Cdc123 and Checkpoint Forkhead Associated With RING Proteins Control the Cell Cycle By Controlling eIF2 γ Abundance*

, homolog of mammalian a new division cycle with a G2 delay at permissive temperature and a terminal, mating-proficient G1 arrest point. Cdc123 by and CHF2 , homologs of mammalian checkpoint forkhead associated with for G2 delay and G1 arrest of cdc123-4 and promote G1 delay when Cell cycle-delaying activity and natural instability of Chf1 and Chf2 integrity of both domains and with Cdc123 . Genetic analysis maps the Chf1 forkhead associated domain-binding site to the conserved Thr274 of Cdc123, suggesting that mammalian D123 is a key target of Chfr. Gcd11, the γ subunit of eIF2, is an additional Cdc123-interacting protein that is an essential target of the Cdc123 cell cycle-promoting and Chf cell-cycle arresting activity whose abundance is regulated by Cdc123, Chf1 and Chf2. Loss of cdc123 activity promotes Chf1 and Chf2 accumulation and Gcd11 depletion, accounting for the essentiality of Cdc123. The data establish the Cdc123-Chf-Gcd11 axis as an essential pathway for nutritional control of START that runs parallel to the Tor-Gcn2-Sui2 system of translational control. of Chf proteins, whose accumulation creates a G1 block that is similar to cdc123 loss of function. Because chf1 chf2 deletion suppressed the cell cycle phenotypes but not the essentiality of cdc123 , we searched for an essential cellular function regulated by Cdc123 and discovered that Gcd11 is another Cdc123-interacting protein whose abundance is regulated by Cdc123 and Chf proteins. Thus we show that nutritional deprivation leads to Cdc123 depletion, which results in accumulation of Chf proteins and depletion of Gcd11, an essential component of translational initiation factor 2, which is a central regulator of the transition between bulk translation and stress-associated translation (55). We also show that cdc123 arrest, which results in Gcd11 depletion, is fully compatible with mating but that mating pheromone does not act by depletion of Cdc123 protein. These experiments establish a Cdc123-Chf-Gcd11 axis of growth and division control functioning in parallel to the well characterized Tor-Gcn2-Sui2 axis of growth and division control (56). Ongoing studies are designed to elucidate the types of stress that lead to Chf-dependent G2 delay and G1 arrest and how these signals bear on eIF2-dependent choices of messages to translate.


Introduction
Eukaryotic cell division cycles consist of a number of dependent and independent gene-controlled sequences that are coordinated by gene-controlled checkpoints that ensure fidelity (1). After cell separation, early in the G1 phase of the S. cerevisiae cell cycle, several genes including those encoding Cdc25 (2,3), a guanine nucleotide exchange factor for Ras, and Cdc35 (4,5), encoding the Ras-stimulated adenylyl cyclase, and Cdc33 (6), encoding eukaryotic initiation factor 4E, function to detect nutrients and generate signals that result in increased protein synthesis necessary for commitment to a new cell cycle. Later in G1, at "START," mating pheromones arrest haploid cell cycles to promote conjugation at the arrest point of cdc28 (7), which encodes cyclin-dependent kinase 1, the highly regulated and highly conserved protein kinase required to drive entry into S phase and M phase in all eukaryotic systems examined (8)(9)(10)(11)(12)(13)(14). While nutrient limitation arrests yeast cells in early G1 prior to a great deal of protein synthesis and pheromone treatment arrests haploids at late G1, many of the connections between nutritional signaling and pheromone signaling and the potential for specific translational requirements in late G1 remain to be explored.
In this study we discover a new conserved cdc gene that is nutritionally regulated at the protein level whose mutants arrest, like cdc28 mutants, as large unbudded cells at the pheromone-sensitive stage in late G1. In addition, reminiscent of many cdc28 alleles and like cdc2 mutants in S. pombe and other systems, cdc123-4 mutants have a G2 delay phenotype in addition to a G1 block. We show that Cdc123 protein is depleted in nutritionally arrested cells but present in dividing cells and in pheromone-treated cells and that it functions to reduce the level of Chf1 and Chf2, the by guest on March 23, 2020 http://www.jbc.org/ Downloaded from two yeast homologs of human checkpoint forkhead associated (FHA) 1 with RING (Chfr) protein, which is required for G2 arrest in response to microtubule-destabilizing agents (15) and is frequently lost in cancer (16)(17)(18)(19)(20)(21). Our study provides a fine genetic mapping of the interaction between the FHA domain of Chf1 and the conserved Thr274 of Cdc123 and shows that this interaction is required for cell cycle delay by Chf1 and RING-dependent destruction of the checkpoint proteins themselves. This result is interesting not only in identification of a specific functional target of the Chfrhomologous FHA domain but also because the finding runs counter to the expectation that a RING protein bound to a phosphoprotein would tag the phosphoprotein for ubiquitin-mediated proteolysis.
Because deletion of chf1 and chf2 relieved the G2 cell cycle delay and G1 cell cycle block but not the essential function of cdc123, we looked for an essential target of Cdc123 regulation and discovered that Gcd11 (22)(23)(24), the γ subunit of eukaryotic initiation factor 2 (eIF2 γ), is an additional protein interactor of Cdc123. The data indicate that mating-proficient arrest of cdc123 is achieved by Chf-dependent alteration of eIF2γ abundance, thus connecting the nutritional and pheromone control of START with regulation of translational initiation.

Plasmid constructions
New plasmids are summarized Table I. All plasmids were made by cloning PCR products using restriction sites incorporated into the primer sequences indicated in Table II. Yeast genomic DNA was used as a template for PCR unless stated otherwise. The yeast CDC123 (YLR215C) gene was amplified with primers 4952 and 4953. The DNA fragment obtained in this reaction was cloned into vector pRS414 to create plasmid pB119. Plasmid pB121 was obtained by cloning of the BamHI-EcoRI fragment from pB119 into vector pRS416. The NdeI site was created in the pB119 plasmid by site-directed mutagenesis using primer 4960. This restriction site was used to insert a linker obtained by annealing and ligating oligonucleotides 5112 and 5113 to generate plasmid pB179 that expresses a 3xFLAG-Cdc123 fusion protein. Human D123 coding sequences were amplified from cDNA made from HeLa cells with primers 5106 and 5107. Plasmid pB174 was made by cloning of this PCR product into BamHI and XhoI sites of the p425GAL1 vector. Plasmid pB319 was obtained by cloning the CDC123 gene from pB179 into vector pRS416. C-terminal HA-tagged alleles of CHF1 and CHF2 were made by PCR with primers 7055 and 7236 (CHF1) or 7009 and 7235 (CHF2). The products of the yeast genomic DNA amplification with these pairs of primers were cloned into vector p425GAL1 and resulting plasmids were named respectively, pB313 and pB312. These plasmids were used to create RING domain mutants by site-directed mutagenesis. Plasmid pB312 and mutagenic primer 7129 were used to create plasmid pB332 that carries the CHF2 gene with two point mutations, C451S and H456A. Plasmid pB333, containing a chf1-C345S, H350A allele, was obtained by mutagenesis of the pB313 plasmid with mutagenic primer 7128. Primers 7055 and 7008 were used to amplify the CHF1 gene which was then cloned into p425GAL1 to generate plasmid pB299.
pB299 was used as a template for site directed mutagenesis using primer 7128 that resulted in plasmid pB331, carrying chf1-C345S, H350A. Plasmid pB323 contains the CHF2 gene cloned as a PCR product obtained with primers 7003 and 7005 in vector p425GAL1. This plasmid and primer 7129 were used to generate plasmid pB334 containing the chf2-C451S, H456A allele. The G192E substitution in the FHA domain of CHF1 was generated from plasmids pB299 and pB313 with primer 7110 resulting in plasmids pB403 and pB405. Plasmids pB404 and pB406, containing the chf1-S220A, H223L allele were obtained by mutagenesis of plasmids pB299 and pB313 with primer 7111. chf1 alleles were amplified from plasmids pB331, pB403 and pB404 by PCR with primers 7008 and 7010 and products of these reactions were cloned into plasmid pACT2 and named pB396, pB408 and pB409, respectively. Plasmid pB310 was obtained by cloning of the PCR fragment generated with primers 7008 and 7010, containing wild-type CHF1, into the pACT2 vector. To make plasmid pB271, CDC123 was amplified with primers 7073 and 7074 and cloned into vector pOBD22. Plasmid pB424 carrying the 3XFLAG-cdc123-T274A allele was made using mutagenic primer 7303 and single stranded DNA obtained from plasmid pB179. The tag-free cdc123 allele of plasmid pB424 was amplified with primers 7073 and 7074 and cloned into plasmid pOBD22. This plasmid was named pB426 and used for twohybrid assays. The HA-tagged allele of GCD11 was created by a two-step PCR reaction. In the first step, the fragment of yeast genomic DNA was amplified with primers 7796 and 7797. The product of this reaction was re-amplified with primers 7796 and 7798 and cloned between SacI and XhoI sites of plasmid pRS416. The resulting plasmid, pB463, contains the GCD11 promoter controlling expression of GCD11 with two C-terminal HA epitopes. To create plasmid pB232, the GCD11 open reading frame was amplified using primers 7830 and 7831. The product of this reaction was cloned in plasmid p425GAL1 using Bam HI and XhoI restriction sites incorporated in the primers. Plasmid pB295 carrying CDC123 gene under GAL1 promoter control was created by cloning of the PCR product generated with primers 7832 and 7833 between Bam HI and XhoI sites of the plasmid p425GAL1. Plasmid pB350 was made by replacing the GAL1 promoter in plasmid pB313 with the CHF1promoter amplified with primers 7045 and 7046. Replacement of the GAL1 promoter in plasmid pB312 with the PCR product generated in a reaction with primers 7047 and 7048 yielded plasmid pB351 that contains the CHF2 gene controlled by its native promoter.

Construction of yeast strains
Gene disruptions were made by direct transformation of the appropriate yeast strains with the PCR products (25). Plasmid pRS400 was used as a template for PCR amplification of the geneticin resistance marker with primers 4949 and 4950. The cdc123-disrupting product from this reaction was used to transform diploid S. cerevisiae strain SEY6210.5. Integration of the marker into the YLR215C locus was confirmed by PCR with primers 4948 and 4951. A resulting cdc123-heterozygous strain BY116 was transformed with the plasmids pB121, pB174 or pB179. Transformants carrying these plasmids were induced to sporulate, and resulting tetrads were dissected using a micromanipulator (26 The GCD11 open reading frame was disrupted in strain BY116 (heterozygous for cdc123::kanMX) by integrative transformation with a DNA fragment obtained by amplification of HIS3 marker from the plasmid pRS413 with primers 7799 and 7800. Correct integration of the marker into the GCD11 locus was confirmed by PCR with primers 7801 and 7802. The resulting strain was transformed with plasmids pB464 and pB463. Selected double transformants were subjected to sporulation and tetrad analysis. This procedure yielded isolates named BY290-1a and BY290-14a with both cdc123 and gcd11 genes disrupted, carrying plasmids pB464 and pB463. Two other strains generated in this procedure, BY290-3a and BY290-4d, are wild type at the CDC123 locus and carry a gcd11::HIS3 mutation covered by plasmid pB463.

Two-hybrid screening and β-galactosidase assays
Two-hybrid screening was performed as described using plasmid pACT2, a yeast cDNA library in lambdaACT2, and yeast strain Y190 (27,28). The bait plasmid was created by cloning the CDC123 open reading frame in plasmid pOBD (29). To create prey plasmids for β-galactosidase assays, wild-type and mutant CHF1 alleles were cloned into plasmid pACT2. Y190 cells transformed with pairs of bait and prey plasmids were grown on selective liquid media to mid-exponential phase. For the assays of β-galactosidase activity, equal numbers of exponentially growing cells were harvested. Assays were performed using 2nitrophenyl β-D-galactopyranoside as a substrate and normalized to units/min/ml/OD according to the methods of Guarente (30). Results are presented as a percentage of the activity in the wild type CDC123-wild type CHF1 sample, which, at 14.4 units/min/ml/OD, was the pair of highest interaction.

Generation of cdc123 temperature sensitive alleles
To generate a library of cdc123 mutants, the gene was used as a template for error-prone PCR using 100µM dATP, dGTP and dTTP and 1 mM dCTP, and amplified products were cloned into vector pRS414. The library was selected for complementation at 28 °C and transformants were screened for temperature-sensitivity at 37 °C as described in Results.

Flow cytometric analysis
DNA was stained with SYBR Green I (Molecular Probes) and flow cytometry was performed using a Beckman Coulter FACScan instrument as described (31).

Western and co-immunoprecipitation analysis
A FLAG epitope was inserted at the N-terminus of Cdc123. HA epitopes were added to C-termini of Chf1, Chf2 and Gcd11. Strains expressing FLAG-CDC123 and GAL1 driven CHF-HA alleles were used to assay Cdc123-Chf interactions. A strain expressing FLAG-CDC123 and GCD11-HA from native promoters was used to test Cdc123-Gcd11 association. Whole cell extracts were prepared by glass bead lysis. For Western analysis, anti-HA and anti-FLAG antibodies and secondary rabbit anti-mouse antibodies conjugated to horseradish peroxidase were from Sigma. For coimmunoprecipitation, lysates were incubated with anti-HA antibodies immobilized on

CDC123 is the Functional and Structural Homolog of Mammalian D123
The human D123 coding sequence was first isolated from a cDNA expression library as a suppressor of the temperature-sensitive mutation in rat cell line 3Y1tsD123, which arrests in G1 phase at restrictive temperature (32). The mutant D123 protein has an Ala109Val alteration in sequence (32) and was reported to result in a polypeptide that is degraded at the restrictive temperature (33) in a lactacystin-sensitive manner (34).
The D123 gene is expressed in the cytosol of human and rat tissues, with expression in testis being highest (35).
Comparison of human D123 protein with sequences deposited in GenBank revealed apparent orthologs in animals, fungi, plants and protists. An alignment of D123 proteins from human, S. cerevisiae, S. pombe, A. thaliana and D. discoideum is presented in Fig. 1. To study function of the 35% identical S. cerevisiae protein, we made a geneticin-resistant disruption of the YLR215C gene (hereafter CDC123), in yeast diploid cells to generate strain BY116, heterozygous at cdc123. After meiosis, sporulation and dissection of spores, we found that haploid isolates segregated 2:2 for viability and none of the viable isolates were geneticin resistant ( Fig. 2A). Examination of the nongrowing isolates revealed micro-colonies of approximately 8 to 16 cells, suggesting, as in the case of CAK1 (36,37), that CDC123 encodes an abundant and/or long-lived essential protein that is depleted in mutant cells after three to four divisions.
To exclude the possibility that the gene is required for spore germination and is later dispensable, a human D123 cDNA from a HeLa cells was amplified and cloned into a yeast vector under control of the GAL1 promoter. Dissection of the BY116 strain transformed with this plasmid yielded tetrads with four viable progeny (Fig. 2B). In these tetrads, geneticin resistance segregated 2:2 and geneticin resistant colonies exhibited galactose-dependent growth (Fig. 2C). Additionally, URA3-based plasmid pB121 constructed to express yeast CDC123 from its own promoter, introduced into heterozygous diploids and recovered in cdc123Δ haploids, could not be cured though the same plasmid could be easily lost from CDC123 wild-type cells (data not shown).
These results show that the CDC123 gene is essential for viability and that the mammalian D123 gene is a structural and functional homolog of CDC123.

Mutation of cdc123 Results in G2 Delay and G1 Cell Cycle Arrest
To determine the essential function or functions of CDC123, the products of errorprone PCR amplification of the gene were cloned in a yeast vector with a TRP1selectable marker to generate a mutagenized library. This library was transformed into the cdc123Δ haploid strain carrying plasmid pB121. Transformants were selected on 5fluoroorotic acid to select for pB121 plasmid loss (38), indicating complementation of cdc123Δ at 28 °C, and then screened for temperature sensitivity at 37 °C. This procedure yielded a number of temperature sensitive alleles. The sequence of the cdc123-4 allele was determined and its phenotypes studied. As shown in Fig. 1, conceptual translation of cdc123-4 indicates eight amino acid substitutions, two altering the strictly conserved residues Asp55 and Glu106.
As shown in Fig. 3A, when analyzed by flow cytometry at the permissive temperature of 28 °C, cdc123-4 cells have an excess of G2 cells compared to isogenic CDC123 wild-type cells and the overall generation time is slowed (data not shown).
Upon shifting to the nonpermissive temperature of 37 °C, G1 cells accumulate, though the terminal phenotype of large G1-arrested cells takes approximately 20 hours to achieve (Fig. 3B). Consistent with the observation that cdc123Δ spores from a CDC123 / cdc123Δ tetrad dissection divide three to four times before dying, these data suggest that CDC123 encodes an essential and abundant and/or relatively stable protein.
Moreover, despite the partial G2 block that is seen at the cdc123-4 permissive temperature, the ability for cdc123-4 temperature-shifted cells to accumulate in G1 indicates that the most stringent requirement for the Cdc123 polypeptide is at G1 cell cycle entry.
Cell division cycle mutants with G1 phenotypes have been put into two classes on the basis of whether they arrest as small, unbudded cells prior to sensitivity to mating pheromone, or, as is the case of the classic START mutant cdc28, concomitant with the pheromone stage (7,39). The large size of arrested cdc123-4 mutants suggested a late G1 block but, as shown in Fig. 4A, temperature-arrested cells released into media with α-mating pheromone at 28 °C formed shmoos rather than proceeding to bud emergence and shmooing after the next cell division. This experiment indicates that the cdc123-4 arrest point is prior to or at the mating pheromone-sensitive point in the cell division cycle. To distinguish between these possibilities, cells arrested at the nonpermissive temperature were treated with pheromone. As shown in Fig. 4B, these arrested cells shmoo upon addition of pheromone. Additionally, as shown in Fig. 4C, arrested cells mate at the nonpermissive temperature. Thus, the data indicate that Cdc123 is a previously unrecognized component of START.

Nutritional Control of Cdc123 Protein Expression
To follow expression of Cdc123 protein, we constructed a plasmid to express Cdc123 with an N-terminal FLAG-tag and introduced this plasmid into a CDC123 / cdc123Δ heterozygous strain. Recovery of viable haploid cells bearing the geneticin-resistance marker after sporulation and tetrad dissection confirmed that FLAG-tagged Cdc123

Proteins
To identify additional components of the cell cycle machinery controlled by Cdc123, we performed a two-hybrid screen and identified both yeast homologs of human checkpoint forkhead associated with RING protein, Chfr. Chfr has been described as a mitotic checkpoint protein (15) whose RING domain is an E3 ubiquitin ligase that reportedly targets Polo-like kinase 1 for destruction (40) and is required for inhibiting chromosome condensation in response to Taxol treatment (41). However, the mechanism of Chfr function in mitotic checkpoint function has recently been questioned with the observation that Chfr participates with the Ubc13-Mms2 heterodimer to lead to auto-polyubiquitination via Lys63-linkages, a type of modification that does not target proteins to destruction but rather functions in stress signaling (42).
The two yeast homologs of Chfr, YHR115C and YNL116W, here named Chf1 and Chf2, are 46% identical to each other at the amino acid sequence level and share the domain structure of Chfr. As shown in Fig. 6A, the two characteristic features are an N-terminal forkhead associated (FHA) domain, typically a module for binding Thrphosphorylated proteins found in proteins in cell cycle arrest and/or DNA repair proteins (43,44), and a C-terminal RING domain, which is thought to be the signature sequence for E3 ubiquitin ligases (45)(46)(47)(48)(49)(50). Indeed the crystal structure of the FHA domain of human Chfr bound to tungstate is consistent with function as a phosphoreceptor (51) though the target of this FHA domain had not been identified. Similarly, the RING domain of Chfr has been investigated from the earliest observations of loss of Chfr in human cancer (15,(40)(41)(42) though the means by which the RING domain mediates a checkpoint function remains incompletely understood.
We made two site-directed mutants in the CHF1 gene to reduce function of the FHA domain and constructed site-directed alleles of both CHF1and CHF2 to inactivate the RING domains. As shown in Fig. 6B, assaying the two-hybrid interaction as measured by a β-galactosidase assay, the chf1-G192E substitution, which was targeted to an absolutely conserved Gly in Box A of FHA domains (51) shown to be functionally important in the kinase-associated protein phosphatase (52), reduced the interaction by more than four-fold, while the double mutation of chf1-S220A,H223L targeted to the FHA domain (52) reduced the β-galactosidase signal by more than 100-fold. When two putative zinc-ligating residues of the RING domain were targeted by the chf1-C345S,H350A allele, a six-fold reduction in two-hybrid signal with CDC123 was observed. Obviously, two-hybrid signals could have arisen from a variety of artifactual phenomena and reductions in two-hybrid signals could indicate reductions in protein stability or other unanticipated effects of the mutations. To test whether wild-type or mutant Chf proteins form physical complexes with Cdc123 whose stability is affected by Chf protein sequences, we expressed HA-tagged CHF genes with the GAL1promoter in a yeast strain in which FLAG-tagged Cdc123 replaces untagged Cdc123. As shown in Fig. 6C, by probing HA immunoprecipitates with an anti-FLAG antibody, we demonstrated that both wild-type Chf1 and Chf2 proteins are physically associated with Cdc123 in vivo. Curiously, however, this assay did not detect a reduction in association of the RING domain alleles, Chf1-C345S,H350A and Chf2-C451S,H456A, in association with Cdc123 though the former allele was shown to reduce the two-hybrid interaction. To explore whether the disparity between twohybrid and immunoprecipitation might be due to differences in protein stability, we simply probed total protein in HA-Chf1 and HA-Chf2 transformants for the abundance of wild-type and mutant Chf proteins. Remarkably, as shown in Fig. 6D

Domains and Conserved Thr of Cdc123
The physiological effects of Chf protein overexpression were explored. As shown in  (43). We therefore scanned Cdc123 homologs (Fig. 1) for conserved T-X-X-D or T-X-X-φ (φ, a hydrophobic amino acid) motifs, and considered that Thr274 was the most likely site of a putative direct association with the FHA domains of Chf proteins.
As shown in Fig. 7C, we constructed a strain carrying cdc123-T274A as the sole source of CDC123 and discovered this to be a hypomorphic allele with a slow growth phenotype. As shown in Fig. 5B, by quantitative β-galactosidase analysis of constructed two-hybrid partners, the Cdc123-T274A allele has a reduced interaction with wild-type and the RING domain mutant of Chf1. Consistent with the idea that

Deletion of chf1 and chf2 Suppresses the cdc Phenotype But not the Essentiality of cdc123
To explore further CHF1 and CHF2-dependent processes, we prepared a chf1 chf2 deletion strain and a strain in which cdc123-4 was accompanied by chf1 chf2 deletion.
The chf1 chf2 deletion was not inviable and the loss of chf genes neither aggravated nor abrogated the temperature-sensitivity of cdc123-4 (data not shown). As shown in Fig.   9, flow cytometric analysis of cdc123-4 versus cdc123-4 chf1Δ chf2Δ strains indicates that while the latter strain is as temperature sensitive as the former, deletion of chf1 and chf2 reduces the G2 peak of cdc123-4 cells grown at the permissive temperature and eliminates the G1 arrest of cdc123-4 cells shifted to the nonpermissive temperature.
Thus, Chf proteins, which are physically stabilized by loss of cdc123 function, enforce the G2 and G1 checkpoints associated with loss of cdc123 function.

The eIF2 Complex is an Essential Target of Chf-Cdc123 Regulation
Though chf1 chf2 deletion makes cdc123 arrest randomly rather than at G1, the checkpoint deletion does not relieve the essential function of CDC123, suggesting that there are other targets of Cdc123 regulation. Additionally, though Chf proteins apparently make use of a FHA domain-dependent interaction with Cdc123 and a RING and FHA-domain dependent destabilization of the Chf proteins themselves, the residual toxicity of Chf protein mutants suggests that there may be an additional target of the Chf protein RING domains. One additional gene, GCD11, which encodes the eukaryotic initiation factor 2 γ subunit, was obtained by two-hybrid screening with Cdc123 as the bait. eIF2 consists in yeast of the Sui2-Sui3-Gcd11 α−β−γ heterotrimeric complex. This complex delivers aminoacylated tRNA Met to the small ribosomal subunit with the γ subunit bound to GTP (53), the β subunit mediating some of the interactions with mRNA (54), and the α subunit as the key target of eIF2 kinases (55).
As shown in Fig. 10A, HA-tagged Gcd11 can be expressed in place of the essential GCD11 gene and when it is immunoprecipitated, FLAG-tagged Cdc123 is specifically detected in the Gcd11-associated fraction. To see if the level of Gcd11 protein is affected by Cdc123 function, we expressed HA-tagged Gcd11 as the sole source of Gcd11 in a cdc123-4 strain and in an isogenic CDC123 wild-type strain. As shown in Fig. 10B, the Gcd11 protein level is not altered by temperature shift in a wildtype strain but shifting cdc123-4 to the nonpermissive temperature leads to loss of Gcd11 protein. As the cycle cycle kinetics of Chf overexpression were shown to be similar to cdc123 loss of function and, indeed, cdc123 temperature shift leads to Chf protein stabilization, we wished to test whether Chf1 and Chf2 overexpression might destabilize Gcd11. As shown in Fig. 10C, GAL1-driven overexpression of Chf1 renders Gcd11 virtually undetectable under the conditions in which a specific, HA-tagged Gcd11 doublet is observed in wild-type cells. Moreover, GAL1-driven overexpression of Cdc123 also reduces the amount of steady-state Gcd11 protein. Surprisingly, GAL1driven overexpression of Chf2, which is almost as toxic as GAL1-driven expression of Chf1, has the opposite effect on eIF2γ abundance, increasing the levels of Gcd11 with respect to wild-type. Because the GCD11 gene is essential (22)(23)(24), reducing the level of the Gcd11 polypeptide is an obvious mechanism for G1 arrest, precedented by arrest of cdc33, encoding mRNA cap-binding eIF4E (6). The fact that too much or too little Cdc123 function depresses Gcd11 levels while Chf2 overexpression boosts Gcd11 levels prompted us to test whether Cdc123 or Gcd11 overexpression are toxic. As shown in Fig. 10D, both of these conditions are strongly deleterious, indicating that having too much or too little eIF2γ is growth-inhibitory.

Discussion
Studies with a mammalian d123-mutant cell line suggested the likely essentiality of the D123 function but no analysis was available to identify the pathways by which D123 acts. In this study, we identify CDC123 as the yeast homolog of mammalian D123 and characterize its functions in cell cycle progression as a molecule positively regulated by nutrients, as a negative regulator of yeast homologs of Chfr, and as a regulator of eIF2γ abundance. Though Chfr is frequently inactivated in human cancer (15)(16)(17)(18)(19)(20)(21), the mechanisms of Chfr function, including the target of its FHA domain and the function of its RING domain were not clear.
As summarized in the model presented in Fig. 11, Cdc123 levels are maintained by the availability of nutrients and this polypeptide, in turn, controls the abundance of the two checkpoint forkhead associated with RING proteins, Chf1 and Chf2, and the γ subunit of eIF2. Our analysis indicates that Chf proteins are normally inabundant polypeptides whose turnover depends on FHA domain-dependent association with Cdc123 and the integrity of their own RING domains. Interestingly, though Cdc123 levels are maintained by nutrients and Cdc123 interacts with Chfr homologs via an interaction of Cdc123-Thr274 and the Chf FHA domains, this interaction does not destabilize Cdc123 but, rather, leads to RING-dependent destabilization of the Chf proteins themselves. Moreover, toxicity of Chf expression is dependent on the integrity of the FHA domain and the RING domain and, for Chf1, this toxicity was shown to depend on the conserved Thr274 of Cdc123, the apparent target of the Chf protein FHA domains. As shown by flow cytometry, in the absence of Chf1 and Chf2, the cdc123-dependent cell cycle delay at G2 and the G1 arrest are not maintained.
Our data indicate that a critical function of Cdc123 is to direct the destabilization of Chf proteins, whose accumulation creates a G1 block that is similar to cdc123 loss of function. Because chf1 chf2 deletion suppressed the cell cycle phenotypes but not the essentiality of cdc123, we searched for an essential cellular function regulated by Cdc123 and discovered that Gcd11 is another Cdc123-interacting protein whose abundance is regulated by Cdc123 and Chf proteins. Thus we show that nutritional deprivation leads to Cdc123 depletion, which results in accumulation of Chf proteins and depletion of Gcd11, an essential component of translational initiation factor 2, which is a central regulator of the transition between bulk translation and stressassociated translation (55). We also show that cdc123 arrest, which results in Gcd11 depletion, is fully compatible with mating but that mating pheromone does not act by depletion of Cdc123 protein. These experiments establish a Cdc123-Chf-Gcd11 axis of growth and division control functioning in parallel to the well characterized Tor-Gcn2-Sui2 axis of growth and division control (56). Ongoing studies are designed to elucidate the types of stress that lead to Chf-dependent G2 delay and G1 arrest and how these signals bear on eIF2-dependent choices of messages to translate.  Three independent isolates exhibited galactose-dependent growth.     Overexpression of wild type CHF1 and, to a lesser degree, CHF2 causes a slow growth phenotype. Though the FHA domain and RING domain mutants of Chf1 accumulate greatly, they exhibit less toxicity. (C) Cells expressing wild type CDC123 or the hypomorphic cdc123-T274A allele were transformed with empty vector or a plasmid overexpressing CHF1. Chf1 toxicity largely depends on Thr274 of Cdc123 and lack of Thr274 of Cdc123 provides a growth advantage in cells overexpressing Chf1.    FHA domain, may also target another protein's destruction. Gcd11, which is also complexed with Cdc123 in vivo, is degraded with temperature-dependent Cdc123 inactivation and Chf1 accumulation.