Experimental and Molecular Pathology 92 (2012) 175–184

Contents lists available at SciVerse ScienceDirect

Experimental and Molecular Pathology

j ourna l homepage: www.e lsev ie r .com/ locate /yexmp

Antibody-directed neutralization of annexin II (ANX II) inhibits neoangiogenesis andhuman breast tumor growth in a xenograft model

Meena Sharma a, Marc R. Blackman b,c,d,e, Mahesh C. Sharma b,d,f,g,h,⁎a University of Pennsylvania, School of Medicine, PA, USAb Research Services, Veterans Affairs Medical Center, Washington DC, USAc Department of Medicine, George Washington University, Washington DC, USAd Department of Biochemistry and Molecular Biology, George Washington University, Washington DC, USAe Department of Medicine, Georgetown, Washington DC, USAf Department of Oncology Georgetown, Washington DC, USAg Department of Surgery, Drexel University College of Medicine, PA, USAh Department of Pathology, Drexel University College of Medicine, PA, USA

⁎ Corresponding author at: Director, Laboratory of Getabolism, Research Service (151), Veterans Affairs MedicWashington, DC 20422, USA. Fax: +1 202 518 4611.

E-mail address: Mahesh.Sharma@va.gov (M.C. Sharm

0014-4800/$ – see front matter. Published by Elsevier Idoi:10.1016/j.yexmp.2011.10.003

a b s t r a c t

a r t i c l e i n f o

Article history:Received 14 September 2011Available online 25 October 2011

Keywords:PlasminAnnexin IIBreast cancerNeoangiogenesistPA

Activation of the fibrinolytic pathway has long been associated with human breast cancer. Plasmin is themajor end product of the fibrinolytic pathway and is critical for normal physiological functions. The mecha-nism by which plasmin is generated in breast cancer is not yet fully described. We previously identifiedannexin II (ANX II), a fibrinolytic receptor, in human breast tumor tissue samples and observed a strong pos-itive correlation with advanced stage cancer (Sharma et al., 2006a). We further demonstrated that tissueplasminogen activator (tPA) binds to ANX II in invasive breast cancer MDA-MB231cells, which leads to plas-min generation (Sharma et al., 2010). We hypothesize that ANX II-dependent plasmin generation in breasttumor is necessary to trigger the switch to neoangiogenesis, thereby stimulating a more aggressive cancerphenotype. Our immunohistochemical studies of human breast tumor tissues provide compelling evidenceof a strong positive correlation between ANX II expression and neoangiogenesis, and suggest that ANX II isa potential target to slow or inhibit breast tumor growth by inhibiting neoangiogenesis. We now reportthat administration of anti-ANX II antibody potently inhibits the growth of human breast tumor in a xeno-graft model. Inhibition of tumor growth is at least partly due to attenuation of neoangiogenic activity withinthe tumor. In vitro studies demonstrate that anti-ANX II antibody inhibits angiogenesis on three dimensionalmatrigel cultures by eliciting endothelial cell (EC) death likely due to apoptosis. Taken together, these datasuggest that selective disruption of the fibrinolytic activity of ANX II may provide a novel strategy for specificinhibition of neoangiogenesis in human breast cancer.

Published by Elsevier Inc.

Introduction

Sprouting of new blood vessels (neoangiogenesis) from existingvasculature is the hallmark of progression and metastasis of humanbreast cancer (Weidner et al., 1992). Clinical studies suggest thatneoangiogenesis is an independent and highly significant prognosticindicator of overall and relapse-free survival in patients with early-stage breast cancer (Weidner et al., 1992). Not surprisingly, targetingneoangiogenesis has been a central focus of the development of singleand multi-drug chemotherapy for breast cancer (O'Reilly et al., 1996).To date, however, the molecular mechanism(s) that triggers neoan-giogenic activity in the breast tumor microenvironment remains

riatric Endocrinology and Me-al Center, 50 Irving Street, NW,

a).

nc.

poorly understood. Angiostatin (AS), an internal fragment of plasmin-ogen, is a powerful inhibitor of angiogenesis. Experimental studiesdemonstrated that recombinant AS inhibits human breast cancerand related bone metastases in xenograft models (O'Reilly et al.,1996). In an attempt to understand AS's mechanism, we discoveredits interaction with the endothelial cell surface receptor, ANX II(Tuszynski et al., 2002). We proposed that the ability of AS to inhibitneoangiogenesis is due to its interaction with cell surface ANX II. Werecently demonstrated selective expression of ANX II in highly inva-sive human breast cancer MDA-MB231 cells, and suggested thatANX II may play an important role in breast cancer progression andmetastasis by facilitating neoangiogenesis (Sharma et al., 2010). Ourinitial findings have been confirmed in various cancer model systemsincluding breast cancer (Chuthapisith et al., 2009) prostate cancer(Braden et al., 2009) angiosarcoma, (Syed et al., 2007) hepatocelluarcarcinoma(Yu et al., 2007), gastrointestinal cancer(Singh, 2007),oral carcinoma(Qi et al., 2007), clear-cell renal cell carcinoma (Ohno

176 M. Sharma et al. / Experimental and Molecular Pathology 92 (2012) 175–184

et al., 2009), renal carcinoma (Zimmermann et al., 2004), lung can-cer,(Brichory et al., 2001), head and neck cancer(Wu et al., 2002),and pancreatic cancer (Diaz et al., 2004).

Because of the frequent overexpression of ANX II in diverse clinicaland experimental models of cancer, investigators are trying to devel-op highly sensitive techniques such as immunosensors or ELISA toassess the utility ANX II in the early diagnosis (Ji et al., 2009; Kimet al., 2009).

Several groups have reported that cell surface ANX II regulates plas-min generation (Diaz et al., 2004) which facilitates extracellular matrix(ECM) degradation, and consequently cell invasion (Brownstein et al.,2004; Diaz et al., 2004) and migration (Tarui et al., 2002), biologicalactivities which are required for neoangiogenesis.

Previously we and others demonstrated that ANX II is criticalfor neoangiogenesis and tumor growth (Semov et al., 2005). ANX IIknockout mice exhibited defective neoangiogenesis (Ling et al.,2004). Many investigators have reported ANX II among the proteinsdifferentially regulated in neoangiogenesis, cancer and metastasis(Aitkenhead et al., 2002; Pei et al., 2007; Zhang et al., 2009) suggest-ing that ANX II may serve as a molecular signature of neoangio-genesis in the tumor progression. ANX II may provide a potentialtherapeutic target to inhibit neoangiogenesis and its dependent tumorgrowth and metastasis in patients with breast cancer and other ma-lignant diseases (Kesavan et al., 2009; Lima e Silva et al., 2010; Linget al., 2004; Zhang et al., 2009). The plasminogen/plasmin systemcomprises of plasminogen activators (PA) [tissue type (tPA) or uro-kinase type (uPA)], their receptors such as uPAR and ANX II respec-tively and plasminogen activator inhibitor (PAI). Although both tPAand uPA bind to their respective cognate receptors and convert inac-tive plasminogen to the highly reactive enzyme plasmin. To datemost of our knowledge about the plasmin generation in breast can-cer is derived from studies uPA/uPAR-dependent mechanism. ANXII is a well established receptor for tPA and known to regulate plas-min generation which physiologically dissolves intravascular fibrinclots, a process known as fibrinolysis (Cesarman et al., 1994). Al-though hyperfibrinolysis is a well-known phenomenon in patientswith advanced stage breast cancer however, the role of tPA/ANXII-mediated fibrinolytic pathway in breast cancer is not known.

Previously, we reported high levels of ANX II expression in advancedstage breast cancer patients concurrent with excessive tPA secretionin the tumor microenvironment (Sharma et al., 2010). These initialfindings from our laboratory have now been confirmed in clinicalstudies, which identified tPA as a single major factor that contributesto the development of metastasis in breast cancer patients (Nainaet al., 2010). Previous clinical studies have also observed an associa-tion of tPA with aggressive and metastatic breast cancer (Grondahl-Hansen et al., 1990; Rella et al., 1993) with neoplastic transformationand invasion of other cancers (Aguilar et al., 2004; Chernicky et al.,2005; Diaz et al., 2002; Goh et al., 2005; Stack et al., 1999).

In the present study, we investigated whether selective targetingof ANX II inhibits human breast cancer growth and neoangiogenesisin a xenograftmousemodel. To test this concept,we used human breastcancer MDA-MB231 cells, which are known to express high levels ofannexin II (Sharma et al., 2006a), to induce tumors in BALB/c nu/numice. Tumor bearing mice were then treated with anti-ANX II mono-clonal antibody to target cell surface ANX II. We now report that anti-ANX II antibody arrests xenograft breast tumor growth by inhibitingneoangiogenesis.

Materials and methods

Human Lys-plasminogen, plasmin and recombinant tPA were pur-chased from Calbiochem, (La Jolla, CA). Electrophoresis reagents wereprocured from BioRad, (Richmond, CA). Anti-tPA monoclonal anti-bodies were purchased from American Diagnostica (Stamford, CT).Antibodies to ANX II were generated in our laboratory as reported

earlier (Sharma et al., 2006b). Chromozyme PL was purchased fromRoche Molecular Biochemicals (Indianapolis, IN). Anti-CD31 mono-clonal antibodies and immunohistochemical staining kit were procuredfrom DAKO Corporation (Carpinteria, CA). Apoptosis kit was procuredfrom Oncogene Research Products (San Diego, CA). Angiostatin wasproduced in our laboratory (Tuszynski et al., 2002) and all otherchemicals used in this study were of analytical grade.

Cell culture and maintenance

The human invasive breast cancer MDA-MB231 cell line was pro-vided by Dr. George Tuszynski, Temple University in Philadelphia, PA.MDA-MB231 cells were grown in RPMI 1640 media containing 10%fetal calf serum (FCS) supplemented with L-glutamine and antibioticsas we reported earlier (Sharma et al., 2006a). Bovine Aortic Endothelial(BAE) cells were grown in Ham's F12 K containing either 10% fetal calfserum (FCS) or serum free media (0.1% BSA) supplemented with L-glutamine and antibiotics (Tuszynski et al., 2002).

Immunoflorescence staining

The surface immunofluorescence staining of MDA-MB231 cellswas performed as we described previously (Sharma et al., 2006b).For surface proteins on non-permeabilized MDA-MB231 cells, cellsgrown on gelatin-coated cover slips were washed three times inPBS and then incubated for 3 h at 4 °C with anti-annexin II monoclo-nal antibodies followed by FITC-conjugated secondary antibody. Cellswere washed three times with PBS and fixed in 3.7% formaldehydefor 5 min and mounted in Vectashield mounting medium (VectorLaboratories, Inc., Burlingame, CA). Cells were viewed under fluores-cence microscope using FITC filter.

Plasmin generation in tumor tissue

Pieces of tumor tissues were excised and snap frozen in liquidnitrogen for plasmin generation assay, as described previously, withminor modifications (Mulligan-Kehoe et al., 2001; Sharma et al.,2010). Tumor tissueswere homogenized in chilled PBS and protein con-centration was determined using the Micro BCA Protein Assay kit(Thermo Fisher Scientific, Rockford, IL). Ten μg of total proteinwas incubated with plasminogen (2 μM final concentration) andplasmin-specific chromogenic substrate (2 mM Chromozyme PL)in a total volume of 100 μl to determine activation of plasminogen(Brownstein et al., 2004; Sharma et al., 2006b). The change in color at405 nm is a direct measure of plasmin generation. Appropriate controlswere included in this experiment to determine non-specific plasmingeneration. Control well 1 was designed to assess self-degradation ofplasminogen to plasmin by incubating plasminogen with ChromozymePL. Control well 2 was included to assess any plasmin activity in tissuesby incubating tumor tissues with Chromozyme PL. Non-specific plas-min generated in these wells was subtracted to calculate tumor specificplasmin generation.

Immunohistochemistry

Tumor tissues were excised and immediately submerged in phos-phate buffered formalin, embedded in paraffin, and 4 μm sectionswere cut on albumin-coated slides. Staining was performed accord-ing to our established protocol (Sharma et al., 2006a, 2010). Briefly,sections were deparaffinized and incubated in 3% H2O2 for 10 minto block endogenous peroxidase activity. Nonspecific protein bind-ing was blocked with 3% BSA/PBS for 1 h. Sections were incubatedwith monoclonal antibodies (1:1000) overnight at room tempera-ture followed by HRP labeled secondary antibody (1:2500) for 1 h.Staining was visualized by diaminobenzidine (DAB) followed by nu-clear counterstaining with hematoxylin. In parallel experiments

177M. Sharma et al. / Experimental and Molecular Pathology 92 (2012) 175–184

control immunostaining was performed without primary antibody.The immunostaining pattern was visualized under a microscope,photographed and analyzed by ImagePro software (MediaCyber-netics, Bethesda, MD). Microvessel density (MVD) was determinedby counting the CD31 positive vessels from five 20× fields from 20different sections and statistical analysis was performed by GraphPadPrizm software (La Jolla, CA).

A

Xenograft model of human breast cancer

All experiments were performed on 4–6-week-old female nudemice purchased from the National Cancer Institute (Frederick, MD).Mice were acclimated to the animal facility for one week prior tostudy. Studies involving animals were conducted in accordance witha code of practice established by the Veterans Affairs Medical Center'sR&D committee and Institutional Animal Care and Use Committee(IACUC). These MDA-MB231 cells have been characterized in ourlaboratory for ANX II expression and plasmin generation (Sharmaet al., 2006a, 2010). MDA-MB231 cells (1×106 cells in 0.1 ml) wereimplanted in the hind flank as reported earlier (O'Reilly et al.,1996; Sharma et al., 2006b). When tumors were about 50–75 mm3,mice were randomly divided into two groups of five animals. Onegroup of mice was injected with anti-ANX II monoclonal antibody(Two doses of 10 mg/kg body weight in 0.1 ml, 15 days apart)and control group of mice was injected with equal amount ofmouse IgG through tail vein. Tumor growth was measured usinga digital caliper and tumor volume was calculated using the formu-la length×width2/2 as reported earlier (Sharma et al., 2006b). Atthe end of the experiment mice were sacrificed and tumors werephotographed and dissected for further analyses.

Control

B

In vitro angiogenesis assay

Because matrigel can spontaneously induce in vitro angiogenesis,we tested the effect of an antibody on in vitro matrigel model of an-giogenesis. The angiogenesis assay was performed as described(Mulligan-Kehoe et al., 2001). Briefly, 50 μl of matrigel was addedper well using 96-well tissue culture plates and allowed to gel at37 °C for 10 min. Low passage BAE cells starved for 24 h in Ham'sF-12 K medium containing 0.1% BSA were detached and gentlyadded as triplicates to wells and allowed to adhere to the gel coatingfor 30 min at 37 °C. After cells adhered on the matrigel, mediumcontaining anti-ANX II antibody (50 μg/ml) was added. Angiostatin(10 μg/ml) and IgG were included in parallel wells as positive andnegative controls, respectively. Plates were incubated for 24–36 hand viewed under a microscope and photographed.

Anti-Annexin II

Fig. 1. ANX II expresses on the surface of invasive human breast cancer cells (MDA-MB231): MDA-MB231 cells were cultured on cover slips and stained with anti-ANX IIantibody followed by FITC labeled secondary. Immunofluorescence staining demon-strated ANX II expression mainly on the cell surface (Panel B, arrows). (A) Isotypematched IgG control staining. Magnifications 200×.

Apoptosis assays

Exposure of phosphatidyl serine (PS) on the cell surface is an indi-cator of apoptosis (Koopman et al., 1994). Cell surface exposed PSwas detected by annexin V immunostaining after treatment of testcompounds. Adherent cells were washed and detached with 0.5×trypsin and centrifuged. The cell pellet was rewashed three timeswith PBS and suspended in PBS. A double immunostaining techniquewas used to detect apoptotic cells using a kit fromOncogene researchproducts (San Diego, CA) according to the manufacturer's instruc-tions. Briefly, cell suspensions were double labeled with FITC-annexinV followed by propidium iodide (PI). The cell suspensions (25–50 μl)were transferred to a microscopic slide and viewed under a fluorescentmicroscope equipped with FITC/Rhodamine filters (Zeiss, Germany).Control cells included in a parallel experiment were treated with IgGand angiostatin as negative and positive controls, respectively.

Endothelial cell viability assay

Promega's cell proliferation/viability assay kit was used to quanti-fy BAE cell viability according to the manufacturer's instructions. Thisassay determines the metabolic activity of live cells by a dehydroge-nase enzyme, which converts MTS (Owen's reagent) into solubleformazan. Anti-ANX II antibody was incubated with BAE cell culturefor 72 h. Relative cell numbers in the absence or presence of anti-bodies were determined by adding 20 μl of viability assay reagentfor 30–60 min or until brown color appeared. The intensity ofbrown color in culture medium is proportional to the number of livecells and was measured at 490 nm in 96-well plate reader

Results

Anti-ANX II antibody inhibited MDA-MB231-induced human breast tumorgrowth in nude mice

Immunofluorescence staining confirmed presence of ANX II onthe surface of MDA-MB231 cells (Fig. 1B). Previously, we and othersreported the presence of ANX II on the surface of Bovine Aortic Endo-thelial Cells (BAEC) (Tuszynski et al., 2002; Zimmermann et al., 2004).

178 M. Sharma et al. / Experimental and Molecular Pathology 92 (2012) 175–184

Consistently, our immunohistochemical analyses demonstrated intensestaining of ANX II mainly on the surface of human breast cancer tissues(Fig. 6B). Immunofluorescence staining confirmed presence of ANX IIon the surface of MDA-MB231 cells (Fig. 1B). Previously we and othershave showed that ANX II is involved in breast cancer progression.

Human MDA-MB231 cells are well-characterized for their ability toform highly vascularized breast tumors in nude mice (Price et al., 1990).MDA-MB231 cells expressing high levels of ANX II were implantedin the flank of nude mice to induce tumor growth. We investigatedwhether targeted disruption of ANX II inhibits neoangiogensis andconsequently tumor growth. Mice bearing MDA-MB231-inducedbreast tumors were treated with anti-ANX II monoclonal antibody.Our data suggested that antibody treatment significantly inhibitedneoangiogenic activity in the tumor microenvironment and resultedin inhibition of tumor growth (Figs. 2A, C and E).

Previous clinical studies have shown that breast tumor growthand invasion are dependent on tumor neoangiogenic activity (Weidneret al., 1992). Interestingly ANX II has been shown to regulate neoan-giogenic activity in vivo (Lima e Silva et al., 2010; Ling et al., 2004).

A

0 25 50 75 100 1250

250

500

750

1000 Control (mouse IgG)

Anti-annexin II monoclonal

Days after tumor implantation

Tum

or V

olum

e (m

m3 )

C

E

Fig. 2. Anti-neoangiogenic and anti-breast tumor effect of Anti-ANX II antibody in nude micemice and treated when palpable tumors appeared. Panel A (arrows) exhibits breast tumor gstrate neoangiogenesis (arrows). Two treatments of anti-ANX II (10 mg/kg of B.W.) significatreated with antibody exhibited pale due to lack of blood (D). Control mice treated with istumor growth (Panels A, B and E). Data were analyzed and plotted using GraphPad Prizm s

We investigated whether selective blocking of ANX II in this mousemodel of breast tumor inhibits neoangiogenesis and tumor growth.Our results demonstrated that MDA-MB231 cells were able to inducerapid tumor growth with intense neoangiogenic activity (Figs. 2Aand B). Tumor-induced neoangiogenic activity as well as tumorgrowth was significantly inhibited by anti-ANX II antibody treatment(Figs. 2C and D). It is noteworthy that the control tumor showingsprouting of numerous new blood vessels from existing vessels (Fig. 2B,arrows) and appeared red due to excessive blood supply. In contrast,anti-ANX II treated tumor appeared pale because of lack of blood supplypossibly due to disruption of neoangiogenesis (Fig. 2D). To confirmthe inhibition of neoangiogenic activity, tumor tissues were immuno-stained with endothelial cell marker CD31. Results in Figs. 3D and Eshowed that anti-ANX II treatment significantly inhibited neoangio-genic activity in the tumor microenvironment as compare to isotypematched IgG controls (Fig. 3C). H&E staining of tumor tissue exhibitedtumor necrosis after anti-ANX II treatment (Fig. 3B). These datastrongly suggests that inhibition of breast tumor growth is likely dueto inhibition of tumor neoangiogenic activity (Ling et al., 2004).

B

D

bearing human breast tumors: MDA-MD231 cells were implanted in the flank of nuderowth with visible neoangiogenesis, (B) skin was removed to expose tumor to demon-ntly inhibited visible neoangiogenesis and tumor growth (C, D and E). Tumors of miceotype matched IgG exhibited intense neovascularization which supported exponentialoftware and are mean±SD of five animals. Experiment was repeated two times.

H&E staining

A

B

CD31 staining

0

100

200

300

400 Control (IgG treated)

Anti-ANX II treated

Num

ber

of v

esse

ls

**

D

H&E staining CD31 staining

C

lortnoClortnoC

Anti-ANX II treated

E

Anti-ANX II treated

Fig. 3. Anti-ANX II antibody inhibited breast tumor-induced neoangiogenic activity: tumor tissues were excised out, fixed in buffered formalin and paraffin embedded for histologyand quantification of neoangiogenic activity. Panels A and B exhibited H&E staining of tumor tissue sections from control (IgG) and anti-ANX II treated mice respectively (magni-fication 100×). Panels C and D demonstrated anti-CD31 immunostaining of tumor tissue sections (brown color, arrows). Anti-ANX II treatment significantly inhibited neoangio-genic activity which is demonstrated by anti-CD31 positive blood vessels in the tumor microenvironment (D, arrows). To analyze the microvessel density (MVD) in the tumormicroenvironment anti-CD31 positive blood vessels were counted using ImagePro software and statistically analyzed by GraphPad Prizm software **=pb0.05(E). Antibody treat-ment demonstrated tumor necrosis (B).

179M. Sharma et al. / Experimental and Molecular Pathology 92 (2012) 175–184

Anti-ANX II antibody inhibited plasmin generation in experimentalbreast tumors

Previously we reported over expression of ANX II in invasivebreast cancer patients and its correlation with disease progression(Sharma et al., 2006a). Consistently ANX II expression has been reportedto directly relate with poor clinical outcome in renal cell carcinoma andbrain tumor (Roseman et al., 1994; Zimmermann et al., 2004). BecausetPA binding to ANX II is known to regulate plasminogen activation(Hajjar and Menell, 1997; Sharma et al., 2010), we investigated in-teraction of tPA and ANX II and analyzed the effect of anti-ANXII treatment on plasminogen activation in the tumor tissues. Our immu-noprecipitaion followed by immunoblotting experiment strongly sug-gested that indeed tPA synthesized in the tumor tissue binds to ANX II(Fig. 4A). Next, we examined biochemical activation of plasminogen

to plasmin. Our results suggest that tumor plasminogen activationwas significantly blocked by anti-ANX II treatments (Figs. 4B lanes4–6 and C). In contrast, IgG treated control tumorswere able to activateplasminogen efficiently (Fig. 4B lanes 2 and 3). Since ANX II and plasminboth are known to induce neoangiogenesis, it is not unreasonable tospeculate that anti-ANX II blocked tPA/ANX II-mediated plasmin gener-ation thereby inhibited neoangiogenesis and consequently breasttumor growth.

Anti-ANX II antibody inhibits angiogenesis in vitro by inducing BAE cellapoptosis

It has been shown that angiostatin induces BAE cell apoptosis andinhibits angiogenesis (Claesson-Welsh et al., 1998). We previouslyshowed that the angiostatin binds to BAE cell surface ANX II. We

1 2 3 4 5 6

Plasminogen

Plasmin

IgG trea

ted

contro

l Anti-ANX

II

treate

d

Purifi

edpl

asm

inog

enB

C

1 2 3 4

HMW IgG

LMW IgG tPA

IP : ANX II IB: anti-tPA

A

16

14

12

10

8

6

4

2

0

Uni

ts/m

g of

pro

tein

Con

trol

(IgG

)

Ant

i-A

NX

II

Fig. 4. Anti-ANX II antibody inhibited tPA-mediated plasmin generation in the tumor microenvironment: tissues from tumor bearing mice were harvested and homogenized andanalyzed interaction of tPA with ANXII by Immunoprecipitaion. Tissue lysate was immunoprecipitated with anti-ANX II followed by immunoblotting with anti-tPA antibody. PanelA demonstrated interaction of tPA with ANX II in the tumor microenvironment. Lane 1 mouse IgG (negative control), lane 2 ANX II immunoprecipitated pellet, lane 3 supernatantafter immunoprecipitaion and lane 4 is purified tPA (positive control). Next we analyzed whether antibody treatments has any effect on plasmin generation in the tumor tissues.The tumors from the mice treated with anti-ANX II demonstrated inhibition of plasmin generation (Panel B, lanes 4–6). In contrast, IgG treated control tumors showed efficientactivation of plasminogen to plasmin (Panel B, lanes 2–3). (C) Quantitative analysis of plasmin generation. Experiments were repeated three times.

180 M. Sharma et al. / Experimental and Molecular Pathology 92 (2012) 175–184

speculated that angiostatin's anti-angiogenic activity may be due toits interaction with ANX II (Tuszynski et al., 2002). We tested anti-angiogenic effect of anti-ANX II antibody on three dimensional an-giogenesis assay on matrigel. Our data showed that antibody-mediated targeted blocking of ANX II disrupts three dimensionaltubular network-like structures on matrigel (Figs. 5, 2A) whereasisotype matched IgG treatment has no effect on tubular network-like growth pattern (Figs. 5, 1A). We examined the mechanism bywhich anti-ANX II inhibited in vitro angiogenesis. BAE cells wereincubated with anti-ANX II for 72 h along with positive and nega-tive controls as described above. Cells were double immunostainedwith FITC-annexin V, a marker for apoptosis, followed by propidiumiodide (PI), a marker for DNA. Anti-ANX II antibody stimulatedapoptotic signals causing BAE cell apoptosis (Figs. 5, 2C and 3C),whereas control cells treated with isotype matched IgG showed nosuch effect (Figs. 5, 1C). The apoptotic effect was dose-dependent(Fig. 5.1) suggesting that antibody-mediated targeting of ANX II in-deed inhibited angiogenesis at least in part by inducing BAE cellapoptosis.

Annexin II correlates with neoangiogenesis in human invasive breastcancer

We and others have reported that annexin II is an important reg-ulator of neoangiogenesis and may be a potential target to inhibitneoangiogenesis in cancer (Lima e Silva et al., 2010; Ling et al.,2004). Consistent with previous findings we now demonstrate signif-icant inhibition of neoangiogenic-activity in our xenograft model ofhuman breast cancer. We thus next sought to determine whetherANX II expression related to neoangiogenesis in patients with clinical-ly diagnosed invasive breast cancers. Immunohistochemical analysisof tissues from patients with invasive and metastatic breast cancerrevealed intense staining of CD31, a marker for neoangiogenic activity(Fig. 6D) in the tumor microenvironment concomitant with over-

expression of cell surface ANX II (Fig. 6B). In contrast, normal hyper-plasic breast tissue showed only large mature existing blood vesselswith no neoangiogenic activity (Fig. 6C). Interestingly normal ductalepithelial cells exhibited little or no ANX II expression (Fig. 6A).Consistent with our immunoflorescence results, ANX II expressionwas detected mainly on the surface of invading breast cancer cells(Fig. 6B inset). These results suggest that the increased fibrinolyticactivity in the tumor microenvironment may be a necessary event forneoangiogenesis.

Discussion

Cancer-related thrombosis represents a complex imbalance of co-agulation and fibrinolysis, the processes of formation and deactiva-tion of fibrin, respectively. During tumor-derived stromal injuryplasma enters the extra vascular space and initiates the clottingcascade to repair injury via induction of neoangiogenesis. It has there-fore been suggested that a tumor may appear to the host as a “woundthat does not heal” (Dvorak, 1986). Proteins of the coagulation andfibrinolytic pathways play a critical role in neoangiogenesis forboth wound repair and tumor progression (Carmeliet, 2001; Nashet al., 2001). The relationship between malignant disease and dys-regulation of blood coagulation was first discovered by Trousseau(Trousseau, 1856). Latter it has been confirmed that disseminatedintravascular coagulation (DIC) and thromboembolism are com-mon phenomena in cancer patients which has been suggested tobe a frequent cause of mortality (Donati, 1995; Sorensen et al.,1998). Activation of both fibrinolytic and coagulation pathwaysare hallmarks of neoangiogenesis in breast, colon, lung and pancre-atic cancers (Sorensen et al., 1998). Therefore proteins that regu-late these pathways are potential therapeutic targets (Huanget al., 1997). To date, the selective dysregulation of specific proteinsof these pathways remains largely unknown. Clinical evidence suggeststhat uPA, an enzyme that activates plasminogen, is strong predictors

1B

2B

3B

44

1

1

2

34

4

3

1

1C

2C

3C

IgG treated (–ve control)

AS treated (+ve control)

Anti-ANX II treated

1A

2A

3A

IgG treated (–ve control) IgG treated (–ve control) )

Anti-ANX II treated

AS treated (+ve control)

Anti-ANX II treated

AS treated (+ve control)

0.0

0.1

0.2

Via

ble

Cel

lsA

bsor

banc

e at

490

nm

5.1A 0.3 IgG control

Fig. 5. Anti-ANX II antibody induces endothelial cell apoptosis and inhibits in vitro angiogenesis: our in vivo data suggests that anti-ANX II treatments inhibited neoangiogenic ac-tivity. We sought to determine whether anti-ANX II has any effect on EC survival in vitro. Data show that anti-ANX II reduces EC viability in a dose-dependent manner (Fig. 5 panel2B and Fig. 5.1A). Reduction of cell viability is due to anti-ANX II induced apoptosis (Fig. 5 panel 2C). We further tested role of anti-ANX II on in vitro angiogenesis. Bovine aorticendothelial cells (BAE) plated in triplicates on matrigel and allowed to form three dimensional capillary networks. Cells were treated with anti-ANX II (Fig. 5 panel 2A), isotypematched IgG (negative control, panel 1A) and angiostatin (positive control, panel 3A). Cells treated with anti-ANX II exhibited disruption of three dimensional endothelial capillarynetworks (panel 2A). Panel C, BAE cells plated in 96-well plates and treated as indicated. After treatments cells were immunostained with annexin V (marker for apoptosis) and PI(marker for DNA). Cells were viewed under immunofluorescence microscope. Anti-ANX II treatment induced BAE cell membrane damage that allowed entry of PI and stained DNAin the cells. Annexin V staining demonstrates damaged cell membrane (green) (panel 2C, arrows) PI staining of DNA (red) (panel 2C). Isotype matched IgG treated cells exhibited PIstaining of dead cells, a normal phenomenon during cell culture (1C). The representative photomicrograph of three independent observations of triplicate wells.

181M. Sharma et al. / Experimental and Molecular Pathology 92 (2012) 175–184

of poor outcome in breast cancer patients (Janicke et al., 1989;Janicke et al., 2001; Li et al., 1998; Look et al., 2002). Consistentwith this, neoangiogenesis is also reported to be a significant inde-pendent prognostic indicator in patients with early stage breastcancer (Weidner et al., 1992), suggesting a common link betweenfibrinolysis and neoangiogenesis in breast cancer. Both uPA andtPA can convert inactive plasminogen to active enzyme plasmin,which degrades fibrin clots and known to initiate neoangiogenesis(Ling et al., 2004; Bajou et al., 2001).

ANX II is a calcium-dependent phospholipid binding protein andthe cell surface receptor for tPA on many cell types including breastcancer (Diaz et al., 2004; Sharma et al., 2010). The interaction of tPAwith ANX II regulates efficient plasmin generation and neoangio-genesis in vitro and in vivo (Diaz et al., 2004; Lima e Silva et al.,2010; Ling et al., 2004; Sharma et al., 2010). Considerable evidenceindicates that the ANX II and tPA contribute to neoangiogenesisand cancer progression including in breast cancer (Diaz et al.,2002; Lima e Silva et al., 2010; Naina et al., 2010; Sharma et al.,

Invasive breast cancer Normal breast with hyperplasia A

D

Invasive breast cancer Normal breast with hyperplasia

BA

C

Fig. 6. ANX II immunostaining correlated with neoangiogenic activity in human breast cancer patients: pre-diagnosed archival tissues from human breast cancer patients were im-munostained for ANX II (A and B) and EC marker CD31 (C and D). Normal hyperplasic breast tissues demonstrated normal ducts with little or no ANX II staining (A, arrow). Insetshowing magnified image of normal duct (A). Tissue sections from invasive/metastatic ductal carcinoma patient demonstrated intense cell surface staining of ANX II (brown stain-ing, arrows) of ductal epithelial cells. Every invading cell showed ANX II staining (B). Inset showing magnified image of cell surface staining of ANX II. (C and D) demonstratesneoangiogenic activity in normal and invasive breast carcinoma. CD31 staining of hyperplasic breast tissue demonstrated staining (brown, arrow) of large mature existing bloodvessels (C) whereas invasive breast carcinoma showed numerous small microvessels as an indication of neoangiogenesis (D, arrows). Representative photomicrographs arefrom twenty different patients and five different normal breast tissues (magnification 200×).

182 M. Sharma et al. / Experimental and Molecular Pathology 92 (2012) 175–184

2010). Previously we have shown that ANX II is selectivelyexpressed on the surface of invasive human breast cancer cells(Sharma et al., 2006a). We extended our observations and con-firmed that tPA binding to cell surface ANX II in MDA-MB231 cellsregulates plasmin generation (Sharma et al., 2010).

To determine if ANX II or tPA is specific to the neoangiogenesisfound in breast cancer patients, we identified ANX II concurrent withsecretion of tPA in the desmoplastic stroma of patients with advancedstage breast cancer. This ANX II/tPA expression coincided with neoan-giogenic activity in desmoplastic stroma (Sharma et al., 2010). It ispossible that the intense neoangiogenic activity is due to tPA/ANXII-mediated plasmin generation (Bajou et al., 2001; Chernicky et al.,2005; Chuthapisith et al., 2009; Diaz et al., 2002; Ling et al., 2004;Sharma et al., 2006a). Our data are consistent with recently pub-lished clinical study which identified tPA as a single major factorinvolved in breast cancer metastasis (Naina et al., 2010). Previousclinical as well as experimental studies also identified ANX II in aggres-sive cancers and linkedwith neoangiogenesis (Chuthapisith et al., 2009;Diaz et al., 2004; Sharma et al., 2006a, 2006b; Syed et al., 2007;Zimmermann et al., 2004). Immunohistochemical analysis of the breastcancer patients demonstrated intense staining of ANX II mainly onthe cell surface (Fig. 6B). In contrast, either little or no expressionwas observed in normal hyperplasic breast ductal epithelial cells(Fig. 6A). ANX II expression coincided with strong neoangiogenicactivity in specimens obtained patients with aggressive breast cancer(Fig. 6D) suggesting a link between ANX II and neoangiogenesis and in-vasive breast cancer. In line with our observations recent clinical study

also reported high levels of ANX II is a predictor of poor pathological re-sponse to neoadjuvant therapy in breast cancer (Chuthapisith et al.,2009). It remains to be seen whether ANX II has prognostic or diag-nostic significance in patients with invasive breast cancer. Taken to-gether, our data strongly suggest that ANX II/tPA interaction in thebreast tumormicroenvironment acts as an “angiogenic switch”, therebytransforming prevascular cancer to an aggressive, highly vascularizedmetastatic phenotype. Because tumors are known to activate localcoagulation, we reasoned that increased ANX II/tPA-dependent fibri-nolysis is possibly host response to neutralize tumor-induced coagu-lation and resulted in increased neoangiogenic activity. The synchronizedactivity of clotting and fibrinolysis is known to initiate neoangiogen-esis (Carmeliet, 2001; Nash et al., 2001). We explored our hypothe-sis by direct targeting of ANX II in a xenograft model. At first weinvestigated whether immuno-neutralization of ANX II inhibits breasttumor growth and neoangiogenesis and secondly whether tumor-synthesized tPA activates plasminogen by binding to cell surface ANXII. We found that anti-ANX II antibody treatment significantly inhib-ited breast tumor growth while concurrently decreasing neoangio-genesis (Jacovina et al., 2009; Lima e Silva et al., 2010; Ling et al.,2004). In contrast, tumors in the control groups were highly vascu-larized and mouse IgG exerted no apparent effect on tumor growth(Figs. 2A and B). Anti-ANX II treatment significantly inhibited visibletumor neoangiogenic activity (Fig. 2D) which was confirmed byimmunostaining and quantification of CD31 positive blood vessels(Figs. 3D and E). These data suggests that immuno-neutralizationof ANX II blocked localized plasmin generation and likely resulted

183M. Sharma et al. / Experimental and Molecular Pathology 92 (2012) 175–184

in inhibition of neoangiogenic activity. Our finding is consistentwith the proposed role of ANX II in stimulating vessel sproutingand neoangiogenesis (Diaz et al., 2004; Jacovina et al., 2009; Kesavanet al., 2009; Lima e Silva et al., 2010; Sharma et al., 2010b). To supportour in vivo data, we tested if anti-ANX II binding to EC cell surfaceANX II affects angiogenic capabilities. The data presented in Fig. 5provides evidence that, (Lima e Silva et al., 2010; Tuszynski et al.,2002), anti-ANX II binding to EC cell surface ANX II induces apo-ptosis and inhibits neoangiogenesis in vitro suggesting that anti-ANXII-mediated inhibition of neoangiogenesis was due to EC apoptosis. Pre-viously, using in vitromodels it has been shown that tPA binding to ANXII regulates neoangiogenesis via localized plasmin generation (Diazet al., 2002, 2004; Ling et al., 2004; Sharma et al., 2006b, 2010). In thisstudy, we demonstrated biochemical interaction between tPA andANX II in the breast tumor microenvironment (Fig. 4A). Moreover,anti-ANX II treatment inhibited plasmin generation in breast tumortissues (Figs. 4B and C). Our data, together with findings from others,suggest that tPA interaction on cell surface ANX II regulates plasmingeneration in breast tumor, which in turn facilitates neoangiogenesisand tumor growth. Because ANX II provides for the assembly of tPAand plasminogen for plasmin generation, the binding of antibody toANX II on EC/MDA-MB231 cells could initiate a negative feed-backloop that blocks ANX II/tPA/plasminogen assembly thereby inhibit-ing plasmin production in the tumor microenvironment. Our pro-posed model is in agreement with recent in vitro and clinical studiessuggesting role for ANX II and tPA in neoangiogenesis and breast cancermetastasis (Chernicky et al., 2005; Naina et al., 2010; Sharma et al.,2010). Consistent with our studies, 123I-anti-ANX II has been shown toaccumulate in neovessels (Jain, 2010). These data suggest both spec-ificity for ANX II in neoangiogenesis and that selective delivery ofanti-ANX II in the tumor vasculature disrupts neoangiogenic activ-ity, thereby resulted in inhibition of breast tumor growth in ourexperimental model.

In conclusion, we now report that the interaction between tPA/ANX II regulates plasmin generation and neoangiogenesis in humanbreast tumors. To our knowledge, this is the first report demon-strating that targeted disruption of ANX II inhibits neoangiogenesisand human breast tumor growth in a xenograft model. Thus ANX IIrepresents a potential candidate target for clinical, diagnostic and ther-apeutic applications in breast cancer andwarrants further investigation.

Conflict of interest

The authors declare that there are no conflicts of interest.

Acknowledgments

This study was supported in part by grants from Department ofDefense idea award W81XWH-07-1-0424 and Concept award #DAMD1703-1-0761 (MCS) and in part by the Research Services of theWashington DC VA Medical Center. We thank Dr. George Tuszynskifor critical reading of this manuscript.

References

Aguilar, S., et al., 2004. Tissue plasminogen activator in murine exocrine pancreas cancer:selective expression in ductal tumors and contribution to cancer progression.American Journal of Pathology 165, 1129–1139.

Aitkenhead, M., et al., 2002. Identification of endothelial cell genes expressed in an invitro model of angiogenesis: induction of ESM-1, (beta)ig-h3, and NrCAM. Micro-vascular Research 63, 159–171.

Bajou, K., et al., 2001. The plasminogen activator inhibitor PAI-1 controls in vivo tumorvascularization by interaction with proteases, not vitronectin. Implications forantiangiogenic strategies. The Journal of Cell Biology 152, 777–784.

Braden, A.R., et al., 2009. Polymeric nanoparticles for sustained down-regulation ofannexin A2 inhibit prostate tumor growth. Journal of Nanoscience and Nanotech-nology 9, 2856–2865.

Brichory, F.M., et al., 2001. An immune response manifested by the common occur-rence of annexins I and II autoantibodies and high circulating levels of IL-6 in

lung cancer. Proceedings of the National Academy of Sciences of the United Statesof America 98, 9824–9829.

Brownstein, C., et al., 2004. Annexin II mediates plasminogen-dependent matrix inva-sion by human monocytes: enhanced expression by macrophages. Blood 103,317–324.

Carmeliet, P., 2001. Biomedicine. Clotting factors build blood vessels. Science 293,1602–1604.

Cesarman, G.M., et al., 1994. An endothelial cell receptor for plasminogen/tissue plasmin-ogen activator (t-PA). II. Annexin II-mediated enhancement of t-PA-dependentplasminogen activation. Journal of Biological Chemistry 269, 21198–21203.

Chernicky, C.L., et al., 2005. Tissue-type plasminogen activator is upregulated in meta-static breast cancer cells exposed to insulin-like growth factor-I. Clinical BreastCancer 6, 340–348.

Chuthapisith, S., et al., 2009. Annexins in human breast cancer: possible predictors ofpathological response to neoadjuvant chemotherapy. European Journal of Cancer45, 1274–1281.

Claesson-Welsh, L., et al., 1998. Angiostatin induces endothelial cell apoptosis and activa-tion of focal adhesion kinase independently of the integrin-binding motif RGD. Pro-ceedings of the National Academy of Sciences of the United States of America 95,5579–5583.

Diaz, V.M., et al., 2002. Tissue plasminogen activator is required for the growth, invasion,and angiogenesis of pancreatic tumor cells. Gastroenterology 122, 806–819.

Diaz, V.M., et al., 2004. Specific interaction of tissue-type plasminogen activator (t-PA)with annexin II on the membrane of pancreatic cancer cells activates plasminogenand promotes invasion in vitro. Gut 53, 993–1000.

Donati, M.B., 1995. Cancer and thrombosis: from Phlegmasia alba dolens to transgenicmice. Thrombosis and Haemostasis 74, 278–281.

Dvorak, H.F., 1986. Tumors: wounds that do not heal. Similarities between tumorstroma generation and wound healing. The New England Journal of Medicine 315,1650–1659.

Goh, K.Y., et al., 2005. Tissue plasminogen activator expression in meningiomas andglioblastomas. Clinical Neurology and Neurosurgery 107, 296–300.

Grondahl-Hansen, J., et al., 1990. Tissue-type plasminogen activator in plasma frombreast cancer patients determined by enzyme-linked immunosorbent assay.British Journal of Cancer 61, 412–414.

Hajjar, K.A., Menell, J.S., 1997. Annexin II: a novel mediator of cell surface plasmin gen-eration. Annals of the New York Academy of Sciences 811, 337–349.

Huang, X., et al., 1997. Tumor infarction in mice by antibody-directed targeting of tissuefactor to tumor vasculature. Science 275, 547–550.

Jacovina, A.T., et al., 2009. Homocysteine inhibits neoangiogenesis in mice throughblockade of annexin A2-dependent fibrinolysis. The Journal of Clinical Investiga-tion 119, 3384–3394.

Jain, D., 2010. Angiogenesis imaging with radiolabelled annexin II antibody. Journal ofNuclear Cardiology A 14.45.

Janicke, F., et al., 1989. Urokinase-type plasminogen activator antigen and early relapsein breast cancer. Lancet 2, 1049.

Janicke, F., et al., 2001. Randomized adjuvant chemotherapy trial in high-risk, lymphnode-negative breast cancer patients identified by urokinase-type plasminogenactivator and plasminogen activator inhibitor type 1. Journal of the National CancerInstitute 93, 913–920.

Ji, N.Y., et al., 2009. Evaluation of annexin II as a potential serum marker for hepatocel-lular carcinoma using a developed sandwich ELISA method. International Journalof Molecular Medicine 24, 765–771.

Kesavan, K., et al., 2009. Annexin A2 is a molecular target for TM601, a peptide withtumor-targeting and anti-angiogenic effects. Journal of Biological Chemistry 285,4366–4374.

Kim, D.M., et al., 2009. Immunosensors for detection of Annexin II and MUC5AC forearly diagnosis of lung cancer. Biosensors and Bioelectronics 25, 456–462.

Koopman, G., et al., 1994. Annexin V for flow cytometric detection of phosphatidylserineexpression on B cells undergoing apoptosis. Blood 84, 1415–1420.

Li, H., et al., 1998. Adenovirus-mediated delivery of a uPA/uPAR antagonist suppressesangiogenesis-dependent tumor growth and dissemination in mice. Gene Therapy5, 1105–1113.

Lima e Silva, R., et al., 2010. Agents that bind annexin A2 suppress ocular neovascular-ization. Journal of Cellular Physiology 225, 855–864.

Ling, Q., et al., 2004. Annexin II regulates fibrin homeostasis and neoangiogenesis invivo. The Journal of Clinical Investigation 113, 38–48.

Look, M.P., et al., 2002. Pooled analysis of prognostic impact of urokinase-type plasmin-ogen activator and its inhibitor PAI-1 in 8377 breast cancer patients. Journal of theNational Cancer Institute 94, 116–128.

Mulligan-Kehoe, M.J., et al., 2001. A truncated plasminogen activator inhibitor-1 proteininduces and inhibits angiostatin (kringles 1–3), a plasminogen cleavage product.Journal of Biological Chemistry 276, 8588–8596.

Naina, H.V., et al., 2010. Systemic fibrinolysis caused by tissue plasminogen activator-producing metastatic breast cancer. Journal of Clinical Oncology 28, e167–e168.

Nash, G.F., et al., 2001. The role of the coagulation system in tumour angiogenesis. TheLancet Oncology 2, 608–613.

Ohno, Y., et al., 2009. Annexin II represents metastatic potential in clear-cell renal cellcarcinoma. British Journal of Cancer 101, 287–294.

O'Reilly, M.S., et al., 1996. Angiostatin induces and sustains dormancy of human primarytumors in mice. Nature Medicine 2, 689–692.

Pei, H., et al., 2007. Proteome analysis and tissue microarray for profiling protein markersassociated with lymph node metastasis in colorectal cancer. Journal of ProteomeResearch 6, 2495–2501.

Price, J.E., et al., 1990. Tumorigenicity and metastasis of human breast carcinoma celllines in nude mice. Cancer Research 50, 717–721.

184 M. Sharma et al. / Experimental and Molecular Pathology 92 (2012) 175–184

Qi, Y.J., et al., 2007. Dysregulation of Annexin II expression in esophageal squamous cellcancer and adjacent tissues from a high-incidence area for esophageal cancer inHenan province. Ai Zheng 26, 730–736.

Rella, C., et al., 1993. Tissue-type plasminogen activator as marker of functional steroidreceptors in human breast cancer. Thrombosis Research 69, 209–220.

Roseman, B.J., et al., 1994. Annexin II marks astrocytic brain tumors of high histologicgrade. Oncology Research 6, 561–567.

Semov, A., et al., 2005. Metastasis-associated protein S100A4 induces angiogenesisthrough interaction with Annexin II and accelerated plasmin formation. Journalof Biological Chemistry 280, 20833–20841.

Sharma, M.R., et al., 2006a. Angiogenesis-associated protein annexin II in breast cancer:selective expression in invasive breast cancer and contribution to tumor invasionand progression. Experimental and Molecular Pathology 81, 146–156.

Sharma, M.R., et al., 2006b. Antibody-directed targeting of angiostatin's receptorannexin II inhibits Lewis Lung Carcinoma tumor growth via blocking of plasmino-gen activation: possible biochemical mechanism of angiostatin's action. Experi-mental and Molecular Pathology 81, 136–145.

Sharma, M., et al., 2010. Breast cancer cell surface annexin II induces cell migration andneoangiogenesis via tPA dependent plasmin generation. Experimental and Molec-ular Pathology 88, 278–286.

Singh, P., 2007. Role of Annexin-II in GI cancers: interaction with gastrins/progastrins.Cancer Letters 252, 19–35.

Sorensen, H.T., et al., 1998. The risk of a diagnosis of cancer after primary deep venousthrombosis or pulmonary embolism. The New England Journal of Medicine 338,1169–1173.

Stack, M.S., et al., 1999. Angiostatin inhibits endothelial and melanoma cellular invasion byblockingmatrix-enhanced plasminogen activation. The Biochemical Journal 340, 77–84.

Syed, S.P., et al., 2007. Angiostatin receptor annexin II in vascular tumors includingangiosarcoma. Human Pathology 38, 508–513.

Tarui, T., et al., 2002. Plasmin-induced migration of endothelial cells. A potential targetfor the anti-angiogenic action of angiostatin. Journal of Biological Chemistry 277,33564–33570.

Trousseau, A., 1856. Phlegmasia alba dolens. Clinique Medicale de IHotel-Dieu de Paris,3, pp. 654–712.

Tuszynski, G.P., et al., 2002. Angiostatin binds to tyrosine kinase substrate annexin IIthrough the lysine-binding domain in endothelial cells. Microvascular Research64, 448–462.

Weidner, N., et al., 1992. Tumor angiogenesis: a new significant and independent prog-nostic indicator in early-stage breast carcinoma. Journal of the National CancerInstitute 84, 1875–1887.

Wu, W., et al., 2002. Identification and validation of metastasis-associated proteins inhead and neck cancer cell lines by two-dimensional electrophoresis and massspectrometry. Clinical & Experimental Metastasis 19, 319–326.

Yu, G.R., et al., 2007. Identification of molecular markers for the oncogenic differentiationof hepatocellular carcinoma. Experimental & Molecular Medicine 39, 641–652.

Zhang, F., et al., 2009. Anxa2 plays a critical role in enhanced invasiveness of themultidrugresistant human breast cancer cells. Journal of Proteome Research 8, 5041–5047.

Zimmermann, U., et al., 2004. Expression of annexin II in conventional renal cell car-cinoma is correlated with Fuhrman grade and clinical outcome. Virchows Archiv445, 368–374.