Metastasis is a common phenomenon in the course of malignant tumor disease and the major reason for cancer-associated morbidity and mortality. In osteotropic malignancies such as breast, prostate, lung, kidney and thyroid cancer, the skeleton is frequently affected as metastatic site. From these cancers, the prevalence of skeletal disease is greatest in patients with breast or prostate carcinoma who in the advanced stage develop skeletal metastases in 70%.1 Breast carcinoma patients experience a median survival time of 20 month after the first appearance of a bone metastasis.2 Such a lesion results in a number of skeletal complications including pathological fractures, bone pain, hypocalcaemia and spinal cord compression. The reasons for the skeleton being the preferential localization site of metastasis in osteotropic cancers are poorly understood.
In patients with primary breast cancer, elevated serum bone sialoprotein (BSP) was recognized as prognostic marker of subsequent bone metastasis3 and was associated with poor survival.4 BSP is a noncollagenous protein of the extracellular bone matrix and a member of the SIBLING (Small Integrin-Binding Ligand, N-linked Glycoprotein) family, which consists of genetically related proteins clustered on human chromosome 4.5 This glycoprotein is relatively specific for the tissues of the skeleton and is known to have a high binding affinity for calcium and hydroxyapatite.5, 6
To further investigate the function of BSP in the course of bone metastasis formation, we set up an animal model that would allow following processes such as tumor cell extravasation, adhesion to the target tissue and formation of osteolytic lesions.
For this complex scenario, animal models are indispensable tools to investigate the pathogenesis of bone metastasis in vivo and to examine the effects of a therapeutic intervention. For inducing breast cancer bone metastasis in rodents, most experimental models require the injection of human cancer xenografts into immunodeficient mice. This is commonly achieved by intracardial injection or intraosseous administration of breast cancer cells.7 Disadvantageously, the former mode of tumor cell administration into the left ventricle of the heart is associated with dissemination of tumor cells into all skeletal and visceral peripheries of the organism. Local intraosseous administration, however, causes bone damage and lacks processes such as tumor cell extravasation and invasion. Finally, methods based on the injection of bone specific tumor cell subclones are more likely to specifically induce and mimic the process of bone metastasis, but metastatic dissemination is still observed at multiple skeletal sites.
Here we describe a new experimental animal model for inducing site-specific osteolytic lesions in the hind leg of nude rats. This method is characterized by intra-arterial injection of human breast cancer cells into an anastomosing vessel between the femoral and the iliac arteries. We applied this model to the effects of an antibody against bone sialoprotein, which resulted in reduced growth of osteolytic lesions in treated animals.
Anti-BSP, polyclonal chicken antibody against human BSP; BSP, bone sialoprotein; CT, computed tomography; FA, femoral artery; GFP, enhanced green fluorescent protein; MDA-MB-231GFP, MDA-MB-231 human breast cancer cells transfected with GFP; SEA, superficial epigastric artery; SIBLING, Small Integrin-Binding Ligand, N-linked Glycoprotein.
Material and methods
Cell lines and culture conditions
The human estrogen-independent breast cancer cell line MDA-MB-231 was obtained from the American Type Culture Collection (ATCC). MDA-MB-231 cells stably transfected with GFP (pEGFP-N1; ClonTech, Palo Alto, CA) were kindly provided by Dr. D.R. Welch (Pennsylvania State University College of Medicine, Hershey, PA). Tumor cell lines were cultured routinely in RPMI-1640 (Invitrogen, Karlsruhe, Germany) supplemented with 10% FCS (Sigma Chemical Co., Taufkirchen, Germany). For the neomycin resistant MDA-MB-231 cells transfected with GFP, 500 μg/ml Geneticin (G418; Invitrogen, Karlsruhe, Germany) was added. All cultures were kept under standard conditions (37°C, humidified atmosphere, 5% CO2) and passaged 2–3 times a week to keep them in logarithmic growth.
A volume of 100 μl RPMI medium per well containing 5×103 MDA-MB-231GFP cells was plated onto 96-well plates (Microtest™, Becton Dickinson, Heidelberg, Germany). After 24 hr, the cells were treated by adding 100 μl medium containing the IgY antibody (Immundiagnostik, Bensheim, Germany). The IgY concentrations were selected to yield final concentrations of 1 to 400 μg/ml. Thereafter, the plates were incubated for 1 to 7 days under standard cell culture conditions. The number of surviving cell was determined by adding 10 μl/well of 3-[4,5-dimethylthiazol-2-yl] -2,5-diphenyltetrazolium bromide; (MTT; 10 mg/ml). The plates were further incubated for 3 hr and thereafter the supernatant was removed, and formazan crystals that had been developed were dissolved by adding of 100 μl acidified 2-propanol/well (0.04 N HCl). Absorption was measured by an automated microtiter plate reader at 540 nm, reference filter 690 nm (Anthos 2001, Anthos GmbH, Krefeld, Germany).
Animals and husbandry
Nude rats (RNU strain) were obtained from Harlan Winkelmann (Borchen, Germany) and Charles River (Sulzfeld, Germany) at an age of 6–8 weeks. They were housed 2 per cage at specific pathogen-free conditions in a minibarrier system of the central animal facility. Autoclaved feed and water was given ad libitum to the animals that were maintained under controlled conditions (21 ± 2°C room temperature, 60% humidity, and 12 hr light-dark rhythm). Experiments were approved by the responsible governmental Animal Ethics Committee (Regierungsprasidium Karlsruhe, Germany).
In vivo metastasis model
Subconfluent tumor cells were harvested using 2 mM EDTA in PBS- (phosphate-buffered saline without Ca++ and Mg++) and 0.25% trypsin (Sigma Chemical Co., Taufkirchen, Germany). Cells were counted in a Neubauer's chamber and suspended in RPMI (5×105 cells in 1 ml). For tumor cell implantation, rats were anaesthetized with a mixture of laughing gas (nitrous oxide; 1 l/min), oxygen (0.5 l/min) and isoflurane (1–1.5 vol. %).
A clear cut of 2–3 cm length was performed in the inguinal region. After preparation of all arterial branches as seen in Figure 1, the flow of the femoral artery (FA; Fig. 1) was temporarily occluded by clips that were placed proximal and distal of the superficial epigastric artery's (SEA; Fig. 1) origin. The deep femoral artery (Fig. 1), which normally branches off the FA and supplies the medial and caudal muscles of the thigh,8 was also clipped in cases of an anatomical variant as seen in Figure 1. In addition, the SEA was ligated distally, which allowed the opening of this vessel without bleeding. In this context, it is interesting to note that a ligation of the distal SEA is possible because it anastomoses with the caudal epigastric artery that arises from the pudendoepigastric trunk, which is a branch of the iliac artery.8 After making an incision proximal of the ligation, a 1% papaverin solution was administered onto the SEA to facilitate the subsequent insertion of a needle (0.3 mm diameter and 42 mm length). After insertion, the needle was fixed in an external support, which reduces irregular movements that would result in perforation of the arterial wall and allows connection with a syringe. Then the distal clip was removed from the FA and placed onto the saphenous artery, which runs superficially and supplies the dorsal and plantar aspects of the foot.8 MDA-MB-231 cells (105 cells suspended in 0.2 ml media) were slowly injected into the SEA and by virtue of the clips directed to the descending genicular and popliteal arteries (Fig. 1), both supplying the knee joint and muscles of the leg.8
Anti-bone sialoprotein immunoglobulin
Polyclonal antibodies developed in chicken against human bone sialoprotein were obtained from Immundiagnostik (Bensheim, Germany). Appropriate dilutions were made using PBS without Ca++ and Mg++.
Radiographic examination and computer tomography
Radiographic examinations of rats were made every 7–14 days under general anesthesia. The animals were fixed in a.p. and p.a. position on a Cronex 5 film (Agfa-Gevaert N.V., Mortsel, Belgium) and exposed to X-rays generated by a Vertix U at 40 kV and 1.4 mA (Siemens, Erlangen, Germany). The X-ray films were processed by an automatic developing machine (Scopix LR 5200; Agfa, Köln, Germany) and the resulting images were scanned using a digital imaging program (Adobe Photoshop 6.0, Adobe Systems Inc.) with a resolution of 300 dpi. After inverting the scanned radiographs (turning positive into negative), analyses were done with a computer based imaging program (UTHSCSA Image Tool, University of San Antonio, TX). This program measures area, perimeter and mean gray scale of the lesion. Lesion sizes were mainly described as the product of area and mean gray scale.
High resolution computed tomography (HRCT) was performed using a multidetector Somatom Plus 4 CT-scanner (Siemens, Erlangen, Germany). Whole animals were investigated with a native spiral scan (increment: 0.5 mm; kernel: 90) at 100 mA and 120 kV. Based on CT slices with 0.5 mm thickness, 3-dimensional volume rendering reconstructions of the skeleton were performed using the Vitrea 2 workstation (Vital Images, Plymouth, MN).
For each animal, the lesion sizes obtained from X-ray analysis were used to draw a curve vs. time after tumor cell inoculation. For statistical comparisons, these individual growth curves were interpolated and lesion sizes were computed for intervals of 10 days. For comparison between 2 groups, the Wilcoxon-Test was applied; p-values < 0.05 were considered significant.
Bone storage and histology
After storage in 70% ethanol, the specimens were embedded in Technovit 9100 New (Heraeus Kulzer, Wehrheim, Germany) according to the instructions for application. Thereafter, the embedded samples were mounted on blocks using Technovit 3040 (Heraeus Kulzer, Wehrheim, Germany) in order to remove the samples from the embedding device. Before cutting the block into 5 μm thick sections (Supercut 2050, Reichert-Jung, Germany), it was wetted with 30% ethanol. The sections were removed and placed on precoated slides using 96% ethanol. Finally, the sections were covered with polyethylene foil, pressed at 37°C overnight and then deplasticised before being stained.
For immunocytochemical staining, the sections were rinsed several times with 0.1 M phosphate buffered saline (PBS, pH 7.4), incubated with 10% normal goat serum (Vector Laboratories, Burlingame, CA) and diluted in PBS for 2 hr. Thereafter, the sections were incubated with the primary antibodies against BSP (anti human bone sialoprotein from Immundiagnostik AG, Bensheim, Germany; mouse monoclonal anti human bone sialoprotein A 423.2, rabbit polyclonal anti bone sialoprotein A 4218.1 and monoclonal anti rat bone sialoprotein were a gift from Jaro Sodek, Toronto, Canada), which were diluted 1:100 in PBS containing 2% bovine serum albumin (PBS-BSA).
After several rinses in PBS, sections were incubated for 2 hr with a 1:200 dilution of secondary antibody goat-anti-mouse-Cy3 (Jackson Immunoresearch Laboratories, Inc., West Grove, PA) in PBS-BSA. The preparations were rinsed again with PBS, counterstained using DAPI (Serva, Heidelberg, Germany), dried and cover slipped. Control sections were treated with nonspecific mouse antibodies (IgG1, DAKO), diluted and applied similarly to the specific antibody. Sections were studied fluorescence microscopically using a photomicroscope equipped with epifluorescence (Axiophot; Zeiss, Germany).
Subsequent sections were stained with hematoxylin and eosine (Merck AG, Dietikon, Switzerland) for light microscopical features.
For our study, 4 groups of nude rats contributed to the total number of 152 animals (Table I). In the pilot study, 23 rats (15.1%; group 1) were used to determine the optimal tumor cell number for inducing lytic lesions as well as the gender showing the higher tumor take rate. The main part (96 animals, 63.16%; group 2) remained untreated until the first lytic lesion was detected and was used to determine the overall tumor take rate. Of these, 27 remained untreated and served as control group. Finally, 2 groups of 7 and 26 rats, respectively, received pretreated cells (4.6%; group 3) or treatment with an antibody against bone sialoprotein (BSP) after tumor cell inoculation (17.1%; group 4).
Table I. Study Design and Model Properties
Rats with tumor cell inoculation and no further treatment until onset of first lytic lesion.
Two rats with spontaneous remissions were excluded.
Circular defects of femur or tibia (circular lysis of cortical bone or bone fracture); circular defects of the fibula were excluded.
The average daily growth rate (day 30–110) of untreated controls was 0.47 mm2.
In a pilot study on the optimum take rate, the parameters tumor cell number and sex were varied. For this purpose, male and female animals were observed for a period of 90 days after administering increasing tumor cell numbers. Two male and 2 female rats, respectively, received an injection into the superficial epigastric artery of the right hind leg containing 2.5×104, 7.5×104, 2.5×105 and 7.5×105 MDA-MB-231 or MDA-MB-231GFP cells. As a result, 50% of all animals developed discernible lytic lesions within an observation period of 90 days. In 5 of 8 males and in 3 of 8 females, overt lytic lesions were observed by X-rays. The tumor take rate did not differ between MDA-MB-231 and MDA-MB-231GFP cells (4 of 8 animals, respectively). Cell numbers between 7.5×104 and 2.5×105 were found appropriate, as these animals developed more and bigger lesions in comparison to rats receiving higher or lower numbers of cells. In an additional group of 7 female nude rats, only 3 developed discernible lytic lesions after inoculation of 1×105 MDA-MB-231GFP cells. As a consequence, 1×105 MDA-MB-231GFP cells were inoculated into male rats as basis for all further studies.
Inoculation of 1×105 MDA-MB-231GFP cells was well tolerated as animals recovered quickly from general anesthesia and did not show weight loss or any signs of walking with a limb. Overt lytic lesions were detected by X-rays in some animals as early as 3 weeks after tumor cell implantation and 89 of 96 animals were positive after day 30. The corresponding tumor take rate was 92.7% (Table I). In the subgroup of control rats, 2 of 27 (7.4%) showed a spontaneous complete remission of an established lytic lesion within the observation period and 1 rat showed a delayed appearance of its lytic lesion at day 50 after tumor cell inoculation.
The appearance of lesions was usually first detected in the distal femur and the proximal tibia of the hind leg inoculated with tumor cells. Thereafter, small single lytic lesions started to increase in extend and to become confluent with adjacent lesions.
Lytic lesions were observed in the femur and the tibia of 24 animals (96%), respectively, with an overall coincidence of 92% (23 of 25). The order of appearance varied, however, with tibial lesions preceding femoral lesions in 15 of 23 rats (65.2%), the reverse order in 1 of 23 rats (4.3%) and a concomitant appearance in 7 of 23 rats (30.4%). The fibula was never affected exclusively but in 15 of 25 rats (60%; Table I) together with tibia and femur.
Advanced lytic tumor growth was associated with development of surrounding soft tissue metastasis in 8 of 25 rats (32%), and with circular defects of cortical bone in 9 of 25 rats (36%; Table I). On average, soft tissue metastasis was detected earlier than circular cortical defects (60 days vs. 80 days after tumor cell inoculation, respectively), and was observed mainly in animals with fast growing lesions of the skeleton.
The mean lesion size of 25 animals inoculated with MDA-MB-231GFP cells in relation to time is shown in Figure 2. The left dependent axis is scaled for both the perimeter (in mm) and the area (in mm2). The mean lesion size is also expressed as product of area and mean gray scale of the measured lesions (in pixel) in order to take the extent of bone lysis into consideration. The mean lesion area initially rose from 17 mm2 (day 30) to 45.5 mm2 (day 60) with a daily average growth rate of 0.95 mm2/day. Thereafter, the average growth rate of lesion size decreased within the observation time of day 60 to 110 to 0.18 mm2/day and the resulting overall growth rate was 0.47 mm2/day. Single animals were continued to be observed for 10 more days. Within this period the osteolytic lesions showed a further increase in size consistent with the growth rate before.
The concept of averaging all lesion sizes, which were detected at a given time interval following tumor implantation, resulted in a steady increase of the average lesion area but was associated with a relatively large standard deviation of the mean which is due to the variability of individual growth curves. At day 30 after tumor cell inoculation, 10 of 25 rats (40%) showed a lesion smaller than 5 mm2 or bigger than 15 mm2, and in 5 of 25 rats (20%) a lesion size between 5 and 15 mm2 was detected (Table I). To come up for differences in lesion size at the first X-ray examination, the rats were stratified according to their lesion size into different classes. Class 1 comprised lesions up to 9.5 mm2, class 2 lesions between 9.5 and 19 mm2, class 3 those between 19 and 28.5 mm2, class 4 those between 28.5 and 38 mm2 and class 5 all lesion sizes bigger than 38 mm2. These classes had been selected according to the initial mean growth rate (0.95 mm2/day), and the growth curves of individual rats with a corresponding lesion size was set to start at the respectively delayed period of time (Fig. 3). This artificial overlay of growth curves resulted in a distinctly lower standard deviation and an almost linear increase compared to that of the normal growth curve.
Figures 4 to 6 illustrate the development of lytic lesions visible in a control rat's hind leg around days 40, 66 and 100 after tumor cell inoculation. When comparing X-ray and computed tomography images, small lytic lesions could easily be detected by X-rays (Fig. 4a) but less well by CT reconstruction (Fig. 4b). On axial slices of a CT scan through the entire animal at selected height, even small lesions of cortical destruction could be detected (Fig. 4c). As lytic lesions expanded, the fibula was partly destroyed by lysis at day 66 after tumor cell inoculation (Fig. 5) and in the proximal tibia a circular defect of cortical bone could be observed (Fig. 5a). Lytic lesions of the femur and tibia, which did not affect cortical bone, could hardly be seen in the CT reconstruction (Fig. 5b). At day 100 after tumor cell inoculation, advanced lysis in the tibia and fibula could be observed by X-rays in a.p. position (Fig. 6a) as well as by CT-reconstruction (Fig. 6b). In contrast to the X-ray in a.p. position, the one in p.a. position (Fig. 6c) did not show the fracture of the proximal tibia. Soft tissue metastasis around tibia, fibula and the knee joint could be observed by both X-rays and CT reconstruction (Fig. 6a,c,d).
Micrographs of histological preparations were taken 120 days after tumor cell inoculation with the aid of light and fluorescence microscopy (Fig. 7). The overview of an osteolytic lesion in the tibia of an untreated control rat stained with hematoxylin and eosine (Fig. 7a) shows cortical bone lysis and tumor cells in the bone marrow cavity.
Immunohistochemistry was performed with antibodies raised against human and rat BSP for which a cross reaction had been excluded. Using the antibody against rat BSP for immunohistochemistry, red immunostaining indicated rat BSP. Thus visualized bone sialoprotein produced by the animals was observed in osteocytes, osteoblasts, at the surface of bone (Fig. 7b) and in cells of the connective tissue (Fig. 7c). In Figure 7d,e, human BSP expressed by MDA-MB-231GFP cells was also labeled with red immunostaining. This red fluorescence of human BSP was always colocalized with the green fluorescence of GFP, which merged into a yellow staining of structures. Two small yellow colored arteries on the ruffled border of bone (arrowheads; Fig. 7d) give evidence for further transport of cells staining positive for both, GFP and human BSP, into bone. These cells invaded also the surrounding tissue thus forming a soft tissue metastasis that showed weaker staining (Fig. 7e).
Effects of the IgY antibody on proliferation
The IgY antibody used was characterized by MTT assay regarding its efficacy on the in vitro proliferation of MDA-MB-231GFP cells. As shown in Figure 8, the concentration of 1 μg/ml medium was ineffective. Higher concentrations gradually decreased cell proliferation with the highest concentration (400 μg/ml), precluding any proliferative activity of MDA-MB-231GFP cells. The IC50 corresponded to 140 μg/ml at 7 days after start of treatment. A control anti IgY antibody did not show any significant effect (data not shown).
Pretreatment and treatment
Animals of groups 3 and 4 (Table I) were examined to investigate the effect of an antibody against bone sialoprotein (BSP) on the growth of lytic lesions. For that purpose, MDA-MB-231GFP cells were incubated with the anti-BSP antibody before inoculation (pretreatment, group 3) and another group of rats received injections of the antibody after inoculation of tumor cells (treatment, group 4; Table I).
In a group of 7 animals receiving pretreatment (group 3; Table I), MDA-MB-231GFP cells were incubated for 2 hr with 600 μg/ml of the antibody against BSP before these cells were inoculated into the superficial epigastric artery. As a result, 1 rat developed no discernible lesion during the observation period of 60 days and the average lesion size in the other rats ranged between 6.4 mm2 (day 30) and 8.6 mm2 (day 60; Fig. 9a). Compared to untreated controls, mean lesion sizes were significantly smaller at days 40–60 after tumor cell inoculation (Fig. 9a; p < 0.05).
Animals of the treatment group (group 4; Table I) received 1 to 3 injections of the antibody within 10 days after tumor cell implantation. Interestingly, low (10–20 mg/kg; n = 5) and high (120–180 mg/kg; n = 6) total doses of the antibody were ineffective regarding the growth of lytic lesions (227 and 200% control at day 60 after tumor cell inoculation, respectively, and not significantly vs. control; data not shown). This observation contrasted with administration of an intermediate dosage (30–80 mg/kg; n = 15), which caused a significantly reduced growth of lytic lesions (p = 0.015). Specifically, in a subgroup of 10 rats receiving a total dose of 60 mg/kg of the anti-BSP antibody s.c. as treatment, the first injection was given directly after inoculation of MDA-MB-231GFP tumor cells, as well as 2 additional injections 2 and 4 days later (injections of 20 mg/kg, respectively). During the observation period of 60 days, 2 animals (20%) did not develop discernible lytic lesions. The remaining 8 rats showed lesions at days 50 and 60 after tumor cell inoculation that were significantly smaller than those of the control group (Fig. 9b; p < 0.05). After 60 days, the average lytic lesion size for the animals of this group was 14 mm2 compared to 45.5 mm2 in the untreated control group and the overall average growth rate was 0.18 mm2 per day vs. 0.95 mm2 per day for the controls (days 30–60, respectively; Figs. 2 and 9b).
Our aim was to establish an animal model that is characterized by site-specific osteolytic lesions that can be used to study the course of bone metastasis and to examine the effects of any specific pharmacological intervention. To this aim, a technique was developed in which 1×105 MDA-MB-231GFP human breast cancer cells were inoculated into the superficial epigastric artery. As this vessel anastomoses with the iliac artery via the pudendoepigastric trunk,8 the blood flow into the region supplied by this vessel is still maintained after its ligation. As a result of this locoregional administration, osteolytic lesions were detected exclusively in the femur, tibia and fibula of the animals, and no further distant visceral or skeletal metastasis was observed.
Established ways for inducing skeletal lesions include injecting breast cancer cells intracardially, subcutaneously into the mammary fat pads or directly into long bones.7 Intracardial administration of breast cancer cells into the left ventricle of the heart leads to systemic metastases at different skeletal sites and visceral organs.9 Orthotopic implantation of tumor cells into the mammary glands usually results in a low incidence of bone metastasis.7 The lack in specific osteolytic metastasis after systemic administration or orthotopic implantation into the mammary fat pad prompted the generation of subclones of breast cancer cells, which display a predilection to bone.10, 11, 12 Another model for site-specific osteolytic tumor growth is the intraosseous injection of tumor cells;13 this technique however is associated with local damage of cortical bone and lacks the processes of tumor cell extravasation and invasion. Recently, a model has been described that is characterized by site-specific osteolytic lesions after injection of MDA-MB-231 cells directly into the femoral artery of nude rats.14 The disadvantage of this model is that the femoral artery has to be ligated, and spontaneous remission of lytic lesions occurred in up to 50% of animals at day 42 after tumor cell inoculation.
In our model, the development of osteolytic lesions can be observed up to at least 110 days after tumor cell inoculation. Immunohistochemical examinations demonstrated tumor cell transport in osseous vessels thus indicating an active metastatic process in control rats even at 4 months after tumor cell implantation. Additionally, the high take rate of 93% enables to study effects of therapeutic intervention at points of time before any discernible metastasis has occurred. Moreover, the advantages include that the model allows to probe for those metastatic steps that are relevant after tumor cells have got access to the circulation, including adhesion, invasion, growth and induction of lytic lesions in the skeleton. Additionally, the monocentric appearance of skeletal lesions in a limb precludes comorbidities such as cachexia and paralysis that are seen in models with disseminated metastases.9 Also, the GFP tag facilitating histological detection of single tumor cells and the relatively long experimental window for up to 3 months can be considered as benefits of the model. The disadvantage of our present model is the need for nude rats that are expensive and require special housing conditions. Also, the technique of tumor cell inoculation needs special equipment and some practice because the procedure has to be done with the aid of an operation microscope.
Radiographic assessment of osteolytic lesions was performed by X-rays and computed tomography. X-rays appeared to be more sensitive in detecting lytic lesions compared to computed tomography, especially when radiographs were made in a.p. and p.a. positions. Radiographs of the whole animal showed conclusively both, interim and at the end of the observation period, that osteolytic lesions were confined to that limb only that had been inoculated with MDA-MB-231GFP mammary carcinoma cells. Reconstructions of the rat's hind leg performed with computed tomography allowed 3-dimensional turning of the skeleton thus enabling views from every direction towards the site of cortical bone lysis. Slices made by computed tomography with a thickness of 0.5 mm through the entire animal at a variable height were useful to detect small cortical lesions. Disadvantageously, osteolytic lesions sparing cortical bone could hardly be detected with computed tomography and the assessment of bone scans is much more expensive and time consuming as compared to that of X-rays. Visualization of bone ultrastructure would require the use of a novel Volume CT or Micro CT scanner.15
Although much has been learned about the pathogenesis of bone metastasis, many important interactions and processes remain to be elucidated. For the aspect of tumor cell extravasation, Mastro et al.16 concluded that mechanical arrest of cancer cells is more common to occur in nonskeletal sites with capillaries of smaller diameters (e.g., lungs) than in the wide-channeled sinusoids of bone. For this reason, specific adhesive interactions between tumor cells and vascular surfaces of sinusoids are additionally needed to explain the high specificity of breast cancer cells to bone.16 Several vascular surface proteins as well as bone- and marrow-derived chemotactic factors like osteonectin, osteopontin and bone sialoprotein certainly play an important role in the pathogenesis of breast cancer bone metastasis.16 The glycoprotein BSP was detected at both protein and mRNA levels in various human breast cancer cell lines such as in MDA-MB-231 cells17, 18 and was also demonstrable by immunohistochemistry in breast and prostate cancer primary tumors and bone metastases.19, 20 Immunostaining for BSP was found to be higher in bone metastases than in visceral metastases of breast and prostate cancer patients.21 In 2 in vivo models, MDA-MB-231 cells overexpressing human BSP demonstrated higher transendothelial migration ability and enhanced subsequent bone metastasis, while repressed expression of BSP or antibodies against BSP inhibited these effects.22, 23 Recently, we were able to show that reduced expression of osteopontin and BSP in MDA-MB-231 cells decreased bone metastasis formation in nude rats distinctly after preexposure to antisense oligonucleotides.24
Here, we report a significant reduction of tumor growth in bone (81% reduction compared to untreated controls at day 60 after tumor implantation) after injection of MDA-MB-231GFP cells preincubated with an antibody against BSP. A similarly high effect was observed in animals receiving the anti-BSP early after tumor implantation. This is evidenced by a by 70% reduced lesion size at day 60 after tumor cell inoculation and a by 80% reduced daily tumor growth rate in comparison to untreated animals. A negative control antibody used for comparison did not exert any anti-lytic efficacy in nude rats implanted with MDA-MB-231GFP cells, nor in vitro (data not shown).
In the context of bone metastasis formation, the most promising binding partners to interact with BSP are the integrins αvβ3 and αvβ5, which have been found to induce adhesion, proliferation and migration responses in MDA-MB-231 cells after binding to BSP.25 The integrin αvβ3 is expressed by the MDA-MB-231 cell line and an αvβ3 overexpressing subclone of these cells showed increased invasion of and adhesion to mineralized bone, collagen and BSP.26 Interestingly, tumor cell integrin αvβ3 is able to mediate tumor cell arrest under dynamic flow conditions by interacting with its binding partners.27, 28 Lately, Fedarko and colleges reported about another specific interaction partner of BSP, the matrix metalloproteinase MMP-2, which is activated after its binding to the sialoprotein.29 Matrix metalloproteinases are a class of hydrolytic enzymes supposed to be important in tumor development and metastasis.30 Taken together, these findings as well as our results support the hypothesis that BSP has the potential to act as an adapter molecule in the pathogenesis of bone metastasis. Antibodies against BSP might prevent interactions between its binding partners such as integrins and matrix metalloproteinases in processes like tumor cell extravasation, adhesion to the target tissue and the formation of osteolytic lesions, thus reducing lytic lesion size in bone.
The dose response data obtained in vitro show that there is a linear decrease in cell growth with increasing antibody concentrations. This is in line with the effects of the antibody on colony formation and migration observed in vitro.31 The U-shaped curve observed in vivo could be consistent with the following considerations: Low concentrations of the antibody will primarily bind to free BSP, which is not attached to other molecules. Higher concentrations may be able to interact with BSP that is already linked to certain binding partners and we propose that the resulting alteration in BSP binding to the skeletal extracellular matrix is responsible for the observed therapeutic effect.
Reasons for the observed decrease in activity at very high antibody concentrations are unknown. Contributing factors could include the polyclonal nature of the antibody, unspecific binding to BSP interacting molecules, inactivation of the antibody by the host, or other still unknown mechanisms. We speculate that the use of a monoclonal antibody may eliminate this problem.
In conclusion, this animal model provides an interesting tool to study the course of metastatic bone disease and its pathogenesis in rats. Using a physiologically relevant approach through the circulation, the model reproducibly shows site-specific lesions which can be observed up to 110 days after tumor implantation. When applied to the effects of an antibody against BSP, the model provides in vivo evidence that pretreatment of MDA-MB-231GFP cells and treatment of animals after tumor implantation with the anti-BSP significantly reduces the size of lytic lesions in bone.
We thank Ms. B. Kahn for excellent technical guidance.