Phenotypic and functional characterization of switch memory B cells from patients with oligoarticular juvenile idiopathic arthritis

Introduction In chronic inflammatory disorders, B cells can contribute to tissue damage by autoantibody production and antigen presentation to T cells. Here, we have characterized synovial fluid and tissue B-cell subsets in patients with oligoarticular juvenile idiopathic arthritis (JIA), an issue not addressed before in detail. Methods B cells from synovial fluid (SF) and peripheral blood (PB) of 25 JIA patients, as well as from PB of 20 controls of comparable age, were characterized by multicolor flow cytometry. Immunoglobulin-secreting cells were detected by ELISPOT. Immunohistochemical analyses of synovial tissue from three JIA patients were performed. Results JIA SF B cells were enriched in CD27+ and CD27- switch memory B cells, but not in CD27+ IgM memory B cells, compared with patient and control PB. Plasma blasts were more abundant in SF and secreted higher amounts of IgG. Lymphoid aggregates not organized in follicle-like structures were detected in synovial tissue sections and were surrounded by CD138+ plasma cells. Finally, transitional B cells were significantly increased in JIA PB versus SF or control PB. CCR5, CCR8, CXCR2, and CXCR3 were upregulated, whereas CCR6, CCR7, and CXCR5 were downregulated on SF CD27+ and CD27- switch memory B cells compared with their circulating counterparts. SF CD27+ and CD27- switch memory B cells expressed at high levels the costimulatory molecule CD86 and the activation marker CD69. Conclusions This study demonstrates for the first time an expansion of activated switch memory B cells and of IgG-secreting plasma blasts in the SF from oligoarticular JIA patients. Memory B cells belonged to either the CD27+or the CD27- subsets and expressed CD86, suggesting their involvement in antigen presentation to T cells. Patterns of chemokines-receptor expression on CD27+ and CD27- switch memory B cells delineated potential mechanisms for their recruitment to the inflamed joints.


Introduction
Juvenile idiopathic arthritis (JIA) is a heterogeneous condition classified into different subtypes according to the symptoms at onset [1]. Oligoarticular JIA is the most frequent form (26% to 56% of all JIA) and is characterized by early disease onset, asymmetric arthritis, high prevalence of iridocyclitis, peculiar HLA association (HLA-DRB1*1101, DRB1*0801, DPB1*0201), and the presence of antinuclear antibodies. In the majority of these patients, the disease remains confined to a limited number of joints (persistent oligoarticular JIA) and has a favorable outcome characterized by a high frequency of selfremission (as reviewed in [2]). Approximately one third of patients with oligoarticular onset experience progression toward a more aggressive form, characterized by the involvement of five or more joints after the first 6 months of disease (extended oligoarticular JIA). In 10% to 30% of JIA patients, the disease shows symmetric involvement of more than four joints, with an erosive course during the first 6 months of disease (polyarticular-onset JIA). A small proportion of these patients (3% to 5% of all JIA patients) display positivity for rheumatoid factor (RF) [2]. The systemic-onset JIA is observed in 4% to 17% of patients and is characterized by a severe systemic involvement (rash, fever, hepatosplenomegaly) associated with arthritis of variable severity that may evolve into an aggressive polyarticular course [2].
A distinctive feature of chronic inflammatory arthritides is the presence of synovial lymphocytic infiltrates that play a role in disease pathogenesis through the release of pro-inflammatory cytokines and other soluble mediators [3][4][5].
In adult rheumatoid arthritis (RA) and occasionally in JIA [6], these infiltrates may organize into follicle-like structures, according to a process known as "ectopic lymphoid neogenesis" [3][4][5]. Both T and B cells are detected in JIA infiltrates [6]. Whereas T cells are likely responding to autoantigens whose nature has been partially defined [7,8], the pathogenic role of B cells in JIA is less clear, because the vast majority of patients test negative for rheumatoid factors [2]. Nonetheless, synovial lymphocytic infiltrates have been recently correlated with the presence of serum anti-nuclear IgG antibodies in JIA patients [6]. The latter observation highlights the relevance of switch memory B cells in the production of these autoantibodies.
Furthermore, activated switch memory B cells can contribute to the pathogenesis of JIA by upregulating the expression of co-stimulatory molecules such as CD80 and CD86 and presenting antigens to T cells [9].
With this background, we here address the immunophenotypic and functional characterization of synovial B cells from JIA patients, with emphasis on switch memory B cells. The results obtained may be translationally relevant because RA patients can benefit from treatment with rituxan (Rituximab), a monoclonal antibody directed to the B cell-specific antigen CD20 [10,11], and preliminary evidence indicates that the same treatment may be efficacious in JIA patients [12].

Patients
This investigation was approved by the Ethical Committee of the G. Gaslini Institute, Genoa, Italy. All biologic samples (blood, synovial fluid, or synovial tissue) from juvenile idiopathic arthritis (JIA) patients or healthy controls were obtained with informed consent of the patients' parents or the legal guardians. JIA individuals were classified according to ILAR Durban criteria [1]. Twenty-three of 25 patients had oligoartic-ular JIA, either persistent or extended, and two had RF-polyarticular JIA. All patients were in articular relapse at study. An intra-articular steroid injection in the previous 6 months was considered an exclusion criterion. The clinical characteristics and ongoing treatment are reported in Table 1. The two RFpolyarticular JIA cases were included in the extended JIA group for statistical purposes, because of the limited number of patients studied.

Cell isolation
Mononuclear cells (MNCs) were isolated with Ficoll-Hypaque density gradients (Sigma Chemical Company, St. Louis, MO) from synovial fluid (SF) and peripheral blood (PB) of 25 patients with JIA or 20 healthy individuals, comparable with respect to mean age (no clinical or laboratory evidence for inflammatory or infectious disorders in the 4 weeks before testing). MNCs were frozen in a solution containing 10% dimethyl sulfoxide (Sigma), and stored in liquid nitrogen until tested. Cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (Sigma).
The gates and the marker combinations used to analyze B-cell subpopulations are detailed later. In all flow-cytometry experiments performed in this study, the first step was the exclusion of non-B cells stained with a combination of the following antibodies and detected in a single channel: CD3, CD14, CD16, and CD56, all labeled with APC-Cy7.
Naïve B cells were detected as IgD + , IgM + cells after gating on CD19 + cells. CD27 + IgM memory B cells were detected as IgG -, IgAcells after gating on CD19 + cells and subsequently on CD27 + cells. CD27 + switch memory B cells were detected as IgG + , IgA + cells after gating on CD19 + cells and subsequently on CD27 + cells. CD27switch memory B cells were detected as IgG + , IgA + cells after gating on CD19 + cells and subsequently on CD27cells. Transitional B cells were detected as IgM + , IgD + cells after gating on CD19 + cells and subsequently on CD24 high , CD38 high cells. Germinal centerlike B cells were detected as CD10 + cells after gating on CD19 + cells. Plasma blasts were detected as CD27 high , CD20 +/cells after gating set on CD19 + cells.
nic diaminobenzidine substrate was applied, and slides counterstained with Mayer's hematoxylin.

ELISPOT assay
Cells secreting IgG, IgM, or IgA were detected with an ELIS-POT assay, as reported [13], by using purified goat antihuman IgG or IgA or IgM from Southern Biotechnology Associates (Birmingham, AL, USA). After washing and blocking with PBS containing 1% BSA for 30 minutes, serial dilutions of cultured B cells were added and incubated overnight at 37°C. Before plating, cultured B cells were washed 5 times with complete medium to eliminate the Ig present in the supernatants. Plates were washed and incubated with isotype-specific secondary antibodies, followed with streptavidin-HRP (Sigma). The assay was developed with AEC (Sigma) as the chromogenic substrate. ELISPOT plates were analyzed blindly with the bioreader 3000 BIOSYS.

Statistical analysis
Data were reported in terms of medians, minimum and maximum values, or first and third quartiles. The nonparametric analysis of variance (Kruskal-Wallis test) was used to compare quantitative parameters among three groups of observations (for example, to compare the number of naïve B cells in SF with respect to PB and to controls' PB); the Dunn test was used as an a posteriori test. The Mann-Whitney U test was used to compare quantitative variables between two groups of observations, and the Bonferroni correction was applied to avoid multiple comparisons error (P B ) (for example, to compare number of CCR5 + , CCR6 + , CCR7 + , CCR8 + , and CCR9 + cells in the SF versus the PB compartment). A P value less than 0.05 was considered statistically significant. Statistical analyses were performed by using Graph Pad Prism 3 software and the statistical package, Statistica 6 (StatSoft Inc., Tulsa, OK, USA). Naïve (CD19 + IgD + CD27 -) B cells [14] were less abundant in SF than in patient and control PB (Dunn test; P < 0.01), whereas they were equally represented in control and JIA PB ( Figure 1A). B cells with a germinal center-like phenotype (CD19 + , CD10 + cells) [15] were virtually absent from both SF andpatient or control PB (data not shown).  Figure 1B). The same cell subset was similarly represented in control and patient PB ( Figure 1B).

B-cell subset characterization in synovial fluid from oligoarticular JIA patients
Recently, a novel subset of switch memory B cells lacking CD27 expression (CD19 + CD27 -IgG + IgA + ) was identified [17][18][19]. CD27switch memory B cells were more abundant in SF than in patient PB (Dunn test; P < 0.01), whereas they were present in similar proportions in control and patient PB ( Figure 1C).
CD19 + CD24 high CD38 high IgM high IgD high transitional B cells [20] were virtually absent in SF compared with patient and control PB (Dunn test; P < 0.01) ( Figure 2) and were more abundant in patient than in control PB (Dunn test; P = 0.05) ( Figure 2). It has been reported that transitional B cells can express CD10 [21]. To address this issue, we stained transitional B cells from patient PB for CD10 and found that the latter marker was expressed by these cells (15% to 20%; range from four experiments with different samples) ( Figure 2). Patients with oligoarticular persistent JIA who were receiving NSAID treatment or were untreated at study did not show significant differences in the percentages of all the mentioned Bcell subsets (naïve B cells, CD27 + and CD27switch memory B cells, and plasma blasts), either in SF or in PB. Similar results were obtained from the comparison of patients with persistent or extended oligoarticular JIA (not shown).
CD19 + CD27 high CD20 +/plasma blasts [22] were increased in SF versus patient PB (Dunn test; P < 0.05) ( Figure 3A). The percentage of plasma blasts in control PB and JIA PB was similar ( Figure 3A). The percentage of plasma blasts was higher in SF from extended than from persistent oligoarticular JIA patients (P = 0.014).
ELISPOT experiments showed higher numbers of CD19 + IgGsecreting cells in SF versus patient PB (P = 0.028) ( Figure  3B). CD19 + IgA-and IgM-secreting cells were equally represented in patient SF and PB ( Figure 3B). Similar percentages of IgG-, IgM-, or IgA-secreting B cells were found in patient and control PB (not shown). Immunohistochemical staining for IgG, IgA, and IgM in synovial tissue sections showed a predominance of IgG-secreting cells. Lower proportions of IgA-and IgM-secreting cells were also detected ( Figure 3C).
Chemokine receptor and costimulatory molecule expression in CD27+ and CD27-switch memory B cells from synovial fluid of oligoarticular JIA patients Next, CD27 + and CD27switch memory B cells were characterized with flow cytometry for the expression of CC chemokine receptors (R) (CCR1-CCR9) and CXCR (CXCR1-CXCR5), as well as for the CD80 and CD86 costimulatory molecules.
CD27 + switch memory B cells CCR5 + , CCR8 + , and CCR9 + cells were significantly increased, whereas CCR6 + and CCR7 + cells were decreased in SF versus patient PB ( Figure 4A and Table 2). CXCR1 + , CXCR2 + , and CXCR3 + cells were significantly more abundant, whereas CXCR5 + cells were less numerous in SF than in patient PB cells ( Figure 4B and Table 2). No difference in CXCR4 expression was observed between SF and patient PB cells.
Finally, CCR and CXCR expression was similar in patient and control PB cells (not shown). SF cells expressed higher levels of the CD86 costimulatory molecule than did patient PB cells (P = 0.0001) ( Figure 4B). In contrast, CD80 expression was similar (SF CD80 median, 7.5; range, 2.0 to 21.9; n = 15; PB CD80 median, 15.2; range, 13.1 to 21.4; n = 10). Likewise, CD86 and CD80 expression on cells from patient and control PB (not shown) was comparable.
To achieve insight into the activation state of CD27 + switch memory B cells, the latter cells from SF and PB were stained with mAbs to CD25, CD69, and HLA-DR. Percentages of CD69 + and CD25 + cells from SF were significantly higher than those in PB (SF CD69 median, 12.9; range, 8. CD27switch memory B cells CCR5 + , CCR8 + , and CCR9 + cells were significantly increased, whereas CCR6 + and CCR7 + cells were decreased in SF versus patient PB ( Figure 4C and Table 3). CXCR2 + and CXCR3 + cells were significantly more numerous, whereas CXCR5 + cells were less abundant in SF than in patient PB cells ( Figure 4D and Table 3). CXCR4 expression was similar in SF and patient PB cells (not shown). Few CXCR1 + cells were detected in SF and patient PB (not shown). Expression of all CCR and CXCR was similar in control (data not shown) and patient PB cells.

Characterization of B-cell infiltrates in synovial tissue from JIA patients
In three JIA cases (one with oligoarticular persistent and two with oligoarticular extended JIA), histologic analysis of synovial tissue sections demonstrated the presence of lymphoid aggregates with predominant perivascular distribution [6]. Clusters of CD20 + B cells ( Figure 5A) and CD3 + T cells (not shown) were detected within lymphoid aggregates, whereas CD138 + plasma cells were found at the periphery of such aggregates ( Figure 5A and 5B). Consistent with a previous study [6], these aggregates were not organized in follicle-like structures, because they tested negative for the expression of the follicular dendritic cell marker CD21 ( Figure 5C).
CD27 + cells were detected both within and around lymphoid aggregates, consistent with CD27 expression by memory B cells and plasma blasts/plasma cells [23], as well as by memory T cells [24] ( Figure 5D).

Discussion
In this study, CD27 + and CD27switch memory B cells from oligoarticular JIA patients were found to be significantly enriched in SF compared with PB, whereas CD27 + IgM + B cells were equally represented in the two compartments.
A previous study showed that CD27switch memory B cells from normal subjects and systemic lupus erythematosus (SLE) patients had substantial levels of Ig mutations but lower than conventional CD27 + switch memory B cells [18]. The former B cells expanded in the peripheral blood from SLE patients, and this expansion correlated with high disease activity and high titers of disease-specific autoantibodies [18]. In contrast, the frequency of CD27switch memory B cells was reported to be normal in the peripheral blood from adult RA patients [18]. Accordingly, in our study, these cells were detected in similar proportions in the peripheral blood of oligoarticular JIA patients and controls.
CD27switch memory B cells from both PB and SF of our patients did not express the FcRH4 surface marker, as reported in SLE patients [18]. FcRH4 -CD27switch memory B cells may be more responsive to activation and may expand more easily in autoimmune diseases than their CD27 -, FcRH4 + counterparts, because FcRH4 is a potent inhibitor of B-cell signaling [25].
All patients tested had an articular disease relapse. No significant differences in the proportions of most SF or PB B-cell subpopulations were detected in patients subdivided accord- Table 2 Chemokine-receptor expression in synovial fluid and peripheral blood CD27 + switch memory B cells ing to ongoing treatment or disease subtype (i.e., persistent versus extended oligoarticular JIA). The only exception was represented by plasma blasts that were significantly more numerous in SF from extended than from persistent oligoarticular JIA patients.
To gain more insight into the mechanisms of CD27 + and CD27switch memory B cell recruitment to the inflamed joints, we investigated their expression of a panel of chemokine receptors. CCR5, CCR8, CCR9, CXCR2, and CXCR3 were upregulated, whereas CCR6, CCR7, and CXCR5 were down-regulated on both SF memory B-cell subsets versus their PB counterparts. In contrast, CXCR1 was found to be upregulated in SF CD27 + but not CD27switch memory B cells, although the mechanisms underlying such a difference are unknown.
These findings suggest that CD27 + and CD27 -switch memory B cells from oligoarticular JIA patients share a common set of chemokine receptors, likely mediating their attraction to the affected joints, where the respective chemokine ligands [28] are abundantly produced as a consequence of chronic inflammation [29][30][31]. Unfortunately, the chemotactic functionality of In adulthood, transitional B cells generated in the bone marrow are released into the bloodstream and transported to the spleen, where they develop into long-lived mature B cells [20,32]. Here, we demonstrated that transitional B cells were virtually absent from the synovial fluid of JIA patients, whereas they were detected in paired blood samples and found to be significantly increased in comparison with control blood. Similar observations have been reported in patients with Sjögren syndrome and SLE [33,34].
In principle, the increased frequency of transitional B cells in peripheral blood from JIA patients as compared with that in normal controls might result from dysfunctional bone marrow production or altered recirculation. Studies performed in SLE and Sjögren disease would favor the latter hypothesis [33,34]. Expansion of transitional B cells in blood from HIV-infected and idiopathic CD4 + T lymphocytopenia patients has been correlated with increased serum levels of interleukin-7 [35,36]. Whether a similar correlation applies also to our patients remains to be established.
In this study, CD138 + plasma cells were detected in synovial tissue at the periphery of lymphoid aggregates, lacking follicu-

Figure 5
Histologic analysis of synovial tissue sections from juvenile idiopathic arthritis (JIA) patients Histologic analysis of synovial tissue sections from juvenile idiopathic arthritis (JIA) patients. Serial synovial tissue sections from three JIA patients were stained with CD20 (a), CD138 (b), CD21 (c), and CD27 (d) mAbs by using the peroxidase method. CD20 + B cells cluster within lymphoid aggregates, whereas CD138 + plasma cells localize at the periphery of such aggregates. Staining for CD21, a follicular dendritic cell marker, is negative, consistent with the absence of follicular organization. CD27 + cells are found both inside and outside the lymphoid aggregates. lar organization and containing CD20 + , CD27 + memory B cells admixed with T cells. Synovial tissue from oligoarticular JIA patients with active, long-standing disease was previously shown to be infiltrated with high numbers of plasma cells [37]. Accordingly, we found an enrichment for IgG-secreting plasma blasts in SF and a large amount of IgG-producing B cells in synovial tissue. The functional significance of these findings remains to be established.
We finally showed that SF CD27 + and CD27switch memory B cells were activated, as assessed by the expression of CD69, and expressed high levels of the CD86, but not CD80, costimulatory molecule. Resting B cells are poor antigen-presenting cells because of the low surface expression of costimulatory molecules. Upregulation of CD86 on synovial memory B cells suggests that these cells can efficiently present antigen to and activate T cells, contributing to the persistence of chronic inflammation [38][39][40]. Increased frequency of B cells expressing costimulatory molecules also was reported in other human autoimmune diseases [41][42][43][44]. The dual role of synovial B cells as antibody-producing and antigen-presenting cells supports their important pathogenic role in JIA and provides a rationale for a clinical trial with the anti-CD20 mAb rituximab [12], which has already shown therapeutic efficacy in adult RA patients [10,11].

Conclusions
This study demonstrates for the first time an expansion of switch memory B cells and IgG-secreting plasma blasts in the SF from oligoarticular JIA patients. These memory B cells belonged to both the CD27 + and the CD27subsets and expressed CD86, suggesting their involvement in antigen presentation to T cells. Analysis of chemokines-receptor expression on CD27 + and CD27switch memory B cells delineated potential mechanisms for their recruitment to the inflamed joints. The dual role of synovial B cells as antibodyproducing and antigen-presenting cells supports their pathogenic role in JIA and provides a rationale for a clinical trial with the anti-CD20 mAb rituximab.