Guadecitabine reduces B16F10 tumor growth in vivo
C57Black6/j mice, challenged subcutaneously with B16F10 cells, received daily treatment with guadecitabine (1mg/kg) starting 3 days post melanoma cell injection until day +13 (Figure 1A), when mice were sacrificed, tumors excised and weighted. Continuous treatment with the drug significantly decreased mean tumor volume at day 13 (Figure 1B). Accordingly, tumor weight was significantly lower in guadecitabine treated mice than in controls, supporting the evidence that continuous administration of low dose guadecitabine reduces tumor growth in vivo (Figure 1C). No sign of evident toxicity was observed since no difference in the mean weight of the mice was detected between guadecitabine and the control group during treatment (Figure 1D).
Guadecitabine in combination with anti-PD-1 and anti-CTLA-4 mAbs greatly reduces tumor growth in vivo
We then assessed the effects of guadecitabine in association to anti-PD-1 and -CTLA-4 mAbs (ICBs). Mice were subcutaneously challenged with B16F10 cells. After three days, mice were randomized in eight groups that received daily treatment with guadecitabine either alone or in association with anti-CTLA-4 and/or anti-PD-1 mAbs and with anti-CTLA-4 and/or anti-PD-1 mAbs (Figure 2A, B). Tumor growth of mice who received guadecitabine/anti-CTLA-4/anti-PD-1 (guadecitabine/ICBs; triple therapy) was compared to control or ICBs treated mice.
Mice treated with guadecitabine/ICBs showed the most significant growth reduction at any day of measurement in comparison to the control group. Guadecitabine showed a greater effect when associated with anti-CTLA-4 than to anti-PD-1 mAb suggesting that epigenetic modulation may preferentially synergize with blockade of a specific immune checkpoint (Figure 2B). Yet triple therapy showed a stronger antitumor effect than either guadecitabine/anti-CTLA-4 (guad/α-CTLA-4), guadecitabine/anti-PD-1 (guad/α-PD-1) or guadecitabine alone (Figure 2B). The analysis of the growth of single tumors (Figure 2C) showed considerable variability without hiding the significant effects that the addition of guadecitabine had on the efficacy of ICBs. No macroscopic sign of toxicity was observed, since mean mice weights were similar between the different groups of treatment (Figure 2D).
Guadecitabine in combination with anti-PD-1 and anti-CTLA-4 mAbs reshapes TME to anti-tumor responsiveness
TME modifications induced by different treatments were analysed by flow cytometry on cell suspensions recovered from tumors. Guadecitabine deeply modified tumor cells within TME inducing the expression of MHC-class I, but not -class II molecules, potentially enabling B16F10 cells to present antigens to CD8+ T cells (Figure 3A). We found that ICBs increased CD39, an ectonucleotidase involved in the conversion of ATP to immunosuppressive adenosine [36,37], on B16F10 melanoma cells but this effect was efficiently counteracted by guadecitabine. Notably, CD39 was found highly expressed by different tumors including melanoma [38,39]. CD39 expression by tumor cells favors resistance to chemotherapy and immunotherapy so that trials testing the therapeutic efficacy of CD39 blockade, alone or in combination with ICBs, are under way [40].
In addition, ICBs, with or without guadecitabine, downmodulated the expression of TIM3, an alternative immune checkpoint, particularly expressed on B16F10 cells (Figure 3A), similarly to what found on human melanoma cells [41]. CTLA-4 and PD-L1 expression on B16F10 cells was negligible in all tested conditions (Figure 3A).
To study TME reshaping we also analyzed the inflammatory tumor infiltrate. Total T cells, as well as CD4+ and CD8+ T cell subset, frequencies were comparable among all the different groups of treatment (Supplementary Figure 1A and 2A), although treatments containing guadecitabine showed a non-significant decrease in CD45+ cells. However, remarkable differences among groups were detected focusing on maturation stages and functions of CD8+ T cells. Guadecitabine alone and in combination with ICBs upregulated T cell responses by increasing granzyme production on tumor infiltrating CD8+ T cells, favoring the maturation of CD8+T cells in effector memory (CD44+CD62L-) cells secreting granzyme (Figure 3B). These data suggest that guadecitabine/ICBs are able to unleash maturation pathways within CD8+ T cells committing these cells to effector populations. In this process guadecitabine seems to have a leading role since mice treated with this drug, but not those treated with ICBs alone, showed expansion of granzyme+ CD8+ T cells (i.e., effector T cells already prompted to cytotoxicity), whose frequency strictly paralleled that of effector memory CD8+ T cells (Figure 3B). Although neither guadecitabine nor ICBs showed effects on maturation stage and acquisition of granzyme-related cytotoxic function by CD4+ T cells (Figure 3B), it is of interest the fact that association of guadecitabine and ICBs was responsible for amplification of both CD8+ and CD4+ T cell subsets highly expressing IFN-g, sign of functional activation (Figure 3C). Collectively, this panel of data suggests that guadecitabine/ICBs impact on the composition of the T cell infiltrate leading to activation and maturation of effector/cytotoxic T cell subsets through mechanisms partly complementary between the two type of agents. Finally, guadecitabine/ICBs did not affect the expression of the alternative immune checkpoint TIM-3 on both CD4+ and CD8+ T cells, whereas ICBs alone strongly upregulated TIM-3, thus potentially driving immune escape (Figure 3D). Guadecitabine induced an increase in CTLA-4 and PD-1 expression on T cells yet this reached statistical significance only for CTLA-4 (Figure 3D). The immunostimulatory activity exerted by the combination of guadecitabine/ICBs assumes even more relevance considering that it induced a significant decrease of CD4+CD25+FoxP3+ Treg cells in the tumor infiltrate, indicating that the net effect of this combination treatment is to shift the balance between effector and regulatory T cell functions toward the former one (Figure 4A).
Guadecitabine/ICBs, in fact, reduced the percentages of immunosuppressive CD4+CD25+FoxP3+ Treg cells compared to any other treatment (Figure 4A). Furthermore, guadecitabine alone induced a remarkable reduction of myeloid derived suppressor cells (MDSC), in particular the monocytic subset (Ly6C+Ly6G- CD11b+, M-MDSC) (Figure 4A and Supplementary Figure 2B), associated with a significant increase of macrophages (Figure 4B and Supplementary Figure 2C). Interestingly, combination treatment with guadecitabine/ICBs determined a significant increase of M1 (CD38+Egr2-) macrophages indicating that triple therapy skews towards pro inflammatory anti-tumor TAM-M1 responses (Figure 4B). Taken together, these data indicate that mice receiving triple therapy underwent a shift from an immune suppressive to an immune responsive TME.
Guadecitabine/ICBs enhances T and NK anti-tumor functions in lymphoid organs, upregulates Th1 responses, and downregulates angiogenic chemokines
To assess signs of functional activation and commitment to cytotoxic function in lymphoid organs we analyzed IFN-g and CD107a expression on T and NK cells from spleen and tumor-draining lymph nodes. Guadecitabine/ICBs significantly expanded the frequencies of T and NK cells producing IFN-g in both spleen and tumor-draining lymph nodes, effect associated with an increased frequency of CD107a+ T and NK cells in the spleen (Figure 5 and Supplementary Figure 3).
To corroborate these data we evaluated systemic modifications of cytokine and chemokine levels in the sera from treated and control mice, 17 days after the injection of tumor cells. A strong increase of Th1 cytokine levels, such as IFN-g, TNF-a and IL-2, was observed in the serum of mice treated with guadecitabine/ICBs in respect to control mice or mice treated with guadecitabine alone (Figure 6A). Among Th2 cytokines, only IL-9 was up regulated by guadecitabine containing therapies, while IL-5, IL-6, and IL-13 were not modified (Figure 6B). The systemic increase of IFN-g observed in mice treated with guadecitabine/ICBs and guadecitabine/anti-CTLA-4 is followed by an increase of IFN-g-dependent anti-angiogenic factors, such as MIG and IP10 (Figure 6C). CXCL5 is a chemokine endowed with angiogenetic and pro-metastatic properties, frequently overexpressed in human cancers where it is considered as a prognostic biomarker [42]. Mice treated with guadecitabine/ICBs and guadecitabine/α-CTLA-4 showed a significant decrease of this chemokine in sera in respect to control mice or mice treated with ICBs (Figure 6C). TNF-a and G-CSF serum levels were upregulated in mice treated with guadecitabine/ICBs, though not in a statistically significant manner when compared to control mice (Figure 6D), VEGF, IL-6 and IL-10 were detected at levels similar to those observed for control mice, while IL-1b was significantly lower in mice treated with guadecitabine/ICBs and guadecitabine/α-CTLA-4 (Figure 6D).
The serum levels of CCL2 and CCL4, chemokines known to stimulate leukocyte and MDSC migration, were respectively up and down regulated in mice receiving guadecitabine/ICBs (Figure 6D). CXCL1 levels increased in mice receiving guadecitabine/α-CTLA-4 and guadecitabine/ICBs (Figure 6D).
Comprehensively, these data show that triple therapy can modify not only TME composition, but also T and NK cell responses in lymphoid organs shifting them towards an anti-tumor Th1 cell response.
Effects of Guadecitabine/ICBs on metastases formation and TME
To assess the ability of guadecitabine/ICBs to reduce the development of metastases, we performed an in vivo experiment in which C57black6/J mice, injected intravenously with 4x105 B16F10 cells, received daily intra peritoneal treatment with guadecitabine (1mg/kg) starting from the day after tumor challenge until day +15. ICBs were administered intra peritoneally at days +2, +5, +8, +11, +14 post intravenous challenge. Mice were sacrificed on day 16 and lungs were analyzed for tumor nodules formation (Figure 7A). Lungs were formalin-fixed and paraffin-embedded (FFPE), mounted on microscope slides and used to measure the numbers of micro metastases, the diameters of tumor nodules and the lung area occupied by the tumors. As shown in Figure 7B the numbers of lung micro metastases were significantly reduced in mice receiving guadecitabine/ICBs compared to control or to mice receiving ICBs alone. The mean maximum diameters of tumor nodules as well as the relative and absolute lung area occupied by the tumors were significantly lower in lungs of guadecitabine/ICBs treated animals, compared to any other treatment (Figure 7B and D). Guadecitabine containing treatments significantly reduced the number of mice presenting extra-lung metastases, in particular metastases in the peritoneum and the spleen, in respect to control mice. Mice treated with guadecitabine/ICBs also showed a reduction of extra-lung metastases over those treated with ICBs alone (Figure 7C).
Lungs from control mice, and from mice treated with guadecitabine, ICBs or guadecitabine/ICBs were dissociated into single cell suspensions and studied by flow cytometry to detect treatment induced modifications of TME composition. ICBs treatment increased CD3+ T cell percentages, mainly CD8+ T cell subset, while guadecitabine and guadecitabine/ICBs treatments did not significantly impact on these cell populations (Supplementary Figure 4). Similarly to what observed in the subcutaneous tumors, guadecitabine/ICBs significantly reduced total MDSC population frequency in lung metastatic lesions, an effect likely due to the synergic activity of the two treatments since not replicated by guadecitabine or ICBs alone. In particular, M-MDSC were completely depleted by guadecitabine and gudecitabine/ICBs (Figure 8A).
Multiplex Immunofluorescence (mIF) allows the simultaneous visualization and quantification of several antigens on single formalin-fixed paraffin-embedded (FFPE) tissue sections, maintaining tissue architecture and morphology [43]. Applying mIF on lung tumor tissue slides we could confirm the significant reduction of MDSC. Also, we observed that very few CD8+ T cells were in proximity (within a radius of 30mm) to MDSC and that the percentage of these CD8+ T cells was significantly different between guadecitabine and guadecitabine/ICBs, the latter showing the lowest percentage. The mean distance between CD8+ T cells and MDSC was higher in guadecitabine and guadecitabine/ICBs groups of treatment compared to control or ICBs (Figure 8B). Reduction of MDSC in guadecitabine/ICBs group was associated with an increase of systemic concentrations of cytokines involved in MDSC and myeloid cell generation such as TNF-a and G-CSF. Guadecitabine treatment increased GM-CSF, IL-10 and IL-1b serum levels compared to control mice or mice treated with ICBs alone. However, IL-10 and IL-1b levels were lower in mice treated with guadecitabine/ICBs in respect to guadecitabine. CCL3, CCL4 and CXCL1 (chemokines involved in leukocyte trafficking, including that of DC and MDSC) were significantly increased upon guadecitabine treatment, while in mice treated with guadecitabine/ICBs CCL3 and CCL4 serum levels were comparable to those of controls (Figure 8C).
Immune regulatory CD4+FoxP3+ Treg cells were identified by mIF and their relative prevalence in the lung TME was not significantly affected by any treatment. It is of note that significantly higher percentages of CD4+FoxP3+ Treg were found in close proximity to CD8+T cells in lungs from mice receiving ICBs, compared to those receiving guadecitabine (Figure 9A), suggesting a possible suppressive effect on CD8+T cells in the ICBs group of treatment. CD8+CD28-CD39+ T cells that have been reported to be immune regulatory and exhausted lymphocytes [44] were significantly decreased upon treatment with guadecitabine/ICBs (Figure 9B), while total CD8+ and CD4+FoxP3- T cells were increased (Figure 9C). Th1 cytokines (IL-2, IFN-g, TNF-a) and IL-17, but not Th2 cytokines (IL-4, IL-5) were upregulated in serum samples from mice treated with guadecitabine and its combination with ICBs (Figure 9D, E and F), as well as IFN-g dependent anti-angiogenic chemokines MIG and IP10 (Figure 9G). A significative decrease in angiogenesis, determined by a reduction of the area occupied by CD31+ cells in the tumor/peritumor areas, was observed in mice treated with guadecitabine/ICBs compared to guadecitabine or ICBs alone (Figure 9H). The serum levels of angiogenic chemokines CXCL5 and LIF were significantly diminished by treatment with guadecitabine/ICBs, compared to control and ICBs alone (Figure 9G). Notably, besides its angiogenic properties, LIF is known to regulate CD206 (a marker for TAM-M2 cell polarization) in TAMs and to prevent CD8+ T cell tumor infiltration, compromising responses to anti-PD1 therapy [45].
mIF analysis also revealed that among F4/80+ macrophages those expressing CD206 were significantly decreased in mice receiving guadecitabine/ICBs or ICBs alone, and that significantly fewer F4/80+CD206+ M2 cells were close to CD8+T cells in a radial distance of 30mm in lung metastases from mice treated with guadecitabine/ICBs, compared to those treated with ICBs alone. Accordingly, lung tumors from mice treated with triple therapy showed higher percentages of CD8+ T cells close to F4/80+CD206- M1-type macrophages, compared to lung nodules from control mice. Finally, mIF analysis indicated that the percentages of CD206 expressing cells close to MDSC were higher in lung metastases of control mice as well as in mice treated with ICBs, than in mice treated with guadecitabine containing combinations (Figure 10 and Supplementary Figure 5A, B).
Total DC percentages in lung TME were affected by guadecitabine containing treatments resulting remarkably decreased (Figure 11A). However, myeloid- (CD11c+IAb+CD11b+) and lymphoid- (Conventional, CD11c+IAb+CD11b-) DC subsets [46] were regulated by guadecitabine in opposite ways: M-DC were significantly reduced, while L-DC were increased, comparing to control mice (Figure 11A). Guadecitabine, but not ICBs, increased the percentages of CD103+CD11b- and CD103+CD8a+ conventional DC populations (Figure 11B). Interestingly, these cells are highly specialized in priming CD8+ T cells independently from their cross-presentation potential, and produce MIG and IP10 chemokines able to attract T and NK cells [47–50]. Serum levels of IL-12 were not affected by any treatment, while IL-15 levels were increased in serum from mice receiving guadecitabine/ICBs (Figure 11C).
Guadecitabine determines a significant reduction of DNA-methylation in experimental tumors
DNA samples isolated from experimental tumors from mice treated with ICBs alone or triple therapy were analyzed by hybridization to Infinium Mouse Methylation BeadChip arrays and analyzed for all methylation sites (“tiling”) or signals derived from CpG-islands, genomic regions containing genes and promoter regions of protein coding genes. Principal component analysis (PCA) of samples exposed to guadecitabine as opposed to those not exposed showed that these two sample types are clearly distinct, irrespective of additional treatments with ICBs (Figure 12 A).
DNA-methylation showed the typical bimodal distribution of high and low methylation with only a minor population of sites of intermediate methylation, especially so for CpG-islands and promoters. Partial pharmacological inhibition of DNA Methyltransferase 1 (DNMT1) in the tumors of mice treated with the combination of guadecitabine and ICBs as compared to those treated with ICBs alone, determined limited effects on highly methylated sites (sites methylated in most or all cells). The drug mainly affects sites of intermediate methylation. As expected, the addition of guadecitabine to the two ICBs determined a reduction of methylated sites, in particular for promoters and genes and less so for CpG-islands (Figure 12 B).
PCA (Figure 12A) showed clear differences between samples obtained from treated with and without guadecitabine. In consistence with PCA, class comparison analysis of these two groups revealed that they form two main clusters of distinct methylation patterns (Figure 13).
The analysis of the DNA methylation data, using the most variable 1% of probes, identified 103 probes (mean difference < 0.16; p <0.01), classifying 334 genes for different transcript/isoforms differentially methylated between tumors arising from guadecitabine-treated mice contrasted with guadecitabine-untreated ones (Figure 13 and Supplementary Table 1). The pathway analysis of these differentially methylated genes highlighted significant immune system related biological networks, among these T cell development, differentiation, and antigen presentation (Figure 14).