Design, Synthesis, and Evaluation of Gadolinium Cationic Lipids As

Histology, University of Verona, Verona, Italy, and UMR 7001 (CNRS/ENSCP/Aventis), E.N.S.C.P., 11 rue P. &. M. Curie, 75231-Paris Cedex, France. Recei...
0 downloads 0 Views 231KB Size
Download PDF
112

Bioconjugate Chem. 2003, 14, 112−119

Design, Synthesis, and Evaluation of Gadolinium Cationic Lipids As Tools for Biodistribution Studies of Gene Delivery Complexes Francoise Leclercq,† Mirit Cohen-Ohana,‡ Nathalie Mignet,† Andrea Sbarbati,§ Jean Herscovici,*,† Daniel Scherman,† and Gerardo Byk*,‡ Bar Ilan University, Department of Chemistry, Laboratory of Peptidomimetics and Genetic Chemistry, 52900-Ramat Gan, Israel, Department of Morphological and Biomedical Sciences, Section of Anatomy and Histology, University of Verona, Verona, Italy, and UMR 7001 (CNRS/ENSCP/Aventis), E.N.S.C.P., 11 rue P. & M. Curie, 75231-Paris Cedex, France. Received July 3, 2002; Revised Manuscript Received October 28, 2002

Gadolinium-chelating cationic lipids have been synthesized to obtain lipoplexes with MRI contrast properties. These compounds were designed to follow the biodistribution of synthetic DNA for gene delivery by nuclear magnetic resonance imaging. The lipid MCO-I-68 was synthesized, and chelate complexes with gadolinium were formed and characterized in terms of physicochemical and DNA binding properties. The transfection activity of MCO-I-68-Gd/DNA complexes was assayed in vitro on NIH 3T3. Different formulations of the product were tested. When up to 5% of the gadolinium lipid complexes were co-formulated with the cationic lipid RPR120535 used as a reference, the transfection levels were maintained as compared to RPR120535 alone. To date, only a liposomal formulation of a gadolinium-cationic lipid chelate without DNA had been observed using magnetic resonance imaging. In vivo intratumoral administration of MCO-I-68-Gd/DNA lipoplexes to tumor model led to an important increase of the NMR signal. It was demonstrated that the new complexes also acted as transfection carriers when they were formulated from liposomes.

INTRODUCTION

The different self-assembling systems presented since the pioneering works of Felgner (1) on nonviral approaches for gene delivery have explored the nature and geometry of the cationic moiety, the type of lipid, the biodegradability properties, the controlled release of DNA from complexes, and the introduction of some targeting elements (2-3). During the last five years, a great effort has been devoted to the elucidation of the physicochemical properties of the different cationic lipid/DNA complexes (named lipoplexes). Moreover, it seems increasingly crucial to develop efficiently tissue-targeted lipoplexes, taking advantage of the growing number of tissue targeting molecules emerging from phage libraries (4) and other selection technologies. Histological processing commonly monitors visualization of gene expression in opaque tissues. Noninvasive techniques such as PET (positron emission tomography), γ cameras, and SPECT (single-photon emission tomography) operate in the range of cubic millimeters or larger. Magnetic resonance imaging (MRI) provides an alternative to these technologies and its use for noninvasive visualization of gene delivery complexes is an exciting emerging technique (5-9). In particular, it was studied with contrast agents whose activity depends of an enzymatic process (7-8). With the aim to follow the biodistribution of lipoplexes and correlate their transfecting ability with in vivo biodistribution, we present * Corresponding authors. G.B.: e-mail, [email protected]; phone, (972)-3-5318325; fax, (972)-3-5351250. J.H.: e-mail, [email protected]; phone: (33)-001-53-10-12-95; fax, (33)001-53-10-12-92. † UMR 7001. ‡ Bar Ilan University. § University of Verona.

here a new strategy based on magnetic resonance imaging using gadolinium-chelating cationic lipid associated to DNA. Our approach included the design and synthesis of a new building block precursor of DTPA in which the carboxylic acids in DTPA are protected as tert-butyl esters and only one of them remains free for coupling to the cationic lipid. This synthesis could be accomplished using a solid-phase methodology. This precursor was coupled to a free amino group placed in the side chain of the protected cationic lipid backbone to result in the desired cationic lipid bearing a gadolinium-complexing moiety as side chain. EXPERIMENTAL PROCEDURES

General. 1H NMR spectra were recorded on Bruker 300 and 600 MHz spectrometers. Samples were dissolved in CDCl3 or CD3OD. Chemical shifts are in parts per million relative to TMS internal standard. We have performed a complete NMR analysis of the final product MCO-I-68, including 1H NMR recorded with an Avance DMX 600 Bruker (2D, COSY, and HOHAHA methods, t ) 40ms) and 13C NMR (hetereonuclear 2D techniques HMQC and HMBC with a delay of 3.45 and 60 ms, respectively, in the reverse mode). Analytical and preparative HPLC were performed on Waters HPLC system equipped with a 717-plus autosampler, a 600-controller pump, a 996-photodiode array detector with tunable wavelength set at 220 nm, and a fraction collector Gilson 202. Mobile phases were (A) H2O (0.1%TFA) and (B) MeCN (0.08% TFA). Separation conditions are as follows. Analytical. Method A: column C18 Vydac-218TP-54, gradient H2O/MeCN, 3 min [80/ 20], 3-25 min [0/100], 25-50 min [0/100], 51min [80/ 20]; flow, 1 mL/min. Method B: Column BU-300 aquapore Butyl 7 m, 30 × 4.6 mm from Perkin-Elmer,

10.1021/bc025567e CCC: $25.00 © 2003 American Chemical Society Published on Web 12/19/2002

Gadolinium Cationic Lipids

gradient H2O/MeCN, 3 min [80/20], 3-25 min [0/100], 25-35 min [0/100], 36 min [80/20]; flow, 1 mL/min. Method C: Column BU-300 aquapore Butyl 7 m, 30 × 4.6 mm, 3 min [100/0], 3-20 min [50/50], 20-30 min [0/100], 30-40 min [0/100], 41 min [100/0]; flow,1 mL/ min. Preparative. Method D: column C4 Vydac-214TP1022, 250 × 26 mm gradient H2O/MeCN, 3 min [60/40], 3-10 min [38/62], 10-15min [34/66], 15-20 min [30/70], 2030 min [60/40]; flow, 10 mL/min. Fluorescence measurements were carried out on a Jobin-Yvon Spex fluoromax-2 spectrofluorimeter (Longjumeau, France). The size distribution of the liposomes and the lipoplexes was determined by dynamic light scattering using a coulter N4 Plus particle sizer (Coulter, Margency, France). Solid supports were purchased from Nova-Biochem (Switzerland), polyamine, and other reagents and solvents from Aldrich. Solvents for HPLC were purchased from Merck (Germany), and DOPE was purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). Solid-Phase Synthesis of Protected DTPA Harboring a Single Free Carboxylate (1). 2-Chloro trityl chloride resin (2 g, 1.5 mmol) was placed in a solid phase synthesis flask, and DCM was added, followed by bromoacetic acid (1.43 g, 10.3 mmol) and DIEA (4 mL). The flask was placed into a motor flask shaker and was shaken overnight at room temperature. MeOH was added as capping reagent, and the reaction was left for 10 additional minutes; the solution was filtered, and the resin was washed alternatively using DCM and iPrOH (x3) and finally MeOH (x2) and then dried. Diethylenetriamine (29.4 mmol, 10 equiv) was dissolved in DCM, added to the flask containing the bromoacetyl resin, and agitated for 4 h. The solvent was filtered, and the resin was washed by alternating between DCM and iPrOH (×3), distilled water (×2), and again DCM and iPrOH (×2). The Kaiser test (90 °C) was positive. The resin was then washed with DMF (×2). tertButyl bromoacetate (5.2 mL, 35.3 mmol) and TEA (6.9 mL, to pH 7-8 on wet pH paper) were dissolved in DMF and added to the flask shaker containing the resin. The reaction was left overnight at room temperature under shaking. The solvent was filtered, and the resin was washed with iPrOH and distilled water. Additional washings were performed alternating between DCM and iPrOH (×3), then MeOH (×2), and finally ether (×1) and dried under vacuo. The Kaiser test was negative. The resin was washed with 20 %TEA in DCM (×2) and alternatively with DCM and iPrOH (×3) and then MeOH (×1) and removed to a round-bottomed flask equipped with a magnetic stirrer. A solution composed of DCM and trifluoroethanol in ratio of 2:1 was added and stirred for 2 h at room temperature. The solution was filtered and the resin washed with DCM. The organic fractions were collected and evaporated to give the expected product 0.81 g (89%). 1H NMR (300 MHz, CDCl3 δ in ppm): 1.451.48 (36H, s, t-Bu), 2.95-3.05 (8H, m, NCH2CH2NCH2CH2N), 3.45 (4H, s, NCH2COO-t-Bu), 3.53 (2H, s, NCH2COO-t-Bu), 3.54 (2H, s, NCH2COO-t-Bu), 3.56 (2H, s, NCH2COOH). HPLC analysis tR ) 17.14 min (analytical method A). HRMS: calcd for C30H55N3O10, 618.3965; found [M + H+], 618.3982. Synthesis of [(Boc)NH(CH2)2]2N(CH2)2-N(Boc)CH2CO-Lys(Z)-ditetradecylamide (2). Ditetradecylamine (1 g, 2.4 mmol) was dissolved in DCM. TEA (1 mL, 7.3 mmol) was added until the solution reached pH 8. BOC-Lys (Z)-OH (0.93 g, 2.4 mmol) was dissolved in DCM and added to the flask. Then PyBOP coupling reagent (1.3 g, 2.4 mmol) was added. The reaction mixture was stirred at room temperature for 3 h. The reaction was

Bioconjugate Chem., Vol. 14, No. 1, 2003 113

followed to completion by TLC (EtOAc:Hex 1:1, ninhydrin). The DCM was then removed under reduced pressure. EtOAc was added and the organic layer was washed with aq. KHSO4 (×3), aq. NaHCO3 (×3), brine (×3), dried over MgSO4, filtered, and evaporated to give 1.76 g (93.3%). TFA (6 mL) was added to the flask-containing product (2). The solution was stirred for 1.5 h at room temperature. The TFA was removed under reduced pressure using ether in order to removed excess of TFA to give 1.48 g (96.9%). This product was used without further purification. H-Lys(Z)-ditetradecylamide (1 g, 1.5 mmol) was dissolved in DCM. The BOC-protected mono functionalized polyamine obtained as previously reported (12) (0.76 g, 1.5 mmol) was then added. TEA was added until the mixture showed pH 8 and then the PyBOP (0.78 g, 1.5 mmol). The reaction mixture was stirred at room temperature for 4 h. The coupling was followed using TLC (EtOAc:chloroform 3:1, ninhydrin) to completion. The DCM was removed under reduced pressure, and EtOAc was added. The organic layer was washed with aq. KHSO4 (×3), aq. NaHCO3 (×3), brine (×2), and dried over MgSO4, filtered, and evaporated, and the product was chromatographed on silica gel (EtOAc:chloroform 3:1) to give 0.63 g (36.4%). This intermediate was used without further purification. Synthesis of [NH3+(CH2)2]2N(CH2)2NH2+ CH2COLys(DTPA)-ditetradecylamide (MCO-I-68). Product 2 (0.6 g, 0.5 mmol) was dissolved in EtOH. Pd/C 10% (0.6 g) was added, and H2 gas was bubbled continuously into the reaction solution. The reaction was followed by TLC (EtOAc:chloroform 3:1, ninhydrin, and fluorescamine) to completion after 1 h. MeOH was added and the reaction solution was filtered and evaporated to give 0.43 g (82%) of the desired product. The crude [BocNH(CH2)2]2N(CH2)2 N(Boc)CH2COLys-ditetradecylamide (0.4 g, 0.4 mmol) was dissolved in DCM, and TEA was added until the mixture showed pH 8. A solution of product 1 (0.4 g, 0.64 mmol) in DCM was added; then BOP (0.25 g, 0.56 mmol) was used as coupling reagent. The reaction was followed by TLC (EtOAc:chloroform 3:1, ninhydrin, and fluram) to completion (overnight). The DCM was evaporated, and EtOAc was added. The organic layer was washed with aq. NaHCO3 (×3), brine (×2), dried over MgSO4, filtered, and evaporated to give 0.67 g (99.2% of crude product). TFA was added, and the solution was stirred at room temperature for 1.5 h. The TFA was evaporated, and the crude was purified by preparative HPLC according to method D. The appropriate fractions were pooled and lyophilized to afford pure product, 0.22 g (48.5%). 1H NMR (600 MHz, CD3OD δ in ppm): 0.9 (6H, CH3), 1.31.35 (44H lipid CH2’s), 1.36 (2H, COCHCHCH2), 1.4-1.6 (2H, NCH2CH2(CH2)11CH3), 1.54 (2H, NCH2CH2(CH2)11CH3),1.6-1.8 (4H, COCHCH2CH2CH2CH2NH), 2.8 (4H, N(CH2CH2NH2)2), 2.8-2.9 (4H, HNCH2CH2N), 3.14 (2H, CONHCH2(CH2)3CH), 3.16-3.48 (2H, CONCH2 lipid), 3.2-3.3 (4H, N(CH2CH2NH2)2), 3.21 (2H, CONCH2-lipid), 3.3 (8H, HO2CCH2N-(CH2CH2N)2), 3.37 (1H, CHCON), 3.64 (2H, (CH2CH2)2NCH2CO2H), 3.72 (2H, Lys-NHCOCH2NCH2CO2H), 3.74 (4H, N(CH2CO2H)2), 3.84-3.9 (2H, Lys-NHCOCH2N), 3.92-4.09 (2H, HNCH2CO NH). 13 C NMR (600 MHz, CD3OD δ in ppm): 14.49 (Me), 23.78 (CH2Me), 27.92-28.0/28.61 (N(CH2CH2CH2(CH2)10Me), 30-31 (central lipid CH2’s), 33.12 (CH2CH2Me), 38.14 (CONHCH2(CH2)3CH), 46.4 (N(CH2 CH2NH2)2), 47.75 (NCH2(CH2)12Me), 49.3 (COCH2NH), 49.3 (NCH2(CH2)12Me), 51.23 (CHCO), 51.55/52.14 (HNCH2CH2N), 52.31 (N(CH2CH2NH2)2), 52.85/53.02/53.21 ((CH2 CH2)2NCH2CO2H), 55.82 (NCH2CONH), 57.25 (N(CH2CO2H)2),

114 Bioconjugate Chem., Vol. 14, No. 1, 2003

57.85 (HO2CCH2 NCH2CONH), 57.88 ((CH2CH2)2NCH2 CO2H), 166.6-173.4 (CO’s). HPLC analysis tR ) 17.14 min (analytical method B) and tR ) 25.38 min (analytical method C). MS analysis: calcd for C56H110N10O11,1098; found [MH+], 1099. Preparation of the Chelate Complex (MCO-I-68Gd). MCO-I-68 (33 mg, 30 µmol) was dissolved in H2O, the pH was adjusted to 5-6.5, and GdCl3‚6H2O (52 mg, 150 µmol) was added. The mixture was stirred overnight and then dialyzed over H2O. After filtration on sephadex G25, the compound was lyophilized. Electrospray mass spectrometry confirmed the mass of the expected product: calcd for C56H107N10O11 (MH+), 1254.78; found, 1254.8. Preparation of MCO-I-68-Gd/DOPE Liposomes. MCO-I-68-Gd (3.7 mg, 3 µmol) was dissolved in sterile water (150 µL) by heating at 30 °C. Dioleoylphosphatidylethanolamine (DOPE, 2.2 mg, 3 µmol) was added to the solution and the mixture was sonicated for an hour (Laboratory Supplies Co., Inc., model G112sp1t). When all the co-lipid had been incorporated, 150 µL of H2O was added to get a 10 mM solution. Integrity of the lipid MCO-I-68-Gd after sonication was checked by mass spectrometry. Peaks compatible with calculated DOPE (744) and MCO-I-68-Gd (1254) were found. Mean size of the particles was measured at 160 nm by dynamic light scattering. Preparation of MCO-I-68-Gd as Micelles. MCO-I68-Gd (3.7 mg, 3 µmol) was dissolved in sterile water (300 µL) by heating at 30 °C. Mean size of the particles was measured at 110 nm by dynamic light scattering. Preparation of Liposomes of RPR120535/ DOPE/ MCO-I-68-Gd. The cationic lipid RPR120535 (5.3 mg, 5 µmol) was dissolved in a DOPE chloroform solution (373 µL, 10 g/L, 5 µmol). A solution of MCO-I-68-Gd was added to get 1% and 5% of the gadolinium lipid incorporated as compared to the cationic lipid (0.05 and 0.25 µmol, respectively). The solvent was removed via rotary vacuum evaporation resulting in the formation of a thin film, which was hydrated at 4 °C in H2O. Slow vortex and sonication of the film lead to particles sizing 100-300 nm. Preparation of the Lipoplexes. For size measurements, complexation, and in vitro experiments, lipoplexes were formed by mixing volume to volume 8 µg DNA (200 µg/mL plasmid pCMV-luc+, pCOR plasmid) with different charge ratio of MCO-I-68-Gd, formulated as micelles or liposomes, in 5% glucose/20 mM NaCl. For in vivo experiments, 25 µL (7.5 mM) liposome or micelle of MCO-I-68-Gd were mixed with 25 µL DNA (400 µg/l in 5% glucose, 20 mM NaCl). Fluorescence Studies. Complexes were prepared as described above. Ethidium bromide (3 µL, 1 mM) was added to the lipoplexes, and fluorescence was read at 590 nm upon a 260 nm excitation wavelength with 5 nm slit width. In Vitro Transfections. 5 × 104 NIH 3T3 cells were seeded in 24-well plates in Dubelcco Modified Eagle Medium (DMEM) at 37 °C, in 5% CO2. After 24 h, cells were washed twice with 0.5 mL of serum free medium and then supplemented with 0.5 mL of medium (( serum). Plasmid DNA (1µg/well) containing the luciferase gene under the dependency of the cytomegalovirus (CMV) immediate early promoter purified according to standard procedure was mixed with 0.5, 1.5, and 6 nmol of cationic lipid per µg DNA in 75 mM NaCl. For each condition a triplicate determination was performed. Extemporaneous prepared complexes were added dropwise to the cells in the medium ( serum. After 2 h of transfection, fetal calf

Leclercq et al.

serum was added to a 10% final concentration in the wells free of serum. The cells were then incubated at 37 °C for an additional period of 24 h, then the medium removed, the cells washed with PBS and lysed with 100 µL/well of passive lysis buffer (Promega, Madison, WI) for 30 min at RT. Protein concentration of cell extracts was determined using the Bradford protein assay kit (Bio-Rad). Luminescence was measured using a luminometer (Multilabel counter 1420 Victor, EG&G Wallac) equipped with a co-injector that delivered 80 µL of luciferase substrate into 20 µL of cells extracts. The indicated luciferase activity corresponds to the ratio between the detected light unit and the protein amount. Animals. Nude mice were bilaterally implanted in the subcutaneous tissue of the flanks with fragments of human lung tumor NCIH-1299. Human tumors have been used because they develop slowly and with less necrosis than murine tumors. As the cationic lipid has no targeting group, injection in the blood flow leads to a wide dispersion in the body, so that the low gadolinium concentration is not sufficient to expect an increasing of the MRI signal. As our purpose was the validation of the DNA/lipid cationic/gadolinium complex as a lipoplex imaging agent, we have tested the distribution of the lipoplex after direct injection into the tumoral mass. After 4-6 weeks from implant, tumors were large enough (about 1 cm in diameter) to be injected with lipoplex. The mice were anaesthetized with aqueous chloral hydrate (8 mg in 100 µL of H2O) administrated intraperitoneally. Then, 50 µL of the lipoplex (formulated as MCO-I-68Gd/DNA or MCO-I-68-Gd/DOPE/DNA) was injected locally in one of the two tumors, the other contralateral tumor being a control. Experiments were performed with 25 mice (12 injected with micelles and 13 with liposomes). In each mouse, imaging was performed after 4 h, 24 h, 48 h, 72 h, 8 days, and 10 days from the lipoplex injection. Before each imaging sequence, the animals were anesthetized with an intraperitoneal injection of pentobarbital (2.4 mg in 400 µL of water). For transfection measurements, the animals were euthanasied 1, 2, 3, or 10 days after injection, and the tumors were removed and homogenized in 1 mL of PLB (Promega, Madison, WI). After centrifugation at 3000 g for 20 min at 4 °C, luciferase was assessed on 10 µL supernatant, using a luminometer (Multilabel counter 1420 Victor, EG&G Wallac). MRI Experiments. Magnetic Resonance images were acquired using a Bruker AMX300 spectrometer equipped with a mini-imaging accessory (RF probe 38 mm, gradient strength 0.5 G/cm/A). A spin-echo sequence was used to obtain T1 weighted MR images (TE/TR ) 10/ 500 ms; NEX ) 6). Typically, a FOV of 3 × 3 cm with a slide thickness of 1 mm was used with a 256 × 256 matrix data. Sixteen consecutive transverse slices, and then eight sagittal slices through the tumor were acquired. RESULTS AND DISCUSSION

Synthesis of Gadolinium Cationic Lipids. According to previously reported results (9) using DTPA dianhydride as precursor, our attempts to synthesize MCOI-68 failed to give good yields of the desired product, since the second anhydride function reacted with another equivalent of the amino-free lipopolyamine to result in a dimeric molecule containing two lipid units linked through a DTPA diamide spacer (Figure 1). To develop our integrated imaging approach, it was necessary to use large quantities of the Gd-cationic lipid. Therefore, we decided to use a protected DTPA harboring

Gadolinium Cationic Lipids

Bioconjugate Chem., Vol. 14, No. 1, 2003 115

Figure 1. Attempts to use DTPA anhydride to obtain MCO-I-68.

a single free carboxylic acid precursor. To our surprise, there are only two methods to obtain the desired DTPA harboring a single free carboxylic acid. In one of them, the building block is obtained by biodegradation of a single methyl ester using an esterase to form a DTPA protected with methyl esters in small quantities (10). This approach is complicated and necessitates the final deprotection by basic ester hydrolysis. The second approach is a seven-step process for obtaining DTPA tert-butyl esters with a free carboxylic acid at one extremity (11). This tedious multistep process prompted us to propose the simplified and versatile method presented here. The approach is based on our previous works for the synthesis of protected monofunctionalized polyamines using solid support techniques (12-13). In those works, we have taken advantage of the dilution effect induced by the coupling of an alkylating agent to a solid support, which prevents polyalkylation when treating a polyamine with the solid supported alkylating agent. Thus, we have attached bromoacetic acid to chlorotrityl chloride resin and reacted the solid supported bromide with an excess of free diethylenetriamine. In a second step, the supported polyamine was reacted with excess tert-butyl-bromoacetate under basic conditions to afford the protected DTPA. Finally, the product was cleaved from the solid support using mild acidic conditions (trifluoroethanol) to prevent tert-butyl ester deprotection, thus resulting in the desired DTPA building block 1 with a single free carboxylic acid (see Figure 2). Unlike our previous works using spermine as polyamine, we did not observe any of the symmetric free carboxymethylene DTPA product which could have been be obtained, in theory, by the reaction between the central secondary amine of the diethylenetriamine and the solid supported bromoacetyl group. This is probably mainly due to the statistical ratio of 1/2 secondary/primary amines in diethylenetriamine, contrary to the 1/1 ratio in the case of spermine. This, together with small steric and inductive effects favoring the reaction of the primary amine, results in the absence of the symmetric product. The advantage of this new method is that no isolation of intermediates is necessary. Thus, product 1 is obtained rapidly without needing any purification step. (Figure 2).

Figure 2. Synthesis of asymmetrically protected DTPA building block 1. (a) Diethylisopropylamine CH2Cl2 overnight, (b) diethylenetriamine CH2Cl2, 4 h, and (c) tert-butyl-bromoacetate triethylamine overnight then trifluoethanol/CH2Cl2, 1 h.

The lipopolyamine precursor was obtained by coupling of previously reported protected building block trisaminoethylcarboxy-methylene to a preformed H-Lys(Z)ditetradecylamide (13). After deprotection of the Lys side chain of product 2 by hydrogenolysis, product 1 was coupled with the free amino group of the obtained product 3 using regular peptide coupling techniques to give 4. The final product was obtained after exhaustive deprotection of 4 using trifluoroacetic acid (see Figure 3). The cationic lipid MCO-I-68 was fully characterized using two different HPLC gradients, multidimensional NMR, and high-resolution mass spectroscopy. The gadolinium ion was complexed to the DTPA moiety of the lipid as previously described (9). The excess of GdCl3.6H2O was removed via subsequent dialysis and gel filtration to remove all traces of toxic GdCl3.6H2O. The full conversion of MCO-I-68 to MCOI-68-Gd was checked by electrospray mass spectrometry. (See Figure 3). Physicochemical Characterization of DNA/MCOI-68-Gd Complexes. Dilution of MCO-I-68-Gd in water gave a micellar suspension. The cationic polyamine moiety of this lipid allowed to form MCO-I-68-Gd/DNA lipoplexes by direct mixing of the micelles with DNA. Liposomes were prepared by direct mixing of the gadolinium lipid with a co-lipid. The lipid chains of the widely used dioleoylphosphatidylethanolamine (DOPE), and the gadolinium lipid or-

116 Bioconjugate Chem., Vol. 14, No. 1, 2003

Leclercq et al.

Figure 3. Synthesis of MCO-I-68 and its complex with gadoliniun.

ganized together under liposome type of structures when both lipids were sonicated together, as described in materials and methods and in (9). On the other hand, the usual way of preparing liposomes via formation of a film under vacuum elimination of CHCl3 did not allow us to form homogeneous liposomal structures. Lipoplexes (MCO-I-68-Gd/DOPE/ DNA were then formed by diluting different amounts of the preformed liposomes with a constant amount of DNA. The ability of MCO-I-68-Gd as micelles or liposomes to complex DNA and to form small particles was studied. The size of lipoplexes obtained either by association of micelles to DNA or by association of liposomes to DNA was evaluated by dynamic light scattering experiments (Figure 4). Results are represented as a function of the ratio of positive charges (MCO-I-68-Gd) to negative charges (DNA Phosphate) equivalents, considering that the MCO-I-68-Gd compound carried three positive charges. Complexes, formulated either from micelles or liposomes, exhibit comparable size for the different studied charge ratio (MCO-I-68-Gd/DNA) studied. Moreover, the characteristics of lipoplexes are similar to those previously observed with lipopolyamine RPR120535 (14). We found that, when lipoplex self-assembly is formed with negative charges in excess (charge ratio MCO-I-68-Gd/ DNA 1 µ,) occurs at lipid/DNA ratio between 1 and 2, corresponding to charge neutrality of the complexes. As the amount of cationic lipid was increasing relative to the DNA, colloidally stable 100 nm size particles are obtained. We also carried out fluorescence experiments in order to estimate DNA entrapment into the lipoplexes. Intercalation of ethidium bromide between the free DNA base pairs gives a fluorescent signal taken as 100%. When DNA is complexed with the cationic lipids, the ethidium bromide signal decreases due to its exclusion from DNA. This allows monitoring the percentage of compacted DNA. Figure 4 shows the diminution of fluorescence depending on the degree of DNA complexation. Fluorescence level decreases to 50% or 60% for

Figure 4. Size of the lipoplexes and ethidium bromide residual fluorescence in function of the charge ratio lipid/DNA ((). Open circles represent the size of the lipoplexes, measured by dynamic light scattering. Full squares represent complexion of DNA, measured by emission fluorescence (100% ) complete intercalation in absence of cationic lipid).

micelle or liposome-based lipoplexes, respectively. This is less than what had been previously observed with

Gadolinium Cationic Lipids

Bioconjugate Chem., Vol. 14, No. 1, 2003 117

Figure 6. Lipid-mediated DNA transfection of NIH 3T3 cells using cationic liposomes RPR120535/DOPE, RPR120535/DOPE + 1% MCO-I-68-Gd, RPR120535/DOPE, /DOPE+5% MCO-I-68Gd in three different charge ratio (RPR120535/DNA) 0.5, 1.5, and 6, in the presence (gray) or absence (white) of calf serum. Data are the mean of triplicate determination (error bars: SD).

Figure 5. Lipoplexes were loaded on a 0.8% agarose gel at the different DNA complexation zone identified in Figure 2. Free DNA was loaded on the first well as a control. Then mixtures of MCO-I-68-Gd /DNA at the lipid/DNA charge ratio 0.4, 1.1, 2.5, 4, and 6 were loaded either when starting from liposomes (A) or micelles (B).

RPR120535 (13), thus suggesting incomplete DNA compaction or association to lipoplexes. However, MCO-I-68Gd /DNA lipoplexes did not migrate on agarose gel electrophoresis (Figure 5). Thus, it appeared in both liposome and micelle-based lipoplexes that a charge ratio of 2.5 was enough to get a full DNA complexion. Nevertheless, the ethidium bromide test indicates a lower affinity of the MCO-I-68-Gd lipid for DNA as compare to the RPR120535 lipopolyamine control. In Vitro Transfection. The gadolinium lipid had been designed as a tracer co-lipid for MRI studies in order to follow the biodistribution of our new homemade DNA lipidic vectors. As this lipid presents the same type of structure as other cationic lipids, it should not alter in vivo lipoplex biodistribution. Moreover, the physico-chemistry shown above indicates that this lipid behaves very closely to previously studied cationic lipids). In vitro transfection activity of MCO-I68-Gd lipid was studied on NIH3T3 cells and compared to reference lipid RPR120535. MCO-I-68-Gd was used either alone or by co-formulation with cationic RPR120535, which represents the way this method will be used in future MRI studies. The gadolinium lipid, to be used as a co-lipid, should not modify the activity, nor reduce the efficiency of the lipid tested. The cationic lipid RPR120535 formulated with DOPE (1/1) was chosen as a reference, since it was shown that this lipid did transfect efficiently NIH 3T3 cells in absence of serum (15). We incorporated 1-5% gadolinium lipid into the cationic formulation (materials and methods) and evaluated the transfection activity of these new gadolinium-based formulations (see Figure 6). With the MCO-I-68-Gd/DNA complex was used, transfection efficiency was reduced as compared to the

RPR120535/DNA complex (not shown). This may be explained by the reduced affinity of the MCO-I-68-Gd for DNA as compared to RPR120535, as measured by ethidium bromide. Complexes formed out of MCO-I-68Gd are likely less stable and release DNA faster than the RPR120535. On the other hand, addition of 1 and 5% of the MCO-I-68-Gd into the RPR120535/DOPE formulation did not significantly affect the DNA transfection efficiency, as compared to DNA complexed with RPR120535/DOPE, indicating that MCO-I-68-Gd does not interfere with in vitro transfection efficiency of RPR120535.(Figure 6). MRI Experiments. The efficiency of a DNA delivery vehicle depends of the pharmacokinetics that determines distribution and concentration in targeted tissue. With the aim to explore the possibility of imaging a DNA transfection event, Nantz et al. (9) brought to the fore a contrast enhancement in lymph nodes after subcutaneous injection of a polymerizable lipid chelated with gadolinium. They formulated afterward this gadolinium lipid with DOTAP and DNA and showed that the lipoplex formed was able to transfect NIH 3T3 cells in vitro. However, visualization DNA complexes were not shown in those works so far. The most important objective of this work was to design and assess a new tool visualizable in MRI for the biodistribution monitoring of cationic lipid/DNA complexes. For this purpose we have tested the ability of MCO-I-68-Gd/ DNA complex to act as a MRI contrast agent and to follow lipoplex biodistribution in vivo. Imaging studies were performed after intratumoral injection of the complexes. Tumors were bilaterally implanted in the flanks of the animals, and only one tumor was injected, the contralateral one serving as a control. Noninjected tumors appeared as multinodular masses with a thin layer of oedematous tissue at their boundary (Figure 7). The neoplastic tissue appeared as rather homogeneous nodules emitting a signal of medium intensity. Hypointense areas were visible within the mass, probably corresponding to necrotic tumoral tissue. In lipoplex injected tumor, areas of hyperintensity were visible. In some cases (2 out of 13), these areas appeared only 24 h after injection, probably because little or no exchange water with the gadolinium could occur in the injection site (Figure 8). The contrast agent diffused slowly (up to 10 days) out of the neoplastic mass, in the healthy tissue. The injection of the contrast agent labeled lipoplexes did not cause any increase of the necrosis, but both with liposomes and micelles small hemorrhages were associated to the injection site. After 24 h, the

118 Bioconjugate Chem., Vol. 14, No. 1, 2003

Figure 7. (A) 48h after injection of the left side tumor with MCO-I-68-Gd/DOPE/ DNA complexes. Right side tumor is thenon injected tumor control. (B) 48 h after injection of the right side tumor with MCO-I-68-Gd/DNA lipoplex. Left side tumor is the noninjected tumor control.

contrast agent diffused in the surrounding neoplastic tissue, creating areas of medium intensity signal with irregular margins. However, injection site was still visible by its hyperintensity. In the cases in which a necrotic core was involved in the injection site, the diffusion of the hyperintensity signal was poor and mainly limited around the injection site. The different behavior of the complex from a mouse to another was correlated to the intern structure of the tumor. No difference could be obviously detected between micelles and liposomes. Diffusion of the hyperintensity signal reached a maximum after 52 h from injection and decreased slowly thereafter. Nevertheless, hyperintensity was still visible after 10 days (Figure 8). The presence of hyperintensity demonstrates that gadolinium complex interacts with water molecules and is not sequestered in small intracellular compartments (i.e., secondary lysosomes). The clearance of the gadolinium from the tumor is probably hampered by the hypovascularity of the tumoral core. The displacement of the contrast agent in the tumor could be due to both diffusive and convective movements of the interstitial fluid. In these MR imaging experiments, the injected plasmid coded for the reporter gene luciferase, thus allowing

Leclercq et al.

assaying transfection efficiency. After 1, 2, 3, and 10 days, mice were euthanasied, tumors were removed, and the luciferase activity was quantified. No significant luciferase activity was detected in the tumors injected with micelle-based complexes. On the other hand, in most of the tumors injected with liposome-based lipoplexes, the luciferase activity was noteworthy although with low levels during the first 3 days. Nevertheless, as shown from the corresponding imaging, the transfection level seems to depend essentially on the possible diffusion of the complex in nonnecrotic zones. Because of the variability between the tumors observed, we could not evaluate the decrease of transfection from day 1 to day 3. No luciferase activity could be detected at day 10. This was expected, as it had been shown that cationic lipid mediated gene transfer results in a temporary transgene expression with a maximal expression after 24 h which decreases rapidly afterward (data not shown). It should be noted that all the MRI experiments were performed using MCO-I-68-Gd to form the DNA complexes, without adding cationic lipid RPR120535. On the other hand the transfection level was lower when using MCO-I-68-Gd/DNA complexes than when diluting 1-5% MCO-I-68-Gd in a RPR120535/DOPE/DNA. Thus, we are currently working on the MRI of the co-formulated cationic lipid, which has been shown to be more efficient in vitro. In this way we will aim at adjusting the in vivo formulation so that gadolinium contrast can be observed with a lower concentration of MCO-I-68-Gd in a coformulation, hence, increasing in vivo transfection and thus correlating the imaging observation to transfection. According to the in vivo results, the transfection occurs only during the first 3 days after injection and is paralleled by hyperintensity at MRI examination. In the following days, there is a mismatch between hyperintensity and transfection. This mismatch seems to suggest that lipoplexes although remain in the tumor, their ability to transfect is lost. We speculate that the loss of transfection ability could arise from DNA damage; however, we cannot demonstrate this assumption to date.

Figure 8. Visualization of lipoplex-Gd in tumor NCIH-1299 at 4 h, 24 h, 72 h, 8 days, and 10 days after injection. (A) Lipoplexes formed from liposomes injected on left side tumor. (B) Lipoplexes formed from micelles injected on right side tumor.

Gadolinium Cationic Lipids CONCLUSIONS

We have designed and applied a new tool based on in vivo magnetic resonance imaging of cationic lipid/DNA complexes. A new versatile solid phase synthesis of asymmetrically protected DTPA building block for covalent conjugation to cationic lipids has been demonstrated. The DTPA-cationic lipid obtained was chelated with gadolinium and complexed with plasmid DNA. We have shown that these complexes display significant levels of transgene expression after in vitro transfection. Our present studies show, for the first time, that cationic lipid/ DNA complexes can be observed in vivo using magnetic resonance imaging after intratumoral administration. We could visualize cationic lipid/DNA complexes and observe their distribution in the tumor area as a function of time. We have shown that lipoplexes formulated from both micelles and liposomes can be observed during long periods of time. However, transgene expression could be detected on tissue extract only when complexes were formulated as liposomes Overall, this technique opens the field of gene delivery to the design and the in vivo observation DNA/cationic targeted to various tissues or tumors. Works exploiting these possibilities are currently ongoing. ACKNOWLEDGMENT

This project was funded by Arc-en-Ciel and AFIRST Israel/France (programs of the French Ministry of Foreign Affairs and the Israeli Ministry of Sciences). Dr. Gerardo Byk is also indebted to the Marcus Center of Medicinal Chemistry and to TEVA Pharmaceuticals Inc. The imaging system was purchased with a grant of the Association pour la Recherche sur le Cancer (ARC). LITERATURE CITED (1) Felgner, J. H., Kumar, R., Sridhar, C. N., Wheeler, C. J., Tsai, Y. J., Border, R., Ramsey, P., Martin, M., and Felgner, P. L. (1994) Enhanced gene delivery and mechanism studies with a novel series of cationic lipid formulations. J. Biol. Chem. 269, 2550-2661. (2) Schatzlein, A. G (2001) Nonviral vectors in cancer gene therapy: principles and progress. Anti-Cancer Drugs 12, 275304. (3) Miller, A. D (1998) Cationic Liposomes for Gene Therapy. Angew. Chem. 37, 1768-1785. (4) Pasqualini, R., and Ruoslahti, E. (1996) Tissue targeting with phage peptide libraries. Mol. Psychiatr. 6, 421-423. (5) Bell, J. D, and Taylor-Robinson, S. D. (2000) Assessing gene expression in vivo: magnetic resonance imaging and spectroscopy. Gene Ther. 7, 1259-1264.

Bioconjugate Chem., Vol. 14, No. 1, 2003 119 (6) Goffeney, N., Bulte, J. W. M., Duyn, J., Bryant Jr, L. H., and van Zijl, P. C. M. (2001) Sensitive NMR detection of cationic-polymer-based gene delivery systems using saturation transfer via proton exchange. J. Am. Chem. Soc. 123, 8628-8629. (7) Louie, A. Y., Hu¨ber, M. M., Ahrens, A. T., Rothba¨cher, U., Moats, R., Jacobs, R. E., Fraser, S. E., and Meade, T. J. (2000). In vivo visualization of gene expression using magnetic resonance imaging. Nat. Biotechnol. 18, 321-325. (8) Bhorade, R., Weissleder, R., Nakakoshi, T., Moore, A., and Tung, C.-H. (2000) Macrocyclic chelators with paramagnetic cations are internalized into mammalian cells via a HIV-tat derived membrane translocation peptide. Bioconjugate Chem. 11, 301-305. (9) Wisner, E. R., Aho-Sharon, K. L., Bennett, M. J., Penn, S. G., Lebrilla, C. B., and Nantz, M. H. (1997) A modular lymphographic magnetic resonance imaging contrast agent: contrast enhancement with DNA transfection potential. J. Med. Chem. 40, 3992-3996. (10) Burks, E., Koshti, N. Jacobs, H., and Gopalan, A. (1998) Selective Monohydrolysis of Esters of Polyaminocarboxylic acids using Pig Liver Esterase. Synlett 11, 1285-1287. (11) Arano, Y., Uezono, T., Akizawa, H., Ono, M, Wakisaka, K, Nakayama, M., Sakahara, H., Konishi, J., and Yokoyama, A. (1996) Reassessment of diethylenetriaminepentaacetic acid (DTPA) as a chelating agent for indium-111 labeling of polypeptides using a newly synthesized monoreactive DTPA derivative. J. Med. Chem. 39, 3451-3460. (12) Byk, G., Frederic, M., and Scherman, D. (1997) One Pot Synthesis of Unsymmetrically Functionalized polyamines by a solid-phase strategy starting from their symmetrical polyamine counterparts. Tetrahedron Lett. 38, 3219-3222. (13) Byk, G., Dubertret, C., Escriou, V., Frederic, M., Jaslin, G., Rangara, R., Pitard, B., Crouzet, J., Wils, P., Scwartz, B., Scherman, and D. (1998) Synthesis, activity, and structureactivity Relationship studies of novel cationic lipids for DNA Transfer. J. Med. Chem. 41, 224-235. (14) Pitard, B., Oudrihiri, N., Vigneron, J. P., Hauchecorne, M., Aguerre, O., Toury, R., Airiau, M., Ramasawny, R., Scherman, D., Crouzet, J., Lehn, J. M., and Lehn, P. (1999) Structural characteristics of supramolecular assemblies formed by guanidinium-cholesterol reagents for gene transfection. Proc. Natl. Acad. Sci. 96, 2621-2626. (15) Escriou, V., Ciolina, C., Lacroix, F., Byk, G., Scherman, D., and Wils, P. (1998) Cationic lipid-mediated gene transfer: effect of serum on cellular uptake and intracellular fate of lipopolyamine/DNA complexes. Biochim. Biophys. Acta 1368, 276-288.

BC025567E