Regulation of CD8+ T Cell Development by Thymus-Specific Proteasomes

See allHide authors and affiliations

Science  01 Jun 2007:
Vol. 316, Issue 5829, pp. 1349-1353
DOI: 10.1126/science.1141915


Proteasomes are responsible for generating peptides presented by the class I major histocompatibility complex (MHC) molecules of the immune system. Here, we report the identification of a previously unrecognized catalytic subunit called β5t. β5t is expressed exclusively in cortical thymic epithelial cells, which are responsible for the positive selection of developing thymocytes. Although the chymotrypsin-like activity of proteasomes is considered to be important for the production of peptides with high affinities for MHC class I clefts, incorporation of β5t into proteasomes in place of β5 or β5i selectively reduces this activity. We also found that β5t-deficient mice displayed defective development of CD8+ T cells in the thymus. Our results suggest a key role for β5t in generating the MHC class I–restricted CD8+ T cell repertoire during thymic selection.

Proteasomes are multicatalytic proteinase complexes that are responsible for regulated proteolysis in eukaryotic cells and essential for the generation of antigenic peptides presented by major histocompatibility complex (MHC) class I molecules of the immune system in jawed vertebrates (1). Proteolysis is conducted by 20S proteasomes, which are large complexes composed of 28 subunits arranged as a cylinder in four heteroheptameric rings: α1-7, β1-7, β1-7, and α1-7. Among these, the three subunits called β1, β2, and β5 perform the catalytic function. In vertebrates, three additional β1i, β2i, and β5i subunits are induced by interferon (IFN)–γ and are preferentially incorporated into proteasomes. These immunoproteasomes produce antigenic peptides more efficiently than do constitutive proteasomes, and they play an important role in the elimination of virus-infected and tumor cells by CD8+ T cells (2).

During a search for proteasome-related genes in a genome database (, we found a previously unrecognized gene product with high homology to β subunits of 20S proteasomes. The gene encoding it is located adjacent to the gene for β5, and the gene product is encoded by a single exon in both human and mouse genomes (fig. S1A). Orthologous genes are found specifically in vertebrates, implying that this gene emerged synchronously with genes involved in adaptive immunity in evolution, as did the immunoproteasomes (2). The expression of this gene was biased toward the thymus in an expressed sequence tag database (UniGene Mm.32009). Multiple sequence alignments showed that its putative active-site threonine residue was preceded by a propeptide that ends with a glycine residue, a hallmark of active β subunits (36) (fig. S1B). The dendrogram and homologies among catalytic β subunits indicated its close relation with β5 and β5i (Fig. 1A and fig. S1C), and we therefore named this gene b5t, as a thymus-specific β5 family member. Northern blot analysis confirmed the exclusive transcription of β5t in the thymus (fig. S1D). Immunoblot analysis of various mouse organs, with the use of an antibody specific to β5t (anti-β5t), further confirmed that the expression was thymus-specific (Fig. 1B). In comparison, β5i was predominantly expressed in a broad range of immune tissues (including the thymus and spleen and to a lesser extent in lung and liver), and it further differed from the wide distribution of β5 (Fig. 1B).

Fig. 1.

β5t is a catalytic proteasome subunit in the thymus. (A) Percentages of amino acid sequence identities (boxed in black) and similarities (white box) among primary structures of mouse catalytic subunits obtained with the BLAST2 program (28). (B) Immunoblot analysis of various tissues of 3-week-old mice with the indicated antibodies. (C) Extracts of mouse thymus were immunoprecipitated with anti-20S or anti-β5t, followed by immunoblotting with the indicated antibodies. IP, immunoprecipitation. (D) 2D-PAGE analysis of samples in (C) with Coomassie staining. 20S proteasomes of brain extracts represent constitutive proteasomes. All spots were identified by tandem mass spectrometry. One of the β5t spots (represented by arrows) overlaps with that of α6.

We next examined how β5t is incorporated into proteasomes. Thymic extracts were immunoprecipitated with an antibody to the 20S proteasome (anti-20S), (cross-specific for all 20S proteasomes) and with anti-β5t. These antibodies immunodepleted 20S proteasomes and β5t, respectively (fig. S2A). Quantitative immunoblotting of α5, an invariable subunit of 20S proteasomes, revealed that ∼20% of the 20S proteasomes in the thymus contain β5t (fig. S2B). β5 and β5i were barely detected in the β5t-containing proteasomes, indicating that β5t was incorporated into 20S proteasomes, in place of β5 or β5i (Fig. 1C). Moreover, the IFN-γ–inducible subunits β1i and β2i were preferentially incorporated into the β5t-containing proteasomes, in place of constitutive subunits β1 and β2 (Fig. 1C). The above findings were further confirmed by analysis of the precipitated products by two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) (Fig. 1D). Spots for β1, β2, β5, and β5i were faintly detected in the β5t-containing proteasomes, whereas spots for β5t emerged together with those for β1i and β2i (Fig. 1D, middle). We therefore propose to call this distinct subtype of proteasomes “thymoproteasomes.”

Proteasomes are responsible for the production of MHC class I–binding peptides and the sole enzymes that determine the C termini of the peptides (7, 8). Previous crystal structural studies of MHC class I complexes have revealed that the hydrophobic C-terminal anchor residues of the peptides are essential for high-affinity peptide binding into the clefts of MHC class I complexes (9). Some types of MHC molecules prefer basic C termini (1). The 20S proteasomes have at least three types of peptidase activities (chymotrypsin-like, trypsin-like, and caspase-like activities), which cleave peptide bonds after hydrophobic, basic, and acidic amino acids, respectively (10). The specificity is determined by the nature of the amino acids that compose the so-called S1 pocket (11, 12). For the production of high-affinity MHC class I ligands, the chymotrypsin-like activity carried by β5 and β5i, whose S1 pockets are composed mostly of hydrophobic amino acids, is important (13). A comparison of the S1 pockets of the three β5 families revealed that the pocket of β5t is mainly composed of hydrophilic residues, in marked contrast to those of β5 and β5i (Fig. 2A), suggesting that thymoproteasomes have weaker chymotrypsin-like activity. To test this idea, we analyzed the peptidase activities of β5t-overexpressing cells, where ∼90% of β5 was replaced by β5t (Fig. 2, B and C). We observed that the precursor forms of β5t were processed into mature forms after incorporation into 20S and 26S proteasomes (Fig. 2B, bottom left). In β5t-expressing cells, the chymotrypsin-like activity was exclusively reduced by 60 to 70%, without affecting the other two activities in both 20S and 26S fractions (Fig. 2B). Kinetic analysis of the chymotrypsin-like activity revealed that β5t reduced both the maximum velocity and the Michaelis constant values for the activity (Fig. 2D), which is opposite to the effect that β5i has on these quantities (14). However, these cells still possessed normal protein-degrading activity, as assessed by the degradation of ornithine decarboxylase (ODC) (Fig. 2E) (15).

Fig. 2.

β5t selectively reduces the chymotrypsin-like activity of proteasomes. (A) Alignment of amino acid residues organizing the S1 pockets of β5 families. The hydrophobic and hydrophilic amino acid residues are boxed in black and white, respectively. The numbers represent the position from the active-site threonine. (B) Peptidase activities of human embryonic kidney–293 T cells expressing human β5t-Flag (open circles), β5-Flag (black squares), or mock cells (black circles) in the presence (left) and absence (right) of 0.025% SDS. Arrowheads indicate the peak locations of 20S proteasomes and 26S proteasomes, proven by immunoblot analysis of each fraction (bottom left). p, precursor; m, mature; asterisk, nonproteasomal activity. (C) Fraction numbers 25 in (B) were immunoblotted for β5, α6, and Flag. (D) Hanes-Woolf plots for chymotrypsin-like activity of 26S proteasomes (pools of fraction numbers 24 to 26) from mock cells (black circles) and β5t cells (open circles). Suc-LLVY-MCA, succinyl-Leu-Leu-Val-Tyr-4-methylcoumaryl-7-amide; [S], substrate concentration. (E) Cellextracts from (B) were assayed for adenosine triphosphate–dependent protein degradation activity using 35S-labeled ODC. Mean ± SD; n = 3 experiments.

The thymus is responsible for generating a T cell repertoire that specifically recognizes self-peptide–MHC (self-pMHC) complexes and tolerates self-antigens. Thymic stromal cells represent a heterogeneous mixture of cell types and provide a proper microenvironment for developing thymocytes (16, 17). To identify a cell type that expresses β5t, we immunostained sections of mouse thymus for β5t and markers for thymic stromal cells (Fig. 3, A and B). β5t was exclusively expressed in the thymic cortex, which is quite similar to the expression of Ly51, a marker for cortical thymic epithelial cells (cTECs). The distribution of β5t was distinct from that of cells that bind to Ulex europaeus agglutinin I (UEA-I), a marker for medullary TECs; CD11c, a marker for dendritic cells; and MTS15, a marker for fibroblasts (Fig. 3A) (1820). A magnified image of the cortex revealed that the reticular staining pattern of β5t is nearly identical to that of Ly51, indicating that β5t is exclusively expressed in cTECs (Fig. 3B).

Fig. 3.

β5t is specifically expressed in cTECs. (A and B) Cryosections of mouse thymus were immunostained for β5t, together with Ly51, UEA-I, CD11c, or MTS15. (C) Various populations of cells from the thymus were immunoblotted for β5t and α6 (a loading control for proteasomes). (D) Contents of thymoproteasomes in Ly51-positive cells were analyzed by immunoblotting. MEF, murine embryonic fibroblast.

To confirm this conclusion, different cell populations of the thymus were isolated. Immunoblot analysis demonstrated that β5t is specifically expressed in Ly51-positive cells and is not detected in any other stromal cells or in CD45+ cells that are mostly composed of thymocytes (Fig. 3C). A comparison of Ly51-positive cells with murine embryonic fibroblasts that were treated with or without IFN-γ [in which almost all the proteasomes are immuno- and constitutive proteasomes, respectively (21)] revealed that the majority of proteasomes in cTECs were thymoproteasomes containing β5t, together with β1i and β2i. Also, the expression levels of β5 and β5i were markedly down-regulated in cTECs (Fig. 3D).

To clarify the physiological role of β5t, we generated β5t-deficient mice. The β5t-coding sequence was substituted for cDNA encoding the protein Venus to identify β5t-expressing cells (Fig. 4A). The loss of β5t proteins in the β5t–/– thymus and the expression of Venus in the β5t–/– and β5t+/– thymuses were confirmed by immunoblotting (Fig. 4B). An inspection of β5t+/– and β5t–/– mice with fluorescence demonstrated that the expression of Venus (i.e., β5t) was limited to the cortex of the thymus with a reticular pattern similar to that of cTECs (fig. S3, A to C). Cortical and medullary architectures, as well as the size of the thymuses of β5t–/– mice, were indistinguishable from those of β5t+/– mice (fig. S3, B and C). When thymic stromal cells from β5t+/– mice were analyzed by flow cytometry, all Venus-expressing cells were MHC class II I-Ab–positive (Fig. 4C). Furthermore, we examined the relationship between Venus and Ly51 expression in the I-Ab–positive population (Fig. 4D). We identified three distinct populations. Among them, the Venus+Ly51+ population was the largest, constituting 50% of I-Ab–positive cells. There was a smaller population of VenusLy51+ that constituted 11% of I-Ab–positive cells. These results demonstrate that all β5t-expressing cells are I-Ab+Ly51+ cTECs and that this population makes up ∼80% of cTECs.

Fig. 4.

Defect in maturation of CD8+ T cells in β5t-deficient mice. (A) Structure of the β5t genomic locus, targeting vector, and targeted locus (left). KO, knockout; RV, EcoRV; RI, Eco RI; PGK-neo, neomycin resistance cassette; DT-A, diphtheria toxin A. The cDNA encoding Venus replaced the β5t coding sequence. Southern blot analysis of Eco RV–digested genomic DNA from embryonic stem cell clones, using the indicated probe (right), is shown. (B) Immunoblotting of thymic extracts for β5t and Venus. (C) CD45 thymic cells from 4-week-old β5t+/– mice were analyzed for the expression of Venus and I-Ab by flow cytometry. (D) CD45I-Ab+ thymic cells were analyzed for the expression of Venus and Ly51. (E) Dot plot analysis of anti-CD4 and anti-CD8 staining of thymocytes. The percentage of cells in each quadrant is indicated. (F) TCRβ expression on CD8 SP thymocytes. The percentages of TCRβ-positive cells are indicated. (G) Ratios of CD4 SP to CD8 SP thymocytes. n = 5 mice in each group. *P < 0.05; **P <0.01. (H) T cell subpopulation of the thymuses (mean ± SD, n = 5 mice). (I) Dot plot analysis of anti-CD4 and anti-CD8 staining of TCRβ(+) splenocytes. The percentages of CD4+ and CD8+ cells are indicated. (J) Ratios of CD4+ to CD8+ splenocytes. n = 5 mice in each group. **P <0.01.

It has been suggested that cTECs are mainly responsible for positive selection by presenting ligands on MHC molecules to CD4+CD8+ double-positive (DP) thymocytes (2225). We therefore examined how β5t is involved in the function of cTECs. Flow cytometry analysis of thymocytes revealed that β5t deficiency was associated with a markedly low percentage of CD8 single-positive (SP) thymocytes but not with the percentage of CD4 SP thymocytes (Fig. 4E). A large proportion (80.7%) of wild-type (WT) CD8 SP thymocytes expressed high levels of T cell receptor β (TCRβ), resembling positively selected cells (Fig. 4F, left) (26). In contrast, a much smaller proportion (50.2%) of CD8 SP thymocytes expressed high levels of TCRβ in β5t–/– mice (Fig. 4F, right), suggesting that many of the CD8 SP thymocytes of β5t–/– mice are immature thymocytes that are transitory intermediates between double-negative (DN) and DP thymocytes (27). Consequently, the ratio of CD4 SP cells to CD8 SP cells in β5t–/– mice was markedly higher than that in WT mice (Fig. 4G). The ratio was also significantly elevated in β5t+/– mice, although the difference was much smaller (Fig. 4G). Total numbers of DN, DP, and CD4 SP thymocytes in β5t–/– mice were nearly identical to those of β5t+/– and WT mice. However, a significantly low total number of CD8 SP thymocytes was observed in β5t–/– mice (Fig. 4H). Selective reduction of CD8 SP T cells was also observed in TCRβ(+) splenocytes of β5t–/– mice (Fig. 4, I and J). These results demonstrate that β5t is required for the development of CD8+ T cells in the thymus and suggest the possibility that β5t deficiency is associated with defective positive selection of CD8+ T cells.

During positive selection, DP cells that interact with self-pMHC complexes expressed on cTECs with sufficiently low affinity or avidity are rescued from intrathymic death and induced to differentiate into CD4 or CD8 SP thymocytes. The recognition of MHC class I molecules results in commitment to the CD8 lineage (2226). In contrast, DP cells that interact with high affinity or avidity with self-pMHC complexes are eliminated by the induction of apoptosis (2224). To date, however, information about whether and how cTECs can offer specialized signals that are suitable for positive selection has been elusive. The present work demonstrates that β5t, which is specifically expressed in cTECs, plays a pivotal role in the development of CD8+ T cells. What, then, is the mechanism for CD8+ T cell development regulated by β5t? The thymic architecture was apparently normal in β5t–/– mice (fig. S3, B and C), suggesting that β5t is not essential for the differentiation and proliferation of cTECs. Normal development of CD4 SP thymocytes observed in β5t–/– mice supports this argument. Surface expression levels of MHC class I on β5t–/– cTECs were also comparable to those of β5t+/– cTECs (fig. S4). It is assumed that other β5 family members (e.g., β5 and β5i subunits) are incorporated in place of β5t in cTECs of β5t–/– mice. Considering that proteasomes are essential for the production of MHC class I ligands and that β5t specifically attenuates the peptidase activities that cleave peptide bonds after hydrophobic amino acid residues, it is possible that thymoproteasomes predominantly produce low-affinity MHC class I ligands rather than high-affinity ligands in cTECs, as compared with constitutive- and immunoproteasomes, thereby supporting positive selection.

Supporting Online Material

Materials and Methods

Figs. S1 to S4


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article