Hydrogel viscoelasticity modulates migration and fusion of mesenchymal stem cell spheroids

Abstract Multicellular spheroids made of stem cells can act as building blocks that fuse to capture complex aspects of native in vivo environments, but the effect of hydrogel viscoelasticity on cell migration from spheroids and their fusion remains largely unknown. Here, we investigated the effect of viscoelasticity on migration and fusion behavior of mesenchymal stem cell (MSC) spheroids using hydrogels with a similar elasticity but different stress relaxation profiles. Fast relaxing (FR) matrices were found to be significantly more permissive to cell migration and consequent fusion of MSC spheroids. Mechanistically, inhibition of ROCK and Rac1 pathways prevented cell migration. Moreover, the combination of biophysical and biochemical cues provided by fast relaxing hydrogels and platelet‐derived growth factor (PDGF) supplementation, respectively, resulted in a synergistic enhancement of migration and fusion. Overall, these findings emphasize the important role of matrix viscoelasticity in tissue engineering and regenerative medicine strategies based on spheroids.

Recent studies have revealed that matrix viscoelasticity regulates multiple cellular processes, including spreading, differentiation, and migration. [8][9][10][11][12]14 In particular, these findings have shed light on the impact of the viscous character of these matrices by highlighting the role of stress relaxation (i.e., time-dependent decrease in stress under a constant deformation of the matrix). For instance, when embedded within hydrogels with a similar elasticity, fibroblasts exhibited enhanced spreading in the hydrogels with a more rapid stress relaxation behavior. 12 Multicellular spheroids and organoids have been shown to be effective tools for in vitro modeling and in vivo regeneration of tissue defects. [17][18][19] In particular, spheroids made of mesenchymal stem cells (MSC) have demonstrated promising results for in vitro biofabrication of tissue-like constructs or in vivo regeneration of tissue defects. 7,18,[20][21][22] These cellular clusters are often embedded in hydrogel matrices to enable inter-spheroid fusion, thereby capturing complex aspects of native in vivo environments such as spatial organization and cellular heterogeneity. 19,[23][24][25][26] For the biofabrication of in vitro tissue models, recent investigations have successfully demonstrated formation of tissue-like constructs by spatial positioning of multiple spheroids within a hydrogel matrix. [24][25][26] In these strategies, the ability of neighboring spheroids to interact and fuse together plays a pivotal role in engineering the formation of complex tissue models. These fusion phenomena largely depend on inter-spheroid cell migration resulting in the formation of cellular bridges between neighboring spheroids. 27 Similarly, based on the fusion of multiple organoids, assembloids have recently emerged as powerful in vitro tools for gaining deeper insight into human development and disease. 28 Furthermore, for in vivo regeneration of tissue defects, the ability of cells to remain as aggregates within spheroids or to migrate into the defect space is an important factor in determining the regenerative outcome. 29 Despite the increasing evidence regarding the influence of matrix viscoelasticity on cellular behavior, 9,17 the effect of hydrogel viscoelasticity on cell migration for spheroid fusion has not been investigated.
Moreover, growth factors such as platelet-derived growth factor (PDGF) provide potent biochemical cues that can activate migration of various cell types including MSCs. 30 Clinically, recombinant human PDGF-BB is one of only two growth factors approved by the US Food and Drug Administration (FDA) for clinical application in craniofacial regenerative medicine and can act by recruiting endogenous MSCs to a defect site. 31,32 However, the interplay between biochemical and biophysical cues on spheroid behavior remains largely unknown.
Here, we investigate the role of hydrogel viscoelasticity on cell migration and fusion of MSC spheroids and elucidate the interplay between the effects of matrix stress relaxation and PDGF on spheroid behavior. We hypothesize that fast stress relaxation behavior would facilitate the fusion of MSC spheroids, and PDGF would further enhance this phenomenon. Alginate-based hydrogels have been previously shown to enable decoupling of the elastic response from the viscous behavior of hydrogels. 11,14 Two alginate hydrogel compositions of similar elasticity, but different stress relaxation behavior were used to explore this hypothesis. Murine MSCs were encapsulated in these hydrogels to study changes in spheroid area and fusion over time. Rho-associated protein kinase (ROCK) and Rac family small GTPase 1 (Rac1) inhibitors were subsequently used to block actomyosin contraction and actin polymerization in cells, to probe their role. Finally, to evaluate the broader clinical applicability of these findings, spheroids composed of human bone marrow-derived MSCs (hBM-MSCs) and human dental pulp stem cells (hDPSCs) were studied. Modification of alginate chains with arginine-glycine-aspartate (RGD) ligands was carried out through covalent coupling of GGGGRGDSP peptides to alginate chains based on carbodiimide chemistry using N-hydroxysulfosuccinimide (Sulfo-NHS; Thermo Scientific Pierce) and N-(3-dimethylaminopropyl)-N 0 -ethylcarbodiimide hydrochloride (EDC; Sigma-Aldrich) as reported previously. 34,35 Using this strategy, $20 RGD motifs were coupled to each HMW alginate chain.
For the RGD modification of LMW alginate chains, the RGD content per mass of alginate was kept the same as that for the HMW alginate ($145 μmol/g). Thereafter, RGD-modified alginate was dialyzed for 3-4 days against decreasing concentrations of aqueous sodium chloride solutions (7.5-0.0 g/L) using regenerated cellulose dialysis tubing (Spectrum Laboratories, Inc.) with a MW cutoff of 3.5 kDa. Thereafter, the alginate solutions were treated with activated charcoal (Sigma-Aldrich), vacuum sterile filtered (0.22 μm pore-size filters), lyophilized, and stored in dry form at À20 C until further usage.
Prior to hydrogel formation, alginates were dissolved in serum-free Dulbecco's Modified Eagles Medium (DMEM; Gibco). To prepare 2% w/v alginate gels, 0.8 ml of an alginate solution composed of 2.5% w/v LMW/HMW alginate was loaded into a 3 ml syringe. At the same time, 0.2 ml of serum-free DMEM containing 91.5 mM (for HMW alginate) or 183.0 mM (for LMW alginate) of calcium sulfate was loaded into another 3 ml syringe. After removing air bubbles from the syringes by manual tapping, the two syringes were connected using a female-female Luer-lock connector. Thereafter, the contents of the two syringes were mixed rapidly to initiate the gelation process, and the mixture was immediately deposited on a surface for gelation.
A two-step casting process was employed based on the modification of a previously reported methodology for the encapsulation of multicellular spheroids within a single geometric plane in order to facilitate their confocal imaging and quantitative analyses. 36 For this process, standard 6-well plates were used, and plastic inserts ( Figure S1A) were fabricated using an Objet30 3D Printer (Stratasys) based on the VeroBlue material (Stratasys). As illustrated in Figure S2, the first alginate layer (1 mm thick) was cast by deposition of the alginate mixture into a well and immediate placement of an insert.
The mixture was then maintained for an hour at room temperature to fully gelate. Thereafter, the insert was removed from the well, and a sterile gauze pad was delicately placed on the top surface of the alginate disk. An aqueous solution containing 100 mM of sodium citrate (Sigma-Aldrich) and 30 mM of ethylenediaminetetraacetic acid disodium salt (EDTA; Sigma-Aldrich) was sterile filtered and was added dropwise onto the gauze surface to wet the whole pad without an excess. After 2 min, the pad was gently removed from the gel surface.
This process results in the availability of uncrosslinked alginate chains at the gel surface to achieve full and uniform bonding with the second layer. 100 μl of 2% w/v alginate (with or without spheroids) were then uniformly distributed onto the gel surface. Prior to casting the second gel layer, a 1 mm thick ring ( Figure S1B) was placed into the insert to allow for a gap height of 2 mm inside the wells (instead of 1 mm gap used for the first layer). Thereafter, the second alginate layer (1 mm thick) was cast by depositing an alginate mixture into the well, immediately followed by the placement of a height-adjusted insert into the well. To ensure uniform ionic cross-linking of the alginate construct, serum-free DMEM containing a calcium concentration similar to that of the gel was added into the well. The plate was incubated for 30 min in a cell culture incubator ( Figure S2). Thereafter, the insert and medium were removed, and alginate disks (8 mm diameter and 2 mm thickness) were obtained by punching the constructs using an 8 mm biopsy punch.

| Rheological test
Viscoelastic properties of hydrogels were evaluated with a Discovery Hybrid Rheometer (TA instruments) using a parallel plate stainlesssteel geometry (diameter = 20 mm). Immediately after mixing the alginate solution with the calcium sulfate dispersion, the mixture was deposited onto the Peltier plate of the rheometer and the gap height was set to 1 mm. Low viscosity silicon oil was used to seal the gap to minimize hydrogel drying during the rheological tests. A time-sweep was then performed at 25 C for 1 h at 1% strain and 1 Hz frequency to allow for complete gelation of the sample. The temperature was then increased to 37 C and another time sweep (1% Strain and 1 Hz) was performed for 15 min to ensure temperature equilibration. The final time point of this step was used to determine the storage moduli of the hydrogels. Afterward, a stress relaxation experiment was carried out at 37 C by applying 15% strain and monitoring the generated stress for 3 h. The stress relaxation data were normalized to the stress recorded at 0.1 s for each measurement. The stress relaxation halftime (τ 1/2 ) was determined as the time at which the initial stress decayed to half of its initial value (at 0.1 s).

| Nanoindentation
A G200 nanoindenter (Keysight Technologies) was used to evaluate the viscoelastic properties of cross sections of hydrogels (n = 4) prepared using the two-step casting process. Hydrogel cross sections were prepared by sectioning each sample using a surgical blade. The cross sections were transferred to glass slides for performing the indentation tests, and droplets of cell culture medium were added around the cross sectioned samples to minimize drying. Dynamic indentations were carried out at three locations of the cross section of each sample: middle region (0.0 mm from midline), middle of first casted layer (À0.5 mm from midline), and middle of second casted layer (+0.5 mm from midline) ( Figure S3). The tests were carried out in air at room temperature using an oscillation frequency and amplitude of 110 Hz and 500 nm, respectively. Each indentation was performed at an indentation depth of 4.5 ± 0.2 μm using a spherical indenter with a diameter of 400 μm. 1% P/S. hDPSCs were cultured using a DPSC BulletKit (PT-3928 and PT4516, Lonza). Both primary human cell types were cultured at subconfluency (maximum 70%) to maintain stemness, passaged (up to P5 maximum), and medium was changed every 2 days.

| Spheroid formation
Spheroids of controlled size and morphology were generated based on a forced aggregation technique using AggreWell™ 400 6-well plates (STEMCELL Technologies). Cells were detached from culture flasks using 0.05% trypsin EDTA and collected by centrifugation at 1200 rpm for 5 min. AggreWell plates were consequently treated with an anti-adherence rinsing solution (STEMCELL Technologies), warm basal medium, and then complete medium. 5.90 Â 10 6 or 2.95 Â 10 6 cells in 5 ml of medium were seeded into each well to achieve a cell density of about 1000 or 500 cells per D1 MSC/hDPSC or hBM-MSC spheroid, respectively. After cell seeding, the plates were centrifuged at 100 g for 3 min to achieve the forced aggregation of cells into the wells. The plates were next incubated overnight at 37 C with 5% CO 2 to allow for spheroid formation. This short incubation period was chosen to ensure sufficient nutrients for cell viability in the AggreWell plates and to avoid the need for refreshing the medium, which could disrupt spheroids localization in microwells. Subsequently, spheroids were harvested by pipetting the medium gently to resuspend the spheroids from the microwells. Harvested spheroids were collected into 50 ml Falcon tubes, centrifuged at 100 g for 3 min, and resuspended in complete DMEM.

| Spheroid encapsulation
For migration and fusion studies, spheroid encapsulation was carried out using the two-step casting process described in Section 2.1. To this end, spheroids were dispersed in 2% w/v alginate solutions, which were applied after casting the first alginate layer ( Figure S2).
Each punched hydrogel disk (8 mm diameter and 2 mm thickness) contained $350 spheroids, which were randomly distributed throughout its middle plane. Due to the consistent dimensions of the microwells employed for spheroid formation, spheroids made of each cell type were of similar size and mass. Therefore, despite the influence of gravity during spheroid culture, the encapsulated spheroids in different hydrogel compositions can largely remain within a geometrical plane ( Figure S4). For cell proliferation studies using DNA assay, spheroid encapsulation was carried out in a one-step casting process as described previously ( Figure S5). 14 Hydrogel disks (8 mm diameter and 1 mm thick) were obtained using a biopsy punch, each of which contained $350 spheroids randomly distributed throughout their matrix.

| Spheroid culture
Following spheroid encapsulation, hydrogel disks were transferred to 24-well plates where they were immersed in 1 ml of corresponding complete growth medium for each cell type and cultured for up to 5 days. To assess the effect of biochemical cues on MSC migration for spheroid fusion, mouse (rmPDGF-BB) or human (rhPDGF-BB) platelet-derived growth factors were included in the cell culture media at a concentration of 10 ng/ml. The growth factor supplemented media were refreshed every 48 h.
To elucidate the cellular mechanism involved in 3D MSC migration for spheroid fusion, small-molecule inhibitors were employed following previous literature. 14 For these experiments, 10 μM of Y-27632 (ATCC) was used to inhibit ROCK and block actomyosin contraction, and 50 μM of NSC-23766 (Selleckchem) was used to inhibit Rac1 and block actin polymerization. These inhibitors were included in the media used for the culture of encapsulated spheroids, which were refreshed every 48 h.

| Spheroid fixation and immunocytochemistry
Hydrogels containing MSC spheroids were rinsed three times with PBS containing 10 mM calcium (cPBS) and fixed in 4% paraformaldehyde (PFA) in cPBS for 30 min, treated with EDTA for 15 min, and permeabilized in 0.5% Triton X-100 in 3% goat serum overnight. Phalloidin-Alexa Fluor™ 488 (AF488, Life Technologies) at a dilution of 1:100 was added to stain the actin cytoskeleton overnight. After rinsing with PBS, Hoechst at a dilution of 1:1000 was added to stain nuclei. Gels were transferred to microscope slides and Prolong Gold antifade reagent (Invitrogen) was added and were maintained at room temperature overnight.

| Confocal microscopy
Confocal fluorescent microscopy was carried out using an upright LSM 710 confocal microscope (Zeiss) and involved imaging of Phalloidin-AF488 and Hoechst channels. For quantification of spheroid area and fusion, a 4x air objective was used to perform tile scans of complete area of each sample. These scans involved Z-stacks images from the middle region of each disk where spheroids were located. Subsequently, images were processed for further analysis via maximum intensity projection of the z-stacks of each tile scan. For visualization purposes, higher magnification images were acquired using a 10x air objective.

| Quantification of spheroid area and fusion
Analyses of spheroid area, spheroid fusion, and inter-spheroid distance were carried out using CellProfiler™ Software. 37 Prior to the analysis, ImageJ software was employed to convert fluorescent images to 8-bit greyscale format. Thereafter, to enhance the accuracy of the automatic quantification using CellProfiler, "Brush Tool" was employed in ImageJ to manually annotate the center of each spheroid, and "Selection Tools" were used to exclude high brightness defects from samples. Next, a CellProfiler Pipeline was utilized to analyze each image involving "identification of primary objects" (i.e., spheroid cores) and "identification of secondary objects" (i.e., total area of each spheroid including migrated cells). For the "identification of secondary objects," a "propagation" method was employed in the CellProfiler settings to define the boundaries between outgrowth area of different spheroids. In this method, the boundary lines are determined based on distance to the spheroid cores (primary objects) and intensity gradients of the outgrowth area. More specifically, the "propaga- The samples were sonicated for 20 s and centrifuged at 12,000 g for 15 s. The supernatant was transferred to a new tube. Total DNA content from each sample was determined according to the assay manufacturer instructions ( Figure S5).

| Statistics
Statistical analyses were performed using GraphPad Prism 9 software.
Statistical comparisons among experimental conditions for spheroids studies were carried out using analysis of variance (ANOVA) tests.
Equality of variance for these tests was evaluated using Bartlett's and  (Figure 1b), respectively. G 0 values of the two hydrogel compositions exhibited no statistical difference, indicating a similar elastic response for the LMW and HMW hydrogels formulated in this study. In contrast, when subjected to a constant strain, the stress generated in these hydrogels decayed with different relaxation profiles (Figure 1c). Quantification of the stress relaxation halftime (τ 1/2 ) revealed an average τ 1/2 of 89 ± 68 s or 467 ± 164 s for the LMW or HMW hydrogels, respectively. Given these stress relaxation profiles, hereafter, we refer to the LMW or HMW hydrogel compositions as FR or slow relaxing (SR) hydrogels, respectively.
Hydrogel materials can undergo swelling and degradation upon exposure to biological fluids, thereby changing their mechanical properties. 39 Nevertheless, the alginate-based hydrogel system employed in this work has been shown in several previous investigations to be mechanically stable, displaying negligible degradation and swelling when incubated in cell culture media for 3 weeks. 35 To enable facile confocal imaging of spheroids and quantification of inter-spheroid distances in 2D images, a two-step hydrogel casting methodology was used to encapsulate spheroids within a single geometrical plane. Oscillatory nanoindentation tests confirmed that hydrogel viscoelasticity was uniform across the cross section of the hydrogel disks ( Figure S3).

| Migration behavior of spheroids encapsulated within viscoelastic hydrogels
MSC spheroids were formed and encapsulated in FR or SR hydrogels, and area per spheroid (μm 2 ) was analyzed as a marker for cell migration from the spheroids across a 5-day time course (Figure 2a). The spheroids cultured in FR gels exhibited significantly higher increase in area as compared to those cultured in SR gels over 5 days (Figure 2b).  In addition, PDGF is a mitogen for MSCs, and therefore cell proliferation may be a contributing factor for the observed enhancement of spheroid area. 47 To determine the potential impact of cell proliferation on these results, we cultured spheroids in the two hydrogel types in the presence or absence of PDGF for 5 days and quantified the total DNA content in each group as a measure of cell number. While there was a trend for greater DNA content in FR gels with and without PDGF, no statistically significant differences were found among these different groups ( Figure S5).
Altogether, these results demonstrated that matrix viscoelasticity plays a critical role in MSC migration from spheroids. FR hydrogels were permissive to cell migration, in contrast to SR hydrogels that were restrictive. The addition of PDGF-BB significantly enhanced MSC migration from and spreading of spheroids in FR matrices. Other growth factors may also stimulate MSC migration including fibroblast growth factor-2 (FGF-2), insulin-like growth factor-2 (IGF-2), and stromal cell-derived factor 1 (SDF-1) but are not the subject of this study. 48,49 In addition, a previous study demonstrated that deletion of the PDGF receptor in MSCs leads to decreased migratory and mitogenic responses. 47 Future studies will be required to assess the effects of various growth factors and their receptor blockade on MSC migratory potential and MSC spheroid fusion in viscoelastic matrices.

| Fusion of spheroids encapsulated within viscoelastic hydrogels
The fusion of spheroids with an initial inter-spheroid gap would require bridging of proliferating and migrating cell populations (Figure 4a), and this bridging phenomenon is anticipated to be impacted by the gap size or distance between the spheroids, which is similar to in vitro wound healing assays. 50 To gain a deeper insight into the fusion behavior of neighboring spheroids, the area of spheroids was next plotted, as well as their fusion status, as a function of their distance from the spheroid in closest proximity (i.e., inter-spher-

| Role of Rac1 and ROCK
Cell motility is mediated by both actomyosin contraction at the rear of a migrating cell, and actin polymerization at the protruding edge, which involve ROCK and Rac1 pathways, respectively. 14,51 Accordingly, two small-molecule inhibitors were next used to block actomyosin contraction (Y-27632) and actin polymerization (NSC-23766), 14 to probe for mechanisms by which viscoelasticity impacts MSC migration ( Figure 6a). Surface area per spheroid (μm 2 ) was again used as a marker for cell migration from the spheroids (Figure 6b). Spheroids cultured in FR gels in medium without supplementation of the inhibitors exhibited a statistically significant increase in area at Day 5, as expected. In contrast, both FR hydrogel groups treated with inhibitors exhibited a decrease in spheroid area at Day 5, similar to spheroid behavior in restrictive SR gels or FR gels without RGD (Figure 2b).
These data suggest that Rac1 GTPase and ROCK are key mediators of MSC migration from spheroids in viscoelastic hydrogels. These findings are in agreement with previous work in the mechanobiology assessing the role of matrix viscoelasticity on breast epithelial cells spheroids, MSC spheroids, and intestinal organoid behavior. 13,14,17,19 However, in contrast to previous findings in which ROCK inhibition did not have a significant effect on the breast epithelial cell (MCF10A) migration, 14   studies. [54][55][56] In addition to hBM-MSCs, hDPSCs were used as they are derived from a more accessible source, the oral cavity. MSC harvest from the bone marrow involves invasive surgical procedures such as bone marrow aspiration from the iliac crest, in contrast to harvesting from an oral source. 57,58 Thus, for future clinical application and reduction of morbidity associated with harvest, we investigate the use of a less invasive cell source.
For hBM-MSCs, the area of spheroids encapsulated in SR gels did not increase with or without the presence of rhPDGF-BB. In fact, spheroid shrinkage was observed at Day 5 for both conditions. In contrast, spheroids encapsulated in FR gels exhibited a significant F I G U R E 5 Quantification of inter-spheroid fusion for spheroids encapsulated in SR or FR hydrogels cultured for 1 or 3 days in PDGF-free (ÀPDGF) or PDGF-supplemented (+PDGF) media. Values were obtained for spheroids positioned within D avg < inter-spheroid distance < 2 Â D avg of their closest neighbor, where D avg is the average spheroid diameter at Day 0. ***, ****, and ns indicate p ≤ 0.001, p ≤ 0.0001, and statistically not significant (p > 0.05), respectively; Brown-Forsythe and Welch ANOVA tests, followed by Games-Howell's multiple comparisons test. Data points represent individual spheroids, based on n = 51-374 spheroids analyzed per group from three to four biologically independent experiments. All spheroids consisted of mouse bone marrow MSCs.
increase in area only when supplemented with PDGF ( Figure 7b). Similarly, spheroids encapsulated in SR gels did not undergo fusion despite PDGF supplementation. In comparison, spheroids encapsulated in FR gels displayed a significant increase in fusion when supplemented with PDGF ( Figure 7c).  Figure 7f).
Finally, when it comes to differences between hBM-MSCs and hDPSCs, a previous study demonstrated that hDPSCs exhibited higher expression of E-cadherin, and lower expression of Snail, an E-cadherin repressor, than hBM-MSC. 60 Interestingly, an increase in human embryonic stem cell migratory capacity was associated with E-cadherin downregulation. 61 This could explain the difference between the migratory behavior between hBM-MSC and hDPSC.
Together, these findings suggest potential differences in response to mechanical and biochemical cues of MSCs from different sources, and further studies will be needed to elucidate these mechanistic differences.
On a separate note, hydrogel viscoelasticity and PDGF supplementation are also capable of impacting a wide range of cell behaviors, including proliferation and differentiation, which might in turn affect migration and fusion behavior of spheroids. 11,62 While investigating these cellular behaviors were beyond the scope of this study, future studies will be needed to obtain a deeper understanding of the potential interplay between these factors. Additionally, while this study was focused on hydrogel compositions with a fixed storage modulus of $3.5 kPa, it will be relevant to investigate in future studies the impact of matrix stress relaxation on spheroid behavior in hydrogels with lower or higher elasticities. More specifically, previous investigations have shown that hydrogels with elastic moduli of $17 kPa and a rapid stress relaxation are more favorable to osteogenic differentiation of MSCs. 11 Therefore, for application of spheroid-based systems in bone tissue regeneration, it will be relevant to study spheroid behavior in matrices with such high elasticities.