CD109 and squamous cell carcinoma

In tumor tissues, CD109 was immunohistochemically detected in SCCs [11‐18] as well as urothelial carcinomas [19], malignant melanomas [20], basal-like breast carcinomas [21], myxofibrosarcoma [22], epithelial sarcomas [23] and glioma [24]. In particular, Shiraki et al. [24] reported CD109-positive perivascular tumor cells in human lower-grade glioma tissues and in a mouse model recapitulated human glioma, suggesting a key role of CD109 for this disease. Previous studies have shown that the high expression of CD109 in SCCs and the limited expression in normal squamous cells (Table 1) [17]. Furthermore, CD109 is highly expressed in well-differentiated SCCs rather than in moderately or poorly differentiated SCCs, thus the expression level of CD109 is inversely correlated with tumor grade [15, 17].

The TGF-β signaling pathway is involved in many cellular processes including cell growth, cell differentiation apoptosis, and cellular homeostasis [40, 41]. The family of TGF-β ligands, TGF-β1, TGF-β2 and TGF-β3, binds to specific transmembrane type I and type II serine/threonine kinase receptors (TGF-βR1 and TGF-βR2) [42], resulting in activation of TGF-βR1 kinase activity [42, 43]. The activated TGF-βR1 then propagates the signal by phosphorylating its intracellular substrates, R-SMADs (Smad2 and Smad3) [44]. Smad2 and Smad3 interact with TGF-βR1 and SARA (Smad anchor for receptor activation), a FYVE domain protein that interacts directly with Smad2 and Smad3, SARA functions to recruit Smad2 to the TGF-β receptor [45], then the phosphorylated R-SMADs form heteromeric complexes with Co-SMAD (Smad4) [44]. After the phosphorylation and subsequent complexes with Smad4, these R-Smads complexes are released from TGF-βR1 and SARA [45], then translocate into the nucleus where they interact with transcription factors that recruit them to specific promoter elements of target genes [41, 44, 46].

Receptor endocytosis is a pivotal regulatory mechanism in signal transduction. TGF-β receptors are internalized via both clathrin- and caveolae-dependent pathways. Internalization of the TGF-β receptors via the clathrin-coated pits has been linked with signaling via Smad2/3 and receptor recycling. In contrast, TGF-β receptor localization in caveolae is associated with downregulation of Smad2/3 signaling and receptor degradation following ubiquitination by the E3-ubiquitin ligase Smurf2 [47]. However, inhibitory Smads (Smad6 and Smad7) form a distinct subclass of Smads that act in an opposing manner to R-Smads and antagonize signaling [48]. They may compete with R-Smads for binding to activated TGF-βR1 and thus to inhibit the phosphorylation of R-Smads [41]. In addition, they recruit E3-ubiquitin ligases to the activated TGF-βR1, resulting in receptor ubiquitination and degradation, and termination of signaling [44].

Dysregulation of the TGF-β pathway has been implicated in multiple types of cancer [49]. Studies have demonstrated that TGF-β signaling elicits a preventative effect during the earlier stages of tumorigenesis, but a suppressive effect during the later in tumor development [50]. Mutations in the TGF-βR1 gene have also been found in SCCs of the skin, suggesting that the inactivation of TGF-β leads to the initiation of SCCs [2].

CD109 is a TGF-β co-receptor [11], and modulates TGF-β signaling receptor activity in a cell-specific manner [49, 51]. On the cell surface, CD109 negatively regulates the TGF-β1 signaling pathway via formation of a receptor complex with TGF-βR1 and TGF-βR2 in human keratinocytes [5]. TGF-β receptors degradate following ubiquitination by the E3-ubiquitin ligase Smurf2 [47], and are internalized via both clathrin-dependent and caveolae-dependent pathways [52]. Bizet et al. [47] demonstrated that CD109 associates with caveolin-1 and promotes TGF-β receptor endocytosis. In addition, CD109 promotes localization of the TGF-β receptors into the caveolar compartment in the presence of ligand and facilitates TGF-β receptor degradation. CD109 also regulates the localization and the association of Smad7/Smurf2 with TGF-βR1. The inhibitory effects of CD109 require Smad7 expression and Smurf2 ubiquitin ligase activity [49]. Furthermore, CD109 can be released from the cell surface by cellular lipases such as phosphatidylinositol-specific phospholipase C (PI-PLC). The soluble form of CD109 retains its ability to bind TGF-β1 and confiscate it away from the TGF-β receptors [53].

However, Vorstenbosch et al. [54] reported that CD109 differentially regulated TGF-β-induced ALK1-Smad1/5 versus ALK5-Smad2/3 pathways (ALK1 and ALK5 are all TGF-β type I receptors). They found that TGF-β signaling inhibits endothelial cell proliferation and migration, while TGF-β signaling also induces these processes via ALK1-Smad1/5 [54]. They demonstrate that ALK1 is expressed and co-localizes with CD109 in mouse keratinocytes and that mice overexpressing CD109 in the epidermis display enhanced ALK1-Smad1/5 signaling, but decreased ALK5-Smad2/3 signaling [54].

Besides, TGF-β1 is a potent inhibitor of growth in most epithelial cells [5]. Hagiwara et al. [17] demonstrated that oral SCC cell lines overexpression CD109 accelerated cell proliferation and impaired the anti-proliferative effect mediated by TGF-β1. In contrast, SCC cells with CD109 knockdown exhibited slower cell growth [17]. A high level of CD109 expression inhibited Smad2 phosphorylation, thus attenuated TGF-β1/Smad2 signaling and impairs TGF-β1-mediated suppression of cell growth, CD109 knockdown increased Smad2 phosphorylation by TGF-β1 stimulation [17]. Although CD109 also regulates Smad1/5 signaling [54], it has not been connected with the development of SCC. Together, CD109 facilitates the development of SCCs via inhibition of the TGF-β-Smad2/3 pathway (Fig. 2).

Signal transducer and activator of transcription factor 3 (STAT3) is critical to cell proliferation, differentiation, migration, survival, and oncogenesis [37, 55]. Litvinov et al. [56] reported that the expression of CD109 protein was markedly decreased in psoriatic epidermis as compared to adjacent uninvolved skin. However, CD109 mRNA expression is unchanged in psoriatic plaques in comparison with normal skin, suggesting a possibility that CD109 protein release is enhanced in psoriatic keratinocytes [56]. They suggested that released/soluble CD109 is able to induce molecular changes that are known to occur in psoriasis [56]. In vitro, they found that transfection of CD109 siRNA down-regulates STAT3, release of CD109 from the cell surface of cultured human keratinocytes. In addition, exogenous/recombinant CD109 induces STAT3 signaling in human keratinocytes [56]. Besides, Chuang et al. reported that CD109 expression was dramatic upregulated in metastatic lung adenocarcinoma cells, and cells expressing a CD109 shRNA (shCD109) showed a dramatic reduction in STAT3 phosphorylation. STAT3 knockdown greatly reduced metastases, and restoration of STAT3 activity increased the ability of shCD109-expressing cells to metastasize [57]. Upon activation, STAT3 is phosphorylated by the non-receptor protein tyrosine kinases janus kinase 2 (JAK2), leading to the formation of STAT3 dimer and translocation into the nucleus [58, 59]. However, inhibition of JAK kinase activity in fibroblasts overexpressing CD109 reduced phosphorylated STAT3 to a level similar to that in the parental cells expressing low levels of CD109, suggesting that CD109-induced STAT3 phosphorylation requires JAK kinase activity. Thus, JAK/STAT3 signaling might mediate the effects of CD109 in tumor growth and metastasis [57].

Although knockdown of CD109 in human keratinocytes and lung adenocarcinoma cells downregulates STAT3 signaling in vitro [56, 57], the CD109-deficient mice displayed opposite results. Mii et al. [60] generated CD109-deficient mice, which exhibited skin abnormalities including epithelial hyperplasia, and inflammatory cell infiltration. They reported that STAT3 phosphorylation in CD109-deficient mice was significantly higher compared to the wild type mice. In addition, up-regulation of STAT3 signaling is associated with increased proliferation and impaired differentiation of keratinocytes [60].

The discrepancy of the results from in vitro and in vivo studies might be caused by the systematic changes of microenvironment in the tissues of CD109 deficient mice. Loss of CD109 in all the cells in mice might modify the subcutaneous microenvironment which activates STAT3 signaling in keratinocytes. In addition, CD109 may exert distinct regulatory effects in different cell types, leading to cell-type specific modification in STAT3 signaling. In addition to keratinocytes, CD109 is expressed in endothelial cells, epithelial cells, and fibroblasts [3, 5, 19, 42], which participate in constitute the skin tissue. However, to date the relationship of CD109 and STAT3 signaling have not been explored in these cell types.

Epidermal growth factor receptor (EGFR) is a member of the ErbB family of receptors. Upon ligand binding by EGF, EGFR forms dimers, either homodimers or heterodimers with another member of the ErbB family HER2 [61]. The dimerized receptors auto-phosphorylate each other and then phosphorylate the non-receptor protein tyrosine c-Src kinase, which activates STAT3 [58]. The activation of EGFR promotes cell migration, survival, and proliferation. In malignant tumors EGFR over-expression is correlated with the depth of invasion of the tumor and linked to poorer prognosis [61]. Mutations that lead to EGFR overexpression are detected in lung SCC [62], head and neck SCC [63], and esophagus SCC [64]. The membrane-anchored CD109 in SK-MG-1 cells directly interacts with EGFR and enhances EGF signaling, which subsequently increases cell migration and invasion, while the secreted CD109 has no effect on EGF signaling [65]. EGFR might mediate the effects of CD109 on STAT3 signaling, which requires further studies to elucidate (Fig. 3).

To date the broad picture of tumorigenesis and development of SCC remain incomplete. There are cross-talks between TGF-β and STAT3 or EGFR signaling pathways. CD109 is likely one of the key effectors of the signaling network regulating SCCs. However, no direct evidence has been reported to delineate the roles of CD109 in pathogenesis of SCCs.

Studies from human tissue samples indicate that CD109 is highly expressed in SCCs of multiple organs, particularly in well-differentiated malignant squamous cells [15, 17]. By detection of CD109 expression with immunohistochemistry in human tissues, CD109 may potentially acts a biomarker to determine the progression of SCC. Current studies suggest that CD109 is highly expressed in well-differentiated SCCs and its expression is lower in un-differentiated SCCs [15, 17]. However, it is not clear whether CD109 is associated with vascular invasion, metastasis, and prognosis after surgery. Therefore, further studies are needed to explore the clinicopathological significance of CD109 in SCCs in a larger sample size.

CD109 is a GPI-linked glycoprotein, which enables it to be released from the membrane [56]. The soluble form of CD109 also affects the binding of TGF-β to its receptors, and subsequently modulate SCC progression [53, 56]. Litvinov et al. [56] found that CD109 released from the cell surface into the extracellular milieu, and the released form of CD109 retains its ability to induce intracellular signaling pathways. Besides, Sakakura et al. [66] reported that serum CD109 was released by xenografted tumor and it increases proportionally with the volume of tumor xenograft. Therefore, detection of serum CD109 level might help monitor tumors overexpressing CD109 including SCCs.

Exosomes communicate primary tumor lesion and its niche via its package containing selected proteins or other molecules [67]. The isolation and analyses of circulating tumor-associated exosome may serve as biomarkers for diagnosis of cancer patients. Exogenous CD109 has been identified as a component of exosome secreted from transfected 293 cells, making it a promising target for exosome diagnosis [66]. Still, to date CD109 have not been reported in exosomes derived from SCC or other tumor cells. The underlying mechanisms of CD109 packaging into exosomes deserve more detailed investigation.

In addition to the potential as a biomarker, CD109 might be a target for therapeutic approaches. CD109 is a membrane protein [3], which can be targeted directly by specific antibodies or enzymes. CD109 might also be recognized by targeted vehicles for drug delivery. However, the detailed roles of CD109 in pathogenesis of SCCs are still unclear. For example, the relationship of lower expression of CD109 and undifferentiated SCCs needs to be defined. CD109 intervention might only be considered in clinical practice after the risks and benefits are evaluated carefully.