Adaptable sliding hydrogels enable pericellular pocket formation while enhancing MSC chondrogenesis and survival in 3D

Abstract

Hydrogels that recapitulate the dynamic mechanical cues of native extracellular matrix are powerful tools that can be leveraged for tissue engineering. Despite growing recognition that cues such as stress relaxation and plasticity modulate cell–matrix interactions, the influence of these properties on mesenchymal stromal cell (MSC) chondrogenesis has yet to be elucidated across a broad range of relaxation timescales and in the absence of confounding biochemical cues. Here, we report the adaptable sliding hydrogel (ASG) with tunable stress relaxation and plasticity as a novel MSC cell niche. By incorporating reversible hydrazone crosslinks into polyethylene glycol (PEG)–based sliding hydrogels (SG), ASG achieves a wide range of tunable stress relaxation and plasticity that are distinct from other dynamic hydrogels used for MSC chondrogenesis. Notably, increasing stress relaxation and plasticity in ASG promotes rapid and robust cartilage formation by human MSCs and supports long-term cell viability. Mechanistically, ASG facilitates local matrix remodeling and enables MSCs to form “pericellular pockets” in 3D that correlate with enhanced nascent extracellular matrix deposition and reorganization, integrin signaling, and nuclear dynamics. Overall, the ASG platform provides a tunable, synthetic microenvironment that helps probe the relationship between dynamic mechanical cues and stem cell fate and informs next-generation material design within the field of tissue engineering.

Graphical abstract

Highlights

• •The adaptable sliding hydrogel (ASG) is a novel viscoelastic 3D hydrogel system.
• •ASG has dynamic crosslinks and displays tunable stress relaxation and plasticity.
• •Increased stress relaxation and plasticity promote stromal cell chondrogenesis.
• •Pericellular pocket formation positively correlates with increased niche remodeling.
• •Cellular mechanotransduction contributes to cell survival and differentiation.

1. Introduction

Hydrogels are versatile biomaterials used across a wide range of applications including in vitro disease modeling [2,3], drug delivery [4,5], and tissue engineering [[6], [7], [8]]. These water-swollen polymeric networks can be designed to mimic key properties of the extracellular matrix (ECM) in which cells reside [9]. One characteristic that has received considerable attention is viscoelasticity, a dynamic property of materials that behave as both viscous liquids and elastic solids and exhibit time-dependent stress relaxation of applied forces via molecular rearrangement [[10], [11], [12], [13]]. In the field of tissue engineering, stress-relaxing hydrogels formed by reversible crosslinking, polymer entanglement, or weak physical interactions have been leveraged to enhance dynamic cell behaviors that require local hydrogel and ECM reorganization. For example, increasing stress relaxation in ionically crosslinked alginate hydrogels permits chondrocyte volume changes and enhances cartilage formation [14]. Similarly, collagen hydrogels with increasing stress relaxation promote mesenchymal stromal cell (MSC) spreading, stress fiber formation, and long-term chondrogenic lineage commitment [15]. Mechanical plasticity, or stress- or strain-induced irreversible hydrogel deformation, is another relevant property of viscoelastic hydrogels that promotes cell spreading and migration by allowing cells to permanently deform the hydrogel interface [[16], [17], [18], [19]]. Together, the time- and strain-dependent properties of hydrogel materials support dynamic cell behaviors that require local niche remodeling.

To sense and respond to dynamic matrix properties such as stress relaxation and plasticity, cells transform extracellular mechanical cues to biochemical signals that regulate cell fate through a process known as mechanotransduction. Key mediators of cellular mechanotransduction include integrin binding, focal adhesion assembly, cytoskeletal tension and dynamics, and nuclear mechanical loading [[20], [21], [22], [23]]. For example, actomyosin contractility upregulates downstream pathways such as RhoA/Rho-associated protein kinase (Rho/ROCK) and Yes-associated protein and transcriptional coactivator with PDZ-binding motif (YAP/TAZ) [21,24]. In turn, these pathways contribute to changes in cell survival, apoptosis, and differentiation outcomes—often in a context-dependent manner. While numerous studies have investigated the influence of dynamic mechanical properties on MSC differentiation, prior research has largely focused on osteogenesis or adipogenesis [21,25,26]. For the few studies that focus on cartilage formation, they either use chondrocytes to assess the response to stress relaxation [14,27] or use materials with a limited range of dynamic behaviors [1,15,28]. Furthermore, modulation of mechanical cues in some of the previous hydrogel systems cannot be decoupled from simultaneous changes in biochemical cue presentation, complicating the interpretation of cellular responses [15,29]. As such, there remains a need to thoroughly characterize the effect of stress relaxation and plasticity on MSC-based cartilage regeneration in 3D hydrogels.

Recent work from our lab has demonstrated that polyethylene glycol (PEG)-based sliding hydrogels (SG) with sliding, irreversible crosslinks enhance MSC chondrogenesis by promoting “cell tumbling,” a rapid, seconds-to-minutes scale cell behavior that increases actomyosin contractility and nuclear mechanical loading as cells move and gyrate within their 3D hydrogel niche [30]. Unlike alginate or collagen hydrogels, SG does not exhibit bulk stress relaxation. Instead, its irreversible crosslinks slide along the PEG backbone in response to applied forces, increasing local matrix permissibility [31,32]. By accommodating cell-mediated matrix reorganization, sliding rings and viscoelasticity both have been shown to support a range of dynamic, time-dependent cellular behaviors. We therefore hypothesize that introducing reversible crosslinks to SG will further accelerate MSC-based cartilage formation and chondrogenic differentiation.

To test this hypothesis, we report the development of adaptable SG (ASG) with reversible, dynamic hydrazone crosslinks as a 3D cell niche for enhancing stem cell chondrogenesis and cartilage formation. We first characterized the viscoelastic and plastic properties of ASG using stress relaxation and creep recovery tests. To determine whether dynamic crosslinks modulate MSC chondrogenesis, MSCs were encapsulated in three ASG formulations with varied stress relaxation and in SG as a non-dynamic control. We demonstrate that ASG accelerated MSC chondrogenesis relative to SG and that the fastest-relaxing ASG group (ASG Fast) supported the most robust chondrogenesis and improved long-term cell viability. Mechanistically, ASG promoted nuclear dynamics and enabled the formation of “pericellular pockets” of space, which correlated with enhanced nascent protein deposition and remodeling, accelerated cartilage formation, and enhanced integrin signaling.

2. Results and discussion

2.1. ASG with dynamic crosslinks exhibits tunable stress relaxation and plasticity

First, the precursors for forming SG and ASG were synthesized, and the dynamic mechanical properties including stress relaxation and plasticity of the resulting hydrogels were characterized. PR-alkyne was synthesized as previously described and was used as a precursor for both PR-norbornene (PR-NB) and PR-hydrazine (PR-Hyd) for SG and ASG, respectively (Fig. 1A) [30]. Following ¹H NMR characterization of PR-alkyne, Copper-Catalyzed Azide-Alkyne Click Chemistry (CuAAC) was utilized to add either norbornene (NB) or hydrazine (Hyd) to PR-alkyne to form PR-NB and PR-Hyd (Fig. 1B). SG crosslinker, 1.5 kDa-PEG-dithiol (1.5k-SH) and ASG crosslinkers, 8-arm-10kDa-PEG macromers functionalized with either aliphatic aldehyde (8-arm-AA) or benzaldehyde (8-arm-BA) were all synthesized as previously reported [30,32], [[33], [34], [35]].

Fig. 1.

Synthesis and formation of SG and ASG. A) Schematic illustration of synthesis of both Polyrotaxane-norbornene (PR-NB) and Polyrotaxane-hydrazine (PR-Hyd) from Polyrotaxane (PR). PR is reacted with glycidyl propargyl ether and propylene oxide to yield PR-alkyne, and subsequently, CuAAC (Copper-Catalyzed Azide-Alkyne Cycloaddition) click chemistry is utilized to synthesize both PR-NB (1) and PR-Hyd (2). B) Schematic of PR-NB (1) and PR-Hyd (2) showing functional groups. C) Sliding hydrogels (SG) are assembled by mixing stoichiometric amounts of PR-NB and 1.5 kDa-polyethylene glycol (PEG) dithiol (1.5k-SH) and UV crosslinked in the presence of the photoinitiator lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP). D) ASG gelation is achieved by mixing stoichiometric amounts of PR-Hyd and 10 kDa 8-arm-PEG-aldehydes. Various ratios of 8-arm-PEG-aliphatic aldehyde (8-arm-AA) to 8-arm-PEG-benzaldehyde (8-arm-BA) can be utilized to vary bond formation and dissociation kinetics.

To form SG, a 1:1 M ratio of PR-NB to 1.5k-SH functional groups were mixed in the presence of 0.1 wt% lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) in PBS and exposed to UV light (365 nm, 4 mWcm⁻²) (Fig. 1C). The formation of ASG with tunable stress relaxation was achieved using hydrazone-based crosslinks: hydrazone bonds have association and dissociation rates that are dependent on the chemical species of aldehyde used for crosslinking, with aliphatic aldehyde (AA) being more dynamic than benzaldehyde (BA) [27]. To balance hydrogel dynamics with long-term culture stability, the following ratios of 8-arm-AA:8-arm-BA were used for ASG Fast, Int, and Slow: 80:20, 40:60, and 0:100, respectively [27]. To form ASG Fast, ASG Intermediate (ASG Int), and ASG Slow, the ratio of PR-Hyd and 8-arm-aldehyde was adjusted to achieve a 1:1 ratio of hydrazine:aldehyde functional groups, the solution was gelled for 20 min at 37 °C, and a biopsy punch was used to create individual cylindrical gels (Fig. 1D).

For each batch of polymer, the weight percent was adjusted between 4 and 6 wt% for ASG and 6.5 and 9 wt% for SG to ensure that the resulting hydrogels had a compressive modulus of ≈5-6 kPa as confirmed by unconfined compression testing (Fig. 2A). Both SG and ASG were formulated without cell-adhesion ligands as biochemical cues can also influence cell behavior [26,36]. To investigate whether nutrient diffusion or protein adsorption influenced biological outcomes, we characterized the physical properties of the SG and ASG hydrogels. Fluorescence recovery after photobleaching (FRAP) showed comparable recovery half-times across all groups, indicating that protein diffusion was similar across all groups (Fig. S1A–B). While all hydrogel formulations also remained intact over time, an increase in the swelling ratio of SG was observed between day 1 and day 14. Despite this, the swollen mass did stabilized after approximately three days and therefore was not likely indicative of gel dissolution (Fig. S1C–D). Non-specific adsorption of BSA was slightly higher in SG but was similar across all ASG groups (Fig. S1E). Overall, SG and ASG hydrogels have comparable stiffness, diffusion, swelling ratios, and non-specific protein adsorption levels.

Fig. 2.

ASG with dynamic crosslinks exhibits increased stress relaxation and plasticity. A) Unconfined compression testing was used to measure the compressive modulus of hydrogels (N = 3). To characterize the stress relaxation behavior of elastic SG and viscoelastic ASG, shear rheology was used. B) Normalized stress was recorded for 10 h and C) of the initial stress, the percentage of stress relaxed within 10 h was quantified (N = 3). D) Schematic showing the design of shear rheology-based creep recovery tests used for evaluation of hydrogel plasticity. Gels were subjected to a 100 Pa stress for 1 h and strain was recorded. After 1 h, stress was set to 0 Pa and strain was measured for 2 h. The degree of permanent plastic deformation was quantified as the normalized strain percent at the end of the 2 h. E) Creep recovery tests for all groups (N = 3). D) Quantification of permanent strain as a percentage of the initial strain at the start of the recovery portion of the test. ns, not significant; ∗P < 0.05, ∗∗P < 0.005, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001. The P value is obtained using one-way ANOVA with Tukey's multiple comparisons test.

Stress relaxation was characterized using shear-based rheology (Fig. 2B). Based on previous reports and the long timescale of stress relaxation for hydrazone-based hydrogels, the stress relaxation was reported as a percentage of initial stress relaxed in 10 h rather than as a relaxation half time (τ_1/2) [13,27]. While SG and ASG Slow exhibited slow stress relaxation (relaxing only 28.5 ± 2.6% and 28.8 ± 4.0% of the initial stress, respectively), ASG Int and ASG Fast displayed significantly faster stress relaxation rates and relaxed 43.0 ± 2.2% and 74.7 ± 2.8% of the initial stress within 10 h, demonstrating that ASG exhibits a tunable range of stress relaxation (Fig. 2C).

In addition to stress relaxation, plasticity is another dynamic mechanical property of hydrogels that can influence cell behavior. Plasticity is defined as irreversible hydrogel deformation under applied stress and can be quantified using shear-based rheology and creep recovery tests. Consistent with previous reports, the parameters of the creep recovery test were chosen to match the timescale of cell behaviors such as cell movement, protein deposition, and hydrogel reorganization [19]. A force of 100 Pa was applied to the gels for 1 h, and strain recovery was recorded over the course of 2 h under 0 Pa of applied stress (Fig. 2D). The creep recovery curves for all groups were plotted, and the permanent strain percentage was quantified (Fig. 2E–F). During a typical creep test, the instantaneous increase in strain after loading is due to the elastic character of the material while the time-dependent response represents the viscous character. After unloading, irreversible deformation is observed only in ASG due to permanent reorganization of the reversible crosslinks, and the strain does not return to zero. ASG Int and ASG Fast displayed significantly greater irreversible deformation (12.5 ± 1.0% and 18.6 ± 1.3%, respectively) compared to SG and ASG slow (−1.7 ± 2.1% and 6.6 ± 0.1%, respectively). These results confirm that introducing dynamic crosslinks to SG resulted in changes to not only stress relaxation but also to plasticity.

It is worth noting that the dynamic properties of ASG are distinct from other dynamic hydrogels such as alginate and collagen gels that have been shown to support MSC chondrogenesis. The τ_1/2 range reported in alginate and collagen hydrogels is ≈1-120 min and ≈0.4-0.9 s, respectively [15,26]. In contrast, even the fastest-relaxing ASG has a τ_1/2 of over 3 h, consistent with hydrazone-based hydrogels used for other applications [27,34]. To directly compare the relaxation and plastic behavior of ASG to those of a commonly used alginate platform, we used a faster stress-relaxing low molecular-weight (LMW) alginate and a slower stress-relaxing high molecular weight (HMW) alginate. Over a 2.5-h duration, HMW alginate exhibited a stress relaxation rate similar to that of ASG Fast, whereas LMW alginate demonstrated the highest relaxation at 64% (Supplementary Table 2). Interestingly, despite differences in relaxation rates, LMW alginate showed irreversible deformation comparable to ASG Fast during creep recovery testing (Fig. S2A–B). This overlap in plasticity between materials with different relaxation profiles underscores the necessity of characterizing both properties independently. Our findings also highlight the unique combination of stress relaxation and plasticity within the ASG platform and motivate its use as a 3D cell niche.

2.2. Increasing stress relaxation and plasticity in ASG accelerates MSC-based cartilage formation in 3D

While the positive correlation between stress relaxation in hydrazone-based hydrogels and chondrocyte-based cartilage formation in 3D has been explored, its effect on MSC chondrogenesis in 3D has yet to be determined [27]. To evaluate MSC chondrogenesis and cartilage formation, we first quantified changes in chondrogenic gene expression. Day 7 reverse transcription-quantitative polymerase chain reaction (RT-qPCR) analysis showed that SRY-Box Transcription Factor 9 (SOX-9), aggrecan (ACAN), and type-2 collagen (COL2) expression were all upregulated by approximately twofold in ASG compared to SG (Fig. 3A–C). Additionally, at day 7, there was no significant difference in expression of type-1 collagen (COL1) (Fig. S3A). Sulfated glycosaminoglycans (sGAG) are important components of healthy cartilage and were visualized using Safranin O (Saf O) staining. ASG accelerated sGAG deposition in a dose-dependent manner by day 7, with ASG Fast having significantly more sGAG per scaffold as observed in Saf O staining and as quantified using a dimethylmethylene blue (DMMB) assay (Fig. 3D–E). This trend persisted until day 21 where all ASG groups had greater positive Saf O staining than SG, and ASG Fast and ASG Int had significantly more sGAG per scaffold (Fig. 3D–F). Total collagen was visualized in the hydrogels at day 7 and day 21 using Masson's Trichrome staining and correlated with the trend in sGAG deposition (Fig. S4A–B). Immunohistochemical staining of type-2 collagen, a desirable mature hyaline cartilage marker, displayed a similar trend by day 28 (Fig. 3G). Consistent with previous reports, type-1 collagen was secreted by MSCs in all groups (Fig. S3B) [32]. Given that all ASG groups showed comparable diffusion, swelling ratios, and protein adsorption yet promoted very different chondrogenesis and cartilage formation outcomes, these results indicate that diffusion and protein adsorption are not responsible for the observed biological findings (Fig. S1, 3D-G). Overall, introducing dynamic crosslinks into sliding hydrogels accelerates MSC chondrogenesis and cartilage ECM deposition, with ASG fast supporting the most rapid and robust MSC-based neocartilage formation.

Fig. 3.

Increasing stress relaxation and plasticity in ASG accelerate MSC-based cartilage formation in 3D. Reverse transcription quantitative-polymerase chain reaction (RT-qPCR) analysis of chondrogenic markers for differentiation including A) transcription factor SRY-box transcription factor 9 (SOX9) and cartilage matrix proteins B) aggrecan (ACAN) and C) type-2 collagen (COL2) relative to SG (N = 3-4) and normalized to GAPDH. D) Safranin O staining of cryosectioned hydrogels at day 7 and day 21. Scale bar 75 μm. Quantification of E) day 7 and F) day 21 sulfated glycosaminoglycan (sGAG) content per hydrogel using a dimethylmethylene blue assay (DMMB) (N = 3). G) Immunohistochemical staining for type-2 collagen at day 28. Scale bar 300 μm ns, not significant; ∗P < 0.05, ∗∗P < 0.005, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001. The P values are obtained using one-way ANOVA with Tukey's multiple comparisons tests.

To assess the long-term phenotype of encapsulated cells, we analyzed expression of hyaline and hypertrophic markers at day 28. Compared to day 0, Type-II collagen (COL2) gene expression was upregulated across all groups, with ASG Fast exhibiting a significantly higher level than SG and ASG Intermediate (Fig. S5A). Aggrecan (ACAN) expression remained comparable across all groups (Fig. S5B). We further assessed markers of hypertrophy, characterized by upregulation of type-X collagen (COLX), type-I collagen (COL1), and matrix metalloproteinases (MMPs) [37]. At day 28, COL1 and COLX expression were comparable across all groups, while MMP3 and MMP13 levels were either unchanged or decreased compared to the day 0 controls (Fig. S5C–F). These findings were further supported by IHC staining, which showed only trace levels of type-X collagen in ASG Slow (Fig. S5G). Similarly, ARS staining showed mineralization was minimal or absent across all groups (Fig. S5H). Together, these results demonstrate that ASG enhances hyaline cartilage formation without inducing an undesirable hypertrophic or mineralized phenotype.

2.3. Increasing stress relaxation and plasticity reduces apoptosis and improves MSC survival over time

As stress relaxation is known to support MSC survival and enhance proliferation [15,26,38], we characterized the effect of stress relaxation and plasticity in ASG on MSC survival, apoptosis, and proliferation. To assess cell viability over time, live/dead staining was quantified at day 1 and day 7. On day 1, high cell viability (over 95%) was observed across all groups, indicating all hydrogels had minimal cytotoxicity (Fig. 4A–B). Though there was a drop in cell viability by day 7 in each group, ASG Fast maintained the highest viability over time (Fig. 4C–D). The incorporation of faster dynamic crosslinks in ASG thus improves cell viability in a dose-dependent manner, with ASG Fast showing about a 100% increase in cell viability compared to SG.

Fig. 4.

Increasing stress relaxation and plasticity reduces apoptosis and improves MSC survival over time. A) Initial cell viability one day after encapsulation is shown using live/dead staining (green = live, red = dead). Scale bar 100 μm. B) Quantification of day 1 live/dead staining (N = 3). C) Live/dead staining was used to show cell viability after 7 days of culture. Scale bar 100 μm. D) Quantification of day 7 live/dead staining (N = 3). Scale bar 100 μm. E) Cleaved caspase 3 (CC3) immunofluorescence staining at day 1 and day 7. Scale bar 50 μm. Quantification of the percentage of cells that stain positive for CC3 at F) day 1 and G) day 7 (N = 3). H) Immunofluorescent staining for the nuclear proliferation marker Ki67 at day 1 and day 7. Scale bar 50 μm. Quantification of the percentage of cells positive for Ki67 at I) day 1 and J) day 7 (N = 3). ns, not significant; ∗P < 0.05, ∗∗P < 0.005, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001. The P values are obtained using one-way ANOVA with Tukey's multiple comparisons tests for B and D and using unpaired two-tailed t-tests for F, G, I, and J.

For more thorough characterization, the groups were narrowed to SG and ASG Fast as they showed the greatest difference in cell viability and cartilage tissue formation by day 7 (Fig. 3, Fig. 4D). To assess apoptosis, immunostaining for cleaved caspase 3 (CC3) was carried out at day 1 and day 7. Consistent with the trend in live/dead staining, both SG and ASG Fast had low levels of apoptosis at day 1, and significantly more CC3 was present in SG than ASG Fast at day 7 (Fig. 4E–G). Proliferation was assessed using immunostaining for the marker Ki67, a nuclear protein associated with cell division. No significant differences were observed between SG and ASG Fast on both day 1 and day 7 (Fig. 4H–J). Overall, these findings suggest that improved cell viability is one key mechanism by which ASG enhances MSC-based cartilage formation.

2.4. ASG increases chromatin condensation and nuclear dynamics but not cortical actin dynamics

Having established the importance of cell survival in the observed tissue formation outcomes, we next characterized mechanotransduction pathways known to influence cell fate in 3D hydrogels. In more permissive hydrogel networks, cells reorganize their pericellular niche [12,21] and increase cytoskeletal tension and nuclear dynamics [[39], [40], [41], [42], [43]]. Using SG, we previously demonstrated that sliding crosslinks promote MSC chondrogenesis by permitting enhanced seconds-to-minutes scale F-actin dynamics, nuclear loading, and chromatin condensation [30]. To determine whether stress relaxation and plasticity in ASG would further enhance these cell behaviors, cells with fluorescently labeled F-actin were imaged using confocal live-cell imaging. Images were taken at 5 min intervals for 25 min with representative images of cortical F-actin shown (Fig. 5A). Cortical F-actin dynamics were quantified using the difference in cortical F-actin positioning between each image frame, where the percent dynamic area is expressed as a percentage of total cortical F-actin area (Fig. 5B). There was not a significant difference in average F-actin dynamics between SG and ASG Fast which suggests that over short timescales, cytoskeletal dynamics may not be increased by stress-relaxation and plasticity in ASG. Previous findings from our group have shown that differences in cell-tumbling and cytoskeletal dynamics in SG can emerge over the course of several hours [30]. We therefore monitored cortical F-actin dynamics for 6 h at 2.5 min intervals to determine whether differences between SG and ASG arise over longer durations. Our results revealed no significant differences in F-actin dynamics between SG and ASG Fast groups (Fig. S6A–B). Furthermore, F-actin organization was observed to occur primarily at the cell boundary (Fig. 5A).

Fig. 5.

ASG increases chromatin condensation and nuclear dynamics but not cortical actin dynamics. (A) Left and middle: F-actin dynamics are shown using representative live cell confocal images taken 15 min apart. Right: The cortical F-actin dynamics are shown using color-coded heatmaps representing the local persistence and spatial organization of cortical F-actin over the course of 25 min with images taken every 5 min. Scale bar 10 μm. B) Quantification of dynamic F-actin area as a percentage of the total cell area (5 min intervals over 25 min, N = 9 cells from 3 independent hydrogels). C) Representative western blots of day 1 protein levels for nuclear mechanotransduction marker Lamin A/C and chromatin condensation marker H3K9Me3. GAPDH is included as a loading control and was run on the same gel as the corresponding protein. D-E) Quantification of western blots for Lamin A/C and H3K9Me3 normalized to GAPDH (N = 3, where each replicate includes two hydrogels pooled). F) Representative immunofluorescent images for Lamin A/C at day 1. Scale bar 5 μm. G) Quantification of nuclear localization of Lamin A/C based on immunofluorescent staining (N = 30 cells across 3 independent gels). H) Representative immunofluorescent images for H3K9Me3 at day 1. Scale bar 5 μm. I) Quantification of nuclear localization of H3K9Me3 based on immunofluorescent images (N = 30 cells across 3 independent gels). J) Superimposed, color-coded outlines of a single nuclei over time (5-min intervals for 25 min) for SG (top) and ASG Fast (bottom). K) Average linear velocity of centroids of nuclei imaged at 5-min intervals for 25 min (N = 18 cells across 3 independent gels). ns, not significant; ∗P < 0.05, ∗∗P < 0.005, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001. The P value is obtained using an unpaired two-tailed t-test for B, D, E, and K.

To further assess nuclear mechanical loading and chromatin accessibility, immunostaining of Lamin A/C and H3K9Me3 was performed as previously reported [30]. Lamin A/C is an important connection point between the nucleus and the cytoskeleton, and under increased loading, Lamin A/C is expressed and localized to the nuclear boundary to regulate nuclear shape and downstream gene expression [[44], [45], [46]]. Western blotting and immunostaining revealed comparable Lamin A/C expression in ASG Fast and SG (Fig. 5C–D, S7A-B). Additionally, Lamin A/C is localized predominantly at the nuclear boundary in both groups (Fig. 5F–G). While ASG Fast does not significantly impact F-actin dynamics or Lamin A/C expression, histone methylation and chromatin condensation were previously identified as critical mediators of MSC chondrogenesis in SG [30]. Indeed, significantly higher levels of H3K9Me3, a marker for increased chromatin condensation, were observed in ASG Fast compared to SG at both day 1 and day 3 (Fig. 5C–E,H-I, S7A,C) [47].

Based on the findings, we speculated that an increase in nuclear movement within the cells may correlate with observed differences in chromatin condensation. To track nuclear dynamics, live-cell imaging was performed over the course of 25 min. Unlike the trend observed in F-actin dynamics, ASG Fast significantly enhanced nuclear dynamics compared to SG (Fig. S6C). By tracking the nuclear location and boundary, we mapped out the position of the centroid of each nucleus at 5 min intervals and found the average linear velocity was increased in ASG Fast (Fig. 5J–K). Taken together, we observed a positive correlation between nuclear dynamics, chromatin condensation, and enhanced MSC chondrogenesis and cartilage formation in ASG Fast.

2.5. ASG enables MSCs to create pericellular pockets in 3D and increases local niche remodeling

One hallmark of viscoelastic materials is the time-dependency of their characteristic behaviors: stress relaxation and hydrogel reorganization occur more slowly in ASG than in other reported hydrogels used for MSC chondrogenesis. Seeing as F-actin dynamics were similar over short, seconds-to-minutes timescales in ASG Fast and SG, we moved to characterize processes that occur over longer periods of time including cell volume and morphological changes, pericellular hydrogel reorganization, and nascent protein deposition [14,21]. Day 7 live/dead images show that cells in ASG Fast have more extended, bleb-like morphologies compared to cells in SG or ASG Slow (Fig. S8A–B). Consistent with this observation, brightfield images show that ASG permitted MSCs to create a unique “pericellular pocket” of space day 7. As SG and ASG hydrogels are transparent, the cell-hydrogel boundary could be clearly defined, and the area of the pocket was quantified (Fig. 6A, S9A). Compared to SG, ASG increased the pericellular pocket size as early as day 3, and the difference became even more significant by day 7 (Fig. 6B). To confirm that the hydrogel boundary was consistent with the observed pericellular pocket boundary in brightfield images, fluorescent beads were encapsulated in the gels (Fig. S9B). Quantification of the area of pericellular space devoid of beads confirmed the same trend: ASG Fast has increased pericellular pocket area between day 1 and day 3, and the hydrogel boundary was pushed back by cells over time (Fig. S9C). This reorganization of the hydrogel interface occurs over a timescale of days, highlighting that cell-mediated changes to the ASG local niche accumulate slowly over time and occur on timescales more consistent with the relatively slow stress relaxation and plasticity.

Fig. 6.

ASG enables MSCs to create pericellular pockets in 3D and increases local niche remodeling. A) Representative brightfield images of cells in SG and ASG Fast at days 1, 3 and 7 showing the hydrogel boundary and B) quantification of the “pericellular pocket” as defined by the cell-hydrogel boundary (N = 15 cells across 3 independent hydrogels). Scale bar 10 μm. C) Representative fluorescent images of F-actin organization within cells imaged in panel A in SG and ASG Fast at days 1, 3 and 7 showing the cell body boundary. Scale bar 10 μm. D) Quantification of cell body area (N = 15 cells across 3 independent hydrogels). E) Schematic showing representative cell pocket composition. Light red represents the cell body area and gray represents the area of space beyond the cell body within the pocket at days 1, 3 and 7. Scale bar 10 μm. F) Quantification of cell pocket distribution between cell body area (solid) and space (striped) as percentages of the total pocket area. Gene expression of nascent ECM proteins G) type-1 collagen (COL1) and H) fibronectin (FN) at day 1 normalized to SG (N = 4). I) Representative images of immunofluorescent staining for fibronectin at day 3. Images are shown as a single z-plane taken from z-projections used for J) quantification of relative fluorescent intensity (N = 15 cells across 3 independent hydrogels). Scale bar 5 μm. K) Relative gene expression of matrix metalloproteinase 13 (MMP13) at days 1, 3 and 7 normalized to a day 0 control in 2D (N = 4). L) Cropped images of human protease array membrane for MMP13 from conditioned media collected between day 0 and day 2 and M) quantification of the relative pixel intensity of the average of the two technical replicate spots (N = 1, technical duplicate). N) Relative gene expression of matrix metalloproteinase 3 (MMP3) is shown at days 1, 3, and 7 and normalized to a day 0 control in 2D (N = 4). O) Cropped images and P) quantification of the protease array membrane showing MMP3 spots (N = 1, technical duplicate). For panels B, D, K and N, the P value is obtained using a two-way ANOVA with Tukey's multiple comparisons test; a, P < 0.05 between SG and ASG Fast at a given time; b, P < 0.05 between SG at a given time and SG day 1; c, P < 0.05 between ASG Fast at a given time and ASG Fast day 1. For J, the P value is obtained using an unpaired two-tailed t-test; ns, not significant; ∗P < 0.05, ∗∗P < 0.005, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

The mechanosensitive ion channel transient receptor potential vanilloid 4 (TRPV4) has previously been identified as a key regulator of cell volume, hMSC chondrogenesis, and chondrocyte matrix formation [[48], [49], [50]]. To determine whether cell volume changes contribute to pericellular pocket formation in ASG Fast, we next assessed TRPV4 expression. Interestingly, no significant difference in TRPV4 expression was observed between SG and ASG Fast at day 1 (Fig. S9D) Additionally, it was noted that the cells in SG remained more spherical, taking up nearly the entire space of the pocket, while cells in ASG Fast occupied only a portion of the pocket in 3D by day 7 and extended sizable protrusions out into the surrounding space (Fig. 6C–S10). As a proxy for cell volume, the cell body area was defined in a single z-plane and quantified [14]. Overall cell area changes are minimal in both SG and ASG Fast (Fig. 6D). Though TRPV4 activity can also influence other mechanosensing pathways and affect cytoskeletal organization [50,51], the change in cell area was modest compared to the significant increase in pocket space. Therefore, we proceeded to characterize this pericellular pocket space.

Interestingly, the composition of the pockets changed over 7 days in ASG Fast due to a sharp increase in pericellular pocket space beyond the cell area boundary (Fig. 6E–F). Additionally, these pockets in ASG Fast persisted over time and appeared to be correlated with type-2 collagen immunostaining at day 14 and day 28 (Fig. 3G–S9E). Cellular protrusions and an extended actin morphology also persisted over time and remained visible at day 21 (Fig. S9F). We therefore postulated that the growth of the pericellular pocket space may correlate with nascent protein deposition as cells cocooned themselves with matrix.

Indeed, ASG Fast induced upregulation of gene expression of type-1 collagen (COL1) and fibronectin (FN) as early as day 1 (Fig. 6G–H). Immunofluorescent staining of F-actin and fibronectin agrees with this trend and shows that beyond the cell F-actin boundary, an increased layer of fibronectin is visible (Fig. 6I). Quantification of fibronectin and type-1 collagen immunofluorescent images also confirmed that nascent protein accumulation is increased in ASG Fast compared to SG (Fig. 6J–S11A-B). Similarly, sGAG deposition appears to occur initially within the pericellular region as observed in day 7 Saf O histology (Fig. 3D). Overall, the timing of ECM deposition is consistent with the formation of pericellular pockets, providing additional evidence that nascent protein accumulation at least partially contributes to this niche reorganization. Interestingly, by day 7, type-1 collagen is restricted to a tight pericellular layer in SG but is distributed throughout the pocket space in ASG Fast, suggesting that cells in ASG Fast may also more actively reorganize secreted ECM within the pocket space (Fig. S11A and C). While increased distribution of cell-secreted ECM has been previously described in alginate hydrogels used for chondrocyte-based cartilage formation [14], the occurrence of a pericellular pocket between cells and the hydrogel interface has not yet been reported and therefore represents a novel mechanism supported by ASG that correlates with enhanced MSC chondrogenesis.

Following nascent protein deposition in 3D, cells dynamically remodel their niche through various mechanisms including enzymatic degradation. To determine if ASG Fast increases ECM remodeling within the pericellular space, we characterized MMP13 and MMP3, two matrix metalloproteinases (MMPs) that chondrocytes utilize to degrade pericellular proteins such as collagen and fibronectin [52]. Our results show that ASG Fast induced only a transient upregulation of MMP13 and MMP3 gene expression, which peaked at day 1 and day 3, respectively. Both levels dropped by day 7 and returned to day 0 baseline by day 28 (Fig. 6K–N, S5E-F). Furthermore, protein levels measured over the first two days of culture using a human protease array supported the trends from RT-qPCR data (Fig. 6L–M,O-P, S12A-B). This is also consistent with previous reports that have demonstrated that enzymatic niche reorganization correlates with enhanced nascent ECM deposition and local hydrogel dynamics in 3D [21].

In native cartilage, chondrocytes reside in lacunae, specialized pockets of space that contain a dense network of pericellular matrix (PCM) proteins including aggrecan, type-IV collagen, and perlecan that can sequester growth factors and facilitate mechanosensing [53]. While the pericellular pockets observed in ASG structurally resemble these native lacunae and permit local ECM accumulation, their formation is not an inherent feature of the ASG platform. In contrast with other strategies that engineer hydrogel propert to mimic the composition [54,55] or spatial patterning of lacunae [[56], [57], [58]], ASG supports cell-mediated niche reorganization and irreversible remodeling. To further describe this dynamic environment, we next characterized additional mechanosensing pathways that may be responsible for driving observed pocket formation and enhanced ECM deposition.

2.6. Integrin β-1 expression and cytoskeletal tension correlate with MSC survival and chondrogenesis in ASG

As MSCs are adhesion-dependent and SG and ASG Fast are PEG-based polymers without cell-adhesive ligands, the cell-secreted ECM becomes an important component of the hydrogel-cell interface [21]. Through integrin binding and focal adhesion formation, cells create connections between their cytoskeleton and the local ECM. Integrin β-1 recognizes type-1 collagen and fibronectin and activates signaling pathways that contribute to cell survival and differentiation including Rho/ROCK and mitogen-activated protein kinase 1 and 3 (MAPK1 and MAPK3) [59,60]. To confirm that cells in ASG Fast interact more strongly with nascent ECM than those in SG, we quantified the gene expression of integrin β-1 (ITGB1), downstream targets such as ROCK and MAPK, and focal adhesion proteins vinculin (VCL) and talin (TLN). All genes were upregulated in ASG Fast by day 1 relative to SG (Fig. 7A, S13A-F). Protein levels of integrin β-1 were also quantified at day 3 by western blotting and were significantly higher in ASG Fast (Fig. 7B–C). Immunofluorescent staining further validated that ASG Fast increased integrin β-1 protein levels compared to SG (Fig. 7D–E). Interestingly, integrin binding and cytoskeletal force generation have also been linked to chromatin condensation [61,62] and may contribute to increased nuclear dynamics and H3K9Me3 levels in ASG Fast. Another mechanotransduction pathway that can be activated by cytoskeletal reorganization, integrin signaling, and mechanical force generation [[63], [64], [65]] is the Hippo-Yes-associated protein (YAP)/transcriptional coactivator with PDZ-binding motif (TAZ) pathway. Upon activation, YAP and TAZ are translocated to the nucleus where they regulate downstream gene expression. We found that nuclear localization of YAP increased from day 1 to day 3 in ASG Fast and was greater at day 3 compared to SG (Fig. S14A–B). Though the effect of YAP/TAZ in 3D hydrogels on MSC differentiation has been shown to be context-dependent [26], nuclear localization in ASG Fast correlated with increased chondrogenesis. Taken together, these results suggest that several mechanotransduction pathways including integrin binding, Rho/ROCK, MAPK, and Hippo-YAP/TAZ are activated by stress relaxation and plasticity in ASG Fast.

Fig. 7.

Integrin β-1 expression and cytoskeletal tension correlate with MSC survival and chondrogenesis in ASG. A) Relative gene expression of integrin β-1 (ITGB1) on day 1 (N = 3). B) Representative western blots for integrin β-1 at day 3 and C) quantification (N = 3). GAPDH is used as a loading control and was run on the same blot. D) Representative immunofluorescent images of z-projected integrin β-1 staining on day 3. Scale bar 5 μm. E) Quantification of relative fluorescent intensity of integrin β-1 staining (N = 15 cells across 3 hydrogels). F) Schematic of experimental timeline for drug studies. CM = chondrogenic media; bleb = blebbistatin. Drugs were administered for the first three days of culture in CM and samples were harvested at day 7 and day 21. G) Cell viability is evaluated using live/dead staining at day 7 and H) quantified as a percent viability (N = 3). Scale bar 100 μm. I) Safranin O staining of final tissue formation outcomes at day 21. Scale bar 100 μm ns, not significant; ∗P < 0.05, ∗∗P < 0.005, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001. The P value is obtained using an unpaired two-tailed t-test for A, C, and E and using a two-way ANOVA with Tukey's multiple comparisons test for H.

As integrin binding can promote actomyosin contractility and activate downstream biochemical pathways that contribute to MSC survival in 3D hydrogels [66], we speculated that perturbation of these pathways would modulate cartilage formation in ASG Fast. In our previous work, cytoskeletal force generation via actomyosin contractility contributed to tissue formation outcomes in SG [30]. Blebbistatin, a myosin II ATPase inhibitor, was therefore chosen to inhibit actomyosin contractility. Downstream of integrin β-1 and several other mechanotransduction pathways, Rho/ROCK signaling has also been extensively studied in the context of MSC differentiation and can be inhibited using the small molecule Y-27632. Interestingly, the effect of ROCK inhibition on MSC chondrogenesis is context-dependent: ROCK inhibition mitigated MSC apoptosis and improved cartilage formation in slower stress-relaxing collagen hydrogels [15] but reduced MSC chondrogenesis in gelatin microribbon (μRB) hydrogels [67]. To evaluate the effect of blebbistatin and Y-27632, the pharmacological treatments or a vehicle control were supplemented in chondrogenic media (CM) for the first three days of culture (Fig. 7F). Both blebbistatin and Y-27632 reduced cell viability in ASG Fast from ≈60% to below 45% but had no effect on viability in SG by day 7 (Fig. 7G–H). Strikingly, tissue formation was also minimal or absent in these groups, with only the ASG Fast vehicle control producing robust sGAG by day 21 (Fig. 7I). Though previous work has shown that actomyosin contractility was critical for enhanced chondrogenesis in SG, we observed low overall cartilage deposition even in the control groups for SG (Fig. 7I). These differences may reflect factors such as donor-to-donor variability in primary MSC chondrogenic potential that are known to influence biomaterial-based tissue formation outcomes, thus highlighting the advantage of introducing stress relaxation into the SG platform [1].

To assess if cell adhesive peptides could enhance MSC survival and tissue formation in SG, hydrogels were functionalized with RGD, a commonly-used adhesive ligand that has been shown to promote cell viability in covalently-crosslinked hydrogels [36]. However, we did not observe significant increases in cell viability or tissue deposition in SG hydrogels with added RGD binding domains (Fig. S15A–C). These results suggest that the enhanced MSC survival and chondrogenesis observed in ASG are driven primarily by stress relaxation and plasticity, which correlate with elevated actomyosin contractility, ROCK signaling, and local matrix reorganization.

3. Conclusion

In summary, we report viscoelastic ASG as a dynamic stem cell niche that supports MSC chondrogenesis and improves cell survival. Increasing stress relaxation and plasticity in ASG enabled cell-mediate niche reorganization highlighted by “pericellular pocket” formation. To the best of our knowledge, this distinct mode of hydrogel remodeling characterized by permanent deformation of the cell-hydrogel interface has not yet been identified. Another unique aspect of this work is its mechanistic investigation across timescales. On the seconds-to-minutes timescale, increased nuclear dynamics were observed in ASG compared to SG and correlated with increased chromatin condensation and chondrogenesis. On an hours-to-days timescale, ASG enhanced cellular mechanotransduction pathways including nascent protein deposition, enzymatic remodeling, integrin binding, and downstream ROCK signaling. Inhibition of actomyosin contractility or ROCK signaling did not affect cell viability and cartilage formation in SG, but significantly reduced both in ASG, underscoring the importance of these pathways in ASG-enhanced MSC survival and chondrogenesis. Over the first 7 days of culture, ASG facilitated pericellular pocket formation that correlated with ECM deposition and early ECM remodeling. While prior research has largely focused on the role of stress relaxation in MSC chondrogenesis, our findings identify plasticity as a critical property for supporting dynamic, cell-mediated niche reorganization. Future studies should build upon this by evaluating the in vivo performance of ASG in disease-relevant models. Such investigations will be essential to assess the long-term biocompatibility, host tissue integration, and functional regenerative capacity of these hydrogels in a complex physiological environment over long timescales.

4. Materials and methods

4.1. Synthesis and characterization of SG and ASG macromers

The synthesis of Polyrotaxane (PR) and Polyrotaxane-alkyne (PR-alkyne) were adapted from previously reported methods [[30], [31], [32]]. Briefly, 35 kDa linear polyethylene glycol (PEG) was first oxidized to have di-carboxyl groups (PEG-COOH) and then dissolved in an aqueous solution at 15 wt% with 15 wt% alpha-cyclodextrin (α-CD). An inclusion complex was formed after overnight stirring. The product, pseudo-polyrotaxane (P-PR) was freeze-dried and subsequently end-capped using 1-adamantylamine in N,N-dimethylmethanamide (DMF) coupled with benzotriazol-1-yloxytripyrrolidinophosphonium hexafluorophosphate (PyBOP). Polyrotaxane (PR) was obtained by crystallization in acetonitrile, hot water washing and lyophilization.

To synthesize PR-alkyne, PR was first dissolved in 1N NaOH. Glycidyl propargyl ether was added at 1 mass equivalent, and to enhance solubility, propargyl groups were added using propylene oxide at 16.6 mass equivalent. The resulting PR-alkyne was neutralized, dialyzed against DI water, and lyophilized. Alkyne and propargyl functionality were confirmed via ¹H NMR. Polyrotaxane-norbornene (PR-NB) was synthesized from PR-alkyne precursors as previously reported [32]. Final α-CD coverage and norbornene functionality were confirmed using ¹H NMR with coverage at ≈10% and functionality at ≈0.3-0.7 norbornene per α-CD ring.

The synthesis of Polyrotaxane-hydrazine (PR-Hyd) was carried out using PR-alkyne and Tri-tert-butyl 2-(2-((3-azidopropyl)amino)-2-oxoethyl)hydrazine-1,1,2-tricarboxylate (azido-Boc₃Hyd). Azido-Boc₃Hyd synthesis was achieved using methods described in previous reports [68]. Coupling between PR-alkyne and azido-Boc₃Hyd was carried out using copper-catalyzed azide-alkyne cycloaddition (CuAAC) click chemistry. PR-alkyne (1g) was dissolved in 18 mL of a 1:1 solution of methanol (MeOH) and water. The reaction was catalyzed using CuSO₄/ascorbic acid. Azido-Boc₃Hyd (1.2 mol ratio to alkyne) was dissolved in 1 mL of MeOH and added to the solution and left overnight. The resulting polymer was dialyzed against water and lyophilized, and hydrazine deprotection was done in a 3.6 N HCl solution at 40 °C for 45 min. Dialysis and lyophilization yielded PR-Hyd with a functionality of ≈0.4-0.6 hydrazine per α-CD ring as confirmed via ¹H NMR.

Crosslinkers for ASG, 8-arm-PEG-benzaldehyde (8-arm-BA) and 8-arm-alkyl aldehyde (8-arm-AA) were synthesized as previously reported and characterized with ¹H NMR [34,35].

4.2. SG and ASG hydrogel formation

To form SG, 20 wt% solutions of PR-NB or PEG di-thiol (1.5kPEG-SH); Mw, 1.5 kDa in PBS with 0.1 wt% lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) were prepared. The precursors were mixed at a 1:1 M ratio (PR-NB:1.5kPEG-SH) and placed under ultraviolet light (365 nm, 4 mWcm⁻²) for 5 min on Teflon-coated slides. Individual hydrogels were made with biopsy punches. Final polymer concentration in the gels was adjusted between 6 and 10 wt% to create hydrogels with a constant stiffness of ≈6 kPa and account for slight differences in functionality. To form ASG, a 10 wt% precursor solution of PR-Hyd was made in PBS. Solutions of 8-arm-BA and 8-arm-AA were made at 20 wt%. Three ratios of 8-arm-AA to 8-armBA were used to create the fast (80:20), intermediate (40:60) and slow (0:100) aldehyde solutions. Precursors were neutralized and kept on ice. A final polymer concentration of ≈4-6 wt% was achieved by combining PR-Hyd and aldehyde solutions at a 1:1 M ratio. Gelation at 37 °C for 20 min yielded ASG.

4.3. Alginate hydrogel preparation

Alginate hydrogels were prepared as previously described [69]. Briefly, high molecular weight (HMW) alginate and low molecular weight (LMW) alginate were prepared as 3 wt% stock solutions from Pronova Ultrapure MVG (>200 kDa) and Pronova Ultrapure VLVG (<75 kDa) in serum-free DMEM. A calcium stock solution of 488 mM calcium sulfate dihydrate in DI water was prepared. Hydrogels were formed with a final concentration of 2 wt% alginate and a calcium crosslinker concentration of 11 mM calcium for HMW and 23 mM calcium for LMW hydrogels.

4.4. Unconfined compression testing

To determine the compressive modulus, cylindrical gels with a diameter of 4.5 mm and a thickness of 2 mm were made and equilibrated at room temperature in PBS for an hour. An Instron 5944 testing system equipped with a 10 N load cell (Interface Inc.) was used to generate a stress versus strain curve where the linear curve fit was taken as the modulus for strain ranges from 10 to 20%.

4.5. Stress relaxation and plasticity characterization

A Discovery HR-2 hybrid rheometer (TA Instruments) was used to quantify the stress relaxation timescale of hydrogels with a 25 mm stainless steel parallel plate geometry. For ASG and alginate, gels were formed in situ at 37 °C, and a time sweep at a 1 rads⁻¹ was conducted to track gelation. Once a stable modulus had been reached, stress relaxation testing using 10% strain was started, and stress was recorded for 10 h or 2.5 h for alginate experiments. For SG, polymer solution was dispensed onto a glass plate affixed to the rheometer and UV light was used underneath the glass for 5 min to form a gel directly between the upper geometry and the glass plate. This set-up did not allow for temperature control, but the same stress relaxation test was carried out. Plasticity was measured using a creep recovery test adapted from previous reports [19]. 100 Pa of stress was applied to the gels for 1 h then removed. Strain was recorded for 2 h, normalized, and plotted.

4.6. Hydrogel diffusion and non-specific protein adsorption

Fluorescence recovery after photobleaching (FRAP) was used to characterize diffusion in acellular hydrogels and adapted from previously reported methods [31]. Briefly, gels were formed and left in a solution of 1 mg mL⁻¹ of fluorescein isothiocyanate (FITC) labeled bovine serum albumin (BSA, Thermo Fisher) in PBS overnight at room temperature. A Leica STELLARIS 5 confocal microscope with HCX PL APO CS × 63.0/1.40 OIL objective was used to bleach a region of interest with a white light laser (WLL) for 20 s (until the region was fully bleached). Fluorescence intensity was recorded every 1.5 s for a period of 5 min and quantified within a circular area around the bleached region with a 25 μm diameter. The fluorescence intensity was normalized to the 100% recovered region, and intensity recovery was plotted over time.

To assess non-specific adsorption, hydrogels were formed and soaked in a PBS solution containing 1 mg mL⁻¹ FITC-BSA overnight. Gels were then washed thrice in PBS for 30 min and the fluorescence was quantified using the STELLARIS 5 confocal microscope. The relative fluorescence intensity was normalized to SG and plotted using FIJI.

4.7. Hydrogel swelling and degradation

Hydrogel constructs were incubated in PBS for 14 days. To track relative swollen mass over time, wet weight was measured at pre-determined time intervals. Increases in swollen mass can indicate gel degradation over time. Wet weight was measured first, and dry weight was obtained using lyophilized gels at day 14. Relative swollen mass was calculated as the ratio of the wet weight to dry weight at day 14 and as the ratio of the wet weight to the calculated polymer mass encapsulated within each scaffold at day 1. The swelling ratio was calculated as previously described using the swollen mass (M_s) and dry mass (M_d): swelling ratio = (M_s-M_d)/M_d x 100 [27].

4.8. Cell culture

Human mesenchymal stromal cells (MSCs) were purchased from Lonza and expanded to passage 6 (P6) for all experiments (Lot 22TL301024, 23-year-old male donor). hMSC growth media consisted of low glucose (1 g L⁻¹) Dulbecco's minimum essential medium (DMEM) (Lonza), 10% fetal bovine serum (FBS) (Gibco), 1X gentamicin/amphotericin B (GA) (MilliporeSigma, Gibco), and 10 ng mL⁻¹ basic fibroblast growth factor (bFGF) (PeproTech). Cells were encapsulated in hydrogels at 1.0x10⁷ mL⁻¹. Chondrogenic differentiation was induced starting immediately after encapsulation using chondrogenic media (CM) composed of high glucose DMEM (4.5 g L⁻¹) (Lonza), 50 μg mL⁻¹ ascorbic-2-phosphate (MilliporeSigma), 100 nM dexamethasone (MilliporeSigma), 1X sodium pyruvate (Fisher Scientific), and 1X GA. Regular mycoplasma testing was conducted using a MycoAlert Mycoplasma Detection Kit (Lonza). Most recent testing results were negative (ratio of A/B was 0.222 where below 0.9 is a negative result).

4.9. Cell viability and quantification

Cell viability was assessed using live/dead staining. Cells were incubated for 30 min at 37 °C in 2 μM calcein AM (Corning) and 4 μM ethidium homodimer (Fisher Scientific) diluted in PBS. A Keyence BZ-X810 fluorescent microscope with 20X objective was used to take z-stacks of scaffolds. After projection in Fiji software, the number of live and dead cells were counted for each sample.

4.10. Fixation and cryosectioning

Hydrogel scaffolds were fixated in 4% paraformaldehyde (PFA) (Boston BioProducts) for 30 min at room temperature and washed twice in PBS. TissueTek optimal cutting temperature (OCT) compound was used to dehydrate samples before cryosectioning. Samples were sectioned at 30 μm.

4.11. Histology

Sectioned samples were immersed in DI water for 5 min before histological staining. To stain sulfated glycosaminoglycans (sGAGs), 0.6% w/v Safranin O in 20% v/v ethanol was used. Phosphotungstic/Phosphomolybdic Acid (5%/5% w/v, Electron Microscopy Sciences) was used as a counterstain, and nuclei were stained using Weigert's Hematoxylin (Electron Microscopy Sciences). Samples were then dehydrated in ethanol, cleared in Clear-Rite-3 (Fisher Scientific), and mounted with Permount (Fisher Scientific). To evaluate mineralization, samples were stained with Alizarin Red S (ARS) (pH 7.4, MilliporeSigma). Total collagen deposition was assessed using Masson's Trichrome staining as adapted from the manufacturer's protocol (Thermo Scientific). Briefly, nuclei were counterstained with Weigert's Hematoxylin (Electron Microscopy Sciences). Cytoplasmic components were stained with Biebrich Scarlet-Acid Fuchsin (MilliporeSigma), followed by treatment with Phosphotungstic/Phosphomolybdic Acid (5%/5% w/v, Electron Microscopy Sciences). Finally, collagen fibers were visualized using Aniline Blue (Electron Microscopy Sciences) before mounting.

4.12. Immunohistochemical (IHC) staining

Staining of Type-II, Type-I, and Type-X collagen was carried out as follows. Sectioned samples were immersed in DI water to remove OCT. Enzymatic antigen retrieval with 0.25% Trypsin (Gibco) for 15 min at 37 °C was followed by PBS washing and additional antigen retrieval with 0.3% hydrogen peroxide for 30 min at room temperature. After PBS washing, samples were blocked for 30 min and then incubated overnight in primary antibody overnight at 4 °C at a 1:100 dilution (Abcam 34712, Abcam 34710, and Thermo Scientific 14-9771-82). HRP-conjugated goat anti-rabbit or anti-mouse secondary (Invitrogen) was used at a 1:1000 or 1:500 dilution, respectively and incubated for 1 h at RT. Color development was done with ImmPACT DAB substrate kit (Vector Laboratories) for 5 min, and samples were washed, dehydrated, cleared with Clearite, and mounted in Permount.

4.13. Whole-gel immunofluorescent staining

Whole-gel constructs were fixed in 4% PFA overnight at 4 °C, washed in PBS, permeabilized for 1 h at RT in PBST (0.25% Triton-X in PBS), blocked for 3 h in a blocking solution of 5% BSA and 0.5% Triton-X, and stained with primary antibody overnight at 4 °C. Anti-Ki67, anti-CC3, anti-Type-1 collagen∗, anti-fibronectin, anti-integrin beta-1, anti-Lamin A/C, anti-H3K9Me3, and anti-YAP were used at a 1:100 dilution (Abcam 16667, CST 9664S, Abcam 34710, Invitrogen PA1-23693, Abcam 52971, CST4777, CST9733, CST14074). Secondary antibody (goat anti-rabbit or goat-anti mouse IgG AlexaFluor 647, 1:1000), Molecular Probes ActinRed 555 ReadyProbes, and DAPI were incubated overnight at 4 °C and washed with PBST before imaging on a Leica STELLARIS 5 confocal microscope with a ×20 or ×63 objective. For type-I collagen, fibronectin, and integrin beta-1 staining, cells were imaged as z-stacks taken at 2 μm intervals. The projected images were used for quantification in Fiji 2.3.0. Analysis of nuclear localization of Lamin A/C and H3K9Me3 was done with a custom Python script as previously described [30].

∗Since purchase, Type-1 collagen Ab34710 has been listed as potentially cross-reacting with Type-3 collagen by Abcam.

4.14. Expression vectors and viral transductions

The plasmid pLenti.PGK.LifeAct-GFP.W was a gift from Rusty Lansford (Addgene plasmid # 51010; http://n2t.net/addgene:51010; RRID:Addgene_51010) and was ordered from Addgene, amplified and packaged into lentivirus (Stanford Gene Vector and Virus Core). Passage 2 hMSCs at 70% confluency were infected with virus particles at a multiplicity of 10 with polybrene at a concentration of 10 μg mL⁻¹ in hMSC growth media for 72 h. Cells were then expanded and used at passage 6.

4.15. Live-cell time-lapse imaging and quantifications

Live-cell imaging was conducted as previously described using a Leica STELLARIS 5 confocal microscope using a HCX PL APO CS x63/1.40 OIL objective. Briefly, imaging experiments were conducted starting 24 h after encapsulation, and an incubation chamber was used to maintain a temperature of 37 °C and 5% CO₂. To evaluate cortical F-actin dynamics across various hydrogel conditions, cells were transfected with LifeAct and visualized via time-lapse microscopy. Imaging was performed at two time scales: short-term dynamics were captured every 5 min for a total duration of 25 min, and longer-term dynamics were recorded at 2.5-min intervals over 6 h. The StackReg plugin in Fiji 2.3.0 was used when needed to correct for drift. A custom Python script was used to quantify the percent area of dynamic cortical F-actin normalized by total cortical F-actin. For nuclear imaging, 1 h before imaging was started, NucSpot Live 650 NucSpot Live Cell Nuclear Stain kit (Biotium) was added to culture media. Z-stacks were made from 10 images taken at 3 μm intervals in the z-axis and projected for analysis. A custom Python script was used to draw color-coded outlines of single nuclei over time and compute average centroid linear. The analysis was done between frames taken 5 min apart for 25 min in this study.

4.16. Brightfield- or fluorescent bead-based pocket and cell area characterization

Images were obtained using a Leica STELLARIS 5 confocal microscope using a ×63 objective either in live-cell chambers or after fixation for end-point analyses. For images with F-actin and nuclear staining, hydrogels were incubated with molecular Probes ActinRed 555 ReadyProbes, and DAPI overnight at 4 °C. For experiments with embedded fluorescent beads, cells were encapsulated with 8% FluoSpheres carboxylate-modified microspheres (0.2 μm, yellow-green, 505/515, Invitrogen).

4.17. YAP nuclear localization quantification

Methods were adapted from previous work [65,70,71]. Briefly, normalized YAP nuclear localization was quantified using FIJI. Whole cell masks were generated using phalloidin staining and nuclear masks were generated using DAPI staining. Cytoplasm masks were generated from the whole cell excluding the nucleus. For each cell, YAP mean fluorescence intensity was measured in the nucleus and normalized to the mean fluorescence intensity in the cytoplasm.

4.18. Reverse transcription quantitative polymerase chain reaction (RT-qPCR)

Hydrogels were transferred to TRIzol (Thermo Fisher Scientific), crushed, and homogenized for phenol-chloroform RNA extraction and reverse transcript PCR (RT-PCR). Following preparation of cDNA using Primescript RT Master Mix (Takara Bio) in a MiniAmp Plus Thermal Cycler, q-PCR was conducted using Power SYBR Green PCR Master Mix (Applied Biosciences) and an Applied Biosystems QuantStudio 6 Pro Real-Time PCR system. Primer sequences are provided in the Supplementary Information (Supplementary Table 1).

4.19. Western blotting

Hydrogels were washed in PBS for 30 min then submerged in 3x RIPA lysis buffer containing 3x Halt protease and phosphatase inhibitor (Fisher Scientific PI78441) and 3% sodium dodecyl sulfate (SDS) to account for the volume of PBS retained in the gels. Gels were manually crushed, homogenized, and centrifuged. The supernatant was collected, and any remaining gel and/or liquid was filtered using nylon centrifugal filters (0.45 μm pore size, VWR). Flow through was collected and combined with the supernatant. A bicinchoninic acid (BCA) assay was used to quantify protein concentration, and normalized protein samples, sample buffer, and reducing agent were combined and boiled at 100 °C for 5 min. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was used to separate lysates with 4-12% Bis-Tris gels (Thermo Scientific) before transfer to PVDF membranes using an iBlot3 (Invitrogen). Blocking with 5% BSA in Tris-buffered saline with Tween 20 (TBST) for 1 h at room temperature was followed by overnight incubation with primary antibodies diluted in SuperBlock blocking buffers (Thermo Fisher Scientific) at 4 °C. Membranes were washed thrice in TBST and incubated in secondary antibody diluted in blocking buffer for 1.5 h at room temperature followed by another round of washing in TBST. Development of the membrane was done with SuperSignal West Femto Maximum Sensitivity Chemiluminescent Substrate (Thermo Fisher Scientific) and imaged using an Invitrogen iBright CL1500 imaging system. Quantification of bands was done in Fiji 2.3.0. Primary and secondary antibodies and dilutions are listed as follows: anti-Lamin A/C (CST 2032, 1:1000), anti-H3K9Me3 (CST 13969, 1:1000), anti-integrin beta 1 (CST, 1:1000), anti-GAPDH (CST 5174, 1:10000), and goat anti-rabbit IgG (HRP) (Abcam 6721, 1:100000). For figures, bands were cropped to show proteins of interest and matched with corresponding loading control from the same gel. Full blot images are available upon request.

4.20. Protease array

Protease protein levels were analyzed in conditioned media collected from the first 48 h after encapsulation. Three hydrogels were pooled, and the array was carried out according to the manufacturer's instructions. The kit was the Proteome Profiler Human Protease Array (ARY021B, R&D Systems) and the arrays were imaged using an Invitrogen iBright CL1500 imaging system. Spots were quantified using Fiji 2.3.0.

4.21. Drug treatments

The following compounds were added to chondrogenic media for the first three days of culture: 50 μM Blebbistatin (MedChem Express); 10 μM Y27632 (Sigma-Aldrich); and DMSO was also added to relevant control groups.

4.22. RGD conjugation in SG

To conjugate RGD cell-binding domains to SG polymers, the adhesion peptide cysteine–arginine–glycine–aspartate–serine (CRGDS, GenScript) was added to the polymer precursor solution at a concentration of 1 mM. Gelation of SG was completed as outlined above.

4.23. Statistical analysis

All statistical analyses were done using GraphPad Prism 10 software and details are included in the figures and figure legends. Error bars represent the standard deviation. Lines in violin plots represent the median and quartiles.

4.24. Custom Python Scripts

Scripts were adapted from previous work [30]. Updated code can be found via GitHub at https://github.com/Stanford-Fan-Yang-lab/CellPocket.

CRediT authorship contribution statement

Sarah J. Loveland: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Xinming Tong: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Funding acquisition, Conceptualization. Manish Ayushman: Writing – review & editing, Methodology, Investigation, Conceptualization. Hung-Pang Lee: Writing – review & editing, Investigation, Conceptualization. Julia M. Johannsen: Investigation, Conceptualization. Callie M. Weber: Writing – review & editing, Software, Investigation, Conceptualization. Jake Song: Writing – review & editing, Investigation. Michelle Tai: Writing – review & editing, Investigation. Georgios Mikos: Software. Yara Chaib: Investigation. Ovijit Chaudhuri: Writing – review & editing, Supervision, Resources. Fan Yang: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Funding acquisition, Data curation, Conceptualization.

Data availability statement

The data that supports the findings of this study is available from the corresponding author upon reasonable request.

Ethics approval and consent to participate

This study does not include any clinical investigations, animal experimentation, or research involving human subjects, including the use of donated organs or tissues.

Funding

This work was supported by the National Institutes of Health [R01DE024772 (F.Y.), R01AR074502 (F.Y.), NIH F31AR083254 (S.J.L.), NIH F32AR084286 (J.S.), NIH NIAMS R01AR081993 (O.C.)], the National Science Foundation Graduate Research Fellowship Program [NSF-GRFP DGE-1656518 (S.J.L.)], the Bio-X Stanford Interdisciplinary Graduate Fellowship (M.A. and M.T.), the Stanford Taiwan Science and Technology Hub Fellowship (HP.L), the Propel Postdoctoral Scholars Program (C.M.W.), the Stanford School of Medicine Dean's Postdoctoral Fellowship (C.M.W), and the Stanford Graduate Fellowship (M.T.).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: