Regulation of Growth Factor Receptor Degradation by ADP-Ribosylation Factor Domain Protein (ARD) 1

Regulation of growth factor receptor degradation by ADP-ribosylation factor domain protein (ARD) 1

Victor Meza-Carmen1, Gustavo Pacheco-Rodriguez, Gi Soo Kang, Jiro Kato, Chiara Donati2, Chun-Yi Zhang3, Alessandro Vichi, D. Michael Payne, Souheil El-Chemaly4, Mario Stylianou, Joel Moss, and Martha Vaughan5

Cardiovascular and Pulmonary Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD 20892

Contributed by Martha Vaughan, March 9, 2011 (sent for review October 10, 2010)

ADP-ribosylation factor domain protein 1 (ARD1) is a 64-kDa pro- lular sites specified by targeting sequences, posttranslational tein containing a functional ADP-ribosylation factor (GTP hydro- modifications, and/or associated molecules (6, 7). Mammalian lase, GTPase), GTPase-activating protein, and E3 ubiquitin ligase ARFs (ca. 20 kDa) are grouped as class I (ARFs 1–3), class II domains. ARD1 activation by the guanine nucleotide-exchange (ARFs 4 and 5), and class III (ARF6), based on similarities of factor cytohesin-1 was known. GTPase and E3 ligase activities of gene structure, protein sequences, and phylogenetic relationships ARD1 suggest roles in protein transport and turnover. To explore (8). The ARFs cycle between active GTP- and inactive GDP- this hypothesis, we used mouse embryo fibroblasts (MEFs) from bound forms dependent on the actions of GEFs, which accelerate ARD1-/- mice stably transfected with plasmids for inducible expres- the replacement of ARF-bound GDP with GTP, and GTPase- sion of wild-type ARD1 protein (KO-WT), or ARD1 protein with activating proteins (GAPs) that enhance intrinsic ARF GTPase inactivating mutations in E3 ligase domain (KO-E3), or containing hydrolysis of GTP (9). Activated ARF–GTP can interact stably persistently active GTP-bound (KO-GTP), or inactive GDP-bound with specific effector molecules, such as vesicle coat protein sub- (KO-GDP) GTPase domains. Inhibition of proteasomal proteases units of COPI coatomer complex (10), and Golgi-associated, in mifepristone-induced KO-WT, KO-GDP, or KO-GTP MEFs resulted γ-adaptin ear-containing, ARF-binding adapter proteins of cla- in accumulation of these ARD1 proteins, whereas KO-E3 accumu- thrin-coated vesicles (6). lated without inhibitors. All data were consistent with the conclu- ADP-ribosylation factor domain protein 1 (ARD1), an atypical

sion that ARD1 regulates its own steady-state levels in cells by member of the ARF family, is a ca. 64-kDa molecule with a BIOCHEMISTRY autoubiquitination. Based on reported growth factor receptor- C-terminal (amino acids 391–565) approximate 18-kDa ARF cytohesin interactions, EGF receptor (EGFR) was investigated in domain that is ca. 60% identical to class I ARFs (11). As ARF induced MEFs. Amounts of cell-surface and total EGFR were higher proteins lack detectable GTPase activity and require a GAP to in KO-GDP and lower in KO-GTP than in KO-WT MEFs, with levels terminate activation, the intrinsic GTPase activity of ARD1 was in both mutants greater (p ¼ 0.001) after proteasomal inhibition. soon recognized (12). Direct functional interaction of recombi- Significant differences among MEF lines in content of TGF-β recep- nant 18-kDa ARF domain and ca. 46-kDa remainder (GAP do- tor III were similar to those in EGFR, albeit not as large. Differences main) of ARD1 was also shown (12). The 46-kDa N-terminal in amounts of insulin receptor mirrored those in EGFR, but did not region of ARD1 (1–390), termed “GAP domain,” although it in- reach statistical significance. Overall, the capacity of ARD1 GTPase cludes sequences with additional diverse functions (Fig. 1A), can to cycle between active and inactive forms and its autoubiquitina- activate ARF-domain GTPase activity to generate ARD1–GDP tion both appear to be necessary for the appropriate turnover of (12, 13). Three structural motifs (RING, B box, and coiled-coil or EGFR and perhaps additional growth factor receptors. RBCC) in this region gave ARD1 the designation TRIM23 in the tripartite motif (TRIM) family (14). RBCC sequences participate tyrosine receptors ∣ ubiquitination in protein–protein interactions, and the ARD1 RING exhibits E3 ubiquitin ligase activity demonstrated as apparent autoubiqui- esicular trafficking, which is responsible for intracellular tination in assays with recombinant ARD1 or fragments, E1, and Vtransport of proteins and/or lipids between and among orga- an appropriate E2, plus ubiquitin and ATP (15). A recent report nelles, is of major importance in regulating cell function and fate that the protein NF-κB essential modulator (NEMO) is a sub- (1). Extracellular stimuli, either via direct ligand interaction with strate for the E3 activity of ARD1 (16) did not relate that activity specific cell-surface receptors (e.g., growth factor) or in response to its ARF domain. to mechanical or environmental signals (e.g., pH, oxygen), also To explore the biological function(s) of ARD1, we prepared use vesicular trafficking to produce needed alterations in cellular from ARD1-null mice mouse embryo fibroblast (MEF) lines stably metabolism, growth, and/or motion (2). For example, after ligand transfected with plasmids for mifepristone (Mfp)-inducible expres- binding, plasma membrane receptors, such as the epidermal sion of ARD1, wild type, or with specific mutations that prevent growth factor receptor (EGFR) are internalized in clathrin- or caveolin-coated vesicles (3) and can be recycled, after ligand dissociation, to the plasma membrane via multivesicular bodies. Author contributions: V.M.-C., G.P.-R., J.M., and M.V. designed research; V.M.-C., G.P.-R., G.S.K., J.K., C.D., C.-Y.Z., A.V., D.M.P., and S.E.-C. performed research; V.M.-C., G.P.-R., M.S., Receptor degradation after covalent modification (e.g., phos- J.M., and M.V. analyzed data; and V.M.-C., G.P.-R., J.M., and M.V. wrote the paper. phorylation and/or ubiquitination) can occur through lysosomal The authors declare no conflict of interest. or proteasomal pathways (4). 1Present address: Instituto de Investigaciones Químico-Biológicas, Universidad Michoacana Vesicle formation is initiated by guanine nucleotide-exchange de San Nicolás de Hidalgo, Morelia, Michoacán, 58030, Mexico. factor (GEF)-catalyzed activation of ADP-ribosylation factor 2Present address: Dipartimento di Scienze Biochimiche, Università di Firenze, 50134 (ARF) that results in its specific association with membrane sites Florence, Italy. – where ARF GTP recruits coat proteins and cargo, resulting in 3Present address: Epitomics, Inc., 863 Mitten Road, Suite 103, Burlingame, CA membrane deformation and initiation of budding. After move- 94010-1303. ment of a completed vesicle to its target and tethering/docking, 4Present address: Division of Pulmonary and Critical Care Medicine, Brigham and Women’s each transport step ends with membrane fusion (5). Biogenesis of Hospital, 75 Francis Street, Boston, MA 02115. several types of vesicles uses specific guanine nucleotide-binding 5To whom correspondence should be addressed. E-mail: [email protected]. GTPases as molecular switches for initiation and termination of This article contains supporting information online at www.pnas.org/lookup/suppl/ diverse pathways that deliver molecules to appropriate intracel- doi:10.1073/pnas.1103867108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1103867108 PNAS Early Edition ∣ 1of6 Downloaded by guest on September 26, 2021 ARD1 and mutants (see Fig. 1A) of the ARF domain that were persistently inactive or active. Cells constitutively overexpressing ARD1 protein failed to proliferate, which seems likely, in retro- spect, to have been a consequence (at least in part) of elevated levels of its E3 ligase activity. Endogenous ARD1 was finally detected by two different antibodies in HepG2 cells grown overnight with, but not without, proteasome inhibitors (Fig. 1B, arrowhead). This behavior mirrors reports of other RING-type E3 ubiquitin ligases capable of autoubiquitination that causes their instability in cells (17, 18) and detection only after inhibition of proteasomal proteases. Results were similar with fibroblasts grown from WT MEFs, although KO MEFs always, unfortu- nately, showed extensive nonspecific staining with our antipeptide antibodies (Fig. 1B). Our ARD1 KO mouse (C57/BL6) was initially generated for alternative exploration of the biological role of ARD1. Mice lacking the ARD1 gene appeared grossly normal and thus far present no obvious pathological phenotype or histopathology of major organs including brain, spleen, heart, kidneys, lung, and axial lymph nodes, or problems with breeding. Fatty liver without iden- tified cause appeared in some older KO mice. ARD1-null MEF lines that inducibly express WT or mutant ARD1 proteins under control of the CMV promoter were, therefore, prepared.

Induced Expression of ARD1 in KO MEFs. RT-PCR confirmed apparent absence of ARD1 mRNA in KO MEFs (Fig. S1). We used ARD1-/- MEFs for Mfp-inducible expression of murine WT ARD1 or its mutant forms, which do not cycle between GTP- and GDP-bound forms. Mutation T418N interferes with GTP binding, locking ARD1 in its inactive GDP-bound form, whereas K458I abolishes GTPase activity, producing persistently active ARD1–GTP (19). The Mfp-induced double E3 mutant with alanine replacing C34 and H54, amino acids critical for E3 ubiquitin Fig. 1. ARD1, a multifunctional protein. (A) Functional domains of 574 ligase activity, was also generated (15). We report data from eight amino acid human ARD1 with C-terminal (402–574) ARF sequence. ARD1 stably transfected MEF lines referred to, respectively, as KO-WT, contains an E3 ubiquitin ligase motif (positions 31–75) with critical C34 and KO-GDP, and KO-GTP (two lines of each), and one line each of H53, which were replaced with alanine in E3-inactive mutants. GAP function KO-E3 and KO-GFP (empty vector). No ARD1 protein was de- was assigned to amino acid 101–190, with amino acid 190–333 additionally tected in any of these MEFs before induction (Fig. 1C), but after required for domain interactions that result in GTP hydrolysis. A region 18 h with 10 μM Mfp, Western blot analysis revealed ARD1 in all required for lysosomal targeting (amino acid 301–402) includes amino acid – cells (except KO-GFP) with a KO-E3 level sevenfold that of 387 402 corresponding to the N-terminal amphipathic helix of an ARF C molecule, which can act as a GDP-dissociation inhibitior (GDI). Mutations KO-WTand the ARF-domain mutants (Fig. 1 and Fig. S1). Pro- in the ARF region (402–574) that abolish GTP binding (T418N) or GTPase teasome inhibition during Mfp induction increased amounts of activity (K458I) are shown. (B) Effect of proteasome inhibitors on ARD1 con- overexpressed proteins dramatically in all lines except KO-E3, tent in HepG2 cells and MEFs. HepG2 cells were incubated 18 h without (0) or which was increased ≤100% (Fig. 1C). Despite the large differ- with (+) lactacystin and MG132, before preparation of lysates and Western ence in ARD1 protein levels between KO-E3 and the other MEF blotting of proteins (40 μg) with ARD1 antibodies 5525 or 5662. MEFs lacking lines, ARD1 mRNA levels did not differ among these cell lines ARD1 (KO) or from wild-type mice (WT) were incubated ca. 18 h without or (except for KO-GFP, in which no ARD1 mRNA was detected) with proteasome inhibitors before Western blotting using ARD1 antibodies (Fig. S1). 5662. Arrowhead, endogenous ARD1 protein. (C) ARD1-/- MEF lines stably transfected with indicated GFP–ARD1-V5/His constructs (KO-WT, KO-GDP, KO-GTP, KO-E3) were incubated 18 h without (0) or with (+) Mfp or Mfp plus Intracellular Localization of ARD1. Consistent with results of lactacystin and MG132 before Western blotting (40 μg proteins) with antibo- Western blotting after induction without proteasome inhibition, dies against V5 or GAPDH. Results were replicated in three experiments. ARD1 (GFP-tagged) was barely visible by fluorescence micro- (D) MEFs were incubated with Mfp without or with proteasome inhibitors scopy (Fig. 1D) in any MEFs except KO-E3, where it was clearly (as in C) before fixation and preparation for inspection of EGFP–ARD1 by more abundant in multiple scattered puncta. With proteasome confocal immunofluorescence microscopy. inhibition, however, GFP–ARD1 was seen in punctate areas throughout cytoplasm of all MEFs with discrete concentrations usual function of the ARF or E3 ligase domains. Here, we report of KO-WT and KO-GDP proteins in a perinuclear region. evidence that continuous function of the ARF-domain GTPase cy- Because ARD1 has E3 ligase activity and had been associated cling is likely required for degradation of endogenous EGF and with lysosomes (13, 20), localization of different recombinant TGF-β III; receptors in these MEFs. EGFR degradation, although ARD1 proteins with that of LysoTracker was assessed. WT and involving proteasomes, was seemingly independent of ARD1 E3 GDP- or GTP-bound ARD1 seemed to colocalize with Lyso- ubiquitin ligase activity that catalyzes its automodification. Tracker, but the E3 ligase mutant did not coincide with either LysoTracker (Fig. S2) or lysosomal membrane protein LAMP1 Results and Discussion (Fig. S3). E3 ligase mutant ARD1 in cell fractions separated Endogenous ARD1 Is Unstable in Cells. For a decade, virtually all of by centrifugation appeared to be largely soluble (Fig. S3), our numerous attempts to detect ARD1 in many cells had failed although a significant portion was pelleted after centrifugation consistently. To identify an ARD1 function, we had transiently for 18 h at 100;000 × g. These data are consistent with the notion transfected several cell lines with plasmids encoding wild-type that E3 mutant ARD1 protein was associated with intracellular

2of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1103867108 Meza-Carmen et al. Downloaded by guest on September 26, 2021 structures resembling exosomes (i.e., sedimented only after prolonged high-speed centrifugation), although a considerable amount of the overexpressed protein in MEFs did remain “soluble.” Analysis of protein composition of the “exosome” fraction is clearly important. When ARD1 E3 ubiquitin ligase activity was reported (15), its autoubiquitination was suggested, despite failure of the multiu- biquitinated protein to react with ARD1 antibodies. Evidence here of ARD1 accumulation in cells treated with proteasome inhibitors (Figs. 1 C and D) is consistent with autoubiquitination of intracellular ARD1 and seems to parallel descriptions of MARCH7 (21) and MARCH5 9 (17) as E3 ligases requiring enzymatic deubiquitination for stabilization or detection. Thus, it appears that ARD1 self-regulates its degradation, presumably subject to additional molecular interactions that also influence ARD1 stability and thereby biological function.

ARD1 Effects on EGFR mRNA and Protein. Our studies of ARD1 molecular interactions had revealed its association with the GEF cytohesin-1 (19), which activated its ARF domain. As this cytohesin and others (cytoheins-2, -3) have been implicated in mod- ulating growth factor receptor function (22–24), in addition to activating ARFs, we explored a relationship of ARD1 to kinase growth factor receptors (R), first EGFR. Levels of EGFR mRNA (Fig. 2A) were similar in KO-WT, KO-GDP,KO-GTP,and KO-E3 MEFs. There were, however, clear differences among MEF lines (Fig. 2 B and C) in amounts of EGFR protein. EGFR levels in 60% ca. KO-GDP were > higher, and in KO-GTP 50% lower than BIOCHEMISTRY those in KO-WT (Fig. 2C). Overall, lack of significant differences in amounts of mRNA perhaps suggests rapid turnover of EGFR protein that enables precise regulation, temporal and spatial, of EGF action. EGFR protein content seemed most related to the activation state of the ARD1 ARF domain, whereas absence of ARD1 E3 ligase activity had no effect. The E3 ligase domain appears responsible for autoubiquitination and, likely, regulation of ARD1 degradation, as well as accumulation of the E3 ligase mutant and other ARD1 proteins after proteasome inhibition. It is tempting to speculate that the ARF domain of ARD1 acts Fig. 2. EGFR mRNA and protein in ARD1 KO MEFs overexpressing WT or similarly to the class I ARFs in vesicular transport, although we mutant ARD1 proteins. (A) EGFR mRNAs in stably transfected MEFs induced do not know how ARD1 interaction with membranes is regulated. with Mfp were quantified by real-time PCR. RNA values were normalized to To determine whether amounts of EGFR at the plasma mem- α-tubulin (also used for protein normalization in experiments below). Data brane paralleled total cell content, we quantified biotinylated were analyzed using Sequence Detector version 1.6.3 program. Data are 1∕2 cell-surface EGFR. Even with large differences in total EGFR averages range for each cell line in two independent experiments. No significant differences were detected. (B) Samples (40 μg) of lysate proteins content of cells overexpressing different ARD1 proteins, – from eight MEF lines induced with Mfp were analyzed by Western blotting amounts at the cell surface were 79 83% of the total in all MEFs with EGFR and GAPDH antibodies. (C) For EGFR, each densitometric value (Fig. 2D), consistent with a functional relationship between was expressed relative to that of KO-GFP cells in the same experiment equal GTPase cycling of ARD1 and EGFR cycling via cell surface to to 100. Data are means SEM of values for each line in three experiments. degradation after endocytosis. Fruitless interaction of endogenous *p < 0.001 for difference between each of lanes 3 and 4 (KO-GDP) or 5 and GEF (cytohesin-1) with overexpressed ARD1–GDP that could not 6 (KO-GTP) from KO-GFP. (D) Biotinylated proteins from MEF lines induced be activated would interfere with cytohesin-1 activation of endo- with Mfp, were separated for quantification of immunoreactive EGFR in cell Ca. genous ARD1 (presumably absent in these KO cells) and addi- surface (B) and in total lysate (T) samples (2.5% of each). 80% of EGFR was biotinylated in each cell line regardless of differences in total amounts. tional ARF substrates. Overexpression of ARD1–GTP may have permitted more EGFR degradation than did similar amounts (Fig. 1C) of other ARD1 mutants, although its capacity to support degradation (2), we looked for accumulation of EGFR in MEFs EGFR transport is presumably limited to a single turnover by the treated with proteasome (lactacystin and MG132) or lysosome inability of this ARD1 to hydrolyze GTP. EGFR activation and (chloroquine) inhibitors (Fig. 3A). EGFR content was clearly subsequent degradation requires ligand binding and probably greater after incubation of KO-GDP or KO-GTP cells with pro- – homodimerization (25 27), although the existence of active dimers teasome inhibitors (Fig. 3B), but levels of EGFR in KO-WT or even in the absence of ligand has been proposed (28), so that con- KO-E3 cells were not appreciably affected and chloroquine centrated EGFR, for example, in plasma membranes of KO-GDP lacked significant effects. Multiplicity of ubiquitinated proteins MEFs could become activated independent of EGF. A role for accumulated in all MEF lines (Fig. 3C) with proteasomal degra- ARD1 in regulation of EGFR transport begins to emerge from exploring molecular interactions of the overexpressed proteins, dation inhibited obscures detection of differences in relatively although caution in extrapolation from such observations to the small contributions from EGFR or other specific protein(s). Most behavior of endogenous molecules is always required. of the EGFR appeared to be targeted for proteasomal degradation, possibly with a role for ARD1 GTPase action in its delivery Effects of Proteasome and Lysosome Inhibitors on Degradation of to proteasomes. All observations are consistent with ARD1 par- EGFR. As EGFR is subject to both proteasomal and lysosomal ticipation in EGFR degradation.

Meza-Carmen et al. PNAS Early Edition ∣ 3of6 Downloaded by guest on September 26, 2021 Fig. 4. Insulin and TGF-β receptor proteins in ARD1–KO MEFs overexpres- Fig. 3. Effects of proteasome and lysosome inhibitors on EGFR content of sing WT or mutant ARD1. (A) Samples (40 μg) of proteins from eight MEF MEFs expressing ARD1 proteins. (A) After induction with Mfp without (C) lines containing an indicated ARD1 or empty vector (KO-GFP) construct and or with proteasome (P) or lysosome (L) protease inhibitors, the indicated induced with Mfp were analyzed by Western blotting with antibodies ARD1 KO MEFs were analyzed by Western blotting with antibodies against against insulin or TGF-β receptor. Densitometric values in each experiment EGFR. (B) Densitometric values for EGFR in P and L samples expressed relative were expressed relative to that of the same protein in KO-GFP cells equal to to that of α-tubulin in the same sample were normalized to the value for the 100. Data are means SEM of values from three experiments. The amounts same cells (C) without inhibitors equal to 100%. Data are from three experi- of insulin R data (B) did not reach significance. * p < 0.01 and p < 0.05 for ments. Only after Mfp induction of ARD1–GDP and ARD1–GTP mutants with differences, respectively, between amounts of TGF-βR(C) in KO-GDP and proteasome inhibitors were ARD1 levels significantly (p < 0.05) different KO-GTP MEFs. from (higher than) those of the same cells without inhibitors. (C) Immunoblot from a representative experiment with antibodies against ubiquitin (Ub) perhaps insulin, seems clearly consistent with our findings, and and α-tubulin. the effects described here were demonstrably dependent on con- Effects on Insulin and TGF-β Receptors. Our EGFR data prompted tinued cyclic activation of its ARF domain. us to also evaluate effects of ARD1 action on levels of endogen- It may be relevant that cytohesin-1, the only known ARD1- ous insulin receptor (IR) and TGF-βR in MEF lines. Findings GEF (19, 34), was translocated to plasma membrane of PC-12 were similar to those for EGFR, without differences among cells (derived from rat adrenal pheochromocytoma) in response KO-GFP (empty vector), KO-WT, and KO-E3, but with signifi- to EGF and nerve growth factor (22, 24). Perhaps translocation cantly more TGF-βR in KO-GDP and less in KO-GTP MEFs of cytohesin-1 to the plasma membrane in response to specific than in KO-GFP (Fig. 4C). Although the same trend was appar- stimuli is associated with activation of ARD1, which was influ- ent for the IR (Fig. 4B), the differences were not statistically enced by specific phospholipids (34). In addition, ARAP1, an significant. Each of these kinase receptors is activated and inter- ARF GTPase-activating protein, regulates EGFR endocytosis, nalized after agonist binding, but pathways of their endocytosis trafficking, and signaling (35, 36), although the ARD1 molecule and signaling from endosomes thereafter are not identical does contain its own GAP domain. All data are consistent with (29, 30). Some of the numerous differences in endosomal signal- the participation of ARFs, including ARD1, in trafficking and ing mechanisms among tyrosine kinase, G-protein-coupled, and function of plasma-membrane-associated receptors. In addition, toll-like receptors were reviewed by Murphy et al. (29). Differ- the finding that ARD1 participates in antiviral defense by ubiqui- ences in usage of signaling pathways also exist among growth tinating NEMO suggests that ARD1 has more than one function factor receptors, which we had thought potentially similar enough related to innate immunity (16). to allow identification of shared ARD1 actions. Understanding of such ARD1 function(s), however, may await elucidation of spe- Materials and Methods β cific molecular interactions of each of those receptors. TGF- R Cell Culture. HepG2 cells were grown at 37 °C (5% CO2/95% humidity) in III, a TGF-β coreceptor also degraded in proteasomes (31), is DMEM high glucose (4.5 g∕L D-glucose, Invitrogen) with 50 units∕mL peni- internalized along with TGF-βR I and II (31–33). Our data are cillin, 50 mg∕mL streptomycin, and 10% FBS (Hyclone). MEF1 medium con- consonant with the notion that ARD1 GTPase could have a role tained DMEM high glucose, with 10% mesenchymal growth supplement in regulating internalization of and signaling by dimeric or mono- (MCGS, Lonza) replacing FBS, penicillin, and streptomycin. MEFs stably trans- meric growth factor receptors. An action of ARD1 in the regula- fected with ARD1 plasmids were grown in MEF2 [MEF1 plus Zeocin, tion of growth factor receptors such as those of EGF, TGF-β, and 0.8 mg∕mL (Invitrogen) and hygromycin B, 0.2 mg∕mL, (Invitrogen)].

4of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1103867108 Meza-Carmen et al. Downloaded by guest on September 26, 2021 Generation and Growth of MEFs. The animal studies were performed accord- ARD1 protein or Ab no. 5562 against ARD1 (301–319, RQEEMALSVVDAHV- ing to the animal protocol approved by the National Institutes of Health, REKLC), were diluted 1∶1;000. Horseradish peroxidase-linked secondary anti- National Heart, Lung, and Blood Institute animal user and care committee. bodies (Promega) were detected using Supersignal West Pico (or Femto) Day 14 embryos of WT or ARD1-/- (KO) C57BL6 mice were euthanized with Chemiluminescent Substrate Kits (Pierce). Images captured using a LAS4000 CO2 and washed in sterile phosphate-buffered saline. Skin was removed, system were quantified with ImageGauge 4.0 (Fujifilm). minced, incubated (30 min, 37 °C) in 2 mL of trypsin/EDTA (Lonza), washed in PBS, and incubated for 2 wk in MEF1 medium (20 mL). Adherent MEFs Fluorescence Microscopy. MEFs (3.5 × 104) in 0.5 mL of MEF2 in four-well Lab- were split one-fifth at 80% of confluence for continued growth in the same TeK II chamber slides were allowed to adhere overnight (15 h) before incuba- – medium and studies at passages 10 20. tion (18 h) with 10 μM Mfp in MEF2 without or with MG132 and lactacystin, Genomic DNA from MEFs (Wizard Genomic DNA Purification Kit, Prome- 50 μM each. After washing with PBS containing CaCl2 and MgCl2 (1 mM each), ca. ga) was template for PCR with primers F1-G and R1-G to generate an cells were fixed (20 min) with 4% paraformaldehyde (Electron Microscopy, 2.4-kb WT ARD1 allele fragment or F1-G and R-NEO for an 2.1-kb fragment Inc.), washed with PBS, covered with Vectashield (Vector Laboratories) con- from the neomycin-disrupted allele (Table S1). DNA was amplified using the taining DAPI, and sealed with coverslips or processed for immunostaining Advantage 2 PCR System (BD Clontech) for 30 cycles of 40 s, 95 °C∕45 s, as described (19). To visualize lysosomes (acidic interior), cells were incubated 60 °C∕4 min, 68 °C, followed by 10 min, 68 °C. PCR products were separated with the fluorophore 50 nM LysoTracker for 15 min at 37 °C. Images were col- in 1% agarose gel containing ethidium bromide and viewed by ultravio- lected using a Zeiss LSM 510 laser scanning confocal microscope. let light.

Biotinylation of Cell-Surface Proteins. A Cell-Surface Protein Isolation Kit Preparation of Murine ARD1 cDNA, Inducible Plasmids, and Stably Transfected (Pierce) was used for biotinylation of proteins according to the manufac- MEFs. WT mouse ARD1 cDNA, reverse transcribed from total RNA isolated turer’s protocol. Briefly, 5.2 × 106 MEFs were grown overnight (ca. 15 h) in (RNeasy Mini Kit, Qiagen) from a 100-mm dish of 80–90% confluent of WT MEF2, incubated 18 h in MEF2 with 10 μM Mfp, rinsed three times (20 mL MEF, was amplified using primers F-RT and R-RT (Table S1). Purified mARD1 each) in ice-cold PBS, and incubated with EZ-link Sulfo-NHS-SS-Biotin cDNA (QIAquick Gel Extraction Kit, Qiagen) was cloned into TOPO vector 0 24 ∕ 500 μ ∕ (Invitrogen) for EcoRI excision of an ca. 1.6-kb fragment that was purified ( . mg mL, 1 h) before adding quenching solution ( L plate). Samples 1∕40 – and ligated in frame to pGENE-V5/His-C vector (Invitrogen) already cut with ( ) of proteins were separated in 4 20% Tris-glycine gels and transferred EcoRI yielding pGENE-ARD1-V5/His encoding EGFP in pGENE-ARD1-V5/His to nitrocellulose for densitometric quantification. by PCR (primers, Table S1). The plasmid pGENE-EGFP–ARD1-V5/His was used 5 to generate ARD1 KO MEFs expressing ARD1–WT protein. Mutations were EGF and EGFR in MEFs. MEFs (2 × 10 ∕100-mm dish) were incubated overnight made in this plasmid using a QuickChange Site-Directed Mutagenesis Kit (ca. 15 h) in MEF2, then 18 h in serum-free MEF2 (minus MCGS) containing (Stratagene; primers, Table S1) to generate ARD1–E3 encoding C34A and 10 μM Mfp. Medium was replaced with serum-free MEF2 containing purified H53A, replacements that abolish E3 ligase activity. ARD1–GTP with ARF- murine EGF (BD Bioscience), 100 ng∕mL, for the indicated time before wash- BIOCHEMISTRY domain mutation K458I is persistently active and ARD1–GDP with a T418N ing cells with 10 mL of ice-cold PBS and collection by scraping in ice-cold replacement is inactive. Excision of ARD1 from pGENE-EGFP–ARD1-V5/His M-PER™ (100 μL, Pierce) containing protease (Roche) and phosphatase (Sig- produced the control empty vector. Complete coding regions (EGFP through ma) inhibitors. After 5 min on ice, samples were sonified (20 s), immediately His tag) of all plasmids were sequenced (sequencing primers Table S1) using frozen on dry ice, and stored at −80 °C. ABIPIRM377 DNA sequencer (Perkin Elmer). ARD1 KO MEFs were cotransfected with pSwitch regulatory plasmid and Inhibition of Intracellular Protein Degradation. MEFs (2 × 105) were grown a pGene-GFP-ARD1-V5/His plasmid (37). Each plasmid (8 μg) in 800 μLof overnight in 100-mm plates in MEF2 before incubation (18 h, 37 °C) with Opti-MEM (Invitrogen) was incubated (15 min, room temperature) with 10 μM Mfp without or with proteasome inhibitors (lactacystin and MG132, 30 μL of Plus reagent before addition of Lipofectamine (32 μL, Invitrogen); each 50 μM, Sigma) or 100 μM chloroquine (Sigma) to inhibit lysosomal 15 min later, the mixture was added to 70–80% confluent ARD1 KO MEFs proteases. MEFs harvested by scraping in ice-cold M-PER (Pierce) containing – (100-mm dishes) previously washed with PBS and incubated 1 h with 3 5mL protease and phosphatase inhibitors (100 μL per plate) were sonified (20 s) to of Opti-MEM. After 4 h with cells, medium was replaced with fresh MEF1. break cells. Cell lysates were analyzed by Western blotting and densitometric MEFs were transfected a second time 24 h later and, after 24 h, cells were quantification. split 1∶4 for growth in MEF2. Stably transfected MEF lines were screened for expression of expected proteins by incubation of cells for 4–12 h with Isolation of RNA and Quantitative PCR. Total RNA (5 μg), separated from lysates 10 μM Mfp before analysis of mRNA and protein by RT-PCR and Western using a total RNeasy isolation kit (Qiagen), was reverse transcribed using a blotting, respectively. After 3–4 wk, six MEF lines (from 24 initiated) for each Transcription III Superscript Kit (Invitrogen). For specific PCR amplification, mARD1 construct were selected for further study. TaqMan Universal PCR master mix and gene-specific TaqMan probes (Applied Biosystems) were used. Data were analyzed using the Sequence Detector ver- Centrifugal Fractionation of Cells. Each cell line was plated on 10 150-mm sion 1.6.3. In most experiments, α-tubulin was an internal control. To evaluate dishes and grown to 70–80% confluency. Mfp (final concentration 10 μM) differences among MEF lines, cycle thresholds (C ) for ARD1and EGFR from was added to each dish 24 h before cells were rinsed twice with 7 mL of t each were expressed relative to that for α-tubulin in the same sample. Data ice-cold PBS and harvested by scraping in 5 mL of ice-cold PBS. Unless other- for each line were then expressed relative to that for control MEFs (usually wise stated, all subsequent procedures were carried out at 4 °C. Cells were “KO-WT”) in the same experiment. pelleted, dispersed, and homogenized (Dounce tissue grinder) in 10 mL of ice-cold 20 mM Tris • HCl buffer containing 1 mM NaN3, 250 mM sucrose, 0.5 mM EDTA, 1 mM DTT, protease inhibitors (BD Biosciences, catalog no. Statistical Analysis. Densitometer readings (raw data) were normalized to 554779), and phosphatase inhibitors (Sigma, P5726). Cell lysate was sonicated the GAPDH readings (internal control). Each EGFR, IR, or transforming growth and fractionated by sequential centrifugation of each supernatant (after the factor receptor III reading was divided by the GAPDH value for that gel lane initial 15 min at 1;000 × g to yield 1S) 3;000 × g, 15 min (3S); 20;000 × g, (e.g., EGFR-1/GAPDH-1 or EGFR-2/GAPDH-2, etc.). Experiments were repeated 20 min (20S); 100;000 × g, 60 min (100S); and 100;000 × g,18h(100 × S). three times. Data are reported as means SD of values from the indicated Samples (0.1% of total lysate, supernatant, and pellet) were analyzed by number of experiments. (In later experiments, α-tubulin replaced GAPDH Western blotting with antibodies against V5 (ARD1), plus markers for nuclei as “sample control” after we recognized GAPDH variations and employed (histone H4), Golgi (GM130), lysosomes (LAMP1), and cytosol (GAPDH). especially conservative interpretation of all receptor values.) ANOVA was used to detect overall differences among group means. If an overall difference was Western Blotting and Quantification of Proteins. Proteins (40–60 μg), separated detected (p < 0.05), we proceeded to all pairwise group comparisons using the by SDS-PAGE in 4–20% gels, were transferred to nitrocellulose membranes Tukey-Kramer post hoc test with significance at p ≤ 0.05. SAS 9.2 (SAS Insti- that were incubated (1 h, room temperature) in Tris-buffered saline containing tute, Inc.) and GraphPad Prism version 4.00 for Windows (GraphPad Software) 5% nonfat dry milk (BioRad) and 0.1% Tween 20, then overnight at 4 °C were used for statistical analyses. with antibodies against EGFR, IR or TGF-βR III (rabbit, 1∶1;000 dilution, Cell Sig- 1∶1 000 nalling), nucleoporin and GM130 (mouse monoclonal, ; ; BD Bioscience), ACKNOWLEDGMENTS. We thank Drs. Christian Combs and Daniela Malide V5 (mouse monoclonal, 1∶1;000; Invitrogen), GAPDH, and α-tubulin (mouse from the light microscopy core facility National Institutes of Health, National monoclonal, 1∶5;000; Sigma). ARD1 antibodies (21st Century Biochemicals, Heart, Lung, and Blood Institute (NHLBI). The research was supported by the Inc.), affinity-purified Ab no. 5525 from rabbits immunized with full-length Intramural Research Program of the National Institutes of Health, NHLBI.

Meza-Carmen et al. PNAS Early Edition ∣ 5of6 Downloaded by guest on September 26, 2021 1. Rothman JE, Wieland FT (1996) Protein sorting by transport vesicles. Science 20. Vitale N, Horiba K, Ferrans VJ, Moss J, Vaughan M (1998) Localization of ADP- 272:227–234. ribosylation factor domain protein 1 (ARD1) in lysosomes and Golgi apparatus. Proc 2. Eden ER, White IJ, Futter CE (2009) Down-regulation of epidermal growth factor Natl Acad Sci USA 95:8613–8618. receptor signalling within multivesicular bodies. Biochem Soc Trans 37:173–177. 21. Nathan JA, et al. (2008) The ubiquitin E3 ligase MARCH7 is differentially regulated by 3. Sorkin A, Goh LK (2009) Endocytosis and intracellular trafficking of ErbBs. Exp Cell Res the deubiquitylating enzymes USP7 and USP9X. Traffic 9:1130–1145. 315:683–696. 22. Bill A, et al. (2010) Cytohesins are cytoplasmic ErbB receptor activators. Cell 4. Ravid T, Hochstrasser M (2008) Diversity of degradation signals in the ubiquitin- 143:201–211. proteasome system. Nat Rev Mol Cell Biol 9:679–690. 23. Hafner M, et al. (2006) Inhibition of cytohesins by SecinH3 leads to hepatic insulin Nature – 5. Spang A (2008) The life cycle of a transport vesicle. Cell Mol Life Sci 65:2781–2789. resistance. 444:941 944. 6. Bonifacino JS, Glick BS (2004) The mechanisms of vesicle budding and fusion. Cell 24. Venkateswarlu K, Gunn-Moore F, Tavare JM, Cullen PJ (1999) EGF-and NGF-stimulated 116:153–166. translocation of cytohesin-1 to the plasma membrane of PC12 cells requires PI 3-kinase J Cell Sci – 7. Pfeffer SR (2007) Unsolved mysteries in membrane traffic. Annu Rev Biochem activation and a functional cytohesin-1 PH domain. 112:1957 1965. 76:629–645. 25. Chung I, et al. (2010) Spatial control of EGF receptor activation by reversible dimer- Nature – 8. Tsuchiya M, Price SR, Tsai SC, Moss J, Vaughan M (1991) Molecular identification of ization on living cells. 464:783 787. ADP-ribosylation factor mRNAs and their expression in mammalian cells. J Biol Chem 26. Alvarado D, Klein DE, Lemmon MA (2010) Structural basis for negative cooperativity in growth factor binding to an EGF receptor. Cell 142:568–579. 266:2772–2777. 27. Leahy DJ (2010) The ins and outs of EGFR asymmetry. Cell 142:513–515. 9. Moss J, Vaughan M (1998) Molecules in the ARF orbit. J Biol Chem 273:21431–21434. 28. Zhang X, Gureasko J, Shen K, Cole PA, Kuriyan J (2006) An allosteric mechanism 10. Kirchhausen T (2000) Three ways to make a vesicle. Nat Rev Mol Cell Biol 1:187–198. for activation of the kinase domain of epidermal growth factor receptor. Cell 11. Mishima K, Tsuchiya M, Nightingale MS, Moss J, Vaughan M (1993) ARD 1, a 64-kDa 125:1137–1149. guanine nucleotide-binding protein with a carboxyl-terminal ADP-ribosylation factor 29. Murphy JE, Padilla BE, Hasdemir B, Cottrell GS, Bunnett NW (2009) Endosomes: domain. J Biol Chem 268:8801–8807. A legitimate platform for the signaling train. Proc Natl Acad Sci USA 106:17615–17622. 12. Vitale N, Moss J, Vaughan M (1996) ARD1, a 64-kDa bifunctional protein containing an 30. Schlessinger J (2000) Cell signaling by receptor tyrosine kinases. Cell 103:211–225. 18-kDa GTP-binding ADP-ribosylation factor domain and a 46-kDa GTPase-activating 31. Blobe GC, Liu X, Fang SJ, How T, Lodish HF (2001) A novel mechanism for regulating domain. Proc Natl Acad Sci USA 93:1941–1944. transforming growth factor beta (TGF-beta) signaling. Functional modulation of type 13. Vitale N, Ferrans VJ, Moss J, Vaughan M (2000) Identification of lysosomal and Golgi III TGF-beta receptor expression through interaction with the PDZ domain protein, Mol Cell Biol localization signals in GAP and ARF domains of ARF domain protein 1. GIPC. J Biol Chem 276:39608–39617. – 20:7342 7352. 32. Blobe GC, et al. (2001) Functional roles for the cytoplasmic domain of the type III 14. Reymond A, et al. (2001) The tripartite motif family identifies cell compartments. transforming growth factor beta receptor in regulating transforming growth factor EMBO J – 20:2140 2151. beta signaling. J Biol Chem 276:24627–24637. 15. Vichi A, Payne DM, Pacheco-Rodriguez G, Moss J, Vaughan M (2005) E3 ubiquitin ligase 33. Chen W, et al. (2003) Beta-arrestin 2 mediates endocytosis of type III TGF-beta receptor Proc activity of the trifunctional ARD1 (ADP-ribosylation factor domain protein 1). and down-regulation of its signaling. Science 301:1394–1397. Natl Acad Sci USA 102:1945–1950. 34. Pacheco-Rodriguez G, Meacci E, Vitale N, Moss J, Vaughan M (1998) Guanine nucleo- 16. Arimoto K, et al. (2010) Polyubiquitin conjugation to NEMO by triparite motif protein tide exchange on ADP-ribosylation factors catalyzed by cytohesin-1 and its Sec7 23 (TRIM23) is critical in antiviral defense. Proc Natl Acad Sci USA 107:15856–15861. domain. J Biol Chem 273:26543–26548. 17. Karbowski M, Neutzner A, Youle RJ (2007) The mitochondrial E3 ubiquitin ligase 35. Daniele T, Di Tullio G, Santoro M, Turacchio G, De Matteis MA (2008) ARAP1 regulates MARCH5 is required for Drp1 dependent mitochondrial division. J Cell Biol 178:71–84. EGF receptor trafficking and signalling. Traffic 9:2221–2235. 18. Lerner M, et al. (2007) The RBCC gene RFP2 (Leu5) encodes a novel transmembrane E3 36. Yoon HY, Lee JS, Randazzo PA (2008) ARAP1 regulates endocytosis of EGFR. Traffic ubiquitin ligase involved in ERAD. Mol Biol Cell 18:1670–1682. 9:2236–2252. 19. Vitale N, et al. (2000) Specific functional interaction of human cytohesin-1 and 37. Kato J, Zhu J, Liu C, Moss J (2007) Enhanced sensitivity to cholera toxin in ADP- ADP-ribosylation factor domain protein (ARD1). J Biol Chem 275:21331–21339. ribosylarginine hydrolase-deficient mice. Mol Cell Biol 27:5534–5543.

6of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1103867108 Meza-Carmen et al. Downloaded by guest on September 26, 2021