Spatial control of viscoelasticity in phototunable hyaluronic acid hydrogels

Viscoelasticity has emerged as a critical regulator of cell behavior. However, there is an unmet need to develop biomaterials where viscoelasticity can be spatiotemporally tuned to mimic the dynamic and heterogeneous nature of tissue microenvironments. Toward this objective, we developed a modular hyaluronic acid hydrogel system combining light-mediated covalent and supramolecular crosslinking to afford spatiotemporal control of network viscoelastic properties. Covalently crosslinked elastic hydrogels or viscoelastic hydrogels combining covalent and supramolecular interactions were fabricated to match healthy and fibrotic liver stiffness. LX-2 human hepatic stellate cells cultured on viscoelastic substrates displayed reduced spreading, less actin stress fiber organization, and lower myocardin-related transcription factor A (MRTF-A) nuclear localization compared to cells on elastic hydrogels. We further demonstrated the dynamic capabilities of our hydrogel system through photomediated secondary incorporation of either covalent or supramolecular crosslinks to modulate viscoelastic properties. We used photopatterning to create hydrogel models with well-controlled patterned regions of stiff elastic mechanics representing fibrotic tissue nodules surrounded by regions of soft viscoelastic hydrogel mimicking healthy tissue. Cells responded to the local mechanics of the patterned substrate with increased spreading in fibrosis-mimicking regions. Together, this work represents an important step forward toward the creation of hydrogel models with spatiotemporal control of both stiffness and viscoelastic cell-instructive cues.

an important step forward toward the creation of hydrogel models with spatiotemporal control of both stiffness and viscoelastic cell-instructive cues.

Introduction
The interplay between cells and their surrounding extracellular matrix (ECM) plays a critical role in regulating development, wound healing, and disease progression [1][2][3]. Through mechanisms such as mechanotransduction, a process in which mechanical forces are converted into biochemical signals, cells are constantly probing and responding to their evolving microenvironment [4]. Cell-ECM interactions are especially important in pathologies such as fibrosis, a heterogeneous pathological scarring process that can lead to irreversible loss of tissue function and organ failure. During fibrosis progression, healthy tissue mechanics transition from softer and viscoelastic to stiffer and less viscous [5,6]. Moreover, fibrosis progresses in a heterogeneous manner, leading to microscale spatial heterogeneity in the form of patchy, stiff fibrotic nodules surrounded by areas of softer, less affected tissue where nodule size often directly correlates with the severity of fibrosis [7][8][9]. The presence of a stiff microenvironment can guide mechanotransduction by providing necessary biophysical cues for activation of resident cells into fibrosis-promoting myofibroblasts [10], and elevated stiffness alone has been shown to drive progression of both fibrosis [11] and cancer [12].
Hydrogels have become valuable model systems to better understand the contributions that matrix biophysical properties play in regulating cell behaviors through their ability to mimic salient properties of natural tissue, including soft tissue mechanics and high water content [13,14], and numerous systems have already investigated the influence of hydrogel mechanics on cell behavior [3,9,[15][16][17][18][19][20][21]. In particular, many groups have shown a direct correlation between increasing hydrogel Young's modulus (stiffness) and elevated cell spreading in two-dimensional (2D) cultures [10,[22][23][24][25]. Although many studies have developed homogenous substrates to study cell-ECM interactions, healthy and especially diseased tissues are inherently heterogeneous. During pathologies such as fibrosis, changes in the physical environment have direct implications on cell mechanotransduction, where activated cell patches begin depositing excessive amounts of ECM proteins, resulting in nodules of nonfunctional scar tissue [7]. Therefore, it is necessary to develop methods to recapitulate tissue heterogeneity in hydrogel models. Recent work using light-based chemistries to spatially pattern elastic substrates has shown that cells will exhibit behavior correlating to their local mechanics such as increased spreading on stiffer areas [9,26,27].
While these findings are informative, they typically involve covalently-crosslinked hydrogels that primarily behave as elastic solids and do not display time-dependent tissue-relevant mechanical properties. The majority of native tissues exhibit viscoelastic behaviors including stress relaxation, which can occur through both external and cell-mediated forces exerted onto the matrix. For this reason, viscoelasticity has recently emerged as a critical parameter for probing cell behaviors and functions. Viscoelastic hydrogels have been developed using ionic [16,17], supramolecular [28], and dynamic covalent crosslinking [29] mechanisms. Viscoelastic hydrogels with stress relaxation properties similar to native tissues have been shown to affect cell spreading, focal adhesion organization, proliferation, and differentiation in comparison with elastic hydrogels [16][17][18][19]30,31]. This can be attributed in part to cell-mediated reorganization and/or relaxation of the energy-dissipative viscoelastic hydrogel network. Recent work from Charrier et al. [19] showed changes in the behavior of hepatic stellate cells, the primary cellular source of hepatic myofibroblasts, when cultured on viscoelastic hydrogels. Stellate cells displayed lower spread area and reduced expression of α-smooth muscle actin (α-SMA), a hallmark of myofibroblast activation, with increasing hydrogel loss modulus [19]. This study highlighted the importance of how hydrogel viscoelasticity can regulate disease-relevant cellular behaviors.
While the importance of incorporating viscoelasticity into hydrogel cellular microenvironments is clearly established, an approach to spatially control viscoelastic properties in a manner that mimics heterogeneous tissue has not been developed. The ability to pattern regions of hydrogel stiffness and/or viscoelasticity in a manner that captures both the dynamic stiffening that occurs during fibrosis progression and the overall heterogeneity of fibrotic tissue would help establish more robust disease models to study pathological cell behaviors. Here, we designed a phototunable viscoelastic hydrogel system where stiffness and viscoelasticity can be independently tuned through control of network covalent and supramolecular interactions. Using this modular approach, we developed photopatterned substrates where stiffness and viscoelasticity could be spatially controlled and investigated the role that matrix mechanical properties played in regulating cell behavior in an in vitro model of fibrosis.

2.2.
-CD-HDA synthesis. The synthesis of β-cyclodextrin hexamethylene diamine (β-CD-HDA) followed the procedure outlined previously [33]. p-Toluenesulfonyl chloride (TosCl) was dissolved in acetonitrile and added dropwise to an aqueous β-cyclodextrin (CD) suspension (5:4 molar ratio of TosCl to CD) at 25°C. After 2 hours, the solution was cooled on ice and an aqueous NaOH solution was added dropwise (3.1:1 molar ratio of NaOH to CD). The solution was reacted for 30 minutes at 25°C before adding ammonium chloride to reach a pH of 8.5. The solution was cooled on ice, precipitated using cold water and acetone, and dried overnight. The CD-Tos product was then charged with hexamethylene diamine (HDA) (4 g/g CD-Tos) and dimethylformamide (DMF) (5 mL/g CD-Tos), and the reaction was carried out under nitrogen at 80°C for 12 hours before being precipitated with cold acetone (5 × 50 mL/g CD-Tos), washed with cold diethyl ether (3 × 100 mL), and dried. The degree of modification was 61% as determined by 1 H NMR ( Figure S2).

2.3.
-CD-HA synthesis. β-cyclodextrin modified hyaluronic acid (β-CD-HA) was prepared through coupling of β-CD-HDA to HA-TBA. A reaction containing HA-TBA, 6-(6aminohexyl)amino-6-deoxy-β-cyclodextrin (β-CD-HDA), and BOP in DMSO was carried out at 25°C for 3 hours. The reaction was quenched with cold water, dialyzed for 5 days, filtered, dialyzed for 5 more days, frozen, and lyophilized. The degree of modification was 27% as determined by 1 H NMR ( Figure S3).  2.6. Rheological characterization. All rheological measurements were performed at 25°C on an Anton Paar MCR 302 rheometer using a cone-plate geometry (25 mm diameter, 0.5°, 25 µm gap). Rheological properties were tested using oscillatory time sweeps (1 Hz, 1% strain) with a 2 minute UV irradiation (5 mW/cm 2 ), oscillatory frequency sweeps (0.001-10 Hz, 1% strain), and cyclic stress relaxation and recovery tests alternating between 0.1% and 5% strain (1 Hz). [34], Millipore Sigma) were used between passages 6-8 for all experiments. Culture media contained Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10 v/v% fetal bovine serum (FBS, Gibco) and 1 v/v% penicillin/streptomycin/amphotericin B (10,000 U/mL, 10,000 g/mL, and 25 g/mL respectively, Gibco). For cell seeding, swelled thin film hydrogels (18 x 18 mm) were sterilized using germicidal UV irradiation for 2 hours and incubated in culture media for at least 30 minutes prior to cell seeding. Cultures were treated with 5-10 μg/mL mitomycin C (Sigma-Aldrich) in serum-free media for 2 hours, washed thrice with PBS, and incubated in complete culture media for at least 1 hour prior to cell seeding. Cells were seeded atop hydrogels placed in untreated 6-well plates at a density of 2 x 10 4 cells per hydrogel. For all experiments, media was replaced every 2-3 days for 7 day cultures.
The hydrogels were washed 3 times in PBS and incubated with secondary antibodies (AlexaFluor 488 goat anti-rabbit IgG, 1:800; AlexaFluor 555 goat anti-mouse, 1:800) for 2 hours in the dark at room temperature. The hydrogels were then rinsed 3 times with PBS and stained with a DAPI nuclear stain (1:10000) for 1 minute before rinsing twice with 3% BSA. Stained hydrogels were stored in the dark at 4°C until imaging. Microscopy was performed on a Zeiss where is the cell area and is the cell perimeter. MRTF-A nuclear/cytosolic ratio was determined using the formula: where the signal intensities were taken and normalized to their respective areas. Force relaxation tests were performed to study viscoelasticity of patterned substrates. Following indentation, the tip was held at a constant indentation depth for 10-30 seconds at a 500 Hz sampling rate. Indentation force and depth were recorded as a function of time [35]. Significance was indicated by *, **, or *** corresponding to P < 0.05, 0.01, or 0.001 respectively.

Viscoelastic hydrogels were synthesized with a combination of covalent and supramolecular
crosslinks. Hyaluronic acid was functionalized with norbornene groups (NorHA) to produce hydrogels containing a high degree of reactive sites (~ 20% of repeat units). Compared to common functional groups such as (meth)acrylates, which can react with each other to form kinetic chains, norbornene groups have high reactivity to thiyl radicals and low reactivity to themselves, allowing rapid and controllable thiol-ene click addition of both pendant and multifunctional thiolated groups [32]. This biorthogonal system was also chosen for its ability to easily synthesize hydrogels with a wide range of tissue-relevant mechanics by simple tuning of parameters such as crosslinker concentration or light intensity. In this study, soft (G' ~ 0.5 kPa) and stiff (G' ~ 5 kPa) thin film hydrogels were fabricated to represent healthy and fibrotic liver tissue, respectively [10,11]. Di-thiol molecules (DTT) were used to provide stable hydrogel crosslinks and thiolated RGD peptide (GCGYGRGDSPG) was incorporated to allow for cell attachment. Elastic NorHA hydrogels were fabricated via ultraviolet (UV)-light mediated thiolene addition between norbornenes on HA and thiols on DTT to create stable covalentlycrosslinked networks.
Viscoelasticity was introduced to the system by incorporating reversible guest-host interactions between adamantane (guest) and β-cyclodextrin (host) groups. The adamantane (Ad) guest moiety has a high affinity to the hydrophobic cavity of β-cyclodextrin ( ~ 10 5 M -1 ) and has previously been exploited to make viscoelastic, shear-thinning hydrogels [33,36,37]. For the viscoelastic hydrogel groups, β-cyclodextrin HA (CD-HA) and thiolated Ad peptide were mixed in solution (1:1 molar ratio of CD to Ad) to introduce supramolecular guest-host interactions, followed by the addition of NorHA and DTT (Figure 1). This particular methodology involving Ad peptide allowed for a more modular approach to fabricating hydrogels due to its detachment from the HA backbone prior to the thiol-ene addition, making the hydrogel precursors less viscous and easier to pipet and mix. Following mixing of the Ad peptide, CD-HA, NorHA, and DTT, the thiols on the cysteine residues of the CD-associated Ad peptide reacted with the norbornenes to form stable supramolecular connections between HA chains while the DTT formed covalent crosslinks, creating a viscoelastic hydrogel network with both covalent and supramolecular crosslinks.

Figure 1. Overview of hydrogel synthesis and crosslinking. (A)
Hyaluronic acid was first converted to HA-TBA salt before modification with norbornene or -cyclodextrin groups using BOP coupling chemistry to synthesize NorHA and CD-HA. (B) For the elastic hydrogel system, covalent crosslinks between the norbornene groups were introduced using di-thiol crosslinkers via light-mediated thiol-ene addition. For the viscoelastic hydrogel system, thiol-ene photochemistry was used to introduce both stable supramolecular interactions between CD-HA and Ad groups on thiolated peptides in addition to di-thiol-mediated covalent crosslinks between the norbornenes.

Viscoelastic hydrogels display stress relaxation and frequency-dependent behavior.
Hydrogel mechanical properties were characterized through shear oscillatory rheology ( Figure   2). In situ gelation of hydrogel precursor solutions demonstrated rapid gelation kinetics controlled by light exposure, resulting in a nearly immediate plateau in storage and loss moduli once light irradiation was stopped (Figures 2A and 2B).  ). Notably, the G" values for the viscoelastic hydrogels were within an order of magnitude of their G', similar to the ratios observed in native viscoelastic tissue [19,38].
Hydrogel frequency sweeps revealed relatively constant storage and loss moduli for the elastic groups ( Figure 2C). However, the viscoelastic hydrogels showed frequency-dependent behavior; at higher frequencies, the loss modulus increased, demonstrating that guest-host interactions were being disrupted with less time to re-associate. Stress relaxation and recovery tests showed that at a constant applied strain of 5%, the elastic hydrogels showed no stress relaxation over time due to their stable covalently-crosslinked network ( Figure 2D). In contrast, the viscoelastic hydrogel groups showed cyclic stress relaxation in which high stress was observed, followed by a plateau to a final stress value equal to the corresponding elastic groups. The ability for the hydrogels to fully recover their mechanical properties upon repeated bouts of applied strain highlighted their viscoelasticity as opposed to viscoplasticity. For the frequency and stress relaxation tests, the soft groups are shown; similar trends were seen for the stiff groups ( Figure S6). to what has previously been reported for elastic substrates of increasing stiffness (Figure 3). In comparison the viscoelastic hydrogels, which had the same storage moduli as the corresponding elastic groups but higher loss moduli, supported decreased cell spreading and more rounded morphologies as measured by cell shape index (CSI) compared to their corresponding elastic substrates for both the soft and stiff groups. The differences in cell spreading and circularity were the greatest between cells cultured on the stiff elastic hydrogels, which became more elongated and extended protrusions (average spread area: 6623 m 2 , CSI: 0.17), and cells cultured on the soft viscoelastic hydrogels, which showed smaller, more rounded morphologies (average spread area: 2982 m 2 , CSI: 0.26). The reduction in stellate cell spreading is similar to results from a recent study where stellate cells showed reduced spreading and reduced expression of α-smooth muscle actin (α-SMA), a marker of myofibroblast activation, when cultured on polyacrylamide substrates with higher loss moduli [19]. Similarly, while cells on our viscoelastic hydrogels showed positive α-SMA staining, we also observed reduction in the organization of α-SMA stress fibers that is typical of activated myofibroblasts [19,39,40]. While around 85% of cells on stiff elastic hydrogels displayed at least some α-SMA stress fiber organization, only 6% and 22% of cells on soft and stiff viscoelastic hydrogels respectively displayed organized α-SMA stress fibers ( Figure S7).

Figure 3. Cell spreading is modulated by both stiffness and viscoelasticity. (A)
Representative images of LX-2 hepatic stellate cells stained for α-SMA (red) and nuclei (blue). Scale bar 100 m. (B) While cell spreading was increased on stiff elastic compared to soft elastic hydrogels, spreading was significantly reduced on both soft and stiff viscoelastic hydrogels compared to the stiff elastic group. (C) Cell shape index was significantly higher for cells on soft hydrogels compared to their respective stiff counterparts, indicating that the cells displayed more rounded morphologies. *: P < 0.05, **: P < 0.01, ***: P < 0.001.
Since α-SMA expression is a relatively late marker of myofibroblast activation, we also sought to investigate earlier markers of fibrogenic mechanotransduction. Myocardin-related transcription factor A (MRTF-A), a transcriptional co-activator implicated in the regulation and progression of fibrosis, has been shown to drive -SMA expression and subsequent myofibroblast activation [41][42][43]. Specifically, activation of mechanotransduction pathways through cell-matrix interactions can promote RhoA/ROCK signaling, actin polymerization, and subsequent MRTF-A nuclear translocation. MRTF-A then interacts with serum response factor (SRF), the transcription factor that promotes upregulation of the Acta2 gene encoding for α-SMA [42,[44][45][46]. We measured the ratio of MRTF-A nuclear to cytosolic signaling intensity and found elevated MRTF-A nuclear localization for cells on stiff compared to soft elastic hydrogels ( Figure S8). However, cells cultured on the viscoelastic hydrogel groups showed reduced MRTF-A nuclear localization compared to the stiff elastic group. Overall, both soft and stiff viscoelastic hydrogels promoted reduced stellate cell spreading, -SMA stress fiber organization, and MRTF-A nuclear localization. A possible explanation for these results could be the viscous dissipation of cell-generated forces into the matrix prevents spreading and activation of the mechanoresponsive signaling pathways investigated here [47].

Light-mediated thiol-ene addition enables secondary incorporation of covalent or
supramolecular crosslinks. We demonstrated the dynamic capabilities of our viscoelastic hydrogel system through specific secondary introduction of either covalent or supramolecular interactions. First, we fabricated elastic hydrogels with increasing covalent crosslinking density controlled by sequential bouts of light exposure, permitting further thiol-ene crosslinking.
Rheological analysis indicated that each additional irradiation corresponded with increasing storage modulus but relatively little change in loss modulus as expected for an elastic network ( Figure 4A). Next, we made initially soft viscoelastic hydrogels containing unreacted norbornene and -cyclodextrin groups and introduced additional supramolecular crosslinks through sequential thiol-ene addition of thiolated adamantane peptide. Each additional light irradiation led to increases in both the storage and loss moduli as the hydrogel maintained its viscoelastic nature ( Figure 4B). Overall, the unique amenability of our system to the light-mediated introduction of either new covalent or supramolecular crosslinks sets the stage for creation of dynamic, heterogeneous viscoelastic hydrogels. Fluorescence microscopy confirmed pattern fidelity, with alternating fluorescent and nonfluorescent regions present in the hydrogel (Figure 5B).

Figure 5. Photopatterning of hydrogels to introduce heterogeneous properties. (A)
Schematic of the photopatterning process. NorHA hydrogels were swollen with thiolated molecules, covered with a photomask, and exposed to UV light, resulting in regions that underwent secondary crosslinking via light-mediated thiol-ene addition. A model thiolated fluorescent peptide was used to demonstrate patterning capabilities. Color intensity profiles showed high pattern fidelity across pattern features for (B) 200 m diameter circles and (C) 200 m stripe patterns; signal intensity profiles were quantified along the white dotted lines. Scale bars: 500 m.
After establishing the photopatterning approach, we wanted to develop a patterned hydrogel model of fibrotic tissue. During the heterogeneous progression of fibrosis, the aberrant shift in healthy tissue mechanics from soft and viscoelastic to stiff and more elastic highlights the need for in vitro models enabling independent spatial control of both stiffness and viscoelasticity.
Given the ability for multiple light-mediated thiol-ene click reactions to occur in series,  Figures 6B, 6C), similar to bulk rheological measurements for homogeneous hydrogels.
Importantly, this novel method for patterning viscoelasticity can be decoupled from changing stiffness. As a demonstration of this, viscoelasticity can be patterned into a stiff elastic substrate through the introduction of supramolecular crosslinks to produce regions of patterned, stiff viscoelasticity without changing the overall Young's modulus (initial stiff elastic = 10.6  0.38 kPa, patterned stiff viscoelastic = 10.7  0.49 kPa) ( Figure S9). (E) Cells on the patterned (stiff elastic) region showed significantly increased spread area compared to those in the non-patterned (soft viscoelastic) regions. (F) Cells in the patterned regions also showed significantly lower cell shape index, indicating a more elongated morphology, compared to more rounded cells in the non-patterned regions. *: P < 0.05, **: P < 0.01, ***: P < 0.001.
Next, we seeded LX-2 stellate cells onto soft viscoelastic hydrogels with patterned regions of stiff elastic mechanics ( Figure 6D). Cells responded to the local mechanics of the patterned substrate and showed significantly increased spreading ( Figure 6E) and significantly lower cell shape index (Figure 6F) on stiffer patterned regions. These results demonstrate the utility of this hydrogel system as a model of heterogeneous tissue mechanics.

Conclusions
This work developed an approach to make viscoelastic hydrogels via light-mediated thiol-ene addition of both covalent and supramolecular crosslinks. The use of light as a trigger for crosslinking enabled secondary modification of the hydrogel network to both increase stiffness (mimicking initiation of fibrosis) and/or modulate viscoelasticity (through the introduction of covalent and/or supramolecular crosslinks). We showed that LX-2 human hepatic stellate cells responded to the viscoelastic hydrogels by displaying reductions in spread area, MRTF-A nuclear translocation, and organization of actin stress fibers. We also used photopatterning to create hydrogels with stiff, elastic areas surrounded by soft, viscoelastic regions to mimic a heterogeneous fibrotic environment and showed that cells spread more in the stiffer patterned regions. Moving forward, we expect that this hydrogel system affording spatiotemporal control of stiffness and viscoelasticity will be useful to model a range of healthy and diseased cellular microenvironments.
Supporting Information. 1 H NMR spectra for NorHA, CD-HDA, and CD-HA, MALDI spectra for peptides, and additional hydrogel mechanical characterization and cell analysis can be found in the supplemental file.

Notes
The authors declare no competing financial interest.