1
Supplementary Information
Supplementary text - RANKL mediates directed osteoclast migration
The establishment and expansion of breast cancer metastasis in bone involves osteoclastic resorption2. Therefore, we considered the possibility that RANKL, in addition to its well established roles in stimulating osteoclast formation and activity7,8, directly recruits osteoclasts. RANKL was found to induce chemotaxis of isolated rat osteoclasts (supplementary Fig. 4a). The migration of osteoclasts towards a source of RANKL was significantly oriented, in contrast to the random movement of control osteoclasts (supplementary Fig. 4b). Moreover, osteoclasts migrated significantly further towards RANKL compared to vehicle (supplementary Fig. 4c), with mean rate of migration to RANKL of 24 μm/h. This mean migration rate to RANKL is comparable to that elicited by other known chemotaxins for osteoclasts such as TGF-β (27 m/h)ref.23. Osteoclast chemotaxis to RANKL was blocked by OPG (supplementary Fig. 4c). In contrast to the effects of RANKL on osteoclasts, there was no chemotactic response by stromal cells. Thus, RANKL may regulate the recruitment of both osteoclasts and tumor cells to sites of bone metastases.
Methods
Tumor cell lines and primary cell cultures
B16F10 murine melanoma cells, MCF-7 human breast cancer, Hs578T human breast cancer, Colo205 human colon cancer, SW480 human colon cancer, LNCaP human prostate cancer, Du145 human prostate cancer, and T47D human epithelial breast tumor (all from ATCC) were maintained in culture in DMEM plus 10% fetal bovine serum (FBS; Invitrogen, Canada) at 37oC within humidified 5% CO2 air. MDA-MB-231 human breast cancer cells were grown at 37oC in Leibovitz’s media (Invitrogen) plus 10% FBS without CO2 supplementation in air. Non-transformed MCF10A mammary gland epithelial cells and primary mouse mammary gland epithelial cells freshly isolated from non-pregnant C57Bl6 mouse mammary glands were cultured as described24. Osteoclasts were isolated from the long bones of neonatal Wistar rats as described23. Osteoclasts stained positive for tartrate-resistant acid phosphatase (TRAP) and retracted in response to calcitonin.Animal care was carried out in accordance with the guidelines of the Council on Animal Care at the University of Toronto and at The University of Western Ontario.
RANKL and RANK expression analysis
Total RNA was isolated from cell lines and mouse tissues using Trizol (Invitrogen) and reverse transcribed into cDNA (Clonetech). PCR analysis of transcript expression was performed using primers for murine RANK (5’-GCAACCTCCAGTCAGCA-3’and 5’-GAAGTCACAGCCCTCAGAATC-3’), murine RANKL (5’-ATCAGAAGACAGCACTCACT-3’ and 5’-ATCTAGGACATCCATGCTAATGTTC-3’), human RANK (5’-GGGAAAGCACTCACAGCTAATTTG-3’ and 5’-GCACTGGCTTAAACTGTCATTCTCC-3’), and human RANKL (5’-TGGATCACAGCACATCAGAGCAG-3’ and 5’-TGGGGCTCAATCTATATCTCGAAC-3’). Amplification of RANK and RANKL from mouse tissues involved an initial denaturation for 1 minute at 94oC, followed by 35 cycles of denaturation at 94oC for 30 seconds, annealing at 56oC for 30 sec and extension for 30 seconds at 72oC. Human RANK and RANKL expression was detected as above, except for a change in annealing temperature to 57oC for RANK and 55oC for RANKL. -actin was used as internal control. In some experiments, RANK (TNFRSF11A, GenBank accession number NM_003839) transcripts were confirmed by quantitative real-time RT-PCR using a primer/TaqMan probe set specific for hRANK on RNA isolated from MDA-MB-231, MCF7, and LnCap cell lines, and mRANK on RNA isolated from RAW 264.7 and MC3T3 cell lines and the 7500 Sequence Detection System (Applied Biosystems, Foster City, CA). RANK mRNA levels were normalized to the -actin levels (endogenous control, human for MDA-MB-231, MCF7, and LnCap and mouse for RAW 264.7, and MC3T3) and expressed relative to levels in MC3T3 cell line. Detection of cell surface expression of RANK protein via FACS utilized FITC-conjugated human RANKL (aa159-317; Amgen, Thousand Oaks).
RANK signaling, proliferation, and cell death assays
Cancer cells were serum starved for 12 hours and then stimulated with recombinant murine RANKL (recombinant aa158 to aa316) in the presence or absence of recombinant murine OPG-FC protein (aa22-401; rOPG, both from Amgen, Thousand Oaks)11, SDF-1(R&D Systems), or recombinant prolactin (Sigma). In addition, commercially available RANKL (R&D Systems) was used with similar results in osteoclastogenesis indicating that the observed effects were not due to secondary effects of recombinant RANKLaa158-aa316. Cell lysates (10 mM Tris pH 7.6, 5 mM EDTA, 50 mM NaCl, 1% triton-X, 30 mM tetra-sodium pyrophosphate, 200 M sodium orthovanadate, 1 mM PMSF, 5 g/ml aprotinin, 1 g/pepstatin, 2g/ml leupeptin)were resolved on 8-12% polyacrylamide gels under reducing conditions and proteins transferred to Hybond ECL nitrocellulose membranes (Amersham Pharmacia Biotech, Buckinghamshire, UK). Primary antibodies reactive to ERK1/ERK2, active ERK1/ERK2 (phospho-Thr202/Tyr204), PKB/AKT, active PKB/AKT (phospho-PKB/AKT-Ser473), STAT5A/B, phospho-Stat5A/B (phosphor-Tyr694)(Cell Signaling and Transduction Lab),and actin (Sigma) were used. For actin polymerization studies, tumor cells were stimulated with RANKL or SDF-1 and actin polymerization determined using phalloidin-FITC.Tumor cell proliferation was determined using H3-thymidine uptake. Cell death was detected via FACS using propidium iodide/AnnexinV-FITC double staining
Tumor cell chemotaxis
Migration of cancer cells was assessed using a 96-well chemotaxis chamber (Neuro Probe Inc. MD) with fibronectin (Sigma) coated polycarbonate filters (8 and 12 m pore size). All cells were starved for 12 h in DMEM (10 mM Hepes, 0.1% BSA) media, detached using 5 mM EDTA in Ca2+/Mg2+-free Hank’s buffer, counted and re-suspended for each assay. DMEM media (10mM Hepes, 0.1% BSA) containing rRANKL, rOPG, or the chemokines SDF-1, 6Ckine, and CTACK (all chemokines were purchased form R&D Systems) was placed in the lower wells and 5 X 105 B16F10 or 2 X 105 human breast, prostate or colon cancer cells were placed in the upper wells. Migration of cells was determined at 37oC for 16 h (B16F10 cells) or 6 h (human cancer cells), fixed, and stained as previously described for chemokine assays6. Migration was quantified using a plate reading spectrophotometer (Beckman). It should be noted that tumor cells migrated in a dose response starting from doses as low as 100ng/ml of rRANKL and that two different sources of RANKL (Amgen and R&D) gave similar results.Inhibition of chemotaxis with the PI3K inhibitor wortmannin, the PLC blocker U73122, the PKC inhibitor GF109203X and the MEK1 inhibitor PD98059 (Calbiochem)was performed as previously described25-27.
RANK detection on human breast cancer tissue arrays
Paraffin embedded specimens of tumors, lymph node metastasis, and adjacent normal tissue were collected from 59 female breast cancer patients who underwent surgery in 1988-1994, and analyzed retrospectively under protocols approved by the institutional review board of the Medical University of Vienna. Triplicate core biopsies of 0.6mm were taken from each donor paraffin block and arrayed into a recipient block with a MTA1 manual tissue puncher/arrayer (Beecher Instruments, Silver Spring, MD) as described28. 5 μm thick paraffin sections were treated in xylene and rehydrated in a gradient of ethanol.After antigen retrieval by 10 mM sodium citrate (pH6.0), endogenous peroxidase activity was blocked with 3% H2O2. The sections were then incubated with 20g/ml of a goat polyclonal anti-RANKantibody (M-20;Santa Cruz Biotechnology, Santa Cruz, CA). Goat and rabbit IgG antibody were used as a negative control. After one hour incubation at RT and washing, the sections were incubated with biotinylated anti-goat/rabbit IgG antibodies for 30 minutes, followed by incubation with streptavidin–peroxidase for 15 minutes and 3,3’-diaminobenzidine. Sections were counterstained with hematoxylin. Immunostaining wasscored on triplicate tissues by two independent observers (T.N. and R.S.) using the following arbitrary scale: 0, no staining; 1, weak staining; 2, medium staining; 3,strong staining. It should be noted that all of the cancer tissues showed staining in greater than 50% of the total tumor area.
Osteoclast chemotaxis
Osteoclasts were plated on 35-mm culture dishes or 12-mm glass coverslips in medium 199 (25 mM HEPES, 26 mM HCO3-, supplemented with 15% FBS) and incubated for 1 hour (37°C, 5% CO2). Dishes and coverslips were washed to remove non-adherent cells and osteoclasts were cultured as described. Because mature osteoclasts are not amenable to Boyden chamber assays, chemotaxis was assessed in a single-cell chemotaxis assay23using an inverted phase contrast microscope coupled to a time-lapse video recorder (AG-6720, 1 frame/2 sec; Panasonic, Secaucus, NJ, USA). A glass micropipette containing solublerRANKL (1 μg/ml), rRANKL premixed with OPG (1 g/ml and 2.5 g/ml) or vehicle in supplemented media was positioned 200-400 µm from the cell under study. The contents of the pipette flowed passively into the bath and migration was monitored for 6-12 h. Images of osteoclasts were digitized and centroids determined using Optimas 3.1 (Bioscan, Inc., Edmonds, WA, USA). Polar data were analyzed using Rayleigh's and modified Rayleigh's tests.
In vivo tumor metastasis
Murine B16F10 melanoma cells and human colon cancer cells that do not express RANK were injected into the left cardiac ventricle of 7-10 wk old female C57BL/6 mice (Harlan Sprague Dawley, Houston, TX) or immunodeficient nude mice, respectively, under avertin anesthesia as described9. Simultaneously, mice were daily treated with either vehicle (PBS), 20 g/day rOPGor zolendronic acid (s.c.g/mouse per day) as previously described18. After final treatment, mice were sacrificed and bones (femur, tibia, humerus, and lumbar vertebrae) and organs (brain, ovary, spleen, kidney and adrenal glands) collected for histological analysis. Serum was collected for calcium (SeCa), phosphorous (SeP), alkaline phosphatase (ALP) and tartrate-resistant acid phosphatase (TRAP) analysis. Bones were fixed in either 70% ethanol or were processed through 10% formalin before being rinsed thoroughly and decalcified in formic acid solution or decalcified using 10% EDTA in PBS (pH 7.5) for TRAP staining. Radiographic and histomorphometric analysis of all bones was as previously described7,21.Briefly, tissues were fixed in 10% formalin, sectioned and stained with hematoxylin and eosin (H&E) to determine the presence or absence of tumor metastases. Midline longitudinalsections of long bones were stained for TRAP activity. Two non-serialsections of each bone were assessed. The total tissue sectionarea and the tissue area occupied by tumor cells were measuredusing an Osteomeasure bone analysis program (Osteometrics Inc.,Decatur, GA).All mice were kept at the Ontario Cancer Institute Animal Care Facility according to institutional guidelines.
Supplementary references
23. Pilkington, M.F., Sims, S.M.& Dixon, S.J. Transforming growth factor-β induces osteoclast ruffling and chemotaxis: potential role in osteoclast recruitment. J. Bone Miner. Res.16,1237-1247 (2001).
24. Medina, D. & Kittrell, F. in Methods in Mammary Gland Biology and Breast Cancer Research (Kluwer Academic/Plenum Publishers, New York, 2000).
25. Curnock, A.P., Logan, M.K. & Ward, S.G. Chemokine signaling: pivoting around multiple phosphoinositide 3-kinases. Immunology105(2), 125-136 (2002).
26.Wells, A., Ware, M.F., Allen, F.D. & Lauffenburger, D.A. Shaping up for shipping out: PLCgamma signaling of morphology changes in EGF-stimulated fibroblast migration. Cell Motil. Cytoskeleton44(4), 227-233 (1999).
27. Rabinovitz, I., Toker, A. & Mercurio, A.M. Protein kinase C-dependant mobilization of the 64 integrin from hemidesmosomes and its association with actin-rich cell protrusions drive the chemotactic migration of carcinoma cells. J. Cell Biol. 146(5), 1147-1160 (1999).
28. Kononen, J., Bubendorf, L., Kallioniemi, A., Barlund, M., Schraml, P., Leighton, S., Torhorst, J., Mihatsch, M.J., Sauter, G. & Kallioniemi, O.P. (1998) Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat. Med, 4, 844-847 (1998).
Supplementary Figures
Supplementary Figure 1. RANK expression on primary and metastatic breast cancer cells and epithelial cancer cell lines
a) Human breast and lymph node tissue-array slides were staining for RANK expression (see methods). Staining was assessed in normal breast tissue biopsies from a patient subset (n=10), primary breast cancers (n=59), and breast cancer cells that have metastasized to the local draining lymph nodes (n=30). Staining intensity was scored in arbitrary units: 0(no staining), 1 (weak staining), 2 (medium staining),and 3 (strong staining). Statistical significance of increased RANK expression as compared to normal mammary gland epithelial cells was determined using an ANOVA-Tukey test. ns = none significant;* p<0.001.b) Expression of RANK in (1) Colo205, (2) SW480, (3) LNCaP, (4) Du145, (5) MDA-MB-231, (6) Hs578T, (7) MCF-7 cancer cell lines, and primary human breast tumor samples (8-11). Expression was detected by RT-PCR.
Supplementary Figure 2. Expression of functional RANK on epithelial cancer cells
a)Expression of RANK onT47D breast cancer cells (solid line). Background is shown as a dotted line.RANK expression was detected using RANKL-FITC. b,c)RANKL induces ERK1/ERK2 phosphorylation on MDA-MB-231 (b) and PKB/Akt and ERK1/ERK2, but not STAT5 activation, in T47D (c) humanbreast cancer cells. Serum-starved cells were stimulated with rRANKL [1g/ml] or prolactin [5g/ml] for the indicated time periods. PKB/AKT activation (Ser473 phosphorylation; p-AKT), ERK1/ERK2 activation(Thr202/Tyr204 phosphorylation; p-ERK), and STAT5A/B (Y694 phosphorylation; p-STAT5) were detected byWestern blotting. Total PKB/AKT, STAT5A/B, and control -actin protein levels are shown. Similar to MDA-MB-231 cells, the human epithelial breast tumor cell line T47D expresses high levels of the prolactin-receptor and RANK but does not express detectable levels of RANKL as determined by RT-PCR and protein analyses. Prolactin-induced activation of downstream signaling pathways is shown for specificity. One experiment representative of 5 different experiments using different stimulation conditions is shown. d,e) RANKL has no apparent role in proliferation (d) or cell death (e) of humanMDA-MB-231 breast cancer cells. Cells were serum starved and proliferation measured on day 1 and day 3 following rRANKL [2.5 g/ml] addition using H3-thymidine uptake. For cell death, RANKL-activated [2.5 g/ml] MDA-MB-231 breast cancer cells were left untreated (control) or treated with anisomycin [50 M], sorbitol [0.4M], or UV irradiation [240mJ/cm2].% survival was detected by FACS using propidium iodide/Annexin5.
Supplementary Figure 3. RANKL triggers chemotaxis of MDA-MB-231 breast cancer cells
Migration of MDA-MB-231 human breast cancer cells in response to different doses of rRANKL, SDF-1, or both rRANKL plus SDF-1. Percentage migration (mean values of triplicate cultures +/- SD) compared to non-stimulated control cells is shown.
Supplementary Figure 4. RANKL triggers chemotaxis of mature osteoclasts
a-c) Migration of osteoclasts monitored by time-lapse microscopy. a) Videomicrographs illustrate fields containing 3 different osteoclasts (lower left of each frame) and micropipette tips outlined in black (upper right) at time 0 (time of introduction of micropipette containing vehicle (Control) or RANKL (1 μg/ml)). White outlines represent positions of the osteoclasts at the times indicated. Video files are available as supplementary material. b) Polar plots in which origin represents the position of the centroids of all cells at time 0 and axes are oriented to set the tip of the micropipette on the positive x-axis (90°). The positions of the centroid of each cell at 6 h are illustrated by closed symbols for vehicle (left panel, 7 cells) and open symbols for rRANKL (right panel, 7 cells). Arrow illustrates the mean vector. The migration of control osteoclasts was not significantly oriented. In contrast, the mean direction of migration induced by RANKL was significantly oriented towards the tip of the micropipette (p<0.004). c) Bars illustrate mean distances traveled by osteoclasts from 0 to 6 h in the direction of the micropipette tip (determined as the projection on the x-axis). Osteoclasts migrated significantly further towards the RANKL-containing micropipette (RANKL) than towards the micropipette containing vehicle (control) as assessed by Student's t-test (* p<0.03). OPG blocked RANKL-induced chemotaxis of isolated osteoclasts (RANKL + OPG). Data are representative means SEM for 8 osteoclasts for vehicle, 16 osteoclasts for RANKL and 6 osteoclasts for RANKL+OPG from 5 different cell preparations for each condition.
Supplementary Figure 5. RANKL induces chemotaxis of B16F10 melanoma cells
a) Expression of RANK and -actin mRNA in (1) untreated (2) RANKL[2.5 g/ml]treated B16F10 melanoma cells. Expression was detected using RT-PCR. b) Additive effects of chemokines and rRANKL in tumor cell migration. Migration of B16F10 melanoma cells in response to rRANKL (2.5 g/ml) and the chemokines 6Ckine (120 ng/ml), SDF-1 (80 ng/ml), and CTACK (100 ng/ml). Percentage migration (mean values of triplicate cultures +/- SD) compared to non-stimulated control cells is shown. Data are representative of 3 different experiments. Treatment of cells with 6Ckine plus RANKL or CTACK plus RANKL significantly increased migration above treatment with chemokine alone (* p<0.05). c) Migration of B16F10 melanoma cells in response to RANKL (2.5 g/ml) or SDF-1[80 ng/ml] in the absence or presence of rOPG [10 g/ml] or the CXCR4 blocking Ab[25 g/ml, clone 4471].Percentage migration (mean values of triplicate cultures +/- SD) compared to non-stimulated control cells is shown. Note that OPG does not block chemokine-induced cell migration and that anti-CXCR4 does not block RANKL-induced migration. RANKL-induced migration of cells is blocked with rOPG (* p<0.05) and SDF-1-induced migration of cells is inhibited by anti-CXCR4 (* p<0.01). Results are representative of 3 different experiments. d) Stimulation of B16F10 melanoma cells with RANKL (2.5 g/mL) or SDF-1 (1g/ml) triggers phosphorylation of ERK1/2 (Thr202/Tyr204 phosphorylation; p-ERK) and PKB/AKT (Ser473 phosphorylation; p-AKT). Total ERK1/2 and PKB/AKT expression are shown as controls.
Supplementary Figure 6. B16F10 melanoma cells do not activate osteoclasts
a) In line with previous results16, B16F10 melanoma cells (asterix) that have metastasized into the long bones and vertebrae are not associated with osteoclasts. Bone sections were stained with TRAP to detect osteoclasts in situ(arrows).b,c) B16F10 melanoma tumors did not decrease bone density as determined by radiography (a typical femur radiogram is shown) (b) nor changed total, trabecular, and cortical bone densities as determined by pQCT measurements(c).There was also no increase in TRAP activity, and no evidence of altered serum calcium, phosphorous or alkaline phosphatase activity in mice that carried confirmed B16F10 tumor metastases in the bones as compared to non-injected controls, confirming that B16F10 melanoma tumors are non-osteoclastogenic.