Mutually Cooperative Binding of Eukaryotic Translation Initiation Factor (eIF) 3 and eIF4A to Human eIF4G-1*

Eukaryotic translation initiation factor 4G-1 (eIF4G) plays a critical role in the recruitment of mRNA to the 43 S preinitiation complex. The central region of eIF4G binds the ATP-dependent RNA helicase eIF4A, the 40 S binding factor eIF3, and RNA. In the present work, we have further characterized the binding properties of the central region of human eIF4G. Both titration and competition experiments were consistent with a 1:1 stoichiometry for eIF3 binding. Surface plasmon resonance studies showed that three recombinant eIF4G fragments corresponding to amino acids 642–1560, 613–1078, and 975–1078 bound eIF3 with similar kinetics. A dissociation equilibrium constant of ∼42 nm was derived from an association rate constant of 3.9 × 104 m −1 s−1 and dissociation rate constant of 1.5 × 10−3 s−1. Thus, the eIF3-binding region is included within amino acid residues 975–1078. This region does not overlap with the RNA-binding site, which suggests that eIF3 binds eIF4G directly and not through an RNA bridge, or the central eIF4A-binding site. Surprisingly, the binding of eIF3 and eIF4A to the central region was mutually cooperative; eIF3 binding to eIF4G increased 4-fold in the presence of eIF4A, and conversely, eIF4A binding to the central (but not COOH-terminal) region of eIF4G increased 2.4-fold in the presence of eIF3.

The initiation of translation in eukaryotes requires multiple initiation factors that stimulate the binding of mRNA and Met-tRNA i 1 to the 40 S ribosomal subunit to form the 48 S preinitiation complex (1). The binding of Met-tRNA i occurs as a ternary complex with eIF2 and GTP. The binding of mRNA is stimulated by the eIF4 factors (eIF4A, eIF4B, eIF4E, and eIF4G). Joining of the 60 S subunit to form the 80 S initiation complex requires hydrolysis of the GTP bound to eIF2, dissociation of the ternary complex, and release of the eIF2⅐GDP binary complex. eIF5 and eIF5B promote these events by stimulating GTP hydrolysis within the ternary complex bound to the 40 S ribosomal subunit (2). eIF1 and eIF1A act synergistically to mediate assembly of initiation complexes at the initiation codon (3).
Mammalian eIF4G can be divided into three domains, roughly corresponding to cleavage site by picornaviral 2A proteases (39,24). The NH 2 -terminal one-third contains the eIF4E-and PABP-binding sites. The central domain contains the binding sites for eIF3, RNA, and one of the two sites for eIF4A. The COOH-terminal domain contains a second eIF4Abinding site and also a site for Mnk1. The central region of eIF4G serves as autonomous "ribosome recruitment core" in vivo (40) and in vitro (38), whereas the COOH-terminal domain has been proposed to serve as regulatory domain (41,42).
Despite the exceptional progress that has been made in identifying ligands for this pivotal initiation factor, little is * This work was supported by Grant GM20818 from the National Institute of General Medical Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: Met-tRNA i , methionyl initiator tRNA; eIF, eukaryotic initiation factor; aa, amino acid residue; BR, molar binding ratio of eIF4A or eIF3 to eIF4G fragments; EMCV IRES, encephalomyocarditis virus internal ribosomal entry site; PABP, poly(A)binding protein; PAGE, polyacrylamide gel electrophoresis; RU, response unit(s); SPR, surface plasmon resonance. known about whether binding of one ligand to eIF4G influences the binding of others. Such knowledge may provide insight into the ordered series of events that results in proper placement of Met-tRNA i at the initiation codon. As a first step toward understanding the relationships between the various eIF4G ligands, we have more precisely defined the binding sites, developed methods to measure rates and stoichiometries of binding, and studied the mutual influence of eIF4A and eIF3 on their binding to eIF4G. Our data provide evidence that binding of eIF3 and eIF4A to the central domain of eIF4G occurs in a cooperative manner.
eIF3, eIF4A, and eIF4F Preparation-Purification and 14 C labeling of eIF4A by reductive methylation was performed as described previously (43). eIF3, eIF4A, and eIF4F were purified from the ribosomal high salt wash of rabbit reticulocyte lysate by m 7 GTP-Sepharose and Mono Q chromatography (44). The eIF4A peak was rechromatographed on Mono Q with a shallower salt gradient. The eIF3 peak from the initial Mono Q chromatography was further purified by gel filtration on an SW300 column (Waters, Milford, MA) in buffer A (20 mM HEPES-KOH, 150 mM KCl, 2 mM ␤-mercaptoethanol, 0.1% (v/v) Tween 20, and 2 mM EDTA, pH 7.5) plus 5% (v/v) glycerol.
Prior to performing binding experiments with S-protein-agarose or m 7 GTP-Sepharose, purified eIF4G fragments, eIF4A and eIF4F, were passed over desalting Econo-Pac 10 DG disposable chromatography columns to replace the buffer with buffer A plus 5% (v/v) glycerol. Prior to SPR analysis, purified eIF4G fragments and eIF3 were passed over the same columns except they were equilibrated with buffer B (20 mM HEPES-KOH, 150 mM KCl, 2 mM EDTA, 0.05% (v/v) Tween 20, and 0.5 mM ␤-mercaptoethanol, pH 7.5). After buffer exchange, the concentrations of proteins were determined using the Bio-Rad protein assay kit, using bovine serum albumin as standard.  , RNAsin (10 units), 30 mM HEPES (pH 7.5), 30 mM potassium acetate, 0.6 mM dithiothreitol, and 120 M spermidine were preincubated for 10 min at 37°C. Reactions were then spotted onto Parafilm and irradiated on ice in the GS Gene Linker UV chamber (Bio-Rad) for 999 s. Reactions were transferred to Eppendorf tubes and incubated for 15 min at 37°C in the presence of RNase A (25 g/ml) and RNase V1 (0.4 units/l). Samples were analyzed on SDS-PAGE followed by autoradiography.
Protein Binding Assays on m 7 GTP-Sepharose-eIF4F was incubated with eIF3 in the presence or absence of the recombinant eIF4G fragments eIF4G-(642-1560), S-eIF4G-(877-1078), or S-eIF4G-(975-1078) for 40 min on ice. Reactions contained at least a 20-fold molar excess of recombinant fragment over intact eIF4G. Proteins were then mixed with m 7 GTP-Sepharose in the presence of 1% milk proteins in buffer A and incubated for 2 h at 4°C. Following washing four times with 300-l aliquots of buffer A, bound material was eluted from the resin with SDS-electrophoresis buffer and analyzed by SDS-PAGE (47), with detection by Coomassie Blue staining.
Protein Binding Assays on S-Protein-Agarose-Binding of S-eIF4G-(613-1078), S-eIF4G-(877-1078), S-eIF4G-(975-1078), and S-eIF4G-(1078 -1560) with eIF3 (and in some cases eIF4A) was performed using S-protein-agarose. After a 40-min preincubation of the S-peptide-tagged eIF4G fragments with eIF3 (and in some cases eIF4A) on ice, proteins were mixed with at least a 10-fold molar excess of S-protein-agarose and incubated for 2 h in buffer A containing 1% milk proteins at 4°C. The resin was washed four times with 200-l aliquots of buffer A, and the bound proteins were analyzed by SDS-PAGE as described above.
Quantitation of Binding Data-Quantitation of eIF4G fragments, eIF3 and eIF4A separated by SDS-PAGE was performed using a Scan-Maker III laser densitometer (Microtek) and ImageQuaNT software, version 3.3 (Molecular Dynamics). Experimental data were compared with standard curves, run on the same gel, of purified eIF4F, recombinant eIF4G fragments, eIF4A or eIF3 for which the concentrations had been determined with the Bio-Rad protein assay kit. Curve fitting was performed using SigmaPlot software version 4.01 (SPSS, Inc.). In cases of eIF3 binding with S-eIF4G-(613-1078) or S-eIF4G-(975-1078), the data were fit with an equation describing the Langmuir isotherm: where BR is the binding ratio, i.e. the molar ratio of bound eIF3 to the eIF4G fragment, n is the number of eIF3-binding sites on the eIF4G fragment, [eIF3] f is the concentration of eIF3 not bound to the resin, and K d is the dissociation equilibrium constant for the eIF3⅐eIF4G complex. A non-linear least squares fit was performed in which n and K d were allowed to vary. The number of binding sites was verified by replotting the data according to Scatchard (48).
In the case of eIF3 binding to eIF4G in the presence or absence of eIF4A, Eq. 1 was used assuming that n ϭ 1 for eIF3, even when saturation had not been achieved. The 1:1 stoichiometry for binding of S-eIF4G-(613-1078) to eIF3 in the presence of eIF4A was confirmed in experiments using higher concentrations of components, when saturation of all binding sites was achieved (data not shown).
In the case of eIF4A binding to S-eIF4G-(613-1078) or S-eIF4G-(1078 -1560) in the presence or absence of eIF3, Eq. 2 was used assuming that n ϭ 1 for eIF4A in each eIF4G fragment, even when saturation was not achieved: The 1:1 stoichiometry for eIF4A binding to each eIF4G fragment in the presence of eIF3 was similarly confirmed in experiments using saturating concentrations of factors (data not shown). SPR Analysis of eIF4G⅐eIF3 Interactions-SPR was carried out using BIAcore 2000 instrument (BIAcore, Inc., Piscataway, NJ). eIF3 was immobilized on a research grade CM5 sensor chip using the aminocoupling kit supplied by the manufacturer in 10 mM sodium acetate, pH 3.5. The surface density of immobilized eIF3 was 1500 -1800 RU. One RU corresponds to an immobilized protein density of 1 pg/mm 2 (49). The portion of the sensor chip in the first flow cell, used as a control, was subjected to activation and blocking in the same way as the eIF3containing cells but without added protein. The signals generated in the control flow cell were subtracted from the experimental signals to correct for refractive index changes and nonspecific binding.
All kinetic experiments were carried out in buffer B at 25°C and a flow rate of 20 l/min. At least six different concentrations of each eIF4G fragment were injected for each experiment. The first injection contained buffer without the eIF4G fragment. Between injections, the surface was regenerated with buffer C (20 mM HEPES-KOH, 500 mM KCl, 3 mM EDTA, 0.1% (v/v) Tween 20, 2 mM ␤-mercaptoethanol, pH 7.5) at a flow rate of 40 l/min and contact time of 3 min, followed by buffer B for 1 min.
Kinetic and equilibrium constants were calculated using the curvefitting facility of the BiaEvaluation software, version 3 (BIAcore, Inc.). Binding data were globally fit to the 1:1 Langmuir binding model (A ϩ B N AB) as described elsewhere (42). Values for the statistical closeness of fit, 2 , were always below 10, indicating that the simple 1:1 model of interaction correctly described the experimental data.

eIF4G Interactions with eIF3
Another approach to answer the same question was to incubate eIF3 with S-eIF4G-(613-1078) in the presence of micrococcal nuclease and then subject the mixture to S-proteinagarose, as in Fig. 1B. The results indicated that eIF3 was bound equally, regardless of the presence of micrococcal nuclease (data not shown), confirming that there is no RNA bridge.
S-eIF4G-(613-1078) and S-eIF4G-(975-1078) Bind eIF3 with a Stoichiometry of 1:1-To estimate the stoichiometry for the eIF4G⅐eIF3 interaction, we carried out titration experiments and then used S-protein-agarose to capture the complex. S-eIF4G-(613-1078) was incubated with either no eIF3 or a range of concentrations varying from a 0.5-to 10-fold molar ratio. The S-eIF4G-(613-1078)⅐eIF3 complex was adsorbed to S-protein-agarose, eluted, and analyzed by SDS-PAGE with Coomassie Blue staining (Fig. 4A). The binding data were quantitated using standard curves derived from eIF3 and the eIF4G fragment run on the same gel. A non-linear leastsquares fit of the experimental data was performed using Eq. 1 (Fig. 4B). The results indicated that the number of eIF3-binding sites on S-eIF4G-(613-1078), n, was 1.3 Ϯ 0.1. A linear transform of the data indicated the existence of a single-binding site (Fig. 4B, inset).
Kinetic Measurements of eIF3-eIF4G Fragment Interactions Using SPR-The S-protein-agarose method of analysis does not yield accurate K d values because it represents a non-equilibrium situation; the eIF3 concentration during the washes is much lower than during the initial incubation, giving an overestimate for the K d . We therefore turned to SPR, a more quantitative method, to measure both affinities and rate constants for binding of eIF4G fragments to eIF3. eIF3 was immobilized on the sensor chip by the aminocoupling procedure, and eIF4G-(642-1560), S-eIF4G-(613-1078), or S-eIF4G-(975-1078) fragments were passed over it (Fig. 5). As a control, a protein containing the same NH 2terminal tags (thioredoxin, His 6 , and S-peptide) as the S-eIF4G fragments but no eIF4G sequences was passed over the immobilized eIF3. The results indicated no binding of the control protein to eIF3 (data not shown).
The experimental data for binding of eIF4G-(642-1560), S-eIF4G-(613-1078), and S-eIF4G-(975-1078) to eIF3 were fit to the simple 1:1 binding model (thin lines in Fig. 5 (A-C); see "Experimental Procedures"). The low values for the statistical closeness of fit, 2 , and for the maximal residuals, which represent deviation of actual data from theoretical, indicate that FIG. 3. The RNA-binding site is located between aa 613 and 876 and is therefore distinct from the eIF3-binding site. Recombinant eIF4G fragments were incubated in the presence of 32 P-labeled RNA representing the EMCV IRES, cross-linked to RNA using UV light, digested with a mixture of RNases, and subjected to SDS-PAGE on 10% gel followed by staining with Coomassie Blue (A) and autoradiography (B). Unreacted proteins (lanes 1-4) and UV cross-linked proteins (lanes 5-8) correspond to eIF4G-(642-1560) (lanes 1 and 5), S-eIF4G-(613-1078) (lanes 4 and 8), S-eIF4G-(877-1078) (lanes 3 and 7), and S-eIF4G-(975-1078) (lanes 4 and 8).  (Table I). The rate constants for association (k a ), rate constants for dissociation (k d ), and equilibrium dissociation constants (K d ϭ k d /k a ) are also shown in Table I. The results show that all three eIF4G fragments had similar kinetic and equilibrium characteristics for binding to eIF3, suggesting that, within experimental error, the S-eIF4G-(975-1078) fragment contains the entire eIF3-binding site.
eIF4A Enhances the Binding of eIF3 to S-eIF4G-(613-1078)-To explore the possible influence of eIF4A on the binding of eIF3 to the central region of eIF4G, two types of titration experiments were carried out. In the first, S-eIF4G-(613-1078) was incubated with a fixed amount of eIF3 and varying amounts of eIF4A. The bound proteins were then analyzed as in Fig. 4 (Fig. 6A). The molar ratio of bound eIF3 to S-eIF4G-(613-1078) for each reaction was plotted against the concentration of eIF4A present in each reaction (Fig. 6B). The results indicated that, at higher concentrations of eIF4A, more eIF3 was retained on the resin through the interaction with S-eIF4G-(613-1078). eIF4A enhanced the binding of eIF3 to S-eIF4G-(613-1078) by more than 4-fold over the range of concentrations tested.
The effect of eIF4A on the affinity of eIF3 for eIF4G could be due to an eIF4A-induced change in the structure of either eIF4G or eIF3. To distinguish between these two possibilities, we used an eIF4G fragment that had an eIF3-binding site but no eIF4A-binding site, S-eIF4G-(975-1078). We incubated this fragment with eIF3 and [ 14 C]eIF4A and then subjected the mixture to S-protein-agarose. eIF3 was efficiently retained on the resin through S-eIF4G-(975-1078), but there was no detectable retention of [ 14 C]eIF4A, as determined by autoradiography (data not shown). This confirmed that eIF3 does not bind to eIF4A directly. It also rules out the possibility that eIF4A binds to eIF3 only after the latter is bound to eIF4G, because the eIF3⅐eIF4G complex retained by S-protein-agarose con-   Fig. 5, on a sensor chip with eIF3 immobilized by amino-coupling. Seven different concentrations of the indicated recombinant eIF4G fragments, ranging from 0 to 200 nM, were used in each experiment. The apparent dissociation constants (K d ) for the interactions were calculated from k d /k a . 2 2-6, respectively) and then fractionated on S-protein-agarose. As a control, 6 M eIF4A and 0.5 M eIF3 were incubated with S-protein-agarose in the absence of an eIF4G fragment (lane 7). Material bound to the resin was subjected to SDS-PAGE on a 10% gel followed by staining with Coomassie Blue. B, the amounts of S-eIF4G-(613-1078) and eIF3 in A were quantitated by scanning stained bands and comparing the signals to standard curves of the two purified proteins electrophoresed on the same gel (not shown). The binding ratio is the molar ratio of eIF3 to S-eIF4G-(613-1078), and [eIF4A] is the total eIF4A concentration present in the reaction. C, S-eIF4G-(613-1078) was incubated at a concentration of 0.05 M with varying amounts of eIF3 (0.04, 0.075, 0.15, 0.2, 0.5 M) in the presence (lanes 6 -10) or absence (lanes 1-5) of 0.1 M eIF4A and fractionated on S-protein-agarose. Material bound to the resin was subjected to SDS-PAGE on a 7% gel followed by staining with Coomassie Blue. D, the binding data in C in the presence (q) or absence of (OE) eIF4A to were fit to Eq. 1 by least squares minimization in which K d was allowed to vary and assuming that n ϭ 1 in both cases. The R 2 values were 0.91 (ϪeIF4A) and 0.96 (ϩeIF4A). tained no eIF4A. Thus, the contrary hypothesis is supported, that the increase in eIF3 binding to S-eIF4G-(613-1078) upon addition of eIF4A (Fig. 6B) is due to an eIF4A-induced change in eIF4G.
The second type of experiment was titration of S-eIF4G-(613-1078) with eIF3 in presence or absence of a fixed concentration of eIF4A (Fig. 6, C and D). The presence of 0.1 M eIF4A in the reaction mixture enhanced the amount of eIF3 bound to S-eIF4G-(613-1078) by 2-to 3-fold at low concentrations of eIF3, but the enhancement decreased as the eIF3-binding site became saturated.
Because there are two independent variables involved in studying the influence of one ligand on binding of another ligand (viz., eIF3 concentration and eIF4A concentration), it is necessary to fix one arbitrarily while varying the other. These two experiments do not explore eIF3⅐eIF4G binding over the full range of possible eIF4A and eIF3 concentrations. Nonetheless, both types of experiments show that eIF4A increases eIF3 binding to the central domain of eIF4G, probably through a conformational change in eIF4G structure.
eIF3 Enhances the Binding of eIF4A to S-eIF4G-(613-1078), but Not the Binding of eIF4A to S-eIF4G-(1078 -1560)-To study the reciprocal relationship, viz., the effect of eIF3 on binding of eIF4A to eIF4G, we performed titration experiments using two different eIF4G fragments. One represented the central domain, S-eIF4G-(613-1078), and therefore contained both eIF4A-and eIF3-binding sites. The other represented the COOH-terminal domain, S-eIF4G-(1078 -1560), and therefore contained only an eIF4A-binding site. Thus, if eIF3 affects the interaction between eIF4A and eIF4G by changing the conformation of the latter, one would expect to see a difference with S-eIF4G-(613-1078) but not S-eIF4G-(1078 -1560).
S-eIF4G-(613-1078) and S-eIF4G-(1078 -1560) were incubated separately with varying amounts of [ 14 C]eIF4A in the presence or absence of eIF3. Proteins bound to eIF4G were retained on S-protein-agarose. The amounts of the eIF4G fragment and eIF3 bound to the resin were analyzed by SDS-PAGE and Coomassie Blue staining (Fig. 7, A and D), whereas the amounts of [ 14 C]eIF4A bound were analyzed by autoradiography (Fig. 7, B and E). Quantitative analysis was performed by fitting the experimental data to Eq. 2, assuming that each eIF4G fragment has one eIF4A-binding site (Fig. 7, C and F). This assumption has been independently validated (42). The results indicated that the presence of 0.1 M eIF3 in the reaction mixture increased the binding of eIF4A to the central domain (decreased the K d ) by 2.4-fold (Fig. 7C) but did not alter the binding of eIF4A to the COOH-terminal domain (Fig. 7F).
Differences in eIF4A binding to eIF4G in the presence versus absence of eIF3 are more pronounced at low eIF4A concentrations, when the eIF4A-binding site is not saturated. However, it was necessary to show that the stoichiometry of eIF4A to the central domain of eIF4G remained 1:1 in the presence of eIF3, which requires high concentrations of eIF4A. Therefore, we performed an experiment similar to that shown in Fig. 7C except that the maximum concentration of eIF4A was 4 M and the fixed concentration of S-eIF4G-(613-178) was 0.5 M. The results indicated that, within experimental error, the stoichiometry of eIF4A binding to eIF4G was still 1:1 (data not shown).
Finally, we tested natural rabbit reticulocyte eIF4F for cooperativity of eIF4A and eIF3 binding to rule out artifacts due to the use of recombinant proteins. eIF4F (0.02 M) was incubated with 0.05 M eIF4A in the presence or absence of 0.03 M eIF3. Bound proteins were captured on m 7 GTP-Sepharose. The results indicated that eIF3 increased the binding of eIF4A to eIF4F by 1.6-fold (data not shown). Thus, eIF3 stimulates the binding of eIF4A to full-length eIF4G present in the eIF4F complex.

DISCUSSION
The finding that the eIF3-binding site on eIF4G is located between aa 975 and 1078 (Fig. 1B), combined with a recent study showing that it is located between aa 672 and 1065 (41), indicates that it is actually between aa 975 and 1065, although this specific fragment has not been tested directly. Despite the fact that this is a rather small portion (104 aa) of the 1560-aa eIF4G molecule, the similarity of K d values for eIF4G-(642-1560), eIF4G-(613-1078), and eIF4G-(975-1078) ( Table I) suggests that all of the binding determinants for eIF3 are located within this region. This places the eIF3-binding site between the two eIF4A-binding sites with no overlap (Fig. 1A). Similarly, there is no overlap between the RNA-binding site and the eIF3-binding site, because S-eIF4G-(975-1078) does not bind RNA (Fig. 3B). The latter finding indicates that the interaction between eIF4G and eIF3 does not require RNA.
Contrasting with the model of eIF4G with widely separated, non-overlapping binding sites (Fig. 1A) is the finding that binding of some ligands to eIF4G positively influence the binding of others. In the current work, we showed that eIF4A enhances the binding of eIF3 to eIF4G (Fig. 6) and, reciprocally, that eIF3 enhances the binding of eIF4A to the central (but not COOH-terminal) domain of eIF4G (Fig. 7). In other studies, it was shown that binding of eIF4A to eIF4G fragments containing both the central and COOH-terminal eIF4A-binding sites was greater than the sum of binding to fragments containing the individual sites, suggesting positive cooperativity between the two eIF4A-binding sites (42). There may also be cooperative binding of eIF4B and RNA to eIF4G; mammalian eIF4B increases binding of eIF4G-(613-1088) to RNA (38). Also, wheat germ eIF(iso)4G in complex with eIF(iso)4E binds PABP better than eIF(iso)4G alone (30), suggesting that eIF4E and PABP bind eIF4G cooperatively. Titration experiments were not performed in these published studies involving binding of RNA, eIF4B, PABP, and eIF4E to eIF4G, so we can only speculate that there is a cooperative change in the respective K d values. However, titration experiments were performed in the case of eIF4A (Fig. 7C) and eIF3 (Fig. 6D), allowing us to draw this conclusion. This paints a picture of eIF4G as a dynamic motor facilitating the stepwise interaction of initiation factors, ribosomes, and mRNA rather than as a static scaffold for assembly of the initiation complex. The cooperativity likely involves a conformational change in eIF4G, because eIF4A and eIF3 do not appear to interact directly.
Several previous results appear at first to be incompatible with the existence of non-overlapping eIF4A-and eIF3-binding sites. eIF4G variants in which the central eIF4A-binding site is changed by site-directed mutagenesis also fail to bind eIF3 in cell extracts (31). eIF4G fragments with truncations extending into the central eIF4A-binding site but containing an intact eIF3-binding site (eIF4G-(702-1090), eIF4G-(733-1090), and eIF4G-(762-1090)) lose not only eIF4A binding but also eIF3 binding (41). Conversely, eIF4G fragments containing an intact central eIF4A-binding site but lacking the eIF3-binding site (eIF4G-(642-947) (Ref. 41) and eIF4G-(642-877) (Ref. 40)) fail to bind not only eIF3 but also eIF4A. The findings reported in the present study may resolve these apparent contradictions: positive cooperativity between eIF4A and eIF3 predicts that the absence of binding of one ligand would weaken binding of the other ligand.
Our studies indicate that the K d for binding of mammalian eIF3 to eIF4G is about 42 nM in the absence of eIF4A (Table I). The K d for the wheat germ eIF3ϪeIF(iso)4F interaction was found to be 460 nM in the absence of salt and 1.28 M in the presence of 100 mM KCl and 2 mM MgCl 2 (50), or about 10-to 32-fold weaker. Interestingly, the eIF4F and eIF(iso)4F complexes from wheat germ do not contain eIF4A (51)(52)(53), suggesting that the affinity of eIF4A for the central domain of wheat germ eIF4G is also weaker than for mammalian eIF4G. Furthermore, plant eIF4G does not contain the second eIF4Abinding site in the COOH terminus, which acts cooperatively to stimulate binding of eIF4A to the central site (42) and recruitment of mRNA to the ribosome (41). Both the lower affinity of eIF3 for eIF4G and the absence of a COOH-terminal eIF4Abinding site would lead to a weaker binding of eIF4A to the central domain, according to the finding of positive cooperativity involving both sites ( Fig. 7 and Ref. 42). The weaker affinity makes the loss of eIF4A from plant eIF4F more likely than from mammalian eIF4F.
It is also interesting to note that, although the COOH-terminal domain of eIF4G serves to stimulate eIF4A binding (42) and enhance function of the central domain in protein synthesis both in vivo (40) and in vitro (41), it does not influence eIF3 binding. As shown in Fig. 5, the 918-aa eIF4G fragment S-eIF4G-(642-1560), which contains both central and COOHterminal domains, binds eIF3 with the same affinity as the 103-aa fragment S-eIF4G-(975-1078), which contains only a small portion of the central domain. Thus, the COOH-terminal domain is somehow able to affect binding of the distal, 46-kDa eIF4A molecule without affecting binding of the proximal, 600-kDa eIF3 molecule.