A host-adapted auxotrophic gut symbiont induces mucosal immunodeficiency

Abstract

INTRODUCTION:

Secretory immunoglobulin A (SIgA) is a crucial component of mucosal barriers. Decreased SIgA levels are associated with increased vulnerability to infections and excessive inflammation in response to mucosal damage. The production of intestinal SIgA depends on the microbiome and specific microorganisms can drive the magnitude of the overall immune response. Conversely, mice have been observed with spontaneous reduction in intestinal SIgA levels; this phenotype has been proposed to be mediated by IgA degradation directed by gut symbiotic bacteria. However, the specific bacteria contributing to low levels of intestinal SIgA in mice remain unknown.

RATIONALE:

We developed an in vitro functional biochemical assay to screen gut bacteria from mice with low levels of SIgA. Our goal was to identify bacterial symbionts contributing to IgA degradation and to understand their relationship with the host and other components of the microbiome.

RESULTS:

We conducted a functional screen of the bacterial microbiota in wild-type (WT) mice with spontaneous low levels of intestinal SIgA. This screening led to the discovery and identification of a previously unidentified Gram-negative bacterium belonging to the Muribaculaceae family. The proposed name is Tomasiella immunophila, and it exhibits strong proteolytic activity against IgA. We found that T. immunophilla was auxotrophic for N-acetylmuramic acid (MurNAc), a critical component of bacterial cell walls, which was essential for its optimal growth in vitro. T. immunophila alone did not colonize WT conventionally raised or germ-free mice. However, fecal slurries from IgA-high mice facilitated T. immunophila colonization in WT mice, suggesting that helper strains are crucial for its successful colonization in the mouse intestine. As a result, mice colonized with T. immunophila exhibited reduced levels of SIgA in the intestine and increased susceptibility to mucosal pathogens such as Salmonella Typhimurium and Candida albicans. Furthermore, these mice also showed delayed mucosal barrier repair in response to dextran sulfate sodium-induced intestinal injury. Additionally, mucosal exposure to T. immunophila in mice induced the production of intestinal SIgA specific to this bacterium. T. immunophila secreted multiple types of immunoglobulin-degrading proteases associated with its outer membrane vesicles and these proteases were observed to specifically degrade all isotypes and subclasses of mouse antibodies; the enzymatic activity did not extend to unrelated proteins. The degradation of immunoglobulins by T. immunophila was particularly selective for rodents. Recombinant antibodies that contain the mouse kappa chain were cleaved by T. immunophila regardless of the species of the heavy chain. Notably, T. immunophila preferentially degraded antibodies harboring kappa light chains while sparing those with lambda light chains.

CONCLUSION:

This study provides evidence regarding the critical role played by the degradative capabilities of the gut microbiome in specific aspects of the mucosal immune and intestinal barrier system. Our research highlights how certain bacterial species, here T. immunophila, are particularly pivotal in this regard. The nutrient requirements of T. immunophila for MurNAc underscores its prominent role within the gut ecosystem, highlighting the intricate and complex nature of polymicrobial interactions. The challenges associated with isolating auxotrophic microorganisms further underscore the complexity of studying these organisms. Furthermore, the host species specificity of IgA degradation suggests a coevolutionary relationship between the gut microbiome and the host. These findings emphasize the important role of symbiotic bacteria such as T. immunophila in mucosal immunodeficiency, providing potential insights into related human diseases. Our study also highlights the importance of employing functional rather than descriptive techniques to identify microorganisms associated with host phenotypes or diseases.

Harnessing the microbiome to benefit human health requires an initial step in determining the identity and function of causative microorganisms that affect specific host physiological functions. We show a functional screen of the bacterial microbiota from mice with low intestinal immunoglobulin A (IgA) levels; we identified a Gram-negative bacterium, proposed as Tomasiella immunophila, that induces and degrades IgA in the mouse intestine. Mice harboring T. immunophila are susceptible to infections and show poor mucosal repair. T. immunophila is auxotrophic for the bacterial cell wall amino sugar N-acetylmuramic acid. It delivers immunoglobulin-degrading proteases into outer membrane vesicles that preferentially degrade rodent antibodies with kappa but not lambda light chains. This work indicates a role for symbionts in immunodeficiency, which might be applicable to human disease.

Graphical Abstract

Screening and functional study of immunoglobulin-degrading bacteria in the intestine. Screening of WT mice with low levels of fecal SIgA identified a previously unreported bacterium that degrades immunoglobulins. This bacterium, proposed as Tomasiella immunophila, secretes outer membrane vesicles (OMVs) containing IgA-degrading enzymes. Mice colonized with T. immunophila showed reduced IgA levels and enhanced susceptibility to Salmonella infection post vaccination. [Figure created with BioRender.com]

The ability to analyze and compare microbiomes using next-generation sequencing has facilitated prediction of associations between host and microbial components. These studies support the concept that targeting components of the microbiome can provide new therapies for a variety of diseases (1, 2). The primary limitation in the development of omics-based therapeutic approaches is the discovery and isolation of causative microorganisms that affect host phenotypes and/or disease. To this end, one challenge is the recognition and harnessing of taxa, as most gut microorganisms can only be assigned to the family or genus level; many unknown species have never been isolated in pure culture, thus impeding mechanistic studies (3). Another difficulty is that phenotypes can be driven by polymicrobial communities and those ecological interactions among microbial species are complex and dynamic, making isolation challenging (4). Microorganisms can also substantially differ in metabolic capacities, such as auxotrophs that lack essential metabolic pathways; this can further challenge the isolation of functional microorganisms on standard media during phenotypic screening (5). To manipulate the gut microbiome to improve human health, causality and mechanism-based studies are paramount. The isolation of novel taxa, along with study of their ecological interactions within microbial communities, should be integrated to identify causative microorganisms and their products responsible for complex phenotypic traits. Specifically, experimental Koch’s postulates, with some modifications, should be fulfilled in microbiome studies (6).

We apply these methods to screen for IgA-degrading microorganisms and demonstrate methods to overcome the challenges described above. One important component of the intestinal milieu is secretory IgA (SIgA), which is produced by plasma cells in the lamina propria and transported by the polymeric immunoglobulin receptor (Pigr) on the intestinal epithelium to the lumen (7, 8). Many studies have shown that SIgA has numerous important roles in intestinal health including shaping host-microbiota mutualism by regulating the composition, geographical distribution, and function of the gut microbiota (9–12). The production of intestinal IgA largely depends on specific intestinal symbionts (13, 14). Loss of intestinal SIgA can occur in both mice and humans through currently unidentified microorganisms (15, 16).

Results

Tomasiella immunophila degrades IgA in vitro

Anaerobic culture of fecal samples from IgA-low, but not IgA-high mice, completely degraded exogenous IgA, whereas bovine serum albumin (BSA) remained unchanged (Fig. 1, A and B). Screening anaerobic bacteria from IgA-low mice for IgA degradation (fig. S1A) revealed that a single colony (#8) from a specific pool (group #11) exhibited complete proteolysis of IgA (fig. S1, B and C). However, subcultures of colony #8 showed activity on only 2 of 5 anaerobic blood agar plates obtained from different sources (Fig. 1C). 16S ribosomal RNA (rRNA) gene amplicon sequencing showed that all subcultures consisted of a consortium of bacteria including Sutterella, Clostridiaceae, Allobaculum, and two related Muribaculaceae species with varying ratios in each subculture (Fig. 1D). We tested isolates of Sutterella (fig. S1D), a biomarker of IgA-low mice (15) and found they did not degrade either IgA or secretory component derived from Pigr (fig. S1, E and F). Notably, neither individual nor combinations of bacterial isolates from the original colony #8 degraded IgA (fig. S1G). We next screened each isolate and the parent colony (#8) against a panel of antibiotics (table S1 and fig. S1H) and found that none of the isolates grew on tetracycline plates. However, bacteria from the original colony cultured on tetracycline plates degraded IgA (Fig. 1E). After 7 to 14 days of culture, pinpoint colonies appeared on these plates and showed full IgA-degrading activity (Fig. 1F). 16S rRNA gene sequencing identified a previously unreported bacterial species within the family Muribaculaceae, named strain X. A strain X–specific primer set targeting the 16S rRNA gene detected this bacterium in fecal samples from IgA-low mice (Fig. 1G), subcultures of colony #8 containing IgA-degrading activity (Fig. 1C), and all antibiotic-treated cultures harboring IgA-proteolytic activity (Fig. 1E). Importantly, strain X was sensitive to the same antibiotics (table S1) that inhibited IgA degradation in the colony #8 consortia (Fig. 1E). Taken together, these results support Muribaculaceae sp. strain X as the agent of IgA-proteolytic degradation in IgA-low mice.

Fig. 1. Functional screening of IgA-low mouse feces allows identification of a previously unreported IgA-degrading bacterium.

(A) to (C) and (E) and (F) Immunoblot analysis for IgA. (A) and (F) (Bottom) Ponceau S staining for BSA. (C), (E), and (G) (Bottom) PCR for Muribaculaceae sp. strain X (XP) and total bacteria (16S). (D) Relative abundance of family/genus 16S rRNA gene amplicon sequences from bacterial composition of colony #8. (G) (Top) Fecal IgA levels from IgA-high (n = 6) and IgA-low (n = 6) mice. H, heavy chain; L, light chain of IgA; N, negative control; P, positive control.

Based on the full-length genome of strain X that we generated, the genome taxonomy database toolkit (GTDB-Tk) identified strain X as the initial isolated representative of a previously undescribed genus in the Muribaculaceae family. Phylogenomic analysis by GTDB-Tk placed strain X within a proposed genus, CAG-873, containing 76 species. We named this Gram-negative bacterium Tomasiella immunophila (Fig. 2A). The irregular morphology of T. immunophila (Fig. 2B) correlated with its genome alterations in the peptidoglycan biosynthetic pathway. Specifically, T. immunophila lacked the enzymes GlmS, MurA, and MurB (table S2), which are required for de novo biosynthesis of both N-Acetylglucosamine (GlcNAc) and N-Acetylmuramic acid (MurNAc) from fructose 6-phosphate (17). MurNAc and GlcNAc are the building blocks of peptidoglycan, a key component of the bacterial cell wall. These genetics predict that T. immunophila would be dependent on the environmental salvage of MurNAc to synthesize its cell wall, much like Tannerella forsythia, the only previously described MurNAc auxotroph (18). MurNAc, muramyl dipeptide, and peptidoglycan were confirmed to enhance T. immunophila growth in culture whereas GlcNAc had no effect. Furthermore, the addition of the MurA-specific inhibitor fosfomycin did not block the growth of T. immunophila (Fig. 2C), which further supports that this bacterium salvages but does not synthesize MurNAc. Further analysis of the T. immunophila genome revealed a gene cluster resembling the MurNAc transport and utilization pathway found in T. forsythia (Fig. 2D), which suggests how T. immunophila could harvest MurNAc from an external milieu. When cultured with exogenous MurNAc, the morphology of T. immunophila changed to a homogenous rod shape (Fig. 2B). The T. immunophila genome also encoded components of peptidoglycan turnover and recycling as well as MurNAc/GlcNAc metabolic pathway genes (table S2), suggesting mechanisms for survival as a cell wall auxotroph in the mouse intestine. These findings suggest that T. immunophila has evolved to adapt to its microbial niche in the mouse intestine by developing strategies to compensate for its cell wall auxotrophy.

Fig. 2. T. immunophila represents a previously uncharacterized genus and species within the Muribaculaceae family.

(A) Phylogenomic tree of T. immunophila, placed within the context of the family Muribaculaceae. All validly named species, as well as representatives from major metagenome-assembled genome–inferred genera from the genome taxonomy database (GTDB) (>10 genomes) are included to represent genera which have yet to be cultured. Purple circles are sized proportionally with the number of inferred species within each genus and are placed at the split in the tree for the genus or at the tip if only a single representative is included. (B) Representative Gram stain images of T. immunophila without (left) and with (right) MurNAc. Scale bars are 5 μm. (C) T. immunophila cultures were serially diluted and spotted on anaerobic blood agar plates containing the indicated additives. (D) Schematic to scale of predicted MurNAc transport and utilization gene loci in T. immunophila and T. forsythia 92A2. XthA, predicted Xanthan lyase; GT, protein with glycosyltransferase domains; DUF4922, protein with domain of function 4922; LytB, predicted amidase enhancer; AmpG, predicted muropeptide transporter; DUF1343, protein with domain of function 1343; FOXRED, FAD-dependent oxidoreductase domain-containing protein; MurT, putative MurNAc transporter; MurK, putative MurNAc kinase; MurQ, putative N-acetylmuramic acid 6-phosphate etherase. SIAE, putative sialate O-acetylesterase. Gene functions annotated from National Center for Biotechnology Information Reference Sequence Database (NCBI RefSeq) and protein-protein Basic Local Alignment Search Tool (BLASTp). (E) Representative transmission and scanning electron microscopy images of T. immunophila and its OMVs. Scale bars are 200 nm. (F) Mouse IgA incubated with T. immunophila culture, supernatant (Sup), OMVs, or OMV supernatant, followed by immunoblot analysis. N denotes IgA negative control. (G) Time course incubation of T. immunophila OMVs, either intact or disrupted by sonication or Tween 20, with IgA followed by immunoblot analysis.

Transmission and scanning electron microscopy of T. immunophila showed abundant outer membrane vesicles (OMVs), a structure secreted by Gram-negative bacteria enriched with bioactive proteins and virulence factors (19). OMVs from T. immunophila contained intact membranes ranging from 40 to 400 nm in diameter (Fig. 2E). The extracellular IgA-degrading activity of T. immunophila was associated with the OMVs, as the supernatant lost IgA proteolytic activity after OMVs were pelleted by centrifugation (Fig. 2F). As a control, OMVs prepared from Bacteroides thetaiotaomicron and Muribaculum intestinale did not exhibit proteolytic activity against IgA (fig. S2A). The association of T. immunophila IgA proteases with OMVs has important implications: OMVs are predicted to stabilize proteins and widen their distribution, which allows enzymes to reach more distant targets (20). T. immunophila may deliver its enzymes to OMVs for efficient degradation of large amounts of SIgA in the mouse intestine. Indeed, disruption of OMVs by sonication or detergent (2% Tween 20) (21) resulted in attenuated IgA cleavage (Fig. 2G and fig. S2B).

T. immunophila colonization reduces intestinal IgA

We found that cohousing IgA-low with IgA-high mice resulted in an IgA-low phenotype and detection of T. immunophila in feces of all mice (Fig. 3, A and B). This effect was reversible; treatment of IgA-low mice with vancomycin depleted the vancomycin-sensitive T. immunophila (table S1 and Fig. 3C), which restored high IgA levels and converted IgA-low to an IgA-high phenotype (Fig. 3D). Despite the transfer of T. immunophila by cohousing, it did not colonize mice under various experimental conditions, including antibiotic-pretreated wild-type (WT) C57BL/6J (B6) mice (fig. S3, A and B), germ-free mice, and gnotobiotic mice colonized with B. thetaiotaomicron and/or the five-strain consortium from colony #8 (fig. S3, C to F). Similar fecal SIgA levels were found in phosphate-buffered saline (PBS)- and T. immunophila–gavaged mice (fig. S3D). As T. immunophila is auxotrophic for MurNAc (Fig. 2, B and C), we coadministered T. immunophila and MurNAc to streptomycin-pretreated WT B6 mice. Mice fed with exogenous MurNAc were still not permissive for T. immunophila colonization (fig. S3, G and H), possibly due to insufficiency of MurNAc in the T. immunophila niche or lack of other metabolites from potential helper strains to support the growth of T. immunophila in the host intestine. We further explored whether potential helper strains facilitate T. immunophila colonization in mice. We found that T. immunophila colonized conventionally raised WT B6 mice and reduced fecal IgA levels when coadministered with an IgA-high fecal slurry. By contrast, administration of an IgA-high fecal slurry alone strongly induced intestinal SIgA production over the same time frame (Fig. 3E). Consistently, T. immunophila was only detected in mice with low fecal IgA levels (Fig. 3F), suggesting that components of the IgA-high fecal microbiome could promote T. immunophila colonization in mice. Taken together, these results support a model in which the transmissible and dominant IgA-low phenotype is T. immunophila-dependent and requires factors from helper strains for its colonization in mice.

Fig. 3. T. immunophila–mediated IgA degradation renders mice more susceptible to infection.

(A) and (B) Fecal IgA levels (A) and PCR detection of T. immunophila (Ti) and total fecal bacteria (16S) (B) in cohoused IgA-low (n = 4) and IgA-high mice (n = 6). (C) and (D) PCR detection of Ti and 16S (C) and fecal IgA levels (D) pre- and post-vancomycin treatment. (E) and (F) Fecal IgA levels (E) in mice gavaged with IgA-high fecal slurry, either without (control) or with Ti; PCR detection of Ti and 16S (F) pre- and postgavage. (G) PCR detection of Ti and total 16S along the intestinal tract of IgA-low mice. (H) FISH for localization of Ti in intestinal sections of IgA-low mice; scale bars are 15 μm. (I) IgA levels of luminal contents along the intestinal tract in IgA-high and IgA-low mice. (J) Workflow for Salmonella vaccination and infection. (K) Fecal IgA levels under specified conditions. (L) Survival curve of Salmonella vaccination/infection mouse model. (M) Colony-forming unit assay of fecal Salmonella in mice after vaccination. (A), (D), (E), (I), (K), (L), and (M). Error bars represent mean ± SEM and each dot represents an individual mouse. Statistical tests: one-way analysis of variance (ANOVA) with Tukey’s test (A), Mann-Whitney U-test (D) and (I), two-way ANOVA with Tukey’s test (E) and (K), log-rank test (L), and mixed-effects model with Tukey’s test (M). Statistical significance denoted as not significant (ns), **P < 0.01, ***P < 0.001, ****P < 0.0001.

Next, we explored the biogeographic distribution of T. immunophila in IgA-low mice and found that T. immunophila was present in both the small and large intestines but was more abundant in the cecum and colon by polymerase chain reaction (PCR) (Fig. 3G) and fluorescence in situ hybridization (FISH) experiments (Fig. 3H); IgA-high mice served as negative controls (fig. S4). A survey of luminal SIgA levels along the intestine in both IgA-high and IgA-low mice revealed that IgA levels were comparable in the small intestines of both groups; however, IgA-high mice exhibited higher luminal IgA levels in the cecum and colon compared with IgA-low mice (Fig. 3I). These findings indicate an inverse correlation between the abundance of T. immunophila in cecum and colon of IgA-low mice and SIgA levels. Lastly, we validated the abundance of T. immunophila in the gut of IgA-low mice by 16S rRNA gene-targeted quantitative realtime PCR. Analysis of fecal material from IgA-low mice showed that DNA copies of T. immunophila were about 4×10⁸/g feces (fig. S5).

T. immunophila colonization enhances intestinal infection and damage

Nontyphoidal Salmonella infection induces protective intestinal IgA that restricts enteric Salmonella growth and mucosal access (22). Oral vaccination of mice with peracetic acid–killed Salmonella Typhimurium induces high-avidity intestinal IgA and provides protection by enchaining growing Salmonella cells (23); we validated this model in our facility (Fig. 3,J to L). Vaccinated mice treated with the IgA-high fecal slurry and T. immunophila showed reduced protection to subsequent Salmonella infection compared with those treated with the IgA-high fecal slurry alone (Fig. 3M). SIgA is also induced and binds to Candida albicans, reducing its intestinal abundance and promoting commensalism (24, 25). Mice colonized with the IgA-high fecal slurry cleared C. albicans more rapidly than those colonized with the IgA-high fecal slurry and T. immunophila (fig. S6).

Pigr^−/− mice and IgA-low mice are more susceptible to colonic damage by dextran sodium sulfate (DSS) (15, 26). After 7 days of 2.5% DSS treatment followed by a 10-day washout period of no DSS, WT B6 mice inoculated with the IgA-high fecal slurry plus T. immunophila lost a greater percentage of body weight and exhibited less weight recovery during the recovery period compared with control mice inoculated with the IgA-high fecal slurry alone (fig. S7A). Histologic analysis confirmed that T. immunophila colonization delayed the repair from DSS-injury (fig. S7, B and C), similar to IgA-low and Pigr^−/− mice (15). These data demonstrate that T. immunophila-mediated SIgA depletion elicits similar phenotypes to those previously captured by the bacterial consortium of IgA-low mice, thus fulfilling experimental Koch’s postulates that T. immunophila is a causative microorganism for the low SIgA phenotypes in mice.

T. immunophila induces SIgA to itself

T cell–dependent SIgA combats mucosal pathogens (27) which in turn evade this defense by secreting proteases capable of degrading a variety of proteins including IgA, thereby enhancing mucosal colonization (e.g., Neisseria meningitidis and Streptococcus pneumoniae) (28). We tested whether T. immunophila fits this paradigm. Oral administration of T. immunophila to WT B6 mice elevated fecal SIgA levels with high affinity for T. immunophila compared with controls (Fig. 4, A to C, and fig. S8A) and high specificity for T. immunophila as demonstrated by minimal cross-reactivity to bacteria within the same family (Muribaculaceae) or several unrelated families (Fig. 4D and fig. S8, B and C). These results demonstrate that T. immunophila effectively induces production of SIgA that specifically targets itself. Salmonella and Candida-induced SIgA reduce the burden of these pathogens through direct binding (22–25). We tested whether SIgA alters T. immunophila colonization in mice using host genetic loss-of-function strains. Colonization of T. immunophila was reduced in Rag1^−/− [no T and B lymphocytes and Pigr ^−/− (no luminal IgA and immunoglobulin M (IgM)] mice compared with WT mice (Fig. 4, E and F). These results suggest that IgA can modulate colonization but is not required for T. immunophila colonization in mice. The microbiome of these mouse strains may also play a role in facilitating T. immunophila colonization. This finetuning mechanism of T. immunophila–mediated IgA induction and degradation suggests that release of amino acids from degraded IgA into the mouse intestine may mediate this effect. Although total amino acid levels in the cecum and colon were similar between IgA-high and IgA-low mice (fig. S9), further investigation is needed to explore how IgA degradation products interact with the gut microbiome.

Fig. 4. Administration of T. immunophila induces the production of intestinal IgA specific to itself.

(A) Fecal IgA levels in mice gavaged with T. immunophila (Ti) compared with PBS controls. (B) and (C) Heat-killed Ti and (D) several bacterial species (Bt, Bacteroides thetaiotaomicron; Bf, Bacteroides fragilis; Ef, Enterococcus faecalis; Mi, Muribaculum intestinale; and two Muribaculaceae species isolates (Ms1 and Ms2) from IgA-low mice (table S1) were incubated with monoclonal mouse IgA (Mono-IgA) or IgA from fecal pellets of PBS- or Ti-administered mice (in triplicate). Samples were analyzed by flow cytometry with anti-IgA-PE staining. (E) and (F) Realtime PCR determination of Ti copy number in fecal samples from WT or Rag1^−/− (E) and WT or Pigr^−/− mice (F) gavaged with IgA-low fecal slurry (E) or IgA-high fecal slurry plus Ti (F). Error bars show mean ± SEM (A), (C), (E), and (F); each dot represents an individual mouse (A), (E), and (F), Statistical tests: two-way ANOVA with Tukey’s test (A) and (C) or uncorrected Fisher LSD test (E), and Mann-Whitney U-test (F). Significance denoted as not significant (ns), *P < 0.05, ****P < 0.0001.

T. immunophila preferentially degrades kappa light chains

Intestinal microbes can induce other types of immunoglobulins (29). T. immunophila degraded all mouse immunoglobulin isotypes to varying degrees. IgA, IgE, and IgM were completely degraded while all four mouse IgG subclasses retained proteolytically resistant fragments (Fig. 5A). N-terminal sequencing of these fragments revealed conserved terminal cleavage sites following positively charged amino acids [two lysines (KK) in IgG1, IgG2a, and IgG2b; lysine and arginine (KR) in IgG3] near the hinge region (Fig. 5B). Tagged fusion proteins harboring WT or mutated sequences (K95A/K96A) from IgG1 (Fig. 5B) were cleaved by T. immunophila (fig. S10). However, although the cleavage site in the WT fusion protein was identical to that of IgG1, the mutated fusion protein was cleaved at a new upstream lysine residue (K92, Fig. 5B). Similar results were observed when full-length recombinant mouse IgG1 antibodies (WT and two mutants) were incubated with T. immunophila (Fig. 5C). These results suggest that multiple cleavage sites are present on the Fab fragments of IgG antibodies. Subsequent experiments using full-length (FL), a fragment with two antigen-binding portions (F(ab′)2), and the fragment crystallizable region (Fc) forms of IgG confirmed that only the Fab fragments, not the Fc, were cleaved by T. immunophila (Fig. 5D).

Fig. 5. T. immunophila preferentially degrades mouse immunoglobulins with kappa light chains.

(A) Mouse antibody isotypes were incubated with T. immunophila (Ti) and analyzed by stain-free SDS-PAGE. The N-terminal sequences of IgG cleavage fragments (black arrows) were determined by Edman degradation. (B) Schematic representation of the mouse IgG heavy chain constant region (CH), showing the Ti-mediated cleavage site (magenta triangle). Partial amino acid sequences depicting Ti-mediated cleavage sites (magenta triangle) in the WT antibodies, with an additional cleavage site at K92 (cyan triangle) in the mouse IgG1 mutant (K95A/K96A). Numbers denote the positions of amino acid residues in the constant region of mouse IgG1. (C) Recombinant mouse IgG1 antibodies (WT, K95A/K96A, and K92A/K95A/K96A mutants) were incubated with Ti and analyzed by immunoblotting. (D) FL, F(ab′)2, and Fc fragments of mouse IgG were incubated with Ti and analyzed by stain-free SDS-PAGE. (E) Mouse IgG1, IgG2a, IgG2b, and IgG3 with κ or λ light chains were incubated with Ti and analyzed by stain-free SDS-PAGE. (F) Recombinant mouse IgG1 antibodies with κ, λ1, or λ2 light chains were incubated with Ti and analyzed by stain-free SDS-PAGE. (G) Mouse IgM with κ or λ light chains was incubated with Ti and analyzed by stain-free SDS-PAGE. (H) Mouse serum incubated with Ti and analyzed by immunoblotting for IgG, IgA, IgM, light chains (κ, λ), and albumin. (I) Fecal slurry from IgA-high mice incubated with Ti cultures or OMVs under different culture conditions. The levels of proteins were determined by immunoblotting. (A, C, E, F, and G), H, heavy; L, light chain IgA. (A), (D), (E), and (F). Black arrows denote cleaved fragments. All SDS-PAGE and immunoblots shown are representative of two to three independent experiments.

Digestion of IgG by T. immunophila revealed multiple cleavage sites, but the initial cleavage site remained unclear. Specifically, the light chains of all tested antibodies were fully cleaved (Fig. 5A). Nevertheless, we found that T. immunophila targeted immunoglobulins with kappa (κ) rather than lambda (λ) light chains, including all four subclasses of mouse IgG (Fig. 5E). To control for variable region sequences, we engineered three recombinant mouse IgG1 antibodies containing identical variable region sequences, and these antibodies differed only in the light chain constant region (κ, λ1 or λ2). Again, only the recombinant mouse IgG1 with a κ light chain was completely digested by T. immunophila. By contrast, the recombinant antibodies with either a λ1 or λ2 light chain remained resistant to cleavage (Fig. 5F). Additionally, IgM with λ light chains also showed resistance to cleavage (Fig. 5G). These data support the conclusion that T. immunophila initiates cleavage at the light chain.

To further test substrate specificity, we performed an in vitro assay using a more complex substrate system in which mouse serum was incubated overnight with T. immunophila. Immunoblot analysis confirmed efficient cleavage of IgG, IgA, and IgM, predominantly targeting κ light chain antibodies in serum. Notably, λ light chains mostly remained intact. Albumin, as a control, showed no digestion by T. immunophila (Fig. 5H). In addition, the IgA-high fecal slurry was used as a substrate for T. immunophila. Because T. immunophila is auxotrophic for MurNAc, we cultured it with or without MurNAc and purified OMVs under identical culture conditions. In vitro assays detected comparable IgA proteolytic activity in both T. immunophila cultures with and without MurNAc supplementation (fig. S11A). Fecal microorganisms in the IgA-high fecal slurry did not promote T. immunophila–mediated IgA degradation (fig. S11B), despite their ability to aid in the colonization of T. immunophila in mice (Fig. 3F). When T. immunophila cultures and OMVs were incubated with the IgA-high fecal slurry, IgA containing κ light chains were completely degraded, whereas λ light chains were only slightly degraded. Other fecal proteins, including the secretory component, Trypsin 2, and Mucin 2 remained intact (Fig. 5I). Similarly, when purified fecal SIgA was used as a substrate, T. immunophila and its OMVs preferentially degraded IgA with κ light chains, leaving λ light chains, the secretory component, and J chain largely intact (fig. S12, A and B). Furthermore, oral administration of OMVs to WT B6 mice resulted in specific targeting of IgA with κ light chains; λ light chains and secretory component remaining unaffected (fig. S12C). Taken together, these results support a model in which T. immunophila preferentially targets the κ light chain over the λ light chain antibodies. Notably, when mouse IgGs were incubated with T. immunophila in the presence of the reducing reagent of Tris (2-carboxyethyl) phosphine, further cleavages were observed in IgG with κ light chains, and some IgG with λ light chains were also partially cleaved by T. immunophila (fig. S13). These results suggest that both amino acid sequence and three dimensional (3D) structure determine whether an antibody can be cleaved by T. immunophila.

To investigate T. immunophila’s enzymatic properties in cleaving mouse immunoglobulins, we tested various protease inhibitors using IgG1 as the substrate. N-ethylmaleimide (NEM), a cysteine protease inhibitor, completely blocked the IgG-cleaving activity of T. immunophila whereas the metal-chelating reagents ethylenediaminetetraacetic acid (EDTA) and ethylene glycol-bis(2-aminoethylether)-N,N,N′, N′-tetraacetic acid (EGTA) partially blocked cleavage (fig. S14, A and B). NEM, EDTA, or EGTA also completely blocked T. immunophila-mediated degradation of IgA (fig. S14B). Notably, when fusion proteins with the fragments from IgG2a, IgG2b, and IgG3 constant regions were used as the substrates, both EDTA and EGTA completely blocked cleavage whereas NEM did not (fig. S14C). These findings support a model in which multiples enzymes are involved in the cleavage of mouse antibodies. The enzyme which initiated the cleavage from the N terminus of the substrates appears to be an NEM-sensitive cysteine protease as NEM completely blocked T. immunophila-mediated IgG/IgA cleavage (fig. S14, A and B). Other enzyme(s) involved in the subsequent cleavage events require cationic cofactors (e.g., metal binding proteases or metalloproteases) to complete the catalytic process. This mode of action may suggest that IgA, IgE, and IgM were fully cleaved because multiple enzymes acted together to cut at various sites on these substrates. By contrast, IgGs might lack these cleavage sites on their Fc fragments (Fig. 5, A and D).

A kinetic study of IgA and IgG degradation by T. immunophila revealed simultaneous digestion of the heavy and light chains of IgG, leaving a detectable Fc fragment. No intermediate bands were observed with either stain-free SDS-PAGE or immunoblot analysis. The cleavage rate was higher under reducing conditions compared to non-reducing conditions (fig. S15, A to D), indicating that the cleavage sites were more accessible to T. immunophila-derived enzymes in reduced conditions. We observed an inverse relationship between reaction temperature and substrate degradation rate. Additionally, no intermediate bands were observed at lower temperatures (fig. S15, E and F). We next confirmed that OMVs from T. immunophila can cleave all isotypes of mouse immunoglobulins, similar to the T. immunophila culture (fig. S16). IgA and IgG were completely degraded by equal amounts of OMVs purified from cultures with or without MurNAc (fig. S17, A and B). Subsequently, similar amounts of OMVs were separated by SDS-PAGE and subjected to mass spectrometry analysis for protein identification (fig. S17C and data S1). Comparative genomic analysis of T. immunophila and M. intestinale, a close relative of T. immunophila without IgA-degrading activity (fig. S17D), identified 23 secreted candidate proteases encoded only by T. immunophila, of which 11 proteins were identified from mass spectrometry analysis of T. immunophila OMVs (data S2). However, none of the 23 candidate proteases were able to cleave IgA or IgG either alone or in combination (fig. S18).

Antibody light chains determine cleavage by T. immunophila

A survey of immunoglobulins from other host species revealed that only closely related species (rat, hamster, and guinea pig) were susceptible to T. immunophila–mediated antibody degradation (Fig. 6A). In addition, subclasses of rat IgG showed varying cleavage efficiencies by T. immunophila (fig. S19A). T. immunophila also completely degraded IgAs from various mouse strains and rats (fig. S19B). Similar to mouse antibodies, hamster IgGs with κ (but not λ) light chains were targeted by T. immunophila (Fig. 6B). All tested human antibodies of different isotypes and subclasses were resistant to cleavage (Fig. 6C). As our earlier findings indicated that the light chain determines antibody cleavage by T. immunophila, we engineered and purified recombinant mouse-human chimeric IgG antibodies, containing either a mouse or human κ light chain. We found that T. immunophila cleaved chimeric antibodies with mouse κ light chains regardless of the origin of the heavy chains. By contrast, chimeric IgG antibodies harboring human κ light chains were resistant to cleavage (Fig. 6, D and E). Sequence alignment of human and mouse IgG heavy chains revealed that the cleavage site in mouse IgGs is also present in human IgGs (fig. S20A). In vitro reactions showed that T. immunophila degraded the human IgG heavy chains to varying degrees under reducing conditions, whereas the light chains remained intact (fig. S20, B and C). These results suggest that T. immunophila initiated antibody cleavage at the light chain, which may vary in sequence among host species. This may explain why T. immunophila degrades most rodent antibodies but not antibodies from other phylogenetically distant species tested.

Fig. 6. Light chains determine species-specific immunoglobulin cleavage by T. immunophila.

(A) IgG/Y from various species were incubated with T. immunophila (Ti) and analyzed by stain-free SDS-PAGE. (B) Hamster IgG1 and IgG2 antibodies with κ or λ1 light chains were incubated with Ti and analyzed by stain-free SDS-PAGE. (C) Human antibody isotypes and subclasses were incubated with Ti and analyzed by stain-free SDS-PAGE. Mouse IgG1 (mIgG1) is a positive control. (D and E) Recombinant chimeric IgG antibodies, including mouse heavy chain with mouse (m-κ) or human κ light (h-κ) chains and human heavy chain with human or mouse κ light chains, were incubated with Ti and analyzed by immunoblotting. H, heavy; L, light chains of immunoglobulins; C, cleaved fragments. (A), (B), and (C) Black arrows indicate cleaved fragments. All SDS-PAGE and immunoblots shown are representative of two to three independent experiments.

In summary, our study supports the idea that T. immunophila cleaves immunoglobulins with κ light chains from mice (or other phylogenetically related species) at multiple sites simultaneously, resulting in undetectable small fragments (fig. S21, A and C). Mouse immunoglobulins with λ light chains were resistant to cleavage, suggesting that λ light chains protect heavy chains from enzymatic digestion by T. immunophila (fig. S21, B and D).

Discussion

Mouse models are commonly used to investigate the impact of gut microbiome on human diseases, despite their limited translatability to humans (30). Our findings demonstrate that T. immunophila exclusively degrades antibodies from rodents, suggesting that IgA-degrading microorganisms have coevolved with their hosts and are thus distinct in taxonomically or phylogenetically divergent hosts. Similar taxa may vary in presence, potentially influencing phenotypic diversity in animal research. Screening for IgA-degrading strains in other hosts will require in vitro studies. This principle could extend to other studies involving microbiome dependent host traits. Our study underscores the complex interplay between microbiome and hosts, as well as interactions among microbial species.

This study also provides a roadmap for isolation of challenging microorganisms that confer important phenotypes. We showed that T. immunophila was co-isolated with five other species, suggesting that polymicrobial interactions may facilitate isolation of hard-to-culture microorganisms. We suggest that microbial screens are best designed if they are permissive for multipartite interactions. Furthermore, antibiotics proved effective in obtaining pure cultures of the causative microorganisms. It is noteworthy that our initial sequencing/abundance-based approach identified Sutterella as a biomarker for IgA-low mice (15), which in practice was merely a bystander. There is a growing need to prioritize functional insights over taxonomic distinctions. These isolates with proper taxonomic identification will serve as valuable benchmarks for future studies.

Our study shows the challenge of fulfilling experimental Koch’s postulates for a given phenotype (here, SIgA loss in the colon). Discovering T. immunophila in a screen was challenging as a result of its auxotrophy for MurNAc. To isolate causative microorganisms that confer a phenotype or human disease, the metabolic requirements of different microorganisms need to be taken into consideration. Optimization of growth media with nutritional additives could help isolate hard-to-grow or even “uncultivable” microorganisms, many of which may be auxotrophs. These approaches will add to existing collections of genome-sequenced isolates of human (31) and mouse (32, 33) intestinal bacteria.

This study highlights the potential need for helper strains to successfully colonize mice. Such polymicrobial interactions are widespread in nature (34, 35). Though mechanisms are unclear, studies show that microbial consortia contribute to host phenotypes (36, 37). Our study supports the concept that gut microorganisms might be metabolically interdependent. To this point, an Enterococci pathobiont provides fermentable amino acids to increase C. difficile fitness (and thus pathogenesis) in the antibiotic-perturbed gut (38). In our study, a microbial community including T. immunophila and other yet-to-be-identified symbionts induce a stable IgA-low phenotype in mice. Because the luminal amino acid levels are similar in both IgA-low and IgA-high mice, it is unclear whether T. immunophila can locally increase amino acid levels by degrading SIgA and thereby nurture neighboring microbial communities, which could provide MurNAc and other metabolites for T. immunophila. Degradation of IgA can also release glycans from SIgA, and glycan metabolism can shape the composition of host microbiota (39). T. immunophila and its helper strains could serve as an ideal model system to study the complexity and dynamics of polymicrobial metabolic interactions. A recent study established hundreds of in vitro microbial communities cultured from a variety human stools, which may serve as a powerful system for microbiota research (40).

Gut microorganisms have coevolved with a variety of hosts, which has led to a mutualistic symbiosis (41). How this mutualism is maintained is poorly understood, but our study provides a potential explanation. T. immunophila preferentially degrades mouse immunoglobulins with κ light chain that represent 95% of antibodies in this host (42), suggesting that T. immunophila coevolved with the host to shape this host phenotypic trait. Discovering the enzymes for antibody degradation and defining the mechanisms by which the enzymes specifically target κ light chain harboring antibodies will provide knowledge and tools for antibody degradation and engineering. Our data suggests that multiple types of enzymes may be involved in substrate degradation and that these enzymes are delivered into OMVs; this mechanism ensures that T. immunophila can efficiently break down large amounts of SIgA in the mouse intestine. Through comparative genomic analysis between T. immunophila and M. intestinale, we obtained 23 protease candidates specific to T. immunophila. Escherichia coli and HEK 293T cell lysates harboring overexpressed protease candidates failed to identify key players in degrading mouse IgA or IgG. Thus, methods need to be developed for the identification of the OMV-bound enzymes that degrade immunoglobulins. Reported or commercial immunoglobulin proteases typically cleave antibodies into F(ab′)2 and Fc fragments (43–46), By contrast, the enzymes of T. immunophila specifically digest IgGs into Fc fragments, completely abolishing antibody functions. Identifying these enzymes could offer alternative tools for antibody engineering.

Selective IgA deficiency (SIgAD) is the most common primary immunodeficiency and is characterized by greatly reduced serum IgA (47); variable levels of fecal IgA were observed in patients with SIgAD (16). Notably, secretory IgM was found to partially compensate for reduced IgA at mucosal surfaces in these patients (16, 48, 49), which was supported by findings using IgA knockout mice (50). By contrast, the digestion of both IgA and IgM by T. immunophila promotes a more complete state of mucosal adaptive immunodeficiency. We propose that the future identification of human gut microorganisms that degrade both secretory IgA and IgM will be of high importance in regards to their relationship with disease.

Methods summary

Animal care and experimentation were consistent with National Institutes of Health (NIH) guidelines and approved by the Institutional Animal Care and Use Committee at Washington University School of Medicine (IACUC protocols 20180221 and 22-1040) and the Cleveland Clinic (IACUC protocol 2281). To screen for IgA-degrading bacteria from SIgA-low mice (15), fecal samples were cultured anaerobically and plated on blood agar. Pools of bacterial colonies were incubated with mouse monoclonal IgA and subjected to sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblot analysis. Pooled bacterial colonies showing IgA-degrading activity were tested individually for cleavage activity. A consortium of bacteria consisting of five species with IgA-cleavage capability was identified. This consortium was cultured with various antibiotics, and subsequent incubation with IgA for the isolation of a pure colony with IgA-degrading activity. 16S rRNA gene sequencing revealed this IgA-degrading bacterium as a previously unreported species. Using GTDB-Tk, this bacterium was classified as the initial isolated representative of a previously undescribed genus within the Muribaculaceae family. Protologger (v1.3) was used to describe this novel taxon (51), proposed to be named Tomasiella immunophila. T. immunophila and its OMVs were visualized using transmission and scanning electron microscopy. WT B6 mice were gavaged with IgA-high fecal slurry ± T. immunophila to test in vivo IgA-degrading activity. We performed mouse models of Salmonella Typhimurium vaccination/infection (23), C. albicans infection (24, 25), and the DSS-induced colitis (15) to test the biological role of T. immunophila IgA degradation. WT B6 mice were orally administered T. immunophila to evaluate its capability to induce IgA production. Rag1^−/− (no T/B cells) and Pigr^−/− (no luminal IgA/IgM) mice were colonized with T. immunophila to investigate the dependency of T. immunophila colonization on IgA in mice. All mouse antibody isotypes and subclasses were incubated with T. immunophila and its OMVs in vitro to evaluate their susceptibility to degradation. The terminal cleavage sites of mouse IgG antibodies by T. immunophila were identified using N-terminal sequencing. All mouse antibody isotypes and subclasses, including those with κ or λ light chains, were incubated with T. immunophila and its OMVs to assess susceptibility to degradation. Recombinant mouse IgG1 antibodies with κ or λ light chains, mouse serum, and mouse fecal slurry were utilized to investigate the specificity of T. immunophila-mediated antibody cleavage. Additionally, antibodies from various species, including humans, were incubated with T. immunophila. Furthermore, recombinant chimeric IgG antibodies, consisting of mouse heavy chains with mouse (m-κ) or human κ light chains (h-κ), and human heavy chains with human or mouse κ light chains, were also incubated with T. immunophila to validate the light chain-dependent antibody cleavage.

Data and materials availability:

The source data used for analysis in this study are available at Dryad (52, 53).

The GenBank accession number for the assembled T. immunophila genome is OX620685. The ATCC accession number for T. immunophila is TSD-438.