Phosphorylation of Gβ is crucial for efficient chemotropism in yeast

In metazoans, directed cell movement in response to a chemical gradient (chemotaxis), plays a vital role in embryogenesis, postnatal development and immunity. Moreover, the survival of many single-celled organisms depends on their ability to detect and move in response to chemical stimuli. The related phenomenon, directed cell growth in response to a chemical gradient (chemotropism), is also essential in many species. In mammals, chemotropism is integral to axon guidance (Hong and Nishiyama, 2010; Tojima et al., 2011) and angiogenesis (English et al., 2001; Basile et al., 2004). Pollen tube guidance is also a chemotropic process (Palanivelu and Preuss, 2000; Kim et al., 2004), and there is an increasing appreciation for the role of chemotropism in the life cycles of fungal species, including plant and human pathogens (Snetselaar et al., 1996; Daniels et al., 2006).

Many chemosensing cells express G-protein-coupled receptors (GPCRs) to detect the presence of the extracellular signaling molecules and determine the direction of their source (Weiner, 2002). Using the spatial information encoded in the distribution of activated receptors, the cell establishes a landmark for polarization. Actin and/or microtubule cytoskeletal elements are then assembled at, or recruited to, this site to promote directional movement or growth. Additionally, the cell must be able to stabilize the axis of polarity, while retaining the plasticity to change its orientation in response to changes in the direction of the gradient. Because physiological gradients of chemoattractant are typically very shallow (Mato et al., 1975; Tranquillo et al., 1988; Segall, 1993), establishment of the polarity site requires the conversion of small differences in receptor occupancy into a substantially steeper intracellular signaling gradient. This is thought to depend on feedback loops, such as those first discovered in Dictyostelium and neutrophils, that couple local excitation with global inhibition (LEGI) (Iijima et al., 2002; Weiner, 2002; Devreotes and Janetopoulos, 2003).

In comparison to chemotactic models, the amplification mechanisms that underlie directional sensing in chemotropic systems are less well understood. The best-characterized example of eukaryotic chemotropism to date occurs during the mating response of the budding yeast, Saccharomyces cerevisiae. In the haploid phase of its life cycle, budding yeast exist as two mating types, MATa and MATα. Each mating type constitutively secretes a peptide mating pheromone that binds to GPCRs on cells of the opposite type. Liganded receptors activate the mating-specific Gα protein, Gpa1 (henceforth Gα), via guanine nucleotide exchange and the subsequent dissociation of Gα–GTP from the Gβγ dimer, Ste4/Ste18. The signal is then transmitted by Gβγ via the Fus3 MAP kinase (MAPK) cascade that ultimately induces cell-cycle arrest, changes in gene expression and morphogenesis (Dohlman and Thorner, 2001; Bardwell, 2005). Cells polarize their growth and form mating projections, becoming pear-shaped ‘shmoos’, in response to both isotropic pheromone and pheromone gradients. In mating mixtures, cells find and contact the closest potential mating partner by determining the direction of the most potent pheromone source and growing toward it (Jackson and Hartwell, 1990).

Polarized growth in S. cerevisiae, like that in higher eukaryotes, requires the marking of a growth site at the cell cortex and alignment of the actin cytoskeleton towards it. Cargo bound for the polarized structures is transported along the actin cables (Pruyne and Bretscher, 2000a; Pruyne and Bretscher, 2000b). Actin polarization depends on Cdc42, which is thought to activate a formin protein, Bni1, that nucleates and tethers actin cables to the polarization site (Evangelista et al., 2002; Sagot et al., 2002). Bni1 is part of the ‘polarisome’ complex, together with Spa2, Bud6 and Pea2 (Pruyne and Bretscher, 2000a). Cdc42 is activated by the exchange factor Cdc24, which itself undergoes localized activation during both vegetative budding and mating. In vegetative cells, cortical tags promote localized activation of the monomeric GTPase, Bud1/Rsr1 (hereafter Bud1), which binds directly to Cdc24 (Park et al., 1997) and Cdc42 (Kang et al., 2010). In cells exposed to pheromone, Gβγ interacts with the Far1–Cdc24 complex (Butty et al., 1998; Nern and Arkowitz, 1998; Nern and Arkowitz, 1999). Interaction of Gβγ with Far1 is thought to activate Cdc24 (Wiget et al., 2004), leading to localized GTP-loading of Cdc42, recruitment of Bni1, nucleation of actin cables, and polarized growth, to form the mating projection. The Gβγ–Far1–Cdc24 complex is presumed to assemble in the region of the cell surface that experiences the highest concentration of pheromone and to mark this area for growth in preference to the predetermined bud site (Nern and Arkowitz, 2000). In this way, the cell is thought to orient its growth toward the source of pheromone, although the feedback loops that amplify the directional signal have not been well characterized. When cells are unable to sense a gradient of pheromone, on the other hand, they form a mating projection at the Bud1-marked site that would have been used for the next bud (Dorer et al., 1995; Nern and Arkowitz, 1999). This is called the default mating projection site.

Yeast exhibit a remarkable ability to interpret chemical gradients. It has been estimated that a 1% difference in receptor occupancy across the 5 µm length of a yeast cell in a pheromone gradient is sufficient to elicit robust orientation toward the pheromone source (Segall, 1993), and microfluidic studies suggest an even greater acuity (Moore et al., 2008). How is this very slight asymmetry in activated receptor and G-protein amplified internally to establish a unique shmoo site, and how is this positional information continually communicated to the polarisome to ensure properly oriented growth over time?

Since it was first reported that pheromone induces the phosphorylation of Gβ on multiple sites (Cole and Reed, 1991), the function of this modification has been elusive. A double point mutant form of Gβ that cannot be phosphorylated was found to have no measurable effect on signal transmission, adaptation, or diploid formation (,Li et al., 1998). More recently, we discovered that the activated forms of the mating-specific Gα protein and the Fus3 MAPK interact directly. A mutant form of Gα that is severely defective in binding Fus3, Gα^DSD (Gα-Docking Site Disrupted; previously Gα^{K21E R22E}), confers a defect in partner discrimination, indicating a problem in directional sensing and/or directed growth (Metodiev et al., 2002; Strickfaden and Pryciak, 2008; Yu et al., 2008). Gα^DSD also results in hypo-phosphorylation and reduced levels of Gβ, as does fus3Δ. These observations raised the possibility that the phosphorylation of Gβ plays a role in chemotropism. Here we show that Gβ phosphorylation is critical for this process.

The findings that Gα^DSD confers partial defects in partner discrimination, mating efficiency, shmooing and actin polarization (Metodiev et al., 2002; Matheos et al., 2004), suggest that the Gα–Fus3 interaction plays a role in chemotropism. Moreover, Gα itself has been implicated in chemotropism (Strickfaden and Pryciak, 2008; Yu et al., 2008). We therefore asked whether Gα^DSD bud1Δ cells can shmoo. Cells can form mating projections if either the default or chemotropic shmoo pathways are inactivated, but are unable to polarize their growth in response to pheromone if neither pathway is functional (Nern and Arkowitz, 2000). Thus, components required for chemotropic growth can be uncovered by examining mutations in a bud1Δ background. For example, cdc24-m1 bud1Δ and far1-H7 bud1Δ cells cannot shmoo (Nern and Arkowitz, 2000) because Cdc24 and Far1 are essential chemotropic elements (Valtz et al., 1995; Nern and Arkowitz, 1998). Similarly, Gα^DSD bud1Δ cells were completely unable to form mating projections (Fig. 1A). Whereas the Gα bud1Δ cells and Gα^DSD BUD1 cells (not shown) formed normally shaped shmoos, the Gα^DSD bud1Δ cells either enlarged uniformly, or exhibited highly aberrant morphologies, consistent with the idea that the Gα–Fus3 interaction is required for chemotropic growth.

As Gβ phosphorylation and level are greatly reduced in pheromone-treated Gα^DSD and fus3Δ cells, Gβ may be a target of Gα–Fus3. Because Gα^DSD also confers defects in partner discrimination and in the genetic assay for ‘chemotropic’ shmooing described above, we wondered whether the phosphorylation of Gβ plays a role in the chemotropic response. As a first test of this idea, we created strains in which the native STE4 was replaced with ste4^T320A/S335A (,Li et al., 1998), which encodes a mutant form of Gβ that is not phosphorylated in vivo (henceforth Gβ^P− ), in BUD1 and bud1Δ backgrounds. The Gβ^P− bud1Δ, Gβ bud1Δ, Gβ^P− BUD1 and Gβ BUD1 (henceforth WT, for wild type) cells were treated with a range of pheromone concentrations and examined over 5 hours. In isotropic pheromone, Gβ^P− BUD1 cells were indistinguishable from WT cells. In contrast, Gβ^P− bud1Δ cells exhibited a variety of shmoo abnormalities at all doses and time points. Representative images are shown in Fig. 1B.

A high proportion of the Gβ^P− bud1Δ shmoos were shorter and broader than those formed by the Gβ bud1Δ control cells, as if their growth was less well focused. Their mean lengths were 0.89 and 0.84 that of the Gβ bud1Δ cells 4 and 5 hours after treatment, respectively (P<0.0001; n = 300 for each time point), and most were abnormally shaped. Instead of narrowing smoothly to a pointed tip, Gβ^P− bud1Δ mating projections often extended almost straight out from the cell body and terminated in a bulbous curve. Although aberrant shmoo morphologies were also seen in the Gβ bud1Δ culture, likely due to the absence of Bud1, their occurrence was greatly increased by Gβ^P− (80.7% as compared to 16.7%; n≥260; P<0.0001).

Wild-type cells treated with high pheromone concentrations form successive mating projections with regular periodicity such that the growth of the first projection ends as the second begins (Bidlingmaier and Snyder, 2004). In isotropic pheromone, the second growth site is usually established far from the first one, so that the two projections ultimately form a wide angle (distal projections). Another striking phenotype of the Gβ^P− bud1Δ cells was their tendency to form second and third mating projections earlier than the control cells and in abnormal positions (Fig. 1C). In the control strain, we observed a small fraction of cells with two or more projections adjacent to one another (proximal projections). However, the occurrence of such cells was dramatically increased in the Gβ^P− bud1Δ strain, with the ratio of proximal to distal shmoos more than 10-fold greater in Gβ^P− as compared to control cells. Similar results were observed at all pheromone concentrations and incubation times. Interestingly, the Spa2–GFP marker of polarized growth simultaneously localized to more than one shmoo tip in a high proportion of Gβ^P− bud1Δ cells (Fig. 1D), suggesting that Gβ phosphorylation is involved in switching growth from one site to another.

Although Gβ^P− bud1Δ cells were not completely incapable of polarized growth in isotropic pheromone, as are Gα^DSD bud1Δ and cdc24-m1 bud1Δ cells, a significant fraction were unable to sustain unidirectional growth long enough to form a projection. Such cells were usually quite small, and formed multiple protrusions unlike true mating projections. In Fig. 1C, these cells are counted in the ‘multiple proximal’ category; we refer to them as small multiple protrusion (SMP) cells. [The SMP percentages (mean ± s.e.m.) were 16.2±1.4 and 4.0±0.5 for the Gβ^P− bud1Δ and Gβ bud1Δ cultures, respectively; n≥200; P<0.0001.] Together, the results presented in Fig. 1B–D suggest that, in cells unable to shmoo via the Bud1-dependent default pathway, Gβ phosphorylation is critical for stabilization of the axis of polarity (hence for focused growth of the mating projection) and for proper growth-site switching.

Nern and Arkowitz (Nern and Arkowitz, 2000) found that double mutants lacking both the default and chemotropic pathways are unable to shmoo because they cannot stabilize their axis of polarity long enough to polarize their growth, not because they cannot establish polarity. The recent demonstration that the Bem1–GFP polarity marker wanders rapidly around the membrane of pheromone-treated cdc24-m1 bud1Δ cells (Dyer et al., 2013) supports this conclusion. Could the blunt shmoo and SMP phenotypes be the result of a less severe wandering-axis phenotype? To test this, we mixed MATa Gβ^P− bud1Δ and MATa Gβ bud1Δ cells expressing Spa2–GFP with WT MATα cells, and assayed reporter localization over time (Fig. 2A). In the control cells, Spa2–GFP localized tightly to the tip of the growing mating projection, moving steadily outward along the axis of polarity. In Gβ^P− bud1Δ cells, Spa2–GFP moved more than twice as far, even though the mutant shmoos elongated at the same rate as the control cells. The increased mobility of Spa2–GFP correlated with an increased tendency to change direction (Table 1). Multiple Spa2–GFP spots were also commonly observed in the mutant, but rarely in control cells. Thus, the axis of polarity exhibited confined wandering within the broad mating projections formed by Gβ^P– bud1Δ cells during mating.

Essential regulators of actin cable polymerization (e.g. Cdc24–Cdc42, Bem1, Spa2) cluster tightly in a patch at the tips of shmooing cells (Pruyne and Bretscher, 2000a; Arkowitz, 2009). In pheromone-treated WT cells, this ‘polarity patch’ wanders along the cell cortex, becoming restricted in its movement around the incipient growth site (Dyer et al., 2013). Polarity patch movement is thought to be retarded by free Gβγ, whose concentration is ultimately determined by the local density of activated receptor. The role of the receptor and Gβγ in stabilizing the polarity patch can be bypassed by a chimeric protein consisting of Bem1 and the transmembrane domain of Snc2. Expression of Bem1–GFP–Snc2 dramatically slowed movement of the polarity patch, focusing the growth of WT cells and restoring the ability of cdc24-m1 bud1Δ cells to shmoo (Dyer et al., 2013). If the shmoo morphology and maintenance defects displayed by Gβ^P− bud1Δ cells are also due to a wandering axis of polarity stemming from a weakened link between the receptor and polarity patch, we would expect these phenotypes to be suppressed by Bem1–GFP–Snc2 as well. This proved to be the case: The shmoos formed by Gβ bud1Δ and Gβ^P− bud1Δ cells transformed with the Bem1–GFP–Snc2 construct were indistinguishable (Fig. 2B).

Pheromone triggers polarization of the receptor and G-protein to the mating projection (Suchkov et al., 2010) – the same area where the growth site wanders in Gβ^P− bud1Δ cells. This confined wandering might underlie the formation of proximal projections. To determine whether receptor polarity influences the tendency of Gβ^P− bud1Δ cells to form proximal mating projections, we constructed Gβ^P− bud1Δ and Gβ bud1Δ strains expressing the Ste2^7XR/GPAAD mutant form of the receptor. Ste2^7XR/GPAAD cannot be internalized (Ballon et al., 2006) and therefore does not polarize in response to pheromone (Suchkov et al., 2010). Although cells expressing Ste2^7XR/GPAAD also formed blunt shmoos, preventing receptor polarization suppressed the formation of proximal projections by Gβ^P− bud1Δ cells (Fig. 2C), consistent with our hypothesis.

Together, the data shown in Fig. 2 indicate that, although the axis of polarity wanders in Gβ^P− bud1Δ cells, the movement of the growth site is limited to the region of highest receptor density. This implies that Gβ phosphorylation helps link the position of the polarity proteins to that of the receptor.

The phenotypes we observe in Gβ^P− bud1Δ cells could result from improper localization of the mutant Gβ. To test this, we tagged the N-termini of WT Gβ and Gβ^P− with GFP in situ. In isotropic pheromone, GFP–Gβ^P− polarized to the incipient shmoo site later and less stably than did GFP–Gβ. Similar but more pronounced defects were observed in mating mixtures (Table 2; Fig. 3). Notably, a substantial fraction of mating GFP–Gβ cells switched from one polarization site to another, presumably to align with the strongest ambient gradient. In contrast, mating GFP–Gβ^P− cells were defective in site switching, as well as in consolidating Gβ in a single region of the membrane. These data suggest that Gβ phosphorylation plays an important role in the pheromone-induced redistribution of Gβ, particularly in gradient-stimulated cells.

The inability to phosphorylate Gβ could also impact the cell by affecting the strength of Gβ interactions. For example, Ste5 preferentially binds the phosphorylated form of Gβ in yeast (Feng et al., 1998), and phosphorylation affects the activity of a mammalian Gβ (Chakrabarti and Gintzler, 2003). Given the phenotypes conferred by Gβ^P− , the interaction of Gβ with the Far1–Cdc24 chemotropic complex is a good candidate to be regulated by Gβ phosphorylation. To test this, we used a well-established genetic assay (Whiteway et al., 1990; Cole et al., 1990; Bar et al., 2003; Draper et al., 2009) to ask whether Far1 can discriminate between Gβ and Gβ^P−. The assay is based on the observation that by triggering the pheromone-responsive MAPK pathway, Gβ overexpression blocks cell-cycle progression. Co-overexpression of a protein that binds Gβ can prevent pathway induction and thereby rescue the vegetative growth of cells expressing excess Gβ. At the level of transcriptional induction, cell-cycle arrest and shmooing, Gβ and Gβ^P− appear to be equally potent (,Li et al., 1998). Nevertheless, Far1 overexpression specifically rescued the cell growth defect due to the overexpression of Gβ^P−, and not that of Gβ (supplementary material Fig. S1), suggesting that Far1 has a greater affinity for the unphosphorylated form of Gβ.

As a complementary way of assaying the effect of Gβ phosphorylation on Gβγ–Far1 interaction, we performed pull-down assays. Cells overexpressing His-tagged Gγ and either Gβ or Gβ^P− were treated with pheromone to induce phosphorylation of WT Gβ before purifying Gβγ and Gβ^P− γ on nickel beads. Because Far1 is a low-level protein (Ghaemmaghami et al., 2003), we used mass spectrometric analysis rather than western blotting to measure the amount of co-purifying Far1. Consistent with the results of the genetic assay, Far1 was detectable on the Gβ^P− γ but not on the Gβγ beads (supplementary material Fig. S2).

To further investigate the possibility that Gβ phosphorylation affects the dynamic interactions of the chemotropic complex, we used fluorescence photobleaching approaches. First, we compared recovery times of the Spa2–GFP signal after photobleaching the shmoo tips of Gβ^P− bud1Δ and Gβ bud1Δ cells. As a component of the polarisome, the positional stability of Spa2 is likely linked to that of the chemotropic complex. In FRAP analysis, the recovery time is a function of the protein's dynamics: The slower the exchange into and out of a complex, the longer the signal will take to recover. As shown in Fig. 4A, the signal recovered in bleached Gβ bud1Δ shmoo tips faster and to a greater degree than in bleached Gβ^P− bud1Δ shmoo tips. This suggests that the duration of Spa2 interactions at the shmoo tip is increased when Gβ cannot be phosphorylated.

To assess the effect of Gβ^P− on the stability of the chemotropic complex itself, we performed iFRAP analysis of GFP–Cdc42 at the shmoo tips of Gβ^P− bud1Δ and Gβ bud1Δ cells. In this technique, the area that serves as a source of molecules bound for the complex of interest is photobleached, and the rate at which the signal intensity decreases at the site of complex formation is measured. The rate of signal loss is inversely related to the stability of the complex. In our experiment, the rate of GFP–Cdc42 signal loss at the shmoo tip was determined after bleaching the cell body. As shown in Fig. 4C, the Cdc42–GFP signal decreased faster in Gβ bud1Δ than in the Gβ^P− bud1Δ shmoo tips. Consistent with the Spa2–GFP FRAP results, this suggests that the duration of Cdc42 interactions at the shmoo tip are increased when Gβ cannot be phosphorylated. Together, these data suggest that unphosphorylated Gβ binds more tightly than phosphorylated Gβ to components of the chemotropic complex and/or polarisome. Thus, Gβ phosphorylation may negatively affect the stability of the chemotropic complex and/or polarisome.

In bud1Δ cells, mating projection formation depends on chemotropic components (Nern and Arkowitz, 2000). Therefore, the observation that Gβ^P− bud1Δ cells shmoo aberrantly in isotropic pheromone implicates Gβ phosphorylation in chemotropism. To test this, we set up dilute bilateral mating mixtures in which the cells best able to form stable, chemotropic mating projections were most likely to find and fuse with a partner, and analyzed images of newly formed zygotes (Fig. 5A). How precisely a cell orients towards a potential partner can be inferred by measuring the angle created when two cells fuse. When orientation is optimal, the two cells of a mating pair grow directly toward one another, and consequently, their angle of fusion is ∼0°. In contrast, large fusion angles are indicative of poor orientation. In the zygote formation assay, Gβ^P− cells exhibited a clear orientation defect (Fig. 5C).

After initial orientation toward a pheromone source, yeast can alter their growth in response to a directional change in the gradient (Segall, 1993; Moore et al., 2008) – either by initiating a second projection, or by bending the existing projection toward the new source (Moore et al., 2008). These phenomena are collectively referred to as reorientation. The mechanisms underlying reorientation are likely to overlap and/or interact with those involved in orientation and maintenance. As our data implicate Gβ phosphorylation in initial orientation and stable chemotropic growth, we asked whether it is involved in reorientation as well. WT and Gβ^P− bilateral mating mixtures were shaken vigorously to induce non-directional (default) shmooing while preventing cell fusion, then allowed to produce and respond to pheromone gradients on solid medium. Gβ^P− conferred a reorientation defect (Fig. 5B,C). Whereas pre-stimulated WT cells often bent dramatically to contact a mating partner, Gβ^P− cells formed zygotes that were predominantly the result of ‘collision matings’, i.e. two shmoos appeared to have simply grown into one another without having significantly changed their direction of growth.

One explanation for the results of the orientation and reorientation assays is that Gβ^P− confers a defect in directional sensing and/or chemotropic shmooing. However, it is also possible that Gβ^P− cells signal their positions less effectively than WT cells because their pheromone secretion is less focused. We distinguished these possibilities by comparing the ability of Gβ and Gβ^P− cells to grow toward a pheromone source in vitro, using a microfluidic device (Brett et al., 2012). As previously reported, the Gβ control cells formed normal mating projections in the artificial gradient (Fig. 6A) and oriented toward its source with an accuracy similar to that observed in other published microfluidic experiments (Brett et al., 2012). Surprisingly, the Gβ^P− cells were unable to sustain growth in a single direction. Rather, they formed multiple small protrusions, often without growing much in overall size – very much like the SMP cells in Gβ^P bud1Δ cultures treated with isotropic pheromone (Fig. 1B). The first protrusions formed by the gradient-stimulated Gβ^P− cells were positioned randomly. These data strongly support the idea that Gβ phosphorylation plays a role in the positioning and maintenance of the chemotropic growth site.

The inability of Gβ^P− cells to maintain a single axis of polarity in an artificial gradient, combined with their ability to form normal mating projections in isotropic pheromone, provides the first example, to our knowledge, of a gradient-specific shmoo phenotype. We therefore wished to confirm this observation using natural gradients. To look for gradient-dependent destabilization of polarized growth under physiological conditions, we compared WT and Gβ^P− shmoos formed in liquid culture (isotropic treatment) with those formed in bilateral crosses. Although the Gβ^P− cells in mating mixtures were not as severely compromised in chemotropic shmooing as the Gβ^P− cells in the microfluidic device, they exhibited a number of obvious shmoo defects (Fig. 6B,C). In fact, the response of Gβ^P− BUD1 cells exposed to natural pheromone gradients was the same as that of Gβ^P− bud1Δ cells exposed to isotropic pheromone: Gβ^P− mating mixtures accumulated fewer shmoos, shorter shmoos (their mean length was 0.852 of the control cells after 5 hours of mating; n≥105; P<0.0001), and a significantly higher proportion of aberrant shmoos. In contrast, the Gβ^P− and WT shmoos induced by isotropic treatment were indistinguishable at all concentrations and time points (Fig. 1B). These data suggest that the phosphorylation of Gβ plays a critical role in chemotropic, but not default, shmooing.

The rapid phosphorylation of Gβ in yeast cells responding to mating pheromone was observed over 20 years ago (Cole and Reed, 1991), and yet its purpose has remained enigmatic. Two early studies failed to elucidate its role (Cole and Reed, 1991; ,Li et al., 1998). We were impelled to revisit this issue by the discovery that the activated forms of Gα and Fus3 interact, and that a double mutation that uncouples them, Gα^DSD, conferred defects in mating and partner discrimination (Metodiev et al., 2002). Indeed, we show here that Gα^DSD bud1Δ cells cannot form mating projections, which identifies Gα^DSD as a chemotropic-defective allele (Fig. 1A). This suggests that the Gα–Fus3 interaction, like the Cdc24–Far1 interaction, is critical for chemotropism. Because Gα^DSD also results in reduced Gβ phosphorylation and protein levels in stimulated cells, we hypothesized that Gβ phosphorylation plays a role in chemotropism. Our results demonstrate that this is the case. Specifically, the modification of Gβ appears to promote communication between activated receptors and downstream components essential for polarized growth. Cells unable to phosphorylate Gβ exhibit defects in pheromone-induced morphogenesis, and both orientation and reorientation towards mating partners. We propose that Gβ phosphorylation provides a means to amplify the spatial signal and that the Gβ phosphorylation/dephosphorylation cycle is an updating mechanism that conveys the status of the receptor to the chemotropic complex.

Although not completely unable to form mating projections, Gβ^P− bud1Δ cells exhibited a variety of shmoo abnormalities (Fig. 1B,C), including short and blunt shmoos, cells that formed multiple projections proximal to the first, and very small cells with multiple protrusions (SMP cells). Time-lapse imaging of Spa2–GFP in these cells indicated that the axis of polarity wandered, albeit in a confined area (Fig. 2A). This broader movement of Spa2 correlated with the generation of proximal projections in the region expected to be highest in receptor density. It is therefore of significance that preventing receptor polarization in Gβ^P− bud1Δ cells dramatically suppressed the proximal-projection phenotype, and that expression of Bem1–GFP–Snc2, which slows polarity site wandering by a mechanism independent of the receptor and G-protein, restored their ability to shmoo normally (Fig. 2B,C). These observations suggest that although the axis of polarity is destabilized in pheromone-treated Gβ^P− bud1Δ cells, its wandering is limited by the high concentration of receptor and Gβγ in mating projections. This confined wandering likely underlies the formation of short, blunt shmoos, and in combination with the apparent defect in regulating growth-site switching (Fig. 1D; Table 2), may result in the emergence of multiple projections proximal to the first. The occurrence of ≥2 Spa2–GFP spots in Gβ^P− bud1Δ shmoos also correlates with the eventual formation of proximal projections (Table 1).

What causes confined wandering? Yeast chemotropism requires the translation of an extracellular pheromone gradient into a gradient of activated receptors across the cell surface, and the communication of this spatial information to the actin cytoskeleton. The receptor does not communicate with actin directly, but rather via its Gβγ, which recruits Far1–Cdc24. Along with the polarisome, Cdc42 and Bem1, the Gβγ–Far1–Cdc24 chemotropic complex ultimately nucleates actin cables at the incipient mating projection site. In cdc24-m1 and far1-H7 cells, the link between the receptor and chemotropic complex is completely severed. Polarization sites are established in such cells (Nern and Arkowitz, 2000), but they lack the constraint provided by interaction with receptor-activated Gβγ (Dyer et al., 2013). Hence, their position is not sufficiently stable to sustain polarized growth in one direction. Gβ^P− bud1Δ cells exhibit a similar, although less severe, phenotype. Instead of global wandering and uniform growth, wandering is confined to the area of most concentrated receptor. This implies that the link between the receptor and chemotropic complex is weakened, but not broken.

We explored the effect of Gβ phosphorylation on the stability of the chemotropic complex and polarisome. A genetic assay suggested that phosphorylation of Gβ reduces its affinity for Far1 (supplementary material Fig. S1), which was supported by the results of a Gβγ pull-down experiment (supplementary material Fig. S2). Photobleaching approaches indicated that both Spa2 and Cdc42 are more mobile in WT cells than in Gβ^P− cells, consistent with shorter-lived associations at shmoo tips (Fig. 4B,C). Thus, the phosphorylation of Gβ appears to decrease its affinity for the chemotropic complex and destabilize the association of Spa2 with the polarisome. This suggests an explanation for the confined wandering observed in Gβ^P− bud1Δ cells.

Immediately after exposure to pheromone, the activated receptor and G-protein are distributed almost uniformly on the plasma membrane, as are the first chemotropic and polarisome complexes to be assembled. As the receptor polarizes, however, so do Gα and Gβγ, leading to the eventual concentration of the signaling proteins at the incipient shmoo site. The high density of free Gβγ in this area biases the recruitment of Far1–Cdc24, and thus the localized formation of the chemotropic complex. Moreover, as chemotropic complexes dissociate, they are much more likely to reassemble where free Gβγ is concentrated. The half-life of the chemotropic complex is critical. If it is too short, the cell will not be able to initiate polarized growth. If it is too long, the complex will wander out of the region of high-density receptor and G-protein before the axis of polarity is established. In WT cells, phosphorylation of Gβ decreases the half-life of the complex, thereby limiting the distance it can move before dissociating. This increases the chance that the complex components will reassemble within the active growth area, where receptor and G-protein concentration are highest. In Gβ^P− cells, on the other hand, the half-life of the complex is long enough to allow movement across the entire region of high-density receptor and G-protein. Additionally, the zone of Gβ appears to be less well consolidated in Gβ^P− cells (Table 2; Fig. 3). A key aspect in this scenario is the recycling of the G-protein – the re-association of the heterotrimer with the receptor. Gβγ cannot report the position of active receptor until it releases from the chemotropic complex and binds inactive Gα. Conversely, the chemotropic complexes can move away from the growth site, blind to local Gβγ density, as long as they remain intact. Interestingly, mutations that disrupt the interaction of Gβ with the N-terminal interface of Gα, or prevent inactivation of Gα, also implicate G-protein cycling in chemotropic shmooing (Strickfaden and Pryciak, 2008). We infer that by increasing the rate at which the position of the activated receptor is reported, the Gα-GDP/GTP and the Gβ-phosphorylation/dephosphorylation cycles strengthen the indirect link between the receptor and the chemotropic complex, and consequently influence the position of the polarity proteins. Posttranslational modification of Gβ may therefore extend the well-established paradigm that cycling is central to the function of G-proteins.

Our results suggest that Gβ phosphorylation is critical for pheromone-induced Gβ polarization (Fig. 3; Table 2). Although the GFP–Gβ^P− polarization defect could result simply from the drifting axis of polarity, a number of observations raise a more interesting possibility: Gβ phosphorylation could play a key role in the genesis of pheromone-induced intracellular signaling gradients. Given that full pheromone-induced phosphorylation of Gβ depends on Fus3 and on Gα–Fus3 interaction (Metodiev et al., 2002), and that gradients of active Fus3 emanate from the tips of shmooing cells (Maeder et al., 2007), we propose that Gα recruits Fus3 to phosphorylate Gβ at the incipient shmoo site. As it is also known that pheromone-induced polarization of Gα and Gβ requires their co-internalization with the receptor (Suchkov et al., 2010), it will be interesting to determine whether unphosphorylated Gβ is preferentially internalized along with Gα and the receptor while phosphorylated Gβ is left on the membrane. A synergistic combination of these two mechanisms – localized Gβ phosphorylation at the shmoo site and preferential internalization of unphosphorylated Gβ at the back of the cell – would be expected to rapidly generate a steep gradient of phosphorylated Gβ via what is essentially a LEGI-type mechanism. An intracellular gradient of phosphorylated Gβ underlying yeast chemotropism would provide an interesting analogy to the intracellular gradient of the phosphorylated lipid, PIP₃, found in numerous chemotactic systems (Jin et al., 2000; Servant et al., 2000; Comer and Parent, 2002).

To determine whether Gβ phosphorylation contributes to directional sensing in addition to its role in mating projection formation, we examined how Gβ^P− affects chemotropism in both natural and artificial pheromone gradients. These experiments were performed with BUD1 strains, as we wished to determine how well the chemotropic shmoo pathway would override the default shmoo pathway to orient and sustain growth up a gradient. In mating mixtures, Gβ^P− conferred defects in initial orientation (Fig. 5A) and polarized growth (Fig. 6B). Similar but more pronounced defects were observed in artificial pheromone gradients (Fig. 6A). Remarkably, the polarized growth defects were gradient-specific. Gβ^P− cells formed mating projections when treated with isotropic pheromone, but could not maintain polarized growth –even at the default site – in either artificial or natural gradients. It is noteworthy that the types and proportions of aberrant shmoos formed by Gβ^P− BUD1 cells in mating mixtures were very similar to those formed by Gβ^P− bud1Δ cells in isotropic pheromone (compare Fig. 6B,C and Fig. 1B,C). In both cases, the cells were forced to use the chemotropic pathway without being able to phosphorylate Gβ. The resulting defects reveal that the cells attempt to shmoo chemotropically, but cannot maintain a stable axis of polarity. These data strongly suggest that Gβ phosphorylation is critical for the initial positioning and maintenance of the chemotropic growth site.

In addition to mechanisms that establish and stabilize directional growth, chemotroping cells must have a means to alter their direction and track dynamic gradients in mating mixtures. To successfully fuse with the shmoo tip of a partner, a yeast cell must not only determine the direction of the strongest source of pheromone and orient its growth accordingly, it must continually reassess the position of the target cell while ignoring weaker signals. The shape of the gradient and concentration of pheromone most likely change as the two cells grow towards each other, and as zygotes form in the mating mixture. Although the mechanisms underlying reorientation are unknown, it is easy to appreciate that the establishment of a chemotropic growth site and ongoing adjustments to its position pose distinct challenges. To date, only one mutation that specifically affects reorientation has been reported: ste2^T236 (Vallier et al., 2002). It is therefore of considerable interest that Gβ^P− confers a dramatic defect in reorientation. This suggests that Gβ phosphorylation plays a critical role in gradient tracking.

In summary, we propose that the phosphorylation of Gβ contributes to chemotropism in two ways: (1) Gβ modification is used to rapidly generate a signaling gradient in which phosphorylated Gβ is concentrated at the incipient shmoo site. This may facilitate the amplification of other intracellular signaling gradients; (2) Phosphorylation of Gβ decreases its affinity for Far1 and perhaps other chemotropic components, thereby shortening the cycle time between the receptor and polarisome. More frequent updating of receptor status increases sensitivity to changes in the pheromone gradient.

Standard methods were used for microbial and molecular manipulation (Sherman et al., 1986; Ausubel et al., 1994; Guthrie and Fink, 2002). Strains and plasmids used are listed in supplementary material Tables S1,and S2, respectively. Yeast cells were cultured in rich medium (YEPD) at 30°C unless noted otherwise.

MATa bud1Δ Gβ or Gβ^P− cells expressing SPA2-GFP were incubated together on s.d. complete agar pads containing 150 nM α-factor; one strain was labeled with ConA-Alexa Fluor 594 (Molecular Probes). Six fields were imaged at 15-minute intervals with 0.5 µm z-sections acquired using a DeltaVision deconvolution system (Applied Precision) on an Olympus IX-70 microscope with an NA 1.4×60 objective. The images were deconvolved using Huygens Deconvolution Software V. 3.7 (Scientific Volume Imaging), sum projected (GFP), or average projected (DIC), and converted into 8-bit TIF files using ImageJ. The scale bar equals 5 µm in all images.

WT MATα cells were labeled with ConA-Alexa Fluor 594 prior to mixing with MATa bud1Δ Gβ or Gβ^P− cells expressing SPA2-GFP. Mating mixtures were incubated on s.d. complete agar pads. Images were acquired 25 minutes after mixing and at 15-minute intervals thereafter. Image acquisition and analysis was as described above. Tracking analysis was performed using the ImageJ Manual Tracking plugin.

MATa bud1Δ GFP–Gβ and MATa bud1Δ GFP–Gβ^P− cells were incubated together on s.d. complete agar pads containing 150 nM α-factor. The strains were distinguished by labeling one or the other with ConA-Alexa Fluor 594. Cells were first imaged 15 minutes after exposure to pheromone and at 10-minute intervals thereafter. Six fields were imaged at each time point with 15 0.3 µm z-section acquired using an ANDOR Revolution XD spinning disk laser confocal system with a motorized Olympus IX-81 microscope, a Yokogawa CSU-X1 spinning disk unit, motorized XYZ control (piezo) and an iXon897 EMCCD camera, all controlled by Andor iQ2 software. A UplanSApo NA 1.4×100 objective was used with 488-nm laser excitation. Voxel size was 133×133×280 nm. An Okolab chamber was used to maintain cells at 30°C. The images were then sum projected and converted into 8-bit TIF files using ImageJ and false colored. The intensity profile along the major axis of the cells was quantified using BudPolarity (Vernay et al., 2012). The line graphs in Fig. 3 represent the output from this analysis. In Table 2, cells were scored as exhibiting polarized Gβ if they met all of the following criteria: peak signal intensity ≥2× the baseline value over less than one half the cell surface for at least three consecutive time points. Cells were scored as exhibiting polarized growth when they were judged to be elongating predominantly along one axis at the center of the DIC z-stack. The ImageJ ellipse tool was used for this determination.

WT MATα cells were mixed 1∶1 with MATa bud1Δ GFP–Gβ or MATa bud1Δ GFP–Gβ^P− cells. Mating mixtures were incubated on s.d. complete agar pads and images were acquired every 15 minutes. Image acquisition and analysis were identical to that described for isotropic pheromone conditions. In Table 2, cells were scored as exhibiting polarized Gβ if they met all of the following criteria: peak signal intensity ≥1.5× the baseline value over less than one half the cell surface for at least three consecutive time points. Cells were scored as exhibiting polarized growth as described above.

MATa bud1Δ Gβ or Gβ^P− cells expressing SPA2-GFP were grown to mid-log phase and then exposed to 30 nM α-factor at 30°C for 1 hour. FRAP analysis was carried out on a Zeiss LSM 510 META confocal Axiovert 200M microscope (Carl Zeiss) using an NA 1.4×63 Plan-Apo objective and 488-nm LASER excitation. Images were captured every 1 second at 2–5% maximum laser intensity and 10×0.5 milliseconds photobleaching scans at 100% laser intensity were performed on a circular area of 1 µm² at the shmoo tip. Data analysis was carried out as described (Bassilana and Arkowitz, 2006).

MATa bud1Δ Gβ or Gβ^P− cells expressing GFP-CDC42 were grown to mid-log phase and exposed to 30 nM α-factor at 30°C for 1 hour. iFRAP was carried out as described for FRAP analyses except that the cell body was photobleached rather than the shmoo tip. The average intensity of the bleached or unbleached area was normalized for photobleaching during image acquisition, using the average intensity of the cell with MatLab. Loss of fluorescence intensity was then fitted to an exponential curve.

To study mating projection formation and maintenance under isotropic pheromone conditions, mid-log phase cells were exposed to α-factor at the concentrations and times indicated. DIC images for experiments carried out in liquid cultures were acquired using a Zeiss Axioskop 2 microscope fitted with an NA 1.4×63 Plan-Apo oil immersion objective and a Zeiss AxioCam digital camera, and processed with Zeiss AxioVision software. To study mating projection formation and maintenance in physiological pheromone gradients, cultures were grown to mid-log phase at 30°C in YEPD. Dilute bilateral Gβ and Gβ^P− mating mixtures containing ∼1×10⁷ MATα cells and ∼1×10⁷ MATa cells were then spread on YEPD agar. Time points were taken at 1 hour intervals for 5 hours, fixed with 3.7% formaldehyde, and 630×DIC images of shmoos were acquired using the Zeiss equipment described above. Mating projection lengths were measured using ImageJ. Percentages in each shmoo class were scored on a hemocytometer using a Leitz Laborlutz S phase-contrast microscope fitted with an NA 0.65×40 Plan objective.

To measure the ability of cells to orient their growth in physiological pheromone gradients, cultures were grown to mid-log phase. Dilute bilateral Gβ and Gβ^P− mating mixtures containing ∼5×10⁶ MATα cells and ∼5×10⁶ MATa cells were then incubated for 3 hours on filters (0.45 µm Millipore) placed on YEPD agar. Filters were washed in 1 ml sterile H₂O to harvest cells. 630×DIC images of zygotes were acquired using the Zeiss Axioskop 2 microscope and camera described above. Zygote angles were measured using ImageJ. The response of cells to artificial pheromone gradients was assayed using a microfluidic device, as described (Brett et al., 2012).

Mid-log phase liquid cultures containing ∼10⁷ MATα cells and ∼10⁷ MATa cells were shaken at 250 r.p.m. at 30°C for 3 hours to induce shmooing prior to mating. Cultures of either WT MATα and MATa cells or Gβ^P− MATα and MATa cells were then spread onto YEPD agar and incubated for 3 hours. Images of zygotes were collected using an Olympus BH-2 inverted microscope fitted with a CDPlan 40 objective and an Olympus digital camera (magnification 400×). Zygote angles were measured using ImageJ.

The P-values for the comparisons of length, angle and time measurements were determined using unpaired, two-tailed t-tests (Graphpad). The P-values for the comparisons of percentages were determined using the chi-square test (Graphpad).

The authors would like to thank Jayme Dyer and Daniel Lew for yeast strains and plasmids, for helpful discussions and for communicating results prior to publication.