Human Microbiota Flagellins Drive Adaptive Immune Responses in Crohn’s Disease

Abstract

BACKGROUND AND AIMS:

Crohn’s disease and ulcerative colitis are characterized by dysregulated adaptive immune responses to the microbiota in genetically susceptible individuals, but the specificity of these responses remains largely undefined. Therefore, we developed a microbiota antigen microarray to characterize microbial antibody reactivity, particularly to human-derived microbiota flagellins, in inflammatory bowel disease.

METHODS:

Sera from healthy volunteers (n 87) at the University of Alabama at Birmingham and from patients recruited from the Kirklin Clinic of University of Alabama at Birmingham Hospital, including patients with Crohn’s disease (n 152) and ulcerative colitis (n 170), were individually probed against microbiota bacterial flagellins of both mouse and human origin and analyzed for IgG and IgA antibody responses. Circulating flagellin-reactive T effector (CD4⁺CD154⁺) and T regulatory (CD4⁺CD137⁺) cells were isolated and evaluated in selected patients. Resulting adaptive immune responses were compared with corresponding clinical data to determine relevancy to disease behavior.

RESULTS:

We show that patients with IBD express selective patterns of antibody reactivity to microbiota flagellins. Patients with Crohn’s disease, but not patients with ulcerative colitis, display augmented serum IgG to human ileal-localized Lachnospiraceae flagellins, with a subset of patients having high responses to more than 10 flagellins. Elevated responses to CBir1, a mouse Lachnospiraceae flagellin used clinically to diagnose CD, correlated with multi-Lachnospiraceae flagellin reactivity. In this subset of patients with CD, multi-flagellin reactivity was associated with elevated flagellin-specific CD154⁺CD45RA⁻ T memory cells, a reduced ratio of flagellin-reactive CD4⁺ T regulatory to T effector cells, and a high frequency of disease complications.

CONCLUSIONS:

Patients with Crohn’s disease display strong adaptive immune response to human-derived Lachnospiraceae flagellins, which may be targeted for prognosis and future personalized therapies.

Graphical Abstract

Regulation of the host response to the intestinal microbiota is a primary function of the mucosal immune system. Multiple mechanisms, including the secretion of antimicrobial peptides and the production of IgA and mucus, serve to confine immune responses to the microbiota within the gastrointestinal tract, thereby preventing systemic immune activation to microbiota constituents.^1–6 Dysregulated responses to the normal microbiota can lead to inflammation and are associated with numerous immune-mediated inflammatory disorders.⁷ This includes the inflammatory bowel diseases Crohn’s disease (CD) and ulcerative colitis (UC), which are generally accepted to occur as the result of dysregulated adaptive immune responses to the microbiota in genetically susceptible individuals.^8–10 Despite overlap of a large number of genetic polymorphisms, ¹¹ UC and CD have distinct seroreactivities to microbiota antigens, with patients with UC specifically reacting to a microbial antigen cross-reactive to perinuclear antineutrophil cytoplasmic antibodies¹² and patients with CD responding to several antigens, including Saccharomyces cerevisiae carbohydrate, Escherichia coli outer membrane porin C, and to microbiota flagellins.⁸ Flagellin proteins and glycoproteins are the building blocks of the bacterial flagellum, which is comprised of 50,000–80,000 flagellin monomers.¹³ Flagellin monomers have a characteristic structure of conserved amino and carboxy domains, which polymerize internally in the flagellum, separated by a highly diverse, exterior-facing intermediate domain.¹⁴ The flagellin operon has chemosensing genes that, in concert with the motor apparatus of the flagellum, allow bacteria to move into favorable microenvironments, such as the intestinal mucus layer. Flagellin is a potent immune activator and antigen and is the only known microbial protein that has 3 receptors—Toll-like receptor 5, nucleotide-binding oligomerization domain-like receptor family CARD domain-containing protein 4, and LY6/PLAUR domain containing 8—encoded in the host genome, in addition to immunoglobulin and T cell receptors. Only a subset of bacteria produce flagellins, and most are in the Firmicute and Proteobacteria phyla. Flagella are present on certain pathogens, such as Salmonella, where they serve not only in motility but also as a major pathogenic constituent.¹⁵

Microbiota flagellins have been defined previously as immunodominant antigens in CD.^16–18 Antibodies to CBir1, a flagellin expressed by Lachnospiraceae bacteria isolated from the murine cecum, correlate with complications in adults with CD,^17,19 as well as development of strictures and fistulas in pediatric patients newly diagnosed with CD.^10,12 These defined seroreactivities may be classified as specific biomarkers, and likely participants, in the immunopathogenesis of CD.

In this study, we focused on characterizing the adaptive immunoglobulin response to human-derived microbiota flagellins in CD and UC. We utilized a novel microbiota antigen microarray directed at recombinant bacterial flagellins. Although our laboratory and others have shown elevated responses to mouse Lachnospiraceae flagellins in CD,^17–19 this is the first study to describe IgG antibodies to human-derived Lachnospiraceae flagellins. We have identified 4 Roseburia and 1 Eubacterium species with flagellins homologous to CBir1 and related mouse flagellins. Elevated IgG response to human-derived Lachnospiraceae family flagellins is not evident in UC or healthy controls (HCs), indicating a specific adaptive immune response in patients with CD. A multi-flagellin reactive subset of patients with CD was identified, which exhibited an abnormal CD4⁺ T cell response with significantly elevated flagellin-specific CD4⁺CD154⁺CD45RA⁻ T effector memory (TEM) cells with a reduced regulatory T cell (Treg) to effector T cell (TEFF) ratio. Thus, this study identifies a subset of patients with CD with abnormal B cell and T cell reactivity to multiple human and mouse-derived Lachnospiraceae flagellins, which associates with a complicated disease course.

Materials and Methods

Patient Sample Collection

After informed consent was obtained, peripheral blood samples were collected from patients with UC or CD at the Kirklin Clinic of the University of Alabama at Birmingham Hospital in Birmingham, Alabama, or healthy in-house volunteers in accordance with the University of Alabama at Birmingham’s Institutional Review Board guidelines (IRB X300001155, X140515004, and F081003001) and the Declaration of Helsinki Principles. A database was constructed of the clinical metadata of the patients, including but not limited to surgery, strictures, fistulas, and biologic therapy.

Protein Microarray and Detection

Bacterial proteins and extracts were arrayed in quadruplicate on 16 pad nitrocellulose slides (Maine Manufacturing) at a concentration of 0.2 mg/mL using a Spotbot Personal Microarrayer (ArrayIt). Recombinant proteins were diluted from stocks to 10 mM Tris (pH 7.4), 20% glycerol, and 0.1% sodium dodecyl sulfate. To probe the microarray, pads were blocked for 1 hour in SuperBlock (ThermoFisher) and then probed with patient sera diluted 1 to 100 in SuperBlock for 1 hour. The pads were washed 3 times in phosphate-buffered saline with 0.05% Tween 20; then, a mixture of anti-IgA (Immunoreagents, Inc) and anti-IgG (Invitrogen) secondary antibodies labeled with Dylight 550 and Dylight 650, respectively, were applied at 1 to 1000 (approximately 0.5 μg/mL) for 1 hour in SuperBlock. The pads were then washed 3 times with phosphate-buffered saline with Tween, air-dried, and scanned using a GenePix 4000B imager (Axion).

In Silico Identification of Lachnospiraceae Bacteria in Terminal Ileum in Patients With Inflammatory Bowel Disease and Healthy Controls

Three hundred and seventy-four experiments (patient samples) representing study SRP002479, “Effect of Crohn’s Disease Risk Alleles on Enteric Microbiota,”²⁰ were obtained from the Sequence Read Archive. The resulting database was queried using BlastN at the command line on the UAB Cheaha supercomputer. Sequences having an identity of ≥97% to the 16sRNA of each organism were obtained and sample information was cross-referenced against de-identified patient information provided by Dr Ellen Li, Stony Brook University in New York, and Dr Rodney Newberry, Washington University, St. Louis. Reads from the same patient having >97% identity were combined to provide abundance numbers for each 16sRNA query.

Antibodies.

CD3-FITC (Clone: REA613, Miltenyi Biotec, catalog no. 130-113-138); CD4-PerCp-Vio700 (Clone: REA623, Miltenyi Biotec, catalog no. 130-113-228); CD69-VioGreen (Clone: REA824, Miltenyi Biotec, catalog no. 130-112-611); CD154-APC (Clone: REA238, Miltenyi Biotec, catalog no. 130-113-610); CD154-PE (Clone: 5C8, Miltenyi Biotec, catalog no. 130-092-289); CD137-PE (Clone: 4B4-1, Miltenyi Biotec, catalog no. 130-119-885); CD25-PE-Vio615 (Clone: REA945, Miltenyi Biotec, catalog no. 130-115-537); CD45RA-PE-Vio770 (Clone: REA1047, Miltenyi Biotec, catalog no. 130-117-746); CCR7-VioBlue (Clone: REA546, Miltenyi Biotec, catalog no. 130-117-353); interferon-gamma APC (Clone: 4S.B3, BD Biosciences, catalog no. 551385); Foxp3-Alexa Fluor 647 (Clone: 236A/E7, BD Pharmingen, catalog no. 561184); LIVE/DEAD fixable near-IR dead cell stain kit (ThermoFisher Scientific, catalog no. L10119); FcR Blocking Reagent (Miltenyi Biotec, catalog no. 130-059-901).

Isolation of human peripheral blood mononuclear cells and antigen-specific T cell assays.

Peripheral blood samples were processed within 4 hours of sample collection. Briefly, peripheral blood mononuclear cells (PBMCs) were isolated using density gradient centrifugation with Ficoll-Paque (GE Healthcare, catalog no. 17-1440-02) and cultured overnight at 37°C, 5% CO₂ in RPMI-1640 media (Gibco, catalog no. 11875–093) supplemented with 5% human AB serum (Corning, catalog no. 35-060-CI). Antigen-reactive T cell enrichment assay was adapted from a protocol described previously²¹ for both CD154⁺ T effector and CD137⁺ T regulatory cells. Briefly, PBMCs were stimulated at a density of 10⁷ cells/mL in a 24-well plate in the presence of 1 μg/mL anti-CD28 and 1 μg/mL anti-CD40 (Miltenyi Biotec, 130-093-375 and 130-094-133, respectively) with pooled bacterial flagellin proteins (10–20 μg/mL) for 6–7 hours (for CD154 induction) or 14 hours (for CD137 induction) at 37°C in 5% CO₂. Cells were then washed, labeled with anti-CD154-biotin or anti-CD137-biotin and subsequently anti-biotin magnetic beads (CD154 Microbead kit, Miltenyi Biotec, catalog no. 130-092-658, anti-human CD137-Biotin, Miltenyi Biotec, catalog no. 130-110-762) were used to enrich for CD154⁺ and CD137⁺ cells over magnetic separation columns, respectively. Brefeldin A (Bio-Legend, catalog no. 420601) at 1 μg/mL was added for the final 2 hours of stimulation when needed. Extracellular staining was performed while the cells were incubated on the first column, and intracellular staining was performed on the second column after fixation (eBioscience Foxp3 transcription factor staining buffer set, catalog no. 00-5523-00) when needed. Cells were then washed with MACs separation buffer (Miltenyi Biotec, catalog no. 130-091-376; no. 130-091-222,) and eluted. Enriched cells were then analyzed on Becton Dickenson LSRII flow cytometer.

Statistical Analysis

Statistical analyses were performed with GraphPad PRISM, version 7.0a, with Student t test used for 2-group comparisons, and one-way analysis of variance for multi-group analyses. All statistical tests were 2-sided. Specific tests used and values of n are stated in corresponding figure legends. Data were presented as mean ± SEM of replicate experiments. The differences were considered statistically significant at P < .01 (*P < .01, **P < .001, ***P < .0001, ****P < .00001) and n.s. indicates not significant.

Results

Human-Derived Lachnospiraceae flagellins Are Immunodominant Antigens in Crohn’s Disease

In silico comparison of gene sequences of known Lachnospiraceae flagellins derived from mouse cecal bacteria to flagellin sequences of the human microbiome in the Gen-Bank database identified 4 Roseburia species (intestinalis, hominis, inulinivorans, and faecis) and Eubacterium rectale (also classified as Agathobacter rectale in the Lachnospiraceae family^22,23), which had high homology in the amino- and carboxy-domains to previously defined flagellins from mouse Lachnospiraceae and clustered together phylogenetically (Supplementary Figure 1). These 5 species had a total of 23 flagellins, with 2–6 per bacterium, similar to the murine-origin Lachnospiraceae strains, with 3–8 flagellins per bacterial strain (https://www.ncbi.nlm.nih.gov/bioproject/175997). Based on sequence variation and Western blotting against a pool of Crohn’s sera, several of these flagellins were not further pursued, particularly those flagellins that had nonhomologous sequences most likely acquired via horizontal gene transfer. We ultimately identified 2 flagellins from each bacterium, synthesized the flagellin genes, and expressed them as proteins. A protein antigen microarray was performed on our previously defined mouse flagellins¹⁸ and the newly identified human-derived flagellins against sera from patients with IBD (152 patients with CD grouped according to the Montreal classification,^24,25 170 patients with UC) and 87 healthy controls (Table 1). Our cohort of patients with IBD exhibited strong heterogeneity in their response to microbial flagellins, but patients with CD overall displayed significantly elevated IgG responses to human-derived Lachnospiraceae flagellins in contrast to sera from patients with UC or HCs (Figure 1). CD subjects did not exhibit an elevated IgG response to inflammatory Proteobacteria flagellins E coli FliC or Salmonella dublin FliC (Figure 1), indicating the augmented IgG reactivity to Lachnospiraceae flagellins is not a global flagellin response. Furthermore, the elevated antibody response in patients with CD is not due to selective colonization of Lachnospiraceae bacteria in patients with CD, considering these organisms are represented in the terminal ileum of both patients with CD and patients with UC, as well as non-IBD controls (Table 2).

Table 1.

Study Demographics Characteristics Characteristic

Characteristic	Data
Total study participants	409
Age, y, mean	43
Range	19–81
Sex, %
Female	45
Male	55
Race and ethnicity, %
American Indian	2
Asian	<1
African-American	16
White Non-Hispanic	79
Hispanic	2
BMI, kg/m2, mean	25.4
Total HCs	87
Total patients with UC	170
Total patients with CD	152
CD with history of surgery	100
CD with history of biologic therapy	118
Montreal classification of CD, n (%)
Age of onset
A1, younger than 16 y	24 (15.8)
A2, 17–40 y	102 (67.1)
A3, older than 40 y	26 (17.1)
Location of disease
L1, ileal	52 (34.2)
L2, colonic	29 (19.1)
L3, ileocolonic	71 (46.7)
Behavior of disease
B1, nonstricturing/nonpenetrating	77 (50.7)
B2, stricturing	45 (29.6)
B3, penetrating	30 (19.7)
p, perianal disease	33 (21.7)

NOTE. Study participants (n = 409) were recruited from Kirklin Clinic of the University of Alabama at Birmingham Hospital Birmingham, Alabama, or healthy in-house volunteers in accordance with the University of Alabama at Birmingham Institutional Review Board guidelines.

Figure 1.

Patients with CD have elevated serum IgG to human Roseburia flagellins compared to patients with UC and HCs. Sera from HCs (n 87), CD (n 152) and patients with UC (n 170) were individually probed against 10 microbiota Lachnospiraceae bacterial flagellins (E rectale Fla1 and Fla2, R faecis Fla1and Fla3, R hominis Fla1 and Fla2, R intestinalis Fla1 and Fla2 and R inulinivorans Fla3 and Fla5) and Proteobacteria flagellins E coli FliC and S dublin FliC using a microbiota antigen microarray and developed with fluorescently labeled anti-human IgG in order to determine anti-flagellin antibody reactivity. Each flagellin was printed in quadruplicate and the resulting data are expressed as fluorescence intensity after being scanned on GenePix 4000B imager. Results are expressed as violin plots for each sample along with median (black) and mean (magenta) (*P < .01; **P < .001; ***P < .0001; ****P < .00001). IBD groups and HCs were compared using 1-way analysis of variance with Tukey-Kramer multiple comparison post-test.

Table 2.

Inflammatory Bowel Disease Patient and Healthy Control Expression Lachnospiraceae Bacteria in Terminal Ileum

Variable	R faecis	R intestinalis	R hominis	R inulinivorans	E rectale
Patients with CD (n = 106)
Copies/patient, mean	58.5	44.8	37	32.1	20.2
Patients with ≥1 copy, %	76.40	69.80	59.40	60.40	54.70
Non-IBD patients (n = 36)
Copies/patient, mean	42.9	20.3	15.5	14.6	32.3
Patients with ≥1 copy, %	69.40	50.00	47.20	55.60	69.40
Patients with UC (n = 32)
Copies/patient, mean	26.1	20.2	8.1	28.9	9.2
Patients with ≥1 copy, %	68.80	62.50	59.40	62.50	56.30

NOTE. In silico analysis of study SRP002479, “Effect of Crohn’s Disease Risk Alleles on Enteric Microbiota,” were obtained from the Sequence Read Archive was performed to determine presence of Lachnospiraceae bacteria in the terminal ileum of IBD (n = 106 CD, n = 32 UC) and non-IBD (n = 36) subjects.

The informative isotype of certain seroreactivities to microbe-derived constituents linked to CD, such as E coli outer membrane porin C and antibodies against Saccharomyces cerevisiae is IgA,²⁶ not IgG. In addition, recent methods, such as IgA sequencing have identified immunestimulatory bacteria to be coated with IgA in IBD.^27,28 In mice, IgA has been shown to provide protection against bacterial sepsis by coating invading bacteria, demonstrating a dynamic relationship between the microbiota and the immune system.²⁹ Furthermore, serum IgA has a shorter half-life (5–6 days) than IgG (21 days) and may reflect more recent exposure. In order to determine whether flagellins also induced a serum IgA response, which has not been previously reported, we analyzed sera from patients with CD, patients with UC, and HCs for IgA reactivity. Interestingly, elevated serum IgA was only detected for 4 of the 10 human-derived Lachnospiraceae flagellins examined in patients with CD (Supplementary Figure 2) and was not increased in the serum of patients with UC. In addition, the serum IgA responses to flagellins in patients with CD were of lesser magnitude compared with the IgG response to the same flagellins (Supplementary Table 1).

Patients with Crohn’s Disease Respond to Mouse and Human Lachnospiraceae Flagellins Selectively

Considering the robust IgG responses observed in patients with CD to human-derived Lachnospiraceae flagellins, we next sought to broaden our novel protein antigen microarray to include Lachnospiraceae flagellins derived from the mouse gut. Although we have previously reported augmented serum IgG to specific Lachnospiraceae mouse flagellins,^16,18,30 we expanded our analysis to include additional related flagellins (Supplementary Figure 1). We determined mouse-derived Lachnospiraceae flagellins elicited IgG responses in patients with CD (Figure 2) similar to human-derived Roseburia and Eubacteria flagellins (Supplementary Table 1). CD patient reactivity to Lachnospiraceae flagellins was consistent across mouse- and human-derived species. Furthermore, collective analyses of seroreactivity to human- and mouse-derived flagellins demonstrate that IBD subjects and HCs demonstrate IgA, IgG, or both IgA and IgG to particular flagellins, but the response in subjects with CD is heavily dominated by IgG (Supplementary Figure 3).

Figure 2.

Patients with CD have elevated serum IgG to multiple mouse Lachnospiraceae flagellins. Sera from HCs (n 87), CD (n 152), and UC (n 170) patients were individually probed against mouse microbiota Lachnospiraceae bacterial flagellins (L A4 Fla2 and Fla3, L FlaX, L CBir1 Fla, L CBir11 Fla, L 14–2 Fla, L CBir66 Fla, MDR254 Fla) using a microbiota antigen microarray as in Figure 1 and developed with fluorescently labeled anti-human IgG in order to determine anti-flagellin antibody reactivity. Results are expressed using violin plots for each sample along with median (black) and mean (magenta) (*P < .01; **P < .001; ***P < .0001; ****P < .00001). IBD groups and HCs were compared using 1-way analysis of variance with Tukey-Kramer multiple comparison post-test.

Identification of a Crohn’s Disease Subset With High Multi-Flagellin Reactivity

The data in Figures 1 and 2 show the IgG reactivity to each flagellin separately and not the response of each serum to the panel of flagellins; therefore, we generated a heat map showing the log₁₀ of the IgG response of each CD sera to all of the flagellins on the microarray (Figure 3). This revealed a subgroup of patients with CD with high IgG response to the majority of the Lachnospiraceae flagellins in the microarray, as well as others that had responses to a lesser number of flagellins. As noted in Figure 1, there was little reactivity in this group to the Proteobacteria flagellins from E coli and S dublin, and weak reactivity to the 3 Lachnospiraceae flagellins with less sequence homology to the others. One of the latter was R hominis Fla2, which had little or no IgG reactivity, despite a strong response to R hominis Fla1 from the same bacterium, demonstrating selective reactivity to a dominant cluster of Lachnospiraceae flagellins.

Figure 3.

A subset of patients with CD responds to multiple flagellins and associate with disease complications. Heatmap of log₁₀ IgG response (color) per individual CD patient (row) and flagellin (column). Both rows and columns were hierarchically clustered using Euclidean distance, average linkage and optimal leaf ordering maximizing sum of similarity between every leaf and all other leaves in the adjacent cluster. Patients without complicated disease are indicated by a magenta dot alongside the patient ID (rows). The bar graph to the right of the heatmap indicates the number of positive flagellin responses (out of 21) per patient.

We, and others, have previously identified CBir1 flagellin as an indicator of a more complicated disease course in CD.^10,17,30 Considering the close proximity of human- and mouse-derived flagellins phylogenetically in this study (Supplementary Figure 1), we sought to determine whether response to CBir1 flagellin in CD correlated with responses to multiple Lachnospiraceae flagellins. Indeed, elevated IgG responses to CBir1 flagellin in CD predicted increased IgG seropositivity to other Lachnospiraceae flagellins (Figure 4A). As measured by cumulative mean fluorescent intensity (MFI) in the protein antigen microarray, CBir1 reactivity correlated strongly with IgG reactivity to overall Lachnospiraceae flagellins (Figure 4B), but poorly with responses to Proteobacteria flagellins (Figure 4C and D), further supporting the specificity of flagellin responses in CD. Although CBir1 was an efficient indicator of total Lachnospiraceae IgG response in CD, most of the IgG responses to Lachnospiraceae flagellins analyzed strongly correlated with one another (Figure 4E), suggesting these responses may be due to shared epitopes, reflecting their close homology (Supplementary Figure 1).

Figure 4.

CBir1 antibody response in patients with CD correlates with total Lachnospiraceae flagellin response. Sera from CD (n 152) patients were individually probed for anti-flagellin reactivity against 19 mouse and human microbiota Lachnospiraceae bacterial flagellins and 2 Proteobacteria flagellins using the microbiota antigen microarray as in Figure 1 to 3 and analyzed in the following manner: (A) Correlation of anti-LCBir1 flagellin serum IgG response in MFI with the IgG positivity against 19 Lachnospiraceae flagellins in patients with CD. A vertical dashed line is drawn at mean 2 SD of anti-LCBir1 flagellin serum IgG response of HC, indicating IgG positivity to LCBir1 flagellin. (B) Linear regression analysis of anti-LCBir1 flagellin serum IgG response MFI with the average serum IgG response MFI against the rest of the 18 Lachnospiraceae flagellins in patients with CD. (C) Linear regression analysis of anti-LCBir1 flagellin serum IgG response with anti-S dublin FliC serum IgG response in patients with CD. (D) Linear regression analysis of anti-LCBir1 flagellin serum IgG response with anti-E coli FliC serum IgG response in patients with CD. (B–D) Patients with CD were grouped based as CBir1-positive (red) and CBir1-negative (blue), each symbol represents an individual patient. R² value for each group is shown. (E) Pearson correlation coefficients (color) between pairs of flagellin log-transformed IgG responses; rows and columns hierarchically clustered using average linkage.

An analysis of the disease features of the multi-flagellin–positive group confirmed that flagellin IgG seroreactivity was associated with a complicated course, such as surgical intervention in 67.4%, stricturing behavior (Montreal B2) in 27.9%, or penetrating behavior (Montreal B3) in 11.6%. Moreover, 77.4% of flagellin-reactive patients were on biologic therapy (Supplementary Table 2). Multi-flagellin reactivity was distributed among different disease locations with 37% ileal (Montreal L1), 21% colonic (Montreal L2), and 42% ileocolonic (Montreal L3) distributions. Patients with perianal fistula were 67% flagellin antibody–positive, with 50% having IgG responses to more than 10 flagellins, a greater proportion than in total patients with CD. Due to the largely refractory nature of our patient population (Table 1 and Supplementary Table 2), the incidence of disease complications in the subset of high-flagellin responders in CD did not achieve statistical significance, but there was a prominent trend toward this conclusion in our analysis. Prospective studies will be needed to determine the significance of this multi-flagellin reactivity for disease course.

Circulating Microbiota Flagellin-Specific CD154⁺ T Cells Are Increased in CD With Multi-Flagellin Seroreactivity

Considering the significant increase of seroreactivity to Lachnospiraceae flagellins in patients with CD, we next investigated whether corresponding circulating microbiota flagellin-specific CD4⁺ T cells were elevated. PBMCs were stimulated with pooled recombinant Lachnospiraceae flagellin antigens (Fla mix). Antigen-specific CD4⁺ TEFF and Treg cells were identified based on up-regulation of CD154 and CD137, respectively (Figure 5A, Supplementary Figure 4A), an assay established from the antigen-reactive T cell enrichment protocol.^21,31 Phenotypic analyses verified that CD154⁺ T cells obtained by short-term antigen stimulation significantly up-regulated expression of the activation marker CD69, and moderately up-regulated CD25, a hallmark for T cell activation (Supplementary Figure 4A). CD137⁺ T cells obtained by short-term antigen stimulation consisted of highly enriched CD25⁺Foxp3⁺ Treg_S (Supplementary Figure 4A). Patients selected for the multi-flagellin reactive phenotype were compared to patients with low or no flagellin reactivity. Consistent with the serologic findings, the frequencies of microbiota flagellin-specific CD4⁺ TEFF cells in PBMCs were significantly elevated in patients with serum IgG response against multiple flagellins (reactive to more than 10 of Lachnospiraceae flagellin antigens tested, CD high), but not in patients with low frequency IgG anti-flagellin (CD low) or HCs (Figure 5B). Contrary to the TEFF profile, microbiota flagellin-specific CD137⁺ Treg cells remained at a similar level across the groups (Figure 5C) and resulted in a significantly decreased ratio of CD137⁺/CD154⁺ in CD-high patients compared to CD-low patients and HCs (Figure 5D). Although CD154⁺CD4⁺ T cells could be detected post-flagellin stimulation in HCs, the majority of the antigen-specific cells displayed a naïve phenotype (CD45RA⁺), whereas those in CD high patients were largely CD45RA⁻, consistent with antigen experience (Figures 5E and F). In this connection, we confirmed that circulating flagellin-reactive CD154⁺ cells isolated from patients with CD had significantly higher expression of effector cytokine interferon-gamma post stimulation compared with HCs (Supplementary Figure 4B and C). Microbiota flagellin-specific CD4⁺ T cells displayed a much higher effector memory phenotype (CD45RA⁻CCR7⁻) in both CD154⁺ and CD137⁺ compartments in patients with CD vs HC (Figures 5G to I), whereas no such differences were found in the central memory phenotype (CD45RA⁻CCR7⁺). Previous studies have shown a stronger suppressive function of Treg cells that express CD45RA.^32,33 In agreement with this, we observed a significant decrease of CD45RA⁺ cells in the flagellin-specific CD137⁺ population in CD compared with HC (Figures 5G and J). These data showed fundamental differences in the frequency, function, and phenotype of microbiota flagellin-specific CD4⁺ TEFF and Treg cells in patients with CD compared with HCs, consistent with the seroreactivity data.

Figure 5.

Subset of patients with CD that respond to multiple flagellins have elevated CD4⁺CD154⁺ flagellin-reactive T cells. (A) Schematic of antigen-specific CD4⁺ TEFF and Treg cell enrichment assay. (B) Frequencies of flagellin-specific TEFF (CD154⁺) cells in CD4⁺ PBMCs in patients with CD stimulated with a Lachnospiraceae flagellin mix (Fla mix), including CBir1, 14–2 Fla1, A4 Fla3, and FlaX that have serum IgG response to more than 10 Lachnospiraceae flagellin antigens tested (CD high), patients with CD that have serum IgG response to 0–10 Lachnospiraceae flagellin antigens tested (CD low) and HCs. (C) Frequencies of flagellin-specific Treg (CD137⁺) cells in CD4⁺ PBMCs in CD high, CD low, and HCs. (D) Ratio of flagellin-specific Treg/TEFF (CD137⁺/CD154⁺) in CD high, CD low, and HCs. (E) Representative flow cytometry plot of CD45RA expression in flagellin-specific CD4⁺CD154⁺ cells in CD high (dark area), CD low (light gray area), and HCs (open area). (F) Frequencies of CD45RA^– cells in flagellin-specific CD4⁺CD154⁺ cells in CD high, CD low, and HCs. (G) Representative flow cytometry plots of CD45RA and CCR7 expression in flagellin-specific CD154⁺ and CD137⁺ cells, respectively, in CD high, CD low, and HCs. Subset identification reference plot is shown on the right. (H) Frequencies of the TEM (CD45RA–CCR7⁻) subset in CD154⁺ cells. (I) Frequencies of the TEM (CD45RA⁻CCR7⁻) subset in CD137⁺ cells. (J) Frequencies of the naïve (CD45RA⁺CCR7⁺) subset in CD137⁺ cells. Data are presented as mean ± SEM. *P < .05; **P < .01. (B–D, F, and H–J) each symbol represents an independent individual, 1-way analysis of variance.

Discussion

The current hypothesis on the development of IBD is that CD and UC occur in genetically susceptible individuals with augmented adaptive immune responses to commensal microbiota species, but relatively little is known about the specificity of this response within these 2 disorders. The human-derived Roseburia and Eubacterium bacterial strains in this study are members of the Lachnospiraceae family, Clostridiales order, Firmicutes phylum, as are the mouse-derived flagellated bacteria reported previously as linked to CD.¹⁸ Lachnospiraceae are gram-positive, obligate anaerobes that are abundant in the intestine of humans and other mammals.^34,35 These Roseburia strains and E rectale are present in the mucosa-associated bacteria in the human ileum in most individuals as shown in Table 2. In mice, native Lachnospiraceae have been found in the crypts of the proximal colon³⁶ and in a gnotobiotic setting Roseburia hominis colonized mice similarly³⁷; thus native human Lachnospiraceae likely colonize the human proximal colon in an analogous manner. The specificity of the antibody response to Lachnospiraceae flagellins in patients with CD, despite the presence of these bacteria in patients with UC (Table 2), suggests fundamentally different pathogenic mechanisms in CD and UC. Lachnospiraceae members, such as Roseburia and E rectale, are diminished in the microbiota of some patients with CD,³⁸ usually in concert with a decrease in Faecalibacterium prausnitzii,^39,40 a nonflagellated member of the Ruminococcus family of the Clostridiales. These bacteria selectively colonize the mucus layer⁴¹ and produce the short-chain fatty acid, butyrate. Butyrate is a major metabolic fuel for colonocytes, enhances the epithelial barrier, inhibits inflammatory responses by myeloid cells, and induces regulatory T cells, enhancing tolerance.⁴² These organisms are unquestionably beneficial to the host, and it is counterintuitive for them to be the target of an immune attack. Yet their presence in the ileum and proximal colon correlates with the dominant pattern of inflammation in CD, in which 75% of patients have ileitis or ileocolitis. The pattern of distribution of Lachnospiraceae may explain, in part, the lack of adaptive immune response to them in UC, which typically starts in the rectum. This is the first study to analyze serum antibody responses to a wide range of human Lachnospiraceae flagellins using a novel microbiota antigen microarray in both CD and UC.

Microbiota flagellins are immunodominant antigens, the only microbial proteins that have 3 genome-encoded receptors, and are among the first microbiota antigens to stimulate serum IgG antibodies in human infants.⁴³ Healthy human infants’ serum IgG antibodies to Lachnospiraceae flagellins are equivalent to those of adults with CD, then decline with age.⁴³ The amino and carboxy domain of flagellin sequences are highly conserved within a family like Lachnospiraceae, but are distinct from flagellins of other bacterial families. One can thus use flagellin sequences of these conserved domains to create phylogenies of flagellins and their associated bacteria (Supplementary Figure 1), or to identify bacteria-producing homologous sequences, as we did here to identify the human-derived flagellated Lachnospiraceae linked to CD. The strong IgG response seen in infants⁴³ suggests that the elevated IgG anti-flagellin seen in CD is an aberrant recall response rather than a de novo one. The presence of flagellin reactivity in CD but not UC, despite the presence of these bacteria in patients with UC (Table 2), indicates different pathogenic mechanisms, possibly due to genetic differences between the 2 disorders. The augmented response to flagellins appears to develop slowly, considering increased anti-flagellin and other antimicrobial seroreactivities can be detected years before diagnosis of C.^44–46 For example, the PREDICTS study identified the presence of antimicrobial antibodies, including those directed against CBir1, Fla2, and FlaX Lachnospiraceae flagellins, in the serum of a cohort of healthy military personnel years before clinical diagnosis of CD.⁴⁵ In a follow-up study, the PREDICTS study team established a panel of serum antibodies and proteins, enriched in pathways involving the complement cascade, innate immunity, glycosaminoglycan metabolism, and lysosomes, that accurately predicted a diagnosis of CD within 5 years, but did not identify any serum biomarkers of a diagnosis of UC.⁴⁶

The timing and longevity of the anti-flagellin IgG response, which matures with development, illustrates the importance of understanding how perturbations in the microbiota might alter health and disease, but the role that microbiota flagellin-specific IgG antibodies play in CD is uncertain. Subsets of B cells can play a regulatory role, and one such subset ameliorated experimental colitis by producing antibodies that remove antigen, effectively reducing pathogenic T cell activation.⁴⁷ In other models, IgG anti-flagellin aggravated experimental colitis.⁴⁸ Our data showing that high multi-flagellin IgG reactivity correlates with a high CD4⁺ T memory cell response does not distinguish between these possibilities.

Adaptive immune responses to antigens generate long-lived memory B cells and T cells. Multiple studies have shown CD4⁺ clonotypes, defined by T cell receptor (TCR)–CDR3 sequences, are expanded in CD intestine, present in peripheral blood, and persist during inactive disease.^49–51 The results of this study are consistent with a recent analysis of CD4⁺TCR repertoire in IBD⁵² that identified significant expansions of disease-associated CD4⁺ clonotypes in CD through sequencing of TCRβ CDR3.⁵² The generation of adaptive immune cell memory is the goal of vaccines to provide protection from pathogens; however, immune memory cells are responsible for chronicity in chronic immune-related inflammatory diseases. CD4⁺ T memory cells consist of 2 subsets, T central memory cells, which circulate through blood and secondary lymphoid organs, and TEM, which circulate through the blood and both lymphoid and nonlymphoid tissues, such as the intestine.⁵³ TEM can convert directly to TEFF cells, such as Th1, Th17, and other subsets. In experimental IBD, microbiota-reactive circulating CD4⁺ TEM are generated and cause accelerated colitis on transfer to RAG recipients.⁵⁴ Circulating CD4⁺ microbiota⁵⁵ and flagellin-reactive^56,57 T cells have been reported in patients with CD. Using the antigen-reactive T cell enrichment technique^21,31 to measure circulating flagellin-specific CD4⁺ T cells, we report increased levels of CD4⁺ TEM and a reduced Treg/TEM ratio in patients with multi-flagellin IgG seroreactivity, and a significant decrease in the more potent CD45RA⁺ Foxp3⁺ Treg cells. Because CD4⁺ T cells specific to a single flagellin, CBir1, cause colitis in T cell transfer models,⁵⁸ we propose the circulating flagellin-specific CD4⁺ TEM subset is pathogenic in patients and likely replenishes short-lived T effectors in the gut. Indeed, CBir1 flagellin–reactive CD4⁺ T cell have been demonstrated in the lamina propria of patients with CD,³⁰ and interruptions of such replenishment by the circulating flagellin-specific CD4⁺ TEM subset may explain, in part, the effectiveness of cell trafficking inhibitors in CD.

Patients with CD have reactivity to multiple microbiota antigens, but we propose that the flagellin immune response is a keystone reactivity for at least a subgroup of patients with CD. This is supported by the increased IgG anti-flagellin immune response years before diagnosis,^44,59 the colocalization of the Roseburia species and E rectale with disease location (Table 2), the presence of flagellin-reactive CD4⁺ T cells in the lesions,³⁰ and the association of increased flagellin seroreactivity with complicated outcomes in both adults^17,19 and children.⁶⁰

Assuming that the flagellin response is a keystone reactivity that leads to adaptive immune responses to other microbiota antigens by epitope spreading, we speculate that modulation of the adaptive and regulatory T cell response to flagellin may have beneficial effects on the response to diverse microbiota antigens. Given the increased CD4⁺ TEM and reduced Treg/TEM ratio found in the high IgG-seropositive group, a reduction of TEM and increase of Treg_S could be effective in restoring a balanced, more homeostatic immune response. Indeed, such a novel flagellin-peptide specific immunotherapy has been shown to prevent experimental colitis in mice,⁶¹ providing proof of principle that a flagellin-directed immunotherapy might provide similar benefits in patients.

WHAT YOU NEED TO KNOW.

BACKGROUND AND CONTEXT

Crohn’s disease and ulcerative colitis are characterized by dysregulated adaptive immune responses to the microbiota. A response to multiple microbiota antigens is associated with inflammatory bowel disease complications.

NEW FINDINGS

Patients with Crohn’s disease, but not patients with ulcerative colitis, have augmented serum IgG antibody responses to human ileal-localized Lachnospiraceae flagellins, with a subset that responds to multiple flagellins and has elevated flagellin-specific CD4 T memory cells.

LIMITATIONS

This cross-sectional study identifies a subset of patients with Crohn’s disease with multi-flagellin reactivity. Prospective studies will be needed to determine the role of this multi-flagellin adaptive immune reactivity in disease course.

IMPACT

B and T cell reactivity to multiple flagellins is associated with complicated Crohn’s disease and identifies those likely to benefit from early therapy. These data set the stage for microbiota antigen-directed therapy in Crohn’s disease.

Abbreviations used in this paper:

CD

Crohn’s disease

HC

healthy control

MFI

mean fluorescent intensity

PBMC

peripheral blood mononuclear cell

TCR

T cell receptor

TEFF

effector T cell

TEM

T effector memory cell

Treg

regulatory T cell

UC

ulcerative colitis