Sky1 regulates the expression of sulfur metabolism genes in response to cisplatin

Cisplatin is commonly used in cancer therapy and yeast cells are also sensitive to this compound. We present a transcriptome analysis discriminating between RNA changes induced by cisplatin treatment, which are dependent on or independent of SKY1 function – a gene whose deletion increases resistance to the drug. Gene expression changes produced by addition of cisplatin to W303 and W303-Δsky1 cells were recorded using DNA microarrays. The data, validated by quantitative PCR, revealed 122 differentially expressed genes: 69 upregulated and 53 downregulated. Among the upregulated genes, those related to sulfur metabolism were over-represented and partially dependent on Sky1. Deletions of MET4 or other genes encoding co-regulators of the expression of sulfur-metabolism-related genes, with the exception of MET28, did not modify the cisplatin sensitivity of yeast cells. One of the genes with the highest cisplatin-induced upregulation was SEO1, encoding a putative permease of sulfur compounds. We also measured the platinum, sulfur and glutathione content in W303, W303-Δsky1 and W303-Δseo1 cells after cisplatin treatment, and integration of the data suggested that these transcriptional changes might represent a cellular response that allowed chelation of cisplatin with sulfur-containing amino acids and also helped DNA repair by stimulating purine biosynthesis. The transcription pattern of stimulation of sulfur-containing amino acids and purine synthesis decreased, or even disappeared, in the W303-Δsky1 strain.


INTRODUCTION
Cisplatin (cis-diaminodichloroplatinum) is commonly used in cancer therapy. The molecular mechanisms of cisplatin cytotoxicity include the binding of the drug to DNA and non-DNA targets, and the subsequent induction of cell death. The binding of cisplatin to DNA may interfere with normal transcription and/or DNA replication mechanisms that would trigger the cytotoxic effect. However, it is known that only 5-10 % of covalently bound cisplatin is found in the DNA fraction, whereas the rest interacts with proteins and other molecules (Basu & Krishnamurthy, 2010;Fuertes et al., 2003). Cisplatin resistance is also a multi-factorial process, which might be influenced by changes in mismatch repair mechanisms or cell signalling, increased cisplatin detoxification, or failure to undergo apoptosis, among others (Basu & Krishnamurthy, 2010;Pérez, 1998).
Saccharomyces cerevisiae has been widely used as a powerful eukaryotic model to find genes related to cisplatin sensitivity (Burger et al., 2000;Wu et al., 2004) or cisplatin resistance (Huang et al., 2005;Wu et al., 2004) because libraries of mutant strains containing single deletions of all the genes present in this yeast genome are available. Frequently, genes identified in yeast and human models are coincident. This is the case with SKY1 and SRPK1, which encode proteinspecific serine-arginine (SR) kinases related to the cytotoxicity caused by cisplatin in yeast and human, respectively (Schenk et al., 2001). SRPK1 shows 53 % sequence similarity and is a functional homologue of SKY1, as shown by heterologous complementation of the cisplatin-resistant phenotype of a Dsky1 strain (Schenk et al., 2001).
Interestingly, the responses to cisplatin and to a SKY1 deletion both involve alterations in DNA repair pathways (Schenk et al., 2002). Among the known targets of Sky1 are proteins related to mRNA export from the nucleus (Gilbert et al., 2001;Windgassen & Krebber, 2003;Windgassen et al., 2004) and RNA splicing (Dagher & Fu, 2001;Shen & Green, 2006). Therefore, important changes in cellular mRNA levels as a consequence of SKY1 deletion should be expected.
Recently, the potential of transcriptome analyses to compare gene expression profiles under different conditions has been applied to find specific patterns related to cisplatin sensitivity (Cavill et al., 2011;Wan et al., 2010) or cisplatin resistance (Kim et al., 2011;Wan et al., 2010) in cancerous cells. The major advantage of using the yeast model for transcriptome analyses is that it allows the easy and simultaneous testing of cisplatin treatment and the loss of function of a specific gene. In this work, we present a genome-wide transcriptional study of S. cerevisiae W303 and W303-Dsky1 cells in the presence or absence of cisplatin. The effects of cisplatin treatment were further validated by quantitative (q)PCR. The functional significance of the transcriptome response is discussed with reference to the platinum, sulfur and glutathione content, and related to the results of cytotoxicity assays.
The knockout strain W303-Dseo1 was obtained by one-step replacement with the LEU2 marker. The plasmid YCplac111 (Gietz & Akio, 1988) was used as a template to amplify a linear fragment containing the LEU2 gene and two flanking regions of homology to the ORF ends of SEO1 by PCR using the primers AVV199 (ATGTATTCAAT-TGTTAAAGAGATTATTGTAGATCCGCCACGACTCATCTCC) and AVV200 (TTATTTTTCATCAGATACTGATAAGGTTTCAACGTC-GGTTCCGCGCACATTTCC). After transformation of the W303 strain with the amplified fragment, cells were selected in complete media without leucine. The correct replacement in the S. cerevisiae genome was verified by PCR as described previously (Tizó n et al., 1999) using primers designed inside the LEU2 ORF and the flanking regions of SEO1 external to the recombination event.
Cell culture and treatments. The handling of yeast cells was carried out according to standard procedures. Three biological replicates of cultures and treatments were run. The yeast cells were pre-cultured overnight in 10 ml complete synthetic medium (SD) prepared as described by Zitomer & Hall (1976). The following day, the cells were inoculated at an initial OD 600 0.4 in 70 ml SD and grown in 250 ml Erlenmeyer flasks at 30 uC and 250 r.p.m. When cells reached OD 600 0.6, the cultures from each strain were divided into two aliquots of 25 ml (control and cisplatin treatment). Cisplatin was added to the treated cultures at a final concentration of 600 mM. The treatment was done at 30 uC and 250 r.p.m. during 4 h in darkness. The concentration of cisplatin and the time-course of the treatment were established previously in trial experiments. Under the selected conditions, cell survival was .80 % and an increase of 15 % survival was observed in the W303-Dsky1 strain versus the W303 strain.
RNA preparation and microarray analysis. Three biological replicates of cultures and cisplatin treatments were run. RNA was extracted from a number of cells corresponding to OD 600 3 with the Aurum Total RNA Mini kit (Bio-Rad) following the manufacturer's instructions. Concentration and purity of RNA were evaluated by measuring the ratio R5A 260 /A 280 (always in the range 1.7,R,2.1). RNA integrity was evaluated by the RIN parameter (RNA integrity number) with the 2100 Bioanalyser (Agilent Technologies) according to the manufacturer's instructions. RIN was~9 in all the samples.
Twelve GeneChip Yeast Genome 2.0 arrays from Affymetrix were used and processed in the GeneChip system with Autoloader from Affymetrix. We started from 100 ng total RNA from each sample for successive cDNA, antisense RNA generation, labelling with biotin and fragmentation using the GeneChip 39 IVT Express kit. RNA fragmentation was monitored with the 2100 Bioanalyser, selecting conditions producing fragments from 35 to 200 nt with the majority 100-120 nt. Hybridization, washes and staining were done with the GeneChip HT Hybridization, Wash and Stain kit (Ambion). These kits included RNA poly-A controls (lys, phe, thr and dap) from Bacillus subtilis to monitor the target labelling process, and they served as sensitivity indicators of target preparation and labelling efficiency. They also included the hybridization controls, which comprise a mixture of biotinylated and fragmented RNA of bioB, bioC and bioD (genes from the biosynthesis of biotin in Escherichia coli) and Cre (recombinase from bacteriophage P1). These controls monitored the hybridization, washing and staining steps. Control oligo B2 was included to provide alignment signals for image analysis. Image capture and preliminary data analysis were carried out with Affymetrix Expression Console software (v.1.1).
Statistical data analysis and data mining. Array data were normalized and summarized using the RMA algorithm from Affymetrix. The data were analysed using the web suite Babelomics (version 4.3) (Medina et al., 2010). Statistical analyses to identify differentially expressed genes (DEGs) were performed by using the LIMMA test (Smyth, 2005), a moderated t-test. The false discovery rate was estimated to correct values for multiple comparison (Benjamini & Hochberg, 1995) using an adjusted P50.05.
Gene descriptions and comparative analyses of lists from DEGs were obtained through Yeast Mine (http://yeastmine.yeastgenome. org/yeastmine/begin.do). The functional distribution of genes in the differentially regulated clusters was analysed using FunSpec (http:// funspec.ccbr.utoronto.ca/) developed by Robinson et al. (2002). The MIPS Functional Catalogue Database (FunCatDB) was used in the analyses (http://mips.helmholtz-muenchen.de/proj/funcatDB/). For these analyses, a P,0.01 was selected. In the report of the analyses carried out with FunSpec (Tables 1-4), k is the number of genes from the input cluster in a given category and f is the total number of genes in a given category.
Analysis of expression by qRT (real-time)-PCR. Total RNA isolated as described previously was converted into cDNA and labelled with the KAPA SYBR FAST universal one-step qRT-PCR kit (Kappa Biosystems). PCR primers for individual genes selected after the microarray analysis were designed with the Universal Probe Library Assay Design Center (https://www.roche-appliedscience.com/sis/rtpcr/upl/index.jsp?id=uplct_030000) developed by Roche Diagnostics to generate 60-85 bp amplicons with a T m of 59 or 60 uC. The list of primers is given in Table S9 (available in the online Supplementary Material). The ECO Real-Time PCR System was used for the experiments (Illumina) and calculations were made by the 2 -DDCt method (Livak & Schmittgen, 2001). Three independent RNA extractions were assayed for each strain or condition. The mRNA levels of the selected genes were corrected by the geometric mean of the mRNA levels of HHO1, TAF10 and ALG9 -a selection of control genes which were verified previously to be constitutive in the assayed conditions and not affected by the Dsky1 deletion. A t-test was applied to evaluate the differences between DC t values (C t values normalized with reference genes) of control and treated samples with a P50.05. Cellular accumulation of cisplatin and determination of sulfur content. The yeast cells (W303 and derivatives, W303-Dsky1 and W303-Dseo1) were pre-cultured overnight in 10 ml SD prepared as described above. The following day, the cells were inoculated at initial OD 600 0.4 in SD and grown at 30 uC and 250 r.p.m. When cells Table 1. Functional gene groups over-represented among genes whose expression in the W303 strain treated with cisplatin was higher than in the untreated strain reached OD 600 0.6, the cultures from each strain were treated with cisplatin at a final concentration of 600 mM. The treatment was done at 30 uC and 250 r.p.m. during 4 h in darkness. Cells from 10 ml of each culture were harvested by centrifugation and washed 3 times with ice-cold PBS. The pellets were resuspended in 1 ml PBS. An aliquot of 0.1 ml cell suspension was utilized for the protein assay and the remainder was digested in 70 % nitric acid. Cell lysates were heated for 2 h at 65 uC, diluted to 5 % nitric acid, and assayed for platinum and sulfur content by means of high-resolution inductively coupled plasma MS. An aliquot of 10 ml of each culture was also used for DNA extraction as described by Hoffman & Winston (1987). The final pellet was resuspended in 0.5 ml PBS and the DNA concentration was measured. The samples were digested in 70 % nitric acid for 2 h at 65 uC, diluted to 5 % nitric acid and assayed for platinum as described above. Two biological replicates and two technical replicates were carried out. A t-test was applied to evaluate the differences between means. Statistically significant changes and their P values are indicated in Fig. 3.
Cytotoxicity assay. Sensitivity to the cytotoxic effect of cisplatin was assessed using a colony formation assay. Cultures (1 ml SD) containing a total of 6610 6 cells were exposed for 4 h to cisplatin at concentrations of 0, 0.5, 1.0, 2.0, 2.5 and 5 mM, washed once in PBS, resuspended in YPD medium (2% glucose, 2% Bacto peptone, 1% yeast extract), diluted 1 : 4000, and plated onto YPD agar plates. After 2 days of growth at 30 uC, the number of colonies was counted. The IC 50 was defined as the drug concentration that reduced the number of c.f.u. to 50 % of the value in a control culture not exposed to the drug. Each experiment was repeated three times with duplicate cultures for each drug concentration. A t-test was applied to evaluate the differences between means.
Glutathione determination. The yeast strain W303 and its derivatives, W303-Dsky1 and W303-Dseo1 were treated with 600 mM cisplatin as described above. Aliquots of 50 ml of each culture were harvested and cells resuspended in 10 ml ice-cold PBS. Cells were washed twice with ice-cold PBS solution and resuspended in 350 ml PBS. Aliquots of 50 ml were reserved for protein determination (Bradford, 1976). To the rest, 300 ml 10 % sulfosalicylic acid was added. This cell suspension was mixed with 600 ml glass beads and cells were lysed by vortexing for 3 min, using 30 s pulses with intervals of 30 s on ice. The extracts were spun down in a microfuge at 4 uC to remove cell debris and protein precipitate. The supernatant was then diluted 1 : 5, and total glutathione and GSSG were determined using the 5,59-dithio-bis(2-nitrobenzoic acid) (DTNB)-GSSG reductase recycling assay as described previously (Rhieu et al., 2013). Each experiment was repeated three times with duplicate cultures for each drug concentration. A t-test was applied to evaluate the differences between means.

Effect of cisplatin on the yeast transcriptome
The changes in mRNA levels produced by addition of 600 mM cisplatin in W303 and W303-Dsky1 cells grown in SD media were recorded, normalized and analysed for statistical significance as described in Methods. Only those with a ¢1.6-fold change and a P,0.01 according to the analysis with LIMMA (Smyth, 2005) were considered for further analysis. The original and normalized data from this study are freely available from the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/info/ linking.html); the accession number of the series is GSE41094. Processed data about the DEGs are available in Tables S1-S6). Selected genes were also analysed by qPCR to validate remarkable results as explained along the text.
There is a previous report of a transcriptome analysis describing the effect of cisplatin upon the yeast strain BY4730 (Caba et al., 2005). However, although using the same drug concentration as in our study, the effects observed were limited to only five genes (HUG1, ECM4 and YLR279W upregulated, and HO and SCW11 downregulated). Here, we obtained a set of 122 differentially expressed genes: 69 upregulated (Table S1) and 53 downregulated (Table S2). The observed differences might be explained because the effect of cisplatin treatment is straindependent (Rodríguez Lombardero et al., 2012), and also varies in relation to cell culture and treatment conditions, these being different in the two studies.
The functional distribution of DEGs was analysed with FunSpec as described in Methods. Analysing the effects of cisplatin treatment on W303 cells, functional groups overrepresented among upregulated genes (Table 1) were those related to sulfate assimilation and metabolism of sulfurcontaining compounds, e.g. sulfate transport (GO:0008272), transport of sulfur-containing amino acids (AGP3, MUP1, MUP3, MMP1 and SAM3) or those necessary for methionine (GO:0009086), cysteine (GO:0019344) and S-adenosylmethionine (AdoMet) biosynthesis (GO:0006556). [Gene Ontology (GO) identifiers are given in parentheses.] Several genes related to purine nucleotide biosynthesis were also upregulated (ADE1, ADE17 and MTD1) as well as others from glycine metabolism (GCV3, GCV1 and SMH2) and folic acid biosynthesis (FOL1), as shown in Table 1.
Functional groups over-represented among downregulated genes (Table 2) were related to ribosome biogenesis, and included RPS8A, RPS9B, RPS16B, RPS8B and RPS1B, necessary for maturation of small subunit rRNA from tricistronic rRNA transcript, and RPS17B, NSR1, RPS0A, RPS26A and RPS18B, participating in ribosomal small subunit assembly and/or rRNA export from the nucleus. Among the other genes also downregulated by cisplatin treatment, but not directly related to ribosome biogenesis, were TYE7 (positive regulator of glycolysis), FCY2 (purinecytosine permease), AAH1 (adenine deaminase), ATO2 and MEP3 (ammonium transport), and HO (homothallic switching endonuclease, necessary for mating type switch). FCY2 and HO have been related previously to the yeast response to cisplatin treatment in transcriptome and deletome studies (Caba et al., 2005;Huang et al., 2005).

Effect of Sky1 depletion on the yeast transcriptome
There were 177 genes upregulated (Table S3) and 179 downregulated (Table S4) when SKY1 was deleted. The functional distribution of upregulated genes analysed with FunSpec showed that most of them did not have a previously assigned function. Among the genes with known function (Table 3), there were groups of inter-related genes, which take part in the metabolism of lipids (GO:0006635 and 0006629), response to stress (GO:0006950), those related to cytochrome c haem linkage (GO:0018063), NAD + biosynthesis via the nicotinamide riboside salvage pathway (GO: 0034356) and thiamine biosynthetic process (GO:0009228).
The functional distribution of downregulated genes after SKY1 deletion (Table 4) showed enriched groups, such as those participating in rRNA processing (GO:0006364), ribosome biogenesis (GO:0042254) and positive regulation of transcription from the RNA polymerase I promoter (GO:0045943). Three genes associated with glutamate biosynthesis (GDH3, IDP1 and GLT1) and four genes related to dephosphorylation (PHO11, PHO5, PHO12 and SDP1) were also downregulated, and included in Table 4. A selection of genes regulated by Sky1 in the DNA arrays was validated by qPCR ( Table 5).
Effect of Sky1 depletion on the response of the yeast transcriptome to cisplatin treatment In order to understand the role of Sky1 in the response of the yeast transcriptome to cisplatin, we compared the list of DEGs shown in Tables S1 (genes upregulated by cisplatin in the W303 strain) and S6 (genes downregulated in the W303-Dsky1 treated strain), as well as those from Tables S2 (genes downregulated by cisplatin in the W303 strain) and S5 (genes upregulated in the W303-Dsky1 treated strain). The analysis is represented in the Venn diagrams in Fig. 1, and the functions of the genes found in the intersections are included as Tables S7 and S8. Levels of relative expression of these genes are also shown in Fig. 1. A new issue revealed by this analysis was that the cisplatin-induced upregulation of genes related to the metabolism of sulfur-containing compounds (MET1, MET14, MET16, MET22, MET3, MET32, MET5, MET8, STRE3 and SUL2) was partially dependent on Sky1 function (Fig. 1a, Table S7). Fig. 2 summarizes the metabolic pathways associated with these genes. The expression pattern of selection of genes depicted in Fig. 2 was validated by qPCR and the results (Table 6) confirmed that the increase caused by cisplatin was reversed to a decrease in expression in the comparison of W303 versus W303-Dsky1 treated cells.
The copper transporter Ctr1 has been reported to import cisplatin into yeast cells (Ishida et al., 2002). In the analysis of data from the DNA arrays, we also focused on changes produced upon expression of permease genes that could be related to cisplatin transport. In the transcriptome analysis, the gene encoding a permease and with the highest transcriptional response to cisplatin was SEO1. The changes observed with the DNA arrays in the expression of SEO1 in response to cisplatin (upregulation) and the consequence of SKY1 deletion in this response (downregulation) were also confirmed by qPCR ( Table 6). The actual substrates transported by Seo1 have not yet been described, although it has been suggested that it might be involved in the transport of sulfur compounds (Isnard et al., 1996).
In addition, 11 genes were downregulated in response to cisplatin in a Sky1-dependent manner (Fig. 1b, Table S8). As shown in Table S8, two of these genes encoded subunits of the ASTRA complex and were related to chromatin remodelling (ASA1 and TTI2). Intracellular platinum and sulfur content in the W303 strain and its derivatives, W303-Dsky1 and W303-Dseo1 The yeast strain W303 and derivatives, W303-Dsky1 W303-Dseo1, were treated with 600 mM cisplatin, and intracellular levels of platinum and sulfur were compared with those measured in W303 untreated cells. The levels of platinum incorporation into genomic DNA were also determined. The results (Fig. 3) showed that the content of intracellular platinum after cisplatin treatment did not change significantly in the Dsky1 or Dseo1 deletants, as compared with the W303 strain (Fig. 3a). According to these results and a previous report (Schenk et al., 2002), resistance to cisplatin in Dsky1 cells was probably not produced by decreased intracellular drug concentration. DNA platination was not affected either (Fig. 3b), which indicated that the differences in resistance between the strains were due to signals generated after the formation of DNA adducts. In the W303 and W303-Dsky1 strains, the content of intracellular sulfur did not increase after cisplatin treatment, but in the Dseo1 deletant there was a significant increase after cisplatin treatment compared with the W303 treated strain (Fig. 3c).

Effects of downregulation of the methionine pathway and SEO1 deletion on cisplatin resistance
As reported above for the W303 strain, cisplatin treatment increased the mRNA levels of a series of genes related to sulfur assimilation and the metabolism of sulfur compounds (Table 1) in a Sky1-dependent manner (Figs 1 and  2). Considering that the W303-Dsky1 strain is more resistant to cisplatin than W303, we also tested if the deletion of genes encoding transcriptional regulators of the methionine pathway (MET4, MET28, the two homologues MET31/ MET32 or CBF1) (Lee et al., 2010), as well as the deletion of SEO1, produced changes in the sensitivity to cisplatin. The cytotoxic assays were performed with the selected strains as reported in Methods and the IC 50 was calculated. Results (Fig. 4) confirmed the increased resistance of a Dsky1 strain as reported previously (Huang et al., 2005). No significant changes were observed for the other strains analysed, with the exception of Dmet28, which was more sensitive to cisplatin than the WT.

Intracellular glutathione levels after cisplatin treatment
Levels of glutathione were determined in the W303, W303-Dsky1 and W303-Dseo1 strains as described in Methods. The measured total intracellular glutathione concentration (GSSG+GSH) was in the range of previous determinations in yeast cells (Rhieu et al., 2013). In the absence of cisplatin, the total glutathione content was increased in both mutants as compared with the WT strain (Fig. 5a). Differences in the total glutathione content between treated and untreated cells were not statistically significant in the WT, but significant decreases were found in the mutants (Fig. 5a). In the absence of treatment, values of GSSG were only higher in the W303-Dsky1 strain than in W303 (Fig. 5b). For the W303-Dseo1 treated strain, the comparisons with all the other strains and conditions differed significantly in GSSG content (Fig.  5b). With regard to the GSSG/GSH ratio after cisplatin treatment, the major increase was observed in the W303-Dseo1 strain (Fig. 5c). Cisplatin treatment has been related to oxidative stress (Martins et al., 2008;Pratibha et al., 2006) and changes in the expression of genes encoding enzymes related to oxidation-reduction processes were observed in the yeast strains analysed in this study after cisplatin treatment (Table 1). However, a comparison of the GSSG/ GSH ratios in the W303 and W303-Dsky1 strains could not explain the higher resistance to cisplatin observed in the null mutant by higher levels of available GSH (Fig. 5c).

Sulfur metabolism and transcriptional response to cisplatin treatment
The combination of deletome and transcriptome analyses is useful to identify regulatory processes in response to The P value is indicated when significant differences were found. external stimuli (Jin et al., 2008). A comparison of genes whose expression changed in response to cisplatin in our study (Tables S1 and S2) and those whose deletions changed the sensitivity or resistance of the strain to the drug (Burger et al., 2000;Huang et al., 2005;Wu et al., 2004) confirms that there is no direct correlation between transcriptome and deletome analyses as proposed previously (Birrell et al., 2002;Jin et al., 2008). For instance, several genes related to DNA repair mechanisms, such as RAD or REV genes and others, whose deletions increase cisplatin sensitivity in yeast (Burger et al., 2000;Wu et al., 2004), do not change their mRNA levels after cisplatin treatment in our study. This is consistent with the assumption that many other posttranscriptional events are part of the regulatory responses elicited by cisplatin, which are not detectable in the transcriptome analyses. Nevertheless, our transcriptome approach has identified genes from sulfate assimilation and sulfur metabolism as targets of cisplatin treatment in a response partially dependent on SKY1 function (Tables 1  and 6, Figs 1 and 2), which have not been identified in previous deletome analyses. The reduction of sulfate to sulfide through the sulfate assimilation pathway allows the synthesis of homocysteine. This metabolite is a precursor of methionine and AdoMet through the methyl cycle, and also of cysteine and glutathione through the trans-sulfuration pathway. In turn, AdoMet is required in the biosynthesis of sirohaem, which is necessary for sulfate assimilation. Fig. 2 shows the effects of cisplatin and SKY1 deletion on these pathways as deduced from the transcriptome analysis. As indicated in Fig. 2, increased transcription of MMP1 and MUP3 genes, necessary for the incorporation of methionine or S-methylmethionine from the medium, might also play an important role in the yeast response to cisplatin.
As Met4 is the transcriptional activator of the genes from the assimilation of sulfate and sulfur compound biosynthesis, we also tested if deletion of MET4 or the genes encoding other regulatory proteins (MET31, MET32, CBF1 and MET28), which lack intrinsic transcriptional activation and act as adaptors for recruiting Met4 to appropriate promoters (Blaiseau et al., 1997;Kuras et al., 1996), caused a decrease in cisplatin IC 50 in cytotoxicity assays (Fig. 4). No significant change was observed in most of the deleted strains, thus indicating that the cellular response to cisplatin does not depend on a mechanism related to the transcription factor Met4. We have only found a small, although statistically significant change, from 1.50 to 1.23 mM, in the Dmet28 deletant. It has been proposed that Met4-activated transcription necessarily depends on Met31 and Met32, whereas the dependency on Cbf1 or Met28 selects distinct sulfur metabolic processes (Lee et al., 2010), separating the assimilation of inorganic sulfate (Cbf1) from the synthesis of organic compounds (Met28). Therefore, sensitivity to cisplatin might be affected by dysregulation of the synthesis of specific sulfur organic compounds, rather than by a general increase in all the metabolic pathways downstream of sulfate assimilation.
Although cisplatin elicits a wide transcriptional response affecting genes from the assimilation and metabolism of sulfur compounds, no significant change in total intracellular sulfur content after cisplatin treatment was observed in W303 (Fig. 3c), which might be explained if increased synthesis of sulfur compounds is accompanied by an active export of one or more of them to the medium. One of the genes related to the transport of sulfur compounds and with a high transcriptional response to cisplatin treatment is SEO1. However, the substrate of this permease has not yet been identified (Isnard et al., 1996). The expression of SEO1 in response to cisplatin partially depends on Sky1 (Fig. 1a, Table 6). We have found that the deletion of SEO1 significantly increases intracellular levels of sulfur (Fig. 3c) without affecting platinum levels or DNA platination (Fig. 3a, b). This argues in favour of its participation in the export of sulfur compounds, but not in complex with platinum compounds.
The fact that a cisplatin-induced transcriptional response upregulates the expression of genes related to sulfate assimilation raises the question of the metabolic use of homocysteine for the biosynthesis of glutathione via cystathionine or for the biosynthesis of L-methionine and AdoMet. Cisplatin exhibits a high affinity toward sulfur donors, such as cysteine, methionine or glutathione. Thus, the formation of stable platinum-sulfur bonds with them might contribute to the mechanism of detoxification and resistance. In mammalian cells, data are still controversial with the assumption that glutathione functions in cisplatin chelation and detoxification. As favourable arguments, depletion of glutathione increased sensitization to cisplatin toxicity in osteosarcoma cells (Komiya et al., 1998). In addition, high resistance to cisplatin in human ovarian cancer cell lines was associated with a marked increase of glutathione synthesis (Godwin et al., 1992), and reactions of cis-and trans-[PtCl 2 (NH 3 ) 2 ] with reduced glutathione (GSH) inside intact red blood cells have been reported (Berners-Price & Kuchel, 1990). In contrast, in other studies there was no correlation between cellular resistance and glutathione levels (Boubakari et al., 2004;Bracht et al., 2006). To the best of our knowledge, none of these mechanisms have previously been studied in yeast. We have thus measured the glutathione content in response to cisplatin treatment and observed that there is no increase in total glutathione content after treatment of the W303 strain (Fig. 5a). Accordingly, expression of the genes involved in the biosynthesis of glutathione from cysteine (GSH1 and GSH2) is not enhanced in the transcritome data. In the Dsky1 mutant, the decrease in utilization of homocysteine for glutathione biosynthesis, as suggested by the significant decrease of total glutathione after cisplatin treatment (Fig.  5a), would allow the higher production of cysteine and methionine. This could be advantageous for resistance against cisplatin if, as recently reported, cysteine and methionine are efficient chelants for cisplatin (Zimmermann et al., 2005), whilst major glutathione cisplatin adducts, [Pt(SG) 2 ], can only account for ¡20 % of the total cisplatin adducts formed inside the cell (Kasherman et al., 2009).
If homocysteine is not employed for glutathione synthesis as explained above, it can be expected to be channelled into the biosynthesis of methionine and AdoMet as well (Fig. 2). Thus, in the cellular response to cisplatin it is also possible that AdoMet, derived from methionine, could be used in the biosynthesis of purine nucleotides needed for DNA repair. This view is supported by the transcriptional data reported here. Several genes related to purine nucleotide biosynthesis (ADE1, ADE17 and MTD1) as well as others from glycine metabolism (GCV3, GCV1 and SMH2) and folic acid biosynthesis (FOL1), previously found to be co-regulated with the synthesis of enzymes of the purine biosynthetic pathway (Denis & Daignan-Fornier, 1998), are upregulated after cisplatin treatment (Table 1). This stimulation of purine biosynthesis pathways, as well as that of metabolites consumed during de novo purine biosynthesis like glycine and 10-formyl tetrahydrofolate, could provide the necessary nucleotides for DNA repair mechanisms, which are well known to participate in the response to cisplatin (Basu & Krishnamurthy, 2010). Also supporting our results, transcriptome and metabolome analyses in cancerous cell lines demonstrated a relation between cisplatin sensitivity and nucleotide synthesis by both salvage and de novo pathways (Cavill et al., 2011).

Cisplatin treatment and stress response
Sulfur metabolism is involved in multiple facets of cellular metabolism related to several responses to stress. Indeed, Met4-dependent transcriptional activation of genes related to assimilation and metabolism of sulfur compounds has been reported previously in response to cadmium and other metal-induced stress (Fauchon et al., 2002;Jin et al., 2008). Yeast cells exposed to cadmium adapt to the extreme requirement of sulfur for GSH synthesis by sulfur sparing. Selective translation of proteins with a low content of sulfur, including enzymes of the sulfur metabolic pathway, allows the sparing of sulfur-containing amino acids (Fauchon et al., 2002). The observed transcriptional activation of genes related to assimilation and metabolism of sulfur compounds after cisplatin treatment differs from metal-induced stress (Fauchon et al., 2002), since it is not accompanied by an increase in glutathione levels (Fig. 5).
Activation of genes from sulfur metabolism in response to metals (Fauchon et al., 2002;Jin et al., 2008) or after Met4induced hyper-activation (Lee et al., 2010) has been coassociated with downregulation of ribosome biogenesis genes. This co-association is also observed in our study. The decrease in cell proliferation that usually accompanies the stress might cause the observed downregulation of genes related to ribosome synthesis. We have found that cisplatin treatment and Sky1 depletion (Tables 2 and 4, respectively) cause downregulation of genes related to rRNA processing and ribosome biogenesis. In support of this result, cisplatin inhibits rRNA synthesis in HeLa cell cultures by a mechanism that affects initiation of transcription by RNA polymerase I (Jordan & Carmo-Fonseca, 1998). An interesting result from our study is that in the Dsky1 strain (Table 4) there is a downregulation of three genes (CTK1, UTP15 and NAN1) which encode proteins that cause a positive regulation of transcription from the RNA polymerase I promoter (Table S8). The Sky1-dependent regulation of CTK1 has been verified and confirmed by qPCR (Table 5).

CONCLUSIONS
The transcriptional response to cisplatin and its dependence on Sky1 function has been evaluated. There is a high increase in mRNA levels of genes related to sulfate assimilation and metabolism of sulfur compounds upon cisplatin treatment, which depends partially on Sky1 function. SEO1 expression increases in this response and its deletion increases the intracellular sulfur content.
Several putative mechanisms may explain the increased cisplatin resistance of the W303-Dsky1 strain when considering the data on the platinum, sulfur and glutathione content, and transcriptional regulation of sulfur metabolism. Thus, resistance does not depend on a decrease in the intracellular levels of cisplatin, DNA platination or variations in the GSH/GSSG ratio. Regulated expression of genes from sulfur metabolism, together with the observed decrease of intracellular glutathione levels in the Dsky1 strain, whilst maintaining the total sulfur content, suggests the preferential use of homocysteine for cysteine and methionine production. The formation of cisplatin complexes with these thiols could be envisaged as a mechanism of increased resistance.