Elongation factor-1 alpha mRNA is selectively translated following mitogenic stimulation.

Stimulation of quiescent Swiss 3T3 cells to proliferate leads to a selective 6-fold increase in the rate of translation of protein synthesis elongation factor eEF-1 alpha. Northern blot and solution hybridization protection studies show that levels of eEF-1 alpha mRNA remain constant following serum stimulation, demonstrating that eEF-1 alpha transcripts are not degraded following mitogenic stimulation and that the increase in eEF-1 alpha synthesis is accounted for by pre-existing mRNA. Localization of these transcripts in resting cells shows that they are largely distributed equally between stored mRNA protein particles and monosomes/disomes. Following serum addition, eEF-1 alpha transcripts present in mRNA protein particles redistribute to large polysomes rather than to monosomes and disomes as would be expected. The same is true for those transcripts present in monosomes and disomes. Salt-shift and translational runoff studies indicate that eEF-1 alpha transcripts sedimenting with monosomes and disomes in quiescent cells are associated with actively translating ribosomes. The results suggest that a specific transcript can move within polysome profile as a function of the affinity of translational apparatus for that transcript.

Stimulation of quiescent Swiss 3T3 cells to proliferate leads to a selective &fold increase in the rate of translation of protein synthesis elongation factor eEF-la. Northern blot and solution hybridization protection studies show that levels of eEF-la mRNA remain constant following serum stimulation, demonstrating that eEF-la transcripts are not degraded following mitogenic stimulation and that the increase in eEF-la s p thesis is accounted for by pre-existing mRNA. Localization of these transcripts in resting cells shows that they are largely distributed equally between stored mRNA protein particles and monosomeddisomes. Following serum addition, eEF-la transcripts present in mRNA protein particles redistribute to large polysomes rather than to monosomes and disomes as would be expected. The same is true for those transcripts present in monosomes and disomes. Salt-shift and translational runoff studies indicate that eEF-la transcripts sedimenting with monosomes and disomes in quiescent cells are associated with actively translating ribosomes. The results suggest that a specific transcript can move within polysome profile as a function of the affinity of translational apparatus for that transcript.

Regulation of gene expression at the translational level plays
a key role in the induction of quiescent cells to re-enter the cell cycle and proliferate (1,2). In quiescent cells the amount of translational machinery, as judged by ribosomal RNA content (31, has been markedly reduced, with only 2630% of that remaining engaged in protein synthesis (4). Mitogenic stimulation leads to an up-regulation of the translational machinery and a corresponding 3-4-fold increase in the rate of protein synthesis (2). This increase is not only a prerequisite for reentry of quiescent cells into the cell cycle, but the resulting high rates of translation must also be maintained throughout GI for entry into S phase (5). This acute translational response is initiated by the available translational machinery as well as a large pool of stored mRNA, and can take place in the absence of new transcription (4,6,7). Indeed, during the first 6-8 h of the mitogenic response 80% of the actively translated mRNA is derived from the stored mRNApopulation (4). These changes in mRNA usage lead to a number of alterations in the expression of specific proteins (8,9), several of which have been identified as essential gene products.
The largest change in the pattern of translation is the rapid 6-fold increase in the rate of synthesis of a M , 49,000 protein (8) payment of page charges. This article must therefore be hereby marked * The costs of publication of this article were defPayed in part by the "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. later identified as protein synthesis elongation factor eEF-la (9). The role of eEF-la in protein synthesis is to direct the binding of aminoacyl-tRNA to the ribosome acceptor site (2,10,11). Recent studies suggest, however, that eEF-la may have other roles in the cell besides those associated with protein synthesis (12)(13)(14), which may explain the seemingly higher levels of this protein in cells as compared with other translational factors. Even though eEF-la is an abundant cellular protein, its overexpression has recently been shown to render mammalian cells more susceptible to transformation, arguing that its normal levels of expression must still be tightly regulated (14). Consistent with this finding, loss of expression is paralleled by decreased rates of protein synthesis and the onset of senescence in human fibroblasts (15). Acausal relationship is supported by the finding that transfection of a n additional copy of the eEF-la gene into Drosophila melanogaster can lead to a 41% increase in their life span (16).
Although eEF-la transcripts are rapidly recruited into polysomes following mitogenic stimulation, it has been reported that more than half of the total transcripts are simultaneously degraded (17) and speculated that this may be due to a mitogen-activated eEF-la RNase (18). This last finding is in apparent conflict with earlier studies from our laboratory employing in vitro translation as a n indirect measure of transcript levels (9). Since the level of eEF-la protein appears to play an essential role in cell growth and senescence, it is of critical importance to delineate the regulatory mechanisms which control its level of expression.
Here we have compared the amount of eEF-la transcripts with those of p-actin by Northern blot analysis and Solution hybridization following mitogenic stimulation. Next we have determined the distribution of both transcripts on polysomes derived from quiescent and mitogen-stimulated cells. The results demonstrate that eEF-la transcripts do not decrease following mitogenic stimulation and further suggest that the translation of these transcripts may be monitored by a novel regulatory mechanism.
EXPERIMENTAL PROCEDURES Cell Culture, Radioactive Labeling, and %o-dimensional Gel Analysis-Swiss mouse 3T3 cells were seeded and maintained on 15-cm tissue culture plates as described previously (19). Total [3SSlmethionine-labeled cell proteins were extracted and analyzed by two-dimensional nonequilibrium polyacrylamide gel electrophoresis (8).
Isolation of RNA-Total RNA was extracted from cells using 6 M guanidinium thiocyanate (20). For isolation of RNA from polysomes, cells were harvested with trypsin as described previously (21), except that cycloheximide (100 pg/ml) was added for 15 min prior to trypsinization, and cells were first washed with a 0.05% (w/v) trypsin and 0.02% EDTA solution before detaching with a 0.2% (w/v) trypsin and 0.02% EDTA solution. After centrifugation, cells were washed with hypotonic buffer containing 1.5 m~ KCl, 2.5 m~ MgC12, and 5 m~ Tris-HCl, pH 7.4, and 100 pg/ml cycloheximide, collected by centrifugation, resuspended in 200 pl of the same buffer, and lysed by the addition of 200 pl of hypotonic lysis buffer containing 2% sodium deoxycholate 4367 (Merck), 2% Triton X-100 (Serva), 2.5 m~ DIT,' 100 pg/ml cycloheximide, and 100 units of RNasin (Promega). To ensure that lysis was complete, the extract was homogenized with 10 strokes of a Dounce homogenizer. The extract was then centrifuged a t 2300 x g for 10 min a t 0 "C (Sorvall RC2-B centrifuge, SS-34 rotor) to remove nuclei, and the resulting supernatant was made 1 mg/ml in heparin. The extract was quick-frozen in liquid Nz and stored a t -70 "C. In general, 500 pl of the cytoplasmic extract were applied to a 12.7-ml linear sucrose gradient from either 17.141% or 5.1-30.2% (w/v). Buffer conditions are indicated in the figure legends. Centrifugation was carried out in an SW41 Beckman rotor a t 36,000 rpm for 110 min a t 2 "C for 17.1-51% gradients and a t 40,000 rpm for 120 min a t 2 "C for 5.1-30.2% gradients. Gradient tubes were punctured from the bottom with a hollow needle and the fractions monitored on a Kontron Uvikon 725 spectrophotometer through an LCD 725 flow cell at 260 nm. Each 1-ml fraction was treated with 100 pg/ml proteinase K for 2 h in 100 m~ Tris-HCI, pH 7.4, 50 m~ EDTA, and 5% SDS before phenoVchloroform extraction and precipitation of the RNA by the addition of 0.10 volume of 3 M sodium acetate, 200 pg of calf thymus tRNA (Boehringer Mannheim) as carrier and 2.5 volumes of ethanol. After three rounds of freeze-thawing, the RNA precipitate was pelleted, washed with 70% ethanol, and lyophilized.
Rudiwctive Labeling of Oligonucleotide Probes-The oligonucleotide eEF-la probe is complementary to the first 57 coding bases of eEF-la mRNA. The oligonucleotide p-actin probe is complementary to the first 57 bases of 3'-untranslated region of mouse p-actin mRNA. Oligonucleotide probes were 5' end-labeled in a reaction mix containing 13 pmol of oligomer T4-buffer, 50 m~ Tris-HCI, pH 7.6,lO m~ MgCIz, 5 m~ DIT, 0.1 m~ spermidine, and 1 m~ EDTA, pH 8.0,75 pmol [+"P]ATP, and 10 units of T4-kinase (Boehringer Mannheim) and was incubated for 40 min a t 37 "C. The labeling efficiency of the probes was 4540%. The labeled probe was purified either by polyacrylamide gel electrophoresis for S1 nuclease-protected hybridization or on a G-25 Sephadex spun column for Northern blots (22).
Northern Blot and Solution Hybridization-For Northern blot analysis RNA samples were separated on 1% formaldehyde-agarose gels and transferred to either a nitrocellulose or nylon membrane (Amersham N-Hybond), as described previously (22). The membrane was hybridized with the labeled oligonucleotide for 1618 h a t 56 "C in a solution containing 1 M NaCI, 10% dextran sulfate, 1% SDS, and 10 mg/ml denatured salmon sperm DNA. The membrane was then washed with 2 x SSPE (3 M NaCI, 0.2 M sodium phosphate, 0.02 M EDTA, pH 7.7) + 1% SDS a t room temperature for 2 x 30 min, then with 2 x SSPE + 1% SDS a t 55 "C 2 x 30 min, and finally with 0.2 x SSPE + 0.1% SDS a t room temperature. The blot was then analyzed by autoradiography.
For solution hybridization assays, RNA pellets from the gradient fractions were dissolved in 15 pl of 20% formamide and hybridized for 1618 h a t 56 "C in 600 m~ NaCI, 20 m~ Tris-HCI, pH 7.2,4 m~ EDTA, 0.1% SDS, 10% formamide, and specific labeled oligomer in a final volume of 30 pl overlaid with parafin oil. S1 nuclease (50 units, Boehringer Mannheim) and 1 ml of S1 nuclease buffer (30 m~ sodium acetate, pH 4.2,300 m~ Nacl, 3 m~ zinc acetate, and 75 pg/ml calf thymus DNA) were added, and the reaction mix was incubated at 37 "C for 1 h. The undigested mRNAoligomer heteroduplex was precipitated by adding 100 pl of trichloroacetic acid (100%) on ice for 15 min. The precipitate was collected on Whatman GF/C filters, washed five times with 5% trichloroacetic acid, 5 m~ sodium pyrophosphate, two times with 100% ethanol, and the filters were dried and counted for radioactivity. The S1 nuclease was added in suffcient amounts to digest all unhybridized labeled oligomer to a background value of -300 cpm and the input countdmin was a t least five times the peak values of labeled oligomer protected from S1 nuclease digestion.
In Vivo Polysome Runof-Quiescent Swiss mouse 3T3 cells were incubated with 30 m~ sodium fluoride for 20 min a t 37 "C. Cycloheximide (100 pg/ml) was then added to the medium and the cells incubated a further 10 min a t 37 "C before making a cytoplasmic extract as described above.

RESULTS
Expression of eEF-I a and p-Actin-A comparison of [35S]methionine pulse-labeled cytoplasmic proteins from quiescent and 60-min serum-stimulated Swiss 3T3 cells by two-dimensional nonequilibrium polyacrylamide gel electrophoresis reveals several notable changes in the pattern of translation (Fig. 1). One The abbreviations used are: DIT, dithiothreitol; mRNP, messenger RNA protein. were labeled with [35Slmethionine as described previously (8). QZ3 is the product of a previously described translationally controlled mRNA (see "Discussion"). of the largest effects (8) was the increase in the rate of synthesis of eEF-la (Fig. 1). To determine the fate of eEF-la transcripts following serum stimulation, mRNA levels were measured in cell extracts either by Northern blot analysis (Fig. 2) or by solution hybridization (Fig. 3)  mRNA Turnover-Although unlikely, the results above did not exclude the possibility that the increase in expression of eEF-la protein is due to enhanced transcription of the gene, compensated for by a corresponding loss of stored mRNA. To test this possibility, cells were stimulated in the presence of actinomycin D, a potent inhibitor of transcription, and the fates of eEF-la and 6-actin transcripts were measured following serum stimulation. Under these conditions the level of eEF-la mRNA remains constant during the first 180-min post-serum induction (Fig. 4A), further indicating that serum stimulation does not induce degradation of this message. The same is true for the p-actin mRNA (Fig. 4B), whose increased expression is dependent upon new transcription. It should be noted that similar results were obtained if transcription was blocked with 5,6-dichloro-l~d-ribofuranosylbenzimidazole (28), which elicits its inhibitory effect by a distinct mechanism from that of actinomycin D. Thus, the half-life of the eEF-la mF?.NA in serum-stimulated cells appears to be much longer than the time course of induction, and since general inhibition of transcription has no effect on the increased expression of eEF-la protein (8), these findings demonstrate that the enhanced expression of this protein is due to pre-existing mRNA.
Polysome Distribution of eEF-la mRNA-The simplest ex- planation for the increase in the amount of eEF-la protein following serum stimulation is that the message is recruited from stored messenger RNA protein (mRNP) particles onto polysomes, as described for other proteins (4, 291, allowing messages which are recognized with low efficiency to be expressed at higher levels (30). To test this model, extracts from quiescent and 180-min serum-stimulated cells were fractionated on sucrose gradients and either eEF-la or p-actin mRNA was localized by solution hybridization. In quiescent cells the mean number of ribosomes bound per actin transcript is approximately seven to eight (Fig. FA). Following serum stimulation there is a large increase in the amount of p-actin mRNA and a shift of inactive 80 S ribosomes into polysomes; however, the mean number of ribosomes bound to each p-actin transcript remains constant (Fig. 5B), as has been shown for other m W A s (29, 31, 32). The mean polysome size also remains constant during this time, with seven to eight ribosomes bound per transcript. In the case of eEF-la, most of the mRNA in quiescent cells is apparently localized in stored mRNP particles at the top of the gradient (Fig. 5A). Following serum stimulation, most of the eEF-la mRNA is found to shift to polysomes containing roughly 11-12 ribosomes per transcript, a size significantly larger than those associated with p-actin mRNA (Fig.  5B). Furthermore, at a time when eEF-la is being actively recruited into polysomes, a small fraction of p-actin mRNA is accumulating in stored mRNP particles (Fig. 5B), despite the fact that the coding sequence for p-actin is only slightly smaller than that of eEF-la (24, 26, 33) and that theoretically p-actin mRNA could accommodate more ribosomes per transcript (1,34). Thus in contrast to the expected result, these data indicate that the translational apparatus has a preference for eEF-la transcripts following mitogenic stimulation.
lEuo Distinct Populations of eEF-la-Closer inspection of the data presented in Fig. 5A also revealed that a portion of the eEF-la mRNA may not be sequestered in inactive mRNP particles but instead may be associated with small polysomes. To examine this possibility, extracts from quiescent cells were fractionated under conditions where 80 S ribosomes can be resolved from 60 and 40 S ribosomal subunits (see Fig. 6.4). Employing this method, eEF-la transcripts could be localized by solution hybridization into two distinct populations, one sedimenting with monosomeddisomes and the other sedimenting slightly slower than 40 S ribosomes (Fig. 6A), probably representing stored mRNP particles (35, 36). Results from a Northern blot analysis of a second gradient were almost identical to those in Fig. 6A with the exception that the first eEF-la mRNA peak was shifted by one fraction (Fig. 6 B ). In addition, it should be noted that those transcripts present in stored mRNp particles move to large polysomes rather than to monosomeddisomes following mitogenic stimulation (Fig. 5B 1,  Danslational Control of eEF-1 (fractions 1-14). The fractions were analyzed for eEF-la transcripts (A) or p-actin transcripts by solution hybridization (0) (see Fig. 3 and "Experimental Procedures"). a fact which is inconsistent with the model that a given mRNA shuttles between stored mRNP particles and polysomes of a fixed size (for review see Ref. 1).
Monosomes Versus 80 S Ribosomes-Quiescent cells contain a large pool of 80 S ribosomes with no associated mRNA. These 80 S ribosomes co-sediment with monosomes (4,37,38). In high-salt sucrose gradients, 80 S ribosomes dissociate, whereas monosomes, with their associated mRNA, remain intact (39). 'Ib establish that the faster sedimenting peak of eEF-la mRNA was associated with monosome-disome complexes and not nonspecifically associated with 80 S ribosomes, extracts from quiescent cells were fractionated on high-and low-salt sucrose gradients (see "Experimental Procedures" and Fig. 7). Under either condition, eEF-la mRNAremained in two distinct populations (Fig. 7), despite the fact that under high-salt conditions, the inactive 80 S ribosomes dissociated and sedimented as 40 and 60 S subunits (Fig. 7B). These findings support the argument that a portion of eEF-la mRNA is associated with monosomes and disomes in quiescent cells.
In Vivo Polysome Runoff-'Ib determine whether the eEF-la transcripts associated with monosomeddisomes in quiescent cells were functionally active, the effect of NaF, a potent inhibitor of the initiation of protein synthesis, was examined. Since NaF does not impede elongation, treatment of cells with this salt leads to a net runoff of polysomes (40)(41)(42). Therefore if eEF-la transcripts are translated in quiescent cells, NaF treatment should lead to a shift of those transcripts associated with monosomeddisomes into inactive mRNP particles. The results of such an experiment show that NaF treatment leads to a significant runoff of polysomes and a corresponding shift of eEF-la transcripts from monosomeddisomes to mRNP particles (Fig. 8, compare A and B ). Similar results were obtained with puromycin (data not shown). Thus, eEF-la transcripts are mobilized from smaller to larger size polysomes as a function of the growth state of the cell.

DISCUSSION
Employing dot blot analysis, it has been shown that in stationary-phase Friend erythroleukemia cells, most of the eEF-la transcripts sediment as a broad peak between 50 SI60 S and polysomes of small size (17). By using increasing sedimentation times, it was argued that approximately 60% of these transcripts were sedimenting with inactive mRNP particles with the remaining eEF-la transcripts largely co-sedimenting in the position of small polysomes. However, unlike the data reported here, two distinct populations of eEF-la transcripts were not detected, even when 40 S subunits, 80 S ribosomes, and polysomes could be clearly resolved from one another. The authors also reported that following stimulation of stationary cells to proliferate there was a 2.6-fold increase in the synthesis of eEF-la protein, with all the eEF-la transcripts now sedimenting with large polysomes. But in seeming contradiction, the up-regulation of eEF-la protein expression was paralleled by a 70% loss in the total amount of eEF-la transcripts. The loss of eEF-la was suggested to have taken place selectively on mRNP particles (17,18) and was attributed to degradation by a specific ribonuclease (43). In contrast, under similar growth conditions in Swiss 3T3 cells, we find no measurable change in total eEF-la mRNAlevels (Figs. 2 and 3). The simplest explanation for this difference is that distinct translational regulatory mechanisms operate in the two cell types. However, in initial studies following a protocol similar to that described by Rao and Slobin (17), we also found that eEF-la transcripts derived from stimulated Swiss 3T3 cells were much more susceptible to degradation than those from resting cells.
In subsequent studies it was found that direct extraction of total mRNA by guanidinium isothiocynate (20) or treatment of sucrose gradient fractions with proteinase K prior to concentration by ethanol precipitation (44) circumvents this problem.
A question that arises from these studies is why eEF-la expression is under translational control, whereas p-actin mRNA is transcriptionally regulated. It is clear that eEF-la, along with other translational components, plays an essential role in the obligatory, rapid 3-to 4-fold increase in the rate of protein synthesis required during the early mitogenic response. It may be that transcripts for proteins whose functions are immediately needed reside in mRNP particles or on small polysomes where they can be more rapidly recruited following mitogenic stimulation. This model is consistent with earlier findings demonstrating that within 30 min of stimulation of quiescent Swiss 3T3 cells eEF-la transcripts are translated at near maximum rate, whereas a similar response for p-actin requires at least 60 min (8). Cytoplasmic mRNAs are generally distributed in two distinct functional states: in polysomes or in stored mRNP particles (29). Furthermore, regardless of the number of transcripts of a specific mRNA present in polysomes at any one time, it has been noted that the mean number of ribosomes bound to that transcript is constant (1,31,32). Under these conditions, translational control has been defined as the redistribution of specific mRNA transcripts between polysomes and mRNP particles as a function of an external stimulus (1,321. Indeed, despite the 34-fold increase in the number of ribosomes engaged in protein synthesis following serum stimulation (4,6,45), the overall mean polysome size remains constant, suggesting that most transcripts distribute to the same size polysomes, as for example p-actin transcripts (Fig. 5). In contrast eEF-la! mRNA redistributes from mRNP particles to large polysomes containing 11-12 ribosomes instead of shiRing into monosomes/ disomes, suggesting that the affinity of the translational apparatus for this transcript is enhanced following mitogenic stimulation. Enhanced expression of eEF-la mRNA could be brought about by an alteration in mRNA structure or a component of the translational apparatus. Changes in the 5' cap methylation or 3' end polyadenylation could be speculated to be responsible, but no evidence exists for either modification in this system. A model in which a specific component of the translational apparatus is altered would require differential recognition of a unique structure in the eEF-la transcript. Recently it has been demonstrated that a number of mRNA transcripts nanslational Control of eEF-1 a mRNA contain a stretch of polypyrimidines immediately downstream of the 5' cap, which act as a translational enhancer or suppressor, depending on the cell type (29,46,47). Almost all ribosomal protein mRNAs described to date have such polypyrimidine tracts, which vary in length and composition, but which are a conserved feature of these messages from amphibians to rodents (48,49). This feature is also seen in the 5'-untranslated region sequence of eEF-la mRNAfrom mouse as well as human fibroblasts (23,50). For QZ3 (see Fig. 11, whose function is unknown but whose message is also under translational control (8,51,52), a comparable structure has been described (53). Like eEF-la, QZ3 transcripts are distributed between mRNP particles and monosomes/disomes (data not shown). Thus, it will now be important to determine whether this motif is the structural element that is differentially recognized by the translational machinery following mitogenic stimulation.
The translational component most likely to be involved in recognizing a 5' mRNA structural motif would be one involved in binding message during the formation of the 48 S ribosomal preinitiation complex, the rate-limiting step in initiation (2). This complex is made up of the 43 S ribosomal preinitiation complex (composed of initiation factors eIF-2, eIF-3, Met-tRNAi and the 40 S ribosomal subunit), mRNA, and four additional initiation factors, eIF-4A, eIF-4B, eIF-4Fa, and eIF-4Fy. These latter initiation factors are thought to bind to the mRNA transcript, either independently or in unison with the 43 S initiation complex, and function to unwind the mRNA secondary structure and scan the transcript until the first AUG sequence is encountered (54,55). Furthermore, phosphorylation of a number of these components is known to be associated with altered rates of translation (2,56). Interestingly, preliminary results show that pretreatment of Swiss 3T3 cells with the immunosuppressant rapamycin, which selectively blocks the activation of the p70"6k/p85s6k and 40 S ribosomal protein S6 phosphorylation (571, also blocks the shift ofeEF-la transcripts into polysome with little effect on global protein synthesis (data not shown). Employing a very active in vitro reconstituted protein synthesizing system which has been described recently (58), it should be possible to determine whether 40 S ribosomes containing phosphorylated S6 are involved in preferentially recognizing eEF-la transcripts.
eEF-la appears to play a critical role in such processes as senescence and transformation. Indeed it is an essential gene product that may have pleiotypic roles in cell growth and differentiation (12)(13)(14)59). Here we show that the expression of the protein is closely monitored at the translational level as a function of the growth state of the cell. An understanding of the control mechanisms which modulate its activity should greatly enhance our knowledge of the mitogenic signaling mechanism.