According to the literature 3, C. dubliniensis is relatively sodium sensitive, whereas C. parapsilosis was shown to tolerate high NaCl concentrations 22. In order to estimate the tolerance of Candida species to different alkali metal cations, the growth of four Candida species and a S. cerevisiae wild type (as a non-osmotolerant control) in the presence of increasing concentrations of various salts was estimated. In the absence of salts, S. cerevisiae cells grew more slowly than all four Candida species, and also the growth of C. parapsilosis was not as robust as with the other three Candida species (Figure 1). All yeast species grew equally well in the presence of lower salt concentrations (cf. Methods) but as the amount of alkali-metal-cation on the plates increased, important differences were observed. Of the tested Candida species, C. dubliniensis had the lowest tolerance to all of the tested salts, it is not able to grow when the salt concentrations are above 1600 mM NaCl, 2300 mM KCl or 200 mM LiCl. Nevertheless, C. dubliniensis is more sodium, potassium and rubidium tolerant than S. cerevisiae, but is much more sensitive to toxic lithium cations (Figure 1). Also, the lithium sensitivity of C. glabrata is higher than that of S. cerevisiae but C. glabrata cells can grow in much higher concentrations of the other salts than S. cerevisiae cells (up to 2300 mM NaCl, 2400 mM KCl, 1800 mM RbCl; Figure 1). C. albicans and C. parapsilosis are the most halotolerant species, and C. parapsilosis seems to grow even better in the presence of high salt concentrations than C. albicans (Figure 1). To summarize, the four Candida species have not only different sensitivity to NaCl, as published previously 322 but they differ in their tolerance to alkali metal cations in general. C. albicans and C. parapsilosis are highly halotolerant and C. dubliniensis is halosensitive.

A search in databases revealed the existence of open reading frames homologous to the CaCNH1 gene in the genomes of C. dubliniensis, C. glabrata and C. parapsilosis. In these species, just one homologous sequence was found, suggesting that their plasma-membrane Na⁺/H⁺ antiporters have a broad substrate specificity to alkali metal cations, similar to those of C. albicans and C. tropicalis. We named the identified orthologous genes according to their species of origin, CdCNH1, CgCNH1 and CpCNH1. The CdCNH1 gene is 2454 nt (818 aa) long, CgCNH1 has 2835 nt (945 aa), and the CpCNH1 gene is composed of 2955 nt (985 aa). Neither of them have any introns.

The predicted protein structures of these three antiporters were compared with the two Candida antiporters that have already been characterized, CaCnh1 and CtCnh1. Comparison of the protein length and predicted structure of the Cnh1 proteins from five Candida species and S. cerevisiae revealed that CpCnh1p is the longest and C. dublinienis antiporter the shortest member of the Candida Na⁺/H⁺antiporters' subfamily (Table 1). For all Candida proteins, the Kyte-Doolittle method predicted a similar structure to ScNha1p with highly conserved N-termini and 12 trans-membrane sections (Tables 1 &2). On the other hand, they differ in the length and composition of their hydrophilic C-termini, as do the antiporters from non-Candida yeast species 23. The most significant is the difference in length (approx. 180 aa) between the C-termini of CpCnh1p (555 aa), CaCnh1p (366 aa) and CdCnh1p (388 aa). Of the analyzed proteins, CaCnh1p and CdCnh1p show the highest sequence identity in all parts of the protein, though the hydrophobic trans-membrane domains and connecting loops are highly conserved (approx. 90% identity) in all Candida antiporters except for CgCnh1p (Table 2). C. glabrata Cnh1p is more similar to S. cerevisiae Nha1p than to the antiporters from other Candida species in all the features analyzed, which corresponds to the phylogenetic relationships among these yeast species 24.

Though the highest divergence from identity was found in the C-termini of all the compared antiporters, the existence of six conserved C-terminal regions described previously 18 was also confirmed in C.dubliniensis, C. parapsilosis and C. glabrata species (not shown). A new, approximately 25 aa-long conserved region (K/R)(L/I)SR(S/T)(L/A)SRRS(Y/F)Y(K/R)KDDP(H/N)(K/R)RKVYAHR (in CaCnh1p aa 639–664) preceding conserved region no. 5 18; was found in the four Candida species, except for C. glabrata.

Both the C. dubliniensis and C. parapsilosis species belong to the group of yeasts in which the CTG codon encodes a serine and not a leucine, as in other yeast species (i.e. S. cerevisiae) 25. One CTG codon exists at aa position 621 in CpCNH1. This serine 621 is localized in the antiporter's hydrophilic C-terminus and not in the membrane part of the protein. It is localized in a small weakly conserved area where at a similar position CdCnh1p (aa 568) and CtCnh1p (aa 644) also have a serine and CaCnh1p (aa 552) has an isoleucine.

In order to 1) verify whether the identified and analysed open reading frames encode functional plasma-membrane alkali-metal-cation/proton antiporters, and 2) elucidate whether the species' salt tolerance could reflect the activity of these antiporters, corresponding DNA fragments from the most and least salt tolerant species, C. parapsilosis and C. dubliniensis respectively, were expressed in a S. cerevisiae mutant lacking its own export systems for alkali metal cations (BW31a ena1-4Δ nha1Δ 8). The lack of both Na⁺-ATPases, Ena1-4 and the Na⁺/H⁺ antiporter Nha1 renders these cells extremely sensitive to higher external concentrations of salts, and almost no efflux of sodium or potassium cations from them is observable. The functional expression of heterologous sodium and/or potassium exporters in BW31a cells has clear phenotypes of an increased salt tolerance and measurable alkali metal cation efflux 26.

The CdCNH1 and CpCNH1 genes amplified from genomic DNAs were cloned behind the ScNHA1 promoter and expressed from multicopy vectors, as were the genes ScNHA1 and CaCNH1 719, which served as positive controls in our study. Thus all four antiporters were expressed under the same conditions (strain, vector, promoter) which should ensure similar antiporters' levels in cells. Empty YEp352 and pGRU1 vectors served as negative controls. The functionality of all the constructs were first tested in drop experiments, which showed that 1) the presence of the constructs did not influence the growth rate of cells in standard media, i.e. the heterologous expression of these membrane proteins was not toxic for S. cerevisiae, and 2) the expression of both GFP-tagged and non-tagged CdCnh1 and CpCnh1 proteins brought about the same ability to grow on 800 mM NaCl or 1800 mM KCl, as did the positive controls with ScNha1 and CaCnh1 proteins, whereas the cells without antiporters were not able to grow (not shown). This result also confirmed that the C-terminal GFP-tagging did not influence the activity of the antiporters. In order to estimate the substrate specificity and transport capacity of the antiporters, BW31a cells expressing the four antiporters or transformed with an empty vector were spotted on a series of YNB plates containing increasing NaCl, KCl, LiCl and RbCl concentrations. Cells expressing CpCnhp1p were able to grow in the highest concentrations of salts, as did cells expressing CaCnh1p. Both these Candida antiporters conferred a slightly higher tolerance to the cells than equivalent expression of the native S. cerevisiae antiporter, Nha1p (Figure 2). The tolerance of cells expressing CdCnh1p to high external potassium and rubidium was almost the same as for cells expressing CpCnhp1p and CaCnh1p (Figure 2), but their tolerance to toxic cations was significantly lower, only 1000 mM NaCl, and there was no increase in LiCl tolerance compared to cells with the empty vector (30 mM LiCl in both cases).

The proton-antiport mechanism of these Cnh1 proteins was verified in a series of drop tests on plates with various pH values (Table 3). As was previously thought, the cells expressing antiporters showed the highest salt tolerance (at least for three of their four substrates) when grown at lower external pH, i.e. in conditions where the proton gradient across the plasma membrane is the highest. Surprisingly, the expression of both the S. cerevisiae and C. dubliniensis antiporters did not increase the cell tolerance to lithium cations at pH 3.5, suggesting that Li⁺ was not recognized as their substrate under these conditions. On the other hand, all four antiporters were partially active, even at an externally neutral pH 7.0, as their presence enabled the cells to support higher salt concentrations. This detailed study confirmed again that CdCnh1p has the least ability to improve the salt tolerance of cells (Table 3).

As mentioned above, C-terminal GFP-tagging of the CdCnh1 and CpCnh1 proteins did not affected their functionality. Both antiporters improved the cell salt tolerance to a similar degree as the non-tagged versions, and fluorescence microscopy localized them to the plasma membrane of S. cerevisiae BW31a cells (Figure 3) and not to the membranes of intracellular organelles. This result indicates a high probability of the same localization in their organisms of origin, as was previously shown for C. albicans Cnh1p 20.

To verify whether the use of the same vector and promoter for expression ensures similar levels of the four antiporters in BW31a cells, the GFP-tagged proteins were visualized on western blots. Fig. 4 shows that 1) the size of Sc and Cp antiporters was alike, similarly as the size of Ca and Cd transporters (and in agreement with the size deduced from the gene sequence, cf. Table 1 and 2) the quantity of antiporters in extracts of exponentially growing cells was similar, the highest amount apparently being observed for CdCnh1p and the lowest one for CpCnh1 antiporter. The analysis was repeated three times with the same result. The fluorescence microscopy and western blot analysis confirmed that the observed low activity of CdCnh1p and the high activity of CpCnh1 antiporter did not reflect different protein levels in the cells.

To confirm the results from drop test experiments and to determine the activity and efflux rate of CdCnh1p and CpCnh1p, the efflux of K⁺ and Na⁺ was directly measured. The loss of K⁺ from BW31a cells expressing C. dubliniensis or C. parapsilosis antiporters was measured directly; to measure Na⁺ efflux, cell preloading with 100 mM NaCl was necessary (cf. Methods). Cells expressing the CaCnh1 antiporter served as positive, and cells transformed with an empty vector as negative controls, respectively. The initial internal concentration of K⁺ in exponentially growing cells was almost the same in all strains, in the representative experiment about 549.5 ± 25.7 nmol (mg dry wt)^-1. After preloading, cells contained 110.5 ± 6.8 nmol (mg dry wt)^-1 Na⁺. As shown in Figure 5 and Table 4, CpCnh1p exported Na⁺ and K⁺ much more efficiently than CdCnh1p. Within 60 minutes, cells with CpCnh1p lost 80% of their internal sodium and 32% of their potassium compared to cells with CdCnh1p, which over the same period only lost 38% of their sodium and 20% of their potassium. The C. albicans antiporter was the most effective, exporting 82% of its host cells' sodium and 55% of their potassium in 60 min, which agrees with previously published results 19. The Na⁺ efflux curves for CaCnh1p and CpCnh1p are very similar and almost exponential; most of the sodium is exported in about 40 minutes. The efflux of sodium via CdCnh1p is linear and slow. The sodium efflux via all three antiporters is faster than their potassium efflux, though the initial intracellular concentration of K⁺ is much higher than that of Na⁺, approx. 300 vs. 50 mM. These results suggest that the Candida antiporters have, at least upon heterologous expression in S. cerevisiae, a higher affinity for sodium than for potassium cations.

To determine the salt tolerance of various Candida species, C. albicans SC5314, C. glabrata ATCC2001, C. dubliniensis CD36 and C. parapsilosis CBS604 were used, together with S. cerevisiae S288c as a control. The CdCNH1 and CpCNH1 genes were isolated from C. dubliniensis CD36 and C. parapsilosis CBS604, and heterologously expressed in S. cerevisiae BW31a (ena1-4Δ nha1Δ, W303 derivative, 8). Yeast cells were grown in YPD or YNB-NH₄⁺ media supplemented with 2% glucose at 30°C. Salts were added to the media prior to and auxotrophic supplements after autoclaving.

For DNA manipulations, standard protocols 30 were used. The CdCNH1 and CpCNH1 gene were amplified by PCR with platinum Pfx polymerase with proofreading activity (Invitrogen) using their isolated genomic DNA 31 as a template.

Plasmids for the heterologous expression of these antiporters in S. cerevisiae were constructed by homologous recombination in BW31a cells. The oligonucleotides used are listed in Table 5. Two types of plasmids were constructed. CdCNH1 and CpCNH1 coding sequences were cloned behind the ScNHA1 promoter either in multicopy YEp352 or in pGRU1, enabling C-terminal GFP tagging 7. All constructs were analysed by sequencing in an ABI PRISM 3100 DNA sequencer using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems).

The cell tolerance to various alkali metal cations was determined by drop test experiments. 3 μl of serial 10-fold dilutions of saturated cell cultures were spotted on solid media. For Candida species, YPD – 3% agar plates supplemented with increasing amounts of salts were used (800 – 2600 mM NaCl; 1800 – 2500 mM KCl; 20 – 1000 mM LiCl; 700 – 1800 mM RbCl). Determination of the maximum salt tolerance of S. cerevisiae BW31a cells expressing either of the antiporters was performed on solid YNB media supplemented with salts and adjusted to various pH levels as described previously 7. The following salt concentrations were used: 1) non-adjusted pH, 500–1500 mM NaCl; 1800–2100 mM KCl; 25–40 mM LiCl; 1000–1600 mM RbCl; 2) pH 3.5, 500–1300 mM NaCl; 1600–2200 mM KCl; 20–50 mM LiCl; 3) pH 5.5, 200–1000 mM NaCl; 1900–2200 mM KCl; 15–40 mM LiCl; 4) pH 7.0, 50–200 mM NaCl; 800–1500 mM KCl; 5–10 mM LiCl.

The transport activity of the antiporters was measured as the cation efflux from cells according to 7 with minor modifications. Cells were grown to OD₆₀₀ ≈ 0.2 in YNB, harvested and then a pH 5.5 incubation buffer (20 mM MES, supplemented with 10 mM KCl or 10 mM RbCl to prevent Na⁺ or K⁺ reuptake respectively) was used. For the determination of Na⁺ efflux, the cells were preloaded for 60 min with 100 mM NaCl in YNB adjusted to pH 7.0 with NH₄OH. The K⁺ efflux was measured directly in the harvested cells. Samples were taken from cell incubation cultures at regular time intervals over a period of 60 min and the cellular cation content was determined by atomic absorption spectroscopy 7. Each efflux experiment was repeated at least three times and representative results are shown.

Exponential phase cells (grown in YNB at 30°C, OD₆₀₀ ≈ 0.15) expressing the CdCHN1 or CpCNH1 gene tagged with the GFP sequence were viewed with an Olympus AX70 microscope using a U-MWB cube with a 450–480 nm excitation filter and 515 nm barrier filter. The micrographs were recorded with a DP70 digital camera using the program DP Controller. For whole-cell pictures, Nomarski optics was used.

Exponetially growing (OD₆₀₀ ≅ 0.15) BW31a cells expressing GFP-tagged antiporters were harvested and concentrated by centrifugation to OD₆₀₀ = 3.0. The proteins were extracted according to 32 with some modifications. After resuspension of cell pellet in 150 μl freshly prepared 1.85 M NaOH with 7.5% β-mercaptoethanol and incubation for 15 min on ice, 150 μl of cold 50% trichloroacetic acid were added. After incubation on ice for 20 min the collection of precipates by centrifugation at 20,000 × g for 20 min followed. The pellet was resuspended in 190 μl of 50 mM Tris-HCl buffer (pH 6.8), containing 8 M urea, 5% sodium dodecyl sulfate (SDS), 0.1 mM EDTA and 1.5% dithiothreitol (DTT) + 10 μl 1 M Tris base. After incubation for 30 min at 37°C the samples were centrifuged at 20,000 × g for 30 min. Supernatant (7.5 μl) were directly loaded on 8% glycine gel and separated by polyacrylamid gel electrophoresis (PAGE). Separated proteins were transferred via electroblotting on nitrocellulose membrane. To detect the GFP-tagged proteins on membranes, rabbit polyclonal anti-GFP antibody (Santa Cruz Biotech., diluted 1:200), secondary goat anti-rabbit IgG antibody with conjugated peroxidase (BioRad, diluted 1:10,000) and ECL detection kit (Pierce) were used.