Naturally processed T cell epitopes from human glutamic acid decarboxylase identified using mice transgenic for the type 1 diabetes-associated human MHC class II allele, DRB1*0401.

The identification of class II binding peptide epitopes from autoimmune disease-related antigens is an essential step in the development of antigen-specific immune modulation therapy. In the case of type 1 diabetes, T cell and B cell reactivity to the autoantigen glutamic acid decarboxylase 65 (GAD65) is associated with disease development in humans and in nonobese diabetic (NOD) mice. In this study, we identify two DRB1*0401-restricted T cell epitopes from human GAD65, 274-286, and 115-127. Both peptides are immunogenic in transgenic mice expressing functional DRB1*0401 MHC class II molecules but not in nontransgenic littermates. Processing of GAD65 by antigen presenting cells (APC) resulted in the formation of DRB1*0401 complexes loaded with either the 274-286 or 115-127 epitopes, suggesting that these naturally derived epitopes may be displayed on APC recruited into pancreatic islets. The presentation of these two T cell epitopes in the islets of DRB1*0401 individuals who are at risk for type 1 diabetes may allow for antigen-specific recruitment of regulatory cells to the islets following peptide immunization.

Responses of NOD congenic mice to a glutamic acid decarboxylase-derived peptide.

Type 1 diabetes in man and the NOD (H-2g7) mouse is frequently associated with an autoimmune response to two isoforms of glutamic acid decarboxylase (GAD), GAD65 and GAD67. GAD-specific autoantibodies produced by B cells and GAD-specific T cells have been observed in both species. In the current study, the response to a GAD65-derived peptide, GAD65 524-543, previously reported to be an epitope recognized by spleen cells obtained from 3-week-old NOD mice, was assessed in NOD MHC and non-MHC congenic strains. Although spontaneous reactivity to GAD65 524-543 was not observed in NOD mice, the peptide was immunogenic in NOD mice as well as in two NOD congenic strains which are both H-2g7, B10.H-2g7 and NOD.B6Il2-Tshb. This was surprising since the B10.H-2g7 strain does not develop diabetes or insulitis and fewer than 3% of NOD.B6Il2-Tshb mice develop diabetes. The response to GAD65 524-543 was shown to be controlled by the MHC since neither the B10 nor the NOD.H-2b strain, both of which are H-2b, responded to the peptide. This study demonstrates that T cell responsiveness to GAD-derived peptides can be elicited in strains of mice that are resistant to the development of spontaneous diabetes, suggesting that peripheral tolerance to GAD is not associated with protection from diabetes.

Presentation of antigen by B cells subsets. I. Lyb-5+ and Lyb-5- B cells differ in ability to stimulate antigen specific T cells.

We have examined the antigen presenting cell (APC) function of different B cells. Resident, peritoneal B cells from normal mice were more efficient than splenic B cells in presenting antigen to CD4+ T cell lines. Peritoneal B cells from X-linked immunodeficient (Xid) mice, by contrast, stimulated no detectable responses. Xid splenic B cells were much less efficient APC than normal splenic B cells. B cells from neonatal mice also were very poor APC until the mice were 3 to 4 weeks old. Xid B cells presented antigen to T cell hybridomas as well as normal B cells showing that they process antigen normally. Thus, the defect is most likely in providing secondary signals. The ability of B cells to present antigen efficiently correlates with the percentage of B cells reported to express the Lyb-5 antigen. Anti-Lyb-5 serum and complement abrogated the APC activity of B cells suggesting that Lyb-5+, but not Lyb-5- cells are efficient APC. We also found that activated and resting normal splenic B cells, separated by buoyant density, presented antigen equally. Both populations also contained Lyb-5+ B cells although they were a larger fraction of the activated cells. Lyb-5 is now thought to be an activation antigen rather than a differentiation antigen. If this idea is correct, then our data indicate that anti-Lyb-5 more cleanly separates activated and resting B cells than buoyant density techniques.

Comparison of the T cell receptors on insulin-specific hybridomas from insulin transgenic and nontransgenic mice. Loss of a subpopulation of self-reactive clones.

Transgenic mice expressing the human insulin gene do not produce insulin-specific antibody after injection of human insulin. Nevertheless, they have some peripheral T cells that proliferate to human insulin in vitro. To investigate the nature of these T cells, human insulin-specific T cell hybridomas were produced from transgenic and nontransgenic mice. Transgenic hybridomas required more insulin to achieve maximum responses and they produced lower levels of lymphokines than nontransgenic hybridomas. The majority of nontransgenic hybridomas recognized only human and pork insulin whereas transgenic hybridomas recognized beef, sheep, and/or horse insulin in addition to human and pork insulin. The TCR expressed by transgenic and nontransgenic hybridomas were determined by Northern analysis. Both types of hybridomas used several different V alpha and V beta gene families and no favored association between V alpha and V beta gene usage was detected in either type. V beta 1 was used by 7 of 16 nontransgenic hybridomas but only by 1 of 16 transgenic hybridomas. V beta 6 receptors were predominantly expressed by the transgenic hybridomas and all V beta 6-bearing hybridomas recognized beef as well as human insulin. The differences in Ag reactivity and TCR gene usage suggest that V beta 1-bearing human insulin-reactive T cells were clonally deleted or inactivated in the transgenic animal. Other clones, representing a minor subpopulation in nontransgenic mice, were recovered from transgenic mice.

A peripheral mechanism preserves self-tolerance to a secreted protein in transgenic mice.

We have examined mechanisms of tolerance to circulating self-proteins in mice that are transgenic for human insulin. Normal, nontransgenic mice develop serum antibody responses when injected with human insulin in CFA; syngeneic transgenic mice do not. B cell responsiveness was assessed by immunizing with human insulin coupled to a T-independent Ag, Brucella abortus. No differences were found in the numbers of insulin-specific splenic plaque-forming cells between transgenic and nontransgenic mice suggesting that insulin-specific B cells are not tolerant in transgenic mice. Similarly, APC from transgenic and nontransgenic mice display no differences in their ability to process and present human insulin to human insulin-specific T cells in vitro. However, marked differences were detected between transgenic and nontransgenic T cells. Lymph node T cells from transgenic mice primed with human insulin provided no detectable helper activity for secondary antibody responses to human insulin whereas, lymph node T cells from nontransgenic mice did. Nevertheless, lymph node T cells from transgenic mice developed significant proliferative responses to human insulin. Lymph node T cells obtained from transgenic and nontransgenic mice were fused to BW5147 and human insulin-specific T cell hybridomas were generated. The fact that human insulin-specific T cell hybridomas were obtained from the transgenic mice suggests that these T cells were not clonally deleted. In addition, APC from transgenic mice did not stimulate human insulin-specific hybridomas from normal mice in the absence of exogenous insulin. We suggest that T cells specific for human insulin are not deleted in the thymus of transgenic mice because APC in the thymus do not bear the requisite levels of endogenous human insulin/Ia complexes. Therefore, we conclude that tolerance in the transgenic mice is preserved by peripheral mechanisms.

Tolerance induced by physiological levels of secreted proteins in transgenic mice expressing human insulin.

We have used transgenic mice to study immune tolerance to autologous, non-MHC encoded proteins that are expressed at physiological levels in the circulation. The transgenic mice used in these studies express the human preproinsulin gene and synthesize human proinsulin. Human and mouse insulin are secreted from the pancreatic islets of transgenic mice in response to normal physiological stimuli, such as glucose. Our data demonstrate that the transgenic mice have acquired tolerance to human insulin. The repertoire of T cells specific for exogenous antigens is shaped by the acquired tolerance to autologous proteins since pork but not beef or sheep insulin is also nonimmunogenic in the transgenic mice. We also found that the transgenic mice were tolerant to human proinsulin, the intracellular precursor of insulin. Unresponsiveness to human proinsulin most likely results from tolerance of insulin-specific and proinsulin-specific T cells that recognize the secreted enzymatic cleavage products of proinsulin, insulin and C-peptide.

Helper T-cell clones that recognize autologous insulin are stimulated in nonresponder mice by pork insulin.

Murine antibody responses to various species of insulin are under major histocompatibility complex-linked Ir gene control. Beef insulin differs from pork insulin by only two amino acids in the A-chain loop, yet strain C57BL/10 (B10) mice produce insulin-specific antibodies after immunization with beef insulin and fail to produce antibody after stimulation with pork insulin. Nevertheless, pork insulin primes helper T cells in B10 mice that can be demonstrated if insulin-specific Lyt-1-, -2+ suppressor T cells are removed. Not only do the pork insulin-primed helper and suppressor T cells cross-react with autologous insulin, but also rat insulin (the amino acid sequence of which is identical to mouse insulin) elicits functionally identical helper and suppressor T cells. In this report, we demonstrate that in B10 mice the frequency of helper T cells stimulated by pork insulin is equivalent to that stimulated by beef insulin and that helper T-cell clones induced by beef and pork insulin are major histocompatibility complex-restricted T cells that proliferate, produce lymphokines, and provide helper activity after activation. These helper T-cell clones exhibit different antigenic fine specificities: beef insulin-induced clones respond to beef insulin but not pork or autologous insulin, whereas pork insulin-induced clones cross-react with all species of insulin tested, including rat insulin. In addition, the helper activity of cloned pork insulin-specific T cells is abrogated by pork insulin-primed suppressor T cells. These data support the hypotheses that Ir gene control of antibody responses to certain antigens involves mechanisms used for maintenance of self-tolerance.

Receptor diversity of insulin-specific T cell lines from C57BL (H-2b) mice.

To characterize the T cell receptor repertoire in an immune response in which the Ia and nominal antigenic determinants are defined and limited, we have cloned and sequenced the expressed receptors from four independent, beef insulin-specific T cell lines from C57BL mice. Each of these lines responded to beef but not to the pork insulin, thus defining the nominal antigenic determinant recognized. Furthermore, each of these lines could only be presented antigen by B6 but not mutant B6.C-H-2bm12 antigen-presenting cells, thus defining the requisite Ia recognition or antigen-association site. In spite of this functional similarity in ligand specificity, each of these T cell lines was found to use different V alpha and V beta gene segments. Moreover, structural comparisons of implied protein sequences of each of these receptors showed no stretches of conserved amino acid residues that could be implicated in ligand interaction. However, the V alpha genes used by these four clones appeared considerably more homologous to each other than were their V beta genes.

Mechanism of enhanced fibroblast arachidonic acid metabolism by mononuclear cell factor.

Chronic inflammation is associated with an infiltration of mononuclear cells, fibroblast proliferation, and elevated levels of prostaglandin (PG) E2. Mononuclear cell conditioned factor (MNCF) medium (5%) stimulated a 100-fold increase in basal human dermal fibroblast PGE2 release over 48 h as compared with fibroblasts that were incubated with control medium (conditioned medium prepared without cells). The MNCF-induced PGE2 production was suppressed by protein synthesis inhibitors. Fibroblasts pretreated with control medium released PGE2 only modestly in response to 1 nM bradykinin for 1 h (basal, 50 +/- 7 pg PGE2/micrograms protein; stimulated, 104 +/- 12 pg PGE2/micrograms protein), whereas cells that had been pretreated with MNCF showed a greatly facilitated bradykinin-induced release of PGE2. (basal, 297 +/- 59 pg PGE2/micrograms protein; stimulated, 866 +/- 85 pg PGE2/micrograms protein). The exaggerated agonist response is not specific for bradykinin because platelet-derived growth factor elicits a similar response. Exogenous arachidonic acid conversion to PGE2 was also facilitated (two- to threefold) by MNCF pretreatment as compared with control. Both the enhanced agonist-stimulated and exogenous arachidonic acid-induced PGE2 release from the MNCF pretreated cells were inhibited by actinomyin D or cycloheximide. A kinetic study of microsomal cyclooxygenase prepared from fibroblasts pretreated with MNCF showed a threefold increase in the maximum velocity (Vmax) but the same Michaelis constant (Km) as control-treated cells. This augmented arachidonic acid metabolism and subsequent enhanced PGE2 production may play an important role in macrophage-fibroblast interactions at sites of inflammation.