Dynamically Crosslinked PEG Hydrogels Reveal a Critical Role of Viscoelasticity in Modulating Glioblastoma Fates and Drug Responses in 3D

Abstract

Glioblastoma multiforme (GBM) is the most prevalent and aggressive brain tumor in adults. Hydrogels have been employed as 3D in-vitro culture models to elucidate how matrix cues such as stiffness and degradation drive GBM progression and drug responses. Recently, viscoelasticity has been identified as an important niche cue in regulating stem cell differentiation and morphogenesis in 3D. Brain is a viscoelastic tissue, yet how viscoelasticity modulates GBM fate and drug response remains largely unknown. Using dynamic hydrazone crosslinking chemistry, we report a poly(ethylene-glycol) (PEG)-based hydrogel system with brain-mimicking stiffness and tunable stress relaxation to interrogate the role of viscoelasticity on GBM fates in 3D. The hydrogel design allows tuning stress relaxation without changing stiffness, biochemical ligand density, or diffusion. Our results reveal that increasing stress relaxation promotes invasive GBM behavior, such as cell spreading, migration, and GBM stem-like cell (GSC) marker expression. Furthermore, increasing stress relaxation enhances GBM proliferation and drug sensitivity. Stress-relaxation induced changes on GBM fates and drug response were found to be mediated through the cytoskeleton and transient receptor potential vanilloid-type 4 (TRPV4). These results highlight the importance of incorporating viscoelasticity into 3D in-vitro GBM models and provide novel insights into how viscoelasticity modulates GBM cell fates.

1. Introduction

Glioblastoma multiforme (GBM) is the most prevalent and aggressive primary brain cancer in adults.^[1] The standard-of-care for GBM consists of surgical resection followed by concurrent radiotherapy and chemotherapy. Despite such aggressive treatment interventions, GBM patients face a bleak 5-year survival rate of 7% and a median survival of less than two years.^[2,3] Such dismal outcomes are largely attributed to GBM’s highly diffusive invasion into surrounding brain tissue. GBM’s diffuse nature makes complete resection impossible without interfering with normal brain function, and residual migrating cells outside of the tumor core are generally unaffected by local therapies and invariably lead to recurrence.^[4,5] GBM tumors contain stem-like cells (GSCs), expressing putative markers such as CD133, Nestin, and Sox2, and have been found to be essential in driving invasion and disease propagation into healthy brain parenchyma, ultimately resulting to poor patient outcomes.^[6] To improve treatment strategies for GBM, it is critical to understand key factors driving aggressive GBM invasion and phenotype.

Towards this goal, engineered biomaterials-based hydrogels have been utilized to create 3D in-vitro brain tumor models to elucidate how brain extracellular matrix (ECM) properties modulate GBM progression and drug responses.^[7] Using synthetic and naturally derived materials like poly(ethylene-glycol) (PEG) or collagen-I, respectively, previous studies have demonstrated that matrix stiffness, degradability, and biochemical cues are critical in driving GBM malignant phenotype and progression in 3D.^[8–11] Recently, viscoelasticity has been discovered as a matrix niche cue that modulates multiple biological processes including stem cell differentiation, morphogenesis, and maintenance of neural progenitor stemness.^[12–15] Importantly, the brain tissue is viscoelastic and exhibits stress relaxation, or a time-dependent decrease in storage or elastic modulus when under constant strain.^[13,16] Furthermore, magnetic resonance elastography measurements demonstrate that GBM tumors are viscoelastic as well.^[17,18] However, how viscoelasticity modulates GBM fate and drug response in 3D remains largely unknown. Previously established PEG hydrogels were covalently crosslinked, rendering the resulting hydrogel network to be elastic in their response to mechanical forces, and thereby lack stress relaxation.^[8,9] While collagen-I matrices are viscoelastic, varying the polymer concentration to tune the stress relaxation profile simultaneously induces changes in the stiffness and biochemical ligand density.^[11,19] As a result, there remains a critical need to develop hydrogels with tunable stress relaxation without confounding factors to elucidate the unknown role of viscoelasticity on driving GBM cell fates in 3D.

To introduce viscoelasticity in 3D hydrogels, dynamic covalent chemistries have been recently reported for crosslinking, which allows the hydrogel network to adapt and reorganize to dissipate stresses at the molecular level in response to deformation.^[20] Specifically, hydrazone crosslinking exhibits reversibility under physiological conditions and has shown biocompatibility in supporting 3D culture of myoblasts, chondrocytes, and neurons^[21–23]. Hydrazone bonds (R–HC=NH–NH–R) are formed when a carbonyl electrophile, such as an aldehyde, attacks a nucleophilic hydrazine in a condensation reaction.^[24] The resulting hydrazone bond is reversible, and the chemical equilibria of the bonds, which governs the rate of network reorganization, can be modulated via the choice of the aldehyde end group, either using an alkyl-aldehyde or a benzaldehyde. Consequently, the stress relaxation properties of the resulting hydrogel network can be easily tuned by varying the molar ratios of the resulting alkyl-hydrazone (AH) and benzyl-hydrazone (BH) bonds.^[21,22] However, this chemistry has not been used to culture and study GBM in 3D.

Here, we report a PEG-based hydrogel system with tunable stress relaxation to investigate the role of matrix viscoelasticity on GBM tumor fate in 3D using the hydrazone crosslinking chemistry. In this work, we functionalize PEG macromers with reactive components to form hydrazone hydrogels with AH and/or BH crosslinks. To allow for cell adhesion, collagen-I is incorporated at a constant concentration to form an interpenetrating network (IPN) with the PEG hydrazone-crosslinked network. We demonstrate that increasing stress relaxation in 3D promotes invasive behavior of GBM including cell spreading, migration, and GSC marker expression. Furthermore, increasing stress relaxation leads to enhanced proliferation and drug sensitivity. We further elucidate that stress relaxation-induced changes in GBM cell responses are modulated through the cytoskeleton and transient receptor potential vanilloid-type 4 (TRPV4). Pharmacological inhibition studies were also performed to validate cytoskeletal tension and TRPV4 are required to induce invasive GBM behavior and drug sensitivity in stress relaxing hydrogels.

2. Results and Discussion

Developing dynamically crosslinked PEG hydrogels with tunable stress relaxation and brain-mimicking stiffness

To interrogate the role of viscoelasticity on GBM tumor fate, we first established and characterized dynamically crosslinked PEG hydrogels with tunable stress relaxation and brain-mimicking stiffness. To form the hydrazone-crosslinked PEG network, three 8-arm PEG macromers (MW ~ 10kDa) were synthesized with hydrazine, alkyl-aldehyde (AA), or benzaldehyde (BA) end groups (Figure 1a). PEG hydrazine can react with AA or BA to form either dynamic (alkyl-hydrazone, AH) or stable (benzyl-hydrazone, BH) crosslinks (Figure 1b). Three hydrogel groups with tunable stress relaxation profiles were achieved by varying the molar ratio of AH:BH crosslinks (100:0, 70:30, 0:100) used within the PEG network (Figure 1c). Collagen-I was incorporated at a constant concentration to provide cell adhesive cues (Figure 1c). In GBM tumors, collagen-I has been found to be elevated and plays an important role in tumor progression through providing adhesion sites and a fibrillar architecture ^[25–27].

Figure 1.

Dynamically crosslinked PEG hydrogels demonstrate tunable stress relaxation with brain-mimicking stiffness. a) Schematic representing 8-arm PEG macromers functionalized with reactive groups that enable hydrazone crosslinking. b) Chemical structures of dynamic alkyl-hydrazone (blue) and stable benzyl-hydrazone (red) crosslinks. c) Schematic depicting how varying the molar ratio of alkyl-hydrazone:benzyl-hydrazone (AH:BH) crosslinks allows formation of three hydrogel groups with tunable stress relaxation. Collagen-I (green) is incorporated at a constant concentration to provide cell adhesion. d-g) Characterizing the mechanical property of the three hydrogel formulations using shear rheology. Mouse brain was included as a positive control. d) Representative stress relaxation profiles of fast-, medium-, and slow-relaxing hydrogels and mouse brain. e) Time for the normalized modulus to reduce to half its original value, τ_½, from stress relaxation tests. f) Loss tangent measurements. g) Young’s modulus. One-way ANOVA with Dunnett’s multiple comparisons test was used for analysis of the data in e-g, comparing with mouse brain: ns, not significant; ****p< 0.0001; n = 6, 5, 5, and 5 independent samples (fast, medium, slow, and mouse brain). Data reported in e-g represent mean value ± s.d.

Rheological testing confirmed that varying the molar ratio of AH:BH crosslinks led to hydrogels with tunable stress relaxation (Figure 1d). The τ_1/2 (the time for modulus to be relaxed to half its original value in shear) were ~187 s, ~1600 s, and ~21,000 s for the hydrogels, which were termed fast-, medium-, and slow-relaxing hydrogels, respectively (Figure 1e). In particular, the fast-relaxing hydrogels exhibited comparable stress relaxation and τ_1/2 to that of mouse brain (τ1/2 ~ 133s), demonstrating that fast-relaxing hydrogels recapitulate brain-mimicking stress relaxation (Figure 1d,e). Consistent with the stress relaxation results, further rheological characterization demonstrated that the loss tangent (another viscoelastic property and a measurement of energy dissipation) in fast-relaxing hydrogels was also comparable to that of mouse brain and diminished with decreasing stress relaxation (Figure 1f). To ensure that stiffness was not a confounding factor, the concentration of PEG and collagen-I used was held constant across all hydrogel groups. The results show that all three hydrogel formulations exhibited a Young’s Modulus around 400 Pa, similar to that of mouse brain (Figure 1g). Furthermore, all hydrogel formulations demonstrated minimal degradation, as verified by relative stable swelling ratio, over the course of 6 days – the duration of the longest cell culture experiment in the present study (Figure S2a). By day 6, the fast-relaxing hydrogel group demonstrated a slight decrease in polymer retention, which is likely attributed to the dynamic nature of AH crosslinks (Figure S2b). This was accompanied by a slight increase in swelling ratio in the fast-relaxing hydrogels (Figure S2c). Using confocal reflectance imaging, we further confirmed that tuning stress relaxation does not change the collagen distribution or fibrillar architecture across the different hydrogel formulations (Figure S2d). Moreover, collagen-I was stably retained within the IPN over 6 days (Figure S2d). Lastly, all hydrogel formulations supported high cell viability of GBM cells after encapsulation and throughout culture (Figure S2e). Together, these data confirm that dynamically crosslinked PEG hydrogels can serve as a 3D niche for GBM with tunable stress relaxation and brain-mimicking stiffness, without confounding factors of biochemical cues.

Increasing stress relaxation promotes GBM cell spreading, motility, and proliferation

We first investigated how tuning stress relaxation impacts GBM invasion, a hallmark feature of GBM disease progression.^[28] Since cell spreading is a prerequisite for migration,^[29,30] GBM cell morphology was first assessed. A patient-derived tumor xeno-graft (PDTX) GBM cell line (D-270 MG) was used given its ability to retain critical features of the parental tumor and inherent intratumoral heterogeneity.^[31] After three days of culture in hydrogels, D-270 MG cells displayed distinct morphological differences in response to tuning stress relaxation. Increasing stress relaxation promoted robust cell spreading, whereas cells exhibited rounded morphologies in slow-relaxing hydrogels (Figure 2a). Quantification of cell roundness corroborated these observations, as decreasing stress relaxation resulted in increasing the distribution of rounded cells (Figure 2b). A similar trend in GBM morphology in response to tuning stress relaxation was confirmed using U-87 MG, a well-established GBM cell line (Figure S3a, b).

Figure 2.

Stress relaxation promotes GBM cell spreading, migration, and proliferation in 3D. a) Representative maximum intensity projection images of membrane stained (R18) D-270 MG cells cultured within fast-, medium-, and slow-relaxing hydrogels on day 3. Scale bar, 100 µm. b) Roundness quantification of D-270 MG cells in the three hydrogel groups on day 3. One-way ANOVA with Tukey’s multiple comparisons test was used for analysis of the data: ****p< 0.0001; n = 498, 343, and 216 (fast, medium, and slow) cells across 3 independent biological replicates. The dashed lines in the violin plots represent median values. c) Brightfield time-lapse imaging of a single cell within fast- and slow-relaxing hydrogels over 90 minutes. Times are indicated in min:s. Scale bar, 20 µm. d) Representative 3D track reconstructions for cell migration in the three hydrogel groups from time-lapse imaging. 80 randomly selected cell migration track trajectories are shown for each condition. Grid size, 10 µm. e-g) Analysis of cell migration track data from time-lapse imaging: e) probability of cell migration, f) mean speed, and g) migration track length of cells tracked within the three hydrogel groups. One-way ANOVA with Tukey’s multiple comparisons test was used for analysis of the data: *p<.05, ***p<.001, ****p < 0.0001; n = 855, 500, and 556 (fast, medium, and slow) cells tracked across 3 independent biological replicates. Bars indicate mean value ± s.d. The dashed lines in the violin plots represent median values. h) Representative immunostaining images of D-270 MG cells for EdU staining (green) and nucleus (blue) on day 3. Scale bar, 50 µm. i) Fraction of EdU-positive D-270 MG cells on day 3. One-way ANOVA with Tukey’s multiple comparisons test was used for analysis of the data: **p<.01, ****p< 0.0001; n = 212, 191, and 153 (fast, medium, and slow) cells across 3 independent biological replicates. Data reported represent mean value ± s.d.

We next studied the impact of matrix stress relaxation on cell motility at the single cell and population levels using live-cell time-lapse imaging. Single cell motility was monitored over the course of 90 minutes within fast and slow-relaxing hydrogels. D-270 MG cells quickly spread and migrate in a protrusive manner, whereas cells remain rounded and stationary in slow-relaxing hydrogels (Figure 2c). To confirm the observed motility on a population level, image analysis software was used to track the migration of labeled D-270 MG cells in the different hydrogel groups over 14 hrs (Supplementary Movies 1–3). The projected cell migration tracks indicate that cells in fast-relaxing hydrogels were more migratory compared to those in medium and slow-relaxing hydrogels (Figure 2d). Using analysis from the cell track data, the probability that a cell would migrate within a particular hydrogel was further calculated. Cells within fast-relaxing hydrogels demonstrated a ~40% likelihood of migrating, significantly higher when compared to ~13% and 1% in medium- and slow-relaxing hydrogels, respectively (Figure 2e). Cells in fast-relaxing hydrogels also demonstrate significantly higher migration speeds and track lengths, compared to cells in both medium- and slow-relaxing hydrogels (Figure 2f,g). These observations confirm that fast stress relaxation promotes robust GBM cell invasive phenotype including spreading and migratory ability. In contrast, within slow-relaxing hydrogels, GBM cells were uniformly confined and remained stationary.

Furthermore, we assessed the impact of stress relaxation on cell proliferation, another hallmark feature of GBM. The proliferation of D-270 MG cells, as indicated by EdU (5-ethynyl-2′-deoxyuridine), was enhanced in fast-relaxing hydrogels but reduced as stress relaxation was decreased (Figure 2h,i). This finding demonstrates that stress relaxation directly mediates GBM proliferation. Our finding is consistent with a previous report that stress relaxation promotes breast cancer cell proliferation, which was shown using a 3D alginate hydrogel model with tunable viscoelasticity.^[32] Together, these results demonstrate that brain-mimicking fast stress relaxation facilitates more aggressive GBM phenotypes, such as increased cell spreading, migratory ability, and proliferation. These hallmark features are fundamental for GBM tumor invasion, recurrence and growth - highlighting the important role of stress relaxation on aggressive GBM behavior.^[2,33] One motivation for using a PDTX GBM cell line is their ability to better retain intratumoral heterogeneity, a key feature of GBM in vivo.^[31,34] An ideal biomimetic hydrogel niche should support retention of such heterogeneity. Our results indicate that fast-relaxing hydrogels best support retention of GBM heterogeneity, in which cells exhibit varying degrees of spreading, migration and proliferation (Figure 2). In contrast, such heterogeneity in cell behavior is largely lost in slow-relaxing hydrogels, with minimal spreading, migration, or proliferation observed across all cells. These results suggest that stress relaxation may play an important role in supporting retention of GBM heterogeneity, which is important for drug screening and predicting in vivo response.

Stress relaxation promotes stemness of GBM cells

GBM tumors contain GBM stem-like cells (GSCs), which have been shown to drive aggressive GBM phenotype and contribute to drug resistance and tumor recurrence.^[6,34,35] Given that regulation of GSCs is dependent not only on intrinsic factors but also on microenvironmental niche cues,^[6,36,37] we sought to investigate the role of matrix stress relaxation on GSC phenotype within GBM tumors, which has never been studied before. PDTX GBM cells were encapsulated in the three hydrogel formulations with varying stress relaxation for 3 days, and immunostaining was used to characterize the expression of three putative GSC markers - Nestin, CD133, and Sox2. Strikingly, all GSC markers demonstrated the highest expression in fast-relaxing hydrogels, and expression diminished with decreasing stress relaxation (Figure 3a). GSC marker expression was minimal at day 0 in all hydrogel formulations (Figure S4), indicating that fast stress relaxation promotes GSC phenotype over the 3-day culture period. These trends were further validated on a population level using western blot, with significantly higher levels of GSC marker expression in hydrogels with faster stress relaxation (Figure 3b,c). These results suggest that matrix viscoelasticity promotes GBM cells to be in a more stem-like state, and it is important to incorporate viscoelasticity in 3D in-vitro models for interrogating GSC biology and drug responses.

Figure 3.

Stress relaxation promotes glioblastoma stem-like cell (GSC) marker expression in 3D. a) Representative immunostaining images of D-270 MG cells for GSC marker expression (Nestin, CD133, or Sox2; green), F-actin (red), and nucleus (blue) in fast-, medium-, and slow-relaxing hydrogels on day 3. Scale bar, 10 µm. b) Western blot analysis of GSC marker expression by GBM cells in fast-, medium-, and slow-relaxing hydrogels. c) Quantification of western blot images by normalizing each marker to GAPDH. One-way ANOVA with Tukey’s multiple comparisons test was used for statistical analysis: ns, not significant; *p<.05, ***p<.001, ****p < 0.0001; n = 3 independent biological replicates per group. Data reported represent mean value ± s.d.

Enhanced GSC marker expression has also been shown to strongly correlate with enhanced GBM invasive phenotype in vivo.^[35] Together, the observed trend from this 3D model recapitulates the in vivo GBM phenotype, in which fast-relaxing hydrogels support both higher GSC marker expressions (Figure 3) and enhanced migratory ability of GBM cells (Figure 2d–g). Similar to the trend with GSC markers, expression of GFAP, an astrocyte marker commonly used to identify GBM cells,^[38,39] was also the highest in fast-relaxing gels (Fig S5a–c). Given GFAP expression was reduced in slow-relaxing gels over 3 days of culture (Fig S5a, d), we found that viscoelasticity is also important in preserving GFAP expression in GBM cells. Cumulatively, these results reveal a critical role of viscoelasticity in supporting in vivo-mimicking GBM cell phenotype and promoting GBM stemness, which correlates with an increased invasive GBM phenotype.

Stress relaxation induces enhanced GBM drug sensitivity to chemotherapeutics

An important criterion for assessing the physiological relevance of a 3D cancer model is its potential for drug screening and ability to recapitulate drug responses in vivo. Having established that viscoelasticity plays a critical role in regulating invasive GBM phenotype, we next studied how stress relaxation impacts GBM drug response to clinically used chemotherapeutics. Two model drugs were tested including Temozolomide (TMZ) and Carmustine (BCNU). For each drug, two GBM cell lines (D-270 MG and U-87 MG) were tested. GBM cells were cultured in the different stress relaxing hydrogel groups for 3 days followed by an additional 3 days of drug treatment. The range of drug concentrations were selected based on previous reports.^[9,40] At low TMZ concentration (3 – 100 µM), both GBM cell lines showed minimal response. At 300 µM and 600 µM, cells within fast-relaxing hydrogels displayed enhanced drug sensitivity compared to slow-relaxing hydrogels. At the very high drug concentration (1000 µM), cytotoxicity was saturated, and no significant difference was seen across the hydrogel groups (Figure 4a). A similar trend was observed in U-87 MG cells (Figure 4b). It is important to note that comparable TMZ diffusion was observed across all hydrogel groups (Figure S6a), eliminating drug diffusion as a potential contributor to the observed differences in drug response. The effect of viscoelasticity on GBM drug response was further verified with BCNU, another drug used for treating GBM patients. Similar to the trend observed with TMZ, D-270 MG and U-87 MG displayed enhanced drug sensitivity in fast-relaxing hydrogels compared to slow-relaxing hydrogels (Figure 4c,d). Together, these results reveal that stress relaxation increases GBM drug sensitivity.

Figure 4.

Stress relaxation increases drug sensitivity of GBM cells in 3D. Two model chemotherapeutic drugs were used, including (a,b) Temozolomide (TMZ) and (c, d) Carmustine (BCNU), to treat two GBM cell lines (D-270 MG and U87-MG). Relative cell viability was reported by normalizing treated to untreated GBM cells in fast (blue), medium (purple), and slow (red) stress relaxing hydrogels. Both GBM cell lines demonstrate enhanced chemosensitivity to both TMZ and BCNU as stress relaxation increases. Two-way ANOVA with Tukey’s multiple comparisons test was used for statistical analysis: *p<.05, **p<.01, ***p<.001, ****p< 0.0001; n = 3 independent biological replicates per condition. Data reported represent mean value ± s.d.

Like many chemotherapeutics, TMZ and BCNU’s mechanism of action depends on cell proliferation.^[41–43] Given GBM cells displayed enhanced proliferation in fast-relaxing hydrogels (Figure 2h,i), we speculated that the enhanced drug sensitivity in fast-relaxing hydrogels was driven by the increased proliferation. To test this, D-270 MG cells were treated with mitomycin C to inhibit cell proliferation. Indeed, viscoelasticity-induced GBM drug sensitivity was abolished with mitomycin C treatment, with all groups exhibiting comparable drug response as the slow-relaxing hydrogels (Figure S6b).

Furthermore, GSCs have been postulated to be important mediators in GBM chemotherapy response.^[6] Our data demonstrated that increasing stress relaxation enhanced both GSC marker expression and chemotherapy sensitivity (Figures 3 and 4). This correlation mimics previous in vivo findings, in which tumors with enriched GSCs showed enhanced drug sensitivity to TMZ.^[44,45] However, there exists conflicting reports suggesting GSCs can evade drug treatment, contributing to overall GBM drug resistance and tumor growth.^[46,47] While normal neural stem cells are quiescent, GSCs may present varying degrees of stemness with distinct proliferative capacities, which may explain conflicting findings on GSC susceptibility to chemotherapy.^[48] In addition, it has also been suggested that TMZ treatment can induce a phenotypic shift in which GSCs adaptively transition from a sensitive to a more drug resistant state.^[49] Future studies can harness this 3D model to characterize the drug resistant cell subpopulation after initial chemotherapeutic treatment to better understand the relationship between GSCs and chemotherapy. It should be noted that TMZ response has also been found to be patient cell-line specific and may depend on other factors such as methylated O⁶-DNA methylguanine-methyltransferase (MGMT) status in the patient’s tumor.^[50] Future studies can therefore utilize this 3D viscoelastic GBM model to investigate other potential mechanisms GBM cells use to evade chemotherapeutic treatment.

Cytoskeletal tension is required for stress relaxation-induced GBM spreading, motility, and chemosensitivity

Given that the cytoskeleton plays a major role in governing how cells sense and transduce mechanical cues from their environment,^[51–53] we next probed how tuning stress relaxation impacts D-270 MG cytoskeletal organization. Immunostaining was performed for two key cytoskeletal markers, F-actin and myosin-IIa. Increasing stress relaxation enhanced cell spreading and promoted the formation of robust actomyosin stress fiber bundles (Figure 5a), characteristic of a high-tensional state.^[54] Furthermore, distinct F-actin organization was observed across the hydrogel groups. D-270 MG cells in slow-relaxing hydrogels displayed cortical F-actin with narrow protrusions penetrating the matrix; however, as stress relaxation was increased, F-actin stress fibers were visibly discrete and elongated (Figure 5a).

Figure 5.

Stress relaxation enhances cytoskeletal formation in GBM cells in 3D, and pharmacologically disrupting the cytoskeleton abrogates stress relaxation-induced GBM spreading, migration and drug sensitivity. a-c) Representative immunostaining images of D-270 MG cells for myosin IIa (green), F-actin (red), and nucleus (blue) within fast-, medium-, and slow-relaxing hydrogels on day 3: a) vehicle-alone (control), b) a myosin inhibitor (+ Blebbistatin), or c) an inhibitor of actin polymerization (+ Cytochalasin D). Scale bar, 10 µm. d) Representative 3D track reconstructions for cell migration in fast-relaxing hydrogels with vehicle alone, Blebbistatin, or Cytochalasin D treatment from time-lapse imaging. 80 randomly selected cell migration track trajectories are shown for each condition. Grid size, 10 µm. e-g) Analysis of cell migration track data from time-lapse imaging: e) probability of cell migration, f) mean speed, and g) migration track length of cells tracked within fast-relaxing hydrogels. One-way ANOVA with Tukey’s multiple comparisons test was used for data analyses: ns, not significant; ****p< 0.0001; n = 670, 845, and 710 (control, BLEB, and CytoD) cells tracked across 3 independent biological replicates. Bars indicate mean value ± s.d. The dashed lines in the violin plots represent median values. h) Relative cell viability of D-270 MG cells treated with Temozolomide (TMZ only), TMZ with Blebbistatin (TMZ + BLEB), or TMZ with Cytochalasin D (TMZ + CytoD). Two-way ANOVA with Tukey’s multiple comparisons test was used for analysis of the data: **p<.01, ****p< 0.0001; comparison with TMZ only: #p<.01, ##p<.0001; n = 3 independent biological replicates per condition. Data reported represent mean value ± s.d.

We next assessed whether the actomyosin cytoskeleton is required for GBM cells to assume invasive phenotype in response to stress relaxation. Two common cytoskeletal inhibitors were used - blebbistatin (a myosin II inhibitor) and cytochalasin-D (an actin polymerization inhibitor) - to disrupt the actomyosin cytoskeleton. Both inhibitor treatments significantly diminished D-270 MG cell spreading and reduced myosin-IIa staining signal across all hydrogel groups relative to the control. Interestingly, subtle differences in F-actin formation and organization were observed when treated with either blebbistatin or cytochalasin-D. Although the F-actin stress fibers were completely disorganized due to blebbistatin treatment, cells extended disjointed protrusions into the matrix of all hydrogel groups. However, upon cytochalasin-D treatment, cells were completely rounded with F-actin displaying large puncta-like formations regardless of the degree of stress relaxation (Figure 5a–c). These results suggest that enhanced cell spreading within fast stress-relaxing hydrogels is mediated through the actomyosin cytoskeleton. Since GBM cell migration has been demonstrated to depend on actomyosin-generated contractile forces,^[55] we also examined how disrupting the actomyosin cytoskeleton would impact GBM migratory potential within fast-relaxing hydrogels, which best recapitulated the stress relaxation properties of brain tissues. Live-cell tracking analysis from time-lapse imaging confirmed both blebbistatin and cytochalasin-D treatments drastically reduced GBM cell motility (Figure 5d–g, Supplementary Movies 4–6). Together, these results indicate that the actomyosin cytoskeleton is required for GBM cells to sense and respond to the viscoelastic matrix cues.

Having established the importance of the cytoskeleton in mediating stress relaxation-induced GBM spreading and migration, we further tested if the cytoskeleton is required for cells to respond to TMZ within viscoelastic matrices. Remarkably, blebbistatin or cytochalasin-D treatment largely abolished stress relaxation-induced changes in TMZ response (Figure 5h). A similar trend was observed in U-87 MG cells within fast-relaxing hydrogels (Figure S7). These data validate that stress relaxation sensitizes GBM cells to chemotherapy through the enhanced cytoskeletal organization. Cytoskeletal organization and tension have been shown to impact cell proliferation, and interfering actomyosin contractility can inhibit cell proliferation.^[56,57] Therefore, disrupting stress relaxation induced actomyosin tension could potentially reduce TMZ response by inhibiting cell proliferation.

Stress relaxation promotes TRPV4 expression, and modulating TRPV4 activity impacts GBM cytoskeleton, motility, and drug response

Recent studies have demonstrated that the transient receptor potential vanilloid–4 (TRPV4) calcium ion channel is an important molecular sensor of matrix viscoelasticity.^[32,58] Interestingly, TRPV4 is upregulated in GBM, negatively correlates with patient prognosis, and plays a significant role in glioma motility and invasion.^[59,60] Moreover, in GBM and other cancers, TRPV4 has been found to be involved in cytoskeletal regulation, specifically impacting F-actin organization.^[59,61,62] However, previous studies on TRPV4 in GBM were limited to 2D studies. Therefore, it remains unknown how TRPV4 regulates GBM behavior such as spreading, migration and drug response within a 3D brain-mimicking viscoelastic matrix.

To fill in this gap of knowledge, we first characterized TRPV4 expression in GBM cells in 3D hydrogels with tunable stress relaxation. Both immunostaining and western blot showed increasing stress relaxation led to significantly higher TRPV4 expression by GBM cells (Figure 6a–c). Minimal TRPV4 expression was detected by immunostaining at Day 0 (Figure S8a), suggesting that stress relaxation promotes TRPV4 expression over time. To further interrogate how modulating TRPV4 level regulates GBM cell behavior in fast-relaxing hydrogels, TRPV4 activity was modulated using an agonist (GSK101) and an antagonist (GSK205). For GBM cell spreading, antagonist treatment severely disrupted the F-actin cytoskeletal organization, resulting in a rounded cell morphology, whereas agonist treatment showed no significant change in F-actin organization or cell spreading compared to control (Figure 6d). Within medium- and slow-relaxing hydrogels, modulating TRPV4 activity with GSK101 and GSK205 resulted in minimal to modest changes in F-actin organization compared to control groups (Figure S8). This was expected since TRPV4 expression diminished as stress relaxation decreased (Figure 6a–c).

Figure 6.

Stress relaxation promotes TRPV4 expression in 3D, and modulating TRPV4 activity impacts GBM cytoskeleton, migration and drug response. a) Representative immunostaining images of D-270 MG cells for TRPV4 (green), F-actin (red), and nucleus (blue) within fast-, medium-, and slow-relaxing hydrogels on day 3. Scale bar, 10 µm. b) Western blot analysis of TRPV4 for cells in fast-, medium-, and slow-relaxing hydrogels. c) Quantification of TRPV4 expression normalized to GAPDH from western blot. One-way ANOVA with Tukey’s multiple comparisons test was used for analysis of the data: ns, not significant; **p<.01, ***p<.001, ****p< 0.0001; n = 3 independent biological replicates per group. Data reported represent mean value ± s.d. d) Representative immunostaining images of D-270 MG cells for TRPV4 (green), F-actin (red), and nucleus (blue) within fast-relaxing hydrogels treated with vehicle-alone (control), a TRPV4 agonist (GSK101), or a TRPV4 antagonist (GSK205). Scale bar, 10 µm. e) Representative 3D track reconstructions for cell migration in fast-relaxing hydrogels with vehicle-alone (control), GSK101, or GSK205 treatment from time-lapse imaging. 80 randomly selected cell migration track trajectories are shown for each condition. Grid size, 10 µm. f-h) Analysis of cell migration track data from time-lapse imaging: e) probability of cell migration, f) mean speed, and g) migration track length of cells. One-way ANOVA with Tukey’s multiple comparisons test was used for analysis of the data: ns, not significant; ****p< 0.0001; n = 670, 913, and 777 (control, GSK101, and GSK205) cells tracked across 3 independent biological samples. Bars indicate mean value ± s.d. The dashed lines in the violin plots represent median values. i) Relative cell viability of D-270 MG cells treated with TMZ only, TMZ + GSK101, or TMZ + GSK205. One-way ANOVA with Tukey’s multiple comparisons test was used for analysis of the data: *p<.05, ***p<.001, ****p< 0.0001; n = 3 independent biological replicates per condition. Data reported represent mean value ± s.d. j) Intracellular calcium level in D-270 MG cells in fast-relaxing hydrogels treated with vehicle-alone (control), GSK101, or GSK205 treatment. One-way ANOVA with Tukey’s multiple comparisons test was used for analysis of the data: *p<.05, ****p < 0.0001; n = 43, 37, and 46 (control, GSK101, and GSK205) cells across 3 independent biological samples. The dashed lines in the violin plots represent median values.

We next assessed how TRPV4 modulation impacts stress relaxation induced GBM migration in fast stress-relaxing hydrogels. Given TRPV4 antagonist treatment drastically disrupts F-actin organization, we hypothesized that blocking TRPV4 activity would also reduce GBM migratory ability within fast-relaxing hydrogels, similar to the trend observed when the GBM actomyosin cytoskeleton was disrupted (Figure 5b–e). As expected, inhibiting TRPV4 with GSK205 significantly reduced GBM cell motility based on cell tracking analysis, while activating TRPV4 with GSK101 showed no difference to control group (Figure 6e–h, Supplementary Movies 4,7,8).

Since disrupting the cytoskeleton reduced stress relaxation-induced drug sensitivity (Figure 5f), we further hypothesized that TRPV4 inhibition would reduce drug sensitivity due to its effect on F-actin organization. Indeed, TRPV4 inhibition with GSK205 decreased drug sensitivity (Figure 6i). Interestingly, TRPV4 agonist GSK101 significantly increased TMZ sensitivity (Figure 6i), which correlated with significantly higher intracellular calcium levels in D-270 MG cells treated with GSK101 (Figure 6j). This suggests that GBM drug sensitivity to TMZ in viscoelastic hydrogels is mediated through TPRV4 signaling and correlates with intracellular calcium level. Due to lower TRPV4 expression within medium- and slow-relaxing hydrogels (Figure 6a–c), GSK101 and GSK205 treatment showed modest to no effect on intracellular calcium levels (Figure S8c,d) and minimal effect on GBM TMZ drug response (Figure S8e). Together, we demonstrate that TRPV4 plays a crucial role in stress relaxation-induced GBM spreading, migration, and drug response. In other words, our findings highlight that TRPV4 expression and activity in GBM is dependent on cells sensing a brain-mimicking viscoelastic environment. Moreover, these results suggest pharmacological modulation of TRPV4 activity may offer a promising strategy to reduce GBM invasion and improve chemosensitivity but done in a spatiotemporally controlled manner to minimize undesirable side effects given TRPV4’s role in various physiological processes.^[63]

Consistent with our observation that TRPV4 activation enhances TMZ sensitivity (Figure 6i,j), previous work has demonstrated that calcium influx triggered by activation of another mechanosensitive ion channel, TRPV2, resulted in increased GBM chemosensitivity. Specifically, TRPV2 triggered calcium influx enhanced drug uptake and synergistically induced GBM apoptosis.^[64] Other mechanosensitive ion channels such as TRPM7 and PIEZO1 can be further investigated as they have been shown to be associated with brain cancer progression,^[63,65] and may also be involved in regulating GBM response to stress relaxation.

Increasing stress relaxation reduces nascent protein deposition, which correlates with higher drug sensitivity

Lastly, the tumor ECM can act as barrier shielding cancer cells from cytotoxic agents.^[66] A recent study showed that local nascent protein deposition and remodeling can directly impact mesenchymal stem cell (MSC) mechanosensing and signaling in 3D hydrogels.^[67] However, it remains unknown how stress relaxation impacts GBM nascent protein deposition. We next investigated if stress relaxation-induced changes in GBM drug response is mediated through changes in nascent protein deposition. Using immunostaining, we first determined the effects of tuning stress relaxation on nascent protein deposition by GBM cells in our 3D hydrogel model. Common ECM components in the brain tumor microenvironment were assessed including laminin, fibronectin, and collagen-IV. GBM cells in the slow-relaxing hydrogels demonstrated the highest amount of nascent protein deposition for all tested ECMs, and increasing stress relaxation significantly reduced the amount of nascent protein deposition (Figure 7a).

Figure 7.

Stress relaxation reduces nascent protein deposition by GBM cells in 3D, which correlates with higher drug sensitivity. Pharmacological inhibition of nascent protein deposition increases GBM drug sensitivity in slow-relaxing hydrogels. a-c) Representative immunostaining images of D-270 MG cells for ECM proteins (Laminin, Fibronectin, or Collagen-IV; green), F-actin (red), and nucleus (blue) on day 3 within fast-, medium-, and slow-relaxing hydrogels. Samples were treated with a) vehicle-alone (control), b) an inhibitor of nascent protein deposition (+ Exo-1), or c) an inhibitor for matrix remodeling (+ TIMP-3). Scale bar, 10 µm. d) Relative cell viability of D-270 MG cells treated with Temozolomide (TMZ only), TMZ with Exo-1 (TMZ + Exo-1), or TMZ with TIMP-3 (TMZ + TIMP-3). Two-way ANOVA with Tukey’s multiple comparisons test was used for analysis of the data, comparing with TMZ only condition: ns, not significant; **p<.01; n = 3 independent biological replicates per condition. Data reported represent mean value ± s.d.

To further probe the effect of modulating nascent protein deposition on TMZ drug response, D-270 MG cells were treated with either Exo-1 or TIMP-3, two inhibitors that either disrupt nascent protein deposition or remodeling, respectively. Immunostaining confirmed Exo-1 treatment significantly reduced nascent protein deposition in slow-relaxing hydrogels. Minimal change was observed in cells within fast- and medium-relaxing hydrogels given that the baseline nascent protein deposition was low (Figure 7a,b). Conversely, TIMP-3 treatment enhanced nascent protein accumulation around cells in all groups due to reduced remodeling, especially laminin in slow-relaxing hydrogels (Figure 7c). It is important to note that use of Exo-I or TIMP-3 displayed minimal disruption to the cytoskeleton (Figure 7a–c). In slow-relaxing hydrogels, where nascent protein deposition was the highest, Exo-1 treatment significantly increased drug sensitivity (Figure 7d), indicating that decreasing nascent protein deposition may reduce the cytoprotective barrier to treatment. TIMP-3 treatment within slow-relaxing hydrogels did not significantly impact drug response, suggesting that the baseline ECM deposited was sufficient in providing a cytoprotective barrier. For fast- and medium-relaxing hydrogels, where the baseline ECM was low (Figure 7a), Exo-1 or TIMP-3 treatment had no significant effect in drug response (Figure 7d).

Overall, we find that increasing stress relaxation reduces nascent protein deposition by GBM cells in 3D. Given that the tumor ECM can act as barrier shielding cancer cells from cytotoxic agents,^[66] our results suggest fast stress relaxation-induced drug sensitivity in GBM cells is in part mediated through reduced nascent protein deposition around the GBM cells. Previous work in small cell lung cancer has demonstrated that adhesion to ECM proteins confers resistance to chemotherapeutic agents via β1-integrin dependent survival signals.^[68] Given that modulating nascent protein deposition and remodeling minimally effected the cytoskeleton or drug responses of GBM cells in fast-relaxing hydrogels, these findings suggest both cytoskeletal organization and nascent protein deposition contribute to stress relaxation-induced changes in GBM drug response in a brain-mimicking niche.

3. Conclusion

In summary, this study established a dynamically crosslinked PEG hydrogel system with tunable stress relaxation and brain-mimicking stiffness as a tool to elucidate the role of viscoelasticity on GBM cell behavior and drug response in 3D. Our findings reveal the critical role of stress relaxation in modulating key GBM features and drug responses (Figure 8). Specifically, brain-mimicking, fast-relaxing hydrogels promoted invasive GBM behavior such as cell spreading, migration, proliferation, and GSC marker expression. Strikingly, these hallmark GBM features were largely lost in slow-relaxing hydrogels, indicating that viscoelasticity is a critical niche cue regulating GBM disease progression in vivo. Importantly, we demonstrate that GBM drug sensitivity is directly impacted by stress relaxation, highlighting the importance of incorporating viscoelasticity in 3D in vitro GBM models used for drug screening and prediction of in vivo drug efficacy. Leveraging this hydrogel as a tool for mechanistic study, we identified that the actomyosin cytoskeleton is an essential mediator of stress relaxation in GBM cells. Linked to the cytoskeleton and an important mechanosensitive ion channel, TRPV4 was also found to be regulated by stress relaxation, and modulation of TRPV4 activity impacts GBM behavior. We further elucidated that stress relaxation impacts GBM drug response through enhanced cytoskeletal organization, upregulated TRPV4 activity, and reduced nascent protein deposition. We envision that this dynamically crosslinked PEG-hydrogel GBM model could provide a powerful tool for discovering novel therapeutics for GBM that would be otherwise missed using in vitro GBM models lacking viscoelasticity.

Figure 8.

A summary schematic of the key findings on the effect of stress relaxation on GBM cell fates and drug responses in 3D. Brain-mimicking, fast-relaxing hydrogels (left panel) promote GBM drug sensitivity and enhances GBM invasive phenotype including cell spreading, migration, and proliferation. Stress relaxation induced drug sensitivity is associated with enhanced actomyosin cytoskeletal formations, GSC marker expression, TRPV4 activity, and reduced nascent protein deposition. Conversely, slow-relaxing hydrogels (right panel) reduce GBM drug sensitivity and induces opposite trends in corresponding cell fates.

4. Experimental Section

Macromer Synthesis

Synthesis of eight-arm PEG hydrazine (PEG-H):

Eight-arm PEG amine (10 kDa, 2 g, Jenkem Technology) and tri-Boc-hydrazinoacetic acid (0.93 g, 1.5 eq per amine, Sigma-Aldrich) were dissolved dichloromethane (DCM, 50 mL), followed by the addition of EDC. HCl (0.61 g, 2.0 eq per amine, Sigma-Aldrich). The reaction was stirred overnight at room temperature (RT). The product was precipitated in cold ether and vacuum dried. The obtained product was then added in 50:50 DCM:TFA for 2 hours. The resulting solution was then precipitated in ether, vacuum dried, and then dialyzed (MWCO 1 kDa) against DI water for 2 days. The solution was then lyophilized to get the final product.

Synthesis of eight-arm PEG alkyl-aldehyde (PEG-AA):

Eight-arm PEG-OH (10 kDa, 2 g, Jenkem Technology), TEMPO (12.5 mg, 0.1 eq hydroxyl, Sigma-Aldrich), and diacetoxyiodo benzene (DAIB, 0.77 g, 3 eq hydroxyl, Sigma-Aldrich) were dissolved in DCM (50 mL) and stirred overnight. The solution was then precipitated in cold ether and dialyzed (MWCO 1 kDa) against DI water for 2 days and lyophilized.

Synthesis of eight-arm PEG benzyl-aldehyde (PEG-BA):

Eight-arm PEG-OH (10 kDa, 2 g) and 4-formyl benzoic acid (0.56 g, 1.5 eq hydroxyl, Sigma-Aldrich) were dissolved in dichloromethane, followed by the addition of N,N’-diisopropylcarbodiimide (0.2 g, 2.0 eq hydroxyl). The reaction was stirred overnight at room temperature and then precipitated in ether to get the product. The obtained product was dialyzed (MWCO 1 kDa) against DI water for 2 days and lyophilized. As previously reported, ^{[21,22] 1}H NMR spectroscopy was used to estimate the degree of modification on the PEG macromers (Figure S1).

Rheological Characterization

Rheological measurements were carried out using a Discovery HR-2 hybrid rheometer (TA Instruments) equipped with 25-mm-diameter top and bottom plates. IPN hydrogels, used for mechanical testing, were directly deposited onto the bottom plate of the rheometer immediately after mixing all IPN hydrogel precursor components together (100 µL final volume). The top plate was immediately lowered and gently spun to spread the hydrogel solution across the plates, forming a 25-mm disk hydrogel. Mineral oil (Sigma) was applied at the edges to prevent sample dehydration. During the gelation period, a time sweep was performed in which the rheometer geometry was oscillated at an amplitude of 1% shear strain and 1 Hz frequency for 1 h at 37°C. As the gelation completed and the storage and loss moduli had equilibrated, Young’s modulus (E) was calculated from the equation:

$$E=2(1+v)G^{*},$$ (1)

where Poisson’s ratio (𝜈) is assumed to be 0.5 and G* is the complex modulus found using the equilibrium values of storage and loss moduli measured. G* was calculated using:

$$G^{*}={\left({G\prime }^2+{G″}^2\right)}^{1/2}.$$ (2)

Equilibrium values of storage and loss moduli were also used to calculate the loss tangent, which is defined as:

$$Loss tangent (tan\delta )=\frac{ Loss Modulus \left(G″\right)}{ Storage Modulus \left(G\prime \right)}$$ (3)

Next, frequency sweep was performed from 0.1 to 100 rad s⁻¹ at 1% constant strain and held at 37°C. The resulting storage and loss moduli values were measured as function of oscillation frequency (ω). Lastly, after the frequency sweep, stress relaxation measurements were conducted at which a constant strain of 10% was applied to the hydrogel at 37°C. The resulting storage modulus and stress were recorded over the course of 1 hr for fast and medium IPNs and 7 hrs for slow IPNs.

To demonstrate physiological relevance of the IPN hydrogel system, rheological characterization was conducted on mouse brain. Briefly, FVB domestic mice (female, 11 weeks of age, Charles River Lab) were euthanized in compliance with NIH and institutional guidelines. Brain was collected immediately after euthanization, and the same rheological testing procedure, as previously described above, was conducted except for the following. The rheometer was equipped with 20-mm x-Hatch top and bottom plates to minimize sample slippage.

Hydrogel Preparation

Functionalized PEG macromers were dissolved in phosphate buffered saline (PBS) and neutralized to pH 7.0 (15% w/v). IPN hydrogels were prepared by mixing stock solutions of PEG-hydrazine, reconstituted collagen-I (TeloCol-10, Advanced Biomatrix), and PEG-aldehyde (PEG-AA and/or PEG-BA) in this successive order. The hydrazine to aldehyde ratio was 1:1 with a final PEG polymer concentration of 2.5% (w/v). To tune the stress relaxation by varying the AH:BH crosslink ratio, the PEG-AA to PEG-BA ratio within the PEG-aldehyde component was adjusted to the following: 100:0, 70:30, or 0:100. A final collagen-I concentration of 2.5 mg mL⁻¹ was used and neutralized to pH 7.0 after addition into precursor solution. Lastly, after addition of PEG-aldehyde, the hydrogel precursor solution was gently mixed for 1 minute before either depositing directly for rheological testing or pipetting into a mold for cell-based studies (described further below).

Cell Culture

A patient-derived human glioblastoma xenograft (PDTX GBM) cell line (D-270 MG) was provided by the laboratory of Dr. Gerald Grant at Stanford University Medical Center and derived as previously reported.^[69,70] D-270 GBM cells were cultured in improved minimal essential zinc option medium (Life Technologies) with 10% fetal bovine serum (FBS, Gibco, Life Technologies) and 0.1% gentamicin (Life Technologies). An immortalized GBM cell line (U-87 MG) was purchased from ATCC and were expanded in cell culture medium consisting of Dulbecco’s minimal essential medium (DMEM, Life Technologies), supplemented with 10% (v/v) FBS (Gibco, Life Technologies), 100 U/mL penicillin, and 100 μg/mL streptomycin. Cells were cultured in standard humidified incubator at 37 °C in a 5% (v/v) CO₂ atmosphere.

3D cell encapsulation in hydrogels

For cell encapsulation, GBM cells were trypsinized using trypsin/EDTA, washed, centrifuged, and resuspended in media. Prior to the addition of PEG-aldehyde, trypsinized GBM cells were homogenously mixed into the hydrogel precursor solution at a final concentration of 1.25M mL^-1. The cell-containing hydrogel precursor solution (55 µL) was dispensed into a cylindrical shaped mold (3 mm in height, 5 mm in diameter). Cell-laden hydrogels were incubated at 37°C for 40 minutes and then pushed out into pre-warmed media. All experiments were carried for 3 days in 24 well plates unless otherwise specified. To assess cell viability of GBM cells within hydrogels shortly after encapsulation and at days 3 and 6, D-270 MG cells were stained with LIVE/DEAD^® Viability/Cytotoxicity Kit (Invitrogen) per manufacturer’s instructions for 20 min in PBS. Images were taken using Leica STELLARIS 5 confocal microscope with HC PL APO 10x/0.40 air objective.

Swelling Ratio and Polymer Retention measurements

Acellular hydrogels (55 µL) were prepared as described above, transferred to a 24-well plate with 1mL D-270 MG media per well, and incubated at 37°C. The swelling ratio (Q) was calculated based on the following formula: $Q=\frac{Wwet}{Wdry}$, where W_wet is the wet weight of the hydrogel at a specified time and W_dry is the initial dry weight of the hydrogel at the beginning of the experiment. To measure the initial W_dry for each group, separate hydrogels were collected several hours after hydrogel formation and freeze-dried. After 6 days, hydrogels were freeze-dried to calculate the polymer retention $\left(\% polymer retention =\frac{ Wdry,day 0}{Wdry,day 6}x100\%\right)$ and equilbirum mass swelling ratio $\left(Q=\frac{ Wwet,day6}{Wdry,day6}\right)$.

Cell morphology quantification

To assess the cell morphology and quantify cell roundness, the cell membrane of D-270 MG and U87 cells were stained with octadecyl rhodamine B chloride (R18, 1:1000 dilution, stock 10 mg mL⁻¹; Thermo Fisher Scientific) on day 3. Briefly, R18 dye was added to cell culture medium and allowed to incubate with cell-laden hydrogels for 1 hr. Hydrogels were then washed three times with PBS. Cells within the hydrogels were then immediately imaged using a Leica STELLARIS 5 confocal micrscope with a HC FLUOTAR L 25x/.95 numerical aperture (NA) water immersion objective. ImageJ analyze particles function was used on maximum projections from obtained z-stacks from confocal imaging to quantify cell roundness.

Tracking cell motility in 3D using time-lapse confocal imaging

For cell motility studies, D-270 MG cells were starved overnight in serum-free medium and then trypsinized and encapsulated in hydrogels. For high magnification live-cell time-lapse imaging to capture single cell motility, cell-laden hydrogels were cast directly into wells of a chambered coverglass (LabTek). The hydrogels were allowed to incubate at 37°C for 40 min, and then were incubated with normal D-270 MG growth media. After one day, brightfield confocal microscopy was used to acquire live-cell images at 15 min intervals for a total of 3 hrs using a Leica STELLARIS 5 confocal microscope with HC PL APO 63x/1.40 NA oil immersion objective.

To track and analyze cell motility on the population level, D-270 cells were also starved overnight in serum free medium. Prior to encapsulation, D-270 MG cell membranes were labeled with R18 (1:1000 dilution, stock 10mg mL⁻¹; Thermo Fisher Scientific). Cell-laden hydrogel precursor solutions were then cast into custom made PDMS molds within a 35mm petri dish to provide sufficient medium volume and ensure high cell viability during the imaging time course. Once cast, hydrogels were allowed to incubate at 37°C for 40 min. After hydrogel gelation, serum-free starvation medium was added. For inhibitor studies against the cytoskeleton or TRPV4, serum-free starvation medium was supplemented with vehicle-alone or inhibitor. One day later, serum-free starvation medium was removed and replaced with D-270 MG growth media three hours prior to the start of imaging. For inhibitor studies, serum-free starvation medium with vehicle-alone or inhibitor was replaced by D-270 MG growth media supplemented with vehicle-alone or inhibitor three hours prior to the start of imaging. Live-cell imaging was conducted at 25 min intervals for at least 14 hrs using a Leica STELLARIS 5 confocal microscope with HC PL APO 10x/0.40 air objective. 62.5 µm z-stack images were acquired for each position at each time point. The mark-and-find feature of the Leica software was used to capture multiple fields of views across multiple samples at every time point. For all time-lapse experiments, hydrogels were placed in an incubated chamber (37°C and 5% CO₂). We note that control conditions were added along with the experimental conditions in each time-lapse experiment.

Imaging Analysis

For cell migration studies, Imaris software (Bitplane) was used to track R18 stained D-270 MG cells from live-cell time-lapse confocal imaging. Briefly, a surface rendering analysis in Imaris was performed to track cell migration along the recorded time frames and z-stacks. The parameters that were used were adapted from previously established methods.^[71] Cells that were at the edge of the field of view were excluded from analysis. Built in drift correction in Imaris was used as necessary. A custom MATLAB script was used to reconstruct 3D cell migration trajectories from cell track position output analysis. To calculate the probability of cell migration, cells were binned into nonmotile and motile populations, in which the threshold for a motile cell was a track displacement length of at least 15 µm over the 14 h time-lapse. Lastly, mean track speed and track length were used to further assess differences among cells cultured in different stress relaxing hydrogels.

Inhibitor treatments used in live-cell imaging and drug response studies

The following inhibitors and their respective concentrations were used: blebbistatin (250 µM; Abcam, ab120425), cytochalasin-D (500nM; Sigma-Aldrich, C8273), GSK1016790A (GSK101, 50nM; Abcam, ab146191), GSK205 (10µM; Aobious, AOB1612), Exo-1 (120nM, EMD Millipore, 341220), TIMP-3 (5nM; R&D Systems, 973-TM-010). DMSO was the vehicle-alone control for blebbistatin, cytochalasin-D, GSK101, GSK205, and Exo-1. Deionized water was used as the vehicle control for TIMP-3 treatment.

Drug response

As previously described, GBM cells were encapsulated and cultured within hydrogels for three days. At day 3, fresh medium was added with either temozolomide (TMZ, Sigma-Aldrich, T2577) or Carmustine (BCNU, Sigma-Aldrich, C0400). Concentrations used for both TMZ and BCNU were the following: 3, 30, 100, 300, 600 and 1000 µM. Hydrogels were cultured in medium supplemented with drug for an additional three days. Based on manufacturer’s instructions, Presto Blue cell viability reagent (Thermo Fisher Scientific) was used to measure the cell viability at the end of the drug treatment period (day 6). For both TMZ and BCNU, DMSO was used as the vehicle control. Relative cell viability was calculated by normalizing treatment groups to untreated vehicle alone control.

For drug response studies using inhibitors in combination with TMZ, a similar procedure was used as described above with the following exceptions. Inhibitors targeting the actomyosin cytoskeleton (blebbistatin and cytochalasin-D) or modulating TRPV4 activity (GSK101 and GSK205) were added 24 hrs prior to TMZ administration. Inhibitors for nascent protein deposition and remodeling (Exo-1 and TIMP3) were added shortly after encapsulation and replenished halfway through the initial three-day culture period. All inhibitors were replenished in fresh medium along with TMZ (300 µM) at day 3, the start of drug treatment.

To halt cellular proliferation, D-270 MG cells were treated with mitomycin C (10 µg mL⁻¹; Abcam, ab120797) for 2 hrs prior to trypsinization and encapsulation. Similar to the procedure above, mitomycin C treated cells were then encapsulated within hydrogels, cultured for three days, and then treated with vehicle-alone control or TMZ (300 µM) for an additional three days. Relative cell viability was calculated by normalizing treatment groups to their respective untreated vehicle alone control.

Inhibitor treated hydrogels for immunofluorescence

For immunofluorescence of hydrogels treated with inhibitors, a similar timeline of administration was followed to that of the drug response studies. This was done to assess the cytoskeleton or baseline expression of interested markers at the start of drug treatment. Briefly, actomyosin cytoskeletal inhibitors (blebbistatin and cytochalasin-D) and TRPV4 agonist (GSK101) and antagonist (GSK205) were added 24 hrs prior to sample fixation (or on day 2). Nascent protein deposition and remodeling inhibitors (Exo-1 and TIMP3) were added shortly after encapsulations and replenished halfway through the three-day culture period before fixation on day 3. Hydrogels treated with blebbistatin and cytochalasin-D were saved for whole-gel staining. All other inhibitor treated hydrogels were sectioned for staining.

Immunofluorescence imaging of sectioned and whole-gel samples

To prepare samples for immunofluorescence imaging, samples were harvested at Day 0 or 3, fixed with 4% paraformaldehyde overnight at 4°C, and then washed twice for 15 minutes at RT while gently shaking. Fixed samples were prepared for either sectioned or whole-gel staining. For sectioned samples, fixed samples were dehydrated in Optimal Cutting Temperature (OCT, Tissue-Tek) overnight at room temperature (RT). Samples were embedded into molds surrounded by OCT solution and frozen at −80°C for at least 20 minutes. Sections of 30 µm were cut using a Leica CM3050 S Research Cryostat. For whole-gel staining, fixed hydrogels were immediately used for staining.

For expression of Nestin, CD133, TRPV4, laminin, fibronectin, and collagen-IV, sectioned samples were stained using standard immunohistochemistry protocols. Sectioned samples were washed with DI water for 5 min, permeabilized with 0.25% Triton X-100 (Sigma-Aldrich) in DPBS for 5min, and blocked with 3% bovine serum albumin, 0.1% Triton-X-100 in PBS for 1 hr at RT. The samples were then incubated overnight at 4°C with primary antibody. The following primary antibodies were used: Nestin (EMD Millipore, ABD69), CD133 (Abcam, ab19898), TRPV4 (Abcam, ab39260), Laminin (Abcam, ab11575), Fibronectin (Abcam, ab2413), Collagen-IV (Abcam, ab6586). All primary antibodies were used at 1:100 dilution. After washing 3 times with PBS, samples were incubated with secondary antibody goat anti-Rabbit IgG Alexa Fluor 488 (Invitrogen, A32731, 1:100 dilution) along with Hoechst 33342 (Thermo Fisher Scientific, H3570, 1:1500 dilution) and ActinRed 555 (Thermo Fisher Scientific, R37112) overnight at 4°C to stain for the nucleus and F-actin, respectively. Samples were washed 3 times with PBS and mounted in vectashield (Vector Laboratories). Confocal microscopy was used to acquire images using a HC PL APO 40x/1.30 oil immersion objective.

For expression of Myosin-IIa and Sox2, whole-gel staining was used. Fixed whole-gel samples were washed 3 times with PBS, permeabilized with 0.25% Triton X-100 (Sigma) in DPBS for 1hr, and blocked with 3% bovine serum albumin, 0.1% Triton-X-100 in PBS for 3 hr at RT. The hydrogels were then incubated overnight at 4°C with primary antibody while gently shaking. The following antibodies were used: Myosin-IIa (Cell Signaling Technology, 8824S, 1:100 dilution) and Sox2 (Cell Signaling Technology, 23064S, 1:400 dilution). After hydrogels were washed three times with PBS for 30 min at RT while gently shaking, samples were incubated with same secondary staining solution used for the sectioned samples overnight at 4°C while gently shaking. Samples were washed three times with PBS for 30 min at RT. Images were acquired using a Leica STELLARIS 5 confocal microscope with HC PL APO 40×/1.2 NA oil immersion objective.

Cell Proliferation

To measure cellular proliferation, Click-&-Go Plus EdU Cell Proliferation Kit (Click Chemistry Tools) was used. D-270 MG cells were encapsulated in hydrogels. On day 1, EdU was added to the culture medium (10 µM final concentration) and incubated with the hydrogels for 2 days. On day 3, cell-laden hydrogels were fixed and sectioned using the same method previously described for sample preparation for immunofluorescence. Following the manufacturer’s instructions, EdU was stained along with the nucleus on sectioned samples. Images were acquired using a Leica STELLARIS 5 confocal microscope with HC PL APO 40×/1.2 NA oil immersion objective. Custom script was used to quantify the %EdU positive cells from collected imaging data set.

TMZ diffusion

Hydrogels were loaded with TMZ (2000 µM) and placed in PBS. PBS, along with the diffused TMZ, was collected at .5, 1, 2, 3, 5, 10, 24, and 32 h timepoints. At each collection timepoint, hydrogels were placed into fresh PBS. Using UV-vis spectroscopy, absorbance of the collected PBS samples was measured with a NanoDrop One/OneC (Thermo Scientific) at 328nm for active TMZ. Absorbance values were converted into concentration based on a standard curve. The cumulative TMZ release was then calculated.

Imaging collagen network using confocal reflectance imaging

The collagen network within the different hydrogel formulations was imaged using the reflectance mode of the confocal microscopy. Samples prepared were acellular hydrogels and imaged on Day 0 after hydrogel gelation to assess the initial collagen-I distribution within the matrix. Additionally, hydrogels were imaged on Day 3 and Day 6 to assess collagen-I distribution and retention over time. Images were acquired using a 647nm laser light source along with Leica STELLARIS 5 confocal microscope with HC PL APO 63x/1.40 NA oil immersion objective.

Western Blot

For western blotting analysis, D-270 MG cell-encapsulated hydrogels were transferred to ice-cold 2X RIPA Lysis buffer (Millipore Sigma) containing 2X HaltTM Protease and Phosphatase inhibitor (ThermoFisher Scientific). For each group, western blot were performed with 3 biological replicates for quantitative analysis. For each individual replicate, 4 hydrogel samples were pooled together, with a total of 12 samples per group to ensure data robustness. The hydrogel samples were homogenized in the buffer and centrifuged, and the supernatant containing the lysate was collected. The protein amount in the lysate was quantified using PierceTM BCA Protein Assay Kit (ThermoFisher Scientific). 10μg of protein from the lysate was separated by SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes. The membranes were blocked in blocking solution (5% BSA in 1X tris-buffered saline with Tween-20, TBST (Santa Cruz Biotechnology) at room temperature for 1 h. The membranes were then incubated in primary antibodies diluted in SuperBlockTM blocking buffers (ThermoFisher Scientific) overnight at 4°C. The following primary antibodies were used at the mentioned dilutions: Nestin (EMD Millipore, ABD69, 1:25000), CD133 (Abcam, ab19898, 1: 500), Sox2 (Cell Signaling Technology, 23064S, 1:1000), GFAP (Abcam, ab7260, 1:10000), TRPV4 (Abcam, ab39260, 1:1000), and GAPDH (Cell Signaling Technology, 5174S, 1:50000). Following primary antibody incubation, the membranes were washed 3 times with TBST for 5 minutes per wash and then incubated in HRP-conjugated secondary antibodies (goat anti-rabbit, Abcam, ab6721, 1:100000 and donkey anti-mouse, Jackson ImmunoResearch, 715–035-151, 1:10000) for 1 h at room temperature. The membranes were then washed thrice in TBST for 5 minutes per wash, developed with SuperSignal West Femto Maximum Sensitivity Chemiluminescent Substrate (ThermoFisher Scientific) for 5 minutes, and imaged using Invitrogen iBright CL1500 imaging system. Quantifications of band intensities were performed using ImageJ and band intensities were normalized to GAPDH as the loading control.

Calcium imaging and analysis

For intracellular calcium imaging, D-270 MG cells in hydrogels after day 3 of culture were incubated in D-270 MG growth media with Fura red AM (33 µM, ThermoFisher Scientific) and Fluo-3 AM (20 µM, ThermoFisher Scientific) for 1 hr. The hydrogels were then washed three times with PBS. The intensities of the calcium dyes in the cells were captured using live-cell confocal microscopy with a Leica STELLARIS 5 confocal microscope with HC FLUOTAR L 25x/0.95 NA water immersion objective. Fluorescent intensities of Fura red and Fluo-3 were measured with excitation at 488 nm and detection at >610nm for Fura red and between 515–580nm for Fluo-3. Intracellular calcium concentration level was calculated as the ratio of Fluo-3 intensity to Fura red intensity. Custom MATLAB script was used for image processing and analysis.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism Software. The statistical tests performed and corresponding p and n values are specified in the figure legends.

Data Availability Statement

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