Stiffness around cells was previously shown by our group to depend on T1C concentration 11 and vary with different treatments and between tested cell lines 11,40. Here, we add new information about stiffness and its anisotropy around two cell lines cultured in distinct ECMs, varying in source, porosity and concentration. For all conditions, G’ values were significantly lower when calculated from PMR data than when estimated from AMR data (Fig. S6). These results are in agreement with previously reported data that found passive microrheology to underestimate G’ due to lower signal-to-noise ratio and the assumption of thermal equilibrium that does not account for the influence of the optical trap and external forces from the cells in calculations of the G* modulus 34,39. Further, the calculation of G* assumes the material is in a local continuum, and that pore size is considerably smaller than the probe bead 34,39, which is not the case for rat tail 1.0T1C, 1.5T1C and 2.0T1C hydrogels. However, in cell-free 3.0T1C hydrogels (Fig. 1b), the pore size distribution shows that most pores are smaller than the bead diameter (2 µm). This crossing of spatial scale may influence the interpretation of AMR data, and may explain in part the similarity in stiffness distribution between 2.0T1C and 3.0T1C hydrogels.
Stiffness around DFs and HT1080s was probed in 4 distinct directions elucidating local anisotropies. Stiffness and cell properties, including expressed nuclear/cytoplasmic YAP ratio, cell solidity or circularity and percentage of secreted fibronectin were shown to vary across the tested ECM types and treatments, indicating a complex cell-ECM relationship based on a variety of factors and characteristics of both cells and the ECM. ECM concentration was found to be a dominant predictor of local stiffness for DFs and HT1080s cultured at different concentrations of rat tail T1C (Fig. 1, S2, S3). These findings are seemingly in opposition to our past studies that found peri-cellular stiffness to be comparable around DFs cultured for 24 hours in rat tail 1.0T1C or 1.5T1C hydrogels 11. Peri-cellular stiffening observed 24 hours after hydrogel preparation was not observed in the current study at the 48 hour time point. Further, in contrast to our previous studies 11, stiffness was largely unaffected by distance from the cell, angular position θ, and axes of bead oscillation. Discrepancy in results could potentially stem from the difference in duration of cell culture, as supported by past research that found hydrogel stiffness to vary with cell incubation time 41,42. Reported results are most likely also affected by additional factors outside the scope of this project, including cell seeding density 42, discrepancies in collagen lots 43 and cell area 44–46.
Both DFs and HT1080s were also shown to respond differently when cultured in three distinct types of hydrogels, formulated to have comparable cell-free median stiffness (Fig. 2b). Our results indicate that cell response to an ECM might not be governed by median stiffness levels alone. Past studies found that cells sense local stiffness anisotropies in 3D hydrogels 47 and thus, cells might also be sensitive to the magnitude of local variances in stiffness within the hydrogels (Fig. 2b). Further, all three types of hydrogels exhibited distinct porosities and microarchitectures – factors, which are known to significantly affect cell survival, proliferation, and migration 48–50. In addition to detecting differences in mechanical properties of hydrogels, cell behavior is known to vary with biochemical properties of the ECM 2,51−53, which is also corroborated by our study. The impact of ECM type on cell properties and local stiffness levels was most pronounced when comparing data collected in collagen hydrogels and fibrin hydrogels. For example, while DFs promoted peri-cellular stiffening in fibrin as compared to either rat tail or bovine skin T1C (Fig. 2d), HT1080s prominently degraded local fibrin, but not collagen matrix (Fig. 4a) 54–56. Even though stiffness levels increased following BB94 treatment of HT1080s in fibrin, fibrinolysis was still observed (Fig. 4). This observation could be consistent with the molecular action of BB94, which is a broad spectrum inhibitor of zinc MMPs, whereas the enzymatic breakdown of fibrin by HT1080s is associated with expression of serine proteases and not directly with MMP activity 57,58.
Rat tail T1C and bovine skin T1C were prepared using the same protocol and differed only with the tissue source of telocollagen. Based on results of SPS-Page tests performed by manufacturer of rat tail T1C and bovine skin T1C (uploaded as “Related files” for review purposes), both types of collagen exhibited similar purity with over 85% of T1C contained within contained within α, β and γ bands. Nonetheless, discrepancy in fiber architectures between the two sources of collagen (Fig. 2a-b) could potentially stem from small differences in amino acid compositions, presence of distinct collagen subtypes other than type 1 or different fibrillogenesis dynamics, which were all previously shown to differ with collagen source, including tissue type and species 59–61. In our study, cells embedded in bovine skin T1C hydrogels with larger pore sizes established lower stiffness values than cells cultured inside rat tail T1C hydrogels with smaller pores (Fig. 2b). These findings indicate that local ECM stiffness established by the cells decreases with the pore size. However, the relationship may not be causal because the cells can also respond to biochemical differences between the collagen types 33. While cells were shown to differentially respond to different types of ECM, the small predictive power of MER suggests that a more comprehensive analysis of factors governing peri-cellular stiffness is still required.
Despite observed effect of ECM type and treatment on stiffness around cells, change in nuclear/cytoplasmic YAP was only detected for Y27632 treatment of HT1080 cells (Fig. 4b, Table S2). While YAP expression was shown to be more prominent inside the nuclei than inside cell cytoplasm for all tested conditions, YAP signal was still widely distributed throughout each cell. Translocation of YAP to the nucleus has been widely reported for cells cultured on 2D substrates with increasing stiffness 62–64, yet translocation was shown to occur at different stiffness values based on cell or substrate type or tested treatment 65–68. For 3D cultures, YAP translocation into the nucleus also varied with cell and ECM type 63,69,70. For example, past studies on fibroblasts embedded inside synthetic fibrous hydrogels reported increase in nuclear/cytoplasmic YAP ratio with fiber density 69, indicating a role in mechanotransduction, but mechanotransduction of human breast cancer cells in 3D cultures was found to be independent of YAP 70. We assert that, to further understanding of the role of YAP in mechanosensing requires measurements of local peri-cellular and not bulk stiffness of the ECM. Such studies may clarify the signal-to-stiffness relationship. Our findings presented here do measure the stiffness sensed by the cells and provide new, but far from comprehensive, understanding regarding roles of ECM types and tested treatments on YAP ratio. Despite no prominent difference in nuclear/cytoplasmic YAP between analyzed conditions, lack of change in YAP ratio could also be attributed to a narrow range of tested stiffnesses in our study (G’ = 0.1–1000 Pa), preventing more prominent YAP translocation to nuclei in stiffer hydrogels or to cytoplasm in softer hydrogels.
Comparatively, fibronectin secretion was shown to be more correlated with local stiffness (Fig. 3c, 5c). For instance, DFs exhibited highest peri-cellular stiffness and fibronectin secretion inside fibrin hydrogels as compared to collagen hydrogels. In contrast, HT1080s promoted much lower stiffness levels and fibronectin secretion inside fibrin hydrogels than inside collagen hydrogels. Nonetheless, after 48 hours of cell culture, fibronectin expression, considered to be colocalized with newly secreted collagen 71–73, was not prominent in the extracellular space. Most of the fibronectin signal was detected inside the cells (Fig. 3a), which is in agreement with past studies on fibroblasts in fibrin hydrogels after 48 hours of culture 74,75. Similarly, collagen secreted by fibroblasts cultured in collagen hydrogels was previously found to be limited to the cell perimeter after 48 hours of culture and was present throughout the whole hydrogel only after 12 days of culture 76. While DFs promote formation of fibronectin fibrils 74,75, HT1080s possess limited ability to assemble fibronectin fibrils without dexamethasone stimulation 77,78. In agreement with past research, our study shows that fibronectin secretion by HT1080s was largely non-fibrillar and lesser in extent than observed for DFs (Fig. 5a). Extent of fibronectin expression indicates that ECM probed by AMR was composed of mostly original, not cell-secreted, ECM. Nonetheless, trends in stiffness across tested ECMs were shown to follow trends in fibronectin expression, and further studies are required to explicate the relationship between peri-cellular stiffness and ECM secretion by cells.
Differential effect of ECM type on how cells remodel the matrix is further evidenced by the addition of Y27632 or BB94 treatments. Stiffening of local matrix by the tested cell types is known to be mediated by contractile forces, which can be inhibited by Y27632 10,11,79 and local matrix degradation by MMP secretion, which is inhibited by BB94 29–31. Interestingly, past studies by our group have showed that MMP secretion can also contribute to stiffening of the ECM, most likely by allowing cell elongation within a dense ECM 10. In fact, MMP activity and cell contractility were essential to ECM stiffening for the case of dermal fibroblasts and aortic smooth muscle cells in type 1 collagen 10. Results from our current study support this finding across multiple ECM types (Fig. 2d, 4b, S7, S8), and further show that Y27632 and BB94 treatments also alter morphology of both cell types, yet the effect of Y27632 and BB94 on YAP and fibronectin expression varied with the type of ECM (Fig. 3,5).
In conclusion, we investigated complex relationship between stiffness established by dermal fibroblasts or HT1080 fibrosarcoma cells and ECM properties, such as hydrogel concentration, type, fiber architecture and pore size. While tested cell lines created highly heterogeneous stiffness landscapes, cell response did not vary with the initial concentration of rat tail T1C hydrogel. In contrast, cells responded differentially when embedded inside different types of ECMs with matched initial stiffness. Given that rat tail T1C, bovine skin T1C and fibrin hydrogels were polymerized at concentrations promoting similar stiffness values in cell-free hydrogels, this work provides further evidence of the importance of measuring peri-cellular and not bulk properties of the ECM. As bulk stiffness might not reflect stiffness sensed by the cells, peri-cellular measurements should be included in comprehensive studies on cellular mechanotransduction.