The Cdc34 ubiquitin-conjugating enzyme plays a central role in progression of the cell cycle. Through analysis of the phenotype of a mutant missing a highly conserved sequence motif within the catalytic domain of Cdc34, we discovered previously unrecognized levels of regulation of the Ace2 transcription factor and the cyclin-dependent protein kinase inhibitor Sic1. In cells carrying the Cdc34tm mutation, which alters the conserved sequence, the cyclin-dependent protein kinase inhibitor Sic1, an SCFCdc4 substrate, has a shorter half-life, while the cyclin Cln1, an SCFGrr1 substrate, has a longer half-life than in wild-type cells. Expression of the SIC1 gene cluster, which is regulated by Swi5 and Ace2 transcription factors, is induced in CDC34tm cells. Levels of Swi5, Ace2, and the SCFGrr1 targets Cln1 and Cln2 are elevated in Cdc34tm cells, and loss of Grr1 causes an increase in Ace2 levels. Sic1 levels are similar in CDC34tm ace2Δ and wild-type cells, explaining a paradoxical increase in the steady-state level of Sic1 protein despite its reduced half-life. A screen for mutations that interact with CDC34tm uncovered novel regulators of Sic1, including genes encoding the polyubiquitin chain receptors Rad23 and Rpn10.
UBIQUITIN is a small, 76-amino-acid residue protein. The covalent attachment of ubiquitin to another protein often serves as a signal for the selective degradation of that protein by the 26S proteasome (for review, see Glickman and Ciechanover 2002). Ubiquitylation is initiated by the ubiquitin-activating (or E1) enzyme, forming a high-energy thiolester intermediate with ubiquitin in an ATP-dependent reaction. The E1 transfers the ubiquitin molecule to a ubiquitin-conjugating (or E2) enzyme, which, like the E1, forms a thiolester with ubiquitin. The E2 then transfers ubiquitin onto a substrate. The transfer of ubiquitin to substrate typically requires the activity of a ubiquitin ligase (E3). The bond formed between ubiquitin and substrate is an isopeptide bond that links the COOH-terminal glycine residue of ubiquitin to the ε-amino group of a lysine residue of the substrate. Substrates may be monoubiquitylated or polyubiquitylated with the polyubiquitin chain being linked through one of the seven lysine residues within ubiquitin.
CDC34 was identified as a gene essential for cell viability and the initiation of DNA replication in the yeast Saccharomyces cerevisiae (Goebl et al. 1988). The Cdc34 protein is an E2 that ubiquitylates histones in vitro (Goebl et al. 1988) and itself in vitro (Banerjee et al. 1993) and in vivo (Goebl et al. 1994). Furthermore, Cdc34 genetically and physically interacts with the SCF family of ubiquitin ligases, and this interaction is a requirement for Cdc34 to carry out its ubiquitin-conjugating function in vivo (Mathias et al. 1998). Members of the SCF family of ubiquitin ligases are composed of at least four distinct proteins, including Skp1, Cdc53, Rbx1, and a member of a family of proteins known as F-box proteins. Rbx1 contains a ring-H2-finger domain and is essential for SCF-dependent attachment of ubiquitin to its substrates (Skowyra et al. 1999). Cdc53 is a cullin family member that binds both Skp1 and Cdc34. Skp1 tethers the F-box protein to the SCF complex. The F-box protein is the component of the SCF ligase that determines substrate specificity (for review, see Deshaies 1999). Cdc34 and SCFCdc4 form an E2–E3 complex that mediates conjugation of lysine 48-linked polyubiquitin chains to the cyclin-dependent kinase inhibitor Sic1 as well the cell cycle regulator Far1 (Henchoz et al. 1997; Skowyra et al. 1997; Verma et al. 1997). On the other hand, SCFGrr1 is required for ubiquitylation of the cyclins Cln1 and Cln2 (Skowyra et al. 1997, 1999; Verma et al. 1997).
The Cdc34/Ubc7 family of ubiquitin-conjugating enzymes is defined by a conserved motif within the catalytic domain. This motif allows the Cdc34/Ubc7 family to catalyze both monoubiquitylation and ubiquitin chain extension (Petroski and Deshaies 2005). This motif is defined by two serines and a 12-amino-acid acidic loop, all of which lie in close physical proximity to the catalytic cysteine (Figure 1). In contrast, the majority of E2s, of which Rad6 is a classic example, have a lysine and aspartic acid residue in lieu of the serine residues and lack the acidic loop. Cdc34 mutants lacking the acidic loop monoubiquitylate Sic1 with the same kinetics as the wild-type enzyme but extend ubiquitin chains at a negligible rate in vitro (Petroski and Deshaies 2005). Cells expressing Cdc34 mutants that lack the acidic loop are inviable as are cells harboring S97D or S73K/S97 mutations (Liu et al. 1995). More recently, Li et al. (2007) found the acidic loop of the Cdc34-like ubiquitin-conjugating enzyme Ube2g1 to be essential for polyubiquitin chain preformation and conjugation to the known in vivo target HERPc, a short-lived, ER-associated protein. Interestingly, the polyubiquitin chain preformation and ubiquitin conjugation to HERPc in vitro require the presence of a RING-finger-containing protein, much like Cdc34 enzyme activity in vivo. During in vitro Cdc34 autoubiquitylation or histone ubiquitylation assays that do not require RING-finger-containing proteins, acidic loop deletion mutants function as well as, if not better than, Cdc34 (Pitluk et al. 1995; Varelas et al. 2003). Deletion of the acidic loop, residues 103–114, in combination with S73K and S97D mutations (hereafter referred to as the Cdc34 triple mutant, Cdc34tm), restores full functionality to Cdc34 in all in vitro assays tested to date and in vivo as assessed by cell growth (Liu et al. 1995; Pitluk et al. 1995; Varelas et al. 2003). The goal of this study was to determine if processes and substrates controlled by Cdc34 are appropriately regulated in cells lacking this Cdc34 motif critical for polyubiquitin chain extension.
MATERIALS AND METHODS
Standard methods were used for strain construction (Rose et al. 1990). Strains RC29, RRC73, RRC74, RRC76, and RRC78 containing the CDC34tm allele flanked by the nourseothricin N-acetyltransferase gene were constructed as follows (Table 1). The plasmid pAG25 that contains the nourseothricin N-acetyltransferase gene (Goldstein and McCusker 1999) was amplified with adaptamer primers CDC34F2 (5′-ACTTTTTTCAAGGCTGAGAATCCATCGACAGATTGTAACGAAGCAGCTGAAGCTTCGTACGC-3′) and CDC34R2 (5′-TGCTCTGTATAGTTCAATAGAATCTTACAGTACATCACGCTGCAAGCATAGGCCACTAGTGGATCTG-3′) using the PCR cocktail and conditions described previously (Goldstein and McCusker 1999). The PCR product was transformed into the CDC34tm-containing strain KS418 as previously described (Gietz and Woods 2002). Transformants were selected by plating cells on YPD plus 80 mg/liter of nourseothricin (Werner BioAgents, Jena, Germany). Insertion into the correct chromosomal locus was confirmed by PCR using the primers 34F2t (5′-CAAACTTGAGATGGAGTTGTTGATG-3′) and pAG25Tr1 (5′-GTCAATCGTATGTGAATGCT-3′). This strain was named RC6. DNA containing the CDC34tm allele and the nat1 gene was amplified from RC6 genomic DNA with Phusion DNA polymerase (Finnzymes, Espoo, Finland) according to the manufacturer's instructions using primers 34F2t and 34R3 (5′-ATGAGTAGTCGCAAAAGCACCG-3′). For strains RRC73, RRC74, RRC76, and RRC78, the PCR product was then transformed into the appropriate BY4741 strain containing CLN1-TAP, CLN2-TAP, SIC1-TAP, or FAR1-TAP at the chromosomal locus as described (Ghaemmaghami et al. 2003). Transformation was carried out following the protocol of Johnston et al. (2002) for gene disruptions generated from a yeast genomic template. Transformants were selected by growth on YPD plus 80 mg/liter of nourseothricin.
RC29, the query strain for the synthetic gene array (SGA) screen, was made by a cross of RC6 and MT1901 to create the diploid RC21, which was sporulated and its tetrads dissected to isolate a haploid with the appropriate markers. RC94, the control strain for the secondary SGA screen, was constructed by insertion of the nourseothricin N-acetyltransferase gene into strain MT1901 at the exact chromosomal location as strain RC6. The PCR conditions and transformation were carried out as described for construction of RC6.
RC200 is a haploid made by crossing RC34 and BY4741 [SWI5-TAP(HIS3-MX6)]. Tetrads from the resulting diploid were dissected. RC34 was generated by dissection of the RC26 diploid, which resulted from a cross of strains RC6 and ResGen 11820 from the Matα deletion library.
Media and growth conditions:
Standard rich (YPD) and defined minimal (SD) media were prepared as described previously (Rose et al. 1990). For analysis of Far1 and Cln1 abundance and half-lives, cells were grown in YPD plus 30 mm succinic acid. Standard sporulation and dissection procedures were used as described previously (Rose et al. 1990).
For sulfite sensitivity assays, 2 mm of sodium sulfite and 75 mm of tartaric acid were added to YPD media as previously described (Avram and Bakalinsky 1996).
Cells were grown to a final density between 5 × 106 and 1.5 × 107 cells/ml. For the blots in Figure 3 and Figure 4A, cells were lysed by repeated cycles of vortex mixing with glass beads in breaking buffer containing 150 mm NaCl, 50 mm Tris–HCl, pH 7.5, 5 mm EDTA, 1% Triton X-100, 50 mm NaF, and one Complete Mini protease inhibitor cocktail tablet (Roche Diagnostics, Mannheim, Germany) per 10 ml of breaking buffer. For Figure 4B, protein extract was made by the Horvath–Riezman method as previously described (Horvath and Riezman 1994). Briefly, 3.75 μl of extraction buffer [60 mm Tris, pH 6.8, 10% glycerol, 2% SDS, 5% β-mercaptoethanol, 5 mm EDTA, 50 mm NaF plus one Complete Mini protease inhibitor cocktail tablet (Roche Diagnostics) per 10 ml of extraction buffer] were added per milligram of wet cell pellet. Cells were resuspended in extraction buffer and boiled for 5 min. This mixture was then centrifuged, and the supernatant was saved as protein extract.
Lysates (∼50 μg of protein) from either extraction method were loaded onto a 10% SDS-PAGE gel. Separated proteins were transferred to PVDF membrane. Western immunoblot analysis was performed by standard methods. Antibodies were used at the following dilutions: affinity-purified anti-Cdc34 (1:10,000), affinity-purified anti-Sic1 (1:1000), Ace2 antisera (1:5000) (Sbia et al. 2008), and anti-TAP (1:2000) (Open Biosystems). Primary antibody was detected with an HRP-conjugated goat anti-rabbit secondary antibody at a 1:10,000 dilution (Santa Cruz Biotechnology).
For measurement of Sic1 and Cln1 half-lives, strains Sic1-Tap, Cln1-Tap, RRC73, and RRC74 were inoculated into 150 ml of YPD + 30 mm succinic acid (pH 4.5) grown at 30° until mid-log phase at which time cycloheximide (0.42 mg/ml final concentration) was added. Thirty-milliliter samples were collected at the indicated time points and immediately frozen in liquid nitrogen. Protein extraction was accomplished as described in Western blotting. Immunoreactivity quantification was accomplished using Quantity One (BioRad) software, where the ratio of the adjusted volume of the Sic1 or Cln1 band was normalized to the adjusted volume of the Cdc34 band.
A CDC34tm strain (RRC85) and an isogenic wild-type strain (DBY2059) were inoculated from stationary-phase cultures into a synthetic defined medium containing 2% dextrose, 0.17% yeast nitrogen base minus amino acids and ammonium sulfate, 0.25% l-glutamine, 0.025% magnesium sulfate, and 25.2 mg/liter l-leucine. Four separate cultures of each strain were grown at 30°, allowing between three and four doublings, and cells were collected at ∼8 × 106 cells/ml. Cells were centrifuged and immediately frozen in liquid nitrogen. Total RNA was extracted using a hot acid phenol-chloroform protocol as previously described (Schmitt et al. 1990). RNA quality was verified with OD260/280 readings and agarose gel electrophoresis (1.5%). An Affymetrix Genechip Scanner 3000 and Affymetrix Yeast 2.0 arrays were used for the microarray. A Student's t-test of the log base 2-transformed data was used to establish P-values for Table 2 and Table 3.
Synthetic gene array:
Duplicate SGA screens of the CDC34tm (nat1) query strain, RC29, were performed against the ∼4700 single, nonessential gene deletion strains as previously described (Tong et al. 2001). Computer-based image analysis of colony size allowed genetic interactions to be scored on a scale of 1–4 with 1 = lethal, 2 = severely sick, 3 = sick, and 4 = mildly sick. Candidate synthetic genetic interactions were validated by a second SGA screen, which was done manually. Strains RC29 and RC94 were separately crossed to strains with the candidate gene deletion. RC94 is isogenic to RC29 except the nat1 gene marks wild-type CDC34. Diploids were selected by replica-plating onto YPD + 100 mg/liter nourseothricin + 200 mg/liter G418 and then isolated by streaking for single-cell colonies on the same media. The isolated diploids were sporulated on sporulation media (1% potassium acetate, 2% agar, 0.1% yeast extract, 0.05% glucose supplemented with 1 mm uracil, l-histidine, and l-leucine). Diploids were sporulated at room temperature for 7 days and then struck onto haploid, double-mutant selection media (2% glucose, 0.17% yeast nitrogen base lacking ammonium sulfate and amino acids, 0.1% monosodium l-glutamic acid, 0.2% amino acid mix lacking l-histidine and l-arginine, 50 mg/liter canavanine, 100 mg/liter nourseothricin, 200 mg/liter G418, and 2% agar). Each CDC34tm(nat1) xxxΔ∷KanR genetic interaction was scored visually after 3 days growth at 30° by comparing its growth to that of the CDC34(nat1) xxxΔ∷KanR haploid.
To determine if deletion of SIC1, ACE2, or SWI5 was capable of suppressing the lethality of certain double mutants, we replaced each gene with the Kluyveromyces lactis URA3 gene in a diploid strain that was heterozygous for the gene deletion of interest (for example, RAD23) and heterozygous for the CDC34tm allele. The primers used were as follows: SIC1 (CCACCAAGGTCCAGAGGGACTAGGTACCTTACTAGTGGATCTGATATCAC and GCTCTTGATCCCTAGATTGAAACAATGCCTCGATTTAGGTGACACTAT), ACE2 (TGGATAACGTTGTAGATCCGTGGTATATAAACTAGTGGATCTGATATCAC and TCAGAGAGCATCAGTTTCGTTTGAAAGGGTGCGATTTAGGTGACACTAT), and SWI5 (GATGGATACATCAAACTCTTGGTTTGATGCACTAGTGGATCTGATATCAC and CTTTGATTAGTTTTCATTGGCGAAACCATAC-GATTTAGGTGACACTAT). The plasmid pUG72 (Gueldener et al. 2002), which contains the K. lactis URA3 gene, was amplified with adaptamers for SIC1, ACE2, or SWI5 listed above using the PCR cocktail and conditions described previously (Goldstein and McCusker 1999). The PCR product was transformed into the desired diploids as previously described (Gietz and Woods 2002). Transformants were isolated and patched onto YPD and grown for 24 hr after which cells were replica-plated onto the sporulation media described above. Diploids were sporulated at room temperature for 7 days and then struck onto the haploid, double-mutant selection media described above. Plates were imaged after 72 hr of incubation at 30°.
A partial sequence alignment of the yeast ubiquitin-conjugating enzymes and Cdc34 orthologs is shown in Figure 1. It is apparent that there is a relationship among S73, S97, and a 12-residue loop region that surrounds the catalytic cysteine. Ubc7, like Cdc34, contains serine residues at the positions equivalent to yeast Cdc34 S73 and S97. In contrast, the majority of E2s, of which Rad6 is a typical example, have a lysine and an aspartic acid residue, respectively, at these positions. Cdc34/Ubc7 family members also contain a loop near S97 (residues 103–114 of Cdc34) that other E2s lack. One of these two alternative motifs of serine/serine/loop or lysine/aspartic acid/loopless is present in most eukaryotic E2s (Liu et al. 1995). The amino acid residues that constitute the motif are in close physical proximity. The crystal structure for Ubc7 shows serine residue S91 (Cdc34 S97 equivalent) to lie within 3 Å of the well-conserved aspartic acid residue (D113) of the 12-amino acid loop while serine residue S67 (Cdc34 S73 equivalent) is within 4 Å of an invariant tryptophan residue (W115) (Cook et al. 1997). Even with differences at these amino acid residues, the tertiary fold of the Cdc34/Ubc7 family resembles that of the Rad6 family (Figure 1B).
The Cdc34-loopless protein has been evaluated in SCF-dependent in vitro reactions (Petroski and Deshaies 2005). In the absence of the loop region, the monoubiquitylation of Sic1 occurs much as with wild-type Cdc34; however, both chain formation and the specificity of the lysine-48 linkage are compromised. Cdc34tm has yet to be investigated in an SCF-dependent in vitro reaction. There are important distinctions that can be made between CDC34Δ103-114 and CDC34tm. We have shown that the CDC34Δ103-114 mutation supports growth much more poorly than either CDC34 or CDC34tm and not at all when expressed from a weak promoter (Liu et al. 1995). However, CDC34tm does not notably compromise growth, viability, or cell cycle distribution, even as a precise ORF replacement into the CDC34 chromosomal locus (Liu et al. 1995; data not shown).
CDC34tm differentially affects the abundance and half-life of SCF substrates:
The disparity between the less efficient, Cdc34Δ103-114-catalyzed Sic1 polyubiquitylation and the lack of a notable change in growth and viability in CDC34tm strains led us to test whether cells dependent upon Cdc34tm had different steady-state levels of known Cdc34 substrates in vivo. Strikingly, the levels of Cln1, Cln2, Sic1, and Far1 are altered in a CDC34tm strain (Figure 2A). The steady-state abundances of Sic1, Cln1, and Cln2 are increased while the Far1 steady-state level is reduced in the CDC34tm-expressing strain. Consistent with previous studies, Cln1 protein is visualized as two bands with the slower-migrating species known to represent a modified species (Tyers et al. 1993). A similar pattern of migration is seen when yeast extracts are probed for Cln2 protein and the slower-migrating species are attributable, at least in part, to phosphorylation (Wittenberg et al. 1990).
We measured the half-lives of Sic1 and Cln1 to determine whether the increased abundance of these proteins in the CDC34tm strain reflected an increase in protein stability. Half-life measurements reveal Sic1 to be much less stable while Cln1 is more stable in CDC34tm cells (Figure 2, B and C), a difference that is not attributable to an alteration in cell cycle distribution as assessed by flow cytometric analysis of DNA content and budding index (data not shown). In wild-type cells, the Sic1 half-life is ∼50 min, and this measurement agrees with previous measurements of Sic1 half-life (Bailly and Reed 1999). Strikingly, the Sic1 half-life in CDC34tm-expressing cells is reduced to <10 min. While the half-life of Cln1 is <10 min in wild-type cells, the Cln1 half-life in the CDC34tm strain is nearly 20 min. This is double the half-life in wild-type cells but not as long as the Cln1 half-life (∼60 min) in cdc34 temperature-sensitive cells (Blondel and Mann 1996). Thus the SCFGrr1 substrates Cln1 and Cln2 are stabilized in CDC34tm cells while the SCFCdc4 substrate Sic1 is destabilized. We cannot address the stability of the SCFCdc4 substrate Far1 as it becomes undetectable in CDC34tm cells. However, a likely explanation for this Far1 effect is a decrease in its stability.
An increase in the mRNA level of SIC1 and a coregulated gene cluster in CDC34tm cells can be attributed to the transcription factor Ace2:
To investigate the basis of the increase in Sic1 steady-state abundance in CDC34tm cells and to better understand the functional significance of the Cdc34/Ubc7-specific motif (serine73/serine97/loop), microarray analysis was performed. Four biological replicates were sampled for the CDC34tm strain and an isogenic wild-type strain. FAR1 and CLN2 mRNA levels are not significantly different in CDC34tm cells. However, CLN1 and SIC1 mRNA levels are significantly increased (P-value < 0.01) in the CDC34tm strain. Examination of the cluster of genes coregulated throughout the cell cycle with SIC1, as defined by Spellman et al. (1998), revealed that the expression of >60% of the genes in this cluster is significantly increased (P-value < 0.05) (Table 2). Again, this effect cannot be attributed to an altered cell cycle distribution of the CDC34tm strain as asynchronous CDC34tm cultures have a cell cycle distribution approximating that of wild-type cultures as assessed by DNA content (data not shown). The transcriptional flux of most genes in this cluster is attributed to the transcriptional activities of Ace2 and Swi5. Ace2 and Swi5 share highly similar DNA-binding domains and regulate many of the same genes (Butler and Thiele 1991; Dohrmann et al. 1996). However, certain genes, such as HO and CTS1, depend solely on either Swi5 or Ace2 for their expression (Dohrmann et al. 1992; McBride et al. 1999). Swi5 and Ace2 have both overlapping and independent roles in the regulation of the SIC1 coregulated gene cluster (Doolin et al. 2001). As shown in Table 2, Ace2-dependent genes are the most highly upregulated in CDC34tm cells. Furthermore, CLN3 transcription is repressed by Ace2 in daughter cells, and this is part of the mechanism that allows daughter cells to grow to a larger size prior to START, the cell cycle commitment point (Laabs et al. 2003; Di Talia et al. 2009). CLN3 mRNA levels are decreased 1.41-fold (P-value = 0.006) in CDC34tm cells.
These microarray results suggested that in CDC34tm cells either Ace2 or Swi5 activity is increased. Since known Cdc34 targets include transcription factors, we hypothesized that an increase in either Ace2 or Swi5 abundance is responsible for the increase in SIC1 mRNA levels. To address this hypothesis, Western blot analysis was performed to examine Ace2 and Swi5 levels in wild-type and CDC34tm cells. As seen in Figure 3A, both Ace2 and Swi5 protein levels are elevated in CDC34tm cells as compared to wild-type cells. The increase in steady-state protein level arises post-transcriptionally since the microarray analysis reveals no significant difference in the mRNA levels of ACE2 or SWI5 between wild-type and CDC34tm cells. We next examined the level of Sic1 protein in CDC34tm cells in the absence of ACE2 and SWI5. Disruption of ACE2 in the context of CDC34tm reduces Sic1 protein levels to below that detected in the wild-type strain; on the other hand, deletion of SWI5 in CDC34tm cells has little effect on the steady-state abundance of Sic1 (Figure 3B). In these experiments (Figure 3B), a rapid protein extraction method using SDS and heat was utilized for the analysis of Sic1 protein (compare with Figure 2). This method allows for the detection of protein modifications such as phosphorylation and ubiquitylation. Under these extraction conditions, Sic1 is visualized as two differently migrating species with the lower band migrating at the size observed when other extraction conditions are employed (see Figure 2). The slower-migrating species is likely Sic1 containing phosphorylation or ubiquitylation modification(s) and, importantly, is not visualized in sic1Δ protein extracts. Thus, the increase in the steady-state abundance of Sic1 in CDC34tm cells is due to increased Ace2-dependent transcription of SIC1, which most likely results from an increase in Ace2 protein abundance.
As the increased steady-state abundance of Ace2 in CDC34tm cells is due to a post-transcriptional mechanism, we postulated that Ace2 is targeted for degradation by an SCF/Cdc34 complex. Therefore, the steady-state abundance of Ace2 was assessed in strains lacking each of the genes encoding F-box proteins (namely SAF1, DIA2, MDM30, GRR1, HRT3, UFO1, RCY1, DAS1, MFB1, ELA1, YMR258c, YLR352w, YDR306c, YNL311c, YDR131c, and YLR224w) except the essential genes CDC4 and MET30. Ace2 levels remain unchanged in all strains tested except grr1Δ and mdm30Δ cells, which both have increased levels of the Ace2 protein (Figure 3C, lanes 2 and 4; supporting information, Figure S1). This is consistent with the finding that both Cln1 and Cln2, SCFGrr1 substrates, accumulate in CDC34tm cells and with our conclusion that the CDC34tm strain is partially defective in degradation of SCFGrr1 substrates.
CDC34tm is not a hypomorphic allele—targets of the transcription factor Haa1 are downregulated in CDC34tm cells:
A previous microarray comparing cdc34 and cdc53 temperature-sensitive strains found an increase in the activity of the Met4, Gcn4, and Tec1 transcription factors (Varelas et al. 2006). Met4, Gcn4, and Tec1 are SCF substrates, and their stabilization in the cdc34 and cdc53 temperature-sensitive strains leads to increased activation of their respective target genes. The analysis of CDC34tm cells did not show a comparable change in Met4-, Gcn4-, or Tec1-dependent transcripts (Table S1). However, 57 gene transcripts were significantly (P-value < 0.05) upregulated at least twofold, and 140 were significantly downregulated by at least twofold in CDC34tm cells (Table S1). To determine the transcription factor(s) whose activity is altered in CDC34tm cells, we analyzed our data using the YEASTRACT (http://www.yeastract.com) database, which relates S. cerevisiae transcriptional regulators and their target genes (Teixeira et al. 2006). Ninety percent of documented Haa1 target genes are downregulated more than twofold in CDC34tm cells (Table 3). Haa1 is critical for the transcriptional response to exogenous acetaldehyde and weak organic acids. Although insensitive to exogenous acetaldehyde, haa1Δ strains are sensitive to acetic and sorbic acid (Aranda and del Olmo 2004; Fernandes et al. 2005).
Elevated levels of acetaldehyde can inhibit the growth of S. cerevisiae cells (Stanley et al. 1993). Excess acetaldehyde can be detoxified by allowing it to react with sulfite to form acetaldehyde hydroxysulfonate (Casalone et al. 1992). Acetaldehyde hydroxysulfonate production can also mitigate the growth inhibitory effects of excess sulfite, which can be present exogenously or produced during the intracellular reduction of sulfate to cysteine (Casalone et al. 1992). Microarray studies demonstrate that part of the response to acetaldehyde exposure is an increase of intracellular sulfite levels via upregulation of the homocysteine production pathway and reduction of the sulfite exporter Ssu1 (Aranda and del Olmo 2004). Strains lacking transcriptional regulators of the methionine biosynthetic pathway (MET4, CBF1, and MET28) are sensitive to 1 g/liter levels of acetaldehyde.
We compared the CDC34tm transcriptional profile to an acetaldehyde response (Aranda and del Olmo 2004) and discovered a strong inverse correlation. Many of the transcripts most highly induced by acetaldehyde are significantly downregulated in a CDC34tm strain (Table 3). Conversely, SSU1, which encodes the sulfite extrusion pump, is downregulated 3.3-fold in cells exposed to acetaldehyde but is upregulated 3-fold in CDC34tm cells. Genes encoding the branch points of glucose fermentation that would shunt glucose flux away from acetaldehyde production, namely the NAD-dependent glycerol-3-phosphate dehydrogenases GPD1 and GPD2, are both downregulated ∼3-fold while genes encoding the first and second steps of the pentose phosphate pathways, namely ZWF1 (−2.2) and SOL4 (−7.3), are also both decreased in CDC34tm cells (Figure 4A).
The inverse correlation between cells exposed to acetaldehyde and CDC34tm cells led us to consider whether CDC34tm cells would be sensitive to sulfite because of an inability to detoxify it. Indeed, CDC34tm cells are much more sensitive than their isogenic wild-type counterparts (Figure 4B). It has previously been shown that grr1Δ cells are sensitive to sulfite (Avram and Bakalinsky 1996), which again supports the conclusion that Cdc34tm compromises SCFGrr1 activity. Thus the reduction in Haa1 targets in CDC34tm cells likely reflects reduced intracellular acetaldehyde.
CDC34tm cell viability requires polyubiquitin receptors of the 26S proteasome and the RNA Pol II kinase CTDK-I to regulate Ace2 and Sic1:
As an alternative approach to identifying the functional significance of the Cdc34/Ubc7-specific motif, an SGA screen using the CDC34tm allele as the query gene was performed. This was accomplished by tightly linking the CDC34tm allele to the nourseothricin N-acetyltransferase gene (nat1) from Streptomyces noursei, which confers nourseothricin (nat1) resistance (Goldstein and McCusker 1999). The nat1 start codon was placed ∼1700 nucleotide bases 5′ of the CDC34tm start codon. As expected, the nat1 gene faithfully segregated with the CDC34tm allele in >20 individual segregants derived from a CDC34tm∷nat1/CDC34 diploid (data not shown). The SGA screen was carried out in duplicate, and genetic interactions were scored by computer-based image analysis of colony size (Figure 5A). To our knowledge, this is the first synthetic lethal screen accomplished using a neomorphic mutant rather than a loss-of-function gene deletion. Notably, 14 genes proximal to the CDC34 chromosomal locus scored as synthetically lethal in the primary screen. Reduced recombination between neighboring loci prevents facile generation of double-mutant haploids. Genes proximal to CAN1 and the MFA1∷pMFA1-HIS3 loci also appeared in the primary screen. These interactions result from reduced recombination frequency rather than from true synthetic lethality and were not included in the secondary screen.
Candidate genetic interactions from the primary SGA screen were confirmed by crossing both a CDC34tm∷nat1 strain and a CDC34∷nat1 strain to strains carrying deletions of genes that scored as synthetically lethal in the primary screen (Figure 5A). Diploids were struck onto media that selected for haploids with both nat1 (marking either CDC34 or CDC34tm) and KanR (marking interacting gene deletion). To our knowledge, this is the first time that the control strain, which determines the gene deletions that are lethal when crossed to a wild-type strain, was noted to harbor the resistance gene in exactly the same genomic location as the query strain. In a typical synthetic lethal screen, nat1 marks the gene deletion rather than being present at the same position as in the control strain.
In all, 88 genes were confirmed to be either synthetically lethal or sick with CDC34tm (Table 4). Rpn10 is a non-ATPase subunit of the 19S regulatory particle of the proteasome that binds medium-to-long K48-linked polyubiquitin chains. Deletion of RPN10 stabilizes Sic1 (Verma et al. 2004). The human Rpn10 ortholog, S5a, was the first 26S proteasome subunit implicated in the binding of polyubiquitylated substrates (Deveraux et al. 1994). Deletion of RAD23 is also synthetically lethal with CDC34tm (Figure 5B). Rad23 has a ubiquitin-like domain (UBL) near its N terminus and a ubiquitin associated (UBA) domain that binds K48-linked ubiquitin chains in vitro (Bertolaet et al. 2001). In yeast, Rad23 can bind substrates with shorter polyubiquitin chains (four to six ubiquitin molecules) through its UBA domain and recruit them to the proteasome through an interaction between its UBL domain and Rpn1 (Elsasser et al. 2002). Yeast strains lacking RAD23 are not impaired in their ability to degrade Sic1 in vivo, but 26S proteasomes isolated from rad23Δ strains are defective in Sic1 deubiquitylation and degradation in vitro (Verma et al. 2004). The synthetic lethality between CDC34tm and deletion of either RPN10 and RAD23 is particularly intriguing because utilization of polyubiquitin receptors in different contexts might explain the ability of CDC34tm cells to survive while the polyubiquitin-conjugating activity of the SCF complex is compromised as suggested by in vitro reconstitution of SCF ubiquitylation using the Cdc34Δ103-114 mutant (Petroski and Deshaies 2005). Dsk2 and Ddi1 are also short-chain polyubiquitin receptors of the 26S proteasome and, like Rad23, have both UBL and UBA domains. Neither Dsk2 nor Ddi1 appeared in the initial CDC34tm synthetic lethal screen, and we do not detect a fitness defect in CDC34tm ddi1Δ or CDC34tm dsk2Δ haploids (data not shown).
Two known regulators and an interacting partner of the SCF complexes were identified as synthetically lethal with CDC34tm. CKA2, one of two catalytic subunits of Casein Kinase II, phosphorylates any of six serines within the acidic C terminus of Cdc34, and it is synthetically lethal with CDC34tm (Figure 5B). Phosphorylation of Cdc34 by Casein Kinase 2 increases its activity against Sic1 (Sadowski et al. 2007; Coccetti et al. 2008). Furthermore, Casein Kinase 2 also phosphorylates Ser201 of Sic1, which promotes Sic1 degradation (Coccetti et al. 2006; Tripodi et al. 2007). Another regulator of the SCF complex, LAG2, shares a genetic interaction with CDC34tm (Figure 5B). Lag2 negatively regulates the SCF by binding to and preventing neddylation of Cdc53 (Liu et al. 2009; Siergiejuk et al. 2009). It is interesting that CDC34tm has negative genetic interactions with both positive and negative regulators of Cdc34, further supporting the hypothesis that CDC34tm is a neomorphic mutant. CDC34tm is also synthetically lethal with PRB1, which encodes the vacuolar proteinase B. Multiple studies have revealed a physical interaction between Prb1 and components of the SCF, including Cdc53, Rbx1, and Skp1 (Ho et al. 2002; Kato et al. 2010). Although it is not clear at this time how Prb1 interacts with SCF/Cdc34, the abundance of physical interactions with the SCF complex and its genetic interaction with CDC34tm suggest a physiologically relevant relationship. The discovery of known Cdc34 regulators/interactors raises the intriguing possibility that some of the other genes sharing negative genetic interactions with CDC34tm play a role in Cdc34/SCF regulation.
The three genes encoding proteins of the CTDK-I complex (Ctk1, Ctk2, and Ctk3) are synthetically lethal with CDC34tm (Figure 6, A and B; data not shown). The CTDK-1 complex coordinates transcriptional elongation, pre-mRNA 3′-end processing, and translational fidelity (reviewed in Hampsey and Kinzy 2007). The role of CTDK-1 in transcriptional elongation is well characterized, and its ability to phosphorylate the second serine of the repetitive C-terminal domain of the largest subunit of Pol II (Rpo21) increases the efficiency of transcriptional elongation (Lee and Greenleaf 1997; Patturajan et al. 1999). CDC34tm ctk2Δ haploids die as multiple, attached cells indicative of a cell separation defect (Figure 6C). As Ace2 is required for the expression of CTS1 and cell separation, we hypothesized that mis-regulation of Ace2 and some of its transcriptional targets was the cause of the CDC34tm ctk2Δ lethality. Interestingly, removal of ACE2, SIC1, or SWI5 is able to suppress the lethality of a CDC34tm ctk2Δ mutant. These results suggest that the lethal defect of the CDC34tm ctk2Δ strain is increased production of Sic1 that is alleviated by loss of the SIC1 activators Ace2 and Swi5 or by loss of SIC1 itself (Figure 6D).
This work originated with the discovery that mutations modifying Cdc34 serine residue 97 or deleting amino acid residues 103–114 in the N-terminal catalytic domain make a nonfunctional enzyme, but a combination of mutations at these positions (and S73) yields a functional enzyme, as assessed by in vivo complementation of a cdc34 null strain. However, recent work has shown that the stretch of acidic residues in Cdc34 is necessary for timely and appropriate polyubiquitin chain extension of Sic1 in vitro (Petroski and Deshaies 2005). Therefore, we reasoned that since a strain bearing the Cdc34Δ103-114 mutant does not support growth while a Cdc34tm strain is viable and has a growth rate comparable to wild type, there must be in vivo mechanisms that compensate for the defect in polyubiquitin chain extension. In many cases, multiple ubiquitin-conjugating enzymes must function together to polyubiquitylate a single substrate, with one E2 serving to monoubiquitylate and the other extending the ubiquitin chain (Rodrigo-Brenni and Morgan 2007). However, no genetic interaction between CDC34tm and any of the other ubiquitin-conjugating enzymes was detected in the primary SGA screens, suggesting that other E2s are not compensating for the defective polyubiquitylation activity of the Cdc34tm. In contrast, CDC34tm exhibits strong genetic interactions with the polyubiquitin receptor genes RAD23 and RPN10. Rad23 and Rpn10 have been shown to be required for Sic1 degradation; however, neither is an essential gene. A rad23Δ rpn10Δ strain degrades Sic1 more slowly than strains bearing either single mutation, but is viable. This argues for an additional mechanism of recruiting Sic1 to the proteasome for degradation (Lambertson et al. 1999; Verma et al. 2004). Both CDC34tm rad23Δ and CDC34tm rpn10Δ cells have multiple, elongated buds (data not shown) reminiscent of cdc34ts cells at 37° and indicative of defective Sic1 degradation. CDC34tm rpn10Δ and CDC34tm rad23Δ mutants are not rescued by simultaneous deletion of SIC1 (data not shown), but this is not altogether surprising since cdc34ts sic1Δ cells progress through START but arrest in mitosis as large-budded cells.
The SCFCdc4 substrate Sic1 has a shorter half-life in CDC34tm as compared to wild-type cells. We were unable to determine the half-life of Far1 in CDC34tm cells because it is undetectable. On the other hand, the SCFGrr1 substrate Cln1 is more stable in CDC34tm cells compared to wild-type cells. If steady-state abundance of the Cln proteins is indicative of activity, Cln/CDK activity should be increased in CDC34tm cells. The Cln/Cdc28 complexes phosphorylate Sic1 and Far1 and, on the basis of the number of CDK sites, likely phosphorylate Cln1 as well (Mendenhall et al. 1987; Peter et al. 1993; Henchoz et al. 1997). We postulate that increased Cln/Cdk phosphorylation of the substrates Sic1, Far1, Cln1, and Cln2 enables substrate recruitment to the individual SCF complexes. Upon encountering a Cdc34tm-bound SCF complex, the substrate is ubiquitylated, but with a polyubiquitin chain that may be different from that formed by wild-type Cdc34. This could mean that in Cdc34tm mutants the time between SCF association and dissociation may be shorter for each substrate and thereby available for proteasomal degradation earlier than for substrates that encounter a wild-type Cdc34-containing SCF complex. Alternatively, the polyubiquitin receptors necessary for substrate proteasomal recognition may be different for each of these substrates when ubiquitylated by Cdc34tm. Our data do not reveal the contribution of each step (phosphorylation, ubiquitylation, proteasome binding) to the overall differences in substrate steady-state abundance and half-lives in CDC34tm cells. However, our data suggest that Cdc34tm be considered a tool for both in vivo and in vitro experiments as we continue to dissect the contribution of each step to the rate of the entire process.
An interesting paradox is seen in the simultaneous reduction in Sic1 protein half-life and increase in the steady-state abundance of the Sic1 protein in a CDC34tm strain due to increased Ace2 activity. It is tempting to consider the possibility that Ace2 is an SCFGrr1 and SCFMdm30 substrate. The SCF ubiquitin ligase complexes are responsible for the ubiquitylation of numerous transcription factors. Recent evidence demonstrates that ubiquitylation of these substrates can have both a positive and a negative role in transcriptional regulation but does not always result in proteasome-mediated degradation (Lipford et al. 2005; Chandrasekaran and Skowyra 2008). Gal4, a transcription factor required for GAL gene induction, is ubiquitylated by SCFGrr1 and SCFMdm30 (Muratani et al. 2005). SCFGrr1 targets an inactive pool of Gal4 for proteasome-mediated degradation under non-inducing conditions. SCFMdm30 targets the transcriptionally active pool of Gal4 for proteolysis. Mdm30-dependent degradation of transcriptionally active Gal4 stimulates cotranscriptional mRNA processing of the GAL gene mRNAs (Muratani et al. 2005). In addition, Gal4 transcriptional activity is enhanced by Gal4 monoubiquitylation. The 19S regulatory particle of the 26S proteasome binds to promoter-bound Gal4 and dissociates the Gal4–DNA complex (Ferdous et al. 2007). When Gal4 is monoubiquitylated, it is resistant to the dissociating activity of the 19S regulatory particle.
We measured the half-life of Ace2 in wild-type and CDC34tm cells but found no detectable difference (data not shown). However, the transcription of ACE2 and localization of Ace2 are regulated in a cell cycle-dependent manner. ACE2 mRNA is upregulated during the G2/M transition, and Ace2 enters both the mother cell nucleus and the daughter cell nucleus late in mitosis (Dohrmann et al. 1992). Ace2 is preferentially retained in the daughter nucleus when phosphorylated by the daughter-specific Mob2/Cbk1 kinase complex (Colman-Lerner et al. 2001; Weiss et al. 2002; Mazanka et al. 2008). We suspect that separately measuring the half-life of Ace2 within the daughter nucleus and the mother and daughter cytoplasm will be more informative regarding the mechanism by which Ace2 is upregulated in CDC34tm, grr1Δ, and mdm30Δ cells.
The synthetic lethality of the ctk2Δ CDC34tm strain can be suppressed by deletion of ACE2, SIC1, or SWI5. This finding suggests that the CTDK-1 complex negatively regulates Sic1 activity by affecting Swi5 and Ace2 activity toward SIC1. Swi5 is known to be targeted for proteasome-mediated degradation by SCFCdc4. Phosphorylation of at least one of the eight consensus CDK sites targets Swi5 to SCFCdc4. The Srb10 cyclin-dependent kinase is partially responsible for phosphorylation of Swi5, but it is not the sole kinase involved. Swi5 is stabilized in a srb10Δ strain but not nearly to the same degree as in a strain in which all eight CDK consensus sites of Swi5 have been removed by serine-to-alanine mutations (Kishi et al. 2008). We propose that Ctk1 is also involved in targeting Swi5 to SCFCdc4. Both Srb10 and Ctk1 phosphorylate the CTD of RNA Pol II. Therefore, these kinases are often associated with the transcriptional machinery, and thus both Swi5 and Ace2 may at times be in close physical proximity to these kinases. The fact that deletion of SIC1, SWI5, or ACE2 suppresses the lethality of CDC34tm ctk2Δ cells suggests that the CTDK-I complex is involved in the regulation of both Swi5 and Ace2.
We thank Dorota Skowyra, Bob Harris, and Maureen Harrington for helpful discussions; Dave Stillman for providing Ace2 antibody and helpful discussions; and Paresh Sanghani for assistance rendering the structures. The microarray studies were carried out using the facilities of the Center for Medical Genomics at Indiana University School of Medicine. The Center for Medical Genomics is supported in part by the Indiana Genomics Initiative at Indiana University (which is supported in part by the Lilly Endowment). This work was funded by National Science Foundation grant MCB-0091317 (M.G.).
Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.110.125302/DC1.
↵ 1 Present address: Department of Biology, Indiana University, 915 East 3rd St., Myers Hall, Room 300, Bloomington, IN 47405.
↵ 2 Present address: Xylogenics, Inc., 351 West 10th St., Room 504, Indianapolis, IN 46202.
↵ 3 Present address: School of Biological Sciences, Swann Bldg., Wellcome Trust Centre for Cell Biology, Room 329, University of Edinburgh, Mayfield Rd., Edinburgh EH9 3JR, Scotland, United Kingdom.
Communicating editor: M. Hampsey
- Received November 18, 2010.
- Accepted December 6, 2010.
- Copyright © 2011 by the Genetics Society of America