OUP user menu

Conservation and dispersion of sequence and function in fungal TRK potassium transporters: focus on Candida albicans

Manuel Miranda, Esther Bashi, Slavena Vylkova, Mira Edgerton, Clifford Slayman, Alberto Rivetta
DOI: http://dx.doi.org/10.1111/j.1567-1364.2008.00471.x 278-292 First published online: 1 March 2009


TRK proteins – essential potassium (K+) transporters in fungi and bacteria, as well as in plants – are generally absent from animal cells, which makes them potential targets for selective drug action. Indeed, in the human pathogen Candida albicans, the single TRK isoform (CaTrk1p) has recently been demonstrated to be required for activity of histidine-rich salivary antimicrobial peptides (histatins). Background for a detailed molecular investigation of TRK-protein design and function is provided here in sequence analysis and quantitative functional comparison of CaTrk1p with its better-known homologues from Saccharomyces cerevisiae. Among C. albicans strains (ATCC 10261, SC5314, WO-1), the DNA sequence is essentially devoid of single nucleotide polymorphisms in regions coding for evolutionarily conserved segments of the protein, meaning the four intramembranal [membrane–pore–membrane (MPM)] segments thought to be involved directly with the conduction of K+ ions. Among 48 fungal (ascomycete) TRK homologues now described by complete sequences, clades (but not the detailed order within clades) appear conserved for all four MPM segments, independently assessed. The primary function of TRK proteins, ‘active’ transport of K+ ions, is quantitatively conserved between C. albicans and S. cerevisiae. However, the secondary function, chloride efflux channeling, is present but poorly conserved between the two species, being highly variant with respect to activation velocity, amplitude, flickering (channel-like) behavior, pH dependence, and inhibitor sensitivity.

  • Candida albicans
  • potassium transport
  • chloride channeling
  • TRK proteins
  • MPM motifs
  • sequence dispersion


Whereas coupled exchange of potassium (K+) for sodium (Na+), mediated by a P-type ATPase in cell plasma membranes, is the principal means for K+ accumulation by animal cells, several quite different kinds of transporters impel K+ accumulation in plants, fungi, and bacteria. The fact that resting membrane voltages (Vm) in non-animal systems are often very negative to the K+ equilibrium voltage (EK; see Slayman, 1982) means that pure channel structures can facilitate net K+ uptake and accumulation in many circumstances. ATP-coupled K+-influx pumps also exist, for example the Kdp system in Escherichia coli (Epstein, 1985; Siebers & Altendorf, 1993), but the major devices for K+ accumulation are gradient-driven coupled-ion transporters and uniporters. Best known of these are the so-called TRK and HAK proteins, which – in plants and fungi – are homologues of bacterial Ktr and Kup transporters, respectively (Stumpe, 1996).

The TRK proteins had been assumed to underlie high-affinity K+ accumulation in fungi such as Neurospora (Rodriguez-Navarro, 1986; Blatt, 1987; but see Haro, 1999). They were also recognized, >10 years ago, as having sequence homology with bona fide K+ channels (Stumpe, 1996; Jan & Jan, 1997; Durell, 1999), and were subsequently demonstrated to fold as internal tetramers, thus forming a channel-like pathway for K+ transit (Durell & Guy, 1999; Kato, 2001; Zeng, 2004). The selectivity of this pathway, as explored in both higher plants (Arabidopsis thaliana: Diatloff, 1998; Liu, 2000) and bacteria (Vibrio alginolyticus: Tholema, 1999, 2005) has been shown to depend critically on specific amino-acid residues, whose counterparts in KcsA – the crystallized K+ channel from Streptomyces lividans (Doyle, 1998) – contribute to actual K+-binding sites.

Comparisons among the first few fungal TRK sequences emerging from the genome data revealed an unexpected degree of conservation for residues expected to reside at the surface of the folded structure. This led Durell & Guy (1999) to suggest oligomerization of folded monomers into tetrads, within cell plasma membranes, resulting in an overall configuration similar to that for aquaporins.

Subsequent patch-clamp experiments, on the yeast Saccharomyces cerevisiae and several mutant strains thereof (Bihler, 1999; Kuroda, 2004), identified strange ionic currents mediated via the two TRK proteins in that organism (Trk1p and Trk2p; S. cerevisiae has no homologue of the HAK gene). These currents are not visible as single-channel events, but do display macroscopic channel-like properties in whole-cell records: they are strongly dependent on extracellular pH (pHo), with a ‘gating’ voltage of −267 mV at pHo=7.5 and −157 mV at pHo=4.5; they are very small for Vm's positive to −100 mV, but can be more than 10-fold larger than expected transporter currents at large negative voltages; and they have proven proportional to the intracellular (pipette) chloride (Cl) concentration at all values of pHo. These currents are evidently carried by Cl efflux, and detailed kinetic analysis has suggested that they flow through the central ‘pore’ in the assembled tetrads of Trk1p/Trk2p (Kuroda, 2004; Rivetta, 2005).

Because of the direct medical importance of the yeast Candida albicans, particularly for immunocompromised patients, and because of the importance of K+ regulation for multiple cellular functions, we undertook to clone the TRK gene(s) in Candida by expression in a double-knockout strain of Saccharomyces, and to characterize the TRK protein(s) in Candida itself. Two related events occurred in the same time-frame: (1) sequencing of the Candida genome, which yielded a defective sequence for the single TRK gene (http://www.candidagenome.org/), and (2) discovery that this K+ transporter is a critical element in the killing of Candida by the oral antimicrobial peptide, histatin 5 (Baev, 2004). In the present study, we report analysis of the TRK1 gene sequence, comparative analysis of the protein sequence across fungal species, and a partial physiological characterization of the protein CaTrk1p, in C. albicans.

Materials and methods

Strains and maintenance

HY483, a double-TRK knockout strain of S. cerevisiae (MATαleu2-3,112 ura3-1 trp1-1 his3-11,15 ade2-1 can1-100 GAL+ SUC2+ trk1Δ∷HIS3 trk2Δ∷HIS3; S288C background; Ko & Gaber, 1991) was used for expression cloning of the Candida TRK1 gene. The strain was maintained routinely on plates in YPAD +100 mM KCl at 30 °C, and was grown for transformation in liquid YPAD +50 mM KCl (Sherman, 1991; Kaiser, 1994). A standard C. albicans library, prepared from strain American Type Culture Collection (ATCC) 10261 in the centromeric vector YCp50 (Rose, 1987; Smith, 1992) was amplified in E. coli strain DH5αf1. Transformation of HY483 was carried out with the BIO-101 kit for yeast (MP Biochemicals, Irvine, CA) plus 10 μg of the C. albicans library DNA. The plasmid DNA from recovered colonies was reisolated using the yeast Teeny-prep protocol (http://www.bs.jhmi.edu/MBG/boekelab/Resources/YGM/Protocols/TeenyPrepGenDNA.html). After functional confirmation (growth on low K+), the TRK1 insert was subcloned into YCplac33 (Gietz & Sugino, 1988), for later use.

Other yeast strains, used for functional comparisons, were S. cerevisiae PLY232 (MATahis3-Δ200 leu2-3,112 trp1-Δ901 ura3-52 suc2-Δ9; Bertl, 2003), BS202 (MATaade2-1 can1-100 his3-11,15 leu2-3,112 trp1-1 ura3-1 lys2-ΔNheI; Smith & Roeder, 2000); and C. albicans SC5314 (provided by Dr P.T. Magee, University of Minnesota), CAI4 (Δura3imm434/Δura3imm434;Fonzi & Irwin, 1993), SGY243 (ade2/ade2Δura3ADE2/Δura3ADE2; Kelley, 1987), CaTK1 (Δura3imm434/Δura3imm434Δtrk1/TRK1; Baev, 2004), and DBT3 (Δura3imm434/Δura3imm434 Δtok1/Δtok1; Baev, 2003).

Ion flux measurements

Functional characterization of the endogenous C. albicans TRK protein was carried out in two ways: first, by measurement of chemical fluxes in suspensions of intact C. albicans yeast cells, using rubidium (especially 86Rb+) as a plausible label for K+ influx (Love, 1954; Armstrong & Rothstein, 1967; Läuchli & Epstein, 1970; Aiking & Tempest, 1977; Rodriguez-Navarro, 2000) and second, by measurement of TRK-dependent ion currents via patch-clamping of yeast-cell spheroplasts.

For chemical flux measurements, C. albicans (strain CAI4) was grown in shaking cultures at 37 °C to OD600 nm≈1 (c. 3 × 107 cells mL−1), in commercial YNB medium (QBIOgene, Irvine, CA) plus 200 μM uridine and 140 mM KCl. The resulting log-phase cells were harvested by centrifugation (500 g for 5 min), washed twice with glass-distilled water, and then subjected to a period of general starvation, patterned on that formerly used to ‘stabilize’S. cerevisiae for ion-flux experiments (Armstrong & Rothstein, 1964; Eddy & Hopkins, 1989). The washed cells were resuspended (at OD≈1) in 140 mM glucose plus 1 M sorbitol and incubated at room temperature (c. 23 °C) for 5 h on a rotary shaker (250 r.p.m.). The resulting starved cells were washed twice, resuspended in transport buffer (50 mM Tris-succinate, pH 5.9, plus 140 mM glucose) at a density of 5 × 08 cells mL−1, and equilibrated for 15 min. Rubidium uptake was initiated by injecting 1 mL of this suspension with 25 μL of transport buffer containing 43 mM RbCl (final concentration=1.07 mM) and 0.1 μCi of 86Rb+. Labeled cells were then harvested at intervals, in 200-μL aliquots, by rapid filtration on Durapore membranes (0.45 mm pore diameter; Millipore Corp., Bedford, MA), and rinsed three times with 2 mM MgCl2 to flush out the extracellular 86Rb+. Pellets and filters were immersed in Ecoscint fluid (National Diagnostics, Atlanta, GA) and counted on a Beckman-Coulter scintillation counter (model LS6500; Fullerton, CA). Data were collected as counts min−1 per 108 cells, and converted to mM (mmol L−1 cell volume) via the measured specific activity plus a standard cell volume of 47 fL per cell (Baev, 2002).

Transport (influx) proved essentially linear for the first c. 5 min of sampling, and sampling was routinely carried out for 3 min.

Patch-clamp measurements

The whole-cell ‘patch’-clamp technique was used, slightly modified from the standard methods for Saccharomyces (Bertl, 1998; Baev, 2004). Cells were grown in log-phase cultures as described above, but in YPD medium, washed twice, and resuspended (at OD≈1) in 3 mL of 50 mM KH2PO4 brought to pH 7.2 with KOH, plus 25 mM β-mercaptoethanol. These suspensions were incubated on a slow orbital shaker (64 r.p.m.), for 30 min at 30 °C, then recentrifuged, and resuspended in 6 mL of the same buffer, plus 3.6 U of zymolyase 20T (ICN Biomedicals Inc., Irvine, CA), and incubated for 45 min at 30 °C. The resulting spheroplasts were spun down (500 g for 5 min), gently resuspended in stabilizing buffer1 and incubated stationary, at room temperature (c. 23 °C) until use. A single batch of spheroplasts could be used for patch recording over a 6–8-h period. For actual recording, 1–10 μL of stabilizing suspension was injected into c. 700 μL of sealing buffer2, gently mixed, and then allowed to settle for 10 min in the recording chamber, so that a small number of spheroplasts adhered lightly to the chamber bottom.

Patch pipettes were manufactured as described for Saccharomyces (Bertl, 1998) and filled with an artificial intracellular buffer3. A reference electrode, consisting of a chlorided silver wire (Ag–AgCl) immersed in 1 M KCl, was connected to the efflux-end of the recording chamber via a 1 M KCl–agar bridge.

Light suction on a patch pipette, placed near a clean spheroplast on the chamber bottom, would usually draw the cell onto the pipette tip. With very light further suction, a seal of 10–35 GΩ would normally develop within 4–6 min. The whole-cell configuration was obtained by breaking the membrane patch in the pipette tip, via a brief high-voltage pulse (c. 750 mV for 100 μs). Before recording, a 10-min period was allowed for equilibration between pipette contents and the spheroplast cytoplasm.

The three solutions noted above were as follows: (1) stabilizing buffer=220 mM KCl, 10 mM CaCl2, 5 mM MgCl2, 5 mM 2[N-morpholino]ethanesulphonic acid (MES) titrated to pH 7.2 with Tris base, 11 mM glucose, and 230 mM sorbitol; (2) sealing buffer=150 mM KCl, 20 mM CaCl2, 5 mM MgCl2, 1 mM MES titrated to pH 7.5 with Tris base, and 140 mM sorbitol; and (3) intracellular buffer=175 mM KCl, 1 mM EGTA, 0.15 mM CaCl2 (free Ca2+=100 nM), 4 mM MgCl2, and 4 mM ATP titrated to pH 7.0 with KOH.

A standard staircase of voltage clamp pulses, covering the Vm range of +100 mV to −180 mV, was adopted to generate current–voltage data. Stimulus delivery (the voltage pulses), current recording, and preliminary data analysis were carried out via an EPC9 patch-clamp amplifier (HEKA Elektronik, Lambrecht, Germany) controlled by a PowerMac G4 Computer (Apple Computer Inc., Cupertino, CA), as already described (Bertl, 1998). Data were collected at 2 kHz, filtered at 250 Hz, corrected for nonspecific leakage currents (Kuroda, 2004), and analyzed in detail via Microsoft excel and/or igor pro software (WaveMetrics Inc., Lake Oswego, OR).


Cloning of the C. albicans K+ transporter gene TRK1

Transformation of S. cerevisiae HY483 with the C. albicans DNA library and selection of transformants were performed in a medium lacking uracil but containing 50 mM K+ (SC+50K; see Sherman, 1991; Kaiser, 1994), then replicated to a low-K+ medium (SC without added K+; actual [K+]o=5.7 mM). Five independent clones were obtained, and the plasmid DNA was isolated (Teeny-prep recipe; see Materials and methods) and transferred to E. coli. Only two plasmids proved capable of supporting growth on the low-K+ medium, after retransformation of HY483. Restriction digests of those two plasmids yielded similar patterns with all enzymes tested, and were consistent with the original library construct. One clone (designated YCp50-TRK) having an c. 8-kb insert was selected for subcloning; the 8.0-kb fragment was excised with HindIII–BamH1, and then ligated into HindIII–BamHI-digested YCplac33 (Gietz & Sugino, 1988). The resulting construct was amplified in E. coli, reisolated, and transformed into yeast strain HY483 in order to certify its complementation of K+ transport in Saccharomyces.

Figure 1 demonstrates complementation, by support of robust growth of HY483 on K+ concentrations as low as 0.3 mM, which compares well with wild-type strains of both Saccharomyces and Candida. The entire YCplac33-TRK plasmid was then sequenced. In addition, the TRK1 segments from two other strains of C. albicans were sequenced as controls for comparison with the Candida genome databases.


Drop test to demonstrate that TRK1 from Candida albicans complements the K+-transport deficit in Saccharomyces cerevisiae deleted of both TRK1 and TRK2. The central experiment is represented in columns 3, 4, and 5. Strain HY483 is the TRK1,2ΔΔ strain provided by Ko & Gaber (1991). Untransformed (column 5), or transformed by the empty vector (YCplac33; column 4), HY483 does not grow robustly at [K+]o < ∼30 mM. Transformation by YCplac33-CaTRK1 (column 3) confers robust growth at [K+]o as low as 0.3 mM, nearly equivalent to wild-type Saccharomyces (strain PLY232, column 6), to wild-type Candida (strain CAI4; column 1), or to Candida deleted of a single allele of TRK1 (strain CaTK1; column 2). All strains were grown overnight (to OD600 nm∼2) at 30°C in YPD+50 mM KCl, harvested by centrifugation, washed twice in sterile glass-distilled water, resuspended at OD 1, and serially diluted to give concentrations of roughly 107, 106, 105, and 104 cells mL−1. Single drops (7 μL) were then spotted onto agar containing K+-free low-salt medium (recipe L86: Ramos, 1985; Gaber, 1988) supplemented with KCl, as indicated at the bottom of each panel. Plates were incubated for 2 days at 30°C, and then recorded on a digital scanner.

DNA sequence variations

For comparative purposes, our C. albicans TRK1 sequence from strain ATCC 10261 (NCBI database # AF267125) has been taken as the default sequence. This sequence includes the TRK1 ORF plus 5′ (937 bp) and 3′ (666 bp) untranslated regions (UTRs), as listed in Supporting Information, Fig. S1A, along with the translation provided by the curators (Fig. S1B). Single nucleotide polymorphism (SNP) changes between ATCC 10261 and the genome sequences of strain SC5314 or WO-1 are summarized in Table 1 (listing the shaded residues in Fig. S1). Annotations for Assembly 21, strain SC5314, indicate a 5′ UTR for TRK1 (orf19.600c) much longer than 937 bases, but a 3′ UTR of only 74 bases, followed by the coding sequence for a small ribonuclear protein (orf19.603w). The combined UTRs sequenced, 1011 bases, contain 20 SNPs, or c. 2%, which are roughly equally distributed among the three strains when the majority residue at each site is taken as reference. We assume such variations to be random.

Table 1 Summary of polymorphisms in Candida albicans TRK1

A measure of ‘typical’ nucleotide variability, for referencing Table 1, was obtained by comparing a region of the genome-database sequences for SC5314 and WO-1, spanning from 6000 bases upstream through 6000 bases downstream of TRK1. This region in chromosome R, described especially for SC5314, includes five more putative ORFs with a total coding sequence of 8087 bases, five noncoding intervals with 3250 bases, and 88% of the centromere (3945 bases). Single-base changes are found at 1% of residues in the noncoding intervals, which is not significantly different from 2% in the TRK1 UTRs, combined for the three strains. (The centromere region is more variable, however, with 2.8% of sites differing between SC5314 and WO-1.)

Within the coding sequence itself, SNPs are less frequent, occurring at 26 sites out of 3180 bases, or c. 0.8%, which compares with 0.5% of residues in all six ORFs, between SC5314 and WO-1. Furthermore, the SNP variations within the TRK1 ORF are nonrandomly distributed in at least two respects. First, from strain to strain: four changes from majority in WO-1, four in ATCC 10261, and 18 in SC5314. Second, location within the gene: 25 of the 26 identified SNPs occur in codons for putative cytoplasmic residues (viz., 693 amino acids out of 1059 total), regions that are very poorly conserved across fungal species. The three-amino acid deletion in SC5314 (486-Asp.Asp.Asp.-488) also maps to the major cytoplasmic loop of the protein. Only a single SNP, 2364T in WO-1, maps within the transmembrane or extracellular segments of the protein, which are well conserved across fungal species. Three apparent SNPs in SC5314 have proven to be sequencing errors in the genome database (see boxed residues in Table 1; Fig. S1). Only four SNPs in SC5314 TRK1 and two in WO-1 TRK1 are nonsilent mutations. All these map to unconserved cytoplasmic segments of the protein, where they are predicted to have little or no effect on function. Finally, the silent mutation at base 465 (A/G, Table 1) has been identified as SNP marker 1772/2368 in the Candida SNP map constructed by Forche (2004).

Amino acid conservation across species

As noted in the Introduction, TRK proteins in plants, fungi, and bacteria are sequence-similar to the selectivity-filter core of K+ channels, and have been postulated to fold in a similar manner. This folding is shown in the bead diagram of Fig. 2, by the clusters just below the membrane–pore–membrane (MPM) numbers (#1, #2, #3, and #4). The index of sequence mutability (μ) across fungal species (calculated by Dr H.R. Guy, National Cancer Institute) is represented in Fig. 2 by colors according to the figure key, with red designating best conserved (least mutable), gray designating very poorly conserved, and colorless designating the absence of conservation.


Representation of high sequence conservation within the MPM segments of TRK proteins. Whole-protein alignments and index-of-conservation calculations were carried out by H.R. Guy according to the procedures described by Durell, (1999) and Shrivastava (2004), for the first 19 fungal TRK sequences obtained from genomic data: Candida albicans, Aspergillus nidulans (two isoforms), Debaryomyces occidentalis, Ashbya gossypii, Gibberella zeae (three isoforms), Kluyveromyces lactis, Magnaporthe grisea (three isoforms), Neurospora crassa, Podospora anserina, Schizosaccharomyces pombe (two isoforms), Saccharomyces cerevisiae (two isoforms), and Saccharomyces uvarum. The triplet diagonal arrays designate α helices, and the extended doublets designate β strands, predicted by means of the predict protein software, available at http://www.expasy.org. The bead clusters directly below each MPM number (#1, #2, #3, and #4) represent the pore loops, with each α-helical segment on the left and each filter sequence on the right, just above P1, P2, P2, and P4. In the intact, folded protein, the four filter sequences would cluster radially around a pore, thus forming several binding sites for K+ ions being transported.

The majority of cytoplasmically localized residues, including the N terminus, the C terminus, and the long hydrophilic loop (L23) show little conservation, whereas the transmembrane helices tend to be well conserved, especially the so-called pore loops (P1, P2, P3, and P4). Indeed, the ‘signature’ glycine residues within the putative filter sequences, QAGLN, DLGLT, TVGFS, and TVGMS, appear to be absolutely conserved, not only between species, but also among the separate MPM motifs within each species. More broadly, among the four MPM motifs, the segment TM7 through TM8 is the best conserved.

A detailed view of these results, extended to 48 TRK sequences that are now complete in the fungal (ascomycete) genome databases, is provided in the Fig. S2 (1–4). This information is analyzed and summarized in Fig. 3, via phylogenetic trees for the four separate MPM motifs. The colors designate seven distinct clades, which are roughly conserved in the four MPM motifs. Trk1p for C. albicans (marked by a white dot) relates most closely with the same six TRK proteins (the red block) in all four MPMs, for Ashbya gossypii, Debaryomyces hansenii, Debaryomyces occidentalis, Pichia guilliermondii, Pichia stipitis, and Yarrowia lipolytica, although nearest neighbor arrangements within that group differ considerably among the four MPM motifs. The closest adjacent clade (the blue block), containing S. cerevisiae (two isoforms), Saccharomyces uvarum, Candida glabrata, Vandervaltozyma polyspora (two isoforms), and Kluyveromyces lactis, is also consistent in all four MPM motifs. Despite the obvious variance, these distributions of sequence are approximately compatible with the current understanding of phylogenetic relationships among the ascomycete fungi (Barr, 2001; Berbee & Taylor, 2001; Kurtzman & Sugiyama, 2001). They also emphasize that residue dispersion across species has occurred at very different rates in different portions of the TRK molecule; in particular, MPM4 has been much more stable than the other three MPM motifs, requiring c. 50% fewer nucleotide substitutions to source the entire set of 48 fungal sequences. The possible significance of this finding is treated further in the Discussion.


Comparison of the separately computed phylogenetic trees of the four MPM motifs in TRK proteins from ascomycete fungi. Sequence data assembled in Fig. S2, aligned via the Clustal V algorithm. Trees constructed via the MegAlign algorithm in the lasergene software (DNASTAR Inc., Madison, WI). Note that distances (hundreds of nucleotide substitutions) are reckoned from the common trunk, rather than from the present, and that the scale of major branches, earlier than 8000 substitutions, is compressed fourfold for MPM1, MPM2, and MPM3. The full list of species names, abbreviated in each panel above, is given in the legend of Fig. S2.

The primary function of CaTrk1p: K+ uptake

Transport functions at yeast plasma membranes are well demonstrated to be stabilized by preconditioning of the cells under generalized starvation, for example incubation for several hours in distilled water or lightly buffered glucose solution (Armstrong & Rothstein, 1964; Eddy & Hopkins, 1989). Influx measurements on such preconditioned cells of Candida were routinely initiated by injecting cell suspensions with 86Rb+ in c. 1 mM extracellular chemical Rb+ (but nominally zero K+). Averaged results from six experiments are displayed in Fig. 4a, showing a bound component of 2.6±1.5 mM (ordinate intercept ±1 SE) and a stable influx (slope) of 6.4±1.2 mM min−1. The dependence of this influx upon the TRK gene/protein was demonstrated previously by means of severe haploid insufficiency: deletion of only one of the two alleles of CaTRK1 reduced Rb+ influx by fivefold (Baev, 2004).


Parameters of K+ uptake by Trk1p in Candida albicans. (a) Average results for six independent experiments at 1 mM extracellular RbCl. (b) Separate experiment for kinetic parameters, using three different extracellular concentrations of RbCl. Experimental details are given in Materials and methods.

Concentration dependence of the uptake process was assessed from similar measurements made with 10 μM, 100 μM, and 1 mM extracellular rubidium, as shown in Fig. 4b, from which the linear slopes describe a simple saturation function having a maximal transport velocity (Vmax) of 19.0 mM min−1, and a Michaelis constant (K0.5) of 0.64 mM. These results place the normal function of the Trk1 protein in Candida in almost the same physiological range as the combined actions of Trk1p and Trk2p in Saccharomyces, for cells of that species similarly preconditioned (Armstrong & Rothstein, 1967). The two are directly compared in Fig. 5 (lower two curves), with kinetic parameters in Saccharomyces of Vmax=16.2 mM min−1 and K0.5=0.56 mM.


K+-limited growth induces more vigorous TRK-dependent transport than does generalized starvation: comparison of Candida and Saccharomyces. Data sources: curve 1 (Candida albicans), Fig. 4b; curve 2 (Saccharomyces cerevisiae), Armstrong & Rothstein (1967) (measured flux of 42K+, not 86Rb+); curve 3, Ramos (1985); curve 4, Rodriguez-Navarro & Ramos (1984). Low-K+ growth medium contained 10 mM arginine brought to pH 6.5 with phosphoric acid, 2 mM MgSO4, 0.2 mM CaCl2, 110 mM glucose, standard vitamins+trace elements, and 20 μM K+ (Rodriguez-Navarro & Ramos, 1984). When [K+]O had fallen to 2 μM, cells were harvested and prepared for the Rb+ influx measurements represented in curves 3 and 4. Kinetic parameters, for curves 1–4, respectively: K0.5=0.64, 0.56, 0.086, and 0.078 mM; Vmax=19.0, 16.2, 28.4, and 27.5 mM min−1.

A long-recognized additional property of the yeast TRK system(s), however, is that its detailed kinetic behavior depends significantly on the regimen of preconditioning, in a manner which defies simple separation into functionally high-affinity and low-affinity systems (Borst-Pauwels, 1981, 1993; Rodriguez-Navarro & Ramos, 1984; Ramos & Rodriguez-Navarro, 1986; Ramos, 1994). Thus, for Saccharomyces cells grown overnight in medium limited only by low K+, uptake of K+ (or Rb+) occurred with roughly 10-fold higher affinity (K0.5∼0.08 mM) and twofold higher velocity (Vmax=28 mM min−1) than for cells stabilized by preincubation in distilled water. The explicit comparison is made in Fig. 5, between the upper two curves for Saccharomyces, and the bottom curve, all representing data from the established literature. It is not known with certainty whether the detailed conditions for K+ starvation similarly affect the kinetics of transport in Candida, but that would be expected, as a mechanism to optimize resources under conditions of varying nutrient stress. With regard to other members of the C. albicans clade (red block in Fig. 3), data on TRK-mediated K+ fluxes in K. lactis (Miranda, 2002) and D. hansenii (Prista, 2007) qualitatively resemble those for CaTrk1p and ScTRK1,2p, but do not address the quantitative impact of varying methods of starvation.

Characteristic secondary function of CaTrk1p: Cl channeling

For cells the size of C. albicans, chemical fluxes of K+ or Rb+ such as reported in Figs 4 and 5 would imply ionic currents in the range of 1–2 pA per cell, only marginally large enough to be measured – as steady currents – by whole-cell patch-clamp techniques. However, early patch-clamp studies of Saccharomyces identified the ScTRK proteins with significantly larger currents, which were peculiarly insensitive to extracellular K+ (Bihler, 1999; Kuroda, 2004). Those currents were shown to arise from a stable anion permeability in both Trk1p and Trk2p, plus the action of Cl ions introduced to cytoplasm by the pipette-filling solution (Kuroda, 2004; Rivetta, 2005). Patch-clamp studies on Candida have now demonstrated a similar Cl permeability in that organism, dependent upon the CaTRK1 protein.

Figure 6a shows a typical set of whole-cell patch-clamp records from a single cell of C. albicans, wild-type strain SGY243. Each of the superimposed traces represents current required to clamp the membrane voltage suddenly from the reference value of −40 mV to test values of +100 mV (top trace), +80, +60, …, −160, −180 mV (bottom trace). Figure 6b depicts the actual voltage-clamp pulses (also superimposed), each lasting for 2.5 s, after a 0.5-s ‘hold’ at the reference value. The upward (outward) currents reflect K+ efflux through Candida's plasma-membrane K+ channel, Tok1p, and – as shown in Fig. 6c– those currents disappeared when both alleles of the TOK1 gene were deleted (Baev, 2003). The currents activated with time constants of c. 120 ms (half times of c. 85 ms read from the left end of each trace) at the onset of each voltage pulse, reflecting molecular conformation changes that are customarily referred to as ‘gating movements’ in bona fide channel proteins. The currents deactivated very much faster when the clamp voltage was returned to its reference value (see right end of each trace).


Voltage pulses trigger outward currents via Tok1p and inward currents via Trk1p in the Candida plasma membrane. Patch-clamp traces from whole-cell records, using 2.5-s voltage pulses from a holding value of −40 mV, as shown superimposed in (b). (a and e) Wild-type strain SGY243; (c) TOK1-knockout strain DBT3; (d) TRK1-single-allele knockout strain CaTK1. Standard extracellular buffer (sealing buffer, pH=7.5) was used throughout, as described in Materials and methods. Standard intracellular (pipette) buffer, containing 183 mM Cl, was used in the experiments of (a), (c), and (d); Cl was replaced by gluconate, for the experiment of (e).

The downward (inward) current traces in Fig. 6a reflect ion flow associated with Trk1p, the K+ transporter protein, and these were nearly abolished by deletion of a single TRK1 allele, as demonstrated in Fig. 6d. This finding is fully compatible with the severe reduction of cation influx, produced by single-allele deletion (86Rb+; Baev, 2004), in this diploid organism. (CaTRK1 appears to be an essential gene, and C. albicans does not grow, even on K+-rich medium, when both alleles have been deleted.) As had been found in Saccharomyces, however, these inward currents proved insensitive to extracellular [K+] (data not shown) and were roughly proportional to intracellular chloride ([Cl] in the pipette solutions). Figure 6e demonstrates the nearly complete disappearance of inward currents when [Cl]i was reduced to submillimolar levels.

More detailed experiments, however, have revealed several modes in which the Cl currents, mediated by CaTrk1p, differ very significantly from those mediated by the two TRK proteins in Saccharomyces. Most conspicuous is a large difference in rates of activation during hyperpolarizing voltage pulses. As shown in Fig. 7a, in Saccharomyces the inward currents jumped (downward) essentially as fast as the voltage clamp pulses were imposed. More specifically, the maximal currents for each pulse were attained within a single sampling interval, 63 μs, for all of the records in Fig. 7. Figure 7b, closely resembling the records of Fig. 6, displays much slower activation of the CaTrk1 currents, with time constants of c. 150 ms. The traces in Fig. 7b also display much larger amplitude noise at low frequencies than is apparent for Saccharomyces (in Fig. 7a). Taken together, the slow activation and relatively large low-frequency noise suggest that bursts of Cl ions are admitted through CaTrk1p by typical channel gating movements. For ScTrk1p and ScTrk2p, by contrast, the nearly instantaneous activation and low noise level (Fig. 7a) are more readily compatible with single-ion jumps through the protein, viz., simple Eyring-barrier events (Rivetta, 2005).


TRK-dependent Cl currents are larger, slower, noisier, less pH sensitive, and more sensitive to 4,4′-diisothiocyano-2,2′-stilbene disulfonic acid (DIDS) in Candida than in Saccharomyces. Procedures as in Fig. 6, except that voltage–pulse durations were only 1.5 s for some experiments with Saccharomyces. Extracellular buffer at pH 5.5 was similar to sealing buffer, except that acidic MES was titrated only as far as pH 5.5, with Tris base. For the experiments of (e) and (f), carried out at pHO=5.5, 0.1 mM DIDS was injected into the pipette solution; similar results were obtained with 1 mM extracellular DIDS. Wild-type strain BS202 of Saccharomyces cerevisiae (a, c, and e), and strain SGY243 of Candida albicans (b, d, and f).

Three other properties distinguishing the Cl currents through CaTrk1p from those through the Saccharomyces proteins are their larger amplitude, their pH insensitivity, and their ready blockade by anion-channel inhibitors. Despite the fact that C. albicans cells routinely selected for patch-clamp experiments were significantly smaller than those of S. cerevisiae (diameters of 5–7 vs. 6–8 μm), the measured TRK-mediated currents were conspicuously larger in Candida, as is readily seen in Fig. 7 (cf. b and a). The effects of elevating the pHo from 5.5 to 7.5 are also demonstrated in Fig. 7: that is, a fourfold reduction of current amplitude in Saccharomyces but no change of amplitude in Candida (cf. Fig. 7c with Fig. 5a, and 7d with 7b). (However, the rate of activation was slowed in Candida by about threefold.) Finally, the classic anion-channel inhibitor 4,4′-diisothiocyano-2,2′-stilbene disulfonic acid had little effect on Cl currents through ScTrk1p+ScTrk2p (cf. Fig. 7e with Fig. 7a), but nearly completely blocked the currents through CaTrk1p (cf. Fig. 7f with Fig. 7d; see also Baev, 2004). These results are further evidence that the molecular events determining Cl permeability of the TRK protein in Candida differ significantly from those in Saccharomyces.


Implications from sequence

Comparison of the TRK gene among strains of C. albicans, as summarized in Table 1, shows the strain ATCC 10261 to be more closely related to WO-1 than to SC5314 (which was selected first for Candida genome sequencing), as judged by the frequency of SNP variations in the coding region plus c. 800-base flanks. While the overall incidence of SNPs is consistent with random single events, their actual distribution is clearly nonrandom; as is generally to be expected, SNPs occur in coding regions at only about half of the rate observed in noncoding flanks. But, among the three strains, DNA sequences that correspond to the ‘channel-forming’ MPM motifs – viz. 35% of the TRK protein – contain only 4% of SNPs (1/26) identified in the whole coding region. On this basis alone, selection has clearly occurred for structural stability in those domains of the protein that are directly involved in K+ transport.

The same conclusion has emerged more conventionally by comparison of amino-acid sequences among homologues of CaTrk1p, across fungal species (see Fig. S2). This information reveals MPM4, the most C-ward component of the protein that is folded into the transport structure, to be especially strongly conserved (Figs 2 and 3), accumulating fewer than half the mutations across species as in the other three MPM segments. One possible interpretation of this finding is that MPM1,2,3 have evolved separately from MPM4; but that seems unlikely, because the primary function of TRK proteins in fungi – K+ accumulation – is regarded as essential. However, if MPM4, but not the other three MPM motifs, were involved in a separate function (such as Cl channeling), simultaneous imposition of two selective pressures could retard its evolution. A relevant additional point may be that the selectivity of the actual ionic pathway through TRK proteins, for K+ ions relative to Na+ or other monovalent cations, is only modest (Armstrong & Rothstein, 1967) compared with the selectivity of canonical K+ channels, for example.

A particularly surprising feature of interspecies sequence comparisons for MPM4, as originally noted by Durell (1999), is conservation of residues along TM7 and TM8, which ‘should’ be buried rather nonspecifically in the plasma membrane's phospholipid bilayer. This observation led to a structural picture (see Modeling the unexpected, below) which cogently anticipated the observed secondary function of fungal TRK proteins.

Functions of Trk1p in Candida

Serious functional comparisons of Candida Trk1p can be made with proteins from only one other yeast species thus far, S. cerevisiae. As demonstrated in Table 2 for all four MPM motifs, sequence identity is nearly 60% between CaTrk1p and both Saccharomyces proteins, and similarity is near 75% when conservative substitutions are included. The numbers for MPM4 itself are close to 65% and 85%. While many factors determine the actual functional capability of a protein in situ– including other (binding) proteins and small molecules that may differ from organism to organism – extended identity/similarity between proteins in two separate species is normally expected to reflect a quantitatively similar function. This expectation was certainly satisfied by the data on K+ transport per se (Fig. 5), when C. albicans and S. cerevisiae were similarly preconditioned by generalized starvation.

Table 2 Summary of identities and similarities of primary structure, between CaTrk1p and the two Saccharomyces proteins, ScTrk1p and ScTrk2p

The effect of pure K+ starvation (growth in rich medium containing only μM K+) still needs to be explored in Candida, for comparison with data from Saccharomyces (upper two curves of Fig. 5). Another important property remaining to be explored, in both organisms, is the effect of small changes of sequence on cation selectivity in transport, particularly with respect to the selectivity-filter motifs (QAGLN, DLGLT, TVGFS, TVGMS). Studies on bacterial and plant TRK proteins have shown that cation permeability varies greatly with sequence changes in these motifs, as is to be expected from the large literature on bona fide K+-channel molecules. In KtrB of V. alginolyticus, for example, conversion of any of the four ‘signature’ glycine residues to alanine, serine, or aspartate greatly reduced the absolute transport rates of the protein, and conversion specifically to serine resulted in preferential transport of Na+ rather than K+ (Tholema, 1999, 2005).

The secondary function of TRK proteins, outward conduction of Cl ions (Fig. 6), also confirms general expectation based on similarity of sequence. However, the observed quantitative differences from this function in Saccharomyces are particularly interesting. As shown in Fig. 7, the Cl currents associated with CaTrk1p are slowly activating (in response to voltage shifts), large, noisy, insensitive to changes of extracellular pH, and very sensitive to anion channel blockers. Such differences might arise from any of several general causes: detailed sequence differences between the Saccharomyces and Candida proteins, differences of the membrane environment in the two species, or different cytoplasmic binding proteins and regulatory pathways.

Although this secondary function of fungal TRK proteins may have been an important factor in the strong interspecies sequence conservation of the MPM4 segment, the essential physiological role of such Cl channeling is still speculative. Glycophilic fungi seem to need only trace amounts of Cl, and intracellular concentrations should be kept low – compared with the extracellular solutions – by large steady-state membrane voltages (viz., in the range of −200 mV). In this context, a TRK-mediated Cl pathway should serve as a Cl escape route, perhaps even too efficiently, because Saccharomyces, at least, appears to concentrate Cl (weakly) by means of a formate transporter (Jennings & Cui, 2008). But in yeasts that can adapt to very salty environments, this pathway could become essential to Cl detoxification. The clade containing C. albicans (red block, Fig. 3) is rich in halophilic species, including D. hansenii, D. occidentalis, P. guilliermondii, P. stipidis, and Y. lipolytica. Whether the pathway plays that same role in sustaining C. albicans on mammalian epithelial surfaces, or in physiological saline solutions such as saliva (with <150 mM salt), is not yet known.

Modeling the unexpected

Potential insight into the origin of Cl channeling via the TRK proteins, and specifically to the corresponding functional differences between C. albicans and S. cerevisiae, is afforded by the structural model of fungal TRK proteins originally proposed by Durell & Guy (1999) as an interesting way to accommodate the unexpected degree of sequence conservation in transmembrane segments TM7 and TM8: intramembranal oligomerization of TRK molecules (see Introduction). This model features specific close packing of the TM7 helix from each of four molecules of Trk1p at the center of the assembly, with the four TM8 helices forming a supporting ring (see fig. 5 in Durell & Guy, 1999, and fig. 14 in Rivetta, 2005). The postulated tetrad assembly would thus carry a central ‘channel’ in addition to the four radially arranged K+ pathways formed by the selectivity-filter motifs in each individual molecule.

According to atomic coordinates provided by H.R. Guy, this central channel would possess a wide, positively charged vestibule at the intracellular surface of the yeast plasma membrane, and two uncharged choke points along the channel wall, buried within the membrane bilayer. This structural model provides a way to account quantitatively for Cl efflux currents in S. cerevisiae (see Rivetta, 2005). The essential residues in TM7 of ScTrk1p correspond to residues Arg879 (in the vestibule), Trp887, and Phe894 of CaTrk1p. If this model is generally correct, three other residues in TM7 may also contribute to the special properties of Cl currents in Candida, Asn880, Cys890, and Ala899, which correspond to very different residues in Saccharomyces: lysine or arginine, phenylalanine, and cysteine, respectively.

This ‘central pore’ hypothesis remains to be tested by mutational analysis, as well as by the effects – in Candida– of other chaotropic ions (nitrate, thiocyanate, and bromide), whose permeability via Trk1,2p of Saccharomyces is at least equal to that of Cl (A. Rivetta & T. Kuroda, unpublished data).


Fig. S1. Gene (A) and protein (B) sequences for the Candida albicans potassium transporter, CaTrk1p, as cloned by expression in Saccharomyces cerevisiae, from a plasmid library of C. albicans strain ATCC 10261. Shaded residues differ from those reported for the genome sequences of strain SC5314 (Assembly 21, at) and/or strain WO‐1 (Supercontig 1.2, at). Underlined residues correspond to expected transmembrane helices in the protein. Other details, and summary, are provided in Table 1.

Fig. S2. Alignments of the four MPM motifs across fungal species. Comparison of the 48 ascomycete TRK proteins whose sequences appear to be complete, as inferred from gene sequences. Alignments slightly modified from those obtained via the Clustal V algorithm in Lasergene. Except for a few species (Saccharomyces cerevisiae, Neurospora crassa, Candida albicans, and Schizosaccyaromyces pombe) for which in vivo functional characterization of the proteins has been carried out, arbitrary serial numbers (1–4) are used for similar proteins within a species. A total of ∼35 implied protein sequences have been omitted from the alignment because of gross deficiencies: typically absence of the N terminus and/or MPM1. Five extra residues are shown at the beginning and end of each MPM motif. The stacking order in these alignments is the same for all four MPM motifs, determined by maximizing agreement with the consensus sequence, at the top of the stack for MPM4 (viz., for the best conserved motif). Numbers inserted into the peptide sequences represent short unmatched loops: In segment L1P (MPM1) for Pa1, 4 = GGTA; for Pn3, 8 = GPKEYGKA; and in LP2 FOR Ac1, 23 = GPIDRSRGRGSLTRNVTKCSVDL. In segment P2 (MPM2) for At1, 6 = LVVKRS. In segment L5P (MPM3) for Ci1, 2 = GW; and in TM6 for Ac4, 19 = MVSFVVVLTCPNTRLTTLQ. Red-shaded residues are identical with the consensus sequence. Species names and genome database references for each sequence are below.
Sc2 = Saccharomyces cerevisiae Trk2p, gi|6322903|ref|NP_012976.1|
Vp2 = Vanderwaltozyma polyspora Trk2p, gi|156113782|gb|EDO15326.1|
Ci2 = Coccidioides immitis Trk2p, gi|119182899|ref|XP_001242549.1|
Cgla1 = Candida glabrata Trk1p, gi|50293219|ref|XP_449021.1|
Pg1 = Pichia guilliermondii Trk1p, gi|146420714|ref|XP_001486311.1|
Ps1 = Pichia stipitis Trk1p, gi|150866084|ref|XP_001385561.2|
Bf1 = Botryotinia fuckeliana Trk1p, gi|154317942|ref|XP_001558290.1|
Do1 = Debaryomyces occidentalis Trk1p, gi|7799615|emb|CAB91046.1|
Dh1 = Debaryomyces hansenii Trk1p, gi|50408451|ref|XP_456781.1|
At2 = Aspergillus terreus Trk2p, gi|115437446|ref|XP_001217812.1|
Ac2 = Aspergillus clavatus Trk2p, gi|121719174|ref|XP_001276311.1|
Ag1 = Ashbya gossypii Trk1p, gi|45198303|ref|NP_985332.1|
Ao2 = Aspergillus oryzae Trk2p, gi|83771272|dbj|BAE61404.1|
Vp1 = Vanderwaltozyma polyspora Trk1p, gi|156116658|gb|EDO18129.1|
Af1 = Aspergillus fumigatus Trk1p, gi|70985214|ref|XP_748113.1|
Ca1 = Candida albicans Trk1p, gi|8099700|gb|AAF72203.1|AF267125_1
Yl1 = Yarrowia lipolytica Trk1p, gi|50547129|ref|XP_501034.1|
Sc1 = Saccharomyces cerevisiae Trk1p, gi|151944998|gb|EDN63253.1|
Kl1 = Kluyveromyces lactis Trk1p, gi|4809179|gb|AAD30128.1|AF136181_1
Su1 = Saccharomyces uvarum Trk1p, gi|136230|sp|P28569|TRK1_SACBA
Anig1 = Aspergillus niger Trk1p, gi|145230083|ref|XP_001389350.1|
Pn2 = Phaeospaeria nodorum Trk2p, gi|111056058|gb|EAT77178.1|
Ss1 = Sclerotinia sclerotiorum Trk1, gi|156050117|ref|XP_001591020.1|
Sp2 = Schizosaccharomyces pombe Trk2p, gi|19114846|ref|NP_593934.1|
Gz1 = Giberrella zeae Trk1p, gi|46115984|ref|XP_384010.1|
Ac1 = Aspergillus clavatus Trk1p, gi|121708654|ref|XP_001272203.1|
Pn1 = Phaeospaeria nodorum Trk1p, gi|111062200|gb|EAT83320.1|
Anid1 = Aspergillus nidulans Trk1p, gi|67902084|ref|XP_681298.1|
Mg1 = Magnaporthe grisea Trk1p, gi|145609089|ref|XP_364274.2|
Sp3 = Schizosaccharomyces pombe Trk3p, gi|550526|gb|AAC41667.1|
Pa1 = Podospora anserina Trk1p, gi|18699011|gb|AAL77222.1|
At1 = Aspergillus terreus Trk1p, gi|115389814|ref|XP_001212412.1|
Ao1 = Aspergillus oryzae Trk1p, gi|83775062|dbj|BAE65185.1|
At3 = Aspergillus terreus Trk3p, gi|115443294|ref|XP_001218454.1|
Gz2 = Giberrella zeae Trk2p, gi|46125803|ref|XP_387455.1|
Nf1 = Neosartorya fischeri Trk1p, gi|119482361|ref|XP_001261209.1|
Ci1 = Coccidioides immitis Trk1p, gi|119192950|ref|XP_001247081.1|
Sp1 = Schizosaccharomyces pombe Trk1p, gi|63081157|ref|NP_592860.3|
Nc1 = Neurospora crassa Trk1p, gi|85084587|ref|XP_957340.1|
Pn3 = Phaeospaeria nodorum Trk3p, gi|111061940|gb|EAT83060.1|
Ac3 = Aspergillus clavatus Trk3p, gi|121714993|ref|XP_001275106.1|
Nf2 = Neosartorya fischeri Trk2p, gi|119494465|ref|XP_001264128.1|
Pa2 = Podospora anserina Trk2p, gi|27764333|emb|CAD60613.1|
Ac4 = Aspergillus clavatus Trk4p, gi|121700757|ref|XP_001268643.1|
Ao3 = Aspergillus oryzae Trk3p, gi|83765904|dbj|BAE56047.1|
Gz3 = Giberrella zeae Trk3p, gi|46109050|ref|XP_381583.1|
Pn4 = Phaeospaeria nodorum Trk4p, gi|111057422|gb|EAT78542.1|
Nc2 = Neurospora crassa Trk2p, gi|85092714|ref|XP_959511.1|


This work was supported by research grants GM60696 from the National Institute of General Medical Sciences (to C.S.) and DE10641 from the National Institute of Dental and Craniofacial Research (to M.E.). M.M. was supported in part by grant 5G12 RR 008124, from the National Center for Research Resources, to the Border Biomedical Research Center/University of Texas at El Paso. The authors are indebted to Dr Carolyn W. Slayman, Dr Alan B. Mason, and Mr Kenneth Allen (Yale Department of Genetics) and to Drs Paul T. Magee and Beatrice B. Magee (Department of Genetics and Cell Biology, University of Minnesota, St. Paul, MN) for much helpful advice and provision of strains. We are also indebted to Drs Richard F. Gaber (Department of Biochemistry, Molecular Biology, and Cell Biology, Northwestern University, Evanston, IL) and Per Lungdahl (Karolinska Institute, Stockholm, SE) for strains of S. cerevisiae, to Dr John D. Reid (formerly of Glaxo IMB, Zurich, Switzerland) for the C. albicans DNA library and to Drs Stewart R. Durell and H. Robert Guy (National Cancer Institute, N.I.H., Bethesda, MD) for the computed cross-species indices of conservation of TRK proteins (Fig. 2), as well as for the coordinates of their atomic-scale model of the yeast protein.


  • Editor: André Goffeau


View Abstract