Bone Cell Activity in Clinical Prostate Cancer Bone Metastasis and Its Inverse Relation to Tumor Cell Androgen Receptor Activity

Abstract

Advanced prostate cancer frequently metastasizes to bone and induces a mixed osteoblastic/osteolytic bone response. Standard treatment for metastatic prostate cancer is androgen-deprivation therapy (ADT) that also affects bone biology. Treatment options for patients relapsing after ADT are limited, particularly in cases where castration-resistance does not depend on androgen receptor (AR) activity. Patients with non-AR driven metastases may, however, benefit from therapies targeting the tumor microenvironment. Therefore, the current study specifically investigated bone cell activity in clinical bone metastases in relation to tumor cell AR activity, in order to gain novel insight into biological heterogeneities of possible importance for patient stratification into bone-targeting therapies. Metastasis tissue obtained from treatment-naïve (n = 11) and castration-resistant (n = 28) patients was characterized using whole-genome expression analysis followed by multivariate modeling, functional enrichment analysis, and histological evaluation. Bone cell activity was analyzed by measuring expression levels of predefined marker genes representing osteoclasts (ACP5, CTSK, MMP9), osteoblasts (ALPL, BGLAP, RUNX2) and osteocytes (SOST). Principal component analysis indicated a positive correlation between osteoblast and osteoclast activity and a high variability in bone cell activity between different metastases. Immunohistochemistry verified a positive correlation between runt-related transcription factor 2 (RUNX2) positive osteoblasts and tartrate-resistant acid phosphatase (TRAP, encoded by ACP5) positive osteoclasts lining the metastatic bone surface. No difference in bone cell activity was seen between treatment-naïve and castration-resistant patients. Importantly, bone cell activity was inversely correlated to tumor cell AR activity (measured as AR, FOXA1, HOXB13, KLK2, KLK3, NKX3-1, STEAP2, and TMPRSS2 expression) and to patient serum prostate-specific antigen (PSA) levels. Functional enrichment analysis indicated high bone morphogenetic protein (BMP) signaling in metastases with high bone cell activity and low tumor cell AR activity. This was confirmed by BMP4 immunoreactivity in tumor cells of metastases with ongoing bone formation, as determined by histological evaluation of van Gieson-stained sections. In conclusion, the inverse relation observed between bone cell activity and tumor cell AR activity in prostate cancer bone metastasis may be of importance for patient response to AR and/or bone targeting therapies, but needs to be evaluated in clinical settings in relation to serum markers for bone remodeling, radiography and patient response to therapy. The importance of BMP signaling in the development of sclerotic metastasis lesions deserves further exploration.

Keywords: BMP; androgen receptor; bone; metastasis; osteoblast; osteoclast; prostate cancer.

Figure 1

Principal component analysis of seven assigned gene products associated with bone cell activity in 28 bone metastases samples. (A) score plot for the first principal component. Each dot corresponds to one metastasis sample collected from prostate cancer patients. Bone content was determined within the range of 10–25% or 25–50% by histological examination of tissue sections; (B) loading plot showing included gene probes. Positive loading values (p) represent genes expressed at high levels in samples with positive score values (t) and vice versa.

Figure 2

Percentage of bone surface lined by tartrate-resistant acid phosphatase (TRAP)-positive and runt-related transcription factor 2 (RUNX2)-positive cells, respectively. Each bar represents TRAP-positive bone surface in one patient sample and the corresponding dot represent the RUNX2-positive surface in the same patient. Castration-resistant prostate cancer patients are represented by grey bars and treatment-naïve patients by black bars. Patients who had undergone chemotherapy are denoted by asterisks.

Figure 3

Representative immunostaining of runt-related transcription factor 2 (A,B), tartrate-resistant acid phosphatase (C,D) and van Gieson (E,F) in metastatic lesions with high (A,C,E) and low (B,D,F) bone cell activity. The area with newly formed bone is indicated with an arrow and magnified (E). Bar indicates 200 µm.

Figure 3

Representative immunostaining of runt-related transcription factor 2 (A,B), tartrate-resistant acid phosphatase (C,D) and van Gieson (E,F) in metastatic lesions with high (A,C,E) and low (B,D,F) bone cell activity. The area with newly formed bone is indicated with an arrow and magnified (E). Bar indicates 200 µm.

Figure 4

Orthogonal projections to latent structures analysis extracting the maximal variation in transcriptome data that co-vary with bone cell activity described by the weighted gene expression of ACP5, ALPL, BGLAP, CTSK, MMP9, RUNX2 and SOST t[1] score vector. (A) score plot of 28 bone metastases samples where each dot corresponds to one metastasis sample collected from prostate cancer patients; (B) loading plot of 13847 analyzed gene products. Genes with positive loading (p) are highly expressed in metastases with high bone cell activity and vice versa.

Figure 5

Orthogonal projections to latent structures discriminant analysis of MetaCore suggested upstream regulators of bone cells analyzed in relation to bone formation in metastasis samples. Class membership was set as ongoing or no ongoing bone formation, based on histological examination of tissue section. (A) score plot of 26 bone metastases samples where each dot corresponds to one metastasis sample collected from prostate cancer patients. The label (0–5) describes gradually increasing osteocyte rich tissue areas; (B) loading plot of 231 gene products suggested to regulate bone cell activity in metastasis samples. Genes with positive loading (p) are highly expressed in patient class with ongoing bone formation and vice versa. Gene products involved in bone morphogenetic protein and/or transforming growth factor β signaling are indicated.

Figure 6

Representative sections of bone metastasis sections showing bone morphogenetic protein 4 (BMP4) immunostaining of metastatic tumor cells scored as negative (0) in (A); weak (1) in (B); moderate (2) in (C) and intense (3) in (D), with a distribution between 0 and 4 (not shown) giving immunoreactive scores (IR scores) in the range of 0–12. Bone metastases with ongoing bone formation had significantly higher BMP4 mRNA levels (E) and BMP4 IR scores (F). Bar indicates 200 µm. Circles indicate outliers and extreme values.