B16 Melanoma Cancer Cells with Higher Metastatic Potential are More Deformable at a Whole-Cell Level

Abstract

Introduction

Metastasis is a process in which cancer cells spread from the primary focus site to various other organ sites. Many studies have suggested that reduced stiffness would facilitate passing through extracellular matrix when cancer cells instigate a metastatic process. Here we investigated the compressive properties of melanoma cancer cells with different metastatic potentials at the whole-cell level. Differences in their compressive properties were analyzed by examining actin filament structure and actin-related gene expression.

Methods

Compressive tests were carried out for two metastatic B16 melanoma variants (B16-F1 and B16-F10) to characterize global compressive properties of cancer cells. RNA-seq analysis and fluorescence microscopic imaging were performed to clarify contribution of actin filaments to the global compressive properties.

Results

RNA-seq analysis and fluorescence microscopic imaging revealed the undeveloped structure of actin filaments in B16-F10 cells. The Young’s modulus of B16-F10 cells was significantly lower than that of B16-F1 cells. Disruption of the actin filaments in B16-F1 cells reduced the Young’s modulus to the same level as that of B16-F10 cells, while the Young’s modulus in B16-F10 cells remained the same regardless of the disruption.

Conclusions

In B16 melanoma cancer cell lines, cells with higher metastatic potential were more deformable at the whole-cell level with undeveloped actin filament structure, even when highly deformed. These results imply that invasive cancer cells may gain the ability to inhibit actin filament development.

Supplementary Information

The online version contains supplementary material available at (10.1007/s12195-021-00677-w).

Introduction

Metastasis is the process by which cancer cells spread from the site of primary focus to various other organs. This process begins when some cancer cells leave the primary focus site, passing through a tiny gap in the extracellular matrix5,30 that may even be smaller than the cancer cells, and get into blood or lymphatic circulatory systems. Many experiments have shown that cancerous cells are more deformable than their benign counterparts, and that cellular stiffness decreases with the progression of the disease, as reviewed by Alibert et al.,1 although it remains unclear whether all cancer cells adopt this softening strategy.13 These studies strongly suggest that it is advantageous for cancer cells to be soft to instigate the metastatic process, and low stiffness at the whole-cell level is thought to be key for high metastatic potential.

The mechanical properties of cancer cells have been assessed using size-based filtration method,18,26 microfluidic devices,10,23 atomic force microscopy (AFM),25,41 and optical tweezers.15,33 However, these methods do not directly provide quantitative information on cell mechanical properties at the whole-cell level. Size-based filtration methods and microfluidic devices have high throughput, but are weak in the accurate quantification of mechanical properties. Approaches using AFM and optical tweezers are much more accurate than using size-based filtration methods and microfluidic device measurements; however, they assess only local mechanical properties.

Actin filaments play a pivotal role in organizing the mechanical properties of cells.22,28,38,39 Confocal and AFM images have shown a significant difference in the organization of sub-membrane actin structures between normal cells and cancer cells.19 At early stages, cancer cells display abundant well-organized actin structures, whereas late-stage and more aggressive cancer cells show less organized actin.14,19,32 The Young’s modulus of cancer cells has been shown to be inversely correlated with their malignancy, and the structures of actin filaments are known to become less organized and more fragmented in more invasive cancer cells.25

The disorganized structure of actin filaments in cancer cells may facilitate their easy migration and invasion during metastasis. When cancer cells penetrate the endothelium and the basement membrane in the metastatic process, they get squeezed as a whole and receive a compressive load from the surrounding tissues. Reportedly, the cell nucleus is 3–10 times stiffer than the cytoplasm,4,8 suggesting that the compressive properties of whole cells could be a key to their metastatic potential. Since assessment with AFM is limited to local, relatively small, and near-surface mechanical properties, it remains unclear whether more invasive cancer cells have reduced stiffness at the whole-cell level even when largely deformed.

B16 melanoma cells are a well-established cell line to study the characteristics of metastatic cancer cells.7,31 Two metastatic B16 melanoma variants (B16-F1 and B16-F10) have been widely used.23,41 B16-F1 cells are a cell line obtained by administering B16 cells to mice by a tail vein injection and harvesting cells that have metastasized to the lungs. B16-F10 cells are obtained by the same selective procedure but repeated 10 times, meaning that the B16-F10 line is more highly selected for invasive and metastatic capacity than the B16-F1. In in vivo and in vitro models, B16-F10 cells have a higher metastatic potential than B16-F1 cells.7,24,31 For example, after injecting B16-F1 and B16-F10 cells into the mouse tail vein, the B16-F10 cells formed significantly more lung nodules (170 ± 33) than the B16-F1 cells (29 ± 6).24 Indirect evidence from the Transwell assays also showed that migration of the B16-F10 cells was 3 times higher than that of B16-F1 cells.41

In the present study, we performed compressive tests on these two metastatic B16 melanoma variants (B16-F1 and B16-F10) to characterize the compressive properties of cancer cells with different metastatic potential at the whole-cell level when largely deformed. Differences in their compressive properties were analyzed by examining actin filament structure and actin-related gene expression.

Materials and Methods

Cells

Cells from the mouse melanoma cell lines B16-F1 and B16-F10 were used. As mentioned above, B16-F10 cells show higher metastatic potential than B16-F1 cells.7,31 The cells were cultured in Dulbecco’s modified Eagle’s medium (05919, Nissui) supplemented with 10% fetal bovine serum (172012, SIGMA), 10 mM HEPES, 100 U/ml penicillin, and 100 µg/ml streptomycin (complete medium) in tissue culture dishes. The cells were harvested from the dishes using trypsin (25200-056, Gibco) to float cells.

Actin filaments were disrupted to understand the contributions of actin filaments in defining the compressive properties of cells. Before harvesting, the cells were treated with 10 µg/ml cytochalasin D (CD, 22144-77-0, TOCRIS) that is potent inhibitor of actin polymerization, and incubated for 3 h following the method of Ujihara et al.37

RNA-Sequence Analysis

Mouse genome sequence and RNA-seq data of B16-F1 (3 datasets) and B16-F10 (3 datasets) were downloaded from NCBI database (IDs: GRCm38.p6, SRR968561, SRR9685614, SRR9685616, SRR12439166, SRR12439170, and SRR12439171). The RNA-seq reads were mapped to mouse mRNA data using Salmon software (version 0.8.2). To investigate the enrichment pathways in B16-F10 cells, Kyoto Encyclopedia of Gene and Genomes (KEGG) pathway analysis was performed using the KEGG Automatic Annotation Server.21 Genes upregulating > 2.0 and downregulating < 0.5 in B16-F10 cells relative to B16-F1 were used as the input gene list, and genes expressed in either B16-F1 or B16-F10 cells were used as the background. To ensure the analysis’s reliability, we restricted the analysis to pathways with more than 100 genes expressed in the B16-F1 and B16-F10 cells.

Fluorescent Imaging

Actin filaments in floating cells were observed under a confocal laser scanning microscope. Cells, suspended in Hanks’ balanced salt solution (HBSS, 084-08965, Wako), were placed on a glass bottom dish treated with poly-L-lysine (167-18651, Wako) and fixed with 10% neutral buffered formalin for 10 min at room temperature. This was followed by the second wash with phosphate buffered saline (PBS) and permeabilization with PBS containing 0.2% Triton X-100 for 10 min. After washing, cells were incubated in blocking with 4% albumin from bovine serum (019-23293, Wako) /PBS solution for 15 min. Actin filaments were then stained with Alexa Fluor 488 phalloidin (A12379, Invitrogen), diluted to 1:200 with PBS containing 1% bovine serum albumin for 30 min, and the cells were again washed with PBS. Images were obtained using a confocal laser scanning microscope (FV3000, Olympus) equipped with a 60× oil immersion objective lens [numerical aperture (NA) = 1.35, UPLSAPO 60XO, Olympus]. A laser (OBIS, Coherent) with an excitation wavelength of 488 nm was used. All cells were observed under the same imaging settings. The fluorescent intensity of the actin filaments lining the cell membrane was quantified to compare the amount of intracellular actin filaments between the groups. We measured the fluorescent intensity at 36 locations defined on the periphery of the cell membrane on single confocal slices that pass almost through the center of cells. Their averages were then used as an index representing the amount of actin filaments in each cell. The index was normalized with the mean of the indices of the intact B16-F1 cells.

Compressive Test

The method used for compressive tests is fully described in Ujihara et al.39 Figure 1 shows the experimental setup, adapted with slight modifications from Hashimoto et al.9 The experimental setup consisted of an inverted microscope (IX-73, Olympus), a 60× oil immersion objective lens (PlanApo N 60×, Olympus), a CMOS camera (ORCA flash 4.0, Hamamatsu Photonics), and a pair of glass microplates, each connected to three-axis motorized micromanipulators (EMM2, Narishige). One of the microplates was a compressive plate and the other was a cantilever plate, with the compressive plate being far more rigid than the cantilever plate. The microplates were uncoated and untreated. The cells that had fallen to the bottom of the chamber were picked up with a compression plate. The cell was attached to the plate owing to its natural adhesiveness. The cell was held between the tips of the compressive and cantilever plates and compressed by moving the compressive plate at a rate of 0.5 μm/s. A compressive test for each cell was performed in < 2 min. The force applied to the cell was estimated from the deflection of the cantilever plate whose spring constant was readily measured by a cross-calibration.16,39 Both plates were fabricated from 0.1 × 1 × 100-mm glass plates (custom-made, Matsunami Glass) using a micropipette puller (PC-10, Narishige). The test was carried out at a room temperature.

Figure 1.

Compressive test system for the cells. (a) Overview of the experimental setup. (b) A magnified view of the main part of the instrument to hold and compress the cell. (c) The definition of cell initial length (L₀), cell length (L_c), and deflection of cantilever plate (X).

Data Analysis and Statistical Methods

Positions of the compressive and cantilever plates were measured using MetaMorph Offline software (version 7.7.0.0; Molecular Devices). The cell length, L_c, was defined as the distance between the compressive and cantilever plates (Fig. 1c). The cell deformation, D, was defined as L₀ − L_c, where L₀ is the initial cell length. The cell strain, ε, was defined as D/L₀. The force, F, applied to the cell was calculated by multiplying the deflection of the cantilever plate, X, by its spring constant (1.0–3.0 nN/μm). A cell’s mechanical property was assessed with the Young’s modulus, E, by fitting the hyperelastic Tatara model10,36 to the F − ε curve (see Appendix). The stiffness, S, was defined as the slope of the force (F) − deformation (D) curve for every 0.1 strain ε based on the assumption that the curve is piecewise linear. The strain data from 0 to 0.4 were analyzed. Data are expressed as mean ± standard deviation (SD). Paired data were evaluated using the Student’s t-test. For multiple comparisons, analysis of variance with Tukey–Kramer tests was performed. A value of P < 0.05 was considered statistically significant.

Results

To examine differences in gene expression between B16-F1 cells and B16-F10 cells, we analyzed RNA-seq data. Compared with B16-F1 cells, 1610 genes were highly expressed in B16-F10 cells and 7611 genes were downregulated (Fig. 2a). We examined the enriched pathways using the KEGG functional network analysis and identified the 40 most enriched biological pathways (Fig. 2b). The “Regulation of the actin cytoskeleton,” the 19th most enriched pathway out of 381, was in the top 5%. Some of the higher-ranked pathways were unrelated to cell deformability; “Tight junction” was the only pathway that could directly affect the cell’s stiffness. Therefore, among the factors related to cell deformability, “Regulation of the actin cytoskeleton” ranked near the top.

Figure 2.

RNA-seq analysis of B16-F1 and B16-F10 cells. (a) The differences in gene expression between the B16-F1 (n = 3 datasets) and B16-F10 cells (n = 3 datasets). The horizontal axis represents the log₂ of transcripts per million (TPM). The vertical axis represents the log₂ of each gene’s fold change between the B16-F1 and B16-F10 cells. Red dots, genes expressed in B16-F10 cells at a level more than twice than that in the B16-F1 cells; black dots, comparable gene expression levels in both cells; blue dots, genes expressed in the B16-F10 cells at a level less than half of that in the B16-F1 cells. (b) The 40 most enriched biological pathways according to KEGG analysis. (c) Of the top 20 of fold changes, actin-related genes with a TPM greater than 1. TV1, transcript variant 1.

We also compared the relative differences in the gene expressions of the actin-related proteins between B16-F1 and B16-F10 cells. The relative differences were evaluated with fold changes, calculated by dividing the transcripts per million (TPM) of B16-F1 cells by that of B16-F10 cells. Supplemental Table 1 shows the calculated fold changes in descending order. Genes with a TPM value below 1 indicate that the results could be either artifacts or those genes do not generally affect the cells. Of the top 20 fold changes, genes with TPM value greater than 1 included fascin actin-bundling protein 1 (Fscn1), capping protein (actin filament) muscle Z-line, alpha 1 (Capza1) transcript variant 1 (TV1), actin-related protein 2/3 complex subunit 2 (Arpc2) TV1, anillin, actin-binding protein (Anln), ARP3 actin-related protein 3 (Actr3) TV1, actinin, and alpha 1 (Actn1) TV1 (Fig. 2c). Fscn1 has two major actin-binding sites and organizes filamentous actin into parallel bundles.12 Capza1 is an actin-binding protein that promotes the nucleation of actin polymerization.11 Arpc2 and Actr3 are components of the Arp2/3 complex, a multiprotein complex that induces actin polymerization upon stimulation by a nucleation-promoting factor.42 Anln is an actin-binding protein that bundles actin filaments.27 Actn1 is a filamentous actin-crosslinking protein.29 The expression levels of the actin filament-related genes described above were drastically lower in B16-F10 cells than in B16-F1 cells (Fig. 2c and Supplemental Table 1), suggesting that B16-F10 cells have less developed actin filaments than B16-F1 cells.

To confirm the differences in actin filament structure between B16-F1 and B16-F10 cells, we observed cross-sectional fluorescent images of actin filaments in floating B16-F1 cells and B16-F10 cells. For reference, cross-sectional fluorescent images of actin filaments in adherent cells are shown in Supplemental Fig. 1. As seen in Fig. 3a, B16-F1 cells clearly showed actin filaments outlining the cell membrane, with fibrous forms inside. In contrast, B16-F10 cells showed less outlining actin filaments, while interior actin filaments similarly existed (Fig. 3b). Treatment with CD, potent inhibitor of actin polymerization, caused significant reduction of actin filaments in B16-F1 cells (as seen in Fig. 3c), but not as much in B16-F10 cells, which looked similar to intact B16-F10 cells (Fig. 3d). The amount of actin filaments was quantified based on their fluorescence intensity. The fluorescence intensity obtained for each cell was normalized with the mean of the intact B16-F1 cells. The fluorescence intensity of the B16-F10 cells was significantly lower than that of the B16-F1 cells (Fig. 3e), suggesting that actin filaments in B16-F1 cells were more abundant than in B16-F10 cells.

Figure 3.

Cross-sectional fluorescence images of actin filaments in suspended cells. (a) Intact B16-F1 cells (F1, n = 26). (b) Intact B16-F10 cells (F10, n = 23). (c) B16-F1 cells treated with CD (F1 + CD, n = 20). (d) B16-F10 cells treated with CD (F10 + CD, n = 25). (e) The fluorescence intensity obtained for each cell was normalized with the mean of the intact B16-F1 cells.

Compressive tests were performed to characterize the compressive properties of B16-F1 and B16-F10 cells at the whole-cell level. The snapshots of the cells during the compressive test at a force of 0 and 10 nN are shown in Fig. 4. Cell strain ε is also listed. In all groups, cells at 0 nN were almost spherical, and appeared to be flattened with compression, causing an increase in the contact area between the cell and the cantilever plate. Application of the same compressive force induced more deformation of B16-F10 cells than B16-F1 cells (Figs. 4a and 4b). Treatment with CD resulted in larger deformation in B16-F1 cells (Figs. 4a and 4c), but not as much in B16-F10 cells (Figs. 4b and 4d).

Figure 4.

Snapshots of the cells during the compressive test at different forces. (a) Intact B16-F1 (F1) cells at forces of 0 nN (left) and 10 nN (right). (b) Intact B16-F10 (F10) cells at forces of 0 nN (left) and 10 nN (right). (c) B16-F1 cell treated with CD (F1 + CD) at forces of 0 nN (left) and 10 nN (right). (d) B16-F10 cells treated with CD (F10 + CD) at forces of 0 nN (left) and 10 nN (right). The cell strains are presented above each figure.

The force–strain curves of B16-F1 and B16-F10 cells with and without CD are shown in Fig. 5. In all groups, the force increased significantly as cell compression progressed. The force–strain relationships between intact B16-F10 cells, CD-treated B16-F1 cells, and B16-F10 cells were quantitatively similar (Figs. 5b–5d). It would appear that larger forces were required to compress intact B16-F1 cells for the same strain amount, compared with other groups. Statistically significantt difference was found between intact B16-F1 cells and B16-F10 cells for all the stains including small strains.

Figure 5.

Force–strain relationships in intact and treated B16-F1 and B16-F10 cells. (a) Intact B16-F1 cells (n = 20). (b) Intact B16-F10 cells (n = 20). (c) B16-F1 cells treated with CD (n = 10). (d) B16-F10 cells treated with CD (n = 14). Each line represents an individual cell. Marks and bars represent the mean and SD, respectively. F1, B16-F1 cells. F10, B16-F10 cells. F1 + CD, B16-F1 cells treated with CD. F10 + CD, B16-F10 cells treated with CD.

To compare the mechanical material properties of the cells among cell groups, we estimated Young’s modulus by fitting the hyperelastic Tatara model36 to the force–strain relationships. Representative data with the fitted curves are given in Fig. 6a, demonstrating that the fitted curves closely match the experimental results. The Young’s moduli are presented in Fig. 6b. Mean ± SD of the Young’s modulus was 531.3 ± 268.0 Pa for B16-F1, 181.2 ± 62.7 Pa for B16-F10, 265.4 ± 128.3 Pa for CD-treated B16-F1, and 226.1 ± 108.5 Pa for CD-treated B16-F10. The SD of the Young’s modulus of intact B16-F1 cells was larger than for other groups. The mean Young’s modulus of B16-F1 cells was significantly higher than that of B16-F10 cells. The mean Young’s modulus of B16-F1 cells treated with CD was significantly lower than that of intact B16-F1 cells, and was similar to that of intact B16-F10 cells. The mean Young’s modulus of B16-F10 cells treated with CD was almost the same as that of untreated B16-F10 cells. These results imply that differences in the Young’s modulus between B16-F1 cells and B16-F10 cells are attributable to differences in actin filament development.

Figure 6.

Young’s moduli of intact and treated B16-F1 and B16-F10 cells. (a) The representative data of compressive force–strain curves with regression lines fitted with the hyperelastic Tatara model. F1, B16-F1 cells. F10, B16-F10 cells. F1 + CD, B16-F1 cells treated with CD. F10 + CD, B16-F10 cells treated with CD. (b) A comparison of Young’s moduli of the intact B16-F1 cells (F1, n = 20), intact B16-F10 cells (F10, n = 20), B16-F1 cells treated with CD (F1 + CD, n = 10), and B16-F10 cells treated with CD (F10 + CD, n = 14). NS, not significant (P > 0.05).

Finally, to clarify the strain region in which the actin filaments contribute to the whole-cell deformability, we assessed whole-cell stiffness, S, for every 0.1 strain ε for all groups. All groups showed similar relationships between cell stiffness and cell strain (Fig. 7). B16-F1 cells had significantly higher stiffness than other groups. The difference in stiffness between intact B16-F1 cells and B16-F1 cells treated with CD vanished in the strain range of 0.3–0.4, suggesting that actin filaments contribute to the compressive properties of whole cells, particularly in the low-strain region. Stiffness was similar across groups, except for intact B16-F1 cells, and interestingly the stiffness of B16-F10 cells was not reduced by CD treatment.

Figure 7.

Change in whole-cell stiffness during cell compression. Intact B16-F1 cells (F1, n = 20), intact B16-F10 cells (F10, n = 20), B16-F1 cells treated with CD (F1 + CD, n = 10), and B16-F10 cells treated with CD (F10 + CD, n = 14). NS, not significant (P > 0.05).

Discussion

When cancer cells metastasize, they penetrate the endothelium and basement membrane, and pass through tiny gaps in the extracellular matrix. On their way out, these cancer cells are squeezed by compressive forces from surrounding cells and tissues. Thus, it would be advantageous for these cells to have reduced stiffness at the whole-cell level. In the present study, we performed whole-cell compressive tests to understand cancer cell metastasis, while previous studies evaluated only local cellular compressive properties using AFM.2,34,35,41 Moreover, the present study imposed large deformations as cancer cells largely deform in the metastatic process.43 Our results show that cells from the more invasive B16-F10 cell line had lower Young’s modulus and stiffness than those from the less invasive B16-F1 cell line at the whole-cell level, even when the cells were largely compressed. Such a difference in the compressive properties of cancer cells were analyzed with a focus on actin filaments. Although the role of the cytoskeleton in determining the mechanical properties of cancer cells has been discussed in previous studies, the RNA-seq analysis conducted in the present study helps to further understand the mechanical properties of cancer cells with varying metastatic potentials. The investigation of cellular mechanical properties with RNA-seq analysis provides new insights into understanding the differences between the mechanical properties of cancer cells with varying metastatic potentials.

Generally, cancer cells with higher metastatic potential were softer. Such a trend is consistent with other studies2,25,35 that have evaluated the local compressive properties of other cancer cells with different metastatic potentials. The Young’s modulus obtained with various techniques are not necessarily the same due to differences in experimental setup, i.e., how the measurements are collected and what area of the cell is probed.44 However, interestingly, the Young’s modulus of B16-F1 and B16–F10 obtained in this study were comparable to the values reported by Watanabe et al.,41 who probed the same cell lines, but in an adhered state, using AFM (727.2 Pa and 350.8 Pa). Because cells are adherent to the extracellular matrix during metastasis, it is better to characterize the mechanical properties of adherent cells. AFM2,35,41 is often used to assess the mechanical properties of adherent cells. Although AFM is more accurate than other techniques, including the compressive tests performed here, data obtained with AFM is limited to local, relatively small, and surface-adjacent mechanical properties. However, in the metastatic process, cancer cells are squeezed as a whole and receive a compressive load from surrounding tissues to pass through the collagen fiber networks into the extracellular matrix.43 The compressive tests carried out here allows us to compress cells to a greater degree than AFM and characterize the compressive properties at the whole-cell level. Although the use of floating cells is a limitation of the present compressive test, the results of the present study show that, using the mechanical properties at a whole-cell level, we can draw the same conclusions as those in previous studies that measured local mechanical properties. Extension of biomechanical discussions of adherent cells to those of floating or suspended cells may also be possible, with a care in interpreting the value of the Young’s modulus that could be smaller for suspended cells due to depolymerization of actin filaments upon detachment.25

The relationship between the metastatic potential and mechanical properties of melanoma cancer cells has been investigated using AFM.2,3,34,35 Bobrowska et al.2 examined the Young’s modulus of six melanoma cell lines (WM115, WM793, WM266-4, VM239, 1205Lu, and A375P cells) and a reference cell line (HEMa-LP) by AFM. They found that the cells with higher metastatic potential tended to have higher deformation potential. Sarna et al.34 reported that the in vitro transmigration through narrow barriers was suppressed in human melanoma SKMEL-188 cells due to melanin granules in the cells that drastically increased the Young’s modulus. The present findings are consistent with the reported correlations between metastatic potential and Young’s modulus of non-B16 melanoma cells. Considering this, the correlation between metastatic potential and cell stiffness is not limited to the B16 melanoma cell line but generally works for other melanoma cancer cells as well.

Disruption of actin filaments in B16-F1 cells significantly reduced the Young’s modulus and stiffness. This observation supports the hypothesis that actin filaments play an important role in determining the compression properties of cancer cells with low metastatic potential. B16-F10 cells with high metastatic potential did not decrease their Young’s modulus and stiffness despite disruption of the actin filaments. Moreover, the Young’s modulus of B16-F1 cells with disrupted actin filaments was at the same level as intact B16-F10 cells as well as B16-F10 cells with disrupted actin filaments. These results, along with molecular biological approaches, such as RNA-seq analysis and fluorescence microscopic imaging, strongly suggest that the difference in compression properties between B16-F1 and B16-F10 cells is attributed to the degree of actin filament development.

B16-F1 cells showed greater variation in the Young’s modulus than B16-F10 cells (Fig. 6b), similar to findings reported by Watanabe et al.41 They noticed a wider distribution of the Young’s modulus in B16-F1 cells compared to B16-F10 cells, and attributed this to the greater heterogeneity in metastatic potential of B16-F1 cells. While B16-F10 cells are a more highly selected population of cells that have experienced 10 times metastasis, B16-F1 cells are a group of cells that have undergone a single transfer, in which cells with low and high metastatic potentials coexist. The greater variability in the Young’s modulus of B16-F1 cells is thus a result of their varied metastatic potentials. It is therefore necessary to bear in mind that the Young’s modulus of B16-F1 cells is not representative of low metastatic cancer cells in general.

Previously, we performed the compressive tests in fibroblasts harvested from the patellar tendon of a mature Japanese white rabbit39 and demonstrated that the Young’s modulus of the fibroblasts was 631 ± 380 Pa, which was comparable to that of B16-F1 cells (531 ± 268 Pa) obtained in the present study (P > 0.05, t-test). Although cancer cells are reported to be softer than normal cells,19 we found that B16-F1 cells with low metastatic potential had almost the same Young’s modulus as normal cells (fibroblasts). However, generalizing this result should be done with caution because only one normal cell line was examined. The close match of the Young’s modulus could be attributed to the use of floating cells in our compressive test. As shown in Fig. 3, floating cells have a less developed structure of thick actin bundles inside cells as compared with the adherent cells presented in Supplemental Fig. 1. Consequently, the mechanical properties of the floating cells are mainly determined by the cell membrane, cytoplasm, and nucleus. Although detailed data is not available, basic constituents of these components of B16-F1 cells could be the same as in fibroblasts. These similarities could account for no significant difference between the Young’s modulus in cancer cells and fibroblasts.

Regardless of the cell group, the force–strain relationships of the cells showed a downwardly convex relationship. Since the same has been found in other compressive tests of non-cancer cells,4,39 this is presumed to be a characteristic of the compressive test. As a cell is compressed by a cantilever, the contact area between the cell and the cantilever plate increases. Since a strain is given, the force acting on the cantilever plate increases with enlargement of the contact area. The nonlinearity is also accounted for by the presence of the cell nucleus, which is reported to be 3–10 times stiffer than the cytoplasm.4,8 During cell compression, the cell nucleus exerts more effect on the compressive property of the whole cell, resulting in a nonlinear increase in the force required to compress.

The present study focused on actin filaments as the components likely responsible for the compression properties of whole cells. However, other subcellular components (such as microtubules, intermediate filaments and cell nuclei) may also affect whole-cell compressive properties because actin filaments interact with them mechanically and functionally.6,17,20,40 For example, actin filaments and microtubules are linked via microtubule-associated proteins.20 Actin filaments and vimentin filaments (a type of intermediate filament) interact directly, and their mixed filamentous network has significantly higher stiffness than networks made of only actin filaments or only vimentin filaments.6 Actin filaments have also been shown to be important in nuclear positioning.17 As a consequence, the disruption of actin filaments not only decreases their direct elastic effects but also indirectly weakens the contribution of other components. In other words, it may be the case that the effects observed in the present study included all the interactive effects of subcellular components caused by the disruption of actin filaments. In the future, we plan to investigate the effects of subcellular components other than actin filaments on the compressive properties of whole cells to clarify their contributions to the mechanical properties and metastatic potential of cancer cells.

Conclusions

In this study, we investigated the relationship between metastatic potential and compressive properties of whole cells. Our RNA-seq analysis revealed that the expression levels of actin filament-related genes were drastically lower in B16-F10 cells than in B16-F1 cells. Using fluorescence microscopy, we demonstrated that B16-F10 cells had less organized and structured actin filaments than B16-F1 cells. Further, the Young’s modulus of B16-F10 cells was significantly lower than that of B16-F1 cells. The disruption of actin filaments in B16-F1 cells by CD reduced the Young’s modulus to the same level as that of B16-F10 cells. The Young’s modulus in B16-F10 cells was unaffected by the CD treatment. In conclusion, in B16 melanoma cancer cells, cells with higher metastatic potential were more deformable at the whole-cell level with undeveloped and less organized structures of actin filaments, even when highly deformed. These results suggest that invasive cancer cells may gain the ability to inhibit actin filament development.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

AFM

Atomic force microscopy

KEGG

Kyoto Encyclopedia of Gene and Genomes

CD

Cytochalasin D

TPM

Transcripts per million

KAAS

KEGG Automatic Annotation Server

HBSS

Hanks’ balanced salt solution

PBS

Phosphate buffered saline

SD

Standard deviation

TV1

Transcript variant 1

Appendix

The hyperelastic Tatara model was adopted to describe the mechanical behavior of cells. Here we used the model originally developed by Tatara36 and modified by Hu et al.10 to represent cellular elastic behaviors during the compression test. The hyperelastic Tatara model describes the relationship between the applied force F and compressive strain ε as

$$F=Eg\left(\epsilon \right)$$ A1

and $g\left(\epsilon \right)$ is a function of strain ε given by

$$g\left(\epsilon \right)={\left\{\frac{3\left(1-{\nu }^2\right)A}{2\left(L_0-L_{\text{c}}\right)}\left(1+\frac{2B{a}^2}{D_{\text{deform}}^2}\right)\frac{1}{a}-\frac{2A}{\pi \left(L_0-L_{\text{c}}\right)}\left(1+\frac{4B{a}^2}{5D_{\text{deform}}^2}\right)\frac{1}{f\left(a\right)}\right\}}^{-1}$$ A2

where A and B are the function of strain ε as described below. In Eq. (A2), L₀ and L_c are the lengths of the cell before and after compression, respectively. D_deform is cell deformed diameter (Fig. A1), and ν is a Poisson’s ratio, a is the contact radius, f(a) is the characteristic length of non-spherical geometry after compression, A and B (related to hyperelastic correction) are

$$a=\frac{1}{2}\left(\sqrt{L_0^2-L_c^2}+D_{\text{deform}}-L_0\right)$$ A3

$$f\left(a\right)={\left\{\frac{\left(1+\nu \right)L_0^2}{2{\left(a^2+L_0^2\right)}^{3/2}}+\frac{1-{\nu }^2}{{\left(a^2+L_0^2\right)}^{1/2}}\right\}}^{-1}$$ A4

$$A=\frac{{\left(1-\epsilon \right)}^2}{1-\epsilon +\frac{{\epsilon }^2}{3}}$$ A5

$$B=\frac{1-\frac{\epsilon }{3}}{1-\epsilon +\frac{{\epsilon }^2}{3}}$$ A6

The hyperelastic Tatara model was fitted to the F − ε data by the least squared method to obtain Young’s modulus, E, under the assumption of material incompressibility (ν = 0.5).

Figure A1.

Geometrical parameters for fitting the hyperelastic Tatara model. L₀ and L_c are the lengths of the cell before and after compression, respectively. D_deform is the cell deformed diameter, and a is the contact radius.