Major source of antigenic peptides for the MHC class I pathway is produced during the pioneer round of mRNA translation

Abstract

The MHC class I antigen presentation pathway allows the immune system to distinguish between self and nonself. Despite extensive research on the processing of antigenic peptides, little is known about their origin. Here, we show that mRNAs carrying premature stop codons that prevent the production of full-length proteins via the nonsense-mediated decay pathway still produce a majority of peptide substrates for the MHC class I pathway by a noncanonical mRNA translation process. Blocking the interaction of the translation initiation factor eIF4E with the cap structure suppresses the synthesis of full-length proteins but has only a limited effect on the production of antigenic peptides. These results reveal an essential cell biological function for a class of translation products derived during the pioneer round of mRNA translation and will have important implications for understanding how the immune system detects cells harboring pathogens and generates tolerance.

Presentation of endogenous peptides on MHC class I molecules serves to allow the immune system to deplete T cells that react against self-antigens and to detect pathogen-infected cells. Much is known about the mechanisms that control the processing of antigenic peptide substrates (1–6), but little is known about the mechanisms that govern the production of peptide material for the MHC class I pathway. When the assumption that degradation of “old” functional proteins was the sole basis for antigenic peptide material for the endogenous MHC class I pathway was challenged by new models suggesting that peptide material might, in fact, be derived from alternative sources, such as defective ribosomal products (7), the interest in finding the origin of these peptides and what governs their synthesis was initiated. Observations, such as that antigenic peptides can be derived from cryptic translation products (8), that the presence of mRNA and not the full-length protein determines antigen presentation, and that viruses target mRNA translation to evade the immune system (9–11), have led to a shift of focus from protein degradation to mRNA translation as being the critical process. However, efforts to take these earlier observations further and identify what such polypeptides might actually be, or where they come from, have been hampered by the fact that no translation products derived from an ORF of an mRNA that can directly be linked to antigen presentation have been identified so far.

Cap-dependent mRNA translation is initiated by the cap-binding eIF4E, whereas cap-independent translation is characterized by the direct recruitment of the 40S subunit to internal ribosome entry sites (12, 13). Even though methionine (AUG) is by far the most commonly used initiation codon to generate full-length proteins, other codons can also be used, most notably leucine (CUG) (14, 15). In addition to these two types of translation initiation processes that give rise to full-length proteins, there are translation events that are less well understood. Newly synthesized mRNAs are subject to quality control scanning, and detected faults trigger the nonsense-mediated decay (NMD) pathway (16). Best described is the detection of a pretermination codon (PTC) at least 50 nt upstream of the 3′ of the splice junction by the scanning ribosome. Ribosomes that are stalled at a PTC recruit up-frameshift protein 1 (Upf1) (17) to form a link with Upf2 and Upf3 of the exon junction complex (EJC). This prevents the mRNA from translating full-length proteins and singles out the faulty mRNA for transport to processing bodies and degradation (18). The pioneer round of translation that precedes NMD requires the ribosome to be initiated around the cap-binding complex (CBC), including CBP20 and CBP80 (19), but it is not known if the ribosomes produce peptides during this scanning. Here, we provide evidence that peptides produced during the pioneer round of mRNA translation provides a major source of antigenic peptide substrates for the MHC class I pathway.

Results and Discussion

Antigenic Peptides for the MHC Class I-Restricted Pathway Are Derived from Noncanonical mRNA Translation.

It has been determined that rapid proteasome-dependent degradation of peptide substrates is required for the initial steps in antigenic peptide processing, and it has also been proposed that this involves the ubiquitin-dependent proteasomal pathway (1). The IκBα protein is a key regulator of the NF-κB pathway and is rapidly degraded via ubiquitin-dependent 26S proteasome activity in response to different extracellular stimuli, including TNF-α. This is mediated by the IκB kinase (IKK), which phosphorylates IκBα at Ser32 and Ser36 and leads to its subsequent ubiquitination and immediate degradation (20) (Fig. S1A). Hence, in response to extracellular stimuli, the IκBα protein requires two independent modifications for becoming targeted for degradation. This makes IκBα a suitable reporter protein to test if rapid degradation of full-length properly folded functional proteins via the ubiquitin-dependent proteasomal pathway is a source of MHC class I antigenic peptides. We fused the SIINFEKL (SL8) immunodominant epitope from chicken ovalbumin (Ova) to the C terminus of WT IκBα (IκBα-wt-SL8) and to an IκBα that carries mutations in both the IKK phosphorylation sites (IκBα-mut-SL8) (Fig. S1B). These constructs were coexpressed with the MHC class I K^b molecule in human H1299 cells, and the presentation of the SL8 epitope was determined using the B3Z CD8⁺ T-cell reporter system (21). Treatment with TNF-α resulted in a rapid degradation of the endogenous IκBα as well as the IκBα-wt-SL8 reporter protein, whereas the IκBα-mut-SL8 was unaffected, as expected (Fig. 1A). Following 30 min of treatment with TNF-α, the levels of IκBα-wt-SL8 were reduced to ∼50%, indicating that half of the protein amount, including newly synthesized proteins during this time period, had been targeted for degradation. However, up to 240 min after treatment with TNF-α, we observed only a small nonspecific increase in SL8 presentation from the IκBα-wt-SL8 construct compared with the IκBα-mut-SL8 or the control protein in which the SL8 epitope had been fused to GFP (GFP-SL8) (Fig. 1B).

Fig. 1.

Antigenic peptides are produced by noncanonical mRNA translation initiation. (A) Western blot showing the rapid degradation of properly folded full-length IκBα-wt-SL8 and endogenous IκBα in H1299 cells following treatment with TNF-α (Fig. S1). The expression of the IκBα-mut-SL8, which carries mutations (Ser to Ala) in the IKK phosphorylation sites at Ser32 and Ser36, is not affected by TNF-α treatment. Actin is shown as a loading control. W.B., Western blot. (B) Effect on SL8 epitope presentation after IκBα-wt-SL8 degradation in H1299 cells expressing exogenous K^b after TNF-α treatment for indicated time periods, during which cells were washed and incubated with the B3Z reporter CD8⁺ T cells. Control cells were exposed to exogenous SL8 peptide for 16 h (peptide control). The difference in antigen presentation between IκBα-wt-SL8 and IκBα-mut-SL8 after TNF-α treatment is not significant. N.S., not significant. (C) Cartoon illustrating indicated codon positions in the MBP or in a chicken Ova construct. WT codons (Upper, black) and codon changes (Lower, red) are shown. (D, Upper) Autoradiograph of [³⁵S]methionine metabolic pulse labeling, followed by MBP immunoprecipitation in the presence of 25 μM proteasome inhibitor MG132. The two upper bands are derived from AUG codons at position 1 (full length) or codon 22 (MBP 22AUG). Changing codon 39 from CUU to CUG results in the appearance of a translation product (Codon 39) without affecting the synthesis of the two AUG-initiated polypeptides. The corresponding effect of indicated codon changes on the presentation of the MBP(79–87) epitope in H1299 cells coexpressing the K^k molecule is shown below using MBP(79–87) reporter CD8⁺ T cells. F.L., full length. (E) Effect on MBP(79–87) antigen presentation after indicated codon changes in the MBP ORF. (F, Upper) Autoradiograph of metabolic pulse labeling followed by Ova IP in the presence of proteasome inhibitors of cells expressing Ova constructs with indicated silent point mutations. (Lower) Corresponding T-cell assay using B3Z reporter T cells. Antigen presentation was carried out in human H1299 cells expressing the mouse K^b MHC class I molecule. The results are representative of at least three independent experiments plus SD. **P < 0.005; ***P < 0.001.

We conclude that the rapid degradation of properly folded full-length proteins via the 26S proteasome pathway does not necessarily constitute a source for antigenic peptides, and we next started to investigate which type of translation products might be used instead. Alternative codons, particularly CUG (leucine), can be used to initiate translation (14, 22). Upstream of the MHC class I epitope MBP(79–87) in the myelin basic protein (MBP) (23) ORF, there are leucine codons at positions 16 (CUG), 39 (CUU), and 73 (CUG) and one codon downstream at position 110 (CUG) (Fig. 1C, Left). Metabolic pulse labeling in the presence of proteasome inhibitors (MG132) showed that silent mutations in any of these codons (i.e., different codon for the same amino acid) do not change the synthesis or stability of the full-length MBP, or of a second MBP product that is derived from an in-frame AUG at position 22 (Fig. 1D, Upper). The codon 22 product is synthesized at a similar rate as the full-length MBP but is highly unstable and is difficult to detect unless cells are treated with proteasome inhibitor (Fig. S2A). Interestingly, silent changes in codon 39 from a CUU to CUG resulted in the detection of a highly unstable translation product (codon 39 product) that could be observed in the presence of proteasome inhibitor (Fig. 1D, Upper and Fig. S2B). By using CD8⁺ lacZ T cells toward the MBP(79–87) epitope presented on K^k molecules, we noticed that changing codon 39 from CUU to CUG resulted in an increase of ∼35% in antigen presentation (Fig. 1D, Lower). Introducing AUG or GUG (Val), which can also be used for translation initiation (14) at position 39, resulted in expression of the codon 39 product at different levels that correlated with the presentation of the MBP(79–87) epitope (Fig. 1E and Fig. S2B). Other codon changes at position 39 that have been shown to initiate translation either reduced or had no effect on antigen presentation. Deletion of the second in-frame AUG prevented the production of the codon 22 product, as expected, without affecting antigen presentation from the WT or from the codon 39CUG constructs (MBPΔ22AUG and MBP39CUG-Δ22AUG), demonstrating that the half-life of a polypeptide, per se, does not determine antigen presentation (Fig. 1E). Even though it is not possible to conclude that the codon 39-dependent translation product is indeed processed for the MHC class I pathway, these results underline the fact that the mRNA translation initiation events that govern the production of full-length products and antigenic peptides are distinguishable.

Similar to the MBP ORF, there are numerous CUG codons upstream and downstream of the SL8 epitope in the Ova mRNA (Fig. 1C, Right). When we introduced silent mutations (CUG to CUC) in codons 114 and 124 (relative to +1), which are located upstream of the SL8 epitope, we observed no changes in the expression of the full-length protein (Fig. 1F, Upper and Fig. S2C) but a significant decrease in SL8 antigen presentation (Fig. 1F, Lower). However, when we carried out the same mutations at positions 173 and 182, which are also located upstream of the epitope, we observed no significant changes in SL8 antigen presentation, indicating that not all CUG codons within the Ova ORF have the same impact on antigenic peptide production (Fig. S2D). From these results, we conclude that the effect on antigen presentation caused by silent changes within an ORF does not depend on the epitope or the MHC class I haplotype. Little is presently known about the synthesis of peptides during the early scanning events of the immature messenger RNA protein complex (mRNP), and we are currently investigating the context of codon 39 initiation; it is plausible that it depends on RNA structures to a larger extent compared with the sequence requirements that are characteristic of the canonical AUG initiation. This could offer an explanation as to why not all CUG codons within the Ova ORF, for example, have the same impact in terms of antigenic peptide production and why some alternative initiation codons, such as GUG, ACG, UUG, and AUU, have different effects.

Production of Antigenic Peptides and Production of Full-Length Proteins from an mRNA Are Temporally Different.

Under normal conditions, a peptide presented together with an endogenous MHC class I K^b molecule has a half-life of over 6 h (24), making it difficult to study the dynamics of antigenic peptide production vs. antigen presentation. The mouse K^b molecule, together with a synthetic SL8 peptide, was found to have a half-life of approximately 10 min on the surface of human cells (H1299), presumably attributable to species-related incompatibilities in this MHC–peptide assembly complex (Fig. 2A). This circumstance, together with transfection of in vitro transcribed capped Ova mRNA, makes this a suitable system to study the production of an antigenic peptide substrate vs. the full-length protein from a given mRNA over time. The levels of Ova mRNA (2 μg) following transfection were detected using quantitative RT-PCR (qRT-PCR) and were found to be approximately half after 8 h (Fig. 2B). The presentation of the SL8 epitope from the transfected Ova mRNA was determined by fixing the cells after indicated time points and revealed a peak of antigen presentation after approximately 2.5 h, after which it rapidly declined (Fig. 2C). It has been estimated that the time it takes for a peptide to be produced, processed, loaded onto the MHC class I molecules, and presented on the cell surface is between 30 and 60 min (25). Assuming that the half-life of K^b–SL8 complexes loaded with synthetic vs. endogenous peptides is similar, if one takes into account the turnover rate of the K^b molecule, one can estimate that the production of SL8 precursor peptides from the transfected Ova mRNA reaches its peak less than 1.5 h after RNA transfection. Interestingly, at the same time, we could detect production of full-length Ova protein until at least 8 h after transfection (Fig. 2D and Fig. S3). Hence, the synthesis of antigenic peptides from the Ova mRNA stops well before that of the synthesis of full-length proteins.

Fig. 2.

Synthesis of antigenic peptides and full-length proteins is temporally distinct. (A) Turnover rate of the endogenous K^b MHC class I-SL8 complex in EL4 mouse cells, or mouse K^b-SL8 complex in human H1299 cells, was estimated by measuring SL8 presentation on K^b molecules. Synthetic SL8 peptide was added for 1 h, followed by extensive washing (time 0) and cell fixation at the indicated time points before incubation with B3Z reporter cells. (B) Levels of capped Ova mRNA in H1299 cells following RNA transfection were determined using qRT-PCR. The arbitrary value of 100% was given at 1 h. (C) Antigen presentation after transfection of Ova mRNA in H1299 cells expressing K^b was determined on fixed cells at the indicated time points. Maximum antigen presentation (100%) was observed at 2.5 h. (D) Western blot shows the expression of Ova at different time points after mRNA transfection in H1299 cells. Endogenous actin is used as a loading control. W.B., Western blot. The results are representative of at least three independent experiments plus SD (Fig. S3).

mRNAs Targeted for the NMD Pathway Produce Antigenic Peptide Substrates.

Newly synthesized and spliced mRNAs undergo a quality control before they can produce full-length proteins. To test if mRNAs that undergo the control ribosome-mediated quality scanning that precedes the NMD produce antigenic peptide substrates, we fused the SL8 epitope in the first exon of a β-globin NMD reporter construct (26) that carries a PTC 53 nt upstream of the EJC (Glob1-SL8-PTC) (Fig. 3A). RNA quantification revealed that the Glob1-SL8-PTC RNA is efficiently degraded and that the Glob1-SL8 RNA is stable, as expected (Fig. 3B). We observed the expression of the YFP-globin fusion protein from the corresponding Glob1-SL8 construct that lacks the PTC but not from the Glob1-SL8-PTC construct (Fig. 3C). We also inserted the SL8 and MBP(79–87) coding sequences in the second exon just after the PTC [Glob2-PTC-SL8/MBP(79–87)], which leads to a less efficient NMD process, allowing some mRNAs to escape the NMD and express a truncated protein that is terminated at the PTC (Fig. 3 B and C). The molecular mechanism behind this partial evasion of NMD is not clear but was observed using both epitope sequences. The presentation of the SL8 epitope from the Glob1-PTC-SL8 construct reaches ∼85% of the levels observed from that of the Glob1-SL8 construct (Fig. 3D). Similar results were obtained using different cell lines (Fig. S4). The SL8 and the MBP(79–87) epitopes are not produced when placed downstream of the PTC codon in the Glob2-PTC-SL8/MBP constructs because the PTC prevents the read-through into the SL8 coding sequence. The presentation of the MBP(79–87) epitope on K^k molecules derived from the Glob1-MBP-PTC construct was found to be ∼90% compared with the corresponding control Glob1 mRNA that lacks the PTC codon (Glob1-MBP) (Fig. 3E). These results demonstrate that the production of different antigenic peptide substrates for various MHC class I molecules coincides with the early round of translation that precedes the NMD.

Fig. 3.

mRNAs targeted for NMD produce antigenic peptides but not full-length proteins. (A) Cartoon illustrating different NMD reporter constructs and the position of the PTC and the SL8 or MBP(79–87) epitope. (B) Ratios of normal and PTC-carrying mRNAs from constructs in which the SL8 is inserted in exon 1 or 2 to the mouse urinary protein (MUP) control mRNA are shown. Numbers show the relative amounts of indicated mRNAs 48 h after transfection in HeLa cells. MUP. (C) Western blot shows the expression of YFP-globin fusion proteins in HeLa cells from indicated constructs using α-YFP mAb. Insertion of SL8 in exon 2 just downstream of the PTC (Glob2-PTC-SL8) partially disrupts the NMD process, allowing the expression of a truncated product that terminates at the PTC. TP, truncated product; W.B., Western blot. (D) Antigen presentation of the SL8 epitope from an increasing amount of indicated constructs in HeLa cells expressing mouse K^b. Similar results were obtained using H1299, a HEK293-TK^b stable cell line, and B6 mouse fibroblast (Fig. S4). Approximately 85% antigen presentation is retained from Glob1-SL8-PTC mRNAs that are targeted for NMD compared with the non-NMD Glob1-SL8 construct. The Glob2-PTC-SL8 construct does not produce antigenic peptide substrates because the PTC stops ribosomal read-through of the SL8 epitope. *P < 0.05. (E) Presentation of the MBP(79–87) epitope on K^k class I molecules from the Glob1-MBP-PTC construct (1 μg cDNA) in L929 mouse fibroblasts reached ∼90% compared with the Glob1-MBP. Similar data were obtained using H1299 and HEK293 cells expressing mouse K^k. The results are representative of at least three independent experiments plus SD.

eIF4E Binding to the Cap Structure Is Not Required for the Production of Antigenic Peptides.

To investigate the role of early mRNA translation events in the production of antigenic peptides further, we expressed an RNA aptamer that specifically targets eIF4E (α-eIF4E aptamer) and prevents its interaction with the cap structure (27, 28). Metabolic pulse labeling assays and Western blots showed that increasing amounts of the α-eIF4E aptamer resulted in a decrease in the rate of synthesis of YFP-globin from the Glob1-SL8 and Glob2-SL8 constructs (∼80%) and from the Ova mRNA (Fig. 4 A and B). The α-eIF4E aptamer had no effect on the NMD process, demonstrating that it does not affect the interaction between the CBC and the cap structure (Fig. 4C). The expression of the α-eIF4E aptamer had little effect on the presentation of antigenic peptides relative to MHC class I molecule expression (Fig. 4D). The 4E-BP1 protein prevents the interaction between eIF4E and eIF4G and blocks cap-dependent translation initiation (29). Accordingly, increasing amounts of 4E-BP1 resulted in a decrease in the expression of YFP-globin (Fig. 4E). However, as with the α-eIF4E aptamer, this had little impact on the production of antigenic peptide substrates from any of the tested reporter constructs (Fig. 4F).

Fig. 4.

eIF4G but not eIF4E is required for the synthesis of antigenic peptides. (A) Cells were transfected with the indicated YFP-globin reporter construct for 24 h, followed by expression of the α-eIF4E aptamer (24 h), which prevents interaction between the cap structure and eIF4E. Autoradiograph of a [³⁵S]methinonine metabolic pulse labeling experiment (Left) and corresponding Western blot (Right) in HeLa cells expressing indicated amounts of the α-eIF4E RNA aptamer. W.B., Western blot. (B) Western blot of Ova expression under similar conditions. (C) RNA stability of indicated globin NMD reporter constructs in the presence or absence of the α-eIF4E aptamer. The numbers indicate the relative amounts of mRNAs with SD. (D) Presentation of the SL8 epitope expressed from indicated mRNA constructs in the presence of α-eIF4E aptamer. Data are presented as relative to presentation of the exogenous SL8 peptide. (E) Western blot of YFP-globin expression from the Glob1-SL8 construct in the presence of increasing amounts of 4E-BP1. (F) Corresponding antigen presentation from indicated constructs under similar conditions as in E. (G) Western blot shows that increasing amounts of protease 2A, which cleaves the eIF4G translation factor (Fig. S5), lead to a reduction in YFP-globin protein expression levels. (H) Relative presentation of SL8 epitope derived from indicated constructs compared with cells being exposed to exogenous SL8 peptide for 16 h in the presence of increasing amounts of protease 2A. All results are representative of three independent experiments and SD.

The eIF4G initiation factor is required for the assembly of both the eIF4E- and CBC-dependent ribosomal complexes that initiate ribosomal quality control scanning (19, 30). The polio virus 2A protease specifically cleaves and inactivates eIF4G (Fig. S5); in line with this, we found that increasing amounts of protease 2A resulted in a partial loss of YFP-globin expression (Fig. 4G) and a reduction of ∼60% in antigenic peptide presentation (Fig. 4H). Hence, synthesis of antigenic peptides requires eIF4G but not eIF4E, indicating that the production of antigenic peptides is spatiotemporally different from that of full-length proteins and is governed by a different mechanism of translation initiation (Fig. 5). Taken together, the presented results are in agreement with the notion that production of antigenic peptides takes place during the early translation events that precede the NMD.

Fig. 5.

Model for the synthesis and selection of antigenic peptide substrates during the pioneer round of translation that precedes NMD. mRNAs are checked for PTCs during the initial quality control translation that follows newly synthesized and spliced mRNAs. This translation is initiated by the CBC, including CBP20/80, and produces PTPs that are instantly degraded and selected for antigenic peptide processing for the MHC class I pathway. mRNAs found to contain PTCs are targeted for the NMD pathway (1), and those mRNAs that pass the quality control mature into mRNP complexes that bind the eIF4E cap-binding factor and initiate canonical translation complexes that govern the synthesis of full-length proteins (2). Degradation of full-length proteins plays only a minor role as a source of antigenic peptides for the MHC class I pathway.

Concluding Remarks

It is likely that PTPs represent fragments of the corresponding ORFs, and would therefore be targeted for rapid degradation by the cells to limit any potential harm they might cause by uncontrolled interference with cellular processes. Hence, in this respect, our findings on PTPs would fit well with criteria for what has been postulated to be hallmarks of antigenic peptide substrates, including rapid degradation and cryptic translation. There would be obvious benefits for the immune surveillance process if it has been adapted to recognize PTP-derived products on MHC class I molecules. Because these peptides are the first to be produced from mRNAs, it would make transacting mechanisms used by, for example, viruses to manipulate the MHC class I pathway less effective. A mechanism in which newly synthesized viral mRNAs are scanned by ribosomes using a “loose” initiation requirement for the production of antigenic peptides would form a more difficult challenge for viruses to overcome. Transfected mRNAs, despite not being spliced, only support the synthesis of antigen peptide substrates early after transfection, indicating that the maturation of mRNP complexes takes place irrespective of whether the mRNAs are expressed endogenously or introduced artificially. Hence, it is conceivable that nonspliced viral mRNAs would go through a similar maturation process. There are, however, reports to support that latent viruses have mechanisms in place to overcome the challenge of PTPs being produced during the pioneer round of translation. EBNA1, encoded by the Epstein–Barr virus, inhibits its own translation in cis to suppress antigen production, whereas the full-length EBNA1 has a long half-life to ensure that it is expressed at levels that are sufficient to support its viral functions (9, 31).

The fact that antigenic peptides can be synthesized from mRNAs, although not producing full-length proteins, has other interesting implications. One of the cornerstones in preventing self-recognition of CD8⁺ T cells is the complex process of generating central tolerance in the thymus, whereby maturing T cells are exposed to self-peptides and, if reactive, become eliminated. The possibility that thymic cells could produce the required epitopes for this selection process without having to produce the full-length proteins is attractive. The results presented here by no means prove this hypothesis; however, importantly, they show that it might be possible. Finally, these results imply that similar to different RNA polymerases producing a variety of RNA products with defined cellular functions, different ribosome complexes produce peptide products with defined cell biological functions.

Materials and Methods

T-Cell Hybridoma, Cell Culture, and Transfection.

The SIINFEKL/K^b-specific (B3Z) and the MBP(79-87)/K^k-specific T-cell reporter hybridomas have been described previously (32). Human cell lines were cultivated under standard conditions in RPMI 1640. The human HEK-293 T Kb stable cell line and mouse B6 cell line were gifts from L. Eisenlohr (Thomas Jefferson University, Philadelphia, PA). For mRNA transfection, cells were transfected with 0.5 μg of Kb expression plasmid along with 3 μL of Genejuice according to the manufacturer's protocol (Merck Biosciences). Twenty-four hours later, the cells were transfected with 2 μg (unless stated otherwise) of capped mRNA using lipofectamine according to the manufacturer's protocol (Invitrogen). Details of plasmid construction and qRT-PCR are provided in SI Materials and Methods.

In Vitro Transcription.

Capped mRNAs were synthesized with T7 RNA polymerase (mMessage Machine; Ambion) according to the manufacturer's protocol. The pCDNA3-Ova (described in SI Materials and Methods) linearized with Xba1 was used in a 50-μL final volume for generating mRNAs and was purified using the RNeasy kit (Qiagen).

Metabolic Cell Labeling and Immunoprecipitation.

All mRNA translation assays were carried out in H1299 and HeLa cells transfected with indicated constructs as described previously (9). Transfected cells were cultured for 36 h before being treated with 25 μM MG132 for 1 h in methionine-free medium containing 2% dialyzed FCS in DMEM. [³⁵S]Methionine (0.15 mCi/mL; Perkin–Elmer) was added in the presence of proteasome inhibitor, and the cells were harvested at indicated time points. The peptide products were immunoprecipitated using indicated antibodies and separated on precast Bis-Tris 4–12% acrylamide SDS/PAGE gels (Invitrogen). The relative amount of protein synthesis was determined using a phosphoimager.

T-Cell Assay and Fixation.

T-cell assays in human cell lines were carried out as described previously (9). Briefly, human H1299 and HeLa cell lines were cotransfected with the K^b or K^k expression vector, together with the different reporter constructs. Human HEK-293 T K^b is a stable cell line, and the mouse B6 cell line expresses endogenous K^b. The B3Z or MBP CD8⁺ T-cell hybridoma expresses LacZ in response to activation of T-cell receptors specific for the SIINFEKL peptide (Ova-immunodominant peptide) in the context of H-2K^b MHC class I molecules and for the MBP(79–87) peptide in the context of H-2K^k MHC class I molecules, respectively.

H1299 cell-expressing exogenous K^b (Fig. 2A) was pulse-chased for 48 h with synthetic SIINFEKL peptide and extensively washed before being fixed in 0.05% glutaraldehyde and 0.1 M glycine. The cells were then washed two times in 1× PBS and cultured with B3Z T-cell hybridoma for 16–20 h.

RNA Preparation, RT-PCR, and qRT-PCR.

Total cellular RNAs were extracted and purified with the RNeasy kit following the manufacturer's protocol. RT was carried out with 1 μg of RNA using the M-MLV reverse transcriptase and oligo-dT (Invitrogen). The StepOne (Applied BioSystems) real-time PCR system was used for qRT-PCR, and the reaction was performed with the Perfecta SYBR green Fast mix ROX (Quanta) and specific primer pairs for each gene of interest.