CBL0137 inhibits proliferation in MYC-high triple-negative breast cancer cells.
Considering that MYC mediates tumor intrinsic malignant properties, as a first step, we aimed at characterizing the biologic effects of CBL0137 in TNBC cells. Analysis of a panel of 10 TNBC cell lines (Fig. 1A) showed that CBL0137 significantly inhibited the proliferation of MYC-high (4/6 TNBC) compared to MYC-low (2/4 TNBC) cell lines (Fig. S1A). MYC-high TNBC cell lines were significantly more sensitive to CBL0137 (Mean IC50 1.39 µM ± SEM 0.56) than MYC-low TNBC cells (Mean IC50 13.35 µM ± SEM 4.3) (Fig. 1B, C). Moreover, CBL0137 significantly enhanced death of MYC-high cells (Mean Fold Increase 4.15 ± SEM 0.53 at 1 µM CBL) when compared with MYC-low cells (Mean Fold Increase 1.272 ± SEM 0.18 at 1 µM CBL) (Fig. S1B & S1C).
To investigate whether CBL0137 selectively inhibits the growth of MYC-high TNBC cells, MYC was ectopically over-expressed in two MYC-low cell lines, MDA-MB-361 and MDA-MB-157 (Fig. 1D) that were then treated with different concentrations of CBL0137 for 72 hours. MYC over-expression itself did not affect cell survival, however, it significantly sensitized both MDA-MB-361 and MDA-MB-157 cells to CBL0137-mediated growth inhibition as compared to parental cells (Fig. 1E). In addition to the transcriptional downregulation of MYC [21], CBL0137 is known to exert anti-cancer activity by inhibiting the NF-kβ pathway and via trapping of SSRP1 into chromatin [14, 16, 17]. Hence, to verify in the TNBC cell context whether CBL0137 effect is dependent on MYC, NF-kβ or SSRP1, either MYC, SSRP1, or p65 gene expression was down-regulated by siRNA in MYC-high SUM159PT and SUM149PT cells (Fig. 1F). Interestingly, siRNA-mediated MYC knockdown reduced cell growth by approximately 50–60% in these two MYC-high TNBC (SUM159PT and SUM149PT) cell lines which was comparable to CBL0137 (1 µM) treatment alone in cells transfected with scramble/control siRNAs (Fig. 1G), suggesting that MYC-high lines have survival-dependency on MYC. Interestingly, CBL0137 treatment did not further reduced cell growth in SUM159PT and SUM149PT cells after MYC knockdown, suggesting that CBL0137 effect is epistatic with MYC-depletion in MYC-high TNBC lines (Fig. 1G). In contrast, SSRP1 and p65 knockdown had no effect on cell growth in both MYC-high TNBC cell lines and CBL0137 treatment exerted a comparable growth inhibitory effect in control and SSRP1 or p65 knocked down (siRNA) SUM159PT and SUM149PT cells (Fig. 1G). Taken together, these data suggested that in TNBC cell context, CBL0137-mediated growth inhibition is MYC dependent and SSRP1 and p65 independent. These results are consistent with the survival dependency of MYC-high TNBC cell lines on MYC but not SSRP1 and p65 (Fig. S1D). Lack of NF-kβ and SSRP1 involvement was further demonstrated at the protein level in SUM159PT and SUM149PT TNBC cells treated with 2.5 µM CBL0137 for 0, 4, and 8 hours. Although CBL0137 markedly reduced MYC protein levels within 4 hours of treatment in both cell lines (Fig. S1E), the drug did not change SSRP1 protein levels in the whole cell lysates and did not inhibit phospho-p65 levels in either cell line (Fig. S1E). Since CBL0137 is known to trap SSRP1 into the chromatin and block RNA Pol II-mediated transcription [14, 16–18], SSRP1 and MYC protein levels were also analyzed in the chromatin fraction of SUM159PT cells. Unlike previous studies, CBL0137 did not trap SSRP1 into the chromatin within 4 hours of treatment but significantly reduced the levels of MYC protein in the chromatin fraction (Fig. S1F). These data are consistent with more recent studies showing that CBL0137 can transcriptionally downregulate MYC expression via inhibition of the enhancer-promoter communication in neuroblastoma and medulloblastoma [17, 19, 21]. Together these data indicate that CBL0137 inhibits the growth of MYC-high TNBC cells in a MYC-dependent, but NF-kβ- and SSRP1-independent manner.
Next, we investigated whether MYC down-regulation was associated with the modulation of known MYC target genes in CBL0137-treated SUM159PT and SUM149PT TNBC cells. Four-hour CBL0137 treatment significantly reduced MYC mRNA levels by 90% compared to untreated cells in both SUM159PT (Fig. S1G) and SUM149PT (Fig. S1H) cells. Additionally, CBL0137 also significantly reduced the mRNA levels of several MYC target genes in both cell lines, including MAT2A, PP1A, HK2, PGK1, RPL5, RPL9, PYCR1, and CDK2 (Fig. S1G & S1H). These findings indicate that CBL0137 inhibits proliferation of human MYC-high TNBC cells by strongly down-regulating MYC and its downstream targets at the transcriptional level.
CBL0137 inhibits TNBC growth in vivo via both direct cytotoxicity and immune modulation.
We next analyzed the direct in vivo anti-cancer activity of CBL0137 using the MYC-high MDA-MB-231 TNBC cells (Fig. 1A) orthotopically injected into immunocompromised NSG mice [37, 38]. CBL0137 significantly reduced primary tumor growth (Fig. 2A & S2A) when compared to the vehicle-treated group, with no evident signs of systemic toxicity (Fig. S2B). Therapeutic efficacy of CBL0137 was also tested in the fully immunocompetent and syngeneic 4T1.2 mouse model of MYC-high TNBC (Fig. 1A) that aptly recapitulates human TNBC features including spontaneous metastasis upon orthotopic injection [39]. In vitro data confirmed the anti-proliferative effects of CBL0137 on 4T1.2 cells (IC50 0.74 µM) (Fig. S2C). In vivo, CBL0137 significantly reduced 4T1.2 tumor growth when compared to vehicle-treated mice (Fig. 2B). Since CBL0137 has been recently shown to exert part of its anti-tumor activity by increasing CD8+ T-cell intratumor infiltration and activation in a murine model of colon carcinoma [40], we investigated whether CBL0137 anti-tumor activity in the 4T1.2 model is also at least partly dependent on modulation of adaptive immunity. Towards this goal, 4T1.2 cells were orthotopically injected into Balb/c RAG2 knockout (KO) or WT mice (28) and treated with CBL0137. Although CBL0137 significantly reduced 4T1.2 tumor growth in both mouse strains (Fig. 2B & 2C), the extent of growth inhibition was significantly higher in fully immunocompetent mice (67% vs 25% at day 18) (Fig. 2D). These data suggest that the anti-tumor effects of CBL0137 in the 4T1.2 model are at least in part mediated by T cells and/or B cells.
To decipher the mechanism mediating the in vivo anti-cancer activity of CBL0137, we first confirmed down-regulation of Myc mRNA in vivo by RT-qPCR in 4T1.2 tumors collected 7 days after a single CBL0137 treatment (Fig. 2E). Bulk RNA sequencing (RNA-seq) and Gene Set Enrichment Analysis (GSEA) also confirmed down-regulation of MYC-target genes in the same tumors as compared with controls (n = 4/group) (Fig. 2F; Fig. S3A; Supplementary Data File 1). Interestingly, a significant upregulation of genes associated with both IFNγ- and IFNα-induced responses was also observed in CBL0137-treated tumors (Fig. 2G). Moreover, a single dose of CBL0137 significantly increased IFNγ levels in the serum of 4T1.2 tumor-bearing mice (Fig. 2H). The upregulation of key IFNγ pathway genes (Oas2, Oas3, Ifit1, Ifit2, Ifit3, Ifih1, and Irf7) in CBL0137-treated 4T1.2 tumors was independently validated by RT-qPCR (Fig. S3B). CBL0137-mediated modulation of immune-related genes was further highlighted by Gene Ontology (GO) pathway analysis, where 7 out of 10 most upregulated pathways were associated with immune responses, including responses to viruses and Type I interferon (Fig. 2I). Moreover, mRNAs encoding for cytokines (Il-1b), chemokines (Ccr2 and Cxcl9) as well as cell surface receptors involved in innate immunity (Clec7a and Clec12a) were among those most significantly downregulated in CBL0137-treated 4T1.2 tumors (Fig. S3A & S3C). Downregulation of these genes has been shown to activate anti-tumor immune responses in multiple cancer models [41–43]. These data demonstrated the ability of CBL0137 to induce a robust interferon-mediated response and immune-stimulating effects in a murine model of MYC-high TNBC cells in vivo.
CBL0137 induces immunogenic cell death in human and murine breast cancer cell lines.
The in vivo immunostimulatory effects and the induction of type I IFN responses described above prompted us to investigate whether CBL0137 triggers immunogenic cell death (ICD) in TNBC cells. CBL0137 induced a significant and dose-dependent increase in the proportion of 4T1.2 apoptotic cells, particularly the percentage of early apoptotic cells expressing Calreticulin on their cell surface (Fig. 3A, S4A, S5A & S5B) a known ICD marker [44]. Of note, CBL0137 also induced ICD markers in human TNBC cell lines expressing high levels of MYC (Fig. 3B, 3C, S4B, S5C & S5D), with markedly reduced effects in a MYC-low cell lines (Fig. 3D). Moreover, we also examined the effect of CBL0137 treatment on HMGB1 expression, a late marker of ICD [45], on MYC-high SUM159PT and SUM149PT TNBC cells. CBL0137 treatment significantly increased HMGB1 release in SUM159PT and SUM149PT cells in a concentration-dependent manner (Fig. S6A). To further confirm if CBL0137 induces ICD via MYC downregulation, we examined if siRNA-induced MYC downregulation induces ICD in MYC-high TNBC cells. To this end, SUM159PT and SUM149PT cells were transfected with either control or MYC-specific siRNAs, and HMGB1 release was analyzed 48 hours post-transfection. Results showed that MYC downregulation significantly increased HMGB1 release in both SUM159PT and SUM149PT TNBC cells, suggesting that MYC downregulation induces ICD in MYC-high TNBC cells (Fig. S6B). These data strongly support CBL0137-mediated induction of ICD via MYC downregulation in both murine and human TNBC cells expressing abnormally high levels of MYC. To characterize the immunogenicity of CBL0137-treated cells, we investigated their ability to elicit an anti-tumor immune response in vivo to protect mice from re-challenge with live tumor cells. As shown in Fig. 3E, mice injected with CBL0137-treated 4T1.2 TNBC apoptotic cells showed a significant delay in the growth of tumors in re-challenged mice when compared with those previously injected with 4T1.2 cells killed by freezing and thawing or naïve mice. These data together suggest that CBL0137 induces ICD which stimulates a partially protective anti-tumor immunity in vivo.
In vivo CBL0137 treatment induces tumor-specific immune responses and local immunomodulation.
To investigate the ability of CBL0137 to stimulate the generation of tumor-specific T cell responses in vivo, Balb/c mice were orthotopically injected with 4T1.2 TNBC cells and treated as described before with 2 weekly injections of CBL0137 (Fig. 4A). After confirming tumor growth inhibition in the absence of evident toxicity (Fig. 4B & 4C), spleens were collected from treated and control mice, and splenocytes were re-stimulated ex vivo with CD4 and CD8 epitope peptides derived from universal tumor associated antigens (Telomerase and Survivin). Flow cytometry analysis on re-stimulated splenocytes, showed that mice treated with CBL0137 had significantly higher numbers of tumor-specific T cells, as shown by the higher percentages of T cells expressing IL-2, IFNγ or TNFα, particularly in the CD4+ T cell compartment (Fig. 4D). These data indicate that CBL0137 promotes the generation of tumor-specific T cell responses that can contribute to the observed therapeutic efficacy.
Local induction of ICD not only increases tumor cell immunogenicity, but also promotes the release of soluble mediators such as ATP, HMGB1 and other endogenous danger-associated molecular patterns (DAMPs) that trigger pro-inflammatory responses and promote recruitment and activation of immune cells [46]. To investigate the in vivo immunomodulatory effects of CBL0137 on the tumor microenvironment, we performed Opal multi-plex immunohistochemistry (IHC) and multi-parametric flow cytometry on tumors collected 10 days after the last CBL0137 treatment and characterized infiltration, activation and exhaustion profiles of T and NK cells in treated and untreated tumors [33]. Opal analysis revealed that CBL0137 reduced the relative number of CD8+ T cells and NK (NCR1+) cells per 1000 nuclei with no significant change in the proportion of total CD4+ T cells and Treg (CD4+ CD25+ FoxP3+) cells (Fig. S7A). In line with Opal staining results, flow cytometry profiling also showed a trend towards decreased percentage of T cells and NK cells in CBL0137-treated tumors compared to vehicle-treated tumors, although not statistically significant (Fig. S7B & S7C). Despite the apparent reduction in the relative number of effector cells, CBL0137 increased the proportions of cells expressing the early activation marker CD69 in all immune effector compartments e.g., CD3+CD8+ T cells, CD4+FOXP3neg Tconv cells and CD3negCD49b+ NK cells (Fig. 5A). CBL0137 also significantly increased the proportion of CD8+ T cells and CD4+ Tconv cells expressing the CD44 memory marker (Fig. 5A). Furthermore, a significant increase in CD8+ T cells and NK cells expressing the co-stimulatory and activation marker DNAM1 was also observed in treated tumors. Of note, the level of expression of DNAM1 was also significantly increased by CBL0137 in CD8+ T cells (Fig. S8A). All together, these data indicate activation of both adaptive and innate effector cells infiltrating CBL0137-treated tumors.
To investigate the extent of exhaustions of tumor infiltrating CD8+, CD4+ Tconv and NK cells, the expression of seven inhibitory immune checkpoints (CTLA4, LAG3, NKG2A, PD1, TIGIT, TIM3, and VISTA) was also analyzed within the same flow cytometry panel. We observed a consistent trend towards higher proportions of immune effector cells expressing inhibitory immune checkpoint molecules (ICMs) in CBL0137-treated tumors, with NKG2A and TIGIT being among the most frequently expressed in all effector compartments and upregulated by the treatment (Fig. 5B). Boolean gating was then used to evaluate the extent of co-expression of the seven ICMs in the three immune cell compartments. The progressive upregulation of these inhibitors determines the hierarchical loss of effector functions and acquisition of more dysfunctional/exhausted phenotypes. CBL0137 treatment resulted in a significant increase in the proportions of CD8+, CD4+ Tconv and NK cells co-expressing more than 1 ICM and even up to 6 ICMs (Fig. 5C). Boolean data were further interrogated to identify the most frequently represented ICM combinations in each immune cell compartment and those significantly increased by CBL0137 (Fig. S8A-S10A). As summarized in Figures S8B-S10B, TIGIT and NKG2A were the ICMs most frequently represented in the CD8+ and NK cells cell subpopulations significantly expanded by CBL0137. In particular, all immune effector compartments infiltrating 4T1.2 tumors have high proportions of TIGIT+ cells, and consequently TIGIT is frequently found in combination with other ICMs (Fig. 5B, S8B-S10B). By contrast, despite the low proportions of NKG2A+ cells within CD8+ T cells, NKG2A was found in most subpopulations significantly expanded by CBL0137 (Fig. S9B-S11B). Moreover, the proportion of NKG2A+ cells was significantly higher in tumor infiltrating CD8+ T cells also expressing CD69+ or GZB+ as well as in CD69+ CD4+ Tconv cells (Fig. 6A & S12). Notably, enrichment in NKG2A+ cells was further enhanced by CBL0137 in these subpopulations (Fig. 6A; red asterisks). TIGIT+ cells were also enriched in activated cells, however to a lower extent which did not always reach statistical significance and was not further enhanced by CBL0137.
TIGIT is highly expressed on human and murine immune cells infiltrating several tumor types where it negatively regulates anti-tumor responses by binding CD155 or CD112, which are widely expressed on tumor cells. These same ligands are shared with DNAM1, which, on the contrary, promotes cytotoxicity and enhances anti-tumor responses. Therefore, when co-expressed on the same cell, the balance between these two receptors determines the shift between stimulatory and inhibitory responses. Figure 6B (left panel) shows the distribution of single and double positive cells for DNAM1 and TIGIT. In all three effector compartments (CD8+, CD4+ Tconv and NK cells), CBL0137 significantly reduced the proportion of DNAM1negTIGITneg cells and significantly increases the proportion of DNAM1+TIGIT+ cells. On note, in all DNAM1+TIGIT+cells, no changes in the ratio between DNAM1 and TIGIT expression (MFI) was observed upon CBL0137 treatment (Fig. 6B; right panel). This data suggests that the resulting balance between activation and inhibition mediated by the TIGIT/DNAM1 axis might not be affected by CBL0137 treatment.
CBL0137 upregulates Qa-1 b on tumor cells and synergizes with NKG2A blockade in inhibiting in vivo growth.
NKG2A exerts its inhibitory effects by engaging with HLA-E (in humans) or Qa-1 (in mice) [47]. HLA-E protein levels are generally higher in tumor cells than in healthy tissues and available evidence indicates that IFNγ can upregulate HLA-E expression [48–50].
Given the robust interferon response induced by CBL0137 (Fig. 2), we examined if CBL0137 modulates Qa-1b mRNA levels in 4T1.2 tumors in vivo. As shown in Fig. 6C, significantly higher Qa-1b mRNA levels were observed in CBL0137-treated tumors compared to controls. These results were confirmed in vitro where treatment of 4T1.2 cells with CBL0137 also increased Qa-1b mRNA levels in a dose-dependent manner (Fig. 6D). We next investigated whether upregulation of Qa-1b was mediated by the increased levels of IFNγ induced by CBL0137 in these cells. 4T1.2 cells were transfected with either scrambled siRNAs (siCtrl) or a combination of siRNAs specific for IFNγ receptor 1 and 2 (siIfngR), and then treated with or without 0.5 µM CBL0137 for 24 hours. Notably, CBL0137 treatment of siCtrl cells significantly upregulated (i) the mRNA levels of both IFN receptors (Fig. 6E) and (ii) the levels of secreted IFNγ (Fig. 6F). IFNγ receptor 1 and 2 knockdown, confirmed by RT-qPCR (Fig. 6E), abrogated the increase in IFNγ secretion induced by CBL0137 (Fig. 6F). This is consistent with the possible positive feedback loop sustaining the level of IFNγ production [49, 50]. Interestingly, while CBL0137 significantly increased Qa-1b mRNA levels in siCtrl cells, it failed to increase Qa-1b mRNA levels in IFNγ receptor knocked-down 4T1.2 cells (Fig. 6G). Taken together, our data indicate that CBL0137 increases Qa-1b mRNA expression via IFNγ signaling.
Overall, these results suggest a potential role of the NKG2A/Qa-1b axis in hampering the activation of CD8+ and CD4+ Tconv cells. To exploit these results for therapeutic purposes, mice bearing 4T1.2 tumors were treated with CBL0137 in combination with anti-NKG2A antibody (Fig. 7A). As shown in Fig. 7B & 7C, the combination treatment significantly inhibited tumor growth when compared with untreated control mice or mice receiving single therapies. Notably, IHC analysis for cleaved Caspase-3 revealed that CBL0137 combined with anti-NKG2A significantly increased the percentage of apoptotic tumor cells compared to single agent treatments (Fig. 7D & 7E). These findings are consistent with an efficient cooperation between CBL0137 and NKG2A blockade in promoting the induction of therapeutically relevant anti-tumor immune responses in the 4T1.2 TNBC model.