A molecular genetic approach was undertaken in Saccharomyces cerevisiae to examine the functions of ARL1, encoding a G protein of the Ras superfamily. We show here that ARL1 is an important component of the control of intracellular K+. The arl1 mutant was sensitive to toxic cations, including hygromycin B and other aminoglycoside antibiotics, tetramethylammonium ions, methylammonium ions and protons. The hygromycin-B-sensitive phenotype was suppressed by the inclusion of K+ and complemented by wild-type ARL1 and an allele of ARL1 predicted to be unbound to nucleotide in vivo. The arl1 mutant strain internalized ∼25% more [14C]-methylammonium ion than did the wild type, consistent with hyperpolarization of the plasma membrane. The arl1 strain took up 30-40% less 86Rb+ than did the wild type, showing an inability to regulate K+ import properly, contributing to membrane hyperpolarity. By contrast, K+ and H+ efflux were undisturbed. The loss of ARL1 had no effect on the steady-state level or the localization of a tagged version of Trk1p. High copy suppressors of the hygromycin-B phenotype included SAP155, encoding a protein that interacts with the cell cycle regulator Sit4p, and HAL4 and HAL5, encoding Ser/Thr kinases that regulate the K+-influx mediators Trk1p and Trk2p. These results are consistent with a model in which ARL1, via regulation of HAL4/HAL5, governs K+ homeostasis in cells.
Generation and maintenance of membrane potential are critical for all cells. This property allows for uptake of nutrients, elimination of wastes, generation of cellular energy, and cellular communication in multicellular organisms. This process is highly regulated, allowing cells to respond quickly and effectively to changes in the environment that affect membrane potential. In Saccharomyces cerevisiae, membrane potential (negative inside) is determined primarily by efflux of protons via the H+-ATPase, encoded by the essential PMA1 gene, and the influx of K+ via the Trk proteins, encoded by TRK1 and TRK2 (reviewed in Gaber, 1992; Serrano, 1996). However, a full understanding of the regulation of this important cellular property is far from complete. Here, we present evidence that the guanine nucleotide-binding protein Arl1p, a member of the Arf-like family of proteins, is an important component in the regulation of K+ influx, thus affecting membrane potential.
ARL1 is highly conserved over eukaryotic evolution. Yeast, Arabidopsis thaliana, Caenorhabditis elegans, Drosophila melanogaster and mammalian Arl1 proteins are ∼60% identical to each other. The D. melanogaster ARL1 homolog, arflike, is essential (Tamkun et al., 1991), although the yeast gene is not (Lee et al., 1997; Rosenwald et al., 2002). We have previously shown that yeast ARL1 encodes a regulator of membrane traffic (Rosenwald et al., 2002), which has been independently confirmed by others (Bonangelino et al., 2002; Jochum et al., 2002). A role for Arl1 in membrane traffic in mammalian cells has been documented (Icard-Liepkalns et al., 1997; Lowe et al., 1996; Lu et al., 2001; Van Valkenburgh et al., 2001). Furthermore, Arl1 interacts with golgin proteins in both yeast (Panic et al., 2003; Setty et al., 2003) and mammalian cells (Lu and Hong, 2003). However, in this study, we demonstrate an unprecedented role for this guanine nucleotide-binding protein in K+ homeostasis.
Strains lacking ARL1 were found to be sensitive to several different cationic translation inhibitors, including hygromycin B. The arl1 strains also exhibit retarded growth in medium containing methylammonium chloride (MA) or tetramethylammonium chloride (TMA). These phenotypes were suppressed by K+. In addition, arl1 strains were sensitive to low pH (pH 3) medium. These results correlated with increased uptake of [14C]-methylammonium ion and decreased 86Rb+ uptake by the arl1 strain. HAL4 and HAL5 were isolated as high-copy-number suppressors of the hygromycin-B-sensitive phenotype. These genes encode kinases that function upstream of the K+ transporters Trk1 and Trk2 (Goossens et al., 2000; Mulet et al., 1999). Loss of ARL1 did not affect steady-state levels of Trk1 or localization of a tagged version of Trk1, suggesting that, despite Arl1p's documented role as a regulator of membrane traffic, Arl1p acts in a novel manner to control K+ influx.
Materials and Methods
Yeast strains and growth conditions
The parental strain was PSY316 (MATα ade2-101 his3-Δ200 leu2-3,112 lys2-801 ura3-52 SSD1) (Stearns et al., 1990). Replacement of the ARL1 open reading frame in PSY316 by the HIS3 gene (arl1Δ::HIS3) using a polymerase chain reaction (PCR) disruption method (Baudin et al., 1993) resulted in strain MA03 (Rosenwald et al., 2002). Strains AM300 and AM310 were created from PSY316 and MA03, respectively, by insertion of a fragment containing the 3′ end of TRK1 without a stop codon fused to 13 Myc epitopes (Longtine et al., 1998). This fragment was amplified from a template on plasmid pR341 using oligonucleotides GT105* and GT106* (Table 1). Correct insertion of the fragment at the TRK1 locus was confirmed by PCR using primers GT105, which binds the 3′ end of the TRK1 open-reading frame, and GT128, which binds to a region 3′ of the TRK1 open reading frame beyond the insertion. If the insertion is correct, a band of 2.7 kb is amplified. In addition, GT51 (which binds to the KanMX cassette) was used with GT128. These primers, if the insertion is correct, amplify a band of 1 kb.
The homozygous diploid deletion strains shown in Fig. 4 [in the BY4743 background: MATα/a his3Δ1/his3Δ1 leu2Δ0/leu2Δ0 LYS2/lys2Δ0 MET15/met15Δ0 ura3Δ0/ura3Δ0, deletions created by replacement of the open reading frame with a KanMX cassette (Winzeler et al., 1999)] were from the Deletion Consortium Collection through Research Genetics (Huntsville, AL) or OpenBiosystems (Huntsville, AL).
Strains were routinely grown on YPAD medium or SD medium (Adams et al., 1997; Sherman et al., 1974) at 30°C. In one experiment, galactose (Gal) replaced dextrose in YPAD. SD medium with appropriate supplements to cover auxotrophies was used to confirm genetic markers and for selection. Antibiotics [cycloheximide, hygromycin B, paromomycin, anisomycin (all Sigma Chemical, St Louis, MO) or geneticin (InVitrogen, Rockville, MD)], MA (Sigma) and TMA (Sigma), as well as other cations (as chloride salts; Sigma), were added to YPAD at the concentrations given in the figure and table legends. Antibiotics were added to media after autoclaving. Buffered medium was made as described (Tanida et al., 1995; Withee et al., 1998). Briefly, 0.1 M succinic acid at either pH 3.0 or pH 6.0 was mixed with an equal volume of 2× YPAD, the pH was adjusted with concentrated NaOH to a final pH of 3.0 or 6.0, respectively, then the medium was filter sterilized. Sensitivity to MA (maximum final concentration of 200 mM) was tested in minimal glutamate medium (Soupene et al., 2001).
Plasmids and transformations
The plasmids used in this study are listed in Tables 2 and 3. Oligonucleotides (Table 1) were obtained from InVitrogen (Rockville, MD). The site-directed mutant alleles of ARL1 in YEp352 (2 micron origin; URA3) were described previously (Rosenwald et al., 2002). The high-copy-number suppressor screen was performed using a genomic library (Nasmyth and Reed, 1980) (American Type Culture Collection, Manassas, VA). Library plasmids able to suppress the hygromycin-B-sensitive phenotype of the arl1 strain are detailed in Table 3.
Several subclones were constructed. Restriction enzymes were either from InVitrogen or New England Biolabs (Beverly, MA). First, plasmid pDH53, containing MIG1 and YGL034c, was digested with EagI, which cuts once in the insert and once in the vector, removing a 2.2 kb piece including the promoter and the first 50 bp of the YGL034c coding sequence. This was religated to create pDH100. Second, FUN21 was subcloned by PCR using GT74 and GT75 using pDH15 as template. SphI restriction sites built into the oligonucleotides permitted subcloning to YEp351 at the SphI site (Hill et al., 1986) to make pDH21-18. Third, pMXL244 (vector YEp352), containing the ATC1 gene, was originally isolated from a library as a high-copy-number suppressor of the temperature-sensitive phenotype of strain AR101 (arl1Δ::HIS3 ssd1-100) and includes 737 bp upstream and 802 bp downstream of the ATC1 open reading frame (Rosenwald et al., 2002).
Yeast transformations were performed using the procedure of Ito et al. (Ito et al., 1983). Plasmids were retrieved from yeast using the procedure of Hoffman and Winston (Hoffman and Winston, 1987). Plasmids were electroporated into electrocompetent Escherichia coli (DH10B) following the manufacturer's instructions (InVitrogen). Isolation of DNA from E. coli was performed either with Wizard (Promega, Madison, WI) or Qiagen (Valencia, CA) kits.
Library transformants were selected on minimal medium lacking leucine, then replica-printed onto YPAD with 0.1 mg ml–1 hygromycin B. Plasmids were isolated from transformants that grew on hygromycin B, transformed into E. coli, purified and then analysed by restriction digestion with EcoRI, HindIII and BglII. Plasmids were retransformed into MA03 to confirm the suppressing function was on the plasmid. Subclones of the library inserts were tested the same way, except for pERG20, in which ERG20 is under the control of the inducible GAL1/10 promoter. Strain MA03 containing pERG20 was grown on YPAGal and YPAGal with 0.1 mg ml–1 hygromycin B. Control experiments demonstrated that MA03 transformed with a construct containing ARL1 under the control of the GAL1/10 promoter, plasmid pARY1-8, complemented the phenotype on YPAGal with 0.1 mg ml–1 hygromycin B.
Sequencing of library plasmids was performed with oligonucleotides MP10 and MP11 elongated with Big Dye 3 reagents (Applied Biosystems, Foster City, CA) by PCR. The PCR products were purified on Sephadex G-50 spin columns (Princeton Separations, Princeton, NJ). An Applied Biosystems 377 sequencer was used.
Western blotting for Arl1p using a polyclonal anti-peptide antibody was performed as described (Rosenwald et al., 2002). Trk1-Myc was detected using a commercially available anti-c-Myc antibody (9E10; Roche, Indianapolis, IN). Anti-Vph1p and anti-Pgk1p antibodies were from Molecular Probes (Eugene, OR). Anti-Pma1p antibody was the kind gift of C. Slayman (Yale University). Horseradish-peroxidase-linked secondary antibodies were from Amersham Biosciences (Piscataway, NJ). Detection was performed using the enhanced chemiluminescence (Amersham Biosciences).
Subcellular fractionation was performed essentially as described (Gaynor and Emr, 1997). Briefly, cells were converted to spheroplasts by digestion with zymolyase (25 μg ml–1 in 25 mM Tris-HCl, pH 7.5, 1 M sorbitol, 20 mM NaN3, 20 mM NaF, 10 mM dithiothreitol) for 30-45 minutes at 30°C. Spheroplasts were then lysed by resuspension in a hypo-osmotic buffer (10 mM HEPES-KOH, pH 6.8, 0.2 M sorbitol, 50 mM potassium acetate, 2 mM EDTA, 1 mM dithiothreitol, containing the protease inhibitors phenylmethylsulfonyl fluoride [20 μg ml–1], antipain [5 μg ml–1], leupeptin [0.5 μg ml–1], pepstatin [0.7 μg ml–1] and α2-macroglobulin [10 μg ml–1]) and disrupted using a glass tissue homogenizer. After clearing the lysates of unlysed cells and cell debris by low-speed centrifugation, the lysates were then separated by differential centrifugation. For supernatant samples, proteins were first concentrated by precipitation with trichloroacetic acid (Peterson, 1977) before gel electrophoresis. Proteins were identified in each fraction by western blot analysis.
Uptake of [14C]-MA (Amersham Biosciences) was performed essentially as described (Navarre and Goffeau, 2000). Briefly, cells were grown in SD medium with the necessary supplements to cover auxotrophies. Log-phase cells were washed twice with and resuspended in water to give an OD600 value of 20 and stored on ice until ready for use. Cells were then diluted 1:1 with 2× reaction buffer for a final concentration of 10 mM 2-[N-morpholino]ethanesulfonic acid (MES), pH 6.0 (adjusted with NaOH), containing 50 mM glucose. The reaction mixture was incubated for 5 minutes at room temperature, then the uptake reaction was initiated by the addition of 2.5 μCi ml–1 [14C]-MA (2 mM final concentration). 100 μl aliquots were removed at intervals up to 60 minutes and diluted into 10 ml of ice-cold 20 mM MgCl2. Cells containing [14C]-MA were collected by rapid filtration onto nitrocellulose filters (Schleicher and Scheull, Keene, NH) followed by three 10 ml washes with ice-cold 20 mM MgCl2.
Uptake of 86Rb+ (Perkin-Elmer, Boston, MA) was adapted from Mulet et al. (Mulet et al., 1999). Cells were grown in SD medium with the necessary supplements (including 200 mM KCl for experiments in which trk1 cells were examined). Log-phase cells were washed twice with and then resuspended in water to give an OD600 value of 20 and incubated on ice until ready for use. Cells were then diluted 1:1 with 2× reaction buffer for a final concentration of 50 mM succinic acid pH 5.5 (adjusted with Trizma base) containing 4% glucose and incubated for 5 minutes at room temperature. 86RbCl (0.22 μCi ml–1, 0.2 mM final concentration) was added from a concentrated stock and cells were incubated at room temperature for up to 45 minutes. Cells were collected and washed as above for MA uptake. 86Rb+ efflux was accomplished by loading cells with 86RbCl as above, then washing once with 1× reaction buffer. Cells were then collected and washed as above.
Proton efflux was measured as described (Perlin et al., 1988). Briefly, cells were grown to mid-log phase (OD600 0.8) in YPAD. Cells diluted to give an OD600 value of 40 were washed three times with distilled water, then stored on ice for 1-3 hours. At the end of this period, cells were pelleted by centrifugation and resuspended in 50 ml 25 mM KCl or distilled in water to give an OD600 value of 0.8. A pH electrode was suspended in the cell mixture, which was subjected to constant stirring. The pH was measured until a stable baseline was observed, then glucose was added from a 100% stock in water to a final concentration of 2%. The pH was read every 15 seconds until a new steady state was reached, usually within 20-25 minutes.
arl1 mutants are sensitive to hygromycin B and related compounds
An arl1 strain, MA03, was found to be extremely sensitive to hygromycin B (Fig. 1), a cationic aminoglycoside that inhibits translation (Brodersen et al., 2000). Growth of MA03 on media containing the related compounds geneticin (G418, added to YPAD medium at a concentration of 25 μg ml–1) and paromomycin (0.25 μg ml–1) was also strongly inhibited. The arl1 strain was also sensitive to a structurally unrelated cationic translation inhibitor anisomycin (20 μg ml–1). However, addition of cycloheximide (0.3 μg ml–1), which is not a cation, inhibited growth of parental and arl1 strains to a similar extent (data not shown), suggesting that inhibition of translation per se was not the cause of the phenotype observed.
Eight independent arl1 isolates derived from the same parent exhibited the same hygromycin B and anisomycin phenotypes. The homozygous diploid arl1::KanMX deletion strain from the Deletion Collection was also sensitive to hygromycin B (see below), whereas most of the other strains in the collection were not (G. L. Fell and A. G. Rosenwald, unpublished). Upon backcross of an arl1::HIS3 strain to an ARL1 strain of the opposite mating type and sporulation of the resulting diploid, the His+ phenotype segregated 2:2 and all His+ spores were sensitive to hygromycin B. Transformation of arl1 strains with either a multicopy plasmid (Fig. 2) or a low-copy-number plasmid (data not shown) containing wild-type ARL1 complemented the hygromycin-B-sensitive phenotype. Hygromycin-B sensitivity is therefore tightly linked to the loss of ARL1.
Site-directed mutant alleles of ARL1 and the hygromycin B phenotype
Because ARL1 encodes a protein that binds guanine nucleotides (Lee et al., 1997), we tested the hypothesis that the nucleotide-binding state of Arl1 protein was important for its function with respect to the hygromycin-B-sensitive phenotype. The arl1 strain was transformed with high-copy-number plasmids, each containing a different site-directed mutant allele of ARL1, including those predicted to alter nucleotide binding and hydrolysis. The wild-type allele complemented the phenotype (Fig. 2). The myristoylation mutant ARL1[G2A] was unable to complement this phenotype, demonstrating that this modification was important for activity. The alleles encoding mutants predicted to have high exchange rates for or to bind only GDP, ARL1[N127I] and ARL1[T32N], respectively, did not complement this phenotype. Interestingly, the mutant predicted to be unable to hydrolyse GTP, ARL1[Q72L], complemented the phenotype weakly (Fig. 2). Surprisingly, only the allele encoding a protein predicted to be unbound to nucleotide, ARL1[D130N], complemented as well as the wild type. This mutant, by analogy to Ras (Powers et al., 1991; Powers et al., 1989), Ypt1 (Jones et al., 1995) and EF-Tu (Hwang and Miller, 1987), is predicted to bind xanthine nucleotides preferentially, and so to be `empty' in vivo. As a control, western-blot analysis of lysates with an anti-Arl1-peptide antibody (Rosenwald et al., 2002) demonstrated that, in all strains transformed with an ARL1-containing plasmid, a ∼20 kDa band was observed. This band was not observed in cells transformed with empty vector (data not shown). Lack of complementation by these alleles cannot therefore be ascribed to lack of expression. Taken together, these results suggest that the relevant conformation of Arl1p for complementation of this phenotype might be the nucleotide-free form rather than the GTP-bound form.
arl1 mutants show increased uptake of a toxic cation
Mutations in several genes have been shown to cause sensitivity to hygromycin B and other toxic cations. These include the genes for calcineurin (Withee et al., 1998), the K+ influx transporters Trk1p and Trk2p (Madrid et al., 1998), and two kinases that regulate these transporters, Hal4p and Hal5p (Mulet et al., 1999), all of which are thought to regulate membrane potential. To determine whether the loss of ARL1 caused a phenotype consistent with inability to regulate membrane potential, we performed several additional experiments.
First, inclusion of K+ in the growth medium suppressed the hygromycin-B-sensitive phenotype of the arl1 mutant very well (Fig. 1). Second, as described above, arl1 strains were sensitive to a range of other cationic translation inhibitors in addition to hygromycin B. Additionally, the arl1 strain was sensitive to several other toxic cations that do not inhibit translation, including TMA (Fig. 1). Sensitivity to TMA was also suppressed by K+ (Fig. 1). The arl1 strain was also sensitive to protons. Although ARL1 and arl1 strains grew at the same rate in medium buffered to pH 6.0, the arl1 strain grew substantially more slowly in medium buffered to pH 3.0 than the parental wild-type strain (data not shown). Similarly, the arl1 strain grew substantially more slowly in medium containing 25 mM MA (data not shown). Thus, the arl1 mutant showed a general sensitivity to toxic cations and not to cationic translation inhibitors specifically. Third, because the Ca2+/calmodulin-dependent protein phosphatase calcineurin is a major regulator of cellular responses to environmental ionic change, we transformed the arl1 mutant strain with an activated calcineurin construct, pCAtrB-2, containing full-length CNB1 and truncated CNA1, missing the calmodulin-binding and autoinhibitory domains (Mendoza et al., 1996). This construct suppressed both the hygromycin-B-sensitive and the TMA-sensitive phenotypes of the arl1 mutant (data not shown).
Finally, the ARL1 and arl1 strains were incubated with a radioactive version of one of the toxic cations, [14C]-MA. This cation has been used as a tracer by others to demonstrate hyperpolarity in yeast (Madrid et al., 1998; Mulet et al., 1999; Navarre and Goffeau, 2000). The arl1 mutant strain took up 25% more than the wild type after 60 minutes (Fig. 3). In addition, inclusion of K+ in the buffer decreased the amount of uptake in both wild-type and arl1 strains, but to a greater extent in the arl1 strain, bringing the amount of uptake observed within the amounts seen in wild-type cells (data not shown). The observed increase in [14C]-MA uptake upon deletion of ARL1 was of the same order of magnitude relative to the wild type as has been observed by others for genes with established roles in ion homeostasis (Madrid et al., 1998; Mulet et al., 1999). In summary, because the arl1 strain was sensitive to a range of structurally unrelated cations and the sensitivity phenotypes could be suppressed by inclusion of K+ or calcineurin, these results are consistent with the hypothesis that cells lacking ARL1 are hyperpolarized, which in turn leads to increased uptake of toxic cations.
High-copy-number suppressors of the arl1 hygromycin-B-sensitive phenotype
To identify genes that act downstream of ARL1, we performed a high-copy-number suppressor screen by transforming a genomic library (Nasmyth and Reed, 1980) into the arl1 strain. We obtained several different genes, all of which also suppressed the TMA-sensitive phenotype (Table 3; data not shown). Seven identical plasmids, exemplified by pDH18, suppressed very well and contained a portion of chromosome II that included ARL1. Because multiple ARL1 isolates were found, sufficient numbers of transformants were obtained to cover the genome. We also identified three genes that appeared to function downstream of ARL1.
Plasmid pDH50 contained three genes: ERG20, QCR8 and HAL5. ERG20 encodes farnesyl pyrophosphate synthetase (Anderson et al., 1989; Szkopinska et al., 2000) and was unable to suppress the phenotype. However, HAL5 alone (plasmid pM89) suppressed the phenotype as well as did the original plasmid. A homolog of HAL5, called HAL4 (plasmid pM73), also suppressed, although HAL4 was not found in the screen. HAL4 and HAL5 encode Ser/Thr protein kinases that function upstream of the K+ influx transporters, Trk1p and Trk2p (Mulet et al., 1999). High-copy-number TRK1 in the arl1 strain suppressed hygromycin-B sensitivity relatively weakly (data not shown), suggesting that mere overexpression of Trk1p is insufficient and that Trk1p needs to be regulated by the Hal proteins.
Plasmid pDH32 contained YFR038w, YFR039c, SAP155 and ERJ5. SAP155 alone (CB2643) suppressed as well as the original. Three other SAP genes, SAP4, SAP185 and SAP190, have been isolated based on sequence homology. All four Sap proteins interact with Sit4p (Luke et al., 1996), a phosphatase required for progression through the cell cycle, specifically the G1-S transition (Sutton et al., 1991). However, only SAP155 suppressed the hygromycin-B-sensitive phenotype of the arl1 mutant.
We also isolated several different library plasmids that suppressed weakly compared to SAP155 and the HAL genes (Table 3). Each insert had at least two open reading frames, so the suppressor in each case was identified by subcloning. First, ATC1 (also known as LIC4), a gene involved in regulation of responses of cells to Li+ (Hemenway and Heitman, 1999) was isolated. Other work from our laboratory has shown that arl1 atc1 double mutants are extremely sensitive to Li+ (Munson et al., 2004). Second, MIG1 (encoding a repressor of ENA1 transcription) was isolated (Alepuz et al., 1997). ENA1 encodes the major Na+ efflux pump in cells (Garciadeblas et al., 1993; Marquez and Serrano, 1996). In addition, isolates were found that contained CKI1 [encoding choline kinase (Kim and Carmen, 1999)], YJL193w (Paulsen et al., 1998) (encoding a protein with homology to Sly41p, a triose phosphate transporter) and FUN21 (a gene with no known function).
HAL4, HAL5 and SAP155 are downstream effectors of ARL1
We next investigated homozygous diploid deletion strains missing genes identified as high-copy-number suppressors of the loss of ARL1. Because the arl1 strain was hygromycin-B sensitive and this sensitivity was reversed by K+ (Fig. 1), we screened the selected deletion strains on medium containing hygromycin B with and without KCl. The three deletion strains corresponding to the strong high-copy-number suppressors (the hal4, hal5 and sap155 mutants) were sensitive to hygromycin B (Fig. 4) and TMA (data not shown), and sensitivity was suppressed by K+ (Fig. 4). It was previously shown that hal4 hal5 and trk1 trk2 double mutants are sensitive to hygromycin B (Madrid et al., 1998; Mulet et al., 1999). We tested both the trk1 and the trk2 homozygous diploid single mutant strains; only the trk1 mutant was sensitive to hygromycin B (Fig. 4). Trk1p activity is expressed at higher levels than Trk2p activity under normal growth conditions (Ramos et al., 1994), suggesting that expression of TRK1 in the trk2 mutant was sufficient (Bertl et al., 2003).
The remaining five strains were not particularly sensitive to hygromycin B (Fig. 4) or TMA (data not shown) compared with the wild-type parent. Thus, the weak suppressors ATC1, CKI1, FUN21, MIG1 and YJL193w did not appear to function downstream of ARL1 and were assumed to be bypass suppressors. These were not studied further.
We next determined epistatic relationships among these genes. The arl1, hal4, hal5 and sap155 strains were transformed with ARL1 (pARY1-3), HAL5 (pM89) or SAP155 (CB2643). Although the loss of ARL1 was suppressed by HAL4, HAL5 and SAP155, overexpression of ARL1 was unable to suppress the loss of any of these genes, demonstrating that HAL4, HAL5 and SAP155 were downstream of ARL1. Furthermore, HAL5 appeared to be downstream of SAP155, because HAL5 suppressed the hygromycin-B-sensitive phenotype of the sap155 mutant. Finally, SAP155 failed to suppress the hygromycin-B-sensitive phenotype of the hal5 mutant (data not shown), lending additional support to the notion that HAL5 is downstream of SAP155.
arl1 mutants exhibit a rubidium uptake defect
The evidence cited above is consistent with a model in which ARL1 regulates Trk1 K+ transporter activity. To test this hypothesis, wild-type and arl1 strains were incubated with 86RbCl. Rb+, a congener of K+, has been used extensively to examine K+ homeostasis (Haro et al., 1993; Ko et al., 1990; Madrid et al., 1998; Mulet et al., 1999; Ramos et al., 1994; Yenush et al., 2002). Loss of ARL1 resulted in decreased 86Rb+ uptake (Fig. 5A). However, the decrease in 86Rb+ influx by the arl1 mutant was not as severe as that exhibited by a strain lacking TRK1 but was more similar to the level of uptake observed in strains with deletions of the regulatory genes HAL4 and especially HAL5 (Fig. 5B). Loss of ARL1 did not have an effect on 86Rb+ efflux (data not shown).
arl1 mutants do not exhibit a proton efflux defect
Plasma membrane potential is controlled by regulation of both cation influx, driven in large measure by the K+ transporters Trk1p and Trk2p, and by cation efflux, driven primarily by the H+-efflux pump Pma1p (Gaber, 1992). Hyperpolarization can be caused by either a decrease in cation uptake or an increase in cation efflux. Although the results of the Rb+ uptake experiment are consistent with a decrease in K+ uptake being responsible for the membrane polarity defects of the arl1 strain, this result does not exclude a role for altered cation efflux in this strain. To test the hypothesis that excess proton efflux contributes to the hyperpolarity observed in arl1 cells in addition to the K+ (Rb+) influx defect (Fig. 5), a whole cell proton efflux assay was used. When resuspended in water and assayed using this protocol (Fig. 6), the arl1 mutant (closed circles) was unable to acidify the external medium as well as the wild-type strain (closed squares), counter to the result predicted if proton efflux is contributing to hyperpolarity. However, under these conditions, both efflux and influx of H+ occur (Perlin et al., 1988). To examine the effect of the arl1 mutation on efflux only, the experiment was repeated in the presence of K+ (Perlin et al., 1988). When cells were suspended in 25 mM KCl, both strains exhibited an apparent increase in proton efflux and, importantly, the difference in proton efflux between the two strains was eliminated (Fig. 6, open symbols). The difference in steady-state pH level in the absence of added K+ between the arl1 and wild-type strains can therefore be attributed to increased H+ uptake by the arl1 mutant and not decreased H+ efflux. This result is completely consistent with our earlier observation that the arl1 strain is sensitive to low pH. In summary, these results suggest that ARL1 regulates membrane potential via regulation of K+ influx, rather than by regulation of H+ efflux.
Loss of Arl1 affects neither the steady-state level of Trk1p nor its localization
The results shown above suggest two different models. First, loss of ARL1 could result in decreased levels of Trk proteins at the plasma membrane. This hypothesis is consistent with the known role of Arl1p in regulation of membrane traffic (Bonangelino et al., 2002; Jochum et al., 2002; Panic et al., 2003; Rosenwald et al., 2002; Setty et al., 2003). Second, loss of ARL1 could result in misregulation of a signaling cascade that regulates the activity of the Trk proteins. This hypothesis is consistent with our findings that HAL4 and HAL5 in high copy number suppress the hygromycin-B-sensitive phenotype of the arl1 strain.
To begin to discover which of these models was correct, we verified that ARL1 functions upstream of TRK1. To do this, an arl1 trk1 double mutant was constructed. 86Rb+ analysis (as in Fig. 5) demonstrated that this mutant had similar levels of uptake to the isogenic ARL1 trk1 mutant, suggesting that ARL1 and TRK1 function in the same genetic pathway (data not shown).
Subsequently, we constructed strains bearing a Myc-epitope-tagged TRK1 allele at the TRK1 chromosomal locus in both wild-type and arl1 backgrounds. The tagged allele, containing 13 Myc epitopes fused to the C-terminus of Trk1p, was functional by all measurements. First, when comparing cells with wild-type TRK1 and TRK1-Myc, the strains grow at the same rate as wild-type in the absence of added K+ (trk1 mutants are K+ bradytrophs). Second, 86Rb+ uptake (as in Fig. 5) was unaffected. Finally, the TRK1-Myc ARL1 strain was no more sensitive to hygromycin B than the TRK1 ARL1 strain. Similarly, the TRK1-Myc arl1 strain was indistinguishable from the TRK1 arl1 strain (data not shown).
As shown in Fig. 7, western blot analysis with an anti-Myc antibody revealed a 180 kDa band. This band is only present in cells bearing the TRK1-Myc allele, not cells containing wild-type TRK1 (data not shown). Loss of ARL1 did not appear to affect the steady-state levels of Trk1-Myc in cells. Subcellular fractionation using differential centrifugation was then performed. Similar amounts of Trk1-Myc were present in the P14 fractions of both ARL1 and arl1 strains, where both the plasma membrane marker Pma1p and the vacuolar membrane marker Vph1p are found. These results suggest that Arl1p might be important for regulating the activity of Trk1p at the plasma membrane, rather than regulating delivery of Trk1p to the plasma membrane.
Investigations into the functions of the Arl family began in 1991 with the identification of the arf-like gene of D. melanogaster (Tamkun et al., 1991). Many Arl genes have been identified in many different organisms since, demonstrating, first, that this gene family is highly conserved across eukaryotic evolution and, second, that family members are more divergent than are members of the related Arf family. Although many ARL genes have been identified, the functions of Arl proteins remain relatively unexplored. Data from our laboratory and others have shown that Arl1 is a regulator of membrane traffic in yeast (Bonangelino et al., 2002; Jochum et al., 2002; Rosenwald et al., 2002) and mammalian cells (Eboue et al., 1998; Icard-Liepkalns et al., 1997; Lowe et al., 1996; Lu et al., 2001; Van Valkenburgh et al., 2001). However, in this study, yeast Arl1 was found to have a novel role in regulation of K+ influx that does not appear to be a result of Arl1p's ability to control membrane traffic.
Mutations in several genes result in sensitivity to toxic cations, including hygromycin B. The genes include those coding for calcineurin [the A (catalytic) subunits encoded by CNA1 and CNA2, and the B (regulatory) subunit encoded by CNB1] (Withee et al., 1998), the plasma membrane K+ transporters Trk1p and Trk2p (Mulet et al., 1999), the protein kinases Hal4p and Hal5p (which regulate Trk1p and Trk2p) (Mulet et al., 1999), Nhx1p (the Na+/H+ exchanger at the pre-vacuolar membrane) (Gaxiola et al., 1999), Gef1p (the chloride channel at the pre-vacuolar membrane) (Gaxiola et al., 1999), and Pmp3 (a small hydrophobic peptide) (Navarre and Goffeau, 2000). Mutations in these genes are thought to result in hyperpolarization of the plasma membrane, leading to increased uptake of toxic cations (Goossens et al., 2000), as we observed in the arl1 mutant. In several of these strains, specifically the hal4 hal5 and trk1 trk2 double mutants, sensitivity to toxic cations like hygromycin B can be suppressed by the addition of K+ (Haro et al., 1993; Madrid et al., 1998; Mulet et al., 1999), an effect also observed in the arl1 mutant. By contrast, mutations in other genes, including PMA1 (Goossens et al., 2000; Perlin et al., 1988; Perlin et al., 1989; Serrano et al., 1986) and the phosphatases PPZ1 and PPZ2 (Yenush et al., 2002), led to toxic cation tolerance and decreased potential across the plasma membrane. Because the hyperpolarization phenotype of the arl1 mutant is reversed by the inclusion of a depolarizing cation, K+, our findings are consistent with a model in which the plasma membrane of the arl1 strain is hyperpolarized as a result of disturbances in K+ influx rather than H+ efflux.
We identified three suppressors that are likely to function in the same pathway as ARL1 – HAL4, HAL5 and SAP155. The molecular function of SAP155 is not well understood at present. All four of the Sap proteins (Sap4p, Sap155p, Sap185p and Sap190p) interact with the phosphatase, Sit4p, a regulator of the G1-S transition. Sit4p exists as a monomer in early G1 phase but becomes associated with Sap155p and, in a separate complex, with Sap190p by the end of G1, and the complexes remain together until mid-M-phase (Sutton et al., 1991). The binding of Sap190 and Sap155 to Sit4 appear to be mutually exclusive, and overexpression of SAP155 cannot suppress the double deletion of both SAP190 and its relative SAP185, suggesting that the functions Sap155p provides are different from the functions of Sap185p and Sap190p (Luke et al., 1996). The Sap proteins might be substrates for Sit4p, because they are hyperphosphorylated in the absence of Sit4p (Luke et al., 1996). It has also been observed that overexpression of only SAP155, but not any of the other three SAP genes, protects cells against Kluyveromyces lactis zymocin, an inhibitor that prevents progression through G1 to S (Jablonowski et al., 2001). Furthermore, functional Sit4p is required for expression of the zymocin-sensitivity phenotype (Jablonowski et al., 2001).
Our results with SAP155 suggest a link between regulation of ion homeostasis and regulation of the cell cycle, which has been explored by others. SIT4 expression is induced by Li+, Na+ and K+, and overexpression of SIT4 confers resistance to Li+ (Masuda et al., 2000). SIT4 overexpression stimulates Rb+ efflux and causes a rise in intracellular pH. The Ppz phosphatases also help to regulate the responses of cells to Na+, H+ and K+ (Yenush et al., 2002). Deletion of both PPZ genes renders cells tolerant to a range of toxic cations including Li+, Na+, spermine, TMA and hygromycin B. Cation tolerance depends on the presence of functional TRK1 and TRK2. Overexpression of PPZ1, by contrast, renders cells defective for Rb+ uptake and confers a slow-growth phenotype, implying that it is a negative regulator of cell cycle progression. The slow growth phenotype, however, is partially suppressed by inclusion of extra K+ in medium. Finally, Ppz1p and Sit4p appear to have opposing functions in regulation of the cell cycle (Clotet et al., 1999). Thus, regulation of ion homeostasis is important for cell cycle regulation and might explain the connection we observed between SAP155 and ARL1, an observation we continue to explore. Specifically, we are testing the hypothesis that overexpression of SAP155 leads to decreased K+ efflux in a manner that depends on SIT4 and TOK1, the K+ efflux channel (Bertl et al., 2003; Zhou et al., 1995).
Uptake of K+ by yeast is accomplished by the transporters Trk1p and Trk2p under normal growth conditions (Gaber et al., 1988). Trk2p shares 55% amino acid identity with Trk1 (Ko and Gaber, 1991). It was initially suggested that each has 12 membrane spans (Gaber et al., 1988; Ko and Gaber, 1991) but, more recently, it has been suggested that each Trk protein instead contains four repeats of a membrane-span/P-loop/membrane-span motif, each repeat similar in structure to the K+ channel KcsA, from Streptomyces lividans (Haro and Rodriguez-Navarro, 2002). Loss of both TRK genes results in a conditional lethal phenotype suppressed by K+ (Ko and Gaber, 1991; Madrid et al., 1998). Nonspecific uptake of K+ through other transporters permits growth under these conditions. One nonspecific channel is NSC1, which has been described electrochemically, although the gene(s) that encodes this activity has not yet been identified (Bihler et al., 1998; Bihler et al., 2002). Interestingly, hygromycin B blocks this channel (Bihler et al., 2002).
TRK1 encodes a high-affinity K+ transporter (Gaber et al., 1988), whereas TRK2 (dispensable in the presence of TRK1) encodes a transporter of moderate affinity (Ramos et al., 1994). Trk2p in addition appears to mediate an inward proton current that is regulated by extracellular pH (Bihler et al., 1999). The trk1 trk2 strains are sensitive to hygromycin B and other cations, and exhibit hyperpolarization of the plasma membrane (Madrid et al., 1998). The hal4 hal5 mutants exhibit similar but milder phenotypes (Mulet et al., 1999). As we demonstrate here, the arl1 mutant was similar to the hal4 hal5 double mutant (Mulet et al., 1999) and to the hal4 and hal5 single mutants.
HAL4 (also called SAT4) and HAL5 encode Ser/Thr kinases that are partially redundant, although we observed that the single mutant strains were sensitive to hygromycin B and TMA. In addition, the single mutants take up less 86Rb+ than the wild type. The function of the two Hal proteins in regulating ion stress depends on the presence of TRK1 and TRK2 (Mulet et al., 1999), demonstrating that they act upstream of TRK1 and TRK2. We have shown here that ARL1 is upstream of the HAL genes and upstream of TRK1. HAL gene overexpression leads to toxic cation tolerance, suggesting that these proteins alter the affinity of Trk1p and Trk2p for K+ and for toxic ions such as Na+ and Li+. HAL4 and HAL5 are members of the Npr1 subfamily of kinases (Hunter and Plowman, 1997). This subgroup includes other regulators of membrane transport, including Npr1p, which regulates the Gap1p amino acid permease (Stanbrough and Magasanik, 1995), and Ptk2p and YOR267c, which regulate Pma1p, the H+-ATPase (Goossens et al., 2000). Our genetic evidence linking ARL1 to HAL4 and HAL5 thus suggests that ARL1 acts as an upstream regulator of K+ influx via regulation of HAL4 and HAL5.
These data are consistent with a model in which Arl1p is a positive regulator of the positive regulators Hal4p and Hal5p. Because we found that the inclusion of high-copy-number TRK1 suppressed the hygromycin-B sensitivity of the arl1 mutant weakly, the presence of Trk1 is not sufficient; rather, the transporter must be in the high-affinity form. Our data are not consistent with a model in which Arl1p controls delivery of Trk1p to the plasma membrane, because loss of ARL1 has no effect on the steady-state level of Trk1-Myc, nor on the fraction in which it resides. Although the P14 fraction contains both vacuolar and plasma membranes, and this analysis does not distinguish between the two, we can assert that there is not substantial Trk1-Myc in internal membranes (Golgi, secretory vesicles and endosomes) in the arl1 mutant strain because virtually no Trk1-Myc was observed in the P100 fraction.
In summary, we demonstrate here that Arl1p, a member of the Arl family of proteins, has a novel role in ion homeostasis and propose that it contributes to K+ influx via regulation of Trk proteins. Despite Arl1p's documented role in regulation of membrane traffic, our results suggest that Arl1p's role in regulation of K+ influx proceeds by a mechanism other than by regulation of Trk delivery to the plasma membrane. Future experiments will test the hypothesis that ARL1 encodes a regulator of a phosphorylation cascade that involves the Hal kinases and the Trk K+ transporters.
We thank E. Christy, F. Scaffidi and L. Dicker for help with the initial experiments, and M. Tokic for help with the experiment shown in Fig. 6. We thank K. Arndt (Weyth-Ayerst, Madison, NJ), G. Carman (Rutgers University, Piscataway, NJ), R. Gaber (Northwestern University, Evanston, IL), A. Hopper (Pennsylvania State University College of Medicine, University Park, PA), I. Mendoza (Instituto de Recursos Naturales y Agrobiologia, Consejo Superior de Investigaciones Cientificas, Sevilla, Spain), P. Novick (Yale University, New Haven, CT) and J. Mulet (Instituto de Biologia Molecular y Celular de Plantas, Universidad Politecnica de Valencia, Spain) for plasmids. We also thank C. Slayman (Yale University, New Haven, CT) for the anti-Pma1p antibody. We especially acknowledge the help of R. Rolfes (Georgetown University, Washington, DC) for plasmids, oligonucleotide sequencing and general advice. We thank A. Peña and A. Goffeau for helpful comments about this work. A.R.'s laboratory is funded by the US National Science Foundation (CAREER MCB 9875782).
- Accepted December 9, 2003.
- © The Company of Biologists Limited 2004