Expression levels of SPAG5, YAP and mut-53 in BC subtypes
Recently, we showed that in BC SPAG5 expression is transcriptionally sustained by YAP-TAZ-TEAD interaction (16), while previous reports suggest that mut-p53 proteins frequently crosstalk with YAP or other determinants of the Hippo signalling pathway (14). We therefore investigated whether YAP, mut-p53 and SPAG5 were mutually aberrantly expressed in BCs patients.
We classified 1969 BC patients from METABRIC dataset, a large cohort of molecularly characterized tumors, into four different groups, based on the expression levels of SPAG5 and YAP signatures: SPAG5high/YAPhigh, SPAG5low/YAPlow, SPAG5high/YAPlow, SPAG5low/YAPhigh, (Heatmap Fig. 1A, B, Fig. S1A, B). For each group, we analysed p53 status, wild-type (wt-p53) or mutated, and the prevalence of BCs histotypes (Normal-like, Basal-like, HER-2, luminal-A, luminal-B) as a measurement of tumor malignancies (Pie-charts Fig. 1A, B, Fig. S1A, B). Figure 1A depicts patients characterized by SPAG5high and YAPhigh: in this group higher frequency of mut-p53 (62.4%) was associated with a higher risk to develop more aggressive tumors histotype (Basal-like=58%). In contrast, patients carrying wt-p53 protein with SPAG5high and YAPhigh, mainly developed luminal features, while the frequency of Basal-like histotype decreased to 16% (Table in Fig. 1A). Opposite scenario was observed by analysing patients with SPAG5low and YAPlow. In this group, TP53 mutations were less prevalent compared to the other categories (14.7%) and the incidence of basal-like subtype was decreased to 6% (Fig. 1B). We also considered patients with SPAG5low/YAPhigh or SPAG5high/YAPlow. Beyond similar frequencies of TP53 mutations 28.5% in SPAG5high/YAPlowand 33% in SPAG5low/YAPhighgroups,41% of patients with YAPhigh and mut-p53 were associated with basal-like subtype (Fig. 1SA), while SPAG5 overexpression act as a self-sufficient supporter of tumor growth. Of interest, normal-like phenotype shows the lower prevalence when SPAG5 is highly expressed (Fig. 1SB). Collectively, these observations suggest that in BC patients, SPAG5, YAP and mut-p53 were concomitantly expressed and clinically associated with tumor aggressiveness.
YAP and mut-p53 regulate SPAG5 expression independently
It was previously shown that mut-p53 and YAP recruit transcription factors to the promoters of shared target genes, to support cell cycle progression, and tumor growth (14, 18). To establish the molecular networks between mut-p53, YAP and SPAG5, we performed whole RNA-sequencing of MDA-MB-468 TNBC cells depleted of mut-p53 (R273H) or YAP genes. We then looked for all transcripts, positively or negatively correlated with mut-p53 and YAP. As seen in Figure 1C, SPAG5 mRNA was significantly reduced in TP53- or YAP-depleted cells: among the 1180 or 1729 downregulated genes, respectively (Fig. 1D, E). These results suggest that both YAP and mut-p53 regulate SPAG5 expression.
We used GSEA and KEGG pathway analysis to identify selective and common biological processes regulated by mut-p53 and YAP (Fig. S1C, D). Most of the downregulated pathways in YAP and p53 depleted cells, are associated with cell cycle progression, mitotic checkpoint, DNA replication, E2F and DNA synthesis pathways, consistent with previous reports (18). Among the negatively enriched pathways in p53-depleted cells, we focus on MYC targets, because of its central oncogenic signal in BC and its compelling role in anticancer target therapy (Fig. S1C). Furthermore, concomitant high expression of mut-p53 and MYC is associated with poor prognosis in TNBC patients (19-22).
SPAG5, mut-p53, YAP and TP53-MYC-target gene signature are prognostically relevant for TNBC
To investigate whether MYC transcriptional activity is associated with SPAG5, mut-p53 or YAP expression, we evaluated the expression of a TP53-dependent MYC-target gene signature (TP53-MYC target signature) in BC patients from METABRIC cohort (22). Normalized high expression levels of SPAG5 and YAP were significantly associated with high levels of TP53-MYC target signature in all BC (p=5.1277 e-77) and in TNBC patients (p=0.001) (S1E, Fig. 1F) (22). Notably, 75% of all BC and 65% of TNBC patients with SPAG5high and YAPhigh were also highly enriched in TP53-MYC target signature (Fig. 1G). Furthermore, TP53 mutations prevailed within patients with high levels of SPAG5, YAP and TP53-MYC signature (Fig. 1H, I). COX multivariate analysis adjusted for tumoral stage, nodal stage, histotype, age and menopausal state, further predicts that SPAG5, mut-p53 and YAP are prognostic determinants for BC, independently or in association with p53-MYC signature (Table 1). Collectively, these findings reveal that expression of SPAG5, YAP, mut-p53 and TP53-MYC signature, are mutually connected and significantly prognostic for BC patients.
Mut-p53 depletion reduces TP53-MYC target signature and SPAG5 expression in BC cell lines
We showed that mut-p53 depletion in MDA-MB-468 cells reduced the expression of TP53-MYC target signature (Fig. 2A, B). Silencing of mut-p53 significantly impaired expression of G3BP1, XRCC6 and CCT3, TP53-dependent MYC target genes, as evaluated by qRT-PCR (Fig. S2A) (22); concomitant decrease of SPAG5 expression (Fig. 2A, B), strongly suggests a common transcriptional program. To further assess the impact of mut-p53 on SPAG5 expression, we knocked it down in four BC cell lines harboring different TP53 missense mutations: MDA-MB-468 (R273H), MDA-MB-231(R280K), SUM-159 (R158L) TNBC cell lines, and SKBR3 (R175) HER-2 positive BC cells. Intriguingly, we found that SPAG5 transcript (Fig. 2C, E, G, I) and protein (Fig. 2D, F, H, J) levels were significantly reduced in all p53-depleted cells, highlighting the impact of mut-p53 on SPAG5 expression.
Mut-p53 and YAP recruit transcription factors to SPAG5 promoter
We showed that YAP depletion reduced SPAG5 expression (Fig. 2A, S2B) (16). Chip-Seq data on MDA-MB-231 cells, revealed that mut-p53 and YAP colocalize with c-MYC, E2F1, and TEAD4 binding sites on SPAG5 promoter (www.cistrome.org/db) (Fig. S2C). Furthermore, LASAGNA algorithm (Length Aware Alignment Guided by Nucleotide Association) predicted specific binding sites of cMYC, NF-Y, E2F and TEAD on SPAG5 promoter (Fig. 2K). We validated the in-silico analysis by ChIP-PCR on MDA-MB-468 TNBC cell line. Rousingly, we found a remarkable increase in the recruitment of mut-p53 and YAP to the SPAG5 promoter within c-MYC, E2F1, NF-Y, and TEAD4 binding sites, highlighting novel transcriptional regulators of SPAG5 expression (Fig. 2L-O).
c-MYC depletion reduces SPAG5 expression and clonogenic ability of TNBC cell lines
To further explore the relation between SPAG5 and MYC, we assessed their association in METABRIC dataset. As shown, high levels of SPAG5 and YAP were significantly associated with c-MYC expression in BCs and in TNBC patients (p=8.0932 e-23 and p=0.122221 respectively) (Fig. 3A, B).
Therefore, we knocked down c-MYC in MDA-MB-468, MDA-MB-231 and SUM-159 TNBC cell lines. SPAG5 mRNA (Fig. 3C, F, S3A) and protein (Fig. 3D, G, S3B) levels were markedly reduced in MYC-depleted cells. Expression of NME1 G3BP1, XRCC6, CCT3 and TCP1 (22), were decreased (Fig. S3D), and the number of the colonies formed by MDA-MB-468, MDA-MB-231 and SUM-159 were also significantly reduced (Fig. 3E, H, S3C). These results, together with the downregulation of SPAG5 and cyclin-B1, and upregulation of CDK inhibitor p21 proteins, suggest that depletion of c-MYC reduced SPAG5 expression and consequently BC cell proliferation (Fig. 3D, G, S3B). To corroborate these findings, we used the small molecule MYC-inhibitor, MYCi975 (23). Noticeably, MYCi975 induced a dose dependent downregulation of SPAG5 mRNA and protein levels, (Fig. 3I, J, L, M, S3E, F), reduced expression of MYC-p(Ser62), G3BP1, XRCC6, NME1 and TCP1 (Fig. 3J, M, S3H, I), and the number of colonies formed by in MDA-MB-468, MDA-MB-231 and SUM-159 cells (Fig. 3 K, N, S3G).
SPAG5 is a direct transcriptional target of MYC
Two different c-MYC binding sites were predicted by Lasagna 2.0 algorithm on SPAG5 promoter (Fig. 3O). ChIP-PCR assay demonstrated a significant enrichment of active MYC-p(Ser62) on the selected regions of SPAG5 promoter (Fig. 3P, R), concomitantly with increased Histone H4 acetylation, marker of transcriptional active chromatin (Fig. 3Q, S). Furthermore, p53 or YAP depletion (Fig. S3J), markedly reduced MYC-p(Ser62) recruitment on SPAG5 promoter and histone acetylation on the selected regulatory regions (Fig. 3P, R, Q, S). SF3B3, MYC-target gene, was used as a positive control (Fig. S3K) (22). Collectively these findings show that SPAG5 is a direct transcriptional target of MYC, p53 and YAP axis.
JQ-1 and Dasatinib inhibit SPAG5 expression troughs MYC
Numerous studies have shown that BET inhibitor JQ-1 downregulates MYC transcription and target genes expression (24). Therefore, we examined the ability of JQ-1 to modulate SPAG5 expression and colony formation of TNBC cell lines. As shown, JQ1 markedly reduced SPAG5 transcript and protein levels (Fig. 4A, B, D, E, S4A, B), NME1 G3BP1, XRCC6, CCT3, TCP1 genes expression (Fig. 4A, D, S4A) and the number of the colonies formed by MDA-MB-231, MDA-MB-468 and SUM-159 TNBC cell lines (Fig. 4G, H, S4D). It has been proved that high c-MYC expressing cells are more sensitive to Dasatinib (25-27). We previously showed that by targeting YAP and TAZ, Dasatinib impairs SPAG5 expression in TNBC cells lines (16). Indeed, 48 hr of Dasatinib treatment, significantly reduced SPAG5 expression, as well as NME1 G3BP1, XRCC6, CCT3, TCP1 transcripts, in the three TNBC cell lines (Fig. 4A, C, D, F, S4A, S4C).
ChIP assay demonstrated that JQ-1 and Dasatinib reduced the recruitment of MYC-p(Ser62) and Histone H4 acetylation to the MYC-binding sites on SPAG5 promoter (Fig. 4I, J).
JQ-1 treatment reduced the level mut-p53 and MYC-p(Ser62) proteins in MDA-MB 231 cell line (Fig. 4K, L), but, consistent with previous reports (28), little affect YAP protein and its nuclear localization (Fig. 4K, L); instead, significantly impaired YAP transcriptional activity, as shown by the reduced expression of SPAG5, CTGF, ANKRD1 and CYR61 (Fig. 4K, L, S4E-G) (24, 28, 29). Dasatinib, weakened MYC-p(Ser62) and mut-p53 proteins expression, preserved YAP nuclear localization and caused to downregulation of CYR61, CTGF and ANKRD1 expression (Fig. 4K, L, S4E-G). Increased of p21 CDK inhibitor and poly (ADP-ribose) polymerase-1 (PARP-1) proteins demonstrate the anti-proliferative and pro-apoptotic activity of JQ-1 and Dasatinib in TNBC cell line, respectively. These results together with the reduced expression of SPAG5, MYC-p(Ser62) and mut-p53 strongly suggest that Dasatinib and JQ-1 treatments directly target the oncogenic network of SPAG5-MYC-mut-p53 in TNBC cells (Fig. 4L).
Anthracyclines and taxanes have no effect on SPAG5 expression in TNBC
Currently, standard care for TNBC include non-specific chemotherapy such as taxanes or anthracyclines. To investigate the clinical efficacy of SPAG5 to predict TNBC patient’s response to cytotoxic therapy we used the Rocplot analysis (www.rocplot.org) (30). Non-significant modulation of SPAG5 expression was obtained between responsive and non-responsive cohorts in combination therapy of 5-Fluorouracil (5-FU) plus Cytoxan, Doxorubicin (p=0.29; AUC=0.55) (Fig. 5A) or Taxane (p=0.3; AUC=0.518) (Fig. 5C). Moreover, ROC curve analysis reveals non-predictive value of SPAG5 to establish patient’s response to combination therapy of 5-Fluorouracil plus Cytoxan and Adriamycin (Fig. 5B) or Taxane (Fig. 5D). Accordingly, sub-apoptotic doses of cisplatin (CDDP), 5-Fluorouracile, Doxorubicin, Paclitaxel and Vinorelbine didn’t affect SPAG5 protein expression in MDA-MB-231 and MDA-MB-468 TNBC cancer cell lines (Fig. S5A, B).
BET-inhibitor sensitizes TNBC cell to chemotherapy
SPAG5 knockdown sensitized TNBC and cervical cancer cell lines to Olaparib and Taxol (9, 31). Therefore, we tested whether JQ-1 treatment, which reduces SPAG5 expression, synergies with cisplatin to impair proliferation of TNBC cells.
Indeed, we found that pretreatment with JQ-1 sensitized MDA-MB231 (Table 2A) (Fig. S5C) and MDA-MB468 (Table 2B) (Fig. S5D) cells to cisplatin treatment (Fig. 5E, F, S5C, D). As shown in Fig. 5G, treatment of MDA-MB-231 cells with cisplatin weakly reduced MYC-p(Ser62) protein level. Intriguingly, compared with the single treatments, combination of JQ-1 and cisplatin led to a significant decrease of MYC-p(Ser62) and SPAG5 proteins in MDA-MB-231 cells (Fig. 5G). Likewise, combined treatment of JQ-1 with other cytotoxic based chemotherapy, 5-Fluorouracile, or Paclitaxel, significantly reduced MYC-p(Ser62) and SPAG5 expression in MDA-MB231 cells (Fig. S5H, I), and significantly increased the expression of CDK-inhibitor p21 (Fig. 5G, S5H, I).
Cisplatin, 5-Fluorouracile, and Paclitaxel slightly inhibited colony formation of TNBC cells, when administer as a single agent (Fig. 5H-J, S5E-G, S5J). However, combination of cisplatin, paclitaxel, or 5-Fluorouracile with JQ-1 farther reduce clonogenic ability of MDA MB-231, compared with single agent treatments (Fig. 5H, S5J). These results suggest that the synergistic effect of these compounds is partially mediated by impairment of MYC activity, and consequently of SPAG5 expression.
Next, we examined whether transient ectopic overexpression of SPAG5 could abolish the synergistic effect of JQ-1, cisplatin, paclitaxel, and 5-Fluorouracile in MDA-MB-231 cell line. As shown in Figure S5K, MDA-MB-231 cells overexpressing SPAG5, gained resistance to JQ-1 treatment, and lost the synergy of JQ-1 with cisplatin, paclitaxel, or 5-Fluorouracile as shown by colony formation assay (Fig. S5K). These results suggest that SPAG5 is a leading player in the synergism between JQ-1 and cytotoxic based chemotherapy. To further corroborate these findings, we used more advanced BET inhibitors, which have been recently used in clinical trials, including CPI-0610 (NCT02158858, NCT02157636), INCB054329 (NCT02431260) and OTX015 (NCT02259114). As shown (Fig. S5L, M), treatment of MDA-MB-231 and MDA-MB-468 TNBC cell lines with sub apoptotic doses of CPI-0610, INCB054329, and OTX015, reduced MYC-p(Ser62) and SPAG5 proteins levels in a dose dependent manner. Concurrently, CDK-inhibitor p21 was increased, highlighting the anti-proliferative effects of these compounds. Moreover, combination treatment of CPI-0610 and cisplatin, paclitaxel, or 5-Fluorouracile markedly reduced MYC-p(Ser62) and SPAG5 proteins level, and significantly inhibited clonogenic ability of MDA-MB-231 cell line, compared to single agent treatments (Fig. 5K-P). All together, these results demonstrated that BET-inhibitors reduced SPAG5 expression via MYC, and sensitized TNBC cells to conventional therapy.
Dasatinib compound synergies with cytotoxic chemotherapy in inhibiting SPAG5 expression and TNBC cell clonogenicity
A recent phase II clinical trial documented the safety of combined treatment of Dasatinib, plus Trastuzumab and Paclitaxel in HER-2-positive metastatic BC patients with acquired resistance to Trastuzumab (NCT01306942, EudraCT 2010-023304-27) (32).
To determine whether Dasatinib can synergize with cisplatin to inhibit SPAG5 expression and TNBC cell proliferation, we calculated their combination index in MDA-MB-231 and MDA-MB-468 cell lines (Fig. 5Q, R). As shown, pretreatment with Dasatinib sensitized MDA-MB-231 (Table 2C) and MDA-MB-468 (Table 2D) to cisplatin treatment (Fig. 5Q, R). As shown in Figure 5S, cisplatin or Dasatinib treatment alone, had no or little effect, respectively, on SPAG5 expression but their combination markedly reduced SPAG5 level in MDA-MB-231 cells. Furthermore, a significant reduction of MDA-MB-231 colonies were obtained in response to the combined treatments (Fig. 5T). Collectively, these findings suggest that in preclinical models, undirected pharmacological targeting of MYC-SPAG5 axis perturb essential oncogenic features of TNBC.
Targeting of SPAG5 expression reduces viability of patients derived tumor organoid (PDO)
To translate our findings on three-dimensional (3D) system that simulate clinical behavior of tumoral cells, we used TNBC patients derived tumor organoids (TNBC-PDTOs) (33-35). RNA-seq analysis reveals similar expression levels of SPAG5, MYC, TP53, YAP1, G3BP1, XRCC6, CCT3, NME1 genes between parental tumoral tissues and TNBC-PDTO (Fig. 6A, Table 3),
Indeed, we analyzed organoids viability following JQ-1, CPI-0610 or Dasatinib treatments, alone or in combination with cisplatin. As shown in Figure 6B, when administrated as a single agent treatment, cisplatin slightly reduced viability of TNBC-PDTO. JQ-1 treatment inhibited TNBC-PDTO proliferation by almost 30%, while combination of JQ-1 with cisplatin inhibited TNBC-PDTO viability by almost 60%, highlighting the potency of this combination. We also tested the efficacy of CPI-0610. CPI-0610 alone, reduced the viability of TNBC-PDTO, but more profoundly together with cisplatin (Fig. 6B). Dasatinib had strong effect on PDTO viability as a single drug, and even stronger effect in combination with cisplatin (Fig. 6B).
Altogether, our results, marked SPAG5 expression as a bona fide biomarker to identify patient’s responder and/or non responders to MYC inhibitors and their combination with chemotherapy.