Folate-Targeted, Cationic Liposome-Mediated Gene Transfer Into Disseminated Peritoneal Tumors

Gene Therapy (2002) 9, 1542–1550  2002 Nature Publishing Group All rights reserved 0969-7128/02 $25.00 www.nature.com/gt RESEARCH ARTICLE Folate-targeted, cationic liposome-mediated gene transfer into disseminated peritoneal tumors

JA Reddy1, C Abburi1, H Hoﬂand2, SJ Howard1, I Vlahov1, P Wils2 and CP Leamon1 1Endocyte, Inc., West Lafayette, IN, USA; and 2Gencell, Aventis Pharma, Hayward, CA, USA

A folate-targeted, cationic lipid based transfection complex cated that as little as 0.01 to 0.3% of FA-Cys-PEG-PE was was developed and found to speciﬁcally transfect folate needed to produce optimal targeted expression of plasmid receptor-expressing cells and tumors. These liposomal vec- DNA. Similarly, using a disseminated intraperitoneal L1210A tors were comprised of protamine-condensed plasmid DNA, tumor model, maximum in vivo transfection activity occurred a mixture of cationic and neutral lipids, and a folic acid-cyst- with intraperitoneally administered formulations that con- eine-polyethyleneglycol-phosphatidylethanolamine (FA-Cys- tained low amounts (0.01 mol%) of the FA-Cys-PEG-PE tar- PEG-PE) conjugate. Pre-optimization studies revealed that geting lipid. Overall, folate-labeled formulations produced an inclusion of low amounts (0.01 to 0.03%) of FA-Cys-PEG- eight- to 10-fold increase in tumor-associated luciferase PE yielded the highest binding activity of expression, as compared with the corresponding non-tar- dioleoylphosphatidylcholine/cholesterol liposomes to folate geted cationic lipid/DNA formulations. These results collec- receptor-bearing cells. In contrast, higher amounts (>0.5%) tively indicate that transfection of widespread intraperitoneal of FA-Cys-PEG-PE progressively decreased cellular binding cancers can be signiﬁcantly enhanced using folate-tar- of the liposomes. In vitro studies with cationic geted techniques. lipid/dioleoylphosphatidylethanolamine formulations indi- Gene Therapy (2002) 9, 1542–1550. doi:10.1038/sj.gt.3301833

Keywords: cationic liposome; gene therapy; tumor targeting; folate receptor; lipoplex

Introduction (FA),12–15 transferrin16 or carbohydrates17 to liposomes for the purpose of combining the intrinsic activities of lipids Gene therapy has become an increasingly important with the receptor-mediated uptake properties of the strategy for treating a variety of human diseases, includ- applied ligand. ing inherited disorders, viral infections and cancer. A We have been interested in the use of folic acid (FA) number of studies have shown that introduction of vari- as a targeting ligand to deliver attached therapeutic and ous genes into tumor cells can lead to suppression of imaging agents to cancer cells that over-express the 1 tumor growth. A common method for delivering genetic receptor for FA.18–20 Because FA-linked cargos are material into cells is that which uses cationic liposomes. efficiently bound and internalized by folate receptor (FR)- 2 Since Brigham and co-workers first reported this tech- expressing cells, we have explored the possibility of nique, the use of cationic lipids complexed to plasmid using FA to enhance cationic liposomal vector delivery expression vectors has become one of the most promising to FR-enriched tumors. Thus, in the present study we 1,3,4 non-viral methods for in vivo gene delivery. investigated the opportunity for targeting a syngeneic Although cationic liposomal vectors mediate effective murine tumor in vivo by the regional administration of gene transfer, tissue specific in vivo DNA delivery is still a FA-labeled cationic liposomes. For this purpose, we first major challenge for gene therapy. To date, tissue-specific optimized and compared the amount of FA ligand targeting of cationic liposomal DNA has been required for optimal targeting of both neutral and cat- accomplished by two distinct techniques. The first ionic liposomes to FR-bearing cells. We then used L1210A method involves transfection of selected tissues, such as intraperitoneal tumor-bearing DBA mice to investigate 5 6 7,8 nasal epithelium, arterial endothelium, lung or the in vivo targeting potential of our FA-labeled vectors. tumors9 by locally administering the complexes within defined regions. This method has proven to be a viable option for the clinical treatment of several diseases, Results including cystic fibrosis and cancer.5,10,11 A second method used to enhance the specificity of gene therapy Optimization of folate-mediated liposome targeting is by coupling cell-binding ligands, such as folic acid Previous studies21 with FA-targeted liposomes had indicated that low mole fractions of a folate-polyethyleneglycol-phosphatidylethanolamine (FA-PEG-PE) conjugate Correspondence: CP Leamon, 1205 Kent Avenue, West Lafayette, IN were needed to successfully mediate liposome binding 47906, USA by FR-bearing cells. However, no investigation was ever Received 22 March 2002; accepted 16 June 2002 conducted to optimize the mole fraction of this targeting Folate-targeted gene delivery JA Reddy et al 1543

= Figure 1 Chemical structures of RPR209120, Folate-Cys-PEG3400-DSPE and Folate-Dap-Asp-Cys-PEG3400-DSPE (n 75 for PEG3400).

component. Although we would have preferred to optimize this parameter using cationic liposome formulations, we felt that their inherent propensity to non- speciﬁcally interact with cell membranes would have adversely affected our ability to measure folate-speciﬁc cell binding. Accordingly, we determined the optimal

amount of a related FA-cysteine-PEG3400-distearylphosphatidylethanolamine (FA-Cys-PEG-PE; see Figure 1) conjugate for use in 65:35 dioleoylphosphatidylcholine/ cholesterol (DOPC/Chol) liposomes. Increasing concentrations of FA-Cys-PEG-PE were formulated within calcein-loaded liposomes. Formulations were then incubated with FR-positive M109 cells to allow liposome binding, and the amount of cell-associated fluorescence was quantitated. As shown in Figure 2a, uptake of the fluorescent liposomes was found to criti- cally depend on the amount of FA-Cys-PEG-PE formulated in these vesicles. Cell-associated calcein fluorescence was found to increase with increasing concentrations of FA-Cys-PEG-PE until a maximum was reached at ~0.03 mole %. However, higher mole fractions of FA-Cys-PEG-PE surprisingly caused a reduction of liposome uptake. The aforementioned observation was unexpected, but it prompted us to search for a plausible explanation. It is known that the folate molecule can form dimers, trimers and even self-assembling tubular quartets at higher concentrations.22 Since the FR can only bind one molecule of FA,23 such self-assembled multimers of FA would be incapable of binding to the FR. Thus, we postulated that the FA moieties in liposomes containing high mole fractions of FA-Cys-PEG-PE might actually self-assemble, thereby preventing efficient interaction with the FR. To address this issue, we conducted additional experiments that were aimed to reduce or prevent the FA moieties from interacting with each other. First, we prepared

DOPC/Chol calcein liposomes with increasing amounts Figure 2 Influence of FA-Cys-PEG3400-PE and PEG2000-PE components of the non-targetable PEG3000-DOPE component plus the on PC/Chol liposome targeting. M109 cells were incubated with fluor- apparent optimum mole percent of FA-Cys-PEG-PE escent FA-derivatized liposomes for 4 h at 37°C in the presence (ᮀ)or ᭿ (0.03%), and then incubated them with cultured M109 absence ( ) of 1 mM folic acid. Cells were then washed three times in cells. As shown in Figure 2b, M109 cells had efficiently PBS, pH 7.4, lysed in 1% Triton X-100 and cell-associated fluorescence was quantitated using a fluorimeter. (a) PC/Chol liposomes formulated bound the fluorescent FA-linked liposomes, and the pres- with increasing concentrations of FA-Cys-PEG3400-PE; (b) PC/Chol lipo- ence of up to 2 mole % of the non-targetable PEG3000- somes formulated with 0.03 mol % of FA-Cys-PEG3400-PE and increasing DOPE did not reduce their affinity for the FR-positive concentrations of PEG2000-PE.

Gene Therapy Folate-targeted gene delivery JA Reddy et al 1544 cultured cells. We interpret these results to suggest that PEG-PE molecules should be available to bind to the FR the PEG component of FA-Cys-PEG-PE was not respon- when present in the liposome at Յ0.033 mole %. Interest- sible for the observed reduction in cell association of lipo- ingly, these predictions have seemingly been confirmed somes formulated with high mole percentages of FA-Cys- in our studies (see Figure 2a), where the greatest binding PEG-PE. of liposomes to FR-positive cells was observed at 0.03% Next, in an effort to reduce the potential interactions FA-Cys-PEG-PE. At concentrations greater than 0.033 of neighboring folates, perhaps via a charge repulsion mole %, the probability of folates interacting or multimer- mechanism, we synthesized a new FA-PEG-lipid conju- izing with each other likely increased with increasing gate that contained a few additional charged residues in- amounts of FA-Cys-PEG-PE, thus causing the observed between the PEG and folate moieties. Thus, the novel reduction of liposome binding to the FR-positive cells. folate-diaminopropionyl-aspartyl-cysteinyl-PEG-DOPE (FA-Dap-Asp-Cys-PEG-DOPE) component was prepared Optimization of cationic liposome/DNA complex for (see Figure 1). This conjugate was formulated at 0.01 to transgene expression in vivo 2 mole % into DOPC/Chol calcein-loaded liposomes, Before the onset of our folate-targeted in vitro and in vivo which were subsequently incubated with M109 cells for gene therapy investigations, we decided to determine the the assessment of receptor-mediated uptake. Interest- optimal formulation ratio of cationic lipid and DNA that ingly, the pattern of cell association of the FA-Dap-Asp- could manifest the greatest transfection activity in tissues. Cys-PEG-liposomes was very similar to that of FA-Cys- To accomplish this task we chose to use the novel PEG-liposomes (data not shown). Thus, decreased lipo- RPR20912026,27/DOPE formulation for DNA com- some binding to cells resulted for those formulations plexation. Formulations prepared with increasing bearing the higher amounts of the FA-Dap-Asp-Cys- lipid/DNA ratios were then injected into Balb/c mice. PEG-DOPE component. We therefore concluded that Because it is well known that lung tissue is primarily having additional negative charges located within the fol- transfected by intravenously administered cationic lipid- ate region of the targeting lipid did not change the bind- based DNA formulations,28,29 we measured the levels of ing potential of these liposomes to FR-bearing cells. In luciferase activity in the lungs of the treated mice and other words, the extra negative charges apparently failed then identified the formulation that produced the to block the stacking of pteroate moieties from neighbor- maximum transgene expression activity. As shown in ing FA-Dap-Asp-Cys ligands. Figure 3, formulations with a nmol lipid/␮g DNA ratio Although FA is the key molecule responsible for the of less than 3 yielded very large-sized particles and unde- binding of neutral or anionic liposomes to the FR, folate- tectable lung transfection activity. However, formulations dependent binding will likely be influenced by the prepared with a lipid/DNA ratio of 5 were found to pro- dynamics of the PEG molecule. Thus, we thought that an duce the greatest level of lung transfection. Interestingly, understanding of the theoretical maximum and mini- these complexes were also smaller (~100 nm for formu- mum PEG surface area coverage may help explain the lations with a lipid/DNA ratio of Ն5). At lipid/DNA observed dependence of liposome binding to cells at any ratios greater than 6, the transfection efficiency decreased given FA-Cys-PEG-PE concentration. For example, since despite the fact that the formulation particles remained FA is attached to the distal end of the flexible PEG tether, small in size. Notably, although we observed no fatalities its average position from the liposome surface should be in cohorts that received intravenous injections of the 5 ␮ equal to the Flory radius, RF, or the average PEG-coil nmol lipid/ g DNA formulation, we did observe pro- 24 0.64 dimension in solution. The RF is equal to aN , where gressively increasing fatalities with the higher ratio for- a is the size of an ethylene glycol monomer (3.5 A˚ ) and mulations (e.g. greater than 50% fatalities with Ն8 nmol N is the polymerization index (75), ie the number of eth- lipid/␮g DNA formulations). Since (1) liposomes with an ylene oxide units in the PEG chain. Hence, the RF for the ˚ PEG3400 used in our studies is equal to 55 A, which corre- sponds to a liposome surface area coverage of ~95 nm2 for each FA-Cys-PEG molecule. Assuming that the total surface area of a 100 nm liposome is approximately 3.14 × 104 square nanometers,25 approximately 330 FA-Cys-PEG molecules would be required to adequately cover the entire liposomal surface. If the total number of lipids which make up the outer leaflet of these liposomes is ~4.25 × 104 molecules,25 roughly 0.78 or greater mole % of FA-Cys-PEG-PE would be enough to theoretically allow the folates to interact with one another and possibly dimerize or multimerize, a consequence that would prevent liposomal binding to the FR. Alternatively, if all of the PEG tethers were fully extended, the maximum possible stacking distance, L, (which equals aN, or approximately 263 A˚ ), would correspond to a surface Figure 3 Optimization of cationic lipid/DNA ratio for in vitro and in area of 2173 nm2. Under this condition, only 14 FA-Cys- vivo transfections. Cationic liposomes containing equimolar mixtures of PEG-PE molecules (or the equivalent of 0.033 mol % FA- the RPR209120 lipid and DOPE were complexed at different ratios with Cys-PEG-PE in the formulation) would be needed to luciferase plasmid DNA. Twenty-four hours following the intravenous injection of 50 ␮g DNA per Balb/c mouse, lungs were excised and lucifer- completely cover the surface of a liposome. Since this ase activity (expressed as relative light units) was measured. Notably, the ‘extended’ PEG configuration has a lower chance of size of each injected complex was also measured before injection using a occurring over any finite time, all or most of the FA-Cys- particle size analyzer.

Gene Therapy Folate-targeted gene delivery JA Reddy et al 1545 average diameter of 100–200 nm have been shown to have the greatest tumor accumulation;30 (2) the endocytosis pathway can accommodate only particles with a maximal diameter of 100–150 nm;31,32 and (3) no fatalities were observed in cohorts receiving the 5 nmol lipid/␮g DNA formulations, we chose to use the 5 nmol lipid/␮g DNA formulations for the remainder of our studies. Notably, the highest luciferase-mediated light emission was detected in tissues 24 h post administration, but the activity then decreased rapidly and reduced to 50% of maximum by 48 h. Thus, for subsequent in vivo studies we chose to measure tissue transfection activity 24 h post administration.

Effect of FA-Cys-PEG-PE content on cationic lipid mediated transfection efficiency in vitro The optimal FA-Cys-PEG-PE mole fraction needed to produce serum-stable cationic-based liposome formulations is currently not known. We therefore explored this parameter by measuring the extent of luciferase gene transfection in FR-positive M109 cells and FR-negative 4T1 cells exposed to formulations containing the cationic lipid, RPR209120, DOPE, protamine, and increasing amounts of FA-Cys-PEG-PE (or PEG-PE). As shown in Figure 4a and b, transfection activity was found to decrease with increasing concentrations of PEG-PE when the gene complexes were incubated with the M109 or 4T1 cells in the presence of 10% serum. When FR-negative 4T1 cells were incubated with complexes containing FA- Cys-PEG-PE, the trend remained similar to that of complexes containing the PEG-PE component (ie as expected, the transfection activity decreased with increasing concentrations of FA-Cys-PEG-PE). In contrast, transfection activity in the FR-positive M109 cells had actually increased with complexes containing between 0.1 and 0.3% of FA-Cys-PEG-PE. In fact, transfections were 10– 15-fold higher than the corresponding non-targeted complexes when the liposome formulations contained the ‘optimum’ level of this targeting lipid. We believe that the observed decrease in activity with the non-targetable PEG-PE-containing formulations is due to the steric shielding of the vector’s cationic charge from the highly anionic cell surface proteoglycans.33 However, since the addition of FA-Cys-PEG-PE to the formulation apparently maintained or even increased the vector’s overall transfection efficiency in cells expressing high levels of the FR (M109), the PEG-dependent shielding of charge interactions may have been compensated by the high affinity binding of FA to the FRs on the tumor cells. Similar to our results with the calcein-loaded liposomes (Figure 2), we observed that liposome-dependent transfection activity was reduced to very low levels when Figure 4 Transfection of FR-positive M109 cells and FR-negative 4T1 high amounts of the PEG component (>1% of FA-Cys- cells with FA-targeted cationic lipid formulations. Liposomes consisted of PEG-PE) were included in the formulation. This sug- 50% RPR209120 without PEG (open bars) or with the indicated mole % gested that although FA helped to bind the complexes to of PEG-DOPE (hatched bars) or FA-Cys-PEG-PE (solid bars). Unmodi- fied DOPE was added to bring the total lipid content to 100%. Transfec- the FR-positive cells, high amounts of the accompanying tion particles were prepared by mixing the luciferase plasmid, pCMVluc, PEG presumably hindered the ability of the cationic lip- with protamine (1:0.8 wt/wt). The complexes were then mixed with pre- ids to interact with the endosome bilayer to effectively formed liposomes at a ratio of 5 nmoles RPR209120/␮g DNA. M109 cells release the DNA into the cytosolic milieu. The dilemma (a and c) and 4T1 cells (b) were incubated for 4 h with the transfection that such data creates is that it becomes difficult to dis- particles present at 1 ␮g DNA/ml in media containing 10% FCS (a, b) tinguish between the role of plasmid delivery from that or 60% FCS (c). Cells were analyzed for luciferase activity after 40 h. of transgene expression activity. The latter cannot occur without the former, but it indicates that successful delivery of the DNA into the cells may not guarantee efficient transgene expression. In summary, optimal transfection

Gene Therapy Folate-targeted gene delivery JA Reddy et al 1546 efﬁciency may only be achieved when the proper balance between delivery of the DNA cargo inside the cell and its release from within intracellular vesicular compart- ments (ie endosomes) is established. Because the physicochemical properties of liposome/ DNA complexes and their interactions with serum proteins and cells are not clearly understood, we decided to evaluate the transfection efﬁciency of these liposome/ DNA particles in the presence of a high level of serum. In contrast to the aforementioned transfections that were conducted in 10% serum, analogous studies conducted in the presence of 60% serum revealed that transfection activity decreased between 50- and 1000-fold (see Figure 4c). Despite this large decrease, a three- to 400-fold level of FA-targeted enhancement was still observed with the FA-Cys-PEG-PE containing formulations relative to the PEG-PE-containing control formulations. This apparent serum-dependent reduction in transfection activity could be explained if the DNA complexes became completely coated with serum proteins, a consequence that would have changed the net surface charge on the cationic liposome/DNA complexes from positive to negative.34 Thus, while serum proteins may have coated the non- targeted cationic vector complexes and reduced their intrinsic electrostatic cellular binding ability, the ~250A˚ PEG spacer between FA and the lipid probably enabled the ligand to penetrate through the protein coating and made it available to bind FRs present on the cell surface.

Luciferase expression in L1210A tumors In our laboratory, the L1210A intraperitoneal cancer in DBA mice seems to be similar to the spread of ovarian cancer in patients with advanced disease. In other words, Figure 5 Transfection of intraperitoneal L1210A tumors and ascites- widespread or disseminated cancer and ascites fluid derived cells with FA-targeted liposome–DNA complexes. Ten days fol- develop in the peritoneal cavity. This property makes it a lowing an i.p. inoculation of L1210A cells, groups of five mice received i.p. injections of liposome–DNA complexes containing 250 nmol of useful model to investigate novel experimental therapies. ␮ × 4 RPR209120/DOPE liposomes complexed to 50 g of a luciferase-contain- Thus, we have noted that the inoculation of 2 10 ing plasmid. Mice were killed 24 h later, and ascites cells and selected L1210A tumor cells into peritoneal cavity of mice consist- tissues were harvested and assayed for luciferase activity, as described in ently produced ascites fluid, large intra-abdominal Materials and methods section. (a) Luciferase activity in L1210A ascites- tumors (~200 mg), sub-diaphragmatic tumor deposits derived cells; (b) Luciferase activity in L1210A solid tumor and other and eventually death within ~16 days. peritoneal organs from mice treated with RPR209120/DOPE (open bars), Having recognized that FA-Cys-PEG-PE enhances cat- RPR209120/DOPE/PEG-DOPE (hatched bars), or RPR209120/ DOPE/ FA-Cys-PEG-DOPE (solid bars). ionic lipid/DNA transgene expression in vitro (relative to non-targeted complexes) only at low concentrations, we decided to evaluate the performance of these complexes tumor cell ascites followed by those with solid tumors. in DBA mice bearing L1210A i.p. tumors. As shown in In both tumor forms, FA-Cys-PEG-PE increased the Figure 5a, i.p. injections with lipid/DNA complexes con- transfection activity over that of non-targeted complexes taining 0.01% FA-Cys-PEG-PE yielded the highest lucifer- by eight- to 10-fold. Importantly, very low or even no ase activity (125 ng/mg protein) in the tumor ascites frac- luciferase activity could be detected in the surrounding tion. This activity was approximately 10-fold lower when peritoneal organs such as the spleen, liver and intestine the liposomes did not contain a targeting moiety (i.e. (see Discussion). Thus, by using a combination of FA tar- when FA-Cys-PEG-PE was replaced by PEG-PE in the geting and regional administration of the complexes, an complex). With complexes containing higher concen- approximate 12–50-fold increase in transfection activity trations of FA-Cys-PEG-PE (0.03–0.3%), the level of of solid tumor tissue and a 500–1000-fold increase in luciferase expression observed was lower (16.5 ng/mg ascites tumor cells could be achieved relative to the protein) than that obtained with complexes containing surrounding peritoneal organs. 0.01% of the targeting lipid. Yet, even with complexes containing 0.03 and 0.1% FA-Cys-PEG-PE, there was still Discussion a six- to nine-fold increase in transfection activity over their corresponding non-targeted complexes containing Liposome-based delivery systems have long been ident- only PEG-PE. ified as safe and effective drug carriers that are amenable Luciferase activity was also measured in tumors and to modification with cell-specific targeting ligands. In various peritoneal organs after the i.p. injection of regards to gene delivery, the most common liposomes are RPR209120/DOPE/DNA complexes. As shown in Figure those prepared with various cationic lipids in combi- 5b, the highest luciferase activities were observed in nation with a fusogenic lipid, such as DOPE.35,36

Gene Therapy Folate-targeted gene delivery JA Reddy et al 1547 For our investigation, we prepared FA-targeted lipo- liposomes. Yet at 60% serum concentration, the highest somes where the folate moiety was attached to a lipid transfections were observed with 0.01% FA-Cys-PEG-PE membrane anchor via a cysteinyl-PEG3400 spacer. The (ie similar to the PC/Chol-based formulations). Aside Cys-PEG spacer was employed for two reasons. First, it from the probability that the folates could self-associate was known that effective targeting of liposomes to cancer at high concentrations, it is also very possible that FR 21 cells resulted when a 250 A˚ PEG3400 spacer was used. interactions were perturbed in the presence of both the Secondly, it was recently shown that having an cationic lipids and the serum. In other words, the cationic additional negative charge (such as the cysteinyl car- lipids in the formulation could have affected the presen- boxylic acid) near the para-aminobenzoyl moiety of FA tation of folate molecules. However, such interference enhanced the binding of corresponding conjugates to FR- could have been reduced at high serum levels, where bearing cells.37 serum proteins could more effectively ‘coat’ or block the We began our investigations by closely examining the cationic charges of the lipids. ability of FA to target PC/Chol liposomes to FR-positive When targeted cationic liposomes were tested in the cancer cells under conditions where the mole percentage L1210A intraperitoneal tumor model, the greatest tar- of the FA-Cys-PEG-PE component was varied over a geted transfections were again produced with formu- wide range. Surprisingly, it was found that the greatest lations containing low amounts (0.01%) of FA-Cys-PEG- cell association occurred when the liposomes were for- PE (ie similar to what we observed with cells incubated mulated with very low amounts of FA-Cys-PEG-PE (less in 60% serum in vitro). As reported elsewhere,38,39 non- than 0.03%). Theoretical calculations revealed that the fol- targeted and pegylated control cationic liposomes were ate moieties located at the distal end of the PEG spacers found to transfect the luciferase gene into the intraperi- would likely not interact with each other at concen- toneal tumors and not into the other peritoneal organs. trations lower than 0.033%; but, at concentrations greater The lack of efficient transfection of normal peritoneal than 0.033%, the folate moieties of the FA-Cys-PEG-PE organs is not that surprising because these organs are conjugate would be close enough to interact by possibly known to be covered by the peritoneum and underlying dimerizing or multimerizing with each other. This would connective tissue, both of which constitute additional consequently lead to a decrease of liposome binding to barriers against the entry of lipid/DNA complexes FR-positive cells. Although no direct proof for folate self- through the organ surface.40,41 In contrast, a disseminated association was obtained in this investigation, these peritoneal tumor is devoid of such coverings, so the can- theoretical predictions were supported experimentally cer cells are more easily accessible to the formulations.40,41 since cell-associated liposomal fluorescence was observed Additionally, the transduced gene is more readily to progressively decrease at higher FA-Cys-PEG-PE con- expressed in the rapidly proliferating cancer cells rather centrations. than in the normal cells that display low mitogenic It has previously been shown that co-incubation of fol- activity.42 Thus, when injected i.p., the cationic liposomes ate-targeted liposomes with very high amounts (~1 mM) can efficiently target peritoneal tumors. We found that in of folic acid is required to significantly reduce FR-specific all cases folate targeting increased the transfection cellular liposome uptake, suggesting that folate lipo- activity in the L1210A tumor by at least 10-fold over that somes may have an unusually high affinity for FR-posi- produced by the non-targeted control formulations. Fur- tive cells.21 Indeed, we have observed similar results with thermore, the transgene was preferentially expressed in our folate-targeted PC/cholesterol liposomes (data not both the ascites cells and solid L1210A tumor tissue, shown). This apparent ‘elevated‘ affinity has been whereas the gene expression was very low or undetect- explained by the fact that the large size of the liposomes able in the various organs located in the peritoneal and the PEG spacers enable each liposome to bind to region. Interestingly, we did observe low levels of trans- more than one cell surface FR, thus yielding a multivalent fection activity in the spleens from animals receiving the attachment. Interestingly, we have observed that when 1 folate-targeted liposomes (see Figure 5b); however, this mM of folic acid was used to compete with our folate- was not surprising because the FR-positive L1210A targeted cationic liposomes, transfections were reduced tumor cells are known to metastasize to the spleen. In by as much as 50-fold. However, when transfections of fact, we have confirmed this phenomenon by visually non-targeted liposomes were tested in the presence of 1 detecting L1201A tumor cells in spleens (using laser- mM folic acid, a 10-fold reduction in ‘non-specific’ trans- assisted optical imaging techniques) of mice injected with fection was also observed. We interpreted these findings a folate-fluorescein molecule (manuscript submitted). to suggest that although most of the reduction in folate- Collectively, these data suggest that although non-tar- targeted transfections in the presence of high amounts geted cationic liposomes facilitate the uptake and of folic acid is due to reduced binding to the FRs, high expression of DNA in tumor cells in vitro and in vivo, concentrations of folic acid could also produce a decrease incorporation of the FA-Cys-PEG-PE targeting ligand in cationic lipid/DNA complex stability (perhaps in a within the liposomal formulation clearly enhances the charge-dependent manner). Therefore, as a control for overall tumor transfection efficiency. our gene transfection experiments, we have preferred to It was recently shown that both the Marburg and Ebola use non-targeted complexes which are identically formu- viruses can transduce a wide variety of normal cell types lated to the targeted complexes except they lack the folate specifically through their FRs.43 Yet we have found that moiety at the end of the PEG linker. folate-targeted lipid/DNA particles apparently do not When the effect of FA-Cys-PEG-PE concentration was efficiently transfect normal tissues. One possible expla- tested with the cationic lipid/DNA formulation in the nation for this discrepancy is that these viruses have presence of 10% serum, the highest targeting and trans- evolved into the most efficient transduction systems fection efficiency was observed at slightly higher levels known to man. They are highly effective in their ability (0.1 to 0.3%) than that required for the PC/Chol-based to (1) bind cells; (2) release their nucleic acid from within

Gene Therapy Folate-targeted gene delivery JA Reddy et al 1548 the endosome and into the target cell’s cytoplasm; and reduced pressure. The concentrated solution was next (3) further transport their genetic material into the host added drop-wise to 100 ml of ether, with constant stirring cell’s nucleus. Compounded with these features is the and then placed on ice for 10 min. Following a 2000 g viruses’ ability to replicate within the compromised cells centrifugation step for 10 min at 4°C, the crude precipi- and then spread to other neighboring cells. Therefore, tate was washed twice with excess ether. The crude pow- both Marburg and Ebola virons could conceivably trans- der was then dissolved in ~5 ml of deionized water to duce cells that express low (or barely detectable) levels form a clear yellow micellar solution. This solution was of the FR. Although synthetic lipid-based vectors (such dialyzed for three cycles in 100 kDa molecular weight as the aforementioned folate-targeted formulations) have cutoff tubing at room temperature against 200 volumes been designed to mimic some of these viral features, the of deionized water for a minimum of 8 h per cycle. The high transfection efficiency of viruses has not yet been retentate was then lyophilized to dryness, and purity was achieved. Thus, it is our belief that detection of any sig- confirmed by thin layer chromatography (FA-Cys-PEG- = nificant amount of gene expression is only possible in PE: Rf 0.3 (silica gel in 65:25:4 cells that express relatively high levels of the FR. Conse- chloroform/methanol/water) and FA-Dap-Asp-Cys- = quently, the two possible target tissues that do express PEG-PE: Rf 0.2 (cellulose in 90:10 methanol/water). high amounts of the FR in our model are the kidneys and the tumor. Although FRs in the tumor are accessible to Liposome preparation our folate-targeted formulations, the same is not true FA-derivatized liposomes were constructed by incorpor- with the kidneys since large complexes (ie lipid/DNA ating the FA tethered lipid and non-targeted PEG-DOPE particles) are not filtered through the kidneys and thus at the required mole percentages in a 65:35 molar mixture never come into contact with the apically oriented FRs of of egg phosphatidylcholine (PC) and cholesterol (Chol). the proximal tubules.18,44 Briefly, 15 mg of PC, 4.2 mg of Chol (Avanti Polar Lipids, In conclusion, we have confirmed the observations of Alabaster, AL, USA), FA-lipid and/or DSPE-PEG were other laboratories that folate-mediated endocytosis can dissolved in 1 ml of chloroform. The lipids were dried be successfully exploited to target transgenes to FR- under reduced pressure to form a thin film and then expressing cancer cells. We have also demonstrated that hydrated in 0.6 ml of 20 mM calcein in PBS, pH 7.4. The having a targeting ligand, such as FA-Cys-PEG-PE, in suspension was subjected to 10 cycles of freezing and cationic liposome formulations can significantly improve thawing and then extruded 10 times through a 100 nm the transfection efficiency of a cationic liposomal formu- polycarbonate membrane (Nuclepore, Pleasanton, CA, lation. By carefully formulating the liposomal vector with USA). The resulting calcein-loaded liposomes were then an optimal amount of FA-Cys-PEG-PE, efficient trans- separated from free calcein on a Sepharose CL-4B column gene expression was possible in FR-bearing cells both in (Pharmacia, Uppsala, Sweden) pre-equilibrated with PBS. vitro and in vivo. Since peritoneal cancers often become The final lipid concentration was ~20 mg/ml. disseminated during the advanced stages of tumor pro- gression, particularly in ovarian cancer, FA-targeted cat- Uptake of FA-Cys-PEG-PE containing liposomes by ionic lipid gene therapy may offer a more effective alter- cells native to conventional chemotherapy. Fifty ␮l of calcein-loaded FA-Cys-PEG-PE containing liposomes were diluted in 1 ml of FDMEM and then added to monolayers of M109 cells. The cells were incu- Materials and methods bated at 37°C for 4 h, washed three times with PBS, and then examined under an Olympus phase-contrast/ Lipid synthesis fluorescence microscope. To quantitatively determine the RPR209120 was synthesized as previously described.26,27 cell-associated fluorescence, cells were lysed in 1 ml of

Synthesis of folic acid-cysteine-PEG3400-distearylphos- lysis buffer (1% Triton X-100 in PBS), and the lysis buffer phatidylethanolamine (FA-Cys-PEG-PE) or folic acid- extract was measured in an Aminco Bowman Series 2 Dap-Asp-cysteine-PEG3400-distearylphosphatidylethanolamine fluorescence spectrophotometer. Protein content was esti- (FA-Dap-Asp-PEG-PE) was achieved by placing 110 mg mated by using the bicinchonic acid protein assay (Pierce, ␮ (147 mol in 5.5 ml CHCl3) of 1,2-distearoyl-sn-glycerol- Rockford, IL, USA), and results were expressed as 3-phosphoethanolamine plus 10 ml of additional CHCl3 fluorescence units per mg of lysate protein. in a 50 ml round bottom flask. Two equivalents of triethy-

lamine were added along with 500 mg NHS-PEG3,400- Plasmid DNA MAL powder (147 ␮mol; Shearwater Polymers, Hunts- Plasmid DNA contained the luciferase reporter gene ville, AL, USA). The solution was heated at 60°C for 30 under the CMV promoter. Purified plasmid DNA met the s to clear the solution, and it was then stirred for 2 h following quality criteria: Endotoxin levels Ͻ20 EU/mg; at room temperature. The progress of the reaction was amount supercoiled DNA >90%; E. coli DNA contami- followed by thin layer chromatography using nation Ͻ5%; RNA contamination Ͻ5%; protein contami- chloroform/methanol/water (65:25:4) as eluant. This step nation Ͻ1%. typically proceeds to near completion and the product need not be further purified. A 1.5-fold molar excess of Preparation of cationic liposome/DNA complexes folate-␥-cysteine (120 mg)37 or FA-Dap-Asp-Cys (109 mg) Liposomes for transfection studies were prepared by dry- along with five additional equivalents of triethylamine ing a chloroform solution of RPR209120 (2 ␮moles), ␮ were added to the flask, and the mixture was stirred DOPE (2 moles) and FA-Cys-PEG3400-PE or PEG3000- overnight under argon at room temperature. The solution DOPE, under vacuum to a thin film. The film was was filtered through a fine Buchner filter apparatus, and hydrated by addition of 1.0 ml of filter-sterilized 5% dex- the filtrate was evaporated to minimal volume under trose in 5 mM HEPES, pH 8.0. Liposomes were prepared

Gene Therapy Folate-targeted gene delivery JA Reddy et al 1549 by first vortexing the hydrated lipid for 1 min and then centrifugation of the peritoneal lavage fluid at 400 g for subjecting it to sonication in a bath sonicator until a clear 10 min. Tumor, liver, spleen and intestine were also emulsion was obtained. DNA was diluted in distilled excised, weighed and stored at Ϫ70°C until assayed for water and mixed with diluted protamine at a weight ratio luciferase activity. Tissue extracts were prepared by of 0.8 mg of protamine per mg of DNA.45 The mixture adding reporter lysis buffer at a volume to weight ratio was allowed to stand at room temperature for 10 min. of 5 ␮l/mg of the collected organs. Each organ was Liposome/protamine/DNA complexes were induced to homogenized for 1 min on ice utilizing a PowerGen 125 self-assemble by vortexing equal volumes of the unit (Fisher Scientific, Racine, WI, USA) at maximum RPR20/DOPE suspension and appropriate amounts of speed (20 000 r.p.m). Samples were then subjected to DNA/protamine mixture dissolved in sterile 5% dex- three freeze/thaw cycles, spun for 10 min at 12 000 r.p.m. trose, 5 mM HEPES, pH 8 for 1 min, to yield a final con- in a microcentrifuge, and 20 ␮l of the supernatant was centration of 5 nmoles RPR209120 per ␮g of DNA. The analyzed for luciferase activity. Protein content was esti- sizes of the complexes were measured at 20°Cby mated by using the bicinchonic acid protein assay, and dynamic laser light-scattering detector using a Coulter results were expressed as fluorescence units per mg of N4plus particle sizer (Coulter Electronics, Hialeah, FL, lysate protein. USA). The measurements were made at a scattering angle of 90 after dilution of the particles in 5% dextrose. Acknowledgements Transfections We wish to thank Dr Philip S Low and Dr George Chou M109 cells were seeded at a density of 1 × 104 cells/well for their valuable discussions. We also thank Marilynn in 96-well plates (Corning Costar, Cambridge, MA, USA) Vetzel, Nikki Douglas, Marc Frederic and Shaff Iginla for and maintained for 40–48 h before transfection in folate their technical help. deficient-RPMI medium 1640 (FDRPMI; Life Techno- logies, Carlsbad, CA, USA) supplemented with 10% heat- inactivated fetal bovine serum (HIFBS). RPR209120/ References DOPE liposomes in 5% dextrose were diluted with 1 Dass CR, Burton MA. Lipoplexes and tumours. A review. J FDRPMI containing either 10 or 60% FBS or 60% HIFBS, Pharm Pharmacol 1999; 51: 755–770. to a final concentration of 1 ␮g of DNA per ml for trans- 2 Brigham KL et al. In vivo transfection of murine lungs with a fection. After 4 h of transfection, the medium was functioning prokaryotic gene using a liposome vehicle. Am J removed and replaced with fresh medium containing Med Sci 1989; 298: 278–281. 10% HIFBS, and the cells were maintained for an 3 Felgner PL et al. Improved cationic lipid formulations for in vivo gene therapy. Ann NY Acad Sci 1995; 772: 126–139. additional 40 h. Cells were harvested using 50 ␮lof 4 Templeton NS, Lasic DD. New directions in liposome gene reporter lysis buffer, and the luciferase activity was delivery. Mol Biotechnol 1999; 11: 175–180. determined using the Luciferase Assay Reagent 5 Caplen NJ et al. Liposome-mediated CFTR gene transfer to the (Promega, Madison, WI, USA) and a 96-well MLX nasal epithelium of patients with cystic fibrosis. Nat Med 1995; microtiter plate luminometer (Dynex Technologies, 1:39–46. Chantilly, VA, USA). 6 Stephan DJ et al. A new cationic liposome DNA complex enhances the efficiency of arterial gene transfer in vivo. Hum Tumor cell inoculation and in vivo administration Gene Ther 1996; 7: 1803–1812. Female DBA mice, 6–8-week-old (Harlan, Indianapolis, 7 Wheeler CJ et al. A novel cationic lipid greatly enhances plasmid IN, USA), were maintained on a standard 12-h light–dark DNA delivery and expression in mouse lung. Proc Natl Acad Sci USA 1996; 93: 11454–11459. cycle and fed ad libitum with folate-deficient chow (Dyets, 8 Alton EW et al. Non-invasive liposome-mediated gene delivery Bethlehem, PA, USA). Since normal rodent chow contains can correct the ion transport defect in cystic fibrosis mutant a high concentration of folic acid (6 mg/kg chow), mice mice. Nat Genet 1993; 5: 135–142. used in these studies were maintained on the folate-free 9 Parker SE et al. Plasmid DNA gene therapy: studies with the diet for 2 weeks before tumor implantation to achieve human interleukin-2 gene in tumor cells in vitro and in the serum folate concentrations close to the range of normal murine B16 melanoma model in vivo. Cancer Gene Ther 1996; 3: human serum.46 FR-positive murine L1210A leukemia 175–185. cells were grown in FDRPMI 1640 with 10% HIFBS at 10 Stopeck AT et al. Phase I study of direct gene transfer of an 37°C in a 5% CO humidified atmosphere. For tumor cell allogeneic histocompatibility antigen, HLA-B7, in patients with 2 metastatic melanoma. J Clin Oncol 1997; 15: 341–349. inoculation, 2 × 104 L1210A cells in 400 ␮l were injected 11 Rubin J et al. Phase I study of immunotherapy of hepatic metast- intraperitoneally per mouse. ases of colorectal carcinoma by direct gene transfer of an allogeneic histocompatibility antigen, HLA-B7. Gene Therapy 1997; In vivo transfections and analysis of luciferase activity 4: 419–425. Liposome–DNA complexes (400 ␮l) were injected into 12 Lee RJ, Huang L. Folate-targeted, anionic liposome-entrapped tumor-bearing mice 10 days after tumor cell inoculation, polylysine-condensed DNA for tumor cell-specific gene transfer. with each mouse receiving 50 ␮g of plasmid DNA. The J Biol Chem 1996; 271: 8481–8487. DNA:lipid ratio (␮g of DNA/nmol of total lipid) was 1:5 13 Reddy JA, Low PS. Enhanced folate receptor mediated gene for all RPR209120:DOPE liposomes. These DNA:lipid therapy using a novel pH-sensitive lipid formulation. J Control ratios were selected on the basis of previous studies that Release 2000; 64:27–37. 14 Reddy JA, Dean D, Kennedy MD, Low PS. Optimization of fol- suggested that optimal expression of cells in vitro ate-conjugated liposomal vectors for folate receptor-mediated occurred at ratios between 1:4 and 1:7 (data not shown). gene therapy. J Pharm Sci 1999; 88: 1112–1118. Twenty-four hours after injection, each mouse was killed 15 Xu L et al. Systemic p53 gene therapy in combination with radi- and its peritoneal cavity was lavaged with 8 ml of Hanks’ ation results in human tumor regression. Tumor Targeting 1999; balanced salt solution. Ascites cells were recovered by 4:92–104.

Gene Therapy Folate-targeted gene delivery JA Reddy et al 1550 16 Xu L et al. Transferrin-liposome-mediated systemic p53 gene 31 Straubinger RM, Hong K, Friend DS, Papahadjopoulos D. Endo- therapy in combination with radiation results in regression of cytosis of liposomes and intracellular fate of encapsulated mol- human head and neck cancer xenografts. Hum Gene Ther 1999; ecules: encounter with a low pH compartment after internaliz- 10: 2941–2952. ation in coated vesicles. Cell 1983; 32: 1069–1079. 17 Hwang SH, Hayashi K, Takayama K, Maitani Y. Liver-targeted 32 Watts C, Marsh M. Endocytosis: what goes in and how? J Cell gene transfer into a human hepatoblastoma cell line and in vivo Sci 1992; 103:1–8. by sterylglucoside-containing cationic liposomes. Gene Therapy 33 Mounkes LC et al. Proteoglycans mediate cationic liposome- 2001; 8: 1276–1280. DNA complex-based gene delivery in vitro and in vivo. J Biol 18 Leamon CP, Low PS. Folate-mediated targeting: from diagnos- Chem 1998; 273: 26164–26170. tics to drug and gene delivery. Drug Discov Today 2001; 6:44–51. 34 Zelphati O, Uyechi LS, Barron LG, Szoka Jr. FC. Effect of serum 19 Reddy JA, Low PS. Folate-mediated targeting of therapeutic and components on the physico-chemical properties of cationic imaging agents to cancers. Crit Rev Ther Drug Carrier Syst 1998; lipid/oligonucleotide complexes and on their interactions with 15: 587–627. cells. Biochim Biophys Acta 1998; 1390: 119–133. 20 Leamon CP, Pastan I, Low PS. Cytotoxicity of folate-Pseudo- 35 Gao X, Huang L. Cationic liposome-mediated gene transfer. monas exotoxin conjugates toward tumor cells. Contribution of Gene Therapy 1995; 2: 710–722. translocation domain. J Biol Chem 1993; 268: 24847–24854. 36 Fasbender A et al. Effect of co-lipids in enhancing cationic lipid- 21 Lee RJ, Low PS. Delivery of liposomes into cultured KB cells mediated gene transfer in vitro and in vivo. Gene Therapy 1997; via folate receptor-mediated endocytosis. J Biol Chem 1994; 269: 4: 716–725. 3198–3204. 37 Leamon CP, DePrince RB, Hendren RW. Folate-mediated drug 22 Ciuchi F et al. Self-recognition and self-assembly of folic acid delivery: effect of alternative conjugation chemistry. J Drug Tar- get 1999; 7: 157–169. salts: columnar liquid crystalline polymorphism and the column 38 Aoki K et al. Gene therapy for peritoneal dissemination of pan- growth process. J Am Chem Soc 1994; 116: 7064. creatic cancer by liposome-mediated transfer of herpes simplex 23 Antony AC, Utley C, Van Horne KC, Kolhouse JF. Isolation and virus thymidine kinase gene. Hum Gene Ther 1997; 8: 1105–1113. characterization of a folate receptor from human placenta. J Biol 39 Kikuchi A et al. Development of novel cationic liposomes for Chem 1981; 256: 9684–9692. efficient gene transfer into peritoneal disseminated tumor. Hum 24 Jeppesen C et al. Impact of polymer tether length on multiple Gene Ther 1999; 10: 947–955. ligand-receptor bond formation. Science 2001; 293: 465–468. 40 Kiyasu Y, Kaneshima S, Koga S. Morphogenesis of peritoneal 25 Torchilin VP. Molecular mechanism of liposome and immunoli- metastasis in human gastric cancer. Cancer Res 1981; 41: 1236– posome steric protection with poly(ethylene glycol): theoretical 1239. and experimental proofs of polymer chain flexibility. In: Lasic 41 Niedbala MJ, Crickard K, Bernacki RJ. Interactions of human DDaMF (ed.). Stealth Liposomes. CRC Press: Boca Raton, 1995, ovarian tumor cells with human mesothelial cells grown on pp 51–62. extracellular matrix. An in vitro model system for studying 26 Byk G, Scherman D, Schwartz B, Dubertret C. Lipopolyamines tumor cell adhesion and invasion. Exp Cell Res 1985; 160: 499– as transfection agents and pharmaceutical uses thereof: U.S. pat- 513. ent, 2001. 42 Takeshita S et al. Increased gene expression after liposome- 27 Pitard B. Sterically stabilized BGTC-based lipoplexes: structural mediated arterial gene transfer associated with intimal smooth features and gene transfection into the mouse airways in vivo. J muscle cell proliferation. In vitro and in vivo findings in a rabbit Gene Med 2001 (in press). model of vascular injury. J Clin Invest 1994; 93: 652–661. 28 Zhu N, Liggitt D, Liu Y, Debs R. Systemic gene expression after 43 Chan SY et al. Folate receptor-alpha is a cofactor for cellular intravenous DNA delivery into adult mice. Science 1993; 261: entry by Marburg and Ebola viruses. Cell 2001; 106: 117–126. 209–211. 44 Deen WM, Bridges CR, Brenner BM. Biophysical basis of glom- 29 Liu Y et al. Factors influencing the efficiency of cationic lipo- erular permselectivity. J Membr Biol 1983; 71:1–10. some-mediated intravenous gene delivery. Nat Biotechnol 1997; 45 Li S, Huang L. In vivo gene transfer via intravenous adminis- 15: 167–173. tration of cationic lipid-protamine-DNA (LPD) complexes. Gene 30 Ishida O, Maruyama K, Sasaki K, Iwatsuru M. Size-dependent Therapy 1997; 4: 891–900. extravasation and interstitial localization of polyethyleneglycol 46 Mathias CJ et al. Tumor-selective radiopharmaceutical targeting liposomes in solid tumor-bearing mice. Int J Pharm 1999; 190: via receptor-mediated endocytosis of gallium-67-deferoxamine- 49–56. folate. J Nucl Med 1996; 37: 1003–1008.

Gene Therapy