Abnormal tensional cellular homeostasis is now considered a hallmark of cancer. Despite this, the origin of this abnormality remains unclear. In this work, we investigated the role of tissue transglutaminase 2 (TG2, also known as TGM2), a protein associated with poor prognosis and increased metastatic potential, and its relationship to the EGF receptor in the regulation of the mechanical state of tumor cells. Remarkably, we observed a TG2-mediated modulation of focal adhesion composition as well as stiffness-induced FAK activation, which was linked with a distinctive increase in cell contractility, in experiments using both pharmacological and shRNA-based approaches. Additionally, the increased contractility could be reproduced in non-malignant cells upon TG2 expression. Moreover, the increased cell contractility mediated by TG2 was largely due to the loss of EGFR-mediated inhibition of cell contractility. These findings establish intracellular TG2 as a regulator of cellular tensional homeostasis and suggest the existence of signaling switches that control the contribution of growth factor receptors in determining the mechanical state of a cell.
Cell contractility plays an important role in the regulation of numerous cellular processes (Bordeleau and Reinhart-King, 2016; Wei and Yang, 2016). Interestingly, the range of contractility displayed by cells is known to vary widely depending on cell lineage and function (Engler et al., 2006; McBeath et al., 2004; Fu et al., 2010). However, in several pathologies, including cancer, cells can develop an abnormal contractile phenotype (Huynh et al., 2011; Steucke et al., 2015; Cohen-Naftaly and Friedman, 2011; Lam et al., 2016; Dembo and Wang, 1999). Indeed, there is ample evidence demonstrating that tumor cells are more contractile than their normal counterparts (Paszek et al., 2005; Butcher et al., 2009). Even more so, increased cell contractility is known to correlate with metastatic potential (Kraning-Rush et al., 2012; Butcher et al., 2009). However, it remains unclear what mechanisms cells use to alter their contractile state, and by extension, what mechanisms lead to the tumor cell contractile phenotype.
The ability of cells to properly sense the mechanical features of the underlying extracellular matrix (ECM) depends mainly on focal adhesions (FAs), which act as mechanical couplers (Wei et al., 2008; Provenzano et al., 2009; Trichet et al., 2012). The interplay between intracellular contractility and the ECM mechanical properties at FAs allows mechanotransduction events to take place that enable the cell to strike an exquisite balance between all the intracellular and extracellular forces involved (Bordeleau et al., 2012; Huang and Ingber, 2005; Provenzano and Keely, 2011). Increased cell contractility can perturb this homeostasis, which can result in increased sensitivity to matrix stiffness and downstream activation of mechanosensing pathways, as is exemplified by FAK (also known as PTK2) activation in tumors and highly contractile tumor cells (Paszek et al., 2005; Acerbi et al., 2015). It is increasingly evident, however, that other players are critical in this balance, most notably through the contribution of tyrosine kinase growth factor receptors (TKRs). While the crosstalk between integrin and TKRs has been known for several years (Lamalice et al., 2007; Soung et al., 2010; Asthagiri et al., 2000), more recent evidence has uncovered a more complex relationship, where TKRs such as EGFR, HER2 (also known as ERBB2) and AXL actively contribute to mechanosensing events (Saxena et al., 2017; Muhamed et al., 2016; Yang et al., 2016). However, their contributions to cell contractility and FA dynamics appear to be context specific (Saxena et al., 2017). For example, EGFR can either increase or decrease the activation of pro-contractility signaling pathways depending on cell type and matrix stiffness (Ramis et al., 2012; Saxena et al., 2017; Grasset et al., 2018). Such cell-dependent modulation of cell contractility points to a control switch that can alter the interplay between cell contractility and ECM stiffness.
Tissue transglutaminase 2 (TG2, also known as TGM2) is a member of the transglutaminase family of proteins that is best known for Ca2+-dependent cross-linking of proteins. Interestingly, TG2 is a multi-functional protein that has GTP-binding and GTPase activity, acyl transferase activity and processes scaffolding abilities (Gundemir et al., 2012). Several studies have linked TG2 expression to tumor cell invasiveness and drug resistance (Lee et al., 2018; Meshram et al., 2017; Kumar et al., 2010). Additionally, extracellular TG2 has been shown to crosslink ECM components, such as collagen, resulting in increased matrix stiffness (Fortunati et al., 2014; Stephens et al., 2004). Moreover, extracellular TG2 can bind to integrins and enhance their affinity to fibronectin in fibroblasts (Akimov et al., 2000). TG2 expression is sufficient to promote a more mesenchymal cell phenotype (Ayinde et al., 2017). Interestingly, intracellular TG2 can act as a scaffold for signaling molecules to potentiate the ERK pathway (Li et al., 2010). Despite this, it remains unknown if TG2 contributes to the contractile phenotype of invasive cancer cells.
In this study, we use both 2D and 3D in vitro systems to show that TG2 actively contributes to the tumor cell contractile phenotype. Specifically, we use malignant MDA-MB-231 cells, which are known to express high levels of TG2 (Mehta et al., 2004), or stable TG2-knockdown MDA-MB-231 cells (shTG2) and their counterparts (PLKO), as well as MCF10a cells transfected with TG2–GFP as cellular models. We distinguish between intracellular and extracellular TG2 pools by using cell-permeable (monodansylcadaverine, MDC) or impermeable (T101) TG2 inhibitors. Our results indicate that TG2 regulates FA dynamics, maturation and mechanosensing. Moreover, traction force microscopy (TFM) reveals that TG2 contributes to increased tumor cell contractility. In addition, changes in cell contractility are linked to both altered activation and spatiotemporal localization of RhoA. Finally, we further show that TG2-mediated cell contractility occurs through an EGFR-dependent mechanism. Overall, our results demonstrate that TG2 contributes to tumor cell contractility and that this happens indirectly by decreasing EGFR-mediated inhibition of contractility.
TG2 controls focal adhesion dynamics and signaling
Since TG2 has been reported to modulate cell adhesion in some cell types in an ECM-dependent fashion (Mangala et al., 2007; Akimov et al., 2000; Fortunati et al., 2014), we investigated the contribution of TG2 to FA assembly and signaling in MDA-MB-231 cells through pharmacological and shRNA-based approaches (Fig. 1). To distinguish between intracellular and extracellular TG2 functions, we used either cell-permeable MDC, which inhibits both TG2 pools, or cell-impermeable T101, which inhibits extracellular TG2. Interestingly, inhibition of TG2 with MDC led to decreased number of FAs compared to the control and T101, while selective inhibition of extracellular TG2 with T101 did not produce any visible changes compared to the control (Fig. 1A,B), indicating that intracellular TG2 regulates FAs. Moreover, a major difference was seen in the levels of FAK autophosphorylation at Y397 with decreased signal in MDC-treated cells compared to control and T101-treated cells (Fig. 1B). Accordingly, shTG2 cells showed similar alterations in FA numbers, localization at the cell periphery and phosphorylated FAK signal compared to PLKO cells (Fig. 1C,D).
To further investigate the effects of TG2 on signaling events and mechanosensing, MDA-MB-231 cells were seeded on compliant (1 kPa) or stiff (10 kPa) collagen-coated polyacrylamide substrates and FAK activation was assessed. The use of polyacrylamide substrates allowed us to set the substrate stiffness independently of the ability of TG2 to crosslink collagen. In line with our FA observations, inhibition of the extracellular TG2 pool with T101 did not block the increased FAK phosphorylation that was mediated by increased substrate stiffness compared to the control, while MDC did abrogate the stiffness-mediated FAK activation (Fig. 2A,B). Similarly, shTG2 cells seeded on stiff matrices only showed marginal FAK activation compared to the substantial phosphorylation seen in PLKO cells (Fig. 2C,D). These results indicate that TG2 influences mechanotransduction and the ability of cells to sense matrix stiffness.
Assembly and maturation of FAs require the recruitment of several structural proteins for mechanotransduction and cell contractility to occur (Pasapera et al., 2010; Galbraith et al., 2002). Interestingly, residency time of FAK within FAs is known to be correlated with FA turnover, and is considered to be a good proxy of FA dynamics (Hamadi et al., 2005; Bordeleau et al., 2010). Using FAK–GFP-transfected MDA-MB-231 cells, we investigated whether TG2 influences FA dynamics. Notably, we observed a longer recovery half-life for FAK–GFP in FAs in cells treated with MDC compared with both the untreated controls and T101-treated cells (Fig. 3A,B). Furthermore, the same experiment performed in PLKO and shTG2 cells transfected with FAK–GFP revealed a similar trend. Namely, the recovery half-life of FAK–GFP in FAs was longer in shTG2 cells compared with the PLKO cells (Fig. 3C,D). To validate whether this observation translated to altered FA dynamics, we evaluated FA maturation through the presence of zyxin. Zyxin is found mainly in the most mature FAs, and its accumulation is known to be mechanoresponsive (Zaidel-Bar et al., 2003; Uemura et al., 2011). Interestingly, the amount of zyxin was found to be significantly lowered in FAs of cells that were treated with MDC, while a substantial zyxin signal was seen in control and T101-treated cells (Fig. 3E,F, see Fig. S1A for individual zyxin and vinculin signal quantification). Similarly, very few zyxin-positive FAs were present in shTG2 cells compared to PLKO cells (Fig. 3G,H; Fig. S1B). In addition, we stained for the presence of zyxin within FAs during early cell spreading. In this situation, all of the observable FAs were enriched in vinculin after both 15 and 30 min of cell spreading (Fig. S2A,B). However, early cell spreading was impaired at 15 min in MDC-treated cells compared to the control and T101 treatment (Fig. S2C). Similarly, early cell spreading at 15 min was decreased in shTG2 cells compared to PLKO cells (Fig. S2D). Overall, our data indicate that intracellular TG2 modulates FA assembly, maturation and signaling.
Intracellular TG2 modulates cell contractility in both 2D and 3D via RhoA
Increased cell contractility is a major element of altered tensional homeostasis of tumor cells (Paszek et al., 2005; Kraning-Rush et al., 2012). In addition, increased contractility in tumor cells is known to drive increased FAK signaling (Paszek et al., 2005). Thus, we utilized traction force microscopy (TFM) to investigate whether TG2 regulates cell contractility. Consistent with our previous results on mechanosensing, the level of traction forces generated by MDA-MB-231 cells treated with MDC was significantly decreased compared to the control while T101 had no effect on the traction forces (Fig. 4A,B). Moreover, shTG2 cells showed a similar decrease in traction force compared to their control (Fig. 4C,D), indicating that the force decrease was indeed TG2 mediated. To verify whether TG2 contributes to cellular tensional homeostasis, we extended our TFM analysis to MCF10a cells transfected with TG2–GFP. Notably, expression of TG2–GFP in MCF10a cells increased traction forces compared to eGFP-expressing MCF10a control (Fig. 4E,F). Furthermore, TG2 inhibition with MDC abrogates the increase in traction forces in TG2–GFP cells while it had no effect in cells expressing eGFP (Fig. 4E,F). Interestingly, the levels of traction forces measured in TG2–GFP MCF10a cells were similar to those observed in MDA-MB-231 cells (Fig. 4B,F). Overall, these results show that TG2 can contribute to increased contractility of tumor cells.
RhoA is one of the main regulators of cell contractility and is essential for FA maturation and mechanosensing (Provenzano and Keely, 2011; Parsons et al., 2010), and it is known to contribute to increased tensional homeostasis in tumor cells (Paszek et al., 2005). Consequently, we found that RhoA activity was decreased when TG2 was inhibited with MDC compared with T101 or the untreated control (Fig. 5A), indicating that TG2-mediated cell contractility is RhoA dependent. To better characterize the link between TG2 and RhoA, we tracked the spatiotemporal behavior of RhoA in cells transfected with the RhoA probe rGBD–GFP. Strikingly, two main active RhoA-containing structures were observed, namely lamellipodia-like membrane protrusions, having forward motions, and spindle-like protrusions with rearward motion relative to the cell center (Fig. 5B, see Movies 1 to 9). While no apparent difference was found in the number of cells presenting the lamellipodia-like structures, only a small percentage of cells treated with MDC showed the spindle-like protrusions compared to those subjected to T101 treatment or untreated control cells (Fig. 5B,C), clearly indicating that localization of active RhoA was altered after intracellular TG2 inhibition. Overall, these observations further support our findings that intracellular TG2 modulates cell contractility.
Cells reside in a complex 3D environment in vivo, and previous work has shown that cells can behave differently in 3D compared to 2D (Mabry et al., 2016; Luca et al., 2013; Chitcholtan et al., 2013). With this in mind, we extended our analysis to a 3D collagen scaffold and performed collagen compaction assays to test 3D cell contractility. TG2 inhibition with MDC lowered the amount of collagen compaction measured over the course of 4 days compared to both the control and T101 treatment (Fig. 6A,B). To further quantify contractility in 3D, we used confocal reflectance microscopy to assess the changes and remodeling of the collagen fibrillar architecture. 3D confocal volumetric images showed that TG2-mediated collagen compaction also affected the apparent thickness of the collagen gel (Fig. 6C). In addition, closer inspection of the collagen surrounding the cells further indicated that TG2 inhibition with MDC, but not with T101, prevented cell-mediated collagen remodeling (Fig. 6D). In fact, quantification of the half distance of radial compaction of the collagen surrounding the cell, a metric we have shown to be related to 3D cell contractility (Kraning-Rush et al., 2011), indicated that intracellular TG2 reduced the distance over which cells pull on the collagen fibers (Fig. 6E). Notably, the more contractile cells pull more collagen toward them, which results in the increased reflectance signal as a function of radial distance. Considering that these metrics are indirect measurements of cell contractility, we further expanded our assessment of contractility using a quantitative polarization (QPOL) microscopy-based method we have recently developed (Wang et al., 2018). The QPOL method provides a direct, 3D-culture compatible and linearly proportional readout of cell contractility by measuring optical retardance (Wang et al., 2018). Notably, our QPOL retardance measurements performed on cells embedded in collagen for 24 h indicate that the MDC treatment significantly lowered the 3D cell contractility compared to both the control and the T101 treatment (Fig. 6F,G). Interestingly, these results were similar to the data that we obtained using TFM, with MDC treatment resulting in a 1.62-fold reduction in contractility in 3D compared to a 1.53-fold decrease in 2D compared to their respective controls (Figs 4B and 6G). Moreover, 3D collagen-embedded shTG2 cells showed a similar decrease in the QPOL-measured retardance compared to the PLKO cells (Fig. 6H,I) in line with our previous TFM results. Overall, these results show that intracellular TG2-mediated cell contractility also occurs within 3D scaffolds and enables increased ECM remodeling.
TG2 effects on cell contractility occur through a EGFR- dependent switch
We have previously shown that TG2 acts as a scaffold, and its interaction with Src and keratin intermediate filaments can enhance signaling from EGFR (Li et al., 2010). Conversely, both Src and keratins have been implicated in the regulation of cell mechanics (Bordeleau et al., 2012; Matthews, 2006). In addition, recent work suggests that EGFR directly contributes to mechanosensing and the control of cellular mechanics (Saxena et al., 2017; Muhamed et al., 2016). In this context, we asked whether TG2 modulation of cell contractility was a consequence of EGFR-mediated signaling. To this end, we treated both shTG2 and PLKO MDA-MB-231 cells with the EGFR-specific inhibitor AG1478. Interestingly, EGFR inhibition did not significantly change FAK activation in PLKO cells on stiff (10 kPa) substrates (Fig. 7A). In contrast, EGFR inhibition increased FAK activation in shTG2 cells (Fig. 7A), suggesting that TG2 may act as a switch that turns off an EGFR-mediated inhibition of FAK activation. FAK activation in both PLKO and shTG2 cells could be blocked and brought to proportionally equivalent levels with a broad spectrum Src inhibitor (PP1) (Fig. 7A–C). Of note, ERK1/2 phosphorylation was blocked by Src inhibition but not EGFR inhibition in our experimental conditions, similar to results shown by others (Zhang et al., 2008; 2011; Filardo et al., 2000). In addition, when we investigated the effects of EGFR inhibition on the inclusion of zyxin in vinculin-positive FAs, we observed that the average levels of zyxin that were associated with vinculin were much higher with AG1478, with both the PLKO and shTG2 cells having equivalent zyxin to vinculin levels (Fig. 7D; see Fig. S3 for individual zyxin and vinculin signal quantification).
To confirm whether TG2 could modulate cell contractility through EGFR, we expanded our analysis to both 3D and 2D cellular force measurements. Notably, in the case of 3D collagen compaction, use of the EGFR inhibitor in PLKO cells did not lead to any measurable change in collagen compaction after 4 days, while collagen compaction was significantly increased in shTG2 cells treated with AG1478 compared to control (Fig. 7E). Consistent with our data on FAK activation, Src inhibition blocked collagen compaction in both cases. Furthermore, 2D TFM experiments revealed the same TG2-dependent EGFR-mediated effect on cell traction forces. Specifically, only the shTG2 cells exhibited increased traction forces after AG1478 treatment compared to control (Fig. 7F). In addition, the traction forces measured in shTG2 cells treated with AG1478 were similar to the forces measured in both control and AG1478-treated PLKO cells. In the 2D case, Src inhibition led to a decrease in traction forces generated by PLKO cells to levels similar to those found in shTG2 cells. Together, these data indicate that contribution of TG2 to the increased tumor cell contractility occurs by inhibiting a negative contractility feedback mediated by EGFR.
Here, we provide evidence that intracellular TG2 contributes to cellular tensional homeostasis and, to our knowledge, the first direct evidence of a mechanism that can alter the contribution of TKRs to cell contractility. Our data indicate that intracellular TG2 influences FA formation and maturation. We demonstrate that TG2 can enhance cell response to increased matrix stiffness, and that TG2 contributes to increased cancer cell contractility in both 2D and 3D, a hallmark of aggressive tumor cells. Accordingly, we observe a correspondingly higher TG2-mediated RhoA activation, as well as altered spatiotemporal behavior of active RhoA. We further show that the altered FAK activation and increased cell contractility mediated by TG2 occurs through crosstalk from EGFR (Fig. 8). Together, these results establish intracellular TG2 as an important regulator of the crosstalk between integrins, growth factor receptors and cell contractility.
Cells are known to actively probe the physical properties of surrounding matrices, and as such can reach a balance between the mechanical features of the ECM and their own contractile machinery (Butcher et al., 2009; Levayer and Lecuit, 2012; Berrier and Yamada, 2007). Cells interacting with stiffer ECM usually display higher integrin and FAK activation (Wei et al., 2008; Paszek et al., 2005; Levental et al., 2009). In turn, cell contractility is essential to this mechanosensing and the subsequent maturation process of FAs. Interestingly, FAK residency in FAs has been shown to be proportional to FA turnover (Bordeleau et al., 2010; Hamadi et al., 2005). In this context, our results on longer residency time of FAK upon TG2 inhibition are consistent with impaired FA dynamics. Notably, zyxin is found mainly in mature FAs (Geiger et al., 2001), suggesting that the TG2-mediated accumulation of zyxin in FAs is indicative of an altered maturation processes. Importantly, since both zyxin accumulation in FAs and FAK activation are known to be force dependent (Uemura et al., 2011; Paszek et al., 2005; Wei et al., 2008), it is likely that the TG2-mediated modulation of FAs is a consequence of its effects on RhoA activation and cell contractility. Moreover, the fact that intracellular TG2 has not been shown to localize specifically at FAs supports the idea of a more indirect role for TG2 in modulating FA dynamics.
The idea that growth factor receptors play a role in controlling mechanical homeostasis is a recent concept that highlights the inside-out versus outside-in events involved in fine tuning cell contractility (Muhamed et al., 2016; Saxena et al., 2017). However, evidence points to EGFR as functioning as both a cell contractility inducer and inhibitor. Notably, EGF stimulation induces a transient increase in cell contractility, which will then decrease after a short time (Schneider et al., 2009; Saxena et al., 2017). Interestingly though, EGFR contributes to mechanosensing and increased cell contractility only when the underlying ECM is sufficiently stiff (Saxena et al., 2017). In contrast, another group has reported that EGFR inhibition can instead lead to RhoA activation (Ramis et al., 2012), which is indicative of increased cell contractility. In this context, our findings offer a plausible explanation for these discrepancies. Indeed, we observe different and inverse responses following EGFR inhibition depending on TG2 expression. Strikingly though, EGFR inhibition brought the cells to a similar baseline level of contractility, suggesting that TG2-mediated regulation of the cellular mechanical state involves, at least in part, a modulation of the EGFR mechanoregulating circuit. Overall, we establish that TG2 acts as an enabling switch toward EGFR-mediated cell mechanical homeostasis. Alternatively, it is also possible that other similar RTKs involved in the regulation of cell contractility, such as the ErbB family member HER2 (Saxena et al., 2017), could influence TG2-mediated contractility. Interestingly, both HER2 and EGFR-mediated cellular contractility require Src while HER2 is able to compensate for the absence of EGFR (Saxena et al., 2017). In this context, a TG2-mediated regulation of cellular mechanical state through other RTK signaling circuits is an interesting possibility.
The crosstalk between EGFR and integrins has been studied intensively over the years and several pathways have been linked to the control of cell contractility and FA dynamics (Eberwein et al., 2015; Hetmanski et al., 2016; Iwabu et al., 2004). Conversely, signaling initiated by EGF can lead to Src-mediated activation of FAK and recruitment of zyxin at focal adhesions in keratinocytes (Eberwein et al., 2015). However, p190RhoGAP (herein referring to ARHGAP35 and ARHGAP5) was also found to be highly activated in these conditions, which would translate to decreased RhoA activity and therefore reduced cell contractility. This is consistent with separate findings indicating that EGFR crosstalk with β3 integrins leads to p190RhoGAP activation (Balanis et al., 2011). Interestingly, under those conditions, deletion of EGFR also leads to zyxin recruitment at FAs (Balanis and Carlin, 2012). Alternatively, EGFR can activate RhoA through a phosphoinositide 3-kinase (PI3K)-mediated pathway (Błajecka et al., 2012). Interestingly, both Src and PI3K have been identified as TG2-interacting proteins (Li et al., 2010; Boroughs et al., 2014). Crucially, a major mechanism by which intracellular TG2 can influence signaling events seems to be through its ability to act as a scaffold protein (Gundemir et al., 2012). In addition, TG2 can be targeted to the leading edge of migrating cells (Antonyak et al., 2009), further suggesting that intracellular localization of signaling components could be influenced by TG2. The difference in active RhoA spatiotemporal localization we observed following TG2 inhibition is consistent with these observations. It remains that we do not yet fully understand the functional link between TG2 and EGFR-mediated contractility. However, our previous results indicate that TG2 is at least involved in EGFR-mediated Src activation through the formation of a complex with keratin intermediate filaments (Li et al., 2010). The involvement of keratins is potentially interesting given our prior work indicating they are involved in the regulation of FA dynamics and cell mechanics (Bordeleau et al., 2010; 2012). This suggests that the contribution of TG2 to cell mechanics could be cell lineage dependent based on intermediate filament expression. In fact, this could explain the often opposing effects on cellular functions regulated by TG2 depending on cellular context (Tatsukawa et al., 2016).
The relationship between cell contractility and growth factor receptors is complex in that while growth factors can stimulate contractility, increased cell contractility can also potentiate cells to respond to growth factor stimulation (Kim and Asthagiri, 2011; Paszek et al., 2005; Meyer-Ter-Vehn et al., 2006). In this context, our results may have additional physiological implications. Notably, the presence of TG2 and its associated increase in cell contractility could provide a strong feedforward loop to further exacerbate cell response to growth factors. Moreover, given our present results, we cannot exclude that part of the TG2-mediated increased signaling we have observed previously due to EGF stimulation is in fact related to the crosstalk between cell contractility and EGFR (Li et al., 2010). Conversely, due to EGFR involvement, our results hint at a likely contribution of EGF in TG2-mediated contractility, which will need to be investigated more in depth. While this remains to be tested, these observations together with our results could suggest increased sensitivity to several kinds of soluble growth factors due to TG2-mediated cell contractility.
Overall, TG2 provides a coupling switch that affects integrin–EGFR crosstalk, which in turn regulates cellular mechanical homeostasis. Thus, we provide a previously unidentified mechanism that governs cellular mechanical state and mechanotransduction, and most notably, a means by which tumor cells can achieve their increased contractile state. These events are particularly relevant considering that both TKRs and microenvironmental mechanical properties play a major role in the progression of several diseases, most notably cancer (Leight et al., 2012; Kim and Asthagiri, 2011; Levental et al., 2009; Mattace-Raso et al., 2006). Furthermore, given the integrating role of contractility in regulating cellular behavior (Bordeleau and Reinhart-King, 2016; Leight et al., 2012; Provenzano et al., 2008; Huynh et al., 2011), our observations provide a broader functional link between TG2, metastatic potential and cell fate. Therefore, our findings highlight the plasticity of cellular mechano-regulation and that further understanding of this regulation is needed to design therapeutics targeting pathological cell mechanical states.
MATERIALS AND METHODS
Reagents and antibodies
Primary antibodies used were as follows: rabbit anti-phospho Y397 FAK (#3283), rabbit anti-FAK (#3285), mouse anti-phospho (Thr202/Tyr204) Erk1/2 (#9106; Cell Signaling, Danvers, MA); rabbit anti-zyxin (#Z4751), mouse anti-vinculin (#V9131, Sigma-Aldrich, St Louis, MO); mouse anti-transgutaminase 2 (#ab2386, Abcam, Cambridge, MA); mouse anti-glyceraldehyde-3-phosphate dehydrogenase (#MAB374, EMD Millipore, Burlington, MA). Secondary antibodies used were as follows: Alexa Fluor 488-conjugated donkey anti-mouse-IgG, Alexa Fluor 594-conjugated goat anti-mouse-IgG, Alexa Fluor 594-conjugated goat anti-rabbit-IgG (Life Technologies Corporation, Grand Island, NY) and HRP-conjugated goat anti-rabbit-IgG and anti-mouse-IgG (Rockland Immunochemicals, Limerick, PA). FBS was purchased from Gibco (Life Technologies); the cell impermeable TG2 inhibitor T101 was purchased from Zedira (Darmstadt, Germany). All other chemicals were from Sigma-Aldrich.
Highly metastatic MDA-MB-231 breast adenocarcinoma cells (HTB-26, ATCC) were maintained in DMEM (Life Technologies) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (Life Technologies). MCF10A mammary epithelial cells (ATCC) were maintained in DMEM supplemented with 5% horse serum, 20 ng/ml EGF (Invitrogen, Carlsbad, CA), 10 mg/ml insulin, 0.5 mg/ml hydrocortisone, 100 ng/ml cholera toxin (Sigma-Aldrich) and 1% penicillin–streptomycin. These are hereafter referred to as complete medium. All cells were cultured at 37°C and 5% CO2. Cells were further tested for mycoplasma contamination by PCR analysis (ATCC® 30-1012K™).
Polyacrylamide gel synthesis and traction force microscopy
Polyacrylamide (PA) gels were prepared and characterized as described previously (Califano and Reinhart-King, 2008). The range of substrate stiffness used (1–10 kPa) was chosen to mimic the stiffness of vascular and tumor tissue as described previously (Paszek et al., 2005). TFM was performed as before using 5 kPa PA gels (Huynh et al., 2013). Briefly, cells were cultured overnight on PA gels embedded with 0.5 µm diameter fluorescent beads (Life Technologies). In some case, cells were pretreated 1 h with either MDC (50 µM), T101 (50 µM) or PP1 (1 µM), or pretreated 6 h with AG1478 (1 µM) before starting the acquisitions. The cells were then imaged in a Zeiss Axiovert inverted microscope equipped with an environmental chamber. The stress field resulting from individual cells traction forces was obtained by imaging fluorescent beads beneath the cell. The unstressed bead field was then imaged following removal of the cells with trypsin (0.25%, 10 min at 37°C). Bead displacements were calculated from the stressed and unstressed images and used to compute cellular traction vectors and total magnitudes of force using the LIBTRC analysis library developed by Dr Micah Dembo (Department of Biomedical Engineering, Boston University).
Preparation of collagen gels
Type I collagen was isolated from rat tail tendons (Rockland Immunochemicals) as described previously (Mason et al., 2013). Briefly, type I collagen was solubilized in 0.1% sterile acetic acid (JT Baker) to obtain 10 mg ml−1 stock solutions. A working collagen solution was prepared by neutralizing with 1 M sodium hydroxide in 10× Dulbecco's phosphate-buffered saline (D-PBS; Life Technologies), mixed with HEPES (EMD Millipore) and sodium bicarbonate (JT Baker) to form 1.5 mg ml−1 collagen gels with final concentrations of 1× D-PBS, 25 mM HEPES and 44 mM sodium bicarbonate.
For collagen contraction assays, 24 well-plates were first blocked 1 h with PBS containing 1% BSA. 500 µl collagen gels seeded with 500,000 cells/ml were then allowed to polymerize in the wells for 1 h at 37°C. 1 ml of medium containing the inhibitors was then added to each well after which the gels were released from the side of the wells. The changes in gel area was then quantified over the course of 4 days. The culture medium was carefully replaced each day to provide fresh inhibitors.
Quantification of 3D ECM remodeling
Quantification of ECM remodeling was performed as described previously (Kraning-Rush et al., 2011). Briefly, collagen gels were fixed 24 h after treatment, and single cells were imaged by confocal reflectance microscopy. The collagen intensity was then analyzed with a custom ImageJ macro (available upon request) along the path of a 100 µm line centered on the cell centroid. The resulting line signal was integrated around the entire cell in 2° increments and the edge of the cell was used to define the origin and obtain the collagen intensity profile. The resulting intensity profiles were fitted with an exponential decay model to extract the characteristic compaction length as described previously (Kraning-Rush et al., 2011).
Quantitative polarization microscopy
Quantitative polarization microscopy of cells embedded within 3D collagen scaffolds was used to measure cell contractility as described previously (Wang et al., 2018). Briefly, the polarization microscope consist of a motorized linear polarizer (Thorlabs, Newton, NJ) positioned in the illumination plane above the condenser and a circular polarizer positioned in the imaging plane of a Zeiss Axiovert microscope equipped with an Axiocam 506 color camera and a 20× polarization lens (NA=0.5). Image sequences were acquired with 5° intervals of the rotating polarizer over a range of 0 to 180° using Zen software. The polarized image sequences were then processed with a custom MATLAB code (available upon request) to obtain a pixel-by-pixel retardance map. The retardance signal proportional to cell contractility was quantified by measuring the average retardance over the cell area after background subtraction (Wang et al., 2018).
Cell adhesion assays
Cells were washed three times in Ca2+/Mg2+-free PBS and then allowed to detach in 15 mM PBS-citrate with 0.6 mM EDTA for 10 min at 37°C. Collected cells were pretreated for 30 min with MDC (50 μM), T101 (50 μM) or DMSO (1:1000 dillution) in complete medium or complete medium alone in the case of the PLKO and shTG2 cells. Cells were then seeded on collagen-coated glass slides (5000 cells per slides). After allowing the cell to attach and spread for either 15 or 30 min, cells were fixed with 3.7% formaldehyde in 1× D-PBS for 10 min at room temperature.
Confocal reflectance imaging
The fiber structure of collagen gels was visualized using a Zeiss LSM 700 inverted laser scanning microscope (Carl Zeiss Microscopy, Thornwood, NY) equipped with a 405 nm laser and a 40×/1.1 NA water-immersion objective (Carl Zeiss) (Mason et al., 2013).
Immunofluorescence of fixed samples
Cells were fixed in 3.7% formaldehyde in 1x D-PBS for 10 min at room temperature and subsequently permeabilized in 1% Triton X-100 in D-PBS (Mallinckrodt Baker, Phillipsburg, NJ) for 5 min. Primary antibodies (anti-phospho FAK, anti-zyxin and anti-vinculin) were used at 1:100 dillution in 1× PBS and incubated overnight at 4°C in a humid chamber. Alexa Fluor-coupled secondary antibodies were then incubated at 1:200 dilution for 1 h at room temperature. Samples were then mounted using Vectashield mounting medium (Vectorlabs, Burlingame, CA). Fluorescent images were acquired with a Zeiss Axio Observer Z1 m, Zeiss 710 confocal or Zeiss 800 confocal microscope, and z-stacks were reconstructed with ImageJ.
To generate stable TG2-knockdown or TG2-expressing negative control cells, MDA-MB-231 cells were subjected to lentiviral transduction with either the TG2-targeting MISSION®shRNA plasmid (SHCLND-NM_004613; TRCN0000272816) or with the MISSION® pLKO.1-puro Empty Vector Control Plasmid DNA (SHC001; Sigma-Aldrich), followed by puromycin selection (1.5 µg/ml).
Quantification of FA metrics and generation of the vinculin to zyxin ratio image sets were performed using ImageJ. The images were background subtracted (rolling ball; 16 µm radius) and subjected to a mean filter (1 µm radius). FA identification was undertaken by applying a top hat filter on the vinculin channel followed by a threshold (Otsu method) to obtain an overlay of all FAs. The FA overlay was then used to count the number of FAs or to quantify the p-FAK signal within FAs. To generate the vinculin to zyxin ratio image sets, the images were converted to a 32-bit format and the vinculin channel was subtracted from the zyxin channel. The FA overlay was then used to quantify the vinculin to zyxin ratio in FAs (ΔZyxin:Vinculin). For cell area quantification, the zyxin fluorescence channel was used to determine the intensity threshold and then measure the area.
At 24 h after the transfection, the cells expressing GFP–FAK were seeded on collagen-coated (0.1 µg/ml in ice-cold PBS, 1 h on ice) glass-bottom dishes (MatTec, Ashland, MA). FRAP experiments were performed with a Zeiss 710 confocal microscope equipped with an environmental chamber, according to a protocol described previously (Bordeleau et al., 2010). Photobleached regions consisted of a 2 μm wide circular region enclosing a selected FA. Fluorescence within the region was measured at low laser power before bleaching, and then photobleaching was undertaken with 8 iterations using both the 405 and 488 nm laser lines at 100% laser power. Recovery was monitored at 488 nm at 0.5 s time intervals for 80 s. In some experiments, cells expressing GFP–FAK were pre-treated with either MDC (50 μM) or T101 (50 μM) for 1 h. Fluorescence during recovery was normalized to the prebleach intensity and for the bleaching occurring during image acquisition. Relative recovery rates were computed with MATLAB (MathWorks, Natick, MA). The half-time for fluorescence recovery towards the asymptote was extracted from the plots after curve fitting of the data to a single exponential association algorithm.
Cells cultured on PA substrate of 1 or 10 kPa were lysed with preheated (90°C) in 2× Laemmli sample buffer as described before and subjected to gel electrophoresis (Bordeleau et al., 2012). Briefly, samples were heated for 5 min at 95°C, subjected to SDS/PAGE with a Mini-PROTEAN Tetra System (BioRad, Hercules, CA) and electrotransferred onto a PVDF membrane. Membranes were blocked with 5% (w/v) milk (Nestlé) in Tris-buffered saline (TBS) with 0.1% Tween 20 (JT Baker) or 5% (w/v) BSA. Membranes were then incubated overnight in TBS-Tween with either an rabbit anti-pTyr397 FAK primary antibody, mouse anti TG2 primary antibody or mouse anti pThr202/Tyr204 Erk primary antibody (all diluted 1:1000) at 4°C, and in TBS-Tween with 0.1% milk with an HRP-conjugated secondary antibody (diluted 1:2000) for 1 h at room temperature. Samples were images with a Las-4000 imaging system (Fujifilm Life Science, Stamford, CT) after addition of Immobilon Chemiluminescent Substrate (Millipore). Densitometry was performed with ImageJ and expressed as the fold change of the ratio of the protein of interest to GAPDH. Data are the result of three independent experiments.
RhoA GTPase activity assay
An ELISA-based assay was used to assess RhoA activity (#BK124, Cytoskeleton, Denver, CO) according to the manufacturer's protocol. Briefly, cells were washed with ice-cold wash buffer and lysates were collected using the extraction buffer. Following antibody hybridization and the colorimetric staining, the RhoA activity signal was acquired using a UQuant microplate spectrometer and Gen5 software (BioTek, Winooski, VT). RhoA activity was normalized to the total protein content of the sample using the included Precision Red Protein Assay (#ADV02; Cytoskeleton).
Alternatively, time lapse images of cells transfected with the RhoA activity probe rGBD-GFP (the RhoA-binding domain of rothekin) were acquired using a Zeiss 800 confocal microscope. The localization of the probe to sites of active RhoA was used to manually identify and count the subcellular structures visible in the movies. Lamelipodia-like structures were defined as active RhoA-containing section extending away from the cell center over time. Membrane ruffle-like structures were defined as active RhoA-containing regions that presented a retrograde motion relative to the cell center. Temporally color-coded images were generated in ImageJ by first applying a mean filter (0.3 µm voxel size), followed by a top hat filter (MorphoLib plugin, ball, 0.6 µm voxel size) and the hyperstacks Temporal-Color code function.
Data were analyzed using a one-way analysis of variance (ANOVA) followed by a post-hoc Tukey's honest significant difference test in MATLAB. In the case of discrete values, the data were analyzed based on a binomial distribution and assuming the normal approximation for hypothesis testing. Statistical significance was considered as P<0.05. All values are expressed as the mean±standard error of the mean (s.e.m.) unless otherwise indicated.
We thank Emilie Pic for her help in setting up experiments. We thank Matthew Zanotelli and Lauren Hapach for helpful discussions and critical reading of the manuscript.
Conceptualization: F.B., M.A.A., R.A.C., C.A.R.-K.; Methodology: F.B., M.A.A., R.A.C., C.A.R.-K.; Formal analysis: F.B., W.W., A.S.; Investigation: F.B., W.W., A.S.; Resources: F.B., M.A.A., R.A.C., C.A.R.-K.; Data curation: F.B.; Writing - original draft: F.B.; Writing - review & editing: F.B., W.W., A.S., M.A.A., R.A.C., C.A.R.-K.; Visualization: F.B., W.W., M.A.A.; Supervision: F.B., M.A.A., R.A.C., C.A.R.-K.; Project administration: F.B., R.A.C., C.A.R.-K.; Funding acquisition: F.B., C.A.R.-K.
This work was supported in part by grants from the National Science Foundation (1740900) and National Institutes of Health (HL127499) to C.A.R.-K., and grants from the Scholarship for the Next Generation of Scientists from the Cancer Research Society and National Institutes of Health (CA212270) to F.B. Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.