Breast Cancer–Associated Gene 3 (BCA3) Is a Novel Rac1-Interacting Protein


  • The authors state that they have no conflicts of interest.


BCA3 was identified in a yeast two-hybrid screen as a novel Rac1-interacting partner in osteoclasts. BCA3 binds directly to Rac and, in vivo, binds GTP-Rac but not GDP-Rac. Perinuclear co-localization of BCA3 and Rac1 is observed in CSF-1–treated osteoclasts. Overexpression of BCA3 attenuates CSF-1–induced cell spreading. We conclude that BCA3 regulates CSF-1–dependent Rac activation.

Introduction: Rac1, a ubiquitously expressed GTPase, is a mediator of colony-stimulating factor 1 (CSF-1)–dependent actin remodeling in osteoclasts. Because the role of Rac in osteoclasts has not been fully defined, we undertook a yeast two-hybrid screen to identify Rac-interacting partners in these cells.

Materials and Methods: A yeast two-hybrid screen was undertaken using a cDNA library prepared from osteoclast-like cells as prey and either native Rac1 or constitutively active Rac1 (Q61L) as bait. Radiolabeled breast cancer–associated gene 3 (BCA3) protein constructs were generated in vitro using rabbit reticulate lysates and used in vitro binding assays with Rac1. In vivo binding was assessed using myc-tagged Rac1(Q61L) and HA-tagged BCA3. PBD pull-down assays were used to determine if GTP-loaded Rac1 preferentially bound BCA3. Co-localization of Rac1 and BCA3 in osteoclasts was assessed using confocal immunofluorescence. The functional significance of the BCA3–Rac1 interaction was assessed by examining the effect of overexpressing BCA3 in RAW 264.7 cells on the subsequent spreading response to CSF-1.

Results: One of three positive clones from the wildtype Rac1 screen and all three positive clones from the Rac1(Q61L) screen encoded the same protein, BCA3. BCA3 expression in osteoclasts was confirmed by RT-PCR and immunocytochemistry. BCA3 bound directly to Rac1 in vitro. Deletional analysis indicated that amino acids 76–125 in BCA3 are important for its ability to bind Rac. In vivo association of the two proteins was shown by co-immunoprecipitation of BCA3 and Rac1. Only GTP-bound-Rac but not GDP-bound Rac could interact with BCA3 in vivo. Confocal immunocytochemistry showed perinuclear co-localization of BCA3 and Rac1 in CSF-1–treated neonatal rat osteoclasts but not in resting osteoclasts. Overexpression of BCA3 markedly attenuated the spreading response to CSF-1 in RAW 264.7 cells.

Conclusions: These data establish that BCA3 is a novel Rac1-interacting protein and suggest that it may influence the ability of Rac1 to remodel the actin cytoskeleton.


Rac is a small GTPase that belongs to the Ras superfamily. Rac participates in a wide range of biological processes, including rearrangement of the actin cytoskeleton, cell motility, cell transformation, gene transcription, and cell cycle progression.(1,2)

Mammalian Rac has three isoforms: Rac1, Rac2, and Rac3. Rac1 is ubiquitously expressed, whereas the expression of Rac2 is restricted to cells of hematopoietic origin.(3,4) Rac3 is expressed primarily in the brain.(5) Previous work has shown that Rac1 is a key regulator of actin remodeling in osteoclasts. Inhibiting Rac has been shown to inhibit bone resorption.(6) Colony stimulating factor-1 (CSF-1)–induced actin remodeling in osteoclasts requires Rac-1 because inhibition of Rac activity with a dominant negative construct completely blocks CSF-1 actions on the cytoskeleton.(7) PI-3 kinase and Vav appear to lie upstream of Rac in a signaling cascade initiated from the activated CSF-1 receptor, c-fms.(7,8)

Mature osteoclasts are highly motile, multinucleated, terminally differentiated cells of the monocyte/macrophage lineage that are responsible for resorbing mineralized bone.(9) Bone resorption proceeds by a complex process in which cells attach to bone and become polarized against the bone surface. They secrete acid and proteases into a resorption lacunae, formed beneath the cell and delimited by a specialized dynamic adherence structure called the sealing zone, which is composed of actin, vinculin, and several other proteins.(9,10) Although the cellular mechanisms underlying osteoclast-mediated degradation of bone are incompletely understood, observations both in vivo and in in vitro indicate that osteoclast motility is required for bone resorption to occur. A more complete understanding of the basic mechanisms underlying osteoclast motility is crucial to the development of new strategies directed toward the treatment of osteoporosis.

As noted, available evidence indicates that Rac1 plays an important role in modulating osteoclast motility. Although the mechanism(s) by which Rac1 induces cell migration are not fully understood, several key steps have been identified in this signaling pathway.(11) In osteoclasts, activation of c-fms leads to the recruitment and activation of c-src and PI3-kinase to docking sites on the cytoplasmic tail of the receptor. One of c-src's functions seems to be participating in the full activation of PI3-kinase.(12,13) In turn PI3-kinase activates Vav3, a known guanine nucleotide exchange factor for Rac. Finally, Rac itself is activated as reflected by an increase in the cellular content of GTP bound-Rac, the active form of this GTPase.(7,11,12)

Because Rac1-interacting partners are not well characterized in osteoclasts, nor is its role in osteoclast motility fully understood, a yeast two-hybrid library screen was used to identify osteoclast-derived, Rac1-binding partners using a murine osteoclast-like cell cDNA library as prey and Rac1 as bait. We report here a novel interaction between Rac1 and breast cancer–associated gene-3 (BCA3),(14,15) also known as A kinase interacting protein-1 (AKIP1)(16) or KyoT2 binding protein 1 (KBP1).(17–19) This interaction seems to be direct and requires the amino-terminal two thirds of the molecule.


Preparation of osteoclast-like cells

Osteoclast-like cells (OCLs) were prepared by co-culturing primary murine osteoblasts with bone marrow as previously described.(20) Primary murine osteoblasts were obtained by serial collagenase/dispase digestion of neonatal mouse calvariae. Fifty calvariae were dissected from neonatal mice, pooled, and subjected to sequential digestion with 0.1% bacterial collagenase (Boehringer Mannheim Biochemicals, Indianapolis, IN, USA) and 0.2% dispase (Boehringer Mannheim Biochemicals). Digestions were for 10 minutes at 37°C with rapid shaking. Digests 2–5 were pooled, plated into 100-mm dishes, allowed to come to confluence, trypsinized, and cryopreserved at a concentration of 2 >106 cells/ml. For use in the co-culture system, osteoblasts were plated at an initial density of 2.5 >104 cells/cm2.

Murine bone marrow was obtained for co-culture from adult, 4- to 6-week-old CD1 mice by dissection of the tibias, removal of the epiphyses, and flushing the marrow cavity with a 27-gauge needle. Marrow cells were collected in α-MEM. Cells were sedimented, resuspended in the same medium containing 10% FCS, counted, and plated at an initial density of 1.5 >105 cells/cm2.

OCLs were prepared by seeding primary osteoblasts at an initial density of 2.5 >104 cells/cm2 2 h before the addition of marrow. Marrow cells (at an initial density of 1.5 >105 cells/cm2) were added, and the mixture was co-cultured for 6–7 days in the presence of 10−8 M 1,25-dihydroxyvitamin D3 and 10−5 M prostaglandin E2 with a medium change (α-MEM with 10% FCS, 1% penicillin/streptomycin, 1% l-glutamine, 20 mM HEPES, pH 7.36) every other day. After 6–7 days, contaminating mononuclear cells were removed by treatment with 0.1% EDTA in PBS buffer at 37°C for 5 minutes. This results in a preparation with ∼90% of the cellular material comprised solely of OCLs. OCLs possess the important phenotypic and biochemical markers of authentic osteoclasts. They stain strongly for TRACP, express calcitonin receptors, c-src, and cathepsin k, and most importantly, can resorb bone.(20)

Isolation of RNA

Total RNA (4.5 mg) was isolated from OCLs using the TRIzol reagent (Sigma-Aldrich, St Louis, MO, USA).(21) Poly A+ RNA was separated from total RNA by using oligo(dT)-cellulose spin columns (Invitrogen, Carlsbad, CA, USA).

Construction of a cDNA library for the yeast two-hybrid screen

Poly A+ RNA from OCLs was used to prepare a cDNA library using a commercially available kit and the manufacturer's recommended protocol (SMART; BD Biosciences, Clontech, Palo Alto, CA, USA). The OCLs library used for yeast two-hybrid screening was constructed by co-transforming AH109 yeast cells with OCLs double-stranded cDNA (prepared using the SMART system) and the pGADT7-Rec expression vector (linearized with SmaI). The pGADT7-Rec is the prey vector for the MATCHMAKER GAL4 Two-hybrid System 3 (BD Biosciences Clonetech). OCLs cDNA recombines with pGADT7-Rec in vivo to form fusion proteins with the activation domain (AD) of the GAL4 transcription factor. Of the 8.6 >105 independent transformants generated with 3 μg of pGADT7-Rec vector, >70% contained inserts. The average size of the insert was 1.5 kb.

Yeast two-hybrid screen

Construction of the murine Rac1 bait plasmid

Oligonucleotides were designed to RT-PCR amplify the wildtype full-length murine Rac1 (192 amino acids), using OCL Poly A+ RNA as a template. The forward primer (5′-CCATATGCAGGCCATCAAGTGTGTG-3′) contains a synthetic NdeI site, indicated in boldface type, and the reverse primer (5′-CGGATCCACATTTACAACAGCAGGCAT-3′) contains a synthetic BamHI site, also indicated in boldface type. The 0.6-kbp product was cloned into the pCR2.1 vector (TA cloning kit; Invitrogen, Carlsbad, CA, USA) and sequenced by dyedeoxy-terminator sequencing (ABI-Prism; Perkin-Elmer, Wellesley, MA, USA). The fragment NdeI-BamHI from the pCR2.1 vector encompassing the murine Rac1 coding sequence was cloned in frame downstream between the NdeI and BamHI sites of the GAL4 DNA-binding domain contained in the yeast two-hybrid bait vector pGBKT7. We also used a constitutively active isoform of Rac1, Rac1 Q61L, as bait. Because the Rac1 Q61L-GAL4 DNA-binding domain fusion protein was toxic to AH109 and Y187 yeast strains, murine Rac1 Q61L was cloned in frame downstream of the GAL4 DNA-binding domain (GAL4BD) in the pGBT9 bait vector. This vector produces lower amounts of the fusion protein (BD Biosciences Clonetech).(22) It is possible that the lower level of fusion protein expression somewhat reduced the sensitivity of the yeast two-hybrid screen.

Yeast two-hybrid screen

The OCL library and Rac1 bait plasmids were transformed into the MATα mating type yeast strain AH109 and MATα strain Y187, respectively. AH109 contains three reporters, adenine2 (ade2), histidine3 (his3), and MEL1/LacZ, under the control of distinct GAL4 upstream activating sequences (UAS) and TATA boxes. Y187 contains the LacZ reporter gene under control of the GAL1 UAS. The library screen was performed by mating AH109 harboring the library with Y187 harboring the Rac1 WT or Q61L bait plasmid following the manufacturer's recommended protocol. The mating culture was incubated overnight (20–24 h) in a liquid culture of YPDA medium. After mating, the culture was plated on 50 individual 150-mm SD minimal medium plates lacking tryptophan (trp), leucine (leu), and histidine (his). As a second approach to the yeast two-hybrid screen, the yeast strain AH109 was co-transformed with OCLs double-stranded cDNA, pGADT7-Rec, and pGBKT7-Rac1 WT bait.

Confirmation of protein–protein interactions identified in the initial screen

Clones identified in the initial screen as coding for Rac1-interacting proteins were tested for target specificity by co-transformation as follows. The yeast strain AH109 was co-transformed with full-length Rac1 linked to GAL4BD and each of the plasmids for the possible positive clones fused to GAL4AD, and plated on selective medium lacking trp, leu, and his, but containing X-α-Gal. Plasmid pairs encoding pGBKT7–53 /pGADT7-T and pGBKT7–lam/pGADT7-T were used as positive and negative protein–protein interaction controls, respectively. The specificity of the interactions observed in the initial two screens was verified using the method of Vojtek et al.(23) and Hollenberg et al.(24) Sequential transformation of the bait (Rac1) and each of the possible positive clones was undertaken on double dropout SD/leu/trp media, followed by selection of three colonies for each set. All the possible positive colonies plus the positive and negative control pairs were replated on the same double dropout media (SD/leu/trp) and allowed to grow for 2–3 days. Colonies that grew under these conditions were replated sequentially onto triple dropout (SD/leu/trp/ade) followed by quadruple dropout (SD/leu/trp/ade/his) media. After 2–4 days of culture on the quadruple drop-out media, the negative control did not grow, whereas the positive control grew well, confirming the validity of the initial two yeast two-hybrid screens (see Fig. 1). One of the four clones encoding a possible Rac1- interacting protein did not grow in the confirmation screen and was therefore not pursued further. Three of the initial four colonies grew under these conditions, one of which is described in this report.


RT-PCR assays were performed in the exponential phase of amplification on an PTC-100 Thermal Cycler (MJ Research, Waltham, MA, USA) using SuperScript One-step RT-PCR with the Platinum Taq System (Invitrogen, Carlsbad, CA, USA) and following the manufacturer's recommended protocol.

Primers used were as follows: mouse BCA3 (forward, 5′-GAGACATGGAATACTGTCT-3′; reverse, 5′-TAGTGACGGTCAACATCAC-3′); mouse Rac1 (forward, 5′-ATGCAGGCCATCAAGTGT-3′; reverse, 5′-ATTTACAACAGCAGGCATTT-3′). The primers amplify DNA fragments of 653 and 589 bp in size, respectively.

In vitro binding assay

The plasmid for BCA3 was transcribed and translated in vitro by TNT quick-coupled transcription/translation system (Promega, Madison, WI, USA) in the presence of 35S-methionine to generate labeled BCA3 protein. The Escherichia coli strain BL21 was transformed with pGEX-Rac1 Q61L and grown overnight at 29°C. Overnight cultures were diluted 1:15 into fresh LB medium and grown for 1–3 h at 29°C (such that the final culture OD595 was between 0.3 and 0.6). Protein expression was induced by adding isopropyl-β-d-thiogalactoside (IPTG) at a final concentration of 0.5 mM and growing the bacteria for an additional 4–5 h. Bacterial lysates were prepared in the presence of 5 mM GTPγS and either GST or GST-Rac1 Q61L fusion proteins. Fusion proteins were batch-purified using glutathione-Sepharose 4B (Amersham Pharmacia Biotech, Piscataway, NJ, USA) as described in the manufacturer's instructions. For the binding assay, GST and GST-Rac1 Q61L fusion proteins were left attached to the glutathione-Sepharose 4B beads. Sepharose beads to which were bound either GST or GST Rac1 Q61L were incubated with 10 μl of the 35S-labeled BCA3 in a final of 300 μl volume of binding HEPES buffer (20 mM HEPES-KOH, pH 7.9, 2.5 mM MgCl2, 50 mM KCl,10% glycerol, 1 mM DTT, 0.5% NP40, 1.5% goat serum, and a cocktail of protease inhibitors; Sigma-Aldrich cat no. P8340) for 1–2 h. Beads were pelleted and washed three times with 1 ml of PBS. Bound proteins were eluted with boiling and electrophoresed on a 12% SDS gel. The gel was soaked in Amplify Solution (Amersham Pharmacia Biotech) with gentle rocking for 30 minutes before drying. Exposure with an enhancing screen was performed with Kodak X-Omat AR film at −80°C for 16–48 h.

Interaction of Rac1 Q61L with BCA3 in vivo

MC3T3 cells were transiently transfected with 6 μg of pcDNA4-Myc-tagged-Rac1 Q61L and/or 6 μg of pcDNA4-HA-tagged BCA3 or an empty pcDNA4 vector using Lipofectamine 2000 (Invitrogen). Twenty-four and 36 h after transfection, cells were washed twice with PBS and lysed in HTNG lysis buffer (50 mM HEPES, pH 7.5; 150 mM NaCl; 1% Triton X-100; 10% glycerol, 1.5 mM MgCl2; 1 mM EGTA) containing protease and phosphatase inhibitors (1 mM PMSF, 1 μg/ml pepstatin, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 1 mM sodium vanadate, and 500 mM sodium fluoride). A total of 1000 μg of protein lysate was precleared with either Protein A–agarose or Protein G–agarose beads and subsequently immunoprecipitated overnight at 4°C with 2 μg of anti-HA antibody (Upstate, Lake Placid, NY, USA) either noncovalently bound or conjugated to protein A-agarose beads. The next morning the beads were washed three times with HTNG wash buffer (20 mM HEPES, pH 7.5; 150 mM NaCl; 0.1% Triton X-100; 10% glycerol) and boiled in 2>Laemmli sample buffer before electrophoresis. For Western blotting anti-Myc (Upstate) and anti-mouse horseradish peroxidase (HRP)-conjugated IgG were used as first and secondary antibodies, respectively.

PBD pull-down assay

HEK293Tcells were cultured in DMEM with 10% FBS overnight in 30-mm dishes. Flag-BCA3 or empty Flag-vector constructs were transfected into the cells using Lipofectamine 2000 and following the manufacturer's instructions. The PBD (p21-binding domain of p21-activated kinase 1) binding assay was carried out using a commercially available kit (StressGen Rac1 Activation Kit; StressGen, Victoria, BC, Canada, EKS-450) following the manufacturers recommended protocol. The PBD binding assay takes advantage of the fact that GTP-bound Rac1 but not GDB-bound Rac1 binds to the PBD domain of Pak1 (p21-activated kinase 1). Briefly, the cells were lysed 48 h after transfection, and 500 μg lysates was treated with either GTPγS or GDP for 15 minutes at 30°C. The treated lysates were poured onto PBD-GST bound to glutathione-filters and gently rocked for 1 h at 4°C. The solutions were filtered, and the retained complexes were washed and eluted by boiling with SDS loading buffer. Samples were run on a 15% SDS-polyacrylamide gel, and the membranes were blotted for Rac1 using the antibody supplied with the kit and for Flag using a Sigma M2 monoclonal Flag antibody (Sigma-Aldrich). Twenty micrograms of whole cell lysates was run on SDS gels separately and blotted for Flag and Rac1 expression.

Co-localization of Rac and BCA3 in freshly isolated mature osteoclasts

To analyze the subcellular localization of BCA3 and Rac1, osteoclasts were isolated directly from the long bones of 1-day-old rats by mechanically disaggregating cells from minced bone and allowing them to settle on glass coverslips as described previously.(7,12) Osteoclasts were treated with either vehicle or 2.5 nM CSF-1 for 30 minutes and fixed in 4% paraformaldehyde/PBS for 30 minutes at room temperature. The fixed cells were washed three times in PBS buffer and incubated at room temperature in PBS containing 10% goat serum for 2 h. After this, cells were incubated with a 1:100 dilution of a mouse anti-BCA3 rabbit polyclonal anti-serum,(14) a 1:100 dilution of anti-Rac1 mouse monoclonal antibody (Upstate), or both antibodies in 10% goat serum for 3 h. After this, cells were incubated with a 1:100 dilution of either FITC-conjugated goat anti-rabbit IgG or rhodamine-coupled goat anti-mouse IgG (Molecular Probes, Eugene, OR, USA) or both in 10% goat serum for 2 h. Slides were mounted in VectorShield mounting medium (Vector Laboratories, Burlingame, CA, USA) and examined under a Leitz Laborlux S fluorescence microscope.

Cell-spreading assay

RAW264.7 cells were cultured to 60% confluence in α-MEM containing 10% FBS on gridded glass coverslips placed in 6-well plates. Cells were transiently transfected with cDNA constructs using the pCDNA4 expression vector (Invitrogen). Cells were transfected with the following cDNAs: (1) GFP (pEGFP-N2; BD Biosciences Clontech catalog 6081–1), (2) GFP plus the full-length cDNA for BCA3 in a ratio of 1:5, or (3) a GFP/BCA3 fusion construct in which PEGFP-N2 was fused in frame with the full-length BCA3 cDNA. Cells were transfected using Lipofectomine 2000 (Invitrogen) and the manufacturers recommended protocol. A total of 3 μg of DNA and 6 μl of Lipofectomine 2000 were used per well. For cells co-transfected with separate vectors, the ratio of the two cDNAs was 0.5 μg of the GFP vector to 2.5 μg of the BCA3-containing vector or empty vector. These DNA ratios were chosen to ensure that any cell expressing GFP had an 80% chance of being transfected with either BCA3 or empty vector. If <0.5 μg of GFP cDNA was used in the transfection, the level of GFP protein expression was too low to easily detect by microscopy. For cells transfected with the GFP/BCA3 fusion construct, 3 μg of DNA was used. Cells were incubated with Lipofectoamine 2000 and DNA for 6 h, at which time FCS was added to a final concentration of 10%. Fifteen hours later, media were changed to fresh α-MEM containing 10% FCS. After an additional 15 h (36 h after transfection), GFP-positive cells were analyzed for cell spreading by exposing them to 2.5 nM CSF-1 while being serially photographed using a CCD camera as previously described.(7) The change in cell area was quantified by measuring the change in cell area as a percentage of the baseline area using NIH Image software version 1.62f as reported previously.(7)


Identification of BCA3 as a novel Rac1-binding protein

When the OCL cDNA library was screened using wildtype Rac1 as bait, five clones were identified that grew on triple drop out media in both the co-transformation and mating assay screens. Sequencing showed that all five clones encoded different proteins. One of these clones proved to be a ribosomal protein that is commonly identified as a false positive in yeast two-hybrid assays(25) and therefore was not pursued further. When the OCL cDNA library was screened using Rac1 (Q61L) as a bait, 56 primary transformants were isolated on triple drop out media. The 4 positive clones from the Rac1 screen and the 56 positive clones obtained using Rac1 (Q61L) as bait were further analyzed for growth on quadruple drop out media (SD/leu/trp/ade/his). Three of the four positive clones from the Rac1 screen grew under these conditions, but only three clones grew from the Rac1 (Q61L) screen. These three clones all encoded the same protein. This latter protein proved to be identical with one of the four clones (clone 2) identified in the Rac1 screen (Fig. 1). The identity of clone 2 with the three clones identified in the Rac1 (Q61L) screen was confirmed by direct plasmid DNA sequencing.

Figure Figure 1.

Co-transformation assay using the four positive clones from the initial yeast two-hybrid screen. (A) Double-drop out media. (B) Triple-drop out media. (C) Quadruple drop-out media. Clone 1 did not grow under all conditions and was considered a false positive. +, positive control; N, negative control. Each numbered row represents a different clone. Clone 2 is BCA3.

Clone 2 from the Rac1 screen encompassed a 631-bp open reading frame that encoded amino acids 76–212 of the murine homolog of human BCA3 (GenBank accession no. AK159903), also known as AKIP1(14–16) or KBP1 (GenBank accession no. AF493783).(17–19) The three clones from the Rac1 (Q61L) screen all encoded the first 125 N-terminal amino acids of BCA3. As shown in Fig. 2, the BCA3 transcript is clearly expressed in osteoclasts.

Interaction of Rac1 and BCA3 in vitro

To determine if the interaction between Rac1 and BCA3 detected in the yeast two-hybrid screens was direct, in vitro transcription and glutathione S-transferase (GST) pull-down assays were used. Fragments of BCA3 (Fig. 3A) encompassing amino acids 1–125 (fragments 1 + 2), amino acids 76–212 (fragments 2 + 3), and amino acids 125–212 (fragment 3) were transcribed in vitro and labeled with 35S methionine as described in the Materials and Methods section. Efforts to transcribe fragment 1 (amino acids 1–75) and fragment 2 (amino acids 76–125) in vitro were unsuccessful because of low yield despite repeated attempts. [35S]Labeled full-length BCA3 evidenced strong binding to Rac1–GST in a pull down assay (Fig. 3B). We also found that fragment 2 + 3 (amino acids 76–212) directly bound Rac1 in vitro (Fig. 3B). In contrast, fragment 3 could not bind BCA3 (Fig. 3C). Although we could not express fragments 1 or 2 alone in sufficient amounts for pull down assays, the fact that 1 + 2 and 2 + 3 but not fragment 3 could bind BCA3 suggests that the critical region of BCA3 required for interacting with Rac1 is contained between amino acids 76–125.

Interaction of Rac1 with BCA3 in mammalian cells

To verify that the Rac1–BCA3 interaction also occurs in vivo in mammalian cells, we expressed HA-tagged BCA3 together with Myc-tagged Rac1 (Q61L), HA-tagged BCA3 alone, or Myc-tagged Rac1 (Q61L) alone in MC3T3 cells. MC3T3E1 cells were chosen for these experiments because OCLs cannot be transiently transfected. We used the constitutively active form of Rac for these experiments because MC3T3E1 cells do not express the receptor for CSF-1 (c-fms), and therefore if CSF-1 activation of Rac was required for this interaction, we would not be able to detect it. As shown in Fig. 4A and consistent with our in vitro binding data, HA-BCA3 co-immunoprecipitated with the constitutively active form of Rac1.

Because in the above experiment a constitutively active form of Rac1 was used, additional experiments were undertaken to ascertain if the GTP-bound, active form of Rac preferentially bound to BCA3. A PBD affinity assay was performed because GTP-Rac but not GDP-Rac is known to bind to the PBD sequence from Pak1. HEK293T cells were transfected with Flag-BCA3 or empty Flag-vector, and the cell lysates were subjected to the PBD assay. As shown in Fig. 4B (lane 1, top panel), good expression of Flag-BCA3 was achieved in cells transfected with Flag-BCA3. There was no Flag detected in cells transfected with empty vector (lane 2, top panel). The level of expression of endogenous Rac1 was unaffected by transfection with Flag-BCA3 (lanes 1 and 2, bottom panels). As shown in lanes 3 and 5, a Rac1-specific band was observed in the PBD complex in GTPγS-treated but not in the GDP-treated lysates (lanes 4 and 6), indicating that, as expected, activated Rac1 protein was being pulled down with PBD. In addition, BCA3 co-purified with active Rac1 in this complex indicating that the BCA3 protein interacts with GTP bound Rac1 (lane 3). In contrast, no band was observed in the empty Flag-vector lysates (lane 5).

Co-localization of BCA3 with Rac1 in osteoclasts after CFS-1 treatment

Because the yeast two-hybrid screen was undertaken using osteoclasts, we sought to determine if BCA3 and Rac1 interact at endogenous levels of protein expression in osteoclasts. Therefore, studies were performed using confocal microscopy to determine if the two proteins co-localize in these cells. As expected based on its known role in regulating the actin cytoskeleton, in resting osteoclasts, Rac1 was localized at the plasma membrane and also showed a meshwork-like pattern of staining in the cytosol in osteoclasts (Fig. 5, top panel, left column). BCA3 is thought to be primarily localized to the nucleus. Consistent with this, BCA3 staining was seen almost exclusively in the nucleus in resting osteoclasts (Fig. 5, top panel, middle column). There was no co-localization of the two proteins in the absence of CSF-1 treatment (Fig. 5, top panel, right column). Interestingly, in 34% of cells examined, treatment with CSF-1 induced co-localization of BCA3 with Rac1 in a distinct perinuclear pattern (Fig. 5, bottom panel, right column). This pattern was observed in only 4% of untreated cells.

Overexpression of BCA3 attenuates CSF-1–induced cell spreading

To determine if BCA3 alters the ability of Rac1 to function as a downstream effector of CSF-1–induced cell spreading, transient expression experiments were undertaken using RAW264.7 cells. Transient transfection experiments in authentic osteoclasts are not possible because of their short half-life (12–20 h). The RAW cells were chosen for these studies because they are CSF-responsive and show cytoplasmic spreading in response to the growth factor. As show in Fig. 6, overexpression of BCA3 caused a marked attenuation of CSF-1–induced spreading. Thus, cells co-transfected with GFP plus the empty expression vector showed a 27 ± 8% mean increase in cell area, whereas cells co-transfected with the GFP and BCA3 expression vectors showed only an 8.1 ± 2.7% increase in cell area. For reasons that are not clear, the GFP-BCA3 construct did not express well, and only eight cells could be identified in which GFP expression was intense enough to identify these cells for subsequent spreading assays. In these cells, there was no spreading observed, with a mean increase in cell area of 0.0 ± 3.2%. By two-way ANOVA, there is an overall effect of construct transfected (p < 0.05). Posthoc Bonferroni testing revealed that there were significant differences in the mean spread area when cells co-transfected with GFP and vector were compared with cells co-transfected with GFP plus BCA3 (p < 0.05). Thus, expression of BCA3 significantly attenuated the ability of CFS-1 to induce spreading in these cells. The difference in results with the fusion protein versus co-transfection of the GFP and BCA3 constructs is likely caused by the stoichiometry of the co-transfection assay. Given a DNA ratio in the transfection assay of GFP:BCA3 of 1:5, one would expect 20% of the GFP-expressing cells not to be co-transfected with the BCA3 vector. Thus, some cells expressing GFP in the experiments where both GFP and BCA3 were expressed on separate vectors would be expected to behave like cells expressing GFP plus empty vector.

Figure Figure 2.

Expression of (A) Rac1 and (B) BCA3 in osteoclast-like cells (Oc), as determined by RT-PCR. Neg refers to a PCR reaction with no template RNA. The amplicons are of the expected size in both cases. MWM, molecular weight markers.

Figure Figure 3.

Direct interaction between murine BCA3 and Rac1 in vitro. (A) BCA3 deletion constructs used in the in vitro association experiments. Numbers refer to amino acids in murine BCA3; fragment 1: amino acids 1–75; fragment 2: amino acids 76–125; fragment 3: amino acids 124–212. (B) Autoradiograph showing binding of full-length BCA3 to GTP-loaded Rac1 (lane 3) but not to GST alone (lane 2). “Input” in this and subsequent panels refers to radiolabeled BCA3 or BCA3 fragment. (C) Autoradiograph showing binding of fragments 1 + 2 of BCA3 to GTP-loaded Rac1 (lane 3) but not to GST alone (lane 2). (D) Autoradiograph showing binding of fragments 2 + 3 of BCA3 to GTP-loaded Rac1 (lane 3) but not to GST alone (lane 2). (E) Failure of BCA3 fragment 3 to bind Rac1. Fragment 1 alone and fragment 2 alone translated very poorly in vitro and could not be used in binding experiments. Input signals in B are from an autoradiogram exposed for 1 h. GST pull-down data are from an autoradiogram exposed for 20 h. Both input and pull-down assay BCA proteins migrated at the same molecular weight for the 1 + 2 and 2 + 3 constructs.

Figure Figure 4.

BCA3 and actived Rac1 interact in vivo. (A) One milligram of whole cell lysate was prepared from MC3T3E1 cells transfected with Myc-tagged constitutively active Rac1 (Q61L) alone (lane 2), HA-tagged BCA3 alone (lane 3), or both constructs (lane 1). HA was immunoprecipitated from each lysate and Western blotted for Myc. A band at 25 kDa corresponding to Myc-tagged Rac1 is seen only in lane 1. (B) Flag-BCA3 or Flag empty vector (EV) was transfected into HEK293T cells as indicated, and the PBD binding assay was performed per manufacturer's instructions. The blot was probed with an anti-Rac1 antibody and with an anti-Flag antibody. BCA3 co-purifies with GTP-bound Rac1 but not GDP-bound Rac1. WCL, whole cell lysates.

Figure Figure 5.

Total and cell surface protein expression levels of mutant receptors as determined by ELISA in transfected Cos-7 cells. All readings were corrected for background observed with untransfected cells. Absorbance was read at 360 nm. (A) Total expression levels. The cells were permeabilized before the addition of primary antibody. (B) Cell surface expression levels. The cells were permeabilized after the addition of primary antibody. Data are expressed as the mean ± SE (n = 3). *p value between 0.01 and 0.05; **p value between 0.001 and 0.01; and ***p < 0.001 vs. wtPTHR1 expression levels.

Figure Figure 6.

Overexpression of BCA3 attenuates CSF-1–induced RAW cell spreading. RAW264.7 cells were transiently co-transfected with two expression constructs: one containing a cDNA encoding GFP and a second containing either full-length BCA3 or vector alone. In a third experimental group, cells were transfected with an expression vector in which GFP was fused in frame to full-length BCA3. Thirty-six hours later, cells were treated with 2.5 nM of CSF-1, and cell area was quantified before and after exposure to the growth factor. Shown is mean ± SE percent increase in cell area. There was a significant overall effect of treatment by two-way ANOVA. Post hoc testing showed that transfection with BCA3 significantly attenuated CSF-1–induced cell spreading when comparing the GFP plus vector group to the GFP plus BCA3 group (n = 72, 77, and 8, respectively, for cells transfected with GFP plus empty vector, GFP plus BCA3, and the GFP/BCA3 fusion protein). Because there were fewer cells in the GFP/BCA3 treatment group, posthoc testing was not significant for this group compared with the GFP plus empty vector group. Although the number of cells was low, none of them spread in response to treatment with CSF-1. *Significant treatment effect by two-way ANOVA. † Significantly different from cells transfected with GFP plus empty vector.


We identified BCA3 as a Rac1 binding protein in a yeast two-hybrid screen. We have confirmed this interaction in vitro and in vivo at endogenous levels of protein expression. This interaction is direct and seems to require the midportion of BCA3 (aa 76–125). In mammalian cells, this interaction is facilitated by activation of Rac to its GTP bound state. An interaction between BCA3 and any member of the Ras family of small GTPases has not been previously reported.(26)

In 2002, the genomic sequence for BCA3 was identified as open reading frame 17 on chromosome 11.(27) In 2003, Kitching et al.(14) cloned and characterized the cDNA for the same gene. They identified two putative splice variants in which exon 2 or both exons 2 and 5 are spliced out and also predicted several putative functional domains. They reported homologous sequences in the mouse and zebrafish genomes. A tissue survey identified high levels of the transcript in heart with low levels in other tissues including breast. However, expression levels were high in breast and prostate cancer cell lines. Immunohistochemical studies confirmed high levels of protein expression in breast cancer tissue, and the authors consequently termed the protein BCA3.(14) A bioinformatics analysis of BCA reported by Leon and Canaves(15) suggested that BCA3 would be a nuclear protein. They identified a potential phosphorylation-regulated nuclear localization sequence in the protein located at amino acids 15–21 in the amino terminus of the molecule.(15) Li et al.(17) and Qin et al.(18,19) reported that this protein bound to the LIM protein KyoT2. KyoT2 interacts with the DNA binding protein, RBP-J.(18) They termed the KyoT2 binding-protein, KBP1 for KyoT2 binding protein-1. This same group subsequently reported BCA3/KBP1 binds to the LIM domain of KyoT2.(18)

Sastri et al.,(16) using a yeast two-hybrid screen, identified BCA3 as interacting with the amino terminus of the C subunit of protein kinase A. They termed the protein AKIP1. These investigators confirmed high-level protein expression in human heart and a human breast cancer cell line. They identified a nuclear localization signal in the protein and confirmed its nuclear localization by immunocytochemistry. Using deletional and overexpression analyses, they provided evidence that BCA3/AKIP1 acts to shuttle activated subunit C to the nucleus. Recently, Gao et al.(28) reported that BCA3 is a NEDD8 substrate. Neddylaton refers to a process whereby the protein neural precursor cell-expressed and developmentally downregulated gene (NEDD8) is covalently linked to proteins in a process similar to ubiquitination. Neddylation of BCA3 leads to suppression of NF-κB–dependent transcription by enabling binding of neddylated BCA3 to the NF-κB transcription factor p65 and to the cyclin D1 promoter.

Our findings identify a fourth BCA3-interacting protein. As noted, there are a number of putative functional domains in BCA3 that suggest that this protein may indeed have multiple cellular roles. These include a nuclear localization signal, a proline-rich sequence, five putative SH2 binding motifs, and a PZD binding motif at the carboxy terminus of the molecule. The murine BCA sequence does not contain any sequences similar to the CRIB domain, which is a known Rac binding motif found in several Rac-interacting proteins,(29) nor are there any sequences similar to the PBD binding domain of Pak1, which selectively binds activated Rac1.(30) Plexin-B1 is a Rac binding protein identified in yeast two-hybrid screen that does not contain either CRIB or PBD domains.(31) The fact that activation of Rac facilitates its interaction with BCA3 suggests that the 3D conformational change induced by GTP on loading contributes to the creation of this binding domain in Rac1.

The finding of a Rac1–BCA3 interaction is surprising because, as noted, BCA3 has been identified as a nuclear protein and Rac is generally considered to act primarily in the cytoplasm to regulate actin remodeling. Our localization data suggest a putative shuttle function of BCA3. In quiescent osteoclasts, BCA3 is localized to the nucleus where it stains in large patches or speckles. There is no co-localization of Rac1 and BCA3. After activation of the CSF-1 receptor, c-fms, and consequent activation of Rac1, there is a redistribution of BCA3 to a perinuclear location where it co-localizes with Rac1. These data suggest that BCA3 may be accumulating in the cytoplasm for a function related to the actions of activated Rac. Our expression data support such a conclusion. Overexpression of BCA3 markedly attenuated CSF-1–induced cell spreading. One possibility is that BCA3 acts to sequester activated Rac1 in a perinuclear location, away from the spreading edge of the cell, and by doing so provides a cellular brake on Rac1-induced cytoskeletal remodeling. Our data indicate that BCA3 shuttles in and out of the nucleus in response to CSF-1. There is precedent for cytoplasmic shuttling of nuclear proteins that regulate Rho GTPase activity. As an example, it has been reported that the cyclin-dependent kinase inhibitor, p27kip1, which is localized in the nucleus of nontransformed cells, is exported to the cytoplasm after phosphorylation on Ser10 by human kinase interacting stathmin.(32–35) Cytoplasmic p27Kip1 binds to and inhibits Rho kinase, thereby disrupting RhoA-dependent stress fiber formation.(36) Consistent with this model, hepatocyte growth factor–induced cell scattering is associated with phosphorylation of Ser10 on p27kip1 and the export of p27kip1 to the cytoplasm.(37) CSF-1, which like hepatocyte growth factor acts through a receptor tyrosine kinase, also causes translocation of a nuclear protein, in this case BCA3, to the cytoplasm. As just noted, our data suggest that BCA3 may sequester Rac1 in a perinuclear location. Whether BCA3 can also shuttle Rac1 to the nucleus is unclear. However, it is known that Rac can interact with nuclear proteins. Thus, GTP-bound Rac1 has been shown directly interact with the transcription factors TCF-4 and β-catenin in the nucleus.(38) There is also precedent for nuclear sequestration of proteins that regulate the actin cytoskeleton. For example Far1p, the yeast homolog of p27kip1 and a cell cycle inhibitor, sequesters the Cdc42 GEF, Cdc24p in the nucleus. Activation of the pheromone response pathway induces phosphorylation of Far1p, with subsequent translocation of the Far1p–Cdc24p complex to the cytoplasm.(39)

Because Rac is critically important for lamellipodia formation and cell motility, the role of BCA3 in modulating these effects is an obvious area for future studies, which should improve our understanding of the role of both proteins.


This work was supported by NIH Grants DE12459, DK45228, and P30 AR46032 (to KLI) and by support from CIHR-UI and the Canadian Breast Cancer Research Alliance Special Program Grant on Metastasis (to AS).