The majority of breast cancers (95% or more) are of epithelial origin, arising either from the ductal or from the lobular epithelium [1]. Through histopathological diagnosis both types can be distinguished as in situ or invasive carcinomas. The in-situ carcinomas grow within the inner border of the epithelium and are not aggressive, unlike the invasive ones that penetrate the tissue barriers and tend to form metastases. However, within these two main groups, histological and intra-tumoral heterogeneities have been identified and classified as breast carcinoma subtypes having different clinical outcomes and response to therapy.

Besides the classical histopathology, other methods to classify breast tumors are applied; these include molecular and genetic analysis, clinical pathology and gene expression profiling. However, at present, no adequate markers are available to depict a portrait of bonemetastatic breast cancer cells.

The metastatic process begins when the tumor cells of epithelial derivation detach from the original tissue, penetrate the basal lamina and invade the surrounding connective tissue to enter into the bloodstream or lymphatic vessels and reach distant anatomic sites [2]. To accomplish this program, the initially altered epithelial cells undergo significant changes, from a stationary polarized phenotype towards a mesenchymal migratory phenotype, a process known as “epithelial to mesenchymal transition, EMT” [3,4].

However, neither all the neoplastic cells from a primary tumor will go through the multistep metastatic process, nor all the metastatic cells will colonize tissues and organs to the same extent. In the case of breast cancer one of the most common sites is the bone [5].

Studies on in vitro and in vivo models have demonstrated the complexity of the whole process of bone metastasis, which goes from the osteotropism to the nesting in the bone where the cells lurk, proliferate and migrate out again into other areas of the bone or in adjacent spaces [6-10]. However, the phenotypic traits of cells homing towards the bone have not been adequately explored.

The objective of this work was to study the proteomic profiling expressed by breast cancer cells after their nesting into the bone tissue. With this aim, we have applied an in vitro system of breast cancer cells (SKBR3) cultured in the presence of fragments of bone obtained from healthy informed donors during reparative surgery.

Through comparative and subtractive proteomics between the SKBR3 before (SKBR3-WT) and after the in vitro invasion of the bone tissues (SKBR3-B1), we have derived a collection of overexpressed proteins, likely generating functional networks involved in the mechanisms of cell penetration, homing and survival into the bone.

Cell culture

The SKBR3 cell line has been selected and the MCF10A cell line (ATCC, USA) was used as control for the present experiments. The SKBR3 is a breast cancer cell line derived from a pleural effusion of mammary adenocarcinoma and over-expressing HER2 / c-erb-2 gene. Cells were grown in RPMI-1640 (EuroClone, Italy) with 10% Fetal Bovine Serum (FBS, EuroClone, Italy), 100 U/ml penicillin and 100 μg/ml streptomycin.

The MCF10A (BS CL 174) is a non-tumorigenic epithelial cell line, derived from a non-tumoral mammary gland. Cells were grown in Dulbecco’s modified Eagle’s medium (DMEM, EuroClone, Italy) supplemented with: 10% FBS, 100 U/ml penicillin / 100 μg/ml streptomycin, 20 ng/ml EGF (Sigma-Aldrich, USA), 50 nmol/ml Hydrocortisone (Sigma-Aldrich, USA), 10 μg/ml Insulin (Sigma-Aldrich, USA) and 100 ng/ml Cholera Toxin (Sigma-Aldrich, USA). The cell lines were kept at 37°C in an atmosphere of 5% CO₂.

Both cell lines were recently submitted to a genetic test for their identity confirmation (BMR Genomics, Italy).

Bone fragments in cell culture assays

Bone fragments obtained as surgical waste from consenting informed donors were taken from tibial plates of 3 young patients (age 30-35), randomly selected among subjects operated on after traumatic events.

Soon after their arrival in the laboratory, the fragments were rinsed several times in sterile PBS and transferred in 6-well plates (EuroClone) with RPMI-1640 supplemented with 10% FBS, 100 U/ml penicillin and 100 μg/ml streptomycin. The wells with the bone fragments were kept in a 5% CO₂ incubator at 37°C for seven days to allow the release of weakly attached resident cells. Following this period, the fragments were relocated into new wells, where sub-confluent SKBR3 and MCF10A cells were seeded separately at the concentration of 15.000/cm2. Additional wells containing SKBR3 and MCF10A cells alone were used as controls.

Scanning electron microscopy

Several bone fragments co-cultured with SKBR3 cells and some fragments, not exposed to the neoplastic cell culture and used as controls, were rinsed with PBS and fixed with Karnovsky solution (1.5% glutaraldehyde, 1% paraformaldehyde, 1% cacodylate buffer, pH 7.4) for the SEM observation, performed at the Bologna and Varese laboratories. The bone fragments were then rinsed three times with 0.1% cacodylate buffer, postfixed for 20 min with 1% OsO₄ in cacodylate buffer, dehydrated with ethanol, and finally dried with hexamethyldisilazane (Sigma-Aldrich, USA) for 15 min, as described [11]. The specimens were sputter-coated with 15 nm of gold-palladium alloy system in a Balzers SCD-004 apparatus and then visualized with either a Philips 515 SEM operated at 15 kV or a FEI XL-30 FEG-SEM operated at 7 kV.

Immuno-assays

Western blotting: Following the electrophoretic runs on sodium dodecyl sulphate PolyAcrylamide Gel Electrophoresis (SDS-PAGE), the gels of cell lysates were electrotransferred onto nitrocellulose membrane (HyBond ECL; Amersham) and stained with Ponceau S (Sigma Aldrich). The membranes were then probed with the following monoclonal (mAb) or polyclonal (pAb) antibodies: anti-vimentin (clone 3B4) mouse mAb (1:500; Novocastra); anti-cytokeratin 8 (C51) mouse mAb (1:500; Santa Cruz Biotechnology, SCBT); anti-cytokeratin 18 (DC-10) mouse mAb (1:500; Merck KGaA); anti-actin (Ab-1) mouse mAb (1:20000; Merck KGaA); anti-c-Myc (9E10) mouse mAb (1:500; SCBT); anti-Matrix Metallo Proteases: anti-MMP2 (8B4) mouse mAb (1:500; SCBT); anti-MMP9 (C-7) mouse mAb (1:500; SCBT); anti-MT-MMP1 (MM0027-9E10) mouse mAb (1:500; SCBT); anti-α Enolase (H-300) rabbit pAb (1:10000, SCBT); anti-galectin 1 (1E8-1B2) mouse mAb (1:1000, Novus Biological); anti-glyceraldheyde-3-phospate dehydrogenase (0411) mouse mAb (1:1000, SCBT); anti-S100A7 (47C1068) mouse mAb (1:1000, SCBT).

Following incubation with the appropriate peroxidase-linked antibodies the reaction was revealed by the Enhanced chemiluminescence (ECL) detection system, using high performance films (Hyperfilm ECL; Amersham). Secondary antibodies were the following: horseradish peroxidase-conjugated goat anti-mouse IgG (1:5000; GE Healthcare Europe GmbH); goat anti-mouse IgM, H&L chain specific peroxidase conjugate (1:10000; Merck KGaA).

Immunocytochemistry (ICC): SKBR3 cells were grown on slides until sub-confluent (four days from seeding) and then fixed using methanol-based buffered preservative solution (ThinPrep PreservCyt Solution) according to the manufacturer’s instructions.

ICC analysis was carried out using the anti-Neu (300G9) mouse mAb (1:100; Santa Cruz). Incubation with primary antibody was carried out at 37°C for 30 minutes followed by washing with Phosphate Buffered Saline (PBS) solution (EuroClone). The reaction was carried out with the ImmunoCruz™ mouse ABC Staining System. The cells were counterstained with haematoxylin-eosin (Sigma). Negative controls were performed by excluding the primary antibody.

Two dimensional gel electrophoresis

Protein extracted from cell lysates were submitted to extensive dialysis against ultrapure distilled water at 4°C and lyophilized. Dried samples were solubilized in a buffer containing 4% CHAPS (3-[(3-Cholamidopropyl) dimethylammonio]-1-propanesulfonate) (Sigma-Aldrich, USA), 40 mM Trizma base (Sigma-Aldrich, USA), 65 mM DTE (1,4-Dithioerythritol) (Sigma-Aldrich, USA) and a trace of bromophenol blue in 8 M urea. Aliquots of 45 μg (analytical gels) or 1.5 mg (preparative gels) of total proteins (Bradford assay) were separately mixed with 350 μl of rehydration solution containing 8 M urea, 2% CHAPS, 10 mM DTE and 0.5% carrier ampholytes (Resolyte 3.5-10; Amersham), and applied for isoelectrofocusing (IEF) using commercial sigmoidal immobilized pH gradients (IPG) strips, 18 cm long with pH range 3.0-10 (Bio-Rad, USA). Isoelectrofocusing conditions were from 200 to 3500 V (3 hrs), 8000 V (8 hrs). The focused proteins were then separated on 9–16% linear gradient polyacrylamide gels (SDS-PAGE) with a constant current of 20 mA / gel at 10°C and the separated proteins were visualized by ammoniacal silver staining [12].

Silver-stained gels were digitized using a computing densitometer and analyzed with ImageMaster 2D PLATINUM software (Amersham Biosciences). Gel calibration was carried out using an internal standard and the support of the ExPaSy molecular biology server, as described [13].

Protein identification

In-gel digestion and MALDI-TOF analysis: Mass spectrometry was performed with the Voyager DE-PRO (Applied Biosystems) mass spectrometer as described [14]. Proteins were digested using sequencing-grade trypsin 20 μg / vial), and the peptides were re-dissolved in 10 μl of 0.1% trifluoroacetic acid (TFA) and spotted in HCCA (R-cyano-4-hydroxycinnamic acid) matrix (Sigma-Aldrich). MALDI-TOF mass spectra were recorded in the 500-5000 Da mass range, using a minimum of 150 shots of laser per spectrum. Internal calibration was carried out using trypsin autolysis fragments at m/z 842.5100, 1045.5642, and 2211.1046 Da. Peptide mass fingerprinting was compared to the theoretical masses from the Swiss-Prot or NCBI sequence databases using Mascot (http://www.matrixscience.com/). Typical search parameters were as follows: 50 ppm of mass tolerance, carbamidomethylation of cysteine residues, one missed enzymatic cleavage for trypsin, a minimum of four peptide mass hits was required for a match, methionine residues could be considered in an oxidized form.

In order to support the reliability of the spot identification, 2D-based Western Blot immuno-detection was applied to a selection of 5 mass-detected proteins, 3 of which are house-keeping, actin, glyceraldehide-3phospate and enolase, and 2 are regulated proteins, galectin and S10A7. Figure 1 shows a perfect coincidence of the two sets of profiles.

Figure 1: Panel of western blot assays performed on selected proteins separated by 2D-IPG, to validate the Maldi / TOF identification procedure.ACTB (Actin B), ENOA (alpha-Enolase), G3P (Glyceraldheyde 3P dehydrogenase), LEG1 (Galectin 1), S10A7 (Psoriasin)

Cell culture with bone fragments

Bone fragments obtained and prepared as described in the Material and Methods section, were located into wells where sub-confluent SKBR3 cells were seeded at the concentration of 15.000/cm². Figure 2a shows the SKBR3 cell culture at sub-confluence. In the inset the strong immune-cytochemical reactions of the SKBR3 with the ERBB2 / Neu antibody is shown, testifying the amplification of this gene expression in the cell line. The non-tumoral epithelial mammary cells, MCF10A, were placed in culture with the bone fragments, as a control experiment.

Figure 2: a) SKBR3 cell culture at sub-confluence, showing the apolarized cell morphology, typical of neoplastic cells. The inset shows the strong immune-cytochemical reaction of the SKBR3 with the ERBB2 / Neu antibody, testifying the amplification of this gene expression in the cell line; b) A spontaneous tropism of SKRB3 cells towards the edge of the bone fragments (BF) to which cells adhere.

The culture of cells with bone fragments were monitored daily. After four days an apparent spontaneous tropism of SKRB3 cells was observed towards the edge of the bone fragments to which cells adhere (Figure 2b). The cell culture was continued until the time of the cell confluence (7^th day). No tropic activity was observed for the MCF10A cells (not shown).

The bone fragments placed in culture with SKBR3 cells were then collected in two separated sets and submitted to two different experimental approaches: one set was fixed and prepared for the scanning electron microscopy; the second was placed in a new well with fresh RPMI culture medium in the CO₂ incubator at 37°C and monitored daily.

A parallel set of bone fragments not exposed to the neoplastic cell culture, was submitted to the scanning electron microscope observation.

Scanning electron microscopy

The results obtained with the SEM highlighted the presence of colonies of cells in the bone fragments co-cultured with the SKBR3 cells. Figure 3a shows an expanding cell colony growing within a bone matrix niche. Typically, the cells showed a rather uniform morphology and were grouped closely to each other, still maintaining a roundish outline with a very spiny cell surface. At a higher magnification the cell surface displayed the emission of irregular membrane ruffling and protrusions by which cells appear to cling tenaciously onto the bone tissue. A vesiculation activity, typical of malignant cancer cells with motile phenotype, also appeared (Figures 3b and 3c). Figure 3d reports the scanning micrograph of a control bone fragment showing the basic structure of the bone scaffold.

Figure 3: a) SEM micrograph of SKBR3 cell colonies growing within niches on a bone matrix fragment. The cells are growing in tight clusters and maintain a rounded shape; b) Detail of a small cluster of the SKBR3 breast cancer cells within the bone fragment; c) Higher magnification of the SKBR3 cell surface showing the elongated projections adhering to the bone matrix and producing an active vesiculation. The cells show numerous elongated cytoplasmic projections by which they establish reciprocal contacts and remain adherent to the bone; d) Low magnification SEM picture of a control bone fragment, showing the basic structure of the bone scaffold.

Cell sprouting from bone fragments

As already mentioned the second set of co-cultured fragments was placed into new wells and monitored daily. After 4 days an out-growth of cells was observed. Figures 4a and 4b show the progressive growth of colonies of cells, which, as being derived from a bone colonization, are named SKBR3-B1 (where B1 stands for “Bone-Conditioned”) to be distinguished from the parental cells named SKBR3-WT. By monitoring the time dependent outgrowth of SKBR3-B1 cells from the bone fragments we observed that the dynamic and modality of the bone conditioned cell-growth appeared typical for very aggressive neoplastic cells: as can be seen in Figure 4b, cells grow three-dimensionally one over the other, from the initial sprouting, and form multilayered colonies, independent of cell density and space availability, typical of the very aggressive cancer cells.

Figure 4: a) Outgrowth of the SKBR3 cells from the bone fragment (BF) in which they penetrated following the co-culture experiments. The new cell population sprouting from the bone fragments is named SKBR3–B1, where B1 stands for “bone conditioned”; b) The progressive three-dimensional overgrowth of SKBR3-B1 cell colonies derived from the bone colonization.

At fifteen days of co-culture, the cells released from the bone fragments were properly collected and subjected to protein extraction for the proteomic evaluations.

Immunochemical assessment

The phenotypic nature of the cells released from the bone fragments was further assessed by western blot using cytokeratins 8 and 18 (as epithelial markers) vimentin (mesenchymal marker, related to the epithelial-mesenchymal transition in cancer cells), cMyc oncogene to assess the cancer phenotype of these cells. The result of the assays revealed high expression levels of these proteins, confirming the epithelial origin and the tumoral phenotype of the SKBR3-B1 cells (Figure 5). To verify the potential collagenolytic capacity of the SKBR3-B1 cells, a battery of antibodies against most MMPs involved in the degradation of the extracellular matrix (MMP1, MMP2, MMP3, MMP9, MMP14) was also tested. As shown in Figure 5, strong positivity was obtained for the MMP1 in the SKBR3-B1 cells, in contrast to the SKBR3-WT, while faint or absent signals was obtained by the other MMPs antibodies (not shown).

Figure 5: Panel of western blot assays performed on SKBR3-WT (lane1) and SKBR3-B1 (lane2) respectively, with following antibodies: anti-c-Myc, anti-ACTB, anti-K2C8, anti-K1C18, anti-VIME, and anti-MMP1.

Proteomic profiling

Bone-conditioned cells, SKBR3-B1 were collected and prepared for the proteomic assays in parallel with the parental cell line, SKBR3-WT. Figure 6 shows a prototype of 2D-IPG reference map of the SKBR3-WT, where 373 protein spots were identified by the Maldi / Tof procedure immunologically validated by Western blot with selected antibodies, as reported in the MM section (see Figure 1). Supplementary Table 1 reports the list of identified proteins with the associated parameters.

Figure 6: Miniature of the reference MAP of SKBR3-WT. The 373 identified protein spots are labeled with the Uniprot AC number. When present, different isoforms of the same protein are jointly labeled.

The identified proteins were grouped into 13 functional clusters (Supplementary Table 2), on the basis of their primary function or subcellular localization according mainly to David Bioinformatics Resources [15,16] namely: Calcium Binding; Chaperones / Folding and Vesiculation; Cytoskeleton; Cytoskeleton Organization and Cell Motility; Detoxification; Ionic Homeostasis; Metabolic Enzymes; Mitochondrial Proteins; Nuclear Proteins; Nucleoside-related Processes; Processing of Metabolites and Aminoacids; Protein with Extracellular Activities; Protein Biosynthesis, Degradation and Modulation.

Finally, a “transverse” class of proteins belonging to different functional classes, but sharing an anti apoptotic activity (Negative regulation of apoptosis) was created.

Comparative profiling of SKBR3-B1 and SKBR3-WT: For the comparative profiling between the two cell cultures, the quantitative analysis of protein spots was performed in duplicate maps, and normalized as vol % (integration of optical density over the spot area) to correct for staining variations. The number of differentially expressed protein spots (including isoforms) displaying an increase in the intensity values greater than 50% was of 63 corresponding to 17% of the total proteomic complement. These data are presented in the histograms in Figure 7.

Figure 7: Comparative histograms of the differentially expressed protein spots displaying an increase in the intensity values greater than 50% in SKBR3-B1 proteins versus the SKBR3-WT. Each spot value is the average of 2 separate duplicated assays. Standard deviation is given as vertical error bars. Proteins are indicated by abbreviated names to UniProtKB database. In the ordinate are the % volumes of the protein spots exported from ImageMaster software. Histograms are grouped by intensity ranges (panels a, b, c and d).

The most represented proteins, among the overexpressed, belong to the following categories: Calcium Binding (12,7%), Chaperones / Folding and Vesiculation (14%), Cytoskeleton (9,5%), Organization and Cell Motility (11%), Detoxification (11%) and Protein Biosynthesis, Degradation and Modulation (14%).

In addition, it was observed that 15 out of 63 overexpressed proteins (including isoforms) belonging to the above functional clusters, exerted anti-apoptotic activity.

The major problem in cancer treatment is the formation of metastases for which there is still no adequate diagnostic and prognostic approach. In particular, bone metastases are the most insidious and risky type of metastasis, because of the nature of the host tissue, their rapid spreading inside and outside of the tissue, the difficulty of treatment, and the debilitating consequences and the frequent lethal progression.

In this study we set up a special model of co-culture to simulate the phenotype of breast cancer metastasizing cells, using bone fragments from healthy subjects operated on for trauma, and the breast cancer-derived cell line, SKBR3. As a negative control the non-tumoral cell line MCF10 was used.

The scanning electron microscopy has revealed that the SKBR3 cells, contrary to the MCF10A cells, were able to penetrate into the bone fragments digging small niches and producing colonies. These cells, named SKBR3-B1, were collected and further analyzed.

Morphological and western blot immuno-assays testified the epithelial origin of these cells; their neoplastic nature was confirmed by the modality of multi-layered growth, by the expression of two typical breast cancer oncogenes, ERBB2 and c-Myc, while the production of proteins which are typical expressions of mesenchymal cells, as vimentin and MMP1, gave evidence of their EMT transition.

The comparative proteomics profiles between the SKBR3-B1 cells and the parental cells (SKBR3-WT), revealed that the bone-conditioned cells displayed an overexpression (cut off 50%) of significant proteins commented below.

Calcium binding

This group includes Annexins and S100 proteins. The S100 family is the largest group of EF-hand calcium binding proteins with signaling properties. To date, at least 25 distinct members of this family have been recognized in humans. Mostly, they are expressed in a tissue-specific manner and may perform different intracellular and extracellular functions involved in several biological and pathological processes including cancer [17-20].

In addition, many of the S100 proteins are involved in the carcinogenesis by participating in the process of epithelial-mesenchymal transition [21], activation of cell migration [22] and the induction of the expression of metalloproteinases related to the invasive phase of the tumor [23]. The most representative members of the S100 family in the bone conditioned cells were the following: S10A2, S10A6, S10A7, S10A9, S10AB, S10AD.

It has been reported that S10A2 interacts with p53 so modulating its transcriptional activity [24]. Moreover S10A2 has been shown to interact with Smad3, to regulate TGF-β / Smad3 signaling and to induce EMT in lung cancer [25]. Our results show that the increase of S10A2 is also concomitant with the enhanced MMP1 protein expression, which may be instrumental for the collagenolysis required for the neoplastic cells to home onto the bone fragments. At present, no data have been reported on the role of S10A2 in bone metastasis.

The S10A6 has been demonstrated to be involved in many types of cancers and linked to metastasis. Recent studies in patients with pancreatic cancer have demonstrated a strong link between high expression of nuclear S10A6 and poor survival. While, depletion of S10A6 has been shown to exert a strong effect on the reversion of invasion and motility of pancreatic cancer cells [26]. No information is available yet on the possible role of this protein in the bone metastasis of cancer cells.

Concerning the S10A7, our previous study on a collection of breast cancer surgical fragments demonstrated that this protein is sporadically present in breast cancer patients with the same clinical diagnosis, but when present it was highly expressed both in the cytoplasm and in the nucleus of the neoplastic cells, and absent from the surrounding tissues [27]. Other Authors reported that S10A7 increases invasive capabilities of prostate cancer cells via a regulation of MMPs [28]. Another report shows that S10A7 inhibits β-catenin signaling by promoting β-catenin degradation. In turn, β-catenin signaling negatively regulates S10A7 expression. Thus, the reciprocal negative regulation contributes to their important roles in tumor progression [29].

Regarding S10A9 it is known that it localizes with its partner S10A8 in several biological systems but it may act as an individual player in many cancers [30].

It is believed that the expression of S10A9 correlates with inflammation, cancer or other pathological processes. In gastric cancer S10A8 / A9 have been shown to promote cell migration and invasion through an increase of MMP2 and MMP12 via p38 MAPK dependent NF-κB activation [31].

Other reports indicate that S10AB and S10AD are correlated, respectively, with the progression of pancreatic adenocarcinomas [32] and with the in vitro invasivity of lung cancer cells [33]. It has also been reported that S10AB is required for efficient plasma membrane repair and survival, necessary for the metastatic spreading of invasive cancer cells [34].

Interestingly, among the targets of these signals are the p53 activator of cJun which is co-promoter of MMP1 transcription. In addition another protein overexpressed in SKBR3-B1, SUMO1 (included in the group of Protein Biosynthesis, Degradation and Modulation), has been shown to modulate c-Jun and p53 activity [35].

Chaperones / Folding and Vesiculation. It is known that many stimuli, including oxidative stress and heavy metals, can activate stress responses in cells, besides the originally documented temperature shock. Major effects of cellular stress are protein aggregation and misfolding. A typical cell defense is the increase of heat shock / chaperon expression that helps in the protein refolding and disaggregation. In addition, these proteins may play essential roles in tumor growth, either by promoting autonomous cell proliferation or by inhibiting cell death pathways [36]. Alterations in HSPs expression are known to affect cell behavior including self-renewal, differentiation, sensitivity to environmental stress and aging [37]. Many of the HSPs are expressed at high levels in a wide range of tumors, where their expression is correlated with a poor prognosis and resistance to therapy. The increased transcription of HSPs in tumor cells is thought to be due to the loss of p53 function and to a higher expression of the proto-oncogenes HER2 and c-Myc [38]. Both oncogenes are found overexpressed in the SKBR3 cells. Numerous reports have indicated that some HSPs physically interact with several transcription factors and cofactors through intrinsic and extrinsic signaling pathways.

Mammalian HSPs have been classified into several families according to their molecular weight: HSP100, HSP90, HSP72, and small molecular HSPs (including HSPB1). Among the proteins of these families, the ones significantly over-expressed in the SKBR3-B1 with respect to the MCF10A cells and increased in respect to the SKBR3-WT, are the following proteins or isforms: CH10, HSPB1 (also known as HSP27), HSP74.

CH10 is a 10 kDa, highly conserved, mitochondrion-resident protein, which co-chaperones with another mitochondrial heat shock protein, the CH60. Eukaryotic CH10, with its partner CH60, is essential for mitochondrial protein biogenesis, as well as for the assembly and disassembly of protein complexes [39]. In addition, it has been well documented that the expression of small HSPs, especially HSPB1 and the inducible HSP70, enhances the survival of mammalian cells exposed to numerous types of stimuli that induce stress and apoptosis [36]. HSPB1 and HSP70 with anti-apoptotic roles are abundantly expressed in many malignant human tumors [40].

Other proteins with chaperone (RABP2) and folding (FKB1A, PPIA) activities were also differentially expressed in the SKBR3-B1 cells, but to a lower extent with respect to the HSPs already mentioned.

Cytoskeleton organization and cell motility

Among the overexpressed proteins in the bone conditioned cells, the proteins involved in the cytoskeleton reorganization deserve particular attention. The reorganization of cytoskeleton is an early cellular response to a variety of extracellular signals and is involved in essential cell functions such as: membrane trafficking, motility, retaining cell shape and polarity, cell proliferation and survival. PROF1 and COF1 are actin-binding proteins which play a crucial role in the turnover and restructuring of the actin cytoskeleton and so controlling several cellular processes. PROF1 has been found to be involved in lipid based signaling and in the regulation of glycolysis through the axis of G protein subunit, Gα13 / EGR1 [41]. COF1 has been found to be involved in several cell processes, among which the initiation phase of apoptosis [42]. It has also been shown to contribute to the contact inhibition and proliferation in cancer, through YAP / TAZ activity [43] and to import actin monomers into the nucleus under certain stress conditions. Recent studies propose the cofilin pathway as a major determinant of metastasis [44].

TYB4 is a regulator of actin assembly by its ability of depolymerization of actin filaments. It has been reported that TYB4 increases cell migration, angiogenesis and tumor metastasis of colon cancers [45].

It is reasonable to believe that even in bone metastases the cytoskeleton-associated proteins mentioned above play an important role in the migration and in the homing into the bone of the osteotropic cells.

Another over-expressed protein with multiple functions, including cytoskeleton dynamics, is the macrophage inhibitor factor, MIF, which may function either intracellularly or as a secreted cytokine. It is over-expressed in many solid tumors and is associated with poor prognosis.

The intracellular actions of MIF include the binding of Jab1, which is a co-activator of AP-1 transcription that also promotes the activation of MMPs transcription [46].

As extracellular cytokine, MIF is an important regulator of innate immunity, but it also participates in the promotion of tumor growth and metastasis [47].

Finally, CAPG and STMN1 over-expressed in SKBR3-B1, exert profound influences on cell proliferation, differentiation and cell motility in response to specific signals. All features characterize tumor growth and invasive behavior [48,49].

Detoxification

The detoxification proteins can be mainly classified into Phase I enzymes, Phase II enzymes and antioxidant enzymes according to their functional mechanisms (cfr. DetoxiProt database). Collectively these proteins can partially eliminate the active oxygen species produced by exogenous or endogenous toxins. In the bone conditioned cells the identified detoxification proteins which appeared over-expressed, include CLIC1, PRDX1, PRDX2, THIO.

Some of them, i.e. THIO can play additional functions including the control of cellular growth and apoptosis, and the modulation a number of extracellular inflammatory processes [50].

Protein biosynthesis, degradation and modulation

Among these protein groups, it is worth mentioning the small ubiquitin-related modifier, SUMO. In mammals, four different genetic isoforms, termed SUMO-1,-2,-3 and -4 have been identified so far. SUMO proteins are critically involved in the modulation of nuclear organization and cell viability. Their expression is significantly increased in processes associated with carcinogenesis, such as: cell growth, differentiation, senescence, oxidative stress and apoptosis. In our system we found over-expression of SUMO1 and SUMO4. SUMO1 plays a crucial role as a transcriptional co-regulation in various cellular pathways, including the p53 pathway in colon cancer [51] and HIF1α in prostate cancer [52]. Among the final targets of these pathways is the promoter of some metalloprotease genes, namely MMP1, MMP2 and MMP9, through direct or indirect pathways [53]. SUMO4, upon oxidative stress, conjugates to various anti-oxidant enzymes, chaperones, and stress defence proteins. It may also conjugate to some transcriptional regulators, positively or negatively regulating their activities. It has been shown that SUMO 4 may conjugate SUMO1 through SENP1 (sumo specific protease 1) and JUN therefore contributing to the modulation of the key enzyme of the invasion and metastasis, MMP1, MMP2 and MMP9 [54].

Negative regulation of apoptosis

Finally it is worth noting the observation that 24% of the protein over-expressed (15 out of 63, including isoforms), among the different functional classes have in common an antiapoptotic role. These are: COF1, MIF (Cytoskeleton, Organization and Cell Motility), NDKA, NDKB (Nucleoside-related proteins), CYTB (Nuclear proteins), RL40 (Protein Biosynthesis, Degradation and Modulation), PRDX1, PRDX2 (Detoxification), CH10, HSPB1 (Chaperones, Folding and Vesiculation), TCTP (Ionic Homeostasis).

In conclusion, the clusterization of proteins detected in this study, shows that the complex pattern of proteins over-expressed in the SKBR3-B1 cells can be subdivided into two major key groups: a) proteins primarily strategic for the cell penetration and homing into the bone, and b) proteins functioning as co-actors, which come into play to meet the modified metabolic needs of the cells.

The first group includes proteins of the calcium-binding family, the matrix metalloprotease. MMP1, proteins with anti-apoptotic activities and proteins of cell motility. The synergy among these proteins is reasonably required for bone demineralization and local digestion of collagen fibers to dig the niche where the cells home and proliferate. The anti-apoptotic proteins are instrumental for the continuation of cell proliferation and the formation of colonies, while the cell matrix adhesion, clearly observed in scanning micrograph, is likely to be facilitated by the proteins of the cell motility group, involved in the formation of membrane protrusions.

The second group includes proteins of metabolism, heat shock / chaperon and other proteins involved in the multiplicity of metabolic and basic processes of cell life.

Conclusively, we suggest that not only a single member within these groups of overexpressed proteins, but the inter-connected activity of the majority of them, is able to determine or drive the invasive and metastatic phenotype of breast cancer cells.

We believe that this present work represents a strong innovative effort due primarily to the co-culture system with fresh human healthy bone fragments and breast cancer cells. Indeed to our knowledge no functional proteomic study on bone metastasis utilizing human tissue fragments has been reported in literature while, several reports have been produced on animal models [55]. These results may contribute to the growing knowledge of the breast cancer proteomic atlas, and to the identification of new therapeutic targets for future molecular therapies against metastatic breast cancer.

Dr. Gianluca Di Cara is the Responsible of the Proteomics Laboratory of the COBS.

This work was approved by the Palermo1-Ethical Commitment (resolution N°939 of April 4th 2015).

Prof. Guido Filosto, President of La Maddalena Hospital is gratefully acknowledged for continuous support to the Centro di Oncobiologia Sperimentale (COBS).

The present work was mainly supported by State-made contribution (5x1000) for distinguished no profit research associations.