Research ArticleCancer

Reverse signaling by semaphorin 4C elicits SMAD1/5- and ID1/3-dependent invasive reprogramming in cancer cells

See allHide authors and affiliations

Science Signaling  20 Aug 2019:
Vol. 12, Issue 595, eaav2041
DOI: 10.1126/scisignal.aav2041

Semaphorins in reverse

Members of the large family of semaphorins guide cell migration. As ligands, some semaphorins are implicated in promoting cancer progression and angiogenesis. Gurrapu et al. found an alternative mechanism in which semaphorin 4C (Sema4C) acts as a receptor to promote metastasis. In response to interaction with the extracellular domain of PlexinB2 (which is a receptor when Sema4C acts as a ligand), transmembrane-resident Sema4C intracellularly interacted with and activated TGF-β receptors in invasive breast cancer cell lines. Subsequent changes in gene expression suppressed mesenchymal features and promoted metastasis in a mouse model of disseminated-cell seeding and growth. These findings reveal one way in which Sema4C promotes the cellular plasticity necessary for metastatic disease.


Semaphorins are a family of molecular signals that guide cell migration and are implicated in the regulation of cancer cells. In particular, transmembrane semaphorins are postulated to act as both ligands (“forward” mode) and signaling receptors (“reverse” mode); however, reverse semaphorin signaling in cancer is relatively less understood. Here, we identified a previously unknown function of transmembrane semaphorin 4C (Sema4C), acting in reverse mode, to elicit nonconventional TGF-β/BMP receptor activation and selective SMAD1/5 phosphorylation. Sema4C coimmunoprecipitated with TGFBRII and BMPR1, supporting its role as modifier of this pathway. Sema4C reverse signaling led to the increased abundance of ID1/3 transcriptional factors and to extensive reprogramming of gene expression, which suppressed the typical features of the epithelial-mesenchymal transition in invasive carcinoma cells. This phenotype was nevertheless coupled with burgeoning metastatic behavior in vivo, consistent with evidence that Sema4C expression correlates with metastatic progression in human breast cancers. Thus, Sema4C reverse signaling promoted SMAD1/5- and ID1/3-dependent gene expression reprogramming and phenotypic plasticity in invasive cancer cells.


Tumor development depends on multiple genetic and epigenetic changes, endowing abilities to grow in aberrant manner, surmount normal tissue boundaries, disseminate through the circulation, and colonize distant organs (1). In particular, metastases account for 90% of cancer deaths (2), yet mechanisms facilitating progression from locally invasive to metastatic tumors remain largely unknown. The metastatic spreading of malignant cells is a multistep process (3), which is only partly linked to the events sustaining primary tumor development. An in-depth understanding of mechanisms regulating tumor metastasis holds great promises for prognostic assessment and therapeutic intervention. Previous studies have highlighted the critical steps in carcinoma metastasis formation, starting from the acquisition of invasive properties at the primary tumor site, through a process known as epithelial-mesenchymal transition (EMT), to allow migration through the basement membrane and intravasation in lymphatic and blood vessels. However, once in the circulation, cancer cells need to reattach to vessel endothelium and extravasate at a secondary site, such as lung or liver, where they can seed micrometastatic colonies and lastly grow overt tumor lesions (2). EMT is often considered the main process leading to cancer cell dissemination from the primary site, and one of its major features is the loss of expression of the cell adhesion protein E-cadherin. EMT-inducing transcription factors Twist, Snail1, Snail2, Zeb1, and Zeb2 can independently suppress E-cadherin expression (4). At odds with this assumption, it is unexpected to find that carcinoma metastases often comprise cells that do not display mesenchymal features or even show overt epithelial markers, including E-cadherin expression (59). It has therefore been proposed that a complementary process named mesenchymal-to-epithelial transition (MET), characterized by E-cadherin reexpression, is often involved in metastatic colonization (1012). In particular, mesenchymal-like breast cancer cells MDA-MB231, which basally lack E-cadherin expression, gave rise to E-cadherin–positive metastatic foci in the lungs upon intravenous injection (6, 13), consistent with the idea that gene expression reprogramming and phenotypic changes in cancer cells occur during metastatic seeding and growth. Among the extracellular signals found to promote MET process are fibroblast growth factor receptor 2 (FGFR2) (14, 15), bone morphogenetic protein 2 (BMP2), (16) and BMP7 (17). Recently, it was furthermore shown that transforming growth factor–β (TGF-β) signaling could induce the expression of ID1, a transcriptional regulator known to negatively control the basic helix-loop-helix (bHLH) transcription factors, such as Twist1, and promote metastatic colonization through MET process in mesenchymal-like breast cancer cells (18). In an independent report, ID1/3 knockdown statistically significantly hampered lung metastasis formation by breast cancer cells (19). Thus, unexpectedly, TGF-β signaling plays a major role in cancer progression, reprogramming cell phenotype and behavior not only to induce EMT but also to suppress mesenchymal features, depending on tumor stage and cellular context (20).

A large group of extracellular cues called semaphorins, initially found as axon-guidance molecules, has been implicated in the regulation of cancer cell behavior and of the tumor microenvironment (21, 22). It is now understood that semaphorin signals control various types of cells other than neurons. Both cancer and stromal cells produce semaphorins and are controlled by these signals, featuring a complex cross-talk within the microenvironment that regulates tumor development and invasive, metastatic progression (23). In particular, transmembrane semaphorins can act through multiple signaling modes. When exposed on the cell surface, they can engage short-range cell-to-cell interactions with neighboring cells (24). Moreover, although they are synthesized as single-pass membrane-spanning molecules, their extracellular moiety can also be shed in soluble form and potentially act as a secreted diffusible signal. All semaphorins are known to act through the intracellular domain of the plexins, by a so-called “forward” signaling pathway, which often negatively regulates integrin-mediated adhesion and induces cytoskeletal remodeling. Moreover, transmembrane semaphorins can also mediate a “reverse” signaling mode (by acting as receptors rather than ligands) through their own intracellular domains (24, 25). In particular, it has been shown that semaphorin 4D (Sema4D) and Sema4A intracellular portions are implicated in controlling cancer cell migration, through the recruitment of Rac1 regulatory molecules T cell lymphoma invasion and metastasis 1 (TIAM1) and Scribble (SCRIB), respectively (26, 27). The cytoplasmic tail of other class 4 semaphorins terminates with a consensus sequence anchoring PDZ homology domains (2830). These protein-protein interaction domains mediate receptor clustering in neuronal postsynaptic membranes and, in general, serve as scaffolds for the assembly of multimolecular signaling complexes. Three different class 4 semaphorins have been shown to colocalize and interact with postsynaptic density protein 95 (also known as synapse-associated protein 90): Sema4C in cerebral cortical neurons (24) and Sema4B and Sema4F in hippocampal neurons (29, 30). Sema4C is a relatively less studied member of the family. Aberrant expression of Sema4C has been reported in esophageal, gastric, and colorectal cancer (31), and its expression in tumor-associated lymphatic endothelial cells was found to promote nodal metastasis (32). We have previously shown that Sema4C forward signaling, elicited by the extracellular domain of the protein and mediated by PlexinB2 and RhoA activity, is required to sustain the proliferation of diverse breast cancer cells (33). However, Sema4C reverse signaling has been poorly investigated so far, and its functions in cancer cells are unknown.


Sema4C expression is increased in metastatic breast cancer, and high expression in primary tumors predicts progression to metastatic disease

In a previously reported 70-gene classifier predicting the survival of patients with breast cancer, high SEMA4C levels were found to be statistically significantly correlated with poor outcome (34). By analyzing gene expression in a large dataset of breast cancer samples (, we found that high SEMA4C levels were statistically significantly associated with worse distant metastasis-free patient survival [n = 1747; hazard ratio = 1.28; log-rank P = 0.013]. Furthermore, the correlation between SEMA4C expression and the occurrence of metastasis was particularly prominent in the subset of grade 2 tumors (Fig. 1A), characterized by an intermediate differentiation phenotype and hardly predictable risk of metastatic progression based on histological features; in these tumors, Sema4C expression was also a predictor of overall survival, independent of Ki-67 proliferation marker in a multivariate analysis (Fig. 1B). These findings are consistent with the hypothesis that Sema4C promotes the formation of distant metastasis, which are the main cause of poor outcome in patients with breast cancer. We then used an immunohistochemistry (IHC) staining protocol to specifically assess Sema4C protein expression in a series of samples derived from invasive breast ductal carcinomas and metastatic secondary tumors (fig. S1, A and B). We observed a markedly higher expression of Sema4C in the metastatic secondary than in primary tumor samples (Fig. 1C and fig. S1C).

Fig. 1 Sema4C expression in breast cancer samples is linked to metastatic disease.

(A) Kaplan-Meier analysis (using of the distant metastasis-free survival (DMFS) of patients bearing grade 2 breast cancers split by Sema4C expression (high or low). n = 546; P = 0.0014. (B) Overall survival in grade 2 patients with breast cancer (n = 387) comparing high and low Sema4C expression subgroups. Multivariate analysis compared the distribution of MKI67 (Ki-67) expression. P = 0.025. (C) Graphical representation of the distribution of Sema4C protein expression in a panel of invasive primary breast carcinomas (n = 50) and lymph node metastases (n = 40). P = 0.000018, by chi-square method. Representative IHC images are shown in fig. S1C.

Sema4C reverse signaling reprograms gene expression and functional behavior in invasive cancer cells

We have previously shown that a forward signaling mediated by the extracellular domain of Sema4C is basally required for breast cancer cell viability, and it can promote hormonal independence and metastatic dissemination in otherwise indolent models of luminal subtype (33). Here, we investigated whether bearing different levels of Sema4C could affect the behavior and tune the metastatic potential of invasive mesenchymal-like cancer cells. Moreover, we asked whether Sema4C reverse signaling could have a role in this context. To this end, we stably overexpressed full-length transmembrane Sema4C (Sema4C-Full), its truncated membrane-associated mutant lacking the intracellular C-tail (Sema4C-TM), or its soluble extracellular domain (Sema4C-secr) (Fig. 2A) in MDA-MB231 breast carcinoma cells, which are characterized by overt mesenchymal phenotype and invasive properties. Unexpectedly, Sema4C-overexpressing cells, but not those carrying Sema4C mutants lacking the intracellular domain, phenotypically changed, starting to grow in tightly packed epithelial clusters (Fig. 2B and fig. S2). This suggested a suppression of the mesenchymal phenotype, elicited by the cytoplasmic domain of Sema4C acting in retrograde signaling mode. We then asked whether this ability of Sema4C to promote cell phenotype plasticity could be validated in a different cancer model and performed similar experiments in prostate carcinoma PC-3 cells, which underwent an apparently similar MET in response to Sema4C bona fide reverse signaling (Fig. 2C). In both invasive carcinoma models, the expression of the typical epithelial marker E-cadherin was notably induced by full-length Sema4C but not in the presence of either the secreted from or its mutant lacking the intracellular domain (Fig. 2, D and E). In addition, full-length Sema4C inhibited the expression of transcriptional factors known to induce and maintain the EMT phenotype (such as Zeb1, Zeb2, Snail, Slug, and Twist) and of the classical mesenchymal markers vimentin, fibronectin, and N-cadherin (Fig. 2, F to H). Furthermore, by in situ cell immunostaining, we could confirm the opposite regulation of differentiation markers E-cadherin and vimentin induced by Sema4C in cancer cells (Fig. 2I).

Fig. 2 Sema4C reverse signaling reprograms gene expression and promotes MET in cancer cells.

(A) Schematic representation of the expression constructs for full-length Sema4C (Sema4C-Full; the full protein containing Sema-, PSI-, and Ig-like domain along with transmembrane and intracellular domain), the truncated construct Sema4C-TM (lacking the intracellular domain), and the secretable construct Sema4C-secr (only containing the extracellular domain). (B and C) Representative microscopic images of crystal violet–stained MDA-MB231 (B) and PC-3 (C) cells transduced with mock, Sema4C-secr, Sema4C-TM, or Sema4C-Full constructs. Scale bars, 50 μm. Larger images at higher magnification are in fig. S2. (D and E) Immunoblotting and quantitative reverse transcription PCR (RT-PCR) analyses of the expression of Sema4C, E-cadherin, and vinculin in the same cells described in (B). (F to H) Quantitative RT-PCR (F and G) and immunoblotting (H) analysis of the mRNA and protein expression of the indicated EMT transcription factors and downstream EMT markers in mock- and Sema4C-Full–overexpressing cells. (I) Representative immunofluorescence images of actin, E-cadherin, and vimentin staining in Sema4C-Full–overexpressing (and mock) MDA-MB231 cells. Scale bar, 100 μm. Data (E to G) are means ± SD derived from three independent experiments; images (D, H, and I) are representative of results obtained in three independent experiments. **P < 0.01 and ***P < 0.001, by Student’s t test.

We then analyzed the impact of Sema4C expression in cancer cell behavior by performing functional assays. There was no significant difference in the proliferation rate of overexpressing cells compared to the respective controls (Fig. 3A), consistent with previous observations in other cellular models (33). However, we found a significant reduction of cell migration and invasiveness of Sema4C-overexpressing cells in Boyden chamber assays (Fig. 3, B and C). Together, these data indicated that enhanced Sema4C reverse signaling reprograms diverse invasive carcinoma cells of breast and prostate origin to lose mesenchymal features and typical EMT markers and acquire an apparently more epithelial phenotype. This was totally unexpected, considering previous reports on Sema4C forward signaling in cancer cells (33), and called for further investigation.

Fig. 3 Functional characterization of Sema4C-overexpressing cells.

(A) Growth curves of MDA-MB231 and PC-3 cells overexpressing full-length Sema4C compared with mock-transfected cells. (B and C) Graphic representation of the migratory, invasive, and wound healing capacity of stably transfected MDA-MB231 and PC-3 cells expressing full-length Sema4C compared with mock-transfected cells, assessed by Transwell chamber assays. Data are means ± SD from three independent experiments. **P < 0.01, by t test.

Broad gene expression reprogramming in Sema4C-overexpressing cells and enhanced expression of ID1 and ID3 transcriptional regulators

To get a broader picture of the transcriptional changes induced by Sema4C reverse signaling and identify potential mediators of this unexpected phenotype, we achieved genome-wide expression profiling of cells expressing mock or full-length Sema4C 1 week after gene transfer. Microarray analysis (table S1) identified 418 differentially regulated genes [with at least twofold change variation and statistically significant P < 0.05 and false discovery rate (FDR) of <0.05]. By scrutinizing the top 25 up-regulated and down-regulated genes, we noticed that Sema4C had strongly induced the expression of both ID1 and ID3 bHLH transcriptional factors (Fig. 4A), which were recently reported to switch TGF-β signaling to induce mesenchymal-to-epithelial transition in cancer cells, a phenotype unexpectedly associated with increased ability to give rise to metastasis (18). Gene set enrichment analysis (GSEA) also indicated a significant suppression of the EMT gene expression signature in Sema4C-overexpressing cells compared to controls (Fig. 4B). We validated these transcriptomic data by analyzing individual gene expression at mRNA and protein levels, and we confirmed that ID1 and ID3 levels are notably increased in cells overexpressing full-length Sema4C but not in mock cells or those expressing either its mutant lacking the intracellular domain (Fig. 4, C and D), consistent with what was seen for E-cadherin and with the observed phenotypic change.

Fig. 4 Sema4C-overexpressing cells undergo gene expression reprogramming with up-regulation of ID1 and ID3 transcriptional regulators of E-cadherin.

(A) Heat map and list of top 25 up-regulated/down-regulated genes in mock-transfected and full-length Sema4C-transfected MDA-MB231 cells. Gene expression is expressed by a log2 color scale; −2/+2 were fixed as hue saturation limits; actual values are given in table S1. (B) GSEA analysis of microarray data described in (A), showing the enrichment of EMT gene signatures in mock-transfected cells compared with full-length Sema4C-transfected cells. Normalized enrichment score (NES), 1.68; P < 0.001; FDR, 0.015. (C) Quantitative RT-PCR analysis of Sema4C, E-cadherin, ID1, and ID3 expression in MDA-MB231 and PC-3 cells mock-transfected or transfected with Sema4C-TM or Sema4C-Full constructs. (D) Immunoblotting analysis of ID1 expression in MDA-MB231 cells described in (C). Vinculin, loading control. (E and F) Quantitative RT-PCR (E) and immunoblotting (F) analysis of E-cadherin expression in the cells described in (C) upon silencing ID1, ID3, and both using siRNA. Data (C and E) are means ± SD from three independent experiments; images (D and F) are representative of results obtained in three independent experiments. **P < 0.01 and ***P < 0.001, by t test.

To assess the mechanistic relevance of ID1 and ID3 in Sema4C-induced E-cadherin up-regulation, we transiently silenced these genes, either alone or in combination (fig. S3A); we found that both ID genes are involved in this pathway because a double knockdown was required to achieve almost complete prevention of Sema4C-induced E-cadherin increased expression (Fig. 4, E and F). This functional synergism (and redundancy) of ID genes is in line with previous literature because the silencing of both ID1 and ID3 was required to effectively hinder breast cancer metastasis formation (19). The relevance of Sema4C in the regulation of ID genes in human tumors was further supported by the analysis of The Cancer Genome Atlas gene expression datasets from breast and prostate carcinomas, indicating a statistically significant correlation (fig. S4).

Sema4C reverse signaling depends on the interaction with plexins

Next, we sought to investigate whether plexins are involved in this previously unknown Sema4C reverse signaling pathway. PlexinB1 and PlexinB2 are known to bind Sema4C and act as its effectors in forward signaling cascades (33, 35). Hence, we transiently silenced PlexinB1 and PlexinB2 in cancer cells (fig. S3B) and assessed E-cadherin expression regulation by Sema4C. Knockdown of either plexin strongly curbed Sema4C-induced E-cadherin up-regulation (Fig. 5A), indicating that both molecules are involved in this pathway. Moreover, a combined targeting of both plexins achieved better reversion of E-cadherin–induced expression compared to single targeting (Fig. 5B), suggesting a partial functional redundancy of the two receptors.

Fig. 5 Gene expression regulation by Sema4C reverse signaling depends on interaction with plexins.

(A) Immunoblotting (IB) analysis of E-cadherin expression in MDA-MB231 cells expressing mock, Sema4C-Full, or Sema4C-TM constructs along with siRNAs targeting PlexinB1 or PlexinB2 or a control (Ctrl). (B) As described in (A), assessment of E-cadherin expression in MDA-MB231 cells expressing Sema4C-Full and siRNA targeting both PlexinB1 and PlexinB2 and siRNA targeting PlexinB2 alone or a control. (C) Quantitative RT-PCR analysis of the expression of mRNAs encoding Sema4C, E-cadherin, ID1, and ID3 in MDA-MB231 cells transfected with full-length or mutant (K100G-DT) Sema4C. (D) Immunoblotting analysis of E-cadherin expression in MDA-MB231 transfected with full-length or mutant (K100G-DT) Sema4C. Actin, loading control. (E) Quantitative RT-PCR analysis of the expression of E-cadherin (CDH1), ID1, and ID3 in MDA-MB231 transfected with full-length PlexinB2 (PlexinB2-Full), soluble PlexinB2 extracellular domain (PlexinB2-ecto), or a mock control. Data (C to E) are means ± SD from three experiments; images (A, B, and D) are representative of results obtained in three independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001, by t test.

To further tackle this issue, we point-mutated Sema4C at two adjacent amino acids (Lys100-Asp101) in the protein surface thought to mediate semaphorin binding to plexins (36), generating a new Sema4C-full KD100GT mutant, which lost the ability to interact with PlexinB2 in binding assays (fig. S3C). We found that the mutated Sema4C was unable to induce E-cadherin, ID1, and ID3 expression in cancer cells, despite being expressed at comparable levels as the wild type (Fig. 5, C and D). These results suggest a requirement of plexins, putatively acting as triggering ligands, to mediate Sema4C reverse signaling and the ensuing gene expression reprogramming in cancer cells.

We then asked whether an interaction with the extracellular domain of PlexinB2 may be sufficient to elicit the retrograde signaling of endogenous Sema4C in cancer cells. We found that PlexinB2 overexpression recapitulated the increased expression of ID1, ID3, and E-cadherin, observed upon Sema4C up-regulation (Fig. 5E and fig. S3D). By expressing a soluble form of the PlexinB2 extracellular domain, capable of engaging endogenous transmembrane Sema4C (fig. S3C), we selectively triggered Sema4C reverse signaling in trans, independent of the PlexinB2 intracellular domain (Fig. 5E and fig. S3D).

TGF-β/BMP receptor signaling and SMAD1/5 activation are pivotally implicated in Sema4C reverse signaling

Previous reports have highlighted the role of TGF-β signaling axis to induce ID genes promoting metastatic colonization (18). Thus, we examined the possible involvement of this pathway in Sema4C-induced ID gene up-regulation. To score for TGF-β pathway activation, we assessed the phosphorylation of SMAD effector proteins. We found a prominent increase in SMAD1/5 phosphorylation (p-SMAD1/5), but no changes in p-SMAD2/3 levels (Fig. 6A). In line with our data, it was recently shown that metastasizing tumor cells comprised a large fraction of p-SMAD1/5–positive cells, which also coexpressed ID1 (37). This suggested the ability of Sema4C to promote a selective TGF-β signaling cascade, distinct from that usually reported to induce EMT and cancer cell invasion, which is instead SMAD2/3 dependent (38). Actually, the mechanisms inducing SMAD1/5 phosphorylation are poorly understood. According to the certain studies, the canonical TGF-β receptor complex containing TGFBR2 and TGFBR1–activin-like kinase 5 (ALK5) can elicit the phosphorylation of both SMAD2/3 and SMAD1/5 (39, 40). Others concluded instead that TGF-β–induced p-SMAD1/5 depends on the activity of alternative type I receptor kinases in the complex, such as ALK2-ACVR1 (activin A receptor 1) or ALK3-BMPR1A (BMP receptor 1A) (41, 42). SMAD1/5 phosphorylation can also be induced by TGF-β–like BMPs (43); however, our transcriptomic analysis had not revealed any statistically significant variation in the levels of TGF-β or BMP factors in Sema4C-overexpressing cells (table S1). Together, these data suggest that transmembrane Sema4C elicited a distinctive mode of activation of TGF-β/BMP receptor complexes, leading to selective SMAD1/5 phosphorylation. To start addressing this mechanistic issue, we asked whether transmembrane Sema4C may interact with subunits of TGF-β receptor complexes in cancer cells and trigger their activity in a nonconventional manner. We found that Sema4C associated with TGFBR1-ALK5 kinase by coimmunoprecipitation experiments in cells overexpressing both molecules (fig. S5A); however, we were unable to confirm this in MDA-MB231 cells bearing endogenous TGFBR1 levels, possibly suggesting an indirect coupling. Actually, we detected a clear association of Sema4C with endogenous TGFBR2, the other subunit of TGF-β receptor complex (Fig. 6B and fig. S5B), as well as with other functionally associated kinase members of the type I receptor family, ACVR1-ALK2 and BMPR1A-ALK3 (Fig. 6B). On the basis of current knowledge, the activation of these kinase receptors is fully consistent with the observed Sema4C-induced SMAD1/5 phosphorylation, providing substantial bases for the cross-talk between these signaling pathways.

Fig. 6 TGF-β pathway and SMAD1/5/9 activation are pivotally implicated in Sema4C reverse signaling.

(A) Immunoblotting analysis of phosphorylated levels of SMAD1/5 and SMAD2/3 in MDA-MB231 cells transduced with mock, Sema4C-TM, or Sema4C-Full constructs under serum-starved conditions (1% FBS). Total SMAD1, internal reference. (B) Immunoblotting analysis of Sema4C protein coimmunoprecipitated (IP) with antibodies against TGFBR1, BMPR1, ACVR1, and TGFBR2. Beads only was a control. t.c.l., total cell lysate. (C) Immunoblotting analysis of the phosphorylated levels of SMAD1/5 and SMAD2/3, as well as E-cadherin expression in the cells described in (A) and treated with dimethyl sulfoxide (DMSO) (vehicle) or TGFBR inhibitor SB-431542 (5 μM) for 72 hours under serum-starved conditions (1% FBS). Total SMAD1, internal reference. (D) Immunoblotting analysis of the phosphorylated levels of SMAD1/5, as well as E-cadherin expression in Sema4C-overexpressing MDA-MB231 cells treated with TGFBR inhibitor SB-431542 (5 μM), BMPR1 inhibitor K-02288 (1 μM), or DMSO (vehicle). Vinculin, loading reference. Data (A to D) are representative of at least two experiments. (E) Quantitative RT-PCR analysis of E-cadherin (CDH1), ID1, and ID3 expression in Sema4C-overexpressing MDA-MB231 cells treated as described in (D). (F) Quantitative RT-PCR analysis of E-cadherin expression in the cells described in (A) upon transient siRNA-mediated knockdown of SMAD1 or a control (si-Ctrl). Data (E and F) are means ± SD from three experiments. *P < 0.05, **P < 0.01, and ***P < 0.001, by t test.

We thus investigated next whether inhibition of the TGF-β/BMP signaling pathway could affect ID1/3 and E-cadherin expression induced by Sema4C in cancer cells. In fact, the treatment with either an inhibitor of TGFBR2-associated kinase ALK5 (SB-431542) or, more so, a selective inhibitor for ALK2/ALK3 (K-02288) markedly impaired Sema4C-induced p-SMAD1/5 and increased expression of E-cadherin and ID1/3 (Fig. 6, C to E, and fig. S5, C and D). These findings suggested that Sema4C retrograde signaling in cancer cells is mediated by the same components of TGF-β/BMP receptor complexes with which it was found associated in coimmunopurification experiments. The mechanistic role of SMAD1 transcriptional regulator in this pathway was further demonstrated by gene knockdown experiments (Fig. 6F and fig. S5E). Together, these data indicate that Sema4C interacts on the cell surface with TGF-β/BMP receptors and can direct their activity to induce p-SMAD1/5 and downstream up-regulation of ID genes and E-cadherin.

Sema4C reverse signaling is associated with prominent metastatic colonization in vivo

It was previously shown that TGF-β signaling and ID genes are crucial for the survival and metastatic growth of disseminated breast cancer cells, whereas interfering with these genes impaired cell viability and metastasis formation (19, 4446). Moreover, TGF-β pathway is required for metastasis formation by prostate cancer cells (47, 48). Thus, we investigated the impact of Sema4C enhanced expression on the metastatic seeding and growth of cancer cells in secondary sites such as the lungs. Although MDA-MB231 and PC-3 cells are already capable of giving rise to lung metastasis in mice, gene expression reprogramming induced by Sema4C was associated with a further increase in the number of macrometastatic colonies (Fig. 7A). IHC analysis confirmed the higher expression of E-cadherin and ID1 protein in metastasis derived from Sema4C-overexpressing cells (Fig. 7B), which is consistent with previous literature indicating a causal role of these genes in promoting metastasis formation (6, 18, 19). A Kaplan-Meier analysis of the distant metastasis-free survival of patients with breast cancer indicated that high E-cadherin expression correlates with greater risk of metastasis occurrence, and this was particularly significant in the subset of poorly differentiated estrogen receptor (ER)–negative tumors (fig. S6).

Fig. 7 Sema4C reverse signaling promotes metastatic colonization in vivo.

(A) Quantification of lung macrometastases detected in NOD/SCID mice after tail vein injection with MDA-MB231 or PC-3 cells transduced with mock or Sema4C-Full constructs. n = 5 mice per condition. (B) Representative IHC images and analysis of E-cadherin and ID1 expression in lung metastasis derived from cells described in (A). Scale bar, 100 μm. (C) Quantification, performed with ImageJ, of extravasated MDA-MB231 cells transduced with mock and Sema4C-Full constructs and labeled with Vybrant-DiD fluorescent dye before tail vein injection. Bottom: Representative fields assessed are shown. (D) MDA-MB231 cells expressing mock, Sema4C-TM, or Sema4C-Full constructs, labeled with fluorescent dye (CFSE), and seeded on top of HUVEC (endothelial cell) monolayers for 30 min. The fold change of adherent cells was determined by measuring fluorescence intensity. Bottom: Representative images are shown. Data (A to D) are means ± SD from three experiments. **P < 0.01 and ***P < 0.001, by t test.

To form metastasis in distant organs, circulating cancer cells have to interact with vessel endothelium and accomplish extravasation. We tested the potential impact of Sema4C signaling on this function by tracking fluorescent-labeled MDA-MB231 cells injected in the circulation. We found that a larger number of cancer cells could be found to seed micrometastatic foci in the lungs, 48 hours after injection, upon Sema4C overexpression (Fig. 7C). In addition, we observed that Sema4C-overexpressing cells more efficiently and quickly adhered to a monolayer of human umbilical cord endothelial cells (HUVECs), compared to controls (Fig. 7D); thus, enhanced adhesion to vessel wall could be a mechanism responsible for Sema4C-induced tumor cell extravasation. Moreover, this acquisition was not observed when the cells expressed Sema4C truncated construct unable to mediate the reverse signaling cascade (Fig. 7D). Together, the data suggest that Sema4C reverse signaling sustains the metastatic process by impinging on TGF-β/BMP receptor signaling cascades in invasive cancer cells and eliciting wide gene expression reprogramming, including the induction of ID1 and ID3 prometastatic genes.


Although innovative and targeted drugs have remarkably improved the efficacy of treatments for primary tumors, metastasis remains the leading cause of death in patients with cancer. Understanding the molecular players controlling metastatic dissemination is crucial to predict the risk of developing a systemic disease, as well as to design novel approaches for prevention and targeted therapy. Several studies supported the idea that an EMT, characterized by loss of E-cadherin expression, is a key step in the metastatic process (49). However, recent studies show that most metastatic cells display increased E-cadherin levels, suggesting that this is linked to metastasis formation and it may be regained through a sort of reverse MET, or simply by the acquisition of an intermediate phenotype, sometimes indicated as partial-EMT (50). It was found that pathways triggering EMT and promoting cancer cell invasiveness actually curb the ability to form distant metastasis (51). In a complementary approach, it was demonstrated that the silencing of transcription factors driving EMT does not suppress metastasis formation (18, 52); moreover, both mesenchymal-epithelial transition and E-cadherin expression in cancer cells promote metastasis development (53, 54). By analyzing a public dataset derived from primary breast cancer samples, we observed that elevated E-cadherin expression correlated with higher risk of metastasis development, especially among poorly differentiated ER-negative tumors. Together, these data underscore the importance of cancer cell plasticity and of the acquisition of epithelial traits by invasive cancer cells in the course of metastatic colonization; however, the molecular mechanisms underlying this process are far from understood, and unveiling the implicated molecular players is of paramount importance for the development of improved approaches to curtailing metastasis.

Here, we reported that transmembrane Sema4C interacts with TGF-β surface receptors and elicits a nonconventional signaling cascade, depending on Sema4C intracellular tail, leading to strong induction of ID1 and ID3 transcriptional regulators. The latter were capable of inhibiting bHLH transcriptional factors implicated in EMT, resulting in suppression of the mesenchymal phenotype and restored E-cadherin expression in cancer cells. On the basis of our data, plexins acted as putative ligand counterparts of Sema4C in this context. This conclusion was further supported by the impact of Sema4C mutagenesis in the specific protein interface known to mediate plexin binding in trans. Although in a few studies, transmembrane semaphorins and plexins were furthermore reported to interact in cis on the surface of the same cell, we have not been able to discriminate whether this is also occurring in our models.

To dissect the downstream implicated mechanisms, we analyzed the effectors of TGF-β pathway and found that Sema4C signaling (dependent on the intracellular domain) was sufficient to elicit the p-SMAD1/5 intracellular effectors. The inhibition of TGF-β receptor kinase activity could reverse Sema4C-induced SMAD activation, as well as the ensuing gene expression reprogramming. Thus, on the basis of our data, Sema4C features a previously unidentified mechanism modulating TGF-β signaling at the cell surface, by skewing the pathway toward SMAD1/5 cascade. The intracellular domain of Sema4C was essential for this process; although we currently ignore the underlying effector mechanisms, the motif recruiting PDZ domain containing proteins at the C terminus of the semaphorin could play a relevant role. Two other class 4 semaphorins, Sema4A and Sema4D, have been shown to control cancer cell migration through the interaction with intracellular regulators of Rac1/Cdc42 guanosine triphosphatase activity (26, 27). Our study reported about a semaphorin reverse signaling pathway controlling gene transcription and phenotype plasticity in cancer cells.

We found that Sema4C overexpression impaired cancer cell migration and invasiveness, without statistically significantly affecting proliferation in culture. However, consistent with data implicating ID genes in metastatic colonization (18), we found that Sema4C-high cells were notably efficient in giving rise to lung metastatic colonies compared to control cells. This previously unknown reverse signaling cascade of Sema4C, leading to gene expression reprogramming and enhanced metastatic dissemination, was presently validated in two diverse invasive tumor models: MDA-MB231 breast and PC-3 prostate carcinoma cells. We observed statistically significant correlation between Sema4C and ID1/3 expression in breast and prostate carcinoma samples, supporting the relevance of this regulatory axis in human tumors. We have previously demonstrated that Sema4C forward signaling, elicited by the extracellular domain of the semaphorin and mediated by the intracellular domain of PlexinB2 receptor, is required for survival and growth of diverse cancer cells, including MDA-MB231 (33). Thus, expectedly, Sema4C silencing impaired both the viability in culture and the metastatic capacity of MDA-MB231 cells upon tail vein injection. We concluded that Sema4C, as with other transmembrane semaphorins, is endowed with two distinct signaling modes, which could concomitantly promote tumor progression. In addition to a common forward signaling cascade promoting cancer cell viability and growth, we found that Sema4C-dependent reverse signaling adjusts the TGF-β pathway in mesenchymal-like tumor cells, leading to gene expression reprogramming and phenotypic changes that foster metastatic colonization. It was previously reported that TGF-β signaling leads to different gene regulation outcomes in different stages of the metastatic process (39). We propose that this mechanism could be enabled, with specific timing, at the level of Sema4C/TGF-β receptor complexes that we found to form in cancer cells, skewing TGF-β signals toward SMAD1/5 effector cascade and leading to ID1/3-dependent gene expression reprogramming and cancer cell plasticity.

Emerging data indicate that the phenotypic plasticity of locally invasive cancer cells and a partial loss of mesenchymal features are responsible for promoting metastatic progression (22); however, the underlying mechanisms have not been elucidated. Here, we found a previously unknown Sema4C reverse signaling cascade, dependent on SMAD1/5 and ID1/3 transcriptional regulators, which promotes gene expression reprogramming in invasive cancer cells, coupling the loss of mesenchymal features with burgeoning metastasis formation. Increased Sema4C levels in breast cancers correlate with poor patient prognosis and particularly with the occurrence of metastasis. In sum, our data underscore the importance of Sema4C bidirectional signaling in tumor progression.


Cell lines and reagents

MDA-MB231 and PC-3 tumor cell lines and COS-7 immortalized cells were obtained from the American Type Culture Collection and cultured in a humidified incubator with 5% CO2 at 37°C in Dulbecco’s modified Eagle’s medium or RPMI 1640 media supplemented with 10% fetal bovine serum (FBS), penicillin, and streptomycin, respectively. Antibodies directed against Sema4C (sc-136445), N-cadherin (sc-7939), fibronectin (sc-8422), Zeb1 (sc-25388), SMAD1 (sc-7965), ID1 (sc-488), actin (sc-1616), PlexinB2 (sc-34504), TGF-β–R2 (sc-17792), and ACTR1/ACVR1 (sc-374523) were all from Santa Cruz Biotechnology. Anti-vimentin (VIM-3B4) was from Merck. Anti–p-SMAD1/5/9 (D5B10, no. 13820) and anti–p-SMAD2/3 (D27F4, no. 8828) were from Cell Signaling Technology. Anti-vinculin (V4505) and anti-Flag (clone M2_F3165) were from Sigma-Aldrich. Anti–E-cadherin was from BD Transduction Laboratories (catalog no. 610182). Anti–TGF-β–R1 antibody was from Abcam (ab31013). Anti-BMPR1A was purchased from Proteintech (catalog no. 12702-1-AP). Cell treatments with 1 to 5 μM SB-431542 hydrate (Sigma-Aldrich) were performed for 48 to 72 hours, as described. Purified recombinant Sema4C ectodomain (catalog no. AF6125) was purchased from R&D Systems. A tumor multiple tissue array containing primary and metastatic breast cancer samples (BR1008A) was obtained from US Biomax Inc. (Rockville, MD).


Standard IHC staining protocols were applied. Briefly, tissue sections were incubated with primary antibodies for 1 hour (or overnight), followed by incubation with the secondary anti-mouse or anti-rabbit for 30 min at 37°C. Slides were analyzed using a Leica DM IRB microscope, and digital images were evaluated with METAMORPH software. For immunohistochemical analysis of lung metastasis, paraffin-embedded samples were cut in 10-μm-thick sections and probed with primary antibodies according to the standard protocols.

An anti-Sema4C immunostaining protocol was specifically set up by the use of BenchMark ULTRA IHC System and UltraView Universal DAB staining kit, according to the following notable steps: dewax at 72°C, EZ Prep 1:10; unmasking for 76 min with ULTRA CC1 (950-224); incubation with primary anti-Sema4C antibody (sc-136445; dilution 1:100) for 48 min at 37°C; detection with amplification kit; and final counterstaining with hematoxylin. In specificity validation experiments, the antibody was blocked by incubation with soluble recombinant Sema4C ectodomain (catalog no. AF6125, R&D Systems).

In the analysis of the breast tumor microarray, a semiquantitative H-score method (55) was applied to classify Sema4C expression levels as low, intermediate, or high. Briefly, we analyzed each section at ×100 microscope magnification and counted the cells characterized by different relative staining intensities (scored as no staining, weak, moderate, or strong staining). Five randomly selected microscopic fields were analyzed for each slide, and the following formula was used to calculate a combined score for each sample in a range from 0 to 300: three times the percentage value of the fraction of strongly stained cells + two times the percentage of moderately stained cells + the percentage of weakly or nonstained cells. On the basis of this global score assigned to individual cases, they were sorted into low, intermediate, or high expression, according to the following ranges: low, combined IHC H-score of 10 to 100; intermediate, combined H-score of 100 to 250; and high, combined H-score of >250.

RNA extraction and gene expression profiling

RNA was extracted using miRNeasy Mini Kit (QIAGEN), according to the manufacturer’s protocol. The quantification and quality analysis of RNA were performed on a Bioanalyzer 2100 (Agilent) using RNA 6000 Nano Kit (Agilent). Synthesis of complementary DNA (cDNA) and biotinylated complementary RNA (cRNA) was performed using the Illumina TotalPrep RNA Amplification Kit (Ambion), according to the manufacturer’s protocol using 500 ng of total RNA. Quality assessment and quantification of cRNAs were performed with Agilent RNA kits on Bioanalyzer 2100. Hybridization of cRNAs (750 ng) was carried out using Illumina Human 48,000 gene chips (HumanHT-12 v4 BeadChip). Array washing was performed using Illumina High-Temp Wash Buffer for 10 min at 55°C, followed by staining using streptavidin-Cy3 dyes (Amersham Biosciences). Probe intensity data were obtained using the Illumina GenomeStudio software (GenomeStudio v2011.1). Raw data were normalized with genome studio, according to the cubic spline normalization.

Differential gene expression analysis

Student’s t test was used to select genes with differential expression between MDA-MB231–mock– and MDA-MB231–Sema4C–overexpressing cells (P < 0.05; absolute log2 ratio, >1). This resulted in 418 unique genes being selected. Monte Carlo analysis calculated on all the possible permutation of samples evaluated an FDR of <0.05. Gene expression clustering was performed using Gene Expression Data Analysis Studio (GEDAS) (56).

Gene set enrichment analysis

The GSEA software was downloaded from The input data for the GSEA were the following: (i) a complete table of gene expression in get file format, (ii) cld files indicating sample grouping, and (iii) a catalog of functional gene sets from Molecular Signature Database corresponding to hallmarks of cancer (MSigDB version 3.0; 30 September 2010 release;; a total of 50 curated gene sets were included in the analysis. Default parameters were used to compare Sema4C versus mock cells lines, in reference to the hallmark signature database. Inclusion gene set size was set between 15 and 500, and the genes were permutated 1000 times ( (57, 58).

Real-time quantitative polymerase chain reaction analysis of gene expression

Total RNA from tumor cell lines was isolated using RNeasy Protect Mini Kit (QIAGEN) according to the manufacturer’s instructions. cDNA preparation was performed according to the standard procedures using M-MLV Reverse Transcriptase (Promega) and oligo-deoxythymine primers/random hexamers. Polymerase chain reaction (PCR) was performed by applying the following TaqMan probes: Hs00367063_m1(PLXNB2), Hs00195591_m1(SNAI1), Hs03676575_s1(ID1), Hs00357821_g1(ID1), Hs00954037_g1(ID3), s01023894_m1(CDH1), and Hs00950344_m1(SNAI2). Alternatively, PCR was conducted with SYBR Green Master Mix (Life Technologies) and run in Applied Biosystems 7900HT Fast Real-Time PCR System, by applying the following primer pairs: glyceraldehyde-3-phosphate dehydrogenase [5′-GAAGGTGAAGGTCGGAGTC-3′ (forward) and 5′-GAAGATGGTGATGGGATTTC-3′ (reverse)], Snail [5′-GACTACCGCTGCTCCATTCCA-3′ (forward) and 5′-TCCTCTTCATCACTAATGGGGCTTT-3′ (reverse)], Slug [5′-AGATGCATATTCGGACCCAC-3′ (forward) and 5′-CCTCATGTTTGTGCAGGAGA-3′ (reverse)], Zeb1 [5′-AGCAGTGAAAGAGAAGGGAATGC-3′ (forward) and 5′-GGTCCTCTTCAGGTGCCTCAG-3′ (reverse)], Twist1 [5′-TGTCCGCGTCCCACTAGC-3′ (forward) and 5′-TGTCCATTTTCTCCTTCTCTGG-3′ (reverse)], h–E-cadherin [5′-GTCACCTTCAGCCATCCTGT-3′ (reverse) and 5′-GGGTTATTCCTCCCATCAGC-3′ (forward)], h-PlexinB1 [5′-CACTGAACCCCACACCTTTC-3′ (forward) and 5′-ATAGCCACCACCTCCTCCTT-3′ (reverse)], h-PlexinB2 [5′-CTTGACCTGGGAGATGGTGT-3′ (reverse) and 5′-CTGGGGGATGATGTGGAGTA-3′ (forward)], and h-Sema4C [5′-GACACCTCCTGGCACAACAC-3′ (forward) and 5′-CCACTTCTGGGCTTCCTCA-3′ (reverse)].

Protein analysis

Total cell lysates were prepared in a tris (pH 6.8) to 2.5% SDS solution by heating at 95°C for 20 min. Protein concentration was measured using Pierce bicinchoninic acid protein assay kit as per the manufacturer’s instructions. For Western blotting, 10 to 40 μg of protein was resolved on 7.5 or 10% minigels from Bio-Rad, transferred to nitrocellulose membrane using semidry method, and immunoblotted. Bovine serum albumin (BSA) (10%) was used for filter blocking under all conditions. Immunostaining was performed with primary antibodies listed above, horseradish peroxidase–conjugated secondary antibodies, and enhanced chemiluminescence detection system.

For immunoprecipitation experiments, cells were lysed on ice in cold extraction buffer [containing 1% Triton X-100, 125 mM tris-HCl (pH 6.8), and 150 mM NaCl] adding the following phosphatase and protease inhibitors: 1 mM Na3VO4, 1 mM phenylmethylsulfonyl fluoride, 1 mM NaF, 1 mM aprotinin, 1 mM leupeptin, and 1 mM pepstatin. Cell lysates were quantified and incubated with Protein A Sepharose beads that were previously saturated in phosphate-buffered saline (PBS) with 1% BSA and coated with the corresponding antibody. Immunoprecipitation of Flag-tagged receptors was performed with 1 μg of anti-Flag (M2, Sigma-Aldrich) antibody and 1 μg of rabbit anti-mouse immunoglobulin G (IgG) (Thermo Fisher Scientific). After 2 hours of incubation at 4°C, the beads were washed thrice in immunoprecipitation buffer. Immunoprecipitated proteins were denatured by boiling in Laemmli buffer according to the standard protocols.

cDNA constructs, gene transfer, and RNA interference in mammalian cells

Full-length human Sema4C cDNA (Sema4C-Full) was obtained from the Promega Kazusa library; its mutant lacking the cytoplasmic domain (Sema4C-TM) was generated by PCR with 5′-CGTACTCGAGACATGGCCCCACACTGGGCTGTC-3′ (forward primer) and 5′-CAGGTCTAGATCACAGCCGTCTCCGCAATGACAGCACCA-3′ (reverse primer). Its mutant lacking the transmembrane and cytoplasmic domain (Sema4C-secr) was generated by PCR with 5′-CGTTGCTAGCCATGGGACCACACTGGGCTGTC-3′ (forward primer) and 5′-GCAATCCGGAGCCTGCCACGACAGCCACAAGGTAG-3′ (reverse primer). The expression constructs were subcloned into pLVX (Invitrogen) lentiviral transfer plasmid. For lentiviral particle production, calcium phosphate method was used to cotransfect transfer plasmids, packaging vectors, and constructs expressing vesicular stomatitis virus glycoprotein envelope protein in HEK-293T cells, according to the validated methods (59). Tumor cells were then transduced by incubation with lentiviral particle suspensions, in the presence of polybrene (8 μg/ml), for 8 to 12 hours.

The Sema4C cDNA expression construct obtained from the Promega Kazusa library was opened between the restriction sites Xho I and Sbf I, and the internal wild-type fragment was replaced with a mutated sequence cassette generated by two-step PCR using internal site degenerated oligos: 5′-AGACTGAGTGTATCCAGGACACGAAGAACAACCAGACC-3′ (forward) and 5′-GGTCTGGTTGTTCTTCGTGTCCTGGATACACTCAGTCT-3′ (reverse), encoding for a point mutated KG100-101 → DT amino acid sequence.

A full-length PlexinB2 cDNA expression construct was previously published (28), whereas a construct containing the secretable extracellular domain of PlexinB2 fused to Ig Fc domain (#72128), as well as constructs expressing Flag-tagged TGF-β type I receptor (#14831) and Flag-tagged TGF-β type II receptor (#31719), were obtained from Addgene and transiently transfected by Lipofectamine 2000, according to the manufacturer’s instructions.

Transient gene knockdown in tumor cells was achieved by transfection with Lipofectamine 2000 or Oligofectamine (according to the manufacturer’s protocols) of the following validated small interfering RNAs (siRNAs): control siRNA, 5′-UAAGGCUAUGAAGAGAUAC-3′; SEMA4C siRNA, 5′-CUACGUCAACAUGCUCACCUU-3′; PLXNB2 siRNA, 5′-GCUGAUGCUGCGCAGGUCTT-3′; ID1 siRNA, 5′-CUCGGAAUCCGAAGUUGGA-3′; ID3 siRNA, 5′-UCCUACAGCGCGUCAUCGA-3′; PLXNB1 siRNA, 5′-ACCACGGUCACCCGGAUUC-3′; and SMAD1 siRNA, 5′-GCAACCGAGUAACUGUGUCACCAUU-3′. Stable Sema4C expression down-regulation was achieved by lentiviral-mediated transfer of targeted short hairpin RNAs TRCN0000060694, TRCN0000060695, and TRCN0000060697 carried by puromycin-selectable constructs from Sigma MISSION library.

In situ ligand-receptor binding assay

In situ ligand-binding assay were performed as described previously (60). Briefly, COS-7 cells transiently transfected to express either full-length Sema4C or KG100-101 → DT mutated Sema4C were incubated for 1 hour at 37°C with a recombinant soluble PlexinB2 extracellular domain fused to alkaline phosphatase (AP). The latter had been harvested from the conditioned medium of COS-7 cells transiently transfected with a cDNA construct encoding PlexinB2 sequence comprised between amino acids 1 and 1190, fused to secreted placental AP. After five washes, Sema4C-expressing cells were fixed, heated for 10 min at 65°C to inactivate endogenous phosphatases, and incubated with nitro blue tetrazolium-5-bromo-4-chloro-3-indolyl-phosphate AP substrate (Promega) to reveal bound PlexinB2.

Cell proliferation assay

Tumor cells were seeded in multiple 96-well plates at an initial density of 1.5 × 103 to 3 × 103 cells per well (depending on the cell line) and subsequently allowed to grow in complete culture medium with 10% FBS. Every 24 hours, one multiwell dish was fixed with 11% glutaraldehyde and stained with crystal violet, and the absorbance was read at 595 nm.

Transwell migration and invasion assays

Transwell migration assays were performed using Transwell chamber inserts with a porous polycarbonate membrane (8-μm pore size) (Corning Costar Incorporated, NY, USA). Briefly, the lower side of the filter was coated with fibronectin (10 μg/ml) and blocked with 1% BSA. For invasion assay, the upper side of the filter is coated with Matrigel according to the manufacturer’s protocol. About 5 × 104 cells were added in the upper chamber and allowed to migrate through the filter toward the lower chamber containing the indicated factors. In parallel, the same volume of cell suspension was seeded in cell culture multiwell dishes to check for equal cell loading. After 12 or 24 hours, nonmigrated cells on the upper side of the filter were removed by a cotton swab, followed by fixing with 11% glutaraldehyde and staining with crystal violet. Microscopic images were then quantified either by cell counting or by converting to a binary image and quantifying the integrated pixel values using ImageJ [National Institutes of Health (NIH)]. Experiments were repeated at least three times in replicates, showing consistent results.

Wound healing assay

Wound healing assay was performed in confluent monolayer of cells grown in six-well plates. A pipette tip was used to make three scratches in cell monolayers; cells were washed twice, and images were taken at starting time point, followed by incubation in appropriate media. Later, images were taken after 24 to 48 hours; images were aligned and analyzed to score for wound closure (based on measurement of residual wound area).

Tumor endothelial cell adhesion assay

For tumor endothelial adhesion assay, HUVECs were plated to form a confluent monolayer. Tumor cells that where gently detached with ACCUTASE are labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE) dye and were seeded on monolayer, and attached cells were counted and imaged after washing with PBS after 30 min.

In vivo metastasis and tumor cell extravasation assays

In vivo studies were conducted in 6- to 8-week-old nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice (Charles River Laboratories). For experimental metastasis experiments, 2 × 105 MDA-MB231 breast cancer cells or 1 × 106 PC-3 prostate cancer cells were injected intravenously into the lateral tail vein to evaluate lung colonization. Three weeks after injection, the mice were euthanized, and macrometastasis was counted by stereomicroscope inspection of the lungs upon airway injection with Indian ink and or by visually analyzing the lung section upon hematoxylin and eosin staining using a Leica DM IRB microscope, and digital images were evaluated with METAMORPH software. For in vivo extravasation assay, 1 million cancer cells were injected into the lateral mouse tail vein. Prior tumor cell labeling was performed by incubation with Vybrant DiD (catalog no. V22887, Thermo Fisher Scientific), according to the manufacturer’s specification. Mice were euthanized 48 hours after injection, and quantification of metastatic cells in the lungs was performed by fluorescence microscopy by analyzing at least four microscopic fields per lung using ImageJ software (NIH) to measure signal intensity. Mouse handling was performed according to international guidelines for animal experimentation and approved by the competent national ethical committee according to Italian law.


Results are means of at least three different independent experiments ± SD. Comparisons were made using the two-tailed Student’s t test, indicated in graphs as follows: ***P < 0.001, **P < 0.01, and *P ≤ 0.05.


Fig. S1. Immunostaining to assess Sema4C expression in tumors.

Fig. S2. Phenotypic change of cancer cells upon Sema4C overexpression.

Fig. S3. Experimental controls for data shown in Figs. 4 and 5.

Fig. S4. Correlation between SEMA4C and ID1/3 expression in human cancer samples.

Fig. S5. Additional evidence of Sema4C-dependent regulation of TGF-β/BMP receptor signaling.

Fig. S6. Correlation of patient survival with Sema4C levels in breast cancer samples.

Table S1. Sema4C-induced gene expression reprogramming in MDA-MB231 cancer cells.


Acknowledgments: We are grateful to B. Martinoglio for help with real-time PCR analysis and to S. Giordano and all Tamagnone lab members for advice and support. We also thank R. Bassett for help with the data analysis. Funding: The work was supported by the Italian Association for Cancer Research (AIRC-IG grant no. 19923 to L.T. and 5perMille grant no. 21091 to E.M.) and the Fondazione Piemontese per la Ricerca sul Cancro (grants FPRC-5perMille-MIUR-2013 and FPRC-5perMille-MinSal-2013 to L.T. and A.S. and 5perMille-MinSal-2015 to E.M.). Author contributions: S.G. and L.T. contributed to the study conception and design, analysis, and interpretation of the data. S.G., G.F., D.F., M.A., C.I., E.M., I.S., and A.S. contributed to the acquisition and analysis of the data. S.G. and L.T. were involved in writing the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.

Stay Connected to Science Signaling

Navigate This Article