Colonic epithelial-derived FGF1 drives intestinal stem cell commitment toward goblet cells to suppress inflammatory bowel disease

Abstract

Understanding the molecular mechanisms that regulate intestinal epithelial cell (IEC) renewal provides potential targets for inflammatory bowel disease (IBD). Growing evidence has highlighted the importance of epithelial signals in regulating intestinal stem cell (ISC) differentiation. However, it remains unclear which IEC-derived cytokines can precisely regulate ISC commitment toward specific mature cells. Here we systematically analyze all fibroblast growth factors (FGFs) expression and find that colonic FGF1 levels are inversely correlated with the severity of IBD in mouse models and patients. IEC-specific Fgf1 deletion leads to impaired goblet cell differentiation and exacerbated colitis, while pharmacological administration of recombinant FGF1 (rFGF1) alleviates colitis by enhancing goblet cell differentiation and improving colonic epithelial integrity. Mechanistic studies reveal that rFGF1 directs ISC differentiation toward goblet cells via FGFR2-TCF4-ATOH1 signaling axis. In conclusion, our study identifies an epithelial niche-derived FGF1 that regulates ISC commitment toward goblet cells, shedding light on strategies for treating IBD.

Fig. 1. Colonic FGF1 levels negatively correlate with pathological grades of mouse model and human inflammatory bowel disease (IBD).

A, B Relative mRNA levels of all Fgfs in colon tissues of C57BL/6 J mice and DSS (A)/TNBS (B)-induced colitis were determined by qRT-PCR (n = 5). C, D Analysis of the expression levels of FGF1 (green) and EpCAM (red) in the distal colon sections of control mice and DSS-induced UC or TNBS-induced CD mouse model by IF staining. Nuclei were labeled with 4,6-diamidino -2-phenylindole (DAPI). The FGF1 stains were quantified using ImageJ software (n = 5, the mean value of 2 fields in each mouse). E–J IF staining of FGF1 (upper panel) and PAS-AB staining (lower panel) of human colon sections from UC (E) (n = 8 for Healthy Control group, n = 9 for Mild group, n = 15 for Moderate group, n = 9 for Severe group) and CD (F) (n = 10 for Healthy Control group, n = 8 for Mild group, n = 10 for Moderate group, n = 14 for Severe group) patients. Correlation analysis between IF intensity of FGF1 and UCEIS (G)/ SEC-CD (I) or Mucin granules⁺ cells per crypt in UC (H) and CD (J) patients. Human colon sections were classified by ulcerative colitis endoscopic index of severity (UCEIS) of 1-3 (mild), 4-6 (moderate) and 7-8 (severe) or simple endoscopic score for Crohn’s Disease (SEC-CD) of 3-6 (mild), 7-16 (moderate) and 17-30 (severe). Data was presented as mean ± SEM. (A–D) two-tailed unpaired t-test (Fgf22 (A) and Fgf4, Fgf22, Fgf23 (B) using non-parametric statistical method, two-tailed Mann-Whitney Test); (G–J) Spearman’s correlation. ns, not significance; nd, not detectable.

Fig. 2. Deficiency of Fgf1 in intestinal epithelial cells sensitizes mice to DSS-induced colitis.

A, B Deficiency of Fgf1 in IECs was verified by qRT-PCR (A) and IF staining (B) (n = 5). C–E Weight loss (C), disease activity index (D) and colon length (E) of Fgf1^fl/fl and VilCreFgf1^fl/fl mice that received regular drinking water with or without 2.5% DSS for 7 days. The distal colon tissues were collected and analyzed at the end of this experiment (n = 6 for Fgf1^fl/fl + DSS group, n = 5 for other groups). F H&E staining of distal colon sections of Fgf1^fl/fl and VilCreFgf1^fl/fl mice with or without DSS challenge and its histology score (n = 6 for Fgf1^fl/fl + DSS group, n = 5 for other groups). G Goblet (Muc2⁺, in green), tuft (Dclk1⁺, in green), enteroendocrine (ChgA⁺, in green) and absorptive enterocyte (Fabp1⁺, in green) cells were quantified by IF staining of the distal colon sections of Fgf1^fl/fl and VilCreFgf1^fl/fl mice with or without DSS challenge (n = 6 for Fgf1^fl/fl + DSS group, n = 5 for other groups). H–I PAS-AB staining of the distal colon sections of Fgf1^fl/fl and VilCreFgf1^fl/fl mice with or without DSS challenge (H). Mucin granules⁺ cells per crypt were counted (I) (n = 6 for Fgf1^fl/fl + DSS group, n = 5 for other groups). Data was presented as mean ± SEM. (A, right panel of B) two-tailed unpaired t-test (kidney (A) using non-parametric statistical method, two-tailed Mann-Whitney Test); (C, D) P values on day 7 were calculated using ordinary two-way ANOVA, followed by Sidak; (right panel of E–G, I) ordinary two-way ANOVA, followed by Sidak. ns, not significance.

Fig. 3. Administration of rFGF1 increased goblet cell number and improved the epithelium integrity to resist DSS or TNBS-induced colitis.

A The experimental design. C57BL/6 J mice received regular drinking water with or without 2.5% DSS for 7 days, and daily i.p. injected with PBS or rFGF1 starting from the third day until the end of this experiment. The distal colon tissues were collected and analyzed at the end of this experiment (n = 4). B–D Weight loss (B) disease activity index (C) colon length (D) (n = 4). E, F H&E (upper panel), PAS-AB staining (middle panel) and Muc2 IF staining (lower panel) of distal colon sections (E). Histology scores (left panel; n = 4), counting of mucin granules-positive cells per crypt (middle panel; n = 4) and Muc2-positive cells per crypt (right panel; n = 4) of distal colon sections (F). G Colonic stem cells (Lgr5), goblet cells (Muc2), Tuft cells (Dclk1), enteroendocrine cells (ChgA) and absorptive enterocytes (Fabp1, Krt20, Apoa1) markers were determined by qRT-PCR (n = 4). H–K BALB/c mice were rectally administered with TNBS, followed by PBS or rFGF1 injection. The survival rates (H) weight loss (I) disease activity index (J) and colon length (K) of mice were measured. The distal colon tissues were collected and analyzed at the end of this experiment (n = 6 for Control group, n = 5 for TNBS + PBS group, n = 9 for TNBS + rFGF1 group). L–M H&E (upper panel), PAS-AB staining (middle panel) and Muc2 IF staining (lower panel) of distal colon sections (L). Histology scores (left panel), counting of mucin granules-positive cells per crypt (middle panel) and counting of muc2-positive cells per crypt (right panel) of distal colon sections (M) (n = 6 for Control group, n = 5 for TNBS + PBS group, n = 9 for TNBS + rFGF1 group). N Colonic stem cells (Lgr5), goblet cells (Muc2), tuft cells (Dclk1), enteroendocrine cells (ChgA) and absorptive enterocytes (Fabp1, Krt20, Apoa1) markers were determined by qRT-PCR (n = 6 for Control group, n = 5 for TNBS + PBS group, n = 9 for TNBS + rFGF1 group). Data was presented as mean ± SEM. (B–C) P values on day 12 were calculated using ordinary one-way ANOVA, followed by Dunnett; (I–J) P values on day 4 were calculated using ordinary one-way ANOVA, followed by Dunnett; (D–G, K–N) ordinary one-way ANOVA, followed by Dunnet (Muc2, Dclk1, ChgA and Krt20 (N) using non-parametric statistical method, Kruskal Willis Test with a post hoc Dunn’s test). ns, not significance; nd, not detectable.

Fig. 4. FGF1 drives intestinal stem cell differentiation toward goblet cells.

A, B qRT-PCR analysis of Elf3 and Klf4 mRNA levels in distal colon sections of DSS (A, n = 4) or TNBS (B, n = 6) induced IBD mouse models treatment with PBS or rFGF1. C Representative images (upper panel, n = 5) and IF staining of Ki67 (lower panel, n = 4) in murine colonic organoids derived from Fgf1^fl/fl and VilCreFgf1^fl/fl mice were cultured for 7 days and analyzed. D Goblet cells (Muc2⁺, in green), tuft cells (Dclk1⁺, in green), enteroendocrine cells (ChgA⁺, in green) and absorptive enterocyte (Fabp1⁺, in green) cells in colonic organoids derived from Fgf1^fl/fl and VilCreFgf1^fl/fl mice were quantified by IF staining (n = 5). E–H Colonic organoids from normal C57BL/6 J mice were cultured and stimulated with vehicle or rFGF1 for 7 days (n = 5). E Representative images (upper panel) and Ki67 IF staining (lower panel) of murine colonic organoids. F, G The mRNA levels of Muc2, ChgA, Dclk1 (F) and Fabp1, Krt20 and Apoa1 (G) in colonic organoids were analyzed by qRT-PCR.H Goblet cells (Muc2⁺, in green), tuft cells (Dclk1⁺, in green), enteroendocrine cells (ChgA⁺, in green) and absorptive enterocyte (Fabp1⁺, in green) cells were quantified by IF staining and its semi-quantitation of IF intensity. I–L Human colonic organoids were cultured and stimulated with vehicle or rFGF1 for 7 days (n = 5). I Representative images (upper panel) and Ki67 IF staining (lower panel) of human colonic organoids. J, K The mRNA levels of Muc2, ChgA, Dclk1 (J) and Fabp1, Krt20 and Apoa1 (K) in human colonic organoids were analyzed by qRT-PCR.L Goblet cells (Muc2⁺, in green), tuft cells (Dclk1⁺, in green), enteroendocrine cells (ChgA⁺, in green) and absorptive enterocyte (Fabp1⁺, in green) cells were quantified by IF staining and its semi-quantitation of IF intensity. Data was presented as mean ± SEM. (A–B, F–G, J–K, lower panel of C, E, I, right panel of D, H, L) two-tailed unpaired t-test (Elf3 (B) and Apoa1 (G) using non-parametric statistical method, two-tailed Mann-Whitney Test). ns, not significance; nd, not detectable.

Fig. 5. ATOH1 is required for enhanced goblet cell differentiation induced by rFGF1.

A Heatmap of differentially expressed Atoh1 and its downstream target genes in PBS and rFGF1-treated UC mice (n = 4). The mRNA level of Atoh1 in PBS or rFGF1-treated UC mice (upper panel, n = 4) and CD mice (lower panel, n = 5) was detected by qRT-PCR. B, C IF staining of ATOH1 in distal colon sections of Fgf1^fl/fl and VilCreFgf1^fl/fl mice challenged with DSS (B) or TNBS (C), and its semi-quantitation of IF intensity (n = 4, the mean value of 2 fields in each mouse). D, E IF staining of ATOH1 in distal colon sections of DSS (D) or TNBS (E)-induced two IBD mouse models, followed by PBS or rFGF1 treatment and its semi-quantitation of IF intensity (n = 4, the mean value of 2 fields in each mouse). F Representative images (upper panel) and IF staining of Muc2 (lower panel) in shAtoh1 or shNC-transfected colonic organoids stimulated with vehicle or rFGF1 (n = 6). G, H Deletion of Atoh1 in VilCre^ERT2Atoh1^fl/fl mice was confirmed by qRT-PCR (G) and IF staining of colon tissues (H) (n = 4). I–M VilCre^ERT2Atoh1^fl/fl mice were injected with tamoxifen for five consecutive days and then given drinking water containing 1.5% DSS to induce acute colitis, followed by vehicle or rFGF1 administration for 7 days. At end of the experiment, the distal colon tissues were harvested and examined. The schematic diagram shows the strategy of tamoxifen injection, DSS challenge and rFGF1 administration (n = 4). J–L Weight loss (J), disease activity index (K), colonic length (L) were monitored (n = 4). M H&E (upper panel), PAS-AB staining (middle panel) and Muc2 IF staining (lower panel) of distal colon sections (n = 4). Data was presented as mean ± SEM. (A, D–E, H) two-tailed unpaired t-test; (B, C, F, J–M) ordinary two-way ANOVA, followed by Sidak; (G) Non-parametric statistical method, two-tailed Mann-Whitney test. ns, not significance; nd, not detectable.

Fig. 6. TCF4 mediates the transcriptional regulation of Atoh1 by FGF1.

A Workflow to find the rFGF1-regulated potential transcription factors of Atoh1, and heatmap of expression level of potential transcription factors (n = 4). B The expression levels of potential transcription factors of Atoh1 (Tcf4, Nr4a1, Nfya, Tec, Hoxa13, Ncaph2, Kdm1a) in distal colon sections of PBS or rFGF1-treated UC mice were analyzed by qRT-PCR (n = 4). C, D IF staining of TCF4 in distal colon sections of Fgf1^fl/fl and VilCreFgf1^fl/fl mice challenged with DSS (C) or TNBS (D) and its semi-quantitation of IF intensity (n = 4, the mean value of 2 fields in each mouse). E, F The expression levels of TCF4 in distal colon sections of DSS-treated UC mice (E) and TNBS-treated CD mice (F) after PBS or rFGF1 administration and IF staining (n = 8). G Predicted TCF4 binding sites on the promoter of Atoh1 by the JASPAR database (left panel). Six binding sites with the most significant prediction scores were selected for pairwise deletion mutation analysis (right panel; M1, in purple; M2, in green; M3, in red). H, I Luciferase activity in HT-29 cells transfected with a human Atoh1 promoter reporter plasmid or the predicted TCF4 binding site mutant reporter along with a Tcf4 overexpression plasmid (n = 4). J ChIP experiments. The data were normalized to the corresponding IgG (n = 4). K Images of murine colonic organoids transfected small hairpin RNA targeting Tcf4 (shTcf4) or negative control hairpin (shNT) followed by treated with rFGF1 or vehicle (n = 6). L IF staining of ATOH1 (upper panel) and Muc2 (lower panel) in colonic organoids in different groups and its semi-quantitation of fluorescent intensity (n = 6). Data was presented as mean ± SEM. (B, right panel of E–F, J) two-tailed unpaired t-test (Tec and Ncaph2 in penal B using non-parametric statistical method, two-tailed Mann-Whitney Test); (H–I) ordinary one-way ANOVA, followed by Dunnett; (right panel of C–D and L, lower panel of K) ordinary two-way ANOVA, followed by Sidak. ns, no significance.

Fig. 7. FGFR2 mediates the protective effects of rFGF1 on IBD in mice.

A Single-cell RNA sequencing analysis (GSM4983265) showed that Fgfr2 is highly expressed in stem cells. B Representative image of colon distal stained with Lgr5 (green), FGFR2 (red) and DAPI (blue) from the C57BL/6 J mice (n = 3). C Deletion of Fgfr2 in Lgr5-EGFP-Cre^ERT2Fgfr2^fl/fl mice treated with vehicle or tamoxifen was confirmed by IF staining (n = 5). D Murine colonic organoids derived from vehicle or tamoxifen-treated Lgr5-EGFP-Cre^ERT2Fgfr2^fl/fl mice were stimulated with vehicle or rFGF1. Representative images, IF staining of TCF4, ATOH1 and Muc2 in colonic organoids were shown (n = 6). E–I Vehicle or tamoxifen-treated Lgr5-EGFP-Cre^ERT2Fgfr2^fl/fl mice were challenged with drinking water containing 2.5% DSS to induce UC, followed by colon tissues were harvested and examined (n = 4 for TAM + PBS group, n = 5 for other groups). Weight loss (E), disease activity index (F) and colon length (G) of different groups were monitored. H H&E staining of distal colon sections and histology scores (upper panel), PAS-AB staining of distal colon sections and mucin granules-positive cells per crypt (lower panel). I IF staining of TCF4 (upper panel), ATOH1 (middle panel) and Muc2 (lower panel) in distal colon sections, and its semi-quantification of IF intensity. Data was presented as mean ± SEM. (right panel of C) two-tailed unpaired t-test; (right panel of D–F, right panel of G–I) ordinary two-way ANOVA, followed by Sidak. ns, no significance.