|Home | About | Journals | Submit | Contact Us | Français|
Chemical gradients of peptide mating pheromones are necessary for directional growth, which is critical for yeast mating. These gradients are generated by cell-type specific secretion or export and specific degradation in receiving cells. Spatial information is sensed by dedicated seven-transmembrane G-protein coupled receptors and yeast cells are able to detect extremely small differences in ligand concentration across their ~5-µm cell surface. Here, I will discuss our current knowledge of how cells detect and respond to such shallow chemical gradients and in particular what is known about the proteins that are involved in directional growth and the establishment of the polarity axis during yeast mating.
Chemical gradients play critical roles in a large number of developmental processes. These gradients, which are precisely controlled at both temporal and spatial levels, provide an efficient means of encoding vectorial information. Although diverse fungi generate chemical pheromone gradients during mating, because of its genetic tractability, most studies on chemical gradients in yeast have focused on Saccharomyces cerevisiae. This review will focus on the role of chemical pheromone gradients during S. cerevisiae mating, as our knowledge of the molecular details of this chemotropic process is most advanced. This review is divided into four main sections: gradients, sensing and detection, cellular responses, and signal amplification. The second and third sections are further divided, with the second section addressing the roles of pheromone receptors and hetero-trimeric G-proteins, as well the mechanisms involved in detection of small differences in ligand concentration across a yeast cell. The third section focuses on the different cellular responses during chemotropism, in addition to what is known about how a growth site and an axis of polarity are established in response to an external gradient.
Most fungi produce diffusible peptide mating pheromones and in general, there are two different peptide mating factors (reviewed in Caldwell et al. 1995) in each species. Given the dynamic nature of pheromone gradients, it is not surprising that they have not been directly measured. Rather, various methods have been used to generate such gradients artificially and to determine cellular responses over a range of chemical gradient concentrations and slopes. During S. cerevisiae mating, pheromone gradients are generated by the secretion of a- and α-factor. There are three factors that are likely to be important for the generation of pheromone gradients: first, the physical characteristics of the mating pheromone; second, the site(s) of pheromone secretion or export; and third, the distribution of proteases that degrade mating pheromone. The shape of a mating pheromone gradient is likely to be dependent on the diffusibility of the mating peptide, and as a-factor is farnesylated, its diffusion in aqueous environments is decreased. This posttranslational modification may be critical in hydrophobic milieu, such as cellular membranes or natural habitats including biofilms. The increased hydrophobicity and reduced diffusibility of a-factor (Caldwell et al. 1994; Khouri et al. 1996) may contribute to the further sharpening (increasing the steepness) or stabilizing of its gradient compared with that of α-factor. Furthermore, pheromones are likely to be secreted (α-factor) or exported (a-factor via the Ste6 transporter) from the tip of shmoos (Kuchler et al. 1993), and this site-specific release will dictate the shape of the gradient. Site-specific secretion or export will generate anisotropy in the mating pheromone gradient, which should provide information on cell position. Specifically, the location of a particular shmoo tip should be evident by the position, direction, and shape of its pheromone gradient.
Mating pheromone proteases are also secreted, yet little is known about whether their secretion occurs at a specific site. Proteases, such as the barrier aspartyl protease Bar1, limits the distance of the α-factor tridecapeptide gradient (Barkai et al. 1998). Similarly, a protease activity that degrades a-factor has been identified in MATα cells (Marcus et al. 1991). Each of these pheromone proteases are expressed only in receiving cells. The α-factor protease Bar1 is largely secreted, with 95% of the Bar1 activity found extracellularly (Ciejek and Thorner 1979; Ballensiefen and Schmitt 1997). This protease has also been found, however, attached to the cell wall (Moukadiri et al. 1999). a-factor protease activity is also essentially cell-associated (Marcus et al. 1991). A gradient of secreted α-factor protease, or cell-associated a-factor protease, is likely to sharpen the pheromone gradient. Finally, pheromone degradation via a cell-associated protease has been proposed as a means for distinguishing potential mating partners amongst many cells (Barkai et al. 1998).
Yeast cells must detect a mating pheromone gradient to initiate a range of cellular responses. Critical for such detection are the classical seven-transmembrane domain receptors coupled to hetero-trimeric G-proteins (for review, see Xue et al. 2008), the a-cell specific Ste2 (which binds α-factor), and the α-cell specific Ste3 (which binds a-factor). These receptors do not share substantial sequence homology. Transcription of both receptors is rapidly increased on addition of the respective pheromone (Hagen and Sprague 1984; Hartig et al. 1986). Haploid a-cells have ~10,000 cell-surface binding sites for α-factor with a Kd of 4–6 nM (Jenness et al. 1986; Stefan et al. 1998). This high-affinity receptor resting state is converted to a low-affinity state, with an ~10-fold increase in ligand off-rate and corresponding increase in Kd, in the presence of GTPγS (Blumer and Thorner 1990).
When haploid a-cells bind α-factor, Ste2 receptors are internalized with a tavg. of ~8 min (Schandel and Jenness 1994) and reappear on the cell surface after ~60 min (Jenness and Spatrick 1986; Ayscough and Drubin 1998; Stefan et al. 1998). Both a- and α-factor receptors have substantial—183 and 133 amino acids, respectively—cytoplasmic carboxy termini. Truncation of the Ste2 carboxy-terminal 105 amino acids results in a ~fivefold increase in cell surface pheromone binding sites, a 10-fold increase in pheromone sensitivity, and a defect in pheromone induced morphogenesis (Konopka et al. 1988). Ste2 receptors also oligomerize in vivo, yet this association does not depend on ligand nor hetero-trimeric G-protein (Overton and Blumer 2000; Yesilaltay and Jenness 2000). Ste2 transmembrane domains five and six together with the intervening third intracellular loop are important for G-protein activation, via Gα (Konopka and Jenness 1991; Clark et al. 1994; Stefan and Blumer 1994; Dube et al. 2000; Celic et al. 2003; Lin et al. 2003). The carboxy-terminal cytoplasmic domain of the receptor interacts with Gα in a preactivation complex (Dosil et al. 2000; Wu et al. 2004). The extracellular end of transmembrane domain 1, together with the extracellular loops 1 and 2, interact directly with pheromone (Son et al. 2004; Hauser et al. 2007). Although pheromone receptors are necessary for detecting pheromone, it is likely that their internalization and recycling play an important role in signal amplification and desensitization.
In the absence of mating pheromone, both pheromone receptors localize uniformly over the cell surface as well as endosomal and vacuole compartments (Jackson and Hartwell 1990b; Davis et al. 1993; Ayscough and Drubin 1998; Stefan and Blumer 1999). During the later stages of mating, Ste2 is localized to mating projections, raising the possibility that this polarized distribution is important for chemotropic growth (Jackson et al. 1991). In cells treated with isotropic saturating pheromone concentrations, little receptor was observed on the cell surface after 20 min and instead a punctate cytoplasmic distribution was observed (Ayscough and Drubin 1998). Finally, after ~90 min, the receptor was localized to the mating projection, enriched at the tip (Ayscough and Drubin 1998) (Fig. 1). In these conditions, the actin cytoskeleton is required for Ste2 polarization (Ayscough and Drubin 1998). It is likely that the actin cytoskeleton is critical for receptor exocytosis and endocytosis, which are important for the receptor polarization (via targeting and maintenance). It will be essential to follow receptor distribution during chemotropism (in response to defined pheromone gradients and during mating) to better understand the role of receptor secretion, endocytosis, and polarization during directional growth.
The pheromone receptor functions as a guanine nucleotide exchange factor for Gα. On pheromone binding, the receptor catalyzes the exchange of GDP for GTP in the Gα•GDP Gβγ complex, resulting in the dissociation of Gα•GTP from Gβγ (reviewed in Dohlman and Thorner 2001; Dohlman 2002). Gα and Gβγ are then able to associate and activate downstream effectors. This signaling competent state of the Gα and Gβγ subunits exists until the Gα•GTP is hydrolyzed. RGS (Regulators of G protein Signaling) proteins such as Sst2, function as GTPase accelerating proteins or GAPs facilitating Gα GTP hydrolysis (Apanovitch et al. 1998; DiBello et al. 1998). Receptor and hetero-trimeric G-protein turnover, either by cellular uptake (endocytosis) or posttranslational modification (including modifications that target proteins for degradation) also contribute to down-regulation.
Initial studies indicated that released Gβγ functions in a positive fashion by initiating a range of pheromone responses (Jahng et al. 1988; Whiteway et al. 1989; Cole et al. 1990; Nomoto et al. 1990) including cell cycle arrest, gene transcription, cell morphology changes, and chemotropic growth (Figs. 2 and and3).3). More recently, however, several studies have revealed that Gα, in addition to its role as a negative regulator (via inactivation of released Gβγ), can also activate signaling (Metodiev et al. 2002; Slessareva et al. 2006).
The Gα, Gβ, and Gγ subunits are all localized to the plasma membrane and internal membranes, with Gα and Gγ each having dual posttranslational lipid modifications that result in membrane association (Song et al. 1996; Hirschman et al. 1997). These modifications are necessary for function, with Gα being both palmitoylated and myrisotoylated (Stone et al. 1991; Dohlman et al. 1993; Song and Dohlman 1996; Song et al. 1996; Manahan et al. 2000) and Gγ being palmitoylated and prenylated (Grishin et al. 1994; Whiteway and Thomas 1994; Hirschman and Jenness 1999). Although Gα is thought to associate with the pheromone receptors via its carboxy terminus (Hirsch et al. 1991), its membrane localization does not require Gβγ or receptors (Blumer and Thorner 1990; Song et al. 1996). Conversely, although Gα is required for Gβγ membrane localization, the pheromone receptor is not required (Song et al. 1996; Hirschman et al. 1997). The localization of the Gα (Metodiev et al. 2002), Gβ (Kim et al. 2000), and Gγ (Nern and Arkowitz 2000a; Moore et al. 2008) has been examined in both isotropic pheromone concentrations (Gα and Gβ) and in pheromone gradients during mating or generated artificially (Gγ) (Fig. 1). These three proteins localize preferentially to the mating projection (also appear on internal membranes), similar to the distribution of the pheromone receptor (Jackson et al. 1991; Ayscough and Drubin 1998; Moore et al. 2008). The subcellular distribution of the different hetero-trimeric G-protein subunits and their relationship to one another during the mating process (chemotropic growth) has not been quantitatively examined at high temporal and spatial resolution. Such analyses will be important for understanding the requirements and feedback loops that are necessary for chemotropic growth.
It is highly unlikely that yeast cells use a temporal mechanism to detect small differences in ligand concentration given that they grow relatively slow, i.e., maximally their 5 µm diameter after ~6 h in pheromone (Segall 1993; Vallier et al. 2002). It is largely accepted, that yeast cells use a spatial detection mechanism, as is the case for leukocytes, in which pheromone concentration is compared on two sides of the cell. If a gradient is generated by a micropipet (Fig. 4), using a 70-nM α-factor solution, released at a rate of 2 nL/min and cells are positioned ~50 µm from the tip of the micropipet, then the cell half closest to the pipet is estimated to experience 11.5-nM α-factor and that farthest from the pipet 10.2-nM α-factor (Segall 1993; Vallier et al. 2002). Assuming a uniform distribution of receptors, ~66 more receptors would have pheromone bound closest to the gradient compared with the opposite side (a difference in receptor occupancy of ~1.3%). A recent study using microfluidics chamber-generated pheromone gradients (Fig. 4) has extended these calculations based on the observation that yeast cells can sense the direction of the pheromone gradient with a slope relative to the average concentration ([pheromoneslope]/[pheromoneconc. at the mid point]) as shallow as 0.1%/µm−1 (Moore et al. 2008); a difference of 0.5% in ligand concentration from the cell front and back. When cells are in a 0–1000-nM gradient at an average concentration of 500 nM, then the absolute slope is 2.5 nM/µm. This results in 4951 and 4950 receptors occupied in the front half and the back half of the cell, respectively! Strikingly, a bar1 mutant strain was able to weakly orient (cosine of the angle of orientation ~0.2: angle of ~78º) to a 0–1000-nM pheromone gradient at >900-nM pheromone, in which all receptors should be saturated. Furthermore, this elegant microfluidics study revealed that bar1 mutants are able to orient their growth to a gradient over almost three orders of magnitude (4 nM–930 nM). The calculations of receptor occupancy and gradient detection assume constant number of pheromone receptors, i.e., do not take into consideration receptor induction, exocytosis, and endocytosis. These different processes undoubtedly affect the real difference in receptor occupancy in a pheromone gradient, yet nevertheless, an extremely small difference in ligand concentration is likely to be detected during mating. How this very small difference is translated into much steeper intracellular gradient(s) remains a fundamental question.
Cellular responses can either be dictated by absolute pheromone concentrations or the fractional change in concentration (Δ[pheromone]/Δdistance, i.e., gradient steepness) across the cell (Walter et al. 1990; Goodhill 1998; Goodhill and Baier 1998). In nerve growth cones (Isbister et al. 2003) and in round immobile latrunculin-treated Dictyostelium discoideum (Janetopoulos et al. 2004), directional responses are dictated by the gradient steepness as has also been observed in chemotaxing neutrophils (Zigmond 1977). In yeast, directional growth has been shown to occur in response to the fractional gradient and not the absolute pheromone concentration (Paliwal et al. 2007). Fractional gradient is the ratio of the pheromone concentration gradient (offset by a gradient value γ0; γ − γ0), and the difference between the mean pheromone and minimum pheromone concentration necessary for shmoo formation, and can be thought of as the steepness of the gradient divided by the pheromone concentration. Accuracy of gradient sensing was highest at low pheromone concentrations and high gradients (Segall 1993; Paliwal et al. 2007; Moore et al. 2008). Accuracy of gradient sensing is the cosine of the angle of the mating projection relative to the pheromone gradient; a value of one indicates perfect alignment. Indeed, gradient sensing appears to be optimal near the pheromone concentration that equals the receptor Kd (Segall 1993; Paliwal et al. 2007; Moore et al. 2008). Precision of gradient sensing has been defined as the root mean square (rms) of several cells between the angles of observed mating projections and the direction of the gradient, with precision increasing as rms decreases (Paliwal et al. 2007). Furthermore, the precision of gradient sensing is constant at a constant fractional gradient and increases as the fractional gradient increases (Paliwal et al. 2007). It is thought that the combination of constant pheromone release (although this has not been validated experimentally), together with first-order reaction of pheromone degradation by the extracellular proteases, should result in an exponential gradient (Barkai et al. 1998). Hence, as cells grow toward the source of an exponential gradient, both the gradient steepness and the pheromone concentration would increase, resulting in a constant fractional gradient and hence constant precision of gradient sensing (Paliwal et al. 2007). Although this makes biological sense, it remains to be seen if the precision of directional sensing is indeed constant throughout the mating process.
In response to a micropipet-generated pheromone gradient (Fig. 4), orientation accuracy of yeast cells increases over time (cosine of the angle of orientation increasing from ~0.55; angle of ~57º to 0.85; angle of ~32º from shmoo formation to ~5 h later) (Vallier et al. 2002). These results suggest that mating projection reorientation can occur over time (Paliwal et al. 2007). Very recently, studies using microfluidics chamber-generated pheromone gradients (Fig. 4) confirmed that directional accuracy improves over time (Moore et al. 2008). In the range of 5–40 nM of α-factor (in a 0–100 nM gradient), cells formed broad mating projections which reoriented toward the gradient, with an orientation accuracy of ~0.4 (angle of 66º) to ~0.6 (angle of 53º) at 2 and 6 h, respectively. At pheromone concentrations greater than 50 nM, cells frequently formed a second mating projection that was more oriented than the first (Moore et al. 2008). Assuming that the pheromone gradient is linear, then this increase in the precision of gradient sensing over time can be explained by an increase in the fractional gradient: The pheromone concentration should increase ~10%–20%, given the above parameters at an average pheromone concentration of the receptor Kd over one to two cell lengths.
Using a micropipet-generated pheromone gradient, cells lacking the Bar1 protease orient similarly to wild-type cells, at low pheromone concentration (Segall 1993). The bar1Δ mutant was extensively characterized in a recent microfluidics chamber study (Moore et al. 2008). Compared with a wild-type BAR1+ strain, bar1Δ cells oriented to 0–100 nM and 0–1000 nM pheromone gradients with decreased accuracy. Furthermore, Bar1 functioned even in the constant flow (1 µL/min) in ~1 µL chambers, consistent with cell-wall associated protease (Moukadiri et al. 1999). Surprisingly, both BAR1+ and bar1Δ cells exhibited the best orientation at [pheromone]avg. that were equivalent to the receptor Kd (Moore et al. 2008). However, it should be pointed out that in the case of the bar1Δ cells, this was in a relatively steep portion of a 0–10 nM pheromone gradient, whereas for BAR1+cells, this was at a shallow portion of a 0–100 nM pheromone gradient, suggesting that Bar1 facilitates sensing of shallow gradients. Furthermore, in a microfluidic chamber-generated 0–100 nM pheromone gradient, sst2Δ bar1Δ cells are incapable of orienting toward the gradient (Moore et al. 2008). Consistent with results using a micropipet-generated gradient (Segall 1993), at low pheromone concentrations (0–10 nM gradient), the accuracy of orientation of bar1Δ sst2Δ cells was less than the corresponding bar1Δ cells in the same conditions (Moore et al. 2008).
In response to pheromone gradients, cells induce gene transcription, undergo cell cycle arrest, change shape, and grow toward a mating partner. The first three responses can occur during mating, in which a pheromone gradient is present, as well as in an isotropic pheromone concentration. The latter response, however, only occurs when a pheromone gradient is present. The optimum mating pheromone concentration for gene induction, cell cycle arrest, and cell shape changes does not seem to differ substantially between isotropic and gradient pheromone conditions (Paliwal et al. 2007; Moore et al. 2008). Recently, it has been shown that the pheromone concentration range of gradients at which cells are sensitive for induction of the mating-specific gene FUS1 coincides with that for directional growth (Paliwal et al. 2007; Moore et al. 2008). This may be because some genes essential for gradient sensitivity are induced by this MAP kinase cascade or that, despite the ability to genetically separate the MAP kinase and chemotropism pathways (Schrick et al. 1997; Strickfaden and Pryciak 2008), there may be cross-talk between these pathways. More than 100 genes are induced by pheromone (Erdman et al. 1998). Genes whose induction is more efficient with a pheromone gradient than with an isotropic pheromone concentration would be ideal candidates for modulators of the chemotropic response.
In yeast, as in many organisms, changes in cell shape can be separated from directional growth. In isotropic pheromone conditions, yeast cells become elongated and form shmoos (Fig. 2). A range of proteins are necessary for these morphology changes, some of which are mating-specific, and others are more generally involved in cell polarity yet appear to have specific functions in response to pheromone (Table 1). This process is referred to as “default” polarization (Madden and Snyder 1992; Dorer et al. 1995) and the previous bud site provides the spatial cue for mating projection formation. In addition to the receptor and hetero-trimeric G-protein, the pheromone-dependent MAP kinase Fus3 and the pheromone-induced proteins Afr1 and Pea2 are required for shmoo formation (Chenevert et al. 1994; Valtz and Herskowitz 1996; Matheos et al. 2004). Other proteins required for cell polarity are particularly important for morphology changes in response to pheromone. Pheromone-dependent cell-shape changes can be separated from chemotropic growth. This may be, in part, because of the fact that most studies on cell-shape changes have been performed in saturating pheromone concentrations that may mask chemotropic effects.
Different assays have shown that the pheromone receptor, the hetero-trimeric G-protein, and several proteins that bind specifically to the released Gα and Gβγ convey the spatial information from the pheromone gradient to the downstream proteins critical for pheromone-dependent morphology changes (Table 1). Elegant studies showed that both MATa and MATα cells choose between potential mating partners (Jackson and Hartwell 1990a; Jackson and Hartwell 1990b). Specifically, each cell type chooses the mating partner that produces the highest level of pheromone. The simplest model to account for this is that cells detect a mating pheromone gradient and orient growth toward the cell producing the steepest gradient (Fig. 3). In these studies, two assays were used: the competition mating assay and the discrimination assay. In both assays, a cell is given a choice between two mating partners of the opposite mating type. In the competition assay, there are two types of mating partners present: target and challenger cells. Typically, the target cells are wild-type, and the genotype of the challenger cells is varied. The importance of a given challenger cell is assessed by its effect on the mating of signaling cells with the target (responder) cells. In the discrimination assay, the relative mating efficiency of one cell with two different partners is determined, measuring the ability of one cell to discriminate between two mating partners. The confusion assay examines the importance of the pheromone gradient during mating by determining the effect of disrupting the gradient by α-factor addition (Dorer et al. 1995; Valtz et al. 1995). This results in a substantial decrease in mating efficiency with wild-type cells not observed, however, with chemotropism mutants. Although these assays are extremely useful in the identification and analysis of cellular components required for chemotropic growth, they are indirect measures of chemotropism.
The pheromone receptor, the Gα, Gβ and Gγ subunits, Sst2, and Bar1 have been shown by both the discrimination and confusion assays to be important for chemotropism (Jackson and Hartwell 1990a; Jackson et al. 1991; Schrick et al. 1997). The discrimination assay has also revealed that actin, myosin V, and clathrin are likely to be required for chemotropism (Jackson et al. 1991). The effects of the sst2 and bar1 mutants in the discrimination and confusion assays is probably caused by their altered pheromone sensitivity (Jackson and Hartwell 1990a; Dorer et al. 1995). These assays have also shown that different components of the pheromone-dependent MAP kinase pathway are not required for chemotropism (Schrick et al. 1997; Strickfaden and Pryciak 2008). Interestingly, MAP kinase Fus3 mutants are somewhat defective in mating-partner discrimination (Dorer et al. 1997), whereas fus3Δ kss1Δ mutants are not. In contrast, both of these mutants are unaffected in the pheromone-confusion assay (Dorer et al. 1997). A far1Δ mutant was only slightly defective in the discrimination assay, yet fully defective in the pheromone confusion assay (Dorer et al. 1995; Valtz et al. 1995).
Specific mutants for the α-factor receptor, Gα, and Gβ have also been examined with these assays; the former two are defective in the mating partner discrimination assay (Jackson and Hartwell 1990a; Metodiev et al. 2002), and latter is defective in the confusion assay (Strickfaden and Pryciak 2008). The defect of the Ste2 mutant lacking its cytoplasmic carboxy terminal tail (ste2-T326) in the mating partner discrimination assay is thought to be because of its increased sensitivity to pheromone (Jackson and Hartwell 1990a). The mating partner discrimination defect observed with the gpa1EE mutant, together with the observation that it is defective in binding Fus3, suggests that released Gα•GTP recruits Fus3, which then phosphorylates proteins required for chemotropism. The Gβ mutations ste4[K126E] and ste4[L117R] disrupt interactions with Gα and the pheromone receptor (Strickfaden and Pryciak 2008). Finally, both discrimination and confusion assays have been used to confirm that a range of proteins required for polarized growth and morphology changes in response to pheromone—such as Spa2, Bni1, Fus1, Fus2, Pea2, Rvs161, Bud proteins, Myo4, and tubulin—are not necessary for chemotropism (Jackson et al. 1991; Dorer et al. 1997).
Segall developed a direct measure of chemotropism in which the direction of growth of MATa cells in a micropipet-generated gradient is followed (Segall 1993). This assay was used to show that bar1 mutants orient their growth normally toward the pheromone source, whereas sst2 mutants are somewhat defective. This assay has also revealed that bud1, spa2, kss1, afr1, ste20 and mating defective bem1-m1 mutants all orient their growth normally in a pheromone gradient (Leberer et al. 1997; Butty et al. 2002; Vallier et al. 2002). A ste2-T326 mutant orients growth in a pheromone gradient (at lower pheromone concentrations compared with the wild type), yet is unable to increase its orientation toward the source of pheromone over time as is observed with wild-type cells (Vallier et al. 2002). Far1 (far1-H7 and far1-D1) and cdc24 (cdc24-m1 and cdc24-m3) mutant alleles, which are specifically defective in mating, are defective in the pheromone-confusion assay and in oriented growth toward a pheromone gradient (Valtz et al. 1995; Nern and Arkowitz 1998). Both of these types of mutants, instead, orient their mating projections adjacent to the site of their previous bud scar, i.e., mate by default.
Considered together, results suggest that two different landmarks, either that dictated by previous bud site or by pheromone gradient, activate the small G-protein Cdc42; Cdc42, in turn, is critical for directing the actin cytoskeleton and ultimately the site-specific incorporation of new plasma membrane and cell-wall material. It is presumed that limiting amounts of the Cdc42 activator Cdc24, at either site ensures that budding and shmooing can not occur simultaneously. Recently, pheromone gradients generated in a microfluidics chamber have been used to assess yeast chemotropism (Fig. 4) (Paliwal et al. 2007; Hao et al. 2008; Moore et al. 2008). The advantage of this method is that the pheromone gradient can be exquisitely controlled and its slope and concentration can be easily quantified. These studies have shown quantitatively that optimal chemotropism occurs at average pheromone concentrations at the receptor Kd and, as expected, with steeper pheromone gradients. At low pheromone concentrations, in which bar1 mutant strains are able to orient their mating projections, a kss1Δ bar1 mutant is somewhat defective. Hence, the kss1Δ mutation decreases the range of pheromone concentration at which a directional response is efficient and in addition, this mutant is defective in improving its directional growth over time (Paliwal et al. 2007). Another microfluidics-chamber study showed that fus3Δ BAR1 cells and a Ste5 mutant in which Fus3 docking is defective (ste5ND in combination with kss1Δ) were both defective in chemotropic growth, whereas kss1Δ BAR1 mutants were unaffected (Hao et al. 2008). Together, these studies highlight the importance of the pheromone-dependent MAP kinases in chemotropic growth. It is likely that the different results regarding the chemotropism of kss1Δ strains is caused by the presence or absence of Bar1, which alters the shape of the pheromone gradient.
A recent microfluidics-chamber gradient study (Moore et al. 2008) revealed that a Ste2 receptor truncation mutant, ste2300Δ (in a bar1Δ strain) was unable to orient toward a shallow gradient at high pheromone concentrations (~90 nM α-factor in a 0–100 nM gradient), yet was able to orient toward the same gradient at lower pheromone concentrations. In a 0–1000 nM pheromone gradient, the orientation defect of these cells is more striking. It is likely that these mutants are defective in maintaining oriented growth (Vallier et al. 2002). Together, these defects (Vallier et al. 2002; Moore et al. 2008) suggest that the receptor carboxy-terminal tail may be important for signal amplification and desensitization.
As discussed previously, activation of the pheromone receptor on pheromone binding and concomitant exchange of Gα GTP for GDP results in dissociation of Gα from Gβγ. It is likely that both released Gα and Gβγ recruit proteins that are critical for the establishment of a growth site, leading to a new polarity axis. Gα most likely recruits activated Fus3 (Metodiev et al. 2002), whereas Gβγ most likely recruits the Cdc24–Far1 complex (Butty et al. 1998; Nern and Arkowitz 1998; Nern and Arkowitz 1999). Both Fus3 (van Drogen et al. 2001; Metodiev et al. 2002; Blackwell et al. 2003; Maeder et al. 2007; Slaughter et al. 2007) and Cdc24–Far1 (Toenjes et al. 1999; Nern and Arkowitz 2000b; Shimada et al. 2000) localize to the nucleus in G1 phase cells, and are subsequently enriched at the shmoo tip on pheromone exposure. The dynamics of nuclear Fus3 are unaffected by pheromone (van Drogen et al. 2001; Maeder et al. 2007), whereas the ratio of nuclear to cytoplasmic Fus3 signal in the presence of pheromone (Blackwell et al. 2003), suggesting that export is not affected. Rather, the appearance of a cortical anchor at the incipient shmoo tip, such as Gα or Gβγ–Ste5, is likely to alter the nuclear–cytoplasmic equilibrium (Metodiev et al. 2002; Slaughter et al. 2007). Fus3 is thought to be phosphorylated on binding to Gβγ–Ste5 and this activated kinase then presumably dissociates from this shmoo-tip site; shuttles into the nucleus (van Drogen et al. 2001; Blackwell et al. 2003) and is also captured at the shmoo tip via Gα, resulting in a gradient of phosphorylated Fus3 (Maeder et al. 2007). With respect to the Cdc24–Far1, pheromone appears to trigger nuclear export (Blondel et al. 1999; Nern and Arkowitz 2000b; Shimada et al. 2000). Regarding the cortical recruitment of Fus3 and the Cdc24–Far1 complex, it must be kept in mind that these changes in localization have not been examined during chemotropism. Furthermore, the kinetics of the dissociation of the hetero-trimeric G-protein and of the cortical recruitment of Fus3 and Cdc24–Far1 have not been examined in these same cells, which would reveal whether dissociation of Gαβγ is a prerequisite for recruitment.
Several studies suggest that activated phosphorylated Fus3 at the incipient shmoo tip is important for the establishment of the growth site via phosphorylation of Bni1 and other substrates (Matheos et al. 2004). In addition, a complex of Gβγ–Cdc24–Far1 at the incipient shmoo tip appears to be critical for the establishment of polarity and growth orientation (Butty et al. 1998; Nern and Arkowitz 1998; Nern and Arkowitz 1999). Given that specific Gα and Fus3 alleles appear to be defective in mating partner selection (Metodiev et al. 2002; Matheos et al. 2004), it is likely that there is communication between Gα, Fus3, and Bni1 proteins, on the one hand, and Gβγ, Cdc24, and Far1 proteins on the other. Furthermore, Fus3 is required for the pheromone-dependent cortical localization of Cdc24 and Bni1 (Matheos et al. 2004; Yu et al. 2008), consistent with some level of interaction between these two groups of proteins. It is, however, unclear whether Gα, Fus3, and Bni1 form a stable complex or transiently associate in vivo. In vitro, Gβγ, Cdc24, and Far1 form a complex (Nern and Arkowitz 1999), although the stability and lifetime of this complex in vivo is unknown. Furthermore, whether the communication between these two groups of proteins is direct, for example via complex formation, or indirect, e.g., by feedback loops, is unknown. One possibility is that Gα, Fus3, and Bni1 are required for generating actin cables, whereas Gβγ–Cdc24–Far1 is involved in restricting the location of actin cables on the cell cortex. In such a scenario, it is possible that the two protein complexes activate different Rho G-proteins: For example, Gα, Fus3, and Bni1 with Rho1, via interactions with Bni1 (Dong et al. 2003), and Gβγ–Cdc24–Far1 with Cdc42, via Cdc24. Interestingly, Gβγ has also been shown to bind Rho1 and recruit active Rho1 to the shmoo tip (Bar et al. 2003), perhaps forming a link between these two complexes. This postulated division of labor, one generating and the other localizing actin cables, is consistent with the observation that, in cdc24 and far1, chemotropism-defective mutants, combined with bud-site selection mutants, the site of growth wanders in the presence of pheromone (Nern and Arkowitz 2000a). Particularly relevant is the wandering of the sites of secretion and the actin cytoskeleton. Nonetheless, actin cables are still visible in these mutants. It should be noted that this growth-site drift has only been examined in isotropic pheromone concentrations, and should be investigated in pheromone gradients. Naively, one would assume that chemotropism-defective cdc24 or far1 mutants would be unaffected by the presence of a pheromone gradient, since while they should be able to sense spatial information, they do not respond to it. It will also be critical to look at the downstream readouts of signaling via these two complexes or groups of proteins, in particular the distribution of actin cable and active Cdc42. When and where Cdc42 is activated in response to pheromone gradients remains to be determined.
As discussed, a shallow gradient of mating pheromone is sufficient for a dramatic and robust chemotropic response (Moore et al. 2008). This response can be visualized at the level of oriented growth and also at the molecular level, i.e., tight localization of a range of proteins to the tip of the mating projection. For example, after treatment with saturating isotropic concentrations or a gradient of mating pheromone, the receptors (Jackson et al. 1991; Ayscough and Drubin 1998; Stefan and Blumer 1999; Moore et al. 2008), Gα (Metodiev et al. 2002), Gβ (Kim et al. 2000), Gγ (Nern and Arkowitz 2000a; Moore et al. 2008), Sst2 (Ballon et al. 2006), Fus3 (van Drogen et al. 2001; Metodiev et al. 2002; Maeder et al. 2007; Slaughter et al. 2007), Far1 (Nern and Arkowitz 2000b; Wiget et al. 2004), Cdc24 (Toenjes et al. 1999; Nern and Arkowitz 2000b; Shimada et al. 2000; Wiget et al. 2004), Bem1 (Ayscough and Drubin 1998; Butty et al. 2002), Ste20 (Peter et al. 1996; Leberer et al. 1997; Ash et al. 2003), Cdc42 (Ziman et al. 1993; Ayscough and Drubin 1998), Bni1 (Evangelista et al. 1997; Matheos et al. 2004), and Spa2 (Snyder 1989; Arkowitz and Lowe 1997; Ayscough and Drubin 1998; Moore et al. 2008) are all localized to the mating projection. The receptor and the proteins that bind directly to it (Gαβγ and Sst2) are localized to the entire region of new growth, i.e., over the entire surface of the mating projection, whereas the rest of these proteins are localized to restricted spots at the shmoo tip (Fig. 1). The localization of all proteins examined thus far (Ste2, Ste18, Ste20, Spa2, Cdc24, and Fus3) in cells exposed to a pheromone gradient (during mating or microfluidics chamber-generated) was qualitatively similar to that observed in isotropic saturating pheromone conditions. Although a range of amplification mechanisms and feedback loops have been examined for pheromone-dependent MAP kinase signaling, little is known about the amplification mechanisms that operate during chemotropism. Furthermore, it is likely that spatial amplification mechanisms vary over time, in that initial signal amplification events are distinct from those that occur after substantial exposure to a pheromone gradient (in which cell morphology changes and hence directed secretion play a major role). It appears that the localization of the pheromone receptor and associated Gαβγ reflects the pheromone gradient. However, it remains to be seen whether the cellular distribution of activated receptor, phosphorylated receptor or hetero-trimeric G-protein, and/or dissociated hetero-trimeric G-protein also mirrors the pheromone gradient, or whether the distribution of such activated components already reveals substantial amplification.
One can imagine that endocytosis and exocytosis are critical for signal amplification during chemotropism. Indeed, Ste2 endocytosis is required for optimal mating projection reorientation (Vallier et al. 2002). Such endocytosis and exocytosis amplification mechanisms would likely depend on specific cellular locations (shmoo tip vs. opposite end of cell), pheromone exposure (magnitude and slope of gradient), and a range of posttranslational modifications. Furthermore, membrane traffic could also contribute to amplification by affecting G-proteins (hetero-trimeric and small Rho GTPases), which are presumably modified at the endoplasmic reticulum and/or Golgi complex (Michaelson et al. 2002; Wright and Philips 2006). It has been suggested that amplification occurs at the level of the formation of the chemotropic protein complexes, i.e., Gα–Fus3–Bni1 and Gβγ–Cdc24–Far1. Amplification of spatial signals subsequent to the establishment of the initial pheromone gradient asymmetry is likely to also involve a positive feedback loop comprised of Cdc42-dependent actin polymerization and actin cable-dependent vesicle transport of Cdc42•GTP to the plasma membrane (Wedlich-Soldner et al. 2003; Wedlich-Soldner et al. 2004). Computer simulations of yeast chemotropism suggest that amplification comes from strong cooperativity and positive-feedback loops (Yi et al. 2007; Moore et al. 2008). In such a model, a Hill coefficient of >100 is required for polarization, and the investigators speculate that this may arise from the assembly of large protein complexes and cooperative binding reactions. The positive-feedback loop that Yi et al. modeled involves activated Cdc42 stimulating its own activation via Cdc24, Cdc42, and Bem1 (Butty et al. 2002). It is unlikely that this feedback loop plays a critical role in spatial signal amplification since a bem1-m1 mating mutant that does not bind Cdc24 and Far1 orients its growth normally in a pheromone gradient (Butty et al. 2002). One aspect of chemotropic fine-tuning that has been largely unexplored is the process of reorientation following a change in pheromone gradient direction. The ability to generate a range of pheromone gradients with different slopes and average concentrations will undoubtedly facilitate studies on the fine-tuning of directional responses.
A range of cells and tissues are able to detect chemical gradients, transducing extremely shallow gradients—which result in very small differences in ligand concentration across the cell surface—into steep intracellular signaling protein gradients. These resulting intracellular gradients are essential for directional growth and movement. During mating, yeast cells generate, sense, and respond to chemical gradients by directional growth toward their mating partner. It is likely that the ability of a cell to select the mating partner with the steepest pheromone gradient ensures the proximity of the partner and its maximal mating competence. Although a number of proteins are required for the yeast mating process, relatively few are known to have direct roles in chemotropic growth, from gradient detection to directional response. Despite the large number of studies on yeast chemotropism, many fundamental questions remain:
It will be critical to address these and other questions by examining chemotropism both in a physiological context, i.e., during mating, and in conditions in which the pheromone gradient can be quantitated, such as in microfluidics chambers. The ability to carry out genetic, biochemical, and high-resolution imaging approaches in yeast makes this system ideal for elucidating the basic principles of gradient sensing and directional responses that will undoubtedly be relevant to a range of chemotropic and chemotaxic processes.
I thank P. Follette and M. Bassilana for useful comments on the manuscript and in particular D. Stone for stimulating discussions. I apologize for not citing all primary work because of space limitations. R.A. is grateful for support from the Centre National de la Recherche Scientifique (CNRS), the Agence National de la Recherche (ANR), and the Fondation pour la Recherche Médicale (FRM) - BNP Paribas.
Editors: James Briscoe, Peter Lawrence, and Jean-Paul Vincent
Additional Perspectives on Generation and Interpretation of Morphogen Gradients available at www.cshperspectives.org