Lysosome proteins are redistributed during expression of a GTP-hydrolysis-defective rab5a

Organelles within the endosome/lysosome system play important roles in cellular physiology. The uptake of extracellular ligands and their transport to lysosomes has been extensively studied (Goldstein et al., 1985); however, lysosome function is critical for many other cellular processes. The termination of signaling by cell-surface receptors, such as receptor tyrosine kinases (Stoscheck and Carpenter, 1984) and G-protein-coupled receptors (Moore et al., 1999b; Trejo et al., 1998) can occur by endocytosis and subsequent receptor degradation in lysosomes. In cells of the immune system, antigen presentation may involve degradation of internalized antigens by lysosomal proteases, generating peptides that bind MHC class II molecules (Nakagawa and Rudensky, 1999). Alterations in the endosome-lysosome system are associated with some forms of Alzheimer’s disease and Down’s syndrome, possibly contributing to changes in the production and/or secretion of amyloidogenic protein by neurons (Cataldo et al., 1996; Cataldo et al., 2000).

Considerable effort has been given to understanding the biogenesis of lysosomes in mammalian cells. Proteins regarded as residents of lysosomes can be directed there from the trans-Golgi network in at least two general ways (Hunziker and Geuze, 1996). Lumenal hydrolases bind mannose 6-phosphate receptors (M6PRs) in the trans-Golgi apparatus and are shuttled to the late endosome compartment for eventual transport to lysosomes (Brown et al., 1986; Duncan and Kornfeld, 1988). Alternatively, lysosome membrane proteins may first transit to the cell surface, then internalize into early endosomes and subsequently traffic to lysosomes (Lippincott-Schwartz and Fambrough, 1986). This second itinerary may also involve trafficking of proteins from the TGN to early endosomes (Ludwig et al., 1991; Nielsen et al., 2000; Press et al., 1998), then subsequent recycling to the plasma membrane (Hunziker and Geuze, 1996). Thus, while many proteins are enriched in lysosomes, this usually is a dynamic steady-state condition resulting from complex traffic amongst several cellular compartments.

Endocytic trafficking events are in part governed by ras-related GTPases of the rab family. Rab proteins are 23-25 kDa in mass and tightly bound to membranes via C-terminal geranylgeranyl modifications (Novick and Zerial, 1997). Each rab isoform appears to be associated with a specific subcellular compartment, and some rabs are known to regulate traffic in the endosome-lysosome system. Transport to late endosomes from the trans-Golgi network of lumenal lysosome proteins liganded to M6PRs is regulated by rab9 (Lombardi et al., 1993). Early events in endocytosis are regulated by rab5a (Bucci et al., 1992), while trafficking from sorting endosomes to lysosomes is governed by rab7 (Feng et al., 1995; Mukhopadhyay et al., 1997; Press et al., 1998; Vitelli et al., 1997). The delivery of transferrin receptors from sorting endosomes to perinuclear recycling endosomes appears to be regulated by both rab11 and rab5a (Ren et al., 1998; Ullrich et al., 1996; Wilcke et al., 2000).

Rab5a is a key regulator of endocytosis because it is rate limiting for homotypic endosome fusion (Bucci et al., 1992; Gorvel et al., 1991). Rab5a mutants defective in guanine-nucleotide binding (rab5a S34N or rab5a N133I) are dominant negative in action, and when expressed in transfected cells, inhibit endosome fusion and cause the accumulation of small vesicles at the cell periphery (Bucci et al., 1992). By contrast, the GTPase defective mutant rab5aQ79L is dominant active and causes the formation of greatly enlarged endosomes due to a stimulation of endosome fusion (Stenmark et al., 1994). Overexpression of rab5a wild-type (wt) or rab5aQ79L increases the steady-state accumulation of fluid phase endocytic markers (Li and Stahl, 1993); however, the rate of receptor endocytosis itself may be unaltered (Ceresa et al., 2001; Seachrist et al., 2000). Various lines of evidence suggest that while rab5a-GTP is necessary for endosome fusion, the GTPase activity of rab5a is not (Barbieri et al., 1994; Hoffenberg et al., 1995a; Rybin et al., 1996; Stenmark et al., 1994). It has been proposed that the function of the GTPase activity is to maintain a dynamic equilibrium between rab5a-GDP and rab5a-GTP (Rybin et al., 1996) that can be regulated by GTPase activators (Lanzetti et al., 2000) and proteins that stimulate guanine-nucleotide exchange (Hoffenberg et al., 2000; Horiuchi et al., 1997). Evidence from studies of rab5a (McBride et al., 1999), rab1 (Allan et al., 2000) and the yeast rab-like protein Ypt7 (Ungermann et al., 2000) suggest a functional role for these small GTPases in the recruitment of SNARE proteins to membranes to facilitate budding or fusion reactions.

There is less information available about how rab5a activity influences the development of lysosomes. Because some fractions of lysosomal proteins reach their destinations via endosomes, experimental changes in rab5a activity might be expected to alter the morphology or function of lysosomes. Expression of the dominant-negative rab5aS34N mutant decreases the rate of endocytosis and degradation of epidermal growth factor receptors (Barbieri et al., 2000; Papini et al., 1997); however, it is unclear whether endosome to lysosome transport is affected. In addition, degradation of low-density lipoprotein (LDL) is greatly reduced by the expression of rab5aS34N, possibly due to a defect in LDL-receptor endocytosis (Vitelli et al., 1997). More direct evidence for a role of rab5a in lysosome function comes from studies of macrophage cells, where the rate of phagosome maturation is reduced by antisense inhibition of rab5a activity, while maturation is accelerated during overexpression of wild-type rab5a (Alvarez-Dominguez and Stahl, 1999). Interestingly, expression of a dominant active rab5a in MDCK cells caused the formation of enlarged rab7-positive vesicles (D’Arrigo et al., 1997), suggesting a role for rab5a downstream of early endosomes.

To further explore the role of rab5a in mammalian lysosome biogenesis, we expressed rab5a as a fusion with enhanced green fluorescent protein (EGFP) using an inducible expression system in cultured cells, and looked for changes in the distribution of lysosome proteins and several endocytic tracers. Surprisingly, modest expression of the GTPase-deficient EGFP-rab5aQ79L caused an extensive redistribution of lysosome proteins into large vesicles that also contain EGFP-rab5aQ79L, without an appreciable effect on the endocytosis or recycling of a cell surface receptor. These findings support the idea that rab5a GTPase activity plays a role in the formation of lysosomes.

EcR293 cells were obtained from Invitrogen (Carlsbad, CA) and cultured in Dulbecco’s modified Eagle medium (DMEM) with 10% fetal bovine serum and 200 μg/ml Zeocin (Invitrogen). Antibody sources were as follows: monoclonal anti-bovine cation-independent mannose-6-phosphate receptor (CI-M6PR) antibody 22D4 (a gift of J. Rohrer, F. Miescher Institute); monoclonal anti-LAMP-1 antibody CD107A (Research Diagnostics Inc., Flanders, NJ); monoclonal anti-LAMP-2 antibody CD107B (Pharmingen, San Diego, CA); rabbit polyclonal anti-cathepsin D (Calbiochem, San Diego, CA); monoclonal anti-rab5a antibody, clone 15 (BD Transduction Labs, San Diego, CA). Fluorescent secondary antibodies (goat anti-rabbit IgG and goat anti-mouse IgG), DiI-LDL, DQ Red BSA and dextran Texas Red (10,000 M_r) were purchased from Molecular Probes (Eugene, OR). Ponasterone was obtained from Invitrogen, and Hygromycin B and FuGene 6 transfection reagent were purchased from Roche Molecular Biochemicals. All other reagents were from Sigma Chemical Co. (St Louis, MO) unless otherwise noted.

A rab5a-EGFP fusion was created by subcloning a rab5a wt cDNA fragment (Hoffenberg et al., 1995b) into pEGFP-C1 (Clontech, Palo Alto, CA). pEGFP-rab5a wt was mutagenized using the QuikChange™ site-directed mutagenesis kit (Stratagene, La Jolla, CA) with the primer 5′-GGATACAGCTGGCCTAGAACGATACCATAG-3′ to create the rab5aQ79L mutant and the primer 5′-GTCCGCTGTTGGTAAAAATAGCCTAGTGCTTC-3′ to create the rab5aS34N mutant. These EGFP-rab5a fusions were then subcloned into the vector pIND-Hygro (Invitrogen) for inducible expression in EcR293 cells, and into pcDNA3.1 (Invitrogen) for transient transfections.

The EcR293 line was derived from HEK293 and expresses the regulatory protein VgRXR, a chimeric steroid receptor that is activated by synthetic ecdysteriods such as ponasterone (No et al., 1996). The vector pIND-Hygro has a promoter element responsive to VgRXR, cloning sites for insertion of open reading frames and a hygromycin B resistance gene. pIND-Hygro/EGFP-rab5a wt, pIND-Hygro/EGFP-rab5aQ79L and pIND-Hygro/EGFP-rab5aS34N were transfected into EcR293 cells using FuGENE6, and clones resistant to hygromycin B (200 μg/ml) were screened for ponasterone-inducible fluorescence. For the induction of expression, cells were treated for up to 96 hours with 5 μM ponasterone, while control cells were treated with vehicle alone (0.125% ethanol). The fraction of cells expressing EGFP-rab5a in the presence of ponasterone was 90-95%.

Transfected cells were washed with PBS and then quickly dissolved in Laemmli sample buffer (Laemmli, 1970). Samples of 20 μg total protein were electrophoresed through SDS-PAGE and transferred to Immobilon-P membranes (Millipore, Bedford, MA). The membranes were probed with anti-rab5 monoclonal antibodies at a dilution of 1:500 and bound antibodies were detected by chemiluminescence (Pierce Chemical Co., Rockford, IL). Protein bands visualized on X-ray films were quantified by densitometry using SigmaGel 1.0 (SPSS Science, Chicago, IL).

For labeling with transferrin Texas Red, the cells were washed and incubated in serum-free medium for 30 minutes, pulsed with 20 μg/ml of the tracer for 5 minutes, then rapidly chilled and fixed. For pulse-chase analysis of fluid phase uptakes, dextran Texas Red (1 mg/ml) was added directly to the medium for 1 hour, then the cells were washed and incubated in complete medium for an additional 6 hours before washing and fixation. To assess lysosome to endosome traffic, cells were incubated with dextran Texas Red for 1 hour, then washed and chased for 6 hours, and finally incubated with ponasterone for 72 hours prior to fixation. For LDL uptakes, the cells were cultured in the presence of ponasterone or vehicle for 72 hours in medium containing 10% NuSerum (Life Technologies) in place of fetal bovine serum, then Di-LDL was added to 5 μg/ml for 5 minutes. The labeled medium was removed, the cells washed and then incubated in complete medium for a further 1 hour prior to fixation.

Cells growing on glass coverslips in 6-well clusters were washed in PBS with 1.2% sucrose (PBSS), fixed with 4% paraformaldehyde in PBSS at 4°C for 10 minutes, and then washed again with PBSS. The following steps were done at room temperature, with PBSS used for washes. The coverslips were incubated in 0.34% L-lysine, 0.05% Na-m-periodate for 20 minutes and permeabilized with 0.2% Triton X-100. The cells were then blocked with 10% normal goat serum (NGS) for 15 minutes. Primary antibodies, diluted in PBSS with 0.2% NGS and 0.05% Triton X-100, were added to the cells and left for 1 hour. The coverslips were washed four times before incubation with secondary antibodies using the same procedure as for the primary antibodies. We used the following concentrations of antibodies: anti-cathepsin D, 20 μg/ml; anti-LAMP-1, 5 μg/ml; anti-LAMP-2, 5 μg/ml; anti-CI-M6PR, 1:200; secondary antibodies, 1:100 dilution. The coverslips were mounted in Mowiol and viewed using a DeltaVision deconvolution microscopy system (Applied Precision Inc., Issaquah, WA) in the Baylor College of Medicine Integrated Microscopy Core. Optical sections of 150 nm (10-15 in number) were obtained and deconvolved, and then five sections through the cell center were combined to produce the images shown.

Sectioning and labeling of ultrathin frozen sections (50 nm) of EGFP-rab5aQ79L expressing cells using the technique of Tokuyasu were performed as described in detail elsewhere (Zimmer et al., 1998). Small specimens were cryoprotected by polyvinylpyrrolidone/sucrose, frozen in liquid nitrogen and sectioned with a cryoultramicrotome (LEICA EM Ultracut R FCS) at –100 to –110°C. Thawed sections were incubated at room temperature with the monoclonal antibody against GFP (Scompany, dilution of 1:10) for 45 minutes and goat anti-mouse IgG conjugated with 6 nm gold (Dianova, D-Hamburg, dilution of 1:10). Double-labeling was performed by incubating the sections with a polyclonal antibody against LAMP-2 (a gift of M. Fukuda, San Diego, dilution of 1:50) and goat anti-mouse IgG conjugated with 12 nm gold (Dianova, dilution of 1:50). After labeling, the grids were contrasted, embedded in 2% methylcellulose and examined in a Philips 400 electron microscope (Kassel, Germany).

Cells growing in 24-well clusters were treated with 5 μM ponasterone or vehicle alone (0.125% ethanol) for 72 hours. Isoproterenol (10 μM) was added to the cells in triplicate wells for varying times up to 20 minutes, then the wells were aspirated and washed with ice-cold serum-free medium containing 20 mM Hepes pH 7.4 (DMEM-H). Cell surface receptors were quantified by incubation at 4°C for 90 minutes with 6 nM [³H]CGP12177, a hydrophilic radioligand that selectively binds surface β₂ARs (Staehelin and Hertel, 1992). The cells were washed twice with cold DMEM-H, then lysed in the wells with 0.1% SDS, 0.1% NP-40 and the lysates counted by scintillation spectroscopy. Nonspecific binding was determined by incubations with 3 μM propranol, and was always less than 5%. The fraction of receptors left on the cell surface was plotted versus time of agonist exposure, and the curves fitted by nonlinear regression using the program GraphPad Prism (v. 3). The rate of approach to a steady-state level of surface and internal receptors is determined by the first-order rate constants for receptor endocytosis (k_e) and recycling (k_r). Unique values for these rate constants were estimated by curve-fitting to equation 4 in Morrison et al. (Morrison et al., 1996).

Stably transfected cells inducibly expressing EGFP-rab5aQ79L were grown in the presence of 5 μM ponasterone or vehicle alone for 72 hours. The cells were washed with PBS and then suspended in 1× SHE, which contained 0.25 M sucrose, 10 mM Hepes, pH 7.42, 2 mM EDTA, 1× Complete® proteinase inhibitor (Roche Biomolecular). The cell suspension was homogenized by repeated passage through a 25G needle, and then centrifuged at 500 g to obtain a post-nuclear supernatant. This was overlayed onto a solution of 25% Percoll in 1× SHE with 25 mM ATP and centrifuged at 36,000 g for 65 minutes in a 70.1 Ti rotor at 4°C. Immediately after collection of fractions, 25 μl aliquots were assayed for β-hexosaminidase activity by incubation with 0.3 mM methylumbelliferone in acetate buffer (100 mM sodium acetate and 0.1% Triton X-100) at 37°C for 1 hour in the dark. Trichloroacetic acid was added to a final concentration of 10% to stop the reactions. Samples were diluted 1:20 with 0.5 M glycine and 0.5 M sodium carbonate buffer, added in triplicate to white U-bottom 96-well trays, then read at an excitation of 365 nm and emission of 450 nm on a Dynatech Fluorolite 100.

The ecdysone inducible mammalian expression system was used to control the levels of EGFP-rab5a in EcR293 cells, as described in Materials and Methods. Immunoblot analysis of one EcR293 pIND-Hygro/EGFP-rab5aQ79L clone is shown in Fig. 1A. Incubation in the presence of 5 μM ponasterone for up to 72 hours induced expression of EGFP-rab5aQ79L to an average extent of 3.4-fold over endogenous rab5a, with no detectable expression in uninduced cells, as assessed by densitometry. Similar results were obtained with cells inducibly expressing EGFP-rab5a wt and EGFP-rab5a S34N (data not shown). Cells incubated with vehicle alone showed very little fluorescence of EGFP-rab5aQ79L (Fig. 1B, left), in agreement with the immunoblotting results. By contrast, cells incubated with 5 μM ponasterone for 72 hours showed a pronounced expression of EGFP-rab5aQ79L, as well as the characteristic dilated endosome morphology normally found with expression of this mutant rab protein (Fig. 1B, right). Within 5 minutes of adding transferrin Texas Red to the medium, both EGFP-rab5a wt (Fig. 2A) and EGFP-rab5aQ79L (Fig. 2B) containing endosomes were mostly positive for label. These results are in agreement with similar studies performed previously with BHK-21 and Hela cells (Stenmark et al., 1994).

We next labeled induced and control cells with antibodies against several well-characterized lysosome proteins. Lysosome-associated membrane proteins LAMP-1 and LAMP-2 reach lysosomes by first sorting from the trans-Golgi network to early endosomes, either directly (Hunziker and Geuze, 1996) or via the plasma membrane (Akasaki et al., 1996; Akasaki et al., 1995). In cells expressing EGFP-rab5aQ79L, both LAMP-1 and LAMP-2 localized extensively with EGFP-rab5aQ79L containing vesicles (Fig. 3, right). Interestingly, a considerable fraction of LAMP-1 and LAMP-2 appeared within dilated EGFP-rab5aQ79L-bounded vesicles, giving the appearance of multivesicular bodies (Fig. 3, right). Another lysosome protein, cathepsin D, is a lumenal enzyme that traffics to lysosomes as a ligand of the mannose 6-phosphate receptor (M6PR) (Kornfeld, 1992). Antibodies to cathepsin D also labeled the enlarged EGFP-rab5aQ79L-containing vesicles (Fig. 3, right), primarily labeling the lumen, with outer membrane labeling being much less apparent than for LAMP-1 or LAMP-2. Consistent with this finding, EGFP-rab5aQ79L-containing endosomes also labeled with DQ Red BSA, a fluid-phase marker whose fluorescence is dequenched in the presence of proteases (Reis et al., 1998) (data not shown). The colocalization of cathepsin D and EGFP-rab5aQ79L suggested that the distribution of M6PRs might also be effected. At steady-state, most CI-M6PR shows a perinuclear localization, and to some degree this is apparent in vehicle-treated cells (Fig. 3, left). In induced cells, the perinuclear distribution of CI-M6PR was less apparent and a fraction of receptors localized adjacent to or within EGFP-rab5aQ79L containing vesicles (Fig. 3, right).

Vesicles containing both EGFP-rab5aQ79L and LAMP-1 or LAMP-2 had a multivesicular appearance at the level of light microscopy (Fig. 3). To examine this more closely, cells expressing EGFP-rab5aQ79L were processed for immunogold electron microscopy using antibodies against EGFP and LAMP-2. A representative image (Fig. 4) shows a greatly enlarged endosome with labeling for EGFP (arrowheads) that is also positive for LAMP-2 (arrows). These endosomes contained numerous internal vesicles, characteristic of mutlivesicular bodies, or prelysosomal structures.

In order to determine whether the effect on lysosomal protein distribution of EGFP-rab5aQ79L can be attributed to its lack of GTPase activity, cells expressing EGFP-rab5a wt or the dominant negative mutant EGFP-rab5a S34N were labeled with antibody to LAMP-1. This lysosome protein did not localize with EGFP-rab5a wt (Fig. 5A), in agreement with a previously published finding (Bucci et al., 2000). The distribution of the dominant negative mutant EGFP-rab5aS34N was diffuse and to some degree localized to a perinuclear region; however, there was no apparent localization with LAMP-1 (Fig. 5B). This finding supports the view that the GTPase deficiency of EGFP-rab5aQ79L mutant is causing a mis-sorting of lysosome proteins.

Although expression of EGFP-rab5aQ79L is barely detectable in the stable inducible cell line, it is conceivable that this may be sufficient to cause adaptive changes in the transfected cells. To rule this out, HEK293 cells were transiently transfected for 48 hours with pcDNA3.1/EGFP-rab5aQ79L or pcDNA3.1/EGFP-rab5a wt, then labeled with antibody against LAMP-1 and imaged as described above. The distribution of LAMP-1 in transiently transfected cells was very similar to that observed in the inducible cell lines (data not shown).

The proteins so far examined traffic to lysosomes by various pathways that are believed to originate at the trans-Golgi network. To determine specifically whether traffic between the plasma membrane and lysosomes is affected by expression of EGFP-rab5aQ79L, cells were labeled with fluorescent endocytic tracers that are known to traffic from endosomes to lysosomes, then subjected to imaging. Dextran Texas Red was fed to cells for 1 hour, then washed and chased for 6 hours prior to fixation and imaging. In cells expressing EGFP-rab5a wt, dextran Texas Red was chased from endosomes into punctate vesicles that showed no overlap with EGFP-rab5a wt (Fig. 6A). By contrast, a considerable proportion of dextran Texas Red remained within the lumen of enlarged EGFP-rab5aQ79L containing vesicles after chase (Fig. 6B). Similar results were obtained using BSA Texas Red (not shown). To determine whether trapping of endocytic tracer within endosomes by expression of EGFP-rab5aQ79L was limited to fluid-phase proteins, DiI-LDL was added to cells for 5 minutes, then removed and chased for 1 hour. In cells expressing EGFP-rab5a wt, internalized label sorted from endosomes into discrete, punctate vesicles (Fig. 6C). However, in cells expressing EGFP-rab5aQ79L, the majority of Di-LDL remained within endosomes after a 1 hour chase (Fig. 6D).

Our results indicate that proteins targeted to lysosomes can accumulate in EGFP-rab5aQ79L-containing endosomes. One possible explanation is that in HEK293 cells, EGFP-rab5aQ79L causes an inhibition in recycling from endosomes to the plasma membrane, or affects some other generalized change in endocytosis or recycling. Studies using HeLa cells have shown that expression of rab5aQ79L has no appreciable effect on transferrin receptor endocytosis or recycling (Ceresa et al., 2001) despite the presence of greatly enlarged endosomes. In HEK293 cells, the endocytosis of cell-surface β₂-adrenergic receptors (β₂ARs) in transient co-transfections with rab5aQ79L appears unaffected, although the recycling rate was not measured (Seachrist et al., 2000). To determine whether EGFP-rab5aQ79L expression in our inducible HEK293 cells alters receptor endocytosis or recycling, the line was stably transfected to express human β₂ARs, then the rate of receptor internalization was measured in response to β-agonist treatment. In the presence of ponasterone, this cell line exhibited the same dilated endosomes as the parent line, and internalized β₂ARs localized to these endosomes together with lysosome markers (data not shown). The internalization of β₂ARs was then examined by measuring the loss of surface receptors as a function of time after adding β-agonist (Fig. 7). Internalization curves were modeled to derive first-order rate constants for both endocytosis (k_e) and recycling (k_r) (Morrison et al., 1996). During the relatively short time courses used to measure these rates, the predominant mode of β₂AR trafficking is between early endosomes and the plasma membrane (Moore et al., 1999a; Moore et al., 1999b). While EGFP-rab5aQ79L expression appeared to cause a slight inhibition in the rate of β₂AR endocytosis, there was no significant difference in recycling compared with control cells (Fig. 7 legend). The recycling rate of internalized receptors can also be determined directly by measuring receptor reappearance at the cell surface following the removal of agonist after steady-state receptor internalization (Morrison et al., 1996). The recycling rates determined this way were similar in both the presence and absence of ponasterone (data not shown).

In addition to altering the traffic of proteins from early endosomes to lysosomes, we wondered if movement of materials from lysosomes to endosomes might occur during expression of EGFP-rab5aQ79L. To test this possibility, cells were loaded with dextran Texas Red, washed and chased for 6 hours, then induced with ponasterone for 72 hours. Cells expressing EGFP-rab5a wt showed no overlap between EGFP and Texas Red (Fig. 8A), although the distribution of EGFP-rab5a wt was somewhat diffused, suggesting some cytoplasmic localization. It is possible that the prolonged culture of cells required for this protocol contributes to this apparent cytoplasmic localization. By contrast, cells expressing EGFP-rab5aQ79L showed large fluorescent vesicles containing Texas Red marker (Fig. 8B). Labeling of EGFP-rab5aQ79L-transfected cells with dextran Texas Red in the absence of inducing agent showed a punctate labeling pattern more typical of lysosomes (Fig. 8C).

The aberrant distribution of lysosome proteins in cells expressing EGFP-rab5aQ79L suggested that there might be a defect in the formation of dense lysosomes. Percoll gradient fractionation was used to assess the distribution of the lysosome enzyme β-hexosaminidase in EGFP-rab5aQ79L-expressing cells and in uninduced controls. In fractions from vehicle-treated cells, most of the β-hexosaminidase-specific activity was seen in a peak near the bottom of the gradient, while such a peak was absent or greatly reduced in cells expressing EGFP-rab5aQ79L (Fig. 9).

We present evidence that a GTPase defective rab5a interferes with normal lysosome biogenesis in mammalian cells. Expression of EGFP-rab5aQ79L caused the formation of large vesicles containing integral membrane lysosome proteins (LAMP-1 and LAMP-2), a lumenal protein that is normally a ligand of M6PRs (cathepsin D), and CI-M6PRs (Fig. 3; Fig. 4). Endocytic tracers remained within EGFP-rab5aQ79L-containing vesicles even after prolonged chase, in contrast to cells expressing EGFP-rab5a wt, where such tracers normally sort to separate compartments (Fig. 6). Moreover, material preloaded into lysosomes was found within endosomes after subsequent expression of EGFP-rab5aQ79L (Fig. 8) and there was a pronounced deficiency in β-hexosaminidase-containing dense lysosomes (Fig. 9). However, trafficking between the plasma membrane and early endosomes appeared to occur normally (Fig. 7). These results suggest that specific changes in the trafficking of proteins between endosomes and lysosomes are caused by expression of rab5aQ79L.

The remarkable redistribution of lysosome proteins to EGFP-rab5aQ79L-containing vesicles is reminiscent of what occurs when cells are treated with chloroquine to block endosome-lysosome transport (Lippincott-Schwartz and Fambrough, 1987). This similarity suggests that EGFP-rab5aQ79L inhibits such transport events, causing the accumulation within endosomes of lysosome proteins that normally traverse the endosome compartment before reaching lysosomes. This interpretation would be consistent with findings that expression of rab5aQ79L inhibits transport (of transferrin) from early endosomes to perinuclear recycling endosomes (Ullrich et al., 1996), inhibits degradation of LDL and epidermal growth factor (McCaffrey et al., 2001) and inhibits to some degree the degradation of ricin in MDCK cells (D’Arrigo et al., 1997). In that study, abnormalities in late endosomes/lysosomes was further suggested by the observation of enlarged rab7-positive vesicles in cells expressing myc-tagged rab5aQ79L, and a partial colocalization of these two proteins. We observed a similar morphology in HEK293 cells expressing GFP-rab7 and myc-tagged rab5aQ79L (data not shown). However, the inhibition of transport from endosomes to lysosomes per se may not be sufficient to explain our findings. Expression of a dominant-negative syntaxin-7 blocks traffic from early to late endosomes in NIH3T3 cells, yet does not appear to cause the accumulation of LAMP-2 or cathepsin D in early endosomes (Nakamura et al., 2000). Also, while our data suggest that the expression of EGFP-rab5aQ79L may inhibit trafficking from early endosomes to lysosomes, there also appears to be a stimulation of traffic in the opposite direction (Fig. 8). However, our results do not allow us to distinguish between a stimulation of direct fusion between lysosome and endosomes versus a stimulation of vesicular trafficking from lysosome to endosomes.

Recent evidence suggests that rab7 also plays an important role in lysosome biogenesis. Expression of GTPase-defective rab7 in HeLa cells increases the size of lysosomes and the extent of perinuclear aggregation, while expression of dominant negative rab7 causes dispersion of lysosomes throughout the cytoplasm. These aberrant organelles are not accessible to endocytic tracers; however, there is no apparent change in the early or late endosome compartments (Bucci et al., 2000). In BHK-21 cells, dominant negative rab7 expression causes an increase in the proportion of CI-MPR and cathepsin D in early endosome compartments. By contrast, lgp 120, a homologue of LAMP-1, shows a normal distribution under these conditions (Press et al., 1998). These results suggest that, in BHK-21 cells, CI-M6PR/cathepsin D is delivered to lysosomes through the early endosome compartment, whereas lgp 120 is delivered via late endosomes. The localization of LAMP-1 and LAMP-2 with EGFP-rab5aQ79L observed in our study suggests that, in contrast with BHK-21 cells, these lysosomal proteins traffic to lysosomes via early endosomes in HEK293 cells. Our results are consistent with what was found in rat hepatocytes, where significant fractions of both LAMP-1 and LAMP-2 are sorted to the plasma membrane and early endosomes before eventual transport to lysosomes (Akasaki et al., 1995).

In our experimental system, we do not detect significant changes in β₂AR recycling during EGFP-rab5aQ79L expression (Fig. 7), suggesting a relatively specific effect on trafficking of proteins between endosomes and lysosomes. Previous studies where rab5aQ79L caused changes in transferrin recycling and endocytosis employed acute 5-10-fold overexpression mediated by recombinant vaccinia viruses over a 4-5-hour period after infection (Stenmark et al., 1994), rather than the approximately threefold expression relative to endogenous rab5a after 48-72 hours reported here. Stenmark’s study also used HeLa or BHK-21 cells, so that differences in cell type could be significant. Further, subsequent studies of rab5aQ79L overexpression using recombinant adenoviruses failed to detect changes in the rates of transferrin recycling or endocytosis (Ceresa et al., 2001). Our findings suggest that alteration of intracellular sorting events may be a more significant consequence of inactivating the rab5a GTPase.

The accumulation of cathepsin D within enlarged rab5a-containing endosomes was recently observed in pyramidal neurons from patients with sporadic Alzheimer’s disease (Cataldo et al., 1997). These abnormal endosomes appeared to be positive for rabaptin 5 and EEA-1, proteins that bind rab5a-GTP specifically, suggesting enhanced rab5a activation (Cataldo et al., 2000). Since these changes occurred in preclinical stages of Alzheimer’s disease, inappropriate activation of rab5a may be proposed as an important factor in the pathogenesis of this disease. The enrichment of proteases within early endosomes could conceivably facilitate the proteolytic processing of amyloid precursor protein, and the secretion of amyloid fragments, perhaps by a form of recycling, could be especially rapid from this peripheral compartment.

We are grateful to E. Millman for assistance with the Percoll gradients. This work was supported by NIH grants R01HL50047 (B.J.K.) and K08HL03463 (R.H.M.). J.L.R. was supported by NIH grants F32HL10150 and T32HL07676.