Functional Genomics Analysis of the Saccharomyces cerevisiae Iron Responsive Transcription Factor Aft1 Reveals Iron-Independent Functions
Sharon Berthelet, Jane Usher, Kristian Shulist, Akil Hamza, Nancy Maltez, Anne Johnston, Ying Fong, Linda J. Harris, Kristin Baetz

Abstract

The Saccharomyces cerevisiae transcription factor Aft1 is activated in iron-deficient cells to induce the expression of iron regulon genes, which coordinate the increase of iron uptake and remodel cellular metabolism to survive low-iron conditions. In addition, Aft1 has been implicated in numerous cellular processes including cell-cycle progression and chromosome stability; however, it is unclear if all cellular effects of Aft1 are mediated through iron homeostasis. To further investigate the cellular processes affected by Aft1, we identified >70 deletion mutants that are sensitive to perturbations in AFT1 levels using genome-wide synthetic lethal and synthetic dosage lethal screens. Our genetic network reveals that Aft1 affects a diverse range of cellular processes, including the RIM101 pH pathway, cell-wall stability, DNA damage, protein transport, chromosome stability, and mitochondrial function. Surprisingly, only a subset of mutants identified are sensitive to extracellular iron fluctuations or display genetic interactions with mutants of iron regulon genes AFT2 or FET3. We demonstrate that Aft1 works in parallel with the RIM101 pH pathway and the role of Aft1 in DNA damage repair is mediated by iron. In contrast, through both directed studies and microarray transcriptional profiling, we show that the role of Aft1 in chromosome maintenance and benomyl resistance is independent of its iron regulatory role, potentially through a nontranscriptional mechanism.

LIKE all organisms, the yeast Saccharomyces cerevisiae maintains tight regulation of cellular iron uptake and utilization to prevent toxicity caused by iron overload (reviewed in Kaplan et al. 2006). S. cerevisiae responds to iron depletion through transcriptional remodeling governed primarily by the iron-responsive transcription factor Aft1 (reviewed in Rutherford and Bird 2004). Aft1 is routinely shuttled between the nucleus and the cytoplasm where the export of Aft1 from the nucleus is promoted in the presence of iron-sulfur clusters (ISC) in the cell (Yamaguchi-Iwai et al. 2002; Chen et al. 2004; Rutherford et al. 2005; Ueta et al. 2007). Upon iron depletion and decreased levels of ISCs, Aft1 accumulates in the nucleus where it activates the transcription of 25 genes, referred to as the “iron regulon,” that are required for increasing cellular iron content (Yamaguchi-Iwai et al. 1996; Rutherford et al. 2001; Rutherford et al. 2003; Shakoury-Elizeh et al. 2004; Courel et al. 2005).

The iron regulon genes can be grouped into three categories (extensively reviewed in Kaplan et al. 2006; Philpott and Protchenko 2008). The majority of the genes encode proteins that increase iron uptake from the environment, including genes that encode siderophore transporters (ARN1, ARN2, ARN3, ARN4), cell-wall siderophore binding/uptake proteins (FIT1, FIT2, FIT3), iron-reducing metalloreductase proteins (FRE1-FRE5), and the high-affinity iron transport complex composed of a ferroxidase (FET3) and a permease (FTR1). As copper is required for the activity of Fet3, the iron regulon also includes the copper chaperone ATX1 and copper transporter CCC2. A second class of genes encode proteins that allow the cell to mobilize the significant amounts of iron the cell stores in the vacuole (SMF3, FET5, FRE6, FTH1, COT1) or in the mitochondria as heme or ISC (HMX1, MRS4). A third class of genes encode proteins that allow the cell to remodel its metabolic activities to decrease the use of iron-dependent enzymes/pathways in favor of iron-independent processes. This includes the upregulation of the biotin transporter VTH1, which allows the cell to obtain essential biotin from the environment instead of utilizing the iron-dependent biotin biosynthesis pathway and CTH2/TIS11, which encodes a mRNA binding protein that destabilizes mRNAs that encode enzymes that require iron cofactors.

In the absence of Aft1, its paralog Aft2 can compensate and regulate transcription of many iron regulon genes (Blaiseau et al. 2001; Rutherford et al. 2001; Rutherford et al. 2003; Courel et al. 2005; Rutherford et al. 2005). Although Aft2 and Aft1 have overlapping functions, their roles in the transcriptional regulation of the iron regulon are nonredundant (Blaiseau et al. 2001) with Aft1 having the prominent role in the transcriptional activation of the iron regulon (Rutherford et al. 2003). Additionally while aft1Δ mutant cells exhibit low-ferrous-iron uptake and poor growth under low-iron conditions (Yamaguchi-Iwai et al. 1995; Casas et al. 1997), aft2Δ mutant cells shows no growth defects under these conditions (Rutherford et al. 2003; Courel et al. 2005). However, consistent with the ability of Aft2 to regulate the iron regulon, an aft1Δaft2Δ double mutant is more sensitive to low-iron growth conditions than a single aft1Δ null mutant alone (Blaiseau et al. 2001; Rutherford et al. 2001).

In addition to iron depletion, numerous environmental conditions result in the Aft1-dependent induction of the iron regulon, including zinc (Pagani et al. 2007), hydroxyurea (HU) (Dubacq et al. 2006), and cisplatin treatments (Kimura et al. 2007), during the diauxic shift (Haurie et al. 2003) and upon loss of mitochondrial DNA (mtDNA) (Veatch et al. 2009). The iron regulon is also induced during both the alkaline response (Lamb et al. 2001) and adaptation to lactic and acetic acid (Kawahata et al. 2006); however, in these cases Aft1 dependence has not been confirmed. Under many of these conditions, Aft1 is mediating a critical role as aft1Δ mutants display hypersensitivity to HU (Dubacq et al. 2006), cisplatin (Lee et al. 2005), zinc (Pagani et al. 2007), and high pH (Serrano et al. 2004). Why is Aft1 needed under these diverse conditions? One possibility is that iron is either limited under these conditions and/or additional cellular iron is required to buffer some of these challenges. While it has been shown that iron or iron uptake becomes limiting under alkaline pH conditions (Serrano et al. 2004) and cisplatin treatment (Kimura et al. 2007) and that loss of mtDNA results in decreased ISC (Veatch et al. 2009), in the case of the diauxic shift, the activation of Aft1 is controlled by a Snf1/Snf6-dependent pathway and not by extracellular iron concentrations (Haurie et al. 2003). Therefore, activation of Aft1 is not solely limited to conditions that decrease ISC levels. Further, while it has been shown that increasing exogenous iron levels can suppress the HU sensitivity of aft1Δ cells (Dubacq et al. 2006), it has yet to be established if the Aft1-dependent transcriptional induction of the iron regulon and maintaining cellular iron levels is mediating all of the cellular functions of Aft1 or if Aft1 has additional iron-independent cellular roles.

Indeed, the transcriptional effects of Aft1 may not be limited to the iron regulon genes. Microarray analysis of wild-type vs. aft1Δ cells grown in iron replete YPD media showed that deletion of AFT1 resulted in the upregulation of 239 genes and the downregulation of 350 genes (Pagani et al. 2007). Furthermore, expression of the constitutively active aft1-1up, a mutation that localizes Aft1 to the nucleus, results in transcriptional modulation of more than 200 genes implicated in a variety of processes (Shakoury-Elizeh et al. 2004). Although it is not known how many of these genes are directly regulated by Aft1 or result from downstream transcriptional cascades, it suggests that Aft1 function may not be limited to the transcriptional induction of just iron regulon genes.

In addition, several studies have suggested that Aft1 may play a role in cell-cycle regulation. Systematic screens have determined that aft1Δ mutant cells are significantly larger than wild-type cells (Jorgensen et al. 2002) and are delayed in G1 (White et al. 2009). Further, overexpression of AFT1 or aft1-1up results in G1 arrest due to the inhibition of translation of G1 cyclins by an undetermined mechanism (Casas et al. 1997; Philpott et al. 1998). Aft1 also has been linked to chromosome stability (Measday et al. 2005; Yuen et al. 2007). Synthetic genetic array (SGA) studies determined that aft1Δ mutants could not tolerate either overexpression or loss-of-function of kinetochore genes (Measday et al. 2005). Chromosome transmission fidelity assays, which measure the ability of a cell to maintain an artificial chromosome fragment, determined that aft1Δ mutant cells display an increase in chromosome loss compared to wild-type cells (Measday et al. 2005; Yuen et al. 2007). Furthermore, Aft1 has been shown to colocalize with kinetochore proteins (Measday et al. 2005) as well as interact with kinetochore proteins Cbf1 (Measday et al. 2005) and Iml3 (Wong et al. 2007) in yeast two-hybrid assays. AFT2, or other iron regulon genes, have not been identified in the genome-wide kinetochore genetic or genome instability screens (Measday et al. 2005; Kanellis et al. 2007; Yuen et al. 2007; Andersen et al. 2008), suggesting that the cellular role of Aft1 in chromosome stability may be independent of its role in iron homeostasis. Presently the molecular mechanism by which Aft1 contributes to genome maintenance is unknown.

To explore the global cellular functions of Aft1 under iron replete conditions we performed genome-wide AFT1 synthetic lethal (SL) SGA and synthetic dosage lethal (SDL) SGA analyses. Our genetic interaction map reveals that >70 deletion mutants are sensitive to perturbations in AFT1 levels under normal iron conditions. While some of these genetic interactions are attributable to the role of Aft1 in iron homeostasis, including the RIM101 pH pathway, and DNA damage repair, we determine that the role of Aft1 in chromosome stability is distinct from its role in regulating the iron regulon and cellular iron levels.

MATERIALS AND METHODS

Yeast strains and plasmids:

The yeast strains used in this study are listed in Table 1. The MATa deletion mutant array was purchased from OpenBiosystems (catalog no. YSC1053). The SGA starting strain Y7092 (Tong and Boone 2006) and the media used in the SGA analysis have been described previously (Tong et al. 2001; Tong et al. 2004). Deletion strains made for this study were designed using a standard PCR-mediated gene insertion technique (Longtine et al. 1998). Plasmid pGAL1AFT1 (pKB38) (a galactose-inducible promoter followed by the gene AFT1 carrying a URA3 resistance marker) was isolated from the yeast overexpression array (Sopko et al. 2006) and confirmed by sequencing.

View this table:
TABLE 1

Strain list

Media and dot assay experiments:

Cells were grown in standard YPD or SD medium supplemented with amino acids (Abelson et al. 2004), unless otherwise described. For growth in liquid SD–uracil medium at a specific pH, SD–uracil containing 150 mm Hepes was titrated to pH 4 or pH 8 and filter sterilized. To assess growth under various conditions, wild-type and mutant strains were grown to mid-log phase in YPD at 25° and dot assays were performed by spotting 5 μl of fivefold serial dilutions (OD600 = 0.1, 0.01, 0.001, 0.0001) onto media containing caffeine (Sigma, C0750), calcofluor white (CFW, Sigma, F3543), cisplatin (Sigma, 479306), methyl methanesulfonate (MMS, Sigma, 129925), or benomyl (Sigma, 381586) as indicated. All dot assay experiments were repeated using two or three different isolates of each strain. For iron-limited YPD plates, 90 μm bathophenanthrolinedisulfonic acid disodium salt hydrate (BPS, Sigma, B1375) was used with the addition of FeSO4 (Sigma, F8048) as previously described (Davis-Kaplan et al. 2004).

Synthetic lethal and synthetic dosage lethal screens:

Robotic manipulation of the deletion mutant array was conducted using a Singer RoToR HDA robot (Singer Instruments, United Kingdom). Genome-wide SL–SGA screens were conducted three times at 30° for the query strains aft1Δ∷natMX4 (YKB676), rim101Δ∷natMX4 (YKB1008), aft2Δ∷natMX4 (YKB1010), and fet3Δ∷natMX4 (YKB1009) as described previously (Tong et al. 2001). Double mutants were scored for slow growth or lethality by visual inspection. Putative genetic interactions identified in a minimum of two of three screens were confirmed by tetrad dissection. Genome-wide SDL–SGA screens were conducted in triplicate for the query strain carrying the plasmid GAL1AFT1 (YKB 794) as previously described (Measday et al. 2005). After replica pinning onto galactose media to induce overexpression of AFT1, colonies were grown for 2 days at 16°, 25°, or 37° and scored for slow growth or lethality by visual inspection. Putative genetic interactions identified in a minimum of two of three screens were confirmed by transforming the deletion mutant strain with either the plasmid pGAL1AFT1 or the vector control pRS416 and growth defects were assessed by either streak tests or dot assays on plates containing galactose.

β-Galactosidase assays:

Reporter constructs pMELb2-lacZ and pMELb2-FET3-lacZ (Kimura et al. 2007) were transformed into wild-type (YPH499), aft1Δ (YPH1735), rim101Δ (YKB1110), and aft1Δrim101Δ (YKB1111) cells. For assessing pH effects, yeast cells carrying the reporter plasmids were cultured in SD–URA media to an OD600 of 0.6 and then inoculated at an OD600 of 0.2 in SD–URA media at pH 4 or pH 8 and grown to an OD600 of 0.8. For assessing the effects of benomyl, wild-type, and aft1Δ yeast cells carrying the reporter plasmids were cultured in SD–URA media to an OD600 of 0.6 and then treated with 20 μg/ml benomyl for 1 hr. β-Galactosidase assays were performed in triplicate using crude extracts exactly as described (Burke et al. 2000).

Chromosome transmission fidelity assays:

Quantitative half-sector analysis was performed essentially as previously described (Koshland and Hieter 1987), except strains were streaked onto YPD or YPD + 90 μm BPS + 100 μm FeS04 plates prior to the selection of single colonies for the plating assay. The red pigmentation caused by the addition of BPS in the plates, did not allow for consistent scoring of nondisjunction (white:pink) events; hence only chromosome loss (pink:red) half sector events were scored.

RNA microarray experiment:

Sample preparation:

Wild-type (YKB779) and aft1Δ (YKB673) cells were grown in YPD at 30° to OD600 of 0.4 and benomyl was added to a final concentration of 20 μg/ml. Cells were harvested prior to benomyl addition and at 20 min post-benomyl treatment by centrifugation and flash frozen in liquid nitrogen. For each strain and time point, three independent treatments and microarray hybridizations were performed. Total RNA was isolated using TRI reagent (Sigma) as per the manufacturer's protocol, followed by RNAeasy column purification (QIAGEN). The RNA was quantified using a NanoDrop-1000 spectrophotometer and quality was monitored with the Agilent 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA). Cyanine-3 (Cy3) labeled cRNA was prepared from 0.5 μg RNA using the one-color microarray-based gene expression analysis protocol version 5.7 (Agilent) according to the manufacturer's instructions, except using half of the reagent amount in step 2, followed by RNAeasy column purification (Qiagen). Dye incorporation and cRNA yield were checked with the NanoDrop ND-1000 Spectrophotometer.

Microarray hybridization and scanning:

Using Agilent Technologies' eArray online tool, a total of 13,189 60-mer oligonucleotides (up to 2 per ORF) were designed using 6649 target ORF sequences (orf_coding_all.20080606.fasta downloaded from http://downloads.yeastgenome.org/) to construct custom S. cerevisiae microarray 8X15K slides (Agilent, G2509F). A total of 0.6 μg of Cy3-labeled cRNA was hybridized to each microarray as per manufacturer's instructions and incubated for 17 hr at 65° in a rotating Robbins Model 400 hybridization oven (Robbins Scientific) and an Agilent rotator rack. Following hybridization, microarrays were washed using the wash procedure with stabilization and drying solution (Agilent) as described in the protocol. Slides were scanned immediately after washing using a GenePix 4200A (Molecular Devices) using only the Cy3 channel, scanning each array individually (scan area 2088 × 3112 pixels). The scan resolution was set at 10 μm, lines to average 1, focus position 0 μm. The laser was set at 100% and PMT between 330 and 370 according to strength of the individual array.

Data analysis:

The scanned images were analyzed with Genepix 6.0 (Molecular Devices). A normalization factor was calculated for each array using the “mean of F532 median” acquired in the array quality control report in Genepix. The normalization factor was determined so the average intensity of each array was 3400 (3400/mean of F532 median = normalization factor). This value was applied respectfully to each array in Genepix. Using BRB ArrayTools (http://linus.nci.nih.gov/BRB-ArrayTools.html) and Microsoft Excel, duplicate spots for each gene were averaged and any nonreproducible values were removed for the rest of the analysis. The Log 2(ratio) for each ORF was calculated and P-values were determined using a one-way ANOVA for Multiple Groups. These data sets were used to filter genes that had an observed change of more than twofold and a P -value <0.05 (see supporting information, File S1, for full analyzed data set; raw data can be accessed at GEO, accession no. GSE20531).

Modified chromatin immunoprecipitation assays:

Aft1–TAP-tagged and untagged wild-type strains grown in YPD medium at 30° to an OD600 of 0.8 were collected by centrifugation and modified chromatin immunoprecipitation (mChIP) was performed as previously described (Mitchell et al. 2008). Immunoprecipitated DNA was amplified using multiplex PCR with the following primer pairs: CTF19 F (5′ CCTGGATGAAACCCACTCGAA) and CTF19 R1 (5′ GAGTAACTTGCACAGCTATTGG); FET3 IRE F (5′ GGTCCCTACAGTACGCTGAG), and FET3 IRE R (5′ GGATCGACTGTTTGAGTGCATCC); TEL-V F (5′ GGCTGTCAGAATATGGGGCCGTAGTA) and TEL-V R (5′ CACCCCGAAGCTGCTTTCACAATAC). PCR products were resolved on a 3% agarose gel and visualized with ethidium bromide.

RESULTS

Synthetic lethal and synthetic dosage lethal analysis reveals novel roles for Aft1:

In an effort to identify cellular processes potentially affected by Aft1 under normal or replete iron conditions, we performed two genome-wide genetic screens using SGA methodology (Tong et al. 2001): SL screens to identify mutants that cannot tolerate the deletion of AFT1 and SDL screens to identify deletion mutants that cannot tolerate overexpression of AFT1. The aft1Δ SL–SGA screen was performed in triplicate and any double mutants that resulted in inviability (SL) or reduced fitness (synthetic sickness, SS) that were identified a minimum of two of three screens were confirmed by tetrad analysis (Table 2). The resulting confirmed data set contains 45 genetic interactions of which 22% (10/45) were SL interactions. Similarly, the AFT1 SDL–SGA screen was performed in triplicate and at three different temperatures (16°, 25°, and 37°). Any deletion mutants that displayed inviability (SDL) or reduced fitness (synthetic dosage sickness, SDS) upon overexpression of AFT1 in a minimum of two of three screens were confirmed by streak test and/or dot assay analysis (see material and methods for details). The resulting confirmed data set (Table 3) contains 32 genetic interactions of which only one was SDL, while the remainder were SDS interactions.

View this table:
TABLE 2

AFT1, AFT2, and FET3 synthetic lethal interactions

View this table:
TABLE 3

AFT1 synthetic dosage lethal genetic interactions

Previous work has determined that SL and SDL screens are complementary in nature, identifying largely nonoverlapping sets of genetic interactions (Measday et al. 2005; Baetz et al. 2006). Therefore as expected, there was limited overlap between the AFT1 SL and SDL screens, with only one gene identified in both screens (Figure 1 ). Despite the lack of overlap, both screens identified genes associated with cellular processes previously associated with Aft1, including iron homeostasis, transcription, and chromosome stability. In addition the genetic network suggests that Aft1 function may affect a wider range of cellular processes than previously thought, including the RIM101 pH response pathway, cell-cycle regulation, DNA damage response, cell-wall assembly, mitochondrial function, and protein transport.

Figure 1.—

Genetic interaction network of AFT1, AFT2, FET3, and RIM101. Genome-wide SL–SGA screens were performed using query strains for aft1Δ (YKB676), aft2Δ (YKB 1010), fet3Δ (YKB 1009), and rim101Δ (YKB 1008) and a genome-wide SDL–SGA screen was performed using a query strain containing the galactose inducible pGAL–AFT1 plasmid (YKB794). Genes are represented by nodes that are color coded according to their SGD cellular roles and/or assigned through review of literature. Interactions are represented by edges. AFT1 SDL–SGA central node is indicated by AFT1 whereas SL–SGA central nodes are indicated by Δ. Deletion mutants that are hypersensitive to decreases in iron are indicated by *.

AFT1 genetic interaction map is not fully attributable to iron deficiency:

It has recently been shown that aft1Δ cells have decreased cellular iron levels when cultured under normal iron conditions (Veatch et al. 2009), but it is not known if overexpression of AFT1 increases cellular iron levels. However, as the constitutively active aft1-1up allele results in the induction of the iron regulon and G1 arrest (Philpott et al. 1998; Rutherford et al. 2003; Shakoury-Elizeh et al. 2004), it is likely that overexpression of AFT1, which causes G1 arrest (Casas et al. 1997), is also inducing the iron regulon and iron influx. Hence, the mutants identified in the AFT1 SL and SDL screens may be sensitive to cellular iron fluctuations. Alternatively, some of the AFT1 genetic interactions may reflect novel iron-independent roles for Aft1. In an attempt to differentiate between iron-sensitive and iron-independent roles for Aft1, we undertook secondary chemical and genetic studies.

Numerous studies have systematically screened the yeast deletion mutant arrays for growth sensitivity to iron-limiting conditions (Davis-Kaplan et al. 2004; Dudley et al. 2005; Lesuisse et al. 2005; Jo et al. 2009) or iron toxicity (Jo et al. 2008). As there is low concordance between these screens, we decided to directly assess the growth of the 77 deletion mutants identified in the AFT1 SL and SDL genetic network for growth on low (2.5 μm FeS04) and high (500 μm and 1000 μm FeS04) iron media (Figure 1 and Table 2 and Table 3). Although we did not identify any mutants that were inhibited by elevated iron levels, 15 deletion mutants were sensitive to decreased iron levels. An additional 10 deletion mutants have been shown to be sensitive to low iron levels in media in other screens (Table 2 and Table 3). As has been previously suggested (Jo et al. 2009), the discrepancy between the screens is likely due to media effects. Regardless, between the previously published genome-wide screens and our direct testing, less than one-third of the deletion mutants in the AFT1 genetic network display sensitivity to iron.

Further, we hypothesized that if an AFT1 genetic interaction was the result of sensitivity to limited iron availability, the deletion mutant may also have genetic interactions with other mutants of the iron regulon, in particular AFT2 and FET3. To explore this possibility, genome-wide aft2Δ and fet3Δ SL–SGA screens were performed as described above (Figure 1 and Table 2). As the role of Aft2 in iron response is secondary to Aft1 (Rutherford et al. 2003) and as aft2Δ cells do not display growth defects under low-iron conditions (Blaiseau et al. 2001), we were not surprised to identify only four mutants, spe2Δ, med1Δ, mms22Δ, and yel007wΔ, that displayed synthetic sickness with aft2Δ. In contrast, as Fet3 is an essential component of the high-affinity iron transport complex and fet3Δ mutants are sensitive to iron depletion (Davis-Kaplan et al. 2004), we expected to identify numerous mutants implicated in iron homeostasis in the fet3Δ SL screen. Ten mutants were identified with synthetic genetic interactions with fet3Δ, including deletion of the low-affinity iron transporter FET4 and the copper transporter CCS1, and five of the mutants identified are sensitive to decreases in iron in media (Figure 1 and Table 2). Why did fet3Δ mutants interact only with a subset of iron-sensitive mutants that interact with aft1Δ mutants? While deletion of FET3 eliminates the function of the high-affinity iron transport complex, deletion of AFT1 downregulates not just the high-affinity iron transport from outside the cell, but the mobilization of iron stores in the vacuole or mitochondria and remodeling of cellular pathways to free iron cofactors from enzymes. Hence, the difference in interactions may be reflective of the difference in the cellular iron levels in the mutants. The fact that many of the mutants identified in the AFT1 network are not sensitive to extracellular free iron concentrations or genetically interact with aft2Δ or fet3Δ suggests that Aft1 may affect numerous processes independently of iron homeostasis.

Aft1 and the RIM101 pH response pathway function in parallel:

A striking feature of the AFT1 genetic network is the identification of numerous genes with established roles in the RIM101 pH response pathway (reviewed in Penalva et al. 2008). The RIM101 pH pathway plays a role in the transcriptional response to alkaline pH (Lamb et al. 2001), as well as cell-wall assembly (Castrejon et al. 2006), sporulation (Su and Mitchell 1993; Li and Mitchell 1997), and ion homeostasis (Lamb et al. 2001). The zinc-finger transcription factor Rim101 is a repressor whose primary targets are two transcriptional repressor genes, SMP1 and NRG1 (Lamb and Mitchell 2003). Hence, Rim101 acts as both a repressor of transcription (through direct binding of promoters) and an activator of transcription (indirectly through the inactivation of repressors). In its full-length form, Rim101 is inactive and requires the proteolytic cleavage of the C-terminal region to become an active repressor. The cleavage of Rim101 is tightly regulated by a variety of processing gene products including the putative transmembrane proteins Rim21 and Rim9, the arrestin-like protein Rim8, the protease Rim13, and the protease scaffold protein Rim20 (reviewed in Penalva et al. 2008). The pH signal transduction and activation of Rim101 also requires ESCRT (endosomal sorting complex required for transport) complexes I, II, and the Snf7–Vps20 subcomplex of ESCRT-III (reviewed in Penalva et al. 2008). In addition to RIM101, the AFT1 genetic network identified RIM9, RIM20, RIM21, along with ESCRT components SRN2, STP22, VPS24, and VPS36. Most of the RIM101 pH response pathway mutants identified in the AFT1 screen are sensitive to decreased extracellular iron levels and fet3Δ also displayed genetic interactions with deletion mutants of RIM101, RIM20, RIM21, and STP22. This suggests that the interaction between these two transcriptional pathways is due to decreased cellular iron levels of aft1Δ cells. Indeed, this is the case as we observed that exogenous iron can suppress the slow-growth defects of aft1Δrim101Δ cells (Figure 2A ).

Figure 2.—

Aft1 and Rim101 function in parallel in alkaline response and Aft1 has a role cell-wall stability and ion homeostasis. (A) Exogenous iron suppresses the synthetic sickness of aft1Δrim101Δ. Wild-type (WT) (YPH499), aft1Δ (YPH1735), rim101Δ (YKB1110), and aft1Δrim101Δ (YKB1111) cells were plated in fivefold serial dilution onto YPD or YPD supplemented with exogenous iron (YPD +90 μm BPS, 100 μm FeS04) as indicated. The plates were incubated for 3 days at 25°. (B) Alkaline induction of FET3–lacZ reporter is dependent on Aft1 and independent of Rim101. WT (YPH499), aft1Δ (YPH1735), rim101Δ (YKB1110), and aft1Δrim101Δ (YKB1111) cells were transformed with either the vector control (pMELb2) or FET3–lacZ construct (pMELb2–FET3–lacZ). The transformed cells were grown in SD–uracil to mid-log phase and then grown for at least two doublings in SD–uracil pH 4 or 8 and the specific activity of β-galactosidase (Miller units) was measured. Data are the mean of three independent transformants and the error bar is 1 standard deviation. (C) Aft1 has a role in cell-wall stability and ion homeostasis. WT (YPH499), aft1Δ (YPH1735) cells were plated in fivefold serial dilution onto YPD or YPD supplemented with exogenous iron (YPD + 90 μm BPS, 100 μm FeS04) that was supplemented with calcoflour white (CFW), SDS, caffeine, LiCl, and NaCl as indicated. The plates were incubated for 2 days at 30°.

The strong genetic interactions between aft1Δ and the RIM101 pathway mutants suggest these two transcriptional pathways are functioning in parallel to regulate similar biological processes. Like RIM101 pathway mutants, aft1Δ cells are also sensitive to alkaline pH (Serrano et al. 2004) and have sporulation defects (Gil et al. 1991). Furthermore, microarray studies have shown that upon alkaline pH treatment, the expression levels of iron regulon genes are induced (Lamb et al. 2001) presumably to compensate for decreases in iron availability in alkaline conditions (Serrano et al. 2004). However, while the alkaline induction of some iron regulon genes, like TIS11, appears independent of Rim101, others like ARN4 are dependent on Rim101 (Lamb et al. 2001; Barwell et al. 2005). Hence it is unclear if Rim101 and Aft1 are functioning in parallel or within a single pathway during alkaline response. Nor is it known if the RIM101 pH pathway also plays a role in any other cellular processes that are also affected by Aft1.

To explore these questions, a genome-wide rim101Δ SL–SGA screen was performed (Figure 1 and Table 4). The resulting confirmed data set contains 26 genetic interactions of which 22% (6/26) were synthetic lethal. aft1Δ and rim101Δ only share five common synthetic genetic interactions with IMG2, MED1, SAC7, YOR331C, and MRE11 (Figure 1). However, both screens identified genes implicated in iron regulation, cell-wall assembly, and sporulation, further providing credence that both Aft1 and Rim101 participate in these cellular functions. The other cellular roles of Aft1, such as chromosome stability and cell-cycle regulation, were not identified in the rim101Δ SL–SGA screen suggesting that the participation of Aft1 in these processes is not shared between the pathways. Likewise, the rim101Δ network identified pathways not identified in the aft1Δ network, such as trehalose-6-phosphate synthase (TPS1 and TPS2). The lack of overlap between the screens suggests that although these transcriptional pathways affect similar cellular functions they are doing so in parallel.

View this table:
TABLE 4

RIM101 synthetic lethal interactions

To further confirm that Aft1 and the RIM101 pH pathway are functioning independently during the alkaline response, β-galactosidase assays were performed using a FET3 promoter–lacZ fusion (Kimura et al. 2007) in wild-type, aft1Δ, rim101Δ, and aft1Δrim101Δ cells after growth at pH 4 and pH 8 (Figure 2B). While the lacZ fusion vector with no insert was not active at either pH, as expected the FET3lacZ fusion expression was induced upon alkaline treatment and this induction was dependent on Aft1. Deletion of RIM101 did not suppress induction of the FET3lacZ; rather the FET3lacZ alkaline induction was increased by 50% in the absence of RIM101. Similar phenomena have been reported for ARN1 (Lamb and Mitchell 2003; Barwell et al. 2005) and FRE1 (Lamb et al. 2001), which suggest that Rim101 may be a negative regulator of some iron regulon genes.

The RIM101 pH pathway also contributes to cell-wall assembly (Castrejon et al. 2006) and ion homeostasis (Lamb et al. 2001); hence we were curious to determine if Aft1 plays a role in these processes too. We asked if aft1Δ cells are hypersensitive to cell-wall-damaging agents calcoflour white (CFW), which interferes with cell-wall assembly by binding to chitin, sodium-dodecyl sulfate (SDS), which interferes with membrane synthesis, and caffeine, which activates a map kinase cascade altering cell-wall assembly (Figure 2C). Although aft1Δ cells are only mildly sensitive to 0.002% SDS, they are hypersensitive to 4 mm caffeine and 5 μg/ml CFW treatment indicating that Aft1 contributes to cell-wall maintenance. We also asked if, like rim101Δ cells, aft1Δ cells display growth defects in the presence of cations NaCl and LiCl. While aft1Δ cells are only mildly sensitive to LiCl, they are hypersensitive to NaCl, which suggests that Aft1 also plays a role in ion homeostasis. Exogenous iron suppressed the sensitivity of aft1Δ cells to caffeine, SDS, and LiCl treatment, suggesting that the role of Aft1 in response to these challenges is mediated through iron availability in the cell. In contrast, exogenous iron cannot suppress aft1Δ cells sensitive to CFW or NaCl, which suggests that the role of Aft1 in response to these environmental challenges is mediated through an iron-independent mechanism.

Iron, Aft1 and DNA Repair:

The genetic map links both Aft1 and the RIM101 pH pathway to DNA damage repair and many lines of evidence have already linked Aft1 to this cellular process. A chemical genomics study found that aft1Δ mutant cells are hypersensitive to interstrand cross-linking DNA damaging agents such as carboplatin and cisplatin (Lee et al. 2005), and cisplatin treatment has been shown to induce the iron regulon through activation of Aft1 (Kimura et al. 2007). However, other members of the iron regulon have not been identified as being hypersensitive to DNA-damaging agents (Bennett et al. 2001; Chang et al. 2002; Lee et al. 2005). Nor have mutants of the RIM101 pH pathway been identified as being sensitive to DNA damage in systematic chemical genomic screens. Therefore we were interested in further exploring the connection between Aft1, Rim101, and DNA damage. While aft1Δ cells hypersensitivity to cisplatin could be suppressed by exogenous iron, rim101Δ cells are not sensitive to cisplatin treatment (Figure 3A ). Despite the fact that aft1Δ, rim101Δ, and other mutants in these pathways have not been identified as being hypersensitive to MMS in genome-wide screens (Bennett et al. 2001; Chang et al. 2002; Lee et al. 2005), we decided to test their sensitivities directly. We determined that neither aft1Δ nor rim101Δ cells had significant hypersensitivity to 0.005, 0.02, or 0.035% MMS treatment (data not shown). However, although dramatic growth inhibition occurs for all strains tested at 0.05% MMS treatment, aft1Δ and aft1Δrim101Δ cells displayed increased sensitivity compared to wild-type cells (Figure 3B). Although subtle, rim101Δ cells were slightly more sensitive than wild-type cells to 0.05% MMS treatment. We were surprised to find that MMS effects even on wild-type cells were exacerbated upon depletion of iron (2.5 μm FeS04) and rescued upon increasing iron levels, with aft1Δ cells requiring higher levels of iron for rescue. This strongly suggests that iron has protective effects against DNA damage, likely through its role as a cofactor in a variety of DNA repair proteins (reviewed in Lill and Muhlenhoff 2008).

Figure 3.—

Exogenous iron buffers the effects of MMS and cisplatin. (A) Exogenous iron suppresses the hypersensitivity of aft1Δ mutants to cisplatin treatment. WT (YPH499), aft1Δ (YPH1735), rim101Δ (YKB1110), and aft1Δrim101Δ (YKB1111) cells were 10-fold serially diluted onto YPD or YPD supplemented with exogenous iron (YPD +90 μm BPS, 100 μm FeS04) that contained DMSO (carrier control) or cisplatin as indicated. The plates were incubated for 2 days at 25°. (B) Exogenous iron levels modulate the cellular effects of MMS. The strains indicated above were 5-fold serially diluted onto YPD plates containing DMSO or YPD plates containing 0.05% MMS with varying levels of iron as indicated. The plates were incubated for 4 days at 25°.

The role of Aft1 in chromosome stability is iron independent:

Aft1 is required for faithful chromosome transmission under normal iron media conditions (Measday et al. 2005; Yuen et al. 2007). As many mutants with defects in chromosome stability also have increased sensitivity to the microtubule-destabilizing drug benomyl (Sora et al. 1982), we asked whether aft1Δ mutants are also hypersensitive to this compound. As expected, aft1Δ mutant cells are hypersensitive to benomyl treatment compared to wild-type cells (Figure 4A ). A deletion mutant of AFT2 does not display chromosome segregation defects (data not shown and Measday et al. 2005; Yuen et al. 2007) and as expected aft2Δ cells are not hypersensitive to benomyl. Unlike the hypersensitivity of aft1Δ cells to HU (Dubacq et al. 2006), SDS, caffeine, LiCl (Figure 2C), cisplatin, and MMS (Figure 3), which are suppressed by exogenous iron, the hypersensitivity of aft1Δ cells to benomyl cannot be rescued by increasing levels of exogenous iron in the growth medium (Figure 4A). Further, using a FET3–lacZ reporter assay we determined that unlike iron-restricted conditions or alkaline pH, benomyl treatment does not induce FET3–lacZ; rather we see reduction of reporter activity upon benomyl treatment (Figure 4B). This is in agreement with microarray studies that have not detected changes in expression of iron regulon genes upon benomyl treatment (Lucau-Danila et al. 2005). Our work indicates that aft1Δ hypersensitivity to benomyl is not the result of defects in the induction of iron regulon genes and cellular iron levels.

Figure 4.—

The benomyl hypersensitivity of aft1Δ cells is not due to defects in iron homeostasis. (A) aft1Δ cells' hypersensitivity to benomyl is not suppressed by exogenous iron. Wild-type (WT, YPH499), aft1Δ (YPH1735), and aft2Δ (YKB788) cells were fivefold serially diluted onto YPD plates containing either DMSO or 10 μg/ml benomyl and supplemented with varying levels of iron (FeS04) as indicated. The plates were incubated for 2 days at 30°. (B) Benomyl treatment does not induce a FET3–lacZ reporter. Wild-type (WT, YPH499), and aft1Δ (YPH1735) cells were transformed with either the vector control (pMELb2) or FET3–lacZ construct (pMELb2–FET3–lacZ). The transformed cells were grown to SD–uracil to mid-log phase and collected (untreated) or treated with 20 μg/ml benomyl for 1 hr and the specific activity of β-galactosidase (Miller units) was measured. Data are the mean of three independent transformants and the error bar is 1 standard deviation.

If the role of Aft1 in faithful chromosome segregation is mediated by cellular iron levels, one would predict that exogenous iron could suppress chromosome loss defects in aft1Δ cells. To test this hypothesis, we performed a series of chromosome transmission fidelity (CTF) assays (Koshland and Hieter 1987) in which wild-type, aft1Δ, and ctf13-30 cells, an inner kinetochore mutant with extremely high rates of chromosome segregation defects (Doheny et al. 1993), were plated onto YPD media or YPD + 90 μm BPS + 100 μm FeS04 (Table 5). As the CTF assay is measuring chromosome loss in the first cell division, cells plated from YPD onto YPD + 90 μm BPS + 100 μm FeS04 may not have enough time to readjust intracellular iron levels prior to the first cell division to impact CTF. Therefore, cells were first cultured on media containing exogenous iron (see material and methods). aft1Δ cells have a chromosome missegregation rate ninefold greater than that of wild-type cells, but less than that of the essential kinetochore mutant ctf13-30. Further, the addition of exogenous iron did not affect chromosome loss rates of the wild-type cells and could not suppress CTF defects of aft1Δ or ctf13-30 cells. Together these assays suggest that the role of Aft1 in chromosome stability is distinct from its role in transcriptional regulation of the iron regulon and iron homeostasis.

View this table:
TABLE 5

Rates of chromosome loss events

Microarray experiments suggest the role of Aft1 in chromosome stability and benomyl response is transcription independent:

Numerous microarray studies in iron replete media have indicated that the impact of Aft1 on transcription is not limited to iron regulon genes (Shakoury-Elizeh et al. 2004; Pagani et al. 2007). Hence, we hypothesized that Aft1 could be regulating the transcription of a non-iron regulon gene required for resistance to benomyl and chromosome stability. We first asked if mutants corresponding to any genes that have been reported to be downregulated in aft1Δ cells (Pagani et al. 2007) or upregulated in aft1-1up mutants cells (Shakoury-Elizeh et al. 2004) have been identified as being both hypersensitive to benomyl (as listed on SGD) and displaying CTF defects in genome-wide screens (Yuen et al. 2007). Neither screen identified genes whose mutants are both benomyl sensitive and display CTF defects (Table 6).

View this table:
TABLE 6

Aft1-regulated genes that may have a role in benomyl response or CTF

To further explore the possibility that Aft1 is regulating the transcription of key genes required for benomyl resistance and chromosome stability, we performed a series of microarray experiments to compare the transcriptional response of aft1Δ cells vs. wild-type cells grown in YPD media. As it is possible that the hypersensitivity of aft1Δ cells to benomyl is due to a role of Aft1 in the transcriptional response to benomyl (Lucau-Danila et al. 2005), genome-wide expression profiles were also compared after treatment with 20 μg/ml benomyl for 20 min. The vast majority of benomyl-dependent transcriptional remodeling is detected using these conditions (Lucau-Danila et al. 2005). Similar to other groups, we found that deletion of aft1Δ results in a global transcriptional remodeling under YPD culture conditions (File S1). Of the genes we identified whose transcription is decreased twofold or more (P-value < 0.05) only deletion mutants of CTF19 have been demonstrated both to be hypersensitive to benomyl and to display chromosome transmission defects (Hyland et al. 1999). CTF19 encodes a component of the central kinetochore complex COMA (reviewed in Westermann et al. 2007). Intriguingly, genome-wide ChIP studies had reported an enrichment of Aft1 to the intergenic region of 5′ of CTF19 (Harbison et al. 2004). We were interested in determining if aft1Δ cells hypersensitivity to benomyl could be explained by decreases in CTF19 gene levels. However, neither an extra genomic clone of CTF19 (pKH5) or an HA-tagged CTF19 fusion clone (pKH32) (Hyland et al. 1999) could suppress the benomyl sensitivity of aft1Δ cells (Figure 5A ). Further, although we could detect the localization of Aft1–TAP to the promoter of the iron regulon gene FET3 by ChIP, we could not detect it on the promoter of CTF19 (Figure 5B). Together this suggests that Aft1 is likely not regulating the transcription of CTF19 directly, nor is the decreased levels of CTF19 in aft1Δ cells likely the reason for aft1Δ benomyl sensitivity. Although this candidate approach is limited by both the microarrays and genome-wide data sets available for benomyl sensitivity and CTF, especially for essential genes, it suggests that under YPD conditions Aft1 may not be affecting chromosome stability through transcription.

Figure 5.—

CTF19 does not rescue the benomyl sensitivity of aft1Δ cells. (A) Wild-type (WT, YPH499) and aft1Δ (YKB1095) cells transformed with pRS315 (vector control), pKH5 (genomic fragment containing CTF19), or pKH32 (HA-tagged CTF19 fusion clone) were fivefold serially diluted onto YPD plates containing either DMSO or 10 μg/ml benomyl. The plates were incubated for 2 days at 30°. (B) Aft1 does not localize to the promoter of CTF19. Modified ChIP was performed using untagged (WT; YPH499) and Aft1–TAP (YKB479) strains. Total or immunoprecipitated (IP) DNA was subjected to multiplex PCR amplification using primers specific to the promoter region of CTF19, FET3, and a subtelomeric region of chromosome V (TEL-V). The result of this ChIP was representative of three experiments.

Similar to previous studies, benomyl treatment dramatically affects the transcription profile of wild-type cells, resulting in the twofold induction of 351 genes and the twofold repression of 495 genes (wild type [WT] + BEN/WT; P-value < 0.05; see File S1). Similarly, benomyl treatment of aft1Δ cells resulted in the induction of 421 genes and repression of 1094 genes (aft1Δ + BEN/aft1Δ, P-value < 0.05). In agreement with our reporter assay (Figure 4B) and previous expression profiles (Lucau-Danila et al. 2005), our analysis confirms that benomyl treatment does not induce the transcription of the iron regulon. Despite the dramatic transcriptional remodeling upon benomyl treatment that occurs in these two strains, the transcriptional differences between aft1Δ and wild-type cells treated with benomyl (aft1Δ + BEN/WT + BEN, P-value < 0.05) are mild. Deletion of AFT1 only negatively impacted the transcription of 35 genes and positively impacted the transcription of 90 genes compared to wild type upon benomyl treatment (minimum twofold change, P-value < 0.05). If Aft1 were playing a transcriptional role in the benomyl response, one would predict that a subgroup of the 351 genes whose expression is induced upon benomyl treatment in wild-type cells would no longer be induced in aft1Δ cells. However, of the 35 genes whose expression upon benomyl treatment is significantly reduced in aft1Δ cells compared to wild-type cells, the majority were already downregulated in aft1Δ cells in YPD or the transcription of these genes is decreased in both wild-type and aft1Δ cells upon benomyl treatment, with the effect greater in the mutant (group I, Figure 6 ). Of the nine genes with an observed induction of twofold or greater in wild-type cells upon benomyl treatment (P-value > 0.05), the expression of six of these genes is also upregulated in aft1Δ cells, but to a lesser extent than in wild-type cells (group II, Figure 6). Only the benomyl induction of three genes, SYN8, MET3, and MET14, appears to be dependent on Aft1. If we extend this analysis to genes that are induced 1.8-fold or more, an additional three genes, LST8, GRX8, and MET5, can be added to the group of genes whose benomyl induction is dependent on Aft1 (group III, Figure 6). Interestingly, MET3, MET5, MET14, and GRX8 are genes of the newly defined 45 gene Met4 regulon (Lee et al. 2010). Met4 is a transcriptional activator that in conjunction with its DNA-binding cofactors Met31/32 or Cbf1 tightly regulates the transcription of the sulfur metabolic network or Met4 regulon in yeast. This suggests that the Met4 regulon is induced upon benomyl treatment in an Aft1-dependent manner. Is it possible that the chromosome stability defects and benomyl sensitivity of aft1Δ cells could be due to defects in the sulfur metabolic network? To our knowledge no mutants of Met4 regulon genes, except for Cbf1 (see discussion), display chromosome transmission fidelity defects or are hypersensitive to benomyl. Nor have compounds synthesized by the sulfur biosynthetic network, such as methionine and cysteine, S-adenosylmethionine, or glutathione been implicated directly in chromosome stability in yeast. Our study suggests that although Aft1 may have a role in the benomyl-induction of the Met4 regulon, the overall contribution of Aft1 to the benomyl transcriptional response is minimal. Taken together, our study argues for a novel nontranscriptional role of Aft1 in chromosome stability.

Figure 6.—

The microarray profiles of the 35 genes whose expression in benomyl is reduced twofold or more (P -value < 0.05) in aft1Δ cells compared to wild-type cells (aft1Δ BEN/ WT BEN). The 2D hierarchical cluster analysis of the expression profiles of the 35 genes was performed. Expression data are represented on a log2 scale, with inductions marked with red and repression marked with green. For aft1Δ/WT, aft1Δ BEN/aft1Δ, and WT BEN/WT expression analysis includes expression data with fold-changes less than twofold and/or P-values > 0.05. Genes whose transcript was significantly induced twofold or greater in wild-type cells upon benomyl treatment (WT BEN/WT) are marked with an asterisk (*). Gene groups I, II, and III are discussed in the text.

DISCUSSION

Genome-wide genetic screens identify diverse cellular roles for Aft1:

In an effort to further define the cellular functions of Aft1, SGA methodology was used to perform complementary genome-wide SL and SDL screens (Tables 2 and Tables 3). As expected, the genetic interaction map identified genes encoding proteins implicated in processes previously linked to Aft1 including iron regulation (reviewed in Rutherford and Bird 2004), chromosome stability (Measday et al. 2005), cell-cycle progression (Philpott et al. 1998; Jorgensen et al. 2002; White et al. 2009), and DNA damage repair (Lee et al. 2005; Kimura et al. 2007). Further, the AFT1 genetic interaction map also predicts possible functional roles for Aft1 in cell-wall assembly, protein transport, and the mitochondria.

The diverse range of cellular functions suggests two possibilities. One is that Aft1 is not directly affecting these pathways per say; rather the deletion mutants identified cannot tolerate fluctuations in cellular iron content and likely encode proteins that work in parallel with pathways that functionally require iron cofactors. As a third of the mutants identified in the AFT1 genetic network are sensitive to decreased levels of iron (Figure 1 and Tables 2 and Tables 3), this is likely an explanation for a subset of the interactions. Furthermore, it suggests that aft1Δ cells have decreased intracellular iron levels even when cultured under iron-replete conditions, which is in agreement with a recent study that showed that aft1Δ mutants have twofold decreases in cellular iron content (Veatch et al. 2009). Additionally, many of the chemical sensitivities displayed by aft1Δ mutants, such as to HU (Dubacq et al. 2006), SDS, caffeine (Figure 2), cisplatin, and MMS (Figure 3), can be suppressed by exogenous iron. This suggests that enzymes that are central to DNA replication, DNA damage response, and some cell-wall challenges require iron cofactors to function and the decreased cellular iron levels of aft1Δ cells compromise the function of these pathways. This is certainly the case for the DNA damage response where numerous proteins, like Rad3, a DNA helicase involved in nucleotide excision repair (Rudolf et al. 2006), and Pri2, a subunit of DNA primase involved in both DNA replication and double-strand-break repair (Klinge et al. 2007), require ISCs to function. Although the scope of enzymes that require iron cofactors has not been systematically assessed, it is clear from our and other genome-wide screens that iron is an essential cofactor for a myriad of diverse cellular processes (Davis-Kaplan et al. 2004; Dudley et al. 2005; Lesuisse et al. 2005; Jo et al. 2008; Jo et al. 2009). As mutant sensitivity to iron fluctuations may result in subtle or distinct phenotypes that we or others have not detected, it is likely that we are underestimating the number of iron-sensitive mutants in the AFT1 network. Nonetheless, the majority of the deletion mutants in the AFT1 network do not have detectable sensitivities to either decreases or increases in iron in the media, or display genetic interactions with either fet3Δ or aft2Δ. Hence, our genetic network suggests that Aft1 has cellular roles that are independent from its role in inducing the iron regulon and regulating cellular iron levels.

Aft1 and the RIM101 pH pathway—connected by iron:

One of the most striking features of our genetic screens is the links between AFT1 and the RIM101 pH pathway. Most (7/8) of the mutants implicated in the RIM101 pH pathway that were identified in the AFT1 genetic network cannot tolerate low levels of iron and many also genetically interacted with the fet3Δ mutant (Figure 1 and Table 2 and Table 3). In addition, we show that the synthetic sick interaction displayed by aft1Δrim101Δ cells can be suppressed by exogenous iron (Figure 2A and 3), which suggests that the RIM101 pH response pathway is regulating cellular processes that may work in parallel with iron-dependent pathways. Indeed both rim101Δ and aft1Δ mutants display similar defects and sensitivities. The RIM101 pH pathway is known to be involved in numerous cellular processes: alkaline pH response (Hayashi et al. 2005), sporulation (Su and Mitchell 1993; Li and Mitchell 1997), ion homeostasis (Lamb et al. 2001), and cell-wall assembly (Castrejon et al. 2006). Similarly, Aft1 has been implicated in alkaline pH response (Serrano et al. 2004), sporulation (Gil et al. 1991), cell wall (Figure 2C), and ion homeostasis (Figure 2C). Interestingly, while the aft1Δ cells' sensitivity to SDS, caffeine, and LiCl can be suppressed by exogenous iron, suggesting that iron is a cofactor for key proteins required for resistance to these treatments, exogenous iron cannot suppress the sensitivity of aft1Δ cells to CFW or NaCl. This suggests that Aft1 potentially buffers the effects of these compounds by a novel iron-independent mechanism. These results demonstrate that Aft1 plays a functional role in three cellular processes also regulated by Rim101 and suggests that these two transcriptional pathways work in parallel to govern similar cellular functions.

However, the interplay between these transcriptional cascades is likely more complicated. Reporter assays, Northern blot analysis, and microarray gene expression studies have shown that both the expression in standard YPD and the alkaline induction of the iron regulon gene ARN4 is dependent on Rim101 (Lamb et al. 2001; Lamb and Mitchell 2003; Barwell et al. 2005). In addition, the expression levels of FRE2, FRE3, FRE4, ARN2, and FIT1 are reduced in rim101Δ cells under alkaline conditions (Barwell et al. 2005). In contrast, the expression of ARN1 (Lamb and Mitchell 2003; Barwell et al. 2005), FRE1 (Lamb et al. 2001), and a FET3–lacZ reporter (Figure 2) is increased in the absence of Rim101 in alkaline treatment. Rim101 functions as a repressor through binding of promoters (reviewed in Penalva et al. 2008) and indirectly as an activator through the repression of negative-acting genes SMP1 and NRG1 (Lamb and Mitchell 2003). Could Rim101 be acting as both a repressor and activator of a subset of iron regulon genes? Directed studies did not detect Rim101 on the promoter of ARN4 by chromatin immunoprecipitation (Lamb and Mitchell 2003) and global ChIP studies only detected weak enrichment of Rim101 to the promoters of iron regulon genes FET5 and ARN1, while Nrg1 or Smp1 localization was not detected on the promoters of any iron regulon genes (Harbison et al. 2004). The mechanism through which Rim101 affects iron regulon gene expression will require further investigation.

An iron-independent role for Aft1 in chromosome stability:

We found that increased exogenous iron cannot suppress the chromosome fragment loss (Table 5) or the benomyl sensitivity (Figure 4A) of aft1Δ cells. In addition iron regulon genes are not induced upon benomyl treatment (Figure 4B, File S1, and Lucau-Danila et al. 2005). It is also important to note that, except for AFT1, no other iron regulon genes have been identified in genome-wide screens measuring genome instability by various assays (Kanellis et al. 2007; Yuen et al. 2007; Andersen et al. 2008). Our results indicate that the role of Aft1 in chromosome stability is distinct from its role as a transcriptional inducer of the iron regulon and iron homeostasis.

How could Aft1 be regulating chromosome stability? One possibility is that Aft1 is regulating chromosome stability through transcription of key genes that encode proteins required for genome maintenance and benomyl resistance. Our candidate approach identified only one potential gene (Table 6); however, we show that the transcriptional regulation of CTF19 is likely not the means through which Aft1 contributes to genome stability (Figure 5). Similarly, we determined that Aft1 plays only a minor role in the benomyl transcriptional response (Figure 6). Intriguingly, of the few genes whose induction upon benomyl treatment is dependent on Aft1, most are members of the Met4 regulon (Lee et al. 2010). Met4 cannot bind DNA on its own, but rather localizes to its target promoters through interaction with either the partially redundant zinc finger proteins Met31 or Met32 or through the helix–loop–helix protein Cbf1 (reviewed in Lee et al. 2010). As the expression of MET32 in benomyl is decreased in aft1Δ cells, this may explain the overall decrease in the Met4 regulon genes. Could Aft1 play a more direct role in the Met4 transcriptional pathway? Two-hybrid interaction has been detected between Aft1 and Cbf1 (Measday et al. 2005), which has an alternative function as an inner kinetochore protein directly binding centromeric DNA (Cai and Davis 1990). As Aft1 also interacts with kinetochore protein Iml3 in two-hybrid studies (Wong et al. 2007) and Aft1 can co-immunoprecipitate numerous other kinetochore proteins (performed by A.H. and K.B.), one interpretation is that interaction between Aft1 and Cbf1 is solely reflective of a role of Aft1 at the kinetochore. Alternatively, the two-hybrid interaction between Aft1 and Cbf1 may reflect a direct role of Aft1 in regulation of the sulfur metabolic network. The connection between Aft1 and Met4 pathway will need to be further explored. However, except for Cbf1 (Cai and Davis 1990), no other Met4 cofactors or genes of the Met4 regulon have been implicated in chromosome stability or benomyl resistance. This suggests the sensitivity of aft1Δ cells to benomyl treatment is not the result of defects in the transcription of the Met4 regulon.

Although we cannot rule out a transcriptional role for Aft1 in chromosome stability or benomyl resistance, it is more likely that Aft1 is functioning directly at the kinetochore to regulate chromosome loss. As Aft1 has never been identified in the numerous kinetochore affinity chromatography purifications (Westermann et al. 2007), it suggests that the Aft1–kinetochore interaction is transient and potentially plays a regulatory role. Intriguingly, Aft1 has recently been shown to interact with and regulate the ubiquitination state of Arn3 (Jeong et al. 2009). As ubiquitination of the inner kinetochore protein Ctf13 (Kaplan et al. 1997) and the centromeric histone H3 variant Cse4 (Collins et al. 2004) contributes to the regulation of kinetochore complex formation at centromeres, it is tempting to speculate that Aft1 may play a similar role at the kinetochore. A detailed dissection of the Aft1-kinetochore interaction will be required to fully understand the role Aft1 is playing in regulating chromosome stability.

Acknowledgments

We thank A. Naganuma for the kind gifts of the pMELb2-lacZ and pMELb2-FET3-lacZ plasmids, P.Hieter for the kind gifts of the pKH5 and pKH32 plasmids, and V. Measday and members of the Baetz Lab for their thoughtful comments. This work was supported by grants to K.B. from National Sciences and Engineering Research Council of Canada (NSERC), from the Canadian Cancer Society Research Institute, and an Early Researcher Award from the Ontario Government and grants to K.B. and L.H. by the Agricultural Bioproducts Innovation Program from Agriculture and Agri-Food Canada. K.B. is a Canada Research Chair in Chemical and Functional Genomics. N.M. was supported by a NSERC Undergraduate Student Research Award.

Footnotes

  • Received April 7, 2010.
  • Accepted April 26, 2010.

References

View Abstract