The classes of genes whose expression was either induced or repressed under copper starvation in fission yeast are listed in Table 1 (also see Additional data file 1 [Supplementary table 1]). A major group of genes upregulated by BCS addition was involved in metal ion uptake, including genes encoding copper transporters, namely ctr5 and ctr6, which have previously been reported to be induced in states of low copper 364349. Ctr5p is known to form a functional complex with Ctr4p 49. The gene for the latter protein was not represented on the arrays, but it was found to be highly induced (>24×) in a real-time quantitative polymerase chain reaction (qPCR) performed on the same samples used for the microarray experiment (Additional data file 1 [Supplementary table 1]).

A number of predicted flavoproteins, oxidoreductases, and dehydrogenases were downregulated during copper starvation (Table 1). These enzymes catalyze a wide range of biochemical reactions, and their repression may reflect a need for copper in some of these processes. Reduced expression of the antioxidant genes gst2 and sod1, which encode a glutathione S-transferase and a copper-zinc superoxide dismutase, respectively, is not surprising, considering the aforementioned link between copper and the generation of free radicals. Downregulation of sod1 may also result from the reduced availability of copper, which is needed to convert apo-Sod1p to its active form.

Previous expression studies in budding yeast have identified a number of genes that are consistently differentially expressed in varying copper levels 171819. For our comparison with fission yeast, we used a recent microarray time-course dataset that closely matches ours with respect to experimental setup, allowing direct comparison between the two yeasts 19. In this study, four gene clusters were described whose mRNA expression was altered in copper starvation or excess. Three of these clusters contain genes that are involved in copper uptake, copper detoxification, or iron uptake, which are respectively regulated by Mac1p, Ace1p, and Rcs1p/Aft2p (Figure 1). The late induction of the iron regulon in conditions of low copper is thought to result from a secondary iron starvation 171819. A fourth cluster was downregulated after prolonged copper deprivation and contains genes that function in the mitochondrion, including a large component of the respiratory chain. Regulation of this latter group is believed to be linked to a dependency on copper or iron by these metabolic processes 19. Many of the genes that are implicated in copper and iron metabolism in S. cerevisiae have homologs in S. pombe. For this study we used orthologs from a manually curated list 47; when these were unavailable, homologs were identified on the basis of sequence similarity. To determine the extent to which the S. pombe homologs are similarly controlled at the transcriptional level as their S. cerevisiae counterparts, we compared their expression patterns during varying copper conditions. Figure 1 shows a direct comparison between homologous gene pairs in four transcriptional clusters with a specific role in copper or iron metabolism in either yeast. The same gene clusters are used in Figure 2 to summarize how many genes from each group exhibit conserved regulation between S. pombe and S. cerevisiae in response to changing copper and iron availability. In addition, the expression patterns for homologs that exhibit conserved expression in both S. pombe and S. cerevisiae are indicated for direct comparison of the timing and amplitude of expression changes.

When evaluating the transcriptional profiles of budding and fission yeast in response to copper deprivation, a striking difference was observed in the number of differentially expressed genes (Figure 2a). Of the four copper responsive gene clusters described in S. cerevisiae, major expression changes in S. pombe were only observed for homologs to the cluster involved in copper uptake (ctr4, ctr5, ctr6, and SPCC11E10.01; Figures 1 and 2a). The timing of induction of the copper uptake systems is similar in both yeasts, with strong induction of ctr5 in S. pombe and CTR1 in S. cerevisiae over a period of 3 hours (Figure 2a). The marked upregulation of the complete iron regulon in S. cerevisiae, starting after 2 hours of copper deprivation and peaking at 3 hours, is virtually absent in S. pombe, with the exception of str1 and frp1 (Figures 1 and 2a) 4041. Induction of str1 was confirmed in three independent microarray experiments, whereas induction of frp1 was validated by real-time PCR (data not shown), because of missing data in two experiments. The lack of substantial induction of genes involved in iron uptake suggests that, in the experimental conditions used here, copper deprivation does not lead to a significant secondary iron starvation.

To identify putative novel genes involved in iron metabolism, we treated S. pombe cells with the specific iron chelator ferrozine (300 μmol/l). Iron deprivation caused changes in the expression of 56 genes (Additional data file 1 [Supplementary table 2]), which were of much greater amplitude than was found during copper starvation (Figure 2a,b). Many of the induced genes can be directly linked to iron uptake (eight genes) and processing (one gene), whereas those downregulated are involved in metabolic processes, which is consistent with previous reports on S. cerevisiae (Table 1) 1950. A large overlap was observed between the cluster of mitochondrial genes in S. cerevisiae and their homologs in S. pombe, supporting the initial assumption that the changes in this cluster after copper deprivation in budding yeast are linked to secondary iron starvation 19 (Figure 2).

A core set of nine S. pombe homologs exhibited conserved regulation as compared with the iron regulon in S. cerevisiae (Figure 2b). These include the five previously identified iron regulated genes (frp1, str3, fio1, fip1, and str1) as well as two predicted novel ones: SPBC947.05c and isu1. Both of these can be directly linked to iron metabolism. Isu1 encodes a scaffold protein that is involved in mitochondrial iron-sulfur cluster biosynthesis 51. SPBC947.05c is predicted to encode a ferric reductase similar to Frp1p, suggesting a role in the reduction of iron before its uptake by the Fip1p-Fio1p complex. Two additional genes encoding a vitamin H transporter (vht1) and a predicted mitochondrial iron transporter (SPAC8C9.12c) are homologous to genes induced as part of the S. cerevisiae iron regulon 171952, but they lack a consensus Fep1p binding site. Considering the conserved regulation between the two yeasts in response to iron deprivation, these genes still represent good candidates for a role in iron metabolism.

An interesting finding was the relatively strong upregulation of ctr5 (4.3-fold) together with the iron uptake system, which may occur to ensure the availability of copper for incorporation into the Fio1p oxidase. In the absence of a putative Fep1p binding site in the promoter region, the mechanism behind this induction is as yet unclear.

The genes induced during copper and iron starvation represent putative novel target genes for the transcription factors Cuf1p and Fep1p, respectively. However, these expression changes can also be the result of additional regulatory mechanisms, given the involvement of copper and iron in several metabolic pathways 53. We therefore searched for Cuf1p and Fep1p binding motifs upstream of 11 genes that were upregulated in low-copper conditions and 32 genes that were upregulated in low-iron conditions (Additional data file 1 [Supplementary tables 1A and 2A]). Seven genes contained one or more copies of the CuSE binding motif, which may reflect direct regulation by Cuf1p (Figure 3). Putative Fep1p binding motifs were found in 21 genes, including five out of the six genes encoding previously identified Fep1p targets (fip1, frp1, fio1, str1, and str3; Figure 3) 41. Most of these genes contain multiple putative Fep1p binding sites, although it has been shown that only one of these motifs is sufficient to confer iron dependent regulation by Fep1p 42.

The role of Fep1p and Cuf1p in regulating the putative novel target genes was further evaluated by examination of the expression levels of these genes in cuf1-Δ and fep1-Δ mutants using qPCR (Figure 3). For this purpose, the deletion strains and a wild-type control were grown in yeast extract (YE) rather than Edinburgh minimal medium (EMM) medium, because cuf1-Δ and fep1-Δ growth was found to be impaired in the latter medium 42. Genes were considered valid Cuf1p targets when they exhibited a significant (P ≤ 0.05) and greater than 1.5-fold decrease in expression relative to the wild-type control. The same cut-offs were used to identify putative Fep1p targets, with the exception that induced genes were considered instead, which is consistent with the role of Fep1p as a repressor. We further subjected the cuf1-Δ and fep1-Δ mutants to conditions of copper and iron deprivation, respectively. The absence of significant additional expression changes relative to standard conditions (Figure 3) confirms that the observed target gene regulation is indeed conferred by the copper or iron responsive transcription factors, as opposed to indirect effects related to a reduction in metal availability.

Based on our stringency cut-offs, we can identify two novel Cuf1p targets, namely pex7 and SPAC3G6.05, both of which are predicted to encode peroxisomal proteins. This strongly suggests a role for this organelle in S. pombe copper homeostasis, perhaps linked to its function in reactive oxygen species metabolism 54. Consistent with previous observations 43, frp1 and str1 were significantly induced in the cuf1-Δ mutants. This probably results from a secondary iron starvation in S. pombe and is further discussed below.

The eight novel regulatory targets for Fep1p exhibit a clear functional link to iron metabolism. The genes sib1 and sib2 both encode proteins that were previously implicated in siderophore biosynthesis 55, and our findings confirm that S. pombe induces production of siderophores in iron limiting conditions. The expression of isu1 points to a link to iron-sulfur biosynthesis, which may further involve the sulfiredoxin Srx1p. SPBC947.05c is predicted to have ferric-chelate reductase activity based on sequence similarity, and it is expected to play a role in iron reduction before uptake, analogous to Frp1p. The role of the remaining proteins (Rds1p, SPAC1F8.02c, and SPBC27B12.03c) in iron homeostasis is currently unclear. The considerable induction of SPAC1F8.02c, greater than that for all previously identified Fep1p targets, indicates that this glycoprotein plays an important role in iron uptake.

Exposure of fission yeast to limited copper stress (2 μmol/l CuSO₄) resulted in a rapid (within 15-30 min) but transient transcriptional response involving 93 genes (Figure 2c and Additional data file 1 [Supplementary table 3]). When copper levels were increased to 25 μmol/l CuSO₄, this number rose dramatically to 1,259 genes, and the expression changes persisted for the 2-hour time course, reaching a plateau after 30 min (Figure 2c). The size of the response suggests additional cell stress at these copper levels and is likely to result from secondary effects of elevated copper levels. Considering that S. pombe is able to sustain growth in copper concentrations up to 10 mmol/l 56 and that growth rate was not impaired compared with standard conditions (data not shown), the observed expression changes indicate a physiologic response to copper rather than cytotoxic effects. We focused on the genes that were also differentially expressed in the limited copper experiment, because they were the first to respond to high-copper stress and are therefore more likely to represent direct copper-specific regulation.

The global character of the S. pombe gene expression response to medium and high copper levels is in stark contrast to the limited expression changes found in S. cerevisiae cells treated with copper (Figure 2c). Notably, the changes in fission yeast already occur at much lower levels of copper (2 μmol/l versus 8 μmol/l). The genes that are induced by high copper levels are involved in a variety of functions (Table 1). As expected, these include antioxidants with an established role in heavy metal detoxification such as glutathione S-transferase (SPAC688.04c and SPCC965.07c), thioredoxin (SPBC12D12.07c and trx2), zinc metallothionein (zym1), and superoxide dismutase (sod1).

Interestingly, a number of iron uptake genes, including frp1, str1, and fip1, were induced in response to high copper (Figures 1 and 2c), which is consistent with previous findings 43. A small and transient induction of iron metabolism genes was also observed in budding yeast, peaking after a 30 min exposure to 8 μmol/l CuSO₄ (Figure 2c). The same group, however, is also known to be upregulated in response to other stressors such as cadmium or hydrogen peroxide, with the exception of fip1, which is downregulated 47. Regulation of these genes may therefore be the result of general stress and unrelated to copper metabolism. Another possible explanation for the induction of iron regulon genes is that excess copper triggers iron starvation by competing with iron uptake. It is known that the low-affinity Fet4p iron transporter in S. cerevisiae can be inhibited by elevated concentrations of cobalt and cadmium 57. Fet4p and its S. pombe ortholog (SPBP26C9.03c) may well be similarly affected by copper.

A large proportion of the genes (41%) exhibiting changes in high copper are part of the core environmental stress response (CESR) 47, which is known to be activated in response to several distinct stress conditions (Figure 2c). The major conserved regulators of this general stress response in S. pombe that have been identified to date are the Sty1p kinase and the transcription factor Atf1p. Sty1p is turned on as part of a mitogen-activated protein kinase cascade by a variety of stressors 5859606162. The resulting transcriptional changes are effected, at least in part, by Atf1p, which is phosporylated by Sty1p 6364656667. The majority of the induced CESR genes were indeed part of the set of known Sty1p or Atf1p regulated genes (27 out of 38) 47, suggesting an important role for these proteins in the regulation of at least part of the response to high copper.

Considerable overlap was also found with genes previously described to be induced in response to the heavy metal cadmium 47, and almost all of the genes expressed in response to high copper were also induced by cadmium (data not shown). In particular, genes involved in the sulfur amino acid biosynthetic pathway (Table 1 and Additional data file 1 [Supplementary table 3A]), which is required for both glutathione and phytochelatin synthesis, were upregulated in both experiments. Expression of the S. pombe phytochelatin synthase itself (SPAC3H1.10) could not be determined because it did not produce measurable signals at most time points.

Our results further underscore the general nature of the S. pombe response to high copper, even when only a relatively small subset of genes that reacted early to copper stress is considered. From comparisons with previous microarray experiments in S. pombe subjected to environmental stresses 47, however, we can identify a small subset of genes that are specifically downregulated in response to high copper (ptr2, SPBC13A2.04c, SPAP7G5.06, SPAC5H10.01, SPCC132.04c, SPCC1223.09, SPAC11D3.18c, SPAC11D3.15, and SPAC1039.08). Most of these genes are involved in amino acid metabolism.

Wild-type S. cerevisiae is protected from copper stress by the presence of metallothioneins; when copper concentration increases, induction of metallothionein synthesis is sufficient to neutralize the toxic effect of the metal and prevent oxidative stress. This can be inferred from absence of additional stress induced genes in the S. cerevisiae response to high copper levels 19 (Figure 2c). When metallothionein synthesis cannot be initiated (for example, because of lack of the transcription factor responsible for their activation, as in an ace1-Δ strain), free copper can exert its toxic effect on cellular components, leading to reduced tolerance to high copper 68. Because S. pombe responds to metal accumulation by initiating a general stress response, we were interested to determining whether S. cerevisiae has retained the ability to induce a similar response in the absence of the specific high-copper detoxification system.

Although deletion of ACE1 resulted in a drastic increase in the number of genes that respond to copper stress (212 versus 50 in wild-type cells) as well as the magnitude of their changes (Additional data file 1 [Supplementary table 4]), there were significant differences in the types of genes regulated (Figure 4). Only 6% of the differentially expressed genes were orthologous to the CESR group (named ESR/CER in S. cerevisiae), which accounts for 41% of the S. pombe response to high copper. Even when considering all genes of the S. cerevisiae ESR/CER 4869, this number increases only slightly to 8%. We also directly compared the fission yeast genes induced by high copper levels in the wild-type with those induced in budding yeast ace1-Δ, and we found that only 18 orthologous genes were differentially expressed in both experiments.

Two major classes of genes were induced upon copper stress in ace1-Δ mutants, encoding components of the proteasome and stress response proteins (Table 2). Similar induction of proteasome related genes have been observed in response to diamide (a sulfhydryl oxidizing agent), griseofulvin (antifungal agent), and methyl methanesulfonate (a DNA damaging agent) 487071 and may be indicative of severe stress leading to cell death. The reduction in growth rate observed for ace1-Δ mutants during the 4 hours of exposure to 8 μmol/l CuSO₄ is consistent with this hypothesis. Expression of proteasome genes is also highly induced in S. pombe cells exposed to 25 μmol/l CuSO₄ (data not shown). Taken together, our findings indicate that S. cerevisiae ace1-Δ mutants exhibit a different response to high copper as compared with S. pombe, and this discrepancy may be an important contributing factor to the copper hypersensitivity that has been observed in these mutants 68. Thus, S. cerevisiae cells can only poorly compensate for the absence of metallothioneins, whereas S. pombe cells may have adapted to the lack of a CUP1 ortholog by launching a general stress response.

For the experiments in S. pombe, we used the wild-type strain 972 h-, as well as the leu1-32 h- pREP3X-CUP1 strain, which over-expresses the S. cerevisiae CUP1 metallothionein gene from a plasmid under the regulatable nmt1 promoter 77. The over-expression strain was constructed in this study using a pair of specific primers (CUP1-FWD: 5'-CTCGAGATGTTCAGCGAATTA-3'; and CUP1-REV: 5'-CGTTTCATTTCCCAGAGCAGC-3') and a two-step cloning procedure, as previously described 78. Standard methods 79 were used to culture S. pombe cells in liquid Edinburgh minimal medium (EMM) at 30°C, with shaking at 170 rpm.

For validating the putative S. pombe Cuf1p and Fep1p targets, we used the following strains: wild-type FY435 (h⁺ his7-366 leu1-32 ura4-Δ18 ade6-M210), the cuf1-Δ disruption strain (h⁺ his7-366 leu1-32 ura4-Δ18 ade6-M210 cuf11Δ::ura4), and the fep1-Δ disruption strain (h⁺ his7-366 leu1-32 ura4-Δ18 ade6-M210 fep1Δ::ura4). All strains were grown in YE at 30°C.

The S. cerevisiae ace1-Δ mutant (MATa; met15; ura3; his3 1; leu2) was obtained from the Saccharomyces deletion project (Research Genetics, part of Invitrogen, Carlsbad, CA, USA) 80 and grown in synthetic complete (SC) medium supplemented with 2% glucose (Qbiogene, Irvine, CA, USA). Basal copper (CuSO₄) and iron (FeCl₂) levels were identical in both media, as defined by the manufacturer.

Wild-type S. pombe was grown to an optical density (600 nm) of 0.15 to 0.2 before adding the stimulus, after which samples were collected at regular intervals, depending on the experiment. Cells were harvested by filtration and pellets immediately frozen on dry ice.

The high copper experiment in S. pombe was performed three times independently, using different final concentrations of CuSO₄, namely 2, 10 and 25 μmol/l, and collecting at 0, 15, 30, 60, and 120 min. Conditions of low copper or low iron were induced, respectively, by either adding the copper chelator BCS (Sigma) to a final concentration of 100 μmol/l or the iron chelator ferrozine (Sigma, St. Louis, MO, USA) to a concentration of 300 μmol/l. The low copper experiment was independently repeated three times, whereas the low iron experiment was performed once. All samples were hybridized onto microarrays together with a reference from untreated wild-type cells from the same experiment (time 0).

The S. pombe cuf1-Δ and fep1-Δ experiments were performed twice independently, testing the following experimental conditions: wild-type FY435 untreated, wild-type FY435 100 μmol/l BCS and wild-type FY435 300 μmol/l ferrozine; cuf1-Δ untreated and cuf1-Δ 100 μmol/l BCS; and fep1-Δ untreated and fep1-Δ 300 μmol/l ferrozine. Cells were harvested 180 min after stimulus addition together with the untreated samples grown in parallel.

For the CUP1 over-expression experiment in S. pombe, cells were grown for 2 days in EMM without thiamine for steady-state induction of the nmt1 promoter. CuSO₄ was added to a final concentration of 2 μmol/l, and cells were collected after 30 min. Cells carrying the pREP3X control vector were treated in the same way and used as reference. This experiment was performed once.

S. cerevisiae ace1-Δ mutants were exposed to high copper by adding CuSO₄ (Sigma) to a final concentration of 8 μmol/l, and cells were collected at 0, 15, 60, 120, 180 and 240 min after addition, at an OD₆₀₀ of 0.5 (corresponding to a mid-log growth phase). The experimental setup and concentrations were chosen according to the methods of van Bakel and coworkers 19 to facilitate comparison of expression profiles between the two studies. Samples for microarray analysis were harvested by centrifuging at 2000 g for 3 min, followed by snap freezing in liquid nitrogen.

For S. pombe, total RNA was isolated from all experimental and reference samples using a hot phenol protocol, as previously described 8182. Between 10 and 20 μg total RNA were labeled by direct incorporation of either fluorescent Cy3-dCTP or Cy5-dCTP (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK), and the fluorescently labeled product hybridized to S. pombe cDNA microarrays, as previously described 81. Microarrays were subsequently scanned using a GenePix 4000B laser scanner (Molecular Devices, Sunnyvale, CA, USA) and fluorescence intensity ratios calculated with GenePix Pro (Molecular Devices, Sunnyvale, CA, USA).

For S. cerevisiae, total RNA isolation, cDNA synthesis and labeling, microarray production, and hybridization was done as described previously 83. For each sample, 300 ng cDNA (with a specific activity of 2% to 4% dye-labeled nucleosides) was hybridized for 16 to 20 hours at 42°C. Microarray probes consisted of 70-mer oligonucleotides, and included 3,000 control features and duplicate probes for 6357 S. cerevisiae genes. Slides were scanned in an Agilent DNA Microarray Scanner (model G2565BA; Agilent Technologies, Santa Clara, CA, USA). Spot quantification was carried out using Imagene 4.0 (Biodiscovery, El Segundo, CA, USA).

The entire raw dataset is available from the ArrayExpress database 84, accession number E-TABM-120.

Normalization of the S. pombe microarray data was performed using an in-house script 81. The S. cerevisiae data were normalized by applying a Lowess function per subgrid on all gene spots 85, using the marrayNorm R package v1.1.3 86. Genes for which more than 50% of data points were missing were discarded from further analysis in both datasets.

Genes were considered differentially expressed in the high copper experiment when they exhibited a greater than 1.5-fold change at at least one time point after stimulation with 2 μmol/l CuSO₄ and at least a similar change in response to 25 μmol/l CuSO₄. In the low copper experiment, genes induced or repressed by more than 1.5-fold at at least one time point in two out of three biologic repeats were selected. Similar cutoffs were used for the low iron (300 μmol/l ferrozine) and the high copper (8 μmol/l CuSO₄) experiments in S. cerevisiae ace1-Δ mutants, selecting genes that were induced or repressed by at least 1.5-fold at one or more time points. Hierarchical clustering was done in Genespring 6.1 (Agilent Technologies, Santa Clara, CA, USA). Genes were assigned to functional classes according to the Gene Ontology consortium database 87.

Raw data and lists of S. cerevisiae copper regulated genes were obtained from the report by Van Bakel and coworkers 19. Genes with a prospective S. pombe ortholog were determined using a table of curated orthologs as previously described 88 or by best reciprocal hit BLAST analysis 89. The total number of curated orthologs available at the time of the analysis was 3,655. If an S. pombe ortholog to an S. cerevisiae gene could not be identified based on the above criteria, we instead used the BLAST algorithm 89 to select the most similar S. pombe sequences for our interspecies comparisons of copper metabolism. The best matching hits in S. pombe with an e value less than 1 × e^-20 and a minimum match length of 80% were selected as putative functional homologs.

DNA regulatory patterns were derived from manual alignments of experimentally confirmed binding motifs and determined as KYWGATAW (K = G/T, Y = C/T, and W = A/T) for Fep1p 4142 and WNNNGCTGD (W = A/T, N = any, and D = G/A/T) for Cuf1p 363790. These patterns were subsequently used to search for putative novel binding sites in the 800 base pair region upstream of the transcriptional start site of genes induced in low copper or low iron conditions. When necessary, upstream regions were truncated to prevent overlap with other open reading frames. Both the retrieval of upstream sequences and pattern matching were done at the Regulatory Sequence Analysis Tools website 91.

Putative novel S. pombe targets for Cuf1p and Fep1p were validated by quantitative real-time PCR in cuf1-Δ and fep1-Δ disruption strains. Total RNA was isolated as previously described 81, treated with Turbo DNase (Ambion, Foster City, CA, USA) in order to remove any genomic DNA contamination, and reverse transcribed using random hexamers and Omniscript RT Kit (Qiagen, Venlo, Limburg, The Netherlands), in accordance with the manufacturer's instructions. Expression levels were quantified by using SYBR GreenER qPCR Supermix ABI PRISM (Invitrogen) on an Applied Biosystems (Foster City, CA, USA) 7300 Real-Time PCR System and normalized using act1 expression levels as a reference. Primer sequences used are available as supplementary data (Additional data file 2 [Supplementary table 5]).