Characterization of liposome-mediated gene delivery ... - Nature

Gene Therapy (1997) 4, 226–237  1997 Stockton Press All rights reserved 0969-7128/97 $12.00

Characterization of liposome-mediated gene delivery: expression, stability and pharmacokinetics of plasmid DNA AR Thierry1, P Rabinovich1, B Peng1, LC Mahan2, JL Bryant3,4 and RC Gallo 1,4 1

Laboratory of Tumor Cell Biology, National Cancer Institute; 2Laboratory of Neurochemistry, National Institute of Neurological Disorders and Stroke; and 3 Animal Care Unit, National Institute of Dental Research, National Institutes of Health, Bethesda, MD, USA

We have characterized a new synthetic gene delivery system, termed DLS, which may be suitable for systemic gene therapy. DLS constitutes a lipopolyamine and a neutral lipid and associated plasmid DNA in the formation of lamellar vesicles (DLS–DNA). The ratio of lipids and lipid to DNA as well as the method of preparation were optimized to yield a high in vitro transfection efficiency compared with that previously reported for cationic lipid systems. DLS– DNA showed a rapid cellular uptake and distribution in the cytoplasmic and nuclear (especially in the nucleoli) compartments as determined by laser-assisted confocal microscopy. There was little or no plasmid DNA degradation over a period of 20 min, relatively slow plasma clearance, and effective and rapid cellular uptake of DLS–DNA following intravenous administration in mice. Supercoiled

plasmid DNA could be detected in blood cells up to 1 h after injection. Systemic administration of DLS–DNA yielded transgene expression in mouse tissues, such as in lung or liver. The ratio of DLS:DNA and the procedure used to form DLS–DNA affected both the level and cellular specificity of expression of a luciferase reporter gene showing that in vitro transfection efficiency of DLS–DNA formulations cannot be easily extrapolated to an in vivo setting. Optimization of the formulation of a DNA delivery system was critical to obtain a defined structure resulting in a preparation with high reproducibility and stability, greater homogeneity of particle size and high efficacy following systemic gene transfer. In addition, the DLS system may be formulated for specific target tissues and may have a wide range of applications for gene therapy.

Keywords: gene transfer; liposomes; plasmid DNA; biodistribution; pharmacokinetics

Introduction One of the main impediments to successful gene therapy is the generally poor efficiency of DNA delivery. Most gene therapy efforts involve the use of retroviral vectors because of the usual efficient cell entry and stable integration of the gene. Clinical use of retroviral vectors, however, is hampered by safety issues.1–3 A first concern is the possibility of generating an infectious wild-type virus following a recombination event. A second concern is the consequence of random integration of the viral sequence into the genome of the target cell which may lead to a tumorigenic or a cytotoxic event. In addition, retroviral systems integrate the foreign gene only in dividing cells. DNA viruses, such as adenoviruses and herpes viruses, are potential gene carriers since they are able to infect postmitotic cells. However, these systems are limited by problems of the need for helper virus, restricted host range, cytotoxicity, immunogenicity and potential safety problems. Correspondence: AR Thierry at his present address: Adcell Ther., 10 rue Armand Gobert, 03300 Cusset, France 4 Present address: Institute of Human Virology, University of Maryland Biotechnology Institute, 618 West Lombard St, 2nd Floor, Baltimore, MD 21201, USA Received 13 February 1996; accepted 19 September 1996

Synthetic gene transfer vectors are receiving increasing study as an alternative to viral vectors since this strategy appears to be safe. Potential methods of gene delivery that could be employed include the pneumatic DNA gun,4 DNA–protein complexes5 or lipidic particles.6 The typical genetic material delivered to target cells by these methods are plasmids, although antisense oligonucleotides are currently being tested as well. 8 Plasmid preparations are simple, quick, safe, inexpensive and may be applied in combination with a synthetic carrier. Therefore, gene therapy by this means may be safe, durable and used as a drug-like therapy. The successful use of this genetic tool for in vivo approaches to gene therapy will rely on the development of an efficient cell delivery system and appropriate levels of gene expression. Lipid particles have been shown to be efficient vehicles for plasmid DNA in many in vitro and in vivo applications.6,7 Formation of complexes of DNA with cationic lipidic particles has recently been the focus of research of many laboratories.9–17 Behr18 was first to demonstrate compaction and coating of DNA to synthetic cationic lipids. However, in vivo application of cationic lipids plus DNA, especially systemic gene delivery of these DNA complexes, is not well documented.19–21 The fundamental aim of this study was to design a defined and novel liposomal gene delivery system which we have termed DLS, which derives from an optimized

In vitro and in vivo liposome-mediated gene transfer AR Thierry et al

formulation of a lipospermine (DOGS) and a neutral lipid (DOPE). We have examined different physicochemical aspects in formulating both lipids and DNA in order to obtain an improved gene delivery system suitable for systemic administration. Ratios of lipids and DNA were varied and methods of preparation of DLS-associated DNA (DLS–DNA) were optimized for in vitro transfection efficiency, as well as cytoplasmic and nuclear transport and reporter gene expression. These results were then compared to in vivo evaluation of DNA protection, pharmacokinetics and transgene expression in mouse tissues following intravenous administration.

Results Physical properties of the DLS liposomes In this work, two different methods of preparing DLS– DNA were employed: (1) complex formation between preformed cationic DOGS–DOPE containing vesicles and negatively charged DNA (DLS-1–DNA) or (2) hydration of a DOGS–DOPE dry lipid film with an aqueous DNA solution (DLS-2–DNA). This type of preparation is common for the preparation of conventional liposomes. Liposomes are the result of a rapid hydration of a concentrated form (dried or dissolved in solvent) of phospholipidic compounds. Consequently this formulation favors the entrapment of the solutes contained in the hydration mixture in the hydrophilic or hydrophobic compartments of the forming liposomal vesicles. Negative stains of DLS-1, DLS-1/pRSV-luc, DLS-2 and DOGS (Transfectam reagent; Promega, Madison, WI, USA)/pRSV-luc were prepared and analyzed by transmission electron microscopy (TEM) analysis (Figure 1). DLS-1/pRSV-luc were prepared at a 0.6 DOGS:DOPE molar ratio and a 0.08 DNA:lipid weight ratio and samples were analyzed by TEM (Figure 1c and e) and particle sizing. DLS-1–DNA lipid vesicles were observed throughout this preparation in large quantities. Each of the different vesicles appeared to display a heavily stained core region, which was surrounded by many different layers of membranes exhibiting pleomorphic shape (mainly roughly spherical and rarely tubular). The particle size distribution analysis showed a mean diameter of 240 nm with a coefficient of variation of 48%. DLS-1 particles (Figure 1a) were similar to DLS-1–DNA regarding their shape and presence of layers of membrane surrounding the particles, but their overall size was remarkably smaller (73 nm, 49%). DLS-1–DNA particles with a DNA:lipid ratio ranging from 0.08 to 0.16 look somewhat similar (data not shown). At higher ratio the particles did not exhibit lamellar structures. DLS-2–DNA (0.6 DOGS: DOPE molar ratio and 0.08 DNA:lipid weight ratio) particles were roughly spherical in shape and appeared to be made up of lipid-like material (Figure 1d and f). These particles did not appear to show membrane layers, but at higher magnification microscopy a few membranes could be observed, which appeared to make up the bulk of the lipid particles. DLS-2–DNA showed larger size preparations resulting in a higher mean diameter and especially a higher coefficient of variation (327 nm, 69%). DOGS–DNA (0.16 DNA:lipid weight ratio)12 formed lipid particles which appeared very different from those observed in DLS-1–DNA (Figure 1b). They did not display any ultrastructural lamellar detail. The only

similarity between this sample and DLS-1–DNA was that the lipid particles were heavily stained. DOGS–DNA particles exhibited a mean diameter of 190 nm (50%).

In vitro transfection efficiency Optimal preparation and transfection conditions were determined for the in vitro expression of DLS/pRSV-luc. DLS-1/pRSV-luc at a 0.6 DOGS:DOPE molar ratio yielded a 10- and 20-fold greater luciferase activity when compared with DLS-1–DNA prepared at 0.2 and 1.4 ratios, respectively (Figure 2a). Two micrograms of pRSV-luc transfected with DLS-1 liposomes into HeLa cells showed higher reporter gene expression in HeLa cells compared with the addition of 0.5 or 1.0 mg (Figure 2a). The addition of 2 mg of DLS–DNA resulted in a statistically higher transfection level compared with the addition of 1 mg DLS–DNA (P , 0.05). Thus, 2 mg of plasmid DNA was used throughout all the in vitro studies since higher amounts resulted in cytotoxicity. Cell confluency at the time of transfection did not significantly modify transgene expression (Figure 2a). There is no statistical difference (P . 0.05) between transfection performed at 75 or 25% cell culture confluency level. We observed about the same results regarding the effect of DNA amount per well and cell culture confluency when using either the DLS-1 or -2 preparations. pRSV-luc alone did not express any luciferase activity in HeLa cells (data not shown). Various DNA:lipid weight ratios and DOGS:DOPE molar ratios were tested; the combinations of these parameters yielding the highest transfection efficiency are shown in Figure 2b. Optimal conditions for DLS-1mediated transfection were observed when a 0.6 DOGS: DOPE molar ratio and a 0.16 DNA:lipid ratio were used (1.35 × 109 RLU or 68 mg luciferase per mg of protein). An optimal preparation DLS-2–DNA (1.2 DOGS:DOPE molar ratio and 0.08 DNA:lipid weight ratio) yielded approximately half the luciferase expression when compared with the DLS-1 preparation (Figure 2b). In contrast, this DLS-2–DNA formulation resulted in higher transgene expression than DLS-1–DNA under optimal conditions in primary human vascular endothelial cells and in mouse bone marrow cell cultures (data not shown). At neutral pH, DOGS has three cationic charges per molecule.18 Thus, DLS-1–DNA prepared at a 0.6 DOGS:DOPE molar ratio and a 0.16 DNA:lipid weight ratio contain theoretically 2.5 net positive charges. There was no consistent relationship between transfection efficiency and the net positive charge. pRSV-luc plasmid DNA was also used for a comparative study of the transfection efficiency of DLS-1 liposomes and commercially available lipid reagents for transfection (Figure 3). Experimental conditions were optimized for each method tested. In serum-containing cell culture medium, transduction efficiency in HeLa cells treated with the DLS-1 system appears to be 11-, 10- and 37-fold higher than that of DOGS or Transfectam (Promega), Lipofectin and Lipofectamine (Gibco BRL, Gaithersburg, MD, USA), respectively. In serum-free medium, transduction efficiency using DLS–liposomes appears equivalent to that determined when cells are incubated in serum-containing medium. In contrast, use of Lipofectamine in serum-free conditions produces greater transfection efficiency (Figure 3). There is a dramatic decrease in transduction efficiency (186-

227


228

Figure 1 Transmission electron microscopy of DLS-1 (a), DLS-1–DNA (c and e), DOGS–DNA (b) and DLS-2–DNA (d and f). Magnification × 36 000 (a, b, d, e); × 65 000 (c and f).

fold) using Lipofectamine in a medium containing 10% heat inactivated fetal bovine serum (Gibco), emphasizing the instability of DNA when exposed to serum nucleases and the need for complete DNA protection from enzymatic attack in a biological environment.

The lacZ gene was also used as a reporter gene for in vitro transfection. As shown in Figure 4a, 70–90% of the HeLa cells treated with DLS/pCMV b-gal under optimal conditions are positive for b-gal staining, but the intensity of substrate staining varied greatly among individual


229

Figure 2 Characterization of the in vitro luciferase expression in HeLa cells using the DLS delivery system and the pRSV-luc plasmid DNA. (a) Effect of DNA doses, cell culture confluency level and DOGS:DOPE molar ratio in DLS-1. The effect of DNA amounts was examined with DLS-1–DNA at a 0.08 DNA:lipid ratio and a 0.6 DOGS:DOPE molar ratio on cells at approximately 75% confluency level. Effect of cell culture confluency at the time of the addition of DLS-1–DNA was determined using 1 mg DLS-1–DNA formulated as described for the effect of DNA amounts. Effect of DOGS:DOPE molar ratio was studied using 2 mg DLS-1–DNA in 75% confluency cultured cells. (b) Comparison of the DLS-1 and DLS-2 preparations at 0.08 or 0.16 DNA:lipid weight ratio and at 0.6 or 1.2 DOGS:DOPE molar ratio, respectively. Values are means (± s.e.m.) of triplicate luciferase determinations from at least three different experiments. Statistical differences between 0.08 and 0.16 DNA:lipid weight ratio were evaluated: *P , 0.001; **P , 0.05; ***P , 0.05; ****P . 0.05.

cells. Expression of the b-gal was also observed in Kaposi’s sarcoma (KS)-4 primary cell cultures ranging from 30–50% (Figure 4b) and in various other cell types (Table 1). The stability of DLS-1 liposomes (not complexed with DNA) was monitored by measuring the in vitro transfection efficacy in HeLa cells of various preparations (n = 12) stored up to 9 months at 4°C. The transfection assay was carried out as described in Figure 3. Results revealed a 28.7% coefficient of variation indicating a great stability and reproducibility of DLS-1 preparations.

Intracellular distribution of plasmid DNA The intracellular distribution of plasmid DNA following DLS delivery was observed using laser-assisted confocal microscopy and ethidium bromide labeled pCMV b-gal DNA. Figure 5 presents images of HeLa cells treated with DLS–DNA for 24 h (Figure 5a, b) and post-incubated in DNA-free medium for an additional 24 h (Figure 5c, d). Labeled DNA delivered with DLS-1 (Figure 5a) was distributed in the cytoplasm and in the nucleus, particularly in the perinuclear and the chromosomal/nucleoli compartments. Intense (red-yellow) spots were found in the cytoplasm and are characteristic of endocytic vesicles as previously described using confocal microscopy for the delivery of liposome-delivered molecules such as anticancer agents22,23 or oligonucleotides.24 The presence of DNA in endocytic vesicles was observed 15 min after DLS–DNA addition to the cells, indicating rapid cellular uptake (data not shown). Following a 24-h incubation period, DNA distribution was homogeneous in the cytoplasm while nuclear distribution did not vary (Figure 5c). Some of the staining may be due to partitioning of the

ethidium bromide from the plasmid DNA to the chromosomal DNA especially when observing the fluorescence 48 h after the addition of DLS–DNA. However such an event would buttress the interpretation that plasmid DNA is also partitioning from the liposomes as demonstrated by the homogeneous and diffuse fluorescence pattern in the cytoplasm. Nevertheless this study undoubtedly demonstrated the complete release of the DNA from the endocytic vesicles and supports the notion of the complete or partial release of the DNA from the liposomal carrier. Cells exposed to DLS-2–DNA exhibited a comparable distribution pattern except that DLS–DNA seemed to aggregate slightly on the cell surface (Figure 5b–d). Fluorescence was not detected in HeLa cells treated with naked labeled plasmid DNA alone (data not shown). Similar intracellular patterns were observed in KS Y-1 (a Kaposi’s sarcoma neoplastic cell line), MOLT-3 (a human lymphoma cell line), HepG2 (a human hepatoma cell line), and human macrophages obtained from peripheral blood treated with DLS–labeled-DNA.

In vitro plasmid DNA sensitivity to DNaseI Plasmid DNA expression is mainly in its supercoiled form as compared to the relaxed circular or linear forms.25 Consequently, the fate of supercoiled DNA was analyzed under various conditions to estimate its stability in a biological environment. The integrity of pCMV b-gal alone or with various DNA carriers was analyzed in the absence or the presence of DNaseI (Figure 6). Only relaxed DNA was detected when naked DNA was exposed to DNaseI. None of the carriers we tested (DLS1, DLS-2, lipofectin and Transfectam) completely pro-


230

Figure 3 Comparison of the transfection efficiency in HeLa cells of DLS1 and various commercially available lipid reagents for DNA delivery. Transfectam (T, Promega), DLS-1 (DLS), Lipofectin (LF, Gibco BRL) and Lipofectamine (LFA, Gibco/BRL). Cells were incubated with pRSV-luc delivered with different DNA carriers for 4 h in the presence or absence of 10% heat inactivated bovine serum. Luciferase activity was measured 3 days after transfection. Values represent means (± s.e.m.) of triplicate luciferase determinations of at least three different experiments.

tected the DNA from degradation. Using Transfectam (DOGS) as a carrier only a smear was observed on the gel corresponding to degraded low molecular weight DNA. Using lipofectin, no plasmid DNA was detected. In clear contrast, plasmid DNA in supercoiled form was detected when DLS-2, and to a lesser extent DLS-1, were used as carriers. Plasmid DNA presented either with DLS-1 or DLS-2 liposomes was equally protected in the presence of 25% human serum (data not shown).

Pharmacokinetics in peripheral blood of plasmid DNA after intravenous administration Pharmacokinetic analysis of plasmid DNA in the plasma and in blood cells was conducted following intravenous delivery (75 mg pBKd1RSV-luc per mouse). The integrity of the plasmid DNA delivered in free form or with the DLS-1 liposomes was determined at selected times using agarose gel analysis (Figure 7). Under the conditions used in this study sensitivity limits of detection were determined as 3 ng DNA (Figure 7a, lanes 1, 2, 3). The results indicate that supercoiled DNA is not detected in either plasma or the cell fractions 1 min following injection of naked DNA (Figure 7a). An intense band of a high molecular weight was observed in the cell fraction samples and representing chromosomal DNA. When plasmid DNA was delivered by DLS-1 liposomes, supercoiled forms were readily observed in the plasma fraction at 1 min and declined thereafter up to 60 min (Figure 7b). No supercoiled DNA was detected at 120 min (data not

Figure 4 In situ detection of b-gal expression in HeLa cells (a) and primary Kaposi’s sarcoma spindle cells (b). Cells were transfected with 2 mg DLS-1/pCMV b-gal and b-gal staining carried out 48 h later.

Table 1 Relative b-gal expression in cultured cells following DLS/pCMV b-gal exposure % Positive cells Cell lines HeLa KS Y-1 NIH 3T3 293 Rat 2

Human cervical carcinoma Human Kaposi’s sarcoma Murine fibroblasts Human embryonic kidney Rat embryonic fibroblasts

Primary cultures HUVEC Human umbilical vascular endothelial KS 4 Human Kaposi’s sarcoma spindle

70–90 40–50 30–50 10–25 5–10 40–60 30–50

Cultured cells were transfected at optimal conditions (0.5–2 mg DNA per 0.5 × 106 cells) with DLS/pPCMV b-gal. Positive cells were counted in three different fields at the center of the culture well.

shown). Since the degradation in plasma occurred rapidly under the experimental protocol used in this study it is not possible to estimate the plasma half-life of the supercoiled form of plasmid DNA. However, the plasma half-life of relaxed DNA is approximately 10 to 20 min.


231

Figure 5 Intracellular localization of ethidium bromide-labeled plasmid DNA (pCMV b-gal) delivered alone (c) or with DLS-1 (a, c) and DLS-2 (b, d) liposomes into HeLa cells. Following incubation with 0.1 mg DLS–DNA per ml for 24 h, cells were postincubated in fresh medium for an additional 24 h (c, d). The photographs represent computer-enhanced images from laser-assisted confocal microscopy. Fluorescence intensity could be observed from low (blue), medium (yellow) to high (red). Bar represents 10 mm.

The supercoiled form of DNA was observed in cell fractions up to 60 min after injection with no detection 120 min after injection (data not shown). Cellular uptake of plasmid DNA in blood cells reached its maximum as early as 1 min following injection of DLS-1–DNA (Figure 7b). Although DNA amounts in the cell fraction decreased with time, especially within 8 min following injection, the apparent supercoiled:relaxed ratio seemed identical and comparable to the positive control. In contrast, the linear:supercoiled ratio increased with time following injection demonstrating that DNA degradation yielding linear DNA was relatively stable either in the plasma or in the cell fraction. This result suggests that the main nuclease activity is endonucleolytic. These data

show DNA protection of DNA by DLS, slow plasma clearance of DLS-1–DNA, and effective and rapid cellular uptake in the blood circulation.

In vivo transgene expression following intravenous delivery In order to improve the length of transgene expression we used an episomally self-replicative plasmid (pBKd1RSV-luc) for the in vivo administration of DLS–DNA in mice.21 The efficiency of various DLS/pBKd1RSV-luc preparations in transfecting mouse tissues was determined following intravenous delivery. As shown in Table 2, liver expressed less luciferase activity than lung tissue. When compared with adminis-


232

Figure 6 Comparison of the in vitro protection of plasmid DNA by DLS and various commercially available DNA carriers. Detection by agarose gel electrophoresis of the relaxed (R), linear (L) and supercoiled (S) forms of plasmid DNA (pCMV b-gal) following incubation in the absence (−) or presence (+) of DNase I for 3 h. LF (Lipofectin, Gibco BRL), T (Transfectam, Promega) and free (naked DNA).

tration of DLS-1–DNA, DLS-2–DNA yielded a lower luciferase activity in the lung but higher in the liver. Luciferase activity in the lung increased when liposomes with low DNA:lipid ratio were employed. In contrast, transgene expression increased in the liver when the DNA:lipid ratio was increased in the DLS liposomes. The values of luciferase activity expressed as RLU per protein and RLU per organ taken together showed higher transgene specific activity in the lung than in the liver. Following an intravenous injection of 75 mg DLS-1–DNA (0.6 DOGS:DOPE and 0.16 DNA:lipid ratios), Southern blot analysis of the amount of DNA in mouse tissues approximately yielded 4.7, 2.04 and 0.12 ng of DNA in the lung, liver and heart, respectively. These results when combined with luciferase measurement give an estimation of a specific activity of 16, 3 and 14 pg luciferase per mg DNA ratios in these tissues, respectively. While homogenization and diminution of the overall particle size by sonication resulted in lowering the transfection efficiency in HeLa cells to 10% of that of nonsonicated DLS-1–DNA, sonicated DLS yielded a higher level of transgene expression than nonsonicated DLS in blood cells (data not shown). As detected by hematoxylin-stained histological organ sections, there was no detectable toxicity at the DLS–DNA doses used in this study (data not shown).

Discussion In this study we have described the optimization of a synthetic gene delivery system consisting of DOGS and DOPE (DLS liposomes). Various combinations of DOGS: DOPE and DNA:lipid ratios as well as methods of preparation were tested and these results revealed that even slight variation of these factors can greatly affect transfection efficiency both in vitro or in vivo. Looking at a variety of cultured cells in the presence of serum, a more in vivolike paradigm, DLS/pRSV-luc under optimal conditions yielded higher transgene expression than various other cationic lipids. DOGS is a lipopolyamine, a class of cationic amphi-

philes in which the cationic spermine group is connected to a double chain (C16:0) lipophilic group via an amidoglycyl spacer arm.7 Behr et al found that this compound was especially efficient in compacting DNA18 and in transfecting a variety of cell types without cellular toxicity.12,26,27 DLS liposomes are a combination of DOGS and a neutral lipid, DOPE, which was shown to destabilize the cellular membrane and enhance transfection efficiency when added to other cationic lipids.28 DLS– DNA forms lipid particles which condense and interact through the electrostatic interaction of the negative charges on nucleic acids and the positive charges on DOGS. Electron microscopic analysis showed the presence of lamellar vesicles when DLS–DNA was prepared at a 0.6 DOGS:DOPE molar ratio and a 0.08 to 0.16 DNA: lipid weight ratio. We suggest that DOPE, a lipid with relatively weak surface hydration, with DOGS at a 0.6 DOGS:DOPE molar ratio, forms membrane structures from hexagonal arrangements (hexagonal phase). Alternatively, it might be possible that DOPE and DOGS, both of whose molecules exhibit an inverted cone shape, form a stable bilayer structure. Although some other cationic lipids, such as DOGS alone, Lipofectin or lipofectamine (Gibco BRL) are able to transfect cells, they do not form stable bilayer structures in physiologic salt concentration.7 The term ‘cationic liposomes’ for most of the preparations made of cationic lipid–DNA complexes is improper. It would be more appropriate to describe them as ‘nucleolipidic particles’. The term ‘liposomes’ has been widely used to describe a closed structure comprising of an outer lipid bi- or multi-layer membrane and encapsulating an aqueous space containing typically a drug, but also vesicles where a drug (ie hydrophobic drug) is localized inside layers of phospholipid membranes. Since DNA is assimilated and compacted into layers of membrane in DLS–DNA vesicles, the DLS–DNA may be described as liposomes. Although some cationic lipids, such as DOGS alone, are able to transfect cells, they do not form stable bilayer structures. The DLS–DNA liposomal structures may partially explain the greater efficacy


233

Figure 7 Pharmacokinetics of plasmid DNA administered in free form (a) or with the DLS-1 liposomes in mouse peripheral blood circulation (b). pBKd1RSV-luc 75 mg was injected intravenously in mice and blood was collected at indicated time. DNA was extracted from the plasma or the cell fraction and analyzed by agarose gel electrophoresis. pos, positive control plasmid DNA (0.3, 0.03 and 0.003 mg in lanes 1, 2 and 3, respectively); neg, negative control uninjected mouse blood; R, relaxed; L, linear; S, supercoiled plasmid; C, chromosomal DNA.

in transferring DNA and better protection function for DNA from degradation than other cationic lipid delivery systems. Our results showed that there was no direct relationship between transfection efficiency and net positive charges when altering DOGS:DOPE ratio and the method of preparation. In addition, Remy et al30 have combined DOGS and DOPE at different lipid ratios, DNA:lipid ratio and methods of preparation than those presented in this work. They formed DNA complexes which did not enhance transfection efficiency when compared to DOGS alone.

These observations reveal the importance of the equilibrium of these physicochemical parameters in formulating effective DLS–DNA preparations. Gershon et al29 suggested that at a critical cationic lipid:DNA ratio two processes occur namely, DNA-induced lipidic membrane fusion and liposome-induced DNA collapse, resulting in a marked cooperativity of the entrapment process. DNA delivered using the DLS carrier system are avidly taken up by HeLa cells and rapidly released from endocytic vesicles resulting in a high cytoplasmic and nuclear distribution of DNA. Because of the presence of high


234 Table 2 Transfection efficiency in mouse tissues following intravenous delivery of various DLS–DNA preparations Vehicle

DLS-1

Tissue Transfection efficiency RLU per mg protein (RLU per organ) DNA:Lipid (weight ratio)

Lung Liver

DLS-2

Lung Liver

0.08

0.16

0.32

2014 ± 610 (117 216) 0

598 ± 270 (34 848) 35 ± 29 (13 125)

55 ± 30 (3201) ND

409 ± 401 (27 730) 82 ± 77 (27 060)

202 ± 24 (12 680) 192 ± 180 (13 152)

ND ND

Luciferase activity was determined 3 days after intravenous administration of 75 mg pBKd1 RSV-luc delivered in DLS liposomes. RLU per mg are the mean ± s.e.m. of triplicate determinations of at least three mice. ND, not determined.

concentrations of degradative enzymes in endocytic vesicles, bypassing or reducing the time of contact with this vesicular trafficking could significantly improve the intracellular availability of nucleic acids delivered with a carrier. The intracellular DNA localization study was performed in a cell line which, as most cell lines, exhibits high endocytic activity and may not be relevant in the in vivo situation. In vivo cells greatly differ in their use of an endocytic pathway, eg the lymphocytes are low compared to cells of the reticular endoplasmic system. However, DLS–DNA was also shown to enter the nucleus of peripheral blood mononuclear cells in culture (data not shown). These results suggest that DLS–DNA effectively delivers DNA into the nucleus where gene expression is initiated. DOGS appears to offer a unique combination of properties such as cell membrane destabilization, endosome buffering capacity and cellular caryophily (possibly nuclear tropism).27,28 In addition, DOGS contains the polyamine (spermine) groups having the functional ability to compact DNA and this interaction is quickly reversible in physiological conditions.9,26 These structures protected DNA from degradation and this DNA was slowly cleared from plasma and rapidly taken up by cells following its intravenous administration. DNA protection may be explained by the liposomal nature of the DLS–DNA particles. A pharmacokinetics study in mice showed that supercoiled plasmid DNA was detected for up to 60 min (3 ng, sensitivity limit) but was rapidly eliminated in plasma when DLS-1–DNA was intravenously delivered. However, plasma half-life of relaxed DNA was estimated as 10–20 min. This work showed for the first time the presence of intact plasmid DNA in plasma and in blood cells following systemic injection of plasmid DNA. Lew et al31 using DMRIE (1,2-dimyristoyl-oxypropyl-3-dimethyl ammonium bromide)/DOPE as a delivery system determined a plasma half-life of a few minutes for relaxed plasmid DNA by Southern blot analysis while no supercoiled form could be detected (1 pg of plasmid DNA, sensitivity limit). It seems likely that cationic DNA–lipid particles may bind to serum components such as albumin, heparin, lipopro-

tein or specific opsonins. These interactions may be different with different DNA carriers and this may explain the differences observed in plasma clearance. In addition, our results clearly showed rapid and efficient DNA uptake into blood cells where supercoiled DNA was detected up to 1 h. Thus, functional plasmid DNA uptake in blood cells occurred after intravenous administration of DLS–DNA, albeit with degradation and distribution to other biological compartments. Reporter gene expression was detected in mouse tissues after intravenous administration of DLS–DNA. Its level varied with the DNA:lipid ratio of the injected DLS–DNA, the procedure used in preparing DLS–DNA, and the tissue tested. For example, transgene-specific activity was higher in lung than in the liver. This is probably a function of the ‘first pass’ effect or trapping of liposomes in the lung capillary bed versus the liver sinusoids. Among those tested the optimal formulation for the lung was the DLS-1 system at a 0.08 DNA:lipid ratio and for the liver, the DLS-2 system at a 0.16 DNA:lipid ratio. The DLS–DNA physico-chemical properties differentially affected the efficiency of in vivo and in vitro transgene expression. Consequently, the transfection efficiency of DLS–DNA formulations determined in vitro cannot be extrapolated in an in vivo setting. In addition, these observations point out that specific DLS– DNA formulations might be employed when specific cells or organs are targeted for gene therapy. The bioavailability of DNA to specific tissues could be altered either by the route of administration or by the DLS–DNA binding with serum components which could influence the interaction with cell surfaces or the extravasation in tissues from the vascular to interstitial space. Specific targeting may also be improved using DNA expression vectors with cell-specific promoters21 or gene delivery particles conjugated to specific ligands32 or antibody.33 The presence of a membrane bilayer and of DOPE in DLS makes it possible to anchor conjugates to their surface. Repeated administration or use of self-replicating plasmid vectors21 might extend the duration of transgene expression. The optimization of the formulation is a prerequisite for developing a synthetic gene delivery system. The DLS-1 system appears advantageous because: (1) its preparation is standardized giving high reproducibility, great homogeneity of a low particle size and a defined liposomal structure; (2) its stability at 4°C; (3) its low toxicity; and (4) its biological stability (low plasma degradation rate) allows for systemic delivery and broad biodistribution of transgene expression. 21,34 We also described the efficacy of DLS used for the in vivo expression of the human MDR-1 gene in mouse bone marrow progenitor cells by employing two different approaches: (1) systemic delivery as described here, and (2) an ex vivo approach by transplanting in vitro transfected bone marrow cells.35 This study demonstrated that the transferred gene conferred to these cells the functional activity of the produced protein. In addition, this study showed that careful design of the formulation of a liposome-mediated gene transfer system may also make it possible to produce tissue specific gene therapy.

Materials and methods Cells and plasmid DNA The cell lines and the primary cultured cells used in this study were maintained under the recommendations of


the American Type Cell Collection (Rockville, MD, USA). pCMV b-gal, an E. coli lacZ (b-galactosidase) gene expression plasmid vector driven by the human cytomegalovirus immediate–early gene promoter, was obtained from Clontech Laboratories (Palo Alto, CA, USA). pRSVluc consisted of a full-length firely Photinus pyralis luciferase cDNA inserted into the plasmid pGEM3 (Promega, Madison, WI, USA), under the control of the Rous sarcoma virus long terminal repeat promoter. pBKd1RSVluc is an episomally replicative plasmid containing luc and a genomic fragment of the human papovavirus BKV (Ref. 21 and LCM, unpublished data). Recombinant plasmid DNA was purified by two ultracentrifugations through cesium chloride gradients.

Liposome preparation Two liposomal systems were used in this study. Both have the same chemical composition but differed in their preparation. DLS-1 liposomes were formed by mixing DOGS (generously provided by JP Behr, Universite´ Pasteur, Strasbourg, and also available from Promega and DOPE (Sigma, St Louis, MO, USA). After thorough stirring, the mixture was evaporated to dryness in a round bottomed borosilicate tube using a rotary evaporator (Labconco, Kansas City, MO, USA). The subsequent dried lipid film was resuspended in a low volume of ethanol (20 ml per mg lipid). Formation of liposomes was carried out by adding an excess of distilled water (200 ml per mg lipid). After homogenization by pipetting, the mixture was incubated for at least 15 min. Complex formation of nucleic acids to the liposome bilayer membrane was achieved by simply mixing the preformed DOGS– DOPE liposomes into a solution of nucleic acids. In an Eppendorf tube, DLS liposomes were added to DNA in 150 mm NaCl to a final concentration of 50 mm NaCl. The mixture was slightly mixed and incubated for 1 h at room temperature. Complex formation was very effective and nearly complete since .80% of nucleic acids were assimilated on to the liposomes as determined by using nick-translated 32P-labeled DNA. DLS-2 liposomes were similarly comprised of DOGS and DOPE and a dry lipid film was obtained as previously mentioned. The dried lipid film was then resuspended in a minimal volume (7 ml per mg lipid) of a water solution containing plasmid DNA (1–2 mg/ml). Formation of liposomes was carried out by thorough vortexing. The subsequent liposome preparation was diluted to 50 mm NaCl. Entrapment and/or assimilation of nucleic acid by the DLS-2 liposomes was very efficient and nearly complete as approximately 80% of the nucleic acids are encapsulated or associated with the liposomes. Microscopy TEM was carried out by Advanced Biotechnology (Columbia, MD, USA) after negative staining with 1% uranyl acetate of the sample diluted 1:50 in 50 mm NaCl. Laser-assisted confocal microscopy was carried out using ethidium bromide-labeled pCMV b-gal DNA. Labeled plasmid DNA was obtained by mixing 50 mg EtBr with 1 mg DNA and by extensive dialysis (three times in 1 l). HeLa cells were grown on glass Chamber Slide (Nunc, Naperville, IL, USA) up to 50% confluency. Cells were exposed to liposomal labeled DNA at 0.1 mg/ml for 24 h. Cells were then rinsed three times and mounted with SlowFade reagent (Molecular Probes, Eugene, OR,

USA) and coverslips. Optical sections of 1 mm were obtained for each sample using an MRC-600 (BioRad, Hercules, CA, USA) laser scanning confocal system equipped with an inverted microscope. Images were analyzed with a variable numerical aperture set of NA 1.1 and observed with the GeoLut output system (BioRad).

Particle size distribution analysis Particle size distribution was carried out by Particle Sizing Systems (Santa Barbara, CA, USA). The analysis was performed using a Model 370 Submicron Particle Sizer (Particle Sizing Systems). The simple, two-parameter Gaussian Analysis gave an approximate mean diameter and standard deviation (coefficient of variation expressed as a percentage). Transfection of cultured cells and reporter gene assays Cell lines were cultured and seeded in six-well dishes (Falcon, Franklin Lakes, NJ, USA) at a cell number per well yielding 60–80% confluency the following day. After overnight culture, the transfection mixture was added to the cells and the transgene expression was monitored 2 days later. Cells expressing b-gal were visualized by staining with B-bromo-4-chloro-3-indoyl-b-galactopyranoside (X-gal), as previously described.30 b-Gal expression was estimated by counting blue cells in three random fields of the center of the culture well and expressed as a percentage of total cells. Firely luciferase activity was measured using a luciferase assay kit (Promega). Following transfection and incubation, cells were rinsed twice with phosphate buffer saline (PBS). Lysis buffer 350 ml was added to the cells. Following 15-min incubation at room temperature, the cells were scraped from the culture plate and the cell homogenate was placed in a 1.5 ml Eppendorf tube. Another 350 ml lysis buffer was added to the culture well and after additional scraping, this rinse was added to the Eppendorf tube. Then the cell homogenate (approximate volume, 700 ml) was vortexed vigorously and centrifuged for 30 s at 10 000 g. Luciferase activity in the supernatant was quantified by using a luminometer (Biolumat LB 9500C; Berthold, Nashua, NH, USA) to integrate light emission over a 15 s reaction period. 21 The total protein concentration in the extract was measured by the BCA protein assay reagent (Pierce, Rockford, IL, USA). Luciferase activity in cultured cells was expressed as relative light units (RLU) per mg of protein (5 fg luciferase, 100 RLU). DNA stability in a biological medium DLS-1 and DLS-2 liposomes were prepared at a 0.6 DOGS:DOPE molar ratio and mixed with pCMV b-gal DNA at a 0.08 DNA:lipid weight ratio. Lipid–DNA complexes using Lipofectin (Gibco BRL) and Transfectam (DOGS; Promega) were formed under manufacturer’s conditions at a 0.2 and 0.16 DNA:lipid weight ratio, respectively. Ten micrograms of DNA with carrier were placed in an Eppendorf tube containing 150 mm NaCl and 20 mm MgCl2 in the presence or absence of 80 U of DNaseI (Boehringer Mannheim, Mannheim, Germany). The reaction mixture (400 ml) was incubated for 3 h at 37°C with rotation. Then DNA was isolated by using the phenolchloroform extraction, precipitated twice in the presence

235


236

of EtOH and 3 mm sodium acetate and was resuspended in 30 ml TE buffer. Integrity of plasmid DNA was examined by fractionation through a 1% agarose denaturing gel and ethidium bromide labeling. The study of the degradation of DNA in the presence of human serum was carried out by incubating 10 mg pCMV b-gal DNA in 150 mm NaCl with 25% human serum (vol/vol) for 5 h at 37°C with rotation. Following incubation, the reaction mixture was treated with 100 U Proteinase K (Boehringer Mannheim) in the presence of 0.05 m Tris HCl, pH 8.0, 0.5% SDS for 1 h at 60°C. Then plasmid DNA was extracted as described above and analyzed for integrity by electrophoresis.

In vivo gene delivery and transgene expression in mouse tissues All procedures were carried out in accordance with the National Institutes of Health guidelines for the care and use of laboratory animals. pBKd1RSV-luc and DLS/pBKd1RSV-luc plasmid DNA were delivered by a single injection of 100–200 ml (total volume) in the tail vein of 6-week-old female BALB/c mice. Mice were killed at 3 days after injection and mouse tissues were placed in 2-ml Eppendorf tubes and immediately frozen on dry ice, and stored at −70°C until examined. Extraction of luciferase and detection of its activity were carried out as described.21 Luciferase content was quantified from a standard curve established by analysis of RLU produced from a known amount of purified luciferase (Boehringer Mannheim). Luciferase activity was expressed as RLU per mg of protein. DNA isolation from whole blood Female Balb C (6-week-old) mice were bled from the retro-orbital artery at the indicated times after injection and killed. Blood (about 400 ml each mouse) was immediately mixed with EDTA to a final concentration of 100 mm in microcentrifuge tubes and frozen in dry ice and stored at −80°C until DNA extraction. DNA was isolated from whole blood using the Puregene DNA Isolation Kit (Gentra System, Research Triangle Park, NC, USA) with a modified procedure. Whole blood was centrifuged for 3 min at 13 000 g. The plasma fractions were transferred into the other microcentrifuge tubes, and incubated at room temperature in 300 ml cell lysis solution for 10 min. The cell pellet was incubated at 55°C in 300 ml cell lysis solution for 10 min. The cell pellet was incubated at 55°C in 300 ml lysis solution and proteinase K (final concentration 100 mg/ml) until the clots were lysed completely. Fractions of both plasma and the blood cells were treated with 100 ml protein precipitation solution. The mixtures were vortexed vigorously at high speed before centrifugation at 13 000 g for 15 min. The supernatant of the mixtures containing DNA was collected, and then extracted with equal volumes of buffered phenol. DNA was precipitated with an equal volume of 100% isopropyl alcohol, and recovered by centrifugation at 13 000 g for 15 min. The DNA was washed with 70% ethanol and air dried. The DNA pellet was rehydrated in 50 ml of TE buffer. OD260 readings were taken to determine DNA concentration. The DNA from plasma and blood cells was analyzed by 1% agarose gel electrophoresis with ethidium bromide staining.

Statistical analysis Statistical analyses were performed using Student’s t test for comparison of means. A probability of less than 0.05 was considered to be statistically significant.

Acknowledgements We thank Jean-Etienne Beneton for excellent technical assistance, Jean-Paul Behr for helpful discussion, and Anna Mazzuca for editorial assistance.

References 1 Mulligan RC. The basic science of gene therapy. Science 1993; 260: 926–932. 2 Crystal RG. Transfer of genes to humans: early lessons and obstacles to success. Science 1995; 270: 404–410. 3 Hodgson CH. The vector void in gene therapy. Biotechnology 1995; 13: 222–225. 4 Sun WH et al. In vivo cytokine gene transfer by gun reduces tumor growth in mice. Proc Natl Acad Sci USA 1995; 92: 2889– 2893. 5 Wagner E et al. Transferrin-polycation conjugates as carriers for DNA uptake into cells. Proc Natl Acad Sci USA 1990; 87: 3410– 3414. 6 Ledley FD. Nonviral gene therapy: the promise of genes as pharmaceutical products. Hum Gene Ther 1995; 6: 1129–1144. 7 Behr JP. Gene transfer with synthetic cationic amphiphiles: prospects for gene therapy. Bioconj Chem 1994; 5: 382–389. 8 Thierry AR, Tackle GB. Liposomes as a delivery system for antisense and ribozyme compounds. In: Akhtar S (ed). Delivery Strategies for Antisense Oligonucleotide Therapeutics. CRC Press: Boca Raton, 1995, pp 199–221. 9 Soriano P et al. Targeted and nontargeted liposomes for in vivo transfer to rat liver cells of a plasmid containing the preproinsulin I gene. Proc Natl Acad Sci USA 1983; 80: 7128–7131. 10 Wang CY, Huang L. pH-sensitive immunoliposomes mediate target-cell-specific delivery and controlled expression of a foreign gene in mouse. Proc Natl Acad Sci USA 1987; 84: 7851– 7855. 11 Felgner PL et al. Lipofection: a highly efficient, lipid-mediated DNA-transfection procedure. Proc Natl Acad Sci USA 1987; 84: 7413–7417. 12 Behr JP et al. Efficient gene transfer into mammalian primary endocrine cells with lipopolyamine-coated DNA. Proc Natl Acad Sci USA 1989; 86: 6982–6986. 13 Jarnagin WR et al. Cationic lipid-mediated transfection of liver cells in primary culture. Nucleic Acids Res 1992; 20: 4205–4211. 14 Legendre JY, Szoka FC. Delivery of plasmid DNA into mammalian cell lines using pH-sensitive liposomes: comparison with cationic liposomes. Pharmaceut Res 1992; 9: 1235–1242. 15 Yoshimura K et al. Expression of the human cystic fibrosis transmembrane conductance regulator gene in the mouse lung after in vivo intratracheal plasmid-mediated gene transfer. Nucleic Acids Res 1992; 20: 3233–3240. 16 Nabel GJ et al. Direct gene transfer with DNA–liposome complexes in melanoma: expression, biologic activity, and lack of toxicity in humans. Proc Natl Acad Sci USA 1993; 90: 11307– 11311. 17 Caplen NJ et al. Liposome-mediated CFTR gene transfer to the nasal epithelium of patients with cystic fibrosis. Nature Med 1995; 1: 39–46. 18 Behr JP. DNA strongly binds to micelles and vesicles containing lipopolyamines or lipointercalants. Tetrahedron Lett 1986; 27: 5861–5864. 19 Brigham KL et al. In vivo transfection of murine lungs with a functioning procaryotic gene using a liposome vehicle. Am J Med Sci 1989; 298: 278–281. 20 Zhu N et al. Systemic gene expression after intravenous DNA delivery into adult mice. Science 1993; 261: 209–211.


21 Thierry AR et al. Systemic gene therapy: biodistribution and long term expression of a transgene in mice. Proc Natl Acad Sci USA 1995; 92: 9742–9746. 22 Shuurhuis GJ et al. Quantitative determination of factors contributing to doxorubicin resistance in multidrug resistance cells. J Natl Cancer Inst 1989; 81: 1887–1892. 23 Thierry AR et al. Modulation of doxorubicin resistance in multidrug-resistant cells by liposomes. FASEB J 1993; 7: 572–579. 24 Thierry AR, Dritschilo A. Intracellular availability of unmodified, phosphorothioated and liposomally encapsulated oligodeoxynucleotides for antisense activity. Nucleic Acids Res 1992; 20: 5691–5698. 25 Schleef M, Heimann P. Cesium chloride or column preparation? An electron microscopical view of plasmid preparations. Biotechniques 1993; 14: 544. 26 Loeffler JP et al. Lipopolyamine-mediated transfection allows gene expression studies in primary neuronal cells. J Neurochem 1990; 54: 1812–1815. 27 Barthel F et al. Gene transfer optimization with liposperminecoated DNA. DNA Cell Biol 1993; 12: 553–560. 28 Pinnaduwage P et al. Use of quaternary ammonium detergent in liposome mediated DNA transfection of mouse L-cells. Biochem Biophys Acta 1989; 985: 33–37.

29 Gershon H et al. Mode of formation and structural features of DNA–cationic liposome complexes used for transfection. Biochemistry 1993; 32: 7143–7151. 30 Remy JS et al. Targeted gene transfer into hepatoma cells with lipopolyamine-condensed DNA particles presenting galactose ligands: a stage toward artificial viruses. Proc Natl Acad Sci USA 1995; 92: 1744–1748. 31 Lew D et al. Cancer gene therapy using plasmid DNA following injection in mice. Hum Gene Ther 1995; 6: 553–564. 32 Wu GY, Wu CH. Receptor-mediated in vitro gene transformation by soluble DNA carrier system. J Biol Chem 1987; 262: 4429–4432. 33 Leonetti JP et al. Antibody-targeted liposomes containing oligodeoxyribonucleotides complementary to viral RNA selectively inhibit viral replication. Proc Natl Acad Sci USA 1990; 87: 2448–2451. 34 Baudard M et al. Expression of the human multidrug resistance and glucocerebrosidase cDNAs from adeno-associated vectors: efficient promoter activity of AAV sequences and in vivo delivery via liposomes. Hum Gene Ther 1996; 7: 1309–1322. 35 Aksentijevich I et al. In vitro and in vivo liposome-mediated gene transfer leads to human MDR1 expression in mouse bone marrow progenitor cells. Hum Gene Ther 1996; 7: 1111–1122.

237