Introduction
The skeleton is a favored site for breast cancer metastases due to unique features of the bone microenvironment, including the presence of growth factors and cytokines stored within the bone matrix [
1]. The emergence of bone metastases disrupts normal bone homeostasis by perturbing interactions between bone-forming osteoblasts and bone-resorbing osteoclasts [
2]. Breast cancer metastases in bone have typically been described as osteolytic in nature, and are associated with excessive bone destruction [
3]. This ultimate shift toward bone resorption results from the ability of tumor cells, either directly or indirectly, to influence osteoclast differentiation and activity positively [
4,
5]. Subsequently, elevated bone resorption releases latent growth factors and cytokines that are stored in the bone matrix; these support tumor cell survival and growth that ultimately lead to further bone destruction [
6,
7]. Hence, the crosstalk between breast cancer cells and the bone microenvironment results in a vicious cycle of bone destruction and increased tumor growth in bone.
Breast tumors are heterogeneous, and cancer cells with bone-specific metastatic capabilities may preexist in the primary tumor. Indeed, gene signatures have been generated that predict whether a primary breast tumor will relapse to bone or visceral sites of metastasis [
8]. A Src-related signature has also been proposed to segregate primary breast tumors based on their propensity to relapse to bone [
9]. Numerous studies have identified cancer intrinsic factors that allow tumor cells to colonize and thrive in the bone microenvironment.
In vivo selected breast cancer populations, isolated from bone metastases, have been used to identify unique functional mediators of bone metastasis [
10‐
13].
These approaches have yielded valuable information regarding mechanisms involved in the spread, colonization, and growth of breast cancer cells in bone. However, growing evidence reveals discordance between the expression of specific markers in the primary breast tumor and those in the corresponding bone metastases [
14]. Up to 40% of breast cancer patients displayed discordance in hormone-receptor expression between the primary tumor and the associated bone metastases [
15,
16]. Thus, it is likely that the bone microenvironment plays a considerable role in modulating the gene-expression profiles of breast cancer cells in emerging bone metastases. Hence, a number of important mediators of breast cancer skeletal metastasis will undoubtedly be overlooked in the analysis of primary breast tumors or breast cancer cells explanted
ex vivo from bone metastases.
To circumvent these limitations, we sought novel mediators of skeletal metastasis directly in bone metastatic lesions from breast cancer patients. We applied laser-capture microdissection to isolate RNA from both trephine-biopsies of bone metastases and primary breast tumors. Numerous genes were differentially expressed between primary breast tumors that later relapsed to bone and breast cancer bone metastases, including several members of the ATP-binding cassette (ABC) transporter family that were overexpressed in the bone metastases relative to primary tumors. ABCC5 was found to be functionally involved in the formation of breast cancer bone metastases in two independent cell-based models.
Materials and methods
Primary breast tumor and bone metastases
Unguided or computed tomography (CT)-guided trephine biopsies were performed on breast cancer patients with known bone involvement at the Princess Margaret Hospital (Toronto, ONT, Canada), as previously described [
17]. Biopsy material was immediately flash frozen and embedded in OCT compound. All procedures were performed with approval from the Research Ethics Board at the Princess Margaret Hospital. Primary breast tumor material was collected from patients who underwent surgery at the Montreal General or Royal Victoria Hospital (Montreal, QUE, Canada). Tumor banking was performed with approval from the Research Ethics Board of the McGill University Health Centre under the protocols SDR-99-780 and SDR-00-966. All patients provided written and informed consent.
Laser-capture microdissection
Histologic sections of primary breast tumors or bone metastases were stained with H&E and examined by a clinical pathologist to identify regions within each section suitable for laser-capture microdissection (LCM). Sections (10 μm) were stained and dehydrated by using a HistoGene LCM Frozen Section Staining Kit (Cat KIT0401; Applied Biosystems, Carlsbad, CA, USA). Clusters of invasive mammary epithelial cells were identified and selected by using an ArcturusXT Microdissection System powered by ArcturusXT software v.1.1 (Applied Biosystems, Carlsbad, CA, USA). Breast cancer cells were captured by using an infrared laser adjusted to a diameter of 20 μm, laser power set to 65 mW and a duration of 20 msec, and pulsed through CapSure HS LCM Caps (Cat. LCM0214; Applied Biosystems, Carlsbad, CA, USA). The beam was passed over the sample to be collected with an overlap of 30% for each specimen. RNA was extracted from the microdissected cells by using a PicoPure RNA Extraction Kit (Cat. KIT0204; Applied Biosystems, Carlsbad, CA, USA) according to the manufacturer's instructions. RNA integrity and quantity was evaluated by using a 2100 Bioanalyzer platform (Agilent Technologies, Santa Clara, CA, USA).
RNA amplification, labeling, and hybridization to Agilent microarray chips
Total RNA (1 to 2.5 ng) from microdissected material was subjected to two rounds of linear amplification by using a RiboAmp HSPlus Amplification Kit (Cat. KIT0525; Applied Biosystems, Carlsbad, CA, USA), following the manufacturer's protocol. Profiles of resulting amplified RNA (aRNA) were assessed by using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). The aRNA samples (5 μg) were conjugated to Cy3 dye by using an Arcturus Turbo Labeling Kit (Cat. KIT0609; Applied Biosystems, Carlsbad, CA, USA). Universal human reference RNA (Stratagene) was amplified by using the same procedure and labeled with Cy5 dye (Cat, KIT0619; Applied Biosystems, Carlsbad, CA, USA). RNA concentration and dye incorporation was measured by using a Nanodrop ND-1000 UV-VIS spectrophotometer. Labeled RNA (0.825 μg) was then hybridized to 44K whole human genome microarray gene-expression chips (Cat. G4112F; Agilent Technologies, Santa Clara, CA, USA) by using a Gene Expression Hybridization Kit (Cat. 5188-5242; Agilent Technologies, Santa Clara, CA, USA) at 65°C for 17 hours according to the manufacturer's instructions. Microarray chips were washed, dried, and immediately scanned on a Microarray Scanner Model G2505B (Agilent Technologies, Santa Clara, CA, USA) by using Agilent Scanner Control Software vA7.0.1 (Agilent Technologies, Santa Clara, CA, USA).
Gene-expression analysis
Microarray data were extracted by using Feature Extraction Software v. 9.5.3.1 (Agilent Technologies, Santa Clara, CA, USA). The raw data were then normalized, and differential expression was performed by using the LIMMA package in R/bioconductor [
18]. Specifically, the arrays were normalized by using normexp background correction, loess within array, and quantile between array normalization. The
P values for differential expression were adjusted for multiple testing. Candidate gene lists were generated by filtering the data on the basis of more than twofold difference in expression between bone metastases and primary breast tumors. The microarray data can be accessed through the GEO repository (ID GSE39494) [
19].
Real-time quantitative reverse-transcription polymerase chain reaction
RNA obtained after two rounds of amplification was quantified by using Quant-iT RiboGreen RNA Reagent based on the manufacturer's protocol (Cat. R11491; Invitrogen, Grand Island, NY, USA). Total RNA (25 ng) was converted to cDNA by using a Transcriptor Reverse Transcriptase kit (Cat. 048970300001; Roche, Laval, QUE, Canada) in accordance with the manufacturer's protocol. After reverse transcription, samples were subjected to real-time polymerase chain reaction (PCR) analysis by using SYBR Green PCR Master Mix (Cat. 04887352001; Roche, Laval, QUE, Canada). Primers were designed by using OligoPerfect software (Invitrogen, Burlington, ONT, Canada) in the region of the target gene surrounding the Agilent probes, at a concentration of 0.5 μM. PCRs were performed on a LightCycler 480 system (Roche, Laval, QUE, Canada) under the following conditions: preincubation step (95°C for 10 minutes), 45-cycle amplification sequence (95°C for 10 seconds, 53°C for 10 seconds, 95°C for 6 seconds) and a melting step (95°C for 5 seconds, 65°C for 1 minute). A complete list of primer sequences can be found in the Supplementary Information. Results were analyzed with the absolute quantification method by using the second derivative maximum method feature of LightCycler 480 Software v. 1.5.0 SP4 (Roche, Laval, QUE, Canada).
Immunohistochemistry
OCT-embedded primary breast tumors and bone trephine biopsies were sectioned (10 μm) and fixed in 2% paraformaldehyde. The sections were blocked with 2% BSA and 5% normal goat serum (NGS) and subsequently incubated overnight at 4°C with a primary antibody directed against ABCC5 (1:25; clone M5I-10). This monoclonal antibody was generated in Dr. Scheffer's laboratory (Amsterdam, The Netherlands) after injection of a bacterial fusion protein containing the N-terminal region of mouse ABCC5. The antibody recognizes mouse ABCC5 but also reacts strongly with the human orthologue. The sections were then incubated with Biotin-conjugated secondary antibody and developed with 3-3-diaminobenzidine-tetrahydrochloride (DAB). A standard hematoxylin counterstain was performed to demarcate cellular nuclei.
Primary mammary tumors and hindlimbs were excised from mice and fixed overnight in 4% paraformaldehyde. Bones were decalcified in a solution of 14.5% ethylenediaminetetraacetic acid (EDTA) and 15% glycerol for 4 weeks. Tissues were then paraffin embedded and sectioned. Sections (5 μm) were deparaffinized and stained with a freshly prepared tartrate-resistant acid phosphatase (TRAP) staining solution (naphthol AS-TR Phosphate, fast blue RR salt, and sodium tartrate). Slides were scanned by using a Scanscope XT digital slide scanner (Aperio, Vista, CA, USA) and analyzed with Imagescope software (Aperio, Vista, CA, USA). The number of TRAP-positive cells within breast cancer lesions in bone was counted manually and is presented as the number of osteoclasts per square millimeter of tumor mass.
Immunoblotting
Human and mouse cell lines were lysed in TNE lysis buffer, as previously described [
20]. Total protein concentrations were determined with the Bradford Protein Assay (Cat. 500-006; Bio-Rad Laboratories, Mississauga, ONT, Canada) and 20 to 50 μg of protein was separated with sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes (Cat. IPVH00010; Millipore, Billerica, MA, USA). The membranes were blocked in 5% wt/vol nonfat dry milk containing 0.1% Tween and incubated with the following primary antibodies: ABCC5 (1:100 dilution; M5I-10) and α-tubulin (1:20,000 dilution; Cat. T9026; Sigma, Oakville, ONT, Canada). The blots were then incubated with horseradish-peroxidase-conjugated secondary antibodies, and proteins were visualized with an enhanced chemiluminescence detection system (Cat. 34080; Pierce, Nepean, ONT, Canada).
DNA constructs
Short-hairpin RNA (shRNA) sequences targeting the human and mouse
ABCC5 mRNA, as well as the scrambled control sequence, were designed by using the RNAi central website at Cold Spring Harbor Laboratories [
21]. The sequences of the shRNAs used for
ABCC5 knockdown are as follows (target sequence denoted in
bold text: h, human; m, mouse):
hABCC5 sh: TGC TGT TGA CAG TGA GCG
ACC TCA AAG TCT GCA ACT TTA ATA GTG AAG CCA CAG ATG TAT TAA AGT TGC AGA CTT TGA GGG TGC CTA CTG CCT GGA;
mabcc5 sh1: TGC TGT TGA CAG TGA GCG
ACC TCA TCC TGT CCT GCT GAA ATA GTG AAG CCA CAG ATG TAT TTC AGC AGG ACA GGA TGA GGG TGC CTA CTG CCT CGG A;
mabcc5 sh2: TGC TGT TGA CAG TGA GCG
CCC TGA CTA TGG CAT TCA AGA ATA GTG AAG CCA CAG ATG TAT TCT TGA ATG CCA TAG TCA GGA TGC CTA CTG CCT CGG A; scrambled sh: TGC TGT TGA CAG TGA GCG
AAG TCC ATA CTT AGT CGA TAG ATA GTG AAG CCA CAG ATG TAT CTA TCG ACT AAG TAT GGA CTC TGC CTA CTG CCT CGG A. These sequences were PCR amplified, digested, and cloned into the LMP vector as
XhoI/
EcoRI fragments by following published instructions [
22].
Cell culture and in vitro osteoclastogenesis assay
Parental MDA-MB-231 breast cancer cells were obtained from the American Type Culture Collection and transduced with a triple reporter system, as previously described [
23]. The MDA-MB-231-derived bone metastatic (1833-BM1) and lung metastatic (4175-LM2) populations were derived as described previously [
10,
24]. Human breast cancer cell lines were cultured in DMEM supplemented with 10% fetal bovine serum and MEM Nonessential Amino Acids (1X), gentamycin, and amphotericin B. The 4T1 murine mammary carcinoma cell line was obtained from the American Type Culture Collection. Nonmetastatic 67NR and lung-metastatic 66cl4 murine mammary carcinoma cell lines [
25] were kindly provided by Dr. Fred Miller (Barbara Ann Karmanos Cancer Institute, Detroit, MI, USA) and cultured in DMEM supplemented with 10% fetal bovine serum, 10 m
M HEPES, 1 m
M sodium pyruvate, 1.5 g/L sodium bicarbonate, gentamycin, and amphotericin B.
For the
in vitro osteoclastogenesis assay, 5 × 10
5 4T1-derivative cells were plated in 10-cm cell-culture dishes. The following day, media were changed to DMEM supplemented with 10% fetal bovine serum and subsequently conditioned for 48 hours. Protocols used to establish primary osteoclast cultures from BALB/c mice were performed in accordance with the McGill University guidelines established by the Canadian Council on Animal Care. Bone marrow was collected from tibiae and femurs of mice (BALB/c, male, 6 weeks old; Charles River, Wilmington, MA, USA), as described previously [
26], and cultured in 75-cm
2 tissue culture flasks (15 × 10
6 cells per flask) in α-minimal essential medium supplemented with 10% fetal calf serum, penicillin, streptomycin, and 25 ng/ml of human recombinant macrophage-colony stimulating (M-CSF) factor (Cat. 300-25; PeproTech, Inc.). On day 1, nonadherent cells were collected, plated at 5 × 10
3 cells/cm
2, and supplemented with M-CSF (50 ng/ml) and recombinant GST-RANKL (100 ng/ml). On day 4, fresh media with or without RANKL (100 ng/ml) or conditioned media harvested from 4T1-derivatives (10%) were added to the cultures. M-CSF (50 ng/ml) was present in all conditions tested. On day 6, cells were fixed with 4% paraformaldehyde (10 minutes), washed with phosphate-buffered saline, and stained for tartrate-resistant acid phosphatase (TRAP) (Cat. 387A-KT; Sigma, Oakville, ONT, Canada).
Left cardiac ventricle injections
Female SCID/beige and BALB/c mice (4 to 6 weeks old) were purchased from Charles River Laboratories. The animals were housed in facilities managed by the McGill University Animal Resources Centre. All animal experiments were conducted under a McGill University-approved Animal Use Protocol in accordance with guidelines established by the Canadian Council on Animal Care. Breast cancer cells were harvested from subconfluent cultures and resuspended in sterile PBS. Mice were anesthetized with isofluorane, and 1 × 10
5 human or mouse breast cancer cells, in a volume of 100 μl, were injected into the left cardiac ventricle by using 26G needles [
12]. A successful injection was distinguished by the pumping of arterial blood into the syringe during the injection procedure and confirmed by a uniform luminescent signal throughout the entire animal body after 1833-BM1 cell inoculation.
In vivo bioluminescent imaging
Tumor outgrowth within skeletal sites of mice injected with 1833-BM1 breast cancer cells was monitored by using an IVIS 100 (Caliper Life Sciences, Hopkinton, MA, USA) bioluminescence imaging system, as previously described [
27]. The resulting data were normalized to the signal generated by the initial cell inoculum, which was measured immediately after cardiac injection. Total metastatic burden was measured by setting a uniform scale for each group of mice, outlining regions of interest around all luminescence signals in the body and summing them.
X-ray microcomputed tomography (μCT) imaging
At the end of the cardiac-injection experiments (21 days for 1833-BM1 and 13 days for the 4T1 models), mice were anesthetized and immobilized with tape in the imaging tube of a Skyscan 1178 μCT. All images were obtained with an x-ray source operating at 50 kV (1833-BM1) or 45 kV (4T1) and 615 mA, with an exposure time of 480 msec. Animals were rotated through 180 degrees at a rotation step of 1.26 degrees (1833-BM1 cells) and 0.9 degrees (4T1 cells). Cross-section images from tomography projection images were reconstructed by using the NRecon program package v.1.6.4.7 (SkyScan, Kontich, Belgium). Reconstruction parameters, including smoothing (1), ring artefacts reduction (4), and beam-hardening correction (30%), were fixed for all the samples. The dynamic image range was defined between 0 and 0.045 for all the samples. Bone alignment was adjusted in all specimens by using DataViewer v.1.4.3.2 (SkyScan, Kontich, Belgium). Bone volume was determined in 3D by using CTAn software v.1.11.8.0 (SkyScan, Kontich, Belgium). In brief, for each bone, a volume of interest (VOI) was determined starting under the growth plate and extending 20 (femur) and 25 (tibia) sections below the diaphysis. For each model, the VOI was designed by drawing user-defined polygons on the 2D sections that encompass the bone of interest. In the binary image mode, the histogram was set at minimum 100 to maximum 255 for a given dataset for each specimen. Each 3D model was visualized by using CTvox v.2.3 (SkyScan, Kontich, Belgium). The absolute bone volume was determined for each piece of bone and expressed in cubic millimeters. Control groups, including uninjected mice of similar strain and age as the experimental animals, were used as reference for normal bone volume. The degree of bone destruction was determined as percentage difference between the average bone volumes in experimental groups compared with those in the appropriate control cohorts.
Statistical analysis
Statistical significance values for RT-qPCR expression, whole body luminescence, μCT bone volumes, and osteoclast assays were obtained by performing a two-sample variance two-tailed Student t test.
Supplemental Materials and Methods can be found as Additional file
1.
Discussion
The vicious cycle of bone-metastasis formation argues that breast cancer cells colonizing the bone are influenced by growth factors that are stored in the bone matrix; these are released during bone resorption [
1,
2]. These influences predict that breast cancer bone metastases will differ in their gene-expression profiles from primary breast tumors. To evaluate gene-expression profiles of
in situ bone metastases from breast cancer patients, we performed laser-capture microdissection (LCM) on trephine biopsies and compared them with profiles obtained from LCM material of primary breast tumors. Numerous genes were differentially expressed between primary breast tumors with known relapse to bone and breast cancer bone metastases and were categorized according to their known or proposed functions.
Intriguingly, three members of the ATP-binding cassette (ABC) transporter family (
ABCA5, ABCC5, ABCG2) were overexpressed in breast cancer bone metastases compared with primary tumors that are metastatic to bone (Additional file
7 and Additional file
8). The human genome encodes 49 ABC genes that are arranged in seven subfamilies designated A to G [
60]. These genes encode ATP energy-dependent molecular pumps that transport substrates across membranes, either in or out of cells or into cellular vesicles, against their electrochemical gradient [
61]. Consistent with the role of these transporters in the excretion of diverse compounds, their expression (ABCG2, ABCA1, ABCA7, ABCG1, and ABCG5) has been found in lactating mammary epithelium [
62,
63]. Furthermore, the ABC transporters (ABCG2, ABCB1) have been proposed to be expressed by quiescent cancer stem cells, which allows them to survive cytotoxic or targeted therapies leading to relapse [
64].
We focused on ABCC5 (MRP5) as a candidate mediator of breast cancer skeletal metastases because we validated its expression in bone metastases at both the mRNA and protein levels (Figure
2). ABCC5 represents a membrane-spanning protein belonging to the C subfamily of the ABC transporters, with the ability to transport endogenous cyclic nucleotides [
65]. One concern associated with our identification of ABC transporters was the possibility that these proteins were upregulated as a consequence of the patient's treatment history rather than any specific role that they might play in the establishment of bone metastases [
66,
67]. To address this possibility, we examined ABCC5 expression in breast cancer cell populations that exhibit a bone-metastatic phenotype. We first used a series of
in vivo selected MDA-MB-231 cell lines with different organ-specific metastatic potential and found that ABCC5 was most highly expressed in the bone-metastatic MDA-MB-231 cells when compared with the parental and lung-tropic MDA-MB-231 cells (Figure
3A).
We next determined the level of ABCC5 expression in mouse-derived breast cancer cells with differential metastatic abilities. Consistent with the result in MDA-MB-231 cells, the highly metastatic 4T1 population, which is highly metastatic to bone and other sites, exhibited considerably higher ABCC5 expression in comparison with lung-only metastatic 66cl4 and nonmetastatic 67NR populations (Figure
4A). These data reveal that ABCC5 expression is highest in breast cancer cells that display enhanced bone-metastatic phenotypes, under conditions in which no treatments (antiestrogens, chemotherapy, or bisphosphonates) were used. Thus, it is conceivable that elevated ABCC5 expression is selected for because of a particular function that enables breast cancer cells to colonize the bone microenvironment efficiently.
The physiological functions of ABCC5 are, at present, poorly defined. It was reported that this protein serves as an efflux pump for intracellular cyclic guanosine monophosphate (cGMP) and, to a lesser degree, cyclic adenosine monophosphate (cAMP) [
65]. A growing body of literature suggests that elevated intracellular cGMP levels result in reduced breast cancer cell proliferation, ultimately triggering apoptosis via activation of protein kinase G (PKG) [
68‐
70]. Nitric oxide and natriuretic peptides, which serve as activators of soluble and transmembrane guanylyl cyclises, respectively, are abundant in bone and exert complex effects on bone cells, bone turnover, and bone formation [
71]. Thus, we reasoned that upregulation of a cGMP efflux pump could be advantageous for breast cancer growth and survival in the bone microenvironment. Diminished ABCC5 expression resulted in a significant reduction in the size of osteolytic lesions formed by both 1833-BM1 and 4T1 breast cancer cell models when compared with controls. However, this reduction was not a reflection of general alterations in breast tumor cell growth (Additional file
9). Consistently, no change in breast cancer cell proliferation, in either primary breast tumors or bone metastases, was noted when ABCC5 was knocked down compared with controls (Additional file
11). Moreover, the removal of ABCC5 did not result in elevated rates of apoptosis in either primary tumors or breast cancer bone metastases, as assessed with immunostaining for cleaved caspase-3 (Additional file
12).
Finally, we did not observe any differences in breast cancer cell apoptosis in response to a stimulator of cGMP production (A-350619 hydrochloride) in control or ABCC5 knockdown cells (data not shown). These observations suggest that ABCC5 does not promote breast cancer proliferation and survival in end-stage bone metastases through a mechanism that involves cGMP efflux and reduced PKG activation.
Elevated cGMP levels have also been shown to modulate the expression of matrix metalloproteinases (MMPs), although this regulation appears complex. Some studies suggest that elevated cGMP levels can induce the expression of MMP-2 and MMP-9 [
72,
73], whereas others argue that MMP-9 expression or secretion is suppressed by increasing cGMP concentrations [
74,
75]. Whether breast cancer cells that express ABCC5, and thus maintain low cytoplasmic cGMP levels through active efflux, are associated with elevated MMP-9 expression or secretion requires further investigation.
One interesting result was the observation that the density of TRAP-positive osteoclasts was reduced in lesions formed by breast cancer cells harboring reduced ABCC5 levels compared with bone metastases arising from control cells. Consistent with these observations, conditioned media from ABCC5 knockdown cells were less efficient in inducing in vitro osteoclast differentiation compared with media collected from cells harboring scrambled shRNA. It should be noted that the diminishment of osteoclastogenesis was not complete, and, in the case of one ABCC5 shRNA (shRNA 1), only trended toward a decrease in osteoclastogenesis. These results raise the possibility that the substrate of ABCC5 might directly influence osteoclast differentiation; however, it is also conceivable that the cargo that is pumped out of breast cancer cells by ABCC5 indirectly influences osteoclastogenesis through an intermediate cell type present in the bone microenvironment.
Previous studies showed that high levels of cGMP can negatively regulate the ability of osteoclasts to resorb bone, disrupting their attachment to the bone surface and preventing efficient secretion of HCl [
76,
77]. However, an important phase of the bone resorption mediated by osteoclasts is their ability to break the sealing zone, detach from the bone surface, migrate to a new area of bone, and reinitiate bone resorption. Indeed, nitric oxide or cGMP analogues have been shown to stimulate osteoclast migration [
78]. Thus, it is conceivable that locally elevated levels of cGMP, by virtue of a growing breast cancer metastasis, could contribute to enhanced osteoclast migration. The reduction of ABCC5 could diminish cGMP efflux, leading to impaired osteoclast motility and decreased bone resorption. This hypothesis might explain the specific requirement for ABCC5 expression in breast cancer cells that metastasize to the bone. Alternatively, an as-yet-unidentified ABCC5 cargo may be responsible for enhanced osteoclast differentiation and motility that leads to the formation of osteolytic breast cancer metastases in bone.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AAM carried out the LCM, expression analysis, and experiments focused on ABCC5 function. EA and MC collected the bone-trephine biopsies from breast cancer patients with known bone involvement. ZD performed immunohistochemistry and immunofluorescence staining. KT and SVK conducted the in vitro osteoclastogenesis experiments. SC and MH conducted the gene-expression analysis. AO identified regions of primary tumors and bone metastases suitable for LCM and scored the ABCC5 IHC staining. NB and MP generously provided primary breast tumor material. VO assisted in intracardiac injection of mammary tumor cells. GLS generously provided ABCC5 antibody. AAM and PMS designed the experiments, interpreted the results, and prepared the manuscript. All authors read and approved the final manuscript.