The retromer complex is a conserved endosomal protein sorting complex that sorts membrane proteins into nascent endosomal tubules. The recognition of membrane proteins is mediated by the cargo-selective retromer complex, a stable trimer of the Vps35 (vacuolar protein sorting 35), Vps29 and Vps26 proteins. We have recently reported that the cargo-selective retromer complex associates with the WASH (Wiskott–Aldrich syndrome homologue) complex, a multimeric protein complex that regulates tubule dynamics at endosomes. In the present study, we show that the retromer–WASH complex interaction occurs through the long unstructured ‘tail’ domain of the WASH complex–Fam21 protein binding to Vps35, an interaction that is necessary and sufficient to target the WASH complex to endosomes. The Fam21-tail also binds to FKBP15 (FK506-binding protein 15), a protein associated with ulcerative colitis, to mediate the membrane association of FKBP15. Elevated Fam21-tail expression inhibits the association of the WASH complex with retromer, resulting in increased cytoplasmic WASH complex. Additionally, overexpression of the Fam21-tail results in cell-spreading defects, implicating the activity of the WASH complex in regulating the mobilization of membrane into the endosome-to-cell surface pathway.

INTRODUCTION

Endosomes operate as major sorting stations that receive proteins and lipids from the cell surface via endocytosis as well as from the biosynthetic pathway [1]. Additionally, membrane flux through the endocytic system plays a key role in the regulation of cell size and shape [24]. The fidelity of endosomal protein sorting is therefore of profound physiological importance and there are numerous examples of pathologies resulting from a failure of correct protein sorting, including the Niemann–Pick disease [5], Charcot–Marie–Tooth disease and HSP (hereditary spastic paraplegia) [6,7]. There is also much evidence to suggest a causal link between the endosomal protein sorting of amyloid precursor protein and the mechanisms that govern production of the neurotoxic amyloid β-peptide that is a key pathological event in Alzheimer's disease [8].

Endosomal protein sorting occurs concomitantly with increasing endosomal acidity. The influx of protons into endosomes is mediated by the V-ATPase (vacuolar ATPase) and regulates the association of some ligands (e.g. lysosomal hydrolases) with their respective receptor [9,10]. One of the principal components of endosomal protein sorting is the retromer complex. Retromer is an evolutionarily conserved protein complex that is required for the localization of several membrane proteins of physiological importance, including the CIMPR (cation-independent mannose 6-phosphate receptor) [11,12], the Vps10 (Vps is vacuolar protein sorting)-family sorting receptors sortilin and SorL1 [13,14] and the divalent cation transporter DMT1 [15].

The retromer complex recognizes cargo proteins through a trimeric complex comprising Vps35, Vps29 and Vps26 [16,17] that is recruited to the endosomal membrane by the small GTPase Rab7 [18,19]. The Vps35 protein is believed to play a key role in cargo recognition [20], although the structural similarity between Vps26 and the arrestin family hints that Vps26 may also act in cargo recognition [21,22]. In addition to the cargo-recognition complex, retromer-mediated endosome-to-Golgi retrieval also utilizes the activity of members of the Snx (sorting nexin)-BAR (Bin/amphiphysin/Rvs) subfamily of Snxs, Snx1 and Snx2, in complex with Snx5 and Snx6 [23]. The Snx-BAR sorting nexins mediate the tubulation of endosomal membranes through their C-terminal BAR domains [24].

The cargo-selective retromer complex associates with a number of accessory proteins, including the Rab GTPase-activating protein, TBC1D5 [TBC (Tre-2/Bub2/Cdc16) domain family, member 5], the WASH (Wiskott–Aldrich syndrome homologue) complex and FKBP15 (FK506-binding protein 15) [19,25], a protein implicated in both ulcerative colitis and growth cone collapse in neurons [26,27]. The WASH complex is a protein complex containing WASH1, a member of the WASP (Wiskott–Aldrich syndrome protein)/WAVE (WASP verprolin homologous) family of actin nucleating promoting factors, KIAA1033 (also known as SWIP), Strumpellin and Fam21. Loss of function of the WASH complex results in dysregulation of endosomal tubules, and since the Strumpellin protein is mutated in HSP, the activity of the WASH complex has been of interest to researchers studying the underlying pathology of HSP and related neuropathies [25,28,29]. Although the precise function of the WASH complex in endosomal protein sorting is yet to be determined, recent studies in Dictyostelium discoideum have revealed a role for the WASH complex in regulating the localization of V-ATPase [30].

Initial reports of the WASH complex association with endosomes [28,29] contained a number of interesting discrepancies. For example, the work by Gomez and Billadeau [28] did not detect Strumpellin or KIAA1033 in association with Fam21 and WASH1, whereas the study by Derivery et al. [29] struggled to detect Fam21 that, when observed, was often proteolysed and partially degraded. This could indicate that the WASH complex is not a stable single entity but rather a looser association of the constituent proteins. Studies conducted by our laboratory showed that the WASH complex comprising Strumpellin, KIAA1033, Fam21 and WASH1 bound to retromer as a single unit. Questions remain, however, as to the nature and assembly of the endosomal WASH complex [25].

Structural similarities between the endosomal WASH complex and the analogous plasma membrane-localized WAVE complex have been recently highlighted [31]. An important difference between the WASH and WAVE complexes, however, is the large unstructured ‘tail’ domain of the Fam21 protein. This clearly distinguishes Fam21 from its WAVE complex counterpart, Abi. Although Abi is ~370 amino acids in size, Fam21 is a much larger protein of ~1300 amino acids containing a predicted globular ‘head’ domain (~200 amino acids) at its N-terminus with a long (~1100 amino acids) unstructured ‘tail’ that contains a binding site for the actin capping protein [32].

In the present study, we show that Vps35 directly binds the unstructured tail of Fam21. We show that the interaction between the Fam21-tail and Vps35 is sufficient to target the Fam21-tail to endosomes, consistent with the requirement for the cargo-selective retromer complex to mediate the endosomal recruitment of the WASH complex. Additionally, we report that the Fam21-tail can bind to the FKBP15 protein, mediating its endosomal localization. Overexpression of the Fam21-tail results in displacement of FKBP15 from the endosome membrane and competes with the endogenous WASH complex for binding to retromer, resulting in cell spreading defects, thereby implicating the activity of the WASH complex in regulating cell shape.

EXPERIMENTAL

Antibodies, general reagents and biochemicals

Unless otherwise stated, chemicals and reagents were purchased from Sigma. 125I-labelled Protein A used in Western blotting was obtained from PerkinElmer. Restriction enzymes and other molecular biology reagents were purchased from New England Biolabs. siRNA (small interfering RNA) oligonucleotides were purchased from Dharmacon. Oligofectamine and Optimem used in siRNA transfections were purchased from Invitrogen.

The anti-Fam21 sera (used for immunofluorescence) and the anti-Strumpellin sera were both obtained from Santa Cruz Biotechnology, the anti-FKBP15 serum used for immunofluorescence was purchased from Abcam and the anti-FKBP15 antibody used for Western blotting was generated against a GST (glutathione transferase)–FKBP15 fusion protein as described previously [25]. Anti-WASH1 antibody was purchased from Sigma or Millipore. Anti-Fam21 serum used for Western blotting was produced using a commercial service (Charles River Laboratories; Margate, Kent, U.K.) using a GST–Fam21-head domain fusion protein to immunize a rabbit. The antiserum was affinity purified using GST–Fam21-head domain coupled with CNBr-activated Sepharose (Amersham Pharmacia). Monoclonal anti-Snx1 and anti-EEA1 (early endosome antigen 1) were purchased from BD Biosciences. Monoclonal anti-GFP (green fluorescent protein) antibody was purchased from Invitrogen, as were fluorescently labelled secondary antibodies used in immunofluorescence. Polyclonal antisera against GFP, Vps26, Vps35 and Snx1 have been described previously [12,19].

Cell culture and production of stably transfected cell lines

HeLa M cells [33] were used throughout the present study and are referred to simply as HeLa cells. Cells were cultured in Dulbecco's modified Eagle's medium with the addition of 100 units/ml penicillin and 100 μg/ml streptomycin, 10% (v/v) fetal calf serum and 0.5 mg/ml G418 as necessary. The cells stably expressing Vps29–GFP, GFP–Vps35 and Hrs1–GFP have been described previously [19,25]. Cells stably expressing GFP–Snx3, GFP–Snx1 and GFP–Fam21 were generated by transfecting HeLa cells with the appropriate GFP-tagged construct in the pIRESneo2 vector (Clontech) using Effectene (Qiagen) according to the manufacturer's instructions. Transfectants were selected using G418 (Gibco) at 0.5 mg/ml and colonies that were G418 resistant were screened for expression of the respective construct by fluorescence or immunofluorescence microscopy.

Production of the GFP-tagged constructs

In the present study, constructs to express GFP–Snx3, GFP–Snx1 and GFP–Fam21-head or -tail were generated. Other GFP-tagged constructs have been described previously [25]. An IMAGE consortium [Integrated Molecular Analysis of Genomes and their Expression consortium (at St. Louis, MO, U.S.A., and at the Human Genome Mapping Project, Hinxton Hall, Cambridge, U.K.)] EST (expressed sequence tag) containing full-length human Snx3 was obtained from GeneService and was used as a template for a Pfu-mediated PCR reaction to amplify Snx3 and add BamHI and SalI sites to the 5′- and 3′-ends respectively. The PCR product was cloned first using pCRblunt (Invitrogen) and then sequenced to confirm the identity and fidelity of the PCR reaction product. The Snx3 ORF (open reading frame) was then subcloned into pEGFP-C1 at the BglII and SalI sites. The GFP–Snx3 construct was excised from pEGFP-C1 by digestion with NheI–BamHI and then cloned into pIRESneo2 that had also been digested with NheI and BamHI.

The Fam21–GFP construct was described previously [25]. The Fam21-head and -tail regions were amplified by Pfu-mediated PCR incorporating BamHI and SalI sites at the 5′- and 3′-ends respectively and then cloned into pCRblunt. Following sequencing, the fragments were subcloned into pEGFP-C1 by digestion with BglII and SalI and then subsequently subcloned into pIRESneo2 by excision with NheI and BamHI.

The GFP–Snx1 construct was generated by subcloning the murine Snx1 coding region (except the start methionine) from pGEX [12] into pEGFP C1. The GFP–Snx1 construct was then subcloned into pIRESneo2.

Y2H (yeast two-hybrid) interaction assays

These were performed essentially as described previously [25]. In some instances, constructs were generated by subcloning from pEGFP-C1 into versions of the pGBT9 (‘bait’) and pGAD424 (‘prey’) vectors that had been modified so that the multi-cloning sites of the two vectors were the same reading frame as pEGFP-C1. All constructs used in the Y2H assays were sequenced to confirm that an in-frame fusion was made.

Automated microscopy and cell spreading assay

Flasks of HeLa cells or HeLa cells expressing GFP-tagged Vps35 or Fam21-tail were trypsinized and seeded on to 24-well plates (Greiner) at comparable density, using up to four wells per cell line. After 75, 150, 250 or 480 min incubation in a 37°C humidified incubator, cells were fixed at room temperature (20°C) with 4% (w/v) paraformaldehyde, permeabilized with 0.1% Triton X-100 and stained with Hoechst 33258 and Alexa Fluor® 594-phalloidin. Cell spreading was measured using a Thermo Fisher Cellomics Arrayscan automated microscope, using its cell spreading assay protocol and a ×10 objective lens. For each experiment, 1000 single cells were measured in four replicate wells at each time point for the cells expressing GFP-tagged constructs; for the HeLa cells, 800 cells were measured in duplicate wells. The average single-cell perimeter per well was obtained from Cellomics vHSC software, whereas further averaging and normalization was performed using Origin 8.1 (OriginLab).

Microscopy and bafilomycin treatment

Immunofluorescence microscopy was performed essentially as described previously [25]. For the bafilomycin treatment experiments described in the present study, a 200× stock solution of bafilomycin was made by dissolving 10 μg in 800 μl of DMSO. This was divided into aliquots and stored at −20°C. A 1000× stock solution of chloroquine was made up in water and stored at −80°C.

HeLa cells were seeded on to coverslips 24 h prior to the experiment. At 1 h after replacing the medium with fresh medium, the cells were incubated with 100 nM bafilomycin or 100 μM chloroquine for 4 h at 37°C. Control cells were treated with just DMSO. At the end of the incubation period, the cells were briefly washed with PBS and then fixed for 10 min at room temperature with 4% paraformaldehyde. The cells were then permeabilized with 0.1% Triton X-100 in PBS and then blocked with 3% BSA in PBS prior to incubation with primary and secondary antibodies. After mounting the coverslips using ProLong Anti-fade Gold (Invitrogen), the cells were viewed and imaged using a Zeiss Axioplan epifluorescence microscope operating with a Hamamatsu C10600 CCD camera (charge-coupled-device camera). A total of 100 cells were blind scored for Snx1 tubules in each of three independent experiments.

SDS/PAGE and Western blotting

SDS/PAGE was performed as described previously [12]. Western blotting employed 125I-labelled Protein A for detection and was performed as described previously [12]. Signals were obtained by a phosphoimager (using a GE Typhoon Imager) or by exposure to X-ray film.

Native immunoprecipitation and MS

Native immunoprecipitations were initially performed (in Figures 2B and 3A) as described previously [24] using the PBS+1% Triton X-100 lysis buffer. Subsequently (for Figures 3B–3D) the buffer was changed to 20 mM Hepes/KOH (pH 7.0), 50 mM potassium acetate, 1 mM EDTA, 200 mM sorbitol and 0.1% Triton X-100, but the methodology was identical.

MS analysis of gel bands was performed as described previously [19].

Cell fractionation assay

The assay to separate membrane-associated (pelletable – P) proteins from cytosolic (soluble – S) proteins was performed as described previously [19].

RESULTS

The interactions of Vps35 and Fam21

The Vps35, Vps29 and Vps26 proteins assemble to form the cargo-selective retromer complex that is depicted schematically in Figure 1(A). To better understand the interactions of Vps35, we performed a systematic Y2H-based assay for protein–protein interactions. In Figure 1(B), we show that full-length Vps35 binds to Vps26, Vps29, Fam21 and Snx3. The interactions with Vps26 and Vps29 act as positive controls in the assay. The interaction with Fam21 is consistent with our previously published results [25]. The Vps35–Snx3 interaction is novel, although an association between Snx3 and retromer has been demonstrated by native immunoprecipitation [34,35].

Interactions of Vps35 and Fam21

Figure 1
Interactions of Vps35 and Fam21

(A) A simple schematic diagram showing the Vps35 protein and the approximate regions where Vps26 and Vps29 bind. The association of Vps26 with Vps35 requires the PRLYL motif at residue ~100 in Vps35 [45]. (B) Vps35 in the Y2H ‘bait’ vector (pGBT9) was tested for interactions with a number of retromer and retromer-associated proteins. Interactions were observed for Vps26, Vps29, Fam21 and Snx3. Truncation of Vps35 abolished the interaction with Fam21 but interactions with Vps26A and Vps29 were retained. (C and D) Full-length Fam21 or the head or tail domains were expressed in either the bait vector (pGBT9; C) or prey vector (pGAD424; D) and tested for interactions with retromer and retromer-associated proteins. Expression of the head or tail domains separately resulted in strong auto-activating activity for the head and weaker auto-activating activity for the tail. In the case of the Fam21-head domain, it was necessary to conduct the Y2H assays in the presence of 3-amino-1,2,4-triazole (3-AT) to reduce the growth from auto-activation. The Vps35–Fam21 interaction is mediated through the tail of Fam21, whereas the head is responsible for binding to KIAA1033/SWIP.

Figure 1
Interactions of Vps35 and Fam21

(A) A simple schematic diagram showing the Vps35 protein and the approximate regions where Vps26 and Vps29 bind. The association of Vps26 with Vps35 requires the PRLYL motif at residue ~100 in Vps35 [45]. (B) Vps35 in the Y2H ‘bait’ vector (pGBT9) was tested for interactions with a number of retromer and retromer-associated proteins. Interactions were observed for Vps26, Vps29, Fam21 and Snx3. Truncation of Vps35 abolished the interaction with Fam21 but interactions with Vps26A and Vps29 were retained. (C and D) Full-length Fam21 or the head or tail domains were expressed in either the bait vector (pGBT9; C) or prey vector (pGAD424; D) and tested for interactions with retromer and retromer-associated proteins. Expression of the head or tail domains separately resulted in strong auto-activating activity for the head and weaker auto-activating activity for the tail. In the case of the Fam21-head domain, it was necessary to conduct the Y2H assays in the presence of 3-amino-1,2,4-triazole (3-AT) to reduce the growth from auto-activation. The Vps35–Fam21 interaction is mediated through the tail of Fam21, whereas the head is responsible for binding to KIAA1033/SWIP.

Truncation of Vps35 so that the Vps29-binding domain is separated from the rest of the protein indicated that Snx3 binds to Vps35 at a site distinct from that of Vps29, but Fam21 interacts only with full-length Vps35. Additional truncation of Vps35 to define the Snx3 and Fam21 interaction sites was not possible due to markedly increased auto-activation of the truncated Vps35 constructs (results not shown). As the Vps35–Snx3 interaction and also the Vps35–Fam21 interaction were only observed when Vps35 was expressed from the pGBT9 ‘bait’ plasmid, further investigation of the interactions of Vps35 using the Y2H system was not possible.

In addition to interacting with Snx3, Vps35 also binds to the WASH complex protein Fam21. As stated previously, Fam21 comprises a predicted globular ‘head’ domain of ~200 amino acids and a large (~1100 amino acids) unstructured ‘tail’. We therefore set out to investigate the interactions of the head and tail of Fam21 and determine whether the Vps35–Fam21 interaction occurs via one domain or another. Using the Y2H system we tested the Fam21 head and tail against the retromer proteins and several retromer-interacting proteins. In Figure 1(C), full-length Fam21 protein demonstrates a robust interaction with KIAA1033/SWIP. The Fam21–KIAA1033 interaction is mediated exclusively through the Fam21-head domain, as no interaction was observed for the Fam21-tail with KIAA1033/SWIP. The Fam21-head domain is also able to bind to WASH1 and can weakly interact with both Snx1 and Snx2.

The Fam21-tail interacts with Strumpellin, FKBP15 and the actin capping protein CAPZa {CAPZ [capping protein (actin filament) muscle Z-line], α}, interactions that are not observable for full-length Fam21. By conducting the Y2H assay with the Fam21 constructs expressed from the pGAD424 ‘prey’ vector, auto-activation effects could be avoided, and in this configuration we observe a robust interaction between Fam21 and Vps35 that appears to be primarily mediated through the tail of Fam21, whereas the head-domain of Fam21 demonstrates an interaction with KIAA1033/SWIP (see Figure 1D).

Although we focused primarily on the interactions of Fam21 and Vps35, we also examined the interactions of WASH1, FKBP15 and Strumpellin. In Supplementary Figure S1 (at http://www.BiochemJ.org/bj/442/bj4420209add.htm) we show that WASH1 interacts strongly with KIAA1033/SWIP and also with Vps35. FKBP15 interacts strongly with itself and also with a number of other proteins. Conversely, Strumpellin demonstrated an interaction with KIAA1033/SWIP only.

Localization and in vivo interactions of the Fam21-tail

We next investigated whether the Fam21-head or -tail domain mediate the endosomal localization of the Fam21 protein. GFP-tagged Fam21-head or -tail constructs were generated and transiently transfected into HeLa cells. In Figure 2(A), we show that the GFP–Fam21-tail construct is sufficient for endosomal localization and co-localizes with Vps26, similar to the full-length Fam21–GFP construct. The GFP–Fam21-head domain appears to be cytoplasmic.

The tail of Fam21 is sufficient for targeting to endosomes

Figure 2
The tail of Fam21 is sufficient for targeting to endosomes

(A) HeLa cells were transiently transfected with GFP-tagged constructs of full-length Fam21, or just the head or tail. Both the full-length and the GFP-tail construct targeted to endosomes and co-localized with Vps26 (indicated with arrowheads). Scale bar, 20 μm. (B) A truncated Fam21–GFP construct co-immunoprecipitates FKBP15. HeLa cells expressing either Vps29–GFP or Fam21–GFP (a truncated construct) were lysed and treated with anti-GFP. The immunoprecipitations were analysed by SDS/PAGE, MS and Western blotting. FKBP15 was detected in both immunoprecipitations but more strongly in the immunoprecipitation from Fam21–GFP expressing cells. (C) A schematic diagram showing where the Fam21–GFP construct has been truncated post-transfection and where peptides detected by MS originate from.

Figure 2
The tail of Fam21 is sufficient for targeting to endosomes

(A) HeLa cells were transiently transfected with GFP-tagged constructs of full-length Fam21, or just the head or tail. Both the full-length and the GFP-tail construct targeted to endosomes and co-localized with Vps26 (indicated with arrowheads). Scale bar, 20 μm. (B) A truncated Fam21–GFP construct co-immunoprecipitates FKBP15. HeLa cells expressing either Vps29–GFP or Fam21–GFP (a truncated construct) were lysed and treated with anti-GFP. The immunoprecipitations were analysed by SDS/PAGE, MS and Western blotting. FKBP15 was detected in both immunoprecipitations but more strongly in the immunoprecipitation from Fam21–GFP expressing cells. (C) A schematic diagram showing where the Fam21–GFP construct has been truncated post-transfection and where peptides detected by MS originate from.

To further investigate the interactions of the Fam21 protein, we sought to generate a cell line stably expressing Fam21–GFP. After repeated unsuccessful attempts, a GFP-positive cell line was obtained in which ~50% of the cells expressed a GFP-tagged protein. This cell line was used in a native immunoprecipitation experiment applying methodology we have used recently [25]. In Figure 2(B), the proteins precipitated with anti-GFP from cells expressing either Fam21–GFP or Vps29–GFP were analysed by SDS/PAGE, MS and Western blotting. Vps29–GFP co-immunoprecipitated other retromer proteins and also members of the WASH complex. The Fam21–GFP protein was observed to be much smaller than expected (the expected size on SDS/PAGE is ~220 kDa) and relatively few tryptic peptides from Fam21 were identified by MS compared with endogenous Fam21 in the Vps29–GFP sample (see Figure 2C). The truncated Fam21 did, however, associate with FKBP15. The interaction between Fam21 and FKBP15 was confirmed by Western blotting of samples similar to those analysed by SDS/PAGE. In the experiments presented below, the cells expressing the truncated Fam21–GFP are designated Fam21–GFP(T).

The total absence of peptides from the N-terminal half of Fam21 in the sample from the Fam21–GFP(T) cells is inconsistent with some form of post-lysis proteolysis. Also, as several attempts at generating a cell line stably expressing the Fam21–GFP construct were required before the Fam21–GFP(T) cell line was obtained, we surmised that the truncation of the Fam21–GFP construct occurred stochastically at the time the construct integrated into the HeLa cell DNA. Attempts to define the N-terminus of the truncated Fam21–GFP protein by Edman degradation were unsuccessful due to the N-terminus being blocked. It is, however, possible to assign the likely start methionine residue with some confidence as there are relatively few methionine residues in the Fam21 protein and the only methionine residue that could be employed as a start methionine residue to produce the truncated protein detected is at position 801 and is shown in Figure 2(C).

As the use of full-length Fam21–GFP to generate a stably transfected cell line was problematic, we next used the GFP–Fam21-tail and GFP–Fam21-head constructs to generate stably expressing cell lines. We were repeatedly unsuccessful in producing a GFP–Fam21-head-expressing cell line, but we did obtain several different cell lines expressing the GFP–Fam21-tail construct. These cell lines, designated GFP–Fam21-tail(A–D), were used in native immunoprecipitation/MS experiments along with positive and negative control cell lines.

Attempts to demonstrate an in vivo interaction between the GFP–Fam21-tail and retromer were initially unreliable when we used our standard lysis buffer comprising PBS and 1% Triton X-100, although the GFP–Fam21-tail construct was able to reliably co-immunoprecipitate the actin capping proteins CAPZa and CAPZb (CAPZ, β) (see Figure 3A). We therefore switched to using a Hepes/potassium acetate buffer with 0.1% Triton X-100. The results of the large-scale native immunoprecipitation experiments are shown in Figures 3(A) and 3(B). The GFP–Fam21-tail cell lines A–C all express the full-length Fam21-tail at varying levels. The GFP–Fam21-tail(D) cell line, however, expressed a truncated form that constituted approximately one-fifth of the Fam21-tail (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/442/bj4420209add.htm).

In vivo interactions of the Fam21-tail

Figure 3
In vivo interactions of the Fam21-tail

(A) HeLa cells stably transfected with various GFP constructs were lysed in PBS+1% Triton X-100 (as we have used previously [25]) and the GFP-tagged proteins were immunoprecipitated using anti-GFP. After SDS/PAGE, bands were excised and the proteins were identified using MS. The actin-capping proteins CAPZa and CAPZb are readily detected in the lanes containing full-length Fam21-tail or the truncated form characterized in Figure 2(B). (B) The same as in (A), but the lysis buffer was 20 mM Hepes/KOH (pH 7.0), 50 mM potassium acetate, 1 mM EDTA, 200 mM sorbitol and 0.1% Triton X-100. The negative control was untransfected HeLa cells. Retromer proteins are readily detected in the GFP–Fam21-tail(A-C) lanes (i.e. full-length Fam21-tail). Two proteins of unknown function, CCDC22 and CCDC93, are also detected in the Fam21-tail lanes. (C) Similar samples to those shown in (B) were analysed by Western blotting. (D) Similar lysates to those in (B) and (C) were treated with anti-Vps26 to immunoprecipitate retromer. The immunoprecipitations were analysed by Western blotting and showed that increased Fam21-tail inhibits the association of endogenous WASH complex with retromer. (E) Lysates similar to those for (C) and (D) were analysed by Western blotting. The levels of some proteins (e.g. FKBP15) vary somewhat between samples but differences in the levels do not account for the levels detected in (C) and (D). (F) Lysates similar to those in (A) were treated with antisera against WASH1. The expression of the GFP–Fam21-tail does not affect the assembly of the WASH complex.

Figure 3
In vivo interactions of the Fam21-tail

(A) HeLa cells stably transfected with various GFP constructs were lysed in PBS+1% Triton X-100 (as we have used previously [25]) and the GFP-tagged proteins were immunoprecipitated using anti-GFP. After SDS/PAGE, bands were excised and the proteins were identified using MS. The actin-capping proteins CAPZa and CAPZb are readily detected in the lanes containing full-length Fam21-tail or the truncated form characterized in Figure 2(B). (B) The same as in (A), but the lysis buffer was 20 mM Hepes/KOH (pH 7.0), 50 mM potassium acetate, 1 mM EDTA, 200 mM sorbitol and 0.1% Triton X-100. The negative control was untransfected HeLa cells. Retromer proteins are readily detected in the GFP–Fam21-tail(A-C) lanes (i.e. full-length Fam21-tail). Two proteins of unknown function, CCDC22 and CCDC93, are also detected in the Fam21-tail lanes. (C) Similar samples to those shown in (B) were analysed by Western blotting. (D) Similar lysates to those in (B) and (C) were treated with anti-Vps26 to immunoprecipitate retromer. The immunoprecipitations were analysed by Western blotting and showed that increased Fam21-tail inhibits the association of endogenous WASH complex with retromer. (E) Lysates similar to those for (C) and (D) were analysed by Western blotting. The levels of some proteins (e.g. FKBP15) vary somewhat between samples but differences in the levels do not account for the levels detected in (C) and (D). (F) Lysates similar to those in (A) were treated with antisera against WASH1. The expression of the GFP–Fam21-tail does not affect the assembly of the WASH complex.

Using the Hepes-based lysis buffer, we detected retromer, FKBP15 and the actin-capping proteins CAPZa and CAPZb in the immunoprecipitations from cells expressing the full-length GFP–Fam21-tail. Additionally, two proteins of unknown function, CCDC22 (coiled-coil-domain-containing 22) and CCDC93, were detected associated with GFP–Fam21-tail. These proteins were largely absent in the lane from the GFP–Fam21-tail(D) cell line. Interactions detected by MS were confirmed by Western blotting (see Figure 3C).

The increased expression of the Fam21-tail is able to compete with the endogenous WASH complex for retromer binding. As shown in Figure 3(D), Vps26 co-immunoprecipitates the WASH complex from HeLa cells and other cell lines that do not express the full-length Fam21-tail, but the presence of Strumpellin and endogenous Fam21 in the immunoprecipitations from the GFP–Fam21-tail(A–C) cell lines was much reduced, particularly when compared with the levels of Vps26 and Vps35 in those immunoprecipitations. Protein levels in the lysates of the samples shown in Figures 3(C) and 3(D) were examined by Western blotting and are shown in Figure 3(E).

The reduction in the association of endogenous WASH complex with retromer in the GFP–Fam21-tail expressing cells is not due to effects on the WASH complex assembly as incubation of lysates with anti-WASH sera was able to recover very similar levels of Strumpellin and Fam21 (Figure 3F). This is consistent with the Fam21-head domain mediating the interaction with KIAA1033/SWIP and assembling into the WASH complex, whereas the tail binds to retromer (through Vps35), FKBP15 and the actin-capping proteins.

The cargo-selective retromer complex mediates the localization of Fam21

The tail of Fam21 can bind Vps35 and is sufficient to target to endosomes (see Figures 2 and 3). As shown in Supplementary Figure S3(A) (at http://www.BiochemJ.org/bj/442/bj4420209add.htm), in the GFP–Fam21-tail(B) cell line we observe that the GFP–Fam21-tail is localized to Vps26-positive endosomes. The GFP–Fam21-tail(A) and (C) cell lines demonstrated a similar degree of co-localization (results not shown), but the GFP–Fam21-tail(D) cell line that expresses the truncated Fam21-tail did not exhibit any co-localization with retromer proteins and the construct was instead cytoplasmic and also localized to the nucleus (Supplementary Figure S3B).

Although the tail of Fam21 is sufficient to target to endosomes, the Fam21-tail retains a requirement for retromer to be membrane associated. In Supplementary Figures S3(C) and S3(D), siRNA KD (knockdown) of Vps26 or Vps35 resulted in no membrane-associated GFP–Fam21-tail. The siRNA KD of FKBP15, Strumpellin or endogenous Fam21, on the other hand, did not cause the GFP–Fam21-tail protein to become cytosolic.

Fam21 mediates the localization of FKBP15

The tail of Fam21 binds the FKBP15 protein. We would therefore predict that overexpression of the Fam21-tail will displace FKBP15 from the membrane. In Figure 4(A), cells expressing GFP–Fam21-tail (cell line B) were mixed with untransfected HeLa cells and seeded on to coverslips. In cells expressing GFP–Fam21-tail, the punctate localization of FKBP15 is much reduced and the protein appears more cytosolic. The localization of WASH1 was also affected, although to a lesser degree, whereas the Vps26 localization did not appear to be affected.

The Fam21-tail mediates the membrane association of FKBP15

Figure 4
The Fam21-tail mediates the membrane association of FKBP15

(A) The GFP–Fam21-tail(B) cells were mixed with untransfected HeLa cells and seeded on to coverslips. After 24 h, the cells were fixed and labelled with antibodies against GFP and either anti-Vps26, anti-FKBP15 or anti-WASH1. Overexpression of the Fam21-tail does not affect the localization of Vps26 but localization of both FKBP15 and WASH1 is affected, appearing less punctate and more diffuse in the GFP–Fam21-tail(B) cells. Scale bar, 20 μm. (B) Cells were fractionated into membrane (pelletable – P) and cytosolic (soluble – S) fractions and the samples were subjected to SDS/PAGE and analysed by Western blotting. FKBP15 and (to a lesser extent) Strumpellin are shifted into the cytosolic fraction in both cell lines expressing the GFP–Fam21-tail construct. (C and D) Results from triplicate experiments were quantified and the level of membrane-associated FKBP15 (C) and Strumpellin (D) are shown. (E) Cells treated with various siRNAs were fractionated into membrane (pelletable – P) and cytosolic (soluble – S) fractions and the samples were subjected to SDS/PAGE and analysed by Western blotting. FKBP15 becomes cytosolic after Vps26 or Fam21 KD. (F and G) Results from triplicate experiments were quantified and the level of membrane-associated FKBP15 (F) and Strumpellin (G) are shown.

Figure 4
The Fam21-tail mediates the membrane association of FKBP15

(A) The GFP–Fam21-tail(B) cells were mixed with untransfected HeLa cells and seeded on to coverslips. After 24 h, the cells were fixed and labelled with antibodies against GFP and either anti-Vps26, anti-FKBP15 or anti-WASH1. Overexpression of the Fam21-tail does not affect the localization of Vps26 but localization of both FKBP15 and WASH1 is affected, appearing less punctate and more diffuse in the GFP–Fam21-tail(B) cells. Scale bar, 20 μm. (B) Cells were fractionated into membrane (pelletable – P) and cytosolic (soluble – S) fractions and the samples were subjected to SDS/PAGE and analysed by Western blotting. FKBP15 and (to a lesser extent) Strumpellin are shifted into the cytosolic fraction in both cell lines expressing the GFP–Fam21-tail construct. (C and D) Results from triplicate experiments were quantified and the level of membrane-associated FKBP15 (C) and Strumpellin (D) are shown. (E) Cells treated with various siRNAs were fractionated into membrane (pelletable – P) and cytosolic (soluble – S) fractions and the samples were subjected to SDS/PAGE and analysed by Western blotting. FKBP15 becomes cytosolic after Vps26 or Fam21 KD. (F and G) Results from triplicate experiments were quantified and the level of membrane-associated FKBP15 (F) and Strumpellin (G) are shown.

Using a simple fractionation assay to separate membranes from the cytosol, we observe a clear shift of FKBP15 from the membrane fraction (P) to the cytosolic fraction (S) in the cell lines expressing GFP–Fam21-tail (see Figures 4B and 4C), but other cell lines expressing GFP-tagged proteins (e.g. GFP–Snx1) do not elicit the same effect. Additionally, as the Fam21-tail can bind to retromer through Vps35, it is able to compete with the endogenous WASH complex for retromer binding and results in a partial shift of the WASH complex into the cytosolic fraction as more Strumpellin was detected in the soluble cytosolic fraction in the cell lines expressing the full-length GFP–Fam21-tail (see Figures 4B and 4D).

Although the overexpression of the Fam21-tail causes more FKBP15 and WASH complex to become cytosolic, the effect is not as pronounced as observed after siRNA KD of the Vps26 protein. In Figures 4(E)–4(G), the amount of membrane-associated FKBP15 and WASH complex (Strumpellin) was quantitatively determined for the KDs of Vps26, FKBP15 or Fam21. The siRNA KD of either retromer (Vps26) or Fam21 results in a shift of FKBP15 into the cytosolic fraction. Additionally, as shown in Supplementary Figure S4 (http://www.BiochemJ.org/bj/442/bj4420209add.htm), cells treated with siRNA to KD the expression of Vps26, FKBP15 and Fam21 were fixed and labelled with antibodies against Vps26, FKBP15 and Fam21 and co-stained with anti-Snx1 antibodies. Labelling of endogenous FKBP15 and Fam21, although not especially strong, showed co-localization with Snx1. The loss of Vps26 expression results in both FKBP15 and Fam21 becoming cytosolic. KD of FKBP15 does not, however, affect Vps26 or Fam21 localization, but KD of Fam21 does abolish the localization of FKBP15. Therefore FKBP15 localization to the endosome requires Fam21, consistent with the interaction observed between the tail of Fam21 and FKBP15.

Inhibition of the V-ATPase also displaces FKBP15

Recent studies in Dictyostelium have reported that the WASH complex regulates the luminal pH of endo-/lyso-somes by mediating the sorting of the V-ATPase from lysosomes [30]. We therefore asked whether the activity of the V-ATPase can influence the functioning of the WASH complex using bafilomycin to inhibit the V-ATPase. Following treatment with bafilomycin, we observed that cells exhibited markedly less endosomally localized FKBP15. Comparison of the labelling shown in Figures 5(A) and 5(B) reveals that cells treated with bafilomycin have little punctate FKBP15. Treatment with chloroquine, a weak base that neutralizes endosomal and lysosomal pH, does not produce the same effect (Figure 5C), indicating that the loss of FKBP15 localization is the result of inhibition of the V-ATPase and not the result of perturbation of the pH gradient across the endosomal membrane. We quantified the effect of bafilomycin on the level of membrane-associated FKBP15 and observed a pronounced shift (~2-fold) of FKBP15 into the cytosolic fraction after treatment with bafilomycin (see Figure 5D).

Inhibition of V-ATPase redistributes FKBP15 into the cytoplasm

Figure 5
Inhibition of V-ATPase redistributes FKBP15 into the cytoplasm

(AC) HeLa cells were treated with just DMSO (A), 100 nM bafilomycin (B) or 100 μM chloroquine (C) for 4 h. After fixation the cells were labelled with antibodies against Snx1 and either anti-Vps26, anti-FKBP15 or anti-Fam21. Incubation with bafilomycin results in FKBP15 redistributing to the cytoplasm, but chloroquine treatment does not. Scale bar, 20 μm. (D) Control or bafilomycin-treated cells were fractionated into membrane (pelletable – P) and cytosolic (soluble – S) fractions and the samples were subjected to SDS/PAGE and analysed by Western blotting. FKBP15 becomes cytosolic after incubation with bafilomycin. Results from triplicate blots were quantified and graphed.

Figure 5
Inhibition of V-ATPase redistributes FKBP15 into the cytoplasm

(AC) HeLa cells were treated with just DMSO (A), 100 nM bafilomycin (B) or 100 μM chloroquine (C) for 4 h. After fixation the cells were labelled with antibodies against Snx1 and either anti-Vps26, anti-FKBP15 or anti-Fam21. Incubation with bafilomycin results in FKBP15 redistributing to the cytoplasm, but chloroquine treatment does not. Scale bar, 20 μm. (D) Control or bafilomycin-treated cells were fractionated into membrane (pelletable – P) and cytosolic (soluble – S) fractions and the samples were subjected to SDS/PAGE and analysed by Western blotting. FKBP15 becomes cytosolic after incubation with bafilomycin. Results from triplicate blots were quantified and graphed.

Interestingly, we also observed that cells treated with bafilomycin exhibited almost no Snx1-positive tubules (see Supplementary Figure S5 at http://www.BiochemJ.org/bj/442/bj4420209add.htm), although, unlike FKBP15, Snx1 remained associated with punctate endosomal structures positive for the retromer protein Vps26.

Dominant-negative effects of the Fam21-tail on cell spreading

Over the course of these studies, we noticed that the cell lines stably expressing the full-length GFP-tagged Fam21-tail were slow to spread after trypsinization and seeding on to tissue culture-treated plastic. Using an automated microscope, we therefore examined the cell spreading of the four GFP–Fam21-tail cell lines (A–D) along with cells expressing GFP–Vps35 and untransfected HeLa cells. As shown in Figure 6, the GFP–Fam21-tail cell lines A–C all exhibit slower cell spreading when compared with GFP–Vps35, GFP–Fam21-tail(D) and also untransfected HeLa cells. As the GFP–Fam21-tail(D) cell line differs from the GFP–Fam21-tail(A–C) cell lines only by virtue of expressing a truncated form of the Fam21-tail, the lack of cell spreading defect in this cell line suggests that the cell spreading defect in the GFP–Fam21-tail(A–C) cell lines is the result of the full-length constructs titrating a protein (or proteins) into inactive complexes.

Overexpression of the Fam21-tail causes cell spreading defects

Figure 6
Overexpression of the Fam21-tail causes cell spreading defects

(A) Average single cell perimeter of HeLa cells and HeLa cells expressing GFP–Vps35 or GFP–Fam21-tail(A–D) measured at four time points after seeding and normalized to the first time point (75 min post-seeding). Cells expressing full-length GFP–Fam21-tail(A–C) show markedly slower cell spreading kinetics compared with control HeLa cells or HeLa cells expressing GFP–Vps35. This experiment was repeated three times; a representative experiment is shown. (B) Representative images of the phalloidin (F-actin) stain at three time points post-seeding for HeLa control cells, HeLa cells expressing GFP–Vps35 and HeLa cells expressing GFP–Fam21-tail(B). Scale bar, 300 μm. Note the greater proportion of rounder, smaller cells at 8 h for the cells expressing GFP–Fam21-tail(B). For the cell perimeter analysis in (A), only single cells in the images are analysed.

Figure 6
Overexpression of the Fam21-tail causes cell spreading defects

(A) Average single cell perimeter of HeLa cells and HeLa cells expressing GFP–Vps35 or GFP–Fam21-tail(A–D) measured at four time points after seeding and normalized to the first time point (75 min post-seeding). Cells expressing full-length GFP–Fam21-tail(A–C) show markedly slower cell spreading kinetics compared with control HeLa cells or HeLa cells expressing GFP–Vps35. This experiment was repeated three times; a representative experiment is shown. (B) Representative images of the phalloidin (F-actin) stain at three time points post-seeding for HeLa control cells, HeLa cells expressing GFP–Vps35 and HeLa cells expressing GFP–Fam21-tail(B). Scale bar, 300 μm. Note the greater proportion of rounder, smaller cells at 8 h for the cells expressing GFP–Fam21-tail(B). For the cell perimeter analysis in (A), only single cells in the images are analysed.

DISCUSSION

In the present study, we have investigated the interactions of the retromer protein Vps35 and the WASH complex protein Fam21. The Vps35 protein forms part of the cargo-selective retromer subcomplex and, as such, associates with Vps26 and Vps29 via its N- and C-terminal regions respectively [36]. Vps35 has been directly implicated in cargo selection [20], and it is therefore important to elucidate how the interactions of Vps35 contribute to endosomal protein sorting.

Determination of the distinct interactions of the Fam21-head and -tail domains

In addition to identifying a novel Vps35–Snx3 interaction, we also show that Vps35 binds to the unstructured ‘tail’ of the Fam21 protein of the WASH complex (see Figure 1B). These novel interactions were observed using the Y2H system. This proved a useful tool for examining a large number of binary interactions of retromer and associated proteins. In some instances, however, further examination of the interactions using the Y2H system were curtailed due to the auto-activating effects of particular constructs.

Dissecting Fam21 into its head- and tail-domains revealed interactions that were not observed for full-length Fam21. For example, the interactions between the tail of Fam21 and the FKBP15 and Strumpellin proteins were detected when no interactions were observed for full-length Fam21 (see Figures 1C and 1D). Similarly, the Fam21-head domain bound to WASH1 and the Snx1 and Snx2 proteins when no such interaction was observed for the full-length Fam21 protein. It is possible that the head of Fam21 interacts with a region of the tail and that this interaction prevents full-length Fam21 from interacting with FKBP15 and Strumpellin in the Y2H experiments. A Fam21-head–tail interaction could also explain why the head and tail domains were individually auto-activating but full-length Fam21 did not auto-activate.

Attempts to generate a cell line expressing full-length GFP-tagged Fam21 were unsuccessful, although a cell line expressing a truncated version of the Fam21–GFP construct was eventually obtained that revealed an interaction between Fam21 and FKBP15 (see Figure 2B). This interaction had been predicted from the Y2H results and therefore provides a useful in vivo confirmation of the validity of the Y2H results. Although the function of FKBP15 is yet to be determined, it is a protein of interest to researchers studying ulcerative colitis and has also been implicated in neuronal cell function [26,27].

It seems likely that overexpression of the ‘head’ of Fam21 may be toxic, as no cell line stably expressing the GFP–Fam21-head construct could be obtained. The reason for the apparent cytotoxicity of the Fam21-head domain is currently unknown and merits further investigation.

Vps35 is central to the endosomal recruitment of Fam21

When cell lines expressing GFP–Fam21-tail were generated, the tail of Fam21 was found to interact with a number of proteins in addition to FKBP15. The actin-capping proteins CAPZa and CAPZb were readily detected in native immunoprecipitations of the full-length GFP–Fam21-tail construct, as were the retromer proteins Vps35, Vps29 and Vps26. Two proteins of unknown function, CCDC22 and CCDC93, were also detected and, in the case of CCDC22, confirmed by Western blotting. As CCDC22 was detected only in the GFP–Fam21-tail immunoprecipitations and not when retromer components were immunoprecipitated, the CCDC22 protein (and possibly CCDC93) may associate with Fam21-tail only when the Fam21-tail is not bound to retromer (see Figures 3A–3C).

The tail of Fam21 is sufficient for recruitment to the endosomal membrane, a process that requires the retromer cargo-selective complex (see Supplementary Figure S3). In vivo, an interaction between Vps35 and the WASH1 protein (revealed in the Y2H experiment) may also contribute to the membrane association of the WASH complex. Indeed, the inability to co-immunoprecipitate retromer with the GFP–Fam21-tail under conditions (namely PBS+1% Triton X-100) in which GFP–Vps35 or Vps29–GFP are able to co-immunoprecipitate endogenous WASH complex may be explained by the absence of the contribution of the Vps35–WASH1 interaction when native immunoprecipitations of GFP–Fam21-tail constructs were initially performed. The loss of the Vps35–WASH1 interaction therefore necessitated the use of the lower stringency Hepes/potassium acetate buffer that was able to retain the retromer–Fam21-tail interaction. Alternatively, the Vps35–WASH1 interaction might be involved in regulating the activity of the WASH1 protein. In either case, the central role that Vps35 plays in mediating the recruitment of the WASH complex could explain why loss of Vps35 function in Drosophila results in extensive actin dysregulation along with defects in endocytosis [37].

The elevated levels of Fam21-tail in the cell lines expressing the GFP–Fam21-tail construct disrupted the association of endogenous WASH complex with retromer, resulting in an increased cytoplasmic localization of WASH1 and Strumpellin (see Figures 3D and 4A–4D). This observation significantly extends our original finding that retromer mediates recruitment of the WASH complex, which was based on phenotypes observed after siRNA KD of retromer proteins [25]. Additionally, because no direct interaction has been observed between WASH1 and the Fam21-tail domain, these results suggest that the WASH complex is more likely to be a stable entity in which the Strumpellin, KIAA1033, WASH1 and Fam21 proteins are recruited to the endosome as a single unit.

As the overexpression of the Fam21-tail was also able to induce the displacement of FKBP15 from the membrane, we suggest that the interaction between Fam21 and FKBP15 may be the driving force in mediating the membrane association of FKBP15, a hypothesis supported by the loss of FKBP15 membrane association after Fam21 KD (see Figure 4).

A mechanistic link between endosomal acidification and protein sorting

Studies in D. discoideum have shown that the WASH complex is required to mediate the trafficking of the V-ATPase [30]. Results presented in the present paper demonstrate that FKBP15 requires Fam21 for its membrane association (see Figure 4). Intriguingly, we observed that inhibition of V-ATPase activity by bafilomycin also markedly reduces the membrane association of FKBP15, an effect similar to that observed when the Fam21-tail is overexpressed (see Figures 4 and 5). The coupling of the V-ATPase activity to the association of FKBP15 with endosomes (and thereby the WASH complex) could provide a mechanism to ensure that FKBP15 is only recruited to the membrane when the V-ATPase is active and the endosome has achieved a measure of maturity. At this time, however, the precise role of FKBP15 in regulating WASH complex function, or other aspects of endosomal protein sorting, remains to be determined.

In addition to the displacement of FKBP15 from the membrane after bafilomycin treatment, we also observed an almost total loss of Snx1 tubules (see Supplementary Figure S5). We do not believe, however, that loss of endosomally localized FKBP15 leads to a defect in Snx1-tubule formation as we have not observed any effect on Snx1-tubules or endosome-to-Golgi retrieval after FKBP15 KD [25]. A block in Snx1-tubule formation after inhibition of the V-ATPase is potentially highly significant as it may provide a regulatory mechanism for ensuring that Snx1-tubule formation does not occur too early in the endocytic pathway. This observation may also explain the apparent defect in endosome-to-Golgi retrieval after pharmacological abolishment of endosomal acidity [38].

Assembly and physiological role of the WASH complex

The dominant-negative effect(s) resulting from the overexpression of the Fam21-tail did not extend to assembly of the WASH complex as there was no apparent defect in the ability of WASH1 to co-immunoprecipitate the Strumpellin or endogenous Fam21 proteins, indicating that the WASH complex was intact and that the assembly of Fam21 into the WASH complex is most likely to be mediated by the interaction of the Fam21-head domain with the KIAA1033/SWIP protein (see Figures 1C, 1D and 3F). The results presented here are therefore complementary to the in vitro structural studies of WASH complex assembly [31] and confirm that the assembly of the WASH complex in vivo is similar to the assembly of the WAVE complex. A schematic diagram of the WASH complex and the analogous WAVE complex is shown in Figure 7(A) and the interactions of the various Fam21 constructs are summarized in Figure 7(B).

Schematic diagrams of the WASH complex and its role in endosomal protein sorting

Figure 7
Schematic diagrams of the WASH complex and its role in endosomal protein sorting

(A) Schematic diagram of the WASH and WAVE complexes showing similar overall architecture (based on Jia et al. [31] and the results of the present study). A key difference is the very long Fam21-tail domain that is not present in the orthologous WAVE protein, Abi. (B) Schematic diagram summarizing the different Fam21 constructs analysed in the present study and the interactions detected for the respective constructs. (C) A model for the action of the WASH complex and retromer in regulating endosomal protein sorting. Recruitment of the WASH complex by retromer facilitates WASH complex-mediated sorting of the V-ATPase and other membrane proteins.

Figure 7
Schematic diagrams of the WASH complex and its role in endosomal protein sorting

(A) Schematic diagram of the WASH and WAVE complexes showing similar overall architecture (based on Jia et al. [31] and the results of the present study). A key difference is the very long Fam21-tail domain that is not present in the orthologous WAVE protein, Abi. (B) Schematic diagram summarizing the different Fam21 constructs analysed in the present study and the interactions detected for the respective constructs. (C) A model for the action of the WASH complex and retromer in regulating endosomal protein sorting. Recruitment of the WASH complex by retromer facilitates WASH complex-mediated sorting of the V-ATPase and other membrane proteins.

Although we did not observe any profound changes in the localization of several membrane proteins [e.g. the transferrin receptor, the CIMPR and LAMP-1 (lysosome-associated membrane protein-1)] in cells expressing the GFP–Fam21-tail (results not shown), we did observe a significant cell spreading defect in all three cell lines expressing the full-length GFP–Fam21-tail. This phenotype was not observed in cells expressing the truncated GFP–Fam21-tail or GFP–Vps35 and therefore is likely to be due to dominant-negative effects exerted by the full-length tail (see Figure 6). Cell spreading following loss of adherence requires the mobilization of endocytosed membrane to expand the area of the plasma membrane [24]. Interestingly, the genes encoding Fam21 and WASH1 homologues in the amoeba Naegleria gruberi (annotated as AM46 and AM5 respectively) have been implicated in amoeboid locomotion due to their absence in related organisms incapable of amoeboid motility [39]. Amoeboid motility is a process that is conceptually similar to the spreading of a trypsinized HeLa cell that is newly adhered to the substratum.

It is also noteworthy that the two proteins of unknown function, CCDC22 and CCDC93, that interact with the Fam21-tail, could be involved in this pathway. The Drosophila melanogaster homologue of CCDC93 has been reported to interact with exo70 [40], a protein of the exocyst complex. CCDC22 has been shown to bind to Copine 1 and 4, calcium-binding proteins that have been implicated in membrane trafficking [41,42]. Additionally, mutations in CCDC22 have recently been identified as causal in X-linked inherited learning disability [43].

We have previously demonstrated that loss of WASH-complex function does not profoundly affect endosome-to-Golgi retrieval of a well-characterized retromer cargo protein, namely the CIMPR [25]. We therefore suggest that the function of the WASH complex is more acutely required for membrane trafficking from endosomes to the cell surface, a pathway that is necessary for cell spreading [24] and recycling of membrane proteins such as the β-adrenergic receptor [44]. The retromer complex mediates the endosomal recruitment of the WASH complex through Vps35 binding to the Fam21-tail domain and thereby contributes to the efficiency of protein sorting into the endosome-to-cell surface pathway (see Figure 7C).

Abbreviations

     
  • BAR

    Bin/amphiphysin/Rvs

  •  
  • CAPZ

    capping protein (actin filament) muscle Z-line

  •  
  • CCDC

    coiled-coil-domain-containing

  •  
  • CIMPR

    cation-independent mannose 6-phosphate receptor

  •  
  • FKBP15

    FK506-binding protein 15

  •  
  • GFP

    green fluorescent protein

  •  
  • GST

    glutathione transferase

  •  
  • HSP

    hereditary spastic paraplegia

  •  
  • KD

    knockdown

  •  
  • siRNA

    small interfering RNA

  •  
  • Snx

    sorting nexin

  •  
  • V-ATPase

    vacuolar ATPase

  •  
  • Vps

    vacuolar protein sorting

  •  
  • WASH

    Wiskott–Aldrich syndrome homologue

  •  
  • WASP

    Wiskott–Aldrich syndrome protein

  •  
  • WAVE

    WASP verprolin homologous

  •  
  • Y2H

    yeast two-hybrid

AUTHOR CONTRIBUTION

Michael Harbour performed the Y2H analysis, generated constructs and cell lines and performed preliminary native immunoprecipitation experiments. Sophia Breusegem performed the quantitative cell spreading assay and Snx1-tubule analysis (in the Supplementary online data). Matthew Seaman performed the immunofluorescence experiments, cell fractionation assays and native immunoprecipitations and wrote the manuscript.

We are grateful to Kamburapola Jayawardena (a.k.a. Jay – ‘the mass-spec-guy’) for MS identification of gel bands.

FUNDING

This work was funded by the Medical Research Council through a Senior Fellowship Award [grant number G0701444 (to M.N.J.S.)] and an additional research grant [grant number G0700750]. Funding for MS in the CIMR was provided by the Wellcome Trust through a Strategic Award [grant number 079895/Z/06].

References

References
1
Jovic
M.
Sharma
M.
Rahajeng
J.
Caplan
S.
The early endosome: a busy sorting station for proteins at the crossroads
Histol. Histopathol.
2010
, vol. 
25
 (pg. 
99
-
112
)
2
Balasubramanian
N.
Scott
D. W.
Castle
J. D.
Casanova
J. E.
Schwartz
M. A.
Arf6 and microtubules in adhesion-dependent trafficking of lipid rafts
Nat. Cell Biol.
2007
, vol. 
9
 (pg. 
1381
-
1391
)
3
Jović
M.
Naslavsky
N.
Rapaport
D.
Horowitz
M.
Caplan
S.
EHD1 regulates β1 integrin endosomal transport: effects on focal adhesions, cell spreading and migration
J. Cell Sci.
2007
, vol. 
120
 (pg. 
802
-
814
)
4
Veale
K. J.
Offenhäuser
C.
Lei
N.
Stanley
A. C.
Stow
J. L.
Murray
R. Z.
VAMP3 regulates podosome organisation in macrophages and together with Stx4/SNAP23 mediates adhesion, cell spreading and persistent migration
Exp. Cell Res.
2011
, vol. 
317
 (pg. 
1817
-
1829
)
5
Rosenbaum
A. I.
Maxfield
F. R.
Niemann–Pick type C disease: molecular mechanisms and potential therapeutic approaches
J. Neurochem.
2011
, vol. 
116
 (pg. 
789
-
795
)
6
Salinas
S.
Proukakis
C.
Crosby
A.
Warner
T. T.
Hereditary spastic paraplegia: clinical features and pathogenetic mechanisms
Lancet Neurol.
2008
, vol. 
7
 (pg. 
1127
-
1138
)
7
Blackstone
C.
O'Kane
C. J.
Reid
E.
Hereditary spastic paraplegias: membrane traffic and the motor pathway
Nat. Rev. Neurosci.
2011
, vol. 
12
 (pg. 
31
-
42
)
8
Burd
C. G.
Physiology and pathology of endosome-to-Golgi retrograde sorting
Traffic
2011
, vol. 
12
 (pg. 
948
-
955
)
9
Toei
M.
Saum
R.
Forgac
M.
Regulation and isoform function of the V-ATPases
Biochemistry
2010
, vol. 
49
 (pg. 
4715
-
4723
)
10
Marshansky
V.
Futai
M.
The V-type H+-ATPase in vesicular trafficking: targeting, regulation and function
Curr. Opin. Cell Biol.
2008
, vol. 
20
 (pg. 
415
-
426
)
11
Arighi
C. N.
Hartnell
L. M.
Aguilar
R. C.
Haft
C. R.
Bonifacino
J. S.
Role of the mammalian retromer in sorting of the cation-independent mannose 6-phosphate receptor
J. Cell Biol.
2004
, vol. 
165
 (pg. 
123
-
133
)
12
Seaman
M. N.
Cargo-selective endosomal sorting for retrieval to the Golgi requires retromer
J. Cell Biol.
2004
, vol. 
165
 (pg. 
111
-
122
)
13
Canuel
M.
Lefrancois
S.
Zeng
J.
Morales
C. R.
AP-1 and retromer play opposite roles in the trafficking of sortilin between the Golgi apparatus and the lysosomes
Biochem. Biophys. Res. Commun.
2008
, vol. 
366
 (pg. 
724
-
730
)
14
Nielsen
M. S.
Gustafsen
C.
Madsen
P.
Nyengaard
J. R.
Hermey
G.
Bakke
O.
Mari
M.
Schu
P.
Pohlmann
R.
Dennes
A.
Petersen
C. M.
Sorting by the cytoplasmic domain of the amyloid precursor protein binding receptor SorLA
Mol. Cell Biol.
2007
, vol. 
27
 (pg. 
6842
-
6851
)
15
Tabuchi
M.
Yanatori
I.
Kawai
Y.
Kishi
F.
Retromer-mediated direct sorting is required for proper endosomal recycling of the mammalian iron transporter DMT1
J. Cell Sci.
2010
, vol. 
123
 (pg. 
756
-
766
)
16
Seaman
M. N.
Recycle your receptors with retromer
Trends Cell Biol.
2005
, vol. 
15
 (pg. 
68
-
75
)
17
Collins
BM.
The structure and function of the retromer protein complex
Traffic
2008
, vol. 
9
 (pg. 
1811
-
1822
)
18
Rojas
R.
van Vlijmen
T.
Mardones
G. A.
Prabhu
Y.
Rojas
A. L.
Mohammed
S.
Heck
A. J.
Raposo
G.
van der Sluijs
P.
Bonifacino
J. S.
Regulation of retromer recruitment to endosomes by sequential action of Rab5 and Rab7
J. Cell Biol.
2008
, vol. 
183
 (pg. 
513
-
526
)
19
Seaman
M. N.
Harbour
M. E.
Tattersall
D.
Read
E.
Bright
N.
Membrane recruitment of the cargo-selective retromer subcomplex is catalysed by the small GTPase Rab7 and inhibited by the Rab-GAP TBC1D5
J. Cell Sci.
2009
, vol. 
122
 (pg. 
2371
-
2382
)
20
Nothwehr
S. F.
Ha
S. A.
Bruinsma
P.
Sorting of yeast membrane proteins into an endosome-to-Golgi pathway involves direct interaction of their cytosolic domains with Vps35p
J. Cell Biol.
2000
, vol. 
151
 (pg. 
297
-
310
)
21
Shi
H.
Rojas
R.
Bonifacino
J. S.
Hurley
J. H.
The retromer subunit Vps26 has an arrestin fold and binds Vps35 through its C-terminal domain
Nat. Struct. Mol. Biol.
2006
, vol. 
13
 (pg. 
540
-
548
)
22
Collins
B. M.
Norwood
S. J.
Kerr
M. C.
Mahony
D.
Seaman
M. N.
Teasdale
R. D.
Owen
D. J.
Structure of Vps26B and mapping of its interaction with the retromer protein complex
Traffic
2008
, vol. 
9
 (pg. 
366
-
379
)
23
Attar
N.
Cullen
P. J.
The retromer complex
Adv. Enzyme Regul.
2010
, vol. 
50
 (pg. 
216
-
236
)
24
Carlton
J.
Bujny
M.
Peter
B. J.
Oorschot
V. M.
Rutherford
A.
Mellor
H.
Klumperman
J.
McMahon
H. T.
Cullen
P. J.
Sorting nexin-1 mediates tubular endosome-to-TGN transport through coincidence sensing of high-curvature membranes and 3-phosphoinositides
Curr. Biol.
2004
, vol. 
14
 (pg. 
1791
-
1800
)
25
Harbour
M. E.
Breusegem
S. Y.
Antrobus
R.
Freeman
C.
Reid
E.
Seaman
M. N.
The cargo-selective retromer complex is a recruiting hub for protein complexes that regulate endosomal tubule dynamics
J. Cell Sci.
2010
, vol. 
123
 (pg. 
3703
-
3717
)
26
Viklund
I. M.
Aspenström
P.
Meas-Yedid
V.
Zhang
B.
Kopec
J.
Agren
D.
Schneider
G.
D'Amato
M.
Olivo-Marin
J. C.
Sansonetti
P.
Van Nhieu
G. T.
Pettersson
S.
WAFL, a new protein involved in regulation of early endocytic transport at the intersection of actin and microtubule dynamics
Exp. Cell Res.
2009
, vol. 
15
 (pg. 
1040
-
1052
)
27
Nakajima
O.
Nakamura
F.
Yamashita
N.
Tomita
Y.
Suto
F.
Okada
T.
Iwamatsu
A.
Kondo
E.
Fujisawa
H.
Takei
K.
Goshima
Y.
FKBP133: a novel mouse FK506-binding protein homolog alters growth cone morphology
Biochem. Biophys. Res. Commun.
2006
, vol. 
346
 (pg. 
140
-
149
)
28
Gomez
T. S.
Billadeau
D. D.
A FAM21-containing WASH complex regulates retromer-dependent sorting
Dev. Cell
2009
, vol. 
17
 (pg. 
699
-
711
)
29
Derivery
E.
Sousa
C.
Gautier
J. J.
Lombard
B.
Loew
D.
Gautreau
A.
The Arp2/3 activator WASH controls the fission of endosomes through a large multiprotein complex
Dev. Cell
2009
, vol. 
17
 (pg. 
712
-
723
)
30
Carnell
M.
Zech
T.
Calaminus
S. D.
Ura
S.
Hagedorn
M.
Johnston
S. A.
May
R. C.
Soldati
T.
Machesky
L. M.
Insall
R. H.
Actin polymerization driven by WASH causes V-ATPase retrieval and vesicle neutralization before exocytosis
J. Cell Biol.
2011
, vol. 
193
 (pg. 
831
-
839
)
31
Jia
D.
Gomez
T. S.
Metlagel
Z.
Umetani
J.
Otwinowski
Z.
Rosen
M. K.
Billadeau
D. D.
WASH and WAVE actin regulators of the Wiskott–Aldrich syndrome protein (WASP) family are controlled by analogous structurally related complexes
Proc. Natl. Acad. Sci. U.S.A.
2010
, vol. 
107
 (pg. 
10442
-
10447
)
32
Hernandez-Valladares
M.
Kim
T.
Kannan
B.
Tung
A.
Aguda
A. H.
Larsson
M.
Cooper
J. A.
Robinson
R. C.
Structural characterization of a capping protein interaction motif defines a family of actin filament regulators
Nat. Struct. Mol. Biol.
2010
, vol. 
17
 (pg. 
497
-
503
)
33
Tiwari
R. K.
Kusari
J.
Sen
G. C.
Functional equivalents of interferon-mediated signals needed for induction of an mRNA can be generated by double-stranded RNA and growth factors
EMBO J.
1987
, vol. 
6
 (pg. 
3373
-
3378
)
34
Strochlic
T. I.
Setty
T. G.
Sitaram
A.
Burd
C. G.
Grd19/Snx3p functions as a cargo-specific adapter for retromer-dependent endocytic recycling
J. Cell Biol.
2007
, vol. 
177
 (pg. 
115
-
125
)
35
Harterink
M.
Port
F.
Lorenowicz
M. J.
McGough
I. J.
Silhankova
M.
Betist
M. C.
van Weering
J. R.
van Heesbeen
R. G.
Middelkoop
T. C.
Basler
K.
, et al. 
A SNX3-dependent retromer pathway mediates retrograde transport of the Wnt sorting receptor Wntless and is required for Wnt secretion
Nat. Cell Biol.
2011
, vol. 
13
 (pg. 
914
-
923
)
36
Hierro
A.
Rojas
A. L.
Rojas
R.
Murthy
N.
Effantin
G.
Kajava
A. V.
Steven
A. C.
Bonifacino
J. S.
Hurley
J. H.
Functional architecture of the retromer cargo-recognition complex
Nature
2007
, vol. 
449
 (pg. 
1063
-
1067
)
37
Korolchuk
V. I.
Schütz
M. M.
Gómez-Llorente
C.
Rocha
J.
Lansu
N. R.
Collins
S. M.
Wairkar
Y. P.
Robinson
I. M.
O'Kane
C. J.
Drosophila Vps35 function is necessary for normal endocytic trafficking and actin cytoskeleton organisation
J. Cell Sci.
2007
, vol. 
120
 (pg. 
4367
-
4376
)
38
Chapman
R. E.
Munro
S.
Retrieval of TGN proteins from the cell surface requires endosomal acidification
EMBO J.
1994
, vol. 
13
 (pg. 
2305
-
2312
)
39
Fritz-Laylin
L. K.
Prochnik
S. E.
Ginger
M. L.
Dacks
J. B.
Carpenter
M. L.
Field
M. C.
Kuo
A.
Paredez
A.
Chapman
J.
Pham
J.
, et al. 
The genome of Naegleria gruberi illuminates early eukaryotic versatility
Cell
2010
, vol. 
140
 (pg. 
631
-
642
)
40
Formstecher
E.
Aresta
S.
Collura
V.
Hamburger
A.
Meil
A.
Trehin
A.
Reverdy
C.
Betin
V.
Maire
S.
Brun
C.
, et al. 
Protein interaction mapping: a Drosophila case study
Genome Res.
2005
, vol. 
15
 (pg. 
376
-
384
)
41
Damer
C. K.
Bayeva
M.
Kim
P. S.
Ho
L. K.
Eberhardt
E. S.
Socec
C. I.
Lee
J. S.
Bruce
E. A.
Goldman-Yassen
A. E.
Naliboff
L. C.
Copine A is required for cytokinesis, contractile vacuole function, and development in Dictyostelium
Eukaryot. Cell
2007
, vol. 
6
 (pg. 
430
-
442
)
42
Tomsig
J. L.
Snyder
S. L.
Creutz
C. E.
Identification of targets for calcium signaling through the copine family of proteins. Characterization of a coiled-coil copine-binding motif
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
10048
-
10054
)
43
Voineagu
I.
Huang
L.
Winden
K.
Lazaro
M.
Haan
E.
Nelson
J.
McGaughran
J.
Nguyen
L. S.
Friend
K.
Hackett
A.
, et al. 
CCDC22: a novel candidate gene for syndromic X-linked intellectual disability
Mol. Psychiatry
2011
 
doi:10.1038/mp.2011.95
44
Temkin
P.
Lauffer
B.
Jäger
S.
Cimermancic
P.
Krogan
N. J.
von Zastrow
M.
SNX27 mediates retromer tubule entry and endosome-to-plasma membrane trafficking of signalling receptors
Nat. Cell Biol.
2011
, vol. 
13
 (pg. 
715
-
721
)
45
Gokool
S.
Tattersall
D.
Reddy
J. V.
Seaman
M. N.
Identification of a conserved motif required for Vps35p/Vps26p interaction and assembly of the retromer complex
Biochem. J.
2007
, vol. 
408
 (pg. 
287
-
295
)

Author notes

1

These authors contributed equally to this work.

Supplementary data