Iqg1p links spatial and secretion landmarks to polarity and cytokinesis

A critical step in eukaryotic cell division involves the correct positioning of the division plane, which in turn determines the segregation of one intact genome into each cell progeny. Eukaryotic organisms utilize diverse mechanisms to define their division planes. The budding yeast, Saccharomyces cerevisiae, uses cortical cues to mark the division site by initiating a bud early in G1, thereby committing to its division axis (Chant, 1996). Haploid yeast divide in such a manner that a bud is formed next to the previous division site, thus resulting in an axial budding pattern (Chant and Herskowitz, 1991). It is believed that a septin ring marks the division site at G1 and persists to assemble other axial markers as well as proteins involved in cytokinesis (for review see Chant, 1996, 1999). Genetic analyses have identified Bud3p, Bud4p, Axl1, and the transmembrane protein Bud10p as axial markers involved in bud-site selection (for reviews see Sanders and Fields, 1995; Chant, 1996, 1999; Madden and Snyder, 1998). The molecular mechanism by which the septin ring localizes to the bud site to position these bud site selection markers (future division-site) is poorly understood.

Once a bud site is selected, polarized secretion is directed to that site (Lew and Reed, 1995). Fusion of the secretory vesicles at the target site requires a protein complex, the exocyst (Finger et al., 1998), composed of the Sec3, -5, -6, -8, -10, and -15 proteins. Some of the exocyst members, such as Sec3p and Sec4p, localize to the bud tip (Bowser et al., 1992; TerBush and Novick, 1995). After nuclear division, the exocyst reorients to the mother bud neck to promote cytokinesis (Finger et al., 1998). Thus, budding and cytokinesis both involve directed secretion. Furthermore, some sec3 mutants exhibit a random budding pattern, suggesting the involvement of the complex in bud-site selection (Finger et al., 1998). Nevertheless, no direct molecular link between the exocyst and the bud site selection proteins has yet been reported.

Sec3p localizes to growth sites independent of the other exocyst components, actin, or septins, indicating that Sec3p works as a landmark for secretion (Finger et al., 1998). The polarized localization of Sec3p has been suggested to require the kinase Cdc28p (Finger et al., 1998), as well as the small GTPases Rho1 (Guo et al., 2001) and Cdc42p (Zhang et al., 2001). It is also believed that the positional signal imposed by the septin and bud site selection proteins is interpreted by Cdc42p and other polarity establishment proteins to polarize the actin cytoskeleton (Johnson and Pringle, 1990) and the secretory pathway (Finger et al., 1998). The Cdc42 effector(s) that mediates these functions and the mechanism by which it is achieved remain important questions. However, one intriguing possibility for the Cdc42 effector is a member of the family of IQGAPs.

We and others isolated the mammalian IQGAPs-1 and -2 as putative target/effectors for Cdc42p (Hart et al., 1996; McCallum et al., 1996; Erickson et al., 1997). The mammalian IQGAPs were localized to cell–cell junctions (Hart et al., 1996; Kuroda et al., 1996; Bashour et al., 1997), as well as to Golgi membranes (McCallum et al., 1998), with the latter finding suggesting their possible involvement in protein trafficking events. To better understand the cellular functions of the IQGAP family of proteins, we isolated and characterized the yeast homologue, Iqg1p, and found that the IQG1-null strain produced phenotypes consistent with the involvement of Iqg1p in both polarity and cytokinesis (Osman and Cerione, 1998). Other groups have shown Iqg1 (Cyk1p) to be essential (Epp and Chant, 1997; Lippincott and Li, 1998) and to participate in an actomyosin-based contractile ring function. However, because the actomyosin ring was found to be dispensable for both cytokinesis and yeast cell growth (Bi et al., 1998), Iqg1p must play additional roles in the cell. Here we report that Iqg1p determines the axial budding pattern, interacts with and promotes the localization of the axial markers Bud4p and Cdc12p, and functionally interacts with the secretion marker, Sec3p. Overall, these findings raise the interesting possibility that the Cdc42p target, Iqg1p, serves to interface proteins involved in a key polarity-dependent process (axial budding) with proteins involved in exocytosis/secretion and cytokinesis.

To investigate the role of Iqg1p in polarity and cytokinesis, we used the full-length IQG1gene in a two-hybrid system (James et al., 1996) to isolate candidate binding partners. In this study, we focus on one clone we identified as BUD4. The region of Bud4p that spans amino acids 769–880 appears to be the interacting domain for Iqg1p (Iqg1-Bud4 interacting domain [IBID];^*Fig. 1 A). This region was isolated four times from the two-hybrid library and does not interact with control proteins such as lamin or with the Gal4-binding domain alone. Bud4p was previously identified as an essential protein for axial budding (Chant and Herskowitz, 1991; Chant and Pringle, 1995; Sanders and Herskowitz, 1996). It contains a putative GTP binding motif at its COOH terminus overlapped by a pleckstrin homology domain. At position 819, and overlapped by the IBID, Bud4p has an RGD motif (Fig. 1 A), denoting a specific cell binding tripeptide motif found on adhesive proteins such as fibronectin. Different domains of Iqg1p (Fig. 1 B) were assayed for their ability to interact with Bud4p, and the results indicated that the NH₂-terminal half of Iqg1p (residues 1–823) contains the Bud4-binding site (Table I).

To establish whether Iqg1p binds Bud4p in cells, we transformed a high-copy HA-tagged IQG1 plasmid into the yeast strain (MO3) lacking the chromosomal copy of the IQG1 gene (see Materials and methods). We also cotransformed the same strain with both the HA-tagged IQG1 plasmid and the Gal4-binding domain-tagged BUD4 plasmid (encoding the IBID) that was isolated from the two-hybrid screen described above. As a control, MO3 cells were also transformed with the vector encoding HA alone, or with both the HA-encoding vector and the Gal4-tagged BUD4 plasmid containing the IBID domain (Fig. 1 A). The total protein extract was used to isolate immune complexes of either HA-tagged Iqg1p or Gal4-IBID using a monoclonal α-HA antibody (BAbCO) or an α-Gal4 antibody (Santa Cruz Biotechnology, Inc.), respectively. Western blot analysis was performed to detect the presence of Bud4p and HA-tagged Iqg1p in the immune complexes. Fig. 1 C (top) shows that HA-tagged Iqg1p was efficiently coimmunoprecipitated with Gal4-IBID (lane 2), whereas Iqg1p was not coimmunoprecipitated with the Gal4 binding domain alone (lane 5). The bottom panel shows that the endogenous Bud4p coimmunoprecipitated with full-length HA-tagged Iqg1p (lane 2) but not with HA alone (lane 4). As additional controls, we also used other antibodies, such as α-Intersectin and α-GFP, and found that these were unable to coimmunoprecipitate either HA-Iqg1 or Bud4p (unpublished data).

To assess whether Iqg1p, like Bud4p, influences the budding pattern, we compared the pattern of the bud scars on the surfaces of haploid and homozygous diploid cells lacking the IQG1 gene with their isogenic wild-type counterparts. Chitin rings were visualized using Calcofluor as described in the Material and methods. Interestingly, the majority of the haploid cells lacking Iqg1p (∼60%, n = 400) exhibited a bipolar budding pattern (Fig. 2 A, A–F) similar to cells lacking BUD4 (Fig. 2 A, top right). This budding pattern was not observed in the isogenic wild-type cells which showed axial budding (Fig. 2 A, top left), but it appears similar to patterns observed in cells defective in BUD3 or BUD4 (Chant and Herskowitz, 1991). Approximately 30% of the iqg1Δ cells exhibited a semirandom pattern of budding such that the scars were clustered but not located at the opposite poles (Fig. 2 A, C and D). Another subset of these cells (Fig. 2 A, E and F) showed a distribution of bud scars to the opposite poles but these scars were scattered at the pole and not lined up as in the wild-type cells. Homozygous diploid strains lacking IQG1 showed the expected bipolar pattern of diploid budding (unpublished data).

To determine whether Iqg1p is required for the proper localization of Bud4p to growth sites, we used an affinity-purified Bud4p antibody, a gift from Dr. S. Sanders (Massachusetts Institute of Technology, Boston, MA) (Sanders and Herskowitz, 1996) to compare the localization of Bud4p in wild-type and iqg1Δ cells. As expected, in 75% of wild-type cells (n = 400) Bud4p localized in two rings at the mother–daughter junction and as a ring in unbudded cells (Fig. 2 B, top left; Sanders and Herskowitz, 1996). In 65% of the iqg1Δ cells (n = 400), Bud4p localization at the neck was either totally abolished or significantly diminished (Fig. 2 B, iqg1-labeled panels). In ∼4% of the cells, we detected only one faint ring in the mother cell but not in the daughter cell. In a few cases, we observed a diffuse pattern throughout the cell. However, in no case did we observe a wild-type double ring. These results suggest that Iqg1p is needed to recruit Bud4p to the growth sites.

Bud4p was shown to genetically interact with and require the septin, Cdc12p, for localization (Sanders and Herskowitz, 1996). This prompted us to consider the relationship between Iqg1p and Cdc12p. We first used two-hybrid assays to examine the ability of Iqg1p and Cdc12p to interact. These data showed that Cdc12p interacted with the COOH-terminal region of Iqg1p that contains the RasGAP-like domain (Table II). However, thus far, we have not been able to detect the coimmunoprecipitation of a GFP-tagged Cdc12p on a low-copy plasmid with the HA-tagged Iqg1p on a high-copy plasmid (Fig. 3 A, top, lane 2), suggesting that Iqg1p and Cdc12p may undergo a relatively weak or transient interaction. We detected a band for the endogenous Bud4p after immunoprecipitation of Cdc12p-GFP (Fig. 3 A, bottom, lane 2, compare with lane 5), suggesting that these two proteins are able to interact in cells and may bridge the interaction of Iqg1p with Cdc12p.

To further explore the significance of the septin-Iqg1p interaction suggested from the two-hybrid experiments, we examined the localization of Cdc12p in iqg1Δ cells, using a Cdc12-GFP construct encoded by either a low-copy (Fig. 3 B, top) or a high-copy vector (Fig. 3 B, bottom). Overall, there was a significant decrease in the intensity of the septin ring in iqg1Δ cells (Fig. 3 B, bottoms) compared with wild-type cells (left). The septin ring was completely absent from some iqg1Δ cells (bottom right). At 23°C, the permissive temperature for iqg1Δ cells, we detected a faint septin ring in 69% of these cells (n = 102), whereas a brighter septin ring was observed in 87% of the isogenic wild-type cells (n = 112; unpublished data). At 30°C, the semipermissive temperature for iqg1Δ cells, we detected a faint septin ring in only 27% of the cells (n = 365) compared with 85% of the wild-type cells (n = 129). In contrast, the localization of Cdc12p was not affected in bud4Δ cells (unpublished data) indicating that only Iqg1p (and not Bud4p) affects Cdc12p localization.

Sec3p is involved in both bud-site selection and cytokinesis (Haarer et al., 1996; Finger and Novick, 1997). Our previous work implicating Iqg1p in cytokinesis and protein trafficking (Osman and Cerione, 1998), together with the results from this study suggesting that Iqg1p is involved in bud site selection, led us to examine whether there is a functional interaction between Iqg1p and Sec3p. To determine whether Iqg1p associates with Sec3p, we cotransformed the MO3 strain lacking Iqg1p with the high-copy plasmid encoding HA-tagged Iqg1p (which complements the iqg1Δ phenotype), together with either a SEC3-GFP on low-copy plasmid (Finger et al., 1998) that complements the sec3Δ strains, or the GFP plasmid. Coimmunoprecipitation experiments showed that HA-Iqg1p associated with Sec3p-GFP (Fig. 4 A, lane 2) but not with GFP alone (lane 4), nor was HA-Iqg1p immunoprecipitated with other control antibodies (e.g., α-Intersectin or α-Gal4 antibodies; unpublished data).

To examine the localization of Sec3p in the iqg1Δ strain, we transformed the SEC3-GFP construct (Finger et al., 1998) into the haploid segregants resulting from one of the tetratype tetrads obtained from the cross between sec3Δ and iqg1Δ (see below and Table III). In 65% of the wild-type cells (n = 400 cells), Sec3p-GFP localized to the cell poles and as two rings at the mother–daughter neck (Fig. 4 B, left, arrows), as previously reported (Finger et al., 1998). In ∼65% of the wild-type cells that appeared to be in G1we also observed two or more cortical Sec3-GFP loci and often a punctate pattern (Fig. 4 B; unpublished data). However, in iqg1Δcells (>400 cells examined), Sec3p-GFP localization at growth sites was abolished and was diffusely distributed throughout the cell (Fig. 4 B, right). Thus, these results suggest that Iqg1p is required for the localization of Sec3p to the sites of polarized growth.

To begin to investigate the cellular function of the Iqg1p-Sec3p interaction, we crossed BHY51, which harbors the temperature-sensitive sec3Δ (Haarer et al., 1996), with the temperature-sensitive iqg1Δ (MO3) strain. The resultant diploid was sporulated and twenty-one tetrads were dissected. Each produced four live spores at room temperature. When scored for temperature sensitivity, we obtained 4 PD: 4 NPD: 12 TT tetrads. The tetratype tetrads that produced three temperature-sensitive spores at 37°C were analyzed by PCR to identify the progenies (MOB1–4) for subsequent comparative studies (Table III). In every case examined, one of the temperature-sensitive spores contained the double deletion. The growth and the phenotypes of the segregants from the same tetrad were examined. Compared to wild-type cells (Fig. 5, top left), sec3Δ cells showed normal budding and they were temperature-sensitive at 37°C (Fig. 5, middle left). The iqg1Δ cells displayed elongated buds (Fig. 5, bottom) and they were temperature-sensitive at 37°C as we previously reported (Osman and Cerione, 1998). In contrast, the double mutant cells exhibited a lower nonpermissive temperature (30°C). At room temperature, the iqg1Δ sec3Δ double mutant cells displayed a mixture of budding phenotypes, all consistent with polarity defects (Fig. 5, a–e). These double mutant cells were generally large with wide necks. Some cells showed a bottle-like shape with an elongated mother containing a wide constriction in the middle, often with a small bud at one side or at one end (e and f). Other cells had two buds, where in many cases both buds originated from the mother cell (b, c, and h), and in some cases, a new bud extended from the older bud (d, g, and i).

Because failure of targeted secretion results in more generalized chitin deposition, we used Calcofluor to analyze the secretion of cell wall material in these mutants. Fig. 6 A shows that in the double mutant cells, chitin is uniformly localized in the mother cell wall, but was not directed to the small bud (6 A, bottom right, arrows). This was not the case in iqg1Δ single mutants where chitin was uniformly distributed in both the mother cells and the buds (Osman and Cerione, 1998; unpublished data). In sec3Δ cells, we have consistently observed that chitin is only slightly diffuse (Fig. 6 A, top right) but correctly directed to the buds and the bud scars (Haarer et al., 1996; unpublished data).

We also used Calcofluor to measure the budding pattern on the cell surfaces of the haploid progenies described above. Surprisingly, unlike each of the single mutants (Fig. 2 A; unpublished data), the haploid double mutant iqg1Δ sec3Δ cells all exhibited random budding patterns (Fig. 6 B, A–E) reminiscent of BUD1 or BUD2 defective cells (Chant and Herskowitz, 1991). Only homozygous diploid strains of sec3Δ were previously shown to exhibit a random budding (Haarer et al., 1996; Finger and Novick, 1997). However, the results presented in Fig. 6 B suggest that Sec3p affects the budding of haploid cells as well.

Because Iqg1p binds and helps localize both Sec3p and Bud4p, we further examined the relationship between Bud4p and Sec3p in a number of different ways. First, we took a genetic approach and crossed BHY51 (sec3Δ) with SY298 (bud4Δ). The resultant diploid was sporulated and tetrads yielded four live progenies analyzed by PCR to determine their genotypes. The segregants of a single tetratype tetrad (MO1A–D; Table III) were used for further analyses. Some (∼10%) of the bud4Δ single mutant segregants from this cross revealed a phenotype consistent with a cytokinesis defect (Fig. 7 A, top right). Interestingly, ∼30% of the double mutant bud4Δ sec3Δ cells displayed chains of 3–5 or more elongated cells (Fig. 7 A, bottom right) that were often branched and remained together after sonication, vigorous vortexing or zymolyase treatment. Another 35% of the double mutant cells displayed shorter chains or appeared as two cells which failed to separate. Moreover, at 30°C, the double mutant cells showed substantial lysis, which was not the case for the single mutants.

Second, we examined the localization of Sec3p in bud4Δ cells. The localization of Sec3p-GFP in wild-type cells (Fig. 7 B, middle, inset) was detected in 60% of the total population examined (n = 200); however, its localization was diffuse in bud4Δ cells (Fig. 7 B, left). The localization of Sec3p at the incipient bud was undetectable. In only 3% of the cells (n = 200) were we able to detect localization of Sec3p-GFP at the mother–bud junction; however, even in these cases, the localization of Sec3p was either extremely faint or distorted in such a way that we did not observe a double ring (Fig. 7 B, right, arrow).

To examine whether Bud4p and Sec3p also cooperate in determining the axial budding pattern, we measured the pattern of bud scars on the cell surface of the bud4Δ sec3Δ haploid strain. Approximately 70% of the double mutants displayed bipolar budding pattern (Fig. 7 C, right, chain of cells), similar to that of bud4Δ single mutants. However, in ∼60% of those cells, we observed a branching scar that appeared on the surface of the double mutant cells (Fig. 7 C, left, arrows). This scar appears to be the point of a branch of a shorter chain of cells (Fig. 7 A, bottom right, arrow).

Finally, we investigated whether Sec3p-GFP on a low-copy plasmid (Finger et al., 1998) coimmunoprecipitates with endogenous Bud4p. Fig. 7 D shows that a Bud4p band was detected in the immune complexes with Sec3p-GFP (lane 2) but was not detected when immunoprecipitation was performed with control antibodies or with GFP alone (Fig. 7 D, lane 4).

To study the subcellular effects of the double deletion of IQG1 and either BUD4 or SEC3, we examined the cells by thin-section electron microscopy after shifting to the nonpermissive temperature (37°C) for 30 min. As reported previously (Finger and Novick, 1997), Sec3Δ cells accumulated vesicles all over the mother cell and the bud (Fig. 8, bottom right). Interestingly, in both of the double-deletion strains (n = 150 each), these vesicles were either no longer visible or in some cases, fused into larger vesicular structures, especially in bud4Δ sec3Δ cells (Fig. 8 g, arrows). In contrast to wild-type and single-mutant cells, the double-mutant cells showed septal defects and in some cases, broad necks. In ∼58% of the iqg1Δsec3Δ cells (n = 100), the septa were either absent (Fig. 8, a and b) or aberrant (panels c and d). In bud4Δsec3Δ cells, the picture was slightly different. We either detected no septa (Fig. 8, e and g) or, in ∼3% of the cells, we detected septa that remained intact in a broad neck (Fig. 8, f, arrows, and h), especially within the chains of cells. These defects suggest that Iqg1p, Bud4p, and Sec3p all contribute to proper septum deposition and separation.

From the same two-hybrid screen described earlier, we also isolated the Myosin light chain protein, Mlc1p, using full-length Iqg1p as bait. Table IV shows the mapping of the interaction of Iqg1p with Mlc1p, compared to its interaction with Calmodulin, Cmd1p. Mlc1p interacts with the region harboring the IQ motifs of Iqg1p, whereas Cmd1p interacts with a region within the NH₂-terminal half of Iqg1p distinct from the IQ motif. These in vivo results are consistent with the in vitro results reported by Shannon and Li (1999)(2000). In addition, we detected a two-hybrid interaction between Mlc1p and Bud4p, although thus far, we have been unable to detect this interaction in coimmunoprecipitation experiments.

We have previously shown that Iqg1p influences polarity, cytokinesis and protein trafficking (Osman and Cerione, 1998). Here we present data addressing the mechanism by which Iqg1p influences these processes. We have identified Iqg1p binding partners that function together in axial bud site selection and cytokinesis. Given that Iqg1p serves as a target/effector for Cdc42p (Osman and Cerione, 1998), these results also provide additional molecular evidence for the involvement of Cdc42p in coordinating polarity establishment and bud site selection with cytokinesis.

The axial budding pattern in haploid yeast is spatially programmed by Bud3p, Bud4p, Axl1p, and the transmembrane protein Bud10p (for review see Chant, 1999). Using a two-hybrid screen, we have identified Bud4p as an Iqg1p-binding partner (Table I). We further demonstrated that Iqg1p was co-immunoprecipitated with Bud4p (Fig. 1 C) and was required for its localization (Fig. 2 B). We showed that, like Bud4p, Iqg1p specifies axial budding in yeast. Haploid cells lacking IQG1 displayed bipolar budding (Fig. 2 A) similar to bud4Δ mutants, while the isogenic homozygous diploid cells were unaffected. This axial budding defect (Fig. 2 A) suggests that Iqg1p works together with Bud4p to specify axial bud site selection. However, the fact that Bud4p is localized as a spatial marker in the previous cell cycle, and loses this localization at the neck in iqg1Δ cells (Fig. 2 B), supports the idea that Iqg1p recruits Bud4p as an axial marker.

Iqg1p also appears to interact with and helps localize the septin Cdc12p (Table II; Fig. 3 B), which is necessary for both cytokinesis and axial budding (for review see Chant, 1996, 1999; Madden and Snyder, 1998). Bud4p may mediate the interaction between Iqg1p and Cdc12p, as we have detected the coimmunoprecipitation of Cdc12p and Bud4p, but have not yet been able to detect an interaction between Cdc12p and Iqg1p in coimmunoprecipitation experiments (Fig. 3 A). However, we have found that Iqg1p, and not Bud4p, affects the efficient localization of the septin Cdc12p to the neck (Fig. 3 B; unpublished data), further supporting the idea that Iqg1p assembles the proteins required for axial budding. In addition, the interaction between Iqg1p, Bud4p, and Cdc12p may involve a cooperation in some aspect of cytokinesis, as both Iqg1p and septins have been previously implicated in this process.

Several lines of evidence presented in this study indicate that Iqg1p and Sec3p work together to influence cell polarity, axial budding and cytokinesis in yeast. We have provided evidence that Iqg1p coimmunoprecipitates with (Fig. 4 A) and is required for the localization of Sec3p (Fig. 4 B). The iqg1Δsec3Δ double mutants displayed a lower restrictive temperature and exhibited phenotypes consistent with a polarity defect (Fig. 5) that resulted in random budding (Fig. 6 B). These double mutants also failed to direct growth material to the new bud (Fig. 6 A) and to form correct septa at the mother-daughter junctions (Fig. 8).

Together, these results suggest a role for Sec3p in axial budding. Although earlier studies have suggested that Sec3p is primarily involved in diploid budding (Haarer et al., 1996; Finger and Novick, 1997), our findings are consistent with the idea that targeted secretion is required for budding in both cell types (for review see Finger and Novick, 1998). Our data also suggest that Iqg1p works with Sec3p to maintain cell polarity. Figs. 5 and 6 B show that budding in the iqg1Δ sec3Δ double mutant strain was initiated normally but was not maintained such that another bud originated either from the mother, the bud (Fig. 5) and/or at a random location (Fig. 6 B). This may reflect the actions of polarity establishment proteins such as Cdc42p and Bud1p. Iqg1p may mediate the interplay between the bud site selection function of the Bud1 GTPase and the maintenance of polarity by Cdc42p (Fig. 9). Three independent pieces of evidence seem to support this view. First, the random budding phenotype of the iqg1Δ sec3Δ double mutant cells (Fig. 6 B) is similar to that of bud1Δ or bud2Δ mutant cells (Chant and Herskowitz, 1991). Second, analyses of bud1Δ cdc24Δ cells revealed that the essential role of the polarity establishment molecules in morphogenesis is to stabilize the axis of polarity, and that in their absence, this axis wanders (Nern and Arkowitz, 2000). Third, a recent study demonstrated direct binding between Sec3p and Cdc42p (Zhang et al., 2001) and showed that Sec3p was mislocalized in cells expressing certain mutations of Cdc42p. Together, these data support the idea that the Cdc42p target Iqg1p coordinates the positional signal for budding and secretion with the signal to maintain polarity and promote cytokinesis.

There also appears to be a functional interplay between Bud4p and Sec3p in budding and cytokinesis. Bud4p can be coimmunoprecipitated with Sec3p (Fig. 7 D) and the proper localization of Sec3p is nearly abolished in the absence of Bud4p (Fig. 7 B). Furthermore, the bud4Δsec3Δ double mutants displayed a lower nonpermissive temperature and exhibited budding defects (Fig. 7 C). In addition, the phenotype of bud4Δsec3Δ cells highlights a possible role for Bud4p in cytokinesis (Fig. 7 A), apparently, in the final stages of cell separation (Fig. 8). Iqg1, Bud4p and Sec3p may work together to ensure proper septum formation or deposition, thus explaining the wide and aberrant neck formation reported in some sec3Δ cells (Finger and Novick, 1997), as well as that observed in sec3Δ bud4Δ cells, or the absence of septa in iqg1Δ sec3Δ cells (Fig. 8). Overall, our findings appear to be consistent with the “cytokinesis tag” model for axial budding. This model predicts that components involved in cytokinesis are also involved in selecting the future bud-site. The subsequent activation of the Bud1 GTPase then directs the assembly and the initiation of the new bud at that site (for review see Madden and Snyder, 1998).

In conclusion, we propose that Iqg1p forms a targeting patch that includes Bud4p, Cdc12p, and Sec3p (Fig. 9). This protein complex may play a primary role in cytokinesis by first determining, and then maintaining, the axis of polarity by choosing a bud site. The complex would recruit the exocyst for targeted secretion and other morphogenetic factors such as Mlc1p and Cmd1p (Table IV) to promote bud growth and eventually lead to septum deposition and separation. In this way, the Iqg1 complex would in effect serve as a checkpoint for cytokinesis by preventing another round of budding until cytokinesis is complete (Fig. 9). In the absence of the Iqg1p-targeting complex, alternative pathways for budding can apparently bypass the checkpoint leading to rounds of budding and chains of cells observed in the double mutants.

The PCR-amplified IQG1 gene was cloned in frame into pGBD-C2 to produce the Gal4 binding domain Iqg1p fusion protein. This plasmid was used to screen three yeast genomic libraries, YL2H-C1-3, using the yeast strain PJ694A (James et al., 1996). The interacting library plasmid was subsequently transformed into Escherichia coli, purified, and sequenced to identify Iqg1p binding partners. The Iqg1p domain deletions shown on Fig. 1 B were generated by PCR and cloned into pGBD-C2 to further delineate the binding domain for Bud4p, Cdc12p, Mlc1p, and Cmd1p.

For coimmunoprecipitation experiments, tagged plasmids were cotransformed into a strain lacking the chromosomal copy of IQG1. To examine the coimmunoprecipitation of Iqg1p and Bud4p, cells were grown in synthetic complete media (cm) lacking tryptophan and leucine after transformation with HA-tagged Iqg1p on a high copy plasmid (pA1; Osman and Cerione, 1998) and Gal4-IBID (amino acids 769–880 from Bud4p, Fig. 1 A). Cells harboring HA-Iqg1p (pA1) alone were grown in cm-leucine. For control experiments, cells were transformed with parent plasmids lacking the relevant gene. When examining the interaction between Iqg1p and Cdc12, MO3 cells were transformed with CDC12-GFP on a low copy plasmid (described below) and HA-IQG1, and grown on cm-uracil and leucine. When examining the coimmunoprecipitation of Iqg1p and Sec3p, MO3 cells transformed with SEC3-GFP on a low-copy plasmid (Finger et al., 1998) and HA-IQG1 were grown in cm media lacking uracil and leucine. In all cases, the cell cultures were grown to saturation and the cell pellets washed with and suspended in cold IP-buffer (50 mM Tris, pH 7.5, 1 mM EDTA, 5% glycerol, 0.1% Triton, 1 mM PMSF and fungal protease inhibitor cocktail [Sigma-Aldrich]). Acid-washed glass beads (Sigma-Aldrich) were added to break the cells by vortexing and the cell lysates were collected after centrifugation at 20,000 rpm at 4°C. Two mL of the total cell lysate (scaled up for Sec3p-Iqg1p experiments) was incubated with antibodies (a control antibody was also included) for 3 h with gentle rocking at 4°C. 4 mg of IP buffer-washed protein A Sepharose (Sigma-Aldrich) was added to the immune complexes and incubated overnight at 4°C with gentle rocking. The reaction mix was centrifuged at 10,000 rpm for 30 s and the bead portion was washed four times with ice-cold IP buffer, suspended in 30 μL SDS-loading buffer, boiled at 100°C and separated by 7-10% SDS-PAGE. Western blot analyses were performed using standard methods.

To visualize chitin deposition and bud scars on the yeast cell wall, cells were collected at log phase and stained with Calcofluor (Fluorescent Brightener; Sigma-Aldrich) as described (Pringle, 1991). Cells with three or more bud scars were scored for budding pattern. For Cdc12p localization, a CDC12-GFP fusion gene was constructed by using PCR to fuse an enhanced version of the GFP (Cormack et al., 1997) to the carboxyl terminus of Cdc12p. The CDC12-GFP gene was cloned into the low-copy vector YCplac111 and also into the high-copy vector YEplac181 (Gietz and Sugino, 1988). Wild-type (MO5) and iqg1Δ cells (MO2) carrying the indicated CDC12-GFP plasmid, or a control vector, were grown overnight at 23°C and then adjusted to 10⁶ cells/mL and shifted to 30°C for 6 h. The cells were then examined using an Olympus BH2 fluorescence microscope, photographed with Kodak TMAX400 film, and then the negatives were scanned into digital images under identical conditions. By comparing different exposures, we found that it requires at least a fourfold longer exposure to see similar neck staining in MO2 than in MO5. CDC12-GFP fluorescence was better at 23°C in the MO2 strain, but was still weaker than the wild-type strain.

Strains resulting from single tetratype tetrads of the cross between either SY298 (bud4Δ) or MO3 (iqg1Δ) with sec3Δ were transformed with a SEC3-GFP plasmid (Finger et al., 1998) which encodes GFP fused to the COOH terminus of Sec3p. Log phase cells growing in cm-uracil liquid media were processed as described in (Finger et al., 1998) and visualized under 100× using a Zeiss fluorescence microscope. Images were collected using Axiovision software under identical conditions.

Log phase overnight cultures of wild-type (MOB2 or MO1B), sec3Δ (MOB1), iqg1Δ (MOB4), bud4Δ (MO1A), iqgΔsec3Δ (MOB3), and bud4Δsec3Δ (MO1D) strains growing at room temperature were adjusted to 0.3 A₆₀₀nm units/mL in YEPD and incubated at 37°C for 30 min. Cultures were directly fixed in 0.1 M cacodylate containing 2.5% glutaraldehyde and 2.5% paraformaldehyde for 2 h. Cells were treated with 0.2 mg/mL zymolyase 100T in 0.1 M KPi, pH 7.5. The cell pellets were incubated with ice-cold OsO₄ in 0.1 M cacodylate for 1 h, washed twice with water and incubated for 1 h in 1.5 mL of filtered 2% uranyl acetate at room temperature. A series of ethanol concentrations (50, 70, 90, and 100%) were used to dehydrate the cells followed by acetone treatment. The cells were then incubated for a few hours in 50% acetone/50% SPURR (Electron Microscopy Sciences), changed to 100% SPURR, and incubated overnight at room temperature. After changing to SPURR 2×, cells were baked at 80°C for 24 h. This and the rest of the sectioning and processing were carried out at the Cornell Integrated Microscopy Center (Ithaca, NY).

We thank Drs. Sylvia Sanders for Bud4 strains and antibody, Brian Haarer for strains and plasmids, and Peter Novick for the SEC3-GFP construct. We also thank Drs. Juan Lucas and Sigrid Holmgren for helping with tetrad dissection, Dr. David Holowka for critical reading of the manuscript and helpful comments, and Drs. David Pruyne for helpful discussion, and Jianbin Wang for the α-Intersectin antibody used for controls. Thanks to Cindy Westmiller for expert technical assistance. This work was supported by NIH grant RO1 GM47458. MAO was supported by an NIH Training Grant T32 CA09682.