Leucine-tRNA Initiates at CUG Start Codons for Protein Synthesis and Presentation by MHC Class I

See allHide authors and affiliations

Science  29 Jun 2012:
Vol. 336, Issue 6089, pp. 1719-1723
DOI: 10.1126/science.1220270

Noncanonical Pathway

The textbook view of translation of messenger RNA to protein is that it is always initiated from open reading frames (ORFs) that begin with an AUG codon (encodes methionine) by an initiator methionine-bound transfer RNA (tRNA). There is evidence, however, that some polypeptides are produced from non–AUG-initiated ORFs. Starck et al. (p. 1719; see the Perspective by Dever) used a variety of biochemical techniques to determine the underlying mechanism for such nontraditional translation initiation. Comparison of translation initiation from AUG-initiated ORFs with those beginning with leucine CUG-initiated ORFs revealed that cells can use an elongator Leu-tRNA to initiate translation at CUG codons. CUG-initiated peptides were presented by major histocompatibility class I molecules and could activate T cells.


Effective immune surveillance by cytotoxic T cells requires newly synthesized polypeptides for presentation by major histocompatibility complex (MHC) class I molecules. These polypeptides are produced not only from conventional AUG-initiated, but also from cryptic non–AUG-initiated, reading frames by distinct translational mechanisms. Biochemical analysis of ribosomal initiation complexes at CUG versus AUG initiation codons revealed that cells use an elongator leucine-bound transfer RNA (Leu-tRNA) to initiate translation at cryptic CUG start codons. CUG/Leu-tRNA initiation was independent of the canonical initiator tRNA (AUG/Met-tRNAiMet) pathway but required expression of eukaryotic initiation factor 2A. Thus, a tRNA-based translation initiation mechanism allows non–AUG-initiated protein synthesis and supplies peptides for presentation by MHC class I molecules.

In almost all nucleated cells, newly translated polypeptides supply antigenic precursors for loading major histocompatability complex (MHC) class I molecules (1). Peptide-loaded MHC class I (pMHC I) molecules reveal the presence of viral or mutated proteins to circulating cytotoxic T cells (CTLs), which bind pMHC I through their T cell receptors to eliminate infected or transformed cells. Antigenic precursors are translated from conventional AUG-initiated open reading frames (ORFs) by the canonical initiator transfer RNA (tRNA), Met-tRNAiMet (2, 3). Cryptic, non–AUG-initiated ORFs (47) also generate pMHC I during viral infections (812) and oncogenesis (1315) by unknown mechanisms. Cryptic CUG start codons can also be decoded with leucine at the initiation stage of translation in mammalian cells (57, 16). However, this decoding is incompatible with the current model of translation, which indicates that ribosomes are preloaded with initiator Met-tRNAiMet before recognition of AUG or even non-AUG start codons (17).

Our previous study suggested that translation of antigenic precursors from a CUG start codon using leucine represents a distinct initiation pathway (16). To determine the molecular mechanism of CUG/leucine initiation, we first screened a series of compounds described as inhibitors of eukaryotic protein synthesis (18) using primer extension inhibition analysis (toeprinting) (19) of AUG-YL8 and CUG-YL8 mRNA ribosome initiation complexes (fig. S1). We found that NSC119893, which inhibits Met-tRNAiMet association with eukaryotic initiation factor 2 (eIF2) (20), selectively inhibits AUG-YL8 initiation (table S1) in a dose-dependent fashion, whereas CUG-YL8 toeprints were resistant to NSC119893 treatment (Fig. 1A). Structurally unrelated protein synthesis inhibitors, such as suramine and aurin tricarboxylic acid, also inhibited AUG initiation yet enhanced initiation at the CUG start codon (table S1). In contrast, the small molecule acriflavine inhibited CUG initiation more than AUG initiation in a dose-dependent fashion (Fig. 1A). Thus, a structurally diverse set of compounds can distinguish ribosomal recognition of AUG and CUG start codons.

Fig. 1

Cryptic CUG/leucine initiation is differentially affected by protein synthesis inhibitors. (A) Toeprint analysis of AUG-YL8 and CUG-YL8 mRNAs (fig. S1), which encode a peptide, either MYL8 or LYL8, respectively, by using RRL in the presence of indicated concentrations of NSC119893 or acriflavine. Data are presented as a percentage of untreated samples (means ± SEM; n = 2 or 3). T cell hybridoma response (BCZ103) to (B) AUG-YL8 versus (C) CUG-YL8–translated peptides from COS-7 cells transfected with plasmid DNAs and treated with NSC119893 (50 μM) or acriflavine (10 μM) for 3 hours (data are representative of three independent experiments). Peptides were acid-extracted and analyzed by reversed-phase high-performance liquid chromatography to resolve the methionine (MYL8)– versus leucine (LYL8)–initiated peptides. Activation of BCZ103 T cell hybridoma was used to measure the amount and identity of peptides produced by initiation at AUG or CUG start codons. (D) Peptide abundance from CUG-YL8 in the presence of NSC119893 and acriflavine from (C) (area under the curve) are relative to untreated samples (UT) (means ± SEM; n = 3). Statistical significance was evaluated with the unpaired t test (*P < 0.05; **P < 0.01; ***P < 0.001).

We next assessed the effect of these protein synthesis inhibitors on translation of antigenic precursors in living cells by biochemically analyzing peptides from extracts of cells transfected with the AUG-YL8 or CUG-YL8 plasmids (5). As expected, a single peak of antigenic activity—corresponding to the methionine-initiated peptide (MYL8)—was detected from AUG-YL8–transfected cells (Fig. 1B). Yet, CUG-YL8–transfected cells yielded the leucine-initiated peptide (LYL8) as well as the MYL8 peptide, arising from Met-tRNAiMet “wobble” initiation (Fig. 1C). Although NSC119893 inhibited the expression of MYL8 from AUG and CUG start codons, it did not inhibit decoding of the CUG initiation codon with leucine (LYL8) (Fig. 1, B to D). In contrast, and consistent with the toeprint analysis (Fig. 1A), translation of LYL8 was inhibited by acriflavine, a nucleic acid intercalator, whereas MYL8 initiation from either AUG or CUG codons was unaffected (Fig. 1, B to D). This inhibitor effect was not limited to translation of antigenic precursors, because NSC119893 inhibited AUG-GFP, but not CUG-GFP, expression (fig. S2, A and B). Conversely, CUG-GFP, but not AUG-GFP, expression was inhibited by acriflavine treatment in cultured cells (fig. S2C). Because NSC119893 inhibits global protein synthesis (fig. S3) and induces stress by limiting Met-tRNAiMet availability (20, 21), our analysis suggests that CUG/leucine initiation proceeds independently of the Met-tRNAiMet initiation pathway and persists during host translation shutoff and stress.

To study cryptic translation in professional antigen-presenting cells (APCs), we used primary APCs from transgenic mice expressing a bicistronic ORF with an AUG/methionine-initiated peptide (WI9) followed by a cryptic CUG/leucine-initiated peptide (LYL8) (fig. S4), both detected by T cell hybridomas (6). NSC119893 treatment reduced WI9 presentation to the level observed with the elongation inhibitor cycloheximide (CHX) (Fig. 2A) and induced complete host translation shutoff (fig. S5) in bone marrow–derived dendritic cells (BM-DCs). Unlike inhibition by CHX treatment, however, presentation of the cryptic LYL8 peptide persisted with NSC119893 treatment in BM-DCs (Fig. 2A) and splenocytes (fig. S6A). Thus, NSC119893 inhibited AUG-initiated presentation of the WI9 peptide, such that more than twice the number of NSC119893-treated APCs were required for a half-maximal T cell response (Fig. 2B, blue bar) compared with an equivalent number of APCs presenting the CUG/leucine-initiated LYL8 peptide (Fig. 2B, orange bar). These results show that professional APCs can use distinct initiation mechanisms to translate antigenic precursors during host translation shutoff, such as during stress or viral infection.

Fig. 2

MHC class I presentation of AUG- versus CUG-initiated peptides in primary cells during host translation shutoff and APC maturation and activation. (A) BM-DC presentation of LYL8 to the BCZ103 T cell hybridoma (bottom) and WI9 to the 11p9Z T cell hybridoma (top), in response to treatment with NSC119893 (40 μM) or CHX (100 μg/ml) for 2 hours after mild acid wash to remove preexisting peptide–MHC class I complexes (38). BM-DCs were obtained from WI9*LYL8 transgenic mice (fig. S4), and C57BL/6 (B6) mice were used as transgene-negative controls. T cell hybridoma responses are representative of three independent experiments and are normalized for cell number with an assay to measure metabolically active cells by conversion of the tetrazolium compound MTS into formazan (38). (B) Relative number of BM-DCs required for half-maximal T cell hybridoma response in the presence of NSC119893 from (A) (means ± SEM; n = 3). (C) CUG/leucine-initiated presentation after maturation and activation with LPS (1 μg/ml) for 3 hours (bottom) compared with AUG/methionine-initiated presentation (top). T cell hybridoma responses are representative of at least three independent experiments from three mice. (D) Relative number of BM-DCs required for half-maximal T cell hybridoma response in the presence of LPS (1 μg/ml) from (C) (means ± SEM; n = 3). Statistical significance was evaluated with the unpaired t test (*P < 0.05; **P < 0.01; ***P < 0.001).

Dendritic cells are key for initiating immune responses (22) and, upon activation, undergo numerous modifications (23, 24). We therefore analyzed antigen presentation of LYL8 and WI9 peptides in immature versus lipopolysaccharide (LPS)–activated APCs. Although LPS exposure enhanced AUG/methionine-initiated WI9 presentation (Fig. 2C), CUG/leucine-initiated antigen presentation of LYL8 peptide was substantially enhanced (Fig. 2C), despite comparable expression of pMHC I on LPS-activated cells (fig. S7, A and B). With ex vivo (Fig. 2D) and in vivo LPS exposure (fig. S7, D and E), less than one-quarter of the number of BM-DCs (Fig. 2D, orange bar) and splenocytes (fig. S7, C to E) presenting LYL8 were required for a half-maximal T cell response. This effect is also seen with a lower Met-tRNAiMet inhibitor concentration (NSC119893) (fig. S6B), which indicates that different cell stimuli profoundly increase cryptic CUG/leucine-initiated translation. Thus, the cryptic pMHC I repertoire can be enhanced in vivo by even a short LPS exposure in professional APCs.

To understand the molecular basis of CUG/leucine initiation, we developed a technique to analyze initiator tRNAs from ribosome initiation complexes stalled at the start codons of either the AUG-YL8 or CUG-YL8 mRNAs (fig. S8). Using this technique, we isolated intact ribosomal initiation complexes bound to AUG or CUG start codons as indicated by the enrichment of ribosomal RNA (rRNA) relative to the control with no mRNA (Fig. 3A). In addition, material migrating at the size expected for type I tRNAs (74 to 77 nucleotides), such as initiator Met-tRNAiMet, was observed from both AUG and CUG initiation complexes (Fig. 3A). However, a single nucleotide substitution of A to C (AUG to CUG) in the mRNA yielded at least one additional distinct band (Fig. 3A) migrating with larger type II tRNAs, which typically contain Leu-tRNA isoacceptors bearing longer variable loops (Fig. 3B).

Fig. 3

Leucine-tRNA is enriched at CUG start codons. (A) RNA isolated from equimolar AUG-YL8 (AUG) and CUG-YL8 (CUG) initiation complexes after purification using ribosome complex capture (38) from RRL or (B) total RRL tRNA (class I tRNA contains Met-tRNAiMet and class II tRNA contains Leu-tRNA) were analyzed by [32P]pCp 3′-end labeling and resolved using 10% denaturing urea–polyacrylamide gel electrophoresis. Data are representative of three independent experiments. (C) tRNA microarray analysis of AUG and CUG initiation complexes in the absence (left) or presence (right) of an unlabeled initiator Met-tRNAiMet probe. tRNA was analyzed using at least three independent microarrays. (D) Relative tRNA abundance at AUG and CUG start codons (from C) (presented relative to total tRNA signal quantified) is shown for initiator tRNAs (leucine and methionine) and tRNA present from ribosome elongation (Thr, Phe, and Asn) along the ORF (shown). (E and F) Northern blot analysis of AUG versus CUG initiation complex tRNAs. Minimal mRNAs (38) were used containing only one AUG or CUG start codon. (E) Thr- and Phe-tRNAs are derived from A-site occupancy and elongation, respectively. The same Northern blot was stripped and reprobed with the indicated full-length tRNA probes, and each blot shown is from an equivalent PhosphoImager exposure. Data are representative of four independent experiments. (G) Quantification of Leu-tRNA with anticodon 5′-CAG-3′ (Leu-tRNA-CAG) in CUG initiation complexes from (F) (means ± SEM; n = 4). Statistical significance was evaluated with the unpaired t test (*P < 0.05; **P < 0.01; ***P < 0.001).

Next, tRNA microarray analysis (25) showed that both AUG and CUG initiation complexes contained Met-tRNAiMet hybridization signals that could be blocked with an unlabeled Met-tRNAiMet probe (Fig. 3C, blue boxes). However, Leu-tRNA was enriched in CUG initiation complexes (Fig. 3C, orange boxes, and Fig. 3D). In addition, Thr-tRNA signal was observed from both AUG and CUG initiation complexes, because the second codon of the ORF codes for threonine (ACC) (Fig. 3, C and D). Furthermore, lower amounts of both Phe-tRNA and Asn-tRNA were also detected, which indicated that some ribosomes could participate in up to two rounds of elongation (Fig. 3D).

The nature of tRNA present in AUG and CUG initiation complexes was further confirmed by Northern blot analysis with full-length tRNA probes. As expected, initiator Met-tRNAiMet was detected at both start codons (Fig. 3E), whereas Asp-tRNA (no aspartic acid codons in the ORF) was not detected. The presence of Thr-tRNA and Phe-tRNA indicated that both AUG and CUG initiation complexes were competent for A-site selection and elongation (Fig. 3E). As the tRNA microarray results indicated, the larger migrating bands from the CUG initiation complex (Fig. 3A) were of Leu-tRNA origin, as a strong signal for Leu-tRNA with the anticodon 5′-CAG-3′ (Leu-tRNA-CAG), but not Leu-tRNA-UAG or Pro-tRNA, was enriched in the sample (Fig. 3, F and G).

To test whether Leu-tRNA initiation at CUG start codons can be extended to non–AUG-initiated cellular mRNAs, we analyzed the CUG-initiated isoform of Myc, a factor implicated in oncogenesis (26) and previously found associated with initiation complexes using ribosome profiling (27). Initiation complexes were detected at the Myc CUG start codon by using toeprinting (fig. S9A) and were also found to contain Leu-tRNA-CAG (fig. S9B). Together, these results indicate that leucine initiation (5, 6) is not the result of initiator tRNAiMet misacylation (28, 29) but is a mechanistically distinct pathway for Leu-tRNA decoding of CUG start codons

To our surprise, analysis of ribosome assembly (toeprinting) in the presence of Leu-tRNA-CAG and Leu-tRNA-UAG, but not Leu-tRNA-UAA, inhibited AUG but not CUG toeprints in rabbit reticulocyte lysate (RRL) (fig. S10, A and B). Moreover, added tRNA per se did not inhibit initiation, because an equivalent amount of total tRNA, which contains initiator Met-tRNAiMet, actually enhanced both AUG and CUG toeprints (fig. S10, A and B). Further, when we used tRNA-depleted RRL (fig. S10C) (30), Leu-tRNA-CAG specifically stimulated CUG but not AUG initiation, whereas total tRNA enhanced initiation at both the AUG and CUG start codons (fig. S10D). This suggests that a portion of ribosomal subunits is not preloaded with initiator tRNA and is receptive to both Met-tRNAiMet and Leu-tRNA-CAG loading for initiation at CUG start codons.

We next tested whether Leu-tRNA initiates translation of antigenic precursors in living cells, using a stop-codon suppression technique that utilizes a tRNA suppressor to decode a unique stop codon present within the ORF (31, 32). HeLa H2-Kb–expressing cells were transfected with UAG-YL8 mRNA (UAG replaced CUG) with or without Leu-tRNA suppressor. Activation of the T cell hybridoma by the H2-Kb–presented LYL8 peptide was observed only when Leu-tRNA suppressor was also present (Fig. 4A) and was enhanced in a dose-dependent manner (Fig. 4B). Thus, Leu-tRNA can function as an initiator tRNA in cells to generate an antigenic precursor for presentation by MHC class I molecules.

Fig. 4

Leu-tRNA participates in a distinct translation initiation pathway. (A) HeLa H2-Kb cells (APCs) were transfected with UAG-YL8 mRNA alone or with Leu-tRNA-CAG containing a 5′-CUA-3′ anticodon (Leu-tRNA suppressor) to recognize the UAG stop codon. T cell hybridoma (BCZ103) activation measures the generation of the leucine-initiated LYL8 peptide in the presence of Leu-tRNA suppressor (data are representative of three independent experiments) (B) in a dose-dependent manner (means ± SEM; n = 3). (C) L-cell fibroblasts expressing H2-Kb (APCs) were subjected to control, eIF2D, or eIF2A siRNA-mediated knockdown before transfection with plasmid DNAs encoding the bicistronic WI9*LYL8 sequence (fig. S4) and H2-Db. AUG/methionine-initiated presentation of WI9 was measured with the 11p9Z T cell hybridoma and CUG/leucine-initiated presentation of LYL8 was measured with the BCZ103 T cell hybridoma. (D) Western blot analysis of control, eIF2D, and eIF2A siRNA knockdown in cells from (C). Data in (C) and (D) are representative of three independent experiments. (E) Relative number of APCs required for half-maximal T cell hybridoma response in the presence of either eIF2D (top) or eIF2A (bottom) siRNA knockdown from (C) (means ± SEM; n = 3). Statistical significance was evaluated with the unpaired t test (*P < 0.05; **P < 0.01).

How Leu-tRNA is delivered to the P-site of the ribosome independent of the Met-tRNAiMet•eIF2 pathway is unclear. Previously, the initiation factors eIF2D (formerly called ligatin and functionally complemented by MCT-1 and DENR) (33, 34) and eIF2A (35, 36) have been shown in vitro to stimulate initiation with elongator tRNAs (e.g., Phe-tRNA) in addition to canonical initiator Met-tRNAiMet. Therefore, we tested the effect of small interfering RNA (siRNA)–mediated knockdown of eIF2D or eIF2A on the presentation of AUG-initiated WI9 versus CUG-initiated LYL8. Neither AUG/methionine- nor CUG/leucine-initiated presentation was altered with eIF2D knockdown (Fig. 4, C to E). In contrast, eIF2A knockdown significantly inhibited CUG/leucine-initiated, not AUG/methionine-initiated, presentation (Fig. 4, C and D) and required more than twice the number of eIF2A-siRNA–treated APCs to achieve a half-maximal T cell response (Fig. 4E). The finding that CUG/leucine-initiated presentation required at least eIF2A, yet was insensitive to reduced Met-tRNAiMet•eIF2 levels (Fig. 1C, Fig. 2A, and fig. S2A), indicated that initiation at CUG start codons with leucine utilizes an eIF2A-dependent pathway, which may be necessary during infection (37) or translation of endogenous CUG-initiated proteins.

Degeneracy is a hallmark of biological systems. Just as multiple codons are decoded with the same amino acid, CUG start codons may be decoded by more than one initiator tRNA. We propose that the highly regulated step of translation initiation is governed by at least two mechanistically distinct pathways at the level of the initiator tRNA. The alternative initiation pathway that uses Leu-tRNA to decode CUG start codons can explain why usage of CUG-initiated 5′-untranslated regions occurs frequently in cells (27) and can also provide polypeptide precursors for presentation by MHC class I (4).

Supplementary Materials

Materials and Methods

Figs. S1 to S10

Table S1

References (3942)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank K. Collins, A. H. Bakker, K. Chen, N. Nagarajan, and T. Greene for helpful discussions and advice; Y. Ow for initial experiments with NSC119893; and N. Nagarajan and K. Lind for assistance with intravenous injections. The data presented in this paper are tabulated in the main paper and in the supplementary materials. S.R.S. was supported in part by an NIH training grant, a postdoctoral fellowship from the Cancer Research Institute, and a National Research Service Award (NRSA) fellowship from the NIH. M.P.-E was supported by an NIH training grant and by the Ruth Kirschstein NRSA predoctoral fellowship. This research was supported by grants from the NIH to N.S. and T.P. and the International AIDS Vaccine Initiative to N.S. The authors declare no competing financial interests.
View Abstract

Navigate This Article