QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Vol. 18, Issue 2, 487-500, February 2007

This Article Abstract Full Text (PDF) All Versions of this Article: E06-07-0592v1 18/2/487    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Liu, K. Articles by Graham, T. R. Search for Related Content PubMed PubMed Citation Articles by Liu, K. Articles by Graham, T. R.

Yeast P4-ATPases Drs2p and Dnf1p Are Essential Cargos of the NPFXD/Sla1p Endocytic Pathway

Department of Biological Sciences, Vanderbilt University, Nashville, TN 37235-1634

Submitted July 10, 2006; Revised November 9, 2006; Accepted November 13, 2006
Monitoring Editor: David Drubin

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Drs2p family P-type ATPases (P4-ATPases) are required in multiple vesicle-mediated protein transport steps and are proposed to be phospholipid translocases (flippases). The P4-ATPases Drs2p and Dnf1p cycle between the exocytic and endocytic pathways, and here we define endocytosis signals required by these proteins to maintain a steady-state localization to internal organelles. Internalization of Dnf1p from the plasma membrane uses an NPFXD endocytosis signal and its recognition by Sla1p, part of an endocytic coat/adaptor complex with clathrin, Pan1p, Sla2p/End4p, and End3p. Drs2p has multiple endocytosis signals, including two NPFXDs near the C terminus and PEST-like sequences near the N terminus that may mediate ubiquitin (Ub)-dependent endocytosis. Drs2p localizes to the trans-Golgi network in wild-type cells and accumulates on the plasma membrane when both the Ub- and NPFXD-dependent endocytic mechanisms are inactivated. Surprisingly, the pan1-20 temperature-sensitive mutant is constitutively defective for Ub-dependent endocytosis but is not defective for NPFXD-dependent endocytosis at the permissive growth temperature. To sustain viability of pan1-20, Drs2p must be endocytosed through the NPFXD/Sla1p pathway. Thus, Drs2p is an essential endocytic cargo in cells compromised for Ub-dependent endocytosis. These results demonstrate an essential role for endocytosis in retrieving proteins back to the Golgi, and they define critical cargos of the NPFXD/Sla1p system.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Drs2p is a resident P-type ATPase of the yeast trans-Golgi network (TGN) that is required for vesicle-mediated protein transport from this organelle. Most well-characterized P-type ATPases are cation pumps that control the concentration of ions in both intracellular and extracellular spaces (for example, the Na⁺/K⁺ ATPase, Ca⁺⁺ ATPase, and H⁺/K⁺ ATPase) (Kuhlbrandt, 2004). Drs2p, in contrast, is the founding member of a large P-type ATPase subfamily, called P4-ATPases (Catty et al., 1997), that are proposed to translocate phospholipid rather than ions. This flippase activity is responsible for translocating specific phospholipid molecules from the exoplasmic leaflet to the cytosolic leaflet to establish asymmetry of the membrane bilayer (Graham, 2004; Pomorski et al., 2004; Holthuis and Levine, 2005; Paulusma and Oude Elferink, 2005; Devaux et al., 2006). For Drs2p, ATPase activity and presumably phospholipid translocation are essential, because mutation of the aspartic acid that forms an aspartyl–phosphate intermediate during catalysis (D560N) renders Drs2p nonfunctional in vivo (Chen et al., 1999). In addition, after shifting to the nonpermissive temperature, a drs2 temperature-sensitive (ts) allele causes a rapid loss of exocytic vesicle formation in vivo (Gall et al., 2002) and the loss of an ATP-dependent phosphatidylserine (PS) flippase activity in purified Golgi membranes containing Drs2-ts (Natarajan et al., 2004). Mammalian homologues of Drs2p include the chromaffin granule ATPase II (now called ATP8A1) (Tang et al., 1996), which is likely responsible for a PS translocase activity observed with these exocytic vesicles (Zachowski et al., 1989), and FIC1 (ATP8B1), for which mutations in humans cause an impairment of bile flow through the liver (cholestasis) (Bull et al., 1998; Klomp et al., 2004). In addition, deletions removing the mouse Atp10c gene cause diet-induced obesity and type 2 diabetes phenotypes (Dhar et al., 2004). P4-ATPases are also agriculturally important, because they are required for pathogenesis of the rice blast fungus Magneporthe griseus (Balhadere and Talbot, 2001; Gilbert et al., 2006) and growth of plants at cold temperatures (Gomes et al., 2000).

The yeast Drs2p family of P4-ATPases, including Neo1p, Dnf1p, Dnf2p, and Dnf3p, are all involved in protein transport in the secretory and endocytic pathways, but at different stages (Graham, 2004). Drs2p and the Dnf proteins form an essential group, and at least one of these proteins must be present to sustain yeast viability. The Drs2/Dnf P4-ATPases have both overlapping and nonoverlapping functions in protein transport (Hua et al., 2002). Strains carrying a deletion of DRS2 (drs2) are viable but strongly cold sensitive for growth, and they exhibit defects in forming one of the two classes of exocytic vesicles targeted to the plasma membrane (Gall et al., 2002). The drs2 mutant also exhibits defects in protein trafficking between the TGN and early endosome that are comparable with clathrin mutant phenotypes (Chen et al., 1999; Hua et al., 2002). Thus, the Dnf ATPases cannot compensate for loss of Drs2p in these pathways; moreover, deletion of all three DNF genes does not perturb these Drs2p-dependent pathways. Dnf1p and Dnf2p are 69% identical in amino acid sequence, localize to the plasma membrane and internal membranes (TGN, early endosomes, and transport vesicles), and have redundant functions in the internalization step of endocytosis (at cold temperatures) and an early endosome to TGN transport pathway traveled by the Snc1p soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) (Hua et al., 2002; Pomorski et al., 2003). Drs2p is also required for Snc1p recycling, suggesting that Drs2p and Dnf1,2p are partially redundant in the this pathway. In addition, Drs2p and Dnf1p have redundant functions in the adaptor protein (AP)-3–dependent transport of alkaline phosphatase from the TGN directly to the vacuole. The individual drs2 or dnf1 mutants show little to no defect in the AP-3 pathway, whereas this pathway is blocked in the drs2 dnf1 double mutant (Hua et al., 2002). The mechanism for coupling a specific P4-ATPase to a specific protein transport pathway is unclear, but it likely involves translocation substrate specificity, unique protein interactions, and appropriate localization.

Localization of Drs2p to the Golgi requires interaction with the Cdc50p chaperone subunit for the Drs2p/Cdc50p complex to exit the ER (Saito et al., 2004). In cdc50, Drs2p is retained in the ER, and these cells show protein transport defects at the TGN comparable with drs2 (Chen et al., 2006). Similarly, the Cdc50p homologue Lem3p (also called Ros3p) is required for transport of Dnf1p and presumably Dnf2p to the plasma membrane and so lem3 phenocopies dnf1 dnf2 (Saito et al., 2004). The Drs2p carboxyl-terminal cytosolic tail (C-tail) makes an essential contribution to its function, apparently by mediating protein interactions and/or TGN localization. Drs2p is linked to the vesicle budding machinery by a direct interaction between the ARF-GEF Gea2p and a short motif in the C-tail (called GIM, for Gea2p interaction Motif) (Chantalat et al., 2004). Adjacent to GIM, there is a region highly conserved among all, including mammalian, Drs2p homologues. Function of this conserved motif (CM) is still unknown, although a mutational analysis suggested that the CM is primarily responsible for the essential function of the C-tail (Chantalat et al., 2004). At the membrane distal end of the C-tail, there are two NPFX_(1,2)D motifs (hereafter referred to as NPFXD), which are potential endocytosis signals (Tan et al., 1996; Howard et al., 2002).

In yeast, two types of endocytosis signals have been characterized that recruit membrane proteins into a clathrin/actin-based endocytic pathway for internalization from the plasma membrane: sequences that mediate phosphorylation and ubiquitination of cargo, such as PEST-like sequences, and the NPFXD motif (Tan et al., 1996; Roth et al., 1998; Rotin et al., 2000; Howard et al., 2002; Hicke and Dunn, 2003). The NPFXD signal is recognized by the Sla1p subunit of an endocytic coat complex consisting of clathrin, Pan1p, End3p, Sla2p/End4p (related to mammalian Hip1R), and Sla1p (related to mammalian CIN85 and intersectin) (Tang et al., 1997, 2000; Howard et al., 2002; Newpher et al., 2005; Kaksonen et al., 2006). Pan1p, a member of the Eps15 family of modular scaffolding proteins, interacts with the clathrin binding proteins AP180 and epsin, and it also binds to and stimulates the ARP2/3 complex (Wendland and Emr, 1998; Duncan et al., 2001; Aguilar et al., 2003). Therefore, Pan1p has the capacity to link adaptor-bound cargo proteins to clathrin-coated pits and sites of actin assembly. Pan1p, End3p, and actin assembly are required for both ubiquitin (Ub)-dependent and NPFXD-dependent endocytosis, although Sla1p is only required for endocytosis of cargo bearing the NPFXD signal (Howard et al., 2002; Miliaras et al., 2004). Drs2p not only has the potential to physically interact with the Sla1p/Pan1p/End3p complex but also it is functionally linked to this complex as drs2 is synthetically lethal with the temperature-conditional pan1–20 allele (Chen et al., 1999). However, the nature of these relationships between Drs2p and this endocytic complex is unclear.

Drs2p exhibits a steady-state localization to the TGN, although recent reports showed accumulation of Drs2p on the plasma membrane of a verprolin (vrp1) mutant and the presence of Drs2p in exocytic vesicles, suggesting that Drs2p transits the plasma membrane as part of its trafficking itinerary (Saito et al., 2004; Alder-Baerens et al., 2006). Dnf1p also seems to cycle between the exocytic and endocytic pathways (Hua et al., 2002; Saito et al., 2004). In this work, we further examined the trafficking itinerary of Drs2p and Dnf1p and tested whether the NPFXD motifs contribute to the function and localization of these P4-ATPases.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Media and Strains
Yeast were grown in standard rich medium (YPD) or synthetic defined (SD) minimal media containing the required nutritional supplements (Sherman, 1991). Yeast transformations were performed using the lithium acetate method. Escherichia coli strains DH5 and XL1-Blue were used for plasmid construction and amplification.

Yeast strains used in this study are summarized in Table 1. The yeast knockout strain collection was originally purchased from Research Genetics (Huntsville, AL), which is now Resgen, Invitrogen (Carlsbad, CA). Strains carrying multiple disruptions were generated by standard genetic crosses and tetrad dissection. The genotype of each spore was determined by a PCR method as described by the Saccharomyces genome deletion project (http://sequence-www.stanford.edu/group/yeast_deletion_project/deletions3.html). Strains expressing Myc and HA tagged Dnf1p were generated by PCR-based targeting into BY4741 and BY4741 sla1 by using pPF6a-13Myc-HisMX6 or pPF6a-3HA-HisMX6 as the PCR template (Longtine et al., 1998). Transformants were selected on SD plates without histidine and the integrated tags were confirmed by PCR.

For construction of GFP-DRS2, a 1.3-kilobase (kb) SpeI/ClaI fragment from pGOGFP (Cowles et al., 1997) consisting of the PRC1 promoter and GFP(S65T) was inserted into pRS416 to generate pRS416-GFP. The plasmid pRS315-DRS2 was used as a PCR template with primers SalI-Drs2-F (ACGTAGTCGACAATGACGACAGAGAAACCCCC) and Drs2-CT-R (CCCCTCGAGGTCGACGGTA) to generate a 3.7-kb fragment that placed SalI sites at both the start and end of the DRS2 coding region. This fragment was subcloned into SalI site of pRS416-GFP, creating the plasmid pGFP-DRS2. To eliminate mutations produced by PCR, most of the DRS2 coding sequence in pGFP-DRS2 was further replaced by an AgeI/ClaI fragment from pRS315-DRS2. This form of GFP-DRS2 fully complemented the cell growth defect of drs2 at 20°C. To generate C-terminal tail mutated GFP-DRS2 (pGFP-Drs2-NPF2, pGFP-Drs2-NPF, or pGFP-Drs2-NPW1,2), PCR amplifications of different regions of DRS2 C terminus were used to replace the NheI/ClaI region of pGFP-DRS2. To generate N-terminal–truncated GFP-DRS2 (pGFP-Drs2-N2 or pGFP-Drs2-N3), primers Drs2N2F (GATGAGATCTCATGAAAATCTATTTATGAGCAAT) or Drs2N3F (GACTGAGATCTCGAGCAGTCAAGCCTCCC) were used with Drs2NR (GAACCACAGTTGGGGTATCAG) to produce fragments to replace the BglII region of pGFP-DRS2. The 1.4-kb NheI/ClaI fragment of pGFP-Drs2-NPW1,2 was used to replace the corresponding sequence in pGFP-Drs2-N3 to generate pGFP-Drs2-N3-NPW1,2. The PRC1 promoter is stronger than the DRS2 promoter, and so to avoid accumulation of GFP-Drs2p in the ER (Saito et al., 2004), we cotransformed yeast strains with a multicopy vector carrying CDC50 (pRS425-CDC50).

To generate pGFP-CDC50, pRS315-CDC50 was used as template with primers CDC50KpnIF (CGGTACCGTTTCATTGTTCAAAAGAGGTA) and CDC50KpnIR (CGGTACCCACAAATACCTACAGGCACTA) to produce a 1.2-kb fragment with KpnI sites at both ends of the CDC50 coding region. The fragment was subcloned into the KpnI site of pRS416-GFP.

Microscopy
Cells were observed using an Axioplan microscope (Carl Zeiss, Thornwood, NY). Fluorescent images were captured with a charge-coupled device camera and processed with MetaMorph 4.5 software (Molecular Devices, Sunnyvale, CA). To visualize green fluorescent protein (GFP)-tagged proteins, cells were grown to early mid-logarithmic phase, harvested, and resuspended in imaging buffer (10 mM Tris-HCl, pH 7.4, and 2% glucose). Cells were mounted on glass slides and observed immediately using a GFP (green) bandpass filter set.

To study the kinetics of GFP-Drs2p transport to the plasma membrane, mid-log phase cells were collected and resuspended in SD medium containing 200 µM latrunculin A. Samples of cells were harvested at different time points and imaged. To label endosomes, we incubated cells in ice-cold SD medium containing 10 µg/ml FM4-64 (Invitrogen) for 20 min. Cells were washed twice with ice-cold medium without FM4-64 and then incubated for 30 min at 30°C before microscopic examination.

Subcellular Fractionation and Immunological Methods
For subcellular fractionation experiments, 25 OD₆₀₀ units of each strain were grown to an OD₆₀₀ of 0.5–1.0. The cells were harvested and converted to spheroplasts in HB buffer (1.4 M sorbitol, 50 mM KPi, pH 7.5, 10 mM NaN₃, 10 mM NaF, and 40 mM -mercaotpethanol), by using 200 µg/ml Zymolyase 100T (MP Biomedicals, Irvine, CA) at 30°C for 30 min. The spheroplasts were washed twice with HB buffer and lysed by resuspension in triethanolamine (TEA) lysis buffer (0.5 M sorbitol, 25 mM TEA, pH 8.0, and 1 mM EDTA) containing 1X Complete protease inhibitor cocktail (PIC) lacking EDTA (Roche Diagnostics, Basel, Switzerland). The extract was centrifuged at 400 x g for 5 min, and the resulting supernatant was centrifuged at 13,000 x g for 15 min in a refrigerated microcentrifuge. After each centrifugation step, the supernatant was transferred to a separate tube, and the pellet was resuspended in an equal volume of TEA lysis buffer supplemented with PIC. SDS/urea buffer was added to 1X (20 mM Tris-HCl, pH 6.8, 4 M urea, 0.05 mM EDTA, 0.5% -mercaptoethanol, 2.5% SDS, and 0.125% bromphenol blue), and the samples were heated at 65°C for 10 min before electrophoresis.

Immunoblotting and immunofluorescence experiments were performed as described previously (Chen et al., 1999). The 9E10 mouse monoclonal c-Myc antibody (Oncogene Research Products, Darmstadt, Germany) was used at 1:2000 for Western blot and 1:100 for immunofluorescence. Polyclonal rabbit anti-Pma1p antibody was a gift from Amy Chang (University of Michigan, Ann Arbor, MI), and was used at 1:1000 to detect Pma1p by Western blot. Polyclonal rabbit anti-G6PDH antibody (Sigma-Aldrich, St. Louis, MO) was used at 1:10,000 dilution. Alexa-594 goat anti-mouse IgG (Invitrogen) was used at 1:200 as secondary antibodies for immunofluorescence.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The C-tail Is Essential for Drs2p Function in Protein Transport
Previously, we had shown that the ATPase dead drs2-D560N allele and a drs2 allele bearing a truncation of the last 96 amino acids of the C-tail (drs2-CT) could not complement the cold-sensitive growth defect of drs2 (Chen et al., 1999; Chantalat et al., 2004). To test whether these alleles could complement drs2 trafficking defects, we examined the localization of the exocytic vesicle SNARE Snc1p, which cycles between the plasma membrane, early endosomes, and the TGN (Lewis et al., 2000). In wild-type cells, although a small fraction of GFP-Snc1p localizes to punctate structures within the cell, GFP-Snc1p primarily localizes to the plasma membrane, concentrating in the bud or the regions of polarized growth. In contrast, drs2 cells carrying an empty plasmid exhibited very little GFP-Snc1p at the plasma membrane, and most of this fusion protein was found in internal structures, which may be either early endosomes or the TGN (Figure 1, empty; Hua et al., 2002). Introduction of a plasmid bearing wild-type DRS2 restored normal plasma membrane localization of GFP-Snc1p, but neither the drs2-D560N mutant nor the drs2-CT allele was able to restore normal localization of GFP-Snc1p (Figure 1; D560N, CT).

View larger version (67K): [in this window] [in a new window]   Figure 1. ATPase activity and the C-tail region are essential for Drs2p function in protein trafficking. (A) Requirement for Drs2p ATPase activity and the C-tail region in Snc1p-GFP recycling. Plasmids pRS315 (empty), pRS315-DRS2 (DRS2), pDRS2-D560N (D560N, ATPase dead), and pRS315-Drs2-CT (CT) were introduced into strain ZHY615M2D (drs2) along with pRS416-GFP-SNC1(GFP-Snc1). Transformants were grown at 30°C to mid-log phase and examined by fluorescence microscopy. Snc1-GFP is at the plasma membrane of DRS2 (wild-type) cells and is trapped in internal membranes in the drs2 mutants. (B) Requirement for Drs2p ATPase activity and the C-tail region in ALP transport to the vacuole. The same DRS2 plasmids as in A were cotransformed into strain ZHY2149D (drs2dnf1) with pGO41 (GFP-ALP). Cells were grown at 30°C to mid-log phase, shifted to 15°C for 2 h, and then imaged. Fluorescent rings in the DRS2 (wild-type) cells are vacuoles, whereas GFP-ALP was mislocalized to extravacuolar puncta in drs2dnf1 mutants.

Functional Requirement for the Drs2p NPFXD Motifs
Three motifs have been mapped within the Drs2 C-tail thus far (Figure 2A): the Gea2p interaction motif (GIM), a highly conserved motif (CM), and the two NPFXD motifs, which could potentially interact with the Sla1p homology domain 1 (SHD1) of Sla1p (Howard et al., 2002). A two-hybrid analysis was done to test for an interaction between the Drs2p C-tail and the Sla1p SHD1 domain. The Eps15 homology (EH) domain of Pan1p interacts with NPF motifs (Wendland and Emr, 1998), and so the Pan1-EH domain was also tested for interaction with the Drs2p C-tail. The Drs2p C-tail gave a positive two-hybrid interaction with the Sla1p SHD1 domain but did not interact with the Pan1p EH domain (Figure 2B). The Drs2p C-tail also failed to interact with a fragment of Sla1p containing the SHD2 domain and a number of charged amino acids (Figure 2B, Sla1-charged). Deletion of the C-terminal region containing the two NPFXD motifs abolished the two-hybrid interaction as did mutating both NPFXDs to NPWXD (Drs2-CTNPF and Drs2-CT-NPW1,2). The F-to-W mutation was previously shown to disrupt the two-hybrid interaction of Sla1p with the NPFXD motif of Kex2p and to disrupt the ability of this motif to serve as an endocytosis signal (Howard et al., 2002). In contrast, deletion of the C-tail GIM sequence had no effect on the interaction (Drs2-CTGIM). These data indicate that the interaction between Drs2p and Sla1p is mediated by the NPFXD motifs.

View larger version (57K): [in this window] [in a new window]   Figure 2. C-tail sequences containing the NPFXD motifs bind Sla1p and contribute to Drs2p function. (A) Predicted topology and domain structure of Drs2p based on the crystal structure of the related sarcoplasmic reticulum Ca++ ATPase (A, actuator; P, phosphorylation; N, nucleotide binding) (Toyoshima and Inesi, 2004). Schematic diagram of motifs in the Drs2p C-tail and constructs used in this study (GIM and CM). (B) Two-hybrid test for interaction between Drs2p C-tail and Sla1p SHD1 domain. Bait and prey plasmids used are described in Table 2. Serial dilutions of the cells were spotted on minimal medium with or without adenine (Ade). Growth in the absence of adenine indicates a two-hybrid interaction. (C) Synergistic defect caused by deleting both the GIM and C-terminal 44 amino acids bearing the two NPFXD motifs. Serial dilutions of drs2 strains (ZHY615M2D) expressing the indicated constructs were tested for their ability to grow at 20°C. DRS2 is the wild-type gene and "empty" received the vector without an insert, thus showing the drs2 growth phenotype. (D) Western blot of cell lysates probed for Drs2p and clathrin light chain (Clc1p). Lysates were normalized for cell equivalents and compared with a dilution series from cells expressing wild-type DRS2.

To determine how mutations of the C-tail affect expression of Drs2p, we performed a Western blot with whole cell lysates from the strains indicated in Figure 2C. Most of the mutants were expressed at lower levels than wild-type Drs2p (DRS2), with CT being most affected at significantly <10% expression (Figure 2D). However, we have previously shown that coexpression of CT and the ATPase dead D560N allele from two separate plasmids complements the cold-sensitive (cs) growth defect of drs2, which indicates that this small amount of CT provides sufficient ATPase activity to support Drs2p function reasonably well (Chantalat et al., 2004). Importantly, all other C-tail mutant proteins were more stable than CT. Therefore, whereas protein stability might be a factor that influences the ability of C-tail mutants to complement drs2, each mutant should supply sufficient Drs2p ATPase activity for in vivo function, and loss of specific sequence within the Drs2p C-tail is primarily responsible for reduced Drs2p function. Particularly relevant is the observation that NPF and GIM-NPF are expressed at a similar level, but the GIM-NPF double mutant is much more defective than either single mutant (Figures 2, C and D).

Moreover, we were surprised to find that deletion of even one NPFXD motif was sufficient to cause synthetic lethality with pan1-20 (Figure 3A). For this experiment, viability of a drs2 pan1-20 strain was maintained by the presence of wild-type DRS2 on a URA3-based plasmid (pRS416-DRS2). Various mutant drs2 alleles were introduced into this strain on LEU2 plasmids and cells capable of losing the wild-type DRS2-URA3 plasmid were selected on medium containing 5-fluoroorotic acid (5-FOA). Growth in the absence of 5-FOA shows the phenotype of the pan1-20 single mutant, whereas the 5-FOA plate shows the drs2 pan1-20 double mutant phenotype. The drs2 (empty) and CM alleles are synthetically lethal with pan1-20, although the GIM allele is not. Deletion of the C-terminal 22 residues (End) had little effect on Drs2p function by this assay, but constructs bearing additional deletions removing one (NPF2) or both NPFXDs (NPF) failed to support growth of a pan1-20 drs2 strain. To better define the role of the NPFXDs in this genetic interaction, we mutated them to NPWXD. In this case, each individual drs2-NPW allele supported growth of the drs2 pan1-20 strain (NPW1 and NPW2), but the drs2-NPW1,2 double mutant (NPW1,2) failed to complement the drs2 pan1-20 synthetic lethality (Figure 3A). This allele-specific genetic interaction indicates that a yeast strain compromised for Pan1p activity relies on an NPFXD-dependent function of Drs2p to sustain life.

View larger version (43K): [in this window] [in a new window]   Figure 3. Mutation of Drs2p NPFXD motifs causes synthetic lethality with pan1-20. (A) Serial dilutions of ZHY823 (drs2 pan1-20 pRS416-DRS2) expressing the indicated constructs from LEU2-based plasmids were tested for growth at 30°C on medium with or without 5-FOA. Failure to grow on the +5-FOA medium indicates synthetic lethality between pan1-20 and the drs2 allele expressed from the LEU2 plasmid. (B) The experiment shown in Figure 2C was repeated to include GIM combined with the NPW point mutations. (C) The drs2-npw1,2 allele is not synthetically lethal with arf1. Serial dilutions of CCY2808 (arf1 drs2-2 pRS416-DRS2) expressing the indicated constructs from LEU2-based plasmids were tested for growth at 30°C on medium with or without 5-FOA.

GIM-NPF

To further test whether the genetic interaction is specific to pan1-20, we also examined the synthetic lethality between drs2 mutants and arf1 (Figure 3C). As expected, the ATPase dead drs2-D560N allele failed to support growth of a drs2–2 arf1 strain. The drs2-CT allele causes slow growth when combined with arf1 (data not shown), but drs2 alleles carrying deletions of CM, GIM, or NPFXD motifs do not substantially perturb growth of arf1 cells. Therefore, the synthetic lethal interaction between drs2-npf alleles and pan1-20 is specific.

Functional Requirement for the Dnf1p NPFXD Motif
Interestingly, Dnf1p contains an NPFXD within its N-terminal cytosolic tail (N-tail) that is not present in the closely related Dnf2p. In addition, Dnf1p, but not Dnf2p, has redundant functions with Drs2p at the TGN, suggesting that the NPFXD-dependent endocytosis of Dnf1p may be required for its Golgi function. To test whether the NPFXD is important for Dnf1p function, we mutated NPF to NAI, the sequence found in Dnf2p (Figure 4A). Because there is no significant phenotype associated with deleting DNF1 alone, we tested the dnf1-NAI allele for complementation of growth defects associated with the drs2 dnf1 double mutant. On minimal medium, drs2 dnf1 grows well at 30°C but is strongly cs and ts for growth (Figure 4B, empty). Transformation with wild-type DRS2 completely complements these growth defects (Figure 4B, DRS2), showing the robust growth of a dnf1 single mutant. Transformation with wild-type DNF1 allowed growth at 24 and 37°C, although these drs2 DNF1 cells grew more slowly than the DRS2 dnf1 cells. In contrast, dnf1-NAI weakly complemented the cs growth defect and failed to complement the ts growth defect. Therefore, the NPFXD motif plays an important role in the ability of Dnf1p to compensate for the loss of Drs2p.

View larger version (54K): [in this window] [in a new window]   Figure 4. An NPFXD/Sla1p interaction contributes to Dnf1p function and endocytosis. (A) Sequence alignment of Dnf1p and Dnf2p in the region surrounding the Dnf1p NPFXD motif. (B) Plasmids bearing wild-type DRS2 (pRS315-DRS2), DNF1 (pRS313-DNF1), empty vector, and NPF to NAI mutated DNF1 (pRS313-Dnf1-NAI) were introduced into drs2dnf1 (ZHY2149D). Cell growth on minimal medium at 37, 30, and 24°C was examined. (C) Growth of wild-type (BY4742), sla1 (KLY011), drs2 (ZHY615M2D), and sla1 drs2 (KLY035) was examined at 37, 30, and 23°C. (D) Fluorescence microscopy of wild-type and sla1 cells expressing Dnf1-Myc and stained with a mouse monoclonal anti-Myc antibody. Arrowheads indicate regions of labeled plasma membrane. (E) Distribution of Dnf1p-HA between the plasma membrane and internal membranes. Wild-type, sla1, and end3 cells expressing HA-tagged Dnf1p were osmotically lysed and centrifuged at 400 x g to clear the cell debris. The supernatants were subsequently centrifuged at 13,000 x g for 15 min to generate P and S fractions. Samples from each fraction were immunoblotted for Dnf1p-HA, the plasma membrane H+-ATPase (Pma1p) and the Golgi protein Mnn1p. The arrow indicates a background protein that distributes in the S fraction.

NPFXD-dependent Endocytosis of Dnf1p and Drs2p
To directly determine whether the endocytosis of Dnf1p was dependent on Sla1p, the localization of Dnf1p-Myc was examined in both wild-type and sla1 cells by indirect immunofluorescence (Figure 4D). In wild-type cells, Dnf1p-Myc is localized to both the plasma membrane and internal membranes with a polarized distribution, which could be transient endocytic and/or exocytic vesicles. Dnf1p is concentrated at the emerging bud site, small buds, and the mother-daughter neck of dividing cells (Hua et al., 2002; Pomorski et al., 2003; Figure 4D). Relative to wild-type cells, sla1 cells showed an accumulation of Dnf1p-Myc on the plasma membrane, although the change in the localization pattern was subtle by this method. To more quantitatively address the distribution of Dnf1p, a fractionation approach was used (Figure 4E). Cells expressing HA-tagged Dnf1p were converted to spheroplasts, osmotically lysed, centrifuged at 400 x g to pellet unlysed cells and large membranes, and then centrifuged at 13,000 x g to produce pellet (P) and supernatant (S) fractions. In wild-type cells, most Dnf1p-HA was found in the S fraction, which was relatively devoid of plasma membrane as judged by the distribution of the plasma membrane H⁺-ATPase Pma1p. However, with sla1 and end3, the amount of Dnf1p-HA in the S fraction was diminished with a concomitant increase in the plasma membrane P fraction. As a control, the distribution of an integral membrane glycoprotein of Golgi complex Mnn1p was examined, and no difference was observed between wild-type, sla1, and end3 cells. Moreover, the HA-tagged Dnf1-NAI mutant showed a 1.4-fold increase in the pellet fraction relative to wild-type HA tagged Dnf1p (our unpublished data). The end3 strain showed a more substantial redistribution of Dnf1p to the P fraction than sla1 or wild-type cells expressing Dnf1-NAI. In total, these experiments indicate that the NPFXD/Sla1p interaction significantly contributes to endocytosis of Dnf1p, but other endocytosis signals likely exist in this protein.

Even though Drs2p localizes to the TGN with Kex2p (Chen et al., 1999), the fact that Drs2p contains two NPFXD motifs suggests that it might travel to the plasma membrane and get rapidly endocytosed by the NPFXD/Sla1p pathway to maintain a steady-state TGN localization. If this is the case, we would expect to see accumulation of Drs2p on the cell surface when either the NPFXD motifs are mutated or SLA1 is deleted. To test this hypothesis, wild-type or NPFXD mutated GFP-Drs2 fusion proteins were expressed in either wild-type or sla1 cells (Figure 5). The GFP-DRS2 construct used here fully complements the drs2 cs growth defect and localizes appropriately to the TGN based on its colocalization with Sec7-RFP (Chen et al., 2006). We fused GFP to the N terminus of Drs2p to avoid interference with the potential trafficking signals near the C terminus. Surprisingly, neither truncation of the C-terminal region containing the two NPFXD motifs (GFP-NPF) nor mutation of both NPFXD motifs to NPWXD (GFP-NPW1,2) caused accumulation of Drs2p on the plasma membrane of wild-type cells (Figure 5, WT). Similar results were obtained for sla1 cells, with neither the wild-type nor the NPFXD GFP-Drs2p being mislocalized (Figure 5, sla1).

View larger version (80K): [in this window] [in a new window]   Figure 5. Drs2p does not rely on its NPFXDs and Sla1p for endocytosis unless the endocytic machinery is compromised. A series of plasmids harboring GFP-tagged wild-type (pGFP-DRS2) or NPFXD mutated drs2 alleles (pGFP-NPF2, pGFP-NPF, pGFP-NPW1,2) were cotransformed with pRS425-CDC50 into wild-type (BY4742), sla1 (KLY011), or end3 (BY4742 YNL084C) cells. Transformants were grown to early log phase at 30°C and imaged by fluorescence microscopy at room temperature. Arrowheads indicate plasma membrane fluorescence.

These results were unexpected and suggested that Drs2p did not normally travel to the plasma membrane, but deletion of the NPFXD motifs caused mislocalization of Drs2p to the plasma membrane where it could be trapped behind the end3 block. This was, however, contradictory to a published report showing Drs2p-GFP accumulates on the plasma membrane upon deletion of verprolin (vrp1), a protein required for proper organization of cortical actin patches and the internalization step of endocytosis (Saito et al., 2004). In addition, NPFXD-mediated endocytosis reportedly requires End3p function (Tan et al., 1996). To resolve these discrepancies, we examined localization of GFP-Drs2 in vrp1 as well as end4-1 (also known as sla2), another endocytosis mutant. GFP-Drs2 accumulated at the plasma membrane of both vrp1 and end4-1 cells (Figure 6), in agreement with previously published vrp1 data (Saito et al., 2004). This result led us to suspect that the end3 strain from the yeast knockout collection contained an extragenic suppressor that specifically restored function of the NPFXD/Sla1p pathway. By backcrossing the original end3 strain to wild-type cells, new end3 strains were isolated that exhibited a tighter temperature-sensitive growth phenotype. Using a backcrossed end3 strain (KLY201), wild-type GFP-Drs2 was readily detected on the plasma membrane (Figure 6).

View larger version (64K): [in this window] [in a new window]   Figure 6. Localization of GFP-tagged Drs2p to the plasma membrane of endocytosis mutants. pGFP-DRS2 was cotransformed with pRS425-CDC50 into wild-type (BY4742), vrp1 (BY4742 YLR337C), end3 (KLY201), and end4-1 (TGY1912) cells. Transformants were grown to early log phase at 27°C and examined by fluorescence microscopy.

Cargo-selective Endocytosis Defect of pan1-20
Mutation of both NPFXD motifs in Drs2p results in synthetic lethality with pan1-20 (Figure 3), suggesting that pan1-20 must maintain an active Sla1p/NPFXD-dependent endocytosis pathway at the permissive growth temperature. Otherwise, it is difficult to understand how mutation of an NPFXD endocytosis signal would further exacerbate growth of pan1-20. To test the Pan1p requirement for endocytosis of Drs2p, we examined the localization of GFP-Drs2p and GFP-Drs2-NPW1,2p in pan1-20 at the permissive (27°C) and nonpermissive (37°C) growth temperatures. GFP-Drs2p localized appropriately to the TGN at 27°C, but in stark contrast, GFP-Drs2-NPW1,2p was primarily localized to the plasma membrane. After a 1-h shift to 37°C, both GFP-Drs2p and GFP-Drs2-NPW1,2p accumulated on the plasma membrane. Kex2-GFP, another TGN resident, did not accumulate on the plasma membrane of pan1-20 at either temperature (our unpublished data). These results indicate that at the permissive growth temperature, pan1-20 cells are defective in the Sla1p/NPFXD-independent endocytosis pathway but retain a functional Sla1p/NPFXD-dependent pathway. Both pathways are blocked at the nonpermissive temperature, causing accumulation of wild-type GFP-Drs2p at the plasma membrane.

The Sla1p/NPFXD-independent endocytosis pathway blocked in pan1-20 at all temperatures is mostly likely dependent on Ub-dependent endocytosis signals. To test this possibility, we examined the localization of Ste6p-GFP in pan1-20 cells. Ste6p is an ATP-binding cassette transporter that uses a Ub-dependent signal for endocytosis (Kolling and Hollenberg, 1994; Kelm et al., 2004; Krsmanovic et al., 2005). In wild-type cells, Ste6-GFP accumulates in the vacuole over time, and so the cells show primarily vacuolar patterns of fluorescence. However, Ste6-GFP accumulated at the plasma membrane of pan1-20 cells at 27°C (Figure 7B). Thus, Ub-dependent endocytosis is abrogated in pan1-20 at permissive growth temperatures.

View larger version (82K): [in this window] [in a new window]   Figure 7. The pan1-20 mutant exhibits a constitutive defect in Ub-dependent endocytosis and a temperature conditional defect in NPFXD-dependent endocytosis. (A) Localization of GFP-Drs2p and GFP-Drs2-NPW1,2p in pan1-20 (TGY1906) cells. This pan1-20 strain expresses wild-type Drs2p from its endogenous locus to support viability. Cells were grown to mid-log phase at 27°C with or without shifting to 37°C for 1 h before imaging. Arrowheads indicate plasma membrane fluorescence. (B) Ste6p-GFP localization in wild-type and pan1-20 cells at 27°C. Wild-type (SEY6211) and pan1-20 (TGY1906) cells transformed with a plasmid harboring Ste6p-GFP (pSM1493) were grown at 27°C to mid-log phase and examined by fluorescence microscopy. Arrows indicate vacuoles. Arrowheads indicate plasma membrane fluorescence.

View larger version (61K): [in this window] [in a new window]   Figure 8. Sequences in the Drs2p N-tail bearing PEST-like motifs mediate endocytosis redundantly with the NPFXD motifs. (A) The N-tail of Drs2p contains two potential PEST sequences and one poor PEST sequence. Arrowheads indicate the N-terminal truncation boundaries for the GFP-DRS2-N2 and -N3 alleles. (B) Localization of the GFP-Drs2p with N-terminal truncations in wild-type, sla1, and pan1-20 cells. pGFP-DRS2, pGFP-Drs2-N2, and pGFP-Drs2-N3 were cotransformed with pRS425-CDC50 into wild-type (BY4742), sla1 (KLY011), or pan1-20 (TGY1907) cells. Transformants were grown to early log phase at 30°C and imaged by fluorescence microscopy at room temperature. Arrowheads indicate plasma membrane fluorescence. (C) Localization of the GFP-Drs2p with both the N-terminal truncation and the NPW1,2 mutations in wild-type cells. pGFP-Drs2-N3-NPW1,2 was cotransformed with pRS425-CDC50 into wild-type cells. Transformants were grown to early log phase at 30°C and imaged by fluorescence microscopy at room temperature. Arrowheads indicate plasma membrane fluorescence.

View larger version (93K): [in this window] [in a new window]   Figure 9. Drs2p accumulates slowly on the plasma membrane after disrupting endocytosis. Wild-type cells expressing either GFP-Drs2p or Kex2p-GFP were grown to early log phase at 30°C and then incubated in SD medium containing 200 µM lat-A or dimethyl sulfoxide at 30°C for the times shown. Arrowheads indicate plasma membrane fluorescence.

View larger version (32K): [in this window] [in a new window]   Figure 10. GFP-Drs2p and GFP-Cdc50p do not accumulate in the prevacuolar compartment of class E vps mutants. vps27 and vps4 cells carrying pGFP-DRS2 and pRS425-CDC50 (A) or pGFP-CDC50 and pRS425-DRS2 (B) were grown to early mid-log phase at 30°C, labeled with 10 µg/ml FM4-64 on ice for 20 min, washed with fresh medium, and then chased at 30°C for 30 min before analysis. GFP and FM4-64 images were acquired separately and merged to show the coincidence of the two patterns. Arrowheads indicate the prevacuolar compartment.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Trafficking of Drs2p and Dnf1p
TGN-dwelling proteins have evolved different mechanisms to maintain their steady-state localization in this organelle, although transport to the endosomal system and retrieval back to the TGN seems to be a common theme. Recycling mechanisms are critical for localization of Golgi proteins because a growing body of evidence indicates that Golgi cisternae are not stable structures but instead mature from cis to trans by changing the content of resident enzymes (Losev et al., 2006; Matsuura-Tokita et al., 2006). Cisternae mature with a half-time of 1 to 2 min in the yeast system and the TGN is apparently consumed into transport carriers with multiple destinations. Therefore, TGN resident proteins must recycle back to younger cisternae every few minutes to maintain Golgi residence. Some mammalian TGN proteins, such as the mannose-6-phosphate receptor, TGN38, and furin, cycle to the plasma membrane and endosomes before returning to the TGN. These proteins require endocytic signals for removal from the plasma membrane and retrograde sorting signals to mediate endosome to TGN transport (Thomas, 2002; Ghosh et al., 2003; Traub, 2005). In yeast, the vacuolar hydrolase sorting receptors Vps10p and Mrl1p rapidly cycle between the TGN and the late endosome/PVC (Cereghino et al., 1995; Cooper and Stevens, 1996; Whyte and Munro, 2001). The TGN markers Kex2p and Ste13p are thought to primarily cycle between the TGN and early endosomes, occasionally transiting the PVC but not the plasma membrane (Abazeed et al., 2005; Foote and Nothwehr, 2006).

None of the yeast TGN proteins mentioned above accumulate on the plasma membrane when endocytosis is blocked (Cooper and Bussey, 1992; Roberts et al., 1992; Wilcox et al., 1992; Bryant and Stevens, 1997). In contrast, most (but not all) Drs2p accumulates on the plasma membrane of mutants constitutively defective for endocytosis. However, acute inactivation of endocytosis by latrunculin A treatment of cells causes a rather slow accumulation of Drs2p on the plasma membrane over the course of 3 h. These findings suggest that Drs2p is inefficiently incorporated into exocytic vesicles, and is rapidly endocytosed upon arrival at the plasma membrane. The slow delivery to the plasma membrane and rapid endocytosis leads to an undetectable amount of Drs2p on the plasma membrane of wild-type cells unless endocytosis is inhibited. It is formally possible that disruption of endocytosis causes an aberrant incorporation of Drs2p into exocytic vesicles leading to plasma membrane accumulation. However, other TGN proteins do not share this fate in endocytosis mutants arguing against a reduced fidelity of sorting TGN residents from exocytic cargo. Moreover, Drs2p can be found in exocytic vesicles that accumulate in the sec6 mutant, providing a method independent of disrupting actin or blocking endocytosis to show targeting of some Drs2p to the plasma membrane (Alder-Baerens et al., 2006). Thus, it is much more likely that disrupting endocytosis simply traps Drs2p on the plasma membrane as it undergoes its normal trafficking itinerary.

After endocytosis, Drs2p must also be efficiently removed from the endocytic pathway and transported from the early endosome to the TGN, because we fail to see significant accumulation of Drs2p or its Cdc50p subunit in the late endosome of class E vps mutants. This is similar to the recycling pathway described for the SNARE protein Snc1p and chitin synthase (Chs3p). In contrast, class E vps mutants accumulate most of Vps10p, Mrl1p, Kex2p, and Ste13p in the PVC (Cereghino et al., 1995; Brickner and Fuller, 1997; Bryant and Stevens, 1997; Whyte and Munro, 2001). Moreover, we never detect GFP-Drs2p in the vacuole, which is a common destination for many proteins that use Ub for an endocytosis signal because this modification also directs proteins into the multivesicular body pathway at the late endosome for delivery to the vacuole lumen (Katzmann et al., 2002; Hicke and Dunn, 2003). If Drs2p is ubiquitinated, it must either avoid the late endosome to escape vacuolar delivery, or the Ub must be removed to allow retrieval of Drs2p to the TGN/early endosomal system. A previous report suggested that Drs2p is a late endosomal protein based on a relatively minor localization of Drs2p-GFP to the prevacuolar compartment of a class E vps mutant (Saito et al., 2004). We disagree with this interpretation because at the rate of recycling suggested by the cisternal maturation model, the majority of Drs2p would relocate to the PVC in class E vps mutants if even a small percentage of molecules transited the PVC in each round of recycling (as observed for Ste13p and Kex2p). In contrast, a majority of Cdc50-GFP (C-terminally tagged) was reported to localize to the PVC (Misu et al., 2003), whereas we found that GFP-Cdc50p (N-terminally tagged) was primarily excluded from the PVC. We suggest that the GFP-Cdc50p fusion protein more faithfully represents the localization of the endogenous Drs2p–Cdc50p complex as these data are more consistent with the localization data for Drs2p. In summary, it is likely that Drs2p rapidly cycles between the TGN and early endosome as suggested for several other TGN proteins, and slowly cycles in the TGN plasma membrane early endosome TGN loop.

Dnf1p is localized to both the plasma membrane and cytoplasmic punctate structures typically found near the plasma membrane. The steady-state localization pattern of Dnf1p and functional studies suggest that this protein cycles in a plasma membrane endosome TGN plasma membrane pathway (Hua et al., 2002). This hypothesis was confirmed by studies of the Tanaka group (Saito et al., 2004) and those reported here, which extend these studies to demonstrate that endocytosis of Dnf1p involves the NPFXD motif and Sla1p. Interestingly, the majority of Dnf1p was reported to fractionate in membranes with the same density as the plasma membrane (Pomorski et al., 2003; Alder-Baerens et al., 2006), which we have confirmed (our unpublished data), but most Dnf1p seems to be inside the cell by immunofluorescence localization and is separable from the plasma membrane by differential centrifugation techniques. We suggest that these data could be explained by efficient incorporation of Dnf1p into exocytic vesicles at the TGN (and/or endosome) and efficient endocytosis at the plasma membrane, thus giving a primary steady-state localization to vesicles trafficking to and from the plasma membrane.

Synthetic Lethality between drs2 and pan1
Endocytosis plays an important role in remodeling the plasma membrane, internalizing extracellular nutrients, down-regulating signal transduction pathways by internalizing receptors, and retrieving proteins required for Golgi/endosome function that have escaped to the plasma membrane. The synthetic lethality between pan1-20 and drs2-npw1,2 provides the best example we are aware of for the essential role of endocytosis in retrieving Golgi proteins. The synthetic lethality between drs2 and pan1-20<