Cdc42 Interacts with the Exocyst and Regulates Polarized Secretion

Polarized delivery and incorporation of proteins and lipids to specific domains of the plasma membrane is fundamental to a wide range of biological processes such as neuronal synaptogenesis and epithelial cell polarization. The exocyst complex is specifically localized to sites of active exocytosis and plays essential roles in secretory vesicle targeting and docking at the plasma membrane. Sec3p, a component of the exocyst, is thought to be a spatial landmark for polarized exocytosis. In a search for proteins that regulate the localization of the exocyst in the budding yeast Saccharomyces cerevisiae , we found that certain cdc42 mutants affect the polarized localization of the exocyst proteins. In addition, we found that these mutant cells have a randomized protein secretion pattern on the cell surface. Biochemical experiments indicated that Sec3p directly interacts with Cdc42 in its GTP-bound form. Genetic studies demonstrated synthetic-lethal interactions between cdc42 and several exocyst mutants. These results have revealed a role for Cdc42 in exocytosis. We propose that Cdc42 coordinates the vesicle docking machinery and the actin cytoskeleton for polarized secretion.

exocyst 8 . Based on these observations, it was proposed that Sec3p represents a spatial landmark for polarized secretion 5,8 . In addition to Sec3p, several other components of the yeast exocytic machinery are also localized to sites of polarized growth. For example, Sec4p, the Rab protein essential for post-Golgi secretion, is localized to exocytic sites through its association with secretory vesicles, which are transported via the actin cytoskeleton 31 . Another exocyst protein, Sec15p, associates with secretory vesicles and interacts with the GTP-bound form of Sec4p.
Sec4p promotes protein-protein interactions among the exocyst components, which then link Sec15p to Sec3p 10 . The assembly of the exocyst complex thereby tethers secretory vesicles to specific sites on the plasma membrane for subsequent membrane fusion.
Proper localization of the exocyst complex is important for spatial regulation of secretion. To identify proteins that regulate the polarized localization of the exocyst, we examined the localization of GFP-tagged exocyst proteins in various classes of mutants. Here we report that the polarized localization of the exocyst is controlled by Cdc42. Cdc42 is a member of the Rho family of small GTP-binding proteins, which are master regulators of many cellular activities including polarization, morphogenesis, membrane traffic, cell growth and development 4,15,16,17 . We found in certain cdc42 mutants that the exocyst proteins were depolarized and the exocytosis pattern was randomized. In vitro assays indicated that Sec3p directly interacted with Cdc42 in its GTP-bound form. These results revealed a role for Cdc42 in polarized exocytosis, and help us understand the molecular mechanisms underlying the polarized localization of exocytic machinery and the spatial regulation of secretion.

Yeast Strains and Media for Growth
Yeast cells were grown in YP medium (1% Bacto-yeast extract, 2% Bacto-peptone) supplemented with 2% dextrose (YPD), or grown in SCD medium (synthetic complete plus 2% dextrose) lacking certain supplements. Strains used in this study are listed in Table 1.

Isolation of cdc42-ts mutants
To isolate temperature-sensitive (ts) mutations in CDC42 that may cause distinct morphological changes at either permissive or non-permissive temperature, we randomly mutagenized the entire CDC42 gene by a PCR-based method. The mutagenic PCR conditions (100 µl reaction volume) were as follows: 1 µl template DNA (0.1mg/ml) of plasmid pKS-CDC42, 2 µl the forward (454 bp upstream of the start codon) and reverse (540 bp downstream of the stop codon) primers (50 mM each), 10 µl Taq polymerase buffer (10 X stock, Mg 2+ free), 10 µl Mg 2+ (10 X stock, 25 mM), 10 µl mutagenic dNTPs mixture (10 X stock: 2 mM dATP, 2 mM dGTP, 10 mM dCTP, and 10 mM dTTP), 1ml MnCl 2 (50 mM), 1 µl Taq (5units/µl) (Promega), and 63 µl of ddH 2 O. PCR was carried out with the following program: 94 0 C, 4 min → 94 0 C, 1 min; 45 0 C, 1 min; 72 0 C, 1 min 45 sec for 30 cycles → 72 0 C, 10 min → 4 0 C. Multiple PCRs were combined and the ~1.6-kb PCR-amplified fragment was purified by ethanol precipitation and resuspended in ddH 2 O. The mutagenized PCR products were mixed with an equal amount of linearized pRS314-CDC42 (CEN TRP1), in which the entire CDC42 gene was removed by a restriction digestion with HpaI (67 bp upstream of the start codon) and NsiI (346 bp downstream of the stop codon). This mixture of DNA was transformed into strain YEF1194 (a cdc42∆::HIS3, pRS316-GAL1-CDC42) (pRS316-GAL1-CDC42 is a centromere-based, URA3-marked plasmid carrying CDC42) grown in SC-Ura medium by guest on March 24, 2020 http://www.jbc.org/ Downloaded from containing 2% glucose and 2% galactose to allow "gap repair" between the PCR products and the linearized pRS314-CDC42, and transformants were selected on SC-Trp plates at 24 0 C. The transformants were replicated onto SC-Trp plates containing 5-FOA to select for the loss of plasmid pRS316-GAL1-CDC42 at 24 0 C. These Trp + Uratransformants were replicated onto two sets of SC-Trp plates that were incubated at 24 0 C and 37 0 C, respectively, to allow the identification of ts mutants. Plasmids were rescued from mutants that displayed interesting morphologies at either permissive or non-permissive temperature, and were sequenced to confirm the existence of mutation(s) in CDC42. These plasmids were transformed into strain YEF1194 and processed as described above to confirm that the ts and morphological phenotypes were indeed caused by mutations on the plasmids. YEF1961 (cdc42-201) and YEF1962 (cdc42-13) were constructed through homologous recombination at the cdc42∆::HIS3 locus by transforming a ~1.7-kb BamHI-SalI fragment from 7 methanol at -20 o C, pelleted and washed with acetone at -20 o C. Cells were subsequently rehydrated by washing 3 times with PBS at 4 o C. Cells were then immediately examined by microscopy. For Sec3-GFP localization in cells recovering from G 0 arrest, the cells were first grown overnight in liquid culture. 0.2 ml of the culture was then spread onto YPD agar plates.
After the yeast cells form a lawn (2-3 days at 25 0 C), the cells were scraped from the plates and transferred into YPD liquid medium. The cells were resuspended in 20 ml of 1M sorbitol / 50% YPD and then spun for 1 min at 500 g to pellet budded cells. The supernatant containing unbudded cells was collected. These cells were examined under the microscope. If necessary, the cells were further spun to ensure the cells were uniformly unbudded. Aliquots of the unbudded cells were transferred to YPD liquid media and allowed to grow for 3 hours at 37 0 C.
The cells were then harvested and Sec3-GFP localization was examined as described above.
Immunofluorescence localization of Cdc42 in yeast cells was performed using affinitypurified rabbit anti-Cdc42 antibody as described previously 22 . Fluorescence microscopy was performed with a Leica microscope fitted with a 100 x oil immersion objective and standard filter sets. The images were acquired with a Quantix HCCD camera and OpenLab Imaging Software.

In vitro Binding between Sec3p and Cdc42
Sec3p (aa. 1-501) was fused with maltose-binding protein (MBP-Sec3p). Cdc42 with its C-terminal four amino acids lipid modification sites deleted was fused to GST. The recombinant fusion proteins were purified from bacterial lysates and used in the in vitro binding assay. The

The exocyst proteins are mislocalized in a cdc24 mutant
To identify proteins that regulate the polarized localization of the exocyst, we examined the localization of GFP-tagged exocyst components in mutants that are defective in different aspects of the budding process such as bud-site selection, bud-site establishment, cell polarization, actin organization, cell cycle regulation, and septin assembly 18 . The GFP-tagged exocyst proteins were expressed under the control of their own promoters and were fully functional because yeast cells expressing the tagged version as the sole source of the exocyst protein grew as well as wild-type cells in both rich and synthetic media, and at temperatures ranging from 25 0 C to 37 0 C. We previously found that the exocyst proteins were mislocalized in several rho1 mutants 18 . Rho1 is a member of the yeast Rho family of small GTPases. It is involved in such cellular processes as actin organization, yeast cell wall remodeling, and maintenance of polarized cell growth 17 . Another member of the Rho family of small GTPases is Cdc42. Cdc42 is important for the initial establishment of polarity and functions very early in the signaling cascade that orchestrates polarized morphogenesis 19 . We therefore speculated that Cdc42 may also control exocyst localization. cdc42-1 is the first identified cdc42 mutant strain 36 , and it has been widely used in studying the function of Cdc42 and cell polarity. Surprisingly, a previous study showed that there is no mislocalization of Sec3-GFP in cdc42-1 cells 8 .
To further explore the role of Cdc42 on the exocyst, we decided to examine whether exocyst localization is dependent upon the activity of Cdc24, which is the guanine nucleotide exchange factor (GEF) for Cdc42. Wild-type and cdc24-4 mutant cells synchronized to G 0 phase were incubated for 3 hours in YPD liquid medium at the restrictive temperature (37 0 C).
The localization of Sec3-GFP in the cells was then examined. As shown in Figure 1 The effect of Cdc24 on the exocyst may be a result of its role in activating Cdc42.
Although cdc42-1 did not show exocyst mislocalization, this mutant strain may not be the appropriate allele for revealing the specific defect. Very recently, a number of new cdc42 mutants have been generated which revealed some previously unidentified functions of Cdc42 [22][23][24][25][26] . We examined the localization of GFP-tagged exocyst proteins in some of the mutants.
Several of the cdc42 mutants used in this study were generated as described in the Methods; other mutants were recently characterized 22 . Figure 2 shows the distribution of two GFP-tagged exocyst proteins, Sec3p and Sec5p, after shift to 37 0 C for 2 hours. In addition, F-actin was also examined using rhodaminephalloidin. In wild-type cells ( Figure 2, upper panel), the exocyst proteins were localized to emerging bud sites, the tips of small daughter cells ("bud tip"), or mother/daughter connections ("necks"). However, in cdc42-201 and cdc42-13 mutant cells, Sec3-GFP and Sec5-GFP were either dispersed throughout the cell or formed multiple punctuate spots. In some mutant cells, the GFP-tagged exocyst was concentrated in one or several areas, but not in the bud tip or mother/daughter connection. In the parent wild-type strain, most of the cells scored from a single focal plane had polarized Sec3-GFP (66%) and Sec5-GFP (61%) localization (n=100) ( Table 2). However, in cdc42-201, only 3% and 2% of the cells, respectively, had polarized localization of Sec3-GFP and Sec5-GFP. In cdc42-13, only 9% and 8% of the cells,  Table 2 summarizes the results we obtained using various mutant strains. Factin was generally depolarized in both mutant alleles (Figure 2).
Since Cdc42 is involved in the establishment of cell polarity, by starting with a population of cells from G 0 phase, one should be able to study the effect of Cdc42 on the initial targeting of Sec3p-GFP to the emerging bud tip. We therefore examined the localization of Sec3-GFP in cdc42-201 and cdc42-13 mutant cells after released from G 0 arrest for 3 hours. As shown in Figure 3A, Sec3-GFP was depolarized in the cdc42-201 cells at the non-permissive temperature. For cdc42-13, most of the cells lost polarized Sec3-GFP localization. In fact, less than 6% of the cells still had polarized localization. As a control, Sec3-GFP was well localized in the wild-type cells released from G 0 for 3 hours. These results suggest that initial targeting of Sec3p is affected in cdc42 mutant alleles. In contrast to these observations, it was found that Sec3p can be correctly targeted in actin mutant cells or in cells treated with Latrunculin-A starting from the G 0 phase 8 , indicating that the polarized localization of Sec3p can be established in an actin-independent fashion.
We have also compared the localization of Sec3-GFP and Cdc42 in the yeast cells.
Due to technical difficulties with the yeast cells, we were not able to observe the Cdc42 immunofluorescence and Sec3p signals (by GFP signals or immunofluorescence using anti-GFP antibody) in the same cell by double labeling. However, we were able to perform the localization experiments on Cdc42 and Sec3-GFP separately and then compare their localization in the cells at similar cell cycle stages. As shown in Figure 3B, Cdc42 co-localizes with Sec3-by guest on March 24, 2020 http://www.jbc.org/ Downloaded from GFP at the bud emerging sites or at the tips of the small buds. This observation is consistent with the hypothesis that Sec3p needs Cdc42 for its initial targeting to the bud tip.

Sec3p interacts with Cdc42 in vitro
The effect of cdc42 mutants on Sec3-GFP localization could reflect either a direct interaction of Sec3p with Cdc42 or an indirect effect mediated by one of the Cdc42 effectors. In order to understand the molecular basis for this effect, we tested for the direct binding of Sec3p with Cdc42 with a biochemical assay using MBP-Sec3p and GST-Cdc42 fusion proteins purified from bacteria lysates. Previous domain analysis indicates that the post-coiled-coil region of Sec3p (aa 500-1336) is responsible for binding to another component of the exocyst complex, Sec5p. The N-terminal portion of Sec3p binds Rho1; and Sec3p lacking its Nterminus diffuses inside cells 18 . We tested whether the N-terminal region of Sec3p also binds to Cdc42. As shown in Figure 4A, GST-Cdc42 preferentially bound to the N-terminal region of Sec3p conjugated to the resin in the presence of GTPγS. The amounts of Cdc42 bound to MBP-Sec3p resin are in average about 20% of the input levels in six independent experiments (not shown). As a negative control, we found that GST alone did not bind MBP-Sec3p under the same reaction conditions. As another negative control, MBP itself, even at levels in great excess of MBP-Sec3p, did not bind GST-Cdc42. Since Sec3p is mislocalized in cdc42-201 and, to a less extent, in cdc42-13, we have compared their binding to MBP-Sec3p with wild-type Cdc42. As shown in Figure 4B, comparing with GST-Cdc42 (wild-type), GST-cdc42-201 had a much weaker interaction with MBP-Sec3p in the presence of either GTP or GDP. This may explain why Sec3-GFP is mislocalized in the cdc42-201 mutant allele. We have also detected a decrease in GST-cdc42-13 binding to MBP-Sec3p. However the decrease in binding was not as dramatic as that seen with the GST-cdc42-201 mutant. Sec3-GFP was mislocalized in cdc42-13, though the localization defect was less severe than that of cdc42-201 (see Table 2). Finally, we did not detect any change in cdc42-1 mutant binding (Fig. 4B). Overall, the binding properties seem to correlate with the exocyst localization in these cdc42 mutants. It is possible that the defects in binding account for the Sec3 mislocalization defects in some cdc42 mutants.
However, we do not exclude the possibility that additional mechanisms are involved in mislocalization of Sec3-GFP in these mutant cells.
Since the Rho1 binding region is also localized to the N-terminus of Sec3p 18 , we tested whether Cdc42 competes with Rho1 binding to MBP-Sec3p. The Rho1 and Cdc42 fusion proteins were first loaded with GTPγS. Then various amounts of Cdc42 were added to the Rho1-Sec3p binding reaction as described above. As shown in Figure 4C, the Cdc42 recombinant protein competed with Rho1 in binding Sec3p.

Depolarized pattern of surface protein incorporation in cdc42 mutant cells
Since Cdc42 interacts with Sec3p and controls the localization of the exocyst, the essential secretory vesicle targeting and docking machine, we speculated that Cdc42 might spatially regulate exocytosis in cells. We therefore examined the exocytosis pattern in the wild type and cdc42 mutant cells by monitoring cell wall glycoproteins using the lectin Concanavalin A (Con A). The cell wall was pre-labeled with FITC-conjugated Con A and then thoroughly At 90 min, the majority of the small-and medium-budded cells had dark buds (the entire buds were dark) or dark zones surrounding the bud tips, indicating that exocytosis was localized in the daughter cells. In contrast, the displacement of fluorescence was randomized on the surface of cdc42 and cdc24 mutant cells. For cdc24-4 cells, almost 100% of the mother and daughter cells were uniformly labeled at 90 min. For cdc42-201, at 0 min after FITC-Con A labeling, 100% of the cells were brightly labeled. At 90 min, ~100% of the cells still had bright FITC-Con A staining in both mother and daughter cells. In some cases, the mother cells were even dimmer than the buds, suggesting that new growth had occurred preferentially in the mother cells. It was obvious that cells at this point were larger than those before the shift, indicating that the cells were growing, but in a depolarized fashion. No polarized growth was found in these mutant cells even by 215 min after washing. For cdc42-13, many small-budded cells or cells with tiny buds had polarized growth. However, some of these cells lost polarized growth. Overall, these results indicate that exocytosis occurs in a depolarized manner when Cdc42 is defective.

Discussion
Cdc42 is an important regulator of many biological processes including cell polarization, morphogenesis, membrane traffic, cell growth and development in the eukaryotic cells 4, 15-17 . The budding yeast S. cerevisiae undergoes polarized growth, which requires a sophisticated system to coordinate multiple cellular processes, including cell cycle progression, assembly of the actin cytoskeleton, morphogenesis, and polarized membrane traffic. Cdc42 is localized to bud tips, and plays important roles in all of these processes. Like many other members of the Ras super family of small GTP-binding proteins, Cdc42 has multiple downstream effectors 20, 21 . Through these different effectors, Cdc42 coordinates various cellular activities to carry out specific biological functions. Recently, many new cdc42 conditional mutants have been generated [22][23][24][25][26] . These mutants have been remarkably useful in revealing the functions of Cdc42 in various cellular processes. Since Cdc42 has multiple effectors, experiments using these mutant alleles have great advantage comparing with those using dominant-negative (cdc42N17) or constitutively active (cdc42V12) cdc42 alleles in revealing the individual processes that Cdc42 controls. The phenotypes observed by expressing dominant-negative or constitutively active cdc42 alleles might result from multiple downstream defects rather than a specific defect. It is also possible that the mutant phenotypes are caused by the loss of one effector system prior to the loss of another effector system. In this study, several newly identified cdc42 mutants specifically revealed the role of Cdc42 in polarized localization of the exocyst. The results presented here reveal that Cdc42 spatially controls the vesicle tethering/docking machine, the exocyst. It is also known that Cdc42 controls the polarized organization of the actin cytoskeleton, which provides the track for directional transport of secretory vesicles [28][29][30] . Taken together, we propose that Cdc42 regulates polarized exocytosis through its dual control over the vesicle transport system (actin) and the vesicle docking system (the exocyst). In addition to exocytosis, it is known that Cdc42 also controls cell polarity and cell cycle progression. Cdc42 may therefore serve as a coordinator of different cellular processes, all of which must be intimately coupled.