Current Genetics

, Volume 50, Issue 4, pp 247–255

Expression of heterologous aquaporins for functional analysis in Saccharomyces cerevisiae

Authors

  • Nina Pettersson
    • Department of Cell and Molecular BiologyGöteborg University
  • Johan Hagström
    • Department of Cell and Molecular BiologyGöteborg University
  • Roslyn M. Bill
    • School of Life and Health SciencesAston University
    • Department of Cell and Molecular BiologyGöteborg University
Research Article

DOI: 10.1007/s00294-006-0092-z

Cite this article as:
Pettersson, N., Hagström, J., Bill, R.M. et al. Curr Genet (2006) 50: 247. doi:10.1007/s00294-006-0092-z

Abstract

In this study the yeast Saccharomyces cerevisiae, which is a genetically tractable model for analysis of osmoregulation, has been used for analysis of heterologous aquaporins. Aquaporin water channels play important roles in the control of water homeostasis in individual cells and multicellular organisms. We have investigated the effects of functional expression of the mammalian aquaporins AQP1 and AQP5 and the aquaglyceroporins AQP3 and AQP9. Expression of aquaporins caused moderate growth inhibition under hyperosmotic stress, while expression of aquaglyceroporins mediated strong growth inhibition due to glycerol loss. Water transport was monitored in protoplasts, where the kinetics of bursting was influenced by presence of aquaporins but not aquaglyceroporins. We observed glycerol transport through aquaglyceroporins, but not aquaporins, in a yeast strain deficient in glycerol production, whose growth depends on glycerol inflow. In addition, a gene reporter assay allowed to indirectly monitor the effect of AQP9-mediated enhanced glycerol loss on osmoadaptation. Transport activity of certain aqua(glycero)porins was diminished by low pH or CuSO4, suggesting that yeast can potentially be used for screening of putative aquaporin inhibitors. We conclude that yeast is a versatile system for functional studies of aquaporins, and it can be developed to screen for compounds of potential pharmacological use.

Keywords

Water transportGlycerolAquaglyceroporinInhibitorsYeastOsmoadaptation

Introduction

Maintenance of the water balance is crucial for all living organisms. Historically, water was believed to diffuse freely over cell membranes driven by osmosis, but rapid, inhibitor-sensitive water movements in certain cells suggested the existence of water channels. Such water channels, aquaporins, were later identified and characterized in erythrocytes and kidney (Preston et al. 1992). Aquaporins have since then been described in organisms ranging from archea to human (Hohmann et al. 2001).

Thirteen aquaporins have been identified in mammals and they are expressed in a tissue-specific manner. The aquaporin family is divided into orthodox aquaporins and aquaglyceroporins. The aquaporins, AQP0, 1, 2, 4, 5, 6 and 8 are highly specific for water (Takata et al. 2004), with exception for AQP6 and 8, which additionally transport nitrate (Ikeda et al. 2002) and urea (Ma et al. 1997), respectively. The aquaglyceroporins, AQP3, 7, 9 and 10, mediate transport of glycerol, other polyols and in some cases urea and even arsenic (Liu et al. 2002; Carbrey et al. 2003). AQP11 and 12 have more limited sequence homology to the aquaporin family and are presently poorly characterized (Morishita et al. 2004).

Aquaporins appear to have physiological roles relevant for numerous health conditions (King et al. 2000). For instance, it has recently been reported that human AQP1 is involved in cell migration, which may have implications in cancer metastasis (Saadoun et al. 2005). In addition, it is well known that modulators affecting expression or function of aquaporins in the kidney could serve as diuretics (Verkman 2005), blockers of AQP4 could help preventing fatal brain swelling upon stroke and trauma (Manley et al. 2000), and blockers of AQP5 could help balancing excessive sweat production associated with certain diseases (Nejsum et al. 2002). Aquaporins also play potential roles in body energy balance and hence obesity and metabolic syndrome, since AQP7 is proposed to mediate release of glycerol from adipose cells which subsequently is taken up as a source for energy into the liver by a process that may involve AQP9 (Kuriyama et al. 2002; Carbrey et al. 2003). Both, for further studies as well as for pharmacological intervention of water homeostasis, there is a need for specific inhibitors of aquaporins, which presently do not exist. In this study we have investigated the use of Saccharomyces cerevisiae as a system to study mammalian aquaporins.

S. cerevisiae itself has four proteins belonging to the aquaporin family, two orthodox aquaporins (Aqy1 and Aqy2) and two aquaglyceroporins (Fps1 and Ylf054c) (Hohmann et al. 2000). Curiously, Aqy1 and Aqy2 are inactivated by spontaneous mutations in most commonly-used lab strains (Bonhivers et al. 1998; Laize et al. 2000), without apparent phenotypic effects. Fps1 is a gated aquaglyceroporin, involved in yeast osmoregulation. When yeast cells are exposed to a hyperosmotic shock they shrink due to water loss and compensate this effect by accumulating glycerol. Glycerol accumulation is mediated by different effects: (1) Stimulated expression of GPD1 and GPP2 encoding enzymes in glycerol biosynthesis (Albertyn et al. 1994; Pahlman et al. 2001) (2) Stimulated glycolytic activity by activation of 6-phosphofructo-2-kinase (Dihazi et al. 2004) (3) Closing of the Fps1 export channel (Luyten et al. 1995). Many, but probably not all, of these effects are mediated by the high osmolarity glycerol (HOG) pathway, one of the yeast mitogen activated protein kinase (MAPK) signaling systems (Hohmann 2002). The biological function of the second yeast aquaglyceroporin, Yfl054c, is presently unknown, although it is well conserved among fungi (Pettersson et al. 2005).

In this work we have employed knowledge of yeast osmoregulatory mechanisms to study the aquaporins AQP1 and AQP5 (from human and rat, respectively) and the aquaglyceroporins AQP3 and AQP9 (both from rat).

Materials and methods

Yeast strains and plasmids

The S. cerevisiae strains used in this study are W303-1A (MATa leu2-3/112 ura3-1 trp1-1 his3-11/15 ade2-1 can1-100 GAL SUC2 mal0) (Thomas and Rothstein 1989) plus its isogenic mutants YSH642 (gpd1Δ::TRP1gpd2Δ::URA3) (Ansell et al. 1997) and YSH690 (gpd1Δ::TRP1) (Tamas et al. 2003). For the CAT-reporter assay (see below) strain UTL7A (MATa. ura3-52, trp1, leu2-3/112, gpd1Δ::CAT-URA3) (Eriksson et al. 1995) was used. The plasmid with human AQP1 inserted into pYEDP10 (multicopy 2 μ plasmid, GAL10-CYC1 promoter, URA3 marker) was kindly provided by Vincent Laizé (Laize et al. 1995). This plasmid was used as a template as was rat AQP3, AQP9 (kindly provided by Søren Nielsen) and AQP5 (kindly provided by Peter Deen) for amplification by primers available on request (PCR). The PCR products were inserted between the EcoRI and SmaI sites of the pYX242 vector (multi-copy 2 μ vector, constitutive TPI1 promoter, LEU2 marker, HA-tag). The single mutation in AQP1, A73M, was introduced using the megaprimer method (Sarkar and Sommer 1990) (primers available on request). Transformation of plasmids into yeast cells was performed using the Lithium one step transformation protocol (Chen et al. 1992).

Growth conditions

Yeast cells were grown in 2% peptone, 1% yeast extract, 2% glucose (YPD). Selection and growth of transformants was performed in synthetic medium (YNB, 2% glucose) lacking leucine. The synthetic media was buffered with Na/Succinate and pH adjusted with NaOH. Glucose was replaced with 2% galactose to induce expression from the GAL-promoter. For growth assays cells were pregrown for 2 days on YNB plates, then resuspended in YNB to an OD600 nm = 0.2 and diluted in a tenfold dilution series, of which 5 μl were spotted onto agar plates supplemented with osmotica and inhibitors as indicated. Growth was monitored for 2–5 days in 30°C.

Membrane preparation and immunoblots

Transformed yeast cells were harvested in mid-exponential phase, washed (10 mM Tris–HCl pH 7.5, 0.5 M sucrose, 2.5 mM EDTA) and resuspended in homogenization buffer [50 mM Tris–HCl pH 7.5, 0.3 M sucrose, 5 mM EDTA, 1 mM EGTA, 5 mg/ml BSA, 2 mM DTT, protease inhibitor cocktail (Roche)]. Total membrane preparations were made by disrupting cells in a Fast-prep (BIO101), centrifugation at 10,000 × g for 10 min and then the supernatant underwent centrifugation at 100,000 × g for 90 min. Plasma membanes were prepared as in Tamàs et al. (1999). The pellets were resuspended (10 mM Tris–HCl pH 7.0, 1 mM EGTA, 1 mM DTT, 20% (v/v) glycerol, protease inhibitor cocktail), 15 μg protein was denatured (10 min, 65°C, 40 mM DTT) separated by SDS-PAGE and blotted (Hybond-ECL, Amersham). Membranes were blocked with PBS-5% milk and probed for 1.5 h with 1:2,000 diluted anti-HA mouse monoclonal antibody (Roche), or AQP1 serum (kindly provided by Vincent Laizé) washed and incubated for 1 h with secondary antibody [HRP-conjugated anti-mouse and anti-rabbit IgG (Promega), respectively], diluted 1:2,500, in PBS-5% milk. Membranes were incubated with Lumi-Light for detection. Quantifications of western signals were done using the ImageQuant program.

Hog1-phosphorylation Westerns were done as previously described (Karlgren et al. 2005). Prior to antibody incubation, membranes were stained with Ponceau for control of equal protein loading (data not shown).

CAT-reporter assay

Transformed cells were harvested at mid-exponential phase, washed, resuspended in 100 mM Tris, pH 8.0 and disrupted by glass beads in a Fast-prep (BIO101). After centrifugation at 12,000 × g for 5 min, the supernatant containing the intracellular proteins was collected and protein concentrations determined. To a total volume of 250 μl (1.25 mM Chloramphenicol, 100 mM Tris, pH 8.0) 15 μg of protein was added. The mixture was transferred to a scintillation tube, 0.1 μCi 3H-Acetyl-Coenzyme-A and 5 ml scintcocktail (Econofluor-2) was added. The radioactivity was monitored over 2 h and specific GPD1-promoter activity calculated as the average of the initial slope of three independent measurements.

Water transport assay

Transformed cells were grown to mid logarithmic phase, harvested, washed once in water and once in 1 M Sorbitol, then suspended in SCE buffer (1 M Sorbitol, 0.1 M Sodium Citrate, 10 mM EDTA, 0.2 mM β-mercaptoethanol, pH 6.8) containing 2,000 U of lyticase (Sigma) per milliliter of culture and incubated shaking for 3 h at 30°C. After confirming protoplast formation microscopically, protoplasts were harvested, washed twice and resuspended in STC buffer (1 M Sorbitol, 10 mM Tris–HCl, pH 7.5, 10 mM CaCl2). Protoplasts were diluted to 0.5 M sorbitol at t = 0 and OD600 nm was monitored at subsequent time points as indicated. Values are an average of three independent transformants and each data point has been normalized to its own starting OD.

Results

Heterologous aquaporins expressed in yeast cells are localized to the membrane

Yeast cells were transformed with plasmid constructs for expression of mammalian aqua(glycero)porins. Expression and membrane localization was verified by Western blot analysis (Fig. 1). AQP1 appeared as a band of around 28 kDa as expected for its monomeric form. AQP5 also migrated as monomers, slightly faster than AQP1. This behavior has been observed previously, although the predicted size of AQP5 is 29 kDa (Raina et al. 1995). AQP3 and AQP9 migrated as oligomers, a feature commonly observed for aquaglyceroporins in yeast (Liu et al. 2004). AQP3 is present as dimers and trimers, while AQP9 in addition appear as tetramers, higher oligomers and a weak band of the size of the monomer. For unknown reasons, the apparent membrane expression levels of AQP1, 3, 5 and 9 were very different.
https://static-content.springer.com/image/art%3A10.1007%2Fs00294-006-0092-z/MediaObjects/294_2006_92_Fig1_HTML.gif
Fig. 1

Membrane localization of heterologously expressed aquaporins in yeast. Western blot analysis of the total cell membrane fraction from cells expressing aquaporins, using an anti-HA antibody. Monomers appear at about 30 kDa. The upper bands represent dimers, trimers, tetramers and higher oligomers

Aquaporin expression affects yeast growth

Under normal growth conditions (YNB), yeast cells expressing aqua(glycero)porins grew indistinguishably from those transformed with an empty plasmid. When exposed to hyperosmotic stress, cells expressing AQP3, 5 and 9 displayed reduced growth to a variable degree, an effect that was independent of the solute used (Fig. 2). The strain used in these experiments lacked GPD1, one of the two isoforms of glycerol-3-phosphate dehydrogenase needed for glycerol production. Such mutants are slightly osmosensitive (Larsson et al. 1993; Albertyn et al. 1994) and were employed to enhance the effects of expressed aquaporins. Growth inhibition under hyperosmotic stress was also observed in aquaporin-transformed wild type cells, although the effects were weaker (data not shown). Cells expressing the aquaglyceroporins AQP3 and 9 were more sensitive than cells expressing the orthodox aquaporin AQP5, probably because loss of the osmolyte glycerol is more deleterious than enhanced water transport. In this assay we did not observe any effect following AQP1 expression (data not shown; see further).
https://static-content.springer.com/image/art%3A10.1007%2Fs00294-006-0092-z/MediaObjects/294_2006_92_Fig2_HTML.gif
Fig. 2

Expression of aquaporins in yeast suppress growth at high osmolarity. Yeast cells (mutant gpd1Δ) expressing AQP3, AQP5 or AQP9 show decreased growth as compared to control cells, transformed with an empty vector (pYX242), at high osmolarity. Transformants were spotted in 1:10 dilution series on defined media supplemented with NaCl, KCl or sorbitol as indicated and incubated for 2–5 days at 30°C. Each experiment has been repeated at least three times and a representative example is shown

Yeast protoplasts expressing orthodox water channels burst faster upon hypo-osmotic shock

To monitor water flux through the aqua(glycero)porins used in this study, we prepared protoplasts and monitored the change in optical density (OD) at 600 nm following a hypo-osmotic shock. Such a treatment causes water influx and protoplast bursting, which in turn is recorded as a drop in optical density. Protoplasts expressing the aquaporins AQP1 or AQP5 burst much quicker than protoplasts transformed with empty plasmid (Fig. 3). The aquaglyceroporins do not seem to mediate water transport in this assay as they do not, or only minimally, affect bursting kinetics. Interestingly, while we did not observe a growth phenotype for cells expressing AQP1, the bursting assay indicated similar water transport kinetics for AQP1 and AQP5-expressing cells.
https://static-content.springer.com/image/art%3A10.1007%2Fs00294-006-0092-z/MediaObjects/294_2006_92_Fig3_HTML.gif
Fig. 3

Yeast protoplasts expressing orthodox aquaporins burst faster than control protoplasts when exposed to a hypo-osmotic stress. The cell wall of yeast cells (gpd1Δ) expressing AQP1, AQP3, AQP5 or AQP9 was digested, and the resulting protoplasts subjected to a hypo-osmotic shock from 1.0  to 0.5 M Sorbitol. Bursting of protoplasts as a consequence of water inflow was monitored as a decrease in optical density (600 nm) for 2 min. The starting value for each transformant was set to 1.0 and all further data were normalized to this value. An average of three independent experiments (± standard deviation) is presented

Yeast cells expressing AQP1 from a galactose inducible promoter show osmosensitivity and delayed HOG-signaling

AQP1-expressing cells showed enhanced water transport in the protoplast bursting assay, similar to AQP5-expressing cells, but did not display an osmosensitive phenotype when the gene was under the control of a constitutive promoter. To exclude any compensatory effects mounted by the cell in response to constitutive AQP1 expression, we employed a galactose inducible promoter. The protein expressed from GAL10-AQP1 was present in the plasma membrane while a mutant version, A73 M, which is known to be defective in water transport [although properly localized in Xenopus oocytes (Jung et al. 1994)] was not (Fig. 4a). Indeed, AQP1-expressing cells displayed an osmosensitive phenotype on galactose but not on glucose medium, while AQP1-A73 M-expressing cells did not exhibit osmosensitivity (Fig. 4b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00294-006-0092-z/MediaObjects/294_2006_92_Fig4_HTML.gif
Fig. 4

Expression of human AQP1 under an inducible GAL promoter mediates a hyperosmotic sensitivity phenotype. a Expression of AQP1 and the AQP1-A73 M mutant in total and plasma membrane extracts prepared from cells transformed with plasmids as indicated. Immunoblots were performed in triplicate, and expression levels were quantified relative to the negative control, and normalized to AQP1. The data are presented as mean ± SE (n = 3). b Cells expressing AQP1, but not AQP1-A73 M, are sensitive to a hyperosmotic shock. Cells were spotted in 1:10 dilution on synthetic growth media supplemented with the indicated compounds and incubated for 2–5 days at 30°C. c The phosphorylation profile of dually phosphorylated Hog1 after a hyperosmotic shock is delayed in cells expressing AQP1, but not AQP1-A73 M

Cells that are osmosensitive have previously been reported to display a delay in initiating signaling through the HOG pathway, which monitors osmotic changes (Karlgren et al. 2005; Klipp et al. 2005). Consistent with the observed growth phenotypes, expression of AQP1, but not AQP1-A73 M caused a reproducible delay in increasing the level of dually phosphorylated Hog1 (Fig. 4c), which is an indicator for HOG pathway activity.

Evidence for glycerol transport through heterologous aquaglyceroporins in yeast

We have previously developed a system of “conditional osmotic stress” (Karlgren et al. 2005). In this system the inability of yeast mutants unable to produce any glycerol (gpd1Δ gpd2Δ) to grow in the presence of high glycerol levels is suppressed by expression of an active aquaglyceroporin. Suppression is due to glycerol influx, relieving osmotic stress. This assay provided evidence for activity of AQP3 and AQP9, which both allowed the mutant to grow in the presence of 2 M glycerol. As expected, the aquaporin AQP5 was unable to suppress the growth defect of the gpd1Δ gpd2Δ mutant on 2 M glycerol (Fig. 5).
https://static-content.springer.com/image/art%3A10.1007%2Fs00294-006-0092-z/MediaObjects/294_2006_92_Fig5_HTML.gif
Fig. 5

Expression of aquaglyceroporins suppresses the growth defect of the gpd1Δgpd2Δ mutant at high glycerol levels. This is demonstrated for AQP3 and AQP9, which clearly grows in the presence of high glycerol concentration, while AQP5 doesn’t. Transformants were spotted in 1:10 dilution series on defined media supplemented with NaCl, KCl or sorbitol as indicated and incubated for 3–5 days at 30°C. Each experiment was repeated at least three times and a representative example of these is shown

We have observed previously that mutants defective in glycerol accumulation, for example because of glycerol leakage through an active aquaglyceroporin, are osmosensitive and show prolonged expression of genes stimulated by osmostress (Klipp et al. 2005). A strain expressing the E. coli chloramphenicol acetyl transferase (CAT) reporter gene coupled to the GPD1 promoter was used to follow GPD1-promoter activity. The activity was increased in AQP9-expressing cells, but not in AQP1, AQP3 or AQP5-expressing cells after 2 h of exposure to hyperosmotic stress (Fig. 6). This is consistent with the observation that AQP9 causes strongest osmosensitivity, although the difference to the sensitivity caused by AQP3 is small.
https://static-content.springer.com/image/art%3A10.1007%2Fs00294-006-0092-z/MediaObjects/294_2006_92_Fig6_HTML.gif
Fig. 6

Cells expressing AQP9 have higher GPD1-promoter activity when exposed to a hyperosmotic shock, as measured in a GPD1-CAT reporter assay than control cells transformed with the empty plasmid (pYX242) do. The other aquaporins tested (AQP1, 3 and 5) do not show comparable elevated activity

Inhibition of aquaporin function can be recorded in yeast cells

Several compounds that inhibit aquaporins non-selectively have been reported, including HgCl2 (Preston et al. 1992), CuSO4 (Zelenina et al. 2004), AgNO3 (Niemietz and Tyerman 2002) and tetraethylammonuim (TEA) chloride (Brooks et al. 2000). To evaluate if a screening system for inhibitors could be set up in yeast we used such compounds in growth assays. The inhibitor most commonly used for aquaporins is Hg2+, but since this compound is toxic for yeast cells at inhibitory concentrations (data not shown) we focused on CuSO4, AgNO3 and TEA (Fig. 7).

Cells expressing AQP3 or AQP5, but not AQP9, grew better after osmotic shock with either sorbitol or KCl in the presence of CuSO4 (Fig. 7a). In the presence of AgNO3 the hyper-osmosensitive phenotype was suppressed on NaCl but not on sorbitol (data not shown). This observation was somewhat surprising because a large part of the silver ions are probably precipitated as AgCl under such conditions and little Ag+ is freely available. No suppression was observed when TEA was tested (data not shown).
https://static-content.springer.com/image/art%3A10.1007%2Fs00294-006-0092-z/MediaObjects/294_2006_92_Fig7_HTML.gif
Fig. 7

The hyperosmotic growth defect can be suppressed by either low pH or by addition of the known aquaporin inhibitor CuSO4. a The osmosensitivity of cells expressing AQP3 or AQP5, but not AQP9, is suppressed by the addition of 1 mM CuSO4 to the medium. b Cells expressing AQP3 or AQP5, but not AQP9, are more sensitive to hyperosmotic stress at pH 7 than pH 6. Cells of the gpd1Δ strain transformed with an empty vector (pYX242) and the same plasmid mediating expression of the indicated aquaporins were spotted in 1:10 dilution series on synthetic growth media supplemented with different compounds and incubated for 2–5 days at 30°C. Each experiment has been repeated at least three times and a representative example is shown

Since it has been reported that AQP3 is inhibited at acidic pH (Zelenina et al. 2003), we tested the effect of different medium pH on AQP-mediated growth inhibition. Note that the control strain grew less well in medium with pH 7 as compared to pH 6. Still, AQP3-mediated growth inhibition was clearly stronger at pH 7 as compared to pH 6 (Fig. 7b). The same seems to be true for AQP5, which has previously been reported to be as active in acidic pH as at physiological pH (Zelenina et al. 2003). AQP9 mediates a strong growth inhibition already at pH 6 and hence it is difficult to judge if the stronger effect at pH 7 is due to the generally poorer growth under these conditions or to a pH effect on AQP9.

Discussion

Numerous heterologous membrane proteins have been expressed in yeast for functional studies (reviewed in Bill 2001). Those include sensors and receptors such as the pharmacologically important human G-protein coupled receptors (GPCRs) (Ladds et al. 2005) or the plant cytokinin receptor Cre1 (Reiser et al. 2003). In addition, heterologous transport proteins have been functionally expressed in yeast, examples include transporters for glucose (Wieczorke et al. 2003), amino acids (Matejckova-Forejtova et al. 1999), Ca2+ (Hashimoto et al. 2004) as well as the cystic fibrosis transmembrane conductance regulator (CFTR) (Kiser et al. 2001). S. cerevisiae has also been employed in a high throughput screening assay for inhibitors of, for instance, the mammalian fatty acid transporter mmFATP2 (Li et al. 2005). Members of the aquaporin family have previously been expressed and assayed for direct uptake of radiolabelled polyols (Karlgren et al. 2005), for water transport in yeast secretory vesicles (Coury et al. 1998) or in yeast membrane vesicles (Suga and Maeshima 2004) and for arsenic uptake both directly and in a growth assay (Liu et al. 2004) in S. cerevisiae. Water transport has additionally been monitored in spheroplasts using the yeast Pichia pastoris (Daniels et al. 2006).

In this study we have expressed two water-specific aquaporins and two aquaglyceroporins in yeast wild type cells as well as in cells impaired in glycerol production. Water transport was verified directly using a protoplast bursting assay. Expression of AQP1 and AQP5 caused moderate hyper-osmosensitivity, probably due to more rapid water loss. We did not observe enhanced hypo-osmosensitivity caused by expression of AQP1 in a plate assay (data not shown), probably because the flexible cell wall can control cell swelling and prevent bursting. We note that AQP1-mediated osmosensitivity was not observed using the constitutive expression construct that was functional for AQP5, but was instead observed only with a galactose-inducible construct. Probably cells adapt to AQP1 expression with time. The AQP5-mediated hyper-osmosensitivity was relieved partly by known inhibitors of aquaporin function such as CuSO4 as well as lower pH, indicating that the observed effects are aquaporin-specific and that they probably can be used for screening and testing of novel inhibitors. At the same time we note that we have been unable to score any effects by TEA, an established inhibitor functioning in Xenopus oocytes. It appears that for unknown reasons aquaporins expressed in yeast are not accessible to TEA or that the effect of TEA [20–40% inhibition (Brooks et al. 2000)] is insufficient to score in the yeast system.

Glycerol transport through AQP3 and AQP9 was verified using the previously established system of conditional osmotic stress (Karlgren et al. 2005), which makes use of suppression of the sensitivity to high glycerol levels of a mutant unable to produce glycerol itself. Hyper-osmosensitivity, likely due to glycerol efflux during osmoadaptation (Tamas et al. 1999), was also observed in AQP3 and AQP9 expressing wild type cells or cells with diminished capacity to produce glycerol. In addition, a gene expression reporter system scored the effect of delayed osmoadaptation due to AQP9- but not AQP3-expression. Probably AQP3-mediated glycerol export rates are sufficient to cause a certain degree of osmosensitivity on plates but are not sufficient to stimulate enhanced gene expression activity. AQP3-, but not AQP9-mediated osmosensitivity was strongly affected by CuSO4 as well as lower pH. This suggests that yeast assays can be used to score effects of inhibitory compounds on aquaglyceroporins as well and that the compounds tested here may display specificity towards certain aquaglyceroporins. In addition, it appears that yeast growth assays, as used here, will have to be fine-tuned with respect to expression levels and compound concentrations for each specific protein/compound pair.

Taken together, we have demonstrated that aquaporin and aquaglyceroporin expression in yeast causes distinct phenotypes. These phenotypes can be used to study the function of the proteins, to monitor the effects of mutations and to score effects of inhibitors. The yeast system holds potential for being used as a screening tool in the search of new aquaporin inhibitors, even though it would need to be complemented with other systems, such as Xenopus oocytes, for subsequent verification and quantitative characterization of the inhibitor.

Acknowledgments

We thank members of the group for helpful discussions and for critical reading of the manuscript. This work was supported by grants from the European Commission (contracts BIO4-CT98-0024 and FMRX-CT96-0128 to SH and contract LSHG-CT-2004-504601 to RMB) and the Swedish Research Council (research position and research grant to SH).

Copyright information

© Springer-Verlag 2006