Rheological Characterization of in Situ Cross-Linkable Hyaluronan

Biomacromolecules 2005, 6, 2857-2865

2857

Rheological Characterization of in Situ Cross-Linkable Hyaluronan Hydrogels Kaustabh Ghosh,† Xiao Zheng Shu,‡ Robert Mou,§ Jack Lombardi,§ Glenn D. Prestwich,‡ Miriam H. Rafailovich,| and Richard A. F. Clark*,†,⊥ Departments of Biomedical Engineering, Material Science and Engineering, Dermatology and Medicine, State University of New York at Stony Brook, Stony Brook, New York 11794-8165, Department of Medicinal Chemistry, The University of Utah, 419 Wakara Way, Suite 205, Salt Lake City, Utah 84108-1257, and Physical Chemistry Laboratory, Estee Lauder Research, Melville, New York 11747 Received May 26, 2005; Revised Manuscript Received June 10, 2005

This report investigates the rheological properties of cross-linked, thiol-functionalized HA (HA-DTPH) hydrogels prepared by varying the concentration and molecular weight (MW) of the cross-linker, poly(ethylene glycol) diacrylate (PEGDA). Hydrogels were subsequently cured for either short-term (hours) or long-term (days) and subjected to oscillatory shear rheometry (OSR). OSR allows the evaluation and comparison of the shear storage moduli (G′), an index of the total number of effective cross-links formed in the hydrogels. While the oscillatory time sweep monitored the evolution of G′ during in situ gelation, the stress and frequency sweeps measured the G′ of preformed and subsequently cured hydrogels. From stress sweeps, we found that, for the hydrogels, G′ scaled linearly with PEGDA concentration and was independent of its MW. Upon comparison with the classical Flory’s theory of elasticity, stress sweep tests on short-term cured hydrogels revealed the simultaneous, but gradual, formation of spontaneous disulfide cross-links in the hydrogels. Results from time and frequency sweeps suggested that the formation of a stable, threedimensional network depended strictly on PEGDA concentration. Results from the equilibrium swelling of hydrogels concurred with those obtained from oscillatory stress sweeps. Such a detailed rheological characterization of our HA-DTPH-PEGDA hydrogels will aid in the design of biomaterials targeted for biomedical or pharmaceutical purposes, especially in applications involving functional tissue engineering. Introduction Hyaluronan (HA) is a linear, nonsulfated glycosaminoglycan found ubiquitously in the extracellular matrix (ECM) of virtually all mammalian connective tissues.1 This polyanionic biopolymer is composed of repeating disaccharide units of β-1,4-D-glucuronic acid and β-1,3-N-acetyl-D-glucosamine.2 Previously, HA in tissues was believed to act only as an inert lubricating substance; however, important biological functions of HA are now widely reported in the literature.1,3 HA occurs naturally in a wide range of molecular weights (MWs; 0.1-10 million) and concentrations4 and is stabilized by association with a number of link proteins. These attributes impart HA with its unique rheological properties that allow it to fulfill diverse physicochemical functions in different locations in the body. This property of native HA has been exploited therapeutically in viscosupplementation and viscosurgery.5 However, pure native HA has found limited clinical application, largely due to its poor bio* Corresponding author: phone 631-444-7519; fax 631-444-3844; e-mail [email protected]. † Department of Biomedical Engineering, SUNY at Stony Brook. ‡ The University of Utah. § Estee Lauder Research. | Department of Material Science and Engineering, SUNY at Stony Brook. ⊥ Department of Dermatology and Medicine, SUNY at Stony Brook.

mechanical properties and rapid degradation in vivo. To address this problem, several chemical modifications have been successfully employed to significantly improve the biomechanical properties of HA and, thereby, its ease of handling and residence time in vivo.6-9 One such novel chemical modification involving the synthesis of thiol-functionalized HA (HA-DTPH) has been previously reported6 (Figure 1A). By virtue of their ability to form spontaneous disulfide bonds upon exposure to air, the free thiols on HA backbone act as latent cross-linking agents (Figure 1B). Since the formation of disulfide bonds is slow, Michael-type addition between free thiols and acrylates of poly(ethylene glycol) (PEG) has been utilized to rapidly cross-link HA-DTPH. This chemistry, first reported by Lutolf et al.10 and suitably modified for HADTPH,11 uses homobifunctional PEG diacrylate (PEGDA) to produce an in situ cross-linkable hydrogel in approximately 10 min (Figure 1C). Importantly, the cross-linking reaction between free thiols and PEGDA occurs at physiological pH and room temperature, which ensures complete biocompatibility during cell encapsulation or incorporation of biologically active ligands.11,13 The disulfide and PEGDA crosslinked HA-DTPH biomaterials (denoted hereafter by HAS-S-HA and HA-PEGDA-HA, respectively) and their functional derivatives have great potential in drug delivery and tissue engineering applications.6,11-13

10.1021/bm050361c CCC: $30.25 © 2005 American Chemical Society Published on Web 07/20/2005

2858

Biomacromolecules, Vol. 6, No. 5, 2005

Ghosh et al.

Figure 1. Schematic showing (A) synthesis of thiol-functionalized HA (HA-DTPH),6 (B) spontaneous cross-linking forming HA-S-S-HA linkages, and (C) exogenous PEGDA-mediated cross-linking forming HA-PEGDA-HA linkages (adapted and modified from Shu et al.6).

Previous published data indicate that the rheological properties of most polymeric biomaterials, including HAbased biomaterials, depend not only on the MW and concentration of the macromer but also on the nature and density of effective cross-links.11,14 Furthermore, the rheological properties of biomaterials can modulate their therapeutic utility. For example, rheological modifications of PLGA film-based implants affect drug release profiles.15 The cross-linking density influences the stiffness of PEG hydrogels and thereby the synthesis and distribution of extracellular matrix produced by the seeded chondrocytes.16 Lee et al.17 reported the effect of multifunctional cross-linking on in vitro

degradation of the resulting hydrogels. The viscoelasticity of cross-linked HA has been shown to affect their therapeutic potential in viscosupplementation,18 in significant reduction of postsurgical adhesions,19 and in soft-tissue augmentation.20 These findings underscore the importance of rheological properties of biomaterials. This study, therefore, characterizes the rheological properties of the recently reported HA-PEGDA-HA hydrogels.11 Oscillatory shear rheometry, operated in time, stress, and frequency sweep modes, was used to evaluate the shear storage moduli (G′) of the hydrogels as a function of concentration and MW of the PEGDA cross-linker. Since

Rheology of Cross-Linked HA Hydrogels

the slow HA-S-S-HA and the rapid HA-PEGDA-HA cross-linking reactions proceed at very different rates,11 we also monitored G′ as a function of short-term (hours) and long-term (days) curing times. Effects of PEGDA concentration and MW on the levels of equilibrium swelling of the various hydrogels were also evaluated. Materials and Methods Materials. Fermentation-derived hyaluronan (HA, sodium salt, Mw 1 500 000) was provided by Clear Solutions Biotechnology, Inc. (Stony Brook, NY). 1-Ethyl-3-[3-(dimethylamino)propyl]carbodiimide (EDCI) and poly(ethylene glycol) diacrylate (PEGDA, MW 700, purity 95%) were purchased from Aldrich Chemical Co. (Milwaukee, WI). PEGDA (MW 3400, purity 98%) was purchased from Nektar Therapeutics (Huntsville, AL). Dulbecco’s modified Eagle’s medium (DMEM) was obtained from Sigma Chemical Co. (St. Louis, MO). Dithiothreitol (DTT) was purchased from Diagnostic Chemical Limited (Oxford, CT). 5,5′-Dithiobis(2-nitrobenzoic acid) (DTNB) was purchased from Acros (Houston, TX). 3,3′-Dithiobis(propanoic dihydrazide) (DTP) was synthesized as previously described.6 Distilled phosphatebuffered saline (1× dPBS, pH 7.4) was prepared in the lab according to a standard protocol. Synthesis of Thiol-Functionalized HA. Thiol-functionalized HA, or 3,3′-dithiobis(propanoic dihydrazide)-modified HA (HA-DTPH), was synthesized according to a previously reported procedure.6 In principle, the native carboxylic groups on HA disaccharide units were replaced by thiol-containing DTPH groups. The degree of substitution (% SD), defined as the number of DTPH groups per 100 disaccharide units on the HA molecule, was determined by 1H NMR.6 The free thiol content (percent), defined as the number of free thiols per 100 disaccharide units, was measured in parallel by a modified Ellman method.6,21,22 Both % SD and the free thiol content were found to be approximately 42%, indicating that the remaining 58% of the HA disaccharide units contained the native carboxylic acid group. The pKa of thiols in HADTPH was 8.87 as determined spectrophotometrically on the basis of UV absorption of thiolates. The MW was determined by calibrated gel-permeation chromatography (GPC) to be Mw 158 000 and Mn 78 000 (polydispersity index 2.03). Specimen Preparation: Hydrogels. A 1.25% (w/v) HADTPH solution was prepared by dissolving HA-DTPH in serum-free DMEM supplemented with 1% (v/v) penicillin, streptomycin, and glutamine (antibiotic mix). The HADTPH solution was first pH-adjusted to 7.4 (by addition of 1.0 M NaOH) and then sterilized by filtration through a 0.22 µm filter. A 4.5% (w/v) PEGDA (MW 3400) stock solution was prepared by dissolving PEGDA powder in 1× dPBS. Four volumes of 1.25% HA-DTPH solution were then mixed with one volume of PEGDA solution of varying concentrations (4.5%, 3.0%, 2.25%, 1.5%, and 0.75%) to obtain HA-PEGDA-HA hydrogels of different crosslinking densities (defined in this report as the molar ratio of thiols on HA-DTPH:acrylate groups on PEGDA) of 2:1, 3:1, 4:1, 6:1, and 12:1, respectively. To determine if MW of PEGDA had an influence on the physicochemical proper-

Biomacromolecules, Vol. 6, No. 5, 2005 2859

ties, HA-PEGDA-HA hydrogels with a cross-linking density of 2:1 were also prepared from PEGDA MW 700. The final concentration of HA-DTPH in the hydrogels was always 1% (w/v). As previously reported,11 gelation time was found to be inversely proportional to PEGDA concentration. A 1% HA-DTPH solution was also prepared without addition of any PEGDA. All hydrogels were plated in regular 35 mm tissue culture dishes. Serum-free DMEM was added to the surface of all hydrogels to allow for equilibrium swelling prior to oscillatory shear rheometry. Regardless of the cross-linking densities used in this study, all HA-PEGDA-HA hydrogels ideally contain excess free thiols. We, therefore, investigated the extent of net effective cross-linking in these hydrogels arising from both HA-SS-HA and HA-PEGDA-HA cross-links as a function of curing time. For this study, all HA-PEGDA-HA hydrogels were plated simultaneously (at t ) 0); at the end of each experimental time point, an 11 mM iodoacetamide solution in 1× dPBS was added to the hydrogel surface to block residual free thiols and thereby prevent any additional HAS-S-HA or HA-PEGDA-HA cross-links. Rheological Characterization: Oscillatory Shear Rheometry of Hydrogels. An AR2000 rheometer (TA Instruments Inc.) with a standard steel parallel-plate geometry of 20 mm diameter was used for the rheological characterization of all hydrogel samples. The test methods employed were oscillatory time sweep, stress sweep, and frequency sweep. The time sweep was performed to monitor, within a given time frame, the in situ gelation of the 2:1, 6:1, and 12:1 HAPEGDA-HA hydrogel solutions. The strain was maintained at 5% during time sweeps by adjusting the stress amplitude. The test, which was operated at 1 Hz and terminated after 30 min, recorded the temporal evolution of G′ and the shear loss modulus, G′′. The stress sweep was performed on hydrogels to determine and compare their G′ under the same physical condition. The stress sweep was set up by holding the temperature (25 °C) and frequency (1 Hz) constant while increasing the stress level from 50 to 70 Pa. The applied range of 50-70 Pa was found to be safe-for-use from a prior experiment where we determined the linear viscoelastic region (LVR) profiles of the 2:1, 6:1, and 12:1 hydrogels by shearing them until structure breakdown. In the stress sweep (or “controlled stress”) tests, the stress was locally controlled in every cycle and the strain (and the corresponding G′) was measured, while globally speaking, the hydrogels were subjected to a steady stress ramp. A constant normal compressional force of ∼4g was applied to all samples throughout the stress sweep regime. Both the time and stress sweeps provide G′ and G′′ information on the structural integrity of the crosslinked network, but at two different physical settings. We also subjected the 2:1, 6:1, and 12:1 hydrogels to a frequency sweep at 50% of their respective ultimate stress levels (corresponding to the point of dip on the LVR profile). At this fixed shear stress and temperature (25 °C), the oscillatory frequency was increased from 0.1 to 100 Hz and the G′ was recorded. To avoid dislocation during each test method, all 35 mm dishes containing the hydrogels were fixed to the bottom plate with stable, double-sided tape. The

2860


plots of G′ versus shear stress, reaction time, or frequency from the three sweep tests were obtained directly from the software controlling the rheometer. All samples were done in triplicate. FTIR Analysis. A Nicolet Magna-IR 760 optical bench spectrometer (Thermo Electron Corp.) was used to obtain Fourier transform infrared (FTIR) spectra of pure HADTPH, pure PEGDA, and 8-h cured HA-DTPH-PEGDA hydrogel on calcium fluoride disks. A normalized spectrum was obtained by subtracting the HA-DTPH spectrum from the hydrogel spectrum, which was then compared with that of pure PEGDA. The peaks at 1634 and 1410 cm-1 (corresponding to -CdC- bond stretching and scissoring, respectively) were used to follow the consumption of PEGDA in the HA-PEGDA-HA hydrogel. Equilibrium Swelling of Hydrogels. Identical volumes of hydrogel samples were plated in wells of a 24-well plate. All hydrogel samples were allowed to sit at room temperature for about 2 h before weighing them; this allowed the weakly cross-linked 12:1 hydrogels to set well before the start of the experiment. After the initial weights (Wi) were recorded, all hydrogels were gently transferred to weigh boats filled with distilled, deionized water. To obtain equilibrium swelling, all samples were allowed to swell at room temperature for 48 h. The equilibrium swollen mass (Ws) was then recorded by gently blotting excess water from each sample. The hydrogel samples were subsequently dried for 48 h in a desiccator at room temperature and their dry weights (Wd) were measured. The equilibrium swelling ratio (Q) was defined as the ratio of Ws to Wd. Results and Discussion Thiol-functionalized HA, or HA-DTPH, is synthesized by substituting the native carboxylic group on HA molecule with free and active thiol groups. Upon exposure to air, the thiols on HA-DTPH are oxidized to form a spontaneous, albeit slow (within 4-6 h), HA-S-S-HA cross-linked network. These HA-S-S-HA cross-links are reversible in nature since addition of DTT (a reducing agent) results in the dissolution of the networked structure.6 However, to enhance the rate of cross-linking, Michael-type addition reaction is employed where, by use of homobifunctional PEGDA, a HA-DTPH solution is cross-linked to form a stable hydrogel within approximately 10 min.11 This rapid gelation, also occurring at physiological pH and room temperature, advocates its proposed injectable in vivo use. In the first published report describing the formation of HA-PEGDA-HA hydrogels,11 Shu et al. showed an increase in both cross-linking efficiency of PEGDA (i.e., double-end anchorage, from 76.2% to 100%) and the observed gelation time (from 5 to 19 min) with decreasing PEGDA concentration [from 9% to 3% (w/v)]. On the basis of these data, PEGDA concentration of 4.5% (corresponding to a cross-linking density of 2:1) was found to be optimum for use both in vitro and in vivo.11,13 Importantly, after these optimally cross-linked hydrogels have been formed, about half the original number of thiol groups is still freely available that may potentially form additional HA-S-S-

Ghosh et al.

HA cross-links. Both the extent and rate of formation of these disulfide links are likely to alter the microlevel network structure and thereby the rheological and physicochemical properties of these HA-PEGDA-HA hydrogels. Altering the molar concentration and MW of PEGDA can also produce such effects. A similar trend was observed in HAADH-PEG dialdehyde hydrogels where changes in PEG dialdehyde (cross-linker) concentration altered the network structure and the inherent physicochemical properties.23 These issues have, however, never been addressed before for the HA-PEGDA-HA hydrogels. Therefore, in this study we aimed to determine the effect of the above-mentioned parameters on the rheological behavior and levels of equilibrium swelling of the resulting HA-PEGDA-HA hydrogels. On the basis of these data, we propose plausible schemes of cross-linking occurring in these hydrogels. Rheological Characterization of HA-DTPH Hydrogels. (A) Oscillatory Time Sweep. Oscillatory time sweeps were performed to monitor the in situ gelation of HA-PEGDAHA solutions prepared from PEGDA MW 3400, the MW shown to be optimal for use both in vitro and in vivo.11 Figure 2 shows the time sweep profiles of G′ and G′′ for the 2:1, 6:1, and 12:1 HA-PEGDA-HA hydrogel networks (panels A, B, and C, respectively). Initially, G′′ is larger than G′, which is expected since the samples are still in liquid state where viscous properties dominate, and therefore most (if not all) of the energy is lost as viscous heat. As the solutions begin to gel and a cross-linked network is formed, both G′ and G′′ begin to increase; however, the rate of increase of G′ is much higher than that of G′′ since now the elastic properties of the gelling hydrogel begin to dominate. Consequently, there is a crossover point where G′ becomes larger than G′′. The time required for this crossover to occur is sometimes referred to as the gelation time for the solution.24 Although the apparent gelation times observed by the “test tube inverting” method were greater than those observed in these profiles, they were proportionate for all three cross-linking densities. Furthermore, from Figure 2 we see that with increased PEGDA concentration the crossover point appears sooner, indicating that PEGDA cross-linking of HA-DTPH is the rate-limiting reaction during this early phase of gelation. Although the plot of G′′ plateaus with time, it never decreases to 0, suggesting the viscoelastic nature of these hydrogels under the applied physical conditions. The slightly erratic nature of G′′ observed during the time sweep tests is attributed to grip-slip caused by the release of water from the hydrogels as they undergo shear stress. (B) Oscillatory Stress Sweep. Oscillatory stress sweep allows determination of G′ of the hydrogels as a function of PEGDA concentration and MW. The effect of curing time on G′ can also be similarly evaluated. The data obtained can be further used to predict and compare the rate and extent of formation of effective cross-links in various hydrogels. In compliance with the principle of small deformation rheology,24 the hydrogels must be tested within their respective linear viscoelastic ranges, the length of which determines the structural stability. We, therefore, first determined the LVR profiles of the 2:1, 6:1, and 12:1 HA-PEGDA-HA hydrogels by subjecting them to a stress sweep until structure



Figure 3. Determination of the linear viscoelastic region (LVR) of the 8-h cured 12:1, 6:1, and 2:1 cross-linked HA-PEGDA-HA hydrogels (PEGDA MW 3400). Frequency of the applied oscillatory shear stress was 1 Hz.

Figure 4. Comparison of G′ of HA-PEGDA-HA hydrogels prepared from PEGDA of MW 3400 and 700. Oscillatory shear stress was performed at 1 Hz.

Figure 2. Evolution of shear storage moduli, G′ (]), and shear loss moduli, G′′ (4), as a function of time during the pregelation and early gelation phases of (A) 2:1, (B) 6:1, and (C) 12:1 HA-PEGDA-HA hydrogels (PEGDA MW 3400).

breakdown. Since the HA-PEGDA-HA and HA-S-SHA cross-links, and the corresponding rheological properties, are expected to develop over varying time scales (minutes to hours to days),11 it becomes impractical to perform rheological tests at all time points. Therefore, rheological properties such as the LVR profile and frequency response of G′ (in the following section) were determined at a fixed curing time of 8 h. The choice of this curing time point was justified from previous studies on a similar system (HAPEG monoacrylate conjugation)11 that intuitively suggest that HA-PEGDA-HA cross-links might more or less reach completion within this period. Figure 3 represents the LVR profile of the 8-h cured HA-PEGDA-HA hydrogels, showing clearly that, with increasing cross-linking density, the structure breakdown occurred at higher shear stress levels. On the basis of these data, a stress range of 50-70 Pa was chosen for comparing various hydrogel samples, which was

verified to be lying in the LVR of even the least (2 h) cured hydrogels (data not shown). We next looked at the effect of PEGDA MW on the storage moduli of the hydrogels. PEGDA MW 700 and 3400 were used to prepare 2:1, 6:1, and 12:1 HA-PEGDA-HA hydrogels. These hydrogels were subsequently cured for 8 h before being subjected to stress sweep. Figure 4 illustrates that G′ of HA-PEGDA-HA hydrogels was almost entirely independent of PEGDA MW; G′ was, however, strongly dependent on the number of cross-links formed, which was controlled by PEGDA concentration. We inferred that, in a hydrated state, water molecules act as the most flexible component and therefore the hydrogel stiffness becomes insensitive to the flexibility imparted by the length (or MW) of the PEGDA molecule. As a result, all subsequent oscillatory stress sweeps were performed on hydrogels prepared from PEGDA MW 3400 only. Next, hydrogels of varying cross-linking densities were formed and allowed to cure for 2, 4, 8, 10, and 24 h. Existing theories and numerous published reports have earlier suggested a strong correlation between the measured G′ and the number of effective intermolecular cross-links formed in a hydrogel network.14,15,17,25,26 We, therefore, monitored and compared the evolution of G′ as a function of cross-linking density and curing time to assess the extent of effective intermolecular cross-links formed in the various hydrogel networks. Two additional cross-linking densities of 3:1 and

2862


Ghosh et al.

Figure 5. Evolution of G′ of HA-PEGDA-HA hydrogels as a function of short-term curing time (PEGDA MW 3400).

4:1 were introduced in this study in order to obtain enough data points to allow comparison of our results with those predicted by the classical elasticity theories. From the results of the stress sweep test shown in Figure 5, the development of G′ appears to be solely governed by the cross-linking density (or PEGDA concentration). Additionally, these data also suggests that within the small time frame of this test, the HA-S-S-HA cross-links do not contribute significantly to the network structure properties. This was further confirmed by measuring the G′ of HA-S-S-HA hydrogels (no PEGDA added) at 24 h postcuring. Not so surprisingly, the value was much smaller (13 Pa) that is, about 1/7 the value for the most weakly cross-linked 12:1 HA-PEGDAHA hydrogels (Figure 5). A thorough understanding of the structure of underlying molecular networks requires evaluating differentially the extent and rate of formation of both HA-PEGDA-HA and HA-S-S-HA cross-links. One must, therefore, be able to tell approximately (a) when the HA-PEGDA-HA crosslinks become saturated and (b) when and to what extent the HA-S-S-HA cross-links contribute to the overall G′network structure relationship. The classical Flory’s rubber elastic theory,27 widely used by investigators to validate experimental assumptions and data,24,28 was consequently used as a standard reference to find plausible explanations. Since the rate of conjugation between acrylate and thiol groups is much faster than that between two thiol groups, we looked at both short-term and long-term curing effects on the evolution of G′. Short-Term Curing. When G′ was plotted against PEGDA molar concentration, a linear relationship was obtained initially at the end of the 30-min time sweep (Figure 6A), characterized by high R2 and low χ2 values for the linear fit (Table 1). This linear relationship is representative of the classical Flory’s rubber elasticity theory27 that assumes the cross-linking reaction to be solely cross-linker (PEGDA) governed. As shown in Figure 6B, with increasing curing time (up to 24 h), however, we observed an increasing deviation from the typical first-order linearity to a power law, with R2 decreasing and χ2 increasing for the linear fit and vice versa for the power fit, indicating that the crosslinking reaction was no longer entirely PEGDA governed. This can be attributed to the gradual formation of HA-SS-HA cross-links occurring secondary to HA-PEGDA-

Figure 6. Correlation of experimental G′ values, obtained at (A) 30 min and (B) 24 h postcuring with those predicted from the classical Flory’s rubber elasticity theory. Shown are the linear and power fits, with the solid line indicating the better fit. Table 1. Comparing G′ Values Obtained for Short-Term Cured Hydrogels with Those Expected from Flory’s Elasticity Theory

R2

χ2

short-term curing time linear fita power fitb linear fita power fitb 30 min 4h 8h 24 h

0.9945 0.9905 0.9851 0.9733

0.9583 0.9835 0.9816 0.9948

3.684 37.782 41.238 94.557

95.216 25.162 32.602 2.507

a As predicted by Flory’s elasticity theory. b Deviation from Flory’s prediction.

HA cross-links. It is also plausible that intramolecular crosslinks are occurring. These would, however, not contribute to the increase in G′ and may in fact interfere with formation of intermolecular cross-links. Also from Figure 5, we see that for almost all hydrogel samples the G′ values plateau between 10 and 24 h, with about 90% or more of the final (24 h) value attained in just about 8 h. This, we assumed, was due to the rapid quenching of all available PEGDA molecules that result in the formation of effective double-end-anchored HA-PEGDA-HA crosslinks in the network. To verify analytically, we used FTIR to follow the addition reaction between PEGDA and HADTPH in the 2:1 cross-linked HA-PEGDA-HA hydrogels. Figure 7 shows the normalized spectrum obtained by subtracting the HA-DTPH spectrum from that of the hydrogel, which was used at 8 h postcuring. The arrows indicate the disappearance of the absorbance peaks at 1634 and 1410 cm-1 in the normalized spectrum that correspond to the stretching and scissoring of -CdC- bonds in PEGDA, respectively. This suggests that 90% or more of PEGDA is used up in the reaction with HA-DTPH within 8 h, a finding that correlates well with those obtained from other chemical analyses of these HA-PEGDA-HA hydrogels. By use of GPC and the modified Ellman method, it


Figure 7. Changing FTIR spectrum of PEGDA during Michael-type addition with HA-DTPH. Arrows indicate the disappearance of absorbance peaks at 1634 cm-1 (-CdC- stretching) and 1410 cm-1 (-CdC- scissoring) in the normalized spectrum. HA-PEGDA-HA hydrogels were cured for 8 h.

Figure 8. Evolution of G′ of HA-PEGDA-HA hydrogels as a function of long-term curing time (PEGDA MW 3400).

has previously been shown11 that 94% of the bifunctional PEGDA molecules get effectively incorporated (double-endanchored) into the HA-PEGDA-HA network within a similar time frame. The same study also showed that, at a lower cross-linking density of 3:1, all PEGDA molecules were completely used up in the addition reaction. It is therefore likely that, at lower cross-linking densities of 6:1 and 12:1, all PEGDA become incorporated within the hydrogel network. Long-Term Curing. Since even the most densely crosslinked 2:1 hydrogel should stoichiometrically contain half the original number of free thiol groups, we investigated the extent of formation of additional HA-S-S-HA cross-links as a function of long-term curing. The 12:1, 6:1, and 2:1 hydrogels were chosen for this study where the hydrogels were cured over a period of 1, 3, 6, or 9 days (instead of hours as in short-term curing) and later subjected to oscillatory stress sweep. Figure 8 shows a plot of G′ versus curing time, where G′ of all the HA-PEGDA-HA hydrogels was found to plateau twice, once between 10 and 24 h and again several days later. The second plateau of G′ occurred at 3 days for 2:1 hydrogels and at 6 days for 6:1 and 12:1 hydrogels. The plateau reached between 10 and 24 h arises


from the quenching of all PEGDA molecules, which is shown in detail in Figure 5. This suggests that a further increase in G′ and the corresponding second plateau arises from the formation of HA-S-S-HA cross-links. Since the tendency to form HA-S-S-HA cross-links should be identical in all samples (owing to identical thiol content of HA-DTPH in all samples), data from Figure 8 suggest that PEGDA, besides creating HA-PEGDA-HA cross-links, also strongly facilitates the formation of HA-S-S-HA cross-links that otherwise form at a very slow rate (Figure 5). This concept can be explained as follows. The large segmental fluctuations and the entropic barrier inhibit spontaneous formation of HA-S-S-HA cross-links. This barrier may be overcome if the enthalpy of the crosslinking reaction is highly favorable, as is the case when PEGDA is present. We believe that due to its high concentration in the 2:1 hydrogels, PEGDA potentially creates a “zipping” effect where, during cross-linking, it brings the HA-DTPH chains and, as a result, the pendant free thiol groups in close proximity to each other, thereby facilitating the formation of additional HA-S-S-HA linkages. In this way, PEGDA helps the HA-DTPH chains overcome the entropic barrier that would otherwise preclude the formation of such HA-S-S-HA cross-links. In 6:1 and 12:1 hydrogels, the lower concentration of PEGDA fails to produce this effect to the same extent and hence the rate of formation of spontaneous HA-S-S-HA linkages is much slower. Alternate reasoning comes from the fact that the probability of bond formation between different chains is proportional to the probability of those chains lying in the same small volumetric unit.29 Along this line of reasoning, HAPEGDA-HA cross-links restrict the random flexibility of the HA-DTPH chains, thereby increasing their density per unit volume. This effect may also produce nonuniformities in the cross-linking density since areas of relatively higher density may “nucleate” around regions where the HAPEGDA-HA cross-links occur. A detailed understanding of this effect is relevant to the characterization of this biomaterial since this effect is also responsible for deviations from linearity of the modulus with cross-linking density. More importantly, this may affect the cross-linking density in the near-surface region, which is relevant for cell/substrate interactions. This effect will be studied in the future by finite element analysis and the digital image speckle correlation methods,30-32 which can map the modulus across the sample with submicrometer resolution. (C) Oscillatory Frequency Sweep. Frequency sweep tests are widely used to obtain information about the stability of three-dimensional cross-linked networks.14,24,25,33 Consequently, we subjected our 8-h cured 2:1, 6:1, and 12:1 HAPEGDA-HA hydrogels to a frequency sweep from 0.1 to 100 Hz (100 Hz being the highest operable frequency on the AR2000 rheometer). The shear stresses applied to hydrogels were 50% of their respective ultimate stress levels. Shown in Figure 9 is the plot between G′ and oscillatory frequency (top panel), where the data obtained for 2:1 hydrogels are characterized by G′ exhibiting a plateau in the range 0.1-10 Hz that is indicative of a stable, cross-linked network. The G′-frequency profile for 6:1 and 12:1 hydro-

2864


Ghosh et al. Table 2. Equilibrium Swelling Data PEGDA concn

equilibrium swelling ratio

% (w/v)

-SH:-CdC-

PEGDA MW 3400

PEGDA MW 700

0.75 1.50 4.50

12:1 6:1 2:1

158 ( 21 93 ( 5 58 ( 2

155 ( 6 87 ( 12 64 ( 4

Figure 9. Evolution of G′ and strain as a function of the applied frequency of oscillatory shear stress.

gels failed to plateau within 10 Hz, indicating that the stability of the cross-linked network depends primarily on HA-PEGDA-HA cross-links. The observed difference in G′ arises from the difference in their absolute strength that corresponds well with that obtained from stress sweep tests. At higher frequencies (10-100 Hz), however, all the hydrogels showed an increase in G′ where the rate of increase was highest for 12:1 hydrogel and lowest for 2:1 hydrogel. This we interpreted as follows. The magnitude of viscoelastic response elicited by a polymeric network is governed primarily by two factors: first, the length of the flexible polymer chains, and second, the nature of the imposed mechanical motion.14,24 Longer chains have characteristic longer relaxation times (or equivalently, lower frequencies of molecular motion) and vice versa. In HA-PEGDA-HA cross-linked networks, the HA-DTPH chain segments between cross-links are longest in the less cross-linked 12:1 networks and shortest in the highly crosslinked 2:1 network. This means that the HA-DTPH chains in the 12:1 network will exhibit comparatively lower frequencies of molecular motion than those in the 2:1 network. At higher frequencies though, long chains (as in the 12:1 network) fail to rearrange themselves in the time scale of the imposed motion and therefore stiffen up, assuming more “solidlike” behavior characterized by a sharp increase in G′ (as observed in Figure 9). Since shorter polymer chains (as in the 2:1 network) exhibit smaller relaxation times, they require higher applied frequencies to elicit a similar response, as observed by only a gradual rise in G′ for the 2:1 network. The plot of strain versus applied frequency (bottom panel in Figure 9) further corroborates these data. At lower frequencies (up to 1 Hz), the 2:1 hydrogel network exhibits smaller strain (compared to the 12:1 network) since it is stiffer by nature (as can be seen from stress sweep tests in Figure 3). However, due to chain stiffening at higher frequencies, the 12:1 hydrogel network undergoes less deformation than the 2:1 network and assumes a more “solidlike” structure (as observed in the G′ versus frequency profile). The resolution of the data was limited since 100 Hz was the maximum operable frequency on the AR2000N rheometer used for the study.

Figure 10. Relationship between G′ and levels of equilibrium swelling (Q) of HA-PEGDA-HA hydrogels (PEGDA MW 3400).

Equilibrium Swelling of HA-DTPH Hydrogels. The degree of equilibrium swelling of a polymeric hydrogel is known to be inversely proportional to its mechanical strength (or G′).14,24 The swelling of polymer chains reaches equilibrium at a point where the swelling (osmotic) force is just balanced by the elastic restoring (entropic) force acting in the opposite direction. We investigated the degree of equilibrium swelling as a function of PEGDA MW (3400 and 700) and concentration (corresponding to cross-linking densities of 2:1, 6:1, and 12:1). From the results in Table 2, we see that degree of swelling is a strong inverse function of PEGDA concentration while being largely independent of its MW. This implies that the elastic restoring force is inversely proportional to the overall flexibility of the network chains that, in turn, is inversely proportional to PEGDA concentration. The effect of varying PEGDA MW is neutralized in a hydrated state where water acts as the most flexible component. When G′ was plotted against the swelling ratio (Figure 10), the curve followed a power law with an exponent of -4 that correlated well with the literature (where a similar value was obtained for PEG-mediated hydrogels33) and standard theories.34,35 Conclusion This report describes an in-depth rheological characterization of HA-DTPH-PEGDA hydrogels under varying time scales and force fields. Oscillatory stress, time, and frequency sweeps were performed on an AR 2000N rheometer to monitor the effect of PEGDA concentration and molecular weight on hydrogel rheology (expressed in terms of G′). Stress sweeps indicated that G′ of the hydrogels varied linearly with PEGDA concentration, while being almost independent of its MW. These data were in good agreement with those obtained from the equilibrium swelling studies, which also indicated a similar trend. The linear plot of G′ versus equilibrium swelling was in accordance with the numerous literature reports showing a similar relationship. Since the amount of PEGDA used for exogenous crosslinking was stoichiometrically less than the number of free


thiols available, these hydrogels always contained excess free thiols. We, therefore, monitored the evolution of G′ of the hydrogels, an index of the total number of disulfide and PEGDA-mediated cross-links formed, as a function of curing time. G′ of hydrogels reached a plateau at shorter curing times, indicating that PEGDA-mediated cross-linking dominated the initial network formation. However, when compared to the classical Flory’s elasticity theory, the data suggested a slow but simultaneous formation of disulfide cross-links. The second plateau of G′ was obtained at longer curing times, indicating the saturation of all possible disulfide cross-links. The end-point difference in G′ between the three hydrogels demonstrates the zipping effect of PEGDA, where a higher concentration causes an increase in the rate and extent of the formation of spontaneous disulfide cross-links. G′ values obtained from the time and frequency sweeps revealed a direct dependence of the three-dimensional network stability on PEGDA concentration. A detailed rheological and physicochemical characterization of HADTPH-PEGDA hydrogels such as this will lead to the development of biomaterials suited for numerous in vivo applications. Acknowledgment. Financial support to R.A.F.C. (NIH Grant AG010143), M.H.R. (NSF-MRSEC program and DOE) and G.D.P. (NIH Grant DC004336 and Centers of Excellence Program, State of Utah) is gratefully acknowledged. We thank Dr. Shouren Ge for helpful discussions and consultation in experimental design. References and Notes (1) Fraser, J. R. E.; Laurent T. C.; Laurent U. B. G. J. Intern. Med. 1997, 242 (1), 27-33. (2) Knudson, C. B.; Knudson, W. Semin. Cell DeV. Biol. 2001, 12, 26978. (3) Savani, R. C.; Cao, G.; Pooler, P. M.; Zaman, A.; Zhou, Z.; DeLisser, H. M. J. Biol. Chem. 2001, 276 (39), 36770-78. (4) Laurent, T. C.; Fraser, J. R. E. FASEB J. 1992, 6 (7), 2397-404. (5) Engstrom-Laurent, A. J. Intern. Med. 1997, 242, 157-60. (6) Shu, X. Z.; Liu, Y.; Luo, Y.; Roberts, M. C.; Prestwich, G. D. Biomacromolecules 2002, 3 (6), 1304-11. (7) Luo, Y.; Kirker, K. R.; Prestwich, G. D. J. Controlled Release 2000, 69 (11), 69-84.

Biomacromolecules, Vol. 6, No. 5, 2005 2865 (8) Prestwich, G. D.; Marecak, D. M.; Marecek, J. F.; Vercruysse, K. P.; Ziebell, M. R. J. Controlled Release 1998, 53, 93-103. (9) Kuo, J. W.; Swann, D. A.; Prestwich, G. D. Bioconjugate Chem. 1991, 2, 4232-41. (10) Lutolf, M. P.; Tirelli, N.; Cerritelli, S.; Cavalli, L.; Hubbell, J. A. Bioconjugate Chem. 2001, 12, 1051-56. (11) Shu, X. Z.; Liu, Y.; Palumbo, F.; Luo, Y.; Prestwich, G. D. Biomaterials 2004, 25, 1339-48. (12) Shu, X. Z.; Liu, Y.; Palumbo, F.; Prestwich, G. D. Biomaterials 2003, 24, 3825-3834. (13) Shu, X. Z.; Ghosh, K.; Liu, Y.; Palumbo, F. S.; Luo, Y.; Clark, R. A.; Prestwich, G. D. J. Biomed. Mater. Res. 2004, 68A (2), 365-75. (14) Anseth, K. S.; Bowman C. N.; Brannon-Peppas, L. Biomaterials 1996, 17, 1647-57. (15) Santovena, A.; Alvarez-Lorenzo, C.; Concheiro, A.; Llabres, M.; Farina, J. B. Biomaterials 2004, 25, 925-31 (16) Bryant, S. J.; Anseth, K. S. J. Biomed. Mater. Res. 2002, 59, 6372. (17) Lee, K. Y.; Bouhadir, K. H.; Mooney, D. J. Biomaterials 2004, 25, 2461-66. (18) Milas, M.; Rinaudo, M.; Roure, I.; Al-Assaf, S.; Phillips, G. O.; Williams, P. A. Biopolymers 2001, 59, 191-204. (19) Weiss, C.; Suros, J. M.; Michalow, A.; Denlinger, J.; Moore, M.; Tejeiro, W. Bull. Hosp. Jt. Dis. Orthop. Inst. 1987, 47, 31-9. (20) Weiss, C.; Band, P. Clin. Podiatr. Med. Surg. 1995, 12, 497-517. (21) Ellman, G. L. Arch. Biochem. Biophys. 1958, 74, 443-450. (22) Butterworth: P. H. W.; Baum, H.; Porter, J. W. Arch. Biochem. Biophys. 1967, 118, 716-723. (23) Kirker, K. R.; Prestwich, G. D. J. Polym. Sci. Part B: Polym. Phys. 2004, 42, 4344-4356. (24) Kavanagh, G. M.; Ross-Murphy, S. B. Prog. Polym. Sci. 1998, 23, 533-62. (25) Li, J.; Xu, Z. J. Pharm. Sci. 2002, 91, 1669-77. (26) Vervoort, L.; Vinckier, I.; Moldenaers, P.; Van den Mooter, G.; Augustijns, P.; Kinget, R. J. Pharm. Sci. 1999, 88, 209-14. (27) Flory, P. J. Principles of polymer chemistry; Cornell University Press: Ithaca, NY, 1953. (28) Zhang, Y.; Ge, S.; Rafailovich, M. H.; Sokolov, J. C.; Colby, R. H. Polymer 2003, 44, 3327-32. (29) James, H. M.; Guth, E. J. Chem. Phys. 1947, 15, 669-83. (30) Chen, D. J.; Chiang, F. P. Exp. Mech. 1992, 32, 145-153. (31) Chen, D. J.; Chiang, F. P.; Tan, Y. S.; Don, H. S. Appl. Opt. 1993, 32, 1839-1849. (32) Guan, E.; Smilow, S.; Rafailovich, M.; Sokolov, J. Dermatology 2004, 208, 112-9. (33) Lutolf, M. P.; Hubbell, J. A. Biomacromolecules 2003, 4, 713-22. (34) Peppas, N. A.; Brannon-Peppas, L. J. Membr. Sci. 1990, 48, 28190. (35) Rubinstein, M.; Colby, R. H.; Dobrynin, A. V.; Joanny, J. F. Macromolecules 1996, 29, 398-406.

BM050361C

Rheological Characterization of in Situ Cross-Linkable Hyaluronan

Recommend Documents