Src-dependent repression of ARF6 is required to maintain podosome-rich sealing zones in bone-digesting

Tobias Heckela,1, Cornelia Czupallaa,1, Ana Isabel Expirto Santoa, Mihaela Aniteia, Maria Arantzazu Sanchez-Fernandeza, Kerstin Moscha, Eberhard Krauseb, and Bernard Hoflacka,2

aBiotechnology Center, Dresden University of Technology, Tatzberg 47-51, 01307 Dresden, Germany; and bLeibniz Institute of Molecular Pharmacology, Robert-Ro¨ssle-Strasse 10, 13125 Berlin, Germany

Edited by Joseph Schlessinger, Yale University School of Medicine, New Haven, CT, and approved November 21, 2008 (received for review May 10, 2008) Bone digestion occurs when osteoclasts adhere onto bone surfaces regulates rearrangements in the by stimulat- and polarize to form acidic, hydrolase-rich resorption lacunae. For ing the Rac GTPase via a protein complex comprising the Src this process, they condense their actin-rich podosomes in tight and Syk kinases, the ␣v␤3 , and the Rac guanine nucle- belts to establish sealing zones, which segregate their basal mem- otide exchange factor Vav3 (10). Although this study highlights branes from those facing resorption lacunae. This polarization the importance of Rac signaling, it neglects the impact of other process remains poorly understood. Here, we combined quantita- small or factors regulated by Src to coordinate the tive proteomics and gene silencing to identify new substrates of extensive actin remodeling in osteoclasts. Therefore, we set out the Src tyrosine kinase, a key regulator of function. We to identify new Src substrates or binding partners by using now report that a depletion of the ARF GTPase-activating protein quantitative proteomics. We now report that Src also down- GIT2, which localizes to sealing zones upon Src phosphorylation, or regulates ARF6 activity via the GTPase activating protein GIT2 a lack of GTP hydrolysis on ARF6 impairs sealing zone formation to maintain sealing zones and osteoclast polarity during bone and polarized membrane traffic. Surprisingly, the Rho guanine degradation. ␣ ␤

nucleotide exchange factors and PIX, which usually coordinate BIOLOGY ARF and Rho signaling, were found to be dispensable. We conclude Results that the Src-dependent localization of GIT2 is essential for down- As a cell-model system allowing biochemical analyses of bone- regulating ARF6 activity at sealing zones, and thus for maintaining digesting cells, we used Raw 264.7 cells that differentiate in vitro osteoclast polarity. toward osteoclasts upon treatment with the osteoclastogenic cytokine receptor activator of NF-␬B ligand (RANKL) (11). polarity ͉ GIT2 ͉ podosomes ͉ quantitative proteomics Raw 264.7 cell-derived osteoclasts conform to conventional assays for osteoclast function as primary osteoclasts isolated steoclasts are specialized, multinucleated cells derived from from bones (2, 12, 13). Upon transfer onto bone or hydroxyap- Othe monocyte/ hematopoietic lineage that di- atite-coated bone-mimicking surfaces, these osteoclasts polarize gest the bone matrix when they adhere and polarize onto bone by establishing sealing zones, as seen in primary osteoclasts surfaces. For this process, they form specific adhesive, actin-rich grown on natural-bone matrices. Src expressed as GFP-tagged structures, referred to as podosomes, consisting of small columns protein using a recombinant adenovirus was highly enriched on of F-actin surrounded by proteins such as , talin, and the ruffled border and at the circumference of osteoclast sealing (1). On contact with bone surfaces, osteoclasts con- zones (Fig. 1A), whereas tyrosine-phosphorylated proteins lo- dense their podosomes into a compact, highly dynamic podoso- calized to the circumference and the core of the sealing zones mal belt or ring that determines a sealing zone (2, 3). The sealing (Fig. 1B). Consistent with the established Src function in oste- zone encircles a specialized, highly convoluted membrane do- oclasts, treatment with the Src inhibitors SU6656 (Fig. 1B) and main facing the bone surface, i.e., the ruffled border, thereby PP2 (data not shown) abolished the formation of actin-rich forming a sealed resorption lacuna between this membrane structures enriched in tyrosine-phosphorylated proteins. Ac- domain and the bone. The ruffled border is enriched in lysoso- cordingly, the recovery of tyrosine-phosphorylated proteins in mal membrane proteins and ion transporters, like the V-type lysates and anti-phosphotyrosine immunoprecipitates was dras- Hϩ-ATPase needed to acidify the resorption lacuna and dissolve tically reduced after Src inhibition by SU6656 (Fig. 1C) and PP2. the bone matrix (4). When bone digestion is complete, osteoclasts disassemble Identification of Src Substrates and Src-Dependent Binding Partners. their sealing zone and depolarize to move to another resorption To identify Src substrates and binding partners, we applied a site. Throughout their life span, osteoclasts achieve multiple mass-spectrometry-based proteomic approach using the stable cycles of and migration (4, 5). How these isotope labeling with amino acids in cell culture (SILAC) method different events of resorption and migration, which are linked to sealing zone assembly and disassembly, are coordinated is not Author contributions: T.H., C.C., and B.H. designed research; T.H., C.C., A.I.E.S., M.A., understood yet. A first clue came from the targeted disruption M.A.S.-F., and E.K. performed research; T.H., C.C., and K.M. contributed new reagents/ of the Src gene in mice leading to osteopetrosis due to nonfunc- analytic tools; T.H., C.C., A.I.E.S., and M.A. analyzed data; and T.H., C.C., and B.H. wrote the tional osteoclasts (6). This phenotype is Src specific because the paper. disruption of any other Src kinase family member failed to The authors declare no conflict of interest. reproduce such a phenotype (6). Src interacts with a plethora of This article is a PNAS Direct Submission. binding partners and substrates, which involves this tyrosine- 1T.H. and C.C. contributed equally to this work. kinase in several fundamental cellular processes including cy- 2To whom correspondence should be addressed. E-mail: bernard.hofl[email protected] toskeletal remodeling, mitogenic signaling, cell cycle control, dresden.de. Ϫ/Ϫ and survival (7). The fact that Src osteoclasts are unable to This article contains supporting information online at www.pnas.org/cgi/content/full/ organize their actin cytoskeleton points at a particular involve- 0804464106/DCSupplemental. ment of Src in this process (8, 9). Recent data suggest that Src © 2009 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0804464106 PNAS Early Edition ͉ 1of6 Downloaded by guest on October 2, 2021 Fig. 1. Identification of candidate Src targets in osteoclasts. (A) Raw 264.7 cell-derived osteoclasts were seeded on osteologic discs and infected with recombinant adenoviruses encoding for Src-GFP (green). Twenty-four hours after viral infection, cells were fixed, stained with phalloidin (red) and DAPI (blue), and analyzed by confocal . (B) Osteoclasts grown on osteologic discs were fixed, double-stained with anti-phosphotyrosine antibody (green) and phalloidin (red), and analyzed by confocal microscopy. Note that after treatment with the Src inhibitor SU6656 (10 ␮M), both the sealing zone itself and the tyrosine-phosphorylated proteins enriched at the sealing zone disappear. (Scale bars, 20 ␮m.) (C) Osteoclasts were treated with 10 ␮M SU6656 (ϩ) or left untreated (Ϫ), cell lysates were prepared, and tyrosine-phosphorylated proteins were immunoprecipitated with anti-phosphotyrosine antibody. Cell lysates (Left) and immunoprecipitates (Right) were probed by Western blotting for tyrosine-phosphorylated proteins. (D) Strategy. Osteoclasts were differentially labeled by growing them in medium containing either light (Arg-0) or heavy (Arg-6) arginine during their RANKL-induced differentiation before treatment of one sample with SU6656. After immunoprecipitation with anti-phosphotyrosine antibodies, precipitated proteins were digested with trypsin and analyzed by mass spectrometry. (E) Abundance ratios of proteins identified in the anti-phosphotyrosine immunoprecipitate. An isotope ratio of 0.8 or below was considered to be decreased protein abundance. (F) Cell lysates (Left) and immunoprecipitates (Right) were prepared as in C and probed by Western blotting for the indicated proteins.

(14), which allows the quantification of changes in protein F-BAR protein PSTPIP1 (proline-serine-threonine phos- abundance after Src inhibition. Raw 264.7 cells were grown in phatase-interacting protein 1), the Rho-GEF ␣ PIX, and the 12 medium containing either normal C6 arginine (Arg-0) or the ARF-GAPs GIT1 and GIT2, which usually form complexes with 13 isotopic variant C6 arginine (Arg-6) and differentiated toward the Rho-GEF ␣ or ␤ PIX (18). osteoclasts. Arg-6-labeled cells were treated with SU6656, and To verify the MS data, we first performed Western blot Arg-0-labeled cells served as the control. Combined cell lysates experiments using specific antibodies after immunoprecipitation were incubated with anti-phosphotyrosine antibodies. The pu- with anti-phosphotyrosine antibodies (Fig. 1F). In this analysis, rified phosphotyrosine-containing protein complexes were re- we also considered paxillin, an Src substrate (19) and core solved by SDS/PAGE and analyzed by nano LC-MS/MS (Fig. component of podosomes (1) that we could identify but not 1D). MS/MS spectra of tryptic peptides were used for protein quantify because of missing arginine-containing peptides. identification, whereas the relative intensities of the correspond- Whereas we did not detect differences in the protein amounts in ing Arg-0- and Arg-6-containing peptides in the MS spectra were cell lysates, the abundance of all tested proteins was reduced in used to calculate abundance ratios (Fig. 1E). An isotope ratio of phosphotyrosine-containing complexes after Src inhibition. Be- 0.8 or below was considered as decreased protein abundance. cause chemical inhibitors of kinases are not monospecific, we Among the 117 phosphotyrosine-containing proteins identified, also used siRNA-mediated depletion of Src. Recombinant ad- of which 45 are known core elements of podosomes (1), we enoviruses expressing short hairpin RNAs or Stealth siRNA detected 25 proteins with altered abundance in phosphoty- oligonucleotide duplexes targeting Src were applied to oste- rosine-containing complexes after Src inhibition, of which 13 oclasts, which led to a Ϸ90% reduction of the Src protein (Fig. were quantified with at least 2 peptides (Table 1 and Table S1). S1). Accordingly, sealing zones and phosphotyrosine-containing Among those proteins, we found Src itself, most likely because proteins could not be detected after Src knockdown (Fig. S1A). of decreased autophosphorylation, and several known Src inter- The amount of Src, FAK1, paxillin, Dok3, and GIT2 that was action partners such as the Src substrate p130Cas (Crk- immunoprecipitated with anti-phosphotyrosine antibodies was associated substrate) (15), which interacts with Pyk2 (focal reduced (63%, 97%, 70%, 54%, and 35% reductions, respec- adhesion kinase 2) (15); the Cas-binding protein Bcar3 (breast tively), thereby validating the quantitative proteomic analysis of anti-estrogen resistance protein 3) (16); and the Rho- drug-treated osteoclasts (Fig. S1C). Pyk2 and Vav phosphory- GEF Vav, which associates with the phosphorylated Hcls1 lation was not significantly affected, possibly because of a lack of required for proper Vav localization (17). Most notably, we stimulated integrin signaling in osteoclasts grown under condi- identified potential new Src substrates and/or interaction part- tions allowing proteomic analyses. Together, these results con- ners, namely the adaptor protein Dok3 (docking protein 3), the firm recent findings identifying Pyk2 (20), paxillin (19), and the

2of6 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0804464106 Heckel et al. Downloaded by guest on October 2, 2021 Table 1. Effect of Src inhibition on protein phosphorylation SwissProt Peptides Quantified peptides Normalized Protein accession no. (MS/MS) (peptide ratio) ratio

Tyrosine kinases FAK1 P34152 22 8 0.78 Pyk2 (FAK2) Q9QVP9 7 3 0.68 Src P05480 3 2 0.80 Known tyrosine kinase effectors p130Cas (Bcar1) Q61140 6 2 0.68 Bcar3 Q9QZK2 2 2 0.58 Hcls1 (HS1) P49710 4 2 0.71 Dok3 Q9QZK7 3 3 0.38 Cytoskeletal proteins Desmoplakin-3 Q02257 5 2 0.17 Abp1 Q62418 2 2 0.74 PSTPIP1 P97814 5 3 0.47 GTPase-interacting proteins GIT1 Q68FF6 7 3 0.51 GIT2 Q9JLQ2 2 2 0.64 ␣ PIX Q8K4I3 3 2 0.63

Shown are proteins with altered abundance in phosphotyrosine-containing complexes after treatment of osteoclasts with the Src inhibitor SU6656.

Rho-GEF Vav (21) as Src substrates and point at a role of Src mediated depletion of GIT1 remained without any effect on in the phosphorylation of the ARF-GAPs GIT1/GIT2 and the sealing-zone integrity. Similarly, the knockdown of the Rho GEF Rho-GEF ␣ PIX. ␣ PIX or ␤ PIX, the other PIX member that was not detected as CELL BIOLOGY a Src target, remained without any effect, although podosomal ARF-GAP GIT2 Is a Key Component Regulating Sealing Zone Formation. belts appeared less compact, consistent with the previous report Our subsequent functional analysis essentially focused on GIT- (22) (Fig. S2 A and B). Also, the ␣ and ␤ PIX double knockdown family ARF-GAPs. Because known Src substrates such as pax- did not modify sealing-zone formation (Fig. S2 A and B). Thus, illin and FAK1 are enriched along the sealing zone, we first GIT2 is essential for sealing zone formation in polarized oste- tested whether GIT proteins could be found at the same location oclasts, whereas GIT1 and ␣ and ␤ PIX are dispensable for this (Fig. 2). Indeed, GIT1 and GIT2 localized to the condensed function. podosomal belts of the sealing zone and to some extent to internal punctate structures (Fig. 2). This localization of GIT GIT2 Down-Regulates ARF6 at Sealing Zones of Osteoclasts. GIT2 proteins to podosomal belts, like that of paxillin and FAK1, was stimulates hydrolysis of GTP bound to the different ARF- lost after Src inhibition (Fig. 2). We then reasoned that a loss of GTPase subfamily members, including ARF6 (18). Because GIT or Src function should result in similar phenotypes. There- ARF6 is the only known ARF member operating at the plasma fore, we used siRNAs to reduce the expression of either protein, membrane (23, 24), we speculated that GIT2 might inactivate which yielded an 80–90% depletion at the mRNA and protein ARF6 with respect to sealing-zone formation. To test this level (Fig. 3 A and B). As shown in Fig. 3 C and D, the depletion possibility, we characterized the effects of ectopically expressed of GIT2 or Src resulted in a loss of sealing zones in Ϸ75% of the ARF6. ARF6 was largely excluded from podosomal belts of osteoclasts examined, which exhibited numerous membrane osteoclasts expressing low levels of this GTPase (Fig. 4A), being protrusions as typically observed in spreading or moving cells, found at the circumference of the sealing zone, the ruffled thus resembling SrcϪ/Ϫ osteoclasts (8, 9). The efficient siRNA- border, and basal membranes of osteoclasts. At higher expres-

Fig. 2. Src-dependent localization of GIT proteins to the sealing zone of osteoclasts. Osteoclasts were seeded on osteologic discs; treated with Src inhibitor or left untreated; fixed; and stained with phalloidin (red), DAPI (blue), and with antibodies for the indicated proteins (green). Cells were analyzed by confocal microscopy. (Scale bars, 20 ␮m.)

Heckel et al. PNAS Early Edition ͉ 3of6 Downloaded by guest on October 2, 2021 Fig. 3. RNAi-mediated depletion of GIT2 and Src impairs sealing zone formation in osteoclasts. (A) Osteoclasts were either mock-electroporated or electroporated with stealth RNAi duplexes targeting the indicated genes or with scrambled RNAi duplexes. Electroporated osteoclasts were grown foran additional 43 h, and mRNA was extracted, reverse transcribed, and analyzed by quantitative PCR using the comparative ct method. Shown are relative mRNA levels of Src and GIT2 normalized to GAPDH mRNA levels. (B). Electroporated osteoclasts were grown for an additional 43 h, and total cell lysates were prepared and probed by Western blotting for Src, GIT2, and Tubulin. (C and D) Osteoclasts were electroporated as in A. Electroporated osteoclasts were grown for an additional 43 h on osteologic discs, fixed, stained with phalloidin (red) and DAPI (blue), and analyzed by confocal microscopy. Sealing zones were counted (C). Values are mean Ϯ SD from 3 different experiments (n ϭ 300 osteoclasts/experiment). Representative images are shown (D). *, P Ͻ 0.001. (Scale bars, 20 ␮m.)

sion levels, which were the case in most of the osteoclasts, ARF6 In contrast, LAMP1 was detected in perinuclear structures, was found at the plasma membrane of spreading cells devoid of presumably lysosomes, in ARF6- or ARF6Q67L-expressing oste- actin rings. Because ARF6 over-expression could override its oclasts. The expression of an ARF6 mutant defective in GTP tightly controlled regulation, we reasoned that the expression of loading (ARF6T27N) or the efficient knockdown of ARF6 (Fig. ARF6Q67L, a mutant blocked in GTP hydrolysis, should show S2 C–E) remained without any effect. Thus, these results suggest that active ARF6 destabilizes sealing-zone formation and results that the ARF-GAP GIT2 mediates the down-regulation of in a lack of osteoclast polarization. Indeed, Fig. 4 A and B show ARF6, which is essential to stabilize sealing zones and osteoclast that, like adenovirus-mediated over-expression of wild-type polarity. ARF6, mutant ARF6Q67L impairs the assembly of sealing zones in osteoclasts. Consequently, osteoclasts were unable to segre- Discussion gate their membrane components, as exemplified by the lysoso- Our study demonstrates the importance of GIT2 and ARF6 in mal glycoprotein LAMP1 (Fig. 4C). LAMP1 normally concen- the dynamics of condensed podosomal belts forming sealing trates within the ruffled border of digesting osteoclasts and in zones and the maintenance of membrane components at the membrane-bound structures located above the ruffled border. ruffled border.

Fig. 4. Active ARF6 impairs sealing zone formation and osteoclasts polarization. (A and C) Osteoclasts were seeded on osteologic discs and infected with recombinant adenoviruses encoding for HA-tagged ARF6, ARF6Q67L, and ARF6T27N or mock-infected (control). Forty-two hours after viral infection, cells were fixed; stained with anti-HA or with anti-LAMP1 antibody (green), phalloidin (red), and DAPI (blue); and analyzed by confocal microscopy. (Scale bars,20␮m.) (B) Sealing zones were counted; values are mean Ϯ SD from 3 different experiments (n ϭ 250 osteoclasts/experiment). *, P Ͻ 0.001.

4of6 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0804464106 Heckel et al. Downloaded by guest on October 2, 2021 GITs represent hubs that coordinate ARF and Rho signaling, signaling. We took complementary approaches to identify genes thereby regulating various cellular functions, including mem- whose expression is drastically modified during osteoclastogen- brane trafficking, scaffolding of signaling cascades, and actin esis (35). Besides Vav proteins, this analysis highlighted several remodeling at adhesion sites (18). The recruitment of GIT other regulators of Rho GTPases. Therefore, further studies are proteins to adhesion sites relies on their interaction with paxillin needed to find the molecular links coordinating ARF6/GIT2 and (18, 25). In nonpolarized cells, GIT2 binds to paxillin upon Rho signaling during these regulated cycles of osteoclast adhe- phosphorylation by FAK and/or Src (26). Thus, it is more than sion and migration required for bone digestion. likely that the proper localization of GIT2 to sealing zones of polarized osteoclasts is determined by Src-dependent phosphor- Materials and Methods ylation. Indeed, Src inhibition in osteoclasts abolishes the proper Cell Culture, SILAC, and Osteoclastogenesis. Production of RANKL, cultivation localization of GIT2 and, interestingly, that of GIT1, too. The of Raw 264.7 cells (American Type Culture Collection), and differentiation facts, however, that GIT2 binds more tightly with paxillin than toward osteoclasts were performed as previously described (35). At day 4 GIT1 (25) and that the knockdown of GIT2, but not of GIT1, of differentiation, cells were transferred to BD BioCoat osteologic discs (BD Bioscience) for microscopy. For SILAC experiments, cells were cultured in dramatically affects sealing zone formation support the idea that DMEM lacking arginine (PAN Biotech) supplemented with 10% (vol/vol) GIT2 is the principal mediator of adhesion-site stability (25, 27). dialyzed FCS. Arg-6 and Arg-0 SILAC media were prepared by adding ␣ 13 GITs usually form complexes with the Rac1/Cdc42-GEFs and L-arginine-U- C6 (Cambridge Isotope Laboratories) or the corresponding ␤ PIX (18). Although ␣ PIX was found as an Src target in nonlabeled amino acid, respectively, and cells were cultured for 5 days in osteoclasts, ␣ PIX is dispensable for GIT2 function in sealing SILAC media. zone dynamics. Furthermore, ␣ and ␤ PIX were found redundant in function, a finding that strongly suggests that the coordination Gene Transduction and RNAi in Osteoclasts. Adenoviral vectors and recombi- between ARF6 and Rac1/Cdc42 signaling most likely involves nant adenoviruses carrying Src-GFP, ARF6-HA, ARF6-HAQ67L, ARF6-HAT27N, and other members of the Rho-GEF family, potentially the Vav short hairpin RNAs targeting Src or GFP were generated according to the proteins (10). manufacturer’s protocol (AdEasy Adenoviral System, Stratagene). Raw 264.7- cell-derived osteoclasts were incubated with medium containing recombinant ARF6 is emerging as a common regulator of various actin- adenoviruses for 24 to 35 h. Experiments were performed between 43 h rich, adhesive structures, i.e., focal adhesions in nonpolarized postinfection for ectopic gene expression and 60 h postinfection for RNAi cells, adherens junctions in polarized epithelial cells, invadopo- experiments. Stealth RNAi duplexes were obtained from Invitrogen. Prede- dia in cancer cells (23, 25), and podosomes in osteoclasts, which signed stealth RNAi or scrambled stealth RNAi duplexes (each 800 nM) were are believed to represent distinct actin-rich supramolecular electroporated into osteoclasts by using a CytoPulse PA-4000 electroporator CELL BIOLOGY structures (1, 28, 29). In this respect, ARF6 function seems to be (CytoPulse Sciences). Electroporated cells were resuspended in medium sup- versatile but always following a mechanistic principle, namely plemented with RANKL and allowed to recover for 43 h. Then cells were supporting membrane mass supply for membrane extension. In processed for subsequent analysis. Detailed information regarding stealth tumor cells, this basic feature of active ARF6 is thought to RNAi duplexes and short hairpin RNAs are given in SI Text. promote their invasive capacities, probably by plasma membrane Quantitative PCR. Total RNA was isolated by using the RNeasy Kit (Qiagen). extension to facilitate formation (30, 31). In polar- DNase-I-treated total RNA was reverse transcribed by using the SuperScript II ized epithelial cells, active ARF6 facilitates the turnover of First Strand Synthesis system (Invitrogen). Quantitative PCR was performed adherens junctions by promoting endocytosis of E-cadherin or with a Stratagene Mx4000 QPCR system and the Brilliant SYBR Green QPCR kit ␤1-integrins, which leads to depolarization and cell spreading, as according to the manufacturer’s instructions (Stratagene). QPCR analyses seen during epithelial-to-mesenchymal transition (32–34). In were done in triplicate, and cycle threshold (ct) values were normalized polarized osteoclasts, active ARF6 also destabilizes condensed against the internal control gene GAPDH. Fold differences in expression levels podosome belts, which form the sealing zone; influences the were calculated according to the comparative ct method (36). Primer se- polarized sorting of proteins, as exemplified here by LAMP1; quences are given in SI Text. and appears to favor cell spreading. Thus, ARF6 repression leading to a down-regulation of endocytosis at specific adhesive Immunofluorescence, Confocal Laser Scanning Microscopy, and Image Analysis. Sample preparation and image acquisition were done as described previously sites would be required to stabilize adhesive structures, whereas (37). Images were processed with Adobe Photoshop v7.0 (Adobe Systems). For ARF6 activation would lead to membrane extension, which quantification, more than 300 osteoclasts per condition and per experiment results either in the destabilization of certain adhesive structures were counted and analyzed for their ability to form sealing zones. Means and as seen in polarized cells or in the generation of protrusions for standard deviations (SD) were calculated, and data were analyzed for statis- invadopodia formation. The regulation of ARF6 could, however, tical significance by using one-way ANOVA. differ according to the actin-rich structure or the cell type considered. Src-dependent phosphorylation destabilizes adher- Immunoprecipitation and Western Blotting. For drug-induced inhibition of Src, ens junctions of polarized epithelial cells (34), whereas it pro- osteoclasts were treated with 10 ␮M SU6656 or its vehicle for 90 min. Drug- or motes, together with GIT2, sealing-zone formation in polarized RNAi-treated cells were processed according to Kratchmarova et al. (38). Then osteoclasts. These observations suggest that Src phosphorylates cell extracts were either subjected to SDS/PAGE or immunoprecipitated by incubation with anti-phosphotyrosine antibodies (25 ␮g of 4G10/mg of lysate different ARF6-interacting proteins or effectors in these differ- and 10 ␮l of P-Tyr-100/mg of lysate) for6hat4°C.Proteins were eluted, ent cell types. The experimental strategy used in this study could resolved by SDS/PAGE, blotted, and subsequently incubated with primary potentially be applied to gain further insight into the regulation antibodies. After incubation with secondary antibodies conjugated with of different adhesive structures and the molecules coordinating horseradish peroxidase (Jackson Immuno Research), bands were detected ARF6 and Rho signaling. with enhanced chemiluminescence Western blotting detection reagents (Am- Bone digestion occurs through several, repeated cycles of ersham). Densitometric quantification of Western blots was performed by sealing-zone formation allowing osteoclasts to adhere and po- using ImageJ software (National Institutes of Health). larize onto bone surfaces, and sealing zone destabilization allowing osteoclasts to migrate to other areas to digest (4). Our Mass Spectrometry and Data Analysis. Tryptic digest of proteins and nanoLC- study strongly suggests that these cycles could involve a tight MS/MS experiments were done as described previously (37). Tryptic pep- tides were separated by a reversed-phase Micromass capillary liquid chro- equilibrium between cycles of GIT2-dependent ARF6 down- matography system connected to a Z-spray nanoelectrospray ion source regulation and ARF6 activation, a process that needs to be and a quadruple orthogonal acceleration time-of-flight mass spectrometer coordinated with Rho down-regulation and activation. To elu- Q-TOF Ultima (Micromass). MS/MS data analysis was performed by using cidate this regulatory mechanism, it will be important to clearly MASSLYNX software Version 4.0 (Micromass), and resulting data files were identify the effector molecules coordinating Rho and ARF6 searched using MASCOT Search Engine Version 2.0 (Matrix Science) against

Heckel et al. PNAS Early Edition ͉ 5of6 Downloaded by guest on October 2, 2021 the SwissProt database version 240206 (211104 sequences). Quantification oligonucleotide sequences used in this study. Table S4 lists the primers used was carried out by using the open-source software MSQUANT (Peter in this study. Mortensen and Matthias Mann, http://msquant.sourceforge.net/). Further details are provided in SI Text. ACKNOWLEDGMENTS. We thank D. Thiel and J. Lehmann for technical assis- tance, M. Schu¨mann for help with the MS measurements, R.T. Premont (Duke Online SI. Further information, including the RNAi-mediated depletion of University, Durham, NC) for his generous gift of anti-GIT antibodies, M. Zerial Src impairing protein tyrosine phosphorylation in osteoclasts (Fig. S1), a (Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany) for the Src-GFP construct, and K. Simons and M. Zerial for their critical reading of demonstration that RNAi-mediated depletion of GIT1, ␣ PIX, ␤ PIX, and the manuscript. This work was supported in part by Dresden University of Tech- ARF6 does not impair sealing zone formation in osteoclasts (Fig. S2), and a nology Grant HWP-1207; Sachsisches Ministerium fur Wissenschaft und KunstEu- list of all quantified proteins of the SILAC approach (Table S1), is available ropaische Fond fur Regionale Entwicklung Grant 1203; Bundesministerium fu¨r in the supporting information. Table S2 lists the oligonucleotides used to Bildung und Forschung Grant 0313815B; and Deutsche Forschungsgemeinschaft silence murine Src, GIT2, GIT1, and ␣ PIX. Table S3 lists the short hairpin Grants TRR13/2-08, HO 254/2-1, HO 2584/1-1, HO 2584/6-1, and HO 2584/8-1.

1. Linder S, Aepfelbacher M (2003) Podosomes: Adhesion hot-spots of invasive cells. 21. Turner M, Billadeau DD (2002) VAV proteins as signal integrators for multi-subunit Trends Cell Biol 13:376–385. immune-recognition receptors. Nat Rev Immunol 2:476–486. 2. Luxenburg C, et al. (2007) The architecture of the adhesive apparatus of cultured 22. Gringel A, et al. (2006) PAK4 and ␣PIX determine podosome size and number in osteoclasts: From podosome formation to sealing zone assembly. PLoS ONE 2:e179. through localized actin regulation. J Cell Phys 209:568–579. 3. Jurdic P, Saltel F, Chabadel A, Destaing O (2006) Podosome and sealing zone: Specificity 23. D’Souza-Schorey C, Chavrier P (2006) ARF proteins: Roles in membrane traffic and of the osteoclast model. Eur J Cell Biol 85:195–202. beyond. Nature Reviews 7:347–358. 4. Vaananen HK, Zhao H, Mulari M, Halleen JM (2000) The cell biology of osteoclast 24. Heo WD, et al. (2006) PI(3,4,5)P3 and PI(4,5)P2 lipids target proteins with polybasic function. J Cell Sci 113(Pt 3):377–381. clusters to the plasma membrane. Science 314:1458–1461. 5. Saltel F, Destaing O, Bard F, Eichert D, Jurdic P (2004) Apatite-mediated actin dynamics 25. Sabe H, Onodera Y, Mazaki Y, Hashimoto S (2006) ArfGAP family proteins in cell in resorbing osteoclasts. Mol Biol Cell 15:5231–5241. adhesion, migration and tumor invasion. Curr Opin Cell Biol 18:558–564. 6. Lowell CA, Soriano P (1996) Knockouts of Src-family kinases: Stiff bones, wimpy T cells, 26. Brown MC, Cary LA, Jamieson JS, Cooper JA, Turner CE (2005) Src and FAK kinases and bad memories. Genes Dev 10:1845–1857. cooperate to phosphorylate paxillin kinase linker, stimulate its local- 7. Martin GS (2001) The hunting of the Src. Nature Reviews 2:467–475. ization, and regulate cell spreading and protrusiveness. Mol Biol Cell 16:4316–4328. 8. Schwartzberg PL, et al. (1997) Rescue of osteoclast function by transgenic expression of kinase-deficient Src in srcϪ/Ϫ mutant mice. Genes Dev 11:2835–2844. 27. Frank SR, Adelstein MR, Hansen SH (2006) GIT2 represses Crk- and Rac1-regulated cell 9. Miyazaki T, et al. (2004) Src kinase activity is essential for osteoclast function. J Biol spreading and Cdc42-mediated focal adhesion turnover. EMBO J 25:1848–1859. Chem 279:17660–17666. 28. Linder S (2007) The matrix corroded: Podosomes and invadopodia in extracellular 10. Zou W, et al. (2007) Syk, c-Src, the alphavbeta3 integrin, and ITAM immunoreceptors, matrix degradation. Trends Cell Biol 17(3):107–117. in concert, regulate osteoclastic bone resorption. J Cell Biol 176:877–888. 29. Buccione R, Orth JD, McNiven MA (2004) Foot and mouth: Podosomes, invadopodia 11. Boyle WJ, Simonet WS, Lacey DL (2003) Osteoclast differentiation and activation. and circular dorsal ruffles. Nat Rev Mol Cell Biol 5:647–657. Nature 423:337–342. 30. Hashimoto S, et al. (2004) Requirement for Arf6 in breast cancer invasive activities. Proc 12. Destaing O, Saltel F, Geminard JC, Jurdic P, Bard F (2003) Podosomes display actin Natl Acad Sci USA 101:6647–6652. turnover and dynamic self-organization in osteoclasts expressing actin-green fluores- 31. Tague SE, Muralidharan V, D’Souza-Schorey C (2004) ADP-ribosylation factor 6 regu- cent protein. Mol Biol Cell 14:407–416. lates tumor cell invasion through the activation of the MEK/ERK signaling pathway. 13. Toyomura T, et al. (2003) From lysosomes to the plasma membrane: Localization of Proc Natl Acad Sci USA 101:9671–9676. vacuolar-type Hϩ -ATPase with the a3 isoform during osteoclast differentiation. J Biol 32. Dunphy JL, et al. (2006) The Arf6 GEF GEP100/BRAG2 regulates cell adhesion by Chem 278:22023–22030. controlling endocytosis of beta1 integrins. Curr Biol 16:315–320. 14. Ong SE, et al. (2002) Stable isotope labeling by amino acids in cell culture, SILAC, as a 33. Palacios F, Price L, Schweitzer J, Collard JG, D’Souza-Schorey C (2001) An essential role simple and accurate approach to expression proteomics. Mol Cell Proteomics 1:376– for ARF6-regulated membrane traffic in adherens junction turnover and epithelial cell 386. migration. EMBO J 20:4973–4986. 15. Lakkakorpi PT, et al. (1999) Stable association of PYK2 and p130(Cas) in osteoclasts and 34. Palacios F, Tushir JS, Fujita Y, D’Souza-Schorey C (2005) Lysosomal targeting of E- their co-localization in the sealing zone. J Biol Chem 274:4900–4907. cadherin: A unique mechanism for the down-regulation of cell–cell adhesion during 16. Gotoh T, Cai D, Tian X, Feig LA, Lerner A (2000) p130Cas regulates the activity of epithelial to mesenchymal transitions. Mol Cell Biol 25:389–402. AND-34, a novel Ral, Rap1, and R-Ras guanine nucleotide exchange factor. J Biol Chem 35. Czupalla C, Mansukoski H, Pursche T, Krause E, Hoflack B (2005) Comparative study of 275:30118–30123. protein and mRNA expression during osteoclastogenesis. Proteomics 5:3868–3875. 17. Gomez TS, et al. (2006) HS1 functions as an essential actin-regulatory adaptor protein at the immune synapse. Immunity 24:741–752. 36. Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data using real-time 18. Hoefen RJ, Berk BC (2006) The multifunctional GIT family of proteins. J Cell Sci 119(Pt quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25:402–408. 8):1469–1475. 37. Czupalla C, et al. (2006) Proteomic analysis of lysosomal acid hydrolases secreted by 19. Schaller MD (2001) Paxillin: A focal adhesion-associated adaptor protein. Oncogene osteoclasts: implications for lytic enzyme transport and bone metabolism. Mol Cell 20:6459–6472. Proteomics 5:134–143. 20. Dikic I, Tokiwa G, Lev S, Courtneidge SA, Schlessinger J (1996) A role for Pyk2 and Src 38. Kratchmarova I, Blagoev B, Haack-Sorensen M, Kassem M, Mann M (2005) Mechanism in linking G-protein-coupled receptors with MAP kinase activation. Nature 383:547– of divergent growth factor effects in differentiation. Science 550. 308:1472–1477.

6of6 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0804464106 Heckel et al. Downloaded by guest on October 2, 2021