This study involved the use of an E. coli F4-producing strain W25K (hereafter referred as ETEC; O149:K91, K88ac; LT, STb, EAST), which was isolated from a piglet with diarrhea (17). ETEC W470 (O4:F18; STa, STb, LT), W817 (O107:F18; STb), and W616 (F18; STa) were also isolated from piglets with diarrhea, while the Shiga-like toxin producing E. coli (W197, SLT-IIe) was isolated from a piglet with edema disease (18). These strains of bacteria were cultured in LB medium. L. lactis subsp. lactis (ATCC19435) was cultured in M17 medium. C. rodentium (DBS100) was cultured in LB medium. Antibodies against RAR-related orphan receptor gamma t (RORγt) (Sc-14196), forkhead box P3 (Foxp3) (Sc-28705), growth factor independent 1 (GFI-1) (Sc-8558), early growth response protein 2 (EGR-2) (Sc-20690), p85 (Sc-1637), phosphorylated protein kinase B (Akt) (Sc7985-R), GAT-2 (Sc-7668), actin (Sc-47778), and proliferating cell nuclear antigen (PCNA) (Sc-56) were purchased from Santa Cruz Biotechnology, Inc. (Dallas, TX, USA). Antibodies against mTOR (CST 2972), p-mTOR (CST 5536), p-p70 S6 Kinase (CST 9205), p-4E-BP1 (CST 9451), p-AMP-activated protein kinase (AMPK) (CST 2535), hypoxia-inducible factor 1 α (HIF-1α) (CST 14179), and p70 S6 Kinase 2 (CST 14130) were purchased from Cell Signaling Technology (Danvers, MA, USA).

This study was approved and conducted according to the guidelines of the Institute of Subtropical Agriculture, Chinese Academy of Sciences and Southwest University. Piglets (Landrace Yorkshire; 18 days old) were purchased from ZhengDa Co., Chongqing, China. ETEC infection in piglets was established according to previous reports (19, 20). The jejunum samples were collected. Samples were stored at −80°C until processing.

TCR delta knockout mice were provided by Prof. Zhinan Yin, from Jinan University (Guangzhou, China). Rag 1 knockout mice were bought from Nanjing University (Nanjing, China). Germ-free mice were generated and provided by Prof. Hong Wei, from Third Military Medical University (Chongqing, China), and these mice were maintained in sterile Trexler-type isolators. ICR mice (6 weeks of age) were purchased from SLAC Laboratory Animal Central (Changsha, China). Experiments in mice were conducted according to the guidelines of the Laboratory Animal Ethical Commission of the Chinese Academy of Sciences.

Mice were orally gavaged with 10⁸ CFUs of ETEC or other strains of E. coli. At 6-h postinfection, mice were sacrificed to collect the jejunum, ileum, and mesenteric lymph node. In some experiments, mice were treated with rapamycin (Fisher Scientific) at 2.5 mg/kg/d (i.p. injection) for six consecutive days prior to other manipulations. In experiments involving the inhibition of GABA signaling, mice were intraperitoneally injected with 20 mg/kg bicuculline (Dalian Meilun Bio. Tech. Co., Ltd., Dalian, China), 120 mg/kg CGP-35348 (Tocris Bioscience), 80 mg/kg l-allylglycine (Aladdin Industrial Corporation, China), or 100 mg/kg semicarbazide (Sangon Biotech Co., Ltd., Shanghai, China) at 30 min prior to other manipulations. For the inhibition of S6K1, mice were intraperitoneally injected with 75 mg/kg/d PF-4708671 (S2163, Selleck) for six consecutive days. In some experiments, mice were intraperitoneally injected with GABA (Aladdin Industrial Corporation, China) at dosages of 40 μg//kg to 40 mg/kg at 30 min prior to other manipulations [the physiologic level of GABA in mouse serum is about 2 mg/kg (21), while in human (adults), it is about 100–400 µg/kg]. For antibiotics treatment, mice received drinking water containing streptomycin (1.0 g/L, Sigma), ampicillin (1.0 g/L, Sigma), gentamicin (1.0 g/L, Sigma), and vancomycin (0.5 g/L, Sigma) for 1 week before ETEC infection. For the supplementation of GABA to antibiotics-treated mice, mice received drinking water containing antibiotics (1.0 g/L streptomycin, 1.0 g/L ampicillin, 1.0 g/L gentamicin, and 0.5 g/L vancomycin) and GABA (1.0 g/L or 5.0 g/L) for 1 week before ETEC infection. For L. lactis subsp. lactis inoculation into antibiotics-treated mice, mice received drinking water containing antibiotics (1.0 g/L streptomycin, 1.0 g/L ampicillin, 1.0 g/L gentamicin, and 0.5 g/L vancomycin) for 6 days, and then orally inoculated with L. lactis subsp. lactis at a dosage of 10⁸ CFUs at 24 h before ETEC infection. For L. lactis subsp. lactis inoculation into germ-free mice, mice were orally inoculated with L. lactis subsp. lactis at a dosage of 10⁸ CFUs at 7 days before ETEC infection.

For L. lactis subsp. lactis inoculation into normal mice, mice were orally inoculated with L. lactis subsp. lactis at a dosage of 10⁸ CFUs for five consecutive days. For GABA signaling inhibition, mice were intraperitoneally injected with 120 mg/kg CGP-35348 at 5 days postinoculation. The jejunum samples were collected for further analysis at 5 days postinoculation (6-h post-CGP-35348 treatment).

For L. lactis subsp. lactis inoculation in piglets, piglets were orally inoculated with L. lactis subsp. lactis at dose of 10¹⁰, while the control piglets received some volume of M17 medium. The jejunum samples were collected for analysis at 7 days postinoculation.

Mice (6 weeks of age) were orally gavaged with 10⁸ CFUs of C. rodentium (DBS100). For rapamycin treatment, mice were treated with rapamycin at 2.5 mg/kg/d (i.p. injection) at 1–7 days postinfection. For semicarbazide treatment, mice were treated with semicarbazide at 100 mg/kg/d (i.p. injection) at 1, 3, 5, and 7 days postinfection. In some experiments, mice were intraperitoneally injected with 20 mg/kg bicuculline or 120 mg/kg CGP-35348 at 7 days postinfection. The colon samples from all groups were collected for further analysis at 7 days postinfection (6-h post-bicuculline or CGP-35348 treatment).

ICR mice (6 weeks of age) were intraperitoneally injected with 300 mg/kg 5-fluorouracil (Sigma-Aldrich). For the inhibition of GABA signaling, mice received intraperitoneal administration of 20 mg/kg bicuculline or 100 mg/kg semicarbazide at 30 min prior to 5-fluorouracil treatment. The jejunum was collected at 6-h posttreatment for further analysis.

Lamina propria lymphocytes from jejunum were obtained according to previous procedures (12, 22, 23). Briefly, jejunum tissue was extracted, opened longitudinally to wash luminal contents. Jejunum (cleared the visible PP) was cut into 0.5- to 1-cm segments that were incubated three times (37°C, 20 min, 250 rpm) in HBSS/HEPES medium (Life Technologies) containing 5 mM EDTA to remove the epithelial cells. Thereafter, the tissue segments were minced and digested in digestion buffer [HBSS + 5% FBS containing type VIII collagenase at 1.5 mg/mL (Sigma)] for 30-min incubations at 37°C. Digested material was passed through a 100-µm cell strainer, and the cells were collected by centrifugation, washed twice in DMEM, and then passed through a 40-µm cell strainer to obtain lamina propria lymphocytes from the jejunum.

Lymphocytes isolated from mouse jejunal lamina propria were stained with cell surface markers of CD3 (FITC-CD3, 100203, Biolegend), CD4 (PE-CD4, 100407, Biolegend), or TCR γδ (PE-TCR γδ, 118107, Biolegend). For the analysis of intracellular cytokine, lymphocytes or CD4⁺ T cells were stimulated with PMA (50 ng/mL), ionomycin (1 µg/mL), and monensin (3 µg/mL) for 5 h. After staining with cell surface markers, intracellular cytokine staining was performed with a fixation and permeabilization kit (eBioscience) and IL-17A Ab (APC-IL-17A, 506195, Biolegend) in accordance with the manufacturer’s instructions. Flow cytometry was performed on a FACSCalibur (BD Biosciences) and data were analyzed using the FlowJo Software (Tree Star).

Metabolite concentrations in piglet jejunal samples were quantified using gas chromatography/mass spectrometry (GC–MS) according to the previous work (19).

Real-time PCR was performed according to our previous study (20, 24). Primers (Table S1 in Supplementary Material) were selected according to previous references. β-Actin was used as an internal control to normalize target gene transcript levels.

Immunoblotting was performed according to our previous study (20, 24). Signal intensity was digitally quantified and normalized to actin or PCNA protein abundance.

Naive CD4⁺ T cells were isolated from mouse splenocytes using a CD4⁺CD62L⁺ T cell isolation kit II (Miltenyi Biotec, purity > 95%). For Th17 differentiation, cells were stimulated with anti-CD3 (2 µg/mL, 100313, Biolegend) and anti-CD28 (2.0 µg/mL, 102111, Biolegend) supplemented with 5 ng/ml TGF-β1 (7666-MB-005, RD), 20 ng/mL IL-6 (575702, Biolegend), 10 µg/mL anti-IFN-γ (505812, Biolegend), and 10 µg/mL anti-IL-4 (504107, Biolegend) for 1–3 days. l-Allylglycine, semicarbazide, and rapamycin were used at 5 mM, 4 mM, and 3 µM, respectively.

Data shown are the means ± SEM. Data between two groups were analyzed by unpaired t test (Prism 6.0) if the data were in Gaussian distribution and had equal variance, or by unpaired t test with Welch’s correction (Prism 6.0) if the data were in Gaussian distribution but with unequal variance, or by non-parametric test (Mann–Whitney U test, Prism 6.0) if the data were not normally distributed. Data among more than two groups were analyzed by the one-way ANOVA followed by Dunnett multiple comparisons (Prism 6.0) if the data were in Gaussian distribution and had equal variance, or analyzed by Kruskal–Wallis followed by Dunn’s multiple comparisons (Prism 6.0) if the data were not normally distributed. The Gaussian distribution of data was analyzed by D’Agostino–Pearson omnibus normality test (Prism 6.0) and Kolmogorov–Smirnov test (Prism 6.0). The variance of data was analyzed by homogeneity of variance test (SPSS 22.0) or Brown–Forsythe test (Prism 6.0). The Bonferroni correction was applied for multiple pairwise comparisons. Differences with p < 0.05 were considered significant.

Enterotoxigenic Escherichia coli infection is known to promote IL-17 expression in piglets (15, 16), a finding we also validated. We infected Landrace × Yorkshire piglets with ETEC W25K, a strain originally isolated from a piglet with diarrhea (17). IL-17 mRNA expression in the porcine jejunum was higher (about 17-fold) in piglets infected with ETEC, as compared with uninfected controls (Figure 1A). Increased mRNA expression of other proinflammatory cytokines, including IL-6, IL-8, and TNF-α, was also detected in the jejunum after ETEC infection (Figure 1A).

In mice, at 6-h postinfection, ETEC colonized in the jejunum with load of about 5.9 log₁₀ CFU/g and promoted the expression of Il-17 in the jejunum (Figure 1B). ETEC also promoted the expression of Il-17 in the ileum (Figure 1C). ETEC infection enhanced the abundance of RORγt (a key transcriptional regulator of Th17 cells) in the cytoplasm and nuclei of mouse jejunum (Figure 1D) and ileum (Figure 1E).

Although the best-characterized cellular source of IL-17 is Th17 cells, a number of adaptive and innate immune cells, such as γδ T cells, natural killer T (NKT) cells, and even Paneth cells, are also the sources of IL-17 (11, 13). Similar to previous conclusion that the increased IL-17⁺ cells at 5-h post-E. coli infection are CD3⁺ cells (25), we found that most of IL-17⁺ cells from ETEC-infected mouse jejunum were CD3⁺ cells. Thus, the ETEC-infected mice had a higher percentage of CD3⁺IL-17⁺ cells in the jejunum compared with the controls (Figure 1F), suggesting that the increased expression of IL-17 at 6-h post-ETEC infection comes from T cells. Indeed, ETEC infection had little effect on the expression of Il-17 in the jejunum in Rag 1 knockout mice (mature T cell-depleted mice), while wild-type mice had higher expression of Il-17 in the jejunum compared to Rag 1 knockout mice during ETEC infection (Figure 1G). Unlike the previous study, which shows that the increased IL-17 is mainly produced by γδ T cells during early E. coli infection (25), we observed a higher percentage of CD3⁺CD4⁺IL-17⁺ cells in the jejunum from ETEC-infected mice as compared with uninfected mice, while the percentage of CD3⁺TCR γδ⁺IL-17⁺ cells in the jejunum did not differ between ETEC-infected mice and control mice (Figures 1H,I). Notably, Il-17 expression in the jejunum after 6 h of ETEC infection was similar between TCR delta^−/− mice (γδ T cell-depleted mice) and wild-type mice (Figure 1J). In conclusion, ETEC infection promotes the intestinal mRNA expression of IL-17 in piglets and mice.

A previous study has shown that ETEC infection inhibits the activation of NF-κB and mitogen-activated protein kinases pathways in the jejunum (20), indicating that new pathways are associated with expression of IL-17 during ETEC infection. Previous proteomic analysis indicated that ETEC infection upregulates ribosomal protein S6 kinase in the piglet jejunum (20), which suggests that ETEC infection may activate mTORC1 signaling. The mTORC1 has a critical role in Th17 responses and in IL-17 expression (26). Indeed, the mTORC1 pathway was activated in piglets infected with ETEC, based on the higher abundance of phosphorylated mTORC1 and its downstream target S6K (Figure 2A). Phosphorylated mTORC1 abundance was also higher in ETEC-infected mice than that in uninfected controls (Figure 2B). To validate the possible roles of mTORC1 signaling in the expression of IL-17 during ETEC infection, this study used rapamycin to inhibit the activation of mTORC1. Rapamycin treatment prior to ETEC infection prevented the increase in mTORC1 phosphorylation in the jejunum and ileum (Figure 2C) and reversed the increased Il-17 expression in mice infected with ETEC (Figure 2D). Similar to a previous report (27), rapamycin also inhibited Th17 cell differentiation from naïve T cells under Th17 polarization conditions in vitro (Figures 2E). In conclusion, ETEC infection promotes IL-17 expression through the activation of mTORC1 pathway.

Among the downstream targets of mTORC1 signaling, HIF-1α and S6K (S6K1 and S6K2) can regulate IL-17 expression (26). mTORC1 signaling activates HIF-1α, which promotes IL-17 expression by activating RORγt and mediating Foxp3 degradation (26). ETEC infection decreased HIF-1α abundance and increased Foxp3 abundance in the cytoplasm of mouse jejunum compared to the controls (Figure 3A). Rapamycin treatment before ETEC infection increased HIF-1α abundance and lowered Foxp3 abundance (Figure 3A). S6K2 (the nuclear-localized counterpart of S6K1) binds to RORγt to promote the nuclear translocation of RORγt, which can complex with HIF-1α and p300 in the nucleus to promote IL-17 expression (26). Although ETEC infection increased RORγt abundance, rapamycin had no effect on RORγt abundance (Figure 3B, top). Similarly, rapamycin did not affect on the abundance of HIF-1α and S6K2 in the nucleus of the mouse jejunum (Figure 3B, bottom), although ETEC infection reduced the nuclear abundance of HIF-1α and S6K2 in mouse jejunum (Figure 3B, bottom).

S6K1 promotes the expression of EGR-2, resulting in the inhibition of the GFI-1, which negatively regulates IL-17 expression without affecting Rorc expression (26). Although ETEC infection had little affect on Egr-2 expression in the mouse jejunum, rapamycin reduced Egr-2 expression in the mouse jejunum (Figure 3C). ETEC infection reduced Gfi-1 expression in the mouse jejunum, while rapamycin prevented this reduction (Figure 3C). At the protein level, ETEC infection increased the abundance of EGR-2 and lowered the abundance of GFI-1 in the mouse jejunum, while rapamycin abolished the effects of ETEC infection (Figure 3D). These results indicate that ETEC induces intestinal IL-17 expression may through the S6K1–EGR-2–GFI-1 axis. Intriguingly, ETEC infection increased Il-17 expression in the jejunum, while PF-4708671 (S6K1 specific inhibitor) reduced Il-17 expression in the jejunum in ETEC-infected mice (Figure 3E). ETEC infection increased the abundance of EGR-2 and decreased the abundance of GFI-1 in the mouse jejunum, while PF-4708671 treatment reversed this change during ETEC infection in mouse jejunum (Figure 3F). These data indicate ETEC promotes the IL-17 expression in the jejunum through the mTOR–S6K1–GFI-1 axis.

Mechanistic target of rapamycin complex 1 is activated by amino acids and by the phosphatidylinositol 3-kinase (PI3K)/Akt pathway, while it is inactivated by the AMPK (28, 29). ETEC infection inhibited the PI3K/Akt pathway and activated AMPK in piglet jejunum (Figures 4A,B). Thus, the activation of mTORC1 in response to ETEC infection may result from changes in amino acid concentration in the jejunum. Using GC–MS analysis, we found that eight amino acids significantly changed in abundance after ETEC infection (Figure 4C). The relative abundance of spermidine, glutamine, asparagine, citrulline, and ornithine decreased, while the relative abundance of glycine, taurine, and GABA increased after ETEC infection (Figure 4C). However, there was little change in mRNA expression of the activating transcription factor 4 (Figure 4D), an indicator of amino acid starvation (30), suggesting there was no overall amino acid starvation after ETEC infection.

As GABA was the most significantly increased amino acid (about fourfold), we hypothesized that GABA is responsible for increased IL-17 expression and mTORC1 activation during ETEC infection. Indeed, numerous investigations have shown that GABA signaling promotes mTORC1 activation (31–33). Treating mice with bicuculline (an antagonist of GABAA receptors) or with CGP-35348 (an antagonist of GABAB receptors) lowered Il-17 expression in the jejunum during ETEC infection (Figure 4E). Bicuculline and CGP-35348 treatment also reversed the activation of mTORC1–EGR-2 signaling in mouse jejunum induced by ETEC infection (Figure 4F). GABA injection promoted Il-17 expression in mouse jejunum during ETEC infection, while rapamycin prevented expression of Il-17 caused by GABA injection (Figure 4G). Furthermore, treatment with l-allylglycine or semicarbazide (blockers of synthetic enzyme for GABA) reduced Il-17 expression in mouse jejunum during ETEC infection (Figure 4H). l-Allylglycine or semicarbazide treatment also prevented the activation of mTORC1–EGR-2 signaling in mouse jejunum caused by ETEC infection (Figure 4I).

To investigate whether GABA signaling regulates Th17 cell differentiation, naïve CD4⁺ T cells were isolated and differentiated into Th17 cells under Th17 polarization conditions. In normal Th17 differentiation condition, about 3.0% of CD4⁺ T cells were differentiated into CD4⁺IL-17⁺ T cells after 3 days of differentiation; while l-allylglycine or semicarbazide decreased the percentage of CD4⁺IL-17⁺ T cells (Figure 4J). After 3 days of differentiation under normal Th17 polarization conditions, CD4⁺ T cells had higher abundance of phosphorylated mTORC1, compared to the naïve CD4⁺ T cells (Figure 4K). Also, l-allylglycine decreased the abundance of phosphorylated mTORC1 (Figure 4K). Collectively, GABA signaling promotes IL-17 expression during infection though mTORC1–EGR-2 signaling.

Intestinal microbiota has critical roles in GABA production from glutamate (34, 35) and in regulation of intestinal IL-17 expression (36, 37). Our previous study has found that ETEC-infected piglets had a higher percentage of Lactococcus, compared to the controls (manuscript submitted). This result was supported by our current finding that ETEC-infected mice had higher numbers of L. lactis subsp. lactis, as compared with uninfected controls (Figure 5A). Within the intestinal microbiota, L. lactis subsp. lactis is one of the main GABA-producing strains (38, 39). Thus we hypothesized that the increased abundance of intestinal L. lactis subsp. lactis during ETEC infection is responsible for activated GABA–IL-17 signaling in the context of ETEC infection. Indeed, we found that piglets orally inoculated with L. lactis subsp. lactis (ATCC19435) has higher mRNA expression of IL-17 in the jejunum compared to the piglets without L. lactis subsp. lactis inoculation (Figure 5B). The GABA production activity of L. lactis subsp. lactis (ATCC19435) is about 210.4 µM (38, 39). Similar to the observation from piglets, L. lactis subsp. lactis inoculation in mice significantly promoted the expression of Il-17 in the jejunum compared to the mice without L. lactis subsp. lactis inoculation (Figure 5C). More importantly, inhibition of GABA signaling by CGP-35348 decreased the expression of Il-17 in the jejunum of mice with L. lactis subsp. lactis inoculation (Figure 5C). To further test this hypothesis, we treated mice with a mixture of antibiotics for 1 week to decrease the load of L. lactis subsp. lactis before ETEC infection. To increase the abundance of L. lactis subsp. lactis in antibiotics-treated mice, we orally inoculated the L. lactis subsp. lactis to antibiotics-treated mice. The antibiotics-treated mice had a lower numbers of L. lactis subsp. lactis than the mice without antibiotics treatment (Figure 5D). After ETEC infection, we found that antibiotics-treated mice had lower expression of Il-17 in the jejunum compared to the mice without antibiotics treatment (Figure 5E). Oral inoculation of L. lactis subsp. lactis before ETEC infection increased the numbers of L. lactis subsp. lactis in the content of jejunum and rescued the expression of Il-17 during ETEC infection (Figures 5D,E). Interestingly, in antibiotics-treated mice, higher dosage of GABA supplementation also rescued the expression of Il-17 in the jejunum during ETEC infection (Figure 5F). In germ-free mice, L. lactis subsp. lactis administration before ETEC infection also promoted expression of Il-17 (Figure 5G). Collectively, intestinal GABA-producing bacteria have vital importance in IL-17 expression during ETEC infection.

To validate the role of GABA–mTORC1 signaling in IL-17 expression in other intestinal infectious models, we infected mice with different ETEC strains and with Shiga-like toxin producing E. coli. In mouse model, E. coli W470, W197, W817, and W616 increased Il-17 expression in the jejunum at 6-h postinfection (Figure 6A). Pretreatment with rapamycin prevented expression of Il-17 in the jejunum caused by W470, W197, W817, and W616 at 6-h postinfection (Figure 6A). Similar to our previous W25K infection, bicuculline treatment reversed the increase in Il-17 expression caused by W470 in mouse jejunum (Figure 6B). In C. rodentium infected mice, the expression of Il-17 in the colon was higher than in control mice (Figure 6C). Interestingly, higher abundance of L. lactis subsp. lactis was also observed in the colon during C. rodentium infection (7.7 ± 0.03 log₁₀ CFU/g in infected mice vs. 6.9 ± 0.08 log₁₀ CFU/g in control mice). Rapamycin treatment lowered the colonic expression of Il-17 (Figure 6C), indicating that inhibition of mTORC1 pathway reduced the expression of Il-17 in the colon during C. rodentium infection. Furthermore, the inhibition of GABA signaling by semicarbazide, bicuculline, or CGP-35348 also lowered the expression of Il-17 in the colon in the context of C. rodentium infection (Figure 6C). 5-Fluorouracil (5-FU) is frequently used in cancer treatment and also induces intestinal inflammation (40, 41). 5-FU enhanced Il-17 expression in the jejunum at 6-h postinjection (Figure 6D). Semicarbazide or bicuculline treatment prevented the increase in Il-17 expression caused by 5-FU (Figure 6D). In summary, GABA–mTORC1 signaling also functions in intestinal IL-17 expression in other intestinal inflammatory models.