Requirement for a Complex Array of Costimulators in the Negative Selection of Autoreactive Thymocytes In Vivo1 Ruixia Li and Dawne M. Page2 Autoreactive thymocytes can be deleted at an immature stage of their development by Ag-induced apoptosis or negative selection. In addition to Ag, negative selection also requires costimulatory signals from APC. We recently used a fetal thymus organ culture system to show that CD5, CD28, and TNF cooperatively regulate deletion of autoreactive thymocytes. Although these experiments provided strong evidence for the action of several costimulators in negative selection, we wished to demonstrate a role for these molecules in a physiologically natural model where thymocytes are deleted in vivo by endogenously expressed Ags. Accordingly, we examined thymocyte deletion in costimulator-null mice in three models of autoantigen-induced negative selection. We compared CD5ⴚ/ⴚ CD28ⴚ/ⴚ mice to CD40Lⴚ/ⴚ mice, which exhibited a profound block in negative selection in all three systems. Surprisingly, only one of the three models revealed a requirement for the CD5 and CD28 costimulators in autoantigen-induced deletion. These results suggest that an extraordinarily complex array of costimulators is involved in negative selection. We predict that different sets of costimulators will be required depending on the timing of negative selection, the Ag, the signal strength, the APC, and whether Ag presentation occurs on class I or class II MHC molecules. The Journal of Immunology, 2001, 166: 6050 – 6057.
T
he TCR for Ag possesses an innate ability to recognize MHC molecules that are programmed into the TCR at the level of nucleotide sequence (1– 4). However, mature T cells do not respond to self-MHC molecules, but instead recognize Ag as peptides bound to self-MHC molecules on APC. To achieve this pattern of responsiveness, developing T cells undergo an intricate selection process in the thymus. T cells with a low affinity/ avidity for self-peptide-MHC complexes survive (positive selection), whereas T cells with a high affinity/avidity for self-peptideMHC complexes are deleted (negative selection) (5–7). It has been shown that up to half of the T cells that are selected to mature in the thymus are then deleted because they are autoreactive (8 –10). Moreover, several studies have linked autoantigen expression in the thymus to resistance to autoimmune disease (11–13). Thus, negative selection is a first line of defense against autoimmunity, and it is important to understand how this process is regulated. In addition to a high avidity TCR stimulus, negative selection also requires costimulatory signals from APC (14, 15). The identification of specific costimulators, however, has been very controversial. Although CD28, Fas, and TNFR have been implicated in negative selection (16 –18), thymocyte deletion is apparently intact in mice lacking these molecules (18 –20). The receptor/ligand pair that has emerged as a master regulator of negative selection in several systems is CD40/CD40 ligand (L).3 Negative selection of CD4⫹ T cells by class II MHC molecules is proDepartment of Biology and the Cancer Center, University of California at San Diego, La Jolla, CA 92093 Received for publication December 4, 2000. Accepted for publication March 9, 2001. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by Grant AI36976 from the National Institute of Allergy and Infectious Diseases. 2 Address correspondence and reprint requests to Dr. Dawne M. Page, Department of Biology and the Cancer Center, University of California at San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0687. E-mail address:
[email protected] 3 Abbreviations used in this paper: L, ligand; FTOC, fetal thymic organ culture; SAg, superantigen; Mtv, mouse mammary tumor virus; HSA, heat-stable Ag.
Copyright © 2001 by The American Association of Immunologists
foundly defective in CD40- or CD40L-null mice (21–24). Since CD40 stimulation of APC increases the expression of many costimulatory molecules, we have hypothesized that CD40 regulates several costimuli that are required for negative selection (21, 25). These costimuli could include the CD28 ligands CD80 and CD86, adhesion molecules such as CD54 (ICAM-1) or CD58 (LFA-3), death receptor ligands such as Fas ligand (FasL), and/or cytokines such as TNF and IL-12 (26). Recently, we investigated whether several costimulators might be jointly controlling thymocyte deletion. Using a combination of blocking Abs in fetal thymus organ culture (FTOC), we found that CD5, the CD28 ligands CD80 and CD86, and TNF cooperatively regulated negative selection of CD4⫹ T cells by class II MHC in three different systems (27). Correspondingly, Kishimoto and Sprent (28) found that CD28, CD43 (a receptor for CD54), and Fas were involved in negative selection of medullary thymocytes induced by injection of bacterial superantigen or peptide Ag into TCR-transgenic mice. Taken together, these studies suggest that CD40 stimulation of APC induces costimuli that cooperatively regulate negative selection of CD4⫹ T cells, including at least CD80, CD86, CD54, TNF, and FasL. CD40 stimulation of APC may also induce CD5L; however, the expression and regulation of CD5L are presently unknown (29 –31). Although these experiments provided strong evidence for the action of several costimulators in negative selection, we wished to demonstrate a role for these molecules in an in vivo model where thymocyte deletion occurs in response to an autoantigen. Such a model is both a more stringent test of costimulator involvement in negative selection and also more applicable to studies of autoimmunity. Therefore, we examined thymocyte deletion caused by endogenous superantigens (SAg), which are produced from an open reading frame in the 3⬘ long terminal repeat of various mouse mammary tumor viruses (Mtv) (32, 33). SAg delete thymocytes bearing specific TCR-V chains, and this deletion has been studied extensively as a model of autoantigen-induced negative selection (33, 34). We chose this model for our investigation because it is a physiologically natural model of negative selection, in that there is no manipulation of TCR or Ag expression in the thymus. Moreover, CD40- or CD40L-null mice are defective in SAg-mediated 0022-1767/01/$02.00
The Journal of Immunology
6051
deletion (21, 23) and thus we could directly compare negative selection in costimulator-null mice to that seen in CD40L-null mice. Surprisingly, we found that the costimulators CD5 and CD28 are required in only one of three models of SAg-dependent negative selection examined. Our results further indicate that the involvement of costimulatory molecules in negative selection in vivo is extraordinarily complex and will probably be different for each system that is examined.
background (H2b) and crossed them to BALB/c (H2d) or D1.LP mice (H2b) for all of the experiments reported here. Additionally, we analyzed very young mice (ages 2–3 wk for the BALB/c experiments and 2–5 wk for the D1.LP experiments) to avoid any cumulative effects of peripheral tolerance on the T cell repertoire. To obtain enough littermates for these experiments, the wild-type mice were heterozygous at both loci (CD5⫹/⫺ CD28⫹/⫺), CD5null mice were CD5⫺/⫺ CD28⫹/⫺ and CD28-null mice were CD5⫹/⫺ CD28⫺/⫺.
Materials and Methods Mice CD5-null, CD28-null, BALB/c and D1.LP mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and maintained in our animal facilities under specific pathogen-free conditions. CD40L-null mice that had been backcrossed to C57BL/6J were obtained from Randolph Noelle (Dartmouth College, Hanover, NH).
Abs and flow cytometric analysis Mice were sacrificed by CO2 inhalation, and lymphocytes were released from thymus, spleen, and/or mesenteric lymph nodes as previously described (18). Surface expression of CD5, CD28, CD4, CD8, heat-stable Ag (HSA), CD44, CD62L, and TCRV chains was determined by Ab staining and flow cytometry with collection of 30,000 –100,000 live cells (18). Abs to CD5, CD28, V5.1/5.2 (MR9-4), V6 (RR4-7), V8.1/8.2 (MR5-2), V11 (RR3-15), HSA, CD44, and CD62L were purchased from BD PharMingen (San Diego, CA). Anti-CD4-PE and anti-CD8-Tri-Color were purchased from Caltag (Burlingame, CA).
PCR and Southern blotting Mtv-6, -8, and -9 were detected by PCR of tail DNA (35). Mtv-7 was detected by Southern blot analysis of tail DNA, as described elsewhere (33).
Statistical analyses Statistical analyses were performed by the Biostatistics Shared Resource Facility (Elizabeth Gilpin, Cancer Center, University of California, San Diego). Examination of the data for the variables of interest indicated that an ANOVA would be appropriate for establishing whether there were differences among groups of mice. Accordingly, if the overall F ratio was significant, costimulator-null mice were compared with wild-type mice using Dunnett’s test to control for multiple testing (36). A further comparison of CD5⫺/⫺CD28⫺/⫺ mice to CD5⫺/⫺ mice was also performed in some cases using a modified t test (Duncan’s procedure).
Results and Discussion Experimental design To investigate whether costimulators are required for negative selection in vivo, we examined thymocyte deletion in response to SAg in three model systems. We chose to analyze the roles of CD5 and CD28, since these molecules exhibited a strong cooperative effect on negative selection in our previous experiments in FTOC (27). We obtained CD5⫺/⫺ and CD28⫺/⫺ mice on the C57BL/6
T cell repertoire in CD5- and CD28-null mice No significant differences in total numbers of thymocytes or splenocytes were noted in these animals. However, the percentage of CD4 and CD8 T cells was increased in the thymus and spleen of CD5⫺/⫺ CD28⫺/⫺ mice (Table I). Both CD4⫹ and CD8⫹ thymocytes were increased in CD5-null mice, and CD4⫹ thymocytes were further increased in CD5⫺/⫺ CD28⫺/⫺ mice. Although Table I reports only the results from the H2b mice from the (CD5⫺/⫺ CD28⫺/⫺ ⫻ BALB/c) crosses, similar results were obtained in the H2b/d mice from these crosses and also in the older H2b mice studied in the D1.LP model (data not shown). These results are consistent with several previous observations. Teh et al. (37) noted an increase in CD4⫹ and CD8⫹ thymocytes in CD5⫺/⫺ CD2⫺/⫺ mice. Pena-Rossi et al. (38) showed that the block in CD4 lineage development in CD4⫺/⫺ mice was partially reversed in CD5⫺/⫺CD4⫺/⫺ mice. Because T cells from CD5⫺/⫺ mice are hyperresponsive to TCR stimulation, CD5 has been reported to be a negative regulator of T cell activation; several groups have proposed that in the absence of CD5, positive selection is enhanced (38 – 40). Yet, another explanation is also possible. We found that CD4⫹ thymocytes were increased in FTOC treated with blocking Abs to CD5 and the CD28 ligands; this increase was observed only in thymuses from mice undergoing negative selection (27). Moreover, these CD4⫹ thymocytes were autoreactive (27). Thus, we proposed that the additional CD4⫹ thymocytes in these cultures had escaped from negative selection rather than experienced enhanced positive selection. In support of this notion, Kishimoto and Sprent (28) found that CD5, CD28, and CD43 were the only three costimulatory receptors of 10 tested that could participate in negative selection in vitro. Now we show that CD5⫺/⫺ CD28⫺/⫺ mice likewise have increased numbers of CD4⫹ and CD8⫹ thymocytes and T cells. It is likely that these cells have escaped from negative selection, although we have not been able to show that they are autoreactive as they respond poorly to APC stimulation due to their CD28-null phenotype (data not shown; Refs. 41 and 42).
Table I. CD4 and CD8 populations in costimulator-null mice Thymus Phenotype
a
Wild-type CD5⫺/⫺ CD28⫺/⫺ CD5⫺/⫺ CD28⫺/⫺
Spleen
% CD4
% CD8
% CD4
% CD8
7.0 ⫾ 1.2 9.0 ⫾ 2.2* 9.4 ⫾ 1.8* 13.2 ⫾ 2.3*$
1.9 ⫾ 0.5 2.5 ⫾ 0.4* 1.9 ⫾ 0.4 2.6 ⫾ 0.8*
10.4 ⫾ 3.0 9.3 ⫾ 2.8 11.5 ⫾ 2.2 14.1 ⫾ 2.6*$
4.1 ⫾ 1.3 4.3 ⫾ 1.5 4.5 ⫾ 0.8 5.7 ⫾ 1.5*$
a CD4 and CD8 populations were analyzed by flow cytometric analysis of lymphocytes from the indicated mice. The mean percentage (⫾SD) of cells is reported for 9 –18 H-2b mice per group. No significant differences in total numbers of thymocytes or splenocytes were noted. For example, in four experiments the average number of thymocytes (⫾SD) was 49 ⫾ 10, 37 ⫾ 8, 54 ⫾ 11, and 45 ⫾ 7 million cells for the wild-type, CD5⫺/⫺, CD28⫺/⫺, and CD5⫺/⫺ CD28⫺/⫺ mice, respectively. ⴱ, Statistically significant difference as compared to wild-type mice ( p ⬍ 0.05, Dunnett’s test); $, statistically significant difference as compared to the CD5⫺/⫺ mice ( p ⬍ 0.05, Duncan’s test).
6052
COSTIMULATORS IN THYMOCYTE-NEGATIVE SELECTION
Deletion of V11⫹ thymocytes and T cells by Mtv-8,9 CD5- and CD28-null mice on the C57BL/6 background (H2b, Mtv-8, -9, -17, -30⫹) were crossed to BALB/c mice (H2d, Mtv-6, -8, -9⫹) to obtain mice that were H2b or H2b/d and that expressed two copies of Mtv-8,9 and one copy of Mtv-6. Mtv-6 deletes V3and V5-bearing T cells, and Mtv-8,9 delete V5-, V11-, and V12-bearing T cells (43). Since Mtv-30 is probably not expressed, and since Mtv-17 has very little effect on V11 deletion in comparison to Mtv-8,9 (43), we did not screen for Mtv-17,30. An examination of SAg-mediated deletion previously showed that CD40L-null mice are profoundly deficient in deletion of V5⫹ and V11⫹ thymocytes, partially deficient in deletion of V12⫹ thymocytes, but not at all deficient in deletion of V3⫹ thymocytes (21). Correspondingly, V3⫹ thymocytes were not rescued from SAg-induced deletion in CD5⫺/⫺ CD28⫺/⫺ mice (data not shown). Thus, we focused our examination on Mtv-6,8,9-mediated deletion of V5⫹ and V11⫹ T cells for these experiments (summary in Table II). Since SAg-mediated deletion is more efficient in the presence of H2-E, we first analyzed Mtv-6,8,9-induced-deletion in H2b/d mice. Fig. 1 shows the V11 profile of thymocytes obtained from these mice. In wild-type mice, negative selection induced by Mtv8,9/H2-E caused the percentage of V11⫹ thymocytes to decrease from 6 to 4% among the CD4⫹ cells and from 5 to 2% among the CD8⫹ cells (Fig. 1, compare wild-type H2b to wild-type H2b/d). As expected, CD40L-null, H2b/d mice exhibited a complete rescue from Mtv-8,9-induced deletion of both CD4⫹ and CD8⫹ V11bearing thymocytes (Fig. 1, compare CD40L⫺/⫺ H2b/d to wildtype H2b/d and wild-type H2b). Interestingly, CD28⫺/⫺ or CD5⫺/⫺ CD28⫺/⫺ H2b/d mice exhibited a partial rescue of V11⫹CD4⫹ thymocytes from negative selection (Fig. 1, top, compare mice in H2b/d group). This rescue appeared to be due solely to the lack of CD28, since increased rescue of V11⫹CD4⫹ cells was not consistently observed in mice lacking both CD5 and CD28. Since the total numbers of thymocytes were similar in wild-type or costimulator-null mice, the increased percentage of V11⫹CD4⫹ thymocytes represents an increase in the total number of these cells. For example, in four experiments, the average number of V11⫹CD4⫹ thymocytes obtained was 1.9 ⫾ 0.4 ⫻ 105, 2.0 ⫾ 0.2 ⫻ 105, 2.6 ⫾ 0.7 ⫻ 105, and 3.3 ⫾ 0.2 ⫻ 105 cells for the wild-type, CD5⫺/⫺, CD28⫺/⫺, and CD5⫺/⫺CD28⫺/⫺ mice, respectively. These results suggest that CD28 is involved in Mtv-8,9-induced deletion of V11⫹CD4⫹ thymocytes.
FIGURE 1. Deletion of V11⫹ thymocytes by Mtv-8,9 in costimulatornull mice. The CD4/CD8/V11 profile was examined in the thymi of mice expressing the indicated costimulators and H2 haplotypes. Reported are the percentages of V11⫹ cells in the CD4⫹ (top) or CD8⫹ (bottom) thymocytes, where each dot represents one mouse (77 mice total) and the line represents the mean value. ⴱ, Significant increase in the percentage of V11⫹ cells as compared with the wild-type, H2b/d animals (p ⬍ 0.05, Dunnett’s test).
Although V11⫹CD8⫹ thymocytes were rescued from negative selection in CD40L-null mice, they were not significantly rescued in the costimulator-null mice (Fig. 1, bottom, H2b/d mice). Thus, CD5 and CD28 are not required for Mtv-8,9-induced deletion of V11⫹CD8⫹ thymocytes. Instead, CD40 stimulation of APC apparently induces costimuli that act through other receptors to delete these cells. Fig. 2 shows the V11 profiles of the splenocytes obtained from these mice. In previous reports, the cells that were rescued from negative selection in CD40L-null mice did not accumulate in the periphery; rather, the autoreactive cells were apparently deleted by uncharacterized mechanisms of peripheral tolerance (21, 23).
Table II. Rescue from SAg-mediated negative selection in coreceptor-null micea Rescue of CD4⫹ Cells
Mouse Line
Thymi CD5⫺/⫺ CD28⫺/⫺ CD5⫺/⫺ CD28⫺/⫺ CD40L⫺/⫺ Periphery CD5⫺/⫺ CD28⫺/⫺ CD5⫺/⫺ CD28⫺/⫺ CD40L⫺/⫺
Rescue of CD8⫹ Cells
Mtv-8,9 SAg V11⫹ TCR
Mtv-6,8,9 SAg V5⫹ TCR
Mtv-7 SAg V6⫹ TCR
Mtv-8,9 SAg V11⫹ TCR
Mtv-6,8,9 SAg V5⫹TCR
Mtv-7 SAg V6⫹ TCR
No ⫹⫹ ⫹⫹ ⫹⫹⫹
No ⫹/⫺ No ⫹⫹⫹
No No No ⫹⫹⫹
No No No ⫹⫹⫹
No No No ⫹
No No No ⫹⫹⫹
No No ⫹⫹ ⫹⫹⫹
No No No ⫹⫹
No No No ⫹
No No No No
No No No ⫹
No No No ⫹⫹⫹
a Rescue from negative selection is reported on a subjective scale, with ⫹⫹⫹ denoting complete rescue. For the Mtv-6,8,9 model, results are summarized only for coreceptor-null mice expressing H2-E. Rescue from negative selection was determined by comparing cell populations obtained in H2-E⫹, coreceptor-null mice to those obtained in H2-E-negative, wild-type mice. For the Mtv-7 model, rescue from negative selection was determined by comparing cell populations obtained in Mtv-7⫹, coreceptor-null mice to those obtained in Mtv-7-negative, wild-type mice.
The Journal of Immunology
6053
FIGURE 2. Accumulation of V11⫹ T cells in the periphery of costimulator-null mice expressing Mtv-8,9. The CD4/CD8/V11 profile was examined in the spleens of the mice from Fig. 1. ⴱ, Significant increase in the percentage of V11⫹ cells as compared with the wild-type, H2b/d animals (p ⬍ 0.05, Dunnett’s test).
FIGURE 3. Expression of V8.1/8.2 in thymocytes of costimulator-null mice expressing Mtv-8,9. The CD4/CD8/V8.1,8.2 profile was examined in the thymi of the mice from Fig. 1. ⴱ, Significant decrease in the percentage of V8⫹ cells as compared with the wild-type, H2b/d animals (p ⬍ 0.05, Dunnett’s test).
However, CD4⫹V11⫹ T cells were present in the periphery of the CD40L⫺/⫺ mice in these experiments (Fig. 2, top). One possibility is that CD4⫹ cells were still present because the mice in this analysis were quite young. However, we note that even these young CD40L⫺/⫺ mice still deleted the peripheral CD8⫹V11⫹ cells (Fig. 2, bottom). The mechanisms of peripheral tolerance underlying these differing effects have not yet been characterized. CD5⫺/⫺ CD28⫺/⫺, H2b/d mice likewise showed an accumulation of CD4⫹V11⫹, but not CD8⫹V11⫹, T cells in their spleens (Fig. 2). Although there appeared to be an increased percentage of V11⫹CD4⫹ splenocytes in CD28⫺/⫺ mice, the difference was not statistically significant. Instead, we noted a better accumulation in CD5⫺/⫺ CD28⫺/⫺ mice as compared with CD28⫺/⫺ mice. It is possible that CD5 does act in conjunction with CD28 in thymic deletion, but that the effect is small and is only seen as the cells accumulate in the periphery. Alternatively, CD5 could be working through mechanisms of peripheral tolerance to induce accumulation of these cells. Taken together, the results in Figs. 1 and 2 indicate that both CD5 and CD28 can contribute to Mtv-8,9-induced deletion of CD4⫹V11⫹ cells. However, rescue of V11⫹ cells was always more profound in CD40L-null mice; thus, other costimulators must also be contributing to negative selection in this system (summary in Fig. 7). Because CD4⫹ and CD8⫹ cells were increased in the CD5⫺/⫺ CD28⫺/⫺ mice in general (Table I), we considered the possibility that the rescue of V11⫹ cells in CD5⫺/⫺ CD28⫺/⫺ mice was not Mtv dependent. Therefore, we examined V8.1/8.2 profiles, since these T cells are not deleted by Mtv-6, -8, or -9 (43), and this population is therefore commonly used as a control for Mtv-6,8,9specific deletion (44 – 46). The percentage of V8.1/8.2⫹ cells did not increase in the costimulator-null mice (Fig. 3). In fact, the percentage of V8.1/8.2⫹ cells among the CD4⫹ population decreased significantly in the CD28⫺/⫺ and CD5⫺/⫺CD28⫺/⫺, but not the CD5⫺/⫺ mice (Fig. 3, top). This result would be expected if V11⫹ T cells are being rescued from Mtv-8,9-induced deletion,
because it is well documented that the V8.1/8.2 population behaves in a compensatory fashion in the presence of these SAg (43– 46). Similar results were observed in the V8.1/8.2 profile of the splenocytes as well (data not shown). Thus, the rescue of the V11 population in CD28⫺/⫺ and CD5⫺/⫺ CD28⫺/⫺ H2b/d mice appears to be Mtv-8,9 dependent. Phenotype of CD4⫹ cells rescued from deletion It was possible that the V11⫹CD4⫹ cells rescued from deletion in the costimulator-null mice were not truly mature or expressed an activated phenotype. We therefore examined these cells for their HSA, CD44, and CD62L profile (Fig. 4). HSA is down-regulated as thymocytes progress from the CD4⫹CD8⫹ (double-positive) stage to the mature CD4⫹ or CD8⫹ stage (e.g., compare Fig. 4, G and H). In H2b mice, 6.0% of the CD4⫹ thymocytes were V11⫹, and 32% of these cells were HSAlow (Fig. 4, A and D). In contrast, CD5⫺/⫺, H2b/d mice exhibited deletion of their CD4⫹V11⫹ thymocytes and only 10% of those remaining were HSAlow (Fig. 4, B and E). As expected, V11⫹CD4⫹ thymocytes were rescued in H2b/d, CD5⫺/⫺ CD28⫺/⫺ mice and 28% of these were HSAlow (Fig. 4, C and F). Thus, the HSA profile of the thymocytes rescued from deletion matched the HSA profile of the V11⫹CD4⫹ cells that matured in the H2b mice. Correspondingly, the V11⫹CD4⫹ splenocytes in the CD5⫺/⫺ CD28⫺/⫺ mice appeared to be mature naive T cells, in that they displayed a normal CD44 and CD62L profile (Fig. 4, K and N, and data not shown). In conclusion, the rescued V11⫹CD4⫹ cells in the costimulator-null mice are both mature and naive based on these parameters. Deletion of V5⫹ thymocytes and T cells by Mtv-6,8,9 We next examined V5 profiles in these same mice, expecting to see similar results as those obtained for the V11 population. Surprisingly, quite different results were obtained. First, CD4⫹V5⫹, but not CD8⫹V5⫹, thymocytes were completely rescued from deletion in CD40L-null mice (Fig. 5, H2b/d mice). Thus, CD40L is
6054
COSTIMULATORS IN THYMOCYTE-NEGATIVE SELECTION
FIGURE 5. Deletion of V5⫹ thymocytes by Mtv-6,8,9 in costimulator-null mice. The CD4/CD8/V5 profile was examined in the thymi of the mice from Fig. 1. ⴱ, Significant increase in the percentage of V5⫹ cells as compared with the wild-type, H2b/d animals (p ⬍ 0.05, Dunnett’s test).
FIGURE 4. Phenotype of CD4⫹V11⫹ thymocytes and splenocytes in costimulator-null mice. Thymocytes and splenocytes from mice of the indicated genotypes were examined by four-color flow cytometric analysis for their expression of CD4, CD8, V11, and HSA, CD44, or CD62L. In this experiment, CD5⫺/⫺ littermates were used as the controls since they exhibited normal deletion of V11⫹CD4⫹ thymocytes. Shown are the V11/CD4 profiles of CD4⫹ thymocytes (A–C) or splenocytes (I–K). The HSA profiles of the CD4⫹V11⫹ thymocytes are shown in D–F, and the HSA profiles of all CD4⫹ and double-positive thymocytes are in G and H for comparison. The percentage of mature HSAlow cells is indicated in these panels. The CD44 profiles of the CD4⫹V11⫹ splenocytes are shown in L–N, and the CD44 profile of all CD4⫹ splenocytes is shown in O for comparison. The percentage of cells with an activated CD44high phenotype is indicated.
not significantly involved in negative selection of the V5⫹CD8⫹ population. Moreover, unlike the results above, CD28 and CD5 contributed either marginally or not at all to deletion of CD4⫹V5⫹ thymocytes (Fig. 5, H2b/d mice). Similar results were obtained in splenocytes (data not shown, but see Table II). Thus, it appears that CD5 and CD28 are not required for deletion of CD4⫹V5⫹ thymocytes by Mtv-6,8,9. Rather, CD40 stimulation of APC apparently induces costimuli that act through other receptors to delete these cells (Fig. 7). The reasons for the discrepancy between costimulator contribution to negative selection in these two systems is not known. Mtv-6 could be responsible for the difference, in that Mtv-6-induced deletion of V5⫹ thymocytes may not require CD5 and CD28. However, this is unlikely because in the cases where we screened for Mtv-6 expression, we did not observe a significant effect of Mtv-6 on V5 deletion over and above that already due to Mtv-8,9 (data not shown). Correspondingly, single Mtv mice reveal that Mtv-9 alone can delete 80 –98% of the V5⫹CD4⫹ T cells (43). A more likely possibility is that deletion of V5⫹ or V11⫹ thymocytes occurs at different stages of thymocyte development or is regulated by different APC. V11 deletion occurs somewhere during the
transition between immature CD4⫹CD8⫹ precursors and mature medullary thymocytes (44), whereas V5 deletion apparently occurs in the thymic medulla (47). Interestingly, bone marrowderived APC are required for V11 deletion (45), whereas V5 deletion can be mediated by bone marrow-derived APC or medullary epithelium (48 –50). Thus, it is possible that V5 deletion by medullary epithelium is not CD5/CD28-dependent and that a requirement for participation of these costimulators is thereby overcome and the thymocytes are deleted. However, this explanation does not agree with data indicating that CD28 is involved in other models of medullary thymocyte-negative selection (28). Deletion of V6⫹ thymocytes and T cells by Mtv-7 Due to the differences in costimulator involvement in negative selection in the systems described above, we investigated the role of CD5/CD28 in another system of SAg-mediated deletion. Deletion of V6⫹ T cells occurs in response to Mtv-7 in both H2-E⫹ and H2-E⫺ mice (51). However, deletion in H2-E⫺ mice (e.g., the D1.LP strain) is weaker and occurs over a period of several weeks. This type of negative selection is completely rescued in CD40-null mice, whereas the stronger H2-E⫹/Mtv-7-dependent deletion is only partially rescued in CD40-null mice (23). Our hypothesis was that roles for costimulators would be more apparent in a system of weak negative selection. To our surprise, this was not the case. For these experiments, we compared deletion of V6⫹ cells in the following D1.LP mice (H2-Ab, H2-E⫺) from 2 to 5 wk of age: 1) CD5⫺/⫺ CD28⫺/⫺, Mtv-7⫺ (negative control for deletion); 2) CD40L⫹/⫺, Mtv-7⫹ (positive control for deletion); 3) CD40L⫺/⫺, Mtv-7⫹ (control for rescue from deletion); and 4) CD5⫺/⫺ CD28⫺/⫺, Mtv-7⫹. As expected, nearly complete rescue of CD4⫹V6⫹ and CD8⫹V6⫹ cells from negative selection was observed in the thymi of CD40L⫺/⫺ mice expressing Mtv-7 (Fig. 6). For example, the percentage of V6⫹ cells among CD4⫹ thymocytes in CD40L⫹/⫺ littermates expressing Mtv-7 (large dashed line with filled triangles) declined from 7.6% at age 2 wk to 5.5% at age 5 wk. In contrast, CD40L-null mice expressing Mtv-7
The Journal of Immunology
FIGURE 6. Deletion of V6⫹ thymocytes by Mtv-7 in costimulatornull mice. The CD4/CD8/V6 profile was examined in the thymi and lymph nodes (data not shown) of the following mice (8 –14/group): 1) CD5⫺/⫺ CD28⫺/⫺, Mtv-7⫺ (negative control for deletion); 2) CD40L⫹/⫺, Mtv-7⫹ (positive control for deletion); 3) CD40L⫺/⫺, Mtv-7⫹ (control for rescue from deletion); and 4) CD5⫺/⫺ CD28⫺/⫺, Mtv-7⫹. Also examined were wild-type, Mtv-7⫺ and wild-type, Mtv-7⫹ mice. However, since the wild-type mice responded the same as the negative and positive controls for deletion, these results are not shown. Reported are the percentages of V6⫹ thymocytes in the CD4⫹ (top) or CD8⫹ (bottom) populations, where each symbol represents one mouse and the lines represent the mean values.
retained high levels of these cells (8 –10%, thick line with open triangles). In fact, there was virtually no Mtv-7-induced negative selection in the CD4⫹ or CD8⫹ thymocytes from CD40L-null mice (compare thick line to Mtv-7⫺ control, small dashed line with filled squares). Some of the cells rescued from negative selection in CD40L-null mice also accumulated in the periphery (data not shown, but see summary in Table II). In contrast to the rescue of V6⫹ cells seen in CD40L⫺/⫺ mice, no rescue whatsoever was observed in CD5⫺/⫺ CD28⫺/⫺ mice, even in the youngest mice examined where deletion was not yet complete. CD5⫺/⫺ CD28⫺/⫺ animals lacking Mtv-7 (small dashed lines with filled squares in Fig. 6, top) possessed 8 –10% V6⫹ cells among their CD4⫹ thymocytes or T cells (data not shown). In CD5⫺/⫺ CD28⫺/⫺ animals expressing Mtv-7, deletion of V6⫹ cells proceeded normally (Fig. 6, compare plain line with open squares to CD40L⫹/⫺, Mtv-7⫹ control). Similar results were observed in CD5⫺/⫺ or CD28⫺/⫺ animals as well (data not shown, but see Table II). Thus, CD5 and CD28 are not required for Mtv7-induced thymocyte deletion; again, CD40 stimulation of APC apparently induces costimuli that act through different receptors to delete these cells (Fig. 7). Why are CD5/CD28 involved in deletion of V11⫹ cells by Mtv-8,9, but not in deletion of V6⫹ cells by Mtv-7? An obvious possibility is the strength of the signal induced by the different SAg. However, Mtv-7 deletion in an H2-E-negative environment is thought to be a weaker signal than Mtv-8,9 deletion on H2-E. Thus, one would expect the Mtv-7 model to be more costimulator dependent. For example, Dautigny et al. (22) found that CD54 (ICAM-1) and Fas were involved in the “weak” Mtv-9-induced deletion of V5⫹ thymocytes in H2-E⫺ mice. Thus, signal strength is not a likely explanation for the different costimulator requirements in these systems.
6055
FIGURE 7. Costimulator involvement in three systems of SAg-mediated negative selection. Shown is a model of the costimulators involved in negative selection of V5-, 6-, and 11-bearing thymocytes by Mtv-6,7,8,9 based on the results presented in Figs. 1– 6 and Table II.
Another possibility is that deletion of V6⫹ and V11⫹ thymocytes depend on different APC. Both models of negative selection apparently require bone marrow-derived cells (45). In an FTOC model, both dendritic cells and B cells were required to induce deletion of V6⫹ cells; it was thought that B cells express large quantities of SAg, which they then pass to dendritic cells (46). Similarly, Frey et al. (52) showed that B cells were likely required for deletion of V11⫹ cells. However, V6 and V11 deletion both proceed normally in B cell-deficient mice (53). Moreover, recent studies of thymic dendritic cells show that they do express Mtv-7,8 by PCR and functional studies (54). Finally, Webb and Sprent (55) showed that CD8 T cells were responsible for inducing neonatal tolerance to Mtv-7. Thus, the APC that induce deletion in these systems are still unknown, and we currently do not understand why costimulator requirements are different in these two systems. Summary CD40L was clearly involved in thymic deletion in all three models examined, except for deletion of V5⫹CD8⫹ cells (Fig. 7). Thus, CD40/CD40L has again emerged as a master regulator of class II MHC-mediated negative selection. However, the precise costimulatory molecules regulated by CD40 that are required for negative selection in vivo remain elusive. Only in the case of negative selection of V11⫹CD4⫹ cells were CD5 and CD28 shown to be significantly involved. Here, CD40 is probably regulating the CD28 ligands CD80 and CD86 (26), but the expression and regulation of CD5L are unknown (29 –31). In the other models of negative selection examined here, CD40 may induce other costimuli required for thymocyte deletion, perhaps CD54 (ICAM-1), FasL, or TNF (22, 27, 28). These molecules could regulate negative selection separately or in combination with CD5 and CD28. Based on these results, we predict that a complex array of costimulatory receptors are involved in negative selection in vivo and that the cohort of receptors involved will vary depending on the timing of negative selection, the Ag, the signal strength, the APC, and whether Ag presentation occurs on class I or class II MHC molecules.
6056
Acknowledgments We thank Steve Hedrick, Laura Vogel, Meei-Yun Lin, and Rachel Soloff for their thoughtful review of this manuscript. We also thank Betsy Gilpin (Biostatistics Shared Resource Facility, Cancer Center) for her help with the statistical analyses and Brian Parmeter for mouse husbandry.
References 1. Garboczi, D. N., U. Utz, P. Ghosh, A. Seth, J. Kim, E. A. VanTienhoven, W. E. Biddison, and D. C. Wiley. 1996. Assembly, specific binding, and crystallization of a human TCR-␣ with an antigenic Tax peptide from human T lymphotropic virus type 1 and the class I MHC molecule HLA-A2. J. Immunol. 157:5403. 2. Garcia, K. C., M. Degano, R. L. Stanfield, A. Brunmark, M. R. Jackson, P. A. Peterson, L. Teyton, and I. A. Wilson. 1996. An ␣ T cell receptor structure at 2.5 A and its orientation in the TCR-MHC complex. Science 274:209. 3. Sim, B. C., L. Zerva, M. I. Greene, and N. R. J. Gascoigne. 1996. Control of MHC restriction by TCR V␣ CDR1 and CDR2. Science 273:963. 4. Zerrahn, J., W. Held, and D. H. Raulet. 1997. The MHC reactivity of the T cell repertoire prior to positive and negative selection. Cell 88:627. 5. Alam, S. M., P. J. Travers, J. L. Wung, W. Nasholds, S. Redpath, S. C. Jameson, and N. R. Gascoigne. 1996. T-cell-receptor affinity and thymocyte positive selection. Nature 381:616. 6. Hogquist, K. A., and M. J. Bevan. 1996. The nature of the peptide/MHC ligand involved in positive selection. Semin. Immunol. 8:63. 7. Liu, C. P., F. Crawford, P. Marrack, and J. Kappler. 1998. T cell positive selection by a high density, low affinity ligand. Proc. Natl. Acad. Sci. USA 95:4522. 8. Ignatowicz, L., J. Kappler, and P. Marrack. 1996. The repertoire of T cells shaped by a single MHC/peptide ligand. Cell 84:521. 9. Merkenschlager, M., D. Graf, M. Lovatt, U. Bommhardt, R. Zamoyska, and A. G. Fisher. 1997. How many thymocytes audition for selection? J. Exp. Med. 186:1149. 10. van Meerwijk, J. P., S. Marguerat, R. K. Lees, R. N. Germain, B. J. Fowlkes, and H. R. MacDonald. 1997. Quantitative impact of thymic clonal deletion on the T cell repertoire. J. Exp. Med. 185:377. 11. Egwuagu, C. E., P. Charukamnoetkanok, and I. Gery. 1997. Thymic expression of autoantigens correlates with resistance to autoimmune disease. J. Immunol. 159:3109. 12. Pugliese, A., M. Zeller, A. J. Fernandez, L. J. Zalcberg, R. J. Bartlett, C. Ricordi, M. Pietropaolo, G. S. Eisenbarth, S. T. Bennett, and D. D. Patel. 1997. The insulin gene is transcribed in the human thymus and transcription levels correlated with allelic variation at the INS VNTR-IDDM2 susceptibility locus for type 1 diabetes. Nat. Genet. 15:293. 13. Vafiadis, P., S. T. Bennett, J. A. Todd, J. Nadeau, R. Grabs, C. G. Goodyer, S. Wickramasinghe, E. Colle, and C. Polychronakos. 1997. Insulin expression in human thymus is modulated by INS VNTR alleles at the IDDM2 locus. Nat. Genet. 15:289. 14. Page, D. M., L. P. Kane, J. P. Allison, and S. M. Hedrick. 1993. Two signals are required for negative selection of CD4⫹CD8⫹ thymocytes. J. Immunol. 151: 1868. 15. Aiba, Y., O. Mazda, M. M. Davis, S. Muramatsu, and Y. Katsura. 1994. Requirement of a second signal from antigen presenting cells in the clonal deletion of immature T cells. Int. Immunol. 6:1475. 16. Punt, J. A., B. A. Osborne, Y. Takahama, S. O. Sharrow, and A. Singer. 1994. Negative selection of CD4⫹CD8⫹ thymocytes by T cell receptor-induced apoptosis requires a costimulatory signal that can be provided by CD28. J. Exp. Med. 179:709. 17. Kishimoto, H., and J. Sprent. 1997. Negative selection in the thymus includes semimature T cells. J. Exp. Med. 185:263. 18. Page, D. M., E. M. Roberts, J. J. Peschon, and S. M. Hedrick. 1998. TNF receptor-deficient mice reveal striking differences between several models of thymocyte negative selection. J. Immunol. 160:120. 19. Singer, G. G., and A. K. Abbas. 1994. The Fas antigen is involved in peripheral but not thymic deletion of T lymphocytes in T cell receptor transgenic mice. Immunity 1:365. 20. Walunas, T. L., A. I. Sperling, R. Khattri, C. B. Thompson, and J. A. Bluestone. 1996. CD28 expression is not essential for positive and negative selection of thymocytes or peripheral T cell tolerance. J. Immunol. 156:1006. 21. Foy, T. M., D. M. Page, T. J. Waldschmidt, A. Schoneveld, J. D. Laman, S. R. Masters, L. Tygrett, J. A. Ledbetter, A. Aruffo, E. Claassen, et al. 1995. An essential role for gp39, the ligand for CD40, in thymic selection. J. Exp. Med. 182:1377. 22. Dautigny, N., A. Le Campion, and B. Lucas. 1999. Timing and casting for actors of thymic negative selection. J. Immunol. 162:1294. 23. Ozaki, M. E., B. A. Coren, T. N. Huynh, D. J. Redondo, H. Kikutani, and S. R. Webb. 1999. CD4⫹ T cell responses to CD40-deficient APCs: defects in proliferation and negative selection apply only with B cells as APCs. J. Immunol. 163:5250. 24. Ferrero, I., F. Anjuere, P. Martin, G. Martinez del Hoyo, M. L. Fraga, N. Wright, R. Varona, G. Marquez, and C. Ardavin. 1999. Functional and phenotypic analysis of thymic B cells: role in the induction of T cell negative selection. Eur. J. Immunol. 29:1598. 25. Page, D. M., L. P. Kane, T. M. Onami, and S. M. Hedrick. 1996. Cellular and biochemical requirements for thymocyte negative selection. Semin. Immunol. 8:69.
COSTIMULATORS IN THYMOCYTE-NEGATIVE SELECTION 26. Grewal, I. S., and R. A. Flavell. 1998. CD40 and CD154 in cell-mediated immunity. Annu. Rev. Immunol. 16:111. 27. Page, D. M. 1999. Cutting edge: thymic selection and autoreactivity are regulated by multiple coreceptors involved in T cell activation. J. Immunol. 163:3577. 28. Kishimoto, H., and J. Sprent. 1999. Several different cell surface molecules control negative selection of medullary thymocytes. J. Exp. Med. 190:65. 29. Van de Velde, H., I. von Hoegen, W. Luo, J. R. Parnes, and K. Thielemans. 1991. The B-cell surface protein CD72/Lyb-2 is the ligand for CD5. Nature 351:662. 30. Biancone, L., M. A. Bowen, A. Lim, A. Aruffo, G. Andres, and I. Stamenkovic. 1996. Identification of a novel inducible cell-surface ligand of CD5 on activated lymphocytes. J. Exp. Med. 184:811. 31. Bikah, G., F. M. Lynd, A. A. Aruffo, J. A. Ledbetter, and S. Bondada. 1998. A role for CD5 in cognate interactions between T cells and B cells, and identification of a novel ligand for CD5. Int. Immunol. 10:1185. 32. Marrack, P., G. M. Winslow, Y. Choi, M. Scherer, A. Pullen, J. White, and J. W. Kappler. 1993. The bacterial and mouse mammary tumor virus superantigens; two different families of proteins with the same functions. Immunol. Rev. 131:79. 33. Scherer, M. T., L. Ignatowicz, G. M. Winslow, J. W. Kappler, and P. Marrack. 1993. Superantigens: bacterial and viral proteins that manipulate the immune system. Annu. Rev. Cell Biol. 9:101. 34. Kappler, J. W., U. Staerz, J. White, and P. C. Marrack. 1988. Self-tolerance eliminates T cells specific for Mls-modified products of the major histocompatibility complex. Nature 332:35. 35. Held, W., G. A. Waanders, A. N. Shakhov, L. Scarpellino, H. Acha-Orbea, and H. R. MacDonald. 1993. Superantigen-induced immune stimulation amplifies mouse mammary tumor virus infection and allows virus transmission. Cell 74:529. 36. Winer, B. J. 1971. Statistical Principles in Experimental Design. McGraw-Hill, New York. 37. Teh, S. J., N. Killeen, A. Tarakhovsky, D. R. Littman, and H. S. Teh. 1997. CD2 regulates the positive selection and function of antigen-specific CD4- CD8⫹ T cells. Blood 89:1308. 38. Pena-Rossi, C., L. A. Zuckerman, J. Strong, J. Kwan, W. Ferris, S. Chan, A. Tarakhovsky, A. D. Beyers, and N. Killeen. 1999. Negative regulation of CD4 lineage development and responses by CD5. J. Immunol. 163:6494. 39. Tarakhovsky, A., S. B. Kanner, J. Hombach, J. A. Ledbetter, W. Muller, N. Killeen, and K. Rajewsky. 1995. A role for CD5 in TCR-mediated signal transduction and thymocyte selection. Science 269:535. 40. Chan, S., C. Waltzinger, A. Tarakhovsky, C. Benoist, and D. Mathis. 1999. An influence of CD5 on the selection of CD4-lineage T cells. Eur. J. Immunol. 29:2916. 41. Shahinian, A., K. Pfeffer, K. P. Lee, T. M. Kundig, K. Kishihara, A. Wakeham, K. Kawai, P. S. Ohashi, C. B. Thompson, and T. W. Mak. 1994. Differential T cell costimulatory requirements in CD28-deficient mice. J. Immunol. 152:3254. 42. Green, J. M., P. J. Noel, A. I. Sperling, T. L. Walunas, G. S. Gray, J. A. Bluestone, and C. B. Thompson. 1994. Absence of B7-dependent responses in CD28-deficient mice. Immunity 1:501. 43. Scherer, M. T., L. Ignatowicz, A. Pullen, J. Kappler, and P. Marrack. 1995. The use of mammary tumor virus (Mtv)-negative and single-Mtv mice to evaluate the effects of endogenous viral superantigens on the T cell repertoire. J. Exp. Med. 182:1493. 44. Matsuzaki, G., Y. Yoshikai, M. Ogimoto, K. Kishihara, and K. Nomoto. 1990. Clonal deletion of self-reactive T cells at the early stage of T cell development in thymus of radiation bone marrow chimeras. J. Immunol. 145:46. 45. Yoshikai, Y., M. Ogimoto, G. Matsuzaki, and K. Nomoto. 1990. Bone marrowderived cells are essential for intrathymic deletion of self-reactive T cells in both the host- and donor-derived thymocytes of fully allogeneic bone marrow chimeras. J. Immunol. 145:505. 46. Mazda, O., Y. Watanabe, J. Gyotoku, and Y. Katsura. 1991. Requirement of dendritic cells and B cells in the clonal deletion of Mls-reactive T cells in the thymus. J. Exp. Med. 173:539. 47. Surh, C. D., and J. Sprent. 1994. T-cell apoptosis detected in situ during positive and negative selection in the thymus. Nature 372:100. 48. Burkly, L. C., S. Degermann, J. Longley, J. Hagman, R. L. Brinster, D. Lo, and R. A. Flavell. 1993. Clonal deletion of V5⫹ T cells by transgenic I-E restricted to thymic medullary epithelium. J. Immunol. 151:3954. 49. Woodland, D., M. P. Happ, J. Bill, and E. Palmer. 1990. Requirement for cotolerogenic gene products in the clonal deletion of I-E reactive T cells. Science 247:964. 50. Bill, J., O. Kanagawa, J. Linten, Y. Utsunomiya, and E. Palmer. 1990. Class I and class II MHC gene products differentially affect the fate of V5 bearing thymocytes. J. Mol. Cell. Immunol. 4:269. 51. Webb, S. R., J. Hutchinson, K. Hayden, and J. Sprent. 1994. Expansion/deletion of mature T cells exposed to endogenous superantigens in vivo. J. Immunol. 152:586. 52. Frey, J. R., B. Ernst, C. D. Surh, and J. Sprent. 1992. Thymus-grafted SCID mice show transient thymopoiesis and limited depletion of V11⫹ T cells. J. Exp. Med. 175:1067. 53. Beutner, U., E. Kraus, D. Kitamura, K. Rajewsky, and B. T. Huber. 1994. B cells are essential for murine mammary tumor virus transmission, but not for presentation of endogenous superantigens. J. Exp. Med. 179:1457. 54. Ardavin, C., G. Waanders, I. Ferrero, F. Anjuere, H. Acha-Orbea, and H. R. MacDonald. 1996. Expression and presentation of endogenous mouse mammary tumor virus superantigens by thymic and splenic dendritic cells and B cells. J. Immunol. 157:2789. 55. Webb, S. R., and J. Sprent. 1990. Induction of neonatal tolerance to Mlsa antigens by CD8⫹ T cells. Science 248:1643.
