Inverse Expression of S100A4 and E-Cadherin Is Associated ... · Inverse Expression of S100A4 and E-Cadherin Is Associated with Metastatic Potential in Gastric Cancer Yutaka Yonemura,1

Inverse Expression of S100A4 and E-Cadherin Is Associated withMetastatic Potential in Gastric Cancer

Yutaka Yonemura,1 Yoshio Endou,Keiichi Kimura, Sachio Fushida, Etsurou Bandou,Keizou Taniguchi, Kazuo Kinoshita,Itasu Ninomiya, Kazuo Sugiyama,Claus W. Heizmann, Beat W. Schafer, andTakuma SasakiSecond Department of Surgery, School of Medicine [Y. Y., S. F.,E. B., K. T., Ka. K., I. N.], and Department of ExperimentalTherapeutics, Cancer Research Institute [Y. E., Ke. K., T. S.],Kanazawa University, Kanazawa 920, Japan; Virology Division,National Cancer Center Research Institute, Tokyo 104, Japan [K. S.];and Department of Pediatrics, Division of Clinical Chemistry,University of Zurich, Zurich, Switzerland [C. W. H., B. W. S.]

ABSTRACTS100A4 is known to be involved in cancer cell motility

by virtue of its ability to activate nonmuscle myosin. E-cadherin has an important role in the homophilic cell-celladhesion and is called an invasion suppressor gene. In thecurrent study, we investigate the histological type and met-astatic potential of gastric cancer from the aspect of theinterrelationship of E-cadherin and S100A4 expression.

Expression of E-cadherin and S100A4 in gastric cancercell lines, primary gastric cancers, and their normal coun-terparts were analyzed by reverse transcription-PCR, West-ern blot, and immunohistochemical methods.

S100A4 protein and E-cadherin were expressed in five ofeight gastric cancer cell lines, and inverse expression of the twoproteins are found in four cell lines. In the clinical specimens,E-cadherin mRNA expression in differentiated adenocarcino-mas (88%, 14 of 16) was significantly more frequent than thatin poorly differentiated adenocarcinomas (50%, 22 of 44;P 50.015). Western blot analysis demonstrates that S100A4 pro-tein expression in poorly differentiated adenocarcinomas was1.6-fold higher than in well differentiated adenocarcinoma.Immunohistochemically, S100A4 expression was detected in 51(55%) of 92 primary gastric cancers. Reduced expression ofE-cadherin in primary tumors was found in 66 (72%) of 92tumors. S100A4 expression in the poorly differentiated adeno-carcinomas had a strong relation to positive lymph node in-volvement or peritoneal dissemination. Reduced E-cadherin

expression showed a strong relationship with positive serosalinvolvement and infiltrating type. Tumors classified as a groupwith reduced E-cadherin and high expression of S100A4 revealpositive peritoneal dissemination, serosal involvement, and in-filtrating type in the growth pattern. Furthermore, these tu-mors showed a strong correlation with the poorly differenti-ated adenocarcinoma. In contrast, tumors with preservedE-cadherin and low expression of S100A4 have a close relationto the well differentiated adenocarcinoma and a favorableprognosis. By the Cox proportional hazard model, S100A4 andE-cadherin tissue status was judged as an independent prog-nostic factor. S100A4 and E-cadherin tissue status may be apowerful aid in evaluating metastatic potential or the prognosisof patients with gastric cancer.

INTRODUCTIONThe S100 protein family was first isolated from bovine

brain by Moore (1). Subsequent studies identified 16 membersof this family, based on amino acid sequence homology andsimilar structural properties (2). Each member of the family hasa feature of calcium-binding properties and exhibits a uniquepattern of expression,2 with some cells expressing multiplemembers of the family. The calcium-binding capacity is thoughtto be mediated through a highly conserved calcium-binding loopconsisted of 12 amino acids, which is flanked by twoa helices.This helix-loop-helix motif is called as an EF-hand, and calciumion binding via EF-hands mediates a conformation change inthese proteins, resulting in an alteration of target protein activityand production of intracellular response (3).

Among the 16 S100 proteins, S100A4 has functions in cellcycle progression (4) and cell motility (5). The mechanisms bywhich S100A4 might be exerting control over cell cycle progres-sion are yet unclear. In contrast, it is well known that S100A4protein alters cell motility by virtue of its ability to alter cytoskeletaldynamics (6). The ability of S100A4 to interact with nonmusclemyosin supports the view that it takes part in cellular motility (7).Filamentous actin, nonmuscle myosin, and nonmuscle tropomyosinhave been suggested as target molecules for S100A4 (7–9). Fur-thermore, several investigators reported that S100A4 expression isintimately related to tumor progression and metastasis (4, 10).

E-cadherin is the strongest molecule for homophilic adhesionof epithelial cells and has an important role in the formation ofepithelial architecture. The adhesive function of E-cadherin is de-pendent on the intracellular molecules, like catenin and actin. In theprogression of carcinogenesis, irreversible inactivation of E-cad-herin at the genomic level gene or through methylation of itspromoter is frequently found (11, 12). As a result, the mutualadhesive ability of cancer cells is weakened and cell dispersion

Received 4/20/00; revised 8/7/00; accepted 8/16/00.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisementin accordance with 18 U.S.C. Section 1734 solely toindicate this fact.1 To whom requests for reprints should be addressed, at Second Depart-ment of Surgery, School of Medicine, Kanazawa University, Takara-Machi 13-1, Kanazawa 920, Japan. Phone: 81-762-65-2000; Fax: 81-762-34-4260; E-mail: [email protected].

2 The new nomenclature of S100 protein was adopted from Schaferetal. (Genomics,25: 638–643, 1995).

4234Vol. 6, 4234–4242, November 2000 Clinical Cancer Research

Research. on May 18, 2021. © 2000 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

occurs. Accordingly, E-cadherin is called a metastasis-suppressorgene.

Especially in gastric cancer, E-cadherin is known to be oneof the causative genes because germ-line mutations of E-cadherin gene cause familial gastric cancer (13). We alreadyreported that reduced expression of E-cadherin in gastric canceris closely associated with invasive phenotype, poorly differen-tiated tumor, and poor prognosis (14).

Here, we demonstrate the intimate relationship betweenhighly invasive potential of gastric cancer and the phenotypeshowing simultaneous overexpression of S100A4 and reducedexpression of E-cadherin.

MATERIALS AND METHODSCell Lines. Eight gastric cancer cell lines (KATO-III,

TMK-1, NKPS, KKLS, NUGC-3, MKN-28, MKN-45, and AZ521) were cultured in RPMI 1640 with 10% FCS (Life Tech-nologies, Inc., Gaithersburg, MD).

Patients and Tumor Samples. Ninety-two patients withprimary gastric cancer diagnosed and treated in the Departmentof Surgery II, Kanazawa University Hospital, between 1990 and1997, were entered into the study. All patients underwent gas-trectomy with lymph node dissection.

Immediately after resection of the primary tumor, samplesof about 5 mm in diameter were taken from 60 primary tumorsand normal mucosa, and these samples were stored at280°C forthe RT-PCR3 and Western blot analyses. In addition, smallpieces of tissue about 5–8 mm in diameter were sampled from92 primary tumors, fixed in acetone at220°C overnight, andprepared to make paraffin-embedded blocks by the AMeXmethod (15). All of the resected primary tumors and regionallymph nodes were histologically examined by H&E stainingaccording to the general rules of the Japanese Classification ofGastric Carcinoma (16).

Antibody and Immunohistochemistry. Immunohisto-chemistry was done as described previously (14). Two depar-affinized sections were incubated overnight at 4°C with anti-E-cadherin monoclonal antibody (Takara Biochemicals, Tokyo,Japan) and S100A4 polyclonal antibody (17), diluted 1:100 inPBS. Next, the slides were incubated with biotinylated goatantimouse IgG for 20 min (LSAB kit; DAKO, Copenhagen,Denmark) and biotinylated antigoat rabbit IgG. Immunostainingwas done with diaminobenzidine (DAKO) solution with hydro-gen peroxide for 1 min. Only cases in which at least 20% of thetumor cells were immunoreactive were scored as positive forS100A4 (Fig. 1A). When.60% of all cancer cell were stainedfor E-cadherin on the cell membrane, the tumors were evaluatedas preserved E-cadherin expression (Fig. 2B). In contrast, whenthe E-cadherin immunoreactivity on the cell membrane wasfound in ,60% of all cancer cells or was expressed in thecytoplasm and not on the membrane, they were classified asreduced E-cadherin expression (Fig. 3).

RT-PCR. RT-PCR analysis was done with the modifi-cations of Conboyet al.(18). Briefly, total RNAs were extracted

from 60 primary gastric cancers, normal gastric mucosa, andeight gastric cancer cell lines, using ISOGEN (Nippon Gene,Tokyo, Japan). The prepared RNA was mixed with oligo-dTprimer and was reverse-transcribed with AMV reverse tran-scriptase (Life Sciences, St. Petersburg, FL), followed by PCRamplification (Perkin-Elmer Corp., Norwalk, CT) with specificprimers. PCR amplification was done for 1.5 min at 94°C, 2 minat 48°C, and 2 min at 72°C for 3 cycles, then by 25 cycles of40 s at 94°C, 1.5 min at 48°C, and 1.3 min at 72°C. The PCRproducts were electrophoresed on 2% agarose gels and trans-ferred to a nylon membrane filter. The transferred products werehybridized overnight to a32P-end-labeled probe specific for thetarget cDNA fragment. The autoradiogram was exposed for 4–5h with two intensifying screens at280°C.

Specific primers for theS100A4gene were S100A4-sense (59-GAGGGTGACAAATTCAAGCTCAACAAG-39)and S100A4-antisense (59-GGT CTCTTGGATAAGGA-AGTCT-39 and the target fragment of which is 644 bp), andthe probe oligonucleotide was 59-TGGCAGTGTCTGAC-CCATTCAT-39. Primer sequences for E-cadherin wereECAD-7 (sense, 59-ACCTCTGTGATGGAGGTC-39) andECAD-10 (antisense, 59-CCACATTCGTCACTGCTACG-39and the target fragment of which is 554 bp), and the probeoligonucleotide was 59-AACGTCGTAATCACCACACT-39.

3 The abbreviations used are: RT-PCR, reverse transcription-PCR;MMP, matrix metalloproteinase.

Fig. 1 A, immunohistochemical finding of a poorly differentiated ad-enocarcinoma stained with S100A4 polyclonal antibody. S100A4 isexpressed in the cytoplasm of cancer cells, lymphocytes, and smoothmuscle of blood vessel.Bars, 75 mm.B, E-cadherin expression iscompletely reduced in the same tumor inA (immunohistological findingusing E-cadherin monoclonal antibody).

4235Clinical Cancer Research



As internal standard, primer pairs specific to theb-actingene (b-act-sense, 59-TTGAAGGTAGTTTCGTGGAT-39; b-act-antisense, 59-GAAAATCTGGCACCACACCTT-39; PCR prod-ucts, 591 bp) were used.b-act-probe 59-ACTGACTACCTCAT-GAAGAT-39 was used as the probe.

Western Blot Analysis. Twelve microliters of proteinsample (total protein, 12mg) were mixed with a 6-ml sample buffer[50 mM Tris-HCl (pH 6.5), 10% glycerol, 2% SDS, and 0.1%bromphenol blue] and 0.6ml of DTT before separation by aNuPAGE Electrophoresis System (NOVEX, San Diego, CA). Af-ter completion of electrophoresis, samples were transferred to poly-vinylidene difluoride membrane filters (Immobilon; Millipore,Bedford, MA). The transferred samples were incubated with anti-S100A4 antibody (17) and anti-E-cadherin antibody (1:2000;Takara Biochemicals, Tokyo, Japan) for 2 h. The second antibodies[peroxidase-conjugated affipure F(ab9)2 F(ab9)2 fragment (1:5000,Santa Cruz Biotechnology, Santa Cruz, CA) for S100A4 and an-timouse IgG and horseradish peroxidase-linked F(ab9)2 fragment(1:5000; Amersham Life Science Co., Ltd.) for E-cadherin andb-actin ] were incubated for 40 min, rinsed, and then incubatedwith enhanced chemiluminescence Western blotting detection re-agent (Amersham Life Science Co., Ltd.) for 10 min. Then, themembrane was exposed to X-ray film for 1–10 min.

Data Presentation and Statistical Analysis. All statis-tical calculations were done with SPSS statistical software. Thex2 test and Student’st test for unpaired samples were used to

analyze data. The outcomes of the different groups of patientswere compared by the generalized Wilcoxon test. To clarify theE-cadherin and S100A4 tissue status as the independent prog-nostic factor, we did Cox stepwise regression analysis.

RESULTSS100A4 and E-cadherin mRNA and Protein Expression

in Gastric Cancer Cell Lines. S100A4 mRNA was expressedin KATOIII, TMK-1, KKLS, MKN-28, and MKN-45; butNKPS, NUGC-3, and AZ-521 did not express S100A4 mRNA(Fig. 4). The expression pattern of S100A4 protein was similar

Fig. 2 A, immunohistochemical finding of a well differentiated adeno-carcinoma stained with S100A4 polyclonal antibody. S100A4 is notexpressed.B, E-cadherin expression in the same tumor inA. E-cadherinexpression is localized on the cell membrane.

Fig. 3 A, E-cadherin expression in a poorly differentiated adenocarci-noma. E-cadherin is found as coarse granules in the cytoplasm of cancercells. B, immunohistochemical finding of a tubular adenocarcinomastained with S100A4 polyclonal antibody. S100A4 is faintly expressed.C, reduced E-cadherin expression in the same tumor inB (immunohis-tological finding using E-cadherin monoclonal antibody).

4236S100A4 and E-Cadherin Expression in Gastric Cancer



to those of S100A4 mRNA, but KATOIII did not expressS100A4 protein. E-cadherin mRNA expression was detected inKATOIII, TMK-1, NKPS, NUGC-3, and MKN-45, but was notfound in KKLS, MKN-28 and AZ521. Western blot analysisshowed that E-cadherin was detected in TMK-1, NKPS,NUGC-3, MKN-28, and MKN-45.

Among these cell lines, inverse expression of S100A4protein and E-cadherin was found in four cell lines (NKPS,KKLS, NUGC-3, and MKN-45). MKN-28 expressed low levelsof S100A4 and E-cadherin. In contrast, both proteins were notexpressed in KATOIII, AZ521 and TMK-1.

S100A4 and E-cadherin mRNA and Protein Expressionin Primary Gastric Cancer. E-cadherin mRNA expressionwas detected in 22 (50%) of 44 poorly differentiated adenocarci-nomas (Fig. 5). In contrast, 14 (88%) of 16 differentiated adeno-carcinomas expressed E-cadherin mRNA. There was a statisticalsignificance in the incidence of E-cadherin mRNA expressionbetween differentiated (88%) and poorly differentiated adenocar-cinomas (50%;P 5 0.015). S100A4 mRNA expression was de-tected in differentiated adenocarcinomas (12 of 16, 75%) and

poorly differentiated adenocarcinomas (26 of 44, 59%; Fig. 5).Western blot analysis demonstrated that S100A4 protein was ex-pressed not only from normal gastric mucosa but also primarytumors. Poorly differentiated adenocarcinomas expressed higherS100A4 protein than did the normal gastric mucosa (Fig. 6). By thedensitomertic analysis, the relative mean S100A4 density againstthat of b-actin in poorly differentiated adenocarcinomas was 1.6-fold higher than that in well differentiated adenocarcinoma (Fig. 6).E-cadherin of 124 kDa was detected in all of the normal mucosaand differentiated adenocarcinomas. In contrast, three of six poorlydifferentiated adenocarcinomas did not express the 124-kDa E-cadherin molecule.

Tissue Localization and Expression of S100A4 andE-Cadherin in Primary Gastric Cancers. S100A4 expres-sion was faint in the cytoplasm of the normal gastric mucosa,but strong in the cytoplasm of lymphocyte and smooth muscle.E-cadherin was exclusively expressed on the cell membrane ofnormal gastric mucosa, but not in the stromal cells or smoothmuscle. In gastric cancer, S100A4 protein was stained in thecytoplasm (Fig. 1A). In contrast, E-cadherin expression was

Fig. 4 E-cadherin/S100A4 protein and mRNAexpression from gastric cancer cell lines.

Fig. 5 E-cadherin/S100 mRNA expres-sion from six poorly differentiated adeno-carcinomas and six differentiated adeno-carcinomas.T, cancer.




mainly found on the cell membrane of cancer cells, and welldifferentiated adenocarcinoma tended to show the uniform ex-pression of E-cadherin on the cell membrane (Fig. 2B). Incontrast, the intensity of E-cadherin expression in the poorlydifferentiated tumors was heterogeneous or completely lost (Fig.1B). Some tumors show cytoplasmic staining of E-cadherinwithout stain on the cell membrane (Fig. 3), and this stainingpattern was exclusively found in the moderately differentiatedadenocarcinomas or poorly differentiated tumors.

S100A4 expression in primary gastric cancer was detectedin 51 (55%) of 92 primary gastric cancers. Reduced expressionof E-cadherin in the primary tumors was found in 66 (72%) of92 tumors.

Table 1 shows the correlation between immunohistochem-ical expression of S100A4 and clinicopathological parameters.There was a strong relationship between S100A4 expression andpoorly differentiated adenocarcinomas, and positive lymph nodeinvolvement or peritoneal dissemination.

Fig. 6 E-cadherin/S100 pro-tein expression from 12 pri-mary gastric cancers and theirnormal counterparts.N, normalgastric mucosa;T, cancer.

Table 1 Correlation between clinicopathological parameters and immunohistochemical expression of S100A4

Clinicopathological parameters No. of cases

S100A4 expressionStatistical

significanceNegative Positive

Liver metastasisNegative 84 40 (97%) 44 (86%) NSa

Positive 8 1 (3%) 7 (14%)Peritoneal dissemination

Negative 74 40 (97%) 34 (67%) P , 0.001Positive 18 1 (3%) 17 (33%)

Serosal invasionNegative 25 14 (34%) 11 (22%) NSPositive 67 27 (66%) 40 (78%)

Histological typeDifferentiated 43 28 (68%) 15 (29%) P , 0.001Poorly differentiated 49 13 (32%) 36 (71%)

Infiltrating patternsLocalized 6 4 (9%) 2 (4%) NSIntermediate 47 22 (54%) 25 (49%)Infiltrating 39 15 (37%) 24 (47%)

Lymphatic invasionNegative 25 15 (37%) 10 (20%) NSPositive 67 26 (63%) 41 (80%)

Venous invasionNegative 25 15 (37%) 10 (20%) NSPositive 67 26 (63%) 41 (80%)

Macroscopic typeEarly cancer 22 7 (17%) 15 (29%) NSBorrmann types 1 and 2 17 10 (24%) 7 (14%)Borrmann types 3 and 4 53 24 (59%) 29 (57%)

Lymph node metastasisNegative 32 19 (46%) 13 (25%) P , 0.001Positive 60 22 (54%) 38 (75%)

Total 92 41 51a NS, not significant.




The correlation of E-cadherin expression and clinicopath-ological parameters is shown in Table 2. Gastric cancers show-ing reduced E-cadherin expression had a strong relationship topositive serosal involvement and infiltrating type.

The primary gastric cancers were subdivided according tothe expression of S100A4 and E-cadherin. There were fourpatterns of S100A4 and E-cadherin: (a) E-cad1/S100A42,preserved E-cadherin and low expression and S100A4; (b)E-cad1/S100A41, preserved E-cadherin and high expressionof S100A4; (c) E-cad2/S100A42, reduced E-cadherin and lowexpression of S100A4; and (d) E-cad2/S100A41, reduced E-cadherin and high expression of S100A4. The number of tumorsbelonging to each group was 16 (17%), 10 (11%), 28 (30%), and38 (42%), respectively. Fig. 2 depicts a typical tumor ofE-cad1/S100A42, and Fig. 1 shows a tumor classified as agroup of E-cad2/S100A41, as listed in Table 3.

E-cad2/S100A41tumors tend to reveal positive perito-neal dissemination, serosal involvement, and infiltrating type inthe growth pattern.

The most interesting finding was the strong correlationbetween the histological type and the expression pattern ofS100A4 and E-cadherin. E-cad2/S100A41tumors showed astrong correlation with the poorly differentiated type. In con-trast, 12 (75%) of 16 E-cad1/S100A42tumors were classifiedas the well differentiated adenocarcinomas.

Survival of Patients in Terms of S100A4 and E-CadherinTissue Status after Surgery. The analysis of E-cadherin tissuestatus and survival is depicted in Fig. 7. Patients with tumors of

preserved E-cadherin expression had a significantly better prog-nosis than did those with tumors of reduced E-cadherin expres-sion. With regard to the expression of S100A4, patients withS100A4-positive tumors survived significantly poorer than didthose with S100A4-negative tumors (Fig. 8).

Fig. 9 shows the survival curves of four groups subdividedaccording to the expression of E-cadherin and S100A4. Patientswith E-cad1/S100A42tumors and with E-cad1/S100A41tumors had a significantly better prognosis than those withE-cad2/S100A42or E-cad2/S100A41tumors. Furthermore,patients with E-cad2/S100A41tumor had the poorest survivalthan did the other three groups.

By the Cox proportional hazard model, serosal invasionwas the strongest prognostic factor, followed by E-cadherintissue status, liver metastasis, and S100A4 and E-cadherin tissuestatus (Table 4). Patients with E-cad2/S100A41tumor had a5-fold higher relative risk for death than did those with E-cad1/S100A42tumor.

DISCUSSIONSeveral cancers, like breast cancer, colon cancer, and os-

teosarcoma, are known to produce S100A4, and the S100A4expression is now known to be involved in the malignantpotential of these tumors (4). To our knowledge, there was noreport about the correlation of S100A4 and gastric cancer. Ourpresent study clearly demonstrated that S100A4 tissue statusclosely relates to the lymph node metastasis, peritoneal dissem-

Table 2 Correlation between the immunohistological expression of E-cadherin and clinicopathological parameters


E-cadherin expressionStatistical

significanceReduced Preserved

Liver metastasisNegative 84 59 (89%) 25 (96%) NSa

Positive 8 7 (11%) 1 (4%)Peritoneal dissemination

Negative 74 49 (74%) 25 (96%) NSPositive 18 17 (26%) 1 (4%)

Serosal invasionNegative 25 14 (23%) 11 (42%) P 5 0.041Positive 67 52 (77%) 15 (58%)

Histological typeDifferentiated 43 28 (42%) 15 (58%) NSPoorly differentiated 49 38 (58%) 11 (42%) 22

Infiltrating patternsLocalized 6 4 (6%) 2 (8%) P 5 0.004Intermediate 47 27 (41%) 20 (77%)Infiltrating 39 35 (53%) 4 (15%)

Lymphatic invasion 25 (63%)Negative 25 14 (21%) 9 (65%) NSPositive 67 52 (79%) 17 (35%)

Venous invasion 30 (82%)Negative 25 16 (24%) 9 (35%) NSPositive 67 50 (76%) 17 (65%)

Macroscopic typeEarly cancer 22 14 (21%) 8 (31%) NSBorrmann types 1 and 2 17 11 (17%) 6 (23%)Borrmann types 3 and 4 53 41 (62%) 12 (46%)

Lymph node metastasisNegative 32 19 (29%) 12 (46%) NSPositive 60 47 (71%) 14 (54%)

Total 92 66 26a NS, not significant.




ination, and histological type. Albertazziet al.(19) reported thatin human breast cancerS100A4gene expression correlatesstrongly with the potential to metastasis to axillary lymph nodes.

In human smooth muscle cells, S100 A4 interacts with thesarcoplasmic reticulum and with actin stress fibers in a Ca21-dependent manner, resulting in the regulation of cell deformabilityand morphology (20). In cancer cells, S100A4 activates the non-muscle myosin that participates in cellular motility. This concept issupported by the fact that S100A4 is highly expressed duringembryonic development in the highly invasive mesenchymal ele-

ments (21). Similarly, Merzaket al. (22) reported that the expres-sion of S100A4 correlates with thein vitro invasive potential ofglioma cells. Additionally, the transfection of theS100A4gene intoa benign rat mammary epithelial cell line has been shown to lead tothe metastatic phenotype when the transfected cells are injectedinto the mammary fat pads of syngeneic rats (23, 24). Ambartsum-ianet al.(25) also reported that overexpression of themts1gene inthe mouse mammary carcinoma cells give rise to more aggressivetumors that are able to metastasize. More recently, Kriajevskaet al.(26) reported that the S100A4 protein modulates protein kinase C

Fig. 7 Survival curves of patients according to the E-cadherin expres-sion in primary tumors.

Fig. 8 Survival curves of patients according to the S100A4 expressionin primary tumors.

Table 3 Correlation between clinicopathological parameters and immunohistochemical status of S100A4/E-cadherin in primary tumors


E-cadherin and S100A4 expression

E-cad2/S100A41Statistical

significanceE-cad1/S100A42E-cad1/S100A41or

E-cad2/S100A41

Liver metastasisNegative 84 15 (94%) 35 (92%) 34 (89%) NSa

Positive 8 1 (6%) 3 (8%) 4 (11%)Peritoneal dissemination

Negative 74 15 (94%) 36 (95%) 23 (61%) P , 0.001Positive 18 1 (6%) 2 (5%) 15 (39%)

Serosal invasionNegative 25 8 (50%) 12 (32%) 5 (13%) P 5 0.015Positive 67 8 (50%) 26 (68%) 33 (87%)

Histological typeDifferentiated 43 12 (75%) 19 (50%) 12 (32%) P 5 0.016Poorly differentiated 49 4 (25%) 19 (50%) 26 (68%)

Infiltrating patternsLocalized 6 2 (12%) 3 (8%) 1 (3%) P 5 0.025Intermediate 47 10 (63%) 24 (63%) 12 (32%)Infiltrating 39 4 (25%) 11 (29%) 25 (65%)

Lymphatic invasionNegative 25 8 (50%) 11 (29%) 7 (18%) NSPositive 67 8 (50%) 27 (71%) 30 (82%)

Venous invasionNegative 25 7 (44%) 10 (26%) 7 (18%) NSPositive 67 9 (56%) 28 (74%) 31 (82%)

Macroscopic typeEarly cancer 22 6 (38%) 11 (29%) 5 (13%) NSBorrmann type 1 17 4 (24%) 8 (21%) 5 (13%)Borrmann type 3 53 6 (38%) 19 (50%) 28 (74%)

Lymph node metastasisNegative 32 7 (44%) 17 (45%) 8 (21%) NSPositive 60 9 (56%) 21 (55%) 30 (79%)

16 38 38a NS, not significant.




phosphorylation of the heavy chain of nonmuscle myosin in acalcium-dependent manner. These results strongly suggest S100A4overexpression may increase the motility of cancer cells by acti-vation of cytoskeletal nonmuscle myosin, resulting in conferringenhanced metastatic ability.

Metastasis is considered to be formed through multistepmechanisms and is not regarded merely by an increased motilityvia S100A4. Simultaneous accumulation of matrix-degrading en-zymes and reduced expression of metastasis-suppressor genes isconsidered to be involved in tumor invasion and metastasis. Fur-thermore, an intimate relationship of the expression betweenS100A4 and the matrix-degrading enzymes were reported. Bjørn-land et al. (27) reported that up-regulation of TIMP-2 and down-regulation of MT1-MMP and MMP-2 were introduced by thedown-regulation of S100A4 by an anti-S100A4 ribozyme. Accord-ingly, S100A4 may exert its effects on metastasis formation notonly by stimulating the motility of tumor cells but also by affectingtheir invasive properties through influencing the expression ofMMPs and their inhibitors. Like E-cadherin,nm23 is known as atumor suppressor gene, and reduced expression ofnm23 is associ-ated with lymph node metastasis in the breast cancer (28). Alber-tazzi et al. (19) also reported that in breast cancer simultaneous

overexpression of S100A4 and reduced expression ofnm23 isstrongly associated with a highly metastatic phenotype. The presentstudy clearly demonstrates that gastric cancers with high expressionof S100A4 and reduced expression of E-cadherin have a highlyinvasive and metastatic potential, as compared with the tumors withlow expression of S100A4 or preserved E-cadherin expression.Furthermore, these tumors are associated with a diffuse infiltratingtype, positive serosal invasion, and peritoneal dissemination. TheCox proportional hazard model demonstrated that gastric cancerswith high expression of S100A4 and reduced expression of E-cadherin have a 5-fold relative risk for death than those withpreserved E-cadherin and low expression of S100A4. E-cadherin isan important cell-cell adhesion molecules and controls cell polarityand tissue morphology (29). E-cadherin gene alterations and thesilence of E-cadherin expression by the methylation of the E-cadherin promoter are the frequent findings in gastric cancer (30–32). We already reported that the reduced expression of E-cadherinwas found in 66% of 125 primary gastric cancers (33). The loss orreduction of E-cadherin expression may induce the dissociation ofcells from primary tumors, due to loosened intercellular adhesion.When gastric cancer cells simultaneously obtain both thereduced E-cadherin expression and overexpression of S100A4,these cancer cells acquire high motility and invasive ability, result-ing in the diffuse infiltrating type and the formation of peritonealdissemination.

An interesting finding in the study is that simultaneousup-regulation of S100A4 and down-regulation of E-cadherinwas found more frequently in the poorly differentiated adeno-carcinomas than in the well differentiated ones. This finding wasconfirmed by the immunohistochemistry. The results from RT-PCR or Western blot analysis should be always judged byconsidering the intratumoral contamination of lymphocytes andvascular smooth muscle, which express considering amounts ofS100A4 mRNA and protein.

Gene abnormalities in the functioning domain of E- cadherinare more frequently found in poorly differentiated adenocarcino-mas than in the well differentiated ones (31, 34). Furthermore, ingastric cancer, the poorly differentiated adenocarcinomas areknown to have a higher invasive ability and poorer prognosis thanthe differentiated adenocarcinomas. Keirsebilcket al.(35) reportedthat E-cadherin and S100A4 expression levels are inversely regu-lated in tumor cell lines derived from the mouse mammary gland.Accordingly, in the poorly differentiated adenocarcinoma of stom-ach, down-regulation of E-cadherin may induce up-regulation of

Fig. 9 Survival of patients classified according tothe immunohistochemical tissue status of S100A4and E-cadherin.

Table 4 Results of multivariate analysis. Pathological parametersand E-cadherin/S100A4 tissue status as prognostic parameters in 92

gastric cancers

Clinicopathological parameters P Relative risk

Liver metastasisNegativevs.positive 0.014 3.99

Peritoneal disseminationNegativevs.positive 0.146 1.84

Serosal invasionNegativevs.positive 0.0004 12.7

Histological typeDifferentiatedvs.poorly

differentiated0.76 1.73

Infiltrating patternsLocalized/intermediate/infiltrating 0.386 0.533

Lymph node metastasisNegativevs.positive 0.111 2.4

S100A4 tissue statusNegativevs.positive 0.557 0.779

E-cadherinPreservedvs. reduced 0.008 0.294

S100A4/E-cadherinPreserved E-cadherin/S100A4(2)

vs. reduce E-cadherin/S100A4(1)0.022 5.032




S100A4, and the inverse expression of these two gene productsmay provide the biological features of its histological type.

In conclusion, the present study indicates that the combinedanalysis of E-cadherin and S100A4 may be a good prognosticindicator of patients with gastric cancer and that tumors withoverexpression of S100A4 and reduced E-cadherin can be clas-sified as highly malignant phenotype. Inverse expression ofS100A4 and E-cadherin seems to be associated with the diffusehistological type and with invasive ability of gastric cancer.

REFERENCES1. Moore, B. A. A soluble protein characteristic of the nervous system.Biochem. Biophys. Res. Commun.,19: 739–744, 1965.

2. Schafer, B. W., and Heizmann, C. W. The S100 family of EF-handcalcium-binding proteins: functions and pathology. Trends Biochem.Sci., 21: 134–140, 1996.3. Heizmann, C. W., and Cox, J. A. New perspectives on S100 proteins:a multi-functional Ca(21)-, Zn(21)- and Cu(21)-binding protein fam-ily. Biometals,11: 383–397, 1998.4. Parker, C., Whittaker, P. A., Usmani, B., Lakshmi, M. S., and Sherbet,G. V. Induction of 18A2/mts1 gene expression and its effects on metastasisand cell cycle control. DNA Cell Biol.,13: 1021–1028, 1994.5. Schelinfreund, R. H., Garger, S. W., Welsh, M. J., and van Eldik,L. J. Antisense inhibition of glial S100Bb production results in alterationin cell morphology, cytoskeletal organization and cell proliferation.J. Cell Biol.,111: 2021–2028, 1990.6. Gibbs, F. E. M., Wilkinson, M. C., Rudland, P. S., and Barraclough,R. Interaction of in vitro p9Ka, the rat S-100 related metastasis-inducingcalcium binding protein. J. Biol. Chem.,268: 18992–18999, 1994.7. Davies, B. R., Davies, M. P. A., Gibbs, F. E. M., Barraclough, R., andRudland, P. S. Induction of metastatic phenotype by transfection of amammary epithelial cell line with the gene for p9Ka, a rat calcium bindingprotein, but not with the oncogene EJ-ras-1. Oncogene,8: 999–1008, 1993.8. Ford, H. L., and Zain, S. B. Interaction of metastasis associated mts1protein with nonmuscle myosin. Oncogene,10: 1597–1605, 1995.9. Takenaga, K., Nakamura, Y., Sakiyama, S., Hasegawa, Y., Sato, K., andEndo, H. Binding of pEL98 protein, an S100-related calcium-bindingprotein to nonmuscle tropomyosin. J. Cell Biol.,124: 757–768, 1994.10. Weterman, M. A. J., Stoopen, G. M., Van Muijen, G. N. P.,Kuznicki, J., and Ruiter, D. J. Expression of calcyclin in human mela-noma cell lines correlates with metastatic behavior in nude mice. CancerRes.,52: 1291–1296, 1992.11. Perl, A. K., Wilgenbus, P., Dahl, U., Semb, H., and Christofori,G. A. A causal role for E-cadherin in the transition from adenoma tocarcinoma. Nature (Lond.),392: 190–193, 1998.12. Vleminckx, K., Vakaet, L., Mareel, M., Fiers, W., and Van Roy, F.Genetic manipulation of E-cadherin expression by epithelial tumor cellsreveals an invasion suppressor role. Cell,66: 107–119, 1991.13. Guilford, P., Hopkins, J., Harraway, J., McLeod, M., Mclod, N.,Harawira, P., Taite, H., Scoular, R., Miller, A., and Reeve, A. E.E-cadherin germline mutations in familial gastric cancer. Nature(Lond.), 392: 402–405, 1998.14. Yonemura, Y., Ninomiya, I., Kaji, M., Sugiyama, K., Fujimura, T.,Tsuchihara, K., Kawamura, T., Miyazaki, I., Endou, Y., Tanaka, M., andSasaki, T. Decreased E-cadherin expression correlates with poor survival inpatients with gastric cancer. Anal. Cell. Pathol.,8: 177–190, 1995.15. Sato, Y., Mukai, K., Watanabe, S., Goto, M., and Shimosato, Y.The AMeX method: a simplified technique of tissue processing andparaffin embedding with improved preservation of antigens for immu-nostaining. Am. J. Pathol.,125: 431–435, 1986.16. Japanese Research Society for Gastric Cancer. Japanese Classifica-tion of Gastric Cancer. Tokyo: Kanehara Shuppan Co., Ltd., 1993.17. Ilg, E. C., Schafer, B. W., and Heizmann, C. W. Expression patternof S100 calcium-binding proteins in human tumors. Int. J. Cancer,68:325–332, 1996.

18. Conboy, J. G., Chan, J., Mohandas, N., and Kan, Y. W. Multipleprotein 4.1 isoforms produced by alternative splicing in human erythroidcells. Proc. Natl. Acad. Sci. USA,85: 9062–9065, 1998.

19. Albertazzi, E., Cajone, F., Leone, B. E., Naguib, R. N., Lakshmi,M. S., and Sherbet, G. V. Expression of metastasis-associated genesh-mts1 (S100A4) and nm23 in carcinoma of breast is related to diseaseprogression. DNA Cell Biol.,17: 335–342, 1998.20. Mandinova, A., Atar, D., Schafer, B. W., Spiess, M., Aebi, U., andHeizmann, C. W. Distinct subcellular localization of calcium binding S100proteins in human smooth muscle cells and their relocation in response torises in intracellular calcium. J. Cell Sci.,111: 2043–2054, 1998.21. Kilngelhofer, J., Ambartsumian, N. S., and Lukanidin, E. M. Ex-pression of the metastasis associated mts1 gene during mouse develop-ment. Dev. Dyn.,210: 87–95, 1997.22. Merzak, A., Parker, C., Koochekpour, S., Sherbet, G. V., andPilkington, G. J. Over-expression of the 18A2/mts1 gene and down-regulation of the TIMP-2 gene in invasive human glioma cell linesinvitro. Neuropathol. Appl. Neurobiol.,20: 614–619, 1994.23. Davies, M. P. A., Rudland, P. S. R., Robertson, L., Parry, E. W.,Jolicoeur, P., and Barrclough, R. Expression of the calcium-bindingprotein S100A4 (p9Ka) in MMTV-neu transgenic mice induces metas-tasis of mammary tumours. Oncogene,13: 1631–1637, 1996.24. Davies, B. R., Davies, M. P., Gibbs, F. E., Barrclough, R., andRudland, P. S. Induction of the metastatic phenotype by transfection ofa benign rat mammary epithelial cell line with the gene for p9Ka, a ratcalcium-binding protein, but not with the oncogene EJ-ras-a. Oncogene.,8: 999–1008, 1993.25. Ambartsumian, N. S., Grigorian, M. S., Larsen, I. F., Karlstrøm, O.,Sidenius, M., Rygaard, J., Georgiev, G., and Lukanidin, E. Metastasis ofmammary carcinomas in GRS/A hybrid mice transgenic for the mts1gene. Oncogene,13: 1621–1630, 1996.26. Kriajevska, M., Tarabykina, S., Bronstein, I., Maitland, N.,Lomonosov, M., Hansen, K., Georgiev, G., and Lukanidin, E. Metas-tasis-associated Mts1 (S100A4) protein modulates protein kinase Cphosphorylation of the heavy chain of nonmuscle myosin. J. Biol.Chem.,273: 9852–9856, 1998.27. Bjørnland, K., Winberg, J. O., Odegaard, O. T., Hovig, E., Loen-nechen, T., Aasen, A. O., Fodstad, O., and Maelandsmo, G. M. S100A4involvement in metastasis: deregulation of matrix metalloproteinases andtissue inhibitors of matrix metalloproteinases in osteosarcoma cells trans-fected with an anti-S100A4 ribozyme. Cancer Res.,59: 4702–4708, 1999.28. Bevilacqua, G., Sobel, M. E., Liotta, L. A., and Steeg, P. S.Association of low nm23 RNA levels in human primary infiltratingductal breast carcinomas with lymph node involvement and other his-topathological indicators of high metastatic potential. Cancer Res.,49:5185–5190, 1989.29. Takeichi, M. Cadherin cell adhesion receptors as a morphogeneticregulator. Science (Washington DC),251: 1451–1455, 1991.30. Shiozaki, H., Oka, H., Inoue, M., Tamura, S., and Monden, M.E-cadherin mediated adhesion system in cancer cells. Cancer (Phila.),77: 1605–1613, 1996.31. Becker, K. F., Atkinson, M. J., Reich, U., Becker, I., Nekarda, H.,Siewert, J. R., and Hofler, H. E-cadherin gene mutations provide cluesto diffuse type gastric carcinomas. Cancer Res.,54: 3845–3852, 1994.32. Yoshiura, K., Kanai, Y., Ochiai, A., Shimoyama, Y., Sugimura, T.,and Hirohashi, S. Silencing of the E-cadherin invasion-suppressor geneby CpG methylation in human carcinomas. Proc. Natl. Acad. Sci. USA,92: 7416–7419, 1995.33. Yonemura, Y., Nojima, N., Kaji, M., Fujimura, T., Itoh, H.,Ninomiya, I., Miyazaki, I., Endo, Y., and Sasaki, T. E-cadherin andurokinase-type plasminogen activator tissue status in gastric carcinoma.Cancer (Phila.),76: 941–953, 1995.34. Berx, G., Becker, K. F., Hofler, H., and van Roy, F. Mutations ofthe human E-cadherin (CDH1) gene. Hum. Mutat.,12: 226–237, 1998.35. Keirsebilck, A., Bonne, S., Bruyneel, E., Vermassen, P., Lukanidin,E., Mareel, M., and van Roy, F. E-cadherin and metastasin (mts-1/S100A4) expression levels are inversely regulated in two tumor cellfamilies. Cancer Res.,58: 4587–3491, 1998.




2000;6:4234-4242. Clin Cancer Res Yutaka Yonemura, Yoshio Endou, Keiichi Kimura, et al. with Metastatic Potential in Gastric CancerInverse Expression of S100A4 and E-Cadherin Is Associated

Updated version

http://clincancerres.aacrjournals.org/content/6/11/4234

Access the most recent version of this article at:

Cited articles

http://clincancerres.aacrjournals.org/content/6/11/4234.full#ref-list-1

This article cites 32 articles, 11 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/6/11/4234.full#related-urls

This article has been cited by 18 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/6/11/4234To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/content/6/11/4234.full#ref-list-1

http://clincancerres.aacrjournals.org/content/6/11/4234.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

Inverse Expression of S100A4 and E-Cadherin Is Associated ... · Inverse Expression of S100A4 and E-Cadherin Is Associated with Metastatic Potential in Gastric Cancer Yutaka Yonemura,1