Maintenance of cation homoeostasis is a key process for any living organism. Specific mutations in Glc7, the essential catalytic subunit of yeast protein phosphatase 1, result in salt and alkaline pH sensitivity, suggesting a role for this protein in cation homoeostasis. We screened a collection of Glc7 regulatory subunit mutants for altered tolerance to diverse cations (sodium, lithium and calcium) and alkaline pH. Among 18 candidates, only deletion of REF2 (RNA end formation 2) yielded increased sensitivity to these conditions, as well as to diverse organic toxic cations. The Ref2F374A mutation, which renders it unable to bind Glc7, did not rescue the salt-related phenotypes of the ref2 strain, suggesting that Ref2 function in cation homoeostasis is mediated by Glc7. The ref2 deletion mutant displays a marked decrease in lithium efflux, which can be explained by the inability of these cells to fully induce the Na+-ATPase ENA1 gene. The effect of lack of Ref2 is additive to that of blockage of the calcineurin pathway and might disrupt multiple mechanisms controlling ENA1 expression. ref2 cells display a striking defect in vacuolar morphogenesis, which probably accounts for the increased calcium levels observed under standard growth conditions and the strong calcium sensitivity of this mutant. Remarkably, the evidence collected indicates that the role of Ref2 in cation homoeostasis may be unrelated to its previously identified function in the formation of mRNA via the APT (for associated with Pta1) complex.
An elevated intracellular concentration of sodium ions is harmful for most eukaryotic cells, probably because it interferes with the correct functioning of certain cellular targets . Yeasts, as well as other single-cell eukaryotic organisms, have developed mechanisms to maintain a safe cytosolic concentration of sodium even when the external concentration of sodium ions in their natural environments is high. To this end, these organisms employ three distinct strategies: (i) discrimination among different alkali metal cations at the level of uptake (favouring transport of potassium compared with sodium); (ii) triggering the efficient efflux of toxic cations; and (iii) sequestering the excess of cations in organelles, such as the vacuole.
The budding yeast Saccharomyces cerevisiae is a model organism very often used for cation tolerance studies. In this organism, the capacity to extrude sodium (or other toxic cations, such as lithium) represents a major mechanism to maintain low intracellular levels. This is achieved in two different and complementary ways. First, via the function of an H+/Na+-antiporter, encoded by the NHA1 gene, which is able to extrude sodium, lithium and even potassium cations by exchange with protons and therefore with biological relevance at rather acidic external pHs [2,3]. The second mechanism is based on a P-type ATPase pump encoded by the ENA system (see  for a review). In S. cerevisiae the ENA genes are disposed in tandem repeats with a variable number of copies (ENA1–ENA5) encoding identical or very similar proteins [5–7]. Deletion of the entire ENA cluster in S. cerevisiae results in a dramatic phenotype of sensitivity to sodium and lithium cations or alkaline pHs [5,6,8,9]. It is widely accepted that ENA1 is the functionally relevant component of the cluster on the basis that (i) cation sensitivity is largely restored by expression of ENA1 [6,10] and (ii) whereas under standard growth conditions expression from the ENA genes is almost undetectable, ENA1 expression dramatically increases in response to osmotic, saline or alkaline pH stress [6,8,9,11].
Regulation of ENA1 expression in response to saline or alkaline stress is under the control of different pathways . These include the osmoresponsive MAPK (mitogen-activated protein kinase) Hog1, the calcium-activated protein phosphatase calcineurin, the Rim101/Nrg1, the Snf1 and the Hal3/Ppz1 pathways, among others (see  and references therein). Activation of calcineurin upon saline or alkaline pH stress results in dephosphorylation and nuclear entry of the Crz1 (calcineurin-responsive zinc finger) transcription factor [13–15], which binds to specific sequences, known as CDREs (calcineurin-dependent response elements), in calcineurin-responsive promoters. Two CDREs have been defined in the ENA1 promoter (at −821/−813 and −727/−719 relative to the translation start site), with the downstream element being the most relevant for induction of the gene in cells exposed to saline or high pH stress [16–19].
Protein phosphatase 1 is a conserved serine/threonine protein phosphatase whose catalytic subunit is encoded in S. cerevisiae by a single, essential gene, named GLC7 [20,21]. The Glc7 protein is required for a myriad of cellular functions and specific regulation of these functions is achieved by interaction of the catalytic subunit with a variety of regulatory protein that influence Glc7 intracellular localization and/or substrate recognition. Many of these regulatory proteins have a conserved binding site, with a consensus sequence of (R/K)(V/I)X(F/W), that is necessary for the interaction with Glc7 [22,23]. Databases searches (at the Saccharomyces Genome Database, http://www.yeastgenome.org/) reveal at least 119 proteins that are annotated as Glc7-interacting proteins. However, most of these relationships are based on recent high-throughput analysis, whereas compelling regulatory evidence has been only collected for a limited subset of these proteins.
Among the cellular functions attributed to the Glc7 phosphatase, a role in ion homoeostasis was postulated some years ago as a result of the characterization of the glc7-109 allele , which carried a mutation at Arg-260 found to be responsible for the mutant phenotypes. glc7-109 mutants displayed, among other phenotypes, a marked sensitivity to sodium cations and alkaline pH, which was remedied by inclusion of potassium ions in the medium. Interestingly, Arg-260 lies in the vicinity of Phe-256, a residue in the Glc7 hydrophobic channel that makes contact with the aromatic residue of the (R/K)(V/I)X(F/W) binding-partner motif. Therefore we hypothesized that the defect in ion homoeostasis in the glc7-109 mutant could be due to a deficient interaction of the mutated protein with a regulatory subunit relevant for ion tolerance. To this end, we have analysed the tolerance to diverse cations of 18 known or putative Glc7 regulatory subunits and found that deletion of REF2 (RNA end formation 2) yields a rather severe and unique phenotype of sensitivity to sodium and lithium cations and alkaline pH. REF2 was initially described as a gene encoding a protein required for mRNA 3′-end maturation prior to the final polyadenylation step . Evidence for interaction between Ref2 and Glc7 was obtained by both two-hybrid and affinity purification techniques [26–28]. Subsequently, it has been found that Ref2 is a component of the APT (for associated with Pta1) subcomplex of the large CPF (cleavage and polyadenylation factor)-containing complex (holo-CPF), which is required for formation of mRNA and snoRNA (small nucleolar RNA) 3′-ends and also contains Glc7 . It has been shown that Ref2 is required for the permanence of Glc7 within the complex . The Ref2 protein contains a R368ISSIKFLD376 sequence that resembles the Glc7-binding motif (the bold residues are a perfect match with the consensus sequence) and recent evidence indicates that mutation of Ref2 Phe-374 to an alanine residue disrupts the interaction between the protein and Glc7, thus preventing incorporation of Glc7 to the APT subcomplex 
In the present study, we highlight the cation-sensitivity phenotype of the ref2 mutant and investigate the molecular basis of this defect. Our results indicate that mutation of REF2 results in a deficient efflux of toxic cations that can, at least in part, be attributed to impaired expression of the ENA1 ATPase gene. Remarkably, this role in salt tolerance seems unrelated to the function of Ref2 in the APT complex.
Growth of Escherichia coli and yeast strains
E. coli DH5α cells were used as plasmid DNA hosts and were grown at 37 °C in LB (Luria–Bertani) medium, supplemented with 50 μg/ml ampicillin when plasmid selection was required. Yeast cells were cultivated at 28 °C in YPD medium [1% (w/v) yeast extract/2% (w/v) peptone/2% (w/v) glucose], or in synthetic complete drop-out medium when carrying plasmids. Restriction reactions, DNA ligations and other standard recombinant DNA techniques were performed as described in . Yeast cells were transformed using the lithium acetate method . Sensitivity of yeast cells to lithium, sodium or calcium chloride and alkaline pH was evaluated by growth on YPD or YP-based [1% (w/v) yeast extract/2% (w/v) peptone] plates (drop tests) or in liquid cultures as described previously [8,33]. Growth under limiting potassium concentration was carried out using an YNB (yeast nitrogen base)-based medium (Translucent K-free medium; Formedium) formulated so that potassium content in the final medium is negligible (approx. 15 μM).
Gene disruption and plasmid construction
Single KanMX deletion mutants in the BY4741 background (MATa his3Δ1 leu2Δ met15Δ ura3Δ) were generated in the context of the Saccharomyces Genome Deletion Project . Replacement of the REF2 coding region by the nat1 marker from Streptomyces noursei was accomplished as follows: the 1.40 kbp DNA fragment containing the nat1 gene, flanked by genomic sequences corresponding to −40/−1 and +1603/+1643, relative to the REF2 ATG codon, was amplified from the plasmid pAG25  with the oligonucleotides OJFD021 and OJFD022 (see Supplementary Table S1 available at http://www.BiochemJ.org/bj/426/bj4260355add.htm) and transformed in the wild-type strains BY4741 and its cnb1::KanMX derivative to yield strains YJFD1 and YJFD7 respectively. The ref2::nat1 mutation was also introduced into the wild-type DBY746 background (MATa, ura3-52 leu2-3,112 his3Δ1 trp1Δ239) to yield strain YJFD5, which was used for measurement of intracellular calcium levels. Positive clones were selected in the presence of 100 μg/ml nourseothricin (Werner BioAgents) and the presence of the deletions was verified by PCR. Strains YJFD17 and YJFD18 were made by co-transformation of wild-type BY4741 and its ref2::KanMX derivative respectively with plasmid pJFD10 (containing the glc7-109 allele, see below) and a disruption cassette containing the nat1 gene flanked by genomic sequences corresponding to −41/−1 and +1465/+1509, relative to the GLC7 ATG codon (amplified from plasmid pAG25 with the oligonucleotides OJFD81 and OJFD82). Diverse tests were carried out to ensure that the only Glc7-encoding gene copy corresponded to the plasmid-borne glc7-109 allele. Strains KT1112 (GLC7), KT1935 (glc7-109) and KT2210 (glc7-F256A) were a gift from Professor K. Tatchell (Department of Biochemistry and Molecular Biology, Louisiana State University Health Sciences Center, Shreveport, LA, U.S.A.) and have been reported previously . Strains SB6 (YTH1) and SB7 (yth1-1) , YWK168 (SWD2) and YWK172 (swd2-3) , as well as W303 (SSU72) and YWK186 (ssu72-2) , were a gift from Dr B. Ditchl (Institute of Molecular Biology, University of Zurich, Zurich, Switzerland). Strains PTA1 (PTA1) and pta1 (pta1Δ1–75)  were a gift from Professor C. Moore (Department of Microbiology, Tufts University, Boston, MA, U.S.A.).
To clone the glc7-109 allele in the YCplac33 plasmid (centromeric with a URA3 marker) the Glc7-encoding ORF (open reading frame), flanked by 2068 bp upstream and 148 bp downstream, was amplified by PCR using genomic DNA from strain KT1935 and oligonucleotides Glc7-BamHI.Up and Glc7-PstI.Do, which contain artificial BamHI and PstI restriction sites respectively (see Supplementary Table S1). The amplified DNA fragment was digested by BamHI and PstI and cloned in the same restriction sites of YCplac33 to yield plasmid pJFD10 (YCplac33-glc7-109). Plasmid YEp195-GLC7 was constructed using the same strategy except that genomic wild-type DNA was used for the PCR. These plasmids were then isolated and sequenced.
For low-copy, centromeric expression of REF2, a 2.28 kbp fragment containing the REF2 ORF, flanked by 500 bp and 200 bp from its promoter and terminator regions respectively, was amplified using a specific pair of oligonucleotides containing artificially added BamHI (OJFD027) and SacI (OJFD026) restriction sites (see Supplementary Table S1). The resulting DNA fragment was digested with BamHI and SacI and cloned in the same restriction sites of the pRS415 plasmid (centromeric with a LEU2 marker) to yield pRS415-REF2 (pJFD1).
To generate the F374A allele of REF2 (pJFD2), the pRS415-REF2 construct was used as a PCR template to change the TTT codon, encoding a phenylalanine residue at position 374, to a GCT codon, encoding an alanine residue. Two overlapping DNA fragments (a and b) were generated. The 1.75 bp fragment (a), spanning from −501 to +1140 (relative to the initiating methionine residue), was obtained using oligonucleotides OJFD029 and M13-20 reverse. The 0.7 bp length fragment (b) comprises from +1101 to +1785 and was amplified with oligonucleotides OJFD063 and M13-20 (see Supplementary Table S1). To allow in vivo DNA-recombination, these two fragments were co-transformed in yeast cells together with the linearized pRS415 plasmid digested with BamHI and SacI. Cells carrying the pRS415F374A construct were selected in a medium lacking leucine and the plasmid was recovered and sequenced to ensure the absence of unexpected mutations.
LacZ reporter plasmids used in the present study have been reported previously: ENA1 reporters were pKC201, containing the entire ENA1 promoter  and pMRK212 and pMRK213, containing the ARR1 and ARR2 regions of this same promoter . pMRK-based constructs were derived from plasmid pSLFΔ-178K , which contains a CYC1 minimal promoter and displays virtually no reporter activity in the absence of a transcriptionally active DNA insert. Plasmid pAMS366  contains four copies of a CDRE from the FKS2 promoter.
Lithium content and lithium efflux measurements
Lithium content and efflux were measured as reported previously, with minor modifications . For determination of the intracellular lithium cation concentration, cells were grown overnight in liquid YPD medium. When cultures reached an attenuance of 0.5–0.6, cells were collected (centrifugation at 1620 g for 5 min) and diluted (to a D600 of 0.2–0.3), in fresh YPD medium containing the appropriate concentration of LiCl. Cells were loaded with lithium for 8 h and then samples were filtered, washed with 20 mM MgCl2, and treated with acid overnight (0.2 M HCl and 10 mM MgCl2). Lithium was analysed by atomic absorption spectrophotometry as described previously . Lithium efflux was studied in cells grown overnight as described above. Cells were suspended in fresh medium and loaded with 75 mM LiCl for 2 h. Yeast cells were separated from the culture medium by centrifugation (1620 g for 5 min), washed, and suspended in fresh lithium-free medium. At various time intervals, samples were taken and treated as above for lithium determination.
The different strains were transformed with the reporter plasmids pKC201, pMRK212, pMRK213 or pAMS366. Cultures were grown up to saturation in the appropriate drop-out medium, then they were inoculated on YPD (at pH 5.5) until the D660 reached 0.6–0.7 and collected by centrifugation (5 min at 750 g). Then cells were resuspended in fresh YPD (pH 5.5) containing the indicated concentrations of LiCl, NaCl or CaCl2, or in medium adjusted to pH 8.2 in the presence of 50 mM Taps buffer, and incubated for a further 60 min, unless otherwise stated, at the specified temperatures. β-Galactosidase activity was measured as described previously  and results are expressed as Miller Units [44a].
Intracellular calcium monitoring
The determination of intracellular calcium was based on the use of pEV11/AEQ  which carries the apoaequorin gene under the control of the ADH1 promoter. Evaluation of cytoplasmic calcium was carried out essentially as described previously [18,44]. Briefly, DBY746 wild-type and ref2-mutant cells (YJFD5) were transformed with plasmid pEVP11/AEQ and grown to a D660 of 1.0. Aliquots of 30 μl were incubated in luminometer tubes for 20 min at room temperature (22 °C), a one-tenth volume of 592 μM coelenterazine (Sigma–Aldrich) in methanol was then added to each sample and the mixture was incubated further at room temperature for 20 min. Basal luminescence was recorded every 0.1 s for 5 s using a Berthold LB9507 luminometer. The RLUs (relative luminescence units) were normalized to the number of cells in each sample.
Vacuolar staining and visualization
Vacuole morphology was assessed basically as described previously . Yeast cultures (5 ml at a D660 of 0.8–1.0) were harvested (1620 g for 5 min), washed with YPD, and resuspended in 0.1 ml of fresh YPD. The lypophilic fluorescent dye FM4-64 (Invitrogen) was added at a final concentration of 20 μM and cells were incubated for 15 min at 30 °C. The cells were then washed, resuspended in 3 ml of YPD, and incubated for 15–45 min to allow the internalization, by endocytosis, and accumulation of the dye within the vacuole. Cells were visualized with a FITC filter using a Nikon Eclipse E800 fluorescence microscope (at 400×). Digital images were captured with an ORCA-ER 4742-80 camera (Hamamatsu) using the Wasabi software.
Lack of Ref2 causes defects in cation tolerance
The observation that Glc7 point mutations lying in the vicinity of residues relevant for binding diverse regulatory subunits may cause alterations in ion homoeostasis  prompted us to examine tolerance to NaCl, LiCl, alkaline pH and CaCl2 in 18 strains carrying a deletion of functionally well-documented regulatory subunits of Glc7. Figure 1 shows a selected panel of conditions. As shown, mutation of most regulatory subunits did not result in significantly altered tolerance to the indicated treatments, with the exception of the ref2 and shp1 mutants. However, deletion of shp1 caused a rather severe growth defect even in the absence of stress, as previously described , making it difficult to assess the actual effect of the mutation under the conditions tested. In contrast, lack of Ref2, which only caused a slight growth defect under standard culture conditions, prevented growth in the presence of lithium, sodium and calcium cations, as well as under alkaline conditions. ref2 cells were also markedly sensitive to caesium ions (results not shown). The intensity of the phenotype for sodium and lithium stress was comparable with that of a trk1 trk2 strain, which has defective potassium uptake (results not shown). The phenotype was not caused by the inability to adapt to hyperosmotic stress, as (i) growth was not affected by inclusion of 1 M sorbitol or KCl in the medium and (ii) an obvious growth defect was already observed at a lithium concentration as low as 30 mM, which does not represent an osmotic stress condition (results not shown). Increased sensitivity to sodium and lithium cations is often observed in strains with defective potassium uptake (i.e. in trk1 mutants) [46a]. This prompted us to test the requirements for potassium in the ref2 mutant, by using a recently developed YNB-based potassium-free medium. As shown in Figure 2, ref2 cells display a slow-growth phenotype at limiting potassium concentrations (0.5–1.5 mM) that was alleviated by increasing the amount of potassium in the medium. This suggests that ref2 cells might have a decreased high-affinity potassium uptake, although direct measurement of rubidium influx (an analogue commonly used to assess for potassium uptake capacity [46b]) failed to reveal a significant decrease in rubidium ion uptake (results not shown). In any case, the ref2 strain showed increased sensitivity to several organic toxic cations, such as hygromycin B, spermine and TMA (tetramethylammonium) (Figure 3A), a phenotype which is commonly observed in strains with altered cation homoeostasis [46c].
Effect of the mutation of several Glc7-interacting proteins on cation-related phenotypes
Mutation of REF2 increases potassium requirements for growth
Mutation of Phe-374 within the Glc7-binding consensus sequence abolishes the function of Ref2 in cation tolerance
Ref2 contains a RISSIKFLD sequence (residues 368–376) that closely matches the (R/K)(V/I)X(F/W) Glc7-binding consensus sequence. We considered it necessary to test whether or not the salt-related phenotypes, derived from the absence of Ref2, could be explained by its role as a putative Glc7-regulatory subunit. To this end, Phe-374 was replaced with an alanine residue and the mutated version of the gene (including its own promoter) was cloned into a centromeric plasmid. As shown in Figure 3(A), introduction of the wild-type version of REF2 (pJFD1) in a ref2 mutant was able to increase tolerance to sodium, lithium and alkaline pH, as well as to diverse toxic organic cations. In contrast, mutation of Phe-374 (pJFD2) resulted in a complete inability to do so. It must be noted that this mutation does not affect Ref2 cellular levels, but it is known to disrupt the interaction with Glc7 . Expression of the mutated form of Ref2 from a high-copy number plasmid, which should lead to overexpression of the protein, was also unable to rescue the ref2 defects (results not shown). These evidences suggest that blocking the interaction between Ref2 and Glc7 results in loss of the Ref2 function in salt tolerance. A similar effect was observed when the growth deficiency at limiting external potassium concentrations of the ref2 mutant was tested (Figure 3B). In this case, expression of the F374A-mutated version of REF2 (pJFD2) even aggravated the growth defect of the strain lacking the chromosomal copy of the gene. We then considered the possibility that the cation-related defects of the ref2 strain could be attenuated or even suppressed by a surplus of Glc7. However, expression of Glc7 on a multicopy plasmid did not increase the tolerance of a ref2 strain to LiCl or NaCl (see Supplementary Figure S1A available at http://www.BiochemJ.org/bj/426/bj4260355add.htm). In contrast, the same plasmid was able to abolish the salt-sensitive phenotype of specific Glc7 mutants (Supplementary Figure S1B), indicating proper expression of Glc7.
Increased sensitivity to sodium and lithium cations can be the result of increased uptake, decreased efflux, or inability to sequester these toxic cations into intracellular compartments, such as the vacuole. To get an insight into a possible cause for sensitivity in the ref2 mutant, the intracellular concentration of lithium in cells grown in the presence of 25 or 50 mM LiCl was tested. As shown in Figure 4(A), ref2 mutants accumulate 60–70% more lithium than the wild-type strain. This suggested that lack of Ref2 might favour lithium entry or make detoxification of the cation more difficult by interfering with the efflux systems. To this end, we tested the ability of ref2 cells to extrude lithium. As shown in Figure 4(B), lithium efflux was markedly impaired in cells lacking Ref2, suggesting that the absence of this Glc7 regulatory subunit interferes with the normal efflux mechanisms.
Effects of the ref2 mutation on intracellular lithium accumulation and efflux
Mutation of Ref2 affects expression of the ENA1 ATPase gene under saline and alkaline pH stresses
The Na+-ATPase Ena1 is a major determinant of sodium and lithium tolerance. The expression of the gene is dramatically increased by exposure to high concentrations of sodium or lithium cations or by alkalinization of the medium, and many mutations impairing ENA1 induction upon stress have been shown to lead to hypersensitivity to these cations. Therefore we considered it necessary to monitor the expression of ENA1 in ref2 cells under cation stress conditions. To this end, cells were transformed with plasmid pKC201, which carries the entire ENA1 promoter fused to the lacZ reporter gene. As shown in Figure 5(A), expression driven from the ENA1 promoter in cells exposed to sodium or lithium cations, as well as to alkaline pH, is drastically reduced in ref2 cells. Therefore the increased sensitivity of the ref2 mutant to these stresses could be attributable, at least in part, to a defect in ENA1 induction.
Effects of the ref2 mutation in wild-type and cnb1 strains on the ENA1 promoter activity
The expression of ENA1 under saline or alkaline stress is controlled by a variety of signalling pathways, including activation of calcineurin, which is triggered by a burst of cytosolic calcium. As observed in Figure 5(A), the effect of the ref2 mutation on ENA1 expression can be even more severe than that caused by mutation of CNB1, encoding the regulatory subunit of calcineurin. Furthermore, the cnb1 mutation seems additive to ref2 both with respect to ENA1 expression (Figure 5A) and to lithium, sodium and alkaline pH tolerances (Figure 5B). Therefore ref2 phenotypes cannot be exclusively explained by a hypothetical impairment of calcineurin signalling under stress. Most regulatory inputs acting on the ENA1 promoter target two relatively small regions named ARR1 and ARR2 (see  for a review). To gain insight into the effect of the ref2 mutation, cells were transformed with plasmids pMRK212 and pMRK213. The former contains a region that mostly integrates calcium/calcineurin inputs, whereas the latter is regulated by calcineurin-independent stimuli. As shown in Figure 5(A), deletion of REF2 decreases expression driven from both promoter regions, suggesting that the mutation has a complex and diverse effect. Analysis of the expression from these constructs further supports an additive effect of the cnb1 and ref2 mutations on cation homoeostasis.
ref2 cells display a hyperactivated calcium/calcineurin pathway in the absence of stress and show altered vacuole morphology
A striking phenotype that can be observed in Figure 1 is that lack of Ref2 results in a substantial sensitivity to calcium cations. It has been reported that hyperactivation of calcineurin results in a calcium-sensitive phenotype, whereas deletion of calcineurin-encoding genes yields calcium hypertolerance [46d,46e]. We therefore hypothesized that the calcium-sensitive phenotype of ref2 mutants could be attributed to an unusually high basal calcineurin activity (i.e. in the absence of saline or alkaline pH stress). To test this possibility, we evaluated calcium tolerance in cells lacking both Ref2 and Cnb1, and compared them with the single ref2 and cnb1 mutants. As shown in Figure 6(A), the calcium sensitivity conferred by the ref2 mutation is largely abolished in the absence of Cnb1. This suggests that the ref2 growth defect in the presence of calcium is caused by hyperactivation of calcineurin and leads to the possibility that calcium levels might be higher than normal in ref2 cells. This was directly tested by introducing the calcium-reporter plasmid pEV11/AEQ in wild-type and ref2-mutant cells. Our measurements indicated that the luminescence observed in ref2 cells was 141.9±3.7 RLU per 106 cells, whereas this parameter was 64.8±3.5 in the wild-type strain, thus confirming the existence of higher free cytosolic calcium levels in the mutant. On the basis of this result, one would expect that expression from a specific calcium/calcineurin-sensitive promoter would be increased in ref2 cells grown under basal (non-stressing) conditions. To test this conjecture, wild-type and ref2 strains (as well as their cnb1 derivatives) were transformed with plasmid pAMS366, which carries a tandem-repeat of four copies of the CDRE from the promoter of the FKS2 gene. As shown in Figure 6(B), under basal growth conditions, expression from this synthetic promoter is 4–5-fold higher in ref2 cells than in the wild-type strain. As expected, this effect is abolished in the absence of calcineurin activity (ref2 cnb1 strain). Remarkably, whereas in the wild-type strain addition of calcium ions to the medium triggers a dramatic, calcineurin-mediated increase in expression, this effect is much less prominent in cells lacking Ref2. This could be explained if the ref2 mutation somehow interferes with stress-triggered effects mediated by activation of calcineurin. Increased sensitivity to alkaline pH and calcium is a characteristic phenotype of yeast strains with deficient vacuolar function [46f]. Therefore the possibility that ref2 mutants may display some alteration in this subcellular compartment was considered. As shown in Figure 6(C), incubation of wild-type and ref2 cells with the lypophilic fluorescent dye FM4-64, which stains the vacuole membrane, reveals that vacuolar structure is dramatically altered in the ref2 mutant, with a punctuated fluorescent pattern and lack of discernible vacuolar structure. Because of the similarity between the saline phenotypes of the ref2 deletion and the glc7-109 allele, we considered it interesting to evaluate vacuolar morphology in the latter. As shown in Figure 6(C), FM4-64 staining of strain YJFD17, which carries a centromeric plasmid-borne glc7-109 allele as the only source of GLC7 function, reveals that this strain has vacuoles with essentially wild-type morphology (in a similar manner to the original cation-sensitive strain KT1935; results not shown). Interestingly, combination of the ref2 deletion and the glc7-109 mutation results in cells with a ref2-type phenotype, i.e. altered vacuoles. The disruption of REF2 in the KT1935 background was repeatedly attempted without success. This is probably due to the close genetic relationship of this strain with the W303 genetic background (K. Tatchell, personal communication), in which the REF2 deletion was reported to be lethal .
Lack of REF2 results in altered calcium homoeostasis and vacuolar morphology
Mutants in components of the APT complex do not share the cation-related phenotypes of the ref2 strain
Given the Ref2 function described previously, as a component of the APT complex, it was reasonable to consider whether the striking cation-related phenotypes of the ref2 strain could be attributed to the role of the protein in this complex. We reasoned that, if this was the case, mutations in other components of the complex would yield phenotypes reminiscent of those of the ref2 strain. As many members of the APT complex are essential (including Pta1), we had to resort in most cases to temperature-sensitive mutants cultured at sublethal temperatures. Figure 7(A) shows that deletion of SYC1, encoding a non-essential member of the complex, does not alter cell growth under conditions that clearly affect proliferation of ref2 mutants. The yth1-1 mutant displayed a slight sensitivity to LiCl, but was indistinguishable from the wild-type in its sensitivity to NaCl and alkaline pH. The swd2-3 and ssu72-2 strains did not show any sensitivity to LiCl or NaCl and only a marginal sensitivity to high pH. None of these strains displayed sensitivity to high calcium levels in the medium. Similarly, a temperature-sensitive strain lacking the N-terminal region of Pta1 (strain pta1Δ1–75) exhibited a near wild-type tolerance to cations or alkaline pH even at 37 °C. None of the tested APT complex mutants were sensitive to low glucose or non-fermentable carbon sources (Supplementary Figure S2 available at http://www.BiochemJ.org/bj/426/bj4260355add.htm). We also observed sensitivity to formamide in the yth1-1 strain when grown at 30 °C as reported previously . Interestingly, we observed a similar phenotype in the swd2-3 mutant at 30 °C and the ref2 or pta1Δ1–75 strains at 37 °C, clearly showing that under these conditions the function of the APT complex is compromised. The almost complete lack of overlap between specific ref2 phenotypes and those of the other mutants in components of the APT complex is exemplified in the phenotypic array shown in Figure 7(B).
Comparison of ref2 phenotypes with those of diverse APT-related mutants
We also examined the basal expression level of ENA1 in these strains and the ability of the ENA1 promoter to be activated by salt stress. As shown in Figure 8, changes in expression levels induced by exposure to LiCl or NaCl were smaller in the swd2-3 strain. However, the expression observed in the syc1, ssu72-2 or pta1-1–75 derivatives was virtually identical with that of their respective wild-type strains. Taken together, these results demonstrate that the mutation of diverse components of the APT complex does not mimic the cation-related phenotypes of the ref2 strain.
Expression of ENA1 under saline stress in diverse APT-related mutants
In the present study we have screened for defects in cation tolerance of 18 strains lacking specific regulatory subunits of the yeast type 1 protein phosphatase Glc7. Among them, only the ref2 mutant presented phenotypes which were undoubtedly associated with altered cation homoeostasis, such as decreased tolerance to lithium and sodium, sensitivity to alkaline pH and decreased growth at limiting external potassium concentrations. We also observed that ref2 cells were sensitive to diverse toxic organic cations that interfere with very different cellular functions. It is generally accepted that these compounds are ‘driven’ to enter the cell by the electrochemical gradient generated by the Pma1 plasma membrane proton ATPase and it has been observed that many mutations that affect sodium and/or potassium homoeostasis result in an altered electrochemical gradient (e.g. trk1, ppz1, hal4,5) [47a]. Therefore the observation that ref2 cells are sensitive to hygromycin B, spermine and TMA is consistent with the notion that Ref2 is a relevant component of cation homoeostasis. Our results suggest that the decreased tolerance to alkaline cations in the ref2 mutant could be due, at least in part, to the inability to fully induce expression of the ENA1 Na+-ATPase gene, which is a major determinant in saline tolerance. As lack of Ref2 blocked the response of ENA1 to diverse stresses, which are mediated through a variety of signalling pathways [12,48], the effect of Ref2 on ENA1 expression is possibly pleiotropic and may involve multiple targets.
Ref2 and Glc7 are components of the APT complex and they are required for efficient 3′ processing of mRNAs and transcription termination of certain snoRNA genes [25,29,30,47]. Therefore it would be reasonable to speculate that the cation-related phenotypes of the ref2 mutant could be caused by malfunction of this complex, which is itself a component of the holo-CPF. However, our results show that deletion or temperature-sensitive mutations of diverse components of the APT complex, tested under conditions that impair their function in the complex [37–39], do not consistently mimic the growth defects or impaired ENA1 expression of the ref2 strain (Figures 7B and 8). The same situation is observed for the conditional mutation of YTH1, whose product participates in polyadenylation factor I, a different subcomplex of the CPF . Furthermore, we also observe that, whereas the ref2 strain fails to grow on non-fermentable carbon sources and grows poorly on low-glucose medium, none of the APT-related mutations tested display such phenotypes (Figure 7B and Supplementary Figure S2), thus providing an additional example of disparate phenotypes. On the other hand, Ref2 function has been implicated in COMPASS/Set1 complex function via the Swd2 protein . The function of Swd2 in this complex has been shown to be independent from its role in the APT complex . The COMPASS/Set1 complex is required for histone H3 methylation at Lys-4, which is a fingerprint of actively transcribed genes . Therefore defective COMPASS function could be invoked to explain the effect of the ref2 mutation on ENA1 expression. However, this seems to be unlikely as (i) a survey of the literature [49a] (and our own results, not shown) indicate that mutants in the non-essential components of the complex do not display salt-related phenotypes nor are they sensitive to organic toxic cations; (ii) no evidence for a role of Glc7 on COMPASS function has been reported so far; and (iii) deletion of REF2 does not perturb histone H3 methylation at Lys-4 .
Our observation that ref2 cells are calcium-sensitive is consistent with the identification of REF2 in a screen for mutations sharing multiple pmr1 phenotypes . PMR1 encodes an ATPase that sequesters calcium and manganese ions in the Golgi/secretory pathway, which is pivotal to maintaining proper cytosolic concentrations of these ions. In addition, in a similar manner to vma and pmr1 mutants , the ref2 mutant displays strong sensitivity to the cell-wall perturbing agent Calcofluor White (results not shown). Remarkably, we observe that in ref2 cells grown in the absence of any source of stress there is an increased cytosolic calcium level and higher than normal expression from a synthetic calcium/calcineurin-regulated promoter (Figure 6B). This suggests that basal calcineurin activity is abnormally high in ref2 mutants. Further activation of calcineurin, by exposure to high concentrations of extracellular calcium, would therefore explain the hypersensitive phenotype of ref2 cells to this cation, as it is known that excessive calcineurin activity is detrimental to the cell [52,53]. The observation that deletion of CNB1, which blocks calcineurin signalling, relieves calcium toxicity provides further support to this hypothesis. We also show in the present study that ref2 cells displayed altered vacuolar morphology, which is reminiscent of the class B/C vps mutants [45,54,55]. As the vacuole is the most important calcium store in yeast , it is reasonable to assume that the calcium-related phenotypes displayed by ref2 cells could be derived from their incapacity for normal vacuole assembly. It must be noted, however, that whereas lack of Ref2 increases basal calcineurin activity, the mutation also interferes with stress signals that transiently activate the phosphatase, as deduced from the lower short-term response of the calcineurin-regulatable promoter (Figure 6B).
We show in the present study that a version of Ref2 that cannot interact with Glc7  is unable to rescue the saline phenotypes of the ref2 strain. Therefore it is reasonable to consider that the perturbed cation homoeostasis observed in a ref2 strain is due to a deregulation of Glc7 activity. Altered cation homoeostasis has been described in the past for specific mutations of Glc7, such as the glc7-109 allele, which carries an R260A mutation near the hydrophobic channel that is likely to be responsible for the interaction with the binding motif present in Ref2 . Remarkably, the glc7-109 mutant displays a subset of phenotypes that closely resembles those reported in the present study for the ref2 disruption, such as enhanced sensitivity to sodium, lithium, alkaline pH and organic toxic cations . We have also observed that deletion of REF2 in a strain carrying the glc7-109 allele (YJFD18) does not result in enhanced sensitivity to NaCl or alkaline pH (results not shown). In addition, the saline-stress-related phenotypes of the glc7-109 allele are additive with those caused by disruption of calcineurin activity, a circumstance also observed for the ref2 strain (Figure 5B). All this evidence suggests that the cation-related phenotypes of the glc7-109 allele could be caused, at least in part, by its incapacity to interact with Ref2. It must be stressed, however, that the ref2 and glc7-109 phenotypes are not identical. For instance, the glc7-109 strain hyperaccumulates glycogen , whereas the ref2 mutant does not (results not shown). More importantly, the glc7-109 strain does not display altered calcium levels and increased calcium sensitivity , or altered vacuolar morphology (Figure 6C). Therefore it could be that the R260A mutation in glc7-109 may affect the interaction of the phosphatase catalytic subunit with diverse regulatory subunits, one of these being Ref2. The relevance of the role of Ref2 is highlighted by the observation that the salt-related defects of ref2 cells cannot be overridden by simply overexpressing Glc7.
In conclusion, the results of the present study demonstrate that the role of Ref2 on cation tolerance is not attributable to its function in regulating Glc7 in the CPF complex but, instead, it suggests that Ref2 must direct Glc7 to alternative target(s). This implies that Ref2 may have multiple cellular functions. Although remarkable, this is not fully surprising; a similar scenario has been reported for the Gac1 regulatory subunit, which functions as a molecular scaffold to tether Glc7 to Gsy2 (encoding glycogen synthase), thus controlling glycogen accumulation  but also binds the transcription factor Hsf1, and thus controls the transcription of certain heat-shock responsive genes, such as CUP1 . Therefore Ref2 would represent a novel example of multifunctional Glc7 regulatory subunit.
Jofre Ferrer-Dalmau performed the majority of the experiments. Asier González and Maria Platara performed the lacZ-measurements. Lina Barreto measured the growth of diverse strains at limiting external potassium. Clara Navarrete, José Martínez and José Ramos measured lithium content and efflux. Joaquín Ariño and Antonio Casamayor designed the experiments and wrote the paper.
We thank H. Sychrovà and G. Wiesenberger for help in developing and testing the YNB-based, K-free medium. We also thank C. Moore, B. Dichtl, K. Tatchell and M. Marquina for provision of strains and plasmids. The excellent technical assistance of our colleagues Anna Vilalta and Montserrat Robledo at the Universitat Autònoma de Barcelona is acknowledged.
This work was supported by the Ministry of Science and Innovation, Spain [grant numbers BFU2008-04188-C03-01, GEN2006-27748-C2-1-E/SYS (to J. A.), BFU2007-60342 (to A. C.), BFU2008-04188-C03-03, GEN2006-27748-C2-2-E/SYS (to J. R.)]. J. A. is the recipient of a Generalitat de Catalunya “Ajut de Suport a les Activitats dels Grups de Recerca” award [grant number 2009SGR-1091].