T cell responses to streptococcal antigens in rats: relation to susceptibility to streptococcal cell wall-induced arthritis

In order to investigate the immunological mechanism of the chronic phase of streptococcal cell wall (SCW)-induced arthritis in Lewis rats, we compared the SCW-specific T cell response in arthritis-susceptible (female Lewis) and resistant (F344) rats. We present evidence that this T cell response is absent in F344 rats, while it is clearly present in Lewis rats. The T cell response was analyzed both in the spleen and in lymph nodes. In addition, we show, that injection of SCW in the F344 rat induces a general unresponsiveness in this strain: the response to mitogen was severely suppressed in SCW-injected F344 rats and, furthermore, when SCW was coinjected with ovalbumin, the response to ovalbumin was depressed. The fact that priming with ovalbumin alone induces a normal response in the F344 rat to both mitogen and ovalbumin implies that the observed abnormality after SCW priming is not a general immunological defect in this strain. Additionally, we demonstrate that adherent cells of both Lewis and F344 exert negative effects on an in vitro T cell response after injection with SCW, and that F344-adherent cells are more potent in this effect. Removal of OX8-positive cells leads to a restoration of the SCW-specific T cell response in SCW-injected F344 rats, indicating that the expression of this response is controlled by (SCW-specific?) suppressor T cells. Our results provide suggestive evidence for the obligatory role of SCW-specific T cells in the expression of chronic joint inflammation after systemic injection of SCW.     

Inhibitors of reactive oxygen intermediates suppress bacterial cell wall-induced arthritis.

Peritoneal and peripheral blood monocyte-macrophages from inbred Lewis (LEW) rats generate higher levels of reactive oxygen intermediates (ROI) in response to group A streptococcal cell walls (SCW) than do similar populations of cells from histocompatible Fischer rats. This differential sensitivity of the phagocytes to SCW is reflected in differences in susceptibility of the two strains to the development of arthritis in response to SCW. After systemic administration of the SCW, LEW rats develop acute and chronic erosive polyarthritis, whereas the Fischer rats are arthritis resistant. Inasmuch as these data suggested that the SCW-induced release of inflammatory cell products such as ROI might be an important contributory factor in the pathogenesis of arthritis in the LEW rats, the animals were injected with SCW and treated with ROI inhibitors. A single intraarticular injection of superoxide dismutase or catalase significantly reduced the SCW-induced inflammatory response and evolution of erosive arthritis in the treated animals (articular index 3.6 +/- 0.36 for SCW only vs 1.4 +/- 0.3 for SCW + SOD; p less than 0.001; n = 6). These data indicate that ROI play a pivotal role in synovitis and, furthermore, that suppression of these inflammatory mediators modulates both acute and chronic SCW-induced inflammation of the joint.     

Endothelial cells and the pathogenesis of rheumatoid arthritis in humans and streptococcal cell wall arthritis in Lewis rats

Endothelial cells play a fundamental role in the pathogenesis of chronic inflammatory arthritis in humans such as rheumatoid arthritis (RA), as well as experimental animal models such as streptococcal cell wall (SCW) arthritis in Lewis (LEW/N) rats. This review summarizes data in support of this concept. The earliest apparent abnormalities in synovial tissues of patients with RA and Lewis rats with SCW arthritis appear to reflect microvascular endothelial cell activation or injury. At the molecular level, the abnormalities include enhanced expression by endothelial cells of activation markers such as class II major histocompatibility complex antigens, phosphotyrosine, leukocyte adhesion molecules, oncoproteins such as c-Fos and c-Myc, and metalloproteinases such as collagenase and transin/stromelysin. The development of severe, chronic, destructive arthritis is dependent upon thymic-derived lymphocytes and is accompanied by tumorlike proliferation of cells in the synovial connective tissue stroma (blood vessels and fibroblastlike cells), which results in resorptive destruction of bone and cartilage. Multiple criteria support the analogy to a neoplastic process. Paracrine and autocrine factors such as interleukin-1 (IL-1), platelet-derived growth factor (PDGF), transforming growth factor-beta (TGF-beta), and heparin-binding fibroblast growth factors (HBGF, FGF) appear to play important roles in the generation of these lesions. Finally, in addition to the autocrine and paracrine regulatory factors, neuroendocrine factors, particularly the hypothalamic-pituitary-adrenal axis, appear to be involved in the counterregulation of the inflammatory process. The counterregulatory effects are mediated, in part, by inhibition of endothelial cell activation by corticosteroids.     

Genetic resistance in experimental autoimmune encephalomyelitis. I. Analysis of the mechanism of LeR resistance using radiation chimeras

Experimental autoimmune encephalomyelitis (EAE) is a cell-mediated autoimmune disease of the central nervous system that has been extensively studied in the rat. The Lewis rat is highly susceptible to the induction of EAE, while the Lewis resistant (LeR) rat is known to be resistant. In this paper, we demonstrate that the LeR rat, which was derived from the Lewis strain by inbreeding of fully resistant animals, is histocompatible with the Lewis strain. Radiation chimeras, a tool for distinguishing between immunologic and nonimmunologic resistance mechanisms, were utilized to analyze the cellular mechanisms involved in genetic resistance to EAE. By transplanting bone marrow cells from LeR rats into irradiated Lewis recipients, Lewis rats were rendered resistant to EAE induction. Likewise, transplanting Lewis bone marrow cells into irradiated LeR recipients rendered LeR rats susceptible. Mixed lymphoid cell chimeras using bone marrow, spleen, and thymus cells in Lewis recipient rats revealed individual lymphoid cell types and cell interactions that significantly affected the incidence and severity of EAE. Our results suggest that LeR resistance is mediated by hematopoietic/immune cells, and that cells located in the spleen appear to play a critical role in the resistance/susceptibility to EAE induction. Depletion of splenic adherent cells did not change the patterns of EAE resistance. In vivo cell mixing studies suggested the presence of a suppressor cell population in the LeR spleen preparations which exerted an inhibitory effect on Lewis autoimmune responses. Thus, the mechanism of LeR resistance appears to be different from that in other EAE-resistant animals.     

Environmental modulation of autoimmune arthritis involves the spontaneous microbial induction of T cell responses to regulatory determinants within heat shock protein 65

Both genetic and environmental factors are believed to be involved in the induction of autoimmune diseases. Adjuvant arthritis (AA) is inducible in susceptible rat strains by injection of Mycobacterium tuberculosis, and arthritic rats raise T cell responses to the 65-kDa mycobacterial heat-shock protein (Bhsp65). We observed that Fischer 344 (F344) rats raised in a barrier facility (BF-F344) are susceptible to AA, whereas F344 rats maintained in a conventional facility (CV-F344) show significantly reduced incidence and severity of AA, despite responding well to the arthritogenic determinant within Bhsp65. The acquisition of protection from AA can be circumvented if rats are maintained on neomycin/acidified water. Strikingly, naive unimmunized CV-F344 rats but not BF-F344 rats raised T cell responses to Bhsp65 C-terminal determinants (BCTD) (we have previously shown that BCTD are involved in regulation of acute AA in the Lewis rat); however, T cells of naive CV-F344 and BF-F344 gave a comparable level of proliferative response to a mitogen, but no response at all to an irrelevant Ag. Furthermore, adoptive transfer into naive BF-F344 rats of splenic cells of naive CV-F344 rats (restimulated with BCTD in vitro) before induction of AA resulted in a considerably reduced severity of AA. These results suggest that spontaneous (inadvertent) priming of BCTD-reactive T cells, owing to determinant mimicry between Bhsp65 and its homologues in microbial agents in the conventional environment, is involved in modulating the severity of AA in CV-F344 rats. These results have important implications in broadening understanding of the host-microbe interaction in human autoimmune diseases.     

The non-MHC quantitative trait locus Cia5 contains three major arthritis genes that differentially regulate disease severity, pannus formation, and joint damage in collagen- and pristane-induced arthritis

Cia5 is a locus on rat chromosome 10 which regulates the severity of collagen- and pristane-induced arthritis (CIA and PIA). To refine the region toward positional identification, Cia5 subcongenic strains were generated and studied in PIA and CIA. The protective effect of the telomeric locus Cia5a was confirmed in both models. A second arthritis severity locus (Cia5d) was identified within the most centromeric portion of Cia5. DA.F344(Cia5d) rats had a significantly lower median arthritis severity index in PIA, but not in CIA, compared with DA. On histologic analyses DA.F344(Cia5a) and DA.F344(Cia5d) congenics with PIA preserved a nearly normal joint architecture compared with DA, including significant reduction in synovial hyperplasia, pannus, angiogenesis, inflammatory infiltration, bone and cartilage erosions. Cia5 and Cia5a synovial levels of IL-1beta mRNA were reduced. Although both DA.F344(Cia5) and DA.F344(Cia5a) rats were protected in CIA, the arthritis scores of DA.F344(Cia5) were significantly higher than those of DA.F344(Cia5a), suggesting the existence of a third locus where F344-derived alleles centromeric from Cia5a contribute to increased arthritis severity. The existence of the third locus was further supported by higher levels of autoantibodies against rat type II collagen in DA.F344(Cia5) congenics compared with DA.F344(Cia5a). Our results determined that Cia5 contains three major arthritis severity regulatory loci regulating central events in the pathogenesis of arthritis, and differentially influencing CIA and PIA. These loci are syntenic to regions on human chromosomes 17q and 5q implicated in the susceptibility to rheumatoid arthritis, suggesting that the identification of these genes will be relevant to human disease.     

Resistance to adjuvant arthritis is due to protective antibodies against heat shock protein surface epitopes and the induction of IL-10 secretion

Adjuvant arthritis (AA) is an experimental model of autoimmune arthritis that can be induced in susceptible strains of rats such as inbred Lewis upon immunization with CFA. AA cannot be induced in resistant strains like Brown-Norway or in Lewis rats after recovery from arthritis. We have previously shown that resistance to AA is due to the presence of natural as well as acquired anti-heat shock protein (HSP) Abs. In this work we have studied the fine specificity of the protective anti-HSP Abs by analysis of their interaction with a panel of overlapping peptides covering the whole HSP molecule. We found that arthritis-susceptible rats lack Abs to a small number of defined epitopes of the mycobacterial HSP65. These Abs are found naturally in resistant strains and are acquired by Lewis rats after recovery from the disease. Active vaccination of Lewis rats with the protective epitopes as well as passive vaccination with these Abs induced suppression of arthritis. Incubation of murine and human mononuclear cells with the protective Abs induced secretion of IL-10. Analysis of the primary and tertiary structure of the whole Mycobacterium tuberculosis HSP65 molecule indicated that the protective epitopes are B cell epitopes with nonconserved amino acid sequences found on the outer surface of the molecule. We conclude that HSP, the Ag that contains the pathogenic T cell epitopes in AA, also contains protective B cell epitopes exposed on its surface, and that natural and acquired resistance to AA is associated with the ability to respond to these epitopes.     

Type II collagen-induced murine arthritis: induction of arthritis depends on antigen-presenting cell function as well as susceptibility of host to an anticollagen immune response.

Two sets of ((resistant x susceptible) F1----parent) and (parent----F1) chimeric mice were prepared. In the chimeric combinations involving BALB/c and DBA/1 mice, all (F1----F1) chimeras developed arthritis as well as potent anticollagen responses after immunization with collagen, whereas all (F1----BALB/c) and (BALB/c----F1) chimeras induced neither arthritis nor immune responses. This type of F1 T cells could be activated with APC from DBA/1 but not from BALB/c mice. Thus, the failure of the [F1 in equilibrium with BALB/c] chimeras to mount anticollagen responses was due to a defect at the APC level. Another arthritis-resistant strain, C57BL/6, exhibited adequate APC function, but reduced T cell responsiveness, representing an intermediate responder. In the chimeric combinations involving C57BL/6 and DBA/1 mice, (F1----F1) and (C57BL/6----C57BL/6) chimeras developed very high and very low incidence of arthritis, respectively. (C57BL/6----F1) chimeras developed an appreciable incidence of arthritis under conditions in which this group of chimeras generated intermediate levels of anticollagen responses. In contrast, (F1----C57BL/6) chimeras developed low incidence of disease despite induction of strong responses. Moreover, cells from collagen-immunized (F1----C57BL/6) chimeras, when transferred into T cell-depleted B cell mice of F1 or C57BL/6 strain, produced comparable immune responses in both groups but induced much more severe arthritis in F1 than in C57BL/6 recipients. These results indicate that: i) two types of arthritis-resistant strains can be identified, each of which has anticollagen APC defect as a low responder and reduced T cell responsiveness as an intermediate responder and ii) a discrepancy between the degree of anticollagen responses and clinical arthritis is attributed to the differential susceptibility to anticollagen immune responses.     

Comparison of rat mandible bone characteristics in f344 substrains, f344/du and f344/n.

The characteristics of the mandible bone were compared through DXA methods between two major substrains of F344 rats, F344/DuCrlCrlj and F344/NSlc at around 60 days of age. Since these two substrains are clearly different in survival and mandible morphology, some genetic differences are supposed to exist. In contrast to a previous microsatellite analysis, clear and significant differences were detected in the body and mandible weights, the mandible bone mineral contents (BMC), bone area (AREA), bone mineral density (BMD) and bone mineral ratio (BMR), between F344/DuCrlCrlj and F344/NSlc, with the mandible molar teeth intact in the bone. Thus, care is needed in the experimental use of these substrains, as results may differ between them. The newly proposed parameter, BMR, may especially contribute to the comparison of bone characteristics among species.     

Osteosclerosis in F344/DuCrj rats

Osteosclerosis was observed in the tibia and sternum in F344/DuCrj rats of both sexes at 6, 18 and 30 months of age. The lesion first seen was a proliferation of osteogenic tissues on the marrow surface of the cortical bone and bone trabeculae, resulting in replacement of the marrow cavity by lamellar bone. Most of the affected rats had associated degenerative osteoarthrosis and regressive changes of the growth plate. Osteosclerosis was considered to be an aging change, lesions were observed at 6 months and increased in frequency with age.     

Differences of mu-opioid receptors between Lewis and F344 rats

F344 and Lewis rats show different responses to opioids in several experimental paradigms. In this study we have used the specific mu-opioid agonist DAMGO to find out if these differences could be attributed to heterogeneity of mu-opioid receptors. The density of [H3]DAMGO binding sites was similar in the brain cortex and spinal cord of both strains, but DAMGO affinity for mu-opioid receptors was higher in F344 tissues. Moreover, a parallel study of the effects of DAMGO on electrically-evoked twitches of isolated vasa deferentia revealed that this drug was also more effective in F344 preparations. These results suggest that mu-opioid receptors of F344 rats are more sensitive to pharmacological stimulation in vitro, which could be related to a higher drug affinity.     

Comparative study of alpha2-adrenoceptors in Fischer 344 and Lewis rats. Evidence for clonidine-induced place aversion

Fischer 344 (F344) and Lewis rat strains have been shown to exhibit different vulnerability to development or maintenance of opioid seeking behaviours probably due to differences in the endogenous opioid system. Since opioid and alpha(2)-adrenergic mechanisms closely interact in nociception and substance abuse, strain differences may be expected to affect alpha(2)-adrenoceptor-mediated events. The sensitivity of these two strains to alpha(2)-adrenoceptor-mediated antinociception has been reported to be markedly different. In this work we have further studied the function of alpha(2)-adrenoceptors in F344 and Lewis rats by means of several in vivo and in vitro procedures. Comparative studies of [(3)H]RX821002 and [(35)S]GTPgammaS binding revealed that alpha(2)-adrenoceptors could be slightly more responsive to agonist stimulation in the brain cortex of F344 rats, which is in agreement with previous antinociception studies. However, these differences were modest, not observed in the spinal cord and did not translate into functional differences concerning the effects of clonidine on vas deferens contractility and body temperature. Conditioning experiments showed that a moderate dose of clonidine, which is relevant in antinociceptive and opioid antiwithdrawal studies, induces a robust place aversion which is also equivalent in F344 and Lewis rats. This finding underlies the consistency of the effect and its independency of genetic differences between both rat strains. It seems therefore that the pharmacological properties of alpha(2)-adrenoceptors are similar in F344 and Lewis rats, and thus the previously reported differences in clonidine-induced antinociception could be attributed to other factors such as dissimilar endogenous function of specific noradrenergic pathways.     

Early high expression of IP-10 in F344 rats resistant to Sendai virus-induced airway injury

Weanling F344 and BN rats differ markedly in their susceptibility to Sendai virus-induced airway injury. Early gene expression that controls their differences in susceptibility remains poorly understood. In this study we combined suppressive subtractive hybridization and cDNA library array hybridization to identify genes differentially expressed in virus-susceptible BN and virus-resistant F344 rats during the first 3 days after inoculation. Differential expression of selected clones was further verified by quantitative RT-PCR. Seven virus-induced gene segments were identified. Of them, interferon-gamma-inducible protein 10 (IP-10), Mx1, and guanylate-binding protein-2 mRNA abundance in infected F344 rats was 201.5, 188.2, and 281.7% higher, respectively, than that of infected BN rats at 2 days after inoculation. In situ hybridization indicated that virus-induced IP-10 was expressed mainly in airway epithelial cells of F344 rats. Sendai virus infection can directly induce IP-10 expression in rat tracheal epithelial cells in vitro. IP-10 early high expression might contribute to the resistance to virus-induced airway disease in F344 rats by promoting Th1 responses and increasing antiviral activity.     

Treatment of adjuvant arthritis with granulocyte-colony stimulating factor and peptide derived from heat shock protein 65

Adjuvant arthritis in Lewis rats is induced by the subcutaneous injection of Mycobacterium tuberculosis in mineral oil, and the predominant T cell immune reactivity is against the heat shock protein 65 derived peptide 176-190. We treated Lewis rats with human recombinant G-CSF followed by (i.v) administration of peptide 176-190 after induction of adjuvant arthritis (AA), and observed decreased disease severity, joint destruction, new bone formation and joint ankylosis. Treatment with G-CSF alone was also effective, but to a lesser extent. In addition, we found that splenocytes from rats treated with G-CSF had reduced antigen presenting capacity compared with splenocytes from vehicle treated rats. Primed lymph node cells from G-CSF plus peptide treated rats showed a marked reduction in proliferation and secretion of IFN-gamma after stimulation with the heat shock protein peptide in vitro as compared to controls.     

The regulatory C-terminal determinants within mycobacterial heat shock protein 65 are cryptic and cross-reactive with the dominant self homologs: implications for the pathogenesis of autoimmune arthritis

The 65-kDa mycobacterial heat shock protein (Bhsp65) has been invoked in the pathogenesis of both adjuvant arthritis (AA) in the Lewis rat (RT.1(l)) and human rheumatoid arthritis. Arthritic Lewis rats in the late phase of AA show diversification of the T cell response to Bhsp65 C-terminal determinants (BCTD), and pretreatment of naive Lewis rats with a mixture of peptides representing these neoepitopes affords protection against AA. However, the fine specificity and physiologic significance of the BCTD-directed T cell repertoire, and the role of homologous self (rat) hsp65 (Rhsp65), if any, in spreading of the T cell response to Bhsp65 have not yet been examined. We observed that T cells primed by peptides comprising BCTD can adoptively transfer protection against AA to the recipient Lewis rats. However, these T cells can be activated by preprocessed (peptide) form of BCTD, but not native Bhsp65, showing that BCTD are cryptic epitopes. The BCTD-reactive T cells can be activated by the naturally generated (dominant) C-terminal epitopes of both exogenous and endogenous Rhsp65 and vice versa. Furthermore, certain individual peptides constituting BCTD and their self homologs can also induce protection against AA. These results support a model for the diversification of T cell response to Bhsp65 during the course of AA involving up-regulation of the display of cryptic BCTD coupled with spontaneous induction of T cell response to the cross-reactive dominant C-terminal epitopes of Rhsp65. The identification of disease-regulating cryptic determinants in Ags implicated in arthritis provides a novel approach for immunotherapy of rheumatoid arthritis.     

Investigation of benzene oxide in bone marrow and other tissues of F344 rats following metabolism of benzene in vitro and in vivo.

This study examines the initial activation of benzene, exploring key aspects of its metabolism by measurement of benzene oxide (BO) and BO-protein adducts in vitro and in vivo. To assess the potential influence of various factors on the production of BO, microsomes were prepared from tissues that were either targets of benzene toxicity, i.e. the bone marrow and Zymbal glands, or not targets, i.e. liver and kidneys, of control and acetone-treated F344 rats. No BO or phenol was detected in microsomal preparations of bone marrow or Zymbal glands (less than 0.007 nmol BO/mg protein and 0.7 nmol phenol/mg protein). On the other hand, BO and phenol were readily detected in preparations of liver and kidney microsomes and acetone pretreatment resulted in a 2-fold (kidney) increase or 3.7-fold (liver) increase in production of these metabolites. Initial rates of BO production in the liver isolates were 30 (control) to 50 (acetone-treated) times higher than in the corresponding kidney tissues. The estimated half-life of BO in bone marrow homogenates was 6.0 min and the second-order reaction rate constant was estimated to be 1.35 x 10(-3) l (g bone marrow)(-1) (h)(-1). These kinetic constants were used with measurements of BO-bone marrow adducts in F344 rats, receiving a single gavage dosage of 50-400 mg benzene (kg body weight)(-1) (McDonald, T.M., et al. (1994), Cancer Res. 54, 4907-4914), to predict the bone marrow dose of BO. Among the rats receiving 400 mg (kg body weight) (-1), a BO dose of 1.13 x 10(3) nM BO-h was estimated for the bone marrow, or roughly 40% of the corresponding blood dose predicted from BO-albumin adducts. Together these data suggest that, although BO is not produced at detectable levels in the bone marrow or Zymbal glands of F344 rats, BO is rapidly distributed via the bloodstream to these tissues where it may play a role in toxicity.     

Orchiectomized Fischer 344 male rat models body composition in hypogonadal state

The hypogonadal state in men is accompanied by substantial decreases in muscle and bone mass and by an increase in adiposity. Most of the strains of orchiectomized (ORX) rat that have been used to model this state display substantial losses in bone, but only subtle changes in adiposity and muscle mass. In order to identify a rat model displaying a robust catabolic response to ORX, we studied three strains: Fischer 344 (F344), Brown Norway and Wistar. ORX caused a significant and sustained decrease in weight gained by F344, but only a trend toward reduced weight gain in Brown Norway rats and a modest reduction weight gain in Wistar rats that was significant only after 56days. ORX suppressed food intake in F344 rats, and to a lesser degree in Brown Norway and Wistar rats. ORX reduced muscle mass significantly in F344 rats, but not in Brown Norway or Wistar rats. ORX increased adiposity moderately in F344 rats and substantially in Wistar rats. ORX caused a marked reduction in prostate mass and increase in bone resorption in all three strains. Thus, F344 was the only strain in which ORX produced substantial decreases in food intake, body weight and muscle mass with increased adiposity and increased bone resorption. We conclude that the F344 rat displays a broad range of catabolic effects following ORX and is the best rat model for studying the androgenic pathway and strategies for reversing catabolic changes induced by hypogonadism.     

Streptococcal cell wall arthritis: studies with nude (athymic) inbred Lewis rats

Group A streptococcal cell walls, upon intraperitoneal administration, fail to induce a chronic arthritis in athymic inbred Lewis rats (mu/mu). In contrast, heterozygous euthymic littermates (+/mu) develop a chronic arthritis upon administration of the cell wall material. Chronic arthritis can be readily induced in athymic rats if they are reconstituted intravenously with spleen cells derived from normal heterozygous euthymic littermates. Both athymic and euthymic rats develop the acute arthritis elicited by cell walls. These studies indicate that the requirements in the host for the induction of acute and chronic arthritis are different. Induction of acute arthritis is not dependent on functional T lymphocytes whereas the induction of chronic arthritis is dependent on functional T cells.     

Expression of CD80/86 on B cells is essential for autoreactive T cell activation and the development of arthritis

Depletion of B cells in rheumatoid arthritis is therapeutically efficacious. Yet, the mechanism by which B cells participate in the inflammatory process is unclear. We previously demonstrated that Ag-specific B cells have two important functions in the development of arthritis in a murine model of rheumatoid arthritis, proteoglycan (PG)-induced arthritis (PGIA). PG-specific B cells function as autoantibody-producing cells and as APCs that activate PG-specific T cells. Moreover, the costimulatory molecule CD86 is up-regulated on PG-specific B cells in response to stimulation with PG. To address the requirement for CD80/CD86 expression on B cells in the development of PGIA, we generated mixed bone marrow chimeras in which CD80/CD86 is specifically deleted on B cells and not on other APC populations. Chimeras with a specific deficiency in CD80/CD86 expression on B cells are resistant to the induction of PGIA. The concentration of PG-specific autoantibody is similar in mice sufficient or deficient for CD80/86-expressing B cells, which indicates that resistance to PGIA is not due to the suppression of PG-specific autoantibody production. CD80/86-deficient B cells failed to effectively activate PG-specific autoreactive T cells as indicated by the failure of T cells from PG-immunized CD80/86-deficient B cell chimeras to transfer arthritis into SCID mice. In vitro secondary recall responses to PG are also dependent on CD80/86-expressing B cells. These results demonstrate that a CD80/86:CD28 costimulatory interaction between B cells and T cells is required for autoreactive T cell activation and the induction of arthritis but not for B cell autoantibody production.