CD44 deletion leading to attenuation of experimental autoimmune encephalomyelitis results from alterations in gut microbiome and short-chain fatty acids

INTRODUCTION

Multiple sclerosis (MS) is an inflammatory Tcell–mediated autoimmune disease of the central nervous system[1] known to have an interplay between genetic and environmental factors[2]. Approximately 400,000 individuals in the United States and 2.5 million individuals worldwide were found to be affected by MS[3]. Despite advancements in genomic[3], epigenomic[4] and proteomic[5] approaches to understand MS, the precise etiopathogenesis of MS remains unresolved and clearly additional studies are still needed. Among the variety of immune cells that contribute to the pathogenesis of MS, CD4+ T helper 17 (Th17) cells and CD4+ T helper 1 (Th1) cells have been well characterized for their ability to cross the blood-brain barrier and trigger inflammation in the CNS.

CD44, a ubiquitously expressed type I transmembrane glycoprotein, is implicated in immune cell migration, activation, and differentiation. It is encoded by 20 exons and 19 introns of which ten central exons (variable exons v1–v10) undergo extensive alternative splicing. This may explain the wide range of functions that the CD44 molecule can regulate. Hyaluronan (HA), a main carbohydrate component of the extracellular matrix, is the principal ligand for CD44. Previous studies from our laboratory showed that CD44may also regulate inflammation, in as much as, CD44 deficiency inhibits proinflammatory Th1 and Th17 cells while promoting CD4+ Th2 and Treg differentiation[6, 7]. Thus, CD44 deficiency led to decreased inflammation and amelioration of an experimental form of MS called Experimental Autoimmune Encephalomyelitis (EAE).

Recent studies suggested that microbial dysbiosis is a critical player in the pathogenesis of various autoimmune diseases, including MS. In rodent models, various discrepancies in gut microbiota were found to be associated with EAE-susceptibility and EAE-resistance[8]. In humans, dysbiosis in the gut microbiota was observed in patients with MS[9, 10]. Apart from modulation of the immune system, gut microbiota also plays a significant role in metabolic functions such as retrieval of dietary sugars, production of short-chain fatty acids (SCFAs) such as acetate, propionate, butyrate and deconjugation of bile acids[11]. The SCFAs have been shown to generate IL-10+ regulatory T cells (Tregs), thereby promoting the immune tolerance but can also induce Th1 and Th17 cells, and inflammatory responses[12]. In another study, long-chain fatty acids (LCFAs) were shown to promote the differentiation of Th1 and Th17 cells while dietary SCFAs expanded Tregs, and furthermore, these fatty acids reciprocally regulated EAE[13].

Interestingly, CD44 has also been shown to interact with bacteria. CD44 was shown to serve as a receptor for Group A Streptococcus (GAS) and played a role in the invasion and colonization of sites of infection[14, 15]. Also, CD44 interacted with IpaB, a Shigella protein which led to the invasion of the bacteria into the epithelial cells[16]. These data indicated the possibility that CD44 deletion may cause microbial dysbiosis. In the current study, therefore, we tested the central hypothesis that CD44 knockout (CD44KO) mice may have altered microbiota that favors the induction of Tregs rather than Th1 or Th17 cells and that this may explain why they are more resistant to induction of EAE when compared to CD44 wild type (CD44WT) mice. To test this hypothesis, we addressed the following questions: First, how does the bacterial diversity partition within and between CD44WT and CD44KO mice? Secondly, is there a correlation between bacterial diversity and CD44 deletion? Thirdly, does deletion of CD44 lead to altered bacterial diversity and can such microbiota attenuate EAE in CD44WT mice? Our studies demonstrated that CD44KO mice when compared to CD44WT mice have unique microbiota which may play a critical role in ameliorating EAE.

RESULTS

DISCUSSION

CD44 has been well characterized as a molecule expressed on long-lived memory T cells and has been implicated in cell migration, activation, and differentiation. Several studies from our laboratory demonstrated that CD44 deficiency in mice leads to increased resistance to induction of EAE [20]. This may result from the fact that CD44 deficiency leads to increased induction of Th2/Tregs and decreased differentiation of Th1/Th17 cells [6, 21, 22]. In the current study, we make an exciting observation that CD44 deficiency in mice leads to altered gut microbiome, which in turn may play a crucial role in suppressing inflammatory T cell differentiation, thereby accounting for the amelioration of EAE. Our studies also suggest that certain gene deletions in mice may result in altered microbiota which may impact immune and other functions, thereby indirectly contributing to unique phenotype.

Microbial colonization of mucosal tissues in early stages of life, as well as commensal bacteria found in the mucosa of adults, regulate the immune functions including tolerance, inflammatory diseases, allergies, and autoimmunity [23–25]. Studies showed that a dysbiosis in gut microbiome plays a key role in the inflammatory and autoimmune diseases such as Crohn's [26] and multiple sclerosis [8]. The role of microbiota composition in the regulation of immune functions was further strengthened by the evidence showing that LCFAs, can exacerbate EAE by promoting the differentiation of pro-inflammatory Th1 and Th17 cells through regulation of p38-MAPK pathway. In contrast, SFCAs expanded Tregs by suppressing JNK1 and p38 pathway and treatment with SFCAs ameliorated EAE [13]. Human studies have also suggested moderate dysbiosis in gut microbiota in patients with MS thereby suggesting that altering the gut microbiota may serve as a strategy for the prevention and treatment of MS [9].

Knocking out specific genes in mice has provided an invaluable tool to researchers to study their functions. Similarly, CD44KO mice have been used extensively by us and others to study the role of CD44 on immune functions. Our studies demonstrated that CD44KO mice exhibit a phenotype that is consistent with decreased inflammation, specifically in EAE model [6, 7, 21, 27–30]. While our studies suggested a connection between CD44 deletion in the CD44KO mice and dysbiosis, the direct role of CD44 on immune cells and other functions such as extravasation cannot be ruled out. For example, in another study, CD44KO mice showed delayed onset of EAE when compared with wild-type animals, due to impaired extravasation of CD4+ T cells into inflammatory sites [31]. CD44 may also act through Osteopontin (Opn), a pleiotropic cytokine that acts as a ligand for CD44 and exacerbates EAE by skewing T cell differentiation towards Th1 and Th17 cells. Thus, anti-Opn treatment reduced the clinical severity of EAE [32]. Also, antibodies against CD44 prevented EAE by targeting homing of encephalitogenic T cells [33]. On the other hand, because CD44 serves as a receptor for certain bacteria to invade the tissues, it is also possible that any treatment that activates (Opn) or blocks (antibodies) CD44 may also have an impact on the microbiota. The current study also has a broader impact on other studies using gene knockout mice by suggesting that the phenotype of knockout mice may be influenced indirectly by the gut microbiota in addition to the possible direct effects of gene deletion on specific cells or tissues. Thus, clearly additional studies are necessary to dissect the direct action of CD44 versus its indirect role through modulation of microbiota [14–16].

To monitor the evolution of gut microbial communities in CD44WT and CD44 KO mice, we pyrosequenced, bar-coded, amplified bacterial 16S rRNA in the fecal samples. At the phylum level, classification of OTU sequences, identified three phyla, Bacteroidetes, Firmicutes and Proteobacteria that were found to make up 97% of all the reads (Fig. 2C). These results are in accordance with the results from previous reports that Firmicutes and Bacteroidetes account for >90% whereas Actinobacteria and Proteobacteria are regular but account for <1–5% of the bacterial population in the murinecolon [34, 35]. Results showed a relatively higher average Firmicutes/Bacteroidetes ratio but they were not statistically different between EAE-induced CD44WT and CD44KO groups (Fig. 2C). Among the other diverse phyla that are classified using RDP classifier, three phyla, Candidatus Saccharibacteria, Deferribacteres and Proteobacteria showed a significant difference (p<0.05) in the percentage abundance of OTUs between the EAE-induced CD44WTs and the CD44KOs (Table S2). The phyla Proteobacteria and Deferribacteres harbor genera previously associated with MS and other immune disorders [9, 36, 37] whereas Candidatus Saccharibacteria is the newly identified phylum associated with MS from our study. Candidatus Saccharibacteria phylum is a well-known, ubiquitous bacterial phylum described from 16S rRNA gene sequence and genome data only [38]. Therefore, we did not consider this phylum for our further analysis.

Generic level taxonomic classification of significant phyla showed that the genera Heliobacter (EAE-CD44WT: 0.10 ± 0.03511, EAE-CD44KO: 0.72 ± 0.28148), Parasutterella (EAE-CD44WT: 0.33 ± 0.08185, EAE-CD44KO: 0.17 ± 0.04163) belonging to the Proteobacteria, Mucispirillum (EAE-CD44WT: 1.13 ± 0.41645, EAE-CD44KO: 0.10 ± 0.041633) belonging to the Deferribacteres (Fig. 3A), seven genera belonging to Bacteroidetes, and one genus belonging to the Firmicutes (Fig. 3B and C) showed a significant difference (p<0.05) between the EAE-induced CD44WT versus CD44KOs. Among the significant genera belonging to the Proteobacteria, bacteria belonging to the Heliobacter play a role in the regulation of marginal zone B cells in the absence of IL-10 signaling [39]. Bacteria belonging to the Mucispirillum of the phylum Deferribacteres, on the other hand, play a role in the transient increase of pro-inflammatory responses in the mucosa and elevated levels of inflammatory cytokines and plasma amines [40].

To analyze species level relationship in EAE-induced CD44WTs and the CD44KOs, the samples were subjected to metagenomics analysis using MG-RAST server. Among the significant genera, only species belonging to the Bacteroides of the Bacteroidetes phylum showed the majority of the sequences assigned at the species level. These species included Bacteroides sp. 2_1_16, Bacteroides caccae, Bacteroides sp.D1, Bacteroides fragilis, Bacteroides sp. 3_1_33FAA, Bacteroides vulgatus, Bacteroides uniformis, and Bacteroides sp. 2_1_22. Other species showing high abundance in OTUs included; Listeria monocytogenes, Streptococcus suis, Clostridium botulinum, Listeria seeligeri, Lactobacillus rhamnosus, Parvimonas micra, Heliobacterium modesticaldum and Pseudoflavonifractor capillosus belonging to the Firmicutes, and Curvibacter putative symbiont of Hydra magnipapillata, and Reineke ablandensis belonging to the Proteobacteria (Fig. 4A and D). The heat map representing the relationship between the metagenomic sequences of the EAE-induced CD44WTs and the CD44KOs showed differences in the relative abundance of the species (Fig. 4B and C). Further, we analyzed the top 25 species that were assigned for each OTU sequences to check which is significant among the Bacteroides genera. The results showed that the species, Bacteroides acidifaciens, Bacteroides nordii, Bacteroides plebeius, Bacteroides salanitronis and Bacteroides stercoris (Fig. 4E) had a significant difference in the percentage abundance between the EAE-induced CD44WT and the CD44KO mice, indicating that these species played a potential role in inducing resistance to EAE in the CD44KO mice.

Analysis of core microbial functions, rather than core microbial communities will assist in elucidating the associated alterations in physiological or disease states. Some of these function-based approaches include searching against the public domain databases such as Kyoto Encyclopedia of Genes and Genomes (KEGG) [41] and Clusters of Orthologous Groups (COG) analyses [42]. We have annotated the metagenomic sequences from MG-RAST pipeline against KEGG database. The KEGGs that were most abundant in EAE-induced CD44KO mice were DNA-directed RNA polymerase subunit beta (K03046), molecular chaperone DnaK (K04043), hydroxylamine reductase (K05601), protein Tex (K06959), regulator of G-protein signaling 3 (K07524), pellino (K11964), heterogeneous nuclear ribonucleoprotein R (K13161), E3 ubiquitin-protein ligase (K15710), citron Rho-interacting kinase (K16308) and leucine-rich repeat transmembrane protein FLRT (K16362) (Fig. 5B).

Among the KEGGs, that are predicted through the functional profiling, RGS3 (K07524) plays a critical role in inflammation and immunity [43] by controlling T cell migration [44]. Induction of immune cell chemotaxis by SCFAs is dependent upon the activation of a G protein-coupled receptor [45]. Results from taxonomic profiling showed a change in the Bacteroidetes from phylum level to species, and fermentation of fiber by the microbes belonging to these Bacteroidetes phylum produce large amounts of SCFAs in the colon and blood, which are known to play a vital role in immune regulation [46]. For these reasons, we explored the levels of SCFAs in the EAE-induced CD44WT, CD44KO and EAE-induced CD44 WT mice with a fecal transfer from CD44 KO mice. Among the SCFAs, propionic acid (C3) showed a significant difference in the EAE-induced CD44WTs, CD44KOs and mice that received the fecal transfer. Bacteroidetes and some Firmicutes are good producers of C3, and C3 can directly suppress histone deacetylases thereby generating the Tregs [47].

While dogma dictates that Firmicutes are responsible for butyrate production and Bacteroidetes are responsible for acetate and propionate production, there is evidence for the contrary [21]. Bacteroidetes’ primary fermentation produces amino acids and there are several non-canonical butyrate-producing pathways that use amino acids as major substrates, including the lysine, 4-aminobutyrate and gluterate pathways [20]. Recently, these often over-looked pathways that contribute to butyrogenic potential have been predicted to occur in Bacteroidetes. In fact, using metagenomic analysis of samples from healthy human patients from the Human Microbiome Project, significant overlap was found between genes involved in the lysine and 4-aminobutyrate pathways of butyrate production and those expressed by members of the Bacteriodetes phylum [20]. Specifically and of importance to our findings, members of the Odoribacter genus dominated the lysine and 4-aminobutyrate pathways, and therefore, increases in Odoribacter may account for the increases in butyric acid observed in CD44KO EAE relative to CD44WT EAE mice [20]. Moreover, Odoribacter is known to produce propionic acid as an end product of glucose metabolism, perhaps also contributing to increases in propionic acid [22]. Increases in Parabacteroides and Alkalitalea genera (both of the Bacteroidetes phylum) may account for increases in propionic acid observed in CD44KO EAE relative to CD44WT EAE mice, as the end fermentation product of Parabacteroides includes propionic acid [48], while the main product of beachwood xylan metabolism by Alkalitalea is propionate [48]. Finally, Alkalitalea saponilacus produces minor amounts of butyrate. Therefore, the increases in Alkalitalea genus may also contribute to butyrogenic potential of the CD44KO microbiome [28]. Consequently, increases in the genera Alkalitalea, Odoribacter and Parabacteroides, all of which belong to the Bacteroidetes phylum, in CD44KO EAE compared to CD44WT EAE may be responsible for the observed increases in propionic and butyric acids.

In conclusion, our study demonstrates that CD44 plays a critical role in the regulation of EAE through alterations in the gut microbiome profile and SCFA production. CD44 deletion leads to alterations in bacteria belonging to three major phyla, Bacteroidetes (especially Bacteroides genera), Firmicutes, Proteobacteria, and the SCFA, butyric and propionic acid, which may play a key role in the amelioration of EAE. Our studies also shed new light on the role of altered microbiota in gene-knockout animals which may indirectly influence the mouse phenotype.

MATERIAL AND METHODS

Supplementary Material