Visible light-responsive hydrogels for cellular dynamics and spatiotemporal viscoelastic regulation

Hydrogels with both intrinsic and visible light-responsive viscoelasticity

We designed three crosslinkers containing the TDS moiety, each with a distinct aldehyde chemical structure (Fig. 1), for this study: 4,4’-(disulfanne-1,2-dicarbonothioyl)bis(1-(3-formyl-4-hydroxybenzyl)−1-methylpiperazin-1-ium) (ALYDS), 4,4’-(disulfanne-1,2-dicarbonothioyl)bis(1-(2-formyl-3-oxopropyl)−1-methylpiperazin-1-ium) (MDADS) and 4,4’-(disulfanne-1,2-dicarbonothioyl)bis(1-(4-formylbenzyl)−1-methylpiperazin-1-ium) (PBEDS) as detailed in supporting information. The selection of these aldehydes was based on their documented kinetic rates for imine formation and hydrolysis, which are known to influence the hydrogel’s stress relaxation^19,20. Hydrogels were synthesized by mixing a crosslinker with CMCS under ambient conditions (room temperature, pH 7.4), forming dynamic imine/Schiff-base bonds (Supplementary Fig. 1 and Table 1).

Rheological measurements were employed to determine the hydrogels’ ultimate shear storage modulus (G’). For the ALYDS hydrogel, G’ in the range of 500 Pa to 10 kPa were achieved (Supplementary Fig. 1a, b, and Table 1) by varying the concentration of crosslinker (1 wt% to 4 wt%) and CMCS (3 wt% to 7 wt%). Hydrogels with a G’ of 2 kPa were selected for stress relaxation, while the stress relaxation rate was quantified using the time taken for the initial stress to relax to half its original value (τ_1/2)²¹. This intrinsic stress relaxation in hydrogels is attributed to Schiff base linkages. In contrast, a control hydrogel containing TDS, yet without Schiff base crosslinking, exhibited intrinsic stress relaxation at a rate nearly 8-fold slower than that of the ALYDS hydrogel (Supplementary Fig. 2, and Tables 2 and 3). The intrinsic stress relaxation times of hydrogels crosslinked with ALYDS, MDADS, and PBEDS varied widely, ranging from 3400 s to 500 s, as shown in Fig. 2b. Firstly, we analyzed the imine reaction rate constants (k₁) and (k₋₁) for different aldehydes using UV spectroscopy. The reverse reaction rate (k₋₁) is crucial as it influences the material’s ability to dissipate applied stress, thereby affecting the stress-relaxation velocity²². Notably, the ALYDS hydrogel, crosslinked using salicylaldehyde, exhibited a slower reaction rate constant (k₋₁) of 0.00109·s⁻¹, leading to a longer τ_1/2 of 3400 s. In contrast, PBEDS demonstrated a faster rate constant (k₋₁) of 0.0017·s⁻¹, resulting in a reduced τ_1/2 of 500 s (Supplementary Table 4 and Figs. 3, 4 and details in supporting information). Moreover, hydrogels’ intrinsic stress relaxation properties can be altered by changing the crosslinking density or polymer concentration (Supplementary Fig. 5a, b). Under light exposure at 20 mW/cm² (445 nm laser), all τ_1/2 values decreased to half of their original values (Fig. 2b and Supplementary Fig. 5), indicating a visible-light-induced responsive behavior. The normalized stress (σ/σ_max)²³ decreased more rapidly in response to higher light intensity (Fig. 2c). Take ALYDS hydrogel as an example, τ_1/2 decreased from 3400 ± 410 s to 1500 ± 230 s as the light intensity increased from 0 to 40 mW/cm² (Fig. 2d). In contrast, a control hydrogel, N¹, N³-bis(3-formyl-4-hydroxybenzyl)-N¹, N¹, N³, N³-tetramethylpropane-1,3-diaminium (C3-ALY), which lacked the TDS moiety, did not exhibit the accelerated stress relaxation under identical light exposure conditions, maintaining an intrinsic stress relaxation rate at 2700 s (Supplementary Fig. 6).

a The stress-relaxation of the hydrogels with different dynamic cross-linkers under the dark condition, as measured by rheology, with constant stiffness at 2 kPa. b Analysis of changes in stress-relaxation behavior using different crosslinkers, with the light intensity fixed at 20 mW/cm² as depicted in a. n = 8, 9 (ALYDS), 6 and 7 (MDADS), 4,4 (PBEDDS) independent samples (0, 20 mW/cm²). c Stress relaxation tests on a dynamically crosslinked ALYDS hydrogel subjected to 15% strain, with stress values normalized to the initial stress. d τ_1/2 of the ALYDS hydrogel under various light exposure conditions, as measured by rheology. n = 7, 9, 8, 9, 9 (0, 5, 10, 20, 40 mW/cm²) independent samples. e Time-dependent changes in viscoelastic creep compliance (J) of the ALYDS hydrogel. f Variation in linear average creep rates (η>) of ALYDS hydrogel under different light intensities as depicted in. n = 3 per group from 3 independent hydrogels. g Exchange reactions involving TETD and PBEDS. h HPLC analysis of the mixed PBEDS (0.01 M) and TETD (0.01 M) in CH3CN/H2O over time. i Reactant concentration change in the dark and under visible light (445 nm) in CH3CN/H2O at room temperature. j Rate constant (k) in TDS-exchange reaction under various light intensities. The Rate study of i, j included three replicates for each group (n = 3). The data in b, d, f, and j are presented as mean values ± S.D.

a Spatial heterogeneity of ALYDS-hydrogel patterned by photomasks. b Fluorescence microscope images showcasing a pattern formed on the hydrogel surface through the thiol-ene reaction, utilizing a photomask and 30 min of light exposure. Scale bars represent 100 μm. c Stress-relaxation measurements of the ALYDS hydrogel were performed using atomic force microscopy (AFM). The hydrogel in illuminated region B relaxed faster than in non-illuminated region A. The inset (image) shows the τ’_1/2 and 1-F_p/F_max of the hydrogel. n = 7 per group from 3 independent hydrogels (dark, light). d Illustration of the ALYDS hydrogel’s stress relaxation rates under alternating light (5 mW/cm²) and dark conditions by rheology. The gray area indicates a resting period of 6000 s. e Graph depicting the cyclic stress relaxation behavior (τ_1/2) of the ALYDS hydrogels under various conditions, as measured by rheology. Each cycle involved applying a 15% strain to the hydrogel for 3000 s at an intensity of 20 mW/cm², followed by a 6000-s recovery period in darkness. This process was repeated over five cycles. n = 3 per group from 3 independent hydrogels. The data (c, e) are present as mean values ± S.D.

a Representative images illustrating the spreading response of Hey cells on ALYDS hydrogel and glass substrates to variations in stress relaxation timescale. Scale bar: 20 μm. b Representative images illustrating Hey cells’ spreading response to variations in short timescale stress relaxation following light exposure on PBEDS hydrogel and CMCS-MA substrates. Scale bar: 20 μm. The immunostaining images of actin are highlighted in red and the nuclei (DAPI, 4,6-diamidino-2-phenylindole) in blue. The YAP/TAZ is green and the nuclei outline is sketched in the images. c Statistical analysis of Hey cells’ spreading area as depicted in a and b. Data analysis was performed using Kruskal–Wallis with Dunn’s multiple comparisons tests, comparing with the long-timescale stress relaxation of ALYDS hydrogel: *P = 0.01452, ****P n = 124, 175, 180, 350 (0 mW /cm², 20 mW /cm² and 40 mW /cm² and glass). Hey cells’s statistical analysis in the short-timescale PBEDS hydrogel was depicted in b: *P = 0.03753, ****P n = 150, 300, 150, and 230 cells (0 mW /cm², 20 mW /cm² and 40 mW /cm² and elastic). d Quantification of YAP nucleo-cytoplasmic (N/C) ratio of Hey cells as depicted in a and b. Data analysis was performed using Kruskal–Wallis with Dunn’s multiple comparisons test (**P = 0.00169, ****P n = 350, 230 for glass, elastic; 140, 120, and 70 cells for 0 mW/cm², 20 mW/cm², and 40 mW/cm² for ALYDS hydrogel; n = 93, 172, 135 cells for 0 mW/cm², 20 mW/cm², and 40 mW/cm² for PBEDS hydrogel). The solid lines (black) in the scatter plots indicate median values. All studies included three independent hydrogels for each group.

Hydrogel creep testing²⁴ was conducted to further assess the viscoelasticity properties. We measured shear strain under a 300 Pa stress while the light source was turned on for 1000 s at an intensity ranging from 0 to 40 mW/cm². Similar to the findings in stress relaxation, we noted an increase in photo-induced creep with higher light intensity. To quantify this effect, we tracked creep compliance (J) over time (Fig. 2e) and calculated the average creep compliance rate (1/ ) for the ALYDS hydrogel²⁵. Under light exposure, the ALYDS hydrogel exhibited significantly enhanced creep compliance compared to its performance under dark conditions. With increasing light intensity, the average creep rate increased from 0.037 MPa⁻¹.s⁻¹ to 0.11 MPa⁻¹.s⁻¹ (Fig. 2f).

The responsive stress relaxation and creep induced by visible light were attributed to an increased rate of disulfide metathesis reaction. To quantify the metathesis rate, a small molecule model reaction including tetramethylthiuram disulfide (TETD) and PBEDS, was used²⁶ (Fig. 2g, see detailed in supporting information). The metathesis reaction between TETD and PBEDS was expected to generate 4-((diethylcarbamothioyl)disulfannecarbonothioyl)-1-(4-formylbenzyl)-1-methylpiperazin-1-ium (DDMI). HPLC analysis revealed that a molar ratio of [TETD]:[PBEDS]:[DDMI] = 3:3:8 was reached after 30 min (Fig. 2h), indicating that the complete scrambling of all three thiuram disulfides reached equilibrium. In contrast, the metathesis reaction proceeded slowly in a dark place, reaching equilibrium after 200 min (Supplementary Fig. 7). The metathesis rate constant (k₂)²⁷ was calculated to be 0.021 L·mol⁻¹·s⁻¹ at 0 mW/cm² and increased to 0.35 L·mol⁻¹·s⁻¹ at 40 mW/cm², as shown in (Fig. 2i). This significant enhancement of k₂ under light illumination confirmed that the TDS moiety plays a pivotal role in visible light- accelerated stress relaxation^28,29. Therefore, by manipulating various factors, such as the type of aldehyde groups, crosslinking density, polymer concentration, and light intensity, one can finely tune the physiological modulus and stress relaxation properties of the hydrogel. This method allows for the customization of hydrogel properties to meet the specific needs of the target tissue being examined³⁰.

Spatial-temporal control of hydrogel stress relaxation via visible light activation

The spatial heterogeneity of hydrogel stress relaxation was accurately reproduced by controlling the TDS shuffling reactions with photopatterning³¹. To visualize the spatial resolution, ALYDS hydrogels were immersed in an alkene-containing fluorescent dye 2-Allyl-6-2-dimethylamino-ethyl-amino-1H-benzo[de]isoquinoline-1,3(2H)-dione (ADBD) solution (see supporting information for details) and then exposed to 445 nm light. The produced sulfur radicals on illuminated rectangular region B could react with ADBD through a thiol-ene click reaction, resulting in green fluorescence emission³² (Fig. 3a). Fluorescence microscopy revealed the distinct fluorescent (area B) and non-fluorescent regions (area A) within the hydrogel, disclosing a high pattern fidelity (Fig. 3b). This experiment provides evidence for achieving high spatial sub-cellular resolution in the light-accelerated TDS metathesis, with patterning precision reaching as fine as approximately 10 μm (Supplementary Fig. 8).

Next, atomic force microscopy (AFM) with a colloidal probe [66] was further employed to measure the in-situ stress relaxation and viscoelastic characteristics (see supporting information for details). To illuminate the hydrogel within the AFM setup, we employed a light source (5 mW/cm²) with a bandpass filter (488 nm) attached to the system. In this microscale measurement, a distinct difference in the stress relaxation (Fig. 3c) was observed between illuminated region B and non-illuminated region A, while their elastic modulus were almost the same (Supplementary Fig. 9). The light induces an increase of relaxation amplitude (1-F_p/F_max)^33,34 and a decrease of the half-life time (τ’_1/2) to reach the relaxation plateau (F_p), demonstrating the visible light-activated viscoelastic heterogeneity.

Using light, we can also achieve precise and reversible temporal regulation of stress relaxation in hydrogels. This is possible due to the reversible changes in the metathesis reaction rate of TDS moieties in response to light exposure³⁵. For example, upon 20 mW/cm² lumination, the ALYDS-hydrogel exhibited fast stress relaxation with τ_1/2 = 1500 ± 230 s, and τ_1/2 = 3400 ± 410 s with the light switched off (Fig. 3d). This switching could be repeated up to 5 times (Fig. 3e). Notably, the ability for light-induced stress relaxation switching fully illustrates the temporal regulation.

Opposite cell spreading responses to light-induced stress relaxation across diverse time ranges

Research has shown inconsistent results on how cells spread in reaction to modified stress relaxation, possibly due to differences in the timescales of stress relaxation examined³⁶. It’s important to note that tissue stress relaxation can differ among individuals and change with age³⁷. In our study investigating cellular responses to light-induced stress relaxation across various timescales, we employed ALYDS and PBEDS hydrogel. These hydrogels exhibit distinct stress relaxation times, with ALYDS’s τ_1/2 ranging from 3500 s to 1500 s, and PBEDS from 500 s to 100 s. We functionalized both hydrogels by incorporating the c(RGDfK) (Arg-Gly-Asp-D-Phe-Lys) cell adhesion peptide to facilitate cell experiments (Supplementary Fig. 10). Human ovarian cancer cells (Hey)³⁸ were cultured on both c(RGDfK) functionalized hydrogels. We observed an increase in cell density as the RGD incubation concentration rose from 0 to 0.25 mg/mL (Supplementary Fig. 11). Consistent with previously published findings³⁹, further augmentation of adhesion ligand quantities (up to 0.5 mg/ml) did not result in greater cell adhesion.

We further investigated the spreading behavior of HEY cells in response to the temporal control of stress relaxation in situ. On the ALYDS hydrogel, which exhibits a slower stress relaxation time of approximately 3400 s, cells showed extensive spreading (1300 ± 570 µm²). This degree of spreading was similar to that observed on glass substrates (Fig. 4a). Notably, this spreading significantly surpassed that on elastic substrates with an identical modulus (990 ± 250 µm²) (Fig. 4b, Supplementary Table 2). When exposed to light intensities of 20 and 40 mW/cm², the τ_1/2 of the ALYDS hydrogel reduced to about 1800 and 1500 s, respectively. We selected a time scale of 30 min to align with both the cellular mechanosensitive response time and the rapid stress-relaxation time (1500 s)⁴⁰. After 30 min of accelerated substrate stress relaxation, the cells quickly transitioned to smaller, more rounded shapes. The average cell area (Ã) decreased markedly from around 1300 µm² to 830 µm², as shown in Fig. 4c. In addition to changes in spreading area, the cells also rapidly evolved towards a more circular morphology in response to the faster relaxation of the substrates (Supplementary Fig. 12a). To rule out the interference from cellular response to visible light, similar experiments were performed with Hey cells on control hydrogels without light responsiveness (see supporting information for C3-ALY hydrogel preparation). In these cases, no differences were observed in cell morphology and spreading area under the aforementioned illumination conditions (Supplementary Fig. 13).

Furthermore, the nuclear localization of the YAP (yes-associated protein) transcriptional⁴¹ regulators was significantly influenced by the in situ modulation of stress relaxation. This is illustrated in Fig. 4a, b, where the YAP nuclear-to-cytosolic ratio diminished from 2.3 ± 0.53 to 1.4 ± 0.21 upon irradiation, indicating cellular remodeling in response to changes in the local micro-environment.

Subsequently, we explored the effect of in situ temporal control of stress relaxation on the PBEDS hydrogel. In contrast to the ALYDS hydrogel, the PBEDS hydrogel, characterized by a shorter τ_1/2 of approximately 500 s, exhibited smaller cell morphologies (450 ± 170 µm², Fig. 4b, c), aligning with the findings previously reported⁴². These substrates undergo faster stress relaxation (200 s, 100 s) when exposed to visible light at 20 and 40 mW/cm². Contrary to expectations, we noted an enhancement in cell spreading, with measurements increasing to 520 ± 270 µm² and 640 ± 280 µm² under these conditions, accompanied by prolonged cell circularity (Supplementary Fig. 12b). Additionally, the YAP nuclear-to-cytosolic ratio exhibited an increase from 1.4 ± 0.62 to 2.2 ± 0.49 (Fig. 4d).

A comprehensive viability study evaluated the potential impact of TDS radicals, visible light exposure, and their combination. The study encompassed an assessment of the toxicity of cell proliferation (Supplementary Fig. 14a) and cell viability (Supplementary Fig. 14b, c). Subsequently, we employed immunostaining for Gamma-H2AX to assess DNA damage (refer to Supplementary Fig. 14d). Our experimental parameters did not reveal any adverse effects on cell behavior, indicating minimal cell damage during light exposure. Therefore, we conclude that observed alterations in cell areas result from temporal regulation in the viscoelastic properties of the hydrogel, rather than any toxic effects induced by light exposure⁴³.

Our study demonstrates the complexity of cell interactions with temporal changes in situ substrate viscoelasticity (Fig. 1). It shows that substrates with slower stress relaxation times (e.g., 3400 s) promote cell adhesion and spreading. Longer relaxation periods can lead to higher integrin-ligand bond formation and increased cytoskeletal contraction⁴⁴. Therefore, the shift to 1500 s stress relaxation times result in reduced spreading areas and decreased focal adhesion (Supplementary Fig. 15). In contrast, when stress relaxation occurs over markedly shorter timescales (from 500 to 100 s), cells were not spread out but were instead rounded. Despite this, cells slightly expand their spreading areas, indicating an adaption to the rapid environmental changes (Fig. 4d). We suggest that at notably short timescales, enhanced stress relaxation might result an increase in filopodia length, which may facilitate their migration and result in minor increases in spreading area⁴² (see Supplementary Fig. 16). These could include adjustments in cytoskeletal dynamics or the activation of mechanotransduction processes that differ from those triggered by slower stress relaxation⁴⁴. Further investigation into cell behaviors on MDADS hydrogel revealed that the shift in relaxation time from 1500 s to 500 s in the intermediate timescale is insufficient to elicit a pronounced response (Supplementary Fig. 17). Specifically, changes in basal stress relaxation at this timescale neither facilitate sufficient energy dissipation to induce cell shrinkage nor prompt extracellular viscosity to elicit cell expansion. Our results highlight the intricate link between cell behavior and substrate mechanics, emphasizing the importance of further research in this area.

Cellular adaptation and mechanical memory in response to dynamic viscoelastic changes in the ECM

We further explored how cellular responses can be reversed in reaction to the dynamic viscoelastic alterations in the ALYDS hydrogel, given its significant impact on cell behavior (Fig. 5a). After exposing cells to light for 30 min; they were subsequently incubated in darkness for different durations: 1 h, 2 h, or 4 h. Notably, our observations revealed that cells slowly regained their spreading area and circularity on substrates after a 4-h incubation period. The time to attain a flattened morphology was longer than the duration needed for cells to round up (Fig. 5b, c). During this time, the YAP nuclear-to-cytosolic ratio also exhibited a recovery, shifting from 1.6 ± 0.40 to 1.90 ± 0.32 (Fig. 5d).

a Representative images of Hey cells cultured on ALYDS gels, subjected to 20 mW/cm² light exposure for 0.5 h and rested for 0 h, 1 h, 2 h, and 4 h. Actin and nuclei are visualized as red and blue, respectively. Scale bar: 25 μm. Analysis of the spreading area b and circularity c of Hey cells on 2 kPa-viscoelastic substrates from a. ****P n = 124, 69, 145, 135, 150 cells from left to fight respectively. d Analysis of Hey cells’ YAP nuc/cyt ratio on 2 kPa-ALYDS hydrogels subjected to 20 mW/cm² light exposure for 0.5 h and rested for varying durations (0 h, 2 h, and 4 h). ns P > 0.05, ****P n = 71, 63, 70, 89 cells respectively. The solid lines (black) in the scatter plots indicate median values. All studies included three independent hydrogels for each group. All statistical analysis was conducted using Kruskal–Wallis with Dunn’s multiple comparisons test.

To enhance our understanding of how cyclic viscoelastic mechanics influence cell spreading, which has relevance in vivo^45,46, we subjected the cells to alternating periods of darkness and light exposure over four cycles (Fig. 6a). During dark periods, cells consistently regained their spreading area after each cycle. Notably, starting from the third light exposure cycle, the cells’ spreading area reduction was less pronounced than in the first cycle. This indicates that cells can “remember” mechanical events experienced during viscoelastic changes by YAP/TAZ shutting at the cell-substrate interface. In alignment with this, an increase in YAP/TAZ localization was noted following the third cycle of light exposure, correlating with the observed changes in cellular spreading⁴⁷ (Fig. 6b–d).

a Immunolocalization of YAP (green) and Actin (red) collected following each cycle of regulation under 20 mW/cm² irradiation for 0.5 h, the nuclei outline (blue) is marked in the image. Scale bar: 100 μm. Quantification of cell spreading areas b and circularity c subjected to cyclic viscoelastic regulation alternating for 4 cycles between 0 mW/cm² for 4 hours and 20 mW/cm² light exposure for 0.5 h. n = 36, 68, 135, 102, 117, 157, 134, 144 cells from left to right. **P = 0.0019; *P = 0.013. d Quantification of the YAP nuclear/cytoplasmic ratio. Data were collected from over 60 cells from e in each group, ****P n = 156, 178, 99, 210 cells from three independent hydrogels. e Hey cells were cultured on the slow-relaxing substrate (azure) for 1 to 7 days before being exposed to 20 mW/cm² light, transitioning to fast-relaxing hydrogels (pink). Subsequently, Hey cells were cultured on fast-relaxing hydrogels for 0.5 h or 1 h before collection and analysis (gray). f Cell area response to in situ increasing viscoelastic relaxation (30 min) after different durations of mechanical dosing (1, 3, 5, and 7 days). n = 124, 173, 197, 202, 150 cells (no light, SR1, SR3, SR5, SR7) from three independent hydrogels. g YAP response to in situ viscoelastic changes after 7 days of mechanical dosing on SR-hydrogels. n = 87, 148, 103 (SR7-0 min, 30 min, 60 min) from three independent hydrogels. The solid lines (black) in the scatter plots indicate median values. All statistical analysis was conducted using Kruskal–Wallis with Dunn’s multiple comparisons test.

To further demonstrate our findings, we subjected Hey cells to mechanical dosing on a slow relaxing hydrogel (τ_1/2 = 3400 s) for varying durations before inducing in situ mechanical changes with light, transitioning the hydrogel to a fast-relaxing state (τ_1/2 = 1800 s) in the presence of cells. Initially, Hey cells were cultured on slow-relaxing hydrogels for one day before switching to the fast-relaxing state (Fig. 6e). After 30 min in the fast-relaxing state (SR1-FR30 min), the cell spreading area decreased from 1440 µm² to 637 µm². However, after 3 days on the slow-relaxing hydrogel, Hey cells began to show some memory behavior; the cell area was reduced to around 1040 µm² after illuminating (Fig. 6f). To increase the mechanical dose, Hey cells were cultured on slow-relaxation hydrogels for 5–7 days before the transition. In this case, a more persistent cell spreading was observed (SR5-FR30 min and SR7-FR30 min). The localization of YAP remained above basal levels up to 60 min (2.5 to 2.3) after light exposure (SR-7-FR 60 min and Supplementary Fig. 18). With 7 days of mechanical dosing, Hey cells demonstrated irreversible activation of YAP after 60 min of light exposure (Fig. 6g and Supplementary Fig. 19). This is reminiscent of the recently proposed concept of “mechanical memory”, which allows cells to remember the stiffness of hydrogel on which they had been previously seeded. Our findings raise the fascinating possibility that “mechanical memory” could also be stimulated by viscoelastic cues⁴⁸. We propose that cells could responsively alter the localization of YAP/TAZ to adapt to dynamic viscoelastic changes in the extracellular matrix (ECM), suggesting a mechanism for cellular mechanical memory^41,49.

Light-induced heterogeneous viscoelastic cues for guiding cell migration

Considering that substrate stress relaxation plays a crucial role in regulating cell spreading, we investigated the potential of light modulation to guide cellular movement^50,51. Visible light was employed to permeate a photomask featuring a gradient in translucency, resulting in the creation of a linear gradient on the substrate above (Fig. 7a). To validate the feasibility of generating continuous gradients through our proposed methodology, the fluorophore ADBD solution was incubated with hydrogels exposed to light for 10 min using the gradient photomask. The outcome revealed a consistent gradient of fluorescence intensity along the X-axis (Fig. 7b, c), with minimal variation observed along the Y-axis after light exposure. AFM experiments further directly measure the spatially heterogeneous stress relaxation and viscoelastic characteristics at four different regions along the gradient (Supplementary Fig. 20a). As suggested by the fluorescent gradient, the relaxation amplitude and characteristic time change gradually (Supplementary Fig. 20b, c). The results illustrate that a continuous and precisely gradual gradient can be established along the X-axis of the hydrogel.

a Schematic depicting the procedure for creating a stress-relaxation gradient within the hydrogel. Measurement of fluorescent green intensity b along the gradient hydrogel, as achieved through the method in a. The resulting fluorescence intensity is graphically represented against the length of the gel c. Scale bar = 1 mm. d Tracking of trajectories for approximately 120 randomly chosen migrating cells under each experimental condition. e Analysis of x-displacement over a period of 150 min at 5-min intervals, observed in cells on various gradient hydrogels. f Calculation of the Steady-Wandering (SW) index for cells on gradient substrates after 150 min. Data represents n = 325 cells for 5 mW/cm², n = 342 cells for 2 mW/cm², and n = 352 cells for 0.8 mW/cm², collected from three independent experiments. The SW index is defined as (slow-fast)/(slow + fast), where it indicates the proportion of cells migrating towards either the slow or fast relaxation region of the gradient. Kruskal–Wallis with Dunn’s multiple comparisons test was used for analysis, *P = 0.037. The black line in the box plot represents the median. g Time-lapse images of Hey cells migration in response to 488-nm-light stimulation. The yellow dashed circles represent the region of light illumination. Scale bar: 10 μm. h Cell migration distance as a function of time (15 min) after light stimulation for n = 9 cells.

To investigate whether the gradient of stress relaxation induced the directed migration of cancer cells, we monitored the movement of Hey cells on the substrates. The cell movements were captured through time-lapse microscopy over 150 min at 5-min intervals. The recorded videos were analyzed using the “manual tracking” plugin in the Image J software⁵². Three distinct gradient strengths were achieved by utilizing light with intensities of 5 mW/cm², and 2 mW/cm², 0.8 mW/cm². Figure 7d depicts representative cell trajectories. Moreover, at median viscoelastic gradients (3400 s-2400 s, 3400 s-2600 s), a notable inclination of cell migration was observed towards the left, corresponding to the direction of slower relaxation (Fig. 7e). As the gradient decreases (3400 s-2900 s), the tendency for cell migration towards the left diminishes. We use SW values to determine the proportion of cells that migrate in both directions (Fig. 7f). At the largest migration gradient (5 mW/cm²), an impressive 70% of the cells exhibit migration towards areas of slower stress relaxation. This gradient corresponds to 15 Pa loss-modulus/mm, as measured by rheology (Supplementary Fig. 21). We also tested a viscoelastic gradient at higher light intensities (10 mW/cm²), however, the rapid cell shrinkage observed at this intensity prevented the cells from exhibiting a stronger migratory tendency (Supplementary Fig. 22). To eliminate the possibility of interference from adaptive cellular protrusions due to light irradiation, comparative experiments were conducted. These experiments involved Hey cells placed on ALYDS hydrogels under a uniform light intensity of 5 mW/cm² without a gradient, and C3-ALY hydrogels under the condition of highest gradient strength (Supplementary Fig. 23). In these experiments, no instances of directional migration were observed. These findings confirmed that the observed cell migration behavior was primarily attributed to differences in the viscoelastic properties of the substrate.

Using a confocal laser spot, we achieved in situ control of single-cell migration with high spatial precision. Hey cells were stained with SPY555-Actin for visualization⁵³, as shown in Fig. 7g. Initially, after 5 min, a decrease in cell spreading within the illuminated area was observed. The Hey cells tended to migrate away from the light spot, moving towards areas with slower stress relaxation (Supplementary Movie 1). As the experiment progressed, the cells migrated, seemingly to minimize their overlap with the mechanically actuated region. We then strategically repositioned the illumination area to ensure that at least part of the cell remained exposed to the mechanical stimulus. This approach allowed us to guide the cell mechanically across the field of view. The average speed of cell migration was recorded at 0.43 μm/min, detailed in Fig. 6h. Notably, when photo-illumination was confined to a single cellular protrusion, we observed a retraction of cellular protrusions in response to the photo-induced changes in viscoelasticity, captured with high spatial resolution (Supplementary Fig. 24).

This observed behavior is qualitatively similar to durotaxis⁵⁴, where cells migrate in response to defined stiffness gradients⁵⁵. In such scenarios, cells react to the tension created by actin fibers, leading to a traction gradient that guides their migration direction⁵⁶. This similarity highlights the significance of viscoelastic changes in influencing cellular migration. Our findings specifically suggest that Hey cells, characterized by their rapid metastatic nature and relatively low adhesion strength, tend to migrate towards areas where they can exert greater myosin expression. (as shown in Supplementary Fig. 25)^57,58. This tendency could potentially enhance their metastatic capabilities. Furthermore, these results, combined with the methodologies used, pave the way for a more detailed investigation into how tissues’ heterogeneous viscoelastic cues impact cancer cell migration.