Characterization of tenascin-C-induced signaling in tumorigenesis

During cancer progression, the extracellular matrix (ECM) is extensively remodeled. The ECM molecule tenascin-C is an adhesion-modulating molecule, which is highly expressed in tumor stroma. Tenascin-C was shown to disrupt the interaction of cells with fibronectin, an adhesive ECM molecule, through inhibition of syndecan-4, the co-receptor of the fibronectin binding integrin α5β1. Cells on a mixed substratum of fibronectin and tenascin-C failed to form cell adhesion structures and actin stress fibers and thus remained rounded. Focal adhesion kinase (FAK) (Huang et al., 2001; Orend, 2003) and the small GTPase RhoA (Wenk et al., 2000), two molecules with an important role in formation of focal adhesions and actin stress fibers, were downregulated in the presence of tenascin-C. Furthermore, the actin-binding and filament-stabilizing molecule tropomyosin-1 (TM1) was identified by Ruiz et al., (2004) to be downregulated by tenascin-C and its downregulation contributed to the lack of actin stress fiber formation on a fibronectin/tenascin-C substratum.
Here, we investigated the signaling events, that lead to cell rounding by tenascin-C. In particular, we wanted to understand how tenascin-C prevents the formation of focal adhesions and stress fibers, and how tenascin-C affects expression and function of the three downstream targets FAK, RhoA and TM1. First, we investigated whether inhibition of syndecan-4 by tenascin-C is linked to reduced expression of RhoA and TM1 and inhibition of FAK. By activating syndecan-4 on fibronectin/tenascin-C substratum we observed that indeed this is the case. Thus, repression of FAK, RhoA and TM1 could explain the lack of actin stress fiber formation on fibronectin/tenascin-C.
Whereas expression of TM1 was not regulated by tenascin-C at the transcriptional level, it turned out that tenascin-C repressed RNA levels of two other tropomyosins TM2 and TM3, which are far less expressed in T98G cells than TM1. Apparently, lowered levels of TM2 and TM3 affected TM1 protein heterodimer stabilization through proteasomal degradation that was largely enhanced on fibronectin/tenascin-C. This possibility was supported by our observation that inhibition of the proteasome restored TM1 expression on this mixed substratum. Our data suggest that tenascin-C does not only repress gene expression of tropomyosins but also enhances their proteasome-mediated protein degradation. Repression of TM1, a molecule with a tumor suppressor-like activity, by tenascin-C might be relevant in cancer, since low levels of this molecule can protect cancer cells from apoptosis.
To learn more about the underlying mechanism of tenascin-C-induced cell rounding, we searched for signaling pathways, that enabled cells to spread in the presence of tenascin-C. In addition, we used knockdown and overexpression studies together with chemical inhibitors. Our data suggest that concomittant restoration of the expression and function of all three downstream targets FAK, RhoA and TM1, is necessary to induce cell spreading on a fibronectin/tenascin-C substratum. In particular, we observed that activation of endothelin receptor type B (EDNRB) induced spreading in the presence of tenascin-C. This was dependent on PI3K, PLC and JNK, since chemical inhibitors of these enzymes blocked EDNRB-induced cell spreading on fibronectin/tenascin-C. Signaling by EDNRB was linked to activation of FAK and paxillin and restoration of TM1 and RhoA expression, again supporting our notion that inactivation of these molecules is critical for tenascin-C-induced cell rounding.
Based on the results from Ruiz et al., (2004), that described an enhanced expression of endothelin receptor type A (EDNRA) in the presence of tenascin-C, we wanted to know whether and how EDNRA signaling contributes to cell rounding by tenascin-C. EDNRA expression was triggered by tenascin-C upon contact with the substratum for more than 5 h. We demonstrated that EDNRA signaling is linked to cell rounding on a fibronectin/tenascin-C substratum through inhibition of FAK and repression of RhoA and TM1. Collectively, these data suggest that inhibition of syndecan-4 is responsible for initial cell rounding and that upon induction of EDNRA by tenascin-C, EDNRA maintains cell rounding on a fibronectin/tenascin-C substratum at later time points. In gliomas and other cancers we found a high expression of tenascin-C and EDNRA, that correlated with more advanced stages, which supports the possibility that tenascin-C potentially promotes tumor progression through EDNRA.
In addition to EDNRB, concomittant activation of the receptors for lysophosphatidic acid (LPA) and platelet-derived growth factor (PDGF) also enabled cells to spread on a fibronectin/tenascin-C substratum by a mechanism, which again involved restoration of the expression and function of FAK, paxillin, RhoA and TM1. By using cells lacking syndecan-4, we observed that LPA/PDGF bypassed the requirement for syndecan-4 in cell spreading on a mixed substratum. Knockdown of paxillin prevented LPA/PDGF-induced cell spreading on fibronectin/tenascin-C, which suggests an essential role of paxillin in LPA/PDGF induced cell spreading in presence of tenascin-C. In further support of an important role of paxillin in LPA/PDGF-induced cell spreading, we showed that ectopic expression of syndesmos, a molecule that binds syndecan-4 and paxillin, enabled cells to spread on fibronectin/tenascin-C. We showed that expression levels of TM1 are critical for cell rounding and cell spreading in the presence of tenascin-C, respectively. Whereas ectopic expression of TM1 restored cell
spreading on fibronectin/tenascin-C, knockdown of TM2/3 prevented LPA/PDGF-induced cell spreading on the mixed substratum. We also observed, that TM1 levels were tightly linked to expression of RhoA and activation of FAK, which suggests an interdependent regulation.
We observed that activation of the receptors for endothelin-1 (ET1) (EDNRB), and LPA/PDGF induced spreading in the presence of tenascin-C by distinct pathways. Whereas EDNRB-induced spreading was dependent on PI3K, PLC and JNK, but not on MEK, LPA/PDGF-induced cell spreading was dependent on PI3K and MEK, but not on PLC and JNK. Signaling by these factors was linked to activation of FAK and paxillin and, restoration of TM1 and RhoA expression. Together, our results suggest that at least two (and presumably many more) mechanisms exist that modulate the adhesive strength of a fibronectin/tenascin-C substratum. This supports the notion that combined signaling from the ECM and growth factors can determine cell adhesion and migration.
ET1 (through EDNRB) and LPA/PDGF did not only trigger cell spreading in the presence of tenascin-C but also stimulated cell migration on this substratum. In particular, LPA/PDGF-induced cell migration was PI3K and ROCK dependent. By using cells with ectopic expression of syndesmos and TM1, or reduced expression of TM1-3, we showed that both a strong as well as a loose adhesion blocked LPA/PDGF-induced cell migration on fibronectin/tenascin-C. These observations might be important for cancer diagnosis and may eventually allow to develop novel cancer treatments. In particular, we find that in gliomas a high expression of tenascin-C and of PDGF receptors a and b correlates with malignancy. Moreover, a high expression of syndesmos correlates with a bad 5 year survival prognosis and chemotherapy response rate in patients with oligodendrogliomas.
In summary, here it was shown that cell adhesion and migration on an anti-adhesive fibronectin/tenascin-C substratum can be modulated by additional signaling from growth factors. We identified a minimal set of critical targets of tenascin-C downstream of syndecan-4 that include FAK, paxillin, RhoA and TM1. Induction of EDNRA signaling by tenascin-C provides an additional mechanism that contributes to maintained cell rounding by a mechanism that again affects the same set of tenascin-C targets as those downstream of syndecan-4.