Blockade of individual Notch ligands and receptors controls graft-versus-host disease

Abstract

Graft-versus-host disease (GVHD) is the main complication of allogeneic bone marrow transplantation. Current strategies to control GVHD rely on global immunosuppression. These strategies are incompletely effective and decrease the anticancer activity of the allogeneic graft. We previously identified Notch signaling in T cells as a new therapeutic target for preventing GVHD. Notch-deprived T cells showed markedly decreased production of inflammatory cytokines, but normal in vivo proliferation, increased accumulation of regulatory T cells, and preserved anticancer effects. Here, we report that γ-secretase inhibitors can block all Notch signals in alloreactive T cells, but lead to severe on-target intestinal toxicity. Using newly developed humanized antibodies and conditional genetic models, we demonstrate that Notch1/Notch2 receptors and the Notch ligands Delta-like1/4 mediate all the effects of Notch signaling in T cells during GVHD, with dominant roles for Notch1 and Delta-like4. Notch1 inhibition controlled GVHD, but led to treatment-limiting toxicity. In contrast, Delta-like1/4 inhibition blocked GVHD without limiting adverse effects while preserving substantial anticancer activity. Transient blockade in the peritransplant period provided durable protection. These findings open new perspectives for selective and safe targeting of individual Notch pathway components in GVHD and other T cell-mediated human disorders.

Figure 1. Efficient Notch inhibition in alloreactive T cells but severe intestinal toxicity of GSIs after BM transplantation.

Lethally irradiated (9 Gy) BALB/c mice were transplanted with B6 TCD BM (5 × 10⁶ cells) with or without WT or DNMAML (DN) B6 T cells (10 × 10⁶ splenocytes). The GSI DBZ was administered daily as compared with vehicle (i.p., 10 μmol/kg) (28). (A) Cytokine production by donor-derived H2Kb⁺H2Kd^– CD4⁺ spleen T cells at day 5 after allo-BMT. Representative flow cytometry plots show intracellular IFN-γ and IL-2 after anti-CD3/CD28 restimulation. (B) Relative abundance of transcripts for Ifng, Il2, and the Notch target gene Dtx1 in day-5 donor-derived CD4⁺ T cells after anti-CD3/CD28 restimulation. (C) Short survival of DBZ-treated mice after allo-BMT, even upon transplantation of TCD BM only (P < 0.01, WT vehicle vs. WT DBZ; P < 0.0001, TCD vehicle vs. TCD DBZ). In contrast, DNMAML expression in donor T cells led to markedly prolonged survival (P < 0.0001, WT vs. DNMAML vehicle) (n = 14 for vehicle-treated, n = 6 for DBZ-treated groups). (D) H&E sections and quantification of villus atrophy in the ileum of DBZ-treated mice at day 5 after allo-BMT (n = 5–6/group). (E) Markedly decreased BrdU incorporation upon DBZ treatment (n = 3/group). Scale bars: 100 μm. Bar graphs represent mean ± SD. **P < 0.01.

Figure 2. Notch1 and Notch2 control the production of IFN-γ and IL-2 by alloreactive T cells, with dominant effects of Notch1.

WT or DNMAML B6 T cells were transplanted into irradiated BALB/c recipients (9 Gy). Isotype control, anti-Notch1 (anti-N1), anti-Notch2 (anti-N2), or both anti-Notch1/Notch2 antibodies were administered at day 0 and day 3. DNMAML T cells exposed to isotype control antibodies were a positive control for pan-Notch inhibition. (A) Intracellular staining for IFN-γ in donor-derived H2Kb⁺H2Kd^– CD4⁺ spleen T cells after anti-CD3/CD28 restimulation. MFI of the IFN-γ⁺ cells is shown. (B) Intracellular staining for IL-2 under the same conditions. MFI of IL-2⁺ cells is shown. Representative flow cytometry plots are shown. Numbers indicate the percentage of cells in each quadrant. Bar graphs represent mean ± SD (n = 3) from 1 of 3 representative experiments. **P < 0.01.

Figure 3. Notch1 inactivation in T cells is sufficient to protect mice from lethal GVHD.

Irradiated BALB/c mice (9 Gy) were infused with TCD BM and WT, DNMAML, or Notch1^f/f Cd4-Cre⁺ (N1 KO) spleen B6 T cells. Isotype control or anti-Notch2 antibodies (5 mg/kg) were administered i.p. twice weekly. (A) Survival after transplantation and (B) clinical GVHD score showing protection from lethal GVHD in groups transplanted with DNMAML or Notch1 KO T cells, irrespective of Notch2 blockade (P < 0.0001 for TCD BM, DNMAML, Notch1 KO groups vs. WT isotype and WT anti-Notch2) (n = 8/group). Cross indicates death of all mice by the indicated time point. Preserved in vivo proliferation (C) and expansion (D) of DNMAML and Notch1 KO CD4⁺ and CD8⁺ T cells. Donor-derived H2Kb⁺H2Kd^–CFSE-labeled T cells were tracked in the spleen at day 5 after transplantation. DNMAML-GFP was present in the same channel as CFSE fluorescence. (E) Increased expansion of DNMAML and Notch1 KO T cells at days 14 and 21 after transplantation. Graphs show the number of donor-derived H2Kb⁺H2Kd^– CD4⁺ or CD8⁺ T cells in the spleen. (F) Increased percentage and absolute numbers of donor FoxP3⁺ Tregs upon DNMAML expression or Notch1 inactivation. Representative flow cytometry plots are shown, including a sample stained with isotype control antibodies. Bar graphs represent mean ± SD. **P < 0.01; *P < 0.05.

Figure 4. Dominant role of Notch1 in intestinal regeneration after BM transplantation.

BALB/c mice were lethally irradiated (9 Gy) and transplanted with TCD B6 BM (5 × 10⁶ cells). Monoclonal antibodies (5 mg/kg) were administered i.p. twice weekly. (A) H&E staining and anti-BrdU immunohistochemistry of ileum in mice treated with isotype control or combined anti-Notch1/Notch2 antibodies (day 4). BrdU was given i.p. 2 hours before euthanasia. The number of BrdU⁺ crypts was quantified in 6–10 crypts/mouse. Bar graphs represent mean ± SD (n = 3/group). **P < 0.01. (B) Rapid lethality in anti-Notch1/Notch2–treated mice, as seen with GSI treatment (Figure 1). Isotype control, n = 5; anti-Notch1/Notch2, n = 12. (C) H&E and anti-BrdU staining of ileum in mice treated with isotype control (day 5), anti-Notch1 (day 6), or anti-Notch2 antibodies (day 5). Bar graphs represent mean ± SD (n = 3/group). Scale bars: 100 μm. (D) Rapid lethality in anti-Notch1–treated mice, consistent with the major effects of Notch1 blockade on intestinal regeneration as seen in C. This was not the case with anti-Notch2 alone (see Figure 3A). Isotype control, n = 5; anti-Notch1, n = 6 mice/group.

Figure 5. Dll1 and Dll4 Notch ligands control the production of IFN-γ and IL-2 by alloreactive T cells, with dominant effects of Dll4.

WT or DNMAML B6 T cells were transplanted into irradiated BALB/c recipients (9 Gy). Isotype control, anti-Dll1, anti-Dll4, or anti-Dll1/Dll4 antibodies were administered at day 0 and day 3. DNMAML T cells exposed to isotype control antibodies were a positive control for pan-Notch inhibition. (A) Intracellular staining for IFN-γ in donor-derived H2Kb⁺H2Kd^– CD4⁺ spleen T cells after anti-CD3/CD28 restimulation. MFI of IFN-γ⁺ cells is shown. (B) Intracellular staining for IL-2 under the same conditions. MFI of IL-2⁺ cells is shown. Representative flow cytometry plots are shown. Numbers indicate the percentage of cells in each quadrant. Bar graphs represent mean ± SD (n = 3) from 1 of 3 representative experiments. **P < 0.01.

Figure 6. Transient blockade of Dll1 and Dll4 protects mice from lethal GVHD without inducing intestinal toxicity.

(A) BALB/c mice were lethally irradiated (9 Gy) and transplanted with TCD B6 BM (5 × 10⁶ cells). Anti-Dll1 and anti-DllL4 antibodies (5 mg/kg) were administered i.p. on day 0 and day 3. Preserved intestinal architecture and BrdU incorporation (H&E and anti-BrdU staining, day 5) indicating absence of intestinal toxicity. (B and C) Irradiated BALB/c mice (9 Gy) were infused with TCD B6 BM, with or without WT or DNMAML spleen B6 T cells. Isotype control, anti-Dll1, anti-Dll4, or anti-Dll1/Dll4 antibodies were administered twice weekly for 60 days (long course) or 10 days (short course: days 0, 3, 7, and 10). Scale bars: 100 μm. (B) Survival after transplantation and clinical GVHD score demonstrate increasing protection with Dll1, Dll4, and combined Dll1/Dll4 blockade (P = 0.005, WT vs. anti-Dll1; P = 0.0001, WT vs. anti-Dll4; P < 0.0001, WT vs. anti-Dll1/Dll4). Anti-Dll1/Dll4 antibodies provided nearly as much protection as DNMAML T cells, even upon short-term administration (P = 0.21, DNMAML vs. anti-Dll1/Dll4 groups) (n = 8/group). Crosses indicate death of all mice by the indicated time point. (C) Representative photographs illustrating protection from GVHD (see also Supplemental Figure 7).

Figure 7. Preserved in vivo T cell proliferation and increased expansion of Tregs upon transient Dll1/Dll4 blockade.

Irradiated BALB/c mice (9 Gy) were infused with TCD BM, with or without WT or DNMAML spleen T cells. Isotype control or anti-Dll1/DllL4 antibodies were administered transiently (days 0, 3, 7, and 10). (A) Preserved in vivo proliferation as assessed by tracking CFSE-labeled donor-derived T cells and (B) BrdU incorporation on day 5 after allo-BMT. (C) Expansion of WT, DNMAML, and WT T cells in anti-Dll1/Dll4–treated mice at days 14, 21, and 35 after transplantation. Graphs show the absolute number of donor-derived H2Kb⁺H2Kd^– CD4⁺ or CD8⁺ T cells in the spleen of individual recipients. (D) Short-term Dll1/Dll4 inhibition was associated with persistently elevated percentages and absolute number of Tregs at day 35. Representative flow cytometry plots for intracellular FoxP3 staining are shown, including a sample stained with isotype control antibodies. Numbers indicate the percentage of cells in each quadrant. Bar graphs represent mean ± SD. **P < 0.01; *P < 0.05.

Figure 8. Preserved hematopoietic recovery after allogeneic transplantation in mice treated with anti-Dll1/Dll4 antibodies.

Allo-BMT and transient administration of anti-Dll1/Dll4 or control antibodies (days 0–10) were performed as described in Figure 6. (A) Weekly complete blood counts after allo-BMT showing unimpaired recovery in recipients treated with anti-Dll1/Dll4 antibodies. (B) CFU-GM activity in the BM on day 21 after transplantation. (C) Absolute numbers of CD45.1⁺ cells derived from B6-CD45.1 donor TCD BM at days 14, 21, and 35. This showed preserved engraftment and expansion of CD45.1⁺ donor-derived cells in the BM. Bar graphs represent mean ± SD.

Figure 9. Protection from thymic GVHD upon transient systemic Dll1/Dll4 blockade.

Lethally irradiated (8.5 Gy) BALB/c mice were transplanted with TCD BM (5 × 10⁶ cells) with or without WT or DNMAML T cells (10 × 10⁶ splenocytes). Isotype control vs. anti-Dll1/Dll4 antibodies were administered i.p. at days 0, 3, 7, and 10 (short course). (A) Thymus was assessed using flow cytometry to identify newly formed CD4⁺CD8⁺ DP thymocytes. At day 21, thymopoiesis was inhibited in the presence of anti-Dll4 antibodies (red arrow). At day 35, after antibody clearance, large numbers of DP thymocytes arose in anti-Dll1/Dll4–treated mice (blue arrow), indicating protection from GVHD-induced thymic damage. (B) Absolute number of CD4⁺CD8⁺ DP thymocytes at days 21 and 35 in individual allo-BMT recipients, quantifying preserved thymic recovery at day 35 in anti-Dll1/Dll4–treated mice. *P < 0.05.

Figure 10. Dll1/Dll4 blockade preserves substantial in vivo cytotoxicity and GVT effects.

Allo-BMT and transient administration of anti-Dll1/Dll4 or control antibodies (days 0–10) were performed as described in Figure 6 legend. (A) In vivo cytotoxicity assay. Allo-BMT recipient mice were challenged on day 14 with a 1:1 infusion of CFSE-labeled allogeneic targets and control cells (CFSE^hiH-2Kd⁺ BALB/c and CFSE^lo control H-2Kb⁺ B6-CD45.1 splenocytes, respectively). After 18 hours, elimination of the BALB/c targets was assessed in the spleen by flow cytometry. (B) Summary of in vivo cytotoxicity data in individual mice (n = 6–10/group). **P < 0.01. (C) Bioluminescence imaging was performed at the indicated time points after allo-BMT and infusion of host-type (H-2Kd⁺) A20-TGL tumor cells (10⁶/recipient on day 0). Representative mice are shown. Cross indicates death of all mice in the group of WT T cell recipients. (D) Cumulative incidence of tumor-related death (days 0–100) (n = 12/group, 1 × 10⁶ A20 cells/recipient). (E) Cumulative incidence of tumor-related death (days 0–70) (n = 25–35/group, 5 × 10⁶ A20 cells/recipient).