RESEARCH ARTICLE

Specific upregulation of RHOA and RAC1 in cancer-associatedfibroblasts found at primary tumor and lymph node metastaticsites in breast cancer

Patricia Bortman Rozenchan1,2,3& Fatima Solange Pasini1 & Rosimeire A. Roela1 &

Maria Lúcia Hirata Katayama1 & Fiorita Gonzáles Lopes Mundim4&

Helena Brentani5 & Eduardo C. Lyra6 & Maria Mitzi Brentani1

Received: 2 April 2015 /Accepted: 28 June 2015# International Society of Oncology and BioMarkers (ISOBM) 2015

Abstract The importance of tumor–stromal cell interactionsin breast tumor progression and invasion is well established.Here, an evaluation of differential genomic profiles ofcarcinoma-associated fibroblasts (CAFs) compared to fibro-blasts derived from tissues adjacent to fibroadenomas (NAFs)revealed altered focal adhesion pathways. These data werevalidated through confocal assays. To verify the possible roleof fibroblasts in lymph node invasion, we constructed a tissuemicroarray consisting of primary breast cancer samples andcorresponding lymph node metastasis and compared the ex-pression of adhesion markers RhoA and Rac1 in fibroblastslocated at these different locations. Two distinct tissue micro-arrays were constructed from the stromal component of 43primary tumors and matched lymph node samples, respective-ly. Fibroblasts were characterized for their expression of α-smooth muscle actin (α-SMA) and vimentin. Moreover, weverified the level of these proteins in the stromal compartment

from normal adjacent tissue and in non-compromised lymphnodes. Our immunohistochemistry revealed that 59 % of fi-broblasts associated with primary tumors and 41 % of therespective metastatic lymph nodes (p=0.271) displayed posi-tive staining for RhoA. In line with this, 57.1 % of fibroblastsassociated with primary tumors presented Rac1-positive stain-ing, and the frequency of co-positivity within the lymph nodeswas 42.9 % (p=0.16). Expression of RhoA and Rac1 wasabsent in fibroblasts of adjacent normal tissue and in compro-mised lymph nodes. Based on our findings that no significantchanges were observed between primary and metastaticlymph nodes, we suggest that fibroblasts are active partici-pants in the invasion of cancer cells to lymph nodes and sup-port the hypothesis that metastatic tumor cells continue todepend on their microenvironment.

Keywords Breast cancer . Tumormicroenvironment .

Carcinoma-associated fibroblasts . RhoGTPases .Metastasis

Introduction

Stromal cells, including fibroblasts, endothelial cells, and im-mune cells, play a critical role in supporting breast cancer(BC) growth, survival, and invasion. Fibroblasts representthe major cell type of the stromal compartment and play animportant role in coordinating interactions between stromaland tumor cells by modulating the composition and functionof the extracellular matrix. Carcinoma-associated fibroblasts(CAFs) present different characteristics from those of fibro-blasts found in normal breast tissue (NAFs). CAFs form aheterogeneous population [1], express alpha-smooth muscleactin (α-SMA) upon activation [2], and impact BC biologicalbehaviors [3].

* Patricia Bortman Rozenchanpbrozenchan@hotmail.com

1 Radiology and Oncology Department, School of Medicine of SãoPaulo University, Av. Dr. Arnaldo, 455, sala 4112, SãoPaulo, SP CEP 01246-903, Brazil

2 Colsan—Blood Bank Beneficent Association, São Paulo, SP, Brazil3 Gynecological Department, School ofMedicine of São Paulo Federal

University, São Paulo, SP, Brazil4 Pathology Department, Vale do Sapucaí University, Pouso

Alegre, MG, Brazil5 Psychiatric Department, School of Medicine of São Paulo University

(USP), Rua Ovideo Pires de Campos S/N, São Paulo, SP, Brazil6 Brazilian Institute of Cancer Control, Av. Alcântara Machado 2576,

São Paulo, SP CEP 03102-002, Brazil

Tumor Biol.DOI 10.1007/s13277-015-3727-1

The importance of interactions between tumor cells and thesurrounding stromal cells [4] has been well established. Inparticular, the migratory/invasive behavior of tumor cells hasbeen reported to be influenced by bidirectional signals be-tween cancer cells and tumor-associated stroma fibroblastsand affects human breast cancer cell adhesion, migrationspeed, and direction [5, 6]. During cell invasion, fibroblastsuse contractile force and proteolytic activity to reorganize col-lagen into linear fibers to generate tracks for migration ofcancer cells [7]. Notably, family members of the Rho-smallguanosine triphosphatases (GTPases), RhoA, RAC1, andRAC2 binding proteins, have been implicated in CAF-mediated remodeling of the tumor microenvironment in amanner that enhances cancer cell invasion [8]. Indeed, RhoA,Rac1, and Rac2 induce stress fiber formation whenoverexpressed in fibroblasts [9].

In the present study, we analyzed genes that were differen-tially expressed in primary breast CAFs compared to fibro-blasts that originated from adjacent tissue of benign breastdiseases. This analysis revealed numerous differences ingenes involved in major functional pathways, including focaladhesion and regulation of actin cytoskeleton and tight junc-tions. Specifically, the GTPase family members RhoA,RAC1, and RAC2, as well as collagens and integrins, whichare known to mediate migration and invasiveness, wereshown to be differentially regulated in the context of CAFs.To date, increased expression of these proteins, particularlywithin the epithelial compartment, has been shown to be im-portant for cellular motility, loss of adhesion, invasion, andmetastasis [10–13].

Nodal status represents one of the most powerful inde-pendent prognostic indicators of breast carcinoma [14].Histologically, the presence of fibroblasts in the BC micro-environment of metastatic lymph nodes further reinforcesthe role that CAFs play in tumor growth and dissemination[15]. Le Bedis et al. [16] suggested that lymph node stromaplays an active part in the process of lymph node metasta-sis by creating a dynamic microenvironment that mimicsthe environmental conditions present at the primary tumorsite. Previous studies by Garcia et al. [17] reported a sim-ilar expression of metalloproteases within primary tumorsand the respective lymph nodes. Moreover, tumor–stromacross-talk seems to influence the metastatic lymph nodemicroenvironment, affecting proliferation and migrationof cancer cells [16, 18].

Because the genomic profiles of fibroblasts BC primaryand lymph node metastasis have been reported to be similar[19], we hypothesized that similar levels of RhoA and Rac1expression in the tumor microenvironment at the primary siteand at the lymph nodes may represent an advantage to tumorcell behavior. In this study, we determined whether the expres-sion of RhoA and Rac1 in stromal fibroblasts of primary tu-mors was similar to that found in lymph node fibroblasts.

Using a tissue microarray (TMA) consisting of primary BCsamples and the corresponding lymph node metastasis, wecompared the expression level of these markers in fibroblastsresiding at the two locations, as well as in adjacent normaltissue of primary tumors and in non-compromised lymphnodes.

Materials and methods

Primary cell culture tissue samples

Malignant and benign breast tissue specimens were obtainedfrom consenting patients undergoing surgery for breast dis-ease. Carcinoma samples were obtained from four patientsclinically staged as IIa, and benign samples were obtainedfrom four patients diagnosed as fibroadenoma. All tissue do-nors were patients at Instituto Brasileiro de Controle do Cân-cer, São Paulo, Brazil, a reference center for cancer treatment.This study was approved by the Ethical Institutional Commit-tee, and written explicit informed consent was obtained fromall participants. Invasive breast cancer was confirmedhistopathologically.

Primary cell culture

Fibroblasts were obtained from normal adjacent tissue sam-ples from patients with benign breast diseases (NAF) orwith primary invasive breast cancer tumors (CAF). H&E-stained, frozen histological sections were prepared fromeach tissue sample to confirm benignity or malignancy.After adipose tissue removal, tissue was minced inphosphate-buffered saline (PBS), washed twice in PBS(Na2HPO4 10 mM, NaCl 1.37 mM, KCl 27 mM, KH2PO4

2 mM, Thermo Fisher Scientific Inc., MA, USA) and inculture medium, and then chopped into small 1- to 4-mm3

pieces under sterile conditions. A total of 15–30 fragmentswere transferred to a T25 culture flask (Thermo Fisher Sci-entific Inc.) and covered with Dulbecco’s modified Eagle’smedium (DMEM, Thermo Fisher Scientific Inc.) supple-mented with 20 % FBS (Thermo Fisher Scientific Inc.),100 μg/mL ampicillin, 100 μg/mL streptomycin, and2.5 μg/mL fungizone and maintained at 37 °C in a humid-ified atmosphere containing 5 % CO2. Outgrowth of cellswas recorded after 10 to 20 days, and the medium wasrenewed once or twice a week thereafter. After sufficientoutgrowth, the tissue fragments were removed and the cellswere passaged by mild trypsinization with trypsin 0.5 %(Thermo Fisher Scientific Inc.).

Early passages (passage 3) of all fibroblasts weresubjected to immunocytochemical evaluation. The prolif-eration rates and description of patients were describedbefore [20].

Tumor Biol.

cDNA microarray assembly, hybridization, and analysis

Total RNA from four NAFs and four CAFs were extracted byTRIzol reagent (Thermo Fisher Scientific Inc.) and purifiedwith RNeasyminicolumns and reagents (Qiagen, Hilden, Ger-many). RNA microarray analysis was performed with 10 μgof biotin-labelled cRNA target prepared by a linear amplifica-tion method from a pool of samples. The poly (A)+ RNA(mRNA) subpopulation within the total RNA population wasprimed for reverse transcription by a DNA oligonucleotidecontaining the T7 RNA polymerase promoter 5′ to a d (T)24sequence. After second-strand cDNA synthesis, the cDNAserved as the template for an in vitro transcription (IVT) reac-tion to produce the target cRNA. This cRNAwas hybridizedusing CodeLinkTMHumanWhole Genome 55KBioarray (GEHealthcare, Buckinghamshire, UK), and the hybridization sig-nals were normalized using the CodeLinkTM System SoftwareAnalysis (GE Healthcare) and subjected to a t test with 10,000permutations. Genes with twofold differential expressionlevels in the CAF versus NAF comparison were considereddifferentially expressed. The gene ontology (GO) analysis wasperformed using the GO Tree Machine tool (GOTM) [21],which identifies hyper-represented categories in our gene listsas well as the Kyoto Encyclopedia of Genes and Genomes(KEGG) pathways. For both GO and KEGG, we used ahyper-geometric distribution with a p value≤0.05 [Onto-tolls].

Fluorescence microscopy

Cells were cultured on glass coverslips in 24-well plates. After24 h, cells were fixed with 3.5 % paraformaldehyde in PBS atroom temperature, permeabilized for 5 min with 0.2 % TritonX-100 in PBS, and blocked with 5 % bovine serum albumin(BSA, Thermo Fisher Scientific Inc.) in PBS for 1 h. Theprimary chicken polyoclonal RhoA antibody (ab23687,Abcam Antibodies, Cambridge, UK) and mouse monoclonalRac1 antibody (ab33186, Abcam Antibodies) were incubatedovernight at 4 °C. After extensive washes and secondary an-tibody staining (goat anti-mouse IgG, anti-chicken IgG, andanti-rabbit IgG; Sigma-Aldrich, MO, USA), nuclei werestained with 4′-6′-diamidino-2-phenylindole (DAPI, Sigma-Aldrich). Cells were analyzed using a Zeiss 510 META con-focal laser scanning microscope using a 488-nm argon and a543-nm HeNe laser. Images were acquired using a PlanNeoFluoar 40×/1 lens (Microimaging Inc., Lane Cove, NewZealand).

Immunohistochemistry tissue samples

This study included 43 breast cancer samples that involvedaxillary nodes from patients undergoing breast surgery at Hos-pital Samuel Libânio, in Pouso Alegre, MG, Brazil from 1997to 2005. This study was approved by the Institutional Ethics

Committee of Hospital Samuel Libânio (no. 1119/09). Themedian age of patients was 58 years (range 38–88 years). Allpatients were diagnosed with invasive ductal carcinomas atclinical stages II and III in 37.2 or 62.8% of cases, respectively(two unknown). Statuses of the estrogen receptor (ER), pro-gesterone receptor (PR), HER-2, and histological grade (HG)are listed in Table 1. HER-2 was assessed by the HercepTestTM

(Dako, Glostrup, Denmark A/S) system. Membrane-basedstaining cases with scores 0–1 or chromogenic in situ hybrid-ization (CISH) negative were categorized as negative, whilecases with score ≥3 were considered as positive. The caseswith an immunohistochemistry score of 2 were further con-firmed by CISH analysis (Ventana Medical Systems, Inc., amember of the Roche Group, Tucson, AZ) [22].

Additionally, we included 10 female BC cases (aged be-tween 39–84 years) in spite of analyzing the stromal compart-ment of adjacent normal tissue of primary tumors and notcompromised lymph nodes. All samples were subjected to apathology analysis, where tumor margins were considered toassure consistency within the tissues. The PathologyReporting of Breast Disease was used as guidelines [23].

Construction of TMA

Two distinct TMAs were constructed from the stromal com-ponent of primary tumors and lymph node samples,

Table 1 Clinical and pathological parameters of 43 breast cancerpatients

Characteristic No. of patients (%)

Age median (range) 58 (38–88)

Clinical stage

I 1 (2.3)

II 16 (37.2)

III 26 (60.5)

Histologic grade

Grade I 9 (20.9)

Grade II 15 (34.9)

Grade III 19 (44.2)

ER—primary tumor

Negative 18 (41.5)

Positive 24 (55.8)

Missing 1 (2.3)

PR—primary tumor

Negative 17 (39.5)

Positive 21 (48.8)

Missing 5 (11.6)

HER-2

Negative 33 (76.7)

Positive 9 (20.9)

Missing 1 (2.3)

Tumor Biol.

respectively. Representative areas of each component weresurrounded by a marking pen at the donor blocks and collect-ed using the Manual Tissue Arrayer I (Beecher InstrumentsInc., Sun Prairie, USA). Samples were orderly arranged in agrid, and the first core was represented by a fragment or nor-mal liver used as a reference in both TMAs. The first TMAwas built with samples of stromal component of the tumors inorder to enable the assessment of fibroblast cells within thedesmoplastic contingent of the carcinomas.

The second TMA consisted of 43 majorly compromisedlymph node samples from cases with axillary lymph nodemetastasis. Because the width of the metastatic element atthe lymph nodes was frequently limited to 0.5 to 1.0 cm, itwas seldom possible to separate distinct areas of stromal andepithelial elements. The chosen areas at the donor block in-volved stromal and epithelial elements, which were furtherevaluated separately.

From each of the two TMAs, 3-μm-thick slices were ob-tained and collected on slides with special adhesives(Instrumedics Inc., NJ, USA). Two sets of triplets, separatedby a gap of 40 cuts, were submitted for immunohistochemicalanalyses in order to represent two levels of the same sample.

Immunohistochemistry

The immunohistochemical reactions were performed usingthe complex streptavidin-biotin peroxidase (StreptABC, DakoCorporation, Glostrup, Denmark). After deparaffinization oftissue sections, antigen retrieval was performed using a pres-sure cooker in citrate buffer pH 6.0, followed by blockingendogenous peroxidase with hydrogen peroxide solution(3 %). The sections were incubated with primary antibodies,Rac1 (1:1,000; ab33186, Abcam) and RhoA (1:175; ab23687,Abcam), vimentin (1:2,000; clone v9, Dako Corporation), andα-SMA (clone: HHF-35; Cell Marque, CA, USA). After in-cubation with primary antibody and primary blocking, a poly-mer–peroxidase (Novolink, Leica, Wetzlar, Germany) ampli-fication step was performed. Antigen detection was carriedout in a solution containing 3,3-diaminobenzidine (Sigma-Aldrich) and 6 % H2O2. Counterstaining was performed withHarris hematoxylin (Merck, NJ, USA). Carcinoma ductal in-vasive of breast was used as positive control for RhoA, andlarge intestine adenocarcimona was the positive control usedfor Rac1. Negative controls were performed removing firstantibodies.

Evaluation of immunohistochemical essays

All reactions were assessed by two independent blinded ob-servers. Disparities between the two pathologists werereevaluated by consensus. Results from the epithelial compo-nent and stromal component were reported separately either tocarcinomas or to lymph node samples.

The presence of RhoA and Rac1 staining was assessed inthe stromal population and separately in primary tumors or inlymph node metastases. In relation to the immunohistochem-ical result evaluation, RhoA was scored according to theAllred system [24]. Samples with a score above 4 were con-sidered positive. For Rac1 evaluation, the percentage of pos-itive cells was assessed and classified into three groups: (1)0 %, (2) 1−33 %, and (3) ≥34 %. Samples with greater than10% of cells that were positive forα-SMA and vimentin wereclassified as positive. For all cases, cells presented in 10 fieldsat a magnification of ×400 were counted.

Statistical methods

Correlations between categorical antigen expression and otherclinicopathological parameters were studied with the Fisher’sexact test or chi-square test, where appropriate. Spearman’srank correlation coefficient was calculated to assess the rela-tionships involving categorical antigen expression. All statis-tical tests were two-sided, with significance defined asp<0.05. Analyses were performed using the software SPSSversion 10.0 for Windows (SPSS Inc., IL, USA).

Results

First, we evaluated the gene expression profile of stromal fi-broblasts derived from four benign (NAF) and four malignant(CAF) breast tissues using the CodeLinkTM Human WholeGenome 55K Bioarray. Using a twofold cutoff, we found 1,111 genes differentially expressed in CAFs compared toNAFs. To further examine the biological functions of thesegenes, an analysis of the KEGG database revealed differentialexpression of several pathways, including Boxidative phos-phorylation,^ “focal adhesion,” BMAPK signaling pathway,^Bleukocyte transendothelial migration,^ Bregulation of actincytoskeleton,^ and Btight junctions^ (Fig. 1). Interestingly,all of these pathways are involved in important processes likecell/cell communication, migration, and invasiveness. Focaladhesion was ranked the second top significant pathway, andin microarray analysis, we found that both RhoA and Rac1were upregulated in CAFs compared to NAFs.

Using confocal assays, we confirmed that protein levels ofRhoA and Rac1 were elevated in carcinoma-associated fibro-blasts (Fig. 2). Strong RhoA staining was localized to thecytoplasm, while Rac1 staining was associated with the plas-ma membrane.

To further examine correlations between altered genesfound in the microarray data with patterns of protein expres-sion, we analyzed the expression of RhoA and Rac1 in thestromal component of 43 samples of primary breast carcinomaandmetastatic ipsilateral axillary lymph nodes by immunohis-tochemistry. In a few metastatic lesions cases, it was not

Tumor Biol.

Fig. 1 Distribution ofdifferentially expressed genes inCAFs. Representative pathwayswere identified for genesdifferentially expressed (blackbars) between CAFs and NAFs inour cDNA microarray platform(gray bars). The gene ontology(GO) analysis was performedusing the GO Tree Machine tool(GOTM), which identifies hyper-represented categories in our genelists as well as the KyotoEncyclopedia of Genes andGenomes (KEGG) pathways. Forboth GO and KEGG, we used ahyper-geometric distribution witha p value≤0.05 [Onto-tolls]

Fig. 2 CAF-specific staining ofRhoA and Rac1. CAF (a, c) andNAF (b, d) confocalphotomicrographs of typicalfields are shown. Merged imagesof a cytoplasmatic staining ofRhoA (a, b) and cytoplasmaticmembrane staining of Rac1 (c, d)validated the gene expressionanalysis; Rac1 and RhoAwereupregulated in CAFs

Tumor Biol.

possible to count the preferred number of cells. These proteinswere also analyzed in fibroblasts of adjacent normal tissue ofprimary tumors and in non-compromised lymph nodes of 10breast cancer cases.

To distinguish fibroblasts (CAFs) from tumoral cells, weevaluated two fibroblast cell markers: vimentin and α-SMA.Vimentin staining was observed in 37/43 (86.0 %) of fibro-blasts of primary tumors and in 35/41 (85.4%) of lymph nodalfibroblasts. α-SMA staining was observed in 51.2 % of pri-mary lesions and in 41 % of nodal metastatic lesions (Fig. 3).

Positive RhoA staining in stromal cells was seen in 36/43cases of primary tumors (59 %) and in 25/35 (41 %) of respec-tive metastatic lymph nodes (p=0.271). Similarly, Rac1 stain-ing in stromal cells was observed in 28/39 cases of primarytumors (57.1 %), whereas the frequency of Rac1 staining instromal cells found in the lymph nodes was 21/38 (42.9 %)(Table 2). In summary, the status of RhoA and Rac1 wassimilar between primaries and lymph nodes (Fig. 4), and allmarkers were detected in the cytoplasm. Importantly, RhoA orRac1 expression was not detected in fibroblasts of adjacentnormal tissue or in non-compromised lymph nodes (Fig. 5),suggesting that the presence of RhoA or Rac1 in CAFs mayhelp facilitate tumorigenesis.

Discussion

A large amount of data have shown that the microenvironmentis important to breast carcinoma epithelial cells [25–29]. Ad-ditionally, we have shown that reciprocal changes in geneexpression profiles of important cellular function pathwaysoccur when breast epithelial cells were co-cultured with

primary fibroblasts, reinforcing the idea that interactions be-tween these two cell types impact cell signaling and behavior[20]. In line with these works, we compared gene expressiondata generated in CAFs and NAFs to show that a large numberof genes present in the focal adhesion pathways, including theGTPAses RhoA and Rac1, were differentially expressed inCAFs.

Cell motility and invasiveness require cytoskeleton reorga-nization, which involves formation of the filopodia/lamellipodia and changes in focal adhesion complexes. Rhoproteins are involved in stress fiber formation and focal adhe-sion, while Rac proteins stimulate lamellipodia andmembrane-ruffle formation [30–33]. Increasing evidence sug-gests that the RhoA and Rac1 proteins play an important rolein cell migration, loss of adhesion, invasion, and metastasis intumors [10, 34–37]. Building on these studies, we used im-munohistochemistry to investigate whether the RhoA andRac1 expression patterns were similar in fibroblasts of prima-ry BC tumors and matched lymph node metastases.

Fig. 3 CAF-specific α-SMA andvimentin staining at primary atmetastatic tumor sites.α-SMA (a,b) and vimentin (c, d) staining inCAFs in the primary tumors (a, c)and in their counterpart lymphnode metastasis (b, d). Originalmagnification ×400

Table 2 Correlation among the proportion of biological markerexpression in stromal tissue between the primary tumors and lymphnode metastasis of 43 breast cancer patients

Variable Primary tumor Lymph node p

RhoA

Negative 7 (41.2) 10 (58.8) 0.27Positive 36 (59.0) 25 (41.0)

Rac1

Negative 11 (39.3) 17 (60.7) 0.16Positive 28 (57.1) 21 (42.9)

Stroma tissue: only fibroblast cells were analyzed; Values of p (two-sided)less than 0.05 were considered significant

Tumor Biol.

Early studies noted that RhoA overexpression result-ed in mouse fibroblast transformation [38] and stressfiber formation [39]. It was previously shown that high-ly metastatic mesenchymal sarcoma cells primarily usean ameboid mode of cell invasion that depends on theactivity of the Rho family of GTPases [40]. Some au-thors have described the role of Rho GTPases in fibro-blasts [41–43]. Verghese et al. [41] reported that cyto-skeletal regulation by Rho GTPases in breast fibroblastsenhanced migration and invasion in consequence of mir-26b dysregulation.

In agreement with Halon et al. [10], we show that a highproportion of cells in the BC tumor–stromal compartmentwere positive for RhoA and Rac1. In addition, we have shownthat activated fibroblasts, as identified by α-SMA, were pres-ent in the majority of metastatic lymph node cases, whereasnon-involved lymph nodes were devoid of myofibroblasts.

The epithelial-to-mesenchymal transition (EMT) repre-sents one possible mechanism by which a cancer cell metas-tasizes and may facilitate the re-localization of fibroblasts tothe lymph node. Past work has emphasized RhoA as a medi-ator of this process via integrin β1/TGF β activation [44].

Fig. 4 RhoA and Rac1 stainingin primaries and in lymph nodemetastasis. Representative casesdepicting RhoA and Rac1 (a, c) inCAFs in the primary tumors andin the respective lymph nodemetastasis (b, d). Originalmagnification ×400

Fig. 5 Lack of RhoA and Rac1staining in fibroblasts of adjacentnormal tissue and in non-compromised lymph nodes.Negative expression of RhoA andRac1 in fibroblasts of adjacentnormal tissue (a, c) and non-compromised lymph nodes (b, d).Original magnification ×400

Tumor Biol.

Moreover, EMT is a recognized source of CAFs, as it pro-duces myofibroblast cells with enhanced migratory capacity,invasiveness, and increased expression of ECM proteins [45].In line with these data, we found that 41 % of fibroblasts oflymph nodes origin were positive for RhoA and α-SMA.

CAFs may also arrive at lymph nodes in co-migration withtumor cells, referred to as the collective pathway of invasion[46]. Reorganization of collagen fibers by mammary fibro-blasts can create avenues for invasion [5], and the Rho-family of GTP-binding proteins may regulate fibroblast-mediated collagen reorganization [8]. Therefore, RhoA andRac1 may be involved in both pathways.

Our study brings some clarification regarding the role ofthe microenvironment in lymph node metastasis in BC. How-ever, we acknowledge that our study is limited by the numberof samples used in immunohistochemical assays, as well as bythe presence of non-distinct histological types, as differentsubtypes represent different fibroblasts populations [47].

In conclusion, we demonstrate positive expression ofRhoA and Rac1 in CAFs in paired primary breast cancerand in the respective lymph node metastasis, suggesting thata similar microenvironment may be present at both sites. Im-portantly, as we were unable to detect positive expression ofthese proteins in fibroblasts of normal adjacent tissue or innon-committed lymph nodes, our findings may highlight thestroma as an active participant in the metastatic process andsuggest that metastatic tumor cells may continue to be depen-dent on their supportive microenvironment.

Acknowledgments The authors are grateful to Ana Lúcia Garippo forher technical assistance in confocal microscopy. This research was sup-ported by Fundação de Amparo à Pesquisa no Estado de São Paulo(FAPESP) 01/13513-1, 05/51593-5, 04/04607-8, 05/60333-7, 2014/03090-3 and 09/10088-7 and Conselho Nacional de DesenvolvimentoCientífico e Tecnológico (CNPq).

Conflicts of interest None

Ethical approval All procedures performed in studies involving hu-man participants were in accordance with the ethical standards of theinstitutional and/or national research committee and with the 1964 Hel-sinki declaration and its later amendments or comparable ethicalstandards.

References

1. Sugimoto H, Mundel TM, Kieran MW, Kalluri R. Identification offibroblast heterogeneity in the tumor microenvironment. CancerBiol Ther. 2006;5(12):1640–6.

2. Xouri G, Christian S. Origin and function of tumor stroma fibro-blasts. Semin Cell Dev Biol. 2010;21(1):40–6.

3. Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer.2006;6(5):392–401.

4. Polyak K, Kalluri R. The role of the microenvironment in mamma-ry gland development and cancer. Cold Spring Harb Perspect Biol.2010;2(11):a003244.

5. Dang TT, Prechtl AM, Pearson GW. Breast cancer subtype-specificinteractions with the microenvironment dictate mechanisms of in-vasion. Cancer Res. 2011;71(21):6857–66.

6. Angelucci C, Maulucci G, Lama G, Proietti G, Colabianchi A, PapiM, et al. Epithelial-stromal interactions in human breast cancer:effects on adhesion, plasma membrane fluidity and migration speedand directness. PLoS One. 2012;7(12):e50804. doi:10.1371/journal.pone.0050804.

7. Conklin MW, Eickhoff JC, Riching KM, Pehlke CA, Eliceiri KW,Provenzano PP, et al. Aligned collagen is a prognostic signature forsurvival in human breast carcinoma. Am J Pathol. 2011;178(3):1221–32.

8. Rhee S, Grinnell F. Fibroblast mechanics in 3D collagen matrices.Adv Drug Deliv Rev. 2007;59(13):1299–305.

9. Wheeler AP, Ridley AJ. Why three Rho proteins? RhoA, RhoB,RhoC, and cell motility. Exp Cell Res. 2004;301(1):43–9.

10. Halon A, Donizy P, Surowiak P, Matkowski R. ERM/Rho proteinexpression in ductal breast cancer: a 15 year follow-up. Cell Oncol.2013;36(3):181–90.

11. Wu YJ, Tang Y, Li ZF, Li Z, Zhao Y, Wu ZJ, Su Q. Expression andsignificance of Rac1, Pak1 and Rock1 in gastric carcinoma. AsiaPac J Clin Oncol. 2013; doi: 10.1111/ajco.12052

12. Ridley AJ. Rho proteins and cancer. Breast Cancer Res Treat.2004;84(1):13–9.

13. Sahai E,Marshall CJ. Differingmodes of tumour cell invasion havedistinct requirements for Rho/ROCK signalling and extracellularproteolysis. Nat Cell Biol. 2003;5(8):711–9.

14. Yenidunya S, Bayrak R, Haltas H. Predictive value of pathologicaland immunohistochemical parameters for axillary lymph node me-tastasis in breast carcinoma. Diagn Pathol. 2011;6:18. doi:10.1186/1746-1596-6-18.

15. Cao Y, Paner GP, Rajan PB. Sentinel node status and tumor char-acteristics: a study of 234 invasive breast carcinomas. Arch PatholLab Med. 2005;129(1):82–4.

16. LeBedis C, Chen K, Fallavollita L, Boutros T, Brodt P. Peripherallymph node stromal cells can promote growth and tumorigenicityof breast carcinoma cells through the release of IGF-I and EGF. Int JCancer. 2002;100(1):2–8.

17. García MF, González-Reyes S, González LO, Junquera S, BerdizeN, Del Casar JM, et al. Comparative study of the expression ofmetalloproteases and their inhibitors in different localizations with-in primary tumours and in metastatic lymph nodes of breast cancer.Int J Exp Pathol. 2010;91(4):324–34.

18. Montel V, Mose ES, Tarin D. Tumor-stromal interactions recipro-cally modulate gene expression patterns during carcinogenesis andmetastasis. Int J Cancer. 2006;119(2):251–63.

19. Del Valle PR, Milani C, Brentani MM, Katayama ML, de Lyra EC,Carraro DM, et al. Transcriptional profile of fibroblasts obtainedfrom the primary site, lymph node and bone marrow of breast can-cer patients. Genet Mol Biol. 2014;37(3):480–9.

20. Rozenchan PB, Carraro DM, Brentani H, et al. Reciprocal changesin gene expression profiles of cocultured breast epithelial cells andprimary fibroblasts. Int J Cancer. 2009;125(12):2767–77.

21. Onto-Tools data base [http://vortex.cs.wayne.edu/projects.htm].22. Wolff AC, Hammond ME, Hicks DG, et al. Recommendations for

human epidermal growth factor receptor 2 testing in breast cancer:American Society of Clinical Oncology/College of AmericanPathologists clinical practice guideline update. Arch Pathol LabMed. 2014;138:241–56.

23. Pathology Reporting of Breast Disease. A Joint DocumentIncorporating the Third Edition of the NHS Breast ScreeningProgramme’s Guidelines for Pathology Reporting in BreastCancer Screening and the Second Edition of The Royal Collegeof Pathologists’ Minimum Dataset for Breast CancerHistopathology. NHSBSP Publication No 58. 2005.

Tumor Biol.

24. Allred DC, Harvey JM, Berardo M, et al. Prognostic and predictivefactors in breast cancer by immunohistochemical analysis. ModPathol. 1998;11:155–68.

25. Casbas-Hernandez P, Fleming JM, Troester MA. Gene expressionanalysis of in vitro cocultures to study interactions between breastepithelium and stroma. J Biomed Biotechnol. 2011; 520987. doi:10.1155/2011/520987

26. Hawsawi NM, Ghebeh H, Hendrayani SF, et al. Breast carcinoma-associated fibroblasts and their counterparts display neoplastic-specific changes. Cancer Res. 2008;68(8):2717–25.

27. Casey T, Bond J, Tighe S, et al. Molecular signatures suggest amajor role for stromal cells in development of invasive breast can-cer. Breast Cancer Res Treat. 2009;114:47–62.

28. Singer CF, Gschwantler-Kaulich D, Fink-Retter A, et al.Differential gene expression profile in breast cancer-derived stro-mal fibroblasts. Breast Cancer Res Treat. 2008;110(2):273–81.

29. Buess M, Nuyten DS, Hastie T, Nielsen T, Pesich R, Brown PO.Characterization of heterotypic interaction effects in vitro todeconvolute global gene expression profiles in cancer. GenomeBiol. 2007;8(9):R191.

30. Dummler B, Ohshiro K, Kumar R, Field J. Pak protein kinases andtheir role in cancer. Cancer Metastasis Rev. 2009;28(1–2):51–63.

31. Arias-Romero LE, Villamar-Cruz O, Pacheco A, Kosoff R, HuangM, Muthuswamy SK, et al. A Rac-Pak signaling pathway is essen-tial for ErbB2-mediated transformation of human breast epithelialcancer cells. Oncogene. 2010;29(43):5839–49.

32. Rider L, Oladimeji P, Diakonova M. PAK1 regulates breast cancercell invasion through secretion of matrix metalloproteinases in re-sponse to prolactin and three-dimensional collagen IV. MolEndocrinol. 2013;27(7):1048–64.

33. Shin YJ, Kim EH, Roy A, Kim JH. Evidence for a novel mecha-nism of the PAK1 interaction with the Rho-GTPases Cdc42 andRac. PLoS One. 2013;8(8):e71495. doi:10.1371/journal.pone.0071495.

34. Chan CH, Lee SW, Li CF, Wang J, Yang WL, Wu CY, et al.Deciphering the transcriptional complex critical for RhoA geneexpression and cancer metastasis. Nat Cell Biol. 2010;12(5):457–67.

35. Fritz G, Brachetti C, Bahlmann F, Schmidt M, Kaina B. RhoGTPases in human breast tumours: expression and mutation anal-yses and correlation with clinical parameters. Br J Cancer.2002;87(6):635–44.

36. Chang YW, Marlin JW, Chance TW, et al. RhoA mediatescyclooxygenase-2 signaling to disrupt the formation of adherensjunctions and increase cell motility. Cancer Res. 2006;66(24):11700–8.

37. Spiering D, Hodgson L. Dynamics of the Rho-family smallGTPases in actin regulation and motility. Cell Adhes Migr.2011;5(2):170–80.

38. Khosravi-Far R, Solski PA, Clark GJ, Kinch MS, Der CJ.Activation of Rac1, RhoA, and mitogen-activated protein kinasesis required for Ras transformation. Mol Cell Biol. 1995;15(11):6443–53.

39. Heasman SJ, Ridley AJ. Mammalian Rho GTPases: new insightsinto their functions from in vivo studies. Nat Rev Mol Cell Biol.2008;9(9):690–701.

40. Rösel D, Brábek J, Tolde O, Mierke CT, Zitterbart DP, Raupach C,et al. Up-regulation of Rho/ROCK signaling in sarcoma cells drivesinvasion and increased generation of protrusive forces. Mol CancerRes. 2008;6(9):1410–20.

41. Verghese ET, Drury R, Green CA, Holliday DL, Lu X, Nash C,et al. MiR-26b is down-regulated in carcinoma-associated fibro-blasts from ER-positive breast cancers leading to enhanced cellmigration and invasion. J Pathol. 2013;231(3):388–99.

42. Berenjeno IM, Bustelo XR. Identification of the Rock-dependenttranscriptome in rodent fibroblasts. Clin Transl Oncol. 2008;10(11):726–38.

43. Sells MA, Pfaff A, Chernoff J. Temporal and spatial distribution ofactivated Pak1 in fibroblasts. J Cell Biol. 2000;151(7):1449–58.

44. Tse JC, Kalluri R. Mechanisms of metastasis: epithelial-to-mesenchymal transition and contribution of tumor microenviron-ment. J Cell Biochem. 2007;101(4):816–29.

45. Aboussekhra A. Role of cancer-associated fibroblasts in breast can-cer development and prognosis. Int J Dev Biol. 2011;55(7–9):841–9.

46. Gaggioli C, Hooper S, Hidalgo-Carcedo C, et al. Fibroblast-ledcollective invasion of carcinoma cells with differing roles forRhoGTPases in leading and following cells. Nat Cell Biol.2007;9(12):1392–400.

47. Tchou J, Kossenkov AV, Chang L, Satija C, Herlyn M, Showe LC,et al. Human breast cancer associated fibroblasts exhibit subtypespecific gene expression profiles. BMCMedGenomics. 2012;5:39.

Tumor Biol.