Successful metastasis requires some degree of hospitality from the host organ that is metastasized. In recent years, evidence has been mounting that the primary tumor contributes to adapting this remote organ to increase this hospitality. Interestingly, many in vitro experiments have shown that conditioned media from cancer cells produce physical changes which theoretically could facilitate invasion. These include perturbations in extracellular matrix and of cell to cell adhesions. The conditioned media in these in vitro experiments could be interpreted to be a proxy for either paracrine or endocrine effects, however, when these observations are coupled with evidence from in vivo experiments, it seems possible that primary tumors have remote effects on un-metastasized organs that can elicit changes in them which are pro-metastatic. We have performed a series of experiments designed to characterize and describe changes in osteoblasts (hFOB) which occur as a result of exposure to conditioned medium from breast cancer cells (MDA-MET). Functional experiments aimed at investigating whether these changes are pro-metastatic were included as a means of this characterization. Thus cell to cell adhesions, as well as the ability of the osteoblastic environment to attract MDAMET cells were both assessed. Further, we investigated mechanistically the bases behind the observations from the functional experiments.

Methods:​ Confluent hFOB cells were treated with conditioned media from either MDA-MET, MDA-MB-231 HTERT-HME1 or hFOB cells 24 hrs. This treatment conditioned medium was removed, the cells washed and serum-free medium added to the hFOB cells. hFOB-conditioned medium was collected after 18hrs. This medium was then used in in vitro migration assays measuring number of MDA-MET migratory cells by labeling the cells with DNA-binding dye Cyquant;​ in establishing presence of collagen by western blot;​ in collagenase assays and in a cytokine array where antibodies to 74 cytokines were embedded on a membrane. The presence of Type 1 collagen receptors was shown by immunocytochemistry. The data were analyzed using one way ANOVA and the Student’s T test.

Results:​ We found that conditioned medium from hFOB cells that had been treated with MDA-METconditioned medium attracted more MDA-MET cells than hFOB cells pre-exposed to its own medium only. We hypothesized that Type 1 collagen fragments of specific length were the chemoattractants responsible for our observed effect. This was evidenced by an increase in collagenase in the conditioned medium of hFOB cells which had been exposed to MDA-METconditioned medium. The definitiveness of this evidence was aided by the inherent fidelity of the collagenase enzymes and its Type 1 collagen substrate and the distinctiveness of the product of this enzyme substrate interaction. We also showed that bacterial collagenase, which creates short collagen fragments of non-specific length removed the ability of hFOB - conditioned medium to attract MDA-MET cells. In addition, we showed that at least two known Type 1 collagen receptors UPARAP (urokinase-like plasminogen activator associated protein) and integrin receptor α2β1 were present on MDA-MET cells. The cytokine array performed on hFOB -conditioned medium showed differential protein expression under treatment by MDAMET compared to HTERT-HME1 in one cytokine only (CXCL5). Therefore, MDA-METconditioned medium treatment caused a loss of expression of CXCL5 while HTERT-HME1- conditioned medium exposure caused an increase of CXCL5. We confirmed this result by quantitative RT-PCR on both CXCL5 and on our proposed metastasis suppressor –like candidate pigment-epithelium-derived factor (PEDF). We showed that this loss of CXCL5 expression occurs both in hFOB cells and on the protein level in conditioned medium. PEDF followed the differential expression of CXCL5;​ that is increasing in response to HTERT-HME1- conditioned medium and decreasing in response to MDA-MET-conditioned medium. We investigated whether the knockdown of CXCL5 by siRNA in hFOB cells would decrease chemoattraction of MDA-MET cells. We also added recombinant PEDF to hFOB-conditioned medium to test whether chemoattraction could be decreased. The knockdown of CXCL5 in hFOB cells had no effect (Figure 5) while the addition of PEDF, had a trend towards a decrease in chemoattraction n=5, p<0.1 (Figure 4).

Conclusions:​ We have further defined the concept of the pre-metastatic niche. This particular niche created by secreted factors from breast cancer cells in bone may 1) include chemoattractants, generated by the effect on the extracellular matrix of exposure to secreted factors from primary tumor cells. We show evidence that collagen fragments of Type 1 collagen are a likely candidate responsible for this observation. 2) Include the loss of anti-metastatic factors such as PEDF and CXCL5 by osteoblasts 3) be further characterized by a decrease in the ability of osteoblastic cells to form cell-cell adhesions.