The conserved EGF-like domain and the two follistatin-like domains of TMEFF2 have attracted the attention of most studies. Although the substitution of Arg39 with His in the EGF-like domain strongly reduces its affinity for EGFRs, Yonezawa et al. [11] observed TMEFF2 mildly activating HER-4 receptors in MKN28 gastric cancer cells. This raises the question of whether a mutation in the EGF-like domain could change its functionality. This is unlikely; for example, Ali and Knauper [5] reported that the addition of a soluble TMEFF2 ectodomain to HEK293T cells led to an increase in ERK1/2 phosphorylation, an effect vanishing dramatically upon treatment with the HER1 inhibitor AG1478. On the other hand, the same study also reported that the treatment of HEK293T cells with a TMEFF2 mutant harbouring only the EGF-like domain failed to induce an increase in cell proliferation. Similarly, at least three additional studies [6,7,8] have reported that the deletion of the EGF-like domain of TMEFF2 had no effect on its reported function (Figure 2).

Figure 2. TMEFF2 domain functionality. SP: signal peptide, FLS: follistatin-like domain, EGF: EGF-like domain, TM: transmembrane domain. The N-terminal portion of TMEFF2 with signal peptide and follistatin-like domains is deemed crucial for impairment of adrenocorticotropic hormone (ACTH) production through inhibition of corticotropin release hormone (CRH) signalling in glioma cells, neuroprotection in Alzheimer’s disease, binding with PDGF-AA and induction of proliferation in HEK293 cells, whereas the EGF-like domain is dispensable for all of these functions. The stalk sequence (residues 303–320) is essential for ADAM17-mediated ectodomain shedding and harbours 2 potential cleavage sites for Type II transmembrane serine proteases (TTSPs). TTSPs like matriptase-1 and hepsin can cause TMEFF2 shedding independent of stalk region. Proteolytic processing of the ectodomain and remaining stub could impact TMEFF2 function. The TMEFF2 intracellular region is necessary for interaction with SHP-1 in gastric cancer cells and TMEFF2-induced impairment of cell migration in prostate cancer.

Turning the attention to the follistatin-like domain, follistatin was initially reported as an activin binding protein in the developing oocyte [57]. Activin is one of the 33 members of the transforming growth factor β (TGF-β) family known mainly for its role in augmenting the gonadotropin-releasing hormone (GnRH)-mediated release of follicle stimulating hormone (FSH) from the anterior pituitary. Follistatin was originally known to be a non TGF-β family member which could bind and inhibit the actions of activin. Later, it was discovered that follistatin is expressed by a number of tissues throughout the body and binds other members of the TGF-β family. In addition, it plays a crucial part during embryogenesis, and follistatin-null mice are smaller in size and die soon after birth due to impaired breathing [57]. However, the role of the follistatin-like domain in the follistatin module-containing proteins is not clear. One follistatin module-containing protein (osteonectin) has been reported to bind and neutralize both platelet derived growth factors (PDGFs) and vascular endothelial growth factor (VEGF) [8]. Similar to osteonectin, Lin et al. [8] reported that TMEFF2 could bind PDGF-AA both in solution and on the cell surface. They also showed this binding to prevent PDGF-AA from inducing cellular proliferation in NR6 fibroblasts. Furthermore, the ability of TMEFF2 to prevent PDGF-AA-induced cellular proliferation was abolished upon deletion of its follistatin-like domains. Huang et al. [54] likewise reported binding of TMEFF2 with PDGF-AA in HEK293T cells. In this interesting study, Tmeff2 was found to be specifically upregulated in cells of oligodendrocyte lineage during the differentiation stage. As PDGF-AA had already been shown to prevent this differentiation, it was hypothesized that TMEFF2 could induce oligodendrocyte differentiation by binding and neutralizing PDGF-AA. However, this hypothesis was proven wrong as Tmeff2 knockout mice did not exhibit any defects in oligodendrocyte differentiation. Also, the addition of TMEFF2 to the media containing HEK293T cells could not prevent PDGF-AA-induced phosphorylation of ERK1/2. On the other hand, TMEFF2 overexpression in HEK293T cells itself caused ERK1/2 phosphorylation, pointing towards a pro-proliferative role. Similarly, the addition of recombinant TMEFF2 ectodomain to primary cultures of hippocampal and mesencephalic neurons was reported to enhance their proliferation in a dose-dependent manner [4].

Employing elaborate gene correlation and ontology analyses over multiple platforms (based on the correlation of gene expression in uterine corpus endometrial carcinoma), Gao et al. [58] suggested TMEFF2 to be involved in a diversity of functions; the notable ones include metabolism, embryonic development, cytoskeletal binding, extracellular matrix binding, chromatin binding and the interaction of RNA polymerase II with DNA. Similarly, wide-ranging signalling pathways were reported to involve TMEFF2, which include cGMP-PKG, Hippo, Hedgehog, MAPK and phospholipase D signalling. A closer look at the data indicates that the leading TMEFF2 function is related to neurons and nervous system physiology. The TMEFF2-related gene set comprises the cellular components of both axons and dendrites as well as the pre- and postsynaptic components, GABA signalling and ion channel complexes [58]. Conversely, TMEFF2 was reported to exhibit oncogenic properties in endometrial carcinoma (EC), where its expression increased gradually through clinical stages, nodal metastasis and de-differentiation of endometrial tissue [58]. In addition, knocking down TMEFF2 in endometrial carcinoma cells (Ishikawa) resulted in reduced cell proliferation, invasion and migration, inhibition of MAPK and PI3K-AKT pathways, and downregulation of the signature proteins of epithelial to mesenchymal transition (EMT). The mechanism behind TMEFF2-driven inhibition of MAPK and PI3K-AKT pathways remain unexplored. Similarly, Li et al. [59] implicated TMEFF2 in the development of drug resistance in ovarian cancer cells through its positive regulation by the long noncoding RNA TMPO-AS1 (TMPO-AS1 ⟞ miR-200c ⟞ TMEFF2).

On the other hand, when compared to adjacent normal tissue, TMEFF2 mRNA was downregulated in pancreatic cancer (72 patients) [48], and an inverse correlation was observed between TEMFF2 protein in pancreatic cancer tissue and the clinical stage of the disease. The study also reported that TMEFF2 overexpression in AsPC1 and Panc1 pancreatic cancer cells (i) impairs cell proliferation both in vitro and in vivo (xenograft), (ii) decreases their invasion and migration capacities, and (iii) downregulates EMT signature proteins. These characteristics were attributed to suppression of the p38 MAPK signalling pathway by TMEFF2. The study also claimed that the experimental overexpression of TMEFF2 reduced phosphorylation of RAS, RAF, MEK and ERK in both AsPC1 and Panc1 cells. The onco-suppressive role of TMEFF2 in pancreatic cancer was also observed by Li et al. [60] (the study reports positive regulation of TMEFF2 by LINC01963, LINC01963 ⟞ miR-641 ⟞ TMEFF2). Fan et al. [61] presented similar findings, where TMEFF2 overexpression led to an increase in apoptosis and a decrease in proliferation, cell migration and invasion capacities of the lung cancer cell line A549.

Likewise, not only was TMEFF2 expression downregulated in glioma but also its mRNA levels decreased with increasing grade and worsening treatment outcomes [8]. In support of the onco-suppressive role of TMEFF2, Labeur et al. [6] reported that TMEFF2 protein levels decreased in cotricotropinomas. TMEFF2 overexpression also caused a marked decrease in proliferation of AtT20 human glioma cells when tested through both short-term (WST-1 assay) and long-term (soft agar colony formation) cell proliferation assays. Furthermore, TMEFF2 inhibited the production of adrenocorticotropic hormone (ACTH) in response to CRH (corticotropin releasing hormone) stimulation in AtT20 glioma cells; this action was demonstrated to be the result of TMEFF2-imparted action at multiple levels of CRHR1 signalling. Notably, while the EGF-like domain, the intracellular domain and transmembrane domains were dispensable, the follistatin modules were deemed essential for the ability of TMEFF2 to impart the abovementioned effects.

Another interesting role of TMEFF2 in the nervous tissue was reported by Hong et al. [7]: TMEFF2 could bind amyloid-β protein oligomers (AβO) with high affinity and protected N2a cells from the neurotoxic effects of AβO. Amyloid-β protein (Aβ) self-aggregates into amyloid fibrils and causes amyloid plaque formation, which is a hallmark of Alzheimer’s disease. Amyloid-β protein is derived from proteolytic processing of the amyloid-β protein precursor (AβPP). In addition to fibrillar forms, amyloid-β protein exists in various smaller assemblies, called amyloid-β oligomers (AβO). TMEFF2 binding was not limited to the oligomers (AβO) but could directly bind the precursor AβPP and mature Aβ. Moreover, this binding was observed in solution, intracellularly and in mouse brain. Immunostaining revealed that Tmeff2 was upregulated in neurons exhibiting excessive amyloid-β intraneuronal accumulation, therefore suggesting a neuroprotective role of the protein in vivo. Considerable TMEFF2 immunostaining was also observed in astrocytes. In addition to binding and neutralization of AβO, the neuroprotective role of TMEFF2 was suggested to be the result of its neurotropic actions. MC65 neuroblastoma cell immunofluorescence showed TMEFF2 as a granular cytoplasmic appearance, suggesting its rapid internalization. Once again, removing the EGF-like domain did not have any effect on the neuroprotective role of TMEFF2.

7. TMEFF2 in Prostate Cancer

A substantial proportion of TMEFF2 literature focuses on its role in prostate cancer. Such studies greatly enhance the knowledge on TMEFF2, but at the same time, they impart perplexity about its function, the oncogenic vs. onco-suppressive nature of TMEFF2, the range of tissue expression as well as the possibility of TMEFF2 being a biomarker and/or a therapeutic target.

Numerous studies regarding the role of TMEFF2 in prostate cancer revolve around its progression from the androgen-dependent to androgen-independent stages. As mentioned earlier, Glynne-Jones et al. [12] identified TENB2 while studying the differential gene expression between androgen-dependent and androgen-independent prostate cancer xenograft models. In addition, they also reported that TMEFF2 expression was markedly increased in prostate carcinoma vs. benign prostate hyperplasia and that its mRNA levels increased in higher-grade prostate cancers. These findings point towards an oncogenic role of TMEFF2 in prostate cancer. Such an inference is reinforced by several subsequent studies: for example, Afar et al. [47] reported that both protein and mRNA levels of TMEFF2 increased significantly in prostate cancer compared to normal tissue. The authors also developed an antibody-drug conjugate (ADC) named Pr1-vcMMAE. Compared to the control ADC, Pr1-vcMMAE effectively reduced the tumour size in LNCap prostate cancer xenografts with minimal toxicity. The humanized version of ADC, hPr1-vcMMAE, gave similar results when used against the CWR22 prostate cancer xenografts. In what appears to be a follow-up study, Zhao et al. [15] used 2H8 tomoregulin mAb (developed by Uchida et al. [14]) radiolabelled with the β-emitting isotope ⁹⁰Y and reported a significant size-reduction in the LNCaP xenografts (compared to radiolabelled IgG1). A substantial upregulation of tomoregulin mRNA and protein in prostate cancer tissue was also reported in this study. The injection of TMEFF2 mAb on its own did not have any effect on xenografts, and there was neither a difference in proliferation reported in the tomoregulin-stable PC3 prostate cancer cells nor any difference observed in the proliferation of LNCap or PC3 cells when the extracellular domain of tomoregulin was added to the cell cultures. Similarly, Zhao et al. [52] complexed the TMEFF2 mAb (2H8) with a goat anti-mouse IgG antibody linked to the saporin toxin (Mab-ZAP). When added to TMEFF2-overexpressing PC3 prostate cancer cells (TMEFF2-PC3), Mab-ZAP induced a 50% reduction in the proliferation of TMEFF2-PC3 cells while the native PC3 cells remained unaffected. Again, TMEFF2-stable PC3 cells showed no difference in cell proliferation compared to the control PC3 cells. These studies point towards TMEFF2 exhibiting oncogenic properties in prostate cancer, with higher expression in prostate cancer vs. normal prostate and further upregulation upon development of androgen independence. Moreover, ectopic overexpression of TMEFF2 had no detrimental effect whatsoever on the proliferation of prostate cancer cells in culture or in xenografts.

However, other studies point towards an altogether opposite function. Gery et al. [46] reported that TMEFF2 was expressed in androgen-dependent LNCap cells but not in the androgen-independent DU145 and PC3 prostate cancer cells. Moreover, reduced TMEFF2 mRNA levels were observed in the androgen-independent prostate cancer xenograft models (LAPC9-AI, LAPC3-AI and LAPC4-AI) when compared to their androgen-dependent counterparts. Also, upon treatment of LNCaP cells with dihydrotestosterone (DHT), TMEFF2 expression increased in both time and dose-dependent manners. In addition, when an androgen-dependent prostate cancer xenograft model, CWR22, was propagated in castrated male mice and TMEFF2 expression vanished. Importantly, when TMEFF2 was ectopically expressed in DU145 and PC3 cells, it impaired their proliferation significantly; although TMEFF2 was upregulated by androgens, it exhibited antiproliferative effects in prostate cancer. Chen et al. [19] reported the steady state levels and the responsiveness of androgen receptors (AR) to have a direct correlation with TMEFF2 (mRNA and protein) expression in the xenograft derived from LNCap cells. Further corroboration of the onco-suppressive role of TMEFF2 in prostate cancer comes from the work of Georgescu et al. [62], who reported a positive correlation between high expression levels of TMEFF2 mRNA and recurrence-free survival of prostate cancer patients. The study identified a set of 11 genes downregulated by TMEFF2 and designated them as TMCC11 (TMEFF2 modulated cell cycle 11). The native expression levels of TMCC11 increased upon treatment of LNCap cells with DHT, and endogenous TMEFF2 attenuated the DHT-driven upregulation of TMCC11, thus hampering the pro-growth effects of DHT in these cells. The study establishes high TMCC11 expression as a strong predictor of prostate cancer progression and recurrence, where C-statistics of the TMCC11 signature were reported to outperform the Gleason and pathological scores in one of the four datasets tested [62].

It is indeed intriguing to notice numerous prostate cancer studies reporting opposing data regarding the relation of TMEFF2 to androgens and its onco-suppressive versus potentially oncogenic role. A study by Mohler et al. [18] could offer an explanation for the former. In this study, the CWR22 androgen-dependent xenograft was implanted in the male host. After propagating the xenograft in male mice, the host was castrated, and the xenograft was left in the castrated host for 5 months (in order to find genes involved in the reoccurrence of prostate cancer). Eventually, an androgen-independent phenotype emerged which showed foci of highly proliferating cells. As expected, TMEFF2 was observed to be downregulated upon castration of the hosts. However, both TMEFF2 mRNA and protein were upregulated in the androgen-independent CWR22 xenograft. Importantly, when two subsets of mice were treated with testosterone propionate at 6 and 20 days after castration, TMEFF2 was found upregulated upon androgen treatment. In conclusion, for these authors, TMEFF2 is an androgen-regulated gene and its expression was abolished upon cessation of androgens, but it reappeared with even higher intensity in androgen-independent recurrent prostate cancer. This could, in part, offer an explanation for the apparent discrepancies in the TMEFF2 expression levels upon development of androgen independence in prostate cancer xenograft models. However, the marked expression of TMEFF2 in androgen-dependent LNCap cells and its absence in the androgen-independent DU145 and PC3 prostate cancer cells as well as the absence of TMEFF2 expression in the primary culture of human normal prostatic epithelial (PrEC) cells remain intriguing [15].

The conflicting oncogenic versus onco-suppressive role of TMEFF2 in prostate cancer literature is more complex to untangle. We will discuss some relevant studies and provide some possible explanations for these disagreements. The first study originates from María Ruiz-Echevarría’s group regarding the role of upstream open reading frames (uORFs) in the leader region of TMEFF2 mRNA [21]. The uORFs are short open reading frames present upstream of the translation initiation sites in 5–10% of cellular mRNAs. uORFs can modulate the translation of mRNAs by hampering the translation of downstream coding regions, and more intriguingly, under cellular stress conditions, they can promote translation of the coding region. Under cellular stress, phosphorylation of eukaryotic translation initiation factor 2 (eIF2α) leads to a decrease in global translation, although the translation of certain genes (containing uORFs in their mRNA), such as ATF4 (activating transcription factor 4), which is capable of initiating both pro-growth and pro-death signals [63], is enhanced [64]. TMEFF2, as ATF4 mRNA, contains multiple uORFs. Echevarría’s group [21] established Gaussia luciferase (GLuc) constructs where TMEFF2 5′-uORFs were deleted one by one and placed under a cytomegalovirus (CMV) promoter. In both the androgen-dependent 22Rv1 prostate cancer cells and the androgen-independent PC3 cells, the removal of all four uORFs enhanced the translation of TMEFF2 mRNA; however, when DHT was added to the 22Rv1 cell culture, a concentration-dependent increase in luciferase activity was only observed in the construct harbouring all four uORFs. Thus, although the uORFs in TMEFF2 mRNA hampered the basal transcription of the gene, they were necessary for the androgen-dependent upregulation of TMEFF2. These authors also reported an enhanced eIF2α-dependent translation of TMEFF2 upon treatment with thapsigargin, a drug used to induce ER stress. This suggests that TMEFF2 is a stress-induced gene. There is a reasonable likelihood that, similar to other stress-related proteins, TMEFF2 could function as both an adaptive and apoptotic protein, which in turn would depend upon its expression level and/or posttranslational processing. The growth promoting vs. growth retarding effects of TMEFF2 in prostate cancer cell lines could therefore be decided by the degree of its expression and posttranslational modification state.

Vera Knauper’s group [5] actively searched for answers to the discrepancies between the oncogenic versus onco-suppressive roles of TMEFF2 in prostate cancer. The group specializes in investigating a family of membrane-bound and secreted metalloproteases known as ADAMs (a disintegrin and metalloprotease). One of the many functions of ADAMs is protein ectodomain shedding, a phenomenon where extracellular portions of membrane proteins are cleaved off by proteases and released into the surrounding ECM (or blood). In the case of the membrane bound ligands, chemokines and cytokines, ectodomain shedding leads to the release of the ligand from the cell surface, enabling them to interact with receptors in endocrine, autocrine or paracrine manners. A prime example of this phenomenon is the well-documented ectodomain shedding of the TNFα ligand by ADAM17 [65]. Owing to the previously reported ectodomain shedding of TMEFF2 upon treatment of A172 glioma cells with TNFα and interleukin-1 [66], Knauper’s group investigated TMEFF2 as a substrate of ADAM proteases. Upon treatment of TMEFF2-stable HEK293T cells with phorbol ester α (PMA), the TMEFF2 ectodomain was released into the conditioned media. Moreover, the PMA-induced ectodomain shedding of TMEFF2 was either reduced or abolished upon inhibition of ADAM17. Importantly, TMEFF2 was shown to increase the proliferation of HEK293T cells both through overexpression and through the addition of soluble TMEFF2 ectodomain in the media. However, the simultaneous addition of ADAM17 inhibitor (TAP-1) and TMEFF2 overexpression reversed this increase in HEK293T cells proliferation, that is, the pro-proliferative effect of TMEFF2 was exclusive to its shed ectodomain. Ectodomain shedding can sometimes lead to activation of receptors through a phenomenon known as regulated intramembrane proteolysis (RIP). Here, after cleavage of the ectodomain, the remaining stub is cleaved within the membrane through an intramembrane-cleaving protease. The resulting fragment thus generated participates in subsequent signalling. Upon intramembrane cleavage of the Notch receptor, for example, the liberated intracellular domain is capable of entering the nucleus and regulating the specific target genes [67]. Another important finding in this [5] and a subsequent study by Knauper’s group [17] was the phenomenon of processing the remaining membrane bound stub of TMEFF2 after its ectodomain was shed. TMEFF2 was demonstrated to be a substrate of several proteinases including γ-secretases, matriptase-1, matripase-2 and hepsin, with each enzyme being able to shed the TMEFF2 ectodomain (sometimes through combined sequential action) and, notably, cleaving the remaining membrane bound stub into specific fragments, distinct for each enzyme. Moreover, it was proposed that the fragments produced through the action of various proteases would function differently. The authors suggested it as a mechanism which would decide the oncogenic versus onco-suppressive activities of TMEFF2 depending on the enzymes available to process TMEFF2, which in turn would vary according to the cell type and/or the cellular and tissue microenvironment.

Ruiz-Echevarria’s group reported similar findings where overexpression of the full-length TMEFF2 impaired proliferation of HEK293T cells, sensitised the HEK293T cells for apoptosis by staurosporine and reduced colony formation in soft agar by approximately 5 fold [53]. In contrast, when only the TMEFF2 ectodomain was added to media of HEK293T cells and RWPE1 prostate cancer cells, it led to an increase in cell proliferation. Therefore, the intracellular domain would be necessary for onco-suppressive effects, whilst the shed extracellular domain would be a ligand enhancing cell growth. In addition, the overexpression of full-length TMEFF2-MYC-HIS in HEK293T cells was reported to bind and enhance the activity of sarcosine dehydrogenase (SRDH). Furthermore, a doxycycline-inducible TMEFF2-RWPE1 cell line was generated, and upon co-treatment with sarcosine and doxycycline, TMEFF2 reversed the cell migration enhancement effect of sarcosine. In a related study [20], Ruiz-Echevarria’s group observed that the TMEFF2-induced reduction in cell migration was not merely due to a decrease in the intracellular sarcosine concentration by enhanced SRDH but rather the metabolism of sarcosine was the key factor in this process, more specifically the one-carbon metabolism. One-carbon metabolism comprises a group of chemical reactions where one carbon group is transferred from one metabolite to another. During this process, a carbon carrier is needed, tetrahydrofolate being the most common one. Sarcosine and dimethylglycine are two of the main one-carbon donors in one-carbon metabolism pathways. When the antifolate drug methotrexate (MTX) was used, it abolished the increase in cell migration induced by TMEFF2 knockdown [20]. TMEFF2 knockdown also reduced the mRNA levels of dimethylglycine dehydrogenase (DMGH) and glycine decarboxylase (GLDC), which are important contributors to one-carbon metabolism.

Chen et al. [22] placed their emphasis on the role of potential GPCR signalling motif in the C-terminal portion of TMEFF2 as the putative deciding factor behind its onco-suppressive function in prostate cancer. The group cloned a deletion mutant lacking 13 consecutive basic-rich amino acids in the C-terminus of the protein (TMEFF2_ΔGA) and compared it to the full-length FL_TMEFF2. The overexpression of FL_TMEFF2 in RWE2 prostate cancer cells (i) reduced the expression of integrins αv, β1 and β3; (ii) inhibited RAS homologue family member A (RHOA) activation; (iii) impaired phosphorylation of the focal adhesion kinase (FAK); and (iv) promoted cell rounding, thus preventing lamellipodia protrusions leading to decreased migration of cells. On the other hand, the overexpression of TMEFF2_ΔGA failed to accomplish any of the above.

8. TMEFF2 in Gastric and Colorectal Cancers

Further corroborating evidence on the onco-suppressive role of TMEFF2 came from studies in gastric and colorectal cancers. As mentioned earlier, Sun et al. [41] reported significant downregulation of TMEFF2 through various stages of gastric cancer, which directly correlated with its promoter methylation and worse patient survival. This study identified five TMEFF2 interacting proteins in AGS gastric cancer cells, including tyrosine-protein phosphatase non-receptor type 6 (SHP1), Ras GTPase-activating protein-binding protein 1 (G3BP1), heterogeneous nuclear ribonucleoprotein K (HNRPK), splicing factor 1 and ubiquitin carboxyl-terminal hydrolase 4 (USP4). The interaction of TMEFF2 with SHP1 in AGS and MKN45 gastric cancer cells resulted in cell cycle arrest, induction of apoptosis and a decrease in cell proliferation both in vitro and in vivo (xenograft models). This interaction between TMEFF2 and SHP1 was abolished upon deletion of the TMEFF2 intracellular domain. Moreover, no mutations were found in the TMEFF2 intracellular domain in gastric cancer patients (TGCA data). These findings corroborate the previously mentioned importance of the TMEFF2 intracellular domain as well as the fact that promoter methylation rather than mutation is the primary cause of TMEFF2 loss of function. The investigators also proposed that TMEFF2 had a role in maintaining genomic integrity as TMEFF2 knockdown in GES-1 gastric cancer cells led to an increase in DNA damage. DNA damage-related genes were found to be enriched in TMEFF2 overexpressing AGS cells and gastric tumours with high TMEFF2 expression levels.

In a later study [42], the same research group investigated the role of TMEFF2 in Helicobacter pylori-induced gastric cancer development using cell cultures, mouse models and human gastric tissue pathology. H. pylori infection resulted in both upregulation and phosphorylation (activation) of STAT3, which repressed TMEFF2 expression after binding to its promoter. On the other hand, TMEFF2 overexpression decreased STAT3 phosphorylation in AGS gastric cancer cells both in culture and in xenografts (TMEFF2 overexpression also resulted in a dramatic decrease in the tumour size). TMEFF2-induced downregulation of p-STAT3 was demonstrated to be dependent on its interaction with SHP1.

Further evidence for the involvement of STATs in TMEFF2 function came from David Shibata’s group. Firstly, Elahi et al. [55] reported that TMEFF2 induced apoptosis in HCT-116 colorectal cancer cells and reduced their proliferation through activation of the STAT1 pathway. Interestingly, when TMEFF2-stable HCT-116 colorectal cells were subjected to microarray analysis, 18 of the 42 upregulated genes were found to be interferon inducible. These interferon-inducible genes exhibited the highest fold change among the list of upregulated genes. The treatment of HCT-116 cells with STAT1 siRNA reversed TMEFF2-induced reduction of proliferation both on 2D and 3D cultures. In a subsequent study by the same group [56], it was reported that the TMEFF2 overexpression in HCT-116 cells led to upregulation of genes with both GAS and ISRE harbouring promoters (GAS to a higher degree than ISRE). Additionally, phosphorylation and subsequent nuclear localization of both STAT1 and STAT2 were observed in stably transfected TMEFF2 overexpressing HCT-116 cells. STAT phosphorylation is mediated by four kinases of the Janus family (JAK1, JAK2, JAK3 and TYK2), and TMEFF2 overexpression resulted in the upregulation of JAK1 and JAK2 but not the other two kinases. The HCT-116 cell line is resistant to interferon therapy (like several other cancers), and an interesting observation from this study [56] was that TMEFF2 overexpression marginally sensitized HCT-116 cells to INF-α but not INF-γ. In their latest study [68], David Shibata’s group observed the onco-suppressive and STAT1-inducing actions of TMEFF2 in HCT-116 cells to be dependent upon its proteolytic shedding (ADAM17 was identified as TMEFF2 shedase in HCT-116 cells).

9. Conclusions and Future Perspectives

Despite discrepancies, the overall literature points towards a context specific role of TMEFF2 when it comes to its oncogenic versus onco-suppressive function, with the latter predominating the former. Nevertheless, two factors should be taken into consideration while studying TMEFF2. Firstly, although functional, the expression levels of TMEFF2 in the majority of tissues and cell lines are very low. Even in prostate tissue where TMEFF2 is highly expressed, the inter- and intra-tissue variability is very high both in normal and cancerous tissue [15,62]. This aspect could affect the interpretation of data where the expression levels of TMEFF2 are assayed in tissue samples. Secondly, when it comes to studies involving cell lines, the cumulative evidence suggests that the extent and duration of experimental overexpression/knockdown could have significant effects on the outcome of these experiments. Similarly, for better consistency and reproducibility, cell culture and related transfection/transduction conditions should be factored in while interpreting the results of both past and future studies.

Two aspects of the posttranslational processing of TMEFF2 remain largely unexplored and are of interest in future studies. The first one is the phosphorylation of TMEFF2. Initial identification and characterization studies predicted one potential phosphorylation site on TMEFF2 just preceding the EGF-like domain [12]. However, the PhosphoSitePlus^® (PSP) database [69] reveals eight potential phosphorylation sites in TMEFF2 (Y202, T231, Y309, Y319, T358, Y361, S362 and S363). Differential phosphorylation could indeed have an impact on TMEFF2 function, especially under stress conditions. The second aspect of TMEFF2 posttranslational modification which remains to be investigated is its glycosylation. As the protein has both O-linked glycosaminoglycans and N-linked glycans attached, it could very well be possible that differential glycosylation of TMEFF2 affects its function.

In conclusion, TMEFF2 is methylated in several cancers and could potentially be used as both a prognostic and a diagnostic marker. TMEFF2 is expressed in many tissues, the highest levels being observed in brain and prostate. The oncogenic vs. onco-suppressive function of TEMFF2 is decided by numerous factors including cell type, expression level, proteolytic processing and posttranslational modification state. Thus, TMEFF2 is an intriguing protein with multifaceted functions, the physiological role of which remains to be elucidated.