A role for Gic1 and Gic2 in Cdc42 polarization at elevated temperature

The conserved Rho-family GTPase Cdc42 is a master regulator of polarity establishment in many cell types. Cdc42 becomes activated and concentrated in a region of the cell cortex, and recruits a variety of effector proteins to that site. In turn, many effectors participate in regulation of cytoskeletal elements in order to remodel the cytoskeleton in a polarized manner. The budding yeast Saccharomyces cerevisiae has served as a tractable model system for studies of cell polarity. In yeast cells, Cdc42 polarization involves a positive feedback loop in which effectors called p21-activated kinases (PAKs) act to recruit a Cdc42-directed guanine nucleotide exchange factor (GEF), generating more GTP-Cdc42 in areas that already have GTP-Cdc42. The GTPase-interacting components (GICs) Gic1 and Gic2 are also Cdc42 effectors, and have been implicated in regulation of the actin and septin cytoskeleton. However, we report that cells lacking GICs are primarily defective in polarizing Cdc42 itself, suggesting that they act upstream as well as downstream of Cdc42 in yeast. Our findings suggest that feedback pathways involving GTPase effectors may be more prevalent than had been appreciated.


Introduction
Regulation of cell shape is central to cell proliferation as well as many aspects of cell function. Cell shape is in large part governed by the cytoskeleton, which itself is regulated by multiple signaling pathways. Among the most prominent and widespread cytoskeleton-regulating pathways are those mediated by evolutionarily conserved small GTPases of the Rho family, including Rho, Rac, and Cdc42 [1]. These GTPases are thought to act as molecular switches, toggling between an inactive GDP-bound state and an active GTP-bound state. Intrinsic rates of activation (GDP/GTP exchange) and inactivation (GTP hydrolysis) are slow, and can be greatly enhanced by guanine nucleotide exchange factors (GEFs) and GTPase activating proteins (GAPs), respectively [2]. Rho-family GTPases are prenylated and reside primarily on the cytoplasmic leaflet of cellular membranes, although they can be extracted to the cytoplasm by guanine nucleotide dissociation inhibitors (GDIs) [3,4]. Signaling pathways controlling cell shape often act by regulating and localizing the activities of GEFs and GAPs, leading to specific spatiotemporal patterns of GTPase activity.
In addition to the studies implicating GICs as mediators of Cdc42-induced actin and septin rearrangements, a genetic interaction was identified between GICs and the Ras-family GTPase Rsr1 [31]. Rsr1 mediates communication between various transmembrane "landmark" proteins, which mark preferred sites for subsequent polarization, and the Cdc42-based polarity establishment pathway [32]. Unlike gic1Δ gic2Δ or rsr1Δ mutants, which are viable at 24˚C, gic1Δ gic2Δ rsr1Δ triple mutants were lethal. This suggested that GICs might act upstream of Cdc42, in parallel with Rsr1, as well as downstream of Cdc42.
Here, we have investigated the gic1Δ gic2Δ phenotype in greater detail, using live-cell imaging of cells bearing probes for polarity regulators Cdc42 and Bem1. We found that at 37˚C, a majority of gic1Δ gic2Δ cells failed to polarize at all. This finding provides an alternative interpretation for previous findings in which the mutants failed to polarize actin or septins: these defects could be secondary effects stemming from a more fundamental lack of Cdc42 polarization. A subset of the gic1Δ gic2Δ mutant cells did polarize Cdc42 and Bem1 at 37˚C, and those cells did not display any obvious difficulty in forming a bud, suggesting that downstream cytoskeletal defects (if present) were quite mild. We conclude that, as suggested by Kawasaki et al. [31], a major role of the GICs is to promote and/or maintain Cdc42 polarization.

Results
Polarity establishment in yeast is regulated by the cell cycle. In particular, activation of G1 cyclin-dependent kinase (CDK) complexes at a commitment point called start in G1 promotes Cdc42 polarization [13]. G1 CDK activation at start occurs through a transcriptional positive feedback loop in which rising CDK activity promotes the inactivation and nuclear export of the repressor Whi5, allowing more transcription of G1 cyclins [33][34][35]. Commitment to enter the cell cycle (i.e. start) occurs when 50% of nuclear Whi5 has been exported, at which point the positive feedback loop becomes self-sustaining [36]. We used Whi5-GFP or Whi5-tdTomato as a probe for start, and Bem1-tdTomato [9] or GFP-Cdc42 [16] as a probe for polarization. Haploid cells were grown at 24˚C and arrested in G1 by treatment with mating pheromone. Arrested cells were released to proceed into the cell cycle by washing out the pheromone, and placed on microscope slabs at 37˚C. Live cell imaging by confocal fluorescence microscopy was then employed to monitor probe localization.
In wild-type cells under these conditions, Whi5 nuclear export was closely followed by Bem1 (Fig 1 and S1 Movie) or Cdc42 (Fig 1 and S2 Movie) polarization. However, in gic1Δ gic2Δ cells there was a heterogeneous phenotype: a majority of cells failed to polarize either Bem1 ( Fig 1B and S1 Movie) or Cdc42 (Fig 1C and S2 Movie) after Whi5 nuclear exit. A substantial minority of cells did polarize the probes, although polarity establishment occurred somewhat later than in wild-type cells. Quantification revealed that 30%-40% of gic1Δ gic2Δ cells were able to form buds, but compared to wild-type cells the start-to-budding interval was longer (Fig 2A). Similarly, 30%-40% of gic1Δ gic2Δ cells were able to polarize Bem1 or Cdc42 (Fig 2B), but with a longer interval between start and polarization ( Fig 2C: median 7.5 min for wild-type and 13.5 min for gic1Δ gic2Δ, p<0.01 by Mood's median test). For the subset of gic1Δ gic2Δ cells that did polarize, the interval between polarization and budding was similar to that in wild-type cells (Fig 2D). Thus, the major defect exhibited by gic1Δ gic2Δ cells at 37˚C was an inability to establish and/or maintain polarity.
Polarity establishment requires activation of Cdc42, which is promoted by the GEF Cdc24 and antagonized by the GAPs Bem2, Bem3, Rga1, and Rga2 [32]. Thus, possible bases for the defect in polarity establishment in gic1Δ gic2Δ cells include that they have insufficient GEF, Cdc42, or Bem1, or excess GAPs. As an initial attempt to test that hypothesis, we compared the abundance of these regulators in wild-type and gic1Δ gic2Δ cells. We noted no significant differences, either when the cells were grown at 24˚C (Fig 3A) or 37˚C (Fig 3B). Many of the regulators undergo phosphorylation, which is thought to regulate their activity [13-16, 37, 38]. Although we detected altered-mobility species in many of the blots, we did not find any systematic difference between wild-type and gic1Δ gic2Δ cells.
We next attempted a genetic approach to test whether mutations in regulators might enhance or suppress the phenotype of gic1Δ gic2Δ cells. Although gic1Δ gic2Δ mutants are lethal at 37˚C following tetrad dissection (Fig 4A), growth of gic1Δ gic2Δ mutants at 24˚C led to a rapid accumulation of spontaneous suppressors, such that subsequent transfer to 37˚C was not uniformly lethal. This high frequency of suppression made it impossible to reliably test whether mutations in regulators were able to suppress the phenotype. We speculated that because a subpopulation of mutant cells was able to bud (Figs 1 and 2), strong selection pressure could be applied to the expanding population, yielding a high spontaneous suppression In early G1 phase (pre-start), Whi5 is concentrated in the nucleus (green) and polarity factors are dispersed. As CDK activation occurs, Whi5 is exported from the nucleus (when 50% of Whi5 has been exported the cells commit to enter the cell cycle at "start"). CDK activation triggers localization of polarity factors to a cortical site (red: polarization) from which the bud later emerges (bud). (B) Inverted maximum projection montages of selected timepoints for representative cells from movies of wildtype (WT: DLY19654) or mutant (gic1Δ gic2Δ: DLY20961) cells progressing through the cell cycle at 37˚C. The cells express Whi5-GFP (top row) and Bem1-tdTomato (bottom row) probes. Cells were synchronized in G1 by pheromone arrest-release, and time relative to start is indicated. Wild-type cells polarized shortly after start, whereas 58% of gic1Δ gic2Δ cells failed to polarize (cell 1); the others polarized, often after a delay (cell 2). (C) Display as for (B) but with strains expressing Whi5-tdTomato (top row) and GFP-Cdc42 (bottom row). Wild type: DLY21726. gic1Δ gic2Δ: DLY21728. Wild-type cells polarized shortly after start, whereas 76% of gic1Δ gic2Δ cells failed to polarize (cell 1); the others polarized, often after a delay (cell 2). (D) Wild-type (WT: DLY19654) and mutant (gic1Δ gic2Δ: DLY20961) cells were grown at 24˚C, placed on a slab containing 2 μM α-factor, and incubated at 37˚C for 2 h. Cells that arrested in G1 (scored by the presence of nuclear Whi5) were examined for polarization of Bem1. Inverted maximum projection montages of selected cells are shown: wild-type cells polarized, whereas 60% of gic1Δ gic2Δ cells failed to polarize (middle row); the others polarized (bottom row). Scale bar, 5 μm.
frequency. Such suppression might occur at many loci or just a few, and we reasoned that in the latter case, identification of the basis for spontaneous suppression might be informative with regard to the specific molecular defect that prevents polarization of a majority of gic1Δ gic2Δ cells.
We picked 10 independent unsuppressed haploid gic1Δ gic2Δ colonies growing at 24˚C, and spread a million cells of each colony on a rich media plate that was incubated at 37˚C. Multiple colonies arose spontaneously on each plate, ranging from large to tiny in size ( Fig  4B). We picked a large colony from each plate, and mated them to an unsuppressed gic1Δ gic2Δ of the opposite mating type. Upon sporulation of the resulting diploids, viability at 37˚C segregated 2:2 in tetrads in 9 cases, showing that suppression was due to a single Mendelian locus in these independently derived strains ( Fig 4B and Table 1).
To assess whether the independent suppressors occurred at the same or different loci, we performed pairwise crosses between gic1Δ gic2Δ mutants carrying the different suppressors. In all cases, diploids generated by crossing one suppressed strain to another showed 4:0 segregation for viability at 37˚C (Fig 4B). This indicates that all suppressors are tightly linked, and likely to be in the same locus.
To characterize the suppressed phenotype, we performed live cell imaging of suppressed strains carrying Whi5-GFP and Bem1-tdTomato. We found that suppressed strains were similar to wild-type in terms of the efficiency and timing of polarization relative to start ( Fig 5). Thus, suppression is highly effective in restoring the ability to polarize.  Fig 1B), while (B) additionally plots data for strains expressing GFP-Cdc42 (as in Fig 1C).

Discussion
Previous studies identified roles for Gic1 and Gic2 in regulating actin or septin organization downstream of Cdc42. As GICs are effectors that bind specifically to GTP-Cdc42, it was natural to expect roles of GICs acting downstream of Cdc42. However, the discovery of synthetic lethality between rsr1Δ and gic1Δ gic2Δ mutants [31] indicated that GICs might also act upstream of Cdc42. Our major finding is that GICs are required for efficient and timely polarization of Cdc42 at 37˚C, strongly supporting the conclusion that GICs act upstream of Cdc42. Given the presence of both positive and negative feedback in the polarity circuit [10], such strains. Anti-GFP Western blots to compare abundance of Bem1-GFP (wild-type, DLY10005; gic1Δ gic2Δ, DLY20597) and GFP-Cdc42 (wild-type, DLY21726; gic1Δ gic2Δ, DLY21728). Anti-myc Western blots to compare abundance of Bem3-12myc (wild-type, DLY11483; gic1Δ gic2Δ, DLY22232), Rga1-12myc (wild-type, DLY21093; gic1Δ gic2Δ, DLY22235), and Rga2-12myc (wild-type, DLY11847; gic1Δ gic2Δ, DLY22232) expressed at the endogenous loci. Loading control is a blot of Cdc11 (a septin) in the same lysates. Cells were grown to mid-log phase and lysates were prepared as described in Methods. Quantification of each blot (fluorescence intensity of secondary antibody for each regulator normalized to its corresponding loading control) is shown in the bar graph below each blot. When independent Western blots were performed, the number of blots is indicated and the bar graphs show mean and standard error of the mean. (B) Western blots were repeated using lysates from cells that were shifted to 37˚C for 6 h prior to lysate preparation. Dashed lines indicate instances in which the two lanes were not adjacent in the original gel and the intervening lanes have been removed for clarity.
There are several potential explanations for these findings. First, GICs might play dual roles, acting both upstream of Cdc42 and downstream of Cdc42 in separate pathways. A preliminary examination of the levels of known Cdc42 regulators did not reveal any differences . A diploid strain with the indicated genotype (DLY21711) was sporulated and tetrads (four spores in a vertical column) were dissected onto plates that were incubated at the indicated temperature. Tetrads contain two GIC1 gic2Δ spores and two gic1Δ gic2Δ spores. At 24˚C all four spores were viable and gave rise to colonies, but at 37˚C two spores from each tetrad died. Replica plating confirmed that the dead spores were the gic1Δ gic2Δ cells. (B) Isolation and genetic characterization of spontaneous gic1Δ gic2Δ suppressors. Cells of a gic1Δ gic2Δ strain (DLY20961) were streaked for single colonies. One million cells from each colony were plated on rich media and incubated at 37˚C for 3 days. Although most cells died, several heterogeneously sized colonies were able to grow (example plate, bottom left), and one large colony from each independent plate was picked for further analysis. Suppressed cells were mated to a non-suppressed gic1Δ gic2Δ strain of opposite mating type (DLY21941), and the resulting diploids were sporulated and dissected as in (A). Tetrads showed 2:2 viability (middle panels) at 37˚C indicating segregation of the suppressor as a single Mendelian locus. Independent suppressed strains (from different initial colonies) were then mated to each other and the resulting diploids were sporulated and dissected as in (A). All tetrads showed 4:0 viability at 37˚C (right panels) indicating that the suppressors all map to the same locus. Sequencing confirmed that suppressed strains retained the gic1Δ and gic2Δ mutations.  between wild-type and gic1Δ gic2Δ mutant cells, but it remains possible that GICs affect the activity rather than the abundance of these regulators. Second, GICs may simply act as downstream effectors of Cdc42, mediating cytoskeletal reorganization. Because Cdc42 is known to polarize even in the absence of F-actin [39,40] or Bem1-tdTomato (bottom row) probes. Cells were synchronized in G1 by pheromone arrest-release, and time relative to start is indicated. Scale bar, 5 μm. (B) Time intervals between start and polarization were scored from the time-lapse movies above as in Fig 2B. The number of cells scored for each plot is indicated (n). Data for wild-type strain are from same conditions as in Fig 2B but  polymerized septins [41], this alone would not necessarily yield the observed defects in Cdc42 polarization. However, it could be that the particular cytoskeletal misregulation that occurs in gic1Δ gic2Δ mutants triggers a stress response that blocks effective Cdc42 polarization. Although stress pathways can act to block polarization [42,43], we believe this scenario is unlikely. Third, and perhaps most likely, GICs could operate as part of a positive feedback loop in which GTP-Cdc42 acts to promote further local accumulation of GTP-Cdc42. This would explain why cells lacking GICs have difficulties in polarizing Cdc42, and there is precedent for such feedback in the role of PAKs and Bem1 [10,11]. In addition, one could imagine that GICs modulate the negative feedback loop in which PAK-mediated GEF phosphorylation reduces GEF activity [16], and that in the absence of GICs this inhibitory pathway is too powerful. However, the mechanism by which GICs might function in either positive or negative feedback remains mysterious.
Cells growing at 24˚C do not require GICs for successful proliferation, suggesting that there are parallel pathways that can operate in the absence of GICs. Alternatively, it could be that GIC pathways are only engaged in order to deal with thermal stress. The growth defect of haploid cells lacking GICs can be suppressed by overexpression of Cdc42 [20], and diploid cells lacking GICs are able to proliferate successfully even at 37˚C [44]. Other mutants (e.g. lacking the formin Bni1) display more severe phenotypes in diploids than in haploids [44]. The basis for these differences is unclear. We found that in our strain background, gic1Δ gic2Δ mutants frequently acquired spontaneous suppressors, and a genetic analysis indicated that several independently isolated suppressors all mapped to the same locus. Identification of the suppressor gene may provide insight into the role of GICs in promoting Cdc42 polarization.
While this paper was in revision, an independent study on GICs was published that reached similar conclusions to ours based on parallel experiments [45].

Yeast strains and growth conditions
The yeast strains used in this study are in the YEF473 strain background (his3-Δ200 leu2-Δ1 lys2-801 trp1-Δ63 ura3-52) [46] and are listed in Table 2. Standard yeast molecular and genetic manipulations were used to construct strains, with additional precautions due to the high propensity of strains lacking GICs to become genetically suppressed. GIC1 and GIC2 deletions were generated by the one-step PCR-based method [47] with pRS304 as template for gic1:: TRP1 and pRS403 as template for gic2::HIS3. Deletions were introduced into diploid strains, and diploids containing at least one wildtype GIC gene were used as strain construction intermediates to avoid selection for suppressors. In cases where strain construction involved a haploid gic1::TRP1 gic2::HIS3 intermediate, we introduced a URA3-marked 2 μm plasmid (pDLB2693) carrying wild-type GIC2 into the parent diploid strain and maintained the plasmid in the derived haploid so as to avoid selecting for suppressors. Loss of the plasmid was induced when needed by growth on plates containing 5-fluoroorotic acid (5-FOA) [48].
The Cdc24-3HA allele was generated by the PCR-based gene modification method [51]. Briefly, primers with 50 bp of CDC24 C-terminus and 3'UTR homology were used to amplify the pFA6 3HA kanMX cassette. The PCR product was purified and transformed to tag CDC24 via standard transformation methods. Proper integration was confirmed by PCR and sequencing.
The Bem3-12myc, Rga1-12myc, and Rga2-12myc constructs were made by cloning PCR products encoding the C-termini of the proteins into a pRS306-based integrating plasmid (pSWE1-myc) containing 12 myc tags and the SWE1 terminator [52]. Digestion at a site within the gene was used to target integration of the plasmid at the endogenous loci, and proper integration was confirmed by PCR checks.

Isolation and analysis of suppressor mutants
The MATa gic1Δ gic2Δ strain DLY20961 was streaked for single colonies on YEPD plates at 24˚C. 10 colonies were picked using sterile toothpicks, and the cells were resuspended in 1 mL of sterile distilled water and counted. 1 million cells from each colony were spread onto individual YEPD plates. Each plate was incubated at 37˚C for 3 days, after which plates displayed growth of numerous microcolonies and a few large colonies. A large colony was picked from each of the 10 plates, and mated to non-suppressed MATα gic1Δ gic2Δ strain DLY21941. Mating was conducted by mixing cells on a YEPD plate, with a large excess of the MATa strain, so that most MATα cells would mate. Cells from the mating mix were spread on YEPD plates containing 2 μM α-factor to arrest unmated MATa cells, and colonies were tested to determine whether they could sporulate (indicating successful diploid formation) when transferred to 2% Potassium Acetate plates and allowed to grow for 5-7 days.
Asci from sporulating diploids were digested by treatment with lyticase for 5 min. Tetrads were diluted in sterile distilled water, spread on YEPD plates, and dissected with a micromanipulator. Tetrad plates were incubated at 24˚C or 37˚C as indicated for 3 days. Images of plates were taken on day 3. Spore colonies were replica plated to relevant selective media plates to test for auxotrophic markers. Suppressor strains were then crossed to each other and tetrads were analyzed using a similar procedure.

Cell synchronization
For imaging experiments, the cells were first synchronized by G1 arrest/release. MATa cells were grown overnight at 24˚C in CSM+D, adjusted to 1.5x10 7 cells/mL, and treated with 2 μM α-factor (Genesee Scientific) at 24˚C for 3 h. G1-arrested cells were released from arrest by washing two times with fresh medium, and placed on microscope slabs at 37˚C for imaging.

Microscopy and image analysis
Cells were mounted on a 250 μL slab solidified with 2% agarose on a microscope slide. After putting a cover slip on top, the edges were sealed with petroleum jelly to prevent evaporation. Image acquisition was done using an Andor XD Revolution spinning-disk confocal microscope (Olympus) with a Yokogawa CSU-X1 5000 r.p.m. disk unit, and a 100x/1.4 UPlanSApo oil-immersion objective controlled by MetaMorph software (Universal Imaging). The microscope is enclosed in a temperature-controlled chamber that was set to 37˚C 1 h prior to imaging. Fluorophores were excited with 488 nm and 561 nm diode lasers. Images (stacks of 17 z planes spaced 0.5 μm apart) were collected at 1 min intervals using a iXon3 897 EM-CCD camera with 1.2x auxiliary magnification (Andor Technology). Laser power was set to 10% maximum output to reduce phototoxicity. Exposure time was 200 ms for each image. An EM-Gain setting of 200 was used for the EM-CCD camera.
Collected images were deconvolved using Hyugens Essential software (Scientific Volume Imaging). Images were then processed using ImageJ (National Institutes of Health). Zstacks were collapsed into maximum projection images. Polarization was scored by eye as the first detection of a cluster of the polarity probe (GFP-Cdc42 or Bem1-tdTomato). Whi5 nuclear export was scored using a custom MATLAB graphical user interface (GUI; Nuc-TrackV3.3) as described [53]. This tool allows for designation and tracking of a region of interest at specific times of interest during the course of the time-lapse. For our purposes, regions of interest were individual cells. The coefficient of variation of Whi5 signal intensity between pixels in each cell was measured and used to determine the time point at which 50% of Whi5 exited the nucleus. Calculated values are normalized to peak intensity for each track. This tool is available upon request from Dennis Tsygankov (ude.hcetag.emb@vokna-gysT.sined).

Immunoblotting
Cells were grown overnight in YEPD at 24˚C, and where indicated shifted to 37˚C for 6 h prior to harvesting. Cell pellets (about 10 7 cells) were resuspended in 225 μL cold Pronase buffer (25 mM Tris-HCl, pH 7.5, 1.4 M Sorbitol, 20 mM NaN 3 , 2 mM MgCl 2 ) and 48 μL of 100% TCA. Pellet-buffer mixture was stored frozen at -80˚C. Once thawed on ice, cells were lysed by vortexing with 280 μL of sterile acid-washed glass beads at 4˚C for 10 min. Beads were washed twice with 5% TCA. Lysate was collected and precipitated proteins were pelleted by centrifugation at maximum speed in an Eppendorf centrifuge for 10 min at 4˚C. Pellets were solubilized in Thorner buffer (40 mM Tris-HCl, pH 6.8, 8 M Urea, 5% SDS, 143 mM β-mercaptoethanol, 0.1 mM EDTA, 0.4 mg/ml Bromophenol Blue). 2 M Tris base was used to adjust the pH to 8. Samples were heated at 42˚C for 3 min prior to loading on a 10% Acrylamide/Bis gel and run for 1 h at 40 mA. Following transfer, membranes were probed with anti-cMyc, or anti-HA (12CA5) (Roche) monoclonal antibodies and anti-Cdc11 polyclonal antibodies (Santa Cruz Biotechnologies) used at 1:1000 and 1:2000 dilution respectively. Secondary antibodies IRDye800-conjugated anti-mouse IgG (Rockland Immunochemicals) and Alexa-fluor680-conjugated goat anti-rabbit IgG (Invitrogen) were used at 1:10,000 dilution. After washing, Western blots were visualized using the ODYSSEY imaging system (Li-COR Biosciences).
Western blot quantification was done using ImageJ to measure band intensity in individual color channels. Mutant and wild-type bands were always compared from the same blot using lanes with comparable Cdc11 loading controls. After dividing by the loading controls, bands were normalized to the wild-type signal.