Dispersing Zwitterions into Comb Polymers for Nonviral Transfection

Jan 7, 2016 - Department of Materials Science and Engineering, University of Delaware, 201 DuPont Hall, Newark, Delaware 19716 United States...
2 downloads 0 Views 5MB Size
Download PDF
Article pubs.acs.org/Biomac

Dispersing Zwitterions into Comb Polymers for Nonviral Transfection: Experiments and Molecular Simulation Ahmadreza F. Ghobadi,†,‡ Rachel Letteri,‡,§ Sangram S. Parelkar,§ Yue Zhao,∥ Delphine Chan-Seng,⊥ Todd Emrick,*,§ and Arthi Jayaraman*,†,# †

Department of Chemical and Biomolecular Engineering, University of Delaware, 150 Academy Street, Newark, Delaware 19716 United States § Department of Polymer Science and Engineering, University of Massachusetts, 120 Governors Drive, Amherst, Massachusetts 01003, United States ∥ Quantum Beam Science Center, Japan Atomic Energy Agency, Tokai, Ibaraki 319-1195, Japan ⊥ Institut Charles Sadron UPR22-CNRS, 23 rue du Loess, 67034 Strasbourg, France # Department of Materials Science and Engineering, University of Delaware, 201 DuPont Hall, Newark, Delaware 19716 United States S Supporting Information *

ABSTRACT: Polymer-based gene delivery vehicles benefit from the presence of hydrophilic groups that mitigate the inherent toxicity of polycations and that provide tunable polymer-DNA binding strength and stable complexes (polyplexes). However, hydrophilic groups screen charge, and as such can reduce cell uptake and transfection efficiency. We report the effect of embedding zwitterionic sulfobetaine (SB) groups in cationic comb polymers, using a combination of experiments and molecular simulations. Ring-opening metathesis polymerization (ROMP) produced comb polymers with tetralysine (K4) and SB pendent groups. Dynamic light scattering, zeta potential measurements, and fluorescence-based experiments, together with coarsegrained molecular dynamics simulations, described the effect of SB groups on the size, shape, surface charge, composition, and DNA binding strength of polyplexes formed using these comb polymers. Experiments and simulations showed that increasing SB composition in the comb polymers decreased polymer-DNA binding strength, while simulations indicated that the SB groups distributed throughout the polyplex. This allows polyplexes to maintain a positive surface charge and provide high levels of gene expression in live cells. Notably, comb polymers with nearly 50 mol % SB form polyplexes that exhibit positive surface charge similarly as polyplexes formed from purely cationic comb polymers, indicating the ability to introduce an appreciable amount of SB functionality without screening surface charge. This integrated simulation-experimental study demonstrates the effectiveness of incorporating zwitterions in polyplexes, while guiding the design of new and effective gene delivery vectors.

1. INTRODUCTION Gene therapy involves the delivery of genetic materials, such as DNA or small interfering RNA (siRNA), into target cells to modify the protein expression profile.1−4 DNA carriers should provide protection from enzymatic degradation and facilitate DNA uptake into cells, nuclear translocation, and release. The efficient transfection offered by viral vectors is countered by safety complications5−7 that drive the development of synthetic systems involving lipids and polymeric DNA carriers.8−15 However, while cationic lipids such as lipofectamine afford efficient transfection, their toxicity precludes translation to in vivo studies.16−18 Polymers represent potentially appealing alternatives to viral vectors and cationic lipids due to their large DNA carrying capacity and structural tunability. Polymer-based transfection reagents have advanced significantly in the past decade through innovations in polymer chemistry and architecture.18−22 We previously showed, through both experiment and simulation, © 2016 American Chemical Society

that comb polymers having oligolysine groups pendent to a hydrophobic backbone markedly improve transfection performance relative to linear poly(lysine).23−26 Other groups have recognized favorable attributes of such architectures, including oligopeptide-grafted polymeric methacrylates27,28 and methacrylamides,29−31 as well as polymer-grafted polylysine32−34 and dextran.33,35 Despite these achievements, further improvement of transfection performance must be achieved in polymer-based systems to rival the gene expression level of viruses. Augmenting polycations with neutral hydrophilic groups, like poly(ethylene glycol) (PEG), is intended to mitigate cation toxicity, while providing polyplex (polymer−DNA complex) stability and tailored polymer-DNA binding strength.29,36−38 Unfortunately, PEG-modified polycations suffer from reduced Received: October 30, 2015 Revised: December 18, 2015 Published: January 7, 2016 546

DOI: 10.1021/acs.biomac.5b01462 Biomacromolecules 2016, 17, 546−557

Article

Biomacromolecules

from VWR. Polystyrene sterile culture test tubes (12 × 75 mm with a polyethylene cap) were purchased from Fisher Scientific. Label IT Fluorescein plasmid DNA was purchased from Mirus Bio. 2.2. Polymer Characterization. 1H and 13C NMR spectra were recorded on a Bruker Spectrospin DPX300, Avance400, or 500 spectrometer. Mass spectrometry was performed on a JEOL MStation JMS700 high-resolution two-sector mass spectrometer equipped with fast atom bombardment (FAB) and low-resolution electrospray ionization (ESI). Size exclusion chromatography (SEC) on Bocprotected polymers was performed in N,N-dimethylformamide (DMF) with 0.01 M LiCl at 50 °C. The system was operated at a flow rate of 1 mL/min with a Sonntek HPLC pump (K-501), one 50 mm × 7.5 mm PL gel mixed guard column, one 300 mm × 7.5 mm PL gel 5 μm mixed C column, one 300 mm × 7.5 mm PL gel 5 μm mixed D column, a Knauer refractive index detector (K-2301), and an Alltech solvent recycler 3000. SEC on deprotected cationic polymers was performed in TFE with 0.02 M sodium trifluoroacetate at 40 °C using an Agilent 1200 system equipped with an isocratic pump operated at 1 mL/min, a degasser, an autosampler, one 50 mm × 8 mm PSS PFG guard column (Polymer Standards Service), three 300 mm × 7.5 mm PSS PFG analytical linear M columns with 7 μm particle size (Polymer Standards Service), and an Agilent 1200 refractive index detector. Both systems were calibrated with PMMA standards. 2.3. Synthesis of Dimethylaminoethyl Cyclooctene. (Z)Cyclooct-4-ene-1-carboxylic acid (2.00 g, 12.9 mmol) was dissolved in CH2Cl2 (60 mL). EDC (2.74 g, 14.3 mmol) was added, followed by N,N-dimethylaminoethylamine (1.56 mL, 1.26 g, 14.3 mmol), and the mixture was stirred for 24 h at room temperature under N2(g). The mixture was washed with brine, 5% NaHCO3(aq) (2×), and again brine. The combined organic layers were dried over MgSO4, filtered, and concentrated by rotary evaporation, and then dried under vacuum to yield a yellow oil (2.6 g, 89%). 2.4. Synthesis of Sulfobetaine Cyclooctene (SB-COE). Dimethylaminoethyl cyclooctene (2.55 g, 11.4 mmol) was dissolved in CH3CN (10 mL) in a vial equipped with a stir bar and a gas inlet adapter. 1,3-Propane sultone (1.67 g, 13.7 mmol) was added, and the mixture heated to 50 °C and stirred under N2. After 10 min, a white precipitate was observed. The mixture was stirred approximately 18 h at 50 °C, and the product isolated by centrifugation (4000 rpm, 5 min) as a white powder (1.91 g, 48%). 1H NMR (300 MHz, D2O, δ): 5.54− 5.82 (br m, 2H), 3.60 (t, J = 7 MHz, 2H), 3.37−3.54 (br m, 4H), 3.11 (s, 6H), 2.93 (t, J = 7 MHz, 2H), 2.66−2.44 (br m, 2H), 1.97−2.26 (br m, 4H), 1.42−1.78 (br m, 4H), 1.32−1.42 (br m, 1H); 13C NMR (100 MHz, D2O, δ): 182.01 (CO), 131.03, 129.74, 71.41, 67.73, 61.71, 50.86, 50.54, 47.06, 44.53, 44.31, 32.89, 31.74, 29.53, 27.20, 25.08, 23.40, 18.08. FAB-mass spectrometry: calcd [M + H]+, 347.2005; found, 347.2003. 2.5. Polymerization of 5-Tetralysine(boc)-1-cyclooctene (K4COE). K4-COE (0.20 g, 0.19 mmol) was dissolved in 2,2,2trifluoroethanol (TFE, 0.4 mL) in a scintillation vial equipped with a stir bar and a septum. The solution was subjected to three freeze− pump−thaw cycles. A solution of the 3-bromopyridine-substituted Grubbs’ catalyst58 in CH2Cl2 was prepared (50 mg/mL) and degassed. Catalyst solution (0.003 g, 0.004 mmol, 0.06 mL) was added, and the mixture was stirred for 2 h at room temperature under N2. Ethyl vinyl ether (0.10 mL, 0.075 g, 1.0 mmol) was added, the mixture stirred ∼1 h, then opened to air. The mixture was diluted with approximately 1 mL of TFE and precipitated into 40 mL of diethyl ether. The precipitate was isolated by centrifugation (4000 rpm, 5 min) and dried under vacuum to afford a yellow-brown solid (81% yield). 1H NMR (300 MHz, DMSO-d6, δ): 12.53 (br, 1H), 7.60−8.27 (br m, 4H), 6.70 (br, 3H), 6.39 (br, 1H), 5.30 (br, 2H), 4.07−4.47 (br m, 4H), 2.87 (br, 8H), 0.98−2.35 (br m, 71H). SEC (0.01 M LiCl DMF, PMMA standards): Mn 99200 g/mol, Mw/Mn 1.61. 2.6. Representative Copolymerization of K4-COE and SBCOE. K4-COE (0.20 g, 0.19 mmol) and SB-COE (0.053 g, 0.15 mmol) were dissolved in TFE (0.4 mL) in a scintillation vial equipped with a stir bar and a septum. The monomer solution was subjected to three freeze−pump−thaw cycles. Separately, a solution of 3bromopyridine-substituted Grubbs’ catalyst58 was prepared in

cellular uptake and low transfection efficiency.13,36,39,40 As an alternative, zwitterions offer a hydrophilic component to polymer-based gene delivery vehicles,34,37,41−50 through integration of sulfobetaine and phosphorylcholine into polycations as a neutral hydrophilic block connected to a cationic block,41−45,49,50 as grafts,34,46 as randomly dispersed units in linear polycations,43 and through conjugation to branched polyethylene imine (PEI).47,48 Addition of neutral hydrophilic components to polycations in a diblock sequence gives polyplexes with enhanced colloidal stability, but unfortunately with reduced transfection performance.41,43,51 On the other hand, incorporation of neutral hydrophilic groups on the periphery of branched polymers,47,48,52 and particles50 yields polyplexes with more favorable gene expression, likely due to the accessibility of the cell membrane to the cationic groups (which are shielded in diblock structures). Neutral hydrophilic moieties that enhance DNA release improve transfection in some cases, but in other cases result in poor DNA complexation.24,42,53,54 Thus, it is essential to assess systematically how incorporation of hydrophilic components impacts the transfection process. Here, we exploit the capabilities of our comb polymer platform24−26 to examine the effects of dispersing zwitterionic components into comb polymers. Specifically, copolymers having cationic tetralysine (K4) oligopeptides and zwitterionic sulfobetaine (SB) groups pendent to a polycyclooctene backbone were synthesized by ring-opening metathesis polymerization (ROMP) and used for polyplex formation. Polymer-DNA binding strength as well as polyplex size, surface charge, and surface composition were assessed experimentally and computationally to understand the impact of dispersing zwitterions throughout the comb polymer structures and within the resulting polyplexes.

2. MATERIALS AND EXPERIMENTS 2.1. Materials. Acetonitrile (anhydrous, 99.8%), methanol (anhydrous, 99.8%), 1,4-dioxane (anhydrous, 99.8%), ethyl vinyl ether (99%), N,N-dimethylethylenediamine (≥98.0%), and 1,3propane sultone (98%) were purchased from Aldrich. 2,2,2Trifluoroethanol (TFE, ≥99%) and sodium hydrogen carbonate (ACS, 99.7−100.3%) were purchased from Alfa Aesar. 1-(3(Dimethylamino)propyl)-3-ethylcarbodiimide hydrochloride (EDC, >98.0%) was purchased from TCI America. Methylene chloride (Optima), hexanes (certified ACS), diethyl ether (stabilized by butylated hydroxytoluene (BHT), certified ACS), acetic acid (glacial, certified), hydrochloric acid (certified ACS plus), and dialysis tubing (Spectra/Por, regenerated cellulose membranes with a molecular weight cutoff of 6000−8000 g/mol) were purchased from Fisher Scientific. (Z)-Cyclooct-4-ene-1-carboxylic acid was prepared from 1,5-cyclooctadiene as previously reported.55−57 5-[K(Boc)]4-1-cyclooctene (K4-COE) was prepared by solid phase peptide synthesis.24 3Bromopyridine-substituted Grubbs Generation III catalyst was prepared according to the literature.58 CH2Cl2 was distilled over CaH2. All other materials were used without purification. Dulbecco’s Modified Eagles Medium (DMEM), penicillin and streptomycin, lipofectamine 2000, phosphate-buffered saline (PBS, 1×, pH 7.4), and the Quant-iT PicoGreen (200×) double-stranded DNA reagent were purchased from Life Technologies. Bovine serum albumin (BSA) was purchased from Aldrich. Fetal bovine serum (FBS) was purchased from Atlanta Biologicals. SKOV3 cells were purchased from American Type Culture Collection (ATCC), while mRFP-IRES-Puro plasmid DNA was a gift from Dr. Lawrence Schwartz. Heparin (ammonium salt from porcine intestinal mucosa) was purchased from Aldrich. The CellTiter-Glo Luminescent Cell Viability Assay was purchased from Promega. Sterile syringe filters with a 0.45 μm poly(ether sulfone) (PES) membrane were purchased 547

DOI: 10.1021/acs.biomac.5b01462 Biomacromolecules 2016, 17, 546−557

Article

Biomacromolecules CH2Cl2 (50 mg/mL) and degassed. The catalyst solution (0.003 g, 0.004 mmol, 0.06 mL) was added to the monomer, and the mixture was stirred for 2 h at room temperature under N2. Ethyl vinyl ether (0.10 mL, 0.075 g, 1.0 mmol) was added, and the mixture was stirred 30 min and opened to air. The mixture was diluted with TFE (1 mL) and precipitated into diethyl ether (40 mL). The precipitate was isolated by centrifugation (4000 rpm, 5 min) and dried under vacuum to afford a yellow-brown solid (89−95% yield). Specific polymerization results are given in the Supporting Information, Table S1. 1H NMR (500 MHz, DMSO-d6, δ): 12.52 (br, 1H from K4), 7.60−8.43 (br m, 4H from K4), 6.71 (br, 3H from K4), 6.38 (br, 1H from K4), 5.32 (br, 2H from K4 and 2H from SB), 4.07−4.41 (br m, 4H from K4), 3.23−3.67 (br m, 6H from SB), 3.08 (br, 6H from SB), 2.87 (br, 8H from K4 and 2H from SB), 1.01−2.38 (br m, 71H from K4 and 13H from SB). 2.7. Synthesis of Poly(5-tetralysine-cyclooctene) (K4). Poly[5tetralysine(boc)-1-COE] (0.1 g) was dissolved in methanol (5 mL) and 4 M HCl in dioxane (2 mL). The mixture was stirred for 3 h at room temperature, then precipitated into diethyl ether (40 mL). The polymer product was isolated by centrifugation (4000 rpm, 5 min), dialysis in water (6−8 kDa molecular weight cutoff membrane), and lyophilization to afford a white powder in 56% yield. 1H NMR (300 MHz, D2O, δ): 5.25−5.54 (br m, 2H), 4.17−4.45 (br m, 4H), 3.73 (s, 8H from residual dioxane), 2.87 (br, 8H), 2.24−2.50 (br, 1H), 1.01− 2.12 (br m, 34 H). SEC (0.02 M NaTFAc TFE, PMMA standards): Mn 57600 g/mol, Mw/Mn 1.49. 2.8. Synthesis of Poly(5-tetralysine-cyclooctene-co-sulfobetaine-cyclooctene) (K4SB-X). Poly[5-tetralysine(boc)-1-COE-coSB-COE] (0.1 g) was dissolved in methanol (5 mL) and 4 M HCl in dioxane (2 mL). The mixture was stirred for 3 h at room temperature, then precipitated into diethyl ether (40 mL), isolated by centrifugation (4000 rpm, 5 min), dialysis in water (6−8 kDa molecular weight cutoff membrane), and lyophilization to afford K4SB-X (where X denotes mol % SB) as a white powder in 51−66% yield. 1H NMR (500 MHz, D2O, δ): 5.39 (br, 2H from K4 and 2H from SB), 4.17−4.47 (br m, 4H from K4), 3.73 (s, 8H from residual dioxane), 3.38−3.86 (br m, 6H from SB), 3.17 (br, 6H from SB), 2.88−3.08 (br, 8H from K4 and 2H from SB), 2.16−2.47 (br, 1H from K4 and 3H from SB), 1.05−2.10 (br m, 34 H from K4 and 10H from SB); 13C NMR (125 MHz, D2O, δ): 179.09, 176.54, 173.75, 157.63, 130.76, 129.97, 80.10, 62.77, 61.59, 53.52, 52.83, 51.09, 47.26, 45.87, 39.86, 39.12, 32.91, 32.09, 30.67, 30.57, 30.20, 29.70, 28.12, 26.36, 24.80, 22.23, 22.07, 21.94, 18.27. 2.9. Preparation of Polymer Stock Solutions for Polyplex Formation. Polymer solutions were prepared such that mixing 1/1 v/ v with mRFP-IRES-Puro plasmid DNA solution (1 μg/50 μL) enabled complexation with DNA at various N/P ratios (i.e., the ratio of protonatable nitrogens, N, in the polymer to DNA phosphates, P). Polymer concentrations of N/P 200 solutions were calculated, as described in Supporting Information. The polymers were weighed into low-retention plastic tubes, dissolved in nuclease-free water, and sterilized by filtration through 0.45 μm poly(ether sulfone) (PES) filters. Polymer solution (N/P 10) was prepared by adding N/P 200 stock solution (50 μL) to water (950 μL). Polymer solution (N/P 5) was prepared by mixing N/P 10 stock solution with water in a 1/1 v/v ratio. 2.10. Polyplex Diameter and Zeta Potential Measurements. DNA solution (0.4 mL, 1 μg/50 μL) was added to the polymer solution (0.4 mL, N/P 5), vortexed lightly, and allowed to equilibrate for 40 min at room temperature to form polyplexes at N/P 5. After polyplex equilibration, water (0.1 mL) and NaCl(aq) (0.1 mL, 100 mM) were added to bring the salt concentration to 10 mM. Polyplex diameter and zeta potential measurements were acquired on a Malvern Zetasizer NanoZS. Polyplex diameters represent the mean intensityaverage size over three measurements, with at least 12 runs per measurement. 2.11. Atomic Force Microscopy (AFM) of K4SB-17 Polyplexes. Mica substrates were prepared by adhering mica to glass slides using fast dry nail enamel (Maybelline). K4SB-17 (0.2 mL, N/P 4) was added to DNA solution (0.2 mL, 1 μg/50 μL), vortexed lightly, and

allowed to equilibrate for 40 min at room temperature to form polyplexes at N/P 4. Polyplex solution (200−300 μL) was then dropcast onto freshly cleaved mica and left on the substrate for 5−10 min, then rinsed with 10 mL of water filtered through a 0.45 μm poly(ether sulfone) filter. Excess solution was removed by tilting the substrate onto filter paper. Filter paper was then held at the edges of the substrate to remove excess solution between the mica and glass slide. The substrates were dried gently with compressed air and imaged on a Digital Instrument Dimension 3100 atomic force microscope in tapping mode at 1 Hz. The imaging area was 5 μm × 5 μm, and linewise leveling was applied using SPIP 6.0.6 software for Windows. 2.12. PicoGreen Quantification of Heparin-Induced Decomplexation. DNA solution (20 μL, 0.2 μg/ μL) was added to the polymer solution (N/P 10, 100 μL) and water (80 μL). The mixture was vortexed lightly and allowed to equilibrate for 40 min at room temperature to form polyplexes at N/P 5. PicoGreen (10×, 40 μL), heparin (20 μL, 2 units/μL), and water (140 μL) were added to the polyplex solution such that the Picogreen concentration was 1×. The mixture was added to each of three wells (100 μL per well) in a 96well plate. After incubation of the polyplex mixture for 16 h at 37 °C, Picogreen binding with DNA was quantified using a plate reader in fluorescence mode (BMG Labtech FLUOstar OPTIMA plate reader). PicoGreen dye fluorescence intensity increases significantly upon binding double stranded DNA, allowing the detection and quantification (using fluorescence measured by a plate reader) of plasmid DNA that is either free from, or loosely associated with, the cationic polymer. As a control, naked plasmid DNA was incubated with 1× PicoGreen and heparin (20 units, 2 units/μL) for 16 h. The fluorescence intensity from 1× Picogreen solution containing heparin (20 μL, 2 units/μL) was subtracted as a baseline from all other fluorescence intensities. The average ratio of fluorescence readings for the polyplex and naked DNA gave the relative percentage of DNA accessible to PicoGreen reagent upon heparin incubation. The reported values of [100 − (PG fluorescence/PG fluorescence*)] × 100%, where PG fluorescence* is the Picogreen fluorescence in the absence of cationic polymer, reflecting the relative percentage of bound DNA, represent the average and standard deviation of three independent experiments. 2.13. Cell Culture, Polyplex Formation, and Transfection. Human ovarian cancer (SKOV3) cells were cultured in Dulbecco’s Modified Essential Medium (DMEM), containing 10% FBS and Penicillin and Streptomycin (P/S), at 37 °C in a 5% CO2 incubator. Proliferating cells were seeded at approximately 5 × 103 cells per well into black-walled 96 well plates, and transfected when cells reached approximately 40% cell density. Prior to transfection, the cells were washed with serum-free media. Polyplexes (50 mL polymer solution and 50 mL DNA solution at 1 mg/50 mL) were formed as described above, then diluted with serum-free media (400 μL), and the diluted polyplex solution (125 μL) was directly added to the cells. A transfection experiment consisted of testing each reagent at N/P ratios from 3−10, with 8 replicates at each N/P ratio. Each experiment was repeated on three different days using a different cell passage and plasmid DNA batch. 6 h post-transfection, the media containing transfection reagents was supplemented with 125 μL/well of fresh growth medium containing 20% FBS. 48 h post transfection, the cells were analyzed by fluorescence microscopy (Olympus IX71 fluorescence inverted microscope equipped with an Olympus DP71 digital camera), a plate reader in fluorescence mode (BMG Labtech FLUOstar OPTIMA plate reader), and finally by the CellTiter-Glo Luminescent Cell Viability Assay. 2.14. Protein Expression and Cell Viability. Red fluorescent protein expression was determined 48 h post transfection at different N/P ratios by a plate reader in fluorescence mode. Standard error of the mean was calculated for each data set. Cell viability was determined 48 h post transfection by Promega’s CellTiter-Glo Luminescent Cell Viability Assay, which was performed according to the recommended protocol. The CellTiter-Glo buffer and substrate were equilibrated at room temperature, and the buffer added to the substrate to form the CellTiter-Glo reagent. After equilibrating the 96well plate at room temperature, the cell media was replaced with 548

DOI: 10.1021/acs.biomac.5b01462 Biomacromolecules 2016, 17, 546−557

Article

Biomacromolecules

Figure 1. CG model for DNA (a), K4 (b), and SB (c). CG beads that have positive and negative charges are denoted by (+) and (−), respectively. serum-free media (50 μL), and then the mixed CellTiter-Glo reagent (50 μL) was added to each well. The plate was mixed on an orbital shaker for 5 min. Luminescence was recorded on a plate reader (BMG Labtech FLUOstar OPTIMA plate reader). The average luminescence values found at each condition (i.e., different reagent, N/P ratio) were divided by the average luminescence of the control cells (cells treated with media only) to determine the percent cell viability. 2.15. Percentage Polyplex Positive Cells and Cellular Internalization. SKOV3 cells were seeded in 12 well plates at 104 cells per well. Upon reaching 70% cell density, the cells were incubated with Label IT fluorescein-labeled plasmid DNA (Mirus Bio)-polymer complexes (N/P 4) for 3 h in serum-free DMEM at 37 °C in a 5% CO2 incubator. After 3 h, the cells were washed with cold phosphatebuffered saline (PBS), trypsinized, and resuspended in 0.1% Bovine Serum Albumin (w/v) solution. Cellular distribution and internalization was then measured using a BD LSRII flow cytometer.

bead representing the negatively charged carboxylate (red, −COO−) together form a single K4 unit. The red and the blue CG beads in Figure 1b have a charge of −1 and +1, respectively, while the remaining beads are uncharged. Each K4 unit is grafted to the hydrophobic backbone (black beads in Figure 1b). The backbone was modeled using three uncharged tangent beads of size 1.3σ and mass 1.0m. Each bead represents roughly three carbons and the corresponding hydrogen atoms. The K4 grafts were attached to the center backbone bead. To ensure transferability of the K4 and hydrophobic backbone interaction parameters to the SB-containing copolymers in this paper, the original model parameters26 were refined as detailed in the Supporting Information. The CG model for SB was developed to be compatible with the above coarse-grained model of K4 and the hydrocarbon backbone. Each SB molecule was modeled by four beads (Figure 1c): gray, −C(O)NH−, white, −(CH2)3−, orange, −SO3−−, and light blue, −CH2N+(CH3)2−. The first two groups were taken from the K4 model and share the same model parameters. The orange and light blue beads have a charge of −1 and +1, respectively. Nonbonded Lennard-Jones potentials’ size and energy parameters as well as bonded potentials were adjusted to reproduce the radius of gyration, end-to-end distance, and sulfonate-ammonium radial distribution functions obtained from all-atom simulations (Supporting Information). The above CG models of the DNA and comb polymers were used along with explicitly represented single CG bead monovalent ions and implicit solvent. Counterions and salt ions were modeled by a single CG sphere with mass of m, diameter of σ, Lennard-Jones interaction well depth of 0.6ε, and charge of +1 or −1 for positive and negative ions, respectively. An implicit solvent model was adopted by normalizing electrostatic Coulomb interactions with a dielectric constant set to 80 to mimic water at T = 300 K. All of the CG model details are summarized in Supporting Information. 3.2. Simulation. Molecular dynamics simulations were performed with the LAMMPS package61 using the CG model described above. Simulations were conducted in NVT ensemble in a box size of 300σ and at a reduced temperature of 6.0, roughly corresponding to 300 K. We used Langevin dynamics simulations given by

3. COARSE-GRAINED (CG) MOLECULAR SIMULATIONS 3.1. Model. The DNA and comb polymers containing K4, SB and a hydrocarbon backbone were modeled in a coarsegrained (CG) manner (Figure 1). The values of σ = 0.3 nm, ε = 0.1 kcal/mol, and m = 41.8 g/mol are used to represent dimensionless length, energy, and mass, respectively.26 Each DNA base pair was modeled with a single CG bead of size 2.2σ and mass 6.2m. Each DNA CG bead has a charge of −2 to represent two negatively charged phosphate groups in each base pair. The equilibrium bond distance between adjacent base pairs is 1.1σ to represent the experimentally observed distance of 0.34 nm between DNA base pairs. The bonded potential parameters of DNA were chosen to roughly reproduce the experimentally observed DNA persistence length (∼50 nm) at moderate to high salt concentrations59 (see Supporting Information). Even though this simple generic model of polyanions does not capture the helical structure of DNA or specific base chemistries, it is an appropriate model for this study because (a) the helical structure should have minimal impact on complexation and overall polyplex properties,60 and (b) the complexation is dominated by electrostatic interactions between phosphate groups, not the bases, in DNA and the positively charged groups in the polycation. The CG model for K4 was mapped from atomistic simulations as done in our original work.26 Each lysine monomer was modeled with three CG beads (Figure 1b): gray, −C(O)NHCH−, white, −(CH 2 ) 3 −, and blue, −CH2NH3+−. Four lysine monomers and one additional

F = −∇U ( r ⃗) − 549

m v⃗ + γ

kBTm R (t ) γdt

(1)

DOI: 10.1021/acs.biomac.5b01462 Biomacromolecules 2016, 17, 546−557

Article

Biomacromolecules

sequence, and were conducted using a fixed SB mole fraction in the polymer. 3.3. Analysis. To quantify structural properties of the polymers and DNA, the end-to-end distance, Ree, radius of gyration, Rg, and radial distribution function, g(r) were calculated. The radius of gyration of a selection of beads is the average distance of all beads in the selection to center of mass of that selection:

where F is the force on any CG bead, U is the interaction potential, m is the particle mass, γ is the friction coupling coefficient, v is the particle velocity, kB is the Boltzmann constant, T is temperature, and R(t) is a random number function with a Gaussian distribution. We set the friction coefficient, assumed to be distributed uniformly irrespective of diameter of each bead, equal to 5 (in dimensionless units of time) corresponding to the diffusion coefficient of an isolated linear DNA molecule in water (5 × 10−7 cm2/s).62 The equation of motion was integrated using the Verlet algorithm and a time step (dt) of 0.002 in reduced units, corresponding to 6.0 fs. Electrostatic interactions were treated with the particleparticle-particle-mesh (PPPM) method, with a force tolerance of 10−2 and interpolation order of 2. The Lennard-Jones potential was cut at 10σ, and the CHARMM switching function applied up to 12σ. The initial configurations for these simulations were prepared by packmol63 and also using in-house scripts in VMD.64 Details of the initial configurations are in Supporting Information. The equilibrium runs varied between 2 × 106 and 6 × 106 time steps, and the production runs between 1 × 106 and 4 × 106 time steps, depending on the N/P ratio and SB content in the polymer. In the production run, configurations were stored every 1000 time steps for data analysis. The number of polymers in the simulation box was varied between 1 and 5, and each polymer had 133 K4 monomers and 0, 19, 44, or 80 SB monomers, leading to SB contents of 0, 12.5, 25, and 37.5%, respectively. Only one DNA chain with a fixed length of 200 base pairs was simulated. With these choices for DNA and polymer lengths, the N/P ratio, defined as the number of protonatable amine groups in K4 divided by the number of phosphate groups in DNA, varied between 1.3 and 6, depending on the number of polymers in the simulation box. We note that if we need to simulate longer lengths of DNA and polymer, we would have to choose/develop a coarse-grained model that has far less chemical detail than the model we developed for this work. However, with that heavily coarsegrained model we would not be able to mimic the chemical details distinguishing the zwitterionic parts from the cationic parts of the polymer. Since the main goal of this work was to study effects of SB zwitterions on the polyplex structure, retention of the chemical details at the expense of polymerDNA length was justified. Simulations were carried out either with only neutralizing counterions or in the presence of 50 mM monovalent salt (i.e., 21950 positive and 21950 negative ions). As mentioned in the model, implicit solvent was used in all CG simulations. For simulations involving complexation of DNA with one polymer (N/P 1.3) one trial was performed because, regardless of initial configuration, only globular polyplexes are formed (Figure S11). In this case, the averages are the mean of the production run configurations and the error bars denote standard deviation of the mean. For simulations where more than one polymer interacts with DNA, polyplexes with a wide range of anisotropic shapes are formed (Figure S12). Therefore, 10 independent trials were performed for each simulation to explore different polymer-DNA binding modes. For simulations at N/P greater than 1.3, where 10 independent simulation trials were run, properties are averaged over all 10 trials. In this case, error bars denote the standard deviation of the average of the ten trials. Note that these trials vary in initial configuration, initial velocity, random number seed and SB

N

⟨R g 2⟩ =

1 ⟨∑ mi(R i − R com)2 ⟩ M i

(2)

where N is the total number of beads in the selection, Ri is the position of the bead i, mi is the mass of bead i, Rcom is the center of mass of the selection, M is the total mass of the selection, and brackets denote ensemble average. Polyplex size was determined by calculating Rg of the complexation region, which includes DNA and other CG beads within 0.75 nm (2.5σ) of any DNA bead. This distance corresponds to the Bjerrum length in water at 300 K.26 Polyplex shape was determined by calculating the polyplex RSA, which ranges from 0.0 for perfect spheres to 1.0 for perfect rods. The RSA was determined from eigenvalues in the radius of gyration tensor.26 The polyplex RSA was also calculated for CG beads in the complexation region. Both RSA and Rg are reported as probability distributions with grid sizes of 0.01 (dimensionless) and 0.06 nm, respectively. To quantify polyplex surface composition, polyplex accessible surface area (ASA) was determined using the measure sasa command of the VMD package with a probe radius of 0.5σ. To ensure that the measure sasa command probes the actual surface of polyplex rather than the pores inside the polyplex or unbound polymers, ASA was calculated for species in the polyplex region. Polyplex region was defined as a selection of beads that include all DNA beads and every other bead that within 3 nm of any DNA bead (see rationale in Supporting Information). To calculate polyplex net surface charge, the number of ions adsorbed on the polyplex surface was calculated. Ions were considered adsorbed if they were within the Bjerrum length of any beads in the polyplex region. From the number of adsorbed ions (nabsorbed), the polyplex net surface charge (ζ) was calculated according to ζ = −1e × (n+adsorbed − n−adsorbed)

(3)

where the subscripts refer to positive and negative ions and e denotes units of charge. The free energy change upon complexation was calculated as ΔA = ΔU − T ΔS + ΔA solv

(4)

where A is Helmholtz free energy, U is potential energy, S is entropy, Asolv is solvation free energy, and Δ denotes the difference between bound and unbound states such that for any quantity X: ΔX = X polyplex − X DNA(unbound) − X copolymer(unbound)

(5)

The potential energy, U, in eq 4 is the sum of bonded and nonbonded contributions: U = Ubond + Uangle + ULJ + UCoul

(6)

where subscripts LJ and Coul denote Lennard-Jones and electrostatic contributions to potential energy, respectively. 550

DOI: 10.1021/acs.biomac.5b01462 Biomacromolecules 2016, 17, 546−557

Article

Biomacromolecules The entropy was approximated as a sum of the conformational entropy loss of DNA and polymer upon complexation (conf) and the translational entropy gain of ions upon dissociation (ion): S = Sconf + Sion

(7)

We estimated the conformational entropy as

65

Sconf ≈ −kB ∑ pi (R ee) ln pi (R ee)

(8)

i

where pi(Ree) is the probability distribution of polymer end-toend distance and the summation runs over all observed values of Ree. To calculate probability distributions, a uniform set of scan points was selected for Ree with a grid size of 0.015 nm. The ion dissociation entropy was estimated by66 Sion = (1 − α+)n+ ln[(1 − α+)ϕ+] + (1 − α−)n− ln[(1 − α−)ϕ−] kB

(9)

where α is the time-dependent fraction of adsorbed ions, n is the total number of ions, ϕ is the volume fraction of ions in the simulation box, and the subscripts denote positive and negative ions. The solvation free energy was approximated as67 polar nonpolar ΔA solv ≈ ΔUsolv = ΔUsolv + ΔUsolv

Figure 2. Synthesis and 1H NMR spectrum (acquired in D2O) of SBCOE.

by DMAEA-COE in acetonitrile at 50 °C afforded SB-COE in 48% yield, as confirmed by 1H NMR spectroscopy, noting the presence of a multiplet at 5.5−5.8 ppm corresponding to the cyclic olefin protons, and a triplet at 2.9 ppm corresponding to the methylene protons adjacent to the sulfonate group. Mass spectrometry confirmed the expected molecular weight of SBCOE (m/z 347.2003, calcd = 347.2005 [M + H]+). Copolymerization of K4-COE and SB-COE proceeded readily in a TFE/CH2Cl2 mixture (Figure 3). The K4-COEto-catalyst ratio was held constant at 50, and the amount of SBCOE was varied from 0 to 45 mol %. After polymerization for 2 h at room temperature, the Boc-protected polymers were precipitated into ether and isolated by centrifugation as brown powders in >90% yield. The polymerizations proceeded to quantitative monomer conversion, as determined by 1H NMR spectroscopy by the disappearance of the multiplet at 5.5−5.8 ppm corresponding to the cyclic olefin protons and the appearance of the peak at 5.3 ppm corresponding to polymeric olefin protons (Figure S14). The Boc protecting groups were removed by acidification (4 M HCl dioxane/methanol 2/5 v/v) and the K4SB copolymers isolated by dialysis in water followed by lyophilization in 50−70% yield as white powders. Table 1 lists the composition, molecular weight, and molecular weight distribution of the K4SB copolymers. The mole % SB-COE was within 5% of the targeted values, as determined by 1H NMR spectroscopy in D2O (Figure S14). The GPC-estimated molecular weights of the deprotected copolymers ranged from 55 to 75 kDa, with polydispersity indices from 1.5 to 1.9, and the GPC traces are shown in Figure S15. The resulting series of comb polymers with both K4 and SB pendent groups having comparable molecular weights and varied SB compositions is ideal for comparing to simulations of DNA binding and polyplex behavior. 4.2. Evaluation of Polymer-DNA Binding. The effect of SB zwitterions on polymer-DNA binding strength was examined by a fluorescence-based assay and coarse-grained simulations (Figure 4). The data in Figure 4a used polyplexes formed at N/P 5 for 40 min, then adding Picogreen as a dye that increases its fluorescence intensity upon binding to DNA, and heparin as an anionic polymer that competes with DNA for binding. As heparin displaces DNA, the Picogreen fluorescence increases. The Picogreen fluorescence intensity relative to that

(10)

Considering our simple implicit solvent model with a distance-independent dielectric constant, the polar term, which includes the electrostatic interactions between solvent and solutes, was estimated as68 polar explicit ΔUsolv = UCoul − UCoul = UCoul − εrUCoul = (1 − εr)UCoul

(11)

Uexplicit Coul

where UCoul is the total electrostatic potential and refers to the electrostatic potential between solutes and a hypothetical explicit solvent. The nonpolar contribution is comprised of dispersion interactions between solvent and solutes, ΔUsolvent−solute , and the energy needed to form a cavity in the dispersion solvent, ΔUcavity. The nonpolar term was assumed to be proportional to the accessible surface area of the solute:69 nonpolar solvent−solute ΔUsolv = ΔUdispersion + ΔUcavity ≈

∑ χi

× ASAi

i

(12)

where χi is the bead-specific nonpolar solvation energy, ASAi is the accessible surface area of bead type i, and the summation runs over all bead types of solute. In this work, the bead-specific solvation energies were estimated based on the corresponding all-atom values available in literature (see Supporting Information). For charged polymers, the specific choice of bead-specific solvation energies is not critical because the nonpolar term in eq 10 is almost negligible comparing to the polar term.

4. RESULTS AND DISCUSSION 4.1. Synthesis of Zwitterion-Containing Comb Polymers. Zwitterionic SB groups were dispersed within cationic comb polymers by copolymerization of SB-substituted cyclooctene (SB-COE) with K4-substituted cyclooctene (K4-COE) by ROMP. Boc-protected K4-COE was prepared by solid phase peptide synthesis (SPPS) as reported previously.24 As shown in Figure 2, SB-COE was prepared from the corresponding dimethylaminoethylamido (DMAEA) precursor, synthesized by the coupling of (Z)-cyclooct-4-ene-1-carboxylic acid with N,Ndimethylethylenediamine. Ring-opening of 1,3-propane sultone 551

DOI: 10.1021/acs.biomac.5b01462 Biomacromolecules 2016, 17, 546−557

Article

Biomacromolecules

Figure 3. Synthesis of K4SB comb polymers by ROMP of K4-COE and SB-COE in a mixture of TFE and CH2Cl2. The Boc protecting groups were removed in 4 M HCl dioxane/methanol (2/5 v/v).

Table 1. Molecular Weight and Composition of K4SB Copolymers sample

target SB (mol %)

incorporated SB (mol %)

Mna (g/mol)

Mw/Mna

K4 K4SB-17 K4SB-34 K4SB-49

0 15 30 45

0 17 34 49

57600 58900 70100 72800

1.49 1.89 1.69 1.73

a

Estimated by TFE GPC (PMMA standards) of deprotected copolymers.

in the presence of DNA alone provides a measure of the percent DNA accessible to Picogreen. The plot in Figure 4a of [100 − (PG fluorescence/PG fluorescence*)] × 100%, where PG fluorescence* is the Picogreen fluorescence in the absence of cationic polymer, reflects the relative percentage of bound DNA 16 h after addition of Picogreen and heparin to the polyplexes. Additionally, we note that the values in Figure 4a convey relative percentages of bound DNA upon incubation of the polyplexes with 40 units of heparin; the percentage of bound DNA decreases with increasing heparin concentration (Figure S16). Lysine-based comb polymers were found to bind DNA more loosely than linear PLL, consistent with our previous reports.24,25 Adding SB zwitterions into the K4-based comb polymers decreased the percent bound DNA from 67% for K4 to 34% for K4SB-49, confirming that the zwitterionic groups markedly decrease polymer-DNA binding strength. Simulations also show that polymer-DNA binding strength, or favorable binding free energy, decreases with increasing SB content in the comb polymers (Figure 4b). The enthalpic driving force for polymer-DNA binding is dominated by electrostatic interactions between the negatively charged DNA and net positively charged polymer, as well as the electrostatic interactions between the solvent and charged species in the polyplex (captured through the solvation term in the free energy calculations described in section 3.3). Figure S17 shows each of these contributions to the binding free energy as a function of SB composition. Calculations of the entropic contributions to polymer-DNA binding show that the conformational entropy loss of the polymer chains and DNA upon complexation is negligible compared to the gain in ion dissociation entropy (Table S7 and ref 68). The enthalpic contribution to polymer-DNA binding includes the potential energy change upon complexation. To connect the change in potential energy upon complexation to the molecular contacts within the polyplex, we calculated radial

Figure 4. Effect of SB content on polymer-DNA binding affinity: (a) Polymer-DNA complexation strength assessed by heparin-induced polyplex decomplexation experiments, wherein polyplexes (at N/P 5) are incubated with heparin (40 units) and Picogreen dye (1×). The quantity, [100 − (PG fluorescence/PG fluorescence*)] × 100%, reflects the percentage of bound DNA determined relative to Picogreen fluorescence with no polymer added (PG fluorescence*). (b) Helmholtz free energy of polymer-DNA binding determined using coarse-grained simulations at N/P 4.0. Experiments and simulations were conducted in the absence of salt.

distribution functions of polymer-DNA pairs, or the number of CG polymer beads within the complexation region (