We have investigated the development of CD4+ T cells in mice expressing low levels of transgenic class II MHC molecules (Ab) preoccupied with covalent peptide (Ep), which in the presence of invariant chain (Ii) is extensively cleaved and replaced with self-derived peptides. In these mice, the transgenic Ab molecules, bound with predominant peptide (Ep) and with multiple self-peptides, selected more CD4+ T cells than Ab/self-peptide complexes expressed in wild-type mice. The enhanced outcome of thymic selection was a result of impaired negative selection, rather than more efficient positive selection by an overall lowered abundance of self-derived Ab/peptide complexes. Peripheral CD4+ T cells in the AbEpIi+ mice had memory phenotype, often followed by polyclonal activation of B cells. The AbEpIi+ mice preserved their good health and had a normal life span despite the profound number of activated CD4+ T cells and B cells in peripheral lymphoid organs, moderate hypergammaglobulinemia, and deposited complexes in the kidneys. We propose that CD4+ T cells positively selected due to low avidity for high abundant AbEp complex avoid negative selection on Ab molecules loaded with low abundant peptides and become self-reactive in the peripheral lymphoid organs.
Positive selection rescues T cells with TCRs that bind self-MHC/peptides complexes expressed on thymic epithelial cells with low avidity (1, 2, 3). Thymocytes with TCRs that do not bind self-MHC/peptide complexes die neglected, and thymocytes that recognize these complexes with high avidity are deleted by negative selection (4, 5, 6). Naive T cells that leave the thymus migrate to the periphery where they must perceive self-MHC/peptide complexes to survive and expand (7, 8). Analysis of T cell repertoires in genetically manipulated mice expressing class II MHC molecules exclusively on thymic stromal cells or in association with one peptide clearly showed that positive selection indiscriminately generates a large number of potentially self-destructive T cells. On the average, one-half to three-quarters of positively selected T cells are believed to be negatively selected, implying that most T cells expressing TCRs that can bind MHC die in the thymus (9, 10). Because the conditions for positive selection and survival bias naive T cells to continuously perceive self-MHC/peptide complexes, there is an increased risk that some of these cells may elicit undesirable reactions against host tissue. The autoreactive T cells often express TCRs with low avidity for self-MHC/peptide complexes, indicating that regardless of the high sensitivity of negative selection, self-destructive T cells leak to the periphery (11, 12). Hence, it is essential to determine how an abundance of the specific MHC/self-peptide complexes and their individual properties may diminish the effectiveness of negative selection.
It is imaginable that physiological expression of MHC evolved to display an optimum of different peptides to T cells. An average APC expresses about 104–105 class II MHC molecules bound with 103–104 different peptides derived from self and non-self proteins. The distribution of peptides across class II MHC molecules varies; some peptides capture 10% of total class II MHC molecules and are displayed to CD4+ T cells at high abundance, while other peptides bind only a few MHC molecules per cell and are presented at low abundance (13, 14). The immune system can augment presentation of low abundant MHC/peptide complexes to T cells. For example, infrequently expressed MHC/peptide complexes may selectively relocate, upon TCR/coreceptor engagement, into the cell-cell contact site called the immunological synapse to enhance their display to T cells. It has been also proposed that inside the immune synapse, few specific MHC/peptide complexes serially engage multiple αβTCRs and trigger the T cell activation signaling pathway (15). It is likely that both low and high abundant peptides bound to MHC influence T cell development, as shown by different studies conducted with T cells selected in mutant mice expressing class II MHC molecules occupied with dominant peptide. Whereas one study emphasized the importance of low abundant peptides bound to class II MHC on thymic selection of CD4+ T cells in these mice, the other studies found evidence for the imprint of dominantly expressed peptide that led to the selection of CD4+ T cells with altered self-reactivity (14, 16).
Here, we have examined the development of CD4+ T cells in mice expressing transgenic Abβ-chain covalently associated with Eα52–68 peptide in the absence of endogenously expressed Abβ-chain (9). In these mice expression of class Ab molecules is reduced 10-fold on all bone marrow-derived APCs compared with wild-type mice, and the covalently attached peptide is replaced at large with endogenously derived peptides due to the presence of invariant chain (Ii).4 In these mice the Ab molecules occupied with dominant Ep peptide and low abundant, multiple self-peptides selected many CD4+ T cells. The majority of these CD4+ T cells expressed activation markers, but did not cause damage to host tissues. We propose that these CD4+ T cells were positively selected on the dominant AbEp complex and acquired a memory phenotype due to diminished negative selection and cross-reactivity for low abundant peptides.
Materials and Methods
The C57BL/6 mice (Abwt) and mice deficient in endogenous TCR α-chain (TCRα−) were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice deficient for the wild-type Abβ-chain (Ab−) were provided by D. Mathis (Harvard Medical School, Boston, MA). Mice transgenic for the AbβEp construct and deficient for the wild-type Abβ-chain (AbEpIi+) were generated at the National Jewish Medical and Research Center (Denver, CO) as previously described (9) and backcrossed on the C57/BL6 background. Mice transgenic for αβTCR specific for Ab and pigeon cytochrome c (PCC)43–58 peptide were generated in our laboratory (P. Kraj, unpublished observations) and crossed with mice transgenic for AbβEp and devoid of endogenous Abβ-chain and TCR α-chain (AbEpIi+TgTCR+TCRα−/−). All mice were further bred in the animal care facility at the Medical College of Georgia (Augusta, GA).
Chimeric mice were generated by irradiation of 6-wk-old animals (1100 rad) followed by i.v. reconstitution with 5 × 106 T cell-depleted bone marrow. Chimeras were analyzed at least 8–10 wk later.
Cells were stained for CD4 (GK1.5), CD8 (53-6.7), CD69 (H1.2F3), CD44 (IM7), CD62L (MEL14), CD45RB (16A), and different Vβ segments as previously described (17). The fluorescein- and PE-labeled Abs were purchased from Becton Dickinson (San Diego, CA) or were made in our laboratory. Briefly, cells were suspended in staining buffer (balanced salt solution (BSS), 0.1% sodium azide, and 2% FBS) and were incubated for 30 min at 4°C with the Abs of interest in the presence of 10% normal mouse serum and 10% anti-Fc receptor mAb (2.4.G2). Cells were than washed three times with staining buffer and analyzed using a FACScalibur instrument (Becton Dickinson).
For detection of IgG deposits in kidney glomeruli, kidneys were embedded in OCT compound (Sakura Finetek, Torrance, CA) and snap-frozen. Five- to 7-μm sections were air-dried and fixed with cold acetone for 10 min. These cryosections were and stained with FITC-labeled goat anti-mouse IgG for 1 h at room temperature. Sections were analyzed, and photographs were taken using a Axiophot fluorescent microscope (Karl Zeiss, Thornwood, NY) and video camera (Photometrics, Tuscon, AZ).
Total levels of serum IgM and IgG subclasses were determined by ELISA using alkaline phosphatase-labeled goat Abs specific for mouse Ig classes and subclasses (Southern Biotechnology Associates, Birmingham, AL). The Ig concentrations were determined by referring to standard curves obtained with known concentrations of mouse Ig (Southern Biotechnology Associates).
The responses of CD4+ T cells selected in AbEpIi+ mice for Ab/self-peptide complexes were tested by incubating purified CD4+ T cells with irradiated splenocytes (3000 rad). Lymph node CD4+ T cells from AbEpIi+ were purified by complement depletion as described using cytotoxic cocktail prepared from supernatants or with purified Abs from the following hybridomas cultured in this laboratory: anti-CD8 (clone HO 2.2), anti-MHC class II (clones 25-6-3S and BP107.2.2), anti-CD45 (clone B220), and anti-J11D (clone J11D.2) (17). The proliferative response was measured on the third day using an 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetetrazolium bromide assay (18).
Peptide-specific T cell hybridomas
Assay for graft vs host (GvH) reaction
The 6- to 8-wk-old B6 mice were lethally irradiated with a single dose of 1100 rad and were reconstituted with 5 × 106 T cell-depleted bone marrow from Abm12 or AbEpIi+ mice together with 1 × 107 purified CD4+ T cells from AbEpIi+ mice. Complement depletion with anti-CD4 and anti-CD8 mAbs was used to deplete the donor bone marrow of mature T cells as previously described (17).
Mice expressing fewer Ab molecules bound with various self-derived peptides select more CD4+ T cells than wild-type mice
We have recently described transgenic mice that express a β-chain of mouse class II MHC molecule (Ab) covalently linked with the single Eα52–68 peptide in the absence of endogenous Abβ and invariant chain (9). Due to the exclusive occupancy of all detectable Ab molecules with one covalent peptide, these mice have a severely compromised number of CD4+ T cells in the thymus and periphery. However, the number of CD4+ T cells is high in mice that express the AbβEp transgene in the absence of endogenous Abβ-chain but with normal expression of the invariant chain (AbEpIi+ mice). The presence of the invariant chain leads to cleavage of the covalent peptide, which is replaced with a diverse set of self-peptides (21). In AbEpIi+ mice, around 30% of the Ab molecules are occupied with Ep, while the rest are bound with low abundant, self-derived peptides. Interestingly, the number of selected CD4+ T cells in the thymus or periphery in the AbEpIi+ mice was roughly two times higher than the number of CD4+ T cells found in wild-type mice (Fig. 1,A). The repertoire of TCRs expressed on CD4+ T cells found in the AbEpIi+ mice was polyclonal, and the frequencies of Vβ segments used in these TCRs were similar to the Vβ pattern recorded for wild-type mice (Fig. 1,B). To determine why the reduced number of Ab/peptide complexes facilitates selection of a higher number of CD4+ T cells in vivo, we examined the expression level of Ab molecules on different types of APCs in AbEpIi+ mice. Thymic epithelial cells fromAbEpIi+ mice vs wild-type mice had two times less Ab expressed, while splenic APCs had 10 times fewer Ab molecules expressed than the respective subpopulation of APCs isolated from wild-type mice. The Ab/peptide complexes present in APCs from AbEpIi+ mice stained positively with mAbs specific for Ab, including ones that depend on the expression of particular endogenously derived peptides (Fig. 2,A). In addition, three T cell hybridomas specific for Ab and endogenously derived peptides secreted IL-2 after overnight incubation with spleen cells derived from AbEpIi+ mice (Fig. 2 B). These results implied that in AbEpIi+ mice the Ab molecules are occupied with dominant Ep and low abundant self-peptides.
The greater number of CD4+ T cells found in AbEpIi+ mice could be a result of enhanced positive selection due to reduced expression of Ab/peptide complexes on thymic epithelium. Positive selection is mediated by low avidity interactions between TCR and self-MHC/peptide complexes, and there are experiments implying that positive selection uses low abundant class II MHC/peptide complexes more favorably than high abundant class II MHC/peptide complexes. Alternatively endogenous peptides in AbEpIi+ mice have to compete with cleaved Ep for binding to Ab, which may bias the repertoire of self-peptides toward ones that firmly bind to Ab and potentially better select CD4+ T cells. Finally, in AbEpIi+ mice, the Ab molecules may be poorly expressed on thymic macrophages and dendritic cells, but are only moderately reduced on thymic epithelial cells. This divergence in expression of Ab/peptide complexes on different thymic stromal cells might also cause alterations in the T cell selection processes. Hence, to compare the efficiency of positive selection of CD4+ T cells on thymic epithelial cells in AbEpIi+ or wild-type mice, we lethally irradiated both types of mice and reconstituted them with bone marrow cells from class II MHC-deficient mice (Ab−). In these chimeras positive selection proceeded on thymic epithelium of AbEpIi+ or wild-type hosts in the absence of negative selection on hemopoietic APCs. As shown in Fig. 3, the numbers of CD4+ T cells positively selected by wild-type and AbEpIi+ epithelium were comparable. This experiment indicated that Ab/peptide complexes expressed in AbEpIi+ mice do not positively select more CD4+ T cells than they do in wild-type mice.
Impaired negative selection in AbEpIi+ mice results in an increased number of CD4+ T cells in the thymus
In AbEpIi+ mice, reduction in Ab expression was greatest on all hemopoietic APCs. Therefore, we next examined whether the negative selection in these mice proceeds with the same efficiency as in wild-type mice. For that purpose, we lethally irradiated AbEpIi+ mice and reconstituted them with bone marrow from either AbEpIi+ or Abwt mice or a mixture of both bone marrows. As shown in Fig. 4,A, only mice reconstituted with bone marrow from the autologous donor had an elevated number of CD4+ T cells in the thymus and the peripheral lymphoid organs. Chimeras with thymic epithelium expressing low levels of transgenic Ab/peptide complexes, but reconstituted with wild-type bone marrow, had a normal number of CD4+ T cells. Also, the AbEpIi+ recipients that received mixed AbEpIi+/Abwt bone marrow had a normal number of CD4+ T cells in the thymus, implying that selected CD4+ T cells do not expand in the medulla upon exposure to altered transgenic Ab/peptide complexes. Similarly, the CD4/CD8 T cell ratio in the thymus or peripheral lymph nodes was 2 times higher only in recipient mice reconstituted with autologous bone marrow and not in the two other recipients (Fig. 4 B). The total number of CD4+ T cells found in the periphery of AbEpIi+→AbEpIi+ chimeric mice was nearly three times higher than the number of CD4+ T cells found in the two other chimeras (data not shown). Collectively, these estimates imply that in AbEpIi+ mice some CD4+ T cells eligible for negative selection avoid deletion and proceed to the periphery.
Accumulation of activated CD4+ T cells in the peripheral lymphoid organs in mice coexpressing dominant AbEp complex and Ab bound with low abundant peptides
Experiments described in the previous section implied that in AbEpIi+ mice, negative selection is partially impaired, allowing more CD4+ T cells to depart to the periphery. To test whether peripheral CD4+ T cells in AbEpIi+ mice have a naive or activated phenotype, we analyzed the expression of activation markers on this population of cells. As shown in Fig. 5,A, far more CD4+ T cells in AbEpIi+ than in wild-type animals expressed CD69 and CD44 molecules, while fewer expressed high levels of CD62L and CD45RB. The populations of activated vs naive CD4+ T cells in AbEpIi+ or wild-type mice were followed over a period of 50 wk. As shown in Fig. 5 B, the number of activated CD4+ T cells increased much faster in the AbEpIi+ mice.
Naive CD4+ T cells can acquire an activated phenotype as a result of homeostatic expansion in the absence of Ag. To determine whether expansion and activated phenotype of CD4+ T cells in AbEpIi+ mice was regardless of TCR specificity, we crossed these mice with mice that exclusively express transgenic TCRs specific for Ab and foreign PCC43–58 peptide in the absence of endogenous TCR α-chains. Even though T cells bearing this transgenic αβTCR were selected toward the CD4+ lineage in both Ab wild-type and AbEpIi+ mice, these CD4+ T cells did not acquire an activated phenotype (Fig. 6). Therefore, in AbEpIi+ mice the CD4+ T cells have to express αβTCRs with multiple specificities to get activated.
Peripheral CD4+ T cells from AbEpIi+ mice mediate bone marrow rejection in lethally irradiated recipients
Despite the significant number of CD4+ T cells with an activated phenotype that are found in AbEpIi+ mice, these mice retained good health throughout their lifetimes. The only visible morphological abnormalities noted among these mice were an increased rate of splenomegaly in their third to fourth decade of life and Ab deposits in the kidneys (see below). Moreover, when isolated CD4+ T cells were cultured with APCs expressing autologous or normal level of Ab/peptide complexes, no significant proliferation was detected (data not shown).
To determine whether the CD4+ T cells found in AbEpIi+ mice can mediate GvH reaction, we lethally irradiated the AbEpIi+ mice and reconstituted them with autologous, T cell-depleted bone marrow together with 5 × 106 mature CD4+ T cells isolated from Abwt, Abm12, or AbEpIi+ mice. Chimeras that received bone marrow together with CD4+ T cells from Abwt mice survived well over 4 mo, while the ones that received CD4+ T cells from Abm12 succumbed to GvH disease after 2–3 wk (Fig. 7). In contrast, mice that received bone marrow and CD4+ T cells from AbEpIi+ mice remained healthy for 4 wk, but than all five recipients died during the following week. These results suggested that CD4+ T cells transferred from Abm12 or AbEpIi+ mice mediate GvH, which is delayed, probably due to the lower expression level of Ab/peptide complexes on the AbEpIi+ bone marrow APCs.
In AbEpIi+ mice activated B cells cause splenomegaly and secrete Igs that are deposited in the kidneys
CD4+ T cells that terminally differentiate toward the Th2 lineage secrete IL-4 and IL-5, which activates naive B cells. During the staining of CD4+ T cells for CD25 expression, we noticed that in AbEpIi+ mice, but not in wild-type mice, there is a separate population of cells bearing this marker. These cells also stained for B220 and CD19, but not for Thy1. Because CD25 is an activation marker also expressed on B cells, this result implied that AbEpIi+ mice also have an increased number of activated B cells in the peripheral lymphoid organs. We next determined that these activated B cells secrete more Ig than normally found in the serum of wild-type mice. As shown in Fig. 8,A, the AbEpIi+ mice had elevated serum levels of IgG1 and IgG2b, suggesting that polyclonal activation of CD4+ T cells also results in minor hypergammaglobulinemia. However, these Abs were not directed against ssDNA or chromatin, and we found no pathological deposits of autoantibodies in the joints (data not shown). Immunocomplexes were found in the kidneys of 40% of these mice, but the mice did not develop proteinuria. These results imply that a leak of autoreactive CD4+ T cells, which escaped deletion in the thymus, activates B cells that secrete an increased amount of Ig, leading to early stages of glomerulonephritis (Fig. 8 B).
We have analyzed thymic selection of CD4+ T cells in mice expressing low levels of transgenic Ab molecules concurrently occupied with dominant Ep peptide and many low abundant peptides. In these mice the overall expression level of Ab molecules was only slightly decreased on thymic epithelial cells, but was reduced 10-fold on all thymic and peripheral hemopoietic APCs. Uniquely, CD4+ T cells were selected in the thymus of these mice more efficiently than in wild-type thymus. Moreover, these CD4+ T cells were prone to spontaneous activation in the peripheral lymphoid organs. To determine whether thymic selection processes may be responsible for this phenomenon, we made different radiation chimeras expressing the transgenic Ab molecules on various thymic stromal cells. The quantitative estimates derived from these experiments showed that the decreased expression of Ab molecules cooccupied with dominant Ep peptide and multiple low abundant self-peptides resulted in impaired negative selection of CD4+ T cells. The CD4+ T cells that avoid deletion in the thymus expressed TCRs with low avidity for Ab/self peptides, and we could not elicit a response in vitro to wild-type APCs expressing Ab/peptide complexes at a higher level (see below). Additionally, none of the 80 randomly generated CD4+ T cell hybridomas derived from AbEpIi+ mice produced IL-2 when incubated overnight with wild-type APCs (data not shown). Some of the CD4+ T cells isolated from AbEpIi+ mice expressed CD25 and intermediate levels of CD45, which may be indicative of the presence of regulatory CD4+ T cells, an issue that we are currently investigating (22). Similarly, anergic CD4+ T cells may express these phenotypic markers, but the responses of CD4+ T cells from AbEpIi+ mice and wild-type mice to TCR stimulation were comparable, implying that the CD4+ T cells in AbEpIi+ mice are functional (data not shown). CD4+ T cells in AbEpIi+ mice expressed activation markers such as CD69high, CD44high, CD62Llow, and CD45RBlow only if their αβTCRs were heterogeneous, suggesting that a signal provided by αβTCR recognition of Ab/peptides is required to activate these cells. B cells are also activated in AbEpIi+ mice and secrete more IgG Igs. Because B cell activation and Ab isotype switch are driven by the Th2 type of CD4+ T cells, this observation suggests that chronic exposure to Ab occupied with unknown self-peptides in AbEpIi+ mice might bias the terminal differentiation of CD4+ T cells.
The properties of CD4+ T cells found in AbEpIi+ mice resemble some of the properties of CD4+ T cells found in mice expressing Ab/self-peptide complexes only on thymic epithelium (K14 mice) (2). These latter CD4+ T cells, upon adoptive transfer into nonirradiated recipients with wild-type Ab/self-peptide complexes on hemopoietic APCs, became activated and provoked B cell activation, but did not cause an adverse GvH reaction (23). However, coinjection of autologous bone marrow and CD4+ T cells from K14 mice into lethally irradiated wild-type recipients provoked rapid failure of bone marrow engraftment. Additional experiments in which the original K14 mice were crossed with mice that express few of the transgenic Ab/peptide complexes on dendritic and myeloid APCs showed that a trace amount of Ab was sufficient to eliminate most of the autoreactive CD4+ T cells (24). This last result implies that epithelial cells could not mediate efficient negative selection and that most autoreactive CD4+ T cells will die if selection proceeds on a similar spectrum of low abundant Ab/peptide complexes coexpressed on epithelial and hemopoietic thymic stromal cells.
The low avidity, self-reactive CD4+ T cells were also found in mice where Ab molecules are dominantly occupied with the class II Ii-derived peptide (CLIP) (16). Notably, these CD4+ T cells had some properties indicating that they were positively selected by the dominant Ab/CLIP complex. These CD4+ T cells were unable to reject a wild-type skin graft or bone marrow when the latter was used to reconstitute lethally irradiated autologous recipients. In contrast, the CD4+ T cells mediated a high primary MLR to wild-type Ab APCs, induced splenomegaly when injected into wild-type neonates, and led to the destruction of host bone marrow in lightly irradiated adults. The CD4+ T cells selected on Ab occupied with CLIP peptide had split tolerance because these cells had recognized Ab/CLIP complex with low avidity but cross-reacted with Ab occupied with endogenous peptides.
Published analysis of the repertoire of TCRs selected by AbEp complex showed that these TCRs frequently recognize Ab molecules loaded with self-peptides (9). Because the AbEp complex remained dominantly expressed in AbEpIi+ mice, this complex might positively select a minor population of CD4+ T cells in these mice. Consequently, the question arises of which of the CD4+ T cells, those selected on Ab bound with Ep or those selected on low abundant self-peptides, are spontaneously activated in the peripheral lymphoid organs in these mice. We favor the hypothesis that the AbEp complex selects self-reactive CD4+ T cells that avoid negative selection on the rest of the transgenic Ab molecules bound with different endogenous peptides. These CD4+ T cells are activated in the peripheral lymphoid organs as they continue to recognize Ab bound by low abundant, endogenous peptides. With age, the accumulation of CD4+ T cells with memory phenotype is followed by polyclonal activation of B cells. Although the health or life span of the AbEpIi+ mice is not compromised, an onset of autoimmune disease requires prolonged avidity maturation of the TCR repertoire, and a few of the Ab/peptide complexes may be insufficient to drive such selection in the periphery (25).
We thank Dr. R Markowitz for critical reading of the manuscript.
This work was supported by National Institutes of Health Grants AI41145 and HD36302.
Abbreviations used in this paper: Ii, invariant chain; BSS, balanced salt solution; GvH, graft vs host; PCC, pigeon cytochrome c;.