Role of Rab9 GTPase in Facilitating Receptor Recruitment by TIP47

See allHide authors and affiliations

Science  18 May 2001:
Vol. 292, Issue 5520, pp. 1373-1376
DOI: 10.1126/science.1056791


Mannose 6-phosphate receptors (MPRs) deliver lysosomal hydrolases from the Golgi to endosomes and then return to the Golgi complex. TIP47 recognizes the cytoplasmic domains of MPRs and is required for endosome-to-Golgi transport. Here we show that TIP47 also bound directly to the Rab9 guanosine triphosphatase (GTPase) in its active, GTP-bound conformation. Moreover, Rab9 increased the affinity of TIP47 for its cargo. A functional Rab9 binding site was required for TIP47 stimulation of MPR transport in vivo. Thus, a cytosolic cargo selection device may be selectively recruited onto a specific organelle, and vesicle budding might be coupled to the presence of an active Rab GTPase.

Mannose 6-phosphate receptors (MPRs) deliver newly synthesized lysosomal hydrolases from the Golgi complex to prelysosomes and then return to the trans-Golgi network (TGN) to pick up more cargo (1, 2). Export of MPRs from the Golgi is mediated by the AP-1 adaptor complex, which binds MPR cytoplasmic domains and recruits these receptors into clathrin-coated vesicles. MPR transport from endosomes to the Golgi is mediated by a protein named TIP47 (tail-interacting protein of 47 kD) that binds to a different signal in MPR cytoplasmic domains and is required for their recycling to the Golgi both in vitro and in vivo (3). Finally, MPR endocytosis is mediated by the AP-2 clathrin adaptor at the plasma membrane (1, 2).

A fundamental question in cell biology is how a single receptor is recognized by different transport machineries depending on its intracellular location. One possibility is that organelle-specific proteins corecruit cytosolic, cargo-packaging proteins to a particular organelle. Because Rab proteins are located on distinct membrane-bound compartments, we tested the possibility that TIP47 was recruited selectively onto late endosomes because of the combined presence of both a specific Rab GTPase and MPR cytoplasmic domains.

The Rab9 GTPase is localized predominantly to late endosomes, and like TIP47, it is also required for MPR transport from endosomes to the Golgi complex (4, 5). We first tested whether TIP47 could interact with Rab9. Partially purified, cytosolic TIP47 (6) bound to His-tagged Rab9-GTP in preference to His-tagged Rab9-GDP (Fig. 1A). Binding was specific: addition of a twofold excess of Rab9 protein that was not tagged with His could compete for binding, whereas excess untagged Rab7 GTPase could not (8). Rab7 is the closest relative of Rab9 [57% identical (4, 9)] and functions in late endosome fusion (10), a process that does not appear to involve TIP47.

Figure 1

TIP47 binds Rab9-GTP specifically. (A) His-Rab9 wild-type (50 μg; “Rab9-GDP”) or His-Rab9 Q66L (50 μg; “Rab9-GTP”) was incubated in the presence of either 5 μM GDP or GTP, respectively, with highly enriched cytosolic TIP47 for 2 hours at room temperature in binding buffer (25 mM Tris pH 7.5, 150 mM NaCl, 5 mM MgCl2, 12 mM imidazole). Ni-NTA-agarose slurry (35 μl, Pharmacia) was added for 30 min; beads were collected, washed 3 times with 1.5 ml buffer + 0.1% CHAPS, eluted with 100 mM EDTA, and analyzed by SDS–polyacrylamide gel electrophoresis (12%). TIP47 was detected by immunoblot analysis. Where indicated, non-His-tagged, Rab9 wild-type (100 μg) or Rab7 (100 μg) proteins were included with His-Rab9-GTP in the reactions (+ XS Rab9, +XS Rab7). (B) TIP47 polypeptide recognizes Rab9-GTP. Rab9 protein was preincubated with 50 μM GDP or GTPγS for 80 min. Pure, recombinant His-TIP47 (100 nM) in 50 μM nucleotide, 10 μg/ml bovine serum albumin (BSA), was then added for 60 min at 37°C. Complexes were analyzed by immunoblot using antibodies against Rab9. (C) TIP47 residues 152 to 187 are necessary for Rab9 binding. Proteins were purified and binding determined as in (B); constructs ‘B’ and ‘C’ comprise residues 152 to 434 or 188 to 434, respectively, and are fully active in MPR binding. (D) TIP47 binding to endosome membranes is dependent on membrane-bound Rab9. Endosome-enriched membranes were pretreated with pure bovine brain GDI to remove Rab9 as described (18). Binding of recombinant TIP47 (50 nM) was then determined by incubation with 3 μg membranes, in 0.1 mg/ml BSA and 0.1 mM GTPγS for 5 min at 37°C. The amount of bound TIP47 or membrane-associated Rab9 was determined by immunoblot. The pairs of bars shown reflect reactions containing membranes pretreated with 0, 100, 120, or 160 μg/ml GDI for 5 min at 37°C.

To ensure that the binding of cytosolic TIP47 to Rab9 was direct, we repeated the binding experiments using recombinant TIP47 protein. Recombinant, His-tagged TIP47 bound directly to Rab9 and showed some preference for the GTP-bound form of the Rab9 GTPase (Fig. 1B). Indeed, binding studies using NH2-terminal truncations revealed that TIP47 residues 152 to 187 were important for Rab9 binding (Fig. 1C). These residues are distinct from those that make up the MPR cytoplasmic domain binding site near TIP47's COOH-terminus (3).

If membrane-associated, native Rab9 is important for TIP47-membrane association, TIP47 binding capacity should decrease if Rab9 is depleted from membranes. To deplete Rab9, we used Rab GDI, which binds prenylated Rabs under specific conditions and removes them from membranes (11). Significant amounts of TIP47 bound to endosome-enriched membranes (Fig. 1D). After a brief pretreatment with increasing amounts of purified Rab GDI, the level of Rab9 present on the membranes could be decreased significantly. Concomitantly, TIP47 binding capacity was also decreased (Fig. 1D). Under conditions where we could deplete membranes of all detectable Rab9, a fraction of TIP47 remained membrane-associated. It is likely that the remaining protein was associated with endosomes via its interaction with MPR cytoplasmic domains (3).

We next tested whether recombinant TIP47 could interact concomitantly with the cytoplasmic domain of the cation-independent (CI-) MPR and Rab9. Recombinant TIP47 was retained on glutathione-Sepharose by its direct interaction with the cytoplasmic domain of the CI-MPR [fused to glutathione-S-transferase (GST), Fig. 2A]. Rab9 does not bind the CI-MPR. However, if a ternary complex formed between the CI-MPR, TIP47, and Rab9, Rab9 would be retained on the glutathione-Sepharose resin in a TIP47-dependent manner. Indeed, Rab9 bound to immobilized CI-MPR only in the presence of TIP47 in a concentration-dependent manner (Fig. 2A). Thus TIP47 formed a ternary complex linking the CI-MPR and Rab9.

Figure 2

(A) TIP47 binds both CI-MPR and Rab9 in a ternary complex. GST-CI-MPR (100 nM) was mixed with TIP47 in binding buffer without imidazole plus 100 μM GTP, 10 μg/ml BSA; reactions also contained Rab9 (100 nM) in 300 μl. Binding was for 60 min at 37°C. Complexes were collected after addition of 40 μl glutathione-Sepharose (50% slurry) for 30 min at 20°C, 3 washes, and two elutions with 50 μl, 20 mM glutathione. The amount of bound Rab9 and TIP47 was determined by immunoblot followed by densitometric scanning. Blots in the lower panel were scanned to yield the upper panel curves. (B) Rab9 enhances TIP47:MPR interaction. Data from (A) were fit to a binding isotherm to obtain aK d estimate of 0.3 μM (filled circles). These data are compared with values obtained in the absence of Rab9 (averages of triplicates, open circles) using a fluorescence assay (7), which fit a curve corresponding toK d = 1 μM. (C) Fluorescent TIP47 binding assay. Reactions contained: Texas Red TIP47 (10 nM active protein), 1.5 μM Rab9 (where indicated), 100 μM GTP, and increasing amounts of GST-CI-MPR in 50 mM Hepes, 150 mM KCl, 1 mM MgCl2 10% glycerol, 0.6 mg/ml BSA (7). Reactions were at 37°C for 60 min; glutathione-Sepharose slurry was added for another 30 min at room temperature. Resin was washed with 1 ml buffer and eluted in 0.5 ml, 20 mM glutathione. Backgrounds were not subtracted and are accounted for in the curve fits.

Quantitative evaluation of this immunoblot data (Fig. 2B) revealed that TIP47 bound more tightly to the CI-MPR cytoplasmic domain when Rab9 was present. The data were fit to a curve corresponding to an apparentK d of 0.3 μM, a value that is one-third that obtained for TIP47:MPR interaction in the absence of Rab9 [1 μM (7)]. This difference was confirmed using an entirely different solution-binding assay using fluorescently labeled TIP47 (Fig. 2C). Thus, Rab9 enhanced the interaction of TIP47 with the CI-MPR at least threefold.

Conversely, the CI-MPR also enhanced the association of TIP47 and Rab9. As expected, in the absence of the CI-MPR cytoplasmic domain, Rab9 bound to His-tagged TIP47 (Fig. 3A). Moreover, the amount of Rab9 that was bound to TIP47 increased significantly upon addition of increasing amounts of the CI-MPR cytoplasmic domain (Fig. 3A). The presence of Rab9 in the eluted complexes was absolutely dependent on the presence of TIP47. Thus both Rab9 and the CI-MPR cytoplasmic domain enhanced each other's interaction with TIP47.

Figure 3

(A) CI-MPR enhances Rab9-TIP47 interactions. Reactions contained 100 nM His-TIP47, 100 nM Rab9, 10 μg/ml BSA, 100 μM GTP, and increasing amounts of GST-CI-MPR. Binding was for 60 min at 37°C; complexes were collected using Ni-NTA resin followed by immunoblot analysis. (B) TIP47 Ala167AlaAla mutant binds poorly to Rab9. Rab9 bound to purified, recombinant TIP47 wild-type (wt) or mutant proteins was determined by immunoblot as before. (C) TIP47, but not TIP47 Ala167AlaAla, stimulates MPR transport from endosomes to the TGN in living cells (12). Error bars represent the standard deviation of duplicate determinations obtained in this representative experiment.

Because TIP47 is likely to be involved in transport vesicle formation, it was possible that TIP47 recruited Rab9 into transport vesicles coupled with its conversion to an active, GTP-bound form. However, nucleotide dissociation, as monitored by the ability of Rab9 to bind GTPγ35S after GDP release, was unchanged in the presence of TIP47. Thus, cytosolic TIP47 does not appear to be a nucleotide exchange rate-enhancer for Rab9.

TIP47 function in vivo required TIP47's capacity to bind Rab9 efficiently. When TIP47 residues Ser167ValVal were mutated to Ala167AlaAla, the ability of recombinant TIP47 to bind Rab9 in vitro was decreased to about one-fifth (Fig. 3B). Cells transfected with wild-type TIP47, but not mutant TIP47, showed an increase in transport of MPRs from endosomes to the TGN (12,13); indeed, the mutant was inhibitory at comparable expression levels.

TIP47 represents a Rab9 effector that binds with preference to Rab9-GTP. Furthermore, the binding of Rab9 enhances the interaction of TIP47 with MPR cytoplasmic domains. Thus, Rab9 is likely to enhance the endosome-recruitment of TIP47 as well as the cargo capture process.

We have shown that Rab9 increases the affinity of TIP47 for the CI-MPR: the K d decreases from 1 μM to 300 nM. This modulation may be critical for TIP47 function. TIP47 binds to the CI-MPR with an affinity of 1 μM (7). Given its presence in cytosol at ∼300 nM, if TIP47 used only the MPR for membrane binding, ∼10% would be membrane-associated, but without any organelle specificity. Using Rab9 as an organelle-specific coreceptor, the K d for TIP47 binding to MPRs decreases to the cytosolic concentration of TIP47. In this scenario, 50% of TIP47 should be membrane-associated, and all of that should be associated with endosomes; this is close to the ratio we observe (3). Thus, an endosome-specific Rab protein triggers the organelle-specific recruitment of TIP47 that might otherwise bind MPRs when they are present in other compartments. In addition, the low intrinsic affinity of TIP47 for MPRs also decreases its binding to MPRs located in compartments where the receptors should interact with other cargo selection devices.

Because Rab9-GTP enhances TIP47 endosome association and cargo binding, it appears to function during transport vesicle formation. In addition, GTP-bearing Rabs recruit cytosolic docking factors to the surfaces of transport vesicles to facilitate SNARE pairing before fusion (14, 15). Thus Rabs are likely to function in both vesicle formation and docking (16,17).

Our data support a model in which Rab9 facilitates the recruitment of TIP47 onto MPR-containing endosomes, rather than onto other membrane-bound compartments that may also contain MPRs. Rab9 is likely to stimulate cargo capture by enhancing the affinity with which TIP47 binds to receptor cargo. After vesicle formation, Rab9 may also participate in the process by which transport vesicles dock with their specific, corresponding target.

  • * Present address: Eos Biotechnology Inc., South San Francisco, CA 94080.

  • To whom correspondence should be addressed. E-mail: pfeffer{at}


View Abstract

Navigate This Article