Mass spectrometry-based analysis of rheumatoid factor

Abstract

Introduction: Rheumatoid factor (RF) are autoantibodies that are found in approximately two thirds of patients with rheumatoid arthritis, a chronic autoimmune disease characterized by potentially destructive inflammation of the joints. RF consists of polyclonal antibodies targeting the Fc part of immunoglobulin G. Despite its clinical relevance, RF is not specific for RA, and conventional assays for RF detection, predominantly solid-phase tests detecting IgM RF, suffer from poor harmonization and the disability to test more than one RF isotype.

Methods: We studied RF using a mass-spectrometry-based approach in RF(+), RF(-) rheumatoid arthritis patients and in disease controls. This allowed evaluation of RF at the amino acid level, including the variable and hypervariable region part of RF. RF was captured on Fc coated microwell plates, isolated, digested into peptides and analyzed by liquid chromatography tandem mass spectrometry. An initial proof-of-concept analysis was conducted comprising 12 samples, followed by a larger-scale experiment comprising 86 samples.

Results: Principal component analysis and sparse partial least squares discriminant analysis demonstrated that RF(+) RA patients displayed peptides that were differentially expressed compared with disease control patients. Framework region-derived peptides, variable region-derived peptides as well as de novo sequenced peptides not present in the human proteome database, were found to be enriched in RF(+) sera compared to disease control sera. Interestingly, some of these peptides were also upregulated in sera from RF(-) RA patients. Furthermore, mass spectrometry analysis revealed different RF isotypes. In addition to IgM, also IgA and IgG isotypes were observed. RF-IgG2 isotype was observed in RF(+) as well as in RF(-) RA patients.

Discussion: In summary, our findings highlight that mass spectrometry provides a platform for elucidating the heterogeneity and isotypic diversity of RF autoantibodies in RA, overcoming limitations inherent to current solid-phase RF assays. Upregulated de novo peptides were found, possibly related to the hypervariable regions of RF. Further validation using integrated proteomic and genomic approaches is required to confirm these novel peptides and their localization within the RF hypervariable regions.

Keywords: complementarity determining regions; immunoglobulins; isotypes; mass spectrometry; rheumatoid factor.

Figure 1

Sparse partial least squares discriminant analysis (sPLS-DA) reveals distinct rheumatoid factor (RF) Ig peptide repertoires between rheumatoid arthritis (RA) and controls (proof-of-concept experiment). RF from four RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) RA patients, three RF(−)/anti-CCP(−) RA patients, and four RF(−)/anti-CCP(−) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). Colors were assigned for different groups and consistently applied across the heatmaps and sPLS-DA figures. (A) sPLS-DA displaying the variance of the most discriminant peptides when comparing all Ig features in RF(+)/anti-CCP(+) RA patients, RF(−)/anti-CCP(−) RA patients, and RF(−)/anti-CCP(−) disease control patient samples. (B) Heatmap after hierarchical clustering of the features identified as being most discriminant within the Ig repertoire using sPLS-DA. Heatmap color key indicates the difference in Z-score (standardized normalized abundance). (C) sPLS-DA plotting the variable region repertoire based on the most discriminating features in RF(+)/anti-CCP(+) RA patients, RF(−)/anti-CCP(−) RA patients, and RF(−)/anti-CCP(−) disease control patient samples. (D) Heatmap after hierarchical clustering of the features identified as being most discriminant within the variable region repertoire using sPLS-DA. Heatmap color key indicates the difference in Z-score.

Figure 2

Heatmap representing Z-scores of different rheumatoid factor (RF) isotypes in rheumatoid arthritis (RA) patients and controls (proof-of-concept experiment). Comparison of RF isotypes in four RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) RA patients, three RF(−)/anti-CCP(−) RA patients, and four RF(−)/anti-CCP(−) disease controls after isolation, digestion into peptides, and analysis by liquid chromatography tandem mass spectrometry (LC-MS/MS). Data shown are the Ig isotypes (IgA, IgG subclasses, and IgM), quantified at the protein level based on unique peptides per protein of the Fc Ig part. *p <0.05 using Kruskal–Wallis testing followed by post-hoc Dunn’s test when comparing RF(+)/anti-CCP(+) RA patient samples with RF(−)/anti-CCP(−) disease control samples.

Figure 3

Sparse partial least squares discriminant analysis (sPLS-DA) of rheumatoid factor (RF)-derived peptides in rheumatoid arthritis (RA) and controls. RF from 27 RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) RA patients, five RF(−)/anti-CCP(+) RA patients, 22 RF(−)/anti-CCP(−) RA patients, 28 RF(−) disease controls, and four RF(+) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). Colors were assigned for different groups and consistently applied across the heatmaps and sPLS-DA figures. (A) sPLS-DA plotting the Ig repertoire based on the most discriminating peptides in sera from different groups. (B) Heatmap after hierarchical clustering of the features identified as being most discriminant within the Ig repertoire using sPLS-DA. Heatmap color key indicates the difference in Z-score. (C) sPLS-DA plotting the variable region repertoire based on the most discriminating peptides in sera from different groups. (D) Heatmap after hierarchical clustering of the features identified as being most discriminant within the variable region repertoire using sPLS-DA. Heatmap color key indicates the difference in Z-score.

Figure 4

Volcano plots reveal upregulation of a set of rheumatoid factor (RF)-derived peptides in RF(+) and RF(−) rheumatoid arthritis (RA) versus RF(−) disease controls. RF from 27 RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) RA patients, five RF(−)/anti-CCP(+) RA patients, 22 RF(−)/anti-CCP(−) RA patients, 28 RF(−)/anti-CCP(−) disease controls, and four RF(+)/anti-CCP(−) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). Sequenced features (Ig-related and non-Ig-related features) were incorporated in the volcano plots with log2(fold change) on the x-axis and −log10(adjusted p-value) on y-axis. Upregulated peptides in the RA group, defined as having an adjusted p-value <0.05 and an FC > 5, are presented in blue. Downregulated peptides, defined as an adjusted p-value <0.05 with FC < 0.2, are presented in red. Vertical dotted line represents FC of 0.2 and 5; the horizontal dotted line represents an adjusted p-value of 0.05. (A) Volcano plot comparing seropositive and seronegative RA patient samples versus disease controls. (B) Volcano plot comparing RF(−) RA patient samples versus disease controls. (C) Volcano plot comparing RF(+) RA patient samples versus disease controls. (D) Table summarizing upregulated hits when comparing different patient groups.

Figure 5

Volcano plots comparing rheumatoid factor (RF)-derived peptides from RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) patient samples with RF(−)/anti-CCP(−) control samples, with annotation of de novo hits and variable region hits in the upregulated peptides. RF from 27 RF(+)/anti-CCP(+) rheumatoid arthritis (RA) patients, five RF(−)/anti-CCP(+) RA patients, 22 RF(−)/anti-CCP(−) RA patients, 28 RF(−)/anti-CCP(−) disease controls, and four RF(+)/anti-CCP(−) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). Sequenced features (Ig-related and non-Ig-related features) were incorporated in the volcano plots with log2(fold change) on the x-axis and −log10(adjusted p-value) on y-axis. Shown are volcano plots comparing RF(+)/anti-CCP(+) patient samples with RF(−)/anti-CCP(−) control samples. Vertical dotted lines represent FC of 0.2 and 5, and horizontal dotted line represents an adjusted p-value of 0.05. Features were considered upregulated when having an FC > 5 and an adjusted p-value <0.05. (A) The upregulated de novo peptides that could be identified through de novo sequencing are annotated in purple. (B) Upregulated variable region peptides identified through a Mascot database search, annotated in yellow.

Figure 6

Heatmap illustrating de novo upregulated rheumatoid factor (RF)-derived peptides overexpressed in RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) rheumatoid arthritis (RA) patients compared to RF(−)/anti-CCP(−) disease controls, with hierarchical clustering of peptides based on their Z-scores. RF from 27 RF(+)/anti-CCP(+) RA patients, five RF(−)/anti-CCP(+) RA patients, 22 RF(−)/anti-CCP(−) RA patients, 28 RF(−) disease controls, and four RF(+) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). The heatmap illustrates the Z-scores of the upregulated de novo peptides. Peptides were considered upregulated when having an adjusted p-value <0.05 and an FC > 5 when comparing RF(+)/anti-CCP(+) RA patients with RF(−)/anti-CCP(−) disease controls. A full list of de novo sequences with their associated mass-to-charge values, charges, retention time, FC, and adjusted p-values is given in Supplementary Table 4 .

Figure 7

Heatmap illustrating rheumatoid factor (RF)-derived variable region peptides overexpressed in RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) rheumatoid arthritis (RA) patients compared to RF(−)/anti-CCP(−) disease controls, with hierarchical clustering of peptides based on their Z-scores. RF from 27 RF(+)/anti-CCP(+) RA patients, five RF(−)/anti-CCP(+) RA patients, 22 RF(−)/anti-CCP(−) RA patients, 28 RF(−) disease controls, and four RF(+) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). The heatmap illustrates the Z-score of the upregulated variable region peptides. Peptides were considered upregulated when having an adjusted p-value <0.05 and an FC > 5 when comparing RF(+)/anti-CCP(+) RA patients with RF(−)/anti-CCP(−) disease controls. All peptides were identified through a Mascot database search. A full list of variable region sequences with their associated mass-to-charge values, charges, retention time, FC, and adjusted p-values is given in Supplementary Table 5 .

Figure 8

Volcano plot comparing rheumatoid factor (RF)-derived peptides from RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) patient samples with RF-derived peptides derived from RF(−)/anti-CCP(−) disease control samples, highlighting 13 upregulated peptides revealed by sparse partial least squares discriminant analysis (sPLS-DA). RF from 27 RF(+)/anti-CCP(+) rheumatoid arthritis (RA) patients, five RF(−)/anti-CCP(+) RA patients, 22 RF(−)/anti-CCP(−) RA patients, 28 RF(−)/anti-CCP(−) disease controls, and four RF(+)/anti-CCP(−) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). Sequenced features (Ig-related and non-Ig-related features) were incorporated in the volcano plots with log2(fold change) on the x-axis and −log10(adjusted p-value) on y-axis. Shown are volcano plots comparing RF(+)/anti-CCP(+) patient samples with RF(−)/anti-CCP(−) control samples. Vertical dotted lines represent FC of 0.2 and 5, and horizontal dotted line represents an adjusted p-value cut-off <0.05. Thirteen features, annotated in purple on the volcano plot, situated in the framework or in the complementarity-determining regions, were more abundant in RF(+) RA than in controls, had an adjusted p-value <0.05, and were present after 50 repeats in sPLS-DA in cross-validation. In the right-hand panel, a heatmap is shown illustrating the median Z-score per peptide in each patient or disease control group, with peptides annotated per row.

Figure 9

Heatmap representing Z-scores of different rheumatoid factor (RF) isotypes in rheumatoid arthritis (RA) patients and controls. RF from 27 RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) RA patients, five RF(−)/anti-CCP(+) RA patients, 22 RF(−)/anti-CCP(−) RA patients, 28 RF(−) disease controls, and four RF(+) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). Z-score of Ig heavy chain protein isotypes was compared in different disease groups. p-Value annotation for comparison of seropositive patient samples [RF(+)/anti-CCP(+)RA] versus disease control samples [RF(−)/anti-CCP(−) C] using Kruskal–Wallis testing followed by post-hoc Dunn’s test: *p < 0.05 and **p < 0.001. p-Value annotation for comparison of seronegative patient samples [RF(−)/anti-CCP(−) RA] versus disease control samples [RF(−)/anti-CCP(−) C] using Kruskal–Wallis testing followed by post-hoc Dunn’s test: ^p < 0.05 and ^^p < 0.001. Regarding IgM and IgG1, two UniProt accessions were found in PEAKS software (P0DOX5 and P01857 for IgM and P0DOX6 and P01871 for IgG1). As they have a high correlation and similar sequences, as shown in Supplementary Figure 13 , only one accession number is illustrated in this figure. UniProt accession numbers corresponding with the protein description as listed in this figure are P01876, P0DOX2, P01857, P01859, P01860, P01861, and P01871.

Figure 10

Quantification of Ig heavy chain rheumatoid factor (RF) isotypes in rheumatoid arthritis (RA) patients and controls. RF from 27 RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) RA patients, five RF(−)/anti-CCP(+) RA patients, 22 RF(−)/anti-CCP(−) RA patients, 28 RF(−) disease controls, and four RF(+)/anti-CCP(−) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). Protein abundance of Ig heavy chain isotype was obtained by totaling the peptide abundances of unique peptides per protein. Area under the curve for each protein is listed. Regarding IgM and IgG1, two UniProt accessions were found in PEAKS software (P0DOX5 and P01857 for IgM and P0DOX6 and P01871 for IgG1). As they have a high sequence similarity (shown in Supplementary Figure 13 ), only one accession number is illustrated in this figure. UniProt accession number P01871 represents the heavy mu chain, and accession P01857 represents IgG1 subclass.

Figure 11

Summary statistics and analysis of data presented in Figure 10 . Rheumatoid factor (RF) from 27 RF(+)/anti-cyclic citrullinated peptide (anti-CCP)(+) rheumatoid arthritis (RA) patients, five RF(−)/anti-CCP(+) RA patients, 22 RF(−)/anti-CCP(−) RA patients, 28 RF(−)/anti-CCP(−) disease controls, and four RF(+)/anti-CCP(−) disease controls was isolated, digested into peptides, and analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS). Data shown are the immunoglobulin isotype hits, quantified using unique peptides as relative protein quantification method. (A) p-Values presented in diagonal format matrices per RF isotype. Different groups were compared using Kruskal–Wallis statistical testing followed by a post-hoc Dunn’s test. (B) Number of samples considered positive with cut-off values based on the highest abundant sample in the RF(−)/anti-CCP(−) disease control group. Two UniProt accessions were found for the heavy constant γ1 chain, namely, P01857 and P0DOX5, situated in the same protein group. These are largely the same sequence and only marginally different (illustrated in Supplementary Figure 13 ).