Relevance of risk assessment to exposed communities.

Current environmental policy tends to evaluate potential, theoretical exposure to health risks by evaluating one chemical or hazard at a time. Risk assessment techniques used by the U. S. Environmental Protection Agency (EPA) and other government agencies do not evaluate the cumulative impact of exposure to environmental contaminants. This problem is of particular significance to low-income and minority populations who tend to live in neighborhoods and work in locations that involve exposure to pollutants in air, water, and workplace activity. Certain areas within the border typify this lifestyle. The problem is further complicated by the fact that EPA operates separate programs for different "media." Exposure patterns in the border suggest the need for a cross-media pollution prevention approach. Minority recruitment into health research, a coordination of research approaches and dollars, and new resources for effective monitoring of minority communities could provide a basic assessment of the risks and their sources. Further research into the cumulative impacts of prevalent subsets of chemicals is also needed. Recent efforts in the Great Lakes may provide a model for this type of regional, cross-border effort.

The induction and maintenance of a polarised and differentiated epithelial cell phenotype is a multistage process that appears to depend at least in part on the expression and function of surface adhesion receptors mediating cellsubstratum as well as cell-cell interactions (Rodriguez-Boulan & Nelson, 1989). These adhesion molecules have been classified into four main groups (integrins, cadherins, immunoglobulins, selectins) (Hynes & Lander, 1992) of which integrins and cadherins comprise the main adhesion molecules expressed by normal and transformed epithelial cells (Hynes, 1992;Takeichi, 1991).
The integrins are a, heterodimeric transmembrane proteins comprising of at least 13a chains and 8p chains which are expressed by epithelial cells as well as other cell types (Hynes, 1987(Hynes, , 1989. The 1, integrin subfamily (or Very Late Antigens, VLA) is characterised by a 13 integrin chain noncovalently associated with one of at least eight different a chains to form receptors for extracellular matrix proteins including fibronectin, laminin and collagen (Hemler, 1990).
In addition some P1 integrins (@2P1 and ay P1) have been shown to function as intercellular adhesion molecules in keratinocytes growing in culture medium with low calcium concentration (Larjava et al., 1990;Carter et al., 1991).
The cadherins (cds) are Ca2'-dependent cell-cell adhesion molecules that connect cells via homotypic interactions (Takeichi, 1991). They are divided into subclasses, E-cd (epithelial cadherin or uvomorulin), P-cd (placental cadherin), N-cd (neural cadherin), T-cd and V-cd which share basic structure and show a selective tissue distribution (Edelman & Crossin, 1991). When cds are functionally expressed, the inactivation of other cell-cell adhesion molecules has little effect (Duband et al., 1987), indicating that cds play a major role in intercellular physical adhesion (Takeichi, 1991). E-cd is a 120 kDa transmembrane protein which is expressed by normal epithelial cells (Shiozaki et al., 1991). There is overwhelming evidence that the normal function of both integrins and cds is critical in the induction and maintenance of cell differentiation in vitro (Pignatelli & Bodmer, 1888;Del Buono et al., 1991;Takeichi, 1988;Takeichi, 1991). It also appears that changes in their expression and/or function occur relatively frequently in transformed cells in vivo and this is associated with loss of differentiation and relates to the biological behaviour of tumour cells (Pignatelli & Bodmer, 1990;Pignatelli et al., 1990aPignatelli et al., , 1991Edelman et al., 1989;Shiozaki et al., 1991).
Colorectal cancer is one of the commonest cancer in Western countries in which loss or down-regulation of both integrins (Pignatelli et al., 1990a) and E-cd (Edelman et al., 1989;Shiozaki et al., 1991) has been demonstrated immunohistochemically. Changes in both integrins and E-cd were found more frequently in poorly differentiated tumours in which the architecture and the glandular configuration were greatly impaired. Therefore loss of cell adhesion molecules may explain the phenotype and biological behaviour of poorly differentiated colorectal adenocarcinomas (Pignatelli & Bodmer, 1990).
To investigate the molecular mechanisms controlling the glandular differentiation of colorectal tumour cells and the role played by E-cd and 13 integrins, we have used a human colon carcinoma cell line (SW1222) which organises into glandular structures, with well defined polarity when cultured in three-dimensional collagen gel (Pignatelli & Bodmer, 1888;Pignatelli & Bodmer, 1989). We have investigated the effect of specific monoclonal antibodies recognising human E-cd (HECD-1) and the common P1 integrin chain (mAbl3) on the glandular differentiation of SW1222 cells. Here we show that E-cd and 131 integrins mediate the cell-cell and cell-collagen interactions required for the induction and maintenance of the glandular differentiation of colorectal tumour cells.
Collagen gel preparation Collagen gels were prepared using Vitrogen 100 collagen (Collagen, Palo Alto, CA) according to the manufacturer's instructions. Vitrogen 100 is 95-98% type-I collagen with the remainder being type-III collagen. Cells (2 x 105) were mixed with 2ml of the neutralised Vitrogen 100 collagen solution (pH 7.4 ± 0.2) and plated into 35 mm tissue culture dishes (Nunc, Roskilde, Denmark). Collagen gelation was then initiated by warming the collagen solution to 37°C for 60min. The gel was then overlaid with 1.0 ml of DMEM/ 10% FCS and this mixture was changed twice a week. In some experiments the cells were resuspended in 50 p1l of DMEM containing serial concentrations of the following monoclonal antibodies previously characterised: HECD-1 (human E cadherin, Shimoyama et al., 1989), mAbl3 (pi integrin, Akiyama et al., 1989), W6/32 (HLA class I, Barnstable et al., 1978). Each monoclonal antibody was also subsequently added to the cultured medium for four consecutive days. The plates were scored every day for glandular structures as follows: two hundred colonies were counted and the glandular structures identified under a phase-contrast IMT Diavert Leitz microscope (objective 32L/0.40). Glands were defined as cell aggregates composed of single columnar epithelial cells whose nuclei were polarised toward the basal surface of the cell and where cells were organised around a central lumen. Triplicate dishes were prepared for each experiment. Values were expressed as number of glandular structures per number of cell colonies. After four days, collagen gels were fixed with 10% neutral buffered formalin for 24 h, removed from the dishes, embedded in paraffin for 4 pim histological sections and stained with haematoxylin/ eosin.
Collagen binding assay Microtitre plates (Dynatech) were coated with 50 pI/well of human type-I collagen (Sigma), human type IV collagen (Sigma), mouse laminin (Collaborative Research) and bovine serum albumin (Sigma) at serial concentrations (5, 10, 20, 40 tLg ml-') and left uncovered in a laminar flow hood overnight to allow normal evaporation. The plates were then rinsed with phosphate buffered saline (PBS) and used in the binding assay. Trypsinised cells were washed three times in serum-free DMEM and resuspended with DMEM, 2.5 mg ml-' BSA with serial concentrations of HECD-1 and mAbl3 monoclonal antibodies. Approximately 5 x 104 cells per well were plated into previously coated 96-well Dynatech plates and allowed to attach for 1 h at room temperature. The supernatants were then removed and the unattached cells were washed away three times with PBS. The attached cells were fixed with 3% paraformaldehyde and stained with 0.5% toluidine blue in 3.7% paraformaldehyde. Cell attachment was estimated from absorbance measurements at 580 nm performed using an ELISA reader (Minireader II; Dynatech Labs, Inc., VA). Preliminary experiments had shown that the maximal attachment (60%) of SW1222 cells to collagen coated plates was reached after 1 h incubation (data not shown).
Cell-cell adhesion assay This was performed as described by Benchimol et al. (1990). Briefly a single cell suspension of SW1222 cell line was obtained by 3 min incubation at 37°C with 0.12% Bacto trypsin in PBS. After centrifugation, the cells were put through a 30-gauge needle in DMEM plus 0.8% FCS. A suspension of 3 x 106 cells in 3 ml DMEM in 30 ml polystyrene tubes was magnetically stirred at 37°C in an atmosphere of 5% CO2. The number of single cells was determined using a haemocytometer at time 0 and at 120 min. Duplicate cell suspensions were resuspended in DMEM containing each monoclonal antibody used for the collagen gel experiment (HECD-1, mAbl3). In addition PR3B1O monoclonal antibody which recognises carcinoembryonic antigen (CEA) and the non-specific cross-reacting antigen (NCA) (Pignatelli et al., 1990b) was used.

Beta-i integrin and E-cadherin expression by
immunocytochemical staining Colorectal carcinoma cell lines were grown on glass slides for immunocytochemical staining. For this purpose 2 x 106 cells were resuspended in 20 ml culture medium (DMEM/ 10%FCS) and plated in 9 cm Petri dishes containing autoclaved 4 well 'Multiwell' glass slides [C.A. Hendley Ltd, (Essex)]. Cell adherence and growth on the glass slides appeared identical to that seen in plastic Petri dishes. Cells were cultured on the slides for at least 2 days prior to staining by a standard avidin-biotin-complex indirect immunoperoxidase technique. E-cadherin expression by Western blot analysis Freshly scraped cells were lysed for 5 min at 96°C with 4% SDS, 5% 2-mercaptoethanol in 1 M Tris-HCI buffer (pH 6.8) followed by centrifugation at 10,000. Aliquots of 50 ig of total cell proteins in 30 p1 volume were loaded per lane onto 10% SDS-polyacrylamide gel. After electrophoresis, the samples were electroblotted onto nitrocellulose sheets. The sheets were incubated with 3% bovine serum albumin (BSA) for 30 min and then with HECD-l monoclonal antibody (20-50ogml -') for 1 h at room temperature. After seven washes the sheets were incubated with 1:1000 diluted horseradish peroxidase (HRP)-conjugated rabbit anti-mouse immunoglobulins (Dako, Denmark) for 1 h at room temperature. After further washes, the sheets were stained in Tris buffer (50 mM Tris-Hcl, pH 7.4) containing 1-chloro-4naphthol.

Results
Beta 1 integrin and E-cadherin expression on SW1222 and LSJ74T cells To examine the expression of P, integrins and E-cd SW1222 and LS174T cells were grown in glass slides for 48 h and then stained by avidin-biotin-complex indirect immunoperoxidase technique using specific monoclonal antibodies. In SW1222 cells, both P, integrin chain (mAbl3) and E-cd (HECD-1) were highly expressed on the cell membrane with accentuation in regions of cell-cell contacts (Figure la and b). The expression of E-cd on SW1222 cells was also confirmed by western blot analysis which showed the specific 120 kDa polypeptide ( Figure 2, lane 1).
LS174T, which is a moderately differentiated colon carcinoma cell line with poor intercellular cohesion and no ability to undergo morphological differentiation in collagen gel (Pignatelli & Bodmer, 1989), did not express E-cadherin by western blot analysis (Figure 2, lane 2) and immunostaining (Figure 1c).
The collagen binding of SW1222 cells is mediated by P, integrins SW1222 cells showed specific binding to type I collagen, type IV collagen and to a lesser extent to laminin (Figure 3). The type I collagen binding of SW1222 cells was specifically inhibited by the mAbl3 in a dose-dependent manner ( Figure  4). The monoclonal antibody to E-cd, as predicted, did not show any specific inhibition of the SW1222 collagen binding (Figure 4). showed no E-cadherin reactivity. The unmarked lane represents the molecular weight markers.

E-cadherin is
1978) was used as negative control. Each monoclonal antibody was subsequently added to the cultured medium for four consecutive days. The plates were scored every day for glandular structures as described in Materials and methods.
SW1222 cells grown in the presence of either HECD-1 (Figure 7a), mAbl3 (Figure 7b) formed small non coherent aggregates with ill-defined margins and undifferentiated morphology. The degree of glandular differentiation of SW1222 cells was inhibited up to 61% and 75% by the addition of mAbl3 and HECD-1 respectively ( Figure 6). No change in the morphology of SW1222 cells was seen by adding the control antibody (W6/32) (Figure 7c).

Discussion
In malignant neoplasia cells fail to fully differentiate and show reduced adhesiveness to one another which may be an important factor in enabling them to infiltrate surrounding tissues and subsequently to detach and migrate (metastasis) (Fidler & Hart, 1982). This loss of differentiation and adhesion is reflected in the cytological and architectural structures. The morphological assessment of the glandular configuration and evaluation of the preserved polarity where cell apex and base are readily distinguished are the most Protein concentration (,ug ml-') Figure 3 Binding of SWI222 cells to type I collagen (TYPE I), type IV collagen (TYPE IV), laminin (LM) and bovine serum albumin (BSA) using a cell adhesion assay. Trypsinised cells were plated into 96-well Dynatech plates previously coated with each extracellular matrix protein or BSA and allowed to attach for h at room temperature. Non-attached cells were washed away with PBS, and the attached cells were fixed with 3% paraformaldehyde and stained with 0.5% Toluidine blue in 3.7% formaldehyde. Cell attachment was estimated from absorbance measurements at 580 nm performed using an ELISA reader. Data shown represent the mean ± standard deviations of three determinations. U4aays. ine plates were scoiru vIyu uay lUI guinuumi oLlU%tUI9J under a phase contrast Diavert Leitz microscope (objective 32L/ Antibody concentration (,ug ml-1) 0.40). Values are expressed as number of glandular structures per number of cell colonies. Figure 4 Inhibition of SW1222 cell attachment to type I collagen by mAbl3 (P, integrin subunit). Cells were plated in microtitre wells coated with type I collagen (20 gLg ml-') and containing the indicated concentrations of monoclonal antibody mAbl3 (P, integrin chain) and HECD-1 (E-cadherin). As negative control no monoclonal antibody was added in some wells. Cell attachment was determined as described in Figure 3. reliable criteria to define the grade of malignancy of colorectal tumours (Jass et al., 1986). Thus colorectal carcinomas can be divided in three histological groups of low grade, average grade and high grade according to the degree of tubular differentiation. This classification reflects the behaviour of the tumour and significantly correlates with survival rate (Halvorsen & Seim, 1988). The molecular basis of glandular differentiation is therefore fundamental to our understanding of neoplastic cell behaviour.
In this study we show that both cell-cell and cell-collagen interactions are required for the induction and maintenance of the glandular differentiation of a colon carcinoma cell line (SW 1222) in collagen gel and are primarily mediated by two classes of cell adhesion molecules, E-cd and P, integrins. We have previously shown that the ability of SW1222 cells to undergo glandular differentiation is mediated by binding to collagen I matrix via a specific cell surface receptor (Pignatelli & Bodmer, 1988). Here we demonstrate that the functional collagen receptor mediating the morphological differentiation in 3D-collagen gel is a member of the PI integrin subfamily. The known PI integrin collagen receptors expressed by SW1222 cells are a2PI (VLA-2) and a3PI   (Pignatelli, 1990). However, the lack of sufficient amount of a subunit-specific monoclonal antibodies has not allowed us yet to identify the P, integrin molecule mediating the glandular differentiation in collagen gel.
Cell-cell interactions are also important in morphogenesis. with a non polarised distribution of marker proteins of apical and basal and lateral membrane domains (Sztul et al., 1987).
Recently it has become clear that cells express a multitude of cell-cell as well as cell matrix adhesion receptors which may control these complex mechanisms. Cadherins are considered to be important regulators of morphogenesis by their homophilic binding specificity. E-cd seems to mediate the selective epithelial cell adhesion which is required for the induction of glandular differentiation of SW1222. No inhibition of cell-cell interactions was seen using the anti-PI integrin monoclonal antibody (mAbl3) and the anti CEA/NCA monoclonal antibody (PR3B10) which have been shown to function as cell-cell adhesion molecules (Larjava et al., 1990;Benchimol et al., 1989). These results are in agreement with previous reports showing that as long as cadherins are functioning, the inactivation of other adhesion systems has little effect on cell-cell adhesion (Duband et al., 1987). Normal epithelial cells always express high levels of E-cd and P1 integrins on the cell surface. However both molecules are either lost or down-regulated in poorly differentiated colorectal as well as other malignant epithelial tumours (Pignatelli et al., 1990a;Pignatelli et al., 1991Pignatelli et al., , 1992Shiozaki et al., 1991). Interestingly in some tumours, there is heterogenous E-cd (Edelman et al., 1989;Shiozaki et al., 1991) and PI integrin expression (Pignatelli et al., , 1992 which is often confined to the cytoplasm with no clear cell surface expression. This obviously implies that in transformed cells still expressing cell adhesion molecules in the cytoplasm, the deregulation of cell-cell and cell-matrix interactions is due to loss of function with subsequent disorganisation of the cytoskeletal filaments which are structurally linked to E-cd (Takeichi, 1991) and integrins (Hynes, 1987).
There is increasing experimental evidence that low expression of E-cd and integrins seen in poorly differentiated tumours plays a major role in their biological behaviour. Frixen et al. (1991), have shown that human carcinoma cell lines with a dedifferentiated 'fibroblast-like' phenotype had lost E-cd and were highly invasive in an in vitro assay. Furthermore the invasive behaviour of dedifferentiated breast carcinoma cell lines was corrected by transfection with E-cd cDNA. Alternatively the introduction of a plasmid encoding E-cd-specific antisense RNA into noninvasive transformed cells rendered the cells invasive (Vleminckx et al., 1991) consistent with a suppressive invasive role for E-cd. Similarly Giancotti and Ruoslahti (1989) have shown by transfection experiments that the overexpression of the xj1 integrin in Chinese Hamster Ovary cells reestablishes normal growth control in vitro with loss of tumourigenicity in nude mice. However the functional cooperation of E-cd and integrins is not fully understood. It is likely that E-cd mediates the selective adhesion by their homotype binding allowing close contact of epithelial cells. In responsive cells which also express functional ,B, integrin molecules this will allow them to fully respond to the differentiating effect of extracellular matrix proteins (Pignatelli & Bodmer, 1988). Loss of expression and/or function of E-cd and integrins will therefore allow tumour cells to dedifferentiate and lose cohesiveness, properties which would facilitate invasion and metastasis.