ALBERT

Description: Antigen-specific T lymphocytes can recognize and eliminate aberrant cells. Cancer cells halt this process by hijacking a system of immune checkpoints, the programmed cell death 1 (PD-1) and its ligands (PD-L1/2) pathway, which physiologically regulates the quantity and activity of T cells, establishing peripheral T cell tolerance and limiting tissue damage. PD-L1-expressing cancer cells interact with and inhibit PD-1 positive T cells, thus abrogating anti-cancer immunity, which can be restored by checkpoint inhibitors (CPI). Improved understanding of the regulation of PD-L1 expression will shed further light on how cancer cells escape immune surveillance, and it may help in the design of combinatorial therapeutic strategies that expand the activity of CPI. Oncogenes (e.g., MYC, STAT3, HIF1 and NF-KB) have been shown to directly induce PD-L1 transcription. In addition, pro-inflammatory cytokines, notably IFN-γ, via the JAK/STAT pathway, also increase PD-L1 expression, an intuitive counteracting regulatory axis that prevents unchecked inflammation and auto-immunity. The second messenger cyclic-AMP (cAMP) is a classical mediator of anti-inflammatory and immunosuppressive inputs. However, its putative role in PD-L1 regulation is unknown. Addressing this knowledge gap is especially relevant because this signaling node can be modulated with a class of FDA-approved agents, the phosphodiesterase 4 (PDE4) inhibitors. We have recently reviewed the pleiotropic roles that cAMP/PDE4 plays in diffuse large B-cell lymphoma (DLBCL) biology (BloodPMID: 27756749). Thus, to examine if cAMP modulates PD-L1 expression, we first used DLBCL cell lines (n=10). Raising the levels of intracellular cAMP readily induced PD-L1 expression (measured by WB and FACS) in ABC-DLBCLs but not in GCB-DLBCLs. This cAMP-mediated induction of PD-L1 occurred also at RNA level; however, using reporter assays we found that the canonical cAMP-PKA-CREB pathway does not directly activate the PD-L1 promoter. The immune modulatory activity of cAMP is mediated, at least in part, by transcriptional activation/secretion of cytokines. Thus, we considered that cAMP induction of PD-L1 in DLBCL may be driven by an autocrine loop. In agreement with this idea, cAMP promoted JAK/STAT activation and culturing DLBCL cell lines in conditioned media (CM) from cAMP-high models induced PD-L1 expression. These assays pointed to secreted factor(s) as intermediaries in the cAMP/PD-L1 axis. Therefore, we screened a panel of 105 cytokines to identify those secreted by DLBCL cell lines following cAMP up-modulation - in most models, we detected a significant cAMP-driven increase in IL-6, IL-8, IL-10 and IL-1α secretion. For validation, we focused on IL-10 because this was the most commonly cAMP-induced cytokine across the DLBCL models. We found that recombinant IL-10 induced PD-L1, albeit this induction was significantly less marked than that observed following an increase in intra-cellular cAMP. Concordantly, antibody-based blocking of the IL-10 signals, and pharmacologically inhibiting the JAK/STAT pathway, only partially abrogated the cAMP-mediated induction of PD-L1. We concluded that IL-10 and JAK/STAT signals relay part, but not all, of the cAMP effects on PD-L1 expression in DLBCL. Next, we utilized the Pde4b null mouse model to examine if these observations were present in an organismal level and in non-immortalized immune cells. In these assays, spleens of Pde4b WT, +/- and -/- mice (8-16 weeks old, male and female, n=8) were collected and analyzed by WB and FACS. Spleen cells from Pde4b deficient mice had markedly higher expression of PD-L1 (WB). By FACS, we found that the increase in PDL1 expression in Pde4b null mice derived from T cells, B cells, but from the smaller non-B/T cell population (CD19/CD3 negative). Finally, we found that the PDE4 inhibitor roflumilast used as a single agent in vitro robustly induced PD-L1 expression in DLBCL cell lines. In summary, we identified cAMP as an "actionable" novel regulator of PD-L1 expression in normal and malignant immune cells. Mechanistically, cAMP drives an autocrine loop enacted by cytokines and transduced in part by JAK/STAT. This finding supports the clinical testing of roflumilast to induce PD-L1 expression, a strategy that may improve the activity of checkpoint inhibitors in DLBCL and related tumor types. Disclosures No relevant conflicts of interest to declare.

Description: Mitochondria play a role in epigenetic remodeling by generating intermediate metabolites (αKG, 2-HG, succinate, etc.) and modulating the activity of enzymes that control DNA, RNA and histone demethylation. The mitochondrial dehydrogenases D2HGDH and L2HGDH regulate 2-HG/αKG homeostasis by catalyzing the oxidation of D2-HG and L2-HG, respectively, into αKG. In humans, loss of either enzyme causes neuro-metabolic syndromes and early mortality, pointing to their essential role in physiology. Yet, the regulation of D2HGDH and L2HGDH remains to be characterized. To address this issue, we mapped the D2HGDH and L2HGDH promoter regions. Using reporter and ChIP assays, lymphoma cell lines and a mouse model, we discovered that MYC binds to E-boxes in these promoters and directly induce D2HGDH and L2HGDH transcription. We created CRISPR KOs of D2HGDH or L2HGDH in MYC-inducible P493-6 cells, and quantified αKG, D2-HG and L2-HG by LC/MS. We found that turning MYC expression ON led to significantly higher accumulation of αKG (and decrease in 2-HG) in cells expressing D2/L2HGDH than in the KO models, confirming that MYC's transcriptional activation of D2/L2HGDH directly influences 2-HG/αKG balance. Then, we tested whether the metabolic output of the MYC-D2/L2HGDH axis influenced the activity of the αKG-dependent TET enzymes, and of the RNA demethylases FTO and ALKBH5. In P493-6 cells, B cells from Eμ-Myc mice and DLBCL cell lines, MYC expression led to a significant increase in 5hmC, decrease in 5mC DNA, and reduction in m6A RNA marks. These MYC-induced changes were blunted in D2HGDH or L2HGDH KO cells. In addition, the effects of MYC on 5hmC/5mC and m6A were blocked in cells exposed to cell-permeable D2-HG or L2-HG, while synthetic αKG rescued D2HGDH and L2HGDH KO cells. Next, we quantified TET and FTO/ALKBH5 activity in the nuclear lysates of our models. In all instances, MYC significantly elevated TET/FTO/ALKBH5 activity in D2HGDH/L2HGDH WT cells, an outcome that was significantly dampened in the KO models. These data suggested that the MYC-D2/L2HGDH axis increases the activity of these enzymes by generating αKG. To examine whether additional layers of control were also operational, we quantified the expression of TET1-3, FTO and ALKBH5, and of members of the RNA methyltransferase complex, METLL3, METLL14 and WTAP, and confirmed that their global expression were unmodified by MYC or other genetic and metabolic perturbations. Strikingly, however, when examining subcellular localization, we found that MYC, in a D2/L2HGDH dependent manner, promoted the nuclear localization of TET1-3, FTO and ALKBH5 (but not METTL3/14). Remarkably, these effects were readily recapitulated by αKG, whereas 2-HG retained these proteins in the cytoplasm, effectively reducing their ability to chemically modify DNA and RNA. Lastly, to test if the MYC-driven increase of TET activity was also present in primary tumors, we explored the DLBCL TCGA dataset. We mapped all enhancers that are active in DLBCL and found that enhancer hypomethylation (a putative consequence of TET activity) significantly correlated with MYC expression (r=0.38, p=0.009) or activity (ssGSEA-derived MYC activity score, r=0.46, p=0.0013); this finding remained significant after excluding potentially cofounding TET2-mutant DLBCLs. We next tested if this MYC-driven/TET-mediated enhancer demethylation was functionally relevant by correlating it to the expression of the gene closest to, or overlapping with, the enhancer. A biologically coherent negative correlation between enhancer methylation and target transcript expression was found for multiple genes, including several products previously implicated in lymphoma pathogenesis, including, FOXP1, PIM1, ATF5, KLHL14 and BRD2. In summary, we showed that D2HGDH and L2HGDH are transcriptional targets of MYC, and that the control of αKG levels by the MYC-D2/L2HGDH axis activates TETs and RNA demethylases. We discovered that MYC and intermediate metabolites control the sub-cellular localization of these enzymes, possibly via covalent modifications, thus adding an new layer of complexity to the remodeling of the epigenome and epitranscriptome in cancer. We showed in primary DLBCLs that MYC expression/activity correlates with hypomethylated/active oncogenic enhancers. Thus, we postulate that downstream to MYC TET enzymes may in specific contexts function as oncogenes. Disclosures No relevant conflicts of interest to declare.