Arsenical resistance genes in Saccharomyces douglasii and other yeast species undergo rapid evolution involving genomic rearrangements and duplications

Abstract

We have isolated and characterized three adjacent Saccharomyces douglasii genes that share remarkable structural homology (97% amino acid sequence identity) with Saccharomyces cerevisiae ARR1 (ACR1), ARR2 (ACR2) and ARR3 (ACR3) genes involved in arsenical resistance. The ARR2 and ARR3 genes encoding the cytoplasmic arsenate reductase and the plasma membrane arsenite transporter are functionally interchangeable in both yeast species. In contrast, a single copy of S. douglasii ARR1 gene is not sufficient to complement the arsenic hypersensitivity of a S. cerevisiae mutant lacking the transcriptional activator Arr1p. This inability may be related to a deletion of a 35-bp sequence including the putative Yap-binding element in the ARR1 promoter of S. douglasii. Different mechanisms of regulation of ARR1 genes expression may therefore explain the increased tolerance of S. douglasii to arsenic in comparison with S. cerevisiae. The apparent duplication of the ARR gene cluster in the S. douglasii genome may constitute another factor contributing to the observed differences in arsenic sensitivity. Comparison of ARR genes from the genomes of several yeast species indicates that they are located in subtelomeric regions undergoing rapid evolution involving large-scale genomic rearrangements.

ARR (ACR) gene cluster, Arsenic resistance, Evolution, Genomic rearrangements, Saccharomyces douglasii, Saccharomyces cerevisiae

1 Introduction

Arsenic is a toxic metalloid present in natural and polluted industrial environments. It is a human carcinogen but is also used in treatment of acute promyelocytic leukemia [1,2] and protozoan parasitic diseases, such as sleeping sickness [3]. The abundance of arsenic contributes to the development of resistance to metalloid salts [2]. Hence, it is of great interest to elucidate the mechanisms involved in arsenic biological action and resistance.

Mechanisms of arsenic detoxification have been investigated in various organisms [4]. Arsenical tolerance is well established in prokaryotes, where arsenic detoxification is assured by the ars operon products [5]. In contrast, we have little information about arsenic tolerance in mammals and other eukaryotic species. However, arsenical resistance has been recently reported in the eukaryotic model organism Saccharomyces cerevisiae[6–8].

It has been shown that in S. cerevisiae tolerance to arsenite and arsenate is determined by a cluster of three genes: ACR1, ACR2 and ACR3[6] designated by the Saccharomyces Genome Database [9] as ARR1, ARR2 and ARR3. The ARR1 gene encodes a transcription factor which belongs to the conserved Yap family of bZIP proteins [10]. Arr1p activates the expression of ARR2 and ARR3 genes in response to metalloid exposure [11,12]. The product of the ARR2 gene is an arsenate reductase converting arsenate to arsenite in the presence of glutathione and glutaredoxin [13,14]. The ARR3 gene encodes a plasma membrane efflux transporter which extrudes arsenite out of the cell [7,8]. Another arsenite transporter is Ycf1p, a homologue of the human multidrug resistance-associated Mrp1 protein. Ycf1p catalyzes the transport of arsenite in the form of glutathione conjugates from the cytosol into the vacuole [8]. Inhibition of metalloid influx constitutes an additional level of resistance as the mutant lacking the glycerol channel Fps1p shows increased resistance to antimonite and arsenite [15].

Among known yeast species Saccharomyces douglasii (and its sister species Saccharomyces paradoxus) is the closest relative of S. cerevisiae. The time of divergence between these two species was estimated at around 50 million years ago [16,17]. Comparative studies on the mitochondrial genomes of both species have revealed important differences in gene order and the intron content [18–20]. These changes have led to functional divergence in some of the nuclear genes involved in the processing of the mitochondrial transcripts, in spite of a remarkable level of sequence conservation between the two species [21–24]. In this paper we report the isolation and characterization of S. douglasii ARR genes conferring increased resistance to arsenicals. In view of the recent evolutionary divergence of both yeast species, we compared the structural and regulatory determinants responsible for mechanisms of arsenic tolerance in both species. We have also analyzed the distribution of ARR genes in the known genomic sequences of other yeast species.

2 Materials and methods

2.1 Strains, plasmids and media

S. cerevisiae and S. douglasii strains and plasmids used in this study are listed in Table 1. Standard yeast media as well as standard DNA manipulation methods were employed [25].

1

Yeast strains and plasmids used in this study

Strain/plasmid	Genotype/description	Source
W303-1A	S. c. MAT a ura3-1 leu2-3 112, trp1-1, his3-11 15, ade2-1, can1-100	[47]
RW115	S. c. as W303-1A except acr1Δ::loxP–kanMX–loxP	[36]
EM60	S. c. as W303-1A except acr2Δ::loxP–kanMX–loxP	This study
RW104	S. c. as W303-1A except acr3Δ::loxP–kanMX–loxP	[15]
RW120	S. c. as W303-1A except acr1Δ::loxP–kanMX–loxP, yap1Δ::loxP–kanMX–loxP	[36]
RW124	S. c. as W303-1A except yap1Δ::loxP–kanMX–loxP	[36]
4707-22D	S. d. MAT aaa ura3, leu2, his4, ade1 HO GAL, SUC, mal, αMG	[21]
EM45	S. d. as 4707-22D except arr1Δ::loxP arr1Δ::loxP–kanMX–loxP	This study
EM46	S. d. as 4707-22D except arr2Δ::loxP arr2Δ::loxP–kanMX–loxP	This study
pUG6	Vector containing the loxP–kanMX–loxP deletion cassette	[28]
YCplac33	CEN vector, AmpR, URA3	[48]
YEplac195	2μ, AmpR, URA3	[48]
pFL38	CEN vector, AmpR, URA3	[49]
YEp357R	2μ vector, with lacZ reporter gene, URA3	[50]
pFL38-ACR1	pFL38 with the insert containing 2.56-kb ScARR1	This study
pRW3	pFL39 with the insert containing 2.05-kb ScARR3	[7]
pEM1	YCplac33 with the 1.408-kb insert containing SdARR1	This study
pEM2	YEplac195 with the 1.408-kb insert containing SdARR1	This study
pEM3	YCplac33 with the 0.98-kb insert containing SdARR2	This study
pEM5	YCplac33 with the 1.622-kb insert containing SdARR3	This study
pEM12	YCplac33 with the 0.98-kb insert containing ScARR2	This study
pEM14	Fusion of 330-bp SdARR2 promoter with lacZ in YEp357R	This study
pEM15	Fusion of 286-bp SdARR2 promoter with lacZ in YEp357R	This study
pEM16	Fusion of 329-bp ScARR2 promoter with lacZ in YEp357R	This study
pEM17	Fusion of 285-bp ScARR2 promoter with lacZ in YEp357R	This study
pEM21	Fusion of 307-bp SdARR1 promoter with lacZ in YEp357R	This study
pEM22	Fusion of 342-bp ScARR1 promoter with lacZ in YEp357R	This study

S. c., S. cerevisiae; S. d., S. douglasii.

2.2 Southern analysis

Total DNA was isolated from S. douglasii and digested with EcoRV. The fragments were separated by electrophoresis on 0.7% agarose gel. After electrophoresis, the DNA fragments were transferred to a nylon membrane (Hybond-N⁺, Amersham). The probe was obtained by PCR amplification of a 3395-bp fragment of the S. cerevisiae chromosome XVI using primers PE13 (5^′-CGCCTCTTGCCTTCCCAAT-3^′) and PE3 (5^′- AGATCAATCAGTTGTCCTTC). The probe was labeled with [α-³²P]dCTP by nick translation and hybridized under stringent conditions [26]. The blots were read and quantified using the PhosphorImager (Molecular Dynamics, Sunnyvale, CA).

2.3 DNA sequencing

DNA sequencing was performed by the Sanger chain termination method using the Sequenase Version 2.0 DNA PCR Product Sequencing Kit with [α-³⁵S]dATP as described by the manufacturer (USB, Cleveland, OH). Dideoxynucleotide sequencing reactions were performed on the 3395-bp PCR product amplified with the PE13 and PE3 primers. The sequence of single-stranded templates was determined using custom oligonucleotides derived from the sequenced fragments. Sequences were assembled and analyzed using GCG package version 9.1 (Wisconsin Package Version 9.1, Genetics Computer Group (GCG), Madison, WI) and BLAST on the NIH server [26].

2.4 Disruption of SdARR1, SdARR2 and ScARR2 genes

Deletion of duplicated copies of SdARR1 and SdARR2 genes was performed using the loxP–kanMX–loxP marker flanked by short homology regions to the target gene [27]. Disruption cassettes were obtained in a PCR using the pUG6 (Table 1) plasmid as a template [28] with following primers: for SdARR1: 5^′-GAC CAA GAT AAT GGC AAA ACC GCG TGG AAG AAA AGG CGG CC AGC TGA AGC TTC GTA CGC T-3^′ (forward) and 5^′-TTT GAC GAA AAG ACC TTA ATG AAA TGC TTG AGG CTA ATG CAT AGG CCA CTA GTG GA TCT G-3^′) (reverse) and for SdARR2: 5^′-GCT TGA GGA GCT CAA CCA CTA ACA ATC ATT TAA GGT TAC CAG CTG AAG CTT CGT ACG CT-3^′ (forward) and 5^′-GCG TAA TGG TAA GTT TCA TAA CGT CTA GGC AAC TCA AGG CAT AGG CCA CTA GTG GAT CTG-3^′ (reverse). All disruptions were confirmed by PCR amplification of the novel DNA junction. To delete the second copy of the SdARR1 and SdARR2 genes with the same loxP–kanMX–loxP cassette, the kanMX marker was removed from the single disruptants by inducing homological recombination between the loxP sequences. The cycling conditions and different controls at each step of the procedure have been described previously [15,28]. The disruption cassette obtained with the primers for SdARR2 was also used for deletion of the ARR2 gene in S. cerevisiae (ScARR2).

2.5 Metalloid sensitivity assays

Yeast cells were grown overnight in liquid minimal YNB glucose or complete YPDA medium. The OD₆₀₀ of cultures was adjusted to about 0.3 and 10-fold serial dilutions of the cells suspensions were prepared. Each dilution (3 μl) was spotted on YNB or YPDA agar plates containing various concentrations of metalloid salts. The growth was monitored for 3–5 days at 30 °C.

2.6 Separation of chromosomal DNA

Preparation of DNA samples and separation of chromosomal DNA by the pulsed-field gel electrophoresis (PFG) technique were described by the manufacturer (CHEF-DR^® III Pulsed Field Electrophoresis Systems, Bio-Rad). Southern blot analysis of the separated chromosomes was performed according to Sambrook et al. [25]. The PCR-amplified fragment containing SdARR1, SdARR2 and SdARR3 genes and a plasmid carrying the S. douglasii SUV3 gene were labeled with [α-³²P]dCTP and used as probes.

2.7 β-Galactosidase reporter assay

The strains were transformed with the pEM14, pEM15, pEM16, pEM17, pEM21, pEM22 (Table 1) promoter-reporter plasmid or with YEp357R as a control. The β-galactosidase activity in transformants was determined as described previously [29]. Transformants were cultured in liquid minimal YNB glucose medium at 28 °C to approximate optical density of 1.0 at 600 nm. Cells were grown overnight either with 0.1 mM sodium arsenite or sodium arsenate. For the last 2 h of incubation, the concentration of the drug was increased 10-fold. Measurements were repeated four times for three independent transformants.

2.8 Sequence comparisons and phylogenetic analysis

Sequences from various yeast species analyzed in this study were from [30,31] and were accessed either through their respective websites, or through the Saccharomyces Genome Database [9]. BLAST [26] was used for homology searches. Sequences were aligned using ClustalX [32].

Phylogenetic analyses were performed in PAUP v.4b10 [33] using the parsimony criterion for nucleotide and protein sequences. About 10,000 Heuristic searches were initiated with random addition and tree-bisection-reconnection (TBR) branch swapping. Gaps were treated as missing data. Support was estimated using 1000 replications of jacknife resampling. The final tree printout was obtained with TreeView [34].

3 Results

3.1 Saccharomyces douglasii is more resistant to arsenic compounds than S. cerevisiae

In order to identify genes conferring arsenical resistance in the yeast S. douglasii, we compared the growth of wild-type S. cerevisiae and S. douglasii cells on YPDA plates containing various concentrations of arsenate and arsenite. Two wild-type S. cerevisiae strains, W303-1A and FY1679-28C, were able to grow in the presence of up to 1 mM arsenite or 3 mM arsenate (Fig. 1). The wild-type S. douglasii strain 4707-22D was found, however, to tolerate much higher concentrations of arsenite and could grow in the presence of as much as 3 mM arsenite. In the case of arsenate, S. douglasii cells were somewhat more resistant than S. cerevisiae cells (Fig. 1). These results suggest that the yeasts S. cerevisiae and S. douglasii may possess different mechanisms of arsenite tolerance.

1

The S. douglasii strain 4707-22D displays higher tolerance to arsenic salts than S. cerevisiae W303-1A and FY1979-28C. Plates were prepared as described in Section 2.

3.2 Cluster of ARR genes exists in the S. douglasii genome

Tolerance to high concentrations of arsenicals in S. cerevisiae cells is conferred by the cluster of ARR1, ARR2 and ARR3 genes. To identify arsenical resistance determinants in S. douglasii, we checked for the presence of ARR genes in the S. douglasii genome. Sequences homologous to the ARR were detected in S. douglasii cells using PCR with three pairs of primers binding to the flanking regions of each of S. cerevisiae ARR genes (data not shown). The resulting PCR products were used as probes in Southern blot analysis of S. douglasii genomic DNA digested with EcoRV (Fig. 2). Under stringent conditions used in this experiment we observed that all three probes strongly hybridized with S. douglasii DNA, confirming the presence of all three ARR homologues in S. douglasii.

2

Southern blot analysis of wild-type strains of S. cerevisiae (S. c) W303-1A and S. douglasii (S. d) 4707-22D. The analysis was performed according to Section 2. The ∼3.3-kb fragments of DNA containing the ScARR1, -2, -3 (S. cerevisiae) or SdARR1, -2, -3 (S. douglasii) genes were used as a probes.

In order to obtain the nucleotide sequence of the ARR genes, we amplified a 3324-bp DNA fragment from S. douglasii using the PE13 and PE3 primers, flanking the ARR gene cluster in S. cerevisiae. Analysis of the resulting sequence revealed the existence in the genome of S. douglasii of three open reading frames corresponding to ARR1, ARR2 and ARR3 from S. cerevisiae. The order and orientation of the ARR genes was identical in both species (Fig. 3). The new S. douglasii sequence was registered in the EMBL Data Library under Accession No. AY395569.

3

(A) Organization of the 3324-bp DNA fragment from the S. douglasii genome containing ARR1, ARR2 and ARR3 involved in metalloid resistance and the alignment of the ARR1 promoter region from four related yeast species. In the alignment the reverse-complement (Crick strand) sequence is presented and the putative Yap1-binding motif is shown in bold. (B) Organization of ARR gene clusters in five Saccharomyces species. Genes were identified by their homology to known S. cerevisiae ORFs. Chromosomal localization shown for S. paradoxus and S. kudriavzevii refers to S. cerevisiae chromosomes identified by synteny of ORFS up- or downstream from the ARR cluster.

When the nucleotide sequences of S. cerevisiae and S. douglasii ARR gene clusters were compared, over 95% of sequence identity was revealed (Table 2). Only minor changes, especially in non-coding promoter regions, were detected in S. douglasii. The size of the non-coding promoter region located between SdARR3 and SdARR2 is virtually identical in both yeast species (250 bp in S. cerevisiae and 251 bp in S. douglasii). As listed in Table 4, there is an insertion of one nucleotide and 12 substitutions but the Yap-binding consensus sequence TTAATAA, which is indispensable for the activation of ARR2 and ARR3 expression [35], is perfectly conserved in S. douglasii. In contrast, the non-coding region located between ARR2 and ARR1 is shorter in S. douglasii than in S. cerevisiae (212 and 246 bp, respectively). As shown in Fig. 3A, this is mainly due to a 35-bp insertion in the S. cerevisiae promoter region. Interestingly, this AT-rich sequence which contains a putative Yap-binding site (TTAATAA consensus) is found only in S. cerevisiae and is absent from S. douglasii and other related yeast species.

2

Comparison of the homologous sequences in ARR1, ARR2 and ARR3 of S. cerevisiae and S. douglasii

Parameter	Gene
	ARR1	ARR2	ARR3
Number of nucleotides	885	393	1215
% Homology	98	96	98
Number of nucleotide substitutions	19	15	24
Base transitions	12	10	22
Base transversions	7	5	2
Transitions/transversions ratio	1.71	2.00	11.00

The analysis has been carried out with the DIVERGE program from the GCG 9.1 package.

4

The changes in the ARR2–ARR3 intergenic region of S. cerevisiae and S. douglasii

Nucleotide substitutions in ARR2 ARR3 promoter region

−3

−10

−13

−28

−103

−111

−156

−161

−199

−231

−236

−239

−244

S. douglasii

G

A

C

T

G

A

C

A

A

A

A

G

A

S. cerevisiae

A

T

=

C

A

C

T

T

G

G

G

A

G

-, nucleotide positions are upstream of the ATG start codon (position 0) of ARR3.

This constitutes the most important difference in alignment of ARR gene clusters from S. douglasii and S. cerevisiae. The amino acid sequences of SdARR1, SdARR2 and SdARR3 gene products also show a high degree of homology to their S. cerevisiae orthologues with only 7–9 amino acids replacements detected in each SdARR gene product (Table 3).

3

Polymorphism in Arr1p, Arr2p and Arr3p of S. cerevisiae and S. douglasii

Origin	117	156	188	191	216	226	232	234	293
Amino acid residues in Arr1p
S. douglasii	G	R	S	K	R	S	A	G	R
S. cerevisiae	E	T	T	E	A	P	V	A	K
	11	58	66	80	97	121	128
Amino acid residues in Arr2p
S. douglasii	E	T	N	R	Q	G	A
S. cerevisiae	K	K	S	K	E	E	V
	25	103	106	120	228	232	237	255	297
Amino acid residues in Arr3p
S. douglasii	I	K	V	V	T	A	V	T	S
S. cerevisiae	T	R	I	I	I	G	A	M	L

The S. cerevisiae sequences have the EMBL Accession Nos.: Q06596, Q06597 and Q06598, the S. douglasii sequences have been deposited in the EMBL databank under Accession No. AY395569.

Similar amino acid residues are marked in italics.

3.3 Expression of the ScARR2–lacZ and SdARR2–lacZ reporter genes is induced by arsenite and Acr1p

To explore and compare the role of Acr1p and Yap1p in the regulation of the expression of ScARR2 and SdARR2, we used the lacZ reporter gene fused to the promoter region of ARR2 from S. cerevisiae and S. douglasii. Two constructs of the ARR2 promoter region, differing in length, were tested to verify whether the 41-nucleotide sequence containing a binding motif for the transcriptional factor Yap1 is essential for efficient induction of this promoter. The resulting constructs were then introduced into wild-type and arr1Δ, yap1Δ and arr1Δyap1ΔS. cerevisiae strains. Introduction of the ScARR2–lacZ and SdARR2–lacZ reporter plasmids into the wild-type W303-1A S. cerevisiae strain resulted in a basal level of β-galactosidase activity (Fig. 4). It was, however, significantly stimulated (up to 200-fold) when cells were exposed to arsenite salts. Only very limited induction occurred in the arr1Δ transformant cells treated with arsenite. The strains RW124 and RW120 bearing a deletion of the yap1Δ and arr1Δyap1Δ, respectively, also showed a limited level of induction, especially in the case of the arr1Δyap1Δ double mutant. The results (Fig. 4) indicate that arsenite-dependent expression of the ScARR2–lacZ and SdARR2–lacZ reporters positively correlates with changes of the promoter sequence length (the longer variant significantly increases induction) and is more dependent on Acr1p than on the Yap1p factor.

4

β-Galactosidase activities produced by the ScARR2–lacZ and SdARR2–lacZ reporter genes in long and short versions without induction (no ind.) and after induction by sodium arsenite (AsIII) in wild-type W303-1A and acr1Δ, yap1Δ and acr1Δyap1Δ mutant S. cerevisiae strains. pEM14 and pEM15 plasmids contain the long (330 bp) and short (286 bp) version of the SdARR2 promoter fused to the lacZ gene, respectively. pEM16 or pEM17 plasmids contain the long (329 bp) or short (285 bp) version of ScARR2 promoter fused to the lacZ gene.

3.4 Reciprocal functional complementation of null mutants by ScARR and SdARR

In previous reports, it has been shown that inactivation of S. cerevisiae ARR genes leads to decreased tolerance to arsenite and arsenate salts [6,7,13]. We decided to inactivate the S. douglasii ARR homologues to explore whether disruption of these genes would result in a similar phenotype. The SdARR genes were disrupted by the short-flanking homology method as described in Section 2[27]. During PCR verification of resulting deletion mutants the presence of an additional wild-type copy of each SdARR gene was always observed. It could be explained by the presence of a second copy of the ARR gene cluster in the S. douglasii genome. After excision of the kanMX deletion marker from each initial deletion allele by the expression of Cre recombinase, we were able to disrupt the second copy of SdARR1 and SdARR2 using the same loxP–kanMX–loxP cassette. However, in spite of numerous attempts, we failed to delete the second copy of SdARR3. The reason of this failure remains unknown.

Evaluation of the arsenical-resistance phenotype of the Sdarr1Δ null mutant revealed that it is nine times more sensitive to arsenite and six times more sensitive to arsenate than the wild-type 4707-22D strain, while the Scarr1Δ mutant is four times more sensitive to both arsenite and arsenate than its parental strain W303-1A. In the absence of arsenicals we did not observe any significant growth differences in Sdarr1Δ and Scarr1Δ cells (data not shown). The SdARR2Δ mutant was unable to grow on medium containing arsenate, which corresponds to a phenotype previously observed in the analogous S. cerevisiae arr2Δ strain (Fig. 5B).

5

(A) A single copy of SdARR1 does not restore wild-type resistance to arsenic salts in S. cerevisiae arr1Δ mutant. SdARR2 (B) and SdARR3 (C) are functionally equivalent to their S. cerevisiae counterparts. Ten-fold serial dilutions of yeast cultures were dropped onto minimal YNB glucose medium lacking uracil with various concentrations of arsenicals. Plates were incubated for 3 days at 30 °C. The wild-type strains used as controls were: W303-1A and 4707-22D.

To address the question of functional homology of ARR genes from S. cerevisiae and S. douglasii, we conducted a series of experiments, in which arsenical-sensitive phenotype in the ARR deletion strains of both species was complemented by the plasmid-borne wild-type ARR genes from the same (homologous complementation) or related species (heterologous complementation). The SdARR2 and SdARR3 genes on centromeric plasmids fully complemented arsenic sensitivity of S. cerevisiae arr2Δ and arr3Δ null mutants (Fig. 5B and C). Similarly, the growth of Sdarr2Δ mutant in the presence of arsenate salts was restored by the wild-type ScARR2 gene. These results, summarized in Fig. 5, demonstrate that ARR2 and ARR3 from S. cerevisiae and S. douglasii are functionally interchangeable. Surprisingly, the heterologous ScARR1 gene complemented more efficiently the arsenite and arsenate sensitivity of the Sdarr1Δ strain than the homologous SdARR1 gene. Moreover, the SdARR1 gene expressed from a centromeric plasmid did not abolish the arsenic hypersensitivity phenotype of the Scarr1Δ mutant. SdARR1 was, however, able to restore the wild-type resistance to arsenic in the Scarr1Δ cells, when present on a multicopy plasmid (Fig. 5A). These observations suggest that ScARR1 and SdARR1 are not fully functionally equivalent. It is tempting to associate the higher efficiency of ScARR1 with the presence of an additional Yap binding site in its promoter, which may contribute to stronger expression of the gene. To address this hypothesis, we have examined expression of ScARR1 and SdARR1 using promoter-lacZ fusions and measured β-galactosidase activity after metalloid exposure (Table 5). The level of ScARR1–lacZ and SdARR1–lacZ expression was very low in all strain backgrounds tested and not induced by arsenic compounds. These results indicate that the expression of ScARR1 and SdARR1 is not regulated by metalloids and is not dependent on either Acr1p or Yap1p.

5

Control of the SdARR1–lacZ and ScARR1–lacZ expression in metalloid-exposed cells

Strain	No stress	As(III)b	As(V)b
SdARR1–lacZ/YEp357R (U/OD600)a
Wild-type	1.3 ± 0.1	2.2 ± 0.5	1.9 ± 0.1
Scarr1Δ	1.7 ± 0.4	1.3 ± 0.2	1.4 ± 0.4
yap1Δ	2.8 ± 0.2	2.2 ± 0.5	2.5 ± 0.4
Scarr1Δ, yap1Δ	2.5 ± 0.5	2.1 ± 0.5	2.4 ± 0.3
ScARR1–lacZ/YEp357R (U/OD600)a
Wild-type	2.0 ± 0.2	2.1 ± 0.4	2.6 ± 0.1
Scarr1Δ	1.7 ± 0.4	1.8 ± 0.5	1.5 ± 0.3
yap1Δ	4.7 ± 0.4	3.6 ± 0.3	4.4 ± 0.4
Scarr1Δ, yap1Δ	1.3 ± 0.3	0.9 ± 0.05	1.4 ± 0.3

^aβ-Galactosidase activities are the average of three independent experiments including standard deviation.

^bSdARR1–lacZ and ScARR1–lacZ expression from the YEp357R multicopy plasmid after 20 h of the metalloid preincubation (0.1 mM) followed by 2 h of induction with 1 mM of the respective metalloid.

3.5 Duplication of SdARR in the S. douglasii genome

To confirm the duplication of SdARR in the S. douglasii genome and to compare the molecular karyotypes of these two related species, we isolated the intact chromosomes from both S. cerevisiae and S. douglasii cells. The chromosomes were separated by pulsed-field electrophoresis (PFE), which showed the same number of chromosomes (16) but apparent differences in size of nine chromosomes between the two yeast species. To identify bands corresponding to the chromosomal localization of the ARR gene cluster, a blot of a PFE gel was hybridized with the SdARR1, SdARR2 and SdARR3 probes. The SUV3 gene region [24] localized to the S. cerevisiae chromosome XVI was used as an additional probe in hybridization and, as expected, revealed only one band in the karyotype of each species (Fig. 6). In hybridization with the ARR probes in S. douglasii, we detected two distinct bands, while only one, corresponding as expected to chromosome XVI, was detected in S. cerevisiae (Fig. 6). This analysis strongly indicates that the ARR gene cluster is duplicated only in S. douglasii, and that the chromosomal localization of both copies differs from the one mapped in S. cerevisiae.

6

(A) Sample electrophoretic karyotypes of yeast chromosomes from S. cerevisiae W303-1A (S.c.) and S. douglasii 4707-22D (S.d.). The S. cerevisiae chromosomes were identified using the yeast chromosome PFG marker from New England Biolabs. (B) Southern blots of a PFG gel hybridized with SdSUV3 and SdARR1 probes. The chromosomes were isolated and separated by OFAGE as described in Section 2. Chromosome XVI was identified by the position of the band hybridizing to the SUV3 probe.

3.6 Distribution and evolution of ARR genes in different Saccharomyces species

In order to look into the ARR genes evolution in yeasts, we have searched the databases of Kellis and Cliften [30,31] and the Génolevures project [36] for sequences orthologous to the known ARR genes. In S. cerevisiae, the ARR gene cluster is located in the subtelomeric region on the right arm of chromosome XVI. The synteny viewer tool of the SGD database [9] does not point to any corresponding regions in the sequences [31]. The results of BLAST searches indicate, however, that sequences orthologous to the three ARR genes of S. cerevisiae and S. douglasii can be found in genomes of two Saccharomyces sensu stricto species, S. paradoxus[30] and Saccharomyces kudriavzevii[31], while the genome of a sensu lato Saccharomyces species Saccharomyces castellii[31] contains orthologues to ARR2 and ARR3, but not to ARR1. Sequences orthologous to Arr3p can also be found in some more distant fungal species and in numerous prokaryotic genomes (see [37]). These ARR gene clusters identified in yeast genomes are shown in Fig. 3B.

In S. paradoxus[30], we have identified sequences showing a very high degree of homology with the S. douglasii sequence analyzed in this study. The region spans two contigs (numbers 296 and 517) and contains complete sequences of ARR1 and ARR2, with the ARR3 region truncated close to the 3^′ end of the ORF. This ARR gene cluster can be aligned with the S. douglasii sequence with only minor differences. Most notably, the ARR1 promoter region between the ARR1 and ARR2 ORFs is nearly identical to the S. douglasii equivalent (Fig. 3A). Another S. paradoxus contig (number 494) overlaps with parts of contig 296 and 517 with a few point nucleotide differences. Based on the available sequence data it cannot be determined whether it belongs to the same genomic region (the differences being sequencing errors), or if it corresponds to another paralogous copy of the gene cluster. The S. paradoxus sequence upstream of ARR1 contains an ORF orthologous to the COS6 (YGR295C) gene of S. cerevisiae, which is located in the subtelomeric region of the right arm of chromosome VII. This suggests that a recent subtelomeric rearrangement changed the chromosomal location of the ARR gene cluster. Kellis et al. [30] mention a translocation event between those two chromosomes. The involvement of arsenical resistance genes had not, however, been previously discussed.

In the genome of S. kudriavzevii[31] we found a region (contig no. 1995) containing the cluster of ARR1, ARR2 and ARR3 similar to the one found in S. cerevisiae, S. douglasii and S. paradoxus (Fig. 3B), albeit in a reverse orientation. The region containing the putative Yap-binding site (TTAATAA consensus) is absent from the ARR1 promoter sequence (Fig. 3A). Downstream from the ARR cluster lie three putative ORFs. The first one shows no significant homology to any known S. cerevisiae gene, while the subsequent two are orthologous to YML131W and ERO1 (YML130C) ORFs, which are located in the subtelomeric region of the left arm of S. cerevisiae chromosome XIII. This suggests that another subtelomeric chromosomal rearrangement that occurred during Saccharomyces evolution changed the location of the ARR gene cluster to yet another chromosome.

The genome of S. castellii, a sensu lato Saccharomyces species more distantly related to S. cerevisiae[31], contains a region (contig no. 550) with two ORFs orthologous to ARR2 and ARR3, their relative positions, however, are different (Fig. 3B). No traces of sequences showing even partial homology to ARR1 were found, neither in this region, nor in other S. castellii contigs. Downstream from ARR3 and ARR2 lie an unidentified ORF with no clear orthologues and two ORFs showing high similarity to the HXT and FRE genes of S. cerevisiae. Both HXT and FRE are paralogous families of highly conserved genes, often found in subtelomeric regions (e.g., HXT8, FRE2 and FRE3). This also suggests that subtelomeric rearrangements have shaped the fate of ARR genes in the genome of S. castellii.

Remarkably, no traces on sequences homologous to the ARR genes were found in several known yeast genomes from species such as Saccharomyces mikitae, Saccharomyces bayanus and Saccharomyces kluyveri. Obviously the genomic sequences available have draft status and can be incomplete. One of them, the S. bayanus genome, has however been sequenced independently by two groups [30,31], and the coverage in all cases approached 95%. The possibility that the ARR genes are indeed absent from these genomes is therefore quite real.

In order to compare the evolution of the ARR genes with the established phylogeny of Saccharomyces yeasts, we have inferred maximum parsimony phylogenetic trees based on their nucleotide and amino acid sequences. The resulting single most parsimonious tree obtained using amino acid sequences of all three (two in the case of S. castellii) proteins is shown in Fig. 7. Jacknife support of the tree is very strong for all the branchings. The tree conforms to the conclusions of Cliften et al. [38] and is consistent with the established Saccharomyces phylogeny.

7

Single 377-step most parsimonious phylogenetic tree based on amino acid sequences of Arr1p (except S. castellii), Arr2p and Arr3p. Numbers indicate percent jacknife support.

4 Discussion

In spite of several studies in the past decade, little is known about mechanisms of arsenic tolerance in eukaryotes. In the yeast S. cerevisiae at least three genes have been shown to confer resistance to arsenate and arsenite [4,6,7,11,39]. In this work we report that the yeast S. douglasii is also resistant to arsenic salts especially to arsenite. We have isolated and characterized the S. douglassi ARR1, ARR2 and ARR3 genes orthologous to S. cerevisiae ACR1, ACR2 and ACR3 (ARR). These three genes are closely clustered and their organization resembles that observed in S. cerevisiae. Sequence comparison between ARR genes of S. douglassi and S. cerevisiae has shown the expected high degree of homology (96%). While the functional homology is also essentially maintained, there are significant differences related to the expression and efficiency of ARR1.

The Sdarr1Δ and Scarr1Δ strains are hypersensitive to arsenite and arsenate to the same degree. It has been shown that Acr1p is a transcriptional regulator of ARR2 and ARR3 involved in arsenical resistance [11]. While ScARR1 efficiently complements the phenotype of arr1Δ null mutants in both species, a single copy of SdARR1 is not sufficient to complement the phenotype of the Scarr1Δ strain.

This correlates with the fact that the most important nucleotide sequence change within the ARR cluster is observed in the promoter sequence of ARR1. In S. cerevisiae the promoter contains an additional sequence with a characteristic motif, possibly recognized by transcriptional factors like Arr1p (Yap8p) or Yap1p [12,37]. Different mechanism of regulation of ARR1 expression could therefore form the basis of functional difference observed for this gene between S. cerevisiae and S. douglasii. The level of ScARR1 and SdARR1 expression was, however, similar and unaffected by metalloid presence (Table 5). In addition, Wysocki et al. [35] have shown by the chromatin immunoprecipitation method that Arr1p (Yap8p) was not bound to its own promoter either in the absence or in presence of As(III); in contrast, Acr1p showed constitutive binding to the ScARR3 promoter which contains also the TTAATAA sequence which is flanked, however, by different nucleotides. It has been recently shown that mutations in the regions flanking the Yap1 and Yap2 DNA binding-sites decreased expression of their target genes [40]. By analogy, the different sequences of nucleotide flanking the TTAATAA motif in the ScARR1 promoter might affect binding of putative regulators and expression of this gene. Also we cannot exclude the possibility that this sequence within the ScARR1 promoter is recognized and activated by other Yap-like proteins or yet unknown transcription factors.

Our data demonstrate that ARR2, encoding an arsenate reductase, is structurally and functionally conserved between S. cerevisiae and S. douglasii. Amplification or deletion of the S. douglasii ARR2 gene leads, respectively, to high level of resistance or hypersensitivity to arsenate [6,13]. Wild-type SdARR2 and ScARR2 fully complement the phenotype of both homologous and heterologous Sdarr2Δ and Scarr2Δ null mutants. As expected, similar results have also been obtained for the SdARR3 and ScARR3 genes encoding a putative membrane protein involved in arsenic transport. The data reported in this paper show that S. douglasii and S. cerevisiae ARR2 and ARR3 are completely interchangeable, and no functional differences with respect to their functioning can be detected between the two studied species.

Previous analysis of orthologous pairs of S. douglasii and S. cerevisiae nuclear genes, such as MRS1, SUV3 and CBP2 involved in mitochondrial pre-mRNA processing, has revealed remarkable sequence conservation with some significant functional differences [22,24,41–46]. Our results show that for case determinants of arsenic resistance, the function of both genes (ARR2 and ARR3) encoding the enzymatic components of the system remains conserved, while the function of the regulatory factor ARR1 is more divergent, apparently mainly due to changes in its promoter sequence.

As expected, the coding region sequences of the ARR genes display a high degree of conservation between the yeast sequences we analyzed, and their phylogeny follows the established pattern of the Saccharomyces evolutionary tree.

Analysis of the distribution of the ARR genes in various yeast species reveals traces of many chromosomal rearrangements and translocation events, including the duplication we found in the genome of S. douglasii. A key factor determining the evolutionary plasticity of this region is undoubtedly its subtelomeric chromosomal localization. Gene families located in subtelomeric regions are known to undergo rapid evolution involving frequent changes in number, order and orientation [30]. The results of our study indicate for the first time that the ARR gene cluster belongs to this class and can be an important factor shaping yeast speciation.

Differences found in comparisons of intergenic promoter regions, particularly the unique presence of an additional Yap1-binding motif in the S. cerevisiae ARR1 promoter, indicate that changes in mechanisms controlling gene expression have also been an important factor shaping yeast evolution. Structural and physiological differences in the regulation of ARR1 gene expression have probably arisen after the speciation event that separated the lineage of S. cerevisiae from that of S. douglasii and other species we analyzed. The contribution of the gene encoding the regulatory protein Arr1p to the evolution of arsenical resistance is further demonstrated by the observation that the genome of S. castellii contains genes encoding both enzymatic ARR proteins but seems to entirely lack the orthologue of ARR1.

Our results place the evolution of arsenical resistance within a wider context of both large-scale genomic events and specific changes in gene regulation mechanisms and suggest its significant role in speciation and adaptation to the environment in the genus Saccharomyces.

✩

Supplementary dataa ssociated with this article can be found, in the online version, at doi:10.1016/j.femsyr.2004.03.002.

Acknowledgements

E.M. was the recipient of doctoral fellowship from the CNRS within the framework of the Centre Franco-Polonais de Biotechnologie des Plantes. This work was supported by the State Committee for Scientific Research, Poland, grant HPRN CT 02 002 69 716/930-5105/SPUB/IGMi/03.