Background
About 90% patients with advanced breast and prostate cancer have incurable bone metastases [
1,
2], with a mean survival of one year [
3,
4]. Treatments that aim to either prevent or suppress the growth of bone metastasis are limited. Anti-bone resorptive agents such as zoledronic acid, prevent bone loss and skeletal-related events (SREs), but do not increase survival in unselected patients with advanced disease [
4‐
7]. However, a better understanding of the early mechanisms leading to the development of bone metastasis would allow such therapies to be specifically targeted to early lesions as they arise.
The ‘seed and soil hypothesis’ for metastasis proposes that only the right cells disseminated from primary tumour (the ‘seeds’) landing in the right microenvironment (the ‘soil’) can form secondary (metastatic) tumours successfully [
8]. It has been suggested that prostate and breast cancers contain a small percentage (<1%) of stem cell-like populations and that it is these cells that have the potential to act as the ‘seeds’, leaving the primary site, surviving in the circulation, homing into the bone, where they eventually form bone metastases [
9‐
14]. In this context, bone is the fertile ‘soil’, providing a supportive microenvironment (‘bone metastasis niche’) for tumour cell survival and secondly, under the right conditions, for the proliferation of clinically relevant metastases. New evidence has shown that disseminated prostate cancer cells target the haemopoietic stem cell niche to establish footholds in bone, with the osteoblastic lineages as the key components of the niche [
15,
16]. In addition, bone continuously renews itself with precisely balanced osteoblastic bone formation and osteoclastic bone resorption. Elevated levels of bone turnover have been shown to be correlate with increased numbers of metastases, in various xenograft models and clinical studies [
17‐
22]. Whether this effect is a result of different bone environments increasing tumour cell seeding, or of environments inducing proliferation of resident tumour cells in bone to form overt metastases, remains to be determined.
In this study, we investigated the initiation of prostate and breast cancer bone metastases in young (6-week old) and in mature (16-week old) athymic mice. Both bone turnover rates and the frequency of metastases were lower in the mature compared to the younger animals allowing us to study the initiation of metastases in different bone microenvironments. We used two-photon microscopy to identify initial seeding of tumour cells into bone and compared this with the frequency of growing metastases. This is the first study to directly test the hypothesis that the frequency of bone metastasis is equally dependent on numbers of seeded tumour cells and growth induction by the bone microenvironment.
Methods
Mice
All studies were performed using 6-week or 16-week old BALB/cAnNCrl immunocompromised (athymic nude) mice (Charles River, Kent, UK) as xenograft models of bone metastasis. All procedures complied with the UK Animals (Scientific Procedures) Act 1986 and were reviewed and approved by the local Research Ethics Committees of the University of Sheffield under Home Office project licence 40/3462 (Sheffield, UK).
Cell lines
The human prostate cancer cell line PC3 (ATCC, Middlesex, UK) and human breast cancer cell line MDA-MB-231 (ATCC, Middlesex, UK) were stably transfected with a firefly luciferase gene luc2 (pGL4.51 [luc2/CMV/Neo] vector, Promega, Southampton, UK) and denoted as PC3-NW1 and MDA-MB-231 luc2. Both cell lines were maintained in Dulbecco’s Modified Eagle Medium (DMEM) (Life Technologies, Paisley, UK), supplemented with 100 Units/mL Penicillin, 100 μg/mL Streptomycin and 10% foetal calf serum (FCS) (Sigma Aldrich Co Ltd, Poole, UK).
Xenograft models
Human cancer cells were initially stained with 5 μM lipophilic carbocyanine dye Vybrant DiD (Life Technologies, Paisley, UK) according to the manufacture’s protocol. Using the Vybrant DiD dye, which is lost as cells divide, allows non-dividing/slowly dividing cells to be identified and distinguished from proliferating cells [
15,
23,
24]. A single-cell suspension of 1 × 10
5 DiD labelled PC3-NW1 cells/100μL PBS and 0.75x10
5 DiD labelled MDA-MB-231 luc2 cells/100μL PBS were injected into the left cardiac ventricle (intracardiac (i.c.) injection) of 6- or 16-week old male and female BALB/c nude mice, respectively. Overt tumours were monitored up to 8 weeks post injection using an in vivo imaging system (IVIS, PerkinElmer, Cambridge, UK) and the tumour burden was measured based on radiance of luminescence using the Living Image software (PerkinElmer), followed by anatomical confirmation of tumour metastatic sites post mortem. Cohorts of animals (minimum
n = 6/group) were euthanized on day 1, 7 and 21 post injection and individual non-dividing DiD labelled tumour cells were quantified in the tibiae
ex vivo by two-photon microscopy, to understand the quantity of tumour cells arriving and colonizing within the bone marrow.
Two-photon microscopy
Dissected right tibiae were prepared and sectioned as described previously [
15,
24]. An area of 2104 μm × 2525 μm below the growth plate with 100 μm in depth was imaged using a Zeiss LSM510 NLO upright multiphoton/confocal microscope (Carl Zeiss Inl, Cambridge, UK). A 633nm HeNe laser was used to detect DiD labelled cells while the bone was detected using the 900nm Chameleon multiphoton laser. The number of tumour cells (DiD positive events) and their distance to the nearest bone surface (a parameter indicating the location of tumour cells towards osteoblastic niche) was analysed using the Volocity 3D Image Analysis software 6.01 (PerkinElmer, Cambridge, UK).
Micro-CT analysis
Right femurs were dissected and scanned by SkyScan 1172 desktop micro-CT (SkyScan) at the resolution of 6 μm. Trabecular morphometry, characterized as trabecular bone content (BV/TV) was measured from a 1.0mm thick region 0.2mm above the growth plate where metastatic tumour cells are generally situated. Nomenclature and symbols were used to describe the micro-CT derived bone morphometries according to the published guidelines [
25].
Serum bone remodelling markers
Tartrate-resistant acid phosphatase 5b (TRAP) activity in mouse serum was determined as a measure of bone resorption activity using an IDS MouseTRAP™ Assay (Immunodiagnostic Systems, Tyne & Wear, UK). Type 1 procollagen amino-terminal-propeptide (P1NP) and osteocalcin were determined for bone formation activities, using a Rat/Mouse P1NP competitive immunoassay kit (Immunodiagnostic Systems, Tyne & Wear, UK) and a Mouse Osteocalcin ELISA Kits (Takara Bio Europe, Saint Germain, France), for male and female mice samples, respectively.
Bone histomorphometry
Dissected left tibiae were prepared and TRAP stained as described previously [
26]. The number of osteoblasts (N.Ob/B.Pm), the bone surface covered by osteoblasts (Ob.Pm/B.Pm), the number of osteoclasts (N.Oc/B.Pm), and the bone surface covered by osteoclasts (Oc.Pm/B.Pm) were determined on a 1.5mm length of endocortical surfaces, using a DMRB microscope (Leica Microsystems, Wetzlar, Germany) with the Osteomeasure bone histomorphometry software (OsteoMetrics, Inc. Decatur, GA, USA).
Statistic analysis
All data are expressed as mean ± SEM. Statistical significance was tested for using an unpaired Student’s t-test using the Prism 6 software (GraphPad, La Jolla, USA). P < 0.05 was considered to be significant.
Discussion
The ‘Seeds and soils’ hypothesis suggests that both the metastasis initiating tumour cells and the microenvironment are equally important for the initiation of metastases [
8]. To test this, we used different aged (6-week and 16-week old) immunocompromised mice as xenograft models to investigate the relationship between numbers of tumour cells colonizing bone, bone turnover, and the incidence of osteolytic skeletal metastases caused by prostate and breast cancer.
In the prostate cancer model, bioluminescent signals were clearly visible in the skeleton, three weeks after initial tumour cell injection. As bioluminescence detects bone marrow metastases of ≈ 0.5mm
3 volume [
27], the signals detected were indicative of overt, growing lesions. These lesions were associated with bone destruction that was confirmed by anatomical confirmation and micro-CT analysis post mortem. The results showed that both the frequency of skeletal metastases and tumour burden were higher in young mice compare to mature mice. However, there were no differences in the incidence of metastases in soft tissues. This indicates that the difference in the frequency of metastases across different aged cohorts is unique to the skeletal system and determined by the bone microenvironment. Further examination of the bone phenotype showed that young male mice had lower bone mass compared to mature animals but higher bone turnover. This was characterised by significant increases in the numbers of osteoblasts and osteoclasts, and elevated bone formation and resorption activities. This is consistent with clinical studies which suggested higher levels of bone turnover predict poor outcome in patients with bone metastasis [
20,
22,
28]. To determine whether the higher metastasis frequency is due to higher numbers of metastasis initiating cells taking up residency in bone, we quantified the numbers of prostate cancer cells present in the bone marrow across a three-week period after initial injection, using two-photon microscopy and cell membrane labelling techniques. The results suggested that there were no increase in prostate cancer cells arriving in the bone marrow of young mice compared to mature animals and that these cells were in direct contact or one cell layer away from bone surface osteoblast lineage cells as they were within a range of 50 μm to the nearest bone surface on which an osteoblast has a typical diameter of 15-30 μm [
29]. Surprisingly, there were significantly more tumour cells detected in the bone marrow of mature mice on day 1 and at 3 weeks post injection. This warrants further investigation. One explanation is that fewer DiD labelled tumour cells were detected in younger mice as a result of more proliferation and consequent loss of dye [
24]. Another possibility is that the low bone turnover micro-environment in mature mice is more selective for stem cell-like cancer cells, while high bone turnover micro-environment in younger mice are more supportive to the growth of metastases and may correlated with angiogenesis which will be discussed in following sections. These data suggest that the higher frequency of skeletal metastases for prostate cancer in younger xenograft models is not due to increased tumour cells homing to the bone marrow but supports the hypothesis that it is the bone microenvironment that controls the frequency of growing metastases detected in these xenograft models, through induction of lesion initiation and growth [
30,
31].
Similar results were found in the breast cancer model. Significantly higher frequencies and levels of tumour burden, determined by bioluminescence and anatomical confirmation post-mortem, were found in young female mice compared to mature mice. This was not due to higher number of breast cancer cells arriving into the bone marrow of younger animals, but correlated with the enhanced bone turnover activities confirmed by both ELISA of serum bone turnover marker (osteocalcin and TRAP) and histomorphometric analysis of osteoblasts and osteoclasts. Once again, there was a trend towards more tumour cells being detected in bone marrow of older female mice.
Taken together, the differences in frequency of skeletal metastases between young and mature mice in both prostate and breast cancer models strongly suggested that it is the trigger for the proliferation of resident tumour cells in bone that is linked to the differences in bone turnover in the bone microenvironment. This is consistent with previous findings of our own and of others that showed increased frequencies of bone metastases by prostate and breast cancer xenograft models in mature animals with experimentally induced enhancement of bone turnover [
17,
18,
32]. More importantly, these data correlate with disease outcomes in cancer patients: Breast cancer patients presenting at a younger age (<40) have worse survival rates and more metastases compared to disease presented in older women, while younger men (<44) with high grade prostate cancer, have poorer prognosis compared to older men with a similar grade/stage distribution [
33,
34]. One could argue these differences are due to higher available sex steroid hormone levels rather than the higher bone turnover in younger populations, directly affecting tumour growth. However, our data suggests another mechanism, as we intentionally used sex steroid hormone-independent cell lines: PC3 and MDA-MB-231 in these studies that are unaffected by androgens/oestrogens.
The xenograft model used in this study has limitations: firstly, we could not determine the role of the immune system, specifically T cells, in regulating metastatic frequency. The latter have been implicated in the bone metastatic process [
35]. Secondly, the administration of a single high dose of cancer cells via intracardiac injection only partially mirrors seeding events in humans where low numbers of cancer cells are released into the circulation over prolonged time intervals. Finally, other differences between young and mature mice could also contribute to variations in metastatic frequencies such as the higher growth hormone levels and continuous longitudinal bone growth in younger mice. Particularly, differences in endothelial lineage populations and associated niches could also affect metastasis frequency and should be investigated [
16]. Tumour cells were shown to be rapidly engrafted into bone marrow endothelial microdomains, where endothelial cells express higher level of adhesion molecules such as E-selectin, P-selectin, and intercellular adhesion molecule (ICAM-1) [
36‐
38]. Kusumbe et al. recently showed that there are two distinct bone microvessels based on the expression of CD31 (also known as PECAM1) and Endomucin (Emcn), which were proposed as type H subset (CD31
hiEmcn
hi) and type L subset (CD31
loEmcn
lo) [
39]. The type H subset expresses higher level of metastasis related growth factor transcripts such as
Tgfb1 and
Tgfb3 [
40]. The type H subset is a key to couple angiogenesis to osteogenesis and was strongly reduced in bone from aged animals [
39]. This study suggested that the correlation of higher incidence of skeletal tumours in younger animals with the presence of type H microvessels in addition to higher bone turnover.
In conclusion, our study using two established metastatic cancer cell lines, provides further evidence to support the hypothesis that the frequency of overt skeletal metastases in hormone-independent prostate and breast cancer is determined by growth initiating influences within the bone microenvironment not by numbers of tumour cell initially seeding to these sites and provides further grounds for the use of treatments to suppress bone turnover and prevent development of skeletal metastases of prostate and breast cancer at early stages of disease.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
Study design: NW, KR, PC, IH, and CE. Study conduct: NW, AF, KR. Data acquisition: NW, FD, AF. Data analysis: NW, HB, and CE. Data interpretation: NW, PO, PC, IH and CE. Drafting manuscript: NW and CE. Revising manuscript content: NW, HB, PO, PC, IH, and CE. Approving final version of manuscript: PC, IH, and CE. NW and CE take responsibility for the integrity of the data analysis. All authors read and approved the final manuscript.