Regulation of Initiation Factors during Translational Repression Caused by Serum Depletion COVALENT MODIFICATION*

One to 2 h after transfer of HeLa cells into fresh serum-containing medium, when translation rates are maximal, the initiation factor proteins were examined on immunoblots of two-dimensional gels. Eukaryotic initiation factor (eIF)-2a, eIF-28, and eIF-4A each formed a single immunoreactive spot; eIF-27 formed 2 spots; and eIF-4B formed a complex array of 12-20 spots. After 4 days of growth in unreplenished medium, when translation rates have dropped 4-6-fold, several alterations in the isoelectric forms were observed eIF-2a now occurred in 2 forms, eIF-28 was present in 3-4 forms, and the most acidic cluster of eIF-4B variants was decreased or absent while a new isoelectric variant appeared at the basic end of the array. No changes were observed for eIF-27 or eIF-4A. The 36-SO-kDa subunits of the multiprotein initiation factor eIF-3 also showed no changes when the aforementioned growth states were compared. Resolution of 32P-labeled lysates by isoelectric focusinglsodium dodecyl sulfate-polyacrylamide gel electrophoresis indicated that the eIF-2a modification and the loss of eIF-4B variants reflected changes in phosphorylation states. Stimulation of 4-day grown cells with fresh serum-containing medium caused a reversal of the initiation factor modifications back to the forms prevailing shortly after replating. This analysis indicates that covalent modifications appear concurrently with decreasing initiation rates and suggests that they may be causative.


Regulation of Initiation Factors during Translational Repression Caused by Serum Depletion
COVALENT MODIFICATION* (Received for publication, November 12, 1984) Roger Duncan and John W. B One to 2 h after transfer of HeLa cells into fresh serum-containing medium, when translation rates are maximal, the initiation factor proteins were examined on immunoblots of two-dimensional gels. Eukaryotic initiation factor (eIF)-2a, eIF-28, and eIF-4A each formed a single immunoreactive spot; eIF-27 formed 2 spots; and eIF-4B formed a complex array of 12-20 spots. After 4 days of growth in unreplenished medium, when translation rates have dropped 4-6-fold, several alterations in the isoelectric forms were observed eIF-2a now occurred in 2 forms, eIF-28 was present in 3-4 forms, and the most acidic cluster of eIF-4B variants was decreased or absent while a new isoelectric variant appeared at the basic end of the array. No changes were observed for eIF-27 or eIF-4A. The 36-SO-kDa subunits of the multiprotein initiation factor eIF-3 also showed no changes when the aforementioned growth states were compared. Resolution of 32P-labeled lysates by isoelectric focusinglsodium dodecyl sulfatepolyacrylamide gel electrophoresis indicated that the eIF-2a modification and the loss of eIF-4B variants reflected changes in phosphorylation states. Stimulation of 4-day grown cells with fresh serum-containing medium caused a reversal of the initiation factor modifications back to the forms prevailing shortly after replating. This analysis indicates that covalent modifications appear concurrently with decreasing initiation rates and suggests that they may be causative.
Modulations in the rate of protein synthesis occur during mitosis (l), nutrient, amino acid, and serum factor deprivation (2-8), and stress (e.g. heat shock) (9-ll), to cite but a few examples. The protein synthetic pathway is commonly differentiated into three phases: initiation, elongation, and termination. Modulations virtually always occur at the level of initiation (12,13), with a few possible exceptions (6,14,15). Molecules involved in the initiation of protein synthesis are thus natural candidates for the mediators of translational control, and we have focused on the initiation factors. These proteins transiently associate with mRNA, tRNA, and ribosomal subunits and promote or are required for formation of the initiation complex.
Several alterations in the initiation factors that inhibit translation have been described, the most notable and exten-* 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. sively studied being the phosphorylation of eIF'-2 on its a subunit (12). This phosphorylation apparently prevents catalytic recycling of eIF-2 by inhibiting the dissociation of the eIF-2.eIF-2B complex (16,17). Phosphorylation of eIF-2 is induced by events such as heme deprivation (18), heat shock (11,19), and viral infection of interferon-treated cells (20). A plausible hypothesis is that eIF-2 phosphorylation serves as a very general molecular switch mediating translational control in response to a wide variety of extracellular and intracellular cues. Another example is the degradation of the 220-kDa subunit of eIF-4F that occurs during poliovirus infection roughly coincident with the shutoff of translation of host cell mRNA (21). This initiation factor modification likely causes poliovirus-induced translation inhibition, although conclusive proof is lacking. Finally, initiation factors eIF-2P and eIF-4B occur in multiple isoelectric forms whose distribution is altered by heat shock (ll), but a regulatory function for these alterations remains to be elucidated.
We wish to determine how generally initiation factor modifications are associated with translational control. Our goal is to establish whether the regulatory strategies employed by the cell frequently involve initiation factor alterations and what the repertoire of alterations may be. Here we have examined HeLa cells during the transition from a serum-rich to a serum-depleted state as a model system for studying how serum growth factor limitation inhibits the rate of protein synthesis and cell division. As documented in the accompanying paper (22), serum-depleted cells exhibit a reduced rate of initiation of protein synthesis. We demonstrate there that this inhibition is not likely due to changes in the intracellular levels of 10 initiation factor proteins measured. .In this report we examine the status of the same initiation factor proteins by IEF/SDS-PAGE and nonequilibrium pH gradient gel electrophoresis/SDS-PAGE and immunoblotting. These techniques allow us to detect and quantitate the covalent modification of initiation factor proteins. We report a number of changes in the extent of modification of these factors which correlate with the inhibition of protein synthesis in serumdepleted cells.

EXPERIMENTAL PROCEDURES
All materials and procedures used in this report have been previously described (22)(23)(24)(25). Refer to the accompanying paper (22) for details of the serum factor depletion growth regime and lysate preparation. The techniques involved in the immunoblotting of the twodimensional gels are described in Refs. 22-24. Phosphate labeling of HeLa cells is described in Ref. 11 and the legend to Fig. 3.

Covalent Modification Changes during Serum Depktwn-
The serum depletion regime is described in the accompanying paper (22). Briefly, monolayer cells are replated in fresh 10% serum-containing medium and then allowed to grow for 4 days without further perturbation. Shortly after replating translational activity is maximal with 70-90% of the ribosomes present in the polysome fraction. After 4 days of growth the fraction of polysomal ribosomes drops to 20-40%. In a previous report (23) we analyzed the initiation factor proteins in exponentially growing cells (24 h after replating) by twodimensional electrophoresis and immunoblotting. We observed multiple immunoreactive spots of equal molecular weight but differing charge for many of the initiation factor proteins, including eIF-2q eIF-28, eIF-By, and eIF-4B, which we tentatively identified as variant isoelectric forms. In the exponentially growing cells the eIF-2a and eIF-28 variants comprise <5% of the protein mass, the eIF-2y variant comprises about 33%, and eIF-4B forms a complex array containing 12-20 spots.
T o determine if the distribution of these variant forms is correlated with the changes in activity of the translational machinery that occur during growth in unreplenished medium, crude cytoplasmic lysates were prepared from cells 1-2 h and 4 days after replating, resolved by two-dimensional gel electrophoresis, and initiation factor proteins were examined by immunoblotting using affinity-purified antibodies to eIF-2a, eIF-28, eIF-27, eIF-4A, and eIF-4B. Lysates prepared 1-2 h after replating in fresh serum-containing medium show very simple eIF protein patterns for most of the proteins analyzed. A single form is observed for eIF-Za, eIF-28, and eIF-4A2 (Fig. lA), and two forms are observed for eIF-2y (Fig.  1R). eIF-4R is exceptional, being observed as a complex array of 12-20 spots (Fig. IC). Lysates prepared from cells which have been grown for 4 days in unreplenished medium (serumdepleted cells) reveal a more complex pattern. A second form of eIF-2a is clearly detectable (Fig. 1D) which is likely the phosphorylated derivative (see below). Two additional eIF-28 forms are seen (Fig. ID), one migrating to the acidic side of the major spot and a second (usually) more minor spot migrating to the basic side. eIF-4A and eIF-2y (Fig. l, D and E ) appear to be unaffected by serum depletion.
Two alterations in the complex eIF-4B immunoblot pattern ( Fig. 1F) are noted: (i) the most basic form (see arrow) is more prevalent in depleted cells and it is detected either faintly or not at all in the recently replated cultures; (ii) there is a general shift of the whole array of spots towards the more basic forms in the serum-depleted cells which frequently results in the disappearance or decreased abundance of a cluster of variants at the acidic end of the complex pattern (marked by a bracket in Fig. IC). The complex nature of the eIF-4B immunoreactive pattern makes it difficult to evaluate the significance of these changes. All of these immunoreactive spots are related to the most basic spot, since antibodies affinity-purified from the single, most basic spot react with the entire spot array.3 A quantitative evaluation of the distribution of forms for e1F-2~1, eIF-28, and eIF-2y was achieved by liquid scintillation counting of immunoblot spots. In confirmation of the visual estimates, there were no detectable variant forms for eIF-2a Previous immunoblot analyses of eIF-4A (7) revealed 2-3 immunoreactive spots. Recent antibody preparations used during the past year reproducibly have produced only 1 eIF-4A spot. We presume that the variants observed formerly were contaminating antibodies in the affinity-purified preparations.
R. Duncan and .J. W. R. Hershey, manuscript in preparation. modifications resolved by two-dimensional gel electrophoresis. Cytoplasmic lysates prepared between 1-2 h after plating and after 4 days of growth in unreplenished medium were resolved hy two-dimensional IEF/SDS-PAGE or nonequilihrium pH gradient gel electrophoresis/SDS-PAGE @oneLs R and E only). electrophoretically transferred to nitrocellulose paper, and sequentially reacted with anti-initiation factor protein antihodies followed by '*"I-laheled second antibody. The reacting spots are labeled. Blot sectom were treated with antibodies to eIF-20, -213. and -4A ( p a r w b A and D); elF-2y (paneLs R and E ) ; and elF-4R (paneb C and J?. Panels A-C. 1-2 h after plating; panels D-F, 4 days after plating. The uptpri~pkq in p a n e l s C and F mark the positions of two non-elF-4R proteins and serve as reference location markers. These proteins are barely viRihle in the figure and are somewhat smaller (&ft asterisk) and larger (right asterisk) than the eIF-4R proteins. The arrow marks the most basic eIF-4R variant, and the bracket, the most acidic cluster.

TARLE I
Qmntitation of the distribution of imrnunoreactioe spots on twodimensional imrnunoblots Cells 1-2 h after replating in fresh serum-containing medium (serum-stimulated) and after 4 days of unreplenished growth (serumdepleted) were analyzed by immunoblotting as described in Fig. 1. The immunoreactive spots were excised. using the film as template, and counted in toluene/2,5-diphenyloxazole/1,4-his[2-(5-phen~l~xazolyl)]benzene scintillation fluid. ~~ "This spot is displaced approximately two charges from the major 28 spot. On some films a faint (4th) spot is detectable between the most basic spot and the major 2p form.
Gel sectors encompassing the entire acidic block (as shown by the brackets in Fig. 4) and the entire hasic block (all immunoreactive spots to the higher pH side) were excised and counted to provide a gross measure of how this group is shifted. and eIF-28 in the freshly replated cells (Table I). In the serum factor-depleted cells 10"20% of the eIF-2a occurs in the modified form, and 15-20% of the eIF-28 immunoreactive spots are distributed between the two newly detected forms ( Table I). The distribution of eIF-27 immunoreactive proteins was eqiuvalent in both samples.
eIF-3 is a multiprotein initiation factor containing 7-10 subunits. Immunoblots with antisera to total eIF-3 reveal a pattern very similar to that produced when purified eIF-3 is subjected to IEF/SDS-PAGE and Coomassie blue staining (see Ref. 23 and below). eIF-3 subunits greater than 100 kDa (with the exception of eIF-3~220) tend to streak; only the subunits between 20-50 kDa focus well. The 35-50-kDa region of two-dimensional gel immunoblots from serum-stimulated cells was probed with a mixture of affinity-purified antibodies to eIF-3~35,' -p36, -p40, -p44, and -p47. A relatively complex pattern of spots is revealed (Fig. 2 A ) . Four of these are coincident with previously identified eIF-3 proteins (arrows, Fig. 2 4 ) while the status of the others as eIF-3 proteins or protein variants is unclear (additional discussion of the composition of eIF-3 will be provided elsewhere')). Similar analyses of eIF-3 proteins in lysates from serum-depleted cells reveal an identical pattern (Fig. 2R), indicating that these eIF-3 proteins are not modified during this type of translational repression.
'"P Labeling Patterns-Cells were labeled with [n2P]phosphate for either 30 min or 24 h a t various points during the serum depletion regime to examine phosphorylation changes. Examples of 30-min "P labeling patterns observed in lysates from the recently replated and 4-day depleted cells resolved by IEF/SDS-PAGE are shown in Fig. 3. Two relevant changes ' eIF-3p35 is always purified as an abundant eIF-3 subunit from rabbits, but is recovered not at all or as a very minor component when eIF-3 is purified from HeLa cells. However, the protein itself is abundant in HeLa lysates and reacts strongly with the antiaemm or affinity-purified antibodies made against rabbit eIF-3 (23). can be detected in the serum factor-depleted cells. (i) A labeled spot co-migrates with the acidic form of eIF-'20 in serumdepleted cells, whereas the spot is less intense (often not detected) in serum-stimulated cells. The increased extent of phosphorylation of eIF-2a in serum-depleted cells is not so obvious or extensive as that observed in other instances (e.g. heat shock (ll)), in accord with the fact that the initiation of protein synthesis still proceeds at one-third the maximum rate. (ii) The relative intensity of the more basic (less phosphorylated) eIF-4B variants is enhanced. The identification of these spots was deduced by co-migrations with purified initiation factors (11). Fig. 3 shows result.. with the pulselabeled samples, but similar resulh were observed when the 24-h labeled samples were examined. Intermediate time samples produce patterns suggesting a progressive activation of the modifying enzyme systems (data not shown). Note that the phosphorylation state reflects a balance between phosphorylation and dephosphorylation, and present data cannot differentiate between activations and repressions of either specific enzyme activity.
Reversal of Initiation Factor Modifications on Addition of Fresh Serum-The fraction of active ribosomes in serumdepleted cells is rapidly increased from -30% to 70-905 by transfer of fresh 10% serum-containing medium onto the cells.  Fig. 1. Cells were replated and grown for 4 days without replenishing serum or medium. Cultures were examined prior to addition of fresh serum-containing medium (0 h) and 2 and 4 h thereafter. Autoradiograms of immunoblot sectors treated with affnity-purified antibodies against eIF-Za, eIF-28, and eIF-4B are shown as labeled. eIF-4A and eIF-27, which are not affected by serum factor depletion, were not examined.

Change in the distribution of eIF-2 immunoreactive spots following serum stimulation of serum-depleted cells
Cells were grown and analyzed by immunoblotting as described in the legend to Fig. 1. Factor spota were excised and counted as described in Table I Two hours after transfer, the immunoblot pattern of eIF proteins is altered from the characteristic serum-depleted state ( Fig. 4.0 h; and Fig. 1, panels D and F) back to a pattern representative of exponentially growing cells (Fig. 4, 2 h and  4 h). The modified forms of eIF-2a and eIF-28, which comprise about 17% of the factor protein in serum-depleted cells, are diminished to about 9% within 2 h (Table 11). and the 2(aP) spot drops to about 2% by 4 h. The most acidic (phosphorylated) forms of eIF-4B (bracket, Fig. 4) increase in abundance while the amount of the most basic form is greatly reduced. eIF-4B characteristically showed a tendency to return to its depleted cell pattern between 2 to 4 h, perhaps reflecting a rapid depletion of critical serum factors by the dense (overcrowded) cell layer. The reversal of modification state temporally parallels the serum factor-induced mobilization of ribosomes into polysomes, which reaches maximum recovery 1-2 h after serum stimulation (12).

DISCUSSION
We have utilized a serum depletion regime to examine whether changes in the covalent modification of initiation factor proteins are correlated with changes in the initiation rate. Cells plated a t a low density in serum-rich medium possess 70-90% of their ribosomes in polysomes, but this value progressively decreases with days of growth as serum factors become depleted (22.26, 27). Repression occurs at the level of initiation, as evidenced by the ability of low doses of elongation inhibitor to recruit most ribosomes into polysomes (22). Several covalent alterations of initiation factor proteins which occur during growth in unreplenished medium may contribute to or cause the repression of translation. Alterations of initiation factor proteins have been implicated as regulators of translation in many systems, including rabbit reticulocytes (12,13,16,17) and Ehrlich ascites cells (28), and viral infections with poliovirus (21), Semliki Forest virus (29), and vesicular stomatitus virus (30).
Three principal covalent alterations of the initiation factors are induced by growth in unreplenished medium. eIF-2n develops an acidic variant comprising up to 20% of the eIF-2a mass which corresponds to the phosphorylated form of eIF-2a. eIF-28 develops minor acidic and basic variants comprising -20% of ita mass. eIF-4B exhibits a complex pattern of change indicative of decreased phosphorylation. These changes occur progressively, being most extensive in the 4day cells, but they begin to be discernible in exponentially growing cells 24 h after transfer into fresh serum-containing medium (23).
The eIF-2a variant precisely co-migrates with eIF-20 phosphorylated in vitro (20). An identically migrating "P-labeled spot is induced by heat shock (11). Phosphorylation of eIF-2a has been observed in numerous cell types (31) including HeLa cells (11,32), and no other eIF-2a modifications have been reported or observed by us. eIF-28 is a constitutively phosphorylated protein (12) and is also an ADP-ribosylation substrate (12). We have not determined how many phosphates are present in the major immunoblot-resolved spot, although other data would suggest it contains 2 phosphate groups. Cells labeled with [J2P]phosphate for 30 min or 24 h do not exhibit any labeled spots co-migrating with eIF-28. Whether this reflects a low rate of phosphorylation and/or exchange or an as yet undetermined technical problem is uncertain. %Slabeled lysates show 1-2 eIF-28 spots, corresponding to the major and acidic minor forms (23). The minor basic form detected in serum-depleted cells may be nonphosphorylated eIF-28, and thus the eIF-28 protein kinase activity may be depressed in depleted cells as may occur with the eIF-4B protein kinase. Perhaps the same kinase or phosphatme regulates the phosphorylation state of both proteins. The e 1 F -h and eIF-28 variants are not likely to be cross-reacting, nonfactor proteins. They are not observed in serum-stimulated cells, and thus are not common proteins of exponentially growing cells. It is unlikely that they are novel proteins whose synthesis is induced by serum depletion, since their rapid disappearance following serum stimulation is far more consistent with covalent modification. eIF-4B is an unusual protein which exhibit!! multiple isoelectric forms (Ref. 23 and Fig. 1). All of the spots in the immunoblot pattern, with the exception of the most basic variant, label with (32P]phosphate and the -15 spota are reduced to 2 by treatment with alkaline phosphatase.3 Previous results indicated that rabbit reticulocyte eIF-4B was phosphorylated in oitro (33) and in oioo (34) a t multiple sites (23), and the present results show that the extent of phosphorylation of eIF-4B is positively correlated with protein synthetic activity.
The serum depletion-induced phosphorylation of eIF-2a may be sufficient to account for the decreased rate of protein synthesis. eIF-2a phosphorylation is correlated with inhibited protein synthesis (12), and recent data provides a mechanistic rationale for inhibition. eIF-2(aP) appears to bind stably to another initiation factor, eIF-2B, forming an inactive complex in the presence of GDP (16,17). eIF-2B is less abundant than eIF-2 (35), and thus even a low level of eIF-2a phosphorylation can tie up a large fraction of eIF-2B. The amount of active eIF-2B in serum-depleted HeLa cells will depend on the abundance ratio of eIF-2-eIF-2B, but this value is not known. In reticulocytes, 30% phosphorylated eIF-2a is sufficient to inactivate virtually all eIF-2B (12). We observe 15-20% eIF-Sa phosphorylation in serum-depleted cells, compared to less than 5% in serum-stimulated cells. A decreased concentration of active eIF-2B in serum-depleted cells could be responsible for the decreased rate of initiation, but until an evaluation of eIF-2B levels is made, this remains conjectural. An additional possibility is that direct modifications of eIF-2B alter the levels of its active form. It is known that at least one subunit of eIF-2B (p67) is phosphorylated (36). We plan to pursue this inquiry when antibodies to eIF-2B are prepared.
Other initiation factor modifications may contribute to translational control as well. It has been reported that phosphorylation of eIF-2/3 is required for activity (37), but this modification is generally considered to be irrelevant to translational regulation because it is constitutive (12). We lack any data to suggest whether the minor eIF-28 variants in serumdepleted cells affect activity. The situation with regard to eIF-4B is similar. The complex spot array undergoes some rearrangement during serum depletion, but most forms still occur in both stimulated and depleted cells. If, for example, the maximally phosphorylated forms are preferentially utilized and differ in translational activity, then the observed changes would be quite significant. To date, our examinations of eIF-4B activity in vitro have failed to detect any differences due to the extent of phosphorylation.8 An alternate possibility is that the eIF-4B phosphorylation state influences message selection (38), but we have also failed to detect any shift in the spectrum of protein synthesis in stimulated versus depleted cells by .6 The eIF-3 proteins in stimulated and depleted cells show no alterations in their pattern. Several of the subunits of eIF-3 which fail to focus well on IEF (p120, p67) are known to be phosphorylation substrates (33,34). Their modification may be regulated and may influence em-3 activity. In the context of protein modifications that may influence initiation, phosphorylation of ribosomal protein S6 should be noted. This phosphorylation appears to be tightly coupled with the activation of protein synthesis in quiescent cells and may confer a selective advantage in initiation complex formation to S6-phosphorylated 40 S ribosomal subunits (39,40).
It is of course possible that one of the initiation factor proteins not examined here (e.g. eIF-2B) undergoes a covalent modification which contributes to the inhibition of initiation of protein synthesis of depleted cells. However, we suspect that phosphorylation of eIF-2a will prove to be the mechanism by which protein synthesis is regulated in virtually all cell types and in diverse conditions exhibiting translational control. We have recently shown that during the inhibition of translation by heat shock, eIF-2a, eIF-2/3, and eIF-4B are ' R. Duncan  altered in the same manner as occurs during serum factor depression. Interferon likewise can induce, via a cascade, an inhibition of the translational machinery which is correlated with eIF-fa phosphorylation (2). A highly speculative possibility is that a molecular system analogous to the interferon pathway may function during normal, physiological modulations of translation via eIF-2a phosphorylation, perhaps involving interferon itself or related molecules. The function of the eIF-2/3 and eIF-4B modifications which appear to be coinduced with eIF-2a phosphorylation is an open question. A role in the preferential stimulation of the translation of a subset of mRNAs is an attractive possibility.