T Cell Autoimmunity in Ig Transgenic Mice¹

Systemic lupus erythematosus (SLE)³ is an autoimmune disease characterized by the presence of high titers of autoantibodies directed at a number of common cytoplasmic and nuclear components (1). An important target of autoimmune responses in human SLE and in murine models of the disease is a group of small nuclear ribonucleoprotein particles (snRNPs), designated U1 and Sm. The snRNPs possess a group of proteins common to all RNAs in their class, designated B, D, E, F, and G, and the particle has the critical biological function of splicing pre-mRNA (2).

MRL lpr/lpr mice spontaneously develop an SLE-like disease, with anti-snRNP and anti-DNA responses similar in frequency and specificity to those found in human SLE (3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13). Studies that have carefully mapped the onset and progression of lupus autoimmunity in mice revealed that autoantibody responses can arise with specificity first to the D protein, followed in time by specificities to other snRNP proteins (3). This observation suggested that the D protein might initiate the autoimmune response, which then spreads or diversifies to other components of the snRNP particle. Moreover, the autoimmune pathology of the MRL mouse appears to require both B and T cell subsets. MRL mice made deficient in either B or T cells fail to develop the intense autoantibody response or severe kidney disease characteristic of this murine model of SLE (13, 14, 15, 16).

The purpose of this study was to elucidate the mechanisms underlying the generation of autoimmune T cells, in particular to determine whether B cells, as auto-APCs, function in the deletion, tolerance, or anergy of autoreactive T cells. Toward these ends, we have generated heavy chain transgenic (Tg) mice using a rearranged unmutated V_H J 558 gene, designated 2-12, derived from an MRL/lpr anti-snRNP hybridoma (17, 18).

Autoreactive B lymphocytes in this mouse are not deleted or anergic and constitute approximately 50% of the splenic B cell repertoire. The Tg heavy chain combines with endogenous light chains, which bind snRNP and/or ssDNA. While the majority of autoreactive B cells are of an immature (heat-stable Ag (HSA^high)) phenotype, autoantibody secretion can be provoked with self Ag or with B cell mitogens (18).

Given the controversial role for B cells as APCs in the priming and/or induction of T cell tolerance (14, 19, 20, 21, 22, 23), we investigated the state of T cell responses in mice bearing B cells with receptors for self Ags. Our studies revealed that CD4 T cells from wild-type (wt) autoimmune-prone and nonautoimmune-prone mice respond to self peptides that require processing by anti-snRNP-specific B cell APCs. However, Tg B cells were capable of tolerizing a subset of autoreactive T cells in Tg C57BL/6 mice. In contrast, no active T cell tolerance induction to the snRNP Ag was observed in Tg MRL lpr/lpr mice.

T cell responses were dependent on the presence of autoantigen-specific B cells as APCs, suggesting that this subset of B cells may be critical in the propagation of autoimmune responses or in shaping autoimmune T cell specificity. Since the normal repertoire is replete with autoimmune B cells, this study emphasizes the importance of B cell subsets in the regulatory mechanisms of lupus autoimmunity.

Anti-snRNP B cell hybridoma, designated 2-12, was originally cloned from an MRL lpr/lpr mouse. Abs from the 2-12 hybridoma are specific for the D protein of murine snRNPs and for denatured DNA (17). Ig Tg mice were derived using the rearranged VDJ segment of the 2-12 anti-snRNP hybridoma cloned upstream in a vector containing the Cm region gene segment (18). Tg founders were backcrossed for >10 generations to C57BL/6 mice and to MRL lpr/lpr mice to the sixth backcross generation. The presence of the transgene was identified by PCR analysis of tail DNA using 2-12 specific primers 5′-GAGGTCCAGCTGCAGCAGTCTGGA-3′ in the first coding region of the V region and 5′-CGCTCCACCAGACCTCTCTAGA-3′ complementary to the XbaI site downstream of JH4.

Tg MRL mice were homozygous for the lpr mutation as screened by PCR using lpr-specific primers (5′-GTAAATAATTGTGCTTCGTCAG, 5′-TAGAAAGGTGCACGGGTGTG, and 5′-CAAATCTAGGCATTAACAGTG).

The 2-12 heavy chain was unmutated in its original construct and, when paired with certain light chains, could bind either snRNPs or DNA; a smaller proportion (<5%) bound both. The autoreactive B cells were not deleted and did not appear to be anergic, as their level of surface IgM was 10-fold higher than that of nonautoreactive B cells (18).

Conventional T lymphocyte proliferation assays were performed with CD4⁺ T lymphocytes (2–3 × 10⁶ cells/ml) and irradiated splenocytes (2500 rad) as APCs at 8–12 × 10⁶ cells/ml. Mice used in these assays were 10–15 wk of age. All assays were set up with triplicate samples and incubated with Ag titrations for 4 days. Lymphocyte proliferation was assessed by [³H]thymidine incorporation (1.0 μCi/well; ICN Chemicals, Irvine, CA) during the last 18 h of culture. Sample wells were harvested onto filters, and incorporated radioactivity was counted in a Betaplate liquid scintillation counter (LKB/Wallac, Gaithersburg, MD).

CD4 T cells were purified using a two-step protocol. Unfractionated spleen cells were incubated with a mixture of Abs to CD8 (TIB208), B cells (TIB146), monocytes/macrophages (TIB128), and anti-class II Abs (Y3JP) at 4°C for 1 h. Excess Abs were washed off followed by incubation with sheep anti-mouse/rat Ig-coated magnetic beads (Perseptive Biosystems, Cambridge, MA) at a ratio of 5:1 (beads:cells) for 1 h at 4°C. The unbound CD4 cells were separated from other cell populations on a magnet, and the unbound cells were then subjected to a second round of purification as described above. The purity of CD4 cells was >96% as assessed by flow cytometry.

The ability of MHC class II to present peptide to autoreactive T cells was confirmed with the use of anti-class II blocking Abs in proliferation assays. Cocultures were set up using irradiated splenocytes from Tg mice (5 × 10⁶/ml) and CD4 T cells (1 × 10⁶/ml) from either wt or Tg MRL lpr/lpr mice. Horse anti-mouse Ig (Vector, Burlingame, CA) and mouse anti-mouse I-A^k and I-E^k Abs (PharMingen, La Jolla, CA) were included in the cultures at concentrations from 5–20 μg/ml. All assays were set up in triplicate, and lymphocyte proliferation was assessed by [³H]thymidine incorporation (1.0 μCi/well) during the last 18 h of culture. Sample wells were harvested onto filters, and incorporated radioactivity was counted in a Betaplate liquid scintillation counter (LKB/Wallac).

Recombinant murine D protein expressed in Escherichia coli was used as stimulating Ag in T cell proliferation assays. The cDNA encoding the open reading frame of the Sm D gene was cloned by PCR from a human placental cDNA library. The Sm D cDNA was subcloned into the vector pBluescript II KS (Stratagene, La Jolla, CA). Sequencing by the Sanger dideoxynucleotide method confirmed that the 372-bp cDNA was identical with that of the published sequence of human Sm D (24). Recombinant mouse D protein was purified by anion and cation exchange chromatography as previously described (24) and was absorbed for potential mitogens by anti-LPS column chromatography using agarose beads coated with polymyxin B (Pierce, Rockford, IL).

Cell populations (∼5 × 10⁵) were stained with Abs diluted in PBS with 0.1% azide and 1% FCS and analyzed on a FACScan flow cytometer (Becton Dickinson, Mountain View, CA). The Abs used were anti-B7.1-FITC (1G10), anti-B7.2-FITC (2D10) (25, 26, 27), and anti-class II^k-FITC (Y3JP; PharMingen, San Diego, CA). Analysis of B cell Ag specificity was performed with biotinylated ssDNA or biotinylated snRNPs followed by incubation with avidin-FITC or avidin-PE as previously described (18). The cells were subsequently washed in the same buffer and were fixed with 1% paraformaldehyde in PBS before analysis.

We investigated the response of CD4 T cells from unimmunized C57BL/6 wt and Tg mice to the snRNP D protein in the presence of either wt or Tg APCs. As described above, the Ig transgene was selected for its binding specificity to the D protein component of the murine snRNP complex. Approximately one-quarter of Tg B cells bind snRNPs as assessed by flow cytometry (Table I). Overall, as many as 50% of Tg B cells are autoreactive in binding either snRNPs or ssDNA. As recently described (18), 2-12 Tg C57BL/6 mice do not possess serum autoantibodies (anti-snRNP or anti-DNA) unless immunized with snRNPs in CFA. Moreover, Tg B cells can be stimulated to secrete autoantibodies by LPS stimulation in vitro (data not shown). Tg C57BL/6 B-autoreactive B cells are primarily of an immature phenotype as indicated by HSA expression (Table II) (18).

As illustrated in Fig. 1, CD4 T cells from wt C57BL/6 mice did not respond to snRNP D protein presented by wt autologous APCs. This observation suggests that the autoimmune T cell subset was deleted or anergic to peptides of the D protein presented in vivo by nonspecific APCs.

In contrast, wt C57BL/6 CD4 T cells responded strongly to snRNP D protein presented by Tg APCs (representative proliferative data from seven experiments), suggesting that unique peptides presented by Tg APCs could activate autoimmune T cells from the normal repertoire. In fact, wt CD4 T cells had the strongest proliferative response to the D protein of snRNPs of all four culture groups (Fig. 1 A). This simple observation suggests that the Tg B cells, by virtue of their surface receptor specificity, presented a unique group of snRNP D peptides to which T cells in the wt mouse had never contacted and thus had never been deleted or tolerized.

CD4 T cells originating from Tg C57BL/6 mice also failed to significantly respond to peptide presented by wt APCs, suggesting again that this autoreactive T cell subset had been deleted or anergized in Tg mice via nonspecific APCs (perhaps macrophages or dendritic cells). C57BL/6 Tg T cells also failed to significantly respond to peptide presented by syngeneic Tg APCs. The inability of these Tg CD4 T cells to respond under these conditions suggests that Tg B cells in vivo acted to tolerize T cells to the specific peptides that were displayed on its surface.

These observations were in contrast to similar experiments performed with the autoimmune-prone MRL lpr/lpr mouse, which also carried the 2-12 Ig heavy chain transgene (Fig. 1 B). As in C57BL/6 mice, CD4 T cells from wt MRL lpr/lpr mice failed to respond to snRNP D protein presented by wt APCs, presumably by virtue of intact mechanisms of tolerance induction as described above. In addition, a Tg source of CD4 T cells appeared to be incapable of responding to D protein presented by wt APCs. However, T cells from Tg MRL lpr/lpr mice responded to peptides of the D protein when presented by Tg APCs. In comparisons between the two mouse strains, MRL lpr/lpr T cells, whether from a Tg or a wt source, required Tg APCs to drive proliferation. While several mechanisms may be responsible for these differences, it is possible that a failure of T cell apoptosis to the snRNP Ag in the Tg lpr/lpr mouse may allow the autoreactive repertoire to possess this subset of autoreactive T lymphocytes. T cell activation or anergy in either strain is probably not due to other mechanisms, such as the processing of immune complexes, since Tg mice do not spontaneously secrete autoantibodies (18).

The nature of the Ag presentation by 2-12 Tg MRL lpr/lpr APCs was explored using anti-mouse Ig and anti-MHC class II Abs. The self Ag presented by the Tg APCs to CD4 T cells from both wt and Tg mice was MHC class II restricted, as it could be blocked by anti-MHC class II Abs. While the self Ag was presented by MHC class II on APCs, we sought to examine the role of the Ig receptor on Tg B cells in binding self Ag (snRNP) and internalizing it to be eventually processed and presented by MHC class II molecules. Our experiments revealed that anti-Ig Abs could be used to effectively block self Ag presented by 2-12 Tg APCs at a concentration of 20 μg/ml (Fig. 4). The presence of both anti-Ig and anti-MHC class II Abs together also served to effectively inhibit the proliferation of CD4 T cells that respond to self Ag (Fig. 2).

In addition, the exclusive ability of Tg B cells to stimulate MRL or C57BL/6 T cells (Fig. 1) is not due to other factors, such as enhanced expression of MHC class II on Tg B cells (Table III). Moreover, Tg and wt APCs stimulate control keyhole limpet hemocyanin-specific T cells in a similar manner, suggesting that Tg APCs are not inherently better at presenting Ag (data not shown).

Our studies demonstrated a clear lack of T cell tolerance in lymphocytes from MRL lpr/lpr Tg mice and/or when Tg B cells were the source of APCs. These observations indicated that autoimmune lymphocytes remaining in Tg mice may be partially or fully primed in vivo. The costimulatory molecules B7.1 and B7.2 are known to play a pivotal role in determining the recognition of Ag by T cells and could lead to activation or anergy depending on their presence or absence (25, 26, 27, 29, 30, 31, 32, 33, 34). Therefore, we analyzed the status of these costimulatory molecules on B cells from both wt and Tg mice. Flow cytometric analysis did not reveal any significant differences in the cell surface levels of B7.1 and B7.2 on B cells in the two groups of mice (Table III). However, B cells from the two groups of mice could up-regulate both B7 molecules, indicating that costimulatory molecule expression is not defective in Tg mice (data not shown). These observations were not entirely surprising, since our prior studies indicated that B cell activation, as expressed by serum anti-snRNP responses, required snRNP immunization of Tg mice (18). However, these studies do not preclude the possibility that a small and undetectable subset of stimulated Ag-specific B lymphocytes from Tg mice may be important in establishing the status of T cell tolerance in these mice.

The immune system has developed several mechanisms by which its constituent cells can avoid reactivity toward self proteins. However, the presence of many types of autoimmune syndromes suggest that these mechanisms are not perfect in purging the immune system of autoreactive cells. This phenomenon is dependent upon a number of factors, including the concentration, structure, and nature of the autoantigen. It is thought that self-reactive lymphocytes could be rendered tolerant by undergoing deletion within the bone marrow and thymus (35, 36, 37, 38, 39, 40) or in the periphery (41, 42). In addition, self-reactive B cells can evade elimination by undergoing secondary Ig rearrangements to replace self-reactive light chains for those that respond to foreign Ag in a process referred to as receptor editing (43, 44, 45).

An elegant approach to study the fate of self-reactive lymphocytes has been to generate high frequencies of anti-self lymphocytes through the use of TCR or Ig receptor Tg mice. We have constructed such a Tg mouse carrying the Ig heavy chain specific for the murine snRNP D protein, a target of lupus autoimmunity. This study demonstrated that the presence and activation of autoreactive T cells are dependent on unique autoantigen-processing properties of autoreactive B lymphocytes. Tg B lymphocytes, depending on the environment in which they function, could either tolerize or activate a highly defined set of self-reactive T cells.

The anti-snRNP Ig Tg mouse appeared to be in normal health and did not have circulating anti-snRNP or anti-DNA Abs despite the fact that autoreactive B cells account for approximately 50% of the repertoire (Tables I and II) (18). These cells did not appear to be anergic, as they possessed a 10-fold higher level of surface IgM compared with nonautoreactive B cells. Although of primarily immature phenotype (HSA^high), Tg B lymphocytes have the ability to bind self snRNP Ag and, upon immunization of Tg mice with murine snRNPs, secrete high titers of serum autoantibodies (18). Stimulation of Tg B cells with LPS also drives anti-snRNP Ab secretion. In this regard our findings are similar to those of Goodnow et al. in the analysis of tolerance/immunity in soluble hen egg lysozyme (HEL)/anti-HEL Ig double Tg mice (46, 47, 48). It was demonstrated that soluble HEL did not trigger elimination of HEL-binding B cells, but rendered them functionally inactive. However, in Tg C57BL/6 lpr/lpr mice, B cell tolerance to soluble HEL was incomplete, in that the anti-HEL B cells could undergo activation by Ag (46). In our anti-snRNP Tg mice, we do not know the context in which snRNPs interact with Tg B cells. However, anti-snRNP autoantibodies are elicited upon immunization with mouse snRNPs, demonstrating that Tg B cells are also not functionally tolerant (18).

It is well established that B cells bind Ag via specific cell surface receptors to efficiently function as APCs. Rock et al. and Lanzavecchia have demonstrated that B cells present Ag up to 10,000-fold more efficiently by receptor-mediated uptake than by nonspecific uptake of Ag (49, 50, 51). Moreover, it has been proposed that B LNCs, as APCs, can shape the specificity of responding T cells. This process is based on the ability of surface Ig to either suppress or enhance the presentation of specific peptide determinants on its surface with class II MHC (52).

This study was undertaken to determine whether self Ag, in the presence of B cells with the ability to bind endogenous snRNP autoantigen, could stimulate, delete, or tolerize an autoimmune T cell response. It has been previously suggested that B cells, either activated or resting, fail to prime naive CD4 T cells but instead tolerize naive T cells (19, 20, 21). It has also been shown that naive T cells are dependent on costimulation and are activated primarily by dendritic cells and preactivated B cells, but not by macrophages or resting B cells. However, several studies from this laboratory and others support a role for B cell APCs in priming and activating an autoimmune T cell response (53, 54, 55, 56, 57). Moreover, this mechanism may be important for so-called epitope spreading to both foreign and self Ag (54).

We examined autoreactive T cell responses by coculturing wt or Tg B cell APCs together with either a source of T cells from wt or Tg mice. As might be expected, wt T cells fail to respond to snRNP Ag presented by wt APCs, suggesting that mechanisms of immune tolerance are intact in wt mice. However, wt T cells respond vigorously to snRNP Ag presented by Tg B bearing surface Ig specific for this Ag. The implications of this response are that autoantigen-specific B cells present unique peptides to which wt T cells have never been exposed and thus never tolerized or deleted. In contrast, C57BL/6 Tg T cells fail to respond to Tg (or wt) B cell APCs presenting autoantigen. These observations suggest that Tg B cells do indeed tolerize autoimmune T cells in the C57BL/6 Tg mouse.

These observations were significantly different from parallel studies performed in autoimmune-prone MRL lpr/lpr Tg mice. We found that Tg B cells could present stimulatory peptides to Tg T cells in vitro, although the source of both populations was a Tg mouse. This outcome demonstrates a defect in T cell tolerance, at least as mediated by autoimmune B cells, to this autoantigen in the Tg mouse. While there may be several explanations for this observation, it is possible that the lpr defect in these mice do not allow autoreactive T cells to be anergized from the peripheral repertoire. We are presently initiating studies in Tg C57BL/6 lpr/lpr mice to examine the potential importance of other MRL background genes in this autoimmune response.

We have also not ruled out the possibility that Tg heavy chains may preferentially combine with endogenous light chains contributing to different binding affinities for the Ag. B cells with a high affinity for the Ag do undergo deletion in a transition from immature to mature phenotype. With this as the case, not only would binding of Ag be influenced based on affinity of surface Ig, but also the interaction of Ag-specific CD4 T cells to Ag bound to the MHC class II molecule would also be affected. The binding of the Ig receptor to the autoantigen with different affinities would also presumably lead to differential protection of the peptide bound in the Ig receptor during Ag processing in the endosomal/lysosomal compartment. This not only may influence the binding of Ag to B cell but may also skew CD4 T lymphocytes toward deletion, anergy, or activation. As indicated previously in studies by Watts and Lanzavecchia, B cells have the capacity to specifically suppress the expression of particular determinants based on surface Ig specificity and/or affinity for Ag (52).

The self Ag presented by 2-12 Tg APCs was MHC class II restricted, as it could be blocked by Abs to MHC class II receptors. More significantly, anti-mouse Ig Abs were able to block presentation of self Ag by Tg APCs to CD4 T cells. The finding that blocking was more effective at higher concentrations may be explained by the fact that we used Abs to the whole Ig receptor as opposed to anti-Fab Abs, or more simply, we needed more Ab to block the Ig receptor. This observation also emphasizes the importance of the Ag-specific Ig receptor in the capture and concentration of Ag and its transport to MHC class II Ag-bearing compartments. A combination of anti-MHC class II and anti-Ig Abs, however, did not have any additive or synergistic inhibitory effect on the presentation of self Ag by Tg APCs.

Our prior studies indicate that B7 costimulation, principally B7.2, is an essential component for B cells to prime autoreactive T cells (53). We therefore examined the peripheral B cell populations in Tg mice for the presence of amplified levels of B7. We demonstrated that Tg B cells possess a B7 phenotype similar to that of wt B cells, indicating that there is no overwhelming B cell activation in Tg mice. Tg B7^neg cells are most likely responsible for tolerizing T cells to presentation of native snRNP D protein. We cannot rule out, however, that a small subset of activated Tg B cells may be important in T cell activation. With respect to T cell activation or memory phenotypes, no significant differences were found between wt and Tg peripheral T cells. These observations may not be unexpected, since other Tg T cell systems responsive to self Ag show no overt signs of activation in vivo (58).

Our study demonstrates that the status of B cell tolerance in individuals may be important in mediating autoimmune T cell responses. Once stimulated, autoimmune T cells may perpetuate and amplify the autoantibody synthesis characteristic of human SLE. It is clear that autoreactive B and T lymphocytes are a component of the normal repertoire of mice (59, 60, 61). It is possible that some perturbation of B cell tolerance, perhaps by virtue of molecular mimicry or the presentation of cryptic self determinants, may provoke autoantigen processing by B cells in the initiation of autoimmune responses.