Genetic basis of arsenite and cadmium tolerance in Saccharomyces cerevisiae
© Thorsen et al. 2009
Received: 10 October 2008
Accepted: 12 March 2009
Published: 12 March 2009
Skip to main content
© Thorsen et al. 2009
Received: 10 October 2008
Accepted: 12 March 2009
Published: 12 March 2009
Arsenic and cadmium are widely distributed in nature and pose serious threats to the environment and human health. Exposure to these nonessential toxic metals may result in a variety of human diseases including cancer. However, arsenic and cadmium toxicity targets and the cellular systems contributing to tolerance acquisition are not fully known.
To gain insight into metal action and cellular tolerance mechanisms, we carried out genome-wide screening of the Saccharomyces cerevisiae haploid and homozygous diploid deletion mutant collections and scored for reduced growth in the presence of arsenite or cadmium. Processes found to be required for tolerance to both metals included sulphur and glutathione biosynthesis, environmental sensing, mRNA synthesis and transcription, and vacuolar/endosomal transport and sorting. We also identified metal-specific defence processes. Arsenite-specific defence functions were related to cell cycle regulation, lipid and fatty acid metabolism, mitochondrial biogenesis, and the cytoskeleton whereas cadmium-specific defence functions were mainly related to sugar/carbohydrate metabolism, and metal-ion homeostasis and transport. Molecular evidence indicated that the cytoskeleton is targeted by arsenite and that phosphorylation of the Snf1p kinase is required for cadmium tolerance.
This study has pin-pointed core functions that protect cells from arsenite and cadmium toxicity. It also emphasizes the existence of both common and specific defence systems. Since many of the yeast genes that confer tolerance to these agents have homologues in humans, similar biological processes may act in yeast and humans to prevent metal toxicity and carcinogenesis.
The presence of nonessential metals like arsenic (As) and cadmium (Cd) in the environment is prevalent. Since these metals are highly toxic, they pose a considerable threat to nature and to human health. The main routes of poisoning are through occupational exposure or through ingestion of contaminated food and water. Pollution of soils with toxic agents is a common global problem, and contamination of drinking water by arsenic is a major health concern because of the large number of contaminated sites and people at risk. These metals are implicated in a broad spectrum of degenerative conditions in humans, including neurotoxicity, nephrotoxicity, genotoxicity and carcinogenesis. For example, chronic As exposure induces cardiovascular diseases, neurological disorders and liver injury, and is associated with cancers of the skin, bladder, liver and lung. Cd is considered genotoxic and carcinogenic for lung, kidney and prostate [1–4]. The International Agency for Research on Cancer (IARC) classifies both metals as group I carcinogens .
The toxicity mechanisms of As and Cd are not fully understood at the molecular level. In general, they may act by targeting signalling or regulatory proteins that control cell proliferation, differentiation and cell cycle regulation. Although the mode of action of each metal is likely to involve unique features, some toxicity mechanisms may be shared. Similarly, cells may mount both common and metal-specific responses to counteract toxicity [2, 6, 7, 8]. A common property of As and Cd is their high reactivity with sulphhydryl groups. Hence, they can bind to and affect the activity of many proteins. In addition, these metals are known to generate oxidative stress in cells and their toxicity has partly been attributed to their capability to induce formation of reactive oxygen species (ROS). The damage caused by ROS to lipids, proteins and DNA are likely to contribute to As and Cd toxicity [7, 9]. Nevertheless, neither the exact details of metal-induced ROS generation nor the full set of toxicity targets is known.
Drugs containing arsenicals are currently used in medical treatment. The arsenic-containing drug melarsoprol® is used to treat diseases caused by the protozoan parasite Trypanosoma [10, 11], whereas arsenic trioxide (Trisenox®) is used as a treatment for acute promyelocytic leukaemia and it may also be employed for other haematological and solid cancers . However, the emergence of resistance threatens the efficacy of medical treatment  and hence, there is an increasing demand to identify tolerance mechanisms. Similarly, the prospect of using plants to clean up polluted soils has recently attracted considerable attention. Nevertheless, to develop phytoremediation into a useful approach requires a detailed understanding of the genetics and molecular basis of detoxification and tolerance acquisition [14, 15].
Tolerance and detoxification mechanisms often involve extrusion of the toxic ions from the cell, sequestration within internal organelles, chelation by metal-binding proteins, and reduction of uptake. Common to these systems is that they reduce the cellular content of the toxic agent, although their molecular basis may differ between metals and also between organisms [16–19]. Furthermore, it is clear that not only the detoxification systems themselves, but also the proteins that regulate their expression, localization and/or activity will contribute to cellular metal tolerance.
The aim of this work was to provide a global view of the genetic basis of As and Cd toxicity and detoxification by identifying the molecular/cellular targets of their action and to reveal tolerance acquisition mechanisms. Unveiling metal toxicity and tolerance mechanisms in yeast may prove of value for identifying similar mechanisms in higher eukaryotes.
To pin-point cellular functions that protect cells from metal toxicity, we searched for functional categories (according to FunCat, Munich Information Center for Protein Sequences (MIPS) ) that were significantly enriched (p-value < 0.005) in the sensitive gene set as compared to the whole genome. As was the case with individual genes (see above), this analysis revealed common as well as metal-specific tolerance functions (Fig 1). Prominent among those required for both As and Cd tolerance, were functions related to methionine and sulphur metabolism, mRNA synthesis and transcriptional regulation, environmental sensing, and vacuolar/endosomal transport and sorting. The p-values and fold enrichment factors of genes within individual functional categories are available in Fig 1C and Additional file 1.
Cells respond to As or Cd exposure by inducing expression of genes and enzymes in the sulphur assimilation and glutathione biosynthesis pathways. Moreover, cells also channel a large part of assimilated sulphur into biosynthesis of the low molecular weight thiol molecule glutathione [21–24]. Glutathione is important for metal tolerance since it may protect cells by: 1) metal chelation and sequestration; 2) protection against metal-induced oxidation since glutathione is considered the main redox buffer of the cell; and 3) binding to reactive sulphhydryl groups on proteins (glutathionylation) to shield them from irreversible metal binding and/or oxidative damage [25, 26]. The fact that we identified several metal sensitive mutants having functions in the sulphur assimilation and glutathione biosynthesis pathways as well as transcriptional regulators controlling these pathways, underscores the importance of the sulphur/glutathione biosynthesis pathways for tolerance acquisition. Moreover, the ycf1Δ mutant lacking an ABC-type transporter responsible for sequestration of metal-glutathione conjugates into vacuoles [27–29] was also As and Cd sensitive. The enzymes in the sulphur assimilation/glutathione biosynthesis pathways are transcriptionally regulated by the transcription factors Yap1p and Met4p, and cells lacking any of these factors display As and Cd sensitivity [21, 22]. Met4p is recruited to target promoters by the DNA-binding proteins Met31p, Met32p and Cbf1p forming the complexes Met4p-Met31p/Met32p and Met4p-Cbf1p. An additional protein, Met28p stabilizes the Met4p-containing complexes . Our data confirms the importance of Yap1p for metal tolerance and suggests that different Met4p-complexes may be important during As and Cd exposure; mutants lacking MET28 and CBF1 were As-sensitive whereas cells lacking MET32 and MET28 were Cd-sensitive. Taken together, these findings are consistent with the central role of sulphur metabolism and glutathione biosynthesis for As and Cd tolerance.
As or Cd exposure results in altered expression of a large number of genes [21, 22, 31–33]. Accordingly, mutants defective in mRNA synthesis and transcriptional control were sensitized to both metals (Fig 1). Among the gene-specific regulators that confer metal tolerance was Yap1p (regulates expression of genes with functions in oxidative stress defence and sulphur/glutathione metabolism), Met28p, Cbf1p, Met32p (sulphur/glutathione metabolism) and Rpn4p (protein degradation). Transcriptional activators may stimulate gene-specific expression by recruiting the co-regulator complexes Mediator, SAGA and/or Swi/Snf as well as RNA polymerase II to promoters. Indeed, several metal-sensitive mutants lacked subunits of RNA polymerase II (rpb4Δ, rpb9Δ) or components of the Mediator (pgd1Δ, srb8Δ), SAGA (gcn5Δ, ada2Δ, spt7Δ), and Swi/Snf (swi3Δ, snf2Δ) co-regulator complexes. Although sensitivity of these mutants could be a result of a general impairment of transcriptional activity, both Yap1p and Met4p interact physically with components of these co-regulator complexes [34, 35].
Cells defective in functions related to environmental sensing and signalling were metal sensitive (Fig 1). The limited overlap between the As and Cd sensitive mutants within this category suggests that cells may use distinct sensing/signalling pathways in response to these metals. For instance, cells appear to use distinct MAP kinase pathways for sensing/responding to As and Cd since mutants defective in the cell integrity pathway (slg1Δ, bck1Δ, slt2Δ) were preferentially Cd sensitive whereas cells defective in the osmosensing HOG pathway (ssk1Δ, ssk2Δ, pbs2Δ, hog1Δ) were preferentially As sensitive. We note that cells lacking the HOG pathway components Ssk1p and Hog1p were also Cd sensitive. The As and Cd-sensitivity of the latter mutants may reflect different functions/targets of the corresponding proteins under the two stress conditions. Hog1p is phosphorylated in response to As and contributes to tolerance by at least two mechanisms; first, Hog1p restricts arsenite influx through the aquaglyceroporin Fps1p by phosphorylating this protein (deletion of FPS1 results in arsenite resistance) [36, 37]. Secondly, Hog1p regulates the exit from As-induced G1 arrest . Hog1p is also phosphorylated in response to Cd  but the target(s) and mechanism(s) through which Hog1p mediates Cd tolerance are unknown. Interestingly, Hog1p has a role in the cell integrity pathway and the cellular response to cell wall damage involves both the Hog1p and Slt2p/cell integrity MAP kinase pathways . The fact that cells lacking HOG1 or components of the cell integrity pathway are Cd sensitive, suggests that the role of Hog1p under Cd exposure may be linked to cell wall damage.
The secretory pathway is implicated in many important processes such as lipid biosynthesis, protein targeting and secretion, as well as the unfolded protein response. Consequently, an intact secretory pathway is crucial for the cell to withstand many different environmental conditions. Indeed, mutations impairing proper functioning of the secretory pathway cause sensitivity to a broad range of treatments and growth conditions [41, 42]. Metals may impact one or more aspects of the secretory pathway; first, metals may enter the endomembrane system and disturb various processes by interacting with proteins and/or lipids. Secondly, failure to maintain appropriate secretory pathway function may cause metal sensitivity due to defects in delivery or targeting of transmembrane transporters, sensors or signalling receptors. Thirdly, since vacuolar sequestration is important for tolerance acquisition, a decreased capacity to do so is likely to contribute to the observed sensitivity of secretory pathway mutants.
Arsenite-specific defence functions that were significantly enriched (p-value < 0.005) in our data set were related to cell cycle regulation, lipid and fatty acid metabolism, mitochondrial biogenesis, and the cytoskeleton. The p-values and fold enrichment factors are available in Fig 1C and Additional file 1.
Deletion of genes encoding cell cycle related functions produced As-sensitivity; those genes encompassed functions involved in regulation of cell cycle progression (HOG1, PHO85, PIN4, DBF2, CLB2), as well as spindle body and microtubule formation (BIK1, BIM1, CNM67, JNM1, CIN8, CIK1, CSM1, BNI1, CIN1, CIN2, PAC2, TUB3, SPC72). This indicates that cell cycle progression may be targeted by As, confirming our recent finding that As exposure triggers transient G1 and G2 delays in yeast and that cells defective in Hog1p signalling cannot exit the arsenite-induced G1 checkpoint arrest . Similarly, the mechanisms of As action in cancer cells include cell cycle arrest at G1 and G2 phases, mitotic arrest and subsequent apoptosis. Arsenite can interrupt mitosis by interfering with tubulin polymerization or by disrupting mitotic spindles [12, 43]. Elucidating the molecular details of arsenite-induced cell cycle regulation and how this relates to As toxicity and carcinogenicity is important for understanding the potential of As-containing drugs in medical therapy.
Many mitochondrial proteins are known to be targeted by arsenicals . Indeed, we found several As-sensitive mutants defective in mitochondrial function and biogenesis, including mitochondrial protein biosynthesis and genome maintenance. This observation is consistent with the notion that arsenic toxicity targets mitochondrial processes and that impaired oxidative phosphorylation causes As sensitivity [12, 45, 46].
Deletion of genes encoding functions in lipid and fatty acid metabolism impaired As tolerance; these genes were related to ergosterol (ERG24, ERG2, ERG3, ERG4, ERG6), phospholipid (OPI1, OPI3, ARV1) and sphingolipid (DPL1, CSG2, SUR4, DEP1) biosynthesis. Trace elements and arsenicals appear to affect the fluidity and the thermotropic properties of cell membranes [47, 48]. Hence, As may have a direct impact on the lipid bilayer. Alternatively, the lipid composition of cell membranes might be altered in the mutants listed above, thereby changing the properties of the membrane and impacting on the activities of membrane-embedded transporters, sensors or signalling receptors . Interestingly, the activity of the aquaglyceroporin Fps1p, through which arsenite enters cells [36, 37] is affected by the plasma membrane ergosterol content . Moreover, an fps1Δ mutant has reduced whole-cell and plasma membrane ergosterol levels compared to wild-type cells . Hence, altered plasma membrane ergosterol levels may change Fps1p activity to facilitate arsenite entry.
Loss of genes encoding cytoskeletal proteins caused As sensitivity. In particular, mutations affecting the function of the heterohexameric GIM (Genes Involved in Microtuble biogenesis)/prefoldin protein complex, which is required for formation of functional tubulin, sensitized cells to As. These genes include YKE2 (GIM1), PAC10 (GIM2), GIM3, GIM4, GIM5, and PFD1 (GIM6). Analysis of the entire set of As-sensitive mutants for known protein-protein interactions indicated that the proteins encoded by these genes form a highly interacted molecular machine in cells and that disruption of any one of the components of this complex rendered cells sensitive to As (Fig 2B). The cytoskeleton is a known target of As; As triggers actin reorganization in various mammalian cell lines [51, 52], and directly interacts with actin as well as with the α and β subunits of tubulin [53, 54].
Several other As-specific defence functions were pin-pointed by our analysis (see Additional File 1). Most notable among those is the arsenic resistance gene cluster ACR3-ACR2-YAP8 . ACR3 (also called ARR3) encodes an arsenite export protein , ACR2 (ARR2) an arsenate reductase  and YAP8 (ACR1/ARR1) encodes an arsenic-responsive transcription factor that controls expression of the former two genes [58–60].
Cadmium-specific defence functions that were significantly enriched (p-value < 0.005) in our data set were mainly related to sugar/carbohydrate metabolism, and metal-ion homeostasis and transport. The p-values and fold enrichment factors are available in Fig 1C and Additional file 1.
Genes encoding functions in metal-ion homeostasis were enriched in the Cd-specific data set including the ion transporter genes PCA1 (Cd exporting ATPase), TRK1 (K+ transporter) and SPF1 (ER-localized Ca2+ ATPase). Interestingly, the majority of the genes in this category encode functions related to iron homeostasis including the iron regulated transcription factors AFT2 and AFT1/RCS1, the multicopper oxidases FET3 and FET5 (involved in iron uptake), FRE6 (ferric reductase), FRE8 (iron/copper reductase) and ISA2 (required for maturation of mitochondrial and cytosolic Fe-S proteins). Importantly, while this paper was being written, Ottonello and co-workers reported that iron addition rescues Cd-sensitivity of certain yeast mutants indicating that Cd interferes with iron homeostasis . This is similar to the situation in plants where Cd exposure was shown to provoke iron deficiency . Indeed, expression of several genes related to iron and heme metabolism (e.g. FET3, ATX1, FRE5, ARN1) is stimulated by Cd [22, 32] and during iron deprivation [63, 64] providing further support to the notion that Cd-exposed cells experience iron deficiency. Since Fe-S cluster homeostasis regulates transcription of the iron homeostasis systems , our data indicate that Cd may affect one or more aspects relating to Fe-S clusters, such as their biosynthesis and/or maturation.
Another Cd-specific defence function appears to involve the glucose sensing Snf1p-pathway since cells lacking SNF1 (homologous to mammalian AMP-activated protein kinases), SNF4, REG1, SIP5 (regulators of Snf1p), and MIG2 (transcriptional repressor targeted by Snf1p) were Cd sensitive (see Additional file 1). Snf1p plays a central role in modulating carbon metabolism under glucose limiting and environmental stress conditions. Snf1p also regulates proteins involved in metabolism of reserve carbohydrates such as glycogen and trehalose . Interestingly, we noted that cells lacking either phosphofructokinase (PFK1 or PFK2) were Cd sensitive. Moreover, deletion of FBP1, whose product acts in the reverse step from that catalysed by Pfk1p and Pfk2p also led to Cd sensitivity. Finally, Cd stimulated expression of many genes with functions in glucose metabolism (GLK1, PDC6, ALD4, VID22, FBP1, FBP26) . Hence, appropriate functioning of glycolysis may be crucial for Cd tolerance. Cd could impact the cell in such a way that flux through the glycolytic pathway is altered. For example, Cd could have a direct effect on glucose uptake systems , on glycolysis itself or on other systems that result in an extra demand on the glycolytic pathway.
In agreement with other studies , we found little overlap between genes necessary for As or Cd tolerance and genes whose expression is induced by As or Cd. However, for some defence systems there appears to be such a correlation. For instance, most genes (and enzymes) of the sulphur assimilation and glutathione biosynthesis pathways were strongly up-regulated in response to As and Cd [21, 22, 32], and many mutants in this pathway were metal sensitive. Another example is genes involved in iron homeostasis: expression of several such genes was induced by Cd whereas their absence resulted in Cd sensitivity. Similarly, expression of the arsenic-resistance genes ACR2 and ACR3 was stimulated by As and their presence was required for As tolerance (see Additional file 1).
We next asked whether the gene expression profiles of As- and Cd-treated cells would be more similar to each other than the As and Cd sensitivity profiles. As stimulated expression of 334 genes (0.2 mM arsenite, 1 hour)  whereas Cd exposure increased expression of 305 genes (0.3 mM Cd, 2 hours)  of which 80 genes were common (see Additional file 1). The overlap in those gene sets was about 14% which is similar to the overlap in the sensitive gene sets (18%). Interestingly, the genes induced by both treatments mainly encode functions in sulphur assimilation and glutathione biosynthesis pathways as well as in (oxidative) stress responses and detoxification, and many are transcriptionally regulated by Met4p and Yap1p. Hence, several processes that contribute to both As and Cd tolerance are transcriptionally stimulated by these metals. We also checked the overlap between the common As/Cd sensitivity gene set (106 genes) and the common As/Cd-induced gene set (80 genes). Curiously, these gene sets only had three genes in common (MET16, MET17, CYS3). Hence, although the two gene sets (phenotype and expression) are enriched for genes encoding related functions, the identities of those genes are largely distinct.
We next asked to what extent the genes in our As- and Cd-sensitive gene set have human homologues. Using the HomoloGene database  we found that 43 of the 106 genes in the common As/Cd sensitive gene set have at least one human homologue (41%) (see Additional file 1). These proteins are involved in diverse functions in humans including signalling, transcription, chromatin modification, and vacuolar protein sorting. Moreover, these genes were significantly (p-value < 0.005) enriched for functions related to carbohydrate metabolism as well as protein targeting and sorting. We conclude that similar biological processes may act in yeast and humans to prevent metal toxicity and carcinogenesis.
Next, we wanted to exploit the information in our data sets to discover toxicity and tolerance mechanisms at a molecular level. To do so, we explored the role of two cellular systems identified above in more detail, one related to arsenite and one related to cadmium.
To further explore how As affects the cytoskeleton, we monitored α-tubulin organization by following GFP-tagged Tub1p and Tub3p. In the absence of As, extended fluorescence arrays were visible in most cells. In contrast, the microtubule arrays decreased dramatically in response to As exposure indicating disassembly of cytoplasmic microtubules (Fig 6C). Also the actin cytoskeleton was strongly affected by As. In untreated wild-type cells, actin patches were predominantly localized to the buds whereas As exposure triggered a temporary depolarization of the actin cytoskeleton with an even distribution of patches in the bud and the mother cell (Fig 6D). After 4 hours, the actin cytoskeleton recovered its polar distribution in nearly all cells and this recovery coincided with growth resumption (Fig 6E). Ssk2p is a kinase that activates the MAP kinase Hog1p in response to osmotic stress. In addition, Ssk2p binds to actin and promotes actin cytoskeleton recovery after osmotic stress . We previously demonstrated that ssk2Δ cells are As sensitive and that this sensitivity is unrelated to Hog1p phosphorylation . Instead, Ssk2p might have a role in actin recovery also during As-exposure; actin remained depolarized in As(III)-treated ssk2Δ cells (Fig 6D) and the mutant could not resume growth (Fig 6E). Taken together, these results provide strong support for the cytoskeleton being a direct target of As toxicity and that As inhibits the activity of the GIM complex in vivo. Hence, arsenite has a dual effect on the cytoskeleton; it binds to and disrupts the actin and tubulin cytoskeleton, and it inhibits the GIM complex that is required for folding of de novo synthesised actin and tubulin monomers.
This study has highlighted the importance of individual genes, pathways and cellular functions that protect yeast cells against arsenite and cadmium toxicity. We have pinpointed common as well as metal-specific defence functions and provided novel insight into As and Cd tolerance systems and toxicity targets. In particular, molecular evidence indicates that the cytoskeleton is a direct target of arsenite toxicity, and that the GIM complex is affected by arsenite in vivo. We also demonstrate that phosphorylation of the Snf1p kinase is necessary for Cd tolerance. These aspects of arsenite/cadmium toxicity and tolerance are novel and add to our understanding of metal action and tolerance acquisition mechanisms.
The fact that As and Cd has been included in some recent genome-wide phenotyping screens [31, 33, 61, 73, 74] enabled us to assess the overlap between and the reliability of these studies. Although the overlap on the gene level was rather poor, the corresponding studies identified similar biological processes that are important for tolerance acquisition. The discrepancy at the gene level might be due to variations in experimental settings and/or due to that the screens are not saturated. Nevertheless, genome-wide phenotypic screens provide biologically relevant insight into tolerance mechanisms, and phenotypic profiling under different conditions may enhance the reliability of the data.
What can we conclude about the mode of action of metals based on a genome-wide sensitivity screen? Even though phenotypic profiling pin-points core functions that confer metal tolerance, one cannot firmly conclude that those gene-products represent direct toxicity targets. In fact, many phenotypes could be a result of synthetic effects, e.g. the lacking gene product and the metal may affect parallel pathways/protective systems. A complementary approach may involve a search for genes that confer resistance upon overexpression since a gene that confers resistance when overexpressed and sensitivity when absent is more likely to be a direct target of the metal. However, it is clear that molecular and biochemical studies will ultimately be required to confirm whether a protein is indeed targeted by a metal or not. In this study, we provide both genomic and molecular evidence that the cytoskeleton and the GIM/prefoldin complex are targets of arsenite toxicity.
To conclude, this study has shed light on the genetic basis of arsenite and cadmium tolerance in S. cerevisiae. This catalogue of genes and protective functions will be instrumental for generating hypotheses about the role of individual factors for tolerance acquisition and for providing insights into the modes of metal action. Unveiling metal toxicity and tolerance mechanisms in yeast may prove of value for identifying similar mechanisms in higher eukaryotes.
The complete set of viable yeast mutants in the haploid strain BY4741 (MAT a his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) and the homozygous diploid strain BY4743 (MAT a /MAT α his3Δ1/his3Δ1 leu2Δ0/leu2Δ0 ura3Δ0/ura3Δ0 LYS2/lys2Δ0 met15Δ0/MET15) from EUROSCARF  were screened for growth in the presence of various concentrations of cadmium chloride (Sigma) and sodium arsenite (Sigma). The strains were grown in the absence or presence of metal on YEPD (1% yeast extract, 1% peptone, 2% glucose) or YNB (0.67% yeast nitrogen base) medium supplemented with complete amino acid supplement mix (BIO101) and 2% glucose as carbon source. Tub1p-GFP and Tub3-GFP strains were obtained from Invitrogen, and the Snf1p pathway mutants (in W303-1A strain background) and the centromeric plasmids containing HA-tagged SNF1 and SNF1-T210A are described in . The MIG1 gene was epitope tagged by chromosomal integration of a PCR amplified triple HA-tag using primers 5'-ACA AAC CCC CAT TTC TCA GTC GGA TTC ACA AGT TCA AGA ACT GGA AAC ATT ACC ACC CAT AAG AAG TTT ACC GTT GCC CTT CCC ACA CAT GGA CCG TAC GCT GCA GGT CGA C and 5'-ATT ATT TAT TAT TTA TTA ATT ATT AAT TGT TAA TAT TAT TAA TTC TTG TCT ATT GTC TTT TGA TTT ATC TGC ACC GCC AAA AAC TTG TCA GCG TAT CAA TGC TCG TTA AAG TGT GTG GTT, and plasmid pYM22 .
An ordered array of ~4700 viable gene deletion mutants in the haploid BY4741 background was screened essentially as described in . Strains were replica pinned onto YNB control plates and plates with cadmium chloride (75, 100 and 150 μM) or sodium arsenite (0.5, 1.0 and 1.5 mM) using a 96-floating pin replicator operated by a Biomek 2000 robot (Beckman). The strains were arrayed in quadruplet to create a dilution in a given square giving a total of 96 strains plated per agar plate (see Additional file 2). The plates were incubated at 30°C and photographed after 24, 48 and 72 hours. The images were quantified in the image analysis software ImaGene v.6 (Biodiscovery) and spot area and intensity were extracted. The resulting data was then imported and analyzed in the statistical language R . To remove technical artefacts and non-growing colonies the spots were filtered based on spot area and intensity. First, all control spots with an area less than 16 were removed. Then, spots with an area greater or equal to 16 but with the intensity within the spot less than the background intensity minus one standard deviation were: (1) removed if it was a control spot or (2) area set to 1 if it was a mutant. The logarithm of the fold-change between the mutant and control area was used as an estimate of the effect of metals on growth, where resistance and sensitivity is indicated by positive and negative values respectively. The logarithmic fold-changes were then normalized by subtracting an estimate plate effect, which was based on calculating the logarithmic fold-change between the median of the spots on the plates in question. Finally, the four replicates were added together using a trimmed mean value, where the highest and the lowest values were removed. Mutants with an average fold-change greater than 1.5 were selected for subsequent analysis.
The homozygous diploid BY4743 collection was screened essentially as described in . In brief, cells were grown at 30°C in liquid YEPD medium with or without metal and cell density of each culture (OD600) was measured after 24 and 48 hours. Sensitive mutants were re-screened to confirm slow growth in the presence of metal. Finally, the sensitive gene-sets from both screens (haploid and homozygous diploid) were pooled and used for further analysis.
To pin-point cellular functions that confer metal tolerance, we searched for functional categories (according to FunCat, Munich Information Center for Protein Sequences (MIPS))  that were significantly enriched (p-value < 0.005) in the sensitive gene-set as compared to the whole genome. Gene Ontology (GO) analysis was done using GO Term Finder , putative human homologues of As and Cd sensitive yeast genes were identified using the National Center for Biotechnology Information HomoloGene database , and affinity MS data for protein-protein interactions was obtained from . Hierarchical clustering of metal and oxidative stress sensitive data was performed using Cluster 2.11 and visualized with Java TreeView 1.1.3 .
Exponentially growing cells were either untreated or exposed to Cd, harvested and disrupted as described previously . Proteins were separated by SDS-PAGE and analysed using anti-phospho-Snf1p antiserum (Open Biosystems) or anti-HA antibody (Santa Cruz Biotechnology). IRDye goat 800CW anti-mouse or IRDye 680 anti-rabbit IgG were used as secondary antibodies. Filters were incubated with Odyssey blocking buffer (LI-COR Biosciences) and visualised using Odyssey IR scanner (LI-COR Biosciences). To monitor Tub1p-GFP and Tub3p-GFP, cells were grown in YEPD medium to mid-log phase, washed twice with phosphate-buffered saline (PBS) and the GFP signals were observed in living cells before and after exposure to arsenite. To visualize actin, cells were untreated or exposed to arsenite, and incubated with rhodamine-conjugated phalloidin (Invitrogen) as described in . The GFP and rhodamine signals were observed using a Leica DM R fluorescence microscope.
reactive oxygen species
high osmolarity glycerol
green fluorescent protein
We thank Marc Sohrmann and Matthias Peter at the Swiss Federal Institute of Technology (ETH) in Zurich for providing access to the Biomek 2000 robot and for helping with the screen, Martin Schmidt (University of Pittsburgh School of Medicine) for plasmids, Stefan Hohmann (University of Gothenburg) for strains, and Marcus Krantz and Jonas Warringer (University of Gothenburg) for critical comments on this manuscript. We gratefully acknowledge financial support from the Swedish National Research School in Genomics and Bioinformatics to ON and MJT, the Swedish Research Council and University of Gothenburg Chemical Biology Platform to MJT, the foundation Knut och Alice Wallenberg Stiftelse in form of a travel grant to MT, and the Australian Research Council and Cancer Council of New South Wales to GGP and IWD.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.