Background
Prostate cancer metastasis to bone leads to debilitating fractures and severe bone pain in men with advanced disease for which there is no treatment and is associated with poor prognosis and rapid decline [
1]. Recent studies have shown that 100% of men who die of prostate cancer have bone metastases [
2]. Paget's "seed and soil" hypothesis posits that the affinity that certain cancers have for bone may be due to a supportive microenvironment for tumor growth [
3,
4].
Androgen ablation therapy is standard-of-care for advanced prostate cancer, however, bone metastatic prostate cancer often becomes castration-resistant [
5]. Two treatments that target the bone microenvironment - bisphosphonates, eg. zolendronic acid, and the RANKL inhibitor, denosumab, which inhibit osteoclasts and osteolysis - have been effective in delaying the onset of skeletal related events (SREs) and new bone metastases in bone metastatic cancers with primarily osteolytic bone lesions [
6‐
11]. A characteristic of prostate cancer bone metastases, however, is that they typically produce osteoblastic or mixed osteoblastic/osteolytic bone lesions that are not as efficiently treated with the osteoclast inhibitors [
12‐
16]. There is currently no curative treatment for prostate cancer bone metastases [
1].
A major limitation in understanding and treating prostate cancer bone metastatic disease is that primary human prostate cancer bone metastasis tissues are rarely available for direct analysis or for the development of predictive model systems [
1,
2]. In addition, spontaneous bone metastasis of prostate cancer is a rare event in murine models of prostate cancer [
4,
17‐
19]. Direct injection of prostate cancer cells into the endosteal space of murine leg bones has, thus, provided a robust and reproducible method for studying the growth of prostate cancer in the bone-niche [
20]. Prostate cancer cell lines such as LAPC4, LNCaP, LuCAP23.1 and LuCAP35.1 - all of which originated from lymph node metastases - were directly injected into bone either intra-femorally or intra-tibially where they formed tumors and induced bone lesions [
4,
20,
21]. However, direct injection models into the bone-niche using prostate cancer cell lines that did not originate from patient bone metastases may not reflect physiological interactions. The C4-2B cell line is an improvement in this respect since it arose from the sub-cutaneously xenografted LNCaP tumor that spontaneously metastasized to bone within a SCID mouse and formed mixed osteoblastic/osteolytic lesions [
22,
23]. An intriguing alternative prostate cancer xenograft model assessed metastasis to adult human bone implanted in the hindlimbs of SCID mice [
24]. Prostate cancer cells from xenograft tumors homed to the human bone and induced osteolytic lesions but only at low frequencies [
24].
Direct bone-injection murine xenograft models using patient-derived bone metastatic prostate cancers, on the other hand, are both an orthotopic and highly tractable xenograft model system [
4,
20,
25‐
28]. In patients in whom the bone metastatic tumor is causing pathologic fractures, orthopedic surgery is performed to stabilize the bone and primary prostate cancer bone metastases may be collected for study at this time [
29,
30]. Currently, there are three prostate cancer bone metastasis-derived orthotopic bone xenograft models, PC3, LAPC9 and VCaP [
31,
32]. Xenograft transplantation of these cell lines into bone demonstrated the range of bone lesions produced by prostate cancer bone metastases: PC3 formed purely osteolytic lesions in intra-tibial xenografts [
25,
27,
28,
31], VCaP produced mixed osteoblastic/osteolytic lesions [
32], while LAPC9 formed purely osteoblastic lesions [
25‐
28,
31].
These models have led to important insights, however, it is crucial to expand on the limited number of existing prostate cancer bone metastasis-derived models in order to understand variability between different patient-derived tumors [
21,
33,
34]. Next-generation genomic DNA sequencing and RNASeq profiling of expression and splice isoforms have revealed significant molecular diversity and complexity of prostate cancers [
35‐
37]. In addition, the existing cell lines have been passaged
ex vivo for over a decade which has led to progressive alteration of the cell lines away from the original patient characteristics [
21,
22]. LAPC9 xenografts, for example, generated androgen-independent derivatives that progressed in castrated SCID mice after passaging in mice [
33,
34]. Genome-wide expression and integrative genomic profiling have comprehensively shown that there are differences between cell lines in vitro compared with primary patient tumors [
35‐
37]. Genome-wide analysis of DNA methylation patterns comparing normal prostate tissue to primary prostate cancer and cell lines revealed a complex picture with some methylation patterns consistently retained in prostate cancer tumors and cell lines while others were distinct [
38]. It is essential, therefore, to develop new patient-derived, orthotopic bone metastasis prostate cancer xenograft models that are closer to patients' original tumors especially for determining predictive therapy response profiles [
39‐
42]. In this report we describe the development and characterization of a new patient-derived bone metastatic prostate cancer femoral injection murine model, PCSD1.
Discussion
Prostate cancer progression is marked by metastasis to bone, resistance to androgen deprivation therapy, radiotherapy and chemotherapy as well as the emergence of an apoptosis-resistant, tumor-initiating population for which there is no effective therapy [
34,
60‐
66]. There is a pressing need for new models to investigate prostate cancer interaction with the bone microenvironment and to develop therapies but they have been difficult to establish due to poor take-rates of xenograft transplantation of primary prostate tumors [
17‐
19]. Here we describe PCSD1, a robust new patient bone metastasis-derived prostate cancer intra-femoral xenograft model for studying prostate metastatic bone disease. PCSD1 generated serially-transplantable sub-cutaneous and intra-femoral tumors when transplanted into immunodeficient
Rag2
-/-
;γ
c
-/-
male mice. PCSD1 xenograft tumors were characterized as PSA
+, AR
+, K5
-, K14
-, K8
+, K18
+, AMACR
+, NKX3.1
+, and TMPRSS2:ERG
- human prostate cancer. These biomarkers identified PCSD1 as an advanced luminal prostate cancer bone metastatic cancer [
43‐
47,
56,
63,
64]. MicroCT analyses revealed PCSD1 formed mixed osteoblastic and osteolytic lesions in a murine femoral injection model which closely resembled the bone lesions in the patient [
28,
31,
60].
The PCSD1 xenograft model will be used to understand the development of castrate-resistant prostate cancer in the bone microenvironment. Tumor growth of PCSD1 xenografts in intact versus surgically castrated mice is currently being measured. In culture, PCSD1 cells demonstrated androgen-independence as they can survive and proliferate without the addition on androgens.
Current standard-of-care therapies such as bisphosphonates, radiation, anti-androgens, chemotherapy, such as docetaxel, often eventually fail in patients who develop castrate-resistant prostate cancer [
1,
5,
15,
67‐
70]. The PCSD1 model will be used not only to elucidate mechanisms of failure of standard-of-care therapy but also to develop new therapies alone or in combination with current therapies.
The PCSD1 model will also be used to gain understanding of the unexpected, discordant effects of some new prostate cancer therapies that are being reported for bone metastatic prostate cancer. For example, in a Phase II Study of the new anti-androgen, Abiraterone, it was found that approximately one third of patients with chemotherapy-naive metastatic castration-resistant prostate cancer displayed bone scan flare discordant with PSA serologic response [
71]. In other words, many patients with significantly lowered PSA levels after treatment with abiraterone still showed positive bone scans [
71]. Conversely, some patients treated with the new c-Met tyrosine kinase inhibitor, Cabozantinib (c-Met TKI, XL184), showed dramatic reductions in positive bone scans but, paradoxically, no decrease in their PSA levels [
72]. New bone metastasis models such as PCSD1 are, therefore, essential to understand the complex mechanisms of interaction of prostate cancer with the bone microenvironment and the variation in response to therapies in different patients, types of bone lesions or stages of bone metastatic prostate cancer progression.
Acknowledgements
We especially thank Dr. Dennis Carson for critical reading of the manuscript, essential scientific input and support of the project, Dr. Nissi Varki for pathology expertise and Dr. Norman Greenberg for insight and advice on prostate cancer models. We thank the Moores Cancer Center Histology Core, Brian Crane for expertise in xenograft tumors, Alice Shih and Angela Court for help with breeding mice. We also thank Kim Wilson for help with manuscript preparation, and Jonathan M. Lee for help with preparation of figures.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
OR provided clinical expertise, selected, designed and performed RT-PCR and analyzed immunohistochemical images. AAK provided patient samples, clinical expertise in orthopedic oncology and interpretation of microCT scans. CW provided animal experiment expertise and performed intra-femoral injections assisted by HL. YBJ, performed RT-PCR and analyzed immunohistochemical images. KMS guided RT-PCR analyses and performed sequence confirmation. DG assisted with RT-PCR and primer design. TY and KM performed microCT scanning and analyses and generated 2D and 3D microCT images and movies; KM also performed scanning of intra-femoral PSA IHC slides. CHMJ provided expertise in generating primagraft and xenograft cancer models, bone marrow niche analysis as well as Rag2-/-;γc-/- mice. SM provided clinical expertise and statistical analyses. NAC, provided molecular biology expertise and was involved in data analysis, biomarker selection and provided signal transduction expertise. CJK provided primary prostate tissue and tumor specimens, contributed to writing the manuscript and clinical expertise in prostate cancer. CAMJ is the PI of the study, wrote the manuscript, guided all aspects of generating the model and analysis. In addition, CAMJ performed patient sample preparation, xenograft tumor dissection and cell preparation for IF injections, RNA, DNA purification, tumor fixation, decalcification and mounting for sectioning and cryopreservation of tumor cells and sub-cutaneous injections. All authors have read and approved the final manuscript.