Secreted protein acidic and rich in cysteine (SPARC) belongs to the matricellular group of proteins, that are transiently secreted to the Extracellular matrix (ECM) without becoming part of the ECM mesh. Acting at the interface between the cell surface and the ECM, matricellular proteins regulate cellular processes such as adhesion, migration, proliferation and differentiation [1,2,3]. SPARC, also known as basement membrane-40 (BM-40) tumour protein or osteonectin, is a 32–35 kDa multifunctional ECM protein, with affinity for collagens and calcium ions; a high molecular weight form in the bone (43 kDa) is due to glycosylation [4].

The role of SPARC in tumorigenesis is controversial, and some investigations report a positive relationship between elevated levels of SPARC and more aggressive cancers, with poorer clinical outcome, even if this largely depends on the tumour type [5,6]. While normal mammary tissue has undetectable or lightly detectable amounts of SPARC, and benign breast lesions are weakly positive, 75% of both in situ and invasive breast carcinomas are strongly positive for SPARC in stromal cells (CD-34-negative, α-SMA-positive) [7,8]. Moreover, SPARC is associated with clinicopathological parameters such as histological differentiation, tumour size, depth of invasion, lymph node metastasis, distant metastasis, and Tumor-Node-Metastasis (TNM) stage [6].

Metastatic occurrence in the skeleton, one of the preferential secondary sites, is the leading cause of death for breast carcinoma, and bone metastases develop also ten years after surgical removal of the primary tumour, since osteotropic malignant cells may have a long period of quiescence before secondary outgrowth [9]. Complications of bone metastases include pain, increased risk of fracture, hypercalcemia and decreased levels of blood cell count. Having reached the bone, malignant cells disrupt the remodelling process that normally occurs. Osteoclast number increases, stimulating bone resorption and the release of factors important for metastasis outgrowth, and osteoblast number might decrease in case of osteolytic bone metastasis while increasing in the osteoblastic or mixed type [3,9].

Improved knowledge of the biochemical processes underlying bone metastasis, has prompted investigation whether skeletal events in patients may correlate with levels of serum or urine markers of bone turnover, thus facilitating early detection and screening for such events. The prevention and the development of therapeutic strategies against bone metastasis implicate the understanding of the carcinoma cell preference to metastasize in the bone tissue. The bone microenvironment may provide growth stimulating factors or other proteins that induce preferential adhesion and proliferation. Based on transgenic models of breast cancer, SPARC expression could be used as a potential prognostic biomarker of tumour severity and/or aggressiveness [10].

As already mentioned, the survival rate in the case of patients with bone metastasis from breast carcinoma is very low. The understanding of molecular mechanisms of organotropism would help the prevention of metastasis, and the establishment of a valid therapy in order to improve the patient’s quality of life and to increase survival. Until now there are only few studies regarding breast carcinoma that tried to disclose the roles of SPARC in bone metastasis.

We have demonstrated a favouring function of endogenous SPARC for bone metastasis from breast cancer using a xenograft model, prepared with clone 1833 derived from MDA-MB231 human breast carcinoma cells. SPARC was induced by Endothelin 1 in 1833 cells with bone tropism, but not in parental cells, and the expression of the two biological stimuli is orchestrated by DNA methyltransferases. Of note, SPARC and Endothelin 1 are expressed in the bone marrow of the bone metastasis xenograft model [11].

The bone marrow is a common homing organ, and reservoir for disseminated tumour cells from breast cancer. Although only a small fraction of patients present overt bone marrow metastasis, 70% of advanced breast cancer patients develop metastasis in the bone [12]. Bone marrow is a source of mesenchymal stem cells that can give rise to cells of mesodermal lineages, such as osteocytes [13]. These mesenchymal stem cells play an important pro-tumorigenic role in the microenvironment of bone metastasis [14], and are also known as bone marrow stromal fibroblasts expressing SPARC/osteonectin [15], that home to sites of tumorigenesis and integrate in primary tumour stroma [13].

Recent research is devoted to understand critical events in breast cancer progression to make a molecular classification. Until now the breast carcinomas are classified as Basal and Luminal, and the latter in A and B principally on Ki67 value [16]. Ki67 is a molecular marker expressed in all cell-cycle phases except G0; the optimal Ki67 level discriminating Luminal A from B is 13.2% [17].

Our hypothesis was that SPARC might be a key molecular signal released from primary breast carcinoma and/or locally produced in bone metastasis to influence the bone niche formation in the secondary site. It has been proposed that bone colonization is initiated in a microenvironment niche exhibiting active osteogenesis, and heterotypic adherens junctions take place between metastatic and osteogenic cells [18]. In this context, the present paper deals with the evaluation of SPARC expression during ductal breast carcinoma progression, from the dysplastic lesion until bone metastasis establishment: SPARC distribution in neoplastic cells versus stroma, and the plasma levels were examined. Concomitantly, the expression of Endothelin 1 and of its receptor ET_AR were assayed to clarify whether the Endothelin 1/ET_AR pattern completed SPARC findings as predictive value for bone metastasis. Activation of autocrine and paracrine signalling by Endothelin 1 binding to its receptors elicits pleiotropic effects on tumour cells and on the host microenvironment for immunomodulation and intravasation [19]. Also, Endothelin 1 and ET_AR are expressed in breast carcinoma metastasizing to bone [20], even if the predictive value of this axis has not been clarified until now.

The aim of the present paper was to individuate molecular events to extend the classification of invasive ductal carcinoma, adding predictive markers for the outcome of breast cancer patients related to the development of bone metastasis.

The results indicate that the biological function of SPARC depended on the phase of breast cancer progression, the tissue compartment in which it was expressed, and its body distribution, with plasma release of SPARC being opposite to metastasis engraftment. SPARC was highly expressed in the stroma of bone metastasis, suggesting a positive effect on the outgrowth also due to a cross talk with metastatic cells expressing SPARC. The association of SPARC with Endothelin 1 and ET_AR—starting from dysplasia until the end of the metastatic process—appears to be of fundamental importance as a predictive index of bone metastasis formation.

Here we show that elucidating the molecular basis of ductal breast carcinoma progression might lead to effective ways to predict the bone colonization of metastasis, and to make a valid molecular classification important to devise a preventive therapy.

The ectopic secretion of proteins, known to be microenvironmental stimuli like SPARC and Endothelin 1, appeared to confer osteogenic traits to disseminated breast carcinoma cells, important for the attachment to bone matrix [7,22]. We demonstrated for the first time in bone metastasis of humans that at the site of colonization, SPARC was expressed by metastatic cells as well as by osteoblasts, which were identified on the basis of osteocalcin immunostaining. We suggest that osteoblasts were involved not only in osteogenesis, but also in the recruitment of osteoclasts [22], maintaining an elevated bone turnover with a consequent release of Endothelin 1 in the microenvironment. Thus, the induction of SPARC by Endothelin 1 was likely to occur also in metastatic cells of patients, as reported before for 1833 clone [11]. In addition, SPARC is a marker of differentiation of osteogenic bone cells [22]. Altogether, the events reported here might be critical for microscopic bone metastasis to progress into clinically significant lesions.

Our data are consistent with those obtained by Wang et al. with an experimental model [18], in which he demonstrates that bone colonization is initiated by a microenvironment niche exhibiting active osteogenesis. We hypothesize that bone niche evolution requires SPARC. Since human bone metastasis from ductal breast carcinoma shows E-cadherin expression [24], an heterotypic interaction with osteogenic cells would occur, as reported in a human/mouse experimental system, giving a proliferative advantage [18]. We cannot exclude that the observed high SPARC signal in supportive cells of bone metastasis conferred a proliferative advantage to metastatic cells [25], and that the latter producing SPARC competed with stem cell fibroblasts for the niche becoming hospitable [26].

The strength of our study in patients is the analysis of pair-matched invasive ductal breast carcinoma from the early preneoplastic stages until bone metastasis formation. The function of SPARC seems to depend on the microenvironment, in terms of local composition of matrix molecules, cytokines and protease profile, and on whether it is expressed by the malignant cells themselves or by neighboring stromal cells [3].

Our principal finding useful for a molecular classification was the very high expression of SPARC (70% incidence, as bone metastasis) in dysplastic epithelium, associated with ductal breast carcinoma, and in bone metastatic cells and stroma (ECM and supportive cells). Since SPARC is able to organize collagen and other ECM proteins, we suggest that in patients it would be involved in creating a fibrotic environment favorable for colonization, similar to lysyl oxidase (LOX) [27,28]. However, for bone metastasis predictive value, the association of SPARC signal with those of Endothelin 1/ET_AR seemed essential, starting from their presence in the dysplastic lesions. It is known that cancer cells may disseminate very early, even before patients present with cancer, and might form micrometastasis in a permissive environment of the secondary site such as that of bone. Our hypothesis is that a local production of SPARC in the hospitable bone, possibly regulated by Endothelin 1/ET_AR axis, would be important for bone niche formation and evolution. Notably, the plasma level of SPARC was not elevated in breast cancer patients with bone metastasis outcome.

Moreover, stem cell fibroblasts seem to migrate to primary carcinoma [13], and we suggest that they were responsible for the release of SPARC to tumor cells. In human specimens, breast carcinoma cells and stroma presented SPARC signals, in agreement with the literature [29]. A recycling of SPARC would occur in the primary breast carcinoma, and a role of exosomes cannot be excluded since these microvescicles can modulate nearby or distant target cells by direct contact of their surface molecules, to activate intracellular pathways. Alternatively, on internalization by membrane fusion or endocytosis, exosomes deliver their protein content [30]. Studies are in progress to evaluate this mechanism, since the study with MDA-MB231 cells indicates that in vitro breast carcinoma scarcely produces the matricellular protein SPARC [11]. Consistently, primary breast carcinoma that did not give bone metastasis abundantly release SPARC in the plasma, likely preventing, therefore, the engraftment of disseminated cells in the secondary bone site. In primary tumor, SPARC was likely to exert counter-adhesive roles important for the intermediate state of adhesiveness, and responsible for motility and invasiveness. Except for one, all of these breast carcinoma bearing patients are still alive 60 months after surgery. Circulating tumor cells shed from breast carcinoma, are a heterogeneous population including EpCAM-positive cells and cells with mesenchymal phenotype (EMT), which are studied for the analysis of the therapeutic response to different chemotherapeutic regimens [31].

In the future personalized medicine, the molecular classification based on SPARC associated with Endothelin 1/ET_AR would be of great importance to devise therapies against bone metastasis with nanoparticles, that target drugs with a SPARC-dependent endosomal mechanism [32].