Research ArticleTHYMIC SELECTION

Human thymoproteasome variations influence CD8 T cell selection

See allHide authors and affiliations

Science Immunology  02 Jun 2017:
Vol. 2, Issue 12, eaan5165
DOI: 10.1126/sciimmunol.aan5165

Proteasome polymorphisms regulate T cell selection

Activation of CD8 T cells relies on proteasome-processed peptide loading on to class I MHC; this is regulated by a thymus-specific proteasome subunit, β5t, in cortical thymic epithelial cells (cTECs) which promote positive selection of CD8 T cells. Here, Nitta et al. studied the effect of three human variants of β5t by engineering mice that express these isoforms. Expression of these variants resulted in a dose-dependent impairment in positive selection of CD8 T cells, suggesting that all three isoforms are loss-of-function mutations. They link one of the variants, β5t G49S to a higher risk of Sjögren’s syndrome. The studies suggest that by influencing T cell selection, genetic variations in proteasome genes can contribute to disease susceptibility.

Abstract

The proteasome is a multi-subunit protease complex essential for housekeeping protein degradation and the production of the major histocompatibility complex (MHC) class I-bound antigen peptides that are essential for recognition by CD8 T cells. MHC variations dramatically contribute to T cell selection and autoimmunity, but genetic variations of peptide processing machinery including proteasome genes have been poorly explored in this context. In the computational analysis of human proteasome gene variation, we documented that PSMB11 was highly enriched for nucleotide changes that interfere with protein function. This gene encodes β5t, a thymus-specific catalytic subunit that regulates positive selection of CD8 T cells by producing a distinct set of MHC class I-bound peptides. The introduction of PSMB11 variations into the mouse genome by genome-editing revealed that these variations impaired the development of CD8 T cells in vivo. One of the PSMB11 polymorphisms altered the CD8 T cell repertoire in mice and was associated with a higher risk of an autoimmune disease in humans. Our findings suggest that, in addition to the MHC haplotype, proteasome variations influence T cell repertoire selection and may contribute to the difference in individual susceptibility to autoimmunity.

INTRODUCTION

The proteasome is a multi-subunit protease complex essential for housekeeping protein degradation and the production of the MHC class I-bound antigen peptides that are essential for T cell receptor (TCR) recognition by CD8 T cells (1, 2). The core proteasome complex, termed the 20S proteasome, is composed of two sets of seven different α-type (α1-α7) and β-type (β1-β7) subunits. Under certain conditions, the catalytic subunits β1, β2, and β5 are substituted by the alternative subunits β1i, β2i, and β5i, respectively. β5t represents the thymus-specific subunit which replaces β5. In total, eighteen different 20S proteasome subunit genes exist in the mammalian genome (2).

β5, β1i, β5i and β5t are catalytic subunits with chymotrypsin-like activity that cleaves the C-terminus of hydrophobic residues to produce peptides with a C-terminal hydrophobic anchor residue, which is required for the association with MHC class I proteins (3). β1i, β5i, and β2i are expressed in immune cells and interferon γ-stimulated cells so as to form the immunoproteasome, an alternative type of proteasome that confers increased chymotrypsin-like activity and MHC class I-bound peptide production (2, 4). Another atypical type of proteasome, the thymoproteasome, contains β5t, together with β1i and β2i (5). β5t is a catalytic subunit characterized by lower chymotrypsin-like activity (5) and is expressed exclusively in the thymus, in particular by cortical thymic epithelial cells (cTECs) (57) that regulate development and repertoire selection of T cells (8). The β5t-containing thymoproteasome is required for positive selection of CD8 T cells in the thymus (5, 911), through the production of a spectrum of MHC class I-bound peptides in cTECs (12, 13). Alternatively, the regulation of CD8 T cell development by β5t might be explained by a ‘peptide-switching model’ in which CD8 T cells need to be selected on different sets of MHC class I-bound peptides at positive selection and subsequent negative selection (11).

Genetic variations of proteasome subunits are associated with several diseases such as autoinflammation syndrome (1418). However, the relative contribution of proteasome genetic variations in antigen peptide production and T cell immunity has not been addressed. In this study, we show by a reverse-translational approach that human thymoproteasome genetic variations influence CD8 T cell selection. Our findings provide a paradigm that, in addition to the MHC haplotypes, the proteasome genetic variations affect T cell repertoire selection and modulate T cell immune responses, which may account for the differences in individual susceptibility to diseases.

RESULTS

The PSMB11 gene is highly enriched for variations predicted to alter protein function

Public human genome sequence databases were examined to explore the distribution of proteasome gene variation focusing on 18 genes (PSMA1-7 and PSMB1-11) that comprise the core proteasome complex. To determine whether a ‘non-synonymous’ (mutations that result in amino acid changes) variation might affect protein function, we used the Polymorphism phenotyping version 2 (PolyPhen-2) algorithm that predicts the functional impact of each missense variation (19). In this study, missense variations proposed to alter protein function (PolyPhen-2 score ≥0.95), as well as nonsense, frameshift and insertion/deletion variations, were classified as ‘damaging’ variations, whereas missense variations proposed to have no or less substantial impact on protein function (PolyPhen-2 score <0.95) were classified as ‘benign’ variations.

We performed analyses of the genetic variations from the United States (Exome database) and Japan (HGVD database) and calculated the sum of the frequencies of benign or damaging variations per gene (Fig. 1A and tables S1 and S2). In both populations, the frequency of damaging variations was evidently high in the three genes PSMB8, PSMB9, and PSMB11, respectively encoding the β5i, β1i, and β5t subunits. We found PSMB11 to be the gene most enriched for damaging variations (Fig. 1B).

Fig. 1 High-frequency damaging variations of PSMB11 alter its proteasome activity.

(A) Summary of the proteasome gene variations. Data were obtained from the Exome and HGVD databases. Eighteen genes encoding core proteasome subunits and the substrate specificities of catalytic activities (chymotrypsin-like (Chym), caspase-like (Casp), and trypsin-like (Tryp)) are listed. The graphs show the sum frequency (% per gene) of the benign as well as damaging variations. (B) The ratio of damaging to benign variations in all 18 proteasome genes, PSMB8, PSMB9, and PSMB11. (C and D) Frequencies of benign or damaging PSMB11 variations obtained from the Exome (C) or HGVD (D) database. The structure of the PSMB11/β5t protein and the positions of the variations are shown. Pro, pro-peptide. T50, active site threonine at 50. (E) HEK293T cells were transfected with plasmids encoding WT or the variant β5t, in addition to thymoproteasome subunits β1i and β2i. Proteins were immunoprecipitated with anti-FLAG antibodies and detected by immunoblotting. p and m indicate the precursor and mature form of β5t. The asterisk indicates an aberrantly cleaved form of the β5tG49S protein. The entire blots are shown in fig. S7A. (F) Chymotrypsin-like peptidase activities of the immunoprecipitates used in E (n = 3-6, mean ± SEM). *P < 0.05. NS, not significant. ND, not detected. Shown data are representative of three independent experiments (E, F).

The β5i, β1i, and β5t subunits all possess the chymotrypsin-like catalytic activity which is essential for producing MHC class I-bound peptides (3). β5t is expressed exclusively by cTECs to form the thymoproteasome (57). The β5t-containing thymoproteasome, characterized by lower chymotrypsin-like activity, produces a spectrum of MHC class I-bound peptides in cTECs (12), which enables efficient positive selection of CD8 T cells (5, 911). Given the database findings, we sought to examine the impact of β5t variation on thymoproteasome function.

PSMB11 variations influence proteasome activity

In total 46 non-synonymous variations of PSMB11/β5t were found in the Exome database (table S3). We focused on nine highly frequent PSMB11/β5t variations (defined as >0.03%), including one frameshift variation (S80Hfs*36), four damaging missense variations (G49S, R193C, A208T, and R238H) (PolyPhen-2 score ≥0.95), and four benign missense variations (R115Q, S165N, R169C, and A203T) (PolyPhen-2 score <0.95) (Fig. 1C). Gly49 is located at the C-terminus of the pro-peptide of β5t that is cleaved in order to yield the mature protein during the assembly into the thymoproteasome. The other eight residues are located within the mature protein of β5t. Only G49S was found to be shared by the American and Japanese populations (Fig. 1D).

Wild-type (WT) or nine PSMB11 variants from the Exome database were introduced into HEK293T cells for the reconstitution of the thymoproteasome. Immunoblot analysis demonstrated that WT β5t was proteolytically processed from a 33 kDa precursor to the 28 kDa mature protein as a result of the cleavage of the N-terminal pro-peptide (Fig. 1E). The β5tG49S protein (33 kDa) was processed to an abnormally sized mature protein (30 kDa, indicated with an asterisk in Fig. 1E), demonstrating aberrant cleavage of the pro-peptide. The other variants, except for β5tS80Hfs, which had a premature stop codon at 116, generated normally sized precursor and mature proteins. The WT and all of the missense variant proteins tested were equally co-immunoprecipitated with the subunits α6, β1i, and β2i, indicating that these variant proteins are incorporated into the proteasome complex.

We determined the chymotrypsin-like activity of the variant β5t-containing thymoproteasomes that were immunoprecipitated from the transfected cells. Interestingly, the damaging missense variants, G49S, R193C, A208T and R238H exhibited significantly higher chymotrypsin-like activity than the WT, while the benign variants, R115Q, S165N, R169C and A203T did not (Fig. 1F). These results show that the damaging variations affect the proteolytic activity of the β5t-containing thymoproteasome, suggesting that these variations may influence CD8 T cell selection in the thymus.

Altered MHC class I/peptide complexes displayed on cTECs in Psmb11 mutant mice

Because the nucleotides encoding the Gly49, Ser80 and Ala208 residues were conserved between human and mouse (fig. S1A), we introduced the G49S, S80Hfs or A208T mutation into the mouse Psmb11 gene using the CRISPR/Cas9 genome-editing technique (20) (fig. S1, B and C). Homozygous mutant mice (Psmb11G49S/G49S, Psmb11S80Hfs/S80Hfs and Psmb11A208T/A208T) all exhibited a normal body size with intact fertility. These mutant mice also displayed unaltered thymus organization and normal development of cTECs and medullary TECs (mTECs) (Fig. 2A and fig. S2), indicating that these variations had no inhibitory effects on TEC development, unlike the previously reported mouse dominant-negative Psmb11 mutation (21).

Fig. 2 Normal thymus development but altered MHC class I/peptide complexes displayed on cTECs in Psmb11 mutant mice.

(A) Representative photographs of thymi from 5-week-old WT, Psmb11G49S/G49S, Psmb11A208T/A208T, or Psmb11S80Hfs/S80Hfs mice (top). The total numbers of thymocytes from 4 to 8-week-old mice (n = 4-7) of the indicated genotype are shown (bottom). (B) Total thymus lysates were immunoprecipitated with an anti-β5t antibody followed by immunoblotting for β5t, α6, β1i, β2i, β5i, and β5. p and m indicate the precursor and mature forms of β5t. The asterisk indicates the aberrantly cleaved form of β5tG49S. Mouse fibroblast cells were used as positive control (PC) for β5. The entire blots are shown in fig. S7B. (C) Chymotrypsin-like peptidase activities of anti-β5t immunoprecipitates in B (n = 3-5) in the absence or presence of the proteasome inhibitor MG132 (10 μM). ND, not detected. (D) Surface expression of 25-D1.16 (H-2Kb associated with a population of peptides) and AF6-88.5 (H-2Kb irrespective of associated peptides) antibody epitopes on cTECs from WT (black lines) or Psmb11G49S/G49S (red lines) mice (left). The numbers indicate the mean fluorescence intensity (MFI). The graphs show the MFI in the WT or the indicated homozygous mutant mice (n = 4-8, mean ± SEM). *P < 0.05. **P < 0.01. NS, not significant. Data shown are representative of at least two (A) or three (B, C, D) independent experiments.

Immunoprecipitation of thymus lysate with an anti-β5t antibody revealed that β5tG49S was aberrantly cleaved to generate a 30 kDa mature protein. β5tA208T was processed to the normal size but was reduced in expression, whereas the β5t protein was not produced from the S80Hfs variant (Fig. 2B). These results were confirmed by immunoblotting analysis of purified cTECs (fig. S3). The immunoprecipitated β5tG49S and β5tA208T proteins exhibited a markedly higher chymotrypsin-like activity (Fig. 2C), in agreement with the results in the human variants. The β5tG49S and β5tA208T proteins were co-immunoprecipitated with a small amount of β5i and β5 proteins (Fig. 2B), suggesting that the G49S and A208T variations lead to the formation of ‘hybrid’ thymoproteasomes containing β5i or β5 in addition to β5t, which might account for the aberrant chymotrypsin-like activity of these two variants.

To examine the effects of these variations on peptide processing in cTECs, we used the 25-D1.16 ‘TCR-like’ antibody. This antibody recognizes mouse MHC class I H2-Kb associated with a population of endogenous peptides via the mechanism similar to that by TCR (22, 23). It was previously reported that staining intensity with the 25-D1.16 antibody is significantly different between cTECs from β5t-deficient (Psmb11−/−) mice and those from control mice, which reflects the production of a particular repertoire of MHC class I-bound peptides by β5t-containing thymoproteasomes in cTECs (9). The cTECs from Psmb11G49S/G49S, Psmb11A208T/A208T and Psmb11S80Hfs/S80Hfs mice displayed significantly increased 25-D1.16 staining intensities (149%, 175%, and 191%, respectively Fig. 2D, top panel) that were similar to those in previously described Psmb11−/− mice (approximately 180%), whereas the expression of MHC class I was unaffected in these mice irrespective of peptide binding (Fig. 2D, bottom panel). Collectively, these damaging variations clearly affected the chymotrypsin-like activity of the thymoproteasome and the production of the MHC class I-bound peptide repertoire in cTECs.

Impaired CD8 T cell development in Psmb11 mutant mice

The number of CD8 single positive (SP) thymocytes was significantly reduced to 35.3% in Psmb11G49S/G49S, 51.4% in Psmb11A208T/A208T and 16.1% in Psmb11S80Hfs/S80Hfs mice compared with littermate WT mice (Fig. 3, A and B). Heterozygous mutant mice showed only a mild and nonsignificant reduction of CD8SP thymocytes (87.9% in G49S, 89.3% in A208T and 84.1% in S80Hfs), and there were no marked changes in the CD4SP thymocyte number in any of the mutant mice. Psmb11G49S/A208T compound heterozygous mice also exhibited a significant reduction in CD8SP thymocytes (fig. S4). As the mutant mice displayed phenotypes essentially identical to those of β5t-deficient mice (5, 911), it is likely that these damaging PSMB11 variations are loss-of-function alterations.

Fig. 3 Impaired CD8 T cell development in Psmb11 mutant mice.

(A) Flow cytometry profiles for CD4 and CD8 in the total thymocytes (top) and TCRβ of CD4CD8+ (CD8SP) thymocytes (bottom) from 5 to 6-week-old WT, Psmb11G49S/G49S, Psmb11A208T/A208T and Psmb11S80Hfs/S80Hfs mice. (B) The numbers of thymus CD4SP TCRβ+ and CD8SP TCRβ+ cells from the indicated mice (bottom) (n = 4-7, mean ± SEM). For each mutant line, the littermate WT, heterozygous, and homozygous mice were analyzed. *P < 0.05. **P < 0.01. NS, not significant. Data represent at least two independent experiments.

The number of peripheral CD8 T cells was also decreased in Psmb11G49S/G49S, Psmb11A208T/A208T and Psmb11S80Hfs/S80Hfs mice (Fig. 4, A and B). We examined naïve and effector memory subsets of peripheral CD8 T cells by staining for CD44 and CD62L. Psmb11G49S/G49S, Psmb11A208T/A208T and Psmb11S80Hfs/S80Hfs mice showed a reduced number of CD44loCD62Lhi naïve CD8 T cells and an increased proportion of CD44hiCD62Llo effector memory population of CD8 T cells (Fig. 4, C and D).

Fig. 4 Reduced CD8 T cells in the spleen of Psmb11 mutant mice.

(A) Flow cytometry profiles for CD4 and CD8 in the total splenocytes from 5 to 6-week-old WT or homozygous mutant mice. (B) The numbers of spleen CD4+ TCRβ+ and CD8+ TCRβ+ cells from the indicated mice (n = 4-7, mean ± SEM). For each mutant line, littermate WT, heterozygous, and homozygous mice were analyzed. (C) Flow cytometry profiles for CD44 and CD62L of spleen CD8+ TCRβ+ cells from WT or homozygous mutant mice. (D) The frequency (% of CD8+ TCRβ+ cells) and numbers of spleen CD8+ TCRβ+ CD44lo CD62Lhi and CD8+ TCRβ+ CD44hi CD62Llo cells from WT or homozygous mutant mice (n = 4, mean ± SEM). *P < 0.05. **P < 0.01. NS, not significant. Data shown are representative of at least two (A, B) or three (C, D) independent experiments.

The G49S variation reduces the size and diversity of CD8 TCR repertoire

We focused on the human G49S variation (registered as SNP rs34457782, major allele G and minor allele A), because this is a common genetic variant in human populations. The allele frequency is strikingly high in the Japanese population, attaining approximately 3% in both the HGVD and NCBI databases, and thus we explored its impact on immunity and disease susceptibility. First we examined the structural impact of the G49S variation. Mass-spectrometry analysis revealed that this variation leads to aberrant cleavage within the pro-peptide and then prevents the exposure of active site Thr50 at the N-terminus (fig. S5). This may account for the altered chymotrypsin-like activity and MHC class I-bound peptides in cTECs.

When crossed with MHC class I-restricted TCR-transgenic mouse lines, HY and OT-I, Psmb11G49S/G49S mice exhibited a reduced frequency of post-selected thymocytes, indicating impaired positive selection (Fig. 5, A and B). To examine the TCR repertoire in polyclonal CD8 T cells, we analyzed the usages of a set of TCR-Vβ chains, which were previously reported to alter in Psmb11-deficient mice (9). Psmb11G49S/G49S mice displayed significantly changed frequency of certain TCR-Vβ chains in CD8 T cells compared with littermate Psmb11WT/WT mice, indicating the alteration of TCR repertoire in polyclonal CD8 T cells (fig. S6A). Similar alterations in the TCR-Vβ usage were observed in Psmb11A208T/A208T mice (fig. S6B). These results suggest that the G49S as well as A208T variations affect CD8 T cell repertoire via positive selection in the thymus.

Fig. 5 Impaired positive selection of CD8 T cells in Psmb11G49S/G49S mice.

Flow cytometry profiles of the total and gated thymocytes from WT or Psmb11G49S/G49S Rag2-deficient female HY-TCR (A) or OT-I-TCR transgenic (B) mice (top). The graphs show the frequency of TCR+CD69+ post-selected DP cells and mature CD8SP TCR+ cells (bottom) (HY WT, n = 2; HY G49S, n = 4; OT-I WT, n = 3; OT-I G49S, n = 6). Each circle represents an individual mouse and horizontal bars indicate the mean (A). Bar graphs indicate mean ± SEM (B). *P < 0.05. **P < 0.01. NS, not significant. Data shown are representative of two (A) or three (B) independent experiments.

We further employed a mouse model that expresses a fixed TCRβ chain in T cells so deep-sequencing of variable TCRα chains would allow for high-throughput quantitative analysis of the TCR repertoire at the individual TCR level. Hematopoietic progenitor cells were retrovirally transduced with a TCRβ chain (Vβ5) and reconstituted in irradiated WT or Psmb11G49S/G49S mice. The retrogenic Psmb11G49S/G49S mice exhibited reduced development of CD8SP thymocytes compared with WT counterparts (Fig. 6A). We sorted Vβ5+ CD8SP thymocytes from four individual retrogenic mice per group and deep-sequenced their TCRα chains (table S4). Principal component analysis (PCA) revealed that the usage of TCR V-J chains was not significantly different between the two groups (Fig. 6B), while correlation analysis showed that the TCR repertoire in the G49S retrogenic group was less correlative (more diverse) among individual mice (Fig. 6C). To statistically compare the frequencies of individual TCR clones between the WT and G49S groups, the sequenced TCR reads were filtered to include the ones present in more than three fourths of the mice from either group and present in more than 0.1% in either group (Fig. 6D). Out of the filtered 108 TCRs, 18 TCRs (16.7%) were significantly more highly detected in WT mice than in G49S mice, and in particular, 6 TCRs were detected solely in WT mice, whereas no TCRs were significantly more highly detected in G49S mice compared with WT mice. In order to confirm the validity of our results, we cloned two representative TCRα chains, TCR-1 and TCR-5, a common and WT-specific TCR, respectively (table S5), and generated retrogenic mice expressing these TCRα chains paired with the partner TCRβ chain. CD8SP thymocytes with TCR-1 evenly developed in both the WT and G49S mice, whereas the development of TCR-5-expressing CD8SP cells was significantly reduced in the G49S mice (Fig. 6E). These data collectively show that the size and diversity of the CD8 T cell TCR repertoire are reduced in the G49S mice with a specific repertoire loss and no additional repertoire diversity generated.

Fig. 6 The G49S variation alters the TCR repertoire of CD8 T cells.

(A) Scheme of the generation of TCRβ-retrogenic mice (top). Flow cytometry profiles of CD4 and CD8 expression in the GFP+ Vβ5+ cells are shown (n = 4). GFP+ Vβ5+ CD8SP cells (highlighted with red squares) were sorted for TCRα sequencing analysis. (B) Principal component analysis of TCR V-J combinations. The blue and red dots indicate 4 individual WT and Psmb11G49S/G49S mice, respectively. (C) Heat map shows the correlation index of TCRα V-J combinations between the indicated pair of mice (left). Bar graphs show the mean ± SEM of the correlation index among the WT group or the G49S group, or between the WT and G49S groups. (D) Frequencies of filtered TCR reads in the WT and Psmb11G49S/G49S mice. The dots indicate unique TCR reads. The red dots indicate TCR reads significantly (P < 0.05) higher in the WT than in G49S. Two representative TCRs are shown. (E) Flow cytometry profiles of CD4 and CD8 expression in the GFP+ Vβ5+ thymocytes (left) and the frequency of CD8SP cells (right) from TCR-1 or TCR-5 retrogenic WT or Psmb11G49S/G49S mice (n = 6, mean ± SEM). **P < 0.01. NS, not significant. Cumulative data from three independent sequencing experiments (A-D) or four independent retrogenic experiments (E).

The PSMB11 G49S variation is associated with a higher risk of Sjögren’s syndrome

The above-mentioned data together with the findings in the mutant mice suggest that humans homozygous for the G49S variant might also have a reduced repertoire of CD8 T cells and may have an increased propensity to develop certain immune-related diseases. Therefore, we examined the allele and genotype frequency of the rs34457782 in control subjects and patients affected with autoimmune diseases (Table 1). The frequency of the rs34457782-A allele (G49S) in the control group was 2.9%, which is consistent with the data from the public databases. We obtained no significant association between the allele frequency of rs34457782-A and autoimmune diseases (P > 0.062). Because the G49S allele had recessive effects on CD8 T cell development in mice, we used a recessive model in which the frequency of the AA genotype was compared with that of other genotypes (GA + GG). Among the autoimmune diseases tested, rs34457782 was significantly associated with Sjögren’s syndrome in the recessive model (AA vs. GG + GA, P = 0.00089, odds ratio (OR) = 7.15, 95% CI 1.84-27.7), indicating that the A allele is recessive and that homozygosity for the G49S variant (AA genotype) of PSMB11 confers a risk for Sjögren’s syndrome.

Table 1 Association analysis of rs34457782 with autoimmune diseases.
View this table:

DISCUSSION

Evolving genome-editing techniques allow the reverse-translational evaluation of the functional contribution of individual human genetic variants in model animals. Our results indicate that homozygosity or compound heterozygosity for the damaging variations, G49S, A208T, and S80Hfs in mice engineered to express these variants, resulted in reduced positive selection and imbalanced naïve and effector memory subsets of CD8 T cells, suggesting that these damaging variations are loss-of-function alterations. Homozygosity of these damaging variations was found in the public human genome databases and control subjects in our genome-wide association study (GWAS). Further, our GWAS analysis revealed that homozygosity for the G49S variation (rs34457782-A) is significantly associated with a higher risk of Sjögren’s syndrome, although the mechanistic link between them remains elusive and needed to be studied. Previous studies suggested that CD8 T cell subsets play either disease-provoking (2426) or protecting (27) roles in Sjögren’s syndrome. It is worth examining whether the G49S mutant mice that have a reduction of CD8 T cell number and repertoire diversity exhibit any disease pathology in the steady state or when subjected to various experimental models of Sjögren’s syndrome.

We report that proteasome genetic variations might significantly influence CD8 T cell immunity through antigen peptide production. Given the relative high frequency of proteasome variants in human populations, it is conceivable that such damaging proteasome variations may confer unknown selective advantages in certain endemic or geographic situations; such as possible avoidance of fatal infections or autoimmunity by an altered CD8 T cell repertoire and antigen responsiveness. Psmb11 mutant strains of mice generated here will be a valuable resource to test whether Psmb11 variations confer protection against certain pathogens.

MATERIALS AND METHODS

Study design

This is a reverse translational study that goes from high-throughput analysis of human genome databases toward functional evaluation of the individual genetic variation using model animals. First we collected the information of genetic variations of proteasome genes from public databases and filtered the variations using the variation effect prediction algorithm. We found that the PSMB11 gene was enriched for damaging variations that influence proteasome activity. The physiological significance of the damaging variations of PSMB11 was evaluated by introducing them into the Psmb11 gene in mice by genome-editing. Finally, to study the possible impact of the PSMB11 variation on disease susceptibility, we performed association analysis using control subjects and autoimmune patients. All data are reported. No randomization or exclusion of samples was used. Investigators were not blinded to the sample identity.

Human genome databases

The information on the human genetic variants was obtained from the Exome Variant Server (EVS) of the National Heart, Lung, and Blood Institute (NHLBI) Exome Sequencing Project (http://evs.gs.washington.edu/EVS/), the Human Genetic Variation Database (http://www.hgvd.genome.med.kyoto-u.ac.jp/index.html) and the National Center for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov/). We collected data on the amino acid changes and allele frequencies from the Exome and HGVD databases, and filtered the variations with the PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/) (19) algorithm that predicts the functional impact of each missense variation. We followed the guidelines for investigating causality of sequence variants in human disease (28) to categorize genetic variations.

Mice

C57BL/6N mice were purchased from SLC Japan (Shizuoka, Japan). OT-I-TCR-transgenic mice (29), HY-TCR-transgenic mice (30), Rag2−/− mice (31), and Tcra+/− mice (32) were described previously. All mice were bred and maintained under specific pathogen-free conditions in our animal facility. Animal experiments were approved by the Institutional Review Board at the University of Tokyo and Animal Care and Use Committee of the NCGM Research Institute, and conducted in accordance with institutional procedures.

Antibodies

Monoclonal antibodies specific for CD4 (GK1.5), CD8α (53-6.7), CD44 (IM7), CD69 (H1.2F3), EpCAM (G8.8), Ly51 (6C3), TCRβ (H57-597), and TCR-Vβ5 (MR9-4) were purchased from BioLegend. The mouse TCR-Vβ screening panel and anti-TCR-Vα2 (B20.1) monoclonal antibody were purchased from BD Pharmingen. The anti-FLAG (M2) monoclonal antibody was purchased from Sigma-Aldrich. UEA1 was purchased from Vector Laboratories. Polyclonal antibodies for proteasome subunits were described previously (5, 33).

Cell culture and transfection

HEK293T cells were cultured in DMEM supplemented with 10% fetal calf serum, 100 U/ml of penicillin G and 100 μg/ml of streptomycin, and transfected with plasmid DNA using FuGENE HD (Promega).

Immunoprecipitation and proteasome activity assay

Cells were lysed with lysis buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM MgCl2, 0.5% NP-40, and protease inhibitor cocktail (Sigma)). FLAG-tagged proteins were immunoprecipitated with an anti-FLAG antibody (M2) conjugated with agarose beads (Sigma). For mouse β5t, whole thymus lysates were immunoprecipitated with an anti-β5t polyclonal antibody (5) and Protein G conjugated with sepharose beads (GE Healthcare). The chymotrypsin-like activity of the proteasome was measured using a Proteasome Enrichment and Activity Assay kit (CycLex). Immunoprecipitates were suspended in Assay Buffer and incubated with Suc-LLVY-MCA at 37°C for 50 min. Fluorescence intensity (380/460 nm) was measured using a FlexStation 3 plate reader (Molecular Devices). The addition of the proteasome inhibitor MG132 (10 μM) completely abolished the activities, indicating that the value is specific to proteasome-mediated proteolysis.

Flow cytometry and cell sorting

Flow cytometry analysis and cell sorting were performed with FACSCantoII and FACSAriaIII (BD Bioscience). Thymic stromal cells were prepared by digesting thymic fragments with Liberase TM (Roche) and DNase I (Roche), as described (34).

Histological analysis

Frozen tissues embedded in OCT compound (Sakura Finetek) were sliced into 5 μm-thick sections with a Cryostat (Leica), air-dried, fixed with acetone, and stained with specific antibodies, as described previously (35). Multi-color images were obtained by a BZ-9000 fluorescent microscope (Keyence).

CRISPR/Cas9-mediated genome editing in mice

Preparation of single-guide (sg) RNA and Cas9 mRNA was described previously (21). sgRNA, Cas9 mRNA, and oligonucleotide containing specific mutation were injected into the cytoplasm of the pronuclear stage eggs obtained from C57BL/6N, and then the eggs were transferred into the oviducts of pseudopregnant ICR female mice. The target and oligonucleotide sequences are shown in fig. S1.

Determination of the N-terminus of the β5tG49S protein

The plasmids expressing human β1i, β2i and C-terminally FLAG-tagged β5tWT or β5tG49S were transfected into HEK293T cells, and then stably transfected cells were screened by Western blotting. Cells (3 × 107) were lysed and the proteasome subunits were disassembled with 1% SDS at 95°C for 5 min. FLAG-tagged β5t proteins were immunoprecipitated with an anti-FLAG antibody, eluted with FLAG peptides and separated by 15% SDS-PAGE. Protein bands were visualized by Coomassie brilliant blue staining and excised from the gel. Proteins were eluted from the excised gels and analyzed using a Triple-TOF 5600 Mass Spectrometer and ProteinPilot software (AB SCIEX).

TCRβ retrogenic mice

A cDNA fragment encoding mouse TCRβ (Vβ5.2-Dβ2-Jβ2.6-Cβ2) was PCR-cloned from OT-I-TCR transgenic mice and inserted into the retrovirus vector pMSCV-IRES-EGFP (36). Retroviral plasmids encoding TCRα-P2A-TCRβ were constructed as previously described (37). Retroviral supernatant was prepared by transient transfection of the Plat-E packaging cells with a retroviral plasmid (38). To generate TCRβ retrogenic mice, bone marrow cells were harvested from donor Tcra+/− mice 4 days after intravenous administration of 5-fluorouracil (150 mg/kg). Sca1+ cells were magnetically isolated and cultured in growth medium (IMDM containing 20% FCS, L-glutamine, sodium pyruvate, nonessential amino acids, penicillin, streptomycin, 50 ng/ml SCF, 50 ng/ml IL-6, and 10 ng/ml IL-3). Two days later, the culture medium was replaced with retroviral supernatant containing 10 μg/ml polybrene, and the culture plates were centrifuged at 1000 × g for 90 min at 30°C. Cells were replenished with the growth medium, cultured, and additionally infected on days 3 and 4 post-harvest. After the retroviral infection on day 4 post-harvest, whole cultured cells were intravenously injected into sub-lethally irradiated (7.0 Gy) recipient mice. The mice were analyzed 5 weeks after the transplantation. The frequency of GFP+ cells in total thymocytes from reconstituted mice ranged from 84% to 97%, indicating efficient chimerism and retroviral transduction.

TCR sequencing

GFP+ TCR-Vβ5+ CD8SP thymocytes were purified from retrogenic mice and total RNA was extracted with ISOGEN (Nippon Gene, Tokyo, Japan). Next-generation sequencing was performed with an unbiased TCR repertoire analysis technology developed by Repertoire Genesis Inc. An unbiased adaptor-ligation PCR was performed as described previously (39). In brief, double-stranded cDNA was synthesized with Superscript III reverse transcriptase (Invitrogen), ligated with a 5’ adaptor oligonucleotide, and then PCR-amplified with primers specific for the adaptor and TCRα constant region. After the amplification of TCRα cDNA, index (barcode) sequences were added using a Nextera XT index kit v2 setA (Illumina). Sequencing was performed with the Illumina Miseq paired-end platform (2 × 300 bp). Data processing was performed using the Repertoire Analysis software originally developed by Repertoire Genesis, Inc. TCR sequences were assigned using a data set of reference sequences from the international ImMunoGeneTics information system (IMGT) database (http://www.imgt.org). Nucleotide sequences of the CDR3 region ranging from a conserved cysteine at position 104 (Cys104) to a conserved phenylalanine at position 118 (Phe118) along with the following glycine (Gly119) were translated to amino acid sequences. A unique sequence read (USR) was defined as a sequence read having no identity with the other sequence reads. The copy number of an identical USR was automatically counted by the RG software.

Association analysis of rs34457782 with autoimmune diseases

Association analysis data were obtained from our ongoing GWAS for autoimmune diseases. We enrolled 576 cases affected with idiopathic inflammatory myopathies (including 236 PM and 340 DM cases) and 626 cases affected with Sjögren’s syndrome (including 378 primary Sjögren’s syndrome cases) from 18 medical institutes in Japan under the support of the autoimmune disease study group of Research in Intractable Diseases, Japanese Ministry of Health, Labor and Welfare. We used the data on 4,023 rheumatoid arthritis cases and the 6,269 controls who were enrolled through the Biobank project (40). All the cases fulfilled the criteria for each disease (4143). All of the control individuals were free of autoimmune disease. All of the subjects were of Japanese origin and provided written informed consent for participation in the study as approved by the ethical committee of the institutional review board. We genotyped case samples using the Illumina HumanOmniExpressExome BeadChip, and control samples using the Illumina HumanOmniExpress BeadChip and Illumina HumanExome BeadChip. Quality control filtering for the SNPs and samples was performed as previously described (44).

Statistical analysis

Statistical significance was calculated with an unpaired two-tailed Student’s t-test using GraphPad Prism (GraphPad Software) for all experiments. When indicated, Permutational Multivariate Analysis of Variance (PermANOVA) was performed using the adonis function within the R (vegan) software. Association analysis was performed with a χ2 test using 2 × 2 contingency tables. Statistical significance was determined as P < 0.05 for all experiments.

Correction: We have corrected the explanation for "non-synonymous mutation" and the P value of the association between rs34457782-A and autoimmune disease online and in the PDF. The results and conclusions have not changed.

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/2/12/eaan5165/DC1

Fig. S1. Generation of Psmb11 mutant mice by the CRISPR/Cas9 method.

Fig. S2. Normal development of TECs in Psmb11 mutant mice.

Fig. S3. Detection of the β5t protein in cTECs from Psmb11 mutant mice.

Fig. S4. Reduced CD8 T cell development in Psmb11G49S/A208T heterozygous mice.

Fig. S5. Identification of the N-terminus of the β5tG49S protein.

Fig. S6. Altered TCR-Vα/Vβ usage in Psmb11G49S mice.

Fig. S7. The entire immunoblots for each protein shown in Figs. 1E (A) and 2B (B).

Table S1. Genetic variants of the genes encoding the core proteasome subunits from the United States.

Table S2. Genetic variants of the genes encoding the core proteasome subunits from Japan.

Table S3. Allele frequency and genotype count of PSMB11 variations obtained from the Exome database.

Table S4. Summary of TCR repertoire sequencing.

Table S5. Two representative TCRs obtained from repertoire sequencing analysis.

Table S6. Raw data.

REFERENCES AND NOTES

Acknowledgments: We are grateful to Drs. T. Kobayashi, K. Kawahata, N. Kimura, N. Umezawa, K. Ishigaki, M. Ota, K. Fujio, and H. Kohsaka for insightful discussion and valuable technical assistance. We also thank S. Nitta, Y. Nakayama, T. Suda, Y. Seki, R. Yanobu-Takanashi, and K. Nakano for technical assistance. Funding: This study was supported by Grants-in-Aid for Research from the Japan Society for Promotion of Science (JSPS) (KAKENHI 15H05703, 16H05202, 16K14648, and 17H05788) and from the National Center for Global Health and Medicine (24A-112 and 26A-105), the Naito Foundation, Japan Foundation for Applied Enzymology, and Takeda Science Foundation. Author contributions: T.N., Y.K., and H.T. conceived the ideas and designed the experiments. T.N., Y.K., R.M., and Y.T. performed the experiments. T.O. generated genetically modified mice. S.M. and H.S. provided advice on project design and data interpretation. Y.K., T.S., and K.Y. contributed with data analysis of the human samples. T.N. and Y.K. carried out data and statistical analyses. T.N., Y.K., and H.T. wrote the manuscript. H.T. supervised the project. Competing interests: The authors declare that they have no competing interests.

Stay Connected to Science Immunology

Navigate This Article