Towards Safer Allogeneic HSCT in Autoimmunity: Proof of Concept in Diabetes-Prone Mice

R.S.K and B.N.L contributed equally to this study. The induction of tolerance towards pancreas autoantigens is a promising approach in the treatment of type 1 diabetes mellitus (T1DM). This goal can be partially attained by immuno-ablation followed by autologous HSCT which mitigates the immune reaction leading to diabetes, at least temporarily, if performed soon after diagnosis. However, most T1DM patients transplanted with autologous HSCT eventually relapse (E Snarski et al., Bone marrow Transplantation; 2016; 51, 398-402). Thus, developing methodologies for allogeneic HSCT, to provide a durable non-autoimmune TCR repertoire, could be a promising treatment approach provided that such protocols can safely achieve donor type chimerism. Accordingly, transplantation of T cell-depleted allogeneic HSCT (TD-HSCT) under mild conditioning, associated with minimal toxicity and reduced risk of GVHD, offers an attractive therapeutic option. However, overcoming rejection after reduced conditioning in T1DM represents a major challenge. We previously demonstrated in wild type mice that rejection of TD-HSCT can be prevented using donor-derived veto cells. Here, we show proof of concept of the safety and efficacy of veto cell mediated non-myeloablative mismatched allogeneic TD-HSCT, in the established NOD mouse T1DM model. Veto activity, first defined by Miller (Miller, R. G; Nature; 1980; 287; 544-54), is based on the ability of specific cell populations to attack host CTL-precursors (CTLp) that are directed against the antigens presented by the veto cells themselves. This response spares cells that are not targeted against the veto cells, including those recognizing pathogens. Among different veto cell populations described in the literature, central memory CD8 T cells exhibit the most robust veto activity upon transplantation, but are also endowed with marked GvH activity. We overcame GvH reactivity by expanding naïve or memory CD8 T cells against 3rd party MHC or viral antigens, under culture conditions favoring expression of the central memory phenotype. Such anti-3rd party central memory CD8 T cells (Tcm) are endowed with marked veto activity, while effectively depleted of GVH reactivity in fully mis-matched recipients (Reviewed in Reisner Y, Or-Geva N. Semin. Hematol. 2019; 56(3): 173-182). In this study, Tcm veto cells were generated from splenocytes obtained from Balb/c donors (H2d) cultured against irradiated third-party splenocytes (FVB; H2q), under cytokine deprivation. The selective expansion of CD8 mouse T cells against 3rd party stimulators under these conditions leads to selective 'death by neglect' of bystander anti-host T cell clones that could mediate GvHD; such cells are further diluted out by subsequent expansion of anti-3rd party T clones during further culture in the presence of IL15. In addition to causing selective loss of GvH reactive T cells, these culture conditions induce a central memory phenotype shown to be crucial for robust veto activity in vivo. To evaluate the safety and efficacy of such a transplant for diabetes therapy, 8 week old NOD mice (before diabetes onset) were treated with 4.5 Gy TBI conditioning at day -1, anti-3rd party veto Tcm and megadose nude bone marrow (NuBM) on day 0, and Rapamycin treatment from day -1 to day +4 (Scheme 1). Controls were untreated or received conditioning with no transplant. In four independent experiments (n= 35 NOD mice in the transplanted group), high chimerism from 83.5% to 99.6 % was found in all transplanted mice at 6 months post-transplant. Notably, with a follow-up of 200 days, 72.4% mice in the untreated group, and 96% in the conditioned group died of diabetes, while only 8.5% diabetes-associated mortality (Fig.1A) and no GvHD or other transplant-related mortality was observed in the transplanted mice (Fig.1B). Our results demonstrate a proof of concept for the safety and efficacy of non-myeloablative allogeneic TD-HSCT in type 1 diabetes. A clinical protocol testing the safety and efficacy of anti-3rd party veto cells in the context of low toxicity non-myeloablative TD-HSCT in hematological malignancies is commencing at MD Anderson Cancer Center. If successful, our results support further extension of this platform to autoimmune diabetes and other autoimmune diseases. Disclosures Lustig: Yeda Ltd.: Patents & Royalties. Reisner:Cell Source, Inc.: Consultancy, Equity Ownership, Membership on an entity's Board of Directors or advisory committees, Patents & Royalties.

Proof of Concept for Repair of Sickle Cell Disease By Non-Myeloablative MHC Disparate T Cell Depleted HSCT Combined with Donor-Derived Veto Cells

Although life extending medical treatments are available for sickle cell disease (SCD), allogenic hematopoietic stem cell transplantation (HSCT) is considered a treatment of choice (Ozdogu et al., Bone Marrow Transplant 2018; 53(7): 880-890). However, HSCT is associated with several limitations, including conditioning-related toxicity and graft-versus-host disease (GvHD), especially when using MHC disparate transplants. Thus, the development of safe transplantation protocols for MHC disparate HSCT in sickle cell disease is vital. While the risk of GvHD and conditioning toxicity can be effectively reduced by the use of T-cell-depleted HSCT (TD-HSCT) under reduced intensity conditioning (RIC), rejection of TD-HSCT remains a challenge. We previously demonstrated in wild type mice that this barrier can be overcome using donor-derived veto cells. Here, we demonstrate the safety and efficacy of this approach in a well-defined murine model for sickle cells disease. Veto activity is based on the ability of certain cells to attack host CTL-precursors (CTLp) which are directed against antigens expressed on the veto cells themselves, sparing cells that are not targeted against the veto cells including T cells needed for defense against pathogens. Central memory CD8 T cells exhibit the most robust veto activity upon transplantation; however, these cells are also endowed with marked GvH activity. We overcame this issue by expanding naïve or memory CD8 T cells against 3rd party MHC or viral antigens, respectively, under culture conditions favoring expression of central memory phenotype. Such anti-3rd party central memoryCD8 T cells (Tcm), which are endowed with marked veto activity, also exhibit reduced risk for GvHD in fully mis-matched recipients (Reviewed in Reisner Y, Or-Geva N. Semin Hematol. 2019; 56(3): 173-182.) To generate Tcm veto cells, splenocytes obtained from Balb/c donors (H2d) were cultured against irradiated third-party splenocytes (FVB; H2q) under cytokine deprivation. The selective expansion of CD8 mouse T cells against 3rd party stimulators leads to selective 'death by neglect' of bystander anti-host T cell clones potentialy mediating GvHD, and these are further diluted out by subsequent expansion of anti-3rd party T clones during continued culture in the presence of IL-15. Apart from selective loss of GvH reactive T cells, these culture conditions induce a central memory phenotype shown to be important for attaining robust veto activity in vivo (Ophir et al., Blood. 2013; 121(7): 1220-8). We first calibrated the optimal irradiation dose for sickle mice (Berkeley model, H2b), comparing 4.5 Gy versus 5 Gy TBI in a conditioning protocol also including short term rapmycin treatment (Scheme 1). Higher levels of engraftment and chimerism were found in the group receiving 5Gy TBI (Fig. 1). All mice of both treatment groups survived >140 days with no evidence of GvHD. To further evaluate this treatment approach, 8 week old sickle mice (N=7) were given bone marrow transplants using the protocol described in Scheme 1, including conditioning with 5 Gy TBI (day -1), rapamycin treatment (day -1 to day 4), and transplantation of NuBM (day 0) plus veto cells (day 7; Scheme 1). Notably, at 44 days post-transplant, 6 out of 7 mice receiving NuBM + TCM + Rapa (85.7%) showed donor chimerism in the peripheral blood, ranging between 77-94% (Fig.2A-B), while no chimerism was detected in mice receiving conditioning alone, or conditioning and transplantation with only NuBM or only veto cells . All mice in all groups survived (N=26) , and no GvHD was detected with a follow up of 77 days, even in the transplanted group which exhibited high donor-derived chimerism. Furthermore, reversal of sickle disease symptoms was observed, including reticulocyte levels (p=0.001;Fig.2C) and expression of wild type hemoglobin ( Fig.2D) in all engrafted mice. Our results offer a proof of concept for the treatment of sickle disease by MHC disparate non-myeloablative T cell depleted HSCT in conjunction with anti-3rd party central memory veto CD8 T cells. A clinical trial testing the safety and efficacy of anti-3rd party veto cells in the context of low toxicity non-myeloablative TD-HSCT in hematological malignancies is currently ongoing at MD Anderson Cancer Center. If successful, our present results support further evaluation of this platform for sickle cell disease. Disclosures Lustig: Yeda Ltd.: Patents & Royalties. Reisner:Cell Source, Inc.: Consultancy, Equity Ownership, Membership on an entity's Board of Directors or advisory committees, Patents & Royalties.

Next Generation Veto Cells for Non-Myeloablative Haploidentical HSCT: Combining Anti-Viral and Graft Facilitating Activity

Transplantation of T cell-depleted BM (TDBM) under mild conditioning, associated with minimal toxicity and reduced risk of GVHD, offers an attractive therapeutic option for patients with nonmalignant hematologic disorders or can be used to induce immune tolerance to subsequent organ transplantation. However, overcoming TDBM rejection after reduced conditioning remains a challenge. We previously demonstrated that this barrier can be overcome using donor-derived naïve CD8 T cells if they are expanded against third-party alloantigens under culture conditions inducing a central memory phenotype. Thus, such anti-3rd party CD8 T cells (Tcm) exhibit marked veto activity and reduced risk for GVHD due to the lower frequency of anti-host clones achieved during the culture period. In the present study, we tested the feasibility of generating veto Tcm by stimulation against well-defined peptides including viral antigens. For proof of concept, we used OT1 mice that express a transgenic (Tg) TCR designed to recognize ovalbumin (OVA) residues 257-264 in the context of H2Kb MHC-I. Prior to harvest of OT1 CD8+ T cells, mice were immunized twice with OVA-peptide. Mice were sacrificed 7 to 14 days after immunization, their spleens and lymph nodes removed and crushed, and magnetic bead sorting utilized to isolate the memory cells (CD8+CD44+). The resulting population was subjected to third-party stimulation by co-culture with irradiated splenocytes generated from spleens of OVA-expressing mice, under cytokine deprivation. hIL-15 (10ng/ml) was added to the culture 60 hrs after culture initiation to induce the cells to express a Tcm like phenotype, as previously described for anti-3rd party Tcm generated from WT naïve CD8 T cells (WT Tcm). When tested in vivo, the OT-1 Tcm (H2Kb) were able to enhance engraftment of allogeneic T cell depleted BMT (H2Kb) in sub-lethally irradiated (5.5 Gy TBI) Balb/c recipients (H2Kd), in analogy to the chimerism induced by WT Tcm. This initial successful experiment was then followed by an experiment, which more closely resembled the human setting, in which the CD8+CD44+ cells were isolated from WT C57BL/6 OVA-immunized mice and subsequently introduced to co-culture with irradiated splenocytes generated from spleens of OVA-expressing mice. Results showed that Tcm grown from such a starting population of CD8+CD44+ cells were also able to achieve marked donor chimerism when administered with megadose of T cell depleted (TCD) alloSCT in sublethally irradiated Balb/c recipients (Fig.1A). Notably, infusion of 5x106 Tcm into sublethally irradiated mice (5.5. GY TBI) without BMT, did not cause any GVHD as measured by weight loss, in contrast to CD8+CD44+ T cells used to generate the Tcm (Fig.1B). These results demonstrate that CD8+CD44+ derived from memory CD8 T cells by expansion against cognate peptides exhibit markedly reduced risk for GVHD compared to freshly isolated memory cells, while retaining their veto activity and inducing tolerance. Finally, based on these proof of concept studies, we were able to translate this approach and generate human anti-viral CD8 veto cells with central memory phenotype. Thus, CD8+CD45RO+ memory T cells were selected by depletion of CD4+CD56+CD45RA+ cells from PBMC of normal donors and then co-cultured with donor DCs pulsed with a viral peptide mixture of three prominent viruses (EBV, CMV and Adenovirus). In three large scale experiments using leukapheresis preparations of normal CMV and EBV positive donors, with GMP grade reagents, more than 1x109 Tcm could be attained by the end of 9 days of culture (average expansion of Tcm=13.5±4 fold) with greater than 90% purity of CD45RO+CD62L+CD8+ T cells (Fig 2). The harvested anti-viral Tcm exhibited more than 3 log depletion of alloreactivity, compared to fresh CD8 T cells, as measured by limit dilution analysis of cytotoxic T cell precursors against host type target cells (Fig. 3). In conclusion, our results suggest that potent veto CD8 Tcm can be generated from the memory pool of donors positive for viral reactivity by stimulation against viral antigens. Such veto Tcm could be most attractive for haploidenitcal transplantation of TCDBM, as they can enable engraftment following non-myeloablative conditioning and at the same time provide anti-viral protection. *E.B.L and N.O.G contributed equally Disclosures Bachar Lustig: Yeda LTD: Patents & Royalties. Or Geva:Yeda LTD: Patents & Royalties. Lask:Yeda LTD: Patents & Royalties. Gidron:Yeda LTD: Patents & Royalties. Kagan:Yeda LTD: Patents & Royalties. Reisner:Cell Source LTD: Consultancy, Equity Ownership, Patents & Royalties, Research Funding.

Towards 'Off-the-Shelf ' Universal Chimeric Antigen Receptor (CAR) T Cells: Mouse Anti-3rd Party Central Memory CD8 Veto Cells Prolong Functional Engraftment of Allogeneic Genetically Modified T Cells

We recently demonstrated that mouse anti-3rd party central memory CD8 veto cells (Tcm) enable engraftment of fully mis-matched T cell depleted bone marrow (TDBM) in sublethally irradiated (5.5 Gy TBI) recipients (Eran et al. Blood 2013). Here we found that Tcm generated from (host x donor)F1 or fully allogeneic donors were able to engraft and sustain their presence in such sub-lethally irradiated mice in the absence of TDBM transplants. Limiting dilution analysis revealed that the reactivity of host T cells (HTC) against donor spleen cells was significantly diminished in the chimeric mice treated with Tcm, whereas reactivity against 3rd party was comparable betweenTcm treated and non-treated mice (Fig. 1). This specific immune tolerance indicated that vetoTcm could potentially enable the use of 'off-the-shelf' allogeneic CAR T cells or any other genetically modified T cells from the same donor. To further investigate this possibility, we tested the ability of Tcmveto cells to enable engraftment of CD8 T cells from OT1 mice, expressing a transgenic T cell receptor (TCR) directed againstovalbumin (OVA) residues 257-264 in the context of H2Kb MHC-I. To that end, 1x106 OT1 CD8 T cells (H2b) were infused into Balb/c (H2Dd) recipients sublethally irradiated with 5.25 Gy TBI in the presence or absence of different doses of veto Tcm. As shown in Fig. 2, when tested at 2 months post-transplant, OT1 cells could not be detected in the peripheral blood of untreated recipient mice, while they were readily detectable in mice receiving 2x106 Tcm (0.78±0.36), reaching a higher level (1.18± 0.61) upon infusion of 5x106Tcm. To evaluate the functionality of the engrafted OT1 T cells in the chimeric mice, we developed and calibrated a new murine model, using a melanoma B16-cell line of C57BL background that expresses the ovalbumin peptide and the tdTomatomarker (B16-OVA-tdTomato) (Fig 3A). In this model, we tried to mimic a state of minimal residual disease by injecting a small number of B16-OVAtdTomato melanoma cells (0.25x106) into syngeneic C57BL mice; we then treated the mice with sublethal irradiation (6Gy TBI) to simulate treatment of the tumor, but also to clear out space for the T-cell adoptive transfer. Finally, we injected 1x106 naïve OT-1 cells generated on a (Balb x-OT1-CD45.1+RAG2-) F1 background in the presence or absence of Balb/c veto Tcm (5x106), and followed the development of the tumor. We found thatBalb/c vetoTcm prolonged the engraftment of the transgenic donor-derived F1 OT1 CD8 T cells inBalb/c recipients, thereby enabling significantly improved control of tumor growth when tested at day 20 post-transplant ( p<0.05). In contrast non OVA-expressing tumor cells grew at the same rate in all treatment groups, thereby demonstrating the specificity of the anti-tumor effect (Fig.3B). Previous studies demonstrated that Tcmare depleted of GVH reactivity by virtue of their culture with 3rd party stimulators under cytokine starvation. Our present results demonstrate that suchTcm veto cells can pave the way for functional engraftment of CD8 T cells with genetically modified specificity. Taken together, these results provide a proof of concept for the potential clinical use of such veto cells to enable therapy with 'off-the-shelf ' allogeneic CAR T cells. The relative ease of growing humanTcm in large numbers over a short period of time (9-12 days) suggests that it could be possible to harvest byleukapheresis from a single donor, a sufficient number of cells, to support CAR therapy for more than 50 allogeneic recipients. * N.O.G and R.G equally contributed Disclosures Or Geva: Yeda LTD: Patents & Royalties. Gidron:Yeda LTD: Patents & Royalties. Reisner:Cell Source LTD: Consultancy, Equity Ownership, Patents & Royalties, Research Funding.