Role of eukaryotic initiation factor 5 in the formation of 80 S initiation complexes.

The function of eukaryotic initiation factor 5 (eIF-5) from rabbit reticulocyte lysate has been studied by sucrose gradient preparation of 40 S and 80 S initiation complexes. eIF-5 is required for transfer of initiator tRNA from 40 S preinitiation complexes to puromycin-reactive 80 S complexes. The transfer is dependent upon GTP hydrolysis and is associated with release of eIF-2 and eIF-3 from the 40 S subunit. The GTP-dependent loss of eIF-2 and eIF-3 is catalyzed by eIF-5 in the absence of 60 S subunits or when subunit joining is prevented by edeine, but not when GTP is replaced by GuoPP(NH)P. Unstable 40 S subunit . Met-tRNAf complexes generated by eIF-5 can form puromycin-reactive 80 S complexes when 60 S subunits are added in the absence of added GTP. In addition, kinetic evidence is presented that indicates GTP hydrolysis occurs prior to 80 S complex formation.

The function of eukaryotic initiation factor 5 (eIF-5) from rabbit reticulocyte lysate has been studied by sucrose gradient preparation of 40 S and 80 S initiation complexes.
eIF-5 is required for transfer of initiator tRNA from 40 S preinitiation complexes to puromycinreactive 80 S complexes. The transfer is dependent upon GTP hydrolysis and is associated with release of eIF-2 and eIF-3 from the 40 S subunit.
The GTP-dependent loss of eIF-2 and eIF-3 is catalyzed by eIF-5 in the absence of 60 S subunits or when subunit joining is prevented by edeine, but not when GTP is replaced by GuoPP(NH)P.
Unstable 40 S subunit*Met-tRNAf complexes generated by eIF-5 can form puromycin-reactive SO S complexes when 60 S subunits are added in the absence of added GTP. In addition, kinetic evidence is presented that indicates GTP hydrolysis occurs prior to 80 S complex formation.
With the availability of reticulocyte initiation factors of high purity, it has become possible to assign specific functions or steps during the initiation process to most of the individual protein factors. eIF-5,' formerly called IF-M2A (l), has been prepared in high purity from rabbit reticulocyte ribosomal wash (2). It consists of a single peptide of approximately 125,000 daltons and is required for amino acid polymerization reactions including methionylpuromycin synthesis, hemoglobin synthesis, and polyphenylalanine synthesis. In addition, eIF-5, in the absence of other protein factors, has GTPase activity which requires 40 S and 60 S ribosomal subunits for its expression. The exact function of eIF-5 is not known, but work from several laboratories has suggested this protein has "joining" activity; that is, it is required for the joining of 60 S subunits to 40 S subunits containing template and initiator tRNA (3)(4)(5).
We have recently defined conditions which permit isolation of 40 S and 80 S ribosomal preinitiation complexes in unfixed sucrose gradients with yields of 10 to 60% of starting material (6). This has permitted us to study in greater detail the sequence of reactions leading to formation of an 80 S initiation complex in the AUG-dependent system. In this report we present evidence that eIF-5, in an energy-dependent step, releases eIF-2 and eIF-3 from the 40 S subunit independent * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertiskment" in accordance with 18 U. of, and perhaps prior to, joining of the 60 S subunit. These findings suggest that eIF-5 should not be considered as a factor which actively joins 40 S subunit complexes with 60 S subunits, but rather in a step separate from, but closely linked to, joining of the 60 S subunit to form the complete 80 S initiation complex.

MATERIALS AND METHODS
Preparation of Materials-Initiation factors from rabbit reticulocytes iysate were prepared and purified as previously described; eIF-2 (7). eIF-3 (8). eIF-4C and eIF-4D (9). eIF-5 (2 or pyruvate kinase was added. In addition, eIF-4C, -4D, -5, and 60 S subunits were added to form 80 S complexes as indicated in the appropriate legends. These components were added at levels optimal for the initial l-ml reaction mixture used to generate 40 S complexes and therefore represent a 2to S-fold excess above the amount present in the more dilute column fractions.

RESULTS
Previous studies in our laboratory had determined the requirements for the formation of 40 S preinitiation complexes, as analyzed by sucrose gradients, and 80 S initiation complexes, as analyzed by methionylpuromycin synthesis (3, 6). In order to monitor the fate of radiolabeled initiation factors during the process of 80 S initiation complex formation, conditions for the formation of such complexes were studied. The stringent requirement for eIF-5 and GTP in the formation of 80 S complexes as assayed by methionylpuromycin synthesis is also observed by sucrose gradient analysis ( Fig. 1). For this analysis, 40 S and 60 S reticulocyte ribosomal subunits were incubated with Met-tRNAf, the initiator codon AUG, initiation factors, and either the energy source GTP or the nonhydrolyzable analog of GTP, GuoPP(NH)P. The samples were placed on sucrose gradients and after centrifugation the absorbance at 254 nm was used to identify the location of the 40 S, 60 S, and 80 S subunit complexes within the sucrose gradient. In Panel A is shown an incubation containing eIF-2, eIF-3, GTP, and eIF-4C and -4D. There is an absorbance peak at the top of the gradient due to unbound tRNA and GTP. Approximately 12 pmol of Met-tRNAr are bound to the 40 S region of the gradient and in the absence of eIF-5, no Met-tRNAf is found in the 80 S peak. There is, however, a small 80 S ribosome peak by absorbance. Radiolabeled eIF-2 is also found in the 40 S subunit region of the gradient in an approx-imate 1:l stoichiometry with Met-tRNAf consistent with the idea that the ternary complex of Met-tRNAr, eIF-2, and GTP is bound to the 40 S subunit. In contrast, eIF-3 is found in two places in the gradient; at approximately 18 S where the large eIF-3 complex migrates into the gradient on its own, and also in the 40 S subunit region where approximately 25 pmol are bound to 50 pmol of 40 S ribosomal subunits. With the addition of 4 pmol of eIF-5, there is a large increase in the 80 S ribosomal peak by absorbance and, in addition, approximately 6 pmol of Met-tRNAf is found in the 80 S region (Panel B). Note however, that eIF-2 and eIF-3 are not found in the 80 S region, but remain, in lower amounts, in the 40 S region. The incomplete conversion of 40 S preinitiation complexes to 80 S initiation complexes presumably reflects a lack of excess joining components (i.e. eIF-4C, -4D, -5, and 60 S subunits). The substitution of GuoPP(NH)P for GTP prevents the formation of an 80 S initiation complex as judged by the absence of [3H]Met-tRNAr and an absorbance peak in the 80 S region. Met-tRNAf and eIF-2 remain bound to the 40 S subunits in an amount equal to or greater than the binding seen in the control in Panel A. Thus, eIF-5 is required for transfer of Met-tRNAf from 40 S preinitiation complexes to 80 S initiation complexes and there is an accompanying release of eIF-2 and eIF-3 during the conversion. The formation of an 80 S initiation complex from a 40 S complex apparently requires hydrolysis of GTP since it can be prevented by use of a nonhydrolyzable GTP analog. Thus, eIF-5 could function as a subunit joining factor with GTP hydrolysis being required for 40 S and 60 S subunit joining. Alternatively, eIF-5 might have its effect at the 40 S subunit level alone, independent of the 60 S subunit. In order to examine this question further, we have studied the effect of eIF-5 on 40 S preinitiation complexes in the absence of 60 S subunits.
The effect of eIF-5 on 40 S preinitiation complexes is shown in Fig. 2  Conditions were as specified for Fig. 1, except that unlabeled eIF-2 was used and 60 S subunits were omitted. Gradients were centrifuged for 150 min instead of the 135 min used for Fig. 1. u, incubation with GTP but omitting eIF-5; ---, incubation with 4 pmol of eIF-5 but the GTP is replaced by 0.4 ITIM GuoPP(NH)P; A. . . . A, incubation with 4 pmol of eIF-5 and 0.4 mM GTP.
breaks down the ternary complex on the 40 S subunit in an energy-dependent manner, and it functions catalytically in that 4 pmol of eIF-5 results in loss of approximately 8 to 9 pmol of Met-tRNAf from 40 S subunits. In addition, 60 S subunits are not required for the action of eIF-5 on 40 S subunit preinitiation complexes. If eIF-5 functions prior to 60 S subunit joining or independent of 60 S joining, then the existence of an intermediate 40 S subunit *Met-tRNAf complex might be suspected. Such a complex might be expected to be unstable since eIF-2 and GTP would not be present. Instability of this complex would explain the paradoxic loss of Met-tRNAf as well as eIF-2 from 40 S subunits when eIF-5 acts in the absence of 60 S subunits. We have obtained evidence for the existence of this complex by forming 40 S preinitiation complexes by a preliminary incubation at 30°C for 15 min and then adding eIF-5 and incubating for varying periods of time at 30°C. The results are presented in Table I. In the absence of eIF-5, about 13 pmol of eIF-2 and Met-tRNAf are bound to 30 pmol of 40 S subunits. If eIF-5 is added and the sample is immediately placed on ice (prior to centrifugation) about half of the eIF-2 is lost and 15% of Met-tRNAf is lost. With progressively longer incubations with eIF-5, both eIF-2 and Met-tRNAf are lost from the 40 S subunit complex but Met-tRNAf is selectively retained. This result suggests that the function of eIF-5 is to break down the 40 S subunit preinitiation complex to a 40 S subunit. Met-tRNAf complex which is itself highly unstable and survives the isolation procedure poorly. This eIF-5-directed phenomenon, which can be referred to as "an eIF-5-dependent release," occurs in the presence or absence of eIF-3 and AUG, and in all instances requires the presence of GTP rather than a nonhydrolyzable analog. An indication of the instability of the 40 Se Met-tRNAf bond is the "trailing" of the Met-tRNAf radioactivity from the 40 S region to the top of the gradient in the samples treated with eIF-5 but omitting 60 S subunits (e.g. Fig. 4).
The data shown above suggest that the requirement for eIF-5 in 80 S complex formation directed by AUG may be distinct from the direct joining of the 40 S and 60 S subunits and perhaps involves a prior independent but associated step leading to formation of a 40 S subunit e Met-tRNAf complex not containing initiation factors. It can be argued that the omission of 60 S subunits is a highly artifactual situation of no direct relevance in Go. Secondly, it can be argued that traces of 60 S subunits which may be present in a 40 S preparation may be playing an important catalytic role in our observations. Thus, we chose a second method to distinguish the eIF-5-dependent release of initiation factors from subunit joining, the use of the antibiotic edeine. Edeine has the property of preventing the joining of 40 S and 60 S ribosomal subunits (14). In Fig. 3 . When edeine is present in the absence of eIF-5,40 S subunit binding is reduced to about 3 pmol (closed triangles). Addition of eIF-5 in the presence of edeine yields little or no 40 S subunit binding of Met-tRNAf and none of the Met-tRNAf appears in the 80 S ribosome region (closed circles). Therefore, edeine prevents formation of 80 S complexes containing Met-tRNAf primarily by blocking subunit joining, but it does not prevent the labelization of Met-tRNAf that accompanies the eIF-5dependent release of eIF-2. It is also clear that edeine interferes with the binding of the ternary complex to 40 S subunits, but to a lesser extent than the blockage of subunit joining.
If eIF-5 leads to formation of a 40 S subunit *Met-tRNAf complex free of eIF-2, eIF-3, and GTP, and this complex is an intermediate leading to 80 S ribosome complex formation, it should be possible to isolate such a complex and show that it can participate in a subsequent reaction to yield a functional 80 S initiation complex. We examined the 40 S subunit. Met-tRNAf complex by harvesting it from the 40 S subunit region of a sucrose gradient and testing whether methionylpuromytin synthesis could occur by adding 60 S subunits, eIF-4C, eIF-4D, and puromycin in the presence of GuoPP(NH)P. Normally, methionylpuromycin synthesis is dependent upon eIF-2 (or eIF-2A), eIF-5, and GTP (instead of GuoPP(NH)P). The results are presented in Fig. 4. Panel A is a control sample in which 40 S and 60 S subunits, eIF-2, -3, -4C, -4D, and -5, and GTP were incubated.
The sample was centrifuged through a sucrose gradient containing GTP. The centrifugation was for I35 min in Panel A and 150 min in Panels B and C. The 80 S peak (Panel A) was harvested and puromycin was added to the sample. Salts and [Mg'+] were adjusted to conditions optimal for methionylpuromycin synthesis and the volume was increased to overcome the inhibitory effect of sucrose. The picomoles of methionylpuromycin synthesized are shown by the vertical bar. Forty-seven per cent of the radioactivity in the 80 S peak was converted to methionylpuromycin; most of the remaining radioactivity was present as free methionine as a result of nonenzymatic deacylation. In a similar experiment, the same initiation factors were incubated with just 40 S subunits and subjected to centrifugation in a gradient containing GuoPP(NH)P (to block any subsequent effect of eIF-5). When 60 S subunits, puromycin and eIF-4C and -4D were added to the three peak fractions, methionylpuromycin synthesis occurred to the extent of 16 to 38% of the Met-tRNAf contained in the gradient fractions. Since GuoPP(NH)P and not GTP was in the gradient, it appears that methionylpuromycin synthesis and presumably 80 S initiation complex formation do not require GTP hydrolysis for subunit joining per se, although the prerequisite "eIF-5-dependent release" of eIF-2 and eIF-3 was dependent on GTP hydrolysis (Fig. 1). To evaluate the possibility that eIF-5 was carried into the gradient fractions with the 40 S complex, ['*C]eIF-5 was applied to a separate tube containing 40 S preinitiation complexes. The amount of eIF-5 in gradient fractions from that tube are plotted in Panel B. Clearly, less than 0.1 pmol of eIF-5 could be expected to be in each gradient fraction. The experiment in Panel C is the same as that presented in Panel B except that GuoPP(NH)P replaced GTP in the incubation media. The eIF-5-mediated release of eIF-2 was blocked and 40 S subunit complexes containing Met-tRNAf and ["'C]eIF-2 were obtained. When 60 S subunits eIF-4C, -4D, and puromycin were added to isolated gradient fractions, very little methionylpuromycin was synthesized, amounting to 4 to 5% of the Met-tRNAf contained in the two highest gradient fractions. Thus, when eIF-5 function was pmol of 60 S subunits, 4 pg of eIF-2,25 pg of eIF-3, and 28 pg of mixed eIF-4C and -4D were incubated for 15 min. APL& control without edeine or eIF-5; A-A, addition of 10 pM edeine A, no eIF-5; M, no edeine, + 3 pg of eIF-5; M, 10 pM edeine A + 3 pg of eIF-5. blocked by GuoPP(NH)P, the 40 S preinitiation complexes formed were unable to combine with 60 S subunits to synthesize methionylpuromycin.
These data suggest that a 40 S subunit. Met-tRNAf complex, free of eIF-2, eIF-3, and eIF-5 is nearly as efficient as 80 S ribosomal complexes in methionylpuromycin synthesis when 60 S subunits and eIF-4C and -4D are added. Furthermore, GTP hydrolysis does not appear to be required after eIF-5 has mediated the release of eIF-2 and eIF-3.
While the above experiments gave insight into the function of eIF-5 in protein synthesis initiation, two problems were encountered.
First, due to the loss of radiolabeled GTP during complex isolation on sucrose density gradients, the hydrolysis of GTP could only be inferred by comparative studies with GTP and GuoPP(NH)P.
Second, kinetic studies on the conversion of 40 S complexes to 80 S complexes were not possible as the isolation of 40 S complexes was slow and contained high concentrations of sucrose. To circumvent these problems, 40 S preinitiation complexes were isolated by gel filtration (Fig. 5). It can be seen that such 40 S preinitiation complexes contain approximately equimolar amounts of [y-"'P]GTP and [14C]Met-tRNAf.
While these complexes contain free 40 S subunits, the other components (eIF-2, GTP, and Met-tRNAf) were eluted in subsequent column fractions.
In order to follow the conversion of the 40 S preinitiation complexes to 80 S initiation complexes, eIF-4C, -4D, -5, and 60 S subunits were incubated with the isolated 40 S preinitiation complexes and the formation of 80 S complexes was determined as sensitivity of the [14C]Met-tRNAf to puromytin. At the same time, the hydrolysis of GTP was determined (as organic extractable phosphate). The results of two such experiments are presented in Table II. From an initial input of 40 S complexes with a slight molar excess of Met-tRN& over GTP, approximately one-half of these complexes were converted to 80 S complexes (as judged by methionylpuromycin synthesis) and there was slightly less than an equimolar hydrolysis of GTP (as judged by phosphate released from GTP).
A kinetic analysis of the formation of 80 S initiation complexes from isolated 40 S preinitiation complexes is presented in Fig. 6. As is apparent GTP hydrolysis is rapid with an appreciable degree of GTP hydrolysis occurring at 0°C during the time required to add the "joining components" (eIF-4C, -4D, -5, and 60 S subunits), mix the solutions, and transfer to 37°C. On the other hand, methionylpuromycin synthesis is linear from zero time and is about 2-to 4-fold slower than GTP hydrolysis. Control experiments (data not shown) indicated the following: 1) in the absence of the "joining components," the rate of GTP hydrolysis is 15 to 20% of the rate for the complete system; 2) the reaction of puromycin with 80 S initiation complexes is much more rapid (7-to lo-fold at least) than the rate of subunit joining. Thus, from the kinetic data in Fig. 6 and the controls cited above, it would appear that GTP hydrolysis precedes subunit joining. DISCUSSION The factors required and the sequence of steps in the formation of an 80 S initiation complex are gradually being defined through the work of many laboratories.
The preliminary step appears to be the formation of a soluble ternary complex which contains eIF-2, [15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32]. Despite the general agreement on this point, there are reports that indicate two different observations on ternary complex formation (with highly purified eIF-2). The fist group has observed that ternary complex formation is rapid, relatively insensitive to [Mg2+] and that the ternary complex can bind directly to 40 S subunits (3-5). The second group has found that ternary complex formation is slow, highly sensitive to [Mg2+] and may be stimulated by an additional protein(s) to enhance the yield of ternary complex (23,33-36). Further considerations will be based on the observations of the first group (which are the basis for this study) as at present it is not possible to reconcile the differences between the two groups of reports.
The next general step in protein synthesis initiation is the binding of mRNA to the Met-tRNAr.40 S subunit complex. This step is catalyzed by eIF-1, eIF-4A, and eIF-4B (4,5) and requires the utilization of ATP (4, 5, 37, 38). To date, there has been no report of any similar requirement when AUG codon or other synthetic mRNAs are used as template. Transfer of initiator tRNA from 40 S subunit complexes to 80 S initiation complexes requires 60 S subunits, GTP, and is catalyzed by eIF-4C and eIF-5 (3,4, 6). During this last step, eIF-2 and eIF-3 are released from the 40 S subunit complexes (Refs. 3, 4, and 6, and Fig. 1). In this report, the role of eIF-5 in the initiation process is defined further. The initial observation with eIF-5 indicated that its presence could catalyze the release of eIF-2 and eIF-3 from 40 S preinitiation complexes in a reaction that required the hydrolysis of GTP (Ref. 6 and Fig. 2). The release of eIF-2 from 40 S preinitiation complexes appears to yield an unstable Met-tRNAr.40 S subunit complex which decomposes easily during the isolation procedure (Table I and Fig. 2). A similar observation had been made previously (39, 40), but had not been correlated with the loss of eIF-2 from 40 S preinitiation complexes. The unstable Met-tRNAr. 40 S subunit complex can combine with 60 S subunits to form an 80 S initiation complex in the absence of added eIF-5 or GTP (Fig. 4).
Studies with 40 S preinitiation complexes isolated by gel filtration yielded additional information on the role of eIF-5. The data in Table II and Fig. 6 are consistent with the interpretation that eIF-5 induces a hydrolysis of eIF-2-bound GTP which is associated with (if not the cause of) the eIF-5mediated release of eIF-2 from 40 S preinitiation complexes. Additional supporting evidence for this interpretation has been obtained from comparative kinetic studies on methionylpuromycin synthesis using either eIF-2 or eIF-2A to direct Met-tRNAr binding to 40 S subunits (41). This interpretation is also consistent with the observation that GTP hydrolysis precedes 80 S complex formation (as judged by methionylpuromycin synthesis, Fig. 6). A previous report from this laboratory had indicated that eIF-5 mediated the hydrolysis of GTP only in the presence of both 40 S and 60 S subunits (2). Consistent with this finding is data obtained using gelfiltered 40 S preinitiation complexes which demonstrated that the eIF-5-mediated hydrolysis of GTP in 40 S preinitiation complexes occurs more rapidly in the presence of 60 S subunits (41). Thus it would seem that although eIF-5 may mediate hydrolysis of GTP in 40 S preinitiation complexes, the kinetic preference is for such hydrolysis in the presence of 60 S subunits. Exactly how this occurs and why GTP hydrolysis should so markedly precede methionylpuromycin synthesis are subjects for further research. 30. Merrick, W. C. (1979) Methods Enzymol. 60, 108-123 Safer, B., Kemper, W., andJagus, R. (1978) J. Biol. Chem. 253, 3384-3386 Walthall,