The ␤/Gcd7 Subunit of Eukaryotic Translation Initiation Factor 2B (eIF2B), a Guanine Nucleotide Exchange Factor, Is Crucial for Binding eIF2 In Vivo

MOLECULAR AND CELLULAR BIOLOGY, Nov. 2010, p. 5218–5233 Vol. 30, No. 210270-7306/10/$12.00 doi:10.1128/MCB.00265-10Copyright © 2010, American Society for Microbiology. All Rights Reserved.

The �/Gcd7 Subunit of Eukaryotic Translation Initiation Factor 2B(eIF2B), a Guanine Nucleotide Exchange Factor,

Is Crucial for Binding eIF2 In Vivo�

Kamal Dev,† Hongfang Qiu, Jinsheng Dong, Fan Zhang,Dominik Barthlme,‡ and Alan G. Hinnebusch*

Laboratory of Gene Regulation and Development, Eunice Kennedy Shriver National Institute ofChild Health and Human Development, Bethesda, Maryland 20892

Received 9 March 2010/Returned for modification 13 April 2010/Accepted 26 July 2010

Eukaryotic translation initiation factor 2B (eIF2B) is the guanine nucleotide exchange factor (GEF) foreukaryotic translation initiation factor 2, which stimulates formation of the eIF2–GTP–Met-tRNA i

Met ternarycomplex (TC) in a manner inhibited by phosphorylated eIF2 [eIF2(�P)]. While eIF2B contains five subunits,the �/Gcd6 subunit is sufficient for GEF activity in vitro. The �/Gcd2 and �/Gcd7 subunits function with �/Gcn3in the eIF2B regulatory subcomplex that mediates tight, inhibitory binding of eIF2(�P)-GDP, but the essentialfunctions of �/Gcd2 and �/Gcd7 are not well understood. We show that the depletion of wild-type �/Gcd7, threelethal �/Gcd7 amino acid substitutions, and a synthetically lethal combination of substitutions in �/Gcd7 andeIF2� all impair eIF2 binding to eIF2B without reducing �/Gcd6 abundance in the native eIF2B-eIF2 holo-complex. Additionally, �/Gcd7 mutations that impair eIF2B function display extensive allele-specific interac-tions with mutations in the S1 domain of eIF2� (harboring the phosphorylation site), which binds to eIF2Bdirectly. Consistent with this, �/Gcd7 can overcome the toxicity of eIF2(�P) and rescue native eIF2B functionwhen overexpressed with �/Gcd2 or �/Gcd1. In aggregate, these findings provide compelling evidence that�/Gcd7 is crucial for binding of substrate by eIF2B in vivo, beyond its dispensable regulatory role in theinhibition of eIF2B by eIF (�P).

Initiator methionyl tRNA (Met-tRNA iMet) is recruited to the

small (40S) ribosomal subunit in the ternary complex (TC)containing activated, GTP-bound initiation factor 2 (eIF2). AUGrecognition triggers the completion of GTP hydrolysis, with re-lease of Pi from eIF2-GDP-Pi, insertion of Met-tRNA i

Met in theribosomal P site, and dissociation of inactive eIF2-GDP from thetranslation preinitiation complex. As the binary complex eIF2-GDP dissociates slowly and eIF2 binds more tightly to GDPthan GTP, eIF2B is needed to recycle eIF2-GDP to eIF2-GTPfor rapid reassembly of the TC. eIF2B contains five subunits (�through ε) well conserved between yeast and mammals andoccurs in a 1:1 holocomplex with eIF2 (7, 39). Except foreIF2B� (encoded by GCN3), the other four eIF2B subunitsare essential in yeast. Remarkably, only the C-terminal portionof ε/Gcd6 is sufficient for guanine nucleotide exchange factor(GEF) function in vitro, albeit with lower activity than for theeIF2B holocomplex (22), and contains a key catalytic residueand determinants for binding of the � and � subunits of eIF2(36).

Recycling of eIF2 by eIF2B is impaired by phosphorylationof eIF2 on serine 51 of its � subunit. Phosphorylated eIF2[eIF2(�P)]-GDP is a poor substrate for nucleotide exchange

and binds more tightly to eIF2B than does unphosphorylatedeIF2-GDP, thus acting as a competitive inhibitor. The eIF2�kinase in budding yeast, Gcn2, is activated by amino acid lim-itation and produces eIF2(�P) at levels that do not fully blocktranslation initiation and that specifically increase translationof GCN4, a transcriptional activator of amino acid biosyntheticgenes under general amino acid control. Induction of GCN4translation by eIF2(�P) is mediated by short open readingframes (uORFs) in the leader of GCN4 mRNA (25). Undernonstarvation conditions, ribosomes that translate uORF1 canresume scanning but then reinitiate at uORF2, -3, or -4 andleave GCN4 untranslated. When TC levels are reduced byeIF2(�P), a fraction of 40S subunits that translate uORF1 andresume scanning do not rebind TC until after bypassinguORF2 to -4 and can reinitiate at GCN4 instead (26).

The nonessential �/Gcn3 subunit of yeast eIF2B mediatesthe inhibitory effect of eIF2(�P) on eIF2B, such that gcn3�mutants are defective for derepression of GCN4 in aminoacid-starved cells (Gcn� phenotype). Point mutations confer-ring a Gcn� phenotype but no growth defect on nonstarvedcells (resembling gcn3�) were identified in the essential�/Gcd7 and �/Gcd2 subunits of yeast eIF2B. Both the elimi-nation of �/Gcn3 and such Gcn� substitutions in �/Gcd7 over-come the inhibition of eIF2B by eIF2(�P) without impairingGEF function in vitro (37). �/Gcn3, �/Gcd7, and �/Gcd2 haveregions of sequence similarity, and co-overexpressing all threeproduces a stable subcomplex that can bind eIF2 or recombi-nant eIF2� in vitro in a manner stimulated by Ser-51 phosphor-ylation (32, 37). There is genetic evidence that this regulatorysubcomplex can sequester eIF2(�P) and prevent it from inhib-iting native, five-subunit eIF2B in vivo (37). The �/Gcd7 Gcn�

* Corresponding author. Mailing address: NIH, Building 6, Room230, Bethesda, MD 20892. Phone: (301) 496-4480. Fax: (301) 496-6828.E-mail: [email protected].

† Present address: Department of Biotechnology, Shoolini Univer-sity of Biotechnology and Management Sciences, Solan 173221, Him-achal Pradesh, India.

‡ Present address: Institute of Biochemistry, Biocenter, Goethe Uni-versity Frankfurt, D-60438 Frankfurt am Main, Germany.

� Published ahead of print on 30 August 2010.

5218

on Septem

ber 8, 2015 by guesthttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

mutations and Gcn� mutations in the S1 domain of eIF2�(contains Ser-51) decrease binding of the eIF2B regulatorysubcomplex and eIF2B holocomplex to the phosphorylated �subunit of eIF2 (eIF2�-P) in vitro (13, 32). Thus, it was pro-posed that tight binding of eIF2�-P to the eIF2B regulatorysubcomplex blocks productive interaction between the ε/Gcd6catalytic domain and the GDP binding pocket in eIF2� thatmediates nucleotide release (see model in Fig. 9A and B).

Nonlethal substitutions have been identified in all five eIF2Bsubunits that derepress GCN4 translation in the absence ofeIF2� phosphorylation in gcn2 mutants (Gcd� phenotype),and it is thought that such mutations constitutively impaireIF2B activity in a manner that bypasses the inhibitory func-tions of Gcn2 and eIF2(�P). This provides evidence that all ofthe subunits in eIF2B contribute at some level to the GEFactivity of the complex. However, certain Gcd� substitutions inthe nonessential �/Gcn3 subunit (encoded by gcn3c alleles) arethought to impair eIF2B function by conferring an abnormallytight interaction of the eIF2B regulatory subcomplex with un-phosphorylated eIF2�, mimicking the inhibitory effect ofSer-51 phosphorylation. Hence, Gcd� substitutions in �/Gcd7or �/Gcd2 could similarly interfere with nucleotide exchangerather than eliminating a positive contribution of these sub-units to GEF function.

There is conflicting evidence about whether the eIF2B reg-ulatory subunits make critical contributions to the binding ofunphosphorylated substrate, eIF2-GDP, in addition to medi-ating the inhibitory interaction with eIF2(�P). Supporting thisidea, the Gcn� �/Gcd7 substitutions mentioned above de-crease interaction between eIF2B and unphosphorylated eIF2in vitro (32). Moreover, a rabbit four-subunit eIF2B complexlacking the � subunit did not copurify with eIF2 and hadreduced GEF activity that was recovered by recombinanteIF2B� (8, 48). On the other hand, four-subunit yeast andmammalian eIF2B complexes were found to have normal GEFactivity that was merely insensitive to eIF2(�P) (17, 31, 37).Furthermore, the �/Gcd7 substitution V341D evokes dissoci-ation of the eIF2B regulatory subcomplex and reduces GEFactivity without diminishing eIF2 binding to the �/Gcd1-ε/Gcd6 catalytic subcomplex of eIF2B (42). These and otherfindings (2, 23) suggest that the eIF2B regulatory subcomplexenhances catalysis rather than substrate binding.

In this study, we have characterized mutants allowing con-ditional depletion of �/Gcd7 and �/Gcd2 and thereby obtainedstrong evidence that �/Gcd7 is crucial for efficient eIF2 bindingto eIF2B. Supporting this conclusion, we identified lethal gcd7mutations and synthetic lethal combinations of �/Gcd7 andeIF2� substitutions that impair eIF2 association with the intacteIF2B complex. These results and extensive allele-specific in-teractions between �/Gcd7 substitutions and mutations in theeIF2� S1 domain support the idea that interaction between�/Gcd7 and unphosphorylated eIF2� stabilizes the eIF2-eIF2Bholocomplex and thereby promotes GEF activity in vivo.

MATERIALS AND METHODS

Yeast strains and plasmids. The yeast strains, plasmids, and primers employedin this study are listed in Tables 1, 2, and 3, respectively. Yeast strains wereconstructed using standard yeast transformation (21, 27), plasmid shuffling (4),and degron fusion (14) techniques. Media were prepared essentially as describedpreviously (43).

Strain yKD1-5 was derived from H1894 by integrating pAJ2 (28) (linearizedwith PmeI) at the UBRI locus and selecting transformants on synthetic completemedium (SC) lacking tryptophan. Correct pAJ2 integration was verified by PCRamplification of the UBR1 locus with primers UBRI-UPS and UBRI-UPAS forthe 5� junction between the integrated plasmid and flanking chromosomal se-quences and with primers UBRI-DS and UBRI-DAS for the corresponding 3�junction. Expression of Myc-tagged Ubr1 from the pGAL1 promoter was con-firmed by Western analysis of whole-cell extracts (WCEs) from cells grown withgalactose or raffinose as a carbon source, using anti-Myc antibodies (Roche).The degron plasmid pKD2 (gcd2-td) was derived from pPW66R (14) by replacingthe HindIII-XhoI fragment containing bp 1 to 284 of the CDC28 ORF with theHindIII-XhoI fragment containing bp 1 to 503 of the GCD2 ORF amplified byPCR using GCD2 plasmid p3311 as the template and primers KD2-F andKD2-R. Degron plasmids pKD6 (gcd6-td) and pKD7 (gcd7-td) were constructedsimilarly, using HindIII-XhoI fragments containing bp 1 to 430 of the GCD6ORF and bp 1 to 676 of the GCD7 ORF, respectively, which were PCR amplifiedusing the following plasmids and primers: GCD6, p3150 and primers KD6-F andKD6-R; GCD7, p1558 and primers KD7-F and KD7-R. Degron plasmids wereverified by restriction enzyme digestion and DNA sequencing. Yeast degronmutants yKD2 (gcd2-td), yKD6 (gcd6-td), and yKD7 (gcd7-td) were generatedfrom yKD1-5 by integrating pKD2 (linearized with NruI) at the GCD2 ORF,pKD6 (linearized with BglII) at the GCD6 ORF, and pKD7 (linearized withNruI) at the GCD7 ORF, respectively. Integration of each degron allele wasverified by PCR analysis of genomic DNA and by complementation of thelethality of the degron (td) mutant under nonpermissive growth conditions(SCGal lacking Ura and 36°C) with a plasmid containing the correspondingwild-type allele. The following pairs of primers were used for PCR verification ofthe 5� and 3� junctions between the degron (td) alleles and the flanking chro-mosomal sequences: (i) gcd2-td, td-F and GCD2-R; (ii) gcd6-td, td-F andGCD6-R; (iii) gcd7-td, td-F and GCD7-R. The following plasmids were used forthe complementation tests: p3311 (GCD2 LEU2), p3150 (GCD6 LEU2), p1156(GCD6 LEU2 containing the minimal GCD6 complementing region), p1558(GCD7 LEU2), and p702 (empty LEU2 vector).

To construct plasmid p101 (GCD7-3HA LEU2), plasmid p21-1 (GCD7-SmaI-stop LEU2) was created first by introducing the SmaI site before the stop codon(CCCGGGTGA) in p1558 (GCD7 LEU2) as follows. DNA fragments wereamplified by PCR from p1558 with primers Oligo2 and Oligo3 and primersOligo1 and Oligo4, respectively (Table 3). The PCR products were joined byfusion PCR using Oligo3 and Oligo4, and the resulting fragment was digestedwith NruI and ApaI and cloned into p1558. The insertion of SmaI was confirmedby SmaI restriction digestion and DNA sequencing. To generate p101, a cassettecontaining 3HA was amplified from p3210/pFA6a-3HA-TRP1 using primersHA-F and HA-R, digested with SmaI and ApaI, and cloned into p21-1 digestedwith SmaI and ApaI. Plasmid p101 was indistinguishable from p1558, containinguntagged GCD7, in complementing the Slg� phenotype of gcd7-201 in H1724and in maintaining the 3-ATS and 3ATR phenotypes of H2218 and H2217,respectively, on 3-aminotriazole (3-AT)-containing medium.

Site-directed mutations were introduced into GCD7-3HA on p101 by PCRfusion using the mutagenic primers listed in Table 3. For example, to constructpM3, a fragment was amplified from p101 with primers M1-P1 and Oligo4-SmaIand primers M1-P2 and Oligo3-SacI (Table 3). The PCR products were joined byfusion PCR, digested with SacI and SmaI, and cloned into p101. A comparablestrategy was employed to construct pM1, pM13, and pM22. To generate randommutants of GCD7-3HA, p101 was mutagenized in XL-1 Red bacteria (Invitro-gen) according to the manufacturer’s instructions. A pool of mutagenized plas-mids was transformed into gcn2� gcd7�::hisG mutant yeast strain H2218, andUra� Leu� transformants were replica plated on SC lacking Leu and containing5-fluoroorotic acid (5-FOA) to evict the resident GCD7 URA3 plasmid (p1151).The resulting Ura� Leu� segregants were then screened for growth on SClacking Leu and containing 30 mM 3-AT. Plasmids were rescued from 3ATR

(Gcd�) strains in bacterial strain DH10B (Invitrogen), reintroduced into H2218to identify those conferring 3ATR/Gcd� phenotypes (established for pM109,pM51, pM38, pM103, and pM106), and sequenced using the appropriate GCD7-specific primers.

pKD3153 (SUI3-FLAG URA3) was constructed by cloning the SacI-HindIIIfragment containing SUI3-FLAG from p3153 between the SacI and HindIII sitesof p1989.

TRP1 SUI2 plasmids pKD81, pKD84, pKD88, and pKD1097 were constructedby cloning 2.8-kb fragments PCR amplified from pY20 (SUI2-Y81S), p1242(SUI2-L84F), R28 (SUI2-R88T), and p1097 (SUI2), respectively, into BamHI-digested p1990.

For testing genetic interactions between GCD7 and SUI2 alleles, sui2� mutantstrain H2507 (containing SUI2 URA3 plasmid p919) was transformed with the

VOL. 30, 2010 �/Gcd7 IS CRITICAL FOR eIF2 BINDING 5219

on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

TRPI SUI2 plasmids just described, and p919 was evicted on 5-FOA-containingmedium to generate strains yKD1097 (SUI2), yKD84 (SUI2-L84F), yKD81(SUI2-Y81S), and yKD88 (SUI2-R88T). These strains were transformed withGCD7 URA3 plasmid p1151 and subsequently with a PCR fragment (amplifiedwith primers GCD7-KanF and GCD7-KanR by using p3203) containing thekanMX6 cassette flanked by sequences derived from the regions upstream anddownstream of the GCD7 ORF, selecting on SC lacking Ura and containingG418. Deletion of chromosomal GCD7 in the resulting transformants was in-ferred by their inability to grow on 5-FOA-containing medium and confirmed byPCR using primers specific for the gcd7�::KanMX6 allele. The resulting strains,yKD1097-1 (SUI2), yKD81-1 (SUI2-Y81S), yKD84-1 (SUI2-L84F), and yKD88-1(SUI2-R88T), were transformed with the appropriate LEU2 GCD7-HA plasmids,and p1151 was evicted on 5-FOA-containing medium to generate the strainsanalyzed (see Table 5).

High-copy-number plasmids p3000-1, p1780-1, and p2996-1 were constructedfrom p3000, p1780, and p2996, respectively, using the marker swap plasmidp4228 to convert URA3 to LEU2 (9). Transformation of yeast strains yKD2,yKD6, yKD7, J294, and J295 with pKD3153 (SUI3-FLAG URA3) generatedyeast strains yKD2-1, yKD6-1, yKD7-1, J294-1, and J295-1, respectively (Table1). Transformants were verified by Western blot analysis of WCEs usingFLAG-M2 antibodies (Sigma).

Antibodies. Glutathione-Sepharose beads were obtained from Santa CruzBiotechnology, and antihemagglutinin (anti-HA) beads and anti-HA antibodieswere obtained from GE Healthcare. Gcd11 antibodies were kindly provided by

Ernest Hannig. Rabbit polyclonal antibodies against the other eIF2 or eIF2Bsubunits were described previously (5–7).

Polysome analysis of degron strains. Strains were grown in SC containing 2%raffinose and 100 M CuSO4 at 25°C (permissive conditions) to an A600 of 1.0,followed by addition of galactose to 2% and incubation for 30 min to inducePGAL-UBR1-myc. Cells were then shifted to prewarmed SCRaf/Gal containingbathocuproinedisulfonic acid (BCS) and grown for 16 h at 36°C. Cycloheximide(50 g/ml) was added 5 min prior to harvesting the cells. WCEs were preparedas described previously (19), and 20 U of A260 was separated on 5 to 45% sucrosegradients prepared as already described (28), by centrifugation in a Ti 41 rotorfor 2.5 h at 39,000 rpm. Fractions were collected with continuous scanning atA254 with an ISCO gradient fraction collector.

Western analysis and coimmunoprecipitation. WCEs for Western analysis ofdegron depletions (see Fig. 2) were prepared and analyzed as described previ-ously (41), using the enhanced chemiluminescence system with horseradish per-oxidase-conjugated anti-rabbit secondary antibodies from Amersham. Coimmu-noprecipitations using anti-GCD6, anti-GCD2, or anti-HA antibodies wereconducted as described previously (3, 8), except that 1 mg of WCE was usedinstead of ribosomal salt wash fractions.

Coimmunoprecipitation analysis of TC. Coimmunoprecipitations of eIF2 andMet-tRNAi

Met were performed as described previously (44a). Briefly, WCEswere prepared from strains expressing FLAG-tagged eIF2� or eIF2� by glassbead lysis using buffer A containing 20 mM Tris-HCl (pH 7.5), 100 mM KCl, 5mM MgCl2, 0.1 mM EDTA, 7 mM �-mercaptoethanol, 5 mM NaF, 1 mM

TABLE 1. Saccharomyces cerevisiae strains used in this study

Strain Genotype Source orreference

H1894 MATa ura3-52 leu2-3,112 trp1-�63 gcn2� 30yKD1-5 MATa ura3-52 leu2-3,112 trp1-�63 gcn2� PGAL1-myc-UBR1::TRP1::ubr1 This studyyKD2 MATa ura3-52 leu2-3,112 trp1-�63 gcn2� PGAL1-myc-UBR1::TRP1::ubr1

PCUP1-UBI-R-DHFRts-HA-gcd2td::URA3::gcd2This study

yKD7 MATa ura3-52 leu2-3,112 trp1-�63 gcn2� PGAL1-myc-UBR1::TRP1::ubr1PCUP1-UBI-R-DHFRts-HA-gcd7td::URA3::gcd7

This study

yKD6 MATa ura3-52 leu2-3,112 trp1-�63 gcn2� PGAL1-myc-UBR1::TRP1::ubr1PCUP1-UBI-R-DHFRts-HA-gcd6td::URA3::gcd6

This study

H1608 MAT� ura3-52 ino1 leu2-3,112 GCN2-513 HIS4-lacZ ura3-52� 40H1402 MAT� leu2-3,112 ura3-52 ino1 GCN2 HIS4-lacZ ura3-52� 24H2218 MATa ura3-52 leu2-3,112 trp1-63 gcd7::hisG gcn2� GCN4-lacZ TRP1� p1151 (GCD7 URA3) 46H2217 MATa ura3-52 leu2-3,112 trp1-63 gcd7::hisG GCN4-lacZ TRP1� p1151 (GCD7 URA3) 46yKD2-1 MATa ura3-52 leu2-3,112 trp1-�63 gcn2� PGAL1-myc-UBR1::TRP1::ubr1

PCUP1-UBI-R-DHFRts-HA-gcd2td::URA3::gcd2 p3153 (FLAG-SUI3 LEU2)This study

yKD7-1 MATa ura3-52 leu2-3,112 trp1-�63 gcn2� PGAL1-myc-UBR1::TRP1::ubr1PCUP1-UBI-R-DHFRts-HA-gcd7td::URA3::gcd7 p3153 (FLAG-SUI3 LEU2)

This study

yKD6-1 MATa ura3-52 leu2-3,112 trp1-�63 gcn2� PGAL1-myc-UBR1::TRP1::ubr1PCUP1-UBI-R-DHFRts-HA-gcd6td::URA3::gcd6 p3153 (FLAG-SUI3 LEU2)

This study

yKD1 MATa ura3-52 leu2-3,112 trp1-63 gcd7::hisG gcn2� GCN4-lacZ TRP1� p101 (GCD7-3HA LEU2) This studyyM1 MATa ura3-52 leu2-3,112 trp1-63 gcd7::hisG gcn2� GCN4-lacZ TRP1� p1151 (GCD7 URA3) pM1

(GCD7-3HA-R151A LEU2)This study

yM13 MATa ura3-52 leu2-3,112 trp1-63 gcd7::hisG gcn2� GCN4-lacZ TRP1� p1151 (GCD7 URA3) pM13(GCD7-3HA-T262A K263A LEU2)

This study

yM22 MATa ura3-52 leu2-3,112 trp1-63 gcd7::hisG gcn2� GCN4-lacZ TRP1� p1151 (GCD7 URA3) pM22(GCD7-3HA-P358A S359A F360A LEU2)

This study

yM3 MATa ura3-52 leu2-3,112 trp1-63 gcd7::hisG gcn2� GCN4-lacZ TRP1)� pM3 (GCD7-3HA-E164A I165A LEU2) This studyJ294 MATa leu2� ura3� met15� gcn2�::loxp his3� gcd11�::kanMX pep4�::kanMX pC2887 (GCD11 LEU2) 1J295 MATa leu2� ura3� met15� gcn2�::loxp his3� gcd11�::kanMX pep4�::kanMX pC2858 (GCD11-N135D LEU2) 1J294-1 MATa leu2� ura3� met15� gcn2�::loxp his3� gcd11�::kanMX pep4�::kanMX pC2887 (GCD11-LEU2) pKD3153

(SUI3-FLAG URA3)This study

J295-1 MATa leu2� ura3� met15� gcn2�::loxp his3� gcd11�::kanMX pep4�::kanMX pC2858 (GCD11-N135D LEU2)pKD3153 (SUI3-FLAG URA3)

This study

H2507 MATa ura3-52 leu2-3,112 trp1-�63 sui2� gcn2� p919 (SUI2 URA3) 35yKD1097 MATa ura3-52 leu2-3,112 trp1-�63 sui2� gcn2� pKD1097 (SUI2 TRP1) This studyyKD81 MATa ura3-52 leu2-3,112 trp1-�63 sui2� gcn2� pKD81 (SUI2-Y81S TRP1) This studyyKD84 MATa ura3-52 leu2-3,112 trp1-�63 sui2� gcn2� pKD84 (SUI2-L84F TRP1) This studyyKD88 MATa ura3-52 leu2-3,112 trp1-�63 sui2� gcn2� pKD88 (SUI2-R88T TRP1) This studyyKD1097-1 MATa ura3-52 leu2-3,112 trp1-�63 sui2� gcn2� gcd7::kanMX pKD1097 (SUI2 TRP1) p1151 (GCD7 URA3) This studyyKD81-1 MATa ura3-52 leu2-3,112 trp1-�63 sui2� gcn2� gcd7:: kanMX pKD81 (SUI2-Y81S TRP1) p1151 GCD7 URA3) This studyyKD84-1 MATa ura3-52 leu2-3,112 trp1-�63 sui2� gcn2� gcd7:: kanMX pKD84 (SUI2-L84F TRP1) p1151 GCD7 URA3) This studyyKD88-1 MATa ura3-52 leu2-3,112 trp1-�63 sui2� gcn2� gcd7:: kanMX pKD88 (SUI2-R88T TRP1) p1151 (GCD7 URA3) This study

5220 DEV ET AL. MOL. CELL. BIOL.

on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

phenylmethylsulfonyl fluoride, Complete protease inhibitors (Roche), and 1g/ml each pepstatin A, leupeptin, and aprotinin. The clarified WCEs (1 mg)were immunoprecipitated with anti-FLAG M2 agarose beads (Sigma), and theprecipitated proteins were resolved by sodium dodecyl sulfate-polyacrylamide gelelectrophoresis and subjected to immunoblot analysis with appropriate antibod-ies. For analysis of tRNA i

Met in TC, total RNA was extracted from the immunecomplexes and analyzed by Northern blotting as described previously (44a).

RESULTS

Codepletion of �/Gcd7 and �/Gcd2 reduces eIF2 associationwith eIF2B in vivo. To probe the essential functions of �/Gcd7and �/Gcd2, we constructed mutants harboring temperature-sensitive degron (td) alleles of GCD7 and GCD2 and com-pared their phenotypes to those of a mutant expressing de-gron-tagged ε/Gcd6, the eIF2B catalytic subunit. Each td alleleencodes ubiquitin and a thermolabile dihydrofolate reductasemoiety fused to the N terminus of the cognate protein, ex-pressed from a copper-dependent promoter, and integratedinto the chromosome in a manner that disrupts the residentwild-type (WT) allele. The td mutants also express ubiquitin

ligase Ubr1 from a galactose-inducible promoter. Shifting cellsfrom medium containing copper and raffinose at 25°C (permis-sive conditions) to one containing galactose but lacking copperat 36°C (nonpermissive conditions) represses new synthesisand triggers proteasomal degradation of the existing degron-tagged protein (14, 33).

All three degron mutants grow like the WT under permissiveconditions (Fig. 1A) but do not form colonies under nonper-missive conditions (Fig. 1B), and growth under the latter con-ditions could be rescued by the cognate WT alleles (data notshown). Western analysis of WCEs revealed that the degronproteins were expressed at levels similar to or exceeding thoseof WT proteins under permissive conditions (Fig. 2A to C, toppanels, cf. lanes 1 and 6) and were markedly depleted within 1to 2 h under nonpermissive conditions (Fig. 2A to C, cf. lanes1 and 2). With one exception, depletion of the degron had noeffects on the levels of other eIF2B subunits or the subunits ofeIF2 or eIF3 that we examined, even after 8 h under nonper-missive conditions (Fig. 2A to C, lanes 1 to 5). However,

TABLE 2. Plasmids used in this study

Name/previous name Parent vector Description and yeast genes Source orreference

pAJ2 pKL54 Derivative of pKL54 containing TRP1 in place of HIS3 and PGAL1-myc-UBR1 28pPW66R NA Integrating URA3 plasmid containing PCUP1-UBI-R-DHFRts-HA-cdc28td 14pKD2 pPW66R Integrating URA3 plasmid containing PCUP1-UBI-R-DHFRts-HA-gcd2td This studypKD7 pPW66R Integrating URA3 plasmid containing PCUP1-UBI-R-DHFRts-HA-gcd7td This studypKD6 pPW66R Integrating URA3 plasmid containing PCUP1-UBI-R-DHFRts-HA-gcd6td This studyp3311/pAV1002 GCD2 LEU2 CEN4 plasmid 38p3150/YCpGCD6-WT GCD6 LEU2 CEN6 plasmid 3p702/pRS315 LEU2 CEN6 vector 44p1156/pJB102 GCD6 LEU2 CEN6 plasmid with minimal GCD6 complementing region 5p3000-1 YEp24 SUI2 SUI3 GCD11 IMT4 LEU2 2m plasmid This studyp4228/pUL9 NA URA3 to LEU2 marker swap plasmid 9p2996-1 YEp24 IMT4 LEU2 2m plasmid This studyp1780-1 YEp24 SUI2 SUI3 GCD11 LEU2 2m plasmid This studyp1558/Jp154 pRS315 GCD7 LEU2 CEN6 plasmid 46p21-2 pRS315 GCD7-SmaI-stop LEU2 CEN6 plasmid This studyp101 pRS315 GCD7-3HA LEU2 CEN6 plasmid This studyp3210/pFA6a-3HA-TRP1 pFA6a 3HA-TRP1 tagging plasmid 34pM1 pRS315 GCD7-3HA-R151A LEU2 plasmid This studypM13 pRS315 GCD7-3HA-T262A K263A LEU2 plasmid This studypM22 pRS315 GCD7-3HA-P358A S359A F360A LEU2 plasmid This studypM3 pRS315 GCD7-3HA-E164A I165A LEU2 plasmid This studypM109 pRS315 GCD7-3HA-L214S LEU2 plasmid This studypM51 pRS315 GCD7-3HA-A295V LEU2 plasmid This studypM38 pRS315 GCD7-3HA-I346T LEU2 plasmid This studypM103 pRS315 GCD7-3HA-N368K LEU2 plasmid This studypM106 pRS315 GCD7-3HA-Q371stop LEU2 plasmid This studyp1990/YCplac22 NA TRP1 CEN4 vector 20p3153/YCpSUI3-FL YCplac111 SUI3-FLAG LEU2 CEN4 plasmid 3p1989/YCplac33 NA URA3 CEN4 vector 20pKD3153 YCplac33 SUI3-FLAG URA3 plasmid This studyp1378/pRS425 NA LEU2 2m vector 44p1379/pRS426 NA URA3 2m vector 44p790/pMF12 Ylp5 GCD2 URA3 plasmid 18p1151/pJB99 pRS316 GCD7 URA3 CEN6 plasmid 5p3000/p1780-IMT YEp24 SUI2 SUI3 GCD11 IMT4 URA3 plasmid 3p2996 YEp24 IMT4 URA3 plasmid 6p1780 YEp24 SUI2 SUI3 GCD11 URA3 2m plasmid 12pKD1097 YCplac22 SUI2 TRP1 CEN4 plasmid This studypKD84 YCplac22 SUI2-L84F TRP1 CEN4 plasmid This studypKD81 YCplac22 SUI2-Y81S TRP1 CEN4 plasmid This studypKD88 YCplac22 SUI2-R88T TRP1 CEN4 plasmid This study

a NA, not available.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

depletion of �/Gcd7 in the gcd7-td strain also evoked a strongdepletion of �/Gcd2 (Fig. 2B).

All three degron mutants display a pronounced depletion ofpolysomes and accumulation of 80S monosomes after 16 hunder nonpermissive conditions, with the polysome-to-mono-some (P/M) ratio decreasing from 1.0 in the WT to 0.07 to 0.09in the mutants (Fig. 1C to F). This phenotype indicates theselective impairment of translation initiation, involving “run-off” of elongating ribosomes, a decrease in new initiationevents, and accumulation of excess free subunits as 80S cou-ples. Thus, depletion of �/Gcd2 in the gcd2-td mutant, or thecodepletion of �/Gcd2 and �/Gcd7 that occurs in gcd7-td mu-tant cells, confers a translation initiation defect comparable tothat given by depleting the catalytic subunit ε/Gcd6. The rela-tively low P/M ratio observed for the WT strain in Fig. 1Creflects the fact that this culture was approaching stationaryphase after 16 h under nonpermissive conditions. Indeed, theP/M ratios were �70% higher when WT cells were cultured foronly 8 h under nonpermissive conditions or when growingexponentially under permissive conditions. In addition, therewere no significant differences in the P/M ratios between theWT and the degron mutants under permissive conditions (datanot shown).

Interestingly, we observed only moderate polysome runoff inall three mutants even after 8 h, and a fraction of polysomesstill remained after 12 h under nonpermissive conditions (datanot shown). We interpret these results to indicate that a small

amount of eIF2B is sufficient for significant levels of transla-tion, reflecting the catalytic nature of eIF2B function, and thatdegradation of the degron-tagged proteins occurs in more thanone phase. It appears that the majority of each degron proteinis degraded rapidly, within 1 h (Fig. 2A to C), but more than12 h is required to eliminate the residual fractions, which cansupport significant levels of translation initiation. In subse-quent experiments, therefore, we incubated the degron mu-tants under nonpermissive conditions for 16 h to thoroughlydeplete eIF2B subunits and achieve a more complete elimina-tion of polysomes. Importantly, the degron mutants do not loseviability and retain nearly WT plating efficiencies on permissivemedium after this extended period under nonpermissive con-ditions (data not shown). The small amount of polysomes stillevident even after 16 h of depletion (Fig. 1D to F) might reflectthe spontaneous, noncatalyzed nucleotide exchange that hasbeen observed previously for yeast eIF2-GDP (15).

We asked next whether depleting the degron proteins affectsthe integrity of eIF2B or its assembly into the native eIF2B-eIF2 holocomplex. As expected, all five eIF2B subunits and allthree eIF2 subunits were efficiently coimmunoprecipitatedfrom WCEs with antibodies against ε/Gcd6 (Fig. 3A, lane 6).Depleting �/Gcd2 alone in the gcd2-td strain had little effect onthe integrity of eIF2B-eIF2 (Fig. 3A, lane 7), indicating that�/Gcd2 primarily affects the function and not the integrity ofeIF2B or its ability to bind eIF2, at least in the presence of WTversions of all other eIF2B and eIF2 subunits.

TABLE 3. Primers used in this study

Name Nucleotide sequence (5�–3�)a

KD2-F................................................CGCAAGCTTCCATGAGCGAATCGGAAGCCAAKD2-R ...............................................GCGCTCGAGTTTGGACCCACGGGCATCAKD6-F................................................CGCAAGCTTCCATGGCTGGAAAAAAGGGACAAKD6-R ...............................................GCGCTCGAGCGATGTTAGTCAATACATCACCKD7-F................................................CGCAAGCTTCCATGTCCTCTCAAGCATTCACTKD7-R ...............................................GCGCTCGAGCATTCTTGGTGTTATTAGGGAAUBRI-UPS........................................GGTCACACAAATTACATAGAACATTCUBRI-UPAS.....................................GACTACCTATAGGGCGAATTGGUBRI-DS ..........................................CCACTAGTTCTAGAGCGGCCUBRI-DAS .......................................GGAATTTGATTGCCATCTGCtd-F ....................................................ACTTCTCCCAGAATACCCAGOligo1 ................................................GCATTTGGATAAAAATAAGGCGCCCGGGTGATGATGTGT CTTTTGTACATTACOligo2 ................................................CGCCTTATTTTTATCCAAATGCACATCAATTTGCOligo3 ................................................GATTACTGCTCGCGAACGTAGOligo4 ................................................GCTGGGTACCGGGCCCCCCCTCGAHA-F .................................................GCGCCCGGGTACCCATACGATGTTCCTGACTATGCGGGCHA-R.................................................CGCGGGCCCAGATCTATATTACCCTGTTATCCCTAGCGGM1-P1 ................................................GGGAGCTCTAGTATGAAAACCAAGACTGATTACGCTCAAGT AGCCATTCAGGGTATCM1-P2 ................................................GTAATCAGTCTTGGTTTTCATACTAGAGCTCCCM3-P1 ................................................GCCATTCAGGGTATCAAGGATCTTATAGATGCTGCT AAAAACATTGATGAAGGTATTCM3-P2 ................................................ATCTATAAGATCCTTGATACCCTGAATGGCM13-P1 ..............................................CCGTGTGGGTAAGGTTATTATCGGCGCTGCTGCTGTTTTTGTCAATGGGGGGACTATCTCGTCM13-P2 ..............................................GCCGATAATAACCTTACCCACACGGM22-P1 ..............................................CTACATTACAAACGTCGGTGGGTTCAATGCTGCTGCT ATATATCGTATTGCGTGGGATAATTACM22-P2 ..............................................ATTGAACCCACCGACGTTTGTAATGTAGOligo3-SacI .......................................GTTCAGGGAGCTCTAGTATGAAAACCAAGACOligo4-SmaI......................................GGGTACCCGGGCGCCTTATTTTTATCCAAATGCIMT4..................................................TCGGTTTCGATCCGAGGACATCAGGGTTATGAGCD7-KanF .....................................GCGGAACCAGTAAACACAACACGCCAACTTTCGAACTTTTGCCCAACATAAGCGGATCCCCGGGTTAA

TTAAGCD7-Kan-R....................................GTTTTGATCTACAAGATTTTTATTTATTTTCATGAGTAATGTACAAAAGACACATCAGAATTCGAG

CTCGTTTAAAC

a Restriction enzyme sites are in boldface, and alanine (GCT) substitutions are underlined. For primers used to delete chromosomal GCD7, the GCD7 sequencesupstream of the start codon and downstream of the stop codon are both in boldface and underlined.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

Interestingly, the codepletion of �/Gcd2 and �/Gcd7 occur-ring in the gcd7-td strain does not affect complex formation bythe remaining three eIF2B subunits (Fig. 3A, lane 8). Theseresults are consistent with the previous identification of anε/Gcd6-�/Gcd1 subcomplex in cells overexpressing only thesetwo subunits (37) and further suggest that �/Gcn3 can interactwith this subcomplex in the absence of the other two eIF2Bregulatory subunits. However, the codepletion of �/Gcd2 and�/Gcd7 in gcd7-td mutant cells substantially lowers theamounts of all three eIF2 subunits coimmunoprecipitating withε/Gcd6 (Fig. 3A, lane 8), indicating a weaker association ofeIF2 with the residual ε/Gcd6-�/Gcd1-�/Gcn3 subcomplexcompared to eIF2 interaction with intact eIF2B. This last find-ing, combined with the fact that depleting �/Gcd2 alone doesnot reduce eIF2 association with eIF2B (Fig. 3A, lane 7),suggests that �/Gcd7 might provide an important contact foreIF2 within eIF2B.

We also examined the effect of depleting ε/Gcd6 on eIF2Bintegrity and eIF2B-eIF2 interaction by immunoprecipitating

the gcd6-td WCEs with antibodies against �/Gcd2. Depletingε/Gcd6 did not reduce the association of the remaining eIF2Bsubunits with one another (Fig. 3A, lane 9). However, deple-tion of ε/Gcd6 strongly reduced the association of eIF2 withthe remaining eIF2B subunits (Fig. 3A, lane 9). Similar resultswere obtained in two replicate experiments (data not shown).Considering the recoveries of eIF2� and eIF2�, it appears thatdepleting ε/Gcd6 has a greater effect than codepleting �/Gcd2and �/Gcd7 on eIF2 binding to eIF2B, whereas the results foreIF2� suggested more equivalent effects of gcd6-td and gcd7-tdon eIF2 association with eIF2B (Fig. 3A, cf. lanes 8 and 9). It

FIG. 1. Degron alleles of GCD2, GCD7, and GCD6 confer invia-bility and reduced translation initiation under nonpermissive condi-tions. The WT (yKD1) and gcd2-td (yKD2), gcd6-td (yKD6), andgcd7-td (yKD7) degron mutants were streaked in parallel on the fol-lowing media and grown for 5 days: (A) SC containing 2% raffinoseand 100 M CuSO4 at 25°C (permissive conditions) and (B) SC con-taining 2% galactose, 2% raffinose, and the copper-chelating agentBCS and lacking CuSO4 at 36°C (nonpermissive conditions). (C to F)Polysome profiles of the WT and the indicated degron mutants grownfor 16 h under nonpermissive conditions.

FIG. 2. Depletion of Gcd7 in gcd7-td mutant cells codepletes Gcd2.(A to C) Strains of the indicated genotypes described in Fig. 1 werecultured under permissive conditions to an A600 of 1.5 and then shiftedto nonpermissive conditions for the indicated times. WCEs were sub-jected to Western analysis using antibodies against the indicated pro-teins, including eIF3g (Tif35), which was examined as a loading con-trol.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

is possible that the polyclonal antibodies against ε/Gcd6 or�/Gcd2 used to immunoprecipitate eIF2B-eIF2 from thegcd7-td and gcd6-td WCEs, respectively, have differential ef-fects on binding of the eIF2 subunits to the mutant eIF2Bcomplexes. For example, binding of ε/Gcd6 antibodies toε/Gcd6 could weaken contact between ε/Gcd6 and eIF2� in thegcd7-td complex lacking �/Gcd7 and �/Gcd2, whereas this in-teraction would be unaffected by binding of �/Gcd2 antibodiesto �/Gcd2 in the gcd6-td complex lacking ε/Gcd6.

We examined the effects of the degron mutations on eIF2B-

eIF2 interaction by an independent approach wherein FLAGantibodies were used to immunoprecipitate eIF2�, eIF2�, andthe eIF2B subunits with ectopically expressed FLAG-taggedeIF2� (eIF2�-FL) (encoded by SUI3-FLAG). As expected,eIF2� and eIF2� coimmunoprecipitated with eIF2�-FL to anextent unaffected by any of the degron mutations (Fig. 3B,lanes 6 to 9). Consistent with the results in Fig. 3A, depleting�/Gcd7 reduced the association of ε/Gcd6 with eIF2�-FL, anddepleting ε/Gcd6 reduced the association of both �/Gcd7 and�/Gcd2 with eIF2�-FL (Fig. 3B, cf. lanes 7 and 8 with lane 9).

FIG. 3. Codepleting Gcd2 and Gcd7 or depleting only Gcd6 weakens eIF2-eIF2B interactions and reduces TC levels in vivo. (A) The WT anddegron strains described in Fig. 1 were cultured under nonpermissive conditions for 16 h, and WCEs were immunoprecipitated with antibodiesagainst Gcd6 (lanes 6 to 8) or Gcd2 (lanes 5 and 9). Twenty percent of the input WCE (lanes 1 to 4) and the entire pellet fractions (lanes 5 to9) were subjected to Western analysis for the indicated proteins. (B) Transformants of the WT and degron mutant strains described aboveharboring SUI3-FLAG on plasmid p3153 (lanes 1 to 4 and 6 to 9) and WT SUI3 strain yKD1-5 containing an empty vector (lanes 5 and 10) werecultured as described for panel A, and WCEs were immunoprecipitated with anti-FLAG antibodies and analyzed as described above, except that10% of the input WCE was analyzed. Note that two exposures are shown for ε/Gcd6 in the pellet lanes, a short exposure (Sh. exp.) and a longerexposure; however, only the short exposure is shown for the input lanes. (C) Total RNA was extracted from input and pellet samples preparedas for panel B from the same four (WT or degron mutant) strains harboring SUI3-FLAG described in panel B (lanes 2 to 5 and 9 to 12).Comparable samples derived from transformants of strains J294 (GCD11) and J295 (GCD11-N135D) containing SUI3-FLAG on pKD3153 (lanes6 and 7 and lanes 13 and 14, respectively) and a transformant of J294 (GCD11) with the empty vector (lanes 1 and 8) were analyzed in parallelas controls. RNA was subjected to Northern analysis using a 32P-labeled probe for tRNA i

Met. (D) Northern signals from the pellet fractions in panelC were normalized for the input signals, and the resulting mean ratios and standard errors of the means calculated from four independentexperiments were expressed as a percentage of the WT value derived from lane 12 of panel C for the degron mutants and lane 13 of panel C forthe gcd11-N135D mutant. ��, P � 0.01 (Student’s t test).


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

Similar results were obtained in several replicate coimmuno-precipitation experiments (data not shown). These findingssupport the possibility that �/Gcd7-�/Gcd2 and ε/Gcd6 con-tribute independently to eIF2 binding by eIF2B. At odds withthe results in Fig. 3A, however, is the fact that depleting�/Gcd2 alone reduced the association of the other eIF2B sub-units with eIF2�-FL (Fig. 3B, cf. lane 6 with lane 9), suggestingthat �/Gcd2 also stabilizes eIF2 binding to eIF2B. To explainthe discrepancy between the results for the gcd2-td mutation,we suggest that the combination of the FLAG tag on eIF2�and the elimination of �/Gcd2 from eIF2B by gcd2-td producesa synthetic defect in eIF2-eIF2B association. In any event, theresults in Fig. 3A and B indicate clearly that �/Gcd7, alone ortogether with �/Gcd2, promotes the stable binding of eIF2 toeIF2B.

To determine whether weakening the association of eIF2with eIF2B reduces recycling of eIF2 and attendant TC assem-bly in the degron mutants, the immune complexes obtained byimmunoprecipitating eIF2�-FL were probed by Northernanalysis for tRNA i

Met. We verified that only a low backgroundlevel of tRNA i

Met was immunoprecipitated with anti-FLAGantibodies from the parental strain lacking eIF2�-FL (Fig. 3C,cf. lane 8 with lane 12) and also that the N135D substitution ineIF2� (encoded by gcd11-N135D) reduces the coimmunopre-cipitation of tRNA i

Met with eIF2�-FL by �75% (Fig. 3C, lanes13 and 14, and quantification in panel D). It was shown pre-viously that N135D impairs Met-tRNA i

Met binding to eIF2-GTP in vitro (1). Depletion of �/Gcd2, �/Gcd7, or ε/Gcd6conferred significant reductions in tRNA i

Met association witheIF2 (Fig. 3C, lanes 9 to 12, and D); however, it was surprisingthat depleting ε/Gcd6 produced only an �25% reduction,whereas depleting �/Gcd7 evoked a larger decrease of �40%in tRNA i

Met association with eIF2�-FL. (Highly similar resultswere obtained in replicate experiments when the WT cultureswere harvested at lower densities, and still dividing exponen-tially, after 16 h under nonpermissive conditions [data notshown].)

The fact that depleting ε/Gcd6 produced only a moderatereduction in tRNA i

Met association with eIF2 could be inter-preted to indicate that the four-subunit eIF2B complex lackingε/Gcd6 can catalyze GDP-GTP exchange. While this possibilitycannot be completely excluded, it seems improbable in view ofprevious observations that eIF2B complexes containing mutantε/Gcd6 (36) or lacking this subunit altogether (37) exhibit,respectively, greatly reduced or no detectable GEF activity. Itseems more likely, therefore, that the relatively high level ofTC remaining in the gcd6-td mutant reflects incomplete deple-tion of ε/Gcd6 combined with noncatalyzed nucleotide ex-change by eIF2-GDP. As discussed below, it is also conceivablethat a fraction of the eIF2-tRNA i

Met complexes detected bycoimmunoprecipitation contain GDP or no guanine nucleotideand, hence, are not authentic TCs. Our finding that the gcd7-tdmutant seems to display a greater reduction in tRNA i

Met asso-ciation with eIF2�-FL than does the gcd6-td mutant (Fig. 3D)is consistent with genetic results described next suggesting thatcodepleting �/Gcd7-�/Gcd2 in gcd7-td mutant cells might im-pair both nucleotide exchange and Met-tRNA i

Met binding toeIF2-GTP.

Depletion of �/Gcd2 is rescued specifically by tRNA iMet over-

expression. It was shown previously by Hannig and colleagues

that simultaneously overexpressing all three eIF2 subunits andtRNA i

Met rescued an inviable mutant lacking the genes for allfive eIF2B subunits, but the cells grew poorly unless eIF2�contained a substitution (K250R) that elevates spontaneousGDP dissociation from eIF2 (16). It was suggested that eIF2Bfunction was bypassed under these conditions owing to theelevated spontaneous exchange activity of the mutant eIF2 (15,16), the fact that Met-tRNA i

Met binding stabilizes GTP bindingto eIF2 (29), and the increased concentrations of both eIF2and Met-tRNA i

Met in the mutant cells. Interestingly, thegrowth defect of our gcd6-td mutant was fully complementedby a high-copy-number plasmid encoding the three WT eIF2subunits and tRNA i

Met (hc TC) (Fig. 4A, gcd6-td). A high-copy-number plasmid encoding only eIF2 subunits (hc eIF2) alsorescued the gcd6-td mutant and partially complemented itsgrowth defect, whereas overexpressing tRNA i

Met alone (hc

FIG. 4. Overexpressing tRNA iMet, but not eIF2, specifically sup-

presses the lethality of Gcd2 depletion. Serial dilutions of transfor-mants of the indicated WT and degron mutant strains harboring high-copy-number plasmids for overexpressing TC (A), eIF2 (B), ortRNA i

Met (C) or the empty vector (VEC) were spotted in parallel topermissive (I) or nonpermissive (II) medium and incubated at 25°C for4 days (I) or 36°C for 6 days (II).


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

IMT) did not suppress the lethality of ε/Gcd6 depletion (Fig.4B and C, gcd6-td).

These last results differ from those described previously inthat Hannig et al. found that overexpressing eIF2 or tRNA i

Met

alone did not rescue the deletion mutant lacking all five eIF2Bsubunits, even when eIF2� contained a substitution that en-hanced noncatalyzed, spontaneous exchange (16). One possi-bility, alluded to above, is that a low level of ε/Gcd6, and henceGEF activity, remains in gcd6-td cells under nonpermissiveconditions, which augments the spontaneous nucleotide ex-change by overexpressed eIF2 to achieve a rate of eIF2-GTPformation and TC assembly that is high enough to permit cellgrowth without the need for overexpressed tRNA i

Met.As expected, the hc TC plasmid also fully suppressed the

growth defect of the gcd2-td mutant. Interestingly, however, hcIMT also strongly suppressed the growth defect of this mutant,whereas hc eIF2 had no effect (Fig. 4A to C, gcd2-td). Incontrast, overexpressing tRNAHis from a high-copy-numberplasmid did not improve the growth of gcd2-td cells (data notshown), suggesting that the suppression is specific for overex-pression of tRNA i

Met. Previous work using the tRNAHis plas-mid confirmed that it confers tRNAHis overexpression in vivo(47). In contrast to both the gcd6-td and gcd2-td mutant strains,the growth defect of the gcd7-td mutant was suppressed byhcTc but not by either hc eIF2 or hc IMT (Fig. 4A to C,gcd2-td).

One way to explain the observation that hc eIF2 suppressesthe growth defect produced by depletion of the catalytic sub-unit (ε/Gcd6), but not that associated with depletion of �/Gcd2,is to propose that reduced nucleotide exchange is not therate-limiting defect in the gcd2-td mutant. The finding that hcIMT rescues the growth of gcd2-td but not gcd6-td cells mightthen indicate that depleting �/Gcd2 impairs a second functionof eIF2B in loading Met-tRNA i

Met on eIF2-GTP, which can berescued only by overexpressing tRNA i

Met, and that ε/Gcd6 doesnot contribute to this second eIF2B function. In this view,codepleting �/Gcd7 and �/Gcd2 in the gcd7-td mutant wouldimpair both nucleotide exchange (owing to reduced eIF2 bind-

ing to eIF2B) and Met-tRNA iMet loading (due to the absence

of �/Gcd2). Hence, rescuing the gcd7-td mutant would requireoverexpressing eIF2 to boost eIF2-GTP levels through spon-taneous nucleotide exchange and also tRNA i

Met overproduc-tion to restore Met-tRNA i

Met loading on eIF2-GTP. This hy-pothesis and an alternative explanation for the genetic findingsin Fig. 4 are discussed further below.

�/Gcd7 is a common constituent of all eIF2B subcomplexesthat suppress the toxicity of eIF2(�P) in vivo. We showedpreviously that the Slg� phenotype conferred by the constitu-tively activated form of Gcn2 encoded by the GCN2c-M719VE1537G allele (henceforth GCN2c) was partially suppressed byco-overexpressing the regulatory subunits �/Gcd2, �/Gcd7, and�/Gcn3. The overexpressed subunits form a stable subcomplexin vivo (49) that was shown to bind specifically to phosphory-lated eIF2 in vitro (37). These facts, combined with evidencethat tight binding of eIF2�-P to this subcomplex is required forthe inhibition of eIF2B by eIF2(�P) (32), provided strongevidence that the overexpressed subunits sequester eIF2(�P)and neutralize its ability to inhibit the resident five-subuniteIF2B in yeast cells. The binary combination �/Gcd2-�/Gcd7also suppresses GCN2c, most likely by titrating �/Gcn3 fromeIF2B to form the same trimeric regulatory subcomplex (49).In view of our findings above implicating �/Gcd7 in eIF2 bind-ing by eIF2B and the fact that depleting �/Gcd2 in gcd2-td cellsdoes not reduce eIF2 association with eIF2B (Fig. 3A), wewondered whether other combinations of eIF2B subunits in-volving �/Gcd7 but lacking �/Gcd2 could also suppress thetoxicity of eIF2� hyperphosphorylation engendered by GCN2c

on cell growth.Indeed, co-overexpressing �/Gcd7, �/Gcd1, and �/Gcn3, or

just �/Gcd7 together with �/Gcd1, suppressed the Slg� phe-notype of GCN2c cells to the same extent as that observed forthe combination of all four essential eIF2B subunits (49) (Fig.5A and Table 4). Western analysis confirmed that the high-copy-number GCD1-GCD7-GCN3 plasmid conferred selectiveoverexpression of these three subunits (Fig. 5B). We showedpreviously that overexpressing �/Gcd1, �/Gcd7, or �/Gcn3 in-

FIG. 5. Co-overexpression of �/Gcd7, �/Gcd1, and �/Gcn3 suppresses the toxicity of constitutively activated Gcn2 in vivo. (A) Transformantsof GCN2c mutant strain H1608 or GCN2 mutant strain H1402 harboring high-copy-number plasmids with the indicated genes or empty vectors(V) were streaked in parallel on SD with minimal supplements and incubated at 30°C for 5 days. (B) Western analysis of WCEs of strains frompanel A with two different loadings (1 and 2 ) for each strain.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

dividually, or �/Gcd7 together with �/Gcn3, does not suppressthe Slg� phenotype of this GCN2c mutation (49), and wefound here that the binary combinations �/Gcd1-�/Gcn3,ε/Gcd6-�/Gcd7, and ε/Gcd6-�/Gcd2 likewise failed to do so(summarized in Table 4). Thus, �/Gcd7 is the only subunitpresent in all combinations of eIF2B subunits that suppress thetoxicity of eIF2 hyperphosphorylation, consistent with the ideathat �/Gcd7 interacts directly with eIF2 in the eIF2-eIF2Bholocomplex.

Lethal substitutions in �/Gcd7 reduce eIF2 association witheIF2B. The results presented above suggested that �/Gcd7 iscrucial for tight binding of eIF2 to eIF2B and attendant recy-cling of eIF2-GDP and TC assembly. To obtain further supportfor this conclusion, we set out to identify point mutations thatwould impair an essential binding surface for eIF2 in �/Gcd7.Previously, we modeled the tertiary structure of �/Gcd7 on thecrystal structure of archaeal aIF2B from Pyrococcus horikoshii,whose ortholog in Thermoplasma acidophilum exhibits bothsequence and functional similarities with eIF2B regulatorysubunits—including the ability to interact with yeast eIF2� andthe cognate archaeal aIF2� in vitro (11). This structural modelis reproduced in the left-hand panels of Fig. 6A, with thesequence conservation among eukaryotic eIF2B� homologsprojected on the surface residues. As noted previously, oneface of eIF2B� is predicted to be enriched in conserved resi-dues (shades of red in view I), whereas the opposite face ismuch less conserved (shades of blue in view II). Guided by thismodel, we mutagenized an HA-tagged GCD7 allele to gener-ate Ala substitutions in various residues conserved among eu-karyotic �/Gcd7 homologs, altering single residues or clustersof two or three contiguous residues at a time. The resultingGCD7-HA alleles were introduced into a gcd7� strain by plas-mid shuffling using counterselection on medium containing5-FOA to evict the resident GCD7� URA3 plasmid.

Three of the mutant GCD7-HA alleles were lethal, includinggcd7-358-360, which encodes Ala substitutions of a string ofconserved residues, P358, S359, and F360. (For simplicity, theHA tag will not be included in the designations of GCD7-HAalleles discussed below.) Note that P358 is exposed on theconserved face of the �/Gcd7 model (Fig. 6A, view I, right

side). The other lethal substitutions, R151A (gcd7-151),T262A, and K263A (gcd7-262-263), involve conserved residueson the opposite, less conserved face of the molecule, althoughR151 resides in a relatively more conserved patch on this otherface (Fig. 6A, view II, right side).

To determine the effects of these lethal substitutions on�/Gcd7 expression, we conducted Western analysis of WCEs

FIG. 6. Lethal substitutions in �/Gcd7 impair eIF2 interaction withintact eIF2B. (A, left panels) Model predicting the sequence conser-vation of surface-exposed residues among eIF2B� homologs, pro-jected on one protomer of the P. horikoshii aIF2B homodimer(PH0440) as described previously (11, 37). (Right panels) Predictedlocations of �/Gcd7 residues substituted with alanines (red, in paren-theses) and cognate aIF2B residues (black) for gcd7-151, gcd7-262-263,and gcd7-358-360. (B) Western analysis of transformants of gcn2�gcd7� mutant strain H2218 harboring a GCD7 URA3 plasmid andLEU2 plasmids containing the indicated mutant or WT GCD7-HAalleles, cultured in SC lacking Leu and Ura at 30°C. Increasingamounts of WCE (1 , 2 , and 3 ) were loaded in successive lanes,and blots were probed with anti-HA or anti-Gcd6 antibodies.(C) Strains from panel B containing GCD7-HA alleles were cultured asdescribed for panel B, and coimmunoprecipitation analysis was con-ducted as described for Fig. 3A, except using anti-HA antibodies.

TABLE 4. Effect of overexpressing eIF2B subunits on growth rateof yeast cells harboring an activated GCN2c-encoded kinasea

Plasmid(s) eIF2B subunit gene(s)Relativegrowthrateb

GCN2allele

pRS426 None 6� GCN2pRS426 None 1� GCN2C

p2300, p1874 GCD6, GCD2 1� GCN2C

p2335, p2305 GCD6, GCD7 1� GCN2C

p211, p2304 GCD1, GCN3 1� GCN2C

p211, p2298 GCD1, GCD7, GCN3 4� GCN2C

p211, p2305 GCD1, GCD7 4� GCN2C

p1873, p2298 GCD1, GCD6,GCD7, GCN3

4� GCN2C

a Strain H1608 containing GCN2C-M719V E1537G was transformed with high-copy-number plasmids containing the indicated eIF2B subunit gene(s) or withempty vector pRS426, and the relative growth rates of the transformants weredetermined by streaking as described in the legend to Fig. 8A.

b The relative growth rate of the isogenic WT GCN2 strain H1402 served as acontrol.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

from the viable parental strains prior to eviction of GCD7�,using HA antibodies to detect the mutant proteins. EssentiallyWT levels of expression were observed for all three mutantproteins (Fig. 6B). Furthermore, coimmunoprecipitation anal-ysis revealed that none of the lethal substitutions impaired theassociation of the mutant HA-�/Gcd7 proteins with othereIF2B subunits (Fig. 6C, lanes 5 to 8, eIF2B). Importantly,however, all three reduced the amount of coimmunoprecipi-tating eIF2 subunits (Fig. 6C, lanes 5 to 8, eIF2). These find-ings strongly support the idea that a critical function of �/Gcd7is to stabilize the association of eIF2 with eIF2B in the eIF2B-eIF2 holocomplex.

Genetic evidence that gcd7-164-165 impairs eIF2B by en-hancing nonproductive binding of unphosphorylated eIF2� tothe regulatory subcomplex. In addition to the lethal mutationsjust described, we used both site-directed and random mu-tagenesis to identify viable gcd7 alleles that confer Gcd�

phenotypes, indicating reduced eIF2B function in vivo (26).Gcd� phenotypes reflect constitutive derepression of GCN4translation resulting from decreased TC assembly that occursindependently of eIF2� phosphorylation by Gcn2. Whereasgcn2� cells are sensitive to the inhibitor of histidine biosynthe-sis 3-AT, Gcd� mutations suppress this phenotype by dere-pressing GCN4 translation (26). As shown in Table 5 (columnSUI2), we identified six gcd7 alleles allowing growth on 3-AT-containing medium in the gcn2� background. The gcd7-164-165 allele is noteworthy because it also confers a growth defectevident in nutrient-replete medium (SC) (Fig. 7B), suggestinga relatively stronger reduction in eIF2B function. Consistentwith this, the gcd7-164-165 mutant exhibits substantial poly-some runoff (Fig. 7C). Interestingly, however, coimmunopre-cipitation analysis revealed that the 164-165 mutation does notaffect the stability of the eIF2B-eIF2 holocomplex (Fig. 7D),indicating a defect in eIF2B function mechanistically distinctfrom that conferred by the lethal gcd7 mutations describedabove.

It has been proposed that eIF2� phosphorylation inhibitseIF2B because eIF2(�P)-GDP forms an excessively stable in-teraction with the regulatory eIF2B subcomplex that blocksproductive interaction of the catalytic subunit with the G do-main of eIF2� (32, 37). A key line of evidence for this modelis that mutations altering residues in the S1 domain of eIF2�were identified that suppress the inhibitory effect of eIF2(�P)on WT eIF2B in vivo and also weaken the interaction of

eIF2�-P with the eIF2B regulatory subcomplex in vitro (13, 32,45). Such mutations eliminate the inhibition of general trans-lation and impair the derepression of GCN4 translationevoked by activation of Gcn2, conferring the same (Gcn�)phenotype as does elimination of Gcn2 itself or the Ala sub-stitution of Ser-51 in eIF2�. Interestingly, a number of Gcd�

mutations in GCN3, dubbed gcn3c alleles (24), are suppressedby one such Gcn� substitution in eIF2�, L84F (46). This sup-pression strongly suggests that these gcn3c substitutions mimicthe effect of eIF2� phosphorylation and confer tighter bindingof unphosphorylated eIF2 to the eIF2B regulatory subcomplexas the means of inhibiting nucleotide exchange.

Having found that gcd7-164-165 confers a Gcd� phenotypewithout weakening eIF2-eIF2B interaction in vivo (Fig. 7D),we wondered if this mutation might impair eIF2B function bythe same mechanism elucidated for gcn3c mutations. Consis-tent with this possibility, the L84F substitution in eIF2� com-pletely suppresses the Gcd�/3-ATR and Slg� phenotypes ofgcd7-164-165 (Fig. 8A, cf. 3-AT WT with L84F, and B, cf.sector 4 with 2; Table 5, column sui2-L84F). Interestingly, theGcd�/3-ATR phenotypes conferred by gcd7-368 and gcd7-Q371 were also diminished by sui2-L84F (Table 5). Theseresults suggest that three of the six gcd7 Gcd� mutations in-hibit eIF2B, at least partly, by mimicking eIF2� phosphoryla-tion to enhance a nonproductive interaction of unphosphoryl-ated eIF2 with the eIF2B regulatory subcomplex.

Allele-specific interactions between gcd7 and sui2 mutationsprovide further evidence of direct binding of �/Gcd7 to eIF2�.We proceeded next to test the gcd7-164-165 allele for suppres-sion by two other sui2 alleles that resemble sui2-L84F in elim-inating the toxicity of eIF2(�P) and reducing the binding ofeIF2�-P to WT eIF2B (13, 32). Interestingly, these allelesdisplayed a completely different genetic interaction with gcd7-164-165, exacerbating its Slg� phenotype in the case of sui2-Y81S and conferring synthetic lethality with sui2-R88T (Fig.8A, 5FOA, and B, cf. sectors 3 and 5 with 2; Table 5, last twocolumns). Remarkably, the synthetic lethal phenotype of thegcd7-164-165 sui2-R88T combination is accompanied by amarked reduction in eIF2 association with the eIF2B complexcontaining the mutant HA-�/Gcd7 protein (Fig. 8C). Impor-tantly, like gcd7-164-165, the sui2-R88T mutation alone haslittle effect on the amount of eIF2 that coimmunoprecipitateswith WT HA-�/Gcd7 (data not shown), as expected from thefact that it confers no growth defect on its own (Table 5). Thus,

TABLE 5. Allele-specific genetic interactions between GCD7 and SUI2 allelesa

GCD7-HAallele Mutation(s)

Growth

SUI2 sui2-L84F sui2-Y81S sui2-R88T

SC 3-AT SC 3-AT SC 3-AT SC 3-AT

WT None �� 164-165 E164A, I165A � �� Lethal Lethal Lethal Lethal368 N368K �� Lethal Lethal � �371 Q371stop �� 214 L214S �� 295 A295V �� 346 I346T ��

a The analysis was conducted as described in the legend to Fig. 8A and B for all of the mutant combinations listed. The degree of growth is summarized qualitativelyby the number of plus signs (�� for the WT) or by a minus or a � sign for no detectable or trace growth, respectively.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

gcd7-164-165 and sui2-R88T produce a synthetic reduction ineIF2 binding to eIF2B that parallels their synthetic lethal phe-notype. Below, we present a detailed model that can accountfor the opposite effects on eIF2B function of combining gcd7-164-165 with sui2-R88T versus sui2-L84F.

As shown in Table 5, gcd7-368 and gcd7-371 resemble gcd7-164-165 in that their Gcd� phenotypes are suppressed by sui2-

L84F, but they are synthetically sick in combination with sui2-R88T. The gcd7-368 mutation further resembles gcd7-164-165in being synthetically lethal with sui2-Y81S, whereas gcd7-371is uniquely suppressed by sui2-Y81S. The remaining three al-leles, gcd7-214, gcd7-295, and gcd7-346, did not interact genet-ically with any of the Gcn� sui2 alleles (Table 5). The highdegree of allele specificity evident in Table 5 supports the ideathat �/Gcd7 and eIF2� interact directly in the eIF2B-eIF2complex in a manner that dictates the efficiency and regulationof nucleotide exchange.

DISCUSSION

�/Gcd7 is crucial for eIF2 binding by native eIF2B. Therewas previous evidence that �/Gcd7 contributes to the tightbinding of eIF2(�P) to the eIF2B regulatory subcomplex thatmediates the negative regulation of eIF2B GEF activity. TwoGcn� substitutions in �/Gcd7 were shown to both weakeneIF2B binding to eIF2(�P) and allow nucleotide exchange oneIF2(�P)-GDP in vitro (32, 37). These �/Gcd7 substitutionsalso reduced eIF2B association with unphosphorylated eIF2 invitro, suggesting a positive role for �/Gcd7 in nucleotide ex-change in addition to downregulating the reaction for thephosphorylated substrate (32). However, there was no indica-tion that these Gcn� substitutions in �/Gcd7 impair eIF2Bfunction in vivo (38). Furthermore, the Gcd� substitution in�/Gcd7, V241D, was shown to provoke dissociation of theeIF2B regulatory subcomplex from eIF2B without reducingeIF2 interaction with the �/Gcd1-ε/Gcd6 catalytic subcomplexof eIF2B (42), suggesting that the inhibition of eIF2B functionby V241D does not involve reduced substrate binding. Indeed,ε/Gcd6 has multiple determinants for binding of the � and �subunits of eIF2 (36), making it unclear whether the regulatorysubunits would be critical for the binding of unphosphorylatedeIF2-GDP.

This study presents several lines of evidence that �/Gcd7does play a critical role in the binding of unphosphorylatedeIF2 by eIF2B. First, codepletion of �/Gcd7 and �/Gcd2 in agcd7-td degron mutant confers a marked reduction in eIF2association with the residual �/Gcn3-ε/Gcd6-�/Gcd1 eIF2Bsubcomplex. As depletion of �/Gcd2 alone in a gcd2-td degronmutant did not affect eIF2 association, it appeared that �/Gcd7promotes eIF2 binding to eIF2B. It was conceivable that elim-inating �/Gcd7 would impair eIF2 binding only when �/Gcd2 isalso missing from the complex, as occurs in the gcd7-td mutant.At odds with this possibility, we identified three lethal gcd7substitutions that reduce eIF2 association with eIF2B withoutaffecting the abundance of �/Gcd2 (or any other subunit) in theeIF2B complex. The same biochemical phenotype was ob-served for a synthetic lethal combination of mutations in�/Gcd7 (gcd7-164-165) and eIF2� (sui2-R88T). Although nei-ther gcd7-164-165 nor sui2-R88T impairs eIF2-eIF2B associa-tion on its own, the synthetic lethality of combining thesemutations correlates with loss of stable binding of eIF2 har-boring eIF2�-R88T to the intact eIF2B complex containing thegcd7 E164A and I165A substitutions. Together, these findingsprovide strong evidence that enhancing the rate or stability ofeIF2 binding to eIF2B is an essential function of �/Gcd7.

Previous work showed that sui2-R88T suppresses the inhib-itory effect of phosphorylated eIF2 and reduces the binding of

FIG. 7. gcd7-164-165 confers Slg� and Gcd� phenotypes and im-pairs translation initiation without reducing eIF2-eIF2B association.(A) Surface projection of sequence conservation and predicted loca-tions of �/Gcd7 residues substituted with alanines (orange) and thecognate aIF2B residues (black) for gcd7-164-165, generated as de-scribed for Fig. 6A. (B) Serial dilutions of derivatives of gcn2� gcd7�mutant strain H2218 containing GCD7-HA or gcd7-HA-164-165 as theonly GCD7 allele and a transformant of the isogenic GCN2 strain(H2217) containing WT GCD7-HA were spotted in parallel on SClacking Leu or SC containing 30 mM 3-AT and incubated for 4 days at30°C. (C) Polysome profiles of the gcn2� mutant strains from panel Bgrown in SC lacking Leu at 30°C and analyzed as in Fig. 1C.(D) GCD7-HA and gcd7-HA-164-165 mutant strains from panel B andGCD7 strain H2218 were subjected to coimmunoprecipitation analysisas described for Fig. 3A, except using anti-HA antibodies.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

phosphorylated eIF2� to the eIF2B regulatory subcomplex(32). These findings led to the model in which the eIF2�-R88Tsubstitution weakens the nonproductive mode of binding byeIF2�-P to the eIF2B regulatory subcomplex, which in turnrestores proper interaction of the eIF2B catalytic subunit witheIF2� to enable nucleotide exchange on eIF2(�P)-GDP(model in Fig. 9A). This proposal is consistent with the factthat R88 is close to surface-exposed residues of the eIF2� S1domain that bind directly to the C lobe of eIF2� kinase PKR(10) and are thought to interact with the eIF2B regulatorysubcomplex (13).

To account for our identification of three lethal substitutionsin �/Gcd7 that reduce eIF2 binding to eIF2B, we suggest thatthey impair a distinct and productive mode of eIF2� binding tothe eIF2B regulatory subcomplex. In this view, eIF2� is alwaysengaged with the eIF2B regulatory complex but “toggles” be-tween different interaction surfaces. The nonproductive modeof binding stabilized by phosphorylated Ser-51 is incompatiblewith proper interaction of the catalytic subunit ε/Gcd6 witheIF2� (Fig. 9B). The productive mode of binding is favored byunphosphorylated eIF2� and is conducive to ε/Gcd6-eIF2�interaction and nucleotide exchange (Fig. 9A). In this view, thelethal gcd7 mutations impair the latter, productive mode ofeIF2� binding, explaining their deleterious effects on eIF2Bfunction (Fig. 9C).

This model also provides a framework to explain the syn-thetic lethal effect of combining sui2-R88T with gcd7-164-165.It is thought that eIF2�-R88T destabilizes the nonproductivemode of eIF2� binding as the means of eliminating the inhi-bition of eIF2B by phosphorylated eIF2� (32) (Fig. 9, cf. pan-els E and B). We propose that R88T also moderately impairsthe productive mode of binding by unphosphorylated eIF2� tothe eIF2B regulatory subcomplex. By itself, R88T does notperceptibly reduce eIF2B function, but it exacerbates a similardefect conferred by the gcd-164-165 mutation to produce asynthetic reduction in eIF2 binding to eIF2B in the double

mutant (Fig. 9, cf. panels I and D). The same mechanismwould apply to the sui2-Y81S mutation, which is also Gcn� onits own and synthetically lethal with gcd7-164-165. In contrast,the sui2-L84F mutation would destabilize only the nonproduc-tive mode of eIF2� binding, explaining its Gcn� phenotype(45) (Fig. 9F and G), without exacerbating the Slg� phenotypeof gcd7-164-165 (Fig. 9J). The notion that sui2-L84F exclu-sively destabilizes nonproductive binding by eIF2�-P fits wellwith the fact that it was found only to suppress, and neverexacerbate, growth defects when combined with differentGcd� alleles of GCD7 (Table 5) or GCN3 (46).

Our finding that the Gcd� and Slg� phenotypes due togcd7-164-165 are fully suppressed, rather than exacerbated, bysui2-L84F also provides insight into the mechanism wherebygcd7-164-165 impairs the productive binding of unphosphoryl-ated eIF2. It was shown previously (46) that sui2-L84F cansuppress a subset of Gcd� gcn3c alleles that seem to mimiceIF2 phosphorylation and promote nonproductive binding ofunphosphorylated eIF2� to the eIF2B regulatory subcomplex.Hence, the suppression of gcd7-164-165 by sui2-L84F observedhere suggests that this �/Gcd7 substitution likewise strengthensnonproductive binding by unphosphorylated eIF2� as a meansof impairing GEF function. This deduction is consistent withour conclusion above that gcd7-164-165 decreases the produc-tive binding of eIF2�, as these two modes of binding to theregulatory subcomplex are expected to be mutually exclusive(Fig. 9A and B).

A second function for eIF2B in loading tRNA iMet on eIF2-

GTP? Previous work has shown that co-overexpressing eIF2and tRNA i

Met suppresses the lethality of eliminating all fouressential subunits of eIF2B, although the resulting strains growpoorly (16). It is thought that spontaneous nucleotide ex-change by eIF2, coupled with eIF2 overexpression, producessufficient eIF2-GTP to support a minimally adequate level ofTC assembly when this reaction is driven by mass action withoverproduced tRNA i

Met (16). Our finding that hc TC restores

FIG. 8. Allele-specific synthetic phenotypes of mutants combining Gcd� gcd7 alleles and Gcn� sui2 alleles. (A) gcn2� gcd7� sui2� mutantstrains containing a URA3 GCD7 plasmid and a TRP1 plasmid with SUI2 (WT), sui2-Y81S, sui2-L84F, or sui2-R88T were transformed with a LEU2gcd7-164-165A plasmid and replica plated in parallel to SC lacking Ura and Leu (SC) and SC containing 5-FOA (to evict the URA3 GCD7plasmid). The 5-FOA-resistant segregants were subsequently replica plated to SC lacking His and containing 30 mM 3-AT. (B) The Ura� 5-FOAR

segregants from panel A were streaked in parallel on SC lacking Leu and incubated for 5 days at 30°C. (C) Coimmunoprecipitation analysis ofstrains described in panel A prior to plasmid shuffling on 5-FOA-containing medium, harboring untagged GCD7, the WT or 164-165 allele ofGCD7-HA, and the indicated SUI2 allele (lanes 1 to 3 and 5 to 7). Strain H2507 containing only untagged GCD7 and SUI2 was analyzed as a control(lanes 4 and 8). WCEs were immunoprecipitated with anti-HA antibodies as described for Fig. 6C.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

robust growth in all three degron mutants, gcd6-td, gcd2-td, andgcd7-td, probably indicates that depletion of the cognate pro-teins is incomplete and that the small amount of eIF2B stillbeing produced can raise the level of TC to nearly WT levelswhen eIF2 and tRNA i

Met are both overproduced.It is intriguing that overexpressing eIF2 but not tRNA i

Met

partially suppressed the growth defect of the gcd6-td mutant,whereas overproducing tRNA i

Met but not eIF2 substantiallyrescued the gcd2-td mutant. One way to explain these findingsis to propose that catalysis of GDP-GTP exchange, and notbinding of substrate eIF2-GDP, is rate limiting in the gcd2-tddegron, so that overexpressing eIF2 and thereby increasing theeIF2-GDP concentration cannot restore catalyzed GDP-GTP

exchange, whereas overexpressing tRNA iMet drives the ex-

change reaction forward by mass action. However, since thegrowth of gcd6-td mutant cells can be rescued by overexpress-ing eIF2 but not tRNA i

Met, it appears that increasing the con-centration of eIF2 is the more effective means of compensatingfor decreased GEF activity, even when the catalytic subunit isdepleted and both substrate binding and catalysis are impaired.Presumably, the combination of spontaneous exchange andlow-level catalyzed exchange by the residual ε/Gcd6, coupledwith eIF2 overexpression, increases the eIF2-GTP level highenough to support appreciable TC formation. The fact thatoverexpressing tRNA i

Met, but not eIF2, rescues the gcd2-tdmutant therefore suggests that the absence of �/Gcd2 provokes

FIG. 9. Hypothetical model explaining the effects of gcd7 and sui2 mutations on productive and nonproductive modes of eIF2 binding to eIF2B.(A) GDP-GTP exchange on unphosphorylated eIF2 by WT eIF2B. eIF2 (three subunits in red, orange, and yellow) has two binding surfaces ineIF2B that promote the reaction. The regulatory subcomplex (three green subunits) binds eIF2� in a productive manner that allows the catalyticsubcomplex (two blue subunits) to interact properly with the G domain of eIF2� to catalyze efficient exchange of GDP (ball labeled “D”) for GTP(ball labeled “T”) (thick solid black arrows). (B) Phosphorylation of eIF2� (ball labeled “P”) inhibits GDP-GTP exchange by strengthening anonproductive mode of eIF2� binding to the eIF2B regulatory subcomplex, which prevents proper interaction between ε and Gcd6 in the catalyticsubcomplex and eIF2�-GDP. (C) The lethal substitution in eIF2B� of gcd7-358-360 disrupts productive binding of eIF2� to the regulatorysubcomplex as the means of impairing nucleotide exchange. (D and E) The R88T Gcn� substitution in eIF2� modestly impairs productive bindingby unphosphorylated eIF2�, having little effect on the exchange reaction on its own (D); R88T strongly reduces nonproductive binding byphosphorylated eIF2� to restore nucleotide exchange on eIF2(�P)-GDP (E). (F and G) The L84F Gcn� substitution in eIF2� does not affectproductive binding of eIF2� (F) and only impairs nonproductive binding to restore GDP-GTP exchange on eIF2(�P)-GDP (G). (H, I, and J) Thegcd7-164-165 substitution mimics eIF2� phosphorylation, as in panel B, to stabilize nonproductive eIF2� binding (H). This nonproductiveinteraction is exacerbated by destabilization of productive binding by eIF2�-R88T, as in panel D, to confer a lethal reduction in exchange (I) butis eliminated by eIF2�-L84F to restore exchange to WT levels (J). Thin and gray dotted arrows indicate moderate and strong reductions inexchange, respectively; substitution mutations are indicated by asterisks.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

a defect beyond nucleotide exchange that cannot be rescued byincreasing eIF2 abundance and thereby boosting the eIF2-GTP level in the manner that occurs when ε/Gcd6 is depleted.Hence, we favor the alternative hypothesis that �/Gcd2 per-forms an important function in loading Met-tRNA i

Met on eIF2-GTP and that overexpressing tRNA i

Met compensates for thisdefect by mass action, utilizing the eIF2-GTP generatedthrough nucleotide exchange catalyzed by the four-subuniteIF2B lacking �/Gcd2. This explanation can also account forour finding that neither eIF2 nor tRNA i

Met overexpressionalone rescues the gcd7-td mutant, as the codepletion of �/Gcd2occurring in this strain imparts a requirement for tRNAi

Met

overexpression, and the impaired eIF2-eIF2B interactionevoked by �/Gcd7 depletion necessitates eIF2 overexpression,to achieve adequate eIF2-GTP levels. A compound defect innucleotide exchange and Met-tRNA i

Met loading in the gcd7-tdstrain could also account for our finding that this mutant ap-pears to contain lower levels of TC than does the gcd6-tdmutant under nonpermissive conditions (Fig. 3C). Obviously,additional biochemical experiments are required to test theintriguing hypothesis that �/Gcd2 enhances Met-tRNA i

Met

loading on eIF2-GTP.The amount of tRNA i

Met found associated with eIF2 in cellextracts was reduced by a factor of less than 2 in all threedegron mutants. In contrast, the gcd11-N135D substitution ineIF2�, shown previously to impair Met-tRNA i

Met binding toeIF2-GTP in vitro (1), produced an �4-fold reduction in thisparameter, providing evidence that the assay faithfully mea-sures the abundance of TC in cells. The fact that the threedegron mutants are inviable but gcd11-N135D mutant cells areviable under the conditions of this assay suggests that TCassembly is not the only essential function of eIF2B. Onecaveat, however, is that we cannot determine by this assaywhether Met-tRNA i

Met is bound to eIF2 in its GTP-boundstate. Considering that the affinity of Met-tRNA i

Met for eIF2-GTP is only �15-fold higher than for eIF2-GDP or for apoeIF2 (29), a sizable fraction of the Met-tRNA i

Met that coim-munoprecipitated with eIF2 from the eIF2B degron mutantextracts might not be an authentic TC. Further studies arerequired to distinguish between these possibilities and evaluatethe likelihood that eIF2B additionally acts at a step down-stream of TC assembly.

ACKNOWLEDGMENTS

We thank Tom Dever for helpful discussion and Ernie Hannig forGcd11 antibodies.

This work was supported in part by the Intramural Research Pro-gram of the NIH.

REFERENCES

1. Alone, P. V., C. Cao, and T. E. Dever. 2008. Translation initiation factor2gamma mutant alters start codon selection independent of Met-tRNAbinding. Mol. Cell. Biol. 28:6877–6888.

2. Anthony, T. G., J. R. Fabian, S. R. Kimball, and L. S. Jefferson. 2000.Identification of domains within the epsilon-subunit of the translation initi-ation factor eIF2B that are necessary for guanine nucleotide exchange ac-tivity and eIF2B holoprotein formation. Biochim. Biophys. Acta 1492:56–62.

3. Asano, K., T. Krishnamoorthy, L. Phan, G. D. Pavitt, and A. G. Hinnebusch.1999. Conserved bipartite motifs in yeast eIF5 and eIF2Bε, GTPase-activat-ing and GDP-GTP exchange factors in translation initiation, mediate bindingto their common substrate eIF2. EMBO J. 18:1673–1688.

4. Boeke, J. D., J. Trueheart, G. Natsoulis, and G. R. Fink. 1987. 5-Fluorooroticacid as a selective agent in yeast molecular genetics. Methods Enzymol.154:164–175.

5. Bushman, J. L., A. I. Asuru, R. L. Matts, and A. G. Hinnebusch. 1993.Evidence that GCD6 and GCD7, translational regulators of GCN4, aresubunits of the guanine nucleotide exchange factor for eIF-2 in Saccharo-myces cerevisiae. Mol. Cell. Biol. 13:1920–1932.

6. Cigan, A. M., L. Feng, and T. F. Donahue. 1988. tRNAiMet functions in

directing the scanning ribosome to the start site of translation. Science242:93–97.

7. Cigan, A. M., M. Foiani, E. M. Hannig, and A. G. Hinnebusch. 1991. Com-plex formation by positive and negative translational regulators of GCN4.Mol. Cell. Biol. 11:3217–3228.

8. Craddock, B. L., and C. G. Proud. 1996. The �-subunit of the mammalianguanine nucleotide-exchange factor eIF-2B is essential for catalytic activityin vivo. Biochem. Biophys. Res. Commun. 220:843–847.

9. Cross, F. R. 1997. ‘Marker swap’ plasmids: convenient tools for buddingyeast molecular genetics. Yeast 13:647–653.

10. Dar, A. C., T. E. Dever, and F. Sicheri. 2005. Higher-order substrate recog-nition of eIF2alpha by the RNA-dependent protein kinase PKR. Cell 122:887–900.

11. Dev, K., T. J. Santangelo, S. Rothenburg, D. Neculai, M. Dey, F. Sicheri,T. E. Dever, J. N. Reeve, and A. G. Hinnebusch. 2009. Archaeal aIF2Binteracts with eukaryotic translation initiation factors eIF2alpha andeIF2Balpha: implications for aIF2B function and eIF2B regulation. J. Mol.Biol. 392:701–722.

12. Dever, T. E., W. Yang, S. Åstrom, A. S. Bystrom, and A. G. Hinnebusch. 1995.Modulation of tRNAi

Met, eIF-2, and eIF-2B expression shows that GCN4 trans-lation is inversely coupled to the level of eIF-2.GTP.Met-tRNAi

Met ternarycomplexes. Mol. Cell. Biol. 15:6351–6363.

13. Dey, M., B. Trieselmann, E. G. Locke, J. Lu, C. Cao, A. C. Dar, T. Krish-namoorthy, J. Dong, F. Sicheri, and T. E. Dever. 2005. PKR and GCN2kinases and guanine nucleotide exchange factor eukaryotic translation initi-ation factor 2B (eIF2B) recognize overlapping surfaces on eIF2alpha. Mol.Cell. Biol. 25:3063–3075.

14. Dohmen, R. J., P. Wu, and A. Varshavsky. 1994. Heat-inducible degron: amethod for constructing temperature-sensitive mutants. Science 263:1273–1276.

15. Erickson, F. L., and E. M. Hannig. 1996. Ligand interactions with eukaryotictranslation initiation factor 2: role of the �-subunit. EMBO J. 15:6311–6320.

16. Erickson, F. L., J. Nika, S. Rippel, and E. M. Hannig. 2001. Minimumrequirements for the function of eukaryotic translation initiation factor 2.Genetics 158:123–132.

17. Fabian, J. R., S. R. Kimball, N. K. Heinzinger, and L. S. Jefferson. 1997.Subunit assembly and guanine nucleotide exchange activity of eukaryoticinitiation factor-2B expressed in Sf9 cells. J. Biol. Chem. 272:12359–12365.

18. Foiani, M., A. M. Cigan, C. J. Paddon, S. Harashima, and A. G. Hinnebusch.1991. GCD2, a translational repressor of the GCN4 gene, has a generalfunction in the initiation of protein synthesis in Saccharomyces cerevisiae.Mol. Cell. Biol. 11:3203–3216.

19. Garcia-Barrio, M. T., T. Naranda, R. Cuesta, A. G. Hinnebusch, J. W. B.Hershey, and M. Tamame. 1995. GCD10, a translational repressor of GCN4,is the RNA-binding subunit of eukaryotic translation initiation factor-3.Genes Dev. 9:1781–1796.

20. Gietz, R. D., and A. Sugino. 1988. New yeast-Escherichia coli shuttle vectorsconstructed with in vitro mutagenized yeast genes lacking six-base pair re-striction sites. Gene 74:527–534.

21. Gietz, R. D., A. R. Willems, and R. A. Woods. 1995. Studies on the transfor-mation of intact yeast cells by the LiAc/SS-DNA/PEG procedure. Yeast11:355–560.

22. Gomez, E., S. S. Mohammad, and G. D. Pavitt. 2002. Characterization of theminimal catalytic domain within eIF2B: the guanine-nucleotide exchangefactor for translation initiation. EMBO J. 21:5292–5301.

23. Gomez, E., and G. D. Pavitt. 2000. Identification of domains and residueswithin the epsilon subunit of eukaryotic translation initiation factor 2B(eIF2�) required for guanine nucleotide exchange reveals a novel activationfunction promoted by eIF2B complex formation. Mol. Cell. Biol. 20:3965–3976.

24. Hannig, E. H., N. P. Williams, R. C. Wek, and A. G. Hinnebusch. 1990. Thetranslational activator GCN3 functions downstream from GCN1 and GCN2in the regulatory pathway that couples GCN4 expression to amino acidavailability in Saccharomyces cerevisiae. Genetics 126:549–562.

25. Hinnebusch, A. G. 1996. Translational control of GCN4: gene-specific reg-ulation by phosphorylation of eIF2, p. 199–244. In J. W. B. Hershey, M. B.Mathews, and N. Sonenberg (ed.), Translational control. Cold Spring Har-bor Laboratory Press, Cold Spring Harbor, NY.

26. Hinnebusch, A. G. 2005. Translational regulation of GCN4 and the generalamino acid control of yeast. Annu. Rev. Microbiol. 59:407–450.

27. Hoffman, C. S., and F. Winston. 1987. A ten-minute DNA preparation fromyeast efficiently releases autonomous plasmids for transformation of Esche-richia coli. Gene 57:267–272.

28. Jivotovskaya, A. V., L. Valasek, A. G. Hinnebusch, and K. H. Nielsen. 2006.Eukaryotic translation initiation factor 3 (eIF3) and eIF2 can promotemRNA binding to 40S subunits independently of eIF4G in yeast. Mol. Cell.Biol. 26:1355–1372.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

29. Kapp, L. D., and J. R. Lorsch. 2004. GTP-dependent recognition of themethionine moiety on initiator tRNA by translation factor eIF2. J. Mol. Biol.335:923–936.

30. Kawagishi-Kobayashi, M., J. B. Silverman, T. L. Ung, and T. E. Dever. 1997.Regulation of the protein kinase PKR by the vaccinia virus pseudosubstrateinhibitor K3L is dependent on residues conserved between the K3L proteinand the PKR substrate eIF2�. Mol. Cell. Biol. 17:4146–4158.

31. Kimball, S. R., J. R. Fabian, G. D. Pavitt, A. G. Hinnebusch, and L. S.Jefferson. 1998. Regulation of guanine nucleotide exchange through phos-phorylation of eukaryotic initiation factor eIF2�. J. Biol. Chem. 273:12841–12845.

32. Krishnamoorthy, T., G. D. Pavitt, F. Zhang, T. E. Dever, and A. G. Hinne-busch. 2001. Tight binding of the phosphorylated � subunit of initiationfactor 2 (eIF2�) to the regulatory subunits of guanine nucleotide exchangefactor eIF2B is required for inhibition of translation initiation. Mol. Cell.Biol. 21:5018–5030.

33. Labib, K., J. A. Tercero, and J. F. Diffley. 2000. Uninterrupted MCM2-7function required for DNA replication fork progression. Science 288:1643–1647.

34. Longtine, M. S., A. McKenzie III, D. J. Demarini, N. G. Shah, A. Wach, A.Brachat, P. Philippsen, and J. R. Pringle. 1998. Additional modules forversatile and economical PCR-based gene deletion and modification in Sac-charomyces cerevisiae. Yeast 14:953–961.

35. Lu, J., E. B. O’Hara, B. A. Trieselmann, P. R. Romano, and T. E. Dever.1999. The interferon-induced double-stranded RNA-activated protein ki-nase PKR will phosphorylate serine, threonine, or tyrosine at residue 51 ineukaryotic initiation factor 2alpha. J. Biol. Chem. 274:32198–32203.

36. Mohammad-Qureshi, S. S., R. Haddad, E. J. Hemingway, J. P. Richardson,and G. D. Pavitt. 2007. Critical contacts between the eukaryotic initiationfactor 2B (eIF2B) catalytic domain and both eIF2beta and -2gamma mediateguanine nucleotide exchange. Mol. Cell. Biol. 27:5225–5234.

37. Pavitt, G. D., K. V. A. Ramaiah, S. R. Kimball, and A. G. Hinnebusch. 1998.eIF2 independently binds two distinct eIF2B subcomplexes that catalyze andregulate guanine-nucleotide exchange. Genes Dev. 12:514–526.

38. Pavitt, G. D., W. Yang, and A. G. Hinnebusch. 1997. Homologous segmentsin three subunits of the guanine nucleotide exchange factor eIF2B mediatetranslational regulation by phosphorylation of eIF2. Mol. Cell. Biol. 17:1298–1313.

39. Proud, C. G. 1992. Protein phosphorylation in translational control. Curr.Top. Cell. Regul. 32:243–369.

40. Ramirez, M., R. C. Wek, C. R. Vazquez de Aldana, B. M. Jackson, B.Freeman, and A. G. Hinnebusch. 1992. Mutations activating the yeast eIF-2�kinase GCN2: isolation of alleles altering the domain related to histidyl-tRNA synthetases. Mol. Cell. Biol. 12:5801–5815.

41. Reid, G. A., and G. Schatz. 1982. Import of proteins into mitochondria.Yeast cells grown in the presence of carbonyl cyanide m-chlorophenylhydra-zone accumulate massive amounts of some mitochondrial precursorpolypeptides. J. Biol. Chem. 257:13056–13061.

42. Richardson, J. P., S. S. Mohammad, and G. D. Pavitt. 2004. Mutationscausing childhood ataxia with central nervous system hypomyelination re-duce eukaryotic initiation factor 2B complex formation and activity. Mol.Cell. Biol. 24:2352–2363.

43. Sherman, F., G. R. Fink, and C. W. Lawrence. 1974. Methods in yeastgenetics, p. 61–64. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.

44. Sikorski, R. S., and P. Hieter. 1989. A system of shuttle vectors and yeasthost strains designed for efficient manipulation of DNA in Saccharomycescerevisiae. Genetics 122:19–27.

44a.Singh, C. R., T. Udagawa, B. Lee, S. Wassink, H. He, Y. Yamamoto, J. T.Anderson, G. D. Pavitt, and K. Asano. 2007. Change in nutritional statusmodulates the abundance of critical pre-initiation intermediate complexesduring translation initiation in vivo. J. Mol. Biol. 370:315–330.

45. Vazquez de Aldana, C. R., T. E. Dever, and A. G. Hinnebusch. 1993. Muta-tions in the � subunit of eukaryotic translation initiation factor 2 (eIF-2�)that overcome the inhibitory effects of eIF-2� phosphorylation on translationinitiation. Proc. Natl. Acad. Sci. U. S. A. 90:7215–7219.

46. Vazquez de Aldana, C. R., and A. G. Hinnebusch. 1994. Mutations in theGCD7 subunit of yeast guanine nucleotide exchange factor eIF-2B overcomethe inhibitory effects of phosphorylated eIF-2 on translation initiation. Mol.Cell. Biol. 14:3208–3222.

47. Vazquez de Aldana, C. R., R. C. Wek, P. San Segundo, A. G. Truesdell, andA. G. Hinnebusch. 1994. Multicopy tRNA genes functionally suppress mu-tations in yeast eIF-2� kinase GCN2: evidence for separate pathways cou-pling GCN4 expression to uncharged tRNA. Mol. Cell. Biol. 14:7920–7932.

48. Williams, D. D., N. T. Price, A. J. Loughlin, and C. G. Proud. 2001. Char-acterization of the mammalian initiation factor eIF2B complex as a GDPdissociation stimulator protein. J. Biol. Chem. 276:24697–24703.

49. Yang, W., and A. G. Hinnebusch. 1996. Identification of a regulatory sub-complex in the guanine nucleotide exchange factor eIF2B that mediatesinhibition by phosphorylated eIF2. Mol. Cell. Biol. 16:6603–6616.


on Septem


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

The ␤/Gcd7 Subunit of Eukaryotic Translation Initiation Factor 2B (eIF2B), a Guanine Nucleotide Exchange Factor, Is Crucial for Binding eIF2 In Vivo

Documents