In Vitro Study of Two Dominant Inhibitory GTPase Mutants of Escherichia coli Translation Initiation Factor IF2

We have recently shown that the Escherichia coli initiation factor 2 (IF2) G-domain mutants V400G and H448E do not support cell survival and have a strong negative effect on growth even in the presence of wild-type IF2. We have isolated both mutant proteins and performed an in vitro study of their main functions. The affinity of both mutant proteins for GTP is almost unchanged compared with wild-type IF2. However, the uncoupled GTPase activity of the V400G and H448E mutants is severely impaired, theV max values being 11- and 40-fold lower, respectively. Both mutant forms promoted fMet-tRNAf Met binding to 70 S ribosomes with similar efficiencies and were as sensitive to competitive inhibition by GDP as wild-type IF2. Formation of the first peptide bond, as measured by the puromycin reaction, was completely inhibited in the presence of the H448E mutant but still significant in the case of the V400G mutant. Sucrose density gradient centrifugation revealed that, in contrast to wild-type IF2, both mutant proteins stay blocked on the ribosome after formation of the 70 S initiation complex. This probably explains their dominant negative effect in vivo. Our results underline the importance of GTP hydrolysis for the recycling of IF2.

The first step of prokaryotic mRNA translation requires at least 3 proteins known as the initiation factors IF1, 1 IF2, IF3, and a complexed GTP molecule, in addition to fMet-tRNA f Met (1). Of these three proteins only IF2 has a GTP-binding domain. During 30 S initiation complex formation, the initiation factors, GTP, fMet-tRNA f Met , and mRNA are bound to the small ribosomal subunit. In this context IF2 binds at least three different ligands: GTP, the 30 S ribosomal subunit, and the initiator tRNA (1)(2)(3). The functional 70 S initiation complex ready for elongation contains only the initiator tRNA and mRNA. During its formation the initiation factors are released and GTP is hydrolyzed. The presence and hydrolysis of GTP which are directly linked to the presence of IF2 have been shown to be essential for initiation of protein biosynthesis (4). However, the precise role of GTP is still not clear. IF2 is one of the largest G-proteins known, and, in contrast to other proteins of this class which carry the G-domain in their N-terminal region, the G-domain of IF2 is centrally located in the protein. IF2 has been shown to bind GTP with a 10-fold lower affinity than GDP, in contrast to EF-Tu which binds GTP 100 times less efficiently than GDP (5,6). Its lower overall affinity for guanine nucleotides (1000-fold for GDP, 10-fold for GTP) might enable IF2 to self-cycle from the GDP to the GTP state. Free IF2 does not show any GTPase activity; this is induced upon binding of the 50 S subunit during 70 S complex formation. The IF2 GTPase depends completely on the presence of the ribosome (7,8). However, in Bacillus stearothermophilus, both IF2 and its isolated G-domain were shown to be capable of hydrolyzing GTP in the absence of ribosomes when 20% ethanol was included in the reaction (9).
In order to define the E. coli IF2 G-domain more precisely, we had previously mutated very conserved residues suspected to be involved in GTP binding or hydrolysis and assessed the effect of these mutant proteins in vivo in a strain carrying a null mutation in the chromosomal copy of infB (4). The different mutations globally confirmed a theoretical three-dimensional model of the IF2 G-domain (10) and demonstrated the crucial importance of GTP hydrolysis in translation initiation. Six out of seven mutations in that study were lethal and of those, two (Val 400 to Gly and His 448 to Glu) exhibited a strong dominant inhibitory effect over the wild-type protein (4).
An alignment of the primary structures of several G-domains reveals that the valine 400 residue of IF2 is equivalent to valine at position 20 in E. coli EF-Tu and, from a structural point of view, corresponds to glycine at position 12 in the mammalian protooncogenic Ha-ras protein, p21. Indeed, these residues are found within the G1 region (GXXXXGK), a consensus element of the G-domain specifically involved in the interaction with the ␣ and ␤ phosphates of GDP/GTP (11)(12)(13). A V20G substitution in E. coli EF-Tu (analogous to the V400G mutation in IF2) causes a 5-10-fold reduction in GTPase activity and impairs its stimulation by the ribosome (14,15). The histidine 448 residue of IF2, the last residue of the second consensus element (DXXGH) constituting the G-domain, is well conserved in initiation and elongation factors of different organisms (16,17). Mutations of the corresponding residue of EF-Tu (His 84 ) underline the importance of this residue for polypeptide synthesis and GTP hydrolysis (18,19) but it is not yet clear how this residue, which is not in direct contact with the ␥-phosphate of GTP, participates in GTP hydrolysis (20).
The work presented here follows up on our in vivo studies and describes the major biochemical and enzymatic characteristics of the V400G and H448E IF2 mutants in vitro. Based on their impaired GTPase activity and their incapacity to leave the initiation complex and permit elongation, we present a rationale for the dominant inhibitory effect of the two mutant IF2 proteins.

EXPERIMENTAL PROCEDURES
Strains and Media-Strains and plasmids used are described in Table I. All strains were grown in Luria Bertani (LB) medium. Where indicated, ampicillin (Amp) was added at 100 g/ml, chloramphenicol (Cm) at 10 g/ml, and spectinomycin at 100 g/ml.
Plasmid Constructs-Plasmids overexpressing infB mutant alleles were constructed by exchanging a 1135-base pair SnaB1-SstI DNA fragment in pSL4 (21) with the corresponding fragments of pB18V400G and pB18H448E (4), giving rise to pSL4V400G and pSL4H448E. In these plasmids, expression of the mutant infB genes is under control of the thermoinducible pL and pR promotors. The correct insertion and the presence of the mutation in the inserted fragments were confirmed by DNA sequencing (22).
Construction of Strain SL679R-In order to purify the mutated IF2␣ proteins without contamination of the wild-type protein we constructed a strain (SL679R) that expresses only the short form (␤) of the factor. The 1.6-kilobase BglII-HindIII fragment of pB18⌬45 (23), carrying the mutant infB gene expressing only the ␤ form of IF2 was cloned into the low copy number plasmid pCL1920 (24) digested with the same enzymes. The resulting plasmid, pCL⌬infB45, was used to transform the strain SL598R (21). Curing of the infB transducing phage was performed as described in Ref. 21. This gave rise to strain SL679R that expresses only the short form of the factor.
Protein Purification-Strain SL679R was transformed with the plasmids pSL4V400G and pSL4H448E. The strains were grown at 30°C in LB medium in the presence of ampicillin and chloramphenicol to an OD 550 ϭ 1, then the temperature was shifted to 42°C to allow expression of the mutant proteins from the thermoinducible pLpR promoters and incubation was carried on for 2 more hours. The culture was chilled on ice, the cells harvested, washed, and kept frozen at Ϫ80°C. Overexpression of the mutant IF2 proteins was monitored by 10% SDSpolyacrylamide gel electrophoresis gels and immunoblotting of aliquots at different times during heat induction.
Cell disruption and preparation of a crude extract was performed as described (25). A 50% ammonium sulfate precipitate was resuspended in LSB buffer (20 mM Hepes-KOH, pH 7.2, 1 mM MgOAc, 7 mM ␤-mercaptoethanol, 5% glycerol) and dialyzed against the same buffer for 2 ϫ 1 h. Proteins were then loaded on a Mono Q column (Pharmacia) and eluted with a 200 -500 mM KCl gradient in LSB buffer. Fractions containing IF2␣ protein were pooled and, after lowering the salt concentration to about 100 mM by dilution with LSB buffer, loaded on a Mono S column (Pharmacia) and separated by a 200 -400 mM KCl gradient in LSB buffer.
Preparation of Ribosomes-Ribosome tight couples (70 S) and ribosomal subunits (30 S and 50 S) were isolated from E. coli MRE600 cells, freshly harvested from exponential growth phase, based on methods described in detail elsewhere (26) except that the ribosomal subunits were fractionated on a 10 -30% (w/w) hyperbolic sucrose gradient in a Beckmann SW28 rotor at 27,000 rpm at 4°C for 14 h.
Uncoupled GTPase Activity of IF2-IF2-catalyzed GTP hydrolysis was measured as described (10,27). Reactions (25 l) were stopped with a mixture of 100 l of 1 M perchloric acid and 1 ml of 1 M KH 2 PO 4 and the amount of [ 32 P]phosphate liberated during the reaction was determined as described previously (28). Kinetic constants were determined from Lineweaver-Burk plots based on initial rate measurements carried out with [␥-32 P]GTP concentrations of 5-50 M, which is in the K m range of wild-type IF2.
IF2-dependent Binding of fMet-tRNA f Met to 70 S Ribosomes-Reactions contained limiting amounts of IF2 with respect to 30 S and 50 S ribosomal subunits, initiation factors IF1 and IF3, polyAUG and labeled f[ 3 H]Met-tRNA f Met . The exact concentrations are given in the legend to Fig. 3. The reactions were incubated at 25°C for 5 min. Ribosome bound fMet-tRNA f Met was measured by filter binding assay (29). The effect of GDP on the fMet-tRNA f Met binding reaction was studied by keeping the GTP concentration in the reaction constant at 0.1 mM and varying the GDP concentration in steps from 0 to 10 mM GDP (0.01, 0.1, 1, and 10 mM).

Sucrose Gradient Fractionation of the Puromycin Reactions-The
fMet-tRNA f Met binding reaction was carried out for 15 min and the puromycin reaction for 20 min. The reaction buffer was the same as described for fMet-tRNA f Met binding except that the magnesium acetate concentration was 10 mM. The reactions were diluted with 100 l of cold reaction buffer, chilled on ice, and loaded on a 7-20% sucrose gradient in the same buffer. The gradients were centrifuged at 23,000 rpm in a Beckmann SW41 rotor for 12.5 h. 500-l fractions were collected and proteins were precipitated with trichloracetic acid and dissolved in SDS-polyacrylamide gel electrophoresis sample buffer. The fractions were separated on 10% polyacrylamide gels and analyzed by Western blotting with anti-IF2 antibodies.

RESULTS
Purification of Mutant Forms of IF2-The E. coli translation initiation factor IF2 exists as two forms in the cell, IF2␣ and IF2␤, which support bacterial growth equally well under normal growth conditions (23). In order to purify the large form, IF2␣, of two IF2 mutants (V400G and H448E) without contamination by wild-type protein we constructed a strain (SL679R ,  Table I) expressing only the short form IF2␤. This strain is a derivative of SL598R (21) and carries the same replacement of the chromosomal copy of the infB gene by a cat cassette. Survival of the cell is assured by the recombinant plasmid pCL⌬infB45 expressing solely the ␤ form of the factor. The mutated infB alleles are supplied by a second compatible plasmid and their expression is controlled by the thermoinducible promoters pRpL. Expression of the mutant proteins was induced by shifting the temperature of the culture from 30°C to 42°C. The ␣-form of both mutant proteins was purified from these strains by fast protein liquid chromatography techniques described under "Experimental Procedures" and tested in a variety of in vitro assays.
Uncoupled GTPase Activity-As shown previously (4), both IF2 mutations, V400G and H448E, are lethal to E. coli. To determine whether they were affected in their ability to hydrolyze GTP in vitro, we performed an uncoupled GTPase assay, where the initial rate of [␥-32 P]GTP hydrolysis by IF2 was measured in the presence of 70 S ribosomes, but in the absence of any tRNA or mRNA. As shown in Fig. 1, both mutants have practically lost their ability to hydrolyze GTP.
We also measured the initial rates of GTP hydrolysis by wild-type and mutant forms of IF2 as a function of increasing GTP concentration. The resulting double-reciprocal plot is shown in Fig. 2. Determination of the K m values derived from Fig. 2 revealed that the affinity of neither of the mutant proteins for GTP is reduced compared with wild-type IF2 (Table  II). On the contrary, the respective K m values are slightly lower than that of the wild-type protein, especially in the case of the V400G mutant (15.3 M versus 30 M for the wild-type protein). This work pCL⌬infB45 pCL1920 rep, spc/str, ⌬infB45 This work The velocity of the uncoupled GTPase activity of the mutant proteins is strongly reduced compared with that of the wildtype IF2, however. As determined from Lineweaver-Burk plots (Fig. 2), the V400G and the H448E mutants exhibit 11.4-and 40-fold lower V max values than the wild-type protein, respectively (Table II). This clearly indicates that it is the catalytic activity, and not the ability to bind the substrate, that is impaired in the mutants. On the other hand, this loss of activity is not sufficient in itself to explain their dominant negative behavior over the wild-type protein in vivo.

Binding of fMet-tRNA f
Met to the Ribosome-The main role of IF2 in translation initiation is to promote fMet-tRNA f Met binding to the ribosome. We tested the ability of both mutant proteins to promote fMet-tRNA f Met binding to 70 S ribosomes (30 S ϩ 50 S subunits) in the presence of translation initiation factors IF1 and IF3. Fig. 3 shows the initial rate of factor-dependent fMet-tRNA f Met binding for the different IF2 proteins. The V400G mutant is 1.5-fold more active and the H448E mutant 1.3-fold less active than the wild-type IF2. We also measured the efficiency of fMet-tRNA f Met binding as a function of GTP concentration. The capacity to promote this reaction increases for all three proteins in the 0 to 1 mM GTP range (Fig.  4). In this assay the activity of the V400G mutant is slightly lower than that observed with wild-type IF2. The H448E mutant promotes about 60% of tRNA binding compared with the wild-type protein (Fig. 4), but this level still represents a full stoichiometric binding of fMet-tRNA f Met . As described above, the binding of GTP to the mutant IF2 proteins as analyzed by K m measurements of the uncoupled GTPase activity is not impaired. In order to detect potential differences in the sensitivity of the mutant proteins to the presence of GDP we measured the inhibitory effect of various GDP concentrations on fMet-tRNA f Met binding in the presence of a constant GTP concentration (0.1 mM). The results are summarized in Table III. The two mutant IF2s appear to be slightly more inhibited at very low GDP concentrations, but the overall pattern of inhibition is practically identical for all three proteins.
Puromycin Reaction-The correct positioning of fMet-tRNA f Met is promoted by IF2 and IF3 and is essential for peptide bond formation during the first round of elongation. We measured the capacity of the mutant proteins to promote formation of the first peptide bond in the puromycin reaction, as a function of the GTP concentration (Fig. 5). Upon addition of GTP, the wild-type protein and the V400G mutant quickly reach their full activity with an only slight increase in efficiency in the 0.25 to 1 mM GTP range. In contrast, no formation of fMet-puromycin can be observed, regardless of the GTP concentration, when the initiation complex was formed with the H448E IF2 mutant (Fig. 5).
If one relates these data to the quantity of fMet-tRNA f Met bound to ribosome it becomes apparent that in the presence of the V400G mutant almost 100% of the fMet-tRNA f Met is converted to fMet-puromycin (compare Figs. 4 and 5). On the other hand, based on the quantity of fMet-tRNA f Met initially bound, more than twice as much puromycin-fMet-tRNA f Met is formed in the presence of wild-type IF2. This suggests that IF2 and hence ribosome recycling is more efficient with wild-type factor.
Recycling of IF2-GTP hydrolysis is believed to be essential for the recycling of IF2. We thus tested to see whether the wild-type and the mutants proteins were still associated with the ribosome after the puromycin reaction. This was done by separating the ribosomes contained in the reaction mixture on sucrose gradients and by analyzing the different fractions for the presence or absence of IF2 by Western blot (Fig. 6). In the case of the wild-type protein, practically all the IF2 had left the ribosome after the puromycin reaction and was present in the top fractions of the gradient, where unbound proteins are to be  found. In contrast, both mutant proteins stay bound to the ribosome and are found in the 30 S, 50 S, and 70 S regions of the gradient. We thus conclude that neither of the mutant factors are efficiently recycled following formation of the first peptide bond.

DISCUSSION
The role of GTP and its hydrolysis during translation initiation has never been clear. Earlier in vitro studies, using non-hydrolyzable GTP analogs, had suggested that GTP hydrolysis may be important for the release of IF2 from the 70 S initiation complex (30 -33). We wished to carry out a mutational analysis to test this hypothesis, to precisely define the amino acid residues involved in GTP hydrolysis and to correlate the in vivo phenotype of a mutant protein to its characteristics in vitro. If the analogy between the EF-Tu G-domain structure and the three-dimensional model for the IF2 G-domain holds true, then the IF2 V400G and H448E mutants should only be impaired in their catalytic (GTPase) activity but not in their affinity for the substrate as is the case for equivalent (V20G) or similar (H84G) substitutions in EF-Tu (15,19). This is exactly what we observed, the V max values for the V400G and H448E mutants being 11.4-and 40-fold lower than that observed for the wildtype GTPase, respectively. At the same time, the affinity for GTP was not adversely affected, but rather slightly increased, up to 2-fold in the case of the V400G mutant (Table II). It is interesting to note that the equivalent mutation in EF-Tu (V20G) led, on the contrary, to a less efficient binding of GTP (15). The V20G mutation in EF-Tu caused a 5-10-fold decrease in GTPase activity and impaired its stimulation by the ribo-   5. Effect of GTP concentration on the formation of fMetpuromycin in 70 S initiation complexes containing wild-type or mutant IF2 proteins. Following the incubation at 25°C for 5 min, puromycin was added to a final concentration of 1 mM to the reactions of the fMet-tRNA f Met binding assay shown in Fig. 4, and incubated for another 5 min. Formation of fMet-puromycin was assayed as described (43). Values in Figs. 4 and 5 are from the same experiment and are directly comparable. The background value due to formation of f[ 3 H]Met-puromycin in the absence of IF2 was subtracted. f, wild-type IF2; q, IF2 V400G; OE, IF2 H448E. some (14,15). The impaired GTPase activity of the IF2 V400G mutant thus indicates that the capacity to hydrolyze GTP, although totally dependent on the presence of the 50 S ribosomal subunit, resides within the IF2 protein itself.
The main function of IF2 in the cell is the binding and correct positioning of fMet-tRNA f Met on the 70 S ribosome for the subsequent elongation step. As shown in Fig. 3 both wild-type and mutant proteins promote fMet-tRNA f Met binding to the ribosome in a GTP-dependent manner. In the absence of GTP, very little or no binding was observed. Binding of fMet-tRNA f Met to the ribosome is especially sensitive to the presence of GTP when IF2 is present at low (catalytic) concentrations, as is the case in the experiments reported here (32)(33)(34)(35). Both mutant proteins promote efficient fMet-tRNA f Met binding to the ribosome (Fig. 4). A comparison of initial rate measurements in the presence of 1 mM GTP (Fig. 3) revealed slight differences, with the V400G mutant being 1.5 times faster than the wild-type protein. On the other hand, the H448E mutant was somewhat slower (1.3-fold) than the wild-type protein. At optimal GTP concentrations (1 mM), the H448E mutant is still able to bind fMet-tRNA f Met in stoichiometric quantities. The wild-type protein binds fMet-tRNA f Met most efficiently, due to factor recycling, but the V400G mutant also seems to exhibit some catalytic activity (Fig. 4).
It was observed previously (6) that, at low concentrations of wild-type IF2 and 30 S subunits, GDP acts as a competitive inhibitor of 30 S initiation complex formation. We obtained comparable results when measuring the formation of 70 S initiation complex in the presence of limiting amounts of IF2. The capacity of both wild-type and mutant proteins to promote fMet-tRNA f Met binding to 70 S ribosomes was inhibited by GDP to roughly the same extent in all three cases (Table III). This suggests that the overall affinities for GDP and GTP are not significantly altered in either IF2 mutant relative to the wildtype protein.
Since the mutant IF2 proteins mediate fMet-tRNA f Met binding to the ribosome, this cannot account for their incapacity to support growth. However, a difference between the wild-type and the mutant proteins was observed in the formation of the first peptide bond, measured by the puromycin reaction. Due to factor recycling, wild-type IF2 leads to the formation of 2.5 times more fMet-puromycin than fMet-tRNA f Met initially bound (compare Figs. 4 and 5). The H448E mutant is completely inactive in promoting the formation of fMet-puromycin, the V400G mutant allows the stoichiometric conversion of all bound fMet-tRNA f Met into fMet-puromycin, indicating that it is not recycled. The activity of the IF2 V400G mutant can be explained in two ways. Either the residual GTPase activity is sufficient for some factor release to occur, or the formation of fMet-puromycin can still take place even with the factor bound to the ribosome. While we cannot rule out that some GTP hydrolysis contributes to the overall activity observed, we rather favor the second possibility. Formation of fMet-puromycin was assayed under conditions close to those used for determining the initial rates of the uncoupled GTPase (i.e. reaction time, IF2:ribosome ratio) and where neither of the mutant proteins showed detectable GTPase activity (Fig. 1). Moreover, puromycin is a very small molecule compared with the EF-Tu⅐GTP⅐aa-tRNA ternary complex normally involved in the formation of the first peptide bond and might still be able to react with fMet-tRNA f Met despite the presence of IF2. Experiments with a non-hydrolyzable GTP analog (GDPCH 2 P) support this possibility. First peptide bond formation in the presence of wild-type IF2 and GDPCH 2 P still occurred substantially when assayed via the formation of fMet-puromycin (36). Thus we believe that IF2 V400G does not hydrolyze GTP and leave the ribosome but still permits near stoichiometric conversion of bound fMet-tRNA f Met into fMet-puromycin (compare Figs. 4 and 5). In the case of the H448E mutant we speculate that the fMet-tRNA f Met is not correctly positioned with respect to the peptidyl-hydrolase center, or its conformation is different to that of the wild-type or the V400G IF2, such that the amino acid moiety on the tRNA is shielded, thereby inhibiting the puromycin reaction. The differential behavior of the two GTPase mutants also suggests that there is no direct link between GTP hydrolysis and the first peptide bond formation. This conclusion is supported by previous experiments where GTP was removed from the 30 S initiation complex without altering the ability of the subsequent 70 S complex to participate in peptide bond formation (33).
Studies in the 1970's, using non-hydrolyzable GTP analogues (30 -33) have led to the current belief that GTP hydrolysis is necessary for the rapid release of IF2. Our experiments strongly support this view. When the puromycin reactions are subjected to sucrose density gradient centrifugation it is appar-FIG. 6. Binding of wild-type and mutant IF2 proteins to ribosomes after engagement of the 70 S initiation complex in peptide elongation. Puromycin reactions were carried out in the presence of 1 mM GTP and immediately subjected to sucrose density gradient centrifugation as described under "Experimental Procedures." After gradient fractionation, the ribosomal profile (top of the figure) was obtained by measuring the OD at 260 nm. Individual fractions were submitted to SDS-gel electrophoresis and tested for the presence of IF2 by Western blot. ent that only the wild-type IF2 leaves the ribosome. Several conclusions can be drawn from this observation. First of all, the fact that the mutant factors are not released from the ribosome after formation of fMet-puromycin provides a good explanation for their dominant negative behavior in vivo. Translation factordependent GTP hydrolysis involves an interaction with the ribosomal GTPase center on the 50 S subunit which encompasses the binding sites for L11 and the pentameric protein complex L10⅐(L12) 4 (37,38). IF2 and EF-Tu have both been shown to interact with this region of the ribosome (39 -42). An IF2 molecule which stays blocked on the ribosome would inhibit binding of the ternary EF-Tu⅐GTP⅐aa-tRNA complex and consequently subsequent elongation steps. The difference observed between the two mutants in the puromycin reaction can only account for the correct positioning of the fMet-tRNA on the ribosome, but not for its ability to promote formation of the first peptide bond.
The question remains why the dominant negative effect of the IF2 V400G mutant is stronger than that of the H448E mutant (4), despite a slightly higher residual GTPase activity. A likely answer is provided by the sucrose gradient experiment. After centrifugation, all of the IF2 V400G cosediments with the different ribosomal particles and none has dissociated sufficiently from the 70 S ribosome to be found at the top of the gradient. On the other hand, the H448E mutant clearly dissociates more easily as judged by a steady increase of the IF2 concentration toward the 30 S fractions and the top of the gradient. In the case of the V400G mutant at least, the presence of factor in the 50 S region of the gradient reflects, to some extent, binding of IF2 to this ribosomal subunit. Recentrifugation of the different gradient fractions through a sucrose cushion revealed some IF2 V400G associated with the 50 S subunit (data not shown). The increased ability of the H448E mutant to dissociate from the ribosome, despite its practically inexistent GTPase activity, probably explains why this mutant is less inhibitory to bacterial growth in the presence of wild-type IF2. This observation also underlines the importance of the conformational state of IF2 for binding to the ribosome. Independent of the GTPase activity, even slight alterations of the G-domain can have important consequences for the rest of the protein, its affinity to the ribosome and the positioning and reactivity of the fMet-tRNA f Met . The weak, but inverse effect on the affinity of IF2 for GTP, relative to the corresponding V20G mutation in EF-Tu, highlights some structural differences between the two factors. Nevertheless, our results essentially confirm the close functional relationship between the EF-Tu and IF2 G-domains and show the importance of the Val 400 and His 448 residues for GTP hydrolysis and IF2 recycling during formation of the first peptide bond.