SKD1 is a member of the family of ATPases associated with cellular activities whose yeast homologue Vps4p has been implicated in endosomal/vacuolar membrane transports. When a mutant of SKD1 that lacks ATPase activity [SKD1(E235Q)] was overexpressed in mammalian cells, it induced a dominant negative phenotype characterized by aberrant endosomal structures (denoted as E235Q compartments). Expression of SKD1(E235Q) caused an accumulation of basolateral recycling receptors, such as asialoglycoprotein receptor and low-density lipoprotein in polarized hepatocytes and Madin-Darby canine kidney cells, respectively, in E235Q compartments. In addition, SKD1(E235Q) also abrogated, via endosomes, transport to the trans-Golgi network, as indicated by an accumulation of TGN38 in E235Q compartments. Three lines of evidence further demonstrated that SKD1 participates in the membrane transport from early endosomes to late endosomes/lysosomes: (1) a redistribution of a late endosomal and lysosomal membrane protein endolyn in E235Q compartments; (2) an inhibition of epidermal growth factor receptor degradation, due to an accumulation of the receptors in E235Q compartments; and (3) a mis-sorting of and defect in the proteolytic processing of newly synthesized cathepsin D. An intriguing finding was that the expression of SKD1(E235Q) caused the number of lysosomes to decrease (to one-sixth of control numbers) but their size to increase (2.4-fold larger in diameter than control lysosomes). Indeed, an ultrastructural analysis revealed that the expression of SKD1(E235Q) causes an accumulation of hybrid organelles formed by direct fusion between late endosomes and lysosomes. We conclude that SKD1 regulates multiple steps of membrane transport out of early endosomes and the reformation of lysosomes from a hybrid organelle.
Endosomes are the central sorting site for molecules that traverse either the endocytic or the biosynthetic pathways (Mellman, 1996). Recycling receptors, those endocytosed from the plasma membrane (PM), return to the cell surface via sorting/recycling endosomes, while downregulation receptors are transported to the degradative compartments, namely, the late endosomes and lysosomes (Gruenberg and Maxfield, 1995). Some trans-Golgi network (TGN) resident proteins are also recycled between the TGN and the PM via early and recycling/sorting endosomes. Mannose 6-phosphate receptors (MPRs) capture newly synthesized lysosomal enzymes at the TGN, by recognizing a mannose-6-phosphate residue attached to N-linked oligosaccharide chains of lysosomal enzymes. Subsequently, MPRs release the ligands at endosomes and recycle back to the TGN for further rounds of sorting, whereas the ligands are transported to the lysosomes. The membrane transport through endosomes is highly regulated by a variety of molecules such as adaptor molecules, which are involved in the selection of cargo through the recognition of cytoplasmic tails, tethering molecules, rab GTPases and SNAREs, which are required for vesicle formation, fission and fusion (Robinson and Bonifacino, 2001; Stenmark and Zerial, 2001).
The family of AAA (ATPases associated with cellular activities)-type ATPases has been found to play a crucial role in membrane transport. One such AAA-ATPase is Sec18p/NSF (N-ethylmaleimide-sensitive fusion protein). NSF is one of the most studied AAA family proteins and appears to work as a master for either the assembly or dissociation of the SNARE complex at several steps in the membrane transport process (Hay and Scheller, 1997). Another important AAA molecule, Vps4p/End13p, was found in yeast vacuolar protein sorting (vps) mutants and has been extensively characterized. Vps4p is a yeast protein known to be required for the efficient transport of newly synthesized carboxy-peptidase Y (CPY) molecules from the TGN to the vacuole, which is the yeast counterpart of mammalian lysosomes (Babst et al., 1997). END13 is allelic to VPS4 and mutations in this gene result in deficiencies in both the endocytic and biosynthetic pathways (Zahn et al., 2001) that result in a phenotype known as class E (Raymond et al., 1992). The most characteristic feature of class E mutants is the appearance of a novel prevacuolar-like structure adjacent to vacuoles, the so-called class E compartment. This class of mutants indicates the accumulation of soluble vacuolar hydrolases, certain Golgi resident proteins and an endocytosed membrane that can be monitored using the fluorescent lipid dye, FM4-46. Therefore, Vps4p/End13p is thought to function in late endosomes/prevacuolar compartments by binding and releasing other class E Vps proteins (Vps24p and Vps32p/Snf7p) in a nucleotide-dependent cycle (Babst et al., 1998). End13p is also implicated in the formation of internal vesicles present in the multivesicular body (MVB) (Odorizzi et al., 1998).
The mouse homologue of Vps4p/End13p is the suppressor-of-potassium-transport-growth-defect-1 protein (SKD1) (Perier et al., 1994). That SKD1 is involved in crucial membrane transport events is indicated by recent studies (Bishop and Woodman, 2000; Nara et al., 2002; Yoshimori et al., 2000). Expression of a mutant SKD1 molecule, denoted as SKD1(E235Q), that lacks ATPase activity in mammalian cells, exhibited dominant negative effects on various membrane transport processes and the formation of autolysosomes. These cells developed abnormal membranous bodies with some diversity that contained endocytosed fluid and receptors as well as lysosomal markers, suggesting they are a mixture of compartments bearing either endosomal or lysosomal characteristics. Despite the works that characterized these abnormal compartments, the functional consequences of SKD1(E235Q) expression on the endocytic and biosynthetic pathways are still not clear. We show here that the alterations in membrane transport caused by the expression of SKD1(E235Q) closely resemble features of the class E phenotype in Saccharomyces cerevisiae. Our observations led us to propose that SKD1 is a master molecule that drives vesicle formation and/or membrane fission and thus functions at multiple steps in membrane transport process.
Materials and Methods
Antibodies and reagents
The monoclonal antibodies (mAb) 4A4, HA503 and HA502, specific for the cytoplasmic domain of the low-density lipoprotein receptor (LDLR), the 120 kDa lysosomal glycoprotein (lgp120) and endolyn, respectively, were provided by A. L. Hubbard (The Johns Hopkins University, Baltimore, MD). HA501, another mAb specific for endolyn was a gift from G. Ihrke (University of Cambridge, Cambridge, UK). The 6C4 mAb specific for lysobisphosphatidic acid (LBPA), was kindly provided by H. Kobayashi, RIKEN, Japan. The mouse mAbs that recognize TGN38, γ-adaptin and the epidermal growth factor receptor (EGFR) were purchased from Transduction Laboratories (Lexington, KY), while the rabbit polyclonal antibody (pAb) specific for rat cathepsin D was purchased from WAKO (Tokyo, Japan). An anti-SKD1 antibody was generated by first expressing SKD1 cDNA (Perier et al., 1994) as a fusion protein with Glutathione S-transferase (GST) in E. coli., DH5α. The protein was subsequently purified by affinity purification with a Glutathione-Sepharose 4B column (Pharmacia Biotech, Uppsala, Sweden). SKD1 was then cleaved from the GST portion by cutting with thrombin and used to immunize Japanese White rabbits. Serum was harvested and a specific polyclonal antibody (pAb) was prepared by affinity purification using purified SKD1-coupled Sepharose 4B (Pharmacia Biotech, Uppsala, Sweden). Similarly, a rabbit pAb that recognized rat cathepsin L was prepared using a GST-mature rat cathepsin L fusion protein as an immunogen, as described elsewhere (Fujita et al., 2001). The rabbit pAb specific for asialoglycoprotein receptor (ASGPR) was prepared using ASGPR affinity purified from total rat liver membrane as an immunogen (Nakada et al., 1984). A rabbit pAb against rat LGP107 was described previously (Furuno et al., 1989). Alexa 594-, Cy-3- or Cy-5-labeled goat anti-mouse and goat anti-rabbit secondary antibodies were obtained from Molecular Probes (Eugene, OR), as was lysine-fixable Texas Red dextran (molecular weight 70,000) (TR-dex), which was used to label lysosomes. Human recombinant epidermal growth factor (EGF) was purchased from Calbiochem (San Diego, CA).
Madin-Darby canine kidney (MDCK), A431 human epidermoid carcinoma, normal rat kidney (NRK), and rat fibroblast (3Y1-B) cells were cultured in Dulbecco's modified Eagle's medium (Sigma, St Louis, MO) with 10% fetal bovine serum (Life Technologies, Gaithersburg, MD). The cultured hepatocyte cell line WIF-B was grown in F-12 COON's modified medium (Sigma) supplemented with 10 μM hypoxanthine, 0.04 μM aminopterin, and 1.6 μM thymidine (HAT; Life Technologies), 50 μg/ml of streptomycin, 200 units/ml of penicillin and 0.5 μ g/ml of fungizone (Life Technologies) and 5% fetal bovine serum (Life Technologies) as described elsewhere (Ihrke et al., 1993; Shanks et al., 1994).
The adenovirus encoding wild-type LDLR was obtained from Enrique Rodriguez-Boulan (Cornell University, NY). The preparation of adenoviruses encoding both Cre recombinase and GFP-SKD1(E235Q) has been described elsewhere (Nara et al., 2002). The cells were infected in the appropriate culture medium and incubated for 24 hours prior to conducting the experiments.
Surface binding and internalization of 125I-antibodies
MAb to endolyn (HA501) and control mouse IgG were iodinated with 125I-Na (PerkinElmer Life Sciences, Boston, MA) by using iodogen (Pierce Chemical Co., Rockford, IL), according to the manufacturer's instructions. Labeled proteins were separated from free 125I-Na on a PD10 column (Pharmacia Biotech). NRK cells (approximately 4×104 cells/well) were incubated with iodinated antibodies (20 μg/ml in 0.5% BSA/DMEM, 20 mM Hepes pH 7.4 containing 40 μM leupeptin and pepstatin) either for 1 hour at 4°C (surface labeling) or 37°C (internalization) for the indicated times. Subsequently, cells were washed three times with 0.5% BSA/PBS and PBS, respectively, then solublized with the lysis buffer (0.5% Triton X-100, 0.25% deoxycholic acid, 25 mM Tris-HCl pH 7.5). The amount of cell-associated 125I-mAb and protein concentration in each sample were determined using a γ-counter and BCA protein assay kit (Pierce), respectively. To determine the nonspecific antibody binding or uptake, cells were identically treated with iodinated control mouse IgG, and these values were normalized with the labeling efficiency of both antibodies and subtracted from those of HA501. The amount of surface appearance and internalization of endolyn were represented as the count (cpm)/μg total cell protein.
Immunofluorescence and antibody transport assay
Cells were fixed with 4% paraformaldehyde for 5 minutes on ice and permeabilized with methanol for 10 minutes. For the LBPA staining, cells were fixed with 4% paraformaldehyde for 5 minutes on ice and permeabilized with 0.05% saponin. Fixed cells were rehydrated and washed in PBS for 15 minutes. After blocking in 1% bovine serum albumin (BSA) in PBS, the cells were incubated with primary antibodies in 1% BSA/PBS for 1 hour at the following dilutions: anti-LDLR (mouse ascites; 1:200), anti-ASGPR (rabbit pAb; 1:100), anti-TGN38 (mouse mAb; 1 μg/ml), anti-endolyn (mouse ascites; 1:200), anti-human EGFR (mouse mAb; 1 μg/ml), anti-rat cathepsin L (rabbit pAb, affinity purified; 10 μg/ml), anti-rat lgp120 (mouse ascites; 1:200), and human lamp1 (mouse hybridoma culture sup; 1:200). The secondary goat anti-rabbit or anti-mouse antibodies that had been conjugated with Alexa-594, Cy3 or Cy5 were used at 5-10 μg/ml.
To label lysosomes, cells were preloaded for 1 hour with the appropriate medium containing 40 μg/ml of TR-dex and then chased with normal medium for 12 hours prior to adenoviral transfection. For the antibody transport assay, cells on coverslips were washed once with Hepes-buffered serum-free medium (HSFM), cooled to 4°C and then labeled with primary antibodies in 0.2% BSA/HSFM for 15 minutes. The anti-TGN38 mAb was used at 20 μg/ml and anti-endolyn mAb (HA502; ascites) was diluted 1:100 for surface labeling of the cells. The surface-labeled cells were then transferred, either directly (for TGN38) or after washing three times with ice-cold 0.2% BSA/HSFM (for endolyn), to prewarmed normal medium and incubated for 1 hour at 37°C. The internalized antibodies were detected with either Cy3- or Alexa 594-conjugated goat anti-mouse secondary antibodies (10 μg/ml) and visualized by indirect immunofluorescence.
Adenovirus-infected 3Y1-B cells were fixed at room temperature for 1 hour with 4% paraformaldehyde and 0.25% glutaraldehyde in 0.2 M Hepes-KOH buffer (pH 7.4), and washed with PBS. The fixed cells were briefly stained with 0.001% toluidine blue for 2 minutes, detached from culture dishes with a rubber policeman in the presence of 20% ethanol, and collected by centrifugation. Cell pellets were then suspended in 1% low-melting temperature agarose and centrifuged. The cells in pellets were chilled to facilitate binding to each other. The pellets were cut into small blocks, dehydrated in a graded ethanol series at -20°C, and embedded in LR White (Polyscience, Warrington, PA). Thin sections were cut using a diamond knife in a Reichert Ultracut R and mounted on nickel grids. Sections were immunostained with a combination of the primary antibodies with protein A-gold probes, and further stained with uranyl acetate and lead citrate. Sections were then examined with a Hitachi H7500 electron microscope at an acceleration voltage of 75 kV.
All the immunofluorescence images, except for those with triple labeling, were observed on a Leica DMRB microscope (Wetzlar, Germany) and acquired through a cooled CCD camera, MicroMAX (Princeton Instruments, Trenton, NJ) and digitally processed using IPlab Software (Scanalytics, Fairfax, VA). Images of triple-labeled cells that had been stained with GFP-SKD1(E235Q), TR-dex and Cy5-labeled anti-mouse antibody were acquired by a confocal microsopy system, LSM 5 PUSCAL, Carl Zeiss (Thornwood, NY). All images from immunofluorescence microscopy, western blotting and pulse-chase experiments were assembled and labeled using Adobe PhotoShop (Adobe Systems, Mountain View, CA).
EGF stimulation and EGFR degradation assays
Both adenovirus-infected and uninfected A431 cells were grown in a 3.5 cm dish to ∼80% confluency and incubated for 16 hours with DMEM containing 1% BSA, after which they were incubated with 500 ng/ml of hEGF for a given period. The cells were harvested with a 10 mM Tris-HCl (pH 7.4) 0.15 M NaCl buffer and the protein concentration of the lysate was determined by Lowry's method (Lowry et al., 1951). Total cell lysate (10 μg) of each sample was subjected to SDS-PAGE according to Laemmli's method (Laemmli, 1970) using 7.5% acrylamide under reducing conditions, after which the gel was processed for western blotting according to standard procedures using enhance chemiluminescence (ECL) detection kit (Amersham Pharmacia). The remaining EGFR was detected with anti-hEGFR antibody and ECL bands were quantitated by NIH-image.
Pulse chase and immunoprecipitation
Both adenovirus-infected and uninfected NRK cells were grown in a 3.5 cm dish to ∼80% confluency, after which they were metabolically labeled for 15 minutes with 100 μCi/ml [35S]methionine/cysteine (EXPRESS™ Protein Labeling Mix, [35S]-Easy Tag™, New England Nuclear, Boston, MA) and chased with normal medium for specific periods. The labeled cells were lysed with lysis buffer (10 mM Tris-HCl (pH 7.4), 0.15 M NaCl, 0.1% Triton X-100, 1 mM EDTA and protease inhibitor cocktail (Sigma) and centrifuged in a microfuge for 10 minutes to remove insoluble compounds. The cell extracts and chasing medium were subsequently processed for immunoprecipitation with anti-rat cathepsin D antibody and protein A-agarose (Roche, Indianapolis, IN) as described previously (Tanaka et al., 2000). The immunoprecipitates were analyzed by SDS-PAGE using 10% acrylamide under reducing conditions. Radioactive bands were detected and quantitated with a Fuji BAS 1000 Imaging Analyzer.
We found that the use of a recombinant adenoviral system was the most effective way to express GFP-SKD1(E235Q) in a variety of cells (Nara et al., 2002). Our adenovirus was able to accomplish nearly 100% infection as judged by GFP in all cell types, including a fully differentiated hepatic cell (WIF-B) and a human epidermoid carcinoma cell (A431).
Previous study has shown that the transient expression of SKD1(E235Q) in cultured cells led to aberrant endosomes that were highly vacuolated and mostly positive for SKD1(E235Q) (Bishop and Woodman, 2000; Yoshimori et al., 2000). We found membranous structures that were unusually enlarged and sometimes swollen but lacked SKD1(E235Q) (Figs 1, 6). We denoted both types of compartments as `E235Q compartments', because they had been induced only by expression of SKD1(E235Q).
Basolateral recycling receptors, LDLR and ASGPR, are trapped in E235Q compartments
We have examined the possible involvement of SKD1 in the recycling of the receptors from early and recycling/sorting endosomes to the PM. To examine whether SKD1(E235Q) expression indeed affects the distribution of LDLR, Madin-Darby canine kidney (MDCK) cells were co-infected with two adenoviruses, one expressing LDLR and the other expressing GFP-SKD1(E235Q) and their intracellular distribution was analyzed by immunofluorescence microscopy. In the polarized MDCK cells infected with LDLR-expressing adenovirus alone, virally expressed LDLR localized to the basolateral PM and early endosomes that were closely aligned to the lateral membrane (Fig. 1A,B; arrows). The doubly-infected cells had phase lucent-swollen vacuoles (Fig. 1C; asterisks) and showed a significant redistribution of LDLR from PM to the swollen vacuoles, as well as small vesicles clustered in perinuclear regions, which probably have the E235Q compartments (Fig. 1A,B; arrowheads), whereas the extent to which LDLR and GFP-SKD1(E235Q) colocalized varied between cells. Notably, the cell surface and peripheral endosomal distribution of LDLR was significantly decreased in SKD1(E235Q)-expressing cells (Fig. 1B; asterisks).
The asialogylcoprotein receptor (ASGPR) is a hepatocyte-specific receptor that is localized to perinuclear recycling endosomes as well as to early endosomes very close to the basolateral PM in polarized cultured WIF-B hepatocytes (Fig. 1D,E; arrows). Expression of SKD1(E235Q) in WIF-B cells led to the appearance of swollen vacuoles (Fig. 1F; asterisks) as in MDCK cells, but did not alter the cell polarity as judged by the normal apical PM distribution of 5′-nucleotidase (data not shown). The ASGPR in the SKD1(E235Q)-expressing WIF-B cells was redistributed into E235Q compartments. In contrast, there were few or no ASGPR-positive peripheral endosomes near the PM (Fig. 1D,E; arrowheads) in the infected cells. While some ASGPR colocalized with GFP-SKD1(E235Q), the extent of colocalization varied between the cells, similar to what was observed for LDLR (see above). It is likely that the distribution of ASGPR is affected by SKD1(E235Q) expression in a fashion similar to that of LDLR. Thus we conclude that the recycling of receptors from early and recycling endosomes to basolateral PM in polarized epithelial cells was regulated by SKD1 ATPase activity.
Recycling of TGN38 is abrogated by the expression of SKD1(E235Q)
TGN38 and a furin, which localize in TGN, are known to be recycled between the TGN and the PM via early endosomes (Ghosh et al., 1998; Mallet and Maxfield, 1999; Reaves et al., 1993; Takahashi et al., 1995; Voorhees et al., 1995). TGN38 in SKD1(E235Q)-expressing NRK cells, like LDLR and ASGPR, localized to the E235Q compartments. Although the majority of TGN38 antibody staining still represented typical TGN staining as seen in uninfected cells, there was significant redistribution of TGN38 into a punctate structure near the Golgi and sometimes into the swollen structures in SKD1(E235Q)-expressing NRK cells (arrows in Fig. 2A-C).
To further test the possibility that the TGN38 in the E235Q compartments is accumulated via endocytic routes, we followed the fate of exogenously applied TGN38-specific antibodies in SKD1(E235Q)-expressing NRK cells. After 1 hour of chase, the internalized TGN38-specific mAb had reached the perinuclear TGN area in uninfected cells (see arrowheads in Fig. 2D,E) whereas, in the infected cells, the mAb accumulated significantly in the E235Q compartments (see arrows in Fig. 2D-F) and could not reach the TGN area. This was not due to the destruction of the TGN structure by overexpression of SKD1(E235Q), because there was no difference in the perinuclear tubular and punctate distribution of AP-1 between the infected cells (asterisks in Fig. 2G,H) and uninfected cells. These observations indicate that, in addition to the receptors recycling between PM and early endosomes, the recycling pathway used by TGN38 is also significantly affected by the endosomal dysfunction caused by the expression of SKD1(E235Q).
SKD1(E235Q) alters the intracellular distribution and the cell surface appearance of the endosomal-lysosomal membrane protein endolyn
It is known that some lysosomal membrane proteins continuously recycle between the PM and the lysosomes via early and late endosomes (Lippincott-Schwartz and Fambrough, 1987). Endolyn is one of lysosomal membrane proteins and belongs to the highly O-glycosylated mucin-like protein family (Croze et al., 1989; Ihrke et al., 2000; Chan et al., 2001). It was shown that in polarized hepatic WIF-B cells, a fraction of endolyn internalized from the basolateral surface is delivered to endosomes/lysosomes either directly or indirectly via a subapical compartment (Ihrke et al., 1998). Recent work demonstrated that in polarized MDCK cells newly synthesized endolyn is directed to the apical surface before it reaches lysosomes (Ihrke et al., 2001). Indeed, it has been demonstrated that endolyn contains multiple targeting signals; one is an N-glycan-dependent apical targeting signals in the luminal domain of endolyn and the other is basolateral/lysosomal sorting information contained in the cytoplasmic tail of it (Ihrke et al., 2001). To clarify the effect of SKD1(E235Q) on the recycling pathway between the PM and lysosomes, we examined the distribution and recycling of endolyn in SKD1(E235Q)-expressing NRK cells. We found that the distribution of endolyn was altered by expression of SKD1(E235Q) and the protein was accumulated in E235Q compartments (Fig. 3A,B, infected cells are marked with asterisks). Then we tested the effect of SKD1(E235Q) expression on the endocytic pathway from the PM to the lysosomes by following the fate of exogenously applied endolyn-specific antibodies in SKD1(E235Q)-expressing NRK cells. We found that the surface labeling of the endolyn-specific mAb at 4°C, which indicates the degree of surface appearance of endolyn, was significantly less on SKD1(E235Q)-expressing cells than uninfected cells (Fig. 3C,D). The measurement of a cell surface binding of 125I-mAb to endolyn (HA501) further revealed that it was reduced to approximately 56% of control cells (Fig. 3G). Since about 30% of internalized antibody to endolyn could be recycled, these results suggest that SKD1(E235Q) leads to the accumulation of the internalized endolyn to the E235Q compartments and inhibits endolyn recycling to the PM. This was confirmed by the following results. First, after a subsequent chase for 1 hour at 37°C, the internalized mAb to endolyn was localized to the E235Q compartments in the infected cells, while in the uninfected cells, it was delivered to late endosomes and lysosomes (Fig. 3E,F). Second, an internalization of 125I-mAb in the cells expressing SKD1(E235Q) was reduced to less than 70% of control cells after a 2 hour incubation (Fig. 3H). Thus, we conclude that the expression of SKD1(E235Q) caused the perturbation of recycling of endolyn from endosomes to the cell surface.
EGFR downregulation is inhibited by the expression of SKD1(E235Q)
In yeast, the class E Vps proteins play a pivotal role in a degradation of Ste2p, the mating pheromone (α-factor) receptor that is known to be internalized from the cell surface and degraded in vacuoles. To examine whether SKD1 is involved in the degradative endosome-to-lysosome pathway in mammalian cells, we tested the effect of GFP-SKD1(E235Q) expression on the downregulation of EGFR in A431 human epidermoid carcinoma cells. When uninfected A431 cells were stimulated with EGF, over 80% of the cellular EGFR were degraded within 1 hour. In contrast, in GFP-SKD1(E235Q)-expressing infected cells, even after 4 hours of ligand stimulation, more than 60% of the EGFRs remained in the cells without being degraded (Fig. 4A,B). The quantitative immunoblotting of cell lysates from adenovirus-infected A431 cells revealed that the infection generated 6-8 times more mutant SKD1(E235Q) molecules relative to the endogenous SKD1 in uninfected A431 cells. Although in the infected A431 cells both EGF and EGFR were internalized from the PM at the same rate as in the control cells, both of them were accumulated in the E235Q compartments (Fig. 4C). Furthermore, the fluorescence signal intensity from the antibody specific for EGFR revealed that the endocytosed EGFR was not degraded and could be detected in the E235Q compartments for at least 2 hours (Fig. 4C, see asterisks and arrowheads). In contrast, most of the uninfected cells displayed little or no staining with the anti-EGFR antibody (Fig. 4C, see n.i.). These results indicated that as in the case of yeast class E mutants, SKD1(E235Q) inhibited the downregulation of EGFR following the endocytosis and transport to the degradative compartments.
Processing of a newly synthesized lysosomal acid hydrolase, cathepsin D, into the mature form is impaired by the expression of SKD1(E235Q)
Vps4-deletion and temperature-sensitive Vps4 mutants in yeast exhibit an accumulation of CPY, a soluble vacuole enzyme in class E compartments, which delays their maturation from the p2-form to the mature form and causes the partial secretion of the p2-form (Babst et al., 1997; Zahn et al., 2001). We examined the effect of SKD1(E235Q) expression on the transport of a newly synthesized soluble lysosomal enzyme, cathepsin D, to lysosomes by pulse-chase experiments of NRK cells with [35S]methionine/cysteine. As shown in Fig. 5, in NRK cells, cathepsin D was synthesized as a 45 kDa pro-form (pro-catD), and was subsequently processed to a 43 kDa mature form (mature catD). Although in uninfected cells proteolytic processing of pro-catD to the mature form was completed after a chase of 3 hours, in SKD1(E235Q)-expressing cells more than 50% of the newly synthesized catD remained as a pro-form even after a chase of 3 hours. In addition to the retardation of proteolytic processing, the expression of SKD1(E235Q) also resulted in a hyper-secretion of pro-catD into the medium (Fig. 5, see EQ infection, right). As in the case of CPY-processing in class E mutants in yeast, SKD1(E235Q) led to a delay in the processing and hypersecretion of pro-catD.
Expression of SKD1(E235Q) induces the accumulation of hybrid compartments that bear features of both lysosomes and late endosomes
To analyze the effect of SKD1(E235Q) expression on the morphology of lysosomes in NRK cells, Texas-Red dextran was preloaded into cells and then adenovirus was used to infect cells. In uninfected cells, TR-dex-labeled lysosomes can be observed as fine punctates that are mostly found near the nucleus but are also present throughout the cell (Fig. 6A). However, in infected cells (Fig. 6A, see asterisks) the TR-dexlabeled lysosomes were significantly less frequent (average number/cell is 10.6±4.2 versus 58.4±21.4 in normal cells) and were not concentrated near the nucleus but scattered throughout the cells. The lysosomes in the infected cells were also considerably larger than those in uninfected cells (average diameter is 1.19±0.47 μm versus 0.48±0.26 μm, range is 0.8-2.0 μm versus 0.2-0.8 μm) (Fig. 6E). Interestingly, these enlarged TR-dex-labeled structures were mostly not associated with GFP-SKD1(E235Q). The enlarged lysosomes were also positive for cathepsin L and lysobisphosphatidic acid (LBPA) (Kobayashi et al., 1998), respectively (Fig. 6B,C). Together these observations suggest that the enlarged lysosomes induced by the expression of GFP-SKD1(E235Q) have characteristics of both late endosomes and lysosomes and thus they may be hybrids of these two organelles. Triple labeling of the cells by infection with the GFP-SKD1(E235Q)-expressing adenovirus, pre-loading with TR-dex and staining with anti-lgp120 antibody revealed the existence of two distinct lgp120-positive compartments in the infected cells (Fig. 6D). In uninfected cells, all TR-dex positive compartments perfectly colocalized with lgp120 (see purple markings in Fig. 6D), which indicates that these compartments are lysosomes. In the infected cells (see asterisk in Fig. 6D), the enlarged hybrid organelles (TR-dex+, lgp120+) poorly colocalized with GFP-SKD1(E235Q) (see arrowheads in Fig. 6D). This notion may imply that the accumulation of hybrid organelles is not due to a direct effect of SKD1(E235Q) but an indirect one, such as a depletion of the ADP-bound form of SKD1 from the cytosol. However, in these cells, some lgp120 signals were also observed in the GFP-SKD1(E235Q)-positive compartments (see light blue markings in Fig. 6D) that lack TR-dex (see arrows in Fig. 6D). In addition, there are many structures that are only labeled with GFP-SKD1(E235Q), and that appear to be aberrant and large vacuole-like structures. These latter structures may be derived from the direct association of SKD1(E235Q) with their membrane and correspond to the accumulation of early endosomes and/or intermediate compartments between early and late endosomes.
The immunofluorescence data were supported by immunoelectron microscopy. In control cells (Fig. 7A,B), both electron dense lysosomes (arrows) and MVB-like late endosomes (arrowheads) were labeled with an anti-LGP107 pAb and their diameter was less than 1 μm. In contrast, in GFP-SKD1(E235Q)-expressing cells (Fig. 7C-F), there were many aberrant membranous structures, which were positive for LGP107. We found that MVBs tended to cluster near the nucleus, docked and fused with each other (arrows in Fig. 7C,E,F). In addition, the typical hybrid organelle-like structures, MVBs fused with electron dense lysosomes, were frequently observed (arrowheads in Fig. 7C,E). Interestingly, consistent with the immunofluorescence analysis, the diameter of the MVBs was over 1 μm (asterisks in Fig. 7C,D). Taken together, we assume that SKD1(E235Q) inhibits the reformation of lysosomes from the hybrid organelle that leads to their accumulation.
SKD1 functions in multiple stages of membrane transport in mammalian cells
Based on our findings and others (Bishop and Woodman, 2000; Yoshimori et al., 2000), we propose that SKD1 participates in several steps of membrane transport, as depicted in Fig. 8. First, SKD1 regulates the efflux from the early endosomes to the PM and the TGN (Fig. 8A). Moreover, although it is still unclear whether the transport from early to late endosomes involves vesicle transport or maturation, it is likely that SKD1 regulates this step (Fig. 8A). Second, SKD1 participates in the reformation of lysosomes from late endosome-lysosome hybrid organelles (Fig. 8B). Because the expression of SKD1(E235Q) did not affect the endocytosis and fusion of late endosomes and lysosomes, it is possible that an enlargement of both early endosomes and late endosomes/lysosomes induced by expression of SKD1(E235Q) is due to a failure of membrane traffic from these compartments. We therefore conclude that SKD1 functions in the formation and/or fission of vesicles from endosomal compartments.
SKD1 regulates the membrane transport via endosomes
The accumulation of LDLR in the E235Q compartments is compatible with the observations of a previous study showing that LDL uptake is inhibited in SKD1(E235Q)-expressing CHO cells (Yoshimori et al., 2000). A likely explanation for the deficiency in LDL uptake is the absence of LDLR on the cell surface caused by its sequestration in E235Q compartments. As endocytosis itself from the PM is not impaired in SKD1(E235Q)-expressing cells (as measured by dextran, horseradish peroxidase and EGF uptake), it is likely that the accumulation of the receptors in E235Q compartments is due to the inhibition of vesicle budding and/or fission from the recycling/sorting endosomes to the PM (Fig. 8A).
We demonstrated that despite the fact that the TGN integrity was normal, a significant fraction of TGN38 was localized to the E235Q compartments. The accumulation of TGN38 in E235Q compartments is, therefore, due to the similar reason with LDLR and ASGPR, e.g. the inhibition of membrane efflux from recycling/sorting endosomes to the TGN (Fig. 8A).
We further found that expression of SKD1(E235Q) results in the accumulation of endolyn in the E235Q compartments accompanying with decrease from the PM. Recent study on endolyn traffic in MDCK cells revealed that a significant fraction (up to 50%) of newly synthesized endolyn appeared on the cell surface and about 30% of endolyn endocytosed from the PM are recycling between early endosomes and the PM (Ihrke et al., 2001). Thus we assume that a small but considerable fraction of endolyn is recycling between endosomes and the PM in NRK cells too. SKD1(E235Q) reduced the cell surface appearance of endolyn (56% of control) and the internalization of antibody to endolyn from the PM (70% of control). These reductions are ascribed to the defect in the endolyn recycling from early endosomes by the expression of SKD1(E235Q). There were still significant amounts of cell surface appearance and the internalization of endolyn in the cells expressing SKD1(E235Q). They may be attributed to the supply of newly synthesized endolyn from the TGN and the subsequent internalization of them from the PM, which are not abrogated by SKD1(E235Q). Together, our results suggest that SKD1(E235Q) impairs the membrane traffic out of early and/or recycling endosomes, leading to the accumulation of molecules that pass through there (Fig. 8A).
CI-MPR is known to be recycled between the TGN and PM via early and late endosomes (Mellman, 1996). A previous study has indicated that expression of SKD1(E235Q) altered the distribution of CI-MPR from the TGN to the E235Q compartments (Bishop and Woodman, 2000). Since CI-MPR poorly colocalizes with lgp120, a marker for late endosomes and lysosomes, in the cells expressing SKD1(E235Q), it is believed that CI-MPR is redistributed to the early rather than to late endosomal E235Q compartments. Such a redistribution of CI-MPR from the TGN to the E235Q compartments may lead to the hyper-secretion of lysosomal enzymes (see below). Taken together, the expression of SKD1(E235Q) impairs membrane transport out of early endosomes, leading to an accumulation of molecules that are delivered to the cell surface, the TGN, or late endosomes/lysosomes into early endosomal E235Q compartments (Fig. 8A). We conclude that SKD1 is a master molecule regulating the membrane dynamics of endosomes by its own ATPase activity.
SKD1(E235Q)-expressing cells have a mammalian class E phenotype
SKD1(E235Q) expression blocked the degradation of EGFR and resulted in the accumulation of both EGF and EGFR in the E235Q compartments. This is analogous to the inhibition of Ste2p degradation in several yeast class E mutants. SKD1(E235Q) expression also resulted in a delay in the proteolytic processing of cathepsin D and induced hyper-secretion of the immature form of this molecule. Similarly, several class E mutants inhibited the recycling of Vps10p, which is a receptor for CPY, from pre-vacuolar compartments to the TGN, which resulted in a delay in the processing and partial secretion of CPY (Babst et al., 1997). By analogy, the accumulation of CI-MPR in the E235Q compartments might account for the mistargeting of cathepsin D. Taken together, we conclude that SKD1(E235Q) expression in mammalian cells indicates the phenotype closely related to the class E vps in Saccharomyces cerevisiae. Class E vps mutants exhibit abnormally enlarged and multilamellar structures (known as `class E compartments') that accumulate newly synthesized vacuolar hydrolysase precursors as well as endocytosed plasma membrane proteins and Golgi-resident proteins (which normally recycle to the Golgi from the endosomes) (Piper et al., 1995; Raymond et al., 1992; Rieder et al., 1996; Zahn et al., 2001). E235Q compartments fully satisfy these criteria.
SKD1(E235Q) impairs the reformation of lysosomes from hybrid organelles
One of the most pronounced features of SKD1(E235Q)-expressing cells is the appearance of the enlarged lysosomal compartments. We assume that they are the accumulated hybrid organelles. Previous studies have suggested the existence of hybrid organelles in normal cells (Luzio et al., 2000), which appear to be derived from a direct fusion of late endosomes with lysosomes (Bright et al., 1997; Mullock et al., 1998). Their mean diameter (0.96 μm) is larger than that of either rat liver lysosomes (0.38 μm) or late endosomes (0.34 μm), and they have a density that is intermediate between that of late endosomes and lysosomes (Mullock et al., 1998). In NRK cells, ∼15% of lysosomes appear to be fused with late endosomes to form hybrid organelles and it was found that lysosomes can be reformed from this compartment (Bright et al., 1997). The following evidence also supports the hypothesis that the aberrant lysosomes in SKD1(E235Q)-expressing cells are hybrid organelles. First, the enlarged lysosomes in SKD1(E235Q)-expressing cells contained both LBPA and cathepsin L. LBPA was originally defined as a specific marker for late endosomes/MVBs in which it is concentrated within internal vesicles (Kobayashi et al., 1998). Recent studies have revealed, however, that LBPA also localizes in lysosomes as well as late endosomes/MVBs (Reaves et al., 2000). Therefore, although it is not clear whether LBPA can be used to distinguish late endosomes/MVBs from lysosomes, the fact that LBPA colocalized almost completely with cathepsin L and pre-loaded TR-dex in SKD1(E235Q) expressing cells (data not shown) reveal that the aberrant lysosomes induced by expression of SKD1(E235Q) have characteristics of the hybrid organelle. Second, the diameter of the enlarged lysosomes (preloaded with TR-dextran) in SKD1(E235Q)-expressing cells was consistent with that reported previously (Bright et al., 1997). Third, ultrastructural analysis revealed the clustering and enlargement of MVBs, which occasionally docked and fused with lysosomes. Since the number of typical lysosomes decreased in the SKD1(E235Q)-expressing cells, we speculate that despite the fact that fusion of MVBs and pre-existing lysosomes are not abrogated, the reformation of lysosomes from the hybrid organelle is inhibited by the expression of SKD1(E235Q). Interestingly, Pryor et al. showed that the reformation of lysosomes from the hybrid organelle requires ATP (Pryor et al., 2000). Although the authors concluded that proton pump (ATPase)-dependent acidification is the crucial event requiring ATP, this does not exclude the possible involvement of SKD1 ATPase activity in the reformation process.
We cannot rule out the possibility that the enlargement of lysosomes may be attributed to the secondary effect through the deficiency of either vesicle transport from or the maturation of, early endosomes. However, two previous works demonstrated that dysfunction of early endosomes did not alter the morphology of preexisting lysosomes. While the inactivation of early endosomal compartments by the crosslinking of endosomal proteins with the biotin-avidin complex formation caused the partial dysfunction of early endosomes, no morphological alteration of late endosomes/lysosomes was observed (Parton et al., 1992). The enlargement of early endosomes and retardation of membrane transport from early to late endosomes by overexpression of the invariant chain did not induce the enlargement of pre-existing lysosomes (Romagnoli et al., 1993). These results imply that early endosome dysfunction itself is not involved in the enlargement of lysosomes. Our models showing that SKD1 is involved in multiple membrane transport steps in the mammalian endocytic pathway is in agreement with a recent study on the yeast homologue of SKD1 (End13p/Vps4p), which is required for both early-to-late endosome and late endosome-to-vacuole (Zahn et al., 2001). End13/Vps4 mutant cells indicated the accumulation of FM4-64 and Ste2p in early endosomes as well as the transient accumulation of the precursor form of CPY in the late endosomes. By analogy to this, it is likely that SKD1 also functions in multiple membrane transport steps in the endocytic pathway, not only at a membrane efflux from early endosomes but also at a lysosome reformation from hybrid organelle (Fig. 8). Whether SKD1 regulates the late endocytic membrane transport directly or indirectly should be further tested experimentally in future studies.
The enlarged lysosomal organelle in SKD1(E235Q)-expressing cells resembles giant lysosomes found in cells from patients with Chediak-Higashi syndrome (CHS) (White, 1966). We tested whether SKD1 could be involved in the formation of giant lysosomes in CHS patients by overexpressing the wild-type SKD1 molecule in CHS cells. However, no changes in the frequency or size of the giant lysosomes in these cells were observed (data not shown). This suggests that the SKD1 molecule locates upstream of the CHS protein, which is a lysosomal trafficking regulator denoted as LYST (Barbosa et al., 1996) that participates in the membrane traffic cascade to the lysosomes.
Molecular mechanism of SKD1 function
It is known that SKD1 ATPase activity drives the membrane recruitment of class E Vps proteins (Babst et al., 2000; Babst et al., 1998; Bishop and Woodman, 2001). The dominant negative effect of SKD1(E235Q) is most likely due to the accumulation of the ATP-binding form of SKD1 in the membrane. Moreover, recently it was shown that a charged MVB protein (CHMP1) was recruited to E235Q compartments (Howard et al., 2001). Thus, we speculate that the overexpression of SKD1(E235Q), which is a nucleotide bound form of SKD1, inhibits the dissociation of both class E Vps and CHMP proteins from the membrane and blocks the subsequent re-utilization of these molecules in the next round of vesicle formation and/or fission. This hypothesis may account for the absence of SKD1(E235Q) itself in some of the E235Q compartments (Figs 1, 6). The mutation in another mammalian class E Vps gene, mouse Vps23/TSG101 indicated the inhibition of EGFR degradation and induction of hypersecretion of pro-cathepsin D (Babst et al., 2000), which are similar but not identical phenotypes to those of SKD1(E235Q). As tsg101 cells did not show a change in the morphology of the early endosomes and the recycling of TfR, the differences in phenotype between tsg101 and SKD1(E235Q) may suggest that SKD1 regulates the membrane association of not only class E vps proteins, including Vps23p/TSG101, but also other cytosolic factors that are responsible for the recycling steps. How do class E vps proteins regulate endosomal membrane transport? It was recently shown that the ubiquitin-dependent sorting of carboxypeptidase S into the inner vesicles of the MVB requires a 350 kDa complex, referred to as endosomal sorting complex required for transport 1 (ESCRT-1), which comprises Vps23p, Vps28p and Vps37p (Katzmann et al., 2001). They also revealed the ability of Vps23p to bind ubiquitin. SKD1(E235Q) indeed recruits hVps23p and hVps28p to the E235Q compartments (Bishop and Woodman, 2001), both of which are possible components of mammalian ESCRT-1. Thus, SKD1-dependent membrane recruitment of ESCRT-1 may regulate the degradation of EGFR, which is accompanied by ubiqutination-dependent sorting of EGFR from the limiting membrane to inner membranes of MVBs (Longva et al., 2002). Whether the accumulated EGFR in SKD1(E235Q)-expressing cells is ubiquitinated or not needs to be determined. Further study of SKD1 and its binding partners is likely to provide a greater understanding of the mechanisms involved in membrane transport in the endocytic pathways and the biogenesis of MVBs and lysosomes.
We thank C. A. Vandenberg from University of California, Santa Barbara for generously providing us with a cDNA encoding mouse SKD1; A. L. Hubbard from the Johns Hopkins University for providing WIF-B cells and the mAbs specific for endolyn (HA502), lgp120 (HA503) and LDLR (4A4); G. Ihrke from the University of Cambridge for providing HA501 mAb and technical advice for the endolyn uptake experiments; E. R. Boulan from Cornell University for providing the LDLR-encoding adenovirus; and T. Kobayashi from the RIKEN Institute for generously providing us with the LBPA-specific mAb (6C4). This work was supported in part by grants from a CREST Science and Technology Corporation of Japan, the Ministry of Labor, Health and Welfare of Japan and the Ministry of Education, Science, Sports and Culture of Japan.
- Accepted October 10, 2002.
- © The Company of Biologists Limited 2003