Trinucleotide Insertions, Deletions, and Point Mutations in Glucose Transporters Confer K⁺ Uptake in Saccharomyces cerevisiae

Abstract

Deletion of TRK1 and TRK2 abolishes high-affinity K⁺ uptake in Saccharomyces cerevisiae, resulting in the inability to grow on typical synthetic growth medium unless it is supplemented with very high concentrations of potassium. Selection for spontaneous suppressors that restored growth of trk1Δ trk2Δ cells on K⁺-limiting medium led to the isolation of cells with unusual gain-of-function mutations in the glucose transporter genes HXT1 and HXT3 and the glucose/galactose transporter gene GAL2. ⁸⁶Rb uptake assays demonstrated that the suppressor mutations conferred increased uptake of the ion. In addition to K⁺, the mutant hexose transporters also conferred permeation of other cations, including Na⁺. Because the selection strategy required such gain of function, mutations that disrupted transporter maturation or localization to the plasma membrane were avoided. Thus, the importance of specific sites in glucose transport could be independently assessed by testing for the ability of the mutant transporter to restore glucose-dependent growth to cells containing null alleles of all of the known functional glucose transporter genes. Twelve sites, most of which are conserved among eukaryotic hexose transporters, were revealed to be essential for glucose transport. Four of these have previously been shown to be essential for glucose transport by animal or plant transporters. Eight represented sites not previously known to be crucial for glucose uptake. Each suppressor mutant harbored a single mutation that altered an amino acid(s) within or immediately adjacent to a putative transmembrane domain of the transporter. Seven of 38 independent suppressor mutations consisted of in-frame insertions or deletions. The nature of the insertions and deletions revealed a striking DNA template dependency: each insertion generated a trinucleotide repeat, and each deletion involved the removal of a repeated nucleotide sequence.

Selective permeation across plasma membranes is achieved through a variety of transport proteins, each of which allows a specific clientele of nutrients and ions to traverse the membrane while excluding others. The selectivity of membrane transport proteins also ensures that gradients of metabolites and ions are appropriately maintained.

The uptake of potassium in Saccharomyces cerevisiae is normally mediated by two highly related membrane proteins, Trk1 (13, 54) and Trk2 (29, 30) and is believed to be driven by the large electrical potential (approximately −180 mV) (for a review, see reference 8) generated by the plasma membrane H⁺-ATPase Pma1 (58). Deletion of both TRK1 and TRK2 increases the concentration of K⁺ required to support growth more than 100-fold, resulting in the inability of these cells to grow on synthetic medium that is not supplemented with high concentrations of potassium (30). The K⁺ uptake-defective phenotype reflects the inability of other transport proteins to mediate efficient K⁺ uptake under these conditions.

We have exploited the conditional negative phenotype of trk1Δ trk2Δ cells to select for extragenic suppressor mutations that restore potassium uptake. Surprisingly, most of the suppressor mutations occurred in genes encoding glucose transporters. Although these hexose transporters have extensive sequence identity (reviewed in reference 32), they are completely unrelated to the Trk proteins. Nevertheless, single mutations in hexose transporters can confer the ability to transport potassium to an extent that satisfies the cellular need for this ion.

In S. cerevisiae, glucose uptake is mediated by transporters encoded by members of the large multigene HXT family (6, 32). This results in a functional redundancy that essentially precludes a straightforward loss-of-function analysis of individual glucose transporters in a wild-type background. In contrast, glucose uptake is virtually absent in cells in which HXT1, HXT2, HXT3, HXT4, HXT6, HXT7, and GAL2 are disrupted, and as a consequence, they are unable to grow on glucose as a sole carbon source (37). The reintroduction of any one of the disrupted HXT genes is sufficient to restore growth of this strain on glucose (37). Thus, the effects of the suppressor mutations at HXT1 and HXT3 on their ability to transport glucose could be independently assessed. Because the suppressors conferred a dominant gain-of-function phenotype (uptake of K⁺), trivial effects on transporter expression or localization were avoided. Through this novel approach, we were able to identify new sites that play essential roles in glucose transport.

MATERIALS AND METHODS

Strains and media.

The potassium uptake-defective S. cerevisiae strain CY162 (MATα trk1Δ trk2Δ::HIS3 ura3-52 his4-15) (1) was used to select for spontaneous suppressors that restored growth on potassium-limiting medium. The dominance of the suppressor mutations isolated in CY162 was tested by mating suppressors with strain CY152 (MATa trk1Δ trk2Δ ura3-52 trp1). Meiotic offspring containing the suppressor mutations in a MATα background were obtained from these diploids and used for the recombination tests described in the text. Standard yeast extract-peptone-dextrose (YPD), yeast nitrogen base (YNB), sporulation media, and genetic techniques were as described by Sherman et al. (59). Low-salt (LS) medium (48) (modified as described previously [54]) lacks potassium and sodium and was used to generate potassium-limiting growth conditions. Potassium was added as indicated. Yeast transformation was performed by electroporation (5) using Gene Pulser (Bio-Rad, Richmond, Calif.).

DNA manipulations.

Standard techniques for DNA purification and cloning were as described in Maniatis et al. (38). DNA sequencing analysis was performed by the dideoxy method (55) by using T7 DNA polymerase and synthetic oligonucleotide primers. PCR was performed on the GeneAmp PCR system 9600 (Perkin-Elmer Cetus, Norwalk, Conn.) as described by Innis et al. (21). Oligonucleotide primers were either obtained from the Northwestern University Biotechnology Facility or synthesized with the Oligo Synthesizer 1000 (Beckman, Arlington Heights, Ill.).

Isolation of suppressors of the trk1Δ trk2Δ phenotype.

Suppressor mutations that restore potassium uptake to trk1Δ trk2Δ cells were isolated by replica plating colonies of trk1Δ trk2Δ cells (strain CY152 [30]) that had developed on YPD medium (supplemented with 100 mM KCl) to standard YNB-based yeast growth medium, which contains approximately 7 mM potassium. Suppressor mutations conferring the ability to grow on this medium were identified as papillations within an otherwise nongrowing patch of cells. Independence of suppressor mutations was guaranteed by picking only one mutant from such a patch.

Genetic analysis of RPD mutants.

Approximately 60 independent mutants, provisionally designated RPD for their reduced potassium dependency, were picked for analysis. Dominance tests were performed by mating each of the suppressor mutants with strain CY162 to generate diploid cells homozygous for the trk1Δ trk2Δ mutations but heterozygous for the RPD mutations. Growth of the diploid cells on YNB (7 mM K⁺) medium indicated dominance of the suppressor mutations. Due to the dominance of each of the suppressors isolated in this study, recombination tests were performed to determine which of the suppressors resided at unlinked loci. The absence of rpd⁺ spore colonies among at least 10 tetrads obtained from each RPD × RPD cross was taken as provisional evidence that the suppressors resided at the same or closely linked loci.

Cloning and molecular analysis of RPD suppressor alleles.

RPD5-1 and RPD103-1 suppressor alleles were cloned by screening libraries of genomic DNA fragments originating from the dominant mutant strains for their ability to confer growth on a trk1Δ trk2Δ recipient (CY162 [30]) strain on standard YNB (7 mM K⁺) medium. The wild-type alleles of RPD5 and RPD103 were obtained by the method of gap repair (49) and have been published previously (31). Sequence analysis revealed that RPD5 is identical to HXT1 (36) and RPD103 is identical to HXT3 (31).

Isolation of HXT3 suppressor alleles.

CY162 (trk1Δ trk2Δ) cells harboring a centromeric plasmid, pCK163, which expresses the wild-type HXT3 gene (31) were plated onto YNB medium lacking uracil but supplemented with 100 mM K⁺ and were allowed to develop into colonies. These colonies were replica plated onto standard YNB media (7 mM K⁺) to select for spontaneous suppressors. Mutants able to grow on YNB (7 mM K⁺) were tested for plasmid dependency of this phenotype. Approximately 120 independent mutants were obtained, 41 of which showed plasmid dependence. Plasmids were retrieved from these mutants and were reintroduced into CY162 cells to test for suppression, and the entire sequence of each of the HXT3 mutant alleles was determined.

⁸⁶Rb uptake assays.

Five-milliliter late-log-phase (optical density at 600 nm, 1 to 2) cultures grown at 30°C in YNB-based medium lacking uracil and supplemented with 100 mM K⁺ were harvested by centrifugation and washed twice in starvation buffer (50 mM Tris-succinate buffer [pH 5.9]). Potassium starvation was achieved by incubating cells in 10 ml of starvation buffer for 4 to 6 h at 30°C on a rotary shaker. Cells were harvested, washed once with LS buffer (LS medium containing no added K⁺, uracil, or sugar), and resuspended in 1 ml of LS buffer. The uptake assay conditions were as follows: LS buffer, 1 mM RbCl, 1 mM KCl, ∼6 μCi of ⁸⁶RbCl/ml, 4% glucose, and ∼1 × 10⁸ cells/ml. Cells were added to the assay mixture at time zero. Samples (200 μl) were taken from the assay mixture at the indicated time points, diluted into 5 ml of cold (4°C) LS buffer supplemented with 10 mM RbCl, harvested, and washed twice with 5 ml of cold LS buffer supplemented with 10 mM RbCl by filtering through a 0.45-μm-pore-size nitrocellulose filter (Millipore, Malborough, Mass.). ⁸⁶Rb uptake was assessed by measuring the radioactivity of the filters in a Beckman LS 7000 scintillation counter. The relative amount of ⁸⁶Rb present in each assay was assessed by measuring the radioactivity of 5 μl of the assay mixture prior to the addition of cells. Only background levels of radioactivity were measured on filters to which (cell-free) buffer containing the ⁸⁶Rb was applied.

Plasmid construction.

Cloning of the mutant and wild-type alleles of HXT3 under the control of the ADH1 promoter was carried out as follows. The open reading frames of mutant HXT3 alleles HXT3-23, HXT3-115, HXT3-122, HXT3-156, and HXT3-164 and of the wild-type HXT3 gene were amplified by PCR using primers 5′-AGT CAA GCT TAG ATC TCA TGA ATT CAA CTC CAG ATT-3′ and 5′-AGT CTC TAG AAG ATC TCA GCA CTA CGG TTT AGC GTG A-3′. The DNA products from PCR amplification were subcloned into vector pVT100U (63) at the HindIII and XbaI sites.

RESULTS

Isolation of HXT1-2, HXT3-1, and GAL2-1.

S. cerevisiae cells from which the two highly related potassium transporter genes TRK1 and TRK2 have been deleted are unable to grow on standard YNB medium, which contains only approximately 7 mM potassium. Spontaneous RPD (reduced potassium dependency) mutants that regained the ability to grow on YNB medium were isolated in strain CY162 and genetically analyzed. Each of the mutants was determined to be dominant, and recombination tests revealed the presence of three linkage groups. For the two groups comprising most of the suppressors, RPD5 and RPD103, the ability to grow on YNB or other low-K⁺ media was found to be dependent on glucose. Suppressor alleles RPD5-1 and RPD103-1 were cloned as described in Materials and Methods, and their sequences were determined. In both cases, a single large open reading frame that conferred suppression of the trk1Δ trk2Δ phenotype was found on the smallest subcloned fragment. A comparison of the sequences of these open reading frames with GenBank sequences revealed that RPD5-1 and RPD103-1 were alleles of the glucose (hexose) transporter genes HXT1 (36) (accession no. L07079) and HXT3 (31) (accession no. L07080), respectively. The suppressor allele of HXT1 was designated HXT1-2 and that of HXT3 was designated HXT3-1. The abilities of these mutations to rescue the growth of trk1Δ trk2Δ cells on potassium-limiting medium are illustrated in Fig. 1.

FIG. 1.

Suppression of the K⁺ uptake-defective phenotype of trk1Δ trk2Δ cells by mutations in HXT1, HXT3, and GAL2. trk1Δ trk2Δ cells containing wild-type or suppressor alleles were grown on permissive medium containing 100 mM KCl and 2% glucose (Glu) or 100 mM KCl and 2% galactose (Gal) prior to replica plating to the indicated test media. trk1Δ trk2Δ cells and wild-type cells containing the vector (pRS316) were used as negative and positive controls, respectively. The patches of cells were incubated at 30°C for 2 days before being photographed.

The HXT1-2 and HXT3-1 suppressor mutations alter the sequences of putative membrane-spanning domains of these glucose transporters. The HXT3-1 mutation results in the substitution of glycine for cysteine at amino acid 174, a position predicted to lie within the fourth putative transmembrane domain (TM4). The HXT1-2 mutation results in the substitution of tryptophan for arginine midway through putative TM11. The structural alteration of putative membrane-spanning domains suggested that the mutant glucose transporters suppress the phenotype of trk1Δ trk2Δ cells by conferring K⁺ permeability and increased uptake of the ion.

To determine whether other sugar transporters could be genetically targeted to yield similar suppressor mutations, the selection was repeated on potassium-limiting medium containing 2% galactose as the carbon source by using a trk1Δ trk2Δ strain that expressed the galactose/glucose transporter gene GAL2 (37, 62) from a centromeric plasmid. Among several independent spontaneous mutants, three showed dependence on the plasmid for suppression of the trk1Δ trk2Δ K⁺ uptake-defective phenotype. As expected, suppression by these GAL2 mutants was dependent on the presence of galactose and was not observed on glucose media (data not shown), since expression of GAL2 is induced by galactose and repressed by glucose (62). Figure 1 shows the suppression phenotype conferred by GAL2-1. DNA sequence analysis of GAL2-1 revealed that the suppressor mutation resulted in a substitution of phenylalanine for serine at amino acid position 412 within the ninth hydrophobic region. These results showed that, giving the appropriate conditions for expression, other hexose transporter genes could give rise to mutations capable of suppressing the trk1Δ trk2Δ phenotype.

Analysis of independent HXT3 suppressor alleles.

The hexose transporter mutants suppress the trk1Δ trk2Δ phenotype at least in part by increasing potassium uptake in these cells (see below). The three mutants isolated from three different hexose transporters exhibited an evolving pattern: each mutation resided within a putative TM. In order to reveal which regions of a glucose transporter can undergo mutations to confer K⁺ transport and in order to identify amino acids that might be important for glucose transport, a genetic selection was performed to isolate additional HXT3 mutants. trk1Δ trk2Δ cells (strain CY162) harboring a centromeric plasmid that expresses the wild-type HXT3 gene were selected for growth on K⁺-limiting medium (7 mM K⁺). Thirty-eight independent plasmid-linked suppressor mutants were isolated (see Materials and Methods). Upon reintroduction into the trk1Δ trk2Δ host, each of the mutant plasmids restored growth of the cells on 7 mM K⁺ medium. The HXT3 mutants conferred different degrees of suppression. Some alleles, e.g., HXT3-16 and HXT3-242, showed only weak suppression on 7 mM K⁺ medium, whereas others, e.g., HXT3-115 and HXT3-206, were able to confer growth on media that contained as little as 0.2 mM K⁺ (data not shown).

DNA sequence analysis of the 38 HXT3 suppressor alleles revealed that each contained a single mutation within the coding region (Table 1 and Fig. 2). Together with the HXT3-1 allele, they represent 25 distinct mutations that correspond to changes in 19 different codons. The HXT3-209 mutation changes a glycine codon at position 81 into a valine codon, and the HXT3-2 mutation changes an aspartic acid codon at position 79 into a tyrosine codon. Both of these mutations are located immediately adjacent to TM1 (Fig. 2). Each of the remaining 23 suppressor mutations alters an amino acid(s) residing in putative TMs. Suppressor mutations affecting TM2, TM4 through TM7, TM10, and TM12 were obtained. Thus, the suppressor mutations alter either the structure of putative TMs or the region in the first extracellular loop immediately adjacent to TM1.

TABLE 1.

Suppressor mutations in HXT1 and HXT3

Suppressor allele	Nucleotide (WT→mutant)a	Amino acid (WT→mutant)	TM affected	Glucose uptake
Substitutions
HXT3-2	GAT→TAT	Asp79→Tyr	1-2 Loop	−
HXT3-209	GGT→GTT	Gly81→Val	1-2 Loop	+
HXT3-207	GGT→TGT	Gly122→Cys	2	−
HXT3-156	TGT→CGT	Cys123→Arg	2	−
HXT3-1	GGT→TGT	Gly175→Cys	4	−
HXT3-23, -220, -225, -240, -241	GGT→AGT	Gly175→Ser	4	−
HXT3-169	GGT→GGT	Gly175→Val	4	−
HXT3-242	GGT→GAT	Gly175→Asp	4	−
HXT3-165	GCC→GTC	Ala182→Val	4	−
HXT3-206	CAA→AAA	Gln206→Lys	5	−
HXT3-213, -237	CAA→CGA	Gln206→Arg	5	−
HXT3-205	GTT→CTT	Val253→Leu	6	−
HXT3-16, -131	TCT→TAT	Ser330→Tyr	7	+
HXT3-119	CAA→CAC	Gln332→Thr	7	+
HXT3-231	TTG→TCG	Leu334→Ser	7	+
HXT3-130	GGT→AGT	Gly336→Ser	7	−
HXT3-210	GCT→ACT	Ala438→Thr	10	+
HXT3-122, -214, -222, -224, -238	GCT→GTT	Ala438→Val	10	−
HXT3-141, -217	GCT→GTT	Ala442→Val	10	−
HXT1-2	TGG→CGG	Trp473→Arg	11	−
HXT3-204, -212	TTC→ATC	Phe495→Ile	12	+
Deletions
HXT3-219	ATTGGTTGTGCCATTGGT→ATTGGT	Ile121GlyCysAlaIleGly→IleGly	2	−
HXT3-164, -233	TTCTTC→TTC	PhePhe341→Phe	7	−
HXT3-28	TTCTACTAT→TTCTAT	PheTyrTyr343→PheTyr	7	−
HXT3-5	TTCTACTAC→TTCTAC	PheTyrTyr491→PheTyr	12	+
Insertions
HXT3-114	GGT→GGTGGT	Gly336→GlyGly	7	−
HXT3-115	ATGGTT→ATGGTGGTT	MetVal500→MetValVal	12	+

^aUnderlined nucleotides are those inserted or deleted in the mutant sequence. WT, wild type. 

FIG. 2.

Twelve-TM model of Hxt3. Residues that form α helices traversing the membrane are represented by shaded circles. The amino acids shown are those of the wild-type transporter. Circled sites are those where suppressor mutations resulted in amino acid substitutions. Brackets indicate sites where the suppressor mutation resulted in deletion of the indicated amino acids. Solid circles are sites of insertions. Additional information about these mutations is summarized in Table 1.

Many of the HXT3 suppressor mutations obtained from this selection were single-base pair substitutions that confer an amino acid change. However, nearly 20% of the mutants contained multinucleotide insertions or deletions (Table 1). Each of the insertion mutations was an insertion of GGT to create GGT GGT, which resulted in insertion of glycine in TM7 (HXT3-114) or valine in TM12 (HXT3-115). Four distinct deletions were identified. Three removed one copy of two adjacent trinucleotide repeats to delete a phenylalanine codon (TTC) at position 341 or a tyrosine codon (CTA) at position 343 or 491. The fourth deletion mutant, HXT3-219, harbored a deletion of the 12 nucleotides ATT GGT TGT GCC from the wild-type sequence ATT GGT TGT GCC ATT GGT, resulting in the removal of four amino acids, Ile¹²¹GlyCysAla, from TM2. Thus, HXT3-219, too, involves the deletion of a repeated sequence, ATT GGT. These results revealed that a high percentage of the suppressor mutations are what appear to be template-dependent insertions or deletions: either insertions that generate trinucleotide repeats or deletions of existing repeated sequences.

The HXT3 suppressor mutations alter sites that are highly conserved among members of the hexose transporter family in S. cerevisiae (32). Most are completely conserved in transporters that have been shown to be capable of mediating glucose transport, which thus far include HXT1 through HXT4, HXT6, HXT7, and GAL2 (37, 52). Gal2 is the galactose transporter and, although normally not expressed on glucose, is fully capable of glucose transport when constitutively expressed (37). Only three sites identified by the HXT3 suppressor mutants are not absolutely conserved among these seven transporters: Gly81, Ala182, and Ser330. Hxt1 has a threonine at the position corresponding to Ala182; Gal2 has a serine and a methionine at positions corresponding to Gly81 and Ser330, respectively. In addition, suppressor mutations were identified in regions that are evolutionarily conserved among all sugar transporters. Gly¹⁷⁵ and Ala¹⁸² lie within the consensus motif (i/v)GlGvGgia(vl/av)sPmli(s/a) (lowercase indicates less-than-complete conservation) (2, 32) in TM4, Val²⁵³ lies within the consensus motif (l/v)PESP(ryy/qfl) (2, 32) in TM6, and Ser³³⁰, Gln³³², Leu³³⁴, Gly³³⁶, Phe³⁴⁰ and Tyr³⁴² are clustered within a highly conserved region in TM7. These results suggest that the suppressor mutations alter motifs that might be functionally important in sugar transporters.

Potassium uptake mediated by mutant Hxt3 transporters.

Suppression of the trk1Δ trk2Δ phenotype by HXT3 mutations was found to be inhibited by elevated but nontoxic concentrations of other cations. For example, growth of trk1Δ trk2Δ cells expressing each of the mutant HXT3 alleles on 7 mM K⁺ medium is significantly reduced in the presence of 25 mM Ca²⁺ (data not shown). This suggested that the mutant hexose transporters suppress the K⁺-deficient phenotype by increasing K⁺ uptake and that high concentrations of other cations can block or compete with the K⁺ transport.

To more directly assess the ability of a mutant glucose transporter to transport potassium, trk1Δ trk2Δ cells constitutively expressing either HXT3 or HXT3-23 from the ADH1 promoter were assayed for uptake of ⁸⁶Rb⁺. As shown in Fig. 3, whereas trk1Δ trk2Δ cells expressing the wild-type transporter exhibit essentially the same amount of ⁸⁶Rb⁺ uptake as cells harboring the vector alone, cells expressing the Hxt3-23 transporter exhibited a significant increase in ⁸⁶Rb⁺ uptake. The rate of Rb⁺ uptake was calculated based on time points within the nearly linear portion of the uptake curve (between 10 and 30 min). In cells harboring the vector or expressing wild-type Hxt3 transporter, Rb⁺ uptake occurred at a rate of 3.1 ± 0.2 nmol/min/mg (n = 4), whereas the rate in cells expressing the Hxt3-23 transporter was 6.0 ± 0.1 nmol/min/mg (n = 3). Furthermore, while ⁸⁶Rb⁺ uptake began to plateau in cells harboring the vector or the wild-type Hxt3 transporter, cells expressing Hxt3-23 showed continuous accumulation of the ion during the assay period. Thus, suppression of the trk1Δ trk2Δ phenotype by the mutant glucose transporters is correlated with increased K⁺ (Rb⁺) uptake.

FIG. 3.

⁸⁶Rb⁺ uptake mediated by HXT3-23. CY162 (trk1Δ trk2Δ) cells were transformed with plasmids expressing HXT3 or HXT3-23 under the control of the constitutive ADH1 promoter (pADH::HXT3 and pADHp::HXT3-23, respectively). Cells harboring the vector pRS316 were used as a negative control. Uptake assays were performed as described in Materials and Methods.

In addition to K⁺ and Rb⁺, the mutant glucose transporters appear to be able to transport Na⁺. Expression of the mutant HXT3 alleles in wild-type (TRK1 TRK2) cells causes hypersensitivity to Na⁺. Although higher concentrations of Na⁺ are toxic, wild-type S. cerevisiae cells can grow on media containing as much as 1 M NaCl even under low-potassium conditions (2 mM) (41, 48). Sensitivity to Na⁺ can be enhanced when uptake is increased through various membrane proteins (48, 69). As shown in Fig. 4, wild-type cells expressing HXT3-156 or HXT3-23 exhibit significantly slower growth on media containing supplemental Na⁺ (400 mM) compared to cells expressing the wild-type HXT3 gene or to cells harboring the vector alone. The expression of each of the mutant HXT3 alleles resulted in increased Na⁺ sensitivity (data not shown). The severity of Na⁺ hypersensitivity conferred by a particular HXT3 mutant allele correlated directly with its ability to rescue growth of trk1Δ trk2Δ cells on medium containing low concentrations of potassium; i.e., stronger alleles conferred both stronger growth on K⁺-limiting media and greater Na⁺ hypersensitivity. These results strongly suggest that although glucose transporters are normally highly selective, single mutations can convert them into nonselective cation transporters.

FIG. 4.

Na⁺ hypersensitivity conferred by HXT3-23 and HXT3-156. Wild-type cells (strain R757) transformed with vector or plasmids expressing HXT3-156, HXT3-23, or wild-type HXT3 were grown to log phase in standard synthetic medium. A serial dilution of each culture was spotted on LS medium supplied with 400 mM Na⁺ and 2 mM K⁺ to test for sodium toxicity. Standard synthetic medium (YNB) containing approximately 7 mM K⁺ and approximately 2 mM Na⁺ (17) was used for permissive growth conditions. Cells were incubated at 30°C for 5 days on LS medium containing 400 mM Na⁺ and for 2 days on YNB before being photographed.

The effects of the suppressor mutations on glucose transport.

To investigate the effects of the suppressor mutations on the normal function of Hxt3, we assessed the abilities of the mutant transporters to take up glucose by testing their abilities to confer growth of a glucose transport-deficient strain on medium containing glucose as the sole carbon source (snf3Δ hxt1Δ hxt2Δ hxt3Δ hxt4Δ hxt6Δ hxt7Δ gal2; strain HY133 [37]). Expression of wild-type HXT1 or HXT3 and of some of the mutant HXT3 alleles was sufficient to restore growth of these cells on glucose. However, HXT1-2 and most of the HXT3 mutants appeared to be strongly impaired for glucose transport. They failed to confer growth on the snf3Δ hxt1Δ hxt2Δ hxt3Δ hxt4Δ hxt6Δ hxt7Δ gal2 recipient (HY133) on 2% glucose (Fig. 5 and Table 1). Although unregulated glucose influx can lead to glucose poisoning and cell death (12, 19), the inability of a mutant transporter to restore growth of HY133 on glucose is due to a lack of glucose transport and not to uncontrolled glucose influx, since these mutants do not confer glucose sensitivity in wild-type or trk1Δ trk2Δ cells (Fig. 1 and 6 and data not shown). Because the mutant transporters are capable of mediating potassium transport, the severe impairment in glucose transport caused by the suppressor mutations suggests that the mutated residues are essential for glucose transport per se. The inability of some of the mutants to transport glucose also suggested that suppression of the trk1Δ trk2Δ phenotype was not due to an indirect effect resulting from increased sugar uptake and metabolism.

FIG. 5.

Abilities of representative HXT3 suppressor alleles to restore growth of glucose transport-deficient cells on glucose. HY133 (snf3Δ hxt1Δ hxt2Δ hxt3Δ hxt4Δ hxt6Δ hxt7Δ) cells transformed with suppressor alleles of HXT3 were grown on synthetic medium containing 3% glycerol and 2% ethanol (GE) and were replica plated onto 2% glucose (Glu) medium containing antimycin A (1 μg/ml) to test for the ability to utilize glucose. Cells containing vector (pRS316) or the wild-type HXT3 allele were used as negative and positive controls, respectively. Patches of cells were incubated at 30°C for 3 days before being photographed.

FIG. 6.

Glucose-dependent and -independent suppression conferred by mutant HXT3 alleles. CY162 (trk1Δ trk2Δ) cells transformed with plasmids containing representatives of HXT3 suppressor alleles and plasmids expressing these alleles under the control of the constitutive ADH1 promoter were grown on medium supplemented with 100 mM KCl and were replica plated onto medium containing 7 mM KCl and the indicated carbon sources. Glu, 2% glucose; GE, 3% glycerol plus 2% ethanol. trk1Δ trk2Δ cells and wild-type cells containing vector (pRS316) were used as negative and positive controls, respectively. Patches of cells were incubated at 30°C for 3 days before being photographed.

Among the 25 different HXT3 mutants isolated, only 8 appeared capable of mediating glucose transport (Fig. 5 and Table 1). Although each of these was able to confer growth of HY133 on glucose, the rate of growth was allele specific. The ability of HXT3-16, HXT3-115, HXT3-119, HXT3-210, or HXT3-231 to restore glucose-dependent growth of HY133 was similar to that mediated by the wild-type HXT3 gene. In contrast, HXT3-5, HXT3-204, and HXT3-209 conferred significantly weaker growth (Fig. 5 and data not shown). Interestingly, while the Ala⁴³⁸→Thr mutation in TM10 (HXT3-210) retains glucose transport, a substitution of a valine at the same position (HXT3-122) appears to abolish this activity.

Each of the three mutants that harbor an alteration at TM12, including an insertion and a deletion, retain glucose transport activity (Table 1 and Fig. 2), indicating that this region is not critical for glucose transport. Rather, since the mutants gained the ability to transport K⁺, this region may play an important role in the overall integrity of the transporter. Each of the remaining five mutants that exhibit glucose transport activity is located in a region shown to be responsible for substrate recognition or translocation by other sugar transporters: TM10 is critical for galactose recognition by Gal2 (25); the first extracellular loop is important for substrate specificity by Chlorella hexose/H⁺ symporters (68); and TM7 has been suggested to form part of the glucose channel and to be involved in substrate translocation (71). The importance of these regions in glucose transport is supported by the isolation of suppressor mutations that abolish glucose transport in these regions (Table 1). Asp⁷⁹→Tyr (loop 1), Gly³³⁶→Ser (TM7), and Ala⁴³⁸→Val and Ala⁴⁴²→Val (TM10) result in severe impairment of glucose transport, indicating that these residues are critical for glucose binding or translocation. Mutations in these regions that retain glucose transport activity may expand the glucose binding and translocation unit of the transporter to allow ion permeation without prohibiting glucose transport.

Potassium transport via mutant Hxt3 transporters is not coupled to glucose transport.

As indicated above, suppression of the trk1Δ trk2Δ phenotype by mutations in HXT1 and HXT3 was found to be dependent on glucose. Although the mutant HXT3 alleles are able to restore growth of trk1Δ trk2Δ cells on medium containing 7 mM potassium in the presence of 2% glucose, no suppression was observed when glycerol and ethanol were supplied as the carbon source (Fig. 6 and data not shown). Suppression conferred by HXT1-2 showed a similar dependency on glucose (data not shown). This glucose dependency is likely due to the lack of expression of HXT1 and HXT3 in the absence of glucose, since the expression of these two genes is greatly induced by glucose (50). However, glucose-dependent suppression of the trk1Δ trk2Δ phenotype could also reflect the glucose dependency of K⁺ transport mediated by the mutant transporters.

To determine if potassium uptake conferred by the HXT3 mutants is inherently coupled to glucose transport, the requirement for glucose by several constitutively expressed mutant HXT3 alleles was tested. None exhibited glucose dependency for suppression when expressed from the promoter of ADH1, regardless of their abilities to transport glucose; i.e., they suppressed the trk1Δ trk2Δ phenotype in the absence of glucose. In addition, no significant differences were observed in the degrees of suppression of the trk1Δ trk2Δ phenotype on 2% glucose compared to those on glycerol and ethanol (Fig. 6). Therefore, potassium permeation through the mutant glucose transporters is not obligatorily coupled to the binding or transport of glucose.

DISCUSSION

We have identified mutations in the hexose transporter genes HXT1, HXT3, and GAL2 that confer growth on K⁺ transport-defective cells (trk1Δ trk2Δ) on K⁺-limiting media. Although transporters of other substrates from both yeast and bacteria have been implicated in K⁺ uptake (11, 16, 23, 70), these hexose transporters are normally highly selective (reviewed in reference 6). Our results reveal that hexose transporters can be converted into general cation transporters by single mutations at residues that lie within or immediately adjacent to putative membrane-spanning domains. These mutant proteins appear to be capable of mediating transport of K⁺, Na⁺, and possibly Ca²⁺.

Although we have not ruled out the possibility that the suppressor mutations in the hexose transporters confer increased K⁺ uptake through an indirect mechanism, the data suggest that this is very unlikely. First, suppressor mutations were obtained in at least three different transporters. An indirect role would require that each transporter can be made capable of activating a cryptic ion transporter. Second, the observation that dominant gain-of-function mutations could be obtained at a wide variety of positions within Hxt3 also argues against an indirect effect, since each would have to activate the putative ion transporter. Third, the ability to confer K⁺ uptake does not require an alteration of glucose transport. Some of the suppressor mutations in HXT3 abolished glucose uptake, while others had no significant effect. Finally, the ion transport activity that results from the suppressor mutations is nonselective. An indirect role resulting in increased uptake for K⁺, Na⁺, and Ca²⁺ would require either the simultaneous activation of multiple transporters or the existence of a novel cation transporter in S. cerevisiae. Therefore, we strongly favor the hypothesis that the suppressor mutations at HXT1, HXT3, and GAL2 confer the ability of these transporters to accommodate the nonselective passage of cations.

The mutant sugar transporters do not appear to be converted into obligate cation/glucose symporters. Although some members of the sugar transporter superfamily are ion/sugar symporters (for reviews, see references 3 and 27), glucose transport in S. cerevisiae is apparently not coupled with protons or other cations (for a review, see reference 6). In addition, K⁺ transport mediated by the mutant sugar transporters does not require the presence of glucose. Thus, the conformational changes that presumably occur during glucose translocation are not necessarily required for potassium transport by the mutant proteins.

Among 18 putative hexose transporter genes identified in S. cerevisiae, 7, HXT1 through HXT4, HXT6, HXT7, and GAL2, have thus far been shown to be capable of mediating glucose transport (33, 36). The remaining HXT genes might encode hexose transporters that are not expressed under laboratory conditions, or they might encode transporters of other solutes. Suppressor mutations that allow K⁺ uptake were mapped only to the HXT1 and HXT3 loci in the initial selection. The absence of analogous mutations in the other functional glucose transporter genes is likely due to the fact that they are not expressed under the selection conditions. HXT2, HXT4, HXT6, HXT7, and GAL2 are each repressed by the concentrations of glucose (2%) contained in the selection medium (37, 50, 52, 62, 67). In fact, we have isolated a suppressor allele of HXT6 that is capable of suppressing the trk1Δ trk2Δ phenotype on medium containing glycerol and ethanol but not on medium containing 2% glucose (unpublished results). Similarly, when the selection was carried out on galactose, cells with suppressor mutations in the galactose transporter gene GAL2 were isolated. Mutations in other transporters, including amino acid permeases (70) and an inositol transporter (manuscript in preparation), can also confer suppression of the trk1Δ trk2Δ phenotype by restoring K⁺ uptake. Therefore, the ability to accommodate the transport of K⁺ upon simple mutation may be a general property of nutrient transporters. Nevertheless, the observation that only a few transporters are represented among many independent suppressor mutations analyzed to date argues that mutationally induced uptake of K⁺ is not likely to be an intrinsic property of all polytopic membrane proteins in S. cerevisiae.

In a selection designed to identify which regions of Hxt3 could be altered to confer K⁺ transport, mutations were found that altered seven different TMs and the first extracellular loop. Whether similar mutations can occur in the remaining five TMs is unknown. Interestingly, the two suppressor mutations obtained in Hxt1 and Gal2 map to TMs different from those identified in Hxt3. This may reflect the fact that the mutational analysis of HXT3 was not exhaustive.

Among the 25 distinct mutations identified in Hxt3, 17 severely impair glucose transport. The fact that these suppressor mutants exhibit a dominant gain-of-function phenotype (K⁺ uptake) suggests that the mutations that abolish glucose transport identify sites that are critical for glucose transport per se and not for protein expression or localization. In this regard the structure and function relationships revealed by the effects of the suppressor mutations on glucose transport confirm and significantly extend those described by others. For example, of the many sites conserved among mammalian glucose transporters that have been analyzed by site-directed mutagenesis, only eight have been determined to be critical for glucose transport (14, 22, 26, 44–46, 56, 60, 64). Of these sites, five are conserved in the S. cerevisiae hexose transporter gene family and three are represented among our collection of suppressors (Gln¹⁶¹ and Tyr³⁴³ in Hxt3 and Trp⁴⁷³ in Hxt1). In agreement with the mammalian data, mutations at these three sites abolished glucose uptake. Similarly, the site in the Hup family of Chlorella hexose transporters analogous to Asp⁷⁹ in Hxt3 is conserved and has been shown to be essential for hexose transport (9). The abolition of glucose uptake by point mutations at Gly¹²², Cys¹²³, Gly¹⁷⁵, Ala¹⁸², Val²⁵³, Gly³³⁶, Ala⁴³⁸, and Ala⁴⁴² and by a single-codon deletion mutation (Phe³⁴¹) reveals eight new sites that are critical for glucose transport. Four of these, Gly¹²², Gly¹⁷⁵, Gly³³⁶, and Phe³⁴¹, are conserved in both the mammalian and the fungal transporters.

The profile of HXT3 mutations was very unusual. Seven of 38 independent mutations contained an insertion or deletion of multiple base pairs. Furthermore, these mutations appeared to be template dependent: six were insertions or deletions of trinucleotide repeats, and one was a deletion of 12 nucleotides that are bounded on either side by a 6-nucleotide repeated sequence. Although much smaller in scale, the trinucleotide insertions are reminiscent of the genomic expansions that occur in the triplet repeat sequences observed in genes associated with hereditary diseases (66). Likewise, a large number of insertions or deletions of partially repeated multiple-base pair sequences have been reported in somatic mutations in the p53 (7, 15, 18, 24, 28) and adenomatous polyposis coli (APC) (10, 42, 43, 47, 51) genes isolated from different human tumors and in germ line mutations that cause diseases such as retinitis pigmentosa (4), junctional epidermolysis bullosa (40), and hypertrophic cardiomyopathy (65).

The frequency of spontaneous multinucleotide insertion or deletion events is significantly higher than that reported by others. No insertions and only seven deletions were obtained among 322 spontaneous loss-of-function alleles analyzed in the S. cerevisiae tRNA suppressor gene SUP4-o (34), suggesting that in wild-type cells, under normal growth conditions, the ratio of spontaneous multinucleotide insertions or deletions to single-nucleotide substitutions is about 2%. In another study, no insertions or deletions were obtained in a total of 68 independent spontaneous mutants analyzed at three independent loci (61). In addition, since our selection required the expression of a functional transporter, insertions or deletions that cause frameshifts would not have been obtained. Thus, the ratio of insertions or deletions to single-base pair substitutions may be significantly underrepresented in our collection of mutants.

It is possible that the selection bias for the unique gain-of-function mutations has revealed a frequency of insertions and deletions that has simply not been detected by other mutant hunts. Alternatively, the potassium-limiting conditions under which the selection was performed might be responsible for an increased frequency of such mutations. After replica plating to potassium-limiting media, trk1Δ trk2Δ cells are able to go through several cell divisions due to the pool of stored K⁺. However, during this period of growth the internal K⁺ concentration decreases significantly. Nucleotide insertions and deletions might occur more frequently under these conditions if DNA replication or repair mechanisms require higher K⁺ concentrations for proper function. Although potassium is an essential ion (35, 39, 53) and high doses in mammalian-cell culture can induce mutations (20, 57), little is known about the effect of potassium deficiency on DNA replication or repair. Tishkoff et al. recently reported a high frequency of insertions or deletions of repeated sequences (73% at one locus and 79% at another locus) in cells with mutations affecting DNA repair (rad27Δ) (61). This raises the possibility that suboptimal concentrations of intracellular potassium might inhibit Rad27-mediated DNA repair activity. If potassium insufficiency is indeed the cause of the insertion and deletion mutants that arose from the suppressor selection, this would reveal a new physiological role for K⁺ in the maintenance of genomic integrity.