ARFGAP1 Promotes AP-2-Dependent Endocytosis

ARTICLES

ARFGAP1 promotes AP-2-dependent endocytosis

Ming Bai1,10, Helge Gad2,3,10, Gabriele Turacchio2,4,10, Emanuele Cocucci5, Jia-Shu Yang1, Jian Li1, Galina V. Beznoussenko2, Zhongzhen Nie6, Ruibai Luo7, Lianwu Fu8, James F. Collawn8, Tomas Kirchhausen5, Alberto Luini2,9 and Victor W. Hsu1,11

COPI (coat protein I) and the clathrin–AP-2 (adaptor protein 2) complex are well-characterized coat proteins, but a component that is common to these two coats has not been identiﬁed. The GTPase-activating protein (GAP) for ADP-ribosylation factor 1 (ARF1), ARFGAP1, is a known component of the COPI complex. Here, we show that distinct regions of ARFGAP1 interact with AP-2 and coatomer (components of the COPI complex). Selectively disrupting the interaction of ARFGAP1 with either of these two coat proteins leads to selective inhibition in the corresponding transport pathway. The role of ARFGAP1 in AP-2-regulated endocytosis has mechanistic parallels with its roles in COPI transport, as both its GAP activity and coat function contribute to promoting AP-2 transport.

Coat proteins initiate vesicular transport by coupling vesicle formation RESULTS with cargo sorting1,2. COPI, COPII and the clathrin–AP-2 complex ARFGAP1 acts directly in AP-2 transport are the best-characterized coats3–6. Coatomer was initially identified To gain insight into the function of ARFGAP1, we searched for as a component of the COPI complex7; Sec13p, Sec23p, Sec24p and interacting partners. In one approach, we incubated cytosol from Sec31p are the core components of the COPII complex8; and the rat brains with beads that contained a glutathione S-transferase clathrin triskelion combines with the AP-2 adaptor, a hetero-tetrameric (GST) fusion of ARFGAP1 (Fig. 1a) and identified interacting proteins complex, in forming a core complex9,10. Multiple auxiliary components by mass spectrometry. Unexpectedly, we identified components of have been identified for these major coats11–15. However, despite intense the clathrin–AP-2 complex (Supplementary Table S1). Interaction investigations, these complexes have not been observed to share a of ARFGAP1 with components of the clathrin–AP-2 complex was common component. confirmed by a co-precipitation experiment (Fig. 1b). The ARF family of small GTPases regulates the recruitment An interaction between ARFGAP1 and AP-2 had been detected of coat proteins to membrane surfaces16. These small GTPases previously21,22. Notably however, the functional significance of this are regulated, in turn, by guanine nucleotide exchange factors interaction had not been explored. Thus, we assessed whether key (GEFs) that catalyse ARF activation17, and GAPs that catalyse ARF examples of clathrin-dependent endocytosis would be affected by deactivation18. Besides this regulatory role, the best-characterized perturbation of ARFGAP1 levels. We depleted endogenous ARFGAP1 ARF GAPs are also coat components that act as ARF effectors. using short interfering RNA (siRNA; Supplementary Fig. S1a), and This role was first demonstrated for Sec23p, which is a GAP for found that transferrin (Tf) uptake was inhibited (Fig. 1c). However, the small GTPase Sar1p and a component of the COPII complex8. the uptake of epidermal growth factor (EGF; Fig. 1d) and low- ARFGAP1 has been shown to have a similar function in the COPI density lipoprotein (LDL; Fig. 1e) were unaffected. This pattern of complex19,20. We now show that ARFGAP1 also has an unexpected selective inhibition was similar to that seen previously following the role in AP-2 regulated endocytosis with functional characterization depletion of AP-2 (ref. 23). suggesting that this has mechanistic parallels to its elucidated A previous study had concluded that siRNA against ARFGAP1 did functions in COPI transport. not affect Tf uptake24. However, scrutiny of this study suggested a

1Division of Rheumatology, Immunology and Allergy, Brigham and Women’s Hospital, and Department of Medicine, Harvard Medical School, Boston, Massachusetts 02115, USA. 2Department of Cell Biology and Oncology, Consorzio Mario Negri Sud, 66030 Santa Maria Imbaro (Chieti), Italy. 3Department of Genetics, Microbiology and Toxicology, Stockholm University, 10691 Stockholm, Sweden. 4Institute of Protein Biochemistry, National Research Council, Via Pietro, Castellino 111, 80131 Naples, Italy. 5Department of Cell Biology, Harvard Medical School, and Immune Disease Institute, Boston, Massachusetts 02115, USA. 6Department of Urology, University of Florida College of Medicine, Gainesville, Florida 32610, USA. 7Laboratory of Cellular and Molecular Biology, National Cancer Institute, Bethesda, Maryland 20892, USA. 8Department of Cell Biology, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA. 9Telethon Institute of Genetics and Medicine, Via Pietro Castellino 111, 80131 Napoli, Italy. 10These authors contributed equally to this work. 11Correspondence should be addressed to V.W.H. (e-mail: [email protected])

Received 28 September 2010; accepted 3 February 2011; published online 17 April 2011; DOI: 10.1038/ncb2221

ARTICLES

a b c 25,000

20,000 GST ARFGAP1ARFGAP1 GST ARFGAP1 Cytosol: ++ – 15,000 M (K) r β-COP GST ARFGAP1 250 10,000 CHC 160 Internalized Tf 5,000 105 BARS β2-adaptin (fluorescence units) 0 75 Control ARFGAP1 siRNA GST– ARFGAP1 de30,000 40,000

50 25,000 30,000 20,000 20,000 35 15,000

10,000 10,000 Internalized LDL (fluorescence units) Internalized EGF

30 (fluorescence units) 5,000 GST 0 0 Control ARFGAP1 siRNA Control ARFGAP1 siRNA

Figure 1 Interactions with ARFGAP1 and effects of its knockdown. experiments with standard error is shown. The difference between the two (a) Detection of proteins interacting with ARFGAP1 using a GST-pulldown conditions is significant (P < 0.05). (d) EGF uptake is not markedly affected assay. GST–ARFGAP1 was incubated with cytosol and associated by siRNA against ARFGAP1. BSC-1 cells were bound with fluorescently proteins were analysed by Coomassie blue staining. (b) Detection of labelled EGF and assessed for the level of internalized EGF at 10 min. The ARFGAP1 interactions with coat components using a GST-pulldown assay. mean from three experiments with standard error is shown. The difference GST–ARFGAP1 was incubated with cytosol and immunoblotted for the between the two conditions is insignificant (P > 0.05). (e) LDL uptake is not indicated proteins. ARFGAP1 interacts with components of AP-2, and also markedly affected by siRNA against ARFGAP1. BSC-1 cells were bound with with previously known interacting proteins that are components of the COPI fluorescently labelled LDL and assessed for the level of internalized LDL at complex. CHC, clathrin heavy chain. (c) Tf uptake is reduced by siRNA 10 min. The mean from three experiments with standard error is shown. The against ARFGAP1. BSC-1 cells were bound with fluorescently labelled Tf difference between the two conditions is insignificant (P > 0.05). Uncropped and assessed for the level of internalized Tf at 10 min. The mean from three images of blots are shown in Supplementary Fig. S6. possible explanation. In our study, Tf was bound to the cell surface triple mutant is referred to as EDE), which reduced the binding at 4 ◦C followed by washing to release nonspecific interactions. Cells of ARFGAP1 to clathrin, but not to AP-2 (Fig. 2c). Another set of were then warmed to 37 ◦C to restore transport. Importantly, we mutations involved substitutions at the fourth position in each motif also quantified the level of internalized Tf. In contrast, the previous (F332A/W366A/W385A; the triple mutant is referred to as FWW), study incubated cells with Tf continuously at 37 ◦C, and then assessed which reduced the binding of ARFGAP1 to both AP-2 and clathrin Tf uptake qualitatively, by visual inspection24. When we carried (Fig. 2d). ARFGAP1 binding to coatomer was unaffected by either set of out the Tf uptake assay by continuous incubation at 37 ◦C, we triple point mutations (Fig. 2c,d). Thus, these results revealed distinct could still detect inhibition in Tf uptake induced by siRNA against requirements for the binding of ARFGAP1 to AP-2 and clathrin versus ARFGAP1, by quantifying the level of internalized Tf (Supplementary coatomer (summarized in Fig. 2e). Fig. S1b). We sought further confirmation for our results using short We next determined whether ARFGAP1 could interact directly with hairpin RNA (shRNA) that targeted a different sequence in ARFGAP1 AP-2 and clathrin using purified components. This approach revealed (Supplementary Fig. S1c). Inhibition of Tf uptake was again observed that ARFGAP1 interacted directly with AP-2 (Fig. 3a), but not with (Supplementary Fig. S1d), and was further characterized by kinetic clathrin (Fig. 3b). Sequential incubations revealed that ARFGAP1 could analysis (Supplementary Fig. S1e). interact with clathrin when AP-2 was first incubated with ARFGAP1 We considered the possibility that the inhibition of Tf uptake (Fig. 3c). We also found that the FWW mutation in ARFGAP1 reduced following the depletion of ARFGAP1 could be an indirect effect of its direct interaction with AP-2, whereas the EDE mutation had a milder having perturbed COPI transport that acts in the secretory pathway. To effect (Fig. 3a). The EDE mutation in ARFGAP1 reduced the ability of address this issue, we initially explored whether ARFGAP1 interacted AP-2 to link ARFGAP1 to clathrin, as assessed by sequential incubations differently with AP-2 versus coatomer. When different truncation of AP-2 followed by clathrin (Fig. 3c). These results revealed that mutants of GST–ARFGAP1 were incubated with cytosol, we found that ARFGAP1 could interact directly with the AP-2 adaptor and only coatomer bound to the extreme carboxy-terminal region of ARFGAP1 indirectly with the clathrin triskelion. (residues 401–415; Fig. 2a and Supplementary Fig. S2a). In contrast, Next, we investigated whether ARFGAP1 acted in AP-2 transport AP-2 and clathrin interacted with a more proximal region of ARFGAP1 independently of its role in COPI transport. Cells were depleted of (residues 301–400; Fig. 2b). Within this region of ARFGAP1, we ARFGAP1 to inhibit Tf uptake, and then the different mutant forms noted multiple tryptophan-based (WXXF/W) motifs (Supplementary of ARFGAP1 were assessed for their ability to rescue this defect. These Fig. S2b), which mediate the binding of auxiliary proteins to AP-2 experiments were carried out by transfecting the rat form of ARFGAP1, (refs 25,26). Thus, we examined the effect of mutating residues which is resistant to the targeting sequences used to deplete endogenous within these motifs. One set of mutations involved substitutions ARFGAP1 in primate cells. The two mutant forms of ARFGAP1 (FWW at the second position of each motif (E330N/D364A/E383N; the and EDE), which were found in the binding studies to have reduced

ARTICLES

abc

GST 1–415137–415252–4151–136137–251252–359360–415 Input GST(1:10)1–415137–415252–415301–415360–415380–415301–400301–380 Wild typeE330NE383ND364A/E383ND364A/E330NE330N/E383NEDE Input (1:20) β-COP CHC CHC

AP-2 AP-2

β-COP

GSTs GSTs GSTs

d e COPI AP-2 CHC 1–415 ++ ++ ++ 1–136 – – – –400 –359 1–415 1 1 F332A F332A/W366AF332A/W385AFWW 137–251 – – – CHC 252–359 – – – 360–415 ++ – – 380–415 – – – AP-2 301–415 ++ ++ ++ 301–400 – ++ ++ β-COP 301–380 – + + 1–359 – – – EDE *** ++ ++ – GSTs FWW *** ++ – –

Figure 2 Distinct requirements for ARFGAP1 binding to coatomer versus experiments were carried out as described in a.(d) Interaction of additional AP-2 and clathrin. (a) Interaction of different ARFGAP1 truncation mutants mutants of ARFGAP1 with clathrin, AP-2 or coatomer. Pulldown experiments with coatomer. The different forms of GST–ARFGAP1 were bound to beads, were carried out as described in a.(e) Interactions of ARFGAP1 with incubated with cytosol and immunoblotted for the indicated proteins. GST coatomer, AP-2 and clathrin are summarized. The catalytic domain of fusion proteins were detected by Coomassie blue staining. (b) Interaction of ARFGAP1 is shown in white. Asterisks indicate W-based motifs. Degree different truncation mutants of ARFGAP1 with AP-2 or clathrin. Pulldown of association between ARFGAP1 form and indicated coat component: experiments were carried out as described in a.(c) Interaction of different none/minor (−), moderate (+), strong (++). Uncropped images of blots are point mutants of ARFGAP1 with clathrin, AP-2 or coatomer. Pulldown shown in Supplementary Fig. S6. ability to form a complex with AP-2 and clathrin, could not rescue Surveying transport pathways that require ARFGAP1 defective Tf uptake (Fig. 3d). In contrast, wild-type ARFGAP1 and a The surface level of the transferrin receptor (TfR) is predicted to truncated form, which spanned residues 1–400 and was defective for represent the balance between its rates of endocytosis and recycling. coatomer binding, did rescue this defect (Fig. 3d). We found that TfR surface levels were not appreciably altered by the We also assessed whether the different mutations in ARFGAP1 depletion of ARFGAP1 (Fig. 4a), which indicated that TfR recycling affected retrograde COPI transport. A chimaeric COPI cargo would also be affected by the perturbation of ARFGAP1 levels. We protein consisting of the KDEL (Lys–Asp–Glu–Leu) receptor coupled confirmed this by examining TfR recycling directly (Fig. 4b). To to a temperature-sensitive form of the vesicular stomatitis virus determine whether this inhibition was an indirect effect of having G protein, VSVG-ts045–KDELR, had been generated to track suppressed TfR endocytosis, we again used the ARFGAP1 mutants this transport in vivo27. At the non-permissive temperature, the that were defective in binding to the AP-2 complex. These mutants misfolding of VSVG-ts045 prevents the chimaera from exiting were able to rescue the defective TfR recycling that had been induced from the endoplasmic reticulum. Thus, because the KDELR cycles by the depletion of endogenous ARFGAP1 (Fig. 4c), indicating that between the endoplasmic reticulum and the Golgi, inhibition of ARFGAP1 acts distinctly in TfR endocytosis versus recycling. In light of its anterograde transport allows the unambiguous tracking of its this finding, we next conducted a broader survey to determine whether retrograde transport, which is COPI dependent28,29. We found ARFGAP1 could be acting in additional transport pathways. that the ARFGAP1 mutant that could not interact with coatomer Besides recycling to the plasma membrane, endocytic cargoes are (1–400) also could not rescue defective retrograde COPI transport transported towards two other general destinations, the lysosomes or induced by the depletion of endogenous ARFGAP1 (Fig. 3e). In the Golgi. To survey endocytic transport to lysosomes, we tracked contrast, the wild-type and the two mutant proteins that were dextran, a fluid-phase marker, which also allowed us to examine defective in forming a complex with AP-2 and clathrin rescued pinocytosis, a type of non-clathrin endocytosis30. We found that this defect (Fig. 3e). Thus, the results of the rescue revealed the internalization of dextran from the plasma membrane to early that ARFGAP1 acted in AP-2 transport independently of its endosomes (Supplementary Fig. S3a), followed by its subsequent role in COPI transport. transport to lysosomes (Supplementary Fig. S3b), were not appreciably

ARTICLES

abc

50% GSTinputWild typeEDE FWW 50% GSTinputWild typeEDE FWW GST Wild typeEDE FWW AP-2 (β2) CHC CHC

GST–ARFGAP1 GST–ARFGAP1 AP-2 (β2) GST–ARFGAP1 GSTs GSTs

GST GST GSTs GST

d 100 e 100 80 75 60 50 40 at Golgi 25 20 0 0 Percentage of uptake

EDE Percentage of VSVG–KDELR EDE FWW 1–400 FWW 1–400 Control Control Wild type Wild type shRNA alone shRNA alone

ARFGAP1 ARFGAP1

Figure 3 Disrupting the interaction between ARFGAP1 and either coatomer induced by the depletion of ARFGAP1. BSC-1 cells stably expressing shRNA or AP-2 leads to selective disruption in transport pathways. (a) Pulldown against ARFGAP1 were transfected with the different forms of rat ARFGAP1 assays to assess direct binding of ARFGAP1 to AP-2. Different forms of as indicated, and uptake of biotin–Tf at 10 min was quantified. The mean GST–ARFGAP1 were bound to beads, incubated with purified AP-2 adaptors from three experiments with standard error is shown. The differences among and immunoblotted for the indicated proteins. GST fusion proteins were conditions of ARFGAP1 shRNA, FWW and EDE are not significant (P > 0.05). detected by Coomassie blue staining. (b) Pulldown assays to assess direct The difference between this group and all other conditions is significant binding of ARFGAP1 to clathrin. The different forms of GST–ARFGAP1 were (P < 0.05). (e) Rescue of defective COPI transport induced by the depletion bound to beads, incubated with purified clathrin triskelia and immunoblotted of ARFGAP1. BSC-1 cells stably expressing shRNA against ARFGAP1 were for the indicated proteins. GST fusion proteins were detected by Coomassie transfected with the different forms of rat ARFGAP1 as indicated. The blue staining. (c) Sequential incubations reveal that the AP-2 adaptor redistribution of VSVG–KDELR from the Golgi to the endoplasmic reticulum is needed to link ARFGAP1 to the clathrin triskelion. Different forms of at 30 min was then quantified. The mean from three experiments with GST–ARFGAP1 were bound to beads, and incubated with purified AP-2 standard error is shown. The differences among conditions of wild type, followed by incubation with purified clathrin. Beads were analysed by FWW and EDE are not significant (P > 0.05). The difference between this immunoblotting for the indicated proteins. GST fusion proteins were group and conditions of shARFGAP1 and 1–400 is significant (P < 0.05). detected by Coomassie blue staining. (d) Rescue of defective Tf uptake Uncropped images of blots are shown in Supplementary Fig. S6. altered. To survey endocytic transport to the Golgi, we tracked perturbing other critical components of COPI vesicle formation, such the B subunit of cholera toxin (CTB), which also allowed us to as brefeldin-A ADP-ribosylated substrate28 (BARS) and phospholipase examine raft-dependent internalization, another type of non-clathrin D2 (PLD2; ref. 29). In contrast, perturbation of coatomer33 or ARF1 endocytosis30. We found that the internalization of CTB from the (ref. 34) was previously observed to disrupt Golgi structure. Thus, we plasma membrane to early endosomes (Supplementary Fig. S3c), concluded that ARF1 and coatomer, as well as having a role in COPI followed by subsequent transport to the trans-Golgi network (TGN; vesicular transport, could have additional functions that make them Supplementary Fig. S3d), were also unaffected. Thus, ARFGAP1 did critical for Golgi structure. not have widespread roles in endocytic pathways. We next surveyed the secretory pathway by tracking a well- ARFGAP1 promotes cargo sorting by AP-2 characterized anterograde cargo, VSVG-ts045 (refs 31,32). Transport To gain further insight into how ARFGAP1 could be acting in AP-2 of this cargo from the endoplasmic reticulum to the TGN was slowed transport, we determined its localization with respect to clathrin-coated down in ARFGAP1-depleted cells (Fig. 4d). However, transport from pits. Using the conventional approach of confocal microscopy, we the endoplasmic reticulum to the cis side of the Golgi (Fig. 4e) and found that the Golgi pool of ARFGAP1 emitted a sufficiently strong from the TGN to the plasma membrane was unaffected (Fig. 4f). Taken fluorescence signal to hamper the ability to detect the peripheral pool together, these results suggest that only the intra-Golgi segment of the of ARFGAP1. We therefore used total internal reflection fluorescence secretory pathway was delayed by the depletion of ARFGAP1, as would microscopy (TIR-FM) with live-cell imaging. We confirmed that be predicted by its role in COPI transport. Depletion of ARFGAP1 ARFGAP1 was still functional when coupled to the green fluorescent had minimal effects on Golgi structure, as assessed at the level of protein (GFP), as the fusion protein could still rescue the inhibition both light (Supplementary Fig. S4a) and electron (Supplementary of Tf uptake induced by the depletion of ARFGAP1 (Supplementary Fig. S4b) microscopy. Notably, similar findings had been observed by Fig. S5a). Using TIR-FM, we detected ARFGAP1 in clathrin-coated

ARTICLES

ab200,000 120 c100 Control 100 80 160,000 ARFGAP1 80 siRNA 60 120,000 60 40 80,000 40 internal Tf 20 Percentage of Surface TfR 40,000 20 0 (fluorescence units) 0

0 Percentage of internal Tf EDE Control ARFGAP1 siRNA 0 102030 FWW Control Time (min) ARFGAP1 Wild type

shRNA alone de f 100 100 Control 100 Control Control ARFGAP1 shRNA ARFGAP1 shRNA ARFGAP1 shRNA 75 75 75

50 50 50 25 25 co-localization co-localization 25 co-localization 0 0

Percentage of VSVG/giantin 2 min 5 min 10 min Percentage of VSVG/TGN46 10 min 20 min 40 min 0 10 min 20 min 30 min Percentage of VSVG/PM marker Figure 4 Surveying transport pathways affected by the depletion of ARFGAP1. the endoplasmic reticulum to the TGN. VSVG-ts045–GFP was transfected (a) The surface level of TfR is not affected by ARFGAP1 depletion. An into BSC-1 cells, followed by quantification of its co-localization with a antibody that recognizes the extracellular domain of TfR was bound to TGN marker (TGN46). The mean from three experiments with standard BSC-1 cells, followed by quantification. The mean from three experiments error is shown. The difference at all time points is significant (P < 0.05). with standard error is shown. The difference between the two conditions (e) Depletion of ARFGAP1 does not affect transport from the endoplasmic is insignificant (P > 0.05). (b) TfR recycling is reduced by siRNA against reticulum to the cis side of the Golgi. VSVG-ts045–GFP was transfected ARFGAP1. BSC-1 cells were treated with siRNA against ARFGAP1 and the into BSC-1 cells, followed by quantification of its co-localization with a level of internal biotin–Tf was quantified at the indicated time points. The cis-Golgi marker (giantin). The mean from three experiments with standard mean from three experiments with standard error is shown. The difference error is shown. The difference at all time points is insignificant (P > 0.05). between the two conditions (except time = 0) is significant (P < 0.05). (f) Depletion of ARFGAP1 does not affect transport from the TGN to the (c) Rescue of defective TfR recycling induced by the depletion of ARFGAP1. plasma membrane (PM). VSVG-ts045–GFP was transfected into BSC-1 BSC-1 cells stably expressing shRNA against ARFGAP1 were transfected cells, and then accumulated at the TGN. Cells were then shifted from 20 with the different forms of rat ARFGAP1 indicated. The level of internal to 32 ◦C to allow transport to the plasma membrane. Arrival of VSVG at the biotin–Tf that remained after 10 min of recycling was then quantified. The plasma membrane was detected through co-localization with fluorescently mean with standard error from three experiments is shown. The difference labelled CTB (which was bound to the cell surface). The mean from three between cells expressing ARFGAP1 shRNA and all other conditions is experiments with standard error is shown. The difference at all time points significant (P < 0.05). (d) Depletion of ARFGAP1 inhibits transport from is insignificant (P > 0.05). pits (Fig. 5a). Dynamic imaging revealed that ARFGAP1 associated associated with AP-2 and ARFGAP1 (Fig. 6a). Transfected GFP-tagged transiently with clathrin-coated structures mainly at the later stage of ARFGAP1 similarly associated with surface TfR and AP-2 (Fig. 6b), their lifetime (Fig. 5b). which supported our previous observation that this tagged ARFGAP1 Using quantitative electron microscopy, we found that the level of could be detected in coated pits. In contrast, two other ARFGAPs, coated pits was reduced in cells depleted of endogenous ARFGAP1 ARFGAP2 and ARFGAP3, could not interact with surface TfR (Fig. 6c). (Fig. 5c). Thus, as coated pits internalize a broad range of cargoes13, a Moreover, consistent with our initial finding that the endocytosis of the prediction was that ARFGAP1 would be important for the endocytosis LDL receptor (LDLR) was not affected by the depletion of ARFGAP1, of more cargoes than just that TfR. Supporting this prediction, we found we found that a surface form of this receptor (CD8-LDLR), previously that the endocytosis of cystic fibrosis transmembrane conductance used study the endocytosis of LDLR, (ref. 23) did not interact with regulator (CFTR), which requires AP-2 (ref. 35), was also inhibited ARFGAP1 (Fig. 6d). Previous studies had observed variable effects by the depletion of ARFGAP1 (Fig. 5d). We next examined whether on LDL uptake following the depletion of AP-2, with no inhibition the depletion of ARFGAP1 arrested a particular stage of coated-pit detected by the traditional method of uptake (which involved binding formation. However, no particular stage was observed to accumulate at 4 ◦C followed by incubation at 37 ◦C; ref. 23), whereas inhibition (Fig. 5e). To account for the above observations, we were led by a recent was detected by an alternative method (which involved continuous study that showed clathrin vesicle formation can have two general uptake at 37 ◦C; ref. 39). In contrast, we found that the depletion of outcomes, either productive or abortive36. As cargoes have been shown ARFGAP1 inhibited Tf uptake using either method (see Fig. 1c and recently to promote the productive fate36–38, we next explored whether Supplementary S1b). Thus, these observations confirm the specificity ARFGAP1 promoted cargo binding by AP-2. by which ARFGAP1 functions in the endocytosis of TfR. Studying TfR as the model cargo, we initially determined whether an Cargo sorting involves coat components binding directly to sorting in vivo complex between ARFGAP1 and surface TfR and AP-2 could signals in cargo proteins1,13. Thus, we examined whether ARFGAP1 be detected. To examine only endogenous proteins, biotin-conjugated could interact directly with two internalization sorting signals that had Tf was bound to the cell surface, and following cell lysis, lysates were been identified in the cytoplasmic domain of TfR. One signal requires a incubated with streptavidin-bound beads. Surface TfR was found to be critical phenylalanine at position 13 of TfR, whereas the other requires

ARTICLES

a 1 2 34 c 160 140 d 80 Control 120 1 ARFGAP1 siRNA 100 60 2 80

3 per mm 60 40 40 20 4 20 CFTR endocytosed Number of CC intermediates Percentage of cell surface 0 0 Control siRNA ARFGAP1 siRNA 02468 Time (min) b

e 70 60 50 Control siRNA 12340 30 20 Percentage of CC 10 ARFGAP1 siRNA123intermediates per stage 0 123 123 Control siRNA ARFGAP1 siRNA

Figure 5 The role of ARFGAP1 in coated-pit formation. (a) Co-localization were randomly selected from each condition to obtain the mean with of ARFGAP1 with clathrin in coated pits. BSC-1 cells transfected with standard deviation. The difference between the two conditions is significant GFP-tagged ARFGAP1 and mCherry-tagged clathrin light chain were (P < 0.05). (d) Depletion of ARFGAP1 inhibits the endocytosis of CFTR. examined by TIR-FM. The merged view shows co-localization of the two CFBE41o- cells expressing wild-type CFTR were polarized and treated with proteins; scale bar, 2 µm. Insets highlight examples of coated pits found the indicated siRNA conditions. The level of internalized CFTR was then to have both ARFGAP1 (green) and clathrin (red). (b) Dynamic association quantified. The mean with standard error from three experiments is shown. of ARFGAP1 with clathrin in coated pits. BSC-1 cells were transfected The difference between the two conditions (except time = 0) is significant with GFP–ARFGAP1 and mCherry-tagged clathrin light chain and examined (P < 0.05). (e) Depletion of ARFGAP1 does not induce accumulation of a by TIR-FM with live-cell imaging. Two examples are shown, with images particular stage of coated pit. The different stages of coated-pit formation, captured every 3 s; scale bars, 1 µm. (c) Depletion of ARFGAP1 reduces with representative images for each stage shown (left; bar, 100 nm), were the level of coated pits. HeLa cells were treated with the indicated siRNA detected by electron microscopy, and then quantified. The mean from three conditions. All forms of clathrin-coated (CC) intermediates were counted experiments with standard deviation is shown (right). The difference between and then divided by the length of the plasma membrane. Seven cells corresponding stages is insignificant (P > 0.05). a critical ‘YTRF’ motif (positions 20–23; refs 40,41). We generated sorting signal (containing the ‘YTRF’ motif) abrogated the association a series of truncation constructs that encompassed the cytoplasmic of TfR with AP-2 and ARFGAP1 more strongly than mutations in the domain of TfR (summarized in Supplementary Fig. S5b), and found other sorting signal (Fig. 7d). Notably, we also found that the depletion that ARFGAP1 bound to the portion of the TfR cytoplasmic region that of ARFGAP1 reduced the association of AP-2 with TfR in vivo (Fig. 7e). contained both internalization sorting signals (Fig. 7a). Subsequently, Taken together, the binding results suggested that ARFGAP1 acted in examining truncations of TfR that possessed either of these two sorting promoting cargo sorting by AP-2. signals, we found that ARFGAP1 could bind to either signal (Fig. 7b). In contrast, AP-2 could bind only to the sorting signal defined by the Elucidating the role of the GAP activity in TfR endocytosis ‘YTRF’ motif (Fig. 7c). As ARFGAP1 also possessed a GAP activity, we next examined whether We also examined how mutations at the two sorting signals of this had a role in AP-2 transport. A point mutation in ARFGAP1 TfR affected its association with ARFGAP1 and AP-2 in vivo. TfR had been shown previously to abrogate its GAP activity43. Whereas forms homodimers42, so a transfected mutant TfR could potentially wild-type ARFGAP1 could rescue the defective Tf uptake induced by the pair with endogenous wild-type TfR and thereby mask an effect of depletion of endogenous ARFGAP1, the catalytic-dead mutant could a particular mutation. To overcome this potential hurdle, we used a not (Fig. 8a and Supplementary Fig. S5c). This result not only revealed mutant cell line that lacked endogenous TfR (ref. 41). When different a role for the GAP activity of ARFGAP1 in AP-2 transport, but also forms of TfR were transfected into this cell line, mutations in one implicated an ARF in this process. Consistent with ARF6 having been

ARTICLES

abBiotin–Tf: –+ Biotin–Tf cBiotin–Tf Biotin–Tf Biotin–Tf dCD8 Ab –+ –+ –+ –+ TfR Lys Lys Lys Lys Lys –+ ARFGAP1 TfR TfR ARFGAP1 AP-2 (β2) GFP–ARFGAP1 ARFGAP CD8-LDLR AP-2 (β2) Transfection:

ARFGAP2–HA ARFGAP3–HA ARFGAP1–Myc Figure 6 ARFGAP1 and AP-2 interact speciﬁcally with surface TfR. cells were transfected with Myc-tagged ARFGAP1, or (HA)-tagged forms (a) Endogenous surface TfR interacts with endogenous forms of ARFGAP1 of ARFGAP-2 or ARFGAP3. Cells were bound with biotin-labelled Tf. Cell and AP-2 in vivo. Biotin-labelled Tf was bound to the surface of lysis was followed by incubation of cell lysate with streptavidin beads and HeLa cells, followed by isolation using streptavidin beads and then immunoblotting for the indicated proteins. ARFGAPs were detected using immunoblotting for the indicated proteins. (b) GFP-tagged ARFGAP1 antibodies against the epitope tag. (d) A surface form of LDLR does not associates with endogenous forms of surface TfR and AP-2. BSC-1 associate with ARFGAP1. HeLa cells stably expressing CD8-LDLR were cells were transfected with GFP-tagged ARFGAP1. Cells were then incubated with anti-CD8 antibody. After this surface binding, cells were bound with biotin-labelled Tf, followed by incubation of cell lysate lysed and incubated with protein A beads, followed by immunoblotting with streptavidin beads and immunoblotting for the indicated proteins. for the indicated proteins. Uncropped images of blots are shown in (c) Surface TfR does not associate with ARFGAP-2 or ARFGAP3. HeLa Supplementary Fig. S7. suggested to act in clathrin-mediated endocytosis44–46, the depletion auxiliary coat component to promote cargo sorting by AP-2. We have of ARF6 inhibited Tf uptake (Fig. 8b). The GAP activity of ARFGAP1 also found that ARFGAP1 acts as an ARF6 regulator in AP-2 transport, had only been well documented to act on ARF1. Thus, we revisited as activated ARF6 that binds to TfR must be deactivated by the GAP the in vitro GAP assay, and found that ARF6 could also be a substrate activity of ARFGAP1 to allow optimal binding of TfR by AP-2. (Supplementary Fig. S5d). Furthermore, as coatomer had been shown AP-2 is now appreciated to act in at least two distinct stages. An to stimulate the GAP activity of ARFGAP1 towards ARF1 (ref. 47), we initial stage involves AP-2 engaging a critical lipid, phosphatidylinositol found that AP-2 could also stimulate the GAP activity of ARFGAP1 4,5-bisphosphate (PIP2), on the plasma membrane. A later stage towards ARF6 (Fig. 8c). involves AP-2 subsequently undergoing major conformational changes On seeking further insight into how the GAP activity of ARFGAP1 to open its binding sites for cargoes48,49. ARF6 has been shown to acted in TfR endocytosis, we initially found that ARF6 could bind promote PIP2 production44,50. As such, it is predicted to act in the directly to the cytoplasmic domain of TfR in an activation-dependent early stage of AP-2 action. However, our elucidation of how the GAP manner (Fig. 8d). Moreover, ARF6 synergized with ARFGAP1 in activity of ARFGAP1 promotes TfR endocytosis suggests an additional enhancing the binding of AP-2 to TfR (Fig. 8e). Notably, the GAP role for ARF6. This role is predicted to occur in the later stage of AP-2 activity of ARFGAP1 further enhanced this binding. This was suggested action, which involves a complete GTPase cycle of ARF6 that optimizes by the observation that ARF6 preloaded with GTP allowed better cargo binding by AP-2. ARF1 deactivation has been shown to promote binding of TfR by AP-2 than ARF6 loaded with a non-hydrolysable COPI cargo sorting51, but the underlying mechanism has been unclear. analogue of GTP (GTP-γS; Fig. 8e). As confirmation, we found that As the mechanistic action of ARFGAP1 is similar in both AP-2 and the activation-dependent binding of ARF6 to TfR was abrogated COPI transport, our present elucidation of how the GAP activity of when ARFGAP1 was added to the incubation, whereas this effect was ARFGAP1 promotes AP-2 cargo sorting also provides insights into its prevented when ARF6 was preloaded with GTP-γS(Fig. 8f). We also function in COPI cargo sorting. sought in vivo confirmation for key aspects of these in vitro findings. First, we confirmed that ARF6 bound to TfR in an activation-dependent METHODS manner, as the activating (Q67L), but not the deactivating (T27N), Methods and any associated references are available in the online point mutant of ARF6 associated with surface TfR (Fig. 8g). Second, version of the paper at http://www.nature.com/naturecellbiology/ we confirmed that ARF6 deactivation was important for the binding of AP-2 to surface TfR in vivo, as the expression of ARF6Q67L reduced this Note: Supplementary Information is available on the Nature Cell Biology website association (Fig. 8h). Altogether, these results suggested that at least ACKNOWLEDGEMENTS one mechanism by which the GAP activity of ARFGAP1 promoted TfR We thank B. Wendland for critical comments. This work was financially supported endocytosis involved the optimization of cargo binding by AP-2. by grants from the National Institutes of Health to V.W.H. (GM058615 and GM073016), T.K. (GM075252 and U54 AI057159—NERCE Imaging Resource) and J.F.C. (DK060065). A.L. was financially supported by the Telethon (Italy) and AIRC DISCUSSION (Italy). H.G. was supported by a Marie Curie Fellowship. E.C. was supported by a We have found that ARFGAP1 promotes AP-2-dependent endocytosis. GlaxoSmithKline fellowship.

Characterizing this role, we have uncovered mechanistic parallels to AUTHOR CONTRIBUTIONS the previously reported roles of ARFGAP1 in COPI transport, which M.B., H.G., G.T., E.C., J-S.Y., J.L., G.V.B., Z.N., L.F. and R.L. carried out experiments involve functions as both an ARF effector and regulator6. ARFGAP1 and data analyses. V.W.H., A.L., T.K. and J.F.C. supervised the work. V.W.H., A.L., binds to the two known sorting signals in TfR to promote endocytosis. H.G. and M.B. wrote the manuscript. Moreover, ARFGAP1 is needed to link AP-2 to TfR. These observations COMPETING FINANCIAL INTERESTS suggest that ARFGAP1 behaves as an ARF effector by acting as an The authors declare no competing financial interests.

ARTICLES

a bc

31 19 19 19(22A/23A) 19 19(22A/23A) Δ Δ Δ Δ Δ Δ 10% inputGST TfR N31 N N19 N 50% GSTinputN19 N19(12A/13A)N N 50% GSTinputN19 N19(12A/13A)N N ARFGAP1 ARFGAP1 AP-2 (β2)

GSTs GSTs GSTs

de Wild type 4A 12A/13A 22A/23A siRNA –+ –– ++ Biotin–Tf –+ –+ –+ –+ Biotin–Tf: Lysate –+ –+ TfR TfR ARFGAP1 AP-2 (β2) AP-2 (β2) ARFGAP1

Figure 7 Characterizing the binding of TfR by ARFGAP1 and AP-2. (d) Effects of TfR mutations on its association with ARFGAP1 and AP-2. (a) Pulldown assays were carried out in which the indicated TfR constructs Biotin-labelled Tf was bound to the surface of mutant CHO (TRVb) were incubated with recombinant ARFGAP1, followed by immunoblotting cells expressing different full-length TfR forms (wild type or with the for ARFGAP1. GST fusion proteins were detected by Coomassie blue indicated point mutations) followed by isolation using streptavidin beads staining. (b) Pulldown assays were carried out in which the indicated and immunoblotting for the proteins indicated. The ‘4A’ construct TfR constructs were incubated with recombinant ARFGAP1, followed contains alanine substitutions at residues 12, 13, 22 and 23 of TfR. by immunoblotting for ARFGAP1. GST fusion proteins were detected by (e) Association of surface TfR with AP-2 requires ARFGAP1. BSC-1 cells Coomassie blue staining. Point mutations within particular truncation were treated with siRNA against ARFGAP1. Biotin-labelled Tf was bound constructs are indicated within parentheses. (c) Pulldown assays were to the surface of BSC-1 cells, followed by isolation using streptavidin carried out in which the indicated TfR constructs were incubated with beads and immunoblotting for the endogenous proteins indicated. Cell puriﬁed AP-2, followed by immunoblotting for β2-adaptin. GST fusion lysates were also directly immunoblotted for proteins in the indicated proteins were detected by Coomassie blue staining. Point mutations conditions. Uncropped images of blots are shown in Supplementary within particular truncation constructs are indicated within parentheses. Fig. S7.

18. Inoue, H. & Randazzo, P. A. Arf GAPs and their interacting proteins. Traffic 8, Published online at http://www.nature.com/naturecellbiology 1465–1475 (2007). Reprints and permissions information is available online at http://npg.nature.com/ 19. Yang, J. S. et al. ARFGAP1 promotes the formation of COPI vesicles, suggesting reprintsandpermissions/ function as a component of the coat. J. Cell Biol. 159, 69–78 (2002). 20. Lee, S. Y., Yang, J. S., Hong, W., Premont, R. T. & Hsu, V. W. ARFGAP1 plays a central role in coupling COPI cargo sorting with vesicle formation. J. Cell Biol. 168, 1. Bonifacino, J. S. & Glick, B. S. The mechanisms of vesicle budding and fusion. Cell 281–290 (2005). 116, 153–166 (2004). 21. Schmid, E. M. et al. Role of the AP2 β-appendage hub in recruiting partners for 2. Cai, H., Reinisch, K. & Ferro-Novick, S. Coats, tethers, Rabs, and SNAREs work clathrin-coated vesicle assembly. PLoS Biol. 4, 1532–1548 (2006). together to mediate the intracellular destination of a transport vesicle. Dev. Cell 12, 22. Rawet, M., Levi-Tal, S., Szafer-Glusman, E., Parnis, A. & Cassel, D. ArfGAP1 671–682 (2007). interacts with coat proteins through tryptophan-based motifs. Biochem. Biophys. 3. Lee, M. C., Miller, E. A., Goldberg, J., Orci, L. & Schekman, R. Bi-directional Res. Commun. 394, 553–557 (2010). protein transport between the ER and Golgi. Annu. Rev. Cell Dev. Biol. 20, 23. Motley, A., Bright, N. A., Seaman, M. N. & Robinson, M. S. Clathrin-mediated 87–123 (2004). endocytosis in AP-2-depleted cells. J. Cell Biol. 162, 909–918 (2003). 4. McMahon, H. T. & Mills, I. G. COP and clathrin-coated vesicle budding: differ- 24. Saitoh, A., Shin, H. W., Yamada, A., Waguri, S. & Nakyryama, K. Three homologous ent pathways, common approaches. Curr. Opin. Cell Biol. 16, 379–391 (2004). ArfGAPs participate in coat protein I-mediated transport. J. Biol. Chem. 284, 5. Pucadyil, T. J. & Schmid, S. L. Conserved functions of membrane active GTPases in 13948–13957 (2009). coated vesicle formation. Science 325, 1217–1220 (2009). 25. Mishra, S. K. et al. Dual engagement regulation of protein interactions with the AP-2 6. Hsu, V. W., Lee, S. Y. & Yang, J. S. The evolving understanding of COPI vesicle adaptor α appendage. J. Biol. Chem. 279, 46191–46203 (2004). formation. Nat. Rev. Mol. Cell Biol. 10, 360–364 (2009). 26. Praefcke, G. J. et al. Evolving nature of the AP2 α-appendage hub during clathrin- 7. Waters, M. G., Serafini, T. & Rothman, J. E. ‘Coatomer’: a cytosolic protein complex coated vesicle endocytosis. EMBO J. 23, 4371–4383 (2004). containing subunits of non-clathrin-coated Golgi transport vesicles. Nature 349, 27. Cole, N. B., Ellenberg, J., Song, J., DiEuliis, D. & Lippincott-Schwartz, J. Retrograde 248–251 (1991). transport of Golgi-localized proteins to the ER. J. Cell Biol. 140, 1–15 (1998). 8. Barlowe, C. et al. COPII: a membrane coat formed by Sec proteins that drive vesicle 28. Yang, J. S. et al. A role for BARS at the fission step of COPI vesicle formation from budding from the endoplasmic reticulum. Cell 77, 895–907 (1994). Golgi membrane. EMBO J. 24, 4133–4143 (2005). 9. Pearse, B. M. F. Coated vesicles from pig brain: purification and biochemical 29. Yang, J. S. et al. A role for phosphatidic acid in COPI vesicle fission yields insights characterization. J. Mol. Biol. 97, 93–98 (1975). into Golgi maintenance. Nat. Cell Biol. 10, 1146–1153 (2008). 10. Pearse, B.M. & Robinson, M. S. Purification and properties of 100-kd proteins 30. Conner, S. D. & Schmid, S. L. Regulated portals of entry into the cell. Nature 422, from coated vesicles and their reconstitution with clathrin. EMBO J. 3, 37–44 (2003). 1951–1957 (1984). 31. Presley, J. F. et al. ER-to-Golgi transport visualized in living cells. Nature 389, 11. Maldonado-Baez, L. & Wendland, B. Endocytic adaptors: recruiters, coordinators and 81–85 (1997). regulators. Trends Cell Biol. 16, 505–513 (2006). 32. Bonazzi, M. et al. CtBP3/BARS drives membrane fission in dynamin-independent 12. Ungewickell, E. J. & Hinrichsen, L. Endocytosis: clathrin-mediated membrane transport pathways. Nat. Cell Biol. 7, 570–580 (2005). budding. Curr. Opin. Cell Biol. 19, 417–425 (2007). 33. Guo, Q., Vasile, E. & Krieger, M. Disruptions in Golgi structure and membrane traffic 13. Traub, L. M. Tickets to ride: selecting cargo for clathrin-regulated internalization. in a conditional lethal mammalian cell mutant are corrected by -COP. J. Cell Biol. Nat. Rev. Mol. Cell Biol. 10, 583–596 (2009). 125, 1213–1224 (1994). 14. Hsu, V. W. & Yang, J. S. Mechanisms of COPI vesicle formation. FEBS Lett 583, 34. Dascher, C. & Balch, W. E. Dominant inhibitory mutants of ARF1 block endoplasmic 3758–3763 (2009). reticulum to Golgi transport and trigger disassembly of the Golgi apparatus. 15. Miller, E. A. & Barlowe, C. Regulation of coat assembly–sorting things out at the ER. J. Biol. Chem. 269, 1437–1448 (1994). Curr. Opin. Cell Biol. 22, 447–453 (2010). 35. Weixel, K. M. & Bradbury, N. A. Mu 2 binding directs the cystic fibrosis 16. D’Souza-Schorey, C. & Chavrier, P. ARF proteins: roles in membrane traffic and transmembrane conductance regulator to the clathrin-mediated endocytic pathway. beyond. Nat. Rev. Mol. Cell Biol. 7, 347–358 (2006). J. Biol. Chem. 276, 46251–46259 (2001). 17. Casanova, J. E. Regulation of Arf activation: the Sec7 family of guanine nucleotide 36. Ehrlich, M. et al. Endocytosis by random initiation and stabilization of clathrin- exchange factors. Traffic 8, 1476–1485 (2007). coated pits. Cell 118, 591–605 (2004).

ARTICLES

ab100 120 c 80 80 100 60 80 60 40 60 40 20 40 0

Percentage of Tf uptake 20 20 R50K Percentage of Tf uptake Control shRNA 0 Wild type Control Arf6 siRNA

Percentage of GTP hydrolysed 0 0 50 100 150 200 250 ARFGAP1 (nM)

d GST–GGA3 GST–TfR e ARF6 (GTP) – – + – f γS ARF6 (GTP-γS) – – – + GDP + –– + –– GTP GTP- GTP – + – – + – ARFGAP1 + + + + ARF6 GTP-γS – – + – – + AP-2 (β2) – + + +

ARF6 AP-2 (β2) ARFGAP1

GSTs GST–TfR GST–TfR

ghT27N Q67L Lysates Control Q67L Biotin–Tf: –+ –+ Biotin–Tf: –+–+ T27N Q67L ARF6 TfR

TfR AP-2 (β2)

Q67L

Figure 8 The GAP activity of ARFGAP1 promotes TfR endocytosis. (a) The in a pulldown experiment (left panel). ARF6 forms were also incubated GAP activity of ARFGAP1 is important for TfR endocytosis. BSC-1 cells with GST–TfR in another pulldown experiment (right panel). (e) GAP stably expressing shRNA against ARFGAP1 were transfected with the activity of ARFGAP1 optimizes the binding of AP-2 to TfR. GST–TfR was indicated constructs and uptake of biotin–Tf at 10 min was quantified. The incubated sequentially with ARF6 (containing different bound nucleotides, mean with standard error from three experiments is shown. The difference as indicated), followed by ARFGAP1 and then AP-2. (f) Deactivation of between the condition of shRNA and rescue by the wild type is significant ARF6 by ARFGAP1 releases ARF6 from binding to TfR. ARF6 loaded with (P < 0.05). The difference between the condition of shRNA and rescue by GTP forms, as indicated, were incubated with GST–TfR along with ARFGAP1 R50K is insignificant (P > 0.05). (b) Depletion of ARF6 inhibits Tf uptake. in a pulldown experiment. (g) Activation-dependent binding of ARF6 to BSC-1 cells were treated with siRNA against ARF6 and uptake of biotin–Tf at surface TfR. BSC-1 cells were transfected with point-mutant forms of ARF6 10 min was quantified. The mean with standard error from three experiments as indicated. Surface TfR was isolated through biotin–Tf binding to cell is shown. The difference between the two conditions is significant (P < 0.05). surface, followed by incubation with streptavidin beads and immunoblotting (c) AP-2 enhances GAP activity of ARFGAP1 towards ARF6. The GAP assay was carried out for the indicated proteins. (h) The constitutively activated was carried out using ARF6 as the substrate, and either with (triangles) or form of ARF6 (Q67L) reduces the binding of AP-2 to surface TfR. BSC-1 without (squares) AP-2. The mean from three experiments with standard cells were transfected with ARF6Q67L or mock transfected. Surface TfR was error is shown. (d) Binding of ARF6 to TfR is activation dependent. The isolated as described in g, followed by immunoblotting for the indicated functional activation of ARF6 was confirmed using the effector domain of associated proteins. Uncropped images of blots are shown in Supplementary GGA3 (Golgi-localized, γ-adaptin ear-containing, ARF-binding protein 3) Fig. S8.

37. Loerke, D. et al. Cargo and dynamin regulate clathrin-coated pit maturation. PLoS 44. Krauss, M. et al. ARF6 stimulates clathrin/AP-2 recruitment to synaptic membranes Biol. 7, 628–639 (2009). by activating phosphatidylinositol phosphate kinase type Iγ. J. Cell Biol. 162, 38. Mettlen, M., Loerke, D., Yarar, D., Danuser, G. & Schmid, S. L. Cargo- and 113–124 (2003). adaptor-specific mechanisms regulate clathrin-mediated endocytosis. J. Cell Biol. 45. Palacios, F., Schweitzer, J. K., Boshans, R. L. & D’Souza-Schorey, C. ARF6-GTP 188, 919–933 (2010). recruits Nm23-H1 to facilitate dynamin-mediated endocytosis during adherens 39. Boucrot, E., Saffarian, S., Zhang, R. & Kirchhausen, T. Roles of AP-2 in junctions disassembly. Nat. Cell Biol. 4, 929–936 (2002). clathrin-mediated endocytosis. PloS ONE 5, e10597 (2010). 46. Paleotti, O. et al. The small G-protein Arf6GTP recruits the AP-2 adaptor complex to 40. Collawn, J. F. et al. Transferrin receptor internalization sequence YXRF implicates membranes. J. Biol. Chem. 280, 21661–21666 (2005). a tight turn as the structural recognition motif for endocytosis. Cell 63, 47. Goldberg, J. Decoding of sorting signals by coatomer through a GTPase switch in the 1061–1072 (1990). COPI coat complex. Cell 100, 671–679 (2000). 41. McGraw, T. E., Pytowski, B., Arzt, J. & Ferrone, C. Mutagenesis of the human 48. Jackson, L. P. et al. A large-scale conformational change couples membrane transferrin receptor: two cytoplasmic phenylalanines are required for efficient recruitment to cargo binding in the AP2 clathrin adaptor complex. Cell 141, internalization and a second-site mutation is capable of reverting an internalization- 1220–1229 (2010). defective phenotype. J. Cell Biol. 112, 853–861 (1991). 49. Rapoport, I. et al. Regulatory interactions in the recognition of endocytic sorting 42. Alvarez, E., Girones, N. & Davis, R. J. Intermolecular disulfide bonds are not required signals by AP-2 complexes. EMBO J. 16, 2240–2250 (1997). for the expression of the dimeric state and functional activity of the transferrin 50. Honda, A. et al. Phosphatidylinositol 4-phosphate 5-kinase α is a downstream receptor. EMBO J. 8, 2231–2240 (1989). effector of the small G protein ARF6 in membrane ruffle formation. Cell 99, 43. Szafer, E. et al. Role of coatomer and phospholipids in GTPase-activating protein- 521–532 (1999). dependent hydrolysis of GTP by ADP-ribosylation factor-1. J. Biol. Chem. 275, 51. Lanoix, J. et al. GTP hydrolysis by arf-1 mediates sorting and concentration of Golgi 23615–23619 (2000). resident enzymes into functional COP I vesicles. EMBO J. 18, 4935–4948 (1999).

METHODS DOI: 10.1038/ncb2221

METHODS buffer (0.5 M NaCl and 0.5% acetic acid) to release biotin–Tf that remained bound Proteins and antibodies. GST fusion and 6×-His-tagged recombinant proteins to the cell surface. Cells were then lysed followed by SDS–PAGE analysis, and then were purified as previously described19. AP-2 and clathrin were purified as previously detection with horseradish peroxidase (HRP)-conjugated streptavidin. The level of described52. Alexa 546-conjugated Tf, Alexa 555–EGF, Dil–LDL and Alexa 555- internalized biotin–Tf was normalized to that initially bound to the cell surface in conjugated forms of dextran and CTB were obtained (Invitrogen). calculating the fraction of TfR internalized. The following antibodies were used, with dilution and source as indicated. TfR recycling was carried out by first incubating biotin-conjugated Tf with cells Antibodies against the following proteins have been described previously28,53: at 37 ◦C for 2 h. Cells were then shifted to 4 ◦C for washes with an acid-based ARF6 (1:200), ARFGAP1 (1:1,000), BARS (1:2,000), β2-adaptin (AP.6, 1:5,000), buffer (0.5 M NaCl and 0.5% acetic acid) to release biotin–Tf that remained bound β-COP (M3A5, 1:100, or culture supernatant, 1:10), CHC (TD.1, 1:500, or culture to the cell surface. To measure recycling, cells were then warmed to 37 ◦C to supernatant, 1:30), giantin (1:5,000), Lamp1 (1:500), Myc epitope (9E10, 1:1,000), restore transport, followed by another acid-wash. Cells were then lysed followed TfR (5E9, 1:500, or H68.4, 1:1,000) and VSVG (BW8G65, culture supernatant 1:10). by SDS–PAGE analysis, and then detection with HRP-conjugated streptavidin. For Antibodies against other proteins were obtained: CD8 (1:400, Santa Cruz), EEA1 quantification, the level of internal biotin–Tf was normalized to the level of biotin–Tf (1:200, BD Bioscience), HA epitope (HA.11, 1:1,500, Covance) and TGN46 (1:1,000, within cells at the start of recycling to calculate the fraction of internal Tf remaining AbD Serotec). at different time points. To assess transport from the plasma membrane to the early endosome, and Mass spectrometry. Cytosol was purified from rat brains by initially homogeniz- subsequently to the lysosome, fluorescently labelled dextran was incubated with ing them in buffer B (25 mM Tris at pH 8, 500 mM KCl, 250 mM sucrose, 2 mM cells at 37 ◦C for the times indicated. Co-localization was then carried out with EGTA, 1 mM dithiothreitol (DTT) plus protease inhibitors) at 4 ◦C, followed by markers of the early endosome (EEA1) and the lysosome (Lamp1). To assess centrifugation at 5,000g for 30 min and then 150,000g for 90 min. The resulting transport from the plasma membrane to the early endosome, and subsequently to supernatant was dialysed against buffer C (25 mM Tris at pH 8, 50 mM KCl and the TGN, fluorescently labelled CTB was bound to the cell surface at 4 ◦C. Cells 1 mM DTT), followed by centrifugation at 150,000g for 90 min at 4 ◦C, and then were then washed with PBS, followed by warming to 37 ◦C for the times indicated. stored at −80 ◦C. GST–ARFGAP1 was crosslinked to glutathione sepharose by Co-localization was then carried out with markers of the early endosome (EEA1) and washing with sodium borate, incubation with dimethyl pimelimidate (DMP) for the TGN (TGN46). Confocal images were obtained using an inverted microscope 30 min followed by two washes with ethanol amine at pH 8.2 and two washes with (TE2000, Nikon) with a C1 confocal system. Quantification of co-localization was 0.1 M glycine at pH 2.5. Beads were then incubated with or without 10 mg rat brain carried out as previously described53. cytosol for 4 h followed by extensive washing. Samples were separated by SDS–PAGE Other in vivo transport assays have been described, including anterograde and then stained with silver chloride. Specific bands were excised for analysis by mass transport of VSVG-ts045 (ref. 55) and the redistribution of VSVG-ts045–KDELR spectrometry (MSF facility, University of California, San Francisco, USA). from the Golgi to the endoplasmic reticulum to assess retrograde COPI-dependent transport29. Endocytosis of surface CTFR in polarized CFBE41o cells has also been Plasmids, mutagenesis and RNA interference. Constructs encoding for mutant described previously56. In particular, polarized monolayers were cultured to attain forms of rat ARFGAP1 fused to GST were generated by PCR, digested with a resistance of ∼400 cm2, which typically occurred after 4 days. EcoRI (for forward primers) or XhoI (for reverse primers) and inserted into the pGEX-6P1 vector (Amersham) using the Rapid ligation kit (Promega). Site-directed In vitro GAP assay. This assay was carried out essentially as previously described57. mutagenesis was carried out with the QuikChange site-directed mutagenesis kit Briefly, large unilamellar vesicles consisting of 40% phosphatidylcholine, 25% (Stratagene), and then verified by sequencing. Multi-site mutagenesis PCR was phosphatidylethanolamine, 15% phosphatidylserine, 10% phosphatidylinositol and carried out by addition of Taq ligase (New England Biolabs), 0.5× ligase buffer 10% cholesterol (Avanti Polar Lipids) were prepared by extrusion. Recombinant and mutant primers annealing only to the lower DNA strand. All constructs were ARF1 or ARF6 was loaded with [α-32P]GTP at 30 ◦C for 30 min in the presence verified by DNA sequencing (MWG Biotech). Other constructs have been previously of large unilamellar vesicles with 500 µM lipid in the GTP loading buffer (25 mM described, including ARFGAP1–Myc (ref. 20), mutant forms of the TfR cytoplasmic HEPES, at pH 7.4, 100 mM NaCl, 0.5 mM MgCl2, 1 mM EDTA, 1 mM ATP and domain as GST fusion constructs54, VSVG–KDELR tagged with yellow fluorescent 1 mM DTT). GTP hydrolysis was initiated by adding recombinant ARFGAP1 at protein29 (YFP) and clathrin light chain tagged with mCherry36. Additional different concentrations, either with or without AP-2 (175 nM), to [α-32P]GTP- constructs were obtained: GFP–ARFGAP1 (gift from D. Cassel, Technion, Israel), loaded ARF6 (1 µM) in GAP assay buffer (25 mM HEPES, at pH 7.4, 100 mM NaCl, and VSVG-ts045–GFP and VSVG-ts045–KDELR (J. Lippincott-Schwartz, NIH, 2 mM MgCl2, 1 mM GTP and 1 mM DTT). USA). Nucleotides 206–224 of human ARFGAP1 were targeted for siRNA (50- Electron microscopy. Cells grown on 35-mm glass-bottom dishes (MatTek ACAUUGAGCUUGAGAAGAU-30; Dharmacon), and nucleotides 297–318 of hu- corporation) were fixed with 1% glutaraldehyde in 100 mM HEPES at pH 7.4 man ARFGAP1 were targeted for shRNA (50-ACAGGAGAAGUACAACAGCAGA- overnight, washed several times with 100 mM HEPES at pH 7.4, once with 0 3 ; Thermal Scientific). A lentivirus expression system (Thermal Scientific) was used double-distilled water and post-fixed with 1% OsO4, 1.5% Na4Fe(CN)6 in 0.1 M for the stable expression of shRNA in BSC-1 cells. cacodylate buffer at pH 7.4 for 1 h at room temperature. Samples were washed with double-distilled water, dehydrated with ethanol and embedded in Epon. Serial Transfections. FuGene 6 (Roche) was used to transfect DNA plasmids, and sections were counterstained with 1% lead citrate and then observed by electron Oligofectamine (Invitrogen) or Lipofectamine RNAiMax (Invitrogen) was used to microscopy. Quantification of coated pits was carried out on random images taken transfect siRNA oligonucleotides, all according to manufacturers’ guidelines. at ×43,000 magnification.

Pulldown assays. GST fusion proteins on glutathione beads were incubated with Live-cell imaging by TIR-FM. Live-cell imaging experiments were carried out soluble proteins (4–10 nM) at 4 ◦C for 1 h in 0.5 ml of a buffer that contained 50 mM using BSC-1 cells transiently expressing GFP–ARFGAP1 together with clathrin HEPES (at pH 7.3), 300 mM NaCl, 90 mM KCl, 1 mM EDTA and 0.5% NP-40. After light chain A–mCherry. At 48 h post-transfection, the cells were plated on 25-mm- this incubation, beads were pelleted by centrifugation (1000g for 1 min at 4 ◦C) diameter glass coverslips (#1.5, Warner Instruments), allowed to spread for 3 h and followed by two washes with the incubation buffer and then analysed by SDS–PAGE then imaged. The coverslips were then washed with sterile PBS and immediately followed by western blotting. Coomassie blue staining was carried out to detect the placed into an open perfusion chamber into which pre-warmed Minimum Essential level of GST fusion proteins on beads. In experiments that assessed how GAP activity Medium alpha without phenol red (GIBCO) was then added. The chamber was modulates the binding of AP-2 to TfR, a modified buffer was used (to allow more inserted into a temperature-controlled sample holder (20/20 Technology) and then efficient GAP activity), which contained 50 mM HEPES (at pH 7.3), 300 mM NaCl, placed on the microscope stage enclosed by an environmental chamber (37 ◦C). 90 mM KCl, 1 mM MgCl2 and 0.5% NP-40. The cells were always kept in humidified 5% CO2. TIR-FM imaging was carried out using a Mariana system (Intelligent Imaging Innovations) based on a 200 M In vivo assays. To detect proteins associated with surface TfR, intact cells were inverted microscope (Zeiss) and a TIRF slider with manual angle control58 (Zeiss). initially bound by biotin-labelled Tf at 4 ◦C. Cells were then lysed followed by Fluorescent images were acquired with an oil immersion Alpha Plan-Apo ×100 incubation with beads coated with streptavidin. Beads were then washed and loaded objective (1.46 NA). Solid-state lasers were used as sources of excitation light at onto SDS–PAGE gel followed by immunoblotting. 488 nm (Sapphire, 50 mW, Coherent Inc.) and 561 nm (Jive, 50 mW, Cobolt Tf, EGF and LDL uptake assays were carried out using a fluorescence-based AB) coupled through an acousto-optical tunable filter into a single-mode fibre approach, as previously described32. Tf uptake was also quantified by a biochemical optic, carried to the TIRF slider and reflected towards the objective using a approach. Briefly, biotin-conjugated Tf was bound to intact cells at 4 ◦C, followed by multi-band dichroic mirror (Di01-R405/488/561/635, Semrock). The fluorescence washing with DMEM/PBS to release nonspecific binding. Cells were then warmed emission light was collected by the objective, then passed through the dichroic to 37 ◦C to initiate endocytosis for the times indicated. Cells were then washed with mirror and a multi-band emission filter (FF01-446/523/600/677, Semrock). A ice-cold DMEM/PBS to arrest transport, followed by washes with an acid-based spherical aberration correction unit (Intelligent Imaging Innovations) was inserted

DOI: 10.1038/ncb2221 METHODS into the emission path. Images were acquired with an electron multiplying 53. Li, J. et al. An ACAP1-containing clathrin coat complex for endocytic recycling. J. charge-coupled device (EMCCD) camera (QuantEM, Photometrics). Microscope Cell Biol. 178, 453–464 (2007). operation and image acquisition were controlled using SlideBook software (V4.2.12, 54. Dai, J. et al. ACAP1 promotes endocytic recycling by recognizing recycling sorting Intelligent Imaging Innovations). The time-lapse images were acquired sequentially signals. Dev. Cell 7, 771–776 (2004). for each wavelength using exposures of 30 ms and each set was acquired at 3 s 55. Trucco, A. et al. Secretory traffic triggers the formation of tubular continuities across intervals. The angle of incidence of the excitation light onto the coverslip was Golgi sub-compartments. Nat. Cell Biol. 6, 1071–1081 (2004). 56. Peter, K. et al. Ablation of internalization signals in the carboxyl-terminal tail of adjusted to generate an evanescence field (estimated penetration depth between 100 the cystic fibrosis transmembrane conductance regulator enhances cell surface and 200 nm). expression. J. Biol. Chem. 277, 49952–49957 (2002). 57. Che, M. M., Nie, Z. & Randazzo, P. A. Assays and properties of the Arf GAPs AGAP1, Statistical analysis. Student’s t-test was used to analyse statistical significance. ASAP1, and Arf GAP1. Methods Enzymol. 404, 147–163 (2005). 58. Saffarian, S. & Kirchhausen, T. Differential evanescence nanometry: live-cell 52. Matsui, W. & Kirchhausen, T. Stabilization of clathrin coats by the core of the fluorescence measurements with 10-nm axial resolution on the plasma membrane. clathrin-associated protein complex AP-2. Biochemistry 29, 10791–10798 (1990). Biophys. J. 94, 2333–2342 (2008).

SUPPLEMENTARY INFORMATION

DOI: 10.1038/ncb2221 Figure S1

a c

ARFGAP1 ARFGAP1

GAPDH GAPDH

b d

30000 100 25000 80

20000 60 15000 40 10000 20 5000 Tf uptake (%) 0 0

Internalized Tf (fluorescence units) Internalized Control siRNA control shRNA

e 70 60 50 40 Control siRNA 30

Tf uptake (%) 20 10 0 0 5 10 Time (min)

Figure S1 Effects of depleting ARFGAP1. a. Depletion of ARFGAP1 by immunoblotted for proteins as indicated. d. Tf uptake is inhibited by siRNA. BSC-1 cells were treated with a siRNA sequence that targeted shRNA against ARFGAP1. BSC-1 cells that stably expressed shRNA primate ARFGAP1. Cells were lysed and then immunoblotted for proteins conditions as indicated were assayed for the uptake of biotin-Tf at 10 as indicated. b. An alternate assay for Tf uptake also reveals inhibition minutes, followed by quantitation. The mean from three experiments upon siRNA against ARFGAP1. BSC-1 cells were incubated continuously with standard error is shown. Difference between the two conditions with Tf in medium at 37oC. Quantitation was then performed. The is significant (p<0.05). e. Kinetics of Tf uptake upon siRNA against mean with standard error from three experiments is shown. Difference ARFGAP1. BSC-1 cells were treated with siRNA, followed by quantitation between the two conditions is significant (p<0.05). c. Depletion of of Tf uptake at times as indicated. The mean from three experiments with ARFGAP1 by shRNA. BSC-1 cells that stably expressed a shRNA sequence standard error is shown. Difference between the two conditions (except targeting another region in the primate ARFGAP1 were lysed, and then time=0) is significant (p<0.05).

SUPPLEMENTARY INFORMATION

Figure S2

β-COP

GSTs

b 301 TSDRPLEGHS YQNSSGDNSQ NSTIDQSFWE TFGSAEPPKA KSPSSDSWTC 351 ADASTGRRSS DSWDIWGSGS ASNNKNSNSD GWESWEGASG EGRAKATKKA 401 APSTAADEGW DNQNW

Figure S2 Characterizing ARFGAP1 binding to partner proteins. a. Pulldown assays using different truncations of ARFGAP1 to assess interactions with coatomer. The different forms of ARFGAP1 as GST fusion proteins were bound to beads, incubated with cytosol, and then immunoblotted for proteins as indicated. GST fusion proteins were detected by Coomassie staining. b. Amino acid sequence of ARFGAP1 (residues 301-415), with the tryptophan-based motifs shown in bold.

SUPPLEMENTARY INFORMATION

Figure S3

a b

Control Control 80 80 shRNA shRNA

60 60

40 40 Colocalization of

Colocalization of 20 20 Dextran / Lamp1 (%) Dextran / EEA1 (%)

0 0 5 min 15 min 15 min 30 min 45 min 60 min

c d 100 Control 100 Control

shRNA shRNA 80 80 60 60 40 40 CTB / EEA1 (%) Colocalization of

20 Colocalization of CTB / TGN46 (%) CTB / 20

0 0 2 min 10 min 15 min 30 min 45 min

Figure S3 Effect of depleting ARFGAP1 on endocytic pathways. a. Depletion is insignificant (p>0.05). c. Depletion of ARFGAP1 does not affect raft- of ARFGAP1 does not affect fluid-phase uptake followed by transport dependent endocytosis followed by transport to the early endosome. to the early endosome. Fluorescence-conjugated dextran was added to Fluorescence-conjugated cholera toxin B subunit (CTB) was added to BSC-1 BSC-1 cells for times indicated, and then colocalization with an early cells for times indicated, and then colocalization with an early endosome endosome marker (EEA1) was quantified at times indicated. The mean marker (EEA1) was quantified at times indicated. The mean from three from three experiments with standard error is shown. Difference at all experiments with standard error is shown. Difference at all time points is time points is insignificant (p>0.05). b. Depletion of ARFGAP1 does not insignificant (p>0.05). d. Depletion of ARFGAP1 does not affect transport affect transport from the early endosome through the late endosome and from the early endosome to the TGN. Fluorescence-conjugated CTB was then to the lysosome. Fluorescence-conjugated dextran was added to added to BSC-1 cells for times indicated, and then colocalization with a TGN BSC-1 cells for times indicated, and then colocalization with a lysosomal marker (TGN46) was quantified at times indicated. The mean from three marker (Lamp1) was quantified at times indicated. The mean from three experiments with standard error is shown. Difference at all time points is experiments with standard error is shown. Difference at all time points insignificant (p>0.05).

SUPPLEMENTARY INFORMATION

Figure S4

b Control siRNA siRNA siRNA siRNA

Figure S4 Effects of depleting ARFGAP1 on Golgi structure. a. The Golgi structure is largely intact upon the depletion of ARFGAP1. HeLa cells were treated with siRNA as indicated, and then examined for a Golgi marker (giantin) by immunofluorescence microscopy; bar, 10 um. b. The cisternal morphology of the Golgi is largely intact upon the depletion of ARFGAP1. HeLa cells were treated with siRNA as indicated, followed by examination by electron microscopy; bar, 200 nm.

SUPPLEMENTARY INFORMATION Figure S5

a 100 80 60 40

Tf uptake (%) 20 0 l A 1 ro N P t hR GA Con s F AR P- GF b

13 23

WT:MMDQARSAFSNLFGGEPLSYTRFSLARQVDGDNSHVEMKLAVDEEENADNNTKANVTKPKR

N31:MMDQARSAFSNLFGGEPLSYTRFSLARQVDG DNSHVEMKLAVDEEENADNNTKANVTKPKR N31

N19:MMDQARSAFSNLFGGEPL SYTRFSLARQVDGDNSHVEMKLAVDEEENADNNTKANVTKPKR N19

c d

controlshRNAwt R50K ARFGAP1

GAPDH

Figure S5 Miscellaneous further characterizations. a. Rescue of defective Tf phenylalanines that anchor the two distinct internalization sorting signals uptake by a GFP-tagged form of ARFGAP1. BSC-1 cells that stably expressed in TfR are indicated by arrowheads. c. Assessing efficiency of depleting shRNA against ARFGAP1 were transfected with the rat form of GFP-tagged ARFGAP1 and expression of different rescue constructs. Immunoblotting was ARFGAP1. Uptake of biotin-Tf at 10 minutes was then quantified. The performed on whole cell lysates to assess the level of proteins as indicated. d. mean with standard error from three experiments is shown. Difference Comparison of GAP activity of ARFGAP1 toward ARF6 versus ARF1. The GAP between shRNA and all other conditions is significant (p<0.05). b. Truncation assay was performed using ARF6 (triangles) or ARF1 (squares) as substrate. forms of TfR used to assess direct binding by ARFGAP1 and AP-2. Critical The mean from three experiments with standard error is shown.

SUPPLEMENTARY INFORMATION Figure S6

Fig 1b

160 105 250 75 160 105 50 75

Fig 2a Fig 2b 250 160 -cop

160 105

Fig 2c 250 Fig 2d 160

160 AP2 105 75

160 -cop 105

Coomassie staining

Fig 3a,b Fig 3c

IB: CHC IB: CHC IB: AG1

IB: AP2 Coomassie staining

Coomassie staining

Figure S6 Full gel scans. Specific gels shown are indicated.

SUPPLEMENTARY INFORMATION

Fig 6a Figure S7

Fig 6b Fig

IB: TfR 6b

Fig 6c Fig 6d

83 62 47.5

47.5 IB: myc IB: HA IB: CD8 32.5 Fig 7a Fig 7b,c

62 62 47.5 47.5

Coomassie staining Coomassie staining

Fig 7d Fig 7e

175 175 83 IB: TfR 83

83 IB: AP2

62 83 IB: AG1 47.5 47.5 IB: AG1

Figure S7 Additional full gel scans. Specific gels shown are indicated.

SUPPLEMENTARY INFORMATION

Figure S8 Fig 8d Fig 8e

25 16.5

IB: Arf6 83 62 47.5

Coomassie staining Coomassie staining

Fig 8f Fig 8g 47.5 IB: AG1 25 32 16.5 IB: Arf6 25 16.5 IB: Arf6

Coomassie 175 staining 83

Fig 8h 175 83

175 83 IB: AP2

25 16.5 IB: Arf6

Figure S8 Additional full gel scans. Specific gels shown are indicated.

Supplementary Table 1 | ARFGAP1 interacting proteins identified by mass spectrometry.

PREDICTED PROTEIN APPROXIMATE PROTEIN MASS (kDa) clathrin, heavy polypeptide 193 adaptor-related protein complex 2, alpha 109 subunit adaptor-related protein complex 2, beta subunit 107 coatomer protein complex, gamma subunit 98 nuclear scaffold protein P130/MAT3 95 heat shock 71K 71 beta tubulin 50 adaptor-related protein complex 2, mu subunit 50 predicted similar to centrosome protein cep290 289 centrosomal protein 4 133