Local production of B lymphocyte stimulator protein and APRIL in arthritic joints of patients with inflammatory arthritis.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 48, No. 4, April 2003, pp 982–992 DOI 10.1002/art.10860 © 2003, American College of Rheumatology Local Production of B Lymphocyte Stimulator Protein and APRIL in Arthritic Joints of Patients With Inflammatory Arthritis Soon-Min Tan,1 Dong Xu,1 Viktor Roschke,2 James W. Perry,2 Daniel G. Arkfeld,1 Glenn R. Ehresmann,1 Thi-Sau Migone,2 David M. Hilbert,2 and William Stohl1 and total nucleated cell counts. Although SF and serum BLyS protein levels correlated with each other, SF and serum APRIL levels did not, suggesting that SF BLyS protein levels are more dependent upon systemic factors than are SF APRIL levels. Moreover, in 8 patients who underwent sequential arthrocenteses, changes in SF BLyS protein levels did not immutably parallel changes in SF APRIL levels, indicating their differential regulation. Conclusion. BLyS protein and APRIL are locally produced in inflamed joints. Their respective SF levels are differentially regulated, suggesting that they serve different functions. Together, their local production may foster survival and/or expansion of B cells that produce pathogenic autoantibodies and/or promote local T cell activation and consequent joint destruction. Objective. To determine whether synovial fluid (SF) levels and cell-surface expression of B lymphocyte stimulator (BLyS) protein and SF levels of APRIL are elevated in patients with inflammatory arthritis (IA). Methods. Same-day blood and SF samples from 89 patients with 103 knee effusions (81 knees with IA and 22 with noninflammatory arthritis [NIA]) were evaluated for BLyS protein and APRIL levels by enzyme-linked immunosorbent assay. Blood and SF mononuclear cells were double-stained for surface BLyS protein and surface CD14 (monocyte marker) and were analyzed by flow cytometry. Complete blood cell counts and SF nucleated cell counts were performed by the clinical hematology laboratory. Results. BLyS protein levels were higher in SF than in corresponding serum samples from both IA and NIA patients. SF BLyS protein levels, but not surface expression of BLyS protein, were disproportionately elevated in IA patients. APRIL levels were higher in SF than in corresponding serum samples from most IA patients but not NIA patients. SF BLyS protein and APRIL levels correlated with each other, and each correlated with SF monocyte, lymphocyte, neutrophil, B lymphocyte stimulator (BLyS; trademark of Human Genome Sciences, Rockville, MD) protein is a recently identified 285–amino acid member of the tumor necrosis factor (TNF) ligand superfamily (1–6). Other names for this protein are TALL-1, BAFF, THANK, TNFSF20 (subsequently renamed TNFSF13B), and zTNF4. It is expressed as a type II transmembrane protein, which is cleaved from the cell surface to release a biologically active soluble 17-kd protein (1–5). Mice genetically deficient in BLyS protein harbor markedly reduced numbers of mature B cells in secondary lymphoid organs and manifest reduced baseline serum levels of Ig and Ig responses to T cell–dependent and T cell–independent antigens (7,8). Conversely, in vivo administration of recombinant BLyS (rBLyS) protein to mice induces B cell expansion and polyclonal hypergammaglobulinemia (1), which is, at least in part, consequent to inhibition of B cell apoptosis and enhanced B cell survival (9–12). Constitutive overexpression of BLyS Supported in part by grants from the NIH (AR-41006), the Alliance for Lupus Research, and the Arthritis Foundation, Southern California Chapter. 1 Soon-Min Tan, MD, Dong Xu, MD, Daniel G. Arkfeld, MD, Glenn R. Ehresmann, MD, William Stohl, MD, PhD: Los Angeles County ⫹ University of Southern California Medical Center, and University of Southern California Keck School of Medicine, Los Angeles, California; 2Viktor Roschke, PhD, James W. Perry, BA, Thi-Sau Migone, PhD, David M. Hilbert, PhD: Human Genome Sciences, Inc., Rockville, Maryland. Address correspondence and reprint requests to William Stohl, MD, PhD, Division of Rheumatology, University of Southern California, 2011 Zonal Avenue, HMR 711, Los Angeles, CA 90033. E-mail: firstname.lastname@example.org. Submitted for publication August 5, 2002; accepted in revised form December 13, 2002. 982 LOCAL PRODUCTION OF BLyS AND APRIL IN INFLAMMATORY ARTHRITIS protein in blys-transgenic mice leads to elevated serum titers of multiple autoantibodies, including anti–doublestranded DNA and rheumatoid factor autoantibodies (6,13,14). APRIL (also called TNFSF13A) is a 250–amino acid member of the TNF ligand superfamily that shares substantial homology with BLyS protein and binds to 2 of the 3 BLyS protein receptors (BCMA and TACI) (15–19). APRIL is not expressed on the cell surface but is processed intracellularly and secreted in its biologically active form (20). As is the case for rBLyS protein, recombinant APRIL (rAPRIL) costimulates B cells in vitro and in vivo (16,17), albeit with considerably less potency. Constitutive overexpression of APRIL in april-transgenic mice leads to enhanced T cell survival and enhanced antigen-specific antibody responses (21). Circulating levels of APRIL in patients with rheumatic diseases have heretofore not been reported. However, circulating levels of BLyS protein have been measured in patients with a variety of rheumatic diseases, including systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), and Sjögren’s syndrome, and have been found to be elevated in a substantial proportion of such patients (22–24). Although it is not yet known what effects BLyS protein/APRIL antagonists may have on human disease or on human in vivo immune responses, the experience with such antagonists in murine rheumatic disease models is telling. Mice genetically prone to the development of SLE ([NZB ⫻ NZW]F1 and MRL-lpr/lpr mice) have elevated circulating levels of BLyS protein, and treatment of these SLE-prone mice with a genetically engineered soluble fusion protein between one of the BLyS protein receptors (TACI) and IgG Fc (TACI-Ig) ameliorates progression of disease and improves survival (6). Moreover, the joint inflammation and joint destruction of collageninduced arthritis, a model of noninfectious inflammatory arthritis (IA), are inhibited by treatment with TACI-Ig before or after induction of disease (7,25). Since TACI-Ig binds both BLyS protein and APRIL (16–18), it is not clear whether its inhibition of IA reflects inhibition of BLyS protein activity, inhibition of APRIL activity, or a combination of both. In any case, these observations raise the possibility that local expression and/or production of BLyS protein and/or APRIL in inflamed joints may contribute to the development and/or propagation of disease. We took the first step in addressing this issue by assessing blood and synovial fluid (SF) levels of BLyS protein and APRIL as well as BLyS protein cell surface expression in patients with IA and in patients with noninflammatory arthritis (NIA). 983 PATIENTS AND METHODS Subjects. Patients who were admitted to the Los Angeles County ⫹ University of Southern California Medical Center (LAC⫹USC MC) and were seen in consultation by the Rheumatology Service or patients who were receiving outpatient medical care at the Rheumatology Clinics of LAC⫹USC MC, the Edward R. Roybal Comprehensive Health Center, the USC Ambulatory Health Center, or the USC Center for Arthritis and Joint Implant Surgery were recruited for this study. Criteria for admission to the study were the clinically indicated need for a therapeutic and/or diagnostic arthrocentesis of one or both knees and a willingness to participate in the study. No exclusions were made on any basis other than an inability to give informed consent. Diagnoses were based on established clinical criteria (26). Samples obtained from a total of 103 knee arthrocenteses (89 patients) were evaluated: 22 SF samples from patients with NIA (15 with osteoarthritis [OA] and 7 with traumatic arthritis), 49 from patients with RA, 14 from patients with crystalinduced arthritis (12 with gout and 2 with calcium pyrophosphate dihydrate crystal deposition disease), 18 from patients with other noninfectious IA (5 with ankylosing spondylitis, 7 with reactive arthritis, 4 with SLE, 1 with polymyalgia rheumatica, and 1 with unspecified culture-negative IA). A patient with features of more than one rheumatic disease was classified as having the disease which clinically predominated. Infectious arthritis was excluded by appropriate microbiologic studies. Blood and SF studies. Venous blood and knee SF were collected and sent to the appropriate clinical laboratories; a complete blood cell count with differential cell count was performed on the blood samples, and a nucleated cell count with differential cell count and crystal examination was performed on the SF samples. In addition, nonheparinized and heparinized venous blood and SF samples were used for experimental studies. For serologic studies, sera obtained from nonheparinized blood and SF samples were shipped to Human Genome Sciences (HGS; Rockville, MD) for measurement of BLyS protein and APRIL. BLyS protein levels were determined by enzyme-linked immunosorbent assay (ELISA) as previously described (23,27). APRIL levels were determined by ELISA in a similar manner, using mouse anti-human APRIL monoclonal antibody (mAb) 16F11 (HGS) as the capture antibody and biotinylated rabbit anti-human APRIL polyclonal antibody (HGS) as the detection antibody. For determination of either BLyS protein or APRIL, Immulon-II plates (Labsystems, Franklin, MA) were coated overnight at 4°C with anti-BLyS protein or anti-APRIL capture mAb (100 l at 3 g/ml), washed, and blocked for 2 hours at room temperature with blocking buffer (3% bovine serum albumin [BSA] in phosphate buffered saline [PBS]). Serial dilutions of serum or SF samples were prepared in diluent (0.1% Tween 20 ⫹ 0.1% BSA in PBS) and added to the antibody-coated plates for 2 hours at room temperature. The plates were washed, incubated for 2 hours with the biotinylated detection antibodies (100 l at 0.25 g/ml), washed, incubated for 1 hour with streptavidin–peroxidase conjugate (100 l at 0.25 g/ml in diluent; Vector, Burlingame, CA), and developed with the tetramethylbenzidine substrate kit (Sigma, St. Louis, MO). 984 TAN ET AL Table 1. Demographic and medication data of the 89 study patients, by diagnostic group Parameter Age, years Mean Range Sex, no. Female Male Race, no. Arabic Asian Black Hispanic White Treatment, no. Prednisone Methotrexate Azathioprine Sulfasalazine Hydroxychloroquine Leflunomide Etanercept Infliximab Osteoarthritis/ traumatic arthritis (n ⫽ 22) Rheumatoid arthritis (n ⫽ 39) Other noninfectious inflammatory arthritis (n ⫽ 14) 58.6 31.9–79.8 47.5 19.9–76.1 42.5 20.8–81.4 Crystal-induced arthritis (n ⫽ 14) 52.9 35.9–72.2 12 10 33 6 5 9 2 12 0 0 0 16 6 1 2 0 32 5 0 1 1 11 1 0 1 2 10 1 2 2 0 0 2 0 0 0 18 18 4 5 11 4 2 2 3 2 1 5 4 0 0 0 1 1 0 0 1 0 0 0 Optical density at 450 nm was measured with a SpectraMax 3000 plate reader (Molecular Devices, Sunnyvale, CA). Standards of rBLyS protein (HGS) and rAPRIL (R&D Systems, Minneapolis, MN) at 0.005–100 ng/ml were processed along with test samples, and the concentrations of BLyS protein or APRIL in the experimental samples were calculated from the standard curves. To ascertain the validity of the BLyS protein and APRIL ELISAs for both serum and SF, test sera and SF were intentionally spiked with known amounts of either rBLyS protein or rAPRIL. In all cases, detection of the respective protein was as predicted, with no evidence of ELISA inhibitors and no difference in the degree of recovery between serum and SF (HGS: unpublished observations). For cell-based studies, mononuclear cells were isolated from heparinized blood and SF by Ficoll density-gradient centrifugation and were double stained with fluorescein isothiocyanate–conjugated anti-CD14 mAb (PharMingen, San Diego, CA) plus biotinylated anti-BLyS mAb 9B6 followed by phycoerythrin-conjugated streptavidin (Dako, Carpinteria, CA). The specificity of mAb 9B6 for BLyS protein has been established by the binding of mAb 9B6 to plate-bound rBLyS protein (but not to rAPRIL, BSA, ovalbumin, or human IgG) and by the ability of mAb 9B6 (but not control IgG1) to immunoprecipitate a 36-kd band (detected by Western blotting with anti-BLyS protein polyclonal antibodies) from monocytes and BLyS-transfected 293T cells but not from nontransfected 293T cells or from 293T cells transfected with other membrane-bound TNF family members (HGS: unpublished observations). The cell surface staining observed with mAb 9B6 is not observed with control IgG1 mAb (27). Stained cells were analyzed by flow cytometry, with at least 10,000 events analyzed for each sample. Cell debris, as determined by forward- and side-scatter characteristics, was electronically excluded from the analysis. Patient profiles. Each patient’s sex, race, age, and medications at the time of the arthrocentesis and phlebotomy were recorded. The demographic and medication data of the study patients are listed in Table 1. Statistical analysis. All analyses were performed using SigmaStat software (SPSS, Chicago, IL). Neither serum nor SF BLyS protein or APRIL levels were normally distributed, so they were each log-transformed to achieve normality. Parametric testing between 2 matched or unmatched groups was performed by the paired or unpaired t-test, respectively. Parametric testing among 3 or more groups was performed by one-way analysis of variance (ANOVA). When log transformation failed to generate normally distributed data or the equal variance test was not satisfied, nonparametric testing was performed by the Mann-Whitney rank sum test between 2 groups and by Kruskal-Wallis one-way ANOVA on ranks among 3 or more groups. Correlations were determined by Pearson’s product-moment correlation for interval data and by Spearman’s rank order correlation for ordinal data or for interval data which did not follow a normal distribution. Nominal data were analyzed by chi-square analysis-ofcontingency tables. RESULTS Disproportionate elevation of SF BLyS protein and APRIL levels in IA patients. SF obtained from 103 clinically swollen knees of 89 patients and the corre- LOCAL PRODUCTION OF BLyS AND APRIL IN INFLAMMATORY ARTHRITIS 985 Figure 1. Serum and synovial fluid (SF) levels of B lymphocyte stimulator (BLyS) protein and APRIL in arthritis patients. Top left, BLyS protein levels from concurrently obtained serum and SF from individual patients with osteoarthritis/traumatic arthritis (OA/Tr; 22 serum, 22 SF samples), rheumatoid arthritis (RA; 45 serum, 49 SF samples), other noninfectious inflammatory arthritis (Inflam; 17 serum, 18 SF samples), or crystal-induced arthritis (Crys; 14 serum, 14 SF samples) are depicted as circles. The lines inside the boxes indicate the medians; the outer borders of the boxes indicate the 25th and 75th percentiles; the bars extending from the boxes indicate the 10th and 90th percentiles. Top right, The results are plotted as the ratio of SF to serum BLyS protein levels. Bottom left, APRIL levels from concurrently obtained serum and SF from individual patients with OA/traumatic arthritis (19 serum, 22 SF samples), RA (42 serum, 49 SF samples), other noninfectious inflammatory arthritis (16 serum, 18 SF samples), or crystal-induced arthritis (13 serum, 13 SF samples). Bottom right, The results are plotted as the ratio of SF to serum APRIL levels. sponding serum samples were analyzed for BLyS protein levels. Although the range of serum BLyS protein levels was broad, the collective values were very similar among all the patient cohorts (P ⫽ 0.644) (Figure 1, top). In each patient cohort, BLyS protein levels in SF from clinically affected knees were significantly higher than the levels in concurrently obtained serum samples (P ⱕ 0.001 for each cohort). Nevertheless, BLyS protein levels in SF from each of the IA cohorts were significantly higher than the levels in SF from the NIA patients (P ⬍ 0.001). SF and serum BLyS protein levels in an additional patient with gonococcal arthritis (13 and 3.3 ng/ml, respectively) and in an additional patient with tuberculous arthritis (14 and 5.8 ng/ml, respectively) were similar to those in the IA patients. This dissimilarity between IA and NIA patients was further highlighted by analysis of the SF-to-serum BLyS protein ratios. A ratio of 3 was arbitrarily defined as a marked increased in SF BLyS protein level relative to that in serum. In OA/traumatic arthritis patients, only 1 of 22 SF samples demonstrated a BLyS protein ratio ⱖ3, whereas 18 of the 49 RA, 6 of the 18 other noninfectious IA, and 5 of the 14 crystal-induced arthritis SF samples demonstrated BLyS protein ratios ⱖ3 (P ⫽ 0.041 for RA versus each of the other 3 groups, and P ⫽ 0.009 for the entire IA group versus the NIA group). APRIL levels were measured in 93 SF samples and their corresponding sera. As was the case with BLyS protein, serum APRIL levels were very similar among all the patient cohorts (P ⫽ 0.405) (Figure 1, bottom). In contrast to the case with BLyS protein, SF APRIL levels in clinically affected knees from NIA patients were lower than those in the corresponding sera (P ⬍ 0.001), with an SF-to-serum APRIL ratio of ⬍1 in 17 of 19 cases. Nevertheless, the SF-to-serum ratios in the majority (52 of 74) of IA samples were ⬎1 (P ⬍ 0.001), and SF 986 TAN ET AL Figure 2. Surface expression of BLyS protein by blood and SF mononuclear cells in arthritis patients. Top, Mononuclear cells from concurrently obtained blood and SF from 2 OA, 2 RA, 2 systemic lupus erythematosus (SLE), and 2 gout patients were double-stained for surface CD14 and BLyS protein. Results are presented as contour plots. The numbers in the upper right quadrants of each tracing indicate the percentages of CD14⫹ cells that stained positive for BLyS protein. Bottom, Percentages of CD14⫹ cells that stained positive for BLyS protein from blood and SF from individual patients with OA/traumatic arthritis (12 blood, 11 SF samples), RA (33 blood, 36 SF samples), other noninfectious inflammatory arthritis (16 blood, 17 SF samples), or crystal-induced arthritis (9 blood, 9 SF samples) are depicted as circles. The lines inside the boxes indicate the medians; the outer borders of the boxes indicate the 25th and 75th percentiles; the bars extending from the boxes indicate the 10th and 90th percentiles. See Figure 1 for other definitions. APRIL levels in clinically affected knees from IA patients were collectively higher than those from NIA patients (P ⬍ 0.001 for RA versus each of the other 3 groups, and P ⬍ 0.001 for the entire IA group versus the NIA group). Expression of surface BLyS protein by blood and SF mononuclear cells. To assess whether SF BLyS protein expression is also disproportionately elevated in IA patients, blood and SF mononuclear cells from IA and NIA patients were stained for surface BLyS protein. Staining profiles from representative individuals within each patient cohort are presented in Figure 2 (top). As anticipated, the great majority of BLyS protein–positive cells in both blood and SF was limited to the CD14⫹ cell population (monocytes) (1–3,5,27), although in some LOCAL PRODUCTION OF BLyS AND APRIL IN INFLAMMATORY ARTHRITIS 987 Figure 3. Correlations among levels of B lymphocyte stimulator (BLyS) protein or APRIL in synovial fluid (SF) or blood and counts of nucleated cells. Each circle indicates an individual SF or blood sample. The lines are the calculated regression lines. Top, SF monocyte, lymphocyte, and neutrophil counts were determined along with SF BLyS protein levels in 85 samples. SF nucleated cell counts were determined along with SF BLyS protein levels in 102 samples. White blood cell (WBC) counts in blood were determined along with serum BLyS protein levels in 97 samples. Middle, SF monocyte, lymphocyte, and neutrophil counts were determined along with SF APRIL levels in 83 samples. SF nucleated cell counts were determined along with SF APRIL levels in 100 samples. WBC counts in blood were determined along with serum APRIL levels in 93 samples. Bottom, Serum BLyS protein and APRIL levels were both determined in 90 blood samples. SF BLyS protein and APRIL levels were both determined in 101 SF samples. Serum and SF BLyS protein levels were determined in 103 matched samples. Serum and SF APRIL levels were determined in 93 matched samples. samples, there were small CD14– cell populations that also stained positive for BLyS protein. Overall, there was considerable variability in BLyS protein staining patterns, even among patients within the same cohort. As illustrated in Figure 2 (bottom), there was no consistent increase in surface BLyS protein expression by monocytes in blood or SF from IA patients relative to that in blood or SF from NIA patients. In fact, the percentages of BLyS protein– positive monocytes in RA SF were actually lower than the percentages in SF from the other patient cohorts (including OA/traumatic arthritis, P ⬍ 0.001) despite BLyS protein levels being significantly higher in RA SF than in OA/traumatic arthritis SF (Figure 1). Correlations between local inflammatory cell response and BLyS protein or APRIL levels in SF. The lack of positive association between the percentages of BLyS protein–positive cells and the actual BLyS protein 988 TAN ET AL Table 2. BLyS protein and APRIL levels in patients who underwent concurrent bilateral knee arthrocenteses* BLyS protein, ng/ml APRIL, ng/ml Synovial fluid Synovial fluid Patient Diagnosis Serum Left knee Right knee Serum Left knee Right knee A B C D E RA RA RA RA ReA 1.4 3.2 6.3 3.4 6.6 5.8 4.2 18 13 18 6.6 5.5 20 11 18 ND 35 20 17 23 100 29 19 34 260 98 40 22 43 78 * RA ⫽ rheumatoid arthritis; ND ⫽ not determined; ReA ⫽ reactive arthritis. levels in SF was an unexpected finding. However, the absolute number of BLyS protein–positive cells (monocytes) in the affected joints of IA patients was higher than that in the affected joints of NIA patients. Indeed, there was a significant positive correlation between SF monocyte counts and SF BLyS protein levels (Figure 3, top). Similar positive correlations were also noted between SF lymphocyte or neutrophil counts and SF BLyS protein levels. Importantly, the SF total nucleated cell counts correlated with SF BLyS levels the strongest, suggesting that the intensity of the local inflammatory response in the joint, rather than just the monocyte count alone, is a major determinant of SF BLyS protein levels. Qualitatively similar, and quantitatively more striking, correlations were also observed between SF APRIL levels and SF cell counts (Figure 3, middle). No correlations were noted between serum BLyS protein or APRIL levels and white blood cell counts in blood, which is consistent with circulating BLyS protein and APRIL arising predominantly from extravascular sites (i.e., production in secondary lymphoid tissues and release into the circulation). Lack of immutable concordance between BLyS protein and APRIL levels in SF. SF and serum BLyS protein levels correlated with each other, but SF and serum APRIL levels did not (Figure 3, bottom). This suggested that SF APRIL levels may be less dependent Figure 4. Surface expression of B lymphocyte stimulator (BLyS) protein by blood and synovial fluid (SF) mononuclear cells in arthritis patients who underwent bilateral knee arthrocentesis on the same day. Mononuclear cells from concurrently obtained blood and SF from both knees of a patient with rheumatoid arthritis (RA; patient D in Table 2) and a patient with reactive arthritis (ReA; patient E in Table 2) were double-stained for surface CD14 and BLyS protein. Results are presented as contour plots. The numbers in the upper right quadrants of each tracing indicate the percentages of CD14⫹ cells that stained positive for BLyS protein. LOCAL PRODUCTION OF BLyS AND APRIL IN INFLAMMATORY ARTHRITIS upon global systemic factors than are SF BLyS protein levels. If this were the case, then BLyS protein levels in SF collected at a single time point from anatomically distinct sites might not necessarily be paralleled by corresponding APRIL levels. Moreover, changes in serum and/or SF BLyS protein levels over time might not be paralleled by corresponding changes in APRIL levels. Five patients underwent therapeutic arthrocenteses of both knees on the same day (Table 2). In 2 of these patients, staining of blood and SF mononuclear cells was also performed (Figure 4). In each patient, the BLyS protein staining patterns for SF mononuclear cells from each knee were remarkably similar but clearly distinct from the staining pattern of corresponding blood mononuclear cells. Although serum and SF BLyS protein levels varied considerably among the 5 patients, SF BLyS protein levels for each patient were very similar in each knee. Serum and SF APRIL levels also varied considerably among these patients, with APRIL levels in SF from each knee being similar in 4 of the patients. However, in the fifth patient (patient E), there was a marked disparity in SF APRIL levels in the two knees. Eight patients underwent arthrocentesis on more than one occasion (Table 3). In 5 of them (patients F, H, I, J, and K), changes (or lack of changes) in SF BLyS protein and APRIL levels occurred in parallel. In addition, the expression of BLyS protein on SF mononuclear cells obtained from either arthrocentesis in patient F was very similar (Figure 5, right). However, in 3 patients, there was a clear discordance between interval changes in SF BLyS protein levels and SF APRIL levels. In patient E, BLyS protein levels in serum and SF at the time of the second arthrocentesis were substantially lower than those at the time of the first, despite there being only a 2-day interval between the two procedures. Of interest, BLyS protein expression on SF mononuclear cells from this patient was relatively unchanged (Figure 5, left), pointing to an uncoupling of BLyS protein expression from BLyS protein levels. In contrast to the decrease in SF BLyS protein levels, serum APRIL levels remained unaffected, and SF APRIL levels in 1 joint modestly rose. In patients G and L, SF and serum BLyS protein levels fell over time. In contrast, serum and SF APRIL levels rose by the third and second arthrocenteses, respectively. DISCUSSION In patients with inflammatory arthritis, BLyS protein and APRIL levels in SF from clinically affected knees were significantly higher than the levels in corre- 989 sponding sera (Figure 1). Although BLyS protein levels (but not APRIL levels) in patients with noninflammatory arthritis were also greater in SF from clinically affected knees than in sera, the levels in SF from IA patients were much higher than those in SF from NIA patients, despite all patients having similar serum BLyS protein levels. The difference in SF BLyS protein and APRIL levels between IA and NIA patients almost certainly related in large measure to the intensity of the local inflammatory responses in the affected joints of the respective patients (Figure 3). With regard to BLyS protein, not only did SF BLyS protein levels correlate significantly with SF monocyte counts (the putative BLyS protein–producing cells), but they also correlated significantly with SF lymphocyte counts and SF neutrophil counts. Although other interpretations are possible, the simplest explanation is that increased numbers of BLyS protein–producing cells, in the presence of increased numbers of activated lymphocytes and neutrophils, led to increased levels of BLyS protein. The immediate stimulus to local production of BLyS protein in inflammatory SF is unknown, but proinflammatory cytokines such as interferon-␥ (IFN␥) and IFN␣ are almost certainly contributory (27,28). We did not measure BLyS protein levels in SF from clinically unaffected joints of IA or NIA patients, but we predict that there would be low BLyS protein levels in SF from joints with low-to-absent degrees of inflammation. Given the strong correlation between serum and SF BLyS protein levels, we propose that SF BLyS protein levels are determined by local factors (e.g., degree of inflammation, cytokine milieu) that accentuate a systemic proclivity to BLyS protein production. Of note, there was no consistent pattern of BLyS protein expression by SF mononuclear cells on a per-cell or a percentage basis, even within a given patient cohort (Figure 2, top). The great majority of BLyS protein– positive cells were CD14⫹, although in some patients, CD14– cells staining positively for BLyS protein were detected. Whether these cells represent CD14⫹ monocytes that have down-regulated their surface CD14 in vivo, other myeloid lineage cells that are inherently CD14–, or CD14– nonmyeloid lineage cells that can express surface BLyS protein or to which BLyS protein passively adheres in vivo remains to be established. Also of note, staining of SF mononuclear cells from patient E obtained from the 2 arthrocenteses (Figure 5) failed to reveal a difference in surface BLyS protein expression despite the substantial difference in BLyS protein levels (Table 3). This suggests a divergence 990 TAN ET AL Table 3. BLyS protein and APRIL levels in patients with repeated arthrocenteses Patient, diagnosis,* arthrocentesis E, ReA 1 2 2 F, RA 1 2 G, RA 1 2 3 H, AS 1 2 I, ReA 1 2 J, RA 1 2 K, RA 1 2 L, RA 1 2 BLyS protein, ng/ml Interval, days APRIL, ng/ml Knee Serum Synovial fluid Serum Synovial fluid Ipsilateral Contralateral 26 6.6 6.6 29 18 18 22 23 23 210 78 260† 0 63 Ipsilateral 2.4 2.8 6.4 7.6 26 26 31 37 0 91 290 Ipsilateral Ipsilateral 5.6 4.1 2.6 22 7.0 7.3 12 2.8 27 76 25 64† 0 14 Ipsilateral 3.8 2.3 7.2 9.1 24 21 16 14 0 35 Contralateral 1.7 1.5 5.1 4.1 36 39 6.9 7.0 0 78 Contralateral 2.6 6.3 12 18 18 13 23 39 0 82 Contralateral 1.9 3.2 6.2 8.7 9.3 21 28 30 0 147 Contralateral 4.2 1.0 11 3.2 13 30 9.3 37† 0 2 2 * ReA ⫽ reactive arthritis; RA ⫽ rheumatoid arthritis; AS ⫽ ankylosing spondylitis. † Unambiguous discordance between the interval changes in synovial fluid levels of BLyS protein and APRIL. Figure 5. Surface expression of B lymphocyte stimulator (BLyS) protein by blood and synovial fluid (SF) mononuclear cells in arthritis patients who underwent repeat arthrocentesis of the same knee. At the time of the first arthrocentesis (day 0, time 1), mononuclear cells were isolated from blood and SF from the clinically affected knee of a patient with reactive arthritis (ReA; patient E in Table 3) and a patient with rheumatoid arthritis (RA; patient F in Table 3) and were double-stained for surface CD14 and BLyS protein. Repeat arthrocentesis of the same knee and phlebotomy were performed on day 2 (ReA patient) and day 63 (RA patient) (time 2). Results are presented as contour plots. The numbers in the upper right quadrants of each tracing indicate the percentages of CD14⫹ cells that stained positive for BLyS protein. LOCAL PRODUCTION OF BLyS AND APRIL IN INFLAMMATORY ARTHRITIS between BLyS protein expression and BLyS protein production. Since soluble BLyS protein arises from the cleavage of surface membrane BLyS protein (3), regulation of such cleavage may be critical to the ultimate levels of soluble BLyS protein. At present, very little is known about the regulation of the release of soluble BLyS protein from the membrane-bound form. Indeed, it may be that accelerated cleavage of surface BLyS protein contributes to reduced percentages of BLyS protein–positive monocytes in SF from RA patients (Figure 2), despite elevated BLyS protein levels in these same SF (Figure 1). Development of an assay that identifies individual BLyS protein–producing cells (e.g., enzyme-linked immunospot assay) would permit formal assessment of the relationship between BLyS protein– expressing cells and BLyS protein–producing cells. As was the case for SF BLyS protein levels, SF APRIL levels also correlated significantly with SF monocyte, lymphocyte, neutrophil, and total nucleated cell counts. The identity of the factors which promote APRIL production and the cells actually producing increased amounts of APRIL remain to be established. It is likely that temporal differences in the local production of various cytokines that may differentially affect APRIL and BLyS protein production occur, and such differences may substantially contribute to the discordance between SF APRIL and BLyS protein levels. Given the lack of correlation between serum and SF APRIL levels, we propose that SF APRIL levels are largely locally regulated and highly independent of systemic APRIL production. This presumed divergence between the regulation of SF BLyS protein levels and the regulation of SF APRIL levels is highlighted by the findings in patients from whom more than one SF sample was obtained. In the 5 IA patients who underwent arthrocentesis of both knees on the same day, SF BLyS protein levels were remarkably similar in each joint (Table 2). In the 2 patients whose SF mononuclear cells from both knees were stained, BLyS protein expression was also highly similar in each (Figure 4). In contrast, SF APRIL levels in both knees of one of these patients were markedly disparate (Table 2). Moreover, in patients studied on multiple occasions (Table 3), those whose SF showed considerable sequential reduction in BLyS protein levels were patient E, who was started on an around-the-clock regimen of an antiinflammatory agent, patient G, who received intraarticular corticosteroids, and patient L, who was started on disease-modifying agents and a TNF antagonist. No consistent parallel reductions in SF APRIL levels were observed in these patients. 991 The apparent differences in their regulation notwithstanding, increased SF levels of BLyS protein and APRIL are each likely to be a consequence of the local inflammatory response rather than its proximate cause. Nevertheless, rather than being an epiphenomenon of uncertain biologic importance, there is considerable, albeit inferential, evidence in animal models that points to a vital role for B cell survival factors, such as BLyS protein and/or APRIL, in noninfectious, non–crystalinduced arthritis. First, when CD4⫹ T cells isolated from individual follicles of human RA synovium are injected into SCID mice previously transplanted with major histocompatibility complex–matched RA synovium, production of proinflammatory cytokines (as measured by messenger RNA levels) in the transplanted synovium is dramatically up-regulated. However, if the target synovium is inherently deficient in B cells or if the recipient mice are treated with anti-CD20 mAb to render the synovium deficient in B cells, then adoptively transferred CD4⫹ T cells do not up-regulate the proinflammatory cytokines (29). These results imply that factors, such as BLyS protein and/or APRIL, could facilitate T cell–driven inflammation in RA by enhancing B cell survival in the synovium. In the absence of a chronic T cell–driven inflammatory process, elevated BLyS protein and/or APRIL levels may have little biologic impact. Accordingly, elevated BLyS protein and APRIL levels in crystal-induced arthritis may have no (or little) pathogenetic ramifications due to no (or little) chronic T cell–driven inflammation in this condition. Second, the inflammation and joint destruction associated with murine collagen-induced arthritis (a disease highly dependent upon T cells) are inhibited, even after their induction, by the BLyS protein/APRIL antagonist, TACI-Ig (7,25). Since the mice treated with TACI-Ig underwent a significant reduction in B cells, the observations collectively suggest that B cells are vital to disease pathogenesis. By extension, B cell survival factors, including BLyS protein and/or APRIL, could facilitate the development and/or exacerbation of IA in humans. In situ studies of the spatial relationship between BLyS protein–expressing cells and B cell clusters in synovial tissue are currently in progress. In any case, BLyS protein/APRIL antagonists may have salutary effects on IA and find an important niche in the management of patients with IA. ACKNOWLEDGMENTS The authors thank Hal Soucier for performing the flow cytometry, the attending physicians and house officers of the 992 TAN ET AL Division of Rheumatology for their assistance in identifying and recruiting patients for this study, and all the subjects for their participation. REFERENCES 1. Moore PA, Belvedere O, Orr A, Piwei K, LaFleur DW, Feng P, et al. BLyS: member of the tumor necrosis factor family and B lymphocyte stimulator. Science 1999;285:260–3. 2. Shu H-B, Hu W-H, Johnson H. TALL-1 is a novel member of the TNF family that is down-regulated by mitogens. J Leukocyte Biol 1999;65:680–3. 3. Schneider P, MacKay F, Steiner V, Hofmann K, Bodmer JL, Holler N, et al. BAFF, a novel ligand of the tumor necrosis factor family, stimulates B cell growth. J Exp Med 1999;189:1747–56. 4. Mukhopadhyay A, Ni J, Zhai Y, Yu G-L, Aggarwal BB. Identification and characterization of a novel cytokine, THANK, a TNF homologue that activates apoptosis, nuclear factor-B, and c-Jun NH2-terminal kinase. J Biol Chem 1999;274:15978–81. 5. Tribouley C, Wallroth M, Chan V, Paliard X, Fang E, Lamson G, et al. Characterization of a new member of the TNF family expressed on antigen presenting cells. Biol Chem 1999;380: 1443–47. 6. Gross JA, Johnston J, Mudri S, Enselman R, Dillon SR, Madden K, et al. TACI and BCMA are receptors for a TNF homologue implicated in B-cell autoimmune disease. Nature 2000;404:995–9. 7. Gross JA, Dillon SR, Mudri S, Johnston J, Littau A, Roque R, et al. TACI-Ig neutralizes molecules critical for B cell development and autoimmune disease: impaired B cell maturation in mice lacking BLyS. Immunity 2001;15:289–302. 8. Schiemann B, Gommerman JL, Vora K, Cachero TG, ShulgaMorskaya S, Dobles M, et al. An essential role for BAFF in the normal development of B cells through a BCMA-independent pathway. Science 2001;293:2111–4. 9. Thompson JS, Schneider P, Kalled SL, Wang L, Lefebre EA, Cachero TG, et al. BAFF binds to the tumor necrosis factor receptor-like molecule B cell maturation antigen and is important for maintaining the peripheral B cell population. J Exp Med 2000;192:129–35. 10. Do RKG, Hatada E, Lee H, Tourigny MR, Hilbert D, Chen-Kiang S. Attenuation of apoptosis underlies B lymphocyte stimulator enhancement of humoral immune response. J Exp Med 2000;192: 953–64. 11. Batten M, Groom J, Cachero TG, Quian F, Schneider P, Tschopp J, et al. BAFF mediates survival of peripheral immature B lymphocytes. J Exp Med 2000;192:1453–66. 12. Hsu BL, Harless SM, Lindsley RC, Hilbert DM, Cancro MP. Cutting edge: BLyS enables survival of transitional and mature B cells through distinct mediators. J Immunol 2002;168:5993–6. 13. Mackay F, Woodcock SA, Lawton P, Ambrose C, Baetscher M, Schneider P, et al. Mice transgenic for BAFF develop lymphocytic 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. disorders along with autoimmune manifestations. J Exp Med 1999;190:1697–710. Khare SD, Sarosi I, Xia X-Z, McCabe S, Miner K, Solovyev I, et al. Severe B cell hyperplasia and autoimmune disease in TALL-1 transgenic mice. Proc Natl Acad Sci U S A 2000;97:3370–5. Hahne M, Kataoka T, Schröter M, Hofmann K, Irmler M, Bodmer JL, et al. APRIL, a new ligand of the tumor necrosis factor family, stimulates tumor cell growth. J Exp Med 1998;188:1185–90. Marsters SA, Yan M, Pitti RM, Haas PE, Dixit VM, Ashkenazi A. Interaction of the TNF homologues BLyS and APRIL with the receptor homologues BCMA and TACI. Curr Biol 2000;10:785–8. Yu G, Boone T, Delaney J, Hawkins N, Kelley M, Ramakrishnan M, et al. APRIL and TALL-1 and receptors BCMA and TACI: system for regulating humoral immunity. Nat Immunol 2000;1: 252–6. Wu Y, Bressette D, Carrell JA, Kaufman T, Feng P, Taylor K, et al. Tumor necrosis factor (TNF) receptor superfamily member TACI is a high affinity receptor for TNF family members APRIL and BLyS. J Biol Chem 2000;275:35478–85. Rennert P, Schneider P, Cachero TG, Thompson J, Trabach L, Hertig S, et al. A soluble form of B cell maturation antigen, a receptor for the tumor necrosis factor family member APRIL, inhibits tumor cell growth. J Exp Med 2000;192:1677–84. López-Fraga M, Fernández R, Albar JP, Hahne M. Biologically active APRIL is secreted following intracellular processing in the Golgi apparatus by furin convertase. EMBO Rep 2001;2:945–51. Stein JV, López-Fraga M, Elustondo FA, Carvalho-Pinto CE, Rodriguez D, Gomez-Caro R, et al. APRIL modulates B and T cell immunity. J Clin Invest 2002;109:1587–98. Zhang J, Roschke V, Baker KP, Wang Z, Alarcón GS, Fessler BJ, et al. Cutting edge: a role for B lymphocyte stimulator in systemic lupus erythematosus. J Immunol 2001;166:6–10. Cheema GS, Roschke V, Hilbert DM, Stohl W. Elevated serum B lymphocyte stimulator levels in patients with systemic immunebased rheumatic diseases. Arthritis Rheum 2001;44:1313–9. Groom J, Kalled SL, Cutler AH, Olson C, Woodcock SA, Schneider P, et al. Association of BAFF/BLyS overexpression and altered B cell differentiation with Sjögren’s syndrome. J Clin Invest 2002;109:59–68. Wang H, Marsters SA, Baker T, Chan B, Lee WP, Fu L, et al. TACI-ligand interactions are required for T cell activation and collagen-induced arthritis in mice. Nat Immunol 2001;2:632–7. Klippel JH, Crofford LJ, Stone JH, Weyand CM, editors. Primer on the rheumatic diseases. 12th ed. Atlanta: Arthritis Foundation; 2001. Nardelli B, Belvedere O, Roschke V, Moore PA, Olsen HS, Migone TS, et al. Synthesis and release of B-lymphocyte stimulator from myeloid cells. Blood 2001;97:198–204. Litinskiy MB, Nardelli B, Hilbert DM, He B, Schaffer A, Casali P, et al. DCs induce CD40-independent immunoglobulin class switching through BLyS and APRIL. Nat Immunol 2002;3:822–9. Takemura S, Klimiuk PA, Braun A, Goronzy JJ, Weyand CM. T cell activation in rheumatoid synovium is B cell dependent. J Immunol 2001;167:4710–8.