TAGLN mediated stiffness-regulated ovarian cancer progression via RhoA/ROCK pathway

Background Ovarian cancer (OC) progression is an unmet medical challenge. Since omental metastases were palpated harder than their primary counterparts during cytoreductive surgery of patients with epithelial ovarian cancer (EOC), we were inspired to investigate OC progression from the perspective of biomechanics. Methods Atomic Force Microscope (AFM) was used to measure the Young’s modulus of tissues. The collagen-coated polyacrylamide hydrogel (PA gel) system was prepared to mimic the soft and stiff substrates in vitro. The effect of TAGLN was evaluated both in vitro and in vivo using transwell assay, immunofluorescence, western blot analysis and immunohistochemistry. Results We quantitatively confirmed that omental metastases were stiffer and more abundant in desmoplasia compared with paired primary tumors, and further demonstrated that matrix stiffness could notably regulate OC progression. Remarkably, TAGLN, encoding an actin cross-linking/gelling protein, was identified as a potent mechanosensitive gene that could form a regulation loop with Src activation reacting to environmental stiffness, thus mediating stiffness-regulated OC progression through regulating RhoA/ROCK pathway. Conclusions These data demonstrate that targeting extra-cellular matrix (ECM) stiffness could probably hamper OC progression, and of note, targeting TAGLN might provide promising clinical therapeutic value for OC therapy. Supplementary Information The online version contains supplementary material available at 10.1186/s13046-021-02091-6.


Background
Peritoneal metastasis is an important cause of high mortality of ovarian cancer (OC) [1], especially the metastasis to the omentum [2,3], which results in transformation of this soft pad of tissue to a solid tumor [2]. Intriguingly, during ovarian cancer cytoreductive surgery, we noticed that these omental metastases are palpated significantly harder than their primary tumors of EOC. Since metastasis is a major difficulty of OC therapy [4] and metastases are palpated harder than primary tumors, we are interested in the correlation between ovarian cancer progression and tissue stiffness.
Tumors are often stiffer than normal tissues [5][6][7], frequently detected through physical palpation as a rigid mass residing within a compliant tissue [5]. This altered tissue-level and cellular mechanics is mainly due to extracellular matrix (ECM) remodeling [8,9]. Recently, tumor mechanics have emerged as an important driving factor during tumor progression and chemotherapy sensitivity [10][11][12]. For example, in breast cancer, matrix stiffness drives epithelial-mesenchymal transition and tumor metastasis through a TWIST1-G3BP2 mechanotransduction pathway [10]. Given that, targeting ECM stiffness presents opportunities to attenuate tumor progression and improve the efficiency of chemotherapy [13,14]. Cells perceive their microenvironment not only through soluble signals but also through physical and mechanical cues [15]. Past decades of research have mainly concentrated on the role of various extracellular and intracellular biochemical signals in tumor progression [16,17], but whether and how mechanical cues play a role in the progression of OC and the underlying mechanism remain unclear.
Therefore, we aim to investigate whether and how tissue mechanics exerts influence on OC progression. We confirmed that omental metastases were stiffer than paired primary tumors and verified the impact of matrix stiffness on OC progression. Through transcriptomic array, TAGLN (transgelin) was identified as a mechanosensitive gene and formed a regulation loop with Src gene activation. We further demonstrated that TAGLN mediated stiffness-regulated OC progression through regulating RhoA/ROCK signaling pathway. Collectively, this study deepens our understanding of tumor biomechanics of OC progression and provides novel targets for OC therapy.

Cell culture
The human ovarian cancer cell lines SK-OV-3 and ES-2 were purchased from American Type Culture Collection (ATCC) and were cultured in McCoy's 5 A medium (Sigma-Aldrich) supplemented with 10 % fetal bovine serum (FBS) (Gibco) and 100 units/ml penicillin/streptomycin (Servicebio, Wuhan). All cells were maintained at 37 ℃ and 5 % CO 2 .

Antibodies and reagents
The complete information on all the antibodies and reagents used in the present study is provided in supplementary Tables 1-2.

Immunofluorescence and imaging
Cells were fixed with 4 % paraformaldehyde for 30 min at room temperature (RT) and permeabilized and blocked with 0.1 % Triton X-100 in 5 % BSA for 20 min at RT. All samples were incubated with primary antibody at 4℃ overnight and followed by Alexa-conjugated secondary antibody for one hour at RT. Cells were subsequently stained with 10 µM phalloidin-TRITC (Sigma-Aldrich) for 30 min. Nuclei were counterstained with diaminophenylindole (DAPI) (Sigma-Aldrich). Immunofluorescence staining was visualized using fluorescence microscopy (Olympus DP73, Tokyo, Japan).

Clinical samples
Human ovarian cancer tissues were all obtained from patients undergoing surgery at the Department of Obstetrics and Gynecology, Tongji Hospital, Huazhong University of Science and Technology after obtaining written informed consent of the patients and the authorization of the Ethics Committee of Tongji Hospital (TJ-iRB20181103). Fresh tissues were prepared for the Atomic Force Microscope (AFM) measurement, frozen or formalin fixed.
For Kaplan-Meier analysis, TAGLN expression was detected in our samples of 132 OC patients on the available clinical data including time of diagnosis, death, or last follow-up. Progression-free survival (PFS) was calculated based on the date of initial surgery to the date of progression/recurrence or last known contact. The cutoff for "high" and "low" TAGLN is the mean value of TAGLN expression score. Overall survival (OS) was calculated based on the date of initial surgery to death or last known contact. The status of PFS and OS regarding TAGLN expression was estimated with the Kaplan-Meier methods using GraphPad Prism 6 software, and the log-rank test was used to evaluate the statistical significance. Patient characteristics are summarized in Supplementary Tables 4, 5, 6, 7 and 8.

Preparation of polyacrylamide (PA) gels substrates
The preparation of polyacrylamide (PA) gels substrates was described previously [18]. The Young's modulus of PA gel substrates were measured using an atomic force microscopy (SPM9700 scanning probe microscope, Shimadzu).

Tissue preparation for AFM measurements of ECM stiffness
Tissues for AFM measurement were prepared as Plodinec et al. described [19]. Both human ovarian cancer tissues and tumors removed from xenograft models were immediately immersed in cold PBS and kept at 4°C until an experiment commences. These tissues could be kept up for three days before the AFM measurement. Dissect the sample into 1-3 mm thick slices as parallel and smooth as possible for high quality measurements. To attach the sample to a hard substrate during the measurement, we used two-component epoxy glue (5 min hardening time). Apply a thin layer of mixed epoxy glue onto the glass of 12 mm diameter and dry briefly the bottom side using a lint-free tissue paper, then attach the sample onto the glue. After the epoxy was hard enough and the tissue sample was stable attached, we transferred the tissue-attached glass to conduct the AFM measurement.

Atomic force microscopy measurements
All AFM measurements were performed using an atomic force microscope (SPM9700 scanning probe microscope, SHIMADZU, Japan) in Huazhong University of Science and Technology. The AFM measurements were conducted as previously described [18]. Ten different positions per tissues were measured and averaged.

Animal studies
Female BALB/c-nu mice (5 weeks old) were purchased from the Animal Experimental Center (Shanghai, China). All animal experiments were performed with the approval of the Committee on the Ethics of Animal Experiments in Hubei Province (Institutional Review Board ID: TJ-A20160102). All mice were housed under pathogen-free conditions in the Animal Research Center of Tongji Hospital at 22 C with 12-hour light/dark cycles and free access to water and food. An orthotopic model of ovarian cancer was established as we described previously [20]. Briefly, 2 × 10 6 cells in 25 µL PBS were injected orthotopically under the ovarian bursa. Seven days after cells injection, mice were randomly assigned to treatment groups to avoid treatment bias. For BAPN treatment, BAPN was given i.p. (50 mg/kg) every 2 days for 4 weeks, or saline was used as a control. For dasatinib treatment, dasatinib was given i.p. (15 mg/kg) every 2 days for 4 weeks, or DMSO was used as a control. For TAGLN in vivo experiment, 2 × 10 6 SK-OV-3 cells ectopically knockdown of TAGLN (sh-TAGLN, n = 6 per group), or respective appropriate control (sh-Control, n = 6 per group) were injected orthotopically under the ovarian bursa BALB/c-nu mice. After weeks of injection, the mice were anaesthetized with 1 % pentobarbital sodium, and imaged with the IVIS SPECTRUM system (Caliper, Xenogen, USA). Total flux (photons/s) of xenografts was analyzed utilizing Living Image version 4·3·1 software. After executing the mice, tumors were resected, prepared for AFM measurement, frozen or fixed with paraformaldehyde.

Statistical analysis
All data were presented as the mean ± SEM. Two experiments groups were compared by using a two-tailed paired t-test for paired data or a two-tailed unpaired ttest for unpaired data after confirming the normality of the data. Analysis of variance was used to compare differences among multiple groups (post hoc test: Turkey's multiple comparisons test). Pearson correlation was used to analyze a potential correlation between two different experimental parameters. GraphPad Prism 6.0 software was used to conduct the statistical analysis of the data. P values less than or equal to 0.05 were considered to be significant.

Cell track
Cells were cultured on soft and stiff PA gels (at sixwells) at a density of 8000 cells per well for over 24 h. Cell movement was recorded using live-cell imaging (laser scanning confocal microscopy, IX83 living cell imaging system, Olympus FV1000, Tokyo, Japan) at 6-min intervals over a 16-h period and multiple position imaging was gathered. At least three independent experiments were performed; 20-40 cells were analyzed in each experiment. Cell movement was analyzed using Chemotaxis and Migration Tool plug-in (Ibidi) of ImageJ software. Total length of the cell track (in µm) was generated for multiple cells, averaged and normalized to cell tracks from cells cultured on soft PA gels.

Other methods
Masson trichrome and picrosirius red staining, Immunohistochemistry, Quantitative real-time PCR, Western blot analysis, Transwell assay, Rho GTPase activity, Gene set enrichment analysis and transcriptomic array were in supplementary methods.

Ovarian cancer metastases exert higher stiffness than primary tumors
The surgeons frequently found that the peritoneal metastases of OC harbored higher stiffness than their primary counterparts. To confirm it, we utilized atom force microscope (AFM) to quantify the stiffness of eight paired primary tumors and omental metastases of EOC, showing that the Young's modulus of metastases were higher than that of paired primary tumors [ Fig. 1A]. Besides, increased mechanosignaling expressions including phosphorylated focal adhesion kinase (pFAK residue Tyr397) and myosin light chain 2 (pMLC2 residue Ser19) were detected in metastases when compared with primary tumors [Fig. 1B].
ECM deposition and collagen crosslinking are proven to contribute to increased tissue stiffness [9]. As revealed by trichrome and picrosirius staining, the stroma of metastases contained more fibrillar collagen than primary tumors [ Fig. 1C and D]. Besides, higher expression of collagen Ι was observed in metastases than in primary tumors [ Fig. 1E]. Moreover, bioinformatics analysis of paired and unpaired public gene data (GSE30587 and GSE2109) confirmed the elevated expression of major fibrillar collagens signature [21] in the metastases [ Fig. 1F and Supplementary Table 3]. Lysyl oxidase (LOX), a key enzyme that initiates the crosslink of collagen, was more abundantly expressed in the metastases than in primary tumors [ Fig. 1G], which was confirmed by the analysis of public gene data [ Fig. 1H]. To conclude, these data demonstrated that the metastases of EOC were stiffer than primary tumors, and the increased stiffness could be attributed to the upregulation of desmoplasia in the metastases.

Matrix stiffness modulates ovarian cancer progression and activates Src gene and RhoA/ROCK pathway
Tumor cells from the metastases were reported to have more aggressive biology than those from the primary sites [4]. To explore whether matrix stiffness could modulate EOC progression, the collagen-coated polyacrylamide hydrogel (PA gel) system with calibrated Young's modulus of 0.25 ± 0.05 kPa and 10.5 ± 0.02 kPa was prepared to mimic the soft and stiff substrates in vitro [22]. To further investigate the effects of matrix stiffness on OC progression in vivo, we conducted an orthotopic OC xenograft and used β-aminopropionitrile (BAPN), a classical LOX inhibitor that was commonly used in interrogating mechanical properties of ECM [13], to decrease the tumor stiffness in vivo. By activating mechanosignaling molecules such as Src family kinases [23], cells perceive and respond to physical signals, thereby transforming their biological behaviors [9]. Significantly, higher expressions of p-Src Try416 were observed in OC cells cultured on stiff substrates [ Fig. 2 L, supplementary Fig. 3A]. Consistently, tumors of mouse models that received BAPN displayed decreased expression of p-Src Try416 [supplementary Fig. 3B]. Rho (Ras homology) GTPases family members and the downstream factor ROCK has been demonstrated important in cell migration and mechanotransduction process [24]. Markedly, OC cells cultured on stiff substrates harbored higher expressions of RhoA, particularly activated RhoA, and downstream ROCK1 and ROCK2 compared with cells cultured on soft substrates [ Fig. 2

TAGLN is identified as a stiffness-regulated gene in ovarian cancer
To investigate the molecular mechanisms that OC cells respond to the mechanical cues, we performed transcriptomic microarray analysis of SK-OV-3 cells cultured on soft and stiff substrates. Gene set enrichment analysis (GSEA) indicated the enrichment of "CYTOSKEL-ETON" and "ACTIN_FILAMENT-BINDING" gene sets in cells cultured on stiff substrates [supplementary Fig. 4], further demonstrated the impact of tissue stiffness on cytoskeleton, as we showed in Fig. 2B. Cytoskeleton alternation is an important cellular response to matrix stiffness [8]. Within the pool of differentially expressed genes, TAGLN, an actin cross-linking/gelling protein sensitive to morphologic alternations, was drastically upregulated upon mechanical stimulation [ Fig. 3 A], even most significantly upregulated when cells were cultured on substrates with greater stiffness (40 kPa) [supplementary Fig. 5A]. Significantly, higher TAGLN expression was detected in OC cells cultured on stiff substrates, compared with cells cultured on soft substrates both at the mRNA level [ Fig. 3B, supplementary Fig. 5B] and the protein level [ Fig. 3C-D]. We also detected higher expression of other upregulated genes, such as THBS1, OXTR, in cells cultured on stiff substrates [supplementary Fig. 5C], but they did not show association with cytoskeleton, thus, we identified TAGLN to continue the study. To further elucidate the mechanosensitivity of TAGLN, we treated OC cells with latrunculin A, an inhibitor of actin assembly and polymerization, which resulted in notably weakened TAGLN expression [Fig. 3E]. The inhibition of nonmuscle myosin with Blebbistatin could also downregulate the expression level of TAGLN [ Fig. 3 F]. Besides inhibition of the crosslink of collagen, BAPN could also decrease expression of fibrillar collagen [ Fig. 3G]. For the in vivo experiment, tumors of BAPN-treated mouse models, which has been proved softer, showed significantly less expression of TAGLN than saline-group [ Fig. 3H].
In addition, TAGLN expression of omental metastases of OC patients was significantly greater than that of  Fig. 3I], which was also verified using public paired (GSE30587) and unpaired (GSE2109) profiling data [ Supplementary Fig. 6]. Moreover, analysis of TCGA database indicated that the expression of TAGLN positively correlated with that of COL1A2 and LOX, two pivotal genes involved in matrix stiffness [ Fig. 3 J]. To summarize, these results unraveled that TAGLN was a stiffness-regulated gene in OC that might play an important role in stiffness-related OC progression.

TAGLN mediates stiffness-regulated ovarian cancer progression and correlates with poor patient prognosis
To determine the impact of TAGLN in stiffnessregulated OC progression, we first successfully estab-   Furthermore, upregulating TAGLN levels using an expression plasmid [Fig. 4D, supplementary Fig. 7D] enhanced the capabilities of migration and invasion of both OC cells cultured on soft and stiff substrates [ Fig. 4E-F, supplementary Fig. 7E-F]. Besides, cell proliferation was higher when TAGLN was upregulated [ Fig. 4G] and lower when TAGLN was downregulated in [Fig. 4H] OC cells.
To further investigate the role of TAGLN in OC progression in vivo, we generated an orthotropic OC xenograft by injection of SK-OV-3-ip3-luc cells, in which TAGLN expression was downregulated through lentivirus transfection [ Fig. 4I]. We observed that silencing TAGLN using shRNA restrained tumor progression [ Fig. 4J and K] and lowered the metastasis rate [ Fig. 4L], suggesting that TAGLN mediated stiffness-regulated OC progression.
Given the importance of TAGLN in mediating stiffness-regulated OC progression, we explored whether TAGLN expression correlated with OC prognosis. Analysis of pan-cancer Genomic Profiles indicated that increased TAGLN was associated with poor survival in most epithelial cancers [25], especially OC [supplementary Fig. 8A]. Besides, among 132 patients with EOC, we found that high level of TAGLN protein (defined by IHC) was significantly associated with both shortened PFS and OS [  Fig. 8B and C]. Thus, we confirmed the association between high TAGLN expression and poor prognosis of OC.

TAGLN and Src activation form a regulation loop
Given that the activation of Src was pivotal in mechanotransduction of OC cells [27], we hypothesized that Src modulated the expression of TAGLN. To investigate whether alternations of Src activation could impose influences on the expression level of TAGLN, we treated cells with Src inhibitors (dasatinib or Src inhibitor 1) or constitutively active Src [CA-Src (Try416)]. Intriguingly, we found that inhibition of Src evidently suppressed the expression of TAGLN [ Fig. 5A and B]. Overexpression of constitutively active Src (CA-Src (Try416) with plasmid markedly upregulated TAGLN expression [ Fig. 5 C]. To further illustrate the effect of Src activation on TAGLN expression in vivo, we treated OC xenograft with dasatinib. Compared with those in DMSO-treated group, the expression level of TAGLN and phosphorylation of Src in mice treated with dasatinib were significantly curtailed [ Fig. 5D-E], indicating that Src activation regulated TAGLN expression in vivo. To decipher how Src activation affects TAGLN expression, we investigated YAP, another key component of mechanosignaling pathway [15]. The inhibition of Src activation using dasatinib or Src inhibitor 1 could effectively suppress YAP expression [supplementary Fig. 9 A-B], moreover, silencing YAP could significantly suppress TAGLN expression both in the mRNA and protein levels [supplementary Fig. 9C-D]. Thus, we proposed that Src might indirectly regulate TAGLN expression by harnessing YAP.
Next, we investigated whether TAGLN could regulate Src activation. Silencing TAGLN markedly reduced Src activation [ Fig. 5F], whereas overexpression of TAGLN noticeably enhanced the activation of Src in OC cells [ Fig. 5G]. Consistently, we noted a significant reduction of phosphorylation of Src in TAGLN-silenced xenograft tumors [ Fig. 5H], signifying that TAGLN could regulate Src activation in vitro and in vivo. This regulation loop was interesting, but how could TAGLN regulate Src gene activation, which was an mechanosensing molecular and supposed to be an up-streamer of TAGLN? During our experiment, we noticed that cells silencing TAGLN expression showed slimmer morphology [ Fig. 5I]. and decreased cytoskeleton expression [ Fig. 5J], indicating that TAGLN could modulate cellular mechanical environment, thereby regulating the mechanosensing molecule Src.
To demonstrate the correlation between Src and TAGLN in OC tissues, we analyzed phosphorylate-Src (Try416) and TAGLN expression in 10 primary tumors of HGSOC by western blot analysis. We also analyzed phosphorylate-Src (Try416) and TAGLN expression in 33 primary tumors of HGSOC patients by IHC staining. Pearson's correlation confirmed the positive relationship between phosphorylate-Src (Try416) and TAGLN protein expression (r = 0.7403) [Fig. 5K]. Besides, we observed significantly lower phosphorylate-Src (Try416) expression in TAGLN-low group and higher expression in TAGLN-high group [ Fig. 5L].
To sum up, our results demonstrated that mechanosensitive Src activation and TAGLN expression formed a regulation loop.
TAGLN regulates ovarian cancer progression through RhoA/ROCK pathway Since RhoA/ROCK pathway was pivotal in cell motility and was activated upon mechanical cues in OC, we explored whether it was regulated by TAGLN to abet OC progression. Knockdown of TAGLN restrained RhoA/ ROCK expression [Fig. 6A]. Moreover, overexpression of TAGLN remarkably increased RhoA/ROCK expression [ Fig. 6B]. To verify it in vivo, we analyzed ROCK1, ROCK2 and RhoA expression in TAGLN-knockdown tumors, and observed that expressions of these proteins were decreased in sh-TAGLN group in contrast to shcontrol group [ Fig. 6 C-D]. Moreover, we found that both RhoA inhibitor (Rhosin) and ROCK inhibitor (Y27632) could inhibit the migration and invasion of OC cells, particularly for cells overexpressing TAGLN [ Fig. 6E-F]. Taken together, these data demonstrated that TAGLN could regulate OC progression through RhoA/ROCK pathway.

Discussion
In this study, we confirmed that omental metastases harbored higher stiffness and level of desmoplasia compared with primary tumors of EOC and demonstrated that matrix stiffness could modulate OC progression. Notably, we identified that TAGLN was a novel mechanosensitive gene that formed a regulation loop with Src, a pivotal mechanosignaling molecule. More importantly, TAGLN was revealed to mediate stiffness-regulated OC progression through RhoA/ROCK signaling pathway. Our study provided the evidence that targeting ECM stiffness could hamper OC progression, suggesting that TAGLN could be a promising therapeutic target for counteracting OC.
Increased matrix stiffness during tumor progression has been shown to contribute to malignant phenotypes  μm. E-F Migratory (E) and invasive (F) ability of cells treated with Rhosin or Y27632 after transfected with control vector or TAGLN plasmid (*P < 0.05, **P < 0.01, ****P < 0.0001). G A graphical illustration of the molecular signaling events involved in stiffness-regulated OC progression. TAGLN mediated stiffness-regulated OC progression through RhoA/ROCK pathway of multiple cancers [7,13,21]. Here, by revealing that omental metastases of EOC were higher in stiffness than primary tumors, we uncovered the OC-promoting role of higher matrix stiffness and indicated the correlation between tissue mechanics and OC behaviors. We demonstrated that stiffer ECM could abet OC progression, suggesting that preventing or reversing tumor stiffening is a potential therapeutic strategy [8]. Targeting ECM stiffness using BAPN in our experiments significantly hampered OC progression, however, the high toxicity of BAPN in clinical trials precludes the clinical applications [28]. Alternative approaches such as inhibiting the LOX transcription factor have been proposed [29]. We revealed that the metastases of OC underwent more desmoplasia, resulting in higher stiffness, but the underlying mechanisms remain unclear. In the future, the investigations of the increased desmoplasia in metastases could provide other druggable targets for eliminating ECM stiffness.
Besides preventing ECM stiffening, an alternative method to overcome the adverse effects of microenvironment stiffening is to curb the cellular response to increased matrix mechanics that contribute to tumor malignant phenotypes [30]. Several researches have suggested the therapeutic effects of small molecule inhibitors that target mechanical signals in tumors, such as FAK inhibitor and YAP inhibitor [31,32]. For example, small-molecule inhibitors that directly inhibit YAP/TAZ has been used as mechano-based interventions [31] and showed therapeutic effect in restraining ovarian cancer progression [33,34]. Our previous study showed that targeting YAP using peptide 17 attenuated OC progression through down-regulation of PI3K/Akt/mTOR pathway [33]. Verteporfin, a suppressor of YAP-TEAD complex, has been proved to effectively reduce proliferation and inhibit the growth of OC cells in vitro and in vivo [34]. In our research, we also demonstrated that YAP could mediate Src activation regulated TAGLN expression, further revealed its role in ovarian cancer.
Notably, our study identified a novel mechanosensitive gene, TAGLN, which was demonstrated to form a regulation loop with Src activation. Src is a wellacknowledged mechanosignaling molecule that becomes activated upon physical signal [27]. Our study demonstrated that Src activation regulated TAGLN expression, therefore, we think Src was an up-streamer molecule of TAGLN. We discovered that TAGLN could in turn govern Src stimulation because silencing TAGLN reshaped the cellular morphology and cytoskeleton expression, could resulting in cellular tension alternation, therefore showed influence on Src activation. Importantly, we revealed that TAGLN could mediate stiffness-regulated progression of ovarian malignancy through the regulation of RhoA/ROCK pathway and blocking TAGLN could diminish OC progression in the experiments. However, there is no approved or explored drug that targets TAGLN currently. Therefore, TAGLN holds great promise as a therapeutic target of OC.

Conclusions
Our study highlighted the pivotal role of tissue mechanics in OC progression and supported that targeting ECM stiffness emerged as a promising strategy for OC treatment. Specifically, we revealed that TAGLN formed a regulation loop with Src activation and mediated stiffness-regulated OC progression through the regulation of RhoA/ROCK pathway, proposing that TAGLN was promising as a novel therapeutic target for suppressing OC progression.