Nanostructure Formation and Transition from Surface to BulkDegradation in Polyethylene Glycol Gels Chain-Extended with ShortHydroxy Acid SegmentsSeyedsina Moeinzadeh, Danial Barati, Samaneh K. Sarvestani, Ozan Karaman, and Esmaiel Jabbari*

Biomimetic Materials and Tissue Engineering Laboratories, Department of Chemical Engineering, University of South Carolina,Columbia, South Carolina 29208, United States

*S Supporting Information

ABSTRACT: Degradable, in situ gelling, inert hydrogels withtunable properties are very attractive as a matrix for cellencapsulation and delivery to the site of regeneration. Celldelivery is generally limited by the toxicity of gelation anddegradation reactions. The objective of this work was toinvestigate by simulation and experimental measurementgelation kinetics and degradation rate of star acrylatedpolyethylene glycol (PEG) macromonomers chain-extendedwith short hydroxy acid (HA) segments (SPEXA) as afunction of HA monomer type and number of HA repeatunits. HA monomers included least hydrophobic glycolide(G), lactide (L), p-dioxanone (D), and most hydrophobic ε-caprolactone (C). Chain extension of PEG with short HAsegments resulted in micelle formation for all HA types. There was a significant decrease in gelation time of SPEXA precursorsolutions with HA chain-extension for all HA types due to micelle formation, consistent with the simulated increase in acrylate−acrylate (Ac−Ac) and Ac−initiator integration numbers. The hydrolysis rate of SPEXA hydrogels was strongly dependent on HAtype and number of HA repeat units. SPEXA gels chain-extended with the least hydrophobic glycolide completely degradedwithin days, lactide within weeks, and p-dioxanone and ε-caprolactone degraded within months. The wide range of degradationrates observed for SPEXA gels can be explained by large differences in equilibrium water content of the micelles for different HAmonomer types. A biphasic relationship between HA segment length and gel degradation rate was observed for all HAmonomers, which was related to the transition from surface (controlled by HA segment length) to bulk (controlled by micelleequilibrium water content) hydrolysis within the micelle phase. To our knowledge, this is the first report on transition fromsurface to bulk degradation at the nanoscale in hydrogels.

■ INTRODUCTIONHydrogels are used extensively in medicine for soft tissuerepair.1−4 Due to their high diffusivity of nutrients andbiomolecules,5−7 hydrogels are very useful as a matrix in tissueengineering for in situ delivery of cells to the site ofregeneration and regulation of cell fate.8−11 Only thosehydrogels that degrade and provide free volume for the newlyformed tissue can be utilized as a matrix in regenerativemedicine.12,13 However, viability and fate of the encapsulatedcells are limited by toxic side effects of gelation and degradationreactions in the hydrogel matrix. Consequently, naturalhydrogels derived from the components of the extracellularmatrix (ECM) of biological tissues that physically cross-linkand degrade enzymatically are frequently used as the deliverymatrix in clinical applications.Minor variation in the sequence distribution of natural gels

can dramatically affect the fate of encapsulated cells. With manycell-interactive ligands and regulatory factors, it is difficult totailor the properties of natural matrices to a particularapplication in regenerative medicine.14 For example, differ-

entiation of mesenchymal stem cells (MSCs) shifted fromosteogenic to chondrogenic lineage by changing the matrixfrom collagen type I to type II.15,16 Moreover, due to their lowstiffness, natural gels are limited in practical applications by softtissue compression.17−19 Therefore, there is a need tosynthesize novel hydrogels with tunable physical, mechanical,and biological properties for a wide range of applications inregenerative medicine such as chondrocyte implantation incartilage regeneration or as cardiac patches to treat heartinfarction.Polyethylene glycol (PEG) hydrogels are inert, nonimmuno-

genic, and compatible with stem cells and can be conjugatedwith multiple bioactive peptides to modify the cell micro-environment and regulate cell fate.20−22 However, PEGhydrogels are nondegradable, which limits their use as asupporting matrix in regenerative medicine. PEG macro-

Received: June 7, 2013Revised: July 11, 2013Published: July 16, 2013

Article

pubs.acs.org/Biomac

© 2013 American Chemical Society 2917 dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−2928

monomers copolymerized with hydroxy acid (HA) monomersproduce block copolymers that have limited solubility inaqueous solution18 and self-assemble to form nanoparticles fordrug delivery.23−29 We previously demonstrated that star PEGmacromonomers chain-extended with very short hydrolyzablelactide segments (SPELA) dissolved in aqueous solution andformed a micellar gel.18 Remarkably, the results showed abiphasic relationship between the lactide segment length andgelation or degradation rate. We hypothesized that theobserved biphasic dependence was related to the transitionfrom surface to bulk hydrolysis in the micelle phase withincreasing lactide segment length. To test the hypothesis, weinvestigate in this work gelation and degradation of star PEGmacromonomers chain-extended with short HA segments(SPEXA where “X” is the HA monomer type) with a widerange of hydrophobicity. HA monomers included leasthydrophobic glycolide (G), lactide (L), p-dioxanone (D), andmost hydrophobic ε-caprolactone (C). All HA chain-extendedPEG hydrogels showed the transition from surface to bulkdegradation with increasing HA segment length. Chainextension of PEG macromonomers with short HA segmentslocalized the reactive moieties within micelles and reduced theiraverage distance, thereby increasing gelation and degradationrates with increasing segment length. As the micelle size wasincreased above a certain value, the average distance betweenreactive moieties within the micelles began to increase, anddegradation rates decreased.

■ SIMULATION METHODSolution of SPEXA macromonomers in water was simulated viadissipative particle dynamics (DPD) using an approachdescribed previously.18,30 The molecular structure of themacromonomer was divided into different beads with equalmass, as shown in Figure 1. The beads included L (lactide

repeat unit), G (glycolide), D (p-dioxanone), C (ε-caprolactone), EO (ethylene oxide repeat unit), Ac (acrylatefunctional group), SPEGc (star PEG core), and W (three watermolecules). The meso-structure of the macromonomer is alsoshown in Figure 1. The notations SPEXA-nA or SPEXA-mB areused to identify length of the degradable segment, where A isthe number of repeat units or ester groups per arm, B is the

number of monomers per arm, and X is HA monomer type (G,L, D, and C). When X is C or D, A equals B, but A = 2B whenX is G or L. In DPD, each bead represents a soft particleinteracting with the other beads via a soft pairwise forcefunction given by31,32

∑= + + +≠

f F F F Fiji j

ij ij ij ijC D R S

(1)

where f ij is the total force and FijC, Fij

D, FijR, and Fij

S are theconservative, dissipative, random, and spring components ofthe force, respectively. Different components of the force in acutoff distance (rc) are calculated by33

α=

− | | <

| | ≥⎪⎪⎧⎨⎩

Fr e r

r

(1 ) 1

0 1ijij ij ij ij

ij

C

(2)

γ= − | | ·F w r e v e[ ( )]( )ij ij ij ij ijD D

(3)

σ θ= | |F w r e[ ( )]ij ij ij ijR R

(4)

∑=F Crijj

ijS

(5)

where rij is the vector joining bead i to j, eij and |rij| are the unitvector in the direction of rij and the magnitude of rij,respectively. vij is the velocity vector given by vij = vi − vj. w

D

and wR are weight functions for dissipative and random forces,respectively, and γ and σ are the magnitude of dissipative andrandom forces. Fij

D and FijR act simultaneously to preserve

dissipation and to conserve the total momentum in the system.The dissipative and random force constants and weightfunctions are interrelated by wD(rij) = [wR(rij)]

2 and σ2 = 2kBTγ in order to satisfy the dissipation−fluctuation condition.34

The spring force term imposes geometrical constraints on thecovalently bonded beads. Values of γ and C constants were 4.5and 4, respectively.30 The repulsion between beads i and j ismainly dictated by the constant αij in the conservative forcefunction. By choosing the system density ρ = 3 rc

−3, the DPDlength scale, rc, was 6.74 Å and the values of αij weredetermined using33

α χ= +78 3.27ij ij (6)

where χij is the Flory−Huggins parameter between beads i andj. Values of χij in turn are given by

χδ δ

=− V

RT

( )ij

i j2

(7)

where δi and δj are solubility parameters of beads i and j,respectively, V is bead molar volume, T is absolute temperature,and R is gas constant. Solubility parameters were calculated viaatomistic molecular dynamics simulation performed via Forciteand Amorphous Cell modules, Materials Studio (v5.5,Accelrys)35 using the COMPASS force field, which is an abinitio force field optimized for condensed-phase systems.36

Position and velocity of the beads at each time point wereobtained by solving the following equations of motion using theforce function (eq 1).

= =rt

v mvt

fdd

,dd

ii i

ii (8)

Figure 1. Bead reperesentation of SPEXA (X = L, G, C, or D)macromonomer in DPD. Beads SPEGc (yellow), EO (green), G(blue), D (pink), L (orange), C (purple), and Ac (red) represent starPEG core, ethylene oxide repeat unit, glycolide, p-dioxanone, lactide,ε-caprolactone repeat unit, and acrylate functional group, respectively.

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282918

All DPD simulations were performed in 30 × 30 × 30 rcboxes with three-dimensional (3D) periodic boundaryconditions over 2 × 105 time steps and dimensionless timestep of 0.05. The Mesocite module of Materials Studio (v5.5,Accelrys)35 was used for DPD calculations.

■ EXPERIMENTAL METHODSMaterials. Lactide (L), glycolide (G), and p-dioxanone (D)

monomers with >99.5% purity were purchased from Ortec (Easley,SC). ε-Caprolactone (C) monomer with >99% purity was purchasedfrom Alfa Aesa (Ward Hill, MA). All monomers were dried undervacuum at 40 °C for at least 12 h prior to use. 4-(2-hydroxyethoxy)-phenyl-(2-hydroxy-2-propyl) ketone (Irgacure-2959) photoinitiatorwas obtained from CIBA (Tarrytown, NY). Calcium hydride,tetrahydrofuran (THF), deuterated chloroform (99.8% deuterated),trimethylsilane (TMS), triethylamine (TEA), tin(II) 2-ethylhexanoate(TOC), acryloyl chloride (Ac), dimethylsulfoxide (DMSO), 4-armPEG (SPEG, 4 arm, nominal Mw = 5 kDa), ethylenediaminetetraaceticacid disodium salt (EDTA), and paraformaldehyde were purchasedfrom Sigma-Aldrich (St. Louis, MO). The protected amino acids andRink Amide NovaGel resin for the synthesis of acrylamide-terminatedGRGD peptide were purchased from EMD Biosciences (San Diego,CA). Dichloromethane (DCM, Acros Organics, Pittsburgh, PA) wasdried by distillation over calcium hydride. Other solvents wereobtained from VWR (Bristol, CT) and used as received. Dulbecco’sphosphate-buffer saline (PBS) and Dulbecco’s modified Eagle’smedium (DMEM; 4.5 g/L glucose with L-glutamine and withoutsodium pyruvate) were obtained from GIBCO BRL (Grand Island,NY). Fetal bovine serum (FBS), screened for compatibility with ratMSCs, was purchased from Atlas Biologicals (Fort Collins, CO).TRIzol for isolation of cellular RNA and trypsin were purchased fromInvitrogen (Carlsbad, CA). Penicillin (PN), streptomycin (SP),fungizone (FG), gentamicin sulfate (GS), dexamethasone (DEX),ascorbic acid (AA), and β-sodium glycerophosphate (GP) werepurchased from Sigma-Aldrich. Quant-it PicoGreen dsDNA reagent kitwas obtained from Invitrogen (Carlsbad, CA). QuantiChrom calciumand alkaline phosphatase (ALP) assay kits were purchased fromBioassay Systems (Hayward, CA). Sircol total collagen assay kit wasobtained from Biocolor (Carrickfergus, UK).Characterization. The chemical structure of SPEXA macro-

monomers was characterized by a Varian Mercury-300 1H NMR(Varian, Palo Alto, CA) at ambient conditions with a resolution of 0.17Hz as described previously.18 The sample was dissolved in deuteratedchloroform at a concentration of 50 mg/mL, and 1% v/v TMS wasused as the internal standard. Number- (Mn) and weight-averagemolecular weight (Mw) and polydispersity index (PI) of themacromonomers were measured by gel permeation chromatography(GPC, Waters 717 System, Milford, MA) in THF with 1 mL/min flowrate as previously described.37 Monodisperse polystyrene standards(Waters) with peak molecular weights (Mp) of 0.58−19.9, 66.35, and143.4 kDa and polydispersities of less than 1.1 were used to constructthe calibration curve.Macromonomer Synthesis. SPEG macromer chain-extended

with short HA segments was synthesized by ring-opening polymer-ization (ROP) as described.18 SPEG and TOC were the polymer-ization initiator and catalyst, respectively. Briefly, dry HA monomerand SPEG were added to a three-neck reaction flask with an overheadstirrer and immersed in an oil bath (only SPEG was added to the flaskfor D monomer). Molar ratio of SPEG to monomer was selectedbased on the desired theoretical length of the HA segment. Next, thereaction flask was heated to 120 °C under nitrogen to melt themixture, maintained at that temperature for 1 h to remove moisture,and TOC was added to the mixture. The reaction was run at 140 °Cfor 12 h for C and L monomers and 160 °C for 10 h for G monomer.Since equilibrium was shifted toward monomer in polycondensation ofp-dioxanone for temperatures >100 °C, the SPEG and catalyst mixturewas heated to 130 °C for 10 min to remove moisture, the mixture wascooled to 85 °C, D monomer was added, and the reaction was run atthat temperature for 48 h. After the reaction, the product was purified

by precipitation in ice-cold hexane to remove any unreacted monomer,initiator, and catalyst.

In the next step, chain ends of the macromer were acrylated toproduce SPEXA macromonomer. SPEX macromer (product of thefirst reaction) was dried by azeotropic distillation from toluene. Themacromer was dissolved in DCM, the flask was immersed in an icebath, and the reaction was carried out by the addition of equimolaramounts of Ac and TEA dropwise to the macromer solution under drynitrogen. After 12 h, solvent was removed by rotary evaporation, andthe residue was dissolved in ethyl acetate to precipitate the byproducttriethylamine hydrochloride salt. After vacuum distilling ethyl acetate,the product was redissolved in DCM and precipitated in ice-cold ethylether twice. The product was dissolved in DMSO and dialyzed againstwater in a 3.5 kDa MW cutoff Spectro/Por dialysis tube (SpectrumLaboratories, Rancho Dominquez, CA) to remove any remainingimpurities. SPEXA macromonomer was dried in vacuum to removeresidual solvent and stored at −20 °C.

Macromonomer Gelation and Rheological Measurements.The aqueous SPEXA hydrogel precursor solution was cross-linked byUV free-radical polymerization with Irgacure-2959 photoinitiator aspreviously described.38 To monitor gelation kinetics, the hydrogelprecursor solution of the photoinitiator and macromonomer on thepeltier plate of an AR-2000 rheometer (TA Instruments, New Castle,DE) was irradiated with a long wavelength (365 nm) mercury UVlamp (model B100-AP; UVP, Upland, CA) at a distance of 10 cm fromthe sample. The gap distance between the peltier plate and the upper20 mm transparent acrylic geometry was 500 μm. A sinusoidal shearstrain with frequency of 1 Hz and amplitude of 1% was exerted on thesample via the upper geometry, and the storage (G′) and loss moduli(G″) were recorded with time. The time at which G′ = G″ wasrecorded as the gelation time. The notation X-mN is used for thehydrogels with X representing the HA monomer, m for monomer, andN for the number of HA monomers per macromonomer arm. Forexample, m0 denotes PEG hydrogel without chain extension with HAmonomer and C-m1.8 denotes SPECA hydrogel with average of 1.8 ε-caprolactone monomers per macromonomer arm.

Measurement of Equilibrium Water Content and Mass Loss.Hydrogels of 20 mm diameter and 300 μm thickness were removedfrom the rheometer and dried in ambient conditions for 12 h followedby drying in vacuum for 1 h at 40 °C. Dry samples were swollen indeionized (DI) water for 24 h at 37 °C with a change of swellingmedium every 6 h. After swelling, surface water was removed, and theswollen weights (ws) were measured. The swollen samples were driedas described above and dry weights (wd) were recorded. Theequilibrium water content was calculated by dividing the weight ofwater (the difference between swollen and dry weights) by the swollenweight as described.18 To measure mass loss, the hydrogel precursorsolution was cross-linked in a polytetrafluoroethylene (PTFE) mold (5cm × 3 cm × 750 μm) covered with a transparent glass plate. Disk-shape samples were cut from the gel, and their mass loss was measuredin 5 mL PBS at 37 °C under mild agitation. At each time point,samples were washed with DI water several times and dried undervacuum. The dried sample weight was measured and compared withthe initial dry weight to determine mass loss as described.18

Marrow Stromal Cell Isolation and Encapsulation in Hydro-gels. MSCs were isolated from the bone marrow of young adult maleWistar rats as described.18,39 Cell isolations were performed under aprotocol approved by the Institutional Animal Care and UseCommittee of the University of South Carolina. The suspension ofbone marrow cells was centrifuged at 200g for 5 min, and the cellpellets were resuspended in 12 mL basal medium consisting of DMEMsupplemented with 10% FBS, 100 units/mL PN, 100 μg/mL SP, 50μg/μL GS, and 250 ng/mL FG, and cultured in a humidified 5% CO2incubator at 37 °C. Cultures were replaced with fresh medium at 3 and7 days to remove hematopoietic and other unattached cells. After 10days, cells were detached from the flasks with 0.05% trypsin−0.53 mMEDTA and used for in vitro experiments. The experimental groups forencapsulation of MSCs in SPEXA hydrogels included m0, L-m1.7, D-m1.7, and C-m1.8. SPEGA hydrogel was not used for cellencapsulation because it completely degraded in a few days (see

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282919

Figure 6d). Cell encapsulation and osteogenic differentiation experi-ments were carried out in SPEXA hydrogels while maintaining aconstant compressive modulus of 50 kPa by varying the concentrationof SPEXA macromonomer in the hydrogel precursor solution.Acrylamide-terminated GRGD peptide (Ac-GRGD) was synthesizedon Rink Amide NovaGel resin in the solid phase, purified by high-performance liquid chromatography (HPLC), and characterized byelectrospray ionization (ESI) mass spectrometry as previouslydescribed.21 Ac-GRGD peptide in the amount of 2 wt %, based onthe macromonomer weight, was added to the hydrogel precursorsolution to facilitate cell adhesion to SPEXA matrix, and the mixturewas sterilized by filtration. Next, 1 × 106 MSCs suspended in 100 μLof PBS was gently mixed with the hydrogel precursor solution to reachthe final density of 5 × 106 cells/mL. The mixture was injected

between two sterile glass slides and cross-linked as described above.UV exposure time for all cell-seeded precursor solutions was 200 s,which was the minimum required time for the gel to reach its plateaumodulus. After gelation, samples were incubated in 2 mL PBS for 1 hwith two PBS changes. Next, the medium was replaced with completeosteogenic medium (basal medium supplemented with 100 nM DEX,50 μg/mL AA, 10 mM GP) and cultured for 28 days. To measure theeffect of photoinitiator on cell viability, 10 mg/mL initiator was addedto the suspension of MSCs in SPEXA precursor solution. After 10 min,the suspension was centrifuged, and cells were resuspended andcultured in basal medium (without initiator). After incubation for 2days, MSCs were counted with a hemocytometer, and cell numberswere normalized to those cells cultured in PBS (no exposure toinitiator).

Figure 2. Evolution of core of the micelles in 20% aqueous solutions of SPEXA. Only X and Ac (red) beads are shown for clarity. “n” is the numberof HA repeat units. G, D, L and C beads are shown by blue, pink, orange and purple, respectively. The scale bars are 5 nm.

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282920

Biochemical Analysis and mRNA analysis. At each time point(7, 14, 28 days), gel samples were washed with serum-free DMEM for8 h to remove serum components, washed with PBS, lysed with lysisbuffer (10 mM tris and 2% triton), and sonicated. After centrifugation,the supernatant was used for measurement of total collagen content,ALP activity, and calcium content. Total collagen content wasmeasured with a Sircol collagen assay based on selective binding ofG-X-Y amino acid sequence of collagen to Sircol dye, according tomanufacturer’s instructions.40 Briefly, 1 mL of Sircol dye was added to

the sonicated cell lysate, incubated for 30 min, and centrifuged at 10000 rpm for 5 min to separate the collagen−dye complex. Afterremoving the supernatant, the collagen−dye complex was mixed with1 mL Sircol alkali reagent, and absorbance was measured on a platereader at 555 nm. ALP activity and calcium content as a measure oftotal mineralized deposit were measured using QuantiChrom ALP andcalcium assays as previously described.41 For mRNA analysis, at eachtime point, total cellular RNA of the sample was extracted andconverted to cDNA as described.41 The cDNA was amplified with

Figure 3. Cross section of the micelles formed in 20% aqueous solutions of SPEXA. “n” is the number of HA repeat units. SPEGc, EO, and Ac beadsare shown by yellow, green and red, respectively. G, D, L, and C beads are shown by blue, pink, orange, and purple, respectively. Water beads are notshown for clarity. The scale bars are 2 nm.

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282921

gene specific primers designed using the Primer3 software aspreviously described.42 Expression of collagen type I (Col-I), ALP,and osteocalcin (OC) was measured by performing real-timequantitative polymerase chain reaction (RT-qPCR) using a CXF96PCR system (Bio-Rad, Hercules, CA) with the following primers(synthesized by Integrated DNA technologies, Coralville, IA): Col-1:forward 5′-GCA TGT CTG GTT AGG AGA AAC C-3′ and reverse5′-ATG TAT GCA ATG CTG TTC TTG C-3′; ALP: forward 5′-CCT TGA AAA ATG CCC TGA AA-3′ and reverse 5′-CTT GGAGAG AGC CAC AAA GG-3′; OC: forward 5′-AAA GCC CAG CGACTC T-3′ and reverse 5′-CTA AAC GGT GGT GCC ATA GAT-3′;S16: forward 5′-AGT CTT CGG ACG CAA GAA AA-3′ and reverse5′-AGC CAC CAG AGC TTT TGA GA-3′.40 The expression ratio ofthe gene of interest to that of S16 housekeeping gene was determinedusing the Pfaffl model43 and normalized to the first time point.Statistical Analysis. Data are expressed as means ± standard

deviation. All experiments were done in triplicate. Significantdifferences between groups were evaluated using a two-way ANOVAwith replication test followed by a two-tailed Student’s t test. A value ofp < 0.05 was considered statistically significant.

■ RESULTS AND DISCUSSIONCharacterization of Macromonomers. 1H NMR spectra

of star-PEG (m0), SPEGA (G-m2.8), SPEDA (D-m2.9),SPELA (L-m2.9), and SPECA (C-m2.8) are provided inFigures S1.a to S1.e of the Supporting Information,respectively. The assignment of chemical shifts for PEGmacromer in the NMR spectra of lactide chain-extendedSPEXA was previously described by us.18 The shifts with peakposition at 3.6 and 4.3 ppm in Figure S1.a−e corresponded tomethylene protons of PEG attached to ether and ester groupsof HA repeat units, respectively. The shifts with peak positionat 4.8−4.9 ppm in Figure S1.b corresponded to methyleneprotons of glycolide; those at 3.7, 4.2, and 4.4 ppm in FigureS1.c corresponded to β, α, and γ methylene protons of p-dioxanone, respectively; those at 1.6 and 5.2 ppm in Figure S1.dcorresponded to methyl and methine protons of lactide; andthose at 1.4, 1.7, 2.3, and 4.1 ppm in Figure S1.e correspondedto γ, β, α, and ε methylene protons of ε-caprolactone. Theshifts with peak position from 5.85 to 6.55 ppm were attributedto vinyl hydrogens of the Ac as follows: peak positions in the5.82−5.87 ppm range corresponded to the trans protons ofunsubstituted carbon in Ac; those in the 6.10−6.20 ppm rangecorresponded to the protons bonded to monosubstitutedcarbon; and those in the 6.40−6.46 ppm range corresponded tothe proton of unsubstituted carbon. The number of HA repeatunits per macromonomer (m) was determined from the ratio ofthe shifts centered at 4.8−4.9 ppm (glycolide hydrogens), 3.7,4.2, and 4.4 ppm (p-dioxanone), 1.6 and 5.2 ppm (lactide) and1.4, 1.7, 2.3, and 4.1 ppm (ε-caprolactone) to those at 3.6 and4.3 ppm (PEG). The number of acrylate groups permacromonomer was determined from the ratio of the shiftsbetween 5.85 and 6.55 ppm (acrylate hydrogens) to those at3.6 and 4.2 ppm (PEG). Mn of SPEGA, SPELA, SPEDA, andSPECA macromonomers depending on HA segment lengthwas in the range of 5.6−6.5 kDa, 5.8−7.4 kDa, 5.4−6.3 kDa,and 5.5−6.4 kDa, respectively, and the corresponding fractionof acrylated chain-ends was in the range of 0.78−0.87, 0.75−0.86, 0.73−0.82, and 0.75−0.85. Polydispersity of SPEGAmacromonomers, as measured by GPC, was in the range of1.24−1.44 (1.24, 1.34, 1.42, and 1.44 for G-m0.8, G-m1.2, G-m1.6, and G-m2.8, respectively); that of SPELA was 1.17−1.34(1.17, 1.23, 1.30, and 1.34 for L-m0.8, L-m1.7, L-m2.9, and L-m3.7, respectively); that of SPEDA was 1.20−1.38 (1.20, 1.28,1.38, and 1.35 for D-m0.6, D-m1.1, D-m1.7, and D-m2.9,

respectively); and that of SPECA was 1.22−1.38 (1.22, 1.25,1.34, and 1.38 for C-m0.7, C-m1.2, C-m1.8, and C-m2.8,respectively).

Structure Formation in SPEXA Precursor Solutions.Formation of nanoscale structure by SPEXA macromonomersin aqueous medium is shown in Figure 2. In the absence of HAmonomer, the distribution of acrylate groups attached toSPEXA chain ends was uniform in aqueous medium,30 but theextension of SPEXA arms with hydrophobic X segmentsinduced aggregation as shown in Figure 2. For SPEGA andSPEDA with n = 2, hydrophobic segments were not sufficientlylong to form stable micellar structures. However, SPELA andSPECA macromonomers (see first column of Figure 2), due tohigher hydrophobicity of lactide and ε-caprolactone monomers,formed stable micelles with n = 2. All four SPEXAmacromonomers formed stable micelles for n = 4 (secondcolumn of Figure 2). The aggregate size increased, and numberdensity decreased with increasing n from 4 to 8. According tothe theory of micellization in block copolymers in solution, thedegree of aggregation increases with increasing block size,driven by the decrease in overall surface free energy ofsolvophobic blocks.44 Cross-sectional view of one of theaggregates along with its EO beads is shown in Figure 3 fordifferent number of HA repeat units. In the images of Figure 3,hydrophobic HA segments and hydrophilic EO beads formedthe core and corona of the micelles, respectively. In ourprevious work, we showed that a change in core size andaggregation number for SPELA micelles is dominated byinterfacial free energy (the product of interfacial tension γ andinterface area a).30 Due to the presence of hydrophilic EOsegments at the interface, the effective interfacial tension inSPEXA micelles is different from the interfacial tensionbetween the micelle core and water, γC−W.

45 γ can be calculatedby minimizing the chemical potential of the aqueous solution atequilibrium46

γ γ

χ

= +−−

− − −

+ −

−

−⎜ ⎟

⎡⎣⎢

⎛⎝

⎞⎠⎤⎦⎥

kTs

CC

NN

C C

C C

ln11

1( )

12

34

C W 2i

bi b

W EO i2

b2

(9)

where γC−W is interfacial tension between the micelles’ core andwater, k, T, s, and N are the Boltzmann constant, absolutetemperature, statistical EO segment length, and number ofstatistical EO segments on each SPEXA arm, respectively. Ciand Cb are concentrations of EO segments at the interface andin bulk, respectively. Equation 9 implies that an increase in Cihas a negative contribution to interfacial tension. In otherwords, dense EO coverage of the interface decreases effectiveinterfacial tension between the hydrophobic domains andwater. According to Figure 3, the micelle core size of SPECAmacromonomers with n = 8 was similar to that of othermonomers even though ε-caprolactone was significantly morehydrophobic than the other monomers. This discrepancy canbe explained by higher packing of EO segments in the corona,thus decreasing the effective interfacial tension of SPECAmicelles. Effect of number of hydrophobic X beads on corediameter of micelles, number of macromonomers per micelle(aggregation number), number density of micelles, and fractionof macromonomer free arms is shown in Figure 4a−d,respectively. Assuming there were only X and Ac beads inthe core, the total number of beads taking part in core

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282922

formation per macromonomer equals 4(n + 1) and aggregationnumber is30

ρ=

+n

Vn4( 1)agg

c

(10)

where ρ and Vc are bead number density and micelle corevolume, respectively. Core radius of SPEGA and SPEDAmicelles increased from 0 to 22 Å when n increased from 2 to 8.Core radius of SPELA and SPECA micelles increased from 9and 11 Å to 23 and 24 Å, respectively, with increasing n from 2to 8. Aggregation number showed an increasing trend with nafter micelle formation (n = 2 for L and C and n = 4 for G andD). Average aggregation number of SPECA increased from 4 to19 when n increased from 2 to 8 which was the highestaggregation number among the four macromonomers. SPEGAhad the lowest aggregation number, which ranged from 0 to 14when n increased from 2 to 8. The increase in SPEXAaggregation number with increasing n is attributed to theincrease in the volume of hydrophobic segments and thedecrease in corona thickness of micelles, leading to an increasein effective interfacial tension between the core and water withincreasing n.30

Number density of micelles initially increased with n due to atransition from uniform distribution of macromonomers in thesystem to the formation of micelles. The number density thendecreased with n due to the increase in size and aggregationnumber of micelles. Figure 4d shows the effect of number ofHA monomers on each arm (n) on the fraction of free arms.For n = 0, acrylate groups were uniformly distributed insolution, and the fraction of free arms was unity. The fraction offree arms for SPELA decreased from 1 to 0.70, 0.14, 0.05, and 0as n increased from 0 to 2, 4, 6, and 8, respectively, whereas forSPEDA it decreased from 1 to 0.93, 0.23, 0.09, and 0. Thefraction of free arms decreased at a faster rate for SPECA andreached 0 for n = 6. SPEGA macromonomers with n ≤ 2 did

not form micelles and had a free arm fraction of unity, andfraction of SPEGA free arms decreased to 0.57, 0.18, and 0.04as n increased from 2 to 4, 6, and 8, respectively. The slowerrate of decrease in fraction of free arms in SPEGA wasconsistent with lower hydrophobicity of glycolide compared tothe other monomers.

Gelation Kinetics of SPEXA Precursor Solution. Therate of cross-linking of SPEXA macromonomers in aqueoussolution depended on the proximity of acrylate groups tophotoactivated acrylates, while the rate of photoactivation ofacrylates depended on the proximity of initiator molecules toacrylate groups. Therefore, the rate of cross-linking dependedon average distance between the acrylates and initiatormolecules. Distribution of photoinitiator beads (pink color)in SPELA-m3 solution and cross-section of one of the micellecores are shown in Figure 5a,b, respectively. Simulation imagesindicate that 98% of the photoinitiator beads partitioned to thehydrophobic core of the micelles in the proximity of acrylates(see pink beads in Figures 5a,b). The simulated fraction ofinitiators in the aqueous phase (those not partitioned to the

Figure 4. Effect of number of degradable HA repeat units on each armon (a) core radius (a), aggregation number (b), number density ofmicelles (c), and free arm fraction of the micelles (d) in 20% aqueoussolutions of SPEXA. Error bars correspond to means ± SD for 5simulation runs.

Figure 5. (a) Distribution of photoinitiator molecules in thesimulation box in SPELA-m3 precursor solution and (b) in thecorresponding cross-section of one of the micelles. L and Ac beads inpanels a and b are shown in orange and red, respectively, while theinitiator bead is in purple. EO and water beads are not shown forclarity. (c) DPD simulation of the effect of number of degradablelactide monomers per SPELA arm on fraction of initiators in aqueoussolution. (d) Effect of exposure of MSCs to photoinitiator (10 mg/mL) in 20 wt % SPEXA precursor solution on the fraction of viablecells as a function of HA type. P, G, L, C, and D in panel d representPEG (no HA, control group), SPEXA with glycolide, lactide, ε-caprolactone and p-dioxanone HA monomer, respectively. “star”indicates a statistically significant difference between the test groupand P. (e) Effect of number of degradable HA monomers per arm onsimulated Ac−Ac running integration number and (f) experimentalgelation time of 20% SPEXA precursor solutions. Error bars in panels cand e correspond to mean ± SD for 5 simulation runs. Error bars inpanels d and f correspond to mean ± SD for 3 experiments.

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282923

micelles’ core) in SPELA solutions is shown in Figure 5c.Fraction of initiators in the aqueous phase decreased from100% to 7.4, 3.3 and 2% when m increased from 0 to 1, 2 and 3,respectively. Partition of a large fraction of initiators to themicelles’ core confined the gelation reaction to the micellephase, thus potentially reducing the exposure of encapsulatedcells to toxic initiator molecules. MSCs in 20 wt % SPEXAsolutions were exposed to 10 mg/mL photoinitiator for 10 min,the cells were resuspended in basal medium (without initiator)and cultured for 2 days. Figure 5d compares the normalized cellnumbers for SPEXA solutions after exposure to photoinitiatorto that in SPEXA-m0 (acrylated star PEG without HA chain-extension). Groups included SPEGA (G-m1.6), SPELA (L-m1.7), SPEDA (D-m1.7), SPECA (C-m1.8) and PEG (m0).The cell number was significantly higher for SPELA (0.87 ±0.04) and SPECA (0.84 ± 0.05) solutions compared toSPEXA-m0 (0.77 ± 0.05), which can be explained bypartitioning of initiator molecules to the hydrophobic micellephase (see Figure 5c), whereby the exposure of encapsulatedcells to initiator molecules was reduced.Micelle formation sharply reduced the average interacrylate

and acrylate−initiator distance, leading to a significant increasein the rate of initiation and propagation of the acrylates. Toquantify the average interacrylate distance (related to cross-linking rate) or proximity of water beads to ester groups onSPEXA macromonomers (related to hydrolytic degradationrate), the average number of Ac (or W) beads in a sphere ofradius R around an Ac (or ester) bead or the runningintegration number of beads “a” around beads “b”, INab(R) wascalculated by47

∫πρ=R g r r rIN ( ) 4 ( ) dab b

R

ab0 0

2(11)

where ρb0 is overall number density of type “b” beads and gab(r)is radial distribution function of bead “b” around bead “a”,located at the origin. The running integration number of Ac−Ac beads in SPELA solutions (INAc−Ac) at R = rc (the DPDlength scale 6.74 Å, see Methods section) initially increasedwith increasing m from 0 to 2, as shown in Figure 5e, and thendecreased as m increased from 2 to 3. Conversely, a unimodalincrease in INAc−Ac was observed for SPEGA, SPECA, andSPEDA solutions with increasing m from 0 to 3. The increase inINAc−Ac with m was attributed to a decrease in average Ac−Acdistance in the micelles’ core. As the core of the micellescontinued to increase in size and their separation distancecontinued to increase for m > 2, the average distance betweenthe Ac beads began to increase, leading to a decrease in INAc−Ac,as predicted for SPELA in Figure 5e. Furthermore, SPELA andSPECA macromonomers had higher INAc−Ac than SPEGA andSPEDA for m ≤ 3. The predicted INAc−Ac values are related togelation time of the macromonomers in aqueous solution.Gelation time of 20 wt % SPEXA solutions was measured witha rheometer as a function of number of HA monomers per arm,and the results are shown in Figure 5f. Gelation times of SPELAand SPECA precursor solutions were shorter than those ofSPEGA and SPEDA, as predicted by simulation (see Figure5e). Gelation time of 20 wt % SPEGA, SPEDA, SPELA, andSPECA solutions decreased from 150 s to 61, 64, 28, and 34 s,respectively, with increasing m from 0 to 3 (see Figure 5f). Theinitial sharp decrease in gelation time was attributed toaggregate formation and an increase in INAc−Ac. Simulationresults in Figure 5e predict that the INAc−Ac value for SPELAshould decrease for m > 2. However, a decrease in INAc−Ac for

SPELA at higher m values was offset by an increase in residencetime of the arms in the micelle core, leading to no change ingelation time. Residence time of a hydrophobic segment in themicelle core is proportional to45,48

τ γ∼ ·n2/3(12)

where γ is effective interfacial tension between the hydrophobicdomains and water. As a result, residence time of the unreactedAc groups in the micelles’ core increased with n, whichincreased the rate of cross-linking; thereby gelation time wasreduced. Furthermore, fraction of bridging arms betweenmicelles increased with increasing residence time, which inturn increased the extent of physical gelation. Therefore,gelation time of the macromonomer solutions continued todecrease with increasing n.

Degradation of SPEXA Hydrogels. Degradation ofSPEXA hydrogels depended on the proximity of water beadsto ester links on HA segments. Local distribution of waterbeads around hydrophobic cores of SPEXA-n4 micelles isshown in Figure 6g. Water beads were in close proximity to Gbeads in SPEGA solution. The relatively small size of G coresalong with lower hydrophobicity of G beads compared to otherHA monomers led to a short average distance between G andW beads. When SPEGA was replaced with SPELA, the size ofthe micelles’ core increased and local concentration of W beadsaround the core decreased (Figure 6g, see image “L”), therebyincreasing the average L−W bead distance. A dip in theconcentration of W beads observed proximal to the micellecore in SPELA was attributed to the higher hydrophobicity of Lbeads, compared to G, leading to a higher packing of EO beadsat the core−water interface. Concentration of W beadsproximal to the micelles’ core increased by exchangingSPELA with SPEDA, but the core size in SPEDA wassignificantly larger than that in SPEGA. The lowestconcentration of W beads at the core margins was observedfor SPECA micelles where the higher hydrophobicity of Ccores compared to other HA monomers overcame the energyof chain extension and forced EO beads to undergo high-entropy packing at the core−water interface by repelling waterbeads from the interfacial layer.The running integration number of water beads around ester

links, INester−W, for SPEXA macromononers (data not shown)initially increased with the addition of one monomer toSPEXA-m0. Then, INester−W decreased for all SPEXA solutionswith m > 1 due to increase in micelle size and decrease in totalmicelle surface area. SPEGA and SPECA solutions had thehighest and lowest INester−W for all m values, respectively.However, degradation rate of SPEXA hydrogels depended onthe density of ester groups as well as proximity of ester groupsto water beads. Assuming that the formation of carboxylic acidgroups by ester dissociation did not affect hydrolysis rate (thisis a good assumption since degradation was performed in abuffered aqueous medium), the relative hydrolysis rate (P) atthe mesoscale scale, which is proportional to the rate ofdegradation at the macroscale, is defined by

= ×− −P IN INester W ester ester (13)

In the above equation, INester−W and INester−ester areproportional to the concentration of water and ester groupsin the micelles, respectively. The simulated relative hydrolysisrate in the reaction volume for 20 wt % SPEXA macro-monomers in aqueous solution is shown in Figure 6a as afunction of m. For all m values, SPEGA had the highest relative

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282924

hydrolysis rate followed by SPELA, SPEDA, and SPECA.Relative hydrolysis rate for SPECA and SPEDA solutionsincreased from 0 to 5.2 and 12.5, respectively, with increasing mfrom 0 to 4. Likewise, relative hydrolysis rate of SPELA andSPEGA solutions increased from 0 to 13.4 and 22.5,respectively, with increasing m from 0 to 3 and then decreasedto 12.3 and 21.0 with increasing m from 3 to 4. The relatively

large difference in predicted relative hydrolysis rates betweenSPEXA macromonomers for a given m indicated that hydrolysiswas related to equilibrium water content and concentration ofester groups in the micelles, not to the bulk waterconcentrations (the solutions had similar bulk water contents,see Figure 6h). The predicted biphasic hydrolysis rate forSPELA and SPEGA in Figure 6a was attributed to the lowproximity of water to ester beads in larger micelle cores athigher m values. SPECA with the most hydrophobic micelleshad the lowest predicted hydrolysis rate while SPEGA with theleast hydrophobic micelles had the highest hydrolysis rate. Massloss of 20 wt % SPELA, SPEGA, SPEDA, and SPECAhydrogels with incubation time for different m values are shownin Figures 6c−f. SPELA-0L without lactide chain extension had<5% mass loss after 6 weeks of incubation. Mass loss of SPELAgels was linear with incubation time for all m values. SPELAhydrogels lost 6%, 37%, 80% and 100% mass after 4 weeks as mincreased from 0 to 0.8, 1.6 and 2.9, respectively. However,SPELA mass loss decreased from 100% to 87% as m increasedfrom 2.9 to 3.7, which was consistent with the predicteddecrease in SPELA hydrolysis rate for 3 ≤ m ≤ 4 in Figure 6a(brown curve). SPEGA gels had the fastest degradation rateamong HA monomers. The time for complete degradation ofSPEGA gels initially decreased from 8 to 2 days with increasingm from 0.8 to 1.6. However, SPEGA degradation timeincreased from 2 to 3 days when m increased from 1.6 to 2.8mainly due to a transition from surface (controlled by thenumber of ester groups) to bulk degradation (domination bywater content of the micelles). The difference in experimental(1.6 < m < 2.8) and simulated (3 < m < 4) transition range forSPEGA was attributed to the polydispersity of G segments(1.24−1.44 for SPEGA compared with 1.17−1.34, 1.20−1.38and 1.22−1.38 for SPELA, SPEDA, and SPECA, respectively),leading to a wider distribution of micelle core sizes. Mass loss ofSPEDA and SPECA gels after 42 days ranged between 30−39%and 11−20% respectively. The nonlinear degradation trend forSPEDA was attributed to a wider micelle size distribution,which was not taken into account in the simulations. Theexperimentally measured mass losses for SPEXA hydrogels atsimilar m values (1.6 ≤ m ≤ 1.8) are compared in Figure 6b.SPEGA and SPELA hydrogels completely degraded in 3 daysand 5 weeks, respectively, whereas SPEDA and SPECAhydrogels lost 40% and 20% of their mass in 6 weeks.Equilibrium water content of SPEXA hydrogels as a function ofm is shown in Figure 6h. The difference in water content ofSPEXA gels was not statistically significant (p values for thedifference between water contents were >0.17). Therefore, thewide range of degradation rates from a few days to manymonths observed for SPEXA gels, as shown in Figure 6b, canonly be explained by differences in equilibrium water content ofthe micelles with HA type and number of HA monomers persegment.

Osteogenic Differentiation of Marrow Stromal Cells inSPEXA Hydrogels. MSCs were encapsulated in SPEXA gels,and the effect of HA monomer type on osteogenic differ-entiation of MSCs was evaluated. SPEGA gel due to itsrelatively fast degradation (see Figure 6d) was not used forMSC encapsulation. Groups included 20 wt % SPELA (L-m1.7), SPEDA (D-m1.7), SPECA (C-m.18) and PEG (m0).The effect of HA type on total collagen content with incubationin osteogenic medium is shown in Figure 7a. Total collagencontent of L-m1.7 gel increased from 54 μg/μg DNA at day 7to 83 and 122 μg/μg DNA at days 14 and 28, respectively.

Figure 6. (a) Effect of number of HA monomers per arm (m) onpredicted hydrolysis rate of 20 wt % SPEXA precursor solutions. (b)Effect of HA monomer type on experimental mass loss of SPEXAhydrogels with incubation time. (c−f) Effect of number of HAmonomers per macromonomner on experimental mass loss of SPELA,SPEGA, SPEDA, and SPECA hydrogels, respectively, with incubationtime. (g) Effect of degradable HA monomer type on distribution ofwater beads around core of the micelles. (h) Effect of number of HAmonomers per arm on experimentally measured equilibrium watercontent of SPEXA hydrogels. In panel a, error bars correspond tomean ± SD for five simulation runs. In panels b−f and h, error barscorrespond to mean ± SD for three experiments. In panel g, G, D, L,C, Ac, and water beads are shown by blue, pink, orange, purple, red,and light blue, respectively, and EO beads are not shown for clarity.

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282925

Total collagen content of L-m1.7 gel was significantly higherthan those of C-m1.8, D-m1.7 and m0 gels for all incubationtimes. The higher secretion of collagen by MSCs seeded in L-m1.7 gels is attributed to hydrolytic degradation of the matrixconcurrent with ECM production. Rate of ECM production isrelated to the rate of matrix degradation.13,49 ECM productionis limited by free volume when rate of matrix degradation isslower than that of ECM production. Conversely, ECMproduction is limited by cell adhesion when rate of matrixdegradation is faster than that of ECM production. ALP activityand extent of mineralization of MSCs in SPEXA gels are shownin Figure 7b,c, respectively. Consistent with previous results,ALP activity of all groups increased from day 7 to 14 and thendecreased after 28 days with mineralization.18,21 At day 14, ALPactivity of L-m1.7 was significantly higher than the othergroups. Calcium content of all groups had an increasing trendwith time. Calcium content of L-m1.7 increased from 10.3 ±1.2 to 40.8 ± 8.5 and 224.7 ± 18.5 mg/mg DNA withincubation time from day 7 to 14 and 28. After 28 days, calciumcontent of L-m1.7 was significantly higher than the othergroups. Furthermore, at day 28, calcium content of D-m1.7 andC-m1.8 was significantly higher than that of m0 group. It can beinferred from the results that degradation of SPEXA gelscontributed significantly to mineralization with L-m1.7 gelhaving 2.3-fold higher mineral deposition compared tonondegradable m0 (PEG gel). Higher extent of mineralizationof MSCs in degradable gels is consistent with previous reports.In one study, Human MSCs encapsulated in a hydrolyticallydegradable PEG matrix and incubated in osteogenic mediumdisplayed higher cell−cell contact and cell spreading comparedto nondegradable PEG.8,50 In another study, mineralization ofosteoblasts increased 3-fold with conjugation of lactidesegments in PEG dimethacrylate hydrogels.12

mRNA expression levels of Col-1, ALP and OC are shown inFigure 7d−f, respectively. Col-1 expression for all groups wassignificantly higher at day 28 compared to day 7. For example,Col-1 mRNA expression of L-m1.7 increased from 2.9 ± 0.7 to16.6 ± 1.7 and 45.2 ± 5.0 when incubation time increased fromday 7 to 14 and 28. At day 28, Col-1 mRNA expression of L-m1.7 was 2.5 fold higher than PEG gel. ALP mRNA expression(Figure 7e) showed a trend similar to its bioactivity in Figure7b. At day 14, ALP mRNA expression of L-m1.7 wassignificantly higher than the other gels and 2.2 times higherthan PEG gel. OC expression of all groups increased from day 7to 28, with L-m1.7 and C-m1.8 gels showing significantly higherOC expression at day 28 compared to D-m1.7 and PEG gels.Taken together, the findings of this work demonstrate thatchain extension of PEG hydrogels with short HA segmentsresulted in the formation of micellar hydrogels that supporteddifferentiation and mineralization of MSCs with a wide range ofdegradation rates from a few days to many months.

■ CONCLUSIONS

Structure formation and gelation kinetics of SPEXA macro-monomers in aqueous solution and degradation of SPEXA gelswere investigated by simulation and experimental measure-ment. In the absence of HA monomer, distribution of acrylategroups was uniform in aqueous solution, but the extension ofthe macromonomer arms with hydrophobic X segmentsinduced aggregation and micelle formation. Core radius ofSPEGA and SPEDA micelles increased from 0 to 22 Å whenthe number of HA repeating units per arm (n) increased from 2to 8 while that of SPELA and SPECA micelles increased from 9and 11 Å to 23 and 24 Å, respectively. Aggregation resulted inpartitioning of the polymerization photoinitiator to the micellephase, which confined the gelation reaction to the micelle

Figure 7. (a) Total collagen, (b) ALP activity, and (c) calcium content of MSCs encapsulated in SPEXA hydrogels with incubation time inosteogenic medium. mRNA expression of (d) Col-1, (e) ALP, and (f) OC of MSCs encapsulated in SPEXA hydrogels with incubation time inosteogenicc medium. “star” indicates statistically significant difference (p < 0.05) between the test group and all other groups at the same time point.Error bars correspond to mean ± SD for three experiments.

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282926

phase, thereby improving the viability of encapsulated MSCs.Micelle formation sharply reduced the average interacrylate andacrylate-initiator separation distance, leading to a significantincrease in gelation rates of SPEXA gels. For example, gelationtime of 20 wt % SPEGA, SPEDA, SPELA, and SPECAsolutions decreased from 150 s to 61, 64, 28, and 34 s,respectively, with increasing number of HA monomers per arm(m) from 0 to 3. The simulated hydrolysis rates of SPEXA gelswere strongly dependent on HA type and number of HA repeatunits, consistent with experimentally measured degradationrates. For all HA types, hydrolysis rate of SPEXA gels wasbiphasic. For example, as m increased from 0.7 to 1.2, 1.8, and2.8, mass loss of SPEGA gels after 2 days increased from 20% to46% and 80% and then decreased to 66%, respectively.Similarly, as m increased from 0.8 to 1.7, 2.9 and 3.7, massloss of SPELA gels after 3 weeks increased from 32% to 50%and 62% and then decreased to 46%. All SPEXA gels hadsimilar initial water contents irrespective of HA type and n. Abiphasic relationship between HA segment length and geldegradation rate indicated a transition from surface to bulkhydrolysis in the micelle phase. The micellar SPEXA hydrogelssupported osteogenic differentiation, collagen production, andmineralization of MSCs.

■ ASSOCIATED CONTENT

*S Supporting Information1H NMR spectra of star PEG, SPEGA (G-m2.8), SPEDA (D-m2.9), SPELA (L-m2.9) and SPECA (C-m2.8) macro-monomers. This material is available free of charge via theInternet at http://pubs.acs.org.”

■ AUTHOR INFORMATION

Corresponding Author*Address: Department of Chemical Engineering, SwearingenEngineering Center, Rm 2C11, University of South Carolina,Columbia, SC 29208. Tel: (803) 777-8022. Fax: (803) 777-0973. E-mail: jabbari@cec.sc.edu.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by research grants to E.J. from theNational Science Foundation under Grant Nos. CBET0756394and DMR1049381, the National Institutes of Health underGrant No. DE19180, and the Arbeitsgemeinschaft furOsteosynthesefragen (AO) Foundation under Grant No.C10-44J. The authors thank Dr. John Lavigne (ChemistryDepartment at USC) for the use of GPC system.

■ REFERENCES(1) Lutolf, M. R.; Weber, F. E.; Schmoekel, H. G.; Schense, J. C.;Kohler, T.; Muller, R.; Hubbell, J. A. Repair of bone defects usingsynthetic mimetics of collagenous extracellular matrices. Nat.Biotechnol. 2003, 21 (5), 513−518.(2) Seliktar, D. Designing cell-compatible hydrogels for biomedicalapplications. Science 2012, 336 (6085), 1124−1128.(3) Sharma, B.; Fermanian, S.; Gibson, M.; Unterman, S.; Herzka, D.A.; Cascio, B.; Coburn, J.; Hui, A. Y.; Marcus, N.; Gold, G. E.; Elisseeff,J. H. Human cartilage repair with a photoreactive adhesive-hydrogelcomposite. Sci. Transl. Med. 2013, 5 (167), 167ra6.(4) Joshi, A.; Fussell, G.; Thomas, J.; Hsuan, A.; Lowman, A.;Karduna, A.; Vresilovic, E.; Marcolongo, M. Functional compressive

mechanics of a PVA/PVP nucleus pulposus replacement. Biomaterials2006, 27 (2), 176−184.(5) Lapienis, G. Star-shaped polymers having PEO arms. Prog. Polym.Sci. 2009, 34 (9), 852−892.(6) Sarvestani, A. S.; He, X.; Jabbari, E. Viscoelastic characterizationand modeling of gelation kinetics of injectable in situ cross-linkablepoly(lactide-co-ethylene oxide-co-fumarate) hydrogels. Biomacromole-cules 2007, 8 (2), 406−415.(7) Seck, T. M.; Melchels, F. P. W.; Feijen, J.; Grijpma, D. W.Designed biodegradable hydrogel structures prepared by stereo-lithography using poly(ethylene glycol)/poly(D,L-lactide)-based resins.J. Controlled Release 2010, 148 (1), 34−41.(8) Cushing, M. C.; Anseth, K. S. Hydrogel cell cultures. Science2007, 316 (5828), 1133−1134.(9) He, X. Z.; Jabbari, E. Material properties and cytocompatibility ofinjectable MMP degradable poly(lactide ethylene oxide fumarate)hydrogel as a carrier for marrow stromal cells. Biomacromolecules 2007,8 (3), 780−792.(10) Klein, J. Repair or replacement - A joint perspective. Science2009, 323 (5910), 47−48.(11) Peppas, N. A.; Lustig, S. R. Solute diffusion in hydrophilicnetwork structures. In Hydrogels in Medicine and Pharmacy. I.Fundamentals; CRC Press: Boca Raton, FL, 2004(12) Benoit, D. S. W.; Durney, A. R.; Anseth, K. S. Manipulations inhydrogel degradation behavior enhance osteoblast function andmineralized tissue formation. Tissue Eng. 2006, 12 (6), 1663−1673.(13) Bryant, S. J.; Anseth, K. S. Hydrogel properties influence ECMproduction by chondrocytes photoencapsulated in poly(ethyleneglycol) hydrogels. J. Biomed. Mater. Res. 2002, 59 (1), 63−72.(14) Salinas, C. N.; Anseth, K. S. Mesenchymal stem cells forcraniofacial tissue regeneration: Designing hydrogel delivery vehicles. J.Dental Res. 2009, 88 (8), 681−692.(15) Bosnakovski, D.; Mizuno, M.; Kim, G.; Takagi, S.; Okumura,M.; Fujinaga, T. Chondrogenic differentiation of bovine bone marrowmesenchymal stem cells (MSCs) in different hydrogels: Influence ofcollagen type II extracellular matrix on MSC chondrogenesis.Biotechnol. Bioeng. 2006, 93 (6), 1152−1163.(16) Kundu, A. K.; Putnam, A. J. Vitronectin and collagen Idifferentially regulate osteogenesis in mesenchymal stem cells.Biochem. Biophys. Res. Commun. 2006, 347 (1), 347−357.(17) Aamer, K. A.; Sardinha, H.; Bhatia, S. R.; Tew, G. N.Rheological studies of PLLA−PEO−PLLA triblock copolymer hydro-gels. Biomaterials 2004, 25 (6), 1087−1093.(18) Moeinzadeh, S.; Barati, D.; He, X.; Jabbari, E. Gelationcharacteristics and osteogenic differentiation of stromal cells in inerthydrolytically degradable micellar polyethylene glycol hydrogels.Biomacromolecules 2012, 13 (7), 2073−2086.(19) Pollock, J. F.; Healy, K. E. Mechanical and swellingcharacterization of poly(N-isopropyl acrylamide-co-methoxy poly-(ethylene glycol) methacrylate) sol−gels. Acta Biomater. 2010, 6 (4),1307−1318.(20) Cha, C.; Liechty, W. B.; Khademhosseini, A.; Peppas, N. A.Designing biomaterials to direct stem cell fate. ACS Nano 2012, 6(11), 9353−9358.(21) He, X.; Ma, J.; Jabbari, E. Effect of grafting RGD and BMP-2protein-derived peptides to a hydrogel substrate on osteogenicdifferentiation of marrow stromal cells. Langmuir 2008, 24 (21),12508−12516.(22) Yang, F.; Williams, C. G.; Wang, D. A.; Lee, H.; Manson, P. N.;Elisseeff, J. The effect of incorporating RGD adhesive peptide inpolyethylene glycol diacrylate hydrogel on osteogenesis of bonemarrow stromal cells. Biomaterials 2005, 26 (30), 5991−5998.(23) Graf, N.; Bielenberg, D. R.; Kolishetti, N.; Muus, C.; Banyard, J.;Farokhzad, O. C.; Lippard, S. J. αVβ3 integrin-targeted PLGA−PEGnanoparticles for enhanced anti-tumor efficacy of a Pt(IV) prodrug.ACS Nano 2012, 6 (5), 4530−4539.(24) Jabbari, E.; Yang, X.; Moeinzadeh, S.; He, X. Drug releasekinetics, cell uptake, and tumor toxicity of hybrid VVVVVVKK

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282927

peptide-assembled polylactide nanoparticles. Euro. J. Pharm. Biopharm.2013, 84 (1), 49−62.(25) Kersey, F. R.; Zhang, G. Q.; Palmer, G. M.; Dewhirst, M. W.;Fraser, C. L. Stereocomplexed poly(lactic acid)−poly(ethylene glycol)nanoparticles with dual-emissive boron dyes for tumor accumulation.ACS Nano 2010, 4 (9), 4989−4996.(26) Saito, N.; Okada, T.; Horiuchi, H.; Murakami, N.; Takahashi, J.;Nawata, M.; Ota, H.; Nozaki, K.; Takaoka, K. A biodegradablepolymer as a cytokine delivery system for inducing bone formation.Nat. Biotechnol. 2001, 19 (4), 332−335.(27) Xiao, L.; Xiong, X. Q.; Sun, X. H.; Zhu, Y. H.; Yang, H.; Chen,H. B.; Gan, L.; Xu, H. B.; Yang, X. L. Role of cellular uptake in thereversal of multidrug resistance by PEG-b-PLA polymeric micelles.Biomaterials 2011, 32 (22), 5148−5157.(28) Zhang, Z.; Xiong, X. Q.; Wan, J. L.; Xiao, L.; Gan, L.; Feng, Y.M.; Xu, H. B.; Yang, X. L. Cellular uptake and intracellular traffickingof PEG-b-PLA polymeric micelles. Biomaterials 2012, 33 (29), 7233−7240.(29) Zhao, Z. X.; Gao, S. Y.; Wang, J. C.; Chen, C. J.; Zhao, E. Y.;Hou, W. J.; Feng, Q.; Gao, L. Y.; Liu, X. Y.; Zhang, L. R.; Zhang, Q.Self-assembly nanomicelles based on cationic mPEG-PLA-b-polyarginine(R-15) triblock copolymer for siRNA delivery. Biomate-rials 2012, 33 (28), 6793−6807.(30) Moeinzadeh, S.; Jabbari, E. Mesoscale simulation of the effect ofa lactide segment on the nanostructure of star poly(ethylene glycol-co-lactide)-acrylate macromonomers in aqueous solution. J. Phys. Chem. B2012, 116 (5), 1536−1543.(31) Espanol, P.; Warren, P. Statistical-mechanics of dissipativeparticle dynamics. Europhys. Lett. 1995, 30 (4), 191−196.(32) Hoogerbrugge, P. J.; Koelman, J. M. V. A. Simulatingmicroscopic hydrodynamic phenomena with dissipative particledynamics. Europhys. Lett. 1992, 19 (3), 155−160.(33) Groot, R. D.; Warren, P. B. Dissipative particle dynamics:Bridging the gap between atomistic and mesoscopic simulation. J.Chem. Phys. 1997, 107 (11), 4423−4435.(34) Groot, R. D.; Madden, T. J. Dynamic simulation of diblockcopolymer microphase separation. J. Chem. Phys. 1998, 108 (20),8713−8724.(35) Materials Studio V 5.5; Accelrys Inc: San Diego, CA, 2011.(36) Sun, H. COMPASS: An ab initio force-field optimized forcondensed-phase applications - Overview with details on alkane andbenzene compounds. J. Phys. Chem. B 1998, 102 (38), 7338−7364.(37) Moeinzadeh, S.; Khorasani, S. N.; Ma, J..; He, X.; Jabbari, E.Synthesis and gelation characteristics of photo-crosslinkable starpoly(ethylene oxide-co-lactide-glycolide acrylate) macromonomers.Polymer 2011, 52 (18), 3887−3896.(38) Sarvestani, A. S.; Xu, W. J.; He, X.; Jabbari, E. Gelation anddegradation characteristics of in situ photo-crosslinked poly(L-lactid-co-ethylene oxide-co-fumarate) hydrogels. Polymer 2007, 48 (24),7113−7120.(39) Maniatopoulos, C.; Sodek, J.; Melcher, A. H. Bone formation invitro by stromal cells obtained from bone marrow of young adult rats.Cell Tissue Res. 1988, 254 (2), 317−330.(40) Chen, K. L.; Sahoo, S.; He, P. F.; Ng, K. S.; Toh, S. L.; Goh, J.C. H. A hybrid silk/RADA-based fibrous scaffold with triple hierarchyfor ligament regeneration. Tissue Eng., Part A 2012, 18 (13−14),1399−1409.(41) He, X.; Yang, X.; Jabbari, E. Combined effect of osteopontin andBMP-2 derived peptides grafted to an adhesive hydrogel on osteogenicand vasculogenic differentiation of marrow stromal cells. Langmuir2012, 28 (12), 5387−5397.(42) Henderson, J. A.; He, X.; Jabbari, E. Concurrent differentiationof marrow stromal cells to osteogenic and vasculogenic lineages.Macromol. Biosci. 2008, 8 (6), 499−507.(43) Pfaffl, M. W. A new mathematical model for relativequantification in real-time RT-PCR. Nucleic Acids Res. 2001, 29 (9),e45.

(44) Nagarajan, R.; Ganesh, K. Block copolymer self-assembly inselective solvents - Spherical micelles with segregated cores. J. Chem.Phys. 1989, 90 (10), 5843−5856.(45) Nicolai, T.; Colombani, O.; Chassenieux, C. Dynamic polymericmicelles versus frozen nanoparticles formed by block copolymers. SoftMatter 2010, 6 (14), 3111−3118.(46) Meng, X. X.; Russel, W. B. Structure and size of sphericalmicelles of telechelic polymers. Macromolecules 2005, 38 (2), 593−600.(47) Fioroni, M.; Burger, K.; Mark, A. E.; Roccatano, D. Model of1,1,1,3,3,3-hexafluoro-propan-2-ol for molecular dynamics simulations.J. Phys. Chem. B 2001, 105 (44), 10967−10975.(48) Li, Z. L.; Dormidontova, E. E. Equilibrium chain exchangekinetics in block copolymer micelle solutions by dissipative particledynamics simulations. Soft Matter 2011, 7 (9), 4179−4188.(49) Bryant, S. J.; Anseth, K. S. Controlling the spatial distribution ofECM components in degradable PEG hydrogels for tissue engineeringcartilage. J. Biomed. Mater. Res., Part A 2003, 64A (1), 70−79.(50) Nicodemus, G. D.; Bryant, S. J. Cell encapsulation inbiodegradable hydrogels for tissue engineering applications. TissueEng., Part B 2008, 14 (2), 149−165.

Biomacromolecules Article

dx.doi.org/10.1021/bm4008315 | Biomacromolecules 2013, 14, 2917−29282928