VGLL4 modulates Paneth cells and sustains intestinal homeostasis

Abstract

Paneth cells are defensive cells in the intestinal tract, which secrete niche factors and antimicrobial peptides (AMPs) to maintain the small intestinal stem cell niche and immune homeostasis. Here, we show that Vestigial-like family member 4 (VGLL4) plays a pivotal role in maintaining small intestinal homeostasis and in regulating Paneth cells. VGLL4 expression is downregulated in response to irradiation and DSS-induced colitis. Consistently, public datasets of human colitis show reduced VGLL4 expression. Loss of VGLL4 in the intestinal epithelium decreases Paneth cell numbers and AMPs production, and triggers gut microbiota dysbiosis, impairing intestinal regenerative capacity. Mechanistically, VGLL4 forms a complex with TEAD4 and ATOH1, stimulating GFI1 expression and promoting Paneth cell differentiation. Furthermore, VGLL4 forms a complex with TEAD4 and TCF4 to induce defensin expression, thereby maintaining microbiota composition. Collectively, our findings uncover novel roles for VGLL4 in intestinal homeostasis.

Synopsis

The Vestigial-like family protein VGLL4 is a crucial regulator of intestinal health. Its expression correlates with Paneth cell differentiation and antimicrobial peptide production in the intestine, whereas its deficiency is linked to intestinal dysbiosis.

• VGLL4 downregulation is a feature of intestinal injury across mouse models and human colitis.

• VGLL4 is essential for Paneth cell function and differentiation via a VGLL4-TEAD4-ATOH1 complex that activates GFI1.

• VGLL4 maintains gut microbiota balance by forming a VGLL4-TEAD4-TCF4 complex to induce defensin production.

The Vestigial-like family protein VGLL4 is a crucial regulator of intestinal health. Its expression correlates with Paneth cell differentiation and antimicrobial peptide production in the intestine, whereas its deficiency is linked to intestinal dysbiosis.

Introduction

The intestinal tract is organized into crypts and villi, increasing the surface area for digestion and absorption while forming a crucial barrier against food antigens and microbes (Martel et al, 2022). The epithelial surface is constantly renewed every 5–7 days through crypt cell proliferation (Flier and Clevers, 2009; Vermeulen, 2013). Throughout this dynamic process, Paneth cells (PCs) create an essential environment for sustaining homeostasis (Cheng, 2005). PCs are unique epithelial cells derived from Atonal bHLH transcription factor 1 (ATOH1)-positive secretory progenitors. The determination of Paneth cell fate requires SRY-box transcription factor 9 (SOX9) and growth factor independent protein 1 (GFI1) (Cui et al, 2023; Takahashi and Shiraishi, 2020). In mice, the loss of SOX9 or GFI1 leads to deficiencies in the Paneth cell lineage, and GFI1 acts downstream of ATOH1 to orchestrate Paneth cell differentiation (Bastide et al, 2007; Mori-Akiyama et al, 2007; Shroyer et al, 2005).

Mature Paneth cells are interspersed between intestinal stem cells and provide crucial niche factors, such as Wnt family member 3 A (Wnt3a), transforming growth factor α (TGFα), epidermal growth factor (EGF), R-spondin 1, and Notch ligands (Delta-like 4 and Delta-like 1), which are essential for maintaining the stem cell population (Sato et al, 2011). PCs also mediate the host-microbe interactions through the secretion of enzymes, growth factors, and antimicrobial peptides and proteins (AMPs) (Bevins and Salzman, 2011; Lueschow and McElroy, 2020), making them guardians of the small intestinal innate immune system (Lueschow and McElroy, 2020). The discernible granules in PCs are rich in α-defensins (referred to as cryptdins in mice) and Lysozyme, which are important regulators of the microbiome (Fu et al, 2023; Salzman et al, 2009). Dysfunction or reduced Paneth cell numbers are observed in Crohn’s disease (Liu et al, 2018; Wehkamp et al, 2008). Hence, elucidating the regulatory mechanisms governing Paneth cell development and function is essential for understanding the intestinal homeostasis maintenance, which may offer new strategies for treating intestinal inflammation and diseases.

VGLL4, a member of the Vestigial-like family, is known as either a co-repressor or activator, engaging with various transcription factors (TFs) in a context-specific manner during organ development, tissue differentiation and tumor progression (Jiao et al, 2014; Koontz et al, 2013; Zhang et al, 2014). We and others have demonstrated that VGLL4 competes with Yes-associated protein (YAP) for binding to TEA domain transcription factors (TEADs), inhibiting YAP-TEAD-driven cell proliferation in various cancers (Deng and Fang, 2018; Guo et al, 2013; Jiao et al, 2014; Mickle et al, 2021; Zhang et al, 2014). Conversely, VGLL4 interacts with TEAD4 to recruit C-terminal Binding Protein 2 (CtBP2) to suppress adipogenesis (Zhang et al, 2018). VGLL4 also promotes muscle differentiation through TEAD4 and enhances RUNX2 transcriptional activity during osteoblast differentiation by disrupting the TEAD4/RUNX2 (RUNX family transcription factor 2) complex (Suo et al, 2020). Given the diverse roles in tissue differentiation, VGLL4’s potential involvement in maintaining intestinal homeostasis remains to be elucidated.

In this study, we used lineage tracing and HA-tagged mice to reveal a crypt-specific expression pattern of VGLL4 in the gut, which is crucial for intestinal homeostasis. VGLL4 deficiency in intestinal epithelial cells (IECs) impaired regenerative ability and reduced Paneth cell numbers and AMPs expression. This involves two signaling complexes, one with TEAD4-ATOH1 and another with TEAD4-TCF4 (T-cell-specific transcription factor 4), which separately trigger the expression of GFI1 and AMPs, which are essential for sustaining Paneth cell differentiation and microbiota diversity. Taken together, this study unveils dual functions of VGLL4 in maintaining intestinal homeostasis and highlights its potential as a novel therapeutic target for intestinal diseases like Crohn’s disease and ulcerative colitis.

Results

VGLL4 responds to intestinal regeneration

To identify pathways and genes involved in small intestinal regeneration, acute injury was induced by whole-body irradiation in mice, and ileum samples were collected for transcriptome sequencing (Fig. 1A). Pathway enrichment by KEGG on all differentially expressed genes revealed significant differences in the Hippo pathway between the irradiated and non-irradiated groups (Fig. 1B). Given the well-established role of the Hippo pathway in regulating the intestinal tract (Chen et al, 2012; Deng et al, 2022; Yu et al, 2015), altered genes related to this pathway were interrogated (Fig. 1C). Among the differentially expressed genes, VGLL4 was of particular interest, showing a fluctuating expression pattern with a decrease on day 1 and an increase on day 3 at the transcriptional level (Fig. 1C,D). A similar trend was observed at the protein level, albeit with a 1-day delay (Fig. 1E). To extend this finding, we used the dextran sodium sulfate (DSS) model to induce intestinal damage and inflammation (Troy et al, 2009). Expectedly, VGLL4 expression was downregulated in the acute phase (Figs. 1F–H and  EV1A). Additionally, VGLL4 expression exhibited a downward trend in patient samples with acute inflammation (Fig. EV1B,C). Validation using online public databases (GSE9452, GSE6731 and GSE59071) revealed that VGLL4 was significantly decreased in ulcerative colitis (UC) and inflamed patient tissues compared to normal controls (Fig. EV1D) (Olsen et al, 2009; Vanhove et al, 2015; Wu et al, 2007). These data suggest that VGLL4 expression in the small intestine is associated with acute inflammatory progression.

Figure 1. VGLL4 is expressed in small intestinal crypts and participates in damage repair.

(A) Schematic graph depicts the workflow of wild-type mice receiving 10 Gy irradiation followed by ileum samples harvest for RNA sequencing. (B) KEGG enrichment of differentially expressed genes in the ileum samples in (A). Hypergeometric distribution test, P values adjusted by Benjamini–Hochberg (clusterProfiler package). (C) Heatmap analysis shows changes in the Hippo pathway members after irradiation (4 mice per group). Hypergeometric distribution test. (D) Vgll4 transcripts per million (TPM) in C (4 mice per group). Unpaired Student’s t test. *P = 0.0215. **P = 0.0013. ^nsP = 0.0847. (E) Immunoblot analysis of VGLL4 in ileum samples after irradiation at indicated days. Samples were derived from the same experiment and blots were processed in parallel. (F) Schematic graph depicts the workflow of wild-type mice treated with 2% DSS followed by ileum samples harvest for qRT-PCR and western blot. (G) Immunoblot analysis of VGLL4 in DSS-fed mice ileum samples. (H) qRT-PCR analysis of Vgll4 mRNA levels (3 mice per group). One-way ANOVA, *P = 0.0367. **P = 0.0077. (I) Immunofluorescence staining (IF) for VGLL4 (green) in wild-type mice ileum samples. DAPI, blue. Crypts are indicated by white dotted lines. Scale bars, 30 μm. (J) A model for constructing Vgll4-3*HA-CreERT2-eGFP mice. Exon, pink. (K) Dual-IF for VGLL4 (green) and HA (gray) in crypts of Vgll4-3*HA-CreERT2-eGFP mice. Scale bars, 20 μm. (L) Dual-IF for HA (gray) and Lysozyme (green) in crypts of Vgll4-3*HA-CreERT2-eGFP mice. Arrows represent co-expression signal in the same cell. Scale bars, 10 μm. (M) Dual-IF for HA (gray) and OLFM4 (green) in crypts of Vgll4-3*HA-CreERT2-eGFP mice. Arrows represent co-expression signal in the same cell. Scale bars, 10 μm. Results are shown as mean + SD. ns no significance. Source data are available online for this figure.

Figure EV1. VGLL4 is downregulated in patient samples and expressed in crypt cells.

(A) Changes in mice body weight during DSS treatment (4 mice per group). (B) Immunohistochemistry (IHC) staining of VGLL4 in healthy controls and acute inflammatory patient samples. The red boxes in the left panels denote the target area for magnification in the right panels. Scale bars, 20 μm. (C) Statistical analysis of VGLL4 staining in healthy control (HD) (n = 5) and patients (n = 9) small intestinal samples. IRS, immunoreactive score. Wilcoxon rank-sum test. ***P = 0.0010. Results are shown as mean ± SD. (D) GEO analysis of VGLL4 mRNA levels in controls and patient samples. Wilcoxon rank-sum test. GSE9452: **P = 0.0062. GSE6731: *P = 0.0286. GSE59071: **P = 0.0033. Results are shown as Min to Max. Center line: median (50th percentile). Box bounds: 25th (lower) percentile and 75th (upper) percentile. Whiskers: from box bounds to min and max. (E) IF of VGLL4 in wild-type mouse small intestine organoids. DAPI, blue. Scale bars, 30 μm. (F) Double IF of HA (gray) and SOX9 (green) in Vgll4-3*HA-CreERT2-eGFP mice small intestine. DAPI, blue. Scale bars, 30 μm. (G) Double IF of HA (gray) and Ki67 (green) in Vgll4-3*HA-CreERT2-eGFP mice small intestine. DAPI, blue. Scale bars, 40 μm. (H) IF of VGLL4 (gray) and RFP (red) in small intestinal organoids from Vgll4-3*HA-CreERT2-eGFP, AI9^+/- mice injected with TAM for 4 weeks. DAPI, blue. Scale bars, 40 μm. (I) IF of RFP (red) in Vgll4-3*HA-CreERT2-eGFP, AI9^+/- mice injected with TAM for 4 weeks. DAPI, blue. Scale bars, 30 μm. (J) Model of Vgll4-3*HA-CreERT2-eGFP, mTmG^+/- mouse construction strategy and small intestinal sample harvest strategy. (K) IF of VGLL4 (gray), tdTomato (purple) and GFP (green) in Vgll4-3*HA-CreERT2-eGFP, mTmG^+/- mouse small intestinal samples after TAM injection for a week. DAPI, blue. Scale bars, 25 μm. Source data are available online for this figure.

VGLL4 is mainly localized in the crypts of the small intestine

To determine the cellular localization of VGLL4 in the gut, immunostaining revealed that VGLL4 was predominantly expressed in crypt cells, both in vivo and in intestinal organoids (Figs. 1I and  EV1E). To confirm this pattern, we generated a VGLL4 reporter mouse line (Fig. 1J), in which HA tag, CreERT2, and eGFP were inserted at the end of the exon 5 of Vgll4. Immunostaining using antibodies against HA and VGLL4 confirmed a consistent expression pattern (Fig. 1K). Co-immunostaining against the Paneth cell marker Lysozyme and the stem cell marker Olfactomedin-4 (OLFM4) revealed that Vgll4-3*HA was expressed in both Paneth cells and stem cells (Fig. 1L,M). Furthermore, VGLL4 co-localized with the progenitor cell marker SOX9 and the proliferation marker Ki67 in the crypts (Fig. EV1F,G), indicating that VGLL4-positive cells may contribute to maintaining intestinal integrity. To explore the potential role of VGLL4-positive cells, we crossed the Vgll4-3*HA-CreERT2-eGFP mice with Ai9+ /- reporter mice. Following 4 weeks of tamoxifen (TAM) administration, nearly all epithelial cells in both organoids and intestinal samples exhibited RFP positivity, a marker denoting VGLL4-expressing cells and their descendant progeny (Fig. EV1H,I). Collectively, these observations demonstrate that VGLL4-positive cells could generate all lineages of the intestinal epithelium. To verify the functionality of CreERT2, we also crossed the Vgll4-3*HA-CreERT2-eGFP mice with mT/mG reporter mice (Fig. EV1J). After TAM injection for 1 week, GFP signals were observed along the villus, indicating the remarkable efficiency of Vgll4-CreERT2 (Fig. EV1K).

VGLL4 ablation in epithelial cells significantly impairs intestinal regenerative ability

Our previous work indicated that systemic Vgll4 knockout caused early postnatal mortality in mice, with a few surviving beyond 8 weeks (Yu et al, 2019). Analysis of the small intestines from postnatal 5-day-old Vgll4^KO mouse (Fig. EV2A) and 8-week-old adult (Fig. EV2B) revealed a tendency toward reduced intestinal length. Histological examination via hematoxylin and eosin (H&E) staining in surviving adult Vgll4^KO mouse exhibited decreased villus length compared with wild-type control (Fig. EV2C). To investigate VGLL4’s role in the small intestine and avoid high mortality, we generated an intestinal epithelial cell-specific Vgll4 knockout mouse line (Vgll4^IEC-KO) by crossing Vgll4^fl/fl mice with Villin^cre mice, and confirmed Vgll4 knockout efficiency at transcriptional and protein levels (Figs. 2A,B and EV2D). Vgll4^IEC-KO mice were born at expected Mendelian ratios (Fig. EV2E) and grew normally, with no overt changes in body weight or intestinal structure (Fig. EV2F–I). Further H&E staining showed no differences in villus length in the duodenum, jejunum or ileum (Fig. EV2J,K). E-Cadherin and ZO-1 staining showed the cell–cell junctions of the small intestine were not affected (Fig. EV2L,M). However, VGLL4 deficiency impaired intestinal regenerative capacity. Vgll4^IEC-KO mice showed reduced organoid formation in vitro (Fig. 2C,D) and exhibited exacerbated structural damage and shorter villi after 10 Gy irradiation for 5 days (Fig. 2E–G). They also displayed more severe symptoms in the DSS-induced colitis model, as reflected by greater body weight loss (Fig. 2H). Although the overall length of the small intestine showed no significant change (Fig. 2I,J), morphological analysis revealed shorter villi length in the ileum (Fig. 2K,L), along with a reduction in colon length (Fig. 2M,N) and more severe edema and inflammatory infiltration (Fig. 2O). VGLL4 deficiency exacerbated DSS-induced injury in both the small intestine and colon, further supporting an essential role for VGLL4 in intestinal regeneration. These data indicate that while intestinal epithelial cell-specific VGLL4 ablation does not alter small intestinal structure, it is indispensable for maintaining intestinal homeostasis and regeneration in response to injury and inflammation.

Figure EV2. Loss of VGLL4 does not affect the morphology of the small intestine.

(A) Small intestine length of 5-day-old wild-type mouse and Vgll4 ^-/- mouse. Scale bar, 1 cm. (B) Small intestine length of 8-week-old wild-type, Vgll4 ^+/- and Vgll4 ^-/- mice. Scale bar,1 cm. (C) H&E staining of the small intestine and colon in wild-type and Vgll4 ^−/− mice. SI, small intestine. Scale bars, 300 μm. (D) IF of VGLL4 (red) in Vgll4^fl/fl and Vgll4^IEC-KO mice crypts. DAPI, blue. Scale bars, 20 μm. (E) Statistical analysis of Vgll4-knockout-mice pups genotypes. (F) Outlook of 8-week-old Vgll4^fl/fl and Vgll4^IEC-KO mice. Scale bar, 1 cm. (G) Body weight analysis of Vgll4^fl/fl and Vgll4^IEC-KO mice at 8 weeks (n = 8). (H) 8-week-old gut morphology of Vgll4^fl/fl and Vgll4^IEC-KO mice. Scale bar, 1 cm. (I) Quantification of small intestine length in (H) (n = 3). (J) H&E staining of the small intestine in Vgll4^fl/fl and Vgll4^IEC-KO mice. Scale bars, 300 μm. (K) Quantification of villus length in different regions of small intestine in (J) (n = 4–5). (L) IF of E-Cadherin (red) in Vgll4^fl/fl and Vgll4^IEC-KO mice ileum samples. DAPI, blue. Scale bars, 30  μm. (M) IF of ZO-1 in Vgll4^fl/fl and Vgll4^IEC-KO mice ileum samples. DAPI, blue. Scale bars, 30 μm. (N) Alkaline Phosphatase Staining (APS) showed enterocytes of Vgll4^fl/fl and Vgll4^IEC-KO mice. Scale bars, 50 μm. Results are shown as mean + SD. ns: no significance. Unpaired Student’s t test. Source data are available online for this figure.

Figure 2. Loss of VGLL4 impairs intestinal regeneration ability.

(A) qRT-PCR analysis of Vgll4 mRNA levels in Vgll4^fl/fl and Vgll4^IEC-KO mice’s ileum samples (n = 5). Unpaired Student’s t test. ****P < 0.0001. (B) Immunoblot analysis of VGLL4 in ileum samples from Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 3). (C, D) Representative images (C) and quantification (D) of small intestinal organoids from Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 5). Scale bars, 50 μm. Unpaired Student’s t test in (D), ****P < 0.0001. (E) H&E staining of the small intestine from Vgll4^fl/fl and Vgll4^IEC-KO mice after 10 Gy irradiation for 5 days. Scale bars, 300 μm. (F, G) Quantification of villus length in different regions of the small intestine from Vgll4^fl/fl and Vgll4^IEC-KO mice at 3 days (F) and 5 days (G) post irradiation (n = 3–4). Unpaired Student’s t test. Day 3 in duodenum: ^nsP = 0.8519. Day 3 in jejunum: ^nsP = 0.4479. Day 3 in ileum: ^nsP = 0.8275. Day 5 in duodenum: ^nsP = 0.6801. Day 5 in jejunum: **P = 0.0076. Day 5 in ileum: **P = 0.0070. (H) Body weight analysis of Vgll4^fl/fl and Vgll4^IEC-KO mice during DSS treatment (n = 6–8). Unpaired Student’s t test. ***P = 0.0002. (I) Representative image of small intestines from Vgll4^fl/fl and Vgll4^IEC-KO mice after DSS treatment. Scale bar, 1 cm. (J) Quantification of small intestine length after DSS (n = 6–8). Unpaired Student’s t test. ^nsP = 0.1536. (K) H&E staining of the small intestines from Vgll4^fl/fl and Vgll4^IEC-KO mice. Scale bars in left panel, 1 mm. Scale bars in right panel, 50 μm. (L) Quantification of small intestinal villi length after DSS (n = 6–8). Unpaired Student’s t test. Duodenum: ^nsP = 0.1905. Jejunum: ^nsP = 0.1915. Ileum: **P = 0.0020. (M) Representative image of colons from Vgll4^fl/fl and Vgll4^IEC-KO mice after DSS treatment. Scale bar, 1 cm. (N) Quantification of colon length after DSS (n = 6–8). Unpaired Student’s t test. ***P = 0.0001. (O) H&E staining of the colon samples from Vgll4^fl/fl and Vgll4^IEC-KO mice. Scale bars, 200 μm. Results are shown as mean + SD. ns no significance. Source data are available online for this figure.

Paneth cell dysfunction in Vgll4^IEC-KO mice

To investigate the mechanism of the reduced regenerative ability associated with VGLL4 loss, RNA sequencing was performed on ileum samples from 8-week-old Vgll4^IEC-KO and Vgll4^fl/fl mice to explore altered signaling pathways. Innate immune responses in mucosa and antimicrobial peptides were significantly enriched (Fig. 3A), prompting a focus on Paneth cells, which play a key role in the innate immune defense of the small intestinal epithelium. Immunostaining for Paneth cell markers Lysozyme and defensin 5 (DEFA5) confirmed a decrease in mature Paneth cells in Vgll4^IEC-KO mice (Fig. 3B–E). Fluorescence-activated cell sorting (FACS) analysis also showed a reduction in the proportion of Paneth cells (Fig. 3F,G). A similar reduction in Lysozyme protein level was observed in the crypt samples from Vgll4^IEC-KO mice (Fig. 3H). For a comprehensive assessment of broader epithelial effects, we examined the proportion of other epithelial cell subtypes. Enterocytes, indicated by Alkaline Phosphatase Staining (APS), remained unchanged (Fig. EV2N). While goblet cell numbers, confirmed by mucin 2 (MUC2) staining, were unaffected (Fig. EV3A,B). Chromogranin A (CHGA) staining confirmed the enteroendocrine cells (EECs) proportion did not change (Fig. EV3C,D). These results suggest that VGLL4 specifically affects the number of Paneth cells.

Figure 3. Severe alteration of Paneth cells in VGLL4-deficient intestinal epithelium.

(A) Gene set enrichment analysis of differentially expressed genes during irradiation. Hypergeometric distribution test, P values adjusted by Benjamini–Hochberg (clusterProfiler package). (B) IF staining of Lysozyme (green) in Vgll4^fl/fl and Vgll4^IEC-KO mice. EpCAM, gray. DAPI, blue. Scale bars, 20 μm. (C) Quantification of Lysozyme⁺ Paneth cells in Vgll4^fl/fl and Vgll4^IEC-KO mice crypts (n = 6). Unpaired Student’s t test. ****P < 0.0001. (D) IF staining of DEFA5 (green) in Vgll4^fl/fl and Vgll4^IEC-KO mice. EpCAM, gray. DAPI, blue. Scale bars, 20 μm. (E) Quantification of DEFA5⁺ Paneth cells in Vgll4^fl/fl and Vgll4^IEC-KO mice crypts (n = 5). Unpaired Student’s t test. ****P < 0.0001. (F) Flow cytometry analysis of Paneth cells sorted from Vgll4^fl/fl and Vgll4^IEC-KO mice. (G) Quantification of Paneth cell proportion in Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). Unpaired Student’s t test. ***P = 0.0003. (H) Immunoblot analysis of VGLL4 and Lysozyme in small intestinal crypt samples of Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 3). Samples were derived from the same experiment and blots were processed in parallel. (I) Representative pictures of organoids derived from wild-type Lgr5⁺ stem cells (WT Lgr5⁺) or Vgll4-knockout Lgr5⁺ stem cells (VGLL4-KO Lgr5⁺), following 5 days of co-culture with Paneth cells isolated from Vgll4^fl/fl and Vgll4^IEC-KO mice. Scale bars, 100 μm. (J) Quantification of organoid number in (I) (n = 6). One-way ANOVA. *P = 0.0289. ****P < 0.0001. (K) Quantification of organoid diameter in (I) (n = 6). One-way ANOVA. ^nsP = 0.3386. **P = 0.0051. ****P < 0.0001. Results are shown as mean + SD. ns no significance. Source data are available online for this figure.

Figure EV3. Analysis of small intestinal morphology, cell differentiation, proliferation and cell death.

(A) IF of MUC2 (green) in Vgll4^fl/fl and Vgll4^IEC-KO mice ileum samples. DAPI, blue. Scale bars, 50 μm. (B) Quantification of MUC2⁺ cells per villus in (A) (n = 3). (C) IF of CHGA (red) in Vgll4^fl/fl and Vgll4^IEC-KO mice ileum samples. DAPI, blue. Scale bars, 100 μm. (D) Quantification of CHGA⁺ cells per villus in (C) (n = 4). (E) IF of Ki67 (red) in Vgll4^fl/fl and Vgll4^IEC-KO mice crypts. DAPI, blue. Scale bars, 10  μm. (F) Quantification of Ki67⁺ cells per crypt in (E) (n = 3–4). (G) Model of Control (Lgr5-EGFP-IRES-creERT2, Ai9 + /−) mouse line and ISC-KO (Lgr5-EGFP-IRES-creERT2, Vgll4^{fl/-;-LSL-RFP}) mouse line construction strategies and small intestinal sample harvest strategy. (H) IF of VGLL4 (gray), GFP (green), RFP (purple) and DAPI (blue) in Control and ISC-KO mouse small intestinal samples after TAM injection for a week. Scale bars, 30 μm. (I) TUNEL staining in Vgll4^fl/fl and Vgll4^IEC-KO mice. Scale bars, 30 μm. (J) Quantification of TUNEL-positive cells per crypt-villus structure in (I) (n = 4). (K) IF of cleaved caspase-3 (red) in Vgll4^fl/fl and Vgll4^IEC-KO mice. Scale bars, 30 μm. (L) Quantification of cleaved caspase-3-positive cells per villus in (K) (n = 3). Results are shown as mean ± SD. ns no significance. Unpaired Student’s t test. Source data are available online for this figure.

Given that Paneth cells maintain a niche for intestinal stem cells (Sato et al, 2011), a co-culture system was employed to assess the niche-forming capacity of Paneth cells and further investigate whether VGLL4 knockout impacts intestinal stem cell function. Four groups were compared: WT Lgr5⁺ (wild-type Lgr5⁺ stem cells) with WT PCs (wild-type Paneth cells), WT Lgr5⁺ with KO PCs (Vgll4-knockout Paneth cells), KO Lgr5⁺ (Vgll4-knockout Lgr5⁺ stem cells) with WT PCs, KO Lgr5⁺ with KO PCs (Fig. 3I–K). WT Lgr5⁺ cells co-cultured with WT PCs generated organoids with the greatest number and largest diameter, while KO Lgr5⁺ cells co-cultured with KO PCs formed the fewest and smallest organoids. Compared to the WT-WT co-culture group, Vgll4-KO Lgr5⁺ cells paired with WT PCs exhibited a slight reduction in organoid number and a tendency toward smaller diameters, implying that VGLL4 potentially regulates stem cells. Notably, co-culture of WT Lgr5⁺ cells with Vgll4-KO PCs formed smaller organoids than those formed by Vgll4-KO Lgr5⁺ cells with WT PCs, indicating that VGLL4 deletion in Paneth cells exerts a more pronounced effect on organoid formation than in Lgr5⁺ stem cells, thereby prompting us to focus on the role of VGLL4 in Paneth cells. Consistently, WT Lgr5⁺ stem cells co-cultured with Vgll4-knockout PCs formed fewer and smaller organoids compared to those co-cultured with WT PCs, substantiating the notion that VGLL4 loss impairs the niche-forming ability of Paneth cells, thus contributing to impaired intestinal regeneration.

TEAD4 acts as a scaffold to link VGLL4 and ATOH1 to regulate GFI1 expression

To elucidate the mechanism underlying the reduction in Paneth cell numbers upon VGLL4 loss, we initially investigated cell proliferation and apoptosis. Ki67 staining indicated that cell proliferation in the crypts was unchanged in Vgll4^IEC-KO mice compared with Vgll4^fl/fl controls (Fig. EV3E,F). To rule out a direct role of VGLL4 in stem cells, a stem cell-specific Vgll4-knockout (ISC-KO) mouse line was generated by crossing our previously established Vgll4^{fl/-;-LSL-RFP} mouse line (Yu et al, 2019) with Lgr5-EGFP-IRES-creERT2 mice. The resulting ISC-KO (Lgr5-EGFP-IRES-creERT2, Vgll4^{fl/-;-LSL-RFP}) mice were compared to Control (Lgr5-EGFP-IRES-creERT2, Ai9+/-) mice. Both groups received TAM injections, and small intestinal tissues were collected 7 days later (Fig. EV3G). IF staining confirmed knockout efficiency in the crypts (Fig. EV3H), where Lgr5-creERT2 activity-driven RFP expression was accompanied by a loss of VGLL4 signal in ISC-KO crypts, indicating successful colonization by VGLL4-negative stem cells. We next examined whether VGLL4 deletion in ISCs affected cell proliferation. No statistically significant difference was observed in the number of Ki67⁺ cells in RFP-positive crypts between Control and ISC-KO mice (Fig. 4A,B), suggesting that VGLL4 loss does not impair the proliferative capacity of ISCs in vivo. In contrast, the proportion of Paneth cells was significantly reduced in VGLL4-deficient crypts (Fig. 4A,B), supporting the hypothesis that VGLL4 is essential for Paneth cell maintenance. Together, these results indicate that conditional knockout of VGLL4 in intestinal stem cells does not affect their proliferation under homeostatic conditions but leads to a decrease in Paneth cell numbers. Next, cell apoptosis was assessed. TUNEL assays and cleaved caspase-3 staining showed no difference in apoptosis levels in the ileum (Fig. EV3I–L). Thus, we hypothesized that impaired lineage differentiation might explain the reduced numbers of Paneth cells.

Figure 4. VGLL4 forms a complex with TEAD4-ATOH1 to regulate GFI1.

(A) IF staining of small intestinal samples from Control and ISC-specific knockout (ISC-KO) mice at 1 week after TAM injection. Top 2 panels: Staining for Ki67 (gray), GFP (green), RFP (purple) and DAPI (blue). Bottom 2 panels: Staining for Lysozyme (gray), GFP (green), RFP (purple) and DAPI (blue). Scale bars, 10 μm. (B) Quantification of Ki67⁺ cells and Lysozyme⁺ cells per crypt in (A) (n = 4). Unpaired Student’s t test. ^nsP = 0.5512. **P = 0.0281. (C) qRT-PCR analysis of Gfi1 and Sox9 mRNA levels in Vgll4^fl/fl and Vgll4^IEC-KO mice ileum samples (n = 5). Unpaired Student’s t test. ***P = 0.0008. ^nsP = 0.1057. (D) Immunoblot analysis of GFI1, SOX9, ATOH1, and VGLL4 in ileum samples of Vgll4^fl/fl and Vgll4^IEC-KO mice. Samples were derived from the same experiment and blots were processed in parallel. (E) Quantification of protein levels in (D) (n = 5). Unpaired Student’s t test. GFI1: **P = 0.0018. ATOH1: ^nsP = 0.6427. SOX9: ^nsP = 0.4877. (F) IF staining of GFI1 (red) in Vgll4^fl/fl and Vgll4^IEC-KO mice small intestinal organoids (n = 5). DAPI, blue. Scale bars, 30 μm. (G) Co-immunoprecipitation assay results between ATOH1 and TEAD4 in HEK293T cells. (H) Co-immunoprecipitation assay results between ATOH1 and TEAD4 in HEK293T cells with DNase. (I) Co-immunoprecipitation assay results between ATOH1 and different mutations of TEAD4 in HEK293T cells. (J) Co-immunoprecipitation assay results between ATOH1 and VGLL4 in HEK293T cells with or without TEAD4. (K) Immunoblot analysis of GST pull-down assay between purified ATOH1-ΔN, VGLL4, and TEAD4 proteins. Results are shown as mean + SD. ns no significance. Source data are available online for this figure.

GFI1 and SOX9 are known to be crucial for Paneth cell differentiation (Flier and Clevers, 2009), with ATOH1 activating GFI1 transcription (Lo et al, 2017). We observed decreased GFI1 at both protein and mRNA levels in Vgll4^IEC-KO mice, while ATOH1 (referred to as Math1 in mice) and SOX9 expression remained unaffected (Figs. 4C–E and EV4A). Staining of GFI1 in mouse small intestinal organoids from Vgll4^fl/fl and Vgll4^IEC-KO mice confirmed that the decrease of VGLL4 led to downregulation of GFI1 (Fig. 4F). Co-immunoprecipitation experiments showed that VGLL4 did not bind ATOH1 directly (Fig. EV4B), instead, VGLL4’s canonical binding partner, TEAD4, interacted with ATOH1 (Fig. 4G), and this interaction was independent of DNA binding, as confirmed by nuclease treatment (Fig. 4H). To gain further insights into the binding of TEAD4 and ATOH1, we mapped the crucial domains involved in this interaction. The TEA domain of TEAD4 was sufficient for binding to ATOH1 (Fig. EV4C–E). Further analysis indicated that the N-terminal of ATOH1 was dispensable for binding to TEAD4 (Fig. EV4F,G), with consistent results from in vitro protein pulldown assays (Figs. EV4H and 4K). Key residues R95, K96, S100, and Q103 within TEAD4’s TEA domain are crucial for ATOH1 binding, as mutations of all four sites (TEAD4-4M) abolished the interaction (Fig. 4I). TEAD4 could act as a scaffold to bring VGLL4 and ATOH1 together to form a complex, as indicated by co-immunoprecipitation and purified GST pull-down assay (Figs. 4J,K and EV4I).

Figure EV4. The ATOH1-TEAD4-VGLL4 complex promotes the expression of GFI1.

(A) qRT-PCR analysis of Math1 from Vgll4^fl/fl and Vgll4^IEC-KO mice ileum samples (n = 5). Unpaired Student’s t test. ^nsP = 0.7009. (B) Co-immunoprecipitation assay results between ATOH1 and VGLL4 in HEK293T cells. (C) Schematic illustration of TEAD4 and its short forms. (D) Co-immunoprecipitation assay results between ATOH1 and TEAD4 short forms in HEK293T cells. The black arrow indicates IgG heavy chain. (E) Co-immunoprecipitation assay results between ATOH1 and TEAD4-N and TEA domain short forms in HEK293T cells. (F) Schematic illustration of ATOH1 and its short forms. (G) Co-immunoprecipitation assay results between ATOH1 short forms and TEAD4 in HEK293T cells. (H) Coomassie brilliant blue staining of the GST-pulldown assay between ATOH1-ΔN, VGLL4, and TEAD4. (I) FLAG antibody pulldown assay results between VGLL4-TEAD4-ATOH1 complex in HEK293T cells overexpressing FLAG-VGLL4, MYC-ATOH1 and HA-TEAD4. (J) ChIP-qPCR analysis of TEAD4 enrichment at the GFI1 promoter in HEK293T cells overexpressing FLAG-TEAD4. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. (K) ChIP-qPCR analysis of TEAD4 at the GFI1 promoter in 293T cells with or without VGLL4 overexpression. Unpaired Student’s t test. ****P < 0.0001. Left ***P = 0.0007. Right ***P = 0.0003. Three biological replicates per group. (L) ChIP-qPCR analysis of ATOH1 at the GFI1 promoter in HEK293T cells with or without VGLL4 overexpression. Unpaired Student’s t test. Left ***P = 0.0003. Right ***P = 0.0007. Three biological replicates per group. (M) A model of GFI1 luciferase construction strategy. Red arrow: TBS, TEAD4 binding sequence. Black arrow: ABS, ATOH1 binding sequence. RR: regulatory region. (N) GFI1 Luc reporter activity in HCoEpiC cells transfected with ATOH1, VGLL4, and TEAD4. One-way ANOVA. ****P < 0.0001. Three biological replicates per group. (O) GFI1 Luc reporter activity in HCoEpiC cells transfected with ATOH1, TEAD4 wild-type or mutant form, VGLL4 wild-type or mutant form. One-way ANOVA. ****P < 0.0001. Three biological replicates per group. Results are shown as mean + SD. ns no significance. Source data are available online for this figure.

The VGLL4-TEAD4-ATOH1 complex binds to GFI1 promoter to promote transcription

Analysis of online ChIP-seq data revealed that one-third of ATOH1 target genes could also be regulated by TEAD4 (Fig. 5A). We found TEAD4 binding peaks within the promoters of ATOH1 target genes (Fig. 5B), suggesting potential cooperation between TEAD4 and ATOH1 in regulating their common targets. We therefore hypothesized that ATOH1 could form a complex with TEAD4 to regulate GFI1 transcription. ChIP-seq data showed binding peaks for both proteins at the GFI1 promoter (Fig. 5C). First, we detected the binding of TEAD4 at GFI1 promoter in both HCoEpiC and 293T cells (Figs. 5D and EV4J), and ChIP and re-ChIP experiments confirmed that TEAD4 and ATOH1 co-occupy the same locus at the GFI1 promoter as a complex (Fig. 5E). Overexpression of the VGLL4-TEAD4-ATOH1 ternary complex significantly upregulated the binding of TEAD4, ATOH1 and VGLL4 at GFI1 promoter (Fig. 5F–H), suggesting the crucial function for this complex in GFI1 transcription. Moreover, overexpression of VGLL4 alone was sufficient to strengthen the binding of TEAD4 and ATOH1 (Fig. EV4K,L), indicating the importance of VGLL4 for GFI1 transcription.

Figure 5. The VGLL4-TEAD4-ATOH1 complex promotes the transcription of GFI1.

(A) Venn analysis of target genes from TEAD4 ChIP-seq data (SRX190228) and ATOH1 ChIP-seq data (SRX5394928). (B) TEAD4 ChIP distance from ATOH1 ChIP binding center. (C) ChIP-seq data (TEAD4: SRX190228; ATOH1:SRX5394928) showing binding peaks of TEAD4 and ATOH1 at the promoter of GFI1. (D) ChIP-qPCR analysis of TEAD4 enrichment at the GFI1 promoter in HCoEpiC cells. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. (E) Two-step ChIP-qPCR analysis  showing co-enrichment of ATOH1 and TEAD4 at the GFI1 promoter in HCoEpiC cells. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. (F) ChIP-qPCR analysis of HA-TEAD4 at GFI1 promoter in 293T cells with or without VGLL4 and ATOH1 overexpression. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. (G) ChIP-qPCR analysis of MYC-ATOH1 at GFI1 promoter in 293T cells with or without VGLL4 and TEAD4 overexpression. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. (H) ChIP-qPCR analysis of SBP-VGLL4 at GFI1 promoter in 293T cells with or without ATOH1 and TEAD4 overexpression. Unpaired Student’s t test. Left **P = 0.0036. Right **P = 0.0083. ****P < 0.0001. Three biological replicates per group. (I) Analysis of luciferase reporter activity driven by GFI1 promoter in 293T cells transfected with indicated plasmids. One-way ANOVA. From left to right: Column 2: ^nsP = 0.3859. Column 3: ^nsP > 0.9999. Column 4: ***P = 0.0018. Column 5: ***P = 0.0015. Column 6: ****P < 0.0001. Column 7: ****P < 0.0001. Column 8: ****P < 0.0001. Three biological replicates per group. (J) Analysis of luciferase reporter activity driven by GFI1 promoter in 293T cells transfected with ATOH1, VGLL4, and TEAD4 wild-type or TEAD4-4M mutant. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. (K) Analysis of luciferase reporter activity driven by GFI1 promoter in 293T cells transfected with ATOH1, TEAD4, and VGLL4 wild-type or VGLL4-HF4A mutant. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. Results are shown as mean + SD. ns no significance. Source data are available online for this figure.

To further validate the regulatory role of the ATOH1-TEAD4-VGLL4 complex in GFI1 transcription, we cloned the ATOH1-TEAD4 binding site from the GFI1 promoter and fused it to the luciferase reporter plasmid (Fig. EV4M). Overexpression of ATOH1, TEAD4, or VGLL4 alone, or in pairwise combinations, led to only slight activation of the luciferase reporter. However, the ternary complex markedly upregulated the transcription of GFI1 (Fig. 5I). We validated VGLL4’s importance using a VGLL4-HF4A mutant (H212A/F213A/H240A/F241A) that abrogates its binding to TEADs (Jiao et al, 2014). Mutations in TEAD4 (TEAD4-4M) or VGLL4 (VGLL4-HF4A) abolished activation of the GFI1 reporter (Fig. 5J,K), and we confirmed these findings in HCoEpiC cells (Fig. EV4N,O). Collectively, these data suggest that the ATOH1-TEAD4-VGLL4 ternary complex is crucial for Paneth cell differentiation by activating the transcription of GFI1. Disruption of this complex, such as through VGLL4 loss, results in the decreased number of Paneth cells observed in the Vgll4^IEC-KO mice.

The VGLL4-TEAD4-TCF4 complex promotes the expression of Paneth cell antimicrobial peptides

Through transmission electron microscopy analysis, we found that Vgll4^IEC-KO mice had smaller granules inside single Paneth cells compared with the Vgll4^fl/fl controls (Fig. 6A,B). To investigate if VGLL4 regulates the expression of peptides within these granules, we assessed the expression profile of defensins and other antimicrobial proteins in Vgll4^IEC-KO and Vgll4^fl/fl mice. The expression of Paneth cell defensins was dramatically decreased in the absence of VGLL4, as demonstrated by both RNA sequencing data (Figs. 6C and EV5A) and qRT-PCR (Fig. 6D). Notably, this decrease in antimicrobial peptide expression was not simply due to a reduction in the number of Paneth cells, but also within individual Paneth cells, as the data were normalized to the number of sorted Paneth cells (Fig. 6E).

Figure 6. The VGLL4-TEAD4-TCF4 complex promotes antimicrobial peptides expression in Paneth cells.

(A) Transmission electron microscopy images of granules in Paneth cells in the ileum crypts in Vgll4^fl/fl and Vgll4^IEC-KO mice. Paneth cells are indicated by dotted lines. Scale bar, 1 μm. (B) Quantification of granule diameters in (A) (n = 5). Unpaired Student’s t test. ****P < 0.0001. (C) Heatmap analysis of defensin expression after irradiation for one day with RNA-seq data (n = 4). (D) qRT-PCR analysis of Lysozyme and Defensin mRNA levels in ileum samples from Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). Unpaired Student’s t test. ****P < 0.0001. Defa21: ***P = 0.0001. (E) qRT-PCR analysis of Lysozyme, Mmp7, GFI1, and VGLL4 in sorted Paneth cells from Vgll4^fl/fl and Vgll4^IEC-KO mice. Unpaired Student’s t test. Lysozyme: ***P = 0.0005. Gfi1: **P = 0.0034. ****P < 0.0001. Three biological replicates per group. (F) Co-immunoprecipitation assay results between TCF4, TEAD4 and VGLL4 in 293T cells. (G) ChIP-qPCR analysis of TCF4 enrichment at the DEFA5 promoter in HCoEpiC cells. Unpaired Student’s t test. *P = 0.0277. Three biological replicates per group. (H) ChIP-qPCR analysis of TEAD4 enrichment at the DEFA5 promoter in HCoEpiC cells. Unpaired Student’s t test. *P = 0.0109. Three biological replicates per group. (I) ChIP-qPCR analysis of VGLL4 enrichment at the DEFA5 promoter in HCoEpiC cells. Unpaired Student’s t test. ***P = 0.0002. Three biological replicates per group. (J) ChIP-qPCR analysis of FLAG-TCF4 enrichment at DEFA5 promoter in 293T cells with or without VGLL4 and TEAD4 overexpression. Unpaired Student’s t test. *P = 0.0174. **P = 0.0011. ***P = 0.0003. Three biological replicates per group. (K) A model of the construction of DEFA5 and DEFA6 promoter driven luciferase. (L) DEFA6 Luc reporter activity in 293T cells transfected with indicated plasmids. One-way ANOVA. From left to right: Column 2: ^nsP = 0.0959. Column 3: ****P < 0.0001. Column 4: ^nsP = 0.4546. Column 5–8: ****P < 0.0001. Three biological replicates per group. (M) DEFA6 Luc reporter activity in 293T cells transfected with TCF4, VGLL4 and TEAD4 wild-type or TEAD4-TEA short form. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. (N) DEFA6 Luc reporter activity in 293T cells transfected with TCF4, TEAD4, and VGLL4 wild-type or VGLL4-HF4A mutant form. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. (O) DEFA6 Luc reporter activity in HCoEpiC cells transfected with indicated plasmids. One-way ANOVA. From left to right: Column 2: ^nsP = 0.8371. Column 3: ^nsP = 0.6384. Column 4: ^nsP = 0.1015. Column 5: ***P = 0.0003. Column 6: ***P = 0.0003. Column 7–8: ****P < 0.0001. Three biological replicates per group. (P) DEFA6 Luc reporter activity in HCoEpiC cells transfected with TCF4, TEAD4 wild-type or TEAD4-TEA short form, and VGLL4 wild-type or VGLL4-HF4A mutant form. Unpaired Student’s t test. ****P < 0.0001. Three biological replicates per group. Results are shown as mean + SD. Source data are available online for this figure.

Figure EV5. The TCF4-TEAD4-VGLL4 complex upregulates defensin expression.

(A) Heatmap analysis of defensin expression with RNA-seq data (n = 4). (B) ChIP-qPCR analysis of TCF4 enrichment at the DEFA5 promoter in SW620 cells. Unpaired Student’s t test. ***P = 0.0008. Three biological replicates per group. (C) ChIP-qPCR analysis of TEAD4 enrichment at the DEFA5 promoter in SW620 cells. Unpaired Student’s t test. ***P = 0.0003. Three biological replicates per group. (D) Luciferase reporter activity driven by DEFA5 promoter in HEK293T cells transfected with indicated plasmids. One-way ANOVA. ****P < 0.0001. Three biological replicates per group. (E) Luciferase reporter activity driven by DEFA5 promoter in HEK293T cells transfected with TCF4, VGLL4 and TEAD4 wild-type or TEAD4-TEA short form. One-way ANOVA. ****P < 0.0001. ***P = 0.0004. Three biological replicates per group. (F) Luciferase reporter activity driven by DEFA5 promoter in HEK293T cells transfected with TCF4, TEAD4 and VGLL4 wild-type or VGLL4-HF4A short form. One-way ANOVA. ****P < 0.0001. Three biological replicates per group. (G) Luciferase reporter activity driven by DEFA5 promoter in HCoEpiC cells transfected with indicated plasmids. One-way ANOVA. ****P < 0.0001. **P = 0.0025. Three biological replicates per group. (H) Luciferase reporter activity driven by DEFA5 promoter in HCoEpiC cells transfected with TCF4, TEAD4 wild-type or TEAD4-TEA short form, and VGLL4 wild-type or VGLL4-HF4A short form. One-way ANOVA. ****P < 0.0001. Three biological replicates per group. Results are shown as mean + SD. Source data are available online for this figure.

Previous studies have implicated the role of Wnt/TCF4 signaling pathway in the regulation of defensin expression (van Es et al, 2005; Wehkamp et al, 2007). We hypothesized that VGLL4 might participate in defensin expression regulation, given its interaction with the TEAD4-TCF4 complex in inhibiting colon cancer (Jiao et al, 2017). We confirmed the formation of the TCF4-TEAD4-VGLL4 complex by co-immunoprecipitation (Fig. 6F). Next, ChIP assays were performed to detect the binding of this complex to the defensin promoter. Results showed that TCF4, TEAD4, and VGLL4 could be detected at the DEFA5 promoter region (Figs. 6G–I and EV5B,C). Furthermore, overexpression of the TCF4-TEAD4-VGLL4 complex enhanced the binding of TCF4 to the DEFA5 promoter (Fig. 6J).

Next, two luciferase reporters for DEFA5 and DEFA6 promoters were constructed to evaluate the transcription regulation of the TCF4-TEAD4-VGLL4 complex to defensins (Fig. 6K). We observed maximal activation of reporter activity when all three proteins were co-expressed (Figs. 6L and  EV5D), illustrating the critical role of the TCF4-TEAD4-VGLL4 complex in regulating antimicrobial peptide expression in Paneth cells. Further mechanistic studies revealed that overexpression of the TEA domain of TEAD4, which is necessary for interacting with TCF4 but dispensable for VGLL4 binding, led to a decrease in the luciferase activity (Figs. 6M and  EV5E). This finding was consistent with results obtained using the VGLL4-HF4A mutant, which also abolished the upregulation of luciferase activity (Figs. 6N and  EV5F). Similar results were observed in the HCoEpiC cell line, which exhibited even greater luciferase activity (Figs. 6O,P and  EV5G,H). Collectively, these data highlight the indispensable role of the VGLL4-TEAD4-TCF4 transcriptional complex in maintaining Paneth cell function and intestinal innate immunity by upregulating the expression of antimicrobial peptides. In summary, TEAD4 acts as a scaffold that facilitates the interaction between VGLL4 and TCF4, thereby promoting the transactivation of defensin expression. Disruption of this complex, such as loss of VGLL4, contributes to the observed defects in Paneth cell antimicrobial peptide production and granule size.

VGLL4 is necessary for the maintenance of intestinal microbiota homeostasis

Given the decrease in defensin expression upon VGLL4 loss, we focused on the role of VGLL4 in safeguarding the host against diverse antigens and regulating the enteric microbial community. Fecal samples from Vgll4^IEC-KO mice and Vgll4^fl/fl littermates were collected for 16S rRNA sequencing (Fig. EV6A). Approximately 70% of the amplicon sequence variants (ASVs) were shared between the two groups (Fig. EV6B). Community bar plot analysis revealed alterations in the community composition at the family level (Fig. EV6C), with genus-level abundance shown in Fig. EV6D. Shannon index (P = 0.4712) and Simpson index (P = 0.6889) analyses indicated no significant change in microbiota diversity (Fig. EV6E,F). However, the composition was disrupted according to Principal Coordinates Analysis (PCoA) (R = 0.4796, P = 0.003), Nonmetric Multidimensional Scaling (NMDS) (R = 0.7222, P = 0.003), and Principal Component Analysis (PCA) (R = 0.2778, P = 0.0190) (Fig. EV6G–I). Linear Discriminant Analysis Effect Size (LEfSe) analysis revealed that Firmicute and Bacteroidota were the main altered phyla (LDA > 4) (Fig. EV6J,K). At the genus level, Muribaculaceae increased significantly, whereas Turicibacter decreased after VGLL4 deletion (Fig. EV6L). These findings indicate that the loss of VGLL4 in IECs disrupts the composition of the intestinal microbiota, thus altering the microenvironment.

Figure EV6. VGLL4 deficiency leads to gut microbiota changes.

(A) Rarefaction curves constructed based on observed ASVs of fecal samples from Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). (B) Venn analysis based on observed ASVs of fecal samples from Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). (C) Community bar plot analysis at the family level between Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). (D) Heatmap analysis at the family level between Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). (E) Shannon index at the phylum level in Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). Wilcoxon rank-sum test. (F) Simpson index at the phylum level between Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). Wilcoxon rank-sum test. (E, F) Results are shown as Min to Max. Center line: median (50th percentile). Box bounds: 25th (lower) percentile and 75th (upper) percentile. Whiskers: from box bounds to min and max. (G) Principal coordinates analysis (PCoA) at the species level between Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). ANOSIM test for differences in the coefficient between the two groups. (H) Nonmetric multidimensional scaling (NMDS) analysis at the ASV level between Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). ANOSIM test for difference in the coefficient between two groups. (I) Principal Component Analysis (PCA) at the ASV level between Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 6). ANOSIM test for difference in the coefficient between two groups. (J) Linear discriminant analysis (LDA) identified the significantly abundant genera in different groups (LDA score >4). (K) LEfSe plot results from phylum to genus. (L) Wilcoxon rank-sum test bar plot at the genus level showing the differentially abundant microbiota in Vgll4^fl/fl and Vgll4^IEC-KO mice (n = 5–6). Error bars in the right panel represent 95% confidence intervals around the mean difference in proportions.

Discussion

Intestinal epithelial cells are crucial for maintaining intestinal homeostasis, and numerous studies have explored their differentiation and maintenance mechanisms. VGLL4, known for its multiple functions in tissue development and regeneration, but its role in the small intestine has not been fully validated. This study reveals that VGLL4 is predominantly expressed in crypt cells, and Vgll4^IEC-KO mice exhibit a reduced proportion of Paneth cells and impaired repair ability in response to intestinal damage. VGLL4 promotes Paneth cell differentiation through forming a complex with TEAD4 and ATOH1, and it sustains gut microbiota homeostasis by promoting Paneth cell AMPs expression. Our findings highlight VGLL4’s novel roles in maintaining small intestinal homeostasis, improving our understanding of diseases related to Paneth cell dysfunction and intestinal dysbiosis.

The Hippo signaling pathway, which precisely controls organ development and tissue homeostasis (Misra and Irvine, 2018; Yu and Guan, 2013), plays an important role in intestinal homeostasis. Loss of Mst1/2 in small intestinal epithelial cells results in the expansion of undifferentiated cells and the absence of the secretory lineage (Zhou et al, 2011). Although deletion of YAP has little effect on intestinal epithelial cells under normal conditions, deleting a single YAP allele can reverse the Mst1/2 loss phenotype (Zhou et al, 2011). YAP overexpression can lead to an expansion of undifferentiated progenitor cells (Camargo et al, 2007). The comparable Ki67⁺ signals within crypts of Vgll4^IEC-KO and Vgll4^fl/fl mice (Figs. EV3E,F and 4A,B) indicate that the proliferative ability in crypts is unaffected, implying YAP is likely not engaged in Paneth cell differentiation upon VGLL4 loss, which is consistent with the previous observations that YAP is inactivated under normal homeostasis (Cai et al, 2010). VGLL4’s expression is committed to Paneth cell differentiation rather than progenitor proliferation, warranting further investigation into YAP’s potential involvement.

The small intestinal epithelium consists of six cell types: absorptive enterocytes, mucus-secreting goblet cells, antigen-transporting M cells, AMP-producing Paneth cells, hormone-producing enteroendocrine cells (EECs), and chemosensory tuft cells (Beumer and Clevers, 2020). Cell fate determination initiates with various signals, such as Wnt, Notch, and EGF (Takahashi and Shiraishi, 2020). ATOH1 drives differentiation towards secretory progenitors, producing goblet cells, Paneth cells, and EECs (Yang et al, 2001). ATOH1 is a direct target of the Notch pathway, and its expression is repressed by HES1 (Fre et al, 2005; Suzuki et al, 2005). The negligible changes in ATOH1 indicate that VGLL4 signaling through ATOH1 is independent of the Notch pathway (Figs. 4D,E and  EV4A). GFI1, the bona fide target of ATOH1, commonly drives Paneth cell and goblet cell differentiation (Shroyer et al, 2005). In our study, diminished GFI1 expression upon VGLL4 loss blunts Paneth cell differentiation, while the differentiation of goblet cell lineage was not affected (Fig. EV3A,B). This indicates additional mechanisms potentially involved (Ghaleb et al, 2008; Noah et al, 2010), because only a partial reduction in Paneth and goblet cells is observed in Gfi1-deficient mice (Shroyer et al, 2005). VGLL4 knockout in intestinal crypts did not affect crypt cell proliferation in vivo (Fig. EV3E,F); however, the finding that co-culture of VGLL4-deficient Lgr5⁺ cells with WT PCs formed a smaller number of organoids in vitro suggests functional compensation by the niche microenvironment for the loss of VGLL4 in maintaining intestinal stem cell proliferation.

Paneth cells require both GFI1 and SOX9 for further specification, and GFI1 functions downstream of ATOH1 (Cui et al, 2023). Co-immunoprecipitation and GST pull-down assay confirm that TEAD4 scaffolds ATOH1 and VGLL4, enabling ATOH1 to bind GFI1’s promoter and stimulate its expression, resulting in the differentiation of Paneth cells. Although VGLL4 lacks a DNA-binding domain, it enhances GFI1 transcription via TEAD4. However, ATOH1/TEAD4 assembly is not affected regardless of VGLL4. How VGLL4 cooperates with ATOH1/TEAD4 to selectively transactivate GFI1 during Paneth cell differentiation remains to be clarified, suggesting the potential involvement of additional regulator(s) or epigenetic events. Our study identifies VGLL4 as a regulator of Paneth cell differentiation, linking Hippo signaling to intestinal homeostasis. More broadly, we demonstrate that the VGLL4–TEAD4 interaction serves as a molecular platform for recruiting the pro-differentiation factor ATOH1, thereby redirecting TEAD4 from a driver of proliferation to an initiator of cellular maturation. This represents a paradigm shift in understanding TEAD4 function, revealing that its transcriptional output is not fixed but is dynamically determined by the specific co-factors it recruits.

Paneth cell-derived AMPs serve multiple roles in the small intestine, including regulating gut microbiota and inflammation (Lueschow and McElroy, 2020; Mei et al, 2020). While AMPs sustain the stability of the intestinal microenvironment and facilitate pathogen clearance, they can also promote infection in certain contexts—for instance, by enhancing bacterial adhesion and invasion (Xu et al, 2018). Maintaining balanced AMP levels is therefore critical for preserving intestinal homeostasis. Previously, studies have revealed a correlation between the Wnt pathway and the expression of defensins (Beisner et al, 2014; van Es et al, 2005; Wehkamp et al, 2007). Given VGLL4’s role in inhibiting the Wnt pathway in colon cancer by targeting the TEAD4-TCF4 complex (Jiao et al, 2017), there is a growing interest in understanding VGLL4’s role in defensin regulation. Our findings suggest that VGLL4 interacts with TEAD4 and TCF4 to induce defensin expression, reinforcing the notion that VGLL4 is a versatile cofactor that controls target genes contextually.

One important role of gut defensins is to maintain gut microbiota homeostasis (Vaishnava et al, 2008). Analysis through 16S rRNA sequencing revealed that loss of VGLL4 alters the composition of the microbiota within the intestine. Vgll4^IEC-KO mice had decreased Firmicutes and increased Bacteroidetes, whose ratio is reported to be decreased in inflammatory bowel disease (IBD) patients (Stojanov et al, 2020). Specifically, the abundance of Bifidobacterium had a tendency to decrease in Vgll4-deficient mice, possibly inhibiting stem cell proliferation and intestinal regeneration (Lee et al, 2018). Akkermansia could produce short-chain fatty acids to regulate the intestinal epithelium (Wu et al, 2021), and its decrease is associated with obesity and nonalcoholic fatty liver disease (Cani et al, 2022). VGLL4 depletion caused decrease in Akkermansia, suggesting that alterations in the microbiota in VGLL4-deficient mice may be linked to other diseases and disorders.

In conclusion, we have characterized the expression and function of VGLL4 in the small intestine of mice. Separately engaging in TEAD4/ATOH1 and TEAD4/TCF4 signaling complexes endows VGLL4 with the ability to facilitate Paneth cell differentiation and influence gut microbiota composition, thus providing a potential therapeutic target for patients with intestinal dysfunction.

Methods

Reagents and tools table

Reagent/resource	Reference or source	Identifier or catalog number
Experimental models
HEK293T	ATCC	ACS-4500
HCoEpiC	ScienCell	Cat# 2950
SW620	Laboratory of Yun Zhao	CCL-227
C57BL/6 J mice	Animal Core Facility, CEMCS	N/A
Vgll4fl/fl mice	Laboratory of Lei Zhang	Yu et al, 2019
Villincre mice	Laboratory of Jun Qin	JAX:021504
Vgll4-3*HA-P2A-CreERT2-T2A-eGFP mice	This paper	N/A
Ai9+/- mice	Laboratory of Bin Zhou	Yu et al, 2019
Lgr5-EGFP-IRES-creERT2 mice	Laboratory of Jianfeng Chen	Chen et al, 2021
Vgll4 fl/-;-LSL-RFP	Laboratory of Lei Zhang	Yu et al, 2019
mTmG+/- mice	Laboratory of Zhengjun Chen	N/A
Escherichia coli TOP10	ToloBio	CC96105
Escherichia coli BL21(DE3)	ToloBio	CC96107
Antibodies
Mouse anti-TEAD4	Abcam	Cat#58310, RRID:AB_945789
Rat anti-EpCAM	Abcam	Cat#ab92382, RRID:AB_2049615
Rabbit anti-VGLL4	ABclonal	PMID: 30789911
Rabbit anti-DEFA5	ABclonal	Cat#A18208, RRID:AB_2861984
Mouse anti-CD24(PE)	BioLegend	Cat#101807, RRID:AB_312840
Mouse anti-CD45(PE/Cy7)	BioLegend	Cat#103113, RRID:AB_312978
Rabbit anti-YAP/TAZ	Cell Signaling Technology	Cat#8418, RRID:AB_10950494
Rabbit anti-HA	Cell Signaling Technology	Cat#3724, RRID:AB_1549585
Rabbit anti-Cleaved-Caspase-3	Cell Signaling Technology	Cat#9661, RRID:AB_2341188
Mouse anti-OLFM4	Cell Signaling Technology	Cat# 39141, RRID:AB_2650511
Rabbit anti-CyclinD1	Cell Signaling Technology	Cat#2978, RRID:AB_2259616
Rabbit anti-Lysozyme	Dako Denmark	Cat#A0099, RRID:AB_2341230
Mouse anti-β-Tubulin	DSHB	Cat#E7, RRID:AB_528499
Rabbit anti-MUC2	GeneTex	Cat#GTX100664S, RRID:AB_10728110
Mouse anti-GST	GenScript	Cat#A00865, RRID:AB_914654
Rabbit anti-SOX9	Millipore	Cat#AB5535, RRID:AB_2239761
Mouse anti-GAPDH	Proteintech	Cat#60004-1-Ig, RRID:AB_2107436
Rabbit anti-GFI1	Proteintech	Cat#14198-1-AP, RRID:AB_2109976
Rabbit anti-ATOH1	Proteintech	Cat#21215-1-AP, RRID:AB_10733126
Rat anti-HA	Roche	Cat# 11867423001, RRID:AB_390918
Rabbit anti-IgG	Santa Cruz Biotechnology	Cat#sc-2027, RRID:AB_737197
Mouse-anti-IgG	Santa Cruz Biotechnology	Cat#sc-2025, RRID:AB_737182
Mouse anti-Flag	Sigma-Aldrich	Cat#F3165 RRID:AB_259529
Rabbit anti-Flag	Sigma-Aldrich	Cat#F7425 RRID:AB_439687
Rabbit anti-Myc	Sigma-Aldrich	Cat#C3956, RRID:AB_439680
Mouse anti-HIS	Sigma-Aldrich	Cat#H1029, RRID:AB_260015
Rabbit anti-Ki67	Thermo Fisher Scientific	Cat#PA5-19462, RRID:AB_10981523
Rat anti-CD326 (EpCAM) (APC)	Thermo Fisher Scientific	Cat#17-5791-82, RRID:AB_2716944
Oligonucleotides and other sequence-based reagents
Clone primers
pGL3-GFI1-RR1-F	CGAGCTCTTACGCGTGCTAGCTTTTTTTTTAAGTCAGAGAGATAGAGCG
pGL3-GFI1-RR1-R	ACTTAGATCGCAGATCTCGAGCTCTCGCACCCCGGCCCG
pGL3-DEFA5-F	CGAGCTCTTACGCGTGCTAGCTTGAATGAAAGCGCTGTTGTTT
pGL3-DEFA5-R	ACTTAGATCGCAGATCTCGAGGGGGAGTGAGGAGTCAGCCT
pGL3-DEFA6-F	CGAGCTCTTACGCGTGCTAGCGTGGCCAGCCCCACCTTC
pGL3-DEFA6-R	ACTTAGATCGCAGATCTCGAGGGGGAGTGAGGAGCCAGCC
pcDNA3.1-MYC-ATOH1-F	CTTGGTACCGAGCTCGGATCCATGTCCCGCCTGCTGCAT
pcDNA3.1-MYC-ATOH1-R	TGCTGGATATCTGCAGAATTCCTAACTTGCCTCATCCGAGTCAC
Real-time PCR primers
Mouse Vgll4-F	ATGAACAACAATATCGGCGTTCT
Mouse Vgll4-R	GGGCTCCATGCTGAATTTCC
Mouse Lyz1-F	GAGACCGAAGCACCGACTATG
Mouse Lyz1-R	CGGTTTTGACATTGTGTTCGC
Mouse Sox9-F	GAGCCGGATCTGAAGAAGGA
Mouse Sox9-R	GCTTGACGTGTGGCTTGTTC
Mouse Math1-F	GACCACCATCACCTTCGCACCG
Mouse Math1-R	AACTCTCCGTCACTTCTGTGG
Mouse Gfi1-F	AGAAGGCGCACAGCTATCAC
Mouse Gfi1-R	GGCTCCATTTTCGACTCGC
Mouse Defa3-F	ATCTGGTATGCTATTGTAGAAA
Mouse Defa3-R	GTGGCCTCAGTACTCATGT
Mouse Defa21-F	GAGAGATCTGATCTGCCTTTG
Mouse Defa21-R	CCTCTATTGCAGCGACGA
Mouse Defa22-F	AGCAGCCAGGGGAAGAG
Mouse Defa22-R	CCTCTATTGCAGCGACGT
Mouse Defa23-F	TCTGGTATGCTATTGTAGAAC
Mouse Defa23-R	GACAGCAGAGCGTGTATA
Mouse Defa24-F	GATCTGGTATGCTATTGTAGAG
Mouse Defa24-R	GACAGCAGAGCATGTACAA
Mouse Defa26-F	ATTGTAGAAAAAGAGGCTGTAC
Mouse Defa26-R	AGCAGAGTGTGTACATTAAATG
Mouse Defa39-F	CAATCACCACCCAAGCTCCA
Mouse Defa39-R	TGAGGCTAAGCACAATGGGA
Mouse Mmp7-F	TTCAAGAGGGTTAGTTGGGGGACTG
Mouse Mmp7-R	TTGTCAAAGTGAGCATCTCCGCC
ChIP-PCR primers
GFI1-ChIP-F1	GAGATAGAGCGGTCTCTCCCG
GFI1-ChIP-R1	AGTGCCAGTGCTCTGCGAGT
DEFA5-ChIP-F	AGACTCACGCTTTGATGAAAGCT
DEFA5-ChIP-R	GTCAGCCTGGATTTATAGCTCTGCT
Chemicals, enzymes and other reagents
Fetal bovine serum	ExCell Bio	Cat# 021C0420A
BeyoZonase™ Super Nuclease	Beyotime	Cat# D7121
Protease Inhibitor Cocktail	MedChemExpress	Cat# HY-K0010
Dextran Sodium Sulfate (DSS)	MP Biomedicals	Cat# 02160110-CF
QIAQIUCK PCR Purification Kit	QIAGEN	Cat# 28104
Protein A/G PLUS agarose beads	Santa Cruz	Cat# sc-2003
37% Formaldehyde	Sigma	Cat# 50-00-0
3×FLAG peptide	Sigma	Cat# F4799
Advanced DMEM/F-12	Thermo fisher-Invitrogen	Cat# 12634010
2 × Phanta Max Master Mix	Vazyme	Cat# P515-03
ClonExpress II One Step Cloning Kit	Vazyme	Cat# C112-01
AceQ Universal SYBR qPCR Master Mix	Vazyme	Cat# Q511-02
RNA isolater Total RNA Extraction Reagent	Vazyme	Cat# R401-01
HiScript III RT SuperMix for qPCR ( + gDNA wiper)	Vazyme	Cat# R323-01
TUNEL BrightGreen Apoptosis Detection Kit	Vazyme	Cat# A112-01
Dual-Luciferase reporter assay system	Promega	Cat# E1960
DAPI (40,6-diamidino-2-phenylindole)	Yeasen Biotechnology (Shanghai) Co., Ltd.	Cat# 40728ES03
Software
ImageJ 1.52a	https://imagej.net/	N/S
GraphPad Prism 10	GraphPad	N/S
Adobe Illustrator	Adobe	N/S
Snapgene	https://www.snapgene.com/	N/S
Leica Application Suite	https://www.leica-microsystems.com/	N/S
Others
RNA-seq RAW data	NCBI sequence read archive (SRA) database	PRJNA1321544
16S rRNA RAW data	NCBI sequence read archive (SRA) database	PRJNA1321500
Non-targeted metabolites RAW data	MetaboLights	MTBLS10456

Mouse models

All mice were kept in a specific pathogen-free (SPF) environment at the Center for Excellence in Molecular Cell Science with protocols approved by Institutional Animal Care and Use Committee (Approval number: SIBCB-S328-1511-052-C01), and Shanghai Jiao Tong University with protocols approved by Laboratory Animal Center (Approval number: A2023115-003). The generation of the transgenic Vgll4^−/− and Vgll4^fl/fl mice has been previously described (Yu et al, 2019). Briefly, a knockout-first strategy was employed to generate the Vgll4 first knockout Vgll4^LacZ/+ mice by introducing a LacZ trapping element between exon1/2 and loxP sites flanking exon2. Male Vgll4^LacZ/+ mice were crossed with female Sox2^Cre+ mice to generate heterozygous Vgll4^+/− mice. Vgll4^+/- mice were intercrossed to generate Vgll4^-/- mice. Vgll4^LacZ/+ heterozygotes were bred to Flp transgenic mice to get Vgll4^fl/+ mice, and Vgll4^fl/+ mice were then intercrossed to generate Vgll4^fl/fl mice. Vgll4^fl/fl mice were crossed with Villin^cre mice to generate Vgll4^IEC-KO mice. Lgr5-EGFP-IRES-creERT2 mice were crossed with Vgll4^{fl/-;-LSL-RFP} mice to generate the ISC-KO mice (Lgr5-EGFP-IRES-creERT2, Vgll4^{fl/-;-LSL-RFP}). Vgll4-3*HA-P2A-CreERT2-T2A-eGFP mice were crossed with mTmG^+/- mice to generate Vgll4-3*HA-P2A-CreERT2-T2A-eGFP, mTmG^+/- mice. All mice were in the C57BL/6 J background and had undergone eight generations of backcrossing following standard procedures.

DSS treatment

Ten to twelve-week-old male wild-type, Vgll4^fl/fl and Vgll4^IEC-KO mice were administered 2% DSS in drinking water for 5–6 days, then the DSS-containing water was replaced by normal water. Body weight was measured daily, and gut tissues were harvested at various specified time points.

γ-ray irradiation

Eight-week-old wild-type mice, Vgll4^fl/fl and Vgll4^IEC-KO mice were exposed to a 10 Gy γ-ray irradiation at Fudan University. Gut tissues were harvested at indicated time points.

Organoid culture and Paneth cell-stem cell co-culture assay

The organoid culture assay was performed according to previously described methods with modifications (Sato et al, 2009). Briefly, small intestines of mice were harvested and washed with PBS for three times. Then the tissues were incubated in 5 mM EDTA for 30 min, followed by shaking with cold PBS and passing through a 70-μm strainer to obtain crypt cells. For the organoid culture, the crypts were resuspended into Matrigel (Corning) with medium (1:1 ratio) and plated into 24well plates. 500 μl of advanced DMEM/F12 medium containing 50 ng/ml mouse-EGF (PeproTech), 1×Primocin (Invitrogen), 500 ng/ml R-Spondin-1 (PeproTech), 100 ng/ml Noggin (PeproTech), 1×HEPES (Invitrogen), 1×GlutaMAX^TM (Thermo Fisher Scientific), 1×B27 (Thermo Fisher Scientific), 1×N2 (Thermo Fisher Scientific), 1 mM N-acetylcysteine (Sigma) was added.

For the cell sorting assay, crypt cells were incubated with TrypLE™ Express at 37 °C for 15 min supplemented with 10 mM Y-27632 (MCE) and 250 U/ml BeyoZonase™ Super Nuclease (Beyotime). The dissociated cells were passed through a cell strainer with a pore size of 40 μm and then stained for FACS antibodies. Paneth cells were gated by DAPI^-CD326^hiCD45^lowCD24^hiSSC^hi, and stem cells were gated by DAPI^-CD326^hiCD45^lowGFP^hi. Wild-type Lgr5⁺ stem cells (WT Lgr5⁺) were sorted from Lgr5-EGFP-IRES-creERT2 mice, and Vgll4-knockout Lgr5⁺ stem cells (KO Lgr5⁺) were sorted from Lgr5-EGFP-IRES-creERT2, Vgll4^fl/fl mice. All Lgr5-EGFP-IRES-creERT2 mice and Lgr5-EGFP-IRES-creERT2, Vgll4^fl/fl mice were injected with TAM three times within one week. Wild-type Paneth cells (WT PCs) were sorted from Vgll4^fl/fl mice, and Vgll4-knockout Paneth cells (KO PCs) were sorted from Vgll4^IEC-KO mice. The sorted cells were mixed (1:1) and embedded into Matrigel. Crypt cultural medium was added. For the first 3 days, 30 ng/ml Wnt3a (PeproTech), 10 mM Nicotinamide (MCE), 10 μM Y-27632 (MCE), and 5 μM CHIR99021 (MCE) were added into the medium.

Histology and immunofluorescence

Mice intestinal samples were fixed in 4% paraformaldehyde overnight and then dehydrated with 30% sucrose at 4 °C for 2–3 days. The small intestinal samples were rolled and then embedded in Optimal Cutting Temperature (OCT) compound at −20 °C for 30 min, then stored at −80 °C. Serial sections were prepared at 10 μm thickness and collected on slides. Hematoxylin & Eosin staining was performed according to standard procedures. The pathology of the samples was observed under an Olympus BX53 microscope, and villus length was quantified using ImageJ. For immunofluorescence, the slides were washed with PBS and then blocked in PBS containing 5% BSA and 0.1% Triton X-100 at room temperature. Primary antibodies were incubated at 4 °C overnight. Then slides were washed with PBS three times and then incubated with Alexa Fluor secondary antibodies and DAPI for 1 h at room temperature. Images  were acquired using Leica SP8 confocal microscope.

Cell culture

Human embryonic kidney (HEK)–293 T cells and human normal colonic epithelial cells (HCoEpiC) were maintained in Dulbecco’s Modified Eagle Medium (DMEM) (Corning) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (Gibco) and cultured at 37 °C in humidified incubators containing an atmosphere of 5% CO₂. All cell lines have been tested for mycoplasma contamination.

Immunoprecipitation and immunoblotting

For immunoprecipitation assay, 293T cells were transfected with different plasmids using PEI. After 48 h, the cells were harvested and lysed with IP buffer (50 mM Tris-HCl pH 8.0, 0.1 M NaCl, 1% NP-40, 10% Glycerol, 1.5 mM EDTA, 1% Protease Inhibitor Cocktail). After centrifugation, the cell lysates were divided into input groups and pulldown groups. For the pulldown groups, 2 μg of antibody was added and incubated at 4 °C overnight. Protein A/G beads were then added for 1 h followed by washing with cold IP buffer for 3 times. Subsequently, the samples were prepared in SDS loading buffer (2% SDS, 10% Glycerol, 50 mM Tris-HCl pH 6.8, 0.025% Bromoxylenol Blue, 5% β-Mercaptoethanol) and subjected to SDS–polyacrylamide gel electrophoresis (PAGE), followed by immunoblotting with the indicated antibodies.

Ileum sections were homogenized in lysis buffer (50 mM Tris-HCl pH 7.5, 0.5 mM EDTA, 1% DSS, 1 mM DTT, 1% Protease Inhibitor Cocktail), followed by centrifugation to obtain the protein supernatant, which was then mixed with SDS loading buffer for SDS-PAGE analysis. All immunoblotting results were visualized using Tanon-4200 Multi System.

RNA extraction, reverse transcription, and real-time PCR

Ileum sections were homogenized in Trizol buffer, followed by centrifugation to get the supernatant. The standard procedure was then followed to isolate pure RNA from the samples. RNA was subjected to reverse transcription with HiScript III RT SuperMix for qPCR ( + gDNA wiper) (Vazyme). Real-time PCR was performed with AceQ Universal SYBR qPCR Master Mix (Vazyme), with GAPDH serving as the internal control.

Sample collection

Small intestinal tissues from human patients were collected from Department of Gastroenterology, Shanghai Xuhui Central Hospital. All procedures were performed under the approval of the Ethics Committee of Shanghai Xuhui Central Hospital (SOP-IEC-033-AF05).

Transmission electron microscope assay

Fresh ileum samples were initially fixed with 2.5% glutaraldehyde at room temperature for 1 h and then at 4 °C overnight. Samples were washed in 0.1 M phosphate buffer twice, 15 min at a time. Then samples were fixed in 1% osmic acid for about 1.5 h and then washed in 0.1 M phosphate buffer for three times. Tissues were progressively dehydrated in ethanol, then substituted by acetone, finally by Epon812 overnight. Samples were then embedded in Epon812 at 60 °C for 48 h. Sections were prepared at a thickness of 70 nm and then stained with 2% uranyl acetate dihydrate and lead citrate. Slides were observed using the FEI Tecnai G2 Spirit 120 kV.

Protein purification and GST pull-down assays

Human ATOH1 (451-1065 bp) was cloned into the pGEX-4T-1-GST vector, human VGLL4 was cloned into the pET-28a-HIS-TRX vector, and human TEAD4 was cloned into the pET-28a-HIS vector. All plasmids were expressed in E. coli BL21 (DE3) cells. The proteins were purified using Superdex 200 Increase by AKTA go (cytiva) system. Equal amounts of different proteins were used for the GST pull-down assay. The proteins and Glutathione agarose beads were incubated at 4 °C for 1 h, then washed for 3 times with PBS. The beads were subjected to SDS-PAGE analysis. Results were detected using both western blot and Coomassie Brilliant Blue staining.

Luciferase reporter assay

The promoter regions of GFI1, DEFA5, and DEFA6 were cloned into the pGL3-basic vector by Nhe I and Xho I restriction enzymes. The GFI1 promoter region can be found in the database NC_000001.11 (92485509-92486109), the DEFA5 region can be found in the database NC_000008.11 (7056740-7057039), the DEFA6 region can be found in the database NC_000008.11 (6926077-6926426). Luciferase plasmids (0.4 μg), other indicated plasmids (0.6 μg), and CMV-Renilla plasmid (0.004 μg) were transfected into HEK293T cells and HCoEpiC cells. After transfection for 36 h, cells were measured following the standard protocol of Dual-Luciferase Assay kit (Promega) on a GloMax 20/20 luminometer (Promega).

Chromatin immunoprecipitation-PCR (ChIP-PCR) and two-step ChIP assay

Cells were fixed with 1% formaldehyde for 15 min at 37 °C then neutralized by Glycine at 37 °C for 5 min. Cells were then lysed by lysis buffer (50 mM HEPES-KOH pH 7.5, 140 mM NaCl, 1 mM EDTA pH 8.0, 1% Triton X-100, 0.1% Sodium Deoxycholate, 0.1% SDS). The chromatins were then sheared to 200–500 bp fragments by sonication (Thermo). Antibodies against mouse IgG and the targeted proteins were used for immunoprecipitation overnight. Protein A/G beads were added to enrich proteins for 2 h at 4 °C. For the two-step ChIP, cells expressing FLAG-TEAD4 and MYC-ATOH1 were used. Chromatin was incubated with anti-FLAG magnetic beads (Sigma) for 6 h at 4 °C. Washed beads for four times. 0.5 mg/mL 3×FLAG peptide (Sigma) were used to elute FLAG-TEAD4 for 15 min at 37 °C. Two sequential elution was performed. The eluate was then subjected to a second pull-down using anti-IgG or anti-MYC antibody at 4 °C overnight. The immunoprecipitated DNA was collected with the QIAGEN PCR Purification Kit (250). The purified DNA was used for qPCR .

RNA sequencing and bioinformatics analysis

Ileum samples were used to extract RNA by Trizol reagent, the RNA being assessed by Agilent 5300. RNA sequencing was performed at Maiorbio Co., Ltd (Shanghai, China) according to the manufacturer’s instructions, and the data were analyzed through the free online platform of the majorbio cloud platform (cloud.majorbio.com). The transcriptome library was prepared following Illumina Standard mRNA Prep protocol, and the sequencing was performed on the NovaSeq X Plus platform (PE150) using NovaSeq Reagent Kit. Clean reads were aligned to the reference genome (Ensembl_GRCm38.p6) with HISAT2 (version 2.1.0) software. RSEM (version 1.3.1) software was used to quantified the expression level. Differentially expressed genes were identified by $\mathrm{FDR}<0.05\mid \log 2\mathrm{FC}\mid \geqq 1$ with DESeq2 (version 1.24.0). Genesets enrichment analyses were performed by R package ‘clusterProfiler’, and heatmap analyses were performed by R package ‘ComplexHeatmap’.

16S rRNA sequencing analysis and untargeted metabolomics by LC-MS

Fecal samples from Vgll4^IEC-KO and Vgll4^fl/fl male mice were collected at 12 weeks of age for 16S rRNA sequencing and untargeted metabolomics analysis. For 16S rRNA sequencing, primers targeting the V3-V4 region of the bacterial 16S rRNA gene (338 F: ACTCCTACGGGAGGCAGCAG; 806 R: GGACTACHVGGGTWTCTAAT) were used to amplify the DNA samples. Purified amplicons were paired-end sequenced using Miseq  PE300 (Illumina, San Diego, California, USA) platform according to the standard protocols by Majorbio Co., Ltd (Shanghai, China). Raw data quality control was performed by fastp (version 0.19.6) and merged by FLASH (version 1.2.11). Noise control was performed with DADA2. Sequences were annotated to silva138.1/16s_bacteria database with classify-sklearn (Naive Bayes). Bioinformatic analysis of the gut microbiota was carried out using the Majorbio Cloud platform (https://cloud.majorbio.com). α-diversity was performed with mothur (version 1.30.2), β-diversity was performed with R (version 3.3.1) package vegan (version 2.4.3). LEfSe analysis was performed with LEfSe (http://huttenhower.sph.harvard.edu/LEfSe).

For untargeted metabolomics analysis, metabolites were extracted and analyzed by UHPLC-Q Exactive HF-X system from Thermo Fisher Scientific according to the standard protocols by Majorbio Co. Ltd (Shanghai. China). Raw data were processed by Progenesis QI (Waters Corporation, Milford, USA) platform. Then metabolites were searched and identified against HMDB (http://www.hmdb.ca/), Metlin (https://metlin.scripps.edu/) and Majorbio Database. Bioinformatic analysis of the metabolites was carried out using the Majorbio Cloud platform (https://cloud.majorbio.com). Differential expressed metabolites analysis and PCA analysis were performed by R package ropls (version1.6.2). KEGG kegg_v20221012 was used to analyze the classification of differentially expressed metabolites and related pathways. Heatmap analysis, KEGG pathway enrichment analysis, and correlation analysis were performed with scipy (Python). Ropls (R) and scipy (Python) were used for VIP analysis.

Statistics

Comparisons between two groups were performed by unpaired Student’s t test in GraphPad Prism 10. Comparisons between more than two groups were performed by one-way ANOVA. The results are shown as mean + SD. Statistical significance was denoted as *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, and non-significant results as “ns”.

Author contributions

Haoen Zhang: Data curation; Software; Formal analysis; Validation; Investigation; Methodology; Writing—original draft. Zuoyun Wang: Conceptualization; Formal analysis; Supervision; Funding acquisition; Validation; Writing—review and editing. Xiaodong Wang: Resources; Methodology. Wentao Yu: Resources; Methodology. Guoying Zhang: Resources; Methodology. Haijiao Zhang: Resources. Yi Lu: Resources. Yang Sun: Resources. Tiantian Lu: Resources. Xiaoyu Li: Resources. Ruizeng Yang: Resources. Jiaqi Sun: Resources. Jinjin Xu: Resources. Shuo Huang: Resources. Xueyan Ma: Resources. Jiale Ren: Resources. Nan Tang: Resources. Zhonghua Cheng: Resources. Jing Yu: Resources; Writing—review and editing. Fang Wei: Resources; Writing—review and editing. Hu Zhou: Resources; Methodology. Jinsong Li: Resources; Methodology. Jun Qin: Resources; Data curation; Methodology. Yunyun Jin: Resources; Data curation; Funding acquisition; Methodology; Writing—review and editing. Lei Zhang: Conceptualization; Data curation; Supervision; Funding acquisition; Project administration; Writing—review and editing.

Source data underlying figure panels in this paper may have individual authorship assigned. Where available, figure panel/source data authorship is listed in the following database record: biostudies:S-SCDT-10_1038-S44319-026-00699-3.

Data availability

The data used and analyzed during the current study are available from the corresponding author upon reasonable request. The datasets produced in this study are available in the following databases: RNA-Seq data: Sequence Read Archive (SRA) Database, PRJNA1321544 (https://www.ncbi.nlm.nih.gov/sra/PRJNA1321544). 16S rRNA-Seq data: Sequence Read Archive (SRA) Database, PRJNA1321500. Untargeted Metabolomics Sequencing Data: Metabolights Database, MTBLS10456 (https://www.ebi.ac.uk/metabolights/MTBLS10456).

The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44319-026-00699-3.

Disclosure and competing interests statement

The authors declare no competing interests.

Supplementary information

Expanded view data, supplementary information, appendices are available for this paper at 10.1038/s44319-026-00699-3.