Reduced response to SARS-CoV-2 vaccination is associated with impaired immunoglobulin class switch recombination in SLE patients

Study population

Peripheral blood samples for PBMC and serum isolation were collected from healthy donors and from SLE patients attending the University College London Hospital (UCLH) rheumatology outpatients’ clinic. Ethical approval was obtained from the UCLH Health Service Trust ethics committee, under REC reference no. 14/SC/1200. Healthy controls (n=135) and patients (n=93) were recruited following informed consent. Sample storage complied with requirements of the Data Protection Act 1998. Demographics, clinical features, routine laboratory testing and therapeutic regimen were collected retrospectively from electronical medical files. Dates of vaccination and history of SARS-CoV-2 infection were also recorded for healthy controls and SLE patients. Detailed exclusion criteria are described in online supplemental material.

Quantification of RBD-specific, S1-specific, and total IgG, IgA and IgM titres

RBD- and S1-specific direct ELISA protocols were carried out as previously published 13 with modifications. Similarly, RBD-omicron-specific IgG was an adaption of this protocol using an anti-omicron IgG monoclonal antibody. RBD-omicron-specific IgA was determined using a commercial kit (RAS-T099, AcroBiosystems). To measure the total isotype (IgG, IgA, and IgM) immunoglobulin concentration in serum samples, a sandwich ELISA (88-50550-22 for IgG, 88-50600-88 for IgA, 88-50620-88 for IgM, Invitrogen) was performed according to the manufacturer’s instructions. Detailed protocols are described in the supplemental methods.

Blocking assay

A V-PLEX SARS-CoV-2 panel 30 (ACE2) kit (K15635U, Meso Scale Diagnostics) was used according to the manufacturer’s instructions to measure the capacity of serum samples from SLE patients and healthy controls to block the binding of variants of SARS-CoV-2 to the ACE2 receptor. A dilution of 1:50 was used for all serum samples. Individuals with blocking capacity >50% for any tested SARS-CoV-2 strain were considered as neutralisers.

Ex vivo B cell phenotyping

SARS-CoV-2 antigen-specific B-cells were identified using labelled RBD (793604, Biolegend) and S1 (792906, Biolegend) proteins with the addition of antibodies against phenotypic markers and BCR isotype-specific antibodies to measure class-switch status. Full antibody panels and protocols used for ex vivo B cell phenotyping is described in supplemental methods. Stained samples were acquired using a Cytek™ Aurora spectral cytometer, using SpectroFlo (v3.0.3) with automated unmixing. Data were analysed using FlowJo (TreeStar, v.10.8.1).

IFNα CSR in vitro culture

Naive B cells were sorted from previously frozen PBMCs of healthy controls, using a BD FACSAria™ Fusion and BD FACSDiva (v9.4). Full details of cell sorting protocols are described in supplemental methods. Sorted naive B cells from healthy controls were cultured in vitro under IgG-polarising class-switching conditions, consisting of complete RPMI (10% heat-inactivated fetal calf serum, 1% penicillin/streptomycin), 1μg/ml anti-IgM (309-006-043, Jackson ImmunoResearch), 1μg/ml CD40L (ALX-522-110-C010, Enzo Life Sciences), and 50ng/ml recombinant human IFNγ (285-IF-100/CF, Bio-Techne), in a round-bottom 96-well plate at 37°C and 5% CO₂ for 6 days. IgG polarising class-switch medium was refreshed on day 3 by gently centrifuging the cells (300g for 10min) and adding fresh media including the above-mentioned factors. Staining protocols and antibody panels used are described in supplemental methods. Stained samples were acquired using a Cytek™ Aurora spectral cytometer, using SpectroFlo (v3.0.3) with automated unmixing. Data were analysed using FlowJo (Treestar, v10.8.1).

Statistical analysis

The normality of each dataset was determined using the Kolmogorov-Smirnov’s, Anderson-Darling’s, D’AgostinoPearson’s and Shapiro-Wilk’s tests, and the quantile-quantile (q-q) plot. For non-parametric datasets, Wilcoxon matched-pairs signed rank test was used to analyse paired data sets, Mann-Whitney for unpaired datasets, and Kruskal-Wallis test for multivariable datasets. For parametric datasets, a paired or unpaired t-test was used to analyse data and an ordinary one-way ANOVA was used for multivariable datasets. For parametric datasets where variance differed between groups, unpaired t-test with Welch’s correction was used. For non-parametric datasets where SD differed between groups, a log transformation to achieve normal distribution and tests for parametric data were used. Violin plots display the median, along with the 25th and 75th percentiles. Bar plots graphs are expressed as the mean +/- standard error of the mean. Graphing of data and analysis of statistical significance was performed in Prism (GraphPad v9) and R (version v4.2.2). Results were considered significant at *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

Study cohort

Antibody responses to the SARS-CoV-2 vaccine were measured in a cross-sectional cohort of 93 SARS-CoV-2 naive SLE patients. 34 patients were recruited pre-pandemic (no vaccine), and 59 patients post vaccination after either 2 or 3⁺ doses of mRNA-based vaccine (Pfizer or Moderna, n=34); non-mRNA vector-based vaccine (Astrazeneca, n=7); or a combination of two vaccine types (n=15). 135 healthy volunteers were also recruited; 70 pre-pandemic (no vaccine), and 65 after either 2 or 3⁺ doses of mRNA-based vaccine (Pfizer or Moderna, n=57); non-mRNA vector-based vaccine (Astrazeneca, n=3); or a combination of two vaccine types (n=4). A small longitudinal cohort of 18 SLE patients and 22 healthy controls was also included. Further demographic details for the cross-sectional and longitudinal cohorts can be found in tables 1 and supplemental table 1, respectively.

SLE patients show reduced serum SARS-CoV-2 IgG antibody response compared to healthy individuals

The subunit 1 (S1) of the SARS-CoV-2 spike protein interacts with the ACE receptor on the host cell. This subunit includes the ACE receptor binding domain (RBD). Virus-specific (S1 and RBD) vaccine-derived IgG, IgA, and IgM levels (ng/ml) were assessed using direct ELISA. We report high degree of variability of vaccine responses for all antibody isotypes in both healthy controls and SLE patients (Figure 1A-C).

• Download figure
• Open in new tab

Figure 1. SLE patients have impaired humoral response after SARS-CoV-2 vaccination.

(A-C) Individual data bar plots and cumulative data violin plots showing the titres of RBD-specific (A) IgM, (B) IgA, and (C) IgG antibodies in a cross-sectional cohort of SLE patients and healthy controls (HC), pre-vaccination and following 2 or 3⁺ doses of vaccine. (D) Graphs showing the titres of RBD-specific IgM, IgA and IgA antibodies in a longitudinal cohort of SLE patients and healthy controls (HC), between pre-vaccination (0 doses) and 2 or 3⁺ doses of vaccine. (E,F) Individual data bar plots and cumulative data violin plots showing the titres of Omicron (Om)-specific (E) IgA and (F) IgG antibodies in a cross-sectional cohort of SLE patients and healthy controls (HC), pre-vaccination and following 2 or 3⁺ doses of vaccine. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, (A to F) by Mann-Whitney of selected group pairs or by Wilcoxon matched-pairs signed rank test. Violin plots display the median, along with the 25th and 75th percentiles.

Analysis of specific antibody isotypes showed that IgM responses against both RBD and S1 contributed minimally to the spike-reactivity in healthy and SLE cohort at any dose tested (Figure 1A, supplemental Figure 1A).

IgA-specific responses to RBD and S1 were impaired in SLE patients after 2 and 3⁺ doses as shown by reduced IgA median titers (IgA titre median after 2 doses: RBD HC=527.5 vs SLE=424.1; S1 HC=846.5 vs SLE=0; IgA titre median after 3⁺ doses: RBD HC=1145 vs SLE=437.1; S1 HC=1613 vs SLE=999.7) (Figure 1B, supplemental Figure 1B).

Although the titers of anti-RBD and anti-S1 IgG significantly increased following booster doses (3⁺doses) in the majority of SLE patients compared to the levels detected after the second dose, median levels remained significantly lower than those in healthy individuals (IgG titre median after 3⁺ doses: RBD HC=63k vs SLE=26.7k; S1 HC=113.7k vs SLE=53.5k) (Figure 1C, supplemental Figure 1C).

SLE patients are known to present hyper-globulinemia (14) and treatment-derived hypo-globulinemia (15), which might influence vaccine responses. To consider the “weight” that this defect in total antibodies might have on the vaccine-specific Ig response, we calculated the relative levels of antigen-specific IgM/A/G as a percentage of respective total Ig isotype. SLE patients present lower amount of total IgG and IgM and an increase in total IgA levels compared to healthy controls both before and after vaccination (supplemental Figure 2A). When calculated as percentage of the total respective Ig isotype, SLE patients continued to show lower RBD-IgG and IgM responses compared to healthy controls after 2 and 3⁺ doses, (supplemental Figure 2B,D). No significant differences were observed in the percentages of RBD-IgA/total IgA between SLE patients and healthy controls at any of the doses tested (supplemental Figure 2C).

Analysis of the longitudinal cohort confirmed that although the levels of RBD-specific IgA and IgG were significantly increased in SLE patients after vaccination, the medians of IgG and IgA titres were higher in healthy compared to SLE patients even after 3⁺dose (titre median after 3⁺ doses: RBD IgG HC=74.2k vs SLE=37.5k; IgA HC=704 vs SLE=212.1) (Figure 1D). No changes in IgM levels were recorded after vaccination in these samples (Figure 1D).

Several studies have indicated that immunosuppressant-drugs and or severity of disease are associated to weaker SARS-CoV-2 vaccination responses (9,11). No significant associations with antibody response across different treatments (DMARDs, including hydroxychloroquine (HCQ), or prednisolone) were observed at any of the doses tested (supplemental Figure 3A,B). We report that RBD-IgM titres were increased in patients with active disease (BILAG global scoring ≥5) compared to those with inactive disease (BILAG <5) after 3⁺ doses. No associations were found between disease activity and RBD-IgA nor RBD-IgG titres (supplemental Figure 3C).

SARS-CoV-2 vaccination targeting the ancestral strain has demonstrated limited efficacy against the Omicron variant in healthy individuals (16), however, the ability of SLE patients to respond to new variants remains largely unexplored. Omicron RBD-specific IgA (Om-RBD-IgA) titres were undetectable in the majority of SLE patients and healthy controls for any number of doses (Figure 1E). We confirm that IgG titres against the Omicron RBD were generally lower compared to the response against the original strain of SARS-CoV-2 after 2 doses in both healthy controls and SLE patients (Figure 1F). A moderate increase of IgG titres was seen in both groups after 3⁺ doses with healthy controls showing a higher median (Figure 1F, IgG titre median after 3⁺ doses: Om-RBD HC=22.2k vs SLE=12.2k).

SLE patients exhibit impaired neutralising capacity against variants of concern

Next, we assessed the functional neutralisation capacity of circulating antibodies to block the RBD protein of various SARS-CoV-2 variants from binding to the ACE2 receptor by using an MSD assay in a cross-sectional cohort (SLE n=59, HC n=65) (demographic details in supplemental table 2). The analysis included the original strain (CoV.2) and several variants of concern (VOC), comprising Alpha (B.1.1.7), Beta (B.1.351), Delta, and a range of sub-Omicron lineages (BA.2, BA.4, BA.5, BA.2.12.1, BA.2.75). Reflecting the lower amount of anti-IgG production, viral neutralisation, and ratio of neutraliser individuals against all variants were very low in SLE patients after 2 doses (Figure 2A-D). Despite lower levels of antibodies detected in SLE patients compared to healthy controls after 3⁺ doses (represented here as alignment between serum IgG levels and respective neutralisation capacity in Figure 2B), the inhibition capacity for all but one (BA.4;BA.5) strain was not significantly different from healthy controls (Figure 2C). However, when taking into account the ratio of neutralising individuals, SLE patients showed a lower percentage of neutralisers for all strains after 3⁺ doses (HC=93.55% vs SLE=82.35% for CoV-2, HC=90.32% vs SLE=70.59% for B.1.1.7, HC=64.52% vs SLE=52.94% for B.1.351, HC=83.87% vs SLE=64.70% for Delta, HC=38.71% vs SLE=20.59% for BA.2, HC=51.61% vs SLE=29.41% for BA2.12.1 and HC=48.39% vs SLE=26.47% for BA.4/BA.5), but for the strain BA.2.75 (HC=12.9% vs SLE=11.76%) (Figure 2D). Thus, our results reinforce the need for booster doses to increase neutralizing capacity, mainly driven by IgG titres, against VOC in SLE patients to levels comparable to healthy controls.

• Download figure
• Open in new tab

Figure 2. SLE patients show reduced neutralising capacity against SARS-CoV-2 variants of concern after initial vaccination.

(A) Heatmap showing percentage inhibition values against the RBD of different SARS-CoV-2 strains (CoV-2, B1.1.7, B1.351, Delta, BA.2, BA.2.12.1, BA.4;BA.5, and BA.2.75) of serum isolated from a cross-sectional cohort of SLE patients and healthy controls (HC), after 2 or 3⁺ doses of vaccine. (B) Heatmaps showing the titres of RBD-specific IgG for the corresponding individuals in (A). (C) Cumulative data violin plots showing percentage inhibition values against all tested different SARS-CoV-2 strains (CoV-2, B1.1.7, B1.351, Delta, BA.2, BA.2.12.1, BA.4;BA.5, and BA.2.75) of serum isolated from a cross-sectional cohort of SLE patients and healthy controls (HC), after 2 or 3⁺ doses of vaccine. Dotted line on graph indicates the neutralisation threshold (50%). (D) Stacked bar plots showing the percentage of neutralisers for all tested different SARS-CoV-2 strains (CoV-2, B1.1.7, B1.351, Delta, BA.2, BA.2.12.1, BA.4;BA.5, and BA.2.75) within each group. **P<0.01, ***P<0.001, ****P<0.0001, by (C) by Mann-Whitney of selected group pairs or by Wilcoxon matched-pairs signed rank test. Violin plots display the median, along with the 25th and 75th percentiles.

SARS-CoV-2-specific B cell isotype switching is impaired in SLE patients

SLE patients present with several inherent B cell abnormalities including increases in transitional-2, DN B cells, and plasma cells (17–19). We found increased levels of B cells after 2 and 3⁺doses in SLE patients compared to healthy individuals (Figure 3A). We report decreased levels of total memory B cells (MBC; CD27⁺IgD^-CD24⁺CD38^lo) and double positive B cells (DP; CD27⁺IgD⁺), mirrored by an increased in double negative B cells (DN; CD27^-IgD^-) and plasmablasts (CD27⁺IgD^-CD24^-CD38⁺) in SLE patients compared to healthy controls at baseline, differences that were maintained after vaccination (Figure 3A; gating strategy in supplemental Figure 4A,B).

• Download figure
• Open in new tab

Figure 3. SLE patients show no differences in the frequencies of SARS-CoV-2-specific B cell subsets compared to healthy individuals.

(A) Cumulative data violin plots showing frequencies of total B cells (CD19⁺), memory B cells (MBC, CD27⁺IgD^-CD24⁺CD38^lo), double-negative B cells (DN, CD27^-IgD^-), double-positive B cells (DP, CD27⁺IgD⁺) and plasmablasts (CD27⁺IgD^-CD24^-CD38⁺) within the total CD19⁺B cell population in ex vivo peripheral blood of healthy controls (HC) and SLE patients, pre-vaccination (0 dose) or after 2 or 3⁺ doses of vaccine. (B,C) Cumulative data violin plots showing frequencies of total RBD⁺B cells (B) and S1⁺B cells (C) in ex vivo peripheral blood of healthy controls (HC) and SLE patients, pre-vaccination (0 dose) or after 2 or 3⁺ doses of vaccine. (D,E) Cumulative data violin plots showing frequencies of RBD⁺ (D) and S1⁺ (E) MBC, DN B cells, DP B cells and plasmablasts, within the total CD19⁺B cell population in ex vivo peripheral blood of HC and SLE patients, pre-vaccination (0 dose), or after 2 or 3⁺ doses of vaccine. (*P<0.05, **P<0.01 ***P<0.001, ****P<0.0001, by unpaired t-test, Mann-Whitney test or data transformation according to data distribution and features (see methods). Violin plots display the median, along with the 25th and 75th percentiles.

To understand whether changes in B cell subsets frequency observed in SLE patients, impacted the response to SARS-CoV-2 vaccination, we measured RBD and S1-specific memory B cells, DN, DP, and plasmablasts frequencies in SLE patients and healthy controls (demographics in supplemental table 3; gating strategy in supplemental Figure 4A,B). There was no difference in the frequencies of total RBD⁺ and S1⁺ B cells between SLE patients and healthy controls (Figure 3B,C). Analysis of antigen-specific B cell subsets, amongst total B cells, show that the frequencies of RBD⁺ and S1⁺MBCs in SLE patients were decreased (a trend for RBD and significant for S1) compared to healthy controls even after 3⁺ doses of vaccine (Figure 3D,E). We did not observe any differences between healthy controls and SLE patients in the frequencies of RBD⁺ or S1⁺DN B cells, DP B cells or plasmablasts (Figure 3D,E).

Next, we assessed how repeated vaccinations affects the frequencies of memory unswitched (IgM⁺) or memory switched (IgG⁺/IgA⁺) RBD⁺ and S1⁺ B cells (Figure 4A). There was an increased frequency of RBD⁺/S1⁺ IgM⁺unswitched MBCs following 2 and 3⁺ doses in SLE patients (Figure 4B,C), and an increase in and RBD⁺/S1⁺ IgM⁺ DN B cells after 2 doses (supplemental Figure 4C,D). This was accompanied by a decrease in RBD⁺/S1⁺ IgG⁺ MBCs (Figure 4D,E), and RBD⁺/S1⁺ IgG⁺ DN B cells (supplemental Figure 4C,D) after 2 or 3⁺ doses in SLE patients compared to healthy controls. We also reported increased frequencies of RBD⁺/S1⁺ IgM⁺ early plasmablasts (early-PB) (CD27⁺IgD^-CD24^midCD38⁺) in SLE patients compared to healthy controls (supplemental Figure 4E,F), previously shown to arise from extrafollicular responses (17–19).

• Download figure
• Open in new tab

Figure 4. SARS-CoV-2-specific B cells show lower IgG and IgA class-switching in SLE patients.

(A) Representative contour plots showing the gating strategy for identification of memory B cells (MBCs; CD27⁺IgD^-CD24⁺CD38^lo), early plasmablasts (CD27⁺IgD^- CD24^midCD38⁺), and plasmablasts (CD27⁺IgD^-CD24^-CD38^hi), within RBD⁺ and S1⁺ B cell subsets. IgM⁺ unswitched (CD27⁺IgD^-CD24⁺CD38^loIgM⁺) and IgG⁺ switched (CD27⁺IgD^-CD24⁺CD38^loIgM^-IgG⁺) or IgA⁺ switched (CD27⁺IgD^-CD24⁺CD38^loIgM^- IgA⁺) MBCs were identified within RBD⁺ and S1⁺ B cell subsets. (B,C) Representative contour plots and cumulative data violin plots showing the frequencies of (B) RBD-specific and (C) S1-specific IgM⁺ unswitched MBCs (CD27⁺IgD^-CD24⁺CD38^loIgM⁺) in ex vivo peripheral blood of healthy controls (HC) and SLE patients, after 2 or 3⁺ doses of vaccine. Values in gates indicate the percentage within CD19⁺ cells. (D,E) Representative contour plots and cumulative data violin plots showing the frequencies of (D) RBD-specific and (E) S1-specific IgG⁺ and IgA⁺ switched MBCs (CD27⁺IgD^- CD24⁺CD38^loIgM^- IgG⁺/IgA⁺), in ex vivo peripheral blood of healthy controls (HC) and SLE patients after 2 or 3⁺ doses of vaccine. Values in gates indicate the percentage within CD19⁺ cells. *P<0.05, **P<0.01 ***P<0.001, ****P<0.0001, by Mann-Whitney test or data transformation according to data distribution and features (see methods). Violin plots display the median, along with the 25th and 75th percentiles.

The results for RBD⁺IgA⁺ MBCs were difficult to interpret due to very low frequencies of RBD⁺IgA⁺ B cells (Figure 4D). Nevertheless, S1⁺IgA⁺MBCs were significantly reduced in SLE patients after 2 and 3⁺ doses, compared to healthy controls (Figure 4E). No differences were observed between SLE patients and healthy controls in the frequencies of IgA⁺DN B cells for both RBD⁺ /S1⁺ and populations (supplemental Figure 4C,D).

The accumulation of antigen-specific IgM⁺ early-PB in SLE patients suggest a defect in CSR supported by an extrafollicular differentiation pathway. We calculated the ratio between RBD⁺/S1⁺ switched (IgA or IgG) MBCs and RBD⁺/S1⁺ IgM⁺ early-PB as an index (CSR ratio) of efficient CSR compared to extrafollicular responses. SLE patients showed a significantly lower CSR ratio compared to healthy controls (supplemental Figure 5A,B). A similar CSR impairment was also observed in non-antigen-specific (S1^-) B cells (supplemental Figure 5C-F), suggesting that this defect may not be limited to vaccine-derived antigen-specific B cells.

IFNα impairs class-switch recombination in vitro and promotes IgM⁺ early plasmablasts

Our findings and those from others (20) suggest that SLE patients exhibit impaired CSR and increased frequencies of vaccine-specific IgM⁺ MBCs and IgM⁺ early-PBs. Given the pivotal role of IFNα in SLE pathogenesis and in driving plasma cell differentiation (21,22), we hypothesised that exposure to elevated IFNα levels may contribute to the observed CSR imbalance in SLE patients.

We took advantage of a previously established culture system whereby healthy naive B cells were stimulated in vitro for 6 days with a combination of anti-IgM, CD40L, and IFNγ to induce class-switching to IgG isotypes (23). Increasing concentrations of IFNα were added to the culture system to model the potential effect of IFNα on B cell differentiation and CSR of SLE B cells (Figure 5A). As expected, IFNα reduced memory B cells whilst promoting an increase of plasmablasts in vitro (Figure 5B,C). In addition, high IFNα levels reduced switching of healthy naive B cells to IgG1, IgG2 and IgG3, under CSR-IgG-polarising conditions (Figure 5D,E). Supporting our ex vivo findings in SLE patient B cells, we observed a significant increase of IgM⁺ early-PB frequencies and a decreased ratio of IgG⁺ B cells to IgM⁺ early-PBs (CSR ratio) under these polarizing in vitro conditions (Figure 5F-H). Together this in vitro modelling of SLE B cell CSR indicates that high concentrations of circulating IFNα in SLE patients may contribute to the impaired CSR of B cells and subsequently result in weaker antibody responses following SARS-CoV-2 vaccination.

• Download figure
• Open in new tab

Figure 5. Interferon alpha suppresses healthy naive B cell class switching to IgG.

(A) Schematic of class-switching in vitro experiment. Naive B cells (CD27^-IgD⁺) were isolated from previously frozen PBMCs of healthy controls. Isolated naive B cells were cultured under IgG class switching polarising conditions (anti-IgM, CD40L and IFNγ) together with 0, 500, or 50000 IU/ml of IFNα for 6 days. After 6 days of culture, cells were harvested and stained for flow cytometry to assess their phenotype and CSR status. (B) Contour plots showing the gating strategy for the identification of memory B cells (CD19⁺CD24⁺CD38^lo), mature naive B cells (CD19⁺CD24^midCD38^mid), transitional B cells (CD19⁺CD24⁺CD38⁺), early plasmablasts (CD19⁺CD24^midCD38⁺) and plasmablasts (CD19⁺CD24^-CD38⁺) after 6 days in IgG-polarising class-switch stimulation of isolated healthy naive B cells (CD27^-IgD⁺). (C) Representative contour plots and cumulative data bar graphs showing the frequencies of memory B cells (CD19⁺CD24⁺CD38^lo) and plasmablasts (CD19⁺CD24^-CD38⁺) after 6 days stimulation of isolated healthy naive B cells under IgG-polarising conditions and increasing concentrations of recombinant IFNα. Values in gates indicate the percentage within CD19⁺ cells. (D) Contour plots showing the gating strategy for the identification of IgG1⁺, IgG2⁺ and IgG3⁺ switched B cells after 6 days stimulation of isolated healthy naive B cells under IgG-polarising conditions. (E) Representative contour plots and cumulative bar graphs showing the frequencies of IgG1⁺, IgG2⁺ and IgG3⁺ switched B cells, after 6 days stimulation of isolated healthy naive B cells under IgG-polarising conditions and increasing concentrations of recombinant IFNα. Values in gates indicate the percentage within CD19⁺IgD^-IgM^- cells. (F) Contour plots showing the gating strategy for the identification of IgM⁺ memory, IgM⁺ mature naive, IgM⁺ transitional B cells, IgM⁺ early plasmablasts and IgM⁺ plasmablasts after 6 days in IgG-polarising class-switch stimulation of isolated healthy naive B cells. (G) Representative contour plots and cumulative bar graphs showing the frequencies of IgM⁺ early plasmablasts (CD19⁺CD24^midCD38⁺ IgD^-IgM⁺) after 6 days stimulation of isolated healthy naive B cells under IgG-polarising conditions and increasing concentrations of recombinant IFNα. Values in gates indicate the percentage within CD19⁺ cells. (H) Cumulative data bar graph showing the ratio between IgG⁺ switched memory B cells (sum of IgG1⁺, IgG2⁺ and IgG3⁺ frequencies within CD19⁺IgD^-IgM^- cells) and IgM⁺ early plasmablasts, after 6 days stimulation of isolated healthy naive B cells under IgG-polarising conditions and increasing concentrations recombinant IFNα. *P<0.05, **P<0.01, by Kruskal-Wallis test with Dunn’s multiple comparison. Error bars in bar plots show mean+/-SEM.