NKT Cell–Deficient Mice Harbor an Altered Microbiota That Fuels Intestinal Inflammation during Chemically Induced Colitis

Inflammatory bowel disease (IBD) is a spectrum of chronic disorders of the gastrointestinal tract characterized by recurring inflammatory bouts (1). Crohn disease (CD) and ulcerative colitis (UC) are the two major forms of IBD and broadly differ in the spread and depth of inflammatory lesions, as well as their immune profile, broadly Th1 and Th2, respectively. IBD has genetic and environmental components. More than 160 IBD susceptibility loci have been identified and comprise genes involved in pathways of innate and adaptive immunity, autophagy, and chemotaxis, such as NOD2, ATG16L1, and IL23R (2). However, the functional relevance of most of these genetic variants remains to be established. It is generally admitted that deregulated T cell responses toward the intestinal microbiota are at the core of IBD (3). Reciprocally, perturbations of host–microbiota homeostasis induced by the host genetics and/or environmental factors (e.g., diet, antibiotics) can alter the makeup of the microbiota and fuel inflammation at mucosal surfaces (1, 3).

NKT cells are unconventional T cells that respond to lipid and glycolipid Ags presented by the MHC class Ib molecule CD1d, and they generally express some activating and/or inhibitory receptors found on bona fide NK cells. NKT cells can be divided into type 1 or invariant NKT (iNKT) cells (4, 5) as well as type 2 or diverse NKT (dNKT) cells (6), based on TCR diversity and lipid specificity. iNKT cells respond to the prototypical glycosphingolipid α-galactosylceramide (αGC) or related compounds and can be accurately identified using αGC-loaded CD1d tetramers (7, 8). iNKT cells are innate lymphocytes capable of producing copious amounts of a large array of cytokines and chemokines only minutes following stimulation, and they also exert cytotoxic functions (5). Based on their functional properties, iNKT cells modulate immune responses and have been reported to play protective or deleterious functions during infection, autoimmunity, cancer, and inflammation, and they therefore constitute an appealing target for immunotherapy (4, 5). The biology of dNKT cells remains more elusive due to the paucity of reagents to track these cells. Some of these cells respond to self and/or microbial sulfated β-galactosylceramides (9–11) or phospholipids (12, 13). Whether dNKT cells are innate or more reminiscent of conventional T cells remains an open question.

Only a few studies have explored the role of NKT cells during intestinal inflammation, and most have focused on iNKT cells. It appears that these cells protect from Th1-mediated inflammation (14–17) (CD) but fuel Th2 type responses such as those typically seen in UC (18) and asthma (19), which is consistent with their overall yin-yang functions in disease. Using the T cell transfer colitis model, Hornung et al. (14) showed that CD4⁺DX5⁺ T cells (a phenotype reminiscent of NKT cells) could protect from colitis. In the dextran sodium sulfate (DSS) model, one study found that NKT cell–deficient (CD1d knockout [KO]) mice developed a similar colitis to that in control mice (15). However, another study reported that CD1d KO mice were more sensitive to colitis (17). The basis for this discrepancy is unclear, but it may relate to the controls used. What is more established is that pharmacological activation of iNKT cells through injection of αGC or the Th2-biasing lipid OCH9 confers consistent protection in the DSS model (15, 16), revealing iNKT cells as a potential target for immunotherapies to treat IBD. Moreover, iNKT cell–derived IL-13 mediates the development of intestinal inflammation during oxazolone-induced colitis, a Th2-driven model of intestinal pathology resembling human UC (18). Of note, a recent study demonstrated that deregulation of dNKT cells led to the development of spontaneous colitis in mice (20).

Vertebrates are colonized by complex communities of mutualistic organisms that reside on mucosal surfaces (lungs, intestines, and genitals) and on the skin and outnumber host cells and genes by factors of 10 and 100, respectively. Although the term microbiota classically refers to bacterial species, it also includes fungi, viruses, bacteriophages, and other more elusive microorganisms (21, 22). Perturbations of the microbiota composition, a phenomenon called dysbiosis, have been associated with several human diseases, including metabolic diseases (type I diabetes mellitus and obesity) (23) and IBD (3). Whether dysbiosis is a cause or a consequence of disease remains unclear. However, there is mounting evidence that the transfer of altered microbiota into healthy mice through cohousing, cross-fostering, or fecal transplant can transfer disease in some cases (24–26). Importantly, fecal transplantation from healthy donors has recently been shown to induce remission in UC patients in a clinical trial (27). One mechanism by which the microbiota can influence disease outcome is through the shaping of the immune system, and individual bacteria or groups of bacteria have been shown to modulate the development and/or function of immune cells, including Th17 cells (28), regulatory T cells (29), neutrophils (30), and more recently iNKT cells (31–34).

In this study, we describe that CD1d-deficient mice, which lack iNKT and dNKT cells, harbor an altered intestinal microbiota characterized by the increased abundance of several proinflammatory bacterial groups. This altered microbiota is associated with a colitogenic phenotype at steady-state, which is exacerbated upon DSS treatment. We also provide evidence that the intestinal microbiota modulates iNKT cell function upon activation during intestinal inflammation.

C57BL/6 (B6) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). CD1d^−/− mice were generated and provided by Dr. Chyung-Ru Wang (Northwestern University Feinberg School of Medicine) (35). All animal procedures were approved by the Faculty of Medicine and Pharmacy Animal Care Committee at the University of Toronto (animal use protocols 20009889 and 20010032 to D.J.P., and protocols 20009158, 20009741, and 20010379 to T.M.). Mice were housed in specific pathogen-free conditions at the Division of Comparative Medicine in the Centre for Cellular and Biomolecular Research (University of Toronto). Unless specified otherwise, 12- to 13-wk-old mice were used in all experiments. Littermate and nonlittermate mice were used, as specified.

Mice were administered 2% DSS (MP Biomedical, Santa Ana, CA) in drinking water (w/v) for 7 d followed by regular water, unless specified otherwise. Mice were weighed daily. Histological scoring was performed on colons fixed in 4% formalin, embedded in paraffin, and stained with H&E. Colon histology was assessed for infiltration by immune cells and tissue damage, by a trained pathologist, as described by Smith et al. (36)

Mice were treated with 20 mg of streptomycin following a 3-h fast. Twenty-four hours later, mice were fasted for 3 h and then infected with 5 × 10⁷ CFU S. enterica serovar Typhimurium SL1344. Bacterial load was measured in spleens or livers homogenized in 1% Triton X-100 and plated on 50 μg/ml streptomycin/MacConkey agar plates. Cecum histology was assessed on cecum fixed in 4% formalin, embedded in paraffin, and stained with H&E. H&E-stained, formalin-fixed cecum tissues were assessed for edema, epithelial erosion, neutrophil recruitment, and goblet cell depletion, as described by Geddes et al. (37, 38).

The abundance of bacterial groups was analyzed as described by Robertson et al. (39). Briefly, total DNA was extracted from the colon fecal matter using the MO BIO Laboratories PowerSoil kit following the manufacturer’s specifications. Bacterial DNA was analyzed by quantitative RT-PCR using 16S rDNA primers (Integrated DNA Technologies) for specific bacterial groups (39). The relative abundance of each bacterial group was normalized to Eubacteria using the 2^−ΔΔCt method.

We used a method adapted from Markle et al. (25). Timed pregnancies were set using 7- to 8-wk-old B6 mice. Pregnant females were transferred into clean cages and gavaged with cecal content from adult female B6 or CD1d KO mice. Briefly, the cecum of donor mice was dissected, opened along its length, and the contents were transferred to a sterile tube and diluted 1:50 (v/v) in sterile water. Two hundred microliters of this suspension was given to each recipient by oral gavage using 22- to 24-gauge rounded edge needles. Recipients were rested for 24 h, and this procedure was repeated once. We think that the longitudinal transfer upon natural delivery allows for a greater stability of the altered microbiota in the newborn mice. These mice were used at 7–8 wk of age.

Mice were administered with 2% DSS in drinking water followed by regular water for 24 h. Proximal colon biopsy punches (2 mm) were cultures in complete RPMI 1640 medium supplemented with 10% FCS for 48 h with or without recombinant mouse IL-23 (10 ng/ml). IL-22 was analyzed using a Ready-Set-Go! ELISA set purchased from eBioscience/Affymetrix.

16S rRNA sequencing was performed by the Centre for the Analysis of Genome Evolution and Function at the University of Toronto. The V4 region was amplified using primers V4-515 forward (5′-AATGATACGGCGACCACCGAGATCTACACTATGGTAATTGTGTGCCAGCMGCCGCGGTAA-3′) and V4-806 reverse (5′-CAAGCAGAAGACGGCATACGAGATXXXXXXXXXXXXAGTCAGTCAGCCGGACTACHVGGGTWTCTAAT-3′) (40). PCR reactions were generated in triplicate for each sample using KAPA2G Robust HotStart ReadyMix (Kapa Biosystems, Wilmington, MA) according to the manufacturer’s instructions. A template-minus control was also included for each sample. Replicate PCRs for each sample were pooled. Equal quantities of each pool were determined using the Quant-iT PicoGreen dsDNA assay kit (Thermo Fisher Scientific, Waltham, MA) and combined. The final pool was purified using Agencourt AMPure XP beads (Beckman Coulter, Mississaugua, ON, Canada) before paired-end sequencing. 16S rRNA sequencing was performed on an Illumina MiSeq2 using V2 chemistry.

The PANDAseq (41) assembler was used for the quality trimming and assembly of the paired-end reads, with a minimum overlap of 25 nucleotides between the forward and reverse reads. The USEARCH (42) reference method was used, along with the Greengenes database version 13.8 (43), for chimera detection and removal. The removed chimeric sequences were clustered at 97% identity using USEARCH in Quantitative Insights Into Microbial Ecology (44) version 1.8.0 against the Greengenes database version 13.8 (43). A representative sequence from each operational taxonomic unit (OTU) was selected and aligned by PyNAST (45) to the Greengenes database version 13.8 (43). The Ribosomal Database Project Bayesian algorithm, executed within Quantitative Insights Into Microbial Ecology, produced taxonomy assignment for each OTU. The aligned sequence data were used to generate a phylogentic tree. The above phylogenetic tree was constructed with the FastTree (46) algorithm. An OTU table was created using taxonomy assignment and aligned representative sequences. Sequences that failed to align to the reference database were removed from the OTU table. Low-abundance OTUs with a fraction count <0.005% were removed and the retained OTUs were used for downstream analysis (47). Then, α diversity plots were generated and data were rarefied to a depth of 40,000 based on the plots. Weighted UniFrac distances were calculated through the R phyloseq library and used for principal coordinates analysis.

Student t tests were used when two groups were compared. For four-group comparisons, we used one-way ANOVA tests with a Tukey multiple comparison tests to compare individual pairs. Statistical analyses were performed using GraphPad Prism. A p value <0.05 was considered significant.

iNKT cells are present in many lymphoid (e.g., bone marrow, thymus, spleen, lymph nodes) as well as nonlymphoid (e.g., liver, lungs, skin) tissues. Consistent with previous studies (32, 33, 48), we detected iNKT cells in the small intestine, cecum, and colon lamina propria of B6 naive mice (Supplemental Fig. 1). However, we could not detect iNKT cells within the intraepithelial lymphocyte (IEL) compartment (data not shown). Nevertheless, we were able to detect iNKT cells within the IEL population in Vα14 transgenic mice (not shown), which have increased numbers of iNKT cells, raising the possibility that iNKT cells may be present at very low numbers within the small intestine IELs in B6 mice.

The role iNKT cells play during DSS-mediated colitis has been previously studied but yielded inconsistent results (15, 17). In an attempt to clarify this controversy, we subjected mice deficient for the two CD1d genes (hereafter named CD1d KO) to a 2% DSS regimen. After 8 d, CD1d KO mice showed macroscopic signs of sickness, such as lethargy, reluctance to move, and hunched posture, compared with B6 mice (Supplemental Video 1). Additionally, CD1d KO mice had greater and faster weight loss (Fig. 1A) and shorter colons (Fig. 1B) compared with B6 mice. Finally, histological examination of H&E-stained colon sections revealed more severe inflammation in CD1d KO mice (Fig. 1C, 1D). Together with previous work using both CD1d KO and Jα18 KO mice (17), our data suggested that CD1d and CD1d-restricted T cells, including iNKT cells, exerted protective functions during chemically driven intestinal inflammation.

The controversial role of iNKT cells in colitis could reflect different environmental conditions (e.g., housing conditions, diet). Therefore, we hypothesized that these discrepancies could be attributed to the use of nonlittermate control mice. Indeed, although the use of littermate mice with different genotypes is the best way to control for environmental differences, nonlittermate control mice are often used for practical as well as budgetary reasons. To test this hypothesis, we crossed CD1d KO mice with B6 mice to generate CD1d^+/− mice that were subsequently crossed to generate CD1d^+/+, CD1d^+/−, and CD1d^−/− littermate mice. These animals were then subjected to 2% DSS to trigger intestinal inflammation, as described above. Surprisingly, mice displayed identical weight loss regardless of genotype (Fig. 2A). Additionally, histological examination of H&E-stained colon sections revealed identical inflammation in all mice studied (Fig. 2B). Interestingly, all mice showed increased sensitivity compared with our original B6 mice (compare Figs. 1C, 1D, 2A, 2B). This suggested that environmental factors, and not CD1d and/or iNKT cell deficiency, were dominantly influencing intestinal inflammation in this model.

To substantiate this finding, we studied B6 and CD1d KO nonlittermate mice as well as CD1d^+/− and CD1d^−/− littermate mice during intestinal inflammation driven by infection with S. enterica serovar Typhimurium. Pretreatment of mice with streptomycin allows Salmonella to trigger an acute colitis, and this model has been widely used to tease out bacterial and host factors that regulate inflammation. Using this model, we found that CD1d KO mice developed a marginally yet significantly milder cecal inflammation mainly characterized by reduced goblet cell depletion (Fig. 2C, 2D). However, CD1d^+/− and CD1d^−/− littermate mice showed no difference in intestinal inflammation that was also comparable to that of the original CD1d KO mice (Fig. 2C, 2D). Additionally, bacterial translocation to the spleen, measured by CFUs, was similar between the four groups of mice (data not shown), suggesting that neither CD1d deficiency nor environmental factors in our experimental settings impact Salmonella infection and Salmonella-driven inflammation. Taken together, these observations suggested that a dominant nongenetic factor present in CD1d KO mice influenced colitis susceptibility in a context-dependent fashion.

The composition of the indigenous intestinal microorganisms is a major factor that can vary between wild-type and gene-targeted mice and influence disease outcome. Variations of microbiota composition can be due to a direct impact of genetic landscape variation or to stochastic longitudinal transmission that can result in ecological drifts across generations in adjacently caged animals (49, 50). Additionally, differences in microbiota composition have been associated with many disease models, including colitis (3). To analyze and compare the microbiota between our littermate and nonlittermate mice, we quantified the relative bacterial abundance of several bacterial groups in the colon of naive mice by quantitative real-time PCR using primers targeting 16S rDNA (39). Compared to B6 nonlittermate mice, CD1d KO mice had higher abundance of Bacteroides, Bifidobacterium, and Lactobacillus but similar levels of mouse intestinal Bacteroides (Fig. 3). We also analyzed the mice for the presence of segmented filamentous bacteria (SFB), which have been associated with increased Th17 (28) responses. Whereas SFB were virtually undetectable in B6 mice, they were consistently found in CD1d KO mice (Fig. 3). CD1d^+/+, CD1d^+/−, and CD1d^−/− littermate mice were found to be similar to CD1d KO mice for the Bacteroides, Lactobacillus, and SFB groups, but not the Bifidobacterium group (Fig. 3). Taken together, these findings revealed that adjacently caged B6 and CD1d KO nonlittermate mice had profound microbiota differences, and they suggested that the CD1d KO colitogenic microbiota was both transmissible and dominant.

To directly assess whether the differences in microbiota composition in CD1d KO mice were responsible for the increased colitis sensitivity, we performed fecal transplant experiments. B6 pregnant dams were gavaged with diluted cecal contents from either B6 or CD1d KO female mice. Markle et al. (25) recently demonstrated that the gavage of weanlings resulted in microbiota alterations that were durable until at least 11 wk of age. We reasoned that natural maternal transmission would result in durable engraftment of the altered microbiota in newborn mice. Adult offspring from these gavaged dams were subjected to 2% DSS in the drinking water ad libitum for 8 d, at which point they were switched to regular water. In the absence of DSS, B6 mice grafted with either B6 or CD1d KO cecal contents (B6^B6 and B6^CD1d mice, respectively) had similar weight gain (Fig. 4A). However, although they also showed similar initial weight loss upon DSS treatment, B6^CD1d mice appeared to have delayed recovery upon DSS removal, compared with B6^B6 mice (Fig. 4A). Additionally, B6^CD1d mice had elevated colon weight-to-length ratios (Fig. 4B) and worse pathological scores (Fig. 4C) upon histological examination compared with B6^B6 mice. This indicates that colonization of mice by the CD1 KO microbiota resulted in increased intestinal inflammation. B6^CD1d mice also showed a trend toward elevated colon weight-to-length ratios and pathological scores in the absence of DSS exposure, although the latter did not reach statistical significance. Finally, we found that colons from DSS-treated B6^CD1d mice failed to produce IL-22 (Fig. 4D), a cytokine that is involved in the resolution of intestinal inflammation and wound repair (51, 52). However, this defect could be corrected by the addition of IL-23, which is known to induce IL-22 production. Taken together, these results suggested that the microbiota found in CD1d KO mice was proinflammatory and affected the resolution of inflammation.

Next, we compared microbial communities between B6 and CD1d KO mice pre- and post-transplant by sequencing bacterial 16S rRNA libraries prepared from proximal colon contents from B6 and CD1d KO nonlittermate mice as well as transplanted B6^B6 and B6^CD1d mice. Principal coordinates analysis revealed differences in gut microbiota composition. Although B6 and transplanted B6^B6 mice were essentially indistinguishable from each other, our analysis revealed that CD1d KO mice and transplanted B6^CD1d mice were different from their B6 and B6^B6 counterparts, as well as from each other (Fig. 5A). This suggested that transfer of the CD1d KO microbiota into B6 mice only partly recapitulated the original microbiota, and created a “third state” of microbiota, as previously reported (25). The results showed that the fecal transplant procedure in itself induced a shift in microbiota composition characterized by an overall decrease in Bacteroidetes and increase in Firmicutes (Fig. 5B), although this shift did not parallel increased DSS sensitivity. Next, we reasoned that proinflammatory bacterial groups may be similarly increased in CD1d KO and B6^CD1d mice and/or that anti-inflammatory bacterial groups may be decreased in these mice, compared with their respective controls. At the phylum level, we found that members of the Proteobacteria, Deferribacteres, and the candidate group TM7 were elevated in CD1d KO and B6^CD1d mice compared with their respective controls (Fig. 5B, 5C). Among Proteobacteria, we found that only Betaproteobacteria and Deltaproteobacteria were elevated in CD1d KO and B6^CD1d mice, whereas Alphaproteobacteria and Gammaproteobacteria remained unaffected (Fig. 5D and not shown). Of note, Epsilonproteobacteria, including the genus Helicobacter, were more abundant in CD1d KO mice compared with B6 mice, but these bacteria were not transmissible via fecal transplant (not shown). Deferribacteres contains the mucin-degrading Mucispirillum species, which were significantly elevated in CD1d KO and B6^CD1d mice (Fig. 5E). Interestingly, the other known group of mucin-eating bacteria Akkermansia were found abundantly only in B6^B6 mice (Fig. 5F). Finally, although overall levels of Bacteroidetes were not affected between B6 and CD1d KO mice, members of the genus Prevotella were significantly more abundant in CD1d KO mice and were transmissible to B6 mice upon fecal transplant (Fig. 5G). Collectively, our analysis revealed profound differences in the microbiota found in CD1d KO mice that could be partially transmitted into B6 recipient mice upon fecal transplant.

It was recently shown that the intestinal microbiota affects the phenotype and function of iNKT cells (31–34), including during Th2-driven airway and intestinal inflammation (33). Additionally, pharmacological activation of iNKT cells through the injection of αGC or OCH9 was shown to decrease the severity of DSS-mediated colitis (15, 16). We next assessed whether the protective outcome from lipid-mediated iNKT cell activation would differ based on microbiota differences. For this, B6^B6 and B6^CD1d mice were placed on 2% DSS for 8 d and at the onset of symptoms (day 7) received a single i.p. injection of either OCH9 or vehicle. As described above, B6^CD1d mice treated with vehicle experienced greater weight loss and delayed recovery compared with B6^B6 mice (Fig. 6A), confirming the proinflammatory nature of the CD1d KO microbiota. Contrary to previously published data, we found that OCH9 treatment slightly increased weight loss in B6^B6 mice (Fig. 6B). Finally, we found that OCH9 increased weight loss and impaired recovery in B6^CD1d mice compared with both vehicle-treated B6^CD1d mice (Fig. 6C) and OCH9-treated B6^B6 mice (Fig. 6D). Of note, we found no difference in colon weight-to-length ratios and pathological scores in these settings (Fig. 6E, 6F). Taken together, these results suggested that the intestinal microbiota influenced iNKT cell function upon pharmacological activation during DSS-mediated colitis.

Our study provides a possible explanation for the published discrepancies regarding the function that CD1d-restricted T cells, including iNKT cells, play during DSS-mediated colitis (15, 17), and it emphasizes the need to use littermate control mice in these experimental settings. We found that the microbiota has a major influence on experimental colitis outcome during DSS treatment, and to a lesser extent following S. typhimurium infection. Alternatively, the mere absence of CD1d-restricted T cells appeared to have little effect on inflammation in these models. Finally, our data showed that CD1d-deficient mice harbor a proinflammatory and transmissible microbiota that also modulated iNKT cell function upon pharmacological activation during DSS-mediated colitis.

Perturbations in the gastrointestinal microbiota have been reported in CD and UC patients, generally characterized by a reduction in ecological diversity, expansion of Gammaproteobacteria, and a contraction of Bacteroidetes and certain Firmicutes (3). Additionally, studies comparing conventional and gnotobiotic genetically susceptible mouse strains, as well as fecal transplant and colonization experiments, demonstrated that certain commensal bacterial species can modulate the immune system as well as intestinal inflammation (24, 28, 29, 53–55). In agreement with this, we found that transfer of the CD1d KO microbiota triggered basal inflammation at steady-state, which was greatly exacerbated upon DSS exposure, demonstrating that CD1d KO mice are colonized with commensal bacteria with proinflammatory potential. Our analyses revealed that CD1d KO mice harbor several bacterial groups that have been described as proinflammatory. First, Elinav et al. (24) reported that NLRP6 inflammasome deficiency leads to increased basal colonic inflammation that is exacerbated upon DSS treatment, a phenotype very similar to that of CD1d KO mice. These authors ascribed this elevated risk for colitis to an increased abundance of bacteria from the phylum TM7 and the family Prevotellaceae (24), which were similarly increased in CD1d KO mice. Second, we found that SFB were present in CD1d KO mice, in agreement with a previous study (56), and transmissible to their progeny. SFB were previously reported to increase inflammation (57, 58), perhaps by increasing the development of Th17 cells (28). Third, we found that the mucin degraders of the genus Mucispirillum but not Akkermansia were significantly increased in the CD1d KO microbiota. Although both genera are increased in DSS-treated mice (59, 60), Akkermansia has been shown to protect from the development of colitis (61), and its absence in CD1d KO mice could explain their increased sensitivity. Finally, Betaproteobacteria and Deltaproteobacteria, which are involved in nitrogen fixation and sulfate/sulfur reduction, respectively, were more abundant in the CD1d KO microbiota, although their contribution to inflammation remains elusive (62).

Another interesting question arising from this work is whether the proinflammatory microbiota found in CD1d KO mice results from serendipitous familial transmission (49) or is fostered by the lack of CD1d and/or iNKT cells in these mice. Our analyses show that second generation littermate mice have an identical microbiota regardless of their genotype, which is in stark contrast with another study (56). The basis for these discrepancies is unclear. To the best of our knowledge, Blumberg and colleagues (56) are using CD1d KO mice generated by Sonoda et al. (63), and we are using CD1d KO mice generated by Chen et al. (35). Although both strains lack the two copies of the CD1d gene, it is possible that genetic differences between the mice used play a role. Additionally, housing and husbandry conditions that differ between animal facilities may explain the difference between the two studies. Nonetheless, it is possible that CD1d and/or NKT cell deficiency fosters an ecological drift over time that leads to the accumulation of proinflammatory microorganisms.

iNKT cells are activated by S. typhimurium both in vitro and in vivo (64–67). Oral infection of mice with S. typhimurium was previously shown to activate iNKT cells and induce IFN-γ production in peripheral lymphoid tissues and the cecum lamina propria (67, 68). In this study, we found that CD1d-restricted T cells, which include iNKT cells, do not seem to play a role in S. typhimurium–driven acute inflammation following infection of streptomycin-pretreated mice, which is in agreement with a previously published study focusing on oral infection in the absence of antibiotic treatment (68).

CD1d-sufficient or -deficient mice displayed no differential susceptibility to DSS treatment, which is in agreement with a study from the Blumberg and colleagues (15) but in stark contrast with a more recent study (17). Our work clearly shows that the microbiota is a confounding factor that can lead to misleading conclusions. Despite funding and time constraints in a challenging and aggressive research environment, using littermate control mice (69) and/or standardization of the microbiota (70, 71) is of paramount importance, especially during infection and inflammation studies. Although cohousing of weanlings or adult experimental mice can be an attractive cost-effective alternative, it is well known that the adult microbiota is more stable and resilient to perturbations (72). Additionally, the timing and method of colonization can affect colonization efficiency and alter immune parameters as well (39, 73).

The prevalence of iNKT cells is highly variable in humans and between different strains of mice. Additionally, the existence of functionally discrete iNKT cell populations (i.e., Th1, Th2, Th17, Th10) have recently been revealed in mice (74, 75), with strain-dependent relative prevalence (75). It is not known whether similar functional iNKT cell subsets are present in humans. It is likely that differences in iNKT cell prevalence and functional differentiation are the consequence of both genetic and environmental factors, among which the microbiota could be a contributing factor. In stark contrast with previous studies, we found that iNKT cell activation via OCH9 treatment exacerbated colitis, especially in mice colonized with the CD1d KO microbiota. This suggests that the gut microbiota affects iNKT cell function during intestinal inflammation and is in direct agreement with recent studies (31–34). A better characterization of the impact of the CD1d KO microbiota on iNKT cell homeostasis and function will be the focus of further studies.

In conclusion, we have identified an altered microbiota within NKT cell–deficient CD1d KO mice that exacerbates intestinal inflammation at steady-state and upon DSS treatment. We also found that this proinflammatory microbiota influenced iNKT cell function in this colitis model. Hence, microbiota composition must be carefully controlled to draw meaningful conclusions with regard to iNKT cell function in the context of intestinal inflammation.

We thank Laura Kent (Division of Comparative Medicine, University of Toronto) for maintenance of our animal colony and for monitoring the mice during colitis experiments.

The authors have no financial conflicts of interest.