Biochemical and Biophysical Research Communications 464 (2015) 1275e1281

Contents lists available at ScienceDirect

Biochemical and Biophysical Research Communications

journal homepage: www.elsevier.com/locate/ybbrc

Overexpression of Arl6ip5 in osteoblast regulates RANKL subcellualr localization

* ** Yu Wu a, b, , Miaomiao Wang c, Ying Peng a, Yuedi Ding a, Lili Deng a, Qiang Fu a, a Key Laboratory of Nuclear Medicine, Ministry of Health, Jiangsu Key Laboratory of Molecular Nuclear Medicine, Jiangsu Institute of Nuclear Medicine, Wuxi, Jiangsu, China b Wuxi Medical School, Jiangnan University, Wuxi, Jiangsu, China c Department of Occupational Health, Wuxi Center for Disease Control, Wuxi, Jiangsu, China article info abstract

Article history: The osteoblastic expression of RANKL, which is essential for the communication between osteoblastic Received 22 July 2015 cells and osteoclastogenic cells, is stimulated by locally acting or circulating osteotropic cytokines and Accepted 24 July 2015 hormones such as PTH and 1,25-(OH)2-D3 during the bone remodeling process. However, mechanisms Available online 26 July 2015 those control subcellular trafficking events, membrane expression and extracellular secretion of the newly synthesized RANKL are still not well understood. In our previous study, we have found that the Keywords: deficiency of osteoblastic Arl6ip5 (ADP-ribosylation-like factor 6 interacting 5), an endoplasmic RANKL reticulum (ER)-localized protein belonging to the prenylated rab-acceptor-family, enhanced osteoclas- Subcellular trafficking Osteoclastogenesis togenesis by increasing RANKL transcription in an ER stress dependent signaling. Here we found that over-expression of hemagglutinin (HA)-tagged Arl6ip5 in UAMS32 stromal/osteoblastic cells inhibited osteoclastogenesis, decreased the amount of soluble RANKL in culture supernatant and increased RANKL retention in ER. Moreover, Arl6ip5 bound with RANKL and disturbed the RANKL-OPG complex in UAMS- 32 cells. Finally, 1 to 36 amino acid deletion on the NH2 lumen terminus of Arl6ip5 impaired the interaction between Arl6ip5 and RANKL, restored the level of soluble RANKL and the osteoclastogenic ability. These findings indicated that Arl6ip5 was an anti-catabolic factor by binding with RANKL and disturbing its subcellular trafficking in osteoblast. © 2015 Elsevier Inc. All rights reserved.

1. Introduction osteoclastogenesis [3e7]. RANKL acts on its receptor RANK on the surface of haematopoietic precursor cells and leads to the devel- Bone remodeling process is tightly regulated by the signaling opment of mature osteoclasts. coupling between osteoblast lineage cells and osteoclast in “Basic In stromal/osteoblast cells, RANKL expression can be induced at Multicellular Unit (BMU)” [1,2]. This coordinated remodeling pro- the transcriptional level by factors known to stimulate bone cess is largely mediated by a wide range of paracrine and endocrine resorption such as PTH and 1,25 dihydroxyvitamin D3 (1,25-(OH)2- factors that secreted by bone cells. Among them, receptor activator D3) [4,8e10]. PTH regulates RANKL transcription via protein kinase of NF-kB ligand (RANKL), which is mainly expressed and secreted A (PKA)-cAMP-cAMP response element binding protein (CREB) by osteoblast lineages, is the essential mediator for pathway [4,8]. RANKL is also regulated at its subcellular trafficking level [11e13]. Osteoprotegerin (OPG), which is a very effective in- hibitor of osteoclast formation via its block on RANKL-RANK signaling, mediates lysosome trafficking of osteoblastic RANKL. In Abbreviations: Arl6ip5, ADP-ribosylation-like factor 6 interacting protein 5; ER, osteoblast, OPG interacts with RANKL and helps RANKL form stable endoplasmic reticulum; HA, hemagglutin; OPG, Osteoprotegerin; POBs, Primary fi osteoblasts; RANKL, receptor activator of nuclear factor-kB ligand; sRANKL, soluble complexes with its lysosome traf cking partner Vps33a [11,12]. The RANKL; TRAP, tartrate resistant acid phosphatase. lysosome localized RANKL is further transported to the cell surface * Corresponding author. Jiangsu Institute of Nuclear Medicine, QianRong Road 20, under the actions of Rab27a/b and their effectors slp4a, slp5 and Wuxi, Jiangsu 214063, China. Munc13-4 [13]. However, the detailed mechanisms about RANKL ** Corresponding author. Jiangsu Institute of Nuclear Medicine, QianRong Road 20, regulation under different physiological states are still unclear. Wuxi, Jiangsu 214063, China. E-mail addresses: [email protected] (Y. Wu), [email protected] (Q. Fu). ADP-ribosylation-like factor 6 interacting protein 5 (Arl6ip5) is http://dx.doi.org/10.1016/j.bbrc.2015.07.119 0006-291X/© 2015 Elsevier Inc. All rights reserved. 1276 Y. Wu et al. / Biochemical and Biophysical Research Communications 464 (2015) 1275e1281 an endoplasmic reticulum (ER) localized and a ubiquitously TCTCAGGGATGC; FL-HA-Arl6ip5-antisense: CGGTCTAGAAATCCGCATCA- expressed protein that belongs to prenylated rab-acceptor (PRA) GCCCGTGCAGCATTCCTAT CTCCCTCGCTTTGCTGATGTAG. All plasmids were protein family [14e16]. Studies from its rat homolog GTRAP3-18 verified by sequencing. indicated that it can form complex with glutamate transporter excitatory amino acid carrier 1 (EAAC-1), delay the ER exit of EAAC- 2.3. Quantitative polymerase chain reaction (q-PCR) analysis 1 and inhibit EAAC-1 transportation into the transport complexes in a Rab-1 dependent manner [14,16,17]. In osteoblast, Arl6ip5 q-PCR was done as indicated before [18] and the primers for mainly localizes in ER and its expression is regulated by osteotropic RANKL, Csf-1, Sema3a, Sema3b, Sema7a, Wnt5a and Actb were factors [18]. Moreover, osteoblastic Arl6ip5 deficiency leads to ER retrieved from Primer bank [21]. All detections were in triplicate for stress dependent RANKL transcription and thereby induce osteo- each sample and data were normalized to Actb level (DDcT). clastogenesis [18]. It would be interesting to find whether over-expression of Arl6ip5 in osteoblast has anti-catabolic effect by 2.4. Western blotting and immunoprecipitation analysis regulating the homeostasis of RANKL. Here, with Arl6ip5 over- expressed osteoblastic cells, we found that Arl6ip5 over- Protein was extracted from whole cell pellets with RIPA lysis expression in osteoblast inhibited osteoclastogenesis, moreover, buffer (Beyotime Institute of Biotechnology, Nantong, China) con- Arl6ip5 was a binding partner of RANKL and its over-expression let taining protease and phosphatase inhibitors (Halt Protease Inhibi- to the ER retention of RANKL. tor Cocktail, Thermo Fisher Scientific, Rockford, IL, USA). The same amount of protein was used for immunoblotting as indicated before 2. Materials and methods [18]. Quantification of band intensities was performed using Image J software (National Institute of Health, Bethesda, MD, USA). 2.1. Reagents and antibodies For Immunoprecipitation analysis, cells with plasmids co-transfected were solubilized in IPB buffer (10 mM Tris pH7.6, Fetal bovine serum (FBS), L-glutamine, antibiotics, alpha- 5 mM EDTA, 5 mM EGTA, 1 mM sodium orthovanadate,0.1 mM modified essential medium (a-MEM) and trypsin/EDTA were ob- PMSF) with 0.5% NP40 at 4 C for 30 min with gentle rotation. tained from Gibco (Life Technologies, Grand Island, NY, USA). PTH Following centrifugation of cell debris (12,000 rpm, 15 min), the and anti-FLAG antibody, anti-HA antibody and anti-EGFP antibody supernatants were quantified and incubated with Protein G were from Sigma (SigmaeAldrich, St. Louis, MO, USA). RANKL, Agarose (Upstate, Lake Placid, NY, USA) then primary antibodies LAMP-1, Calnexin, OPG and b-actin antibodies were from Santa were added to supernatant and incubated overnight at 4 C with Cruz (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). Arl6ip5 gentle rotation. The immunobead-bound protein complexes were antibody was used as implied before [19]. washed and dissociated by boiling in 2 X Laemmli sample buffer and the eluted protein were then analyzed by immunoblotting. 2.2. Cells and plasmids

2.5. Co-culture assay and TRAP staining UAMS-32 stromal/osteoblastic was a kind gift from Prof. Charles A O'Brien (University of Arkansas for Medical Sciences, Little Rock, POBs were seeded into 48 wells culture plate with density of AR, USA) and maintained in a-MEM contained 10% FBS, 2 mM 1000 per well for co-cultured with RAW264.7 cells and 5000 per L-glutamine and antibiotics [20]. Primary calvarial osteoblasts well with BMMs, 2 days later, 5000 RAW264.7 or 20,000 BMMs (POBs) were prepared as indicated before [18]. The POBs used in were added. At the same time, 100 nM PTH was added and the present study was 2e4 passages. Bone marrow macrophages medium were half-changed every 2 days with consistent PTH (BMMs) were harvested from mice femurs and tibia as indicated treatment for 4e6 days for RAW264.7 cells and 7e9 days for BMMS, elsewhere [18]. Raw264.7 cell was purchased from ATCC and the cells were evaluated for the osteoclast formation with TRAP maintained in a-MEM with 10% FBS and antibiotics. UAMS-32 cells staining using TRACP & ALP double-stain Kit (TaKaRa Bio Inc., Otsu, with stable plasmid expression were selected under 100 mg/ml Japan). Geneticin (SigmaeAldrich, St. Louis, MO, USA). The cDNA extracted from UAMS-32 cells was used as template for plasmid construction. FLAG-RANKL and HA-Arl6ip5 were con- 2.6. ELISA and immunofluorescence structed into pcDNA-3.1(þ) with FLAG tag in COOH terminus of mouse RANKL and HA sequence in the COOH terminus of mouse Soluble RANKL in culture medium was detected with mouse Arl6ip5. EGFP-RANKL was constructed into EGFP-C1 with RANKL ELISA from R&D(R&D Systems, Inc., Minneapolis, MN, primers (50 to 30): Sense-CTACTCGAGCTATGCGCCGGGGCCAGCCG; USA) followed the protocol provided. For the localization of Antisense-GCCAAGCTTGGTCTATGTCCTGAACTT. The truncates of endogenous RANKL to LAMP-1, Calnexin and HA in HA-Arl6ip5 FLAG-RANKL were constructed on the backbone of full length over-expressed UAMS-32 cells, cells were fixed and blocked in FLAG-RANKL, the primers were showed below (50 to 30): FL-FLAG- PBST contained normal goat serum at RT for 1 h then incubated RANKL-sense: GCCAAGCTTGCCACCATGCGCCGGGCCAGCCG; D1-44 with diluted primary antibody at 4 C overnight. Then cells were sense: GCCAAGCTTATGGCCTCCCGCTCCATGTTC; D1-100 sense: treated with corresponded secondary antibodies with Alexa-488 or GCCAAGCTTATGCAGGACTCGACTCTGGAGAG; FLAG-RANKL-antisense: Alexa-555 tagged and incubated at 37 C for 2e3 h. Images were CGCCTCGAGTCACTTGTCATCGTCGTCCTTGTAGTCGTCTATGTCCTGAACTTTG. acquired with Olympus Laser Scanning Confocal Microscope EGFP-RANKL with types of Cysteine mutations were constructed with (Olympus Corporation, Tokyo, Japan) and merged with Image J QuikChange™ Site-Directed Mutagenesis Kit (Stratagene, La Jolla, California, software. USA) followed the protocol provided. The NH2 and COOH terminus trun- cates of HA-Arl6ip5 were constructed on the backbone of full length HA- 2.7. Statics Arl6ip5, the primers were showed below (50 to 30): FL-HA-Arl6ip5-sense: CGCGGATCCGCCACCATGGACGTGAACCTCGCC; DN-HA-Arl6ip5-sense: Results were expressed as the mean ± SEM. Statistical signifi- CGCGGATCCGCCACCATGGTAGTGAGCAATCTGCTCTAT; DC-HA-Arl6ip5- cance was identified by Student's t test or one-way ANOVA where antisense: CGGTCTAGATTAAGCGTAGTCTGGGACGTCGTATGGGTAGCGAAG- appropriate, with probability P < 0.05 being considered significant. Y. Wu et al. / Biochemical and Biophysical Research Communications 464 (2015) 1275e1281 1277

3. Results of UAMS-32 cells that received PTH treatment was about 10 pg/L, which decreased to 5 pg/L in the medium from Arl6ip5OE cells To access the effect of osteoblastic Arl6ip5 over-expression on (P < 0.05) (Fig. 1C). However, the transcription (Fig. 1D) and protein osteoclastogenesis, a stable stromal/osteoblastic UAMS-32 cell line expression (Fig. 1E) of RANKL that induced by PTH treatment were with HA-tagged Arl6ip5 (HA-Arl6ip5) over-expression (Arl6ip5OE) comparable between Arl6ip5OE cells and controls. The transcrip- was constructed and used for co-culture assay with Raw264.7 cells. tional level of Csf-1, Sema3a, Sema3b, Sema7b and Wnt5a, coupling TRAP positive cells that induced by PTH were 40% reduced factors that had been demonstrated to regulate osteoclast forma- (P < 0.05) in co-cultures with Arl6ip5 OE compared to the co- tion [2], were also not significantly changed (Fig. 1D). cultures with control cells (Fig. 1A). In consistence, in co-cultures To investigate whether osteoblastic Arl6ip5 over-expression with primary calvarial osteoblast (POBs) and bone marrow mac- inhibited sRANKL level via its regulation on RANKL intracellular rophages (BMMs), the formation of TRAP positive cells that induced trafficking, immunostaining was used to analyze the subcellular by PTH treatment was significantly inhibited by re-expressing HA- localization of endogenous RANKL in UAMS-32 cells with Arl6ip5 D D Arl6ip5 into Arl6ip5 knockout POB (Arl6ip5 2/ 2 POB) (P < 0.05) over-expression. As previously reported, most of newly synthesized (Fig. 1B). Furthermore, when treated the BMMs (Fig. S1A)or endogeous RANKL protein was co-localized with lysosome marker Raw264.7 cells (Fig. S1B) with conditional medium collected from LAMP-1 in control cells, but it was mainly localized in ER but not in PTH treated Arl6ip5OE cells, the TRAP positive cells were signifi- lysosome in Arl6ip5OE cells as reflected by the substantial co- cantly decreased compared to controls. localization with ER marker Calnexin (Fig. 2A and B). This subcel- The RANKL is the principal factor for inducing osteoclast for- lular localization pattern of RANKL in Arl6ip5OE cells was also mation [3]. The level of soluble RANKL (sRANKL) in culture medium observed in UAMS-32 cells with stable over-expression of RANKL

Fig. 1. Arl6ip5 over-expression in osteoblast inhibited osteoclast formation. (A) Osteoclast formation in co-cultures of HA-Arl6ip5 over-expressed UAMS-32 cells and D D RAW264.7 cells, treated with PTH (100 nM). (B) Effect of exogenous expression of HA-Arl6ip5 to Arl6ip5 deficient (Arl6ip5 2/ 2) POBs on PTH-induced osteoclast formation in co- cultures with bone marrow macrophages (BMMs). The parallel experiments with immunoblotting were used to verify the effect of exogenous expression. Representative images were shown. Scale bar ¼ 100 mm. (C) The effect of HA-Arl6ip5 over-expression on sRANKL induced by PTH was evaluated by ELISA analysis in UAMS-32 cells. (D) q-PCR analysis was used to analyze the effect of HA-Arl6ip5 over-expression on the transcription of couplings factors (Csf-1, RANKL, Sema3a, Sema3b, Sema7a and Wnt5a) that induced by PTH treatment in UAMS-32 cells. (E) Immunoblotting was used to measure the protein expression of RANKL, HA and b-actin in HA-Vector and HA-Arl6ip5 treated UAMS-32 cells that with PTH treatment. *P < 0.05 by Student's t test in A and B or by one way ANOVA in C. For all panels, each bar represents mean ± SEM (n ¼ 3e4). 1278 Y. Wu et al. / Biochemical and Biophysical Research Communications 464 (2015) 1275e1281

Fig. 2. HA-Arl6ip5 over-expression led to ER retention of RANKL in UAMS-32 cells. HA-Vector or HA-Arl6ip5 were transfected into UAMS-32 cells, 48 h later, the subcellular localization of endogenous RANKL with LAMP-1 (lysosome marker) (A) and Calnexin (ER marker) (B) were analyzed with immunostaining. Representative images were shown. Scale bar ¼ 10 mm.

that tagged enhanced green fluorence protein on NH2 terminus According to TMHMM, a tool for prediction of transmembrane (EGFP-RANKL) (data not shown). Moreover, the EGFP-RANKL co- helices in [23], the AA 1e36 and 148e188 of mouse Arl6ip5 localized with HA-Arl6ip5 in Arl6ip5OE cells (Fig. S2). were in lumen and AA 36e148 was transmembrane (data not Proteineprotein interaction is important for the subcellular shown). The truncates with AA 1e36 deletion (HA-Arl6ip5-DN) but localization of RANKL [11,12]. The endogenous RANKL precipitated not the 148e188 deletion (HA-Arl6ip5-DC) decreased the forma- with Arl6ip5 (Fig. 3A). In UAMS-32 cells with FLAG-RANKL and HA- tion of RANKL-Arl6ip5 complex (Fig. 4A). Moreover, the localization Arl6ip5 co-expression or with HA-Arl6ip5 and EGFP-RANKL co- of RANKL was diffuse and only partially co-localized to HA-Arl6ip5 expression, the interaction between exogenous RANKL and Arl6ip5 in the cells with HA-Arl6ip5-DN over-expression (HA-Arl6ip5- was also found (Fig. S3A-B). Moreover, the formation of RANKL- DNOE) (Fig. 4B). Furthermore, the sRNAKL level in the cultured OPG complex was decreased in Arl6ip5OE cells when compared to medium of HA-Arl6ip5-DNOE was significant higher than its level in the controls (Fig. 3B). Furthermore, 1 to 44 amino acid (AA) deletion HA-Arl6ip5-FLOE or HA-Arl6ip5-DCOE (Fig. 4C). The formation of on NH2 terminus of RANKL retained the interaction between FLAG- TRAP positive cells was also significant higher in co-cultures with RANKL and HA-Arl6ip5, which was lost when the 1 to 100 AA on HA-Arl6ip5-DNOE and Raw264.7 than that in co-cultures with the RANKL were truncated (Fig. 3C). As the deletion of 1e100 AA on HA-Arl6ip5-FLOE or HA-Arl6ip5-DCOE (Fig. 4D). RANKL disrupted the transmembrane domain of RANKL and let to the accumulation of RANKL in cytosol but not in ER where Arl6ip5 4. Discussion localized to (data not shown), the contribution of AA in the trans- membrane and lumen domain of RANKL to the interaction between Arl6ip5 is an ER localized protein and has been demonstrated for RANKL and Arl6ip5 was further analyzed. Cysteine amino acid (Cys) its role in the regulation of the trafficking of membrane proteins is known important for protein trafficking and proteineprotein [16,17]. In present study, we found that Arl6ip5 was a novel binding interaction of secreted proteins [22]. Cys was existed on amino acid partner of RANKL in osteoblast. Its over-expression in osteoblast 63, 87, 115 and 270 of murine RANKL. The interaction between leads to the ER retention of RANKL thereby decreased EGFP-RANKL with all four Cys were mutated to Glycine (Gly) and osteoclastogenesis. HA-Arl6ip5 was dramatically decreased, while keeping Cys 63 and/ Previous studies demonstrated that newly synthesized RANKL or Cys 87 in RANKL protein sequence restored the interaction be- in osteoblast was stored in secretory lysosomes, OPG and Vps33a tween RANKL and Arl6ip5 (Fig. 3D). were involved in the transport of RANKL from Golgi apparatus to Y. Wu et al. / Biochemical and Biophysical Research Communications 464 (2015) 1275e1281 1279

Fig. 3. Arl6ip5 was a binding protein of RANKL. (A) UAMS-32 cells were pretreated with PTH (100 nM) for 24 h and endogenous proteineprotein interaction between Arl6ip5 and RANKL was determined by IP with Arl6ip5 or RANKL antibodies (Ab) followed by immunoblotting. The corresponding IgG was used as a negative control for immunoprecipitation. # indicated the unspecified band and * indicated the target bands. (B) The interaction between endogenous OPG and RANKL in UAMS-32 cells with HA-Arl6ip5 over-expression was analyzed with IP analysis. (C) UAMS-32 cells were co-transfected with Vector, full length (FL) FLAG-RANKL and two truncated FLAG-RANKL (D1-44 and D1-100) with HA-Arl6ip5 for 48 h, Co-immunoprecipitation (Co-IP) using anti-FLAG antibody, and immunoblotting (IB) using the indicated antibodies were carried out to determine the interaction of RANKL and Arl6ip5. (D) The interaction between HA-Arl6ip5 and EGFP-RANKL with different Cysteine (Cys) mutated to Glycine (Gly) in UAMS-32 cells was analyzed with Co-IP. In C and D, b-actin was used as a loading control.

secretory lysosomes (major pathway) [11,12]. When lysosome oligomerization [17]. Evidences from the homologue of Arl6ip5 trafficking pathway was blocked or dysfunction of Vps33a- demonstrated that the COOH-terminal domain of Arl6ip5 con- mediated transport, RANKL was transported directly from Golgi tained the signal for its ER localization and was important for the to membrane (alternative pathway or minor pathway). In present binding with proteins such as TR1 [26,27]. However, it was the NH2- study, we found that both endogenous and exogenous RANKL were terminal but not the COOH-terminal responsible for the interaction mainly localized in ER but not in lysosomes in HA-Arl6ip5OE. between Arl6ip5 and RANKL, as revealed by immunoprecipitation Moreover, these data indicated that Arl6ip5 over-expression in analysis with types of truncates. According to the prediction with osteoblast may interrupt the RANKL major and/or minor trafficking TMHMM [23], both the NH2-terminal and COOH-terminal is inside pathway. As the expression of Arl6ip5 can be regulated by some the membrane where Arl6ip5 localized. However, in prediction, the osteotropic factors such as BMP-2 and PTH [18], it is possible that 1 to 36 AA of Arl6ip5 was inclined to form coil shape structure but these factors might control the RANKL homeostasis by the regula- the 148e188 AA was prone to form a helix structure. Whereas tion on the subcellular trafficking of RANKL that mediated by further biochemical evidence for the topology of Arl6ip5 in cell is Arl6ip5. Further studies on the subcellular localization of RANKL in needed, the potential discrepancy on the secondary structure of OE Arl6ip5 that received osteotropic factors treatments would NH2 and COOH terminal of Arl6ip5 may be important for its se- demonstrate this possibility. lection of binding partners. As Arl6ip5 bound to RANKL, moreover, Dependent on cell types and stimulus, Arl6ip5 has different over-expression of Arl6ip5 in UAMS-32 cells obviously disturbed binding partners such as EAAC-1 [14,17], Rab1 [16] and XRCC-1 [24]. endogenous RANKL binding to OPG, it is possible that Arl6ip5 In osteoblast, we revealed that Arl6ip5 was a novel binding partner competes with the proteins involved in major or minor trafficking of RANKL as both endogenous and exogenous Arl6ip5 immuno- pathways to interact with RANKL and sequesters RANKL in ER in precipitated with RANKL. As RANKL can form homo- or hetero- Arl6ip5 over-expressed cells. multimers [25], our finding supported that Arl6ip5 was inclined to Our research further showed that correct structure of RANKL bind with the proteins that have the property to form was necessary for RANKL-Arl6ip5 complex formation. Mutated all 1280 Y. Wu et al. / Biochemical and Biophysical Research Communications 464 (2015) 1275e1281

Fig. 4. The NH2 terminus of Arl6ip5 was important for the subcellular localization of RANKL. (A) UAMS-32 cells were co-transfected with Vector, full length (FL) HA-Arl6ip5 and two truncated HA-Arl6ip5 (DN and DC) with FLAG-RANKL for 48 h, Co-IP using anti-HA antibody, and immunoblotting with the indicated antibodies were carried out to determine the interaction of RANKL and Arl6ip5. * marked the target bands. (B) UAMS-32 cells with FL and truncated HA-Arl6ip5 expression were analyzed by immunostaining to detect the localization of endogenous RANKL with HA-Arl6ip5. Representative images were shown and scale bar ¼ 10 mm (C) UAMS-32 cells with HA-vector (VectorOE), HA-Arl6ip5-FL (FLOE), HA-Arl6ip5-DN(DNOE) and HA-Arl6ip5-DC(DNOE) stable over-expression were treated with PTH (100 nM) for 24 h, then sRANKL was detected by ELISA assay. (D) Osteoclast formation in cocultures with types of HA-Arl6ip5 over-expressed UAMS-32 cells and RAW264.7 cells, treated with PTH (100 nM). In whole panel, n ¼ 4, bars represents mean ± SEM, **, P < 0.01 and *, P < 0.05 by one-way ANOVA for FLOE vs DNOE; NS means no significance between VectorOE and DNOE by one-way ANOVA.

four Cysteine residues that existed in murine RANKL to Glycine Conflicts of interest residues or mutated Cys 63 and 87 to Gly in RANKL protein dramatically decreased the capacity of RANKL to bind with Arl6ip5. None. Cys 63 is in and Cys 87 is adjacent to the transmembrane region of RANKL (data not shown), Cys63Gly and Cys87Gly mutation may Acknowledgements change transmembrane or lumen structure of RANKL, thereby disrupt its binding with Arl6ip5. Interestingly, Cys amino acid is This work was supported by the National Natural Science important for some post-translational modifications such as pal- Foundation of China (81200645) to Yu Wu. mitoylation [28,29]. The palmitoylation on Cys plays roles in pro- tein trafficking and proteineprotein interactions. Although it was Appendix A. Supplementary data unclear whether RANKL can be palmitoylated, it had been shown that the Cys in FasL, which is one member of TNF super family like Supplementary data related to this article can be found at http:// RANKL, was palmitoylated [30]. Current efforts focus on the post- dx.doi.org/10.1016/j.bbrc.2015.07.119. translational modification especially palmitoylation on Cys amino acids of RANKL in an attempt to gain further insight into the Transparency document mechanisms for the formation of Arl6ip5-RANKL complex and the regulation of RANKL intracellular trafficking. Transparency document related to this article can be found In total, the current findings unraveled an un-recognized role online at http://dx.doi.org/10.1016/j.bbrc.2015.07.119. of osteoblastic Arl6ip5 in the maintenance of RANKL homeostasis by its binding with RANKL, which led to the localization of RANKL References to ER, thereby suppressed osteoclast formation. Taken the critical role of RANKL on the osteoclastogenesis and bone metabolism, [1] N.A. Sims, J.H. Gooi, Bone remodeling: multiple cellular interactions required further study with Arl6ip5 transgenic mice is warranted for the for coupling of bone formation and resorption, Semin. Cell Dev. Biol. 19 (2008) 444e451. in vivo role of over-expression Arl6ip5 on bone metabolism and [2] N.A. Sims, T.J. Martin, Coupling the activities of bone formation and resorp- bone disease. tion: a multitude of signals within the basic multicellular unit, Bonekey Rep. 3 Y. Wu et al. / Biochemical and Biophysical Research Communications 464 (2015) 1275e1281 1281

(2014) 481. [17] A.M. Ruggiero, Y. Liu, S. Vidensky, S. Maier, E. Jung, H. Farhan, M.B. Robinson, [3] W.J. Boyle, W.S. Simonet, D.L. Lacey, Osteoclast differentiation and activation, H.H. Sitte, J.D. Rothstein, The endoplasmic reticulum exit of glutamate Nature 423 (2003) 337e342. transporter is regulated by the inducible mammalian Yip6b/GTRAP3-18 pro- [4] C.A. O'Brien, Control of RANKL expression, Bone 46 (2010) 911e919. tein, J. Biol. Chem. 283 (2008) 6175e6183. [5] J.R. Edwards, G.R. Mundy, Advances in osteoclast biology: old findings and [18] Y. Wu, M. Yang, J. Fan, Y. Peng, L. Deng, Y. Ding, R. Yang, J. Zhou, D. Miao, Q. Fu, new insights from mouse models, Nat. Rev. Rheumatol. 7 (2011) 235e243. Deficiency of osteoblastic Arl6ip5 impaired osteoblast differentiation and [6] T. Nakashima, M. Hayashi, T. Fukunaga, K. Kurata, M. Oh-Hora, J.Q. Feng, enhanced osteoclastogenesis via disturbance of ER calcium homeostasis and L.F. Bonewald, T. Kodama, A. Wutz, E.F. Wagner, J.M. Penninger, H. Takayanagi, induction of ER stress-mediated apoptosis, Cell Death Dis. 5 (2014) e1464. Evidence for osteocyte regulation of bone homeostasis through RANKL [19] Z. Gong, Y. Shi, Z. Zhu, X. Li, Y. Ye, J. Zhang, A. Li, G. Li, J. Zhou, JWA deficiency expression, Nat. Med. 17 (2011) 1231e1234. suppresses dimethylbenz[a]anthracene-phorbol ester induced skin papil- [7] J. Xiong, M. Onal, R.L. Jilka, R.S. Weinstein, S.C. Manolagas, C.A. O'Brien, Matrix- lomas via inactivation of MAPK pathway in mice, PLoS One 7 (2012) e34154. embedded cells control osteoclast formation, Nat. Med. 17 (2011) 1235e1241. [20] C.A. O'Brien, I. Gubrij, S.C. Lin, R.L. Saylors, S.C. Manolagas, STAT3 activation in [8] Q. Fu, R.L. Jilka, S.C. Manolagas, C.A. O'Brien, Parathyroid hormone stimulates stromal/osteoblastic cells is required for induction of the receptor activator of receptor activator of NFkappa B ligand and inhibits osteoprotegerin expres- NF-kappaB ligand and stimulation of osteoclastogenesis by gp130-utilizing sion via protein kinase A activation of cAMP-response element-binding pro- cytokines or interleukin-1 but not 1,25-dihydroxyvitamin D3 or parathyroid tein, J. Biol. Chem. 277 (2002) 48868e48875. hormone, J. Biol. Chem. 274 (1999) 19301e19308. [9] Q. Fu, S.C. Manolagas, C.A. O'Brien, Parathyroid hormone controls receptor [21] X. Wang, A. Spandidos, H. Wang, B. Seed, PrimerBank: a PCR primer database activator of NF-kappaB ligand via a distant transcriptional for quantitative gene expression analysis, 2012 update, Nucleic Acids Res. 40 enhancer, Mol. Cell Biol. 26 (2006) 6453e6468. (2012) D1144eD1149. [10] R.D. Nerenz, M.L. Martowicz, J.W. Pike, An enhancer 20 kilobases upstream of [22] J.E. Smotrys, M.E. Linder, Palmitoylation of intracellular signaling proteins: the human receptor activator of nuclear factor-kappaB ligand gene mediates regulation and function, Annu. Rev. Biochem. 73 (2004) 559e587. dominant activation by 1,25-dihydroxyvitamin D3, Mol. Endocrinol. 22 (2008) [23] A. Krogh, B. Larsson, G. von Heijne, E.L. Sonnhammer, Predicting trans- 1044e1056. topology with a hidden Markov model: application to [11] Y. Kariya, M. Honma, S. Aoki, A. Chiba, H. Suzuki, Vps33a mediates RANKL complete genomes, J. Mol. Biol. 305 (2001) 567e580. storage in secretory lysosomes in osteoblastic cells, J. Bone Min. Res. 24 (2009) [24] S. Wang, Z. Gong, R. Chen, Y. Liu, A. Li, G. Li, J. Zhou, JWA regulates XRCC1 and 1741e1752. functions as a novel base excision repair protein in oxidative-stress-induced [12] S. Aoki, M. Honma, Y. Kariya, Y. Nakamichi, T. Ninomiya, N. Takahashi, DNA single-strand breaks, Nucleic Acids Res. 37 (2009) 1936e1950. N. Udagawa, H. Suzuki, Function of OPG as a traffic regulator for RANKL is [25] E. Douni, V. Rinotas, E. Makrinou, J. Zwerina, J.M. Penninger, E. Eliopoulos, crucial for controlled osteoclastogenesis, J. Bone Min. Res. 25 (2010) G. Schett, G. Kollias, A RANKL G278R mutation causing osteopetrosis identifies 1907e1921. a functional amino acid essential for trimer assembly in RANKL and TNF, Hum. [13] Y. Kariya, M. Honma, A. Hanamura, S. Aoki, T. Ninomiya, Y. Nakamichi, Mol. Genet. 21 (2012) 784e798. N. Udagawa, H. Suzuki, Rab27a and Rab27b are involved in stimulation- [26] M. Abdul-Ghani, P.Y. Gougeon, D.C. Prosser, L.F. Da-Silva, J.K. Ngsee, PRA dependent RANKL release from secretory lysosomes in osteoblastic cells, isoforms are targeted to distinct membrane compartments, J. Biol. Chem. 276 J. Bone Min. Res. 26 (2011) 689e703. (2001) 6225e6233. [14] C.I. Lin, I. Orlov, A.M. Ruggiero, M. Dykes-Hoberg, A. Lee, M. Jackson, [27] T. Arano, S. Fujisaki, M.J. Ikemoto, Identification of tomoregulin-1 as a novel J.D. Rothstein, Modulation of the neuronal glutamate transporter EAAC1 by addicsin-associated factor, Neurochem. Int. 71 (2014) 22e35. the interacting protein GTRAP3-18, Nature 410 (2001) 84e88. [28] J. Charollais, F.G. Van Der Goot, Palmitoylation of membrane proteins (Re- [15] M.E. Butchbach, L. Lai, C.L. Lin, Molecular cloning, gene structure, expression view), Mol. Membr. Biol. 26 (2009) 55e66. profile and functional characterization of the mouse glutamate transporter [29] C. Aicart-Ramos, R.A. Valero, I. Rodriguez-Crespo, Protein palmitoylation and (EAAT3) interacting protein GTRAP3-18, Gene 292 (2002) 81e90. subcellular trafficking, Biochim. Biophys. Acta 1808 (2011) 2981e2994. [16] S. Maier, V. Reiterer, A.M. Ruggiero, J.D. Rothstein, S. Thomas, R. Dahm, [30] F. Guardiola-Serrano, A. Rossin, N. Cahuzac, K. Luckerath, I. Melzer, S. Mailfert, H.H. Sitte, H. Farhan, GTRAP3-18 serves as a negative regulator of Rab1 in D. Marguet, M. Zornig, A.O. Hueber, Palmitoylation of human FasL modulates protein transport and neuronal differentiation, J. Cell Mol. Med. 13 (2009) its cell death-inducing function, Cell Death Dis. 1 (2010) e88. 114e124.