This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Levin, D. E. Search for Related Content PubMed PubMed Citation Articles by Levin, D. E.

 Previous Article  |  Next Article 

Microbiology and Molecular Biology Reviews, June 2005, p. 262-291, Vol. 69, No. 2
1092-2172/05/$08.00+0     doi:10.1128/MMBR.69.2.262-291.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.

Cell Wall Integrity Signaling in Saccharomyces cerevisiae

SUMMARY

INTRODUCTION

The Yeast Cell Wall

Cell Wall as a Target for Antifungal Drug Development

CWI PATHWAY ARCHITECTURE

CWI MAP Kinase Cascade

Rho GTPases: Rho1-5 and Cdc42

Rho1: Master Regulator of CWI Signaling.

Regulators of Rho1.

Targets of Rho1.

Pkc1.

Glucan synthase.

Bni1 and Bnr1.

Skn7.

Sec3.

Cell Surface Sensors: Wsc1-3, Mid2, and Mtl1

Phosphoinositide Metabolism: Stt4-Mss4 Signaling

Nuclear Targets of Mpk1

Rlm1.

SBF (Swi4/Swi6).

Cytoplasmic Targets of Mpk1

Cch1/Mid1 Ca2+ channel.

MAP kinase phosphatases.

Mih1 tyrosine phosphatase: morphogenesis checkpoint.

ACTIVATION OF CWI SIGNALING

Cell Cycle Regulation

Heat Stress

Hypo-osmotic Shock

Pheromone-Induced Morphogenesis

Cell Wall-Stressing Agents

Actin Cytoskeleton Depolarization

Oxidative Stress

Plasma Membrane Stretch

Delocalization of Signaling Components

ALTERNATIVE Pkc1 PATHWAY BRANCHES

Cell Wall Targets of Pkc1

Oligosaccharyl transferase.

Chitin synthase 3: the chitin emergency response.

Phospholipid Biosynthesis Targets

Nuclear Functions of Pkc1

Arrest of secretion.

Mitotic recombination.

G2/M progression and the mitotic spindle.

SPB duplication.

Pkc1 does not control depolarization of the actin cytoskeleton.

INTERFACE WITH OTHER SIGNALING PATHWAYS

Pkh1/2 and Ypk1/2

Tor Regulation of the Actin Cytoskeleton

PERSPECTIVES AND FUTURE DIRECTIONS

ACKNOWLEDGMENTS

REFERENCES

Top Next References

INTRODUCTION

Top Previous Next References

 

The cell wall of the budding yeast Saccharomyces cerevisiae is also required to maintain cell shape (50, 159), which is essential for the formation of a bud and hence cell division. The cell must remodel this rigid structure to accommodate cell expansion during vegetative proliferation, mating pheromone-induced morphogenesis, and nutrient-driven filamentation. Turgor pressure is critical for cell expansion, because it provides the force to overcome molecular cohesion within the cell wall (109). Because fungal cells maintain an intracellular osmolarity that exceeds that of the extracellular environment, water tends to flow into the cell, thereby providing turgor pressure. However, this pressure is equally distributed across the cell surface. Therefore, for growth to produce cell shapes other than spheres, cell wall expansion must be focused to particular regions. Saccharomyces cerevisiae uses an internal actin cytoskeleton for this purpose (77). During periods of polarized cell growth, the wall is loosened by digestive enzymes (e.g., glucanases and chitinases) and expanded at a single point on the cell surface. Wall remodeling must be carried out in a highly regulated manner—the growth site is loosened enough to allow expansion but not so much as to risk rupture.

Yeast cells invest considerable energy toward biogenesis of the cell wall, which comprises some 20 to 30% of the cell dry weight (243, 313). The major features of the Saccharomyces cerevisiae cell wall architecture are now fairly well understood. For a recent review on its molecular organization, the reader is referred to Klis et al. (160). Briefly, the cell wall is a layered structure with an electron-transparent inner layer and an electron-dense outer layer (40, 244). The inner layer is comprised of glucan polymers and chitin (N-acetylglucosamine polymers). This layer is constructed mainly (80 to 90%) of ß1,3-glucan chains with some ß1,6-linked glucan branches. Polymers of ß1,6-glucan chains make up most of the remainder of the inner layer (8 to 18%), with chitin chains representing the smallest fraction (1 to 2%). This layer is largely responsible for the mechanical strength and elasticity of the cell wall owing primarily to the helical nature of ß1,3-glucan chains (270, 313).

The outer cell wall layer is a lattice of highly glycosylated mannoproteins, which functions to protect the glucan layer from wall-degrading enzymes (68, 69, 160, 372). It is also important for cell-cell recognition during sexual agglutination and biofilm formation (40, 186, 273). Two major classes of cell surface glycoproteins comprise the outer cell wall layer. Members of one class, called glycosylphosphatidylinositol (GPI) proteins, are directed through the secretory pathway to the extracellular face of the plasma membrane by lipid anchors at their C termini. GPI-proteins are liberated from the plasma membrane by cleavage of their anchors prior to attachment to the cell wall (164). Among the approximately 70 GPI-proteins identified in the Saccharomyces cerevisiae genome (41), it is estimated that half reside in the cell wall (313). The other major class of cell wall proteins is represented by four related polypeptides, Pir1 to Pir4 (152, 228, 330). Although the Pir proteins appear to be linked directly to the ß1,3-glucan-chitin lattice, GPI-proteins are generally linked to ß1,3-glucan indirectly through a connecting ß1,6-glucan chain (160).

The focus of this review is the regulatory pathways employed by Saccharomyces cerevisiae to maintain cell wall integrity during growth and morphogenesis and in the face of external challenges that cause cell wall stress. Although several signaling pathways contribute to the maintenance of the cell wall, the one principally responsible for orchestrating changes to the wall and responding to challenges to this structure is known as the cell wall integrity pathway, which will be abbreviated hereafter as the CWI pathway. I will also discuss recent advances in our understanding of how this pathway interfaces with other signaling pathways. In particular, several signaling pathways converge to regulate organization of the actin cytoskeleton. Some of these pathways also regulate CWI signaling, presumably to coordinate cell polarization with cell wall biogenesis. I will not discuss in depth the related topic of osmoregulation except as it relates to CWI signaling. An excellent recent review deals with adaptation to both hyper- and hypo-osmotic stress in yeast (121).

CWI PATHWAY ARCHITECTURE

Top Previous Next References

 
The CWI signaling pathway is comprised of a family of cell surface sensors coupled to a small G-protein called Rho1, which activates a set of effectors, each of which will be considered individually. Collectively, these effectors regulate a diverse set of processes including, but not limited to, ß-glucan synthesis at the site of wall remodeling, gene expression related to cell wall biogenesis, organization of the actin cytoskeleton, and secretory vesicle targeting to the growth site. A diagrammatic representation of the core elements of this pathway is presented in Fig. 1.

View larger version (32K): [in this window] [in a new window]   FIG. 1. CWI signaling pathway. Signals are initiated at the plasma membrane (PM) through the cell surface sensors Wsc1, -2, and -3, Mid2, and Mtl1. The extracellular domains of these proteins are highly O-mannosylated. Together with PI4,5P2, which recruits the Rom1/2 GEFs to the plasma membrane, the sensors stimulate nucleotide exchange on Rho1. Relative input of each sensor is indicated by the width of the arrows. Rho1 activates five effectors, including the Pkc1-MAP kinase cascade, the ß1,3-glucan synthase (GS), the Bni1 formin protein, the exocyst component Sec3, and the Skn7 transcription factor. Additional regulatory inputs from the Tus1 GEF, the inhibitory Rho1 GAPs, and the Pkh1/2 protein kinases are indicated. Pkh1/2 are activated by phytosphingosine (PHS). The MAP kinase cascade, which is comprised of Bck1, Mkk1/2, and Mpk1, is activated by Pkc1. Several MAP kinase phosphatases downregulate Mpk1. Two transcription factors, Rlm1 and the SBF complex (Swi4/Swi6), are targets of the MAP kinase.

Briefly, the MAP kinase cascade for CWI signaling is a linear pathway comprised of Pkc1 (180), an MEKK (Bck1) (55, 174), a pair of redundant MEKs (Mkk1/2) (137), and a MAP kinase (Mpk1/Slt2) (173, 204). A combination of genetic and biochemical studies have established that Pkc1 activates Bck1, which activates Mkk1/2, which in turn activate Mpk1. What is the purpose of such an arrangement? Studies by Ferrell and colleagues have demonstrated that MAP kinase cascades serve both to amplify a small signal initiated at the cell surface and to convert a graded input to a highly sensitive, switch-like response (84, 127). A cascade will not be triggered by signals of low magnitude or duration (noise) but will respond rapidly and fully to stimuli that reach their threshold for activation.

Pkc1 phosphorylates Bck1 in vitro at several sites (Ser939, Thr1119, and Ser1134) in a hinge region between its putative regulatory domain and its catalytic domain (178). Significantly, activating mutations in BCK1 also cluster in this region of the protein (174). Of particular importance is the Thr1119 phosphorylation site, which is mutated to proline in the constitutive BCK1-19 allele (174). Mutation of Thr1119 to Ala, Cys, or Tyr also results in constitutive signaling (D. E. Levin and A. K. Sobering, unpublished), suggesting that disruption of an interaction involving Thr1119 (either by phosphorylation or mutation) is the key to activation of this MEKK. Bck1 is presumed to phosphorylate and activate Mkk1/2 based on genetic epistasis studies, two-hybrid interaction, and its requirement for activation of Mpk1 (137, 148, 254). Mkk1/2 phosphorylate Mpk1 on neighboring tyrosyl and threonyl residues in a T-X-Y motif that is diagnostic for MAP kinases. This MAP kinase is conserved in Candida albicans, and is required for maintenance of cell wall integrity in that species (234).

Loss of function of any protein kinase below Pkc1 (or both Mkk1 and Mkk2) results in cell lysis at elevated growth temperature. The growth defects of these mutants are osmoremedial (e.g., with 1 M sorbitol), consistent with a primary defect in cell wall biogenesis. Loss of PKC1 results in osmoremedial cell lysis at all growth temperatures (177, 253), prompting the suggestion that Pkc1 regulates additional targets that are separate from the MAP kinase cascade (174). Secondary Pkc1 targets are discussed in the section on alternative Pkc1 pathway branches. Other phenotypes associated with mutants in the CWI MAP kinase cascade include sensitivity to mating pheromone and cell wall antagonists such as calcofluor white, Congo red, caffeine, and the wall lytic enzyme zymolyase (81, 205), and actin polarization defects (212).

Mpk1 resides predominantly in the nucleus under nonstress conditions but rapidly relocates to the cytoplasm in response to cell wall stress (148). A small pool of Mpk1 localizes to sites of polarized cell growth and shuttles constitutively between these sites and the nucleus (341). During pheromone-induced morphogenesis, a minor pool of Mpk1 can be detected at the shmoo tip (18, 341). Bud tip and bud neck localization of Mpk1 evidently does not require the actin cytoskeleton because it is not disrupted by treatment with the actin antagonist latrunculin A, which interferes with actin polymerization (225). However, polarized localization of Mpk1 during growth and morphogenesis does require Spa2, a component of the polarisome, which functions in actin cytoskeleton organization (197).

Mkk1 and Mkk2 are mainly cytoplasmic proteins, but, like Mpk1, they can be detected at sites of polarized growth in an Spa2-dependent manner. Moreover, Spa2 displays two-hybrid interactions with both Mpk1 and Mkk1/2 (341), leading to the suggestion that Spa2 serves as a scaffold for these protein kinases. However, in contrast to the role of the Ste5 scaffold protein in activation of the pheromone response MAP kinase cascade (79), Spa2 is not required for Mpk1 activation during vegetative growth or in response to pheromone treatment (35, 304). This finding suggests that the function of the Spa2 scaffold with regard to CWI signaling is to focus the action of the kinases to the site of polarized growth. Interestingly, Bck1 has been detected in the cytoplasm but not at polarized growth sites (341). However, Bck1 is expressed at low abundance compared with Mpk1 and Mkk1/2 (D. E. Levin, unpublished), leaving open the possibility that a small pool of polarized Bck1 has escaped detection.

Rho5 was the last family member to be discovered (283), but the absence of a phenotypic defect associated with its deletion has left its function enigmatic. A recent report suggests that it functions to downregulate the CWI pathway (295). This was based on the finding that a rho5 mutant displays elevated basal and stress-induced Mpk1 activity and increased resistance to cell wall stressors. However, if Rho5, like the other Rho family members, serves a function in cell polarity that is separate from CWI signaling, the effect of its loss of function on Mpk1 activity may be indirect.

Rho proteins are C-terminally prenylated, a modification that increases their hydrophobicity and allows their association with membranes. These modifications are essential for their proper localization and function (288). The essential GTPases, Rho1 and Cdc42, are modified by the action of the Cdc43/Ram2 geranylgeranyl transferase (135). The absence of prenylation renders Rho1 soluble and unable to activate or even interact with GS (135). For this reason, geranylgeranyl transferase 1 is regarded as a potential target for the development of antifungal drugs (154). Inspection of their C-terminal sequences suggests that Rho2 and Rho5 are likely also modified by geranylgeranyl transferase 1, whereas Rho3 and Rho4 are probably substrates for the Ram1/Ram2 farnesyl transferase.

Regulators of Rho1. Like other G-proteins, Rho1, the yeast homolog of mammalian RhoA (264), cycles between the active GTP-bound state and the inactive GDP-bound state. The Rho1 cycle is regulated both by GTPase-activating proteins (GAPs) and guanosine nucleotide exchange factors (GEFs) acting in opposition. Among the 11 Rho-GAPs identified in Saccharomyces cerevisiae, four have been shown to act on Rho1 both in vitro and in vivo—Bem2, Sac7, Bag7, and Lrg1 (49, 205, 256, 283, 292, 294, 347). Interestingly, these GAPs appear to regulate Rho in a target-specific manner. For example, Lrg1 is evidently dedicated to regulation of GS (347). Similarly, Bem2 and Sac7 are the only GAPs that regulate the Pkc1 MAP kinase pathway (205, 294). Bag7 and Sac7, which are most similar to each other rather than to any of the other GAPs, collaborate to control the actin cytoskeleton (292, 294). The apparently independent regulation of different Rho1-effector pairs by distinct GAPs indicates some compartmentalization of Rho1 functions. Different effectors may be active through the cell cycle or in response to different types of cell wall stress.

Rho1 is stimulated primarily through the action of the Rom1 and Rom2 GEFs (246). These GEFs provide a redundant function in the activation of Rho1 (and likely Rho2). Loss of ROM2 function results in temperature-sensitive growth, whereas loss of both ROM1 and ROM2 is lethal. Like Rho1, Rom2 (and probably Rom1) resides at sites of polarized growth (201). Rom1 and Rom2 have Dbl homology (DH) domains, which interact with GDP-bound Rho1 and possess the nucleotide exchange activity of these proteins (246). They also possess pleckstrin homology (PH) domains, which bind phosphatidylinositol-4,5-biphosphate (PI4,5P₂) and are responsible for proper localization of Rom1/2 to the plasma membrane (11).

An N-terminal domain of Rom2 (and presumably Rom1) that is separate from either the DH or PH domains is responsible for associating with Wsc1 and Mid2 and likely other cell surface sensors (257). A third Rho1-GEF, Tus1, was isolated recently as a dosage suppressor of the growth and cell lysis defect of a ypk1/2 mutant (291). A tus1 mutant displays an osmoremedial growth defect at high temperature, which is also suppressed by overexpression of CWI pathway components (i.e., Rom2, Rho1, or Rho2). Tus1 possesses both a DH domain that interacts with GDP-bound Rho1 and an uncharacterized PH domain (291). Study of its PH domain has thus far failed to establish a role in phospholipid binding (367). Functional distinctions between Tus1 and Rom1/2 have not yet been made, nor is it clear if Tus1 responds to the cell surface sensor proteins in the same way as its cousins. Clarification of the role of Tus1 in CWI signaling is needed.

Saccharomyces cerevisiae possesses a single Rho-GDP dissociation inhibitor (RhoGDI), designated Rdi1, which associates with both Rho1 and Cdc42 (162, 206). Mammalian RhoGDIs stimulate release of Rho GTPases from membranes (123, 238, 357) and block GDP dissociation and GTP hydrolysis by interfering with the actions of GEFs and GAPs (113). Deletion of Saccharomyces cerevisiae RDI1 results in no detectable phenotypic defects (206), but its overexpression inhibits growth and induces increased steady-state levels of Rho1 and Cdc42, consistent with a role in the inhibition of these G-proteins (162). Rdi1 was shown recently to localize to sites of polarized growth, and its overexpression caused delocalization of Cdc42 from membranes to the cytoplasm (274). Rdi1 may facilitate the relocalization of Rho1 and Cdc42 from the bud tip to the mother/bud neck, but this possibility has not been addressed directly.

Targets of Rho1. Five effectors for Rho1 have been described—the Pkc1 protein kinase, the ß1,3-glucan synthase (GS), the Bni1 and Bnr1 formin proteins, the Skn7 transcription factor, and the Sec3 exocyst component. As indicated above, evidence is accumulating to suggest that each Rho1-effector pair is regulated separately by a different complement of GAPs and perhaps different GEFs as well. Together, these effectors regulate synthesis of cell wall glucans and chitin, expression of genes important for cell wall biogenesis, polarization of the actin cytoskeleton, and perhaps exocytosis.

Pkc1. Mammalian cells possess at least 10 isoforms of protein kinase C (PKC) and two additional PKC-related kinases (215). By contrast, the Saccharomyces cerevisiae genome encodes only a single homolog of mammalian protein kinase C, designated Pkc1 (180). It was the first component of the CWI signaling pathway discovered, and although this protein kinase likely has several intracellular substrates, only its regulation of the Bck1-Mkk1/2-Mpk1 MAP kinase cascade has been well studied. Deletion of PKC1 is lethal under normal growth conditions, but the viability of a pkc1 mutant can be rescued by osmotic support (177, 180, 253). Loss of PKC1 results in a more severe growth defect than that displayed by deletion of any of the members of the MAP kinase cascade under the control of Pkc1, which prompted the suggestion that Pkc1 regulates multiple pathways (174). Pkc1 was recently identified as the target of cercosporamide, a potent natural antifungal product whose antifungal activity was widely recognized but whose mechanism of action was not understood (325).

Electron micrographic images of pkc1 cells maintained in the presence of osmotic support suggest a pleiotropic set of cell wall defects (178, 281). Both the inner, glucan-containing layer and the outer, mannoprotein layer are thinner in pkc1 mutants. These alterations are mirrored in a reduction in both ß1,3- and ß1,6-glucans of approximately 30% and a reduction in mannan of approximately 20% (281, 306). Additionally, the plasma membrane of pkc1 mutants appears to separate from the cell wall at some points (178, 253).

Pkc1 associates with and is activated by GTP-bound Rho1 (149, 240), which confers upon the protein kinase the ability to be stimulated by phosphatidylserine (PS) as a lone cofactor (149). Cofactors that activate conventional PKCs, such as diacylglycerol and Ca²⁺, do not regulate Pkc1 (9, 348). Consistent with this finding, a pkc1 mutant is complemented by human PKC-eta (239), a member of the novel subfamily of PKCs which do not respond to diacylglycerol or Ca²⁺. PKC isoforms from other fungal species studied to date share with Pkc1 the requirements for Rho and phosphatidylserine (176, 325).

Pkc1 is larger than any of the mammalian PKCs owing to an extended regulatory domain that possesses all of the subdomains that are differentially distributed among various PKC and PKC-related kinase isoforms. It possesses two homologous region (HR1) domains (HR1A and B) at its N terminus (215), which are found in proteins that bind to RhoA, including the PKC-related kinases (305). Indeed, HR1 domains have been shown to be sites of RhoA interaction (88, 305), and the HR1A domain of Pkc1 contributes to the interaction of this kinase with Rho1 (297).

Pkc1 possesses a cysteine-rich domain, also known as a C1 domain, which is defined by a pair of zinc finger motifs. The C1 domain in the conventional mammalian PKCs is the site of diacylglycerol and phorbol ester binding (146, 326). However, despite the apparent importance of these putative zinc fingers to the function of Pkc1 (140), biochemical evidence indicates that this enzyme is not responsive to either diacylglycerol or phorbol ester (9, 348), even in the presence of GTP-bound Rho1 (149). Interestingly, the Pkc1 C1 domain lacks several residues that are essential for diacylglycerol/phorbol ester binding (340). Rather, the C1 domain of Pkc1 appears to be a second site for interaction with Rho1 (240, 296).

The C2 domain, also found in the conventional PKCs, is responsible for binding phospholipids in a Ca²⁺-dependent manner (24). This domain is characterized by a set of five conserved aspartate residues that coordinate two Ca²⁺ ions (302). The Ca²⁺-independent PKC isoforms possess C2-like domains that lack one or more of the conserved aspartate residues. Imperfect C2 domains may be responsible for the Ca²⁺-independent phospholipid activation displayed by these isoforms, but this has not been demonstrated directly. Pkc1 possesses two imperfect C2 domains that flank the C1 domain (180, 215). Thus, activation of Pkc1 by phosphatidylserine/Rho1 likely involves the C2 domain (for phosphatidylserine binding) and the HR1 domain and the C1 domain (for Rho1 binding). Some C2 and C2-like domains form specific protein-protein interactions. For example, receptors for activated PKCs bind to the C2 domains of their cognate protein kinases and seem to be important for targeting them to specific membrane compartments (219). In this regard, it is noteworthy that a two-hybrid fusion containing the C1 and C2 domains of Pkc1 interacts with the endoplasmic reticulum luminal domains of several subunits of the oligosaccharide transferase complex (255). This raises the intriguing possibility that some Pkc1 may reside within the lumen of the endoplasmic reticulum to regulate protein glycosylation.

Finally, all PKCs possess a pseudosubstrate site, typically positioned immediately N-terminal to the C1 domain (236). A pseudosubstrate site resembles a PKC phosphorylation site except that it has an alanine at the position that would be the phosphorylation target serine or threonine. Under conditions in which the PKC is not active, the pseudosubstrate site inhibits protein kinase activity through an intramolecular interaction with the active site. Mutational incapacitation of the pseudosubstrate site results in cofactor-independent protein kinase activity. Pkc1 possesses a pseudosubstrate site which, when mutated, yields a constitutive form (348) that can suppress the growth defects of conditional rho1 mutants (240).

An intracellular localization study of Pkc1 revealed that it resides at sites of polarized cell growth (8). Specifically, early in the cell cycle, Pkc1 was detected at the prebud site and at bud tips, a pattern that is very similar to that of Rho1 (265, 359). Later in the cell cycle, it becomes delocalized and finally relocalized at the mother-bud neck. The neck localization of Pkc1 requires an intact septin ring (64). A recent molecular dissection of Pkc1 suggested that each domain was responsible for localizing a pool of Pkc1 to various subcellular sites (64). For example, the HR1 domains are responsible for targeting Pkc1 to the bud tip and neck. This is consistent with the role these domains play in the association of Pkc1 with Rho1. The C1 domain in isolation localized to the cell periphery but in an unfocused manner, perhaps reflecting association of this domain with membrane lipids.

Surprisingly, removal of the HR1 domains resulted in localization of Pkc1 to the mitotic spindle, evidently directed by the N-terminal C2-like domain. Additionally, the C-terminal C2-like domain of Pkc1 (referred to by these authors as the interdomain) in isolation localized specifically to the nucleus, owing to a pair of nuclear import signals present within this region. A nuclear export signal residing within the HR1A domain is apparently responsible for maintaining the pool of nuclear Pkc1 at a low level. The existence of a nuclear pool of Pkc1 may ultimately help to explain some of the functions of this protein that are not connected to cell wall integrity (see below). Interestingly, there are several examples of metazoan PKCs colocalizing with microtubules (78, 175, 231, 345). Perhaps yeast Pkc1 is a better model for understanding some functions of higher eukaryotic PKCs than was previously believed.

Glucan synthase. Vegetative proliferation requires remodeling of the cell wall to accommodate its expansion at the growth site. The main structural components responsible for the rigidity of the yeast cell wall are ß1,3-linked glucan polymers with some branches through ß1,6 linkages (160). The biochemistry of the enzyme complex that catalyzes the synthesis of ß1,3-glucan chains (ß1,3-glucan synthase [GS]) has been studied extensively, and three components of this complex have been identified. The echinocandin antifungal agents, which interfere with the production of ß1,3-glucans and are presumed to target the GS, constitute the leading class of experimental drugs directed at treating life-threatening fungal infections (352). A pair of closely related genes, FKS1 and FKS2 (for FK506-sensitive), encode alternative catalytic subunits of the GS (75, 136, 211, 268). Fks1 and Fks2 are large, multispanning integral membrane proteins, either one of which is sufficient for GS activity and cell viability. Unlike loss of Pkc1, however, loss of Fks1/2 is not suppressed by increased osmotic support. This is presumably because cell wall biosynthesis is completely shut down in an fks1 fks2 mutant. A third gene encoding a homolog of Fks1/2, called Fks3, has not been characterized. Rho1 is an essential regulatory subunit of the complex, serving to stimulate GS activity in a GTP-dependent manner (76, 210, 265). Both Rho1 and Fks1 localize to the plasma membrane at sites of cell wall remodeling—the bud tip during bud growth, and the bud neck during cytokinesis (265, 359). A more detailed localization study revealed that GS colocalizes with cortical actin patches and moves on the cell surface in a manner dependent on actin patch mobility (338). The GS is thought to extrude glucan chains from the plasma membrane for incorporation into the wall.

An elegant intragenic complementation analysis of a dozen conditional rho1 alleles revealed that two of its essential functions could be separated (287). Mutants in one group were defective in GS activity, and mutants in the other group were defective in activating Pkc1. Mutants in one group could complement mutants in the other group. Accordingly, mutants specifically deficient in Pkc1 signaling display cell lysis defects at the restrictive temperature, whereas mutants deficient in GS activity arrested growth without cell lysis. Although some of the rho1 mutations disrupted its physical interaction with the effector of diminished function, others did not. In the latter case, an inability of the mutant Rho1 to induce a conformational change in the effector is suggested.

The FKS1 and FKS2 genes differ primarily in the manner in which their expression is controlled. Under optimal growth conditions, FKS1 is the predominantly expressed gene, and its mRNA levels fluctuate periodically through the cell cycle, peaking in late G₁ (130, 287). Cell cycle-regulated expression of FKS1 is controlled by the SBF and MBF transcription factors but primarily by SBF (136, 211, 268, 318), which is comprised of Swi4 and Swi6 (7). Expression of FKS1 is also weakly regulated by CWI signaling (130, 144) through the Mpk1-activated transcription factor Rlm1 (144).

Expression of FKS2 is low under optimal growth conditions, but its expression is induced in response to treatment with mating pheromone, elevated growth temperature, high extracellular [Ca²⁺], growth on poor carbon sources, entry into stationary phase, or in the absence of FKS1 function (211, 371). The pathway for induction of FKS2 expression by pheromone, CaCl₂, or loss of FKS1 requires the Ca²⁺/calmodulin-dependent protein phosphatase calcineurin (96, 211), the target of FK506 action (89, 187). Because FKS1 and FKS2 provide a redundant but essential function, regulation of FKS2 expression by calcineurin explains the sensitivity of fks1 mutants to FK506 and their synthetic lethality with calcineurin mutants (96).

Overexpression of Mkk1 or the Mpk1-activated transcription factor Rlm1 suppresses the synthetic lethality of an fks1 cnb1 mutant (371). Moreover, expression of a constitutive form of calcineurin suppresses the growth defects of pkc1 and mpk1 mutants (96). This connection to CWI signaling was explained in part by the finding that calcineurin and the CWI pathway function in parallel to regulate FKS2 expression, through separable promoter elements, under conditions of cell wall stress (371) (Fig. 2). In response to elevated growth temperature, the immediate transcriptional induction of FKS2 is mediated by the calcineurin-activated transcription factor Crz1/Tcn1, which binds to a calcineurin-dependent response element (CDRE) within the FKS2 promoter (320, 371). The core consensus binding sequence for Crz1 is a 6-bp motif (GAGGCTG) (366). Maintenance of high levels of FKS2 expression under chronic stress is driven by the CWI pathway but only partially through Rlm1 (144, 371). The sustained transcription of FKS2 in response to wall stress appears to be regulated largely by Swi4 (D. E. Levin, K. Y. Kim, and U. S. Jung, unpublished), a second transcription factor that is activated by Mpk1. Therefore, Rho1 controls both the activity of the GS during normal growth and expression of its catalytic subunit under conditions of wall stress. Recent developments also implicate Rho1 at multiple levels in the activation of Crz1 (see Skn7 and Cch1/Mid1). The complex regulatory network centered on the induced expression of FKS2 is evidently a mechanism to augment Fks1-derived GS activity under emergent conditions.

View larger version (27K): [in this window] [in a new window]   FIG. 2. Coordinate activation of CWI signaling, Ca2+ signaling, and Skn7 to induce FKS2 expression. Activation of Mpk1 results in stimulation of Rlm1 and Swi4 transcription factors to drive transcription of FKS2 (and other cell wall-related genes) and the Mid1-Cch1 Ca2+ channel. Channel activation leads to stimulation of the Ca2+-calmodulin-dependent protein phosphatase calcineurin. Dephosphorylation of the Crz1 transcription factor by calcineurin allows its entry into the nucleus. Rho1 may activate the Skn7 transcription factor (hatched line), which drives both expression of the OCH1 gene and stabilization of the Crz1 transcription factor for induction of FKS2. Thus, CWI signaling and Ca2+ signaling coordinately regulate expression of the GS-encoding FKS2 gene at several levels.

A recent study describing an in vitro assay for ß1,6-glucan synthesis revealed a requirement for GTP and, provocatively, demonstrated enhanced activity in cells overexpressing Rho1 (344). Thus, Rho1 may control the biosynthesis of both ß-glucan polymers. If this is correct, it seems likely that ß1,6-glucan synthesis is carried out at sites of polarized cell growth based on the localization pattern of Rho1. A further connection between ß1,6-glucan synthesis and CWI signaling is the observation that a pkc1 mutant is suppressed by overexpression of KRE6 (281). However, this may also be explained in terms of compensation for one wall defect by fortification of the glucan/chitin layer through an unrelated mechanism.

Bni1 and Bnr1. Formins nucleate the assembly of linear actin filaments in response to activation by Rho-GTPases. In yeast, the assembly of actin cables, which extend along the axis of growth and function as tracks for the transport of secretory vesicles into the bud, requires the functionally redundant formin proteins Bni1 (Bud neck involved) and Bnr1 (Bni1 related) (83, 285, 286). Bni1 and Bnr1 are components of the polarisome that reside at the cell cortex during bud growth (247) and nucleate the assembly of actin cables emanating from the bud tip. Both proteins relocalize to the bud neck during cytokinesis (247), where they are essential for the formation of the actomyosin-based contractile ring (332).

Bni1 and Bnr1 are large, multidomain proteins that have been shown to associate with multiple Rho proteins through an N-terminal Rho-binding domain (RBD). The RBD negatively regulates formin activity by binding to a C-terminal autoregulatory domain (termed DAD) (3, 252). Association of GTP-bound Rho proteins to the RBD relieves the autoinhibitory interaction. Bni1 also interacts in two-hybrid and coprecipitation assays with actin (Act1) and three actin-binding proteins—profilin (Pfy1), Bud6, and translation elongation factor 1 (Tef1/2) (82, 133, 337). All of these actin-binding proteins have been implicated in the assembly or organization of actin filaments. Additionally, Bni1 associates directly with the cell polarity protein Spa2 (94), which is an early marker for sites of wall expansion (314).

Each Bni1-interacting component appears to have a separate binding site on the protein. Profilin binds to Bni1 through a proline-rich formin homology domain 1 (FH1), which is shared with Bnr1 and metazoan formins. A second domain that is highly conserved among family members, called FH2, nucleates actin polymerization in vitro (263, 285, 286). Tef1 and -2 bind in a region between the FH1 and FH2 domains, Bud6 associates with the C terminus, and the Spa2-binding site is located at midprotein. Thus, the emerging picture is one in which the formins serve as a platform for the assembly of an actin filament-producing machine (44).

The GTP-bound forms of Rho1, Rho3, Rho4, and Cdc42, bind to the RBD of Bni1 (82, 94, 163). Additionally, Bnr1 has been reported to interact with Rho4 (133). Recent genetic analyses, summarized below, have provided some insight into the function of various Rho proteins with respect to Bni1 and Bnr1. Rho3 and Rho4 have been known for some time to serve a redundant role in bud growth that involves actin polarization (132, 208). As noted above, a rho3 mutant grows very slowly at all temperatures, and loss of RHO4 function exacerbates this growth defect, resulting in cell lysis during bud expansion. Overexpression of Cdc42 but not Rho1 or Rho2 suppresses the growth and actin polarity defect of a rho3 rho4 mutant, suggesting some functional overlap (208).

It now appears that the essential function of Rho3 and Rho4 is to activate Bni1 and Bnr1 during bud growth, because constitutively activated forms of Bni1 (Bni1RBD) or Bnr1 (Bnr1RBD) suppress the growth defect of a rho3 rho4 mutant (74). By contrast, Rho1 is required for Bni1/Bnr1-mediated actin ring assembly during cytokinesis (332). A rho1-2 mutant failed to form an actin ring at the mother-bud neck after release from nocodazole-induced M-phase arrest. Although neither rho3 nor rho4 mutants were examined in this study, a cdc42-1 mutant was competent for actin ring formation. Thus, it seems that Rho1 drives actin polarization through Pkc1 and the MAP kinase cascade during bud growth and through Bni1/Bnr1 during cytokinesis.

Skn7. The HOG (High Osmolarity Glycerol) pathway is controlled by a phosphorelay switch comprised of a cell surface sensor kinase (Sln1), a histidine phosphotransfer protein (Ypd1), and two response regulators (Skn7 and Ssk1) (155, 183, 199, 245, 261, 262). The Skn7 protein is one of only two yeast proteins related to bacterial response regulators of so-called two-component signal transduction pathways (183). Like many bacterial response regulators, Skn7 is a transcription factor. The other yeast response regulator, Ssk1, activates the MAP kinase cascade of the HOG (High Osmolarity Glycerol response) pathway. Whereas Ssk1 appears to be entirely under the control of Sln1, the lone sensor kinase of yeast, Skn7 activity is only partially regulated by Sln1. Skn7 may also be activated by Rho1 in response to cell wall stress (4) and through an unknown pathway in response to oxidative stress (167). Sln1 is a sensor of turgor pressure which is inactivated under conditions of low turgor (272). Sln1 regulates the HOG signaling pathway by phosphorylating the Ypd1 phosphorelay molecule, which transfers its phosphate to aspartyl residues within the receiver domains of both Ssk1 and Skn7, which constitute two branches of the same pathway (Fig. 3). Under high-osmolarity conditions, inactive Sln1 accumulates, resulting in dephospho-Ssk1, which is the active form of this response regulator. Thus, high extracellular osmolarity stimulates the Hog1 MAP kinase, which mediates, among other things, the biosynthesis and retention of glycerol as a compatible intracellular solute (5, 194, 328).

View larger version (24K): [in this window] [in a new window]   FIG. 3. Sln1 branch of the HOG pathway. The Sln1 sensor kinase is autophosphorylated under conditions of low extracellular osmolarity. Sln1 phosphorylates the histidine phosphotransfer protein Ypd1, which phosphorylates both the Skn7 and Ssk1 response regulators. Phospho-Skn7 is the active form of this transcription factor, which binds to the SSRE of OCH1 and probably other cell wall-related genes. Phospho-Ssk1 is the inactive form of this response regulator. Sln1 is inactivated in response to high extracellular osmolarity. This allows the dephosphorylated forms of Ypd1, Ssk1, and Skn7 to accumulate. Dephospho-Skn7 is inactive, whereas dephospho-Ssk1 activates the Hog1 MAP kinase cascade, which stimulates several transcription factors to drive expression of genes to mount the HOG (High Osmoslarity Glycerol) response.

och1

Skn7 associates with GTP-bound Rho1 through an HR1 domain that resides between the N-terminal heat shock factor-like DNA-binding domain and the response regulator domain (4). Although the significance of this interaction has not been tested directly, several lines of genetic evidence from Howard Bussey's laboratory implicate Skn7 in CWI signaling. The SKN7 gene was isolated initially as a dosage suppressor of the growth defect of a kre9 mutant (Suppressor of kre 9) (7, 34), which is deficient in ß1,6-glucan synthesis. Overexpression of SKN7 suppresses the growth defect of a pkc1 mutant in the absence of osmotic support (33). However, epistasis experiments suggested that if Skn7 functions within the CWI signaling pathway, its function is likely not under the control of Pkc1. Instead, it was suggested that Skn7 acts in parallel with Pkc1 to regulate cell surface growth. Finally, MID2, which encodes a cell surface sensor for CWI signaling (see next section), was isolated as a multicopy activator of an Skn7-LexA-dependent transcriptional reporter (156), suggesting that activation of CWI signaling can stimulate transcriptional activation by Skn7.

Taken together, these data suggest a model in which Skn7 is somehow activated by Rho1 (Fig. 2). However, it is difficult to hypothesize how Rho1, which is thought to reside only at the plasma membrane, might activate Skn7, which is reported to reside only in the nucleus (33, 129, 192, 266). The Ypd1 phosphorelay protein shuttles between the cytoplasm and the nucleus to phosphorylate Ssk1 and Skn7, respectively, in response to hypo-osmolarity (192). Perhaps a minor pool of Skn7 also undergoes nuclear-cytoplasmic shuttling in response to Rho1 activation.

Skn7 appears to be multifunctional, as reflected by its ability to partner with a variety of other transcriptional regulators at distinct promoter sites under different conditions. For example, its overexpression can bypass the requirement for the cell cycle transcription factors SBF and MBF (223), and it can pair with Mbp1, the DNA-binding component of MBF (31). Skn7 also binds to the Hsf1 heat shock factor and its cognate DNA element (HSE), and possibly the Yap1 oxidative stress factor, in response to oxidative stress (171, 222, 266). However, oxidative stress activation of Skn7 evidently does not involve Sln1 and is independent of the phospho-accepting Asp427 residue (183, 222, 266). Similarly, suppression of pkc1 is also independent of Asp427 (33). By contrast, Skn7-dependent transcriptional activation of the OCH1 gene requires phosphorylation of Asp427 through the activity of Sln1 (184). The Skn7 binding site within the OCH1 promoter is a 13-base-pair repeated sequence, ATTTGGCC/TGGC/GCC, called the Skn7 response element (SSRE), that may reflect a DNA-binding activity intrinsic to Skn7, or may indicate a novel partnering of Skn7 with another transcription factor. The OCH1 promoter also possesses a functional CDRE that is separate from the SSRE, indicating that Skn7 and calcineurin collaborate in the hypo-osmolarity induction of OCH1.

In addition to its function as a transcriptional activator, Skn7 associates with and stabilizes the calcineurin-activated transcription factor Crz1 (355). SKN7 answered a genetic screen for activators of a CDRE reporter. Intriguingly, its mechanism of action appears to be through interfering with the turnover of Crz1. Phosphorylation of Asp427 is not required for CDRE activation by Skn7, but mutations in either its HR domain or its DNA-binding domain block this function. This raises the interesting possibility that CDRE activation by Skn7 is driven through a Rho1-dependent, Sln1-independent pathway (Fig. 2). The question of the involvement of Rho1 in the regulation of Skn7 awaits further investigation.

Sec3. Cell surface expansion in yeast is driven by polarized exocytosis, a process that involves transport of post-Golgi secretory vesicles along the actin cytoskeleton toward the cell surface. These vesicles dock with components of the exocytic machinery localized to sites of polarized growth and ultimately fuse with the plasma membrane at these sites. A multiprotein complex called the exocyst, which is involved in vesicle targeting and docking at the plasma membrane, assembles at the exocytosis site in response to the arrival of vesicles. Sec3 is a component of the exocyst with the unusual property of localizing to the site of exocytosis independently of active secretion, the actin cytoskeleton, or other components of the exocyst. Therefore, Sec3 is thought to be a spatial landmark for polarized secretion (85).

The two essential G-proteins, Rho1 and Cdc42, have been proposed to be responsible for the spatial regulation of the exocyst complex. Genetic and biochemical evidence connects Sec3 with these GTPases. Sec3 becomes mislocalized in certain rho1 (rho1-5 and rho1-104) (103) and cdc42 (cdc42-13 and cdc42-201) (369) mutants. Sec3 associates with Rho1, Rho2, Rho3, and Rho4, as judged by two-hybrid analysis, and with Rho1 and Cdc42, as judged by coprecipitation (103, 369). Moreover, Rho1 and Cdc42 compete in vitro for a direct interaction with the N-terminal domain of Sec3, and an N-terminally truncated form of Sec3 fails to localize in a polarized manner, suggesting that this region of Sec3 may receive targeting information from Rho1 and Cdc42. Therefore, Rho1 and Cdc42 appear to collaborate in the process of vesicle delivery to the plasma membrane through dual control of actin cytoskeleton polarization (for vesicle transport) and vesicle docking through the exocyst. Although a rho3/4 mutant is not defective in Sec3 localization, Rho3 has been reported to associate with Exo70, another component of the exocyst important for vesicle docking (1, 279). Thus, most members of the Rho family participate at one or more levels in the exocytic pathway.

The Wsc proteins display sequence similarities with one another, (342, 343), and Mid2 shares 50% sequence identity with Mtl1 (156, 267). However, aside from the gross structural similarities between the two subfamilies, their sequences are not conserved. The Wsc proteins possess an N-terminal cysteine-rich domain that is absent from Mid2 and Mtl1. This region, termed the WSC domain, is found in human polycystin 1 (PKD1), a plasma membrane protein that is defective in autosomal dominant polycystic kidney disease (258). A Trichoderma ß1,3-exoglucanase also possesses a WSC domain (52), suggesting the possibility that this domain binds glucan chains.

Among the constellation of cell wall stress sensors, Wsc1 and Mid2 appear to be the most important. A double wsc1 mid2 mutant requires osmotic support to survive (267). The WSC1 gene was identified through several contemporaneous genetic screens. It was isolated as a dosage suppressor of the temperature-sensitive growth defect of a swi4 mutant (100). Swi4 is a component of the SBF transcription factor, which is a target of Mpk1 phosphorylation (see below) (196). WSC1 and WSC2 were also isolated as dosage suppressors of the heat shock sensitivity of a hyperactive Ras/cyclic AMP pathway mutant (343). The connection to Ras signaling remained obscure until recently, when cell wall defects were discovered to be among the many phenotypes associated with hyperactive Ras pathway mutants (315).

The isolation of WSC1 through a screen for mutants that require a constitutive form of BCK1 (BCK1-20) for growth provided a more satisfying link between the cell surface protein and CWI signaling (139). Deletion of WSC1 results in cell lysis at elevated growth temperatures (e.g., 37 to 39°C), a phenotype that is exacerbated by loss of WSC2 and/or WSC3 and suppressed by overexpression of Rho1, Rom2, or Pkc1 (100, 139, 343). Consistent with the importance of Wsc1 for survival of thermal stress, a wsc1 mutant is deficient in thermal activation of Mpk1 (100, 343). Like most other components of the CWI pathway, Wsc1 localizes to sites of polarized cell growth (63, 129).

The MID2 gene was isolated initially as a dosage suppressor of the growth defects associated with overexpressed cyclic AMP-dependent protein kinase (Tpk1) (59). This is reminiscent of one of the genetic screens that yielded Wsc1 and supports an antagonistic role for Ras/cyclic AMP pathway signaling in the maintenance of wall integrity. MID2 also answered a genetic screen for mutants that fail to survive pheromone treatment, a behavior that led to its most widely recognized name (Mating Induced Death 2) (242). MID2 was additionally isolated as a dosage suppressor of the growth defects associated with loss of the actin-associated protein profilin (203), loss of Wsc1 and Wsc2 (267), and as an activator of the Skn7 transcription factor (156). A mid2 mutant is not temperature sensitive for growth but is nevertheless somewhat impaired for Mpk1 activation in response to mild heat shock, particularly in combination with mtl1 (156, 267). Although MTL1 appears to play a minor role in CWI signaling and has rarely turned up in genetic screens, it was isolated as a dosage suppressor of a temperature-sensitive GS mutant (fks1-1154 fks2) (300). Consistent with the importance of Mid2 for the survival of mating, it is required for Mpk1 activation in response to pheromone (156, 267). It should be noted that pheromone-induced activation of CWI signaling is not a direct response to pheromone but rather a secondary response triggered by morphogenesis (35, 81, 276). Further support for the importance of Mid2 in signaling wall stress is its requirement for activation of Mpk1 in response to calcofluor white (156). An interesting distinction between Mid2 and Wsc1 is the former's uniform distribution around the cell periphery (156, 267, 129). The diffuse distribution of Mid2 in the plasma membrane may reflect its role in signaling wall stress resulting from pheromone-induced morphogenesis, which may be initiated at any point on the surface of a G₁-arrested cell.

Despite the differences cited above, it is clear that Wsc1 and Mid2 serve a partially overlapping role in CWI signaling. Overexpression of Wsc1 suppresses the pheromone-induced death associated with mid2 and, conversely, overexpression of Mid2 suppresses the temperature sensitivity of a wsc1 mutant (156, 267). Additionally, a double wsc1 mid2 mutant displays a severe growth defect, lysing at all temperatures in the absence of osmotic support. The cytoplasmic domains of both proteins display two-hybrid interactions with the N-terminal domain of Rom2 but not with Rho1 (257). This domain is distinct from the Rho1-interacting domain of Rom2, suggesting that the GEF can interact simultaneously with a sensor and with Rho1. Moreover, extracts from wsc1 and mid2 cells are deficient in catalyzing GTP loading of Rho1, providing evidence that the function of the sensor-Rom2 interaction is to stimulate nucleotide exchange of Rho1 (257). As would be expected, the short cytoplasmic domains of both Wsc1 and Mid2 are essential to their function (101, 189, 257, 342).

Domain analysis of Wsc1 indicated that the cytoplasmic domain is phosphorylated (189). Further mutational analysis of the cytoplasmic domain revealed that it possesses two short regions important for Rom2 interaction, one at the extreme C terminus and the other near the transmembrane domain (342). The phosphorylation site resides between these interaction regions and serves to inhibit Wsc1 function, probably by interfering with Rom2 interaction. However, phosphorylation is not the only means of Wsc1 regulation. A Wsc1 phosphorylation site mutant (S319/320/322/323A) is not constitutively active but is potentiated for activation by wall stress. It is interesting that a Wsc1 truncation mutant that has only 16 residues of the cytoplasmic domain (the membrane-proximal Rom2 interaction site) retains partial function (342). Such a short region may only be capable of providing a recruitment site for Rom2 rather than imposing a conformational change upon the GEF.

A small plasma membrane-associated protein, called Zeo1, was identified in a two-hybrid screen for interactors with the cytoplasmic domain of Mid2 (101). The Zeo1 protein associates with the region of Mid2 that is closest to the plasma membrane. It evidently interferes with Mid2 signaling because a zeo1 mutant displays constitutive Mpk1 activity that is dependent on Mid2. Moreover, loss of ZEO1 suppresses the growth defect of a rom2 mutant. This suppression likely results from enhanced activity of Mid2 through Rom1. It is not yet clear if Zeo1 interacts with any of the other sensors.

The extensive mannosylation of the periplasmic ectodomains of at least Mid2 and Wsc1 is also important to their function (191, 257). Both proteins are mannosylated by Pmt2 and Pmt4 (191, 257), members of a seven-isoform family of proteins that catalyze the first step in protein O-mannosylation (323). Although there is considerable overlap in substrate specificity among these various mannosyltransferases, only Pmt2 and Pmt4 are capable of modifying Mid2 and Wsc1. Consistently, a double pmt2 pmt4 mutant undergoes cell lysis in the absence of osmotic support (97). This defect is suppressed by overexpression of Pkc1, Wsc1, or Mid2 (191), suggesting that O-mannosylation of the sensors, although important, can be bypassed.

It is possible that signal transduction from the cell surface sensors is not unidirectional. Bud tip localization of Wsc1 is disrupted by the actin antagonist latrunculin A, indicating that its positioning is dependent on the actin cytoskeleton. Thus, although Wsc1 controls actin polarization through the action of Rho1, this sensor also responds to changes in the actin cytoskeleton. For this reason, it has been suggested that bidirectional signaling occurs between the actin cytoskeleton and Wsc1 (63) and perhaps the other sensors as well. Such bidirectional signaling is similar to that observed for animal cell integrins, plasma membrane proteins that connect the actin cytoskeleton to the extracellular matrix (298).

Although all five sensors appear to converge on Rho1, there may exist differentially regulated pools of this G-protein at the plasma membrane. The observation that overexpressed Wsc1 but not Mid2 suppresses the growth and GS activity defects of an fks1(Ts) fks2 mutant has prompted the suggestion that Mid2 activates a pool of Rho1 dedicated to the stimulation of Pkc1, whereas Wsc1 can drive activation of multiple Rho1 effectors (300).

Stt4 was shown only recently to reside at the plasma membrane, where it catalyzes the synthesis of phosphatidylinositol (PI) 4-phosphate (4P) (11) (Fig. 4). Interestingly, this protein appears to localize to patches at the mother cell cortex. A dosage suppressor of an stt4(Ts) mutant, designated SFK1 (Suppressor of Four Kinase) (11), encodes a plasma membrane-localized protein that appears to function as a tethering factor for Stt4. The SAC1 gene encodes a phosphatidylinositol-4-phosphate, which together with Stt4 regulates the pool of PI4P at the plasma membrane (91). Intriguingly, Sac1 resides predominantly at the endoplasmic reticulum, suggesting the possibility of contact between the peripheral endoplasmic reticulum and the plasma membrane.

View larger version (17K): [in this window] [in a new window]   FIG. 4. Phosphoinositide signaling pathway at the plasma membrane (PM). The sequential actions of Stt4 and Mss4 at the cell surface generate PI4,5P2, which recruits Rom2 to the plasma membrane through its PH domain for interaction with the cell surface sensors. The sensors activate Rom2, a GEF for the Rho1 GTPase. Sfk1 appears to be a plasma membrane tethering factor for Stt4. Mss4 cycles between the plasma membrane and the nucleus for reasons that are not yet clear. Yck1 and -2 are plasma membrane-associated casein kinase 1 isoforms that phosphorylate Mss4 and stabilize it at the plasma membrane.

The MSS4 gene encodes the only PI4P,5-kinase of yeast, and like STT4, it is essential for viability. Mss4 catalyzes the conversion of PI4P at the plasma membrane (the product of Stt4 activity) to PI4,5P₂ (70, 122). This gene was isolated initially in a screen for dosage suppressors of the temperature sensitivity of an stt4(Ts) mutant (364). Similarly to stt4(Ts) mutants, mss4(Ts) mutants display actin organization defects and cell lysis at the restrictive temperature (12, 70). Audhya and Emr (11) identified Rom2 as the first effector of the Stt4-Mss4 pathway by demonstrating that a critical role of PI4,5P₂ production is to recruit this Rho-GEF (and presumably Rom1) to the plasma membrane through its plekstrin homology (PH) domain. This recruitment is evidently integral to the activation of Rom2 GEF activity for Rho1 (Fig. 4). Interestingly, levels of PI4,5P₂ increase transiently in response to mild heat shock (11, 70), a stress that activates Mpk1 (148), suggesting that the concentration of this phosphoinositide in the plasma membrane contributes to the stress-induced activation of CWI signaling.

Although initial immunofluorescence studies localized Mss4 to the plasma membrane (122), a more recent study demonstrated that this protein shuttles between the nucleus and the plasma membrane (12). A mutant form of Mss4 that does not exit the nucleus was nonfunctional, but its function was restored by artificially tethering it to the plasma membrane, confirming that the essential function of this lipid kinase is at the cell surface. The purpose of Mss4 nuclear-cytoplasmic shuttling is not clear, but it may have a nonessential nuclear role. Mss4 is also phosphorylated at the plasma membrane by the casein kinase I isoforms encoded by YCK1/2 (12). This phosphorylation is required for stable membrane association of Mss4. It is possible that Yck1 and -2 are regulated by inputs that activate phosphatidylinositol signaling at the plasma membrane. However, these protein kinases are involved in diverse processes, including morphogenesis, cell wall biogenesis, nutritional signaling, cytokinesis, phosphoinositide metabolism, pheromone response, endocytosis, and turnover of membrane proteins (12, 120, 202, 224, 277, 278), suggesting that they may act simply by virtue of proximity to targets that present themselves at the plasma membrane.

Several additional targets of Stt4-Mss4 signaling have been suggested recently. The Cla4 protein kinase, which is a direct effector of the Cdc42 GTPase and is important for polarized cell growth, possesses a PH domain that binds in vitro to several phosphoinositide species (354). A mutant in STT4 but not MSS4 was defective in localization of Cla4 to the bud tip, suggesting that plasma membrane PI4P is responsible for recruitment of Cla4 to the cell surface for activation by Cdc42. Boi1 and Boi2, two related proteins of undetermined function, have also been suggested as targets of Cdc42 action that are important for polarized cell growth (25). Significantly, Boi1 and -2 possess PI4,5P₂ -binding PH domains (106), suggesting that, like Cla4, these proteins may be recruited to the plasma membrane for interaction with Cdc42. Finally, a recent genomewide synthetic lethality screen with an mss4(Ts) mutant yielded a pair of essential PH-domain proteins, designated Slm1/2 (Synthetic lethal with mss4) (14). Slm1 and -2 are required for polarization of the actin cytoskeleton, and their localization to the plasma membrane requires PI4,5P₂. The lethality of an slm1 slm2 mutant was suppressed by constitutive activation of Rho1 or Pkc1. Interestingly, Slm1 and -2 are also targets for Tor2 protein kinase activity and may be at least partially responsible for regulation of the actin cytoskeleton by Tor2 (see Tor signaling below).

Rlm1. The Rlm1 transcription factor (for resistant to the lethality of Mkk1^S386P) is responsible for the majority of the transcriptional output of CWI signaling. The RLM1 gene was identified in a genetic screen for mutants that could survive the growth inhibition caused by overexpression of a constitutive form of Mkk1 (350, 361). The encoded factor possesses an N-terminal DNA-binding domain related to the MADS (MCM1, Agamous Deficiens, Serum response factor)-box family of transcriptional regulators. Rlm1 is most closely related to mammalian MEF2, sharing the same in vitro binding specificity (CTA[T/A]₄TAG) (72). However, in vivo studies revealed that the binding specificity is relaxed at the terminal G/C base pairs (142, 143). Rlm1 displays two-hybrid interaction with Mpk1 (349) and is phosphorylated in vivo and in vitro by Mpk1 (359). Rlm1 evidently always resides in the nucleus, but its phosphorylation by Mpk1 at two residues within its transcriptional activation domain (Ser427 and Thr439) stimulates its activity (143). A MAP kinase docking site in the activation domain (delimited by residues 324 to 329) is shared with MEF2 and is also essential for activation by Mpk1 (143).

A genomewide survey of gene expression through CWI signaling revealed that Rlm1 regulates the expression of at least 25 genes, most of which encode cell wall proteins or have been otherwise implicated in cell wall biogenesis (144). This analysis was conducted using a constitutively active form of Mkk1 (Mkk1^S386P) expressed under the inducible control of the GAL1 promoter so as to restrict the output to genes that are regulated specifically by this pathway. Remarkably, regulation of the expression of all these genes (both positive and negative) in response to cell wall stress required Rlm1, indicating that this factor can act as either a transcriptional activator or a repressor depending on the context. The 20 genes identified as being activated in response to Mkk1^S386P expression were also activated by mild heat shock in an Rlm1-dependent manner. A similar global gene expression study reported the use of constitutive forms of Pkc1 and Rho1 to identify an overlapping set of CWI signaling-regulated genes (276). In this study, RLM1 was identified among the induced genes, suggesting the existence of an autoregulatory circuit for amplification of the stress response. Consistent with this, the RLM1 gene was also induced in response to cell wall stress associated with an fks1 mutation (36).

One intriguing transcriptional target of Rlm1 is MLP1 (Mpk1-like protein kinase), which encodes a homolog of Mpk1 (350). Its function has remained shrouded, due largely to an absence of phenotypic defects associated with its loss. Mlp1 shares 53% sequence identity with Mpk1, interacts with Rlm1 by two-hybrid analysis, but is lacking several catalytic domain residues recognized to be critical for protein kinase activity. First, the threonyl residue within the dual phosphorylation site of the activation loops of MAP kinases (T-X-Y) is mutated to Lys in Mlp1, indicating that it is not a true MAP kinase. Second, a universally conserved Lys residue within the ATP-binding site of all protein kinases is mutated to Arg. This is a mutation often generated to create a catalytically inactive form of protein kinases. Third, a universally conserved Asp residue within the triplet DFG in subdomain VII is mutated to Asn. Although Mlp1 protein levels increase by approximately 100-fold in response to cell wall stress, efforts to detect protein kinase activity have not been successful (A.<