Viscoelastic stiffening of gelatin hydrogels for dynamic culture of pancreatic cancer spheroids

Abstract

The tumor microenvironment (TME) in pancreatic adenocarcinoma (PDAC) is a complex milieu of cellular and non-cellular components. Pancreatic cancer cells (PCC) and cancer-associated fibroblasts (CAF) are two major cell types in PDAC TME, whereas the non-cellular components are enriched with extracellular matrices (ECM) that contribute to high stiffness and fast stress-relaxation. Previous studies have suggested that higher matrix rigidity promoted aggressive phenotypes of tumors, including PDAC. However, the effects of dynamic viscoelastic matrix properties on cancer cell fate remain largely unexplored. The focus of this work was to understand the effects of such dynamic matrix properties on PDAC cell behaviors, particularly in the context of PCC/CAF co-culture. To this end, we engineered gelatin-norbornene (GelNB) based hydrogels with a built-in mechanism for simultaneously increasing matrix elastic modulus and viscoelasticity. Two GelNB-based macromers, namely GelNB-hydroxyphenylacetic acid (GelNB-HPA) and GelNB-boronic acid (GelNB-BA), were modularly mixed and crosslinked with 4-arm poly(ethylene glycol)-thiol (PEG4SH) to form elastic hydrogels. Treating the hybrid hydrogels with tyrosinase not only increased the elastic moduli of the gels (due to HPA dimerization) but also concurrently produced 1,2-diols that formed reversible boronic acid-diol bonding with the BA groups on GelNB-BA. We employed patient-derived CAF and a PCC cell line COLO-357 to demonstrate the effect of increasing matrix stiffness and viscoelasticity on CAF and PCC cell fate. Our results indicated that in the stiffened environment, PCC underwent epithelial-mesenchymal transition. In the co-culture PCC and CAF spheroid, CAF enhanced PCC spreading and stimulated collagen 1 production. Through mRNA-sequencing, we further showed that stiffened matrices, regardless of the degree of stress-relaxation, heightened the malignant phenotype of PDAC cells.

Graphical Abstract

1. Introduction

Pancreatic cancer, especially pancreatic ductal adenocarcinoma (PDAC), is among the leading cause of cancer-related deaths [1-3]. While accounting for only ~3% of all cancer cases, PDAC contributes to 7% of all cancer-related deaths [4, 5]. Early detection of PDAC remains a challenge as it is asymptomatic until reaching advanced stages. Additionally, only 10% to 20% of the patients have tumors that can be surgically resected [6]. PDAC is highly invasive and quick to develop chemo-resistance [7, 8]. The standard gemcitabine-based treatment has a lower than 25% response rate [9, 10], whereas almost all current combinatorial therapies are accompanied by serious side effects [11]. These complications can be attributed to the complex PDAC tumor microenvironment (TME), or desmoplasia, which contains fibrotic areas with hypovascularity that prevents chemotherapeutics from reaching the tumor [12]. In this dense stroma tissue, cancer-associated fibroblasts (CAF) facilitate cancer growth by secreting multiple pro-tumor molecules, extracellular matrices (ECM), proteolytic enzymes, and immune-suppressive cytokines [12-15]. Understanding the impact of TME on cancer initiation, invasion, and response to anti-cancer therapy is key to developing new and more effective treatments against PDAC [14-16]. However, most of the cancer therapeutics are still evaluated using two-dimensional (2D) tissue culture plastics (TCP), which expose cells to ultra-stiff planner surfaces that lack the structural complexity found in 3D PDAC desmoplasia. On the other hand, cells assembled in multicellular spheroids can mimic aspects of the 3D tumor complexity. Unfortunately, pure spheroids culture does not resemble the physicochemical structures and functions of the TME.

The mechanical properties of TME are complex and evolve over the course of tumor development [16-18]. In PDAC, the increase in stiffness heavily influences cancer malignancy and should be a critical factor to consider. In addition, most current 3D matrices lack time-dependent properties, such as dynamic stiffening and fast stress-relaxation, two important physical properties in tumor stroma. For example, Rubiano et al. reported that, as PDAC progresses, tumor stiffness increases significantly during disease progression, while the viscosity and the degree of stress-relaxation of the tumor tissue remain high regardless of the stages of the disease [19]. In tumor progression and invasion, the stiffness or rigidity of the ECM impacts intracellular signaling, gene expression, and cell behaviors. Viscoelastic or stress-relaxing matrices can facilitate cancer cell movement and invasion into adjacent tissues and can contribute to a pro-tumorigenic and chemo-resistant microenvironment [20-22]. Certain therapies, such as drugs or physical interventions, can target the tumor matrix to prevent or slow tumor progression [23]. However, few studies focus on the effects of viscoelasticity on cancer cells. In this regard, engineered matrices capable of being dynamically stiffened while maintaining fast stress-relaxation are ideal in mimicking the dynamics of TME [24]. Toward this goal, our lab and others have reported hydrogel systems with controllable mechanical properties [25-30]. For example, we have developed gelatin-norbornene (GelNB)-based hydrogels with additional pendant functional groups (e.g., hydroxyphenylacetic acid or HPA, carbohydrazide or CH) amenable for secondary crosslinking and dynamic stiffening of initially soft hydrogels [28, 31]. Of note, while prior biomimetic hydrogels can be dynamically stiffened, they did not exhibit high viscoelasticity/stress-relaxation akin to that found in the tumor tissues. We also reported synthetic and naturally derived hydrogels with tunable stress-relaxation [30]. In one example, we developed gelatin-norbornene-boronate (GelNB-BA) hydrogels with adjustable stiffness and viscoelasticity for culturing pre-osteocytes [30]. In another example, we synthesized BA-containing polymers from reverse addition-fragmentation chain-transfer (RAFT) polymerization and used the polymers to construct stiff and fast-relaxing matrices to demonstrate the importance of matrix relaxation for PDAC cells to gain invasive phenotype [32]. However, no mechanism was built into this synthetic viscoelastic hydrogel system to afford dynamic stiffening.

Recognizing the need to create a stiffening matrix with fast stress-relaxation, we report here an improved dynamic hydrogel system via combining our previously developed GelNB-HPA (Fig. 1A) and GelNB-BA (Fig. 1B) [28, 30]. The two sets of modified gelatins were modularly mixed to create covalently crosslinked hydrogels via light-initiated thiol-norbornene chemistry (Fig 1C). With the addition of Mushroom Tyrosinase (MT), HPA moieties were oxidized, forming DOPA-dimers and stiffened the cell-laden hydrogels (Fig 1D). This stiffening process mimicked aspects of the desmoplastic reaction within the TME. On the other hand, the presence of the BA groups afforded boronate-diol bonding with the DOPA-dimers newly formed through the MT-mediated stiffening reaction (Fig. 1D). As such, hydrogels crosslinked with both GelNB-BA and GelNB-HPA experienced concurrent increases of stiffness and degree of stress-relaxing owing to the additional crosslinks from DOPA-dimers and dynamic boronic ester-diol bonding, respectively. In contrast, hydrogels crosslinked with only GelNB-HPA could be dynamically stiffened without increasing the degree of stress relaxation. With this stiffened viscoelastic hydrogel system, we investigated the effect of dynamic matrix properties on growth and morphogenesis, ECM deposition, and transcriptional changes of encapsulated PDAC cells (Fig. 1E).

Figure 1. Gelatin-based stiffening viscoelastic hydrogels.

(A) Schematic of GelNB-HPA. (B) Schematic of GelNB-BA. (C) Thiol-norbornene photo-click chemistry for primary crosslinking of hydrogels. (D) MT-mediated elastic stiffening via DOPA-dimer formation and viscoelastic stiffening through boronic ester bond formation between BA and DOPA-dimer. NB, norbornene; CH, carbohydrazide; BA, boronic acid; HPA, 4-hydroxyphenylacetic acid. (E) Schematics of the experimental designs (Graphic created from Biorender.com).

2. Materials and Methods

2.1. General materials

Gelatin (Type B, 225 Bloom) was obtained from Electron Microscopy Sciences. 3-Carboxyphenol boronic acid (CPBA) was obtained from Chem-Impex Int’l Inc. 4-hydroxyphenylacetic acid (HPA) was obtained from Acros Organic. 4-(4,6-Dimethoxy-1,3,5-triazin-2- yl)-4-methyl-morpholinium chloride (DMTMM) was obtained from Combi-Blocks. Four-arm PEG–thiol (PEG4SH, 20 kDa) was obtained from Laysan Bio. The photoinitiator lithium phenyl-2,4,6- trimethylbenzoylphosphinate (LAP) was purchased from Sigma-Aldrich. Fluoraldehyde^™ o-Phthalaldehyde Crystals (OPA) was purchased from Thermo Scientific. All other chemicals were obtained from Sigma-Aldrich unless otherwise noted.

2.2. Dual functionalization of gelatin

Gelatin-norbornene (GelNB) and GelNB-HPA was synthesized following published protocols [26, 28]. The NB substitution on gelatin was controlled at ~30% of the available amine (ca. 1.6 mM NB per wt % of gelatin) in order to accommodate the secondary conjugation of BA and HPA groups. To prepare GelNB–BA, 0.5 g of GelNB was dissolved in 15 mL of ddH₂O at 40 °C. In a separate beaker, 3-CPBA and DMTMM (both at 2 mmol) were dissolved in 9 mL of DMF and 6 mL of ddH2O to activate the −COOH groups of BA for 30 minutes. The dissolved mixture was added to the GelNB solution and allowed to react for 24 h. The crude product was dialyzed for 3 days against slightly acidic ddH₂O and lyophilized. The functionalization was confirmed with ¹H NMR (Fig. S1), and the degree of BA substitution was quantified by the Fluoradehyde (OPA) assay. OPA reagent was prepared following the manufacturer’s protocol and added to modified-gelatin samples to detect the primary amine concentration, with unmodified gelatin as the control. The remaining amine concentrations on GelNB and GelNB-BA, GelNB- HPA after conjugation was calculated to estimate the degree of substitution.

2.3. GelNB-BA/GelNB-HPA hydrogels fabrication and characterization

To fabricate hydrogels, 3.5 wt% GelNB-BA along with 3.5 wt% GelNB-HPA, 1.5 wt% PEG4SH, and 2 mM LAP were mixed and exposed to longwave UV light (365 nm) at 5 mW/cm² for 2 minutes (Fig. 1A). We have shown that these photocrosslinking process causes limited cell death for pancreatic cancer cells [27-32]. Gels made with GelNB-HPA only (without GelNB-BA) were used as a control. Gelation kinetics, storage moduli (G’) and loss moduli (G”), and stress relaxation profiles were obtained with MCR 102 Anton Paar rheometer. Strain-sweep test was carried out at 0.1% to 5% strain at 1 Hz oscillation frequency. Frequency sweeps were performed between 0.1 and 10 Hz and at 10% strain. Stress relaxation tests were conducted at 10% strain for 15 minutes.

2.4. MT-mediated stiffening

Prior to MT-mediated stiffening, hydrogels were swollen in PBS for 24 hrs to wash off un-crosslinked species. To induce dynamic stiffening, hydrogels were submerged in 1 kU/ml MT for overnight. Afterwards, MT was removed via swelling hydrogels in PBS for 24 hrs. Moduli of the stiffened gels were measured using oscillation rheometry in strain-sweep mode.

2.5. CAFs and PCCs culture, spheroid formation and encapsulation

GFP-CAF (UH1303-49) was donated by Dr. Melissa Fishel of IU School of Medicine. As reported previously, patient-derived PDAC tumor tissue was minced into 1–3mm fragments, trypsinized for 30 minutes, washed in DMEM with 10% FBS, plated in a petri dish with DMEM containing 10% FBS [33]. Fibroblasts were allowed to grow out of tumor fragments for 2–3 weeks. Cells were infected and immortalized with hTERT by the Hanenberg Lab at IUPUI. Cells were authenticated by IDEXX RADIL^™ and were found to be mycoplasma free and did not genetically match any cell line in the DSMZ database [33]. COLO-357 (a metastatic PDAC cell line) was a gift from Dr. Murray Korc from IU School of Medicine. RFPCOLO-357 was generated by transducing the COLO-357 cells with red fluorescent protein using Incucyte^® Nuclight Red Lentivirus following the manufacturer’s protocol. All cell types were cultured in DMEM High Glucose media supplemented with 10% FBS and 1% Penicillin-Streptomycin. The media for RFP-COLO-357 was supplemented with puromycin to inhibit the growth of non-transfected cells. Cells were cultured in a standard 37°C, 5% CO₂ incubator. CAF and PCC were aggregated into spheroids in a PCC:CAF=1:5 ratio in 6-well AggreWell for 48 hours following the manufacturer’s protocol.

To encapsulate cell (or spheroids) in hydrogels, hydrogel precursors were mixed similarly to the steps as described in section 2.3. Then cell suspension in media was added at a desired concentration. Next, 20 μl of hydrogel precursor with cell/spheroids was added to syringe with top cut and allowed to gel under UV light for 2 minutes. After which cell-laden gels were moved to 48-well plate and submerged in media. Encapsulated cell (or spheroids) were cultured in DMEM/High Glucose media for 24 hours and then stiffened with MT as described above. Cells were cultured in the hydrogels for 4-14 days depending on experiments. CAF was encapsulated at 2 x 10⁶ cells/ml, COLO-357 was encapsulated at 4 x 10⁵ cells/ml. CAF and COLO-357 spheroids were assembled in AggreWell at 2 x 10⁶ cell/ml and 4 x 10⁵ cells/ml (i.e., 5:1 ratio of CAF to COLO-357, respectively, then used for encapsulation.

2.6. Immunofluorescent Staining and Image analysis

Cell-laden gel samples were fixed with 4% Paraformaldehyde for 1 hr at room temperature. Gels were then washed with 0.3% Triton-X 3 times (10 minutes each wash) and blocked overnight in blocking/permeabilizing buffer (PBS, 0.3% Triton-X, 1% BSA). Gels were incubated in primary antibody the next day (anti-collagen I, Cell Signaling #66948, 1:100) overnight at 4°C. The following day, gels were washed 3 times in wash buffer (PBS, 0.3% Triton-X, 1 %BSA), 30 minutes each wash, then incubated in secondary antibody (1:500, Goat anti-mouse IgG, Human ads-AF647, SouthernBiotech). Finally, gels were washed with PBS 3 times, 30 minute each and imaged on Fluoview 1000 confocal microscope. For F-actin staining, cell-laden gels were fixed with 4wt% PFA for 1 hr at room temperature, permeabilized overnight at 4°C and incubated with Acti-stain 670 Fluorescent Phalloidin (Cytoskeleton Inc., #PHDN1) overnight. Image analysis was performed using ImageJ. Spheroids diameter, cell area and circularity were determined in ImageJ using the ‘Analyze Particles’ function with the measurement setting included ‘Area’ and ‘Shape descriptors. Circularity was calculated by ImageJ following the formula:

$$Circularity=4\pi (\frac{Area}{{Perimeter}^2})$$

A circularity value of 1.0 indicates a perfect circle. As the value approaches 0.0, it indicates an increasingly elongated cell.

Live cell imaging of encapsulated co-culture spheroids was done on BioTek Lionheart FX Auntomated Microscope at 4X magnification. Images were captured every 30 minutes for 20 hours.

2.7. Viability and cytotoxicity staining and assay

GFP-CAF cell-laden hydrogels were incubated in Ethidium Homodimer-2 staining solution (Biotium, #40050) for 1 h at room temperature to visualize cells with damaged cell membranes. Cell viability was obtained by counting GFP-CAF and dead cells. The live cell number was divided by total cell count to obtain initial cell viability. For RFP-COLO-357, Cell apoptosis was assessed by caspase 3/7 activity. Caspase staining was done using CellEvent Caspase-3/7 green detection reagent (Invitrogen, C10423) following manufacturer’s instruction. To obtain quantitative result for comparison between gel conditions, 100 mL of Caspase-Glo^® reagent (Promega) mixed with culture media at 1:1 ratio was added to the cell-laden gels and incubated for 1 h. The luminescence was measured by a microplate reader (SpectraMax^® i5).

2.8. EdU cell proliferation staining and analysis

To identify proliferating cells, 10 μM of EdU reagent (Click-IT^® EdU staining kit, Invitrogen, C10337) was added into the culture media, followed by incubation at 37C for 4h. The cell-laden gels were collected and rinsed with PBS twice, fixed with 4% paraformaldehyde solution for 20 min, and permeabilized and blocked with 0.5% Triton X-100, 3% BSA mixture for overnight. The gels were washed with PBS twice and immersed in EdU reaction cocktail prepared following manufacturer's instruction. The gels were counter-stained with DAPI for 1h and the stained gels were imaged by confocal microscopy. For RFP-COLO-357, AlexaFluor 488-azide was used; for GFP-CAF, AlexaFluor 647-azide was used for EdU visualization. For COLO-357, the number of EdU-positive and total cell nuclei within each spheroid were counted. Percentages of Edu-positive cells were determined by dividing the number of Edu-positive cells by the total number of nuclei. For each gel condition, at least 12 spheroids were counted.

For CAF, EdU-positive cells were counted and divided by the total number of cells in an image. For each gel condition, at least 3 images were used for counting with at least 30 cells per image.

2.9. Western Blotting

Cells were lysed in 100 μl cold RIPA lysis buffer (Thermo Scientific, 89901) supplemented with protease inhibitor cocktail (Thermo Scientific, 78443) for Western blot analysis. Total proteins were separated by 4–15% Mini-PROTEAN^® TGX^™ Precast Gel (Biorad, #4561084DC) and transferred to PVDF membranes using a Trans-Blot Turbo Transfer System with Trans-Blot Turbo Mini PVDF Transfer Packs (Biorad, #1704157EDU). The membrane was blocked for an hour at room temperature using blocking buffer (5% non-fat dried milk in 1X TBST). Subsequently, blots were probed with primary antibodies as follows: α-Smooth Muscle Actin (#19245), PDGF Receptor β (#3169), Vimentin (#5741), FAP (#66562), S100A4 (#13018), E-cadherin (#3195), and GAPDH (#97166T) (all were diluted at 1:1000, Cell Signaling Technology). After incubation with appropriate Europium-labeled secondary antibodies (Molecular Devices, ScanLater Eu-Goat anti-mouse IgG, R8208 or ScanLater Eu-Goat anti-Rabbit IgG, R8209), the membrane was scanned using the SpectraMax^® i5 Multi-Mode Detection Platform and analyzed with SoftMax Pro 7.1 software (Molecular Devices).

2.10. Collage 1 detection

Fluorescent intensity from Collagen 1 staining was determined using ImageJ with ‘area integrated intensity’ and ‘mean grey value’ functions selected in ‘Set measurements’ tab. Corrected Mean Fluorescence = Mean Fluorescence of Region of Interest - Mean Fluoresence of Background. The expression of COL1A1 was assessed by quantitative real-time PCR (qRT-PCR). To extract RNA from encapsulated cells and spheroids, gels were first degraded with 100 U/ml collagenase at 37°C. RNA was extracted and process using Takara Bio RNA isolation kit (Nucleospin RNA Catlog. #740955.50) according to the manufacturer’s instructions. Eluted RNA was stored in −80°C. Isolated RNA were converted into cDnA using Takara PrimeScript RT reagent kit (#RR037A). Quantitative real-time (RT) PCR was performed using SYBR Premix Ex Taq II kit. The relative quantity of COL1A1 was normalized to the GAPDH internal control and evaluated by using 2 method. Forward and reverse primers are listed in Table S1.

2.11. RNA extraction and mRNA-sequencing

Total RNA was isolated from cells after 6 days of culturing. For spheroids grown in Aggrewell plate, the spheroids were dislodged and retrieved from the microwells by gentle pipetting. For spheroids encapsulated within hydrogels, cells were retrieved by degrading the hydrogels with 100 U/ml collagenase. All RNA samples were collected and process using Takara Bio RNA isolation kit (Nucleospin RNA Catlog. #740955.50) according to the manufacturer’s instructions. Eluted RNA was stored in −80°C.

mRNA sequencing was carried out in Center for Medical Genomics at Indiana University. RNA sequence Library preparation was performed with KAPA RNA Hyperprep Kit (KK8581). Library was sequenced on Illumina NavaSeq 6000 instrument in Pair-End (PE) configuration. mRNA sequencing data processing and bioinformatics analysis (differential expressed gene (DEG) and gene set enrichment analysis (GSEA) were performed by the Center for Medical Genomics at Indiana University. Quality of the raw RNAseq data was verified with FastQC. Rawfastq file were mapped to the human genome reference (USCS hg38) with STAR aligner software. Read counts per gene were calculated using featureCounts function against the reference genome. Differential expression analysis was performed using DESeq2 R package. Gene set functional enrichment was performed using two methods: 1) pre-ranked GSEA using Molecular Signatures Database (MsigDB), 2) gene set overrepresentation analysis (ORA) via WEB-based Gene Set Analysis Toolkit (WebGestalt) using ‘illumine humanht 12 v4’ reference set. KEGG pathway analysis was carried out via Database for Annotation, Visualization and Integrated Discovery (DAVID) Knowledgebase.

2.12. Statistical Analysis

Student’s t-tests (for two-group experiments) or one-way ANOVA were performed for all experiments using Prism 9 software to evaluate statistical significance. Single, double, triple and quadruple asterisks represent p <0.05, 0.01, 0.001 and 0.0001, respectively.

3. Results and Discussion

3.1. Hydrogel crosslinking, stiffening, and rheological characterization

We have previously shown that either GelNB-HPA or GelNB-BA could be readily crosslinked with PEG4SH into elastic hydrogel via light-mediated thiol-norbornene photochemistry (Fig. 1A-1C) [28, 30]. Through in situ photo-rheometry, we demonstrated that both GelNB-HPA (7 wt%) or a mixture of GelNB-HPA (3.5 wt%) and GelNB-BA (3.5 wt%) reached gel points in under 15s after initiating light exposure (Fig. 2A). Furthermore, the inclusion of GelNB-BA did not interfere with the thiol-norbornene gelation as all gels with the same gelatin macromer contents reached similar plateau moduli (G’ = 3 ± 0.8 kPa. Fig. 2A). The thiol-norbornene hydrogels were further incubated in MT (final concentration: 1 kU/ml) to induce dynamic stiffening. The hydrogels shrunk in size and became dark brown in color (Fig. 2B). This result was similar to that reported in the literature and suggested that the HPA motifs were oxidized by MT into DOPA and DOPA-dimers [27, 28]. In addition to the color change, MT treatment led to around 3-fold increases in gel elastic moduli (from ~2 kPa to ~6 kPa. Fig. 2C). The increased stiffness (G’) in GelNB-HPA-only gel was attributed to the MT-mediated DOPA-dimer formation; and the stiffening of GelNB-HPA/GelNB-BA gel resulted from both DOPA-dimer formation and BA-diol conjugation (Fig. 1D). This result also implied that the incorporation of BA moieties did not interfere with MT-mediated stiffening. Importantly, the degree of stiffening (G’ ~2kPa to ~6 kPa) was equivalent between the two groups of gels and was similar to the values reported for normal and cancerous pancreatic tissues [34-36].

Figure 2. Rheological characterization of GelNB-HPA/GelNB-BA hydrogels.

(A) In situ rheometry of the hydrogels crosslinked by PEG4SH using 2 mM LAP as the photo-initiator and 365 nm light exposure at 5 mW/cm². (B) Photograph of GelNB-BA/GelNB-HPA hydrogels before and after MT-mediated stiffening overnight. (C) & (D) Storage modulus G’ and Tan (d) of +/− BA hydrogels before and after MT treatment. (E-F) Effect of MT treatment on stress relaxation of hydrogels crosslinked in the absence (E) and presence (F) of GelNB-BA. Hydrogels were crosslinked from 7 wt% of GelNB-HPA only or a mixture of GelNB-HPA and GelNB-BA (each at 3.5 wt%). PEG4SH was added at 1.5 wt% (R_thiol/nor =0.75). Hydrogels were stiffened by incubation in 1 kU/ml MT for overnight.

After asserting that hybrid GelNB-HPA/GelNB-BA hydrogels could be stiffened via MT-mediated DOPA-dimer formation, we further characterized other rheological property of these dynamic hydrogels. Although both hydrogel groups were stiffened to similar G’, the loss moduli (G") and hence the Tan(δ) values (i.e., G”/G’) were substantially higher when GelNB-BA was incorporated in the hydrogels. Specifically, Tan(δ) values remained low (<0.01) for GelNB-HPA hydrogels both before and after MT-mediated stiffening, indicating that the hydrogels remained elastic (Fig. 2D). On the other hand, Tan(δ) values increased one order of magnitude for GelNB-HPA/GelNB-BA hydrogels after MT-mediated stiffening (from 0.006 to 0.055) (Fig. 2D). We reasoned that the newly emerged 1,2-diols from MT-mediated HPA oxidation (Fig. 1D) formed reversible BA-diol complexes, allowing stress-induced dissipation of energy, and contributed to the increase of Tan(δ). The introduction of two extra types of secondary crosslinking reactions, namely MT-mediated HPA oxidation and BA-diol dynamic bonding, led to a noticeable increase in stiffness and relaxation property. To assure that the stiffening process increased hydrogels viscoelasticity, we conducted time-dependent relaxation tests for hydrogels without (Fig. 2E) or with (Fig. 2F) GelNB-BA. In the absence of GelNB-BA, MT treatment did not change the stress relaxation rate, with relaxation halftimes (τ_1/2) for both groups approaching infinity. On the other hand, stiffening of GelNB-HPA/GelNB-BA hydrogels led to a significant decrease in relaxation modulus (G/G₀), reaching 60% of the initial modulus in less than 1,000s (stress relaxation rate τ_1/2 = 6,000 s after stiffening vs τ_1/2 = infinity before stiffening, Fig. 2F). Additionally, frequency sweeps also revealed the presence of BA group in the hydrogels became more frequency-dependent after stiffening, a characteristic of viscoelastic materials (Fig. S2A and S2B). At low frequency, the BA-containing stiffened hydrogels exhibited high G” (G” = 400 ± 4 Pa) at 1 Hz, and rapidly decreased to G” = 20 ± 7 Pa at high frequency, which indicated +BA gels behave more viscously over long timescales and elastically over short timescales. This property was not observed before stiffening. In the GelNB-HPA only hydrogels, the G” remained low (G” = 40 ± 5 Pa) before and after stiffening. The value of G” did not fluctuate significantly within the range of 1 Hz and 10 Hz frequency, indicating that within this range, GelNB-HPA gels are more frequency-independent, behaving elastically regardless of timescale magnitude [37]. Typically, hydrogel stiffening only increases storage modulus G’ and not loss modulus G” [38], meaning stiffening would decrease tan (δ) and may result in a loss of gel viscoelasticity and stress-relaxation. By leveraging the diol and BA functional groups within this hydrogel system, hydrogels were stiffened while the newly formed boronate-diol bonds would increase hydrogel viscoelasticity.

3.2. Cell encapsulation and dynamic stiffening of cell-laden hydrogels

Patient-derived pancreatic CAF and a PCC line COLO-357 were used to test the cytocompatibility of the dynamic GelNB-HPA/GelNB-BA hydrogel system. For longitudinal tracking of cell morphology, CAFs and PCCs were separately tagged with green fluorescent protein (GFP) and red fluorescent protein (RFP), respectively. GFP-CAF and RFP-COLO-357 cells were encapsulated in hydrogels crosslinked by GelNB-HPA or a mixture of GelNB-HPA and GelNB-BA, with PEG4SH as the thiol-bearing crosslinker for both cases. To induce dynamic stiffening, the cell-laden hydrogels were incubated in MT (1 kU/ml) overnight. As shown in Fig. 2, only hydrogels containing both GelNB-HPA and GelNB-BA were stiffened with concurrent increase in viscoelasticity (i.e., Stiffened VE), whereas hydrogels containing only GelNB-HPA were stiffened without notable increase in relaxation (i.e., Stiffened/EL). Finally, untreated GelNB-BA containing gels were termed as ‘Soft’ gels. Fig. S3 and 3A and B showed that all hydrogels (Soft, Stiffened/EL, and Stiffened/VE) were highly cytocompatible for in situ encapsulation of PCCs and CAFs as cells were able to grow and spread in the hydrogels. Caspase 3/7 staining indicated minimal apoptosis of COLO-357 cells encapsulated in all three groups of hydrogels (Fig. S4A). Quantitative caspase-3/7 activity assay also showed no significance difference in apoptosis among the three gel conditions (Fig. S4B). Similarly, a cytotoxicity staining with ethidium homodimer-2 showed that CAF obtained high viability across all gel conditions (Fig. S4 C and D). These results demonstrated that the encapsulation process (UV-light exposure at 365 nm at 5 mW/cm² for 2 minutes) and MT treatment (at 1 kU/ml) were not harmful for cells, consistent with previous reports [27, 38]. In term of cellular morphology, PCCs grew into multicellular spheroids over 7 days, whereas CAFs spread out as single cells (Fig. 3A). Measurements of PCC spheroid diameters indicated that the growth in Soft hydrogels was the least restricted, with an average diameter of 68.0 ± 21 μm, followed by Stiffened/VE (61.4 ± 15 μm) and Stiffened/EL (50.8 ± 15 μm) (Fig. 3C). The sizes of spheroids in Stiffened/VE gels were slightly bigger than those in Stiffened/EL gels, implying that even in a stiffened matrix, high viscoelasticity was more permissive to cell proliferation in 3D [30, 32]. Western Blot results showed that stiffened hydrogels (i.e., Stiffened/EL and Stiffened/VE) led to decreased expression of epithelial marker E-cadherin and upregulation of mesenchymal marker Vimentin in the encapsulated COLO-357 cells (Fig. 3E). This was consistent with results demonstrated by Rice et al. [34], where BxPC-3 pancreatic cancer cells responded to increased substrate rigidity by increasing Vimentin expression [34].

Figure 3. Encapsulation of PCC and CAF in hydrogels.

(A) PCC cell line (RFP-COLO-357) and (B) GFP-CAF (UH 1303-49) encapsulation in GelNB-HPA (+/− BA) hydrogels. Soft, untreated with MT; Stiffened, treated with MT. (C) Size analysis of the cancer spheroids at D7. (D) Shape analysis, cell area and circularity of the encapsulated CAF on D4. (One-way ANOVA, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001). (E) Western Blot of EMT markers in encapsulated RFP-COLO-357 and (F) CAF-associated proteins in encapsulated GFP-CAF. (G) Effect of MT treatment on growth of PDAC cells (dispersed CAF & PCC) in hydrogels crosslinked in the absence and presence of GelNB-BA (i.e., Stiffened/EL and Stiffened/VE). MT-mediated stiffening was initiated on Day 1, or 24 hr post-encapsulation.

Next, we evaluated the effect of dynamic stiffening on CAF phenotype. The encapsulated CAF did not form aggregates but exhibited fibroblastic phenotype (Fig. 3B). Shape and size analyses of CAFs in the hydrogels showed that dynamic stiffening resulted in reduction of cell area and a slightly increase in circularity (Fig. 3D), but the degree of spreading was similar between the two stiffened groups. Western Blot results show that dynamic hydrogel stiffening led to reduction of CAF markers PDGF-β, FAP, and α-SMA (Fig. 3F). This result agreed with a prior study by Cao et al. where CAF encapsulated in softer, more compliant hydrogel network demonstrated higher cell spreading and upregulation of CAF hallmark transcripts (e.g., α-SMA) [39, 40]. It stands to reason that the stiffer matrix provided more restrictive environment and limit CAF spreading and activation [39]. The effect of matrix stiffening on cell fate was further assessed by Click-iT EdU, which labels duplicating DNA in the proliferating cells (Fig. S5). We found that both PCCs and CAFs proliferated within the three groups of hydrogels (i.e., Soft, Stiffened/EL, Stiffened/VE). Due to the heterogeneity nature of the cancer spheroids, the percentage of EdU positive cells in the spheroids varied largely from 16% to 52%, 20% to 65%, and 19% to 70% for spheroids in soft, stiffened/EL, and stittened/VE hydrogels, respectively. Due to the high variation, no statistical significance was found between the three hydrogel groups. The percentage of individual EdU positive CAFs were lower, ranging from 7% to 25% in different images and no statistical significance was found between all groups. This result seemed to be in line with a prior study where PANC-1 and AsPC-1 cells were repopulated in soft decellularized healthy pancreas ECM (~ 2 kPa) and stiff PDAC ECM (~6 kPa) and no statistical significance was found in the percentages of proliferating cells [41].

When PCCs and CAFs were co-encapsulated as single cells, both cell types exhibited morphology similar to when they were encapsulated separately even after 14 days of culture (Fig. 3G). This observation signified that the co-encapsulation of CAFs and PCCs within the same gel did not result in substantial alterations to their morphology or functional behaviors. It can be reasoned that the dispersion of single CAFs and PCCs in the gels did not provide sufficient proximity to allow intercellular interactions between the two cell types.

3.3. Co-encapsulation of CAF and PCC in dynamic stiffening hydrogels

Since CAFs and PCCs co-encapsulated as dispersed single cells did not form strong intercellular interactions even after prolonged culture (Fig. 3G), we generated mixed cell spheroids to provide direct cell-cell contact and to elucidate the impact of matrix stiffening on PDAC cell fate. Spheroids of GFP-CAF and RFP-PCC were generated (at a ratio of 5 to 1 [42]) in an AggreWell plate for 2 days, followed by encapsulation in the hydrogels and dynamic stiffening 24 hr later (i.e., on day 3). After culturing for 3 additional days, cell-laden hydrogels were processed for cellular and molecular analyses (Fig. 4A). Due to the adherent nature and larger number of cells, CAFs aggregated in the core, with PCCs lining on the periphery of the spheroids (Fig. 4B). After encapsulating in hydrogels, the mixed spheroids remained spherical with CAF at the center and COLO-357 at the periphery. Interestingly, after 24h, CAFs started to interact with the gelatin-based matrix and migrated out of the spheroids (Fig. 4C), while some PCC clustered closer to each other. Within 24h, significant CAF and PCC spreading out from the initially spherical spheroids into the matrix was notable from live cell tracking (Video S1-S3). Stiffening of the spheroid-laden hydrogels with MT was done on Day 3 (Fig. 4A). Day 4 images revealed that significant cell spreading of CAF was seen in the non-stiffened soft hydrogel control group, while cells in both stiffened hydrogels exhibited less spreading, indicating that the encapsulated CAF were able to remodel the gelatin matrix more easily in the soft condition (Fig. S6). On day 6, however, the spreading of both CAF was seen to the same extent in all hydrogels (Fig 4D, S6). It is worth noting that, at this point, all CAFs had migrated to the border of the spheroids, while PCCs situated as cell clusters at the core, completely contrasting to the initially formed spheroids. More interestingly, the PCC within the co-culture spheroids obtained an invasive morphology in all groups, regardless of stiffness (Fig. 4D). The PCCs became more elongated at the periphery of the spheroids and seemed to follow the path that was directed by the surrounding CAFs. This type of mesenchymal morphology was not observed when CAFs and PCCs were dispersedly co-encapsulated in the hydrogels (Fig. 3G), indicating that close contact between CAFs and PCCs was critical for the two cell types to interact, leading to mesenchymal phenotype of PCCs. The spread area of PCCs in the co-culture spheroid has no significant difference across all group (Fig. S7), suggesting that the presence of CAFs likely ‘overwrote’ the effects of higher stiffness and stress relaxation on promoting growth of PCCs as that seen in the monoculture (Fig.3A). It is also worth noting that no spreading morphology was observed in co-culture spheroids maintained in AggreWell for 1 week (Fig. S8). This demonstrated the importance of matrix in supporting cell adhesion, spreading and migration.

Figure 4. Multicellular PCC/CAF spheroids generation and encapsulation.

(A) Timeline for spheroid formation, encapsulation, stiffening of cell-laden hydrogels, and cellular analysis. (B) & (C) CAF and PCC hetero-species spheroids within AggreWell^™400 plate and hydrogels, respectively. (D) Representative images of encapsulated CAF/PCC spheroids in soft and dynamically stiffened hydrogels and threshold images of red channel demonstrating PCC spreading in soft and dynamically stiffened hydrogels. Spheroids were allowed to grow in soft hydrogels for 1 day prior to stiffening. Scale: 100 μm. (E) Collagen 1 staining in gels with CAF only, PCC only, or CAF/PCC spheroid co-encapsulation. Images were taken 4 days after encapsulation. (F) and (G) qRT-PCR for COL1A1 gene expression of encapsulated CAF, PCC and Co-Culture Spheroids.

Next, we investigated how changes in stiffness and viscoelasticity affected other cellular functions, specifically the secretion of type 1 collagen, the primary matrix protein secreted by activated CAF that contributes to a stiffening TME. Indeed, collagen 1 was detected sparsely in monoculture of CAFs, but not at all in monoculture of PCCs (Fig. 4E, S9 A and B). Furthermore, in co-culture of CAF-COLO spheroids, collagen deposition increased and expressed similarly across all gel conditions (Fig. 4E, S9). Quantification of the fluorescent intensity from immuno-staining images showed higher collagen 1 production in co-culture comparing to mono-culture CAF, although not statistically significant (Fig. S9 D). We also assessed COL1A1 mRNA expression by qRT-PCR. Interestingly, matrix stiffening downregulated COL1A1 mRNA expression in mono-culture CAF, especially in Stiffened/VE gels (Fig. 4F). This was in line with the reduced CAF hallmark protein expression in the Western Blot results (Fig. 3F). Not surprisingly, no COL1A1 was detected in mono-culture of PCCs. However, COLO1A1 expression was restored in the PCC/CAF spheroids in both stiffened hydrogels (Fig. 4G), suggesting that intimate interactions between PCCs and CAFs restored CAF functions that were otherwise suppressed in the stiffened matrices. These results further affirmed that immediate contact between PCCs and CAFs guided PDAC cell fate more than dynamic matrix stiffening and viscoelasticity. It is worth noting that the presence of two cell types in the co-culture spheroids hindered the direct comparison of COL1A1 expression in CAF between mono- and co-culture. This limitation arises from the fact that COL1A1 was normalized to GAPDH expressed in both CAFs and PCCs, while in mono-culture COL1A1 was normalized to GAPDH expressed by CAFs only. To definitively determine whether collagen 1 production was upregulated in co-culture CAF compared to mono-culture CAF, cell sorting separating CAF from COLO-357 should be performed prior to RNA extraction.

Many reports in the past have shown that CAFs could promote the migration of cancer cells indirectly with the use of CAF-conditioned media [43, 44]. In a study done by Ermis et al., cancer-CAF multicellular spheroids cultured in stiff gelatin/hyaluronic acid gels displayed higher expression of markers associated with EMT, mechano-transduction, and proliferation compared to cancer-only spheroids [45]. In another case, a mixture of PDAC cells, pancreatic stellate cells (PSC) and macrophages were cultured together in a peptide amphiphiles (PA) and customized ECM gel [46]. Researchers found that comparing to mono-culture of PDAC cells, the inclusion of PSC and macrophages promoted chemo-resistance in cancer cells, decreasing their apoptosis rate [46]. However, in both studies mentioned above, PDAC cells in the co- and tri-culture settings formed duct-like colonies and not elongated/spindle-shaped morphology observed in the current study. The unique protrusion and migration of PCC from spheroids described in Fig. 4 and 5 were also not observed in the prior studies. We reason that the intimate and direct cell-cell interactions between the stromal and cancer cells as found in the native tumor tissues likely played a more dominating role than matrix stiffening and viscoelasticity.

Figure 5. Transcriptomic comparison of CAF+PCC spheroids.

(A) PCA analysis of CAF+PCC spheroids cultured in AggreWell (naked spheroids, NS) and in Soft gels. (B) Volcano plot for transcriptomic comparison between NS and those in Soft gels; determined by DEseq2 method, DEGs cutoff conditions are FDR < 0.05, ∣Fold Change∣>=1.3. (C) PCA analysis of CAF+PCC spheroids in Soft, Stiffened/EL and Stiffened/VE gels. (D) Volcano plot for transcriptomic comparison between spheroids in Stiffened/EL gels comparing to those in Soft gels; determined by DEseq2 method, DEGs cutoff conditions are FDR < 0.05, ∣Fold Change∣>=1.3.

3.4. mRNA-seq of naked and encapsulated CAF/PCC spheroids

To further elucidate the impact of dynamic matrix mechanics on PDAC cells, we conducted mRNA sequencing (mRNA-seq) to reveal the transcriptional changes associated with increasing matrix stiffness and viscoelasticity. Here, spheroids encapsulated in Soft, Stiffened/EL and Stiffened/VE hydrogels were isolated and extracted for total RNA. Naked spheroids cultured in AggreWell were included as an additional control. After obtaining the mRNA sequencing results, we first conducted bioinformatic analyses on naked and encapsulated spheroids. The principle component analysis (PCA) showed that naked spheroids (NS) had distinct transcriptomic profiles from the encapsulated spheroids (Fig. 5A). Differentially expressed genes (DEG) analysis was performed, with DEG defined as genes with adjusted-p (aka false discovery rate, FDR) smaller than 0.05 and absolute fold-change greater than or equal to 1.3. The results showed drastic differences between spheroids encapsulated in 3D matrix and those that were simply cultured in AggreWell plate. Specifically, 6,143 DEGs (3154 genes were upregulated, 2989 genes were down regulated) were identified between the unencapsulated spheroids and spheroids encapsulated in the soft gels (Fig. 5B). The PCA plot in Fig. 5C confirmed that Soft samples displayed a clear separation from Stiffened (both EL and VE) samples. Between Soft and Stiffened/EL groups, 341 DEGs were identified, with 165 genes upregulated, and 176 genes down regulated in the Stiffened/EL condition (Fig. 5D).

While hydrogel encapsulation led to drastic changes in DEG, surprisingly, no significant difference was detected between spheroids encapsulated in the Stiffened/EL and Stiffened/VE groups (Fig. S10). It was likely that the viscoelasticity/relaxation of the Stiffened/VE gels was not fast enough to affect gene expression in the CAF/PCC mixed spheroids. Adebowale et al. showed that stress relaxation substrate enhanced fibrosarcoma and breast cancer cell adhesion and migration, at a relaxation half time of ~200 s [47] and only minimal differences were found when comparing cell motility on elastic and slow relaxing substrates (relaxation half time = ~2,000 s) [47]. Another explanation for the similarity between cells in Stiffened/EL and Stiffened/VE gels might be that, with CAF remodeling the surrounding environment, the majority of cancer cells did not experience changes in matrix viscoelastic property, which ultimately led to an unaltered transcriptomic profile. Future experiments will focus on increase the relaxation half time of the hydrogels to roughly 100 s [19] to observe how viscoelasticity affect CAFs interaction with and remodeling the matrix. The mechanism by which CAFs reshaped PCCs morphology and potentially phenotype will also require further investigation.

Next, we searched for cellular processes that were affected by cell-matrix interactions using over-representation analysis (ORA), which identifies biological processes that are implicated in the DEGs. Using ORA, we discovered that encapsulated spheroids vs naked spheroids induced DEGs mostly related to cell proliferation (Fig. S11A), extracellular matrix, extracellular structure, and skeletal development (Fig. S11B). This was reasonable as encapsulated spheroids were surrounded by the gelatin matrix and were able to adhere to the cell adhesive ligands (e.g., RGD). As no matrix was provided to the naked spheroids, cells could only interact with adjacent cells.

ORA was also performed on DEGs between Soft and Stiffened/EL groups. Fig. 6A shows biological processes that were associated with genes up-regulated in the Stiffened/EL gels. Most of the identified pathways involved cell cycle and DNA replication. Although EdU staining of encapsulated mono-culture PCC and CAF did not detect significant difference in proliferation (Fig. S5), in co-culture, cells in Stiffened matrix were more prone to proliferate than in the soft hydrogels as suggested by RNA-sequencing data. Notable genes were cell cycle genes (CDC5, CDC20, CDC25, etc.) and replication factor genes (RCF2, RCF3, RCF4, etc.). On the other hand, genes up-regulated in the Soft gels were mostly involved in interferon and immune responses (Fig. 6B). Fig. 6B also reveals that cells in Stiffened/EL gels had reduced IFN- and defense-response compared to those in the soft gels. It can be speculated that cells in the Stiffened environment produced more growth factors and chemokines that created a more immunosuppressive environment [48-51]. Tumors expressing high levels of IFN-alpha and gamma had been reported to response well to immunotherapy-based treatment due to higher levels of IFN-stimulated genes that sustain antitumor immune response [48-52]. In this light, reducing tumor microenvironment stiffness can potentially induce more robust IFN response that can help increase the effectiveness of cancer treatment.

Figure 6 – Enrichment plots of differentially expressed gene in Stiffened/EL versus Soft group.

(A) ORA of Biological Process GO term of genes that up-regulated in Stiffened/EL gels. (B) ORA of Biological Process GO term of genes that up-regulated in Soft gels. (C&D) Pre-ranked GSEA analysis of DEG, the plot’s black vertical lines represent the ranked gene hits in the gene set. The green line represents the normalized enrichment score (NES). (C) NES>0, the gene set is enriched in Stiffened/EL group. (D) NES<0, the gene set is enriched in Soft group.

Next, Gene Set Enrichment Analysis (GSEA) was conducted to identify pathways that were affected by the physical environment surrounding cells. This approach was used to identify significant pathways where the differences in gene expression between groups are small but consistent [53]. For further in silico analysis, we focused only on cells encapsulated in Soft vs Stiffened/EL gels as there was no significant difference in DEG between Stiffened/EL and Stiffened/VE gels. We found that several gene sets were significantly enriched in Stiffened/EL gels, including DNA repair, EMT, MTORC1, E2F, MYC Targets, Miotic spindle, G2M checkpoints (Fig. 6C). These hallmark pathways correlate with a higher proliferative and more aggressive cancer phenotype [54]. On the other hand, IFN response was enriched in Soft gels, corroborating the previous ORA results. KRAS signaling, which activates multiple intracellular pathways that affect proliferation, cell migration, immune evasion and apoptosis [55], was down in Soft gels while the TNFA signaling pathway via Nuclear factor-κB (NF-κB) was upregulated in Stiffened/EL gels (Fig. 6C & D). The overexpression of NF-κB has been linked to KRAS activation, driving tumor initiation and progression in PDAC [56]. The two pathways acted in concert to suggest that the cells in Soft gels were not as aggressive as those in Stiffened/EL gels. Moreover, we also identified doxorubicin drug resistance genes and putative targets of EZH2, were also enriched in spheroids in stiffened environment, which imply these cells would be less sensitive to anti-cancer drugs (Fig. S12) [57]. The high expression of EZH2 has been reported to activate cell survival pathway as well as promote drug resistant in several cancer, including pancreatic cancer [57-59].

Lastly, via KEGG pathway analysis, we identified the differentially expressed genes between Soft and Stiffened EL gels included genes encoding motor proteins such as KIF14, KIF23, KIF15, KIF20, KIF18A, KIF18B, KIF2/24, MYO1, MYO7B, MYO15, MYH and genes involved in the activation of integrin subunit alpha (Table S2 and Fig. S13). Based on KEGG pathway analysis, it is likely that the differences in cell fate processes observed between Soft and Stiffened/EL matrices were mediated by integrin subunit alpha as well as actin and microtubule motors, which are proteins responsible for cell contraction, motility, mitosis. More specifically, the stiffened matrix altered cellular contractility or cytoskeletal tension, which in turn activates integrins that ultimately mature into focal adhesion complexes. These events typically induce a cascade of signaling pathways that leads to the polymerization and reorganization of the actin cytoskeleton/microtubule cytoskeleton, and ultimately dictates cell morphology, movement, and cell fate [60]. However, further experiments are needed to confirm this hypothesis.

In this studies, we utilized COLO-357, a PDAC cell line derived from metastatic tumor, as well as patient-derived CAF. These two cell types represent the major PDAC cells in the later stage of tumor development where substantial desmoplastic reactions have occurred. These cells were different from pancreatic cancer cells (e.g., BxPC-3) that display more epithelial phenotype and pancreatic stellate cells (PSC) that were inactivated stromal cells. Depending on the context of the study, PSCs may be used to probe the events that lead to their activation into CAFs.

Conclusion

In summary, hydrogel stiffening typically results in a decrease in gel viscoelasticity due to an increase in elastic modulus G’ but no change in loss modulus G”. Here, we present an engineered GelNB-based hydrogel platform with a built-in mechanism for simultaneously increasing matrix elastic modulus and viscoelasticity. Our study show that stiffness changes played a significant role in regulating cellular functions in 3D culture, inducing many critical pathways regulating cancer progression and metastasis. In addition, compared to mono-PCC culture, the incorporation of heterogenous CAF and PCC spheroids promoted cancer cell invasion into the matrix and collagen production. The mRNA-sequencing results provided additional insights into the effect of matrix stiffening on the development of tumorigenic phenotype marked by heightened EMT, proliferation, drug resistance, and reduced anti-inflammation characteristics. While we did not detect significant difference in DEG between stiffened matrix with different viscoelasticity, whether matrix stress-relaxation affects transcriptional activity in cancer cells remained unclear. Ongoing efforts include formulating the hydrogels to exhibit PDAC-mimetic fast stress relaxation (relaxation halftime <100 sec) and matrix with other stromal components (e.g., hyaluronic acid). The results from this study underscore the need for a dynamic 3D hydrogel platform where changes in ECM mechanical properties are recapitulated and provide a strategy for stiffening incorporating viscoelastic.

Statement of Significance:

The pancreatic cancer microenvironment is a complex milieu composed of various cell types and extracellular matrices. It has been suggested that stiffer matrices could promote aggressive behavior in pancreatic cancer, but the effect of dynamic stiffening and matrix stress-relaxation on cancer cell fate remains largely undefined. This study aimed to explore the impact of dynamic changes in matrix viscoelasticity on pancreatic ductal adenocarcinoma (PDAC) cell behavior by developing a hydrogel system capable of simultaneously increasing stiffness and stress-relaxation on demand. This is achieved by crosslinking two gelatin-based macromers through orthogonal thiol-norbornene photochemistry and post-gelation stiffening with mushroom tyrosinase. The results revealed that higher matrix stiffness, regardless of the degree of stress relaxation, exacerbated the malignant characteristics of PDAC cells.