (2011) 183–197. [30] Y. Landesman, N. Svrzikapa, A. Cognetta, X. Zhang, B.R. Bettencourt, S. Kuchimanchi, et al., In vivo quantification of formulated and chemically modified small interfering RNA by heating-in-Triton quantitative reverse transcription polymerase chain reaction (HIT qRT-PCR), Silence 1 (2010) 16.

H. Ragelle et al. / Journal of Controlled Release 211 (2015) 1–9 [31] Y. Pei, P.J. Hancock, H. Zhang, R. Bartz, C. Cherrin, N. Innocent, et al., Quantitative evaluation of siRNA delivery in vivo, RNA 16 (2010) 2553–2563. [32] Y. Sakurai, H. Hatakeyama, Y. Sato, M. Hyodo, H. Akita, H. Harashima, Gene silencing via RNAi and siRNA quantification in tumor tissue using MEND, a liposomal siRNA delivery system, Mol. Ther. 21 (2013) 1195–1203. [33] S. Ouasti, P.J. Kingham, G. Terenghi, N. Tirelli, The CD44/integrins interplay and the significance of receptor binding and re-presentation in the uptake of RGDfunctionalized hyaluronic acid, Biomaterials 33 (2012) 1120–1134. [34] F.M. Mickler, Y. Vachutinsky, M. Oba, K. Miyata, N. Nishiyama, K. Kataoka, et al., Effect of integrin targeting and PEG shielding on polyplex micelle internalization studied by live-cell imaging, J. Control. Release 156 (2011) 364–373. [35] S. Hak, E. Helgesen, H.H. Hektoen, E.M. Huuse, P.A. Jarzyna, W.J.M. Mulder, et al., The effect of nanoparticle polyethylene glycol surface density on ligand-directed tumor targeting studied in vivo by dual modality imaging, ACS Nano 6 (2012) 5648–5658. [36] Y.-P. Yu, Q. Wang, Y.-C. Liu, Y. Xie, Molecular basis for the targeted binding of RGDcontaining peptide to integrin αVβ3, Biomaterials 35 (2014) 1667–1675.

9

[37] K.L. Douglas, C. a Piccirillo, M. Tabrizian, Cell line-dependent internalization pathways and intracellular trafficking determine transfection efficiency of nanoparticle vectors, Eur. J. Pharm. Biopharm. 68 (2008) 676–687. [38] M. Lakadamyali, M.J. Rust, X. Zhuang, Ligands for clathrin-mediated endocytosis are differentially sorted into distinct populations of early endosomes, Cell 124 (2006) 997–1009. [39] I.-J. Fang, I.I. Slowing, K.C.-W. Wu, V.S.-Y. Lin, B.G. Trewyn, Ligand conformation dictates membrane and endosomal trafficking of arginine–glycine–aspartate (RGD)functionalized mesoporous silica nanoparticles, Chemistry 18 (2012) 7787–7792. [40] D.M. Shayakhmetov, A.M. Eberly, Z.-Y. Li, A. Lieber, Deletion of penton RGD motifs affects the efficiency of both the internalization and the endosome escape of viral particles containing adenovirus serotype 5 or 35 fiber knobs, J. Virol. 79 (2005) 1053–1061.