Gut Microbiota as a sensor of autoimmune response and treatment for rheumatoid arthritis

Summary

Rheumatoid arthritis (RA) is considered a multifactorial condition where interaction between the genetic and environmental factors lead to immune dysregulation causing autoreactivity. While among the various genetic factors, HLA-DR4 and DQ8, have been reported to be the strongest risk factors, the role of various environmental factors has been unclear. Though events initiating autoreactivity remain unknown, a mucosal origin of RA has gained attention based on the recent observations with the gut dysbiosis in patients. However, causality of gut dysbiosis has been difficult to prove in humans. Mouse models specially mice expressing RA-susceptible and -resistant HLA class II genes have helped unravel the complex interactions between genetic factors and gut microbiome. This review describes the interactions between HLA genes and gut dysbiosis in sex-biased preclinical autoreactivity and discusses the potential use of endogenous commensals as indicators of treatment efficacy as well as therapeutic tool to suppress pro-inflammatory response in rheumatoid arthritis.

Autoimmune diseases are characterized by chronic inflammation as a consequence of immune dysregulation arising from cellular and humoral response against self-proteins. While most autoimmune diseases have a strong genetic predisposition, the etiology of these diseases remains a mystery. For some autoimmune diseases such as rheumatoid arthritis, preclinical autoreactivity could occur many years before the symptomatic disease. We do not understand the events that cause autoreactivity though different disease outcomes in genetically identical individuals and low concordance rates in twins suggests a multifactorial etiology ^1-4. The role of environmental factors, specially smoking and infections, has been studied extensively. Over the last decade, an interest in the contribution of microbiome and its function has burgeoned.

A complex array of risk loci have been associated with various autoimmune diseases, and genes within the Major Histocompatibility Complex (MHC) have shown the strongest association ⁵. The MHC region encodes for the Human leukocyte antigen (HLA) class II alleles, DR and DQ. Certain HLA class II alleles occur with significantly high frequency in autoimmune diseases compared to healthy individuals. Most of the autoimmune disease HLA-loci associations belong to 3 alleles, DR2, DR3 and DR4⁶. These genes have undergone multiple mutational events resulting in extensive polymorphism. Evidence suggests there is a selection of these alleles during evolution with some of the sub-types of these alleles offering protection from pathogens better than the other alleles⁶. However, while these alleles generate a robust response to clear pathogens, they can cause bystander effects or if there is molecular mimicry between the pathogen and self-proteins, it may activate cellular and humoral response to self-proteins. Since antigen presentation is mediated by HLA molecules, it directly implicates the MHCII genes in the pathogenesis of autoimmune diseases. Besides environmental factors, geographical location have also been corelated to autoimmunity. For example, multiple sclerosis has been linked to latitude of residence before the age of 15 years, suggesting the disease process may start earlier in life even if it becomes symptomatic in later years. Additionally, a higher incidence of autoimmune diseases in women point to a role of sex-hormones. Among environmental factors, smoking has been shown to be a contributor to severe disease in most autoimmune diseases suggesting interactions between environmental and genetic factors may be involved in initiation of disease^4,7. This also raises the question about the origin of autoimmunity and whether conditions affecting specific tissue may be initiated elsewhere. There is now a plethora of evidence that suggests a role of mucosal surfaces and microbiome, especially the gut microbiome, in autoimmunity. The review details the role of factors that impact both RA and microbiome in inflammation and use of endogenous commensals for modulation of immunity in RA.

Rheumatoid Arthritis- a multifactorial condition

Rheumatoid arthritis (RA) is a chronic disease which is characterized by autoreactive cellular and humoral responses to self-proteins like, collagen, vimentin and fibrinogen, causing immune dysregulation and consequently progressive joint damage with loss of function and disability. Autoantibodies to self-proteins can be present 1-10 years prior to patients being symptomatic ⁸. Additionally, RA patients present with significant clinical heterogeneity with involvement of other organs, suggesting systemic inflammation⁹. Although the disease affects joints, initiation of disease likely occurs in an extra-articular site such as the lungs, gut or oral mucosa^10,11. The transition from asymptomatic autoreactivity to joint inflammation is still an active area of investigation.

Decades of research using clinical demographics and family history has identified various genetic and environmental factors that could be involved in the onset of autoreactivity or progression of severe disease. Familial cases of RA suggest it is heritable and previous research has been partially successful in identifying certain candidate gene and alleles ^12-14 with HLA-DR4/DQ3 haplotype showing the strongest association with sporadic and familial RA^8,12,15,16.

Using single nucleotide peptide (SNP) mapping, multiomics and clinical factors, genome wide association studies (GWAS) studies from multiple ethnic groups from around the world have observed different genetic loci associated with RA in various populations¹⁷. A compilation of data pointed to many non-HLA loci with the protein tyrosine phosphatase non-receptor type 22 (PTPN22) polymorphism providing the major risk. However, HLA class II alleles provided the strongest association and odds ratio for predisposition to RA; seropositive RA patients showed a much stronger association as compared to seronegative patients¹⁸. HLA-DRB1*0401 allele was observed in around 70% of seropositive RA patients, but not in seronegative patients¹⁹. This suggests that genetic factors along with clinical parameters could predict heterogeneity of disease. RA patients in some ethnic groups carried DRB1*01 or DRB1*10 rather than the commonly associated DRB1*0401. To explain the non-HLA-DRB1*0401 positive RA patients in various ethnic groups, a hypothesis called “shared epitope” was coined. According to the hypothesis, HLA-DR alleles sharing the peptide binding 3^rd hypervariable region with DRB1*0401 also predispose susceptibility to RA while those alleles sharing with DRB1*0402 are resistant^15,16. Substitutions at various amino acids (AA) in the peptide binding region of the DRB1 chain demonstrated that 11, 71 and 74 of HLA-DRβ1 alleles are key positions for susceptibility to RA ¹⁸. These observations indicated that a peptide presented by the shared epitope HLA molecules could generate an autoreactive response. However, no specific epitope has been defined since RA patients are positive for antibodies to many self-proteins and the causality of these self-peptides has not been proven. Importantly, the presence of RA-susceptible HLA alleles in healthy individuals contradicts the hypothesis that presentation of a peptide initiates RA. Interestingly, the presentation of DR4-derived peptides by DQ8 molecule generates an immune response in mice expressing RA-susceptible haplotype DR4/DQ8 ^20,21 supporting a role of self-epitopes presentation in RA. However, whether this contributes to autoimmunity or is required for maintaining tolerance is unclear.

Since polymorphism of many non-HLA loci have also been associated with RA ¹⁸, it could be the genetic burden that predicts the odds ratio of developing RA. Previous GWAS did not take into consideration familial history of other autoimmune diseases and linkage disequilibrium, thus a study using multi-ancestry genetic analysis was conducted to detect a causative candidate gene for RA ²². The study included 37 cohorts of RA patients and control individuals from European, east and south Asian, African, and Arab ancestries. This study identified 34 novel candidate genes, involved in the immune system as well as joint specific genes, which were comparable between European and East Asian ancestries thus providing general genetic predicability in predisposition to RA. Multi-ancestry GWAS identified causal variants in the CD4+ T cell T-bet annotation in all populations though the effect size was bigger in the European population. Besides the novel genes, the study also confirmed the role of genes encoding PTPN22, IL-6R, and tumor necrosis factor alpha induced protein (TNFAIP), identified previously in RA, and pointed to the power of genetic mapping for causal genes in RA.

Among the HLA alleles, HLA class II haplotype DRB1*0401/DQ8 provides the strongest propensity for susceptibility to RA²³. However, only a small percent of individuals positive for DR4/DQ8 develop RA, suggesting a role of other factors including infections ^12,24. The infectious etiology of RA gained attention as certain infectious agents such as Proteus mirabilis, Epstein Barr Virus, Parvovirus, Mycoplasma fermentans, Porphyromonas gingivalis, Klebsiella pneumonia and Rubella virus were linked to RA ^25-30. The concept of infections etiology was based on molecular mimicry between the pathogen-derived products and self-proteins which has the potential to cause a break in tolerance. After the clearance of an infection, the activated immune cells could generate inflammatory response against self. However, the caveat was that no specific pathogen was observed in RA patients from various ethnic populations. Based on recent observations, scientists have come to appreciate that there is a possibility that an endogenous opportunistic gut commensal could become pathogenic.

Recent data using OMICS suggest that microbiome and its function could open the black box to the causative factors in RA. Besides microbes, endogenous or exogenous, RA is impacted by various factors including, genetics (MHC and non-MHC), smoking, aging, lifestyle, diet, circadian rhythm, hormones and epigenetics. ^4,24,31-35. All these factors are also known to impact gut microbial composition. Thus, it is possible that in an individual with RA-susceptible genes, an endogenous opportunistic gut commensal in combination with other factor(s) could become pathogenic and lead to diseased state. Indeed, an immunogenetic control of gut microbiota has been suggested ³⁶. One can speculate that the MHC-based selection of T cells define the immune milieu ³⁵, which could be contributory to the microbial composition and colonization³⁶. This is supported by observations in humanized mice where the presence of the arthritis-resistant DRB1*0402 gene demonstrates a regulatory role in mouse model of arthritis ³⁷ and is associated with dynamic alterations in the gut microbiome. On the other hand, arthritis-susceptible *0401 mice display similar microbiome profile from young to old age suggesting augmented aging and microbiome as factors in RA pathogenesis ^38-40.

Historical perspective on microbiome and disease

The concept that the gut could be the site of origin of inflammation in many diseases is not novel. A famous expression attributed to Hippocrates, the father of modern medicine, is “All diseases begin in gut.” While RA has been considered a disease of the recent world, erosive RA has been described in skeletons of 6500 years ago. Furthermore, paintings from 15^th century onwards document the presence of RA⁴¹. The presence of RA in Indian tribes was also described during 15^th century. However, characterization of the disease may not have been uniform due to a lack of evidence.

Ayurveda is one of the oldest systems of medicine that originated in India and has been practiced since 1500 BC. It is based on the unique constitution of each individual called prakriti which is composed of physiologic, physical and psychological attributes. The factors impacting prakriti are diet, digestion, lifestyle and environment and is divided in to 3 sub categories which differ in biochemical, lipid biosynthesis and TLR signaling pathways, and gene expression, immune response gene and DNA repair etc. ⁴². The presence of various HLA alleles in individuals identified showed differential expression of HLA- DRB1*02, DRB1*13 and DRB1*10 in 3 sub categories. DRB1*02 and DRB1*13 have been associated with protection from autoimmunity while DRB1*10 is present with high frequency in RA patients in Indians ^15,16. These studies point to an ancient system of characterizing the immune response and biochemical pathways which are influenced by the HLA genes and environmental factors, known to impact the gut microbiome. Recent research has shown that many ayurvedic herbal preparations used for decades have a beneficial effect on the gut microbiome. One such preparation is turmeric which affects digestion and microbiome⁴³. Its active compound, curcumin has shown anti-inflammatory properties^44,45. These observations suggest that while recent advances in technology have facilitated our understanding of the impact of gut microbiome for on disease, the notion of the role of gut in health was recognized much earlier.

During late 1700s, scientists were convinced that toxins came from the colon ⁴⁶. Bifidobacteria, identified in 1899, were considered to be part of a healthy gut ⁴⁷. The concept that the gut microbes are beneficial or involved in age-related diseases was later confirmed in the early 1900s by a Nobel Laureate Elle Metchnikoff who proposed that lactic acid producers are beneficial and fermented milk products are related to longevity while bacterial toxins in the gut are related to aging ⁴⁸. Based on his work, he theorized that certain lactic acid producing bacteria, when taken enterally, can alleviate intestinal diseases, a concept which is similar to a fecal transplant. He further suggested that diet strongly influences intestinal microbes directly impacting diseases ⁴⁹. Indeed, recent studies have shown that diet is a strong modulator of gut microbiome. Many studies have documented alterations in gut microbiota and presence of metabolites as humans age ^31,32,50. These observations raised the question of whether changes in immune system during aging cause gut dysbiosis or vice versa.

The Gut Microbiome

There are trillions of microbes on the skin and inside of humans. Microbiome is a term that describes the collective genome of bacteria, viruses, and fungi including symbiont, commensal and pathobionts⁵¹. Microbiota defines microbial communities in a specified environment such as skin, oral, lung and gut. The composition of human gut microbiome has deviated from the primates, like great apes, who are considered ancestral to humans. Among primates, the human microbiome composition shows distinct differences from the gut microbiomes of New World primates and lemurs and is most closely related to Old World monkeys and apes ⁵². Some of the reasons could be the advent of agriculture and the processing of foods, hygiene, medicines, culture, environment, presence of enteric parasites, population size, and physiological changes. Partial variance of microbiome may have occurred in tandem with the expansion of human populations across the globe. For example, the distribution of Helicobacter pylori strains corresponds with the documented human migration patterns⁵³.

Each mucosal site, skin, lung, oral and gut, has its own microbiota though there is a sharing of taxa among the mucosal sites⁵⁴. Among all the mucosal sites, the gut microbiota harbors the maximum numbers and diverse bacteria and is considered our second genome. Colonization of gut microbiota starts in utero and based on the birth, Cesarean or vaginal, infants acquire skin or vaginal microbiome, respectively⁵⁵. Since mothers’ milk also carry bacteria, infants’ microbial colonization is influenced by the feeding, breast milk or commercial formula milk. These factors along with the genetic factors determine initial gut microbial colonization. While many studies suggest that Cesarean birth and formula milk are associated with obesity and other complications later in life, a comprehensive microbiome study established that mother’s microbiome imprinting is visible for first year of infant’s life and is reduced to 27% after 3 years of age. Further, the mode of delivery based differences were lost after 3 years ⁵⁶. However, the gut microbiome transmission from mother to infant was detectable even at older ages, suggesting there might be a core microbiome based on genetics or non-inherited maternal antigens (NIMA). Nonetheless, during adulthood and later years, the microbiome of an individual can be impacted by various factors that include diet, infections, use of antibiotics, transmission from co-inhabitants and other environmental factors.

Humans harbor more than 30 trillions of microbial cells and share strains with other species, However, there are traces of genetic changes that are specific to the strains present in humans⁵², suggesting there is a co-evolution, and microbes are adapting to us. Microbes digest the dietary products that humans are unable to, and produce energy for the host, and in return they have an environment where they can thrive. Besides producing energy from the nutrients, gut microbes support a variety of physiological functions^57,58. The intestine has a large thin membrane for absorption of nutrients and hence needs a robust defense system. Microbiota forms a barrier, physical and immunological, that separates the intestinal tract from the outside environment as well as produce metabolites required for epithelial cell proliferation to repair damage. Commensals compete with pathogens to protect from infections and also induce the production of IgA antibodies. Germ-free (GF) mouse models provide evidence that gut microbiota is required for the immune system development as GF mice are immunodeficient⁵⁹. Recent data suggest that microbes are crucial for metabolization/activation of certain drugs and also for the synthesis of vitamins and neurotransmitters. Thus, gut microbes perform many functions that impact immunity. The symbiotic relationship between the host and gut microbiota has the potential to elicit specific biological responses, both local and systemic, that influence host energy and lipid metabolism⁶⁰. OMIC-based metagenomics and metabolomics have helped make a rapid progress on the impact of gut microbiome on health and disease and how microbes regulate the innate and adaptive immunity ^61-66. This has contributed to understanding of how interaction among genetic and environmental factors, and the gut microbiome influence autoimmunity.

Modulators of microbiome

Diet has been reported to be a significant determinant of microbial diversity. Individuals in countries with westernized diets tend to have lower gut microbial diversity as compared to non-Westernized diets. Individuals on vegetarian diets were observed to harbor abundance of Prevotella Spp as compared to Westernized countries where processed foods are commonly ingested ⁶⁷. A higher numbers of Bacteroides with a lack certain species, such as Treponema, are present in the microbiomes of individuals from Western countries while non-Western microbiomes generally carry an abundance of Proteobacteria and Firmicutes. Further, based on the diet, the gene enrichment of a taxa can differ among omnivores versus vegetarians⁶⁸. However, diet as a sole explanation for microbial changes is probably an oversimplification. Indeed, cohabitation and environmental factors have a prominent role in transmission of microbes and gut microbial colonization⁵⁶.

The observation that there is a core microbiome^57,69 suggests involvement of endogenous factors. However, the role of genetic factors in determining the core microbiome still needs to be defined. A higher strain sharing between monozygotic twins, even if they lived apart, and high strain concordance with maternal microbiome suggests a role of genetic factors in microbial colonization ⁵⁶. Observations with naïve HLA-transgenic mice expressing RA-susceptible DRB1*0401 and RA-resistant-DRB1*0402 suggest a role of MHC genes in colonization of intestinal microbes which could be dependent on the immune response ⁴⁰. Alternatively, the plasticity of T cells would dictate that microbial changes reflect the impact of immune response. Indeed, there is a gradual change in immune response in concordance with microbial composition and function during aging^70,71. MHC-based selection of T cells and antigen presentation support the contention that MHC genes contribute to gut microbial colonization.

We are host to an ecosystem which when perturbed, can trigger dysbiosis leading to shifts in microbial diversity and function, consequently causing dysregulation of the immune system with the risk of developing various diseases. Humans have an internal biological clock that maintains daily and seasonal rhythms. Gut microbiota show both circadian and seasonal rhythms ⁷². Desynchronization of the circadian rhythm such as late night snacking or jet lag can alter microbiota and cause inflammation which has been associated with many diseases suggesting a bidirectional relation between circadian rhythms and gut microbiota⁷². Microbes live in communities which can impact the presence and abundance of other microbes and overall physiology. Dispersal and interaction of microbes from the environment in which infants are born will have differential impact on microbiota. These differences can contribute to microbiota’s resistant and resilience for maintaining a healthy microbiome. Even if humans share strains/species, the genome and gene expression of shared species can be different. Also, individual intestinal hydrolyzation of diet can impact microbial colonization as well as the genes expressed in various taxa⁷³. Additionally, host filtering can result in unique and adapted individual microbiome thereby influencing disease outcomes. Overtime some aspects that have become evident is that oxidative state and gut PH can impact the microbial composition and their genetic content and oxygen tolerance enables the horizontal transmission of gut bacterial lineages. This is important as pathogens likely transmit horizontally ⁷⁴.

There is microbial compartmentalization within the intestinal sections, likely based on the oxygen levels, PH and available nutrition, resulting in diverse microbial communities⁶⁶. For example, the duodenum has an acidic PH and only microbes that can survive the acidic conditions can colonize it. Small intestinal microbes are generally anaerobic, facultative or obligate, and are difficult to culture. On the other hand, in the colon which is more open to the outside environment, there is a mixture of anaerobic and aerobic microbial communities. The gut bacteria play a crucial role in metabolizing amino acids and metabolization is compartmentalized due to the presence of distinct bacteria in various niches ^75-77. Luminal residing bacteria use preformed AA, though adherent bacteria generally metabolize dietary components and synthesize AA ⁷⁵. Generally, the synthesis of AA occurs in the small intestine while catabolism occurs in the large intestine ⁷⁸. Both essential and nonessential AAs are utilized by the intestinal microbiota which results in metabolites production thereby contributing to generation of energy and survival of microbes ^75,79. Microbes in the colon produce metabolites like short chain fatty acids (SCFAs) and also contribute to biotransformation of bile acids which are then absorbed and transported back to liver. Thus, the functional traits of microbes differ based on the site.

Hence, the overall environment specific to an individual determines the microbial community surviving in that condition as well as the genes enriched in those microbes. With so many factors impacting the microbiota, it behaves like a fingerprint unique to each individual. Nonetheless, much about the gut microbiome remains undiscovered due to the challenging nature of culturing many microbes.

Gut microbiome as sensor of RA.

Non-communicable inflammatory diseases have been linked to the gut microbiome as well as other infectious pathogens. This concept raises the question whether endogenous commensals are contagious. There are no data that suggest this. However, transmission of an opportunist pathogen can increase susceptibility to an otherwise uncommon disease in a population. RA is a non-communicable disease which has been associated with various fecal microbes in different populations^80-83. Does that mean that these bacteria are contagious or that transmission into a population which have lower abundance of these bacteria can cause RA? Generally, microbiome is symbiont even though there are opportunist pathogens that have been associated with autoimmunity. Do we consider these opportunist bacteria contagious? Although this is a fascinating concept, the evidence is not yet there to support it.

The pathogenesis of RA is complex and not fully understood. A growing body of evidence suggests a role of mucosal surfaces, specially dysbiosis of the gut microbiome in pathogenesis of RA^{24,41,81,84,85}. Recent advances in biotechnology has led to high throughput analysis of gene expression and biological molecules in a cell using OMICS-based approaches which include, metagenomics, genomics, proteomics, metabolomics and transcriptomics. These are rapidly evolving fields and are facilitating identification of potential causal biomarkers and pathways that can be targeted for therapeutic intervention.

Metagenomic analysis of fecal samples has discovered an expansion of certain taxa in RA patients^86,87, which could be due to microbiota’s inability to regulate itself. There is a high likelihood that the perturbation induces dysbiosis resulting in a functional and metabolic profile shift which could directly contribute to chronic inflammation in RA. In addition, reduced diversity due to dysbiosis could cause low community stability and increased susceptibility to further loss or gain of certain taxa induced by environmental factors. These alterations over the course of disease potentially contribute to the exacerbation of RA. Since commensal bacteria constitute an integral and active component of the intestinal community, exploration of host and microbial interactions should open up new avenues of research for understanding the consequences of perturbations of gut microbiota which holds the potential for discovering novel treatment options.

The first clue about the role of microbiome in RA was realized from the observations of bacterial products in the synovium of RA patients⁸⁸. A few years later a comparison of fecal microbiota carried out by rtPCR of stool samples from established RA patients and controls showed dysbiosis with reduced levels of Bifidobacterium, B. fragilis, Prevotella, Clostridium coccoides Spp. in RA patients as compared to controls ⁸⁹. To understand the gut dysbiosis, metagenome analysis of fecal microbiota of RA patients, who were not yet treated, and healthy comparators was conducted, which demonstrated an abundance of Prevotella Copri in RA patients⁸². Using high throughput sequencing, another study showed dysbiosis with reduced diversity in fecal microbiota of established RA patients when compared with age and sex matched healthy controls and first degree relatives ⁸¹. The observations suggested that certain taxa of Phylum Actinobacteria, Eggerthella lenta and Collinsella aerofaciens, were expanded in RA patients independent of treatments or genetic factors. On the other hand, Faecalibacterium, the most abundant intestinal bacteria in humans, was significantly reduced in RA. Predictive profiling of the fecal microbiome suggested that a combination of increased proportion of E. lenta and C. aerofaciens with reduced Faecalibacterium defined the highest risk of RA, though among these taxa, E. lenta provided the most significant association with disease status. One of the vital observations was that the random healthy controls and the first degree relatives did not differ significantly in microbial composition, indicating that the observed differences were due to the disease status⁸¹. Also, noteworthy was the high proportion of E. lenta in seropositive RA patients with higher autoantibodies, rheumatoid factor (RF) and anti-citrullinated antibodies (ACPAs), production ⁹⁰. An over-abundance of E. lenta and C. aerofaciens in established RA patients was also confirmed in other ethnic populations^91-93. E. lenta is involved in many functions⁹⁴; it produces equol, inactivates digoxin, deconjugates bile acids and metabolizes arginine via the ornithine pathway leading to production of carbamyol phosphate, used in pyrimidine synthesis. This is significant as treatment with a pyrimidine inhibitor has been successfully used in RA patients ⁹⁵. These data suggest that E. lenta can contribute to RA pathogenesis in many ways. In addition to gut dysbiosis, an oral dysbiosis was observed with depletion of Haemophilus spp along with higher numbers of Lactobacillus Salivarius in active RA patients. Interestingly, Lactobacillus is often considered a probiotic ⁹¹.

Causality versus an effect of dysbiosis or expansion of a clade is difficult to prove in humans, thus in vitro assays and animal studies were used to confirm the contribution of these gut bacteria to systemic inflammation. In vitro culture of the human epithelial cell line CACO-2 with E. lenta and C. aerofaciens led to an increase in intestinal permeability^81,90,96. In vivo, this would allow emigration of luminal contents with the possibility of activating the immune system resulting in systemic inflammation. However, these studies pointed to an association but did not prove dysbiosis as a cause or an effect of disease. Collagen-induced arthritis (CIA) has been used as an animal model for RA for decades as it shares the immune component of RA pathogenesis⁹⁷. In animal models, microbial dysbiosis during arthritis development and a potential role of endogenous commensals in immune competent as well as antibiotic treated mice has been shown ^40,98. However, except for mouse segmented filamentous bacteria (SFB), that cause arthritis in germ free conditions, no single endogenous microbe causes arthritis in mice. An increase in Lactobacillus was observed in CIA-susceptible mice and when the fecal microbiome of the susceptible mice was transplanted into germ free mice, there was increased presence of Collinsella and Lactobacillus⁹⁹.

In a humanized mouse model of arthritis, gavaging DQB1*0302 (DQ8) mice with C. aerofaciens augmented gut permeability and enhanced arthritis severity⁸¹. Similar observations were noted in DQ8 mice gavaged with E. lenta⁹⁰. To delineate the effect of an expansion of E. lenta in the preclinical stage in RA patients, E. lenta was inoculated in DQ8 mice before the induction of arthritis. Augmenting the abundance of E. lenta altered the gut microbial composition and metabolic profile consequently enhancing proinflammatory conditions and disease severity⁹⁰. However, just increasing the abundance of E. lenta without another insult to the immune system did not induce arthritis in DQ8 mice. These observations indicate that dysbiosis or abundance of one or more taxa may be sufficient to induce pro-inflammatory conditions, but the onset/progression of RA requires another insult to break tolerance. The observations in RA patients and humanized CIA model provide an outline of how opportunist commensals of intestinal microbiota and genetic factor interactions may be involved in pathogenesis (Fig 1).

Figure 1: The Role of Gut microbiome in Rheumatoid arthritis (RA).

Genetic factors, major histocompatibility complex (MHC II) and others, epigenetics, sex-hormones, environmental factors including lifestyle, diet, smoking and infectious agents, and circadian rhythm shape the microbial composition at various mucosal surfaces of the body including the gut microbiota, thereby establishing an individualized microbiome. Interactions among these various factors, under certain circumstances, can cause dysbiosis resulting in an expansion of some taxa with contraction of beneficial taxa, reducing diversity and altering metabolic profile. This can initiate intestinal innate and adaptive immune response leading to production of pro-inflammatory cytokines. Molecular mimicry of bacterial protein with self-protein in humans can break tolerance, based on the host’s MHCII genes, and cause preclinical autoreactivity, production of autoantibodies (AutoAbs), rheumatoid factor (RF) and anti-citrullinated antibodies (ACPAs), ensuing a pro-inflammatory milieu and an increase in gut permeability. Rheumatoid arthritis is associated with reduced diversity and Faecalibacterium (Faecalibac) concomitant with high abundance of Collinsella aerofaciens (CA) and Eggerthella lenta (EL), and/or Prevotella copri (PC). Gut microbes such as EL have estrogenic properties that can contribute to sex-bias. Reduced amino acids and short chain fatty acids (SCFAs) can decrease epithelial repair and synthesis of certain proteins. Translocation of luminal contents including bacterial products and metabolites amplify the inflammatory cascade including production of TNF-α, IL-6 and other senescence associated secretory proteins (SASP). Emigration of activated T and B cells systemically and in the joints triggers the extra-intestinal inflammation. In the synovium, fibroblast and other cells are activated by the antibody immune complexes and activated T cells ensuing an autoimmune feedback loop. The immune cells specific to self-proteins and the presence of self-antigens in the joints cause epitope spreading amplifying the inflammatory cascade subsequently leading to cartilage loss with joint destruction. Besides gut microbiome, alterations in the lung microbiome due to exposure to environmental factors, smoking and infections, also contribute to inflammation.

Mucosal immunity and integrity is the first line of defense against the exogenous factors entering the intestine. Observations with C. aerofaciens and E. lenta in humanized mice ^81,90 indicated that one way gut microbes play a role in RA pathogenesis is via an increase in intestinal permeability which would allow microbial products to emigrate systemically. This concept was proven in humans where the presence of cellular and humoral response to HLA-DR presented 27 KD peptide (Pc27), derived from P. Copri , was reported in new onset RA patients (NORA) ¹⁰⁰. IgA antibodies to Pc27 correlated with the serum Th1 and Th17 cytokines concentrations as well as the presence of ACPA and RF¹⁰¹ . While the initial study did not show P. copri-induced arthritis in an animal model, a later study reported worse arthritis in germ free mice when utilizing the P. copri isolated from RA patients and not healthy controls ¹⁰². These observations suggest that P. copri function is dependent on the intestinal microbial communities. Indeed, a difference in gene enrichment of P. copri was observed based on the host’s diet, vegetarian vs western diet ⁶⁸.

Whether oral microbes get transferred to the gut and cause an increase in permeability remains unknown. However, a link between the periodontal disease (PD) and RA suggests that oral microbes can contribute to the disease process. Patients with RA produce antibodies to Porphyromonas gingivalis, an oral cavity commensal associated with PD ¹⁰³. P. gingivalis is the only bacterium to carry the enzyme peptidyl arginine deiminase (PAD) for citrullinating bacterial and human proteins. This can lead to presentation of citrullinated antigen causing cellular and humoral response against alpha-enolase and fibrinogen¹⁰⁴. In addition, the DNA of P. gingivalis, has been detected in synovial fluid of RA patients ¹⁰⁵, suggesting a role of oral mucosa in RA. Involvement of various mucosal sites and microbial strains might indicate distinct modes of causality.

Preclinical autoreactivity to microbial proteins in RA

Mechanistic insight into how autoimmunity begins, remains obscure. The mucosal origin of RA hypothesizes initiation and perpetuation of inflammation in the mucosa¹⁰⁶. Most human studies have described associations of various taxa in RA, which expand in patients causing dysbiosis and altered diversity. However, in RA patients autoreactivity could start 10 years prior to the actual symptomatic RA as defined by the ACR criteria ¹⁰⁷. This would suggest that most patients had certain events that initiated the autoreactive response before they were diagnosed, making it hard to determine how dysbiosis came about and why certain taxa become overabundant in RA patients. Thus, the major question that needs to be addressed is, if certain gut commensals are opportunist pathogens, and whether they share similarities with human proteins.

Molecular mimicry with environmental pathogens, though observed, has not shown a consistent response in various RA populations. However, homology between bacterial and self-proteins can generate pro-inflammatory immune response. C. aerofaciens and E. lenta, RA-associated gut bacteria, harbor similarities with certain human proteins ^21,90. C. aerofaciens shares sequences with the HLA-DR4, suggesting it can contribute to pathogenesis via mimicry with a self-peptide. The major facilitator superfamily (MFS) transporter in E. lenta shares homology with human type II collagen (CII) epitopes that have been shown to bind and be presented by DQ8 molecule ^90,108. Naïve DQ8 mice gavaged with E. lenta produced RF, suggesting E. lenta contributes to B cell activation⁹⁰. Splenic cells from naïve DQ8 mice generate in vitro cellular response to E. lenta derived proteins as well as CII-derived peptides suggesting molecular mimicry at play in preclinical autoreactive response. Also, colonization with E. lenta of naïve DQ8 mice led to a change in immune response to Th17⁹⁶, supported by an increase in germinal center CD19+PNAhi and IL-17/IFN-γ producing B cells as well as CXCR5+ T follicular helper (Tfh) cells. Importantly, activated Tfh cells have been shown to emigrate from intestine to periphery resulting in systemic immunity⁸⁴. Since gut microbiota is essential to maintaining homeostasis, perturbation in the gut microbiota of naïve mice after E. lenta gavage can contribute to metabolic and immune alterations. After E. lenta gavage, naïve DQ8 mice demonstrated a significant altered beta diversity accompanied with reduced prevalence of certain taxa, Clostridium, Allobaculum, Adlercreutzia and Staphylococcus and an increase in Prevotella, Arthromitus and Ruminococcus. Microbial compositional alterations of naïve mice after colonization with E. lenta resembled arthritic mice, suggesting E. lenta may contribute to arthritis directly and indirectly. E. lenta is involved in metabolization of Arginine/Ornithine to produce energy and generate citrulline as a byproduct. Metabolic alterations with reduced levels of citrulline after E. lenta gavage in DQ8 mice resemble low citrulline levels in RA patients⁹⁰, suggesting E. lenta causes microbial/metabolic alterations consequently activating autoreactive response to CII-proteins prior to clinical symptoms. RA patients do harbor CII-reactive T cells⁹⁷. However, whether an abundance of E. lenta can predict at risk individuals and disease progression in RA patients still needs to be answered.

A role of microbiota in RA was further demonstrated in a study where the fecal microbiota of early RA patients, dominated by the presence of P. copri, triggered a Th17 response and arthritis when transplanted in SKG mice ¹⁰⁹. Since P. copri did not induce arthritis in other strains of mice, pathogenicity of P. copri may be strain specific. Further, even though RA patients harbor antibodies to P. copri derived peptide Kd27, whether P. copri causes preclinical autoreactivity was unknown. To answer this, discovery based proteomics was used to identify molecular mimicry of self-proteins with P. copri ¹⁰⁰. Two autoantigens, n-acetylglucosamine-6-sulfatase and filamin A, expressed in RA patients, were shown to have homology with Prevotella Sp. In addition, Parabacteroides sp. and Butyricimonas sp. also showed homology with the autoantigens suggesting the involvement of gut commensals in autoreactivity. However, a direct role of these autoantigens in preclinical response remains to be discovered. On the other hand, a direct role of a commensal in inflammation was recently reported where mono-colonization of germ free mice with Subdoligranulum didolesgii, isolated from stool of an individual at risk of developing RA, triggered joint swelling, Th17 response and autoantibodies to CII ¹¹⁰. However, whether this commensal shares mimicry with autoantigens involved in RA or perturbs epithelial integrity is unknown.

These observations indicate that an endogenous bacteria can instigate autoantibodies to RA relevant antigens and generate preclinical autoreactivity. Thus, future studies need to define the epitope(s) that generate cross reactive responses which could provide a therapeutic target antigen for RA. The microbial origin of RA would explain the heterogeneity of RA as each individual might generate a different magnitude of response, based on the HLA genes, to the various autoantigens involved in RA. Ultimately, the idea is to determine at risk individuals, use microbes as predictors of disease progression and identify targets for therapeutic intervention.

We have to be cognizant of the fact that in humans, stool samples are the most commonly utilized specimens, serving as an analogue to the gut microbiome. However, stool samples represent a window into the colon as they are exposed to environment and do not have the anaerobic bacteria which are present in the small intestine. Thus, the basic mechanisms of pathogenicity, treatments and prevention can be answered by using appropriate mouse models and need to be explored before the jump to clinical trials in humans.

Interaction among genetic factors, smoking and microbiome as contributor to rheumatoid arthritis.

Lack of a specific pathogen in RA-susceptible HLA-DR4 positive individuals suggests interaction with other environmental factors could be the contributing factor for RA onset ⁸⁰. There is a plethora of evidence that DR4 positive smokers produce higher levels of autoantibodies and develop severe disease with extraarticular manifestations ^4,80. However, healthy smokers are also positive for ACPA¹¹¹, indicating other factors, besides autoantibodies, are involved. Smoking affects lung microbiome and there is some evidence suggesting that asymptomatic individuals at risk of RA differ in lung microbiome from healthy individuals. Smoking also alters the intestinal microbiome, ¹¹² alluding to a connection between mucosal surfaces. Indeed, RA patients show distal airway dysbiosis in lung microbiota and lack taxa such as Prevotella, Treponema and Porphyromonas¹¹³. Certain species of Prevotella and Porphyromonas can be opportunistic pathogens suggesting either lack of diversity in lungs or dysbiosis rather than specific species in lungs that contribute to immune dysregulation.

In humans, it is difficult to elucidate interactions between the HLA molecules and smoking, due to strong linkage between the DR and DQ genes and other confounding factors. Using transgenic mice expressing HLA-DRB1*0401 and DQ8 mice, in the absence of endogenous class II genes, interaction between DQ8 and cigarette smoking demonstrated an increased production of RF and ACPAs with severe CIA ⁷ as well as emphysematous changes in the lungs of arthritic mice ¹¹⁴. The novel and interesting observation was that severity was enhanced due to the presentation of citrullinated peptides by DQ8 molecules. On the other hand, DRB1*0401 transgenic mice exposed to cigarette smoke did not develop severe CIA. However, in mice expressing both DRB1*0401 and DQ8 genes, there was a strong sex-bias with histopathological and antibodies similar to human RA ^20,37,115. The hypothesis that DR4-restricted antigen presentation and autoantibodies production cause severe disease was further tested using transgenic mice expressing RA-non susceptible HLA-DRB1*0402 gene which differed from the *0401 mice in 3 AAs in the peptide binding region ¹¹¹. Both *0401 and *0402 mice, after exposure to cigarette smoke, generated an equal response to native and citrullinated Vimentin, a self-protein associated with RA, although the downstream cytokine response was different in both strains. This supports a role of HLA molecules in the generation of specific type of immune response to an autoantigen based on the thymic selection of T cells ³⁵. As citrullination of pathogenic antigens is a common strategy to elicit a strong response for effectively clearing infections, the observations imply a potential role of HLA molecules and presentation of citrullinated proteins- whether derived from a pathogen or self-protein- in RA.

Gut bacteria in senescence and RA

The microbial composition undergoes a dynamic change during aging in mice and humans ^31,32,40,50. The median age of onset of RA is 58 years and based on the premise that immune dysregulation underlies the basis of onset of RA, accelerated immunosenescence is considered one of the hallmarks of RA ^116,117. It is possible that genetically susceptible individuals lose the dynamic microbial changes during aging as suggested by the transgenic animal model ⁴⁰. The connection between senescence and gut microbiota in RA is an intense topic of interest for scientists. Both, RA and chronological aging, are associated with dysbiosis and loss of certain taxa with concomitant expansion of other taxa and alterations in microbial/metabolic profile ^{24,32,118,119}. For example, an abundance of C. aerofaciens in RA patients is associated with an increase in alpha aminoadipic acid, a marker for age-associated alterations in collagen ¹²⁰.

Studying the impact of microbial changes on human health during chronological aging poses a formidable challenge. While cross-sectional studies provide important information, there are many confounding factors that can obscure the findings. Mouse models provide a controlled environment and genetic background to understand the age-associated alterations in microbial composition and functional status specific to diseases. The observations will be crucial in providing a window into the mechanism of contribution of microbiota and explore new targets.

The available literature points to a significant impact of the interactions between microbiota and metabolites on the immune system ^81,121,122. Transplant of microbiota from young mice to aged mice altered the microbial profile and immune function restoring the lost cognitive and physical functions by reducing aging associated inflammation (inflammaging) ^123,124. On the other hand, transplant of aged gut microbiota into young GF mice caused inflammaging¹²⁵, suggesting a bidirectional relationship between gut microbiota and immunity. Humanized mice expressing RA-susceptible HLA-DRB1*0401 and arthritis-resistant DRB1*0402 mice show distinct gut microbial profile and gut immunity ^20,24,35,40. Naïve arthritis-susceptible *0401 mice lose the dynamic changes in gut microbiota and develop leaky gut as they age, while arthritis-resistant *0402 mice show age-dependent microbial alterations. These data support the concept that young arthritis susceptible mice harbor microbiomes similar to aged mice, which can cause inflammaging.

Many studies have documented microbial dysbiosis and immune alterations resulting in inflammation in aging mice ^118,125-127. Whether arthritic microbiota represents the aged microbiota is unknown and is crucial to understand the pathogenesis of RA. This was determined by using CIA-susceptible HLA-DQ8 mice. Inoculation of RA-associated E. lenta by gavage into DQ8 mice altered microbial composition with reduced beta diversity and an increase in senescence-associated secretory protein (SASP), G-CSF and CXCL5^90,128,129. In humans, reduced gut microbial diversity with an abundance of phylum Actinobacteria has been observed during aging ^118,127,130. Frailty associated microbiomes also have an abundance of E. lenta ¹³¹. The microbial changes induced by E. lenta led to a decline in AA in DQ8 mice, similar to aged individuals and RA patients, and a reduction in the Tryptophan-Kynurenine pathway and nicotinamide adenine dinucleotide (NAD) production ^{81,118,132,133}. The NAD-sirtuin (SIRT) metabolic pathway provides a connecting link between the immune and metabolic programming that controls the differentiation of T cells into T regulatory (Treg) or TH17 cells and is of great significance in a chronic disease like RA ^134-136. Boosting NAD+ bioavailability has been shown to prevent Th17 differentiation and promote resolution of inflammation via arginine biosynthesis ¹³⁷ and augmenting SIRT1 signaling axis ¹³⁸, suggesting microbial modulation could be used as a treatment to reduce inflammation. In contrast to animo acids decrease, E. lenta gavage in DQ8 mice caused an increase in secondary bile acids (BAs) including deoxycholate (DCA) and lithocholate (LCA), as well as their receptor, sphingosine 1 phosphate receptor 1 (S1PR1). Secondary BAs are associated with gut inflammation and DNA damage by activating many signaling pathways and regulating lipid metabolism ^139-142. Senescence of synovial fibroblasts is associated with an increase in BAs ¹⁴³. These microbial/metabolic alterations along with increased autoantibody production after E. lenta gavage suggest dysbiosis associated with RA mimics microbial/immune senescence.

The observations in mice expressing RA-susceptible HLA genes prove that microbial/metabolic profile dysbiosis increases gut permeability and production of SASP thus augmenting immune senescence⁹⁰, making it an attractive marker to identify at-risk individuals and for RA progression. Similarly, in individuals with RA-susceptible genetic factors, under certain conditions, intestinal dysbiosis can augment senescence-associated alterations and promote proinflammatory conditions leading to a break in tolerance and disease development.

Impact of diet on sex-bias and microbiome

Our understanding of the diverse functions ascribed to the gastrointestinal microbiota continues to evolve. The gut microbiota contributes in host physiology via production of metabolites, competing with the pathogens, producing short chain fatty acids (SCFAs) for defense and generating host dietary metabolites¹⁴⁴. A role of microbiota in sex-bias was proposed as gut microbes metabolize dietary nutrients to enhance systemic estrogen levels ¹⁴⁵.

Based on the diet, vegetarian or non-vegetarian, enterotypes have been defined ^{67 146}. The Mediterranean diet is recognized as an anti-inflammatory diet. In a study of individuals aged 65 and above adhering to a mediterranean diet, distinct sex-specific methylation patterns were observed ¹⁴⁷ , indicating a potential role of diet-based sex-specific immunity through the gut. A prospective study analyzed the association between diet and the risk of developing seropositive and seronegative RA.¹⁴⁸ A long term healthy diet which included omega-3 fatty acids, polyunsaturated fatty acids (PUFA), fruits and vegetables was associated with reduced risk of developing RA as compared to diet consisting of red meat, sweetened beverages and sodium^148,149. This difference was much more significant when stratified according to age with significantly reduced risk of seropositive RA for women <55 years on a healthy diet. This could be related to high intake of fiber in the form of vegetables as fiber requires gut commensals for fermentation leading to the production of SCFAs, important for intestinal health. High fiber intake is linked to a Prevotella enterotype and dietary fiber induced improvement in glucose metabolism has been attributed to an increase in Prevotella ¹⁵⁰. Small intestinal P. histicola metabolizes carbohydrates and prebiotics like inulin, and has anti-inflammatory properties¹⁵¹. A review of the effect of mediterranean diet, healthy oils, fruits and spices on the immune system emphasized the significant association between a healthy diet and risk of inflammatory arthritis.¹⁵². On the other hand, sweetened beverages were associated with a higher risk of RA in individuals >55 years, while moderate alcohol and lower red meat consumption were associated with a decrease in the early onset RA¹⁴⁹. Members of the Clostridiales such as Roseburia, Blautia and Coprococcus exhibit increased abundance when high sugar diets or diets rich in readily fermentable carbohydrates are consumed ¹⁵³. However, the caveat with most studies is that they did not compare sex-specific effects.

Diet requirements for both sexes can be different. The sex-bias observed in RA might also be influenced by the sexually-dimorphic effects of diet on the gut microbiota. An impact of diet-induced differences in microbiota was reported in a study where operational taxonomic units (OTUs) of Fusobacteriaceae were observed to be much more abundant in men than women¹⁵⁴. To understand the sex-bias of RA and autoimmunity, studies should consider the interplay of lifestyle, diet and environment in the context of gender. The new ordinance from the National Health of Institute (NIH) to include both sexes in research and clinical trials, along with omics data should help define the function of the gut microbiota in sex-bias of disease.

Sex-bias in microbiome, immunity and arthritis

Rheumatoid arthritis occurs 2-3 times more often and with higher disease severity in women as compared to men ^155-158. Genetic factors and sex influence RA onset, with men requiring higher genetic load to develop RA ¹⁹. Remission of symptoms during pregnancy followed by a flare after childbirth has led to the speculation that sex-hormones are one of the main culprit of sex-bias in RA. In addition, men with RA have higher levels of estradiol as compared to healthy men¹⁵⁹. The sex hormones-dependent polarization of the immune system as well as differential numbers of various immune cells can contribute to variable responses in sexes ^33,34,160. Estrogen has been shown to augment humoral response by influencing B cells, while testosterone suppresses immune response leading to low production of TNF-α ¹⁶¹. The role of estrogen in autoimmunity was supported in a study using the CIA model in DRB1*0401 mice where males with an exogenous supply of 17Eβ generated immune response and autoimmunity comparable to females, while ovariectomized female mice showed reduced autoimmunity ^20,40,97,162. Gut microbes can also metabolize sex hormones and regulate systemic levels by converting glucocorticoid into androgens ¹⁶³ and 17β reduction of androgen ⁷⁰ as well as inter-conversion of β-estradiol and estrone which can alter intestinal hormonal balance and microbial composition ^164,165. Evidence of a direct impact of sex hormones on the gut microbiome was demonstrated in twin studies, where fraternal twins of opposite sexes exhibited sex-bias in the gut microbiome after puberty ⁷⁰. Reduced levels of estrogen in postmenopausal women may contribute to microbial alterations as depicted by low intestinal SCFAs producing bacteria in menopausal compared to pre-menopausal women, contributing to RA onset ¹⁶⁶. Hormonal contraceptives in women also lead to alterations in microbial communities ¹⁶⁷. Estrogen receptors are expressed by various immune cells including macrophages and lymphocytes which can regulate mucosal immune response. Levels of sex hormones can also be controlled by the microbiota, like certain E. lenta species produce equol, a phytoestrogen with estrogenic properties. This can alter microbial composition and function which can further dictate physiological functions. Importantly, E. lenta proportions are much higher in female RA patients ⁹⁰. However, it still needs to be proven whether the gut microbe-produced sex-steroids impact immune function similar to the host’s sex hormones. In a GWAS Gardnerella vaginalis was observed with high abundance in RA patients from Japan. Abundance of Gardnerella in women with vaginitis suggests it may be involved in sexual dimorphism in RA¹⁶⁸.

There are differential alterations in immune cells during aging, higher NK cells and memory Treg cells are observed in men as compared to women ^169,170. Cellular changes coupled with sex hormones driven immune responses could explain the differences in immune responses between the sexes which can be conducive to inflammation. Sexual dimorphism in innate immunity was demonstrated in mice, where males generated a higher TLR4-driven response ^161,171. A shift in microbial composition during aging based on sex has provided novel insights into disease onset and severity^40,70. Differences in the gut microbiome might be one of the reasons for sex-bias in arthritis as shown in RA patients and humanized mice^40,81,90,172. E. lenta expansion in RA patients was associated with sex-bias and higher levels of the autoantibodies, ACPA+/RF+, in women than men ⁹⁰. In DQ8 arthritic mice gavaged with E. lenta, female mice developed severe CIA with earlier onset and higher IgG-RF and IL-17 as compared to males. Sex-bias in microbiota is also dependent on the mode of delivery.. Indeed, cesarean delivery has been associated with risk of RA¹⁷³.

One reason attributed to the increased incidence of autoimmunity in females is the expression of immune response genes, FoxP3, CD40L, IRAK1, TLR7, IL-13R, on the X chromosome ¹⁷⁴, which can regulate immune responses as well as sex hormones. Interactions among sex hormones, genetic factors and environmental factors control microbial compositions. In mice genotype and sex provide a strong determinant of gut microbial composition^36,40,175. However, observations in humans have not been consistent, which could potentially be due to various lifestyles including diet and not necessarily sex-hormones. The sex-specific differences in microbial composition were confirmed in European population focusing on the microbiota of individuals from diverse age groups and genders where men exhibited higher abundance of Prevotella and Bacteroides compared with women ¹⁷⁶. However, RA patients did not show sex-differences⁸¹.

Even though autoimmunity, including RA, occurs more often in women and a role of sex-hormones is known, the mechanism remains unclear. Whether OMICS data using host’s genome and metagenome can determine a sex-specific profile that demands sex-based treatments still remains open.

Microbiota as indicator of response to treatment

Understanding the metabolic signaling between microbes and host is essential for unraveling the mechanistic basis of their interaction^86,87. The question we need to ask is “can we balance our microbiome from within?” It is known that treatment with drugs can modulate gut microbiota ¹⁷⁷. Indeed, sulfasalazine (SSZ) used to treat RA requires metabolization by intestinal microbes for its active form ¹⁷⁸. Methotrexate (MTX) is the first line of treatment for RA and is known to suppress the immune system by inhibiting purine synthesis. Treatment with MTX partially restores gut microbiota and microbial function including redox environment and preventing pyrimidine synthesis ^21,81,82, suggesting a microbial contribution to therapeutic response. To determine whether microbiota can be used as a predictor for response to treatment, cohorts of patients treated with MTX were analyzed for fecal microbiota^83,179 at baseline and after follow-up. Two studies reported differential microbiota in MTX responders and non-responders, showing an increase in microbial diversity in patients with clinical improvement. NORA patients treated with MTX showed increased diversity and abundance of certain OTUs of Prevotella spp. in responders compared to non-responders¹⁷⁹ . At the baseline, patients with established RA showed a significant higher abundance with Prevotellaceae (family) and Coprococcus (genus), while Eubacterium sp. 3_1_31 (species) was lower in patients exhibiting clinical improvement compared to non-responders⁸³. MetaCyc pathways revealed that tetrahydrofolate and L-methionine biosynthesis, were significantly higher in responders compared to non-responders at baseline. Followup analysis of fecal microbiota after 6-12 months indicated that microbial clade, Prevotellaceae (family) was increased while Bifidobacteriaceae, and Oscillospiraceae (family) were reduced in responders vs non-responders⁸³. While comparing within each group at follow-up, responders showed an increase in Bacteroides vulgatus while non-responders showed higher abundance of Ruminococcus, Anaerotruncus, colihominis, Clostridium leptum, Coprococcus catus, and Ruminococcus sp. 5_1_39BFAA. The main difference between the two groups was sugar metabolism. A neural network model was used to determine the usefulness of the gut microbiota as a predictor of response to treatment, the top 5 features were composed of pathways involving microbial functions and included the sucrose degradation III pathway, Fatty acid & beta-oxidation II pathway and Biotin Biosynthesis I pathway. In addition, metabolization of MTX by NORA gut bacteria prior to treatment could predict clinical response to MTX; non-responders metabolized MTX ex vivo faster than responders ¹⁷⁹. Thus, Prevotellaceae was associated with response to treatment in NORA as well as established RA patients ^83,179. However, fecal Prevotella spp, P. Copri, has been linked to RA suggesting there could be differential effects of Prevotella Spp. based on the gene enrichment or on the speciation of the taxa in the intestine. As discussed below, Prevotella have niche specific roles which suggests a need to be cautious while defining a biomarker from fecal samples in humans.

Microbiome as modulator of autoimmunity

The question arises whether altering the gut microbial profile can positively impact treatment efficacy and if there is a commensal or combination of commensals that can balance the gut milieu rather than suppressing immune response as current treatments do. There are several approaches to restoring microbial eubiosis that include prebiotics, probiotics, diet and microbial modulation (Fig2).

Figure 2: Modulation of gut microbiota for suppressing inflammation.

Microbial surfaces share microbiota and dysbiosis at any mucosal surface leads to immune dysregulation. Dysbiosis can be targeted via alteration of diet, prebiotics e.g., inulin, probiotics, e.g., certain lactobacilli, and endogenous gut commensals with prebiotic like properties such as Prevotella Histicola. Gut microbiota modulation can alter other mucosal surfaces and suppress inflammation in extra-intestinal organs.

Probiotics are safe commensals that can alleviate inflammation, produce nutrients and help keep the intestinal mucosa barrier intact thus preventing colonization with pathogenic bacteria. Many scientists have theorized probiotic effects of Lactobacillus and tested its use for treating RA ^86,87. An intervention with the combination of L. rhamnosus GR-1and L. reuteri RC-14 in a randomized clinical trial led to reduced Th1 proinflammatory cytokines in RA patients. Similar observations were noted in mouse models ⁸⁶. However, the disadvantage of existing available probiotics is that they are nonspecific and have not demonstrated long term benefits.

Besides producing metabolites as byproducts of dietary metabolization, gut commensals also produce SCFAs. The use of biologically active microbial metabolites such as SCFAs is another way of protecting the mucosal barrier. Butyrate is essential for epithelial cells proliferation and repair and can modulate microbial composition. One of the dominant taxa that produces butyrate, Faecalibacterium, in a healthy human gut is reduced in RA patients ^81,86. A higher abundance of butyrate consumers observed in RA patients was linked to autoantibody production and joint deformities and a low ratio of Treg and T follicular helper cells ¹⁸⁰. In an animal model of arthritis, authors showed that butyrate supplementation reduced arthritis severity by increasing Treg cells¹⁸¹. However, reports on the use of Faecalibacterium for treating arthritis are limited, which could be due to difficulty in culturing the bacterium as well as the fact that it is a dominant commensal in humans and might have limited utility.

While most studies have focused on fecal microbes as probiotics, the role of small intestinal commensals has largely been overlooked. One of the reasons is that small intestinal commensals are difficult to isolate due to the need for intestinal endoscopy for tissue collection. However, commensals can be niche specific, based on the nutritional needs, environment and function, indicating a need to define small intestinal commensals in health. Niche specific impact of Prevotella is evident in RA. While fecal P. copri was found to be abundant in NORA patients, a novel species of P. histicola isolated from a duodenal biopsy has shown immunomodulatory properties^82,122.

Oral administration of the novel P. histicola in prophylactic and therapeutic CIA protocols, stopped the progression of arthritis in DQ8 mice¹²². A longitudinal study demonstrated that suppression of arthritis was accompanied with alterations in gut microbial composition ^96,151,182 resulting in partial restoration of the microbial profile with an increase in Allobaculum. Orally administered P. histicola colonized the duodenum signifying a specific niche. The study further defined microbial function by determining fecal SCFA levels. Treatment with P. histicola restored butyrate levels leading to an increase in tight junction proteins and an improvement in epithelial permeability. Additionally, a decrease in IL-17 with an increase in IL-10 production and differentiation of intestinal T cells into Treg cells was observed in P. histicola treated DQ8 arthritic mice. Treatment dampened inflammation systemically also as observed by an increase in splenic Treg cells with reduced cellular and humoral response in arthritic DQ8 mice as compared to control media treated mice. Characterization of P. histicola showed that it can utilize carbohydrates such as sucrose, lactose and inulin leading to production of acetate, folate and biotin¹⁵¹. Importantly, sugar metabolism has been shown to be one of the top five factors in MTX responders⁸³. Acetate can be utilized by Faecalibacterium and Allobaculum to produce butyrate, as was observed in P. histicola treated mice. Folate is an essential vitamin involved in many metabolic functions. Notably, patients on MTX treatment regimen are given folate supplement to avoid its deficiency. Further, pilot data suggest that P. histicola treatment of arthritic mice improves pain behavior in arthritic mice ¹⁸². P. histicola is a small intestinal commensal and it is possible that upper gut bacteria are in close contact with the immune system while colonic or stool bacteria have different functions. Further studies of the upper gut bacteria are essential to understand their function and impact on health. Whether these bacteria are commensal in nature and benefit one or more inflammatory diseases, or promote inflammatory disease remains to be established, however, in either case, they may be viable targets for patient care.

Conventional disease modifying anti-rheumatic drugs (DMARDS) and biologics have helped treat the symptoms of RA by suppressing the immune response. However, not all patients benefit and there are significant side effects. There is an incomplete understanding of many aspects of RA, from the initiating events to the autoantigens involved, and the mechanism of pathogenesis. Thus, there is a considerable need in acquiring new targets for generating novel and safe therapies that lead to homeostasis of the immune system with lower side effects. There is also a need to develop biomarkers that can predict individuals at risk for RA to prevent or delay the onset of disease. While this is a tall order, breakthroughs in this field may not only help patients suffering from RA, but those who are affected by other autoimmune diseases as well.

Based on the human and animal data that reveal sex-specific variations in the gut microbiome, it is conceivable that the interactions among the three axes- genotype, sex hormones, and microbiota- influence the pathophysiology and effectiveness of treatments underscoring the importance of gender-specific therapies for conditions with sex-biased manifestations. Based on the available observations on the gut microbiome and its function, one can surmise that the small intestinal axis is the sensing system in the gut that governs inflammation and immunity throughout the body.