OUP user menu

Genome-wide analysis of the response to protein glycosylation deficiency in yeast

Paul J. Cullen, Rufeng Xu-Friedman, Jeffrey Delrow, George F. Sprague
DOI: http://dx.doi.org/10.1111/j.1567-1364.2006.00120.x 1264-1273 First published online: 1 December 2006


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.

  • MAP kinase
  • signaling
  • filamentous growth
  • cell wall


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, 2000), regulated secretion (Scheiffele and Fullekrug, 2000), and protein localization (Ohnishi, 2003; Watanabe, 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, 1998). In addition, defects in protein glycosylation influence secretion (Travers, 2000), cell integrity (Lee and Elion, 1999; Cullen, 2000), and the activation of cellular signaling pathways (Cullen, 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, 2000; Lecca, 2005) or analyzing mutants partially defective for N-linked glycosylation (Klebl, 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, 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, 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, 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, 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, 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 FUSI reporter expression, invasive growth phenotypes, and viability of the pmi40-101 mutant. Yeast and bacterial strains were propagated using standard methods (Rose, 1990). Yeast peptone dextrose (YPD) and synthetic complete dextrose (SCD) media have been described (Rose, 1990). Yeast transformations were performed as described (Gietz, 1995). Bacterial transformations, bacterial DNA preparations and plasmid constructions were performed by standard methods (Sambrook, 1989). Strains harboring the pmi40-101 allele were maintained on medium containing 50 mM mannose unless otherwise stated.

View this table:

Yeast strains used in this study

1436Wild typeCullen (2000)
1436–13ste4 pmi40-101Cullen (2000)
1004ste4 pmi40-101 ssk1 HOG1–GFP::KanMX6This study
1006Σ1278b ssk1HOG1–GFP::KanMX6This 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, 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, 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 OmniGrid 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 microarrays 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, 1998).

Validation of microarray data by transcriptional fusions

Transcriptional (lacZ) fusions were constructed to a subset of genes that showed differential expression by DNA microarray 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, 2000). Other gene fusions have also been described (Cullen, 2000, 2004). Reporter plasmids were transformed into the pmi40-101 strain, and β-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, 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, 1990).


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, 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, 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, 1992; Cullen, 2000). Spotted cDNA microarrays (DeRisi, 1997; Lashkari, 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 dye 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. β-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.


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. β-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, 1998). Expression profiles using drug treatments that block N-linked glycosylation (Travers, 2000) or in mutants partially defective for N-linked glycosylation (Klebl, 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, 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.

View this table:

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

Gene nameFold change P-valueProcess/function
INO110.7<0.001Inositol biosynthesis; inositol-1-phosphate synthase
ERO14.5<0.001Protein folding; disulfide bond formation in the endoplasmic reticulum
KAR23.0<0.001Secretion; BiP homolog; endoplasmic reticulum protein translocation
HAC13.00.038Transcription factor for UPR
PDI12.8<0.001Protein folding; protein disulfide isomerase
OPI32.2<0.001Phospholipid metabolism; phospholipid synthase
YET12.2<0.001Endoplasmic reticulum 25 kDa transmembrane protein
DER12.1<0.001Protein degradation, endoplasmic reticulum
PCK15.9<0.001Phosphoenolpyruvate carboxykinase; Cat8 target
MDH23.1<0.001Malate dehydrogenase; Cat8 target
LSC22.5<0.001Succinyl-CoA ligase
GLK12.30.001Glycolysis; glucokinase
CIT22.3<0.001Glyoxylate cycle; peroxisomal citrate synthase
NTH12.2<0.001Trehalose metabolism; α-trehalase
SFC12.1<0.001Transport; succinate–fumarate carrier; Cat8 target
IDP22.1<0.001Isocitrate dehydrogenase; Cat8 target
Hexose permease
HXT23.6<0.001High-affinity glucose transporter
HXT132.30.006Hexose transporter
HXT162.1<0.001Hexose transporter
Glycerol metabolism
GUT23.4<0.001Glycerol-3-phosphate dehydrogenase, mitochondrial
ATP synthesis
STF13.4<0.001ATPase stabilizing factor
STF22.90.001ATPase stabilizing factor
ATP32.0<0.001Mitochondrial F1F0 ATP synthase subunit
Amino acid biosynthesis
ARG3−2.0<0.001Arginine biosynthesis; ornithine carbamoyltransferase
CPA2−2.0<0.001Glutamate metabolism; carbamyl phosphate synthetase
HOM2−2.1<0.001Threonine and methionine biosynthesis
SAM3−2.1<0.001Transport, amino acid; S-adenosylmethionine permease
BAT2−2.4<0.001Branched-chain amino acid biosynthesis; transaminase
ARG1−3.0<0.001Arginine biosynthesis; arginosuccinate synthetase
Cell cycle
PCL13.3<0.001G1/S cyclin
CLN11.50.002G1/S cyclin
CLN21.9<0.001G1/S cyclin
CLN31.50.001G1/S cyclin
SIM12.4<0.001Cytoskeleton organization, control of DNA replication
HSL12.3<0.001G2/M; negative regulator of Swe1 kinase
PCL72.0<0.001Glycogen biosynthesis; CDK regulator
CLB2−2.1<0.001G2/M cyclin
CLB1−3.5<0.001G2/M cyclin
AMS15.2<0.001Cell wall catabolism; vacuolar α-mannosidase
KTR24.3<0.001Protein 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/).

  • a results are presented as fold changes in mRNA levels (>2 and <2),

  • b determined by the average of 6 trials; see Materials and methods,

  • c determined by description 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, 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, 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, 1997). Similarly, the genes required for respiration are not induced in the pmi40-101 mutant but are induced during the diauxic shift (DeRisi, 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 ribosome biosynthetic genes occurs (Warner, 1999; Raught, 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, 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, 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 G1-associated transcripts, including the G1 cyclins (Levine, 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, 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, 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, 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, 2002), were modestly induced in the pmi40-101 mutant (Fig. S4).

View this table:

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

Gene nameFold change P-valueProcess/function
PRM512.9<0.001Pheromone-regulated protein, induced by PKC
PIR34.5<0.001Cell wall organization and biogenesis
FKS24.3<0.001Cell wall biogenesis; 1,3-β-d-glucan synthase subunit
CRH14.1<0.001Cell wall biogenesis (putative); cell wall protein
PST13.2<0.001Secreted by regenerating protoplasts
HSP1502.9<0.001Heat shock response; secreted glycoprotein of heat shock protein (HSP) family
DFG52.7<0.001Pseudohyphal growth
SED12.4<0.001Cell surface glycoprotein
CHS32.2<0.001Cell wall biogenesis; chitin synthase III
SLT22.1<0.001MAP kinase for PKC pathway
Filamentous growth
SVS111.4<0.001Vanadate resistance
RTA15.6<0.0017-Aminocholesterol resistance
KTR24.3<0.001Protein glycosylation; putative mannosyltransferase
YPS13.1<0.001Protein processing; GPI-anchored aspartic protease
MSB22.7<0.001Filamentous growth pathway; cell surface component
MPT52.2<0.001Cell wall organization and biogenesis
KSS12.1<0.001MAP kinase for the filamentous growth pathway
DDR482.1<0.001Induced by DNA damage, heat shock, or osmotic stress
WSC22.0<0.001Cell wall biogenesis; α-1,4-glucan-glucosidase
FLO11– 9.6<0.001Flocculation, invasive growth; regulated by the MAP kinase and cAMP pathways
YDL223C2.3<0.001Cellular morphogenesis
RHR22.2<0.001Glycerol metabolism; dl-glycerol-3-phosphatase
HOR22.2<0.001Glycerol 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.

  • a results are presented as fold changes in mRNA levels (>2 and <2),

  • b determined by the average of 6 trials; see Materials and methods,

  • c determined by description from the Saccharomyces Genome Database (http://www.yeastgenome.org/).

Activation of the PKC pathway can occur in response to integrity defects in the cell wall (Levin, 1994; Kamada, 1995). Comparison of the expression profile of the pmi40-101 mutant with an expression profile derived from cells defective for 1,3-β-glucan synthase (fks1; Terashima, 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, 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 β-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, 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, 2000). As general cell wall defects resulting from hypoglycosylation of cell wall mannoproteins (Osmond, 1999), and defects in β-1,6-glucan synthesis (Chavan, 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, 2000). The expression profile of the pmi40-101 mutant confirmed that targets of the filamentous growth pathway (Roberts, 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, 1999) that was downregulated in the pmi40-101 mutant (Fig. S5). The FLO11 promoter is complex (Rupp, 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, 2004), were also induced in the pmi40-101 mutant. Filamentous growth is known to occur in response to nutrient depletion (Gimeno, 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, 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, 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, 1999; Roberts, 2000) and by a report indicating that the HOG and PKC pathways function antagonistically (García-Rodriguez, 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, 1998; Reiser, 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.


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.

Supplementary material

The following supplementary material is available online:

Fig. S1. Genes differentially expressed in the pmi40-101 mutant involved in the UPR, protein glycosylation and cell cycle progression pathways. Heat map without cluster analysis showing differential expression of genes for the six trials of the DNA microarray analysis. As shown in the scale bar, red indicates degree of differential induction, and green indicates degree of differential repression. The Fold Change refers to the average of all six experiments, with the indicated P-value shown. Function refers to gene descriptions from the Saccharomyces Genome database (http://www.yeastgenome.org).

Fig. S2. Overlap between the ESR and glycosylation mutant expression profiles. (a) Expression profile of genes differentially expressed >2-fold in the pmi40-101 mutant (left two panels), and expression profile of known genes induced in the ESR (right panel). The description of the heat map produced for the microarray analysis is as in Fig. S1. (b) Venn diagram illustrating the degree of overlap between the two profiles.

Fig. S3. Nutritionally controlled genes are differentially expressed in the pmi40-101 mutant. Genes are clustered loosely by function. The description of the heat map produced for the microarray analysis is as in Fig. S1.

Fig. S4. PKC, fks1 and cell wall genes are differentially expressed in the pmi40-101 mutant. For the PKC pathway targets, the CTT1, SPS100, MLP1, PGK1 and PIR1 profiles are not shown because they contained high P-values. The description of the heat map produced for the microarray analysis is as in Fig. S1.

Fig. S5. Genes differentially expressed in the pmi40-101 mutant that comprise known filamentous growth and HOG pathway targets. The description of the heat map produced for the microarray analysis is as in Fig. S1.


This work was supported by a research grant from the US Public Health Service (GM-30027 for G.F.S.) and fellowships from the American Heart Association (AHA120635Z and AHA0535393T for P.C.). We thank Charlie Boone and John Pringle for plasmids, and Marc Dorfman for assistance with Hog1–GFP localization.


  • Editor: Andrew Alspaugh


View Abstract