Genome-wide analysis of the response to protein glycosylation deficiency inyeast

Abstract

Protein modification by glycosylation occurs through an essential biochemical pathway that produces mannosyl side chain substrates, which are covalently attached to proteins in the endoplasmic reticulum. We used DNA microarray analysis to characterize the cellular response to a conditional defect (pmi40-101) in the protein glycosylation pathway. Expression profiles were obtained from DNA microarrays containing essentially every gene from Saccharomyces cerevisiae. We validated the microarray analysis by examining the expression patterns of induced genes using transcriptional lacZ fusions. The major class of genes differentially expressed in the glycosylation mutant overlapped significantly with that of a starvation response and included those required for gluconeogenesis, the tricarboxylic acid and glyoxylate cycles, and protein and amino acid biosynthesis. Two mitogen-activated protein (MAP) kinase pathways were also activated in the mutant, the filamentous growth and protein kinase C pathways. Taken together, our results suggest that a checkpoint is activated in response to a protein glycosylation defect, allowing the cell to mount an adaptive response by the activation of multiple MAP kinase pathways.

Introduction

Glycosylation of proteins by mannose moiety side chain addition is a posttranslational modification that occurs in the endoplasmic reticulum and Golgi. The pathway responsible for producing these side chains and transferring them to target proteins is conserved throughout eukaryotes. Protein glycosylation is important for protein function (Spiro, 2002), protein folding (Ng et al., 2000), regulated secretion (Scheiffele and Fullekrug, 2000), and protein localization (Ohnishi et al., 2003; Watanabe et al., 2004). In addition, carbohydrate moieties of glycoproteins are structurally diverse, imparting specific biological information and regulation onto their cognate proteins.

A change in the rate of protein glycosylation causes several characterized responses, the best studied of which is the unfolded protein response (UPR; Chapman et al., 1998). In addition, defects in protein glycosylation influence secretion (Travers et al., 2000), cell integrity (Lee and Elion, 1999; Cullen et al., 2000), and the activation of cellular signaling pathways (Cullen et al., 2000). How a cell responds to a change in the glycosylation rate is not fully understood, but presumably the response results from communication between internal cellular signaling networks. Several genomic studies have been employed to characterize the response of cells to abrogation of one aspect of protein glycosylation, N-linked glycosylation, using the inhibitor tunicamycin (Travers et al., 2000; Lecca et al., 2004) or analyzing mutants partially defective for N-linked glycosylation (Klebl et al., 2001).

We previously isolated partial loss-of-function mutants defective in protein glycosylation in a genetic screen for altered expression of a mitogen-activated protein (MAP) kinase pathway-dependent reporter (Cullen et al., 2000). We also demonstrated that in such mutants activation of the MAP kinase pathway is required for viability, indicating that a defect in glycosylation leads to a compensatory cellular response (Cullen et al., 2000). One of the mutants we identified was defective in the PMI40 gene, which is required for the first committed step of protein glycosylation, the conversion of fructose 6-phosphate to mannose 6-phosphate (Smith et al., 1992). In the present study, we confirm and extend these studies by examining the glycosylation defect using a partial loss-of-function allele of PMI40, called pmi40-101 (Cullen et al., 2000). Expression profiling generated from this study demonstrates that genes encoding structural proteins of two MAP kinase pathways are induced in response to a glycosylation defect. Our data also show that cells defective for glycosylation exhibit a starvation-like response and show growth inhibition. Taken together, our results provide evidence for a checkpoint that is sensitive to the changes in the flux of protein glycosylation in the cell.

Materials and methods

Strains, plasmids, and microbiological techniques

The yeast strains used in this study are listed in Table 1. Integrated green fluorescent protein (GFP) fusions were made by PCR-based methods using plasmids provided by J. Pringle (University of North Carolina; Longtine et al., 1998). Gene disruptions and integrated promoter and protein fusions were confirmed by PCR analysis and by phenotype. GFP-tagged fusion proteins used in this study were functional as assessed by FUS1 reporter expression, invasive growth phenotypes, and viability of the pmi40-101 mutant. Yeast and bacterial strains were propagated using standard methods (Rose et al., 1990). Yeast peptone dextrose (YPD) and synthetic complete dextrose (SCD) media have been described (Rose et al., 1990). Yeast transformations were performed as described (Gietz et al., 1995). Bacterial transformations, bacterial DNA preparations and plasmid constructions were performed by standard methods (Sambrook et al., 1989). Strains harboring the pmi40-101 allele were maintained on medium containing 50 mM mannose unless otherwise stated.

Table 1.

Yeast strains used in this study

Strain	Genotype	Reference
1436	Wild type	Cullen et al. (2000)
1436–13	ste4 pmi40-101	Cullen et al. (2000)
1004	ste4 pmi40-101 ssk1 HOG1–GFP::KanMX6	This study
1006	Σ1278b ssk1 HOG1–GFP::KanMX6	This study

Inductions and RNA isolation

Strain pmi40-101 (1436-13) was grown to stationary phase in YPD medium plus 50 mM mannose. Cells were harvested by centrifugation, washed twice in YPD, and resuspended in YPD (OD 0.5) or YPD plus 50 mM mannose (OD 0.2) medium prewarmed to 30 °C. Cells were incubated at 30 °C until the culture reached OD 0.8 (c. 4 h), at which point cells were harvested by centrifugation and stored at −80 °C. RNA was extracted from cells using hot acid phenol extraction followed by passage over an RNeasy Maxi column (Qiagen, Valencia, CA). Exogenous mannose is not thought to influence the growth rate of wild-type cells (Pitkanen et al., 2004).

Expression analysis

Expression profiles were obtained from microarray analysis performed at the Fred Hutchinson Cancer Research Center. Microarray construction, target labeling and hybridization protocols were performed as described (Fazzio et al., 2001). Microarrays were constructed by employing a set of c. 6200 ORF-specific PCR primer pairs (Research Genetics, Huntsville, AL). Individual PCR products were verified as unique via gel electrophoresis, purified, and mechanically ‘spotted’ onto polylysine-coated microscope slides using an Omni-Grid high-precision robotic gridder (GeneMachines, San Carlo, CA). cDNA targets were generated using a standard amino-allyl labeling protocol, where 30 μg each of ‘experimental’ and ‘reference’ total RNA was coupled to either Cy3 or Cy5 fluorophores. Targets were cohybridized to micro-arrays for 16 h at 63 °C and sequentially washed at room temperature in 1 × SSC and 0.03% sodium dodecyl sulfate (SDS) for 2 min, 1 × SSC for 2 min, 0.2 × SSC with agitation for 20 min, and 0.05 × SSC with agitation for 10 min. Arrays were immediately centrifuged until dry and scanned using a GenePix 4000 scanner (Axon Instruments, Union City, CA). Image analysis was performed using GenePix Pro 3.0.

Statistical analysis of the microarray data

For each array, background-subtracted spot intensities (red channel and green channel) were filtered and removed from further analysis if the spot intensity did not exceed 3 SD above the background signal in at least one channel. Red/green ratios were natural log transformed, and a logical regression normalization strategy (f = 0.67) was applied using S-Plus (MathSoft, Cambridge, MA) to correct for intra-array intensity-dependent ratio bias. Sample comparisons were independently replicated six times, each from a separate cell culture. Pairwise comparisons were made using CyberT (Baldi and Long, 2001) using a default window size of 101 and a confidence value of 6. Significance was determined by ranking the Bayesian P-values and using a false discovery rate methodology (FDR = 0.05) to account for multiple testing. A series of three ‘same vs. same’ hybridizations was performed using independent cultures, and the spot-level averaged normalized ratios were subtracted from the averaged normalized ratio of the six experimental replicates to correct for residual dye effects. A fold-change threshold of ≥1.5 was applied as an additional significance criterion, based in part on the expression pattern of known targets. This threshold was derived from ratios taken from the ‘same vs. same’ observations, where the threshold was set at four times the average SD value. Accordingly, any ORFs with a P-value that met the FDR = 0.05 criterion and where the normalized ratio was ≥1.5 were identified as differentially expressed. Heat maps were generated using the Eisen Cluster/Tree View Packge (Eisen et al., 1998).

Validation of microarray data by transcriptional fusions

Transcriptional (lacZ) fusions were constructed to a subset of genes that showed differential expression by DNA micro-array analysis. Plasmid V84 and plasmids containing filamentous growth pathway targets fused to lacZ for KSS1 (p2987), PGU1 (p2985), SVS1 (p3017) and YLR042c (p2988) were provided by C. Boone (University of Toronto; Roberts et al., 2000). Other gene fusions have also been described (Cullen et al., 2000, 2004). Reporter plasmids were transformed into the pmi40-101 strain, and beta;-galactosidase assays were performed in induced [yeast extract-peptone-dextrose (YEPD) medium] and uninduced (YEPD+50 mM mannose) conditions. Cells harboring the MSB2–lacZ reporter were assessed over multiple time points (2 h, 4 h, 8 h, and 16 h). Assays were performed from cell extracts derived from three separate trials, and the average of the values were expressed in Miller Units as previously described (Cullen et al., 2000). The resultant microarray data have been made available on the SGD Website (http://www.yeastgenome.org/).

Protein localization

Hog1-GFP localization was performed using a C-terminal Hog1–GFP fusion integrated at the HOG1 locus into a pmi40-101 ssk1 strain (1004) or an ssk1 strain in the filamentous background (1006). Cells were visualized by microscopy at ×100 using a fluorescein isothiocyanate (FITC) filter. Nuclei were visualized by 4,6-diamidino-2-phenylindole (DAPI) staining as previously described (Rose et al., 1990).

Microscopy

Differential-interference-contrast (DIC) and fluorescence microscopy using UV and FITC filter sets were performed using an Axioplan 2 microscope (Zeiss, Jena, Germany), a black and white Orca II digital camera (Hamamatsu, San Jose, CA), and the Openlab software program (Improvision, Coventry, UK).

Results and discussion

We turned to DNA microarray analysis (DeRisi et al., 1997) to better understand the cellular response to a protein glycosylation defect and to define the signaling pathways activated under this condition. We focused on the pmi40-101 mutant (Cullen et al., 2000), which exhibits a conditional glycosylation defect: when grown in the presence of mannose, no glycosylation defect is apparent; when grown in the absence of mannose, a glycosylation defect is evident (Smith et al., 1992; Cullen et al., 2000). Spotted cDNA microarrays (DeRisi et al., 1997; Lashkari et al., 1997) representing >96% of all yeast ORFs were used to monitor the changes in expression in the inducible glycosylation defect in the pmi40-101 mutant.

Expression analysis of a glycosylation mutant

cDNA sequences were prepared from the pmi40-101 mutant incubated in medium containing mannose, and compared to those obtained from incubation in medium lacking mannose at a 4 h time point, at which time significant MAP kinase activation occurs (data not shown). Six independent comparisons were performed, and most genes that had differential expression showed a highly correlative pattern between replicates. To validate the expression data, a statistical analysis was used in which variable signals were flagged and removed from the dataset. Significance was determined by a Bayesian method derived for microarray analysis (CyberT; Baldi and Long, 2001) and using an FDR methodology to account for multiple testing. Control hybridizations were performed and subtracted from the averaged normalized ratio of the experimental replicates to account for bye biases. We established a fold-change threshold of ≥1.5 as an additional significance criterion, based in part on the expression pattern of known induced targets. Only ORFs with a P-value that met the significance criterion and where the normalized ratio was ≥1.5 were identified as differentially expressed. The results of this analysis indicate differential induction of c. 740 ORFs, and repression of c. 440 ORFs was observed in the pmi40-101 mutant.

To validate the DNA microarray data, we analyzed a series of transcriptional fusions to a subset of the promoters that showed differential expression. beta;-Galactosidase assays were performed using promoter fusions carried in the pmi40-101 strain. All promoter fusions tested (SVS1, MSB2, FUS1, YLR042c, KSS1, FgTy1, FRE, and PGU1) showed a >2-fold induction of gene expression in mannose-limiting media (Fig. 1), indicating a strong correlation with the expression profile data. Databases containing annotated yeast genes (SGD, http://www.yeastgenome.org/) were used to classify the differentially expressed genes in the pmi40-101 mutant. Differentially induced gene sets fell into distinct functional classes, which were analyzed and compared with other genome-wide expression profiles. These results are summarized below.

Fig. 1.

Verification of a subset of microarray targets using transcriptional reporter gene fusions. The pmi40-101 mutant (strain 1436-13) containing lacZ fusions to the denoted genes was incubated for 4 h in medium containing (light rectangles) or lacking (dark rectangles) 50 mM mannose. beta;-Galactosidase assays were performed as described. The average of two independent experiments is shown.

Unfolded protein response

Altered glycosylation of proteins in the endoplasmic reticulum triggers the UPR, resulting in the induction of an intracellular signaling pathway that functions to promote protein folding and degrade misfolded proteins (Chapman et al., 1998). Expression profiles using drug treatments that block N-linked glycosylation (Travers et al., 2000) or in mutants partially defective for N-linked glycosylation (Klebl et al., 2001) have identified a number of genes whose products contribute to the UPR. The expression profile of the pmi40-101 mutant showed that most of the genes that play a role in the UPR were induced. These genes include HAC1, which encodes the transcription factor (Mori et al., 2000), and a number of its targets, INO1, ERO1, KAR2, PDI1, OPI3, YET1, and DER1 (Table 2). In addition, genes encoding proteins that promote protein folding were differentially induced, including LHS1, MPD1, EUG1, and FKB2 (Fig. S1). Thus, as expected in a mutant defective for protein glycosylation, transcriptional targets of the UPR were induced.

Table 2.

Microarray analysis of UPR, nutrition, cell cycle, and glycosylation-related genes differentially expressed in the pmi40-101 mutant

Gene name	Fold change	P-value	Process/function
UPR
INO1	10.7	<0.001	Inositol biosynthesis; inositol-1-phosphate synthase
ERO1	4.5	<0.001	Protein folding; disulfide bond formation in the endoplasmic reticulum
KAR2	3.0	<0.001	Secretion; BiP homolog; endoplasmic reticulum protein translocation
HAC1	3.0	0.038	Transcription factor for UPR
PDI1	2.8	<0.001	Protein folding; protein disulfide isomerase
OPI3	2.2	<0.001	Phospholipid metabolism; phospholipid synthase
YET1	2.2	<0.001	Enoplasmic reticulum 25 kDa transmembrane protein
DER1	2.1	<0.001	Protein degradation, endoplasmic reticulum
Gluconeogenesis
PCK1	5.9	<0.001	Phosphoenolpyruvate carboxykinase; Cat8 target
MDH2	3.1	<0.001	Malate dehydrogenase; Cat8 target
LSC2	2.5	<0.001	Succinyl-CoA ligase
GLK1	2.3	0.001	Glycolysis; glucokinase
CIT2	2.3	<0.001	Glyoxylate cycle; peroxisomal citrate synthase
NTH1	2.2	<0.001	Trehalose metabolism; α-trehalase
SFC1	2.1	<0.001	Transport; succinate–fumarate carrier; Cat8 target
IDP2	2.1	<0.001	Isocitrate dehydrogenase; Cat8 target
Hexose permease
HXT2	3.6	<0.001	High-affinity glucose transporter
HXT13	2.3	0.006	Hexose transporter
HXT16	2.1	<0.001	Hexose transporter
Glycerol metabolism
GUT2	3.4	<0.001	Glycerol-3-phosphate dehydrogenase, mitochondrial
RHR2	2.2	<0.001	DL-glycerol-3-phosphatase
HOR2	2.2	<0.001	DL-glycerol-3-phosphatase
ATP synthesis
STF1	3.4	<0.001	ATPase stabilizing factor
STF2	2.9	0.001	ATPase stabilizing factor
ATP3	2.0	<0.001	Mitochondrial F1F0 ATP synthase subunit
Amino acid biosynthesis
ARG3	−2.0	<0.001	Arginine biosynthesis; ornithine carbamoyltransferase
CPA2	−2.0	<0.001	Glutamate metabolism; carbamyl phosphate synthetase
HOM2	−2.1	<0.001	Threonine and methionine biosynthesis
SAM3	−2.1	<0.001	Transport, amino acid; S-adenosylmethionine permease
BAT2	−2.4	<0.001	Branched-chain amino acid biosynthesis; transaminase
ARG1	−3.0	<0.001	Arginine biosynthesis; arginosuccinate synthetase
Cell cycle
PCL1	3.3	<0.001	G1/S cyclin
CLN1	1.5	0.002	G1/S cyclin
CLN2	1.9	<0.001	G1/S cyclin
CLN3	1.5	0.001	G1/S cyclin
SIM1	2.4	<0.001	Cytoskeleton organization, control of DNA replication
HSL1	2.3	<0.001	G2/M; negative regulator of Swe1 kinase
PCL7	2.0	<0.001	Glycogen biosynthesis; CDK regulator
CLB2	−2.1	<0.001	G2/M cyclin
CLB1	−3.5	<0.001	G2/M cyclin
Glycosylation
AMS1	5.2	<0.001	Cell wall catabolism; vacuolar α-mannosidase
KTR2	4.3	<0.001	Protein glycosylation; putative mannosyltransferase

Results are presented as fold changes in mRNA levels (≥2 and ≤−2). Gene descriptions are from the Saccharomyces Genome database (http://www.yeastgenome.org)

Environmental stress response (ESR)

The ESR in yeast is a general stress response resulting in the coordinated activation of independent stress-responsive pathways (Gasch et al., 2000). We considered the possibility that a defect in glycosylation induces this general stress response. Comparison of genes induced >2-fold in the pmi40-101 mutant with those induced >2-fold in the ESR showed only a 14% overlap for characterized ORFs (Fig. S2). Most of the overlapping genes encode proteins that play a role in the heat shock response (HSP42, SSE2, and HSP104), protein degradation (PRB1 and PMC1), and nutrition (GLK1, YDR516c, PDE1, GLC3, and NTH1). Therefore, although some ESR genes are induced in the pmi40-101 mutant, the transcriptional response was largely distinct from the ESR.

Nutritional pathways

A large subset of the differentially expressed genes in the pmi40-101 mutant are differentially expressed during the diauxic shift, which represents entry into stationary phase (DeRisi et al., 1997). In particular, the expression of genes required for gluconeogenesis was induced in the pmi40-101 mutant, including Cat8, which encodes a transcription factor for gluconeogenesis (Fig. S3) and the Cat8 targets PCK1, MDH2, IDP2, JEN1, and SFC1 (Table 2). Genes that are required for hexose transport, trehalose, glycogen and glycerol metabolism, the glyoxylate cycle, ATP synthesis and poor carbon source utilization also showed differential induction (Table 2 and Fig. S3). Likewise, some genes known to be controlled by glucose repression (e.g. INO1, PHO89, AMS1, PCK1, MDH2, IDP2, ACS1, HXT2, HXT16, and ADR1) were induced in the pmi40-101 mutant (Fig. S3).

The gene expression profile in the pmi40-101 mutant differs from that of the diauxic shift in several ways. Most genes encoding citric acid cycle enzymes are not affected in the pmi40-101 mutant (13/14), although they are reported to be induced during the diauxic shift (DeRisi et al., 1997). Similarly, the genes required for respiration are not induced in the pmi40-101 mutant but are induced during the diauxic shift (DeRisi et al., 1997). Indeed, genes encoding mitochondrial ribosomal proteins, mitochondrial protein import machinery and proteins known to be associated with respiration were modestly repressed in the pmi40-101 mutant (data not shown). On the other hand, genes that encode components of the protein glycosylation pathway itself and genes required for recycling mannosyl moieties (Table 2) were induced in the pmi40-101 mutant but not in the diauxic shift.

Upon nutrient limitation, coordinate repression of ribo-some biosynthetic genes occurs (Warner, 1999; Raught et al., 2001). We also observed that the expression of ribosome biosynthetic genes was modestly repressed in the pmi40-101 mutant (Fig. S3). The only exception was MPR8, which showed induced expression (Fig. S3). This is consistent with previous reports (Yale and Bohnert, 2001; Parveen et al., 2003) indicating a distinct regulatory function for the protein. Genes involved in amino acid biosynthesis were also repressed in the pmi40-101 mutant (Table 2 and Fig. S3).

Cell cycle

Cell growth and the cell cycle are coupled processes (Tapon et al., 2001). Many genes are cell cycle regulated, and their differential expression by DNA microarray analysis can be used to pinpoint changes in cell cycle progression (Futcher, 2000). We examined the expression profile of the pmi40-101 mutant for such a bias and found differential induction of G­-associated transcripts, including the G1 cyclins (Levine et al., 1995) encoded by the CLN1, CLN2 and CLN3 genes, and the gene encoding the Pho85 G1 cyclin PCL1 (Table 2 and Fig. S1). Genes that control entry into the G2 phase of the cell cycle (Yeong et al., 2001) were repressed, including the cyclins CLB1 and CLB2 (Table 2 and Fig. S1). Delay in the G1 phase of the cell cycle is consistent with entry into stationary phase (Reinders et al., 1998) and with the slow growth defect and higher percentage of round, unbudded cells observed in the pmi40-101 mutant (data not shown). G1 delay is also consistent with the G1 arrest observed in cells containing misfolded proteins (Trotter et al., 2001). Delay in the G1 phase of the cell cycle indicates that a cell cycle checkpoint is initiated in the mutant in response to a glycosylation defect.

Defects in protein glycosylation activate the protein kinase C (PKC) and filamentous growth MAP kinase pathways

Activation of MAP kinase pathways can occur in response to a wide variety of cellular stresses. We examined the expression of known MAP kinase pathway targets to characterize the signaling response initiated in the pmi40-101 mutant. Comparison of the expression profile of the pmi40-101 mutant with targets of the cell wall integrity pathway (or PKC pathway) (Jung and Levin, 1999) revealed that 72% (18/25) of known PKC targets were induced in the pmi40-101 mutant (Table 3; Fig. S4). In addition, genes that encode components of that pathway, including the MAP kinase Mpk1 (Slt2), and transcription factor, Rlm1 (Jung et al., 2002), were modestly induced in the pmi40-101 mutant (Fig. S4).

Table 3.

Microarray analysis of MAP kinase targets differentially expressed in the pmi40-101 mutant

Gene name	Fold change	P-value	Process/function
PKC
PRM5*	12.9	<0.001	Pheromone-regulated protein, induced by PKC
YLR194C	8.2	<0.001	Unknown
PIR3	4.5	<0.001	Cell wall organization and biogenesis
FKS2	4.3	<0.001	Cell wall biogenesis; 1,3-beta;-D-glucan synthase subunit
CRH1	4.1	<0.001	Cell wall biogenesis (putative); cell wall protein
PST1*	3.2	<0.001	Secreted by regenerating protoplasts
HSP150	2.9	<0.001	Heat shock response; secreted glycoprotein of heat shock protein (HSP) family
DFG5	2.7	<0.001	Pseudohyphal growth
CIS3	2.6	<0.001	Unknown
SED1	2.4	<0.001	Cell surface glycoprotein
CHS3	2.2	<0.001	Cell wall biogenesis; chitin synthase III
SLT2	2.1	<0.001	MAP kinase for PKC pathway
Filamentous growth
SVS1	11.4	<0.001	Vanadate resistance
RTA1	5.6	<0.001	7-Aminocholesterol resistance
YLR042C	5.3	<0.001	Unknown
KTR2	4.3	<0.001	Protein glycosylation; putative mannosyltransferase
YLR414C	3.2	<0.001	Unknown
YPS1	3.1	<0.001	Protein processing; GPI-anchored aspartic protease
MSB2	2.7	<0.001	Filamentous growth pathway; cell surface component
MPT5	2.2	<0.001	Cell wall organization and biogenesis
YOR248W	2.2	<0.001	Unknown
KSS1	2.1	<0.001	MAP kinase for the filamentous growth pathway
DDR48	2.1	<0.001	Induced by DNA damage, heat shock, or osmotic stress
WSC2	2.0	<0.001	Cell wall biogenesis; α-1,4-glucan-glucosidase
FLO11	– 9.6	<0.001	Flocculation, invasive growth; regulated by the MAP kinase and cAMP pathways
HOG
YDL223C	2.3	<0.001	Cellular morphogenesis
RHR2	2.2	<0.001	Glycerol metabolism; DL-glycerol-3-phosphatase
HOR2	2.2	<0.001	Glycerol metabolism; DL-glycerol-3-phosphatase

Results are presented as fold changes in mRNA levels (≥2 and ≤ −2). Gene descriptions are from the Saccharomyces Genome database (http://www.yeastgenome.org).

^*Target of both PKC and filamentous growth pathways.

GPI, glycosylphosphatidylinositol.

Activation of the PKC pathway can occur in response to integrity defects in the cell wall (Levin et al., 1994; Kamada et al., 1995). Comparison of the expression profile of the pmi40-101 mutant with an expression profile derived from cells defective for 1,3-beta;-glucan synthase (fks1; Terashima et al., 2000) showed a striking overlap: 91% (20/22) of the genes induced in an fks1 mutant were also induced in the pmi40-101 mutant (Fig. S4). Most of these genes also contain Rlm1-binding sites in their promoter regions (Terashima et al., 2000). The induced cell wall genes include those that encode cell wall constituents such as chitin synthases (CHS1, CHS3, and CHS7) and genes that contribute to chitin biosynthesis (CSI2 and GFA1). The expression of several other genes known to influence cell wall biosynthesis, but not known Rlm1 targets, was induced in the pmi40-101 mutant, including SCW10, which encodes a cell wall glucanase, and KRE11, which encodes a regulator of beta;-1,6-glucan synthesis (Fig. S4). Activation of the PKC pathway in response to a glycosylation defect is consistent with the increased sensitivity of the pmi40-101 mutant to cell wall-perturbing agents (Cullen et al., 2000), with the previously reported induction of FKS2 (GSC2) expression observed in protein glycosylation mutants (Lee and Elion, 1999), and with the requirement for the PKC pathway for viability of glycosylation mutants (Cullen et al., 2000). As general cell wall defects resulting from hypoglycosylation of cell wall mannoproteins (Osmond et al., 1999), and defects in beta;-1,6-glucan synthesis (Chavan et al., 2003), induce similar responses, the data indicate that a defect in protein glycosylation causes a cell wall defect, which in turn triggers a PKC pathway-dependent response to maintain cell integrity.

We previously showed that a second MAP kinase pathway, the filamentous growth pathway, is activated in response to a protein glycosylation defect and is required for viability under this condition (Cullen et al., 2000). The expression profile of the pmi40-101 mutant confirmed that targets of the filamentous growth pathway (Roberts et al., 2000), including SVS1, YLR042c, and PGU1, were induced in the pmi40-101 mutant. One exception was the FLO11 gene, which is a known filamentous growth pathway target (Rupp et al., 1999) that was downregulated in the pmi40-101 mutant (Fig. S5). The FLO11 promoter is complex (Rupp et al., 1999) and contains cis-elements on which multiple signaling pathways converge; thus, its regulation in this instance may reflect the combined influence of multiple pathways. Genes that encode components of the filamentous growth pathway, including MSB2 and KSS1 (Table 3) (Cullen et al., 2004), were also induced in the pmi40-101 mutant. Filamentous growth is known to occur in response to nutrient depletion (Gimeno et al., 1992; Cullen and Sprague, 2000). Since the expression profile of the pmi40-101 mutant indicates that the cells suffer nutritional stress, it is possible that this condition contributes to activation of the filamentous growth pathway.

Filamentous growth pathway activation is distinct from high osmolarity glycerol response (HOG) pathway activation

We determined if targets of a third MAP kinase pathway, the HOG pathway, were induced in the pmi40-101 mutant. Transcriptional targets of the HOG pathway, including GPD1, CTT1, GPH1, HOR2, SIP18, GRE1, PGM2, STL1, and ALD6 (Posas et al., 2000; O’Rourke and Herskowitz, 2004), were not induced in the pmi40-101 mutant (Table 3), even at time points at which HOG pathway activation is reported to be maximal (data not shown). Only three HOG pathway targets were significantly induced: YDL233c, RHR2, and HOR2 (Table 3). Moreover, expression of the gene encoding the HOG pathway MAP kinase, HOG1, was not induced (Fig. S5).

Sho1 is an upstream component of both the filamentous growth and HOG pathways (Cullen et al., 2004). The DNA microarray analysis suggests that Sho1 activation is restricted to the filamentous growth pathway in the glycosylation mutant. This hypothesis is supported by expression analysis of cells undergoing filamentous growth (Madhani et al., 1999; Roberts et al., 2000) and by a report indicating that the HOG and PKC pathways function antagonistically (Garcia-Rodriguez et al., 2000). We determined whether Sho1 functions specifically in the filamentous growth pathway during a glycosylation defect. We used the nuclear localization of the HOG pathway MAP kinase Hog1 (Ferrigno et al., 1998; Reiser et al., 2000) to measure Sho1 activity in the HOG pathway. An integrated and functional Hog1–GFP fusion in the pmi40-101 mutant lacking the Sln1 branch of the HOG pathway (Posas and Saito, 1998) (ssk1) showed predominantly cytoplasmic localization of Hog1–GFP throughout a culture growth cycle in medium lacking mannose (Figs 2a and b). Addition of salt (0.9 M NaCl) caused rapid nuclear localization of Hog1–GFP in >90% of cells (Figs 2a and b). We also examined the specificity of Sho1 function in response to filamentous growth pathway activation. We found that the HOG pathway is not activated during filamentation in the Σ1278b background based on cytoplasmic localization of Hog1–GFP in ssk1 strains in glucose-limited medium (Fig. 2c). Therefore, in the two contexts in which Sho1 functions in the filamentous growth pathway, its activity appears to be restricted to that pathway.

Fig. 2.

Sho1 activation in response to a glycosylation defect or during filamentous growth does not activate the HOG pathway. (a) The pmi40-101 mutant containing ssk1 and an integrated Hog1–GFP fusion (strain 1004) was grown in medium containing mannose (+Man), lacking mannose ( −Man) or lacking mannose and containing 0.9 M NaCl ( −Man, +NaCl). Cells were examined by differential interference contrast (DIC) or fluorescence (FITC) microscopy at ×1000. Bar, 10 μm. (b) Graph of nuclear Hog1–GFP over a culture growth cycle. The strain in (a) was grown in media containing mannose (white squares, −Man), lacking mannose (filled triangles, +Man), or lacking mannose and containing 0.9 M NaCl (filled circles, –Man, −NaCl) over a culture growth cycle and periodically assayed for nuclear Hog1–GFP localization. (c) Filamentous Σ1278b cells lacking ssk1 and containing Hog1–GFP (strain 1006) were grown on minimal semisolid agar media containing or lacking glucose (Glu) for 24 h at 30 °C. A coverslip was added directly to the plate and cells were examined. Where indicated, NaCl (final 200 mM) was also added to the plate. Cells were examined by DIC or fluorescence (FITC) microscopy at ×1000. Bar, 10 μm.

Although Sho1 activity in the filamentous growth pathway does not trigger HOG pathway activation, Sho1 retains the capacity to function in the HOG pathway under conditions in which the filamentous growth pathway is active. Specifically, addition of salt to pmi40-101 ssk1 HOG1–GFP cells in medium lacking mannose (data not shown), or to ssk1 HOG1–GFP cells undergoing filamentous growth, caused rapid nuclear localization of Hog1–GFP (Fig. 2c). Therefore, Sho1 function in the filamentous growth pathway does not preclude its ability to activate the HOG pathway.

In summary, we have identified a large number of genes that are differentially expressed in response to a defect in protein glycosylation. An important subset of these are transcriptional targets of two MAP kinase pathways, indicating that one cellular response to glycosylation deficiency is MAP kinase pathway activation. This response may be a common theme in intracellular regulation in eukaryotic cells.