CXCR2 expression during melanoma tumorigenesis controls transcriptional programs that facilitate tumor growth

CXCR2 correlates with poor prognosis in patient populations and response to checkpoint inhibitors

Using the available Gene Expression Omnibus (GEO) cohort, we evaluated CXCR1, CXCR2, and CXCL1-3, 5 and 8 (CXCR1/CXCR2 ligands) expression in nevi and melanoma. CXCR1 and CXCR2 mRNA exhibited a trend toward increased expression in melanoma compared to nevi, but these differences were not statistically significant (Fig. 1A). This may be partially explained by the analysis being performed on bulk RNA-sequencing data, rather than measuring tumor cell-specific expression. However, CXCL1, CXCL2, CXCL3, CXCL5 and CXCL8 mRNAs were significantly upregulated in melanoma samples compared to benign nevi (Fig. 1B). Furthermore, there were no significant differences in CXCR1 and CXCR2 expression among nevi and melanoma tumors when stratified by BRAF or NRAS mutation status (Figure S1A, B). However, since the number of samples available for analysis of mutation status was small, these findings should be interpreted cautiously.

Fig. 1figure 1

CXCR2 is associated with tumorigenesis and poor prognosis. a GEO dataset analysis of expression of CXCR1 and CXCR2 in nevi as compared to melanoma lesions (not significant, Welch's t-test). b GEO dataset analysis of expression of CXCL1, CXCL2, CXCL3, CXCL5 and CXCL8 in nevi and melanoma tissues (significance determined by Welch's t-test). c Overall survival plot of melanoma patients from the TCGA SKCM dataset indicates significantly improved survival (p = 0.035, log-rank test) in the lowest quartile of CXCR2 expression (blue, n = 107) compared to the highest quartile (red, n = 114). d Analysis of survival of 25 melanoma patients treated with anti-PD-1 in relation to high (red) or low (blue) expression of CXCR2 [p < 0.01, log-rank test; [27]]. e Re-analysis of the Riaz RNA-seq database shows CXCR2 expression is lower in melanoma patients who responded to anti-PD1 treatment (p < 0.05, Welch's t-test)

CXCR2 has been associated with increased tumor growth and poor prognosis across multiple cancers [6]. To define the relationship between CXCR2 expression and the clinical prognosis of melanoma patients, we examined clinical data from the Cancer Genome Atlas (TCGA), and the skin cutaneous melanoma (SKCM) dataset using Gene Expression Profiling Interactive Analysis (GEPIA). Survival analysis comparing patients with high CXCR2 expression (n = 114) to patients with lower CXCR2 expression (n = 107) indicates that CXCR2 expression correlates with decreased overall survival of melanoma patients (p = 0.035, Fig. 1C). Evaluation of survival in a patient cohort treated with anti-PD-1 therapy also suggests that patients with high CXCR2 expression (n = 24) exhibited poor prognosis in response to anti-PD-1 when compared with patients with low CXCR2 expression (n = 23, p < 0.01; Fig. 1D) [31]. Finally, analysis of another immune checkpoint inhibitor-treated cohort showed that responding patients had significantly lower CXCR2 expression than non-responders (Fig. 1E, p < 0.05) [32]. These data indicate that CXCR2 expression correlates with poor therapeutic response in melanoma patients.

CXCR2 influences tumor differentiation status and enhances tumor growth

To evaluate the role of CXCR2 in BrafV600E/Pten−/− melanoma tumorigenesis, we crossed C57BL/6 Tyr-CreER + ::BrafV600E/Ptenfl/fl::mT/mG:: mice (Braf/Pten) [23] with C57BL/6 mice carrying a Cxcr2fl/fl allele [24] to produce Tyr-CreER + ::BrafV600E/Ptenfl/fl::mT/mG::Cxcr2−/− (Braf/Pten/Cxcr2−/−) and Tyr-CreER + ::BrafV600E/Ptenfl/fl::mT/mG::Cxcr2WT (Braf/Pten/Cxcr2WT) littermates. Four-week-old mice were treated with 4-OH tamoxifen (4HT) to induce the tyrosinase promoter-driven Cre-recombinase.

We then utilized flow cytometry to determine whether CXCR2 expression is indeed lost in the tumors that form in the Braf/Pten/Cxcr2−/− mice. Flow cytometry was first performed on the skin of adult Braf/Pten/Cxcr2WT and Braf/Pten/Cxcr2−/− mice immediately after application of 4-HT (prior to tumor formation), and we confirmed that melanocytes do become GFP-positive in both genotypes and that Braf/Pten/Cxcr2−/− mice lose expression of CXCR2 as expected (Figure S2D). However, after tumor formation, the same assay indicated that ~ 30% of GFP-positive cells in Braf/Pten/Cxcr2−/− tumors expressed CXCR2. While this is decreased from ~ 70% of GFP-positive cells in Braf/Pten/Cxcr2WT tumors, is does indicate that CXCR2 positive tumor cells were present during tumor formation in both genotypes (Figure S2E). To confirm these results, we also used immunohistochemistry to co-stain for SOX10 (a melanoma marker) and CXCR2 (Figure S2F). While we do not expect all melanocytic cells in the Braf/Pten/Cxcr2WT tumors to be CXCR2 positive due to variation caused by cell cycle and differentiation status, we should not see any CXCR2 positivity in the Braf/Pten/Cxcr2−/− tumor cells. This may be a result of chimerism in the loss of CXCR2 during recombination. It is expected that different floxed alleles recombine at differing efficiencies given the diversity in designs of the floxed alleles and their varied chromatin states. When 2–3 alleles are present in the same cell and all recombine with high efficiency, chimerism is low. However, when recombination efficiency differs between the alleles with one recombining slower than the other, a higher amount of chimerism is expected. Moreover, tamoxifen induction efficiency can also vary and is less efficient than Cre alone [33].

Despite the presence of  approximately 30% CXCR2 positive melanoma cells in Braf/Pten/Cxcr2−/− tumors, we observed that tumor burden and incidence (Fig. 2A) were significantly reduced in mice with Braf/Pten/Cxcr2−/− tumors (271 ± 361mm3, n = 21) compared to mice with Braf/Pten/Cxcr2WT tumors (615 ± 609mm3, n = 24, p < 0.05) 36 days after 4HT administration. The tumor number per mouse was also reduced upon melanocytic Cxcr2 deletion (0.7 ± 0.9 vs. 2.1 ± 2.3, p < 0.05). These data indicate that CXCR2 signaling plays a role in the induction and growth of BrafV600E/Pten−/− melanoma.

Fig. 2figure 2

CXCR2 knockout decreases melanoma tumor burden. a Tyr-CreER+:: BrafCA/+::Ptenlox4−5/lox4−5::mT/mG C57BL/6 mice were crossed with floxed Cxcr2 mice to obtain mice with inducible tumors with or without CXCR2 expression. Thirty-six days after 4-HT administration, skin tumor volume and count were recorded, and mice were photographed (significance determined by Welch's t-test). Similarly, b Tyr-CreER+::NRasQ61R::Ink4a−/− mice were crossed with floxed Cxcr2 mice, and resulting pups were treated with 4-HT on days 1 and 2 prior to UV irradiation on day 3 to initiate tumor formation (n = 16/genotype). Tumors were measured, counted, and mice were photographed (significance determined by Welch's t-test). RNA was extracted from BrafV600E/Pten−/−/Cxcr2−/− and BrafV600E/Pten−/−/Cxcr2WT tumors and subjected to RNAseq analysis. c A volcano plot showing fold change and significance of differential gene expression in Cxcr2−/− tumors compared to Cxcr2WT tumors. d Gene set enrichment analysis (GSEA) of RNAseq data identifies 8 gene sets enriched in Cxcr2−/− tumors. Point size indicates the gene ratio (percent of genes from the gene set contributing to the enrichment score) and point color represents the FDR q-value

To determine whether Cxcr2 is also important in NRasQ61R/Ink4a−/− melanoma tumors, we crossed Tyr-CreER + ::NRasQ61R/Ink4a−/− mice [34] with the Cxcr2fl/fl mice [24] to produce Tyr-CreER+::NRasQ61R/Ink4a−/−::Cxcr2−/− (NRas/Ink4a/Cxcr2−/−) and Tyr-CreER+::NRasQ61R/Ink4a−/−::Cxcr2WT (NRas/Ink4a/Cxcr2WT) littermates. Newborn pups (1–2 days old) were exposed to 4HT, followed by ultraviolet (UV) irradiation on day three, and tumor growth was evaluated over five months. We observed significantly reduced tumor volume with deletion of Cxcr2 (360 ± 285mm3) when compared to NRas/Ink4a/Cxcr2WT mice (764 ± 601mm3) (Fig. 2B, p < 0.05, n = 16). However, in contrast to the BrafV600E/Pten−/− model, the number of tumors per mouse was not significantly different between NRas/Ink4a/Cxcr2−/− (1.69 ± 1.08) and NRas/Ink4a/Cxcr2WT mice (1.88 ± 1.26, p = 0.654). As the NRas GEM model requires UV irradiation in addition to the genetic alterations, and the 4HT-induction phase occurs shortly after birth in this model as opposed to 30 days post birth in the BRAF model, it is possible that additional pathways that function independent of Cxcr2 are evoked.

To elucidate the mechanism by which Cxcr2 perturbation in melanocytes could alter the initiation and growth of BrafV600E/Pten−/− (Braf/Pten) melanoma, we examined the transcriptome of tumors arising in Braf/Pten/Cxcr2WT (n = 7) and Braf/Pten/Cxcr2−/− (n = 8) mice via RNA sequencing (RNAseq) analysis (Figs. 2C, S3A, S4). Interestingly, gene set enrichment analysis revealed that loss of Cxcr2 expression in Braf/Pten tumors resulted in a significant increase in expression of genes involved in CD4 + T cell activation and lymphocyte activation, with a trend toward increased leukocyte proliferation, immune response, and stem cell differentiation (Fig. 2D). However, there is also a paradoxical change in genes involved in lymphocyte anergy. In addition, we see this complex immune modulation reflected in our most differentially expressed genes, with immune-related genes falling into both the most up-regulated and most down-regulated (Figure S3A).

We next utilized the RNAseq data from Braf/Pten mice with or without loss of CXCR2 to identify the most differentially expressed genes that are associated with favorable or unfavorable outcome in melanoma patients. We identified the top twenty growth related genes with reduced expression and the top twenty genes associated with inhibition of tumor growth and favorable outcome based on their log2 fold change and -log10p-value (Figure S4). Key growth stimulatory (Figure S4A) and tumor suppressive genes (Figure S4B) are indicated by red arrows. Genes in common in both enrichment analyses in Figures S3A and S4 include upregulation of the tumor suppressors Tmprss11e, Adamts18 and Tgm3, as well as induction of the pyroptosis regulating gene GSDMc and the epithelial-specific Ets transcription factor 1 (Elf3). Commonly down-regulated genes include activators of the lectin pathway of the complement system (Fcna), myosin light chain kinase 4 (Mlk4), and pathogen recognition receptors (Cd209). These changes are plausible contributors to difference in tumor growth observed when Cxcr2 is targeted in melanocytes during transformation.

CXCR2 contributes to an immunosuppressive melanoma tumor microenvironment

Due to the GSEA-indicated enrichment in gene sets associated with CD4 + T cell activation, lymphocyte activation, and leukocyte proliferation in Braf/Pten/Cxcr2−/− tumors (Fig. 2D), we then evaluated the immune cell infiltrate between Braf/Pten/Cxcr2WT and Braf/Pten/Cxcr2−/− tumor-bearing mice. We first utilized the murine Microenvironment Cell Population counter (mMCPcounter) [34], an immune deconvolution algorithm developed for bulk murine RNAseq data. mMCPcounter predicted an increase in CD3 + T cells, CD8 + T cells, monocytes, lymphatic vessels, and eosinophils, as well as a decrease in mast cells, NK cells, and endothelial cells (p < 0.05) (Figs. 3A, S5A), suggesting enhanced anti-tumor immunity in the Braf/Pten/Cxcr2−/− TME. To analyze the immune environment in vivo, we defined the profile of CD45 + cells from Braf/Pten/Cxcr2WT and Braf/Pten/Cxcr2−/− tumor-bearing mice using FACS analysis. In agreement with the mMCPcounter predicted leukocytic infiltrates, we observed that deletion of Cxcr2 in melanocytes undergoing transformation skewed the TME toward anti-tumor immunity. FACS analysis of the CD45 + cells in Braf/Pten/Cxcr2−/− tumors revealed a decrease in the immunosuppressive Ly6G + CD11b + (p < 0.01) and CD14 + G-MDSC (p < 0.05) cells, with no change in total CD11b + cells (Fig. 3B), in addition to a trend toward decreased CD25hiCD45 + CD3 + regulatory T cells and a trend toward an increase in the frequency of CD3 + CD8 + T cells. There was also a significant increase in memory CD44 + CD4 + T cells (p < 0.05) and activated CD69 + CD8 + T cells (p < 0.05) within the Braf/Pten/Cxcr2−/− tumors (Fig. 3C, S5D). We also validated these results with immunohistochemical staining of tumor sections (Figure S6A). FACS analysis of peripheral blood cells revealed no significant difference in any immune population between Braf/Pten/Cxcr2−/− mice and Braf/Pten/Cxcr2WT mice before or after tumor formation. (Figure S5B, S6B).

Fig. 3figure 3

The immune infiltrate of BrafV600E/Pten−/− tumors is altered with loss of Cxcr2. a mMCPCounter analysis performed on bulk RNAseq data from BrafV600E/Pten−/−melanoma tumors with or without Cxcr2 predicts significantly enhanced infiltration of T cells, CD8 + T cells, monocytes, NK cells, and lymphatic vessels into Cxcr2−/− tumors. b FACS analysis of CD45 + myeloid cells in BrafV600E/Pten−/− melanoma reveals decreased MDSC-like cells in Cxcr2−/− tumors. c FACS analysis of CD45 + cells in BrafV600E/Pten−/− melanoma tumors identified changes in activated CD4 + CD44 + T cells and CD8 + CD69 + T cells. d Cytokine array for 62 cytokines expressed in TME of BrafV600E/Pten−/− tumors revealed one major cytokine, CCL20, that is strongly upregulated with loss of Cxcr2 (n = 4/genotype) based on net density. These data are complemented by increased Ccl20 mRNA with loss of Cxcr2 in BrafV600E/Pten−/− tumors. e Cxcl9, Cxcl10, and PD-L1 expression based upon RNAseq analysis from BrafV600E/Pten−/− tumors expressing or not expressing Cxcr2 in melanocytes. All statistical significance determined via Welch’s t-test

The identified differences in immune cell infiltrate are highly suggestive of altered cytokine signaling within the TME. Therefore, a 62-cytokine array was performed on Braf/Pten/Cxcr2WT (n = 4) and Braf/Pten/Cxcr2−/− (n = 4) tumor lysates. CCL20, an inflammatory chemokine that is highly chemotactic for CCR6-expressing lymphocytes and dendritic cells, is strongly upregulated (24-fold) in the Braf/Pten/Cxcr2−/− TME (Fig. 3D). In addition, RNAseq analysis revealed a significant increase in PD-L1 expression in tumors from Braf/Pten/Cxcr2−/− mice compared to Braf/Pten/Cxcr2WT mice (Fig. 3E). Furthermore, M-CSF, eotaxin, and MIP-2 were slightly increased, which could contribute to myeloid cell infiltration, and there was a slight decrease in IL-1β in the tumors from Braf/Pten/Cxcr2−/− mice as compared to tumors from Braf/Pten/Cxcr2WT mice (Figure S5C). These data suggest that targeted deletion of Cxcr2 in melanocytes during tumorigenesis results in a marked increase in CCL20 and additional subtle changes in the cytokine milieu of the TME.

CXCR1/CXCR2 antagonist SX-682 inhibits Braf V600E/Pten −/− and NRas Q61R/Ink4a −/− melanoma tumor growth and promotes anti-tumor immunity

Having established the importance of Cxcr2 in the development, growth, and TME of Braf/Pten melanoma tumors, we sought to evaluate the therapeutic potential of systemic CXCR1/CXCR2 inhibition. Thus, chow containing the CXCR1/CXCR2 antagonist SX-682 [35] was administered to four-week-old mice. After two weeks of eating vehicle control or SX-682-containing chow, 4-HT was applied to the backs of the mice for three successive days. Following a month of continuous feeding on control or SX-682-containing chow, we observed that Braf/Pten mice fed SX-682-containing chow exhibited a trend toward reduction in tumor volume compared to mice fed vehicle control chow (Fig. 4A, p = 0.07; 802.5 ± 724.01mm3 for control; 230.20 ± 373.21 mm3 for SX-682). Moreover, there was a trend toward decreased tumor formation in SX-682-fed mice (p = 0.145), where only 40% (4/10) of SX-682-fed mice developed tumors compared to 75% (6/8) of control-fed mice (Fig. 4A). Similarly, NRasQ61R/Ink4a−/− (NRas/Ink4a) mice were fed chow containing SX-682 or control chow, and tumors that developed over five months were counted and measured. We observed that SX-682 treatment significantly suppressed tumor growth (p = 0.041, Fig. 4B) but only trended toward a decrease in tumor incidence (p = 0.111, Fig. 4B). Overall, SX-682 produced inhibition of tumor volume comparable to that of CXCR2 loss in melanocytes but did not impact tumor formation as significantly. This suggests that at our current dosage of SX-682 in the chow, we are unable to achieve complete suppression of CXCR2 at the time of tumor initiation.

Fig. 4figure 4

SX-682 affects BrafV600E/Pten−/− and NRasQ61R/Ink4a−/− tumorigenesis. a BrafV600E/Pten−/− and b NRasQ61R/Ink4a−/− mice were fed chow containing SX-682 or vehicle continuously through tumor formation, and tumors were measured and counted. Significance was determined using a Welch's t-test. c A volcano plot showing fold change and significance of differential gene expression between tumors from SX-682-fed and control-fed BrafV600E/Pten−/− mice. d Gene set enrichment analysis of SX-682 treated or control BrafV600E/Pten.−/− tumors identifies gene sets enriched in SX-682 treated tumors (positive normalized enrichment score) or enriched in control tumors (negative normalized enrichment score)

RNA sequencing analysis of control and SX-682 treated tumors from Braf/Pten mice identified nearly 3000 differentially expressed genes with many trends similar to those observed in Braf/Pten/Cxcr2−/− tumors. A volcano plot shows that a significant number of genes were strongly up or down-regulated (log2 fold change of > 3) with a very high level of significance (-log10(P-adj) > 50) (Fig. 4C). Upregulated genes include those involved in regulation of growth, proliferation, and cell cycle; tumor suppression; differentiation/stemness; immune regulation; and motility and adhesion. Genes downregulated in response to Cxcr1/Cxcr2 antagonism with SX-682 include those involved in cell adhesion and cell proliferation, cell cycle, and growth (Figure S3B).

GSEA of the tumors from Braf/Pten mice treated with SX-682 revealed a significant increase in CD8 + T cell activation, with trends toward increased T cell-mediated immune response to the tumor, immune response to tumor, adaptive immune response, antigen processing and presentation, CD4 + T cell activation, stem cell differentiation, CD8 + T cell proliferation, T cell-mediated cytotoxicity, and lymphocyte activation. There were significant decreases in genes involved in melanocyte proliferation, cell cycle process, cell cycle, stem cell division, and cell cycle G1-S transition (Fig. 4D). mMCPcounter analysis of the tumor RNAseq data predicted an increase in CD8 + T cells (Fig. 5A) and monocytes (Figure S7A), and a decrease in B-derived cells and cells of the lymphatics (p < 0.01) in tumors from the SX-682-treated Braf/Pten mice (Figure S7A).

Fig. 5figure 5

SX-682 alters the immune profile of BrafV600E/Pten−/− melanoma. a mMCPCounter analysis of bulk RNAseq data predicts enrichment for CD8 + T cell infiltrate into tumors following treatment with SX-682 (p < 0.05). b FACS analysis confirms a trend toward increased CD8 + T cells in SX-682 treated BrafV600E/Pten−/− melanoma. c FACS analysis of CD45 + myeloid cells indicated a significant decrease in immunosuppressive CD11b + Ly6G + cells, but no change in total CD11b + cells. d A cytokine array was performed on control and SX-682 treated tumors, identifying a notable decrease in Vegf and an increase in Tnfα. e Cxcl9, Cxcl10, and Pd-l1 expression based upon RNAseq analysis from SX-682 or control treated tumors. All statistical significance determined via Welch’s t-test

FACS analysis of SX-682 treated Braf/Pten tumors revealed a trend toward increased CD8 + T cells (p = 0.17), no change in CD11b + cells, and a significant decrease in CD11b + Ly6G + cells (p < 0.001) (Fig. 5B, C). Additional FACS analysis of tumor CD45 + cells showed a decrease in CD4 + CD3 + cells (p < 0.05) in tumors from the SX-682 chow-fed mice (Figure S7C). In peripheral blood, there was a significant decrease in CD44 + CD4 + T cells and CD62L + CD4 + T cells and a trend toward increased CD69 + CD8 + T cells from mice fed SX-682 chow (p = 0.059; Figure S7B). In addition, a cytokine array of tumor lysates (n = 4 for each genotype) revealed a marked reduction in VEGF, indicating a reduction in tumor angiogenesis, and an increase in TNFα, indicating a more inflammatory tumor microenvironment (Fig. 5D). Moreover, RNAseq analysis of Braf/Pten tumors revealed that SX-682 induces expression of Cxcl9, Cxcl10, and Pd-l1 (Fig. 5E). Altogether, these data indicate that SX-682 alters the TME to stimulate anti-tumor immunity and reduce tumor growth.

When we evaluated the hematological effects of SX-682 in a rodent model during the IND-enabling toxicology assessment, we saw reversible neutropenia following treatment with no significant effect on other peripheral blood components (Figure S7D). These data indicate that SX-682 more widely affects the immune cell population in tumor bearing as compared to tumor free rodents. We also evaluated the effects of short term SX-682 treatment on normal mice. C56BL/6 mice were treated with 50 mg/kg SX-682 daily via oral gavage for 4 days. The peripheral blood leukocytes were analyzed by FACS, and we found that SX-682 reduced the percentage of Ly6G + cells that were CD14 + (p = 0.04) and increased the percentage of CD45 + cells that were CD19 + (p = 0.026) (Figure S7E).

SX-682 treatment of Melan-A, B16F0, and B16F10 cells reveals tumor cell-specific gene modulation

Our murine experiments involved bulk RNA sequencing of tumors that contain tumor cells in addition to stromal and immune cells. To identify the specific effect of SX-682 treatment on tumors without the contribution of other cell types, we investigated the effect of SX-682 on non-tumorigenic Melan-A cells, tumorigenic B16F0 cells, and metastatic B16F10 cells in vitro. First, we evaluated Cxcr2 expression and found that B16F0 and B16F10 cells express significantly more Cxcr2 than Melan-A cells, as evaluated by mRNA levels and surface protein labeling (Fig. 6A, B). We then analyzed the effect of SX-682 (5 μM) on the growth of these cells and observed that SX-682 treatment resulted in a small but significant inhibition of growth in B16F0 and B16F10 cells in vitro based on the percentage of cells staining positively for Ki-67 (Fig. 6C) and cell number (Figure S8A). In addition, SX-682 treatment of B16F0 and B16F10 cells in vitro also reduced production of both Cxcl1 (KC) and vascular endothelial growth factor (Vegf) as evaluated by cytokine array (Fig. 6D), again indicating the potential for SX-682 to impact the immune profile of the TME.

Fig. 6figure 6

Tumor cell-specific impacts of SX-682. CXCL1 and CXCR2 expression on Melan-A, B16F0, and B16F10 cells based on a the NCBI database and b CXCR2 expression in Melan-A, B16F0 and B16F10 cells based on flow cytometry. c Cell lines were treated with 5 µM SX-682 (or DMSO control) for 4 days prior to staining with Pacific Blue-Ki67 for FACS analysis. The percentage of positive staining cells was significantly decreased in the SX-682 treated cells for all cell lines (analyzed using a two-way ANOVA with Benjamini and Hochberg (BH) correction for multiple tests). d Cytokine array of SX-682-treated Melan-A, B16F0 and B16F10 cells shows that SX-682 treatment reduced the expression of KC and VEGF in all three cell lines

To identify tumor cell-specific transcriptional changes following SX-682 treatment, we performed bulk RNAseq analysis on each of the three cell lines. Of the total differentially expressed genes, expression of 4024 genes was altered in all three lines. An additional 860 genes were differentially expressed in both tumorigenic B16F0 and B16F10 lines in response to SX-682 (Figure S3C and 7A ). Commonly upregulated genes include those involved in apoptosis and cell stress response and suppression of gluconeogenesis. In contrast, commonly down-regulated genes include those involved in methylation, RNA splicing, and cell cycle processes (Figure S3C and S8B). Reverse phosphoprotein analysis (RPPA) identified SX-682-induced decreases in phosphoproteins involved in growth (AKT, BRAF, pS445-BRAF, CDC2-pY15, CDC6, GSK-3b, mTOR, mTOR pS2448, MMP14, PAX8, and S6), as well as SX-682-induced increases in immunomodulatory proteins (STING, PD-1, PD-L1, TRIM25, and ANNEXIN I); proteins involved in the regulation of apoptosis (PUMA, BLC2, BLC2A1, BCLxL, Smac); tumor suppressors (TSC2, WTAP); and cell cycle regulators (CDC25, CDC42, PLK1, EGFR, PRAS40_pT246). Of interest, β-catenin expression is increased following SX-682 treatment. This is counter-intuitive for SX-682 inhibition of tumor growth, as the Wnt/β-catenin pathway often drives melanoma tumor growth and metastasis. However, we observed that the phosphorylated forms of β-catenin (pT41 and pS45) that enable its ubiquitin-mediated degradation are increased as well. This indicates that β-CATENIN is marked for degradation, thus diminishing the potential for enhanced tumor growth. There were also increases in proteins involved in motility: MYOSIN-Iia, PAK, CDC-42, MYOSIN Iia-pS1943, and HMGA1 (Figure S8C, D). Finally, there were only subtle changes in cytokine expression in response to SX-682 treatment in vitro, and these were inconsistent across the three cell lines (Figure S8E). Altogether, these results suggest that multiple compounding signals are induced in cells treated with SX-682, including a decrease in growth signaling, modulation of apoptosis, enhanced anti-tumor immunity, and altered cell cycle processes.

Tfcp2l1 distinguishes the Cxcr2WT from the Cxcr2 perturbed phenotype

To better understand the complex transcriptional reprogramming that occurs when CXCR2 activity is diminished via knockout or with SX-682 treatment, we compared differentially expressed genes in Braf/Pten/Cxcr2−/− tumors, SX-682-treated tumors, and SX-682-treated tumorigenic B16F0 and B16F10 cell lines compared to controls. We noted that based upon our search for genes with a minimum of a log2 fold change > 2 and a p-value < 0.05, only one gene stood out as significantly upregulated across all four models compared to the respective controls: Transcription factor CP2 like-1(Tfcp2l1) (Fig. 7A, B). To verify the RNA sequencing results, we performed RT-PCR analysis of RNA samples from MelanA, B16F0, and B16F10 cells to determine Tfcp2l1 expression. With this assay, we show that SX-682-treatment elevates Tfcp2l1 expression in the tumorigenic cell lines (Figure S10).

Fig. 7figure 7

Tfcp2l1 is commonly upregulated across three models of CXCR2 perturbation. a, b In comparing expression data from BrafV600E/Pten−/−/Cxcr2−/− tumors. BrafV600E/Pten−/− tumors treated with SX-682, and B16F0 and B16F10 cell lines treated with SX-682, Tfcp2l1 was consistently upregulated compared to appropriate controls (as determined by Welch's t-test). c Log2 fold change for Tfcp2l1 and related genes across experimental groups based upon RNAseq analysis. d Identification of transcription factors central to Weighted Correlation Network Analysis (WGCNA) co-expressed gene modules (by kME) and significantly differentially expressed between BrafV600E/Pten−/−/Cxcr2−/− and BrafV600E/Pten−/−/Cxcr2WT tumors. TFs are colored by gene module and show varying levels of centrality to each module and importance in distinguishing WT and KO tumors. Turquoise dots represent transcription factors that are up in the BrafV600E/Pten−/−/Cxcr2−/− tumors and blue dots represent transcription factors that are up in the BrafV600E/Pten−/−/Cxcr2WT tumors

TFCP2L1 is a member of the TFCP2/TFCP2L1/UBP1 subfamily of transcription factors that contributes to the maintenance of stemness in pluripotent stem cells and can also exhibit tumor suppressive activity and modulate differentiation [26, 36,37,38,39]. The Krupple-like Factor (KLF) family of transcription factors works with and can be induced by Tfcp2l1 to modulate induction and maintenance of naïve pluripotency in mouse primordial germ cells [40,41,42]. It has been previously reported that TFCP2L1 is positively associated with expression of pluripotency genes including Nanog, Oct4, Sox2, and Esrrb in mouse embryonic stem cells [42]. However, our data suggest a complex relationship between Cxcr2 perturbation and Tfcp2l1-related gene expression. In the Braf/Pten model, stemness marker Esrrb and neural crest markers Foxd3 and Sox10 were decreased when Cxcr2 was deleted in tyrosinase expressing cells. In contrast, stemness markers Tfcp2l1, Klf4 and Hmga2, were increased. In SX-682 treated Braf/Pten model, there was a trend toward a decrease in stemness marker Esrrb, a significant decrease in the neural crest marker Sox10, and a small but significant decrease in the melanoblast marker Mitf (Figs. 7C, S8A-L). The melanocyte differentiation marker Tyr was increased in both the CXCR2−/− and the SX-682 treated Braf/Pten mouse models. In the B16F0 and B16F10 cells, RT-PCR analysis revealed that stemness markers Esrrb, Hmga2, Myc, Sox2, neural crest marker Sox10, and melanoblast marker Mitf were significantly decreased in response to SX-682-treatment in vitro. Foxd3 was significantly decreased in B16F10 and trended toward significant reduction in B16F10. In contrast, stemness markers Tfcp2l1, Nanog and Notch1 were increased, while there was a decrease in Tyr expression (Figure S10). Altogether, these data imply that with ablation of Cxcr2 activity, there is an increase in some stemness markers but a

留言 (0)

沒有登入
gif