Lectin Histochemistry of Metastasizing and Non-metastasizing Breast and Colon Cancer Cells

Lectin Histochemistry of Metastasizing and Non-metastasizing Breast and Colon Cancer Cells BIRTHE SCHNEGELSBERG, UDO SCHUMACHER and URSULA VALENTINER Institute of Anatomy II - Experimental Morphology, University Hospital Hamburg-Eppendorf, D-20246 Hamburg, Germany Abstract. Background: Glycosylation of the tumour cell surface is of importance in metastasis formation as indicated by lectin-binding studies. In particular, binding of the lectin HPA is associated with metastasis formation, both in clinical studies and in xenograft models of breast and colon cancer. Here we examined if there is an association between the HPApositive glycotopes of metastasizing cancer cells and selectinbinding properties. Materials and Methods: Glycotope expression of human breast and colon cancer cells (MCF7, T47D, HBL100, HT29, SW480) grown in culture and xenografted into SCID mice were investigated by histochemical analysis. Results: HPA binding was observed in metastasizing breast and colon cancers and not in non-metastasizing ones. In colon cancer, E-selectin binding and expression of the selectin ligands CD15s and CA19-9 was higher in metastatic HT29 than in non-metastatic SW480 cells, especially when cells were grown in vitro. In breast cancer, E-selectin binding, CD15s and CA19-9 expression were independent of the metastatic potential. P-Selectin binding was slightly higher in metastasizing breast cancer cells (MCF7, T47D) than in nonmetastasizing HBL100 cells. Conclusion: Binding to E-selectin and expression of E-selectin ligands of colon cancer cells grown in vitro is associated with metastasis formation in a xenograft model. However, analysis of selectin ligands is of limited predictive value for the metastatic potential of breast cancer cells in our xenograft model. The formation of metastases is a life-threatening event in cancer progression. In order to understand this process and to predict and analyze metastasis formation, several metastasis Correspondence to: Dr. Ursula Valentiner, Institute for Anatomy II: Experimental Morphology, University Hospital Hamburg- Eppendorf, Martinistraße 52, D-20246 Hamburg, Germany. Tel: +49 407410 53587, Fax: +49 407410 55427, e-mail: valentin@uke.unihamburg.de Key Words: Metastasis, HPA, E-selectin, P-selectin, SCID mouse, lectin, breast carcinoma cells, colon adenocarcinoma. markers have been developed in clinical studies. One of these markers is the lectin HPA (Helix pomatia agglutinin), whose binding to primary tumour cells in tissue sections has been correlated with a poor prognosis in breast and colon cancer patients (1-4). HPA is a sugar-binding protein of nonimmunological origin, which is specific for N- acetylgalactosamine (GalNac) (4). In addition to clinical studies, HPA also identifies the metastatic phenotype of human colon and breast cancer cell lines when subcutaneously transplanted into severe combined immunodeficient (SCID) mice (1, 5). These experiments demonstrated that SCID mice are a well suited model to reflect the clinical situation and hence can be used to study the role of human cancer cell glycosylation in the metastatic cascade (6). Despite their clinical usefulness, the exact glycotope composition of HPA-binding glycoproteins of cancer cells has not been characterized in detail. Brooks et al. (7) described 13 HPA-positive glycoproteins from breast cancer primary tumours whose expression is quantitatively linked to the malignancy of the tumour, and which indicate that metastasizing cancer cells express a specific carbohydrate pattern on their cell surface in comparison to nonmetastasizing cells. Carbohydrate analysis of the HPAbinding glycoproteins revealed that it is most likely a sialylated tetrasaccharide which is responsible for HPA binding (8). Sialyl Lewis x (sle x ) is also a tetrasaccharide which is expressed on the cell surface of leucocytes mediating the attachment of leucocytes to inflamed endothelium via binding to E(ndothelial) and P(latelet) selectins. Sialyl Lewis x includes α-1,3-linked fucose residues and from nuclear magnetic resonance experiments it was concluded that sle x binds to the lectin Aleuria aurantia agglutinin (AAA), which is specific for fucose residues (9). Furthermore, binding of cancer cells to AAA was associated with hepatic metastasis in patients with colorectal cancer (10). It is attractive to hypothesize that the HPA- and AAA-binding carbohydrate residues are also recognized by the selectins. Thus, endothelial attachment and transmigration of metastatic tumour cells would appear to highjack the attachment and transmigration molecules normally used by leucocytes. 0250-7005/2011 $2.00+.40 1589

Table I. Characterization of different cancer cell lines. Cell line Cell line derivation Metastatic in SCID mice* Colon HT29 Adenocarcinoma Yes SW480 Grade 3-4 colon adenocarcinoma No Breast MCF7 Derived from malignant pleural effusion of an infiltrating ductal carcinoma Yes T47D Pleural effusion from breast cancer patient, oestrogen sensitive Yes HBL100 Breast milk cells No *Data taken from (1). Indeed, when E- and P-selectins were knocked out in SCID mice, the number of spontaneous lung metastases fell by 85% as compared to that in wild-type SCID mice (11). The aim of the present study was therefore to characterize the carbohydrate residues of metastasizing and nonmetastasizing cancer cells using lectins, selectin fusion proteins and monoclonal antibodies directed against carbohydrate residues of the sialylated Lewis family representing selectin-binding sites in xenograft models of human breast and colon cancer. Materials and Methods Cell culture. The human colon cancer cell lines HT29 and SW480 and the human breast cancer cell lines MCF7, T47D and HBL100 were used. For further information regarding the cell lines, see Table I. Cells were cultured at 37 C in humidified air containing 5% CO 2 in RPMI-1640 culture medium (Life Technologies, Karlsruhe, Germany). For staining, cells were harvested with a cell scraper (TPP, Trasadingen, Switzerland), centrifuged, washed several times in 0.1 M phosphate-buffered saline (PBS) and were fixed in 4% paraformaldeyde for 24 h at 4 C. After rinsing the cells in PBS, they were centrifuged for 5 min at 1500 rpm and resuspended in 2% agar (Merck, Darmstadt, Germany). The solidified agar blocks of the cell pellets were routinely processed in wax for histology. In addition, tumour cells of the breast and colon cancer cell lines were subcutaneously xenografted into pathogen free BALB/c severe combined immunodeficient (SCID) mice. Each cell line was injected into groups of 8 to 10 SCID mice. Breast cancer cells were injected in female mice and colon cancer cells were injected in female and male mice in equal parts. The mice were 8 to14 weeks old and weighed 20-25 g at the beginning of the experiments. They were housed in filter top cages and provided with sterile water and food ad libitum and treated according to institutional care protocols. The animal testing was approved by the local animal experiment approval committee (Behörde für Soziales, Familie, Gesundheit, Verbraucherschutz; Amt für Gesundheit und Verbraucherschutz, Hamburg, Germany). All manipulations were carried out aseptically inside a laminar flow hood. For injection, tumour cells were harvested by trypsination and viable cells (5 10 6 ) were suspended in 1 ml cell culture medium. An aliquot of 200 μl of this suspension was injected subcutaneously between the scapulae of each SCID mouse. The mice were sacrificed when the primary tumour had reached maximal growth (up to 20% of the body weight of the animal at the beginning of the experiment) or started to ulcerate. Primary tumours were removed, weighed and fixed in 10% PFA in PBS. Histochemistry. Binding of the lectins HPA, AAA and E- and P-selectin, as well as expression of the selectin ligands sle x (CD15s) and sialyl Lewis a (sle a, CA19-9), was analyzed in paraffin-embedded cells from cell culture and paraffin-embedded primary tumours. Paraffin sections (5 μm) were deparaffinized and pretreated as required. For staining with the lectins HPA (Sigma Aldrich, Munich, Germany) and AAA (Vector Laboratories, Burlingame, CA, USA), sections were trypsinized (0.1%; Biochrom KG, Berlin, Germany) for 10 min at 37 C, washed with lectin buffer (0.05 M TRIS-buffered saline (TBS), ph 7.6, with 1 mm CaCl 2 and 1 mm MgCl 2 ) and incubated for 1 h at room temperature (RT) with biotinylated HPA and AAA (10 μg/ml and 50 μg/ml respectively, diluted in lectin buffer). Subsequently, sections were washed with TBS and incubated for 30 min with streptavidin-alkaline phosphate complex (Vectastain ABC kit; Vector Laboratories, Burlingame, CA, USA). Enzyme reactivity of the alkaline phosphatase complex was visualized using naphthol-as-bisphosphate (Sigma Aldrich) as a substrate and hexatozised New Fuchsin (Sigma Aldrich) was used for simultaneous coupling. For staining with P-selectin (4 μg/ml; R&D Systems, Wiesbaden, Germany) and E-selectin fusion protein (20 μg/ml; R&D Systems) and with anti-ca19-9 antibody (1:100; Abcam, Cambridge, UK) sections were pretreated with 0.1% trypsin for 10 min at 37 C. Anti- CD15s (1:50; BD Pharmingen, Heidelberg, Germany) antibody was used without pretreatment. Subsequently, all sections were washed and non-specific binding was blocked by incubating the slides with the respective normal serum (1:10; DAKO, Hamburg, Germany) for 30 min at RT. Slides were incubated for 1 h at RT with selectin fusion proteins, or for 24 h at 4 C with primary antibodies, respectively, then rinsed and incubated for 30 min at RT with the respective secondary biotinylated antibody (DAKO) diluted 1:200. Sections were rinsed, incubated for 30 min with streptavidin-alkaline phosphate kit (Vector Laboratories) or streptavidin-horseradish peroxidase kit (Vector Laboratories). Alkaline phosphate was processed as described above and enzyme reactivity of the horseradish peroxidase complex was visualized by incubating with 0.02% hydrogen peroxide (Merck, Darmstadt, Germany)/ diaminobenzidine (0.5 mg/ml) (Sigma, St. Louis, USA) for 10 minutes. All sections were counterstained with Mayer`s haemalum diluted 1:1 in distilled water for 5 to 10 s, blued under running tap water and mounted using Crystal mount (Biomeda, Foster City, CA, USA). Slides were examined and photographed with a Zeiss Axiophot photomicroscope. Negative controls were treated in the same way except for replacing the lectin by lectin-blocking buffer and the primary antibody by the isotype-matched Ig, respectively (CA19-9 and CD15s: mouse IgM, DAKO; selectin fusion proteins: human IgG1 Fc, R&D Systems). Negative controls showed no immunoreactivity. 1590

Schnegelsberg et al: Lectin Histochemistry of Breast and Colon Cancer Cells Table II. Histochemistry of different colon and breast cancer cell lines and primary tumours with HPA, AAA, P-selectin and E-selectin fusion protein, CD15s and CA19-9. Immunostaining was evaluated using a modified immunoreactive score (IRS) (12). Colon cancer HT29 SW480 Cell pellets Primary tumour Cell pellets Primary tumour HPA Moderate Weak Negative Negative AAA Strong Moderate Strong Moderate to strong P-Selectin Weak to moderate Weak Moderate Weak E-Selectin Weak Negative Negative Negative CD15s Strong Weak to moderate; Signet ring cells strong Moderate Negative to weak CA19-9 Moderate Negative; Single signet ring cells moderate Negative Negative Breast cancer MCF7 T47D HBL100 Cell pellets Primary tumour Cell pellets Primary tumour Cell pellets Primary tumour HPA Moderate Weak to moderate Strong Moderate Negative Negative AAA Strong Moderate Strong Strong Strong Strong P-Selectin Moderate Moderate Moderate Moderate Weak Weak E-Selectin Negative Negative Negative Negative Weak Negative CD15s Strong Weak Moderate Moderate Moderate Moderate CA19-9 Weak Negative Negative Negative Negative Negative Immunohistochemistry was evaluated using a modified immunoreactive score (IRS) used previously (12). IRS is the product of intensitiy of immunostaining (none = 0; weak = 1; moderate = 2; strong = 3) and the percentage of positive tumour cells <5% = 0; 5-20% = 1; 20-50% = 2; 50-80% = 3; >80% = 4). A value of 0 was scored as no expression, values of 1-4 as weak, 5-8 as moderate and 9-12 as strong expression. Results In this study, the binding of different lectins and selectins, and expression of selectin ligands in metastasizing and nonmetastasizing colon and breast cancer cell lines were examined by immunohistochemistry. Staining results are summarized in Table II. HPA binding to tumour cells grown in vitro differed in staining intensity between the metastasizing and nonmetastasizing cell lines. Whereas the metastasizing cell lines (HT29, MCF7 and T47D) showed a moderate or intense HPA labelling, non-metastasizing cell lines (SW480 and HBL100) did not bind HPA (Figure 1). In the primary tumours, these differences in HPA binding of metastasizing and non-metastasizing colon and breast cancer cell lines were confirmed. However, staining intensity was in general weaker in the primary tumours than in in vitro grown cells and within single primary tumours, differences in lectinbinding intensity were detected. Thus, HPA bound to HT29, MCF7 and T47D cells in some areas of the primary tumour with moderate intensity, but in some areas, tumour cells were HPA negative (Figure 1). In SW480 and HBL100 primary tumours, HPA binding was restricted to very few cells only (<5%) (Figure 1). In contrast to HPA staining, the binding of AAA to colon and breast cancer cells (Table II) did not show pronounced differences between non-metastasizing and metastasizing cells. All colon and breast cancer cells grown in vitro showed strong AAA staining and the corresponding primary tumours were moderately to strongly stained by AAA (Figure 2). For the E-selectin-binding properties, a slight difference concerning metastatic and non-metastatic colon cancer cells was observed (Figure 3). Binding of E-selectin to in vitro grown non-metastasizing colon SW480 cells was absent compared to its weak binding in metastasizing HT29 cells. However, the primary tumours of these cell lines grown in SCID mice did not show any difference in E-selectin binding as no reactivity was observed. In breast cancer cells, in vitro grown non-metastasizing HBL100 cells showed weak staining with E-selectin fusion protein, whereas metastasizing MCF7 and T47D cells were negative. None of the corresponding primary tumours bound to E-selectin fusion protein. 1591

Figure 1. Lectin histochemistry of HT29 (A) and SW480 (B) cells grown in vitro and of HT29 (C) and SW480 (D) primary tumours. Note the strong HPA binding of metastasizing HT29 cells grown in vitro (A) compared to HPA-negative non metastasizing SW480 cells (B). SW480 primary tumours were also HPA negative, whereas HT29 tumours showed a heterogeneous staining pattern within a single tumour, with both HPA-negative and - positive areas. Arrows mark cells moderately stained with HPA. Figure 2. Lectin histochemistry of HT29 (A) and SW480 (B) cells grown in vitro and of HT29 (C) and SW480 (D) primary tumours. Cells showed moderate to strong staining with lectin AAA; staining intensity of in vitro-grown cells was slightly more pronounced. Staining intensity showed no considerable difference between the different cell lines. 1592

Schnegelsberg et al: Lectin Histochemistry of Breast and Colon Cancer Cells Figure 3. E-Selectin binding of HT29 (A) and SW480 (B) cells grown in vitro. Of HT29 cells grown in vitro, 5 to 20% showed strong staining with E-selectin fusion protein (A), whereas non-metastasizing SW480 cells were completely negative (B). Figure 4. P-Selectin binding of MCF-7 (A), T47D (B) and HBL-100 (C) cells grown in vitro. P-Selectin binding showed only slight differences between metastasizing MCF-7 (A) and T47D (B) and non-metastasizing HBL-100 (C) breast cancer cells. Figure 5. CD15s immunohistochemistry of HT29 (A) and SW480 (B) cells grown in vitro and of HT29 (C) primary tumours. Expression of CD15s was stronger in HT29 (A) than in SW480 (B) cells grown in vitro. In HT29 primary tumours (C), expression of CD15s is lower; only single signet ring cells (arrows) showed strong staining. Figure 6. CA19-9 immunohistochemistry of HT29 (A) and SW480 (B) cells grown in vitro. A percentage of HT29 cells grown in vitro showed strong staining with anti-ca19-9 antibody (A), whereas non-metastasizing SW480 cells were completely negative (B). The staining pattern is similar to that shown in Figure 3. 1593

Staining with P-selectin fusion protein exhibited differences between in vitro and in vivo grown metastatic and non-metastatic breast cancer cells. HBL100 cells and tumours demonstrated weak staining, while MCF7 and T47D cells and tumours bound to P-selectin fusion protein with a moderate intensity (Figure 4). P-Selectin binding to HT29 and SW480 colon cancer cells was nearly identical, cells grown in vitro reacted moderately (SW480), and weakly to moderately (HT29) with P-selectin fusion protein. In vivo staining showed weak P-selectin binding of the primary tumours. CD15s immunohistochemistry of tumour cells demonstrated there to be slight differences in staining intensity with anti-cd15s between the metastasizing and non-metastasizing colon cancer cell lines. Whereas HT29 cells grown in vitro showed strong staining, the nonmetastasizing SW480 cells reacted moderately with anti- CD15s (Figure 5). In the respective primary tumours staining intensity was reduced. HT29 tumours reacted weakly to moderately with the anti-cd15s antibody, with single signet ring cells showing strong staining. SW480 tumours were negative or showed few areas with weak staining intensity (Figure 5). MCF7 breast cancer cells grown in vitro reacted strongly with anti-cd15s, whereas T47D and HBL100 cells were moderately stained. Staining intensity of MCF7 primary tumours was weak; T47D and HBL100 tumours showed moderate staining intensity. Expression of CA19-9 was different in metastatic and nonmetastatic colon cancer cells (Figure 6). In vitro binding of non-metastasizing colon SW480 cells to CA19-9 was negative compared to weak to moderate binding of metastasizing HT29 cells. CA19-9 expression pattern of HT29 was similar to the staining pattern with E-selectin fusion protein; a proportion of the cells showed strong staining, the rest of the cells were negative (Figure 3 and 6). However, 20 to 50% of the HT29 cells expressed CA19-9, whereas only 5 to 20% of them bound to E-selectin fusion protein. In HT29 primary tumours, only single signet ring cells showed moderate staining with the anti-ca19-9 antibody (Figure 6). Apart from this observation, primary tumours of HT29 were CA19-9 negative, as were SW480 tumours (Figure 6). Metastasizing MCF7 cells grown in vitro exhibited weak CA19-9 binding site expression, whereas T47D and non-metastasizing HBL100 cells were CA19-9 negative. None of the primary tumours expressed binding sites for the CA19-9 antibody. Discussion HPA binds primarily to GalNac residues and with a lower affinity to GlcNac residues. It is a suitable tool to differentiate between metastasizing and non-metastasizing breast and colon carcinomas in both clinical and in xenograft studies (13). The present study was undertaken to investigate whether HPA-positive breast and colon cancer cells, which were metastatic in SCID mice, are also able to bind to selectins. The possibly overlapping binding specificities of HPA and some or all of the selectins might help to explain why HPA-positive cells are able to metastasize. This approach seems warranted as the identification of the physiological ligands for the selectins has been challenging because, like many other lectins, the selectins adhere to a variety of carbohydrate structures in vitro. This observation also applies to HPA (7, 8). Selectins are important for physiological processes, such as inflammation, immune response and haemostasis/ thrombosis, and they are also involved in a number of pathophysiological processes, including cancer metastasis (14). Initial adhesion events of cancer cells facilitated by selectins result in activation of integrins and release of chemokines, and are possibly associated with the formation of a microenvironment, which permits metastasis. Cancer cell interactions with selectins are possible due to the frequent presence of carbohydrate determinants acting as selectin ligands on the cell surface of tumour cells from various types of cancer. The present study shows that E- selectin fusion protein highlights differences in binding of metastatic and non-metastatic colon cancer cells grown in vitro, but not in vivo. About 5% to 20% of the metastatic HT29 cells grown in vitro bound strongly to E-selectin, whereas the non-metastatic SW480 cells were completely E- selectin negative. This finding is not surprising, as malignant cells are believed to bind directly to vascular E-selectin, thereby inducing extravasation and seeding of metastatic cells (14). This hypothesis is strengthened by the fact that a soluble E-selectin protein reduced experimental lung colony formation of HT29 cancer cells in cytokine-treated nude mice (15). Consistent with this finding, E-selectin binding of HT29 cells grown in vitro indicates their metastatic potential in this study. HT29 cells have been found to bind to E-selectin only, but not to P- or L-selectins (16). In this study, HT 29 cells grown in vitro expressed both E- and P-selectin-binding sites. However, no considerable difference in binding capacity for P-selectin was observed between metastasizing HT29 and non-metastasising SW480 colon cancer cells in vitro and in vivo. It has been shown that P-selectin promotes platelet tumour cell binding and facilitates metastasis in colon cancer, but a direct binding of colon carcinoma cells to endothelial P-selectin mediating their extravasation was not demonstrated (17, 18). Accordingly, HT29 cells did not firmly adhere to P-selectin, but only to E-selectin in cell flow assays in vitro (11). The influence of P-selectin binding on the metastatic potential of colon cancer cells seems to be complex. This complexity can explain why binding of colon cancer cells to P-selectin in histochemistry and their metastatic potential are not directly correlated, although P- selectin has been shown to facilitate metastatic initiation (17-1594

Schnegelsberg et al: Lectin Histochemistry of Breast and Colon Cancer Cells 19). P-Selectin is also essential for tumour cell adhesion to the endothelium and metastasis formation in breast cancer and P-selectin deficiency attenuates tumour growth and metastasis (20). Here, both metastatic MCF7 and T47D cells and non-metastatic HBL100 cells and their primary tumours, respectively, bind to P-selectin with a slight difference in labelling intensity. Several studies demonstrated an additional critical role for E-selectin in regulating tumour cell transendothelial migration in breast cancer (21-23). However, a previous study has already revealed that MCF7 and T47D cells did not bind to E-selectin (23). This is in accordance with our results showing no binding of metastatic MCF7 and T47D cells to E-selectin fusion protein when grown in vitro and in vivo. In vitro-grown HBL100 cells showed weak E-selectin binding, xenografted HBL100 cells were negative for E-selectin histochemistry. Thus, binding of breast cancer cells to E-selectin fusion protein is not correlated with development of pulmonary metastases in our xenograft model. The in vivo selectin ligands of the cancer cells are as yet not well characterized. The tetrasaccharide sialyl Lewis x (CD15s) has been identified as a prototype carbohydrate ligand for both P- and E-selectin, although all three selectins can bind sle x and sle a under appropriate conditions (13, 24, 25). Previous studies showed a correlation between expression of the E-selectin ligands sle x and sle a and adhesion to E-selectin (15, 16). Furthermore, the degree of selectin ligand expression is generally correlated with metastatic spread and hence poor prognosis of cancer patients (14, 15, 26). Consistent with this observation, metastatic HT29 cells grown in vitro expressed CD15s and CA19-9 and bound to E-selectin fusion protein. Nonmetastatic SW480 cells expressed no CA19-9, less CD15s than HT29 cells did and did not bind to E-selectin. These results confirm a correlation between CD15s and CA19-9 expression, in particular, and E-selectin binding in colon cancer cells. The tumour cell-endothelium adhesion via CA19-9 and E-selectin has been shown to be an important step in the metastatic cascade of colon cancer (26). Thus, the amount of the selectin ligand sle a on three colon cancer cell lines correlated with their metastatic potential and blocking of sle a reduced metastasis formation (26). However, in this study the percentage of HT29 cells expressing CA19-9 was higher than those binding to E-selectin and non-metastatic SW480 cells also expressed CD15s, probably indicating that the carbohydrate residues are linked to diverse protein backbones which have an influence on the ligand function. SLe a and sle x expression showed a positive correlation with the metastatic risk in breast cancer patients and was an independent prognostic indicator of survival, regardless of the size of the primary tumour and lymph node involvement (27-30). In our study, cells of all breast cancer cell lines expressed sle x, but no considerable difference between the cell lines was detected. Expression of sle a was also very similar in breast cancer cells; MCF7 cells grown in vitro reacted weakly, T47D and HBL100 cells grown in vitro and primary tumours of all three cell lines did not react with with anti-sle a antibody. Thus, the analysis of selectin binding and ligand expression was not able to differentiate between metastatic breast cancer cells developing pulmonary metastases and non-metastatic breast cancer cells, when xenografted into SCID mice. HPA and selectin histochemistry of the investigated cancer cells were different in staining intensity and staining pattern. These results suggest that HPA and selectins did not bind to identical glycotopes. Although the importance of selectins for tumour cell adhesion and metastatic spread is beginning to be well established, binding of malignant cells to selectins in histochemistry and expression of selectin ligands is of limited predictive value for the metastatic potential of these cells in our xenograft model, also depending on the tumour entity. The lectin AAA, produced by an orange-cup fungus, binds to α-1,6-linked fucose residues, and with a lower affinity to α-1,2-linked fucose residues and sialyl Le x, which expresses α-1,3-linked fucose residues (9, 31, 32). Thereby, the sialic acid residue of sialyl Le x bound to AAA adopts an orientation similar to that in the corresponding sialyl Le x /E-selectin complex (9). The experiments performed here in xenograft models did not show any correlation between binding of cells to AAA and E-selectin fusion protein. Furthermore, AAApositivity of breast and colon cancer cells was not associated with their metastatic potential. In parallel no significant differences in reactivity for the lectin AAA were shown for normal and malignant colorectal tissues (31). These results indicate that fucose residues, recognized by the AAA lectin, are at least not exclusively linked to glycoproteins specifically involved in processes of metastasis formation. In colon cancer cells, histochemistry with the lectins, selectins and selectin ligands showed generally stronger staining when cells were grown in vitro than in corresponding primary tumours. This observation confirms that the expression of carbohydrate structures on tumour cells grown in vitro and in vivo can vary as a result of transformational effects such as hypoxia, 3D growth and high cell density (33). Thus, HPA- and selectin-binding of cancer cells, indicating the metastatic potential of these cells, can also be influenced by the environment of the cells (34). It is now established that the epithelialmesenchymal transition (EMT) and the reverse process (MET) are central regulators of cellular plasticity in carcinomas and play an important role in cancer metastasis (35-37). When epithelial tumour cells disseminate, they lose their contact with the neighbouring epithelial cells and basal lamina, and acquire a more mesenchymal character with migratory and invasive properties (35-37). Cells grown in vitro and cells in the primary tumours grown in 1595

SCID mice might represent such different states of tumour cells undergoing EMT and MET, respectively, and hence differ in their carbohydrate residues. In summary, the carbohydrate residues of the breast and colon cancer cells recognised by HPA seem to be different to the glycotopes binding to the selectins. Histochemical analysis of selectin fusion protein binding and ligands for selectins are of limited predictive value for the metastatic potential in our xenograft model, depending on the tumour entity. Thus, E-selectin binding, as well as CA19-9 and CD15s immunohistochemistry, revealed differences in metastatic and non-metastatic colon cancer cells grown in vitro, while metastatic and non-metastatic breast cancer cells exhibited only slight differences in P-selectin binding. The different glycotope expressions, in particular of especially colon cancer cells grown in vitro and in vivo indicate that selectin- and HPA-binding properties of carcinoma cells are not static, but dynamic and might be related to the EMT and MET of cancer cells. Thereby, cancer cells grown in culture seem to reflect the mesenchymal and thus metastasizing phenotype, whereas xenografted cells in solid tumours rather possess epithelial characteristics. References 1 Schumacher U and Adam E: Lectin histochemical HPA-binding pattern of human breast and colon cancers is associated with metastases formation in severe combined immunodeficient mice. Histochem J 29: 677-684, 1997. 2 Brooks SA and Leathem AJ: Prediction of lymph node involvement in breast cancer by detection of altered glycosylation in the primary tumour. Lancet 338: 71-74, 1991. 3 Brooks SA, Lymboura M, Schumacher U and Leathem AJ: Histochemistry to detect Helix pomatia lectin binding in breast cancer: methodology makes a difference. J Histochem Cytochem 44: 519-524, 1996. 4 Schumacher U, Higgs D, Loizidou M, Pickering R, Leathem A and Taylor I: Helix pomatia agglutinin binding is a useful prognostic indicator in colorectal carcinoma. Cancer 74: 3104-3107, 1994. 5 Valentiner U, Hall DM, Brooks SA and Schumacher U: HPA binding and metastasis formation of human breast cancer cell lines transplanted into severe combined immunodeficient (SCID) mice. Cancer Lett 219: 233-242, 2005. 6 Schumacher U and Mitchell BS: Use of clinically relevant human-scid-mouse models in metastasis research. Trends Biotechnol 15: 239-241, 1997. 7 Brooks SA, Hall DM and Buley I: GalNAc glycoprotein expression by breast cell lines, primary breast cancer and normal breast epithelial membrane. Br J Cancer 85: 1014-1022, 2001. 8 Dwek MV, Ross HA, Streets AJ, Brooks SA, Adam E, Titcomb A, Woodside JV, Schumacher U and Leathem AJ: Helix pomatia agglutinin lectin-binding oligosaccharides of aggressive breast cancer. Int J Cancer 95: 79-85, 2001. 9 Haselhorst T, Weimar T and Peters T: Molecular recognition of sialyl Lewis(x) and related saccharides by two lectins. J Am Chem Soc 123: 10705-10714, 2001. 10 Konno A, Hoshino Y, Terashima S, Motoki R and Kawaguchi T: Carbohydrate expression profile of colorectal cancer cells is relevant to metastatic pattern and prognosis. Clin Exp Metastasis 19: 61-70, 2002. 11 Köhler S, Ullrich S, Richter U and Schumacher U: E-/P-selectins and colon carcinoma metastasis: first in vivo evidence for their crucial role in a clinically relevant model of spontaneous metastasis formation in the lung. Br J Cancer 102: 602-609, 2010. 12 Remmele W and Stegner HE: Recommendation for uniform definition of an immunoreactive score (IRS) for immunohistochemical estrogen receptor detection (ER-ICA) in breast cancer tissue. Pathologe 8: 138-140 (in German), 1987. 13 Laidler P and Litynska A: Tumor cell N-glycans in metastasis. Acta Biochim Pol 44: 343-357, 1997. 14 Konstantopoulos K and Thomas SN: Cancer cells in transit: the vascular interactions of tumor cells. Annu Rev Biomed Eng 11: 177-202, 2009. 15 Mannori G, Santoro D, Carter L, Corless C, Nelson RM and Bevilacqua MP: Inhibition of colon carcinoma cell lung colony formation by a soluble form of E-selectin. Am J Pathol 151: 233-243, 1997. 16 Mannori G, Crottet P, Cecconi O, Hanasaki K, Aruffo A, Nelson RM, Varki A and Bevilacqua MP: Differential colon cancer cell adhesion to E-, P-, and L-selectin: role of mucin-type glycoproteins. Cancer Res 55: 4425-4431, 1995. 17 Borsig L, Wong R, Feramisco J, Nadeau DR, Varki NM and Varki A: Heparin and cancer revisited: mechanistic connections involving platelets, P-selectin, carcinoma mucins, and tumor metastasis. Proc Natl Acad Sci USA 98: 3352-3357, 2001. 18 Borsig L, Wong R, Hynes RO, Varki NM and Varki A: Synergistic effects of L- and P-selectin in facilitating tumor metastasis can involve non-mucin ligands and implicate leukocytes as enhancers of metastasis. Proc Natl Acad Sci USA 99: 2193-2198, 2002. 19 Laubli H and Borsig L: Selectins as mediators of lung metastasis. Cancer Microenviron 3: 97-105, 2010. 20 Kim YJ, Borsig L, Varki NM and Varki A: P-selectin deficiency attenuates tumor growth and metastasis. Proc Natl Acad Sci USA 95: 9325-9330, 1998. 21 Shaker OG, Ay El-Deen MA, Abd El-Rahim MT and Talaat RM: Gene expression of E-selectin in tissue and its protein level in serum of breast cancer patients. Tumori 92: 524-530, 2006. 22 Tozeren A, Kleinman HK, Grant DS, Morales D, Mercurio AM and Byers SW: E-selectin-mediated dynamic interactions of breast and colon cancer cells with endothelial cell monolayers. Int J Cancer 60: 426-431, 1995. 23 Zen K, Liu DQ, Guo YL, Wang C, Shan J, Fang M, Zhang CY and Liu Y: CD44v4 is a major E-selectin ligand that mediates breast cancer cell transendothelial migration. Journal 3: e1826, 2008. DOI: 10.1371/journal.pone.0001826 24 Renkonen R, Mattila P, Majuri ML, Rabina J, Toppila S, Renkonen J, Hirvas L, Niittymaki J, Turunen JP, Renkonen O and Paavonen T: In vitro experimental studies of sialyl Lewis X and sialyl Lewis A on endothelial and carcinoma cells: crucial glycans on selectin ligands. Glycoconj J 14: 593-600, 1997. 25 Fukuda MN, Ohyama C, Lowitz K, Matsuo O, Pasqualini R, Ruoslahti E and Fukuda M: A peptide mimic of E-selectin ligand inhibits sialyl Lewis X-dependent lung colonization of tumor cells. Cancer Res 60: 450-456, 2000. 1596

Schnegelsberg et al: Lectin Histochemistry of Breast and Colon Cancer Cells 26 Sato M, Narita T, Kimura N, Zenita K, Hashimoto T, Manabe T and Kannagi R: The association of sialyl Lewis(A) antigen with the metastatic potential of human colon cancer cells. Anticancer Res 17: 3505-3511, 1997. 27 Renkonen J, Paavonen T and Renkonen R: Endothelial and epithelial expression of sialyl Lewis(x) and sialyl Lewis(a) in lesions of breast carcinoma. Int J Cancer 74: 296-300, 1997. 28 Matsuura N, Narita T, Mitsuoka C, Kimura N, Kannagi R, Imai T, Funahashi H and Takagi H: Increased level of circulating adhesion molecules in the sera of breast cancer patients with distant metastases. Jpn J Clin Oncol 27: 135-139, 1997. 29 Jeschke U, Mylonas I, Shabani N, Kunert-Keil C, Schindlbeck C, Gerber B and Friese K: Expression of sialyl lewis X, sialyl Lewis A, E-cadherin and cathepsin-d in human breast cancer: immunohistochemical analysis in mammary carcinoma in situ, invasive carcinomas and their lymph node metastasis. Anticancer Res 25: 1615-1622, 2005. 30 Nakagoe T, Fukushima K, Itoyanagi N, Ikuta Y, Oka T, Nagayasu T, Ayabe H, Hara S, Ishikawa H and Minami H: Expression of ABH/Lewis-related antigens as prognostic factors in patients with breast cancer. J Cancer Res Clin Oncol 128: 257-264, 2002. 31 Fernandez-Rodriguez J, Feijoo-Carnero C, Merino-Trigo A, Paez de la Cadena M, Rodriguez-Berrocal FJ, de Carlos A, Butron M and Martinez-Zorzano VS: Immunohistochemical analysis of sialic acid and fucose composition in human colorectal adenocarcinoma. Tumour Biol 21: 153-164, 2000. 32 Stelck S, Robitzki A, Willbold E and Layer PG: Fucose in alpha(1-6)-linkage regulates proliferation and histogenesis in reaggregated retinal spheroids of the chick embryo. Glycobiology 9: 1171-1179, 1999. 33 Warren L, Buck CA and Tuszynski GP: Glycopeptide changes and malignant transformation. A possible role for carbohydrate in malignant behavior. Biochim Biophys Acta 516: 97-127, 1978. 34 Koike T, Kimura N, Miyazaki K, Yabuta T, Kumamoto K, Takenoshita S, Chen J, Kobayashi M, Hosokawa M, Taniguchi A, Kojima T, Ishida N, Kawakita M, Yamamoto H, Takematsu H, Suzuki A, Kozutsumi Y and Kannagi R: Hypoxia induces adhesion molecules on cancer cells: A missing link between Warburg effect and induction of selectin-ligand carbohydrates. Proc Natl Acad Sci USA 101: 8132-8137, 2004. 35 Berx G, Raspe E, Christofori G, Thiery JP and Sleeman JP: Pre- EMTing metastasis? Recapitulation of morphogenetic processes in cancer. Clin Exp Metastasis 24: 587-597, 2007. 36 Polyak K and Weinberg RA: Transitions between epithelial and mesenchymal states: acquisition of malignant and stem cell traits. Nat Rev Cancer 9: 265-273, 2009. 37 Thiery JP, Acloque H, Huang RY and Nieto MA: Epithelial mesenchymal transitions in development and disease. Cell 139: 871-890, 2009. Received February 7, 2011 Revised March 31, 2011 Accepted April 1, 2011 1597