Reduced to lost expression of S1PR2 in colorectal cancer
To investigate the involvement of S1PR1, S1PR2, and S1PR3 in CRC, we firstly analyzed their mRNA levels in a cohort of CRC patients (within stages II and III). In the normal mucosa, expression levels of S1PR1 were higher compared to S1PR2 and S1PR3 levels, whereas S1PR2 and S1PR3 expression was comparable to each other/similar (Supplementary Fig. 1A). Noticeably, in CRC samples, S1PR2 was significantly decreased as compared to normal mucosa (p = 0.043), whereas no difference was observed for S1PR1 and S1PR3 levels (Supplementary Fig. 1A), indicating that only S1PR2 expression is reduced in CRC. Unexpectedly, no difference in the levels of S1P ligand was found between tumor and adjacent healthy tissue (Supplementary Fig. 1B). To address S1PR2 involvement in CRC, we first quantified protein levels, which confirmed the drastic reduction of S1PR2 expression in the tumor (TN0) compared to matched normal mucosa (p < 0.01) (Fig. 1A), and second we characterized the tissue distribution of the receptor. Although different cell types, including monocytes and endothelial cells, express S1PR2 in the healthy mucosa (Fig. 1B), we found that S1PR2 was strongly present in epithelial compartment. Specifically, it was more expressed in differentiated luminal/apical epithelial cells compared to undifferentiated cells residing at the bottom of the crypts (Fig. 1B). The comparison analysis of S1PR2 between the epithelial compartment and the whole tissue confirmed that the receptor is constitutively present on the healthy epithelium and strongly down-regulated on tumor cells (Fig. 1C). To further confirm the above data, we analyzed S1PR2 expression in 40 primary CRC samples and adjacent normal mucosa (Table 1), as well as in a tissue microarray (TMA) cohort consisted of 36 CRC tumors. S1PR2 immunostaining enforced the evidence of its marked reduction in tumor lesions, where it was confined to the epithelial compartment, displaying a heterogeneous modulation varying from a low (intensity score 3) or moderate reduction (intensity score 1–2) to a complete loss (intensity score 0) of the receptor in CRC tissues compared to adjacent normal mucosa (Fig. 1D). Based on a score of immunoreactivity that combines intensity and percentage of S1PR2 immunoreactivity (Table 2), 25 (32,89%) out of 76 patients presented no reactivity (score 0); 23 (30,26%) displayed a low (score 1–2); 19 (25%) a medium (score 3–4) and only 9 (11,85%) a high (score 5–6) reaction comparable to healthy tissue (Fig. 1E), with no correlation with CRC stage (p = 0.4338) (Fig. 1F). Of relevance, we observed a strong significant reduced immunoreactivity for S1PR2 in CRC samples carrying KRAS mutation (p < 0.0001) (Fig. 1G).
Genetic ablation of S1PR2 increases the susceptibility to develop neoplastic lesions in an in vivo model of colitis-associated colorectal cancer
To explore the functional involvement of S1PR2 in CRC development, we took advantage of a mouse model of colitis-associated cancer induced in S1PR2 knockout (S1PR2−/− or KO) mice. S1PR2−/− mice did not show any significant worsening neither in the inflammatory clinical parameters (such as body weight and disease activity index, DAI) nor in the inflammation score compared to S1PR2+/+ (Fig. 2A-C). These data were consistent with the results obtained in the acute colitis model, displaying no differences between S1PR2−/− and S1PR2+/+ littermates, thus confirming that the loss of S1PR2 does not affect mouse susceptibility to DSS-induced colitis (Supplementary Fig. 2A-C). On the other hand, S1PR2−/− mice exhibited a higher tumor incidence than S1PR2+/+ mice (100% vs. 63%; p = 0.034) Fig. 2D. Furthermore, both endoscopic and microscopic examination revealed that S1PR2-deficient developed a significantly higher number of tumors compared to littermate’s WT animals (p < 0.001 and p = 0.028, respectively; Fig. 2E up panel and Fig. 2F). Besides, the histological analysis highlighted a marked increased number of high-grade adenomas (HGA) (1.3 ± 2.109 S1PR2−/− vs. 0.3 ± 0.483 S1PR2+/+ mice, p < 0.05) with a diameter between 0.3–0.4 mm and an increased number of carcinomas with diameter > 0.4 mm in S1PR2−/− compared to S1PR2+/+ mice (2.5 ± 3.6 S1PR2−/− vs. 1 ± 1.2 S1PR2+/+mice, p < 0.05) (Fig. 2E middle panels and Fig. 2G). The difference in the number of low-grade adenomas (LGA) between the two groups did not reach significance (0.3 ± 0.8 S1PR2−/− vs. 0.3 ± 0.6 S1PR2+/+ mice, p = 0.1) (Supplementary Fig. 2D). Altogether, these data point to an enhanced cancer susceptibility, coupled to a faster/higher tumor growth rate in the S1PR2−/− background under inflammatory conditions. Interestingly, the immunostaining for nuclear ß-catenin was strongly positive in KO mice compared to littermate WT mice (Fig. 2E, low panels). We next assessed whether the higher incidence of tumors in S1PR2−/− was also associated with a deregulated cell growth by intraperitoneally injecting BrdU 24 hours before sacrifice. No difference was observed in healthy conditions between naive and tumor-bearing mice. In contrast, S1PR2−/− tumor-bearing mice displayed an increased cell proliferation in tumor lesions compared to tumor bearing S1PR2+/+ ones (Fig. 2H).
Loss of S1PR2 following APC mutation in colon carcinogenesis
To further gain deeper insight into the role of S1PR2 in intestinal tumorigenesis, we exploited also an ApcMin/+ mouse model that spontaneously develop multiple polyps in the small intestine23. S1PR2 deficiency in Apcmin/+ at 21 weeks old led to a significant increase of total tumor load over the entire gastrointestinal tract compared with their littermates S1PR2+/+Apcmin/+ mice (p < 0.05) (Fig. 3A), which was well evident in the distal colon (p = 0.033) (Fig. 3B-C). In keeping with the observation in the DSS mouse model, we also observed a significant increase in the size of tumors in S1PR2−/−Apcmin/+ compared to S1PR2+/+Apcmin/+ mice (p = 0.005) (Fig. 3C). In parallel, histologic examination highlighted a significantly higher number of carcinomas in S1PR2−/−Apcmin/+ (2.40 ± 0.51) compared to S1PR2+/+Apcmin/+ mice p = 0.041 (Fig. 3D). No difference was observed in LGA and HGA lesions number between the two groups (Fig. 3D). Accordingly, the Ki67 immunostaining showed a significantly increased cell proliferation in the colon of S1PR2−/−Apcmin/+ compared to S1PR2+/+Apcmin/+ (p = 0.02) mice (Fig. 3E). We found no difference in the small intestine tumor burden between the two groups (Fig. 3E). To address the role of S1PR2 during the early phases of tumor development, we inhibited pharmacologically S1PR2 in ApcMin/+ mice at ten weeks of age, before the appearance of both intestinal and colonic tumors, by the specific S1PR2 inhibitor (JTE013). JTE103 inhibitor accelerated tumor formation in Apcmin/+ mice in comparison to the vehicle (Fig. 3F) along the gastrointestinal tract. Moreover, while vehicle-treated Apcmin/+ mice developed only low-grade adenomas, the Apcmin/+ mice treated with the JTE013 inhibitor broadened high-grade adenomas and carcinomas (Fig. 3G), corroborating the loss of S1PR2 as an accelerator of tumor development and de-differentiation. S1PR2 immune-histochemical analysis in intestinal tissue of Apcmin/+ mice revealed a strong decrease of the receptor in the epithelial compartment of both high-grade adenomas and carcinomas compared to the normal epithelium (Fig. 3H).
Loss of S1PR2 is an early event in the pathogenesis of colorectal cancer
To validate this feature in humans, we analyzed S1PR2 expression in 5 human tubulovillous adenomas with moderate focal dysplasia, which is considered as an early precancerous lesion. All adenoma samples showed a significant reduction of S1PR2 compared to healthy mucosa (Fig. 3H-I), thus corroborating the hypothesis that the loss of S1PR2 in the epithelial compartment plays a key role in colorectal tumorigenesis and that it likely occurs in the early phase of cancer development.
The overexpression of S1PR2 reduces the tumorigenicity of human CRC-derived epithelial cells in vivo
To gain insight into the mechanisms by which the loss of S1PR2 promotes tumorigenesis, we explored the role of S1PR2 as a brake of tumor proliferation and a potential tumor suppressor gene in vivo. To this end, we firstly examined the endogenous expression of S1PR2 in four metastatic colon cancer cells SW620, RKO, HCT116, and HT29. In line with our previous data, S1PR2 expression was deficient in all cancer cells (Supplementary Fig. 3A). Then, to test whether S1PR2 acts as a brake of tumor proliferation, we used lentivirus-mediated overexpression of S1PR2 in RKO (RKO S1PR2 OE), which exhibited, at least in our hands, highest infection efficiency among all cell lines negative for S1PR2. Overexpression efficiency was verified by RT-PCR (Fig. 4A). The effect of S1PR2 overexpression on the in vivo tumor cell growth was measured over 23 days after subcutaneous injection of RKO cells in female CD-1 nude mice. S1PR2 overexpression attenuated tumor growth with statistical significance (p < 0.05) compared to the scramble (Fig. 4B). Cell cycle analysis on the recovered tumors by flow cytometry demonstrated that most RKO cells overexpressing S1PR2 arrested at the G0/G1 phase, while a substantially higher fraction of cells from scramble tumors were in S and G2/M phases [compared to S1PR2 overexpressing ones] (p < 0.05) (Fig. 4C). Based on in vivo experiments in which the loss of S1PR2 promoted the pathological accumulation of nuclear ß-catenin that, in turn, can control cell cycle, we examined whether S1PR2 exerted a direct role in this event. Interestingly, in contrast to scramble cells, RKO-S1PR2-OE displayed significantly higher levels of Axis inhibition protein 2 (Axin2) gene (Fig. 4D) that enhances the formation of the beta-catenin destruction complex and therefore prevents the nuclear translocation of ß-catenin.
S1PR2 has been shown to inhibit cell migration in cancer cell lines19, 20. To address whether S1PR2-overexpression affected migratory and invasive properties of metastatic RKO cells, we performed transwell migration and invasion assays and analyzed the expression of some genes that support invasive capacities. S1PR2- overexpression did not affect the in vitro migratory capacity of RKO cells (Supplementary Fig. 3B), neither in the mRNA expression levels of matrix metalloproteinases such as MMP1 and 2 (Supplementary Fig. 3C), both genes involved in the distant metastasis development in CRC24. In line with these features, no difference was found between RKO scramble and RKO S1PR2 OE cells in distant organs such as liver, mesenteric lymph nodes, adipose tissue, and colon (Supplementary Fig. 3D) of mouse xenograft model. Overall, these findings supported the key role of S1PR2 in arresting tumor growth and excluded its potential function in controlling the migratory capacity of epithelial tumor cells. Recently S1PR2 has been involved in the growth of hepatocellular carcinoma cells through the activation of PI3K/AKT signaling25. To investigate whether, in CRC cells, the modulation of S1PR2 can also drive the activation of the PI3K/AKT pathway, which is highly expressed in the RKO cell line26, we quantified the protein levels of the phosphorylated AKT in RKO scramble and RKO S1PR2 OE cells. The overexpression of S1PR2 significantly reduced the phosphorylation of AKT (Fig. 4E). It augmented the levels of Phosphatase and tensin homolog deleted on chromosome ten (PTEN), a negative regulator of PI3K/AKT pathway (Fig. 4F). These results point out S1PR2 as a regulator of PTEN. To validate this hypothesis in an in vivo tumor model, we quantified pAKT in S1PR2+/+ and S1PR1−/− tumor-bearing mice. As expected, the loss of S1PR2 augmented AKT levels and its phosphorylation in the mucosa of KO mice (Fig. 4G).
S1PR2 inhibition impacts intestinal stem cell expansion
The observation that the S1PR2 receptor is mainly expressed at the top of the intestinal crypts, while its expression is lower at the bottom where reside intestinal stem cells, may support a differential expression of S1PR2 between differentiated intestinal epithelial and stem cells.
To verify whether S1PR2 is functionally involved in the proliferation and differentiation of intestinal stem cells, organoids isolated from naive WT mice were maintained in vitro for six days in the absence or presence of JTE013. Typically, intestinal organoids cultures tend to exhibit extensive budding of crypt-like domains (Fig. 5A up panels). In the presence of JTE013 organoids appeared with more cyst-like morphology characterized by a small number of uncomplete branches (Fig. 5A low panels). The round-shape of organoids raised the suspicion that JTE013 could prevent the differentiation of epithelial cells. Indeed, the levels of Olfm4 and Lgr5 stemness markers were significantly higher after JTE013 compared to untreated organoids (Fig. 5B) proving that the loss of S1PR2 maintains the organoids in an undifferentiated status. To gain deep insight into this aspect, we analyzed the expression of S1PR2 in stem (EpCAM + Lgr5-GFP+) and differentiated (EpCAM + Lgr5- GFP-) intestinal epithelial cells isolated from Lgr5-EGFP-IRES-creERT2 mice (Fig. 5C). Accordingly, EpCAM + Lgr5-GFP + stem cells expressed lower levels of S1PR2 compared to differentiated cells (EpCAM + Lgr5-GFP-) (Fig. 5D). In addition, Lgr5 + GFP immunostaining confirmed an increased number of Lgr5 positive cells in S1PR2−/− mice compared to their littermates Lgr5-EGFP-S1PR2+/+ (Fig. 5E).
Loss of S1PR2 impairs mucosal regeneration in vivo
To assess whether the deregulation of intestinal stem cell proliferation and differentiation participates to the impairment of mucosal regeneration in S1PR2 deficient mice in vivo, which may contribute to intestinal tumorigenesis, we analyzed the regeneration of mucosal structure in S1PR2−/− and S1PR2+/+ mice after irradiation. Based on previous studies showing the time of mucosal regeneration following the irradiation27, the destruction of the normal crypt-villus axis starts after two days in association with the expansion of undifferentiated Lgr5 + cells for replacing proliferating cells that, within 6–7 days, renew the intestinal mucosa structure by migrating from the bottom to the top along the crypt-villus axis (Fig. 5F). Both S1PR2−/− and S1PR2+/+ mice groups displayed substantial bodyweight loss within seven days after irradiation, reflecting the damage to the intestinal mucosa (Supplementary Fig. 4). After this time, both groups of mice started to gain their body weight in support of the recovery of mucosal damage. Although no difference was observed in the bodyweight recovery between S1PR2+/+ and S1PR2−/− mice, the intestinal mucosa structure presented differently. Despite the presence of signs of a regenerative process still in progress, the mucosa of S1PR2+/+ mice recovered intestinal damage. It restored intestinal integrity by normalizing the villus height and crypt depth. Differently, the mucosa of S1PR2−/− mice, while showing a recovery of villi height, displayed elongated and enlarged crypts characterized by an increased number of undifferentiated cells as evidenced by the strong immune-positivity for Oflm4 (Fig. 5G). To further corroborate the key role of S1PR2 in arresting the expansion of intestinal stem cells, we analyzed in RKO S1PR2OE and scramble cells the expression of transcription factor Sex-determining region Y (SRY)-box 9 (SOX9), which is linked to stem cell maintenance 50 and implicated in CRC. The overexpression of S1PR2 reduced significantly SOX9 levels compared to RKO scramble cells (Fig. 5H) supporting S1PR2 as a brake for the expansion of intestinal stem cells.