SMAD4 TGF-β-independent function preconditions naive CD8+ T cells to prevent severe chronic intestinal inflammation

Abstract

SMAD4, a mediator of TGF-β signaling, plays an important role in T cells to prevent inflammatory bowel disease (IBD). However, the precise mechanisms underlying this control remain elusive. Using both genetic and epigenetic approaches, we revealed an unexpected mechanism by which SMAD4 prevents naive CD8+ T cells from becoming pathogenic for the gut. Prior to the engagement of the TGF-β receptor, SMAD4 restrains the epigenetic, transcriptional, and functional landscape of the TGF-β signature in naive CD8+ T cells. Mechanistically, prior to TGF-β signaling, SMAD4 binds to promoters and enhancers of several TGF-β target genes, and by regulating histone deacetylation, suppresses their expression. Consequently, regardless of a TGF-β signal, SMAD4 limits the expression of TGF-β negative feedback loop genes, such as Smad7 and Ski, and likely conditions CD8+ T cells for the immunoregulatory effects of TGF-β. In addition, SMAD4 ablation conferred naive CD8+ T cells with both a superior survival capacity, by enhancing their response to IL-7, as well as an enhanced capacity to be retained within the intestinal epithelium, by promoting the expression of Itgae, which encodes the integrin CD103. Accumulation, epithelial retention, and escape from TGF-β control elicited chronic microbiota-driven CD8+ T cell activation in the gut. Hence, in a TGF-β-independent manner, SMAD4 imprints a program that preconditions naive CD8+ T cell fate, preventing IBD.

Keywords: Cellular immune response; Gastroenterology; Immunology; Inflammatory bowel disease; T cells.

Figure 1. SMAD4 in T cells prevents severe chronic intestinal inflammation in a TGF-β–independent manner.

(A) Scheme representing the different pathways of TGF-β signaling and the mouse models used. (B) On the left panel, body weight of mice from 1 to 7 months of age (n = 3 or more for each time point) and on the right panel, weight of the mice at 5 to 7 months of age (WT n = 27, TKO n = 9, SKO n = 21, STKO n = 15, and R2SKO n = 16). All mice were male. (C and D) Representative pictures of colon and duodenum, colon length, and duodenum enlargement of the different strains of mice at 7 months of age (n = 6–17). (E) Representative H&E staining of duodenum and colon sections at 7 months of age. Scale bars: 50 μm. Original magnification, ×20. (F–I) Irradiated RAG2-KO mice were reconstituted with WT, R2KO, SKO, or R2SKO BM cells. Percentage change in body weight between the beginning and the end of experiment (n = 6–19) (F), colon length (n = 4–18) (G), histological intestinal damage score (n = 4–8) (H), and representative H&E staining of duodenum and colon sections (I). Scale bars: 50 μm. Original magnification, ×20. Red arrows highlight crypt abscesses. All data represent at least 3 independent experiments (C–I) and are presented as mean ± SD. Each symbol represents an individual mouse (n = 4 or more for each group). Data were analyzed using 1-way ANOVA with Tukey’s test. **P < 0.01, ***P < 0.001, ****P < 0.0001. NS, not significant (P > 0.05).

Figure 2. CD8αβ⁺ T cell depletion prevents intestinal inflammation upon SMAD4 deletion in T cells.

(A) Scheme of the in vivo CD8β⁺ and CD4⁺ T cell depletion. (B) Body weight on days 35–40 after reconstitution with WT or SKO BM cells and treatment with anti-CD8β or anti-CD4 depleting antibody (n = 5–20). (C) Representative pictures of colons and colon length measurement of BM-reconstituted mice, treated with anti-CD8β or anti-CD4 depleting antibody (n = 7–14). (D and E) Representative H&E staining of duodenum and colon sections (D) and histological damage score (E) of irradiated mice reconstituted with WT or SKO BM cells and treated with anti-CD8β or anti-CD4 depleting antibody (n = 3–4). Scale bars: 50 μm. Original magnification, ×20. Red arrows highlight crypt abscesses: yellow arrows highlight immune infiltrate, green arrows highlight crypt irregularity. All data represent at least 3 independent experiments and are presented as mean ± SD. Each symbol represents an individual mouse and n = 3 or more for each group. Data were analyzed by unpaired 1-way ANOVA followed by Tukey’s test. *P < 0.05; ****P < 0.0001. NS, not significant (P > 0.05).

Figure 3. SMAD4 in T cells prevents microbiota-mediated accumulation and epithelial activation of CD8⁺ T cells.

(A) Representative flow cytometry showing the frequency of CD8αβ⁺ T cells among CD45⁺ cells in the spleen and epithelium from the colon and small intestine of mice at 5–7 months of age (WT n = 14–29, TKO n = 5–7, SKO n = 8–19, STKO n = 4–8, and R2SKO n = 6–10). (B) Representative pictures showing immunofluorescence staining of CD8β (green), E-cadherin (red), and DAPI staining (blue) in the small intestine and colon sections of WT and SKO mice at 7 months of age. Scale bars: 50 μm. Original magnification, ×20. (C) Flow cytometry staining of GZMA, GZMB, and IFN-γ among splenic CD8αβ⁺ T cells and colonic CD8αβ⁺ intraepithelial lymphocytes (WT n = 8–21, TKO n = 4–13, SKO n = 5–26, STKO n = 2–12, and R2SKO n = 8–13). (D and E) Effect of antibiotic (ATB) treatment on the frequency, numbers, and activation of colonic intraepithelial CD8αβ⁺ T cells from WT and SKO mice at 5 months of age (WT without ATB n = 3–9, WT plus ATB n = 3–9, SKO without ATB n = 8–11, and SKO plus ATB n = 7–12). All data represent at least 3 independent experiments and are presented as mean ± SD. Each symbol represents an individual mouse. Data were analyzed by unpaired 1-way ANOVA with Tukey’s test. *P < 0.05, **P < 0.01, ****P < 0.0001. NS, not significant (P > 0.05).

Figure 4. In the absence of TGF-βR signaling, SMAD4 restrains TGF-β signature in naive CD8⁺ T cells inversely of TGF-βR signaling.

(A) Venn diagram showing the numbers of differentially expressed genes between WT (n = 3), R2KO (n = 3), and SKO (n = 3) naive F5 CD8αβ⁺ T cells after conducting whole-transcriptome sequencing. (B) Heatmap showing the hierarchical clustering of differentially expressed genes between WT, SKO, and R2KO F5 naive CD8αβ⁺ T cells. (C) Fold change (FC) (logarithmic scale) of gene expression of SKO over WT (in orange) and R2KO over WT (in green). DEGs correspond to those shown in heatmap in Figure 4B. (D) Volcano plot of RNA-seq data from R2KO (n = 3), SKO (n = 3), and R2SKO (n = 4) naive F5 CD8αβ⁺ T cells. The data for all genes are plotted as log₂FC versus the –log₁₀ of the adjusted P value. Genes selected as significantly different are highlighted as green and red dots. (E) Heatmap showing the log₂FC expression of genes of cluster II and III highlighted in Figure 2B, and for each condition, the heatmap value corresponds to the KO relative to WT (average of 3 biological replicates). (F) Heatmaps showing the expression of genes linked to CD8⁺ T cell effector functions and genes linked to the naive and quiescent stage in WT, R2KO, SKO, or R2SKO F5 naive CD8αβ⁺ T cells. (G) Violin plot showing the relative expression of effector genes from R2KO (n = 3), SKO (n = 3), and R2SKO (n = 4) as compared to WT (n = 4) CD8⁺ T cells.

Figure 5. SMAD4 depletion promotes expression of TGF-β repressors and impedes TGF-β response in CD8⁺ T cells.

(A) Volcano plot showing TGF-β inhibitory genes in SKO (n = 3, orange), R2KO (n = 3, green), and R2SKO (n = 4, brown) F5 naive CD8αβ⁺ T cells, all relative to WT (n = 4). (B) Quantitative RT-PCR analysis of the expression of indicated TGF-β regulatory genes in F5 naive CD8αβ⁺ T cells from spleen of WT, R2KO, SKO, and R2SKO mice. These mice are different from those used for RNA-seq data (n = 5–9). (C and D) Flow cytometry data showing inhibition of GZMB (n = 11–12 per group) and TBET (n = 8–9 per group) after anti-CD3/anti-CD28 stimulation of WT or SKO CD8αβ⁺ T cells with or without recombinant TGF-β at 10 ng/mL (C) or different concentrations (n = 5 mice per group) (D). The percentage of CD8αβ⁺ T cell inhibition was determined by calculating the ratio between anti-CD3/anti-CD28 plus TGF-β and anti-CD3/anti-CD28 alone. All data represent at least 2 independent experiments and are presented as mean ± SD. Each symbol represents an individual mouse. Data were analyzed by unpaired Student’s t test (C) and 1-way ANOVA with Tukey’s test for the other panels. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. NS, not significant (P > 0.05).

Figure 6. Overactivation of the remaining TGF-β signaling pathways in SMAD4-deficient mice does not rescue mice from intestinal immunopathologies.

(A and B) Flow cytometry plots showing the frequency of CD8αβ⁺ T cells among CD45⁺ cells present within the colonic epithelium of WT (n = 14), RCA (n = 13), SKO (n = 13), and SKO-RCA (n = 9) mice (A), and intracellular staining for GZMB among colonic epithelial CD8αβ⁺ T cells (B). (C) The body weight in grams of WT (n = 8), RCA (n = 5), SKO (n = 11), and SKO-RCA (n = 14) male mice. (D) H&E staining of duodenum and colon sections from mice at 8 months of age. Scale bars: 200 μm. Original magnification, ×20. All data represent at least 3 independent experiments and are presented as mean ± SD. Each symbol represents an individual mouse and n = 5 or more for each group. Data were analyzed by 1-way ANOVA with Tukey’s test.*P < 0.05, **P < 0.01, ****P < 0.0001. NS, not significant (P > 0.05).

Figure 7. SMAD4 promotes homeostatic survival of CD8αβ⁺ T cells in an opposite way to TGF-βR signaling.

(A) Flow cytometry staining of CD127 (IL-7Rα) on WT (n = 14), R2KO (n = 12), SKO (n = 12), and R2SKO (n = 7) F5 naive CD8αβ⁺ T cells from spleen and MLNs and histograms showing mean fluorescence intensity (MFI) of CD127 relative to WT. (B) Flow cytometry staining of phosphorylated STAT5 (p-STAT5) in WT (n = 8), R2KO (n = 8), SKO (n = 7), and R2SKO (n = 6) F5 naive CD8αβ⁺ T cells after in vitro IL-7 treatment and histograms showing relative MFI of p-STAT5 relative to WT. NT, not treated. (C) Survival monitoring of WT (n = 6), R2KO (n = 7), SKO (n = 6), and R2SKO (n = 3) naive F5 CD8αβ⁺ T cells treated or not with IL-7. (D) Flow cytometry data showing the frequency with absolute numbers of F5 naive CD8αβ⁺ T cells among CD45⁺ cells in the spleen of 3-month-old WT (n = 9), R2KO (n = 6), SKO (n = 9), and R2SKO (n = 3) F5 mice. All data represent at least 3 independent experiments and are presented as mean ± SD. Each symbol represents an individual mouse. Data were analyzed by 1-way ANOVA with Tukey’s test. *P < 0.05, ***P < 0.001, ****P < 0.0001. NS, not significant (P > 0.05).

Figure 8. SMAD4 promotes gut epithelial retention of CD8αβ⁺ T cells in an opposite way to TGF-βR signaling.

(A) Representative histograms of CD103 expression by F5 naive CD8αβ⁺ T cells — WT (n = 12), R2KO (n = 12), SKO (n = 13), and R2SKO (n = 7) — and histogram of the relative MFI of CD103 expression compared to WT. (B) Flow cytometry–assessed expression of CD103 on R2KO (n = 3), SKO (n = 14), and R2SKO (n = 6) CD8⁺ T cells from spleen, shown as relative to WT counterpart, after mixed BM experiment. (C–G) Experimental procedure for anti-CD103 blocking treatment (C), CD8⁺ T cell absolute number (D), body weight change from initial weight (E), colon length relative to WT (F), and H&E staining of duodenum and colon sections (G) of irradiated mice reconstituted with WT (n = 11) or SKO BM cells and treated (n = 9) or not (n = 9) with anti-CD103 blocking antibody. Scale bars: 200 μm. Original magnification, ×20. All data represent at least 3 independent experiments and are presented as mean ± SD. Each symbol represents an individual mouse. Data were analyzed by an unpaired Student’s t test (F) and 1-way ANOVA followed by Tukey’s test for other panels. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. NS, not significant (P > 0.05).

Figure 9. The TGF-β–independent function of SMAD4 that restrains CD103 and IL-7R occurs in CD8⁺ T cells at the naive stage.

(A) Experimental procedure for TAT-CRE experiment. Naive and memory CD8⁺ T cells were purified from either Smad4^fl/fl Stop^fl/fl Rosa26^EYFP or Smad4^wt/wt Stop^fl/fl Rosa26^EYFP mice and treated with TAT-CRE recombinant protein prior to transfer into RAG2-KO mice and 3 weeks later, cells were recovered and analyzed by flow cytometry. (B) Histograms illustrating the expression of CD103 and CD127 in YFP⁺ (recombined) in the naive (n = 7) or memory state (n = 6) CD8⁺ T cells and YFP^– (nonrecombined) counterpart CD8⁺ T cells from Smad4^fl/fl Stop^fl/fl Rosa26^EYFP mice. (C) Similarly, histograms depicting the expression of CD103 and CD127 in YFP⁺ in the naive (n = 7) or memory state (n = 4) CD8⁺ T cells and YFP^– counterpart CD8⁺ T cells from Smad4^wt/wt Stop^fl/fl Rosa26^EYFP mice. The data in panels B and C are expressed as percentage or MFI. Each symbol represents an individual mouse and n = 4 or more for each group. All data represent at least 3 independent experiments and are presented as mean ± SD. Data were analyzed by a paired Student’s t test. **P < 0.01, ***P < 0.001. NS, not significant (P > 0.05).

Figure 10. In the absence of TGF-βR signaling, SMAD4 largely binds to DNA and mediates epigenetic control of TGF-β target genes in naive CD8⁺ T cells.

(A) Venn diagram showing the number of SMAD4 common peaks between WT (pool of 3 mice) and R2KO (pool of 5 mice) naive CD8αβ⁺ T cells. (B) The proportions of SMAD4 peaks associated with promoter, 5′ UTR, 3′ UTR, exon, intron, and intergenic regions in WT and R2KO naive F5 CD8αβ⁺ T cells. (C) Enriched heatmaps showing the SMAD4 occupancy signals in genomically aggregated TSS regions in WT and R2KO CD8⁺ T cells. Each panel represents 2 kb upstream and downstream of the TSSs. (D) Venn diagram showing the overlap between SMAD4 ChIP-seq peaks and RNA-seq DEGs. (E) SMAD4-binding ChIP-seq peaks in WT (blue), R2KO (green), or SKO control (orange), in corresponding genes. (F) Transcription factor (TF) top motifs in SMAD4-binding sites in WT and R2KO CD8⁺ T cells. The x axis represents the log(P value) of the motif enrichment, and the y axis represents the fold change of the motif enrichment. (G) The 3 top motifs found by hypergeometric optimization of motif enrichment (HOMER) analysis among SMAD4-binding peaks in WT and R2KO CD8⁺ T cells. (H) qPCR-based ChIP analysis of SMAD4 on the promoters/enhancers of Smad7, Skil, and Itgae in WT, R2KO, and SKO F5 naive CD8αβ⁺ T cells. Each data point represents a pool of 3–5 mice. (I) qPCR-based ChIP analysis of H3K27ac on the promoters/enhancers of Smad7, Skil, and Itgae in WT, R2KO, SKO, and R2SKO F5 naive CD8αβ⁺ T cells. Each data point represents a pool of 3–5 mice.