The Role of Far1p in Linking the Heterotrimeric G Protein to Polarity Establishment Proteins During Yeast Mating

See allHide authors and affiliations

Science  20 Nov 1998:
Vol. 282, Issue 5393, pp. 1511-1516
DOI: 10.1126/science.282.5393.1511


Heterotrimeric guanosine triphosphate (GTP)–binding proteins (G proteins) determine tissue and cell polarity in a variety of organisms. In yeast, cells orient polarized growth toward the mating partner along a pheromone gradient by a mechanism that requires Far1p and Cdc24p. Far1p bound Gβγ and interacted with polarity establishment proteins, which organize the actin cytoskeleton. Cells containing mutated Far1p unable to bind Gβγ or polarity establishment proteins were defective for orienting growth toward their mating partner. In response to pheromones, Far1p moves from the nucleus to the cytoplasm. Thus, Far1p functions as an adaptor that recruits polarity establishment proteins to the site of extracellular signaling marked by Gβγ to polarize assembly of the cytoskeleton in a morphogenetic gradient.

Asymmetric cellular organization or cell polarity is a central feature of morphogenesis and is controlled by both internal and external signals (1). In the yeast Saccharomyces cerevisiae, mating pheromones trigger a mitogen-activated protein kinase (MAPK) signal transduction pathway, culminating in arrest of the cell cycle, changes in gene expression, and altered cell polarity and morphology (2). These responses are initiated by a cell-surface receptor coupled to a G protein. Activation of the pheromone receptor triggers dissociation of the heterotrimeric G protein into subunits Gα and Gβγ, which in turn signal to downstream effectors to induce cellular responses (3). Cells use a pheromone gradient to locate their mating partner and polarize their actin cytoskeleton toward the site of the highest pheromone concentration (4). Far1p and Cdc24p are necessary for oriented cell polarity: specific alleles ofFAR1 (far1-s) and CDC24(cdc24-m) have been identified which cause a specific mating defect, because these cells are unable to locate their mating partner (5, 6). Genetic experiments also implicate the G protein in the regulation of cell polarity during mating (7).

Morphological changes during mating depend on reorganization of the actin cytoskeleton by a group of proteins including Cdc24p, Bem1p, Cdc42p, and the two Gic proteins (Gic1p and Gic2p) that are necessary for establishment of cell polarity (1, 8). Cdc24p functions as a GDP-GTP exchange factor (GEF) for Cdc42p (9), whereas the Gic proteins are effectors of Cdc42p involved in organizing the actin cytoskeleton (8). Bem1p contains two SH3 domains and interacts with Cdc42p, Cdc24p, Ste20p, Far1p and, Ste5p (10–13). Bem1p coimmunoprecipitates with actin (12), suggesting that it is involved in recruiting the actin cytoskeleton to the site of polarization. Little is known about how the polarity establishment proteins are targeted to the site of polarization.

To examine the role of Far1p in orienting polarized growth during mating, we tested by two-hybrid analysis whether Far1p was able to interact with the polarity establishment proteins Bem1p, Cdc42p, Cdc24p, and Gic2p (14, 15). Far1p interacted with Bem1p, Cdc24p, and Cdc42p but not with Gic2p or a truncated Cdc24p, which lacks a portion of the NH2-terminal domain (Table 1). Far1p was unable to interact with the guanosine triphosphatase (GTPase) Rho1p, demonstrating that Cdc42p is a specific GTPase-binding partner of Far1p. Far1p preferentially bound to Cdc42p in its active GTP-bound state, whereas little binding was observed when Cdc42p was bound to guanosine diphosphate (GDP). Because Bem1p also interacted with Cdc42p in a GTP-dependent manner (Table 1) (10, 11), we tested whether the interaction between Far1p and Cdc42p-GTP was dependent on the presence of Bem1p. The interaction between Cdc42p and Far1p was abolished in a strain deleted for BEM1, whereas Cdc24p was able to interact with Far1p under these conditions (Table 1). Thus, Far1p interacts with the polarity establishment proteins Bem1p and Cdc24p, and Bem1p may bridge the interaction between Far1p and Cdc42p-GTP.

Table 1

Two-hybrid interactions among Far1p, Bem1p, Cdc24p, Cdc42p, Rho1p, and Ste4p in wild-type cells (EGY48) or derivatives deleted for BEM1 or STE7. The activation domain fusions were carried on pJG4-5–based vectors; the LexA DNA-binding domain fusions were carried on pEG202-based vectors (14). Miller units with standard deviations are presented; assays were done as described (15).

View this table:

To examine whether Bem1p could directly bind to Far1p in vitro, Far1p was fused at its NH2-terminus to two copies of the polyoma epitope (PT-Far1p) and purified from yeast (16). Cells expressing a hemagglutinin (HA) epitope–tagged version of Far1p (HA-Far1p) were used as a control. PT-Far1p was immobilized on a polyoma antibody affinity column and probed with Bem1p expressed as a 6His protein in Escherichia coli. After extensive washing, Far1 protein was eluted using polyoma-peptide, and bound Bem1p was detected by immunoblotting. Bem1p readily bound to the PT-Far1p column (Fig. 1A), whereas no binding was detected with HA-Far1p, demonstrating that Far1p and Bem1p interact specifically in vitro. Treating the cells expressing PT-Far1p with α factor did not alter the efficiency of binding between Far1p and Bem1p, suggesting that the interaction between Far1p and Bem1p was not modified by pheromone induction (13, 15). Coimmunoprecipitation experiments also confirmed that Far1p interacted with Cdc24p in vivo: cells expressing epitope-tagged Cdc24p (HA-Cdc24p) or control cells expressing untagged Cdc24p were treated with α factor, and then HA-Cdc24p was precipitated with HA11 antibodies and analyzed for the presence of Far1p expressed from the inducible GAL promoter by immunoblotting (17). Far1p readily coprecipitated with Cdc24p (Fig. 1B), demonstrating that Far1p bound to Cdc24p in vivo.

Figure 1

Far1p interacts with the polarity establishment proteins Bem1p, Cdc42p, and Cdc24p, as well as Ste4p, the β subunit of the heterotrimeric G protein. (A) Complex formation between Far1p and Bem1p in vitro (16). Far1p tagged with two copies of the polyoma epitope (PT-Far1p; lanes 1 and 2) or with an HA-epitope as a control (HA-Far1p; lane 3) was purified from cells which were either treated (+; lanes 2 and 3) or not treated (–; lane 1) with α factor and incubated with 6His-tagged Bem1p purified from E. coli. Bound proteins were eluted with polyoma peptide and subjected to immunoblot analysis with antibodies specific for Bem1p (12). (B) Far1p interacts with Cdc24p in vivo. HA-tagged Cdc24p (lanes 2 and 3) or untagged Cdc24p as a control (lane 1) was immunoprecipitated with HA11 antibodies from extracts prepared from Δfar1 cells, which express Far1p from the inducible GAL promoter (17). Expression of Far1p was induced by addition of galactose (+) or repressed by the addition of glucose (–). Immunoprecipitates were blotted with polyclonal antibodies against Far1p (upper panel) or Cdc24p (lower panel). The bracket marks the position of Far1p; the arrow points to the position of HA-Cdc24p. (C) Far1p interacts with Ste4p in vivo. Extracts prepared from cells expressing Ste4p tagged with an HA-epitope (lanes 1 and 2) or untagged Ste4p as a control (lanes 3 and 4) were incubated with antibodies to HA, and the immunoprecipitates were analyzed for the presence of bound Far1p (upper panel) or Ste4p (lower panel) by immunoblotting with polyclonal antibodies. Extracts prepared from cells lacking Far1p (lane 5) or Ste4p (lane 6) confirm the specificity of the antibodies. Note thatste4Δ cells express low amounts of Far1p, because basal levels of Far1p depend on a functional signaling pathway (31). SN, soluble extract prior to immunoprecipitation; IP, immunoprecipitate with HA11 antibodies. The arrow points to the position of Far1p; the brackets mark the different forms of HA-Ste4p or untagged Ste4p. The asterisk marks the position of immunoglobulin G. (D) Far1p is specifically required for the interaction between Ste4p and Cdc24p. The interaction between Cdc24p and Ste4p was measured in different two-hybrid strains (14, 15). Bars show mean β-galactosidase activity ± SD for four independent transformants. Plasmids were pBTM-CDC24 (DBD-Cdc24), pGADXP (vector), pGAD-STE4 (AD-Ste4, low expression), and pGADXP-STE4 (AD-Ste4, high expression) (14). (E) Far1p and Cdc24p function in a common pathway during cellular orientation (19).far1-c or cdc24-m single mutants mate with comparable efficiency to far1-c cdc24-m double mutants, suggesting that the two mutant proteins are defective in the same cellular function. In contrast, far1-c pea2-2 cells are synthetic sterile (5).

To determine whether Far1p might be an effector of Gβγ, we next tested the ability of Far1p to coprecipitate with Ste4p, the β subunit of the yeast heterotrimeric G protein (Fig. 1C). Far1p was readily detectable in HA11 immunoprecipitates from cells expressing epitope-tagged Ste4p (HA-Ste4p) but not in immunoprecipitates from control cells expressing untagged Ste4p. In addition, a specific interaction between Far1p and Ste4p was detected by two-hybrid analysis (Table 1 and Fig. 2A). Because expression of Ste4p activates the pheromone response pathway, these experiments do not address whether the interaction between Far1p and Ste4p is regulated by pheromones. However, Gβγ bound to Far1p and could use Far1p to orient the cytoskeleton toward the mating partner.

Figure 2

Analysis of wild-type and Far1-s proteins for their ability to interact with Ste4p and the polarity establishment proteins Bem1p, Cdc24p, and Cdc42p. (A) Domains required for the interaction between Far1p and Bem1p, Cdc24p, Cdc42p, and Ste4p were determined by two-hybrid analysis. Wild-type or various mutant Far1p fused to the activation domain are schematically represented on the left (20); expression of the β-Gal reporter was quantified and shown as Miller units with standard deviations as described (15). Far1p1–389correspond to Far1p-c (5, 31). Note that the binding sites for Ste4p and Cdc24p are separable. (B) A schematic representation of the binding domains between Far1p and Bem1p, Cdc24p, and Cdc42p. Ste4p binds to the RING finger domain in the NH2-terminal part of Far1p (gray box), whereas Bem1p and Cdc24p require sequences in the COOH-terminal domain of Far1p. The striped box indicates a domain of Far1p that shows homology to Ste5p (28). Binding of Far1p to Cdc42p-GTP is dependent on the presence of Bem1p (Table 1).

Cdc24p interacts with Ste4p (6, 18), and mutants of Cdc24p have been identified which fail to interact with Gβγ (6). The interaction of Cdc24p and Ste4p in vivo is likely to depend on Far1p. First, the interaction between Cdc24p and Ste4p assessed by the two-hybrid system was abolished in strains deleted for FAR1 (Fig. 1D). Second, Far1p interacted independently with both Cdc24p and Ste4p (Table 1 and Fig. 2), and Cdc24 mutant proteins that were unable to interact with Ste4p (6,18) were all unable to bind Far1p (Table 1), indicating that the domain of Cdc24p that mediated interaction with Far1p was also required for the interaction with Ste4p. Finally, genetic analysis indicated that cdc24-m and far1-s mutants are defective in the same pathway (Fig. 1E) (19). Thus, Far1p is needed to bridge the interaction between Gβγ and the polarity establishment proteins in vivo.

To address the functional importance of the interaction between Far1p, Gβγ, and the polarity establishment proteins, we tested whether any of the mutant Far1 proteins unable to orient polarization in vivo (5) failed to interact with Bem1p, Cdc42p, Cdc24p, or Ste4p (Fig. 2) (20). Far1p1–389, which lacked the COOH-terminal half of the protein, was defective for interacting with Bem1p, Cdc24p, and Cdc42p, although it was able to bind Ste4p. Similarly, the Far1p-H7 mutation (5) was unable to bind to Bem1p and Cdc42p and exhibited strongly reduced binding to Cdc24p, but still allowed efficient interaction with Gβγ (Fig. 2). Analysis of additional Far1p mutants indicated that the RING finger domain of Far1p was necessary and sufficient for interaction with Gβγ. Mutants lacking all or part of the RING finger domain were unable to form oriented mating projections,suggesting that binding of Gβγ to Far1p is essential for the mating function of Far1p in vivo (21). Thus, Far1p contains separable binding sites for Gβγ, Cdc24p, and Bem1p (Fig. 2B), and binding of Far1p to both Gβγ and the polarity establishment proteins is likely to be required for oriented cell polarity during mating.

Although pheromones did not alter the interaction between Far1p, Gβγ, and the polarity establishment proteins when assayed by coimmunoprecipitation or two-hybrid experiments, these interactions may be regulated in vivo by compartmentalization of Far1p. In the absence of pheromone, Far1p was found predominantly in the nucleus of G1 cells (Fig. 3) (22). Because Bem1p and Cdc42p are localized at the site of polarized growth during budding (23), Far1p may be unable to interact with these proteins during vegetative growth in the absence of α factor. We found that a fraction of Far1p was distributed throughout the cytoplasm in cells treated with pheromones (shmoos; Fig. 3), indicating that Far1p relocalizes from the nucleus to the cytoplasm in response to pheromones (24). Far1p did not accumulate at shmoo tips of α factor-treated cells, suggesting that only a small fraction of Far1p interacts with Gβγ or that its interaction with Gβγ might be transient. Like wild-type Far1p, truncated Far11–389 protein was localized to the cytoplasm in pheromone-treated cells (Fig. 3), indicating that the COOH-terminal domain, which mediated the interaction with Bem1p, Cdc24p, and Cdc42p, was not required for the redistribution of Far1p. Relocalization of Far1p appears to require activation of the mitogen-activated protein (MAP) kinase signaling pathway triggered by α factor (25). Thus, Far1p changes its localization in an α factor–dependent manner, suggesting that compartmentalization prevents Far1p from interacting with Gβγ and the polarity establishment proteins in the absence of pheromones.

Figure 3

Far1p relocalizes from the nucleus to the cytoplasm in response to pheromones. Cells expressing either full-length Far1p (upper row) or Far1p1–389 (Far1p-c, lower row) fused to GFP were treated (right panels, +) or not treated (left panels, –) with α factor. No staining was detected in cells expressing untagged Far1p (41). Photographs show GFP fluorescence overlaid with the corresponding phase contrast image. Note that Far1p is nuclear in G1 cells in the absence of α factor but accumulates in the cytoplasm of cells treated with α factor.

Our results support the following model for how growth is directed toward the mating partner. In the absence of pheromones, Far1p is localized in the nucleus, and the polarity establishment proteins organize the actin cytoskeleton toward the bud site. The presence of pheromones activates the receptors, leading to dissociation of Gβγ from Gα at the plasma membrane. Gβγ then activates the pheromone response pathway resulting in redistribution of Far1p from the nucleus to the cytoplasm. Cytoplasmic Far1p is recruited to the site of the incoming signal by binding to Gβγ, thereby targeting the polarity establishment proteins Cdc24p, Bem1p, and Cdc42p to the site of polarization. Cdc42p becomes locally activated, which triggers polymerization of the actin cytoskeleton. We propose that Far1p functions as a scaffold molecule linking Gβγ to the polarity establishment proteins Cdc24p, Bem1p, and Cdc42p. Far1p is thus analogous to Ste5p, which links Gβγ to the MAP kinase cascade by directly interacting with Fus3p, Ste7p, and Ste11p (26,27). In addition to this functional similarity, Ste5p and Far1p share two domains with significant sequence similarity (28): an NH2-terminal RING finger that is necessary for Gβγ binding (29), and a short stretch in the COOH-terminus. Although both Far1p and Ste5p interact with Bem1p, neither of the two conserved domains appears to be involved in this binding (Fig. 2). Mutations in Ste4p have been identified which function efficiently for signal transduction, but exhibit a severe mating defect presumably because they are unable to orient cell polarity toward the mating partner (30), suggesting that specific Ste4p mutations may be able to distinguish between several effectors. It is not known whether these mutant Ste4 proteins are defective for their interaction with Far1p.

Far1p was originally identified because it is required to arrest the cell cycle in response to pheromones, probably by inhibiting the activity of the cyclin-dependent kinase Cdc28p-Clnp (31). As we show, Far1p also functions as an effector of Gβγ which is involved in cytoskeletal polarization during mating. These two activities are mutationally separable (5, 32). Thus, distinct binding partners of Far1p are necessary to mediate the different functions, suggesting that multiple Far1p complexes execute these responses in vivo (5).

Heterotrimeric G proteins determine cell polarity in a variety of organisms, for example, in orientation of cell division axes in earlyCaenorhabditis elegans embryos (32) or in response to chemotactic cytokines in leukocytes (33). InDrosophila melanogaster, signaling mediated by the G protein–coupled receptor, Frizzled, polarizes precursor cells and specifies asymmetric cell divisions by properly orienting mitotic spindles (34). Although the effectors of the G proteins in these systems are not known, the domain of Cdc24p required to interact with Far1p is conserved in mammalian exchange factors such as the DBL proto-oncogene (6), suggesting that Far1p-like molecules may link G protein coupled receptor signaling pathways to polarized cell growth in all eukaryotes.

  • * To whom correspondence should be addressed. E-mail: matthias.peter{at}


Stay Connected to Science

Navigate This Article