Distinctive Properties of the 5′-Untranslated Region of Human Hsp70 mRNA

A relaxed cap-dependence of translation of the mRNA-encoding mammalian heat shock protein Hsp70 may suggest that its 5′-untranslated region (UTR) possesses an internal ribosome entry site (IRES). In this study, this possibility has been tested in transfected cells using plasmids that express dicistronic mRNAs. Using a reporter gene construct, Renilla luciferase/Photinus pyralis luciferase, we show that the 216-nt long 5′-UTR of Hsp70 mRNA acts as an IRES that directs ribosomes to the downstream start codon by a cap-independent mechanism. The relative activity of this IRES (100-fold over the empty vector) is similar to that of the classical picornaviral IRESs. Additional controls indicate that this high expression of the downstream reporter is not due to readthrough from the upstream cistron, nor is it due to translation of cryptic monocistronic transcripts. The effect of small deletions within the 5′-UTR of Hsp70 mRNA on the IRES activity varies in dependence on their position within the 5′-UTR sequence. With the exception of deletion of nt 33–50, it is small for the 5′-terminal half of the 5′-UTR and rather strong for the 3′-terminal section. However, neither of these small deletions abolishes the IRES activity completely. Excision of larger sections (>50 nt) by truncation of the 5′-UTR from the 5′-end or by internal deleting results in a dramatic impairment of the IRES function. Taken together, these data suggest that the IRES activity of the 5′-UTR of Hsp70 mRNA requires integrity of almost the entire sequence of the 5′-UTR. The data are discussed in terms of a model that allows a three-dimensional rather than linear mode of selection of the initiation region surrounding the start codon of Hsp70 mRNA. A relaxed cap-dependence of translation of the mRNA-encoding mammalian heat shock protein Hsp70 may suggest that its 5′-untranslated region (UTR) possesses an internal ribosome entry site (IRES). In this study, this possibility has been tested in transfected cells using plasmids that express dicistronic mRNAs. Using a reporter gene construct, Renilla luciferase/Photinus pyralis luciferase, we show that the 216-nt long 5′-UTR of Hsp70 mRNA acts as an IRES that directs ribosomes to the downstream start codon by a cap-independent mechanism. The relative activity of this IRES (100-fold over the empty vector) is similar to that of the classical picornaviral IRESs. Additional controls indicate that this high expression of the downstream reporter is not due to readthrough from the upstream cistron, nor is it due to translation of cryptic monocistronic transcripts. The effect of small deletions within the 5′-UTR of Hsp70 mRNA on the IRES activity varies in dependence on their position within the 5′-UTR sequence. With the exception of deletion of nt 33–50, it is small for the 5′-terminal half of the 5′-UTR and rather strong for the 3′-terminal section. However, neither of these small deletions abolishes the IRES activity completely. Excision of larger sections (>50 nt) by truncation of the 5′-UTR from the 5′-end or by internal deleting results in a dramatic impairment of the IRES function. Taken together, these data suggest that the IRES activity of the 5′-UTR of Hsp70 mRNA requires integrity of almost the entire sequence of the 5′-UTR. The data are discussed in terms of a model that allows a three-dimensional rather than linear mode of selection of the initiation region surrounding the start codon of Hsp70 mRNA. At least three mechanisms are thought to exist for eukaryotic translation initiation, namely 5′→ 3′ scanning, “shunting,” and internal ribosomal entry. In the scanning model (1Kozak M. Annu. Rev. Cell Biol. 1992; 8: 197-225Crossref PubMed Scopus (418) Google Scholar), the small ribosomal subunit binds to the 5′-end and is believed to subsequently travel in a 5′→ 3′ direction until it encounters an initiator AUG codon in a favorable context. The shunting mechanism (2Futterer J. Kiss-Laszlo Z. Hohn T. Cell. 1993; 73: 789-802Abstract Full Text PDF PubMed Scopus (179) Google Scholar, 3Yueh A. Schneider R.J. Genes Dev. 1996; 10: 1557-1567Crossref PubMed Scopus (170) Google Scholar) is also 5′-terminus-dependent but differs from scanning in that the 40 S subunit bypasses part of the 5′-UTR by “jumping” to a region at or near the authentic initiation codon. Translation initiation by internal ribosome entry involves the binding of the 40 S ribosomal subunit to an internal ribosome entry site (IRES) 1The abbreviations used are: IRES, internal ribosomal entry site; UTR, untranslated region; EMCV, encephalomyocarditis virus; HRV (hrv), human rhinovirus; nt, nucleotides; eIF, eukaryotic initiation factor; Rluc, Renilla luciferase; Fluc, firefly luciferase; PIPES, 1,4-piperazinediethanesulfonic acid. at or near the authentic AUG, thereby eliminating the requirement for scanning through the greater part of the 5′-UTR (for recent reviews see Refs. 4Hellen C.U. Sarnow P. Genes Dev. 2001; 15: 1593-1612Crossref PubMed Scopus (806) Google Scholar and 5Vagner S. Galy B. Pyronnet S. EMBO Rep. 2001; 2: 893-898Crossref PubMed Scopus (236) Google Scholar). Whether these three modes of translational initiation are mechanistically quite different or else they are just versions of the same mechanism of eukaryotic translational initiation is not clear. IRESs were originally identified in the 5′-UTRs of picornaviral RNAs where these complex structural elements allow ribosomes to enter at a considerable distance from the 5′-end of the viral mRNA. Later, IRESs were also identified within the 5′-UTRs of cellular mRNAs. The list of cellular IRESs is constantly growing (5Vagner S. Galy B. Pyronnet S. EMBO Rep. 2001; 2: 893-898Crossref PubMed Scopus (236) Google Scholar), giving an impression that every long and structured 5′-UTR of eukaryotic mRNAs may harbor an IRES or employ the shunting model, whereas a “purely” scanning mechanism may operate only for mRNAs with short and unstructured 5′-UTRs. The cellular IRESs, as a rule, are also complex and highly organized structures. Of those known to date, the only cellular IRES that structurally stands by itself is that from the mRNA-encoding human immunoglobulin heavy chain binding protein (BiP) (6Masejak D.G. Sarnow P. Nature. 1991; 353: 90-94Crossref PubMed Scopus (423) Google Scholar). The attempts to experimentally identify a specific structure within the relatively short 5′-UTR of BiP mRNA (210 nt) have been unsuccessful (7Yang Q. Sarnow P. Nucleic Acids Res. 1997; 25: 2800-2807Crossref PubMed Scopus (85) Google Scholar). Even more surprising is that the 5′-UTR of mammalian Hsp70 mRNA has been reported to be devoid of IRES properties (8Yueh A. Schneider R.J. Genes Dev. 2000; 14: 414-421PubMed Google Scholar, 9Vivinus S. Baulande S. van Zanten M. Campbell F. Topley P. Ellis J.H. Dessen P. Coste H. Eur. J. Biochem. 2001; 268: 1908-1917Crossref PubMed Scopus (46) Google Scholar), despite the fact that its length, GC-content, and relaxed cap-dependence (10Duncan R. Hershey J.W.B. J. Biol. Chem. 1984; 259: 11882-11889Abstract Full Text PDF PubMed Google Scholar, 11Joshi-Barve S. De Benedetti A. Rhoads R.E. J. Biol. Chem. 1992; 267: 21038-21043Abstract Full Text PDF PubMed Google Scholar) are similar to the 5′-UTR of BiP mRNA, and the corresponding mRNAs encode proteins with related functions (chaperones). The high G+C content of the 5′-UTR of mammalian Hsp70 mRNA seems more compatible with operation via an IRES than by means of the classical scanning mechanism. The Drosophila homologue of the mammalian Hsp70 mRNA has been studied much more extensively. However, the results obtained for its 5′-UTR may be extended to the mammalian Hsp70 mRNA only with great caution. Indeed, the 5′-UTR of the former is strikingly enriched in adenylic residues (>50%), which should greatly facilitate the scanning process. Furthermore, a higher discrimination in translation between Hsp and normal mRNAs is observed in Drosophila as compared with mammalian cells (12Duncan R.F. Hershey J.W.B. Mathews B. Sonenberg N. Translational Control. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York1996: 271-294Google Scholar). Nevertheless, like the Drosophila homolog, the mammalian Hsp70 mRNA demonstrates a reduced dependence on eIF4F, a principal initiation factor that recognizes the cap at the 5′-end of mRNA and initiates the search for the authentic initiation codon. Under heat shock conditions and soon after heat shock, the mammalian cap-binding initiation factor eIF4E is impaired (13Paniers R. Henshaw E.C. Eur. J. Biochem. 1984; 140: 209-214Crossref PubMed Scopus (31) Google Scholar, 14Lamphear B.J. Panniers R. J. Biol. Chem. 1991; 266: 2789-2794Abstract Full Text PDF PubMed Google Scholar, 15Panniers R. Stewart E.B. Merrick W.C. Henshaw E.C. J. Biol. Chem. 1985; 260: 9648-9653Abstract Full Text PDF PubMed Google Scholar, 16Duncan R. Hershey J.W.B. Mol. Cell. Biol. 1987; 7: 1293-1295Crossref PubMed Scopus (29) Google Scholar, 17Duncan R. Hershey J.W.B. J. Cell Biol. 1989; 109: 1467-1481Crossref PubMed Scopus (153) Google Scholar) and the abundance of eIF4F complexes is reduced (14Lamphear B.J. Panniers R. J. Biol. Chem. 1991; 266: 2789-2794Abstract Full Text PDF PubMed Google Scholar). In addition, 4E-BPs, repressors of eIF4E, become activated under these conditions due to their hypophosphorylation (18Feigenblum D. Schneider R.J. Mol. Cell. Biol. 1996; 16: 5450-5457Crossref PubMed Scopus (67) Google Scholar, 19Vries R.G. Flynn A. Patel J.C. Wang X. Denton R.M. Proud C.G. J. Biol. Chem. 1997; 272: 32779-32784Abstract Full Text Full Text PDF PubMed Scopus (60) Google Scholar). Thus, like mRNAs encoding other heat shock proteins, the translational activity of human Hsp70 mRNA is resistant to changes in activity of eIF4F. The relaxed cap-dependence of the 5′-UTR of human Hsp70 mRNA along with the current lack of evidence for an IRES within this 5′-UTR leaves open the possibility of translation initiation occurring via a shunting mechanism. This is usually revealed by a low sensitivity of translation initiation to insertion of a stable hairpin near the 3′-end of 5′-UTR. Indeed, insertion of a stable hairpin into the 3′-end of the 5′-UTR of human Hsp70 mRNA still allows the mRNA to direct translation of a reporter cistron. In addition, two short sequences complementary to a 3′-terminal hairpin of 18 S rRNA have been identified at positions 96–102 and 194–205 of this 5′-UTR. These sequences have been postulated to participate in a primary binding of the 40 S ribosomal subunit followed by its shunting to the authentic AUG codon (8Yueh A. Schneider R.J. Genes Dev. 2000; 14: 414-421PubMed Google Scholar). However, insertion of the 3′-hairpin showed a large suppression (3-fold) of translation directed by the modified 5′-UTR, whereas, on the contrary, deletion of the sequences complementary to 18 S rRNA did not abolish completely the initiation activity of the 5′-UTR of Hsp70 mRNA (8Yueh A. Schneider R.J. Genes Dev. 2000; 14: 414-421PubMed Google Scholar). In addition, it is not immediately clear why a 5′-UTR that uses for ribosomal binding sequences complementary to 18 S rRNA is not capable of internal initiation. Given incompleteness of current data on the translation initiation of mammalian Hsp70 mRNA, we have reinvestigated the issue using plasmids that express dicistronic mRNAs, a conventional approach to test eukaryotic 5′-UTRs for the presence of an IRES. Here we show that the 5′-UTR of human Hsp70 mRNA represents an IRES with the relative activity similar to that of the classical picornaviral IRESs. However, unlike many other IRES-elements, the activity of the Hsp70 mRNA IRES requires the integrity of almost the entire sequence of the 5′-UTR. Plasmid Constructs—The initial dicistronic vector pGL3R containing reporter genes for Renilla luciferase (first cistron, Rluc) and Photinus pyralis firefly luciferase (second cistron, Fluc) M. 16: PubMed Scopus Google Scholar). This vector an with the for and that has the 5′-UTR of Hsp70 mRNA in the a with the 5′-UTR of Hsp70 mRNA and nt of the region of the between the with and in the The from the dicistronic by the and the Hsp70 5′-UTR in the site to a site using the with the in The with and a of the nt of gene and the nt of the 5′-UTR of Hsp70 mRNA by from using the and The with with and into which with and with plasmids and of the gene and a part of the Hsp70 mRNA were by with from using the and the and These were with of a part of the gene and a part of the 5′-UTR of Hsp70 mRNA were by from using the and and for and These were with The corresponding and were between the and in in The dicistronic of the were plasmids and and of the 5′-UTR of Hsp70 mRNA were by from using the for and for and These were with and and between the and in with and with and a the gene and nt of the 5′-UTR of Hsp70 mRNA by from using the and This with and and between the same in with and with and The containing the IRES from human in the region of pGL3R as as the with a the were a from A. The used to the with the 5′-UTR of Hsp70 mRNA in the position a containing a of the IRES with a deletion of nt by from Nucleic Acids Res. 1991; PubMed Scopus (46) Google Scholar) using the and The with and with and into pGL3R with and The the as in B. Mol. Cell. Biol. PubMed Scopus Google cellular from transfected cells by the the from using and This into site of A using with an The were with and on a The with cellular of in of and containing the and at for and containing of were and the at for of and of the at for and to and with rRNA as a The were and on a The and were by were using which were by from a on the vector containing the gene under of the to this with or and or cells were on a in modified with The cells were transfected using the as R.M. in New Scholar), with of pGL3R and of as a were after luciferase expression using the luciferase and expression with the were for a on a the in were by luciferase activity to were in on three to of a by the 5′-UTR of Hsp70 mRNA in a the 5′-UTR of Hsp70 mRNA an internal ribosome entry a dicistronic reporter The 5′-UTR of Hsp70 mRNA between the two of the dicistronic vector pGL3R M. 16: PubMed Scopus Google Scholar) The vector the and firefly as the and controls for the expression of the cistron, the highly structured IRES with a site Nucleic Acids Res. 1991; PubMed Scopus (46) Google Scholar, J. Biol. Chem. Full Text Full Text PDF PubMed Scopus Google Scholar) used along with the initial vector a for the IRES a with the human IRES in the position of the vector The activity of in transfected human cells after and the activity of firefly luciferase (second to that of Renilla luciferase (first to for the downstream expression that were of in the translation of the downstream in cells is over after insertion of the 5′-UTR of human Hsp70 mRNA into the and the activity of the IRES. the IRES activity for the IRES. The IRES properties of the Hsp70 5′-UTR are not by the of This is from of the in cells with that of human cells the IRES activity of the 5′-UTR of Hsp70 mRNA is in cells than in the relative of the Hsp70 IRES to the HRV IRES the IRES of the 5′-UTR of Hsp70 mRNA due to or the of or in the that the internal ribosome entry not observed due to ribosomal readthrough or the presence of cryptic or in the dicistronic three were In the vector M. 16: PubMed Scopus Google Scholar) This vector is to dicistronic pGL3R but a structure with a to in the of the 5′-UTR of Renilla cistron. The effect of this insertion on translational for the and The Renilla and firefly luciferase were to from a and relative to those obtained from a similar but the hairpin is that the hairpin the translation of the reporter whereas translation of the In the to that only dicistronic from the dicistronic A the 3′-end of Renilla luciferase the region and the of firefly luciferase reporter nt) to mRNA from cells that been transfected or transfected with the vector with the Hsp70 5′-UTR in the position followed by The data in show that the to the In the to that is cryptic within the reporter gene or within the the of as in B. Mol. Cell. Biol. PubMed Scopus Google Scholar, and cells were transfected with the activity of Renilla or firefly in cells not Taken these results evidence that the 5′-UTR of human Hsp70 mRNA an IRES with a strong translation initiation the Hsp70 the position of the IRES within the 5′-UTR of Hsp70 mRNA, a of dicistronic containing deletions of different in the 5′-UTR of Hsp70 mRNA between the two The plasmids were transfected into cells and the relative firefly luciferase activity for of these from the IRES activity of the with relatively small deletions varies in dependence on their position within the 5′-UTR sequence. With the exception of deletion of nt the effect is small for the 5′-terminal half of the 5′-UTR of the did not the IRES and much for the 3′-terminal section. The dramatic effect for However, neither of these deletions completely the IRES activity of Hsp70 mRNA. This with the effect of small deletions or on viral IRESs, at least those to data Nucleic Acids Res. 1991; PubMed Scopus (46) Google Scholar, Sarnow P. A. J. PubMed Google Scholar, Mol. Cell. Biol. 1996; 16: PubMed Scopus Google Scholar, S. J. 1997; PubMed Google Scholar, R.J. Genes Dev. PubMed Scopus Google Scholar, S. R.M. J. 73: PubMed Google Scholar, PubMed Scopus Google Scholar, S. 2000; PubMed Scopus Google Scholar, R. N. J. 2000; PubMed Scopus Google Scholar). of completely abolish their translation initiation activity for the data for of the IRES in of deletions within the 5′-UTR of Hsp70 mRNA on its IRES The relative were as and as a of the activity for the Hsp70 IRES which at A represents the effect of whereas demonstrates the of larger IRES of the 5′-UTR of Hsp70 mRNA of the of the data in the may the impression that the IRES of Hsp70 mRNA is in the 3′-terminal half of the 5′-UTR and the 5′-terminal half is this is the of the 5′-terminal part of the 5′-UTR of Hsp70 mRNA should not the IRES The results in show that this is not the of just the nt from the 5′-UTR in the which of the entire of the 5′-UTR, to a dramatic of the IRES Even larger truncation of the 5′-UTR from the 5′-end the IRES and in which is elements from the 3′-terminal half it a internal deletions larger than those used in the in almost the IRES may that the 3′-terminal half of the 5′-UTR of Hsp70 mRNA is more to almost the entire sequence of the 5′-UTR of Hsp70 mRNA is for the IRES The data to a specific by the of Hsp70 mRNA where not of its sequence to a different to a activity in the cap-independent mode of translation initiation. This on the mechanism that allows Hsp70 mRNA to be the cap under normal Thus, the of dicistronic mRNAs in this with the of Hsp70 mRNA in the completely of translation of the reporter sequence directed by the 5′-UTR of Hsp70 mRNA under normal of the of Hsp70 mRNA that it from Hsp70 mRNA from Drosophila is its This it compatible with the of the classical scanning model for mRNAs and of an of translation initiation. Here we evidence that the of Hsp70 mRNA possesses a strong IRES with that of picornaviral However, unlike many other viral and cellular mRNAs studied to where IRESs an internal position within their long the IRES of Hsp70 mRNA to be by almost the entire of this mRNA. large of the 5′-UTR be a dramatic effect on the expression of the cistron. This with data of S. Baulande S. van Zanten M. Campbell F. Topley P. Ellis J.H. Dessen P. Coste H. Eur. J. Biochem. 2001; 268: 1908-1917Crossref PubMed Scopus (46) Google Scholar), have the of the integrity of Hsp70 5′-UTR for its effect on translation of reporter genes in monocistronic should be that data obtained by deletion should be with great in the structure of the corresponding region is and the are not Excision of a sequence may a of the structure of the corresponding and the structure may have in with the initial a fact that is in many on and of This is also to the in this deletion of sequences or a effect on the IRES activity of Hsp70 mRNA. These deletions nt 96–102 and that have been postulated to be in complementary with 18 S rRNA (8Yueh A. Schneider R.J. Genes Dev. 2000; 14: 414-421PubMed Google Scholar). In data only indicate that these sequences are of a larger sequence the IRES of the 5′-UTR of Hsp70 mRNA. The effect of their may be for by impairment of the of the 40 S ribosomal subunit with this 5′-UTR and a specific structure in its much structural is to between these two Nevertheless, into these may as to the IRES of Hsp70 mRNA is The viral IRESs to data and highly structural C.U. Sarnow P. Genes Dev. 2001; 15: 1593-1612Crossref PubMed Scopus (806) Google Scholar). These highly specific binding for initiation proteins, or the 40 S ribosomal subunit C.U. Sarnow P. Genes Dev. 2001; 15: 1593-1612Crossref PubMed Scopus (806) Google Scholar). is why short deletions or in many of these IRESs are to completely Nucleic Acids Res. 1991; PubMed Scopus (46) Google Scholar, Sarnow P. A. J. PubMed Google Scholar, Mol. Cell. Biol. 1996; 16: PubMed Scopus Google Scholar, S. J. 1997; PubMed Google Scholar, R.J. Genes Dev. PubMed Scopus Google Scholar, S. R.M. J. 73: PubMed Google Scholar, PubMed Scopus Google Scholar, S. 2000; PubMed Scopus Google Scholar, R. N. J. 2000; PubMed Scopus Google Scholar). This not to be the for the IRES of Hsp70 mRNA. With the exception of deletions within this IRES at strong rather than dramatic This that it a relatively with of binding for mRNA translational that the mode of cap-independent of the scanning is similar for the 5′-terminal and positions of the Hsp70 mRNA 5′-UTR, we suggest a model for the cap-independent of the Hsp70 mRNA the 40 S ribosome In this model, the mRNA binds to the of the Hsp70 which its and 3′-terminal as The binding in a that a three-dimensional of the 5′-UTR of Hsp70 mRNA a of the sequence surrounding the start codon to the scanning large deletion of the internal or of the IRES binding of the mRNA or the of the IRES. is a of the three-dimensional by a part of of the sequence surrounding the initiation codon. This is the for the IRES. Here the mRNA to have their binding on the IRES J. Biol. Chem. Full Text Full Text PDF PubMed Scopus Google Scholar, A. M. J. PubMed Google Scholar) and not to The IRES seems to a specific three-dimensional that allows selection of the initiation only within a A. R.J. EMBO J. PubMed Scopus Google Scholar, T. J. Mol. Biol. PubMed Scopus Google Scholar). should be that the translation initiation not three-dimensional of the initiation In that a more of eIF4F with an mRNA the only facilitate a three-dimensional This model differs from the shunting mechanism of translation initiation of mammalian Hsp70 mRNA that has been (8Yueh A. Schneider R.J. Genes Dev. 2000; 14: 414-421PubMed Google Scholar). neither binding of the mRNA at the cap nor a scanning of the sequence of the of an mRNA followed by a to the initiation codon. The model also why it is not to by the conventional approach for of shunting This approach is known to be on insertion of a hairpin structure between the 5′-end of an mRNA and its initiation codon. A small effect of on translational activity is as evidence in of the ribosomal However, it is to an of the 5′-UTR relative positions of the scanning and the initiation codon. These may to why insertion of a stable hairpin in the position of the 5′-UTR of mRNA results in a considerable of the translational activity (8Yueh A. Schneider R.J. Genes Dev. 2000; 14: 414-421PubMed Google Scholar). The classical scanning model (1Kozak M. Annu. Rev. Cell Biol. 1992; 8: 197-225Crossref PubMed Scopus (418) Google Scholar) is should be that it tested only for 5′-UTRs did not of the a three-dimensional of selection of the sequence near the initiation codon by initiation on the 40 S ribosomal Thus, we that the model is not only to the of the Hsp70 IRES but may have a more a of the scanning to the initiation codon may not be the only principal the and of translation initiation. A and a of the sequence surrounding the initiation codon should greatly of the process. mRNAs used in with their unstructured and relatively short three a low and a are giving a strong for the initiation are that the model of translation initiation discussed in this is However, we that it may to be in why the of the IRESs has been identified in and highly structured of mammalian mRNAs, and the of cellular IRESs is and they may be for with plasmids and for of the and are to and for to their also for