Notch signaling mediated by Delta-like ligands 1 and 4 controls the pathogenesis of chronic GVHD in mice

Abstract

Chronic graft-versus-host disease (cGVHD) is a major complication of allogeneic hematopoietic cell transplantation (allo-HCT) and remains an area of unmet clinical need with few treatment options available. Notch blockade prevents acute GVHD in multiple mouse models, but the impact of Notch signaling on cGVHD remains unknown. Using genetic and antibody-mediated strategies of Notch inhibition, we investigated the role of Notch signaling in complementary mouse cGVHD models that mimic several aspects of human cGVHD in search of candidate therapeutics. In the B10.D2→BALB/c model of sclerodermatous cGVHD, Delta-like ligand 4 (Dll4)-driven Notch signaling was essential for disease development. Antibody-mediated Dll4 inhibition conferred maximum benefits when pursued early in a preventative fashion, with anti-Dll1 enhancing early protection. Notch-deficient alloantigen-specific T cells showed no early defects in proliferation or helper polarization in vivo but subsequently exhibited markedly decreased cytokine secretion and enhanced accumulation of FoxP3⁺ regulatory T cells. In the B6→B10.BR major histocompatibility complex-mismatched model with multi-organ system cGVHD and prominent bronchiolitis obliterans (BO), but not skin manifestations, absence of Notch signaling in T cells provided long-lasting disease protection that was replicated by systemic targeting of Dll1, Dll4, or both Notch ligands, even during established disease. Notch inhibition decreased target organ damage and germinal center formation. Moreover, decreased BO-cGVHD was observed upon inactivation of Notch1 and/or Notch2 in T cells. Systemic targeting of Notch2 alone was safe and conferred therapeutic benefits. Altogether, Notch ligands and receptors regulate key pathogenic steps in cGVHD and emerge as novel druggable targets to prevent or treat different forms of cGVHD.

Graphical abstract

Figure 1.

Notch blockade in mature T cells prevents sclerodermatous cGVHD. BALB/c mice were lethally irradiated (8-8.5 Gy), followed by transplantation of B10.D2-Thy1.2 TCD BM (light green circle), TCD BM with 12 × 10⁶ B10.D2-Thy1.1 WT splenocytes (red circle), or 12 × 10⁶ B10.D2-DNMAML splenocytes (blue circle), followed by monitoring with systemic or skin-specific clinical GVHD scores. B10.D2-DNMAML T cells cannot respond to canonical Notch signals. (A) Mean systemic clinical GVHD scores. *P < .01 (WT vs DNMAML T cells and WT T cells vs TCD BM only; 1-way ANOVA). (B) Mean skin-specific clinical GVHD scores; *P < .01 (WT vs DNMAML T cells and WT T cells vs TCD BM only; 1-way ANOVA). (C) Overall survival; *P < .001 (WT vs DNMAML T cells and WT T cells vs TCD BM only; log-rank test). n = 20 mice/group, pooled from 4 experiments. (D) Representative images of recipient mice on day 35 show absence of Scl-cGVHD features in TCD BM and DNMAML T-cell recipients. (E) Histopathological assessment of dermal thickness, a hallmark of skin cGVHD (n = 14, 3 experiments). *P < .01 (1-way ANOVA). (F) Flow cytometric assessment of inflammatory cytokine production and FoxP3 expression in donor-derived spleen T cells (day 6). Representative contour plots (left). Numbers indicate the percentage of events falling within indicated rectangular gates. Cumulative quantification (right) (n = 10 mice/group, from 3 experiments). *P < .001 (2-tailed unpaired Student t test). IFN, interferon; IL, interleukin; TNF, tumor necrosis factor.

Figure 2.

Dll1/Dll4 Notch ligands are critically active during the earliest posttransplant period and can be therapeutically targeted to prevent Scl-cGVHD. B10.D2→BALB/c chimeras were generated by transplanting B10.D2-Thy1.2 TCD BM (light green circle) vs TCD BM plus 12 × 10⁶ B10.D2-Thy1.1 WT splenocytes. T-cell recipients were treated with 4 peritransplant doses of isotype control (red circle), anti-Dll1 (purple-rimmed circle), anti-Dll4 (purple circle), or anti-Dll1 plus anti-Dll4 (blue circle) antibodies (5 mg/kg i.p. on days 0, 3, 7, and 10). (A-B) Biweekly systemic and cutaneous GVHD scores, showing that Dll4 is the dominant Notch ligand in the Scl-cGVHD model and that its targeting induces almost complete disease protection. *P < .01 (1-way ANOVA) for TCD BM and all groups receiving anti-Dll4 vs groups receiving isotype control or anti-Dll1 alone (cumulative data from 2 experiments). (C-D) B10.D2→BALB/c mice were generated as described above. T-cell recipients were treated with 4 doses of isotype control (red circle) or anti-Dll1/4 (blue circle) antibodies at days 0 to 10 vs anti-Dll1/Dll4 at later time points (days 10-20, purple-rimmed circle; days 40-50, purple circle). (C) Systemic GVHD scores were monitored longitudinally. Note lack of GVHD protection with delayed administration of anti-Dll1/4. (D) Histopathological analysis of dermal thickness in mice depicted in panel C. *P < .01 (1-way ANOVA). Note Scl-cGVHD protection with early anti-Dll1/Dll4 administration and trend toward benefit with delayed anti-Dll1/Dll4 treatment (P = .07). (E-F) B10.D2→BALB/c chimeras were generated as described above. In one cohort, anti-Dll1/Dll4 treatment was delayed by 48 hours (purple circle). Note loss of protection from systemic (E) and cutaneous (F) GVHD with delayed anti-Dll1/4 administration. *P < .01 (1-way ANOVA) for TCD BM and group receiving early anti-Dll1/4 vs groups receiving isotype control or delayed anti-Dll1/Dll4 (n = 12, cumulative data from 3 experiments).

Figure 3.

In the Scl-cGVHD model, immunodominant and pathogenic alloreactive Vβ3⁺T cells expand rapidly post-transplant and their elimination mitigates Scl-cGVHD. BALB/c mice were lethally irradiated and transplanted with B10.D2-Thy1.2 TCD BM (10⁷ cells) supplemented with 12 × 10⁶ CFSE-labeled B10.D2-Thy1.1 WT splenocytes. (A) Expression of Vβ3 in donor-derived CD90.1⁺CD4⁺ T cells in spleen or lymph nodes at the indicated time points after transplant. Vβ3⁺ T cells proliferated and expanded rapidly to become the majority of donor T cells on day 6, before their subsequent contraction. Representative flow cytometry plots, with numbers indicating the percentage of Vβ3⁺ cells among donor-derived CD4⁺ T cells (left). (B-C) B10.D2→BALB/c chimeras were generated by transplanting TCD BM only (light green circle) vs TCD BM with 3 × 10⁶ WT T cells depleted of Vβ3⁺ T cells (blue circle) or sham manipulated (red circle). (B) Cutaneous cGVHD manifestations were monitored biweekly. Note significant cutaneous GVHD score reduction in mice receiving Vβ3-depleted T cells as compared with whole T-cell recipients. *P < .05 for all comparisons (1-way ANOVA). (C-D) Histopathologic analysis of animals depicted in panel (B). (C-D) Cumulative data show decreased pathological hallmarks of cGVHD (C) with transplantation of Vβ3-depleted T cells, without corresponding impact on the severity of aGVHD (D). *P < .05 (1-way ANOVA). (E) Representative microphotographs of hematoxylin and eosin–stained skin sections showing increased dermal thickness in the whole T cell as compared with TCD BM recipients, with decreased thickness in recipients of Vβ3-depleted T cells (ear lobes, day 60). Scale bar represents 100 µm.

Figure 4.

Notch signaling enhances proinflammatory functions of alloreactive T cells and suppresses the expansion of alloantigen-specific regulatory T cells after transplantation. BALB/c recipients were lethally irradiated (800 cGy) and transplanted with B10.D2-Thy1.2 TCD BM (10⁷ cells) plus 12 × 10⁶ B10.D2-Thy1.1 WT or B10.D2-DNMAML splenocytes (labeled with eFluor450 to allow for tracking of cell division). Tissues were retrieved for flow cytometric analysis of alloantigen-specific Vβ3⁺ T-cell populations. (A) CD25 expression in CD4⁺Vβ3⁺ T cells at days 2 and 6 after transplantation. *P < .01 (2-tailed unpaired Student t test). (B) eFluor450 dilution at days 2 and 6 posttransplant showing preserved early proliferation of DNMAML as compared with WT CD4⁺Vβ3⁺ T cells. (C-D) Flow cytometric analysis of cytokine production by WT vs Notch-deprived DNMAML CD4⁺Vβ3⁺ T cells isolated from spleen or liver at day 6 after allo-HCT (ex vivo anti-CD3/CD28 restimulation before staining for intracellular cytokines). Representative contour plots are shown in panel C and cumulative data in panel D. Numbers indicate the percentage of events falling within indicated rectangular gates. *P < .01 (2-tailed unpaired Student t test). (E) Notch inhibition in Vβ3⁺CD4⁺ T cells alters transcription of Notch targets but does not affect Th-lineage polarization. RNA was isolated; a complementary DNA library was generated from sort-purified Vβ3⁺CD4⁺ T cells on day 6 after allo-HCT and then subjected to quantitative reverse-transcription PCR. While Notch targets were depressed (Dtx1), canonical Th1 (Tbx21) and Th2 (Gata3) transcription factors remained unaffected. (F-G) Notch inhibition enhances expansion of Vβ3⁺CD4⁺FoxP3⁺ Tregs after allo-HCT. Data show expression of FoxP3 in Vβ3⁺CD4⁺ T cells 6 days after allo-HCT (F) and total Vβ3⁺ Tregs at that time point (G). *P < .01 (2-tailed unpaired Student t test). Data in panels A-F are representative of at least 2 experiments with 10 animals per cohort.

Figure 5.

Notch signaling mediated by Dll1 and Dll4 is necessary for the development and persistence of BO-cGVHD. B10.BR recipients were conditioned with total body irradiation (8.3 Gy, day −1) and cyclophosphamide (120 mg/kg/day i.p., days −3 and −2) before transplantation with B6 WT TCD BM only (10⁷ cells) vs TCD BM plus 80 × 10³ B6 WT or B6-DNMAML T cells. (A) Pulmonary function testing (resistance, elastance, and compliance) was performed at day 56 after allo-HCT. Recipients of WT, but not DNMAML, T cells showed profound changes consistent with BO-cGVHD airway disease. *P < .05 (1-way ANOVA). (B) B10.BR recipients were transplanted as indicated above. After 4 weeks, systemic treatment was started with isotype control vs anti-Dll1, anti-Dll4, or a combination of anti-Dll1 and anti-Dll4 antibodies (5 mg/kg i.p. twice weekly, days 28-56). Pulmonary function tests were performed at day 56. *P < .05 for mice treated with isotype control antibodies (BO-cGVHD) vs mice receiving only TCD BM (GVHD-free control) and all groups receiving Notch ligand neutralizing antibodies. (C) Dll1 and/or Dll4 inhibition alters GC B-cell homeostasis, significantly decreasing the frequency of B cells with a GL7⁺Fas^hi GC B-cell phenotype in the spleen. *P < .05. Representative data from one of 2 experiments is shown with 8 animals per group.

Figure 6.

Notch1 and Notch 2 receptors deliver nonredundant pathogenic signals in T cells during BO-cGVHD. B10.BR recipients were conditioned with irradiation and cyclophosphamide and transplanted with B6 WT TCD BM cells (10⁷) supplemented with 80 × 10³ B6 WT, Cd4-Cre × Notch1^f/f (N1Ko), Cd4-Cre × Notch2^f/f (N2Ko), and Cd4-Cre×Notch1^f/fNotch2^f/f (N1Ko/N2Ko) T cells. (A) Notch1 and Notch2 receptors are nonredundant in driving pathogenic Notch signaling in T cells to cause BO-cGVHD. Pulmonary function tests at day 49 showing preservation of pulmonary functions in mice receiving Notch1-deficient, Notch2-deficient, or Notch1/2-deficient vs WT T cells. *P < .05 (1-way ANOVA). (B-C) Notch1 and/or Notch2 loss in T cells decreases the accumulation of GC B-cell (B) and Tfh cell (C) posttransplant in spleens of recipient animals. Spleens were harvested on day 60 posttransplant and analyzed for GC B cells as well as Tfh and Tfr cells. The Tfr/Tfh ratio was calculated by dividing the percentages of respective subsets in parental populations. *P < .05 (1-way ANOVA).

Figure 7.

Notch2 inhibition is a potential therapeutic strategy in BO-cGVHD without the detrimental side effects characteristic of Notch1 targeting. Following conditioning, B10.BR recipients were transplanted with B6 WT TCD BM cells (10⁷ cells) supplemented with 74 × 10³ B6 WT T cells. Treatment with control vs anti-NRR2 antibodies was given with biweekly injections for 7 weeks, before pulmonary testing was completed. (A) Pulmonary functions tests at day 49 after allo-HCT showing beneficial effects of Notch2 blockade with anti-NRR2 antibodies. (B-C) Anti-NRR2 blockade decreases peribronchiolar collagen deposition in BO-cGVHD. Representative microphotographs of Masson’s trichrome stained lung sections taken at ×200 magnification (B). Cumulative data (C) showing quantification of collagen deposition expressed as a ratio of blue over total stained area. (D-E) Peanut agglutinin staining in the spleens of TCD BM recipients vs recipients of TCD BM plus T cells treated with isotype control or anti-NRR2 antibodies show dampened GC B-cell responses in the spleens of anti-NRR2 treated recipients. Representative microphotographs taken at ×200 magnification (D) and cumulative data (E) quantifying average GC size. *P < .05 (1-way ANOVA).