ReviewCell Biology

Hog1: 20 years of discovery and impact

See allHide authors and affiliations

Science Signaling  16 Sep 2014:
Vol. 7, Issue 343, pp. re7
DOI: 10.1126/scisignal.2005458

Abstract

The protein kinase Hog1 (high osmolarity glycerol 1) was discovered 20 years ago, being revealed as a central signaling mediator during osmoregulation in the budding yeast Saccharomyces cerevisiae. Homologs of Hog1 exist in all evaluated eukaryotic organisms, and this kinase plays a central role in cellular responses to external stresses and stimuli. Here, we highlight the mechanism by which cells sense changes in extracellular osmolarity, the method by which Hog1 regulates cellular adaptation, and the impacts of the Hog1 pathway upon cellular growth and morphology. Studies that have addressed these issues reveal the influence of the Hog1 signaling pathway on diverse cellular processes.

Introduction

Two decades have passed since the protein kinase high osmolarity glycerol 1 (Hog1) was discovered in yeast (1), followed shortly by the discovery of the p38 and JNK (c-Jun N-terminal kinase) orthologs of this kinase in mammals (2, 3). These kinases play a central role in stress-activated signaling and are core components of the stress-activated protein kinase (SAPK) pathways. The SAPK pathways sense potentially harmful cellular environmental shifts and initiate either protective or apoptotic cell death responses to those conditions. During the more than 20 years of exploration into Hog1-based signaling, experimental approaches have matured from traditional gene discovery, genetic analysis, mutational studies, and biochemical studies of signaling components. New approaches to the study of this signaling pathway have included systems biology–based studies of integrated signaling networks (4, 5), comparative evolutionary biology (6), the therapeutic implications of modulating SAPK activity (7, 8), the regulation of core cellular processes including replication, transcription, translation, and metabolism (9), and mathematical modeling of signaling dynamics within living systems (10).

Discovery of the high osmolarity glycerol (HOG) pathway was published in a study defining two kinases that play a central role in mediating adaptive responses to osmotic stress (1). The kinases are members of the mitogen-activated protein (MAP) family of kinases with Pbs2, a MAP kinase kinase (MAPKK), serving as an upstream regulator of Hog1, a MAP kinase (MAPK). Genes encoding these two proteins were identified in a classic mutagenesis-based screen. Although Hog1 was initially discovered as a regulator of the yeast adaptation to osmotic stress, subsequently, Hog1 and its homologs in other organisms were found to regulate a wide array of stress responses, acting like a switchboard controlling cellular behavior and survival. This review is organized around key questions that arose during the initial characterization of Hog1, and the answers that have come since.

How Do Eukaryotes Regulate Their Internal Osmolarity and Cell Volume Under Osmostress?

Cells manage the cytoplasmic concentration of small molecules, called osmolytes, to counterbalance changes in extracellular osmolarity. Increasing solute concentrations outside the cell stimulates yeast to increase the intracellular glycerol concentration up to molar amounts (11, 12). To uncover a mechanism for osmosensing, we screened mutagenized yeast cells for reduced growth and glycerol accumulation in high-osmolarity medium (1). We anticipated identifying genes encoding a mechanosensitive ion channel mutant in the screen (13), but instead found a previously unknown MAPK cascade that mediated this stress response (1).

Osmotically sensitive mutants segregated into four complementation groups, three of which displayed low amounts of glycerol accumulation during osmotic stress. Sequencing of the wild-type alleles of the three genes identified two of the three as previously reported genes: HOG2, which was previously reported as TPS2 (trehalose-6-phosphate synthetase 2) (14), and HOG4, which was previously reported as PBS2 (polymyxin B sensitivity 2) (15). Analysis of the gene sequences revealed that HOG1 and PBS2 were highly homologous to genes encoding protein kinases, with HOG1 encoding a previously unreported homolog of the MAPK encoded by FUS3 (cell fusion defective 3) (16). Hog1 was phosphorylated on both threonine and tyrosine in response to osmotic stress, and this phosphorylation depended on Pbs2 activity (Fig. 1A). Within a year, the discovery of human orthologs p38 and JNK was announced (2, 3), revealing functional conservation in osmoregulation. Despite the distant evolutionary relationship between yeast and mammals, expression of either human ortholog of HOG1 within yeast cells lacking a functional copy of HOG1 restored osmoregulation in the mutant cells. Orthologs of HOG1 have been identified in all eukaryotic organisms examined. Indeed, this pathway defines a central stress response signaling network for all eukaryotic organisms.

Fig. 1 The developing model of osmoregulation in S. cerevisiae.

(A) The initial identification of the kinase Hog1 in signaling osmotic adaptation revealed a responsiveness that was dependent on the kinase Pbs2. (B) The Sln1 branch of the osmoregulatory pathway was identified by a lethal gene mutation that could be suppressed by protein tyrosine phosphatase 2 (Ptp2). Additional suppressors of sln1Δ lethality defined a branch of the osmoregulatory pathway. (C) The Sho1 branch of the osmoregulatory pathway was revealed through genetic manipulation of ssk2Δ ssk22Δ cells (lacking the Sln1 branch). (D) A summary of the current model of the HOG pathway in yeast. Osmotic stress releases Sln1-dependent inhibition of Ssk2/22 to activate the pathway. The Sho1 branch activation requires the membrane-embedded mucin proteoglycans Msb2 or Hkr1 to interact with Sho1 and Ste20 in a complex with the MAPK components. Opy2 is a transmembrane glycoprotein that serves as an anchor for Ste50. Cdc24 and Cdc42 are the cytosolic guanosine triphosphatase (GTPase) and guanine nucleotide exchange factor (GEF) that activate Ste20.

CREDIT: V. ALTOUNIAN/SCIENCE SIGNALING.

How Does Osmostress Activate the HOG Pathway MAPK Cascade?

The HOG pathway contains redundant signaling components upstream and downstream of Hog1 and Pbs2, making identification of these components challenging. Saito and colleagues played a key role in identifying and characterizing additional pathway components. While characterizing the protein phosphatase genes PTP1, PTP2, and PTC1 of Saccharomyces cerevisiae (17), Saito’s laboratory used a synthetic lethal screen to identify recessive gene mutations that required PTP2 overexpression for growth (18). Within the discovered genes was SLN1, a gene with sequence homology to prokaryotic two-component signaling proteins (19). Prokaryotic two-component systems sense environmental stimuli and signal cellular responses. Previously unknown in eukaryotes, these prokaryotic stress pathways involve a “sensor” protein that transfers high-energy phosphate through a histidine phosphorelay to a “receiver” protein, a response regulator (RR) that mediates the cellular response (20). Consistent with prokaryotic examples, Sln1 was localized within the cell membrane and contained a histidine kinase (HK) domain that autophosphorylates Sln1 to initiate a phosphorelay signaling mechanism (Fig. 1B) (21, 22). The lethality associated with deletion of SLN1 identified this protein as vital in the physiology of eukaryotic cells.

The connection between Sln1 and Hog1 signaling began with the identification of additional suppressors of the lethality associated with SLN1 deletion (sln1Δ). Three genes, loss of function of which suppressed sln1Δ lethality, defined core components of the HOG MAPK signaling pathway: SSK2, PBS2, and HOG1. YPD1 was subsequently identified in a synthetic lethality screen (22) and defined as encoding a histidine-containing phosphocarrier protein in the Sln1 two-component signaling pathway (Fig. 1B). A multistep phosphorelay between Sln1, Ypd1, and the RR Ssk1 is active in yeast experiencing normal turgor pressure. Active Sln1 autophosphorylates His576 followed by a transfer to Asp1144; this enables a phosphorelay to His64 on Ypd1 and subsequent transfer to Asp554 on the receiver domain of Ssk1. Phosphorylated Ssk1 is inhibited with regard to its binding and activation of the Ssk2 and Ssk22, which are homologous and functionally redundant MAP kinase kinase kinases (MAPKKKs) and referred to collectively as Ssk2/22. Osmotic stress arrests Sln1 HK activity, and nonphosphorylated Ssk1 associates with and activates the kinase activity of Ssk2/22 (18). Once active, Ssk2/22 binds and activates Pbs2 by phosphorylation of Ser514 and Thr518. Active Pbs2 phosphorylates Hog1 at Thr174 and Tyr176 of a Thr-Gly-Tyr (TGY) motif that is conserved in Hog1 homologs from yeast to humans (23, 24). The tyrosine phosphatase Ptp2 dephosphorylates Hog1 and inhibits the pathway after osmoadaptation (25), consistent with the ability of Ptp2 to suppress the lethality associated with a loss of SLN1 function.

In evaluating the Sln1-regulated pathway, the Saito group predicted that double mutant ssk2Δ ssk22Δ cells should display no activation of Hog1 in response to osmotic stress and that the cells would be osmotically sensitive. They were surprised to note nearly normal osmoregulation in these cells, indicating that another redundant osmostress signaling pathway must also regulate the HOG pathway. To find the second pathway, they mutagenized the ssk2Δ ssk22Δ cells and isolated mutants with sensitivity to osmotic stress. One gene that was identified was named Sho1 (synthetic, high osmolarity–sensitive). The encoded protein with four transmembrane domains and a single, cytosolic SH3 domain was localized to the cell membrane (26) (Fig. 1C). SH3 domains are commonly associated with protein-protein interactions during the formation of protein complexes. As expected, triple mutant cells (ssk2Δ ssk22Δ sho1Δ) displayed osmotic sensitivity and a lack of HOG pathway activation after osmotic stress exposure. The two key upstream osmoresponsive branches of Hog1 activators had now been identified.

The mechanism for Sho1 interaction with Pbs2 was initially unclear, although a MAPKKK similar to Ssk2/22 in the Sln branch of the pathway was anticipated to mediate this signal. A docking function was soon identified for Pbs2, revealing that it bound Hog1, Sho1, and the MAPKKK Ste11 (21). The kinase Ste11 was initially discovered as key component of the yeast mating response pathway (27). Ste11 was identified in a manner similar to the discovery of Sho1: cells deficient for the Sln1 pathway (ssk2Δ ssk22Δ) that could osmoregulate exclusively through the Sho1 sensor were mutagenized and screened for a loss of osmoregulation. In ste11Δ ssk2Δ ssk22Δ cells, osmotic stress did not activate Hog1 or Pbs2, and the triple mutant did not grow in high-osmolarity medium. Coimmunoprecipitation experiments revealed that Pbs2 was in a complex with Sho1, Ste11, and Hog1, thus defining an osmosensing MAPK signaling complex localized to the cell membrane that was similar to the mating pheromone-responsive MAPK pathway, which involved Ste11 (MAPKKK), Ste7 (MAPKK), Ste5 (scaffold), and Fus3 (MAPK). The formation of such a complex is a model for controlling crosstalk among signaling pathways, enabling a single protein (Ste11 in this case) to play a similar role in each pathway without inappropriate cross-regulation of separate signaling networks.

How Is Osmotic Stress Detected to Activate the HOG Pathway?

A central question surrounding the search for yeast osmoregulation genes was the sensing mechanism upstream of glycerol production and osmoadaptation. Shifting osmolarity in the medium can cause dramatic movements of water in and out of the cell, and survival requires the ability to rapidly respond to these changes. The S. cerevisiae cell wall surrounds the plasma membrane and is composed of a complex network of polysaccharides (glucans and chitin), proteins, and glycoproteins (28, 29). Sensing osmotic gradients across membranes begins with changes in membrane tension or “stretch.” Mechanosensitive cation channels were first discovered in mammalian cell membranes using patch-clamp techniques (30). These channels are present in all kingdoms and are responsible for complex regulation of cell size and morphology (31). Animal cells display receptors with different sensitivity to membrane tension, which converge upon shared signaling pathways (32). Sln1 and Sho1 are considered the osmosensors in yeast, but the physical method of sensing osmotically induced mechanical stress is not well understood. Despite this gap in mechanistic information regarding sensing the osmotic stress, genetic studies have revealed what is likely to be most of the proteins involved in this osmotic stress pathway, and a working model is developing (Fig. 1D).

The Sln1 HK is an active homodimer under normal levels of cellular turgor. Increasing extracellular osmolarity decreases cellular turgor pressure, dehydrating the yeast cell and causing a reduction in cell volume. Under these conditions, the phosphorelay-based inhibition of the Ssk1 RR protein is arrested, stimulating the HOG pathway. Turgor pressure itself serves as the primary control mechanism for the HK activity of Sln1. A common theme within two-component signaling pathways is the sensory domains that mediate interactions with auxiliary signal transduction proteins and detect stimulation through this interaction (33). Mutagenesis studies revealed that the osmotic regulation of Sln1 depends on the N-terminal extracellular domain and one of the associated transmembrane domains (34). This region is hypothesized to sense macromolecular crowding through both membrane-associated and cell wall–associated components. Removal of the cell wall, ablation of the gene encoding the cell wall protein Ccw12, or permeabilization of the cell membrane with nystatin all inactivate Sln1 HK such that it is not responsive to modulation of extracellular osmolytes (9, 35). Imagining a turgor-induced regulatory interaction of cell membrane proteins with this structure is a particularly intriguing possibility. Normal cellular conditions include a turgor-induced pressing of the cell membrane against the cell wall, potentially maintaining the inhibitory HK activity of Sln1 through Sln1–cell wall interactions. Osmotic stress would cause the membrane to pull away from the cell wall, potentially suppressing Sln1 activity through this loss of interaction and activating the HOG pathway. However, additional structural studies of Sln1 to evaluate regulation of its HK activity and an in vivo examination of the putative interactions of Sln1 with cell wall components are needed.

The transmembrane protein Sho1 plays a critical role in regulating two MAPK pathways: the pathway involved in activating filamentous growth and the pathway involved in the response to osmotic stress. In addition, Sho1 functions in the formation of focal cytokinesis sites during cell division (36). During osmoregulation, Sho1 participates in Hog1 activation through assembly and regulation of a membrane-associated complex with the scaffold protein Pbs2, Ste11, and Hog1. Sho1 with its four transmembrane domains anchors the complex to the cell membrane and interacts with Pbs2 through a cytosolic SH3 domain within residues 306 to 367 and with Ste11 through residues 172 to 211 (37). The SH3 domain of Sho1 also binds to Fus1, a cell wall–associated protein required for mating competency. This interaction may serve as a negative regulator of Hog1 signaling during mating, with activated Fus1 competing with Pbs2 for binding to Sho1 (38).

Sho1 is now recognized as more than just an osmosensor; it also serves as a critical adaptor protein connecting signals from the cell wall to MAPK signaling through the formation of distinct membrane-associated complexes. Sho1 recruits proteins necessary for Ste11 phosphorylation and activation, a GTPase protein and its regulators: the GTPase Cdc42, the GEF Cdc24, and the functionally redundant protein kinases Ste20 or Cla4 (39, 40). The regulation of filamentous growth and osmoregulation through Sho1 requires Msb2 and Hkr1, heavily glycosylated mucin-like transmembrane proteoglycans, which stimulate the coordinated scaffolding functions of Sho1 with Pbs2 (41). Msb2 and Hkr1 have the capacity to activate HOG pathway signaling through Sho1, although the mechanisms of interaction with the pathway are distinct. Activated Msb2 interacts with cytosolic Bem1 to recruit Ste20 or Cla4 into the complex, enabling their activation of Ste11. Hkr1 activation of the HOG pathway does not require Bem1 (42). The plasma membrane protein Opy2 has a single transmembrane domain, and its cytoplasmic C terminus has multiple, functionally distinct binding sites for Ste50 (4345), an adaptor needed for Ste11-Pbs2 signaling downstream of Sho1 (46).

Mucin proteins in mammals comprise a primary component of mucus. Some membrane-spanning forms of mammalian mucin proteins regulate cell signaling pathways (47). Mucin activation of signaling often involves posttranslational processing, such as glycosylation modifications or proteolytic cleavage of the mucins, to remove the inhibitory influences of the extracellular domain (40, 48). The interaction of the lengthy, glycosylated extracellular domains of Msb2 and Hkr1 with the chitin and glucan network is a potential source of osmosensing. Dehydration of the cell would pull the cell membrane away from the cell wall, an action that could be physically sensed by Msb2 and Hkr1 and then potentially stimulate assembly of the osmoresponsive MAPK signaling complex.

A recurring theme from the study of signal transduction pathways is the localization of signaling complexes to enable signal amplification, limitation of crosstalk, and increased control. At the cell membrane, protein complexes can be anchored to the membrane by membrane association or insertion. Sphingolipid- and sterol-enriched microdomains (also called lipid rafts) provide some partitioning of cell membranes and serve as a site of regulated protein localization (49). Both Sho1 and Sln1 copurify with detergent-resistant lipid rafts, with osmotic stress differentially influencing this localization: Sln1 decreases and Sho1 increases (50). Impeding sphingolipid biosynthesis, which gradually disrupts lipid rafts, activates both the Sho1 and Sln1 branches of the osmoregulatory pathways in the absence of any osmotic stress. These results offer an intriguing possibility that raft localization may play a role in regulating these adaptor proteins.

What Are the Acute Responses to Hog1 Signaling?

The initial isolation of Hog1 was tied to its requirement for adaptive up-regulation of glycerol biosynthesis after osmotic stress. A central question that remained after the discovery of the HOG pathway was how glycerol production was rapidly induced to avoid cell volume decrease in hypertonic solutions. After stimulation, phosphorylated Hog1 translocates to the nucleus, where it modulates transcription factor activity by phosphorylation, localizes with the RNA polymerase II (RNA Pol II) complex at various promoters, and enhances the nuclear export of stress-responsive gene transcripts (51). The expression of more than 300 genes is regulated by Hog1 in response to osmotic stress, but only a small subset directly affects cytosolic glycerol concentrations. The glycerol-regulating osmoresponsive genes include the STL1 encoding the glycerol uptake transporter (52) and the enzymes that convert dihydroxyacetone phosphate (DHAP) to glycerol:GPD1 and GPD2 encoding glycerol-3-phosphate dehydrogenases, and GPP1 and GPP2 encoding glycerol-3-phosphatases (53).

Despite the profound effect of the localization of active Hog1 within the nucleus on transcriptional regulation, elimination of Hog1 nuclear localization revealed this response to be dispensable for acute osmoregulation (54). In an experiment designed to assess crosstalk among MAPK pathways, Thorner and coworkers generated yeast cell strains with cytosol-limited Hog1 either by using cells lacking Nmd5, an importin β homolog required for nuclear import of Hog1 (55), or by anchoring Hog1 to the cell membrane. These yeast strains exhibited none of the known Hog1-dependent transcriptional regulation but were surprisingly tolerant to osmotic stress and capable of dynamically increasing cytosolic glycerol to osmoregulate. This striking result indicated that Hog1 does not mediate osmoregulation primarily through transcriptional responses, but rather through regulation of cytosolic proteins. Analysis of the cells with cytosol-limited Hog1 revealed which genes are essential for osmoregulation under this unusual condition. The glycerol channel glyceroporin (encoded by FPS1) is inhibited by phosphorylation of the Fps1-interacting protein Rgc2 (56), thereby reducing glycerol release from the cell and contributing to glycerol accumulation during osmoregulation. Unexpectedly, deletion of FPS1 or AQY1 and AQY2, which encode the water transporters aquaporin1 and aquaporin2, was surprisingly ineffective at generating osmotic sensitivity in these cells. The phosphorylation of Rgc2 has been attributed to both Hog1-dependent and Hog1-independent mechanisms (56, 57).

Mutating the cells with cytosol-limited Hog1 to generate defects in glycerol biosynthesis (by deletion of GPD1 or GPP1 and GPP2) or the production of DHAP (by deletion of TPI1, encoding triosephosphate isomerase) conferred osmosensitivity, consistent with Hog1 cytosolic signaling exerting its osmoregulatory influence through these enzymes. Activation of the Hog1 pathway is also responsible for stimulating the activity of phosphofructokinase 2 (Pfk2), which increases the production of fructose 2,6-bisphosphate, a metabolite that allosterically stimulates phosphofructokinase (Pfk1) (58) (Fig. 2).

Fig. 2 A summary of Hog1-based glycerol accumulation by regulation of glycolytic enzymes in the cytosol.

Rapid glycerol accumulation after osmotic stress is accomplished through the Hog1-based inhibition of Ypk1 and stimulation of Pfk2. This rapid response is followed by Hog1-induced increases in the expression of GPD1, GPP1, and GPP2.

CREDIT: V. ALTOUNIAN/SCIENCE SIGNALING.

Increased glycolysis may feed demand for DHAP during stress-induced glycerol production. DHAP metabolism is a convergence point for multiple metabolic pathways. The reduction of DHAP to glycerol-3-phosphate by glycerol-3-phosphate dehydrogenase 1 (Gpd1) is normally repressed by target of rapamycin complex 2 (TORC2)–dependent kinases Ypk1 and Ypk2. Osmotic stress results in the phosphorylation and inhibition of Ypk1 in a manner dependent on the Hog1 pathway (59), thereby stimulating Gpd1 activity and glycerol synthesis.

Thus, Hog1 signaling plays an acute role in regulating metabolic pathways in a manner that is sufficient for immediate glycerol accumulation and osmotolerance. The transcriptional response is critical for osmoregulation over time, but is nonessential for acute adaptation.

How Does Hog1 Signaling Mediate Long-Term Adaptation to Osmotic Stress?

Osmotic stress induces active Hog1 transport into the nucleus within seconds, and that localization is sustained for 20 min before gradually declining to basal levels within 1 hour even if the cells remain exposed to high osmotic conditions (60). Not only has examination of Hog1 activity in the nucleus revealed insight regarding the eukaryotic osmotic stress response, but understanding this transcriptional regulatory function has provided fundamental insights into the regulation of gene expression in eukaryotic cells. One of the first genes identified as a Hog1-responsive gene was CTT1, encoding catalase 1, which is controlled by the stress response element (STRE) that is found within the promoter of a number of stress-responsive genes in yeast (24). Identifying the STRE as a genetic element regulated by Hog1 signaling revealed that Hog1 could induce genes that had functions that were not primarily involved in the adaptation to osmotic stress. Several groups began to assemble the list of substrate proteins regulated by Hog1 and the genes regulated by this network, eventually revealing a remarkably large and diverse regulatory system.

Genetic studies of osmoregulation revealed several transcription factors that are regulated by nuclear Hog1 activity. These include Hot1, Msn1, Sko1, and the functionally redundant Msn2 and Msn4 (collectively referred to as Msn2/4) (61). The STRE regulatory sequence is primarily controlled by the Msn2 and Msn4 transcription factors, key mediators of Hog1-controlled changes in gene expression. With the development of microarray technology to assess global patterns of gene expression, the Hog1-induced transcriptional targets during osmostress were fully documented (52, 62). These gene expression studies revealed a cluster representing ~10% of the entire yeast genome as responsive to diverse stresses (osmostress, pH, temperature, and H2O2), and this cluster was called the environmental stress response (ESR) (63). Exposure to one type of stress can generate growth arrest and the adoption of protective adaptations by the cell, presumably in preparation for the possibility of subsequent environmental stresses. During osmotic stress, Hog1 signaling affects ~70% of ESR genes in S. cerevisiae.

Analysis of active Hog1 within the nucleus has revealed functionality beyond the regulation of transcription factors. Hog1 localizes with RNA Pol II, stimulates chromatin remodeling in the form of stress-induced redistribution of nucleosomes from the site of osmoresponsive gene promoters, and promotes the export of stress-responsive mRNAs from the nucleus (51, 64). A role for Hog1 in chromatin remodeling was identified by studying yeast with mutations in Rpd3, the catalytic subunit of yeast histone deacetylase complexes (HDACs), which were osmosensitive and did not appropriately induce osmoresponsive genes (65). Active Hog1 bound Rpd3, triggering its localization to the promoters of osmoresponsive genes. Rpd3 is found within the large multisubunit protein complex RSC, named for its function of remodeling the structure of chromatin and is recruited to the open reading frames of osmoresponsive genes in a Hog1-dependent manner (66), promoting rapid nucleosome eviction and RNA Pol II recruitment to these sites.

A genome-wide assessment of RNA Pol II localization during osmotic stress and the impact of nuclear Hog1 were performed using the chromatin immunoprecipitation and sequencing (ChIP-Seq) methodology (64). This study revealed that Hog1-mediated stress signaling repositioned RNA Pol II away from housekeeping genes in addition to promoting the localization of RNA Pol II at stress-responsive genes. In addition, repositioned RNA Pol II was physically associated with active Hog1. These results have established a model in which active Hog1 enters the nucleus and induces a dramatic redistribution of RNA Pol II and the eviction of nucleosomes from stress-responsive genes.

What Have Modeling and Computational Biology Studies Revealed About HOG Pathway Signaling?

Systems-level analysis, based on mathematical modeling, has been used to investigate the relative role of different negative feedback mechanisms in mediating transient activation of HOG pathway by osmostress and in the homeostatic control of the osmotic gradient across the cell plasma membrane (6769). Combined experiment and modeling revealed that Sho1 oligomerization and activity are negatively regulated by Hog1-mediated phosphorylation (69) and that basal amounts of Hog1 activity enable a faster response to osmostress than could be achieved if Hog1 was completely inactive in nonstressed cells (68).

Computational analyses have indicated that different aspects of the HOG pathway have more or less importance in the adaptation to osmotic stress. Fitting of data to different mathematical or engineering control-based models provides a strong argument (70, 71) that Hog1-mediated regulation of the activity of cytoplasmic enzymes that catalyze glycerol synthesis from sugar represents the key mechanism by which homeostatic control of HOG pathway is achieved and osmotic gradients in periods of osmotic stress are maintained (54, 58, 72). Less important in perfect adaptation (where steady-state signaling is equal to pre-stress signaling) to osmostress (70, 71) is Hog1-induced expression of glycerol synthesis genes, such as GPD1 (73), and negative regulation of the Fps1 glycerol channel by Hog1 and other signaling proteins (56, 74, 75).

Why Do Hog1-Deficient Yeast Have Such a Strange Morphology?

During the original examination of hog1Δ and pbs2Δ mutants, we noticed that both displayed an unusual morphology when stressed with high osmolarity. The cells displayed abnormal bud-like extensions, growing large and structurally complex, with multiple nuclei apparent within each structure (Fig. 3A). Using both wild-type and HOG pathway mutant strains, analysis of osmotic stress upon bud-site selection, the cytoskeleton, and cell division indicated that osmotic stress affected both growth control and the cell cycle (76, 77). The unusual morphology was ultimately revealed to originate from crosstalk among upstream components of yeast MAPK pathways. When hog1Δ or pbs2Δ mutant cells are exposed to high-osmolarity medium, the Ste11 MAPKKK upstream of the pathway defect is continually stimulated, and, in the absence of pathway feedback, activates other MAPK pathways, in particular the pheromone response pathway and the filamentous growth pathway. Ste11 (anchored to the Ste50 adaptor protein) is essential for the Sho1 branch of osmotic stress signaling, and this Ste11-Ste50 complex also serves as the MAPKKK in the pheromone response and filamentous growth pathways. Indeed, all three pathways include Ste20 [a p21-activated kinase (PAK) family member], the Cdc42 membrane-associated GTPase and partnered Cdc24 GEF, the Ste50 adaptor protein, and Ste11 as common MAPK pathway components, which communicate with pathway-specific downstream components (Fig. 3B). The MAPKK Pbs2 is functionally restricted to the osmoresponsive pathway, whereas Ste7 is the MAPKK in both the pheromone response and filamentous growth pathways. The unique MAPK in each pathway includes Hog1 in osmosensing, Fus3 in the pheromone response pathway, and Kss1 in filamentous growth signaling. Thus, hog1Δ or pbs2Δ mutant cell lines exposed to osmotic stress exhibit abnormal mating behavior and invasive filamentous growth through pathway crosstalk among shared upstream components (7880). Perhaps, the most challenging area of HOG pathway characterization is unraveling interactions among MAPK pathways.

Fig. 3 Osmotic stress in hog1 mutant cells results in a breakdown of MAPK pathway insulation.

(A) Osmotic stress induces abnormal pseudohyphal morphology in cells carrying a null mutation in HOG1 or PBS2 genes. (B) Three MAPK pathways in yeast, the pheromone-responsive, osmoregulatory, and filamentous growth pathways, share common signaling components at the cell membrane. Pathways are notably distinct with regard to the signal receptor and the MAPK for each pathway. Hog1-dependent feedback and cross-inhibition signaling are illustrated. (C) Deletion of HOG1 generates cells that are sensitive to osmotic stress and inappropriately activate other MAPK pathways potentially through hyperstimulation of shared upstream components, although the molecular nature of this crosstalk remains unresolved.

CREDIT: V. ALTOUNIAN/SCIENCE SIGNALING.

Although yeast MAPK pathways share several components, they control completely different cellular processes. Exposure of a haploid cell to peptide pheromone from the opposite mating type stimulates a receptor-mediated MAPK pathway to generate mating competency. Pheromone-induced responses include arrest of the cell cycle and the formation of a “shmoo,” a structure that facilitates fusion between the two mating-competent cells (type a with type α). The capability for S. cerevisiae to enter into an invasive filamentous growth or pseudohyphal invasive phase was poorly recognized before the announcement that nitrogen starvation triggered this phenomenon as a method of foraging (81). The genetic dissection of the filamentous growth pathway revealed the MAPK Kss1 as activated through the MAPKKK Ste11 and the MAPKK Ste7 (82), components that previously only associated the pheromone response pathway.

Perhaps, the most striking examples of interchangeable components are Ste50 and Ste11, the adaptor and MAPKKK in each of the three pathways. In all three pathways, Ste20 phosphorylates Ste11 from a membrane-associated complex that includes Cdc24 and Cdc42. This activation event is pathway-specific because each pathway involves the induced recruitment of the Ste20/Cdc42/Cdc24 complex to a specific scaffold-bound complex that includes Ste11 and the pathway-specific MAPKK and MAPK. Pathway components that are shared among the three pathways are thus insulated from the other pathways through their sequestration within a membrane-anchored protein complex (Fig. 3B), with Pbs2 serving a key scaffolding role in the HOG pathway (83), Ste5 serving this role in the pheromone pathway (84), and an anticipated, but as yet undiscovered protein serving this role in the filamentous growth pathway. Continual pathway stimulation or loss of feedback responses creates conditions in which pathway crosstalk is increased. One possible mechanism for the aberrant crosstalk could be through the diffusion of active Ste11 or other components from the hyperstimulated scaffold.

Creative new experimental approaches have yielded critical insight into crosstalk among MAPK pathways. In wild-type cells, HOG pathway activation after mild osmostress is transient, with Hog1 activity being sustained for ~20 min, before adaptive responses (glycerol accumulation), increased phosphatase activity, and negative feedback inhibit the pathway (9, 25, 69, 80, 85). Using crystal structure data for both Hog1 and p38, an alanine substitution within the ATP (adenosine 5′-triphosphate) binding site was predicted to confer highly specific susceptibility of the mutant kinase to an ATP analog. This “analog-sensitive” inhibitor approach had previously been used to characterize other kinases (86). The rapid and specific inhibition of a single kinase enables an inducible control over the target pathway and is, therefore, superior to genetic mutants that survive in the absence of the enzyme in question. To address the mechanism through which active Hog1 provides negative feedback to upstream components, an ATP analog-sensitive T100A mutant form of Hog1 (Hog1-as) was engineered (87). Yeast strains expressing the hog1-as allele from the normal HOG1 locus displayed a completely wild-type phenotype, including normal osmoregulation. However, in the presence of the ATP analog/inhibitor, Hog1 activity was lost, and osmoregulation was deficient in a manner similar to hog1Δ null mutants. One of the more remarkable results provided by this study is that Hog1 activity during osmoregulation is required for stimulation of Hog1 dephosphorylation and for inhibiting the inappropriate activation of the filamentous growth and pheromone response pathways (87). Another report revealed that Hog1 phosphorylated Ste50 at several sites, decreasing its affinity for the Opy2 transmembrane protein and inhibiting inappropriate activation of Kss1 during osmotic stress (80). Ste50 is similarly phosphorylated by active Fus3 and Kss1. Mutation of Ste50 to ablate known phosphorylation sites resulted in extended activation of Hog1 during osmotic stress and inappropriate activation of the filamentous growth pathway (44). Active Hog1 also phosphorylates Sho1, generating an inactive monomeric form of this osmosensor (69). These studies reveal a critical role for a MAPK in feedback inhibition of upstream pathway components to limit hyperstimulation of the pathway and minimize inappropriate activation of pathways that share components.

Two models have been investigated to explain how pathways with shared components minimize inappropriate nontarget pathway activation: cross-inhibition and pathway insulation. Hog1 activity inhibits Kss1-mediated stimulation of the transcription factor Tec1 (88), supporting the cross-inhibition model, in which activation of one pathway limits the activation of other MAPK pathways (89). Assessing two pathways simultaneously and within a single cell is indeed technically challenging, and a unique approach was used to test the MAPK cross-inhibition model (90). Activation of the osmosensing and pheromone response pathways each stimulates relocalization of a pathway component: Hog1 moves into the nucleus during osmosensing, and Ste5 moves to the cell membrane during the pheromone response. Recombinant constructs were generated for each gene, fusing a fluorescent tag with each protein (Hog1-RFP and Ste5-GFP). Stimulation of either pathway generated a clearly quantifiable change in component localization, and dual stimulation surprisingly did not show cross-inhibition. Indeed, the dynamics of signaling in this study support a model of pathway insulation, in which the scaffolding and feedback signaling within each pathway protect common components from incorrectly activating the off-target pathway. Evaluation of these yeast MAPK pathways continues to be a productive and interesting area of study in the characterization of crosstalk among parallel signal transduction pathways.

Is Hog1 Functionally Conserved Among Higher Eukaryotes?

With the identification of Hog1 as a key mediator of signaling dehydration, we anticipated that orthologous proteins would be found in other eukaryotes, noting the potential importance of this signaling in plants adapting to drought conditions. Mammalian orthologs (SAPKs) were initially discovered through their activation in response to different stress-related signals, including endotoxin (2), osmotic stress (3), ultraviolet light (9193), and proinflammatory cytokines, such as tumor necrosis factor–α (TNFα) (91, 94, 95). The mammalian p38 family of kinases is part of a signaling pathway that mediates an increase in intracellular concentration of compatible osmolytes in response to hypertonic stress (9698). Announcement of these mammalian orthologs of Hog1 included data that showed that any of them were able to substitute for Hog1 in yeast hog1Δ cell lines, restoring osmoregulation and survival in high-osmolarity conditions (2, 3). This revealed a surprising functional conservation of SAPKs over extensive periods of evolutionary time.

The MAPK genes encode proteins that are central signaling components in all eukaryotes, and the SAPK genes comprise a key subcategory of the MAPK genes. The p38 (SAPK1 family) and JNK (SAPK2 family) genes have been observed in all metazoans, including sponges, the simplest examples of multicellular animals (99). Phylogenetic analysis has traced the likely origin of all SAPKs to a primordial HOG1 gene and has indicated that an early duplication event occurred after the Urmetazoa split from yeast, but before the split from Porifera (99, 100). JNK genes diverged further from HOG1 than did p38 genes, including a change in the phosphorylation site in the activation loop from a TGY motif in the SAPK1 family (Hog1, p38) to a TPY motif in the SAPK2 family (JNK). Gene duplication events have added copies of p38 and JNK in higher eukaryotes: mammalian genomes include four p38-encoding genes (p38α, p38β, p38γ, and p38δ) and three JNK-encoding genes (JNK1, JNK2, and JNK3).

Sequenced fungal genomes have enabled an examination of the evolutionary conservation of osmoregulation genes across the kingdom. Although these genes are well represented in each genome, interesting exceptions include fungi that lack the Sln1 HK branch for osmosensing or that have a SHO1 gene that lacks the proline-enriched Pbs2 interaction domain (101). The targets of Hog1 activity identified in S. cerevisiae were divergent among the other species of fungi evaluated, revealing flexibility in pathway output. When examining sequence conservation of specific components across diverse fungal examples, the kinase Hog1 and the GTPase Cdc42 were the most highly conserved genes. Computational assessment of the evolutionary rates of all orthologous HOG pathway genes (19 genes total) performed using 10 sequenced yeast genomes (6) revealed a clear correlation between the position of a gene in the pathway and its divergence. The HOG1 genes and genes encoding the immediate upstream regulators were most highly conserved, whereas more distal sensors and pathway activators displayed higher rates of evolutionary change. Selective constraints upon HOG1 evolution are hypothesized to be due to the complexity of converging stress signals and the central role of Hog1 in distributing that signal.

There are five MAPK pathways in S. cerevisiae, each responsible for cellular responses to extracellular stimuli. MAPK components of each pathway share a high degree of homology and are thought to have arisen by gene duplication events early in eukaryotic evolutionary biology. Through evolutionary time, these kinases have gained new signaling specificities, binding partners, and downstream targets. In one notable study, the Hog1 and Fus3 kinases were comparatively evaluated by domains within the protein sequence, assessing the impact of interchanged domains upon kinase responsiveness and substrate specificity (85). Protein kinases regulate other proteins and are themselves regulated, each through protein-protein interactions that facilitate recognition. Plasticity within MAPK signaling pathways over evolutionary time has been hypothesized to arise within surface domains that primarily mediate interactions with other proteins. Catalytic and internal structural domains are highly conserved. Mody et al. (85) examined structural organization in the five MAPK proteins of yeast, and noted six key domains in each kinase (A to F), each containing one surface domain. To evaluate the importance of each domain upon kinase activation and specificity, a combinatorial approach was undertaken, creating 64 distinct kinases with different combinations of Hog1 and Fus3 domains (ranging from 100% Hog1 to 100% Fus3). Activation of osmoregulation was assessed in hog1Δ cell lines carrying the hybrid kinase proteins. Alternatively, pheromone response was assessed in fus3Δ cell lines carrying the hybrid kinase proteins. One-third of the hybrid constructs were functional in a signaling pathway, revealing symmetrical functional roles along the sequence of each protein. The substantially unique surface domains could convert the activating interactions or downstream targets, or both, of each kinase. Constructs were identified with HOG pathway responsiveness, but that signaled to both Hog1 and Fus3 downstream targets. This broadening of substrate specificity was achieved by the substitution of rather small surface domains and substitution of as few as seven amino acids. The implications for MAPK modularity include an improved understanding of the evolutionary origins of signaling pathway specificity. Applying this work to the rather complex array of signaling responsibilities of the p38 and JNK families will provide valuable insight into the evolutionary biology of regulatory networks.

In addition to stress-induced gene expression, the mammalian SAPKs JNK and p38 regulate diverse responses to stress, including apoptosis (102, 103), autophagy (104), ectodomain shedding from transmembrane proteins (105), endocytosis (106), and cell cycle progression (107, 108). p38 mediates responses to TNFα (95) and stimulates the expression of the genes encoding TNFα and other proinflammatory cytokines (109), thus forming part of a positive feedback loop that promotes inflammation. Inhibitors of p38 have been tested as anti-inflammatory drugs in the treatment of rheumatoid arthritis and other inflammatory diseases but have been disappointing due to the lack of long-term clinical efficacy (110). As an alternative, inhibitors of signaling proteins upstream and downstream of p38 in the TNFα pathway are being developed as anti-inflammatory drugs.

Each SAPK cascade plus upstream signaling proteins form highly branched pathways allowing integration of different input signals (111, 112). Analysis of the signaling dynamics of different MAPK and SAPK pathways showed that JNK responds in an all-or-none fashion to input signals (113) and that pulses of extracellular signal–regulated kinase (ERK, a MAPK) activity mediate cell proliferation response to mitogens (114). Both p38 and JNK inhibit ERK activity (115117), a response proposed to contribute to cell cycle arrest in response to stress. Overall, the stress signals and cellular responses are more complex in mammalian SAPK signaling than for Hog1 in yeast. A future challenge will be to use experimental and systems biology approaches to determine how stress signal inputs are integrated and how SAPK dynamics determine stress responses in mammalian cells.

Looking Forward

Here, we focused on the literature related to HOG1 since its discovery, which reveals a gene whose characterization has been broadly influential in the biological sciences. The simple question of how a single-celled eukaryotic cell might sense and adapt to environmental shifts in the osmolarity of growth medium has yielded fundamental discoveries in biology and revealed a signaling network that effectively models similar processes in more complex systems. The ability to genetically manipulate S. cerevisiae was clearly a fundamental requirement for much of the characterization of this pathway, as was the collaborative nature of scientists within the research community. Because of its extensive characterization, the HOG pathway has often been a popular choice for the initial uses of new technologies. The ease of pathway manipulation and the dynamic nature of HOG pathway signaling ensure that this regulatory network will continue to be a focal point in the study of cellular adaptation.

Looking forward, one can anticipate that HOG pathway signaling will continue to surprise and to provide fundamental insight into signal transduction. The key osmostress-sensing components within the cell membrane and cell wall are now known, although their mechanism of sensing turgor loss remains unclear. Revealing this mechanism may occur through collaboration between cell biologists and biophysicists. We note that orthologous genes are known in plants, including an SLN1 homolog from Arabidopsis, which can substitute for the yeast gene in osmosensing (118); however, their role in sensing or influencing adaptation to osmotic stress (dehydration) is not understood (119). Analysis of these genes in plants will shed valuable insight into the evolution of osmoregulation. Although not a central focal point for this Review, the mathematical modeling of cellular adaptation to osmotic stress has become a prevalent topic in the systems biology literature. A number of important systems biology studies are based on the HOG pathway. Despite years of intensive study and a number of valuable discoveries, fundamental questions remain unresolved. We look forward to the surprises that lie ahead.

REFERENCES

Acknowledgments: We thank J. Fritsch of Pepperdine University for his assistance in the generation of the chemical structures used in Fig. 2.
View Abstract

Stay Connected to Science Signaling

Navigate This Article