|Home | About | Journals | Submit | Contact Us | Français|
Arsenic is a human toxin and carcinogen commonly found as a contaminant in drinking water. Arsenite (AsIII) is the most toxic inorganic form, but recent evidence indicates that the metabolite monomethylarsonous acid (MMAIII) is even more toxic. We have used a chemical genomics approach to identify the genes that modulate the cellular toxicity of MMAIII and AsIII in the yeast Saccharomyces cerevisiae. Functional profiling using homozygous deletion mutants provided evidence of the requirement of highly conserved biological processes in the response against both arsenicals including tubulin folding, DNA double-strand break repair, and chromatin modification. At the equitoxic doses of 150μM MMAIII and 300μM AsIII, genes related to glutathione metabolism were essential only for resistance to the former, suggesting a higher potency of MMAIII to disrupt glutathione metabolism than AsIII. Treatments with MMAIII induced a significant increase in glutathione levels in the wild-type strain, which correlated to the requirement of genes from the sulfur and methionine metabolic pathways and was consistent with the induction of oxidative stress. Based on the relative sensitivity of deletion strains deficient in GSH metabolism and tubulin folding processes, oxidative stress appeared to be the primary mechanism of MMAIII toxicity whereas secondary to tubulin disruption in the case of AsIII. Many of the identified yeast genes have orthologs in humans that could potentially modulate arsenic toxicity in a similar manner as their yeast counterparts.
Arsenic (As) is a human carcinogen ubiquitous in the environment. Millions of people worldwide are exposed to this metalloid primarily through drinking water, where it is present as a natural contaminant. Chronic exposure to As is associated with several adverse effects on human health that range from alterations in skin pigmentation and non malignant pulmonary disease to cardiovascular disease and cancer of the skin, lung, bladder, liver, and kidney (Ferreccio et al., 2000; Mazumder et al., 2005; Smith et al., 1992, 1998; von Ehrenstein et al., 2005). The cancer risks estimated with lifetime exposure to high concentrations of As in water are very high and comparable with those for cigarette smoking and high concentrations of radon in homes (Smith et al., 1992). Therefore, the presence of As in the environment poses a serious risk to human health.
Some mammals, including humans, methylate the inorganic As (AsIII) ingested in water to form the more toxic monomethylarsonous acid (MMAIII) (Mass et al., 2001; Petrick et al., 2000, 2001; Styblo et al., 2000). People that excrete high levels of monomethylated arsenicals in urine have a higher risk of As-induced cancer (Steinmaus et al., 2006). Moreover, humans excrete relatively more MMAIII in urine than any other animal species and are more sensitive to As carcinogenicity, suggesting that MMAIII may be central to As toxicity.
Various mechanisms have been proposed to explain arsenic's carcinogenicity, including spindle disruption, induction of chromosomal aberrations, formation of reactive oxygen species, inhibition of DNA repair, alteration in DNA methylation patterns, and promotion/progression in carcinogenesis (Kitchin, 2001; Kligerman and Tennant, 2007). Although the molecular mode of action of arsenic remains unclear, it is likely to involve several of these mechanisms.
The baker's yeast Saccharomyces cerevisiae shares many fundamental cellular processes with humans. Homozygous yeast deletion mutants of nonessential genes can be analyzed simultaneously to interrogate their growth phenotype and functionally profile the yeast genome under selective conditions of interest (Giaever et al., 2002). Previous studies have shown a low correlation between gene expression levels and genetic requirements for growth in several conditions. For example, very few DNA repair genes required for growth in the presence of DNA-damaging agents are upregulated in yeast (Birrell et al., 2002). Considering that growth is a better indicator for the requirement of a gene in the presence of a toxicant than its expression level, we conducted a functional profiling of the yeast genome to help determine the genes required for yeast's fitness in MMAIII and AsIII, and gain insight into potential mechanisms underlying their toxicity.
All yeast strains used in this study were of the BY4743 background (MATa/MATα his3Δ1/his3Δ1, ura3Δ0/ura3Δ0, leu2Δ0/leu2Δ0, lys2Δ0/+, met15Δ0/+). Growth was conducted in either liquid rich (1% yeast extract, 2% peptone, and 2% dextrose, YPD) or synthetic defined (SD) media, at 30°C with shaking at 200 rpm.
Monomethylarsine oxide (MMAIIIO) was a generous gift from Professor Miroslav Styblo. MMAIIIO hydrolyzes to MMAIII in solution (Petrick et al., 2001). Sodium arsenite (Sigma-Aldrich, St Louis, MO) and MMAIIIO stock solutions were prepared in sterile Milli-Q water (Millipore, Billerica, MA), protected from light and stored at −80°C until use. Under the growth conditions utilized, approximately 30 and 15% of the MMAIII was oxidized to MMAV in the presence and absence of yeast cells, respectively. Likewise, about 60% of AsIII was converted to the less toxic AsV when cells were present (Supplementary Table 1). There was no evidence of AsIII or MMAIII methylation in yeast wild type.
Yeast strains were pregrown to mid-log phase, diluted to an optical density at 595 nm (OD595) of 0.0165, and dispensed into different wells of a 48-well plate. Arsenical stock solutions were added to the desired final concentrations with at least three replicates per dose. Plates were incubated in a Tecan Genios spectrophotometer set to 30°C with intermittent shaking. OD595 measurements were taken at 15-min intervals for a period of 24 h. Raw absorbance data were averaged for all replicates, background corrected, and plotted as a function of time. The area under the curve (AUC) was calculated with Prism version 5.01 (GraphPad Software, La Jolla, CA), as a measure of growth, and expressed as a percentage of the control. AUCs were compared with either one- or two-way ANOVA, as appropriate, followed by Dunnett or Bonferroni post-tests, respectively.
Pool growth, genomic DNA extraction, barcode amplification, and hybridization were performed as previously described (Pierce et al., 2006), with minor modifications. Briefly, homozygous diploid deletion mutants (n = 4607) were grown in YPD at different arsenical concentrations for 5 and 15 generations (5g, 15g). Cells were collected and genomic DNA was extracted using the YDER kit (Pierce Biotechnology, Rockford, IL). The strain-specific barcodes in the DNA were amplified by PCR using a set of biotinylated primers, and reactions hybridized to TAG4 arrays (Affymetrix, Santa Clara, CA). Arrays were incubated overnight and then stained and scanned at an emission wavelength of 560 nm using a GeneChip scanner (Affymetrix). Raw and processed data files are available at the Gene Expression Omnibus (GEO) database.
Raw TAG4 array data were log2 transformed, corrected for signal saturation as previously described (Pierce et al., 2006), and corrected for mean chip background using robust location and scale estimators for log2-transformed intensities of null features (total of 18,000 equally distributed on the array). To account for variability in strain growth, data from each treatment array were paired to data from 12 controls (5g or 15g) for analysis. Treatment-control pairs were normalized with locally weighted scatterplot smoothing (global normalization), and the differentially growing strains identified using an alpha-outlier approach (Loguinov et al., 2004). Data from three biological replicates were combined, resulting in 36 treatment-control data pairs per treatment group. Residual variances (with a robust scale estimator) of log2 (treatment/control) for each 36 pairs were inspected using box plots. The “effective pairs” were then determined by excluding pairs with abnormally high residual variance, or with suspected serial correlation in variability (regular patterns in the box plots) (see Supplementary Methods for sample plots).
Significant genes (strains) were statistically inferred using an exact binomial test, assuming that the outcomes for each gene in all effective treatment-control pairs were independent binary variables with the same probability of success (p=0.5) for all trials (Bernoulli trials). For a particular gene n, outcomes were considered as “successful” if they were significant in the outlier analysis with q-values ≤ 0.05 in each of all effective pairs with log2 ratios of the same sign, simultaneously. The corresponding raw p values based on the exact binomial test were then corrected for multiplicity of comparisons using q-value approach and only the genes with q-value ≤ 0.05 were considered for further analysis. This approach does not apply a scale estimator and, as a result, it does not require between-chip pair normalization for the statistical inference.
Data sets were verified for enrichment for any particular biological attribute by identifying significantly over represented Gene Ontology (GO) categories by a hypergeometric distribution using the Functional Specification resource, FunSpec (http://funspec.med.utoronto.ca/), with a p value cutoff of 0.01. Fitness data were visualized onto the yeast interaction (Kiemer et al., 2007) and regulatory networks (www.yeastract.com) (accessed on July 2007) using Cytoscape version 2.5.1 (www.cytoscape.org).
Control and treated cultures were centrifuged and cell pellets washed in Milli-Q water. Samples were deproteinated by resuspension of the cell pellets in 5% metaphosphoric acid followed by incubation with shaking for 15 min at room temperature. The resulting cell lysates were then centrifuged and the supernatants saved for reduced glutathione (GSH) and oxidized glutathione (GSSG) quantitation. For liquid chromatography tandem mass spectrometry (LC-MS-MS) quantitation of GSH, calibration standards, quality control samples and supernatant samples were precipitated with 2× volumes of ice-cold acetonitrile and centrifuged at 6100 × g for 30 min at 4°C. A volume of 10 μl of supernatant was injected to a 50 × 2 mm hydrophilic interaction liquid chromatography column and subsequently analyzed using a MDS SCIEX API 3000 Mass Spectrometer (Applied Biosystems, Foster City, CA) in MRM scan mode. Total GSH (GSH + 2*GSSG) was also determined using the Glutathione Assay Kit (Cayman Chemical, Ann Arbor, MI) following manufacturer's specifications. To determine GSSG concentration in this assay, 2-vinylpyridine was added to a final concentration of 10mM, incubated for 1 h to derivatize the GSH, and assayed for total GSH.
MMAIII showed a higher potency than AsIII to inhibit the growth of the wild-type yeast strain (Fig. 1). In order to study and compare the genetic requirements of yeast for growth in the presence of these arsenicals, we selected equitoxic concentrations that resulted in 20% growth inhibition (IC20), as well as 50 and 25% of this IC20. These doses corresponded to 150, 75, and 37.5μM for MMAIII, and 300, 150, and 75μM for AsIII, respectively (Supplementary Figs. 1 and 2). We exposed pools of yeast homozygous deletion mutants for 5g and 15g of growth, totaling six treatments per arsenical, with three biological replicates each (Table 1).
We used a differential strain sensitivity analysis (as described in “Methods”) to identify the strains that exhibit significant changes in growth after treatments with MMAIII and AsIII, and determine the genes that influence yeast's fitness in their presence. The number of identified strains (genes) correlated directly both to the dose and to the number of generations of growth, ranging from few genes after 5g to several hundred after 15g (Fig. 2, Supplementary Tables 2 and 3), and showing different degrees of overlap (Figs. 3A and 3B).
For each arsenical, we considered the genes that were significant in the majority of the six treatments as the ones having the most influence on yeast fitness. The genes that were essential for growth in at least five of the six MMAIII treatments (Table 2) were associated with cadmium ion response (YCF1, GSH1); multivesicular body membrane disassembly (ATG15); and response to mercury ion/hydrogen peroxide (GSH1). Similarly, genes identified in at least five of the six the AsIII treatments (Table 3) belonged primarily to several biological processes including tubulin complex assembly (GIM4, YKE2, RBL2); protein import into peroxisome matrix, receptor recycling (PEX4, PEX6); postchaperonin tubulin folding pathway (RBL2, CIN2); histone exchange (VPS71, SAS2); chromatin modification (SAS4, VPS71, SAS2, SAS5); microtubule-based process (TUB3, CIN2); regulation of transcription, DNA dependent (SGF29, SRB8, SAS4, SSK1, SAS2, SSN8, SAS5, ARR1); arsenic resistance (ARR1, ARR3); chromatin silencing at telomere (SAS4, SAS2, SAS5); and protein ubiquitination (PEX4, BUL1).
About 15% of genes were common to the MMAIII and AsIII treatments after 5g of growth and approximately 35% after 15g. In general, genes associated with cadmium ion transport, cytoskeleton organization and biogenesis, chromatin modeling, intracellular protein transport, regulation of transcription, and protein catabolism were common to both arsenical treatments, although there were also genes that belonged to these processes but were specific to one of them (Table 4). Glutathione biosynthesis, sulfur and methionine metabolism, and nicotinamide adenine nucleotide phosphate (reduced) (NADPH) regeneration were enriched for genes from the MMAIII treatments. Interestingly, the number of genes specific to MMAIII increased about fourfold from 5g to 15g, whereas the one for AsIII was only slightly increased, as almost all of the newly identified genes in AsIII overlapped with those of MMAIII (Fig. 3C, Supplementary Table 4). The genes specific to MMAIII were mainly associated with transcription and GSH metabolism after 5g and 15g of growth, respectively; whereas the ones found only in AsIII were associated to DNA double-strand break (DSB) repair after both 5g and 15g (Table 5).
We identified genes encoding known components of AsIII resistance in yeast. Among these, the ARR genes (ARR1, transcription factor for arsenic resistance; ARR2, arsenate reductase; ARR3, AsIII transporter) were specifically essential for growth in the presence of AsIII. The identification of ARR2 for resistance was likely due to AsV formed by oxidation of AsIII. YCF1, encoding a vacuolar glutathione S-conjugate transporter, was essential for growth in both arsenicals. On the other hand, deletion of the aquaglyceroporin gene FPS1, whose product transports AsIII into the cell, resulted in resistance not only to AsIII but also to MMAIII (Fig. 4). The identification of FPS1 and YCF1 suggests a common entry path of arsenic into yeast cells and detoxification mechanism to the vacuole between these arsenicals, respectively.
Two important groups of genes identified as essential for growth with MMAIII and/or AsIII were associated with tubulin folding (CIN2, CIN4, RBL2) and biogenesis (PFD1, GIM4, PAC10, YKE2). The genes in the later group encode subunits of the heterohexameric GIM/prefoldin complex, a component of the unfolded protein response that also participates in actin and tubulin biogenesis. Another related gene, TUB3, encodes for α-tubulin and was also essential for growth in the presence of AsIII in the screen and to MMAIII only after growth assay (data not shown). Unlike the other tubulin genes TUB1 and TUB2, TUB3 is not required for viability; however, these results suggest that it plays an important role in the presence of arsenicals.
Biological processes associated with DNA DSB repair were important for resistance to AsIII (Table 5). Members of the RAD52 epistasis group (RAD27, RAD50, RAD54) were required for growth in at least one treatment with MMAIII or AsIII. The proteins encoded by these genes are involved in homologous recombination and in the repair of DSBs. Rad50p is a subunit of the Mre11p-Rad50p-Xrs2p (MRX) complex, associated with several DNA repair processes. Individual strain analysis confirmed that deletion of any of the three subunits of this complex resulted in growth sensitivity to MMAIII and AsIII (Fig. 5), providing evidence of its involvement in the response against As toxicity.
We identified multiple subunits of multimeric complexes including SAS, SAGA, SWR1, and Itc1p-Isw2p, associated to chromatin remodeling processes as involved in response to both arsenicals. We evaluated the phenotypes of the selected mutants htz1Δ, itc1Δ, sas2Δ, swr1Δ, and yaf9Δ, containing deletions of genes encoding key (and associated) components of these complexes (Fig. 5). Htz1p is a histone variant, H2AZ, involved in transcriptional regulation. Itc1p is part of the Itc1p-Isw2p chromatin remodeling complex, required for gene repression. Sas2p is the catalytic subunit of the SAS complex, which has histone 4 lysine 16 acetyltransferase activity. Swr1p is the catalytic subunit of the SWR1 complex, which replaces H2A-H2B by Htz1p-H2B histone dimers at specific chromosomal locations. Yaf9p is present in both the NuA4 histone H4 acetyltransferase and SWR1 complexes and contains a YEATS domain often found in subunits of chromatin-modifying complexes and in the human leukemogenic protein AF9. In addition, we tested the phenotype of gcn5Δ. Gcn5p is the catalytic subunit of the SAGA complex, which acetylates N-terminal lysine residues in histones H3 and H2B. Except for sas2Δ that was only sensitive to AsIII, all the mutants tested were sensitive to both arsenicals. MMAIII was about twofold more potent than AsIII in inhibiting mutant growth, suggesting a similarity of these compounds in toxicity mechanisms but differences in potency.
Deletion of GSH1 negatively affected yeast growth upon exposure to MMAIII but not AsIII (Fig. 6A). GSH1 encodes γ-glutamylcysteine synthetase, an enzyme that catalyzes the first step in GSH biosynthesis; thus, gsh1Δ mutants are deficient in GSH. In individual growth assay, this strain exhibited a significant decrease in growth at doses as low as 5μM of MMAIII, being approximately 30-fold more sensitive than the wild type (Supplementary Fig. 3). Although MMAV was formed during incubation with MMAIII, treatment with MMAV alone did not inhibit growth of wild type at 1mM (IC20 > 1mM), or gsh1Δ at the same MMAIII concentrations (data not shown). Therefore, the observed effects on growth are likely to be due to MMAIII. On the other hand, gsh1Δ was slightly more sensitive than wild type to AsIII only when grown at concentrations above the IC50 (600μM) (Supplementary Fig. 4), being approximately 60-fold more sensitive to MMAIII than to AsIII (IC20’s of 5 and 300μM, respectively).
To further determine if the lack of GSH in gsh1Δ was the cause of its sensitivity to MMAIII, we treated the wild-type strain with buthionine sulfoximine (BSO), an inhibitor of Gsh1p and thus, GSH biosynthesis. In support of an important role of GSH in the protection against MMAIII, depletion of GSH significantly increased the sensitivity of the wild type to MMAIII but did not affect its growth even at AsIII concentrations that were eightfold higher (Fig. 6B).
Ycf1p transports AsIII-GSH conjugates into the vacuole and was essential for fitness not only in AsIII but also in MMAIII. The strain ycf1Δ was able to grow in the presence of MMAIII but became significantly sensitive after treatment with BSO (Fig. 6C). Therefore, the requirement of GSH is unlikely to be associated with this detoxification pathway.
Treatment of the wild-type strain with MMAIII for 8 h resulted in a significant increase in total GSH of more than threefold at the highest exposure concentration of 150μM (Table 6). This increase was in agreement with the observed requirement of the transcription factors Met28p and Met32p, responsible for the induction of genes involved in the uptake of the sulfur and biosynthesis of the methionine utilized in GSH biosynthesis (Fig. 7A).
GSSG levels were increased after 8 h of MMAIII treatment when measured using an enzymatic-based method (Supplementary Table 5), and were in agreement with the observed sensitivity of glr1Δ to MMAIII (Fig. 6A). GLR1 encodes GSH reductase, which is involved in the regeneration of GSH from its oxidized form GSSG. Other genes identified in our screen were RPE1 and TKL1, encoding proteins involved in NADPH regeneration and thus, necessary in the reaction catalyzed by Glr1p.
Because of the existing evidence that arsenicals are genotoxic, we further examined the role of RAD27, RAD50, and XRS2 in the resistance to AsIII. Deletion of any of these genes resulted in significant decrease in growth relative to the wild type, as shown by individual strain growth analysis (Fig. 5). Exposure to AsIII concentrations as high as 1mM for 4 h did not show any apparent formation of DSBs in the wild-type or deletion strains after assessment with pulsed-field gel electrophoresis (Supplementary Fig. 5). In spite of the fact that 1mM of AsIII inhibited wild-type growth by approximately 90% (Fig. 1), the absence of DSBs at this high concentration indicates that increased DNA damage is not the primary mechanism of AsIII toxicity. The observed arsenical sensitivity of the RAD mutants could then be the result of other mechanisms of arsenic toxicity such as inhibition of DNA repair. Alternatively, the negative results could reflect a limitation of this method to detect low levels of this type of damage.
We identified the genetic requirements of yeast for resistance to MMAIII and AsIII by using a genome-wide phenotypic analysis of yeast mutants containing deletions in nonessential genes. Functional profiling in yeast mutants has been previously reported for AsIII (Haugen et al., 2004; Jin et al., 2008; Thorsen et al., 2009; Vujcic et al., 2007). Because MMAIII is more potent and formed in the human body after metabolism of AsIII, the screening of AsIII alone may not fully reflect the toxicity of As. Because of the poor overlap in identified gene sets between different AsIII screens, possibly influenced by the different techniques used, we directly compared the genetic determinants for resistance to MMAIII and AsIII in the same study. Comparison of the genes associated with the cellular response to these compounds could then provide insight into the adverse effects of As and identify conserved toxicity pathways in humans (Fig. 8).
Yeast cells are more resistant to arsenicals than human cells probably due to the presence of the AsIII transporter, encoded by ARR3, not present in human cells, and/or the low transport rate of MMAIII into yeast cells (Liu et al., 2006). In order to compare MMAIII and AsIII treatments, we used concentrations that produced the same degree of growth inhibition in the BY4743 wild-type strain (20% inhibitory concentrations, IC20, and 25 and 50% of this value). Further, we evaluated growth after 5g and 15g to account for the effect of the duration of exposure on the growth of certain strains and on overall fitness profiles. In some cases, the differences in fitness can be due to accumulation of cellular damage, such as DNA damage (Birrell et al., 2002) or to the activation of general stress response pathways after 15g of growth. In addition, very small differences in growth between strains may not be detected after few generations but only after being amplified following many generations of growth.
Interestingly, our results showed that both arsenicals induced distinct but overlapping mutant sensitivity profiles, despite yeast's inability to convert AsIII to MMAIII. The absence of AsIII metabolism to MMAIII in yeast allows dissecting their genetic requirements for resistance and studying their toxicity separately. However, those requirements arising from synergistic effects between them or from the metabolism of AsIII, which may affect methyl donor or GSH pools, may not be identified in this organism. The genes found to be essential for fitness are associated with biological processes that are consistent with proposed mechanisms of arsenic toxicity. We identified tubulin metabolism, DNA repair, chromatin remodeling (epigenetic changes), and glutathione biosynthesis as key processes, as discussed below.
The requirement for multiple genes associated with tubulin assembly for growth in MMAIII and AsIII can be related to the fact that As targets tubulin (Menzel et al., 1999; Zhang et al., 2007). Trivalent arsenicals have high affinity to cellular thiols due to the reactive pair of unshared electrons present in their outer shell. Among the many intracellular proteins that As binds, tubulin is probably one of the most common because of its abundance and high thiol content. In human cells, As blocks polymerization of tubulin by binding to vicinal cysteine residues and blocking the active site of GTP required for this process (Li and Broome, 1999; Zhang et al., 2007). The finding that genes involved in tubulin metabolism were also required in MMAIII was not unexpected, as MMAIII also binds to proteins (Styblo and Thomas, 1997) and is capable of inhibiting several enzymes more effectively than AsIII (Lin et al., 2001; Petrick et al., 2001; Styblo et al., 1997). In yeast, deletion of CIN2 and CIN4 increased the rate of chromosome loss and sensitivity to the microtubule-disrupting fungicide benomyl (Hoyt et al., 1990; Stearns et al., 1990). These genes were also required in As, and provide further support that As targets microtubules and the cytoskeleton.
Although the requirement of the MRX complex and some RAD genes in MMAIII and AsIII suggested the induction of DNA damage, we did not observe any apparent increase in the formation of DSBs in several mutant strains after treatment with high doses of AsIII. Arsenic has been shown to inhibit DNA repair enzymes and decrease DNA repair efficiency, possibly by disrupting the zinc fingers in these proteins (Piatek et al., 2008; Takahashi et al., 2000). Considering the occurrence of background levels of DSBs, the presence of As could disadvantage deletion strains that are already defective in DNA repair. Therefore, one explanation to our observations is that, by inhibiting DNA repair, As could stress DNA repair networks, resulting in sensitivity of certain RAD deletion mutants.
Epigenetic modifications are mediated by multimeric molecular complexes with highly conserved functions across species, including humans and yeast. Interestingly, several subunits of the yeast chromatin-modifying complexes SWR1, SAGA, SAS, and Itc1-Isw2, were essential for optimal growth in the presence of arsenicals (Supplementary Table 6), suggesting that their functions are necessary for the adaptation to As-induced stress. The histone Htz1p, which is incorporated into nucleosomes by SWR1, is expressed at higher levels after arsenic treatment (Dr W. Jo, unpublished data). The SWR1 complex genes and HTZ1 are therefore needed for the induction of epigenetic changes associated with yeast's resistance to arsenicals. Because incorporation of Htz1p into chromatin is associated with rapid transcriptional activation under certain conditions (Zhang et al., 2005), adaptation to arsenic may implicate differential expression of specific genes. In addition, these findings suggest that the other complexes mediate additional epigenetic changes upon arsenic exposure are also required for resistance.
At equitoxic low doses, the requirements for genes associated to GSH metabolism and related pathways differed between MMAIII and AsIII. The increase in GSSG levels at high MMAIII concentrations, together with the requirement of the GSH reductase gene, suggests that this arsenical promotes oxidation of GSH, which could be the result of its ability to inhibit GSH reductase in vitro (Styblo et al., 1997) and/or produce reactive oxygen species (Liu et al., 2001).
Several studies have reported the protective role of GSH against AsIII (Han et al., 2008; Ortiz et al., 2009). Compared with previous yeast studies (Haugen et al., 2004; Thorsen et al., 2007, 2009), there was limited evidence in our data that AsIII induces the GSH cellular response. Treatment with AsIII slightly increased GSH levels but did not affect the growth phenotype of most mutants with a defect in GSH-related pathways. Based on the integrated phenotypic data and GSH measurements presented in this study, it can be concluded that the requirement of GSH in MMAIII is much higher than in AsIII, suggesting that GSH is more important in the protection against MMAIII- than AsIII-induced toxicity. Therefore, our results indicate that special attention must be given to MMAIII when assessing the effects of arsenic on GSH levels and susceptibility to arsenic toxicity. Because GSH is linked to cellular antioxidant status, oxidative stress may constitute an important component of MMAIII toxicity and may be responsible for its higher potency than AsIII, at least in yeast.
The cellular responses against AsIII at the concentrations tested demanded the function of genes primarily associated with tubulin metabolism, with minor involvement of those related to GSH metabolism. These findings are consistent with the hypothesis that reactive oxygen species are not involved in tubulin disruption and constitute another mechanism of arsenic toxicity (Kligerman and Tennant, 2007). Based on the relative sensitivity (at equitoxic doses) of deletion strains deficient in GSH metabolism and tubulin folding processes, oxidative stress appeared to be the primary mechanism of MMAIII toxicity and secondary to tubulin disruption in the case of AsIII.
Because yeast and humans share in common many cellular metabolic pathways, a better understanding of the toxicity of arsenicals in yeast should help to understand their toxicity in humans and identify biomarkers of susceptibility. Several human orthologs to yeast genes associated with mitochondrial processes and involved in protection against arsenic-induced toxicity have been identified (Vujcic et al., 2007). Similarly, several of the genes that we identified as essential in yeast's response to arsenic have at least one human homolog. These human genes could potentially modulate As toxicity in a similar way to their yeast counterparts and thus, influence susceptibility (Table 7). The MRX complex, from which we found its three subunits to be essential for yeast growth in arsenicals, is homologous to the mammalian Mre11-Rad50-Nbs1 (MRN) complex. Among the components of the MRN complex, RAD50 was found to be upregulated and Mre11 phosphorylated after treatment with AsIII (Perez et al., 2008; Yuan et al., 2002), suggesting that this complex is involved in the response to AsIII in human cells. Moreover, the E185Q polymorphism in NBS1 has been associated with increased risk of basal cell carcinoma in men exposed to As (Thirumaran et al., 2006). Therefore, the functional conservation between the genes that we identified in this study and their human homologs raises the question whether or not the later could be necessary for As resistance in humans. Our findings in this study represent a starting point for more focused, confirmatory gene- or pathway-specific studies.
National Institute of Environmental Health Sciences Superfund Basic Research Program (P42 ES004705-19); and University of California Berkeley Faculty on Research Grant awarded to C.D.V.
We thank Professor Miroslav Styblo for generously providing MMAIIIO, Dr Jeung H. Kim for guidance on methods development, Dr Matthew North for helpful discussions, Eric Taylor for developing the methods and conducting MS analysis, and Onita Bhattasali for technical assistance. W.J.J. is a trainee in the Superfund Basic Research Program at the University of California, Berkeley.