SPINT1 genetic alterations are associated with poor prognosis of SKCM patients and altered tumor immune microenvironment
To study the impact of SPINT1 in promoting SKCM progression and aggressiveness, an in silico analysis of human SKCM samples of the TCGA cohort was performed. This analysis revealed that genetic alterations occurred in 10% of SKCM patients; a relevant percentage comparing with major SKCM driven oncogenes and tumor suppressors (Fig. 1a). Among these genetic alterations, an increased mRNA level was the most prevalent alteration (7%), while 1.9% missense mutations of unknown significance and 1.9% deep deletions were also observed. Notably, these genetic alterations of SPINT1 significantly correlated with poor SKCM patient prognosis (Fig. 1b) and SPINT1 expression was significantly inhibited in human SKCM comparing with nevus and normal skin (Fig. 1c). In addition, the analysis of patient survival with increased SPINT1 transcript levels also revealed their poor prognosis but, unfortunately, the analysis of deep deletions and missense mutations separately gave no statistical significance because the number of patients for each condition was very low (data not shown). We next performed a GO enrichment analysis of biological process (Fig. 1d), analyzing the differentially expressed genes in SKCM samples of the TCGA cohort with missense mutations or copy-number alterations of SPINT1. The results showed that regulation of immune system, inflammatory response, cell cycle, cell adhesion, and extracellular matrix organization represent key pathways significantly affected in human SKCM with these SPINT1 genetic alterations. Collectively, these results suggest that both high and low levels of SPINT1 result in an unbalanced crosstalk between tumor cells and their microenvironment promoting higher aggressiveness.
The tumor microenvironment contains diverse leukocyte populations, including neutrophils, eosinophils, dendritic cells, macrophages, mast cells and lymphocytes [26]. It is known that tumor-associated macrophages (TAM) are able to interact with tumor cells and can promote cancer progression [27,28,29]. As shown in Fig. 1e, the number of TAM in human SKCM samples correlated with the mRNA levels of SPINT1 in metastatic SKCM. However, the number of tumor-associated neutrophils (TAN) was independent of SPINT1 levels in both primary and metastatic SKCM. These data further confirmed the role of SPINT1 in the regulation of the crosstalk between tumor and inflammatory cells in human SKCM.
The expression of SPINT1 positively correlates with both inflammation and macrophage markers in human SKCM biopsies
In order to further understand the role of SPINT1 in SKCM, the RNA Seq database of the large TCGA cohort of SKCM was analyzed in terms of the expression of SOX10, TYR and DCT genes, that have been shown to be important in melanocyte development [30, 31]. In addition, SOX10 is a recognized biomarker for the diagnosis of SKCM [30]. It was found that SPINT1 expression positively correlated with those of SOX10 and TYR, while a negative correlation was found between the expression of SPINT1 and DCT (Fig. 2a). The expression of the epithelial to mesenchymal transition (EMT) markers ZEB1, ZEB2 and TWIST1, but not TWIST2, negatively correlated with that of SPINT1 in SKCM (Fig. 2b).
SKCM cells release several cytokines and chemokines that recruit and polarize macrophages [32]. Therefore, several inflammation markers were analyzed and only the expression of the genes encoding the receptor of the pro-inflammatory cytokine TNFα (TNFR1) and the receptor of the pro-inflammatory chemokine interleukin 8 (CXCR2), positively correlated with SPINT1 levels (Fig. 2c). Notably, the macrophage marker MFAP4 also positively correlated with SPINT1 expression (Fig. 2d). However, the M2 polarization marker CD163 (Fig. 2d) and several interferon-stimulated genes (ISGs) (Fig. 2e) were all unaffected by SPINT1 levels. Collectively, these results further suggest that SPINT1 regulates SKCM differentiation and aggressiveness, and macrophages infiltration.
Inflammation accelerates the onset of SKCM in zebrafish
Given the strong correlation between alterations of SPINT1 levels with the progression of SKCM and the crosstalk with the tumor immune microenvironment, we crossed the zebrafish line kita:Gal4;HRAS-G12V, which expresses the human oncogene HRAS-G12V in melanocytes and spontaneously develops SKCM [8], with the zebrafish mutant line spint1ahi2217Tg/hi2217Tg [19], which presents chronic skin inflammation (Fig. 3a). Firstly, we quantified by fluorescence microscopy the number of early oncogenically transformed goblet cells, which also expressed the kita promoter [33], in spint1a-deficient larvae and their wild type siblings (Fig. 3b). The results showed that spint1a deficiency resulted in increased number of HRAS-G12V+ cells (Fig. 3c). To determine if the enhanced Spint1a deficiency-driven oncogenic transformation was also able to promote SKCM aggressiveness, SKCM development in spint1ahi2217Tg/hi2217Tg fish were compared with wild type (spint1a+/hi2217Tg) from the end of metamorphosis stage (between 28 and 30 dpf) to 120 dpf (adult stage) (Fig. 3d-f). The resulting Kaplan-Meier curve showed a significant increased incidence of melanoma in the Spint1a-deficient fish, which developed SKCM in more than 50% of cases at 50–60 dpf compared with their wild type siblings which reached only 30% at this age (Fig. 3f). These data suggest that Spint1a deficiency increases oncogenic transformation and accelerates SKCM onset in vivo.
Spint1a deficiency is required at cell autonomous and non-autonomous levels to enhance SKCM cell dissemination in a zebrafish larval allotrasplantation model
To assess the in vivo role of Spint1a deficiency in SKCM invasiveness, SKCM tumors from spint1ahi2217Tg/hi2217Tg; kita:Gal4;HRAS-G12V and spint1a+/hi2217Tg; kita:Gal4;HRAS-G12V were disaggregated, after staining the cells were transplanted into the yolk sac of 2 dpf casper zebrafish larvae (Fig. 4a). The results showed that Spint1a deficiency in SKCM cells enhanced the dissemination of SKCM, assayed as the percentage of invaded larvae and the number of foci per larva, compared to control SKCM cells (Fig. 4b-d). We next examined whether Spint1a deficiency in the stroma, i.e. in a non-autonomous manner, also promoted SKCM aggressiveness. Spint1a wild type SKCMs were transplanted into the yolk sac of Spint1a-deficient and their wild type siblings larvae (Fig. 4e). Strikingly, it was found that Spint1a deficiency in the tumor microenvironment also promoted a significantly higher dissemination of SKCM compared to control tumor microenvironments (Fig. 4f-h).
To further confirm a role of Spint1a in both SKCM and tumor microenvironment cells, we next sorted tumor (eGFP+) and stromal (eGFP−) cells from both genotypes and then mixed in equal proportions (~ 90% of tumor and ~ 10% of stromal cells) in the 4 possible combinations (Fig. 5a), since it was found that all tumors had ~ 90% of tumor and ~ 10% of stromal cells (data not shown). Notably, both Spint1a-deficient tumor and stromal cells were able to increase SKCM cell invasion (Fig. 5b and c). Collectively, these results suggest that Spint1a deficiency enhances SKCM invasion by both cell autonomous and non-autonomous mechanisms.
Spint1a-deficient SKCM cells showed enhanced aggressiveness in adult zebrafish allotransplantation model
The results obtained in allotransplantation assay in larvae prompted us to analyze the role of Spint1a in SKCM aggressiveness and metastasis in adult casper zebrafish to directly visualize tumor cell proliferation and dissemination over time. spint1ahi2217Tg/hi2217Tg and spint1a+/hi2217Tg SKCMs were sampled, disaggregated and subcutaneously injected (300,000 cells) in the dorsal sinus of adult casper recipients previously irradiated with 30 Gy (Fig. 6a). Tumor engraftment was visible as early as 7 days post-transplantation in both genotypes. While 90% engraftment was obtained with wild type SKCM cells, Spint1a-deficient cells showed a significant enhancement of tumor engraftment rate, around 95% (Fig. 6b). In addition, adult zebrafish recipients transplanted with Spint1a-deficient SKCMs developed tumors with a significant higher growth rate than those injected with wild type SKCMs (Fig. 6c). Notably, Spint1a-deficient SKCM cells were able to invade the entire dorsal area, part of ventral cavity and the dorsal fin (Fig. 6c).
We next performed additional transplant assays following the same work-flow but injecting an increased number of cells (500,000 cells per recipient fish), that ensured a 100% of engraftment was for both genotypes (data not shown). From the first week of analysis, Spint1a-deficient SKCM tumor size was significantly larger than their control counterparts (Fig. 6d). In addition, the recipients injected with Spint1a-deficient SKCM cells developed larger tumors spanning the entire dorsal area and even exceed the notochord line and grew vertically, a clear aggressiveness signature of SKCM (Fig. 6d).
To further investigate the aggressiveness potential of Spint1a-deficient SKCMs, a serial dilution assay was performed following the work-flow previously described in Additional file 1: Figure S1. Cells from both Spint1a-deficient and wild type SKCMs were serially diluted and 3 different numbers of cells (30,000 cells, 100,000 cells and 300,000 cells) were transplanted into recipients as described above. Notably, while 30,000 and 100,000 Spint1a-deficient SKCM cells were able to engraft and the tumor grew over the time, wild type SKCMs hardly grew (Additional file 1: Figure S1A, S1B). However, injection of 300,000 Spint1a-deficient SKCM cells resulted in large tumor spanning the entire dorsal area and invading part of the ventral cavity (Additional file 1: Figure S1C), confirming previous results. Collectively, all these results confirm that Spint1a deficiency enhances SKCM aggressiveness.
Spint1a deficiency promotes SKCM dedifferentiation and inflammation
To understand the mechanisms involved in the Spint1a-mediated aggressiveness of SKCM, the expression of genes encoding important biomarkers was analyzed by RT-qPCR. The mRNA levels of sox10, tyr, dct and mitfa were lower in Spint1a-deficient SKCMs than in their wild type counterparts (Additional file 1: Figure S2). In addition, while the transcript levels of mmp9 and slug were similar in Spint1a-deficient and wild type SKCM, cdh1 levels significantly decreased in Spint1a-deficient compared to wild type SKCM (Additional file 1: Figure S2).
We next analyzed genes encoding key inflammatory molecules and immune cell markers, including the pro-inflammatory cytokine Il1b, the neutrophil markers Lyz and Mpx, the macrophage marker Mpeg1, and the ISGs B2m, Mxb and Pkz, in Spint1a-deficient and wild type SKCMs (Additional file 1: Figure S2). Curiously, it was found that while il1b, lyz and mpx mRNA levels were not affected by Spint1a deficiency, those of mpeg1 were elevated in Spint1a-deficient SKCMs. Furthermore, the ISGs b2m, mxb and pkz genes showed enhanced mRNA levels in Spint1a-deficient SKCMs. These results point out to altered immune surveillance and tumor cell dedifferentiation promoted by Spint1a-deficiency in SKCM.