Saccharomyces cerevisiae responds to a wide array of environmental signals through a variety of mechanisms, including induction and repression of specific transcriptional programs. For example, the response to oxidative stress is well characterized and involves the induction of antioxidant species such as superoxide dismutase, catalase, and peroxidase, all of which improve the ability of the cell to survive exposure to reactive oxygen species (Costa and Moradas-Ferreira, 2001 ). In this work, we sought to identify components of the pathway by which S. cerevisiae senses and detoxifies exogenously supplied nitric oxide (^.NO).

^.NO is a membrane-permeable free radical that is biologically produced by the nitric-oxide synthase (NOS) family of enzymes (Sessa, 1994 ) and by denitrifying bacteria (Xu and Verstraete, 2001 ). ^.NO reacts in a concentration- and environmental-dependent manner, leading to the formation of reactive nitrogen intermediates (RNIs). ^.NO and resulting RNIs have been shown to have both cytostatic and cytotoxic activity due to the inhibition of ATP production (Brown, 1997 ; Stevanin et al., 2000 ; Chenais et al., 2002 ), altered iron metabolism (Chenais et al., 2002 ), (Martinez et al., 2001 ; D'Autreaux et al., 2002 ), direct inhibition of enzymes (Gardner et al., 1997 ), and DNA damage (Martinez et al., 2001 ; Kow, 2002 ). ^.NO has been strongly implicated as a component in higher eukaryotic nonspecific immune response to parasites, fungi, bacteria, and viruses (Fang, 1997 ). Host inducible NOS expression and endogenous ^.NO levels increase in response to infection by a wide range of pathogens, including Plasmodium falciparum (Kun, 2003 ), Leishmania major (Bogdan et al., 2000 ), Candida albicans (Elahi et al., 2001 ), Borrelia burgdorferi (Harter et al., 1999 ), and cytomegalovirus (Tanaka and Noda, 2001 ). For these pathogenic organisms, the ability to detoxify ^.NO may be important for their survival, proliferation, and virulence.

Microorganisms have developed mechanisms to detoxify ^.NO, including the conversion of ^.NO to nitrate. This reaction is catalyzed by the Escherichia coli flavohemoglobin protein hmp (Gardner et al., 1998 ). Homologues of hmp are present in many bacteria and yeast species and are induced by reactive nitrogen species in Escherichia coli (Poole et al., 1996 ), Salmonella typhimurium (Crawford and Goldberg, 1998 ), and C. albicans (Ullmann et al., 2004 ). Deletion of flavohemoglobin in the human pathogens C. albicans and Cryptococcus neoformans leads to a decrease in virulence in mouse models of infection (de Jesus-Berrios et al., 2003 ; Ullmann et al., 2004 ).

The hmp ortholog in S. cerevisiae is Yhb1p (Zhu and Riggs, 1992 ), which is required for metabolism of ^.NO (Liu et al., 2000 ). yhb1Δ strains are hypersensitive to exogenous ^.NO, implying an important detoxification role for Yhb1p (Liu et al., 2000 ).

We hypothesized that S. cerevisiae might possess mechanisms capable of responding to exogenous ^.NO or other forms of nitrosative stress. To test this hypothesis, we used cDNA microarrays to track changes in S. cerevisiae RNA transcript abundance levels after exposure to exogenously supplied ^.NO. From this genome-wide survey, we were able to identify and characterize a specific nitrosative stress response. This response includes the induction of the YHB1 and SSU1 genes as well as three other uncharacterized open reading frames. We also show that the transcription factor Fzf1p is necessary for this response and further characterize the cis-acting determinants sufficient for Fzf1p-dependent transcriptional activation. Finally, we show both that YHB1 and SSU1 contribute to nitrosative stress resistance, depending on growth conditions and that absence of FZF1 results in hypersensitivity to nitrosative stress.

Table 1. S. cerevisiae strains used in this study

Poole et al., 1996

Time-course response experiments were conducted using strain DBY7283 in SCD media after mock exposure, exposure to 100 μM APD or exposure to 100 μM DPTA NONOate (Figure 1, A-C). Similarly, time course response experiments were conducted using YHB1-GFP and YHB1-GFP fzf1Δ strains after exposure to DPTA NONOate (Figure 1, D and E). Additionally, time-course response experiments were conducted in YPD media using DBY7283, diploid yhb1Δ/yhb1Δ, and diploid fzf1Δ/fzf1Δ strains starting at OD_1.0 after exposure to 1 mM DPTA NONOate. In addition to the first five time points, cells also were harvested at 120 min (Figure 1, F-H).

Figure 1. DNA microarray analyses. Data from eight independent time-course response experiments were analyzed using the Cluster Program. The lower region is a detailed view of the RNI responsive FZF1-dependent cluster. Although not part of this cluster, the expression (more ...)

A single time-point microarray experiment compared a DBY7283 culture exposed to chemically generated nitric oxide (5 ml of gas per 200 ml of culture) for 120 min to an untreated DBY7283 culture (Figure 1I).

For the Fzf1p overexpression profiling experiments, DBY7283 + pLacZ, and DBY7283 + pFzf1p strains were both grown to OD₆₀₀ 0.4 in Synthetic Complete media without uracil containing 2% raffinose. Galactose was added to a final volume of 2% and fractions were collected 0, 40, 80, and 120 min after addition for each time course.

DeRisi et al., 1997

For each set of time courses, Cy5-labeled cDNA derived from each time point was compared with a Cy3-labeled reference pool representative of the time points within each set of time courses. The time-course profile was then obtained by dividing the ratios for each time point by the corresponding ratio for the 0-h time point.

Microarray data were stored and extracted using the NOMAD database (http://ucsf-nomad.sourceforge.net/). Unflagged spots were included for cluster analysis only if the sum of the median intensities was greater than the median background plus 2 times the SD of the background. The Cluster program (Eisen et al., 1998 ) was used for data analyses and Treeview (Saldanha, 2004 ) was used for data visualization. Microarray data used to generate Figure 1 are in supplemental Table 4, and all raw microarray data from this study are freely available at the National Center for Biotechnology Information Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/). To be included in the nitrosative stress response overview (Figure 1), data needed to be present in all 0-h time points, and transcript levels were required to change greater than 5.6-fold (2^2.5) in two or more time points of the eight time courses shown.

To calculate the correlations to other environmental stress responses, data (Gasch et al., 2000 ) were merged with the DPTA NONOate response data and genes with a greater than 5.6-fold (2^2.5) response in two or more arrays were selected to calculate Pearson correlations (Table 2).

Table 2. Pearson correlations between time-course microarray data sets

Ghaemmaghami et al., 2003

Ausubel et al., 2004

Huh et al., 2003

Exponentially growing DBY7283 pFzf1p, DBY7283 pLacZ, and fzf1Δ pLacZ strains were exposed to DPTA NONOate (0-2 mM) in SG-URA media. After 12 h, final cell density was monitored by OD₆₀₀ and compared with untreated cell density.

Russell et al., 1986

Gietz and Schiestl, 1991

To construct the SSU1 promoter reporter strain AS101, EcoR1 and XhoI sites were introduced into the promoter region -568 to -45 of the SSU1 locus by PCR amplification using the primers 5′-GGAATTCATGTGGAAAAAGAAGGGGTGG and 5′-ATACCGCTCGAGAATTGCGTATTGTCTGAG with genomic DNA as template, and cloned into placZi. This plasmid was then linearized with ApaI and integrated into the ura3 locus in YM4271. AS102, AS103, AS106, and AS108 were constructed similarly, using the underlined restriction sites (Table 3).

Table 3. Strain construction PCR primers

To construct AS104, a SSU1 promoter reporter strain -568 to -45 with the region -389 to -370 removed, the PCR product used to create AS106 was cloned into the plasmid created for the construction of AS103.

The SSU1 CS2 reporter strain AS105 was constructed by annealing the oligos encoding the CS2 with sticky ends 5′-AATTCCTGCAAACTATCATTTTTT and 5′-TCGAAAAAATGATAGTTTGCAGG into the multiple cloning site of pLacZi, which was integrated as described above.

The YHB1 CS2 reporter strain AS109 was constructed as described above for AS105 except oligos encoding the YHB1 CS2 sequence with sticky ends 5′-AATTCTGAAAATGATAGTCTGCGCT and 5′-TCGAAGCGCAGACTATCATTTTCAG were used.

To construct the fzf1Δ strains AS201-209, the FZF1 gene was deleted in AS101-109 and YHB1-GFP using a PCR product constructed using primers 5′-TACGCTGGTGTGCACAAGTGGTACCAGAATACGTGGCCAAAACAATCGGATCCCCGGGTTAATTAA and 5′-ATAGTTCGAATCACATGAGTAGAGGACGGAAATTGCTCTTCTATGGCGTGAATTCGAGCTCGTTTAAAC, with a Pringle kanamycin resistance plasmid as template (Longtine et al., 1998 ). This PCR product also was used to remove the FZF1 gene from the YHB1-GFP strain (Huh et al., 2003 ).

The DBY7283 + pFzf1p galactose-mediated Fzf1p overexpression strain was created by amplifying the complete 900-base pair coding sequence for the FZF1 gene using the primers 5′-ACAATGACGGATATAGGGAGAACCA and 5′-TCAGTATTCGAATAAATCCCAGACGCT, which was cloned into the pYES2.1 vector (Invitrogen, Carlsbad, CA), and the resulting plasmid was transformed into DBY7283. To construct YHB1-GFP + pFzf1p the pYES2.1 Fzf1p plasmid was transformed into YHB1-GFP. To construct DBY7283 + pLacZ the pYES2.1 LacZ plasmid was transformed into the DBY7283 strain. To construct the DBY7283 pLacZ strain, the pYES2.1 LacZ (Invitrogen) plasmid was transformed into DBY7283. To construct the fzf1Δ pLacZ strain, the pYES2.1 LacZ (Invitrogen) plasmid was transformed into fzf1Δ.

Mooradian et al., 1995

A distinct transcriptional response occurred after treatment with 100 μM DPTA NONOate in SCD media. The response was reproducible: an independent biological replicate experiment showed a very similar genome-wide transcriptional response (correlation 0.77). After mock treatment or treatment with the DPTA NONOate breakdown product APD, expression profiles differed substantially from those produced after DPTA NONOate addition (correlation mock to DPTA1, 0.35 and DPTA2, 0.37) (correlation APD to DPTA1, 0.45 and DPTA2, 0.46). The mock and APD time courses were very similar to each other (correlation 0.80) and showed very little response overall (Figure 1, A-D).

Examination of the response to DPTA NONOate treatment revealed that the strongest and most lasting induction occurred in a cluster which contained YHB1, SSU1, and three uncharacterized open reading frames (ORFs): YFL061w, YNL335w, and YNR064c. Yhb1p catalyzes the reaction of ^.NO with oxygen to create nitrate, limiting exposure of the cell to ^.NO (Liu et al., 2000 ). The SSU1 gene encodes a transmembrane sulfite efflux transporter that confers resistance to sulfite when present in multiple copies, and when deleted produces sulfite hypersensitivity (Park and Bakalinsky, 2000 ). YNR064c is a conserved member of the α/β hydrolase superfamily with orthologues in Homo sapiens (25% identical) (1 × e^-9), Arabidopsis thaliana (24% identical) (1 × e^-10), and Mycobacterium tuberculosis (24% identical) (1 × e^-14). YNL335w and YFL061w are homologous open reading frames, which differ by only one silent mutation. They are located within a 15-kb duplication in the telomeric regions of chromosomes 6 and 14. The function of these two genes is unknown although they have homology to a cyanimide hydratase from Myrothecium verrucaria (36% identical) (2 × e^-24).

Examination of other genes with altered expression profiles after DPTA NONOate treatment revealed that the majority belong to a large set of genes associated with the yeast general stress response (Gasch et al., 2000 ). We calculated correlations between the DBY7283 response to DPTA NONOate and other stresses as measured by Gasch et al. (2000 ). The highest correlation was seen with the response to the thiol oxidant diamide (correlation 0.67) and a strong positive correlation also was found with the response to heat shock (correlation 0.61). However, YHB1, SSU1, YNR064c, and YNL335w/YFL061w transcript levels did not change significantly after treatment with diamide, heat shock, or other environmental stresses (Gasch et al., 2000 ), indicating that they may be specific to ^.NO-mediated nitrosative stress. On the basis of this data, we labeled this cluster RNI responsive.

Avram et al., 1999

To confirm that FZF1 was also necessary for the induction of YHB1, SSU1, and the ORFs observed in YPD media, we conducted a time course of transcriptional response after treatment with 1 mM DPTA NONOate in a diploid fzf1Δ strain. Similarly to the response observed in SCD media, transcript profiles of a diploid fzf1Δ strain were comparable to wild type (r = 0.72), except that the FZF1-dependent gene set was not induced (Figure 1H).

Because Yhb1p has been implicated in the metabolism of ^.NO, we wished to determine whether the presence of Yhb1p was necessary for induction of the FZF1 dependent gene set. We conducted a time-course experiment to measure the transcriptional response after 1 mM DPTA NONOate treatment in a diploid yhb1Δ strain. Expression profiles in YPD were similar to wild type (r = 0.80), with the exception that the YHB1 gene was silent because the gene had been deleted. Aside from YHB1, relative mRNA transcript abundance level increases were observed for the FZF1-dependent gene set (Figure 1G). We noted that the stress response after exposure to DPTA NONOate was qualitatively more pronounced in the yhb1Δ strain than in wild type. These results indicate that DPTA NONOate-mediated induction of the FZF1-dependent gene set occurs independently of the YHB1 gene.

Ghaemmaghami et al., 2003

Figure 2. Protein levels after exposure to DPTA NONOate and .NO gas. (A) Western blot using rabbit anti-CBP antibody of TAP-tagged Yhb1p and Ssu1p in the presence or absence of DPTA NONOate (100 μM). (B) Flow cytometry of Yhb1p-GFP of fluorescein isothiocyanate (more ...)

To quantitatively measure the relative abundance change for Yhb1p induction, a YHB1-GFP strain (Huh et al., 2003 ; John Newman, personal communication) was exposed to bolus addition of 100 μl of ^.NO gas generated by acidification of sodium nitrite and monitored by flow cytometry. After exposure to ^.NO gas for 180 min, median fluorescence intensity of the cell population increased by 7.5-fold (Figure 2B). To quantify relative Yhb1p levels as a function of DPTA NONOate concentration, the YHB1-GFP strain was exposed to increasing amounts of DPTA NONOate (2.5-520 μM) and observed by flow cytometry 180 min after treatment. YHB1-GFP fluorescence increased in a dose-dependent manner and at the highest concentration (520 μM) we measured a 10-fold increase in YHB1-GFP fluorescence (Figure 2C). In a YHB1-GFP fzf1Δ strain, an increase in YHB1-GFP fluorescence was not observed at any concentration of DPTA NONOate, indicating that FZF1 was required for ^.NO-mediated increases in Yhb1p expression (Figure 2C).

Avram et al., 1999

Figure 3. Overexpression of FZF1 results in induction of the RNI-responsive gene cluster. (A) Transcript levels of FZF1-dependent RNI-inducible transcripts after overexpression of either Fzf1p or lacZ (control). For the Fzf1p time course, transcript levels are (more ...)

To determine whether overexpression of Fzf1p also affected levels of Yhb1p, the same GAL-driven FZF1 plasmid was introduced into the YHB1-GFP strain. One hour after galactose induction, YHB1-GFP fluorescence increased by threefold (Figure 3B). In the absence of the Fzf1p overexpression plasmid, this induction was not observed.

Cliften et al., 2003

Kellis et al., 2003

Avram et al., 1999

Figure 4. Comparison of upstream sequences reveals conserved motifs. (A) Multiple alignment of the promoter region of the SSU1 gene for closely conserved Saccharomyces species. Conserved sequences in the four species are shaded. (B) Sequence motif found in promoter (more ...)

To investigate the possibility of a second regulatory motif, we used the MEME analysis tool (Bailey and Elkan, 1994 ) to probe 600 base pairs upstream of each member of the FZF1-dependent gene set. A consensus motif (5′ YGSMNMCTATCAYTTYY) was found in each promoter sequence that also coincided with a second island of sequence conservation (CS2) in the SSU1 promoter (Figure 4, A and B). Strikingly, a search for the CS2 consensus sequence in the promoter regions of the entire S. cerevisiae genome revealed that this motif is only found upstream of the five FZF1-dependent genes.

Figure 5. Two separate conserved sequence motifs are sufficient for DPTA NONOate-dependent reporter activation. (A) Representation of SSU1 and YHB1 promoter constructs cloned into the minimal CYC promoter driving lacZ expression. (B and C) lacZ activity in Miller (more ...)

As noted above, the other members of the FZF1-dependent gene set, including YHB1, do not contain a recognizable CS1 motif. To test whether the CS2 motif found in YHB1 was sufficient to drive the FZF1-dependent gene set, LacZ reporters were constructed using either the full-length (AS108) upstream sequence or the YHB1-derived CS2 motif (AS109) (Figure 5A). Consistent with the results for SSU1, the wild-type upstream sequence and the YHB1 CS2 sequence were capable of mediating an FZF1-dependent response (3.4- and 8.8-fold, respectively) to DPTA NONOate treatment (Figure 5C).

Liu et al., 2000

Figure 6. Genotype-dependent sensitivity and resistance after DPTA NONOate treatment. (A) Hypersensitivity to growth inhibition in response to DPTA NONOate (0-1 mM) in YPD media for yhb1Δ, fzf1Δ and ssu1Δ strains compared with wild type. (more ...)

In SCD media, both wild-type and fzf1Δ strain growth was more severely affected by DPTA NONOate treatment. Interestingly, a yhb1Δ mutant strain revealed a growth inhibition that was indistinguishable from wild-type, yet deletion of SSU1 resulted in hypersensitivity. Although the relative sensitivity of yhb1Δ and ssu1Δ strains seems reversed in SCD media, deletion of FZF1 led to an intermediate phenotype similar to what was observed in YPD media (Figure 6B).

To determine whether overexpression of Fzf1p would lead to a physiologically relevant response, we compared a GAL1-10-driven FZF1 strain to a control GAL1-10-driven LacZ strain after exposure to DPTA NONOate (0.1-2 mM). Overexpression of Fzf1p resulted in a protective advantage relative to wild type in range of 0.5 to 1.5 mM DPTA NONOate, whereas deletion of FZF1 in the LacZ control strain resulted in hypersensitivity to DPTA NONOate (Figure 6C).

S. cerevisiae uses a complex network of signaling systems and transcriptional regulons to recognize and respond to environmental pressures. We report the transcriptional response of S. cerevisiae to ^.NO-mediated nitrosative stress. Although treatment with DPTA NONOate led to a general stress response common to other perturbation experiments, DPTA NONOate and ^.NO gas also induced a genetically separable, physiologically relevant set of genes. This set of genes is comprised of YHB1, SSU1, YNR064c, and YNL335w/YFL061w.

Yhb1p catalyzes the reaction of ^.NO with oxygen to create nitrate, limiting the exposure of the cell to ^.NO and thus confers a growth advantage after ^.NO treatment in YPD media (Liu et al., 2000 ). In contrast to previous studies in which S. cerevisiae was exposed to DETA NONOate (Liu et al., 2000 ), nitrosoglutathione (Ullmann et al., 2004 ), or sodium nitrite (Ullmann et al., 2004 ), we found that both mRNA and protein levels of YHB1 were responsive to ^.NO exposure. This discrepancy might be due to insufficient ^.NO availability, cytotoxic ^.NO effects, media effects, or differences in strain background. In addition, we found that the deletion of the YHB1 gene seems to lead to an increase in the general stress response after DPTA NONOate treatment yet does not affect the induction of SSU1 and the other ORFs in the RNI-responsive gene cluster.

Ssu1p has been reported to be a sulfite efflux transporter and to be located at the outer membrane (Park and Bakalinsky, 2000 ). We found that the SSU1 transcript and protein levels increased after nitrosative stress. Furthermore, the presence of the SSU1 gene conferred a growth advantage after exposure to DPTA NONOate in SCD media. We speculate that in addition to transporting sulfite, SSU1 also may transport ^.NO metabolites out of the cell.

The predicted ORFs YNR064c and YNL335w/YFL061c also were induced after nitrosative stress. The presence of these genes did not confer significant growth advantages (our unpublished data). The function of these genes with regard to nitrosative stress remains unclear. It is possible that these ORFs may have roles in ^.NO detoxification that were not revealed by the laboratory conditions used for the growth advantage assay.

FZF1 encodes a Zn-finger DNA binding transcription factor necessary for SSU1 transcription, and until this study, SSU1 was the only known target. We found that the FZF1 gene was required for the induction of YHB1, SSU1, and the other ORFs after nitrosative stress. Furthermore, overexpression led to induction of YHB1, SSU1, and the other uncharacterized ORFs of the RNI-responsive gene cluster. Importantly, the Fzf1p-overexpressing strain exhibited a growth advantage relative to wild type after DPTA NONOate treatment. Deletion of FZF1 resulted in a growth disadvantage that was less profound than the growth disadvantage caused by deletion of YHB1 in YPD media, or deletion of SSU1 in SCD media. This is likely due to FZF1-independent basal transcription of YHB1 and SSU1.

Overexpression of Fzf1p resulted in increased transcription of the target genes, despite FZF1 mRNA levels remaining constant after DPTA treatment. The mechanistic explanation for these observations may be similar to the regulation of Pho4p and other transcription factors for which phosphorylation or some other posttranslational modification leads to a differential subcellular localization. The situation could be analogous in that the overexpression of Fzf1p may stoichiometrically outcompete a regulatory mechanism that would ultimately result in transcriptional activation.

Although it may be that additional components are required for sensing ^.NO or ^.NO-derived metabolites before activation of Fzf1p, it also may be the case that these capabilities are inherent to Fzf1p alone. It is plausible that nitrosylation of Fzf1p could lead to its modulation as an activator. Interestingly, induction of the FZF1-dependent gene set did not occur after treatment with the thiol oxidant diamide, heat shock, or in other oxidative stresses (Gasch et al., 2000 ).

Previous studies have shown that after treatment with methyl methane sulfonate (MMS), SSU1, YNR064c, and YNL335w/YFL061c are induced raising the possibility that DNA damage also may activate the FZF1-dependent gene set. However, YHB1 mRNA was not significantly induced after this treatment (Gasch et al., 2001 ), and a genome-wide screen of the deletion library for MMS hypersensitivity did not find that ssu1Δ, yhb1Δ or fzf1Δ strains were sensitive (Chang et al., 2002 ).

It has previously been reported that Fzf1p specifically binds the SSU1 promoter in vitro (Avram et al., 1999 ). The conserved sequence motif CS1 is contained within the region protected from DNAse I cleavage, yet this sequence could not be located in the promoter regions of the other ^.NO-responsive genes. Sequence comparisons revealed a CS2 in the promoters of the FZF1-dependent gene set that was sufficient for DPTA NONOate-mediated, FZF1-dependent induction. These data imply that Fzf1p possesses the ability to interact with at least two distinct consensus binding sequences, given that CS1 and CS2 have no obvious similarity. Because we have not shown a direct biochemical interaction between CS2 and Fzf1p, it is formally possible that induction via CS2 is an indirect effect of Fzf1p action. Further in vitro and in vivo DNA binding studies will directly address this issue.

The growth inhibition effect of ^.NO-mediated stress seems to be partially dependent on environmental factors and understanding the effect of growth conditions is essential for proper interpretation of these assays. In minimal media, 10- to 15-fold less DPTA NONOate or ^.NO gas was necessary to induce a specific response compared with experiments conducted in YPD. An obvious difference between the composition of YPD and SCD includes higher thiol concentrations, which may account for the differential sensitivity of yeast in these two media.

S. cerevisiae is a relevant model organism for studying the response to ^.NO because pathogenic fungi C. albicans and C. neoformans are likely to use similar molecular signaling mechanisms to induce Yhb1p levels in response to ^.NO. The orthologous flavohemoglobin in C. albicans has already been shown to respond to ^.NO (Ullmann et al., 2004 ), and the C. albicans genome-wide transcriptional response bares significant similarity to the response we observe in S. cerevisiae, although the corresponding transcription factor has not been identified (Bethann Hromatka, personal communication). Further dissection of the mechanism by which S. cerevisiae senses and responds to ^.NO may shed light on this important molecule.

We thank Anne Sarver, Charlie Kim, Anita Sil, Paige Nittler, members of the DeRisi laboratory, Bethann Hromatka, and Alexander Johnson for careful reading of the manuscript and discussions. This work was supported by an award from the David and Lucile Packard Foundation.