• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Oct 29, 2002; 99(22): 14368–14373.
Published online Oct 21, 2002. doi:  10.1073/pnas.182549099
PMCID: PMC137890

Despite ubiquitous autoantigen expression, arthritogenic autoantibody response initiates in the local lymph node


K/BxN mice develop an inflammatory joint disease with many features characteristic of rheumatoid arthritis. In this model, the KRN transgenic T cells and nontransgenic B cells both recognize the glycolytic enzyme glucose-6-phosphate-isomerase (GPI) as an autoantigen. Here, we followed the anti-GPI B cell response that naturally arises in K/BxN mice. The anti-GPI B cell response was robust and arose at the same time as the development of serum anti-GPI autoantibody and joint inflammation. Surprisingly, although GPI was expressed systemically, the anti-GPI B cell response was focused to the lymph nodes (LN) draining the distal joints where arthritis was evident. In lymphotoxin-β receptor-Ig-treated mice, which lack LNs, the development of arthritis was completely inhibited up to 5–6 weeks. At later times, some arthritis did develop, but at a significantly reduced level. Thus, in this spontaneous model of autoimmunity, the LNs draining the distal joints are essential for both the inhibition and amplification of the arthritogenic B cell response. These findings imply that the immune physiology of a joint is unique, resulting in a local immune response to a systemic autoantigen.

The K/BxN murine arthritis model shares many disease characteristics with rheumatoid arthritis (RA), including joint inflammation and eventual destruction of the synovial joints (1, 2). This disease affects primarily distal joints, leaving other joints unaffected. The arthritic K/BxN mouse was generated by crossing KRN T cell receptor transgenic (Tg) mice on a C57BL/6 background with nonobese diabetic (NOD) mice. In this model, K/BxN mice develop arthritis by about 1 month of age. Disease incidence is 100% in BxN mice with the KRN Tg, whereas no disease is detected in Tg(−) BxN mice (1). KRN T cells recognize a peptide from the glycolytic enzyme glucose-6-phosphate-isomerase (GPI) bound to I-Ag7 (3, 4). B cells from these mice also recognize GPI as an autoantigen (3). The importance of both the KRN T cells and endogenous B cells to the development of arthritis has been shown by using a series of lymphocyte transfer, Ab depletion, and lymphocyte deficient mouse experiments. Furthermore, anti-GPI serum Ab alone transferred arthritis to naive recipient mice, highlighting the importance of anti-GPI B cells in the initiation phase of disease (1, 5).

How a ubiquitously expressed antigen can lead to the development of a joint-specific autoimmune disease has been a puzzling question. An obvious possibility is that the GPI expressed in the joint is different in form or quantity compared with that found elsewhere in the body. The explanation does not appear to be this simple, as no difference has been found in the coding sequences of GPI transcripts, posttranslational modification of GPI protein, or in GPI transcript/protein levels in the joint versus a tissue that is not affected by disease, the kidney (6). As an alternative, it has been proposed that the GPI found in the joint is no different from that found in other organs (e.g., kidney) and therefore, autoantibodies to GPI would bind to GPI wherever it is exposed and initiate the complement cascade. In most organs, the complement cascade is inhibited by the presence of regulatory proteins. However, the cartilage in the joint might lack these complement regulatory proteins to keep the inflammatory response in check, resulting in overt disease only in the joint (7). This hypothesis implies that the joint specificity lies at the level of the inflammatory cells and their regulators, not at the B and T cell level.

The requirement for inflammatory mediators and the innate immune system (alternative complement pathway, Fc receptors, neutrophils, etc.) in the disease process in K/BxN mice has been clearly documented (7–9); however, the inciting factors that lead to the activation of the autoreactive B and T cells are less clear. T and B cell responses to both foreign and self Ags are primed in lymphoid tissue in close proximity to the source of their Ag (reviewed in ref. 10). Immunization of foreign Ags at various locations on the body has delineated a distinct drainage pattern to local lymph nodes (LNs). For example, Ags administered in the footpad drain to the popliteal LN, whereas those injected into the tailbase drain to the inguinal LNs (11). This distinct pattern has been useful in defining responses to foreign Ags (12). Lymphocyte responses to organ specific self Ags have also been demonstrated preferentially in the LNs draining the target tissue (e.g., myelin basic protein in experimental allergic encephalomyelitis, pancreatic islet Ags in the NOD model of insulin-dependent diabetes, and kidney Ags in experimental glomerulonephritis) (13–15). However, ubiquitously expressed self Ags would be expected to be presented in lymph nodes throughout the body as well as in the spleen. For example, in lupus-prone mice, B cell responses to double-stranded DNA are detectable systemically (16). It follows, then, that autoimmune responses to a ubiquitously expressed Ag such as GPI should also be detectable in all lymphoid organs.

To determine whether the B cell response to GPI is systemic or instead focused to the local lymph nodes draining the joint, we followed anti-GPI B cells before disease, as disease developed, and after disease was well underway. Our findings suggest that, although GPI is expressed systemically, the draining LNs of the distal joints serve as focal points to allow for the efficient initiation, amplification, and pathogenesis of the autoreactive anti-GPI B cell response.

Materials and Methods


KRN TCR Tg mice on a C57BL/6 background have been described (1). NOD mice were purchased from Taconic (Germantown, NY). To obtain arthritic mice, homozygous KRN Tg C57BL/6 mice were crossed with NOD mice yielding KRN (C57BL/6 × NOD)F1 mice designated K/BxN. Age-matched Tg(−) C57BL/6 or Tg(−) (C57BL/6 × NOD)F1 mice were used as controls in all experiments. All mice were bred and housed under specific pathogen-free conditions in the animal facility at the Washington University Medical Center (St. Louis). Studies were performed in accordance with National Institutes of Health and Association for Assessment and Accreditation of Laboratory Animal Care guidelines.

Recombinant GPI–GST Fusion Protein.

The coding sequence of the glucose-6-phosphate isomerase (Gpi-1b) was amplified from cDNA generated from B6.AKR spleen RNA. The GPI coding region was amplified by using primers that encoded EcoRI sites at the 5′ and 3′ ends of the GPI cDNA (amino acids 1–558 of the full-length protein) and added a linker of 7 aa (including a factor Xa cleavage signal) at the 5′ end of GPI. The linker+GPI insert was ligated into plasmid pGEX-2T to yield plasmid pGEX-GPI encoding GST followed by the short linker and GPI residues 1–558. The sequence of the linker and GPI coding region was confirmed by Big Dye terminator mix sequencing (Applied Biosystems). Large scale cultures of XL-1 Blue Escherichia coli carrying pGEX-GPI were grown at 37°C with shaking to OD 0.6, shifted to a 25°C shaker and induced overnight with 100 μM IPTG (Sigma). GPI–GST fusion protein was recovered by using glutathione-Sepharose 4B (Amersham Pharmacia) and eluted with 10 mM reduced glutathione in 50 mM Tris, pH 8.0. Protein concentration was determined by bicinchoninic acid (BCA) protein assay (Pierce).

Flow Cytometry.

Spleen, draining LN (pooled from 2 each of popliteal, axillary, and brachial LNs), or nondraining LN (pooled from 2 each of inguinal, cervical, and mandibular LNs) cells (1 × 106) were surface stained according to standard protocols. RA3-6B2-PE (anti-B220) (BD PharMingen, San Diego) was used as a primary antibody. GPI–GST fusion protein, control CD2AP-GST fusion protein (gift from A. Shaw, Washington University, St. Louis), and rabbit GPI (Sigma) were conjugated to biotin using N-hydroxysuccinimidobiotin (Sigma) in our laboratory. Streptavidin-PerCP (BD PharMingen) was used as a secondary Ab.

All samples were analyzed on a FACScalibur flow cytometer (BD Biosciences, Mountain View, CA) with cellquest software. Gating on live lymphocytes based on forward and side scatter, 20,000 events were collected for each sample.


Spleens or LNs were suspended in OCT, frozen in 2-methyl-butane cooled with liquid nitrogen, sectioned, and fixed with acetone, and sections were stored at −20°C before staining. The sections were blocked with 5% normal goat serum (Vector Laboratories)/0.1% Tween-20 and then stained with RA3-6B2-PE (anti-B220) and GPI–GST-biotin, CD2AP-biotin, or PNA-biotin (Vector Laboratories). Similar staining was observed with recombinant mouse GPI–GST and rabbit GPI (data not shown). Streptavidin–FITC was used as a secondary Ab. Immunofluorescence was visualized under a fluorescent microscope, and pictures were taken with a Nikon DXM1200 digital camera and ACT-1 software.

Arthritis Incidence.

The two rear ankles of K/BxN mice were measured starting at the age of 3 weeks. Measurement of ankle thickness was made above the footpad, axially across the ankle joint by using a Fowler Metric Pocket Thickness Gauge (Ralmikes Tool-A-Rama, NJ). Ankle thickness was rounded off to the nearest 0.05 mm.


Mice were bled once a week between 3 and 11 weeks of age. Sera were stored at −20°C before analysis. These serum samples were plated at an initial dilution of 1:100 and diluted serially 1:5 in Immulon II plates (Fisher Scientific) coated with GPI–GST or CD2AP-GST (2 μg/ml). Similar results were obtained with plates coated with rabbit GPI (data not shown). Donkey anti-mouse total Ig-HRP (Jackson ImmunoResearch), goat anti-mouse IgM–horseradish peroxidase (HRP), IgG1–HRP, IgG2a–HRP, IgG2b–HRP, or IgG3–HRP (Southern Biotechnology Associates) were used as secondary Abs. Serum Ab was detected by using ABTS substrate (2,2′-Azino-di-[3-ethylbenzthiazoline sulfonate] diammonium salt; Roche Molecular Biochemicals). The serum titer was defined as the reciprocal of the last dilution, which gave an OD >3× higher than that of the background.

Anti-GPI Enzyme-Linked Immunospot (ELISpot) Assay.

Spleen or LN cells were plated at 4 × 105 cells per well and diluted serially 1:4 in Multiscreen HA mixed cellulose ester membrane plates (Millipore) coated with GPI–GST or control CD2AP-GST (2 μg/ml). The cells were incubated on the Ag-coated plates for 4 h at 37°C. The Ig secreted by the plated cells was detected by AP-conjugated goat anti-mouse total Ig secondary Ab (Southern Biotechnology Associates) and visualized by using NBT/BCIP substrate (nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate; Sigma).

Lymphotoxin-β Receptor-Ig Fusion Protein (LTβR-Ig) Treatment.

Timed pregnant NOD females, bred to KRN homozygous C57BL/6 males, were injected i.v. with 100 μg LTβR-Ig (gift of J. Browning, Biogen) or control human Ig (Jackson ImmunoResearch) on embryonic day 11 (E11) and E14 as described (17). This method generated K/BxN pups born without most LNs, including the brachial, axillary, inguinal, popliteal, mandibular, parathymic, iliac, and sciatic LNs (17). LTβR-Ig-treated and control Ig-treated mice were measured starting at three weeks for the incidence of arthritis, serum anti-GPI titers, and anti-GPI antibody secreting cells.

Statistical Analysis.

Statistical significance was determined by using an unpaired Mann–Whitney nonparametric test.


Disease Onset and Anti-GPI Seroconversion in K/BxN Mice.

Given the requirement for B cells and serum anti-GPI autoantibody for the development of arthritis in the K/BxN model (1, 5), we wanted to understand when and where the B cell response to GPI initiates. Before following the GPI-specific B cell response, it was first necessary to establish the time course of disease and anti-GPI seroconversion in our colony. Three litters of K/BxN mice were followed for the onset and progression of arthritis starting at the age of 3 weeks (Fig. (Fig.11a). No disease was detected in 3- to 4-week-old mice. Ankle swelling first became measurable between weeks 4 and 5 and was present in all mice by 5 weeks. Disease continued to develop in mice aged 5–7 weeks and then plateaued and declined to a reduced level that was significantly different from control mice in weeks 8–11. Control Tg(−) BxN mice never showed any indication of ankle swelling. This time frame is slightly later than that previously reported for the onset of arthritis in K/BxN mice, and may be a result of variation between mouse colonies (1). The same three litters of mice were also followed for the development of serum anti-GPI titers. Mice were serially bled once a week from 3–11 weeks, and anti-GPI titers were measured by ELISA (Fig. (Fig.11b). Serum anti-GPI Ig was detectable in K/BxN mice starting in week 4, and all mice were positive by week 5. Anti-GPI titers continued to rise through week 7, and then remained high in weeks 8–11. Anti-GPI titers were not measurable in Tg(−) BxN mice. The emergence of serum anti-GPI autoantibody correlated well with the development of arthritis in the ankles. Together, these data establish three distinct stages for disease in K/BxN mice, a prediseased stage (3–4 weeks), an acute stage of robust inflammation (5–7 weeks), and a chronic less aggressive disease stage (8–11 weeks).

Fig 1.
Kinetics of arthritis and anti-GPI Ig seroconversion. Three litters of K/BxN (○) or Tg(−) BxN (•) mice were followed from 3 to 11 weeks of age. (a) Both rear ankles were measured as an indication of arthritis. The data are represented ...

Anti-GPI B Cells Are Detectable in K/BxN Mice.

GPI can be detected at circulating levels (400 ng/ml) in the serum of mice, making it systemically available to the immune system (6). However, the arthritis that develops in K/BxN mice is localized to the front and rear paws (1). Therefore, we examined whether GPI-specific B cells could be detected in K/BxN mice during the development of disease and whether their response would show a bias toward the LNs draining the affected joints (popliteal, axillary, and brachial LNs). To follow GPI-specific B cells, we used a recombinant mouse GPI–GST fusion protein. Similar results were obtained with GPI isolated from rabbit muscle (data not shown). A control GST fusion protein was used as a negative control.

First, we looked for anti-GPI B cells in the spleen, draining LN, and nondraining LN by immunofluorescence microscopy (Fig. (Fig.2).2). Anti-GPI B cells were not detectable in the spleen and LN of K/BxN mice before disease (data not shown). As the inflammatory response began in the ankles (5–7 weeks), GPI-specific B cells could be detected within the medullary chords of the draining LN, but not in the spleen or nondraining LN (Fig. (Fig.22a). Consistent with being the site of an active immune response, the draining LNs were increased in cellularity (K/BxN 31.0 ± 9.7 × 106 cells; Tg(−) BxN 7.4 ± 2.9 × 106 cells) compared with the nondraining LNs (K/BxN 14.4 ± 5.3 × 106 cells; Tg(−) BxN 15.5 ± 4.0 × 106 cells). By 8–11 weeks, the GPI-specific B cell pool expanded, becoming more systemic, and anti-GPI B cells were found in the spleen, draining LN, and nondraining LN (Fig. (Fig.22a). The anti-GPI B cells filled up the medullary chords in the draining LN and were also found within the B cell follicle. Staining with PNA confirmed these areas to be germinal centers (data not shown). In the spleen, the anti-GPI B cells were more confined to germinal centers and the red pulp (Fig. (Fig.22a). The GPI–GST staining in the B cell follicles of the LN and spleen consisted of both a cellular staining pattern (GPI-specific germinal center B cells) and a more punctate pattern indicative of Ag-Ab complexes on follicular dendritic cells (18, 19). Specific GPI–GST staining was detected on the walls of the cortical sinus in the draining LNs, also probably due to deposited anti-GPI Ig. No staining was detected in Tg(−) BxN mice (Fig. (Fig.22a), or in sections stained with control GST fusion protein (Fig. (Fig.22b). These data demonstrated that autoreactive anti-GPI B cells could be specifically detected in the endogenous B cell repertoire of K/BxN mice during the development of disease.

Fig 2.
Localization of anti-GPI B cells in spleen and draining LN. Spleen and LN sections were stained with (red) B220 and (green) either GPI–GST or control GST and analyzed by fluorescent microscopy. The GPI–GST staining cells also express a ...

Anti-GPI B Cell Response Localizes to Draining LNs.

The histological staining suggested that the anti-GPI B cells were differentiating into antibody secreting cells (ASCs) in the spleen and LNs, especially in the draining LNs. To directly measure the number of GPI-specific B cells producing anti-GPI autoantibody, an ELISpot assay was performed with cells freshly isolated from the spleen, draining LNs, and nondraining LNs (Fig. (Fig.3).3). In agreement with serum anti-GPI Ig titers (Fig. (Fig.11b) and histological staining (Fig. (Fig.22 and data not shown), anti-GPI ASCs were undetectable in Tg(−) BxN mice and prediseased (3–4 weeks) K/BxN mice (Fig. (Fig.3).3). Importantly, anti-GPI ASCs were first evident at the time point when anti-GPI autoantibody was present in the serum and arthritis began to develop (5–7 weeks). Strikingly, the response was focused to the LNs draining the arthritic joints. Anti-GPI ASCs were elevated 5-fold in the draining LN (570.5 ± 191.9) compared with the nondraining LN (118.8 ± 46.1) or the spleen (88.0 ± 46.3). As disease progressed (8–11 weeks), anti-GPI ASC numbers continued to increase and were now detectable in both the spleen (310.4 ± 234.7) and nondraining LNs (636.8 ± 703.4). However, the vast majority of the anti-GPI ASCs still were found in the draining LN (1,952.0 ± 1,298.7). For each mouse examined, the number of ASCs was always higher in the draining LN compared with the nondraining LN or spleen. By 8–11 weeks, 10–50% of the total cells in the draining LNs of K/BxN mice were secreting anti-GPI autoantibody (Fig. (Fig.3).3). This large percentage of GPI-specific ASCs was surprising given that this response was occurring in a non-Ig Tg B cell repertoire. Together, these data show that, in K/BxN mice, the native anti-GPI B cell response is both robust and highly enriched in the LNs draining the distal joints where arthritis is found. This finding suggests that although GPI is a ubiquitously expressed Ag, it is preferentially presented to the B cells in the local LNs.

Fig 3.
Numbers of anti-GPI ASCs are elevated in draining LN. The number of anti-GPI ASCs was determined by ELISpot. Cells from the spleen ([open triangle]), draining LNs (pooled from 2 each of popliteal, axillary, and brachial LNs; •), and nondraining LNs ...

LNs Are Important for Disease Process.

The focused anti-GPI ASC response suggested that the draining LNs were important in driving the autoimmune response to GPI. To determine whether the LNs were necessary for the initiation and/or amplification of the anti-GPI response, we measured disease progression and anti-GPI B cell responses in mice deficient of LNs. Mice genetically deficient in LTα, LTβ, or LTβR are LN-deficient, but also lack distinct T and B cell areas, marginal zones, and follicular dendritic cells in the spleen. Additionally, they fail to induce GCs and produce limited Ab responses to T-dependent Ags (reviewed in refs. 20 and 21). In contrast, mice with LT-signaling blocked only during gestation by using a receptor-Ig fusion protein (LTβR-Ig) are born without LNs, but exhibit normal splenic architecture and T cell-dependent B cell responses (17, 22). Therefore, to generate K/BxN mice without LNs but with an otherwise intact immune system, NOD females, bred to KRN C57BL/6 males, were treated with LTβR-Ig at day E11 and E14 of gestation to block LT-signaling only during development. Pups from LTβR-Ig-treated mice were compared with pups from mice treated with control-Ig in the same timecourse. As previously reported, LTβR-Ig-treated mice were born without brachial, axillary, inguinal, and popliteal LNs (17) and data not shown). Control Ig-treated mice exhibited normal LN development (data not shown).

K/BxN mice treated with control Ig developed arthritis with the same kinetics and intensity as untreated K/BxN mice (compare Fig. Fig.44 with Fig. Fig.1).1). Strikingly, the development of arthritis in LTβR-Ig-treated mice was greatly delayed and reduced in severity (Fig. (Fig.44a). At 4–5 weeks of age, when control Ig-treated mice began to exhibit profound arthritis, LTβR-Ig-treated mice exhibited little to no disease. Ankle swelling was not observed in LTβR-Ig-treated mice until several weeks later, and was significantly reduced in intensity (Fig. (Fig.44a). Although an additional effect of a small amount of residual fusion protein during the first 3 weeks of life (before weaning) cannot be ruled out, the short half-life of the protein (<5 days) (17) suggests that the observed effects at 4–8 weeks were caused by the absence of LNs.

Fig 4.
Arthritis induction and serum anti-GPI levels are reduced in LTβR-Ig-treated K/BxN mice. Pregnant females were injected i.v. with LTβR-Ig (•) or control Ig (○) on days E11 and E14, and their progeny were followed starting ...

To determine whether the reduction in arthritis was reflected in the amount of anti-GPI autoantibody in the serum, LTβR-Ig- and control Ig-treated K/BxN mice were followed for development of serum anti-GPI Ig by ELISA. Control Ig-treated mice developed serum anti-GPI Ig starting at 4 weeks, and all mice were positive by 5 weeks. In contrast, in LTβR-Ig-treated mice, anti-GPI Ig was not detectable until 5–6 weeks. Furthermore, serum anti-GPI Ig titers remained 10- to 100-fold lower in LTβR-Ig-treated mice compared with control Ig-treated mice (Fig. (Fig.44b). Because LTβR-Ig-treated mice still had some serum anti-GPI Ig, it was possible that their lack of arthritis was caused by a difference in the isotype of the anti-GPI present. It has been previously demonstrated that arthritis can be transferred to naive mice by using only the IgG fraction of K/BxN serum (5). To determine whether the lack of LNs had altered the isotype of anti-GPI Ab in the serum, a subset of LTβR-Ig-treated and control Ig-treated mice were tested for anti-GPI Ig isotype by ELISA (Fig. (Fig.44c). The majority of anti-GPI Ab in the serum of control Ig-treated K/BxN mice was of the IgG1 isotype, with a lesser amount of IgG2b also present (Fig. (Fig.44c). Although at reduced titers, LTβR-Ig-treated mice had the same anti-GPI Ig isotypes (IgG1 and IgG2b) and at similar ratios in their serum (Fig. (Fig.44c). Therefore, the reduced disease severity was not simply caused by the anti-GPI autoantibody being of the wrong Ig isotype. Together, these data suggest that the LNs are important in the initial stages of disease, but not absolutely required for its initiation. However, in the absence of LNs, the anti-GPI response is delayed in time of onset and diminished in intensity, resulting in a dramatic decrease in disease severity.

Reduced serum anti-GPI Ig levels suggested that the anti-GPI B cell response was limited in LN-deficient LTβR-Ig-treated mice. To directly measure the anti-GPI B cell response, the number of anti-GPI ASCs was determined by ELISpot (Fig. (Fig.5).5). Similar to untreated K/BxN mice, anti-GPI ASCs were detectable in the spleen (77.7 ± 33.2) and nondraining LNs (148.0 ± 69.3) of 6 weeks old control Ig-treated K/BxN mice, and elevated 5-fold in draining LNs (734.7 ± 381.2) (Fig. (Fig.5).5). In contrast, anti-GPI ASCs were low to undetectable in the spleens of 6-week-old LTβR-Ig-treated K/BxN mice (18.7 ± 6.1). By 10 weeks of age, anti-GPI ASC numbers were elevated in the spleen (426.7 ± 179.5) and nondraining LNs (464.0 ± 185.2) of control Ig-treated K/BxN mice with the highest numbers still in the draining LN (1,605.3 ± 381.0). However, even in 10-week-old LTβR-Ig-treated mice, anti-GPI ASC numbers in the spleen remained low (97.3 ± 26.9). Taken together, the data from LTβR-Ig-treated mice demonstrate that LNs, most likely the draining LN, are important both in the initiation and amplification of the B cell response to GPI. This finding implicates the draining LN as an important site where the autoimmune B cell response to GPI begins in the K/BxN arthritis model.

Fig 5.
Anti-GPI ASCs are reduced in LTβR-Ig-treated K/BxN mice. The number of anti-GPI ASCs in the spleen ([open triangle]), draining LN (•), and nondraining LN (□) was determined by ELISpot as described in Fig. Fig.3.3. The data are ...


In this study, we have measured the B cell response to the autoantigen GPI in the K/BxN model of RA. The induction of arthritis followed a distinct time course that was remarkably consistent between individual mice. This uniformity and precise timing of disease onset is somewhat unusual for a spontaneous model of in vivo autoimmune disease and allowed us to track the autoreactive B cells in a cohort of mice as the disease progressed from inception, through the acute inflammation phase to the chronic state. We found that anti-GPI B cells were detectable in the naturally arising B cell repertoire of K/BxN mice by immunofluorescence microscopy and ELISpot analysis. The anti-GPI B cell response correlated well with rising serum anti-GPI autoantibody titers and disease induction. Importantly, although the autoantigen GPI was ubiquitously expressed, the B cell response was focused specifically to the lymph nodes draining the distal joints where arthritis was found. The number of B cells secreting anti-GPI autoantibody was quite high and, in fact, comprised up to half of the total cell population in the draining LNs by 8–11 weeks of age. Depletion of these LNs, using LTβR-Ig fusion protein treatment during gestation, led to reduced serum anti-GPI titers, diminished anti-GPI B cell responses, and dramatically attenuated arthritis. Therefore, this study demonstrates that the LNs draining the distal joints are a critical site for the initiation and amplification of the autoimmune B cell response to GPI.

Our finding that the anti-GPI B cell response is preferentially localized to the lymph nodes draining the distal joints is in contrast to that reported by Mathis and colleagues (23). They used a hybridoma-based approach, and recovered an increased number of anti-GPI hybridomas from the spleens compared with the LNs of diseased K/BxN mice. These data led them to conclude that the anti-GPI B cell response occurred primarily in the spleen. In the current study, we have followed primary unmanipulated B cells and assayed them directly out of the mouse. This method allowed us to directly measure the effector cells, the GPI-specific ASCs, bypassing problems of low fusion efficiency and potential bias against nondividing cells such as plasma cells inherent in hybridoma analysis (24).

This study demonstrates that the anti-GPI B cell response is enriched in the LNs draining the distal joints both at the initiation/amplification stage of inflammation and later on during the chronic phase of disease. Unlike lymphocyte responses against organ-specific autoantigens or foreign antigens immunized in a particular location, which should occur primarily in the draining LN, responses to ubiquitously expressed Ags should be detectable in all LNs and in the spleen. In models of collagen-induced arthritis in rats or Ag-induced arthritis in mice, Ag-specific lymphocyte responses were primarily detected in the draining LNs (25, 26). However, these two examples both used an immunization protocol that placed the Ag in a specific location, and the responses were measured in the LN draining the immunization site. In the K/BxN model, we followed the B cell response to endogenous GPI that is expressed at measurable levels circulating in the serum (6) and found that the autoreactive B cells expanded and produced anti-GPI autoantibody specifically in the LNs draining the distal joints. The necessity of local LNs in the development of the organ-specific disease diabetes has been recently reported (27). By removing the pancreatic LN, this study formally demonstrated the importance of the local LN in initiating the autoreactive T cell response to an organ-specific Ag. Importantly, our study adds the surprising finding that, in the K/BxN model of arthritis, the local LN is also the driving force for the amplification of the autoimmune response to a systemic Ag, GPI.

The findings of this study imply that the local immune response is caused by the unique immune physiology of the joint. An important question raised by our study is why the GPI-specific B cells are stimulated in the local LN. It is possible that the GPI that is presented to the B cells in the draining LN is at a higher local concentration or in a more immunogenic form. GPI, a crucial component of the glycolytic pathway is expressed in every cell throughout the body as an intracellular protein, but has also been detected as a secreted protein (autocrine motility factor, neuroleukin) (28, 29). In humans, higher levels of GPI protein have been measured in the synovial fluid around inflamed joints (30). Furthermore, in both normal and arthritic mice, GPI has been reported to be bound to the articular surface of the cartilage lining the joint (6). These data suggest that extracellular GPI in the joint is potentially accessible to the immune system. This extracellular GPI could then drain to the local LN via the lymphatics, or be picked up by dendritic cells and carried there (18, 31). A higher local Ag concentration could result from an increased release of GPI and other joint Ags upon normal movement of the small distal joints. Within each paw there are 29–30 synovial joints, compared with just one knee and one hip joint (32). Therefore, there is an increased amount of synovial tissue per unit volume in the small joints of the paws. Because each rear paw drains to a single popliteal LN and each front paw drains to the one axillary and brachial LN (11), the local concentration of GPI in these LNs could be slightly higher than that found in LNs that drain other locations (e.g., inguinal LNs). This increase in Ag, even if it is only 2-fold, could then trigger the autoreactive B cell response (33, 34). The concept of joint movement leading to differential Ag release from small versus large joints has been suggested previously (35, 36). It is interesting to speculate that movement-induced release of joint autoantigens could also be involved in the development of other forms of arthritis.

In this study, we have demonstrated that the LNs draining the distal joints are important for both the initiation and progression of the B cell response to the ubiquitously expressed autoantigen GPI. This autoimmune response requires anti-GPI T cells and the GPI Ag be present in addition to the anti-GPI B cells. Introduction of the KRN TCR Tg elevates the potential frequency of anti-GPI T cells in the mouse, but how these autoreactive T cells as well as the anti-GPI B cells avoided tolerance induction is not known (37–42). The selective priming of anti-GPI B cells in the LNs draining the distal joints is consistent with there being an increase in GPI Ag draining to this site. In this local LN, the KRN T cells would encounter the GPI Ag and provide help for the anti-GPI specific B cells, initiating the autoimmune response. Once this autoimmune response is underway, it can readily amplify in the local LN, producing higher titers of anti-GPI autoantibody to trigger the inflammatory response initiating joint destruction. Thus, our findings demonstrate that although GPI is a ubiquitously expressed autoantigen, the immune system responds to it as though it were an organ-specific Ag, resulting in a joint-specific autoimmune disease.


We thank Dr. J. Browning for the LTβR-Ig fusion protein, Dr. A. Shaw for the CD2AP-GST fusion protein control, D. Kreamalmeyer for technical assistance and managing the mouse colony, J. Smith for administrative assistance, and Drs. J. Erikson, L. Norian, J. P. Atkinson, and J. Browning for critical reading of the manuscript. This work was supported by grants from the National Institutes of Health. F.F.S. is supported by a grant from the Pediatric Scientist Development Program (National Institute of Child Health and Human Development).


  • ASCs, antibody secreting cells
  • GPI, glucose-6-phosphate isomerase
  • HRP, horseradish peroxidase
  • K/BxN, KRN (C57BL/6 × NOD)F1
  • LT, lymphotoxin
  • LTβR-Ig, LT-β receptor-Ig fusion protein
  • NOD, nonobese diabetic
  • RA, rheumatoid arthritis
  • LN, lymph node
  • Tg, transgenic
  • En, embryonic day n


1. Kouskoff V., Korganow, A.-S., Duchatelle, V., Degott, C., Benoist, C. & Mathis, D. (1996) Cell 87, 811-822. [PubMed]
2. Feldmann M., Brennan, F. M. & Maini, R. N. (1996) Cell 85, 307-310. [PubMed]
3. Matsumoto I., Staub, A., Benoist, C. & Mathis, D. (1999) Science 286, 1732-1735. [PubMed]
4. Basu D., Horvath, S., Matsumoto, I., Fremont, D. H. & Allen, P. M. (2000) J. Immunol. 164, 5788-5796. [PubMed]
5. Korganow A.-S., Ji, H., Mangialaio, S., Duchatelle, V., Pelanda, R., Martin, T., Degott, C., Kikutani, H., Rajewsky, K., Pasquali, J.-L., et al. (1999) Immunity 10, 451-461. [PubMed]
6. Matsumoto I., Maccioni, M., Lee, D. M., Maurice, M., Simmons, B., Brenner, M., Mathis, D. & Benoist, C. (2002) Nat. Immunol. 3, 360-365. [PubMed]
7. Ji H., Ohmura, K., Mahmood, U., Lee, D. M., Hofhuis, F. M. A., Boackle, S. A., Takahashi, K., Holers, V. M., Walport, M. J., Gerard, C., et al. (2002) Immunity 16, 157-168. [PubMed]
8. Wipke B. T. & Allen, P. M. (2001) J. Immunol. 167, 1601-1608. [PubMed]
9. Solomon S., Kolb, C., Mohanty, S., Jeisy-Walder, E., Preyer, R., Schöllhorn, V. & Illges, H. (2002) Eur. J. Immunol. 32, 644-651. [PubMed]
10. Zinkernagel R. M., Ehl, S., Aichele, P., Oehen, S., Kündig, T. & Hengartner, H. (1997) Immunol. Rev. 156, 199-209. [PubMed]
11. Tilney N. L. (1971) J. Anat. 109, 369-383. [PMC free article] [PubMed]
12. Bogen S. A., Weinberg, D. S. & Abbas, A. K. (1991) J. Immunol. 147, 1537-1541. [PubMed]
13. Chabannes D. & Borel, J. F. (1990) Transplant. Proc. 22, 2591-2593. [PubMed]
14. Höglund P., Mintern, J., Waltzinger, C., Heath, W., Benoist, C. & Mathis, D. (1999) J. Exp. Med. 189, 331-339. [PMC free article] [PubMed]
15. Lan H. Y., Nikolic-Paterson, D. J. & Atkins, R. C. (1993) Clin. Exp. Immunol. 92, 336-341. [PMC free article] [PubMed]
16. Mandik-Nayak L., Seo, S., Sokol, C., Potts, K. M., Bui, A. & Erikson, J. (1999) J. Exp. Med. 189, 1799-1814. [PMC free article] [PubMed]
17. Rennert P. D., Browning, J. L., Mebius, R., Mackay, F. & Hochman, P. S. (1996) J. Exp. Med. 184, 1999-2006. [PMC free article] [PubMed]
18. Szakal A. K., Holmes, K. L. & Tew, J. G. (1983) J. Immunol. 131, 1714-1727. [PubMed]
19. Kosco-Vilbois M. H. & Scheidegger, D. (1995) Curr. Top. Microbiol. Immunol. 201, 69-82. [PubMed]
20. Fu Y. X. & Chaplin, D. D. (1999) Annu. Rev. Immunol. 17, 399-433. [PubMed]
21. Matsumoto M. (1997) Immunol. Rev. 156, 137-144. [PubMed]
22. Yamamoto M., Rennert, P., McGhee, J. R., Kweon, M.-N., Yamamoto, S., Dohi, T., Otake, S., Bluethmann, H., Fujihashi, K. & Kiyono, H. (2000) J. Immunol. 164, 5184-5191. [PubMed]
23. Maccioni M., Zeder-Lutz, G., Huang, H., Ebel, C., Gerber, P., Hergueux, J., Marchal, P., Duchatelle, V., Degott, C., van Regenmortel, M., et al. (2002) J. Exp. Med. 195, 1071-1077. [PMC free article] [PubMed]
24. Shen G. L., Liu, E. X. & Wu, A. R. (1986) Sci. Sin. Ser. B 29, 165-172. [PubMed]
25. Rahman J. & Staines, N. A. (1991) Clin. Exp. Immunol. 85, 48-54. [PMC free article] [PubMed]
26. Petrow P. K., Thoss, K., Henzgen, S., Katenkamp, D. & Bräuer, R. (1996) J. Autoimmun. 9, 629-635. [PubMed]
27. Gagnerault M.-C., Luan, J. J., Lotton, C. & Lepault, F. (2002) J. Exp. Med. 196, 369-377. [PMC free article] [PubMed]
28. Gurney M. E., Apatoff, B. R., Spear, G. T., Baumel, M. J., Antel, J. P., Bania, M. B. & Reder, A. T. (1986) Science 234, 574-581. [PubMed]
29. Haga A., Niinaka, Y. & Raz, A. (2000) Biochim. Biophys. Acta 1480, 235-244. [PubMed]
30. Schaller M., Burton, D. R. & Ditzel, H. J. (2001) Nat. Immunol. 2, 746-753. [PubMed]
31. Thomas R. & Lipsky, P. E. (1996) Immunol. Today 17, 559-564. [PubMed]
32. Popesko P., Rajtova, V. & Horak, J., (1992) A Colour Atlas of the Anatomy of Small Laboratory Animals: Rat, Mouse, Golden Hamster (Wolfe, London).
33. Batista F. D. & Neuberger, M. S. (1998) Immunity 8, 751-759. [PubMed]
34. Mongini P. K., Blessinger, C. A., Highet, P. F. & Inman, J. K. (1992) J. Immunol. 148, 3892-3901. [PubMed]
35. Simkin P. A. & Benedict, R. S. (1990) Arthritis Rheum. 33, 73-79. [PubMed]
36. Simkin P. A. (1995) Scand. J. Rheumatol. 24, 13-16. [PubMed]
37. Melamed D. & Nemazee, D. (1999) Curr. Dir. Autoimmun. 1, 1-30. [PubMed]
38. Kouskoff V. & Nemazee, D. (2001) Life Sci. 69, 1105-1113. [PubMed]
39. Goodnow C. C. (1996) Proc. Natl. Acad. Sci. USA 93, 2264-2271. [PMC free article] [PubMed]
40. von Boehmer H. & Jaeckel, E. (2001) Adv. Exp. Med. Biol. 490, 41-48. [PubMed]
41. Hanahan D. (1990) Annu. Rev. Cell Biol. 6, 493-537. [PubMed]
42. Kishimoto H. & Sprent, J. (2000) Clin. Immunol. 95, S3-S7. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...