INTRODUCTION

In all studied cells, initiation of translation involves a specialized tRNA^Met species that bears identity determinants that allow it to interact with the components of the initiation apparatus and prevent it to play with the elongation factors. In eubacteria, the initiator tRNA is called tRNA^Met_f because the normal pathway of initiation begins with the addition of a formyl group to the methionylated tRNA (1, 2). The importance of the formylation reaction for the efficiency of the translational process has been early recognized (3, 4). Recently, the disruption of the fmt gene encoding the formylating enzyme has brought direct evidence that formylation of the initiator tRNA is needed for insuring the rapid growth of Escherichia coli (5).

The nucleotides of E. coli initiator tRNA determining its function have been studied by identity switching experiments; species of elongator tRNAs (tRNA^Met_m and tRNA^Gln₂) have been converted into initiator ones after receiving the acceptor stem and the anticodon stem and loop of tRNA^Met_f (6, 7). These regions are respectively known to be important for (i) formylation of the aminoacyl-tRNA (8, 9) and (ii) interaction with initiator factor IF3¹ and the ribosomal P site (10, 11, 12). In particular, the absence of pairing between bases 1 and 72 of the acceptor stem, a unique feature of initiator tRNA in procaryotes (13), governs both the recognition of the aminoacyl-tRNA by methionyl-ARNt^Met_f formyltransferase and its low affinity for the elongation factor EFTu (8, 14, 15).

Reciprocally, several variants of tRNA^Met_f have been shown to suppress an amber mutation inside a reporter gene and therefore to be active in elongation (6, 8, 16, 17, 18). In addition to an amber anticodon (CUA), most of these tRNAs bear a mutation that restores a Watson-Crick base pair, U-A or C-G, at positions 1-72. However, some of the elongator variants of tRNA^Met_f keep their C¹A⁷² pair unchanged (6, 8, 9, 16). Consequently, the lack of a base pairing at positions 1-72 does not appear sufficient per se to prevent the initiator tRNA from entering the elongation pathway. Moreover, a tRNA^Met_f with only the amber anticodon can become active in elongation provided that cells are rendered partially deficient in formyltransferase activity (6). This latter result has led us to consider that formylation of methionyl-tRNA^Met_f could be important also in preventing its misappropriation by the elongation apparatus.

In combination with IF3, IF2 favors the specific binding of the initiator tRNA species to the 30 S mRNA complex (19). N-Acylation of the aminoacylated tRNA is required to observe the formation of a binary complex with IF2 (20, 21, 22). However, IF2 stimulates the binding of unformylated tRNAs to the 30 S ribosomal subunit in vitro (23). Because the 30 S subunit displays a high affinity binding site for IF2 (24), IF2 is thought to act while bound to the 30 S subunit rather than free like a tRNA carrier (23, 25). Therefore, the precise mechanism of action of IF2 remains unclear.

To address the questions concerning with the interaction of tRNA^Met_f with EFTu or IF2 in vivo, we have overproduced each of these two factors in the cell and studied the consequences on the fates of various tRNAs. In addition we have examined the effect of the overproduction of tRNA^Met_f and of several derived tRNA^Met variants on the growth rate of an fmt strain.

MATERIALS AND METHODS

Oligonucleotides were synthesized on a Pharmacia gene assembler and purified by anion-exchange chromatography (Mono Q, Pharmacia Biotech Inc.).

The EcoRV-PstI fragment from the pTUB1 plasmid (Ref. 26, kindly provided by Dr. M. Springer) bearing the tufB gene was inserted between the SmaI and PstI restriction sites of pUC18 to give pUCtuf. The insertion of the SacI-HindIII fragment from pUCtuf between the corresponding sites of the pACFatg plasmid (6) gave pACtuf. The insertion of the XbaI-XhoI fragment from pBSM547WA461AV451QP213DA449 (27) between the XbaI and HindIII sites of pACtuf gave pACMTS*tuf, which also expressed a mutant form (MTS*) of E. coli methionyl-tRNA synthetase. The factor of overexpression of EFTu in cells harboring either the pACMTS*tuf or the pACtuf plasmids was quantitated by Western blot and found equal to 1.5 ± 0.1-fold.

A polylinker that provided the restriction sites NotI, HindIII, StuI, BglII, SalI, and XbaI was inserted between the NotI and XbaI sites of plasmid PACFatg (6). A fragment containing the infB gene was created by digestion of the pB18.1 plasmid (28) with BglII and HindIII and further cloned between the corresponding sites of the polylinker yielding pACinfB. To obtain the pACinfBfmt plasmid, a DNA fragment containing the fms-fmt operon was amplified from pBS936 plasmid (5) by polymerase chain reaction with the use of the following primers: 5-CTTTGCGATTGGCTAGCGATGCTGTCAA-3 and 5-TGTTCAACGGCCTCGAGCGCCATGCTAC-3. The resulting fragment was restricted with NheI and XhoI and then inserted into the SalI and XbaI sites of pACinfB. To obtain the pACfmt plasmid, the infB gene was deleted from pACinfBfmt using HindIII and BglII and Klenow polymerase fill-in. The factor of overexpression of IF2 in cells harboring either the pACinfBfmt or the pACinfB plasmids was quantitated by Western blot and found equal to 5 ± 2.5-fold. The factor of overexpression of formyltransferase in cells harboring either the pACinfBfmt or the pACfmt plasmid was assayed as described (5) and found equal to 27 ± 4-fold.

The genomic library from PAL13.3Tr was constructed using pHC79 (29) essentially as described previously (30). The metZ region was amplified by polymerase chain reaction using the oligonucleotides 5-CATGCCAAATCCGAATTCGGGGTAAAAAA-3 and 5-CGCTTTTTTATGGTCGACCAGAAACAAAA-3. The amplified fragment was then cleaved with EcoRI and SalI and inserted between the corresponding sites of pHC79, yielding pHCmetZ.

Bulk tRNA extracts were prepared according to Meinnel and Blanquet (31). The amounts of formyl-methionine and lysine accepting species in such extracts were measured according to Refs. 8 and 32, respectively. The relative amount of tRNA^Met_f in an extract was defined as the ratio between the amount of tRNA^Met_f and that of tRNA^Lys. The factor of tRNA^Met_f overproduction in an extract is the ratio of the relative amount of tRNA^Met_f in this extract to the relative amount of tRNA^Met_f in a control extract derived from JM101Tr cells grown in the same conditions.

-Galactosidase activity was measured in Miller units from toluene-treated cells (33) as described (6). In each strain, four independent clones were systematically assayed, and the three closest results were used to calculate an average value with the associated deviation given in the table. For each combination of plasmids, it was verified that their presence did not impair bacterial growth of the indicator strains PAL125R and UF121R, thus excluding an indirect effect on -galactosidase activity measurements that could have resulted from a decrease in the overall protein synthesis rate.

RESULTS

The initiator and elongator activities of various tRNAs bearing an amber anticodon (CUA) could be previously evaluated through the suppression of an amber mutation located either inside or at the first position of a reporter gene (6). Here, we further measure and compare both the initiator and the elongator activities of these tRNAs as a function of the amount of EFTu present in the host cell. The suppression of an internal amber codon in the lacI-lacZ gene fusion was followed in the UF121R indicator strain (Table I; 32). In a second assay, the reporter lacZ_TAG1 gene, starting on an amber codon (Fig. 1), was introduced into PAL125R (an F strain derived from UF121R) through transformation with pRSlacZtRNA^Met_f amber plasmid, carrying both the lacZ_TAG1 and the tRNA^Met_f amber genes.

Table I. Description of strains and plasmids used in this study Strain or plasmid Description PAL13Tr supE thi (lac-pro) recA56 srl-300::Tn10 fmt1::kan F(traD36 proA+B+ lacIq lacZM15) (5) PAL13TrFatg Same as PAL13Tr but Fatg; expressing the formylase from the fmt-1 (ATG) gene under the control of the lac promoter (6) UF121R ara argEam lac-proB nalA rif thi recA56 srl-300::Tn10 FproA-B+ lacI-Z UAG 181; indicator strain carrying the reporter gene in the elongation assay (42) PAL125R F derivative of UF121R; host strain in the initiation assay (6) pBStRNA Derivatives of pBStNAV overexpressing tRNAs (43) pRStRNA Plasmids expressing tRNAs in the elongation assay in UF121R (6) pRSlacZtRNA Plasmids expressing tRNAs in the initiation assay, also carrying the reporter lacZTAG1 gene (6) pACMTS* pACYC184 derivative overproducing a MetRS variant (MTS*) capable of methionylating amber tRNAs (6) pACtuf pACYC184 derivative overproducing the elongation factor Tu pACMTS*tuf pACYC184 derivative overproducing both the MTS* MetRS variant and the elongation factor Tu pACinfB pACYC184 derivative overproducing the initiation factor 2  pACfmt pACYC184 derivative overproducing the methionyl-tRNAMetf formyltransferase pACinfBfmt pACYC184 derivative overproducing both the methionyl-tRNAMetf formyltransferase and the initiation factor 2

The three tRNA^Met_f amber variants (U¹, G⁷², and G⁷³), which can participate in both the elongation and the initiation of protein synthesis (6), were studied. The formylability and interaction with EFTu of these tRNAs are affected to different extents (8, 9, 15).

UF121R cells were transformed by the five pRStRNA plasmids (encoding tRNA^Met_m amber, tRNA^Met_f amber, or the G⁷², G⁷³, and U¹ variants of the latter). PAL125R cells were transformed by the corresponding set of pRSlacZtRNA plasmids (Fig. 1). In the case of each studied tRNA^Met amber, levels of -galactosidase activity were measured in the above UF121R and PAL125R derivatives, overexpressing or not EFTu. In order to distinguish the fate of the tRNAs actually aminoacylated in vivo with methionine, plasmids expressing either a methionyl-tRNA synthetase variant (MTS*) capable of aminoacylating a tRNA^Met bearing the amber anticodon (pACMTS*) or both MTS* and EFTu (pACMTS*tuf) were used. In this case, cells were transformed with either the pACMTS*tuf plasmid, bearing the tufB gene, or the pACMTS* vector as a control.

The overproduction of EFTu had minor effects on the rate of suppression in initiation obtained with tRNA^Met_f amber or on that in elongation obtained with tRNA^Met_m amber (Table II). In contrast, the overproduction of EFTu resulted in (i) a significant suppression with tRNA^Met_f amber in elongation, (ii) a 2-fold increase of the suppression in elongation with the U¹ and G⁷³ variants, and (iii) a decrease of the suppression in initiation with the U¹, G⁷², and G⁷³ variants by factors ranging from 1.5 to 3-fold.

Table II. Effect of an overproduction of EFTu on the suppression during elongation or initiation by tRNAMetf variants tRNAMet amber  -Galactosidase activity in extracts of the following strain carrying the indicated plasmid UF121R (elongation) PAL125R (initiation) pACMTS* pACMTS*tuf pACMTS* pACMTS*tuf f <0.02 0.08  ± 0.01a 33.1  ± 2.5 49.1  ± 4.0 fU1 4.5  ± 0.3 8.0  ± 0.4 34.4  ± 1.7 23.8  ± 1.0 fG72 28.0  ± 0.5 29.5  ± 0.4 6.2  ± 0.3 1.7  ± 0.1 fG73 1.4  ± 0.1a 2.5  ± 0.1a 2.1  ± 0.1a 1.4  ± 0.1a m 71.7  ± 1.6 65.2  ± 1.7 <0.02 <0.02 a  Values are the same as those obtained in the absence of MTS* with either pACYC184 instead of pACMTS* or pACtuf instead of pACtufMTS*.

From these results, it could be concluded that the amount of intracellular EFTu influenced the selection of an amber tRNA by either the initiation or the elongation apparatus. However, in vivo, tRNA^Met_f amber variants behave as substrates of glutaminyl-tRNA synthetase (17). Consequently, in the case of the G⁷³ species, a poor substrate of E. coli methionyl-tRNA synthetase, the measured suppression had to be attributed to the sole glutaminylated tRNA (6, 34). With the other tRNAs, the efficiencies of suppression could be compared in the presence or the absence of MTS* by transforming the cells with either pACYC184 or pACMTS* and with either pACtuf or pACtufMTS*. In the case of tRNA^Met_f amber, the lack of effect of MTS* suggested that suppression in elongation should be attributed to a glutaminylated tRNA species. In all other cases, the relative extents of suppression were lower in the absence of MTS*, thus indicating that the suppression was actually mediated for the most part by a methionyl-tRNA species.

IF2 was overproduced from the pACinfB plasmid and the effect on the initiator activity of several tRNA^Met_f amber variants was assayed in the PAL125R indicator strain (Table III). In agreement with the data recently reported by Mangroo and RajBhandary (35), an important positive effect of IF2 on initiator activity was observed in the case of tRNA^Met_f amber and of its U¹ and G⁷³ variants. To know whether this effect of overproduced IF2 involved formylated tRNA species, initiation activities were also compared in the presence of an excess of the formylase enzyme overexpressed from the pACfmt or the pACinfBfmt plasmids. By causing a higher formylation rate, an overproduction of the formylase is known to increase the efficiency of amber tRNAs in translation initiation (6, 35). In these conditions, an overproduction of IF2 showed little or no effect on initiator efficiencies (Table III), thereby indicating that overproduced IF2 recognized the unformylated tRNA species.

Table III. Effect of an overproduction of IF2 on the suppression during initiation by tRNAMetf variants tRNAMet amber  -Galactosidase activity in extracts of PAL125R carrying the indicated plasmid pACYC184 pACinfB pACfmt pACinfBfmt f 2.3  ± 0.1 4.0  ± 0.5 4.2  ± 0.3 3.5  ± 0.2 fU1 0.2  ± 0.1 0.6  ± 0.1 0.9  ± 0.1 0.8  ± 0.1 fG72 <0.02 <0.02 0.06  ± 0.01 0.06  ± 0.01 fG73 1.7  ± 0.1 4.3  ± 1.0 12.5  ± 0.2 11.4  ± 0.6

To further probe this hypothesis, we examined the effect of overproducing IF2 on the elongator activity of tRNA^Met_f amber variants in the UF121R indicator strain. With the tRNA^Met_f amber variants U¹, G⁷², and G⁷³, a decrease in elongator activity was systematically observed upon overproduction of either IF2 or of both formyltransferase and IF2 (Table IV). No such decrease could be observed with elongator tRNA^Met_m amber. The case of the tRNA^Met_f amber could not be studied because of its lack of activity in elongation. These results brought additional support to the idea that IF2 can interact with unformylated aminoacyl-tRNA apart from the P site of the ribosome.

Table IV. Effect of an overproduction of IF2 on the suppression during elongation by tRNAMetf variants tRNAMet amber  -Galactosidase activity in extracts of UF121R carrying the indicated plasmid pACYC184 pACinfB pACfmt pACinfBfmt fU1 5.7  ± 0.1 4.2  ± 0.1 0.16  ± 0.01 0.12  ± 0.01 fG72 12.1  ± 1.0 9.8  ± 0.2 11.8  ± 0.9 8.2  ± 0.2 fG73 3.3  ± 0.3 2.8  ± 0.1 0.32  ± 0.01 0.25  ± 0.02 m 32.4  ± 3.7 31.6  ± 1.0 not determined not determined

An Overproduction of Initiator tRNA Partially Restores the Growth of an fmt Strain

The fmt strain PAL13Tr is characterized by a very slow growth and a thermosensitive phenotype at 42 °C (5). Fast growers spontaneously appear in liquid cultures. One of these fast growing cell line (PAL13.3Tr) has a generation time of 1 h 40 min instead of 3 h 20 min in the case of PAL13Tr as measured in LB at 37 °C. Like in the case of PAL13Tr, formylase activity and formylation of initiator tRNA were not detectable in PAL13.3Tr. Interestingly, however, PAL13.3Tr constitutively overexpresses initiator tRNA by a factor of 2.5, with respect to PAL13Tr or to the isogenic fmt⁺ strain JM101Tr.

A chromosomal library from PAL13.3Tr was constructed in pHC79 in order to isolate DNA fragments able to complement in trans the thermosensitive phenotype of PAL13Tr. From 500 tested clones, 2 positive clones growing steadily at 42 °C were isolated. These clones harbored hybrid cosmids, called pHC74 and pHC132, carrying inserts of about 30 kilobase pairs. Sau3a deletions were performed in these two cosmids, and the smallest cosmids still able to complement were designated pHC74.96 (7.5 kilobase pairs) and pHC132.46 (4.5 kilobase pairs). The sequencing and comparison of a 100-base pair fragment shared by the two cosmids with the GenBank^TM release 90 allowed the identification of a DNA fragment contiguous to the metZ region. A 400-base pair fragment corresponding to the entire metZ region carrying the promotor, the three tRNA genes, and the terminator was then amplified by polymerase chain reaction from the PAL13.3Tr chromosomal DNA and cloned into pHC79, yielding pHCmetZ. pHCmetZ cured the growth defect of PAL13Tr, thus confirming that this region was indeed responsible for the phenotype conferred by pHC74 and pHC132. The three metZ genes from pHC74 were then sequenced, and the sequence matched that previously reported. Although this finding did not explain the origin of the tRNA^Met_f overproduction in PAL13.3Tr, it clearly established that the addition in multicopy of the metZ region was enough to restore the growth of the fmt strain PAL13Tr.

PAL13Tr was then transformed with one of the cosmids pHC74, pHCmetZ, or pHC74.96. From measurements of the generation times of the resulting strains in LB medium at 37 °C (80, 68, and 47 min, respectively) and of the relative amounts of overproduced tRNA^Met_f (6-, 8-, and 13-fold, respectively) in the cells harvested in late exponential phase, it could be concluded that the increase of growth rate conferred by each plasmid varied in parallel with the degree of overproduction of the initiator tRNA.

To find precisely which determinants of initiator tRNA are the most important to sustain cell growth in a fmt context, we transformed the PAL13TrFatg strain (5) with a set of pBStRNA^Met plasmids. Cells streaked from LB plates containing 12.5 mg/ml ampicillin and 1 mM isopropyl-1-thio--D-galactopyranoside were further examined for growth at 37 °C on LB plates containing 0.2% glucose, conditions in which formylase activity can no more be detected in cell extracts and in which the recipient strain does not grow (5). The assayed tRNAs were tRNA^Met_f, tRNA^Met_f U¹, tRNA^Met_f G⁷², and tRNA^Met_f G¹C⁷² on the one hand and tRNA^Met_m, tRNA^Met_m G²⁹G³⁰G³¹C³⁹C⁴⁰C⁴¹, tRNA^Met_m C¹A⁷²G²⁹G³⁰G³¹C³⁹C⁴⁰C⁴¹, tRNA^Met_fasm, tRNA^Met_fasm G²⁹G³⁰G³¹C³⁹C⁴⁰C⁴¹, and tRNA^Met_fasm G¹C⁷²G²⁹G³⁰G³¹C³⁹C⁴⁰C⁴¹ on the other hand (8). All tRNAs had a positive effect on the growth of PAL13TrFatg, except tRNA^Met_m and tRNA^Met_fasm, a derivative of tRNA^Met_m with the acceptor stem of tRNA^Met_f (8), in which cases colonies could not be detected after 48 h. Noticeably, the anticodon stems of the two latter tRNA species lack the G²⁹G³⁰G³¹C³⁹C⁴⁰C⁴¹ motif required for IF3 recognition. Authentic tRNA^Met_f and the tRNA^Met_fasm G²⁹G³⁰G³¹C³⁹C⁴⁰C⁴¹ variant were the most efficient in supporting the growth of PAL13TrFatg. Noticeably, nonformylatable species like tRNA^Met_f G⁷² and tRNA^Met_f G¹C⁷² also had a positive effect on the growth rate of the indicator strain.

DISCUSSION

In the present study, we show that in a fmt⁺ strain, a moderate overproduction of elongation factor Tu, is enough to change the fate of tRNA^Met_f amber variants modified or not in the acceptor stem. The 1.5-fold increase in EFTu cellular concentration both improves the activity of such tRNAs in the elongation of translation and precludes their participation in the initiation process. This behavior may reflect a competition between EFTu and the formylase for the taking over of an aminoacyl-tRNA in the cell. In particular, the overproduction of EFTu is sufficient to allow the glutaminylated tRNA^Met_f amber to acquire an elongator activity. However, our results show that in the fmt⁺ context, methionyl-tRNA^Met_f amber does not participate in the elongation. This behavior probably reflects the favored stronger interaction of methionylated tRNAs with the initiation apparatus, in particular with the formylase (36). In the absence of formyltransferase activity, one can imagine that part of the nonformylated methionyl-tRNA^Met_f would become complexed with EFTu, thus lowering the amount of tRNA available for translation initiation.

The question of whether IF2 forms a complex with initiator fMet-tRNA in solution has long remained unclear. In vitro, the formation of a binary complex was reported to require the absence of Mg²⁺ (22, 37), whereas Petersen et al. reported a protective effect of IF2 on the spontaneous deacylation or digestion of fMet-tRNA^Met_f by ribonucleases in the presence of magnesium ions (20, 21). From calculations of the intracellular concentrations of IF2 and of the 30 S ribosomal subunits and the measurement of their affinity constant (24, 38), it was proposed that most of the IF2 molecules and small ribosomal subunits are associated together in vivo (19, 39). In this context, an overproduction of IF2 such as that obtained in the present study is expected to increase the fraction of free IF2. Consequently, the observed effects on the initiator or elongator activity of a tRNA may be supposed to reflect the interaction of this tRNA with free IF2 and to indirectly establish that a productive binary complex between these macromolecules occurs in the cell. In wild-type conditions, however, the formation of such a binary complex might not be the major pathway leading to IF2-tRNA recognition.

A 5-fold overproduction of IF2 results in an increase in initiator activity and in a decrease in elongator activity of all studied tRNA^Met_f amber variants. This effect is more pronounced without than with an overproduction of formyltransferase. Taken together, these results strongly suggest that overproduced IF2 interacts with the unformylated species, thereby allowing their participation in the translation initiation process. A straightforward interpretation of this behavior is that the increase in IF2 concentration enables the fraction of unformylated tRNA complexed to this factor to significantly increase.

Translation Initiation in an fmt Cell Is Likely to Occur through the Residual Recognition of Unformylated Methionyl-tRNA^Met_f

Formylation of initiator tRNA is not detectable in the fmt PAL13Tr strain (5). This result raises the question to know how the initiation of translation can occur in such a strain. The capacity of IF2 to engage nonformylated aminoacyl-tRNA^Met_f in the initiation of translation has to be considered to account for the residual growing of PAL13Tr. A decreased affinity of IF2 for unformylated methionyl-tRNA^Met_f would explain the leaky character of the growth. In agreement with this idea, the growth rate of PAL13TrFatg is decreased by an overproduction of EFTu, as measured in liquid medium (result not shown). A straightforward explanation is that sequestration of unformylated methionyl-tRNA^Met_f by the excess of EFTu decreases the concentration of free tRNA available for complexation with IF2. Surprisingly, however, an overproduction of IF2 also inhibits the growth of the PAL13TrFatg (result not shown). This unexpected result may be related to the previous report that the growth of an fmt⁺ strain is impaired by IF2 overproduction (40). IF2 in excess would start binding aminoacylated tRNAs other than the initiator species and the resulting IF2-aminoacyl-tRNA interaction, lacking the identity features for recognition by IF3, might behave as a poison of the 30 S ribosomal subunit.

In a fmt Context, the Amount of Free Aminoacyl-tRNA^Met_f Available for Initiation Limits the Growth Rate

The stimulation of the growth of the fmt strain PAL13Tr by overproduction of tRNA^Met_f sustains the above idea that in the absence of formylation, the free concentration of this tRNA limits the efficiency of translation initiation. Accordingly, tRNA^Met_f overproduction may compensate for both a misappropriation of the initiator tRNA by EFTu on one hand and the low affinity of IF2 for unformylated Met-tRNA^Met_f on other hand.

The positive effect on the growth of PAL13TrFatg of the overproduction of tRNA^Met_f allowed us to detail the features of this tRNA, which are important for initiation in the absence of formylation. The accessory character of the formylability of tRNA was particularly clear in the case of the G⁷² and G¹C⁷² species, which are not substrates of the formylase. On the contrary, the occurrence of a G²⁹G³⁰G³¹C³⁹C⁴⁰C⁴¹ motif in the anticodon stem of the overproduced tRNAs was needed to score a positive effect on growth. This tends to further establish that in a fmt context, IF3 remains crucial to select the initiator tRNA whether it is formylated or not.