The budding yeast Saccharomyces cerevisiae can respond to nutritional and environmental stress by implementing a morphogenetic program wherein cells elongate and interconnect, forming pseudohyphal filaments. This growth transition has been studied extensively as a model signaling system with similarity to processes of hyphal development that are linked with virulence in related fungal pathogens. Classic studies have identified core pseudohyphal growth signaling modules in yeast; however, the scope of regulatory networks that control yeast filamentation is broad and incompletely defined. Here, we address the genetic basis of yeast pseudohyphal growth by implementing a systematic analysis of 4909 genes for overexpression phenotypes in a filamentous strain of S. cerevisiae. Our results identify 551 genes conferring exaggerated invasive growth upon overexpression under normal vegetative growth conditions. This cohort includes 79 genes lacking previous phenotypic characterization. Pathway enrichment analysis of the gene set identifies networks mediating mitogen-activated protein kinase (MAPK) signaling and cell cycle progression. In particular, overexpression screening suggests that nuclear export of the osmoresponsive MAPK Hog1p may enhance pseudohyphal growth. The function of nuclear Hog1p is unclear from previous studies, but our analysis using a nuclear-depleted form of Hog1p is consistent with a role for nuclear Hog1p in repressing pseudohyphal growth. Through epistasis and deletion studies, we also identified genetic relationships with the G2 cyclin Clb2p and phenotypes in filamentation induced by S-phase arrest. In sum, this work presents a unique and informative resource toward understanding the breadth of genes and pathways that collectively constitute the molecular basis of filamentation.
THE budding yeast Saccharomyces cerevisiae is dimorphic, exhibiting both a unicellular growth form and a multicellular filamentous state generated presumably as a foraging mechanism under conditions of nutritional stress (Gimeno et al. 1992; Liu et al. 1993; Roberts and Fink 1994; Cook et al. 1996). In S. cerevisiae, nitrogen stress (Gimeno et al. 1992), growth in the presence of short-chain alcohols (Dickinson 1996; Lorenz et al. 2000a), and glucose stress (Cullen and Sprague 2000) can induce the transition to a filamentous form characterized morphologically as follows. Yeast cells undergoing filamentous growth are elongated in shape, due to delayed G2/M progression and prolonged apical growth (Gimeno et al. 1992; Kron et al. 1994; Ahn et al. 1999; Miled et al. 2001). Some reports indicate that these cells bud in a preferentially unipolar fashion (Gimeno et al. 1992; Kron et al. 1994), and, most distinctively during filamentous growth, daughter cells bud from mother cells but remain physically connected after septum formation (Gimeno et al. 1992). As a result, the interconnected cells form filaments that are termed pseudohyphae since they superficially resemble hyphae but lack the structure of a true hyphal tube with parallel-sided walls (Berman and Sudbery 2002). Depending on the induction condition and strain ploidy, pseudohyphal filaments can spread outward from a yeast colony over an agar surface and can also invade the agar (Gancedo 2001). This pseudohyphal growth response is not unique to S. cerevisiae; the related pathogenic fungus Candida albicans also exhibits pseudohyphal and hyphal morphologies, and the ability to switch between yeast, pseudohyphal, and hyphal growth forms is generally considered to be necessary for virulence in C. albicans (Braun and Johnson 1997; Lo et al. 1997; Jayatilake et al. 2006).
Pseudohyphal growth in S. cerevisiae is mediated by at least three well-studied signaling pathways encompassing the mitogen-activated protein kinase (MAPK) Kss1p, the AMP-activated kinase family member Snf1p, and cyclic AMP-dependent protein kinase A (PKA). The filamentous growth MAPK cascade consists of Ste11p, Ste7p, and Kss1p (Liu et al. 1993; Roberts and Fink 1994; Cook et al. 1997; Madhani et al. 1997). Ste11p is a substrate of Ste20p, and Ste20p is itself regulated by the small rho-like GTPase Cdc42p and the GTP-binding protein Ras2p (Mosch et al. 1996; Peter et al. 1996; Leberer et al. 1997). In yeast, PKA consists of the regulatory subunit Bcy1p and one of three catalytic subunits Tpk1p, Tpk2p, or Tpk3p; Tpk2p is required for pseudohyphal growth (Robertson and Fink 1998; Pan and Heitman 1999). The adenylate cyclase Cyr1p is regulated by Ras2p (Minato et al. 1994); thus, Ras2p acts upstream of both the filamentous growth MAPK and PKA pathways. The serine/threonine kinase Snf1p regulates transcriptional changes associated with glucose derepression, mediates several stress responses, and is required for pseudohyphal growth (Cullen and Sprague 2000; Vyas et al. 2003). Snf1p, Kss1p, and Tpk2p regulate the activity of FLO11/MUC1, which encodes a GPI-anchored cell surface flocculin, which is a key downstream effector of pseudohyphal growth (Lo and Dranginis 1998; Rupp et al. 1999; Guo et al. 2000; Kuchin et al. 2002; Pan and Heitman 2002; Karunanithi et al. 2010).
The genetic basis of the yeast pseudohyphal growth response extends well beyond the core signaling modules outlined above (Li and Mitchell 1997; Mosch and Fink 1997; Madhani et al. 1999; Ma et al. 2007a,b; Granek and Magwene 2010; Xu et al. 2010). By transposon-mediated gene disruption of 3627 genes, we have previously identified 309 genes required for pseudohyphal growth in a haploid strain under conditions of butanol induction (Jin et al. 2008). The Boone laboratory has generated genome-wide collections of single gene deletion strains in a filamentous genetic background and has identified 700 genes required for the formation of surface-spread filaments in a diploid strain under conditions of nitrogen stress (Dowell et al. 2010; Ryan et al. 2012). Thus, loss-of-function studies identify a broad set of genes that contribute to the filamentous growth response; however, even these studies are limited in that: (1) essential genes cannot be easily analyzed other than for haploinsufficiency; (2) some deletion phenotypes may be below a threshold that can be easily observed by standard assays; and (3) many phenotypes may be obscured by compensatory buffering effects in mutational analyses that rely on single gene deletions/disruptions. Obviously, no single genetic approach can be expected to yield comprehensive results, and in this light, gene overexpression-based studies have proven to be an effective complement to loss-of-function analyses (Sopko et al. 2006; Douglas et al. 2012). Upon integration with results from the studies above, the analysis of filamentation phenotypes from gene overexpression should identify more completely the genetic scope of pseudohyphal growth.
We present here the first genome-wide overexpression analysis of yeast pseudohyphal growth. For this study, we systematically overexpressed 4909 yeast genes and identified 551 genes that enable pseudohyphal growth under conditions of normal vegetative growth. The data set was analyzed computationally to identify enriched pathways and signaling cascades, highlighting networks mediating MAPK signaling and cell cycle progression. Subsequent studies address a function for nuclear localization of the high osmolarity pathway MAPK Hog1p in repressing pseudohyphal growth, relevant to recent reports that the nuclear localization of Hog1p is not required for osmotolerance. We also identify genetic relationships with the G2 cyclin Clb2p and genes required for filamentation induced by S-phase arrest. Collectively, the work provides a valuable information resource for studies of yeast pseudohyphal growth.
Materials and Methods
Strain and growth conditions
The filamentous strains Y825, Y825/6, and HLY337 used in this study are derived from the Σ1278b genetic background (Gimeno et al. 1992), and all strains are listed in Supporting Information, Table S1. The genotype of haploid Y825 is MATa ura3-52 leu2Δ0; the genotype of diploid Y825/6 is ura3-52/ura3-52 leu2Δ0/leu2Δ0; and the genotype of HLY337 is MATα ura3-52 trp1-1. Gene deletion mutants were generated using PCR-mediated gene disruption with pFA6a-kanMX6 (Longtine et al. 1998) or pUG72 (Gueldener et al. 2002). In addition to synthetic complete (SC) and SC drop-out media (US Biological), media for specific applications are described below.
The overexpression collection and high throughput plasmid transformations
The plasmids utilized for overexpression are high-copy yeast shuttle vectors, with each construct containing a yeast open reading frame (ORF) cloned under transcriptional control of a galactose-inducible promoter. The 3′ end of the open reading frame is fused in-frame with sequence encoding a triple affinity tag of His6, an HA epitope, a protease 3C cleavage site, and the IgG binding domain from protein A. In total, this plasmid collection encompasses 5854 yeast ORFs, including 4973 verified protein-coding ORFs as currently annotated in the Saccharomyces Genome Database (www.yeastgenome.org). It should be noted that the affinity tags may perturb protein folding at the carboxy terminus of some of the gene products, but we expect that the majority of genes in this collection (∼80–90%) should encode fully functional proteins, extrapolating from large-scale protein localization and affinity purification studies (Gavin et al. 2002; Ho et al. 2002; Kumar et al. 2002a; Ghaemmaghami et al. 2003; Huh et al. 2003; Bharucha et al. 2008). To generate overexpression strains for phenotypic analysis of filamentous growth, we introduced the plasmids individually in 96-well format into a diploid strain of the filamentous Σ1278b genetic background by a modified form of lithium acetate-mediated transformation as described (Kumar et al. 2000, 2002b; Ma et al. 2007a,b; Bharucha et al. 2008; Jin et al. 2008). All transformants were selected on SC −Ura, and glycerol stock solutions (15% glycerol) were prepared. In total, we performed 6894 plasmid preparations and yeast transformations to generate a collection of 5854 strains (∼85% efficiency).
By design of the overexpression vectors, galactose induction was used to regulate transcription of the plasmid-based target genes as follows. Yeast strains were sequentially cultured in a 30° shaking incubator in nitrogen-sufficient minimal liquid media containing glucose, raffinose, and galactose for 2–3 days, overnight, and 6 hr, respectively. Minimal liquid media consisted of 0.67% yeast nitrogen base (YNB) without amino acids and ammonium sulfate (Difco), 2% carbon source (glucose, raffinose, or galactose), 5 mM ammonium sulfate (nitrogen sufficiency), and additional amino acids to correct for auxotrophies as necessary. Following galactose induction for 6 hr, yeast cultures were spotted using a multichannel pipette onto agar plates consisting of 2% galactose, 0.67% YNB without amino acids and ammonium sulfate (Difco), 5 mM ammonium sulfate, and additional amino acids to correct for auxotrophies. Galactose induction typically drives gene expression to levels 1000-fold of those observed in the presence of glucose (St John and Davis 1981), and we estimate similar levels of inducible expression here by Western blotting (Figure S1).
The presence of galactose in the medium significantly diminished the degree of observed surface-spread filamentation for all strains, including control wild-type strains under conditions of low nitrogen, consistent with results reported in Lorenz et al. (2000b). Strong levels of invasive filamentation, however, were still observed, and we used this filamentation phenotype as an indicator of pseudohyphal growth. Invasive growth was assessed by a standard plate-washing assay as follows. Spotted cultures were incubated at 30° for 7 days and photographed before plate washing. Agar plates were rinsed with a gentle stream of water to remove noninvasive cells, and the remaining cells were photographed. Invasive clones were recorded and rescreened with the same protocol in a second pass. The degree of invasive growth was quantified by the pixel intensity ratio of spotted cultures pre- and postwashing.
Quantification of screen results
The level of invasiveness for each clone in the second pass was quantitatively measured using the integrated density feature of ImageJ. Images of individual clones pre- and postplate washing were analyzed after background subtraction. Scores indicate the ratios of post- to prewash pixel intensity for each indicated clone.
For galactose-independent overexpression, yeast ORFs with 1 kb upstream sequence and 300 bp downstream sequence were cloned into pRS426 (Sikorski and Hieter 1989) using standard restriction enzyme digestion and ligation techniques.
Plasmids pFA6a-GFP(S65T)-CAAX-kanMX6 and pFA6a-GFP(S65T)-CAAX-HIS3MX6 carrying GFP-CAAX modules were modified from pFA6a-GFP(S65T)-kanMX6 and pFA6a-GFP(S65T)-HIS3MX6 (Longtine et al. 1998). The Pac1 and Asc1 restriction sites of these plasmids were used to replace the GFP module with a GFP-CAAX module, generated by PCR amplification using a 3′ primer encoding the nine carboxy-terminal residues of the budding yeast Ras2p CAAX box (Westfall et al. 2008). The following forward and reverse primers were used: HOG1_RAS_F1: CGGTAACCAGGCCATACAGTACGCTAATGAGTTCCAACAGCGGATCCCCGGGTTAATTAA and HOG1_RAS_R1: TCTTTTTTTTTTTGTTTCCTCTATACAACTATATAC GTAAGAATTCGAGCTCGTTTAAAC. All plasmids are available upon request.
Y825/6 strains with overexpression vectors (pRS426 carrying yeast ORFs with 1 kb upstream sequence and 300 bp downstream sequence) were streaked on synthetic low ammonium dextrose (SLAD) plates (2% glucose, 0.67% YNB without amino acids and ammonium sulfate, 50 μM ammonium sulfate, and supplemental leucine to correct for auxotrophy). Plates were incubated 5 days at 30° prior to imaging.
Verification of overexpression by Western blotting
Selected moveable open reading frame (MORF) strains of the Y825/6 background were cultured in 5 ml SC −Ura media overnight at 30°, followed by back dilution into nitrogen sufficient minimal media with galactose for 4 hr. Following protein extraction, SDS–PAGE resolution and transfer to nitrocellulose membrane by standard procedures, membranes were incubated with protein A antibody against the tandem affinity purification (TAP) tag in a 1:10,000 dilution. Blots were developed using SuperSignal West Dura Extended Duration Substrate (Thermo Scientific).
Identifying network modules and signaling cascades by computational analysis
The gene set identified from this overexpression screen was submitted to the functional analysis tool DAVID (Huang Da et al. 2009) to identify enriched Kyoto Encyclopedia of Genes and Genomes (KEGG)-annotated pathways. Since the overexpression screen was genome-wide in scope, the default background set was used. The most enriched pathways, mediating cell cycle progression (sce04111), meiosis (sce04113), and MAPK signaling (sce04011), were selected for further analysis. The KEGG.xml files of these pathway maps were downloaded and parsed using an in-house program that generates nodes and edge lists. The KGML pathway .xml (kgml) file consists of “entry” tags, which can be represented as nodes, and “relation” tags, which can be represented as edges of a network. The entry tags consist of genes, compounds, and complexes, while the relation tags include manually curated molecular interactions, reaction networks, genetic and environmental information processing, and cellular networks (Kanehisa et al. 2012). Cytoscape (Killcoyne et al. 2009), which constructs a ball-and-stick representation of a network using an edge list, was used to visualize the network. Since the three pathways possess overlapping gene sets, the edge lists were concatenated and visualized as one single network.
Cell cycle analysis
For CLB2 epistasis analysis, a sampling of 10 genes was selected that fulfilled the following criteria: (1) genes that had been identified in this overexpression screen as yielding filamentous growth under noninducing conditions of nutrient sufficiency; and (2) genes that had either not been placed in a clear signaling pathway or that functioned in a pathway with unclear upstream components. The sampled gene set was not intended to be comprehensive, but rather served as a probe for additional genes that may contribute to filamentation through mechanisms that impact CLB2 and cell cycle progression. Homozygous diploid double deletion mutants were constructed for this analysis, and the mutant strains were streaked onto SC agar plates, incubated overnight at 30°, and then photographed. For surface-spread filamentation using hydroxyurea treatment, hydroxyurea was added to SC agar plates to a final concentration of 100 mM (Kang and Jiang 2005). Strains were photographed following overnight incubation at 30°.
Invasive growth analysis of strains with modified HOG1 alleles
Yeast strains of the Y825 genetic background containing integrated HOG1-GFP, HOG1-GFP-CAAX, and hog1Δ alleles were assayed for invasive growth by plating spotted cultures on SLAD plates (2% glucose, 0.67% YNB without amino acids and ammonium sulfate, 50 μM ammonium sulfate, and supplemental amino acids to correct for auxotrophies) with 1% (vol/vol) butanol. Plates were sealed in parafilm and incubated for 7 days at 30°. Strains were assayed for invasive growth by rinsing with water and rubbing away nonadherent cells (Cullen and Sprague 2000). Spotted cultures were photographed pre- and postwash, and invasiveness was measured using the integrated density feature of ImageJ.
The FRE-lacZ reporter construct (Madhani and Fink 1997) was used to measure filamentous growth signaling using the Yeast β-Galactosidase Assay kit (Thermo Scientific) according to protocols described previously (Ma et al. 2008; Xu et al. 2010).
Fluorescence images were taken using a DeltaVision-RT Live Cell Imaging system (Applied Precision). Image capture was conducted using Applied Precision’s SoftWorx imaging software.
Generating a mutant collection for genome-wide overexpression analysis
Standard laboratory strains of S. cerevisiae (e.g., derivatives of S288c) are nonfilamentous and, consequently, inappropriate for studies of pseudohyphal growth. The Σ1278b strain has emerged as the preferred background for studies of filamentation, since it undergoes a significant and easily controlled transition to filamentous growth (Grenson 1966; Gimeno et al. 1992); however, no genome-wide mutant collections suitable for this study have been generated previously in the Σ1278b background (Coelho et al. 2000). Here, we sought to construct an extensive reagent base for overexpression studies of pseudohyphal growth, generating a collection of yeast strains in Σ1278b with each mutant carrying a single plasmid enabling galactose-inducible gene overexpression. For this purpose, we utilized the plasmid collection constructed in Gelperin et al. (2005) and introduced the plasmids individually by transformation into diploid yeast. Of the 5854 plasmids encompassed in this overexpression collection, we identified 4909 clones that: (1) contained an ORF corresponding to an annotated and verified yeast gene and (2) allowed for sufficient cell growth upon galactose induction in the Σ1278b background such that invasive growth assays could be performed.
The design of the phenotypic screen is outlined in Figure 1A and detailed in Materials and Methods. In brief, we drove gene overexpression by growth in galactose under conditions of nitrogen sufficiency; this approach identifies genes that upon overexpression can enable pseudohyphal growth in the absence of stimuli capable of inducing filamentation. To ensure that gene overexpression was efficient, we analyzed resulting protein levels by Western blotting for a sampling of seven strains in the constructed mutant collection (Figure S1). Filamentation was assessed by invasive growth analysis, as surface filamentation is lessened in the presence of galactose (Lorenz et al. 2000b). To confirm that invasive filamentation was indeed an effective indicator of diploid pseudohyphal growth, we cloned five genes along with native promoters into a high-copy yeast shuttle vector such that gene overexpression phenotypes could be measured without effects from galactose induction. Strains carrying the high-copy number vector clones yielded surface-spread filamentation phenotypes matching the corresponding invasive growth phenotypes observed in the screen (Figure S2). In addition, a positive control consisting of galactose-induced overexpression of eight genes known to affect pseudohyphal growth yielded exaggerated invasive phenotypes as shown in Figure 1B. The phenotypic difference between a wild-type strain and the indicated overexpression mutants is clear and establishes an easily identifiable threshold for positive results.
A collection of yeast genes capable of inducing pseudohyphal growth
By the systematic genome-wide overexpression analysis described above, we identified 551 genes that resulted in invasive filamentation upon galactose induction under conditions of nitrogen sufficiency (Figure 2A); the full gene list is provided in Figure S3. This gene set is comparable in size to the complement of genes that yield pseudohyphal growth defects upon gene deletion under conditions of butanol induction (Jin et al. 2008) and nitrogen deprivation (Ryan et al. 2012). The individual genes, however, vary between these gene sets, and the distinctions between overexpression-based screens vs. loss-of-function screens are presented in Discussion.
By simple Gene Ontology (GO) term analysis, the invasive growth overexpression gene set was not enriched for any molecular functions or protein-associated subcellular components; however, we did identify several enriched biological process terms (Figure 2B). Genes annotated as contributing to the regulation of metabolic processes associated with nitrogenous compounds were enriched in the overexpression gene set (P-value of 8.2 × 10−7). This is not surprising since nitrogen availability is an important regulator of pseudohyphal growth. This GO term is broad in scope; among the associated genes are several known pseudohyphal growth regulators, including the STE12 and TEC1 genes that collectively encode a transcriptional complex acting downstream of Kss1p to activate gene promoters with filamentation-and-invasion response elements (FREs) (Madhani and Fink 1997). Genes involved in the cellular response to nutrient levels were also enriched in the results from our screen, as were overlapping gene sets associated with cytoskeletal organization and spindle pole body organization. The nutrient-responsive gene set encompasses the SNF1 kinase gene, which plays an established role in regulating pseudohyphal growth (Kuchin et al. 2002). Interestingly, a large cohort of 79 functionally uncharacterized genes that lack a standard gene name indicative of function was identified in this screen. Mutant alleles of these genes lack extensive phenotypic characterization, and, for many of the indicated genes, the overexpression studies presented here offer initial insight into the regulatory consequences of increased transcription, particularly in a filamentous genetic background.
To consider the possibility that genes mediating specific cellular processes may affect pseudohyphal growth to differing degrees, we analyzed our quantified screening results for GO term enrichment within genes grouped by the intensity of observed overexpression-induced invasive growth (Figure 2D). The degree of invasive growth was estimated by the pixel intensity postwash to prewash of each spotted overexpression culture. The pixel intensities were binned into categories (ranging from 0.94 to 0.99 and above), and the associated genes within each grouping were assessed for enrichment of GO terms. By this analysis, genes annotated as being associated with M phase of the meiotic cell cycle (GO:0051327) were enriched in the gene set that yielded the strongest level of invasive growth upon overexpression (postwash:prewash pixel intensity greater than 0.99).
Signaling pathways that regulate pseudohyphal growth
While the GO term analysis above provides broad indications of cellular processes contributing to pseudohyphal growth, we also sought to identify specific pathways that affect filamentation by searching for KEGG signaling pathways overrepresented in the overexpression screen results. KEGG is an online database that provides annotated and manually drawn signaling pathway maps for a broad range of eukaryotes (www.genome.jp/kegg/). For our purposes, KEGG provides the largest set of yeast pathway annotations.
To identify enriched KEGG-annotated signaling pathways in our overexpression data set, we implemented a computational approach utilizing the functional annotation tool DAVID. By this analysis, we found pathways controlling cell cycle progression (sce04111), meiosis (sce04113), and MAPK signaling (sce04011) to be the most highly enriched in the overexpression data (Figure 3A). As these pathways encompass overlapping gene sets, we constructed network connectivity maps to better visualize the signaling modules (Figure 3B). Genes identified in the overexpression screen were used as core “seeds” along with other annotated components of the three pathways. The pathways were parsed using an in-house program and subsequently reassembled into a network using Cytoscape (Materials and Methods). Connections are visualized as ball-and-stick representations in Figure 3B. From this analysis, gene sets exhibiting genetic and/or physical interactions with components of the KEGG-annotated cell cycle and meiosis pathways were densely overlapping; this is not surprising since progression through the cell cycle and meiosis are obviously related processes. Genes exhibiting connections with MAPK signaling pathways share extensive connectivity with genes associated with meiosis and cell cycle progression. Interestingly, strictly from connections reported in the KEGG resources, the Kss1p and osmosensing Hog1p MAPK cascades link the MAPK signaling connectivity map with larger gene sets associated with cell cycle progression and meiosis (Figure 3B, inset).
Thus, from this analysis we identified core networks enriched in the data set mediating (1) MAPK signaling and (2) cell cycle progression/meiosis. Consequently, in the following sets of experiments we further investigated (1) the role of nuclear Hog1p in regulating pseudohyphal growth and (2) the genetic basis of pseudohyphal growth phenotypes resulting from altered cell cycle progression.
Hog1p-mediated repression of pseudohyphal growth
Genes annotated as contributing to MAPK signaling were enriched in the results of our overexpression screen; in particular, the screen identified several genes known to regulate the activity of Hog1p. The Hog1p kinase is a MAPK best studied for its role in producing glycerol as a compensatory osmolyte in response to increased levels of extracellular osmolarity (Kultz and Burg 1998; Westfall et al. 2008); however, Hog1p is also known to repress pseudohyphal growth in the absence of filamentation-inducing stimuli (O’Rourke and Herskowitz 1998; Pitoniak et al. 2009). A simplified representation of the Hog1p pathway is presented in Figure 4A. In yeast, high extracellular osmolarity stimulates two putative osmosensors, Sho1p and Sln1p (Posas and Saito 1997). Sho1p activates the P21-activated kinase family member Ste20p, which in turn activates a cascade of the MAPKKK Ste11p, the MAPKK Pbs2p, and Hog1p (Raitt et al. 2000). Sln1p, Ypd1p, and Ssk1p are components of a phosphorelay signaling system that activates the partially redundant kinases Ssk1p and Ssk22p upon osmostress; these kinases in turn activate Pbs2p, resulting in activation of Hog1p (Brewster et al. 1993; Maeda et al. 1994; Posas et al. 1996). Upon activation, Hog1p is rapidly translocated to the nucleus through a process that requires the importin-β family member Nmd5p (Ferrigno et al. 1998). Nuclear Hog1p has been identified in complexes at hundreds of promoters and genes, influencing chromatin remodeling and transcription (O’Rourke and Herskowitz 2004; Pokholok et al. 2006; Zapater et al. 2007). Subsequently, Hog1p is largely dephosphorylated by the phosphatases Ptc1p, Ptc2p, Ptc3p, Ptp2p, and Ptp3p (Robinson et al. 1994; Wurgler-Murphy et al. 1997; Mattison et al. 1999). Dephosphorylated Hog1p is exported into the cytosol through interaction with the karyopherin Crm1p (Ferrigno et al. 1998).
Interestingly, three genes involved in the nuclear export of Hog1p (NBP2, PTP2, and CRM1) were identified in the overexpression screen as enabling invasive growth under conditions of nitrogen sufficiency. To further consider the possibility that the nuclear export of Hog1p promotes pseudohyphal growth, we cloned the genes and promoters for CRM1, NBP2, PTC3, PTP2, and PTP3 into a high-copy vector, enabling analysis of overexpression phenotypes without galactose induction. Each of these genes contributes to the dephosphorylation and nuclear export of Hog1p; NBP2 is not illustrated in Figure 4A, but it recruits Ptc1p to the Pbs2p-Hog1p complex (Mapes and Ota 2004). We introduced these plasmids into a diploid strain of the filamentous Σ1278b genetic background and assayed for surface-spread filamentation under conditions of nitrogen limitation. In each case, the strains exhibited hyperactive surface filamentation relative to a wild-type strain carrying an empty vector control (Figure 4B).
Classically, the nuclear form of Hog1p had been thought to mediate osmotolerance; however, Westfall et al. (2008) reported that cells lacking NMD5 and/or cells with a plasma membrane-tethered form of Hog1p survive hyperosmotic stress. This raises an interesting question regarding the functional contributions of nuclear Hog1p. Considering the overexpression results above, one function of nuclear Hog1p may be to repress pseudohyphal growth in a filamentation-competent strain of S. cerevisiae, although it should be noted that other pathways will also be affected by overexpression of genes such as CRM1 and NMD5.
To investigate more directly the effect of spatial compartmentalization on Hog1p function, we constructed a haploid yeast strain in the Σ1278b background wherein endogenous HOG1 was fused at its 3′ end to sequence encoding GFP and the nine C-terminal residues of Ras2p (designated CCAAXRas2p) as in Westfall et al. (2008). By virtue of S-palmitoylation and S-farnesylation of the cysteine residues in the appended Ras2p carboxy-terminal tail, the translated Hog1p-GFP-CCAAXRas2p chimera should localize at the plasma membrane, and we did observe concentrated fluorescence at the cell periphery in this strain (Figure 4C) relative to an otherwise isogenic strain containing an integrated HOG1-GFP allele. It should be noted that some Hog1p-GFP-CCAAXRas2p may be present at the nuclear membrane, although we did not observe any nuclear Hog1p chimera by fluorescence microscopy. Under pseudohyphal growth-inducing conditions of nitrogen stress and butanol treatment, the strain containing the Hog1p-GFP-CCAAXRas2p chimera showed slightly exaggerated invasive growth relative to a strain containing Hog1p-GFP, with invasive growth levels comparable to those observed in a hog1Δ strain. Under conditions of nitrogen sufficiency, the mutant strain containing Hog1p-GFP-CCAAXRas2p is hyperfilamentous with respect to a strain containing Hog1p-GFP, although not quite to the level of a hog1Δ strain; the degree of filamentous growth activity is measured in Figure 4C using a filamentation MAPK Kss1p pathway-specific FRE-lacZ reporter.
If nuclear Hog1p does repress pseudohyphal growth, filamentation should not be observed upon activation of the Hog1p pathway by high osmolarity. Under conditions of nitrogen stress and high salt, inducing both pseudohyphal growth and the Hog1p pathway, we find that a wild-type diploid strain of the filamentous Σ1278b background grows poorly and shows no signs of filament formation. Collectively, these findings are consistent with the notion that nuclear Hog1p contributes to the repression of pseudohyphal growth.
Genes contributing to exaggerated filamentous phenotypes from prolonged apical growth
Yeast cells undergo a switch from isotropic to apical growth upon progression through Start and a subsequent return to isotropic growth upon transition through G2/M (Hartwell et al. 1970; Lew and Reed 1993; Chant and Pringle 1995; Pringle et al. 1995). Genetic perturbations and/or chemical treatments that delay the G2/M transition in the filamentous Σ1278b strain result in a prolonged period of apical growth, as well as increased unipolar budding and decreased cell separation (Sheu et al. 2000; Miled et al. 2001; Rua et al. 2001). In yeast, the mitotic cyclins Clb1p and Clb2p antagonize polarized growth and are key in the molecular events underlying the onset of mitosis (Fitch et al. 1992; Booher et al. 1993), and a homozygous diploid clb2Δ/Δ strain is hyperfilamentous. Considering the importance of genes that regulate the apical–isotropic transition in affecting pseudohyphal growth phenotypes, we sought to determine if additional genes identified in our screen contributed to the hyperfilamentous phenotype of a clb2Δ mutant. We selected a sampling of 10 genes identified in our screen with unclear pathway designations and/or unclear roles in promoting pseudohyphal growth (HMS1, HMS2, MGA1, MSB2, MSN1, NPR1, PTP3, SNF1, YAK1, and YCK1) and generated homozygous diploid deletions of these genes in the clb2Δ/Δ background for phenotypic analysis; results are shown in Table 1 and Figure S4A. Notably, under conditions of nitrogen sufficiency, the majority of double mutants yielded phenotypes mirroring clb2Δ/Δ; however, the clb2Δ/Δmsb2Δ/Δ mutant exhibited a reduction in surface-spread filamentation relative to the clb2Δ/Δ parent. Increased apical growth resulting from hydroxyurea-induced cell cycle arrest in S phase has also been shown to drive surface-spread filamentation (Lorenz and Heitman 1998; Kang and Jiang 2005). Consequently, we also tested homozygous diploid single deletions of the same 10 genes indicated above for the absence of surface-spread filamentation upon hydroxyurea treatment. As indicated in Table 1 and Figure S4B, the mga1Δ/Δ, msn1Δ/Δ, ptp3Δ/Δ, and msb2Δ/Δ strains exhibited decreased surface filamentation in response to hydroxyurea treatment relative to the wild-type parent. From these results, the msb2Δ/Δ strain exhibited the most significant decrease in hydroxyurea-induced filamentation of the strains tested. Collectively, these studies highlight the contribution of Msb2p under genetic perturbations and chemical treatments that induce filamentation through prolonged apical growth.
Here we implemented a systematic and genome-wide analysis of yeast invasive filamentation induced by gene overexpression. Interestingly, as compared against the results from large-scale deletion/disruption screens, systematic overexpression screens typically identify overlapping, but decidedly nonredundant, data sets (Sopko et al. 2006). We observe similar results here. In this screen, we identified 61 genes that were also reported in the targeted gene deletion screen by Ryan et al. (2012) and 79 genes that yielded pseudohyphal growth defects in a previous transposon-based disruption screen (Jin et al. 2008); a full listing of these overlapping genes is provided in Figure S5. Comparisons between these data sets are inexact, however, since: (1) filamentation phenotypes were assayed slightly differently in each screen; (2) the transposon-based study was smaller in scope, encompassing ∼60% of the annotated yeast gene complement; (3) butanol treatment as opposed to nitrogen stress was used to induce filamentation in the transposon mutagenesis study; and (4) a haploid strain was used for transposon mutagenesis, while we used diploid cells for this overexpression screen. The partial overlap between loss-of-function and overexpression results likely stems from the fact that many genes can be required for a given cell process without being sufficient to induce that process upon overexpression. Consequently, we expected to identify a greater number of regulatory genes by this overexpression screen, and we did identify many such genes, including several that regulate cell cycle progression and MAPK signaling. However, we did not observe any statistically significant enrichment for transcription factors, kinases, and/or nutrient sensors in the data set, and the set of gene hits from this overexpression screen that overlapped the genes identified by targeted deletion and transposon-based loss-of-function screening was not significantly enriched for any GO terms. This overlapped gene set does encompass several key pseudohyphal growth genes, including STE12, TEC1, SNF1, and SHO1.
In interpreting the results from this study, it is important to bear in mind two caveats. First, 4909 genes (of 4973 verified ORFs) were analyzed by overexpression; thus, we do not consider the screen to be comprehensive, although it is the largest overexpression-based screen of pseudohyphal growth to date. Second, the plasmid library used in this study is a gene fusion library, and for a subset of the genes tested, the carboxy-terminal modification may result in dominant effects that can confound the interpretation of results. It is difficult to estimate the degree of this effect, but previous studies indicate that ∼97% of the cloned gene products do encode full-length proteins (Gelperin et al. 2005), which may mitigate concerns regarding phenotypes from truncated proteins.
The gene set identified in this study appears large at first glance. However, systematic deletion studies in haploid and diploid strains of the Σ1278b background have identified comparably large sets of genes yielding pseudohyphal growth phenotypes. Our previous transposon-based disruption screen of 3627 genes identified 309 that were required for butanol-induced surface filamentation. Similarly, our previous smaller-scale overexpression screen identified 199 genes of 2043 tested that yielded exaggerated pseudohyphal growth under conditions of butanol induction (Jin et al. 2008). Extrapolating these results to the genome as a whole, we arrive again at comparably sized data sets. From these systematic disruption and overexpression screens, it is clear that many cellular processes need to occur effectively for surface filaments to appear. Cell cycle progression, cell budding, polarized growth, cytoskeletal organization, nutrient sensing/responses, and numerous metabolic/biosynthetic processes all contribute to, and are required for, the formation of extensive surface filaments. The pseudohyphal growth response represents an integrative output, the magnitude of which is modulated by a diverse complement of signaling pathways and genetic networks. In sum, the complexity and scope of the genetic machinery underlying yeast pseudohyphal growth makes it an ideal subject for genomic analysis.
Several genes that impacted the timing of G2/M progression in yeast were capable of inducing invasive growth upon overexpression. The results presented here are consistent with the importance of enhanced apical growth in establishing, at minimum, morphological phenotypes resembling those observed during pseudohyphal growth. We induced prolonged apical growth by genetic means (clb2Δ) and chemical treatment (hydroxyurea). Both sets of results highlight contributions from Msb2p, a mucin family member that promotes activation of the MAPK Kss1p while also serving as an osmosensor for the HOG pathway; msb2Δ strains have been shown previously to exhibit decreased expression of a FRE-lacZ reporter and decreased filamentation (Cullen et al. 2004; Chavel et al. 2010). Interestingly, Msb2p was identified as a potential Cdc28p substrate through kinase assays using an analog-sensitive allele of Cdk1-Clb2p and lysate from a strain containing an Msb2p-GST fusion (Ubersax et al. 2003). With respect to the hydroxyurea-based results, Kang and Jiang (2005) previously screened the yeast deletion collection in a nonfilamentous genetic background for loss of what they termed to be semifilamentous growth induced by hydroxyurea treatment, identifying 16 genes that were required for the process. Each of those 16 genes was also required for hydroxyurea-induced filamentous growth in the filamentous Σ1278b background, and we found 4 of 10 genes independently selected in our study that yielded hydroxyurea-induced filamentation defects. Thus, a broader genome-wide screen in the Σ1278b background would likely reveal a large set of genes that contribute to this response, although some genetic networks required for nitrogen stress-induced pseudohyphal growth may not be required. It would further be interesting to determine if the filamentous response to hydroxyurea stems strictly from the S phase arrest or if genetic networks independent of cell cycle regulation also affect the observed filamentation.
MAPK signaling pathways are key pseudohyphal growth regulators, and the scope of genes identified in this screen that affect filamentation by impacting a MAPK signaling pathway may be large, as suggested from the network analysis presented here. In particular, our data encompass the known MAPK regulators/effectors SHO1, MSB2, PTP3, STE12, and TEC1, and 17 genes identified in this overexpression screen also exhibited increased mRNA transcript abundance upon induction of MAPK pathway activity (Roberts et al. 2000).
With respect to understanding MAPK cascade activity, considerable research efforts have been expended to consider the mechanisms ensuring MAPK signaling specificity during pseudohyphal growth (Bao et al. 2004; Maleri et al. 2004; O’Rourke and Herskowitz 2004; Hao et al. 2008). In particular, the Hog1p MAPK pathway is known to inhibit pseudohyphal growth in the absence of filamentous growth-inducing stimuli, and HOG1 deletion mutants exhibit hyperactive surface filamentation under nutrient-rich conditions (O’Rourke and Herskowitz 1998). Interestingly, several genes that promote nuclear export of the osmoregulatory MAPK Hog1p, induced invasive growth upon overexpression. The nuclear-localized form of Hog1p is generally thought to mediate the hyperosmotic response and presumably also represses pseudohyphal growth. This notion, however, has been called into question, as Westfall et al. (2008) have reported that a yeast strain containing an allele of hog1 encoding a plasma membrane-tethered form of the protein is still resistant to hyperosomotic stress. If nuclear-localized Hog1p is not required for resistance to hyperosmotic stress, the possibility exists that one function of nuclear Hog1p may be to represses pseudohyphal growth. As presented here, our overexpression screen results are consistent with this possibility, and results using a plasma membrane-tethered form of Hog1p are also consistent with this model. Identification of the nuclear targets of Hog1p that mediate a repressive effect remains a future goal. The transcription factor Tec1p is a strong candidate, as Shock et al. (2009) propose that Hog1p prevents Tec1p binding to DNA.
The importance of Hog1p signaling in regulating pseudohyphal growth is not limited to S. cerevisiae. In C. albicans, Hog1 is involved in oxidative stress responses, osmotic stress responses, and in cell wall biosynthesis, and functional Hog1 represses the yeast-to-hyphal transition (San Jose et al. 1996; Alonso-Monge et al. 1999, 2003; Enjalbert et al. 2006). Notably, mutants defective for Hog1p function display reduced virulence in mice and increased susceptibility to phagocytic cells (Gonzalez-Parraga et al. 2010; Cheetham et al. 2011). The Hog1 MAPK cascade in C. albicans encompasses the MAPKK Pbs2 and the MAPKKK Ssk2 (Cheetham et al. 2007), although additional upstream and downstream regulators have not been elucidated as extensively as in S. cerevisiae. As this Candida network becomes more clearly delineated, it will be interesting to determine if the emerging model of Hog1p-mediated regulation of filamentous growth in S. cerevisiae is borne out and even further, the degree to which corresponding filamentous growth regulatory networks in baker’s yeast contribute to hyphal development and virulence in C. albicans.
In sum, we present here the first systematic overexpression screen for genes capable of inducing yeast pseudohyphal growth. Our results provide overexpression phenotypes for a large number of genes with uncharacterized function, and it is tempting to speculate that some of these genes may exhibit functions in the filamentous Σ1278b background but not in nonfilamentous strains. We further identified several signaling networks that modulate pseudohyphal growth levels upon overexpression-based perturbation, and collectively, the work presents a significantly enhanced foundation for the mapping of genetic relationships within the pseudohyphal growth regulatory network.
We thank Damian J. Krysan, Daniel J. Klionsky, William Stanaforth-Donahue, and Paul J. Cullen for reagents and/or helpful discussions regarding the manuscript. We thank Lois Weisman for the use of her fluorescence microscope. This work was supported by grants 1R01-A1098450-01A1 from the National Institutes of Health and 1-FY11-403 from the March of Dimes (to A.K.).
Communicating editor: D. Voytas
- Received November 21, 2012.
- Accepted February 5, 2013.
- Copyright © 2013 by the Genetics Society of America