Smad2 and Smad6 as predictors of overall survival in oral squamous cell carcinoma patients

The Smad family of proteins, Smads 1 to 8, are key molecules in Transforming Growth Factor-β (TGFβ) signaling, eventually modulating both TGFβ tumor suppressive and oncogenic effects [1]. Among them, Smad2 and Smad3 are known as receptor regulated Smads (R-Smads) and are phosphorylated in response to TGFβ itself. The phosphorylated protein, in conjunction with the common Smad (Co-Smad), Smad4, translocates to the nucleus eliciting the transcription of other genes [2‐4]. The Inhibitory Smads (I-Smads) Smad6 and 7, on the other hand, prevent the activation of R-Smad by phosphorylation and/or interfering with its nuclear translocation [5‐7].

Smad signaling seems to be relevant to the pathogenesis of several epithelial cancers. Smad4 and Smad2 functions are disrupted in pancreatic, esophageal, gastric, colon and lung cancer [8‐12]. Over-expression of inhibitory Smad6 and Smad7 was described in pancreatic cancer and in pancreatic cancer cell lines [13, 14]. Smad2 and 3 present different targets and have distinctive roles, as shown in skin tumors of transgenic mice [15].

Concerning head and neck squamous cell carcinoma (HNSCC), however, data on Smads are still scarce. Studies done with HNSCC samples have shown alterations of individual Smad expression as measured by immunohistochemistry [16, 17]. In addition, evidence obtained in in vitro studies indicates that Smad signaling may enhance invasiveness in HNSCC [18].

We have previously suggested that, in oral SCC but not in other HNSCC sites, the tumor supressive effect of TGFβ was absent in lymph node positive (pN+) but still present in lymph node negative (pN0) patients [19]. Therefore, we assumed that the extent of expression of individual Smad mRNAs might reflect the degree of TGFβ resistance, and in this way, correlate with progression in oral SCC and consequently with survival. In this work, we found that Smad family mRNA expression was globally increased in HNSCC as compared to adjacent tissue. In addition, among all Smads, Smad2 and Smad6 were suggested to be prognostic markers, correlating with overall survival.

In this study, we provide evidence that Smads may have a key role in head and neck cancer, influencing survival. Specifically, we show, prospectively, that high Smad6 mRNA expression and low Smad2 mRNA correlate with better survival in oral SCC patients submitted to curative surgery, independent of nodal status. In addition, Smad7 high mRNA expression correlated with shorter survival in patients with larynx SCC submitted to curative surgery. The fact that these correlations were restricted to specific tumor locations supports the hypothesis that underlying biological heterogeneity exists between different subsites within head and neck.

TGFβ is known as a potent tumor supressor in normal epithelial cells and in early-stage tumors but during tumor progression it becomes an oncogenic factor, mainly in advanced tumors [21]. Since Smad6 has a blocking effect on TGFβ signaling either by direct binding to the TGFβ receptor, by competing with Co-Smad for R-Smad complex formation or by targeting receptor for degradation [6, 7, 22, 23], we can speculate that this effect on survival occurs because Smad6 would inhibit the TGFβ tumorigenic signaling, thus favouring a better outcome. The favorable prognosis presented by Smad6⁺ oral SCC patients agrees with that described in 115 esophageal SCC using immunohistochemistry [24]. Reinforcing this idea, we found that even in the poor prognosis group, pN+, the presence of Smad6⁺ was associated with a survival advantage similar to that of pN0 patients.

Smad2 is the classically R-Smad of the TGFβ pathway [1], our own data suggest Smad2 negativity reinforces the effect of Smad6 on survival of oral SCC. Smad2 may be a key factor for the interruption of TGFβ tumorigenic signaling in this group of patients, together with Smad6. In accordance with that, we found TGFβ1 positive tumors were also Smad2 negative. Our finding regarding the association between lack of Smad2 expression and a favourable clinical outcome is in disagreement with previous studies done in head and neck cancer as well as other tumor types. In oral squamous cell carcinoma, a study involving 125 patients suggested a link between decreased expression of both activated Smad2 (p-Smad2) and TGFβ receptor II (TβR-II), with aggressive tumor features, which suggests TGFβ signaling exerts a protective role possibly through Smad 2 [25]. In accordance, the Smad expression profile assessed by tissue array in 170 head and neck squamous cell carcinoma pointed to the loss of TGFβ/Smad2 signaling as a possible cause of adverse outcome [17]. A protective role of Smad 2 was also suggested for esophageal, colon and breast cancer [26‐28]. In line with this effect, others have identified a missense mutation of Smad2 in the squamous cell line SCC-15 suggesting that the loss of Smad 2 may be part of head and neck carcinogenesis [29].

There are, however, several studies in line with our data. Matrix metalloproteinases play roles in cancer progression by degrading the extracellular matrix and basement membrane. TGFβ1 signaling induces MMP-9 expression via Smad 2/3 [30]. Accordingly, the TGFβ1/Smad 2/3 axis regulates MMP-9 expression through the transcriptional factors Snail and Ets-1, contributing to oral cancer progression [31]. In addition, the metastasis-associated protein metastatin can physically and functionally interact with Smad2/3, and enhance TGFβ mediated MMP-9 induction [32]. Activation of Smad 2/3 signaling has also been linked to enhancement of MMP-13 expression and invasion of head and neck squamous carcinoma cells [18].

Smad 2 accumulation seems to parallel an elevation of H-ras, both of which are essential for epithelial mesenchymal transition (EMT), an essential step during carcinogenesis. Having undergone EMT, others have shown that fibroblastoid carcinoma cells with elevated levels of activated Smad2 gain the capability to spread to a wide variety of tissues by a further increase in Smad2 expression [33]. In addition, a prominent expression of alpha(v)beta(6) integrin at tumor stroma interface in xenograft model, which resembles human head and neck carcinomas, has been connected with cancer progression. Alpha(v)beta(6) interacts with TGFβ, as an Alpha(v)beta(6) blocking antibody can inhibit TGFβ mediated Smad2/3 phosphorylation, which leads to inhibition of tumor growth in vivo, suggesting a role for the microenvironment for this effect [34].

It is not easy to reconcile these conflicting results regarding the role of Smad2, in head and neck tumors, since they reflect the dual roles of TGFβ itself, acting both as a tumor suppressor and a tumor enhancer, depending on the context. What is not clear, however, is what constitutes this context [35]. In our head and neck patient series, Smad2 seems to be a TGFβ signaling node that enhances tumor aggressiveness (Figure 4). An intriguing possibility, that merits further study, is that HPV infection may be interacting with TGFβ network as shown in cervical carcinoma [36].

Taking into account that Smad6 is at the crossroad of many signaling pathways, being regulated not only by family members of bone morphogenetic protein (BMP) and TGFβ, but also by epidermal growth factor and Ras/MAPK [6, 7, 37], its expression may be key to shift signaling from oncogenesis to tumor supression, towards or against proliferation, independent of Smad3 and Smad4, at least in oral SCC (Figure 4).

If we consider Smad6 and Smad7 as interchangeable TGFβ blocking Smads, our data in oral SCC is also in keeping with others, showing that a combination of increased immunohistochemical expression of Smad7 and decreased Smad4 expression are markers of good prognosis in gastric cancer, with 67.5% survival rate versus 52.2%, (P = 0.0011) [38]. However, data concerning Smad7 seems to be more controversial since our own present data indicates that Smad7 mRNA low expression correlates with shorter survival in larynx SCC (Figure 3). In agreement with our results, patients with colon and gastric cancers, with Smad7 gene deletion or low Smad7 protein expression, were described as having prolonged survival as compared to patients with higher Smad7 expression [39, 38].

Concerning the lack of influence of Smad4 mRNA expression on patient survival in our study, data in the literature suggests that there is no single rule for this. Low Smad4 expression, detected by immunohistochemistry, and poorer five-year survival was shown in 249 patients with advanced gastric cancer and in 258 esophageal squamous cell carcinomas (P < 0.05) [9, 10]. In colon cancer, such an influence was not shown, which is in accordance with our present data [39]. Taking into account that Smad4 mRNA was the most ubiquitous among Smads we can argue that perhaps the amount of Smad4 was not a crucial element for these patients.

Finally, we found higher Smad mRNA expression in SCC primary tumors as compared to adjacent tissue in agreement with previously published data, showing that HNSCC cell lines present multiple defects in TGF-beta signaling [40]. In HNSCC samples, data obtained from 170 tumors using tissue array technology, shows expression of Smad2, Smad3 and Smad4 proteins [17]. Another small study comparing the expression between tumor and adjacent tissue did not find differences at least with respect to Smad4, Smad6 and Smad7 protein expression in 13 head and neck SCC tumors [16].

To sum up, our results suggest that increased Smad6 mRNA expression and low Smad2 mRNA expression might be markers of better outcome in oral SCC but not in larynx cancer submitted to curative surgery. These effects may be linked to aberrant TGFβ signalling. In larynx cancer, a similar relationship was found for Smad7 mRNA low expression. Our findings need to be validated in larger prospective studies, and may in the future, help to stratify candidate patients for adjuvant treatment in head and neck cancer.

TGFβ is classically known as tumor supressor in normal epithelial cells that turns into a malignant factor during tumor progression favoring tumor growth and metastasis. Here we propose that the interruption in TGFβ tumorigenicity by Smad2 dowregulation or by Smad6 overexpression confers a better outcome. Although both might be considered as prognostic markers, the molecular mechanism envolved in this process is not clear. Further studies are warranted to explore the mechanisms involved.