Indoles derived from intestinal microbiota act via type I interferon signaling to limit graft-versus-host disease

Abstract

The intestinal microbiota in allogeneic bone marrow transplant (allo-BMT) recipients modulates graft-versus-host disease (GVHD), a systemic inflammatory state initiated by donor T cells that leads to colitis, a key determinant of GVHD severity. Indole or indole derivatives produced by tryptophan metabolism in the intestinal microbiota limit intestinal inflammation caused by diverse stressors, so we tested their capacity to protect against GVHD in murine major histocompatibility complex-mismatched models of allo-BMT. Indole effects were assessed by colonization of allo-BMT recipient mice with tryptophanase positive or negative strains of Escherichia coli, or, alternatively, by exogenous administration of indole-3-carboxaldehyde (ICA), an indole derivative. Treatment with ICA limited gut epithelial damage, reduced transepithelial bacterial translocation, and decreased inflammatory cytokine production, reducing GVHD pathology and GVHD mortality, but did not compromise donor T-cell-mediated graft-versus-leukemia responses. ICA treatment also led to recipient-strain-specific tolerance of engrafted T cells. Transcriptional profiling and gene ontology analysis indicated that ICA administration upregulated genes associated with the type I interferon (IFN1) response, which has been shown to protect against radiation-induced intestinal damage and reduce subsequent GVHD pathology. Accordingly, protective effects of ICA following radiation exposure were abrogated in mice lacking IFN1 signaling. Taken together, these data indicate that indole metabolites produced by the intestinal microbiota act via type I IFNs to limit intestinal inflammation and damage associated with myeloablative chemotherapy or radiation exposure and acute GVHD, but preserve antitumor responses, and may provide a therapeutic option for BMT patients at risk for GVHD.

Graphical abstract

Figure 1.

Microbiome-derived indoles mitigate morbidity and mortality in a mouse model of GVHD. (A) Left panel: Changes in urinary 3-IS before (day −1) and after (day 1, 2, 7) lethal TBI at 11 Gy (2 × 5.5 Gy) without subsequent BMT (combined data from 2 experiments, n = 10, sampled at multiple time points). Right panel: 3-IS before (day −6) and 1 day after (day 1) completion of 6-day chemo regimen with busulfan (80 mg/kg total) and cyclophosphamide (200 mg/kg total), without subsequent BMT. Combined data from 2 experiments, n = 10, sampled at multiple time points. (B-F) B10.BR recipient mice were treated with streptomycin and then colonized with streptomycin and nalidixic acid resistant K12 or K12ΔTnaA E coli 1 week prior to lethal irradiation and allo-BMT with TCD-BM + T cells from C57Bl/6 donor mice. Various parameters were tracked before irradiation/allo-BMT (day −1) and then weekly for 5 weeks following transplant. Single experiment with n = 15 per condition and 5 mice censored per condition on day 21 for various assays. (B) Colonization as measured by colony-forming unit (CFU)/g of bacteria in feces assessed on E coli selective plates containing streptomycin and nalidixic acid (n = 15 per condition). (C) 3-IS levels in urine (n = 15 per condition). (D) Weight loss. (E) Bacterial translocation to MLN (day 21, n = 5 per condition). (F) Kaplan-Meier survival curve with ticks indicating mice censored on day 21. Statistics: Mantel Cox Log-rank (survival curve), Mann-Whitney or Kruskal-Wallis ANOVA. ****P < .0001; ***P = .0001 to .001; **P = .001 to .01; *P = .01 to .05.

Figure 2.

ICA reduces GVHD-related morbidity and mortality in a dose-dependent manner in allo-BMT recipients. (A-G) B10.BR recipients were lethally irradiated and subjected to allo-BMT with TCD-BM alone or in combination with purified T cells (TCD-BM + T) from C57Bl/6 donor mice to induce GVHD. Mice received daily oral gavage with 100 mg/kg or 150 mg/kg ICA or vehicle (VEH) starting 1 day prior to irradiation. All data from day 21 posttransplant. Representative data from 2 experimental repeats at each ICA dose; n = 15 to 20 per group, 5 mice censored per condition on day 21 for various assays. (A) Weight loss (n = 15-20). (B) Kaplan-Meier survival curve (n = 15-20) with ticks indicating mice censored for histological studies. (C) Hematoxylin and eosin (H&E) staining of distal colon in animals receiving 100 mg/kg ICA or VEH. (All images are ×200 magnification; “L” indicates the lumen). (D-F) Quantitation of colon histology. For each parameter, an average value per mouse was determined as detailed in supplemental Methods (n = 4 to 8 per condition). (D) Degree of crypt loss (0 = none → 3 = severe). (E) Apoptotic cells per crypt. (F) Degree of infiltrating immune cells (0 = none → 3 = severe). (G) Quantitation of CFU per gram MLN. (H) Image of representative result of dilution plating of MLN homogenates to determine CFU/g MLN. (I) TER measured across Caco-2 cell monolayers after treatment with increasing concentrations of ICA. Combined data from 3 experiments. (J) TER measured across Caco-2 cell monolayers damaged with TNF-α and treated with ICA (100 µM). Combined data from 2 experiments. Statistics: Mantel Cox Log-rank (survival curve), ANOVA. ****P < .0001; ***P = .0001 to .001; **P = .001 to .01; *P = .01 to .05.

Figure 3.

ICA decreases GVHD-associated inflammatory cytokines in allo-BMT recipients. (A-F) Lethally irradiated B10.BR recipients were transplanted with TCD-BM in combination with purified T cells from C57Bl/6 mice to induce GVHD. Mice received daily oral gavage of 150 mg/kg ICA. Samples were collected at day 21 posttransplant (n = 5 mice per condition, assayed in triplicate). Representative data from 1 of 3 experimental repeats. Cytokines (A) and chemokines (B) significantly altered in colon homogenates in mice transplanted with T-cell-depleted BM + T and following treatment with ICA (day 21). (C) Cytokines significantly altered in plasma in mice transplanted with T-cell-depleted BM + T and following treatment with ICA (day 21). (D) IL-22 in colon (left) and plasma (right) in mice transplanted with T-cell-depleted BM + T and following treatment with ICA (day 21). (E) Percentages of CD69 and PD1 positive CD4 donor T cells (left) or CD8 donor T cells (right) in mice treated with ICA or VEH. (F) Percentages of cytokine expressing CD4 donor T cells (left) or CD8 donor T cells (right) in mice treated with ICA or VEH. Statistics: Mann-Whitney rank sum. Error bars represent standard deviations. *P = .01 to .05; NS, not significant. Mantel Cox log-rank (survival curve).

Figure 4.

ICA treatment of allo-BMT recipients leads to sustained survival after removal of ICA and induces tolerance in allogeneic T cells. (A) Lethally irradiated B10.BR recipients were transplanted with TCD-BM in combination with purified T cells (TCD-BM + T) from C57Bl/6 (B6) mice to induce GVHD. Mice received daily oral gavage of 150 mg/kg ICA or vehicle through day 45 posttransplant, at which point ICA delivery was terminated. Weight loss and survival were tracked through day 60. (A) Treatment schema (top). Weight loss (middle). Survival curves representing a subset of vehicle-treated recipients (n = 9) and ICA-treated recipients (N = 10) from Figure 2, followed for an additional 15 days after ICA treatment was terminated (bottom). (B) Lethally irradiated B10.BR recipients were transplanted with TCD-BM in combination with purified T cells from C57Bl/6 mice to induce GVHD. Top panel: Treatment schema. Early ICA mice received daily oral gavage of ICA from day −2 to day 12 and vehicle day 13 to day 53. Late ICA mice received daily oral gavage of vehicle from day −2 to day 12 and ICA day 13 to day 53. Control mice received ICA or vehicle throughout. Middle panel: Weight loss. Lower panel: Kaplan-Meier survival curve. Early ICA and late ICA, n = 15 per group. Controls, n = 4 per group. (C-D) T cells were harvested from spleens of ICA-treated survivors (n = 8, from panel A) at 60 days posttransplant (15 days after termination of ICA delivery), labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE), and used for secondary transfer. Splenocytes were harvested from secondary recipients 3 days after transfer, and proliferation profiles assessed by Flow Cytometry. (C) CFSE proliferation profiles of CD8 T cells in control B6 → B6 syngeneic transfer (upper left), N = 3; control B6 → B10.BR allogeneic transfer (lower left), N = 4; marrow-derived T cells from ICA-treated survivors → B10.BR recipients (upper right) and donor spleen-derived T cells from ICA-treated survivors → B10.BR recipients (lower right), N = 4. Flow cytometry gating was used to quantify T cells which were marrow hematopoietic stem cell (HSC) derived vs donor splenic T-cell derived. (D) Replication indices for the CD8 CFSE profiles depicted in panel C as well as CD4 CFSE profiles. (E) Splenocytes from Early ICA and late ICA survivors at day 53 posttransplant (from panel B) were tested, along with splenocytes from B6 control and B6 → B10.BR control with chronic GVHD, in MLR. Stimulators included irradiated splenocytes from B6, B10.BR, and FVB. At 48 hours, culture wells were harvested and analyzed for donor-derived CD4 (left) and CD8 T-cells (right) and Ki67 expression. Error bars represent standard deviations. Statistics: Mantel Cox log-rank (survival curve), ANOVA, Student t test. **P = .001 to .01; *P = .01 to .05.

Figure 5.

ICA does not inhibit the GVL activity of allogeneic T cells. (A-D) B6 albino recipients were injected with C1498ff on day −2, lethally irradiated on day −1, and transplanted on day 0 with 5 × 10⁶ TCD-BM cells alone (n = 6) or in combination with 3 × 10⁶ splenocytes (TCD-BM + T, n = 9 per group) from B10.BR donor mice. Mice received daily oral gavage with 150 mg/kg ICA or vehicle (VEH) starting 5 days prior to irradiation. (A) Weight loss. (B) Kaplan-Meier survival curve. (C) Visualization of C1498ff by in vivo bioluminescent imaging on day 22 (left) and day 56 (right) posttransplant. Two different luminescence scales are used to best show detectable tumors in different cages of mice. (D) Quantitation of C1498ff bioluminescence on day 22 (left) and day 56 (right) posttransplant. Statistics: Mantel Cox log-rank (survival curve), Kruskal-Wallis ANOVA. ****P < .0001; ***P = .0001 to .001; **P = .001 to .01; *P = .01 to .05; ns, not significant.

Figure 6.

ICA regulates expression of genes in the type I interferon (IFN1) and circadian rhythm pathways. (A) Venn diagram indicating the schema for identification of both GVHD-dependent and GVHD- and ICA-dependent gene expression changes assessed 21 days after B6 → B10.BR transplant. (B) Hierarchical clustering of 97 z-score normalized GVHD- and ICA-dependent genes in all 4 experimental conditions (TCD-BM or TCD-BM + T treated with either VEH or ICA). Each column represents averaged z-score data from n = 4 (TCD-BM) or n = 5 (TCD-BM + T) animals. Group A represents genes downregulated by ICA in TCD-BM + T transplanted animals. Group B represents genes upregulated by ICA in TCD-BM + T transplanted animals. (C) Principal component analysis of GVHD- and ICA-dependent genes; n = 4 (TCD-BM) or n = 5 (TCD-BM + T). (D) Gene ontology analysis (GoAmigo) of group A and group B GVHD- and ICA-dependent genes. Statistics: 1-way and 2-way ANOVA tests were performed to determine significant differential expression of genes among groups and identify common genes. Differential expression filtering was performed using an unadjusted P value <.05 and absolute fold change >1.5.

Figure 7.

ICA reduces intestinal damage and extends survival following TBI. (A-G) C57B6 WT or C57B6 IFN-αR^−/− mice were lethally irradiated (11 Gy) without subsequent BMT. Control animals were not irradiated (NRC, no radiation control). Mice received daily oral gavage with 150 mg/kg ICA or vehicle (VEH) starting 1 day prior to irradiation. Representative data from 2 experiments. (A) Kaplan-Meier survival curve (NRC, n = 5; VEH and ICA, n = 10). (B) Severity of diarrhea (day 3 postirradiation) (NRC, n = 5; VEH and ICA, n = 10). (C) Quantitation of CFU per gram MLN (day 3 postirradiation) (NRC, n = 5; VEH and ICA, n = 9). (D) Villus height (distal ileum, day 3 postirradiation) measured in H&E-stained sections (NRC, n = 3; VEH and ICA, n = 5). (E) H&E staining of distal ileum on day 3 postirradiation. Typical villus height measurements are indicated by yellow lines. Images are ×200 magnification. (F) Regenerating crypt foci. Area of regenerating crypt foci were determined by measuring height and width of foci, as identified by H&E staining (n = 5 all groups). (G) H&E staining of distal ileum on day 4 postirradiation. Yellow lines represent typical height measurements of regenerating foci. Width measurements were determined similarly. Images are ×400 magnification. Statistics: Mantel Cox log-rank (survival curve) or Kruskal-Wallis ANOVA. ****P < .0001; ***P = .0001 to .001; **P = .001 to .01; *P = .01 to .05; ns, not significant.