Because TFEB plays an essential role in the colon under bacterial infection and the majority of research focuses on macrophages (Gray et al., 2016; El-Houjeiri et al., 2019; Rao et al., 2020; Schuster et al., 2022), we were eager to explore the detailed distribution of TFEB, which might clarify the TFEB-mediated anti-bacterial responses. According to the Human Protein Atlas (HPA), TFEB is highly expressed in enterocytes (Figure 1A). To further verify this finding, the colon tissues of healthy C57/B6 mice were dissected and subjected to immunohistochemical (IHC) staining to determine the expression of TFEB (Figure 1B). The results were consistent with the dataset. In addition, the primary colonic epithelial cells of mice were isolated and used to examine the protein levels of TFEB, in comparison with BMDM, colon tissue without epithelium, and whole colon tissue. Western blotting analysis showed that TFEB was less expressed in the macrophages and more expressed in the epithelial cells (Figures 1C,D). Therefore, the epithelium may be the cardinal battlefield for TFEB to exert its anti-bacterial functions.

To investigate the effects of infection on the regulation of TFEB expression and functions, LPS derived from S.typhimurium was used in the stimulation of a colonic epithelial cell line Caco 2 for 0 h, 6 h, 12 h, 24 h, and 36 h. TFEB, the lysosomal marker LAMP1, and the autophagic molecule LC3 were determined in these treated cells using Western blotting. The results revealed that LPS enhanced the expression of all the determined proteins in a time-dependent manner (Figures 2A–D). NF-κB and mTOCR signaling pathways usually function as the two major responders to the stimulation of LPS (Temiz-Resitoglu et al., 2017; Lund et al., 2022). Subsequently, to figure out the potential mechanisms underlying the increase of TFEB, the phosphorylation of NF-κB and mTOCR1 signaling pathways (S6 and 4EBP1) was detected at different time points. Interestingly, the activity of P65 was remarkably potentiated by LPS, indicating that epithelial cells underwent cellular responses to LPS. The levels of p-S6 were slightly increased within 6 h, while they were dramatically decreased after longer treatment with LPS. However, the phosphorylation of 4EBP1 was constantly reduced (Figures 2A,F–H). Moreover, the Caco 2 cells were directly infected with Salmonella to figure out whether bacterial infection has a similar effect on the regulation of TFEF expression and functions as LPS. The results indicated that Salmonella also inhibited mTOR activation and increased TFEB, LAMP1, and LC3 (Figures 2I–O). To further demonstrate this upregulation of LAMP1 and LC3 by LPS and Salmonella infection, we used shRNA to silence the expression of TFEB in the Caco 2 cells and then treated these engineered cells with LPS or Salmonella infection. As expected, downregulating TFEB inhibited the induction of LAMP1 and MAP 1LC3B (the gene for LC3) expression (Figure 2P). Additionally, gene set enrichment analysis (GSEA) was conducted, revealing that the autophagy- and lysosome-related genes were enriched after LPS stimulation in another colonic epithelial cell line, HT29 cells, as indicated by a Gene Expression Omnibus (GEO) database (Figures 2Q,R). Together, these observations suggest that LPS enhances TFEB and autophagy-lysosome protein expression, which is likely due to the inactivation of mTOR.

As a TF, TFEB should be translocated to the nucleus to exert its transcriptional functions (Napolitano and Ballabio, 2016). To evaluate the alteration of TFEB activity after infection, TFEB and DAPI were determined in the Caco 2 cells following stimulation with LPS by immunofluorescence staining. The images showed that both the total fluorescence intensity of TFEB and the proportions of nuclear TFEB were significantly increased after the treatment with LPS (Figures 3A–C). To further confirm these observations, we determined the TFEB levels in the cytoplasm and nucleus of the Caco 2 cells after LPS stimulation. Consistently, we found significantly increased TFEB expression in both the cytoplasm and nucleus following the stimulation with LPS for 12 h and 24 h. However, there were no changes in TFEB levels within 6 h (Figures 3D–F). Furthermore, infection with Salmonella has a similar effect on the TFEB translocation (Figures 3G–I). Collectively, these results suggest that both LPS and Salmonella infections enhance the transcriptional activity of TFEB in the colonic epithelium.

To further substantiate that TFEB can be activated by infection in vivo, C57/B6 mice were infected with Salmonella (ampicillin-resistant) by oral gavage for 48 h or 120 h. After the indicated time, the mice were euthanized under anesthesia. The length of the colon was measured, and it showed a shrinkage of colon length in a time-dependent manner (Figures 4A,B). Additionally, the weight of the spleen also increased after infection (Figure 4C). Moreover, by culturing the tissue lysis solutions in LB plates containing ampicillin, we found a large number of Salmonella that survived in the feces, colon, and spleen of mice (Figures 4D–F). Thus, these observations confirmed that Salmonella caused colitis and invaded the colon of mice. Next, we sought to investigate whether the pathogenic bacteria modulated the activity of TFEB in the epithelium using immunohistochemistry (IHC) staining. The results showed a significant increase of the nuclear TFEB in the epithelium of these mice infected with Salmonella (Figures 4G,H). Moreover, we isolated the colonic epithelial cells and determined TFEB using Western blotting. In line with the histochemical analysis, Salmonella infection enhanced the nuclear translocation of TFEB (Figure 4I). Hence, these findings indicate that the Salmonella infection results in the activation of TFEB in colonic epithelial cells in vivo.

By clarifying the predominant distribution of TFEB in the epithelium and demonstrating its activation by LPS and Salmonella in vitro and in a mouse model, we subsequently aimed to determine the distinct functions executed by TFEB. Therefore, we used the GEO dataset and conducted a Gene Ontology (GO) analysis and found that most of the upregulated genes with LPS treatment were related to anti-bacterial response (Figure 5A). The GSEA of this dataset showed that post-administration with LPS, there was an enrichment of anti-bacterial response- and peptide-related genes (Figures 5B,C). Furthermore, the heatmap graph showed that LPS increased a large group of anti-bacterial peptides, such as PI3, LCN2, HTN1, and S100A9, that were positively related to TFEB (Figure 5D). To confirm the critical role of TFEB in bacterial defense, we utilized the Cancer Cell Line Encyclopedia (CCLE) gene set, which measured whole gene expression with RNA-seq in a series of cell lines and conducted GSEA. The results showed that genes linked to anti-bacterial response and peptides were enriched in TFEB highly expressed intestinal epithelial cell lines (Figures 5E,F). More importantly, the levels of several critical genes that encode anti-bacterial peptides, including PI3, LCN2, HTN1, and S100A9, were significantly increased after LPS treatment. However, these genes were dramatically decreased by the deficiency of TFEB (Figure 5G). To investigate whether these upregulated anti-bacterial peptides can suppress bacterial replication, we collected the supernatant of control Caco 2 and shTFEB Caco 2 cells, with or without LPS treatment, and cultured Salmonella. After 12 h of culture, we found that fewer bacteria survived in the LPS-treated culture medium than in the non-treated medium, and this effect was diminished upon the reduction of TFEB expression (Figure 5H). In addition, to investigate other possible mechanisms that may contribute to the anti-bacterial response mediated by TFEB, we conducted GSEA to examine the relationship between TFEB and the tight junction and polarity of epithelial cells based on the CCLE dataset. The results revealed a less significant correlation between TFEB level and tight junction and polarity (Supplementary Figures S1A,B). We also tested the association between TFEB and inflammatory cytokine expression. GSEA and heatmaps showed that TFEB also did not affect the expression of cytokines such as IL-1, IL-6, and TNF in the epithelial cells (Supplementary Figures S1C,D). Therefore, epithelial TFEB contributes to the production of anti-bacterial peptides to fight infection.

The IHC antibody anti-TFEB (ab2636) was purchased from Abcam. Western-blot antibodies, including anti-LAMP1 (sc-20011, Santa Cruz Biotechnology), anti-LC3 (3,868, Cell Signaling Technology), anti-p-P65 (3,033, Cell Signaling Technology), anti-P65 (8,242, Cell Signaling Technology), anti-p-S6 (4,858, Cell Signaling Technology), anti-S6 (2,217, Cell Signaling Technology), anti-p-4EBP1 (2,855, Cell Signaling Technology), anti-4EBP1 (9,452, Cell Signaling Technology), anti-Actin (3,700, Cell Signaling Technology), anti-Histone H3 (ab1791, Abcam), donkey Anti-Goat IgG H&L (FITC, ab6881), and goat anti-rabbit IgG H&L (HRP, ab6721) were purchased from Abcam. DAPI (d9542) and LPS (L6386) were purchased from Sigma Aldrich.

The Caco 2 cells (SCSP-5027) were purchased from the National Collection of Authenticated Cell Cultures. The cells were cultured with DMEM (10,566,016, Thermo Fisher Scientific) containing 10% FBS, streptomycin, and penicillin at 37°C, 5% CO₂. The Caco 2 cells were planted onto six-well plates and stimulated with 100 ng/mL of LPS for 0 h, 6 h, 12 h, 24 h, or 36 h. The cells were collected for Western blotting or immunofluorescence staining. BMDMs were differentiated and cultured as previously described (Xia et al., 2019).

The LPS-treated Caco 2 cells, BMDMs, primary murine colonic epithelial cells, colon tissue without epithelium, and integrated colon tissue were lysed with RIPA buffer containing proteases and phosphatase inhibitor cocktail (78,440, Thermo Fisher Scientific). The concentration of cell lysate was measured using a BCA kit (A53226, Thermo Fisher Scientific). After adding 5 × protein loading buffer and boiling for 5 min, the protein samples were loaded onto a 10% SDS-PAGE gel. Separated proteins were transferred onto a PVDF membrane post-electrophoresis for 2 h. The PVDF membrane containing proteins was blocked with 5% BSA and incubated with primary antibodies overnight. The blots were visualized using electrochemiluminescence (ECL) after incubation with HRP-conjugated second antibody. Quantification was performed using Image J.

Infected or non-infected murine colonic tissues were fixed with formalin and embedded in paraffin. The tissues were sectioned into slices and then baked for 1 h before being dewaxed with dimethylbenzene. Dehydration was performed using ethanol and endogenous peroxidase was eliminated with 3% H₂O₂. The sections were incubated with a 5% BSA and then an anti-TFEB antibody (1:200). After staining with an HRP-conjugated second antibody, the sections were visualized with diaminobenzidin (DAB) and hematoxylin. An Olympus microscope was used to capture images.

The LPS-treated Caco 2 cells were fixed with 4% paraformaldehyde for 20 min. The cell membrane was permeated with 0.5% Triton X-100 after being washed with PBS. The cell slides were blocked with a 5% BSA before being incubated with anti-TFEB. The slides were stained with FITC-conjugated secondary antibody and DAPI and then photographed using confocal microscopy (Zeiss, Germany).

To separate the cell cytoplasm and nucleus, a commercial kit (P0028, Beyotime) was used, and the detailed experiments were conducted according to the manufacturer’s instructions.

The total RNA was isolated using TRIZOL (15,596,026, Invitrogen™), and the cDNA was obtained using a commercial kit (4,374,967, Applied Biosystems™). Next, the qRT-PCR was performed using SYBR green (A46110, Applied Biosystems™), and the primers used are as follows:

TFEB, F-CCTGGAGATGACCAACAAGCAG; R-TAGGCAGCTCCTGCTTCACCAC.

PI3, F-CGCTGCTTGAAAGATACTGACTG; R-ACGGCACAGGTGCAGCAAGGA.

LCN2, F-GTGAGCACCAACTACAACCAGC; R-GTTCCGAAGTCAGCTCCTTGGT.

HTN1, F-CATCATGGGTATAGAAGAAAATTCC; R-TGCCCCATGATTACTAAGGATATC.

S100A9, F-GCACCCAGACACCCTGAACCA; R-TGTGTCCAGGTCCTCCATGATG.

GAPDH, F-GTCTCCTCTGACTTCAACAGCG; R-ACCACCCTGTTGCTGTAGCCAA.

LAMP1, F-GGCCTCTTGCGTCTGGTAAC; R-AAAGGTACGCCTGGATGGTG.

MAP1LC3B, F-GAGAAGCAGCTTCCTGTTCTGG; and R-GTGTCCGTTCACCAACAGGAAG.

shRNA lentivirus construction and shTFEB Caco 2 cells generation.

The scrambled shRNA lentivirus target human TFEB was constructed using pLVX-shRNA2 plasmids (Clontech Laboratories, Inc., 632,179), following the procedure described previously (15). These plasmids were transfected onto 293 T cells to generate lentivirus in the supernatant. For the generation of shTFEB Caco 2 cells, the epithelial cells were infected with lentivirus for approximately 5 days, and the TFEB level was determined by RT-PCR. The shRNA sequences are as follows:

Forward: GATCCCCACTTTGGTGCTAATAGCTTTCAAGAGAAGCTATTAGCACCAAAGT.

GGGTTTTTG;

Reverse: AATTCAAAAACCCACTTTGGTGCTAATAGCTTCTCTTGAA AGCTATTAGCAC.

CAAAGTGGG.

Salmonella (SL1344, ampicillin-resistant) was a gift from Xiang-ping Yang lab, and the protocol for culturing bacteria was the same as before (Rao et al., 2020). For the infection in vivo, Salmonella was cultured in 3 mL of LB medium for 4 h and measured using a spectrophotometer (Thermo Fisher Scientific). The infection model was performed as previously described (Xia et al., 2019) and in accordance with Institutional Animal Care guidelines. In brief, 30 C57/B6 mice were randomly arranged into three groups (5 mice per group). Two groups of mice were intragastrically infected with 1 × 10⁸ of Salmonella for 48 h or 120 h. Infected mice or healthy mice were euthanized under anesthesia. The length of the colon and the weight of the spleen were determined. The tissues were homogenized and resuspended in sterile PBS. After serial dilution, the suspensions were plated onto LB plates with ampicillin. The number of bacterial colonies was quantified.

For the extraction of colonic epithelial cells, normal or Salmonella-infected colon tissues of C56/B6 mice were collected from the mouse model (5 mice per group). After three washes with sterile PBS, the colonic lumens were injected and filled with trypsin (Thermo Fisher Scientific, 25,200,072). The two ends of the colon lumen were sealed with surgical thread ties and incubated in a cell incubator for 1 h at 37°C. After digestion, these epithelial cells in the colonic lumen could be pipetted and ejected. Then, a 10% FBS was added to the cell suspensions, and the colonic epithelial cells and non-epithelial cell colon tissues were obtained. After centrifugation and two washes with ice-cold PBS, these newly isolated epithelial cells were cultured in DMEM (10% FBS) for approximately 30–60 min. The suspended epithelial cells in the culture medium were then collected for further determination. The purity of epithelial cells was determined by flow cytometry using EpCAM antibodies (Biolegend, 118,207), a marker of epithelial cells.

The control Caco 2 cells and shTFEB Caco 2 cells were treated with LPS (100 ng/mL) for 48 h, and the culture media were collected and centrifuged at 10000 rpm. The supernatants were used to culture Salmonella in a concentration of 5 × 10⁵/mL. After 12 h of culture, the media containing Salmonella were collected and plated on LB plates at an appropriate dilution. The CFU of Salmonella was recorded and quantified.

The gene expression profile of HT29 cells was downloaded from the GEO database, specifically from the GSE113581 dataset.¹ HT29 cells were treated with LPS or vehicle (DMSO). We downloaded and processed the RNA-seq data of 57 intestinal epithelial cell lines from the CCLE.²

We downloaded related gene sets from the Molecular Signatures of GSEA official website³ and used GSEA software to determine the different pathways related to the target genes. The GO annotation for the targeted genes was performed using the R packages “clusterProfiler” and “enrichplot” to determine the upregulated genes in epithelial cells following LPS treatment. The CCLE was used to interrogate the transcriptomics data of the inflammatory cytokine-related genes in a panel of 57 intestinal epithelial cell lines (see text footnote 2, respectively).

Statistical analyses were performed using GraphPad Prism 8. An unpaired two-tailed Student’s t-test was used to compare two groups, while a one-way ANOVA with Dunnett’s test was used to compare at least three groups. Results were presented as mean values ± standard error of the mean (s.e.m). A p-value less than 0.05 was considered significant.

R (version 4.0.4) was used to carry out differential analysis of the RNA expression profile, and the “limma” package was used to standardize and analyze all data. Differential genes were screened based on corrected p-values less than 0.05 and absolute log fold change greater than 1.