|Home | About | Journals | Submit | Contact Us | Français|
In the yeast Saccharomyces cerevisiae, a ring of myosin II forms in a septin-dependent manner at the budding site in late G1. This ring remains at the bud neck until the onset of cytokinesis, when actin is recruited to it. The actomyosin ring then contracts, septum formation occurs concurrently, and cytokinesis is soon completed. Deletion of MYO1 (the only myosin II gene) is lethal on rich medium in the W303 strain background and causes slow-growth and delayed-cell-separation phenotypes in the S288C strain background. These phenotypes can be suppressed by deletions of genes encoding nonessential components of the anaphase-promoting complex (APC/C). This suppression does not seem to result simply from a delay in mitotic exit, because overexpression of a nondegradable mitotic cyclin does not suppress the same phenotypes. Overexpression of either IQG1 or CYK3 also suppresses the myo1Δ phenotypes, and Iqg1p (an IQGAP protein) is increased in abundance and abnormally persistent after cytokinesis in APC/C mutants. In vitro assays showed that Iqg1p is ubiquitinated directly by APC/CCdh1 via a novel recognition sequence. A nondegradable Iqg1p (lacking this recognition sequence) can suppress the myo1Δ phenotypes even when expressed at relatively low levels. Together, the data suggest that compromise of APC/C function allows the accumulation of Iqg1p, which then promotes actomyosin-ring-independent cytokinesis at least in part by activation of Cyk3p.
Cytokinesis is the process that divides the cell surface and cytoplasm of one cell into two cells. Although the list of proteins known to be involved in cytokinesis has expanded significantly in recent years, a complete molecular understanding of this process remains elusive. Among the mysteries are the roles of the septins and of the actomyosin contractile ring. The septins are a family of GTP-binding proteins that have been found at the division site in all fungal and animal cells examined (Longtine et al., 1996 ; Hall and Russell, 2004 ; Gladfelter, 2006 ). Although their roles are still imperfectly understood, the septins seem to function as both a scaffold and a diffusion barrier for the localization and organization of other proteins (Gladfelter et al., 2001 ; Longtine and Bi, 2003 ; Dobbelaere and Barral, 2004 ; Versele and Thorner, 2005 ; Spiliotis and Nelson, 2006 ). Surprisingly, although the septins are indispensable for cytokinesis in some cell types, they are dispensable in others (Longtine et al., 1996 ; Adam et al., 2000 ; Nguyen et al., 2000 ; Kinoshita and Noda, 2001 ; An et al., 2004 ). Similarly, it has recently become clear that the actomyosin ring at the division site is also dispensable for cytokinesis in a variety of cell types (Bi et al., 1998 ; Nagasaki et al., 2002 ; Kanada et al., 2005 ). In the budding yeast Saccharomyces cerevisiae, the septins are essential for formation of the actomyosin ring. However, this cannot be their only role, because they are essential for cytokinesis in this organism, whereas the actomyosin ring is not essential in most strain backgrounds (Bi et al., 1998 ; Schmidt et al., 2002 ; Lord et al., 2005 ; our unpublished results). A major challenge at present is to elucidate the actomyosin-ring-independent role(s) of the septins in cytokinesis in yeast and (presumably) in other cell types.
In S. cerevisiae, Myo1p (the only type II myosin in this organism) forms a ring at the presumptive budding site in late G1 (Bi et al., 1998 ; Lippincott and Li, 1998 ). This ring remains at the mother-bud neck until the onset of cytokinesis, when actin and other proteins are recruited to it to form the mature actomyosin ring, which soon contracts. Concurrent with this contraction, the plasma membrane invaginates and the primary cell-wall septum is synthesized, principally by the chitin synthase Chs2p (Cabib et al., 2001 ; Schmidt et al., 2002 ). Secondary septa are then deposited on both sides of the primary septum to form the mature trilaminar septum, and the mother and daughter cells are separated by the action of a chitinase that partially hydrolyzes the primary septum (Kuranda and Robbins, 1991 ; Colman-Lerner et al., 2001 ). In viable myo1Δ cells, no actomyosin ring forms, and the septa that form are typically disorganized and often lack well defined primary septum-like structures (Schmidt et al., 2002 ; Tolliday et al., 2003 ; our unpublished results). These disorganized septa are presumably the reason that myo1Δ cells grow more slowly than wild type and fail to separate efficiently, resulting in the formation of multicell clusters.
Among the proteins recruited to the division site in a septin-dependent manner just before cytokinesis are Iqg1p and Cyk3p. Iqg1p is the only member of the IQGAP family (Brown and Sacks, 2006 ) in S. cerevisiae, and it has been reported to be essential both for formation of the actomyosin ring and for cytokinesis (Epp and Chant, 1997 ; Shannon and Li, 1999 ; Luo et al., 2004 ). Because the actomyosin ring itself is not essential for cytokinesis, Iqg1p must have at least one cytokinetic function that is actomyosin-ring independent. The function of Cyk3p is not known, but its overexpression suppresses the iqg1Δ lethality without restoring the actomyosin ring, suggesting that Cyk3p also promotes cytokinesis through an actomyosin-ring-independent pathway (Korinek et al., 2000 ; our unpublished results).
For successful cellular reproduction, cytokinesis must be coordinated with other late cell-cycle events such as the completion of chromosome segregation and the exit from mitosis. Not surprisingly, cells have sophisticated regulatory pathways to ensure this coordination. Regulation of late mitotic events depends largely on the anaphase-promoting complex (or cyclosome) (APC/C), an essential multisubunit ubiquitin ligase that targets specific cell cycle-related proteins for degradation (Peters, 2006 ). The APC/C, with its activating subunit Cdc20p, initiates anaphase by triggering sister-chromatid separation via degradation of the securin Pds1p (Nasmyth, 2005 ), which unleashes the protease separase. In S. cerevisiae, separase activation also initiates release of the protein phosphatase Cdc14p from its nucleolar inhibitor Net1p. Activation of the mitotic-exit network further activates Cdc14p, which then activates Cdh1p, a second APC/C-activating subunit that targets Clb2p and other mitotic cyclins for degradation in late mitosis and early G1. In addition to these major roles, the APC/C also regulates other cell-cycle proteins, such as the spindle-associated kinesins Kip1p and Cin8p (Harper et al., 2002 ).
Thirteen subunits of the APC/C have been identified in S. cerevisiae. Many of the core subunits are essential for APC/C function and thus for viability. However, some subunits, including Cdc26p, Apc9p, Doc1p, Swm1p, and Mnd2p, are not essential. Cdc26p and Apc9p seem to be involved in the assembly of the APC/C (Passmore, 2004 ), Doc1p is involved in promoting the association between the APC/C and its substrates (Carroll and Morgan, 2002 ; Passmore et al., 2003 ; Carroll et al., 2005 ), Swm1p is required for full catalytic activity of the APC/C (Schwickart et al., 2004 ; Page et al., 2005 ), and Mnd2p seems to have little role in mitotic control but is important in regulating the APC/C during meiosis (Oelschlaegel et al., 2005 ; Page et al., 2005 ).
To investigate the mechanisms of septin-dependent, actomyosin-ring-independent cytokinesis in S. cerevisiae, we have been conducting synthetic-lethal and dosage-suppression screens starting with septin and myo1Δ mutants. During these studies, we unexpectedly observed that mutations in nonessential subunits of the APC/C could suppress the phenotypes of myo1Δ mutants. We present evidence that Iqg1p is a direct substrate of the APC/C and that the elevation of Iqg1p levels in APC/C mutants accounts for the suppression of myo1Δ phenotypes.
The strains and plasmids used in this study are listed in Tables 1 and and2;2; their construction is described below or in the tables. Yeast were grown on liquid or solid synthetic complete (SC) medium lacking specific nutrients as needed to select plasmids or transformants, YP rich liquid or solid medium, or YM-P rich, buffered liquid medium (Lillie and Pringle, 1980 ; Guthrie and Fink, 1991 ). Two percent glucose was used as carbon source except for experiments involving induction of gene expression under GAL promoter control, for which 1 or 2% raffinose plus 0.5 or 2% galactose was used, as indicated. Yeast strains were grown at 23°C except as noted. The antibiotic Geneticin (G-418; Lonza Walkersville, Walkersville, MD) was used to select for cells containing the kanR marker, and 5-fluoroorotic acid (5-FOA; Research Products International, Mt. Prospect, IL) was used to select for ura3-mutant cells. To arrest cells in G1, α-factor (Sigma-Aldrich, St. Louis, MO) was used at the concentration indicated. The microtubule-depolymerizing drug benomyl (DuPont, Wilmington, DE) was used at 50 μg/ml to arrest cells in M phase. To block translation, cycloheximide (MP Biomedicals, Solon, OH [Figure 6] or Sigma-Aldrich [Figure 9]) was added to the growth medium at the concentration indicated. Standard procedures were used for growth of Escherichia coli, genetic manipulations, polymerase chain reaction (PCR), and other molecular biological procedures (Sambrook et al., 1989 ; Guthrie and Fink, 1991 ; Ausubel et al., 1995 ).
Genes were deleted using the PCR method (Baudin et al., 1993 ; Longtine et al., 1998 ), and the primers are indicated in Supplemental Table 1; in each case, the entire coding region was deleted. The success of each deletion was confirmed by two PCR tests that used check primers that were upstream and downstream, respectively, of the deleted region together with primers internal to the selectable markers (Longtine et al., 1998 ; Supplemental Table 1). C-terminal tagging with sequences encoding the 3HA epitope or GFP(F64L,S65T,V163A) was also done using the PCR method with plasmid pFA6a-3HA-His3MX6 (Longtine et al., 1998 ) or pFA6a-GFP(F64L,S65T,V163A)-His3MX6 (see below) as template. The success of the tagging was confirmed by two PCR tests, essentially as described above. To construct strains DOY138-141, an IQG1-TAP:His3MX6 C-terminal fragment was PCR amplified using genomic DNA from strain IQG1-TAP as template and the primers indicated in Supplemental Table 1. This cassette was then transformed into appropriate parental strains as indicated in Table 1.
Plasmid pFA6a-GFP(F64L,S65T,V163A)-His3MX6 was constructed by subcloning an MscI–BstBI fragment containing the three mutations from YEpGFP*-BUD8F (Schenkman et al., 2002 ) into MscI/BstBI-cut pFA6a-GFP(S65T)-HIS3MX6 (Wach et al., 1997 ). Plasmid pTSV31A-MYO1 was constructed by subcloning the 7.0-kb SalI–BamHI MYO1 fragment from pBS-MYO1 (a gift from E. Bi, University of Pennsylvania, Philadelphia, PA) into SalI/BamHI-cut pTSV31A (a 2μ URA3 ADE3 plasmid; Tibbetts and Pringle, unpublished data). Plasmid pGT04 was constructed using two steps of PCR. In the first step, a fragment of IQG1 (nucleotides −262 to +3 relative to the A of the start codon) was amplified from genomic DNA with a BamHI site incorporated into the 5′ primer and a 3′ primer that included nucleotides corresponding to positions +127 to +141 of IQG1. A second fragment (nucleotides +127 to +277) was also amplified from genomic DNA using a 5′ primer that included nucleotides corresponding to positions −15 to +3 of IQG1 and a 3′ primer that included an XbaI site. In the second step, the PCR products from the first step were purified and used as template with the BamHI site-containing 5′ primer and the XbaI site-containing 3′ primer. The resulting product, which contained 262 nucleotides of the IQG1 promoter, a start codon, and 151 nucleotides (from +127 to +277) of open reading frame sequence, was cut with BamHI and XbaI, gel purified, and inserted into BamHI/XbaI-cut pRS305 (Sikorski and Hieter, 1989 ). Plasmids pGAL-IQG1-TAP and pGAL-iqg1Δ42-TAP were constructed by transforming yeast cells with BamHI/HindIII-cut pRSAB1234 (see Supplemental Materials and Methods of Gelperin et al., 2005 ) with PCR-amplified full-length or truncated (lacking codons 2–42) IQG1; the amplified fragment contained 22 (5′) and 21 (3′) base pairs of flanking vector sequences to allow the in vivo recombination.
To determine growth rates, exponential phase cultures (OD600 ≈ 0.4) in YM-P medium were diluted twofold with fresh YM-P, and incubation was continued. The times needed to return to the original OD600 were recorded as the doubling times.
To determine cell-cluster indices, strains were streaked onto the indicated media, grown overnight, scraped from the plates, washed once with water by centrifugation, resuspended, sonicated briefly, and observed by differential-interference-contrast (DIC) microscopy. Each unbudded (one cell body) or budded (two cell bodies) cell was scored as one nonclustered unit, and entities with three, four, five, or six or more cell bodies were scored as one, two, three, or four clusters, respectively. Each count was continued until the number of clusters plus nonclustered units was 100, and the number of clusters was recorded as the cluster index. To minimize the possibility that the cell-cluster index determined for the myo1Δ single mutant would be influenced by spontaneously arising suppressors, we examined nine different myo1Δ strains that were obtained as segregants from RNY112 and its double-mutant derivatives.
The colony-sectoring assay was based on that described by Bender and Pringle (1991) . ade2-1 ade3Δ ura3-3 myo1Δ strains carrying a URA3 ADE3 MYO1 plasmid were grown on SC-Ura plates and then streaked onto YP or YPGalRaf plates to observe sectoring or nonsectoring single colonies. In some experiments, a LEU2-marked plasmid (empty or carrying IQG1 or CYK3) was also present. In these cases, the strains were first grown on SC-Ura-Leu plates and then streaked onto SC-Leu plates to observe sectoring or nonsectoring single colonies.
To assess the extent of G2 delay by using FACS, cells were grown to exponential phase (OD600 ≈ 0.4) in YM-P medium, and then they were collected, fixed, stained with SYTOX Green (Invitrogen, Carlsbad, CA), and examined using standard procedures (Haase and Reed, 2002 ).
A W303-background myo1Δ haploid strain, RNY798, which contained an ADE3 MYO1 plasmid and could only form nonsectored viable colonies (see above), was transformed with a genomic-DNA library in the low-copy plasmid YCp50-LEU2 (Bi and Pringle, 1996 ; the library was constructed using DNA from an S288C-background strain and was kindly provided by F. Spencer and P. Hieter, Johns Hopkins University, Baltimore, MD). From ~20,000 transformants screened on SC-Leu plates, 85 reproducibly sectoring transformants were identified. Isolation of plasmids and retransformation of strain RNY798 yielded 14 plasmids that rescued the lethality of the myo1Δ strain. Sequencing and subcloning of the inserts revealed that nine plasmids contained full-length MYO1 or C-terminal MYO1 fragments, and that in three other plasmids, the gene responsible for suppression was either IQG1 (two cases) or CYK3 (one case). The suppressing genes in the final two plasmids have not yet been identified.
Iqg1p-3HA levels were determined by immunoblotting in both asynchronous and synchronous cultures. For synchronous cultures, the supersensitive bar1Δ (Sprague, 1991 ) strains were grown to exponential phase in SC-His medium, treated for 3 h with 70 ng/ml α-factor to arrest cells in G1, released from arrest by centrifugation and resuspension in SC-His medium, and sampled at 20-min intervals. At each time point, one sample was taken for protein extraction, and a second sample was fixed for 10 min in 70% ethanol at 0°C, resuspended in phosphate-buffered saline (PBS), stained with 4,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich), and used to evaluate cell-cycle progression by scoring the percentages of large-budded cells with two well separated chromosome sets. To extract proteins, cells were collected by centrifugation, resuspended in 1.85 M NaOH containing 2% β-mercaptoethanol, and incubated for 10 min at 0°C. Trichloroacetic acid was then added to 50%, and incubation was continued for 15–25 min at 0°C. Insoluble material was collected by centrifugation, mixed with SDS-sample buffer, boiled for 5 min, and analyzed on 7% SDS-polyacrylamide gels. After transferring proteins electrophoretically to nitrocellulose transfer membrane (GE Healthcare, Piscataway, NJ), Iqg1p-3HA was detected using the rat monoclonal anti-HA antibody 3F10 (Roche Molecular Biochemicals, Indianapolis, IN) as primary antibody. Protein bands were then visualized using either mouse anti-rat-immunoglobulin (IgG) secondary antibody (Jackson ImmunoResearch, West Grove, PA), alkaline phosphatase-conjugated goat anti-mouse-IgG tertiary antibody (Sigma-Aldrich), and the AttoPhos AP fluorescent substrate system (Promega, Madison, WI) (Figure 6D), or horseradish peroxidase (HRP)-conjugated goat anti-rat-IgG secondary antibody (GE Healthcare) and the enhanced chemiluminescence (ECL) system (GE Healthcare) (Figure 7). Actin was detected using either a goat polyclonal anti-actin antibody (Karpova et al., 1993 ) (Figures 6D and and7B)7B) or the mouse monoclonal anti-actin antibody MAB1501 (Chemicon International, Temecula, CA) (Figure 7A) as primary antibody. Protein bands were then visualized using alkaline phosphatase-conjugated rabbit anti-goat-IgG secondary antibody (Sigma-Aldrich) and the AttoPhos AP system (Figure 6D), HRP-conjugated donkey anti-goat-IgG secondary antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and the ECL system (Figure 7B), or alkaline phosphatase-conjugated goat anti-mouse-IgG secondary antibody and the AttoPhos AP system (Figure 7A). In Figure 6D, the intensities of the protein bands were measured using the Storm Scanner model 840 (GE Healthcare), and the values for Iqg1p-3HA were normalized using the actin bands from the same samples.
Iqg1p-TAP levels were determined by immunoblotting using the peroxidase anti-peroxidase soluble complex produced in rabbit (catalog no. P1291; Sigma-Aldrich) and the SuperSignal West Pico chemiluminescence system (Pierce Chemical, Rockford, IL). For the experiments depicted in Figure 6, B and C, cell lysates were prepared as described previously (Ostapenko and Solomon, 2005 ), and Iqg1p-TAP was precipitated from equal amounts of lysate by incubation for 90 min at 4°C with IgG-Sepharose (GE Healthcare) in 10 mM Tris-Cl, pH 7.5, 150 mM NaCl, containing 1% NP-40 and protease inhibitors (10 mg/ml each of leupeptin, chymostatin, and pepstatin; all from Chemicon International). Precipitated proteins were separated on 10% SDS-PAGE, transferred to Immobilon-P membranes (Millipore, Billerica, MA), and detected as described above using an overnight incubation at 4°C in 10 mM Tris-Cl, pH 7.5, 150 mM NaCl, containing 0.1% Tween 20 and 5% dry milk. For the loading control in Figure 6B, Cdc28p was detected using a rabbit polyclonal anti-PSTAIR primary antibody (Solomon, unpublished data), an HRP-conjugated goat anti-rabbit-IgG secondary antibody (Santa Cruz Biotechnology), and the SuperSignal West Pico system. For the experiments depicted in Figure 9, C and D, protein extracts were prepared and analyzed by 10% SDS-PAGE and immunoblotting as described previously (Ubersax et al., 2003 ). Iqg1p-TAP and Iqg1Δ42p-TAP were detected as described above, and Clb2p was detected using a rabbit polyclonal anti-Clb2p primary antibody (Kellogg and Murray, 1995 ), an HRP-conjugated donkey anti-rabbit-IgG secondary antibody (GE Healthcare), and the SuperSignal West Pico system.
To perform ubiquitination assays, reaction components were expressed and purified as described previously (Charles et al., 1998 ; Carroll and Morgan, 2002 , 2005 ). Substrates were produced in rabbit reticulocyte lysates by coupled transcription and translation in the presence of [35S]methionine, following the manufacturer's instructions (Promega). Truncated IQG1 constructs with a T7 promoter sequence and Kozak site added upstream were made using two steps of PCR. In the first step, the desired IQG1 fragment was PCR amplified using a 5′ primer containing a sequence of 23 nucleotides that overlapped the 5′ primer used in the second step plus 30–37 nucleotides of IQG1 coding sequence beginning with a start codon. The 3′ primer contained ~20 nucleotides of coding sequence ending with a stop codon. In the second step, the product from the first step was used as template with the same 3′ primer and a 5′ primer of 119 nucleotides that included a T7 promoter and a Kozak site. Point mutations were introduced by incorporation into the 5′ primer used in the first step. Similarly, full-length PDS1 (from the start codon to 99 nucleotides downstream of the stop codon) and CYK3 (from the start codon to 64 nucleotides downstream of the stop codon) constructs were generated with the T7 promoter sequence and Kozak site upstream of the start codons. Ubiquitination reactions were performed and monitored using a Molecular Dynamics PhosphorImager (GE Healthcare) as described previously (Carroll and Morgan, 2002 , 2005 ; Carroll et al., 2005 ).
DIC and fluorescence microscopy were performed using a Nikon (Tokyo, Japan) Eclipse E600-FN microscope and an ORCA-2 cooled charge-coupled-device camera (Hamamatsu Photonic Systems, Bridgewater, NJ). To quantitate Iqg1p-GFP signal intensities, cells were grown to exponential phase in SC-His medium, collected by centrifugation, and resuspended in water before observation by DIC and fluorescence microscopy. The average intensities of the GFP signals at the mother-bud neck were determined for each strain using MetaMorph version 5.0 (Molecular Devices, Sunnyvale, CA); a box of fixed size was drawn to contain the neck of each large-budded cell with a detectable signal, and the total fluorescence of the boxed area was measured and recorded using the regional-measurement function of MetaMorph.
A myo1Δ mutation is not lethal in the S288C strain background at 23°C (see Introduction). We used the colony-sectoring method of Bender and Pringle (1991) to screen for mutations synthetically lethal with myo1Δ under these conditions (our unpublished results). The first several mutants analyzed proved to be temperature sensitive for growth (i.e., unable to grow at 37°C even when the MYO1 plasmid was present), which seemed to offer an easy method to clone the genes harboring the synthetic-lethal mutations. Using a genomic-DNA library to rescue the temperature sensitivity of one mutant, we recovered CDC26, which encodes a subunit of the APC/C that is nonessential at 23°C but essential at 37°C. Further investigation revealed that the parental myo1Δ strain used in the screen was itself temperature sensitive and harbored a cdc26 mutation [a YJRWdelta11 or YJRWdelta13 sequence (338 nucleotides; Saccharomyces Genome Database) inserted at nucleotide −7 relative to the start site of the CDC26 ORF] that accounted for the temperature-sensitive phenotype. Because myo1Δ strains grow less well than wild type (Rodriguez and Paterson, 1990 ; Bi et al., 1998 ; Lippincott and Li, 1998 ) and can accumulate spontaneous suppressor mutations (Tolliday et al., 2003 ), this suggested that the cdc26 mutation might have been selected because it alleviated the phenotype of the original myo1 mutant. Indeed, when we intentionally introduced deletions of CDC26 or other nonessential APC/C subunits into freshly prepared myo1Δ strains, we found that these mutations could suppress both the clustering and slow-growth phenotypes in the S288C background (Figure 1) and the lethal phenotype (Tolliday et al., 2003 ; our unpublished results) in the W303 background (Figure 2). Similar results were obtained with deletions of CDH1, the APC/C activator for exit from mitosis (Figures 1, B and C, and and2A).2A). In contrast, deletion of PDS1, which encodes securin, the APC/C target whose degradation triggers anaphase onset, did not alter the phenotypes in either background (Figures 1, B and C, and and2A).2A). Together, these results suggested that APC/C mutations do not suppress myo1Δ phenotypes by affecting the timing of anaphase.
It also seemed possible that APC/C mutations might suppress myo1Δ phenotypes by delaying mitotic exit and thus allowing more time for an inefficient process of cytokinesis to be completed successfully. To explore this possibility, we first introduced a deletion of NET1 into an S288C-background strain that was heterozygous for a myo1Δ mutation. Because NET1 encodes an inhibitor of Cdc14p, which activates the APC/C for mitotic exit, it seemed possible that a net1 mutation might exacerbate the myo1Δ phenotypes by accelerating mitotic exit. Indeed, the doubly heterozygous diploid yielded only synthetic-lethal or synthetic-sick myo1Δ net1Δ double-mutant segregants (Figure 3A), and the poor viability was suppressed by deletion of a nonessential APC/C subunit (Figure 3B). These data were consistent with the hypothesis that APC/C mutations might suppress the myo1Δ phenotypes by delaying mitotic exit. However, this hypothesis was difficult to reconcile with the observation that deletion of MND2, which seems to have little effect on mitotic exit (Oelschlaegel et al., 2005 ; Page et al., 2005 ), suppressed the myo1Δ phenotypes as effectively as did mutations of other APC/C subunits (Figures 1, B and C, and and22A).
To explore this matter further, we wanted to delay mitotic exit by means independent of effects on the APC/C. To this end, we constructed strains in both backgrounds in which the nondegradable mitotic cyclin Clb2pΔDB (Amon et al., 1994 ) is expressed from the inducible GAL1 promoter. FACS analysis on asynchronous cultures showed that in glucose medium, APC/C-mutant strains showed a higher ratio of G2/M cells to G1 cells than did wild-type strains (Figure 4A and B, panels 1 and 4), indicating a delay in mitotic exit in the APC/C mutants. A shift to galactose medium produced a modest decrease in the G2/M-to-G1 ratio in these strains (Figure 4, A and B, panels 2, 3, 5, and 6), presumably reflecting a delay in cell-cycle initiation on the poorer carbon source (Pringle and Hartwell, 1981 ). In contrast, the PGAL-clb2ΔDB strains showed no accumulation of G2/M cells on glucose medium (Figure 4, A and B, panels 7), but, as expected, they showed a significant accumulation of such cells when production of Clb2pΔDB was induced by growth in galactose medium (Figure 4, A and B, panels 8 and 9). The apparent delay in mitotic exit was similar to (W303 background) or greater than (S288C background) the delay produced by APC/C mutations. However, in contrast to APC/C mutations, induction of Clb2pΔDB produced little or no suppression of either the myo1Δ clustering phenotype in the S288C background (Figure 4C) or the myo1Δ lethal phenotype in the W303 background (Figure 4, D and E). Thus, neither reduced degradation of the APC/CCdh1 target Clb2p nor the resulting delay in mitotic exit seems to suppress myo1Δ phenotypes. These results suggested that some other APC/C target(s) and/or pathway(s) might be involved in the suppression.
In parallel with the synthetic-lethal screen, we also screened for dosage suppressors of myo1Δ lethality in the W303 background (see Materials and Methods). From this screen, low-copy plasmids carrying IQG1 and CYK3 were isolated. Further investigation showed that both low-copy and high-copy IQG1 and CYK3 plasmids could indeed suppress myo1Δ inviability in the W303 background, as judged by tests either of spore viability (Figure 5A) or of the ability of vegetative myo1Δ cells to lose a MYO1 plasmid (Figure 5B). Sequencing of both genes in the W303 background revealed no mutations (our unpublished results), suggesting that the suppression of myo1Δ lethality was due simply to increased amounts of one or the other wild-type protein. In further support of this hypothesis, either low-copy or high-copy plasmids containing either IQG1 or CYK3 could also suppress the clustering phenotype of the myo1Δ mutant in the S288C background (Figure 5C). These results suggested that Iqg1p, Cyk3p, or both might be a previously unrecognized target(s) of the APC/C, so that APC/C mutations would suppress myo1Δ phenotypes by increasing the amount of that protein(s).
To ask whether Iqg1p or Cyk3p might be a target of the APC/C, we first searched their sequences for the known APC/C recognition motifs called the destruction box (DB) (Glotzer et al., 1991 ; King et al., 1996 ; Burton and Solomon, 2001 ; Burton et al., 2005 ), KEN box (Pfleger and Kirschner, 2000 ), A box (Littlepage and Ruderman, 2002 ), O box (Araki et al., 2005 ), and GxEN motif (Castro et al., 2003 ). No such sequences were found in Cyk3p; one 5-amino-acid sequence resembles the core of the O box, but the adjacent amino acids are very different. In contrast, two possible DBs, a possible KEN box, and a GKEN sequence (a possible KEN box or GxEN motif) were found in Iqg1p (Figure 6A). Thus, we focused on this protein. Wild-type cells arrested in G1 phase had significantly lower levels of Iqg1p and a significantly higher rate of Iqg1p turnover than cells arrested in M phase (Figure 6B; Figure 6C, top two panels), suggesting that the APC/C might target Iqg1p for degradation at the end of mitosis, as it does other proteins. Consistent with this hypothesis, we found that Iqg1p levels were significantly elevated in asynchronous populations of APC/C-mutant cells (Figure 6D) and that the rapid rate of Iqg1p degradation in G1 cells was eliminated in APC/C mutants (Figure 6C, bottom two panels). Moreover, in synchronous cultures, Iqg1p levels were found to increase before mitosis and drop precipitously after mitosis in wild-type cells (Figure 7, A and B, left), and the postmitotic decrease was largely eliminated in an APC/C mutant (Figure 7, A and B, right).
Finally, we also examined cells expressing Iqg1p-GFP from the chromosomal IQG1 promoter. As reported previously (Epp and Chant, 1997 ; Shannon and Li, 1999 ), Iqg1p localized to a ring at the mother-bud neck in medium-budded and large-budded wild-type cells. However, under the conditions used, the neck signal was rather weak (Figure 8A), and it was observed in only ~50% of the medium-budded and large-budded cells. Moreover, the Iqg1p-GFP did not seem to persist through cell division, because signal was never observed in unbudded cells (Figure 8B, left). In contrast, in APC/C mutants examined under identical conditions, Iqg1p-GFP signal was observed at the neck in a higher percentage (~80%) of medium-budded and large-budded cells, and it was significantly brighter there (Figure 8A). Moreover, Iqg1p-GFP seemed to persist through cell division in at least some cells, because patches of signal were sometimes observed in unbudded cells (Figure 8B, right).
Together, the results strongly suggest that Iqg1p is degraded in a cell-cycle-dependent manner that depends, directly or indirectly, on the APC/C.
To ask whether Iqg1p is a direct target of the APC/C, in vitro ubiquitination assays were performed using various fragments of Iqg1p and wild-type APC/C with its activator Cdh1p. Under conditions in which the well characterized APC/C substrate Pds1p was ubiquitinated effectively (Figure 9A, lanes 1 and 2), Iqg1p(1-750) and Iqg1p(33-250) were ubiquitinated in an APC/C-dependent manner (Figure 9A, lanes 3 and 4, and B, lanes 1 and 2), whereas Iqg1p(43-750), Iqg1p(400-1100), and Iqg1p(741-1495) were not (Figure 9A, lanes 5–10). These data indicated that Iqg1p is indeed a direct target of the APC/C and that a recognition sequence between amino acids 1 and 42 is required for this ubiquitination. Although this region contains no clear DB or other known APC/C-recognition sequence (see above), amino acids 34–42 have some resemblance (RxxxxxxxN) to previously characterized DBs (RxxLxxxxN/D/E), and mutating residues within this sequence reduced (single substitutions: Figure 9B, lanes 3–10) or eliminated (multiple substitutions: Figure 9B, lanes 11–14) APC/CCdh1-dependent ubiquitination. We also performed similar experiments using full-length Cyk3p. However, consistent with the lack of recognizable APC/C-recognition sequences in this protein, no ubiquitination was detected (data not shown).
To ask whether the apparent APC/CCdh1-recognition sequence in Iqg1p is also important for its turnover in vivo, we compared the stabilities of full-length Iqg1p and Iqg1p(43-1495) in both the W303 and S288C strain backgrounds. As expected, the truncated protein was substantially more stable both during prolonged arrest in G1 (Figure 9C) and at the end of mitosis in a synchronized population (Figure 9D).
The data presented above suggest that the elevated levels of Iqg1p resulting from APC/C mutations might explain the suppression of myo1Δ phenotypes by such mutations. In this case, synthesis at normal levels of a nondegradable Iqg1p should also be able to suppress myo1Δ phenotypes. In agreement with this prediction, we found that expression of Iqg1p(43-1495) from the chromosomal IQG1 promoter could suppress both the myo1Δ lethality in the W303 background (Figure 10A) and the myo1Δ clustering phenotype in the S288C background (Figure 10B). Somewhat surprisingly, segregants expressing the nondegradable Iqg1p in an otherwise wild-type background showed little or no defect in growth rate (Figure 10A) or in cell division, although subtle defects in septum formation were observed (our unpublished data). However, overexpression of the nondegradable Iqg1p from a GAL promoter did cause severe growth defects (our unpublished data).
Although an actomyosin contractile ring is involved in cytokinesis in most if not all animal and fungal cell types, it has recently become clear that actomyosin-ring-independent mechanisms are also critical, and in some cases sufficient, for cytokinesis (see Introduction). In an attempt to elucidate these mechanisms, we are performing genetic studies in S. cerevisiae. In this organism, the phenotypes associated with loss of the actomyosin ring (due to deletion of the single myosin II gene, MYO1) vary greatly in severity depending on the strain background, presumably due to differences among the strains in the efficiency of the actomyosin-ring-independent mechanisms, as we will discuss in more detail elsewhere.
During our studies, we made the surprising observation that mutations in genes encoding nonessential subunits of the APC/C could suppress both the lethal phenotype of myo1Δ in the W303 strain background and the slow-growth and delayed-cell-separation phenotypes of myo1Δ in the S288C strain background. Initially, we thought that this suppression probably resulted simply from changes in the timing of late-cell-cycle events; for example, a delay in mitotic exit due to APC/C malfunction might allow more time for an inefficient process of cytokinesis to be completed successfully. However, several lines of evidence (the suppression by cdh1 and mnd2 mutations; the lack of effect of a pds1 mutation or of expression of a nondegradable Clb2p; see Results for details) suggested strongly that the suppression was not due to changes in the timing of either anaphase or mitotic exit, and hence that it was likely to involve some novel APC/C target(s), pathway(s), or both. Indeed, we have now obtained strong evidence that suppression occurs because the APC/C defects result in an increased abundance of Iqg1p, a novel APC/CCdh1 target that is important in actomyosin-ring-independent cytokinesis, as discussed further below.
This model also provides a plausible explanation for the observation that a net1 mutation exacerbates the myo1Δ phenotype in the S288C strain background. We initially thought that an acceleration of mitotic exit due to loss of Net1p might allow insufficient time for an inefficient cytokinesis process to be completed. However, it now seems more likely that premature activation of Cdc14p resulting from the absence of Net1p leads to premature activation of Cdh1p (Visintin et al., 1998 ) and thus to targeting of Iqg1p for degradation by APC/CCdh1 before its role in cytokinesis can be completed. Of course, the effects of net1 mutations could also be more complex.
In a dosage-suppressor screen, we observed that even low-copy plasmids containing IQG1 or CYK3 could suppress the myo1Δ phenotypes in both the S288C and W303 genetic backgrounds. These observations suggested that Iqg1p, Cyk3p, or both might be previously unidentified targets of APC/C. Thus, mutations that compromised APC/C function might allow the intracellular concentrations of Iqg1p and/or Cyk3p to rise to levels that could suppress the myo1Δ phenotypes, by mechanisms that we will consider in more detail elsewhere. Although it remains possible that Cyk3p is an APC/C target, there is as yet no evidence to support this possibility: Cyk3p contains no clearly recognizable APC/C-recognition sequences, and we failed to detect ubiquitination in vitro by APC/CCdh1. In contrast, a variety of in vivo and in vitro experiments indicate that Iqg1p is a target of APC/CCdh1. First, in wild-type cells, both the levels of Iqg1p and the rates of Iqg1p degradation fluctuate during the cell cycle in a manner consistent with APC/C-triggered degradation at the end of mitosis. Second, in APC/C (including cdh1) mutants, the rates of Iqg1p degradation are drastically reduced; correspondingly, Iqg1p is present at higher levels and persists abnormally through cell division. Finally, in vitro assays showed that Iqg1p can be ubiquitinated directly by APC/CCdh1. The observation that the myo1Δ phenotypes can be suppressed by expression from the chromosomal IQG1 promoter of a stabilized Iqg1p (lacking the APC/CCdh1 recognition site) provides strong support for the hypothesis that the suppression of myo1Δ by APC/C mutations indeed results from the elevated and more persistent levels of Iqg1p in the mutant strains.
Interestingly, recognition of Iqg1p by APC/CCdh1 does not seem to depend on any of the sequences that correspond to previously characterized APC/C-recognition motifs. Instead, the ubiquitination of Iqg1p depends on a novel sequence (33-LRPQSSSKIN-42) near the N terminus that is similar but not identical to the consensus DB motif (RxxLxxxxN/D/E; Burton and Solomon, 2001 ). We observed only modest effects on ubiquitination when single residues in the Iqg1p sequence were replaced by alanine, but mutants in which positions 34, 37, and 42 were all altered showed an essentially complete loss of ubiquitination in vitro. Further dissection of the amino acids important for recognition of this site by APC/CCdh1 should shed light on the mechanisms of APC/C-target interaction. In the meantime, the results highlight the danger of attempting to identify APC/C targets solely on the basis of the known recognition motifs.
Another interesting conclusion from this study concerns the role of Mnd2p. Although this APC/C subunit seems to have little or no role in mitotic control (Oelschlaegel et al., 2005 ; Page et al., 2005 ), we found that an mnd2 mutation suppressed the myo1Δ phenotypes as strongly as did other APC/C mutations. This apparent discrepancy might be explained if particular APC/C subunits are important for the interactions with different specific targets. However, it also seems possible that the explanation might simply involve differential strengths of the interactions of the APC/C with different targets. In particular, because the APC/C roles in anaphase promotion and mitotic exit are essential for cell survival, the interactions with the targets relevant to these processes might be stronger than the interaction with a target like Iqg1p whose APC/C-mediated degradation is not essential for cell survival. The loss of an accessory subunit like Mnd2p might have little effect on the strong interactions but significantly compromise the weak interactions. An attraction of this model is that it might also help to explain the otherwise puzzling observation that loss of each of the nonessential APC/C subunits produces a very similar suppression of the myo1Δ phenotypes, even though each subunit seems to have a distinct role in APC/C function (Carroll and Morgan, 2002 ; Passmore et al., 2003 ; Schwickart et al., 2004 ; Carroll et al., 2005 ; Page et al., 2005 ).
Another interesting question concerns the mechanisms by which the APC/C, which resides predominantly or exclusively in the nucleus (Sikorski et al., 1993 ; Zachariae et al., 1996 ; Jaquenoud et al., 2002 ; Melloy and Holloway, 2004 ), ubiquitinates proteins, such as Hsl1p (Burton and Solomon, 2000 , 2001 ) and Iqg1p (this study), that seem to reside predominantly or exclusively in the cytoplasm (Epp and Chant, 1997 ; Barral et al., 1999 ; Shannon and Li, 1999 , 2000 ; Shulewitz et al., 1999 ; Longtine et al., 2000 ; Theesfeld et al., 2003 ). Although it remains possible that some APC/C acts in the cytoplasm, a more attractive model is that cytoplasmic APC/C targets must enter the nucleus in order to be ubiquitinated. At least for APC/CCdh1 targets, such nuclear entry may occur after the target is already bound to Cdh1p, a model supported both by studies on the order of assembly of APC/C-Cdh1p–substrate complexes (Burton et al., 2005 ) and by the observation that Cdh1p, unlike other APC/C components, shuttles in and out of the nucleus during the cell cycle (Jaquenoud et al., 2002 ). In this regard, it is also striking that a substantial fraction of the cytoplasmic Cdh1p is concentrated at the mother-bud neck (Jaquenoud et al., 2002 ), the very site to which both Hsl1p and Iqg1p are localized before their ubiquination-dependent degradation.
An interesting and unresolved question is the role of the APC/C-dependent proteolysis of Iqg1p, a question that is underscored by the observation that expression of the nondegradable Iqg1p from the chromosomal IQG1 promoter had little effect on the growth or division of wild-type cells. One possible answer involves the multiple roles of Mlc1p, which is a light chain for Iqg1p, Myo1p, and the type V myosin Myo2p, and whose interaction with Myo2p is critical for septum formation, cell separation, and new bud formation (Stevens and Davis, 1998 ; Boyne et al., 2000 ; Shannon and Li, 2000 ; Wagner et al., 2002 ; Luo et al., 2004 ). S. cerevisiae cells are highly sensitive to the cellular levels of this protein: heterozygous mlc1Δ diploid cells display haploinsufficiency that can be suppressed by reducing the copy number of MYO2 (Stevens and Davis, 1998 ), and overexpression of a Myo1p C-terminal tail (containing the Mlc1p-binding IQ motifs) leads to cytokinetic defects that can be alleviated by overexpressing Mlc1p (Tolliday et al., 2003 ). In addition, our own recent studies have shown that overexpression of wild-type Iqg1p from its own promoter on a high-copy plasmid in wild-type cells can cause abnormalities in cytokinesis (our unpublished results). Finally, overexpression of the nondegradable Iqg1p from the GAL promoter on a high-copy plasmid seems to cause severe growth defects (see above). Together, these results suggest that the association of Mlc1p with each binding partner may need to be fine-tuned temporally and that the timely APC/C-mediated degradation of Iqg1p may be important to release Mlc1p, particularly for interaction with Myo2p during secondary-septum formation and/or the formation of the new bud.
IQGAPs and the APC/C are both widely conserved, so it is possible that IQGAPs in other organisms are also regulated by the APC/C. This hypothesis is particularly attractive for Rng2p in Schizosaccharomyces pombe, IQGAP1 in mammalian cells, and PES-7 (F09C3.1) in Caenorhabditis elegans, because these proteins are also involved in cytokinesis (Eng et al., 1998 ; Wu et al., 2003 ; Skop et al., 2004 ). However, although each protein has one or more DB consensus sequences, none of them has a sequence corresponding to the novel APC/C-recognition motif identified in Iqg1p. The possible regulation of these proteins by the APC/C will require further investigation.
We thank I-Ching Yu for initiating the synthetic-lethal screen; Erfei Bi, Forrest Spencer, Phil Hieter, Erin O'Shea, Jonathan Weissman, and John Cooper for strains, plasmids, and antibodies; Steve Haase and members of his laboratory for help with the FACS analysis; and Akanksha Gangar for a protein extraction protocol. We also thank Danny Lew, Pat Brennwald, Steve Haase, Matt Sullivan, and members of their and our own laboratories for helpful suggestions and support. This work was supported by National Institutes of Health grants GM-31006 (to J.R.P.), GM-53270 (to D.O.M.), and GM-76200 (to M.J.S.) and by March of Dimes grant 0455851T (to M.J.S.). R.N. was supported in part by a long-term fellowship from the Human Frontier Science Program.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E07-05-0509) on October 17, 2007.
The online version of this article contains supplemental material at MBC Online (http://www.molbiolcell.org).