Selective Induction of Homeostatic Th17 Cells in the Murine Intestine by Cholera Toxin Interacting with the Microbiota Free

This article is featured in In This Issue, p.1

There are trillions of microbes residing in the normal mammalian intestine (1). These microbes include bacteria, archaea, fungi, and viruses, of which bacteria are the most abundantly studied. The role of intestinal microbiota in the induction and modulation of the host immune system has been increasingly recognized in recent years (2, 3). At steady-state, the mammalian intestine is the biggest reservoir of activated effector T cells, which provide protection against potential intestinal pathogens together with other immune cell subsets. Th17 cells are among the most abundant effector CD4⁺ T cells in the intestinal lamina propria (LP) (4). They are characterized by the expression of master transcription factor Rorγt and the production of cytokines including IL-17A, IL-17F, IL-21, and IL-22 (5). Th17 cells show heterogeneity in their gene expression profile in different research models and conditions. For instance, they are involved in the pathogenesis of an array of autoimmune diseases, including rheumatoid arthritis, multiple sclerosis, experimental autoimmune encephalomyelitis, psoriasis, and inflammatory bowel diseases (IBD). Such pathogenic Th17 cells frequently coexpress IFN-γ and TNF-α (6). In other studies, Th17 cells have been shown to be indispensable in fighting against respiratory fungal infections and intestinal bacterial pathogens via the production of IL-17A and IL-22 (7).

Segmented filamentous bacteria (SFB) are a group of Gram-positive, spore-forming commensal bacteria that form long filaments and preferentially adhere tightly to the epithelium and Peyer’s patches of the terminal ileum. Colonization with SFB strongly induces endogenous Th17 cells, as well as a broad range of other proinflammatory cytokines in the murine small intestine (8, 9). Such induction is considered as a beneficial stimulation of the immune system and is shown to provide mice with better protection against Citrobacter rodentium infection. However, SFB may be only one among a large pool of microbes that are likely to elicit Th17 cells (10), and other bacterial species that could do so still await identification.

Cholera toxin (CT), which is a potent mucosal immunogen and adjuvant, also has the capacity to induce mucosal Th17 cell differentiation after intranasal administration (11). Mucosal delivery of CT induces strong mucosal and systemic humoral immune responses to itself and coadministered protein Ags (12), but does not cause intestinal inflammation. CT does so by activating Gsα and the nucleotide-binding oligomerization domain containing 2 expressed in CD11c⁺ cells (13, 14). Th17 cells can convert into T follicular helper cells in the Peyer’s patches, thus enhancing high-affinity intestinal IgA production (15), suggesting an important role of Th17 cells in the induction of Ag-specific IgA in the gut, which in turn is involved in the protection against microbial Ags at the mucosal surface (16).

IgA deficiency is the most commonly seen primary immunodeficiency (17) and has been shown to relate to IBD pathogenesis on genome-wide association studies (18). Interestingly, in mice that are deficient of IgA, we observed increased Th17 cells in their normal small intestine, which can be further augmented with mucosal CT immunization without inducing any inflammation. This increased Th17 and Ab response to CT in IgA^−/− mice was dependent on intestinal microbiota. Experiments with gnotobiotic mice indicate that not only SFB, but also other commensal bacteria participate in CT induction of the Th17 response. Moreover, we performed mRNA sequencing analysis and found that CT-induced intestinal Th17 cells share a similar expression profile with endogenous intestinal Th17 cells at homeostasis, and both of them are distinctly different from pathogenic Th17 cells.

C57BL/6 mice and IL-6^−/− mice were purchased from The Jackson Laboratory. IgA^−/− mice were kindly provided by Dr. I.N. Mbawuike from Baylor College of Medicine (Houston, TX). 10BiT.Foxp3^gfp mice were kindly provided by Dr. C.L. Maynard (University of Alabama at Birmingham [UAB]). IgA^−/−.10BiT.Foxp3^gfp mice were generated by crossing 10BiT.Foxp3^gfp mice with IgA^−/− mice. IL-17A^hCD2 mice were generated and kindly provided by Dr. S.N. Harbour and Dr. C.T. Weaver (UAB). IgA^−/−.IL-17A^hCD2 mice were generated by crossing IL-17A^hCD2 mice with IgA^−/− mice, using the latter as the dam. All mice were on C57BL/6 background and were maintained in the Animal Facility at the UAB under specific pathogen-free (SPF) conditions. Germ-free (GF) mice derivation and maintenance were as described previously (19). For inoculation of GF/altered Schaedler flora (ASF) mice with SFB, fecal pellets were purchased from the Center for Gastrointestinal Biology and Disease Gnotobiotic Core of the University of North Carolina (collected from SFB-monocolonized mice using sterilized test tubes in the vinyl-isolator) and were preserved frozen under dry ice or at −80°C until immediately before oral administration. SFB colonization was performed by oral gavage with 300–400 μl of suspension obtained by homogenizing the fecal pellets from SFB-monocolonized mice in water. Fecal pellets of the recipients were collected 2 wk after gavage, and SFB colonization was confirmed with PCR (primers: forward, 5′-ACG CTA CAT CGT CTT ATC TTC CCG C-3′; reverse, 5′-TCC CCC AAG ACC AAG TTC ACG A-3′). Vancomycin treatment was performed with indicated schedules with 0.5 mg/d in the drinking water. For intragastric immunization, 10 μg of CT and 1 mg of OVA in 500 μl 0.2 mM NaHCO₃ was gavaged on days 0, 7, and 14. Mice were sacrificed on day 28 for examination. Intracolonic immunization was performed with 10 μg of CT and 1 mg of OVA in 100 μl of PBS, with the same immunization schedule. Short-chain fatty acid repletion was performed with 100 mM butyrate or 150 mM propionate in the drinking water from day 0 of CT immunization until mice were sacrificed. Eight- to 12-wk-old mice were used in these experiments unless specified otherwise. All experiments were reviewed and approved by the Institutional Animal Care and Use Committee of UAB.

Reagents and materials were purchased from the following sources: anti-mouse CD4 (RM4-5), CD25 (PC61) Abs were purchased from BD Biosciences; anti-mouse IL-17A (TC11-18H10.1), IFN-γ (XMG1.2), CD90.1 (OX7) Abs were purchased from BioLegend; anti-mouse Foxp3 (FJK-16s) and IL-22 (1H8PWSR) Abs, as well as Foxp3 staining buffer set, were purchased from eBioscience; anti-mouse CD90.1 (19E.2) and anti-human CD2 (OKT11) were purchased from the hybridoma laboratory of Epitope Recognition and Immunoreagent Core at UAB; live/dead staining kit was purchased from Invitrogen; mouse LP dissociation kit and regulatory T cell isolation kit were purchased from Miltenyi Biotec; mouse lipocalin-2 (LCN-2) ELISA kit was purchased from R&D Systems; fecal DNA extraction kit was purchased from Zymo Research.

Small and large intestines were removed, sliced, and washed with ice-cold PBS to remove fecal content. After removing epithelium by gentle shaking in Ca²⁺/Mg²⁺-free HBSS supplemented with 5 mM EDTA, 1 mM DTT, and 2% FBS for 40 min at 37°C, the tissue was washed and resuspended in digestion medium following Miltenyi protocol and then incubated for 30 min at 37°C by gentle shaking. Cells were passed through a mesh, centrifuged, and the pellet was resuspended in 40% Percoll and carefully overlaid onto 70% Percoll. The interface containing the LP leukocytes was collected and washed with PBS supplemented with 10% FBS. A total of 10⁶ LP lymphocytes (LPLs) were cocultured with 5 × 10⁴ APCs isolated from naive C57BL/6 mouse spleen (CD4⁻ fraction) with the addition of either SFB, CT B subunit (CTB), or OVA CD4 T cell epitope peptide (SFB 568–580, DVQFSGAVPNKTD; CTB 89–100, NNKTPHAIAAIS; OVA 323–339, ISQAVHAAHAEINEAGR) (20–22) for 5 d. Culture supernatant was then collected and measured for cytokine production.

As described previously (23), isolated cells were stimulated for 4–5 h with PMA (50 ng/ml; Sigma) and ionomycin (750 ng/ml; Sigma) in the presence of GolgiStop (10 μg/ml; BD Biosciences). After staining of cell surface Ags, cells were fixed and permeabilized using Foxp3 staining buffer set (eBioscience), followed with intracellular cytokine staining. Dead cells were excluded by surface staining with Live/Dead staining kit. Flow cytometry was performed on BD LSR II and analyzed with FlowJo software (V9.3.3; Tree Star).

Ninety-six-well ELISA plates (MaxiSorp; Thermo Fisher Scientific) were coated with capture Ab in PBS or coating buffer at 4°C overnight. Plates were washed and blocked; then samples and standards were added and incubated at room temperature for 2 h. After washing, biotin-labeled detection Abs and streptavidin-HRP were added and incubated for another hour. Followed by final washes, plates were developed with TMB solutions for 15 min and read at 450 nm. Sample concentrations were calculated according to the standard curve. Milliplex mouse Th17 magnetic kit was purchased from Millipore. Multiplex assay was conducted following manufacturer’s instructions. Nine cytokines including IFN-γ, IL-4, IL-5, IL-13, IL-6, IL-17A, IL-21, IL-22, and IL-10 were measured simultaneously.

Naive IgA^−/− mice at 9 mo old and CT-immunized mice at 12 wk old (immunization started at 8 wk old with the schedule described earlier) were sacrificed for histopathologic assessment. At necropsy, the small intestine, cecum, and colon were separated and cut open longitudinally. Tissues were fixed in 10% buffered formalin and paraffin embedded. The sections (5 μm) were stained with H&E. All slides were scored by an experienced veterinary pathologist (T.R. Schoeb, UAB) without knowledge of their origin.

Bacterial genomic DNA from mouse ileal and cecal contents was extracted with ZR Fecal DNA MiniPrep kit following manufacturer’s protocol. The V4 region of 16S rRNA gene was targeted for PCR amplification using a modified F515/R806 primer (24). Duplicates were run for all PCRs. The 150 paired-end sequencing reaction was performed using the MiSeq platform (Illumina) at the Gut Microbiome and Large Animal Biosecurity Laboratories, Department of Animal Science, University of Manitoba (Winnipeg, MB, Canada). The FLASH assembler was used to merge overlapping paired-end Illumina fastq files (25). The output fastq file was then further analyzed by downstream computational pipelines of the open source software package QIIME (26). Taxonomies were assigned to the representative sequence of each OTU using RDP classifier and were aligned with the Greengenes Core reference database using PyNAST algorithms. Within-community diversity (α-diversity) was calculated using QIIME. An α rarefaction curve was generated using Chao 1 estimator of species richness with 10 sampling repetitions at each sampling depth. An even depth of ∼13,100 sequences per sample was used for calculation of richness and diversity indices. To compare microbial composition between samples, we measured β-diversity by calculating the weighted and unweighted Unifrac distances (27) using QIIME default scripts to generate principal coordinate analysis plot and generated nonmetric multidimensional scaling plots using Bray-Curtis distance in R with package “picante.” Permutational multivariate ANOVA was used to calculate p values and test for significant differences of β-diversity among treatment groups.

CD4⁺CD25⁻ cells were purified from spleen with regulatory T (Treg) cell isolation kit (Miltenyi Biotech) and polarized to pathogenic Th17 cells with plate-bound anti-CD3 (5 μg/ml; 145-2C11; BD Biosciences) and anti-CD28 (2 μg/ml; 37.51; BD Biosciences) in the presence of recombinant human TGF-β1 (5 ng/ml; R&D Systems), recombinant mouse IL-6 (20 ng/ml; R&D Systems), recombinant mouse IL-23 (20 ng/ml; R&D Systems), and recombinant mouse IL-1β (20 ng/ml; R&D Systems). Cells were cultured for 5 d, and CD4⁺hCD2⁺ cells were surface stained with fluorescent Abs and sorted with BD FACSAria and stored in lysis buffer. mRNA was isolated using Dynabeads mRNA DIRECT purification kit (Life Technologies) following manufacturer’s protocol. RNA sequencing (RNA-Seq) and data analysis were performed at the Broad Institute of MIT and Harvard. In brief, cDNA libraries were generated using the Smart-seq2 protocol, modified for 500+ cells of input, and sequenced on a MiSeq (Illumina) per manufacturer’s instructions. Sequences were aligned against the Mus musculus genome (mm10) using TopHat + Bowtie (28). HTSeq-count (29) was used to count the transcripts associated with each gene, and a counts matrix containing the number of counts for each gene across different samples and stimulations was obtained. The counts were normalized using the TMM method (30). To analyze differential expression across different stimulations, we analyzed the counts matrix with a generalized linear model, using the edgeR package (31) with a threshold of q ≤ 0.05.

Unpaired Student t test was used for the determination of difference between groups unless other methods were stated. Data were presented as mean ± SEM of replicate experiments. The differences were considered statistically significant at p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Intestinal LPLs isolated from 9-mo-old naive C57BL/6.IgA^−/− mice were compared with sex- and age-matched naive C57BL/6 wild-type (WT) mice and found to have a significantly enlarged Th17 population in the small intestine compared with WT mice (Fig. 1A–C), a difference not observed in 8-wk-old naive mice (Fig. 1E–G, control group). Correspondingly, there was a significantly higher serum LCN-2 level in elderly IgA^−/− mice compared with matched WT mice (Fig. 1D), which is an antimicrobial protein that is downstream of IL-17A stimulation and plays important roles in epithelial homeostasis (32–34).

Next, 8- to 12-wk-old IgA^−/− and WT mice were immunized with CT and CBir1 flagellin (a microbiota Ag) intragastrically. Th17 cells in small intestinal LPLs (siLPLs) of both strains increased considerably compared with control mice, and this increase was significantly higher in IgA^−/− mice than in WT mice. The induction of Th17 cells in the large intestine was much less profound than that in the small intestine (Fig. 1E–G). Different from a previous report investigating cellular response in the peripheral lymphoid organs after intranasal and i.v. CT immunization (13), we did not observe significant changes of Th1, Th2, or Treg cells either in the mucosal or in the peripheral sites (data not shown). siLPLs isolated from IgA^−/− mice secreted almost 10-fold more IL-17A than those from WT mice upon Ag restimulation in vitro (Fig. 1H). Also, IgA^−/− mice developed 10-fold higher serum IgG responses against CTB (Fig. 1I). Although Th17 cells have been shown to contribute to pathogenicity in various colitis models, this high percentage of endogenous or CT-induced Th17 cells in the small intestine did not cause spontaneous inflammation in IgA^−/− mice in that all sections of intestinal tissue from duodenum to distal colon were normal on histologic evaluation by an experienced pathologist (T.R.S.) (Fig. 1J, data not shown), consistent with a homeostatic role of gut Th17 cells at steady-state.

It has been shown that long-term isolated breeding of mutant and WT mice can cause marked compositional differences in their gut microbiota (35), which plays a critical role in immune system development. Thus, we asked whether the gut microbiota or the lack of IgA accounted for the enlarged endogenous and CT-induced Th17 responses in IgA^−/− mice. First, we tested whether the gut microbiota in IgA^−/− mice differed from WT mice by comparing the bacterial composition in the ileum and cecum of age- and sex-matched adult mice. Pie charts of the phyla show that IgA^−/− mice had higher proportions of Firmicutes than WT mice in both ileum and cecum (Fig. 2A). In contrast, WT mice had more Bacteroidetes, Tenericutes, and Actinobacteria. Principal coordinate analysis shows that bacteria in IgA^−/− mice cluster separately from WT mice in both ileal and cecal samples (Fig. 2B), indicating that the composition of the intestinal microbiota of IgA^−/− mice differs from that of WT mice. Because Firmicutes and Bacteroidetes are the two major phyla of intestinal commensals, and are closely related to intestinal immune homeostasis, we compared the ratio of Firmicutes over Bacteroidetes in IgA^−/− and WT mice. In both ileum and cecum, this ratio was significantly higher in IgA^−/− mice (Fig. 2C). Further analysis showed that a large part of Firmicutes in the ileum of IgA^−/− mice was SFB, comprising ∼40% of the total detected bacteria, much higher than that in WT mice (Fig. 2D). Because SFB is a potent inducer of endogenous Th17 cells in the small intestine, we asked whether differences in microbiota composition, in particular elevated SFB in IgA^−/− mice, could promote stronger CT-induced Th17 and Ab responses. To test this hypothesis, we equalized the intestinal microbiota of these two strains by crossing female IgA^−/− mice to male WT mice and then interbreeding of the heterozygous first generation (F1) mice to obtain littermate IgA^−/− and IgA^+/+ mice in the second generation (Fig. 2E). The intestinal microbiota of all offspring were transmitted from maternal IgA^−/− mice and remained equivalent in adulthood (Supplemental Fig. 1). Homozygous IgA^−/− and WT littermates in the F2 generation were immunized with CT and OVA intragastrically. Surprisingly, the previously seen differences in intestinal Th17 induction and Ag-specific serum IgG responses did not occur (Fig. 2F, 2G), indicating it was the microbiota composition in IgA^−/− mice that led to the elevated host Th17 responses. Furthermore, naive littermate IgA^−/− and WT mice had comparable levels of serum LCN-2 as well (Fig. 2H). Interestingly, because offspring WT mice were colonized with intestinal microbiota from the original IgA^−/− mothers, their immune responses upon CT immunization resembled that of IgA^−/− mice, but not the parent WT strain, providing strong evidence that IgA-deficient mice are not intrinsically colonized by higher levels of Firmicutes, including SFB, which is different from a previous report (36).

To investigate how intestinal microbiota modulates immune responses to CT immunization, we treated IgA^−/−.10BiT.Foxp3^gfp mice, which were derived from IgA^−/− mice and express a Thy1.1 (CD90.1) reporter for IL-10 expression and a GFP reporter for Foxp3 expression (37), with antibiotics in their drinking water for 2 wk to ablate Gram-positive, spore-forming bacteria, including SFB. Ablation of SFB was confirmed with PCR using DNA extracted from their fecal pellets (data not shown). IgA^−/−.10BiT.Foxp3^gfp mice receiving vancomycin or regular water were immunized intragastrically with CT and OVA as described earlier. On day 28 of immunization, mice receiving vancomycin developed 1 log less anti-CTB and anti-OVA IgG responses in their sera compared with mice kept on regular water (Fig. 3A), indicating that Gram-positive intestinal bacteria are actively regulating systemic immune responses to CT/OVA immunization.

Two-week treatment with vancomycin did not reduce intestinal Th17 on its own (data not shown); however, when we treated parent IgA^−/−.10BiT.Foxp3^gfp mice with vancomycin for 1 mo before breeding and during gestation, offspring had greatly decreased intestinal Th17 cells, as well as IL-17A and IL-10 double-expressing cells in all parts of their small intestine compared with sex- and age-matched SPF IgA^−/−.10BiT.Foxp3^gfp mice (Fig. 3B, 3C). Th17 cells in the colonic LP or other T cell subsets in the intestine were not significantly affected, indicating that Gram-positive bacteria ablated by vancomycin are mainly present in the small intestine and locally modulate siLPLs, consistent with our microbiome sequencing data showing SFB outgrowth in the terminal ileum of parent IgA^−/− mice (Fig. 2D).

To test whether SFB is playing a role in modulating siLPLs in response to intragastric CT immunization, we immunized 8-wk-old IgA^−/−.10BiT.Foxp3^gfp mice born from vancomycin-treated parents lacking SFB and divided the small intestine equally into three parts when examined. Using naive littermate controls that either had SFB or not, we found that SFB alone was able to induce endogenous Th17 cells as well as IL-17A and IL-10 double-producing cells in the small intestine, but this induction was the most prominent in the middle and lower thirds. In contrast, CT immunization increased Th17 cells in all parts of the small intestine, but did not induce IL-17A and IL-10 double-producing cells (Fig. 3D).

Consistent with previous reports (38, 39) that intracolonic CT immunization induces systemic humoral immune responses, our data show that intracolonic CT immunization induced intestinal Th17 cells as well. The Ab response to intracolonic CT immunization was also microbiota dependent, because vancomycin-treated C57BL/6 mice had impaired anti-CTB fecal IgA as well as total fecal IgA after immunization (Fig. 3E, data not shown). Addition of the bacterial fermentation products butyrate or propionate to the drinking water (40–42) partially rescued the production of anti-CTB fecal IgA, as well as total fecal IgA (Fig. 3E), indicating that decreased immune responses to CT after ablation of Gram-positive bacteria was partially due to deficiency of bacterial metabolic products in the intestine. Collectively, these findings indicate that Gram-positive bacterial ablation with vancomycin impairs the host intestinal Th17 response and its related Ab responses to CT immunization. SFB specifically induces an IL-17A⁺IL-10⁺ population in the distal small intestine, indicating that the intestinal microbiota can modulate the host Th17 response in some ways that are different from CT.

To better define the effect of SFB versus other microbes on the ability of CT to induce Th17 cells, we used gnotobiotic mice. GF and ASF colonized C57BL/6 mice (43) were selectively colonized with SFB to further test whether SFB alone is sufficient to modulate immune responses against CT or whether other commensal bacteria also contribute to this regulation. In both GF and ASF mice, SFB colonization alone induced Th17 cell differentiation in the small intestine and cecum (Fig. 4A). However, the increase of Th17 cells in the small intestine in SFB monocolonized mice versus GF mice was not statistically significant (Fig. 4A, 4B). ASF colonization did not induce Th17 cells by itself and largely suppressed Th1 cell differentiation in both small intestine and cecum (data not shown). In the absence of SFB, CT/OVA immunization significantly induced Th17 cells in the small intestine and cecum of ASF mice, but only slightly in GF mice, indicating that single or multiple bacteria belonging to ASF also have the potential to regulate Th17 response upon CT immunization (Fig. 4A, 4B). SFB predominantly enhanced Th17 cell induction in response to CT immunization in monocolonized mice, although not significantly in ASF mice, which already had a higher level (Fig. 4A). Cytokine production (including Th1-, Th2-, and Th17-related cytokines) was measured by coculturing of isolated siLPLs with naive APCs and either SFB, CTB, or OVA CD4 T cell epitope peptide (see 2Materials and Methods). Intragastric CT immunization specifically induced LP IL-17A response (Fig. 4C) and slightly suppressed IFN-γ response, without a significant effect on IL-4, IL-5, IL-13 response (data not shown). Although peptide restimulation resulted in increased IL-6 production in CT-immunized groups, IL-6 appeared not to be the sole pathway that contributed to Th17 differentiation in vivo, because there was no difference in Th17 cell induction or Ab responses with CT immunization in IL-6^−/− mice (Supplemental Fig. 2). Upon SFB colonization, both GF and ASF mice receiving CT/OVA immunization significantly increased IL-17A production specific to SFB peptide stimulation, to a much greater extent than either CTB or OVA peptide, indicating that CT serves as an adjuvant for the proliferation of Th17 cells that are predominantly reactive to SFB. Collectively, these data provided clear evidence that CT induction of intestinal Th17 response is dependent on the gut microbiota. Indeed, siLPLs isolated from CT-immunized SPF WT mice responded significantly more strongly than those of any group of gnotobiotic mice (Supplemental Fig. 3A, 3B), indicating that a more diverse intestinal microbiota provides optimal support for the development of host Th17 response to CT immunization. Although CT elicits strong B cell and T cell responses to itself, tetramer staining showed that most of the induced Th17 cells are not CTB-specific, indicating that CT induces Th17 response in a polyclonal fashion (Supplemental Fig. 3C, 3D).

SFB colonization is a strong inducer of both fecal and serum IgA in gnotobiotic mice, whether naive or immunized (Fig. 4D). In contrast, ASF colonization enhanced fecal secretory IgA levels, but not serum IgA, indicating that different commensals have distinct mechanisms to regulate host IgA production. Surprisingly, in contrast with our vancomycin experiment data, SFB colonization did not increase host Ab responses to CT or OVA in either GF mice or ASF mice; that is, SFB did not act as an adjuvant for the response to these Ags. However, regardless of the presence of SFB, ASF-colonized mice developed greater CTB- and OVA-specific Ab responses upon immunization than GF mice both mucosally and systemically (Fig. 4E), indicating that the decreased Ab responses upon CT immunization in vancomycin-treated mice were due to vancomycin-sensitive commensal bacteria other than SFB. These experiments suggested that, in addition to SFB, which is a known potent inducer for endogenous Th17 cells in the intestine, other commensal bacteria are also involved in the regulation of host Th17 and Ab responses upon stimulation.

To test the hypothesis that intestinal Th17 cells under steady-state conditions are different from pathogenic Th17 cells, we used IgA^−/−.IL-17A^hCD2 reporter mice, in which the IL-17A promoter directs the transcription of a bicistronic mRNA such that IL-17A–producing cells are marked by surface membrane expression of the human CD2 molecule (Supplemental Fig. 4A; S.N. Harbour, H. Turner, C.L. Maynard, C.L. Zindl, R.D. Hatton, and C.T. Weaver, manuscript in preparation), allowing us to recover IL-17A–producing cells for RNA-Seq. Of note, almost none of the IL-17A–producing cells coexpress Foxp3 (Supplemental Fig. 4B). We isolated CD4⁺hCD2⁺ (Th17) and CD4⁺hCD2⁻ (non-Th17) cells from the small intestine of either naive or CT-immunized IgA^−/−.IL-17A^hCD2 mice and used splenic CD4⁺CD25⁻ cells from naive mice as negative controls for gene expression profiling. Comparing CD4⁺ cells isolated from naive and CT-immunized IgA^−/−.IL-17A^hCD2 intestines, we found that Th17 cells induced by CT only have one gene differently expressed compared with endogenous LP Th17 cells; similarly, non-Th17 cells isolated from CT-immunized mice only had two genes differently expressed compared with non-Th17 cells isolated from naive mice. Thus, although mucosal CT immunization enlarges the Th17 population in the intestinal LP, CT-induced Th17 cells share the same expression profile as endogenous intestinal Th17 cells. The gene expression of Th17 cells isolated from the healthy intestine or CT-immunized intestine is very different from pathogenic Th17 cells induced with IL-6+IL-23+IL-1β⁴⁴, as well as Th17 cells isolated from the LP of colitic tissues (Fig. 5, data not shown). In particular, a series of genes related to Th17 pathogenicity, including Tnf, Ifng, Il23a, Il24, Il12rb1, and Gzmb, were significantly less expressed in endogenous and CT-induced intestinal Th17 cells compared with pathogenic Th17 cells, whereas genes involved in anti-inflammatory activities, such as Ctla4, Icos, Il22, Ahr, Maf, and Ikzf3, showed at least 4-fold higher expression in homeostatic intestinal Th17 cells, consistent with previously reported Th17 signatures (44–47). Additionally, functional analysis reveals a stronger immune-regulatory profile of homeostatic intestinal Th17 cells, whereas pathogenic Th17 cells highlight pathways that are involved in various inflammatory diseases. In summary, comparison of gene expression levels shows great similarity between CT-induced Th17 cells and endogenous Th17 cells isolated from the healthy small intestine, and both of them are distinct from the signature of pathogenic Th17 cells.

In this study, we showed that endogenous Th17 cells present in the normal intestine have a homeostatic phenotype. The increase of this homeostatic population and the elevated Th17 response upon CT immunization in IgA^−/− mice are dependent on intestinal microbiota. Using gnotobiotic mouse models, we found that SFB, ASF, and other commensal bacteria are actively involved in the regulation of CT immunogenicity and adjuvanticity. Furthermore, our RNA-Seq data indicate that CT-induced intestinal Th17 cells share a similar gene expression profile with endogenous Th17 cells in the intestine, both of which are quite different from the signature of pathogenic Th17 cells.

The age-related increase of intestinal Th17 cells in IgA^−/− mice suggested an environmental effect of Th17 cell development, in this case most likely the intestinal microbiota. Indeed, with activation-induced cytidine deaminase knockout mice, Suzuki et al. (36) concluded that the lack of hypermutated IgA resulted in the excessive overgrowth of anaerobic bacteria in the intestine, the most predominant being SFB. Although conflicting data have been reported with Igh-J^−/− mice (48), neither of them were perfect models for the study of IgA deficiency. Targeted deletion of the IgA C region in our IgA^−/− mice leads to specific deficiency of IgA, thus providing a more appropriate model of the effects of IgA deficiency on microbiota (49). Our study shows that increased Firmicutes, including SFB, led to the exaggerated intestinal Th17 responses and humoral Ab induction upon CT immunization in IgA^−/− mice. However, this microbiota difference is not intrinsic to IgA deficiency, in that crossing the WT with IgA^−/− mice equalized the intestinal microbiota in the offspring, indicating a familial transmission, rather than IgA deficiency, caused microbiota shift (35, 36, 50).

CT has been known as a potent mucosal immunogen and adjuvant for several decades; we confirm and extend a recent report linking these CT properties with the intestinal microbiota (14). With SFB as a known stimulator of endogenous Th17 cells in the intestine, we found that in addition, other commensal bacteria such as those belonging to the ASF can also participate in the regulation of intestinal Th17 and Ab responses to CT mucosal immunization. When the CTB subunit binds to the GM1 ganglioside present on cell membranes, CT induces strong activation of STAT3 signaling and massive production of IL-6 (11, 51, 52), which is known to promote the differentiation of Th17 cells in vitro. However, in agreement with previously published in vitro studies (53, 54), IL-6^−/− mice did not show impaired intestinal Th17 cell induction upon CT immunization compared with littermate WT mice (Supplemental Fig. 2), implying the involvement of other possible cytokine pathways such as IL-21 or IL-1β. Although both SFB and CT are strong inducers of intestinal Th17 cells, the induction tends to reach a saturated level at ∼35–40% of CD4⁺ T cells in siLPLs, even in IgA^−/− mice. In addition, CT had little effect on other CD4⁺ T cells in the intestine, including Th1, Th2, and Treg cells. This difference from the conclusions of a previous report (13) was probably due to different microbiota in mice maintained in isolated facilities.

In this study, we show that both intragastric and intracolonic CT immunization induces profound Th17 cell increase in the intestinal LP, without causing any intestinal inflammation. Data from a collection of studies suggest that Th17 cells play a major role in the pathogenesis of IBD and experimental models of colitis, via secretion of various cytokines, particularly the coexpression of IL-17A and IFN-γ (23, 55–57). However, IgA^−/− mice, which lack the Treg-IgA homeostatic pathway (58) and have an expanded Th17 population in their intestine, do not develop spontaneous intestinal inflammation under SPF conditions throughout life. Our data showed that a subset of endogenous Th17 cells induced by SFB colonization coexpress IL-17A and IL-10. IL-10 is an anti-inflammatory cytokine that restrains the activities of effector T cells. Although CT immunization did not show induction of IL-17A/IL-10–coproducing Th17 cells on the cytokine level, RNA-Seq data indicated an immune-regulatory transcriptome of CT-induced Th17 cells, as well as endogenous Th17 cells, explaining why CT does not cause intestinal inflammation despite strongly expanding Th17 cells. For example, in keeping with a recent report (59), we saw upregulated Foxp3 gene expression in endogenous intestinal Th17 cells, as well as in CT-induced intestinal Th17 cells. Ctla4 and Ikzf2 (encoding Helios), which are normally highly expressed in Treg cells, and the products of which contribute to the suppressive function of Treg cells (60, 61), were more highly expressed in homeostatic intestinal Th17 cells as well. Also, these homeostatic intestinal Th17 cells have an upregulated immune gene network that promotes IL-10 production, including Icos, Maf, Prdm1, and Ahr (62–66). Single-cell RNA-Seq is needed in future studies to fully reveal the nature of Foxp3-expressing Th17 cells, although we saw very little coexpression of Foxp3 and IL-17A on the protein level (Supplemental Fig. 4B). Interestingly, pathogenic Th17 cells had lower expression of both Il17a and Il17f, but higher expression of Ifng compared with endogenous and CT-induced Th17 cells, consistent with what has been shown by Harbour et al. (67) that Th17 convert into Th1 cells and contribute to the pathogenesis of colitis. Most recently, Gaublomme et al. (47) and Wang et al. (68) published their studies of Th17 cell pathogenicity using the single-cell RNA-Seq technique, in which they found an intestinal Th17 cell signature similar to ours, although we did not notice any differential expression of Cd5l in our RNA-Seq data. The ratio of homeostatic Th17 cells to pathogenic Th17 cells might be critical in overt inflammation; it is an important topic of future research to understand whether these are separate lineages, and if not, what triggers the conversion from homeostatic Th17 cells to pathogenic Th17 cells and/or Th1 cells, and what determines the conversion from Th17 to Treg cells. This will provide us with new insights into the pathogenesis, prevention, and treatment of diseases of dysregulated immunity and inflammation.

We thank Dr. I.N. Mbawuike and Dr. C.L. Maynard for generously providing the IgA^−/− and 10BiT.Foxp3^gfp mice, respectively; M.L. Spell for cell sorting; and Dr. D. Botta for helping with the multiplex assay.

The authors have no financial conflicts of interest.