Calmodulin interacts with Rab3D and modulates osteoclastic bone resorption

Calmodulin is a highly versatile protein that regulates intracellular calcium homeostasis and is involved in a variety of cellular functions including cardiac excitability, synaptic plasticity and signaling transduction. During osteoclastic bone resorption, calmodulin has been reported to concentrate at the ruffled border membrane of osteoclasts where it is thought to modulate bone resorption activity in response to calcium. Here we report an interaction between calmodulin and Rab3D, a small exocytic GTPase and established regulator osteoclastic bone resorption. Using yeast two-hybrid screening together with a series of protein-protein interaction studies, we show that calmodulin interacts with Rab3D in a calcium dependent manner. Consistently, expression of a calcium insensitive form of calmodulin (i.e. CaM1234) perturbs calmodulin-Rab3D interaction as monitored by bioluminescence resonance energy transfer (BRET) assays. In osteoclasts, calmodulin and Rab3D are constitutively co-expressed during RANKL-induced osteoclast differentiation, co-occupy plasma membrane fractions by differential gradient sedimentation assay and colocalise in the ruffled border as revealed by confocal microscopy. Further, functional blockade of calmodulin-Rab3D interaction by calmidazolium chloride coincides with an attenuation of osteoclastic bone resorption. Our data imply that calmodulin- Rab3D interaction is required for efficient bone resorption by osteoclasts in vitro.


Results
Calmodulin interacts with Rab3D. We have previously established a yeast two-hybrid approach to successfully uncover novel Rab3D interacting partners such as Tctex-1 17 . Here we identify calmodulin as an additional binding partner of Rab3D. The interaction of calmodulin with Rab3D was verified by a yeast two hybrid assay, using a histidine-deficient plate (Fig. 1A). To further examine the interaction of calmodulin and Rab3D, we generated Rluc-camodulin and EYFP-Rab3D fusion protein constructs and performed BRET protein-protein interaction assays. As shown in Fig. 1B, co-expression of Rluc-calmodulin and EYFP-Rab3D resulted in a significant BRET signal when compared to the co-expression of Rluc and EYFP. To further confirm the interaction, an in vitro calmodulin sepharose-pull down assay was performed. Rab3D was cloned into a mammalian expression vector with an N terminal Flag-tagged (Fig. 1C). Flag-Rab3D proteins were prepared from COS cells transfected with pcDNA3.1-Flag-Rab3D expressing plasmids. COS cell lysates were harvested and subjected to immobilized calmodulin sepharose in the presence or absence of 2 mM calcium. As shown in Fig. 1C, Flag-Rab3D proteins bound immobilized calmodulin saphorose in the presence (but not in the absence) of calcium, indicative of a calcium dependent binding dependency.
Calmodulin calcium insensitive mutant perturbs its interaction with Rab3D. Considering that calmodulin has four calcium binding sites via four aspartic acid residues 18 and acts as a calcium modulator in the calcium sensitive regulation of many cellular processes, we next examined if calcium binding site of calmodulin is required for the interaction of calmodulin with Rab3D. For this, we generated a Rluc-calmodulin construct in which four aspartic acid residues at position 23, 59, 96, 132 were substituted with alanine, mimicking a  calcium insensitive form of calmodulin 18 (Fig. 1D). BRET assay results showed that the calcium insensitive form of camodulin attenuated the interaction with Rab3D (Fig. 1E).

The preferential interaction between Calmodulin and Rab3D in its GTP-bound conformation.
Rab GTPases embed in organelle membranes via C-terminal prenylation moties where they function as molecular switches that oscillate between GTP "active" and GDP "inactive" conformations. In their active state, Rabs recruit GTP-dependent effector proteins through which they elicit their biological function at various stages of vesicular transport. Therefore, we next asked whether the interaction between Calmodulin and Rab3D was dependent on the nucleotide and/or prenylation status of Rab3D. To access this, we employed several well characterised Rab3D variants 16 , which selectively disrupt the GDP/GTP exchange i.e. GTP-bound Rab3D (Rab3DQ81L), nucleotide empty RAB3D (Rab3DN135I) and prenylation motif deletion of Rab3D (Rab3D CXC) compared to wildtype Rab3D ( Fig. 2A,B). These constructs were successfully expressed as EYFP fusion proteins in transfected COS cells as confirmed by western blot analyses (Fig. 2C). As with other bona fide Rab effector protein, calmodulin exhibited a preferential association with Rab3D when locked in is GTP-bound form (Rab3DQ81L) when compared to wild-type Rab3D, nucleotide-empty (Rab3DN135I) and prenylation motif deletion of Rab3D (Rab3D CXC) in BRET assays (Fig. 2D). These data imply that the interaction of calmodulin with Rab3D is largely influenced by its active GTP-bound state.
Calmodulin and Rab3D are co-expressed during osteoclast formation and co-sediment in membrane fractions of osteoclasts. To begin to probe the potential relevance of the calmodulin-Rab3D association in osteoclasts, we first compared the gene expression profile of calmodulin and Rab3D in osteoclasts and their precursor cells. Bone marrow macrophages (BMM) were cultured in the present of macrophage colony stimulating factor (M-CSF) and receptor activator of NF-κ B ligand (RANKL) for a period of 0, 1, 3, 5 days and then fixed and stained for tartrate resistant acid phosphatase (TRACP) activity, showing the presence of that calmodulin exhibited an enhanced association with a GTP-bound Rab3D (Rab3DQ81L) when compared to wild-type Rab3D, nucleotide empty RAB3D (Rab3DN135I) and prenylation motif deletion of Rab3D (Rab3DΔ CXC) in BRET assays. *Indicates p Value < 0.001 when compared with EYFP and Rluc. # indicates p Value < 0.05 when compared to wild-type Rab3D, nucleotide-empty (Rab3DN135I) and prenylation motif deletion of Rab3D (Rab3DΔ CXC).
Scientific RepoRts | 6:37963 | DOI: 10.1038/srep37963 multinucleated TRACP positive osteoclast like cells (Fig. 3A). In a parallel experiment, semi-quantitative RT-PCR was performed. Calmodulin gene expression appears to be constitutive during osteoclastogenesis by RANKL at an expression kinetic similar to that of Rab3D (Fig. 3B). Osteoclast marker gene expression of TRACP, Cathepsin K, V-ATPase d2, and calcitonin receptor (CTR) were induced by RANKL as compared to 36B4 gene expression as an internal control (Fig. 3B). Further, Western blot analysis showed that Calmodulin protein is constitutively expressed during osteoclastogenesis similar to Rab3D protein expression (Fig. 3C).
Next, sucrose gradient sedimentation assays were performed to examine calmodulin and Rab3D co-fractionation. Rab3D is present in small vesicles (F9-10) and large plasma membrane fractions (P) (Fig. 3D). Interestingly, calmodulin co-fractionated with Rab3D but only in the large membrane faction (P) (Fig. 3D). By comparison, V-ATPase (d2) was also present in both vesicle and membrane fractions (Fig. 3D). Moreover, an association between Rab3D and calmodulin was further confirmed in bone-resorbing osteoclasts by immunofluorescence confocal microscopy (Fig. 4). In this instance colocalisation (yellow colour) was observed upon overlay of individual fluorescent channels for Rab3D (green) and Calmodulin (red), which were detected using antibodies specific to Rab3D and calmodulin, as validated in the Western blot analysis above (i.e. Fig. 3C). Interestingly, a subset of Rab3D-calmodulin colocalisation was observed within the F-actin ring/sealing zone (blue) that typically denotes the ruffled border membrane (Fig. 4A, region circumscribed by red line). This colocalisation was confirmed by correlative linescan analysis (Fig. 4B) which revealed overlap between the fluorescent peaks of Rab3D and calmodulin within the ruffled border region (Fig. 4A, white dashed line).
Functional blockade of the calmodulin-Rab3D interaction by calmidazolium chloride attenuates osteoclastic bone resorption. Finally, we set out to define the impact of the interaction of calmodulin with Rab3D on osteoclast function. To this end, we first tested the effect of calmidazolium chloride on the interaction of calmodulin and Rab3D using BRET assays. Interestingly, calmidazolium chloride perturbs the association of calmodulin and Rab3D, Rab3DQ81L, Rab3DN135I and Rab3DΔ CXC (Fig. 5A). Further, to examine the effect of calmidazolium on osteoclastic bone resorption, BMM derived osteoclasts were seeded into bone slices in the presence and absence of calmidazolium chloride for 24 hours. Treatment of osteoclasts with calmidazolium chloride inhibited osteoclastic bone resorption (Fig. 5B) but did not affect the total number (Fig. 5C) and morphology of TRACP positive osteoclastic like cells at 1 μ M, and 5 μ M (Fig. 5D). Taken together, these data suggest that

Discussion
Calmodulin has versatile roles in regulating intracellular calcium homeostasis and diverse cellular processes including osteoclastic bone resorption. In this study we document that calmodulin interacts with Rab3D in a calcium-dependent manner. Both calmodulin and Rab3D proteins co-occupy membranes factions by a sucrose gradient ultracentrifugation sedimentation assay and colocalise in the ruffled border by confocal microscopy. Functional blockade of calmodulin and Rab3D interaction by calmidazolium chloride resulted in an attenuation of osteoclastic bone resorption. Considering that calmodulin is concentrated on the ruffled border membrane in osteoclasts 8 , and that Rab3D is a functional requirement for ruffled border maintenance 17 , it is plausible that calmodulin, through its interaction with Rab3D facilitates the delivery and/or calcium sensitively of Rab3D-bearing vesicles at the ruffled border membrane during bone resorption (Fig. 6).
Bone resorption by osteoclasts is a multi-step process which culminates in the removal of an inorganic mineral layer primarily composed of cystralline hydroxylapatite and subsequent degradation of the underlying organic phases. This process involves continual content delivery and membrane recycling through vesicle trafficking, governed largely small Rab GTPases. Rab proteins have been implicated in the regulation of distinct events in vesicle transport on the exocytotic, endocytotic and transcytotic pathways 19,20 . Because of this, in recent years, there has been mounting interest surrounding the role of Rab3D in intracellular transport. Rab3D was shown to mediate exocytosis in mast cells 21 , adipocytes 22 , and acini cells 23 . GTPase-deficient Rab3D decreased nutrient induced insulin release 24 . Moreover, actin coating of secretory granules during regulated exocytosis has been shown to correlate with the release of Rab3D 25 . In addition, Rab3D has been suggested to play a role in the regulation of apically directed transcytosis in rat hepatocytes 26 and appears to be essential for apical transport in polarized epithelia 27 . Our previous data indicates that Rab3D modulates a post-TGN trafficking step that is required for osteoclastic bone resorption 16 . Furthermore, we have shown that Rab3D interacts with Tctex-1 which regulates microtubule-directed trafficking of Rab3D vesicles via cytoplasmic dynein 17 . The present study further extends the role of Rab3D in bone resorption by binding to calmodulin; a molecule previously implicated in acid secretion and bone resorption [10][11][12] .
During the formation of ruffled border membrane domains, the osteoclastic plasmalemma can be further divided into several specialised functional domains. These include the basolateral domain and a functional secretory domain 28,29 . In fact more recent evidence suggests that the ruffled border is not a continuous domain but rather segregated into an "uptake" and "secretory" domain reflecting its dynamic endo-exocytic intracellular trafficking routes 30 . Therefore it is likely that Rab3D mediates bone resorption along the exocytic pathway through its direct interaction with calmodulin that is located in the ruffled border membranes. In other systems, calmodulin has been shown to stimulate GTP binding to Rab3A that is complexed with GDI, which leads to the formation of an active GTP-bound form of the Rab3A-Ca (2+ )/Calmodulin complex in synaptic membranes of activated nerve termini 31 . Similarly, interaction between Rab3B and calmodulin was Ca (2+ )-dependent 32 . These findings provide evidence that Rab3B is primarily localized with the particulate fraction and that Ca (2+ )/calmodulin could regulate function of this GTPase in platelets 32 . It remains to be elucidated whether other Rab proteins might also interact with calmodulin and facilitate these molecular pathways in osteoclasts and other cells.
Previously, structural analysis revealed four Ca (2+ )-binding domains in calmodulin that are important for the function of calmodulin, together with hydrophobic regions represent the sites of interaction with pharmacological agents 33 . The three-dimensional structure of calmodulin has been determined crystallographically at 3.0 A resolution. The molecule consists of two globular lobes connected by a long exposed alpha-helix, and each lobe is able to bind two calcium ions through helix-loop-helix domains 34 . The flexibility of the protein may explain the fact that calmodulin is able to bind many different targets 35 . We have found that the interaction of calmodulin with Rab3D is calcium dependent. During osteoclastic bone resorption, free calcium is to be released in the resorption compartment, which could in turn further facilitate the interaction of calmodulin with Rab3D. It has been suggested that a Rab3-calmodulin complex generated by elevated Ca (2+ ) concentrations mediated at least some of the effects of the GTPase and limited the number of exocytic events that occurred in response to secretory stimuli 36 . We propose that the recruitment of calmodulin by Rab3D is an important requirement for osteoclast-mediated bone resorption. Identification of interacting partners to Rab proteins will be important for drug design, for instance, Plekhm1 was found to co-localize with Rab7 to late endosomal/lysosomal vesicles, a putative function in vesicular transport in the osteoclast. This has been implicated in the development of osteopetrosis 37 . Interestingly, alendronate inactivates osteoclasts by mechanisms that impair their intracellular vesicle transport, with apoptosis being only a secondary phenomenon to this 38 . The anti-resorptive activity of NE10790 is thus likely due to disruption of Rab-dependent intracellular membrane trafficking in osteoclasts 39 . Given that Rab-dependent intracellular membrane trafficking in osteoclastic bone resorption has been proposed to be a target of nitrogen-containing bisphosphonate drug NE10790 39 , defining Rab3D interacting partners might facilitate the design of the next generation of bisphosphonate drugs.

Materials and Methods
Two-hybrid screening. Mouse Rab3D cDNA was inserted into pBTM116 and used to screen a pVP16based yeast two-hybrid cDNA library as previously described 17 . Briefly, for cDNA library screening, the yeast reporter strain L40 was first transfected with baits pLexA-Rab3D and subsequently transfected with the pVP16 mouse embryo cDNA library using lithium acetate and polyethylene glycol. Library plasmids were grown in the presence or absence of histidine. Positive clones isolated from the cDNA library were further analysed by co-transfection with pLexA-Rab3D and DNA sequencing.
In vitro protein interaction. The full length of mouse Rab3D cDNA was subcloned into Flag-tagged expression vector with CMV promoter to generate a Flag-Rab3D plasmid. To express Rab3D protein, 5 × 10 5 COS-7 cells was transfected with 10 μ g of Flag-Rab3D plasmid using effectant reagents (Qiagen, Sydney). After incubation for 48 hours, transfected cells were washed twice with PBS and lysed for 30 mins on ice with 750 μ l of lysis buffer (50 mM Tris.Cl, pH 7.5, 150 mM NaCl, 0.1% Nonidet P-40, 1 M EDTA, 1 μ g/ml pepstatin A). Following centrifugation at 12,000 RPM for 5 mins at 4 °C, supernatant was collected and 750 μ l of binding buffer (20 mM Tris. Cl, pH 7.5, 50 mM KCl, 100 mM NaCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 5 mM DTT) was added to the supernatant and mixed. One half of the sample was added to 50 μ l of calmodulin immobilized in sepharose beads (Sigma, Sydney) in the presence and absence of 2 mM of calcium. The mixture was incubated overnight at 4 °C with a shaker. The beads were washed three times with 300 μ l of buffer containing equal volumes of lysis buffer and binding buffer. The bound proteins were boiled with 1 x SDS-PAGE sample buffer and analysed by Western blot analysis using an anti-FLAG monoclonal antibody (Sigma, Sydney).