Clickable microgel scaffolds as platforms for 3D cell encapsulation

Abstract

While microporous scaffolds have been increasingly used for regenerative medicine and tissue repair applications, the most common techniques to fabricate these scaffolds use templating or top-down fabrication approaches. Cytocompatible bottom-up assembly methods afford the opportunity to assemble micro-porous systems in the presence of cells and create complex polymer-cell composite systems in situ. Here, microgel building blocks with clickable surface groups were synthesized for the bottom-up fabrication of porous cell-laden scaffolds. The facile nature of assembly allowed for human mesenchymal stem cells to be incorporated throughout the porous scaffold. Particles were designed with mean diameters of approximately 10 μm and 100 μm, and assembled to create varied microenvironments. The resulting pore sizes and their distribution significantly altered cell morphology and cytoskeletal formation. This microgel-based system provides numerous tunable properties, which can be used to control multiple aspects of cellular growth and development, as well as providing the ability to recapitulate various biological interfaces.

1. Introduction

Hydrogels are a versatile class of polymeric networks that have been utilized for a wide variety of cell culture and tissue repair applications. These highly water-swollen polymer networks have served as a platform to support cell growth in vitro,^[1,2] and to improve viability and engraftment in cell transplantation therapies.^[3,4] Additionally, hydrogels can be modified to mimic many important facets of the native extracellular matrix or selectively functionalized to present specific chemical^[5–7] or mechanical^[8–10] cues to trigger desired cellular responses. Hydrogels can also be spatially confined as microscopic shapes (microgels) and functionalized with specific moieties to affect cell function. Microgels have found numerous uses in cell culture, including as drug delivery vehicles,^[11] structural or bioactive components of bulk hydrogels^[12,13] or cellular aggregates,^[14,15] and as 3D cell culture platforms.^[16–18]

More recently, however, microgels have been utilized as the building blocks for cell culture scaffolds. In this way, hydrogel scaffolds are produced via bottom-up fabrication, by assembling microgels into a fully percolated network. While traditional bulk networks have a porosity (mesh size) on the order of nanometers, microgel-assembled networks contain an inherent porosity on the size scale of tens to hundreds of microns, allowing for rapid cell proliferation, infiltration, and motility.^[19,20] Top-down approaches to creating porous scaffolds, that is, pore formation by degrading part of a bulk network, have been traditionally used, and have been successful at creating biocompatible scaffolds with highly tunable pore characteristics.^[21,22] While top down approaches have proven useful in numerous applications, bottom-up network fabrication can aid in the recreation of complex tissue architectures. Bottom-up approaches have already been used to create heterogeneous tissue mimics,^[23] and the inclusion of multiple “building blocks” could allow for numerous cues to be rapidly incorporated into a porous cell culture scaffold. Elbert and colleagues have successfully demonstrated this concept, by creating microgel assembled scaffolds with three types of microgel components, functioning as structural supports, porogens, or drug delivery systems.^[24] Similarly, microgels have been functionalized with cell adhesive peptides allowing for increased cell invasion to facilitate more rapid wound healing.^[19] Finally, the Zhang group has explored sub-micron sized gels for similar purposes, demonstrating how particle size and interconnectivity within porous scaffolds can be utilized to control cell growth.^[25]

Thus, these scaffolds provide a wealth of opportunities to investigate cell interactions with both chemical and mechanical cues. Fabricating a scaffold with surface features on the size scale of a cell allows for the examination of how micro-structured environments (for example, trabecular bone or alveoli) affect cellular function and growth. The ability to co-assemble cells with microgel building blocks allows for facile creation of these environments in vitro, as well as providing numerous opportunities for cell transplantation and wound healing. Finally, these microgel building blocks can be functionalized with active moieties to impart specific cues to cells contained within the assembled network. Thus, a bottom-up type fabrication, which we present in this study, can be used, not only to create porous scaffolds, but also to permit the assembly of heterogeneous particles and cells that can create varied complex culture environments.

In this contribution, we present a method to fabricate cell-laden microporous hydrogels by the co-assembly of reactive hydrogel particles with primary cells. Macromolecular poly(ethylene glycol) monomers end functionalized with azide and alkyne moieties were synthesized and reacted using an inverse suspension polymerization method to generate microgels with average diameters either on the order of 100 μm or 10 μm. Particles were synthesized with excess alkyne or azide functional groups, and then introduced into a cell suspension. Upon centrifugation, the particle-cell composite formed a macroscopic, but microporous, hydrogel via a strain-promoted azide alkyne cycloaddition (SPAAC) with a cell density of 3×10⁶ cells mL⁻¹. Introduction of a fibronectin-derived adhesive ligand, GRGDS, into the microsphere formulation allowed for cell-particle interactions, cell spreading and process extension. The described microgel networks may have widespread applications in both in vitro cell culture and regenerative medicine.

2. Results

2.1 Microgel synthesis

Microgels containing excess DBCO (dibenzocylcooctyne) or azide groups were formed using an inverse suspension polymerization method (Figure 1). Briefly, aqueous precursor solutions, containing PEG-DBCO (8-arm, 20kDa), PEG-N₃ (4-arm, 10kDa), an azide functionalized fluorophore (AlexaFluor-594-N₃), and an azide functionalized adhesive peptide (N₃-GRGDS) were prepared with 11mM excess of either functional group. Immediately after the addition of the limiting group (PEG-DBCO or PEG-N₃), the aqueous solution was transferred to a continuous phase of hexane, Span-80, and Tween-20. In order to control microgel size, solutions were exposed to either low shear (vortexing) or high shear (sonication) for 5 minutes, which is sufficient time for modulus evolution (Figure S1). Particles were subsequently washed with hexanes (3×), isopropanol (3×), and transitioned to PBS.

Figure 1. Inverse suspension polymerization of microgel building blocks.

Scheme for microgel formation. 8arm 20kDa PEG-DBCO, 4arm 10 kDa PEG-N₃, and azide labeled RGD were combined in PBS and dispersed in hexane. Solutions were exposed to either continuous vortexing or sonication, while polymerizations were performed off-stoichiometry to yield microgels with excess DBCO or N₃ surface functionalities.

Next, we categorized the size distribution of the microgels for each formulation using a custom Matlab script (Figure 2, Figure S2). Microgels formed using low shear (hereafter referred to as “10² μm microgels”) had a broad distribution, ranging between 30 μm to 350 μm, with mean particle diameters of 120 μm and 130 μm for gels containing excess DBCO and azide groups, respectively (Figure 2b). Microgels formed using high shear (hereafter referred to as “10¹ μm microgels”) were an order of magnitude smaller, with average particle sizes of 16 μm and 15 μm for DBCO and azide gels, respectively (Figure 2c). Microgel mechanical properties were approximated by measuring the moduli of bulk gels at identical cross-linking densities. Gels with excess DBCO functional groups were 15 kPa in compressive modulus, while excess azide gels had a compressive modulus of 8.7 kPa (Figure S1c).

Figure 2. Size characterization of vortexed and sonicated microgels.

a. Fluorescently labeled (Alexa Fluor 594) microgels were imaged and sized using a custom Matlab script (scale bar 100 μm). b. Size distribution of 10² μm gels with varying stoichiometry of reactive groups. Microgels exhibited a heterogeneous population for both excess DBCO and N₃ gels, with average diameters of 120 μm and 130 μm, respectively. DBCO-excess and N₃-excess gels had polydispersity indices of 0.29 and 0.38, respectively. c. Size distribution of 10¹ μm gels with varied excess surface functionalities. 10¹ μm microgels had average diameters of 16 μm and 15 μm and PDIs of 0.15 and 0.11 for the excess DBCO and excess N₃ cases, respectively.

2.2 Assembly of microgels into microporous scaffolds and characterization

To assemble the PEG microgels into a microporous, covalently linked material, we prepared microgel suspension containing equal densities of both DBCO and N₃ functionalized microgels. In order to increase the number of particle interactions and covalent cross-linking between microgels, solutions were concentrated via centrifuged for scaffold formation (Figure 3). While solutions containing both DBCO excess and N₃ excess microgels formed networks in both the 10² μm and 10¹ μm particle cases (Figure 3b, c), solutions containing homogenous particles (only DBCO or N₃ surface reactive groups) at identical densities exposed to the same centrifugation did not form full networks (Figure S4). The mechanical properties of the assembled gels were assessed by measuring their compressive moduli. Networks composed of 10² μm or 10¹ μm microgels formed networks with compressive moduli of approximately 2.1 kPa and 3.3 kPa, respectively (calculated via the slope of the linear stress-strain plot) (Figure 3c). The bulk scaffold had a lower modulus (approximately 3–7 times lower) than the compressive moduli of the microgels.

Figure 3. Assembly of microgels into microporous scaffolds.

a.For both the 10² μm and 10¹ μm particles, equal volumes of microgels containing excess DBCO or N₃ groups were combined and centrifuged to form a percolated network. Interactions between particles bearing surface DBCO and N₃ results in triazole formation and a covalently connected microgel network b. Macroscopic scaffold formed from fluorescently labeled 10² μm particles (scale bar = 10mm). c. Values for the average compressive moduli of networks formed from 10² μm and 10¹ μm particles. Compressive modulus was calculated from the slope of the stress-strain curve, yielding networks of 2.1 kPa and 3.3 kPa for the 10² μm and 10¹ μm particles, respectively.

We next sought to categorize the overall porosity, pore connectivity, morphology, and size distribution. We first incubated each hydrogel in a high molecular weight fluorescein-labeled dextran solution for one hour to visualize the interconnectedness of the void space. The 250 kDa dextran readily diffused throughout the open pore structures, but did not penetrate the microgels themselves on the time scale of the experiment (Figure 4). Three-dimensional renderings of the scaffold demonstrate a continuous network of pores in both the 10² μm and 10¹ μm gel conditions. To quantify the overall porosity of the scaffolds, as well as the distribution and size of the pores, the overall void fraction and selected morphological characteristics of the pores were calculated using a custom Matlab script (Figure S3). In brief, Z stack images were collected using a laser scanning confocal microscope of fluorescently labeled microgel assembled networks and analyzed at 12 μm intervals over 400 μm. A custom Matlab script was used to analyze the porous region in each image, identifying individual contiguous pores (Figure 4b), and measure each pore’s area, as well as major and minor axes lengths. The void space was calculated for each slice (area of pores/total area), averaged across the entire scaffold, and reported as the overall void fraction for each microgel type. The 10² μm networks had a void fraction of 29±3%, while the 10¹ μm gels had a lower void fraction of 12±2% (Figure 4c).

Figure 4. Pore analysis of microgel-assembled networks.

a. Three-dimensional images of scaffolds with a fluorescently labeled high molecular weight dextran solution to demonstrate the interconnectivity of pores within the scaffold. b. Matlab analysis of pores within microgel-assembled networks (10² μm particles shown, scale bar 100 μm). c. Void fraction analysis of microgel-assembled networks. Void space was consistent among networks, with 29% and 12% negative space in the 10² μm and 10¹ μm networks, respectively.

While the void space distribution was relatively non-disperse across the gels, the pore areas within the gel were relatively heterogeneous. The 10¹ μm networks contained a higher number of pores than did the 10² μm networks, while the average pore size was much larger in 10² μm networks compared to 10¹ μm networks (Figure 5). In the case of the 10² μm system, 75% of the pores were less than 3300 μm² in area, while the corresponding amount of pores were less than 60 μm² in the 10¹ μm system (Figure 5a). For reference, these sizes correspond to circles with diameters of approximately 65 μm and 9 μm, respectively. However, as the pores in this gel were irregularly shaped and typically acircular, the major axes lengths were measured (Figure 5b), and the aspect ratio (major: minor) of the pores in each gel type is reported (Figure 5c) were assessed to better categorize the dimensions of the void space in the gels. The majority (75%) of pores in the 10² μm and 10¹ μm gels had major axes at or below 120 μm and 14.0 μm, respectively (Figure 5b). Despite the difference in average pore size, the shape of pores (as determined by aspect ratio) was very similar in both networks (Figure 5b). In both cases, the average aspect ratio was approximately 2, indicating elongated, acircular pore shapes (Figure 5c).

Figure 5. Analysis of pore dimensions in assembled networks.

a. Analysis of pore area within 10² μm and 10¹ μm scaffolds. 75% of pores in 10² μm particle gels were less than 3300 μm², while the corresponding percentage of pores in 10¹ μm gels were below 60μm². b. Analysis of the major axis of pores within assembled networks. The majority (75%) of pores within 10² μm networks had a major axis length smaller than 120 μm. Networks containing 10¹ μm particles had much smaller pores, with the same percentage of pores being less than 14 μm in length. c. Aspect ratio of pores within microgel scaffolds. Pores in both networks had heterogeneous distributions of pore shapes, with networks consisting of 10² μm particles having an average aspect ratio of 2.1. Networks containing 10¹ μm microgels had a slightly smaller average aspect ratio of 1.9.

2.3 Co-assembly of cells and microgels

To assess the ability to co-assemble primary cells and microgels during the scaffold fabrication process, bone marrow derived mesenchymal stem cells (hMSCs) were selected as a model system. Specifically, hMSCs were suspended in the microgel formulations prior to centrifugation, and encapsulated at approximately 3×10⁶ cells mL⁻¹. Upon assembly, the cell-laden microporous scaffolds were cultured for 96 hours, and overall cell viability, cytoskeletal structure, and cell morphology were assessed (Figure 6). Cells showed high survival in both cases, with approximately 95% viability observed in both the 10² μm and 10¹ μm networks (Figure 6c). In the 10² μm networks, hMSCs spread robustly, stretching between multiple microgels or spreading across the surface of a single microgel (Figure 6a insert, left). Similarly, the cells in the 10¹ μm networks also adopted a spread morphology, with an increase in thin, fibular-like projections from cells (Figure 6a insert, right). We next set out to categorize cellular morphology by assessing cytoskeletal structure. Cells in the 10 μm² microgel networks spread and exhibited visible actin fibers (Figure 6b, left), while those in the 10 μm¹ microgel network had only small protrusions into the matrix, with diffuse actin staining (Figure 6b, right). Finally, cell shape was quantified by analyzing area:perimeter² ratio in ImageJ, with cells in the 10¹ μm microgel networks significantly more circular than those in the 10² μm microgel networks (Figure 6c).

Figure 6. Cell encapsulation within microporous assembled scaffolds.

a. Representative images of cells encapsulated within 10² μm (left) and 10¹ μm (right) microgel networks. hMSCs are stained with calcein AM (green, live cells) and ethidium homodimer (red, dead cells) prior to imaging. Insert: Higher magnification image demonstrating cell morphology and interaction with microgels (gray). Scale bar denotes 100 μm. b. Cell staining for cytoskeletal morphology. Cells were stained with DAPI (nuclei, blue) and phalloidin (actin, green). Scale bars in inserts denote 100 μm. Insert: Higher magnification image showing actin structure. Scale bar denotes 100 μm. c. Quantification of cell viability and morphology. Viability (top) in both networks was similar, with approximately 95% of cells viable after 4 days in culture. However, cells in the 10¹ μm gel network were significantly more circular (bottom) than those encapsulated in 10² μm networks (** denotes p<0.01).

3. Discussion

In the presented work we introduce a novel, self-assembling microgel network for cell culture applications. Our approach allows for the simple assembly of microgel components into a porous network without the need for porogens or post fabrication processing. We used the described SPAAC chemistry for microgel synthesis, as well as system assembly, without the addition of external initiators or cross-linking agents (Figure 3). Furthermore, stable networks did not form when only microgels containing a single surface functionality were centrifuged (Figure S4), indicating that the networks form via covalent interactions rather than simple physical entanglements. Finally, we demonstrated that the scaffolds could be assembled in the presence of living cells, allowing for facile encapsulation and high cellular viability (Figure 6a). By varying particle size, and thus network porosity and pore size, we could control cell shape, varying from a fibroblast type morphology in 10² μm microgel networks to a more rounded morphology with thin protrusions in the 10¹ μm microgel networks (Figure 6c). This resulted in variations in cytoskeletal organization, with cells in larger pore sizes being able to form visible actin fibers, while those encapsulated in smaller pores showing only diffuse actin (Figure 6b).

In this study, we chose to modify the microgels with the adhesive ligand RGD to allow for rapid cell spreading within the network. These cell-microgel interactions can be advantageous for network formation and engraftment in in vivo studies. Furthermore, the ability to functionalize microgels with bioactive moieties can be extended to numerous applications. In particular, functionalizing microgels with specific differentiation factors or using microgels as drug depots could allow for a new class of cell instructive materials. Microgel scaffolds have already been synthesized with deliverable growth factors,^[24,26] and the fabrication of scaffolds with multiple chemical cues could allow for the creation of complex culture environments.

Beyond the inclusion of chemical cues, we can also tune many material properties of this system. While nanoporous networks can achieve similar, or even higher void fractions,^[27,28] as our microgel assembled system, we have demonstrated how particle size, and thus network pore size, affect cell growth. Both morphology and actin structure were significantly different between the 10² μm and 10¹ μm microgel networks. In the former, we observed fibroblast morphology and defined cytoskeletal actin fibers, while in the latter case cells were more contracted with thin protrusions into the network and diffuse cytosolic actin (Figure 6). This is similar to reported differences observed in cell morphology and actin structure as culture dimensionality and network elasticity is varied.^[29,30] We expect that the smaller pore size in 10¹ μm microgel networks restricts cell spreading to thin processes, limiting cytoskeletal organization. We plan to further investigate the potential effects these differences may have on genetic expression and cell phenotype decisions.

One advantage of this material system is the ability to manipulate the bulk versus local mechanical properties. We observed a local microenvironmental stiffness greater than that of the macroscopic scaffolds, with microgels approximately 3–7 times the moduli of the bulk material. This stands to reason, as we expect the significant percentage of void space to detract from the compressive modulus of the assembled networks. We also expect the dispersity of microgel sizes to play an important role in determining both porosity and scaffold mechanical properties, as it is likely that this dispersity allows for tighter microgel packing than monodisperse samples. More importantly, this tunability can be highly advantageous, as one can recapitulate stiffer, yet porous, cellular environments, which may have broad reaching implications for regenerative medicine for different tissue environments (e.g., trabecular bone). Furthermore, as porous hydrogels allow for more rapid cell infiltration^[31], ECM deposition^[32], and a mitigated immune response^[19], this system could aid in cell transplantation therapies. Thus, our microgel-assembled networks present a highly manipulatable system that could allow for a better understanding of cell function in native tissues, such as the bone marrow stem cell niche, and for in vivo tissue regeneration.

Finally, while our initial formulation relied on microgels only differing in their size and surface functionality (to allow for interparticle cross-linking), this formulation could be readily adapted to include a more heterogeneous particle distribution to recapitulate aspects of native tissues. As material stiffness affects cellular function and fate,^[8,9,21,33] microgels with varied stiffness, or even chemical moieties, might be included within the formulation to probe cell growth fate decisions in unique ways. While homogenous particles can be incorporated ubiquitously within the formulation, they could also be segregated to produce sub-regions within the material. Based on the nature of assembly, gradients or even distinct zones of functionalized microgels could be created via a layer-by-layer deposition method during centrifugation. These scaffolds could then be utilized to control specific cellular responses to recapitulate complex tissue interfaces both in vitro and in vivo.^[34] This approach may be highly beneficial for recapitulating complex tissue interfaces (e.g., an osteochondral interface), where cell growth and fate decisions must be precisely regulated.

4. Conclusion

We have designed a bottom-up assembled microgel network suitable for cellular encapsulation. Clickable microgels of varied size were formed without the need for device fabrication, with varied surface functionalities allowing for spontaneous network formation. This system proved a useful cell culture platform, allowing for facile encapsulation, as well as controllable morphology, of mesenchymal stem cells. These porous scaffolds offer a high degree of tunability over both mechanical and chemical properties, and can be used to recapitulate highly varied or complex tissue environments. Collectively, the bio-click functionalized microgels and their cytocompatible assembly processes offer a unique platform to study and direct cell growth, interactions and function for both in vitro and in vivo applications.

5. Experimental Section

Monomer synthesis

Eight-arm poly(ethylene glycol) (PEG) amine (M_n ~ 20000 Da) was end functionalized with dibenzylcyclooctyne (DBCO) using standard HATU chemistry. Briefly, dibenzocyclooctyne acid (Click Chem Tools, USA), HATU (1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate) coupling agent (1:1 ratio with DBCO, Sigma Aldrich, USA), and di-isopropylethylamine (DIPEA) (2:1 ratio with DBCO, Sigma Aldrich, USA) were dissolved in N,N-dimethylformamide (peptide synthesis grade) and combined with PEG amine (1:15 ratio to DBCO). The reaction was run overnight at room temperature, subsequently concentrated under reduced pressure by rotoevaporation, and precipitated in diethyl ether at 4°C. The resulting product was resuspended in DI H₂O, dialyzed for 72 hours, and lyophilized. The extent of end-group functionalization was confirmed by ¹H NMR (Bruker AV-III) to be approximately 85%; ¹H NMR (400MHz, CDCl₃, δ): 7.70 (d, J=7.5Hz, 1H), 7.45 (m, 8H), 5.18 (d, J= 13.8Hz, 1H), 3.84 (q, J=4.0Hz, 2H), 3.72 (s, 1H), 3.67 (m, PEG), 3.36 (q, J=5.3Hz, 2H).

Four-arm poly(ethylene glycol) (PEG) azide (M_n ~ 10000 Da) was synthesized as previously reported.^[35] Briefly, 4-arm poly(ethylene glycol) was dissolved in dichloromethane (DCM) and pyridine (Sigma Aldrich) and cooled to 0°C. Methanesulfonyl chloride (20 fold excess to PEG) (Sigma Aldrich) dissolved in DCM was then added dropwise, and allowed to react overnight. The product was washed with aqueous sodium bicarbonate, dried with MgSO₄, and precipitated in diethyl ether (Fisher). The mesylate activated PEG was then dissolved in anhydrous DMF along with sodium azide (5 fold excess to PEG) (Sigma Aldrich) and stirred overnight at 80°C under argon. The product was filtered, concentrated under reduced pressure, resuspended in DI H₂O, dialyzed for 72 hours, and lyophilized. The extent of end group functionalization was confirmed by ¹H NMR (Bruker AV-III) to be >98%; ¹H NMR (400MHz, CDCl₃, δ): 3.65 (m, PEG), 3.41 (m, 2H).

The fibronectin derived adhesive peptide GRGDS (RGD) was synthesized on a Protein Technologies Tribute Peptide Synthesizer using standard Fmoc chemistry and a Rink Amide MBHA resin (Chempep Inc, USA). Azide functionalized RGD was synthesized by coupling 4-azidobutanoic acid to the free N-terminus using standard HATU chemistry. The peptide was purified using reverse phase High Pressure Liquid Chromatography (HPLC, Waters Corporation, mobile phase: water and acetonitrile) on an XSELECT CSH C18 column and confirmed via mass spectrometry using a Voyager DE-STR MALDI-TOF (matrix-assisted laser desorption/ionization – time of flight).

Microgel synthesis

Off-stoichiometry monomer solutions were prepared by combining PEG-DBCO, PEG-N₃, and RGD-N₃ in Phosphate Buffered Saline (PBS) (total volume 50 μL). Solutions were made on ice with either excess DBCO or N₃ groups at a concentration of 11mM. In the case of DBCO excess gels, 8-arm 20kDa PEG-DBCO (3 mM), 4-arm 10kDa PEG-N₃ (3 mM) and RGD-N₃ (1 mM) were used, while for N₃ excess gels PEG-DBCO (2 mM), PEG-N₃ (6.5 mM), and RGD-N₃ (1 mM) were used in the formulation. Each macromer solution was rapidly mixed and transferred to a continuous phase consisting of hexane with Span-80 (2.25% v/v) and Tween-20 (0.75% v/v) (20:1 ratio of continuous phase: aqueous phase). Solutions were then exposed to shear stress in the form of vortex mixing or bath sonication for 5 minutes to allow for complete polymerization (based on in situ gelation data (Figure S1a,b)). Microgels were concentrated via ultracentrifugation (18,000 rcf, room temperature, 10 minutes) and washed with hexane (3×), isopropyl alcohol (3×), and PBS (1×), followed by resuspension in 1mL of PBS. After isopropyl alcohol washes, all gels were maintained in sterile conditions.

Rheological measurements

In situ gelation data (Figure S1a,b) was obtained by pipetting 20 μL of monomer solution (for either DBCO-excess or N₃-excess conditions) (cooled on ice) between the bottom plate and an 8mm flat tool on a shear rheometer (TA DH-R3). Time sweeps were performed at 1% strain and 1 rad s⁻¹ for 300 seconds to observe full modulus evolution.

In order to assess the swollen moduli of microgels, 30 μL of monomer solutions (for either DBCO-excess or N₃-excess conditions) were pipetted into a 5mm mold and allowed to swell overnight in PBS after formation. Swollen gels were then tested using an MTS Synergie 100. Tested gels were approximately 2mm in height and 5.3mm in diameter. Gels were exposed to compression up to 15% strain at a rate of 0.5mm min⁻¹, and the compressive modulus was taken to be the slope of the reported stress-strain curve (linear region). Reported values are taken from four separate gels for each condition (DBCO-excess or N₃-excess) (Figure S1c).

Microgel size categorization

For imaging purposes, AlexaFluor594-N₃ (Life Technologies) was included in the DBCO-excess and N₃-excess formulations in the aqueous phase at 40 μM. After washing and transitioning to PBS, the washed microgels were diluted and placed between a glass slide and a cover slip. Gels were imaged on a Zeiss LSM710 scanning confocal microscope with a 10× objective. Images were analyzed using a custom Matlab script (Figure S2) to quantify particle sizes and distributions, analyzing at least 200 microgels (in total) originating from three separate syntheses. PDI values are calculated by (σ*d⁻¹)², where σ is the standard deviation and d is the average particle diameter. Particle size distributions were fit to Gaussian curves using Graph Pad Prism software.

Assembly of microgels into macroscopic, porous scaffolds

Microgel-based scaffolds were assembled by co-centrifuging equal quantities of DBCO-excess and N₃-excess microgels together in 15mL conical tubes. In both conditions, DBCO-excess and N₃-excess gels were mixed in 2 mL of PBS to reach final particle densities of 8×10⁵ and 9×10⁷ particles mL⁻¹ for 10² μm and 10¹ μm particle networks, respectively. These densities correspond to equal microgel volumes (50 μL starting monomer volume). The resulting microgel suspension was mixed and centrifuged (room temperature, 1,000 rcf for 10 minutes, and 3,000 rcf for 3 minutes) to form the covalently-linked microgel-assembled scaffold. Centrifugation speeds were chosen to ensure network formation, without limiting viability in subsequent cell studies. The microporous gel assembly was carefully removed from the conical tube and left to equilibrate overnight in PBS. Gel volumes were determined by displacement, with both 10² and 10¹ μm gel networks having average volumes of 160 μL (pre-swollen) and 180 (final swollen volume) (average measurements of at least 3 gels per condition).

Characterization of the microporous gel assembly

After equilibration in PBS, the porosity of the scaffolds containing the fluorescently labeled particles was assessed using quantitative image analysis techniques. First, images were collected on a Zeiss LSM710 scanning confocal microscope using a 10× objective; z-stacks were taken every 3–4 μm. Next, gels were incubated with high molecular weight fluorescein labeled dextran (250 kDa, Sigma Aldrich) to assess pore interconnectivity. The high molecular weight prevents dextran diffusion into the microgels, while still allowing for transport through the pores. The images were analyzed using a custom Matlab code (Figure S3) to quantify the dimensions and size of each pore, as well as the overall scaffold porosity of the scaffolds. In brief, slices were taken every 12 μm, converted to binary, and thresholded to identify individual contiguous pores. In both cases, 400 μm stacks were imaged from three separate microgel-assembled scaffolds. Over 1500 identified pores were then categorized for each case (10² μm or 10¹ μm scaffolds). The area, major and minor axes lengths for each pore were then identified and averaged across each condition.

The macroscopic mechanical properties of the reacted microgel assemblies were tested using an MTS Synergie 100. Tested microporous hydrogels were conical, approximately 7.55mm in height and 9mm in diameter at the base. The microporous hydrogels were exposed to compression up to 15% strain at a rate of 0.5mm min⁻¹, and the compressive modulus was taken to be the slope of the reported stress-strain curve (linear region). Reported compressive moduli are taken from five gels from each condition (10² μm and 10¹ μm gel networks).

Cell culture

Human mesenchymal stem cells (hMSCs) were isolated from bone marrow aspirates (Lonza Biosciences) as previously reported.^[36] Bone marrow samples were plated on 10 mm tissue culture polystyrene plates (Corning, USA) and cultured in growth media (low-glucose DMEM (1 g L⁻¹) with 10% (v/v) fetal bovine serum, penicillin (50 U mL⁻¹), streptomycin (50 μg mL⁻¹), amphotericin B (500ng mL⁻¹), and basic fibroblast growth factor (bFGF) (1 ng mL⁻¹)) for 72 hours at 5% CO₂ and 37°C. Media was aspirated to remove non-adherent cells; adherent cells were expanded in growth media, trypsinized, and frozen until use in experiments. Prior to encapsulation, cells were similarly plated, expanded (not exceeding 80% confluency), and trypsinized, with all experiments using cells at passage three.

Assembly of microgel-cell composite scaffolds and cell categorization

Cell-laden microporous networks were then fabricated by co-assembling hMSCs with both DBCO-excess and N₃-excess microgels. hMSCs were mixed with DBCO-excess and N₃-excess gels and centrifuged at high speed (room temperature, 1,000 rcf for 10 minutes, and 3,000 rcf for 3 minutes) to form cell-laden gels at 3×10⁶ cells mL⁻¹ . The resulting 180 μL gels were placed in wells containing hMSC experimental media (growth media without bFGF). We then set out to quantify cell viability and morphology to assess the potential of the network as a cell culture scaffold. Cell viability was quantified at 96 hours after encapsulation via calcein (0.5 μM, green, live) and ethidium homodimer (1 μM, red, dead) staining. Cell morphology was characterized by measuring the aspect ratio and circularity of each cell, where circularity is defined as 4π*Area*Perimeter⁻². Four separate gels for each condition (10² μm or 10¹ μm microgel networks) were analyzed with at least 100 cells from each gel (totaling 500 cells per condition). Cell circularity was averaged for each gel, and statistical analysis was then performed using an unpaired t-test with Welch’s correction (to account for differing standard deviations) using Graph Pad Prism software. Finally, all images used are maximum intensity projections of 300–400 μm stacks taken with a 10× objective.

Cytoskeletal morphology of cells within microporous scaffolds was also assessed after 96 hours in culture. Microgel networks were fixed in 10% formalin (Sigma Aldrich) for 30 minutes, permeabilized in 0.1% Triton (×100, Sigma Aldrich) in PBS for 1 hour, blocked with 1% bovine serum albumin solution, and stained with DAPI (300 nM, Life Technologies) and rhodamine-phalloidin (22 nM) overnight. All images are maximum intensity projections of 200–300 μm stacks taken with a 20× objective.