Peripheral antigen-expressing (cells)
Medullar thymic epithelial cells
Transgenic adenocarcinoma of mouse prostate
Large T antigen
Activation-induced cell death
Many tissue-specific antigens are expressed in specialized cells called peripheral antigen-expressing cells (PAE) in the thymus and can induce central tolerance. While thymic medullary epithelial cells are the prototypic PAE that express peripheral antigens via an aire-dependent mechanism, some studies also describe bone marrow (BM)-derived dendritic cells (DC) and macrophages as PAE in both the thymus and secondary lymphoid organs. However, the role of these cells in development of tolerance to tissue-specific antigens has not been elucidated. Here we use BM radiation chimeric mice to study the existence of hematopoietic PAE and their contribution to tolerance to tissue-specific antigens. Our results reveal that BM-derived PAE exist in both central and secondary lymphoid organs and that the expression of peripheral antigens in the BM-derived cells does not correlate with aire expression. Using double-transgenic mice expressing TCR specific for a model antigen expressed under the control of a prostate-specific promoter, we show that expression of self antigen in PAE of non-hematopoietic origin is both necessary and sufficient to induce clonal deletion. Surprisingly, while BM-derived PAE fail to induce clonal deletion, they do cause the activation-induced cell death of autoreactive cells in the secondary lymphoid organs. Thus, BM-derived PAE have a distinct function in the maintenance of tolerance to tissue-specific antigens.
The expression of peripheral antigen in the thymus by PAE serves to project a “shadow of immunological self” in the thymus 1. By constitutively expressing antigens assumed to be limited in the peripheral organs, PAE ensure that tolerance to tissue-specific antigens is imposed during T cell development and thus reassert the importance of central tolerance in self-nonself discrimination 2–10.
Recent studies established medullar thymic epithelial cells (mTEC) as the prototypic PAE cell phenotype in the thymus 6–9. The expression of a diverse set of genes that encompass cell surface proteins, enzymes, hormones, and structural proteins, which are all restricted tissue distribution, has been detected in mTEC 9, 11, 12. Expression of these genes correlates to the risk of autoimmunity in experimental models of autoimmune diseases. More recently, it has been demonstrated that mutation of aire, a nuclear protein with preferential expression in mTEC, abrogates the expression of a large array of the peripheral antigens in the thymus 1 and prevents deletion of T cells specific for antigens expressed under the control of tissue-specific promoters 13.
Meanwhile, it has also been reported that dendritic cells (DC) and macrophages are possible PAE candidates. In support of this contention are thymus cell-fractionation studies 3, 4, which show that thymic insulin-expressing cells segregate into a low-density fraction that is enriched with DC and macrophages. In addition, several reports demonstrated the co-localization of insulin and other pancreatic hormones with markers of the DC and macrophage lineages in the murine and human thymus 5, 10. However, the functions and mechanisms of these putative PAE in self-tolerance have not been systematically analyzed.
In analysis of the mechanism of immune tolerance in the transgenic mouse prostate cancer model (TRAMP), we have demonstrated that DC constitute a substantial portion of thymic PAE expressing the SV40 large T antigen (Tag) under the control of a prostate-specific promoter 14. While the PAE in TRAMP mice caused complete deletion of SV40 Tag-specific T cells, the subset of PAE responsible for the clonal deletion was not identified. Here we used irradiation chimeric mice to demonstrate that radio-resistant PAE are necessary and sufficient to cause clonal deletion of the Tag-specific T cells. Surprisingly, bone marrow (BM)-derived PAE caused activation-induced cell death (AICD) of self-reactive T cells in the spleen. Our results establish novel function and mechanisms of BM-derived PAE in the induction of tolerance of T cells to tissue-specific antigens in the secondary lymphoid organs.
2.1 Radio-resistant PAE are necessary and sufficient to induce clonal deletion in the thymus
TG-B mice express, at high levels, a T cell receptor from a CD8+ cytotoxic T cell clone that recognizes SV40 Tag presented by the MHC class I molecule H-2Kk15. We recently demonstrated that the transgenic T cells are deleted in TRAMP/TG-B double-transgenic (H-2bxk) mice due to PAE in the thymus 14. In this study, chimeric mice were produced by transferring BM from TG-B+ mice (with or without the SV40 Tag transgene) to lethally irradiated TG-B– mice (with or without SV40 Tag) (Table 1). With the exception of the recipients in group I, all groups of chimeric mice synthesized the SV40 Tag in the non-hematopoietic cells of the recipients and/or in BM-derived donor cells. To confirm that both radio-resistant recipient cells and BM-derived donor cells expressed the Tag in the thymus, we used a previously described method 14, based on RT-PCR plus probing of products by Southern blot, to determine expression of Tag in the thymi of the chimeric mice. As shown in Fig. 1, Tag mRNA was detectable in the thymi of chimeric mice from groups II, III and IV, although the amounts detected in group III were significantly lower than those in groups II and IV. This result is consistent with our immunohistochemical analysis of SV40 Tag protein expression in the thymus 14, which revealed that while both DC and non-DC express Tag, most Tag-expressing cells lack CD11c. As expected, no Tag mRNA was detected in group I thymi.
To analyze the functions of two different lineages of PAE in the thymus, we studied the fate of Tag-reactive T cells by flow cytometry. In comparison to group I, the total thymocytes recovered from reconstituted thymi were also clearly reduced in groups II and IV, but not in group III (Table 1). Moreover, thymi from groups I and III had essentially identical subset distributions, while those from groups II and IV were depleted of CD8+CD4– T cells (data not shown). Among the T cells that express high levels of transgenic TCRβ, the reduction of the mature antigen-specific T cells was even more pronounced (Fig. 2A, B). Groups I and III had comparable numbers of antigen-specific T cells with similar responsiveness to peptide stimulation (Fig. 2C). However, a more than five-fold reduction in the number of mature CD8+ T cells was observed in groups II and IV. Since the common feature of groups II and IV is their shared origin of radio-resistant PAE, our results demonstrate that expression of Tag in the recipient thymic stromal cells is sufficient to induce clonal deletion. In addition, since no clonal deletion was observed in group III chimeras that have BM-derived PAE, expression of Tag in the recipient radio-resistant PAE is also necessary for the clonal deletion of autoreactive T cells.
2.2 BM-derived PAE cause AICD of self-reactive T cells in the secondary lymphoid organs
As shown in Fig. 3, in comparison with group I chimeras, groups II and IV had substantially reduced numbers of transgenic T cells in the spleen. This reduction roughly correlated to that found in the thymi (Fig. 2). Surprisingly, although groups I and III had essentially identical numbers of mature transgenic T cells in the thymi (Fig. 2), the number of transgenic T cells in the spleens of group III chimeras was approximately three- to five-fold lower than found in group I (Fig. 3). These results suggest that BM-derived PAE can reduce the number of autoreactive T cells in the secondary lymphoid organs.
The reduction in the number of T cells resulted in reduced responses of T cells to in vitro stimulation by the cognate peptide. As shown in Fig. 4A, spleen cells from groups II and IV did not proliferate in response to the Tag peptide, while group III mice mounted a significant, although much reduced, proliferation. After in vitro stimulation, potent CTL could be elicited from group I, but not group IV, spleen cells (Fig. 4B). In most experiments, a recall CTL response was not detectable in group II spleens, while a much reduced (about 100-fold less as judged by E/T ratio) CTL activity could be elicited from group III. In some experiments, however, a low but significant CTL response was detected in groups II and III. Thus, optimal tolerance to self antigens requires both lineages of PAE.
It has been demonstrated that host APC can cross-present tissue-specific antigens and thereby cause AICD of T cells 16. However, it has not been tested whether hematopoietic cells can express tissue-specific antigens and induce AICD in the secondary lymphoid organ. To determine if this is the case, we compared spleen cells from groups I and III for activation markers and signs of programmed cell death. As shown in Fig. 5A, the transgenic T cells from group III were clearly being stimulated, as substantial proportions expressed CD69, CD25, and CD24, which are T cell activation markers. Interestingly, the difference in down-regulation of the memory marker CD62L was much less pronounced, which is consistent with the fact that activation of T cells in group III did not lead to stronger CTL recall responses (Fig. 4B).
Instead of inducing strong memory markers, we observed that group III T cells in the spleen had elevated expression of Fas and Fas ligand, which were critical for AICD (Fig. 5B, C). Tunnel assay revealed about a five-fold increase in the proportion of cells undergoing programmed cell death (Fig. 5D). This result demonstrates that PAE of hematopoietic origin induce AICD of self-reactive T cells.
2.3 Peripheral antigen expression by PAE in the secondary lymphoid organs does not correlate with aire expression
It has been demonstrated that aire is preferentially expressed in the mTEC and functions as a transcriptional regulator to control the peripheral organ-specific antigen expression in the thymus 1, 13. Interestingly, significant expression of aire can be detected in the spleen, although the level is about 10% of what is found in the thymus 1. To determine whether aire expression correlates with the synthesis of autoantigens in the secondary lymphoid organs, we compared CD11c+, CD11c–, and total spleen cells for expression of aire and a panel of autoantigens that are found in the PAE in the thymus. As shown in Fig. 6, enrichment of CD11c+ cells increased the aire mRNA by about ten-fold, while elimination of the CD11c+ cells reduced aire mRNA by more than ten-fold. Thus, DC are the primary aire-expressing cells in the spleen. SV40 Tag expression was neither enriched nor depleted in the CD11c+ population (Fig. 6A). Of the three “organ-specific” autoantigens that are found in mTEC but not in DC and macrophages in the thymus 17, insulin and cytochrome P450 1a2 mRNA were found at significant and comparable levels in total spleen cells as well as CD11c+ and CD11c– spleen cells. In contrast, GAD67 mRNA was barely detectable in CD11c+ cells, and depletion of CD11c+ cells did not reduce GAD67 mRNA (Fig. 6B).
We further isolated mRNA from TRAMP mice and compared the RNA expression of aire, SV40 Tag, insulin, P450, and an endogenous mouse prostate protein probasin (mPB) in thymus and spleen. As reported before 1, 13, the aire expression in thymus was ten-fold more than in the spleen. The insulin and P450 showed higher mRNA expression in the thymus, while the SV40 Tag and endogenous prostate protein mPB showed higher mRNA expression in the spleen (Fig. 6C). These results make two points. First, expression of tissue-specific antigens in the spleen is not limited to the SV40 Tag transgene. Secondly, the expression of “organ-specific” autoantigens does not correlate with aire expression in the spleen.
PAE in the thymus constitutively express antigens assumed to be limited to the peripheral organs 4. Although thymic medullar epithelial cells are considered the major PAE in the thymus 7, 9, several groups have reported de novo synthesis of peripheral antigens in hematopoietic cells in the thymus 5, 10, 18. However, the relative contribution of the two types of PAE for clonal deletion in the thymus has not been addressed. Our previous study established that TRAMP mice express prostate-specific antigen in both lineages of PAE 14. In order to identify which subset of PAE induces clonal deletion, we made BM chimeric mice that express the peripheral antigen in only one lineage. Using deletion of SV40 Tag-specific transgenic T cells as a basic readout, we demonstrated that expression of the peripheral antigens in thymic epithelial cells is sufficient to induce clonal deletion. Since no clonal deletion was observed in mice that had only BM-derived PAE, expression of peripheral antigens in the thymic epithelial cells is also necessary for clonal deletion. Given the fact that thymic mTEC are the major PAE in the thymus for the majority of antigens analyzed 9, our conclusion may be generally applicable to clonal deletion of tissue-specific antigens. This is also compatible with recent genetic data showing that mutation of aire, which prevents the expression of many tissue-specific antigens in the mTEC 1, inhibits clonal deletion of tissue-specific T cells in the thymus 13.
In light of previous reports that the expression of minor H, allogeneic MHC, or viral superantigen in either thymic epithelial cells or BM-derived cells is sufficient to induce clonal deletion 19–22, it is surprising that expression of peripheral antigens by BM-derived cells in the thymus does not cause clonal deletion. Since our analysis indicates that the level of expression is lower among the hematopoietic APC (this study), and since the number of BM-derived PAE is substantially lower than the number of thymic epithelial cells 14, the requirement for de novo synthesis by thymic stromal cells may simply be due to their higher expression of peripheral antigen.
Another important observation documented in this study relates to expression and function of PAE in the secondary lymphoid organs. Although the existence of PAE in secondary lymphoid organs had been suggested 10, it was unclear if antigen expression is aire dependent. Our data demonstrate that the expression of peripheral antigens by the APC does not correlate with the level of aire expression. As such, the expression is unlikely to be controlled by aire. Our data further demonstrate that BM-derived PAE reduce the number of autoreactive T cells in periphery and cause AICD of self-reactive T cells in the secondary lymphoid organs. The localization of cells undergoing programmed cell death is consistent with the notion that PAE in the secondary lymphoid organs cause AICD, although an imprinting by thymic hematopoietic PAE during T cell development cannot be ruled out at this stage. These two lines of evidence establish a novel function and mechanism of BM-derived PAE in the induction of tolerance of T cells to tissue-specific antigens. It is likely that this mechanism complements the previously established mechanism by which host APC cross-present tissue antigen to induce AICD 23, 24.
Taken together, we have demonstrated that two populations of PAE play distinct roles in the induction of tolerance of self-reactive T cells. The non-hematopoietic PAE, presumably medullar epithelial cells, induce clonal deletion in the thymus. Perhaps because of lower levels of antigen expression, the hematopoietic PAE are neither necessary nor sufficient to induce clonal deletion in the thymus but induce AICD in the secondary lymphoid organs. Since clonal deletion induced in the thymus is rarely complete under physiological conditions, optimal tolerance may require concerted actions of both populations of PAE.
4 Materials and methods
4.1 Experimental animals
TRAMP mice expressing the SV40 Tag controlled by rat probasin regulatory elements (C57BL/6 background) were purchased from The Jackson Laboratory (Bar Harbor, ME). TG-B mice on the B10.BR background were kindly provided by Dr. T. Geiger from St. Jude's Children's Hospital 25. TRAMP and TG-B mice were bred at the animal facilities of the Ohio State University (Columbus, OH). Mice were typed for SV40 Tag or TCR expression by isolation of mouse tail genomic DNA. The PCR-based screening assays were described previously 14.
4.2 Generation of irradiation BM chimeras
Four groups of chimeric mice with different donor and recipient combinations from TRAMP×TG-B (H-2bxk) F1 phenotypes are presented in Table 1. Briefly, the lethally irradiated (1,000 rad) mice were reconstituted with BM from femurs and tibias of the donor mice after the T cells were depleted with anti-CD4 (Gk1.5) and anti-CD8 (TIB210) monoclonal antibodies. A total of 1.0×107 T cell-depleted BM cells were injected i.v. through tail vein into the recipient mouse. All experiments were performed 8 weeks after BM reconstitution.
The fluorescence-conjugated antibodies anti-CD4 (RM4.5), anti-CD8 (53–6.7), anti-Vβ8.1/8.2 (MR5–2), anti-CD25 (PC61), anti-HSA (M1/69), anti-CD69 (H1.2F3), anti-CD62L, anti-CD28 (37.51), anti-Fas Ligand (MFL3), anti-Fas (Jo2), and the APO-DIRECT Kit were purchased from BD PharMingen (San Diego, CA).
4.4 Peptide synthesis
Peptides (SV40 Tag 560–568 SEFLLEKRI 14 and HSV gB peptide gB498–505 SSIEFARL 26) were synthesized by Research Genetics, Inc. (Huntsville, AL). The peptides were dissolved in dimethyl sulfoxide (DMSO) at a concentration of 10 mg/ml and diluted in PBS or culture medium before use.
4.5 Proliferation of T cells to antigenic peptides and CTL assays
The T cell proliferation and CTL assays have been previously described 14.
4.6 Fractionation of splenic CD11c+ and CD11c– cells and real-time PCR
Splenic CD11c+ and CD11c– cells were fractionated according to a previously described protocol 27. Briefly, collagenase (Sigma) solution (1 mg/ml in 10 mM Hepes-NaOH, pH 7.4) was injected into the spleen before the spleen was sliced and incubated with additional collagenase solution for 60 min at 37°C. A single-cell suspension was obtained by passing the digested spleen through a steel mesh. Red blood cells were lysed with 1× lysis buffer (0.15 M NH4Cl, 1.0 mM KHCO3, 0.1 mM Na2EDTA, pH 7.4). The viable cells were used as the total spleen cell population. CD11c+ and CD11c– cells were magnetically separated using CD11c microbeads and LS+ positive selection columns according to manufacturer's protocol (Miltenyi Biotec Inc., Auburn, CA). Total RNA was extracted from total spleen, CD11c+, and CD11c– cells, and 1 μg/sample was pretreated with RNase-free DNase I before cDNA synthesis using Superscriptase II and oligo(dT) (Invitrogen, Carlsbad, CA). The real-time PCR was carried out in ABI PRISM 7700 Cycler (Applied Biosystems, Foster City, CA) using the QuantiTect SYBR green PCR kit (Qiagen) according to manufacturers’ protocols. The oligonucleotide primers used in real-time PCR were: SV40 Tag, F: 5′-GCTACACTGTTTGTTGCCCA-3′ and R: 5′-CCCCCACATAATTCAAGCAA-3′; Aire, F: 5′-ACCATGGCAGCTTCTGTCCAG-3′ and R: 5′-GCAGCAGGAGCATCTCCAGAG-3′ 9; Insulin I, F: 5′-TATAAAGCTGGTGGGCATCC-3′ and R: 5′-GGGACCACAAAGATGCTGTT-3′; Insulin II, F: 5′-TTTGTCAAGCAGCACCTTTG-3′ and R: 5′-GTCTGAAGGTCACCTGCTCC-3′; GAD67, F: 5′-ATCGTGCAAGCAAGGAAGCA-3′ and R: 5′-GCAAGAGACCTCGGATAGAAGAGT-3′; Cytochrome P450 1a2, F: 5′-GCTGCCATATCTAGAGGCCTTCAT-3′ and R: 5′-TGGTTGACCTGCCACTGGTTTA-3′; the ribosome L-19: F: 5′-CTGAAGGTCAAAGGGAATGTG-3′and R: 5′-GGACAGAGTCTTGTGATCTC-3′. The relative amount of the gene was normalized using the CT values of the sample and the corresponding standard curve. The target gene expression level was quantified using the ratio between the target gene and the ribosome L-19 as a reference gene.
This work was supported by grants from the Department of Defense (DAMD17–03-1-0013 to P. Z.), National Institute of Health (CA82355 to P. Z., CA69091 and CA58033 to Y. L.), and by the Ohio State University Comprehensive Cancer Center. We thank Dr. Yongjun Liu for helpful discussion and Ms. Lynde Shaw for editorial assistance.