Frequency of CD19⁺CD24^hiCD38^hi regulatory B cells is decreased in peripheral blood and synovial fluid of patients with juvenile idiopathic arthritis: a preliminary study

A total of 32 patients from the Division of Rheumatology of Children’s National Medical Center were recruited in this study, including 21 JIA patients (13 poly-JIA, 5 oligo-JIA, 2 systemic, 1 psoriatic) and 11 children with growing pain but no known rheumatic diseases as controls. JIA patients were diagnosed according to the ILAR criteria [2]. Growing pain was diagnosed after known diseases were excluded with negative immunologic test findings. Peripheral blood (PB) samples were collected from the JIA patients and controls. One patient was followed longitudinally and PB samples were collected at both active and inactive phase. Synovial fluid (SF) samples were collected from 4 JIA patients who required intra-articular steroid injection as a part of treatment protocol. Of these subjects, both PB and SF samples were collected from 1 patient on the same day. Disease activity was assessed and inactive disease was defined according to Wallace’s criteria [35]. Patients who didn’t meet the definition for inactive disease were defined as having active disease. Demographic and clinical data of the patients were collected. The study was conducted in compliance with the Helsinki Declaration and ethical approval was obtained from the Institution Review Board of Children’s National Medical Center (Pro00005055). All patients were enrolled after obtaining informed consent from parents and assent from patients older than 7 years old.

Peripheral blood mononuclear cells (PBMCs) and synovial fluid mononuclear cells (SFMC) were isolated from heparin-treated PB and SF by Ficoll-Paque Plus (GE Healthcare, Uppsala, Sweden) gradient centrifugation. PBMCs and SFMCs were cultured in RPMI 1640 containing L-glutamine (Life Technologies, Paisley, UK) supplemented with 100 U/μg/ml penicillin/streptomycin (Life Technologies, Paisley, UK), and 10% fetal bovine serum in 48-well flat-bottom plates for 48 h at 37 °C in 5% CO₂. According to previous study [12], combination of CpG and CD40L stimulation could generate the most of B10 cells in human. Therefore, the cultured cells were stimulated with 10 μg/ml CpG ODN2006 (Invivogen, San Diego, USA) and 1 μg/ml CD40L (R&D Systems, Minneapolis, USA), or with phosphate-buffered saline (PBS) as control. For the last 6 h, 50 ng/ml phorbol myristate acetate (PMA) and 1 μg/ml ionomycin (Sigma-Aldrich, USA) were added to the stimulated cells; Brefeldin A (BFA, eBioscience, San Diego, USA), Golgi transport blocker, was added to all wells.

The following anti-human monoclonal antibodies (mAbs) were used for surface markers and intracellular IL-10 detection: fluorescein isothiocyanate (FITC)-conjugated anti-CD19; phycoerythrin (PE)-conjugated anti-CD24 (BD Biosciences, San Diego, USA); PE-Cyanine7-congugated anti-CD38; allophycocyanin (APC)-conjugated anti-IL-10 (Biolegend, San Diego, USA); and isotype-matched and fluorochrome-matched control antibodies. Cells were stained with combinations of CD19-FITC, CD24-PE and CD38-PE-Cy7 mAbs for surface phenotype. For intracellular IL-10 detection, cultured cells were washed, fixed, permeabilized, and stained with IL-10-APC mAb. APC-conjugated isotype control was used for gate setting for cytokine expression. Stained cells were analyzed on an eight-color FACSCanto II flow cytometer (BD Biosciences) using FACSDiva software (BD Biosciences).

Statistical analyses were performed using GraphPad Prism Version 6 (GraphPad Software, La Jolla, CA, USA). Chi-square tests were performed for discrete variables. Student t test was used for parametric test when comparing two groups with equal variances. Welch’s t test was used for parametric test when comparing two groups with unequal variances. Mann-Whitney U-test was used for non-parametric test when comparing two groups. Spearman’s correlation was performed to determine correlation. A p value of < 0.05 was considered statistically significant.

The demographic and clinical features of 21 JIA patients and 11 controls are summarized in Table 1. There was no significant difference between the 2 groups with the exception that none of the control subjects were positive for rheumatoid factor. Table 2 shows the demographic and clinical features of 4 JIA patients from whom SF samples were collected.

Peripheral blood and synovial fluid mononuclear cells from subjects were phenotypically analyzed by flow cytometry for their expression of CD19, CD24, and CD38 surface markers. B cells were defined as CD19⁺ lymphocytes. Within CD19⁺ B cells gate, CD24^hiCD38^hi cells were defined as CD24^hiCD38^hi Bregs. The gate strategy for CD19⁺CD24^hiCD38^hi cells was illustrated by a representative staining of cells in a healthy control subject (Fig. 1a). The percentages of CD24^hiCD38^hi Bregs subset were calculated as the ratios of gated targeted cells to total CD19⁺ B cells. The results were expressed as mean values ± standard deviation (SD).

As shown in Fig. 1b, we observed a significant decrease in the levels of CD24^hiCD38^hi Bregs in the PB of JIA patients compared to those in controls (16.11 ± 1.09% vs. 23.83 ± 1.73%, p < 0.001). An even more significant decrease was seen in the SF of JIA patients compared to those in the PB (3.23 ± 1.92% vs. 16.11 ± 1.09%, p < 0.0001). The data of PB and SF samples from the same patient supported this result. The percentage of CD24^hiCD38^hi Bregs in the SF was lower than that in the PB of the same patient collected on the same day (1.41% vs. 13.00%) (Fig. 1c). Of note, there was no significant difference of B cells percentages in PB between JIA patients and controls (15.22 ± 1.28% vs. 16.35 ± 1.22, p = 0.5788). However, B cells population in SF was significantly decreased compared with PB (1.47 ± 0.27% vs. 15.22 ± 1.28%, p = 0.0001). Probably due to this decrease, CD24^hiCD38^hi Bregs subset was almost absent in SF mononuclear cells (SFMCs). In peripheral blood mononuclear cells (PBMCs) of JIA patients, the percentage of CD24^hiCD38^hi Bregs was 2.41 ± 0.27%, while this percentage was 0.04 ± 0.02% in SFMCs, which was highly significant (p = 0.0004).

We next examined the frequencies of CD24^hiCD38^hi Bregs in poly-JIA and non-poly-JIA patients. As in total JIA patients, the frequencies of CD24^hiCD38^hi Bregs in poly-JIA and non-poly-JIA patients were significantly decreased compared with controls (p < 0.01 and p < 0.05, respectively) (Fig. 1b). No significant difference of frequency of CD24^hiCD38^hi Bregs was observed between poly and non-poly JIA patients (p = 0.6177).

Next, we examined whether CD24^hiCD38^hi Bregs level correlated with disease activity or treatment. Treatments for all patients are shown in Table 1. As shown in Fig. 2a, there was no significant difference in the CD24^hiCD38^hi Bregs levels between active JIA and inactive JIA patients (p = 0.6238). In addition, no difference of CD24^hiCD38^hi Bregs level was observed between patients with and without methotrexate(MTX) or TNF antagonist treatment (p = 0.1358, p = 0.1469, respectively) (Table 3).

We further examined the possible correlations between the frequencies of CD24^hiCD38^hi Bregs and laboratory parameters. As shown in Fig. 2b, a significantly lower frequency of CD24^hiCD38^hi Bregs was found in the PB of RF-positive patients than in RF-negative patients (10.81 ± 1.80% vs. 17.11 ± 1.14%, p < 0.05). However, no significant correlation was found between the frequencies of CD24^hiCD38^hi Bregs and ESR (Spearman’s r = − 0.5108, p = 0.0519) (Fig. 2c).

As a subgroup of Bregs, B10 cells level was also examined in PB and SF of JIA patients and PB of controls. By ex vivo stimulation of CpG + CD40L for 48 h and PMA + ionomycin for the last 6 h, intracellular production of IL-10 by CD19⁺ B cells was observed, while few B cells produced IL-10 when cultured with PBS only (Fig. 3a). A representative experiment showing intercellular IL-10 staining of B cells in a JIA patient is shown in Fig. 3a.

Unlike the remarkable reduction of CD24^hiCD38^hi Bregs levels in JIA patients, there was no significant difference of B10 cells levels between the PB of JIA patients and controls (7.43 ± 0.99% vs. 4.56 ± 0.71%, p = 0.0527) (Fig. 3b). In SF of JIA patients the B10 cells level was not significantly different from that in PB (4.32 ± 1.14% vs. 7.43 ± 0.99%, p = 0.2419) either. However, when we examined B10 cells frequency in the poly-JIA subgroup, we noticed a significant increase compared with controls (p = 0.0249) (Fig. 3b).

We then examined whether the frequencies of B10 cells correlate with disease activity. Significantly lower frequency of B10 cells was found in active JIA patients compared with that in inactive patients (5.77 ± 1.13% vs. 9.72 ± 1.49%, p < 0.05), while no significant difference was found between active patients and controls (5.77 ± 1.13% vs. 4.56 ± 0.71%, P = 0.3762) (Fig. 3c). One patient was followed longitudinally and the same trend was noted: the percentage of B10 cells increased from 2.93% when active and increased to 5.96% when inactive, supporting the association between B10 level and JIA disease activity.

Same trend was also found in poly-JIA patient. As shown in Fig. 3c, when we subdivided poly-JIA patients into active and inactive subgroups, we observed a significantly lower frequency of B10 cells in the active poly-JIA subgroup compared with inactive poly-JIA subgroup (6.00 ± 1.48 vs. 11.87 ± 1.56, p = 0.0239), while no significant difference was found between B10 cells frequencies in the active poly-JIA subgroup and controls (p = 0.3531).

When we compared the frequencies of B10 cells with different treatment groups, no significant difference was observed between groups with and without MTX or TNF antagonist (p = 0.8408, p = 0.3683, respectively) (Table 3).

Since B10 cells are a specific subgroup of Bregs, we examined whether there was a correlation between the level of CD24^hiCD38^hi Bregs and the level of B10 cells in JIA patients. As shown in Fig. 2d, no notable correlation was observed (p = 0.7438). Also, there was no correlation between the level of B10 cells and ESR level (p = 0.3073) (Fig. 2e). No difference of levels of B10 cells was found between RF-positive and RF-negative patients (8.07 ± 2.48% vs. 7.50 ± 1.52%, p = 0.8494) (Fig. 2f).