Available online at www.sciencedirect.com

www.elsevier.com/locate/ybbrc

Biochemical and Biophysical Research Communications 366 (2008) 724–730

AP-1 and retromer play opposite roles in the trafficking of sortilinbetween the Golgi apparatus and the lysosomes

Maryssa Canuel a, Stephane Lefrancois b, Jibin Zeng a, Carlos R. Morales a,*

a Department of Anatomy and Cell Biology, McGill University, 3640 University Street, Montreal, Que., Canada H3A 2B2b Centre de Recherche de l’hopital Maisonneuve-Rosemont, Montreal, Que., Canada

Received 23 November 2007Available online 17 December 2007

Abstract

Sortilin has been implicated in the sorting of one soluble hydrolase and two sphingolipid activator proteins to the lysosomes. Whilethe GGA adaptor proteins have been demonstrated to play a role in the targeting of sortilin to the endosomes, the recycling of sortilinhas not yet been elucidated. Here we examine the role of two adaptor protein complexes, AP-1 and retromer. Our results demonstratethat AP-1 is required for the transport of sortilin to the endosomes and retromer for the recycling of sortilin to the Golgi apparatus.While inhibition of AP-1 causes accumulation of sortilin in the Golgi apparatus, RNAi depletion of retromer results in retention of sor-tilin in the lysosomes. We also demonstrate that the interaction of sortilin with retromer occurs through a YXXU site in its cytosolic tail.In conclusion, our observations indicate that retromer and AP-1 play opposite roles in the trafficking of sortilin.Crown copyright � 2007 Published by Elsevier Inc. All rights reserved.

Keywords: Sorting receptors; Sortilin; Retromer; AP-1; Prosaposin

Efficient delivery of lysosomal sorting receptors andtheir cargo depends upon interaction with adaptor proteins[1–5]. Following release of their cargo into the endosomallumen, sorting receptors are suspected to recycle back tothe Golgi via the retromer complex [6,7,10]. The mamma-lian retromer complex consists of Vps26, Vps29, Vps35,and sorting nexins (SNX) 1 and 2 [8,11].

The mannose 6-phosphate receptor (M6PR), responsi-ble for sorting most soluble hydrolases [12,13], has beendemonstrated to reach the endosomal compartmentthrough interactions with the Golgi-localized, c ear-con-taining ARF binding adaptor proteins (GGAs), as well aswith AP-1 [14–19]. AP-1 is a heterotetrameric adaptor thatconsists of two large subunits (c and b1), a medium subunit(l1), and a small subunit (r1) [20]. The b-subunit of AP-1has been demonstrated to interact with clathrin, whileinteraction with the receptor occurs through the l-subunit

0006-291X/$ - see front matter Crown copyright � 2007 Published by Elsevie

doi:10.1016/j.bbrc.2007.12.015

* Corresponding author. Fax: +1 514 398 5047.E-mail address: carlos.morales@mcgill.ca (C.R. Morales).

[21–23]. Upon release of its cargo, the M6PR binds retro-mer and recycles back to the Golgi apparatus [24]. Theinteraction between retromer and the M6PR is purportedto occur through the Vps35 subunit of retromer [24].

Sortilin is a sorting receptor that has been implicated inthe trafficking of sphingolipid activator proteins (SAPs)[25]. Interestingly, the cytoplasmic tails of the M6PR andsortilin exhibit sequence and functional homology [27].These observations have led us to hypothesize that theM6PR and sortilin share a conserved sorting and traffick-ing mechanism. Herein we demonstrate a role for both retr-omer and AP-1 in the trafficking of sortilin. We alsoidentify a new sequence in the cytoplasmic tail of sortilinthat governs the interaction with retromer.

Materials and methods

Cell culture. COS-7 cells were cultured in DMEM (Invitrogen) sup-plemented with 10% FBS, 5% penicillin and streptomycin, and L-gluta-mine. Cells were maintained at 5% CO2 at 37 �C. The cells were

r Inc. All rights reserved.

M. Canuel et al. / Biochemical and Biophysical Research Communications 366 (2008) 724–730 725

transfected using PolyFect and HiPerFect Transfection Reagents(Qiagen).

Antibodies and constructs. The sortilin-myc construct was a gift fromDr. C. Petersen (University of Aarhus). The Y14A and L17A mutationswere generated using the QuikChange II XL Mutagenesis kit (Stratagene).Sortilin was detected with a monoclonal antibody from BD TransductionLaboratories. Sortilin constructs were detected with a monoclonal anti-myc antibody (Santa Cruz Biotechnology). The rabbit polyclonal anti-body to prosaposin was produced in our laboratory [28]. Cathepsin Bantibody was a rabbit polyclonal antibody (Santa Cruz Biotechnology).Monoclonal anti-LAMP-1 antibody was from Santa Cruz Biotechnology.The previously characterized rabbit polyclonal anti-Vps26 antibody was agift from Dr. J. Bonifacino (NIH, Bethesda, MD) [8]. Monoclonal anti-c-adaptin and b-actin antibodies were from Sigma–Aldrich. Anti-rabbit andanti-mouse IgG antibodies conjugated to Alexa 488 or 594 were fromMolecular Probes. Goat anti-rabbit and anti-mouse IgG antibodies con-jugated to HRP were from Santa Cruz Biotechnology.

Co-immunoprecipitation. COS-7 cells were transfected with sortilin-myc, Y14A, or L17A sortilin constructs, or pcDNA3.1. Twenty four hourssubsequently the cells were harvested and lysed in 50 mM Tris–HCl (pH6.0), 0.15 M NaCl, 0.5% NP-40, and protease inhibitor cocktail (Roche).The cellular supernatants were incubated with protein A Sepharose beadscoupled to anti-myc antibody at 4 �C overnight. The resulting complexeswere washed and boiled in 3· SDS–PAGE sample buffer. Proteins wereseparated on a 10% SDS–PAGE gel, transferred to a nitrocellulosemembrane and immunoblotted with either anti-Vps26 diluted to 1:5000 oranti-c-adaptin antibody diluted to 1:2500. Anti-rabbit and anti-mousesecondary antibodies coupled to HRP were used at 1:10,000.

Yeast two-hybrid assays. The sortilin c-terminal tail (residues 779–831)was subcloned into either the pB42AD or pGAD424 vector. Truncationmutations were obtained by inserting a stop codon after residue 789, 799,809, or 819. Vps26, Vps29, and Vps35 were cloned into the pLexA,pGAD424, or pGBKT7 vectors. The co-transformed yeast strains wereplated on synthetic medium plates lacking leucine and tryptophan (GAL4,two-hybrid) or lacking histidine and tryptophan (LexA, two-hybrid).

Fig. 1. Sortilin interaction with retromer and AP-1. The interaction of sortilinanti-c-adaptin antibodies (A). Immunoprecipitation was done using anti-myc avector did not immunoprecipitate Vps26 or c-adaptin (A). Lane 2 shows thatLane 3 shows the cell lysate (A). Yeast were co-transformed with the retromerthe LexA system (B). The interaction of the l1 subunit of AP-1 was tested focontrols for yeast growth (C).

Colonies after 72 h of growth were re-plated on synthetic medium plateslacking leucine, tryptophan, and histidine (GAL4, two-hybrid) or lackinghistidine, tryptophan, and leucine (LexA, two-hybrid).

RNAi. The sequence selected to silence Vps26 has been previouslycharacterized [24]. The sequence for l1A (AP-1) has also been tested [29].Both of these RNAi sequences were produced with a 3 0-Alexa Fluor 488tag (Qiagen). A negative control RNAi was purchased from Qiagen with a3 0-Alexa Fluor 488 tag. Inhibition of expression was tested by Westernblot.

Immunofluorescence microscopy. Cells were seeded onto coverslips andimmediately transfected with the siRNA using HiPerfect (Qiagen). Forty-eight hours later, the cells were transfected with the sortilin-myc constructand 24 h after fixed in 4% paraformaldehyde. The cells were permeabilizedwith 0.25% Triton X-100. Blocking was performed in 3% goat serum and0.1% Triton X-100. Primary antibody incubation was performed at a 1:300dilution overnight at 4 �C for anti-myc, anti-prosaposin, anti-cathepsin B,anti-LAMP1, anti-c-adaptin, and anti-Vps26p antibodies. The cells werethen washed with PBS and incubated with Alexa-conjugated secondaryantibodies diluted to 1:300 for 1 h at RT. The cells were again washed andtreated with Hoeschst 33342. The cells were mounted to coverslips usingGeltol (Thermo Electron Corporation). The immunofluorescence stainingwas visualized with an LSM 510 confocal microscope (Carl Zeiss).

Similarly, the distribution of the Y14A and L17A sortilin-myc con-structs was assessed by confocal microscopy. Cells expressing Y14A orL17A constructs were prepared as described above. These cells werestained with anti-myc and anti-LAMP-1 antibodies followed by Alexa-coupled secondary antibodies.

Metabolic labeling. COS-7 cells were transfected with control, Vps26,or l1A siRNA. Seventy-two hours post-transfection, the cells were washedand starved for 30 min in methionine-free, cysteine-free DMEM (Invit-rogen) supplemented with 5% dialyzed fetal bovine serum. The cells werepulse-labeled for 30 min in medium supplemented with 1023 lCi/ml [35S]methionine (Amersham Biosciences). The chase was carried out usingDMEM with 10% fetal bovine serum, 5 nM unlabeled methionine, and2.5 nM unlabeled cysteine for 0, 30, 60, and 90 min. The culture medium

with retromer and AP-1 complexes was tested in cells using anti-Vps26 andntibody. Lane 1 shows that the lysate from cells transfected with an emptythe sortilin-myc construct immunoprecipitated Vps26 and c-adaptin (A).subunits and sortilin c-terminal fragments and tested for interaction usingr interaction with sortilin c-terminal fragments. p53 and LT are used as

726 M. Canuel et al. / Biochemical and Biophysical Research Communications 366 (2008) 724–730

was collected and incubated with protein A Sepharose beads coupled toanti-prosaposin or anti-cathepsin B antibody at 4 �C overnight. The beadswere then washed and the proteins eluted with 3· SDS–PAGE samplebuffer. The samples were subjected to reducing SDS–PAGE and analyzedby autoradiography.

Results and discussion

Sortilin interaction with retromer and AP-1

It has been established that the M6PR interacts with theretromer complex via its cytoplasmic tail to be recycledfrom the endosome to the Golgi [24]. Given that sortilinpossesses a high degree of homology to the M6PR cytosolictail, we hypothesized that both receptors share a conservedtrafficking mechanism. The use of a chimeric construct con-sisting of CD8 and the cytosolic tail of sortilin suggestedthat this hypothesis may be correct [30]. However, the chi-meric nature of this probe did not permit analysis of therequirement of these adaptor complexes for the traffickingof sortilin and its cargo. To overcome this deficiency weemployed a co-immunoprecipitation (Co-IP) assay toexamine if full-length sortilin interacts with AP-1 and retr-omer. Given that AP-1 and retromer have been shown tobind to the cytosolic tail of the M6PR and the CD8-sortilinchimera [31], we first examined whether full-length sortilininteracts with these adaptors. To this end, COS-7 cells weretransfected with sortilin-myc construct or pcDNA3.1B.The cell lysates were immunoprecipitated using anti-mycantibody conjugated to protein A Sepharose beads. The

Fig. 2. Mutational analysis of sortilin–retromer and sortilin–AP-1 interactionY14A and L17A sortilin constructs. Cells were transfected with these constexpressed, producing bands at 100 kDa (A). The interaction of Y14A and Lantibody (B,C). Both mutant constructs were unable to interact with Vps26 viaband at 35 kDa (B). In contrast, Y14A and L17A constructs had no effect on biexamined by immunofluorescent confocal microscopy (D–F). The sortilin conssecondary antibody conjugated to Alexa 488 (green). The cells were also labelY14A (E) and L17A (F) constructs increased retention of sortilin in the perin

complexes were resolved on a 10% acrylamide gel, trans-ferred to a nitrocellulose membrane, and immunoblottedfor Vps26 or c-adaptin (Fig. 1A). Our results showed thatsortilin-myc immunoprecipitated retromer (Vps26) andAP-1 (c-adaptin), producing bands at 35 and 90 kDa,respectively (Fig. 1A).

To identify the regions in the cytosolic tail of sortilinthat interact with retromer or AP-1, a Yeast Two-Hybridapproach was employed. Because retromer is not an abun-dant protein we used the LexA system. When the entirecytoplasmic tail of sortilin (residues 779–831) was tested,a specific interaction with Vps35 was observed, as sortilindid not interact with either Vps26 or Vps29 (Fig. 1B). Toidentify the residues involved we designed various c-termi-nal deletion constructs of sortilin. We found that residues789–799 were required for the recruitment of Vps35(Fig. 1B).

Using the Gal4 system, we demonstrated that therecruitment of AP-1 to the cytoplasmic tail of sortilin ismediated by l1 (Fig. 1C). To determine which residueswere important we used the same c-terminal truncationsof sortilin as described above (Fig. 1B). Interestingly, resi-dues 789–799 also were required to mediate the interactionwith l1 (Fig. 1C).

Mutational analysis

While the results obtained with a CD8 chimeric sortilinconstruct suggest a role for an FLV motif found in the

s. The binding between sortilin and retromer or AP-1 was studied usingructs as well as with wild-type sortilin-myc (A). Y14A and L17A were17A constructs was examined by co-immunoprecipitation with anti-mycretromer, while the binding of wild-type sortilin was unaffected producing anding to AP-1 and bands were visible at 90 kDa (C). In addition, cells weretructs were detected by anti-myc antibody, followed by incubation with aed for LAMP-1 (red). When compared to wild-type sortilin (D), both theuclear region. Scale bars = 8 lm.

M. Canuel et al. / Biochemical and Biophysical Research Communications 366 (2008) 724–730 727

cytosolic tail of sortilin [30], a more common sorting motifis found between residues 789–799, namely YSVL. Thissequence is of interest because YXXU has been implicatedin the trafficking of sorting receptors and in the recruitmentof accessory proteins [7,27]. Thus, to characterize the inter-action between sortilin and retromer or AP-1, two muta-tions, Y14A and L17A, were introduced in the YSVLsequence of sortilin. The interaction between Y14A orL17A mutated sortilin and Vps26 or c-adaptin was testedin COS-7 cells. Protein A Sepharose beads conjugated toanti-myc were incubated with the cell lysate. The resultingimmunoprecipitated complexes were run on a 10% acryl-amide gel, transferred to a nitrocellulose membrane andimmunoblotted with anti-Vps26 or anti-c-adaptin antibody(Fig. 2A–C). The results showed that while the Y14A orL17A mutations inhibited interaction with Vps35/Vps26,the interaction with c-adaptin was unaffected, producingbands at 90 kDa (Fig. 2B and C). These results indicatethat the interaction between sortilin and retromer wasdependent upon the YSVL sequence. However, the interac-

Fig. 3. Effect of retromer and AP-1 RNAi treatment. To verify the inhibitionmicroscopy were performed. Cells expressing either siRNA had no Vps26 or c-a(A,B). These images were compared to cells transfected with control siRNA, wsiRNAs inhibited expression of their targets as is seen by the absence of bandswas unchanged (A,B). To determine and compare the effect of retromer and APcargo (cathepsin B), pulse-chase analysis was performed subsequent to Vps26anti-cathepsin B, or anti-prosaposin antibody at 0, 30, 60, and 90 min (C,D).presence of cathepsin B and prosaposin in the media (C,D).

tion of sortilin with AP-1 relies on a distinct sequencewithin residues 779–789 of sortilin, or more likely, AP-1binds to the dileucine motif of sortilin downstream of thetyrosine motif.

Distribution of Y14A and L17A

To determine the effect of the Y14A and L17A muta-tions on the trafficking of sortilin, immunofluorescencemicroscopy was employed. Cells were fixed in 4% parafor-maldehyde prior to staining with both anti-myc and anti-LAMP-1 antibodies and compared to cells expressingwild-type sortilin. Secondary antibodies conjugated toAlexa 488 or 594 were used to specifically detect myc(green) and LAMP-1 (red). While wild-type sortilin-mycwas detected in the perinuclear region and in punctatestructures labeled with LAMP-1 (Fig. 2D), the presenceof the Y14A construct was decreased in LAMP-1 positivestructures (Fig. 2E). Similar results were observed withL17A (Fig. 2F). These results suggest that while Y14 and

of retromer and AP-1 by siRNA, Western blotting and immunoconfocaldaptin expression, as demonstrated by the absence of red immunostaining

hich showed normal expression of Vps26 and c-adaptin (A,B). In addition,at 35 and 90 kDa by Western blot. Conversely, the loading control, b-actin-1 depletion on the transport of a sortilin cargo (prosaposin) and a M6PRor l1 siRNA treatment. The culture media was immunoprecipitated withThe inhibition of both Vps26 and l1 by siRNA resulted in the increased

728 M. Canuel et al. / Biochemical and Biophysical Research Communications 366 (2008) 724–730

L17 may be essential for the interaction of sortilin withretromer, these amino acid residues are also important inmediating the exit of sortilin from the TGN.

Retromer and AP-1 RNAi

To characterize the roles of both retromer and AP-1 inthe trafficking of full-length sortilin, two siRNAs wereselected to target the Vps26 domain of retromer and thel1 subunit of AP-1 [24,29]. When these 3 0-Alexa Fluor488 tagged siRNAs were transfected in COS-7 cells, therewas complete inhibition of retromer and AP-1 expressionas detected by Western blot using antibodies for Vps26and c-adaptin (Fig. 3A and B).

To further confirm these results, immunofluorescencemicroscopy was performed. After fixation, cells were trea-

Fig. 4. RNAi inhibited transport of sortilin cargo. Cells were transfected withor LAMP-1 (A–D). Sortilin staining was reduced in the perinuclear region andVps26 siRNA (A). Sortilin-myc was localized to the perinuclear Golgi region wabsent in cells expressing Vps26 or l1 siRNA (C,B). LAMP-1 expression waseither siRNA (D). Scale bars = 8 lm.

ted with anti-Vps26 or anti-c-adaptin antibody followedby incubation with secondary antibody labeled with Alexa594. Green staining within the cells showed siRNA expres-sion in control as well as Vps26 and l1 siRNA treated cells(Fig. 3). In cells transfected with retromer or AP-1 specificsiRNAs, staining of Vps26 and c-adaptin was abolished(Fig. 3A and B). These results confirmed that the siRNAconstructs specifically inhibited retromer and AP-1expression.

To determine the effect of the absence of functional retr-omer and AP-1 on the transport of the sortilin ligand pro-saposin, pulse-chase analysis was performed subsequent toVps26 or l1 siRNA expression (Fig. 3D). Given that retr-omer has been implicated in the transport of the M6PR,the fate of cathepsin B, a ligand of the M6PR, was alsotested (Fig. 3C) [24].

Vps26 or l1 siRNA and stained for sortilin-myc, cathepsin B, prosaposin,increased in punctate lysosomal structures when cells were transfected withhen cells expressed l1 siRNA (A). Prosaposin and cathepsin B staining wasexamined as a negative control (D). LAMP-1 staining was unaffected by

M. Canuel et al. / Biochemical and Biophysical Research Communications 366 (2008) 724–730 729

Cells expressing siRNA were pulsed with [35S] methio-nine-cysteine and chased with DMEM. Immunoprecipita-tion of the culture medium was done at 0, 30, 60, and90 min using anti-cathepsin B, or anti-prosaposin antibod-ies (Fig. 3C and D). The resulting immunoprecipitates wereresolved on 10% SDS–PAGE gels and visualized by auto-radiography. Ablation of Vps26 expression resulted inthe increased presence of cathepsin B and prosaposin inthe media (Fig. 3C and D). Comparable results wereobtained for cathepsin B and prosaposin when expressionof the l1 subunit of AP-1 was inhibited (Fig. 3C and D).These results suggest that without functional sorting andrecycling pathways, cargo proteins are secreted.

RNAi inhibited transport of sortilin

The consequence of inhibition of Vps26 and l1-adaptinby RNAi on the transport of full-length sortilin and itscargo was studied by confocal microscopy (Fig. 4). Cellsexpressing Vps26 or l1 Alexa 488 tagged siRNA were fur-ther transfected with sortilin-myc and compared tountransfected cells. After fixation, cells were incubated withanti-myc (Fig. 4A), anti-cathepsin B (Fig. 4B), anti-pro-saposin (Fig. 4C), or anti-LAMP-1 antibody (Fig. 4D) fol-lowed by secondary antibody labeled with Alexa 594.When cells expressing Vps26 siRNA were stained for sorti-lin-myc, there was a notable absence of sortilin in the per-inuclear Golgi region and increased expression in punctatestructures (Fig. 4A). Thus, Vps26 siRNA resulted in the re-localization of sortilin from the TGN to the endosomes(Fig. 4A). In contrast, inhibition of l1 resulted in retentionof sortilin in the TGN (Fig. 4A). This difference in sortilinsorting suggests that the trafficking of full-length sortilin tothe endosomes is dependent upon AP-1 whereas its recy-cling requires retromer. Interestingly, prosaposin wasabsent from cells transfected with Vps26 or l1 siRNA(Fig. 4C). The lack of prosaposin staining appears to bea direct consequence of the sortilin sorting inhibition,which in turn, caused the release of prosaposin to the extra-cellular space (Fig. 3). Retromer and AP-1 abrogation alsoabolished cathepsin B staining. In contrast, cathepsin Bstaining was observed in punctate structures of control cells(Fig. 4B). Using the same approach, LAMP-1 localizationwas unaffected by Vps26 or l1 siRNA (Fig. 4D). The sim-ilarity between prosaposin (Fig. 4C) and cathepsin B(Fig. 4B) indicates that both sortilin and the M6PR usethe same trafficking mechanisms. In conclusion, our resultsdemonstrate a critical role for both retromer and AP-1 inthe trafficking of sortilin and identify the YXXU motif asan essential sequence of sortilin for the recruitment ofretromer.

Acknowledgment

This paper was supported by a CIHR operating grant toC.R. Morales.

References

[1] J. Hirst, M.R. Lindsay, M.S. Robinson, GGAs: roles of the differentdomains and comparison with AP-1 and clathrin, Mol. Biol. Cell 12(2001) 3573–3588.

[2] J. Hirst, W.W. Lui, N.A. Bright, N. Totty, M.N. Seaman, M.S.Robinson, A family of proteins with gamma-adaptin and VHSdomains that facilitate trafficking between the trans-Golgi networkand the vacuole/lysosome, J. Cell Biol. 149 (2000) 67–80.

[3] G. Costaguta, C.J. Stefan, E.S. Bensen, S.D. Emr, G.S. Payne, YeastGga coat proteins function with clathrin in Golgi to endosometransport, Mol. Biol. Cell 12 (2001) 1885–1896.

[4] E.C. Dell’Angelica, R. Puertollano, C. Mullins, R.C. Aguilar, J.D.Vargas, L.M. Hartnell, J.S. Bonifacino, GGAs: a family of ADPribosylation factor-binding proteins related to adaptors and associ-ated with the Golgi complex, J. Cell Biol. 149 (2000) 81–94.

[5] O. Zhdankina, N.L. Strand, J.M. Redmond, A.L. Boman, YeastGGA proteins interact with GTP-bound Arf and facilitate transportthrough the Golgi, Yeast 18 (2001) 1–18.

[6] M.N. Seaman, J.M. McCaffery, S.D. Emr, A membrane coat complexessential for endosome-to-Golgi retrograde transport in yeast, J. CellBiol. 142 (1998) 665–681.

[7] A.A. Cooper, T.H. Stevens, Vps10p cycles between the late-Golgi andprevacuolar compartments in its function as the sorting receptor formultiple yeast vacuolar hydrolases, J. Cell Biol. 133 (1996) 529–541.

[8] C.R. Haft, M. de la Luz Sierra, R. Bafford, M.A. Lesniak, V.A. Barr,S.I. Taylor, Human orthologs of yeast vacuolar protein sortingproteins Vps26, 29, and 35: assembly into multimeric complexes, Mol.Biol. Cell 11 (2000) 4105–4116.

[10] R.C. Kurten, A.D. Eddington, P. Chowdhury, R.D. Smith, A.D.Davidson, B.B. Shank, Self-assembly and binding of a sorting nexinto sorting endosomes, J. Cell Sci. 114 (2001) 1743–1756.

[11] B.F. Horazdovsky, B.A. Davies, M.N. Seaman, S.A. McLaughlin, S.Yoon, S.D. Emr, A sorting nexin-1 homologue, Vps5p, forms acomplex with Vps17p and is required for recycling the vacuolarprotein-sorting receptor, Mol. Biol. Cell 8 (1997) 1529–1541.

[12] P. Lobel, K. Fujimoto, R.D. Ye, G. Griffiths, S. Kornfeld, Mutationsin the cytoplasmic domain of the 275 kd mannose 6-phosphatereceptor differentially alter lysosomal enzyme sorting and endocyto-sis, Cell 57 (1989) 787–796.

[13] S. Kornfeld, Structure and function of the mannose 6-phosphate/insulin like growth factor II receptors, Annu. Rev. Biochem. 61 (1992)307–330.

[14] S. Waguri, F. Dewitte, R. Le Borgne, Y. Rouille, Y. Uchiyama, J.F.Dubremetz, B. Hoflack, Visualization of TGN to endosome traffick-ing through fluorescently labeled MPR and AP-1 in living cells, Mol.Biol. Cell 14 (2003) 142–155.

[15] H. Takatsu, Y. Katoh, Y. Shiba, K. Nakayama, Golgi-localizing,gamma-adaptin ear homology domain, ADP-ribosylation factor-binding (GGA) proteins interact with acidic dileucine sequenceswithin the cytoplasmic domains of sorting receptors through theirVps27p/Hrs/STAM (VHS) domains, J. Biol. Chem. 276 (2001)28541–28545.

[16] R. Puertollano, R.C. Aguilar, I. Gorshkova, R.J. Crouch, J.S.Bonifacino, Sorting of mannose 6-phosphate receptors mediated bythe GGAs, Science 292 (2001) 1712–1716.

[17] Y. Zhu, B. Doray, A. Poussu, V.P. Lehto, S. Kornfeld, Binding ofGGA2 to the lysosomal enzyme sorting motif of the mannose 6-phosphate receptor, Science 292 (2001) 1716–1718.

[18] R. Le Borgne, G. Griffiths, B. Hoflack, Mannose 6-phosphatereceptors and ADP-ribosylation factors cooperate for high affinityinteraction of the AP-1 Golgi assembly proteins with membranes, J.Biol. Chem. 271 (1996) 2162–2170.

[19] M. Boehm, J.S. Bonifacino, Adaptins: the final recount, Mol. Biol.Cell 12 (2001) 2907–2920.

[20] T. Kirchhausen, Adaptors for clathrin-mediated traffic, Annu. Rev.Cell Dev. Biol. 15 (1999) 705–732.

730 M. Canuel et al. / Biochemical and Biophysical Research Communications 366 (2008) 724–730

[21] A. Gallusser, T. Kirchhausen, The beta 1 and beta 2 subunits of theAP complexes are the clathrin coat assembly components, EMBO J.12 (1993) 5237–5244.

[22] W. Shih, A. Gallusser, T. Kirchhausen, A clathrin-binding site in thehinge of the beta 2 chain of mammalian AP-2 complexes, J. Biol.Chem. 270 (1995) 31083–31090.

[23] H. Ohno, J. Stewart, M.C. Fournier, H. Bosshart, I. Rhee, S.Miyatake, T. Saito, A. Gallusser, T. Kirchhausen, J.S. Bonifacino,Interaction of tyrosine-based sorting signals with clathrin-associatedproteins, Science 269 (1995) 1872–1875.

[24] C.N. Arighi, L.M. Hartnell, R.C. Aguilar, C.R. Haft, J.S.Bonifacino, Role of the mammalian retromer in sorting of thecation-independent mannose 6-phosphate receptor, J. Cell Biol.165 (2004) 123–133.

[25] S. Lefrancois, J. Zeng, A.J. Hassan, M. Canuel, C.R. Morales, Thelysosomal trafficking of sphingolipid activator proteins (SAPs) ismediated by sortilin, EMBO J. 22 (2003) 6430–6437.

[27] M.S. Nielsen, P. Madsen, E.I. Christensen, A. Nykjaer, J. Gliemann,D. Kasper, R. Pohlmann, C.M. Petersen, The sortilin cytoplasmic tailconveys Golgi-endosome transport and binds the VHS domain of theGGA2 sorting protein, EMBO J. 20 (2001) 2180–2190.

[28] Q. Zhao, C.R. Morales, Identification of a novel sequence involved inlysosomal sorting of the sphingolipid activator protein prosaposin, J.Biol. Chem. 275 (2000) 24829–24839.

[29] K. Janvier, J.S. Bonifacino, Role of the endocytic machinery in thesorting of lysosome-associated membrane proteins, Mol. Biol. Cell 16(2005) 4231–4242.

[30] M.N. Seaman, Identification of a novel conserved sorting motifrequired for retromer-mediated endosome-to-TGN retrieval, J. CellSci. 120 (2007) 2378–2389.

[31] R. Le Borgne, A. Schmidt, F. Mauxion, G. Griffiths, B. Hoflack,Binding of AP-1 Golgi adaptors to membranes requires phosphor-ylated cytoplasmic domains of the mannose 6-phosphate/insulin-likegrowth factor II receptor, J. Biol. Chem. 268 (1993) 22552–22556.