Association of Rab1B with GDP-dissociation inhibitor (GDI) is required for recycling but not initial membrane targeting of the Rab protein.

We have identified the Rab1B effector-domain mutant (D44N) that, when geranylgeranylated by Rab:geranylgeranyltransferase (GGTase II) in cell-free systems or intact cells, fails to form detectable complexes with GDP-dissociation inhibitors (GDIs). GDI-Rab complexes were collected on anti-FLAG affinity beads after incubating recombinant geranylgeranylated Rab1B with FLAG epitope-tagged GDI in vitro, or transiently coexpressing Myc-tagged Rab1B with FLAG-GDI-alpha or FLAG-GDI-2 in human embryonal kidney 293 cells. [3H]Mevalonate labeling and immunoprecipitation studies confirmed that the inability of Myc-Rab1BD44N to associate with GDI in vivo was not due to failure of the mutant to undergo geranylgeranylation. Immunofluorescence localization and immunoblot analysis of subcellular fractions indicated that expressed Myc-Rab1BD44N was efficiently delivered to intracellular membranes in 293 cells. This was confirmed when the fate of the prenylated pool of Rab1BD44N in 293 cells was traced by labeling the geranylgeranyl groups attached to the nascent protein with [3H]meval onate. However, in contrast to the prenylated Rab1BWT, which was distributed in both the membrane and soluble fractions, the prenylated Rab1BD44N was completely absent from the cytosol. Overexpression of Myc-Rab1BD44N did not impair ER --> Golgi glycoprotein trafficking in 293 cells, which was assessed by monitoring the Golgi-dependent processing of coexpressed beta-amyloid precursor protein. The current findings suggest that nascent prenylated Rab1B can be delivered to intracellular membranes in intact cells without forming a stable complex with GDI, but that recycling of prenylated Rab1B to the cytosolic compartment is absolutely dependent on GDI interaction.

Low molecular mass GTP-binding proteins encoded by the rab gene family play important roles in protein trafficking in mammalian cells. More than 30 Rab proteins have been identified and many of these have been localized to discrete vesicle populations or organelles (1,2). Individual Rab proteins have been implicated as essential mediators of vesicle targeting and/or fusion events in specific segments of the exocytic and endocytic pathways. For example, studies employing dominant-negative Rab mutants have demonstrated that Rab1A and Rab1B function in the transport of glycoproteins between the endoplasmic reticulum (ER) 1 and cis-Golgi (3)(4)(5), whereas Rab5 is required for the fusion of clathrin-coated endocytic vesicles with early endosomes (6,7). It is presently unclear how Rab proteins interact with the components of the vesicular transport machinery. However, the most widely held view is that geranylgeranylated Rab proteins initially associate with membranes and/or membrane-protein complexes of the budding donor vesicle where they are activated by exchange of GDP for GTP, facilitated by proteins(s) that stimulate guanine nucleotide dissociation (8 -10). In a manner yet to be defined, the activated Rab protein helps to ensure correct targeting of the transport vesicle. As the vesicle fuses with the appropriate acceptor compartment, GTP hydrolysis is presumed to occur, and the inactive Rab-GDP cycles off the membrane into a cytosolic pool from which it can be recruited into additional rounds of transport (reviewed in Refs. [11][12][13]. Mounting evidence suggests that the recycling of Rab-GDP involves a family of proteins termed GDP-dissociation inhibitors (GDI's), which form 1:1 complexes with geranylgeranylated GDP-bound Rab proteins, solubilize them from membranes, and slow the rate of nucleotide exchange (reviewed in Ref. 14).
The first Rab-GDI was isolated from bovine brain and was shown to associate with the GDP form of Rab3A and inhibit dissociation of bound nucleotide (15,16). Subsequent studies established that post-translational geranylgeranylation of the Rab protein was required for this interaction (17) and that bovine Rab3A-GDI can solubilize a variety of other Rab proteins from cellular membranes (18). The bovine Rab-GDI is now recognized as the prototype for a family of mammalian proteins encoded by at least three distinct genes. The ␣ GDIs include the original bovine Rab3A-GDI and the closely related rat GDI-␣ (19), human GDI-␣ (20), and mouse GDI-1 (21). The ␤ GDIs are comprised of the rat GDI-␤ (19), human GDI-␤ (19), and the recently identified mouse GDI-␤ (22). A third mouse GDI, designated GDI-2, has been identified (21). It shares 86 and 95% amino acid identity with mouse GDI-1 (21) and GDI-␤ (22), respectively, and thus appears to be a distinct gene product.
Geranylgeranylation of nascent Rab proteins is clearly essential for their entry into the complex cycle of functional associations with membranes and regulatory proteins. The modification of Rab proteins is catalyzed by Rab:geranylgeranyltransferase (also termed GGTase II) (23)(24)(25). In order for Rab proteins to serve as substrates for the catalytic ␣␤ sub-units of GGTase II, they must first associate with Rab Escort Proteins (REP-1 or REP-2) (26,27). In the absence of detergents, REP's remain associated with geranylgeranylated Rab proteins (RabGG) in vitro, suggesting that in addition to their role in the prenylation reaction, REP's may also serve to escort newly modified Rab proteins to downstream components of the vesicular transport machinery (26). Cytosolic GDIs have been proposed as potential intermediates that can accept RabGGs from REPs and deliver them to specific intracellular membranes (13, 14,26). This view is supported by studies showing that endogenous GDI in reticulocyte lysates can serve as an acceptor for geranylgeranylated Rab5 (28), and that pre-formed GDI-RabGG complexes can deliver functional geranylgeranylated Rab proteins to transport vesicles in cell-free systems and permeabilized cells (29 -31). However, the recent finding that REP itself can deliver prenylated Rabs directly to membranes in perforated cells (32) suggests an alternative possibility that GDI may not be required for the initial delivery of nascent RabGG to membranes in vivo.
In an effort to clarify the role of GDI in the initial membrane targeting of geranylgeranylated Rab proteins in intact cells, we have identified a Rab1B effector-domain mutant (Rab1B D44N ) that, when geranylgeranylated, fails to form detectable complexes with GDI in vitro and in transfected HEK 293 cells. Studies comparing the subcellular distribution of Rab1B D44N with that of Rab1B WT indicate that the initial association of the mutant with intracellular membranes is not significantly impaired. However, over time, the prenylated Rab1B D44N fails to accumulate in the cytosolic compartment. These findings suggest that stable association of nascent prenylated Rab1B with GDI is not required for its initial delivery to intracellular membranes, but that GDI is essential for returning the prenylated Rab protein to the cytosolic compartment after it enters the vesicular transport cycle.

EXPERIMENTAL PROCEDURES
Mutagenesis of Rab1B-Mutations were introduced into cDNA encoding Rab1B by means of overlap-extension PCR (33), using Taq DNA polymerase (Perkin Elmer) and pGEM3Zrab1B (34) as the template. The cDNA encoding Rab1B D44N was generated as described (35). The cDNAs encoding Rab1B WT and Rab1B D44N were modified by PCR so that the expressed proteins contained an amino-terminal Myc-epitope (EQKLISEEDL) and were subcloned into pCMV5neo (36,37) for expression in intact cells. Recombinant Rab1B proteins that could be purified by Ni 2ϩ chelation chromatography were obtained by modifying the rab1B cDNAs by PCR so that the expressed proteins contained a hexahistidine tag at their amino termini. The sequences of all rab1B constructs were verified by the dideoxy chain termination technique, using Sequenase 2.0 (U. S. Biochemical Corp.).
Epitope-tagging of GDIs-The cDNA encoding bovine GDI-␣ (16) was obtained by PCR from a cDNA template, reverse transcribed from bovine brain mRNA (Clonetech). The PCR product was cloned into pCRII (Invitrogen) and modified by addition of a 5Ј sequence encoding the FLAG epitope (DYKDDDDK). The FLAG-GDI-␣ construct was then subcloned into pCMV5neo to generate pCMVGDI␣. The cDNA encoding mouse GDI-2 (21) was obtained from Assia Shisheva (University of Massachusetts) and subcloned into pTrcHisB (Invitrogen), which had been modified to include a FLAG sequence immediately upstream of the EcoRI site. The FLAG-GDI-2 construct was then cloned into pCMV5.
Affinity Purification of FLAG-GDI-Recombinant FLAG-GDI-␣ was purified from bacterial lysate (1 ml) using an anti-FLAG M2 affinity gel column as described by the manufacturer (Eastman Kodak). The fraction containing the 55-kDa FLAG-GDI-␣ was dialyzed against Buffer A and stored at Ϫ80°C.
Cell-free Assay for the Formation of GDI-Rab Complexes-Aliquots of E. coli lysate containing equal amounts of Myc-tagged or His 6 -tagged Rab1B WT or Rab1B D44N (determined by immunoblot analysis) were incubated in Buffer A supplemented with 100 M GDP, 0.5 mM Nonidet P-40, 80 g/ml bovine serum albumin, 400 ng of affinity-purified FLAG-GDI, 200 ng of REP-1, 200 ng of GGTase II ␣␤, and 2 Ci of [ 3 H]GGPP (15 Ci/mmol, American Radiochemical Corp.). Recombinant REP and GGTase II ␣␤ were provided by Miguel Seabra (University of Texas Southwestern Medical Center). After a 1-h incubation at 37°C, SDS sample buffer was added to half of the reaction mixture and the extent of Rab1B prenylation was assessed by SDS-PAGE and fluorography. The other half of the reaction was added to 50 l of a 1:1 (v/v) slurry of anti-FLAG M2 affinity gel (Eastman Kodak Co.) which had been prewashed twice with 0.1 M glycine, pH 3, and 4 times with 1 ml of 10 mM NaPO 4 , 1 mM MgCl 2 , 200 M GDP, 0.15 M NaCl, pH 7.4 (phosphate wash buffer). Samples were mixed with the anti-FLAG resin on an end-over-end rotator for 30 min at 4°C and the resin was collected by centrifugation at 12,000 ϫ g for 1 min. The supernatant solution was removed and the resin was washed 3 times with 1 ml of the phosphate wash buffer. FLAG-GDI complexes were then eluted with 100 l of 0.1 M glycine, pH 3, and the eluates were neutralized by addition of 10 l of 1 M Tris-HCl, pH 8. The eluates were mixed with SDS sample buffer and subjected to SDS-PAGE on 12.5% polyacrylamide gels. Rab protein co-eluting with the FLAG-GDI was visualized as [ 3 H]GG-labeled protein on the fluorogram of the dried gel. To check for the presence of non-prenylated Rab1B in the FLAG-GDI complex, aliquots of eluates from the anti-FLAG resin were subjected to SDS-PAGE and a [␣-32 P]GTP blot overlay assay (40).
Metabolic Labeling Assay for Geranylgeranylation of Rab1B Expressed in Mammalian Cells-Transformed human embryonic kidney (HEK) 293 cells were grown in Dulbecco's modified Eagle's medium with 10% (v/v) fetal calf serum and plated at a density of 1.8 ϫ 10 4 cells/cm 2 on the day before transfection. All transfections were carried out using the calcium phosphate precipitation technique (41). Three hours after addition of the DNA, cultures were shocked by exposure to 15% (v/v) glycerol in phosphate-buffered saline for 30 s and were allowed to recover for 1 h in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. Cultures (6 cm) were transfected with 15 g of pCMVrab1B WT , pCMVrab1B D44N , or pCMVrab1B ⌬CC . Parallel cultures transfected with each Rab1B construct were incubated for either 12 or 24 h in 3 ml of medium containing 200 Ci/ml [ 3 H]mevalonolactone (3.4 Ci/mmol), beginning immediately after recovery from the glycerol shock. During the labeling period, 10 M lovastatin was added to the culture medium to block endogenous isoprenoid synthesis. Cells did not exhibit any change in morphology or viability during the incubation period, indicating that the radiolabeled mevalonolactone added to the medium (approximately 60 M) was sufficient to satisfy the cellular requirement for mevalonate (MVA). Labeled cells were harvested and washed three times with Hanks' balanced salt solution. Cells were lysed in 100 l of RIPA (100 mM Tris-HCl, pH 7.4, 2 mM EDTA, 0.1% (w/v) SDS, 0.5% (w/v) deoxycholate, 0.5% (v/v) Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 5 M leupeptin, 1 M pepstatin, 0.3 M aprotinin) and each lysate was diluted 1:5 by addition of 100 mM Tris-HCl, pH 7.4, 2 mM EDTA, with protease inhibitors. Insoluble material was removed by centrifugation at 12,000 ϫ g for 30 min. Cleared cell lysates were incubated for 2 h at 4°C with the 9E10 anti-Myc monoclonal antibody (Oncogene Sciences) at a concentration of 5 g/ml lysate, and the immune complexes were collected by addition of protein A-Sepharose coated with goat anti-mouse IgG (Cappel). The Sepharose beads were pelleted by centrifugation and washed once with RIPA, twice with 5 ϫ diluted RIPA, and once with 100 mM Tris-HCl, pH 7.4, 2 mM EDTA, 100 mM NaCl. Myc-tagged Rab1B proteins were eluted from the beads in SDS sample buffer and resolved by SDS-PAGE. Proteins were transferred to PVDF membrane and tritium incorporation into the Rab proteins was quantitated with a Molecular Dynamics phosphorimaging system and ImageQuant software. Each 3 H value (peak volume) was then normalized to the amount of immunoprecipitated Myc-Rab1B on the same blot. This was accomplished by incubating the membrane with antibody to Rab1B, followed by 125 I-labeled goat anti-rabbit IgG. Bound 125 I-IgG was then quantitated by PhosphorImager analysis. Before forming a ratio between the tritium and bound 125 I-IgG values, tritium values were corrected for any nonspecific radioactivity in the immunoprecipitates by subtracting corresponding values obtained for Myc-Rab1B ⌬CC in parallel [ 3 H]MVA-labeled cultures.
Coexpression Assay for Interaction of Rab1B with GDI in Intact Cells-HEK 293 cells growing in 10-cm dishes were transfected with DNA mixtures containing 15 g of pCMVFLAG-GDI (GDI␣ or GDI-2) combined with pCMVrab1B WT (15 g), pCMVrab1B ⌬CC (40 g), or pCMVrab1B D44N (15 g). At 48 h after transfection, cultures were harvested and homogenized in 0.5 ml of lysis buffer (100 mM Tris-HCl, pH 7.4, 0.1 mM GDP, 1 mM MgCl 2 , 50 mM sodium fluoride, 1 mM phenylmethylsulfonyl fluoride). Cytosol was obtained by centrifuging the lysates at 100,000 ϫ g for 30 min at 4°C. A 50-l aliquot of cytosol was reserved for quantitation of expressed GDI and Rab1B proteins by immunoblot analysis. The remaining cytosol was mixed with 100 l of a 50% suspension of anti-FLAG M2 affinity gel in phosphate wash buffer. Samples were incubated on a rotating rack for 30 min at 4°C and the affinity gel was collected by centrifugation and washed twice with phosphate wash buffer. Bound FLAG-GDI protein complexes were then eluted by suspending the gel in 100 l of 0.1 M glycine, pH 3. After removal of the gel by centrifugation, 10 l of 1 M Tris-HCl, pH 8.0, was added to the glycine supernatant solution and the samples were mixed with 25 l of 5 ϫ Laemmli sample buffer. Proteins were resolved by SDS-PAGE and transferred to PVDF membrane. The upper half of the membrane was immunoblotted with anti-FLAG antibody (Eastman Kodak) and 125 I-labeled goat anti-mouse IgG. The lower half of the membrane was immunoblotted with the antibody to Rab1B and 125 Ilabeled goat anti-rabbit IgG.
Subcellular Distribution of Immunodetectable Myc-Rab1B Expressed in 293 Cells-Myc-Rab1B proteins were coexpressed with FLAG-GDI-␣ as described above. Cells were harvested 48 h after transfection and homogenized in 0.5 ml of lysis buffer supplemented with 0.05% Nonidet P-40. Nuclei and unbroken cells were removed by centrifugation at 500 ϫ g for 5 min at 4°C. The low-speed supernatant fraction was removed and centrifuged at 100,000 ϫ g for 30 min at 4°C to obtain a crude high-speed membrane fraction. The latter was solubilized in 100 l of Laemmli SDS sample buffer (42). The high-speed supernatant was designated the cytosol and was mixed with 100 l of 5 ϫ SDS sample buffer. 50% of the membrane protein and 25% of the cytosolic protein were subjected to SDS-PAGE and transferred to PVDF membrane. Immunoblot analysis was performed with the Rab1B antibody and horseradish peroxidase-goat anti-rabbit IgG (Bio-Rad). Chemiluminescence detection was performed using the Amersham ECL kit.
Subcellular Distribution of Geranylgeranylated Myc-Rab1B in 293 Cells-Myc-tagged Rab1B proteins (WT, D44N, or ⌬CC) were transiently coexpressed with FLAG-GDI-␣ in 293 cells as described above. Each culture was incubated for 16 h in 3 ml of medium containing 200 Ci/ml [ 3 H]MVA and 10 M lovastatin to label the geranylgeranyl groups attached to newly synthesized proteins. Cells were then harvested, washed three times with Hank's balanced salt solution, and homogenized in 0.5 ml of lysis buffer without Nonidet P-40. Cells from three identically transfected 10-cm cultures were pooled for each determination. Subcellular fractionation was carried out as described above and cytosol was combined with 300 l of lysis buffer and 200 l of RIPA. The membranes were suspended in 200 l of RIPA and mixed with 800 l of lysis buffer. Myc-Rab1B proteins were immunoprecipitated from the cytosol and membrane fractions, subjected to SDS-PAGE, and transferred to PVDF. Tritium-labeled Myc-Rab1B proteins were quantitated by PhosphorImager analysis. To correct for any variations in Myc-Rab1B recovery in the immunoprecipitates, the same blot was incubated with antibody to Rab1B and each tritium value was normalized to the amount of immunodetectable Rab1B (bound 125 I-IgG). In a separate study, [ 3 H]MVA-labeled Myc-Rab1B proteins immunoprecipitated from the membrane and cytosol fractions were resolved by SDS-PAGE and visualized by fluorography.

Post-translational Processing of ␤-Amyloid Precursor Protein (␤APP)
in 293 Cells-Cells were cotransfected with 2 g of phCK751 (which encodes ␤APP 751 ) and pCMVrab1B WT (10 g), pCMVrab1B D44N (10 g), or pCMVrab1B N121I (30 g) as described previously (43). Sixteen hours after transfection, the cultures were pulse-labeled for 10 min at 37°C with 1 ml of methionine-free Dulbecco's modified Eagle's medium containing 100 Ci of [ 35 S]methionine/cysteine (Tran 35 S-label, 1100 -1200 Ci/mmol, ICN Inc.). Cells were washed twice with phosphate-buffered saline and subjected to a 1-h chase in medium containing 2 mM methionine, and 2 mM cysteine. Immunoprecipitation of ␤APP was then carried out as described by Dugan et al. (43). Immunoprecipitates were analyzed by SDS-PAGE and fluorography. Prior to immunoprecipitation of ␤APP, one-tenth of each lysate was retained for verification of Myc-Rab1B expression by immunoblot analysis.

Identification of a Rab1B Mutant That Fails to Associate
with GDI in Vitro-To explore the potential roles of GDI versus REP in the delivery of nascent prenylated Rab proteins to intracellular membranes, we wished to identify a Rab1B mutant that was capable of undergoing REP-dependent prenylation by GGTase II (and thus by inference was competent to associate with REP), but was incapable of forming a stable complex with GDI. To screen for such mutants, we developed a cell-free immunoprecipitation assay that measures the association of geranylgeranylated Rab proteins with FLAG epitopetagged bovine GDI␣ (Fig. 1). Recombinant Rab1B was geranylgeranylated in a reaction mixture containing GGTase II, REP, and [ 3 H]GGPP. Recombinant FLAG-GDI␣ was added as an acceptor for 3 H-labeled Rab1BGG generated during the reaction. Upon completion of the incubation, the FLAG-GDI was collected on anti-FLAG-agarose affinity beads and any associated 3 H-labeled Rab1BGG was detected by SDS-PAGE and fluorography. As shown in Fig. 1, when this assay was applied to Myc-Rab1B WT , a substantial portion of the geranylgeranylated Rab1B was collected with FLAG-GDI␣ on the affinity beads. Association of Rab1BGG with the affinity beads was specifically dependent on the presence of FLAG-GDI. When FLAG-GDI was omitted from the reaction mixture, neither prenylated Rab1B (detected by 3 H fluorography) nor nonprenylated Rab1B (detected by 32 P-GTP blot overlay) was collected on the anti-FLAG beads. Consistent with previously published observations, the association of Rab1B with FLAG-GDI␣ was entirely dependent on prenylation of the Rab protein. Thus, when the wrong prenyltransferase was added to the reaction (e.g. GGTase I instead of GGTase II), there was no incorporation of [ 3 H]GGPP into the Rab protein, and no Rab protein was detected in eluates of the anti-FLAG beads by [ 32 P]GTP overlay ( Fig. 1) or immunoblot assay (not shown).
Using this assay, we screened a number of Myc-Rab1B point mutants and found that one particular mutant with an amino acid substitution in the effector domain (D44N) met the criteria of being competent to undergo prenylation by GGTase II but incompetent to associate with GDI. Similar results were obtained when the His 6 tag was substituted for the Myc tag on the Rab substrates. Fig. 2 illustrates the results typically obtained when (His) 6 Rab1B D44N and (His) 6 Rab1B WT were compared with respect to their ability to undergo geranylgeranylation and associate with FLAG-GDI-␣ in the coprecipitation assay. As expected on the basis of our previous attempts to prenylate Rab1B D44N translation products in reticulocyte lysates (35), recombinant Rab1B D44N was prenylated less efficiently than Rab1B WT . Nevertheless, by incubating the protein for 1 h with purified REP and GGTase II ␣␤, sufficient amounts of 3 H-labeled geranylgeranylated Rab1B D44N were generated so that association of the prenylated protein with FLAG-GDI on the anti-FLAG beads should have been readily detected.
Rab1B D44N  Rab1B proteins were isolated from other endogenous prenylated proteins by immunoprecipitation with an antibody directed against the Myc epitope. To control for variations in recovery of the wild-type and mutant Rab1B proteins in separate cultures, the amount of [ 3 H]MVA incorporated into the immunoprecipitated protein was normalized to the total Myc-Rab1B protein recovered in each sample. This was done by first subjecting the immunoprecipitated Myc-tagged Rab1B to SDS-PAGE and electrophoretic transfer to PVDF membrane. The incorporation of [ 3 H]MVA into the Rab1B protein was then quantitated by radiometric scanning of the blot. Finally, the same blot was probed with an affinity purified antibody to Rab1B. By using an 125 I-labeled secondary antibody, we were able to quantitate the total immunodetectable Rab1B in each 3 H-labeled band.
As shown in Fig. 3A, immunoprecipitation of Myc-Rab1B D44N clearly revealed that the protein was prenylated in 293 cells. Parallel studies with cells expressing Myc-Rab1B ⌬CC , which cannot undergo geranylgeranylation, showed virtually no 3 (Fig. 3B)  nylgeranylated Rab1B D44N accumulates in transfected 293 cells to permit an analysis of the fate of the prenylated mutant with respect to its associations with GDI and cellular membranes.
Rab1B D44N Fails to Form Detectable Complexes with GDIs in Transfected Cells-Based on the results of cell-free assays (Fig.  2) we predicted that prenylated Rab1B D44N would be unable to associate with GDI in the cytosolic fraction of intact cells. To test this prediction, Myc-tagged versions of Rab1B WT , Rab1B D44N , or Rab1B ⌬CC were transiently coexpressed with FLAG-tagged GDI-␣ in 293 cells. The FLAG-GDI-␣ was subsequently collected from the cytosol on anti-FLAG affinity beads and the eluates were subjected to immunoblot analysis for the presence of Rab1B. As shown in Fig. 4, a prominent band corresponding to Myc-Rab1B (28 kDa) was readily detected in eluates from the anti-FLAG affinity beads collected from samples where Myc-Rab1B WT was coexpressed with FLAG-GDI-␣. No Rab1B signal was observed in eluates from samples where Myc-Rab1B WT was expressed without FLAG-GDI-␣, ruling out nonspecific association of the overexpressed Myc-Rab1B with the anti-FLAG beads. When FLAG-GDI-␣ was coexpressed with Myc-Rab1B ⌬CC , which lacks the carboxyl-terminal cysteine prenylation sites, no detectable Myc-Rab1B signal was observed at 28 kDa, confirming that the interaction of Rab1B with FLAG-GDI-␣ was dependent on geranylgeranylation of the Rab protein. Finally, in agreement with the cell-free studies described earlier, we were unable to detect any Myc-Rab1B signal associated with the FLAG-GDI-␣ complex collected from cells where Myc-Rab1B D44N was coexpressed with FLAG-GDI-␣ (Fig. 4). It should be noted that longer exposures of the blots revealed a minor 26-kDa band corresponding to endogenous (i.e. untagged) Rab1B in these samples. However, even at these long exposures, we did not see a band at 28 kDa corresponding to Myc-Rab1B D44N .
To determine whether the forgoing observations might be specific for the interaction between Rab1B D44N and the ␣-form of GDI, we performed a similar study wherein Myc-Rab1B D44N was coexpressed with FLAG-GDI-2. As shown in Fig. 5, the D44N mutant also failed to form a detectable complex with FLAG-GDI-2, suggesting that the change induced by the D44N substitution impairs the ability of Rab1B to associate with all forms of GDI.
Rab1B D44N Is Delivered to Intracellular Membranes-The preceding results suggest that if GDI plays an essential role in the delivery of nascent prenylated Rab proteins to intracellular membranes, there should be a depletion of Rab1B D44N in the membranes of cells expressing this mutant. To test this hypothesis, we performed immunofluorescence localization of the Myc-tagged Rab1B proteins expressed in 293 cells (Fig. 6). When cells were stained with the antibody to the Myc epitope, most of the Myc-Rab1B WT appeared to be concentrated in punctate perinuclear structures, consistent with the expected localization of Rab1B in ER and Golgi membranes. In contrast, cells expressing Myc-Rab1B ⌬CC exhibited a diffuse cytoplasmic staining pattern, reflecting the inability of this mutant to associate with membranes. The staining pattern observed for Myc-Rab1B D44N was not discernably different from that observed for Myc-Rab1B WT .
To obtain additional information about the localization of Rab1B D44N , we carried out immunoblot analyses of cytosol and membrane fractions obtained from 293 cells that were transiently expressing Myc-Rab1B WT , Myc-Rab1B ⌬CC , or Myc-Rab1B D44N (Fig. 7). The results confirmed that Myc-Rab1B ⌬CC was localized predominantly in the cytosol, whereas a significant portion of the Myc-Rab1B WT was localized in the membrane fraction. The absence of Myc-Rab1B ⌬CC in the highspeed membrane fraction is noteworthy because it suggests that potential aggregates of overexpressed non-prenylated Myc-Rab1B do not partition in this fraction. In agreement with results of the immunofluorescence analysis, the proportion of Myc-Rab1B D44N localized in the membrane fraction was not noticeably different from that observed in cells expressing the wild-type protein. Taken together, these observations suggest that the apparent inability of Rab1B D44N to form stable complexes with GDI does not prevent delivery of the mutant protein to intracellular membranes.
Geranylgeranylated Rab1B D44N Does Not Accumulate in the Cytosol-The conclusion that newly synthesized Rab1B D44N can be targeted to membranes without forming a GDI complex assumes that the behavior of this mutant with respect to the coexpressed FLAG-GDIs reflects a similar inability to form stable complexes with any endogenous GDIs that may be present in the 293 cell line. As discussed previously, GDI is thought to play a key role in returning geranylgeranylated Rab proteins to the cytosol after they have hydrolzyed GTP in conjunction with vesicle fusion with the appropriate acceptor compartment. Hence if Rab1B D44N is incapable of associating with all GDIs (both epitope-tagged and endogenous), one would predict that any geranylgeranylated Rab1B D44N that is initially delivered to intracellular membranes would be unable to recycle. Consequently, at steady state, the cytosol should be markedly depleted of prenylated Rab1B D44N . The immunoblot of total expressed Rab1B depicted in Fig. 7 shows that transfected 293 cells harbor a substantial cytosolic pool of Myc-Rab1B D44N . However, this type of analysis does not discriminate between geranylgeranylated Rab1B and non-prenylated (i.e. non-functional) protein that may accumulate in the cytosol if overex-pression of Rab1B overloads the REP/GGTase II machinery. To obtain a more precise indication of the membrane partitioning and recycling of the geranylgeranylated pool of Rab1B, we labeled the prenyl groups attached to nascent proteins by incubating 293 cells in medium containing [ 3 H]MVA for the first 16 h after transfection. The radiolabeled Myc-tagged Rab1B was then immunoprecipitated from the cytosol and membrane fractions and analyzed by SDS-PAGE and fluorography.
As shown in Fig. 8A, prenylated Myc-Rab1B WT and Myc-Rab1B D44N were both readily detected in the membrane fraction. However, in contrast to the wild-type protein, there was no [ 3 H]MVA-labeled Myc-Rab1B D44N in the cytosolic pool. In a separate experiment, the immunoprecipitated Rab proteins were transferred to PVDF membranes and tritium was quantitated by PhosphorImager analysis. The same blots were then probed with an antibody to Rab1B to estimate the relative amount of Rab1B collected in each of the immunoprecipitates. The results, which are depicted in Fig. 8B, confirmed that the lack of a tritium signal in the cytosolic D44N sample was not due to an absence of Myc-Rab1B protein in the immunoprecipitate. The results also indicated that the specific radioactivity of the membrane associated pools of Rab1B WT and Rab1B D44N (i.e. the ratio of 3 H to bound 125 I-IgG) were comparable (Fig.  8B), suggesting that if aggregated non-prenylated protein contributes to the total Rab1B detected in the membrane fraction, the proportion of such protein is not substantially greater in cells expressing the D44N mutant than it is in cells expressing wild-type Rab1B. Parenthetically, the results of this experi- ment also show that the relative amount of [ 3 H]MVA incorporated per unit of immunodetectable Myc-Rab1B WT was substantially higher in the membrane fraction than in cytosol. Thus, it is reasonable to conclude that much of the immunodetectable Myc-Rab1B in the cytosol from transfected 293 cells is not post-translationally modified. This reinforces the importance of the [ 3 H]MVA metabolic labeling studies, which can specifically track the fate of the geranylgeranylated pool of Rab1B.
Overexpression of Rab1B D44N Does Not Impair ER 3 Golgi Trafficking in 293 Cells-We used the post-translational maturation of the Alzheimer's ␤-amyloid precursor protein (␤APP 751 ) to assess the consequences of overexpression of Myc-Rab1B D44N on ER 3 Golgi trafficking of glycoproteins in 293 cells. ␤APP undergoes an increase in its apparent molecular mass on SDS gels after it is translocated from the ER to the Golgi apparatus and undergoes O-glycosylation (44,45). We have previously shown that when ␤APP 751 is transiently coexpressed with a Rab1B mutant with reduced affinity for guanine nucleotides (Myc-Rab1B N121I ), the Golgi-dependent post-translational maturation of ␤APP is blocked (43), consistent with the established ability of such Rab mutants to suppress the function of their endogenous counterparts (3,5). The pulse-chase studies depicted in Fig. 9 show that when Myc-Rab1B D44N was coexpressed with ␤APP 751 at a level comparable to that observed in cells expressing Myc-Rab1B WT , there was no detectable inhibition of ␤APP 751 maturation. In contrast, Myc-Rab1B N121I was a potent inhibitor of ␤APP 751 maturation, even when expressed at a much lower level than the WT and D44N proteins.

DISCUSSION
Recent studies have suggested two alternative models to describe how REP and GDI may function in the delivery of nascent geranylgeranylated Rab proteins to intracellular membranes and in the recycling of Rab proteins once they are engaged in the vesicular transport machinery. One model predicts that REP remains tightly associated with RabGG upon completion of the prenylation reaction and then escorts the modified Rab protein directly to the appropriate intracellular membrane compartment (32). According to this model, the primary function of GDI would be to solubilize and recycle GDP-bound prenylated Rab proteins after they have begun to function in vesicular transport. An alternative view originates from the finding that GDI can serve as an acceptor of prenylated Rab proteins from the REP-GGTase II enzyme complex in cell-free systems (28) (also see Figs. 1 and 2). It envisions that GDI not only functions in Rab recycling, but also mediates initial membrane targeting of nascent Rab proteins by serving as an obligatory intermediate between the translation/prenylation machinery and the acceptor membrane. In the present study we have attempted to discriminate between these possibilities by first identifying a Rab1B mutant (Rab1B D44N ) that undergoes REP-dependent geranylgeranylation but does not form a complex with GDI, and then examining the fate of this protein in transfected 293 cells. Our results support a model in which delivery of newly synthesized and geranylgeranylated Rab1B D44N to intracellular membranes can be accomplished without the formation of a stable Rab-GDI complex, implying that this function is performed by REP or other unidentified accessory proteins (Fig. 10). However, our metabolic labeling studies also show that without the capability to associate with GDI, prenylated Rab1B D44N fails to accumulate in the cytosolic compartment. This provides a direct in vivo confirmation of the proposed role of GDI in returning prenylated Rab proteins to the cytosol during the vesicular transport cycle (Fig. 10).
Geranylgeranylation of Rab1B D44N -We previously reported that Rab1B proteins bearing mutations in the effector domain are not efficiently geranylgeranylated when translated in rabbit reticulocyte lysates (35,46). However, in the present study we found that recombinant (His) 6 Rab1B D44N incubated with REP and GGTase II ␣␤ in vitro could be geranylgeranylated to levels approximating 40 -50% of those attained in similar reactions with Rab1B WT (see Fig. 2). Similarly, when Myc-Rab1B D44N was transiently overexpressed in 293 cells, the incorporation of [ 3 H]MVA per unit of immunodetectable protein was 60 -70% of that observed for the wild-type protein (Fig. 3). One possible explanation for these different findings is Our results with Rab1B D44N indicate that: 1) nascent Rab proteins are geranylgeranylated and then delivered directly to intracellular membranes without the obligatory formation of a Rab-GDI complex. The properties of REP suggest that it may perform this function. The prenylated Rab protein then participates in a round of vesicular transport, with GTP hydrolysis presumably occurring at the end point of the cycle (vesicle fusion with the acceptor membrane). Our results indicate, that 2) the eventual return of the geranylgeranylated GDP-bound Rab protein to the cytosol is absolutely dependent on its ability to associate with GDI.
that Rab1B D44N has an altered affinity for either REP or the GGTase II catalytic components. Thus, when relatively high concentrations of substrate protein or enzyme components are combined in vitro, or when Rab1B D44N is overexpressed for 12-24 h in transfected cells, it is possible to obtain significant accumulation of geranylgeranylated protein. While a detailed examination of this issue is beyond the scope of the present study, two important inferences can be drawn: first, since prenylation of Rab proteins by GGTase II is completely dependent on the formation of a Rab-REP complex (24,26,27), the reasonably efficient prenylation of Rab1B D44N compared to Rab1B WT in transfected 293 cells implies that the effectordomain mutant is capable of productive interaction with REP in vivo. Second, since prenylation of Rab proteins is sensitive to conformational perturbations (35,(47)(48)(49) and requires occupancy of the guanine nucleotide binding site (50), it is highly probable that the pool of Myc-Rab1B D44N that has undergone geranylgeranylation is not grossly misfolded or aggregated.
Inability of Rab1B D44N to Associate with GDI-Although Myc-Rab1B D44N can undergo geranylgeranylation, thereby meeting one of the major criteria for interaction of Rab proteins with GDI (17,51,52), we were unable to demonstrate the formation of stable complexes between this mutant and epitope-tagged GDI-␣ or GDI-2 in cell-free systems and intact cells. It is important to emphasize that the coprecipitation assay used in these studies readily detects complexes formed between both forms of GDI and Myc-Rab1B WT (Figs. 1, 4, and 5), as well as a variety of other epitope-tagged Rab1B point mutants. 2 Thus, the disruption of Rab1B interaction with GDI appears to be very specifically related to the introduction of the D44N substitution in the predicted ␤2/loop-2 junction of the effector domain. This finding agrees with a previous study of Rab6, in which the exchange of an 11-amino acid segment of the effector domain of Rab6 with the corresponding region from H-Ras rendered Rab6 incapable of being solubilized from membranes by recombinant GDI (48).
The molecular basis for the disruption of GDI interaction with geranylgeranylated Rab1B D44N is presently unknown. Although incomplete prenylation of the mutant (i.e., incorporation of only one geranylgeranyl group) is possible, this probably would not explain the absence of GDI-Rab1B D44N complexes, since previous studies have shown that mono-geranylgeranylated Rabs can be incorporated into GDI complexes (18,52). Another possibility is that the D44N alteration prevents nucleotide binding. However, several lines of evidence suggest that this is unlikely: first, studies of Rab3A (53) and Ypt1 (the yeast homolog of Rab1) (54) have shown that substitutions at the position equivalent to Asp 44 do not significantly affect nucleotide binding. Second, in blot overlay assays we have found that recombinant Myc-Rab1B D44N binds [␣-32 P]GTP to the same extent as Myc-Rab1B WT . 2 Third, as mentioned above, prenylation of Rab proteins by REP/GGTase II requires that the Rab substrate contain bound nucleotide. Thus, the effective prenylation of Myc-Rab D44N in intact 293 cells (Fig. 3) implies that guanine nucleotide binding is not significantly compromised. It remains possible that Rab1B D44N is able to bind GTP but, because of altered GTPase activity, is unable to assume the GDP state that interacts preferentially with GDI. Indeed, Becker et al. (54) have shown that Ypt1 D44N is not stimulated by GAP activity to the same extent as the wild-type protein.
Although we cannot completely discount this possibility, it would be difficult to reconcile this notion with the fact that recombinant Myc-Rab1B D44N fails to associate with GDI even when prepared and prenylated in buffers containing exclu-sively GDP (Fig. 2). Moreover, we have found that when [␣-32 P]GTP is bound to Myc-Rab1B D44N on nitrocellulose membranes, the rate at which the mutant protein converts the bound GTP to GDP is essentially the same as for Myc-Rab1B WT . 2 We favor the hypothesis that the ␤2/loop-2 domain represents a site of direct interaction between Rab1B and GDI, and that conformational alterations conferred by the D44N substitution directly interfere with association of the two proteins.
Delivery of Newly Prenylated Rab Proteins to Intracellular Membranes-The present studies show that when recombinant GDI is added to a cell-free prenylation reaction containing REP and GGTase II, the GDI is capable of serving as a soluble acceptor for nascent Rab1BGG (Figs. 1 and 2). This observation is consistent with a recent report describing the transfer of newly prenylated Rab5 to GDI in reticulocyte lysate (28). However, our studies with the D44N mutant strongly suggest that transfer of nascent Rab1BGG to GDI is not an obligatory intermediate step for delivery of the newly prenylated protein to intracellular membranes in intact cells. Three separate experimental approaches provided evidence that the delivery of Myc-Rab1B D44N to membranes was not impaired by the inability of this mutant to form stable complexes with GDI in 293 cells. First, the pattern of immunofluorescence staining of Myc-Rab1B D44N was not substantially different from that of Myc-Rab1B WT , whereas the same technique easily revealed the failure of Myc-Rab1B ⌬CC to associate with intracellular membranes (Fig. 6). Second, comparable amounts of Myc-Rab1B D44N and Myc-Rab1B WT were detected in membrane fractions by immunoblot analysis, whereas the same fractions were devoid of overexpressed Myc-Rab1B ⌬CC (Fig. 7). Third, using a metabolic labeling approach that specifically traces the fate of the geranylgeranylated pool of Rab1B, we found that comparable amounts of prenylated (i.e. [ 3 H]MVA-labeled) Myc-Rab1B D44N and Myc-Rab1B WT could be immunoprecipitated from membranes obtained from cells expressing these proteins (Fig. 8). The general immunofluorescence staining pattern for Myc-Rab1B D44N was consistent with ER/Golgi localization and gave no indication that the mutant was grossly mistargeted (e.g. to peripheral vesicles or plasma membrane). However, the inability of immunofluorescence light microscopy and simple centrifugation steps to resolve ER membranes, transitional vesicles, and Golgi subcompartments leaves open the possibility that the precise membrane localization of Myc-Rab1B D44N was different from that of Myc-Rab1B WT . Indeed, as discussed below, one might predict that the failure of Rab1B D44N to undergo GDI-mediated recycling at the end of each round of ER 3 Golgi transport might result in its disproportionate accumulation in the acceptor compartment (i.e. the cis-Golgi).
Role of GDI in Recycling of Prenylated Rab Proteins-Our studies of the subcellular distribution of geranylgeranylated Rab1B D44N in 293 cells (Fig. 8) highlight the key role of GDI in maintaining a steady-state cytosolic pool of prenylated Rab proteins. However, insofar as the same studies also suggest that GDI may not function as an obligatory soluble intermediate in the delivery of newly prenylated Rab proteins to intracellular membranes, they raise a perplexing question for future investigation. That is, if the GDIs are able to accept prenylated Rab proteins from REP in cell-free systems, how might REP be able to deliver newly prenylated Rab proteins to membrane acceptors in vivo without competitive interference from the relatively abundant cytosolic pool of GDI? One possibility is that nascent Rab proteins associated with REP in intact cells are in the GTP-bound state and thus are not in the correct conformation for recognition by GDI. This seems unlikely in light of studies suggesting that GGTase II preferentially mod- ifies Rab proteins that are in the GDP-bound state (50). Another possibility is that upon completion of the prenylation reaction the REP-RabGG complex associates with as yet unidentified accessory proteins that participate in site-specific membrane targeting of individual Rab proteins, while simultaneously masking RabGG from GDI. Finally, it is conceivable that not all of the GDI in intact cells is in an active state that is competent to bind prenylated Rab proteins, so that the potential for GDI to compete with REP for binding of nascent RabGG is limited. In this regard, reports of phosphorylation (55) and isoelectric shifts of GDIs (56) are intriguing, since they suggest that post-translational modifications may be involved in regulating the activity of the GDI pool in vivo.
ER 3 Golgi Transport in Cells Overexpressing Myc-Rab1B D44N -We have been unable to detect any perturbation of glycoprotein trafficking in 293 cells overexpressing Myc-Rab1B D44N (Fig. 9), using an established pulse-chase assay that measures the Golgi-dependent processing of coexpressed ␤APP 751 in this cell line (43). These findings agree with those of Tisdale et al. (3), who observed that Rab1B D44N had no effect on vesicular stomatits virus-G protein processing in HeLa cells. Since these assays readily reveal trans-dominant suppression of ER 3 Golgi transport by Rab1B mutants that have reduced affinity for guanine nucleotides (e.g. N121I) or are predominantly in the GDP state (e.g. S22N) (3, 43) (see Fig. 9), the ␤APP processing studies conducted with Rab1B D44N further support the assumption that the D44N mutation does not markedly impair the ability of Rab1B to bind GTP. In light of these findings, we speculate that geranylgeranylated Myc-Rab1B D44N delivered to intracellular membranes is competent to support a single round of vesicular transport between the ER and Golgi complex. The inability of the D44N mutant to undergo GDI-mediated recycling might therefore have little overall impact on ER 3 Golgi transport in transfected cells that are continuously delivering newly prenylated Rab1B D44N to the donor compartment (i.e. the ER). If Rab1B D44N does in fact accumulate in the acceptor compartment (i.e. the Golgi complex) as a result of its inability to associate with GDI, this apparently does not interfere with subsequent rounds of vesicle fusion as might be expected if the mutant were to irreversibly occupy key components of the docking/fusion machinery. Further insight into this issue should be forthcoming as we learn more about the precise subcellular localization and properties of Rab1B proteins with alterations in the ␤2/loop-2 region.