Abstract
Multidrug transporters are often responsible for failure of medical treatment, since they expel a variety of structurally and functionally unrelated drugs out of the cell. We found that the fluorescent probe diS-C3(3) is a substrate of not only Pdr5p of Saccharomyces cerevisiae (ScPdr5p) but also of its less-explored Kluyveromyces lactis homologue (KlPdr5p). This enabled us to compare the ability of azoles to competitively inhibit the Pdr5p-mediated probe efflux in the two species. In K. lactis, these azoles completely inhibit probe transport by KlPdr5p and also compete with each other for transport. This indicates that the probe and the azoles are bound by the same site(s) of the KlPdr5p binding pocket. On the other hand, the azoles’ capacity to inhibit the probe transport by ScPdr5p is limited, as a result of their partial cotransport with the probe. While the azoles bind to only one or two separate binding sites, the probe is able to bind to all three of them. Moreover, the bulky ScPdr5p substrate enniatin B, which effectively inhibits both probe and azole transport by the pump, has negligible effect on KlPdr5p. Our data point to a tighter arrangement of the KlPdr5p binding pocket compared to that of ScPdr5p.
INTRODUCTION
Multidrug resistance (MDR) is an evolutionary conserved cellular mechanism providing defence against chemical toxicity of environmental toxins and drugs. In yeast, MDR is the result of the overexpression of drug efflux pumps belonging to the ATP-binding cassette (ABC) superfamily of plasma membrane proteins (Prasad and Goffeau 2012). One of the main MDR pumps of Saccharomyces cerevisiae is ScPdr5p, which is able to rid the cells of an astonishingly wide range of structurally and functionally distinct compounds (Kolaczkowski et al.1996). The function and structure of ScPdr5p have been studied and described in detail over the last 25 years (reviewed in Golin and Ambudkar 2015). The results of these studies have provided us, among other things, with knowledge about the way substrates bind to the pump. Based on studies using rhodamine 6G as a benchmark substrate, Kolaczkowski et al. (1996) proposed that ScPdr5p makes use of more than one site to bind its substrates. This idea has been elaborated upon by others and the current view is that the binding pocket of ScPdr5p has at least three distinct binding sites, each apparently using different chemical properties to transport the substrates, as shown by [3H]-chloramphenicol accumulation (Golin et al.2003;Hanson et al.2005).
Owing to its industrial application (e.g. in lactose fermentation and as a host for heterologous protein expression) and metabolic peculiarities (Crabtree-negative, petite-negative), Kluyveromyces lactis has become an alternative model organism to the more traditional S. cerevisiae, especially for other unconventional yeast species. The two species are phylogenetically closely related, both being members of the Saccharomycetaceae family. It has been proposed by Wolfe and Shields (1997) that S. cerevisiae underwent a whole-genome duplication after its divergence from K. lactis. While the majority of the duplicated genes have been deleted from the genome over time, most of the duplicates that have remained in S. cerevisiae only have a single homologue in K. lactis. The number of genes coding ABC transporters is also somewhat lower in K. lactis (22 vs 29 in S. cerevisiae) (Gbelska, Krijger and Breunig 2006). There have been extensive studies on the nature of the ScPdr5p binding pocket concerning its size and number of binding sites. However, similar studies of KlPdr5p, the K. lactis homologue of ScPdr5p (Chen 2001), have not been carried out to date, leading to complete absence of data on this important pump parameter.
To fill some of the gaps, we used a novel approach utilising the diS-C3(3) fluorescence assay originally developed in our laboratory to study the activity of ScPdr5p in cell suspensions (Gaskova et al.2002). The fluorescence assay is based on comparing the accumulation of the benchmark pump substrate diS-C3(3) in Pdr5p-expressing and Pdr5p-deficient cells in the presence and absence of a substrate/inhibitor (Hendrych et al.2009). The difference in staining between Pdr5p-expressing and Pdr5p-deficient cells reflects the pump's capacity for probe transport. Hence, monitoring of how this difference is affected by the addition of a substrate provides us with information about possible inhibition of probe transport and substrate-probe co-transport, as is the case of ScPdr5p-mediated transport of ketoconazole (Gaskova et al.2013). From such data, conclusions about the character of the Pdr5p binding pocket can be drawn.
In this study, we show that the diS-C3(3) fluorescent probe is a valuable benchmark substrate not only for monitoring of the transport activity of ScPdr5p but also of KlPdr5p. This allows us to select a series of common ScPdr5p and KlPdr5p azole substrates and monitor their ability to competitively inhibit the Pdr5p-mediated probe transport in both species. The main aim of this study is to bring insight into the differences between the binding pockets of KlPdr5p and ScPdr5p, focusing on their size and the number of binding sites.
MATERIALS AND METHODS
Yeast strains
The Saccharomyces cerevisiae strains AD1–3 (MATα, PDR1–3, ura3, his1, yor1Δ::hisG, snq2Δ::hisG, pdr5Δ::hisG) and AD12 (MATα, PDR1–3, ura3, his1, yor1Δ::hisG, snq2Δ::hisG) (Decottignies et al.1998) are derived from the parental strain US50–18C (MATα, PDR1–3, ura3, his1). The Kluyveromyces lactis strains CK373/1 (MATa, uraA1, ade2, RAG1+ (pKD1+), pdr5Δ::kan) (Chen 2001) and CK373/1 + KlPDR5 (MATa, uraA1, ade2, RAG1+ (pKD1+), pdr5Δ::kan, p26) (Takacova et al.2002) are derived from the parental strain PM6–7A (MATa, uraA1, ade2, RAG1+ (pKD1+)) (Chen, Wesolowski and Fukuhara 1992).
For convenience, the Pdr5p-expressing strains AD12 and CK373/1 + KlPDR5 are in text referred to as PDR5+ strains and the Pdr5p-deficient strains AD1–3 and CK373/1 as pdr5- strains.
Media and cell growth conditions
Yeast precultures were grown in YPD medium (2% yeast extract, 1% bactopeptone and 2% glucose) at 30°C for 24 h. A small volume (10–40 μl for S. cerevisiae and 1–10 μl for K. lactis) of this preculture was added to 10 ml of fresh YPD medium, and the main culture was grown until it reached the desired phase of growth.
Fluorescence measurements of diS-C3(3) accumulation in cells
Exponential yeast cells were harvested, washed twice with distilled water and resuspended in citrate-phosphate buffer (10 mM HNa2O4P·12H2O titrated to pH 6.0 with citric acid). The optimal concentration of K. lactis cells was set to OD578 = 0.13 (2.5 × 106 cells/ml). Since K. lactis cells are slightly smaller than S. cerevisiae cells, the concentration of S. cerevisiae cells was set to OD578 = 0.11 (1.8 × 106 cells/ml) to keep the overall biomass in the samples constant. This adjustment is important when comparing different yeast strains/species using the diS-C3(3) fluorescence assay.
The diS-C3(3) (3,3΄-dipropylthiacarbocyanine iodide) probe was added to 3 ml aliquots of the yeast cell suspension in cuvettes to a final concentration of 2 × 10−8 M. Fluorescence emission spectra of the stained cell suspensions were measured using a FluoroMax-4 spectrofluorimeter (Horiba JobinYvon) in intervals of 2–5 min. The excitation wavelength was 531 nm, fluorescence emission range was 560–590 nm and scattered light was eliminated by an orange glass filter with a cut-off wavelength of 540 nm.
The rate and extent of intracellular probe accumulation are reported by the dependence of the diS-C3(3) fluorescence emission maximum wavelength (λmax) on the time of staining, the so-called staining curve (Denksteinova et al.1997; Gaskova et al.1998). When appropriate, the tested compound was added to the desired final concentration after ∼20 min of staining. The samples were kept at room temperature and occasionally gently stirred.
The activity of the pump was determined as the difference between the equilibrium intracellular probe concentration (i.e. the equilibrium level of staining, Λ) of pdr5– and PDR5+ strains in the absence and presence of substrates.
Confocal fluorescence microscopy
Cell suspensions were prepared in the same way as described above for the spectrofluorimetric measurements and stained with 2 × 10−8 M diS-C3(3) probe for 30 min. After this incubation interval, micrographs were captured using an Olympus IX83/FV1200 laser scanning microscope with a water immersion objective UPLSAPO 60x/1.2.
Disc diffusion assay
To determine the substrate specificity of KlPdr5p and ScPdr5p, disc diffusion assays were performed as described before (Gaskova et al.2013). Tested chemicals (3–7 μl) were spotted onto Whatman paper discs lying on the top of the cell-containing top agar. After 48 h at 30°C, the plates were photographed and the sizes of the growth inhibition zones were measured. To determine the possibility of competitive inhibition between two substrates, we used a ‘double addition’ mode of the classical disc diffusion test (Gaskova et al.2013), where one of the tested pair of compounds was added onto a disc 15 min before the other (and vice versa). When appropriate, the fluorescence probe diS-C3(3) was added 15 min after the addition of the second substrate (a ‘triple addition’ mode).
Chemicals
The following materials were purchased from the respective companies: diS-C3(3) (3,3΄-dipropylthiacarbocyanine iodide), DMSO (dimethyl sulfoxide), DMF (dimethyl formamide) and aceton (Fluka, Prague, Czech Republic), yeast extract (Serva, Heidelberg, Germany), bactopeptone (Oxoid, Brno, Czech Republic), glucose, ethanol for UV spectroscopy and Na2HPO4·12H2O (Sigma-Aldrich, Prague, Czech Republic), citric acid (Penta, Prague, Czech Republic), agar (Dr Kulich, Pharma, Hradec Kralove, Czech Republic). Pdr5p substrates, protonophore and lysosomotropic compound were obtained from the following sources: azoles (clotrimazole, miconazole, bifonazole, ketoconazole, itraconazole), enniatin B, 2-(methyl-trityl-amino)-ethanol, tritylimidazole, tetrabutyltin, rhodamine 6G and protonophore CCCP (carbonyl cyanide m-chlorophenylhydrazone) (Sigma-Aldrich, Prague, Czech Republic), and the lysosomotropic compound DM-11 (2-dodecanoyloxyethyldimethylammonium chloride) was synthesised in the laboratory of Prof. S. Witek (Univ. Wroclaw; Witek et al.1997) and kindly provided by Dr A. Krasowska.
RESULTS AND DISCUSSION
Fluorescent probe diS-C3(3) and tested azoles are substrates shared by ScPdr5p and KlPdr5p
In order to determine if the diS-C3(3) fluorescent probe is a substrate of KlPdr5p, we compared its accumulation in the Pdr5p-expressing and Pdr5p-deficient (henceforth referred to as PDR5+ and pdr5– respectively) Kluyveromyces lactis strains using confocal fluorescence microscopy (Fig. 1A) and disc diffusion tests (Fig. 1B). The lower probe accumulation in PDR5+ cells compared to pdr5– cells and greater growth inhibition zones of pdr5– cells clearly show that diS-C3(3) is indeed a substrate of KlPdr5p, just like it is a substrate of ScPdr5p (Fig. 1A and B). It is worth noting that the dark compartments inside the cells in Fig. 1A are vacuoles that do not accumulate the probe. Their membrane potential is, in contrast to plasma membrane potential, positive inside, because the vacuolar V-ATPase pumps protons into the vacuole.
DiS-C3(3) is a substrate of both ScPdr5p and KlPdr5p. (A) Intracellular accumulation of the probe in S. cerevisiae (both PDR5+ and pdr5–) and K. lactis (both PDR5+ and pdr5–) cells was imaged by confocal microscopy after 30 min incubation with the fluorescent probe. (B) Growth inhibition zones measured in a disc diffusion test; all strains were exposed to 1 mM diS-C3(3).
To get insight into the possible differences in the arrangement of Pdr5p substrate-binding pocket in both species, we used a series of relatively simple azoles (clotrimazole, miconazole, bifonazole, ketoconazole and itraconazole) that are known substrates of both ScPdr5p (Rogers et al.2001; Golin et al.2003; Hanson et al.2005) and KlPdr5p (Takacova et al.2002; Balkova et al.2009), see also Fig. 2. Monitoring their ability to competitively inhibit the Pdr5p-mediated probe transport in both species enabled us to decide whether these substrates and diS-C3(3) bind to the same site(s) in the binding pocket of the pump.
The tested azoles are substrates of both ScPdr5p and KlPdr5p. Growth inhibition zones of pdr5- and PDR5+ cells of both species exposed to clotrimazole (5.8 mM), miconazole (4.8 mM), bifonazole (6.4 mM), ketoconazole (3.8 mM) and itraconazole (2.8 mM).
Different effectiveness of clotrimazole to competitively inhibit Pdr5p-mediated probe transport in Saccharomyces cerevisiae and Kluyveromyces lactis
To optimise our diS-C3(3) fluorescence assay (Hendrych et al.2009; Gaskova et al.2013) for comparing the probe transport activity of the studied pumps after addition of a shared substrate, we used clotrimazole as a model compound, since it is known to bind to two (referred as site 1 and site 2) out of the three binding sites in the ScPdr5p binding pocket, having different affinities for each (Golin et al.2003; Golin, Ambudkar and May 2007).
To achieve comparable extent of clotrimazole-induced inhibition of Pdr5p-mediated probe efflux in both species, it was necessary to add higher concentrations of the substrate to S. cerevisiae cells than to K. lactis cells, as indicated by the staining of the strains (Fig. 3A and B). This difference originates in higher overall ScPdr5p activity compared to that of KlPdr5p, which is probably a consequence of the overexpression of Pdr5p in the used S. cerevisiae strain, AD12 (Balzi et al.1994). The lower Pdr5p expression in K. lactis is reflected in the smaller difference of diS-C3(3) staining of PDR5+ and pdr5– control cells (Fig. 3A and B) and also in the less striking difference in the inhibition zone sizes of the strains in the presence of clotrimazole (Fig. 2).
Clotrimazole completely inhibits probe transport by KlPdr5p but not by ScPdr5p. The inhibitory effect of clotrimazole (added at t = 20 min) on the activity of (A)KlPdr5p and (B)ScPdr5p. Data are representative of five independent repeats. (C) Comparison of growth inhibition zones of the PDR5+ K. lactis (left) and S. cerevisiae strain (right) exposed to clotrimazole (0.01 mM for K. lactis/4.4 mM for S. cerevisiae), diS-C3(3) (1 mM) and their respective combinations. (D) Extent of inhibition of the probe transport mediated by KlPdr5p (red) and ScPdr5p (blue) after the addition of clotrimazole. Clotrimazole was added at t = 20 min, Λ values measured and the extent of probe transport inhibition calculated according to Eq. 1. Means and SDs were calculated from five independent repeats.
Clotrimazole addition to cells of both species diminishes the difference of staining levels between the PDR5+ and pdr5– strains, demonstrating the concentration-dependent character of the probe transport inhibition (Fig. 3A and B). In general, staining of MDR pump-deficient cells is dependent solely on their membrane potential (Gaskova et al.2002). The addition of clotrimazole to the pdr5– strains has only a small effect on their staining, indicating slight hyperpolarisation of K. lactis cells (Fig. 3A) and weak depolarisation of S. cerevisiae cells (Fig. 3B). We therefore conclude that the compound does not have a significant effect on membrane potential. Note that since membrane potential is not affected by the presence or absence of Pdr5p (and vice versa) (Gaskova et al.2002), the change in membrane potential caused by clotrimazole (or generally any compound) is the same in pdr5– and PDR5+ cells, i.e. the absolute difference between the staining of the strains is not affected by a change in their membrane potential. To confirm that the increase in staining of Pdr5p-expresing cells after clotrimazole addition is indeed the result of inhibition of probe transport and is not connected to a rise in the fraction of permeabilised cells, we added the diagnostic CD cocktail (10 μM CCCP + 10 μM DM-11). The coincidence of staining curves after the addition of the cocktail clearly demonstrates the absence of permeabilised cells in the sample (Gaskova et al.2013). For illustration, the competition of clotrimazole and the diS-C3(3) probe for transport is also documented by the increased size of the growth inhibition zones when they are applied together (Fig. 3C).
where Λ denotes the emission maximum wavelength in equilibrium, lower indices ‘substrate’ and ‘control’ indicate the presence and absence of a substrate, respectively, and upper indices ‘pdr5–’ and ‘PDR5+’ indicate the absence and presence of Pdr5p, respectively. The second term in the brackets denotes the residual probe pumping activity after the addition of a substrate.
Equation 1 enables us to readily compare the ability of various substrates to inhibit the probe transport mediated by Pdr5p in both studied species. As is evident from Fig. 3D, even a low clotrimazole concentration (1.7 μM) is sufficient to cause complete inhibition of the probe transport in K. lactis, suggesting that clotrimazole binds to the same site(s) as the probe.
On the other hand, the addition of clotrimazole to S. cerevisiae cells, even to the highest used concentration (30 μM), is not sufficient to fully suppress the ScPdr5p-mediated probe transport, amounting to only ∼60% inhibition. This saturating and incomplete competitive inhibition of ScPdr5p-mediated probe transport by clotrimazole is a typical feature for binding pockets with multiple binding sites (Hanson et al.2005).
This saturation contrasts with complete clotrimazole-induced inhibition of ScPdr5p-mediated efflux of rhodamine 6G (Hanson et al.2005) which is known to bind only to site 1 (Golin et al.2003). We therefore conclude that the saturating incomplete inhibition of ScPdr5p-mediated probe transport by clotrimazole is the result of diS-C3(3) probe binding not only to clotrimazole-binding sites 1 and/or site 2, but also to site 3 which does not bind clotrimazole. In this case, incomplete inhibition of probe transport by clotrimazole could be explained by co-transport of the two compounds.
In ScPdr5p binding pocket the probe diS-C3(3) binds to all three identified substrate-binding sites
The multisite nature of ScPdr5p binding pocket was discovered using the approach introduced by Golin et al. (Golin et al.2003; Hanson et al.2005; Golin, Ambudkar and May 2007). Based on the study of competitive inhibition of relatively simple substances, several Pdr5p substrates have been identified that bind specifically only to one of the binding sites. Rhodamine 6G, chloramphenicol, 2-(methyl-trityl-amino)-ethanol and (2-chloro-ethyl)-methyl-trityl-amine defined a single transport site, referred to as site 1. Tritylimidazole and tritylamine are the site 2 substrates, and tetrabutyltin binds only exclusively to site 3 (Golin, Ambudkar and May 2007).
To confirm our statement that diS-C3(3) binds not only to site 1 and/or site 2 but also to site 3 (see above), we used diS-C3(3) fluorescence assay and 2-(methyl-trityl-amino)-ethanol (MTAE), tritylimidazole (TI) and tetrabutyltin (TBT) as benchmark substrates for the single binding sites. The choice of MTAE and TI as substrates for sites 1 and 2, respectively, was given by their minimal direct interaction with the probe. Any strong interaction complicates the interpretation of the results due to shift of λmax towards longer wavelengths (a red shift) and can thus falsely indicate Pdr5p inhibition.
Like clotrimazole, all benchmark substrates of single binding sites cause saturating, incomplete competitive inhibition of ScPdr5p-mediated probe transport, amounting to ∼40%–50% inhibition (Fig. 4A–C). This indicates that the probe binds to all three binding sites, possibly with distinct affinities. Therefore, complete inhibition of probe transport cannot be accomplished by a substrate that binds only to one or two of them.
DiS-C3(3) binds to all three substrate-binding sites of ScPdr5p. Concentration-dependent saturating inhibitory effect of (A) MTAE, site 1 benchmark substrate; (B) TI, site 2 benchmark substrate; and (C) TBT, site 3 benchmark substrate, on ScPdr5p-mediated probe transport, calculated according to Eq. 1. Means and SDs were calculated from six independent repeats. (D) λmax changes in pdr5– and PDR5+ cells with sequential addition of individual benchmark substrates at saturation concentrations, control cells—empty blue symbols. Data are representative of four independent repeats. (E) Extent of inhibition of ScPdr5p-mediated probe transport after addition of benchmark substrates in different combinations, calculated according to Eq. 1. Means and SDs were calculated from four independent repeats.
To clearly demonstrate that the addition of substrates that bind only to two different ScPdr5p binding sites, such as clotrimazole, is not sufficient for a complete competitive inhibition of probe transport, we have performed model experiments in which we added individual substrates in different combinations (Fig. 4D and E). The main function of Fig. 4D is to show the overall trend of λmax changes in ScPdr5p-deficient and ScPdr5p-expressing cells after exposure to individual benchmark substrates and their combinations. As can be seen from Fig. 4E, only the addition of all three individual benchmark substrates leads to complete inhibition of ScPdr5p-mediated probe transport. In addition, it should be emphasised that the combined addition of benchmark substrates for sites 1 and 2 (as well as any other combination) results in about 60% inhibition, comparable with the effect of clotrimazole that binds to sites 1 and 2. Comparison of the extent of probe transport inhibition after the addition of two (60%) and a single substrate (40%–50%) suggests that the increase in inhibition is not merely additive in character. One probable explanation of this incomparability appears to be that the addition of a benchmark substrate of a given binding site also adversely affects the binding of the probe to the remaining ones. This could be caused by a spatial overlap of at least some binding sites, where the presence of one substrate, e.g. TI, reduces the affinity of the second one, i.e. the probe in this case (Pascaud et al.1998; Hanson et al.2005).
DiS-C3(3) is partially co-transported with azoles by ScPdr5p but not KlPdr5p
Our results indicate that the diS-C3(3) probe binds to all identified binding sites of both ScPdr5p and KlPdr5p. This allowed us to use it as a sensitive tool for identifying the binding sites of other azoles (Fig. 5). Experiments analogous to those performed with clotrimazole were carried out with miconazole (Mico), bifonazole (Bifo), ketoconazole (Keto) and itraconazole (Itra).
Azoles are effective inhibitors of probe transport in K. lactis, but not S. cerevisiae. The concentration-dependent inhibitory effect of (A) miconazole (Mico), (B) bifonazole (Bifo), (D) ketoconazole (Keto) and (E) itraconazole (Itra) on Pdr5p-mediated probe transport in K. lactis (red) and S. cerevisiae (blue), calculated according to Eq. 1. Means and SDs were calculated from six independent repeats. (E, F) Comparison of growth inhibition zones of the PDR5+ of both species exposed to azoles, diS-C3(3) (1 mM) and their respective combinations. Stock solutions of azoles for K. lactis/S. cerevisiae: miconazole (5/5 mM), bifonazole (6.5/6.5 mM), ketoconazole (4/7.5 mM) and itraconazole (3/5.7 mM).
The inhibition effectiveness of all tested azoles is comparable to that of clotrimazole, i.e. complete inhibition of KlPdr5p-mediated probe transport and saturating, incomplete competitive inhibition of ScPdr5p (Fig. 5). In the case of miconazole (Fig. 5A) and bifonazole (Fig. 5B), the inhibition of ScPdr5p is the same as that of clotrimazole, i.e. ∼60%. Ketoconazole (Fig. 5D) and itraconazole (Fig. 5E) achieve saturation at a lower extent of inhibition, about 50% and 43%, respectively. Furthermore, the maximum attainable inhibition with ketoconazole and itraconazole is achieved at higher concentrations. In particular, ketoconazole requires eight times higher concentration than miconazole to achieve saturation. The requirement for relatively high concentrations of the last two azoles is documented also by the disc diffusion tests performed with PDR5+ cells of both species (Fig. 5C and F). The combined addition of the probe with either ketoconazole or itraconazole to PDR5+ K. lactis cells causes an enlargement of the growth inhibition zones compared to the addition of the respective substrates alone. In the case of S. cerevisiae, however, no such enlargement is visible (Fig. 5F). This is most likely due to a diffusion-controlled decrease in the azole concentration around the disc below the effective limit.
Various concentrations of azoles are needed to achieve saturated incomplete inhibition of ScPdr5p-mediated probe transport as a consequence of the relative affinity of the probe and the tested azole to the binding sites. Nevertheless, the saturating character itself (Fig. 5) indicates that none of the azoles binds to more than two binding sites of ScPdr5p binding pocket. The probe binding to the remaining one/two sites is sufficient for its transport, although at diminished efficiency. To confirm our conclusion unambiguously, we performed a series of disc diffusion tests on PDR5+ S. cerevisiae cells in which we studied the competition of the azoles with the benchmark substrates of sites 1, 2 and 3 (Fig. 6A). In these experiments, the same benchmark substrates as for the former experiment were used for site 2 (TI) and for site 3 (TBT). MTAE (site 1) was substituted by rhodamine 6G (R6G), as MTAE diffuses poorly through agar. As expected, the tested azoles exhibit binding to only one (itraconazole—site 3) or to two binding sites (clotrimazole, miconazole and ketoconazole—sites 1 and 2; bifonazole—sites 1 and 3), as indicated by the enlargement of the growth inhibition zones after the combined addition of two substrates. A schematic representation of the binding of individual azoles together with diS-C3(3) and benchmark substrates to the binding sites of ScPdr5p is shown in Fig. 6B.
Azoles exhibit binding to only one or two out of the three binding sites in the ScPdr5p binding pocket. (A) Growth inhibition zones of the PDR5+ S. cerevisiae strain exposed to azoles, benchmark substrates of individual binding sites and their respective combinations. Stock solutions of azoles: clotrimazole (4.4 mM), miconazole (3.6 mM), bifonazole (1.6 mM), ketoconazole (7.6 mM) and itraconazole (4.2 mM). Stock solutions of benchmark substrates: rhodamine 6G (10 mM), TI (1 mM) and TBT (2.8 M). Asterisks highlight combinations of azoles and benchmark substrates leading to an enlargement of the growth-inhibition zones. (B) Schematic overview of the binding sites of tested azoles, fluorescent probe diS-C3(3) and benchmark substrates in the ScPdr5p binding pocket.
KlPdr5p has only one detectable binding site
Complete competitive inhibition of KlPdr5p-mediated probe transport by all tested azoles indicates that, unlike in the case of ScPdr5p, the probe binds to the same binding site(s) as the azoles. In other words, there is no additional binding site in KlPdr5p that would be exclusive to the probe.
To get more information on the number of binding sites in KlPdr5p, we performed experiments analogous to those with ScPdr5p. As we found that the benchmark substrates of the individual binding sites of ScPdr5p (MTAE, TI and TBT) are also KlPdr5p substrates (Fig. 7A–C), we tested their inhibition effectiveness of probe transport by KlPdr5p (Fig. 7D–F). In contrast to the case of ScPdr5p, all benchmark substrates cause complete competitive inhibition of probe transport indicating that they bind to the same binding site(s) of the binding pocket as the probe and azoles. In addition, simultaneous addition of azoles in various combinations to a paper disc in a disc diffusion test always results in an enlargement of growth inhibition zones compared to that of the addition of the respective azole alone. This indicates that they are binding to the same binding site(s). When the probe is added to the same paper disc as the two azoles, an additional increase in the zone size is observed in all cases (Fig. 7G). All of these data indicate the existence of only one detectable binding site for substrate recognition and binding in the KlPdr5 binding pocket (Fig. 7H). This finding is indeed unexpected and interesting because KlPdr5p transporter (1525 amino acids) shares 63.8% aa sequence identity with ScPdr5p (1522 amino acids). In addition, structurally, these two proteins have a similar molecular architecture with the same organisation of their functional domains (Chen 2001). Although these pumps do not display completely overlapping substrate specificities, amino acids whose mutations lead to loss of transport activity or decreased expression of ScPdr5p are strictly conserved in KlPdr5p (Chen 2001; Dou et al.2016).
All KlPdr5p substrates bind to the same binding site(s) in the binding pocket. Growth inhibition zones of pdr5– and PDR5+ K. lactis cells exposed to (A) MTAE (1 mM), (B) TI (0.1 mM) and (C) TBT (0.9 M). The concentration-dependent inhibitory effect of (D) MTAE, (E) TI and (F) TBT on Pdr5p-mediated probe transport in K. lactis cells, calculated according to Eq. 1. Means and SDs were calculated from four independent repeats. (G) Growth inhibition zones of the PDR5+ K. lactis strain exposed to azoles and their respective combinations with and without the addition of the fluorescent probe diS-C3(3) (1 mM). Stock solutions of azoles: clotrimazole (0.01 mM), miconazole (0.02 mM), ketoconazole (0.03 mM), itraconazole (0.1 mM) and bifonazole (0.2 mM). (H) Schematic illustration of the same binding site(s) of all tested substrates in the KlPdr5p binding pocket.
Taken together, the published knowledge and our results indicate that the arrangement of the binding sites in KlPdr5p binding pocket is tighter than that of ScPdr5p. This in turn means that an already bound substrate may sterically hinder the binding of another to a distinct binding site.
The specificity of individual binding domains as well as actual arrangement of the Pdr5p binding pocket, particularly in KlPdr5p, could be elucidated by studying Pdr5p mutants with exchanged binding sites and will be a challenge for our future studies.
KlPdr5p and ScPdr5p binding pockets differ in both size and number of binding sites
The higher number of distinct binding sites in the ScPdr5p binding pocket compared to KlPdr5 naturally evokes the idea of its larger cavity size. We have tried to prove this by using a relatively bulky substrate of ScPdr5p, enniatin B (Hiraga et al.2005), that we have previously found to effectively inhibit the probe transport by ScPdr5p (Hendrych et al.2009), see also Fig. 8A. As seen in Fig. 8A, enniatin B completely inhibits the ScPdr5p-mediated probe transport already at the concentration of 5 μM. On the other hand, its effect on KlPdr5p is negligible. The difference in inhibition efficiency is further demonstrated by the size of growth inhibition zones after the probe addition to PDR5+ cells of both species in both the presence and absence of enniatin B (Fig. 8B). While the combined addition of the diS-C3(3) probe and enniatin B gives rise to a growth inhibition zone in S. cerevisiae, there are no zones in K. lactis regardless of the identity of the used compounds or their combination (Fig. 8B). Furthermore, the effect of the combined addition of enniatin B with the studied azoles is in character similar to its combined addition with the probe, i.e. it leads to an enlargement of growth inhibition zones only in the case of ScPdr5p (Fig. 8C). A possible explanation for the lack of effect of enniatin B on KlPdr5p-mediated transport of the probe or other substrates could be that enniatin B does not enter these cells. However, we excluded this possibility due to the fact that this ionophore antibiotic (Hiraga et al.2005) causes slight depolarisation of K. lactis cells, which is even greater than that of S. cerevisiae cells. This assertion is based on the decrease in the staining curves of pdr5– cells of both species after addition of enniatin B, which are not shown here.
Enniatin B, a bulky substrate of ScPdr5p, has a negligible inhibitory effect on KlPdr5p-mediated transport of the probe and azoles. (A) The concentration-dependent inhibitory effect of enniatin B on Pdr5p-mediated probe transport in K. lactis (red) and S. cerevisiae (blue), calculated according to Eq. 1. Means and SDs were calculated from four independent repeats. Comparison of growth inhibition zones of the PDR5+ K. lactis and S. cerevisiae strains exposed to (B) enniatin B, diS-C3(3) (1 mM) and their respective combination and to (C) enniatin B, individual tested azoles and their respective combination. Stock solutions for K. lactis/S. cerevisiae: enniatin B (0.13/0.8 mM), ketoconazole (0.03/1.9 mM), clotrimazole (0.01/4.4 mM), bifonazole (0.2/3.2 mM), itraconazole (0.1/2.8 mM) and miconazole (0.03/1.2 mM). (D) Schematic representation of the different sizes of the binding pocket of ScPdr5p and KlPdr5p demonstrating different requirements for the size of their substrates.
Taken together, both the diS-C3(3) transport inhibition data (Fig. 8A) and the results of the disc diffusion assay performed with the probe (Fig. 8B) and with azoles (Fig. 8C) document the inability of the relatively large molecule enniatin B to inhibit the transport of KlPdr5p substrates and indicate that the molecule is too bulky to be accommodated by the KlPdr5p binding pocket (see Fig. 8D). On the other hand, it is large enough to completely block probe binding to all three ScPdr5p binding sites (or binding of other substrates, such as azoles, interacting with multiple binding sites in the pocket) (Fig. 8D). This drug size-dependent cut-off limit resembles the cut-off effect observed in the case of alcohol-induced anaesthesia caused by their interaction with receptors (Pringle, Brown and Miller 1981; Franks and Lieb 1984) and supports the proposed difference in the size of the binding pockets of ScPdr5p and KlPdr5p.
CONCLUSIONS
We compared the ability of different azoles to competitively inhibit the ScPdr5p- and KlPdr5p-mediated diS-C3(3) probe transport. Although these MDR pumps share numerous substrates (such as tested azoles and the diS-C3(3) fluorescent probe), it was not clear whether they have the same arrangement of their binding pockets. The possible differences include the number of binding sites and the overall size of the binding pocket, which seem to be of great importance for substrate–transporter interactions. Our data indicate that the substrate specificities and affinities for various substrates of the two pumps differ. We demonstrate that the Pdr5p binding pocket in Saccharomyces cerevisiae houses more distinct binding sites and is larger than that of Kluyveromyces lactis, as schematically illustrated in Fig. 9, summarising both the already published knowledge and the findings presented in the current study.
Schematic overview of both the already published knowledge of the ScPdr5p and KlPdr5p pumps and the findings presented in the current study.
The fact that the diS-C3(3) fluorescent probe, used as a benchmark substrate in the study, is also expelled by Cdr1p, a Candida albicans homologue of ScPdr5p (Szczepaniak, Lukaszewicz and Krasowska 2015), and by the mammalian transporter P-gp (also homologous to ScPdr5p) (unpublished data) makes our approach very promising for future studies of multiple transport sites of these medically important MDR pumps in relation to substrates with chemically complex structure such as enniatin B.
Acknowledgements
We acknowledge Andre Goffeau (Universite Catholique de Louvain, Louvain-la-Neuve, Belgium) for the isogenic mutant strains.
FUNDING
This work was supported by Charles University [SVV 260 323], EU supported Operational Programme ‘Research and Development for Innovation’ [OP VaVpI no. CZ.1.05/4.1.00/16.0340], Slovak Research and Development Agency [APVV-0282-10] and VEGA [1/0077/14 and 2/0111/15].
Conflicts of interest. None declared.
