Modified Linear Polyethylenimine−Cholesterol Conjugates for DNA

D. W. (1984) The human LDL receptor: a cysteine-rich protein with multiple Alu ..... A. Eltoukhy , Christopher G. Levins , Yingxia Wang , Daniel G...
0 downloads 0 Views 140KB Size
840

Bioconjugate Chem. 2003, 14, 840−847

Modified Linear Polyethylenimine-Cholesterol Conjugates for DNA Complexation Darin Y. Furgeson, Winter S. Chan, James W. Yockman, and Sung Wan Kim* Department of Pharmaceutics and Pharmaceutical Chemistry, Center for Controlled Chemical Delivery, University of Utah, Salt Lake City, Utah 84112-5820. Received April 14, 2003; Revised Manuscript Received May 27, 2003

Linear polyethylenimine (LPEI) is an effective nonviral gene carrier with transfection levels equal or above branched polyethylenimine (BPEI) and exhibits a lower cytotoxicity profile than BPEI. High molecular weight LPEI Mw 25 k was modified with cholesterol in three different geometries: linear shaped (L), T-shaped (T), and a combined linear/T-shaped (LT) forming the LPEI-cholesterol (LPC) conjugates LPC-L, LPC-T, and LPC-LT, respectively. Physical characterization of LPC/pDNA complexes included particle size, zeta potential, DNase protection, mIL-12 p70 expression, and cytotoxicity. The particle size was further confirmed by atomic force microscopy (AFM). The LPC-T/pDNA complexes were optimal at N/P 10/1 that resulted in a particle size of ∼250 nm, which was confirmed by AFM, and a surface charge of +10 mV. These complexes also effectively protected the pDNA for up to 180 min in the presence of DNase I. B16-F0 cells transfected with LPC-L and LPC-T showed protein expression levels higher than LPEI alone and twice that of BPEI but without any significant loss in cell viability. These results were confirmed with EGFP flow cytometry and transfection of Renca cells. The differences in rates of transfection of the LPC/pDNA complexes is due in part to conformational changes from the point of complex formation to interaction with the plasma membrane. These conformation changes provide protection for unprotonated secondary amines in the LPEI backbone by hydrophobic protection of the cholesterol moiety that we termed “unprotonated reserves”. Finally, we show that LPC conjugates exploit receptor-mediated endocytosis via the LDL-R pathway with transgene expression levels decreasing nearly 20% after saturating the LDL-R sites on MCF-7 cells with hLDL-R-Ab.

INTRODUCTION

Polyethylenimine (PEI) has long been used as the standard for nonviral gene delivery; however, the ideal molecular weight and geometry for PEI have not been determined (1). The molecular weights of commercial PEIs vary from 423 Da-800 kDa with linear and branched geometries. The molecular weight of BPEI has been shown to directly affect transfection efficiency while the pH of the BPEI before complexation does not (2). Recently, transfection efficiencies increased and cytotoxicity decreased with the synthesis of low molecular weight PEI (Mw 11 900) with minimized branching (3), a step toward LPEI. BPEI 25 k is limited in its use primarily due to its high cytotoxicity, presumably due to the high cationic state of the numerous primary amines (4, 5); however, the best results for in vivo BPEI transfection were shown to be with Mw 25 k (6). LPEI has been shown to be an effective nonviral gene carrier (7-15) with increased gene expression and decreased toxicity compared to BPEI of comparable molecular weight. The putative method of PEI gene expression is through endosomal release by osmotic swelling by the proton sponge effect; however, this endosomal release may cause local cytotoxicity due to the release of endo/lysosomal enzymes (16). LPEI Mw 423 cannot efficiently buffer the low pH found in the secondary lysosome after endocyto* To whom correspondence should be addressed: University of Utah, Center for Controlled Chemical Delivery, 30 S 2000 E Rm. 201, Salt Lake City, UT 84112-5820. Tel: (801) 581-6654. Fax: (801) 581-7848. E-mail: [email protected].

sis; consequently, we have shown that a colipid is needed to facilitate gene expression (17). However, LPEI 25 k has over 520 secondary amines capable of protonation and subsequent pH buffering; therefore, this polymer does not require the use of a colipid for endosomal escape. LPEI Mw 25 k should have lower cytotoxicity than BPEI Mw 25 k as a higher charge density was equated with higher cytotoxicity (5). Modifications to LPEI have been rather sparse while BPEI has been modified extensively including PEGylation (4), targeting moieties (18, 19), and lipids (20). Cholesterol conjugation to a primary amine of BPEI Mw 1800 resulted in water-soluble lipopolymer (WSLP) that has shown to dramatically increase transfection efficiencies and promote tumor regression when combined with a therapeutic plasmid (20-22) beyond that of the polymer itself. Conjugation of cholesterol to a secondary amine of BPEI Mw 1800 has also been completed with favorable results (23). The only known LPEI carrier used in the literature is ExGen 500, an LPEI 22 k conjugate with 510 monomer units (7), that has also been modified with mannose, galactose, and RGD ligands by PolyPlustransfection (Illkirch, France). In addition, the effect of the geometry of cholesterol conjugation with cationic lipids (17) and LMW cationic polymers (20, 23) has previously been explored; however, studies with HMW cationic polymers have not been initiated. LPEI Mw 25 k can adequately provide pH buffering, similar to BPEI Mw 25 k, but without the deleterious effects of the high cationic charge density found with BPEI. LPEI Mw 25 k should

10.1021/bc0340565 CCC: $25.00 © 2003 American Chemical Society Published on Web 06/21/2003

Polyethylenimine−Cholesterol Conjugates

have a lower charge density compared to BPEI Mw 25 k (5) resulting in decreased cytotoxicity. Upon the basis of the success of cholesterol conjugation with WSLP and knowing that LPEI Mw 25 k should provide pH buffering similar to BPEI Mw 25 k without the degree of cytotoxic side effects, we synthesized three LPEI-cholesterol (LPC) conjugates: LPC-L (linear), LPC-T (T-shaped), and LPC-LT (combined linear/T-shaped). The introduction of the cholesterol moiety should provide favorable interactions with the cell membrane as seen with WSLP in addition to sequestering regions of the LPEI backbone for future lysosomal buffering in the late endosomolytic stages. EXPERIMENTAL PROCEDURES

Materials. Linear polyethylenimine (LPEI, Mw 25 000) was purchased from Polysciences, Inc. (Warrington, PA). Cholesteryl chloroformate and chlorotrimethylsilane were purchased from Aldrich Chemical Co., Inc. (Milwaukee, WI). Anhydrous methylene chloride, CDCl3, D2O, acetone, ethyl acetate, triethylamine (TEA), and trifluoroacetic acid (TFA) were purchased from Sigma Chemical Company (St. Louis, MO). RQ1 RNase-free DNase and JM109 competent cells were purchased from Promega (Madison, WI). pmIL-12e was purchased from Invivogen, Inc. (San Diego, CA). pCMS-EGFP was purchased from Clontech Laboratories, Inc. (Palo Alto, CA). SeaKem GTG agarose, molecular biology grade water, Luria Broth (LB), and ethidium bromide were purchased from ISC Bioexpress (Kaysville, UT). Rosewell Park Memorial Institute (RPMI 1640) medium and heat-inactivated fetal bovine serum were purchased from Hyclone (Logan, UT). Trypsin-EDTA, gentamicin reagent solution, and penicillin-streptomycin-glutamine were purchased from Gibco-BRL (Gaithersburg, MD). The Endofree Maxi Plasmid Purification Kit was purchased from QIAGEN (Valencia, CA). The PELCO mica disks, 9.9 mm diameter, were purchased from Ted Pella, Inc. (Redding, CA). BDOptEIA ELISA kits for murine interleukin-12 (mIL12) p70 were purchased from Pharmingen (San Diego, CA). LDL-receptor antibodies were purchased from Oncogene Research Products (Boston, MA). Spectra/Por membrane tubing MWCO 6-8000 was purchased from Spectrum Laboratories, Inc. (Rancho Dominquez, CA). Methods. Synthesis of LPC-L and LPC-LT. LPEI Mw 25 000 (1.0 g, 0.04 mmol) was added to 150 mL of anhydrous methylene chloride and heated in a hot water bath for ∼20 min to facilitate solubility. The sample was cooled, and additional methylene chloride was added to return the volume to 150 mL. For LPC-L, 0.018 g (0.04 mmol) of cholesteryl chloroformate was dissolved in 3 mL of anhydrous methylene chloride and added to the LPEI dropwise over 10 min; consequently, 0.036 g (0.08 mmol) of cholesteryl chloroformate was added for the synthesis of LPC-LT. These reactions were stirred overnight, and the solvent removed by rotary evaporation and purified by solvent precipitation in acetone, ethyl acetate, and excess methylene chloride. The conjugates were dried and lyophilized prior to analysis by 1H NMR (Varian Mercury 400, Inc., Palo Alto, CA) and stored at -20 °C. Because of the high cationic charge and multiple charge states of the LPEI, mass spectrometric analysis could not be completed. Synthesis of LPC-T. LPEI Mw 25 000 (1.0 g, 0.04 mmol) was added to 150 mL of anhydrous methylene chloride and heated in a hot water bath for ∼20 min to break the intra- and intermolecular hydrogen bonds of the LPEI. The clear solution was removed from the hot water bath

Bioconjugate Chem., Vol. 14, No. 4, 2003 841

and cooled, and additional methylene chloride was added to return the volume to 150 mL. The mixture was stirred for 15 min at room temperature after which 200 µL of anhydrous TEA was added and stirring continued for 5 min. To protect the terminal hydroxyl group, 60 µL (0.06 mmol) of 1 M Si(CH3)3Cl was added, forming an HCl(g) cloud over the solution. The mixture was stirred overnight. Cholesteryl chloroformate (36 mg, 0.08 mmol) was dissolved in 3 mL of anhydrous methylene chloride and added dropwise to the TMS-protected LPEI over 10 min; consequently, the cholesterol was conjugated to a secondary amine on the LPEI by a carbamate bond. The reaction was stirred overnight. Deprotection of the LPCT-TMS was completed by adding 3 mL of TFA at which point LPC-T fell out of solution. The sample was dried on the rotavaporator, dissolved in 20 mL of ultrapure water, and purified by dialysis (MWCO 6-8000) against deionized water for 3 days. The conjugate was lyophilized and analyzed by 1H NMR (Varian Mercury 400, Inc., Palo Alto, CA) and stored at -20 °C. As with LPC-L and LPCLT, the high cationic charge and multiple charge states of the LPEI prevented mass spectrometric analysis. Amplification and Purification of pmIL-12e. The pmIL12e vector was purchased from Invivogen (San Diego, CA) in which the p35 and p40 are linked by an elasti-site and driven by a human T-cell leukemia virus (HTLV) promoter. Plasmids were amplified using JM109 competent cells and purified using the QIAGEN Endofree Maxi Plasmid Purification Kit according to the protocol. UV spectrophotometry at 260/280 nm and gel electrophoresis determined the concentration, integrity, and purity of the amplified pmIL-12e. Purity was greater than 1.8, and appropriate bands were seen with restrictive enzyme digestion. Preparation of LPC/pmIL-12e Complexes. LPC conjugates and BPEI were concentrated at 1.25 mg/mL in ultrapure water and stored at 4 °C for future use. Turbidity was seen with the LPC-L and more so with the LPC-LT most likely due to intermolecular hydrogen bond formation. No visual turbidity was seen with LPC-T. The LPC solutions were heated in a warm water bath ∼55°C for 15 min to break the intra- and intermolecular hydrogen bonds of the conjugates. After being cooled to room temperature, both the LPC solutions and pmIL12e were separately diluted to a final concentration of 5% glucose at 100 µL, each at a pDNA concentration of 0.1 mg/mL. The LPC solution was added to the pmIL12e solution and thoroughly mixed. Complexation was allowed for 15 min at ambient conditions. Significant aggregation of LPC-LT/pmIL-12e complexes was seen after 30 min. Gel Retardation Assay. pDNA condensation by the LPC conjugates was evaluated by a gel retardation assay. The LPC/pmIL-12e complexes were electrophoresed on a 1% agarose gel pretreated with 0.5 mg/mL ethidium bromide in 1 × Tris-base-acetate-EDTA (TAE) buffer at 84 V. Naked pmIL-12e was used for the marker lane. DNase Protection Assay. LPC/pmIL-12e complexes were prepared at N/P ratios 5/1, 10/1, 20/1, and 30/1 at a final pDNA concentration of 0.1 mg/mL and 500 µL total volume. The samples were incubated at ambient conditions for 20 min. Fifty microliters of the stop solution (200 mM sodium chloride, 20 mM EDTA, and 1% SDS) was added to eight PCR tubes for each LPC/ pDNA sample, representing the appropriate DNase stop times: 0, 2, 5, 15, 30, 60, and 180 min. Fifty microliters of each LPC/pDNA complex was removed and added to the 0 min incubation tube and gently mixed. To the remaining LPC/pDNA complexes, 50 µL of RQ1 RNase-

842 Bioconjugate Chem., Vol. 14, No. 4, 2003

free DNase (Promega, WI) was added, gently mixed, and incubated at 37 °C. At the appropriate stop times, 50 µL of each sample was removed and added to the stop solution tube. To dissociate the pDNA from the LPCs, the samples were incubated overnight at 60 °C. The free pDNA was purified by phenol/chloroform extraction followed by ethanol precipitation. The pellets were redissolved in 10 µL of molecular biology grade water and analyzed by 1% agarose gel electrophoresis. Particle Size and Zeta Potential. LPC/pmIL-12e complexes were measured at several N/P ratios for particle size and zeta potential on a Brookhaven Instruments Corp. (Holtsville, NY) ZetaPALS. Experimental conditions were 37 °C using a 677 nm wavelength at a constant angle of 15°. Smoluchowski’s formula was used to calculate the zeta potential from the electrophoretic mobility. Values for the particle size are effective mean diameters. Atomic Force Microscopy. A method for determining surface morphology by AFM has been previously described (17). Briefly, 9.9 mm mica disks were soaked in 33 mM magnesium acetate for a minimum of 24 h to promote stronger pDNA binding by the divalent magnesium ions rather than the monovalent potassium ions. The mica was sonicated for 30 min in ultrapure water and subjected to glow discharge for 15 s in a vacuum between 100 and 200 mTorr. Upon exposure to air, 20 µL of 0.1 mg/mL LPC-T/pmIL-12e complexes was placed on the mica surface for 2 min after which the mica was gently rinsed with distilled water and slowly blown dry with nitrogen. Imaging was completed at room temperature using a Digital Instruments Nanoscope II SFM (Santa Barbara, CA) in tapping mode. Tumor Cell Lines. B16-F0 murine melanoma and Renca murine renal cell carcinoma cells were a gift from Dr. Wolfram Samlowski of the Huntsman Cancer Institute (Salt Lake City, UT). MCF-7 human breast carcinoma cells were donated by Dr. You Han Bae of the University of Utah (Salt Lake City, UT). The cell lines were grown and maintained in RPMI 1640 medium supplemented with 10% FBS, 100 U/mL penicillin, 100 U/mL streptomycin, and 50 µg/mL gentamycin at 37 °C and humidified 5% CO2. In Vitro Transfection. Five thousand cells (B16-F0 or Renca) were seeded on 96-well plates in RPMI 1640 containing 10% FBS and antibiotics. The plates were incubated at 37 °C and humidified 5% CO2 until cell confluency reached ∼70%. At this point, cells were transfected with LPC/pmIL-12e prepared at various N/P ratios ranging from 1/1-30/1. pDNA (0.1 µg) was loaded per 100 µL of RPMI + 10% FBS. Cells were incubated for 4 h in the presence of complexes and 10% FBS at standard incubator conditions. After 4 h, the cell media was replaced with 100 µL of fresh RPMI + 10% FBS, and the cells were further incubated for an additional 20 h at the same conditions, resulting in a total transfection time of 24 h. The cell media was removed for ELISA as mIL-12 p70 is a secreted protein. Untreated cells in addition to cells treated with naked pDNA alone were used as controls. For the EGFP studies, 300 000 Renca or B16-F0 cells were seeded on six-well plates in RPMI 1640 supplemented with 10% FBS and antibiotics. The plates were incubated at 37 °C and humidified 5% CO2 overnight. The following day the cells were transfected with 2 µg of pCMS-EGFP in the presence of serum for 4 h after which the media was replaced with fresh RPMI 1640 + 10% FBS for an additional 44 h. Cells were trypsinized, centrifuged, and suspended in PBS prior to flow cytometry analysis.

Furgeson et al.

Cell Viability. A Cell Counting Kit-8 (CCK-8) (Dojindo Molecular Technologies, Inc., Gaithersburg, MD) was used to evaluate the cytotoxicity of the LPC conjugates. Five thousand cells (B16-F0 or Renca) were seeded on a 96-well plate with RPMI +10% FBS and incubated at 37 °C and humidified 5% CO2 until confluency reached ∼70%. LPC/pmIL-12e complexes were prepared at various N/P ratios ranging from 1/1-30/1. Cells were transfected with 0.1 µg of pmIL-12e in the presence of 10% FBS for 4 h after which the media was changed, and transfections proceeded for an additional 20 h in the presence of 10% FBS. After removal of the surrounding media containing the excreted mIL-12 p70, 10 µL of thawed CCK-8 solution was added to each well. Plates were incubated for 1-1.5 h at the same incubator conditions after which the absorbance was read at 450 nm with a reference wavelength of 600 nm. Cell viability was calculated as

cell viability (%) ) (OD450(sample)/OD450(control)) × 100 where OD450(sample) is the absorbance at 450 nm of the transfected cells and OD450(control) is the absorbance at 450 nm of the negative control (nontransfected cells). ELISA. mIL-12 p70 levels were measured with a BDOptEIA ELISA set for mIL-12 p70 (Pharmingen, San Diego, CA) as per the manufacturer’s instructions. Cells were transfected as per the method previously described. Following the 24 h transfection time, the media was removed and assayed for secreted mIL-12 p70 by ELISA. Levels of mIL-12 p70 were reported as pg/mL. Flow Cytometry. Flow cytometric analysis of EGFPlabeled Renca and B16-F0 cells was carried out using a FACScan flow cytometer and analyzed with accompanying Cell Quest software (Becton Dickinson, Franklin Lakes, NJ). LDL-R Ab Saturation. MCF-7 (human breast carcinoma) cells were used in the LDL-R study to determine whether receptor-mediated endocytosis of LPC-T/pDNA complexes was also possible. To calculate the concentration of anti-LDL-R Ab needed for saturation, we assumed 10 000 LDL-R sites per cell based on a previous flow cytometry study by Li et al. (24). The concentration of anti-LDL-R Ab was determined to be approximately 0.05 µM, assuming a fraction bound (fb) of 0.99 and KA of 2 × 109. Five thousand MCF-7 cells were seeded onto a 96well plate with 100 µL of RPMI + 10% FBS. Cells were incubated at standard conditions to ∼70% confluency after which a saturating concentration of the anti-LDL-R Ab was added in 100 µL of RPMI + 10% FBS. After an incubation of 1 h, the media was removed and replaced with 100 µL fresh RPMI + 10% FBS containing LPC-T/ pmIL-12e. Cells were transfected with 0.1 µg of pmIL12e per 100 µL for 4 h followed by replacing the wells with fresh media and an additional 20 h for transfection. The pmIL-12e vector was used to eliminate background levels of IL-12 p70. RESULTS

We synthesized three LPC conjugates in varying geometries by the addition of cholesterol to the terminal hydroxyl of LPEI (LPC-L), a secondary amine (LPC-T), or both the terminal hydroxyl and a secondary amine (LPC-LT). Following synthesis and purification, we determined the degree of cholesterol conjugation for each LPC conjugate. The target and resulting cholesterol conjugation, based upon 1H NMR were LPC-L, target 1.0, actual 0.991; LPC-T, target 1.0, actual 1.401; and LPCLT, target 2.0, actual 1.782. The 1H NMR results for

Polyethylenimine−Cholesterol Conjugates

Figure 1.

1H

Bioconjugate Chem., Vol. 14, No. 4, 2003 843

NMR spectra for LPC-L (CDCl3).

LPC-L are as follows (Figure 1): 1H NMR (400 MHz, CDCl3) δ 0.70 [s, 3.24 H of CH3 from cholesterol (b)]; δ 1.4-2.1 [587.62 H of secondary amines of LPEI backbone (c)]; δ 2.5-3.15 [1808.55 H of methylene units of LPEI backbone (d)]; and δ 5.38 [s, 1.00 H of dCdCHC from cholesterol (a)]. The peak assignments were confirmed by D2O exchange with LPC-L in which the peaks for the secondary amines disappeared. To show that the assumed geometries were correct, we first attempted to use two-dimensional NMR studies, heteronuclear multiple bond coherence (HMBC) and heteronuclear multiple quantum coherence (HMQC). Because of the HMW LPEI used in this study, the ability to identify the protons near the conjugated secondary amine was limited by the large number of methylene and secondary amine protons. Second, the terminal hydroxyl of the LPEI molecule is more reactive than any of the secondary amines. Consequently, we assumed that the reaction between LPEI and cholesteryl chloroformate would result in a linear geometry (LPC-L). This assumption proved correct as our D2O exchange studies showed (data not shown). Using the base LPEI molecule, we identified the secondary amine peaks by comparing the NMR spectra of the LPEI before and after addition of D2O. With these studies completed, we next performed a D2O exchange study with LPC-L. Again, comparing the spectra before and after addition of D2O, the secondary amine peaks were no longer visible. Finally, we showed that the terminal hydroxyl peak dropped out with conjugation of cholesterol and the TMS protecting group used in the synthesis of LPC-T. Secondary amines are reactive, albeit less than primary amines; therefore, the 1H NMR spectrum of the LPC-LT conjugate showed the loss of the hydroxyl peak as well. The only other possible site for additional cholesterol conjugation is at a secondary amine. The physicochemical properties of the LPC/pDNA complexes were further analyzed by gel retardation, particle size, zeta potential, and AFM. Gel retardation was positive for LPEI/pDNA, LPC-L/pDNA, LPC-T/ pDNA, and LPC-LT/pDNA complexes at N/P 5/1 with complete DNA exclusion at ∼N/P 20/1 (data not shown). The particle size distribution of LPC/pDNA complexes was determined by dynamic light scattering (DLS). The mean particle size for LPC-T/pDNA complexes remained relatively constant over the range of N/P 5/1-30/1 with a mean diameter of ∼275 nm (Figure 2A). A dramatic particle size change was seen for the LPEI, LPC-L, and LPC-LT complexes from N/P 10/1 to 20/1. The LPC-LT complexes showed a particle size inappropriate for use

Figure 2. (A) Particle size data for LPC/pmIL-12e complexes. Data reported as mean ( SD, n ) 3. (B) Zeta potential data for LPC/pmIL-12e complexes. Data reported as mean ( SD, n ) 3.

Figure 3. AFM image of LPC-T/pmIL-12e, N/P 10/1, 90 min after complexation.

in vivo (.300 nm). The slopes of the particle size curves for LPC-LT, LPEI, and LPC-L from N/P 10/1 to 20/1 were quite similar; moreover, this trend was retained in the zeta potential measurements. LPC-T complexes were negatively charged at an N/P ratio of 5/1; however, the complexes quickly became cationic upon doubling the number of nitrogen residues to N/P 10/1 (Figure 3). Results from the AFM study show the complex morphology to be roughly spherical with a discrete particle size of ∼250 nm at 30 and 60 min (data not shown). However, at 90 min postmixing, small satellites were seen to be merging with the ∼250 nm LPC-T complex. We noted that the particle size for LPC-T/pmIL-12e N/P 10/1 from DLS and AFM were both ∼250 nm. The high cationic

844 Bioconjugate Chem., Vol. 14, No. 4, 2003

Figure 4. DNase protection assay for LPC-T/pmIL-12e N/P 10/1 and BPEI/pmIL-12e N/P 10/1.

Furgeson et al.

Figure 6. PEI/pmIL-12e transfection data of Renca cells. mIL12 p70 ELISA data reported as mean ( SD, n ) 8. Table 1. Flow Cytometry Data of PEGFP Expression Following Transfection of Renca and B16-F0 Cellsa Renca % gated average SD

0.38 0.18

average SD

0.42 0.09

average SD

1.09 0.28

average SD

2.08 0.23

average SD

32.14 1.77

average SD

2.23 0.12

geom mean C72.73 4.03 Naked pDNA 69.57 4.41 BPEI 211.96 17.37 LPEI 549.12 77.70 LPC-L 2380.42 90.40 LPC-T 804.26 130.63

B16-F0 % gated

geom mean

0.43 0.27

50.74 6.03

0.45 0.08

48.69 1.87

27.23 1.02

1552.44 20.52

10.14 0.27

1416.54 58.43

30.80 0.32

6484.99 8347.57

13.47 0.06

1269.71 49.18

Figure 5. (A) PEI/pmIL-12e transfection data of B16-F0 cells. mIL-12 p70 ELISA data reported as mean ( SD, n ) 8. (B) PEI/ pmIL-12e cytotoxicity data for B16-F0 cells. CCK-8 data reported as mean ( SD, n ) 8.

a The % gated indicates the percentage of cells (per 10 000) that were positive for EGFP while the geometric mean is indicative for the range of values from these gated cells. Data reported as mean ( SD, n ) 3

density of BPEI, compared to LPEI, was confirmed with the DNase protection assay (Figure 4). After incubation with excess DNase, BPEI fully protected the pDNA up to 180 min while LPEI was successful up to only 60 min; however, LPC-T showed results comparable to BPEI at N/P 10/1. In vitro transfection of B16-F0 cells (murine melanoma) showed ∼2-fold higher mIL-12 p70 expression levels for LPEI and two of the LPC conjugates compared to BPEI over the N/P ratios that we studied (Figure 5A). All transfections were completed in the presence of 10% FBS with transgene expression levels as LPC-L > LPC-T > LPEI > BPEI > LPC-LT. The cytotoxicity profile was evaluated for the same samples used in the ELISA studies. We found a 10% difference in cytotoxicity, on average, between BPEI and the LPEI-LPC group. Renca cells (murine renal cell carcinoma) were also transfected with pmIL-12e (Figure 6), which showed the highest levels for LPC-T/pmIL-12e at N/P 20/1. There were no significant differences between the conjugates at N/P ratios below 20/1 with this cell line; moreover, the cytotoxicity profile was consistent with that previously found with B16-F0 cells (data not shown). The flow cytometry studies (Table 1) showed that LPC-L/pEGFP consistently gave the highest levels of positively trans-

fected cells in addition to the highest geometric mean of EGFP expression. For the Renca cells the percentage of gated cells was nearly 2-times that of BPEI for LPC-T and LPEI. LPC-L levels were nearly 32-times that of BPEI; however, there is not a direct relationship between the percentage of cells gated and geometric mean of the fluorescence detected. Both BPEI and LPEI gave significantly high percentages of B16-F0 cells that were positive for EGFP, but the geometric mean for LPC-L was more than 4-times that of BPEI. B16-F0 cells transfected with LPC-T/pEGFP showed half the percentage of positive cells, but the geometric means were quite close. The final assay was a hLDL-R saturation study showing that receptor-mediated endocytosis via the LDL-R is possible with an LPC conjugate. The LDL-R sites of human mammary carcinoma cells (MCF-7) were saturated with hLDL-R-Ab followed by transfection with LPC-T/pmIL12e. On average mIL-12 p70 levels dropped by ∼20%, strongly indicating receptor-mediated endocytosis via the LDL-R pathway, in addition to the passive pathways of adsorptive endocytosis and pinocytosis (Figure 7). There was a sharp drop of mIL-12 p70 expression levels from N/P 5/1 to 10/1; however, levels remained relatively constant up to N/P 30/1 for the LPC-T and LPC-T plus LDL-R-Ab groups.

Polyethylenimine−Cholesterol Conjugates

Figure 7. Anti-LDL-R Ab saturation study. MCF-7 cells transfected with LPC-T/pmIL-12e (N/P 10/1). mIL-12 p70 ELISA data reported as mean ( SD, n ) 8. DISCUSSION

Although BPEI has been used for years as the nonviral gene delivery standard of choice, LPEI is emerging as a superior alternative due to its favorable charge density and comparable pH buffering as BPEI. To better understand the structural and functional differences between BPEI and LPEI Mw 25 k, we designed a number of experiments to further examine the benefits of HMW LPEI. In addition, upon the basis of the success with the LMW BPEI conjugation to cholesterol resulting in WSLP, we sought to explore whether such favorable changes could also be found with a HMW LPEI. We synthesized three new LPC conjugates in varying geometries from linear (LPC-L), to T-shaped (LPC-T), and finally to a combined conformation incorporating both the linear and T-shaped cholesterol conjugation sites (LPC-LT). The highly reactive terminal hydroxyl of LPEI provided an easy target for cholesterol conjugation, resulting in LPCL. Protection of this reactive group was necessary to facilitate cholesterol conjugation to a secondary amine (LPC-T). Finally, saturating the reaction with excess cholesterol resulted in conjugation to both a secondary amine and the terminal hydroxyl (LPC-LT). The actual degrees of cholesterol conjugation were as expected based upon the 1H NMR studies. The LPC conjugates were characterized by a number of methods including particle size, zeta potential, surface morphology by AFM, and gel retardation. The dramatic change in particle size for the LPEI, LPC-L, and LPCLT complexes from N/P 10/1 to 20/1 is probably due to a conformation change of some sort resulting in a higher particle size due to intermolecular hydrogen bonding between neighboring complexes. The similar slopes within the particle size curves for LPC-LT, LPEI, and LPC-L from N/P 10/1 to 20/1 further suggest a conformation change within the LPC conjugates upon doubling the number of nitrogen residues per pDNA. Zeta potential measurements show the same trend in slope from N/P 10/1 to 20/1 for LPC-L, LPC-T, and LPC-LT (Figure 2B). LPC-T complexes exhibited a negative charge at N/P 5/1, suggesting that the total pDNA is not fully condensed. The surface charge becomes positive for LPC-T after increasing the N/P ratio to 10/1. Again, conformational changes could explain the abrupt change in surface charge for LPEI, LPC-L, and LPC-LT from N/P 5/1 to 10/1. Reorientation of the polymer complexes at this point could account for the dramatic increase in surface charge for the majority of the complexes from N/P 10/1 to 20/1. The initial negative charge of the LPC-T/pDNA complex at N/P 5/1 was interesting in that it could be accounted

Bioconjugate Chem., Vol. 14, No. 4, 2003 845

for if the LPEI backbone is shielding the hydrophobic cholesterol moiety from the surrounding aqueous environment. Knowing that the charge density is less than BPEI of the same molecular weight, the number of protonated secondary amines with LPC-T available for electrostatic condensation of the pDNA is less, hence additional LPC-T conjugates are needed to effectively condense the pDNA thus producing a positive surface charge. Upon mixing 1.25 mg/mL stock solutions of the LPC conjugates, LPC-T was the only conjugate that did not require brief heating in a hot water bath to remove turbidity resulting from LPC aggregation. Aggregation effects with LPC-T/pDNA complexes were evaluated with an AFM study of the surface morphology of LPC-T/pDNA complexes at N/P 10/1. Images were taken at 30, 60, and 90 min post-mixing of LPC-T with pDNA in order to determine the potential for aggregation due to intermolecular hydrogen bonding between neighboring LPC-T complexes. The appearance of the small satellites with the AFM image suggests that LPC-T/pDNA complexes at N/P 10/1 are stable up to 90 min at ambient conditions. To evaluate the protective nature of pDNA condensation with the LPC conjugates, a DNase protection assay was completed. BPEI, due to its higher charge density, would condense pDNA more tightly than LPEI at the same N/P ratio. This supposition was confirmed with the DNase protection assay (Figure 4), which shows BPEI providing comparable pDNA protection up to 180 min in the presence of excess of DNase. LPEI was able to effectively protect pDNA up to 60 min while LPC-T could protect the pDNA up to 180 min, similar to BPEI. These data suggest that LPC-T forms a favorable complex conformation that shuttles the pDNA away from the deleterious DNase without the use of a high cationic charge density like that found with BPEI. B16-F10 cells (murine melanoma cells) are easily transfected with polyethylenimine (PEI) (25); consequently, a close relative of this cell line, B16-F0, was specifically used to evaluate the structural differences between LPEI and BPEI. Because of the high levels of transfection, effects due to the plasmid construct were assumed to be minimal while the conformation of the PEI carriers was the dominant force in transfection levels. This finding is significant, and thus we can conclude that the different degrees of transfection are due to the PEI molecule and not from the vector (pmIL-12e). The transfections were carried out in RPMI + 10% FBS, thus the high cationic state of the BPEI/pDNA complexes may contribute to the lower transfection values due to unfavorable electrostatic interactions with serum components. Levels were expected to be low for the LPC-LT complexes due to the unfavorable complex size (Figure 2A). These data strongly suggest that LPEI and analogues may be superior to BPEI of the same molecular weight for nonviral gene delivery. The same samples used in the mIL-12 p70 study were also evaluated for cytotoxicity (Figure 5B). On average the differences in cytotoxicity were over 10%, further confirming the advantages of LPEI and its LPC analogues over BPEI. Cytotoxicity with the BPEI/pDNA complexes is due to the high cationic state of the complexes, whereas cytotoxicity with the LPEI/pDNA complexes is due to possible aggregation of the complexes by intermolecular hydrogen bonding of the LPEI backbones. This aggregation could explain the drop in cell viability at high N/P ratios. As the number of complexes increase, the ability for intermolecular hydrogen bonding also increases. Ongoing in vivo studies with the LPC conjugates are for the treatment of subcutaneous tumors and pulmonary metastases. We decided upon

846 Bioconjugate Chem., Vol. 14, No. 4, 2003

Renca cells for our target cell line as they are somewhat immunogenic and are not necessarily easily transfected by PEI. Figure 6 shows the results from pmIL-12e transfections, where the highest levels were seen with LPC-T/pmIL-12e at N/P 20/1. Appreciable differences between mIL-12 p70 levels were not seen below N/P 20/ 1; however, the trend of LPEI and LPC conjugates providing higher transfection efficiencies was retained. The high levels of EGFP expression seen in the flow cytometry studies can be due to increased endocytosis rates, possibly due to LDL-R mediated endocytosis, increased protection of the pDNA prior to endocytosis and during lysosomal degradation, and high rates of release in the cytosolic compartment followed by nuclear translocation. We would expect a large number of cells to test positive for EGFP due to the high electrostatic binding between BPEI and pDNA; however, this tight binding of the pDNA could also prevent the pDNA from releasing quickly from this acidic environment. LPEI and the LPC conjugates possess enough cationic charge to adequately condense the pDNA, however, the release of the pDNA from the endosomal compartment should be quicker due the less compact binding. Furthermore, cationic charge density may be even further reduced by the addition of cholesterol to the LPEI backbone (26). Upon sequestering the cholesterol from the aqueous environment, the LPC conjugate could further use the region shielding the cholesterol for endosomal buffering after endocytosis. Cleavage of the cholesterol moiety from the LPEI backbone by lipase could result in an immediate increase in proton buffering. An additional purpose of this study was to further demonstrate mechanisms by which PEI-cholesterol conjugates could be internalized. Mechanisms by which polyplexes may be internalized into a cell include pinocytosis, adsorptive endocytosis, and receptor-mediated endocytosis. It is well-known that the human low density lipoprotein receptor (LDL-R) consists of five domains of which the outermost domain, the “ligand-binding domain”, is negatively charged (27). Furthermore, Goldstein et. al confirmed that cholesteryl esters are cleaved from lipoproteins by lipase, a lysosomal acid (28). Therefore, it is plausible that LPC/pDNA complexes could exploit both of these characteristics for internalization and endosomal release of the transgenic pDNA based upon our results. Conjugation of highly hydrophobic cholesterol to a hydrophilic polymer such as PEI followed by hydration results in a complex in which the PEI reorients itself so as to shield the cholesterol from the surrounding aqueous environment. With a high molecular weight, LPEI such as that used in these studies, a number of conformations were possible dependent upon the site and degree of cholesterol conjugation. We hypothesized that the LPC-T conjugate would provide the highest transfection levels based upon a number of factors including increased water solubility by decreasing the degree of hydrogen bonding between neighboring LPEI backbones, favorable protection of the cholesterol by the hydrophilicinteractions of the two neighboring strands, and increased, late-stage pH buffering capacity after cleavage of the cholesterol from the LPEI backbone by lipase. After the late endosome fuses with the lysosome, the ethylenimine monomers used to shield the cholesterol could become available for further pH buffering, a process we termed “unprotonated reserves”. Future work will include biodistribution studies in normal and diseased mice in addition to further optimization of the LPC gene carriers by the addition of targeting ligands. PEGylation of LPC-L should provide

Furgeson et al.

adequate water solubility and a resulting decrease in particle size. LPC-L showed consistently higher transgene expression; however, the particle size and degree of cytotoxicity are currently not appropriate for in vivo studies. Finally, optimization of the size and degree of conjugation of the hydrophobic moiety on the LPEI will be completed. ACKNOWLEDGMENT

We thank Jay Olsen for his assistance with the NMR studies and Andras Pungor for his assistance with the AFM. This project was generously supported by Expression Genetics, Inc. LITERATURE CITED (1) von Harpe, A., Petersen, H., Li, Y., and Kissel, T. (2000) Characterization of commercially available and synthesized polyethylenimines for gene delivery. J. Controlled Release 69, 309-322. (2) Godbey, W. T., Wu, K. K., and Mikos, A. G. (1999) Size matters: Molecular weight affects the efficiency of poly(ethylenimine) as a gene delivery vehicle. J. Biomed. Mater. Res. 45, 268-275. (3) Fischer, D., Bieber, T., Li, Y., Elsasser, H. P., and Kissel, T. (1999) A novel nonviral vector for DNA delivery based on low molecular weight, branched polyethylenimine: effect of molecular weight on transfection efficiency and cytotoxicity. Pharm. Res. 16, 1273-1279. (4) Petersen, H., Fechner, P. M., Martin, A. L., Kunath, K., Stolnik, S., Roberts, C. J., Fischer, D., Davies, M. C., and Kissel, T. (2002) Polyethylenimine-graft-poly(ethylene glycol) copolymers: influence of copolymer block structure on DNA complexation and biological activities as gene delivery system. Bioconjugate Chem. 13, 845-854. (5) Fischer, D., Li, Y., Ahlemeyer, B., Krieglstein, J., and Kissel, T. (2003) In vitro cytotoxicity testing of polycations: influence of polymer structure on cell viability and hemolysis. Biomaterials 24, 1121-1131. (6) Abdallah, B., Hassan, A., Benoist, C., Goula, D., Behr, J. P., and Demeneix, B. A. (1996) A powerful nonviral vector for in vivo gene transfer into the adult mammalian brain: polyethylenimine. Hum. Gene Ther. 7, 1947-1954. (7) Goula, D., Remy, J. S., Erbacher, P., Wasowicz, M., Levi, G., Abdallah, B., and Demeneix, B. A. (1998) Size, diffusibility and transfection performance of linear PEI/DNA complexes in the mouse central nervous system. Gene Ther. 5, 712-717. (8) Coll, J. L., Chollet, P., Brambilla, E., Desplanques, D., Behr, J. P., and Favrot, M. (1999) In vivo delivery to tumors of DNA complexed with linear polyethylenimine. Hum. Gene Ther. 10, 1659-1666. (9) Chollet, P., Favrot, M. C., Hurbin, A., and Coll, J. L. (2002) Side-effects of a systemic injection of linear polyethylenimineDNA complexes. J. Gene Med. 4, 84-91. (10) Goula, D., Becker, N., Lemkine, G. F., Normandie, P., Rodrigues, J., Mantero, S., Levi, G., and Demeneix, B. A. (2000) Rapid crossing of the pulmonary endothelial barrier by polyethylenimine/DNA complexes. Gene Ther. 7, 499-504. (11) Zou, S. M., Erbacher, P., Remy, J. S., and Behr, J. P. (2000) Systemic linear polyethylenimine (L-PEI)-mediated gene delivery in the mouse. J. Gene Med. 2, 128-134. (12) Uduehi, A. N., Stammberger, U., Kubisa, B., Gugger, M., Buehler, T. A., and Schmid, R. A. (2001) Effects of linear polyethylenimine and polyethylenimine/DNA on lung function after airway instillation to rat lungs. Mol. Ther. 4, 52-57. (13) Wightman, L., Kircheis, R., Rossler, V., Carotta, S., Ruzicka, R., Kursa, M., and Wagner, E. (2001) Different behavior of branched and linear polyethylenimine for gene delivery in vitro and in vivo. J. Gene Med. 3, 362-372. (14) Brunner, S., Furtbauer, E., Sauer, T., Kursa, M., and Wagner, E. (2002) Overcoming the nuclear barrier: cell cycle independent nonviral gene transfer with linear polyethylenimine or electroporation. Mol. Ther. 5, 80-86. (15) Kircheis, R., Ostermann, E., Wolschek, M. F., Lichtenberger, C., Magin-Lachmann, C., Wightman, L., Kursa, M.,

Polyethylenimine−Cholesterol Conjugates and Wagner, E. (2002) Tumor-targeted gene delivery of tumor necrosis factor-alpha induces tumor necrosis and tumor regression without systemic toxicity. Cancer Gene Ther. 9, 673-680. (16) Boussif, O., Lezoualc’h, F., Zanta, M. A., Mergny, M. D., Scherman, D., Demeneix, B., and Behr, J. P. (1995) A versatile vector for gene and oligonucleotide transfer into cells in culture and in vivo: polyethylenimine. Proc. Natl. Acad. Sci. U.S.A. 92, 7297-7301. (17) Furgeson, D. Y., Cohen, R. N., Mahato, R. I., and Kim, S. W. (2002) Novel water insoluble lipoparticulates for gene delivery. Pharm. Res. 19, 382-390. (18) Suh, W., Han, S. O., Yu, L., and Kim, S. W. (2002) An angiogenic, endothelial-cell-targeted polymeric gene carrier. Mol. Ther. 6, 664-672. (19) Benns, J. M., Maheshwari, A., Furgeson, D. Y., Mahato, R. I., and Kim, S. W. (2001) Folate-PEG-folate-graft-polyethylenimine-based gene delivery. J. Drug Targeting 9, 123139. (20) Han, S.-O., Mahato, R. I., and Kim, S. W. (2001) Watersoluble lipopolymer for gene delivery. Bioconjugate Chem. 12, 337-345. (21) Mahato, R. I., Lee, M., Han, S., Maheshwari, A., and Kim, S. W. (2001) Intratumoral delivery of p2CMVmIL-12 using water-soluble lipopolymers. Mol. Ther. 4, 130-138. (22) Yockman, J. W., Maheshwari, A., Han, S., and Kim, S. W. (2003) Tumor regression by repeated intratumoral delivery of water soluble lipopolymers/p2CMVmIL-12 complexes. J. Controlled Release 87, 177-186.

Bioconjugate Chem., Vol. 14, No. 4, 2003 847 (23) Wang, D. A., Narang, A. S., Kotb, M., Gaber, A. O., Miller, D. D., Kim, S. W., and Mahato, R. I. (2002) Novel branched poly(ethylenimine)-cholesterol water-soluble lipopolymers for gene delivery. Biomacromolecules 3, 1197-1207. (24) Li, Y., Wood, N., Parsons, P. G., Yellowlees, D., and Donnelly, P. K. (1997) Expression of alpha2-macroglobulin receptor/low-density lipoprotein receptor-related protein on surfaces of tumour cells: a study using flow cytometry. Cancer Lett. 111, 199-205. (25) Wightman, L., Patzelt, E., Wagner, E., and Kircheis, R. (1999) Development of transferrin-polycation/DNA based vectors for gene delivery to melanoma cells. J. Drug Targeting 7, 292-303. (26) Suh, J., Paik, H.-J., and Hwang, B. K. (1994) Ionization of poly(ethylenimine) and poly(allylamine) at various pH's. Bioorg. Chem. 22, 318-327. (27) Yamamoto, T., Davis, C. G., Brown, M. S., Schneider, W. J., Casey, M. L., Goldstein, J. L., and Russell, D. W. (1984) The human LDL receptor: a cysteine-rich protein with multiple Alu sequences in its mRNA. Cell 39, 27-38. (28) Goldstein, J. L., Dana, S. E., Faust, J. R., Beaudet, A. L., and Brown, M. S. (1975) Role of lysosomal acid lipase in the metabolism of plasma low-density lipoprotein. Observations in cultured fibroblasts from a patient with cholesteryl ester storage disease. J. Biol. Chem. 250, 8487-8495.

BC0340565