Precipitating anti‐dsDNA peptide repertoires in lupus

Summary

Anti‐double‐stranded (ds)DNA autoantibodies are prototypical serological markers of systemic lupus erythematosus (SLE), but little is known about their immunoglobulin variable (IgV) region composition at the level of the secreted (serum) proteome. Here, we use a novel proteomic workflow based on de novo mass spectrometric sequencing of anti‐dsDNA precipitins to analyse IgV subfamily expression and mutational signatures of high‐affinity, precipitating anti‐dsDNA responses. Serum anti‐dsDNA proteomes were oligoclonal with shared (public) expression of immunoglobulin (Ig)G heavy chain variable region (IGHV) and kappa chain variable region (IGKV) subfamilies. IgV peptide maps from eight subjects showed extensive public and random (private) amino acid replacement mutations with prominent arginine substitutions across heavy (H)‐ and light (L)‐chains. Shared sets of L‐chain complementarity determining region 3 (CDR3) peptides specified by arginine substitutions were sequenced from the dominantly expressed IGKV3‐20 subfamily, with changes in expression levels of a clonal L‐chain CDR3 peptide by quantitative multiple reaction monitoring (MRM) paralleling the rise and fall of anti‐dsDNA levels by Farr radioimmunoassays (RIA). The heavily mutated IgV peptide signatures of precipitating anti‐dsDNA autoantibody proteomes reflect the strong selective forces that shape humoral anti‐dsDNA responses in germinal centres. Direct sequencing of agarose gel precipitins using microlitre volumes of stored sera streamlines the antibody sequencing workflow and is generalizable to other precipitating serum antibodies.

Introduction

Anti‐double‐stranded (ds) DNA are prototypical autoantibodies in systemic lupus erythematosus (SLE) and form part of the American College of Rheumatology (ACR) and Systemic Lupus International Collaborating Clinics (SLICC) criteria for the classification of SLE [1, 2]. Although anti‐dsDNA are used widely as diagnostic and longitudinal biomarkers in routine clinical practice, controversies persist regarding their diagnostic significance in nephritis, correlation with disease activity and prediction of flares in SLE and explanations of their origins and pathogenicity [3, 4, 5]. This relates, in part, to different anti‐dsDNA detection assays, with the more specific precipitating Farr radioimmunoassays (RIA) and Crithidia luciliae (CLIF) assays considered more likely to detect high‐affinity antibodies related to nephritis than those detected by solid‐phase immunoassays such as enzyme‐linked immunosorbent assay (ELISA) [6, 7, 8, 9]. However, many laboratories cannot perform the putative gold standard Farr RIA for anti‐dsDNA detection because of restrictions on radioactivity, raising a need for alternative approaches to profile precipitating anti‐dsDNA populations in patients with SLE [10, 11].

Considerable effort has gone into the isolation and sequencing of monoclonal anti‐dsDNA antibodies in mouse and humans, with an emphasis on the selection of arginine replacement mutations in the complementarity determining regions (CDRs) that favour DNA binding [12, 13, 14, 15, 16]. However, single‐cell analysis provides at best a snapshot of an antibody repertoire; moreover, cellular transcripts of monoclonal antibodies do not translate necessarily to secreted (serum) antibodies. Consequently, little is known about functional anti‐dsDNA repertoires at the level of their autoantibody proteomes.

We have developed proteomic methodology within our facility to analyse immunoglobulin (Ig)V peptide repertoires of anti‐Ro/La autoantibodies in primary Sjögren’s syndrome (SS) and anti‐Sm and anti‐ribosomal P responses in SLE, based on high‐resolution mass spectrometric sequencing of specific autoantibodies purified from stored serum samples [17, 18, 19, 20, 21, 22, 23]. In brief, anti‐Ro/La proteomes comprised oligoclonal expression patterns of shared, heavily mutated heavy (H)‐ and light (L)‐chains with anti‐Ro60 IgV molecular peptide fingerprints of anti‐Ro/La determinant spreading [22, 23, 24]. Conversely, lupus‐specific anti‐SmD and anti‐ribosomal P proteomes were biclonally restricted with shared and individual amino acid replacement mutations [17, 18]. This evolving technology is potentially transformative because it can identify molecular fingerprints and clonal turnover of pathogenic autoantibodies that are invisible to immunoassays used in current clinical practice [20]. Thus, a practical goal of autoantibody proteomics is to identify sets of surrogate IgV peptides for clonal tracking and monitoring of pathogenic autoantibody responses to biological therapeutics.

In the present study, we have adapted the proteomic antibody workflow for molecular profiling of precipitating anti‐dsDNA responses in the sera of lupus patients, using the streamlined approach of direct mass spectrometric sequencing of anti‐dsDNA purified from agarose gel precipitins, a method used by several groups in the past to precipitate anti‐dsDNA from lupus sera [25, 26, 27, 28, 29]. We find that anti‐dsDNA proteomes are comprised of restricted H‐ and L‐chain IgV subfamilies and express unique sets of IgV peptides specified by arginine substitutions.

Materials and methods

Overall proteomic workflow

An updated 2018 version of the serum autoantibody proteomic workflow is shown in Fig. 1 and includes additional front‐end antibody purification steps from agarose gel precipitins, double enzyme digests and quantitative multiple reaction monitoring (MRM) for tracking clonotype expression using surrogate CDR3 peptides.

Figure 1.

Proteomic workflow for serum anti‐dsDNA IgV peptide mapping. Anti‐dsDNA autoantibodies are isolated from systemic lupus erythematosus (SLE) serum by agarose gel double immunodiffusion with circular plasmid dsDNA in the central well. Well‐defined precipitins are excised and solubilized and heavy (H)‐ and light (L)‐chains separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS‐PAGE). Gel bands are digested by trypsin and chymotrypsin to generate immunoglobulin (Ig)V peptides for liquid chromatography–mass spectrometry (LC‐MS/MS). Germline and mutated IgV peptide sequences are derived by combined de novo sequencing and database matching. IgV subfamilies are assigned according to the presence of unique peptides corresponding to each gene family and the presence of additional supporting peptides. Identified CDR3 peptides can be used for clonotype tracking by using multiple reaction monitoring (MRM).

Patient and control sera

Residual sera from eight patients testing positive for anti‐dsDNA by Farr RIA (Trinity Biotech, Bray, County Wickow, Ireland) from the diagnostic laboratory were used for proteomic anti‐dsDNA profiling, and case notes reviewed retrospectively to confirm the diagnosis of SLE prior to de‐identification. Subjects fulfilled at least four of the 11 ACR revised classification criteria for SLE [1]. Demographic characteristics and serological findings are shown in Table 1. Anti‐dsDNA‐negative control sera were obtained with informed written consent from six healthy donors, eight primary SS, eight rheumatoid arthritis (RA) and three mixed connective tissue disease (MCTD) patients. The study was approved by the Southern Adelaide Clinical Research Ethics Committee with a waiver of consent for de‐identified anti‐dsDNA‐positive diagnostic sera (approved 17 November 2016; no. 39.034).

Table 1.

Characteristics of patients with SLE and their immunoglobulin (Ig)V subfamily gene usage of precipitating anti‐dsDNA autoantibody proteomes

Patients					IGHV§	IGHJ§	IGKV§	IGKJ§	IGLV§	IGLJ§
Code	Age/sex	Anti‐dsDNA (IU/ml)*	ENA†	ACR‡
SLE1	38/M	98	Sm	5	1–3, 3–23, 3–7, 3–74	2, 3, 4, 6	1–33, 1–5, 3–20, 4–1	1, 2, 3, 4	2–11, 3–21	1, 2
SLE2	20/F	96	negative	4	3–23, 3–7, 3–74, 4–34	3, 6	1–33, 1–39, 2–28, 3–20, 4–1	1, 2, 4, 5	–	–
SLE3	39/F	94	Ro52, Ro60	4	1–3, 1–69, 3–23, 3–7, 3–74, 3–9, 4–34,	3, 6	1–33, 1–5, 2–28, 3–15, 3–20, 4–1	1, 2, 3, 4	3–21	2
SLE4	57/M	70	negative	4	3–23, 3–7,3–74	3, 4, 6	1–33, 3–20, 4–1,	1, 2, 4, 5	–	–
SLE5	46/F	98	Ro52, Ro60, La	6	1–69, 3–15, 3–23, 3–7, 3–73, 3–74	3, 6	3–20, 1–5, 4–1, 1–33	1, 2, 4, 5	3–21	2
SLE6	25/F	94	Ro52, Ro60, Sm	8	1–69, 3–23, 3–7, 3–73, 3–74	3, 6	1–33, 3–20, 4–1,	1, 2, 3, 5	–	–
SLE7	62/F	49	Ro52, Ro60, La	5	3–15, 3–23, 3–7, 3–74, 3–9	3, 6	3–20, 4–1, 1–33	2, 3, 4, 5	–	–
SLE8	33/F	96	Sm	4	3–23, 3–48, 3–7, 3–74	2, 3, 4, 6	3–20, 4–1, 3–15, 1–33	2, 3, 4, 5	–	–

^*Anti‐dsDNA were measured by Farr radioimmunoassay;

^†ENA were measured by line‐blot immunoassay;

^‡American College of Rheumatology Criteria for Classification of systemic lupus erythematosus [1];

^§Public gene families (shared in four of eight patients) are in bold type. SLE = systemic lupus erythematosus; IgV = immunoglobulin variable region; anti‐dsDNA = anti‐double‐stranded DNA; ENA = extractable nuclear antigen; IGHV = IgG heavy chain variable region; IGHJ = IgG heavy chain joining region; IGKV = IgG kappa chain variable region; IGLV = IgG lambda chain variable region; IGKJ = IgG kappa chain joining region; IGLJ = IgG lambda chain joining region; M = male; F = female; – = not detected.

Mass spectrometry (MS) sequencing and protein sequence data analysis of anti‐dsDNA precipitins from SLE sera

Anti‐dsDNA antibodies were purified from serum precipitins reactions between circular plasmid dsDNA (Arotec, Wellington, New Zealand) and SLE serum by a modified Ouchterlony technique [30]. Briefly, double diffusion was performed in 1% agarose (SeaKem, Lonza, MD, USA) in a barbital buffer of pH 8·2 at 37°C for 48 h. Precipitins were excised, washed extensively in distilled water and solubilized using 1% sodium dodecyl sulphate (SDS) by boiling at 95°C for 5 min, and Igs fractionated by reduced SDS‐polyacrylamide gel electrophoresis (PAGE) (criterion stain‐free TGX gels; Bio‐Rad, Hercules, CA, USA). In‐gel tryptic and chymotryptic digests were performed on the 50 kD H‐ and 25 kD L‐chain bands and peptides analysed using a TripleTOF 5600+ mass spectrometer (AB Sciex, Framingham, MA, USA) coupled to an Ekspert nano LC 415 high‐performance liquid chromatography (HPLC) (Eksigent; AB Sciex), as detailed previously [17, 18, 21, 22, 23, 24]. Peptide de novo sequencing was performed by Peaks studio version 8.0 software (Bioinformatics Solution Inc., Waterloo, ON, Canada) using combined International ImMunoGeneTics (IMGT), National Center for Biotechnology Information (NCBI) and Uniprot 2017‐08 databases (Fig. 1). Parameters for database searches, data refinement and IgV gene family assignments have been described previously [17, 18, 21, 22, 23, 24, 31]; in brief, a maximum of two missed cleavages, precursor tolerance of < 15 parts per million, product ion tolerance of 0·02 Da, precursor charge state of +2 to +4, fixed modification carbamidomethylation, variable modifications oxidation and deamidation, a maximum of three modifications allowed and non‐specific cleavage at one end. High‐quality de novo peptides were selected based on sequences having an average local confidence score threshold greater than or equal to 75% and inspected manually to ensure correct assignments. A false discovery rate (FDR) threshold of 0·0% was applied at the peptide level to each data set. The IgV‐region subfamily is assigned from the presence of a unique peptide corresponding to the gene family. Purification of anti‐dsDNA Igs from individual patient sera was carried out on at least two independent occasions, and the purified Igs were run for MS at two technical replicates.

Multiple reaction monitoring (MRM)

MRM quantitative analysis was performed targeting specific anti‐dsDNA clonotypes. A TripleTOF 5600+ mass spectrometer (AB Sciex) coupled to an Ekspert 415 nano HPLC (Eksigent) was used for this analysis, which involved a repeating 2·8‐s cycle that comprised a 0·15‐s MS scan of all intact peptide ions, followed by the isolation and fragmentation of a suite of 12 anti‐dsDNA peptides with a dwell time of 0·1 s. The first four contain clonotypical light‐chain complementarity determining region (LCDR)3 sequences used subsequently for clonotypical monitoring, and the remaining eight are IgG kappa chain variable region (IGKV)3‐20 peptides representing their L‐chain subfamily. The annotated spectra of each individual peptide are shown in Supporting information, Fig. S2. Processing of MRM data files involving peak detection and integration on five different fragment ions per anti‐dsDNA peptide was performed using Multiquant version 1.2 software (AB Sciex). Specific fragment ions were selected, as they were sufficient to represent a unique spectral signature of each peptide. One point of Gaussian smooth width and one point of peak splitting was applied to fragment ion extracted ion chromatograms (XICs) that were all inspected manually to ensure that correct assignments were made. The XICs of individual L‐chain peptides were visualized in Skyline 3.6.0.10162 (MacCoss Laboratory, Department of Genome Sciences, University of Washington, Seattle, WA, USA) to demonstrate time alignment and signal strength (Supporting information, Fig. S2). The peak areas for all five fragment ions were averaged to provide a quantitative measure of LCDR3 peptide levels expressed as spectral counts per second (c.p.s.).

Results

Serum anti‐dsDNA autoantibody proteomes sequenced from agarose‐gel precipitins express shared IgV subfamilies

MS sequencing of tryptic and chymotryptic digests of IgG H‐chains of agarose gel‐isolated anti‐dsDNA immune precipitins revealed an oligoclonal repertoire with shared expression of the IGHV3‐23, 3‐7 and 3‐74 subfamilies in all subjects and minor expression of IGHV1‐69 subfamilies in some subjects. Low levels of IGHV4‐34 peptide expression were noted in precipitins from only two of eight subjects, probably related to the low‐affinity, polyreactive properties of these anti‐dsDNA subfamily autoantibodies in SLE [32]. Furthermore, we were unable to detect specific IGHV4‐59‐ and IGHV4‐39‐encoded peptides identified recently from anti‐dsDNA‐specific monoclonal antibodies (mAbs) derived from SLE plasmablasts [33]. Constant region sequencing identified IgG1‐ and IgG3‐specific peptides, consistent with previously reported anti‐dsDNA subclasses [34, 35]. L‐chain sequencing revealed that serum anti‐dsDNA were mainly kappa L‐chain restricted with dominant expression of IGKV3‐20 in all subjects followed by IGKV4‐1, 1‐33 subfamilies based on V‐region spectral counts. IGLV3‐21‐specific peptides were also detected, but these were minor species with low spectral counts (Table 1). Minor H‐ and L‐chain species were found in a few or individual patients and tabled as private H‐ and L‐chain clonotypical species (sequencing data not shown). H‐chain and L‐chain joining (J) sequences were similarly restricted to particular gene families (Table 1). Because of the multiple H‐ and L‐chain signatures derived from the ‘bottom‐up’ sequencing performed herein, it has not been possible to determine H/L chain pairings in the present study.

To investigate the specificity of anti‐dsDNA precipitin reactions, we tested healthy and disease control sera negative for anti‐dsDNA by Farr assay by double immunodiffusion. No precipitin lines were observed with dsDNA antigen on the gels by match sera from six healthy donors, eight patients with primary SS and anti‐Ro/La, eight with seropositive RA and three with MCTD and anti‐ribonucleoprotein (RNP) (data not shown).

IgV peptide maps of anti‐dsDNA precipitins reveal shared mutational signatures

IgV peptide signatures of anti‐dsDNA proteomes from eight SLE subjects are shown in Supporting information, Fig. S1, revealing complex patterns of public and private (random) amino acid replacement mutations in both H‐ and L‐chains consistent with ongoing dsDNA‐driven intraclonal selection and diversification. The frequencies of shared amino acid substitutions are collated in a proteomic heat map in Fig. 2. Arginine replacements were noted frequently across H‐ and L‐chains in the CDRs and flanking framework regions, with eight of eight subjects showing a serine to arginine substitutions at position 30 in the CDR1 of IGHV3 families and eight of eight subjects with serine to arginine substitutions at position 54 flanking CDR2 of IGKV3‐20. Lysine and asparagine substitutions, as reported in early studies of mouse monoclonal anti‐dsDNA antibodies [13], were also shared across the H‐ and L‐chains of serum anti‐dsDNA in our subjects.

Figure 2.

Immunoglobulin (Ig)V peptide heat map of compiled sequencing data from eight patients with systemic lupus erythematosus (SLE) revealing shared mutations. Combined de novo amino acid sequencing and database matching was performed on purified anti‐dsDNA IgGs. H‐ (a) and L‐chain (b) sequences were aligned with indicated germline V‐region gene families. Shared mutations divergent from the germline are shown with colour codes based on the frequency. Dots indicate homology with germline sequence derived from the ImMunoGeneTics (IMGT) database. Germline complementarity determining regions (CDRs) are underlined. Full peptide sequences are provided for each of the eight patients in Supporting information, Fig. S1.

A limitation of the current proteomic workflow serum antibody mixtures relates to the difficulty in obtaining accurate HCDR3 data (discussed in [31]). Conversely, we were able to efficiently fragment and identify serum IgV peptides specified by NH2‐terminal LCDR3 and flanking sequences from the dominant IGKV3‐20‐encoded L‐chains, thereby identifying clonotypical L‐chain peptides for potential tracking of anti‐dsDNA autoantibodies. These LCDR3 peptides were characterized by arginine substitutions at positions 92, 94, 96 and 97, enabling their efficient fragmentation and identification in tryptic anti‐dsDNA digests (Supporting information, Fig. S1). Notably, one or a combination of the four de novo sequenced peptides which contain clonotypical LCDR3 sequences (shown in bold type): PEDFAVELYYCQQYGSSPR (eight of eight subjects), PEDFAVELYYCQQYGR (seven of eight), PEDFAVELYYCQQYGSSR (six of eight) and PEDFAVETVYYCQQR (four of eight) were found in all eight SLE subjects, highlighting their potential utility as a panel of surrogate L‐chain peptide markers of secreted anti‐dsDNA autoantibody responses.

Expression of an anti‐dsDNA LCDR3 peptide parallels changes in anti‐dsDNA levels detected by Farr RIA

A technical comparison was performed between quantitative multiple reaction monitoring/mass spectrometry (MRM/MS) and Farr RIA to determine whether changes in anti‐dsDNA peptide expression paralleled those of anti‐dsDNA by conventional testing. In brief, additional paired serum samples from four de‐identified anti‐dsDNA‐positive patients were selected based on either a fall or rise in precipitating anti‐dsDNA levels by Farr RIA over time. Clinical details were not available for these patients. The anti‐dsDNA LCDR3 peptide PEDFAVELYYCQQYGSSPR, identified in all eight SLE subjects, was the most highly expressed and therefore chosen as the surrogate L‐chain clonotypical marker for charting against anti‐dsDNA levels. In the two paired sera showing a fall in anti‐dsDNA by Farr RIA there was a concordant decrease in the expression of the LCDR3 peptide by MRM/MS (Fig. 3a,b). In contrast, the paired serum sets with increased anti‐dsDNA levels over 3–4 months showed parallel rises in LCDR3 peptide expression (Fig. 3c,d), indicating that changes in anti‐dsDNA levels by quantitative proteomics are comparable to those of Farr RIA. The XICs of this surrogate peptide demonstrated consistent patterns among the four different patients, in terms of its relative fragment ion intensity, peak shape and width and similar retention times (Supporting information, Fig. S3), indicating the reliability of using this peptide in clinical application. Expression profiles for the less abundant clonotypical peptides also paralleled total anti‐dsDNA values, although these peptides were detected at lower levels than the surrogate L‐chain clonotypical peptide (Supporting information, Fig. S4).

Figure 3.

Comparison of clonotypical immunoglobulin (Ig)G kappa chain variable region (IGKV)3‐20 CDR3 peptide expression and anti‐dsDNA levels. Paired sera (a) and (b) with a fall in anti‐dsDNA over 3–4 months show a concomitant decrease in light‐chain complementarity determining region (LCDR)3 peptide expression by multiple reaction monitoring/mass spectrometry (MRM/MS), while paired sera (c,d) with a rise in anti‐dsDNA over 3 months show a concomitant increase in CDR3 peptide. Serum LCDR3 peptide is expressed as spectral counts per second (c.p.s.) and anti‐dsDNA levels are measured by Farr radioimmunoassays (RIA). The clonotypical CDR3 sequence contained within the monitored tryptic peptide is shown in bold type.

Discussion

In this study, we have reported the first MS sequencing of high‐affinity, precipitating anti‐dsDNA autoantibodies from SLE sera, heralding an innovative approach to the study of these ‘quintessential biomarkers of SLE’ [10] by discovering shared and individualized IgV peptide signatures among unrelated patients. Notably, lupus anti‐dsDNA proteomes are of greater complexity than the recently reported biclonal anti‐Sm and anti‐ribosomal P serum profiles in terms of their multiple IgV subfamily gene usage and intraclonal diversification [17, 18], suggesting different clonal origins and pathways of production for these three prototypical lupus‐specific autoantibodies. Similarly, in primary SS, some commonly used IgV subfamilies such as IGHV3‐23 and IGKV3‐20 are shared among anti‐Ro/La and anti‐dsDNA responses, but both express individual IgV subfamily and mutational profiles [21, 22, 23, 24]. Individual lupus patients showed specific anti‐dsDNA IgV peptide signatures for potential utility in the clinical setting as personalized molecular biomarkers of anti‐dsDNA responses. Prominent networks of conserved H‐ and L‐chain IgV arginine replacement mutations are reported here for the first time at the level of the serum (secreted) anti‐dsDNA repertoire. This reflects the strong selective forces shaping humoral anti‐dsDNA responses in SLE and the importance of arginine substitutions in binding to the negatively charged DNA molecule [12, 13, 14, 15, 16].

Here, we have developed a novel one‐step precipitin method to purify and sequence precipitating anti‐dsDNA from microlitre quantities of fresh or archived lupus sera. This streamlined workflow represents a significant advance over conventional solid‐phase affinity purification methods that cannot distinguish between low‐ and high‐affinity binding autoantibodies and require 10‐fold higher volumes of sera to yield comparable amounts of antibody for MS sequencing. While the present study uses circular plasmid dsDNA (as used in the Farr RIA) for anti‐dsDNA purification, we plan to extend these studies by comparing IgV peptide maps specific for other sources of DNA that may induce autoantibody production in lupus, such as mitochondrial (mt) dsDNA and nucleosomes. Molecular profiling of anti‐mtDNA will be of particular interest, given the recent emphasis on oxidized, mitochondrial DNA in expelled neutrophil extracellular traps (NETs) as pathogenic factors in lupus and potential drivers of anti‐dsDNA production [36, 37].

MS‐based antibody proteomics is a powerful serum‐based approach for mapping V‐gene usage and identifying shared mutated IgV peptides of diagnostic utility at the level of protein expression [19, 38]. However, the high‐resolution bottom‐up de novo LC‐MS/MS‐IgV peptide‐mapping workflow used herein cannot capture the complete repertoire of a serum antibody, most notably clonotypical HCDR3 regions, as these peptides are generally hydrophobic, ionize and fragment poorly and lack published HCDR3 matching databases [39]. Identification of HCDR3 sequences may be aided by searching serum antibody repertoires against personalized B cell receptor (BCR) databases by deep sequencing of peripheral blood mononuclear cells (PBMCs) [40, 41]. While these proteome–transcriptome matching approaches using PBMCs may be informative for vaccine responses in which peripheral blood can be timed to correspond with peak plasmablast responses, their utility in systemic autoimmunity is unproven, as autoreactive B cells and plasma cells are thought to reside mainly in tissues, bone marrow and/or lymphoid organs.

While it is considered preferable to define and monitor secreted antibody populations by HCDR3 profiling in view of their importance in antigen specificity, as aforementioned these peptides are often not feasible for use in proteomic studies. Because arginine mutations in the CDRs of both L‐chains and H‐chains of human anti‐dsDNA hybridomas appear to be involved in their interactions with dsDNA [14, 16], an alternative approach is to utilize clonotypical L‐chain CDR3 peptides. In this regard, clonotypical L‐chain peptides sequenced from serum paraproteins by LC‐MS/MS have been demonstrated in a proof‐of‐concept study to be sensitive proteomic markers for monitoring minimal residual disease in multiple myeloma [42]. Here, we have identified shared panels of de novo sequenced L‐chain CDR3 peptides specified by multiple arginine replacement mutations from the dominantly expressed IGKV3‐20 subfamily. In a pilot study, we have found that changes in expression levels of a clonal IGKV3‐20 CDR3 peptide using quantitative MRM/MS parallel the rise and fall of anti‐dsDNA levels by Farr RIA, indicating the potential utility of CDR3 peptides as clonotypical biomarkers for monitoring anti‐dsDNA responses. Individualized anti‐dsDNA IgV peptide signatures, in terms of their clonal restriction, IgV subfamily usage and mutational profiles, also deserve investigation as proteomic biomarkers for stratifying patients into treatment responder or non‐responder groups.

Serum IgV peptide profiling poses a challenge to traditional solid‐phase and fluid‐phase serological methods that are unable to resolve antibody responses at a molecular level. The technology can be adapted readily to infectious diseases and vaccine responses. In a recent proof‐of‐concept study, we have sequenced anti‐H1 antibodies in archived sera from H1N1pdm09 influenza‐vaccinated subjects and demonstrated convergent IgV subfamily landscapes of potential clinical utility [31]. Secreted antibodies can also be sequenced from other compartments, as shown recently for paired saliva and serum anti‐Ro60 precipitins from patients with primary SS [43]. In the longer term, temporal analysis of anti‐dsDNA proteomes in larger patient populations, including newly diagnosed and asymptomatic subjects with anti‐dsDNA, will be required to observe the dynamics of these secreted autoantibody repertoires. As the high‐affinity, somatically mutated anti‐dsDNA autoantibodies profiled herein can be considered to reflect germinal centre‐dependent responses, a priority will be to investigate IgV peptide signatures as companion diagnostics for biological therapeutics thought to block autoimmune germinal centre reactions, notably via CD40 ligand blockade [44, 45]. Quantitative MRM using the panel of surrogate anti‐dsDNA clonotypical L‐chain peptides identified herein will probably take the lead in these studies. In this regard, we have recently used MRM/MS with HCDR3 peptides to track serum rheumatoid factor (RF) clonotypes in patients with primary SS and mixed cryoglobulinaemia, and shown that pathogenic RF signatures appear years before clinical presentation [46]. Overall, this first comprehensive proteomic analysis of precipitating anti‐dsDNA IgV peptide repertoires in lupus serum offers a fresh approach for characterizing and monitoring anti‐dsDNA clonal populations with MS precision and accuracy, and for comparing their molecular signatures with other high‐throughput ‘omics’ technologies [47].

Disclosure

The authors have no financial and commercial conflicts of interest to disclose.

Author contributions

J. J. W., A. D. C., T. K. C., and T. P. G. conceived and designed the study. J. J. W. and A. D. C. performed the experiments. J. J. W., A. D. C., D. B., T. K. C, and T. P. G. analysed and interpreted the data. D. B. recruited and characterised serum samples. J. J. W. and T. P. G. wrote the paper and all authors revised it critically for important intellectual content.

Supporting information

Fig. S1. De novo sequencing of purified anti‐dsDNA IgV peptide repertoires from 8 patients with SLE. Individual peptide sequences derived from bottom up MS/MS were aligned with indicated germline V‐region gene families (a‐f). Dots indicate homology with the germline sequence derived from the IMGT database; amino acid substitutions divergent from germline are indicated in single letter code; germline complementarity determining regions are underlined; superscript indicates insertion; ‐ indicates deletion.

Fig. S2. Representative annotated MS/MS spectra (left panel) and their corresponding extracted ion chromatograms (XICs; right panel) for each of the fragment ions monitored in MRM analyses. The peptides used in MRM analyses are (a) peptides containing LCDR3 sequences (shown in bold) and (b) peptides derived from IGKV3‐20. C indicates a carbamidomethylated cysteine. b ions are indicated in green and y ions in red, with those ions used for MRM quantification indicated by an Asterix. m=mass; z=charge; ALC=average local confidence; ppm=part per million.

Fig. S3. Representative XICs from the clonotypic LCDR3 peptide (PEDFAVELYCQQYGSSPR) by MRM/MS to demonstrate the consistent patterns among four different patients (a‐d).

Fig. S4. Expression of the three less abundant clonotypic LCDR3 peptides derived from anti‐dsDNA repertoire as detected by MRM/MS in four different paired sera. Paired sera (a) and (b) had a fall in anti‐dsDNA over 3‐4 months, and paired sera (c) and (d) showed a rise in anti‐dsDNA over 3 months. Serum LCDR3 peptide is expressed as spectral counts per second (cps). The clonotypic CDR3 sequences contained within the monitored tryptic peptides are shown in bold.