Report

Sorting of Mannose 6-Phosphate Receptors Mediated by the GGAs

See allHide authors and affiliations

Science  01 Jun 2001:
Vol. 292, Issue 5522, pp. 1712-1716
DOI: 10.1126/science.1060750

Abstract

The delivery of soluble hydrolases to lysosomes is mediated by the cation-independent and cation-dependent mannose 6-phosphate receptors. The cytosolic tails of both receptors contain acidic-cluster-dileucine signals that direct sorting from the trans-Golgi network to the endosomal-lysosomal system. We found that these signals bind to the VHS domain of the Golgi-localized, γ-ear–containing, ARF-binding proteins (GGAs). The receptors and the GGAs left the trans-Golgi network on the same tubulo-vesicular carriers. A dominant-negative GGA mutant blocked exit of the receptors from the trans-Golgi network. Thus, the GGAs appear to mediate sorting of the mannose 6-phosphate receptors at the trans-Golgi network.

Lysosomal hydrolases are posttranslationally modified in the Golgi complex by the addition of mannose 6-phosphate groups that function as signals for sorting to lysosomes (1). The mannose 6-phosphate groups are recognized in the trans-Golgi network (TGN) by a cation-independent mannose 6-phosphate receptor (CI-MPR) or a cation-dependent mannose 6-phosphate receptor (CD-MPR). Both mannose 6-phosphate receptors (MPRs) mediate recruitment of the lysosomal hydrolases to clathrin-coated areas of the TGN, from which carrier vesicles deliver the MPR-hydrolase complexes to endosomes. The acidic pH of endosomes induces release of the hydrolases from the MPRs, after which the hydrolases are transported to lysosomes while the MPRs return to the TGN for additional rounds of sorting.

Sorting of both MPRs from the TGN to endosomes is mediated by signals present in the cytosolic tails of the receptors (Fig. 1A). These signals consist of a cluster of acidic amino acid residues followed by two leucine residues (2, 3). Early studies suggested that sorting of MPRs at the TGN was mediated by the clathrin-associated adaptor protein (AP) complex, AP-1 (4, 5). However, AP-1 does not appear to bind the acidic-cluster-dileucine signals from the MPRs (6, 7). In addition, a recent study has shown that disruption of AP-1 impairs retrograde transport of the MPRs from endosomes to the TGN (8). Thus, clathrin-associated proteins other than AP-1 might be responsible for the signal-mediated sorting of MPRs at the TGN. Prime candidates for this role are theGolgi-localized, γ -ear–containing,ARF-binding proteins (GGAs) (9–12). Three GGAs have been identified in humans (GGA1, GGA2, and GGA3) and two in yeast (Gga1p and Gga2p). These proteins are monomeric and display a modular structure consisting of a VHS (VPS27, Hrs, and STAM) domain of unknown function, a GAT domain that interacts with the guanosine 5′-triphosphate–bound form of ADP-ribosylation factors (ARF) (9, 10, 13), a hinge domain that interacts with clathrin (13), and a GAE domain that interacts with γ-synergin and other potential regulators of coat assembly (11) (Fig. 1B). Disruption of the two yeast genes encoding GGAs results in impaired sorting of pro-carboxypeptidase Y to the vacuole, the equivalent of the mammalian lysosome (10,11). These characteristics of the GGAs prompted us to test whether they could bind the cytosolic tails of the MPRs.

Figure 1

Analysis of interactions between MPR cytosolic tail sequences and the GGAs with the yeast two-hybrid system. (A) COOH-terminal sequences of the CI-MPR and CD-MPR cytosolic tails. Acidic-cluster-dileucine signals are underlined. (B) Domain organization of the GGAs. (C to G). Yeast cells were cotransformed with plasmids encoding GAL4bd fused to the cytosolic tail sequences or control proteins indicated on the left with GAL4ad fused to the proteins or protein domains indicated above each panel. The MPR constructs contained the terminal 113 of the 154–amino acid cytosolic tail of the CI-MPR and the terminal 41 of the 69–amino acid CD-MPR tail. Mouse p53, SV40 large T antigen (T Ag), μ2, and ARF1 Q71L were used as controls. Mutated residues are underlined. Cotransformed cells were spotted onto histidine-deficient (−His) or histidine-containing (+His) plates and incubated at 30°C as described (14). Growth on −His plates is indicative of interactions. TYR: tyrosinase; LDLR: LDL receptor; APP: β-amyloid precursor; TfR: transferrin receptor (19).

Analyses with the yeast two-hybrid system showed that the cytosolic tail of the human CI-MPR interacted with full-length GGA3 as well as with the ∼150–amino acid VHS domain, but not the GAT or GAE domains of GGA3 (Fig. 1C). Further analyses revealed that the CI-MPR tail interacted with the VHS domains of all three GGAs and that the CD-MPR tail interacted with the VHS domains of GGA1 and, more weakly, GGA3 (Fig. 1D). These interactions were highly specific, because neither MPR tail bound to the VHS domains of STAM1, HRS, TOM1, and TOM1L1 (Fig. 1D). Likewise, we did not detect interactions between the GGA VHS domains and the cytosolic tails of TGN38, TRP2, tyrosinase, TRP1, LAMP-2, LIMP II, low density lipoprotein (LDL) receptor, β-amyloid precursor, and transferrin receptor (Fig. 1D), which contain either tyrosine-based sorting signals or dileucine-based sorting signals devoid of acidic clusters.

Deletion of the acidic-cluster-dileucine signals from the CI-MPR and CD-MPR tails abolished interactions with GGA VHS domains (Fig. 1E). Conversely, placement of the acidic-cluster-dileucine signal of the CI-MPR at the end of a 23–amino acid segment of the TGN38 cytosolic tail (14) conferred binding to the GGA VHS domains (Fig. 1E). The acidic-cluster-dileucine signal was thus both necessary and sufficient for interaction with the GGA VHS domains. Mutational analyses of the signals showed that the two leucines, as well as some of the acidic residues, were required for interactions (Fig. 1, F and G).

To corroborate the results of the yeast two-hybrid assays, we performed in vitro binding experiments. Glutathione-S-transferase (GST) pull-down assays showed specific binding of the recombinant VHS domain of GGA1 to the cytosolic tails of both MPRs (Fig. 2A). These interactions were abolished by deletion of the signals (Fig. 2A). Analysis by surface plasmon resonance spectroscopy confirmed the interaction of the recombinant GGA1 VHS domain with a biotinylated peptide comprising the acidic-cluster-dileucine signal of the CI-MPR (Fig. 2B). This interaction was completely abrogated by substitution of the two leucines with alanines (Fig. 2B). In agreement with other studies (6, 7), a mixture of AP-1 and AP-2 purified from bovine brain did not show appreciable binding to the acidic-cluster-dileucine signal (Fig. 2C). Binding of various concentrations of GGA1-VHS to the CI-MPR tail peptide (Fig. 2D) allowed an estimation of the equilibrium dissociation constant at ∼1 μM.

Figure 2

Interactions of MPR signal sequences with the VHS domain of GGA1 in vitro. (A) Recombinant His6-GGA1-VHS (25, 50, and 100 μg/ml) was tested for interactions with GST or GST fusion proteins bearing the tails of the CI-MPR or CD-MPR, or these tails with deletion of the acidic-cluster-dileucine signals shown in Fig. 1A (GST-CI-MPRΔ constructs). The GST fusion proteins were visualized by Coomassie blue staining, and bound His6-GGA1-VHS was detected by immunoblotting with antibody to His6 (20). (B) His6-GGA1-VHS (6 μg/ml, ∼0.3 μM) was injected (arrow 1) onto streptavidin-coated sensor chips loaded with N-biotinylated NKSSFHDDSDEDLLHI or NKSSFHDDSDEDAAHI peptides derived from the CI-MPR tail or with no peptide (blank). Dissociation of the protein was induced by injection of buffer containing 20 μM biotinylated NKSSFHDDSDEDLLHI to prevent rebinding (arrow 2). Flow rate: 20 μl/min. (C) His6-GGA1-VHS (6 μg/ml) or a ∼1:7 mixture of AP-1 and AP-2 (80 μg/ml) purified by gel filtration from extracts of bovine brain clathrin-coated vesicles (APs) were injected onto streptavidin-coated sensor chips loaded with N-biotinylated NKSSFHDDSDEDLLHI peptide. The blank corresponded to His6-GGA1-VHS injected onto sensor chips without peptide. Flow rate: 20 μl/min. (D) Various concentrations of His6-GGA1-VHS as indicated were injected onto streptavidin-coated sensor chips loaded with biotinylated NKSSFHDDSDEDLLHI peptide. The apparent equilibrium dissociation constant was calculated by nonlinear least-squares fitting of the data, assuming a single-site interaction model. Flow rate: 2 μl/min (21).

Immunofluorescence microscopy analyses showed good colocalization of endogenous CI-MPR (Fig. 3, A to C), as well as CD-MPR tagged at the COOH-terminus with the green fluorescent protein (GFP) (Fig. 3, D to F), with stably expressed MYC-GGA1 at the TGN and peripheral vesicles of MDCK cells. Time-lapse, confocal imaging of live MDCK cells expressing GGA1 tagged with the yellow fluorescent protein (YFP) revealed budding of tubulo-vesicular structures containing GGA1 from the TGN [Fig. 3G and video 1 (15)]. After detaching from the TGN, these structures migrated toward the peripheral cytoplasm at speeds of 1 to 4 μm/s [Fig. 3G and video 1 (15)]. Coexpressed CD-MPR tagged at the COOH-terminus with the cyan fluorescent protein (CFP) was present in the GGA1-containing tubulo-vesicular structures (Fig. 3, H and I), suggesting that these corresponded to intermediates that carry MPRs from the TGN.

Figure 3

Immunofluorescence microscopy analysis of GGA1 and MPR localization. (A to C) MDCK cells stably expressing MYC-GGA1 were stained with mouse monoclonal antibody (mAb) to the CI-MPR (MA1-066; Affinity BioReagents, Golden, CO) (A) and rabbit polyclonal antibody to the MYC epitope (BabCo, Richmond, CA) (B) followed by Alexa 488 anti-mouse immunoglobulin G (IgG) and Cy3 anti-rabbit IgG. (D to F) MDCK cells stably expressing MYC-GGA1 were transiently transfected with a construct encoding CD-MPR-GFP (D) and stained with mouse mAb to the MYC epitope (9E10; BabCo, Berkeley, CA) (E) and Cy3 anti-mouse IgG (22). (G) Time-lapse microscopy imaging sequence of live MDCK cells expressing YFP-GGA1. Images were acquired on a TILL Photonics microscope at the intervals indicated. Arrows track individual vesicles budding from the TGN. N: nucleus. For more details, see video 1 (15). (H and I) Confocal microscope imaging sequences of live MDCK cells coexpressing YFP-GGA1 and CD-MPR-CFP. Images were acquired at the intervals indicated. Arrows point to tubulo-vesicular structures containing both fluorescently labeled proteins. Bars: (A to F) 5 μm; (G to I), 1 μm. Transiently transfected cells grown in chambers were imaged as described (13).

Expression of moderate levels of a dominant-negative GGA1 VHS-GAT construct lacking the hinge and GAE domains caused accumulation of CD-MPR at the TGN and its depletion from the periphery (Fig. 4B, arrow), as previously observed for the CI-MPR (13). In contrast, this construct did not affect the localization of LAMP-1 to lysosomes (Fig. 4E), Tac (14) to the plasma membrane (Fig. 4H), and TGN38 to the TGN (Fig. 4K). At these moderate levels of GGA1 VHS-GAT expression, there was also no visible effect on the localization of AP-1 to the TGN and peripheral foci (Fig. 4N). Thus, GGA function was apparently required for CD-MPR exit from the TGN.

Figure 4

Effect of GGA1 VHS-GAT expression on CD-MPR localization. COS-7 cells were transfected with plasmids encoding GFP-tagged GGA1 VHS-GAT (GFT-VHS-GAT) alone (A toF, M to O), GFP-VHS-GAT and Tac (G to I), or GFP-VHS-GAT and HA-TGN38 (J to L). Cells were immunostained with rabbit anti–CD-MPR (A to C), mouse anti–LAMP-1 (AC17) (D to F), rabbit anti-Tac (from our lab) (G to I), mouse anti-hemagglutinin epi-tope (BabCo) (J to L), or mouse anti–AP-1-γ (100/3, Sigma) (M to O) antibodies, followed by the corresponding Cy3 anti-rabbit or anti-mouse IgG. Bar: 15 μm.

We have identified highly specific interactions of acidic-cluster-dileucine signals involved in sorting of MPRs at the TGN with the VHS domain of the GGAs. These interactions have the hallmarks of bona fide signal-recognition events, including (i) similar sequence requirements for interactions and for function of the signals in vivo (2, 3); (ii) an affinity comparable to that of other well-characterized sorting signals for their recognition molecules (14, 16, 17); (iii) colocalization of the GGAs and the MPRs to the TGN and peripheral vesicles, and their exit from the TGN on the same tubulo-vesicular carriers; (iv) the properties of the GGAs as clathrin adaptors (13) and the requirement for clathrin for MPR sorting at the TGN (18); and (v) the ability of a dominant-negative GGA mutant to interfere with sorting of MPRs at the TGN. The elucidation of a specific signal-recognition function for the VHS domain completes the assignment of at least one function to each of the four GGA domains and establishes the GGAs as genuine sorting adaptors.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: juan{at}helix.nih.gov

REFERENCES AND NOTES

View Abstract

Navigate This Article