Thymic medullary epithelium and thymocyte self-tolerance require cooperation between CD28-CD80/86 and CD40-CD40L costimulatory pathways

Abstract

A critical process during thymic development of the T cell repertoire is the induction of self-tolerance. Tolerance in developing T cells is highly dependent on medullary thymic epithelial cells (mTEC), and mTEC development in turn requires signals from mature single-positive thymocytes, a bidirectional relationship termed thymus crosstalk. We show that CD28-CD80/86 and CD40-CD40L costimulatory interactions, which mediate negative selection and self-tolerance, upregulate expression of LTα, LTβ, and receptor activator for NF-κB in the thymus and are necessary for medullary development. Combined absence of CD28-CD80/86 and CD40-CD40L results in profound deficiency in mTEC development comparable to that observed in the absence of single-positive thymocytes. This requirement for costimulatory signaling is maintained even in a TCR transgenic model of high-affinity TCR-ligand interactions. CD4 thymocytes maturing in the altered thymic epithelial environment of CD40/CD80/86 knockout mice are highly autoreactive in vitro and are lethal in congenic adoptive transfer in vivo, demonstrating a critical role for these costimulatory pathways in self-tolerance as well as thymic epithelial development. These findings demonstrate that cooperativity between CD28-CD80/86 and CD40-CD40L pathways is required for normal medullary epithelium and for maintenance of self-tolerance in thymocyte development.

Figure 1

Altered thymic microenvironment in CD40/CD80/86 KO mice. Thymic cryosections from WT, CD80/86 KO, CD40 KO, CD40/CD80/86 KO and TCRα KO were stained with a combination of (left panel) anti-keratin 14 (medullary epithelium, red) and anti-keratin-8 (cortical and some medullary epithelium, green) or (right panel) with UEA1 (medullary epithelium, green) and anti-Ly51 (cortical epithelium, red). Bars, 2 mm. Thymic cross sections from 3 mice per strain were stained and representative images are shown. Mice were all 4 weeks of age.

Figure 2

Number of thymic medullary epithelial cells is decreased in CD40/CD80/86 KO mice while number of CD4 SP thymocytes is increased. (A) Thymic stromal cells from BALB WT, CD80/86 KO, CD40 KO and CD40/CD80/86 KO mice were prepared as described in Materials and Methods and stained with anti-CD45, UEA-1, anti-Ly51 and anti-IA/IE. After first gating on CD45^neg, IA/IE⁺ cells, mTECs were identified as UEA⁺ and cTEC as UEA^negLy51⁺. Flow cytometry dot plots are representative of > 9 mice per strain. (B) Thymic stromal cells were prepared as in (A). After gating on CD45^neg, IA/IE⁺ cells, numbers of mTEC (UEA⁺) and cTEC (UEA^neg, Ly51⁺) were calculated for each group. Data shown are mean ± SE for ≥ 9 mice per strain. All mice were between 3-5 weeks of age. **, p≤ 0.01; ***, p≤ 0.001. (C) Thymocytes from 4 week old WT, CD80/86 KO, CD40 KO and CD40/CD80/86 KO mice were stained with anti-CD4 and anti-CD8 and numbers of CD4 and CD8 SP were calculated after gating on the respective single positive populations. *, p≤ 0.05; **, p≤ 0.01; ***, p≤ 0.001 ****, p≤ 0.0001. (D) mTEC population in CD40/CD80/86 KO mice fails to expand after birth. Thymic stromal cells from BALB WT and CD40/80/86 KO mice were prepared from mice 1-32 days of age and stained as in B. After gating on CD45^neg, IA/IE+ cells, numbers of mTEC (UEA-1+) were determined. Data shown are the mean ± SE for ≥ 3 mice per strain.

Figure 3

Quantitative RT-PCR analysis of sorted BALB WT, CD80/86 KO, CD40 KO, and CD40/CD80/86 KO CD4 SP thymocytes, and WT and CD28/CD40L mTECs. (A) mRNA expression of LTα, LTβ, RANKL, and CCR7 in CD4 SP thymocytes was normalized to GAPDH mRNA and values for the WT DP samples were arbitrarily set to 1. Data shown are mean ± SE for 5 independent measurements for each strain (*, p ≤0.05; **, p ≤0.01; ****, p ≤0.0001). All mice were 4 weeks of age. (B) mTEC were sorted as described in Material and Methods by gating on CD45^neg, MHC II⁺, UEA⁺ cells. mRNA expression of RANK, CCL19, CCL21, AIRE, and LTβR was normalized to GAPDH. Data shown are mean ± SE for n=5 (WT) or n=4 (CD28/CD40L K0) independent measurements. For each independent measurement, RNA was prepared from mTECs sorted from a minimum of 4 pooled thymuses per strain. All mice were 2 weeks of age. (p=0.1 for CCL21; **, p<0.01; ****p<0.0001).

Figure 4

Number of thymic medullary epithelial cells in CD40/LTβR KO mice is reduced relative to LTβR KO mice. Thymic stromal cells from 3-5 week old mice were prepared as described in Materials and Methods and stained with anti-CD45, UEA-1, anti-Ly51 and anti-IA/IE. After gating on CD45^neg, IA/IE⁺ cells, numbers of mTEC (UEA⁺), and cTEC (UEA^neg, Ly51⁺) were calculated for each group. Data shown are mean ± SE for ≥ 3 mice per strain. (*, p<0.05; **, p<0.01, ****p<0.0001), ns – not significant.

Figure 5

High affinity TCR-MHC interactions are not sufficient to overcome requirement for costimulation. (A) Thymocytes from CB6F1 DO11tg^neg (H2^d×b), CB6F1 DO11 TCRtg⁺(H2^d×b), BALB DO11tg^neg (H2^d), and BALB DO11 TCR tg⁺(H2^d) mice were stained with anti-CD4, anti-CD8, KJ126 (DO11 TCRtg⁺ samples only), and anti-CD5 and analyzed by flow cytometry. Histograms show CD5 expression on gated CD4 SP thymocytes from CB6F1 DO11tg^neg (H2^d×b) and BALB DO11tg^neg (H2^d) and KJ126⁺ CD4 SP thymocytes from and CB6F1 DO11 TCRtg⁺ (H2^d×b), and BALB DO11 TCR tg⁺(H2^d) mice. (B) Preparations of thymic stromal cells from costimulation intact as well as CD40/CD80/86 deficient CB6F1 DO11tg^neg (H2^d×b), CB6F1 DO11 TCRtg⁺ (H2^d×b), BALB DO11 TCR tg^neg(H2^d), and BALB DO11 TCR tg⁺(H2^d) mice were stained with anti-CD45, UEA-1, anti-Ly51 and anti-IA/IE to identify mTEC (CD45^neg, IA/IE⁺, UEA⁺) and cTEC (CD45^neg, IA/IE⁺,UEA^neg, Ly51⁺). Thymocytes were stained with anti-CD4 and anti-CD8 to identify CD4 SP cells. Numbers of mTEC, cTEC, and CD4 SP thymocytes for each strain are shown. Bar graphs show the mean ± SE for ≥ 4 mice per strain (*, p < 0.05;**, p < 0.01; ***, p < 0.001).

Figure 6

Changes in number (left panels) and frequency (right panels) of thymocyte subpopulations in the absence of costimulatory interactions. Thymocytes from 4 week old BALB WT, CD80/86 KO, CD40 KO and CD40/CD80/86 KO mice were stained with anti-CD3 and CD1-PBS-57 tetramers to identify iNKT cells (top panels), or anti-CD4, anti-CD8 and anti-FoxP3 to identify Treg cells (bottom panels) and analyzed by flow cytometry. Bar graphs show the mean ± SE for 4-9 mice per strain. (*, p≤ 0.05; **, p≤ 0.01; ***, p≤ 0.001 ****, p≤ 0.0001).

Fig 7

CD40/CD80/86 KO thymocytes proliferate in syngeneic MLRs in a costimulation-dependent manner. (A) CD8-depleted thymocytes from BALB WT, CD80/86 KO, CD40 KO and CD40/CD80/86 KO mice were cultured with T cell-depleted BALB or B6 WT splenocytes for 72 hr prior to addition of 3H-thymidine for 16 hr. Results shown are means of triplicate wells ± SE and are representative of 8 independent experiments. (B) CD8-depleted BALB CD40/CD80/86 KO thymocytes were cultured with T-depleted BALB WT, CD80/86 KO, CD40 KO and CD40/CD80/86 KO splenocytes for 72 hr prior to addition of 3H-thymidine for 16 hr. Results shown represent means of triplicate wells ± SE and are representative of 2 independent experiments.

Figure 8

Transfer of CD40/80/86 KO thymocytes into syngeneic BALB nu/nu mice results in rapid death. Total thymocytes were purified from BALB WT, CD80/86 KO, CD40 KO and CD40/80/86 KO mice and 1 × 10⁷ (1×) or 0.5 × 10⁷ (0.5×) cells were transferred i.v. to BALB nu/nu recipient mice. (A) Mice were monitored daily for 13. 5 weeks and the proportion of surviving mice are shown plotted against days post transfer. 15-20 mice were included in each group. (B) Splenomegaly and increased numbers of activated CD8 T cells in BALB nu/nu mice receiving CD40/CD80/86 KO thymocytes. Total thymocytes were purified from BALB WT, CD80/86 KO, CD40 KO and CD40/80/86 KO mice and transferred i.v. to BALB nu/nu recipient mice as in Fig. 8A. Single cell suspensions were prepared from spleens harvested from BALB nu/nu recipient mice 4-6.5 weeks after receiving thymocytes and stained with anti-CD4, anti-CD8 and anti-CD69. Data shown are means ± SE for WT (n=9), CD80/86 KO (n=6), CD40 KO (n=5) and CD40/CD80/86 KO (n=17). 1 × 10⁷ thymocytes were transferred for WT, CD80/86 KO and CD40 KO, while the data for the CD40/CD80/86 KO transfers represent both 1× (1 × 10⁷) and 0.5× (0.5 × 10⁷ ) cells transferred (*, p≤ 0.05; **, p≤ 0.01; ***, p≤ 0.001 ****, p≤ 0.0001).