Patterns of ANA+ B cells for SLE patient stratification

Abstract

IgG antinuclear antibodies (ANAs) are a dominant feature of several autoimmune diseases. We previously showed that systemic lupus erythematosus (SLE) is characterized by increased ANA+ IgG plasmablasts/plasma cells (PCs) through aberrant IgG PC differentiation rather than an antigen-specific tolerance defect. Here, we aimed to understand the differentiation pathways resulting in ANA+ IgG PCs in SLE patients. We demonstrate distinct profiles of ANA+ antigen-experienced B cells in SLE patients, characterized by either a high frequency of PCs or a high frequency of IgG+ memory B cells. This classification of SLE patients was unrelated to disease activity and remained stable over time in almost all patients, suggesting minimal influence of disease activity. A similar classification applies to antigen-specific B cell subsets in mice following primary immunization with T-independent and T-dependent antigens as well as in lupus-prone mouse models (MRL/lpr and NZB/W). We further show that, in both lupus-prone mice and SLE patients, the classification correlates with the serum autoantibody profile. In this study, we identified B cell phenotypes that we propose reflect an extrafollicular pathway for PC differentiation or a germinal center pathway, respectively. The classification we propose can be used to stratify patients for longitudinal studies and clinical trials.

Keywords: Autoimmunity; B cells; Immunology.

Figure 1. Distinct phenotypes of ANA⁺ B cells and PC in SLE patients.

(A–F) Frequencies of ANA⁺ and total B cell subsets in healthy controls (n = 15) and SLE patients (n = 36). Each dot indicates an individual, and the bars represent the median. *P < 0.05; **P < 0.01, using Mann-Whitney test. (G and H) Principal component analysis of all B cell parameters analyzed (frequencies of ANA⁺ and total B cell and PC subsets). The percentage indicated on the axis is the percentage of variance explained by that principal component. SLE patients were separated based on whether they displayed an increase in the frequency of ANA⁺ IgG PCs compared with healthy controls (quartile 3 + 1.5 × interquartile range). Patients without expansion are denoted as “cluster 0.” (G) The variables contributing to each dimension in principal component analysis. The length and direction of each arrow shows the strength of their contribution to each PC. (H) The coordinates of each healthy individual and SLE patient. Ellipses represent the 95% confidence interval for each group. (I) Frequency of ANA⁺ IgG⁺ memory B cells and ANA⁺ PCs in SLE patients with an expansion of ANA⁺ PCs. Each dot indicates a patient (n = 22). (J) Relative proportion of ANA⁺ IgG memory B cells and ANA⁺ IgM and IgG PCs in healthy controls and SLE patients. The median and range of SLEDAI scores is shown below each circle. Patients were defined as “cluster 1” when their ANA⁺ PCs were >20% of the total ANA⁺ antigen-experienced cells (memory B cells and PCs). Patients were defined as “cluster 2” when their ANA⁺ PCs were <20% of the total ANA⁺ antigen-experienced cells. (K) Principal component analysis as in C and D, here only showing patients from cluster 1 and 2. ANA, antinuclear antibody; HC, healthy control; PC, plasmablast/plasma cell; SLE, systemic lupus erythematosus; SLEDAI, SLE disease activity index.

Figure 2. Stability of clustering over time.

SLE patients were classified as described in Figure 1, and a follow-up was done between 1 month and 1 year. (A) Relative proportion of ANA⁺ IgG memory B cells and ANA⁺ IgM and IgG PCs in healthy controls and SLE patients. The SLEDAI scores at the time of each measurement are shown below each circle. Patients indicated in red had a change in classification between the first and second assessment. (B) Principal component analysis as in Figure 1, including all initial measurements for healthy subjects and patients, was used to predict the follow-up measurements in each patient. Individual dots represent only those patients with a follow-up measurement. Arrows indicate the follow-up measurement. Dotted black arrows indicate 2 patients (SLE 19 and SLE 23) for whom the classification based on the proportions in A changed over time. ANA, antinuclear antibody; HC, healthy control; PC, plasmablast/plasma cell; SLE, systemic lupus erythematosus; SLEDAI, SLE disease activity index.

Figure 3. Frequencies of memory B cells and PCs upon primary immunization in mice.

(A) Model for differentiation of PCs. Two pathways can result in the differentiation of ANA⁺ IgG⁺ PCs, extrafollicular and germinal center, each leading to different proportions of IgG⁺ memory cells, IgM PCs, and IgG PCs. (B–G) C57BL/6 mice were immunized with NP-Ficoll or NP-CGG and analyzed at day 7, 14, and 21 after immunization. (B) Representative flow cytometry plots showing the frequency of NP⁺ cells among GC B cells, memory B cells, and PCs (the latter was analyzed upon cytoplasmic staining for NP). (C–F) Frequencies of NP⁺ GC B cells, NP⁺ memory B cells, NP⁺ IgM PCs, and NP⁺ IgG PCs. Each dot shows an individual mouse, and the bar represents the median. (G) Proportion of NP⁺ memory B cells and PCs These were calculated after subtracting the number of NP-specific cells in each population in unimmunized mice. Each pie chart shows the median for each strain (n = 4–6 mice per group). P values were calculated using χ² test. ANA, antinuclear antibody; CGG, chicken γ globulin; FVD, fixable viability dye; GC, germinal center; NP, 4-hydroxy-3-nitrophenylacetyl; PC, plasmablast/plasma cell; SHM, somatic hypermutation.

Figure 4. Frequencies of ANA⁺ memory B cells and PCs in MRL/lpr and NZB/W mice.

(A–G) Splenocytes from MRL/lpr and NZB/W mice were stained with nuclear extract to analyze the percentage of ANA⁺ B cells. Each dot indicates an individual mouse (n = 5–8 for each group), and the bars represent the median. (A) Proportion of ANA⁺ antigen-experienced cells in the spleens of each mouse strain. NZB/W “Change during seroconversion” was calculated by subtracting the maximal measurements in young mice (quartile 3 + 1.5 × interquartile range) from the measurements in old mice. (B–D) The frequencies of total and ANA⁺ IgG3⁺ PCs. (E) dsDNA-reactive IgG3 in serum of MRL/lpr and NZB/W mice was determined by ELISA. (F and G) The relative number of IgG3⁺ and IgG1/2⁺ cells within the population of IgG ANA⁺ PCs of MRL/lpr and NZB/W mice. (H and I) The relative number of IgG3⁺ and IgG1/2⁺ cells within the population of IgM^– NP⁺ PCs of C57BL/6 mice immunized with NP-Ficoll (day 7) and NP-CGG (day 7 and day 21). *P < 0.05; **P < 0.01; ***P < 0.001, using χ² test (A), Mann Whitney U test (B–E), or 2-way ANOVA with Bonferroni post-hoc test (G and I). ANA, antinuclear antibody; CGG, chicken γ globulin; NP, 4-hydroxy-3-nitrophenylacetyl; PC, plasmablast/plasma cell.

Figure 5. Characteristics of serum autoantibodies in lupus mice.

(A–D) Relative affinity of anti-dsDNA IgG and IgM was measured in serum of MRL/lpr and NZB/W mice by inhibition ELISA. (A and B) Relative OD (average ± SEM) of each strain. (C and D) IC₅₀ values for each individual mouse. (E–H) Profiles of recognition of nuclear antigens by serum IgG (E and F) and IgM (G and H). (E and G) Heatmaps, including hierarchical clustering based on the profile of antigen recognition. (F and H) The number of nuclear antigens recognized. Each dot indicates an individual mouse (n = 6–9 for each group), and the bars represent the median. *P < 0.05; **P < 0.01, using Mann Whitney U test. ANA, antinuclear antibody; cyto, cytoplasmic; PC, plasma cell; SS-A/B, anti–Sjögren’s syndrome–related antigen; Sm, Smith antigen.

Figure 6. Characteristics of serum autoantibodies in SLE patients.

(A and B) Profiles of recognition of nuclear antigens by serum IgG. Patients are separated based on classification, as described in Figure 1, into cluster 1 and cluster 2. (A) Heatmap, including hierarchical clustering based on the profile of antigen recognition. Blocks that show 2 colors represent the patients in which the classification changed over time (Figure 2), in which case the top left triangle represents the classification in the first measurement and the lower right represents the classification in the second measurement. (B) The number of nuclear antigens recognized. Each dot indicates an individual patient, and the bars represent the median. (C) Total ANA-IgG was determined by ELISA. (D and E) Relative affinity of anti-dsDNA IgG was measured in serum of SLE patients by inhibition ELISA. (D) Relative OD (average ± SEM) of each group is shown. (E) IC₅₀ values for each individual patient. (F and G) Principal component analysis of all parameters analyzed (frequencies of ANA⁺ and total B cell and PC subsets as well as ELISA recognition of nuclear antigens). The percentage indicated on the axis is the percentage of variance explained by that principal component. A selection of the vectors for direction of parameters is shown here. The parameters omitted here (ANA⁺ and total transitional and naive B cells) are shown in Figure 1G. Their directionality remained the same. SLE patients were clustered as described in Figure 1. (F) The variables contributing to each dimension in principal component analysis. The length and direction of each arrow shows the strength of their contribution to each PC. (G) The coordinates of each SLE patient. Ellipses represent the 95% confidence interval for each group. *P < 0.05, using Mann Whitney U test. ANA, antinuclear antibody; PC, plasmablast/plasma cell; SS-A/B, anti–Sjögren’s syndrome–related antigen; Sm, Smith antigen; SLE, systemic lupus erythematosus.