ALBERT

Description: Graft-versus-host disease (GVHD) remains a major cause of morbidity and mortality after allogeneic hematopoietic stem cell transplantation (allo-HSCT). GVHD involves complex interactions of immune cells, induction of host-reactive donor effector T cells, and donor T cell-mediated injury to normal tissues. Epigenetic changes have been implicated in T cell-mediated GVHD. We previously described that genetic deletion of Ezh2, which catalyzes trimethylation of histone H3 at lysine 27 (H3K27me3), reduced GVHD in mice but preserved graft-versus-leukemia (GVL) responses. Several selective inhibitors of Ezh2 have been recently discovered (e.g. GSK126, UNC1999 and EPZ6438), which specifically reduce the levels of H3K27me3 but not EZH2 protein. Unexpectedly, our preliminary studies showed that administration of GSK126 failed to prevent GVHD in mice. This stands in contrast to our findings that genetic deletion of T cell Ezh2 leads to GVHD inhibition, and suggest that Ezh2 may regulate GVHD through a mechanism independent of H3K27me3. Identifying an optimal method to target T cell Ezh2 for controlling GVHD remains an unmet need. Using experimental mouse models, we demonstrate that functional heat shock protein (Hsp)90 is critical for maintaining Ezh2 protein stability and function in activated T cells. Pharmacological inhibition of Hsp90 destablizes Ezh2 protein in alloreactive T cells, reduces GVHD but preserves GVL effects in mice. To determinethe molecule(s) that is critical for maintaining Ezh2 protein stablility in T cells, we performed mass spectrum (MS) analysis and identified 25 Ezh2-interacting proteins that showed higher intensities than others in T cell receptor (TCR)-activated CD8+ T cells. Among them, we found a group of proteins associated with protein folding and degradation, including Hsp90. Hsp90 is a molecular chaperone required for the stability and function of several key signaling intermediates (e.g., AKT, Raf1 and ERK1/2). Using reciprocal co-immunoprecipitation assay, we confirmed that Ezh2 and Hsp90 directly interacted with each other in TCR-activated CD8+ T cells. Pharmacological inhibition of Hsp90 using its specific inhibitor AUY922, which is currently in phase II clinical trials for cancer therapy, effectively reduced Ezh2 protein without decreasing H3K27me3 24 hours after treatment. This effect was accompanied by decreased proliferation and survival of TCR-activated T cells in vitro. Retroviral overexpression of Ezh2 in T cells markedly improved their proliferation in the presence of AUY922, suggesting that reducing Ezh2 by Hsp90 inhibition is an important mechanism that reduces proliferation and survival of activated CD8+ T cells. Building on these observations, we examined the impact of inhibiting Hsp90 on GVHD by administering AUY922 to B6 mice receiving MHC-identical minor histocompatibility antigen-mismatched C3H.SW mouse CD8+ T cells and T cell-depleted bone marrow (BM). While about 80% of control B6 recipients died from severe GVHD, 80% of AUY922-treated B6 recipients survived without clinical signs of severe GVHD by 84 days after transplantation. In vivo AUY922 administration reduced the survival and expansion of alloreactive T cells, and decreased the fequency of alloreactive T effector cells producing IFN-g and TNF-a. To rule out the model-specific effect of AUY922, we used a haplo-identical B6 into BDF1 mouse model of GVHD. Using CFSE-labeled donor T cells, we first validated that in vivo administration of AUY922 to unirradiated BDF1 mice receiving parent B6 T cells selectively reduced the expansion of alloantigen-reactive donor T cells, but did not impair the expansion and survival of donor T cells that did not respond to alloantigens. In lethally irradiated BDF1 mice receiving B6 T cells and BM, AUY922 administration reduces lethal GVHD, with approximately 50% of them surviving long-time. Importantly, AUY922 treatment preserved GVL activity of donor T cells, leading to significantly improved survival of BDF1 recipients challenged with A20 leukemic cells (Fig.1). Taken together, our findings identified a previously unrecognized molecular mechanism by which Ezh2 and Hsp90 are integrated to regulate alloreactive T cell responses and GVHD. Targeting the Ezh2-Hsp90 complex using AUY922 represents a novel and clinically relevant approach to reduce GVHD while preserving GVL effects, thereby improving the efficacy of allo-HSCT. Disclosures No relevant conflicts of interest to declare.

Description: Key Points Ezh2 requires Hsp90 to maintain Ezh2 protein stability and function in alloreactive T cells. Pharmacological inhibition of Hsp90 destabilizes Ezh2 protein in alloreactive T cells and reduces GVHD but preserves graft-versus-leukemia effects.

Description: Plasmacytoid dendritic cells (pDCs) derived either from adoptive transfer from the donor graft or stem cell reconstitution can attenuate and prevent graft-versus-host disease (GVHD) in both pre-clinical and clinical settings. However, the reconstitution of donor pDCs is severely impaired during GVHD via an unknown mechanism. Here we demonstrate that the histone methyltransferase Dot1l, which specifically catalyzes methylation of histone H3 at lysine 79 (H3K79me), is critical for regulating the commitment and differentiation of pDCs from hematopoietic stem cells (HSCs) and we observed its function was severely impaired in GVHD mice. We have previously demonstrated that Flt3L-induced DCs can program allogeneic T cells to reduce their GVHD toxicity (Blood 2016). Using this platform, we explored candidate histone methyltransferase(s) that affected pDC development. We found the inhibitor specific to Dot1l dramatically decreased the frequency and number of pDCs in cultures compared to other chemical probes that inhibit Ezh2, MLL1, G9a and Jmjd3, respectively. Under steady-state condition, pDCs develop from HSCs through successive steps of lineage commitment and differentiation: multiple potent progenitors (MPP) → macrophage and DC progenitors (MDP) → common DC progenitors (CDP). Upon culturing in the presence of Flt3L+SCF, MPP produced 2-fold and 5-fold more pDCs than MDP and HSCs, respectively, over 6 days' incubation. For this reason, we focused on evaluating the effect of Dot1l deletion on pDC development from MPP and CDP, which represent the early and later stage of pDC progenitors, respectively. To examine the specific role of Dot1l in DC development, we crossed Dot1l conditional knock mice (Dot1lf/f) to ER-Cre B6 mice to generate ER-Cre.Dot1lf/f B6 mice. We administered tamoxifen to ER-Cre.Dot1lf/f B6 mice at day 0 and day +1 to delete Dot1l, highly purified MPP and CDP, and cultured them in the presence of Flt3L+SCF. Deletion of Dot1l led to significant decrease of (3- to 5-fold) pDCs from MPP but marginally decreased CDP production of pDCs. These results suggest that Dot1l is required for the initial commitment and differentiation of MPP into pDCs or the expansion of selected pDCs. To test it, we added Dot1l inhibitor SGC to the cultures containing wild-type (WT) mouse-derived MPP, CDP and pDCs. Inhibiting Dot1l with SGC decreased both the frequency (3-fold) and number (5-fold) of pDCs in the MPP culture compared to control, however, it did not affect the generation of pDCs from the cultures of either CDP or pDCs. Thus, Dot1l regulates the commitment and differentiation of pDCs during the MPP stage. To understand the molecular mechanism by which Dot1l regulates pDC commitment and differentiation, we examined the expression of transcription factor Tcf4, which promotes pDC development, and Id2, which antagonizes Tcf4 effects on pDCs. Dot1l inhibition led to 2-fold reduction of Tcf4 expression and 5-fold increase of Id2 in Flt3L-induced bone marrow cells. ChIP assays confirmed the presence of high amount of Dot1l-catalyzed H3K79me2 at the promoter regions of Id2 and Tcf4 loci. Retroviral introduction of Tcf4 into hematopoietic progenitors lacking Dot1l restored their capacity to produce pDCs in cultures. These findings clearly establish that Tcf4 is the down-stream effector of Dot1l to control pDC development. Building on these observations, we finally examined whether GVHD may impair the reconstitution of donor pDCs through a mechanism of reducing Dot1l expression and function. Using the C57BL/6 (B6) into Balb/c mouse GVHD model, we confirmed the loss of donor pDCs in GVHD recipients at day 14 and day 21 after transplantation and, observed that these GVHD mice had 3- to 12-fold fewer MPP and CDP compared to normal donor mice. Additionally, both HSCs and MPP derived from GVHD mice showed significant reduction of H3K79me2 compared to their normal counterparts. Finally, adoptive transfer of donor BM-derived pDCs attenuated the incidence and severity of GVHD in Balb/c mice receiving B6 T cells. Collectively, these data suggest that GVHD-mediated inflammation profoundly decreases Dot1l function in engrafted donor DC progenitors, leading to damaged reconstitution of donor pDCs. Novel strategies that target Dot1l and its-regulated transcriptional programs may improve the reconstitution of donor pDCs to inhibit GVHD while also promoting anti-leukemia activity in recipients undergoing allo-HSCT. Disclosures No relevant conflicts of interest to declare.

Description: Graft-versus-host disease (GVHD) remains a major barrier for the success of allogeneic hematopoietic stem cell transplantation (allo-HSCT). We have identified the central role of the histone methyltransferase Ezh2 in regulating allogeneic T-cell expansion, differentiation and function. Conditional loss of Ezh2 in donor T cells inhibits GVHD in mice due to the inability of alloreactive T cells to persist. However, the molecular mechanism by which Ezh2 deficiency causes alloreactive T cell death remains unknown. Here we demonstrate that genetic deletion of Stromal Interaction Molecule (Stim) 1, a dynamic endoplasmic reticulum Ca2+ sensor and regulator of Ca2+ signaling, rescues antigen-activated Ezh2-null (Ezh2-/-) T cells, leading to restored persistence of alloreactive effector T cells in mice and severe GVHD. Using RNA-sequencing analysis, we found Ezh2-deficiency led to the upregulation of multiple genes (e.g., Ifng, Prf1, Ccl5, Ccl4, Upp1 and Spp1) known to be regulated by Ca2+ signals through calcineurin (CN), the primary target of the immunosuppressant cyclosporine A (CsA). This reverse correlation between Ezh2 inhibition and CsA-treatment for gene expression suggests that Ezh2 may antagonize Ca2+ signaling in activated T cells. Calcium signaling assays revealed higher cytosolic Ca2+ uptake and more frequent Ca2+ oscillations in Ezh2-/- T cells. Moreover, Ezh2-/- T cells exhibited significantly increased polarization of Stim1 and Orai1 in the cellular membrane. These data reveal an unexpected role of Ezh2 as a negative regulator of Ca2+ entry, thereby serving as a 'brake' for Ca2+ signaling. Using the C57BL/6 (B6) into Balb/c mouse GVHD model, we found significantly fewer Ezh2-/- or Stim1-/- IFN-g-secreting effector T cells compared to the WT counterparts on day 8 or 14 post-transplantation. In contrast, deleting Stim1 from Ezh2-/- donor T cells rescued the cells in the spleen and liver, producing even more donor T cells and IFN-g-secreting effector T cells compared to WT T cells and inducing severe GVHD. We further examined the cell autonomous effect of Stim1 deletion on the rescue of Ezh2-/- T cells by mixing WT T cells (B6/SJL, CD45.1) with Ezh2- and/or Stim1- conditional knockout T cells (i.e., Ezh2-/-, Stim1-/- or Ezh2-/- x Stim1-/- B6 T cells (CD45.2)) at a ratio of 1:1 before transferring into the Balb/c mice. While loss of either Ezh2 or Stim1 led to lower frequency of IFN-g+IL-2+ effector T cells, combined deletion of both genes restored the frequency and number of IFN-g+IL-2+ effector T cells to that of WT T cells. Thus, Stim1-mediated Ca2+ signals are crucial for mediating cell death in alloantigen-driven Ezh2-/- effector T cells. To further determine whether the inhibition of CN-NFAT contributes to the rescue, we treated T cell receptor (TCR)-activated Ezh2-/- T cells with CsA or the calcium release-activated channel specific inhibitor BTP2, respectively, in vitro. While BTP2 dramatically improved the survival of IFN-g-producing effector T cells, CsA did not, suggesting the involvement of CN-NFAT-independent pathways. Ca2+ overload is known to impair mitochondrial function and cause massive cell death. As compared to TCR-activated WT T cells, activated Ezh2-/- T cells displayed significantly less ATP, lower mitochondrial membrane potential, enlarged mitochondrial mass, and decreased capacity to upregulate oxidative phosphorylation. Stim1 deletion largely reversed the metabolic defect in Ezh2-/- T cells, indicating the critical role of mitochondrial metabolism in rescuing these T cells. Considered together, our findings identify the remarkable coordination between Ezh2- and Stim1-regulated effector T cell persistence. As such, these investigations may lead to new approaches to inhibit GVHD, with broad implications to defining fundamental mechanisms of T cell differentiation for control of adaptive immunity, such as tumor immunity and autoimmunity. Disclosures Reshef: Incyte: Consultancy; Takeda Pharmaceuticals: Consultancy; Pfizer: Consultancy; Kite Pharma: Consultancy; Atara Biotherapeutics: Consultancy; Bristol-Myers Squibb: Consultancy.

Description: Promoting donor T cell tolerance to host non-hematopoietic tissues remains the ultimate therapeutic goal in allogeneic hematopoietic stem cell transplantation (allo-HSCT). Dendritic cells (DCs) play dual functions in regulating alloimmunity. DCs can elicit alloreactive T cell responses to mediate graft-versus-host disease (GVHD), but are also implicated in reducing GVHD. In patients, the depletion of plasmacytoid DCs (pDCs) from donor BM grafts resulted in GVHD acceleration. On the other hand, acute GVHD causes complete failure of donor pDC reconstitution after allo-HSCT, and low levels of donor pDC correlate with significantly increased GVHD severity. Thus, the impairment of pDC reconstitution by GVHD may be responsible for the dysfunctional immune regulation. Delineation of the mechanism involved may allow therapeutic intervention to reduce GVHD and improve the efficacy of allo-HSCT. In this study, we demonstrate that alloreactive T cells produce GM-CSF to impair reconstitution of donor pDCs by inhibiting Flt3 expression and its-regulated transcription programs in DC progenitor cells. Using murine GVHD model, we confirmed GVHD severely impaired reconstitution of both donor pDCs and conventional DCs (cDCs). Adoptive transfer of donor-type pDCs rather than cDCs prevented the occurrence of severe GVHD in mice, suggesting donor pDC reconstitution is important to restore tolerance of donor T cells against host tissues. Flt3 is required to induce pDC production through a successive differentiation pathway: HSC → multiple potential progenitors (MPP) → common DC progenitors (CDP) → precursor DCs (pre-DCs). GVHD mice produced significantly less MPP, CDP and pre-DCs compared to normal donor mice and allogeneic mice receiving T cell-depleted BM. Ex vivo culture with Flt3 ligand (Flt3L) showed that those MPP and CDP derived from GVHD mice dramatically decreased the capacity to produce pDCs. Thus, GVHD not only causes decreased numbers of MPP and CDP but also their intrinsic defect in producing pDCs. While both MPP and CDP gave rise to similar numbers of pDCs within 3 days of culture with Flt3L, MPP produced 40-fold more pDCs than CDP by day 9 of culture. This indicates the impairment in GVHD MPP may have much more profound long-term impact on pDC reconstitution than that in CDP. Based on surface expression of Flt3, normal MPP contained two subsets: CD135high MPP and CD135mod MPP. CD135high MPP produced 4-fold more pDCs than CD135mod MPP. As compared to CD135mod MPP, CD135high MPP expressed lower levels of Ink4 family genes, which are cyclin-dependent inhibitors restraining cell proliferation and survival, suggesting that CD135high MPP represent earlier stage differentiated progenitors with greater proliferative capacity. Intriguingly, although GVHD mice generated similar amount of CD135mod MPP as did normal mice, they failed to reconstitute highly proliferative CD135high MPP. Thus, the failure of donor pDC reconstitution may largely result from GVHD-mediated inhibition of CD135high MPP. Alloreactive T cells are known to produce high levels of effector molecules, such as IFN-γ, TNF-α, GM-CSF and other cytolytic molecules. We observed that GVHD effector T cells significantly reduced the production of pDCs from Flt3L-induced normal MPP. Blocking GM-CSF using neutralizing antibody but not other effector molecules markedly inhibited this repressive effect of GVHD T cells on pDC production. GM-CSF dose-dependently decreased the expression of Flt3 and its-regulated transcription factors Irf8 and Tcf4, which are important for development of functional pDCs. However, GM-CSF failed to inhibit the conversion of SiglecH+ pre-pDCs into pDCs. These data suggest that alloreactive T cells produce GM-CSF to block pDC reconstitution by targeting DC progenitors (e.g., MPP and CDP). Building on these findings, we established a novel optimized culture system to produce adequate numbers of SiglecH+ pre-pDCs. Adoptive transfer of these pre-pDCs prevented GVHD, leading to significantly improved overall survival of mice undergoing allo-HSCT. Our findings identify for the first time that selective restoration of donor pDCs early after allo-HSCT may represent an effective cellular therapy to prevent GVHD. Further delineation of the molecular pathway(s) involved in GVHD inhibition of DC progenitors may allow the development of novel approaches to circumvent mortality and morbidity associated with GVHD. Disclosures Zheng: Pfizer: Research Funding.

Description: Despite pharmacological prophylaxis using calcineurin (CN) inhibitors (i.e., cyclosporin A and Tacrolimus), graft-versus-host disease (GVHD) remains a major barrier to the success of allogeneic hematopoietic stem cell transplantation (allo-HSCT). Upon antigen stimulation, activated T cell receptor (TCR) signaling triggers rapid Ca2+ influx. This induces both CN-mediated NFAT activation, and increase of mitochondrial Ca2+ content, a major driver of metabolic activity. However, mitochondrial Ca2+ overload triggers opening of the mitochondrial permeability transition pore and cell death. We hypothesize that pharmacologically increasing mitochondrial Ca2+ load may decrease T cell survival capability, thereby reducing the GVH reaction. If this can be accomplished, it may lead to new strategies for inhibition of GVHD, which is conceptually different from the use of CN inhibitors. To test this hypothesis, we employed a high throughput drug screening system with the proliferation and cytotoxicity of TCR-activated human T cells as the readout, and screened the NPL-800 library (https://www.timtec.net) composed of 800 pure natural compounds. With a stringent criterion in consideration of dose-dependent effect, 26 compounds stood out for reducing the count of activated human T cells with a reduction rate of at least 30% at both 1.0uM and 10.0uM. Positive hits included inhibitors of DNA synthesis, the Na-K-ATPase and mitochondrial metabolism. We were particularly interested in artesunate (ART), which is a derivative of artemisinin that has been used for treating malaria in patients. While artemisinin acts by inhibiting sarco/endoplasmic reticulum Ca2+-ATPase (SERCA) in P. falciparum malaria, which causes passive endoplasmic reticulum (ER) Ca2+ depletion and the subsequent cytosolic Ca2+ influx, ART did not inhibit SERCA in T cells. Ex vivo culture assays showed that ART dose-dependently reduced the survival of TCR-activated murine T cells. This inhibitory effect of ART was abrogated by inhibiting Ca2+ influx using BTP2, a potent inhibitor of store-operated Ca2+ channels. Furthermore, treatment of murine CD8 T cells with ART induced significant increases in mitochondrial Ca2+ loading upon TCR activation. These data suggest that inhibition of T cell survival by ART was dependent on TCR activation-induced Ca2+ influx and associated with enhanced mitochondrial Ca2+ uptake. We examined the impact of ART on GVHD in Balb/c mice receiving C57BL/6 (B6) mouse T cell-depleted bone marrow (TCD-BM) and CD4+ T cells. Intraperitoneal injection of ART (10 mg/kg, every other day) from day 1 to day 28 after transplantation reduced clinical signs of GVHD in these recipients and significantly improved their overall survival. Similar inhibition effects of ART on GVHD were observed in miHA-mismatched B6 anti-Balb/b and haplo-identical B6 anti-BDF1 mouse models of GVHD. Further investigations showed that in vivo administration of ART caused significant decreases in the number of host-reactive donor T cells in the spleen and liver of Balb/c mice 7 days after transfer of B6 TCD-BM plus CD4+ T cells. ART treatment did not affect the capacity of donor T cells to produce effector cytokines (e.g., IFN-g and TNF-α) in individual cells. Importantly, in vivo administration of ART preserved anti-leukemia activity of donor T cells and did not impair the reconstitution of hematopoiesis and lymphocytes. Collectively, our findings indicate that pharmacologically increasing mitochondrial Ca2+ loading may have significant implications in the development of novel strategies to prevent GVHD and other T cell-mediated inflammatory disorders in a broad context. Since ART therapy has been clinically approved, this work could be immediately translated into patients. Disclosures No relevant conflicts of interest to declare.