The self-assembly of biodegradable cationic polymer micelles as vectors for gene transfection

ARTICLE IN PRESS

Biomaterials 28 (2007) 5358–5368 www.elsevier.com/locate/biomaterials

The self-assembly of biodegradable cationic polymer micelles as vectors for gene transfection Yong Wanga,b, Chyan-Ying Kea,c, Cyrus Weijie Beha,d, Shao-Qiong Liua,e, Suat-Hong Gohb, Yi-Yan Yanga, a Institute of Bioengineering and Nanotechnology, 31 Biopolis Way, Singapore 138669, Singapore Department of Chemistry, National University of Singapore, 3 Science Drive 3, Singapore 117543, Singapore c Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD 21218, USA d Division of Bioengineering, National University of Singapore, 7 Engineering Drive 1, Singapore 117574, Singapore e Department of Pharmacy, National University of Singapore, 18 Science Drive 4, Singapore 117543, Singapore b

Received 12 April 2007; accepted 7 August 2007 Available online 30 August 2007

Abstract Cationic micelles self-assembled from a biodegradable amphiphilic copolymer, poly{(N-methyldietheneamine sebacate)-co[(cholesteryl oxocarbonylamido ethyl) methyl bis(ethylene) ammonium bromide] sebacate} (P(MDS-co-CES)) have recently been reported for efficient gene delivery and co-delivery of drug and nucleic acid. In this study, poly(ethylene glycol) (PEG) of various molecular weights (Mn ¼ 550, 1100 and 2000) was conjugated to P(MDS-co-CES) having different cholesterol grafting degrees to improve the stability of micelle/DNA complexes in the blood for systemic in vivo gene delivery. DNA binding ability, gene transfection efficiency and cytotoxicity of P(MDS-co-CES), PMDS, PEGylated PMDS and PEGylated P(MDS-co-CES) micelles were studied and compared. As with P(MDS-co-CES), PEG–P(MDS-co-CES) polymers could also self-assemble into stable micelles of small size. However, PMDS and PEG–PMDS without cholesterol could not form stable micelles but formed large particles. PEGylation of polymers significantly decreased their gene transfection efficiency in HEK293, HepG2, HeLa, MDA-MB-231 and 4T1 cells. However, increasing N/P ratio promoted gene transfection. An increased cholesterol grafting degree led to greater gene expression level possibly because of the more stable core–shell structure of the micelles. PEG550–P(MDS-co-CES) micelles induced high gene transfection level, comparable to that provided by P(MDS-co-CES) micelles. PEGylated polymers were much less cytotoxic than P(MDS-co-CES). PEGylated P(MDS-co-CES) micelles may provide a promising non-viral vector for systemic in vivo gene delivery. r 2007 Elsevier Ltd. All rights reserved. Keywords: Cationic polymer micelles; PEGylation; Self-assembly; Gene transfection

1. Introduction Since Fraley et al. [1] reported liposome-based non-viral gene delivery vector in 1980, many natural and synthetic materials have been explored as alternatives, including cationic polymers [2], block copolymer-coated calcium phosphate nanoparticles [3], inorganic nanoparticles such as amine-coated silica [4] and metal [5] nanoparticles as well as carbon nanotubes [6]. Of these vectors, cationic polymers are the most attractive because they can be easily Corresponding author. Tel.: +65 68247106; fax: +65 64789084.

E-mail address: [email protected] (Y.-Y. Yang). 0142-9612/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2007.08.013

synthesized and tailored to suit the special requirements such as biocompatibility, DNA binding ability and endosomolytic property for gene delivery. In general, the cationic polymers such as polyethylenimine (PEI) and its derivatives [7,8], polylysine [9], polyamidoamine dendrimers [10], poly(beta-amino esters) [11], and chitosan [12] are water-soluble and would directly condense DNA in aqueous media. Although many of these cationic polymers provide high in vitro gene transfection efficiency, the stability of the cationic polymer/DNA complexes and their ability in inducing sufficient in vivo gene transfection level still remain a challenge. Hydrophilic polymers such as poly(ethylene glycol) (PEG) have been grafted onto various

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368

cationic polymers to improve the stability of the polymer/ DNA complexes in the blood by preventing protein absorption and uptake of macrophages, mononuclear phagocytes and reticuloendothelial systems (RES) [13,14]. Although the introduction of PEG leads to a decrease in in vitro gene transfection efficiency possibly due to the lower endolysosomal escaping capacity and weaker DNA binding ability caused by the steric hindrance of PEG, PEG-conjugated cationic polymers induced higher in vivo gene transfection efficiency due to the improved stability in the blood. Cationic polymer particles self-assembled from cholesterol-grafted PEI have also been investigated for gene delivery [15]. The gene transfection efficiency induced by cholesterol-grafted PEI (Mw 1800) was higher than that provided by unmodified PEI (Mw 1800), and the cytotoxicity of cholesterol-grafted PEI was also reduced. It was suggested that the cholesterol-grafted PEI/DNA complexes might be internalized by cells through a cellular cholesterol uptake pathway. We have recently synthesized a biodegradable, cationic and amphiphilic copolymer, which consists of cholesterol side chains and a cationic main chain carrying quaternary ammonium for DNA binding and tertiary amine for endosomal buffering. This copolymer easily formed core– shell nanoparticles (i.e. micelles) having a cholesterol core and a cationic shell in aqueous media by a self-assembly process. Hydrophobic drugs can be incorporated into the core during the self-assembly process. The cationic shell of the resulting drug-loaded micelles can be employed to bind DNA. These micelles were used to co-deliver both drugs and genes [16], and a synergistic effect was achieved by co-delivering an anticancer drug (i.e. paclitaxel) and IL-12 encoded plasmid or BCl-2 targeted siRNA. In addition, the micelles induced high gene transfection efficiency in various cell lines and primary human cells, which was comparable to that provided by PEI (Mw 25 kDa). In particular, they yielded much higher gene transfection efficiency in mouse and human breast cancer cell lines. In this study, PEG with various number-average molecular weights (i.e. 550, 1100 and 2000 Da) was capped onto the ends of the polymer to improve the stability of micelle/DNA complexes for systemic in vivo gene delivery. The gene transfection efficiency induced by PEGylated polymer micelles was studied in HEK293, HepG2, HeLa, MDA-MB-231 and 4T1 cells in comparison with the unmodified polymer micelles. The effects of PEGylation, core–shell structure, cholesterol grafting degree and N/P ratio on particle size, zeta potential, DNA binding ability and in vitro gene transfection efficiency were investigated. In addition, the cytotoxicity of PEGylated polymer micelles was also evaluated in these cell lines. 2. Materials and methods

5359

N,N-dimethylformamide (DMF), and anhydrous dimethyl sulfoxide (DMSO) were purchased from Aldrich, USA. Triethylamine (X99%), 2-bromethylamine hydrobromide (499%), monomethoxy poly(ethylene glycol) (mPEG) [i.e. poly(ethylene glycol) monomethyl ether] with number-average molecular weights of 550, 1100 and 2000 Da, ethidium bromide and anhydrous sodium carbonate were obtained from Sigma, USA. Tetrahydrofuran (THF), ether, toluene and chloroform of ACS grade were purchased from Tedia or Merck, USA. Magnesium sulfate, sodium acetate and anhydrous acetic acid were obtained from Merck, USA. Agarose powder of biological grade was purchased from Bio-Rad, USA. HEK293, HepG2, HeLa, MDA-MB-231 and 4T1 cells were purchased from ATCC, USA. Plasmid DNA encoding the 6.4 kb firefly luciferase (pCMV-luciferase VR1255_C) driven by the cytomegalovirus (CMV) promoter and GFP reporter gene encoding the GFPmut1 variant (pEGFP-C1) with 4.7 kb driven by the SV 40 early promoter were obtained from Carl Wheeler (Vical, USA) and Clontech (USA), respectively. Cholesteryl chloroformate and 2-bromethylamine hydrobromide were used as received. N-Methyldiethanolamine and sebacoyl chloride were purified by distillation under vacuum. Triethylamine was treated with toluene sulphonyl chloride to remove primary and secondary amine, and then distilled and freshly dried with sodium prior to synthesis. THF was freshly dried with sodium and distilled before use. Benzophenone was used to indicate that the moisture was removed completely. Toluene was dried by sodium before use. Sodium acetate buffer (0.02 M, pH 4.6) was selfprepared.

2.2. Synthesis of poly(N-methyldietheneamine sebacate (PMDS) N-Methyldiethanolamine (5.958 g, 0.05 mol) and 50.5 g of triethylamine (0.5 mol) were added to a 250-ml round-bottom flask with 50 ml of freshly dried THF in a dry ice/acetone bath (below 30 1C). Freshly dried THF (40 ml) containing 11.945 g of sebacoyl chloride (0.05 mol) was added dropwise to the flask with stirring. The flask was removed 1 h later, and the reaction was allowed to proceed at room temperature overnight. The reaction mixture was filtered to harvest the solution. The solvent and residual triethylamine were removed using a rotavapor. The crude product dissolved in 100 ml of toluene was extracted four times with 50 ml of NaClsaturated aqueous solution, and then dried with anhydrous NaCO3. It was further dialyzed in acetone using a membrane with a molecular weight cutoff of 3.5 kDa. Acetone was subsequently removed from the dialysate using the rotavapor, and the final product was dried in a vacuum oven for 2 days. The yield was 75%.

2.3. Synthesis of N-(2-bromoethyl) carbarmoyl cholesterol Chloroform (50 ml) dried with a molecular sieve was put into a 100-ml round-bottom flask in a dry ice/acetone bath. Cholesteryl chloroformate (4.34 g, 0.0097 mol) and 2.18 g of 2-bromoethylamine hydrobromide (0.0106 mol) were then added with stirring. Next, 3 ml of freshly dried triethylamine was added to the flask, which was moved after 30 min for the reaction to proceed at room temperature for 12 h. The organic solution was washed three times with 20 ml of 1 N HCl solution saturated with NaCl, and once with 30 ml of NaCl-saturated aqueous solution to remove residual triethylamine. The organic phase was collected and dried with 5 g of anhydrous magnesium sulfate. The solution was then filtered and distilled. The crude product was recrystallized with anhydrous ethanol once, and with anhydrous acetone twice. The final product was dried with a vacuum oven for 24 h. The yield was 78%.

2.1. Materials

2.4. Synthesis of poly{(N-methyldietheneamine sebacate)-co[(cholesteryl oxocarbonylamido ethyl) methyl bis(ethylene) ammonium bromide] sebacate (P(MDS-co-CES))

Cholesteryl chloroformate (98%), N-methyldiethanolamine (99%), sebacoyl chloride (97%), water-free PEI (branched, 25 kDa),

PMDS (2.85 g, 0.01 mol) and 5.5 g of N-(2-bromoethyl) carbarmoyl cholesterol (0.01 mol) were dissolved in 50 ml of dry toluene, and refluxed

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368

5360

for 2 days under argon. Diethyl ether (250 ml) was then added to precipitate the product. To completely remove unreacted N-(2-bromoethyl) carbarmoyl cholesterol, the product was washed with diethyl ether four more times. The yield was 70%.

The nitrogen content of the polymers was determined by elemental analysis using Perkin-Elmer Instruments Analyzer 2400.

2.5. PEGylation of PMDS

2.10. Preparation of polymeric micelles

N-Methyldiethanolamine (5.958 g, 0.05 mol), 0.00125 mol pre-dried mPEG (2.5 g, 1.375 g, 0.8125 g for Mn of 2000, 1100 and 550 Da, respectively) and 0.5 mol triethylamine (50.5 g) were added to 150-ml round-bottom flask with 50 ml of freshly dried THF in a dry ice/acetone bath (below 30 1C). Freshly dried THF (40 ml) containing 11.945 g sebacoyl chloride (0.05 mol) was added dropwise to the flask with stirring. The flask was removed 1 h later, and the reaction was allowed to proceed at room temperature for 1 more day. The solvent and residual triethylamine were removed using a rotavapor. The crude product was washed three times with 300 ml of THF and the solution was collected by filtration. The solvent was then removed using the rotavapor. The crude product was semi-solid, which was put in a vacuum oven overnight to further remove triethylamine. Thereafter, the crude product was washed by using ether to remove the oligomers and triethylamine residues, and dried under vacuum overnight. PMDS conjugated with PEG of 550, 1100, 2000 Da were labeled as PEG550–PMDS, PEG1100–PMDS and PEG2000–PMDS. The PEGylated products were dissolved in acetone and dialyzed against acetone using a dialysis membrane with a molecular weight cut-off of 3.5 kDa for 3 days to further remove the unreacted PEG and other impurity. The yield of PEGylated PMDS was 70%.

Polymeric micelles were prepared by a membrane dialysis method using the cationic polymers. Briefly, 15 mg of polymer was dissolved in 5 ml of DMF, which was then dialyzed against 500 ml of the sodium acetate buffer for 24 h using a dialysis membrane with a molecular weight cut-off of 2 kDa (Sigma, D-7884 or Spectrum 2000). The external aqueous solution was changed every hour for the first 8 h and then every 8 h.

2.6. Synthesis of PEGylated P(MDS-co-CES) PEGylated PMDS was firstly characterized by using 1HNMR to determine the ratio of PEG block to PMDS block, and the amount of cholesteryl derivative N-(2-bromoethyl) carbarmoyl cholesterol (Be-chol) was added according to the amount of PMDS units in a molar ratio of 1:1. For example, the content of PMDS calculated by 1H NMR was 80% in weight. Thus, 2.5 g PEGylated PMDS contained 2.0 g PMDS, i.e. 0.007 mol of repeated PMDS units (2.0  285 ¼ 0.007). Therefore, the amount of Be-chol added was 3.74 g (0.007 mol). By changing the molar ratio of PMDS units to Be-chol, PEGylated P(MDS-co-CES) with different grating degree of cholesterol groups can be obtained. PEGylated PMDS and Be-chol were dissolved in 100 ml of dry toluene and refluxed at 120 1C for 24 h under argon. Toluene was removed by distillation using the rotavapor. The crude product was washed with diethyl ether four times to remove unreacted N-(2-bromoethyl) carbarmoyl cholesterol, and dried overnight in a vacuum oven. The yield was 50%.

2.7. 1HNMR analysis 1

The HNMR spectra of the polymers in CDCl3 were recorded on a Bruker AVANCE 400 spectrometer (400 MHz). Chemical shifts were expressed in parts per million (d) using tetramethyl silane in the indicated solvent as the internal standard.

2.8. Gel permeation chromatography (GPC) analysis The weight- and number-average molecular weights of PMDS, P(MDS-co-CES), PEGylated PMDS and PEGylated P(MDS-co-CES) were determined by GPC (Waters 2690, MA, USA) with a differential refractometer detector (Waters 410, MA, USA). Polymer sample (10 mg) was dissolved in 5 ml of THF and the solution was then filtered. The mobile phase was THF with a flow rate of 1 ml/min. Weight and numberaverage molecular weights were calculated from a calibration curve using a series of polystyrene standards (Polymer Laboratories Inc., MA, USA, with molecular weight ranging from 1300 to 30,000).

2.9. Elemental analysis

2.11. Particle size and zeta potential measurements The particle size and zeta potential of the freshly prepared micelles were measured by ZetaPals equipped with a He–Ne laser beam at 658 nm (scattering angle: 901) (Brookhaven Instruments Corp, USA) at 25 1C. Each measurement was repeated five times. An average value was obtained from the five measurements.

2.12. Agarose gel electrophoresis assays The DNA binding ability of the polymeric micelles was analyzed by agarose gel electrophoresis. The micelle/DNA complexes containing 0.28 mg luciferase reporter gene were prepared at various N/P ratios. The N/P ratio means the molar ratio of amine groups in the cationic polymer, which represent the positive charges, to phosphate groups in the plasmid DNA, which represent the negative charges. The DNA complex solutions at various N/P ratios were diluted to an identical volume (i.e. 8 ml) by using the same buffer employed for the preparation of micelles. Five times DNA loading buffer (2 ml) was added to the complex solutions. The mixtures were allowed to stay at room temperature for 45 min. Thereafter, the complexes were loaded into individual wells of 1.0% agarose/1  TAE gel containing 0.5 mg/ml ethidium bromide, and electrophoresed at 100 V for 90 min. The naked DNA diluted with the same buffer without adding the micelles and the micelles without adding DNA were used as the controls. The resulting DNA migration patterns were revealed under UV irradiation (Vilber Lourmat, France) and the photos were taken.

2.13. Protein adsorption of P(MDS-co-CES), PEGylated P(MDS-co-CES) micelles and their DNA complexes The main purpose of PEGylation onto the polymer was to form a PEG shell on the surfaces of micelles, reducing the opsonization of surfaces by hindering the approach of the proteins, which would otherwise result in a protein coat that can then be detected by the RES in the body. Bovine serum albumin (BSA) was used as a model protein to simulate non-specific protein adsorption onto the surfaces of the micelles. BSA was dissolved in water, giving a 5 mg/ml stock solution with 0.05% sodium azide. This was further diluted 10 times using phosphate-buffered saline (PBS) to a 0.5 mg/ml solution before addition to 0.8 mg/ml polymer/complex solutions. 0.5 ml of each solution was used, which was made up to 2 ml using PBS. The final concentration of BSA and polymer was 0.125 and 0.2 mg/ml, respectively. The pH of the test solution was approximately 7, which approximated the physiological conditions. After incubation for the stipulated period of time, 200 ml of each sample was aliquoted after vortexing for 10 s to ensure homogeneity, and then centrifuged at 10,000g for 15 min to precipitate the micelles. A control sample without any polymer was prepared, with the polymer solution being replaced by the sodium acetate buffer. After centrifugation, the supernatant was carefully removed to prevent resuspending the pellet, and tested using the BCA

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368 protein assay (Bio-Rad, USA). An uncentrifuged BSA solution is used to normalize the results.

5361

the cells were suspended in 0.3 ml of 1% paraformaldehyde for fixation prior to analyses by a cell cytometer (EPICS ELITE ESP, Coulter, USA).

2.14. Preparation of micelle/DNA complexes

3. Results and discussion The micelles for the in vitro gene transfection studies were prepared as describe above. Thereafter, the micelle solution was harvested, and the volume was measured to calculate the concentration of the micelles. The solution was sterilized by filtration using 0.22 mm filter, and a certain volume of micelle solution was added into an identical volume of the plasmid to form micelle/DNA complexes. The complex solution was diluted using the buffer to an identical volume (50 ml per well), and allowed to incubate at room temperature for 45 min.

2.15. Cytotoxicity test The cytotoxicity of PEI, P(MDS-co-CES) and PEGylatd P(MDS-coCES) (PEG Mn 550 or 2000) micelles was studied against HEK293, HepG2, HeLa, MDA-MB-231 and 4T1 cells. Cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (for HEK293, HepG2 and HeLa cells) or RPMI 1640 (for 4T1 cells) or L-15 (for MDA-MB-231 cells) supplemented with 10% FBS, 2 mM L-glutamine, 100 U/ml penicillin and 100 mg/ml streptomycin at 37 1C under an atmosphere with 5% CO2. Cells were seeded onto 96-well plates at a density of 10,000 cells/well. The plates were then returned to the incubator. In the morning of the tests, the media in the wells were replaced with 150 ml of fresh media. The micelles or PEI solution (50 ml) with varying concentration was then added to each well. The sodium acetate buffer of an equivalent volume was used as the negative control. The plates were then returned to the incubators. Each sample was tested in eight replicates per plate. After 24 h, aliquots of MTT solution (20 ml) were added into each well after the designated period. The plates were then returned to the incubator. After 3 h of incubation, the growth medium in each well was removed, and 150 ml of DMSO were added to each well to dissolve the internalized purple formazan crystals. An aliquot of 100 ml was taken from each well, and transferred to a new 96-well plate. The plates were then assayed at 550 and 690 nm using a microplate reader (PowerWave X, Bio-Tek Instruments). The absorbance readings of the formazan crystals were taken to be that at 550 nm subtracted by that at 690 nm. The results were expressed as a percentage of the absorbance of the negative control.

2.16. In vitro gene expression The in vitro gene transfection of the micelle/DNA complexes was performed in HEK293, HepG2, HeLa, MDA-MB-231 and 4T1 cells. Cells were seeded onto 24-well plates at a density of 8  104 cells/well for luciferase gene transfection, and 12-well plates at a density of 2  105 cells/ well for GFP gene transfection, and cultivated in 0.5 and 1.0 ml of growth medium, respectively. After 24 h, the growth media were replaced with fresh ones, and the DNA complex solution containing 2.5 mg luciferase reporter gene or 3.5 mg GFP reporter gene was added to each well. After 4 h of incubation, the culture media were replaced with fresh ones. The culture media were removed after 2 days, and the cells on the 24-well plates were washed with 0.5 ml of PBS. Reporter lysis buffer (0.2 ml) was then added to each well to lyse the cells. The cell suspension was next frozen in 80 1C for half an hour and thawed, followed by centrifugation at 14,000 rpm for 5 min. The relative light units (RLU) were measured using a luminometer (Bio-Rad, USA), and normalized to the protein content using the BCA protein assay (Bio-Rad, USA). The PEI/DNA complexes at N/P ratio of 10 were used as the positive control. For the GFP gene transfection, the cells were harvested by a different protocol: after 2 days of incubation, the cells on the 12-well plates were washed with 1.0 ml of PBS. A 0.3 ml of 1  trypsin solution was then added to each well, which was incubated at room temperature for 10–15 min to detach the cells. The cell suspension was centrifuged at 14,000 rpm for 5 min, and re-suspended in PBS (pH 7.4). Upon separation from PBS by centrifugation,

3.1. Synthesis and characterization of cationic amphiphilic copolymer The cationic amphiphilic copolymer was synthesized by a three-step synthesis as reported previously [16]. First, the main chain, PMDS, was produced by condensation polymerization between N-methyldiethanolamine and sebacoyl chloride in dry THF. Excess triethylamine was used to remove hydrochloride and limit protonation of the tertiary amine. Next, cholesteryl chloroformate was allowed to react with 2-bromoethylamine hydrobromide in an amidation reaction. The resulting hydrophobic N-(2-bromoethyl) carbarmoyl cholesterol (Be-chol) was then grafted onto the hydrophilic poly(N-methyldietheneamine sebacate) main chain through a quaternization reaction to obtain the cationic amphiphilic copolymer, P(MDS-co-CES). The main chain was a polyester, and the pendant chain contained potentially hydrolytically labile urethano groups, rendering this copolymer degradable. The product was characterized by 1H nuclear magnetic resonance (NMR) and infrared (IR) spectroscopies. Based on 1HNMR studies, the degree of cholesterol grafting was determined to be about 30%. P(MDS-co-CES) had a weight average molecular weight (Mw) of about 9.1 kDa with a polydispersity of 2.0, as measured by GPC. Like other amphiphilic copolymers [17], P(MDS-co-CES) also formed micelles in aqueous solutions. Its critical association concentration (CAC) in de-ionized water was determined to be about 1.9 mg/l by fluorescence spectroscopy using pyrene as a probe. The nitrogen content of the polymer was measured to be about 4.3% by an elemental analyzer. Mw of PMDS was about 18.5 kDa. The Mw of P(MDS-co-CES) was lower than that of PMDS, from which the P(MDS-co-CES) was synthesized, indicating that the grafting reaction at the high temperature caused the degradation of the main chain. 3.2. Synthesis and characterization of PEGylated P(MDSco-CES) The introduction of PEG onto the cationic polymer is aimed to prevent protein adsorption and increase the stability of micelle/DNA complexes in the blood. PMDS was first PEGylated by adding PEG to the reaction monomers of PMDS (Scheme 1). The successful synthesis of PEGylated PMDS was verified by 1HNMR spectra (see Fig. S1 in Supplementary Information). The peaks at d 2.71–2.73 (signal a), d 1.62 (signal b) and d 1.32 (signals c and d) were attributed to the protons of four different –CH2– groups from the sebacate units. The peaks at d 4.17–4.19 (signal e) and d 2.30–2.37 (signals f and g) were

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368

5362

O

CH3

O

Cl-C (CH2) 8 C -Cl sebacoyl chloride

N

(CH 2)2

+ HO

triethylamine (CH 2)2

N-methyldiethanolamine O

CH3O (CH2CH2O) n

(CH2CH2O) n H

OH + CH3O

methoxy PEG (mPEG)

CH 3

O O (CH 2)2

C (CH2) C 8

(CH 2)2

N

O m

PEGylated poly(N-methyldietheneamine sebacate)(PEG-PM DS) H3C

H3C

CH3

H3C

H

H

Br

H

CH3

H

O

triethylamine

O

Cl

H3C

CH3

H

H3C

BrCH 2CH 2NH2 + bromoethylamine

CH3

H3C

H

O

NH

O

N-(2-bromoethyl)carbarmoyl cholest erol (Be-chol)

cholest eryl chloroformat e

CH3 H3 C CH3

H3 C H H3 C H H

O

PEG-PMDS + Be-chol

O NH

CH 3O

(CH 2CH 2O)n

O

O

C

(CH 2) 8 C

CH 3 O (CH 2)2

N (CH2)2

O

O C (CH2)

O

8

C

O (CH2)2

q

Br + N (CH2)2 CH3

O

p

PEGylated poly(N-methyldietheneamine sebacate)-co-((chloesteryl oxocarbonylamido ethyl) methyl bis(ethylene) ammonium bromide) sebacate)(PEG-P(MDS-co-CES)) Scheme 1. PEGylation of PMDS and P(MDS-co-CES).

due to protons of two different –CH2– groups and the –CH3 group linked to the nitrogen atom. The peak at d 3.65–3.7 (signal h) was from –O–CH2– in PEG. An increased Mw of PEG led to an increased relative intensity of the peak when compared to signal b, indicating an increase in the relative length of PEG in the polymer. PEGylated PMDS was then reacted with Be-chol to form PEGylated P(MDS-co-CES) (Scheme 1). Signals a–h appeared in the 1HNMR spectra of PEG550, PEG1100 and PEG2000 conjugated P(MDS-co-CES) at similar positions when compared to those in the 1HNMR spectra of PEG-conjugated PMDS (see Figs. S1 and S2 in Supplementary Information). Besides, the characteristic peaks of cholesterol groups were observed at d 0.7–1.2. The single peak at d 0.69 (signal i) was attributed to the methyl group linked to the cyclic hydrocarbon. These results prove that cholesterol grafting and PEG conjugation were successful. The weight ratio of PEG to PMDS was calculated based on the integrated intensity of signal h from PEG and signal b from the sebacate units of PMDS, which is listed in Table 1. From Mw PEG:Mw PEG–PMDS

data, it can be seen that for PEG550 and PEG2000, PEG was not capped onto every polymer chain but for PEG1100, each polymer chain contained about one PEG molecule. Mw of PMDS was about 18.5 kDa. After PEGylation, Mw of the polymer decreased especially when PEG2000 was used (Table 1). This is because PEG acted as a chain ending agent. On the other hand, based on the peak areas of signals a and i, cholesterol grafting degree was calculated (Table 1). The grafting degree of cholesterol in PEGylated PMDS was much lower than that in P(MDSco-CES), indicating that the presence of PEG affected the quaternization reaction of Be-chol with the tertiary amines on the PMDS main chain. The conjugation of PEG onto PMDS changed the hydrophobicity of the polymer, reducing the compatibility of the polymer with the reaction solvent toluene. This might inhibit full stretching of the polymer chain in the solvent, resulting in steric hindrance for the quaternization reaction. After PEGylation, cholesterol grafting degree was difficult to control. Although the feed molar ratio of Be-chol to PMDS was kept constant, the final grafting degree of cholesterol was not the same for

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368

5363

Table 1 Molecular weight, cholesterol-grafting degree and PEG content of the polymers Polymer

Mwa

Mna

Mw/Mna

Mw PEG:Mw PEG–PMDS or Mw PEG:Mw PEG–P(MDS-co-CES) (%)b

Grafting degree (%)b

P(MDS-co-CES) PEG550–PMDS PEG1100–PMDS PEG2000–PMDS PEG550–P(MDS-co-CES) PEG1100–P(MDS-co-CES) PEG2000–P(MDS-co-CES) A PEG2000–P(MDS-co-CES) B

9100 6000 9500 6400 4732 8074 3900 5100

4600 3400 3800 3500 3500 4761 2600 3300

2.0 1.8 2.5 1.8 1.4 1.7 1.5 1.5

– 6.2 13.0 20.0 1.8 7.5 7.6 8.2

30 – – – 12.2 4.3 16.4 8.6

a

Obtained based on GPC analysis. Calculated based on 1HNMR spectra.

b

Table 2 Particle size and zeta potential of PEGylated PMDS and PEGylated P(MDS-co-CES) nanoparticles Polymer

Particle size (nm)

Zeta potential

P(MDS-co-CES) PMDS PEG550–PMDS PEG1100–PMDS PEG2000–PMDS PEG550–P(MDS-co-CES) PEG1100–P(MDS-co-CES) PEG2000–P(MDS-co-CES) A PEG2000–P(MDS-co-CES) B

8273 800710 50975 832712 58675 14274 7171 19471 16172

8475 0 0 1171 774 3673 2972 771 2172

PEGylated P(MDS-co-CES) with different PEG length. Further studies will be conducted to optimize reaction conditions so that cholesterol grafting degree can be readily controlled. 3.3. Self-assembly of PEGylated PMDS and PEGylated P(MDS-co-CES) P(MDS-co-CES) was easily self-assembled into core– shell structured micelles having an average size of 82 nm and zeta potential of 84 mV in the sodium acetate buffer (Table 2). Similarly, PEGylated polymers were also dialyzed against the same buffer to form nanoparticles. The size of PEGylated P(MDS-co-CES) micelles ranged from 71 to 194 nm, depending on PEG molecular weight and cholesterol grafting degree (Table 2). PEG2000– P(MDS-co-CES) A and B had a similar cholesterol grafting degree but yielded bigger micelles as compared to PEG550–P(MDS-co-CES) due to a thicker shell formed by PEG2000. In addition, a higher cholesterol grafting degree provided stronger driving force for micelle formation, resulting in greater aggregates of polymer molecules and thus larger particles [PEG2000–P(MDS-co-CES) A versus PEG2000–P(MDS-co-CES) B]. Furthermore, the size of PEGylated PMDS nanoparticles was the largest,

ranging from 509 to 832 nm. During the self-assembly process of PEGylated PMDS, PMDS block acted as a hydrophobic unit to form the core of micelles. However, at pH 4.6, a portion of tertiary amines in PMDS block was protonated, decreasing the hydrophobicity of PMDS. Therefore, the driving force for micelle formation was weakened, and more PMDS units were needed to form each micelle, leading to larger particle size. These findings indicate that the presence of cholesterol group in PEGylated P(MDS-co-CES) was essential for the formation of small and stable nanoparticles. The zeta potential of PEGylated PMDS nanoparticles could hardly be measured because PMDS, where the positive charges came from, was assembled into the core of the nanoparticles (Table 2). The zeta potential of PEGylated P(MDS-co-CES) micelles was also significantly reduced when compared to P(MDS-co-CES) micelles due to the shielding effect of PEG. 3.4. Protein adsorption of polymer and PEGylated polymer micelles The incorporation of PEG550 in the micelles markedly prevented the protein adsorption, with more than 90% of the proteins retained in the solution even after 3 days

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368

5364

110

0 Day 1 Day 2 Day 3

Percentage Protein (%)

100

1

2

4

6

8

10

Blank

A

90 0

80

1

2

3

4

6

8 10

12

14 16 18 Blank

B

70 60 50 1

2

3

4

5

6

7

Fig. 1. Percentage proteins remained in the blank solution (7) and the solution incubated with PEG550–P(MDS-co-CES) micelles (6), PEG550–P(MDS-co-CES) micelle/DNA complexes at N/P ratio of 15 (5), PEG2000–P(MDS-co-CES) B micelles (4), PEG2000–P(MDS-co-CES) B micelle/DNA complexes at N/P ratio of 15 (3), P(MDS-co-CES) micelles (2) or P(MDS-co-CES) micelle/DNA complexes at N/P ratio of 15 (1) as a function of time.

(Fig. 1). The protein content of the control sample was reduced to about 98%, while the micelles without PEG retained only 74%. Complexation with DNA further decreased the amount of protein adsorbed. No significant protein adsorption was observed on PEG2000–P(MDS-coCES) B micelles due to efficient shielding effect of PEG2000. These findings prove that PEGylation inhibited protein adsorption.

3.5. Particle size and zeta potential of micelle/DNA complexes The particle size of micelle/DNA complexes increased with increasing N/P ratio (see Fig. S3 in Supplementary Information). After reaching a maximum, it gradually decreased to 184, 330, 437 and 310 nm at N/P ratio of 25 for PEG550–P(MDS-co-CES), PEG1100–P(MDS-coCES), PEG2000–P(MDS-co-CES) A and PEG2000– P(MDS-co-CES) B, respectively. The zeta potential of the complexes also increased with N/P ratio, and reached a constant level at N/P ratio of 10, 5 and 15 for PEG550–P(MDS-co-CES), PEG1100–P(MDS-co-CES) and PEG2000–P(MDS-co-CES) B, respectively (see Fig. S4 in Supplementary Information). It did not reach a constant level for PEG2000–P(MDS-co-CES) A micelle/ DNA complexes in the N/P ratio range between 1 and 25. These findings suggest that PEG550–P(MDS-co-CES) and PEG1100–P(MDS-co-CES) condensed DNA more efficiently than PEG2000–P(MDS-co-CES) A and PEG2000– P(MDS-co-CES) B, and PEG2000–P(MDS-co-CES) B possessed a relatively greater DNA binding ability than PEG2000–P(MDS-co-CES) A.

C

D

Fig. 2. Electrophoretic mobility of plasmid DNA in micelle/DNA complexes loaded with (A) PMDS, (B) PEG550–PMDS, (C) PEG1100–PMDS and (D) PEG2000–PMDS at N/P ratios specified (1–18). Lane 1: naked DNA; last lane: blank micelles.

3.6. DNA binding ability of PMDS, PEGylated PMDS, P(MDS-co-CES) and PEGylated P(MDS-co-CES) micelles As shown in Fig. 2, PMDS and PEGylated PMDS nanoparticles provided very weak DNA binding ability, and complete retardation of DNA was not observed even at N/P ratio of 10 and 18, respectively. As discussed in Section 3.3, PEG–PMDS polymers did not form stable core–shell particles. The PMDS block carrying positive charge acted as the hydrophobic block to form the cores, resulting in lower surface charge and thus weaker DNA binding ability. In addition, PMDS was unable to form well-defined core–shell structure, and thus the zeta potential of its nanoparticles was neutral (i.e. 0 mV). In sharp contrast, P(MDS-co-CES) nanoparticles exhibited strong DNA binding ability (Fig. 3), and complete retardation of DNA was observed at N/P ratio of 2. This is because the presence of cholesterol ensured the formation of small and stable particles having a high zeta potential. Similarly, PEG550–P(MDS-co-CES), PEG1100– P(MDS-co-CES) and PEG2000–P(MDS-co-CES) B also showed strong DNA binding ability, and DNA mobility was completely inhibited at N/P ratio of 2, 1 and 2, respectively (Fig. 3). Interestingly, PEG2000–P(MDS-co-CES)

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368

A

0

0

1

1

2

2

3

3

4

6

4

8

10

6

12

8

14

Blank

16 18 Blank

B

C

D

E

Fig. 3. Electrophoretic mobility of plasmid DNA in micelle/DNA complexes loaded with (A) P(MDS-co-CES), (B) PEG550–P(MDS-coCES), (C) PEG1100–P(MDS-co-CES), (D) PEG2000–P(MDS-co-CES) B or (E) PEG2000–P(MDS-co-CES) A at N/P ratios specified (1–18). Lane 1: naked DNA; last lane: blank micelles.

A with a higher cholesterol grafting degree (16.4%) exhibited much lower DNA binding ability when compared to PEG2000–P(MDS-co-CES) B (8.6%). Unlike conventional PEGylated cationic polymers, PEGylated P(MDSco-CES) formed a well-defined core–shell structure and the positively charged surface was shielded with a PEG layer, preventing DNA binding. PEG550–P(MDS-co-CES) had a short PEG chain. Therefore, the positively charged surface might be still accessible to DNA binding, resulting in good DNA binding ability. PEG1100–P(MDS-co-CES) had the lowest grafting degree (4.3%) of cholesterol among all the PEGylated polymers, forming the smallest micelles (71 nm). On the other hand, the core of the PEG1100– P(MDS-co-CES) micelles might be loosely packed owing to the low cholesterol grafting degree. Therefore, the positively charged surfaces might be accessible to DNA molecules. These two factors are the possible reasons for their strong DNA binding ability. It is also observed that although PEG2000–P(MDS-co-CES) B had a much longer PEG chain than PEG550–P(MDS-co-CES), they had similar DNA binding ability. This might be because PEG2000–P(MDS-co-CES) B had a lower cholesterol grafting degree (12.2% versus 8.6%), forming a more loosely packed core so that the positively charged surface was easily accessible to DNA molecules. In sharp contrast, PEG2000–P(MDS-co-CES) A had a high cholesterol grafting degree and a long PEG chain, forming large

5365

particles with a stable core–shell structure. The positively charged surface might be well-surrounded by a PEG layer, leading to very weak DNA binding ability. 3.7. Cytotoxicity of P(MDS-co-CES) and PEGylated P(MDS-co-CES) micelles Cytotoxicity is an important parameter of cationic polymers as non-viral vectors. It was hypothesized that cytotoxicity was related to gene transfection efficiency [18], and might be caused by electrostatic interactions with negatively charged glycocalyx of the cell surface [19]. As shown in Fig. 4, the cytotoxicity of P(MDS-co-CES) was much lower than that of PEI against HEK293, HepG2, HeLa, MDA-MB-231 and 4T1 cells. In addition, PEGylation significantly reduced the cytotoxicity of the polymers due to the reduced surface charge of the micelles (Table 2). 3.8. In vitro gene transfection As shown in Fig. 5, P(MDS-co-CES) micelle/DNA complexes induced high gene transfection efficiency especially in MDA-MB-231 and 4T1 cell lines, which was higher than that provided by PEI. This may be because the positively charged surface of P(MDS-co-CES) micelle/ DNA complexes with a higher zeta potential (around 70 mV at N/P ratio of 15) induced greater cellular uptake. In sharp contrast, the main chain PMDS nanoparticle/ DNA complexes gave very low gene transfection efficiency even at N/P ratio of 25 because their large size and neutral surface charge resulted in weak DNA binding and low cellular uptake (Fig. 6). After PEGylation of P(MDS-coCES), gene transfection efficiency decreased significantly. In particular, PEG2000–P(MDS-co-CES) B micelle/DNA complexes yielded very low gene expressions in all the five cell lines. However, PEG550–P(MDS-co-CES) nanoparticles induced high gene expression level, comparable to that provided by P(MDS-co-CES) nanoparticles in HepG2, HeLa, MDA-MB-231 (at N/P ratio of 30, data not shown) and 4T1 cells but slightly lower in HEK293 cells. In addition, it increased with increasing N/P ratio, and did not reach the maximum level up to N/P ratio of 25, indicating that a higher N/P ratio was needed to increase the surface charge of the complexes and reduce the particle size in order to improve cellular uptake. PEG1100– P(MDS-co-CES) and PEG2000–P(MDS-co-CES) nanoparticles followed the same trend. Interestingly, PEG1100–P(MDS-co-CES) and PEG2000– P(MDS-co-CES) B micelles having stronger DNA binding ability and smaller size exhibited much lower gene expression efficiency when compared to PEG550– P(MDS-co-CES) and PEG2000–P(MDS-co-CES) A micelles, respectively (Figs. 5 and 7). PEG550–P(MDSco-CES) and PEG2000–P(MDS-co-CES) A carried higher amount of cholesterol, forming a more stable core–shell structure, which improved the stability of the micelle/DNA complexes. Another possible reason is that the cellular

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368

5366

120 PEI Polymer Polymer-PEG550 Polymer-PEG2000

Cell Viability (%)

100 80 60 40 20 0 0

50 100 150 200 250 Polymer Concentration (ppm)

C

B 140

160 Cell Viability (%)

Cell Viability (%)

120 100 80 60 40 20

140 120 100 80 60 40 20 0

0 0

50 100 150 200 Polymer Concentration (ppm)

250

D

0

50 100 150 200 Polymer Concentration (ppm)

250

0

50 100 150 200 Polymer Concentration (ppm)

250

E 140

120

120

100

100

Cell Viability (%)

Cell Viability (%)

180

80 60 40

80 60 40 20

20 0

0 0

50 100 150 200 Polymer Concentration (ppm)

250

Fig. 4. Cytotoxicity of PEI, P(MDS-co-CES) and PEGylated P(MDS-co-CES) (PEG Mn 550 or 2000) micelles fabricated in the sodium acetate buffer by a membrane dialysis method against HEK293 (A), HepG2 (B), HeLa (C), MDA-MB-231 (D) and 4T1 (E) cells.

uptake of PEG550–P(MDS-co-CES) and PEG2000–P (MDS-co-CES) A micelle/DNA complexes having a greater amount of cholesterol might be higher than that of PEG1100–P(MDS-co-CES) and PEG2000–P(MDS-coCES) B micelle/DNA complexes based on an enhanced cellular cholesterol uptake pathway [15]. GFP transfection experiments were performed in HEK293 cells to study cellular uptake of DNA complexes induced by PEGylated micelles. Compared to the data reported previously [16], PEGylation significantly reduced the percentage of cells transfected with GFP. It decreased with increasing PEG length (see Fig. S5 in Supplementary Information). In addition, compared to PEG2000–P

(MDS-co-CES) B micelles, PEG2000–P(MDS-co-CES) A micelles mediated higher GFP transfection efficiency due to their more stable core/shell structure and higher amount of cholesterol. 4. Conclusions PEGylation of an efficient gene delivery carrier, cationic P(MDS-co-CES) micelles has been attempted to improve the stability of micelle/DNA complexes. PEGylated P(MDS-co-CES) micelles were much less cytotoxic but they provided lower gene transfection efficiency. However, the chain length of PEG, cholesterol grafting degree and N/P

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368

A

1010

109

Luciferase Expression Level (RLU/mg protein)

Luciferase Expression Level (RLU/mg protein)

1011

107 105 103 10

B

102 1

C

107 105

104 102

109

107 106 105 104 103 102 10 1

C

103

109

Luciferase Expression Level (RLU/mg protein)

Luciferase Expression Level (RLU/mg protein)

106

108

104

10

D

107 105 103

109 108 107 106 105 104 103 102 10 1

10 1010

Luciferase Expression Level (RLU/mg protein)

108

1

106

109 Luciferase Expression Level (RLU/mg protein)

B

108

Luciferase Expression Level (RLU/mg protein)

Luciferase Expression Level (RLU/mg protein)

1010

5367

A

E

1

5

15 10 N/P Ratio

20

25

30

Fig. 6. Comparison in luciferase transfection efficiency between PMDS and P(MDS-co-CES) (A, N/P: 15) micelles. (A) HEK293; (B) HepG2; (C) HeLa.

108 106 104 102 1 5

10

15 N/P Ratio

20

25

Fig. 5. Luciferase expression level in HEK293 (A), HepG2 (B), HeLa (C), MDA-MB-231 (D) and 4T1 (E) cells transfected with ( ) PEI, ( ) P(MDS-co-CES), ( ) PEG550–P(MDS-co-CES), ( ) PEG1100–P(MDSco-CES) and ( ) PEG2000–P(MDS-co-CES) B micelles.

ratio could be modulated to achieve high gene transfection efficiency. PMDS and PEGylated PMDS particles without cholesterol gave weaker DNA binding ability and thus

lower gene transfection efficiency when compared to welldefined core–shell structured P(MDS-co-CES) and PEGylated P(MDS-co-CES) micelles. However, strong DNA binding ability did not always lead to high gene transfection efficiency. The stability of micelles self-assembled from PEGylated P(MDS-co-CES) played a crucial role in gene transfection. High gene transfection efficiency was achieved by using micelles with a more stable core–shell structure, which self-assembled from the polymers having a relatively high cholesterol grafting degree, although they possessed lower DNA binding ability and relatively larger particle size. PEG550–P(MDS-co-CES) micelles induced high gene transfection efficiency in various cell lines at high N/P

ARTICLE IN PRESS Y. Wang et al. / Biomaterials 28 (2007) 5358–5368

5368

(RLU/mg protein)

Luciferase Expression Level

1010

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.biomaterials. 2007.08.013.

108 106 104

References

102 1

(RLU/mg protein)

10 Luciferase Expression Level

Appendix A. Supplementary material

A

8

B

106 104 102 1

(RLU/mg protein)

Luciferase Expression Level

108

C

106 104 102 1

(RLU/mg protein)

Luciferase Expression Level

107

D

105

103

10 5

10

15 20 N/P Ratio

25

30

Fig. 7. Comparison in luciferase transfection efficiency between PEG2000–P(MDS-co-CES) A ( ) and PEG2000–P(MDS-co-CES) B ( ) micelles with different cholesterol grafting degrees (A: 16.4%, B: 8.6%). (A) HEK293; (B) HepG2; (C) HeLa; (D) 4T1.

ratios, which may provide a promising carrier for systemic in vivo gene delivery. Acknowledgments This work was financially supported by Institute of Bioengineering and Nanotechnology, Agency for Science, Technology and Research, Singapore.

[1] Fraley R, Subramani S, Berg P, Papahadjopoulos D. Introduction of liposome-encapsulated SV40 DNA into cells. J Biol Chem 1980;255: 10431–5. [2] Luo D, Saltzman WM. Synthetic DNA delivery systems. Nat Biotechnol 2000;18:33–7. [3] Kakizawa Y, Furukawa S, Ishii A, Kataoka K. Organic–inorganic hybrid-nanocarrier of siRNA constructing through the self-assembly of calcium phosphate and PEG-based block aniomer. J Control Rel 2006;111:368–70. [4] Gemeinhart RA, Luo D, Saltzman WM. Cellular fate of a modular DNA delivery system mediated by silica nanoparticles. Biotechnol Prog 2005;21:532–7. [5] Salem KA, Searson PC, Leong KW. Multifunctional nanorods for gene delivery. Nat Mater 2003;2:668–71. [6] Singh R, Pantarotto D, McCarthy D, Chaloin O, Hoebeke J, Partidos CD, et al. Binding and condensation of plasmid DNA onto functionalized carbon nanotubes: toward the construction of nanotube-based gene delivery vectors. J Am Chem Soc 2005;127: 4388–96. [7] Godbey W, Wu K, Hirasaki G, Mikos A. Improved packing of poly(ethylenimine)/DNA complexes increases transfection efficiency. Gene Ther 1999;6:1380–8. [8] Neu M, Fischer D, Kissel T. Recent advances in rational gene transfer vector design based on poly(ethylene imine) and its derivatives. J Gene Med 2005;7:992–1009. [9] Jeon E, Kim HD, Kim JS. Pluronic-grafted poly-(L)-lysine as a new synthetic gene carrier. J Biomed Mater Res A 2003;66:854–9. [10] Wood KC, Little SR, Langer R, Hammond PT. A family of hierarchically self-assembling linear-dendritic hybrid polymers for highly efficient targeted gene delivery Angew. Chem Int Ed 2005; 44:2–7. [11] Akinc A, Lynn DM, Anderson DG, Langer R. Parallel synthesis and biophysical characterization of a degradable polymer library for gene delivery. J Am Chem Soc 2003;125:5316–23. [12] Liu WG, Yao KD. Chitosan and its derivatives—a promising nonviral vector for gene transfection. J Control Rel 2002;83:1–11. [13] Jiang X, Dai H, Leong KW, Goh SH, Mao HQ, Yang YY. Chitosang-PEG/DNA complexes deliver gene to the rat liver via intrabiliary and intraportal infusions. J Gene Med 2006;8:477–87. [14] Hong JW, Park JH, Huh KM, Chung H, Kwon IC, Jeong SY. PEGylated polyethylenimine for in vivo local gene delivery based on lipiodolized emulsion system. J Control Rel 2004;99:167–76. [15] Han S, Mahato RI, Kim SW. Water-soluble lipopolymer for gene delivery. Bioconjug Chem 2001;12:337–45. [16] Wang Y, Gao SJ, Ye WH, Yoon HS, Yang YY. Co-delivery of drugs and DNA from cationic core–shell nanoparticles self-assembled from a biodegradable copolymer. Nat Mater 2006;5:791–6. [17] Coulembier O, Mespouille L, Hedrick JL, Waymouth RM, Dubois P. Metal-free catalyzed ring-opening polymerization of beta-lactones: synthesis of amphiphilic triblock copolymers based on poly(dimethylmalic acid). Macromolecules 2006;39:4001–8. [18] Merdan T, Kopecek J, Kissel T. Prospects for cationic polymers in gene and oligonucleotide therapy against cancer. Adv Drug Deliv Rev 2002;54:715–58. [19] Fischer D, Li Y, Ahlemeyer B, Krieglstein J, Kissel T. In vitro cytotoxicity testing of polycations: influence of polymer structure on cell viability and hemolysis. Biomaterials 2003;24:1121–31.