Studies on the Role of Eukaryotic Nucleotide Exchange Factor in Polypeptide Chain Initiation*

Interactions of eukaryotic 5-dimethylaminonaph-thalene- 1-sulfonyl-initiation factor 2 (eIF-2) from rab- bit reticulocytes and the guanine nucleotide exchange factor (GEF), Met-tRNAf, GTP, and GDP were moni-tored by changes in fluorescence anisotropy and radio- active filtration assays. At 1 mM Mg2+, radioactive filtration assays demonstrate that GEF is necessary for nucleotide exchange. We did not observe a GDP dependence in the association reaction of eIF-2-GEF for GDP concentrations from 0.01 to 20 PM. This is in disagreement with the model: eIF-2.GDP + GEF + eIF-2-GEF + GDP. The addition of GTP caused a de- crease in fluorescence anisotropy which is interpreted as a dissociation of eIF-2-GEF. We propose an asym- metrical model of ternary complex (eIF-2-GTP.Met-tRNAf) formation where 1) GDP does not displace GEF and 2) GTP replaces GEF and presumably GDP. For reticulocyte eIF-2, phosphorylation of the CY subunit greatly inhibits protein synthesis. This inhibition de-rives neither from failure of GEF to bind to ~ I F - ~ ( c Y P ) nor from greatly enhanced

Several mechanisms have been proposed for the nucleotide exchange reaction and the effects of phosphorylation of eIF-2 (8,(12)(13)(14). In this paper, we present data from membrane filter assays showing the overall reactions and a mechanism based on direct measurement by fluorescence anisotropy of the interaction of eIF-2 and eIF-21aP) with GEF. These results are used to interpret the overall nucleotide exchange reactions and the effect of phosphorylation on the catalytic utilization of eIF-2 and GEF.
Dansylation of elF-2"Reticulocyte eIF-2 and eIF-B(aP) were labeled by the same procedures. A 100-fold molar excess of dansyl chloride in 5 MI of ethanol was added to 25 pl of eIF-2 or eIF-B(nP) containing 0.3-0.6 mg/ml of protein (1 p M = 0.127 mg/ml). The solution was incubated for 20 min a t 4 "C and chromatographed on a Sephadex G-25 column (0.4 X 15 cm) equilibrated with buffer A containing 0.05 mM EDTA and eluted with the same buffer. The first fluorescent hand was collected (0.1-0.2 ml) and applied to a DE52 column (0.5 X 2 cm) equilibrated with the same buffer. Upon chromatography on the Sephadex G-25 column, the A2W and activity peaks eluted earlier than the fluorescence intensity maximum. The second column (DE52) removed additional noncovalently bound dye so that the A, , , eIF-2 activity, and fluorescence intensity peaks corresponded and free fluorescent material was retained.
Activity Assays-Activity assays for labeled and unlabeled protein were carried out as previously described (15). The dansyl-labeled protein showed the same activity for binary and ternary complex formation as unlabeled protein.
Fluorescence Measurements-Fluorescence anisotropy was measured as previously described (19,20). Titrations were carried out in buffer A. Concentrations of eIF-2 and eIF-B(aP) ranged from 3 to 17 nM.
Data Fitting-Equilibrium constants for eIF-2 andeIF-Z(aP) binding to GEF were obtained by fitting fluorescence anisotropy titrations to equations described previously for prokaryotic ribosomal protein S1 binding to 70 S ribosomes (20).  (20).
The fluorescence intensity was unchanged when GEF was added to eIF-2 or eIF-2(nP). The total eIF-2 and GEF concentrations were known, and therefore the equilibrium constant in principle could be determined from any point on the anisotropy curve once TeIF-2 and reIp GEF were known. The value for relp.2 was easily measured before addition of GEF, and the value for relF~2 (;EF was determined from the end point of the anisotropy titration. The robs uersus [GEFIr data were fit with a Fletcher-Powell sum of squares minimization using only one variable (KeJ for the entire titration curve (20).

Role of Nucleotide Exchange
Factor in Protein Synthesis 7375 RESULTS Binary (eIF-2. GDP) and ternary (eIF-2.GTP.Met-tRNAf) complex formation with phosphorylated and nonphosphorylated eIF-2 is shown in Table  I. In the absence of Mg2+, reticulocyte eIF-Z(aP) formed binary and ternary complexes as efficiently as the nonphosphorylated form. However, the addition of Mg2+ reduced these reactions 70-86%. In the presence of a factor, GEF, isolated from either the postribosomal supernatant or the high salt wash of ribosomes, [3H]GDP binding or ["5S]Met-tRNAf binding occurred to the same extent in the presence or absence of 1 mM M$+. When the LY subunit of reticulocyte eIF-2 was phosphorylated, addition of GEF in the absence or presence of Mg2+ did not promote significant binding of [3H]GDP or [3sS]Met-tRNAf.
Labeling of eIF-2 with dansyl chloride permitted direct measurement of the interaction of eIF-2 and eIF-B(aP) with GEF. Dansyl-eIF-2 showed a large change in fluorescence anisotropy when GEF was added to the solution (Fig. 1). The apparent equilibrium constant (Table 11)

TABLE I1
The equilibrium constants for GEF and phosphorylated and nonphosphorylated eIF-2 from rabbit reticulocytes For A, [eIF-2] was 0.005-0.009 p~. No GDP was added for the titration; however, some GDP was usually present in eIF-2 preparations. These concentrations were low enough that less than 10% of the eIF-2 was complexed with GDP. For B, [eIF-21 was 0.007-0.015 p~; GDP was 0.1-0.2 p~. The GDP concentration was adjusted so that initially 85-95% of the eIF-2 was present as an eIF-2.GDP complex. For C, [eIF-21 was 0.008-0.02 p~; GDP was 20 p~ before addition of GEF. All titrations were performed in buffer A (see text).

Additions
Experimental conditions  about three to four times more tightly than the binary (eIF-2 . GDP) complex.
The effects of GTP on this system were as follows. For nonphosphorylated reticulocyte eIF-2, the addition of GEF caused a fluorescence anisotropy change from about 0.11 to 0.19-0.22. At low levels of GDP (stoichiometric, 0.005-0.015 p~ to about 0.5 p~) , the addition of GTP (10-15 p~) caused a decrease in anisotropy to about 0.08-0.09, presumably from dissociation of the eIF-2.GEF complex. The phosphorylated eIF-2 behaved quite differently. The GTP level was increased approximately 10-fold t o 100 p~ before dissociation took place. Fig. 2 shows the kinetics of ["HIGDP release from eIF-2 or eIF-2(cuP). GDP was released from eIF-2 but not from Role of Nucleotide Exchange Factor in Protein Synthesis eIF-2(aP) by the addition of 125 FM GTP and GEF. In the absence of GTP, the addition of a &fold excess of GEF over eIF-2. ['HIGDP did not cause the release of [3H]GDP within the 10-min assay.

DISCUSSION
For reticulocyte eIF-2, the addition of GEF was necessary to promote significant exchange of GTP for GDP in the presence of Mg".
A symmetrical model was proposed to represent the GEF catalytic cycle of ternary complex formation (8,14). The model can be represented as follows.

GEF][GDP])/([eIF-2.GDP][GEF]).
Here, we assume' there is negligible free eIF-2, since KD for  (Fig. 3 solid line). We did not observe a GDP dependence in the dissociation reaction of eIF-2.GEF for GDP concentrations from 0.01 to 20 p~. If binding of GEF to eIF-2 were a displacement reaction, high levels of GDP should reverse the binding and promote dissociation of the complex. The observed data are clearly not in accord with this model.
An alternative model of the GEF catalytic cycle of ternary complex formation is (13) as follows. eIF-2. GDP. GEF eIF-2. GTP GD eIF-2. GTP. GEF In this model, eIF-2. GEF forms a stable complex first with GDP and then with GTP during the exchange reaction. Our data are not in conflict with the first part of the model, formation of a stable eIF-2 I GDP. GEF. We cannot measure directly release of GDP and therefore cannot distinguish, from anisotropy measurements, between an eIF-2. GEF complex and eIF-2 -GDP . GEF. The proposal of a stable elF-2.GEF. GTP complex is not in accord with our data. The addition of GTP caused a decrease in anisotropy, which is most reasonably interpreted as a dissociation of eIF-2. GEF since the end point corresponded to the anisotropy of labeled eIF-2.GTP (see Fig. 4).

GT>C
We represent the reactions of eIF-2, GEF, and GDP by a single equilibrium given by K (above) since, according to the model being tested, the ternary complex (eIF-2.GDP.GEF) is not allowed and GDP and GEF are present in large enough amounts to ensure that we may neglect unassociated eIF-2. Calculation shows that for the datum in Fig. 3 at 0.01 /IM GDP, unassociated eIF-2 was 4 2 % of the total; for the other points, the percentage was <2%. In this model, GEF binding does not displace GDP. This conclusion is supported by the fact that the association equilibrium for eIF-2.GEF does not depend on GDP concentration in the range 0.010-20 PM, a >lO-fold excess of GDP over eIF-2 and GEF.
The binding of G T P releases GEF and presumably GDP, so that a stable eIF-2 .GTP.GEF is not formed. In terms of nucleotide binding, the model presented is asymmetrical: the first reaction, GEF associating with eIF-2 .GDP, does not require release of nucleotide, and the second reaction, GTP-GDP exchange, causes release of GEF. This model incorporates the first part of the model of Safer et al. (13) and the second part of the symmetrical model (8,14). Phosphorylation of the reticulocyte eIF-2 inhibits the second part of the reaction.
Two opposed explanations were provided for the inhibition of protein synthesis in rabbit reticulocyte lysates. 1) Phosphorylated eIF-2 binds GEF much more tightly than nonphosphorylated eIF-2 and sequesters it from the active pool (17). This explanation is supported by the observation that even low levels of phosphorylation (10-20%) will inhibit protein synthesis as much as 90-95%. 2 ) Phosphorylated eIF-2 fails to interact with GEF, and therefore nucleotide exchange does not take place, and consequently a ternary complex cannot be formed (9).
Our equilibrium data ( Table 11) show that the second interpretation (non-interaction) is not correct. We have measured directly the interaction of GEF and eIF-P(aP) and find tighter binding and a 10-fold larger KO than for nonphosphorylated eIF-2. In the presence of GDP, the difference is only 2-fold. With regard to the first explanation (sequestering), it appears that phosphorylated eIF-2 can indeed sequester GEF, but not because binding to GEF is intrinsically much tighter. Rather, our data show that GTP replaces GEF and probably GDP but this replacement reaction requires much higher levels of G T P for phosphorylated eIF-2. This can be written as an equilibrium analogous to the GDP replacement of GEF discussed ). The complex eIF-2.GEF.GDP(?) is written with a (?) because we cannot distinguish from these measurements how much GDP is bound to the eIF-2 .GEF complex. The anisotropy, however, is a measure of the GEF bound. Addition of GTP favors dissociation of the complex. For eIF-2(aP), higher levels of GTP are required for the same degree of dissociation, and in terms of this mechanism the equilibrium constant must be smaller. This observation is in accord with data (17) that show inhibition of protein synthesis can be overcome by increasing the GTP/GDP ratio.