Photochemical Control of Network Topology in PEG Hydrogels

Abstract

Hydrogels are often synthesized through photoinitiated step-, chain-, and mixed-mode polymerizations, generating diverse network topologies and resultant material properties that depend on the underlying network connectivity. While many photocrosslinking reactions are available, few afford controllable connectivity of the hydrogel network. Herein, we introduce a versatile photochemical strategy for tuning the structure of poly(ethylene glycol) (PEG) hydrogels using macromolecular monomers functionalized with maleimide and styrene moieties. Hydrogels are prepared along a gradient of topologies by varying the ratio of step-growth (maleimide dimerization) to chain-growth (maleimide-styrene alternating copolymerization) network-forming reactions. The initial PEG content and final network physical properties (e.g., modulus, swelling, diffusivity) are tailored in an independent manner, highlighting configurable gel mechanics and reactivity. These photochemical reactions allow high fidelity photopatterning and 3D printing and are compatible with 2D and 3D cell culture. Ultimately, this photopolymer chemistry allows facile control over network connectivity to achieve adjustable material properties for broad applications.

Graphical Abstract

In this work, maleimide- and styrene-functionalized poly(ethylene glycol) macromers are used to prepare photocrosslinked hydrogels along a continuum of network topologies, from step- to chain-growth. This simple approach provides precise control over hydrogel nanostructure, decoupling gel modulus and diffusivity from the initial polymer content and presenting new opportunities for (bio)materials science and the study of soft matter physics.

Introduction

In covalently crosslinked polymer networks, the crosslinking chemistry dictates bond connectivity with dramatic effects on material characteristics¹. Notably, in hydrogels synthesized from macromolecular monomers of fixed functionality and molecular weight, chain-growth polymerizations densify crosslinks, which embrittles materials and also decreases the functional group conversion required to reach the gel point compared to step-growth polymerizations². To complement these properties, other methods have emerged for toughening hydrogels including the development of interpenetrating and highly entangled networks^3-5. In addition to the mechanical consequences of network topology, mesh size is affected by crosslink structure with implications for hydrogel swelling, controlled release and diffusion of nutrients and secreted factors in hydrogel-based cell culture⁶. Further, there is significant interest in directing the topological structure of hydrogels for developing accurate models of the underlying reaction-structure-property relationships which inform rational material design^7-12.

Across the vast library of photocrosslinking reactions used in soft polymeric materials, the thiol-ene, thiol-acrylate, and acrylate polymerizations are among the most popular for their rapid and predictable reaction kinetics and structural control. Notably, these chemistries result in distinct topologies: where the thiol-ene reaction generates comparatively more homogenous step-growth networks, the acrylate family of polymerizations form long kinetic chains with comparatively lower molecular weight between crosslinks and increased network heterogeneity, and the radical-mediated thiol-acrylate mixed-mode reaction yields intermediate structures^13-16. However, radical-mediated thiol-acrylate reactions require careful tuning of the thiol to acrylate ratio, as the kinetics of acrylate homopolymerization typically result in incomplete conversion of thiols^17-19. In an alternative approach for generating mixed-mode crosslinks, West and coworkers developed alloc-containing diacrylate peptides to favor termination events in photopolymerized hydrogels, accessing lower-modulus chain-growth materials^20-22. Additionally, our group recently reported another such system wherein dithiolane-crosslinked hydrogels were varied from step-growth to chain-growth by incorporating increasing ratios of alkene-containing macromers, although this approach also resulted in changes in viscoelasticity²³. Consequently, the ability to photopolymerize crosslinked networks that vary along more of a continuum between step- and chain-growth without affecting other properties (e.g., viscoelasticity, soluble fraction) has remained challenging in both the biomaterials community and at large.

As a less popular step-growth reaction, the photo-reversible [2+2] maleimide cycloaddition was first applied for on-demand hydrogel synthesis and functionalization in 2017^24,25. In these studies, the authors showed both one-photon (1P) and two-photon (2P) activation of maleimides in the absence of exogenous photoinitiator (PI), in which [2+2] cycloadditions were induced by 365 nm light (1P) or 800 nm light (2P) but surprisingly not by 400 nm 1P or 730 nm 2P excitation. This same chemistry was also used to achieve dynamic wrinkling in elastomers and site-specific labeling in hydrogels^26,27. Maleimides are an incredibly versatile functional group, capable of dimerizing to [2+2] and 2-2’ adducts, polymerizing into poly(maleimides), forming alternating copolymers with styrene, and reacting with nucleophiles (thiol-Michael, aza-Michael), dienes (Diels-Alder), and tetrazoles and azirines (1,3-dipolar cycloadditions)^28-31. Herein, we exploit this suite of compatible functional groups to synthesize highly customizable photopolymer networks, specifically using maleimide dimerization and alternating copolymerization with styrene to tailor the network topology.

By rationally combining distinct populations of poly(ethylene glycol) (PEG) macromers functionalized with either maleimide or styrene, we demonstrate photochemical control over hydrogel nanostructure. Adding exogenous PI (lithium phenyl-2,4,6-trimethylbenzoylphosphinate; LAP) increases the reaction rate for both maleimide dimerization and maleimide-styrene alternating copolymerization such that hydrogels form with reduced (10 to 100-fold lower) light doses, including with 405 nm irradiation, compared to PI-free formulations. The comonomer reactivity and parallel reaction mechanisms lead to topologically distinct networks with unique properties, including gel moduli that vary 5 to 10-fold at fixed macromer content and diffusion rates of macromolecular cargo that provide insight into the underlying network architecture. The semi-orthogonal kinetics between rapid alternating copolymerization and slower homopolymerization of the reactive components are leveraged to preserve post-gelation reactivity for hydrogel labeling as demonstrated by high-fidelity patterning of multiple fluorophores. Finally, we incorporate cell-adhesive peptide ligands in networks with varied topology to facilitate human mesenchymal stem cell (hMSC) matrix interactions. In sum, the reactivity of maleimide and styrene yields PEG hydrogels with well-defined, molecularly tunable topologies and properties that should prove useful for fundamental studies in areas ranging from soft matter physics to cell mechanobiology³².

Results and discussion

Tuning polymer network topology via photoinitiated crosslinking of PEG-maleimide macromers

Maleimides are known to participate in a variety of crosslinking reactions and even function as PIs for radical polymerizations³³. As such, we reasoned that multifunctional PEG-maleimide macromers could serve as a combined photoinitiator and crosslinker to form hydrogel networks with both step-growth (PEG-maleimide dimerization, Fig. 1a) and chain-growth (PEG-maleimide alternating copolymerization with PEG-styrene, Fig. 1b) mechanisms under irradiation with 365 nm light. While this was indeed the case, we also found several limitations relating to maleimide reactivity, especially in phosphate-buffered saline (PBS; Supplementary Note and Supplementary Figs. S1 and S2). However, introducing exogenous photoinitiator (LAP) resulted in rapid crosslinking in PBS without altering hydrogel mechanics, and enabled photocrosslinking with 405 nm light, although the precise mechanism remains somewhat unclear and may be a combination of photosensitization in addition to radicals generated by photoinitiator fragmentation³⁴. Step-growth PEG-maleimide gels prepared by this route displayed visible darkening during photocrosslinking compared to gels prepared without LAP, which may be the result of absorptive products formed between initiator fragments and maleimide. With the compatibility of maleimide-based hydrogel crosslinking with buffered solutions and the improved kinetics of gel formation in mind, we elected to focus on the LAP-facilitated crosslinking approach and did not study the photoinitiator-free system further.

Fig. 1 |. Maleimide-based crosslinking for step- and chain-growth network architectures.

a,b, PI-mediated maleimide dimerization (a) and alternating copolymerization with styrene (b). For simplicity, PI fragments are excluded from the crosslink structures. c, Monofunctional PEG macromers (MW = 2000 Da) bearing maleimide and styrene functional groups are used as a model non-network forming system. d, Monitoring alkene conversion as a function of irradiation in the presence of PI (17 mM LAP) by NMR illustrates increased reaction rate in the alternating copolymerization (blue, 1:1 maleimide to styrene) relative to homopolymerization of individual functional groups (red). Maleimide and styrene protons are represented by squares and diamonds, respectively. Dashed lines are present to guide the eye. e, GPC of PEG standards (black dashed line) compared to the reactants and products from d. f, Tetra-functional PEG macromers (MW = 20,000 Da) bearing maleimide and styrene functional groups are used to form crosslinked hydrogels with varied network topology. g, In situ oscillatory rheology reveals rapid photocrosslinking of 15 wt% macromer solutions in the presence of 17 mM LAP under either 365 or 405 nm light, where chain-growth gels (1:1 maleimide to styrene) have a higher final modulus (n = 3 gels for each condition). Dashed line indicates plateau storage modulus after measurement was ceased.

To assess the conversion and degree of polymerization of maleimide and styrene moieties in aqueous conditions, we utilized monofunctional linear 2 kDa PEG macromers functionalized with these alkenes (Fig. 1c). These macromers were dissolved at a concentration of 30 mM in a solution of D₂O containing 0.5 wt% (17 mM) LAP, and ¹H NMR was employed to monitor disappearance of the alkene protons during irradiation with 365 nm light at 5 mW/cm². As expected, the favorable reactivity of alternating copolymerization between the electron-deficient maleimide and electron-rich styrene³⁰ resulted in significantly faster conversion of both alkenes in a 1:1 stoichiometric mixture compared to either molecule in isolation; the mixed macromers reached ~100% conversion in <100 s of irradiation while several minutes of exposure were required to reach similar conversion for each individual macromer (Fig. 1d). Next, the reactants and products of the NMR study were characterized with gel permeation chromatography (GPC) to determine an approximate degree of polymerization. Using linear PEG standards, we observed that the starting materials were ~2 kDa, and that chain-growth crosslinking resulted in products with dramatically increased molecular weight relative to their step-growth counterpart (Fig. 1e). We also confirmed that the maleimide-only reaction in the presence of LAP did not produce any higher-MW species using matrix-assisted laser desorption/ionization mass spectrometry, finding evidence of LAP fragment-conjugated dimers but not trimers. An additional NMR study using ethylmaleimide in CDCl₃ suggested the presence of cyclobutane dimers in systems with and without exogenous photoinitiator (Supplementary Fig. S3). Finally, 20 kDa tetra-PEG macromers functionalized with maleimide and styrene (Fig. 1f) were dissolved at 30 mM with respect to reactive endgroups (15 wt% macromer) in PBS containing 17 mM LAP and photocrosslinked with 365 nm or 405 nm light at 10 mW/cm², confirming that the in situ mechanics of step- vs. chain-growth PEG-maleimide hydrogels vary in accordance with the crosslink density expected from our GPC results, and that photoinitiation with LAP enables crosslinking with visible light (Fig. 1g). All step- and chain-growth hydrogels presented throughout the remainder of the study were prepared using these 20 kDa tetra-PEG macromers, although mechanics could be further tuned by altering the molecular weight and geometry of the starting materials.

Next, we tested how varying the ratio of maleimide- to styrene-functional tetra-PEG macromers would affect crosslink density and ensuing mechanical properties. We prepared five formulations of 15 wt% macromer ranging from 0% to 100% step-growth, with 0% step-growth (or 100% chain-growth) defined as a 1:1 ratio of maleimide to styrene (Fig. 2a). When polymerized with 405 nm light at 15 mW/cm² while monitoring network evolution with in situ oscillatory shear rheology, the hydrogels followed a predictable continuum of crosslinking kinetics and moduli, with decreasing time to reach a plateau in storage modulus and increasing stiffness as the fraction of chain-growth crosslinks increased (Fig. 2b). Rheological characterization was repeated after 24 hours of equilibration in water for these same hydrogel formulations, and gels were re-weighed to determine the mass swelling ratio. As with the in situ rheological measurements, hydrogel properties varied along a gradient of moduli and swelling ratios (Fig. 2c), with 100% step-growth gels having the lowest storage modulus (G′ ~ 9±0.5 kPa) and highest mass swelling ratio (q ~ 30±2) and 0% step-growth gels having the highest storage modulus (G′ ~ 38±4 kPa) and lowest swelling ratio (q ~ 10±0.5).

Fig. 2 |. Tuning physical properties of photocrosslinked hydrogels along a step- to chain-growth continuum.

a, Five formulations of PEG hydrogel precursors varying across a gradient of step- and chain-growth topologies. b, In situ oscillatory rheology of the formulations in a at a fixed polymer content of 15 wt%. Inset shows the same data with linear scaling on the y-axis. Dashed lines indicate plateau storage moduli after measurement was ceased. c, Storage moduli (G’) and equilibrium mass swelling ratio (q) of hydrogels in b after 24 hours of equilibration in water. Error bars represent standard deviation. d, Tetra-PEG macromers functionalized with thiol, norbornene, or acrylate used as model photopolymerizations for preparing step- and chain-growth hydrogels. e,f, In situ monitoring of hydrogel formation via thiol-ene and acrylate polymerizations with 0, 50, and 100% step-growth conditions overlaid in the dashed lines (e) and rheological and diffusion-based characterization of equilibrated gels (f) reveal consistent network properties between maleimide-based and conventional photocrosslinking strategies. n.s. = not significant, *p < 0.05, **p < 0.01, multiple unpaired t tests with Welch correction. g, Varying macromer content between 2.5-15 wt% results in an approximate 10-fold variation in equilibrated hydrogel storage modulus while tuning network topology affords the same change in mechanics at fixed macromer content (n = 3 gels for all formulations). Dashed lines in c and g are present to guide the eye.

Given the ubiquity of thiol-ene (step-growth) and acrylate (chain-growth) photopolymerizations in modern hydrogel-based biomaterials science (Fig. 2d), we next sought to compare the maleimide/styrene-based crosslinking modalities to these more conventional systems. Using 20 kDa tetra-PEG macromers functionalized with thiols, norbornenes, and acrylates, in situ rheology was again applied to follow the kinetics of network formation and hydrogel mechanics. PEG-acrylate homopolymerization resulted in similar plateau moduli to the 0% step-growth condition and thiol-ene crosslinking yielded mechanics akin to the 100% step-growth condition, although the maleimide crosslinking reaction required longer irradiation times to reach the gel point (i.e., crossover between G′ and G″), indicating lower reactivity and limited propagation of radicals in the maleimide-based networks compared to thiol-ene polymerizations (Fig. 2e). Moreover, increasing photoinitiator concentration resulted in reduced plateau moduli in the thiol-ene reaction compared to the maleimide dimerization reaction, suggesting that endcapping of thiols or norbornenes by LAP fragments may be more detrimental to this step-growth crosslinking modality than maleimide dimerization. Notably, copolymerization of PEG-maleimide and PEG-norbornene resulted in hydrogels with very similar mechanical properties to the maleimide-only crosslinking mode, but with a significantly decreased time to the gel point (Supplementary Fig. S4), which we attribute to improved radical propagation from norbornene compared to maleimide. Additional screening of reactivity between maleimides and other electron-rich alkenes may give rise to further opportunities for tuning the kinetics of hydrogel formation and the final network connectivity/topology.

To characterize the network structures produced by our maleimide/styrene crosslinking approach relative to thiol-ene and acrylate polymerizations, we estimated an idealized mesh radius^35-38 of the swollen gels based on their equilibrium storage modulus. We then determined the diffusion coefficient of 10 kDa PEG-FITC in these gel formulations using fluorescence recovery after photobleaching (FRAP) measurements. Consistent with our rheological results, 0% step-growth and acrylate polymerizations had similar mesh radius and resulting diffusivity of the PEG-FITC, as did 100% step-growth and thiol-ene crosslinking reactions (Fig. 2f). These networks are a promising platform for tuning diffusivity of solutes in a length-scale-dependent manner³⁹, as other work has shown that both mesh radius and polymer- and solute-specific interactions can affect diffusion in PEG hydrogels^40,41. We also prepared hydrogels with 2.5 and 7.5 wt% macromer for thiol-ene, acrylate, and 0% and 100% step-growth polymerizations, observing equilibrium storage moduli that varied by nearly an order of magnitude at fixed initial macromer content in gels with varied ratios of maleimide to styrene (Fig. 2g). Taken together, these results demonstrate that maleimide-based crosslinking grants access to a broad continuum of hydrogel network topologies without requiring synthesis of macromers of varying molecular weight or functionality or sacrificing conversion of any reactive group.

Semi-orthogonal kinetics enable selective and sequential labeling of maleimide/styrene-crosslinked hydrogels

Many hydrogel applications require post-modification of the network, whether this is done to incorporate proteins or peptide ligands^42,43, create regions of variable stiffness⁴⁴, or for purposes such as encrypting information⁴⁵. We tested three 405 nm light intensities (15, 7.5, and 2.5 mW/cm²) to quantify how the initiation rate with 17 mM LAP affects the final storage modulus in hydrogels synthesized with 15 wt% macromer as a proxy for conversion in these crosslinked networks. Consistent with the results of our NMR study (Fig. 1d), hydrogels formed by maleimide/styrene alternating copolymerization reached the gel point at light doses <0.1 J/cm² while the styrene homopolymerization and maleimide dimerization required between 3 to 20-fold longer irradiation times (Fig. 3a). As a step-growth reaction with limited propagation, maleimide dimerization naturally required the highest light dose and conversion to reach the gel point, and the network evolution (G′) as a function of time effectively collapsed into a single curve irrespective of intensity. Intriguingly, styrene homopolymerization was strongly dependent on light intensity compared to the copolymerization reaction with maleimide, with faster initiation rates delaying the gel point. We next employed variable light doses to pattern freestanding microscale hydrogels using a laser scanning confocal microscope to deliver irradiation to regions of interest in step- and chain-growth gel precursors. Using fixed pixel dwells that varied 10-fold for step- (12.7 μs) versus chain-growth (1.27 μs) networks, both formulations were patterned with high fidelity (Fig. 3b).

Fig. 3 |. Maleimide and styrene photocrosslinking kinetics enable selective labeling and patterning applications.

a, Evolution of storage modulus during photocrosslinking of 15 wt% macromer solutions indicates decreasing reaction kinetics from alternating copolymerization to styrene homopolymerization to maleimide dimerization. b, Photopatterning of freestanding 50x50 μm hydrogels illustrate a 10-fold increase in light dose required to produce comparable step-growth gels relative to chain-growth gels. Gels contain 2 μM FITC-maleimide for visualization. c, Fluorescence micrograph of macroscale woven 3D hydrogel lattice prepared via DLP. Inset: photograph of DLP-printed construct. d, Hydrogels bearing pendant maleimide or styrene groups labeled by a fluorogenic PI-free tetrazole-ene photoclick reaction. e,f, Microscopic image of resolution test in a hydrogel with pendant maleimides patterned with the tetrazole-ene reaction (e). Quantification of resolution test showing high fidelity for positive and negative micron-scale features (f). g,h, Workflow for patterning alkene-bearing hydrogels with maleimide-functional molecules (g). Representative example of iterative hydrogel patterning with blue, green, and red maleimide-functional fluorophores to prepare a ~400 μm-wide reproduction of Da Vinci’s Mona Lisa (h). Panels (i-iv) are individual channels including brightfield illumination with merged fluorescence channels in panel (v).

Since the maleimide/styrene chain-growth reaction proceeds rapidly compared to either styrene homopolymerization or maleimide dimerization, we reasoned that either functional group could be retained in networks partially crosslinked by alternating copolymerization. Chain-growth hydrogels were prepared with 50% excess styrene or maleimide (i.e., a 3:1 ratio of maleimide:styrene and vice versa) and photocrosslinked in the presence of 17 mM LAP for 20 s with 405 nm light at 15 mW/cm², and post-functionalized with fluorescent PEG-thiol by addition of base (thiol-Michael) in gels containing excess maleimide or addition of LAP and light (thiol-ene) in gels containing excess styrene (Supplementary Fig. S5). Thus, these networks can be designed to retain specific reactivity via the semi-orthogonal kinetics of homo- and copolymerization. Next, we used a commercial digital light processing (DLP) apparatus to fabricate structured hydrogels from a 9:1 maleimide:styrene macromer formulation, which resulted in gels with primarily step-growth crosslinks but decreased time to reach the gel point due to the fast-reacting chain-growth component (Fig. 3c, Supplementary Fig. S6 and S7), allowing for crosslink density, modulus, and swelling ratio to be systematically varied by simply increasing the irradiation time without needing to change photopolymer functionality or content. These results collectively show that the maleimide/styrene photocrosslinking approach is easily extended to a variety of light-based lithography and additive manufacturing techniques, rendering this strategy useful for diverse applications where a range of properties can be achieved from a single photopolymer system.

Because photoinitiation with LAP results in crosslinking reactions between pendant alkene functionalities, we also implemented an alternative strategy for spatiotemporal labeling of either maleimides or styrenes in the absence of exogenous radicals. To achieve this, we employed the tetrazole-ene photoclick reaction⁴⁶, which produces self-reporting fluorogenic cycloadducts under near-UV light (Fig. 3d). Returning to patterning with the laser scanning confocal microscope, positive and negative features ranging from 1–50 μm in their smallest dimension were patterned into a hydrogel containing pendant maleimides (Fig. 3e). Quantification of feature resolution was within 1–2 μm in all intended dimensions (x-y), although negative features were slightly undersized, likely due to diffusion of activated tetrazole outside of the pattern boundaries (Fig. 3f). Finally, we demonstrated iterative patterning in a hydrogel containing excess maleimide (Fig. 3g) using three maleimide-functional fluorophores to generate a red-green-blue reproduction of the Mona Lisa (Fig. 3h). This image demonstrates that tuning light (and subsequent radical) dose allows for multiple patterning steps, even in overlapping regions of interest, via controlled conversion of the reactive macromer endgroups.

Applying maleimide-crosslinked hydrogels as cellularized biomaterials

We next asked whether the PEG-maleimide hydrogel-forming reactions were compatible with 2D and 3D cell culture for exploring cellular responses to network topology. Using hMSCs as a model cell type, we found that cells readily adhered to hydrogels functionalized with a CRGDS peptide, which mimics the integrin-binding domain of fibronectin (Fig. 4a). By culturing hMSCs on stiffness-matched (G’ ~ 10 kPa) step-growth, chain-growth (i.e., the 0% step-growth formulation), and PEG-styrene-only gels containing CRGDS (Supplementary Fig. S8), we observed no distinct trends in single-cell morphology, as measured by cell area and aspect ratio after 24 hours of culture (Fig. 4b). These results suggest that the morphology of individual cells is minimally affected by the underlying hydrogel network topology, at least on short timescales. Although previous work comparing hMSC responses to stiffness-matched thiol-ene and acrylate hydrogels showed dramatic differences in cellular adhesion and spreading, this was likely attributable to inconsistencies in RGD peptide incorporation and accessibility⁴⁷. We also used a maleimide-RGD⁴⁸ to photopattern cell adhesive regions into PEG networks containing pendant maleimides (in the same manner as Fig. 3), allowing spatial control over attachment of hMSCs (Fig. 4c). Post-functionalization reactions combined with tailored network topology suggests versatility of these materials for applications where directing spatiotemporal deposition of cells in material microenvironments is desirable. Finally, hMSCs were encapsulated in step- and chain-growth gel formulations prepared with 7.5 wt% macromer and 8.5 mM LAP before staining with calcein AM (live cells) and ethidium homodimer (dead cells) to probe cell viability in the presence of these reactions. We observed high (>90%) hMSC viability in both conditions several hours after encapsulation (Fig. 4d), meaning that these photochemical reactions are compatible with light-based biofabrication techniques. However, as the step- and chain-growth hydrogels were prepared with matched polymer wt%, these hydrogels have different final properties, with increased moduli in the chain-growth gels compared to the step growth gels. While long-term 3D culture of cells encapsulated in these matrices was beyond the scope of this study, these differences in physical properties would have a dramatic effect on cellular responses including motility and differentiation if these hydrogels were prepared with degradable crosslinks to permit cellular remodeling of the network. In a final demonstration, hMSCs stained with different colors of CellMask^™ were sequentially encapsulated via photopatterning with a confocal microscope (Fig. 4e), illustrating that the maleimide-based photocrosslinking is amenable to biomaterial applications involving bulk or structured sequestration of cells in both two and three dimensions.

Fig. 4 |. Maleimide-based hydrogel photocrosslinking is compatible with multiple biomaterial implementations.

a,b, Application of maleimide- and styrene-functional hydrogels as 2D biomaterials for culture of hMSCs via incorporation of CRGDS (RGD-SH) through thiol-Michael and thiol-ene crosslinking. hMSCs adhere and spread on PEG-maleimide hydrogels containing CRGDS (a). Hydrogels with matched stiffness (G’ ~ 10 kPa) prepared with maleimide dimerization (red), styrene homopolymerization (black), and maleimide/styrene alternating copolymerization (blue) display similar hMSC area and aspect ratio (b; n = 3 gels per formulation). c, Photopatterned RGD-maleimide results in selective adhesion of hMSCs to RGD-functionalized regions of the hydrogel. d, 3D encapsulation of hMSCs into step- and chain-growth maleimide-based hydrogels shows high (>90%) viability after photocrosslinking (n = 3 gels per condition). e, 3D encapsulation is easily extended to multi-stage microscale biofabrication; hMSCs stained with different CellMask^™ labels are used in each stage to illustrate user-dictated spatial distribution of cells.

Conclusions

PEG hydrogels were prepared along a continuum of network topologies which varied from step- to chain-growth using maleimide-based photocrosslinking. Addition of LAP to the gel precursors improved crosslinking kinetics and compatibility with different environmental conditions, and extended accessible wavelengths for initiating these reactions. By tuning network connectivity, storage modulus was varied over an order of magnitude with fixed initial macromer content. Preparing off-stoichiometry hydrogels allowed for retention of unreacted functional groups and subsequent post-gelation network modifications. We demonstrated that these reactions are compatible with photolithography approaches including laser stereolithography and DLP, expanding the range of chemistries commonly used with these technologies. Finally, these hydrogels were implemented as cellularized matrices in 2D and 3D. This strategy is a highly versatile method to prepare topologically tailored functional polymer networks for a diverse array of soft matter applications.

Materials

All PEGs were purchased from JenKem Technology USA and used as received unless otherwise noted. All other reagents were purchased from Sigma and used as received unless otherwise noted. Plots were made in Python and GraphPad Prism; statistics were performed in GraphPad Prism.

Methods

Macromer synthesis

To prepare mono- and tetra-functional PEG-styrene, mPEG-amine (2 kDa) and 4-arm PEG-amine (tripentaerythritol core, 20 kDa) were dissolved in DMF at a concentration of 0.1M with respect to amino end groups. HATU (4 eq. per end group) and vinylbenzoic acid (4 eq. per end group) were added to this solution and stirred for 10 minutes at room temperature before N-methylmorpholine (8 eq. per end group) was added dropwise to initiate amide coupling. These reactions were stirred overnight before precipitation in excess ice-cold diethyl ether, centrifugation, and dialysis in sealed regenerated cellulose tubing (SpectraPor) with daily water changes for 3 days. Dialyzed macromer was flash-frozen in LN2 and lyophilized on a Benchtop Pro freeze dryer (SP Scientific) until all water content was removed from the macromer (3-7 days). Quantitative functionalization of amine groups with styrene moieties was observed via ¹H NMR in CDCl₃ (20 mg macromer per 500 uL deuterated solvent; Supplementary Fig. S9). PEG-norbornene was prepared as previously described²³ in the same manner as PEG-styrene using norbornene carboxylic acid in place of vinylbenzoic acid.

Chemical characterization of macromers and crosslinking reactions

All NMRs were acquired using a 400 MHz Bruker spectrometer. For measuring alkene conversion over time, replicate samples of mPEG-maleimide and styrene (and a 1:1 mixture of the two polymers) in D₂O containing 17 mM LAP were placed under a collimated 365 nm light source (Omnicure) set to an intensity of 5 mW/cm² and removed at specified timepoints. After exposure to light, NMR spectra were obtained and peaks representing single protons of maleimide and styrene carbon-carbon double bonds were integrated and analyzed for double bond conversion. ¹H NMR (400 MHz, D₂O, δ): 6.79-6.81 (s, ¹H, maleimide proton), 5.85-5.95 (d, ¹H; styrene proton). GPC was performed using a EcoSEC HLC-8230 (Tosoh) and Agilent PEG standards, with samples prepared at concentrations 1-5 mg/ml in water and passed through 0.2 μm PTFE filters. Absorbance measurements were performed using a Thermo Scientific Evolution 300 UV-Vis spectrophotometer using LAP and tetra-PEG-maleimide dissolved at 1, 5, and 10 mM in water. Measurements were collected at wavelengths 250-500 nm with a resolution of 0.5 nm through quartz cuvettes (path length = 1 cm). Beer-Lambert law was used to calculate the molar extinction coefficient at each concentration, which were subsequently averaged to yield the absorptivity spectrum. MALDI-TOF was performed using a Shimadzu 8020 mass spectrometer by diluting polymers (prepared in the same manner as for NMR and GPC) in acetonitrile to 10 mg/mL and spotting them sequentially with dithranol (5 mg/mL) and sodium trifluoroacetate (10 mg/mL) on a stainless-steel plate. MALDI data was acquired using positive ion and linear mass tuning modes; observed molecular weight distributions were analyzed within the Shimadzu Data Acquisition application utilizing minimal baseline correction and smoothing. For GPC, NMR, and MALDI, photocrosslinking of the 2 kDa monofunctional PEG-maleimide and PEG-styrene was performed at concentrations 15-30 mM (30-60 mg/mL) in water or D₂O before dilution for further analysis.

Hydrogel formulation and photocrosslinking

Tetra-functional 20 kDa PEG macromers bearing maleimide, styrene, thiol, norbornene, and acrylate end groups were dissolved in sterile PBS at 20 wt% and mixed and diluted as specified in the text for each experiment. For experiments using high concentrations of LAP (>17 mM), macromer was directly dissolved in LAP solution at the desired concentration. LAP stocks were dissolved in sterile PBS at 68 mM. Hydrogel photocrosslinking was performed using a variable-intensity collimated light source (Omnicure) with 365 nm and 400-500 nm filters. CRGDS and RGD-maleimide were synthesized via Fmoc solid phase peptide synthesis with a Tribute peptide synthesizer (Protein Technologies) and characterized using mass spectrometry as previously described⁴⁸. Hydrogels for 2D cell culture were prepared with 2 mM RGD.

Physical characterization of hydrogels

Oscillatory shear rheology was performed using a DHR-30 rheometer (TA Instruments). For in situ photorheology, the instrument was equipped with an 8 mm parallel plate and a quartz stage connected to a light source via a liquid light guide. For swollen rheology, the instrument was equipped with a sandblasted 8 mm parallel plate and fine-grit sandpaper was fixed to the quartz stage to prevent sample slippage. All measurements were acquired using 1% strain and a frequency of 1 rad/s. The idealized mesh radius was calculated using by estimating the idealized network mesh size from the equilibrium storage modulus^35-37 (ξ = [6RT/G’πN_av]^1/3) and multiplying this value by a correction factor of 0.816 to account for tetra-functional junctions³⁸. Hydrogels were weighed immediately after curing, following equilibration in water or PBS for 24 hours, and after drying to calculate the mass swelling ratio (q, defined as dry polymer mass/equilibrated hydrogel mass). Fluorescence recovery after photobleaching (FRAP) was performed as previously described⁴⁹. Briefly, pre-formed hydrogels were equilibrated overnight in fluorophore solution (5 μM 10 kDa PEG-FITC) before performing photobleaching (100% laser power, circular bleach area with 50 μm radius) and acquiring fluorescence baseline and recovery measurements (2% laser power) at 1 frame per second for 90-120 seconds. To calculate the diffusion coefficient, recovery curves were fit to an exact closed formula describing fluorescence recovery to calculate the characteristic timescale of diffusion⁵⁰.

Hydrogel photolithography, photopatterning, and DLP

Laser micropatterning was performed using a Zeiss LSM 710 confocal microscope equipped with a 10X air objective and 405 nm laser (maximum output power ~1 mW). Custom digital photomasks prepared in MATLAB⁵¹ were applied to samples in the ZEN software “regions” module and patterned with 405 nm light as specified in the text. Patterning fidelity was quantified by measuring patterned regions in ImageJ. For photopatterning maleimide-conjugated fluorophores and naphthyl tetrazole, small molecules were dissolved in DMSO and diluted in water (for maleimide fluorophores, also containing LAP). Hydrogels were equilibrated in these patterning buffers for at least 10 minutes before laser scanning. Naphthyl tetrazole was synthesized as previously described⁵² and verified by observing a fluorogenic reaction with acrylamide under illumination with 405 nm light, although interested readers should note that a more recently published route utilizing an alternative solvent is better-optimized for this synthesis⁵³. 3D DLP printing was performed on a Lumen Alpha bioprinter (Volumetric, Inc.) with a 100 μm step size using 405 nm light at 15 mW/cm². No photoabsorber was added to the macromer solution, which was composed of 15 wt% tetra-functional 20 kDa PEG at a ratio of 9 maleimide:1 styrene endgroups dissolved in 17 mM LAP; to better visualize samples, 2 μM FITC-maleimide was added to the macromer solution. After printing was complete, samples were gently removed from the build plate and swelled in PBS overnight before imaging on a Nikon TE2000 fluorescence microscope. For 2D light projection patterning using the DLP apparatus and laser micropatterning, macromer solution was placed on the surface of an acrylated coverslip⁵⁴ to ensure strong adhesion between printed hydrogel objects and the substrate.

hMSC culture, encapsulation, and labeling

hMSCs were isolated from female human bone marrow (Lonza) and cultured at 37°C and 5% CO₂ on tissue culture polystyrene in low-glucose DMEM supplemented with 10% FBS, 50 U/ml penicillin/streptomycin, and 1 μg/mL amphotericin B. hMSCs were cultured up to passage 4 before encapsulation. Before encapsulation, hMSCs were resuspended at twice the desired concentration and mixed 1:1 with 15 wt% macromer containing 17 mM LAP. For live/dead quantification, cells were stained with 2 μM calcein AM and ethidium homodimer and 10 μM Hoechst 33342 for at least 20 minutes before imaging. Cell morphometry was performed by thresholding the live channel of live/dead images in ImageJ, performing watershed segmentation, and measuring cell area and aspect ratio using the “Analyze Particles” tool. For visualization of the actin cytoskeleton, cells were fixed in 4% paraformaldehyde, washed with PBS, and incubated with rhodamine phalloidin (1:500, Invitrogen). Cells labeled with CellMask^™ (Thermo Fisher Scientific) before 3D encapsulation were incubated in staining solution (1:1000) for 20 minutes before mixing with macromer.