The breast cancer cell lines MCF-7 (luminal A), ZR75.30 (luminal B) and the monocyte cell line U937 were obtained from ATCC (Manassas, VA, USA) and maintained in RPMI-1640 medium supplemented with 10% FBS (both from Gibco-BRL, Grand Island, NY, USA), penicillin 10,000 U/ml, streptomycin 10 mg/ml and amphotericin B 25 μg/ml (PAA Laboratories GmbH/GE Healthcare Bio-Sciences Corp., Piscataway, NJ, USA) in a humidified atmosphere of 5% CO₂ at 37°C.

HUVECs were isolated and cultured as previously reported (20) by mixing cells from at least three human umbilical cords. The resulting cell cultures were maintained in M199 medium (Gibco-BRL) supplemented with 10% FBS, 2 mM L-glutamine (Gibco-BRL), 20 μg/ml endothelial mitogen (Biomedical Technologies, Inc., Stoughton, MA, USA), 5 U/ml heparin (Laboratorios PISA S.A. de C.V), penicillin 10,000 U/ml, streptomycin 10 mg/ml and amphotericin B 25 μg/ml, under a humidified atmosphere of 5% CO₂ at 37°C. All HUVEC cultures used for the experiments were at the third passage. The local Ethics and Research Committees of the Hospital General Dr Manuel Gea González, Ministry of Health (Mexico) approved this protocol (11-62-2014), and all participants signed an informed written consent form.

CM was isolated as previously described (17). Briefly, breast cancer cell lines were cultured in 100-mm plates until they reached 80% confluence. The cell layer was first washed 10 times with 10 ml of PBS/RPMI-1640 (1:1 v/v) without phenol red (Laboratorios Microlab S.A. de C.V., D.F. Mexico, Mexico) to remove serum components. Then, cells were maintained in 8 ml of serum-free RPMI without phenol red, after 48 h the culture medium was collected and lyophilized. The resulting powder was dissolved in water (1/10 of the original volume) and dialyzed using a PM-3 Ultrafiltration Membrane (EMD Millipore, Billerica, MA, USA). The solution was filtered through a 0.22-μm Millex-GS syringe filter unit, and a protease inhibitor cocktail was added (cOmplete™ Protease Inhibitor Cocktail; Roche Applied Science, Indianapolis, IN, USA). The protein concentration was determined using the Bradford reagent assay (Bio-Rad, Hercules, CA, USA). The resulting concentrated preparation was maintained at 4°C until further use.

CM (50 μl), was analyzed with the Bio-Plex suspension array system (Bio-Rad) against 26 proteins, following the manufacturer's instructions.

For adhesion assay, microarrays and western blots, confluent HUVECs were treated with 9 μg/ml of the indicated CM or with 10 ng/ml of human recombinant TNF (R&D Systems, Inc., Minneapolis, MN, USA) for the indicated time frames depending on the experiment. At the end of the treatments, the corresponding assay was performed as described. Inhibition of IKKs was performed by pre-incubating HUVECs for 1 h with 10 μM BAY 11-7085 (Calbiochem/Merck KGaA, Darmstadt, Germany). After pre-incubation, HUVECs were stimulated with 9 μg/ml of ZR75.30-CM. Electrophoretic mobility shift assay (EMSA) and IκBα western blotting were performed after 20 min, while CAM western blotting and cell adhesion were evaluated after 3 h, as described. For time course assay, HUVECs were starved 4 h previous to stimulation with 9 μg/ml of ZR75.30-CM, and cell lysates were used for western blot analysis.

This assay was performed as previously described (17), adherent cells were visualized in a TMS-F phase-contrast inverted microscope Nikon Eclipse TS100 (Nikon Instruments, Inc., Melville, NY, USA) and counted in a β counter (1600TR liquid scintillation analyzer; Canberra-Packard, Meriden, CT, USA).

TRIzol reagent (Invitrogen Corp., Camarillo, CA, USA) was used to obtain total RNA from three independent biological replicates of confluent HUVECs (60-mm plates) treated as indicated. The preparation of cRNA hybridization to Human Gene 1.0 ST and data analysis were performed according to Affymetrix™ recommendations. Differentially expressed genes were determined using the Partek^® Genomics Suite software (Partek, Inc., St. Louis, MO, USA) with a p<0.05 and a differential fold change of 1.5 on either positive or negative directions. Gene Ontology (GO) classification was performed through the use of National Cancer Institute-Database for Annotation, Visualization and Integrated Discovery (NCI-DAVID) (http://david.abcc.ncifcrf.gov) (21) and the Search Tool for the Retrieval of Interacting Genes (STRING) software was used to build the functional gene association networks (string-db. org) (22). Enriched canonical pathways within the networks of differentially expressed genes were carried out using the Ingenuity Pathway Analysis (IPA) (Ingenuity^® Systems; www.ingenuity.com) and Protein ANalysis THrough Evolutionary Relationships (PANTHER) software (www.pantherdb.org) (23).

The generation of cDNA was performed using the First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Inc., Waltham, MA, USA) with 2 μg of total RNA as a template. PCR reactions were performed with Maxima SYBR-Green/ROX qPCR Master Mix (Thermo Fisher Scientific, Inc.) on a 7300 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) using a conventional amplification protocol. The housekeeping gene ACTIN was used as an internal control. Primer sequences used for gene expression analysis are shown in Table I. The data were analyzed using the 2^{− Δ ΔCq} method (24).

Confluent cultures of HUVECs treated as indicated were lysed using RIPA buffer (150 mM NaCl, 50 mM Tris-HCl pH 7.4, 1% NP-40, 0.25% sodium deoxycholate, 1 mM EDTA pH 8.0, 1 mM PMSF, 1X complete protease inhibitors cocktail, 2 mM Na₃VO₄, 10 mM Na₂MoO₄ and 5 mM NaF) and protein concentration was determined by Bio-Rad DC protein assay (Bio-Rad). Total cellular proteins (10–40 μg) were separated via SDS-PAGE, transferred onto PVDF (EMD Millipore), membranes blocked for 1 h at room temperature in TBS-Tween 0.1%, with 5% non-fat milk and probed overnight at 4°C with specific antibodies against anti-VCAM-1, ICAM-1, E-selectin, IκBα, STAT3 (all from Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), phospho-STAT3 (Y705), phospho-VEGFR2 (Y1175), and VEGFR2 (all from Cell Signaling Technology, Inc., Danvers, MA, USA) and β-actin (Sigma-Aldrich, St. Louis, MO, USA). The membranes were incubated in the presence of HRP-conjugated secondary antibodies for 1 h at room temperature (Sigma-Aldrich), and the signals were detected using enhanced chemiluminescence (SuperSignal West Pico Chemiluminescent Substrate; Thermo Fisher Scientific, Inc.). The membranes were stripped as reported by Yeung and Stanley (25) and re-blotted. Optical densitometric scanning was performed using NIH ImageJ software.

Nuclear protein extracts and cytoplasmic fractions from HUVECs and NFκB translocation to the nucleus were determined as previously described (26). Briefly, 10 μg of the nuclear protein extracts were incubated with γ-³²P-ATP-labeled oligonucleotide containing the consensus NFκB site (5′-AGTTGAGGGGACTTTCCCAGGC-3′) (Santa Cruz Biotechnology, Inc.) in the presence of 100-fold excess of unlabeled specific probe as specific competitor. Samples were fractionated on a 5% non-denaturing polyacrylamide gel in 1X Tris-borate-EDTA buffer and DNA-protein complexes visualized on a Storm PhosphorImager (Molecular Dynamics, San Francisco, CA, USA). For supershift analysis, 1 μg of anti-p65 antibody (Santa Cruz Biotechnology, Inc.) was incubated with the nuclear extract for 30 min at room temperature prior to adding the reaction mixtures.

The results are expressed as the means ± SE, and statistical analyses were performed with GraphPad Prism 5.0 software (GraphPad Software, Inc., San Diego, CA, USA). Multiple comparisons were analyzed using one-way ANOVA with Dunnett's post-hoc test and two-way ANOVA with Bonferroni correction. Significance was set at p<0.05.

Previous study by our group (18) showed that CM secreted from the breast cancer ZR75.30 cells promotes a pro-adhesive endothelial phenotype. Interestingly, CM secreted from the breast cancer MCF-7 cells did not promote this phenotype nor induced the expression of CAMs: ICAM-1, VCAM-1 and E-selectin, therefore we focus on characterizing the transcriptome of ECs treated with ZR75.30-CM (Fig. 1).

For the transcriptome analysis a new batch of CM was tested for induction of the pro-adhesive phenotype in ECs and for the content of soluble factors using a multiplex assay. We found enrichment in the inflammatory cytokines TNF, IFN-γ and IL-6, the hematopoietic cytokine G-CSF, the chemokine IL-8, the growth factor VEGF and, to a lesser extent, the anti-inflammatory cytokine IL-1Ra for ZR75.30-CM. MCF-CM contained lower amounts of these factors. The molecules with the highest differences between ZR75.30-CM and MCF-7-CM were VEGF (>39,592 vs. 14,231 pg/ml), G-CSF (>28,728 vs. 9,594 pg/ml), IL-8 (>24,800 vs. 8,429 pg/ml), IL-6 (2,891 vs. 1,081 pg/ml), IFN-γ (2,068 vs. 1,315 pg/ml) and TNF (1,274 vs. 771 pg/ml). For VEGF, G-CSF and IL-8 from ZR75.30-CM, the values were above of detection range. Results from the composition analysis were similar to those previously reported (18). We used TNF as a positive control for the induction of a pro-adhesive endothelial phenotype accompanied by CAM expression. ZR75.30-CM altered the expression of 54 genes in ECs (53 upregulated and 1 down-regulated), whereas TNF treatment altered the expression of 299 genes (249 upregulated and 50 downregulated). Of the total number of genes affected, 50 were common to both treatments, 249 were exclusive to the TNF treatment and 4 were exclusive to the ZR75.30-CM treatment (Fig. 2). Among the 10 genes most upregulated by ZR75.30-CM, are genes related to chemotaxis and inflammation (CCL20, CXCL2, CXCL1, F3, IL-6, IL-8 and TNFAIP3) and cell adhesion (VCAM-1, SELE and RND1). Likewise, the most upregulated genes after TNF treatment correspond to cell adhesion process (VCAM-1, SELE, ICAM-4 and EFNA1), chemotaxis and inflammation (CCL20, CCL5, CXCL2, TNFAIP3 and TNFAIP2) and IFN-γ (IRF1) (Fig. 3A). All genes affected by both treatments showed significantly higher expression levels with TNF compared to those with ZR75.30-CM with the exception of CXCL2, for which ZR75.30-CM induced higher expression (Fig. 3A). The genes most repressed by TNF were PPP1R3C, PRICKLE1, ALX1, RUNX1T1, ZNF792, ID2, MBOAT1, ID1, RANBP6 and HOXA9. Interestingly, PPP1R3C showed the strongest down-regulation with TNF and it was the only downregulated gene with ZR75.30-CM (Fig. 3B).

The induction of a pro-adhesive phenotype directed the initial approach of this study, however, the bioinformatic analysis of the microarray by NCI-DAVID software associated the expression profile with other cellular processes such as: inflammatory response (p-values of 1.1e–10 for ZR75.30-CM treatment and 1.2e–11 for TNF), response to wounding (p-value of 4.1e–10 for ZR75.30-CM treatment and 1.3e–11 for TNF) and extracellular space (8.5e–10 for ZR75.30-CM) (Fig. 4A). While SELE was present in the three processes identified by NCI-DAVID, ICAM-1 and VCAM-1 were present only in the extracellular space process. The genes associated to these processes also shared chemokines, cytokines and components of the NFκB pathway known to regulate and promote a pro-adhesive phenotype (Fig. 4B).

In a similar analysis using PANTHER software the most representative biological process was the immune system (Fig. 4C). Among the main pathways identified by IPA software we found granulocyte adhesion and diapedesis (p-value of 1.72e–12) as part of ZR75.30-CM treatment (Table II). NFκB emerged as the principal regulatory molecule with the highest score (p-value of 8.93e–38) in ZR75.30-CM treatment analyzed by IPA and PANTHER (Table III and Fig. 4C). Bioinformatic analysis of TNF-treated ECs also showed NFκB as the main regulator and shared some biological processes and pathways with ZR75.30-CM treatment (data not shown). Finally, we generated a functional gene association network with the STRING software. The resulting network (Fig. 5) confirmed a cluster of interactions between TNF, NFKB, and CAMs and revealed a second cluster of interactions among chemokines and cytokines. These two clusters are interconnected through TNF, NFKB, IL-6, IL-8, CCL2 and CCL20.

To validate the data obtained in the micro-array and to compare the expression levels of selected genes between treatments, we used real-time PCR to quantify the mRNA of 15 genes involved in processes like cell adhesion (VCAM-1, SELE, ICAM-1), NFκB pathway (NFKBIA, NFKB2, MAP3K8, CDKN1B), chemotaxis and inflammation (CXCL2, CCL20, TGFB3, IL-6, TNFAIP2, TRAF1) and metabolism (PPP1R3C, SORBS1). Two of these genes were induced exclusively with ZR75.30-CM (TGFB3 and SORBS1), one with TNF (CDKN1B) which was repressed. Real-time PCRs validated the overexpression of the genes reported in the microarray. However, downregulation of PPP1R3C and CDKN1B genes observed in the microarray analysis was not replicated in the qPCR assays, although their expression level was lower than control for either TNF or ZR75.30-CM treatment (Fig. 6).

As NFκB emerged as the principal regulator of gene expression changes after ZR75.30-CM treatment associated to the pro-adhesive phenotype, and to test the relevance of this pathway we used a pharmacological inhibition of IκBα phosphorylation by pre-treating ECs with BAY 11-7085 prior to stimulation with ZR75.30-CM. EMSA revealed a faint signal from a basal NFκB/DNA complex present in control cells. The presence of this complex increased significantly when ECs were treated with ZR75.30-CM (first and second lanes, respectively). The basal complex disappeared when ECs were treated with the inhibitor (third lane) and was barely visible in cells treated with ZR75.30-CM plus inhibitor (fourth lane). Excess unlabeled NFκB probe completely eliminated the signal, indicating the specific detection of this complex (fifth lane) (Fig. 7A). Supershift assays confirmed that ZR75.30-CM triggered canonical NFκB activation, evidenced by markedly enhanced supershifted band in the presence of anti-p65 (sixth lane) (Fig. 7A). Western blotting against IκBα showed that the basal level of expression in control cells disappeared when treated with ZR75.30-CM (first and second lane, respectively). A slight reduction in IκBα expression was observed in the presence of the inhibitor and CM plus inhibitor (third and fourth lanes) (Fig. 7B). Western blotting revealed that the expression of ICAM-1, VCAM-1 and E-selectin increased with ZR75.30-CM treatment and that the expression was prevented when the cells were treated with the inhibitor (third and fourth lanes) (Fig. 7C). Finally, we confirmed the importance of the NFκB pathway in the adhesion process by showing that the pro-adhesive phenotype induced by ZR75.30-CM was prevented when the cells were treated with the inhibitor (Fig. 7D).

Despite the enrichment in VEGF (>39,592 pg/ml), G-CSF (>28,728 pg/ml), IL-8 (>24,800 pg/ml) and IL-6 (2,891 pg/ml) in ZR75.30-CM that could affect endothelial gene expression through STAT3, microarray analysis did not contain classic target genes related to these pathways. We verified that early signaling events of these systems were not impaired by analyzing the state of phosphorylation at residues related to functional activation of VEGFR2 (Y1175) and STAT3 (Y705) (Fig. 8A). In response to ZR75.30-CM VEGFR2 phosphorylation presents a biphasic response between 5–180 min. The signal presented a first peak after 5 min that reached a maximum at 30 min and returned to the value of the first peak after 1 h. STAT3 phosphorylation also presented a biphasic response between 5–180 min with a maximum signal at 30 min that became faint close to 120 min. Taken together these results indicate that early signaling of these two systems was not affected. However, when we analyzed the adhesion of HUVECs in response to recombinant human cytokines IL-6 and VEGF the adhesion was not significant (Fig. 8B).