Glucagon-like peptide 2 for intestinal stem cell and Paneth cell repair during graft-versus-host disease in mice and humans

Abstract

Acute graft-versus-host disease (GVHD) is a life-threatening complication after allogeneic hematopoietic cell transplantation (allo-HCT). Although currently used GVHD treatment regimens target the donor immune system, we explored here an approach that aims at protecting and regenerating Paneth cells (PCs) and intestinal stem cells (ISCs). Glucagon-like-peptide-2 (GLP-2) is an enteroendocrine tissue hormone produced by intestinal L cells. We observed that acute GVHD reduced intestinal GLP-2 levels in mice and patients developing GVHD. Treatment with the GLP-2 agonist, teduglutide, reduced de novo acute GVHD and steroid-refractory GVHD, without compromising graft-versus-leukemia (GVL) effects in multiple mouse models. Mechanistically GLP-2 substitution promoted regeneration of PCs and ISCs, which enhanced production of antimicrobial peptides and caused microbiome changes. GLP-2 expanded intestinal organoids and reduced expression of apoptosis-related genes. Low numbers of L cells in intestinal biopsies and high serum levels of GLP-2 were associated with a higher incidence of nonrelapse mortality in patients undergoing allo-HCT. Our findings indicate that L cells are a target of GVHD and that GLP-2-based treatment of acute GVHD restores intestinal homeostasis via an increase of ISCs and PCs without impairing GVL effects. Teduglutide could become a novel combination partner for immunosuppressive GVHD therapy to be tested in clinical trials.

Graphical abstract

Figure 1.

Loss of enteroendocrine L cells and its peptide GLP-2 upon GVHD induction. (A) Representative immunohistochemistry staining for GLP-2 (red) in colon sections. Shown are sections from untreated animals, mice undergoing syn-HCT on day 10, mice undergoing allo-HCT on day 10 and allo-HCT mice on day 10 conditioned with chemotherapy. Scale bars, 20 μm. (B) GLP-2⁺ cells quantification in untreated C57BL/6 (n = 5), untreated BALB/c mice (n = 10), TBI/syn-HCT BALB/c mice on day 10 (n = 10), chemotherapy/allo-HCT BALB/c mice on day 10 (n = 10), TBI/allo-HCT C57BL/6 mice on day 10 (n = 10) and TBI/allo-HCT BALB/c mice on day 10 (n = 10). Results from 1 or 2 independent experiment(s) are shown. (C) Relative gene expression of the L-cell marker Cldn4 mRNA in the small intestine of untreated BALB/c (n = 10), syn-HCT on day 10 (n = 10), and allo-HCT BALB/c mice on day 5 (n = 10). Gene expression was normalized to β-actin used as reference gene. Results are derived from 2 independent experiments. (D) Relative gene expression of the GLP-2R (receptor) in the large intestine in BALB/c mice under GVHD conditions euthanized at day 2 (n = 6), day 4 (n = 6), day 7 (n = 11), day 14 (n = 13) compared with untreated BALB/c mice (n = 10). Gene expression was normalized to β-actin used as reference gene. Results are derived from 2 independent experiments. (E) Representative western blot showing the GLP-2 protein levels in the small intestine of BALB/c mice undergoing TBI-based allo-HCT on day 10 in comparison with untreated mice. β-actin was used as loading control. (F) Quantification of GLP-2 total protein. Fold change was normalized to the β-actin levels and relative to the untreated controls (set as “1”). Results are from 2 independent experiments at 2 different time points, day 5 (n = 11) and day 10 (n = 8) after allogeneic HCT. (G) Representative western blots from the gastrointestinal tract (colon) of chemotherapy-based allo-HCT BALB/c mice on day 3 in comparison with untreated mice. (H) Quantification of GLP-2 protein levels in chemotherapy-based allo-HCT BALB/c mice (n = 10) vs the untreated group (n = 8). Results are shown from 2 independent experiments. The P values in this figure were calculated using the unpaired, 2-sided Student t test.

Figure 2.

Teduglutide treatment leads to improved survival and reduced histological scores in mice with acute SR-GVHD. (A) Percentage survival of C57BL/6 allo-HCT mice treated with teduglutide or vehicle (n = 17 per group). Mice were treated with teduglutide or vehicle from day −3 until day +3 unless differently specified. Results are derived from 3 independent experiments. P value was calculated using the Mantel-Cox test. (B) GVHD histopathological scores from target organs of allo-HCT C57BL/6 animals on day 10, mice were treated with teduglutide or vehicle (n = 7 per group). Results are derived from 1 experiment. The P values were calculated using a 2-sided Mann-Whitney U test. (C) Percentage survival of allo-HCT BALB/c mice treated with teduglutide or vehicle (n = 12 per group). Results are derived from 2 independent experiments. P value was calculated using the Mantel-Cox test. (D) GVHD histopathological scores from small and large intestine and liver from allo-HCT BALB/c mice on day 10, animals were treated with teduglutide or vehicle (n = 8 per group). Results are derived from 2 independent experiments. The P values were calculated using a 2-sided Mann-Whitney U test. (E) Percentage survival of allo-HCT BALB/c mice treated with vehicle or teduglutide from day −3 to day 3 + prednisolone from day 4 to day 13 (n = 10 per group). Data are derived from 2 independent experiments. P value was calculated using the Mantel-Cox test. (F) Histological GVHD scores from BALB/c mice on day 10, treated as the schedule above described with teduglutide + prednisolone or vehicle + prednisolone (n = 9 per group). Data are derived from 2 independent experiments. The P values were calculated using a 2-sided Mann-Whitney U test. (G) Percentage survival of syn-HCT BALB/c mice treated with teduglutide or vehicle (n = 5 per group). Results are derived from 1 experiment. P value was calculated using the Mantel-Cox test. (H) Relative Cldn4 gene expression in the small intestine of allo-HCT BALB/c mice on day 5, treated with vehicle (n = 10) or teduglutide (n = 11). The P value was calculated using the unpaired, 2-sided Student t test. (I) Relative Chga gene expression in the small intestine of allo-HCT BALB/c mice on day 10, treated with vehicle or teduglutide (n = 8 per group). Data are derived from 2 independent experiments. The P value was calculated using the unpaired, 2-sided Student t test. (J) Glp-2r gene expression from microarray-based analysis from BALB/c mice on day 10 treated with vehicle (n = 3) or teduglutide (n = 2) from day −3 until day +10. Results indicate an increase Glp-2r gene expression in teduglutide-treated mice compared with vehicle-treated animals. The experiment was performed once. Adjusted P = .02.

Figure 3.

Teduglutide treatment prevents PCs loss caused by GVHD in mice. (A) Representative immunohistochemistry staining for the PC marker, lysozyme (red), in small intestine from allo-HCT BALB/c mice on day 10. Mice treated with vehicle or teduglutide were compared with untreated controls. Mice were treated with teduglutide or vehicle from day −3 until day +3 unless differently specified. Scale bars, 50 to 100 μm. (B) Quantification of PCs/crypt of untreated BALB/c mice (n = 5) or from allo-HCT BALB/c mice on day 10 treated with vehicle or teduglutide (n = 8 per group). Results were pooled from 2 independent experiments. The P values were calculated using the unpaired, 2-sided Student t test. (C-E) qPCR-based quantification of lysozyme, Reg3γ, and Defα-4 expression in small intestine from untreated or allo-HCT BALB/c mice on day 5 treated with vehicle or teduglutide (n = 11 per group). Data were pooled from 2 independent experiments. The P values were calculated using the unpaired, 2-sided Student t test. (F) Volcano plot shows the top 25 regulated genes in BALB/c mice with GVHD on day 10 post–allo-HCT that were treated with teduglutide from day −3 to day +10 and relative to the vehicle-treated group. The experiment was performed once. (G) Gene-expression array indicates the top 10 positive regulated genes involved in response to gram-positive bacterium within the teduglutide-treated mice in comparison with the GVHD group. Mice were treated with teduglutide or vehicle from day −3 to day +10. The experiment was performed once. *Nonadjusted value of P < .05.

Figure 4.

GVHD and antibiotic treatment promote gastrointestinal dysbiosis. (A) The Bray-Curtis metric test shows the dissimilarity of the microbiome β diversity between the 3 groups (untreated, vehicle-treated, and teduglutide-treated mice). Stool samples from colon were isolated on day 5 after allo-HCT from BALB/c mice. Mice were treated with teduglutide or vehicle from day −3 until day +3 unless differently specified. Representative data from 2 independent experiments. (B-C) Relative abundance of specific bacteria at the genus level. Increased Firmicutes bacterial load has been associated with GVHD. We observed an upregulation of Flintibacter bacteria (belonging to the Firmicutes phylum) in the BALB/c allo-HCT vehicle-treated group (n = 6), and those decline after teduglutide treatment (n = 6). On the other hand, the loss of unclassified Bacteroidales due to GVHD was compensated by the treatment with teduglutide (n = 6 per group). Representative data from 2 independent experiments. The P values were calculated using a 2-sided Mann-Whitney U test. (D) Percentage survival of C57BL/6 mice undergoing allo-HCT treated with teduglutide from day −3 to day 3 + broad spectrum antibiotics or normal water (vehicle) from day −14 to day −1 (n = 10 per group). Data are derived from 2 independent experiments. P value was calculated using the Mantel-Cox test. (E) Gene expression of interferon-related genes in the intestinal tract of BALB/c mice treated from day −3 until day 10 with teduglutide compared with mice treated with vehicle. *Nonadjusted value of P < .05. The experiment was performed once.

Figure 5.

Teduglutide treatment promotes KGF production and protects ISCs from GVHD. (A) Gene-expression array from BALB/c mice treated with teduglutide or vehicle from day −3 to day 10 shows increased expression of different growth factors, one of them is KGF (FGF7) in the teduglutide-treated mice in comparison with the control mice. Results are shown from 1 experiment. *Nonadjusted value of P < .05. (B) Relative expression of KGF analyzed on day 5 in the small intestine of BALB/c allo-HCT mice treated with teduglutide or vehicle compared with the untreated control mice (n = 11 per group). Mice were treated with teduglutide or vehicle from day −3 until day +3 unless differently specified. Results are shown from 2 independent experiments. The P values were calculated using the unpaired, 2-sided Student t test. (C) Quantification of primary intestinal organoids that were cultured for 2 days with different teduglutide concentrations reveals an increased surface area in the wells treated with 250 and 500 nM teduglutide when compared with the untreated cells. Results are shown from 2 independent experiments. The P values were calculated using the unpaired, 2-sided Student t test. (D) Representative microscopy images of mouse primary intestinal organoids. Figures indicate crypt cell proliferation in a dose-dependent manner upon teduglutide treatment in contrast to the untreated cells. Results are shown from 2 independent experiments. Scale bars, 50 µm. (E) Representative immunofluorescence stain images stained for Lgr5⁺ GFP-ISCs (green) taken on day 6 from allo-HCT B6-Lgr5-EGFP^creER reporter mice that were treated with vehicle or teduglutide from day −3 to day 3. Images indicate reduced Lgr5⁺ cells in mice with GVHD in comparison with the untreated control mice. In contrast, the teduglutide-treated mice exhibit an enhanced number of Lgr5⁺ cells during GVHD. (F) Quantification of Lgr5⁺ cells/crypt of untreated B6-Lgr5-EGFP^creER (n = 7) or B6-Lgr5-EGFP^creER mice that underwent allo-HCT and were treated with teduglutide or vehicle (n = 8 per group). Teduglutide treatment protects ISCs from their loss due to GVHD. Results are shown from 2 independent experiments. The P values were calculated using the unpaired, 2-sided Student t test. (G-H) qPCR-based quantification of the ISC markers Olfm4 and Prom-1 expression in small intestine from untreated (n = 10) or allo-HCT BALB/c mice on day 5 treated with vehicle or teduglutide (n = 16 per group). Data were pooled from 2 independent experiments. The P values were calculated using the unpaired, 2-sided Student t test. (I) Gene-expression array of ISC markers analyzed on day 10 from the small intestine of allo-HCT BALB/c mice treated with teduglutide or vehicle from day −3 to day 10. Experiment was performed once.

Figure 6.

Teduglutide confers protection to mice with defective PCs and decrease intestinal cell apoptosis without loss of GVL effects. (A) Percentage survival of Irgm-1^−/− mice that underwent allo-HCT and were treated with teduglutide or vehicle (n = 9 per group) from day −3 to day 3. Survival curves shown an increased survival rate in mice with defective PCs treated with teduglutide. Results are shown from 2 independent experiments. P value was calculated using the Mantel-Cox test. (B) Histological scores from allo-HCT Irgm-1^−/− mice, small and large intestine were analyzed on day 10. Results indicate lower organ damage when GVHD mice were treated with teduglutide (n = 4) in comparison with the vehicle-treated group (n = 3). Mice were treated with teduglutide or vehicle from day −3 until day +3 unless differently specified. Results are shown from 1 experiment. The P values were calculated using a 2-sided Mann-Whitney U test. (C) Gene expression–based arrays analyzed on day 10 from the small intestine of allo-HCT BALB/c mice treated from day −3 to day 10 indicates the downregulation of different cell death–related genes in the teduglutide-treated group. Results are shown from 1 experiment. (D) Caspase 3 expression analyzed by qPCR in the small intestine of allo-HCT BALB/c mice treated from day −3 to day 10 with teduglutide or vehicle. Results display a downregulation of caspase 3 in mice treated with teduglutide (n = 10) in comparison with the high caspase 3 expression during GVHD (n = 11). Results are shown from 2 independent experiments. The P values were calculated using the unpaired, 2-sided Student t test. (E) Percentage quantification of apoptotic nuclei determined by a terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) staining. BALB/c mice underwent allo-HCT and were treated with teduglutide or vehicle from day −3 to day 3. Small intestine samples were stained to detect DNA brakes. Results indicate less apoptotic cells in GVHD mice treated with teduglutide (n = 7) compared with GVHD mice treated with vehicle (n = 8). The P value was calculated using the unpaired, 2-sided Student t test. (F) Percentage survival of BALB/c-WEHI-3B leukemia-bearing mice that received allo-HCT/T cells and treatment with teduglutide or vehicle or BM and leukemia cells alone (n = 10 per group). Data are derived from 2 independent experiments. P value was calculated using the Mantel-Cox test. (G) Representative bioluminescence images taken on day 22 from BALB/c mice transplanted with Ba/F3-ITD Luc⁺ cells + allo-BM or Ba/F3-ITD Luc⁺ cells + allo-BM followed by T-cell transplantation on day 2 and treated with teduglutide or vehicle. Bioluminescence imaging signal indicates the tumor growth in mice according to the group. Representative data from 2 independent experiments. (H) Percentage survival of C57BL/6-AML^{MLL PTD FLT3 ITD} leukemia bearing mice that received allo-HCT + leukemia cells or in addition T cells and were treated with teduglutide or vehicle (n = 10 per group). Data are shown from 1 experiment. P value was calculated using the Mantel-Cox test.

Figure 7.

Low numbers of L cells in the gut and high GLP-2 levels in blood correlate with poor patient outcome after allo-HCT. Sigmoid colonic biopsies from 29 patients with initial diagnosis of acute gastrointestinal GVHD were stained for GLP-1 for quantification of L cells. Median L cell number per crypt was 1.05, therefore 1 L cell per crypt was used as cutoff for stratification in high (≥1 L cell per crypt) and low L-cell counts. (A) The 2-year cumulative incidence of NRM in patients with low vs high L-cell counts. (B) Cumulative incidence of SR-GVHD at 6 months in patients with low vs high L-cell counts. (C) Representative immunohistochemistry stain images of colonic biopsies stained for GLP-1 (brown). Left, patient with high L-cell count who responded well to steroid treatment and stayed alive. Right, patient with low L-cell count who developed SR-GVHD and NRM. For GLP-2 blood-level correlation, serum prepared from fasting blood samples of 82 consecutive patients with newly diagnosed acute GVHD was analyzed for GLP-2 concentration. Median GLP-2 concentration was 3.7 ng/mL, which was used as cutoff for stratification in high (>3.7 ng/mL) and low GLP-2 levels. (D) Cumulative incidence of SR acute GVHD at 6 months in patients with high vs low serum GLP-2 concentrations. (E) Cumulative incidence of NRM at 2 years in patients with high vs low serum GLP-2 concentrations. GLP-2 level in the peripheral blood as a continuous variable was also associated with a higher risk of SR-GVHD (HR, 1.09; 95% CI, 1.00-1.19; P = .05) and NRM (HR, 1.17; 95% CI, 1.03-1.32; P = .01). (F) Further subgroup analysis was performed including only patients with no gastrointestinal symptoms at time of diagnosis of GVHD (n = 53). Cumulative incidence of SR-GVHD at 6 months again tended to be higher for patients with high GLP-2 levels: 54% (95% CI, 34% to 74%) vs 26% (95% CI, 8% to 44%), P = .05. (G) Cumulative incidence of NRM at 2 years was higher for patients with high GLP-2 concentrations: 40% (95% CI, 20% to 60%) vs 4% (95% CI, 0% to 12%) (P = .002).