Aberrant intestinal microbiota due to IL-1 receptor antagonist deficiency promotes IL-17- and TLR4-dependent arthritis

Background Perturbation of commensal intestinal microbiota has been associated with several autoimmune diseases. Mice deficient in interleukin-1 receptor antagonist (Il1rn −/− mice) spontaneously develop autoimmune arthritis and are susceptible to other autoimmune diseases such as psoriasis, diabetes, and encephalomyelitis; however, the mechanisms of increased susceptibility to these autoimmune phenotypes are poorly understood. We investigated the role of interleukin-1 receptor antagonist (IL-1Ra) in regulation of commensal intestinal microbiota, and assessed the involvement of microbiota subsets and innate and adaptive mucosal immune responses that underlie the development of spontaneous arthritis in Il1rn −/− mice. Results Using high-throughput 16S rRNA gene sequencing, we show that IL-1Ra critically maintains the diversity and regulates the composition of intestinal microbiota in mice. IL-1Ra deficiency reduced the intestinal microbial diversity and richness, and caused specific taxonomic alterations characterized by overrepresented Helicobacter and underrepresented Ruminococcus and Prevotella. Notably, the aberrant intestinal microbiota in IL1rn −/− mice specifically potentiated IL-17 production by intestinal lamina propria (LP) lymphocytes and skewed the LP T cell balance in favor of T helper 17 (Th17) cells, an effect transferable to WT mice by fecal microbiota. Importantly, LP Th17 cell expansion and the development of spontaneous autoimmune arthritis in IL1rn −/− mice were attenuated under germ-free condition. Selective antibiotic treatment revealed that tobramycin-induced alterations of commensal intestinal microbiota, i.e., reduced Helicobacter, Flexispira, Clostridium, and Dehalobacterium, suppressed arthritis in IL1rn −/− mice. The arthritis phenotype in IL1rn −/− mice was previously shown to depend on Toll-like receptor 4 (TLR4). Using the ablation of both IL-1Ra and TLR4, we here show that the aberrations in the IL1rn −/− microbiota are partly TLR4-dependent. We further identify a role for TLR4 activation in the intestinal lamina propria production of IL-17 and cytokines involved in Th17 differentiation preceding the onset of arthritis. Conclusions These findings identify a critical role for IL1Ra in maintaining the natural diversity and composition of intestinal microbiota, and suggest a role for TLR4 in mucosal Th17 cell induction associated with the development of autoimmune disease in mice. Electronic supplementary material The online version of this article (doi:10.1186/s40168-017-0278-2) contains supplementary material, which is available to authorized users.

We questioned the role of IL-1Ra in regulation of the intestinal microbiota and the involvement of mucosal immune response as an underlying mechanism for the spontaneous autoimmune arthritis in IL1rn −/− mice, which is dependent on T cells and IL-17 [4,8]. Several studies have associated commensal microbiota with autoimmune disease in mouse models of rheumatoid arthritis (RA), diabetes, and multiple sclerosis [9][10][11][12][13]. Importantly, the diversity and the composition of commensal intestinal microbiota are altered in patients with psoriatic and RA compared with healthy individuals [14][15][16][17][18]. One of the most prominent effects of microbiota is to define the balance between the proinflammatory CD4 + T helper 1 (Th1) and Th17 cells and protective regulatory T (Treg) cells, both at mucosal surfaces and systemically [19][20][21]. In this context, specific subsets of intestinal microbiota, such as the vancomycin-sensitive segmented filamentous bacteria (SFB), robustly induce differentiation of Th17 cells in small intestine LP (SI-LP) [22,23]. Th17 cells are considered to play a pathogenic role in a subset of patients with RA by producing proinflammatory mediators, such as IL-17, and inducing osteoclastogenesis [24][25][26][27][28]. Interestingly, SFB colonization has been shown to exacerbate arthritis in K/BxN mice, an autoimmune model of arthritis arising from T cell auto-reactivity to the glycolytic enzyme glucose-6-phosphate isomerase [13,29]. However, given that SFB were not found in human adults [30,31], it is important to investigate the involvement of other indigenous microbiota in arthritis.
We previously described that arthritis in IL1rn −/− mice is diminished under germ-free (GF) condition [12]. We also showed that IL1rn −/− arthritis is dependent on the activation of Toll-like receptor 4 (TLR4), which affected systemic Th17 cell differentiation [12]. Here, we characterized the intestinal microbiota present in autoimmune-prone IL1rn −/− mice to clarify the nature of the microbiota that trigger arthritis and the underlying mucosal immune pathways. We also examined the role of TLR4 in the intestinal mucosal immune responses associated with arthritis.
Using high-throughput 16S ribosomal RNA (rRNA) gene sequencing of fecal microbiota, we demonstrate a critical role for IL-1Ra in maintaining the natural diversity and composition of commensal intestinal microbiota. We show that the aberrant IL1rn −/− microbiota increases intestinal Th17 cell differentiation, a phenotype that is transferable to wild-type (WT) mice by the microbiota. We also provide evidence that tobramycinsensitive indigenous commensal intestinal bacteria contribute to arthritis in IL1rn −/− mice and identify a significant role for TLR4 in mucosal induction of IL-1β and IL-17 prior to the onset of arthritis.

IL-1Ra maintains the biodiversity and richness of commensal intestinal microbiota
To identify intestinal microbiota associated with arthritis, we sequenced fecal bacterial 16S rRNA genes of IL1rn −/− and age-and gender-matched WT control mice. Fecal microbiota were analyzed as an unselected representation of the overall microbial communities in the intestines. Considering differential roles of TLR2 and TLR4 in IL1rn −/− arthritis [12], we sequenced samples of The average sequencing depth and total numbers of reads and operational taxonomic units (OTU) per experimental group as well as the hierarchical weighed UniFrac cluster analysis at the genus level are shown in Additional file 1: Table S1 and Additional file 1: Figure S1.
Principal coordinates analysis (PCoA) based on an unweighted UniFrac analysis of intestinal microbiota showed that IL1rn −/− microbiota is profoundly different from the WT microbiota (Fig. 1a). WT and Il1rn −/− mice formed clear, separate clusters regardless of the cage or litter of origin (Fig. 1a). Strikingly, microbial composition of IL1rn −/− and IL1rn −/− Tlr2 −/− mice were indistinguishable, while IL1rn −/− Tlr4 −/− mice formed another distinct cluster (Fig. 1a). To assess the effects of familial transmission and lineage origin versus the effect of the genotype (WT or IL1rn −/− ), we compared the UniFrac distances within a litter with UniFrac distances across litters of the same genotype as well as the opposite genotype, similar to the study by Ubeda et al. This analysis showed that the effect of the lineage origin and litter was limited in our experimental setting, because, as long as the genotype remained the same, the UniFrac distances across different litters were very similar to the UniFrac distances within the litters (Additional file 1: Figure S2A). This was true for both WT and Il1rn −/− groups. Importantly, the UniFrac distances were significantly higher when mice from different genotypes were compared, indicating a higher level of dissimilarity (Additional file 1: Figure S2B). Therefore, the effect of the genotype (Il1rndeficiency) on the overall microbiota composition was significantly higher than any litter and cage effect. In addition, IL1rn −/− and IL1rn −/− Tlr2 −/− mice showed significantly reduced number of OTUs and loss of microbial diversity based on the Shannon index, the rarefaction curves of phylogenetic distance (PD) whole tree, and the diversity index bootstrapped for the number of retrieved sequences (Fig. 1b-e). IL-1Ra deficiency also resulted in loss of species richness estimated by Chao index (Fig. 1f). These effects were fully or partially restored in IL1rn −/− Tlr4 −/− mice ( Fig. 1b-f). Altogether, these data strongly suggest that IL-1Ra plays a critical role in maintaining the intestinal microbial diversity, and that the loss of diversity in IL1rn −/− mice partially depends on TLR4.
Specific taxonomic alterations characterize the dysregulated microbiota of autoimmune-prone IL1rn −/− mice The phylogenetic tree in Fig. 2 summarizes the observed alterations in relative abundances of microbial taxa. Compared with WT microbiota, we found a highly significant overrepresentation of the genus Helicobacter (P = 0.004, Bonferroni corrected), and a significant underrepresentation of the genus Prevotella (P = 0.008, Bonferroni corrected) ( Fig. 2 and Additional file 1: Table S2). In addition, IL1rn −/− intestinal microbial composition was characterized by   Figure S1 and Additional file 1: Table S1 expansion of Butyricimonas, Rikenella, and Streptococcus by 10, 3.7, and 2.4 folds (P = 0.0048, P = 0.0022, and P = 0.0032, respectively, Bonferroni uncorrected), along with a decrease in Parasutterella, Xylanibacter, Ruminococcus, and Barnesiella by 10, 6.9, 2.7, and 1.4 folds (P = 0.040, P = 0.0004, P = 0.0099, and P = 0.0005, respectively, Bonferroni uncorrected), respectively ( Fig. 2 and Additional file 1: Table S2). Notably, we were unable to identify any OTUs in our dataset that could be classified as SFB (family Clostridiaceae, genus Candidatus Arthromitus). Moreover, none of the 27 present OTUs assigned to the family Clostridiaceae aligned with the known SFB 16S gene sequences in The Ribosomal Database Project [30]. However, SFB were detectable by qPCR in fecal samples of all WT mice and most of the Il1rn −/− mice (Additional file 1: Table S3). Although WT mice tended to have slightly more SFB, the level of SFB colonization was not significantly different between the groups (Additional file 1: Table S3).
Altogether, these data suggest that multiple yet specific microbial taxa are regulated by the physiologic expression of IL-1Ra. Therefore, a complex set of aberrant microbiota may affect the (mucosal) immune response and contribute to the autoimmune disease in IL1rn −/− mice.

IL1rn −/− intestinal microbiota potentiate IL-17 production by intestinal lamina propria lymphocytes
To assess the effect of IL-1Ra deficiency on the mucosal T cell response, we cultured enzymatically isolated lamina propria lymphocytes (LPL) ex vivo in the presence of PMA and ionomycin. The production of the Th1 signature cytokine IFNγ was low and not altered by the IL-1Ra deficiency (Fig. 3a, gating strategy shown in Additional file 1: Figure S3); however, we observed a marked increase in the production of IL-17 by IL1rn −/− LPLs compared with WT LPLs (Fig. 3b). Flow cytometry analysis of lamina propria cells of WT and IL1rn −/− mice verified a significant, clear increase of IL-17-producing TCRβ + CD4 + cells in IL1rn −/− mice, while TCRβ − cells showing specific changes in intestinal microbial community at different taxonomic levels induced by IL-1Ra deficiency. Nodes represent taxa, and the size of each node represents its relative abundance. The color red indicates a decrease and blue represents an increase of relative abundance in IL1rn −/− compared with WT mice. The thickness of the green border indicates the degree of the statistical significance by Mann-Whitney U test. See also Additional file 1: Table S2 in LP produced similar amounts of IL-17 in WT and IL1rn −/− mice (Additional file 1: Figure S4). This suggests that Th17 cells, not γδ T cells, are the source of increased IL-17 production in LP of IL1rn −/− mice. Production of IL-4, IL-6, and TNFα was not affected ( Fig. 3c and data not shown). Interestingly, the production of IL-17 but not IFNγ by lymphocytes in jointdraining lymph nodes (dLN) was significantly increased in IL1rn −/− mice compared with WT mice (Fig. 3d, e). This was paralleled by a concomitant decrease in the production of the Th2-related cytokine IL-4 in IL1rn −/− mice (Fig. 3f ).
To identify a potential causative relationship between the aberrant microbiota and enhanced mucosal IL-17 production, we transferred IL1rn −/− microbiota to WT mice by oral gavage followed by immediate co-housing of the two mouse strains for up to 6 weeks. Transfer of IL1rn −/− microbiota clearly potentiated IL-17 production by SI-LP T cells in WT recipients as early as 10 days post fecal transfer and co-housing, without affecting IFNγ and IL-4 ( Fig. 3g- , and Th2 cell cytokines (IFNγ, IL-17A, and IL-4, respectively) by SI-LP (a-c and g-i) and draining lymph node (dLN) lymphocytes (d-f). Cells were isolated from WT and IL1rn −/− mice (a-f) or WT mice transplanted with IL1rn −/− feces and co-housed with IL1rn −/− mice for 10 days (g-i). Cells were stimulated ex vivo with PMA and ionomycin in duplicates for 5 h, and cytokines were measured by Luminex assay. Data represent mean + SEM of a representative experiment with n = 5 (a-c) and n = 3 (d-i) mice per group, each stimulated in duplicate. n.s. not significant, *P ≤ 0.05 and ***P ≤ 0.001, by Mann-Whitney U test. See also Additional file 1: Figure S2 animals during the 6-week follow-up period. This suggests that additional (genetic) susceptibility of the host, as in IL1rn −/− mice, is required for the development of arthritis. Furthermore, co-housing with WT mice did not affect the development of arthritis in IL1rn −/− mice (not shown).
Potentiated Th17 response and spontaneous arthritis in IL1rn −/− mice highly depend on the presence of commensal microbiota To determine whether the increase in intestinal Th17 cells and spontaneous arthritis in IL1rn −/− mice depends on commensal microbiota, we established GF IL1rn −/− mice. Flow cytometry analysis of LPLs showed that germ-free condition had no significant effect on the percentage of Th1 cells while reducing the numbers of Th1 cells in SI-LP ( Fig. 4a, b). In contrast, both the percentage and the number of SI-LP Th17 cells were substantially reduced in GF compared with conventional (CV) mice ( Fig. 4c, d). This strongly suggests that the skewed intestinal T cell balance in IL1rn −/− mice is largely microbiota dependent.
In agreement with IL-17-dependence of IL1rn −/− arthritis [4,8] and in line with our previous observations [12], GF IL1rn −/− mice showed a clear sustained protection from arthritis with on average 3 weeks delay in disease onset (Fig. 4e). In addition, transfer of conventional IL1rn −/− microbiota to GF IL1rn −/− mice re-induced arthritis and resulted in a severe disease comparable to that in conventional IL1rn −/− mice (Additional file 1: Figure S5). Therefore, IL1rn −/− commensal microbiota, although not sufficient to induce arthritis in a WT host, are critical for the full development of arthritis in IL1rn −/− mice. Consistently, we also observed a robust reduction of IL-17, but not IFNγ, production in the spleen and most notably in the joint-adjacent lymph nodes of GF mice (Fig. 4f, g). These effects were accompanied by a significant reciprocal increase in Th2-related cytokines IL-4 and IL-10 as well as IL-2 in the spleens of GF mice ( Fig. 4h and Additional file 1: Figure S6). These data support modulation of extra-intestinal immune response by intestinal microbiota during arthritis.
Tobramycin-induced alteration of intestinal microbiota suppresses arthritis in IL1rn −/− mice The lack of microbiota in GF mice is not limited to the intestines. To determine whether intestinal microbiota serve as a relevant trigger for arthritis, we first depleted intestinal microbiota in conventionally housed mice using a cocktail of metronidazole, neomycin, and ampicillin. Treatment of 5-week-old IL1rn −/− mice for only 1 week suppressed arthritis over a sustained period, i.e., 6 weeks after ceasing antibiotics (Fig. 5a). This indicated that abrogation of arthritis in GF mice (Fig. 4c) is not due to an immature immune system and, more importantly, can be reproduced by the sole eradication of intestinal microbiota. Interestingly, colonization of the antibiotic-treated mice with SFB as model organisms inducing SI-LP Th17 cells was sufficient to fully restore arthritis (Fig. 5a).
To determine which subset of IL1rn −/− microbiota triggers arthritis, we first compared the effects of treatment with ampicillin, broadly targeting aerobic bacteria, and metronidazole, broadly targeting anaerobic bacteria. To our surprise, only metronidazole showed efficacy in reducing arthritis severity (Fig. 5b). This suggests involvement of anaerobic bacteria in the progression of arthritis. We next compared the effects of more selective antibiotics tobramycin and vancomycin, the latter of which has been reported to eradicate SFB and inhibit SFB-induced lamina propria Th17 cells and arthritis [13,22,32]. These experiments revealed that although SFB were able to exacerbate arthritis in IL1rn −/− mice (Fig. 5a), only tobramycin but not vancomycin significantly diminished arthritis (Fig. 5c). To understand the changes in the microbiota induced by tobramycin treatment, we compared 16S rRNA gene sequences of fecal microbiota at the end-point of tobramycin treatment with the baseline microbiota. Among taxa with >0.1% relative abundance, tobramycin treatment resulted in a near-complete elimination of the genera Helicobacter and Flexispira (both belonging to the family Helicobacteraceae). In addition, a strong and highly significant reduction in the genera Clostridium and Dehalobacterium was observed (Additional file 1: Figure S7 and Table S4). Other changes in the microbiota did not reach the statistical significance after Bonferroni correction for multiple testing. Therefore, tobramycininduced alterations in these indigenous IL1rn −/− microbiota taxa resulted in suppression of arthritis. This was confirmed by histological examination of arthritic joints which showed a significant reduction of synovial inflammation as well as cartilage destruction and a nonsignificant reduction in bone erosion upon treatment with tobramycin (Fig. 5d, e).
Aberrations of the intestinal microbiota and LP IL-17 production in IL1rn −/− mice partly depend on TLR4 TLR4 plays a major role in recognition of Gram-negative bacteria [33]. We previously showed that IL1rn −/− Tlr4 −/− mice have a marked and sustained reduction of arthritis [12]. Therefore, we assessed whether TLR4 plays a role in alterations of the intestinal microbiota and the induction of LP Th17 cells. A detailed analysis of the intestinal microbiota showed that in addition to the TLR4dependent loss of microbial diversity in IL-1Ra-deficient mice (Fig. 1b-f), alterations in Ruminococcus, Streptococcus, and Xylanibacter were partially dependent on TLR4 and were restored in IL1rn −/− Tlr4 −/− mice (Fig. 6a). Abundance of Prevotella was also restored to a statistically significant, yet minor extent (Fig. 6b). In total, 11 out of 44 taxa significantly altered in IL1rn −/− mice were normalized toward the WT levels in IL1rn −/− Tlr4 −/− mice (Additional file 1: Table S2).
We next examined the role of TLR4 in the mucosal T cell response in IL1rn −/− mice. Th17 cells require transforming growth factor-β and IL-6, plus IL-1β in mouse, for initial differentiation, and IL-23 for their functional maturation and pathogenic function [24,34]. To determine the role of TLR4 in response to IL1rn −/− intestinal microbial antigens, we cultured SI-LP mononuclear cells from IL1rn −/− and IL1rn −/− Tlr4 −/− mice ex vivo with autoclaved IL1rn −/− intestinal microbiota. SI-LP mononuclear cells from IL1rn −/− Tlr4 −/− mice produced significantly less IL-1β (Fig. 7a, P = 0.0042). Furthermore, the induction of IL-23 and IL-6 by IL1rn −/− fecal microbiota was partly TLR4-dependent (Fig. 7b, c, P = 0.0014 and P = 0.009, respectively). Reduced cytokine production in IL1rn −/− Tlr4 −/− mice was not due to an altered composition of mononuclear cells in the SI-LP, because the percentage and abundance of CD11c + MHCII + DCs as well as distinct subsets of CD103 + CD11b + , CD103 +   Figure S8). This suggests that the altered composition of microbiota in IL1rn −/− Tlr4 −/− mice as such is not responsible for the lower production of these cytokines. These observations imply a significant role for TLR4 in intestinal production of the cytokines involved in LP Th17 differentiation in IL1rn −/− mice.

Discussion
The intestinal microbiome has emerged as a key determinant of health and disease. Although advanced sequencing techniques have enabled microbiome profiling in rheumatic patients, study of the underlying mucosal responses and the functional impact on arthritis is limited in human subjects due to the requirement of invasive techniques. The animal studies presented here demonstrate aberrations in intestinal microbiota in mice developing spontaneous autoimmune arthritis, introduce commensal tobramycin-sensitive microbiota as potential triggers for arthritis, and suggest a role for TLR4 activation in mucosal induction of inflammatory pathways including Th17 induction associated with arthritis.
Our study identifies loss of microbial diversity and specific taxonomic alterations in the microbiota of autoimmune-prone IL1rn −/− mice. Importantly, loss of intestinal microbial diversity and richness also coincides with human autoimmune diseases such as diabetes and rheumatoid and psoriatic arthritis [14,16,35]. Among the microbiota increased in IL1rn −/− mice, Streptococcus species are known inducers of chronic TLR-mediated arthritis in animal models when injected intra-articularly [36,37]. Furthermore, a commensal Helicobacter (H. hepaticus) has been shown to induce IL-23 and mediate T cell-dependent gut inflammation in immunocompromised mice [38]. The decreased Barnesiella in IL1rn −/− mice is consistent with a previous study associating the abundance of Barnesiella with resistance to arthritis in HLA-DRB1*0402 mice [39]. A specific species of Prevotella, P. copri, is overrepresented in patients with new-onset RA [14] and was recently shown to increase colonic Th17 cells and promote arthritis in SKG mice after co-exposure to the fungal component zymosan [18]. On the other hand, P. histicola has been reported to suppress collagen-induced arthritis in transgenic mice expressing RA-susceptibility gene HLA-DQ8 [40]. Therefore, the immunomodulatory effects of the gut microbiota, including Prevotella, are species-and sometimes even strain-dependent. Due to inherently limited resolution of 16S rRNA gene sequencing, our data on the abundance of Prevotella is limited to the genus level and the exact Prevotella species altered in IL1rn −/− mice remain unclear. Overall, it is tempting to speculate that complex alterations in several taxa determine the net mucosal response to affect arthritis. It should also be noted that fecal bacterial community structures do not fully mirror the site-specific luminal or mucosaassociated microbiota profiles and were used in this study as a proxy of the gut microbiota of the Il1rn −/− mice.
Our previous studies showed that IL1rn −/− Tlr2 −/− mice develop a more severe arthritis compared with IL1rn −/− mice [12]. Given that additional TLR2 deficiency did not affect the microbiota of IL1rn −/− mice to a major extent (Fig. 1), we speculate that severe arthritis in IL1rn −/− Tlr2 −/− mice is due to the altered host immune response, specifically reduced function of Treg cells [12], rather than alteration in the microbiome. However, the data regarding lack of a major influence of TLR2 deficiency on Il1rn −/− microbiota should be interpreted with caution due to the absence of littermate Il1rn −/− Tlr2 +/+ mice in our studies.
IL1rn −/− mice had specific expansion of intestinal Th17 cells. The pathogenic relevance of IL-17 in the development of arthritis in IL1rn −/− mice has been demonstrated before, since both IL-17 gene deficiency and treatment with neutralizing anti-IL-17 antibodies inhibit arthritis [4,8]. A previous study showed that γδ T cells rather than Th17 cells represent most IL-17-producing T cells in the inflamed joints of IL1rn −/− mice [41]. While γδ17 and Th17 cells may have complementary pathogenic roles in the development of IL1rn −/− arthritis, our data suggest that IL-17-producing cells located in lamina propria and induced by IL1rn −/− intestinal microbiota are TCRβ-expressing CD4 + Th17 cells (Fig. 4 and Additional file 1: Figure S4).
The expansion of LP Th17 cells in IL1rn −/− mice was caused by the dysregulated microbiota as confirmed by fecal transfer experiments (Fig. 3g-i). A critical pathogenic link to the spontaneous arthritis was revealed by our germ-free and antibiotic treatment studies (Figs. 4 and 5). Other previous studies which demonstrated the involvement of the gut microbiota in exacerbation of autoimmune arthritis found SFB as the responsible microorganisms. One study showed a role for vancomycinsensitive microbiota including SFB in the induction of IL-17-and autoantibody-driven arthritis in K/BxN mice [13], and another showed that SFB can lower the activation threshold of self-reactive T cells and promote the differentiation of arthritogenic Th1 cells in a T cell transfer model of arthritis [42]. Our data are the first to demonstrate that although SFB colonization exacerbates arthritis, among the dysregulated indigenous microbiota present in the IL1rn −/− mice, those sensitive to tobramycin, i.e., Helicobacter, Flexispira, Clostridium and Dehalobacterium, are potential candidates to promote arthritis in a genetically susceptible host. This is relevant for human disease, given that SFB were not found in genome-wide sequences of 263 gut metagenomes of human adults [30,31].
Our data also provide the first evidence for the involvement of TLR4 in defining the intestinal mucosal T cell phenotype. TLR4 activation of LP mononuclear cells by IL1rn −/− microbiota-induced IL-1β, IL-23 and IL-6 ( Fig. 7a-c). Microbiota-induced IL-1β is critical for the development of steady-state Th17 cells in the gut [43]. IL-1 also synergizes with IL-6 and IL-23 to regulate early differentiation of Th17 cells and maintain cytokine expression in effector Th17 cells [44]. It was recently shown that infectious triggers such as influenza lung infection and colitis trigger an IL-1β-induced Th17 differentiation and promote arthritis induced by KRN transgenic T cells [45]. Interestingly, a subset of human CD14 + CD163 low lamina propria cells expressing both macrophage and DC markers has been found to express TLR4, produce IL-1β and IL-6 upon TLR4 stimulation, and induce Th17 differentiation [46]. However, the specific subset of LP phagocytes that orchestrates the phase-dependent TLR4-mediated mucosal response to microbiota in our studies remains to be determined.
Several studies have shown that TLR4 deficiency and systemic inhibition of TLR4 using specific antagonists or neutralizing antibodies can suppress experimental arthritis [12,[47][48][49]. Importantly, TLR4 is believed to be hyper-responsive in both blood monocyte-derived DCs and CD14 + synovial fluid macrophages of RA patients compared with healthy controls [50,51]. Pathways associated with TLR signaling are upregulated in synovial fluid macrophages of patients with RA. A proinflammatory role for TLR4 during arthritis has previously been widely attributed to TLR4 activation by endogenous damage-associated molecular patterns present in the joint rather than microbial agonists [12,49,52,53]. Our observations suggest that TLR4-mediated modulation of the mucosal immune response in intestinal LP may be another function involving TLR4 in arthritis.

Conclusions
Our study reveals a crucial role for IL-1Ra in regulation of the diversity and the composition of intestinal microbiota and a balanced T cell response in the intestinal LP. We show that the aberrant microbiota in IL1rn −/− mice have the capacity to enhance LP Th17 cells which are associated with arthritis, likely via TLR4-induced production of IL-1β, IL-6, and IL-23. Although IL1rn −/− intestinal microbiota do not cause arthritis in a normal (WT) host, these microbiota, in particular tobramycinsensitive bacteria, contribute to the development of arthritis in IL1rn −/− mice. Our data suggest that the interplay between IL-1Ra, intestinal microbiota, TLR4, and mucosal T cells may serve as a potential predisposing or initiating event in the context of autoimmune disease and provide opportunities to control RA.

Mice
IL1rn −/− mice on a BALB/c background were kindly provided by Dr. M. Nicklin (Sheffield, UK) [54]. WT BALB/c mice were purchased from Harlan, UK. Mice were co-housed in filter-top non-individually ventilated (non-IVC) cages in the same room in our animal facility for at least 8 weeks prior to feces collection for pyrosequencing. IL1rn −/− Tlr4 −/− mice and their IL1rn −/− Tlr +/+ littermates were generated as described before [12] and used for microbiota sequencing. IL1rn −/− Tlr2 −/− mice were compared to non-littermate IL1rn −/− mice in this study.

Microbiota sequencing and data analysis
Fecal bacterial DNA from 15-week-old mice was isolated using phenol-, chloroform-, isoamyl alcohol-based extraction (Sigma). Sequencing was performed by DNAVision (Charleroi, Belgium) on a Roche 454 GS-FLX System using 16S rRNA gene bar-coded primers targeting the V5-V6 conserved DNA regions (forward primer 784F: 5′-AGGATTAGATACCCTGGTA-3′, reverse primer 1061R: 5′-CRRCACGAGCTGACGAC-3′) [55]. For gene sequence analysis, a customized workflow based on Quantitative Insights Into Microbial Ecology (QIIME version 1.2) was adopted (http://qiime.org/) [56]. Settings recommended in QIIME 1.2 tutorial were applied. Additionally, reads were filtered for chimeric sequences using Chimera Slayer as described before [57]. OTU clustering was performed with settings as recommended by QIIME [58] using an identity threshold of 97%. The Ribosomal Database Project classifier version 2.2 was used for taxonomic classification [59]. Hierarchical clustering of samples was performed using UPGMA with weighted UniFrac as a distance measure as implemented in QIIME 1.2. For statistical analysis and generation of figures, a custom QIIME implemented R-package, SciPy [60] (www.Scipy.org), Graphpad Prism version 5.0, and Microsoft® Office Excel® 2007 were adopted. Presence of SFB was assessed by real-time quantitative PCR (qPCR) on fecal DNA using SFB-specific primers as described before [61]. The delta Ct (cycle threshold) value was calculated for SFB-specific rRNA gene relative to the total (conserved) bacterial 16S rRNA genes amplified using universal bacterial primers to correct for the total bacterial DNA input. Data are presented as delta Ct (ΔCt) and relative SFB expression calculated as 2 −ΔCt × 10,000 (Additional file 1: Table S3).

Microbiota transfer and co-housing
IL1rn −/− microbiota were transferred to WT mice by oral gavage of 200 μl of a homogenized IL1rn −/− fecal suspension prepared in sterile PBS. Immediately hereafter, the gavaged WT mice were co-housed with IL1rn −/− mice in the same individually ventilated cage for a period of 6 weeks to ensure sustained microbiota transfer by coprophagy. The control WT mice were gavaged with their own fecal suspension and housed separate from IL1rn −/− fecal-transplanted mice in IVC cages. To verify microbiota of conventional IL1rn −/− mice can trigger arthritis, GF IL1rn −/− mice received either 200 μl of sterile water or 200 μl fecal suspensions of conventional IL1rn −/− mice and were monitored for the development of arthritis for 8 weeks. In some studies, IL1rn −/− mice were co-housed with IL1rn −/− Tlr4 −/− mice for 10 days before analysis of LP T cells.

Antibiotic treatments and reconstitution with SFB
Intestinal microbiota were depleted using a cocktail of metronidazole (Acros Organics), neomycin trisulfate (Sigma), and ampicillin sodium salt (Sigma) (all 1 g/l) provided in drinking water for 1 week. Indicated groups received 200 μl fecal suspensions of SFB-monocolonized mice by oral gavage 1 week after ceasing antibiotics. For single antibiotic treatments, ampicillin sodium salt (1 g/l), metronidazole (1 g/l), vancomycin hydrochloride (0.5 g/l, Fisher Scientific), or tobramycin sulfate (1 g/l, Centrafarm) was added to drinking water for 8 weeks and refreshed once a week. Sucrose (6 g/l) was added to drinking water of all groups including controls during treatments.

Isolation of lamina propria cells
Lamina propria mononuclear cells were isolated from the small intestine and colon after removing mesenteric fat and Peyer's patches, followed by incubation with 33 mM EDTA on ice for 30 min to remove epithelial cells, and subsequent digestion with 1 mg/ml collagenase-D (Roche) and 10 μg/ml DNAse I (Sigma) at 37°C for three cycles of 15 min. LP lymphocytes were then harvested at the interphase of a 40:80% percoll gradient (Sigma), washed thoroughly, and used in culture as described below.

Assessment of arthritis
Severity of arthritis was scored using a previously standardized arbitrary scoring system on a 0-2 scale per paw [12]. Arthritis developed only in ankle joints (maximum score of 4). For histology, total ankle joints were isolated and fixed in 4% formaldehyde for 4 days, thereafter decalcified in 5% formic acid and embedded in paraffin. Tissue sections of 7 μm were stained using hematoxylin and eosin to study synovial inflammation, cartilage destruction, and bone erosion. Each parameter was scored on a scale from 0 to 3 in a blinded manner.

Statistics
Group measures are expressed as mean + SEM. Statistical significance was tested using an unpaired two-tailed Mann-Whitney U test to compare two and ANOVA to compare more groups, with Bonferroni correction for multiple testing when applicable (GraphPad Prism 5.0). Arthritis scores were compared using repeated measures ANOVA with Bonferroni correction. Significance is indicated on figures as follows: n.s. (not significant), *P < 0.05, **P < 0.01, ***P < 0.001.