Inhibition of LSD1 in Myelodysplastic Syndrome Progenitors Restores Differentiation of CD141^Hi Conventional Dendritic Cells

Background: Immunotherapeutic approaches for myelodysplastic syndrome (MDS) show promise, but progress is limited by our incomplete understanding of the immunologic milieu. In a recent Phase I trial, we found that MDS patients with higher numbers of CD141^Hi conventional dendritic cells (cDCs) were more likely to respond to NY-ESO-1 vaccination. In solid tumor models, the CD141^Hi cDC is critical for initiating anti-tumor immune responses but its impact in myeloid malignancies is unknown. In studies of primary human specimens and mouse models, we tested the hypothesis that MDS patients exhibit decreased quantity and quality of CD141^Hi cDCs due to impaired myeloid differentiation. Methods: Bone marrow (BM) cells were collected from MDS patients (pre-treatment) and age matched healthy donors (HD; defined as absence of hematologic malignancy). We quantified DC populations, stem, progenitor cells and interferon regulatory factor-8 (IRF-8) expression using flow cytometry and RT-qPCR. Histone modifications were assessed by chromatin immunoprecipitation. To assess DC differentiation of progenitors, human CD34+ and mouse c-kit+ cells were expanded and differentiated in vitro. Results: We found fewer CD141^Hi cDCs (p<0.0001), CD1c⁺ cDCs (p<0.005) and plasmacytoid DCs (CD123⁺ pDCs; p<0.005) overall in BM samples from MDS patients (n=71) compared to HD (n=17). We stratified MDS patients based on the relative number of DCs and found that only those patients with highest number of CD141^Hi cDCs had superior survival (p < 0.05). No differences in survival were seen in patients stratified by the CD1c⁺ cDCs (p = 0.96) and CD123⁺ pDCs (p = 0.32) populations. We hypothesized that decreased numbers of CD141^Hi cDCs and adverse survival in MDS patients resulted from impaired differentiation of DC progenitors. We showed that MDS patients (n=19) have fewer monocyte-DC progenitors (MDP) and common DC progenitors (CDP) compared to HD (n=11; p<0.01). We then hypothesized that MDS progenitors express lower levels of IRF8, a master regulator of CD141^Hi cDC differentiation. IRF8 expression was significantly lower in CDPs from MDS patients compared to HD (p<0.05). Furthermore, MDS patients with lower levels of IRF8 (n=8) in their MDPs showed a trend towards production of fewer CDPs and significantly fewer CD141^Hi cDCs compared to those with higher levels of IRF8 (n=10; p<0.005). These results suggest that approaches to increase IRF8 expression could enhance CD141^Hi cDC differentiation. We hypothesized that inhibition of lysine-specific histone demethylase 1A (LSD1), which increases IRF8 expression in myeloid leukemia cells, would induce CD141^Hi cDC differentiation. Pharmacologic inhibition of LSD1 increased IRF8 expression (both mRNA and protein) in KG-1 cells, a model of human CD34⁺ cells, and in HD and MDS CD34⁺ progenitors (p<0.05). LSD1 inhibition in KG-1 cells resulted in increased H3K27 acetylation (3861-fold change) and H3K4 dimethylation (922.4-fold change) compared to PBS (p<0.05) at a region -70 kb to the IRF8 transcriptional start site, a putative regulatory element that demonstrated the highest level of LSD1 binding. These data indicate that LSD1 inhibition alters histone modifications at the IRF8 locus, resulting in increased expression of IRF8. Pharmacologic inhibition of LSD1 in HD CD34⁺ cells increased the number of mature CD141^Hi cDCs in 92% of specimens (n = 12; 3.4 fold-change) compared to PBS. Similarly, LSD1 inhibition in MDS CD34⁺ cells increased the number of CD141^Hi cDCs (n = 12) in 75% of patient specimens (16.8 fold-change) compared to PBS. IRF8 function is conserved between mice and humans. To test whether the effect of LSD1 inhibition on cDC differentiation was dependent on IRF8, we compared the effect of LSD1 inhibition on BM c-kit⁺ cells from Irf8 knock-out mice (Irf8-KO) and littermate controls (WT). LSD1 inhibition in WT c-kit⁺ cells resulted in increased numbers of CD141^Hi cDCs in vitro (p<0.05). By contrast, LSD1 inhibition of Irf8-KO c-kit⁺ cells did not result in differentiation of CD141^Hi cDCs. These data suggest that LSD1 inhibition drives CD141^Hi cDCs differentiation through IRF8. Conclusion: These data reveal a previously unrecognized determinant of the immune microenvironment in MDS. The opportunity for epigenetic regulation of CD141^Hi cDC differentiation in MDS offers an opportunity for intervention and a potential adjunct to immunotherapy for patients.