Cellular Processes and Pathways That Protect Saccharomyces cerevisiae Cells against the Plasma Membrane-Perturbing Compound Chitosan

Abstract

Global fitness analysis makes use of a genomic library of tagged deletion strains. We used this approach to study the effect of chitosan, which causes plasma membrane stress. The data were analyzed using T-profiler, which was based on determining the sensitivities of groups of deletion strains to chitosan, as defined by Gene Ontology (GO) and by genomic synthetic lethality screens, in combination with t statistics. The chitosan-hypersensitive groups included a group of deletion strains characterized by a defective HOG (high-osmolarity glycerol) signaling pathway, indicating that the HOG pathway is required for counteracting chitosan-induced stress. Consistent with this, activation of this pathway in wild-type cells by hypertonic conditions offered partial protection against chitosan, whereas hypotonic conditions sensitized the cells to chitosan. Other chitosan-hypersensitive groups were defective in RNA synthesis and processing, actin cytoskeleton organization, protein N-glycosylation, ergosterol synthesis, endocytosis, or cell wall formation, predicting that these cellular functions buffer the cell against the deleterious effect of chitosan. These predictions were supported by showing that tunicamycin, miconazole, and staurosporine (which target protein N-glycosylation, ergosterol synthesis, and the cell wall integrity pathway, respectively) sensitized Saccharomyces cerevisiae cells to chitosan. Intriguingly, the GO-defined group of deletion strains belonging to the “cytosolic large ribosomal subunit” was more resistant to chitosan. We propose that global fitness analysis of yeast in combination with T-profiler is a powerful tool to identify specific cellular processes and pathways that are required for survival under stress conditions.

Chitosan, a linear β-1,4-d-glucosamine polymer, is a deacetylated derivative of chitin. When added exogenously, it is known to cause cell leakage and to inhibit the growth of fungi and bacteria; importantly, it seems to be less toxic to mammalian cells. These properties make it a valuable antimicrobial compound with potential applications in medicine and in the food industry (37, 45, 46, 48, 56, 63). Chitosan has a pK_a value of around 6.5. Thus, at neutral and acidic pHs, chitosan molecules become positively charged as a result of the protonation of the amino groups in the glucosamine residues, enabling them to interact with negatively charged components of the plasma membrane, such as phospholipids (37). Chitosan also induces the formation of mass transfer channels in artificially created lipid bilayers, thus providing additional evidence for its disturbing effect on the plasma membrane (61). When Saccharomyces cerevisiae is challenged with sublethal concentrations of chitosan, it induces a specific transcriptional expression program comprising the environmental stress response and three more major transcriptional responses mediated by the transcription factors Cin5p, Crz1p, and Rlm1p, respectively (63). This is accompanied by structural changes in the cell wall, as reflected by the increased resistance of living cells to β-1,3-glucanase.

Here, we report a genome-wide screen of heterozygous essential and homozygous nonessential deletion strains for altered sensitivity to chitosan. This type of approach is complementary to transcriptional profiling (4, 6, 21, 23, 38, 43, 44, 60, 64). Homozygous nonessential deletion strains have been extensively used to examine the response of yeast deletion strains to various environmental conditions. Strains that are heterozygous for a deletion mutation of an essential gene are assumed to produce less of the essential gene's product and therefore tend to be hypersensitive to drugs that target that protein. For this reason, they may be used to identify novel drug targets.

We have used T-profiler (5) to identify predefined groups of gene deletion strains that are hypersensitive to chitosan stress. Chitosan-hypersensitive categories of deletion strains included groups with defects in RNA synthesis and processing, actin cytoskeleton organization, protein N-glycosylation, ergosterol synthesis, endocytosis, and cell wall formation. The results of this analysis were supported by using drugs that target specific cellular processes. For example, tunicamycin, which blocks the N-glycosylation of secretory proteins and thus causes the hypoglycosylation of yeast mannoproteins, strongly sensitized yeast cells to chitosan. Chitosan sensitivity was also elevated when it was combined with miconazole (an inhibitor of ergosterol biosynthesis) and with staurosporine, which inhibits the Pkc1p complex, a protein kinase complex that controls the cell wall integrity pathway. In addition, we demonstrate that various forms of plasma membrane stress, such as that caused by hypotonic conditions, treatment with chlorpromazine (a drug that causes membrane stretch), or a high growth temperature, also dramatically increase the sensitivity of the cells to chitosan. We conclude that analyzing global fitness experiments with T-profiler is a powerful approach to identify cellular processes and pathways that are essential for survival under drug-induced or other stress conditions.

MATERIALS AND METHODS

Strains and growth conditions.

S. cerevisiae strain BY4743 (MATa/MATα his3Δ1/his3Δ1 leu2Δ0/leu2Δ0 MET15/met15Δ0 LYS2/lys2Δ0 ura3Δ0/ura3Δ0) and individual deletion strains in this genetic background were used in this study. Synthetic complete (SC) medium (0.67% yeast nitrogen base without amino acids, 0.5% Casamino Acids, 2% glucose, with 25 mM phthalate-NaOH buffer [pH 5.5]) was used to culture the wild-type (WT) strain, the homozygous deletion strains (in the case of nonessential genes), and the heterozygous deletion strains (in the case of essential genes). Cells were precultured overnight in SC medium (in the case of the WT), for 12 h in yeast-peptone-dextrose (homozygous deletion pool), or for 5 h in SC medium (heterozygous essential pool). Batch fermentors were inoculated with cells to an optical density at 600 nm (OD₆₀₀) of 0.07 (∼0.7 × 10⁶ cells/ml). The cultures were grown in 1 liter of SC medium at 30°C, with an aeration rate of 1.0 liter/min, fast stirring (200 rpm), and monitoring of pH. Chitosan (final concentration, 25 μg/ml) was added at an OD₆₀₀ of 0.1 (∼1 × 10⁶ cells/ml). Growth was measured by monitoring the OD₆₀₀. Samples for genomic DNA isolation were taken after 5 and 9 h of growth. Reference samples were taken from a culture grown without chitosan at the same time points. An appropriate volume (20 to 25 ml) of the culture was centrifuged at 4,000 rpm for 5 min, and the cell pellets were stored at −20°C.

Preparation and analysis of chitosan.

Chitosan was obtained from Sigma (crab shells, minimum 85% deacetylated; average molecular mass of ≥600 kDa) (41). It was fragmented using nitrous deamination. Chitosan was dissolved in 10% acetic acid and incubated with sodium nitrite at a concentration of 20 mg per gram of chitosan at room temperature for 17 h. To stop the reaction, the pH of the preparation was adjusted to 5.5 with NaOH. The final concentration of chitosan was measured as glucosamine equivalents after hydrolysis in 6 M HCl at 100°C for 17 h (31). The average size of the chitosan fragments was defined as the ratio of the number of glucosamine residues measured after acid hydrolysis of chitosan and the number of reducing ends measured after fragmentation of chitosan by using a method described previously by Lever et al. (35) and was ∼50 glucosamine residues. This was supported by gel filtration, which indicated a molecular mass of about 10 kDa (∼60 glucosamine residues). We used chitosan fragments because we reasoned that native chitosan (average molecular mass of ≥600 kDa) might be too large to easily pass the cell wall (14) and because a previous study had shown that fragmented chitosan had higher antifungal activity than native chitosan (46).

SYTOX green uptake.

Cell samples from exponential-phase cultures of S. cerevisiae wild-type strain BY4741 grown in either the absence or presence of 25 μg/ml chitosan were spun down at 3,500 rpm, and the cell pellet was washed twice with 50 mM MES (morpholineethanesulfonic acid)-NaOH buffer (pH 5.5). Ninety microliters of cell suspension in buffer (OD₆₀₀ of 1) was transferred onto a 96-well plate with flat-bottom wells, and SYTOX green was added to a final concentration of 1 μM. The SYTOX green fluorescence was read at 1-min intervals for 30 min in a GeminiXS microtiter plate reader using 488 nm as the excitation wavelength, 544 nm as the emission wavelength, and 530 nm as the cutoff. To calculate relative fluorescence units, all measurements were expressed as a percentage of the fluorescence of a sample of cells treated with 70% (vol/vol) ethanol for 5 min for complete cell permeabilization, spun down at 3,500 rpm, resuspended in buffer, and measured as described above.

PCR amplification, hybridization, and data acquisition.

Yeast cells were lysed at 37°C with 0.5 mg/ml Zymolyase 100T for 60 min in sorbitol buffer (1 M sorbitol, 100 mM sodium EDTA, 14 mM β-mercaptoethanol). Genomic DNA was isolated with the QIAGEN DNeasy kit. The PCR amplification of the DNA tags was performed on 200 ng of genomic DNA template. Both UPTAGs and DOWNTAGs were amplified in separate reactions using biotinylated PCR primers complementary to common regions in the replacement cassette. PCR conditions and primers used were described previously (21). The final labeled UPTAGs and DOWNTAGs were purified on a YM10 Microcon column, combined, and hybridized to custom-made oligonucleotide microarrays (DNA TAG3; Affymetrix, Santa Clara, CA) as described previously (21), with the exception that four blocking primers were used in the hybridization mixture. After 16 h of hybridization at 42°C, the arrays were stained with streptavidin-phycoerythrin (Molecular Probes) and scanned with an Affymetrix GeneChip scanner. The hybridization intensities for each probe were determined using Affymetrix GeneChip operating software.

Data processing.

The majority of knockout strains are represented by four values of signal intensity (sense and antisense array elements for each UPTAG and DOWNTAG). We have calculated the mean of the signal intensities of tags representing each open reading frame (ORF) on the basis of two biological replicates of the experiment. The signal intensities of the UPTAGs and DOWNTAGs were calculated separately. Next, the ORFs whose averaged signal intensities were below the background level in the control experiment were discarded from further analysis. We defined the background as the average of all signal intensities of nonhybridizing spots and found that, for all arrays, it was below 20. To quantify the strain susceptibility to chitosan, we calculated the log₂ ratio (log₂ R) of normalized signal intensities for each strain in the control condition to those with chitosan treatment. This results in positive values for sensitive strains. To define significantly sensitive or resistant mutants in strains homozygous for the deletion of nonessential genes or heterozygous for essential genes, we used a cutoff of ≥1 for the log₂ R at 5 h (corresponding to a ratio of 2) and a cutoff of ≥1.585 for the log₂ R at 9 h (corresponding to a ratio of 3).

Data analysis using T-profiler.

We used an unpaired t test, which gives a measure of significance to the difference between the mean of a specific group of deletion mutants and the mean of the remaining deletion mutants of the total data set (5). To increase the robustness of the t test, we discarded the highest and lowest mutant log₂ R of all deletion mutant groups. This method is comparable to the “jackknife” procedure and reduces the effect of outliers, which might cause false-positive or false-negative results. Only groups consisting of at least five members were used for analysis. The P values obtained in this way were Bonferroni corrected for multiple testing by multiplying them by the number of mutant groups tested in parallel. Resulting E values that were ≤0.05 were considered to be significant.

The following two types of predefined groups of deletion strains were used in our analysis. The first type of predefined groups consisted of Gene Ontology (GO) categories. These are defined according to function, biological process, or cellular localization (15). Deletion mutant groups containing more than 100 members were left out. In total, 961 categories were tested. The second type consisted of synthetic lethality (SL)-based deletion mutant groups. These groups have been obtained by crossing a query deletion strain into a set of ∼4,900 viable deletion mutants and by screening the resulting double mutants for synthetic lethality or slow growth (54, 55). There have been ∼500 deletion strains used as queries in such experiments, resulting in the 462 deletion mutant groups used in our analysis. The complete gene composition of each SL- and GO-based deletion mutant group can be found on the SGD website (27).

Growth experiments.

Exponential-phase cultures of Saccharomyces cerevisiae wild-type strain BY4741 were diluted to an OD₆₀₀ of 0.1, and cultures of 200 μl were grown further in 96-well plates at 30°C. Either no drug, 25 or 50 μg/ml chitosan, 1.25 μg/ml tunicamycin, 250 μM chlorpromazine, 1.6 μM staurosporine, 2 μg/ml miconazole, or a combination of chitosan with one of the drugs was added. To combine chitosan with hypo-osmotic stress, the cells were suspended in SC medium (control culture) or 75% SC medium, and chitosan was added at 50 μg/ml. To test growth at 37°C, cells were suspended in SC medium, with or without 50 μg/ml chitosan, and placed in a 96-well microtiter plate, which was positioned in the reader that was set at 37°C. Growth was monitored for 16 h by measuring the OD₆₀₀ at 20-min intervals. The relative growth rates were calculated by linear regression from logarithmic plots of the OD₆₀₀ data versus time.

Chitosan susceptibility assays were carried out as described previously (63). In short, 10-fold serial dilutions of yeast cells were spotted onto solid SC medium prepared by using 2% agarose instead of agar, and the plates were incubated at 30°C for 3 days.

RESULTS

Overview of the chitosan susceptibilities of all deletion mutants.

Treatment of the S. cerevisiae wild-type strain with 25 μg/ml or 50 μg/ml chitosan resulted in a mild reduction in its relative growth rate (Table 1). SYTOX green is a dye that fluoresces when bound to nucleic acids but that can pass the plasma membrane only when that membrane's integrity is compromised. When exponentially growing cells were treated with 25 μg/ml chitosan and stained with this dye, the fluorescence increased from about 0.5% of the fluorescence of ethanol-treated cells to ∼6% at 15 min and increased further with time (7% at 30 min, 11% at 60 min, and 13% at 120 min), whereas the fluorescence of untreated cells remained unchanged (Fig. 1). These results suggest that chitosan at 25 μg/ml slightly affected the integrity of the plasma membrane. However, after a 4-h exposure to chitosan, the fluorescence decreased to 8% and to 2% at 6 h, suggesting that the cells may partially recover from treatment with this mild chitosan concentration.

TABLE 1.

Conditions and drugs that sensitize yeast cells to chitosan^a

Treatment	RGR (h−1)
	−Chitosan	+25 μg/ml	+50 μg/ml
Control	0.52	0.46	0.32
Decreased osmolarity (0.75 SC)	0.52	ND	0
Chlorpromazine (250 μM)	0.25	0	ND
Miconazole (2 μg/ml)	0.33	0.13	ND
Growth at 37°C	0.59	ND	0
Tunicamycin (1.25 μg/ml)	0.35	0.28	0.24
Staurosporine (1.6 μM)	0.48	0	ND

^aWT cells were cultured in SC medium (pH 5.5) in the absence (−) or presence (+) of the indicated drugs. Growth was monitored as OD₆₀₀ in a 96-well microtiter plate for 16 h. The data are from two independent experiments; the differences in values between the two experiments were generally 0.01 to 0.02 and never exceeded 0.04. RGR, relative growth rate; ND, not determined.

FIG. 1.

SYTOX green uptake of yeast cells grown in the presence of chitosan. Yeast cells were grown exponentially in the absence or presence of 25 μg/ml chitosan. Samples for fluorescence measurements were taken at the indicated time points. RFU, relative fluorescence units expressed as a percentage of the fluorescence obtained by treating the same number of cells with 70% ethanol. After 4 and 6 h of incubation, the relative fluorescence unit values were 8% and 2%, respectively. Closed triangles, control; closed squares, 25 μg/ml chitosan. Means and standard deviations are shown (n = 3).

To identify the genes and the cellular processes involved in chitosan sensitivity, a global fitness experiment was carried out in the presence of 25 μg/ml of chitosan. The abundance of the deletion strains in the control culture and in the chitosan-treated culture was determined after 5 and 9 h of growth, which corresponds to 3 to 4 and 6 to 7 generation times of the wild-type strain, respectively. These late time points were chosen to be able to detect even relatively small differences in growth. Among the ∼4,900 homozygous deletion mutants of nonessential genes, we found 184 strains with a log₂ R of ≥1 (this corresponds to a twofold reduction in strain abundance in the presence of chitosan) at 5 h and 153 strains with a log₂ R of ≥1.585 (this corresponds to a threefold reduction) at 9 h (see Fig. 2 for Venn diagrams) (for the whole list of mutants, see Table SD1 in the supplemental material). A total of 101 mutants were chitosan hypersensitive at both time points. To establish how reliable the results obtained by the genomic approach were, we retested a number of individual strains that varied widely in their sensitivities to chitosan. Both approaches generally gave the same results for both the sensitive and the resistant strains (see Table SD2 in the supplemental material). As shown previously (63), rlm1Δ cells were less sensitive to chitosan than wild-type cells. Interestingly, mnn4Δ and mnn6Δ cells, which are responsible for the introduction of (negatively charged) phosphodiester groups in N- and O-linked protein side chains, were also less sensitive to chitosan, indicating that the binding of chitosan to phosphodiester linkages enhances its effect. Conceivably, the binding of chitosan to such groups in plasma membrane proteins, and the resulting local increase in the concentration of chitosan, is largely responsible for this stimulatory effect.

FIG. 2.

Overview of chitosan-hypersensitive deletion mutants at 5 and 9 h. (A) Venn diagram summarizing the results for strains heterozygous for the deletion of an essential gene. (B) Venn diagram summarizing the results for strains homozygous for the deletion of a nonessential gene. The cutoffs for the log₂ R were 1 at 5 h (ratio of 2), and 1.585 at 9 h (ratio of 3). The numbers in parentheses refer to the total numbers of sensitive strains identified at the indicated time points.

Among the ∼1,100 strains heterozygous for the deletion of an essential gene, we detected only 11 and 3 chitosan-hypersensitive mutants at 5 and 9 h, respectively. In addition, the highest observed sensitivity of the mutants, expressed as log₂ R, was much lower than in the homozygous deletion strains both at 5 h (1.46 and 9.57, respectively) and at 9 h (1.75 and 9.39, respectively). These observations indicate that in most heterozygotes, the activity of the remaining copy of the gene is sufficient to withstand chitosan-induced stress to a large extent. Since previous studies have shown that heterozygous deletants are often highly sensitive to drugs that specifically target the product of the deleted gene (20, 22, 23, 38), our data also indicate that chitosan does not have a unique protein target.

Identification of GO-based groups of deletion strains with altered fitness in the presence of chitosan.

To acquire easily interpretable biological information about the response of deletion strains to chitosan treatment, we used T-profiler (which is based on t statistics) to identify changes in susceptibilities of predefined groups of mutants (see Materials and Methods). We first tested groups of strains with a deletion of a gene with a common regulatory motif in their upstream promoter region and groups with a deletion of a gene bound by a common transcription factor as determined by transcription factor occupancy analysis (25, 34). None of those groups of deletants had significant t values, indicating that there is no simple relationship between the transcriptional response of coregulated genes in response to stress and the fitness of the corresponding deletion mutants under the same conditions.

Next, we investigated the response of GO- and SL-based deletion mutant groups to chitosan. GO categories were included in our study in order to investigate whether the genes that were deleted in chitosan-hypersensitive GO-based groups of mutants share any functional features or participate in the same biological processes. The advantage of this approach is that all yeast genes are annotated in these categories, thus giving a comprehensive overview of the whole yeast genome. On the other hand, due to the hierarchical structure of the categories, there is considerable overlap between them. We further used 462 SL-based gene groups in our analysis, representing a wide spectrum of biological processes and cellular functions (for a detailed discussion of these results, see below). The major advantage of this approach is that it represents experimentally validated relationships between genes rather than the researchers' view of those relationships. However, it is incomplete, as only around 10% of the genes have been tested as query genes (54, 55).

We detected 13 significant GO groups, 12 of which were hypersensitive to chitosan, whereas 1 of them contained resistant mutants (Table 2). These 13 groups can be further ordered into several biologically relevant categories.

TABLE 2.

Identification of GO-based gene groups with altered fitness in the presence of chitosan^a

Gene group	No. of ORFs	t value
		5 h	9 h
General RNA polymerase II transcription factor activity	14	3.4	5.0
DNA-directed RNA polymerase II, holoenzyme	14	2.8	4.7
RNA polymerase II transcription mediator activity	5	2.8	4.4
RNA splicing via transesterification reactions with bulged adenosine as nucleophile	22	1.6	4.8
Endosome membrane	10	2.5	4.7
Retrograde transport, Golgi apparatus to ER	8	1.6	4.5
Nuclear envelope-ER network	64	5.2	4.4
Glycoprotein metabolism	44	3.9	4.6
Response to osmotic stress	40	1.7	4.7
Osmosensory signaling pathway via two-component system	8	1.7	4.2
Signal transducer activity	44	2.2	4.5
Receptor signaling protein activity	10	2.0	4.3
Cytosolic large ribosomal subunit (sensu eukaryota)	59	−3.0	−4.2

^aSignificant t values (t ≥ 4) are in boldface type.

Four groups of strains that carried deletions of genes involved in RNA polymerase II transcription or in RNA splicing were found to be hypersensitive to chitosan. The five most sensitive mutants are paf1Δ, snf5Δ, rtf1Δ, srb5Δ, and med1Δ. Interestingly, Paf1p and Rtf1p belong to the Paf1p complex, which is required for the full transcription of several cell wall biosynthetic genes (9). Ten mutants with a deletion of genes involved in processes occurring at the endosome membrane (snf8Δ, snf7Δ, vps20Δ, vps28Δ, vps22Δ, vps36Δ, vps25Δ, vps24Δ, did4Δ, and srn2Δ) are highly sensitive to chitosan. Deletion mutants of five genes that encode proteins participating in retrograde transport from the Golgi apparatus to the endoplasmic reticulum (ER) (get1Δ, get2Δ, get3Δ, glo3Δ, and sec22Δ) also showed increased sensitivity to chitosan. In addition, the mutants in the “nuclear envelope-ER network” group are hypersensitive to chitosan. These observations indicate that the proper functioning of the intracellular transport machinery is crucial for counteracting chitosan stress. We also found that mutants from the “glycoprotein metabolism” group, which includes N-glycosylation, are highly sensitive to chitosan. In addition, we identified mutants in the “response to osmotic stress” category to be hypersensitive to chitosan. These findings will be the subject of a more extended discussion later in this paper.

We also detected one GO functional category that contained genes whose deletants exhibited increased resistance to chitosan. This category, designated “cytosolic large ribosomal subunit,” comprises genes coding for structural proteins of the large ribosomal subunit. Interestingly, a similar behavior of deletion mutants in ribosomal proteins was observed in other genome-wide fitness studies of yeast (58, 60).

Deletion mutants of genes that synthesize a truncated lipid-linked oligosaccharide for protein N-glycosylation are highly susceptible to chitosan.

The pathway of protein N-glycosylation in yeast begins with the assembly of a lipid-linked oligosaccharide, which is later transferred in the ER to suitable protein N-glycosylation sites. The first part of the lipid-linked oligosaccharide assembly in yeast takes place on the cytoplasmic side of the ER membrane. The second part, which involves the addition of the final four mannose residues and three glucose residues, occurs in the ER lumen (7). None of the heterozygote deletion strains of essential genes showed increased sensitivity to chitosan (log₂ R of ∼0), consistent with our previously noted observation that most heterozygous deletants were not strongly affected by chitosan. We found that three of six homozygous deletion mutants of genes involved in the second part of lipid-linked oligosaccharide biosynthesis (alg9Δ, alg6Δ, and alg8Δ) are hypersensitive to chitosan (Table 3). All three mutants exhibit hypoglycosylation of secretory proteins. We found that the alg5Δ deletion mutant, which lacks the gene coding for Dol-P-Glc synthase, is also highly sensitive to chitosan. Dol-P-Glc is the sugar donor of the three glucose residues in the N-chain precursor.

TABLE 3.

Chitosan sensitivities of deletion strains of genes that synthesize a truncated lipid-linked oligosaccharide for protein N-glycosylation^a

ORF	Gene	Log2 R
		5 h	9 h
YOR002W	ALG6	1.17	1.93
YPL227C	ALG5	0.91	1.66
YNL219C	ALG9	1.55	1.50
YOR067C	ALG8	0.92	1.44
YGR227W	ALG10	0.40	0.89
YNR030W	ALG12	0.11	0.21
YBL082C	ALG3	0.25	0.15

^aSignificant values are in boldface type. The values are organized according to log₂ R at 9 h.

The sensitizing effect of incomplete N-glycosylation was also observed when yeast cells were treated with a combination of chitosan and tunicamycin, which blocks N-glycosylation (Table 1). Hypoglycosylation probably leads to a lower efficiency of folding of secretory proteins in the ER and thus to lower levels of the mature proteins (47). As many known cell wall polysaccharide synthases and cell wall construction enzymes are N-glycosylated, tunicamycin-induced hypoglycosylation may result in a weakened cell wall and, indirectly, in increased membrane stress, potentially rendering the cells more prone to the membrane-perturbing effect of chitosan.

HOG pathway mutants are highly susceptible to chitosan.

The HOG (high-osmolarity glycerol) pathway is crucial for the regulation of the intracellular osmotic balance of the yeast cell in response to osmotic changes in its surrounding environment and also for cell wall maintenance (1, 18, 32). The HOG pathway comprises two upstream signaling branches mediated by the osmosensing transmembrane proteins Sln1p and Sho1p and a downstream mitogen-activated protein (MAP) kinase cascade. In the Sln1p branch, the signal is transferred from Ssk1p to two redundant MAP kinase kinase kinases, Ssk2p and Ssk22p, which activate the MAP kinase kinase Pbs2p, which in turn activates Hog1p. We found that the homozygous deletion mutants of genes in this branch (ssk1Δ) and in the MAP kinase cascade (ssk2Δ, pbs2Δ, and hog1Δ but not the redundant gene ssk22Δ) are extremely sensitive to chitosan (Table 4). Consistent with these observations, the PBS2 gene was initially identified by the sensitivity of the deletion mutant to polymyxin B, a decapeptide that (like chitosan) is a cationic molecule and causes plasma membrane disintegration and leakage of cellular components (50). The hypersensitivity of mutants of the HOG pathway to chitosan indicates that the proper functioning of this pathway is required for counteracting chitosan stress. This was experimentally validated by incubating yeast cells before and during chitosan treatment in the presence of 1 M sorbitol, which partially protected the cells against the deleterious effect of chitosan on growth (Table 5). Conversely, hypotonic growth conditions made the cells more vulnerable to chitosan (Table 1).

TABLE 4.

Chitosan sensitivities of strains homozygous for the deletion of a nonessential gene from the HOG pathway^a

ORF	Gene	Log2 R
		5 h	9 h
YJL128C	PBS2	1.85	4.46
YNR031C	SSK2	1.72	4.13
YLR113W	HOG1	1.55	3.99
YLR006C	SSK1	1.15	2.33
YDL006W	PTC1	0.33	1.99
YCR073C	SSK22	0.09	0.09

^aSignificant values are in boldface type. The values are organized according to log₂ R at 9 h.

TABLE 5.

Pretreatment with 1 M sorbitol protects yeast against chitosan-induced stress^a

Treatment	RGR (%) at chitosan concn of:
	25 μg/ml	50 μg/ml
No sorbitol added	77	66
Sorbitol added forb:
1 h	93	80
2 h	92	82
3 h	89	80

^aYeast cells were exposed to chitosan after pretreatment with 1 M sorbitol for the indicated times. Relative growth rates (RGR) were expressed as a percentage of the relative growth rate in the absence of chitosan. The results are the means of three experiments. The standard errors were ≤1%. The relative growth rate of the control culture grown without sorbitol was ∼0.52 ± 0.02 h⁻¹.

^bPreincubation time.

Identification of chitosan-hypersensitive SL-based groups of deletants.

In synthetic genetic analysis, a strain bearing a mutation in a query gene is crossed with a set of ∼4,700 deletion mutants, and the resulting double mutants are screened for synthetic lethal (SL) or slow-growth (“sick”) (SS) phenotypes (54). Thus, an SL group is composed of the strain with a deletion of the query gene and all mutants that have an SL or SS interaction with this deletion mutant. Twenty-five SL-based deletion mutant groups were hypersensitive to chitosan (Table 6), comprising 61 chitosan-hypersensitive mutants. The membership of these SL groups only partially overlaps with that of the GO-based mutant groups. For example, two SL groups are involved in protein N-glycosylation (the HOC1 and the CWH41 groups), in agreement with the results obtained with the GO group analysis (Table 2). In addition, both GO and SL groups identify intercompartmental transport in the secretory pathway as being important for counteracting chitosan-induced stress. On the other hand, both approaches point to several unique cellular functions that are needed to counteract chitosan-induced stress, indicating that the two approaches complement each other.

TABLE 6.

Identification of chitosan-hypersensitive SL deletion mutant groups^a

SL group and query gene	No. of ORFs	t value
		5 h	9 h
Cell wall integrity
BNI4	16	1.53	4.12
CHS3	41	5.51	4.74
SKT5/CHS4	34	7.31	5.61
SMI1	44	3.62	4.5
Glycoprotein biosynthesis
CWH41	9	3.27	7.46
HOC1	32	4.18	4.16
Plasma membrane integrity
ERG11	23	3.83	6.19
CNB1b	26	2.57	4.09
Intercompartmental transport in secretory pathway
CHS5	59	5.37	8.45
CHS6	19	8.37	7.74
GYP1	24	3.1	4.41
RIC1	118	4.3	6.66
YPT6	117	3.96	6.56
Cytoskeleton related
ARC40	33	5.3	6
ARP2	40	4.18	5.64
GIM3	73	3.42	5.07
GIM5	102	5.27	6.46
PAC10/GIM2	107	3.72	4.76
RVS161	45	2.88	4.64
RVS167	37	3.43	5.42
YKE2/GIM1	106	3.72	4.99
Nuclear proteins
ARP6	37	3.36	5.64
CDC73	94	3.69	5.57
CTF4	84	2.52	4.37
SET2	31	3.49	4.9

^aSignificant values (t ≥ 4) are in boldface type.

^bOur own unpublished data.

The first four chitosan-hypersensitive groups are based on query genes that are involved in the formation of the cell wall and its localization and regulation (Table 6). CHS3 and CHS4 are directly involved in chitin synthesis, whereas BNI4 is involved in the polymer's localization (13). The SMI1 gene encodes a signaling protein that is involved in regulating cell wall synthesis (3). These observations indicate that cell wall integrity is required to withstand chitosan-induced stress. Consistent with this, staurosporine (a potent inhibitor of the protein kinase C complex, which controls the cell wall integrity signaling pathway) (36, 59, 62) showed a dramatic growth-inhibitory effect when combined with chitosan (Table 1).

Plasma membrane integrity also seems to be a major contributing factor to the yeast cell's ability to cope with chitosan-induced stress. Members of both the ERG11-based SL group and the CNB1-based SL group are chitosan hypersensitive. Ergosterol is the main component responsible for the rigidity of the plasma membrane, and ERG11 encodes an essential enzyme in ergosterol biosynthesis, the cytochrome P-450-dependent C14 lanosterol demethylase. The hypersensitivity of the ERG11-based SL group to chitosan was experimentally confirmed by incubating yeast cells in the presence of a combination of chitosan and miconazole (a specific inhibitor of Erg11p), resulting in a synergistic effect on growth (Table 1). CNB1 encodes the regulatory subunit of calcineurin, which is involved in coping with plasma membrane stress (11). The identification of the CNB1-based SL group is in agreement with our previous study, where cells exposed to the calcineurin inhibitor FK506 showed increased chitosan sensitivity (63). Additional experimental validation for a key role of plasma membrane integrity in coping with chitosan-induced stress was obtained by subjecting the cells to hypotonic stress and to chlorpromazine, a cationic amphipathic drug that inserts itself into the plasma membrane lipid bilayer, thus leading to plasma membrane stretching (30). Table 1 shows that both treatments acted synergistically in combination with chitosan. Consistent with this, cells grown at 37°C, a temperature that is believed to increase membrane fluidity, were also more sensitive to chitosan.

Finally, eight groups consist of mutants that show SL with mutants of genes encoding various cytoskeleton-related proteins. These include the four members of cytoplasmic prefoldin protein complex (GIM1, GIM2, GIM3, and GIM5), which participates in the folding of tubulin and actin; RVS161 and RVS167, both subunits of the same complex that regulates cell polarity, actin cytoskeleton polarization, and endocytosis; and, finally, ARC40 and ARP2, the components of the actin nucleation center, involved in endocytosis, membrane growth, and polarity. In addition, we detected four groups of chitosan-hypersensitive mutants that have SL interactions with CFT4, SET2, ARP6, and CDC7, genes participating in DNA binding and histone methylation.

DISCUSSION

Like many natural antimicrobial compounds found in vertebrates, invertebrates, and plants, chitosan is positively charged (at least at neutral and acidic pHs) and perturbs the plasma membrane of bacteria and fungi (51-53). Cationic peptides and other cationic compounds, including chitosan, have been shown to facilitate the entry of molecules into mammalian cells (17). Chitosan may thus function as an antimicrobial compound itself and as a model compound to understand the deleterious effects of other cationic antimicrobial compounds on microorganisms. In a previous study, we analyzed the transcriptional response of yeast cells to chitosan (63). Here, we have complemented the transcriptional analysis of chitosan-treated cells with a global fitness study. Global fitness studies make use of a genomic library of deletion strains, allowing the parallel measurement of their growth response to a chosen stress condition (43, 44, 54). For the analysis and interpretation of our data, we have applied and extended the approach developed for global transcriptional analysis, called T-profiler, which is based on the study of the transcriptional behavior of related groups of genes using t statistics (5, 63). In this study, we have determined the response to chitosan of predefined groups of mutant strains carrying a deletion in related genes. Two types of deletion mutant groups were successfully tested: (i) groups based on GO categories (2), and (ii) groups based on global SL screens (54, 55). GO groups have the advantage that they cover all deletions, in contrast to the incomplete coverage by SL groups. However, the SL groups have the intrinsic advantage that they are defined by the yeast cell rather than the investigator.

Both approaches led to testable predictions. For example, the analysis based on GO groups identified a role for the Sln1 signaling branch and the Hog1 MAP kinase module of the HOG pathway in protecting the cell against chitosan-induced stress. In contrast, the Sho1 signaling branch is not involved. We hypothesize that chitosan may make the plasma membrane leaky, for example, by inducing the formation of mass transfer channels as has been observed in artificial lipid bilayers (61). This would result in the uptake of (acidic) medium and, consequently, cytosolic acidification. The HOG pathway genes SSK1, SSK2, PBS2, and HOG1, which function in the Sln1 branch and the Hog MAP kinase module, respectively, might then be expected to offer protection not only against chitosan (this paper) but also against cytosolic acidification resulting from other stresses. Evidence for this comes from the work of Mollapour and Piper (40), who found that mutants with a deletion of either SSK1, PBS2, or HOG1 are highly sensitive to acetic acid/acetate at pH 4.5. At this pH, which is below the pK_a value of acetic acid, undissociated, and thus uncharged, acetic acid molecules form the majority. The uncharged acetic acid molecules can easily pass the plasma membrane. They will immediately dissociate upon arrival in the cytosol, resulting in acid stress. Mollapour and Piper (40) also found that the sho1Δ strain does not show increased sensitivity to acetic acid/acetate, demonstrating that the Sho1 signaling branch is not involved in coping with acid stress. Similar results have been described previously by Lawrence et al. (33), who found that ssk1Δ, pbs2Δ, and hog1Δ cells, but not sho1Δ cells, were more sensitive to citric acid/citrate at pH 3.5, and by Kapteyn et al. (32), who showed that Hog1p is required for a cell wall-strengthening response induced by culturing cells at pH 3.5. Using global transcript analysis, we have previously shown that in the presence of chitosan, yeast cells strongly and continuously activate the cell wall integrity pathway (63). This raises the question of how cells challenged with chitosan can have the cell wall integrity pathway turned on, whereas simultaneously, the HOG pathway is functional in protecting the cells against chitosan. Interestingly, Mollapour and Piper (40) showed that in yeast cells challenged with acetic acid/acetate at pH 4.5, the doubly phosphorylated forms of Slt2p and Hog1p can be present in the cell simultaneously, indicating that at least under some stress conditions, the simultaneous activation of both MAP kinases is possible. This further suggests that the HOG pathway might have functions other than organizing a response to hypertonic stress, as also indicated by the observations that the HOG pathway is induced by cold stress and dimethyl sulfoxide (26, 42).

Analysis of both types of deletant groups identified defective protein N-glycosylation as an important factor contributing to chitosan hypersensitivity, and this was supported experimentally by the observation that tunicamycin sensitized yeast cells to chitosan (Table 1). As defective protein N-glycosylation affects protein folding in the ER, it seems likely that many secretory proteins involved in cell wall construction will be produced in insufficient amounts, affecting cell wall integrity and indirectly resulting in increased membrane stress.

Several SL groups are related to actin cytoskeleton organization. Previous studies have shown that actin cytoskeleton depolarization occurs upon heat and cell wall stress (12) as well as under hypo- and hyperosmotic conditions (10, 24). Interestingly, actin cytoskeleton depolarization was also observed upon treatment with two plasma membrane perturbants, sodium dodecyl sulfate and chlorpromazine (12). The actin cytoskeleton is also required for stress-dependent cellular redistribution of the cell wall biosynthetic protein Fks1p (12). The transient depolarization of the actin cytoskeleton may therefore act as a homeostatic mechanism to repair cell wall and membrane damage, which can explain why mutants defective in actin cytoskeleton organization are hypersensitive to membrane stress caused by chitosan. Interestingly, actin patches colocalize with sites of endocytosis and associate with endosomes, and their movement is mediated by polarized actin cables (28). Endocytosis in S. cerevisiae is essential for plasma membrane-associated functions, because it is required for the recycling of plasma membrane components, uptake of nutrients, and regulation of cell surface signaling receptors (16, 57). The observation that mutants that are defective in endocytosis are hypersensitive to chitosan-induced plasma membrane perturbation supports the idea that endocytosis is crucial for the maintenance of plasma membrane integrity and functioning.

Recent studies of S. cerevisiae have demonstrated that on the level of individual genes, there is often no correlation between the transcriptional activation of a gene upon stress and the sensitivity of the corresponding deletion strain to the same stress (4, 21). Similarly, we have found that there is no clear relationship between the transcriptional responses of coregulated genes in case of stress and the stress sensitivities of the corresponding groups of deletion strains. The transcriptional response to stress is unavoidably confounded by a slow-growth response (8). We have separated these two responses by focusing our analyses at the level of cellular functions and pathways, thus gaining results in which transcriptional and fitness profilings show a larger degree of similarity.

We (63) showed previously that the main transcriptional responses of yeast cells during the first 180 min of incubation in the presence of chitosan are mediated by the transcription factor Msn2p/Msn4p, which mediates a general stress response (19); Crz1p, which probably mediates the transcriptional response to plasma membrane stress (49); Rlm1p, which activates a group of mostly cell wall-related genes in case of cell wall stress (29); and Cin5p, which mediates salt tolerance (39). Although the global fitness experiment described in this paper covers a period of 9 h and the data are thus not directly comparable to the transcriptional profiling data obtained during a 180-min chitosan treatment, our current results clearly point to a major role for cell wall and plasma membrane integrity in counteracting and recovering from chitosan-induced stress. They also show the benefit of analyzing the data on the level of functional categories of genes using T-profiler.

In summary, our genome-wide analysis of fitness of all yeast deletion mutants in response to chitosan treatment indicates that the loss of genes that are involved (directly or indirectly) in maintaining plasma membrane integrity is the primary cause of chitosan hypersensitivity (summarized in Fig. 3). In addition, we have shown that T-profiler analysis of yeast fitness data, based on GO- and SL-based gene groups, is highly effective in identifying cellular processes and pathways that are crucial for defending the cell against chitosan-induced stress. This approach is equally effective for the analysis of other fitness profiles (our own unpublished data) and may eventually provide a useful platform for the comparison and integration of other large-scale genomic studies. Finally, it will also allow the systematic identification of combinations of drugs to prevent or minimize fungal infections in food and hosts.

FIG. 3.

Factors contributing to chitosan hypersensitivity in S. cerevisiae. Δmutants refers to deletion mutants.