Portal de Periódicos da CAPES

Temporal Translational Control by a Metastable RNA Structure

2001; Elsevier BV; Volume: 276; Issue: 38 Linguagem: Inglês

10.1074/jbc.m105347200

1083-351X

Jakob Møller‐Jensen, Thomas Franch, Kenn Gerdes,

RNA Research and Splicing

Programmed cell death by thehok/sok locus of plasmid R1 relies on a complex translational control mechanism. The highly stablehok mRNA is activated by 3′-end exonucleolytical processing. Removal of the mRNA 3′ end releases a 5′-end sequence that triggers refolding of the mRNA. The refolded hok mRNA is translatable but can also bind the inhibitory Sok antisense RNA. Binding of Sok RNA leads to irreversible mRNA inactivation by an RNase III-dependent mechanism. A coherent model predicts that during transcription hokmRNA must be refractory to translation and antisense RNA binding. Here we provide genetic evidence for the existence of a 5′ metastable structure in hok mRNA that locks the nascent transcript in an inactive configuration in vivo. Consistently, the metastable structure reduces the rate of Sok RNA binding and completely blocks hok translation in vitro. Structural analyses of native RNAs strongly support that the 5′ metastable structure exists in the nascent transcript. Further structural analyses reveal that the mRNA 3′ end triggers refolding of the mRNA 5′ end into the more stable tac-stem conformation. These results provide a profound understanding of an unusual and intricate post-transcriptional control mechanism. Programmed cell death by thehok/sok locus of plasmid R1 relies on a complex translational control mechanism. The highly stablehok mRNA is activated by 3′-end exonucleolytical processing. Removal of the mRNA 3′ end releases a 5′-end sequence that triggers refolding of the mRNA. The refolded hok mRNA is translatable but can also bind the inhibitory Sok antisense RNA. Binding of Sok RNA leads to irreversible mRNA inactivation by an RNase III-dependent mechanism. A coherent model predicts that during transcription hokmRNA must be refractory to translation and antisense RNA binding. Here we provide genetic evidence for the existence of a 5′ metastable structure in hok mRNA that locks the nascent transcript in an inactive configuration in vivo. Consistently, the metastable structure reduces the rate of Sok RNA binding and completely blocks hok translation in vitro. Structural analyses of native RNAs strongly support that the 5′ metastable structure exists in the nascent transcript. Further structural analyses reveal that the mRNA 3′ end triggers refolding of the mRNA 5′ end into the more stable tac-stem conformation. These results provide a profound understanding of an unusual and intricate post-transcriptional control mechanism. post-segregational killing fold-back inhibition translational activator polymerase chain reaction dithiothreitol Shine-Dalgarno sequence RNA molecules fold into highly ordered structures essential to their diverse biological functions. Accordingly, the question of how the linear sequence of ribonucleotides dictates the overall folding of an RNA has received much attention (1Westhof E. Masquida B. Jaeger L. Folding Des. 1996; 1: R78-R88Abstract Full Text Full Text PDF PubMed Scopus (141) Google Scholar, 2Draper D.E. Nat. Struct. Biol. 1996; 3: 397-400Crossref PubMed Scopus (43) Google Scholar, 3Draper D.E. Trends Biochem. Sci. 1996; 21: 145-149Abstract Full Text PDF PubMed Scopus (107) Google Scholar, 4Brion P. Westhof E. Annu. Rev. Biophys. Biomol. Struct. 1997; 26: 113-137Crossref PubMed Scopus (432) Google Scholar, 5Tinoco Jr., I. Bustamante C. J. Mol. Biol. 1999; 293: 271-281Crossref PubMed Scopus (710) Google Scholar, 6Woodson S.A. Cell. Mol. Life Sci. 2000; 57: 796-808Crossref PubMed Scopus (99) Google Scholar). Folding of RNA, whether occurring from a denatured state or sequentially in concomitance with its synthesis, involves the formation of a number of hierarchically ordered intramolecular interactions leading to increasing levels of structural organization. In both cases, structural rearrangements of kinetically favored folding intermediates may occur before the thermodynamically most stable conformation is reached (7Kramer F.R. Mills D.R. Nucleic Acids Res. 1981; 9: 5109-5124Crossref PubMed Scopus (83) Google Scholar,8Pan J. Thirumalai D. Woodson S.A. J. Mol. Biol. 1997; 273: 7-13Crossref PubMed Scopus (166) Google Scholar). In the course of folding, RNA molecules face the risk of being trapped in nonequilibrium conformations. Because of the high thermodynamic stability of RNA secondary structures, rearrangement of such nonequilibrium conformations can constitute a substantial energy barrier, thus leading to kinetic trapping of the RNA in thermodynamically suboptimal conformations termed metastable structures (9Walstrum S.A. Uhlenbeck O.C. Biochemistry. 1990; 29: 10573-10576Crossref PubMed Scopus (74) Google Scholar, 10Emerick V.L. Woodson S.A. Biochemistry. 1993; 32: 14062-14067Crossref PubMed Scopus (50) Google Scholar, 11Pan J. Woodson S.A. J. Mol. Biol. 1998; 280: 597-609Crossref PubMed Scopus (167) Google Scholar). Metastable folding intermediates have proved important to a number of biological processes including plasmid replication (12Gultyaev A.P. van Batenburg F.H. Pleij C.W. Nucleic Acids Res. 1995; 23: 3718-3725Crossref PubMed Scopus (29) Google Scholar), replication of RNA by Qβ replicase (13Biebricher C.K. Luce R. EMBO J. 1992; 11: 5129-5135Crossref PubMed Scopus (110) Google Scholar, 14Fernandez A. Belinky A. Biophys. Chem. 1996; 61: 101-105Crossref PubMed Scopus (1) Google Scholar), human immunodeficiency virus-1 RNA export to the cytoplasm (15Belinky A. Fernandez A. Biophys. Chem. 1992; 42: 1-6Crossref PubMed Scopus (1) Google Scholar), ribozyme activity (16Lazinski D.W. Taylor J.M. RNA (N. Y.). 1995; 1: 225-233PubMed Google Scholar, 17Cao Y. Woodson S.A. RNA (N. Y.). 1998; 4: 901-914Crossref PubMed Scopus (21) Google Scholar, 18Matysiak M. Wrzesinski J. Ciesiolka J. J. Mol. Biol. 1999; 291: 283-294Crossref PubMed Scopus (34) Google Scholar, 19Song S.I. Silver S.L. Aulik M.A. Rasochova L. Mohan B.R. Miller W.A. J. Mol. Biol. 1999; 293: 781-793Crossref PubMed Scopus (16) Google Scholar, 20Chadalavada D.M. Knudsen S.M. Nakano S. Bevilacqua P.C. J. Mol. Biol. 2000; 301: 349-367Crossref PubMed Scopus (68) Google Scholar), and viroid replication (21Loss P. Schmitz M. Steger G. Riesner D. EMBO J. 1991; 10: 719-727Crossref PubMed Scopus (92) Google Scholar, 22Gultyaev A.P. van Batenburg F.H. Pleij C.W. J. Mol. Biol. 1998; 276: 43-55Crossref PubMed Scopus (30) Google Scholar, 23Repsilber D. Wiese S. Rachen M. Schroder A.W. Riesner D. Steger G. RNA (N. Y.). 1999; 5: 574-584Crossref PubMed Scopus (69) Google Scholar). The hok/sok locus of plasmid R1 mediates plasmid maintenance by the killing of plasmid-free cells, also termed post-segregational killing (PSK)1 (24Gerdes K. Gultyaev A.P. Franch T. Pedersen K. Mikkelsen N.D. Annu. Rev. Genet. 1997; 31: 1-31Crossref PubMed Scopus (171) Google Scholar). The PSK mechanism, which restricts synthesis of Hok toxin to newborn plasmid-free cells, is controlled entirely at the post-transcriptional level. The hok/sok locus, presented schematically in Fig. 1, specifies two transcripts: the toxin-encoding hok (hostkilling) mRNA and the labile antisense inhibitor Sok RNA (suppression of killing).mok (mediator of killing) is a reading frame, the translation of which is required for the translation of Sok (because of translational coupling ofhok to mok). Sok RNA is complementary to the hok mRNA 5′ end and inhibitshok translation indirectly by occluding themok ribosome-binding site (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar). Full-length versions of hok mRNA (1 and 2) are unusually stable because of extensive secondary structure formation and specific base pairings between their 5′ and 3′ ends (26Franch T. Gerdes K. Mol. Microbiol. 1996; 21: 1049-1060Crossref PubMed Scopus (51) Google Scholar). The structure ofhok-2 mRNA is shown in Fig.2 B. The 5′-to-3′ base pairing in the full-length molecules yields compact structures, andhok mRNA-1 and -2 are inactive with respect to translation and antisense RNA binding (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar, 26Franch T. Gerdes K. Mol. Microbiol. 1996; 21: 1049-1060Crossref PubMed Scopus (51) Google Scholar, 27Gerdes K. Thisted T. Martinussen J. Mol. Microbiol. 1990; 4: 1807-1818Crossref PubMed Scopus (78) Google Scholar, 28Thisted T. Nielsen A.K. Gerdes K. EMBO J. 1994; 13: 1950-1959Crossref PubMed Scopus (52) Google Scholar). Translation ofhok mRNA-2 is activated by slow 3′-exonucleolytical removal of the fbi (fold-back inhibition) element (Fig. 2 B). The 3′ processing releases the 5′ tac (translational activator) element and thereby triggers a series of structural rearrangements prerequisite for translation and rapid Sok RNA binding (27Gerdes K. Thisted T. Martinussen J. Mol. Microbiol. 1990; 4: 1807-1818Crossref PubMed Scopus (78) Google Scholar, 28Thisted T. Nielsen A.K. Gerdes K. EMBO J. 1994; 13: 1950-1959Crossref PubMed Scopus (52) Google Scholar, 29Thisted T. Sorensen N.S. Wagner E.G. Gerdes K. EMBO J. 1994; 13: 1960-1968Crossref PubMed Scopus (82) Google Scholar). Thus, the activated refolded hok mRNA contains the energy-rich tac stem and the antisense RNA target hairpin (Fig. 2 C), the latter of which is required both for translation and rapid antisense RNA binding (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar, 26Franch T. Gerdes K. Mol. Microbiol. 1996; 21: 1049-1060Crossref PubMed Scopus (51) Google Scholar, 29Thisted T. Sorensen N.S. Wagner E.G. Gerdes K. EMBO J. 1994; 13: 1960-1968Crossref PubMed Scopus (82) Google Scholar, 30Gultyaev A.P. Franch T. Gerdes K. J. Mol. Biol. 1997; 273: 26-37Crossref PubMed Scopus (40) Google Scholar). The presence of tac in the mRNA 5′ end suggested the existence of a regulatory element that would prevent initiation of translation during transcription. Nucleotide covariations in the aligned family ofhok mRNAs indicated that the mRNAs could form small metastable hairpins at their 5′ ends (30Gultyaev A.P. Franch T. Gerdes K. J. Mol. Biol. 1997; 273: 26-37Crossref PubMed Scopus (40) Google Scholar). However, the function of these hairpins is not known. Here we present evidence thathok mRNA specifies a structure that simultaneously prevents antisense RNA binding and synthesis of Hok toxin during transcription. This molecular safeguard consists of two small metastable hairpins at the mRNA 5′ end (Fig. 2 A). During transcription, the metastable hairpins prevent formation of the tac stem and ultimately inhibit formation of the target hairpin. Our data indicate that the metastable hairpins exist until completion of transcription and that their disruption is triggered by the fbi sequence at the mRNA 3′ end. Thus, we reveal here thathok mRNA follows a specific folding pathway that seems to have evolved to allow translation of hok in plasmid-free cells only. Ampicillin was added to a concentration of 30 μg/ml for mini-R1 plasmids and 100 μg/ml otherwise. Rifampicin (Ciba-Geigy) was added to cells prior to RNA extraction at a concentration of 250 μg/ml. Restriction endonucleases were purchased from Roche. All other enzymes were purchased from Promega. The partially Hok-resistantEscherichia coli mutant strain NWL37 (HokR, CmlR, KanR) (31Poulsen L.K. Larsen N.W. Molin S. Andersson P. Mol. Microbiol. 1992; 6: 895-905Crossref PubMed Scopus (27) Google Scholar) was used for the establishment of plasmids carrying mutated hok/sokloci. The E. coli K-12 strain CSH50 (Δ(lac pro)rpsL) (32) was used for plasmid stability tests (33Gerdes K. Larsen J.E. Molin S. J. Bacteriol. 1985; 161: 292-298Crossref PubMed Google Scholar) and rifampicin induction experiments. Both strains were cultured in Luria-Bertani broth (34Bertani G. J. Bacteriol. 1951; 62: 293-300Crossref PubMed Google Scholar) at 37 °C. Plasmid pPR633 carries the 580-base pair wild-typehok/sok locus from plasmid R1 cloned between the EcoRI and BamHI sites of pBR322. Plasmids pJMJ210 (U21A), pJMJ216 (A9U), and pJMJ218 (U21A/A9U) are mutant derivatives of pPR633. The probe plasmid, pGEM342, carries thehok/sok 240 base pairs downstreamSau3A-EcoRI fragment cloned under the T7 promoter of pGEMblue. The U21A,A9U, and U21A/A9U mutations were introduced using double PCR as described in Ref. 35Barettino D. Feigenbutz M. Valcarcel R. Stunnenberg H.G. Nucleic Acids Res. 1994; 22: 541-542Crossref PubMed Scopus (106) Google Scholar using the following primers: −20 primer, 5′-GTAAAACGACGGCCAGT; EcoRI clockwise, 5′-GTATCACGAGGCCCTTTCG; U21A, 5′-CACGTCATGTGGCAGAAAGCC; and A9U, 5′-GGCAGAAAGCCACAAGCGCCGGGCAC (mutant nucleotides are printed in bold face). The preparation of total RNA from E. coli and Northern transfer analysis were performed as described previously (27Gerdes K. Thisted T. Martinussen J. Mol. Microbiol. 1990; 4: 1807-1818Crossref PubMed Scopus (78) Google Scholar). The preparation and purification of uniformly 3H- or32P-labeled RNA molecules were performed essentially as described by Franch and Gerdes (26Franch T. Gerdes K. Mol. Microbiol. 1996; 21: 1049-1060Crossref PubMed Scopus (51) Google Scholar). The RNA species were synthesized using T7 RNA polymerase and templates generated by PCR. In all PCRs, pPR633 was used as template. The T7 promoter region and wild-type mRNA 5′ end was specified by the upstream oligonucleotide T7-hok, 5′-TGTAATACGACTCACTATAGGCGCTTGAGGC, whereas the upstream oligonucleotides of hok metastable andhok super-tac was T7-meta, 5′-TGTAATACGACTCACTATAGGCGCTTGAGGCTTTCTGCCTCAAGCGCCAAGGTGGTTTGTTGCC, and T7-perfect, 5′-CGGGATCCTGTATACGACTCACTATAGGGGCTA- TCTTCTTTCTGCCTCATGACGTGAAGGTGG, respectively. The downstream oligonucleotides specifying the 3′ end of the RNA run-off transcripts were: T7-3N, 5′-AAGGCGGGCCTGCGCCCGCCTCCAGG for truncated molecules; T7-4, 5′-AAGGCGCTTCAGTAGTCAG forhok mRNA-2; and T7-2, 5′-GCAAGGAGAAAGGGGCTAC forhok mRNA-1. Sok RNA was prepared as described elsewhere (36Dam M.N. Gerdes K. Mol. Microbiol. 1997; 26: 311-320Crossref PubMed Scopus (54) Google Scholar). The oligonucleotides used to generate PCR templates for Sok RNA synthesis were T7-sok wild type, 5′-CGGGATCCTGTAATACGACTCACTATAGACTAGACATAGGGATG- CCTCG, and NDM8, 5′-AGAAGATAGCCCCGTAGTAA. Translation of purified and native wild-type and mutated truncated hok mRNAs was performedin vitro by use of the E. coli-coupled transcription/translation system (37Zubay G. Annu. Rev. Genet. 1973; 7: 267-287Crossref PubMed Scopus (463) Google Scholar) purchased from Promega. For translational analysis of native mRNA in vitro, mRNA was synthesized from purified PCR-generated templates by T7 RNA polymerase. RNA synthesis was performed by incubation of 5 pmol of DNA template with 10 mm DTT, 40 mm Tris (pH 7.9), 6 mm MgCl2, 2 mm spermidine, 10 mm NaCl, 16 μm CTP, 10 μCi of [α-32P]CTP, 18 units of RNAguard (Amersham Pharmacia Biotech), and 15 units of T7 RNA polymerase for 12 min at room temperature. After incubation, 2-μl aliquots were transferred to prewarmed E. coli S30 extract followed by the addition of 0.2 μml-[35S]methionine. Sample preparation and subsequent SDS-polyacrylamide gel electrophoresis was carried out as described by Franch and Gerdes (26Franch T. Gerdes K. Mol. Microbiol. 1996; 21: 1049-1060Crossref PubMed Scopus (51) Google Scholar). Binding experiments were performed as described previously (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar). DNA templates for in vitro transcription were prepared by PCR. The PCR templates were identical to those used for the purification of labeled in vitro transcripts (as described above). PCR products were purified by a PCR purification kit (Qiagen), extracted with phenol/CHCl3, and precipitated with NH4OAc and ethanol. Transcription assays were carried out by incubation of 5 pmol of DNA template with 10 mm DTT, 40 mm Tris (pH 7.9), 6 mm MgCl2, 2 mmspermidine, 10 mm NaCl, 16 μm of CTP, 10 μCi of [α-32P]CTP, 125 mm GTP and UTP, 18 units of RNAguard (Amersham Pharmacia Biotech), and 20 units of T7 RNA polymerase for 8 min at room temperature. After incubation, 125 mm ATP was added to the preincubation mixture for initiation of transcription. Four-microliter aliquots were removed for RNase H probing at various time points after the initiation of transcription. Probing mixtures contained 4 μl of transcription mix, TMK-glutamate (see above) supplemented with 5 mm DTT, 10 pmol of DNA oligo, and 1 unit of RNase H. After 30 s of probing, samples were frozen in dry ice/ethanol, extracted with phenol/CHCl3, and precipitated. The transcripts were redissolved in formamide dye and analyzed by autoradiography on a 6% denaturing acrylamide gel. The oligonucleotide sequence used was 5′-CCTCGTGGTG. As a molecular marker, uniformly 32P-labeled purified truncated hok mRNA (100 fmol) was equilibrated in TMK-glutamate buffer (see above) supplemented with 5 mmDTT for 3 min at 37 °C prior to cleavage with RNase H. PCR-generated templates for RNA synthesis were prepared as described above. Native RNA transcripts were generated by incubating 5 pmol of DNA template with 10 mm DTT, 40 mm Tris (pH 7.9), 6 mmMgCl2, 2 mm spermidine, 10 mm NaCl, 1 mm nucleoside triphosphates, and 20 units of T7 RNA polymerase for 15 min at 37 °C. Native and purified RNA molecules were purified using gel-filtration spin columns (Bio-Rad) equilibrated in TMK-glutamate buffer (see above) and probed immediately. RNase T2 cleavage reactions were carried out in TMK-glutamate buffer for 5 min at 37 °C. Cleavage patterns were analyzed by primer extension using the hokA primer, 5′-GTGGTAGTTTCATGGC, according to Ref. 28Thisted T. Nielsen A.K. Gerdes K. EMBO J. 1994; 13: 1950-1959Crossref PubMed Scopus (52) Google Scholar. Based on folding simulations and phylogenetic analyses of the hok family of mRNAs, two hairpin structures were proposed to form at the hok mRNA 5′ end (30Gultyaev A.P. Franch T. Gerdes K. J. Mol. Biol. 1997; 273: 26-37Crossref PubMed Scopus (40) Google Scholar). The secondary structure of the nascent hoktranscript is shown in Fig. 2 A. By completion of transcription the hairpins must give way for the thermodynamically more stable structures, characteristic of the full-length molecule shown in Fig. 2 B. Notably, the full-length molecule contains the fbi-tac interaction and the top of the long tac stem. The U21A mutation that destabilizes the proposed metastable structure (Fig.2 A) while increasing the energy of the alternative tac stem (Fig. 2 C) is expected to increase the rate of the structural transition from the metastable structure to the tac stem (see Fig. 2,A and C). Along with the U21A mutation, we introduced the second-site A9U mutation, which restores the possibility of forming the metastable hairpin. The effects of U21A and A9U mutations on the stability and processing pattern of hokmRNA in vivo was assayed by Northern analysis. In the wild-type case (Fig. 3 A,first panel), the processing pattern is explained as follows (24Gerdes K. Gultyaev A.P. Franch T. Pedersen K. Mikkelsen N.D. Annu. Rev. Genet. 1997; 31: 1-31Crossref PubMed Scopus (171) Google Scholar): truncated hok mRNA is slowly and constitutively generated by exonucleolytical processing of hok mRNA-2. In the presence of Sok RNA, truncated hok mRNA is removed rapidly by RNase III cleavage of the hok/sokduplex RNA. Therefore, truncated hok mRNA is not observed in cells growing in steady state (i.e. before the addition of rifampicin in Fig. 3 A). However, after rifampicin, Sok RNA decays rapidly. This in turn allows for accumulation of the stable truncated hok mRNA. The U21A mutation dramatically reduced the levels of both full-length molecules (Fig. 3 A, second panel) and inactivated the ability of the system to mediate plasmid stabilization by the PSK mechanism (data not shown). During steady-state cell growth, a band corresponding to the major product of Sok-mediated RNase III cleavage of hok mRNA is visible, indicating that the predominant fraction of the molecules is degraded by RNase III (-2 minsample in second panel). This in turn indicates that mutatedhok mRNA binds Sok RNA prematurely or, less likely, that Sok RNA exhibits enhanced binding to the full-length RNAs. We then investigated whether disruption of the metastable structure by the U21A mutation could be back-complemented by the U9A mutation (Fig.2 C). As seen from Fig. 3 A (fourth panel), the U21A/A9U double mutation restored the processing pattern to that of the wild-type mRNA. This provides genetic evidence for the existence of the 5′ metastable hairpin. The A9U single mutation had no effect on the mRNA processing pattern, probably because it simultaneously reduces the energy of the metastable and tac stems (see "Discussion"). To further investigate the function of the metastable structurein vivo, we analyzed hok mRNA band patterns in the absence of Sok RNA (Fig. 3 B). In the wild-type case, truncated hok mRNA is now present in steady state, because it cannot be removed by antisense RNA-mediated RNase III cleavage (first panel). Importantly, the mRNA level of the U21A is no longer reduced (second panel), implying that the U21A mutation accelerates Sok RNA binding to hokmRNA. The band patterns of hokA9U andhokU21A/A9U were also similar to that of the wild-type mRNA (Fig. 3 B). We find this result consistent with the proposal that the metastable structure functions to keep the nascent hok transcript in an inactive configuration. To gain more direct insight, we turned to in vitro analyses. Two mutant mRNAs were constructed (Fig.4). In one RNA termed "hoksuper-metastable" mRNA, the metastable hairpins were forced by the introduction of five base changes in the 5′ stem. In the second RNA termed "hok super-tac," the tac stem was forced by five different base changes (Fig. 4). The secondary structures of the forced mutant RNAs were confirmed experimentally (data not shown). Sok RNA and variants of truncated hok mRNA were synthesized in vitro, gel-purified, and renatured. Using an in vitro binding assay (38Persson C. Wagner E.G. Nordstrom K. EMBO J. 1988; 7: 3279-3288Crossref PubMed Scopus (97) Google Scholar), the apparent second-order binding-rate constants (K app) of Sok RNA association with truncated wild-type and mutant hokmRNAs were determined (Table I). Sok RNA bound with similar high rates to truncated wild-type and super-tachok mRNAs. However, the binding rate was reduced ∼10-fold in the case of the super-metastable hok mRNA. Thus in this assay the forced metastable structure reduced antisense RNA binding significantly. This is consistent with the notion that the super-metastable structure prevents the formation of the antisense RNA target hairpin (Fig. 2 C). In contrast, Sok RNA bound rapidly to both purified wild-type and super-tac hok mRNAs, indicating that these molecules contain the antisense RNA target hairpin that mediates that highest rate of antisense binding (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar, 39Franch T. Petersen M. Wagner E.G. Jacobsen J.P. Gerdes K. J. Mol. Biol. 1999; 294: 1115-1125Crossref PubMed Scopus (135) Google Scholar). Note that the molecules used above were denatured and renatured during their purification.Table IIn Vitro Sok RNA binding ratesTruncated RNASecond-order binding-rate constantsRelative binding rates105M−1 s−1hok wild type8.01super-tac8.01super-metastable0.90.11The numbers were obtained by standard RNA binding assays as described previously (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar).The numbers given are averages of at least three independent measurements. Open table in a new tab The numbers were obtained by standard RNA binding assays as described previously (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar). The numbers given are averages of at least three independent measurements. To test the effect of the metastable structure on hoktranslation, wild-type and mutant hok mRNAs were translated in a cell-free S30 extract (37Zubay G. Annu. Rev. Genet. 1973; 7: 267-287Crossref PubMed Scopus (463) Google Scholar). When gel-purified and renatured, the wild-type truncated hok mRNA was translated efficiently (Fig. 5, +,upper panel). This result is consistent with the finding that renatured truncated hok mRNA folds into the translatable configuration that contains the energy-rich tac stem (26Franch T. Gerdes K. Mol. Microbiol. 1996; 21: 1049-1060Crossref PubMed Scopus (51) Google Scholar) (Fig. 2 C). To examine if this was also the case for the native molecule, transcripts were synthesized in vitro and added directly to the S30 extract without denaturation and renaturation. Strikingly, truncated wild-type hok mRNA in its native form was not translated at all (Fig. 5). Native truncated hok mRNA carrying the forced metastable structure was also translated very inefficiently. In contrast, the native forms of hokmRNAs carrying the super-tac and U21A mutations were translated efficiently. Truncated hok mRNA carrying the A9U was not translated, probably because the mutation destabilizes the tac stem required for translation. As expected the two full-lengthhok mRNAs also were not translated (Fig. 5). These results indicate that mutations or conditions that favor the metastable structure prevent translation, whereas the presence of the tac stem allows translation. An important inference from the above-described results was that gel-purified truncated hok mRNA should contain the tac stem, whereas the native form of the molecule should contain the metastable structure. To investigate this directly, we performed a structural analysis. As seen in Fig.6 A, the RNase T2 cleavage patterns (T2 cleaves 3′ of unpaired nucleotides) of the two forms of the RNA are strikingly different. The purified form has prominent bands corresponding to the tac-stem loop (tac in Fig.6 A), whereas these bands are reduced in the native isoform of the mRNA. In contrast, the native isoform exhibits enhanced cleavage at bases corresponding to the loops of the metastable hairpins (m1 and m2 in Fig. 6 A). The clear differences between the RNAs indicate that the metastable hairpins exist in truncated hok mRNA. Leaving the native truncated hok mRNA at room temperature for 30 min before probing does not change the probing pattern (data not shown), indicating that this configuration is stable throughout the probing experiment. To study the refolding kinetics of native hok mRNA, a coupled transcription/RNase H-probing time-course assay was conducted (Fig. 7; see "Experimental Procedures"). Native wild-type and mutated truncated hok mRNAs were generated with T7 RNA polymerase and structure-probed with RNase H using an oligonucleotide complementary to the region of the mok ribosome-binding site (see Fig. 2, A and C). Samples were withdrawn from the transcription mix at various time points after initiation to monitor structural rearrangements. Native hok mRNA containing the forced tac stem exhibited efficient cleavage (Fig. 7,third panel), indicating that in this RNA the mokribosome-binding site exists in an open configuration. This is consistent with the translatability of the RNA (Fig. 5). In contrast,hok mRNA containing the forced metastable structure was not cleaved at all (Fig. 7, second panel), indicating that the mok ribosome-binding site is in a closed configuration. Again, the closed structure is consistent with the observed lack of translation (Fig. 5). Note that the mok SD region is located 70 nucleotides downstream of the mutational changes in the RNAs, showing that the mutations impose long-range structural changes in the RNAs. Most importantly, wild-type native hok mRNA exhibited the same RNase H cleavage pattern as the RNA containing the forced metastable structure, and the pattern did not change in time (Fig. 7,first panel). These results support that the metastable structure exists stably in native truncated hok mRNA and that it keeps the mok SD region in a closed configuration. Fig. 6 B shows a structural analysis of purifiedversus native full-length hok mRNA-1. Both the purified and native isoforms were cleaved at positions corresponding to the loop of the tac stem, indicating that the native molecule contains this structure. This stands in contrast to truncated wild-type hok mRNA, in which case the native form exhibited a cleavage pattern compatible with the metastable structure (Fig. 6 A). Thus, the presence of the 3′ fbi element eliminates the existence of the metastable hairpins. By inference, this suggests that the 3′ fbi element triggers the disruption of the metastable hairpins in the native full-length mRNA. The previous identification of a translational activator element (tac) in the 5′ end of hok mRNA posed the following question: during transcription, how is lethal translation and Sok RNA-mediated degradation of hok mRNA avoided? A possible answer to this question came from the phylogenetic observation that allhok homologous mRNAs (including hok itself) have the possibility to form local hairpins at their 5′ ends. In all RNAs, these metastable hairpins would preclude formation of the longer tac stems (30Gultyaev A.P. Franch T. Gerdes K. J. Mol. Biol. 1997; 273: 26-37Crossref PubMed Scopus (40) Google Scholar). By preventing tac-stem formation, the metastable hairpins would favor the formation of downstream inhibitory structures, thereby keeping the mRNA translationally inert and relatively inaccessible for antisense RNA binding. Recently, the metastable structure comprising the two local hairpins (Fig. 2 A) has been shown to exist in vitro in a minimal molecule consisting of the 5′-end 74 nucleotides of hok mRNA (40Nagel J.H. Gultyaev A.P. Gerdes K. Pleij C.W. RNA (N. Y.). 1999; 5: 1408-1418Crossref PubMed Scopus (36) Google Scholar). The presence of this structure in the native molecule in vivo as well as its function was, however, not investigated. Here, a single nucleotide change that simultaneously disrupted the metastable structure and favored the tac stem (U21A) reduced the intracellular amount of hok mRNA. This reduction was accompanied by a PSK-negative phenotype. Furthermore, the reduction depended on Sok RNA, indicating increased antisense RNA binding to the mutated hok mRNA (Fig. 3). These results are consistent with the proposal that the U21A-mutated hoktranscript refolds prematurely into the tac-stem configuration. The tac stem favors the formation of the antisense RNA target hairpin that in turn would increase the binding rate of Sok RNA during transcription (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar). The deleterious effect of the U21A mutation on hok mRNA stability was complemented by the A9U mutation (Fig. 3). This observation provides genetic evidence for the formation of the metastable structure in vivo (Fig. 2 A). The A9U mutation itself had no effect on the processing pattern ofhok mRNA. This indicates that destabilization of the metastable structure per se does not account for the observed effect of the U21A mutation. However, U21A also increases the energy of the tac stem (Fig. 2 C). Hence, during transcription U21A may increase the rate of refolding into the stable tac stem that in turn would favor rapid antisense RNA binding. The in vivo results presented in Fig. 3 corroborate the proposed function of the metastable structure. To obtain more direct evidence, we employed in vitro techniques. Enforcement of the metastable structure of hok mRNA led to a 10-fold decrease in the in vitro Sok RNA binding rate (Table I). The enforced metastable structure precludes the formation of the tac and target stem loops (Figs. 2 A and 6 A). The target hairpin contains a specialized structure (the U-turn) that is required for rapid antisense binding (39Franch T. Petersen M. Wagner E.G. Jacobsen J.P. Gerdes K. J. Mol. Biol. 1999; 294: 1115-1125Crossref PubMed Scopus (135) Google Scholar). As in tRNA, the U-turn exposes three bases in the target loop, thereby enhancing the Sok RNA binding rate ∼10-fold. Thus, if the target hairpin cannot form, the U-turn structure cannot form either, and rapid antisense binding is prevented. This explains the slower binding rate for a target RNA that contains the enforced metastable structure. On the other hand, purified wild-type and super-metastable hok mRNAs bound Sok RNA rapidly, indicating that both of these RNAs are in the tac-stem form (Table I). Previously, we found that gel-purified truncated hokmRNA was translated efficiently, whereas full-length hokmRNA-1 and -2 were translated poorly (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar, 26Franch T. Gerdes K. Mol. Microbiol. 1996; 21: 1049-1060Crossref PubMed Scopus (51) Google Scholar, 28Thisted T. Nielsen A.K. Gerdes K. EMBO J. 1994; 13: 1950-1959Crossref PubMed Scopus (52) Google Scholar). Here we expand our analyses using native RNAs that were not denatured and renatured before being subjected to translation (Fig. 5). Native full-length mRNA-1 and -2 were translated poorly, consistent with the finding that native full-length mRNA-1 did not contain the target stem-loop known to be present in the active translatable mRNA (Fig.6 B). This is consistent with the native full-length RNAs being in the closed configuration containing the fbi-tac interaction but lacking the target stem-loop (Figs. 2 B and6 B). Gel-purified truncated hok mRNA was translated efficiently. In contrast, the native isoform of the RNA was not translated at all (Fig. 5). Structure mappings of the two isoforms showed patterns consistent with the metastable structure being present in the native RNA and the tac/target stem-loops in the purified species (Fig. 6 A). Thus, we conclude that the metastable structure prevents translation in native truncated hok mRNA. This conclusion is corroborated by translational analysis of the series of mutant mRNAs (Fig. 5). Mutations favoring the metastable structure (super-metastable and U21A/A9U) reduce translation, whereas mutations favoring the tac/target configuration (super-tac and U21A) favor translation. The native mRNA carrying the A9U mutation was translated at a lower rate than the one carrying U21A. Both mutations disrupt the metastable structure, but only U21A increases the stability of the tac stem. The low translation rate conferred by A9U may be caused by secondary effects on tac-stem stability (e.g.disruption of a possible noncanonical GA/AG base pairing at the bottom of the tac stem, see Fig. 2 C). Native and purified full-length hok mRNA-1 was also structure-probed (Fig. 6 B). In this case, the cleavage patterns were indistinguishable. The cleavage pattern is consistent with previous structure-probing of purified RNAs (25Franch T. Gultyaev A.P. Gerdes K. J. Mol. Biol. 1997; 273: 38-51Crossref PubMed Scopus (85) Google Scholar) and indicates that the native full-length molecules do not contain the metastable structure but rather the top of the tac stem (Fig. 2 B). Thus, in contrast to native truncated hok mRNA, native full-length molecules do not exhibit a cleavage pattern compatible with the metastable configuration (Fig. 6, A and B). However, both molecules are in closed configurations lacking the antisense target hairpin. A reasonable inference from these results is that the 3′-end fbi element triggers disruption of the metastable structure by forming the fbi-tac interaction (the transition fromA to B in Fig. 2). In turn, this triggers formation of the top of the tac stem in the full-length molecule. Hence, in the course of transcription, one inactive form of the mRNA (containing the metastable structure) is replaced actively by the formation of a second inactive form (the full-length transcript). In other words, co-transcriptional repression ofhok translation by the metastable structurein vivo does not rely on refolding kinetics but rather assumes a "lock and key" mechanism, in which the fbi element constitutes the key for unlocking the metastable structure. This is in keeping with hok/sok biology, because spontaneous refolding of the 5′ end into a translatable conformation could be lethal to the cell. Because of the considerable thermodynamic stability of the two metastable hairpins, refolding is unlikely to involve complete unwinding of the two hairpins prior to the formation of thermodynamically more stable structures. Such a transition would require a large activation energy and thus becomes extremely slow at physiological temperatures. Rather the refolding occurs through a number of folding intermediates by an RNA strand exchange mechanism in which one secondary structure is formed at the expense of another. Such mechanisms have been proposed to account for the rapid interchange of RNA secondary structures in Leptomonas collosoma spliced leader RNA (41LeCuyer K.A. Crothers D.M. Biochemistry. 1993; 32: 5301-5311Crossref PubMed Scopus (77) Google Scholar) and the P(−1)/P1 region of Tetrahymenagroup I intron (17Cao Y. Woodson S.A. RNA (N. Y.). 1998; 4: 901-914Crossref PubMed Scopus (21) Google Scholar). One possibility is that the extreme 5′-end nucleotides serve as a toehold for nucleation of fbi-tac formation and concomitant unwinding of the metastable hairpin I. Analysis of a hok G30C mutant, however, suggests that this is not the case. The G30C mutation, which creates a base pair with the extreme 5′-end guanosine, has no effect on PSK, indicating that refolding is unaffected by this substitution (data not shown). Alternatively, refolding can be nucleated by the formation of a pseudoknot intermediate by long range interaction between loop nucleotides (U13-C16) of metastable hairpin I and their downstream complementary subset (G60-A63). Such an intermediate structure has been proposed for the spliced leader RNA (42LeCuyer K.A. Crothers D.M. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 3373-3377Crossref PubMed Scopus (73) Google Scholar). The present identification of a metastable structure formed during transcription extends the concept of hok mRNA folding dynamics to involve two refolding events that are separated in time. Initially, during early stages of transcription two metastable hairpins are formed in the very 5′ end of the mRNA. As the rate of RNA secondary structure formation exceeds that of RNA synthesis by orders of magnitude, their formation is favored kinetically. By sequestering the 5′-end tac element, the metastable hairpins facilitate subsequent formation of downstream inhibitory structures, protecting the RNA against premature degradation and the cell against Hok-toxin synthesis. Upon completion of transcription, the long range fbi-tac interaction together with the partial tac stem replace the metastable structure. Eventually, slow constitutive processing of the full-length mRNA leads to the second refolding event, which produces the active version of the hok mRNA (Fig. 2 C). Finally, depending on the presence of Sok RNA, the mRNA is either translated or inactivated. We thank C. W. A. Pleij, Alexander P. Gultyaev, and other members of the Leiden group for discussions and comments to the manuscript.