YIH1 is an actin-binding protein that inhibits protein kinase GCN2 and impairs general amino acid control when overexpressed.

The general amino acid control (GAAC) enables yeast cells to overcome amino acid deprivation by activation of the alpha subunit of translation initiation factor 2 (eIF2alpha) kinase GCN2 and consequent induction of GCN4, a transcriptional activator of amino acid biosynthetic genes. Binding of GCN2 to GCN1 is required for stimulation of GCN2 kinase activity by uncharged tRNA in starved cells. Here we show that YIH1, when overexpressed, dampens the GAAC response (Gcn- phenotype) by suppressing eIF2alpha phosphorylation by GCN2. The overexpressed YIH1 binds GCN1 and reduces GCN1-GCN2 complex formation, and, consistent with this, the Gcn- phenotype produced by YIH1 overexpression is suppressed by GCN2 overexpression. YIH1 interacts with the same GCN1 fragment that binds GCN2, and this YIH1-GCN1 interaction requires Arg-2259 in GCN1 in vitro and in full-length GCN1 in vivo, as found for GCN2-GCN1 interaction. However, deletion of YIH1 does not increase eIF2alpha phosphorylation or derepress the GAAC, suggesting that YIH1 at native levels is not a general inhibitor of GCN2 activity. We discovered that YIH1 normally resides in a complex with monomeric actin, rather than GCN1, and that a genetic reduction in actin levels decreases the GAAC response. This Gcn- phenotype was partially suppressed by deletion of YIH1, consistent with YIH1-mediated inhibition of GCN2 in actin-deficient cells. We suggest that YIH1 resides in a YIH1-actin complex and may be released for inhibition of GCN2 and stimulation of protein synthesis under specialized conditions or in a restricted cellular compartment in which YIH1 is displaced from monomeric actin.

Phosphorylation of the ␣ subunit of translation initiation factor 2 (eIF2␣) 1 is a key regulatory mechanism for down-regulating protein synthesis in response to starvation or stress in eukaryotic cells. Four different eIF2␣ kinases have been identified in mammalian cells that are activated by different stimuli: HRI by hemin deprivation (1), PKR by double-stranded RNA (2), PEK or PERK by unfolded proteins in the endoplasmic reticulum (3), and GCN2 by amino acid starvation (4,5). eIF2 is necessary for delivery of charged initiator methionyl-tRNA (Met-tRNA i Met ) to the 40 S ribosomal subunits in an eIF2-GTP-Met-tRNA i Met ternary complex and is released from the initiation complex in the inactive GDP-bound form. Phosphorylation of eIF2␣ at serine 51 by eIF2␣ kinases converts eIF2 from a substrate to an inhibitor of its guanine nucleotide exchange factor, eIF2B. Since only eIF2-GTP can bind Met-tRNA i Met , the inhibition of eIF2B evoked by eIF2 phosphorylation leads to a decrease in ternary complex formation and a general reduction in protein synthesis (4).
GCN2 is the sole eIF2␣ kinase in the yeast Saccharomyces cerevisiae, where it was first identified as being required for growth under amino acid starvation conditions. Phosphorylation of eIF2␣ by GCN2 induces the translation of GCN4 mRNA, encoding a transcriptional activator of amino acid biosynthetic enzymes, while dampening the translation of most other messages. Four short open reading frames in the GCN4 mRNA leader underlie a specialized reinitiation mechanism that allows efficient translation of this mRNA when the ternary complex level drops (4). The increased expression of GCN4 and its amino acid biosynthetic target genes in starved cells is known as general amino acid control (GAAC) (6).
GCN2 is present as a latent kinase under nonstarvation conditions and is activated by uncharged tRNAs that accumulate in amino acid-starved cells. The uncharged tRNA binds to a regulatory domain in GCN2 that resembles histidyl-tRNA synthetase, and this interaction is believed to induce a conformational change that overcomes an intrinsic defect in the adjacent kinase domain (7)(8)(9)(10). The products of GCN1 and GCN20 are also necessary for the activation of GCN2 by uncharged tRNA in starved cells (11,12). GCN1 and GCN20 form a protein complex (12) that binds to the N-terminal domain (NTD) of GCN2, which is highly conserved among all GCN2 orthologs. The GCN2 NTD is required for kinase function in vivo, and its overexpression impairs complex formation between native GCN1 and GCN2 and produces a dominant Gcn Ϫ (general control noninducible) phenotype (13,14). A C-terminal segment in GCN1 is necessary and sufficient for complex formation with the GCN2 NTD and, when overexpressed in vivo, competes with full-length GCN1 for GCN2 binding and likewise confers a dominant Gcn Ϫ phenotype (14,15). Arginine 2259 in the GCN2-binding domain of GCN1 was shown to be essential for GCN1/GCN2 association in vivo, and the Gcn Ϫ phenotype produced by mutating this residue was suppressed by overexpressing GCN2 (15). Other point mutations in the GCN2-NTD and GCN1 C-terminal region that impair interaction of these segments in the yeast two-hybrid assay also conferred a Gcn Ϫ phenotype and diminished eIF2␣ phosphorylation in starved cells (16). Together, these findings demonstrate that association between the NTD of GCN2 and the C-terminal region of GCN1 is required for eIF2␣ phosphorylation by GCN2 and the GAAC response in amino acid-starved cells.
GCN1 and GCN2 can interact independently with ribosomes, and these interactions are thought to be crucial for activation of GCN2 in starved cells (8,15,17). Genetic evidence further suggests that GCN1 binds near the decoding (A) site of the ribosome. These findings have led to a model in which GCN1 and GCN2 are tethered to the ribosome in a GCN1-GCN20-GCN2 complex, wherein GCN1 facilitates transfer of uncharged tRNA from the A-site to the tRNA-binding domain in GCN2 for kinase activation (15,18). This proposed mechanism has similarities with that demonstrated in Escherichia coli for activation of RelA protein by uncharged tRNA in the A site, which mediates the stringent response to amino acid starvation (19,20). YIH1, a yeast homologue of the mouse gene IMPACT (an imprinted gene of unknown function), has a domain related in sequence to the GCN2 NTD and was shown to interact in the yeast two-hybrid assay with the same region of GCN1 that binds to GCN2 (14). If YIH1 and GCN2 compete for the same binding site on GCN1 in vivo, then overexpressing YIH1 might displace GCN2 from GCN1 and impede the activation of GCN2 and attendant derepression of GCN4 translation, producing a Gcn Ϫ phenotype. In fact, it was found that overexpression of YIH1 had a dominant negative Gcn Ϫ phenotype, conferring sensitivity to an inhibitor of histidine biosynthesis, 3-aminotriazole (3AT) (14). However, no further evidence was provided that YIH1 functions by dissociation of the GCN1-GCN2 com-plex or that it serves as an authentic negative regulator of GCN2 when expressed at native levels in yeast cells.
In this report, we provide direct evidence that overexpressing YIH1 impedes the GAAC response by competing with GCN2 for GCN1 binding, thus reducing the concentration of GCN1-GCN2 complexes in the cell and decreasing phosphorylation of eIF2␣. However, deletion of YIH1 does not have a detectable effect on GCN2 kinase function under nonstarvation or starvation conditions, indicating that YIH1 at native levels is not a general inhibitor of GCN2. Interestingly, we found that YIH1 copurifies with monomeric actin (G-actin) and that a reduction in actin levels produces a Gcn Ϫ phenotype that is at least partly dependent on YIH1. These findings raise the possibility that YIH1 inhibits GCN2 only under certain growth conditions or in restricted locations in the cell, where YIH1 is displaced from G-actin.

MATERIALS AND METHODS
Strains and Plasmids-Yeast strains and plasmids used in this study are summarized in Tables I and II, respectively. Details of their construction are as follows. To generate HIS3 ϩ versions of strains BY4741, 5780, 4562, 5688, and 249, the wild type HIS3 gene was excised from plasmid pRS413 (21) as an Eco47III/NsiI fragment and used to transform each stain to His ϩ . The conversion of his3⌬ to HIS3 was verified by PCR amplification of genomic DNA (22) using primer his3a, annealing to sequences within the Eco47III/NsiI fragment (5Ј-GAG AAA GTA GGA GAT CTC TCT TGC G-3Ј), and primer his3b, annealing to sequences near the HIS3 gene outside of this fragment (5Ј-AAG CGC GCC TCG TTC AGA ATG ACA C-3Ј).
For construction of yeast strain ESY-11b, the yih1⌬::Kan R strain MSY-Y2 was transformed with SpeI-digested pES197-1, and the resulting transformants were grown on 5-fluoroorotic acid medium to select for eviction of the plasmid. A Ura Ϫ G418-sensitive transformant was isolated, and the replacement of yih1⌬::Kan R with FLAG-YIH1 was verified by PCR amplification of genomic DNA using the appropriate primers.
The ACT1 gene and its promoter were amplified from yeast strain BY4741 using primers designed to contain EcoRI sites that anneal to sequences upstream (Act1-UP, 5Ј-CAC ACG AAT TCA AAT AAG ACA CAC GCG AGA AC-3Ј) and downstream (Act1-DOWN, 5Ј-GAG AGG AAT TCA TAT GAT ACA CGG TCC AAT GG-3Ј) of the ACT1 coding region. The PCR product was digested with EcoRI and cloned into the EcoRI site of YCplac33 (24) to create plasmid pMJS1.
For construction of plasmids expressing the YIH1 gene from a heterologous promoter, the YIH1 open reading frame was PCR-amplified from yeast strain BY4741 using primer ES400-4 (CGC CCC GGG ATG GAC TAC AAG GAC GAT GAC GAC AAG GGA TCC GAT GAC GAT CAC GAA CAG TTG), containing at bases 4 -9 an XmaI site, at 10 -12 an ATG codon, at 13-36 the FLAG tag, at 37-42 a BamHI site, and at 43-63 YIH1 codons 2-8, and primer ES400-2 (CGA AAG CTT GTC GAC AGA ACT TGA AAT CGG ATT TCA TT), containing at bases 4 -9 a HindIII site, at 10 -15 a SalI site, and at 16 -39 a sequence from ϳ220 bases downstream of YIH1. The resulting PCR fragment was digested with BamHI and HindIII and cloned into the same restriction sites of pET28a (Stratagene) and pES128-9-1 (15), yielding pES189-D1A and pES187-B1, respectively. For construction of pES191-H1, the PCR fragment was cut with XmaI and HindIII and inserted into p2444 digested with the same enzymes. 3 p2444 is a derivative of pEMBLyex4 (25) where the URA3 marker is replaced by the TRP1 marker.
For construction of a low copy plasmid containing YIH1 under its own promoter, the YIH1 gene was PCR-amplified as outlined above, except that primer ES400-1 (TAC GGC GGC CGC TCT GCT TTT CAA ATT GGC TCA T, containing at bases 5-12 a NotI site and at 13-34 a sequence corresponding to 170 -190 bases upstream of YIH1) and primer 400-2 were used. The PCR fragment was digested with NotI and SalI and introduced into pRS316 (21) digested with the same enzymes, resulting in pES185-1. For introduction of a FLAG tag N-terminal to YIH1, a fusion PCR was performed using as the outside primers ES400-1 (described above) and primer ES400-8 (CGG GAA GGA GAT CTG CAG T), containing at bases 9 -14 the unique BglII site in YIH1 and bases 132-150 of the YIH1 open reading frame. The mutagenic primers for introduction of the FLAG tag were antisense primer ES400-5 (TC CAT TGA GCT TTT CTT TCC TC), containing at bases 3-5 the start codon and bases 6 -22 from immediately upstream of YIH1, and sense primer ES400-6 (GAG GAA AGA AAA GCT CAA TGG ACT ACA AGG ACG ATG AC), containing bases 1-17 from immediately upstream of YIH1, at bases 18 -20 the start codon, and at bases 21-38 the first six codons of the FLAG tag. Two separate PCRs were performed, one amplifying the sequence extending from the upstream cloning site NotI to the YIH1 ATG using primers ES400-1 and ES400-5 and pES185-1 as template and the second reaction amplifying sequences extending from the YIH1 ATG to the BglII site in the YIH1 gene with primers ES400-6 and ES400-8, using pES191-H1 as template. The two PCR products were fused together by performing a third PCR with primers ES400-1 and ES400-8 and the products of the first two PCR reactions as template. The resulting PCR fragment was digested with NotI and BglII and used to replace the NotI-BglII fragment in pES185-1, resulting in pES196-6-4. The SalI-SacI fragment of pES196-6-4, containing the FLAG-YIH1 gene including surrounding sequences, was introduced into YIplac211 digested with the same enzymes (24), resulting in pES197-1.
The YIH1::hisG::URA3::hisG deletion cassette was constructed as follows. A 100-bp fragment immediately upstream of YIH1 was PCRamplified from yeast strain BY4741 using primers ES400-22 and ES200-23. . By a subsequent PCR using both PCRamplified fragments just described as templates and primers ES400-22 and ES400-25, the 100-bp sequence upstream and 130-bp sequence downstream of YIH1 were fused together and separated by only a BglII site. The fusion product was digested with XbaI and SalI and introduced into the similarly digested plasmid pBluescript (Stratagene), yielding pES221-6. The hisG::URA3::hisG cassette was excised from 3 G. Pavitt and A. G. Hinnebusch, unpublished observations.
This study plasmid pHQ221 2 with BamHI and inserted into BglII-digested pES221-6, resulting in pES222-202-2. Quick Preparation of Genomic DNA from Yeast Cells for PCRs-Cell pellets from 5 ml of yeast overnight cultures were resuspended in 200 l of TE buffer. 100 l of glass beads (0.5-mm diameter) and 200 l of phenol were added, and the samples were vortexed for 2 min. The supernatant was extracted once with 200 l of chloroform and then used directly as a template in PCRs.
Purification of the YIH1-containing Complex-Yeast strains were grown in YPD medium to exponential phase (A 600 ϭ 10) in a 10-liter fermentor, harvested via continuous flow centrifugation, resuspended in buffer A (1ϫ phosphate-buffered saline, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 0.5 mM 2-mercaptoethanol, 1ϫ complete protease inhibitor tablets (Roche Applied Science) to about 1 g of cells/ml, and passed twice through a French pressure cell. Cell debris was removed (10,500 rpm, 15 min), and the extract was cleared by centrifugation at 42,000 rpm for 45 min in a Beckman Ti45 rotor). After incubation of the extract with 500 l of (100%) FLAG resin (Sigma) overnight, the protein complexes were eluted three times with FLAG peptide according to the manufacturer's protocol. 50 l of the second FLAG-elution was applied to a Superdex200 column (24-ml bed volume; Amersham Biosciences), and 80-l fractions were collected. For analysis in a glycerol gradient, 100 l of the second FLAG elution was applied to a 10-40% gradient and separated by velocity centrifugation (Beckman rotor SW60; 55,000 rpm, 4 h), and 200-l fractions were collected. Using the same separation conditions, a molecular size standard (Bio-Rad) was resolved on Superdex200 and a glycerol gradient.
Mass Spectrometry Identification of Proteins-Samples were separated on SDS-PAGE gels and stained with Coomassie, and the resulting bands were excised after washing the gel with 1% acetic acid. The bands were digested in-gel with trypsin, and the peptides were extracted as previously described (30). The extracted peptides were loaded onto a 75-mm inner diameter reverse phase high pressure liquid chromatography column (Poros R2; Perceptive Biosystems) equilibrated in 0.5% acetic acid. Peptides were eluted using a linear gradient of 0 -40% acetonitrile over 60 min followed by 40 -60% over 10 min at a flow rate of 0.3 ml/min. Eluting peptides were analyzed by electrospray ionization tandem mass spectrometry using an ion trap mass spectrometer (LCQ DECA; Thermofinnigan). All tandem spectra were searched against the S. cerevisiae open reading frames data base (Stanford University) using the SEQUEST algorithm (31). Data processing of the SEQUEST output files identified the proteins in the SDS-PAGE bands as previously described (32).

RESULTS
Overexpressed YIH1 Competes with GCN2 for GCN1 Binding in Vivo-Binding of the GCN2 NTD to amino acids 2052-2427 (region D) of GCN1 is essential for activation of GCN2 under amino acid starvation conditions (15). YIH1 can bind to the same region of GCN1 in the yeast two-hybrid assay, and overexpression of YIH1 produced a dominant Gcn Ϫ phenotype, manifested by sensitivity to 3AT (14). These findings suggested that YIH1, if overexpressed, can compete with GCN2 for GCN1 binding and prevent activation of GCN2 in cells starved for histidine with 3AT. If so, then overexpression of GCN2 should overcome this competition and suppress the Gcn Ϫ phenotype of overexpressing YIH1. To test this prediction, we added a GST or FLAG tag at the N terminus of the YIH1 open reading frame and placed it under a galactose-inducible promoter. As reported, overexpression of the tagged YIH1 alleles conferred dominant 3AT sensitivity (3AT s ) in an otherwise wild type strain (Fig. 1, A and B). Importantly, overexpression of GCN2 from a high copy plasmid (hcGCN2) suppressed the Gcn Ϫ phenotype associated with GST-YIH1 overexpression (Fig. 1B), consistent with the hypothesis that YIH1 impairs GAAC by competing with GCN2 for the same binding site on GCN1.
To provide additional proof for this conclusion, we took advantage of the fact that a constitutively active GCN2 protein encoded by GCN2 c -E601K-E1591K (33) causes a slow growth (Slg Ϫ ) phenotype resulting from hyperphosphorylation of eIF2␣ and the consequent general inhibition of protein synthesis. The Slg Ϫ phenotype of GCN2 c -E601K-E1591K is suppressed by deletion of GCN1 (33), suggesting that GCN2 c -E601K-E1591K requires GCN1 binding for its activation. Therefore, if overexpressed YIH1 disrupts GCN1-GCN2 interaction, it should reverse the Slg Ϫ phenotype associated with the GCN2 c allele. In accordance with this prediction, overexpression of GST-YIH1 suppressed the Slg Ϫ phenotype of GCN2 c -E601K-E1591K (Fig. 1C). Taken together, these data provide strong genetic evidence that overexpressed YIH1 impairs activation of GCN2 and blocks the GAAC response by competing with GCN2 for binding to GCN1.
Thus far, YIH1-GCN1 interaction has been demonstrated by the yeast two-hybrid assay and with an in vitro binding assay using recombinant GCN1 fragments encompassing the GCN2/ YIH1 binding site (16). To provide physical evidence that overexpressed YIH1 sequesters native GCN1 in vivo, we asked whether endogenous GCN1 can be co-precipitated with GST-YIH1 on glutathione-agarose beads from extracts of yeast cells overexpressing GST-YIH1. We found that GCN1, but not GCN2, coprecipitated specifically with GST-YIH1 (Fig 2A), providing evidence that overexpressed GST-YIH1 binds to native GCN1 but not to the GCN1-GCN2 complex in vivo. This conclusion is consistent with the idea that YIH1 competes with GCN2 for GCN1 binding.
Next, we sought physical evidence that overexpressing YIH1 reduces GCN1-GCN2 complex formation in vivo. Cell extracts of strains overexpressing GST-tagged YIH1 or GST alone were immunoprecipitated with GCN1 antibodies, and the immune complexes were probed with GCN1 and GCN2 antibodies. The results in Fig. 2B indicated that about 53% of cellular GCN2 was complexed with GCN1 in the strain expressing GST alone, in agreement with our previous results (15). When GST-YIH1 was overexpressed in the cell, the amount of GCN2 associated with GCN1 was reduced by almost one-half (28% of cellular GCN2 complexed with GCN1), demonstrating that overexpressing YIH1 reduces GCN1-GCN2 complex formation in vivo.
It was important to establish that the Gcn Ϫ phenotype of high copy YIH1 results from decreased phosphorylation of eIF2␣ on Ser-51, the site of phosphorylation by GCN2. In accordance with this expectation, we found that YIH1 overexpression reduced the levels of eIF2␣ phosphorylated on Ser-51 (eIF2␣-P) by ϳ40% under starvation conditions and by about 90% in nonstarved cells (Fig. 2C). Together, the results in Figs. 1 and 2 show that overexpressed YIH1 inhibits GCN2 kinase activity by disrupting GCN1-GCN2 interaction.
Arg-2259 in GCN1 Is Essential for YIH1 Binding in Vitro and in Vivo-We showed previously that substitution of Arg-2259 in GCN1 abolished association of native GCN1 and GCN2 in vivo and impaired interaction of the recombinant N terminus of GCN2 with region D of GCN1 in vitro (15). The fact that the binding sites for YIH1 and GCN2 mapped to almost the same region in GCN1 (14,15) suggests that GCN2 and YIH1 might interact with the same residues in GCN1. To test whether YIH1 also requires Arg-2259 for GCN1 interaction, we tested a recombinant His 6 -YIH1 fusion protein expressed in E. coli for interaction with bacterially expressed GST-GCN1-(2052-2428), which contains the minimal GCN1 fragment sufficient for GCN2 binding (15). We found that GST-GCN1-(2052-2428) is sufficient for binding to His 6 -YIH1 in vitro and that this interaction was abolished by the R2259A mutation in the GCN1 fragment (Fig. 3A). In order to show that Arg-2259 in native GCN1 is essential for YIH1 binding in vivo, we performed GST pull-down assays on extracts of gcn1⌬ strains overexpressing GST-YIH1 and containing either GCN1 or gcn1-R2259A on a low copy plasmid. As shown in Fig. 3B, GCN1, but not gcnl-R2259A, coprecipitated with GST-YIH1 from cell extracts. These findings suggest that YIH1 and GCN2 share a key binding determinant in GCN1 region D, supporting the idea that they compete directly for complex formation with GCN1.
Is YIH1 a Negative Regulator of the GAAC When Expressed at Native Levels?-The results presented above suggest that YIH1 is a negative regulator of GCN2 that, presumably, must be disabled to permit a strong GAAC response in starved cells. If so, then deletion of YIH1 should lead to constitutive activation of GCN2. To test this possibility, we asked whether deletion of YIH1 confers resistance to the tryptophan and histidine analogs 5-fluoro-DL-tryptophan (5FT) and triazolealanine (TRA), respectively, a phenotype conferred by known GCN2 c mutations that lead to constitutive induction of GCN4 and the tryptophan and histidine biosynthetic genes under GAAC (Gcd Ϫ phenotype) (33). Surprisingly, the yih1⌬ strain showed the same sensitivity to 5FT and TRA as the isogenic wild-type strain (Fig. 4A, bottom sector labeled vector). Introduction of a plasmid-borne GCN2 c allele into the YIH1 strain produced the expected 5FT/TRA resistance, indicating a Gcd Ϫ phenotype (Fig. 4A, sector labeled GCN2 c -M788V). Again, deletion of YIH1 did not enhance the Gcd Ϫ phenotype of the GCN2 c -M788V strain. Thus, under the conditions tested, YIH1 does not appear to be a negative regulator of GCN2. We also tested the unlikely possibility that YIH1 at native levels of expression functions as an activator of GCN2 and the general control response. As shown in Fig. 4, B and C, the yih1⌬ and YIH1 cells were indistinguishable in their ability to grow in the presence of 3AT or sulfometuron (SM), an inhibitor of leucine, isoleucine, and valine biosynthesis. As expected, isogenic strains with deletions in known GCN genes conferred sensitivity to both 3AT and SM.
We asked next whether deletion of YIH1 would increase the level of eIF2␣ phosphorylation in nonstarvation conditions or decrease the time required for eIF2␣-P to return to the basal level following the addition of histidine to 3AT-starved cells. We saw no increase in the level of eIF2␣-P in yih1⌬ versus wild-type cells under nonstarvation conditions (Fig. 5A, lanes  1-4 versus lanes 5 and 6) or in cells starved for histidine or glutamine by the addition of the antimetabolites 3AT or Lmethionine-S-sulfoximine (MSX), respectively (Fig. 5B). In addition, we found that eIF2␣-P levels declined at indistinguishable rates in wild-type and yih1⌬ cells, returning to the basal level within 6 min in both strains (Fig. 5C). The experiments described above were carried out using synthetic medium supplemented with all 20 amino acids as the nonstarvation condition. We also found that yih1⌬ did not alter the levels of eIF2␣-P in cells during the early logarithmic phase of growth on the nutrient-rich medium YPD (data not shown). Thus, there is no evidence that YIH1 is required to prevent GCN2 activation in nonstarvation conditions, to limit the extent of GCN2 activation in starved cells, or to increase the rate at FIG. 1. Genetic evidence that overexpressed YIH1 competes with native GCN2 for GCN1 binding in vivo. A, overexpression of FLAG-YIH1 causes a Gcn Ϫ phenotype. Strain H1511 harboring the FLAG-YIH1 allele under the galactose-inducible promoter on high copy plasmid pES191-H1 or the corresponding 2 empty vector p2444 was grown on medium containing glucose and replica-plated to the same medium (Glc) or to medium containing galactose (Gal) and 30 mM 3AT, as indicated. B, overexpression of GST-YIH1 causes a Gcn Ϫ phenotype that can be suppressed by GCN2 overexpression. Strain H1511 containing GST-YIH1 or GST under the galactose-inducible promoter on plasmids pES187-B1 and pES128-9-1, respectively, were transformed with either high copy GCN2 plasmid pAH15 or empty vector YEp13 and characterized for sensitivity to 30 mM 3AT as described in A. The isogenic gcn1⌬ strain H2556 harboring pES128-9-1 was examined as a control for the 3AT s phenotype. C, the Slg Ϫ phenotype associated with a constitutively active GCN2 protein is partially suppressed by YIH1 overexpression. The galactose-inducible GST-YIH1 or GST alleles were introduced into isogenic strains H1402 and H1613 harboring genomic GCN2 or GCN2 c -E601K-E1591K, respectively. One, two, or four of the resulting transformants were streaked on medium containing glucose or galactose as carbon source, as indicated. which eIF2␣ phosphorylation declines to basal levels when starved cells are replenished with the limiting amino acid.
The yeast genome encodes two other proteins containing domains related to the GCN1-binding domains in GCN2 and YIH1, known as YDR152W and YLR419W. We considered the possibility that YIH1 is functionally redundant with one of these predicted proteins for down-regulation of GCN2. To test this idea, we examined strains deleted for these genes singly or in combination with yih1⌬ for resistance to 5FT and TRA. The results showed that the strains lacking only YDR152W or YLR419W, the double mutant lacking both of these genes, the double mutants lacking YIH1 and YDR152W or YLR419W, and the triple mutant lacking all three genes, were all essentially indistinguishable from the isogenic WT strain in resistance to 5FT (data not shown). Thus, it appears that none of the proteins containing domains related to the GCN1-binding domain in GCN2 have an obvious function in negatively regulating GCN2 activity in nonstarvation conditions, at least under the growth conditions tested.
YIH1 Forms a Complex with Actin in Vivo-We sought additional clues about the cellular function of YIH1 by identifying proteins that interact with YIH1 expressed at native levels. To this end, we introduced a FLAG tag at the N terminus of genomic YIH1 and purified the FLAG-YIH1 protein from whole cell extract (WCE) using anti-FLAG resin, eluting it from the resin with FLAG peptide. The resulting eluate was resolved further by gel filtration on a Superdex200 column or by velocity sedimentation through a glycerol gradient. Fractions were collected, and aliquots were resolved by SDS-polyacrylamide gels and subjected to silver staining and immunoblot analysis with anti-FLAG antibodies. In both separation procedures, a protein of about 48 kDa co-fractionated precisely with FLAG-YIH1 (Fig. 6, A and B). The copurifying protein was excised from the gel and subjected to trypsin digestion, and the recovered peptides were sequenced using tandem mass spectrometry. The results identified unambiguously the copurifying protein as actin, encoded by ACT1, and we confirmed this conclusion by Western blot analysis using antibodies against actin (Fig. 6, A  and B). Using GCN1 antibodies, we could not detect any GCN1 co-purifying with FLAG-YIH1, suggesting that under the growth conditions used for purification of the YIH1 complex (YPD medium, native amounts of YIH1) YIH1 does not interact with GCN1, or else the YIH1-GCN1 interaction is too weak to withstand the purification procedures. We showed above that

FIG. 2. Overexpressed YIH1 binds to GCN1 and actin, reduces GCN1-GCN2 association, and lowers eIF2␣ phosphorylation in vivo.
A, overexpressed YIH1 binds to GCN1 and actin in vivo, but not to GCN2. Transformants of wild-type strain H1511 containing the galactoseinducible genes GST-YIH1 (on pES187-B1) or GST alone (on pES128-9-1), respectively, were grown in minimal medium containing galactose as carbon source. WCEs were prepared, and aliquots with equivalent amounts of protein were subjected to GST pull-down assays using glutathione-Sepharose beads. The precipitated complexes were probed with antibodies against GCN1, GCN2, and actin in Western blots (upper three panels). The bottom panel shows Ponceau S staining of the membrane, revealing the amounts of GST-YIH1 and GST that were precipitated. The input lanes contained 10% of the WCEs used in the assays; the pellet lanes contained 100% of the precipitated proteins. B, overexpressed YIH1 reduces GCN1-GCN2 complex formation in vivo. WCEs from gcn1⌬ strain H2556, harboring the galactose-inducible gene GST-YIH1, and GCN1 strain H1511, containing GST-YIH1 or GST alone, respectively, were prepared as outlined in A and immunoprecipitated with GCN1 antibodies. The immune complexes were probed for GCN2 and GCN1 in Western blots (upper two panels). 10% of input (I) and supernatant (S) and 100% of precipitated protein (P) were loaded on the gel. The histogram shows the average percentage of GCN2 immunoprecipitated after normalizing for the amount of GCN1 that immunoprecipitated from the same samples (set to 100%). The mean values and S.E. values (shown by error bars) were calculated from at least eight independent assays. C, overexpressed YIH1 reduces eIF2␣-P levels under starvation and nonstarvation conditions. Yeast strains described in A were grown to exponential phase in medium containing galactose and then starved for histidine by the addition of 3AT (20 mM) for 4 h. WCEs of these cultures were subjected to Western blot analysis, using antibodies against eIF2␣ phosphorylated on Ser-51 (eIF2␣-P) and antibodies against eIF2, respectively. Two different exposures are shown for the eIF2␣-P detection. The Western signals from two independent strains and two independent blots for each strain were quantified using NIH Image software, and the results and S.E. values are presented at the bottom as the mean eIF2␣-P to total eIF2␣ ratios relative to the corresponding mean ratio measured for the unstarved GST control (lanes 3 and 4).
GCN1 is associated with overexpressed GST-YIH1 in WCEs. In the same assay, we found that actin also co-precipitated specifically with GST-YIH1 ( Fig. 2A). Taken together, our results indicate that YIH1 forms a stable complex with monomeric actin in vivo that does not contain GCN1 as a stable constituent at native levels of YIH1 expression.
Deletion of YIH1 Does Not Alter the Morphology of the Actin Cytoskeleton-The association of YIH1 with monomeric actin in vivo raises the possibility that YIH1 regulates actin polymerization or depolymerization. Hence, we asked whether deletion or overexpression of YIH1 would alter the appearance of the actin cytoskeleton. To address this question, we stained WT cells, yih1⌬ cells, and WT cells overexpressing GST-YIH1 with rhodamine-phalloidin to visualize filamentous actin structures. Growing yeast cells contain asymmetric distributions of actin, with mother cells containing mostly actin cables along the longitudinal axis and buds containing mostly actin patches along the cell surface (34). The total amount and the relative distribution of actin patches in the mother versus the daughter cell were determined. We reproducibly observed a lower proportion of yih1⌬ cells that showed strong staining, 27.2% versus 68.1% staining for the YIH1 cells (Fig. 7); however, it is unknown whether this phenotype can be attributed to a defect in the actin cytoskeleton. Of the cells that showed strong staining, we observed no change in the number of actin patches or preferential localization of the patches in the buds and no obvious morphological differences in the actin cables between yih1⌬ and YIH1 cells. Similarly, overexpression of YIH1 did not affect the distribution or total amount of actin patches, nor did it significantly change the structure of actin cables.

Reduction of the Cellular Actin Level Causes a Gcn
Ϫ Phenotype-Knowing that YIH1 interacts with actin independently of GCN1, and having found that overexpression of YIH1 inhibits GCN2, we wished to determine whether a decrease in the cellular actin content could release YIH1 from the YIH1actin complex and inhibit GCN2. To test this possibility, we took advantage of the haplo-insufficient phenotype of act1⌬/ ACT1 heterozygous diploid strains. These strains exhibit a slight growth defect at 37°C that is exacerbated by the addition of NaCl (35). We confirmed by Western analysis of WCEs that the act1⌬/ACT1 heterozygote contains a lower level of actin compared with an isogenic ACT1/ACT1 homozygous strain and also is hypersensitive to latrunculin A, a drug that inhibits actin polymerization (36) (data not shown). According to our prediction, reduced amounts of actin in the act1⌬/ ACT1 strain at 37°C should produce sensitivity to SM, a Gcn Ϫ phenotype. (We employed SM versus 3AT for these tests, because the act1⌬/ACT1 strain is a histidine auxotroph and thus cannot grow on 3AT medium lacking histidine.) Interestingly, at the semipermissive temperature of 37°C, but not at the permissive temperature of 30°C, the act1⌬/ ACT1 strain was sensitive to SM (SM s ) compared with the WT ACT1/ACT1 strain (Fig. 8A). Introduction of ACT1 on a single copy plasmid reversed the SM s phenotype of the act1⌬/ ACT1 strain (not shown), ensuring that it results from actin insufficiency.
If YIH1 mediates the inhibition of GCN2 resulting from

FIG. 3. Binding of YIH1 to GCN1 is dependent on Arg-2259 in the GCN1 C-terminal region in vitro and in vivo.
A, GST-GCN1-(2052-2428) directly interacts with His 6 -YIH1 in vitro dependent on Arg-2259. The indicated GST-GCN1 fusions (wild type (wt) or R2259A) or GST alone encoded by plasmids pES123-B1, pES164-2A, and pGEX-6p-1, respectively, were expressed in E. coli, immobilized on glutathione-Sepharose beads, and incubated with an E. coli extract containing His 6 -YIH1 encoded by plasmid pES189-D1A. Proteins bound to the beads were subjected to immunoblot analysis using His 6 antibodies (upper panel). GST proteins were visualized via Ponceau S staining (lower panel). B, YIH1 interacts with full-length GCN1 in vivo, dependent on GCN1 Arg-2259. GST pull-down assays were performed as described in the legend to Fig. 2A using gcn1⌬ strain H2556 containing the galactose-inducible gene GST-YIH1 or GST and a second plasmid carrying either GCN1 (p1832) or gcn1-R2259A (pES200-2). reduced actin levels, then YIH1 overexpression should exacerbate the SM s phenotype of the act1⌬/ACT1 strain. Consistent with this prediction, we found that overexpression of GST-YIH1, but not GST alone, from a galactose-inducible promoter greatly exacerbated the SM s phenotype of the act1⌬/ACT1 strain at 37°C (Fig. 8B). Another key prediction of the model that YIH1 down-regulates GCN2 in the act1⌬/ACT1 strain is that the SM s phenotype of the latter should be suppressed by inactivation of YIH1. To test this possibility, we deleted both YIH1 alleles in the act1⌬/ACT1 strain and compared the SM s phenotypes of transformants of the resulting act1⌬/ACT1 yih1⌬/yih1⌬ double mutant harboring plasmid-borne YIH1, plasmid-borne ACT1, or empty vector. As shown in Fig. 8C  (rows 3 and 4), the transformant containing YIH1 showed strong sensitivity to SM compared with that containing ACT1. This comparison recapitulates the SM s phenotype associated with act1⌬/ACT1 versus ACT1/ACT1 in the presence of YIH1 that was shown above in Fig. 6A. Importantly, the transformant of the act1⌬/ACT1 yih1⌬/yih1⌬ strain containing empty vector was less sensitive to SM than that containing YIH1 (Fig.  8C, rows 4 and 5) but more sensitive than that containing ACT1 (rows 3 and 5). These findings indicate that YIH1 mediates a portion of the Gcn Ϫ phenotype conferred by act1⌬/ACT1 hemizygosity. The residual SM sensitivity of the act1⌬/ACT1 FIG. 5. Deletion of YIH1 does alter levels of eIF2␣ phosphorylation in vivo. A, yih1⌬ cells exhibit wild-type levels of eIF2␣ phosphorylation in nonstarvation conditions. Overnight cultures of yih1⌬ strain MSY-Y2 and wild-type strain MSY-WT2 were grown in YPD to exponential phase and harvested at an A 600 of below 0.5 and subjected to Western analysis as described in the legend to Fig. 2C. B, yih1⌬ cells exhibit wild-type levels of eIF2␣ phosphorylation in starvation conditions. Overnight cultures of the strains described in A were diluted 50-fold in SD medium and grown for 2 h before supplementing with 10 mM 3AT and/or 0.5, 1, or 4 mM methionine-S-sulfoximine as indicated. After an additional 6 h, the cells were harvested and subjected to Western analysis as described in Fig. 2C. C, yih1⌬ cells do not show a delayed decline of eIF2␣ phosphorylation with replenishment of the limiting amino acid. Exponentially growing yih1⌬ strain MSY-Y2, gcn2⌬ strain EMSY6053-3-1, and wild-type strain MSY-WT2 were subjected to histidine starvation as outlined in B, but by using 20 mM 3AT (final concentration). After 4 h, histidine (His) was added where indicated (1.2 mM final concentration), and 1, 3, 6, or 12 min thereafter cells were killed instantly by adding trichloroacetic acid to a final concentration of 5%. The cells were washed once with 20% trichloroacetic acid and resuspended in 5% trichloroacetic acid. WCEs were generated by breaking cells with glass beads for 2 min on a vortexer. After removal of the glass beads, the insoluble fraction was pelleted and resuspended in Laemmli buffer, followed by neutralization with 1 M Tris base. Samples were boiled for 3 min, and the supernatant was subjected to Western analysis as outlined in the legend to Fig. 2C.   FIG. 6. Actin co-purifies with YIH1. Wild-type strain ESY-11b harboring genomic FLAG-YIH1 was grown in a 10-liter fermentor in YPD to exponential phase. The YIH1 complex was purified from WCE using anti-FLAG resin and eluted with FLAG peptide. The resulting eluate was subjected to one of two further purification steps as follows. A, the FLAG eluate was separated on a Superdex200 column, and aliquots of the collected fractions were resolved in an SDS-polyacrylamide gel. The gel was subjected to silver staining (upper panel) or Western analysis using antibodies against FLAG or actin (lower panels). B, the FLAG eluate was subjected to velocity sedimentation through a 10 -40% glycerol gradient. Fractions were collected from top to bottom, and the same assays were performed as in A. In A and B, a mixture of size standards was subjected to the same separations, and the fractions containing each standard are indicated at the top. Fractions containing no detectable proteins were omitted from the figure.
FIG. 8. Haplo-insufficiency of cellular actin confers a Gcn ؊ phenotype that is partially dependent on YIH1. A, the act1⌬/ACT1 diploid strain and other relevant isogenic diploid strains indicated (from top to bottom: 33642, BY4743, 27075, and 35780) were tested for sensitivity to SM as described in the legend to Fig. 4B, using undiluted culture and consecutive 10-fold dilutions at 30 and 37°C, as indicated. B, overexpression of GST-YIH1 exacerbates the SM s phenotype associated with insufficient actin levels. The act1⌬/ACT1 strain 27075 and isogenic wild-type strain BY4743, both harboring plasmid-borne galactose-inducible gene GST-YIH1 or GST alone on plasmids pES187-B1 and pES128-9-1, respectively, were analyzed for SM s phenotypes at 37°C as in A except using medium with galactose as the sole carbon source. C, inhibition of GAAC by actin is partially mediated through YIH1. Saturated cultures were prepared from strains as indicated (from top to bottom, ESY12A-a and ESY6F-b) harboring low copy vector (pRS316), plasmid-borne low copy YIH1 (pES196 -6-4), or single copy ACT1 (pMJS1). Subsequently, 5 l of 10fold dilutions were transferred to solid medium containing additional supplements as outlined below and incubated at 36 -37°C. FIG. 7. Deletion of YIH1 reduces the proportion of cells stainable with rhodamine-phalloidin but does not alter the morphology of the actin cytoskeleton. A, cells were grown in YPD to exponential phase to an A 600 less than 0.5 absorbance units, subjected to rhodamine-phalloidin staining (42), and visualized in a Nikon E1000 light microscope using a Sensicam QE digital CCD camera (Cooke Corp.) and a customized filter set (excitation 540/25 nm; emission 605/55 nm, beam splitter 565 nm, Chroma Technology) controlled by IP Labs software (Scanalytics Inc.). B, cells in A stained weakly and strongly by rhodamine-phalloidin were counted, and the averages and S.E. values are indicated as percentage of the total number of cells. In total, 389 wild-type cells and 758 yih1⌬ cells were analyzed.
yih1⌬/yih1⌬ strain compared with the ACT1/ACT1 strains shown in Fig. 8C represents a YIH1-independent component of the Gcn Ϫ defect associated with actin haplo-insufficiency. DISCUSSION It was shown previously that YIH1 interacts with the isolated GCN2-binding domain in GCN1 and that overexpression of YIH1 confers sensitivity to 3AT, an inhibitor of histidine biosynthesis. This finding led to the prediction that YIH1 can negatively regulate GCN2 by preventing formation of the GCN1-GCN2 complex (14). In this report, we provided several lines of evidence demonstrating that the 3AT s phenotype conferred by YIH1 overexpression results from dissociation of the GCN1-GCN2 complex and consequent impaired activation of GCN2. First, we showed that overexpression of YIH1 can suppress the growth defect conferred by a constitutively activated form of GCN2 that inhibits general translation initiation by high level phosphorylation of eIF2␣ (33). This result provides strong genetic evidence that YIH1 overexpression specifically inhibits GCN2 function. Biochemical evidence for this conclusion came from our finding that cells overexpressing YIH1 contain reduced levels of eIF2␣-P in nonstarvation conditions and also in response to histidine starvation. Evidence that overexpressed YIH1 competes with GCN2 for GCN1 binding came from our in vivo GST pull-down experiments showing that overexpressed GST-YIH1 forms a complex with native GCN1, but not with GCN2, and from our coimmunoprecipitation experiments showing that YIH1 overexpression reduces the level of GCN1-GCN2 complexes in vivo. We also showed that binding of YIH1 to GCN1, both in vitro and in vivo, is dependent on the same residue in GCN1 (Arg-2259) that is critical for GCN1-GCN2 interaction (15). Thus, YIH1 and GCN2 share a critical binding determinant in the C-terminal region of GCN1.
These findings point to a potential role for YIH1 as a negative regulator of GCN2 and the GAAC response. However, we were unable to demonstrate complex formation between YIH1 and GCN1 in vivo unless YIH1 was overexpressed. In addition, we did not observe any effect of deleting YIH1 on the GAAC response or on the levels of eIF2␣ phosphorylation under starvation or nonstarvation conditions. We considered the possibility that YIH1 could be required only to ensure a rapid dephosphorylation of eIF2␣ when starved cells are replenished with the limiting amino acid. However, we did not detect any difference between WT and yih1⌬ cells in the rate of eIF2␣ dephosphorylation when 3AT-treated cells were supplied with histidine. Thus, we have no evidence that YIH1 functions as a general inhibitor of GCN2 under normal growth conditions. We also considered the possibility that the negative regulatory function of YIH1 could be carried out redundantly by one of the other predicted proteins in yeast that contain a domain related to the GCN1-binding domains in GCN2 and YIH1, namely YDR152W and YLR419W. However, deleting YDR152W and YLR419W, either together or in combination with a deletion of YIH1 had no significant effect on the GAAC response.
In an effort to uncover the function of YIH1 in vivo, we purified a FLAG-tagged YIH1 expressed from the chromosomal allele under its own promoter and found that the protein occurs in a 1:1 complex with monomeric actin containing no trace of GCN1. This finding raised the possibility that the ability of YIH1 to inhibit GCN2 might be regulated by its association with G-actin. This possibility was supported by our finding that actin haplo-insufficiency in an ACT1/act1⌬ heterozygote confers sensitivity to an inhibitor of isoleucine, leucine, and valine biosynthesis (SM) and that this was exacerbated by YIH1 overexpression. This Gcn Ϫ phenotype was partially suppressed by deletion of YIH1, indicating that the defect in GAAC associated with actin insufficiency was mediated at least partly by YIH1. Thus, the reduced actin levels in the ACT1/act1⌬ heterozygote may reduce the concentration of monomeric actin, leading to release of YIH1 from the YIH1-actin complex and consequent formation of the YIH1-GCN1 complex with attendant inhibition of GCN2. The residual SM S phenotype observed in the ACT1/act1⌬ yih1⌬/yih1⌬ strain presumably reflects a YIH1independent reduction in the expression or activation function of GCN4 in cells with an actin insufficiency.
One difficulty with the idea that actin insufficiency leads to YIH1-mediated inhibition of GCN2 is that we have been unable to detect a reduction in eIF2␣-P levels in the ACT1/act1⌬ heterozygote (data not shown). One mitigating factor here is that a portion of the Gcn Ϫ phenotype of the ACT1/act1⌬ strain is independent of YIH1 and thus may not involve impaired activation of GCN2. Considering that actin is a component of several transcriptional coactivator complexes (37), a defect in transcriptional activation by GCN4 could be responsible for the YIH1-independent component of the Gcn Ϫ phenotype of the ACT1/act1⌬ strain. However, another intriguing possibility is that YIH1-mediated inhibition of GCN2 could be a localized phenomenon, restricted to the site of bud emergence, where an optimal level of eIF2 activity and translation initiation would support rapid bud growth. This model could explain why we failed to detect an increase in the total cytoplasmic pool of eIF2␣-P in yih1⌬ cells. There are precedents for localized regulation of eIF2␣ phosphorylation and translational control in mammalian cells (38,39).
This last model assumes that the concentration of YIH1actin complexes would be locally reduced at the site of bud growth. Most of the filamentous actin in yeast cells occurs in the cortical actin patches located preferentially near regions of exocytosis in the growing bud. There is increasing evidence that actin patches are sites of "compensatory" endocytosis needed to recycle plasma membrane and the protein secretion machinery utilized in exocytosis during bud emergence. The actin patches are highly dynamic structures, containing many actin-binding proteins and regulatory factors, and are constantly turning over (40). It has been predicted that an actin cable assembly machine also would be located near the sites of exocytosis in the bud to anchor the growing end of actin cables as they project outward into the mother cell to support polarized vesicle transport (41). It seems possible that the high concentration of other G-actin binding proteins involved in actin polymerization at the cortical patches and at the growing ends of actin cables would displace YIH1 from G-actin. This would enable localized YIH1-GCN1 complex formation and inhibition of GCN2 activity and provide a mechanism for polarized up-regulation of protein synthesis in the buds.