ABC Transporter Pdr10 Regulates the Membrane Microenvironment of Pdr12 in Saccharomyces cerevisiae
- First Online:
- Cite this article as:
- Rockwell, N.C., Wolfger, H., Kuchler, K. et al. J Membrane Biol (2009) 229: 27. doi:10.1007/s00232-009-9173-5
- 593 Downloads
The eukaryotic plasma membrane exhibits both asymmetric distribution of lipids between the inner and the outer leaflet and lateral segregation of membrane components within the plane of the bilayer. In budding yeast (Saccharomyces cerevisiae), maintenance of leaflet asymmetry requires P-type ATPases, which are proposed to act as inward-directed lipid translocases (Dnf1, Dnf2, and the associated protein Lem3), and ATP-binding cassette (ABC) transporters, which are proposed to act as outward-directed lipid translocases (Pdr5 and Yor1). The S. cerevisiae genome encodes two other Pdr5-related ABC transporters: Pdr10 (67% identity) and Pdr15 (75% identity). We report the first analysis of Pdr10 localization and function. A Pdr10-GFP chimera was located in discrete puncta in the plasma membrane and was found in the detergent-resistant membrane fraction. Compared to control cells, a pdr10∆ mutant was resistant to sorbate but hypersensitive to the chitin-binding agent Calcofluor White. Calcofluor sensitivity was attributable to a partial defect in endocytosis of the chitin synthase Chs3, while sorbate resistance was attributable to accumulation of a higher than normal level of the sorbate exporter Pdr12. Epistasis analysis indicated that Pdr10 function requires Pdr5, Pdr12, Lem3, and mature sphingolipids. Strikingly, Pdr12 was shifted to the detergent-resistant membrane fraction in pdr10∆ cells. Pdr10 therefore acts as a negative regulator for incorporation of Pdr12 into detergent-resistant membranes, a novel role for members of the ABC transporter superfamily.
KeywordsPdr10ABC transporterSphingolipidLipid raftDetergent-resistant membraneLateral segregationPdr12Lem3Dnf1Dnf2
It has been appreciated for quite some time that the two leaflets of a eukaryotic plasma membrane have distinct lipid compositions (Devaux 1991; Gordesky and Marinetti 1973). Now there is much evidence for additional layers of structural complexity in the plasma membrane. For example, fluorescently tagged annexins with different lipid specificities will label distinct regions of the plasma membrane in smooth muscle cells (Draeger et al. 2005). Such distinct subdomains in the plasma membrane are also reminiscent of much more specialized structures, like the rhodopsin-rich photosensory discs in the vertebrate retina. Similarly, the plasma membrane of the yeast Saccharomyces cerevisiae contains at least two microscopically distinct subdomains (Malinska et al. 2003, 2004). One of these compartments contains the plasma membrane ATPase Pma1, whereas the other contains a number of permeases, including Can1 and Fur4 (Bagnat et al. 2000; Dupré and Haguenauer-Tsapis 2003; Hearn et al. 2003; Malinska et al. 2003, 2004). By contrast, the amino acid permease Gap1 is found in both compartments of the yeast plasma membrane (Lauwers and Andre 2006; Lauwers et al. 2007).
It has been proposed that these subdomains of the plasma membrane can be equated with structures variously termed lipid rafts or detergent-resistant membranes (DRMs), which have been proposed in a variety of eukaryotic systems (Bagnat et al. 2000; Simons and Toomre 2000). This proposal remains quite controversial (Kenworthy 2008; Munro 2003). In particular, it is unclear whether there is any relationship between plasma membrane microdomains observed in vivo by fluorescence microscopy and DRM fractions obtained in vitro (Lichtenberg et al. 2005). By definition, a DRM fraction can only be observed after extraction of the membrane with detergent and thus may not correspond to structures in the native membrane (Lichtenberg et al. 2005). Based on studies of model membranes, it has been argued that DRM fractions might be artifactually created by detergent treatment (Heerklotz 2002; Heerklotz et al. 2003; Keller et al. 2005; Lichtenberg et al. 2005), particularly in the presence of sterols or sphingolipids. Attempts to perturb detergent-resistant subdomains in vivo via depletion of sterols are similarly subject to alternative interpretations (Kenworthy 2008).
Despite the questionable correspondence between in vivo observations of lateral segregation or microdomain formation and in vitro examination of DRM components, it is clear that the formation, turnover, and biological significance of membrane microdomains are important topics of current research. Establishment of these unique regions, whether highly specialized or rudimentary, requires lateral segregation of both lipid and protein components within the plane of the plasma membrane (Bogdanov et al. 2008). S. cerevisiae has proven to be a valuable system for studying this phenomenon, due largely to thorough elucidation of lipid biosynthetic pathways in this organism (Carman and Henry 2007), particularly for sphingolipids (Daum et al. 1998; Dickson 2008).
Advances have been made in our understanding of the underlying biophysical basis of the lateral segregation found in plasma membranes. Sphingolipids and sterols seem important both for “raft” assembly in vivo and for formation of the DRM fraction in vitro (Bagnat et al. 2000; Dupré and Haguenauer-Tsapis 2003; Hearn et al. 2003). There is also evidence that certain glycerophospholipids are important in establishment of discrete plasma membrane subdomains, including phosphatidylethanolamine (PE) (Opekarova et al. 2005) and a specialized phosphatidyinositol with a very long fatty acid substituent that is a component of GPI-anchored proteins (Fujita et al. 2006). It has been reported that localization of certain permeases to one of the plasma membrane subdomains in S. cerevisiae depends on maintenance of the transmembrane electrochemical potential, whereas other segregated plasma membrane proteins, including the eisosome component Sur7 (Walther et al. 2006) and the plasma membrane ATPase Pma1, are not affected by membrane depolarization (Grossmann et al. 2007). Pma1 is able to form high molecular weight oligomers in the presence of ceramide (Bagnat et al. 2001; Lee et al. 2002), suggesting that, at least in some cases, protein-protein interactions can contribute to lateral segregation.
In eukaryotic cells, aminophospholipids, like PE, are normally markedly enriched in the inner leaflet of the plasma membrane. Some of the enzymes that contribute to the maintenance of such plasma membrane asymmetry have been elucidated in yeast. For PE and phosphatidylcholine (PC), there are both inward-directed and outward-directed lipid translocases (Kean et al. 1993; Pomorski et al. 2003). Thus, establishment of the observed asymmetric distribution demands balance between these opposing catalytic activities. The inward-directed PE translocases in the plasma membrane of S. cerevisiae are proposed to be Dnf1 and Dnf2, two members of a novel subgroup (class 4) of membrane transporters of the P-type ATPase family (Hua et al. 2002; Pomorski et al. 2003), although there is some ambiguity as to whether these proteins are responsible for translocation of phosphatidylserine (Chen et al. 2006; Stevens et al. 2008). For their proper localization and function, Dnf1 and Dnf2 associate with a noncatalytic subunit, Lem3/Ros3 (Kato et al. 2002; Saito et al. 2004). Dnf1 and Dnf2 also catalyze inward translocation of other compounds, including PC, lysophospholipids, and ether-linked lipid antifungals (Elvington et al. 2005; Hanson et al. 2003; Riekhof et al. 2007). Dnf1, Dnf2, and Lem3 all have paralogues that reside in intracellular membranes and that have been implicated in endocytosis, intracellular vesicular trafficking, and polarized growth (Chen et al. 2006; Furuta et al. 2007; Hua et al. 2002; Pomorski et al. 2003; Saito et al. 2007). Interestingly, in some instances, these biological processes do not seem to require the translocase function of these proteins per se (Natarajan et al. 2004; Noji et al. 2006).
The outward-directed PE translocases in the plasma membrane of S. cerevisiae are thought to be members of a different superfamily of membrane transporters, the ATP-binding-cassette (ABC) transporters (Pomorski et al. 2003). The S. cerevisiae plasma membrane likely contains as many as 10 ABC transporters; of these, Pdr5 and Yor1 have been specifically implicated as outward-directed PE translocases (Pomorski et al. 2003). Both had been identified previously via their ability to mediate resistance to a broad spectrum of antimicrobial drugs (Balzi et al. 1994; Bissinger and Kuchler 1994; Cui et al. 1996; Decottignies et al. 1998; Katzmann et al. 1995; Kralli et al. 1995). The S. cerevisiae genome encodes two other ABC transporters that are closely related to Pdr5, which have been designated Pdr10 (67% identity to Pdr5) and Pdr15 (75% identity to Pdr5). Our work has shown that Pdr15 has considerable functional overlap with Pdr5 and can also mediate drug export (Schüller et al. 2007; Wolfger et al. 2004). Unexpectedly, a pdr5∆ pdr15∆ double mutant strain exhibited increased outer-leaflet PE (Schüller et al. 2007), indicating that loss of the activity of two putative outward-directed translocases was somehow offset, presumably by decreased activity of the inward-directed translocases. Our findings suggest, therefore, that maintenance of plasma membrane PE asymmetry is under some type of feedback control, reminiscent of the interplay observed between glycerophospholipid asymmetry and sphingolipid transport (Kihara and Igarashi 2004).
Pdr10 has been presumed to play a role in pleiotropic drug resistance only by virtue of its resemblance to Pdr5 and Pdr15 (Abolmaali et al. 2008; Decottignies et al. 1998) and to have a function that overlaps with those of Pdr5 and Pdr15 only because expression of the PDR10 gene involves the same transcription factors (Pdr1 and Pdr3) that control PDR5 and PDR15 expression (Wolfger et al. 1997). By contrast, for example, the ABC transporter Pdr12 (39% identity to Pdr5) is under the control of a dedicated transcription factor (War1) and has a well-defined and unique function, namely, export of weak organic acids such as sorbate (Kren et al. 2003; Piper et al. 1998; Schüller et al. 2004). In any event, the properties, subcellular localization, and function of Pdr10 had not heretofore been directly examined experimentally.
Here we describe our analysis of the physiological role of Pdr10. First, we constructed a functional Pdr10-GFP chimera to visualize its distribution in vivo. Second, we also examined the behavior of Pdr10 upon biochemical fractionation of cell extracts. Third, we constructed a null mutation and compared the phenotypic characteristics of pdr10∆ cells to those of otherwise isogenic PDR10+ cells. Fourth, we analyzed various properties of the cell envelope (plasma membrane and cell wall) and performed additional genetic analysis to pinpoint the underlying molecular causes of the defects and phenotypes observed. Our findings demonstrate that Pdr10 contributes in a significant way to maintaining proper lateral segregation of Pdr12 by helping to exclude it from the DRM fraction. To our knowledge, the function of Pdr10 in establishment of the normal plasma membrane microenvironment for another protein is a novel role for an ABC transporter in cell biology.
Materials and Methods
Strains, Plasmids, and Growth Conditions
S. cerevisiae strains used in this study
MATaura3-1 leu2-3,112 his3-11,15 trp1-1 ade2-1 can1-100
Sherman et al. (1986)
MATα ura3-1 leu2-3,112 his3-11,15 trp1-1 ade2-1 can1-100
Sherman et al. (1986)
MATa/MATα ura3-1/ura3-1 leu2-3,112/leu2-3, 112 his3-11,15/his3-11,15 trp1-1/trp1-1 ADE2/ade2-1 LYS2/lys2::hisG can1-100/can1-100
Shulewitz et al. (1999)
Kuchler et al. (1989)
Schüller et al. (2007)
Schüller et al. (2007)
W303-1A pdr5∆::URA3MX4 pdr10∆::HIS3MX6
W303-1A pdr10∆::HIS3MX6 pdr15∆::TRP1
W303-1A dnf1∆::URA3MX4 dnf2∆::kanMX4
W303-1A pdr10∆::HIS3MX6 lem3∆::URA3MX4
W303-1A pdr10∆::HIS3MX6 dnf1∆::URA3MX4 dnf2∆::kanMX4
W303-1A ADE2 lys2::hisG
W303-1B ADE2 lys2::hisG
W303-1A pdr10∆::HIS3MX6 LEM3-mh::kanMX6
NRY802 pdr10∆::URA3MX4 PDR5-GFP::HIS3MX6
NRY802 pdr10∆::URA3MX4 LEM3-GFP::HIS3MX6
NRY802 pdr10∆::HIS3MX6 PDR12-GFP::kanMX6
NRY802 pdr10∆::URA3MX4 DNF2-GFP::HIS3MX6
NRY802 pdr10∆::URA3MX4 DNF1-GFP::HIS3MX6 LYS2
MATaade2-101ochis3-∆200 leu2-∆1 lys2-801a trp1-∆63 ura3-52
Sikorski and Hieter (1989)
Plasmids pFAkanMX4, pFA6aHIS3MX6, pAG60, pFA6aGFP(S65T)kanMX6, and pFA6aGFP(S65T)HIS3MX6 have been described previously (Goldstein et al. 1999; Wach et al. 1994, 1997). To construct pNR99A, which permits PCR-based C-terminal tagging of yeast proteins with a tandem c-Myc-His6 tag followed by a STOP codon and transcriptional terminator (the “mh” tag), plasmid yGALSte6(∆A-box)-myc-his (gift from E. E. Swartzman, Thorner Lab), which encodes a functional Ste6 with a C-terminal mh tag, was amplified by PCR with appropriate primers to permit cloning of a fragment containing 165 bp from the 3′ end of STE6 followed by the mh tag coding sequence and a STOP codon, which was inserted into pFA6aGFP(S65T)kanMX6 using unique BamHI and AscI sites. The presence of a single, correctly cloned insert was verified by restriction digestion and nucleotide sequencing. This plasmid permits integration of the mh tag with or without linker sequence derived from the C-terminus of Ste6. In the present work, the Lem3-mh fusion lacks any linker, whereas the Pdr10-mh chimera incorporates a pentapeptide (–SNQSS–) sequence derived from the STE6 gene that is situated between the C-terminal residue of Pdr10 and the mh tag. Both Pdr10-mh and Lem3-mh were functional, as judged by a phenotypic test (normal Calcofluor sensitivity compared to untagged strains and pdr10∆ or lem3∆ strains).
Rabbit polyclonal antisera used in this study were anti-Ste2 (David et al. 1997), anti-Chs2 and anti-Chs3 (R. Schekman, University of California, Berkeley, USA), anti-Drs2 (T. Graham, Vanderbilt University, Nashville, TN, USA), anti-Pdr12 (Piper et al. 1998), anti-Pdr5 (Egner et al. 1995), anti-Snq2 (Mahé et al. 1996b), anti-Ycf1 and anti-Yor1 (S. Moye-Rowley, University of Iowa, Iowa City, USA), ant-Gas1 (A. Conzelmann, University of Fribourg, Fribourg, Switzerland), and anti-Hxt1 (E. Boles, Goethe-Universität Frankfurt, Frankfurt-am-Main, Germany). Mouse monoclonal antibodies used in this study were anti-Fks1 mAb (Y. Ohya, University of Tokyo, Toyko, Japan), anti-Pma1 mAb (F. Portillo, Universidad Autonoma de Madrid, Madrid, Spain), anti-Vph1 mAb (T. Stevens, University of Oregon, Eugene, OR, USA), anti-ALP/Pho8 mAb (Molecular Probes), and mAb 9E10 directed against the c-Myc epitope used in this work (Evan et al. 1985).
Two techniques were used to quantitate immunoblots. For samples analyzed by chemiluminescence, exposures were taken using Kodak type MS film, which was scanned and quantitated using densitometry and open-source image analysis software (NIH Image, NIH, Bethesda, MD, USA). Alternately, blots were probed with infrared dye-labeled secondary antibodies and scanned using an infrared scanner (Odyssey; Li-Cor, Inc., Lincoln, NE, USA). Results obtained by these two techniques were crosschecked using samples of equivalent protein content from the total membrane fraction of wild-type cells and pdr10∆ mutants, which were blotted with anti-Pdr12 antiserum. When determined by either method, the value obtained for the ratio of Pdr12 content in pdr10∆ mutants to that in wild-type cells was essentially the same. Thus, measurements of relative protein content made using either procedure were comparable.
All spot-dilution assays were performed using overnight cultures grown in YPD. Aliquots from such cultures were adjusted to an A600 nm = 1 with sterile H2O. Three 10-fold serial dilutions were made in sterile H2O, and aliquots (5 μl) of all four cell suspensions were spotted onto YPD plates lacking or containing the additive to be tested. After the liquid in each spot was absorbed, the plates were inverted, incubated at 30°C for 2 days, and then photographed. The following additives were tested at the indicated concentrations: Calcofluor White (Fluorescent Brightener 28; Sigma Chemical Co., St. Louis, MO, USA), 0 to 8 μg/ml; Congo Red, 100 μg/ml; ketoconazole, 2 μM; rhodamine 6G, 20 μM; and cycloheximide, 100 ng/ml. For assay of sorbate sensitivity, the cell suspensions were spotted onto YPD plates that were buffered using 50 mM sodium succinate (pH 3.5) and contained various concentrations of potassium sorbate (0 to 4 mM), and the resulting plates were photographed after incubation at 30°C for 3 days.
Detergent sensitivity was assessed using a diffusion (“halo”) assay. Cells of the strain to be tested (0.1 A600 nm unit) from overnight cultures grown in YPD were diluted into 3 ml of YPD top agar and spread on the surface of a YPD plate. Sterile filter disks (1-cm diameter; Whatman 3MM) were placed on the seeded cell lawn and samples (10 μl) of solutions (10%, w/v) of the detergents to be tested were spotted onto the disks. Plates were then incubated right-side-up at 30°C for 2 days.
A functional GFP-tagged chitin synthase (Chs3-GFP) was expressed from plasmid pCRP12 with simultaneous overexpression of Chs7 from pRS315-CHS7, as previously described (Valdivia et al. 2002). Exponentially growing cultures of strains W303-1A and NRY251 carrying these plasmids were treated with α-factor mating pheromone to induce cell cycle arrest and shmoo formation, and then pheromone was washed out with fresh, prewarmed medium to permit synchronized re-entry into the cell cycle. The cellular distribution of Chs3-GFP was visualized under a fluorescence microscope (Olympus BH-2) equipped with a 100× objective and appropriate bandpass filters (Chroma Technology). Digital images were captured using an Olympus charge-coupled device (CCD) camera and Magnafire-SP software (Optronics). To visualize the distribution of fluorescently labeled chitin in the cell wall, cells were stained with Calcofluor White following a published procedure (Pringle 1991). Briefly, exponentially growing cells were collected by centrifugation, washed twice in H2O, incubated in aqueous Calcofluor White (0.1 mg/ml) for 5 min at room temperature, washed three times with H2O, recollected by centrifugation, resuspended at an appropriate dilution in H2O, and viewed using appropriate filters under the same fluorescence microscope described above. The intensity of Calcofluor staining in photomicrographs was quantitated using NIH Image; pilot experiments examining chitin levels in wild-type, pdr10∆, and cho1∆ cells by fluorescence-activated cell sorting were in agreement with quantitation of photomicrographs.
To examine uptake of the water-soluble fluorescent dye Lucifer Yellow as a measure of fluid phase endocytosis, cells (0.5 A600 nm unit) from exponentially growing cultures of W303-1A and NRY243 grown in YPD were collected by centrifugation and resuspended in fresh YPD medium (90 μl). An aliquot (10 μl) of an aqueous stock solution (40 mg/ml) of Lucifer Yellow CH (Molecular Probes, Inc., Eugene, OR, USA) was added and the cell suspension was incubated with gentle agitation at 30°C for 1 h. After this incubation, the cells were recollected, washed with 1 ml H2O, and viewed using appropriate filters under the Olympus fluorescence microscope. Similarly, for uptake of the lipophilic fluorescent dye FM 4-64, this compound (16 μM final concentration) was added from a concentrated stock (16 mM) in DMSO directly to cultures of exponentially growing W303-1A and NRY243 cells and the suspension was incubated with gentle agitation at 30°C for 40 min, then washed with H2O and viewed using appropriate filters under the Olympus fluorescence microscope. To examine filipin staining and sterol polarization, log phase cells were resuspended in SCD medium buffered to pH 3.5 with 25 mM sodium succinate. Aliquots were treated with α-factor as described above or with solvent control. After pheromone treatment, cells were washed twice in SCD buffered with 1× lysis buffer (defined below) and stained with 10 μg/ml filipin for 5 min in the dark at room temperature. Cells were then washed with SCD buffered with lysis buffer and examined on the Olympus fluorescence microscope. To visualize actin, 134 μl of 37% formaldehyde (commercial) was added to1-ml log phase cultures of appropriate strains and the mixture incubated for 30 min at room temperature with gentle agitation. Cells were washed twice in phosphate-buffered saline supplemented with 1 mg/ml bovine serum albumin and then resuspended in 50 μl of the same buffer. Ten microliters of rhodamine-phalloidin from a stock of 300 units dissolved in methanol was added, and cells were incubated for 30 min in the dark at 4°C with gentle agitation. Cells were then washed three times with the same buffer, resuspended in 10 μl buffer supplemented with 20 μl Citifluor, and visualized by fluroescence microscopy.
The Pdr10-GFP, Pdr12-GFP, Pdr5-GFP, Lem3-GFP, Dnf1-GFP, and Dnf2-GFP chimeras were constructed by generating in-frame fusions of the coding sequence for GFP to the coding sequence of each protein at its resident locus in its respective chromosome using a PCR-based insertion method and plasmids containing the cognate genes as the template DNA (Wach et al. 1997). The primers used for each construction were designed to retain the native C-terminal sequence of each protein, followed by a nonapeptide linker (–AAGRRIPGL–) and then by GFP. Proper integration was verified by PCR. Retention of biological function of each chimera was assessed by its ability to complement the phenotype of the corresponding mutant: for Pdr5-GFP, restoration of resistance to rhodamine 6G; for Pdr12-GFP, restoration of resistance to sorbate; and for Pdr10-GFP, Lem3-GFP, Dnf1-GFP, and Dnf2-GFP, restoration of resistance to Calcofluor White.
To visualize the subcellular localization of each chimera (except Pdr10-GFP), cells from exponentially growing cultures were washed once in 10 mM Tris–HCl (pH 7.0) and then viewed using a Nikon TE300 fluorescence microscope equipped with a 100× objective and appropriate bandpass filters (Chroma Technology). Digital images were captured with a bottom-ported Orca 100 CCD camera (Hamamatsu, Bridgewater, NJ, USA) and Phase 3 imaging software (Northern Exposure, Glen Mills, PA, USA). To examine Pdr10-GFP, we determined empirically that the most reliable results were obtained for stationary phase cells (i.e., cultures grown in YPD for~2 days), which were then, as for the others, washed once in 10 mM Tris–HCl (pH 7.0) prior to viewing under the fluorescence microscope. For time-lapse photomicroscopy, cells were immobilized by spotting onto polylysine-coated coverslips to minimize their movement and short exposure times (at high gain) were taken to capture as many frames as possible to minimize the loss of signal intensity due to photobleaching. When cells were treated with the actin filament-depolymerizing drug latrunculin A, an aliquot (2 μl) containing this compound was added from a concentrated stock (10 mM) in DMSO to a sample (150 μl) of cells from a culture of strain NRY912 that had grown in YPD for 2 days. After incubation with end-over-end mixing for 10 min at 30°C, the cells were collected by centrifugation, washed in 10 mM Tris–HCl (pH 7.0), and examined using the Nikon fluorescence microscope.
Uptake of fluorescent C6-7-nitrobenz-2-oxa-1,3-diazol-4-yl-PE (NBD-PE) and C6-NBD-ceramide (NBD-Cer) (Avanti Polar Lipids) was monitored in exponentially growing cells from cultures grown in synthetic complete (SC) medium containing glucose (dextrose) as the carbon source (SCD), as follows. Cells were collected by centrifugation and resuspended 10-fold concentrated in 300 μl of prewarmed SCD, and then latrunculin A (4 μl) was added from a concentrated stock (10 mM) stock in DMSO. After incubation by end-over-end mixing for 5–10 min at 30°C, a sample (2 μl) of either NBD-PE or NBD-Cer, each from a concentrated stock (10 mM) in DMSO, was added while the suspension was subjected to vigorous vortex mixing (Grant et al. 2001; Pomorski et al. 2003). After a further incubation with end-over-end mixing for 30 min at 30°C, the cells were recollected and washed three times with ice-cold SC medium containing 2% sorbitol and 20 mM NaN3 (abbreviated SCNaN3) prior to viewing under the Nikon fluorescence microscope adjusted for optimal viewing of fluorescence emission in the same range as for fluorescein.
Preparation of Total Yeast Membranes
For membrane preparations, cells were inoculated into fresh, prewarmed YPD medium and grown at 30°C for one to three doublings. The cells were then chilled on ice at the same time that NaN3 and NaF were added (each at a 20 mM final concentration). After 10 min on ice, cells were harvested by centrifugation, washed with 10 ml ice-cold aqueous 20 mM NaN3–20 mM NaF, and resuspended in aliquots (0.12–0.15 g wet weight) that were flash-frozen in liquid N2 and stored at –80°C until used. For lysis, frozen cell pellets were resuspended in 650 μl ice-cold lysis buffer (50 mM Tris–HCl [pH 7.4], 150 mM NaCl, 5 mM EDTA) containing a protease inhibitor mix (Complete; Roche), 1 mM PMSF, 200 μM pepstatin A, and 5 μM MG-132 and then ruptured by five 1-min pulses of vigorous vortex mixing with glass or silica/zirconium beads (Biospec, Bartlesville, OK, USA) separated by 1 min of cooling on ice between each pulse. The resulting lysate was removed and then combined with the solutions generated by extracting the beads three times with 1-ml portions of ice-cold lysis buffer. The combined crude extract was subjected to centrifugation at 500 g for 5 min at 4°C to remove any broken beads, unlysed cells, cell walls, and other large debris. The clarified extract was subjected to centrifugation at 43,000 rpm for 1 h at 4°C in the TLA 100.3 rotor of a TL100 tabletop ultracentrifuge (Beckman). The resulting total membrane pellets were resuspended in 100 mM Tris–HCl (pH 7) containing 5% SDS, 40 μM pepstatin A, and 1 mM PMSF and warmed briefly to 37°C. The concentration of protein in these membrane preparations was determined using the bicinchoninic acid (BCA) method (Pierce Chemical Co.). Samples (typically 50 μg) were adjusted to 1× SDS-PAGE (sample buffer final concentration) by addition of an appropriate volume of 6×-concentrated SDS-PAGE sample buffer (Laemmli 1970) and then divided into aliquots, which were frozen in liquid N2 and stored at –80°C before use. For analysis, an aliquot was thawed for 5 min at 37°C and then subjected to SDS-PAGE and immunoblotting.
Density Gradient Analysis of Detergent-Resistant Membranes
Analysis of DRMs was carried out using modifications of the Optiprep flotation procedure first developed by Simons and co-workers (Bagnat et al. 2000). Total membrane pellets, prepared as described in the preceding section, were resuspended in 230 μl ice-cold buffer lysis buffer containing 1% Triton X-100 in a 4°C room and kept on ice for 1 h. A sample (170 μl) of this material was then added to LOAD solution (2500 μl Optiprep, 150 μl 10× lysis buffer, 150 μl 10% Triton X-100, containing PMSF, pepstatin A, and MG-132 at the same concentration as in lysis buffer), and the resulting mixture was transferred to ultracentrifuge tubes. The material in the tubes was then overlaid with 45% Optiprep (3 ml), 40% Optiprep (3 ml), 30% Optiprep (2.5 ml), and 0% Optiprep (200 μl); each of these solutions was in 0.5× lysis buffer, 0.5% Triton X-100, plus PMSF, pepstatin A, and MG-132. The resulting step gradient was subjected to centrifugation at 4°C for 18 h at 39,000 rpm in an SW41 rotor (Beckman). After centrifugation, fractions (600 μl) were collected, starting from the top of the gradient, and the material in each was precipitated by the addition of 1.2 ml 15% trichloroacetic acid (TCA) followed by immediate vortex mixing and incubation on ice for 45 min (Dupré and Haguenauer-Tsapis 2003). The resulting precipitates were collected by centrifugation at maximum speed in a microfuge for 15 min, washed twice with ~250 μl ice-cold acetone, and solubilized by resuspension in 100 μl 5% SDS containing 0.1 M Tris base and brief incubation at 37°C. For ease of analysis, adjacent fractions were pooled so as to divide each gradient into nine separate portions. Aliquots of each of the nine portions were flash-frozen in liquid N2, some of which were used for protein determination and the remainder of which were used for analysis by SDS-PAGE and immunoblotting as described in the preceding section.
Lipid Analysis and TNBS Labeling
Total phospholipid composition of yeast strains
% Total phospholipid (as incorporated 32P label)
7.6 ± 0.6
20.2 ± 1.5
35.5 ± 2.1
24.1 ± 0.6
7.3 ± 0.9
7.7 ± 0.5
20.3 ± 1.2
34.9 ± 1.1
23.9 ± 1.1
7.3 ± 1.3
2.5 ± 0.6
7.5 ± 0.5
20.5 ± 0.5
36.3 ± 0.7
22.9 ± 0.5
7.8 ± 0.6
2.0 ± 0.6
pdr10∆ dnf1∆ dnf2∆
7.7 ± 0.5
20.0 ± 0.5
35.9 ± 0.5
23.8 ± 0.5
7.4 ± 0.5
2.3 ± 1.0
7.9 ± 0.1
20.7 ± 0.7
35.1 ± 1.5
23.6 ± 1.2
7.3 ± 0.4
2.7 ± 0.5
7.8 ± 0.2
20.3 ± 1.1
35.0 ± 0.5
24.1 ± 1.9
7.2 ± 0.4
2.7 ± 1.0
9.2 ± 0.5
22.7 ± 0.5
32.5 ± 0.5
19.8 ± 0.8
7.4 ± 0.5
0.9 ± 0.2
7.9 ± 1.0
20.3 ± 2.0
38.7 ± 1.7
18.7 ± 0.7
6.4 ± 1.3
0.3 ± 0.2
7.5 ± 0.5
19.7 ± 0.5
36.0 ± 0.5
24.0 ± 0.5
7.7 ± 0.8
1.6 ± 0.6
8.1 ± 0.5
19.9 ± 0.5
35.0 ± 1.1
24.0 ± 0.5
7.6 ± 0.5
1.5 ± 0.6
8.2 ± 0.5
19.1 ± 0.5
34.2 ± 0.6
23.3 ± 0.5
10.0 ± 0.5
1.1 ± 0.2
8.2 ± 0.5
19.7 ± 0.5
35.6 ± 0.7
22.5 ± 1.2
9.2 ± 0.6
1.1 ± 0.2
11.0 ± 3.0
52.7 ± 6.2
26.9 ± 2.8
5.6 ± 1.2
2.7 ± 0.8
To analyze the composition of mature sphingolipids, cells were grown and labeled as described above, and sphingolipids were extracted using a previously published procedure with minor modifications (Dickson et al. 1997). To samples (1 ml) of cultures incubated on ice was added 53 μl of 100% TCA. After 15 min, the acid-treated cells were washed twice with water (1 ml each), resuspended in 400 μl solvent E (diethyl ether/95% ethanol/water/pyridine, 5:15:15:1), and incubated at 60°C for 60 min with occasional vortex mixing. To separate the phases, the mixtures were subjected to centrifugation at maximum speed in a microfuge for 2 min and the sphingolipids present in the resulting organic phase were separated by ascending TLC with CHCl3:CH3OH:4.2 N NH4OH (9:7:2), which yields an acceptable resolution of the major sphingolipid species (Hama et al. 2000), which were quantified as described in the preceding paragraph.
Nonradioactive TNP-PE was synthesized according to a published procedure (Gordesky and Marinetti 1973) with minor modifications (Schüller et al. 2007). PE (25 mg) was dissolved in 2.5 ml CHCl3 and added to a solution composed of 2.5 ml CH3OH, 1 ml CHCl3, 250 μl 5% aqueous NaHCO3, and 390 μl 5% aqueous TNBS. This resulting reaction mixture was incubated at room temperature for 2 h under argon gas and then mixed with 2 ml CHCl3 and 20 ml aqueous saturated NaCl (brine). After separating the phases, the organic phase was withdrawn, acidified with aqueous 1 N HCl, and re-extracted with 5 ml brine, and the resulting organic phase was dried over Na2SO4 and stored at –20°C under argon.
Pdr10 is a Plasma Membrane Protein
The markedly punctate distribution in the plasma membrane displayed by Pdr10-GFP has been observed for several other integral plasma membrane proteins. In some cases, these puncta are mobile because they represent proteins (for example, the glucan synthase Fks1) associated with sites where actin-mediated endocytosis occurs (“actin patches”) (Utsugi et al. 2002). In other cases, the puncta are immobile because they represent proteins (for example, the arginine permease Can1 and the tetraspanin Sur7) associated with stable, static plasma membrane subdomains such as the MCC compartment (Grossmann et al. 2007; Malinska et al. 2003; Young et al. 2002). Therefore, we examined whether the Pdr10-GFP puncta were stable over time and whether they were affected by inhibition of actin function. We found that treatment of cells with latrunculin A did not disrupt the punctate localization displayed by Pdr10-GFP (Fig. 1b). Some of the Pdr10-GFP puncta in a typical cell were immobile for at least 20 min (Fig. 1c). However, other Pdr10-GFP puncta seemed more mobile, particularly in the bud. There is evidence that the mother and bud compartments have distinct properties. For example, PE exposure is reportedly greater at the bud tip than in the mother cell (Iwamoto et al. 2004), and the collar of septin filaments at the bud neck acts as a diffusion barrier to confine certain plasma membrane proteins uniquely in the bud (Takizawa et al. 2000). However, the presence of mobile patches in the mother cell as well indicates that there is a mobile population of Pdr10 throughout the cell. Pdr10-GFP did not behave like an actin patch-associated protein, but the presence of both mobile and immobile populations argues against its exclusive localization to previously defined subdomains of the yeast plasma membrane.
Can1 and other structurally related permeases not only display a punctate distribution in the plasma membrane, but also are found in the detergent-resistant membrane (DRM) fraction (sometimes equated with lipid rafts) when cells are subjected to biochemical fractionation (Malinska et al. 2003, 2004). To permit facile examination of the behavior of Pdr10 during biochemical fractionation, we tagged the chromosomal allele of PDR10 at its C-terminus with c-Myc and His6 epitopes in tandem (the mh tag) under control of the native PDR10 promoter. We then prepared total membranes from such cells, extracted them with cold detergent (Triton X-100), and fractionated them by density gradient flotation. When subjected to this procedure, the residual lipid associated with the DRM fraction causes it to appear at the top of the gradient, while detergent-soluble or aggregated proteins exhibit higher apparent density and are found at the bottom of the gradient (Bagnat et al. 2000; Malinska et al. 2003, 2004). Previous studies using this assay have identified the plasma membrane H+-ATPase Pma1 and the glycosylphosphatidylinositol (GPI)-anchored protein Gas1 as markers for the DRM fraction (Bagnat et al. 2000; Malinska et al. 2003). Markers for the detergent-soluble fraction of the plasma membrane have proven more controversial: the hexose transporter Hxt1 has been reported to primarily localize to the detergent-soluble fraction by Tanner and colleagues (Malinska et al. 2003, 2004), using anti-Hxt1 antisera for detection, but Hxt1-GFP has instead been found in the detergent-resistant fraction by Lauwers and André (2006) using anti-GFP antisera for detection. These studies also use different strain backgrounds, which may explain the discrepancy. We find that Hxt1 is primarily (~90%) in the detergent-soluble fraction (Fig. 1d), in good agreement with the results reported by Tanner and colleagues (Malinska et al. 2003, 2004).
To detect the inabundant Pdr10 in flotation gradients, we found that it was necessary to load substantially more material than has typically been reported. Unfortunately, the greater amounts loaded resulted in loss of resolution in the small Optiprep (iodixanol) density gradients pioneered by Simons and colleagues (Bagnat et al. 2000). We therefore developed a modified procedure using a larger gradient with a consequently longer centrifugation (described in detail under Materials and Methods). The DRM marker Gas1 exhibits essentially identical behavior in both gradients, as does the chitin synthase Chs3 (data not shown). However, while more than 90% of Pma1 was found in the DRM fraction in the smaller gradients, a substantial amount of Pma1 was instead associated with the soluble fraction in these expanded gradients, with only 25%–50% of Pma1 remaining in the DRM fraction (Fig. 1d). We have not attempted to ascertain whether this reflects slow solubilization or aggregation of Pma1 in the longer ultracentrifugation procedure we employ, because Gas1 and Hxt1 serve as robust markers for the detergent-resistant and detergent-soluble fractions in this gradient.
Using this procedure, we were able to detect Pdr10-mh with antisera directed against the c-myc epitope (Fig. 1d). Tellingly, Pdr10-mh displayed a distribution quite similar to that of Pma1, with substantial Pdr10-mh residing in the DRM fraction. The presence of both detergent-soluble and detergent-resistant Pdr10 is reminiscent of the presence of immobile and mobile populations of Pdr10-GFP as seen by fluorescence microscopy, but the nature of the DRM experiment precludes assignment of the in vitro populations to in vivo populations (Lichtenberg et al. 2005).
Absence of Pdr10 Causes Distinct Phenotypes
We also challenged pdr10∆ cells with a number of detergents. As judged by a halo assay, a pdr10∆ mutant strain in the W303 genetic background reproducibly exhibited modestly elevated resistance to detergents containing C12 alkyl chains, including dodecyl maltoside and dodecyl phosphocholine (Fig. 2b, upper two rows). However, a pdr10∆ mutant in the YPH499 genetic background did not exhibit this effect (Fig. 2b, lowest row). We therefore used pdr10∆ cells derived from the W303 lineage for further study.
In addition to changes in membrane composition, changes in cell wall structure can affect the detergent sensitivity displayed by S. cerevisiae cells; indeed, cell wall alterations can affect the sensitivity of yeast cells even to a strong anionic detergent like SDS (Bickle et al. 1998; Machi et al. 2004; Santos and Snyder 2000). The yeast cell wall is composed of three types of polymers: mannoproteins, glucan, and chitin (Klis et al. 2006; Lesage and Bussey 2006). We found (Fig. 2c) that pdr10∆ mutants were abnormally sensitive to two agents, Calcofluor White and Congo Red, that inhibit the growth of cells that have abnormally high amounts of chitin (Roncero et al. 1988). In contrast, otherwise isogenic wild-type cells (Fig. 2c), or cells expressing either Pdr10-GFP or Pdr10-mh from the chromosomal PDR10 locus (data not shown), did not display such hypersensitivity (and, thus, both Pdr10-GFP and Pdr10-mh are functional). If the sensitivity of the pdr10∆ cells is due to abnormal chitin, then we expected that sensitivity to Calcofluor White might be ameliorated by compromising chitin production. The chitin synthase responsible for producing the majority of the chitin generated in a vegetatively growing cell is Chs3, but cells can survive in its absence due to the function of the homologous chitin synthase Chs2 (Cabib 2004). Chs3 is normally delivered to the plasma membrane via the action of dedicated escort factors, including Chs6 (Valdivia et al. 2002; Valdivia and Schekman 2003). Consistent with the conclusion that the Calcofluor White sensitivity of pdr10∆ cells arises from elevated cell wall chitin, we found that pdr10∆ chs3∆ and pdr10∆ chs6∆ double mutants were not hypersensitive to Calcofluor White (Fig. 2c). Likewise, in the W303 strain background, a pdr10∆chs3∆ double mutant was not detergent resistant (Fig. 2b, upper rows). These findings indicate that both the Calcofluor White sensitivity and the detergent resistance displayed by pdr10∆ cells are largely due to enhanced chitin production by the integral membrane protein Chs3 in the absence of this ABC transporter.
Although Pdr10 shares its highest degree of homology with Pdr5 and Pdr15, it is also closely related to other ABC transporters in the S. cerevisiae genome, including Pdr12 (36% identity to Pdr10). Pdr12 is located in the plasma membrane and is responsible for the export of weak organic acids (Piper et al. 1998). Consequently, pdr12∆ mutants are hypersensitive to the growth inhibitory action of weak organic acids used as food preservatives (e.g., sorbate or benzoate) (Piper et al. 1998). We found that pdr10∆ cells exhibited weak resistance to sorbate (Fig. 2d); however, this behavior required the function of Pdr12, because the modest sorbate resistance observed was eliminated in a pdr10∆ pdr12∆ double mutant (Fig. 2d). Thus, here, too, the absence of Pdr10 somehow promoted the function of a sister integral membrane protein, Pdr12.
Pdr10 Is Required for Normal Endocytic Trafficking of Chs3
In vegetatively growing cells, Chs3 cycles between the trans-Golgi network and endosomal compartments during most of the cell cycle (Valdivia et al. 2002) but displays significant accumulation at the plasma membrane primarily at the small-bud stage (Chuang and Schekman 1996). The intracellular pool of Chs3 can also be redistributed to the plasma membrane after treatment of cells with mating pheromone, an effect we exploited to trap Chs3 at the plasma membrane in wild-type and pdr10∆ cells. Upon washout of pheromone, cells re-entered the cell cycle and Chs3 moved to the bud neck, as expected (Fig. 4b). At later time points, Chs3 was efficiently internalized in wild-type cells, whereas in pdr10∆ cells a small amount of Chs3-GFP persisted in discrete patches at the plasma membrane, even though the majority was intracellular. Thus, in pdr10∆ cells, endocytosis of Chs3 was somewhat less efficient than in wild-type cells (although this block was by no means complete). The prolonged residence time of Chs3 at the plasma membrane in pdr10∆ cells could well explain the increased amount of chitin produced in these cells.
Lack of Dnf1 and Dnf2 Results in Defective Endocytosis of Chs3
It has been proposed that proper maintenance of membrane asymmetry is important for endocytosis at low temperatures and that Pdr5 and another ABC transporter, Yor1, are candidate outward-directed PE translocases (Pomorski et al. 2003). Moreover, pdr5∆ pdr15∆ cells exhibit increased accessibility of PE to the small-molecule reagent TNBS (Schüller et al. 2007), which is able to pass through the cell wall. TNBS labeling has been used to assess exposure of PE and PS on the outer leaflet of the plasma membrane (Gordesky and Marinetti 1973; Pomorski et al. 2003; Siegmund et al. 1998), although subcellular fractionation data raise the possibility that there is considerable leakage of the reagent into the secretory pathway (Pomorski et al. 2003). We analyzed the total phospholipid composition (Table 2) and accessibility of aminophospholipids to TNBS (Fig. 6b) in wild-type and pdr10∆ cells grown on 32P-phosphate. We found that a pdr10∆ mutation had no discernible effect on overall phospholipid composition (Table 2), but caused a significant increase in the fraction of the PE that was accessible to TNBS labeling (Fig. 6b). Thus, pdr10∆ plasma membranes either are significantly more permeable to TNBS or exhibit a defect in maintenance of plasma membrane asymmetry.
The ABC transporters with established transport activities that are most closely related to Pdr10, including Pdr5 and Snq2, mediate outward-directed lipid transport and/or export of xenobiotics from the cell (Balzi et al. 1994; Bissinger and Kuchler 1994; Decottignies et al. 1995; Egner et al. 1998; Mahé et al. 1996a; Pomorski et al. 2003; Schüller et al. 2007). If Pdr10 had a similar function, then loss of Pdr10 should reduce, not increase, the fraction of PE that is surface-exposed. In contrast, if the absence of Pdr10 somehow indirectly crippled the function of the inward-directed PE translocases, then the biochemical phenotype observed in pdr10∆ cells would be more readily explained. In support of this, we observed that the level of PE accessible to TNBS labeling in pdr10∆ cells was comparable to that in cells in which the genes for both putative inward-directed plasma membrane PE translocases, DNF1 and DNF2, were deleted (Fig. 6b); similar results were also obtained for cells lacking the LEM3 gene, which encodes an integral membrane protein that is required for exit of Dnf1 and Dnf2 from the endoplasmic reticulum (Saito et al. 2004). Furthermore, no additive increase in exposure of PE was observed in pdr10∆ lem3∆ or pdr10∆ dnf1∆ dnf2∆ strains (Fig. 6b). The overall phospholipid compositions of all of these strains were quite comparable (Table 2). The absence of any additive or synergistic effect of the dnf1∆dnf2∆ (or lem3∆) mutations in a pdr10∆ mutant demonstrated that pdr10∆ phenocopied dnf1∆ dnf2∆. That is, regardless of whether TNBS labeling is measuring PE exposure on the outer leaflet or permeability, the loss of Pdr10 function mimicked the loss of Dnf1 and Dnf2 function in these cells.
Complex Interplay Among Pdr10, Lem3, and Pdr12
Cells lacking Pdr10 also exhibit a modest increase in their resistance to sorbate that requires the sorbate transporter, Pdr12 (Fig. 2d). We were interested in whether this was also a phenotype of dnf1∆ dnf2∆ cells or lem3∆ cells. Interestingly, we found not only that lem3∆ cells were not sorbate resistant, but that a lem3∆ mutation eliminated the modest sorbate resistance conferred by the pdr10∆ mutation; no such epistasis was found with the dnf1∆ dnf2∆ double mutant (Fig. 9c and data not shown). Thus, the apparent increase in Pdr12 action found in cells lacking Pdr10 does not mimic a defect found in dnf1∆ dnf2∆ or lem3∆ cells but, rather, requires the presence of functional Lem3.
Cells lacking Pdr10 exhibit increased sensitivity to the chitin-directed agent, Calcofluor White (Fig. 2c), and that is not further enhanced by the absence of Dnf1 and Dnf2 (Fig. 9a). Strikingly, as observed for the sorbate resistance of pdr10∆ cells, we found that loss of Pdr12 eliminated the Calcofluor White sensitivity of pdr10∆ (Fig. 9d). We also found that the introduction of a pdr12∆ mutation into pdr10∆ cells substantially reduced the accessibility of PE to TNBS labeling (Table 2). Collectively, these findings indicate complex interactions among Pdr10, Pdr12, and Lem3, which produce secondary effects on Chs3 endocytosis.
Pdr10 Action Excludes Pdr12 from the Detergent-Resistant Membrane Fraction
Association of membrane proteins with the detergent-resistant membrane fractiona
33 ± 11
54 ± 10
61 ± 6
28 ± 11
17 ± 9
62 ± 6
19 ± 9
56 ± 10
19 ± 5
75 ± 7
24 ± 4
72 ± 1
47 ± 5
37 ± 1
45 ± 2
36 ± 2
57 ± 11
23 ± 10
69 ± 9
14 ± 8
Interaction Between Pdr10 Function and Sphingolipids
Sphingolipids are known to be important factors for lateral segregation and DRM formation in yeast membranes (Bagnat et al. 2000). We were therefore interested in the interplay between sphingolipid composition and Pdr10 function, because of its role as a negative regulator of Pdr12 association with the DRM fraction. In S. cerevisiae, the headgroups of mature sphingolipids contain inositol phosphate. Inositol phosphate is added to ceramide by Aur1, producing inositolphosphorylceramide (IPC). AUR1 is an essential gene, demonstrating that IPC or lipids derived from IPC are essential (Nagiec et al. 1997). IPC is converted to mannosyl-inositolphosphoryl-ceramide (MIPC) by the redundant action of Sur1 (also known as Csg1) and Csh1 (Beeler et al. 1997; Uemura et al. 2003). MIPC is then converted to mannosyl-di(inositolphosphoryl)-ceramide (M[IP]2C) by the subsequent action of Ipt1 (Dickson et al. 1997).
Composition of mature sphingolipids
35 ± 4
17 ± 2
47 ± 4
41 ± 2
18 ± 2
41 ± 1
61 ± 3
13 ± 1
26 ± 3
46 ± 3
48 ± 2
7 ± 1
We have shown that the ABC transporter Pdr10 is a plasma membrane protein that plays a role in maintaining the proper distribution and function of a subset of other membrane proteins in the yeast S. cerevisiae. In the absence of Pdr10, yeast cells exhibit a modest defect in the endocytosis of the chitin synthase Chs3, resulting in increased cell wall chitin and consequent sensitivity to Calcofluor White. The increased chitin deposition and Calcofluor sensitivity are equivalent to those seen in dnf1∆dnf2∆ cells, which lack the putative inward-directed PE translocases of the yeast plasma membrane (Pomorski et al. 2003). Both pdr10∆ cells and dnf1∆dnf2∆ cells exhibit increased accessibility of PE to TNBS labeling, consistent with a defect in maintenance of plasma membrane asymmetry. However, Dnf1 and Dnf2 are present at normal levels in pdr10∆ cells, and these cells are able to internalize NBD-PE. It thus seems likely that these phenotypes of pdr10∆ cells arise indirectly, perhaps as a consequence of a modest defect in trafficking of plasma membrane proteins.
The absence of Pdr10 function also results in increased resistance to the weak organic acid sorbate. Sorbate is directly exported from yeast cells by the plasma membrane ABC transporter Pdr12, itself a Pdr10 paralogue. We have shown that pdr10∆ cells accumulate higher levels of Pdr12, despite no changes in PDR12 transcription. Additionally, we have shown that the behavior of Pdr12 upon detergent extraction is altered in pdr10∆ membranes: while Pdr12 in wild-type membranes is mainly in the detergent-soluble fraction, it is mainly in the DRM fraction in pdr10∆ membranes. There are complex interactions among PDR10, PDR12, and LEM3, because pdr12∆ is epistatic to pdr10∆ for Calcofluor sensitivity and for TNBS labeling of PE, while lem3∆ is epistatic to pdr10∆ for sorbate resistance. By contrast, the dnf1∆ dnf2∆ double mutant is equivalent to pdr10∆ for Chs3-related phenotypes, with no additive effect in the pdr10∆ dnf1∆ dnf2∆ triple mutant, but does not affect the Pdr12-related phenotype of sorbate resistance. Moreover, pdr5∆ is epistatic to pdr10∆ for both phenotypes. Pdr5, Pdr12, and Lem3 thus seem to be proteins intimately linked to the molecular function of Pdr10. Interestingly, sur1∆ and ipt1∆, which lack genes required for elaboration of sphingolipid headgroups, are epistatic to pdr10∆ for Calcofluor sensitivity, chitin deposition, sorbate resistance, Pdr12 accumulation, and TNBS labeling of PE. The aberrant behavior of pdr10∆ membranes thus requires at least three specific proteins (Pdr5, Pdr12, and Lem3) and also requires mature mannosylphosphosphingolipids.
In the presence of these components, pdr10∆ cells significantly overincorporate Pdr12 into the DRM fraction. Pdr10 thus functions as a negative regulator of Pdr12 incorporation into DRMs. Slight increases in Pdr5 and Lem3 association with the DRM fraction are also seen, while Gas1 and Chs3 distributions are unchanged. Pdr12 has been estimated to be present at approximately 750 molecules per cell (Ghaemmaghami et al. 2003), and approximately one-third of the total is shifted to the DRM fraction in pdr10∆ membranes relative to wild-type membranes (Table 3). Therefore, approximately 250 additional Pdr12 molecules per cell would become incorporated into the DRM fraction in the absence of Pdr10. In the same study, levels of Ste20 were estimated at approximately 260 molecules per cell, and both Dnf1 and Dnf2 exhibited lower but still detectable signals; however, Pdr10 could not be detected at all (Ghaemmaghami et al. 2003). Therefore, Pdr10 levels must be substantially lower than 250 molecules per cell, which means that Pdr10 is substoichiometric with the Pdr12 population that is shifted to the DRM fraction in pdr10∆ membranes. It thus seems unlikely that the overincorporation of Pdr12 in the DRM fraction of pdr10∆ membranes reflects the loss of a stoichiometric association between Pdr10 and Pdr12.
One could also envision the redistribution of Pdr12 into the DRM fraction as arising from an endocytic defect, such that an intracellular pool of Pdr12 would effectively relocalize to the plasma membrane in the absence of Pdr10. The proposed link between endocytosis and exit from the DRM fraction (Lauwers and Andre 2006; Lauwers et al. 2007) would imply that such an intracellular pool would become available for partitioning into the DRM fraction upon such relocalization. In this model, the plasma membrane Pdr12 would be in equilibrium between a microenvironment that gives rise to a DRM fraction and a microenvironment that does not, while intracellular Pdr12 would instead be entirely associated with the detergent-soluble fraction. The absence of Pdr12 endocytosis in pdr10∆ cells would then result in the total Pdr12 population equilibrating between these two microenvironments, rather than just a plasma membrane population doing so. Were this to be the case, the association of approximately 60% of total Pdr12 with the DRM fraction in pdr10∆ cells would reflect the intrinsic affinity of Pdr12 for this fraction, while no intracellular Pdr12 would be found in the DRM fraction. The amount of Pdr12 in the DRM fraction would thus vary between 0% and 60%. The association of approximately 30% of total Pdr12 with the DRM fraction in wild-type cells would thus imply the presence of approximately 50% of total Pdr12 in an intracellular pool. However, we demonstrated that a functional Pdr12-GFP chimera is predominantly localized to the plasma membrane (Fig. 10b). Only ≤ 30% of cells exhibited any detectable intracellular Pdr12-GFP, and the vast majority of such cells also displayed plasma membrane Pdr12-GFP. Moreover, this localization pattern was equivalent in both wild-type and pdr10∆ cells. Therefore, the intracellular population of Pdr12 is likely to be much less than 50% of the total, and there is no evidence that pdr10∆ cells lack this population. These observations are inconsistent with the idea that Pdr12 accumulation in the DRM fraction of pdr10∆ membranes arises from relocalization of Pdr12 to the plasma membrane. We favor, therefore, the idea that the composition of plasma membrane in cells lacking Pdr10 is itself different from that of the wild-type plasma membrane in a way that results in overincorporation of Pdr12 into the DRM fraction.
The gradient flotation assay we used to examine DRM incorporation does not provide information about the actual size of any detergent-resistant structures present in the yeast plasma membrane, but is only sensitive to density during fractionation. It is not rigorously established whether microdomains of native membranes can be stably maintained during the lysis, extraction, and flotation procedure, and it has been argued that detergent treatment can induce DRM formation in homogeneous model membranes (Lichtenberg et al. 2005). However, it is hard to see how such an effect could be specifically enhanced by the absence of such a rare protein as Pdr10. Therefore, the higher occupancy of Pdr12 in the DRM fraction observed in pdr10∆ membranes would represent either its increased partitioning into a specialized membrane subdomain or its increased interaction with boundary lipids that are resistant to removal by detergent. It thus seems that the microenvironment of Pdr12 in the native membrane is altered in the absence of Pdr10.
The most parsimonious model to accommodate all of the observations we have made is that the normal function of Pdr10 is to counteract some of the protein or lipid movements that are needed to establish a specific microdomain in the plasma membrane. Hence, when Pdr10 is absent, there is aberrant or excess formation of this microdomain, and Pdr12 becomes sequestered. Chs3 and Gas1 do not get trapped in this way. Pdr12 sequestered in such a compartment may be more resistant to the processes responsible for its endocytosis, explaining why it accumulates to a higher than normal level in pdr10∆ cells. If Pdr12, Lem3, and Pdr5 are all required for formation of this microdomain, the observed epistasis among these membrane proteins would be explained, because removal of a required microdomain component would disrupt the microdomain and mimic the loss of a negative regulator. Similarly, if the proposed microdomain regulated by Pdr10 requires mature sphingolipids for its formation, then alteration of sphingolipid composition could also ameliorate the effects resulting from the absence of Pdr10, as we observed. Lem3 and Pdr5 are both present at higher levels than Pdr12 (Ghaemmaghami et al. 2003), which provides a possible explanation for why their shift to the DRM fraction (Table 3) is not statistically significant: if Pdr12 is also required for microdomain formation and is less abundant than Lem3 or Pdr5, it will be limiting for microdomain formation, such that the amount of Lem3 and Pdr5 shifted to the DRM fraction is smaller.
The precise nature of the putative sphingolipid-dependent microdomain on which Pdr10 has its effect is unclear. Pdr12, Dnf1, Dnf2, Lem3, and Pdr5 do not accumulate in the ER in pdr10∆ cells, and we have shown that Pdr12 and Pdr5 are active in these cells. The absence of Pdr10 also does not activate the unfolded-protein response pathway (Ron and Walter 2007) (data not shown), providing further evidence that these membrane proteins are not misfolded. Therefore, this microdomain should be viewed as an accretion of functional proteins rather than as an aggregate of unfolded proteins. It has been reported that association of permeases into ~300-nm patches in the membrane requires maintenance of the transmembrane electrochemical potential (Grossmann et al. 2007). Also, it is now generally accepted that segregation of membrane proteins into the DRM fraction in yeast and other eukaryotes depends on the presence of both sterols and sphingolipids (Bagnat et al. 2000; Dupré and Haguenauer-Tsapis 2003; Hearn et al. 2003). Formation of large oligomers of Pma1 is required for the segregation of this protein into the DRM fraction (Bagnat et al. 2001; Lee et al. 2002), and there is ample evidence for certain members of the ABC transporter family that their native state in the plasma membrane is as a large homomultimer (Poruchynsky and Ling 1994; Xu et al. 2007). Thus, both protein-protein and lipid-protein interactions can contribute in a positive sense to microdomain formation, and indeed our epistasis analysis provides indirect evidence for both. Pdr10 acts as a negative regulator of the establishment and/or maintenance of plasma membrane microdomains. This suggests that microdomain assembly and breakdown are quite dynamic processes in vivo.
If microdomains have a critically important role in the physiological functions of the plasma membrane, one might question why a pdr10∆ mutant is viable. However, all of the evidence we have presented indicates that microdomain formation still occurs in cells lacking Pdr10 and that the defect is in the efficiency of microdomain disassembly. Moreover, there are yet other ABC transporters encoded in the S. cerevisiae genome whose function may be semiredundant with that of Pdr10 and, hence, partially mask the true effect of the problems that might be caused if microdomains remained too static. The latter consideration may explain why we observed some strain-background dependence in the phenotypes conferred by a pdr10∆ mutation. In this regard, it is also interesting to note that the noncoding region of the PDR10 gene has been identified as a mutational hotspot in the S. cerevisiae genome (Fay and Benavides 2005). Thus, different lineages of this yeast may have adapted to cope with different demands for Pdr10 function. It is also important to note that the functional significance of the putative Pdr10-regulated microdomain is unclear; it is quite possible that this structure is actually a pathological association of functional Pdr12, Pdr5, and Lem3 in the presence of sphingolipids, and we do not propose to identify it with one of the previously defined compartments of the yeast plasma membrane.
The function of Pdr10 at the molecular level and the means by which its biochemical activity contravenes microdomain formation or maintenance have yet to be elucidated. Possibly, Pdr10 catalyzes outward translocation of an as yet unidentified lipid substrate, by analogy to its paralogous ABC transporter, Pdr5. Alternatively, the fact that Pdr10 itself localizes in discrete puncta in the plasma membrane may be important. For example, the microdomain compartment that appears to accumulate when Pdr10 is absent might be formed from (or beside) the subdomain normally occupied by Pdr10 itself. Hence, Pdr10 could help to occlude or exclude other integral membrane proteins from this same region without invoking any need for it to have transport activity. Another possibility is that, when Pdr10 is embedded in the plasma membrane, it dynamically associates with a specific set of lipids (or even other proteins); in this way, Pdr10 may act as a “detergent” to “solubilize” these components and thus enhance their dispersal within the plane of the membrane.
The authors thank Françoise Roelants, Victor Cid, Matthias Versele, Daniel Ballon, and other members of the Thorner and Kuchler laboratories for helpful discussions. We also wish to thank Gustav Ammerer for many stimulating discussions; William Dowhan, Günther Daum, and Dennis Voelker for protocols for lipid analysis and useful advice; and Todd Graham and the reviewers for constructive criticism. We thank Randy Schekman, Scott Moye-Rowley, Todd Graham, John McCusker, Andreas Conzelmann, Paco Portillo and Pilar Eraso, Eckhard Boles, Tom Stevens, and Yoshi Ohya for antibodies and/or plasmids used in the course of these studies. This work was supported by postdoctoral fellowship PF98-141-01 from the American Cancer Society and NCI postdoctoral traineeship CA09041 to N.C.R., by grants from the Austrian Science Foundation (“Fonds zur Förderung der wissenschaftlichen Forschung” project SFB035-04) and the Marie-Curie training network RTN-Flippases (CT-MC-RTN-2004-005330) to K.K., and by NIH Research Grant GM21841 and facilities provided by the Berkeley campus Cancer Research Laboratory to J.T.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.