In situ forming, resorbable graft copolymer hydrogels providingcontrolled drug release

Derek J. Overstreet,1 Richard Huynh,1 Keith Jarbo,2 Ryan Y. McLemore,2 Brent L. Vernon1

1Center for Interventional Biomaterials, School of Biological and Health Systems Engineering, Arizona State University,

Tempe, Arizona 85287-97092Orthopaedic Residency, Banner Good Samaritan Medical Center, Phoenix, Arizona 85006

Received 26 April 2012; revised 30 July 2012; accepted 28 August 2012

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/jbm.a.34443

Abstract: In situ forming hydrogels are promising drug deliv-

ery vehicles due to their ease of delivery as liquids and their

ability to be used in sites with irregular geometries. In this

work, we report on in situ forming, resorbable hydrogels

based on N-isopropylacrylamide (NIPAAm) as a fluid-like con-

trolled release gel. These gels are the first resorbable

NIPAAm-based gels providing controlled release without rely-

ing on affinity between the drug and device. Therefore, these

gels provide a more flexible delivery system which can be

used to deliver any drug at a controlled rate. The polymers

contain repeat units of NIPAAm with (R)-a-Acryloyloxy-b,b-di-methyl-c-butyrolactone (DBLA) and varying amounts of

hydrophilic JeffamineVR

M-1000 acrylamide (JAAm) grafts.

The graft copolymer architecture allows the water content of

the hydrogels to be tuned over a wide range while keeping

the initial gelation temperature below body temperature.

Incorporation of JAAm in the polymers led to greater water

content, faster gel degradation, and reduced burst release. Sus-

tained release of the antimicrobial drugs cefazolin and vanco-

mycin (over about 5 and 7 days, respectively) was observed

from gels containing an intermediate amount of grafts which

combined reduced phase separation with a degradation time of

40 days. The degradation byproducts of one hydrogel formula-

tion were cytocompatible to NIH 3T3 fibroblasts at concentra-

tions up to 2.5 wt %. This class of terpolymer hydrogels is a

promising local delivery system for a wide variety of drugs, par-

ticularly for applications involving irregular geometries such as

implant interfaces. VC 2012 Wiley Periodicals, Inc. J Biomed Mater Res

Part A: 00A:000–000, 2012.

Key Words: N-isopropylacrylamide, controlled release, graft

copolymer, antimicrobials, injectable

How to cite this article: Overstreet DJ, Huynh R, Jarbo K, McLemore RY, Vernon BL. 2012. In situ forming, resorbable graftcopolymer hydrogels providing controlled drug release. J Biomed Mater Res Part A 2012:00A:000–000.

INTRODUCTION

Currently, injectable microsphere suspensions based on pol-y(lactic-co-glycolic acid) (PLGA) provide one of the mainclinically available degradable drug delivery vehicles.1,2 Insitu forming hydrogels share this system’s advantages ofinjectability and potential for controlled release, but canundergo degradation much more uniformly with reducedgeneration of charges.3–5 Hydrogels also provide viscoelasticproperties that cannot be replicated with particulate sys-tems, which may enable their use in additional applicationsin areas under cyclic mechanical loads, since wear debrisfrom harder materials in joint spaces can lead to local toxic-ity.6,7 Because hydrogels contain a higher fraction of watercompared to hard nonhydrogel materials (e.g., PLGA, hy-droxyapatite), they offer potential for faster hydrolytic deg-radation. The slow degradation of hard materials may alsobe a drawback when administration of a second dose isdesired, or in orthopedic applications, as bone healingoccurs within 6–8 weeks.8,9 Using soft in situ forminghydrogels with tunable degradation may therefore enable

drug delivery in applications which are not amenable tosustained release carriers in currently approved products,especially on implant surfaces.

Temperature-responsive hydrogels based on N-isopropy-lacrylamide (NIPAAm) are particularly well-suited to drugdelivery applications due to rapid physical crosslinkingwhich takes place in physiological conditions without chemi-cal reaction. Poly(NIPAAm) has a sharp lower critical solu-tion temperature (LCST) in aqueous media near 30�C.10–12

As a sufficiently concentrated solution of poly(NIPAAm) isheated above the LCST, a hydrogel is formed. Copolymers ofNIPAAm can be made to be degradable by incorporatingcomonomers with side chains which become more hydrophilicupon degradation either by water3–5,13,14 or specificenzymes,15 causing an increase in LCST. When the LCSTincreases to above body temperature, the polymer redissolves.

Effective controlled drug delivery from NIPAAm-basedhydrogels is limited primarily by high burst release follow-ing gelation. As a poly(NIPAAm)-based gel forms as a resultof physical crosslinking, a homogeneous polymer-rich phase

Correspondence to: B. L. Vernon; e-mail: brent.vernon@asu.edu

Contract grant sponsor: National Science Foundation GK-12 Down to Earth Science Graduate Fellowship

VC 2012 WILEY PERIODICALS, INC. 1

(i.e., the gel) separates from a fraction of the original sol-vent and then tends toward an equilibrium composition.10,16

Macroscopically, this is observed as shrinking and expulsionof solvent. While the polymer solutions tend to be formu-lated at 30 wt % or less for ease of handling, the final poly-mer concentration in the gel is about 50 wt %.17–20 Uponphase separation, hydrophilic drugs are prone to distributepredominantly into the solvent phase rather than the gelphase. As a result of shrinkage and drug distribution intothe expelled solvent, most of the payload of both low molec-ular weight hydrophilic drugs21 or proteins22–24 is typicallyreleased quickly and completely (within 3 h or less) uponthe phase transition of NIPAAm-based gels.

Several methods have been reported for slowing or con-trolling drug release from temperature-responsive materials.However, these methods often involve covalently cross-linked, non-injectable gels.21,22,25,26 Electrostatic interactionbetween the polymer and drug27 and covalent conjugationof the drug by a labile linkage5 have been demonstrated forprolonged protein release from resorbable gels, but theseapproaches are severely limited because they only are com-patible with drugs having specific chemistries. It would bemore desirable to have an injectable degradable hydrogelwhich could provide controlled release of any drug, simplyby mixing the drug into the polymer solution prior to use.

In this work, we demonstrate the first resorbableNIPAAm-based gel capable of providing controlled releaseon its own, without requiring covalent crosslinking, intermo-lecular interactions, covalent conjugation of drug, or sepa-rately prepared embedded particles. Such a system wouldbe simple to reliably manufacture and use and would havepotential for a variety of local or sustained drug deliveryapplications. In particular, these materials may be especiallyuseful for drug-eluting device coatings or in joint spaces,where hard materials or particulate systems are limited. Wehave previously demonstrated that incorporating hydro-philic grafts of JeffamineV

R

M-1000 on a poly(NIPAAm) back-bone yields weak, viscoelastic physical gels which exhibitcontrolled swelling and slow protein release.24 Here, wehypothesized that using JeffamineV

R

M-1000 grafts in aresorbable copolymer gel would reduce burst release upongelation and, instead, cause any entrapped drugs to be sub-sequently released by diffusion or hydrogel degradation.These copolymers were synthesized and characterized toevaluate the effect of graft content on hydrogel propertiesand the feasibility of using these gels for delivery of lowmolecular weight drugs was evaluated.

MATERIALS AND METHODS

MaterialsAll materials were reagent grade and obtained from Sigma-Aldrich unless otherwise noted. NIPAAm was recrystallizedfrom hexane. Azobisisobutyronitrile (AIBN) was recrystal-lized from methanol. HPLC grade tetrahydrofuran (THF)was used for polymerizations and as the mobile phase formolecular weight determination. JeffamineV

R

M-1000 wasdonated by Huntsman Corporation (The Woodlands, TX,USA). JeffamineV

R

M-1000 is a random copolymer of ethylene

oxide (EO) and propylene oxide (PO) in a 19:3 EO:PO ratioand about 1000 Da molecular weight, with one methoxyend group and one primary amine end group.

SynthesisJeffamineV

R

M-1000 acrylamide (JAAm) was synthesizedfrom JeffamineV

R

M-1000 polyetheramine by acryloylationwith acryloyl chloride.28 JeffamineV

R

M-1000 (20 g, 20mmol) was dissolved at 10 w/v% in dichloromethane (DCM)along with triethylamine (3.3 mL, 24 mmol) and maintainedat 0�C under nitrogen atmosphere. Acryloyl chloride (1.95mL, 24 mmol) was then added dropwise into the solutionunder stirring and the reaction was allowed to proceed for 1h at 0�C and then at room temperature for at least 5 h undernitrogen atmosphere. Following the reaction, DCM was evapo-rated and the remaining residue was dissolved in 100 mMNaHCO3. The product was extracted into DCM and the organicphase purified by column chromatography. After evaporationof DCM, JAAm was obtained as a viscous clear to light yellowliquid. JAAm was solidified by cooling on ice, vacuum driedovernight, then collected and stored at �20�C until use.

Copolymers of NIPAAm, (R)-a-Acryloyloxy-b,b-dimethyl-c-butyrolactone (DBLA), and JAAm were synthesized by rad-ical polymerization with 80% dioxane / 20% THF as thesolvent. This solvent blend was chosen to obtain a weight-average molecular weight near 40 kDa. The molar amountof DBLA relative to NIPAAm was held fixed at 92.8:7.2, asthis was shown in previous work to provide a material withan LCST after degradation that exceeds 37�C.3 JAAm was fedinto the reaction as 0, 15, or 30% of the mass relative toNIPAAm. Monomer solution (10 w/v%) was bubbled withnitrogen for at least 20 min prior to addition of initiator. Afterthis time, AIBN was added (0.007 mol AIBN/mol of total mono-mer) and the reaction allowed to proceed for 18 h at 65�Cunder nitrogen. Copolymers were collected by precipitation into10- to 15-fold excess of chilled diethyl ether. Following precipi-tation, the product, poly(NIPAAm-co-DBLA-co-JAAm) (abbrevi-ated PNDJ) was collected by filtration and dried overnightunder vacuum. The product was then dissolved in deionizedwater and dialyzed against 3500 MWCO at 4�C for 20 h withthe excess water replaced three times. Materials were then ly-ophilized and stored under nitrogen at�20�C until use.

1H NMR spectroscopy1H NMR (Varian Inova, 300 MHz) was used to confirm suc-cessful synthesis and determine the chemical composition ofthe synthesized polymers. CDCl3 was used as the NMR solvent.

Molecular weight and polydispersity determinationThe molecular weight and polydispersity of the synthesizedpolymers was determined by gel permeation chromatogra-phy (Shimadzu Corporation) in conjunction with static lightscattering (MiniDawn, Wyatt Technology Corporation) withTHF as the mobile phase.

Cloud point determinationSynthesized copolymers were dissolved at 0.25 wt % in 150mM PBS (pH 7.4) and analyzed for LCST properties by cloud

2 OVERSTREET ET AL. RESORBABLE TEMPERATURE-RESPONSIVE HYDROGELS

point determination. None of the polymer solutions satu-rated the detector when heated above the LCST. Cuvettes con-taining the polymer solutions were allowed to reach thedesired temperature in a water bath for at least 90 s prior toeach measurement. Absorbance at 450 nm was measured ev-ery 1�C by a UV/Vis spectrometer throughout the range ofthe LCST transition with buffer alone as the reference. Somepolymer solutions reached a temperature at which the ab-sorbance began to decrease due to precipitation and settlingout of polymer particles from the solution. In this case, thelast value of absorbance before observed aggregation wasrecorded as the maximum value and all previous values werenormalized relative to this maximum value. Absorbance val-ues for polymers that did not show a decrease in absorbancewith heating were normalized to the absorbance at 55�C.

DegradationSolutions of PNDJ15 and PNDJ30 were dissolved at 5 wt %in 150 mM PBS (pH 7.4) and several samples of each wereincubated separately at 37�C. After various time points, singlesamples of polymer solution were removed and frozen at�20�C to stop degradation. The study was stopped for eachmaterial after the polymer solution was completely transpar-ent at 37�C. After the end of the study, all of the frozen sam-ples were thawed, diluted to 0.25 wt %, and measured forLCST by cloud point determination as described above. LCSTis reported as the temperature at which the absorbance wasat least half of the maximum observed. The degradationkinetics of PND have been reported previously.3

RheometryHydrogels (30 wt %) were characterized by rheometry. Foreach run, about 400 lL of polymer solution was placedbetween the flat 25 mm plates of a rheometer (Anton PaarMCR-101), with a gap height of 0.5 mm. For frequencysweeps, polymer solutions were placed on the rheometer at20�C and then heated to 37�C for 60 s before measure-ments were taken. Because PND gels underwent significantshrinking throughout the first 15 min after gelation, PNDgels of similar size to the other gel samples were quicklyplaced on the rheometer stage after being incubated inexcess PBS at 37�C for 24 h to allow the gels to reach equi-librium. Gels were evaluated at 37�C with normal forcemaintained throughout the measurement at 100 6 50 mNand a humidity chamber placed over the samples to reduceevaporation. The gels were then subjected to 1% oscillatorystrain and the frequency varied from 0.1 to 10 Hz. For tem-perature sweeps, 400 lL of polymer solution was placed onthe rheometer stage maintained at 10�C. The gels were eval-uated under 1% oscillatory strain at 1 Hz frequency as therheometer stage was heated to a final temperature of 50�Cat a rate of 2�C/min with a humidity chamber used asabove.

In vitro drug releaseRelease of model drugs cefazolin (454.5 g/mol) and vanco-mycin (1449.3 g/mol) from copolymer gels was evaluatedat 37�C. The model drugs were chosen because they can be

easily measured by UV spectrophotometry and have differ-ent hydrophilicity. The hydrophilicity of a drug can substan-tially affect its release kinetics.29 Cefazolin exhibits consider-able water solubility and is available as the sodium salt.Vancomycin is much less soluble at neutral pH, and is usedin clinical preparations as the hydrochloride salt. In the PBS(pH 7.4) used both for hydrogel preparation and as therelease medium, the solubilities of cefazolin and vancomycinwere determined to be 28 and 2.5 wt %, respectively.

Polymer solutions were prepared at 30 wt % andallowed to dissolve overnight at 4�C. Gel samples were pre-pared by measuring one gram of polymer solution into a 20mL glass vial (inner diameter � 25 mm, resulting in gelthickness of � 2 mm), and then the desired amount of drug(either 5 or 50 mg) was added to the solution and vortexmixed to dissolve or suspend the drug. The mass of drug(60.1 mg) loaded into each sample was recorded due tosome error in adding 5.0 mg of drug to each low-dose sample.Only the low dose cefazolin-loaded polymer solutions weretransparent after mixing; the high dose samples containedevenly distributed but insoluble drug. Gels were formed byincubation in a 37�C water bath for 15 min. After 15 min,vials were filled to the top with 20 mL prewarmed PBS andmaintained at 37�C. PND hydrogels evaluated for releaseused only polymer batch PNDb, while the batches of PNDJ15and PNDJ30 were used interchangeably for various studies.

At selected time points, aliquots were removed andstored at �20�C, and the release buffer was completelyreplaced with prewarmed PBS to maintain infinite sink con-ditions. Drug concentration in the aliquots was measured atthe end of the study by UV spectrophotometry (BMG Lab-tech, Fluostar Omega) at 280 nm for vancomycin and 272nm for cefazolin. Calibration curves for each drug were usedon each plate. It was verified that degraded polymers exhib-ited minimal absorbance at 1.5 wt % (corresponding tocomplete gel dissolution in the release medium in a singletime point) at the wavelengths used for drug detection.

In vitro cytocompatibilityThe cytocompatibility of degraded poly(NIPAAm-co-DBLA-co-JAAm) (batch PNDJ30a) was characterized at variouspolymer concentrations between 1 and 25 mg/mL. We thinkthese concentrations are likely to adequately represent therange of concentration of soluble polymer byproducts towhich cells in local tissues would be exposed for a sus-tained period of time (i.e., days). Studies on similar andlower concentrations have been reported in the literature toassess the cytocompatibility of the soluble precursors ordegradation byproducts of resorbable hydrogel materi-als.30,31 Polymer was dissolved at 10 wt% in PBS titrated topH 12 with 1 N NaOH and degraded by stirring for 3 daysat room temperature with the pH titrated to pH 12 twicedaily. Following degradation, the polymer solution was dia-lyzed for 5 days against deionized water (3500 MWCO)with frequent water changes, and then lyophilized. Degrada-tion was confirmed by 1H NMR. Lyophilized polymer andglass scintillation vials were then treated with ethanol andexposed to UV light in a laminar flow hood for at least 1 hr.

ORIGINAL ARTICLE

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH A | MONTH 2012 VOL 00A, ISSUE 00 3

Then the polymer was dissolved at various concentrationsin sterile DMEM supplemented with penicillin (100 units/mL), streptomycin (100 lg/mL), and L-alanyl-L-glutamine (2mM). Solutions were then sterile filtered using a 0.45 lmfilter and placed into scintillation vials. NIH3T3 mousefibroblasts were seeded at 10,000 cells/well in separate 24well plates and allowed to adhere for 24 hr (37�C, 5% CO2) in500 lL control media (media described above þ 10% bovinecalf serum) per well. The media was then replaced with eitherdegraded polymer solution (n ¼ 4 for each concentration) orcontrol media without polymer. Calf serum (50 lL) wasadded to the polymer-media solutions in each well individu-ally. NIH3T3 cultures were analyzed after 3 days. Cells werethen rinsed with PBS and evaluated for live cell number usingthe CellTiter 96VR AQueous Cell Proliferation Assay (Promega,Inc.) according to the manufacturer’s instructions. Beforeevaluation of cell number, cells were imaged by light micros-copy to observe any differences in cell morphology.

RESULTS

Polymer compositionThe polymer feed ratios, composition, molecular weight, ini-tial LCST as measured by cloud point, and degradation timeare shown in Table I. Polymer batches are abbreviated interms of the JAAm weight fraction in the feed (0, 15, or 30wt %). Additional batches of polymer made using the samefeed ratios are denoted by a letter. The molecular weightdistributions of all the polymer batches were relatively con-sistent, with number-average molecular weight between 13and 28 kDa and weight-average molecular weight between28 and 54 kDa. Because the polymer backbone remainsintact after degradation of these materials, their molecularweight should be near or below 40 kDa in order to facilitaterapid clearance through the kidneys.32

Successful synthesis was confirmed by 1H NMR. JAAmcontent in the copolymers was calculated from the integra-tion ratios of the peak at 3.5 ppm ascribed to the oxyethy-lene protons of the EO units (CH2CH2O) of JAAm relative tothe peak at 3.7 ppm (1H) of the lone isopropyl proton ofNIPAAm (CH(CH3)2). DBLA content was determined usingthe peaks near 5.3–5.5 ppm (1H) of the lone proton on theacryloyloxy-substituted carbon of the butyrolactone ringgroup. All copolymers exhibited lower incorporation ofJAAm relative to the feed ratio, and DBLA content was also

slightly lower than feed. Batch-to-batch variability in contentwas low.

LCST properties and degradationThe initial LCST of all synthesized materials was between18 and 27�C, with JAAm incorporation being positively cor-related with LCST. Solutions of these polymers will thus beflowable at ambient or slightly cooler temperatures and willgel at body temperature. Degradation kinetics of PNDJ30aand PNDJ15a at 5 wt % are shown in Figure 1. Similar dataon PND copolymers without JAAm have been previouslyreported by our lab using an accelerated degradation study3

due to the longer degradation time of these polymers.Arrhenius kinetics suggest a degradation time for PND of �180 days at 37�C. In this work, degradation time of thecopolymers varied greatly with JAAm content, with JAAmleading to faster degradation. PND gels exhibit an initiallyslow increase in LCST, followed by a linear increase in LCSTas the hydrophilicity of gels increases.3,33 A similar butfaster process was observed for PNDJ15, with minimal LCSTchange over the first two weeks of degradation, followed bya linear (R2 ¼ 0.995) increase in LCST with time betweenday 18 and day 54, and a final degradation time of about

TABLE I. Selected Properties of Degradable Temperature-Responsive Polymers Poly(NIPAAm-co-DBLA-co-JAAm)

Polymer

Feed Ratio(mol %)

Compositiona

(mol %)Molecular Weight

(g/mol)Initial LCST

(�C)Degradation Time

(37�C, pH 7.4)DBLA JAAm DBLA JAAm Mn Mw

PNDa 7.2 – 5.6 0.0 13,100 32,100 18 –PNDb 7.2 – 6.4 0.0 14,890 28,940 19 � 180 daysb

PNDJ15a 7.1 1.7 6.8 1.2 28,000 54,440 21 40 daysPNDJ15b 7.1 1.7 6.6 1.2 19,060 36,980 21 –PNDJ30a 6.9 4.1 6.4 2.8 19,100 35,330 27 9 daysPNDJ30b 6.9 4.1 6.4 2.8 19,820 49,680 27 –

a Measured by 1H NMR.b Estimated from accelerated degradation reported previously.3

FIGURE 1. Degradation kinetics of poly(NIPAAm-co-DBLA-co-JAAm)

hydrogels at 5 wt % (PBS, pH 7.4). Increased JAAm content resulted in

higher initial LCST and faster degradation. [Color figure can be viewed

in the online issue, which is available at wileyonlinelibrary.com.]

4 OVERSTREET ET AL. RESORBABLE TEMPERATURE-RESPONSIVE HYDROGELS

40 days. PNDJ30a exhibited a linear (R2 ¼ 0.992) increasein LCST throughout the first 14 days of degradation.

The degradation time of this system at 5 wt % can thus betuned between about 9 and 180 days depending on JAAm con-tent. This agrees well with previous work showing that JAAmfunctions to retain water (reducing shrinkage) within the poly-mer-rich ‘‘gel’’ phase at temperatures above the LCST.24 Asmore water is retained within the polymer-rich phase, the rateof hydrolysis increases. While these polymers did not form gelsat 5 wt%, we found that 30 wt% gels, including those used inrelease studies, exhibited degradation kinetics similar to theresults reported in Figure 1. Whereas 30 wt % p(ND) gelsshowed no marked change in appearance in over 15 weeks at37�C, PNDJ15b gels degraded over about 4–5 weeks at 37�C,and PNDJ30a gels with 2.8 mol % JAAm (about 21 wt %)became translucent within as little as 3 days.

Inclusion of JAAm grafts in the polymer design leads tomore uniform degradation throughout the resulting hydro-gel (Fig. 2). Gels of 30 wt % PNDJ30 maintained their initialvolume (or slightly increase in volume) and underwentgradual degradation. Gels of 30 wt % PND that have lowequilibrium water content underwent rapid shrinking andslower degradation (Fig. 2, top row). Interestingly, PNDJgels degrade over a similar period of time but in a muchdifferent process than poly(NIPAAm-co-DBLA-co-acrylicacid) (PNDA) gels. Whereas PNDJ gels exhibit degradationwhich appears to be approximately uniform, the degradationof PNDA gels (which shrink) has been shown to occur pri-marily at the surface, leading to re-swelling of the peripheryof the gel and formation of an inward-moving front betweenthe swollen, translucent outer phase and the shrunken innergel core.33,34 After the swelling front reaches all parts of the

PNDA gel, the swollen phase then disintegrates to yield asolution.33

Rheological propertiesRheological properties of 30 wt % hydrogels at 37�C as afunction of frequency are shown in Figure 3. All of the gelswere viscoelastic above the LCST, with phase angles greater

FIGURE 2. Swelling and degradation behavior of 30 wt % PND (top row) and PNDJ30 (bottom row) hydrogels incubated in PBS (pH 7.4) at

37�C. Gels were 1 mL with 1.5 mL excess PBS added after gelation. PND gels shrink and expel solvent and do not degrade within 13 days,

whereas PNDJ30 gels maintain their initial volume, but dissolved in the excess PBS when buffer was replaced at 13 days. PBS was replaced at

1, 3, 7, and 13 days. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 3. Frequency sweep showing the storage (solid icons) and

loss (open icons) moduli of 30 wt % gels of PNDJ30a (circles),

PNDJ15b (triangles), and PNDb (squares) at 37�C. PNDJ gels are

weaker and more sensitive to frequency than PND gels. [Color figure

can be viewed in the online issue, which is available at

wileyonlinelibrary.com.]

ORIGINAL ARTICLE

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH A | MONTH 2012 VOL 00A, ISSUE 00 5

than 45� (i.e., the gels are more viscous than elastic). Gelscontaining JAAm had storage and loss moduli in the 100–500 Pa range at 1 Hz, while PND gels were considerablystronger, particularly at lower frequencies. All of the gelsshowed increasing modulus with frequency which is charac-teristic of physical gels.14,24,35

Rheological properties of 30 wt % hydrogels as a functionof temperature are shown in Figure 4. All of the gels wereviscoelastic and PNDJ gels showed G00 > G0 over the rangetested. PND in both the sol and gel state had the highest mod-uli. PNDJ15 and PNDJ30 gels were both considerably weakerin both the sol and gel states, but both showed approximately50-fold increase in complex modulus when heated from a solto a gel. PNDJ15 became stronger over the temperature range15–21�C, whereas PNDJ became stronger over 20–40�C. Theslight decrease in modulus observed at temperatures above37�C for PNDJ gels may have to do with stronger physicalcrosslinking at these temperatures leading to shrinking andpoor contact with the rheometer resulting in a measurementof lower modulus, a finding consistent with other gels whichdo not shrink at 37�C but do shrink at higher temperatures.27

In vitro drug releaseDrug release profiles from 2 mm-thick discs of PND,PNDJ15, and PNDJ30 hydrogels are shown in Figure 5. Data

are reported as the cumulative amount of drug released(Mt) normalized to the cumulative amount of drug releasedafter 168 h (M168 hr), because the release rate by 168 h wasvery slow for all hydrogels tested. The percentage of the

FIGURE 4. Temperature sweep showing the storage (solid icons) and

loss (open icons) moduli of 30 wt % gels of PNDJ30a (circles), PNDJ15b

(triangles), and PNDb (squares) during heating. PNDJ solutions form less

viscous solutions below the LCST and less strong gels above the LCST

compared to PND. Both PNDJ solutions increase in modulus by about

50-fold through the sol–gel transition. [Color figure can be viewed in the

online issue, which is available at wileyonlinelibrary.com.]

FIGURE 5. Release of model drugs from 2 mm thick discs of resorb-

able temperature-responsive hydrogels (30 wt %). Release is depend-

ent on JAAm graft content, drug, and drug loading. Data are reported

as mean 6 s.d. (n ¼ 3). [Color figure can be viewed in the online

issue, which is available at wileyonlinelibrary.com.]

6 OVERSTREET ET AL. RESORBABLE TEMPERATURE-RESPONSIVE HYDROGELS

loaded drug that was released from each material within168 h is reported in Table II. For each hydrogel/drug com-bination, at least 65% of the drug loaded into the gels wasreleased within 168 h. Cumulative release of cefazolin wasover 85% from PNDJ gels, and as low as 65% from PNDgels, whereas only 70–85% of vancomycin was measured ashaving been released from any of the gel formulations.Incomplete release over this timeframe may be due to somecombination of drug retention within the gel or degradationof the drugs—vancomycin36,37 and cefazolin38,39 both areknown to be unstable under conditions similar to thoseused in these studies. PNDJ30 gels became translucent by72 h and were dissolved in the release media by 120 or168 h. Both PNDJ15 and PND gels remained intact through-out the studies.

When loaded with 5 mg/mL cefazolin, PND gels pro-vided high burst release, with 94% release observed within1 h after gelation. Gels with JAAm showed reduced burstrelease (under 40% within 1 h) but over 85% release after24 h. When loaded with 50 mg/mL cefazolin, similar releaseprofiles were observed both from PND and PNDJ30, withslower release observed over about 5 days from PNDJ15gels. PNDJ15 released 12% within 1 h, 61% within 24 h,and 90% within 72 h. Gels loaded with 50 mg/mL vancomy-cin showed slower release from each hydrogel formulation.PND gels released drug with high sample-to-sample variabil-ity, but on average released 31% within 1 h, 67% within 24h, and 92% in 72 h. PNDJ30 gels provided low burstrelease, but release was 98% complete within 72 h. PNDJ15gels provided the slowest release of vancomycin, with 54%release within 24 h and approximately linear release for 6days thereafter.

In vitro cytocompatibilityResults from the in vitro cytocompatibility study onPNDJ30a degradation byproducts are shown in Figure 6.The number reported represents the number of live cellspresent per well of a 24-well plate. The number of livingNIH3T3 fibroblasts after 3 days was greatly increased fromthe initial seeding of 10,000 cells in every concentration ofthe polymer solution tested, and no significant differencesin cell number between negative control media withoutpolymer and any of the individual polymer concentrationswere found. Representative images of high cell density areasin the presence of control media or 2.5 wt % degradedPNDJ30 for both cell types prior to performing the MTSassay are shown in Figure 7. All of the wells (including con-trols) were observed to have variable cell density from loca-tion to location within a single well, though cell morphology

was consistent. Cells cultured in up to 2.5 wt % degradedpolymer displayed spread or spindle-like morphology thatwas similar to that of cells cultured in standard media.

DISCUSSION

The inclusion of hydrophilic JAAm grafts in temperature-re-sponsive resorbable copolymers results in increased watercontent in the corresponding hydrogels. JAAm content cantherefore be used to control gel properties relevant to localdrug delivery (degradation kinetics, drug release, and rheo-logical properties). The PNDJ gels reported in this work dis-play a unique combination of attributes (viscous, controlledrelease, and degradable over 1–8 weeks) for an injectableand degradable drug carrier. While the relatively fast degra-dation time and viscous nature of the hydrogels limits theirfeasibility for some applications, it may enable their feasibil-ity in others, particularly in sites with irregular geometriesor sites under mechanical load which are unsuitable forpre-formed solid implants or drug-loaded microspheres.

The degradation and swelling behavior of PNDJ hydro-gels changes greatly due to JAAm. Incorporation of JAAm inthe polymer led to reduced shrinking and more uniformdegradation throughout the gel. The initially slow change inLCST observed in PNDJ15 gels agrees well with previouspublished data on PND gels.3 The acceleration of the LCSTincrease observed in PNDJ15 is thought to be attributed tothe degradable ester groups on the butyrolactone ring beinghydrolyzed to pendent hydroxyl and carboxyl groups stillbound to the polymer. Along with increasing the LCST, thisincreases the accessibility of the water to the remainingintact esters of the polymer. In PNDJ30 gels, the polymer isso hydrophilic that the degradation occurs at a faster rateeven before hydrolysis. Despite the fact that the degradation

TABLE II. Cumulative Fraction of Entrapped Drug

(M168 hr/Loading) Released After 168 h

Hydrogel5 mg/mLCefazolin

50 mg/mLCefazolin

50 mg/mLVancomycin

PND 65.3 6 1.2 78.1 6 4.8 72.8 6 11.7PNDJ15 86.3 6 6.6 99.4 6 8.8 70.5 6 2.2PNDJ30 100.0 6 1.7 92.6 6 9.7 86.6 6 2.6

FIGURE 6. Cytocompatibility of degraded PNDJ30a to NIH3T3 fibro-

blasts. No significant differences in cell numbers were observed when

cells were cultured in degraded PNDJ30a at concentrations up to 2.5

wt %. [Color figure can be viewed in the online issue, which is avail-

able at wileyonlinelibrary.com.]

ORIGINAL ARTICLE

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH A | MONTH 2012 VOL 00A, ISSUE 00 7

time of NIPAAm-based gels can be easily controlled usingsmall molecules like AAc,4,14,33 the graft copolymer approachused in this work results controls swelling to a greater extent,which is important for maintaining the interfacial contactbetween the gel and the implantation site as well as for pro-viding controlled release of low molecular weight drugs.

It is important to note that gelation of these copolymersis concentration-dependent and the polymer moleculeswithin each batch are heterogeneous in both content and mo-lecular weight. Accordingly, the gels are likely to weaken ormacroscopically degrade slightly prior to the average LCSTincreasing to body temperature. In vitro, this results in the gelbecoming a translucent solution, and eventually a transparentsolution. While undisturbed PNDJ30 gels remained somewhatintact at 7 or even 13 days, we observed that even gentlereplacement of release media on PNDJ30 gels in the releasestudies caused PNDJ30 gels to partially mix with the releasemedium as early as 5 days postgelation.

A novel aspect of the PNDJ gels is that they behavemore like very viscous liquids than elastic gels, even at tem-peratures above the LCST. PND gels macroscopicallybehaved as solid materials after shrinking—the gelsretained their shape and were not prone to flow. On theother hand, PNDJ15 and PNDJ30 gels had much lower mod-ulus, and were even prone to slow flow over a period ofdays at 37�C. The gels would not noticeably flow in aninverted vial for 30 seconds, but if left overnight, the gelswould flow. These gels therefore would be unsuitable foruse in sites where external forces would deform the gelundesirably or cause displacement of the gel from the site.All of the hydrogels tested would be prone to creep. How-ever, the fluid-like nature of these gels is likely to be accept-able or even advantageous in irregular geometries, in siteswhere the material would remain in the site due to its vis-cosity (such as a thin coating on an implant), or in siteswhere gel cohesion is desired rather than strength, limitingthe potential for immune reaction to wear debris as is seenin many implant environments.

Drug release kinetics from the gels varied depending onthe drug choice, drug loading, and polymer JAAm content.Vancomycin release was sustained for a period of 3–7 daysfrom gels based on all three polymers. On account of vanco-mycin’s higher molecular weight and lower solubility at neu-tral pH, vancomycin release was slower than cefazolin releasefor all gels tested. While PND gels did show the greatestamount of release over the first hour when gel shrinkingoccurs, they nevertheless provided sustained release over atleast 5–7 days, which is evidence that vancomycin distributespredominantly into the polymer-rich gel phase when thepolymer solution is heated above its LCST. The shortest dura-tion of vancomycin release was observed from PNDJ30 gels.These gels became translucent within 72 h of release and par-tially mixed with the release media by 120 h. Therefore, weattribute the faster release from PNDJ30 to partial degrada-tion of the gel, resulting in increased diffusivity of the drug inthe gel and faster release.

When heated above the LCST, PND gels released a largefraction of the loaded cefazolin within 15 min regardless ofdrug loading. Because cefazolin is hydrophilic, it is likely topartition predominantly into the solvent phase as the PNDphase-separates following gelation. PNDJ30 gels showedslower cefazolin release over 24–48 h without the large ini-tial burst observed from PND gels. Although the release wasstill relatively rapid, a majority of the cefazolin was retainedin the gels at 1 h, indicating controlled release. The slowestcefazolin release was observed from PNDJ15 gels, which ex-hibit intermediate hydrophilicity and reduced phase separa-tion. Even though a small fraction of water separates fromPNDJ15 gels, the release data suggests that most of the cefa-zolin was retained within the gel after the transition.Release then occurs almost entirely by diffusion, as thesegels did not degrade much within the first week.

The release rate being the slowest from PNDJ15 gels isevidence that a balance was achieved in these gels betweenlow phase separation (achieved by more JAAm) and slowdegradation (achieved by less JAAm). Even with further

FIGURE 7. Light microscopy photographs of NIH3T3 fibroblasts. Cell density and morphology is similar when cultured in standard media with-

out polymer (top) or 2.5 wt % degraded PNDJ30a (bottom). Scale bars are 100 lm.

8 OVERSTREET ET AL. RESORBABLE TEMPERATURE-RESPONSIVE HYDROGELS

optimization, we anticipate that the release of small hydro-philic drugs such as cefazolin could not be prolonged bymuch longer than 5–7 days by this system, simply due tothe diffusivity of the drug within the polymer gel. This dura-tion of release is shorter than what might be achieved byinjectable gels with much longer degradation times contain-ing drug-loaded microspheres.5 Differences in the final poly-mer content in the gel (after shrinking) did not seem tostrongly affect the diffusivity of the drugs remaining in thegel phase, as the release of vancomycin was similar betweenPND and PNDJ15 gels. While release could be somewhatcontrolled by polymer concentration in the gel, the concen-tration must not exceed about 40 wt % because the poly-mer solution can become too viscous and eventually thepolymer itself would not be soluble. Alternatively, therelease of hydrophobic drugs could be prolonged for amuch longer time, perhaps up to 20 weeks from PND gels,particularly if the drug were loaded well above its solubility,resulting in partition-controlled release.40 Sustained releasebeyond the degradation time of PND would require changesto the polymer composition not explored in this work.

It is important to note that the release kinetics fromthese gels in vivo are expected to depend on the gel shapeand the surrounding fluid environment (viscosity, flow, pH).We chose to evaluate release from discs with 2 mm diame-ter in infinite sink conditions to reliably compare gel formu-lations under conditions prone to fast release. Thus, weanticipate that release would be sustained over a longertime if the gel had a lower surface area-to-volume ratio or ifthe drug were cleared slowly from the local tissue.

Prolonging the release of low molecular weight drugslike cefazolin for 5–7 days has not previously been demon-strated from temperature-responsive injectable gels andcould have significant clinical impact. For example, antimicro-bial-loaded bone cement used for the treatment of estab-lished joint infections shows release which is similar to PNDJhydrogels over the first 7 days, with much slower releasethereafter.41 A resorbable material with similarly effectivedrug might increase the effectiveness of retention debride-ment in treating infection, without requiring two-stage revi-sion arthroplasty42 as is the standard today. This would leadto reduced cost and improved patient quality of life. Alterna-tively, an implanted device or surgical site surface could becoated with antimicrobial-loaded gel to prevent infection,without limitations inherent to hard materials (very slow orno degradation, trade-off between drug loading and fixationstrength,43 generation of wear debris) or particulate drugdelivery systems (wear debris, acidic byproducts).

CONCLUSIONS

In situ forming, resorbable graft copolymers were success-fully synthesized and characterized for the controlledrelease of the model drugs cefazolin and vancomycin. Toour knowledge, these are the first resorbable gels based onNIPAAm demonstrated for the sustained release of hydro-philic low molecular weight drugs. Gel degradation anddrug release was controlled by varying the amount ofhydrophilic JAAm grafts on the polymer backbone. Gels with

greater JAAm content shrank less upon heating above theirLCST, which resulted in reduced burst release, particularlyfor cefazolin. Gels with JAAm were viscoelastic fluids at bodytemperature with storage and loss moduli in the 100–500 Parange at 1 Hz. Drug release from PNDJ15 gels was sustainedfor both drugs tested, and nearly complete elution occurredwithin 7 days. The degradation byproducts of one hydrogelformulation were shown to be cytocompatible to NIH 3T3fibroblasts and MC3T3 osteoblasts at concentrations up to2.5 wt %. Given their unique combination of controlledrelease properties, viscoelasticity, and degradability, thesenew hydrogels are promising for drug-eluting device coatingsor at implant interfaces, where hard materials or particulate-based controlled release systems are not suitable.

ACKNOWLEDGMENTS

The authors acknowledge Kyle Oxley for his help with conduct-ing degradation studies.

REFERENCES

1. Van Dyke TE, Offenbacher S, Braswell L, Lessem J. Enhancing the

value of scaling and root-planing: Arestin clinical trial results. J

Int Acad Periodontol 2002;4:72–76.

2. Peeling WB. Phase III studies to compare goserelin (Zoladex) with

orchiectomy and with diethylstilbestrol in treatment of prostatic

carcinoma. Urology 1989;33(5 Suppl):45–52.

3. Cui Z, Lee BH, Vernon BL. New hydrolysis-dependent thermosen-

sitive polymer for an injectable degradable system. Biomacromo-

lecules 2007;8:1280–1286.

4. Li Z, Wang F, Roy S, Sen CK, Guan J. Injectable, highly flexible,

and thermosensitive hydrogels capable of delivering superoxide

dismutase. Biomacromolecules 2009 14;10:3306–3316.

5. Nelson DM, Ma Z, Leeson CE, Wagner WR. Extended and sequen-

tial delivery of protein from injectable thermoresponsive hydro-

gels. J Biomed Mater Res A 2012;100:776–785.

6. Morscher EW, Hefti A, Aebi U. Severe osteolysis after third-body

wear due to hydroxyapatite particles from acetabular cup coating.

J Bone Joint Surg Br 1998;80:267–272.

7. Trumpy IG, Lyberg T. In vivo deterioration of proplast-teflon tem-

poromandibular joint interpositional implants: A scanning elec-

tron microscopic and energy-dispersive X-ray analysis. J Oral

Maxillofac Surg 1993;51:624–629.

8. Alt V, Bitschnau A, B€ohner F, Heerich KE, Magesin E, Sewing A,

et al. Effects of gentamicin and gentamicin–RGD coatings on

bone ingrowth and biocompatibility of cementless joint prosthe-

ses: An experimental study in rabbits. Acta Biomaterialia 2011;7:

1274–1280.

9. Landes CA, Ballon A, Roth C. In-patient versusin vitro degradation

of P(L/DL)LA and PLGA. J Biomed Mater Res B 2006;76:403–411.

10. Heskins M, Guillet JE. Solution properties of poly(N-isopropyla-

crylamide). J Macromol Sci Chem 1968;2:1441–1455.

11. Schild HG, Tirrell DA. Microcalorimetric detection of lower critical

solution temperatures in aqueous polymer solutions. J Phys

Chem 1990;94:4352–4356.

12. Schild HG. Poly(N-isopropylacrylamide): Experiment, theory and

application. Prog Polym Sci 1992;17:163–249.

13. Lee BH, Vernon B. Copolymers of N-isopropylacrylamide, HEMA-

lactate and acrylic acid with time-dependent lower critical solu-

tion temperature as a bioresorbable carrier. Polym Int 2005;54:

418–422.

14. Ma Z, Nelson DM, Hong Y, Wagner WR. Thermally responsive

injectable hydrogel incorporating methacrylate-polylactide for

hydrolytic lability. Biomacromolecules 2010;11:1873–1881.

15. Overstreet DJ, Dhruv HD, Vernon BL. Bioresponsive copolymers

of poly (N-isopropylacrylamide) with enzyme-dependent lower

critical solution temperatures. Biomacromolecules 2010;11:

1154–1159.

ORIGINAL ARTICLE

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH A | MONTH 2012 VOL 00A, ISSUE 00 9

16. Leon C, Solis FJ, Vernon B. Phase behavior and shrinking kinetics

of thermo-reversible poly(N-isopropylacrylamide-2-hydroxyethyl

methacrylate). Mater Res Soc Symp Proc 2009;1190.

17. Kaneko Y, Nakamura S, Sakai K, Aoyagi T, Kikuchi A, Sakurai Y,

et al. Rapid deswelling response of poly (N-isopropylacrylamide)

hydrogels by the formation of water release channels using poly(-

ethylene oxide) graft chains. Macromolecules 1998;31:6099–6105.

18. Stile RA, Burghardt WR, Healy KE. Synthesis and characterization

of injectable poly(N-isopropylacrylamide)-based hydrogels that

support tissue formation in vitro. Macromolecules 1999;32:

7370–7379.

19. Vernon BL, Kim SW, Bae YH. Thermoreversible copolymer gels

for extracellular matrix. J Biomed Mater Res A 2000;51:69–79.

20. Lee BH, West B, McLemore R, Pauken C, Vernon BL. In-situ inject-

able physically and chemically gelling NIPAAm-based copolymer

system for embolization. Biomacromolecules 2006;7:2059–2064.

21. Gutowska A, Bark JS, Kwon IC, Bae YH, Cha Y, Kim SW. Squeez-

ing hydrogels for controlled oral drug delivery. J Control Rel

1997;48(2–3):141–148.

22. Gutowska A, Bae YH, Feijen J, Kim SW. Heparin release from

thermosensitive hydrogels. J Control Rel 1992;22:95–104.

23. Milasinovic N, Kalagasidis Krusic M, Knezevic-Jugovic Z, Filipovic

J. Hydrogels of N-isopropylacrylamide copolymers with con-

trolled release of a model protein. Int J Pharm 2010;383(1–2):

53–61.

24. Overstreet DJ, McLemore RY, Doan BD, Farag A, Vernon BL. Tem-

perature-responsive graft copolymer hydrogels for controlled

swelling and drug delivery. Soft Materials. In Press.

25. Okano T, Bae YH, Jacobs H, Kim SW. Thermally on-off switching

polymers for drug permeation and release. J Control Rel 1990;

11(1–3):255–265.

26. Bae YH, Okano T, Kim SW. ‘‘On-Off’’ thermocontrol of solute

transport. II. Solute release from thermosensitive hydrogels.

Pharm Res 1991;8:624–628.

27. Garbern JC, Hoffman AS, Stayton PS. Injectable pH- and tempera-

ture-responsive poly(N-isopropylacrylamide-co-propylacrylic acid)

copolymers for delivery of angiogenic growth factors. Biomacro-

molecules 2010;11:5891–5905.

28. Kizhakkedathu JN, Janzen J, Le Y, Kainthan RK, Brooks DE. Poly(-

oligo(ethylene glycol)acrylamide) brushes by surface initiated po-

lymerization: Effect of macromonomer chain length on brush

growth and protein adsorption from blood plasma. Langmuir

2009;25:3794–3801.

29. Jeong B, Bae YH, Kim SW. Drug release from biodegradable

injectable thermosensitive hydrogel of PEG–PLGA–PEG triblock

copolymers. J Control Rel 2000;63:155–163.

30. Potta T, Chun CJ, Song SC. Design of polyphosphazene hydro-

gels with improved structural properties by use of star-shaped

multithiol crosslinkers. Macromol Biosci 2011;11:689–699.

31. Galperin A, Long TJ, Ratner BD. Degradable, thermo-sensitive

poly(N-isopropylacrylamide)-based scaffolds with controlled po-

rosity for tissue engineering applications. Biomacromolecules

2010;11:2583–2592.

32. DeNardo SJ, Yao Z, Lam KS, Song A, Burke PA, Mirick GR, et al.

Effect of molecular size of pegylated peptide on the pharmacoki-

netics and tumor targeting in lymphoma-bearing mice. Clin Can-

cer Res 2003;9:3854S–3864S.

33. Henderson E, Lee BH, Cui Z, McLemore R, Brandon TA, Vernon

BL. In vivo evaluation of injectable thermosensitive polymer with

time-dependent LCST. J Biomed Mater Res A 2009;90:1186–1197.

34. Peppas NA, Khare AR. Preparation, structure and diffusional

behavior of hydrogels in controlled release. Adv Drug Deliv Rev

1993;11(1–2):1–35.

35. Lin HH, Cheng YL. In-situ thermoreversible gelation of block and

star copolymers of poly(ethylene glycol) and poly(N-isopropyla-

crylamide) of varying architectures. Macromolecules 2001;34:

3710–3715.

36. Mathew M, Das Gupta V. Stability of vancomycin hydrochloride

solutions at various pH values as determined by high-perform-

ance liquid chromatography. Drug Dev Ind Pharm 1995;21:

257–264.

37. Antipas AS, Velde DV, Stella VJ. Factors affecting the deamida-

tion of vancomycin in aqueous solutions. Int J Pharm 1994;109:

261–269.

38. Ahmed I, Day P. Stability of cefazolin sodium in various artificial

tear solutions and aqueous vehicles. Am J Hosp Pharm 1987;44:

2287–2290.

39. Galanti LM, Hecq JD, Vanbeckbergen D, Jamart J. Long-term sta-

bility of cefuroxime and cefazolin sodium in intravenous infu-

sions. J Clin Pharm Ther 1996;21:185–189.

40. Vernon BL, Fusaro F, Borden B, Roy KH. Partition-controlled pro-

gesterone release from waterborne, in situ-gelling materials. Int

J Pharm 2004;274(1–2):191–200.

41. Miller R, McLaren A, Leon C, McLemore R. Mixing method affects

elution and strength of high-dose ALBC: A pilot study. Clin

Orthop Relat Res 2012;470:2677–2683.

42. Hsieh P-H, Shih C-H, Chang Y-H, Lee MS, Shih H-N, Yang W-E.

Two-stage revision hip arthroplasty for infection: Comparison

between the interim use of antibiotic-loaded cement beads and a

spacer prosthesis. J Bone Joint Surg Am 2004;86:1989–1997.

43. Moran JM, Greenwald AS, Matejczyk M-B. Effect of gentamicin

on shear and interface strengths of bone cement. Clin Orthop

Relat Res 1979;141:96–101.

10 OVERSTREET ET AL. RESORBABLE TEMPERATURE-RESPONSIVE HYDROGELS