Lupus anti‐ribosomal P autoantibody proteomes express convergent biclonal signatures

Summary

Lupus‐specific anti‐ribosomal P (anti‐Rib‐P) autoantibodies have been implicated in the pathogenesis of neurological complications in systemic lupus erythematosus (SLE). The aim of the present study was to determine variable (V)‐region signatures of secreted autoantibody proteomes specific for the Rib‐P heterocomplex and investigate the molecular basis of the reported cross‐reactivity with Sm autoantigen. Anti‐Rib‐P immunoglobulins (IgGs) were purified from six anti‐Rib‐P‐positive sera by elution from enzyme‐linked immunosorbent assay (ELISA) plates coated with either native Rib‐P proteins or an 11‐amino acid peptide (11‐C peptide) representing the conserved COOH‐terminal P epitope. Rib‐P‐ and 11‐C peptide‐specific IgGs were analysed for heavy (H) and light (L) chain clonality and V‐region expression using an electrophoretic and de‐novo and database‐driven mass spectrometric sequencing workflow. Purified anti‐Rib‐P and anti‐SmD IgGs were tested for cross‐reactivity on ELISA and their proteome data sets analysed for shared clonotypes. Anti‐Rib‐P autoantibody proteomes were IgG1 kappa‐restricted and comprised two public clonotypes defined by unique H/L chain pairings. The major clonotypic population was specific for the common COOH‐terminal epitope, while the second shared the same pairing signature as a recently reported anti‐SmD clonotype, accounting for two‐way immunoassay cross‐reactivity between these lupus autoantibodies. Sequence convergence of anti‐Rib‐P proteomes suggests common molecular pathways of autoantibody production and identifies stereotyped clonal populations that are thought to play a pathogenic role in neuropsychiatric lupus. Shared clonotypic structures for anti‐Rib‐P and anti‐Sm responses suggest a common B cell clonal origin for subsets of these lupus‐specific autoantibodies.

Introduction

Autoantibodies binding the P0, P1 and P2 ribosomal P phosphoproteins (Rib‐P) are highly specific for systemic lupus erythematosus (SLE) and have been reported in association with neurological, renal and hepatic manifestations, with some conflicting results 1. Anti‐Rib‐P autoantibodies are detectable before SLE onset and bind an immunodominant COOH‐terminal epitope common to all three P proteins in addition to other regions on the native complex 2, 3, 4. Active and passive immunization studies in mice have underlined the clinical importance of anti‐Rib‐P by enhancing lupus nephritis and mediating depression, smell and memory deficits, potentially by cross‐reacting with a neuronal antigen expressed highly in the hippocampus termed the neuronal surface P antigen (NSPA) 5, 6, 7, 8. Functional antibodies against the major 11‐mer COOH epitope (11‐C) bind NSPA and appear to mediate cognitive impairment by enhancing glutamatergic transmission in the hippocampus 9. While an early immunochemical study reported limited heterogeneity and mainly immunoglobulin (Ig)G1 kappa restriction for COOH‐terminal responses 10, nothing is known at the protein sequence level of the clonotypic structure and Ig variable (V) gene usage of these potentially pathogenic lupus autoantibodies, and whether they share V‐region signatures with other systemic autoantibodies.

Human anti‐Rib‐P antibodies purified from the native complex and anti‐Rib‐P anti‐serum raised in mice have been reported to cross‐react with Smith (Sm) autoantigen, suggesting an immunological link between these two lupus‐specific autoantibodies, but the molecular basis of this cross‐reactivity is unknown 6, 11. We have recently developed a novel proteomic workflow based on plate‐purification of antigen‐specific Igs, two‐dimensional gel electrophoresis (2‐DE) and high‐resolution mass spectrometric sequencing to identify two public immunodominant clonotypes that drive anti‐SmD immunity, considered to have the highest specificity for SLE 12, 13. In the present study, we employ direct antibody sequencing to compare clonotypic structures and V‐region signatures of serum anti‐Rib‐P and anti‐SmD autoantibody proteomes. The findings reveal a novel immunological link between subsets of anti‐Rib‐P and anti‐SmD autoantibodies based on shared public clonotypes. Furthermore, we identify an anti‐Rib‐P monoclonal population specific for the immunodominant COOH‐terminal epitope that may play a pathogenic role in neuropsychiatric lupus.

Materials and methods

Patients and controls

Serum samples positive for anti‐Rib‐P autoantibodies by commercial line blot immunoassay (Euroline ANA profile 5; Euroimmun, Lubeck, Germany) were retrieved from the clinical immunology serum repository at the Flinders Medical Centre from six female patients (median age 51 years; range 22–76 years) who met the American College of Rheumatology revised criteria for SLE 14. All patients were positive for antibodies to double‐stranded DNA by Farr assay (Trinity Biotech, Bray, County Wicklow, Ireland). Demographic characteristics and serological findings in the patients are shown in the Supporting information, Table S1. Control sera were obtained from four female healthy donors (median age 50 years; range 21–72 years). The study was approved by the Clinical Ethics Committee of the Flinders Medical Centre.

Preparation and specificity analysis of anti‐Rib‐P autoantibodies

Serum anti‐Rib‐P or 11‐C peptide Igs were purified from enzyme‐linked immunosorbent assay (ELISA) plates (Maxi‐Sorp; Nunc, Roskilde, Denmark) using a recently reported elution method after coating with 1 μg/ml purified bovine native Rib‐P hetercomplex consisting of the P0 (38 kDa), P1(19 kDa) and P2 (17 kDa) phosphoproteins (Arotec Diagnostics, Wellington, New Zealand) or 10 μg/ml of the 11‐C peptide (SDEDMGFGLFD) synthetized as a 4‐branch MAP (Mimotopes, Notting Hill, VIC, Australia) 12. The activity and specificity of ELISA plate‐purified Igs for Rib‐P were determined by testing starting sera (diluted 1 : 100), unbound fractions (normalized to each starting serum) and bound/eluted Igs (2·5 μg/ml) for reactivity against native Rib‐P and native Ro60 control (Arotec Diagnostics) by ELISAs, as described previously 12. Purified anti‐Rib‐P Igs were tested for reactivity with native Rib‐P antigen by indirect immunofluorescence of human epithelial type 2 (HEp‐2) cells (Immunoconcepts, Sacramento, CA, USA). Purified anti‐Rib‐P and anti‐SmD Igs 12 were tested for cross‐reactivity on ELISA‐coated native Rib‐P, native SmD (Arotec Diagnostics), with recombinant (rec) Ro52 (Arotec Diagnostics) and rec influenza haemagglutinin from H1N1/2009 (HA‐09) (Protein Sciences Corporation, Meriden, CT, USA) as controls. Prior to mass spectrometric sequencing, the purity of affinity‐isolated Igs were verified by sodium dodecyl sulphate‐polyacrylamide gel electrophoresis (SDS‐PAGE).

2‐DE of purified autoantibodies

2‐DE was performed as described previously 12, with the following modifications: 13‐cm, non‐linear immobilized pH 3‐11 IPG strips (GE Healthcare, Uppsala, Sweden) were used in the first dimension and 8–15% gradient Criterion stain‐free TGX gels (Bio‐Rad, Hercules, CA, USA) at 300V using a Criterion electrophoresis system (Bio‐Rad) in the second dimension. Imaging was performed using a Gel Doc EZ Imager (Bio‐Rad).

Mass spectrometry (MS)

In‐solution digests were performed on plate‐purified Igs, while in‐gel digests were performed on heavy (H)‐ and light (L)‐chain bands and spots excised from SDS‐PAGE and 2‐DE gels using Trypsin Gold (Promega, Madison, WI, USA), as described previously 12. Analysis of peptides was carried out using a mass spectrometer (AB Sciex, Framingham, MA, USA) coupled to an Eksigent nanoLC 400 HPLC. Samples were applied to a C18 trap (Eksigent, Dublin, CA, USA) and eluted onto a 15‐cm C18 column (Nikkyos Technos, Tokyo, Japan) using a 2–40% acetonitrile gradient for 33 min. The instrument was operated in high‐sensitivity positive ion mode; charge state of +2 to +5 ions selected; with one MS scan followed by 20 MS/MS scans. At least two technical and biological replicates were performed for each sample.

Protein sequence data analysis

MS data were analysed using Peaks Studio version 7·5 software (Bioinformatics Solution Inc., Waterloo, ON, Canada) using the combined ImMunoGeneTics (IMGT) (http://www.imgt.org), NCBI and Uniprot 2010‐06 databases. The database search parameters were as follows: a maximum of two missed cleavages; precursor m/z tolerance of ≤10 parts per million (ppm); product ion error tolerance of 0·01 Da; precursor charge state of +2 to +4, as described previously 12.

Results

The anti‐Rib‐P autoantibody population is biclonal and shares a cross‐reactive clonotype with anti‐SmD

Monospecificity of eluted Igs for anti‐Rib‐P was confirmed by testing starting sera (anti‐Rib‐P/Ro60‐positive), unbound and eluted fractions on individual Rib‐P/Ro60 ELISAs (Fig. 1a). Furthermore, eluted Igs gave the typical anti‐Rib‐P cytoplasmic and nucleolar immunofluorescence staining pattern on HEp‐2 cells (Fig. 1b). Anti‐Rib‐P Igs purified from SLE patients with anti‐Rib‐P specifically bound Rib‐P and SmD on ELISA but not Ro52 and HA‐09 present in starting sera while, conversely, purified anti‐SmD Igs from anti‐Sm‐positive patients bound SmD and Rib‐P, indicating a weak two‐way cross‐reactivity between these autoantibody populations (Fig. 1c). SDS‐PAGE of the affinity‐purified Igs showed single H‐ and L‐chain bands, which were excised and sequenced by high‐resolution MS (Supporting information, Fig. S1).

Figure 1.

Specificity of anti‐ribosomal P (anti‐Rib‐P) immunoglobulins (IgGs) purified from native Rib‐P‐coated enzyme‐linked immunosorbent assay (ELISA) plates from sera of six patients with systemic lupus erythematosus (SLE). (a) Purified IgGs tested by ELISA using native Ribo‐P and Ro60 proteins. Starting, bound and unbound fractions are compared. Bars show the mean ± standard deviation (s.d.) of duplicate optical density values. (b) Indirect immunofluorescence of human epithelial type 2 (HEp‐2) cells probed with anti‐Rib‐P IgGs (2·5 μg/ml). (c) Reciprocal reactivity between anti‐Rib‐P and anti‐SmD IgGs against Rib‐P and SmD, and Ro52 and HA‐09 control proteins. S stands for starting serum and P for plate‐purified IgGs. Bars show the mean ± s.d. of duplicate optical density values.

The clonality of plate‐purified anti‐Rib‐P Igs was then assessed by high‐resolution 2‐DE, which revealed two H‐chain clusters of spots shared between patients with pIs ranging from 6·0 to 7·6 (identified as IGHV3·7‐JH6 on sequencing of gel plug digests) and 8·0 to 9·0 (IGHV1·3‐JH4) linked with two shared clusters of kappa (K) L‐chain spots ranging from pI 5·8 to 7·4 (IGKV1·39‐JK4) and pI 7·8 to 9·0 (IGKV3·20‐JK2) (Fig. 2a,b). Strikingly, the 11‐C peptide‐specific autoantibody population purified from patient SLE1 was monoclonal in terms of H‐ and L‐chain usage and matched the same IGHV1·3‐JH4/IGKV1·39‐JK4 signature seen within the total anti‐Rib‐P population of this patient (Fig. 2a,c). This identical 11‐C peptide‐specific monoclonal pairing signature was detected in an additional three anti‐Rib‐P‐positive patients, SLE3, SLE5, SLE6 (data available on request), consistent with the presence of a dominant clonal population of anti‐Rib‐P directed against the common COOH‐terminal epitope.

Figure 2.

Clonal restriction of total anti‐ribosomal P (anti‐Rib‐P) immunoglobulins (IgGs) and comparison with 11‐C peptide‐ and SmD‐specific IgGs. (a) and (b), Representative two‐dimensional gel electrophoresis (2‐DE) of native Rib‐P heterocomplex‐purified IgGs from two systemic lupus erythematosus (SLE) patients showing two H‐chain sets of spots sequenced as IGHV3·7‐JH6 (1) and IGHV1·3‐JH4 (2) together with two sets of L‐chain spots sequenced as IGKV1·39‐JK4 (3) and IGK3·20‐JK2 (4). (c), 2‐DE of 11‐C peptide‐ purified IgGs from patient SLE1 reveals persistence of H‐chain 2 and L‐chain 3 clonotypic sets (circled in red). (d) 2‐DE of SmD‐purified IgGs from an anti‐Sm‐positive SLE patient highlighting a shared IGHV3·7‐JH6/IGKV3·20‐JK2 clonotype with anti‐Rib‐P (circled in blue). The other anti‐SmD H‐ and L‐ chain spots (circled in green) are anti‐SmD‐specific and sequenced as IGHV1·69/IGKV2·28.

Analysis of purified anti‐SmD Igs run in parallel by 2‐DE revealed a clonotypic population specified by IGHV3·7‐JH6/IGKV3·20‐JK2 with the same electrophoretic mobility and clonal signature as the second anti‐Rib‐P clonotype (Fig. 2 a,b,d). Significantly, 11‐C peptide‐purified Igs did not react with SmD protein by ELISA, nor did SmD‐purified Igs bind the 11‐C peptide (data not shown).

Rib‐P‐reactive clonotypic IgGs express convergent H‐and L‐chain signatures

Mass spectrometric analysis was performed on solution trypsin digests of plate‐purified whole anti‐Rib‐P Igs from the six SLE patients to confirm 2‐DE assignment of clonotypes and obtain high‐resolution V‐region sequencing and mutational status. This confirmed the common expression of two IgG1 kappa‐restricted clonotypic populations specified by IGHV3·7‐JH6/IGKV3·20‐JK2 and IGHV1·3‐JH4/IGKV1·39‐JK4 pairing signatures. No lambda L‐chain tryptic peptides were detected in plug or solution digest of anti‐Rib‐P Ig, verifying kappa L‐chain restriction at the level of direct protein sequencing. Complete anti‐Rib‐P V‐region tryptic peptide maps are shown in the Supporting information, Fig. S2a–f, and reveal multiple homologous clonotypic peptides with public (shared among patients) and private amino acid replacement mutations consistent with selection of intraclonal variants by persistent antigen stimulation. Public (common) mutations are present in both framework regions (FR) and complementary determining regions (CDRs) of the H‐ and L‐chains and are tabulated by a proteomic heat map (Fig. 3). In contrast, no IgG tryptic peptides were detected by mass spectrometry in control experiments from Rib‐P‐coated ELISA plates treated with sera from healthy donors (n = 4).

Figure 3.

Variable (V)‐region peptide heat map of compiled de‐novo sequencing data from six patients with systemic lupus erythematosus (SLE) showing public (shared) amino acid replacement mutations. (a) H‐chain V‐region sequences align with germline IGHV3‐7 and IGHV1‐3 [ImMunoGeneTics (IMGT)] database. (b) L‐chain V‐region sequences align with germline IGKV3‐20 and IGKV1‐39. (c) H‐chain J‐region align with JH4 and JH6 germline sequence and L‐chain J‐regions align with JK2 and JK4 germline sequence. Shared amino acid replacement mutations divergent from the germline sequence are indicated in the text and colour‐coded according to the prevalence. Dots indicate amino acid matching to the germline sequence. Germline complementary determining regions (CDRs) are underlined.

Discussion

This study has utilized high‐resolution electrophoretic and mass spectrometric techniques to discover two public clonotypical autoantibodies that dominate humoral Rib‐P immunity in SLE. The proteomic data are consistent with humoral anti‐Rib‐P responses being driven by a dominant biclonal population, with one clone specific for the immunodominant COOH‐terminal epitope and the second accounting for cross‐reactivity with SmD protein. The expression of specific H/L chain pairings and sharing of V–J clonal signatures, combined with V‐region mutations common to different patients, appear to be general properties of systemic autoantibodies and emphasize the importance of recombinatorial bias and antigen‐driven clonal selection in shaping autoantibody repertories 12, 15. Recent studies based on deep sequencing of antibody repertoires following vaccination and infection have revealed a similar convergence of H‐chain responses in unrelated subjects, suggesting that natural selection against pathogens may guide autoantibody responses towards clonotypes readily available in the primary Ig repertoire 16.

Human and rabbit antibodies purified on the 11‐C peptide have been reported recently to impair memory in mice by binding NSPA on hippocampal neurones 7, implicating the IGHV1·3‐JH4/IGKV1·39‐JK4 clonotype as a pathogenic species in neuropsychiatric lupus. Independent passive transfer studies have implicated anti‐Rib‐P in renal and neurological manifestations, although it is unclear whether the transferred Ig was monospecifc for the COOH‐terminal P epitope 5, 6. If the 11‐C‐peptide‐specific clone proves to have pathogenic and functional properties in human disease, then removal or selective silencing of the clone may be a future therapeutic option in SLE. Early work suggests that this major autoreactive clonotype may bind to other cell surface proteins sharing homology with the COOH‐terminal P epitope on hepatocytes and T lymphocytes 1.

The IGHV3·7‐JH6/IGKV3·20‐JK2 clonotype, which binds an epitope outside the COOH‐terminus yet to be mapped, is notable for its immunological cross‐reactivity and shared germline H/L chain pairing signature with a recently identified anti‐SmD clonotype 12. These findings provide a novel molecular explanation for cross‐reactivity between two lupus‐specific autoantibodies based on shared clonotypic structures, suggesting a common clonal origin for at least a subset of these autoantibodies. We hypothesize that naive germline‐encoded IGHV3·7‐JH6/IGKV3·20‐JK2 B cells in the primary repertoire evade early B cell checkpoints in SLE patients and escape to the periphery, where they can undergo either Rib‐P‐ or Sm‐driven clonal selection, expansion and affinity maturation in germinal centres. The secretion of mature autoantibodies with different maturation pathways and V‐region mutation profiles would account for the weak cross‐reactivity detected on immunoassay.

The stereotyped immunoglobulin gene rearrangements and conserved H/L‐chain pairings in humoral anti‐RibP responses recapitulate findings for anti‐Ro/La and anti‐Sm proteomes in primary Sjögren's syndrome and SLE and support clonotypic sharing of autoantibodies as a unifying feature of systemic autoimmune diseases 12, 17, 18, 19. The marked clonal restriction and conserved V‐region gene usage observed for both anti‐Rib‐P and anti‐Sm Igs support a unifying mechanism of pathogenic autoantibody production in unrelated patients with SLE, based on highly similar, if not identical, B cell activation pathways from the original stimulus through to the generation of clones of autoantibody‐secreting cells. A corollary of the serum autoantibody proteome analysis is that these lupus‐specific autoantibodies are derived from a limited number of B‐clonal precursors that have evaded early tolerance checkpoints and undergone antigen‐driven clonal selection, expansion and affinity maturation in germinal centres. In a broader context, stereotyped B cell receptors are becoming recognized increasingly in infections, B cell cancers and autoimmune diseases, challenging the paradigm that immunoglobulin responses against the same antigen are determined randomly in different individuals 20. It remains unclear as to why identical determinants are selected on autoantigens among different patients 13, 21, 22, and why their structural features appear to channel responses to a few shared clones. While stereotypy at the level of both autoantigen and cognate autoantibody points to a deterministic model of highly similar pathways of autoantibody production from patient to patient 23, the presence of intraclonal variants with individual somatic mutations and different antibody levels in individual patients is consistent with a secondary role for somatic selection in shaping the final autoantibody response. Both public‐ and patient‐specific V‐region peptides have potential as diagnostic biomarkers using mass spectrometric multiple reaction monitoring platforms, and may also be useful to track autoreactive B cell clonal turnover.

Finally, this work reinforces the analytical power of direct mass spectrometric sequencing to determine molecular characteristics of serum autoantibody biomarkers, enabling identification of stereotyped responses at the level of secreted (serum) immunoglobulin proteomes. Prospective studies on SLE patients who have undergone formal neuropsychiatric assessment, together with new experimental approaches in animal models, will be required to unravel the diagnostic potential and pathogenic properties of these novel Rib‐P immunoreactive clonotypes.

Disclosure

This work was supported by an Australian National Health and Medical Research Council grant 1041900 to T. P. G. There is no commercial support for the work reported on in the manuscript.

Supporting information

Additional Supporting information may be found in the online version of this article at the publisher's web‐site:

Fig. S1. Representative sodium dodecyl sulphate‐polyacrylamide gel electrophoresis (SDS‐PAGE) of affinity‐purified immunoglobulins (IgGs) from patient systemic lupus erythematosus (SLE1) showing a single H‐ and a single L‐chain band, which were excised and sequenced by high‐resolution mass spectrometry.

Fig. S2. De‐novo sequencing of affinity‐purified anti‐ribosomal P (anti‐Rib‐P) immunoglobulins (IgGs) from six patients with systemic lupus erythematosus (SLE) reveals clonal restriction with common heavy (H)‐ and light (L)‐chain variable (V)‐ and joining (J)‐regions. (a) IGHV3‐7 gene family usage. (b) IGHV1‐3 gene family usage. (c) IGKV3‐20 gene family usage. (d) IGKV1‐39 gene family usage. A number of public V‐region mutations were identified which are tabulated by proteomic heat map in Fig. 3. A number of privately mutated peptides within each affinity‐purified anti‐Rib‐P IgG sample were also sequenced. (e) Sequencing of JH‐regions reveals an IGHJ4 and IGHJ6. (f) Sequencing of JK‐regions reveals an IGKJ2 and IGKJ4. Dots indicate homology with the germline sequence derived from the ImMunoGeneTics database; mutations divergent from germline are indicated in the text; germline complementary determining regions are underlined; _ indicates deletion; spaces indicate areas of incomplete sequence.

Table S1. Characteristics of patients with systemic lupus erythematosus (SLE).