Interaction of Wheat Germ Protein Synthesis Initiation Factor eIF-(iso)4F and Its Subunits p28 and p86 with m 7 GTP and mRNA Analogues*

The binding of p28, p86, and native wheat germ eIF- (iso)4F with m 7 GTP and oligonucleotides was measured and compared. The purified subunits (p28, 28 kDa and p86, 86 kDa) of wheat germ protein synthesis initiation factor eIF-(iso)4F have been obtained from Escherichia coli expression of the cloned DNA (van Heerden, A., and Browning, K. S. (1994) J. Biol. Chem. 269, 17454–17457). The binding of the 5 (cid:42) -terminal cap analogue m 7 GTP to the small subunit (p28) of eIF-(iso)4F as a function of pH, temperature, and ionic strength is described. The mode of binding of p28 to cap analogues is very similar to the intact protein. Assuming that all tryptophan residues contribute to p28 and eIF-(iso)4F fluorescence, iodide quenching shows that all 9 tryptophan residues in p28 are solvent-accessible, while only 6 out of 16 tryptophan residues are solvent-accessible on the intact eIF-(iso)4F. The fluorescence stopped-flow studies of eIF-(iso)4F and p28 with cap show a concentration-independent conformational change. The rate of this conformational change was approximately 10-fold faster for the isolated p28 compared with the native eIF-(iso)4F. From these studies it appears that cap recognition resides in the p28 subunit. However, p86 enhances the interaction

Recognition of the m 7 -cap structure of mRNA by eucaryotic initiation factors and formation of 48 S initiation complex are important steps in initiation of protein synthesis in eucaryotic cells. In mammalian cells, formation of the 48 S initiation complex is catalyzed by initiation factors eIF 1 -4A, eIF-4B, eIF-4E, and eIF-4␥. The complex termed eIF-4F contains polypeptides eIF-4A, eIF-4E, and eIF-4␥ (Grifo et al., 1983;Edery et al., 1983), but a complex of only eIF-4E and eIF-4␥ has also been determined (Etchison and Milburn, 1987;Buckley and Ehrenfeld, 1987). In addition, eIF-4E can be isolated as a separate polypeptide (Rinker et al., 1992). Recent evidence suggests that eIF-4E and eIF-4␥ form a complex concurrent with the binding of mRNA to the 40 S ribosomal subunit (Joshi et al., 1994).
Such a stepwise addition suggests specific functional roles for the separated polypeptides. In contrast to the mammalian eIF-4F complex, the individual polypeptides of the eIF-4F complex from wheat germ cannot be separated except under denaturing conditions. Wheat germ eIF-(iso)4F is one of the two wheat germ initiation factors that has cap binding ability, the other is wheat germ eIF-4F. Wheat germ eIF-(iso)4F consists of two subunits (p28 and p86) in a 1:1 molar ratio (Lax et al., 1985), while wheat germ eIF-4F consists of a 26-kDa and a 220-kDa subunit in a 1:1 molar ratio. Some structural and functional similarity exists between these two factors. Both wheat germ eIF-4F and eIF-(iso)4F have RNA-dependent ATPase activity. Only one wheat germ factor is required for ATP hydrolysis and stimulation of protein synthesis in an eIF-4F or eIF-(iso)4F deficient translation system (Lax et al., 1985(Lax et al., , 1986b. Wheat germ eIF-(iso)4F can substitute for mammalian eIF-4F in an RNA-dependent ATPase activity and in cross-linking of mammalian eIF-4A to the cap of oxidized mRNA (Abramson et al., 1988).
We have obtained purified subunits of wheat germ eIF-(iso)4F from expression in E. coli and examined the binding and functional properties of the separated subunits. Separate subunits of eIF-(iso)4F were not functionally active as measured by the ability to stimulate polypeptide synthesis. However, when both subunits were added, activity was equal to native eIF-(iso) 4F (van Heerden et al., 1994). These results indicate that the two subunits are able to associate to form an active complex. In addition, both subunits are required for activity. In order to understand the RNA binding ability of the intact protein and the function of the individual subunits, we have studied the binding and physical properties of the separated subunits, reconstituted protein, and native eIF-(iso)4F.

MATERIALS AND METHODS
Buffer A, used for fluorescence measurements, consisted of 20 mM HEPES-KOH, 100 mM KCl, 1 mM dithiothreitol, and 1 mM MgCl 2 and was adjusted to the appropriate pH. Buffer B, used in isolation of the wheat germ factors, consisted of 20 mM HEPES-KOH, pH 7.6, 0.1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, and KCl as indicated. All chemicals were reagent grade or better. m 7 GTP and m 7 GpppG were purchased from Sigma (St. Louis, MO). m 7 GTP-Sepharose was obtained from Pharmacia Biotech Inc. mRNA analogues (capped and uncapped oligonucleotides) were transcribed from partially double-stranded deoxyoligonucleotide templates containing a T7 RNA polymerase primer according to the procedures of Milligan et al. (1987) and were purified according to the method of Draper et al. (1988); the oligonucleotides used for the binding study, I and II, are shown in Fig. 1. The oligonucleotides used for the helicase assay, III and IV, are shown below.
Wheat germ initiation factor eIF-(iso)4F was purified according to the protocol of Lax et al. (1986aLax et al. ( , 1986b with the following modifications. The 40% ammonium sulfate fraction was loaded onto a DE52 column, equilibrated with buffer B containing 40 mM KCl (B-40). The column was washed until the OD was less than 0.2. The buffer was then * This work was supported by National Science Foundation Grants DMB-9105353 (to K. S. B.) and GER-9023681 and MCB-9303661 (to D. J. G.), the American Heart Association (AHA-NYC Established Investigatorship), and a PSC-CUNY Faculty Award (to D. J. G.). 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 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed. Tel.: 212-772-5330; Fax: 212-772-5332; E-mail: goss@mvaxgr.hunter.cuny.edu. 1 The abbreviations used are: eIF, eucaryotic initiation factor; m 7 G, 7-methylguanosine. changed to B-80 (80 mM KCl), and the peak containing eIF-(iso)4F was collected and concentrated with 80% ammonium sulfate precipitation and centrifuged at 11,000 rpm in a Sorvall centrifuge; the precipitate, which contains eIF-(iso)4F, was resuspended in 10 ml of B-120 and dialyzed overnight against B-120. The sample was then applied to an m 7 GTP-Sepharose column, and the pure factor was eluted as described by Lax et al. (1986aLax et al. ( , 1986b; in order to maximize yields, the m 7 GTP solution used to elute the pure eIF-(iso)4F factor was freshly prepared. The pure p28 and p86 subunits were prepared as described previously (van Heerden et al., 1994).
Fluorescence measurements were carried out at 23°C, unless otherwise noted, and data were collected and analyzed as described previously in detail (Carberry et al., 1989. Background fluorescence emission was subtracted. The double-stranded RNA (T m ϭ 65°C) for the helicase assay of eIF-(iso)4F was prepared by annealing the partially complementary III and IV single-stranded RNA. Equal molar amounts of RNA III and IV were mixed and heated to 80°C for 10 min, slowly cooled to 50°C for 30 min, and was then cooled to 40°C and kept overnight. The doublestranded RNA is shown as follows.
Oligonucleotide III: 3Ј-GCGACGCGACGGGG-5Ј Oligonucleotide IV: The unwinding reaction for the helicase assay was performed by incubating partially double-stranded RNA with eucaryotic initiation factors (5 M eIF-4A, 2 M p28, p86, and eIF-(iso)4F were used) in 20 mM Tris-HCl buffer (pH 7.5) containing 2 mM ATP, 1 mM magnesium acetate. The reaction mixture was incubated at 37°C for 20 min and then transferred to an ice bath and 2 ml of 5 ϫ stopping solution containing 50% glycerol and 10% SDS was added. SDS was used to separate the protein from the RNA and to avoid the formation of a large molecular weight complex. The sample was loaded onto a 15% nondenaturing polyacrylamide gel. The double-and single-stranded RNA on the gel was silver-stained (Bassam et al., 1991), and the bands were quantitatively determined by a laser scanning densitometer.
Stopped-flow fluorescence experiments were performed on a Hi-Tech SF-51 stopped-flow spectrofluorometer equipped with a Berger-type mixing chamber and a 2 ϫ 2 ϫ 10-mm flow cell with a dead time ϳ1.5 ms. The excitation wavelength was 280 nm, and the slit width was 2 mm. Light emitted from the reaction mixture was monitored after passage through a cut-off emission filter (WG-320 provided by Hi-Tech). A series of stopped-flow experiments were carried out at 23°C in buffer A, pH 7.05, under different concentrations of cap analogue m 7 GpppG. After rapid mixing of the protein with the mRNA cap, the time course of the intrinsic fluorescence intensity was recorded.

RESULTS
Fluorescence of p28⅐Oligonucleotide Complex-The fluorescence emission spectra of the p28 subunit of eIF-(iso)4F⅐ oligonucleotide complex as a function of oligonucleotide concentration is shown in Fig. 2. The fluorescence emission spectra of p28⅐m 7 GTP complex was similar (data not shown). Upon complex formation, there was a decrease in the protein fluorescence intensity at 331 nm. Such protein fluorescence quenching has been attributed to thestacking interactions between an aromatic amino acid residue and the nucleic acid base (Ishida et al., 1983;Brun et al., 1975;Lawaczek and Wagner, 1974).
The K eq of p28⅐oligonucleotide and p28⅐m 7 GTP complex formation can be calculated from the relative fluorescence intensity changes in free and complexed p28 emission spectra by construction of an Eadie-Hofstee plot, as shown in the inset of Fig. 2. K eq was found to be (1.24 Ϯ 0.04) ϫ 10 6 M Ϫ1 for oligonucleotide II in Fig. 2.
pH Dependence-The pH dependence of m 7 GTP binding to the p28 subunit of eIF-(iso)4F is shown in Fig. 3. The pH optimum for m 7 GTP was found to be 7.05 for p28, compared with the pH optimum of 7.6 for eIF-(iso)4F⅐m 7 GpppG complex (Carberry et al., 1991).
Temperature Effects-A Van't Hoff plot of Ϫln(K eq ) versus the reciprocal of temperature (T Ϫ1 ) can be used to calculate the thermodynamic parameters of entropy (⌬S) and enthalpy (⌬H). Fig. 4 shows the Van't Hoff plot based on m 7 GTP binding to p28; the values of ⌬H and ⌬S were obtained from the intercept and slope, respectively. The values of 5.8 Ϯ 0.4 kcal/mol and 44.7 Ϯ 1.9 cal/(mol°C) were obtained for ⌬H and ⌬S, respectively. These results are similar to the eIF-4E⅐m 7 GTP interaction, where ⌬H was 6.25 Ϯ 0.25 kcal/mol, and ⌬S was 46.1 Ϯ 1.8 cal/(mol°C) (Carberry et al., 1989). Similar values were also obtained for the eIF-4F⅐m 7 GTP interaction, where ⌬H was 6.84 Ϯ 0.7 kcal/mol, and ⌬S was 47.4 Ϯ 5.0 cal/(mol°C) (Carberry et al., 1991). An interpretation has been given (Ross and Subramanian, 1981) that positive ⌬H and ⌬S values suggest either hydrophobic (Gill et al., 1967(Gill et al., , 1976 or ionic (Pimentel and McClellan, 1971) interactions. The ionic strength dependence of m 7 GTP binding with p28 was investigated in order to determine if the interaction was hydrophobic or ionic. Ionic Strength Dependence-Debye-Huckel theory predicts that for charge-charge interactions, a plot of log K eq versus the square root of the ionic strength will give a linear plot. The slope equals 1.02Z A Z B for ionic interactions where Z A and Z B are the charges of the reactants. The binding of m 7 GTP to p28 as a function of the concentration of KCl and KC 2 H 3 O 2 was measured. The treatment of these data according to Debye-Huckel theory is shown in the Fig. 5. For the interaction of a single positive and negative charge, Z A Z B ϭ Ϫ1. For m 7 GTP binding with p28 in KC 2 H 3 O 2 , Z A Z B ϭ 0.008, demonstrating that the binding of m 7 GTP to p28 has little dependence on ionic strength. There are no significant ionic interactions involved in the binding of m 7 GTP to p28. Using KCl to vary ionic strength, the value of Z A Z B was Ϫ0.47. This difference between KCl and KC 2 H 3 O 2 suggests a possible uptake of Cl Ϫ by the p28. The special effect of anions on protein-RNA binding was reported earlier with the wheat germ eIF-4B protein (Sha et al., 1994).
Iodide Quenching Effect-Iodide quenching to determine the solvent accessible tryptophan residues was described previously (Sha et al., 1994). The modified Stern-Volmer plot (Fig. 6) shows quantitative results for quenching of p28 and eIF-(iso)4F. The reciprocal of the y intercept represents the fraction of accessible tryptophan residues among all tryptophan in the protein which contribute to the fluorescence (Lakowicz, 1983). A Y intercept of 1.02 Ϯ 0.05 and 2.65 Ϯ 0.50 was obtained for p28 and eIF-(iso)4F, respectively. Reciprocal values gave the fraction of accessible tryptophan residues as 98.0% Ϯ 4.5% and 37.7% Ϯ 6.0% for p28 and eIF-(iso)4F. Thus, if all tryptophan residues contribute to the protein fluorescence, all 9 tryptophan residues in p28 are solvent accessible and only 6 tryptophan residues from a total of 16 in eIF-(iso)4F are solventaccessible, indicating some tryptophan residues on p28 are buried during binding with the large subunit, p86, to form eIF-(iso)4F.
Comparison of Oligonucleotide Binding to Wheat Germ eIF-(iso)4F and Its Subunits-The binding of p28, p86, and wheat germ eIF-(iso)4F with m 7 GTP and mRNA analogues was measured and compared in Table I. The association of p28 with p86 to form eIF-(iso)4F did not significantly increase the K eq for m 7 GTP and capped RNA binding, yet both subunits were required for activity.
Helicase Activity-The eIF-4A-dependent helicase activity of eIF-(iso)4F and its subunits was measured by unwinding the partially double-stranded RNA annealed from oligonucleotide III and IV. It was found that neither the p28 nor the p86 subunit alone had helicase activity, while the combination of p28 and p86 gave full helicase activity as compared with eIF-(iso)4F (Table II). These data demonstrated that p86, the large subunit of eIF-(iso)4F, although it did not significantly enhance the RNA binding, was necessary for mRNA secondary structure unwinding during the translation process.

FIG. 6. Modified Stern-Volmer plot of iodide quenching to p28 (f) and to eIF-(iso)4F (q).
Stopped-flow Fluorescence Kinetics-The stopped-flow data for the binding of m 7 GpppG cap to the eIF-(iso)4F and p28 were plotted as ⌬F versus time as shown in Fig. 7. Fitted curves correspond to the following single exponential equation (Olsen et al., 1993), where k obs is the observed first-order rate constant, and the ⌬F ϱ is the maximum fluorescence change.
Mechanisms for eIF-(iso)4F-cap Interaction-Stopped-flow experiments were conducted using high concentrations of m 7 GpppG and limiting concentrations of eIF-(iso)4F to ensure that the bimolecular combination of mRNA cap analogue with protein was pseudo first-order. The mechanisms considered involved a one-and a two-step binding process (Garland, 1978).
The one-step reaction is as follows, where k 1 and k Ϫ1 are forward and reverse rate constants, respectively; P and C refer to eIF-(iso)4F and m 7 GpppG, respectively. Under the pseudo first-order condition, the observed rate constant is predicted to be a linear function of substrate concentration, i.e. k obs ϭ k 1 [C] ϩ k Ϫ1 . The two-step reaction is as follows, which involves a fast association of protein, eIF-(iso)4F, and cap analogue m 7 GpppG followed by a slow change of conformation of the first association complex, (P⅐C)*, to the stable complex, P⅐C, giving rise to the fluorescence change.
The interaction of eIF-(iso)4F (0.5 M) with m 7 GpppG under different concentrations of cap analogue (2.5, 5, and 10 M) gave about the same k obs (10.7, 11.5, and 12.1 s Ϫ1 , respectively). The reaction is not a single-step pseudo first-order reaction (Mechanism i). Earlier CD experiments have shown that eIF-(iso)4F undergoes conformational changes upon binding to m 7 GpppG 2 in agreement with Mechanism ii. Mechanism ii can be written as follows (Olsen et al., 1993), where K 1 ϭ k Ϫ1 /k 1 . Equation 2 rearranges as follows.
If one assume k Ϫ2 Ͻ Ͻ k obs , then A plot of 1/k obs versus 1/[C] will give an intercept of 1/k 2 (Fig. 8).
This model gives a k 2 value of 12.2 Ϯ 0.5 s Ϫ1 . Mechanisms for p28-cap Interaction-Stopped-flow experiments of p28 binding with cap similar to those described above were also conducted. Earlier CD experiments have shown that p28 undergoes conformational changes upon binding to m 7 GpppG 2 in agreement with Mechanism ii. The same Equation 4 was used here for the plot of Mechanism ii, the plot of 1/k obs versus 1/[C], which gave a k 2 value of 123.5 Ϯ 8.6 s Ϫ1 for p28 binding with the m 7 GpppG cap analogue. While more complex models with additional conformational changes will also fit the data, at present there is no experimental data to suggest such a mechanism is required.

DISCUSSION
The general mechanism of interaction of the cap with the p28 subunit of eIF-(iso)4F is very similar to the wild type protein.
The pH optimum for cap binding to p28 was found to be 7.05;  compared with the pH optimum of 7.6 for eIF-(iso)4F⅐m 7 GpppG complex (Carberry et al., 1991). The pK a values of amino acids are known to be environmentally sensitive and this could account for the shift in binding profile between p28 and eIF-(iso)4F. The shift in pK a implies the binding site has become more positively charged in the separated subunit. The previous sequence comparison (Allen et al., 1992) of wheat p28 with the cap-binding protein eIF-4E from mammals and yeast had shown about 38% sequence homology. There are three highly conserved histidine residues (position 33, 91, 193 from NH 2terminal) and 9 tryptophan residues in p28 (positions from NH 2 -terminal: 39, 42, 55, 72, 101, 112, 126, 161, 178), one more tryptophan residue than in mammalian and yeast cap-binding proteins (Trp-178). Site-directed mutagenesis of yeast eIF-4E showed that tryptophan 1, 2, and 8 were required for cap binding activity (Altmann et al., 1988). Site-directed mutagenesis of human eIF-4E showed that tryptophan 5 and the glutamic acid residue three amino acids to the carboxyl-terminal side of tryptophan 5 were involved in cap recognition. None of these studies examined the tertiary structure of the mutant protein; however, Ueda et al. (1991) proposed that the base stacking of the tryptophan and hydrogen-bond pairing by the glutamic acid are responsible for binding the m 7 G cap of mRNA. The p28 of eIF-(iso)4F has a glutamic acid residue four amino acids to the carboxyl-terminal side of tryptophan 5 (Glu-105) which could participate in the binding proposed by Ueda et al. (1991). Their data, however, fail to account for the pH dependence of binding. The small subunit, p28, has cap binding properties very similar to native eIF-(iso)4F. This observation is similar to the relative affinities of mammalian eIF-4E and eIF-4F binding to globin mRNA (eIF-4E: K eq ϭ (20.9 Ϯ 1.0) ϫ 10 5 M Ϫ1 ; eIF-4F: K eq ϭ (18.6 Ϯ 1.1) ϫ 10 5 M Ϫ1 , Goss et al., 1990). In the mammalian system eIF-4E has about the same affinity for mRNA as the larger eIF-4F complex. These data suggest that the essential role of the large subunit is involvement in other interactions such as protein-protein interactions and helicase activity.
The p86 subunit, which also binds to m 7 GTP, although less tightly, is unlikely to be the specific cap binding subunit in eIF-(iso)4F for the following two reasons: p86 shows no significant difference between capped RNA, and uncapped RNA binding and p86 does not distinguish m 7 GTP from GTP. The fact that p86 has a relatively high binding affinity for RNA suggests the native protein may have an RNA site involving both subunits and that p86 stabilizes the contact with RNA during the course of protein synthesis. In addition, photoaffinity labeling with m 7 GTP analogues label only the small subunit. 3 The stopped-flow kinetic data showed that the small subunit, p28, changed conformation approximately 10 times faster than eIF-(iso)4F (p28: k 2 ϭ 123.5 Ϯ 8.6 s Ϫ1 ; eIF-(iso)4F: k 2 ϭ 12.2 Ϯ 0.5 s Ϫ1 ). The faster conformational change rate of the p28-cap may be partially caused by an agile movement of its small mass (p28, 28 kDa compared with eIF-(iso)4F, 110 kDa). These results are surprising, since one would expect p86 to increase the rate of complex formation and hence increase the rate of interaction. However, if the initial equilibrium between the cap and protein is approximately the same for p28 and eIF-(iso)4F, then the off rate for eIF-(iso)4F would also be much slower than the off rate for p28. This would allow for interaction of other initiation factors or ribosomes and favor formation of an active initiation complex.
A working model is that p28 contains the binding site for the cap and locates the complex at or near the 5Ј end of the mRNA. While p86 does not affect the equilibrium affinity for cap, it does affect the rate of complex formation. In order for the eIF-(iso)4F complex to perform helicase activity (which requires both p28 and p86), the complex must process down the RNA. This would in all probability require release of the cap while maintaining contact with the RNA. p86 and other factors may stabilize the interaction with mRNA and prevent the protein complex from dissociating from the RNA. While p28 binds capped RNA specifically, the intact protein (or sum of p28 ϩ p86) is necessary to stimulate protein synthesis (van Heerden et al., 1994). This suggests p86 is involved in proteinprotein interactions to form a complete initiation complex.