Hbp

NMa

b

c

a

ARRAA

KAPcBI

1

htPtiacome

t

mtcc

1h

The International Journal of Biochemistry & Cell Biology 50 (2014) 1–9

Contents lists available at ScienceDirect

The International Journal of Biochemistry& Cell Biology

jo ur nal home page: www.elsev ier .com/ locate /b ioce l

igh-throughput molecular profiling of a P-cadherin overexpressingreast cancer model reveals new targets for the anti-cancer bacterialrotein azurin

uno Bernardesa,b, Ana Sofia Ribeirob, Sofia Abreua, André F. Vieirab, Laura Carretoc,anuel Santosc, Raquel Serucab, Joana Paredesb, Arsenio M. Fialhoa,∗

Institute for Biotechnology and Bioengineering, Center for Biological and Chemical Engineering, Instituto Superior Técnico, Lisbon, PortugalInstitute of Molecular Pathology and Immunology of the University of Porto (IPATIMUP), Porto, PortugalDepartment of Biology and CESAM, University of Aveiro, Aveiro, Portugal

r t i c l e i n f o

rticle history:eceived 9 October 2013eceived in revised form 18 January 2014ccepted 28 January 2014vailable online 6 February 2014

a b s t r a c t

Azurin is a bacterial protein from Pseudomonas aeruginosa which exerts an inhibitory activity in cancercells. In P-cadherin-overexpressing models, a bad prognosis marker in breast cancer increasing invasionand other malignant features, azurin decreases the invasion of cancer cells.

We performed a microarray analysis to compare the expression profile of azurin treated cells with dif-ferent P-cadherin expression levels. Azurin up-regulated apoptosis mediated by p53 protein, endocytosis

eywords:zurin-cadherinDNA microarraysreast cancer

ntegrin

and vesicle-mediated transport. In the contrary, in invasive MCF-7/AZ.Pcad cells, azurin decreased theexpression of genes associated with cell surface receptors and signal transduction, as well as biologicaladhesion. Further, azurin decreased adhesion of cells to proteins from the extracellular matrix (ECM) andaltered protein expression of integrins �6, �4 and �1 and interfered with the ability of these cells to formmammospheres. Altogether, our results further enlighten the anti-cancer effects mediated by azurin inP-cadherin overexpression breast cancer models.

. Introduction

Azurin is a protein secreted by Pseudomonas aeruginosa, whichas already been demonstrated that harbors in vitro and in vivo anti-umor progression properties (Yamada et al., 2002, 2004, 2009;unj et al., 2003). This protein demonstrates the ability to pene-rate preferentially in cancer cells compared to normal cells andnduces a cytotoxic response that mostly relies in a p53-dependentpoptosis-induction. Recently, it has been demonstrated that somehemical compounds, commonly used to inhibit the caveosomer late endosome formation, or that remove cholesterol from cellembranes, were significantly effective inhibiting the azurin pen-

tration in cancer cells (Yamada et al., 2009; Taylor et al., 2009).P-cadherin is a cell–cell adhesion molecule that has been shown

o promote aggressiveness in epithelial breast cancers and is

Abbreviations: E-cad, E-cadherin (or epithelial cadherin); ECM, extracellularatrix; EGFR, epidermal growth factor receptor; CXCR1, chemokine (C-X-C) recep-

or 1; FAK, focal adhesion kinase; CCR4, chemokine (C-C motif) receptor 4; CXCR4,hemokine (C-X-C) receptor 4; MMPs, metalloproteases; P-cad, P-cadherin (or pla-ental cadherin); VEGFR-2, vascular endothelial growth factor receptor 2.∗ Corresponding author. Tel.: +351 21 8417684.

E-mail address: afialho@ist.utl.pt (A.M. Fialho).

357-2725/$ – see front matter © 2014 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.biocel.2014.01.023

© 2014 Elsevier Ltd. All rights reserved.

recognized as marker for poor prognosis and worse patient survival(Paredes et al., 2005, 2007). In a context of wild-type E-cadherin,P-cadherin expression induces invasion and promotes the expres-sion and activity of matrix metalloproteinases (MMPs) 1 and 2(Ribeiro et al., 2010). In fact, P-cadherin exerts an inhibitory rolefor normal E-cadherin functional properties, which prevent tumorinvasion. Recently, in mouse models of highly invasive breast can-cers that co-express both cadherins, knocking down one of thecadherins rendered the cells to a less tumorigenic behavior whencompared to their simultaneous presence (Ribeiro et al., 2012).Moreover, analyzing a series of tumor samples, the co-expressionof both cadherins was associated with a poor clinical outcomeof the patients (Ribeiro et al., 2012). In the subgroup of basal-like breast cancers, where P-cadherin expression is of particularimportance, this protein is also associated with the phenotype ofcancer stem cells, being co-expressed with CD44 and CD49f (�6integrin). Cell populations enriched for P-cadherin demonstrateda higher ability to grow in anchorage-independent conditions, asmammospheres, than those with a lower P-cadherin expression;its presence still conferred resistance to X-ray-induced cell death

(Vieira et al., 2012).

We and others have shown different levels of action for azurinor its derived peptide in different models. In a previous study,we addressed the possible effects in an invasive model of breast

2 al of B

cdealaaaa

aPtvmvcuapucpccacbwc

2

2

((aawTab

2

uMSolms(sp(

hlcll

N. Bernardes et al. / The International Journ

ancer that depends on P-cadherin. We demonstrated that azurinecreases cell invasion through MatrigelTM in P-cadherin over-xpressing breast cancer cells, an effect that is associated with

specific decrease in P-cadherin protein levels and membraneocalization in both p53 wild type and mutant cell lines, withoutffecting E-cadherin expression. Additionally, azurin decreased thectivity of the extracellular protease MMP2, and the activity of Srcnd FAK non-receptor tyrosine kinases in the invasive cell modelsnalyzed (Bernardes et al., 2013).

In this work, we have performed a microarray analysis to inferbout the signaling pathways that are altered in a p53 wild-type and-cadherin overexpressing breast cancer cell model upon azurinreatment, comparing with the mock condition. According to pre-ious results, we showed that the entry of azurin in cancer cellsost surely involves endocytic pathways that are dependent on

esicle trafficking and membrane organization, since genes asso-iated with these cellular processes were found to be significantlyp-regulated in azurin treated conditions. Also an enrichment inpoptosis-related signaling pathways was a common consequence,robably as a result of the p53 wild type status of the cell modelsed. Finally, we also showed that the addition of azurin to breastancer cells, particularly to the transformed cell line that overex-resses P-cadherin, causes a down-regulation of genes coding forell surface receptors, linked to a down-regulation of the their intra-ellular signaling cascades. The results suggested by the microarraynalysis have been validated by some functional assays in bothell lines tested, as well as in another P-cadherin overexpressingreast cancer cell line – the triple negative basal-like SUM 149 PT,here we have also previous demonstrated that azurin reduces P-

adherin and invasion through MatrigelTM (Bernardes et al., 2013).

. Methods

.1. Antibodies

The following antibodies were used [integrin subunits �61:200, sc-13543), �4 (1:200, sc-6629) and �1 (1:200, sc-18887)Santa Cruz Biotechnologies)]. �-actin (1:1000, sc-1616) was useds a loading control. In order to evaluate azurin expression, an anti-zurin antibody was produced through immunization of one goatith purified azurin, obtained as described above (1:1000 dilution).

he resulting immunized serum was then purified by protein Affinity chromatography (SicGen, Portugal) and purity was checkedy SDS–PAGE.

.2. Cell culture and growth conditions

The following human breast cancer cell models have beensed in this study: MCF7/AZ [kindly provided by Prof. Marcareel (Ghent University, Belgium)] (Bracke et al., 1991) and

UM 149 [kindly provided by Prof. Stephen Ethier (Universityf Michigan, MI, USA)] (Willmarth and Ethier, 2006). Both cellines were routinely maintained at 37 ◦C, 5% CO2, in the following

edia (Invitrogen Ltd, Paisley, UK): 50% DMEM and 50% HamF12,upplemented with 10% heat-inactivated fetal bovine serumLonza, Basel, Switzerland), 100 IU/ml penicillin and 100 mg/mltreptomycin (Invitrogen). SUM149 medium was additionally sup-lemented with 5 �g/ml of insulin and 1 �g/ml of hydrocortisoneSigma–Aldrich, St. Louis, MO, USA).

MCF-7/AZ cell line was retrovirally stable transduced to encodeuman P-cadherin cDNA together with EGFP (MCF-7/AZ.Pcad cell

ine), as previously described (Paredes et al., 2004). MCF-7/AZ.Mockell line, encoding only EGFP, was used as a control. SUM149 celline constitutively expresses high levels of P-cadherin. All the cellines used express normal levels of E-cadherin.

iochemistry & Cell Biology 50 (2014) 1–9

2.3. Bacteria growth media and protein purification

Bacteria growth and protein expression and purification wereperformed as previously described in Bernardes et al. (2013).

2.4. cDNA microarray analysis

Total RNA from MCF-7/AZ cell lines (MCF-7/AZ.mock and MCF-7/AZ.P-cad), from treated with 100 �M of azurin or untreatedconditions, was extracted with the RNeasy Extraction kit (Qia-gen) (three independent samples per condition) and qualitywas analyzed with BioAnalyzer (Agilent Technologies, SantaClara, CA, USA). Samples were hybridized onto Agilent 44Kmicroarrays with probes for the Human Genome (AgilentHPAG4112F), following the manufacturer’s instructions, using theAgilent One-Color Microarray-Based Gene Expression AnalysisProtocol Version 5.7 (Quick Amp Labeling Kit). After obtain-ing microarray images, fluorescence intensity was measuredwith the Agilent Feature Extraction software (version 10.5.1.1)and signal processing was performed according to the Agilentrecommendations (GE1-v5 95 Feb07 Protocol). Processed sig-nals were annotated and filtered using BRB Array Tools 4.2.1(http://linus.nci.nih.gov/BRB-ArrayTools.html). Differences in geneexpression were assessed using Student’s t-test implemented inBRB-tools, with a p-value cut-off of 0.05. Only genes with alteredexpression by at least 1.7-fold in relation to untreated cell lineswere indicated as differentially ArrayExpress microarray databaseexpressed. A hierarchical clustering method was applied to identifygroups of differentially expressed genes (DEGs) between untreatedand treated samples. DAVID software was used to analyze geneontology and pathway enrichment (http://david.abcc.ncifcrf.gov)(Huang da et al., 2009). The Gene Expression Omnibus (GEO) aces-sion number for the microarray data associated with this paper isGSE54319.

2.5. Quantitative real-time reverse transcription-polymerasechain reaction (qRT-PCR) analysis

Total RNA from MCF-7/AZ and SUM149 cell lines was extractedusing the RNeasy Extraction kit (Qiagen), according to the manu-facturer’s instructions. Samples were subjected to treatment withDNase (Qiagen) during the extraction procedure. A list of probesis provided in the Supplementary Data. Assay relative quantifica-tions between treated and untreated samples were determinedby the ��Ct method using the internal standard human GAPDH(Hs.PT.51.19405051) to normalize for cDNA quantity.

2.6. Protein extraction and Western blot analysis

Protein extraction and Western blot were performed as previ-ously described (Bernardes et al., 2013). For a more detailed pleasesee the Supplementary Data.

2.7. Adhesion assay to ECM substrates

Cell adhesion was performed in 96-well plates coatedwith laminin 332 (Sigma–Aldrich, St. Louis, MO), fibronectin(Sigma–Aldrich), type-I or IV collagen (Sigma–Aldrich) (5 �g/ml)overnight at 4 ◦C. Afterwards, plates were washed three timeswith PBS and non-specific binding sites were blocked with 0.5%BSA (w/v) in PBS containing PenStrep (Invitrogen) for 2 h at 37 ◦C.After washing again, 100 �L of untreated or azurin (50 or 100 �M;

48 h) treated cells (106 cells/ml) were seeded in serum-free mediafor 30 min. Non-adherent cells were removed by washing platesthree times with PBS, and the attached cells were fixed with ace-tone:methanol (1:1) for 10 min at 4 ◦C. Adhesion was determined

N. Bernardes et al. / The International Journal of Biochemistry & Cell Biology 50 (2014) 1–9 3

Fig. 1. Gene expression profile of MCF-7/AZ.Mock and MCF-7/AZ.Pcad treated with azurin. (a) Azurin enters breast cancer cells as observed by Western blot. (b) Venn diagramo eatedl comma

fmasu

2

(i

2

at(

3

3c

o74osp

f genes with modified expression levels in MCF-7/AZ.Mock and MCF-7/AZ.Pcad trines. (c) Selected gene ontology assignment and enriched KEGG pathways of the

nalyzed with DAVID software (p-value < 0.05).

ollowing a colorimetric method described by Busk et al. (1992),easuring absorbance at 570 nm with a microplate reader. BSA

nd plastic uncoated wells were used as controls. Results are pre-ented as the percentage of adhesion between azurin treated andntreated cells (100%).

.8. Mammosphere assay

Mammosphere assay was performed as described in Vieira et al.2012). Azurin at 100 �M was added at day 1 to the media contain-ng cells.

.9. Statistical analysis

Regarding in vitro assays, data are expressed as mean values oft least three independent experiments ± SD. Student’ two-tailored-tests were used to determine statistically significant differencesp < 0.05).

. Results

.1. Gene expression profile of azurin treated MCF-7/AZ breastancer cell models

We performed a microarray analysis comparing mRNA profilesf azurin treated and untreated MCF-7/AZ.Mock and MCF-/AZ.Pcad cell lines (Fig. 1a). After exposition to azurin (100 �M,

8 h), total RNA was extracted and the mRNA profiles werebtained. DAVID software (Huang da et al., 2009) was used toeparately analyze up- and down-regulated Gene Ontology andathway enrichment through KEGG database.

with azurin (48 h, 100 �M). 415 genes were found commonly altered in both cellon genes with modified expression in both cell lines after treatment with azurin,

As a first step of the analysis of the microarray data, weidentified 1030 genes whose expression was significantly de-regulated (DEGs), upon azurin treatment in MCF-7/AZ.Mock cells,and 3164 DEGs in MCF-7/AZ.Pcad cells (p < 0.05, 1.7-fold change),in comparison to untreated cells. In these data sets, we found415 genes that were altered in both cell lines after azurin treat-ment; from those, 238 genes were commonly up-regulated and128 genes were commonly down-regulated (Fig. 1b). Among thesecommon genes differentially expressed, we identified genes asso-ciated with the p53 signaling pathway and apoptosis, whichexpression was up-regulated in comparison with untreated cells(enrichment score = 6.5; p = 0.021); in addition, genes associatedwith membrane organization (enrichment score = 3.0; p = 0.006),vesicle-mediated (enrichment score = 2.6; p = 0.004) and endosometransport (enrichment score = 7.9; p = 0.013), or with lysosome(enrichment score = 4.7; p = 0.019), were also found significantlyup-regulated by azurin; on the contrary, a major category signifi-cantly associated with genes commonly down-regulated by azurinis the regulation of transcription (enrichment score = 1.5; p = 0.045)(Fig. 1c).

3.2. Identification of the differentially expressed cancer-relatedgenes in P-cadherin overexpressing MCF-7/AZ breast cancer cellsafter azurin-treatment

Based on the previously described anti-invasive effects medi-ated by azurin in P-cadherin overexpressing breast cancer models,the next stage of analysis was focused primarily on the data set

comprising the 3164 de-regulated genes in MCF-7/AZ.Pcad azurin-treated vs untreated breast cancer cells. As can be observed in Fig. 2,the molecular pathways and biological processes that statisticallysignificantly changed by the azurin treatment were the following:

4 N. Bernardes et al. / The International Journal of Biochemistry & Cell Biology 50 (2014) 1–9

F wn-ret

vafcliacmbeMFtwddi

3p

cmaOpsc

ig. 2. Selected gene ontology and enriched pathways (Kegg database) of up- and doo untreated cells, obtained with DAVID (p-value < 0.05).

esicle-mediated transport, membrane organization, induction ofpoptosis, endocytosis, p53 signaling pathway, and lysosome wereound has being significantly up-regulated by the treatment; con-erning the down-regulation, we found the cell surface receptorinked signal transduction, biological adhesion, response to wound-ng, cell motility, cytokine–cytokine receptor interaction, and focaldhesion (Fig. 2). The most upregulated and downregulated genesategorized by the above-mentioned functions are listed in Supple-entary Table 1. To independently validate the alterations found

y microarrays, we have chosen 4 random genes with varyingxpression profiles between treated and untreated conditions inCF-7/AZ.P-cad breast cancer cell model (namely, TWIST1, CXCR1,

GFR1 and BAX) to evaluate by real-time RT-PCR analysis. Impor-antly, the results obtained for these selected genes were consistentith the microarray data (Fig. 3), allowing its validation. A furtheriscussion about some of the genes that were highly upregulated orown regulated within these identified pathways is more detailed

n the following sections.

.2.1. DEGs associated with the induction of apoptosis and of the53 signaling pathway induced by azurin treatment

The previous described effects of azurin in p53 wild-typeells (Yamada et al., 2002, 2004, 2009) were confirmed by thisicroarray analysis, since we observed an up-regulation of genes

ssociated with an instruction of apoptosis, observed by both Gene

ntology and KEGG pathways (Figs. 1c and 2). BAX, for exam-le, which is a major pro-apoptosis inducer, was found has beingignificantly up-regulated upon azurin treatment in both breastancer cell models (2.1-fold, p-value 0.006 for MCF-7/AZ.Mock and

gulated genes in MCF-7/AZ.Pcad cell line treated with azurin (48 h, 100 �M) related

2.3-fold, p-value 0.007 for MCF-7/AZ.Pcad). CASP9 mRNA was alsofound up-regulated in MCF-7/AZ.Pcad cell line (2.2-fold, p-value0.002), also reflecting the cell entrance in apoptosis induced byazurin. Importantly, although cell viability remained above 80%(Bernardes et al., 2013) in the experimental conditions used, itseems that intracellular azurin leads to p53 signaling cascadeswhich, upon repeated doses, is likely to terminate in cell death.

3.2.2. DEGs associated with the up-regulation of vesicle-mediatedtransport, membrane organization, endocytosis and lysosomeupon azurin treatment

Upon azurin exposition, genes associated with endosomeformation, membrane organization, transport through vesiclesand lysosome degradation were found has being up-regulatedin both cancer models (MCF-7/AZ.Mock and MCF-7/AZ.Pcad)(Figs. 1c and 2). These results are also in accordance with previousresults obtained by others, in which azurin, or its derived peptide,was shown to enter cancer cells in an energy-dependent manner,without plasma membrane disruption (Taylor et al., 2009).

3.2.3. DEGs associated with the down-regulation of thecell-surface receptor-linked signal transduction induced by azurin

The most representative group of genes in this data set belongsto the cell surface receptor linked signal transduction family, with155 genes associated to this group (Fig. 2). This family of genes

is best characterized by their ability to detect extracellular sig-nals, providing a link to the inside of the cell, culminating inprocesses often associated with changes in the transcription of tar-get genes. Amongst these genes, we found cell surface receptors

N. Bernardes et al. / The International Journal of Biochemistry & Cell Biology 50 (2014) 1–9 5

F F-7/AZ( ents.

adt0r

3gd

fi(ais

trcmf0e(i

CmEP(

b

ig. 3. Validation of the microarray data of selected genes by qRT-PCR between MCmRNA levels relative to control GAPDH transcript) from three independent experim

nd their ligands, which are often overexpressed in cancer, as beingown-regulated by azurin. Particularly, we found that the recep-ors EGFR (2.04-fold; p-value 0.029), CXCR1 (2.08-fold; p-value.030), and CCR4 (5.9-fold; p-value 0.006) were significantly down-egulated by this bacterial protein (Supplementary Table 1).

.3. Identification of the differentially expressed cancer-relatedenes up-regulated by P-cadherin overexpression andown-regulated by azurin treatment

As another approach, we decided to compare the mRNA pro-les of the untreated controls MCF-7/AZ.Pcad and MCF-7/AZ.Mockp < 0.05, 1.7-fold change), in order to identify the genes that wereltered by P-cadherin overexpression. This analysis was performedn order to find, in a second step, if some of these genes were alsopecifically altered, but in an opposite way, by azurin treatment.

Interestingly, P-cadherin overexpression in MCF-7/AZ cells ledo a significant increase in genes associated with cell surfaceeceptor linked signal transduction, namely genes coding for twohemokine receptors associated with breast tumor progression andetastasis, such as CCR4 (3.1-fold, p-value 0.01) and CXCR4 (3.3-

old, p-value 0.0003) (Jin et al., 2012), but also END1 (2-fold, p-value.01). Also, overexpression of P-cadherin led to an increase in thexpression of genes associated with lipid localization and transportSupplementary Table 2), suggesting that it could lead to a changen the lipid profile expression or localization at the cell membrane.

The validation by qRT-PCR was performed to CDH1/E-cadherin,DH3/P-cadherin and CXCR4 genes. As can be seen in Fig. 4, theicroarray data has been validated, with no alterations found in

-cadherin codifying gene, but confirming the up-regulation of

-cadherin and CXCR4 mRNA transcripts in MCF-7/AZ.Pcad cellsFig. 4).

Finally, we compared the global gene mRNA expression profilesetween azurin-treated and untreated MCF-7/AZ.Pcad cells with

.Pcad + azurin 100 �M and control cells. Bars represent mean gene expression ± SD

the differentially expressed genes between MCF-7/AZ.Mock andMCF-7/AZ.Pcad cell lines. 430 genes were commonly de-regulatedin both conditions (Fig. 5a). Interestingly, from those, almost 90% ofthe genes had opposite effects: 186 genes that were up-regulatedby CDH3/P-cadherin were then down-regulated upon exposure ofMCF-7/AZ.Pcad to azurin. Inversely, 188 genes that were down-regulated by P-cadherin were then up-regulated by azurin inMCF-7/AZ.Pcad cells (Fig. 5b). As an example, genes involved inthe promotion of apoptosis were found as down-regulated byP-cadherin, being specifically associated with the p53 signalingpathway, such as ZMAT3, CDK1NA (p21/WIP1) or FAS; interestingly,these same genes were then up-regulated in MCF-7/AZ.Pcad cellline after the treatment with azurin. Of note, form the genes withopposite changes, we observe that those belong mainly to the cellsurface receptor liked transduction family and to a negative regula-tion of apoptosis (Fig. 5c), suggesting that P-cadherin up-regulatesthose genes in order to promote the invasive and tumor progres-sion cascade and negatively regulates apoptosis. However, treatingthese cells with azurin counteracts these effects.

3.4. Azurin decreases adhesion of P-cadherin overexpressingbreast cancer cells to ECM proteins and interferes with integrinexpression

Biological adhesion was one of the Gene Ontology categoriesthat was found as significantly enriched in the data set of genesdown-regulated by azurin in MCF-7/AZ.Pcad cells. Within thiscategory, we found genes associated with cell–cell adhesion, orwith adhesion between cells and the extra-cellular matrix (ECM),namely to laminins and collagens. In order to independently ver-

ify and validate the alterations caused by azurin in this process,we investigated its ability to interfere with the adhesion betweencancer cells several components constituting the ECM, such aslaminin-332, collagen type I, collagen type IV and fibronectin.

6 N. Bernardes et al. / The International Journal of Biochemistry & Cell Biology 50 (2014) 1–9

Fig. 4. Validation of the microarray data of selected genes by qRT-PCR between MCF-7/AZ.Mock and MCF-7/AZ.Pcad. Bars represent mean gene expression ± SD (mRNA levelsrelative to control GAPDH transcript) from three independent experiments. CDH1 was not differentially expressed the cDNA microarray analysis.

Fig. 5. Common altered gene transcription modulation by P-cadherin in MCF-7/AZ.Pcad related to MCF-7/AZ.Mock and MCF-7/AZ.Pcad cells treated with azurin (48 h, 100 �M)related to untreated cells. (a) Venn diagram of genes with modified expression in both conditions under analysis. 430 genes were found commonly de-regulated betweenboth data sets (left panel); cross-comparison between up- and down-regulated genes in both data sets (right panel). (b) Selected gene ontology assignments and enrichedKegg pathways.

N. Bernardes et al. / The International Journal of Biochemistry & Cell Biology 50 (2014) 1–9 7

Fig. 6. Azurin alters adhesion of breast cancer cells to some extracellular matrix (ECM) components. (a) Azurin treatment (50 �M and 100 �M, MCF-7/AZ cells 48 h, SUM1 ock,( on assp and S

Awaiafiataimwd

aarmwaBaa

49PT 24 h) caused a reduction in the percentage of adhesion of cells (MCF-7/AZ.Madhesion time = 20 min; *p-value < 0.05). (b) Azurin (same conditions as for adhesilastic conditions or a matrix formed by collagen type-I in invasive MCF-7/AZ.Pcad

s previously described for global gene expression analysis, cellsere first exposed to azurin and left to adhere to ECM proteins;

dhesion was then measured by the crystal-violet assay. Interest-ngly, cell adhesion was significantly decreased (about 20–30%) inzurin treated condition, particularly to laminin, collagen type I andbronectin (Fig. 6a). BSA and plastic uncoated wells were used as

control and no significant adhesion capacity was found betweenreated and untreated cells (supplementary data). Since we havelready demonstrated that azurin can also lead to a decrease in thenvasion and P-cadherin levels in SUM 149 PT breast cancer cell

odel (Bernardes et al., 2013), we also performed adhesion assaysith this cell line and observed the same pattern of results, withecreased adhesion of treated cells to ECM proteins (Fig. 6a).

The above results prompted us to investigate if the ability ofzurin to decrease adhesion to ECM matrices was dependent onlterations in integrin receptors. Thus, we studied some integ-in subunits, such as �6, �1 and �4. �1 integrin is present inost integrin heterodimers and recognizes most ECM components,hereas �6 and �4 bind more specifically to laminin and are

ssociated with breast cancer promotion (Lipscomb et al., 2005;issell et al., 2011). Interestingly, exposing both MCF-7/AZ.Pcadnd SUM 149 PT invasive breast cancer cell lines to azurin led to

decrease in these integrin subunits, as observed by Western blot

MCF-7/AZ.Pcad and SUM 149 PT) to laminin-332, fibronectin and collagen type-Iays) reduces protein expression of integrin subunits �6, �4 and �1 under normal

UM 149 PT cell lines.

(Fig. 6b); however, in the case of integrin subunit �6, this result wasdependent if cell growth and azurin treatment was performed overa layer of collagen-I. Integrins �4 and �1 were decreased in bothnormal plastic conditions and collagen type-I growth conditions(Fig. 6b).

3.5. Azurin decreases mammosphere forming efficiency inP-cadherin overexpressing breast cancer cells

We have recently demonstrated that P-cadherin is associatedto a breast cancer stem cell phenotype, being co-expressed withCD49f (�6 integrin subunit) and being a regulator of the mammo-sphere forming efficiency (Vieira et al., 2012) of basal-like breastcancer cells. Integrins, particularly integrin �6, are also neces-sary to the stem cell-like phenotype in MCF-7 cells (Cariati et al.,2008). Therefore, according to our previous results in which azurincauses a specific decrease in P-cadherin and �6 integrin expres-sion, we investigated if azurin was also capable to reduce themammosphere forming efficiency (MFE) of breast cancer cells. In

anchorage-independent growth conditions, and upon azurin treat-ment, we observed a decrease in the MFE for both MCF-7/AZ.Mockand MCF-7/AZ.Pcad cells, with a more prominent impact in thelatter. As already described, MCF-7/AZ.Pcad cells harbor a higher

8 N. Bernardes et al. / The International Journal of B

Flv

Mainm

4

taAftsr�ermnAtalwoai

asscatt

ear(c

ig. 7. Azurin decreased the mammosphere forming efficiency of breast cancer cellines in anchorage-independent growth conditions in breast cancer cell lines (*p-alue < 0.05; **p-value < 0.01).

FE than mock control cells; however, treating with the samemount of azurin, there was more pronounced decease in MFEn P-cadherin overexpressing cells. Accordingly, azurin also sig-ificantly decreased the MFE of the SUM 149PT breast cancer cellodel (Fig. 7).

. Discussion

Azurin is a bacterial protein that harbors anticancer proper-ies, either by promoting apoptosis, or by blocking invasion orngiogenesis (Yamada et al., 2004, 2005, 2009; Mehta et al., 2011).zurin (or its derived peptide – p28) penetrates in cancer cells

aster than in normal cells (Yamada et al., 2005), by a mechanismhat doesn’t cause plasma membrane disruption but depends onome of its components. For example, it is known that cholesterolemoval from the plasma membrane of cancer cells, using methyl--cyclodextrin, significantly reduced the azurin entry (Yamadat al., 2009). Additionally, treatment with nocodazole, which dis-upts membrane caveolae by disrupting microtubules, or withonensin, that inhibits late endosome/lysosome activity, also sig-

ificantly affect the azurin penetration rate (Yamada et al., 2009).ccordingly with this data, our microarray analysis of azurin

reated breast cancer cells (MCF-7/AZ.Mock and MCF-7/AZ.Pcad)lso revealed an up-regulation of genes associated with these cel-ular processes, such as vesicle transport and pathways associated

ith the lysosome (Fig. 1). Moreover, as identified by gene ontol-gy, genes associated with endocytosis, membrane organizationnd endosome transport also displayed an increased expressionnduced by azurin treatment.

A significant number of genes coding for cell surface receptors,s well as their intracellular partners, decrease their expres-ion upon azurin treatment, down-regulating their downstreamignaling, which usually sustains cell proliferation and aberrantonstitutive signaling (Figs. 1 and 2). In fact, cancer cells have thebility to grow even in the absence of external growth stimula-ory signals, frequently by overexpressing growth factor receptoryrosine kinases (Hanahan and Weinberg, 2011).

Among these DEGs, we have identified the gene EGFR, whosexpression was down-regulated by 2-fold (p-value 0.029) after

zurin treatment in MCF-7/AZ.P-cad cells. EGFR is a tyrosine kinaseeceptor, particularly overexpressed in triple-negative basal-likeTNBL) breast cancers (Eccles, 2011), a subset of tumors where P-adherin was also described as a useful marker. This receptor is

iochemistry & Cell Biology 50 (2014) 1–9

also associated with nodal or distant metastasis, and was alreadyassociated with P-cadherin expression (Albergaria et al., 2011).

Interestingly, this study still showed that the P-cadherin over-expression in MCF-7/AZ.Pcad cells leads to elevated expressionof genes encoding cell surface receptors involved in signal trans-duction. Receptor tyrosine kinases may become overly activatedeither by genomic amplification, overexpression or by mecha-nisms that inhibit their degradation. Such deregulations can leadto aberrant accumulation of these receptors on the cell surfaceof cancer cells (Abella and Park, 2009). Thus, our results sug-gest that the more aggressive cellular phenotype displayed byP-cadherin-overexpressing breast cancer cells is strongly linkedto the up-regulation of multiple cell-surface receptors, which inpart is then down-regulated by azurin treatment (Fig. 5). Also, inresistance to apoptosis, azurin has a beneficial role in P-cadherinoverexpressing cells. P-cadherin not only promotes invasion andmigration of cancer cells, but also resistance to cell death. In BT-20 cell line, a basal-like triple negative breast cancer model whichendogenously express elevated levels of this protein, there was asignificantly decreased cell survival after an apoptotic stimulus ifa knocking-down of the expression of CDH3 gene with a specificsiRNA was performed (Ribeiro et al., 2012).

The results observed here suggest that the mechanism by whichazurin exerts its anti-cancer effects depends on its route of cancercell entry, disrupting caveolae and removing from the cell mem-brane selective receptors that may be overactivated. In cancer cells,the removal of functional receptors from cell surface and their tar-geting to lysosome was proven to be an important mechanism bywhich their permanent activation and consequent tumorigenesisis prevented, particularly to EGFR (Abella and Park, 2009).

Altered expression of ECM components or their cell surfacereceptors, such as integrins, are key players in tumor initiationand progression (Jinka et al., 2012). Luminal cells contacting thestromal ECM, such as collagen I, is a feature known to lead tosignaling into aberrant transformations, up-regulation of MMPs,invasion and metastasis (Bissell et al., 2011). Also, elevated expres-sion of laminin promotes invasion in breast carcinomas (Carpenteret al., 2009). Additionally, cell adhesion and migration are alsoimportant processes that can be mediated by cell membrane orga-nization and lipid composition. In agreement with these results,our microarray data revealed that membrane organization andendocytic trafficking processes, as well as cell adhesion, displayedsignificant alterations in response to azurin (Figs. 1 and 2). In orderto validate these findings, we analyzed the ability of azurin tointerfere with adhesion to ECM proteins (laminin, fibronectin andcollagen type I) and integrin expression (subunits �6, �4 and �1).We observed that azurin treatment, in both MCF-7/AZ.Pcad andSUM149 PT invasive breast cancer cells, led to decreased adhesionto ECM components with a concomitant reduction of the integrinprotein levels (Fig. 6). Accordingly, we have previously demon-strated that phosphorylated FAK and its partner Src were decreasedin these models, concomitantly with decreased invasion and P-cadherin levels (Bernardes et al., 2013). FAK and Src are importantnon-receptor tyrosine kinases that can be activated by integrinengagement to the ECM. Once more, this result suggests a globalanti-cancer phenomenon triggered by azurin, in which a varietyof cellular receptors and intracellular signaling cascades becomede-activated.

In anchorage-independent conditions, ECM is present withinthe spheroid structure formed by the cells. The role of P-cadherin,as a breast cancer stem cell marker and a promoter of cell survivalin these conditions, has been previously established (Vieira et al.,

2012). Furthermore, FAK signaling is important for the survival ofcells under these conditions, since FAK ablation reduced the poolof cancer stem/progenitor cells in primary tumors of FAK-targetedmice, impairing their self-renewal and in vitro migration (Luo et al.,

al of B

2oaaPw

eila

A

aa

C

A

(ag(RJaT

A

f2

R

A

A

B

B

B

B

ization of bacterial redox protein azurin in mammalian cells: entry domain and

N. Bernardes et al. / The International Journ

009). We tested the ability of azurin to interfere with the capacityf these breast cancer cells to form mammospheres and observed

decrease in the MFE of all three breast cancer models used, withn increased ability of reduce in those that express higher levels of-cadherin, MCF-7/AZ.Pcad and SUM 149 PT, which were the onesith the higher MFE (Fig. 7).

The blockage of these signaling pathways in P-cadherin over-xpressing breast cancer cell models leads to the abrogation ofnvasive and cancer stem cell/progenitor features mediated, ateast in part, by this protein, providing a possible new therapeuticpproach to this specific type of breast cancer.

uthor contributions

This study was designed by RS, JP and AMF. NB, ASR, SA, AFVnd LC performed the experiments. NB wrote the manuscript. Alluthors read and approved the final manuscript.

onflict of interest

The authors declare no conflict of interests.

cknowledgments

The work presented was supported by a scientific projectPTDC/EBBBIO/100326/2008) financed by the Portuguese Sciencend Technology Foundation (FCT). FCT also provides researchrants as follows: PhD research grant for Nuno BernardesSFRH/BD/48763/2008), Post doc research grant for Ana Sofiaibeiro (SFRH/BPD/75705/2011) and Programa Ciência 2007 for

oana Paredes [POPH – QREN – Tipology 4.2]. IPATIMUP and IBBre Associate Laboratories of the Portuguese Ministry of Science,echnology and Higher Education and partially supported by FCT.

ppendix A. Supplementary data

Supplementary data associated with this article can beound, in the online version, at http://dx.doi.org/10.1016/j.biocel.014.01.023.

eferences

bella JV, Park M. Breakdown of endocytosis in the oncogenic activation of receptortyrosine kinases. American Journal of Physiology Endocrinology and Metabolism2009;296:E973–84.

lbergaria A, Ribeiro AS, Vieira AF, Sousa B, Nobre AR, Seruca R, et al. P-cadherinrole in normal breast development and cancer. International Journal of Devel-opmental Biology 2011;55:811–22.

ernardes N, Ribeiro AS, Abreu S, Mota B, Matos RG, Arraiano CM, et al. Thebacterial protein azurin impairs invasion and FAK/Src signaling in P-cadherin-overexpressing breast cancer models. PLoS ONE 2013;8(7):e69023.

issell MJ, Hines WC, Berenblum I. Why don’t we get more cancer? A proposed roleof the microenvironment in restraining cancer progression. Nature Medicine2011;17:320–9.

racke ME, Van Larebeke N, Vyncke BM, Mareel MM. Retinoic acid modulates bothinvasion and plasma membrane ruffling of MCF-7 human mammary carcinomacells in vitro. British Journal of Cancer 1991;63:867–72.

usk M, Pytelas R, Sheppard D. Characterization of the integrin avP6 as a fibronectin-binding protein. Journal of Biological Chemistry 1992;267(9):5790–6.

iochemistry & Cell Biology 50 (2014) 1–9 9

Cariati M, Naderi A, Brown JP, Smalley MJ, Pinder SE, Caldas C, et al. Alpha-6 integrinis necessary for the tumourigenicity of a stem cell-like subpopulation within theMCF7 breast cancer cell line. International Journal of Cancer 2008;122:298–304.

Carpenter PM, Dao AV, Arain ZS, Chang MK, Nguyen HP, Arain S, et al. Motilityinduction in breast carcinoma by mammary epithelial laminin 332 (laminin 5).Molecular Cancer Research 2009;7:462–75.

Eccles SA. The epidermal growth factor receptor/Erb-B/HER family in normaland malignant breast biology. International Journal of Developmental Biology2011;44:685–96.

Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell2011;144:646–74.

Huang da W, Sherman B, Lempicki R. Systematic and integrative analysis oflarge gene lists using DAVID bioinformatics resources. Nature Protocols2009;4:44–57.

Jin F, Brockmeier U, Otterbach F, Metzen E. New insight into the SDF-1/CXCR4 axisin a breast carcinoma model: hypoxia-induced endothelial SDF-1 and tumorcell CXCR4 are required for tumor cell intravasation. Molecular Cancer Research2012;10:1021–31.

Jinka R, Kapoor R, Sistla PG, Raj TA, Pande G. Alterations in cell-extracellular matrixinteractions during progression of cancers. International Journal of Cell Biology2012;2012:219196.

Lipscomb E, Simpson KJ, Lyle SR, Ring JE, Dugan AS, Mercurio AM. The alpha6beta4integrin maintains the survival of human breast carcinoma cells in vivo. CancerResearch 2005;65:10970–6.

Luo M, Fan H, Nagy T, Wei H, Wang C, Liu S. Mammary epithelial-specific ablationof the focal adhesion kinase suppresses mammary tumorigenesis by affectingmammary cancer stem/progenitor cells. Cancer Research 2009:466–74.

Mehta RR, Yamada T, Taylor BN, Christov K, King ML, Majumdar D, et al. Acell penetrating peptide derived from azurin inhibits angiogenesis and tumorgrowth by inhibiting phosphorylation of VEGFR-2, FAK and Akt. Angiogenesis2011;14:355–69.

Paredes J, Stove C, Stove V, Milanezi F, Van Marck V, Derycke L, et al. P-cadherin isup-regulated by the antiestrogen ICI 182,780 and promotes invasion of humanbreast cancer cells. Cancer Research 2004;64:8309–17.

Paredes J, Albergaria A, Oliveira JT, Jerónimo C, Milanezi F, Schmitt FC. P-cadherinoverexpression is an indicator of clinical outcome in invasive breast carcino-mas and is associated with CDH3 promoter hypomethylation. Clinical CancerResearch 2005;11:5869–77.

Paredes J, Correia AL, Ribeiro AS, Albergaria A, Milanezi F, Schmitt FC. P-cadherinexpression in breast cancer: a review. Breast Cancer Research 2007;9:214.

Punj V, Das Gupta TK, Chakrabarty AM. Bacterial cupredoxin azurin and its interac-tions with the tumor suppressor protein p53. Oncogene 2003;312:109–14.

Ribeiro A, Albergaria A, Sousa B, Correia A, Bracke M, Seruca R, et al. Extracellu-lar cleavage and shedding of P-cadherin: a mechanism underlying the invasivebehaviour of breast cancer cells. Oncogene 2010;29:392–402.

Ribeiro AS, Sousa B, Carreto L, Mendes N, Nobre AR, Ricardo S, et al. P-cadherinfunctional role is dependent on E-cadherin cellular context: a proof ofconcept using the breast cancer model. Journal of Pathology 2012;229(5):705–18.

Taylor BN, Mehta RR, Yamada T, Lekmine F, Christov K, Chakrabarty AM, et al. Non-cationic peptides obtained from azurin preferentially enter cancer cells. CancerResearch 2009;69:537–46.

Vieira AF, Ricardo S, Ablett MP, Dionísio MR, Mendes N, Albergaria A, et al. P-cadherinis coexpressed with CD44 and CD49f and mediates stem cell properties in basal-like breast cancer. Stem Cells 2012;30:854–64.

Willmarth NE, Ethier SP. Autocrine and juxtacrine effects of amphiregulin onthe proliferative, invasive, and migratory properties of normal and neoplastichuman mammary epithelial cells. Journal of Biological Chemistry 2006;281:37728–37.

Yamada T, Goto M, Punj V, Zaborina O, Chen ML, Kimbara K, et al. Bacterial redox pro-tein azurin, tumor suppressor protein p53, and regression of cancer. Proceedingsof the National Academy of Science 2002;99:14098–103.

Yamada T, Hiraoka Y, Ikehata M, Kimbara K, Avner BS, Das Gupta TK, et al. Apo-ptosis or growth arrest: modulation of tumor suppressor p53 specificity bybacterial redox protein azurin. Proceedings of the National Academy of Science2004;101:4770–5.

Yamada T, Fialho AM, Punj V, Bratescu L, Das Gupta TK, Chakrabarty AM. Internal-

specificity. Cellular Microbiology 2005;7:1418–31.Yamada T, Mehta RR, Lekmine F, Christov K, King ML, Majumdar D, et al. A peptide

fragment of azurin induces a p53-mediated cell cycle arrest in human breastcancer cells. Molecular Cancer Therapeutics 2009;8:2947–58.