Portal de Periódicos da CAPES

Major Histocompatibility Class I Molecules Can Present Cryptic Translation Products to T-cells

1995; Elsevier BV; Volume: 270; Issue: 3 Linguagem: Inglês

10.1074/jbc.270.3.1088

1083-351X

Nilabh Shastri, Vu Thuong Nguyen, Federico González,

T-cell and B-cell Immunology

Self or foreign cellular proteins provide peptides for presentation by major histocompatibility complex (MHC) class I molecules on the surface of antigen presenting cells (APC). Surprisingly, several studies have shown that T-cells can recognize APC transfected with antigen genes that were not present in the appropriate translational context. To understand the basis of this phenomenon, APC were transfected with DNA constructs encoding the OVA257-264 (SL8) peptide, but with varying translation initiation codons. We report that, in addition to ATG, 6 other codons (ATT, ACG, CTG, GCG, TGG, GAT) also allowed presentation to SL8•Kb-specific T-cells. Significantly, this set includes 3 of 4 known non-ATG translation initiation codons strongly suggesting that cryptic translation accounts for this phenomenon. Although expression of the SL8•Kb complex was readily detected by T-cell activation, the amount of processed peptides was below detection limit (<30 copies/cell) in cell extracts. Thus, the fortuitous presence of these cryptic translation initiation sites in transcribed genes can explain how peptide•MHC complexes were obtained in sufficient amounts for T-cell activation. The translation initiation codons identified here could also be useful for identifying potential open reading frames that possess biological and/or immunological activities. Self or foreign cellular proteins provide peptides for presentation by major histocompatibility complex (MHC) class I molecules on the surface of antigen presenting cells (APC). Surprisingly, several studies have shown that T-cells can recognize APC transfected with antigen genes that were not present in the appropriate translational context. To understand the basis of this phenomenon, APC were transfected with DNA constructs encoding the OVA257-264 (SL8) peptide, but with varying translation initiation codons. We report that, in addition to ATG, 6 other codons (ATT, ACG, CTG, GCG, TGG, GAT) also allowed presentation to SL8•Kb-specific T-cells. Significantly, this set includes 3 of 4 known non-ATG translation initiation codons strongly suggesting that cryptic translation accounts for this phenomenon. Although expression of the SL8•Kb complex was readily detected by T-cell activation, the amount of processed peptides was below detection limit ( 2× over background), and 20 with the highest activity were selected for further analysis. Nucleotide sequences of these recombinants revealed seven different codons at the X(NNN) position, ATG (7Jardetzky T.S. Lane W.S. Robinson R.A. Madden D.R. Wiley D.C. Nature. 1991; 353: 326-329Crossref PubMed Scopus (733) Google Scholar), CTG(5Falk K. Rotzschke O. Stevanovic S. Jung G. Rammensee H.-G. Nature. 1991; 351: 290-296Crossref PubMed Scopus (2097) Google Scholar), ATT(3Rammensee H.-G. Falk K. Rotzschke O. Annu. Rev. Immunol. 1993; 11: 213-244Crossref PubMed Scopus (708) Google Scholar), TGG(2Yewdell J.W. Bennink J.R. Adv. Immunol. 1992; 52: 1-123Crossref PubMed Scopus (362) Google Scholar), ACG(1Brodsky F.M. Guagliardi L.E. Annu. Rev. Immunol. 1991; 9: 707-744Crossref PubMed Scopus (318) Google Scholar), GCG(1Brodsky F.M. Guagliardi L.E. Annu. Rev. Immunol. 1991; 9: 707-744Crossref PubMed Scopus (318) Google Scholar), and GAT (1Brodsky F.M. Guagliardi L.E. Annu. Rev. Immunol. 1991; 9: 707-744Crossref PubMed Scopus (318) Google Scholar) (Fig. 1b). The frequency (7/367 = 1:53) of recombinants containing ATG was close to the theoretical frequency (1:64 codons) and thus validates the random representation of X codons tested in this screen. In addition, four other randomly picked recombinants that did not score positive in the screen were selected as negative controls.The relative efficiency of generating the SL8•Kb complex from these seven constructs was determined by transfecting DNA into Kb-COS cells and then measuring their ability to stimulate SL8•Kb-specific B3Z T-cells. Among these recombinants, the construct containing ATG as the translation initiation codon was clearly the most efficient (Fig. 2b). When compared at the optimal DNA concentration (100-300 ng/well), the T-cell stimulatory activity generated by the non-ATG constructs was within 2-10-fold that of the ATG construct and was within 10-100-fold when compared at half-maximal response. By contrast to these seven constructs, negligible T-cell stimulation was obtained with constructs containing CTA, TCT, GAA, or TTG as the X translation initiation codons (Fig. 2, b and c). These results clearly demonstrate that the usage of initiation codons in translation was not random. The seven codons identified by this assay included, besides ATG, 3 of the 4 (CTG, ATT, ACG, GTG) known non-ATG translational initiation codons(20Kozak M. J. Cell Biol. 1991; 266: 19867-19870Google Scholar, 21Mehdi H. Ono E. Gupta K.C. Gene (Amst.). 1990; 91: 173-178Crossref PubMed Scopus (80) Google Scholar). To rule out the possibility that the fourth known initiator codon, GTG, could have been used but may have been missed among the recombinants screened, a construct encoding GTG at the X position was prepared with synthetic oligonucleotides (Fig. 2a). This construct was transfected into Kb-COS cells, but, like the other negative control constructs, it was also inactive, further validating the screening strategy and emphasizing the non-random usage of X codons (Fig. 2c). Moreover, this result also indicates that relative efficiency differences exist among the non-ATG translation initiation codons. We conclude that Kb-MHC can present endogenous SL8 peptide synthesized via cryptic non-ATG translation initiation codons.We next assessed whether the translation initiation codons identified by the transient expression assay were also active when the DNAs were stably integrated. Stable transfectants were generated with murine Kb L-cells with these seven constructs. To control for possible differences in gene expression due to variable integration sites, pools ( ∼50-100 colonies) of G418-resistant cells were tested for their ability to stimulate SL8•Kb-specific B3Z T-cells. As a further internal control, the same transfectants were also tested in the same experiment with another T-cell hybrid, 27.5Z, that recognizes an endogenous and constitutively expressed peptide presented by Kb-MHC(17Sanderson S. Shastri N. Int. Immunol. 1994; 6: 369-376Crossref PubMed Scopus (316) Google Scholar, 18Rotzschke O. Falk K. Faath S. Rammensee H.-G. J. Exp. Med. 1991; 174: 1059-1071Crossref PubMed Scopus (205) Google Scholar). Transfectants with each of the seven constructs generated the SL8•Kb complex, while the parental Ltk - and Kb L-cells were incapable of stimulating B3Z T-cells (Fig. 3, a and b). Again, transfectants with the ATG construct were the most stimulatory, and relative differences in presentation with transfectants expressing the non-ATG constructs were similar to those obtained in transient transfections. The reasons for the somewhat different hierarchy of the non-ATG codons in stable versus transient transfectants are not clear, but may be due to species (murine versus simian) or to cell type (fibroblast versus kidney epithelium) differences. These relative differences in presentation efficiency were, however, limited to expression of the SL8•Kb complex, because the same transfectant cells yielded superimposable response profiles with Kb-restricted 27.5Z T-cells (Fig. 3, c and d). This result confirmed that Kb-MHC can present cryptic translation products derived from either transiently or stably expressed genes. The fortuitous presence of one or more of these alternate initiation codons upstream of the antigen coding sequences can also now satisfactorily explain how antigen presentation activity could have been obtained in our own (15Shastri N. Gonzalez F. J. Immunol. 1993; 150: 2724-2736PubMed Google Scholar) and in other previous studies (11Fetten J.V. Roy N. Gilboa E. J. Immunol. 1991; 147: 2697-2705PubMed Google Scholar, 13Lurquin C. Van Pel A. Mariamé B. De Plaen E. Szikora J.-P. Janssens C. Reddehase M.J. Lejeune J. Boon T. Cell. 1989; 58: 293-303Abstract Full Text PDF PubMed Scopus (256) Google Scholar, 14Boon T. Van Pel A. De Plaen E. Chomez P. Lurquin C. Szikora J.-P. Sibille C. Mariamé B. Van Den Eynde B. Lethe B. Brichart V. Cold Spring Harbor Symp. Quant Biol. 1989; 54: 587-596Crossref PubMed Google Scholar).Figure 3:Mouse Kb L-cells, stably transfected with indicated XGSL8 constructs, stimulate SL8•Kb-specific B3Z T-cells. The nucleotide triplets shown represent the X translational initiation codon. Varying numbers of parental Ltk -, Kb L-cells (K89), or G418r cells obtained by transfecting Kb L-cells with the indicated XGSL8 plasmids, were co-cultured overnight with 1 × 105 B3Z (a and b) or unknown peptide•Kb-specific 27.5Z (c and d) T-cells. Antigen-specific response was measured by the induced lacZ activity in the T-cells as in the legend to Fig. 1.View Large Image Figure ViewerDownload Hi-res image Download (PPT)The high SL8•Kb expression in APC transfected with non-ATG constructs was surprising. It was possible that, despite predominance of ATG as the most efficient translation initiation codon known(19Kozak M. J. Biol. Chem. 1989; 108: 229-241Google Scholar, 20Kozak M. J. Cell Biol. 1991; 266: 19867-19870Google Scholar), significant amounts of processed SL8 peptide may have been generated in cells expressing the non-ATG constructs. To directly measure the amount of processed peptides in cells, we assayed trifluoroacetic acid extracts prepared from the stable L-cell transfectants (Fig. 4). High levels of processed peptide were present in extracts of transfectants expressing the ATG construct. However, extracts of transfectants expressing either of the six non-ATG constructs were completely inactive (>300-fold lower yield). This result directly shows that the amounts of the SL8 gene product present in cells was profoundly influenced by the identity of the translation initiation codon, with ATG being by far the most effective. Nevertheless, despite the lack of detectable peptide in the extracts, APC transfected with non-ATG constructs presented the SL8•Kb complex in amounts sufficient to stimulate T-cells (Fig. 3). Due to the complexity of these trifluoroacetic acid extracts (4Van Bleek G.M. Nathenson S.G. Nature. 1990; 348: 213-216Crossref PubMed Scopus (587) Google Scholar, 5Falk K. Rotzschke O. Stevanovic S. Jung G. Rammensee H.-G. Nature. 1991; 351: 290-296Crossref PubMed Scopus (2097) Google Scholar, 6Hunt D.F. Henderson R.A. Shabanowitz J. Sakaguchi K. Michel H. Sevilir N. Cox A.L. Appella E. Engelhard V.H. Science. 1992; 255: 1261-1263Crossref PubMed Scopus (1043) Google Scholar), the exact peptide amounts are difficult to assess accurately. However, using synthetic SL8 peptide as a standard (sensitivity ∼1-3 pM)(15Shastri N. Gonzalez F. J. Immunol. 1993; 150: 2724-2736PubMed Google Scholar), and assuming quantitative recovery, we estimate that the amount of processed SL8 peptide obtained from non-ATG constructs was equal to or less than 30 SL8 copies/cell. This estimate sets the threshold of peptide•MHC required for T-cell stimulation within a magnitude of the theoretical limit of one peptide•MHC complex on the APC surface. Furthermore, this result directly demonstrates that T-cell assay is an extraordinarily sensitive measure of gene expression in APC.Figure 4:Quantitation of processed peptides extracted from stable Kb L-cells transfected with XGSL8 constructs. Processed SL8 peptides are detected only in extracts of cells expressing the ATG construct. Serial dilutions of trifluoroacetic acid extracts were prepared as described under "Materials and Methods" and were assayed using B3Z T-cells and untransfected Kb L-cells as APC.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Why do cells initiate translation at non-ATG codons? Note that although the fidelity of translation initiation in vitro is often less stringent and was the basis for the discovery of the genetic code, translation initiation in living cells virtually always occurs at the ATG codon(22Kozak M. Nucleic Acids Res. 1987; 15: 8125-8148Crossref PubMed Scopus (4151) Google Scholar). Examples of rare exceptions to this rule that allowed the discovery of non-ATG codons include transcription and growth factors, as well as viral proteins of unknown function(21Mehdi H. Ono E. Gupta K.C. Gene (Amst.). 1990; 91: 173-178Crossref PubMed Scopus (80) Google Scholar, 23Taira M. Iizasa T. Shimada H. Kudoh J. Shimizu N. Tatibana M. J. Biol. Chem. 1990; 265 (16397): 16491Abstract Full Text PDF PubMed Google Scholar, 24Florkiewicz R.Z. Sommer A. Proc. Natl. Acad. Sci. U. S. A. 1989; 86: 3978-3981Crossref PubMed Scopus (444) Google Scholar, 25Xiao J.H. Davidson I. Matthes H. Garnier J.-M. Chambon P. Cell. 1991; 65: 551-568Abstract Full Text PDF PubMed Scopus (301) Google Scholar, 26Kingsley C. Winoto A. Mol. Cell. Biol. 1992; 12: 4251-4261Crossref PubMed Scopus (487) Google Scholar). One possible explanation proposed is that these genes use non-ATG initiation codons, despite their inefficient translation capacity(21Mehdi H. Ono E. Gupta K.C. Gene (Amst.). 1990; 91: 173-178Crossref PubMed Scopus (80) Google Scholar), because expression of these potent gene products must be kept extremely low(20Kozak M. J. Cell Biol. 1991; 266: 19867-19870Google Scholar). Thus, while the occurrence of cryptic translation is firmly established, its mechanism and its significance for biological regulation is yet to be clarified. By contrast to these normal genes where usage of the non-ATG codons is apparently promoted by stable RNA hairpin structures that cause stalling of the ribosome(21Mehdi H. Ono E. Gupta K.C. Gene (Amst.). 1990; 91: 173-178Crossref PubMed Scopus (80) Google Scholar), no such RNA structures were apparent in the constructs used by us and by others. Thus, it is unlikely that our artificial constructs were fortuitously favored for cryptic translation. Rather, a more attractive possibility is that cryptic translation could have resulted as a consequence of the high levels of constitutive transcription from viral promoters widely used in expression vectors. Whether similar conditions are possible for normal genes and whether any of the thousands of peptides displayed by MHC include cryptic translation products is not yet known. Conceivably, expression of viral genes in infected cells could be similar to the conditions tested in our experiments.Regardless of the mechanism of cryptic translation, the ability of MHC molecules to present vanishingly small amounts of endogenous peptides is relevant to the efficiency of the immune surveillance mechanism. Ideal surveillance of viral infection or of transformation events within cells requires that all cellular proteins be available as peptide•MHC for recognition by T-cells. However, MHC are constrained to present only consensus motif-bearing peptides(3Rammensee H.-G. Falk K. Rotzschke O. Annu. Rev. Immunol. 1993; 11: 213-244Crossref PubMed Scopus (708) Google Scholar), of which only a small subset are apparently used(27Lipford G.B. Hoffman M. Wagner H. Heeg K. J. Immunol. 1993; 150: 1212-1222PubMed Google Scholar). Thus, the presentation of rare cryptic translation products by MHC molecules could be rationalized as a property of the antigen processing mechanism whose function is to ensure that the largest possible set of peptides is made available for survey by the T-cell repertoire. Indeed, the mechanism of cryptic translation and the capture of these small amounts of precursors by the antigen presentation pathway remain to be elucidated. Regardless of the mechanism, our results suggest that the identity of antigenic peptides for candidate vaccines (28Takahashi H. Takeshita T. Morein B. Putney S. Germain R.N. Berzofsky J.A. Nature. 1990; 344: 873-875Crossref PubMed Scopus (399) Google Scholar, 29Hill A.V.S. Elvin J. Willis A.C. Aidoo M. Allsopp C.E.M. Gotch F.M. Gao X.M. Takiguchi M. Greenwood B.M. Townsend A.R.M. McMichael A.J. Whittle H.C. Nature. 1992; 360: 434-439Crossref PubMed Scopus (568) Google Scholar) may not be limited only to those found within the most abundant ATG defined open reading frames. The alternate translational initiation codons identified here could prove useful in these pursuits. INTRODUCTIONMajor histocompatibility complex (MHC) 1The abbreviations used are: MHCmajor histocompatibility complexAPCantigen presenting cellOVAovalbuminNPnucleoprotein. class I molecules display peptides on the antigen presenting cell (APC) surface. CD8+ cytolytic T-cells probe these peptide•MHC complexes to identify and to subsequently eliminate cells that express foreign peptides. The fact that these peptides are derived from cellular proteins allows T-cells to detect intracellular pathogens or transformation events due to expression of new proteins (reviewed in (1Brodsky F.M. Guagliardi L.E. Annu. Rev. Immunol. 1991; 9: 707-744Crossref PubMed Scopus (318) Google Scholar, 2Yewdell J.W. Bennink J.R. Adv. Immunol. 1992; 52: 1-123Crossref PubMed Scopus (362) Google Scholar, 3Rammensee H.-G. Falk K. Rotzschke O. Annu. Rev. Immunol. 1993; 11: 213-244Crossref PubMed Scopus (708) Google Scholar).The thousands of distinct peptides that are constitutively displayed by MHC depends upon the corresponding diversity of donor proteins that provide these peptides(4Van Bleek G.M. Nathenson S.G. Nature. 1990; 348: 213-216Crossref PubMed Scopus (587) Google Scholar, 5Falk K. Rotzschke O. Stevanovic S. Jung G. Rammensee H.-G. Nature. 1991; 351: 290-296Crossref PubMed Scopus (2097) Google Scholar, 6Hunt D.F. Henderson R.A. Shabanowitz J. Sakaguchi K. Michel H. Sevilir N. Cox A.L. Appella E. Engelhard V.H. Science. 1992; 255: 1261-1263Crossref PubMed Scopus (1043) Google Scholar, 7Jardetzky T.S. Lane W.S. Robinson R.A. Madden D.R. Wiley D.C. Nature. 1991; 353: 326-329Crossref PubMed Scopus (733) Google Scholar). The mechanism by which cellular proteins are targeted to the processing pathway remains poorly understood. Some studies have proposed that these peptides are generated as by-products of normal turnover of cellular proteins(8Townsend A. Bastin J. Gould K. Brownlee G. Andrew M. Coupar B. Boyle D. Chan S. Smith G. J. Exp. Med. 1988; 168: 1211-1224Crossref PubMed Scopus (271) Google Scholar, 9Michalek M.T. Grant E.P. Gramm C. Goldberg A.L. Rock K.L. Nature. 1993; 363: 552-554Crossref PubMed Scopus (283) Google Scholar). The generality of this mechanism is, however, not clear because differences in antigen presentation activity often do not correlate with differences in stability of donor proteins(8Townsend A. Bastin J. Gould K. Brownlee G. Andrew M. Coupar B. Boyle D. Chan S. Smith G. J. Exp. Med. 1988; 168: 1211-1224Crossref PubMed Scopus (271) Google Scholar, 10Del Val M. Schlicht H.-J. Ruppert T. Reddehase M.J. Koszinowski U.H. Cell. 1991; 66: 1145-1153Abstract Full Text PDF PubMed Scopus (280) Google Scholar, 11Fetten J.V. Roy N. Gilboa E. J. Immunol. 1991; 147: 2697-2705PubMed Google Scholar). As an alternate mechansim, it has been proposed that these peptides are generated via a specialized pathway that does not depend upon protein turnover but instead involves novel transcriptional and/or translational mechanisms. Evidence in support of this latter mechanism, referred to as the "pepton hypothesis"(12Boon T. Van Pel A. Immunogenetics. 1989; 29: 75-79Crossref PubMed Scopus (108) Google Scholar), comes from several independent studies that peptide•MHC complexes are generated in APC transfected with antigen genes despite the absence of obvious promoters or translation initiation sites(11Fetten J.V. Roy N. Gilboa E. J. Immunol. 1991; 147: 2697-2705PubMed Google Scholar, 13Lurquin C. Van Pel A. Mariamé B. De Plaen E. Szikora J.-P. Janssens C. Reddehase M.J. Lejeune J. Boon T. Cell. 1989; 58: 293-303Abstract Full Text PDF PubMed Scopus (256) Google Scholar, 14Boon T. Van Pel A. De Plaen E. Chomez P. Lurquin C. Szikora J.-P. Sibille C. Mariamé B. Van Den Eynde B. Lethe B. Brichart V. Cold Spring Harbor Symp. Quant Biol. 1989; 54: 587-596Crossref PubMed Google Scholar, 15Shastri N. Gonzalez F. J. Immunol. 1993; 150: 2724-2736PubMed Google Scholar). Again, which one or more of the several possible mechanisms that can account for this phenomenon has not been yet determined.Our previous study confirmed that presentation of either ovalbumin (OVA)•Kb or influenza nucleoprotein (NP)•Db complexes did indeed occur in APC transfected with OVA or NP genes that were in inappropriate expression context (15Shastri N. Gonzalez F. J. Immunol. 1993; 150: 2724-2736PubMed Google Scholar). By contrast to other studies, our experiments suggested that presentation of peptide•MHC complexes was related to the translational mechanism. First, the relative differences in presentation activity correlated with particular translational reading frames of the peptide coding sequence. Second, and more significantly, the T-cell stimulating activity was completely abrogated by placing translational stops immediately upstream of the peptide coding sequence. How translation was obtained in the absence of the ATG translation initiation codon was not ascertained.In this report, we have analyzed the mechanism that accounts for the generation of these endogenous peptide•MHC complexes. We identified the cryptic translation initiation codons by screening a random set of codons that allowed presentation of the OVA257-264 (SL8)•Kb complex to T-cells. Interestingly, the seven codons thus identified include ATG, as well as three of four known non-ATG translational initiation codons. Furthermore, we show that despite the ability of these non-ATG codons to allow peptide/MHC expression, the amount of processed peptides in the cells remained below biochemical detection limits estimated to be less than 30 copies/cell. These findings explain previously surprising results and are important for understanding the mechanism of the endogenous peptide/MHC presentation pathway.