Commensal orthologs of the human autoantigen Ro60 as triggers of autoimmunity in lupus

Abstract

The earliest autoantibodies in lupus are directed against the RNA binding autoantigen Ro60, but the triggers against this evolutionarily conserved antigen remain elusive. We identified Ro60 orthologs in a subset of human skin, oral, and gut commensal bacterial species and confirmed the presence of these orthologs in patients with lupus and healthy controls. Thus, we hypothesized that commensal Ro60 orthologs may trigger autoimmunity via cross-reactivity in genetically susceptible individuals. Sera from human anti-Ro60-positive lupus patients immunoprecipitated commensal Ro60 ribonucleoproteins. Human Ro60 autoantigen-specific CD4 memory T cell clones from lupus patients were activated by skin and mucosal Ro60-containing bacteria, supporting T cell cross-reactivity in humans. Further, germ-free mice spontaneously initiated anti-human Ro60 T and B cell responses and developed glomerular immune complex deposits after monocolonization with a Ro60 ortholog-containing gut commensal, linking anti-Ro60 commensal responses in vivo with the production of human Ro60 autoantibodies and signs of autoimmunity. Together, these data support that colonization with autoantigen ortholog-producing commensal species may initiate and sustain chronic autoimmunity in genetically predisposed individuals. The concept of commensal ortholog cross-reactivity may apply more broadly to autoimmune diseases and lead to novel treatment approaches aimed at defined commensal species.

Conflict of interest statement

Competing interests: The authors declare that they have no competing interests.

Copyright © 2018 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works.

Figures

Fig. 1. The human commensal microbiota contains multiple species with Ro60 orthologs

(A) Phylogenetic tree showing commensal bacterial ortholog Ro60 sequence homology to hRo60, with commensal niches identified by colors of each branch (see legend). (B) Structure of hRo60 with a major B cell epitope [amino acids (aa) 169 to 190] mapped in pink and T cell epitope (amino acids 316 to 335) mapped in orange, and the protein sequence alignments between human and commensal Ro60 at these epitopes compared below. Green color indicates residues that interact with Y RNA. Blue color indicates residues that interact with misfolded RNA. P. prop, P. propionicum; C. amyc, C. amycolatum; A. mass, A. massiliensis.

Fig. 2. Ro60 commensal bacteria are common among lupus and healthy subjects without overt dysbiosis of the fecal, oral, or skin microbiome

(A) 16S V4 sequencing was performed from the fecal, oral, and skin microbiomes of 15 subjects with lupus and 7 healthy controls. Principal coordinates analysis of weighted UniFrac distances represents β-diversity by body site but not by the presence of serum anti-Ro60 antibodies. (B) Vertical bars represent relative abundance of microbial phyla in individual fecal, oral, and skin microbiome samples. Legend indicates most abundant phyla. No significant differences in linear discriminant analysis effect size were found between groups. (C) Heat map of relative abundance of four Ro60 commensal bacteria from human microbiome samples measured by bacterial Ro60-specific qPCR. Each row represents a study subject: healthy (NOR), SLE, or SCLE. (+) or (−) indicates serum anti-Ro60 immunoglobulin G (IgG) autoantibodies. Subjects completed up to three longitudinal visits, shown by column. White space indicates no sample. Legend indicates color relative to ΔΔCt value. There was no significant difference in the mean abundance between Ro60(+) and Ro60(−) individuals except P. prop from the chest (two-sample t test, *P = 0.02) (plot shown at the bottom right). (D) SCLE patient cutaneous lesional biopsies stained with a P. prop 16S rDNA–specific FISH probe (green) as well as the eubacterial probe EUB338 (red). DAPI, 4′,6-diamidino-2-phenylindole.

Fig. 3. Commensal-reactive T cell clones from lupus patients cross-react with hRo60 protein and a pathogenic Ro60 T cell peptide

Human memory T cells from an anti-Ro60–positive SLE patient were sorted into (A) CCR6− and (B) CCR6+ subsets and stimulated with the Ro60 commensal P. prop. X axis indicates SI, and y axis indicates proliferation as counts per minute (cpm). Each point on the graph represents one clone. Dotted line indicates autologous, irradiated monocyte control. (C) Restimulation of CCR6+ clones with SI ≥ 5 with recombinant hRo60 or the pathogenic Ro60 T cell peptide p316-335. (D) Identical experiment as in (A) to (C), stimulating CD4+ memory CCR6− and (E) CCR6+ lupus T cells from another anti-Ro60–positive SLE patient with recombinant bacterial Ro60 from the Ro60-containing gut commensal B. theta. (F) Similar to (C), CCR6+B. theta Ro60 (BtRo60)–reactive T cell clones were restimulated with recombinant hRo60 or the pathogenic Ro60 T cell peptide p316-335.

Fig. 4. hRo60-reactive patient T cell clones generated by a CD4+ T cell library assay cross-react with the Ro60 commensal P. prop

Human lupus memory T cells were sorted into (A) CCR6− and (B) CCR6+ subsets and stimulated with hRo60. X axis indicates SI, and y axis indicates proliferation as counts per minute (cpm). Each point on the graph represents one clone. (C) Three clones from each subset had SI ≥ 5 and were restimulated with the Ro60 ortholog–containing commensal bacteria B. theta and P. prop (two-sample t test, *P < 0.05). (D) Cytokine concentrations (in picograms per milliliter) from the supernatant of the cross-reactive clone 72 hours after stimulation. In (C) and (D), measurements were performed in duplicates, as indicated by the dots in each group. Dotted line indicates monocyte control.

Fig. 5. hRo60-reactive T cell clones cross-react with commensal Ro60-reactive mimic peptides and whole commensal bacteria

(A) Alignment of hRo60 T cell autoepi-tope peptide 369–383 with the corresponding amino acid sequences in commensal Ro60 orthologs of P. prop., C. amyc, and B. theta. CCR6− (B) and CCR6+ (C) memory CD4+ T cell subsets of an anti-Ro60–positive SLE patient stimulated with hRo60 protein. Y axis indicates proliferation as reactive light units (RLU) using a nonradioactive adenosine 5′-triphosphate (ATP) release assay. Dotted lines indicate monocyte control, which was set as baseline to zero for restimulation assays. (D) hRo60-reactive T cell clones of the same patient proliferated to hRo60 peptide 369–383, commensal ortholog mimic peptides, and whole commensal bacteria. (E) CD4+ T cell clone isolated from the peripheral blood of an anti-Ro60–positive SLE patient using a hRo60 peptide–specific tetramer cross-reacts with the Ro60 commensal C. amyc (*P < 0.05, two-sample t test).

Fig. 6. Human lupus sera immunoprecipitate YrlA RNA–containing RNPs from P. prop lysates

(A) After incubating P. prop lysates with human sera, RNAs in immunoprecipitates were extracted and subjected to Northern blotting to detect YrlA. First lane, molecular size markers (nucleotides). Total P. prop, total RNA extracted from the input lysate. Serum from a healthy donor is shown in the next lane. Six SLE sera and two SCLE sera are labeled with + or − representing the anti-Ro60 IgG antibody status by ELISA. (B) Predicted structure of P. prop YrlA. (C) As a negative control, the blot was reprobed for P. prop tRNA- proline-GGG. (D) Lupus sera from independent Harvard patient cohort (listed by identification numbers).

Fig. 7. Anti-Ro60–positive SLE sera bind the recombinantly expressed BtRo60 ortholog by Western blot

(A) Sera from four Yale SLE patients with positive serol-ogies for anti-hRo60 IgG were subjected to Western blotting with recombinant hRo60 (60kDa), recombinant BtRo60 (~56.5 kDa) and bacterial lysates from B. theta, as well as two skin/gut commensal strains that do not carry Ro60 orthologs, P. acnes and R. intestinalis (R. intes). (B) Sera from three Harvard SLE patients with positive serologies for anti-hRo60 IgG were subjected to Western blotting as in (A).

Fig. 8. Monocolonization of the GF mouse gut with B. theta leads to hRo60 T and B cell responses

(A) hRo60 ELISA of B. theta monocolo-nized C57Bl/6 mice (n = 15) after 3 to 5 months of coloni-zation compared to GF age-and sex-matched controls of the C57Bl/6 strain (n = 2). X axis shows twofold serial dilutions of sera from 1:250 to 1:4000, with the mean and SD of duplicate measurements plotted for each mouse. (B) Proliferation, measured by an ATP release assay, of cells from MLNs and (C) spleens from B. theta–monocolonized C57Bl/6 mice, stimulated with B. theta lysate or recombinant BtRo60 protein for 72 hours in triplicate. Each point represents the SI (proliferation divided by the mean background proliferation from cells with no antigen) of MLNs and spleens pooled by cage. P values were calculated using two-sample t tests. (D) hRo60 ELISA of B. theta– monocolonized, autoimmune- prone NOD mice (n = 6) after 2 weeks of colonization, compared to two age-matched GF mice (6-week-old C57Bl/6 females). X axis shows twofold serial dilutions of sera from 1:250 to 1:4000, with the mean and SD of duplicate measurements plotted for each mouse. (E) Proliferation assay of cells from MLNs and spleens (F) from monocolonized NOD mice stimulated with B. theta lysate, BtRo60, hRo60, and control proteins (BSA and β2GP1, an unrelated autoantigen) for 72 hours in triplicate as above. (G to L) Eight-week-old GF C57Bl/6 mice were monocolonized with B. theta followed by topical treatment with or without a TLR7 agonist (IMQ) for 8 weeks. Monocolonized mice were compared to age- and sex-matched GF C57Bl/6 mice with and without IMQ (n = 3 to 7 in each of the four groups). (G and H) Comparison of body weight and relative spleen weight in B. theta–monocolonized mice with and without IMQ versus GF controls. (I to K) Anti-hRo60, anti-dsDNA, and anti- RNA IgG ELISAs of B. theta–monocolonized and GF mice with or without IMQ after 8 weeks of treatment. B. theta monocolonization together with IMQ. (L) Renal immunofluorescence of B. theta–monocolonized and GF mice treated with or without IMQ. Immunofluorescence of kidneys was performed on kidney tissue from B. theta–monocolonized mice with or without IMQ and GF with IMQ. Anti-C3 (green), anti-C1q (teal), and anti-IgG, anti-IgA, and anti-IgM (red). OD, optical density; N.S., not significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, two-sample t test.