Use of the novel technique of analytical ultracentrifugation with fluorescence detection system identifies a 77S monosomal translation complex

Abstract

A fundamental problem in proteomics is the identification of protein complexes and their components. We have used analytical ultracentrifugation with a fluorescence detection system (AU-FDS) to precisely and rapidly identify translation complexes in the yeast Saccharomyces cerevisiae. Following a one-step affinity purification of either poly(A)-binding protein (PAB1) or the large ribosomal subunit protein RPL25A in conjunction with GFP-tagged yeast proteins/RNAs, we have detected a 77S translation complex that contains the 80S ribosome, mRNA, and components of the closed-loop structure, eIF4E, eIF4G, and PAB1. This 77S structure, not readily observed previously, is consistent with the monosomal translation complex. The 77S complex abundance decreased with translational defects and following the stress of glucose deprivation that causes translational stoppage. By quantitating the abundance of the 77S complex in response to different stress conditions that block translation initiation, we observed that the stress of glucose deprivation affected translation initiation primarily by operating through a pathway involving the mRNA cap binding protein eIF4E whereas amino acid deprivation, as previously known, acted through the 43S complex. High salt conditions (1M KCl) and robust heat shock acted at other steps. The presumed sites of translational blockage caused by these stresses coincided with the types of stress granules, if any, which are subsequently formed.

Introduction

Proteomic analysis centers upon the key issue of determining which proteins associate together, under what conditions they do so, what are the sizes of the complexes that are formed, and how the composition of the complexes change in response to the environment. Protein complexes usually have been sized by gel exclusion chromatography or by sucrose gradient analysis. The former technique is limited in terms of identifying complexes of sizes much greater than two megadaltons and lacks resolution. Furthermore, interactions with the separating medium and dilution may lead to loss of material and dissociation of complexes. For both methods, a time-consuming and tedious secondary analysis (usually Western/Northern analysis) must be conducted to specifically identify the proteins/RNA present. Moreover, Western analysis is usually limited to about 15 or so samples, resulting in a tremendous loss of size resolution in the analysis.¹ Mass spectroscopic studies can also detect potential protein groupings but by its nature does not inform as to the size or purity of such complexes.

Analytical ultracentrifugation analysis (AUC), in contrast to the above techniques, allows the extremely rapid and precise (at least an order of magnitude better than sucrose gradient analysis) determination of the size of protein/RNA complexes. AUC is both able to sample several hundred positions across the cell and to take more than 200 scans to identify protein complexes. Moreover, AUC analysis is typically conducted at 20°C, a temperature closer to physiological conditions than that is used for sucrose gradient analysis (4°C). Also, AUC is conducted in physiological buffers without the presence of high concentrations of sucrose.

To maximize the advantages of AUC, we have used an analytical ultracentrifugation instrument with a fluorescence detection system (AU-FDS).^{2, 3} In this case, fluorescently tagged proteins or individually marked mRNA species (with GFP, for example) can be identified uniquely and all of the protein/RNA complexes containing a particular entity can be determined unambiguously and precisely.

In this article, we have used AU-FDS to identify translational complexes from Saccharomyces cerevisiae. The initiation of eukaryotic protein translation occurs through multiple steps.⁴ The mRNA is believed to form a closed-loop structure linking its 5′ end to its 3′ end in which eIF4E, the mRNA cap binding protein, interacts with the bridging protein, eIF4G, which in turn binds the poly(A)-binding protein (PAB1) that is bound to the poly(A) tail of mRNA.⁵ The resultant complex interacts with the 43S complex that consists of the 40S small ribosomal subunit, translation initiation factors eIF1, -2, -3, -5 and the charged methionine tRNA to form the 48S complex.⁶ This 48S complex then scans the mRNA for the initiation codon and brings in the 60S large ribosomal subunit to form the 80S ribosome that commences protein synthesis.

Under periods of environmental stress, translation ceases and mRNP complexes may be observed to form P bodies, a site of mRNA degradation, and large stress granules where mRNA is kept translationally silent until the stress ends.^7–11 The type of stress granule formed, if any, in yeast is highly dependent on the stress. For example, neither the stress of amino acid deprivation nor that of osmotic stress (1M KCl) results in stress granules, although osmotic stress does cause P body formation.^{7, 10, 11} Glucose deprivation, in turn, results in stress granules that contain mRNA, closed-loop structural components, and several other proteins but lack the 40S ribosome and the translation initiation factors eIF3, eIF1, and eIF5 present in the 43S complex.^{7, 9, 10} Robust heat shock, in contrast, results in stress granules that contain the 40S ribosome and the 43S translation initiation factors as well as the closed loop structural components.⁸ Since amino acid deprivation is dependent on eIF2α phosphorylation for its effects on decreasing translation, whereas the other three stresses are not, some of these differences in P body and stress granule formation may relate to the site of translational blockage.

The identification of translational complexes and their components has relied primarily on in vitro reconstitution experiments, which by their nature may not be indicative of the in vivo situation.¹² Techniques have also been developed to isolate the 43S, 48S, and other translational complexes from crude extracts by sucrose gradient analyses and thereby come closer to characterizing the complexes as they would exist in vivo.^13–16 To expedite the identification of translational complexes from crude extracts, we subjected extracts to a single step of affinity purification using two different points of entry into these complexes: Flag-tag attached to PAB1¹⁷ and to the ribosomal large subunit protein, RPL25A.^{18, 19} Using either affinity purification step followed by AU-FDS analysis, we primarily detected a 77S complex consistent with a monosomal translating complex: it contained mRNA, the 80S ribosome, and the closed-loop structural components of eIF4E, eIF4G, and PAB1 but no other translation initiation components. The translational stress of glucose deprivation that results in stress granule formation^7–9 caused the rapid decrease in abundance of the 77S complex and did so dependent on the presence of a functional eIF4E protein. In contrast, the stress of amino acid deprivation, known to act via eIF2α and to affect the ternary complex involved in 43S formation,¹⁸ required a functional eIF3 complex, a component of the 43S complex. Osmotic stress (1M KCl) acted to repress translation independently of these factors, and the stress of extreme heat shock displayed no dependency on PAT1, a repressor protein that we had shown to act in conjunction with eIF4E. These data begin to elucidate the site of action of multiple stress conditions on repressing translation initiation and establish that they work by multiple mechanisms. Importantly, the putative sites of action of these stresses relative to 77S complex formation were consistent with the type of stress granule that was formed. This application of AU-FDS to detect translational complexes following a single step purification of crude extracts should prove widely adaptable to the identification of protein complexes in other biological systems.

Results

Identification of translation complexes with AU-FDS

We first used the analytical ultracentrifuge (AU) to detect the sizes of ribosomal material in crude extracts. Cell extracts were subjected to AUC at a rotor speed of 15,000 rpm in order to readily differentiate complexes ranging from 20S to 250S in size. As visualized in Figure 1(A), following cycloheximide treatment to minimize polysomal run-off, a typical A₂₆₀ ribosomal profile was obtained that included well-demarcated 40S, 60S, 80S, and polysomal material. The complexes larger than 80S were confirmed to be polysomal material, as they decreased in abundance upon subjecting the cells to glucose deprivation [Fig. 1(A)], a treatment known to cause translational stoppage and formation of stress granules.^{7, 9, 21} The mean decrease of polysomal material following glucose deprivation was to 22% of that observed under normal growth conditions (standard error of the mean, 6.8%, six determinations), a reduction similarly observed by others using sucrose gradient analysis.^{7, 9, 10, 21, 22} Cells that were not treated with cycloheximide before cell lysis and AUC analysis, displayed about half the levels of the disomal and trisomal material and slightly increased levels of the 80S complex [Supporting Information Fig. 1(A,B)], indicative of some polysomal run-off occurring in the absence of cycloheximide which is in agreement with previous results.^{18, 23} In contrast to the polysomal fractions, the 80S complex increased in abundance with glucose deprivation, as observed previously.^{7, 21, 24} EDTA treatment caused nearly complete disappearance of the 80S complex [Supporting Information Fig. 1(C)], as EDTA is known to dissociate the 80S ribosome.²⁵

Figure 1.

AUC analysis identifies a 77S translation complex. A. Crude extracts prepared from strain AS319/YC504 were subjected to AUC analysis at 15,000 rpm and with absorbance optics (A₂₆₀). Cells were pregrown to mid-log phase on medium containing 2% glucose (glc +) and then shifted to medium containing no glucose for 10 min (glc + −). A 50S complex is also indicated that was often observed irrespective of carbon source treatment and may relate to a 50S complex previously described following AUC analysis of yeast ribosomal material but not observed following sucrose gradient analysis.²⁰ B. Flag pull downs using Flag-PAB1 were conducted from crude extracts before AUC analysis. Flag-PAB1-strain AS319/YC504; PAB1-strain AS319/YC360. C. Same as B except only strain AS319/YC504 was used: glc + (growth in glucose-containing medium); glc + − (pregrowth in glucose-containing medium followed by growth for 10 min in medium lacking glucose. D. AU-FDS analysis was conducted with strain XW1-2 (eIF4E-GFP) compared with AS319/YC504 (eIF4E). E. Same as D except strain XW1-2 (eIF4E-GFP/Flag-PAB1) was compared with YMK885 (eIF4E-GFP). Western analyses using anti-Flag antibody directed against Flag-PAB1 was used to verify that equivalent levels of Flag pull down material were present in cells treated with different growth conditions.

We subsequently used AU-FDS to study the sizes of translational complexes utilizing GFP fusions to various known components of the translation complex. Because yeast crude extracts contain background fluorescence that migrates in the 40S to 60S region, we identified translationally competent complexes following the affinity purification of PAB1 tagged at its N-terminus with Flag peptide.¹⁷

To characterize the sizes of the PAB1-mRNP complexes that we were purifying, we first subjected the Flag-purified material to AUC analysis using A₂₆₀ absorption to identify RNA-containing complexes. The major complex that was isolated migrated at about 80S [Fig. 1(B)]. Other complexes, migrating at about 40S, 60S, and 116S, were routinely detected [Fig. 1(B,C)]. Use of a PAB1 tagged at its C-terminus with Flag resulted in the identification of the same approximately 80S complex, indicating that the location of the Flag tag was not interfering with normal PAB1-mRNP complex formation [Supporting Information Fig. 1(D)]. Control Flag pull downs from an isogenic strain carrying PAB1 without the Flag tag, displayed very little RNA containing material [Fig. 1(B)]. The 80S complex, determined to migrate at 77S (standard error of the means, SEM, of 0.55S), was sensitive to glucose depletion [Fig. 1(C) and Supporting Information Fig. 1(D)]. This result indicates that the 77S complex corresponds to a translating complex and not to the 80S free ribosome. Conversely, this result further indicates that the 80S material observed in crude extracts that increases in abundance upon glucose deprivation, unlike the 77S complex that is associated with mRNA, consists principally of free 80S ribosomes and not the monosomal translation complex. The 77S complex also disappeared completely and shifted to smaller s-values upon pretreatment with EDTA [Supporting Information Fig. 2(A)]. The inability of the Flag immunoprecipitation of Flag-PAB1 to identify high levels of polysomal material may suggest that the Flag agarose beads bind polysomes poorly¹⁹ or that polysomes are not well detected by AU analysis [see Fig. 1(A)], as sucrose gradient analysis of crude extracts usually detects much more polysomal material than does AU analysis. Stabilizing polysomes with cycloheximide addition before conducting the Flag pull down did not enhance the abundance of the polysomal material that was identified [cf. Fig. 2(E) with that of Supporting Information Fig. 2(D)].

Figure 2.

AU-FDS analysis of 77S complex components using GFP fusions to translation factors. Growth conditions and AUC analysis were the same as described in Figure 1 except fluorescence optics was utilized to visualize GFP-tagged proteins. Glucose depletion was conducted as described in Figure 1(C), and amino acid depletion was conducted for 30 min (glu+, aa−). All strains carried Flag-PAB1. A. Strain XW1-2 (eIF4E-GFP). B. Strain XW3-1 (eIF4G1-GFP). C. Strain XW2-1 (eIF4G2-GFP). D. Strain RP1950-RPS4B/YC776 (RPS4B-GFP). E. Strain RPL6B/YC776 (RPL6B-GFP). Western analyses were conducted as described in Figure 1 to verify that equivalent levels of Flag pull down material was analyzed in each Figure.

To determine which if any of these PAB1-containing complexes identified by AUC contained the eIF4F translation factors, eIF4E and eIF4G, we used AU-FDS analysis to characterize complexes containing GFP fusions to the translation factors eIF4E and eIF4G that are components of the putative closed-loop structure involving PAB1 and mRNA. Strains carrying Flag-PAB1 and either eIF4E-GFP, eIF4G1-GFP, or eIF4G2-GFP were subjected to Flag-pull downs and AU-FDS analyses. All GFP fusion proteins used in this study have been shown to be functional and to not cause any apparent growth defects (data not shown).^{7, 9, 26} All three translation initiation proteins, eIF4E-GFP, eIF4G1-GFP, and eIF4G2-GFP, were found to migrate primarily in a complex of 77S [Figs. 1(D,E) and 2(A–C)] and much less so in a complex about 116S in size. All three of these proteins were either absent or diminished from the complexes that migrated around 40S, and 60S that were observed in the AU A₂₆₀ analysis. In control experiments, Flag pull downs from strains expressing Flag-PAB1 and no GFP fusion displayed no protein complexes by AU-FDS [Fig. 1(D)]. Similarly, conducting Flag immunoprecipitations with strains expressing PAB1 and tagged-GFP factors (such as eIF4E) coimmunoprecipitated no complexes [Fig. 1(E) for eIF4E]. The complex that migrated around 80S was determined to be 77S (with an SEM value of 0.67S) in size following the averaging of 15 analyses with eIF4E and eIF4G components. These analyses indicate that a one-step affinity purification of protein complexes containing GFP fusion proteins can avoid the fluorescence background difficulties observed in crude extracts and specifically detect complex multimeric associations.

The 77S and the 116S Flag-PAB1 purified complexes contained both the 40S and 60S ribosomes, as these complexes were also detected from strains carrying Flag-PAB1 and either small ribosomal subunit proteins RPS4B-GFP [Fig. 2(D)] or large ribosomal subunit proteins RPL6B-GFP [Fig. 2(E)] following Flag pull downs and AU-FDS analysis. Finally, to examine which of the Flag-PAB1 complexes carried mRNA, we coexpressed in yeast along with Flag-PAB1 the U1A RNA binding protein fused to GFP (U1A-GFP) and one of three different mRNA each carrying U1A binding sites in their 3′ UTRs: PGK1p-U1A, MFA2p-U1A, and tet off-MFA2-U1A.^{28, 29} After purification of Flag-PAB1, the resultant complexes were subjected to AU-FDS. As shown in Figure 3(A), each of the mRNA migrated in complexes around 25S, 77S, and 116S. Control AU-FDS experiments with strains expressing just U1A-GFP and Flag-PAB1 lacked these complexes [Fig. 3(B)].

Figure 3.

AU-FDS analysis identifies mRNA in the 77S translation complex. Growth conditions and AUC analysis were the same as described in Figure 2. In C and D, parental strains and strains carrying the cdc33-1 or prt1-1 alleles were analyzed at 37°C following growth at this temperature for 1 hr and 20 min, respectively. A. Strain AS319/YC504 containing plasmids RP1193 (MFA2-U1A), RP2037 (PGK1-U1A), or RP1291 (tetoff-MFA2-U1A) and plasmid RP1194 (U1A-GFP). B. Strain AS319/YC504/RP2037/RP1194 compared with control AS319/YC504/RP1194. C. Strain 21R/YC776 (PRT1) compared with TP11B/YC776 (prt1-1). D. Strain AS319/YC504 (CDC33) compared with AS1881/YC504 (cdc33-1). E. Strain XW3-1 (eIF4G1-GFP) compared with strains carrying GFP fusions to several translation initiation factors as indicated. F. Strain RP1730/YC504 (XRN1-GFP) compared with XW1-2 (eIF4E-GFP).

These data indicate that the 77S and 116S complexes contain the 80S ribosome, the eIF4F translation initiation factors, PAB1, and mRNA. Because of low yields in regards to the 116S complex, we focused our subsequent analysis on the 77S complex. Other possible PAB1 interacting proteins that are of equal or greater abundance in the cell than eIF4E or eIF4G, such as eIF4A (an unstable component of yeast eIF4F),^30–32 XRN1 (the mRNA degradation 5′-3′-exonuclease),²⁴ and RRP5 (nucleosomal component),³³ did not migrate in a 77S complex [see e.g., Fig. 3(E), eIF4A, and Fig. 3(F), XRN1]. Moreover, other translation initiation factors such as eIF2α, eIF3b, and eIF5 were not present in this complex [Fig. 3(E)]. The absence of these latter translation initiation factors suggests that the 77S complex does not contain both the 80S ribosome and large amounts of the 48S initiation complex.

The sensitivity of the 77S complex to two blocks in translational initiation were investigated. As shown in Figure 3(C), the prt1-1 allele, encoding a defect in the eIF3b subunit,¹⁵ reduced the level of the 77S complex to 66% of wild-type. The cdc33-1 allele, which encodes a defect in the eIF4E protein³⁴ reduced the 77S complex to 49% of wild-type [Fig. 3(D)]. These effects on 77S complex abundance are similar to the previous known effects of these same alleles on polysomal abundance. It should be noted that prt1-1 has been shown to decrease the abundance of polysomes with a large number of ribosomes to a much greater extent than it does polysomes containing only a few ribososmes.^{15, 25} The cdc33-1 allele shows a related, albeit, less pronounced bias in primarily affecting polysomes with many ribosomes.¹⁵

The 77S complex decreases in abundance upon glucose depletion

Upon the stress of glucose depletion, mRNA translation ceases, polysomes decrease in number, and mRNA, PAB1, and eIF4F factors move into stress granules.^{7, 9, 21} Consequently, we analyzed the effect of glucose deprivation on the complexes purified by Flag-PAB1. As shown in Figures 1(C) and 2 and Supporting Information Figure 2(F), each 77S complex purified with Flag-PAB1, whether it contained eIF4E-GFP, eIF4G-GFP, RPS4B-GFP, RPL6B-GFP, or mRNA, showed a dramatic decrease in abundance upon the removal of glucose (e.g., for eIF4E-GFP, 17% of wild-type, seven samples, SEM, of 3.7%; RPS4B-GFP, 19% of wild-type, three samples, SEM of 5.6%).

Importantly, readdition of glucose for 10 min following glucose depletion resulted in restoration of the 77S complex [RPS4B-GFP shown in Fig. 2(D), and eIF4E-GFP, PAB1-GFP, and mRNA in Supporting Information Fig. 2(B,C,F), respectively].²¹ When these experiments were repeated using A₂₆₀ analysis of the 77S complex, the 77S complex on average was reduced to 23% of wild-type (SEM of 1.2% for seven samples) [Fig. 1(C)]. Notably, in contrast, the 60S and 40S complexes, routinely observed with absorbance AUC analysis, displayed reduced sensitivity to glucose depletion: they decreased to only 66% and 63% of wild-type, respectively. Therefore, the 77S complex, similar to polysomes, is sensitive to glucose depletion.

Pretreatment with cycloheximide before shifting cells to glucose-depleted medium is known to block the subsequent stress granule formation.²⁴ We similarly found that cycloheximide completely blocked the disappearance of the 77S complex upon glucose depletion [Supporting Information Fig. 2(D)], indicating that polysomal run-off is a required step for disappearance of the 77S complex. Moreover, because blocking elongation with cycloheximide did not alter the abundance of the 77S complex that we observed, the 77S complex cannot be viewed as a remnant complex resulting from polysomal run-off. Treatment of the cells with cycloheximide after glucose depletion, in contrast, had no effect on the reduction in abundance of the 77S complex upon glucose deprivation [Supporting Information Fig. 2(E)], indicating that the decrease in 77S complex abundance occurred in vivo before preparation of cell extracts.

Identification of the 77S complex by AU-FDS using RPL25A-Flag

To analyze these translation complexes further, we conducted affinity purification with the large ribosome subunit protein RPL25A-Flag that is known to copurify yeast ribosomes.^{18, 19} Following a single step purification of RPL25A-Flag in strains containing RPS4B-GFP, we identified a large amount of an 80S complex and some polysomal material that migrated around 116S [Fig. 4(A)]. Similar results were obtained with RPL25A-Flag immunoprecipitations with RPL6B-GFP [Fig. 4(B)] and RPL7A-GFP (data not shown). These 80S complexes identified by RPL25A-Flag, however, were insensitive to glucose depletion [Fig. 4(A,B)], indicating that they were not the 77S complex identified by affinity purification of Flag-PAB1 but were similar to the 80S complex observed in crude extracts.

Figure 4.

AU-FDS analysis of translation complexes using RPL25A-Flag. Growth conditions and Flag pull downs were the same as described in Figure 2 except RPL25A-Flag (plasmid JC288)^{18, 19} was used in place of Flag-PAB1. A. Strain RP1950-RPS4B. B. Strain RPL6B. C. Strain YMK885 (eIF4E-GFP). D. Strain YMK1172 (eIF4G1-GFP). E. Strain RP2191 (PAB1-GFP).

Because the high abundance of free 80S ribosomes might be masking a comigrating 77S translation complex, we subsequently used RPL25A-Flag to immunoprecipitate eIF4E-GFP that would be expected to be a component of the 77S translating ribosome but not to be a component of the free 80S ribosome. RPL25A-Flag in combination with eIF4E-GFP identified a 77S complex containing eIF4E-GFP that became reduced upon glucose depletion (reduction to about 13% of nonstressed abundance) [Fig. 4(C)], indicating that this complex behaved the same as that identified with Flag-PAB1 [Fig. 2(A)]. The exact same results were obtained when RPL25A-Flag was used to pull down eIF4G1-GFP and PAB1-GFP [Fig. 4(D,E), respectively], confirming that eIF4E, eIF4G1, and PAB1 associate in a 77S complex identifiable by immunoprecipitation by either RPL25A-Flag or Flag-PAB1 that is sensitive to the effects of glucose depletion. This complex is distinct from the free 80S ribosome that is insensitive to the effects of glucose depletion and which is detected either with RPL25A-Flag when following GFP tagged ribosomal components or in crude extracts.

The ability of the stress of glucose deprivation and PAT1 to repress translation requires a functional eIF4E-mRNA cap interaction

The above data indicate that our AU-FDS analysis can specifically detect the monosomal translation complex, a form not previously visualized by sucrose gradient analysis. We subsequently used this detection to quantitate the changes in abundance of the monosomal 77S complex in response to different stress conditions. To initiate this analysis we first determined at what step in the initiation process glucose depletion repressed 77S complex formation. We hypothesized that the stress of glucose depletion might be targeting components of the putative closed-loop structure since PAT1 has been shown to be required for the translation repression by glucose depletion²⁴ and PAT1 affects 48S formation in vitro.³⁵ We initially analyzed different deletions in PAB1, as the RRM2 domain (residues 125–203 of 577 total residues) of PAB1 is known to be required for stably interacting with eIF4G.^36–38 Deleting the RRM2 domain of PAB1, however, had no effect on 77S complex abundance under glucose growth conditions (data not shown) and did not affect the decrease in 77S complex abundance that occurs following glucose depletion [Supporting Information Fig. 3(A)]. Deleting RRM1 (residues 1–116), which dramatically impairs the ability of CCR4, the primary yeast mRNA deadenylase, to remove the mRNA poly(A) tail by apparently blocking PAB1 dissociation from the poly(A) tail,^{17, 39} also had no effect on 77S complex reduction upon glucose deprivation [Supporting Information Fig. 3(B)]. Therefore, the reduction in abundance of the 77S complex that occurs upon glucose deprivation does not appear to involve interfering with PAB1-eIF4G interactions or PAB1-poly(A) interactions.

At the other side of the putative closed-loop structure, eIF4E contacts the 5′ mRNA cap structure, and the cdc33-1 mutation in the gene encoding eIF4E has been shown to block this interaction and reduce translation.^{15, 17, 34, 40} When we conducted the same experiment with the cdc33-1 allele, the 77S complex was reduced to only 50% of the nonstress conditions abundance following glucose deprivation as compared with the reduction to 23% of the nonstress condition observed in the wild-type strain [Fig. 5(A); Table I]. These results suggest that a defect in eIF4E significantly attenuates the ability of the stress of glucose depletion to repress translation. In contrast, following the stress of glucose deprivation in a prt1-1 background, we observed that the 77S complex was reduced to 30% of its nonstress abundance (in a wild-type background glucose depletion reduced the 77S complex to 23% of its nonstress abundance) [Fig. 5(B); Table I], indicating that a block in eIF3b function had a more marginal effect. These results indicate that translational repression by glucose deprivation requires a functional eIF4E-mRNA cap interaction.

Figure 5.

Effect of glucose depletion on 77S complex abundance as affected by different mutations. Growth conditions and AUC analysis were the same as described in Figures 1 and 2 using strains containing Flag-PAB1. Strains carrying the cdc33-1 or prt1-1 alleles were analyzed at 37°C following growth at this temperature for 1 hr and 20 min, respectively. Wild-type strains coanalyzed at 37°C displayed the same effects of stress conditions as did wild-type strains grown at 30°C. The stress of glucose depletion was as described in Figure 2 and used in A–D. For E and F, cells were subjected to extreme heat shock (shift from 30° to 46°C for 10 min). A. Strain AS1881/YC504 (cdc33-1). B. Strain TP11B/YC776 (prt1-1). C. Strain BY5797/YC776 (pat1Δ). D. Strain XW6-1/YC788 (pat1Δ cdc33-1). E. Same as Figure 1(C). F. Same as “C” above.

Table I.

Abundance of 77S Complex as Percent of that Observed in the Unstressed Growth Condition

Strain	Glucose depletion	Amino acid depletion	Osmotic stress	Heat shock
Wild-type	23 ± 1.2	53 ± 9.0	45 ± 3.3	17 ± 3.5
cdc33–1	52 ± 4.5	64 ± 8.1	50 ± 10	N.D.
prt1–1	30 ± 4.1	83 ± 6.6	52 ± 8.8	ND
pat1Δ	50 ± 6.6	65 ± 4.4	ND	20 ± 5.4
pat1Δ cdc33–1	55 ± 0.51	ND	ND	ND

The abundance of the 77S complex was determined following AU analysis as described in the text. All stress analyses were conducted at the same time as were the nonstressed conditions. The values represent the average (± SEM). For the glucose deprivation studies with wild-type, cdc33–1 (defective in eIF4E), and prt1–1 (defective in eIF3b) alleles, four to seven separate analyses were conducted For the similar set with amino acid depletion, three to four analyses were conducted. For the experiments with pat1Δ and for the osmotic and heat shock stress experiments (addition of 1M KCl and shift to 46°C, respectively, two to three analyses were conducted.

ND, not determined.

We subsequently analyzed the effect of a PAT1 deletion on the ability of glucose depletion to reduce 77S complex abundance. As shown in Figure 5(C), the pat1Δ resulted in the 77S complex being reduced to only 49% of its nonstress levels (compared with the reduction to 23% of the nonstress condition observed in the wild-type strain), confirming the requirement of PAT1 for the repression of translation by glucose deprivation. When we combined the pat1Δ with that of the cdc33-1 allele, the 77S complex abundance was reduced to 55% of the nonstress condition [Fig. 5(D); Table I], a value that was almost the same as observed with either the pat1Δ or cdc33-1 allele alone. These results suggest that PAT1 acts at a site requiring eIF4E-mRNA cap interactions for repressing translation.

Amino acid deprivation, osmotic stress, and robust heat shock act to repress translation by different mechanisms than glucose deprivation

Because of the above results with glucose deprivation, we used the effects of the two different blocks in translation initiation (cdc33-1 and prt1-1) to identify specific steps in translation initiation that were being targeted by other stress-induced translational blockages. By this means we could determine if a particular stress required a functional eIF4E or eIF3b protein for its effects.

In contrast to the results from glucose deprivation, amino acid starvation causes the 77S complex to be reduced to 53% as compared with nonstress conditions [Fig. 2(A)]. Subsequent analysis suggested that amino acid deprivation operated independently of eIF4E but required a functional eIF3b protein. In a prt1-1 background, the stress of amino acid depletion caused the 77S complex to be reduced to only 83% of the nonstress condition (indicating an attenuation of the stress response) whereas the cdc33-1 allele had no significant effect on the ability of amino acid depletion to reduce the abundance of the 77S complex (the 77S complex was reduced to 64% as compared with the nonstress condition, which is similar to the effect of amino acid depletion observed in the wild-type strain) [Fig. 6(A,B), respectively; Table I]. These results are in agreement with the known effect of amino acid stress on eIF2α function and 43S complex formation. In confirmation of this result, the pat1Δ had no significant effect on the ability of amino acid deprivation to repress 77S complex abundance [Fig. 6(C); Table I].

Figure 6.

Effect of amino acid depletion and osmotic stress on 77S complex abundance. Amino acid depletion (aa + −), A–C, was conducted for 30 min as described in Figure 2. Osmotic stress (D–F) was initiated with addition of 1M KCl to cells for 30 min (KCl − +). A and E. Strain TP11B/YC776 (prt1-1). B and F. Strain AS1881/YC504 (cdc33-1). C. Strain BY5797/YC776 (pat1Δ). D. Strain AS319/YC504.

To expand on these patterns, we assayed the effect of osmotic stress (addition of 1M KCl) on 77S complex abundance. Increased KCl has been suggested to act through an eIF2α-independent mechanism in stopping translation.^{41, 42} Osmotic stress is also dissimilar from glucose deprivation in that it results in P bodies but not stress granules.^{10, 11} Addition of 1M KCl reduced the 77S complex abundance to 45% of nonstress conditions [Fig. 6(D)]. This degree of reduction of 77S complex abundance was unaffected by either the prt1-1 or cdc33-1 alleles [Fig. 6(E,F), respectively; Table I]. Osmotic stress appears, therefore, to act at a site other than requiring eIF4E-mRNA cap interactions or eIF3b function.

Finally, we addressed the issue as to where the stress of extreme heat shock affected translational processes.⁸ Heat shock, like glucose deprivation, inhibits translation by an eIF2α-independent mechanism and causes stress granules to be formed. However, the stress granules formed upon heat shock contain the 40S ribosome and eIF1 and eIF3 translation initiation components that are lacking in stress granules found following glucose removal.^7–9 The stress of extreme heat shock was found to reduce 77S complex abundance to 17% of the nonstress condition [Fig. 5(E); Table I]. Because the very nature of heat shock obviated our ability to use the temperature sensitive prt1-1 and cdc33-1 alleles, we instead used the pat1Δ to determine if heat shock was acting through eIF4E. In a pat1Δ background, heat shock resulted in a reduction of 77S complex to 20% of the nonstress condition, a value not significantly different than the effect of heat shock in the wild-type strain [Fig. 5(F); Table I]. This result suggests that heat shock works independently of PAT1 and eIF4E contacts to the mRNA.

Discussion

Identification of a 77S translation complex

We have used the technique of AU-FDS to identify a 77S protein translation complex from affinity-purified extracts. This complex contained mRNA, the closed-loop structure components of eIF4E, eIF4G, and PAB1, and the 80S ribosome. All these components were verified to be part of the 77S complex based on two principal criteria: GFP fusions to these components were visualized in a 77S complex following affinity purification, and their presence in this complex decreased upon the stress of glucose depletion and increased with the reintroduction of glucose. As this stress results in translational cessation, the depletion of the 77S complex is consistent with it representing a critical functional translational complex. Blockage with cycloheximide, which inhibits polysomal run-off, correspondingly inhibited the effect of glucose deprivation on 77S complex abundance, implying that the 77S complex is actively engaged in translation. In addition, defects either in eIF4E binding to the mRNA cap or in eIF3b function that target different steps in the initiation of translation reduced 77S complex abundance. Two different handles were used to gain access to the 77S complex, Flag-PAB1 and RPL25A-Flag, establishing the robustness of the analysis. Previous studies on translation complexes have not purified significant quantities of a monosomal translating complex containing the 80S ribosome, eIF4E, eIF4G, and PAB1, establishing the uniqueness of AU-FDS to identify novel or difficult to observe complexes.

Current models of eukaryotic translation suggest that upon formation of the 80S translational mRNA structure, translation initiation factors that aid in AUG scanning and 60S binding to the mRNA disengage from the mRNA. We found no evidence for stable association of eIF2, eIF3, or eIF5 components in the 77S complex, implying that a 48S initiation complex is either not present or not abundant in our 77S complex. Other experiments involving the use of RPS4B-Flag as a point of entry into the 77S complex has shown that there exists essentially only one RPS4B component to each 77S complex (unpublished observation).

This identification of a 77S complex that contains the 80S ribosome bound to mRNA and associating with eIF4E, eIF4G, and PAB1 also suggests that the 77S complex displays significant asymmetry. The average mass of such a group of proteins would be about 5.0 MDa whose spherical s-value would be about 93S. Given that an mRNA bound to an 80S ribosome would not be spherical, its migration would be considerably slowed. A 5.0 MDa complex with a ratio of length to width of about 2:1 or 3:1 would migrate at around 77S.

In contrast to the above results with affinity purification of Flag-PAB1 complexes, we observed in the AU analysis of crude extracts that polysomes become depleted and the 80S free ribosomal component (as well as the 40S and 60S complexes) becomes enriched following glucose deprivation [Fig. 1(A)]. This result is identical to that obtained in ribosomal profiles following sucrose gradient analyses of crude extracts^{21, 24} and recent purifications of the 80S free ribosome.⁴³ Since the 80S complex observed in crude extracts does not behave in the same manner as polysomal material in regards to glucose deprivation and behaves similarly to the 40S and 60S complexes, it is probable that the 80S complex represents the free 80S ribosome and not simply the monosomes attached to mRNA. Using RPL25A-Flag as our entry into translational complexes we further verified that we could separately detect the free 80S ribosome which was insensitive to the effects of glucose depletion [see Fig. 4(A,B)] and the 77S complex which was sensitive to the effects of glucose depletion [Fig. 4(C–E)]. Therefore, the 77S complex is functionally related to polysomal material whereas the 80S complex identified by AU-FDS when following RPL25A-Flag pull downs in the presence of GFP tagged ribosomal components is not.

Flag-PAB1 would also be expected to be present in a 48S translation initiation complex that we did not readily detect. Other smaller complexes containing Flag-PAB1 were routinely detected with AU analysis using A₂₆₀ optics, albeit in much reduced and variable abundance; these complexes were around 20S, 40S, and 60S [Fig. 1(A) and unpublished observation]. It remains to be determined whether any of these complexes represent the 48S complex. The use of other Flag tagged translation factors as handles may be required to purify the 48S complex or to obtain better yields of these other complexes.

Effect of stresses on the 77S translation complex

The abundance of the 77S complex decreased significantly upon the stress of glucose deprivation. We showed that blocking eIF4E binding to the mRNA cap attenuated this response by at least twofold. In contrast, the inactivation of eIF3b had a much more minor affect on the ability of glucose depletion to repress translation. We interpret these results to indicate that the translational repression caused by glucose removal acts to a major extent through eIF4E. This would place the translational repression occurring either in 48S complex formation or in the initial association of the eIF4E with the mRNA. In agreement with our data, the PAT1 protein, shown to be required for the translational repression by glucose deprivation, blocks 48S complex formation in vitro.³⁵ We additionally demonstrated that deletion of the PAT1 gene attenuated the translational repression by glucose deprivation, as expected based on the previous findings. Moreover, when the PAT1 gene deletion was combined with the cdc33-1 allele defective in eIF4E function, no additional effect on translational repression was observed. This result suggested that PAT1 was acting through eIF4E function in its repression of translation.

An alternative hypothesis to that of the cdc33-1 allele attenuating the translation defect caused by glucose deprivation is that under glucose starved conditions, about 25% of the total wild-type translation occurs by an eIF4E-independent mechanism. In this scenario, the translation that occurs in the cdc33-1 allele background could not be reduced under glucose deprivation conditions beyond this point (50% reduction by the cdc33-1 allele and at most another 50% reduction upon glucose deprivation), thereby resulting in the apparent attenuation of the glucose deprivation effect on the 77S complex with the cdc33-1 allele. This hypothesis seems less likely for the following reason. The deletion of the PAT1 gene which reduced the 77S complex to about 78% of wild-type under glucose growth conditions (data not shown, an observation consistent with the known small distinct effect of the pat1Δ in reducing polysomal abundance),¹ still displayed only a 50% reduction in 77S complex abundance upon glucose deprivation, similar to that observed with the cdc33-1 allele. The resultant reduction in abundance of the 77S complex to about 39% of the wild-type levels is significantly above the hypothesized limit to eIF4E effects (23% of wild-type). This implies that such a limit does not exist.

An additional model explaining the mechanism by which the prt1-1 and cdc33-1 alleles reduce 77S complex abundance would be that these alleles also increase the rate of mRNA degradation.⁴⁴ This degradation of the mRNA could be expected to reduce the pool of mRNA bound to the ribosome. Changes in mRNA degradation rates may, however, not exert large effects on the 77S complex abundance. For example, the deletion of the RRM1 domain of PAB1 causes a dramatic decrease in mRNA degradation (Yao et al., 2007),¹⁷ and yet, we observed no effects on 77S complex abundance.

In contrast to these effects on translation initiation by the stress of glucose deprivation, the starvation for amino acids did not appear to act principally through eIF4E. The overall effect of amino acid depletion on 77S complex abundance was significantly less than that of glucose depletion. In addition, the prt1-1 allele attenuated the effect of amino acid depletion on 77S complex abundance while the cdc33-1 allele had no significant effect on this process. Furthermore, the pat1 deletion had no significant effect on the ability of amino acid removal to repress translation. These data indicate that amino acid depletion does not act through eIF4E, in agreement with previous data identifying its principal site of action as involving eIF2α.⁴⁵

The third stress that we analyzed, osmotic stress (1M KCl), in contrast to the above two stresses, appeared to exert its effects on translation initiation separately from both eIF4E and eIF3b function. Finally, the stress of extreme heat shock had the most effect on 77S complex abundance and this effect, in turn, was unaffected by deleting the PAT1 gene. Heat shock, therefore, appears to repress translation independently of the pathway involving PAT1 and presumably eIF4E. Previous studies had already shown that heat shock acted in an eIF2α-independent manner.⁸

The suggested sites of action of stresses upon 77S complex abundance correlate with the types of stress granule, if any, which are formed following translation cessation

The differences in stress-induced reductions in translation described above imply that there are multiple mechanisms by which this process can be initiated. Given that there exists a number of potential targets within the initiation process for translational repression,^{46, 47} other known mutations in initiation factors will need to be studied to more precisely determine how osmotic stress and heat shock block translation. This variety in mechanistic targets within the translational process may have consequential effects on subsequent steps involving P body and stress granule formation, resulting in the wide variation that is observed in the formation of these mRNP structures. For example, because glucose deprivation results in stress granules lacking 43S complex components, including eIF3 and the 40S ribosome, it is likely that its effect on translation occurs before 43S association with mRNP structure and therefore on the structure just containing mRNA-PAB1-eIF4E-eIF4G. This presumed site of action agrees with our results indicating that glucose deprivation acts through eIF4E. In contrast, since amino acid deprivation acts through the 43S complex formation, it would be predicted to not result in mRNA entering stress granule formation, which is what is observed. Similarly, extreme heat shock, which acted independently of PAT1 (and presumably of eIF4E), forms stress granules containing 43S components and eIF4E, eIF4G, and PAB1, consistent with it blocking translation at the function of the 48S complex. Based on these correspondences, we suggest that osmotic stress, in not forming stress granules, must be acting to reduce translation before association of the 43S complex with the mRNA-eIF4E-eIF4G-PAB1 complex. These correspondences are summarized in Figure 7. One future goal in this field would be to clarify not only how translation ceases but also how that cessation influences subsequent steps in maintaining mRNA in a quiescent state.

Figure 7.

Model of 77S translation complex and target site for the stresses of glucose deprivation, amino acid depletion, extreme heat shock, and osmotic stress. Three PAB1 molecules are displayed with the assumption that the average yeast mRNA has a full-length poly(A) length of 75 A's and that PAB1 binds 25 A's. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Methods

Yeast strains and growth conditions

Yeast strains are listed in Supporting Information Table I. Cells were grown to mid-log phase in synthetic complete medium with appropriate amino acids as described before.¹⁷ Generally, for absorption AUC analyses 200 mL of cells were used and 400 mL for AU-FDS analyses. Cell lysis and Flag pull downs have been described.^{48, 49} Generally, 1 mL of a 15 to 25 mg/mL crude extract was incubated initially with the Flag beads. After Flag peptide elution, the protein concentration that was analyzed by AUC was in the 0.1 to 0.3 mg/mL range. Control experiments conducted with strains lacking a Flag-tagged protein resulted in protein concentrations in the 0.02 mg/mL range following Flag peptide elution. Growth was at 30°C except for temperature-sensitive mutants, which were first grown at 26°C before shifting to 37°C for 20 min or 1 hr depending on the strain. For glucose depletion or amino acid starvation, cell pellets from the undepleted medium were washed and then resuspended in fresh medium lacking glucose or amino acids for 10 min and 30 min, respectively. Glucose readdition experiments were conducted by adding the requisite amount of glucose (2%) directly to cultures. Osmotic stress was initiated by resuspending cells in media containing 1M KCl. The stress of heat shock was monitored following shifting of cells to 46°C for 10 min.⁸ Cycloheximide was added to growing cultures at a concentration of 100 μg/mL as described.²¹

AU analyses

Flag eluted samples (350 μL) were subjected to AU analysis by using A₂₆₀ absorption to identify RNA-containing complexes or fluorescence detection system (AU-FDS) to detect GFP-fusion proteins. Twelve micrometers Spin60 centrifuge cells were used for the AU experiments. All analytical ultracentrifugation experiments were conducted at 20°C and at a rotor speed of 15,000 rpm usually for at least 150 scans for AU-FDS experiments and at least 75 scans for AU analysis. Data were analyzed by SEDFIT software. The sedimentation coefficient S* was converted to standard conditions (pure water at 20°C, S_20,w) using the equation below. The modified equation for our experiments is S_20,w = 1.51S*.

(1)

where ρ, solvent density; η, viscosity; -protein's partial specific volume.

Western analyses

Western analyses using anti-Flag antibody directed against Flag-PAB1 or RPL25A-Flag was used to verify that equivalent levels of Flag pull down material was present in cells treated with different growth conditions. A contaminant protein reacting with anti-Flag antibody was used to verify that Flag pull downs were conducted similarly when comparing strains containing Flag-PAB1 and containing only PAB1. Similar Western analysis comparisons were conducted for all experiments whenever two mutants or growth conditions were directly compared in the same centrifugation analysis.

Supplementary material

Additional Supporting Information may be found in the online version of this article.