Cancer Letters, 60 (1991) 143- 152 Elsevier Scientific Publishers Ireland Ltd.
143
Specific adhesion of carcinoembryonic antigen-bearing colorectal cancer cells to immobilized carcinoembryonic Larissa
V. Levin
Dioision 01655
and Thomas
ojHematology/Oncology,
antigen
W. Griffin
University
of Massachusetts
Medical School,
55 Lake Avenue
North,
Worcester,
MA
(U.S.A.)
(Received
9 July
(Accepted
9 August
1991) 1991)
Summary
CEA-producing cell lines (LoVo, HT29, LSl74T and LS174-S), correlated with membrane CEA expression as determined by FACS analysis. The results of these experiments add support to the concept that CEA may function as a specific homotypic cellular adhesion molecule for colorectal cancer cells.
Recent characterization of the genomic structure of carcinoembryonic antigen (CEA) is consistent with that of a cellular adhesion molecule. To examine this function in colorectal cancer, the adherence of cell lines to microtiter wells coated with CEA and welldescribed adhesive molecules was determined. The CEA-positive cell line LoVo and the CEAdevoid cell line H-Meso-l did not differ in adherence to the extracellular matrix proteins laminin, collagen and fibronectin, whereas LoVo cells adhered to CEA (10 pg/well) in a specific manner (43% bound cells vs. 1.5 % bound cells with BSA or cY-acidglycoprotein controls, P < 0.01) while H-MESOshowed no adhesion to CEA (C 0.6% bound cells). This adhesion of LoVo cells to CEA was not affected by co-incubation of cells with EDTA, sodium azide, or at 23OC. Howeuer, the CEA to CEA adhesive interaction was inhibited by a monoclonal antibody directed against an epitope in the N-terminal domain of the CEA molecule, and decreased by enzymatic removal of CEA from the LoVo cell membrane. The extent of adhesion to immobilized CEA by.four Correspondence to: Hematology/Oncology,
Larissa V. Levin, Division of University of Massachusetts Medical
School, 55 Lake Avenue,
North Worcester,
0304.3835/91/$03,50 Published and Printed
MA 01655,
Keywords: carcinoembryonic adhesion; colorectal cancer
cell
Introduction Carcinoembryonic antigen (CEA) , originally discovered by Gold and Friedman [6], is a human tumor marker widely used in clinical oncology. Clinical immunoassays of CEA in sera of patients are used for surveillance after colorectal cancer surgery and to monitor response to treatment. In recent years, the presence of CEA on colorectal cancer cells has been used for the immunolocalization of primary tumors and their metastasis [7,8], and as a target antigen for radioimmunotherapy [7,13] and therapy with immunotoxins [9,15]. CEA is a complex immunoreactive glycoprotein with a molecular weight of 180 000 Da containing 60% carbohydrate. The antigen is produced during human embryogenesis and is reexpressed in tumors of epithelial origin. The
U.S.A.
0 1991 Elsevier Scientific Publishers in Ireland
antigen;
Ireland
Ltd
144
highest tissue concentrations are found in adenocarcinomas of digestive system, as well as tumors of breast and lung. CEA is present on the surface of tumor cells and is also released into the extracellular fluid. CEA membrane expression is perhaps the most prevalent of phenotypic changes documented for human epithelial cancer cells. Although it is 25 years since its discovery, the biological function of CEA is still not welldefined. The high degree of glycosylation of CEA has made its characterization difficult. Recently, cDNA sequencing of the gene for the protein core of CEA has revealed a highly significant homology with members of the immunoglobulin gene superfamily [ 1,25,29]. These proteins include immunoglobulins, Tcell antigens, growth factor receptors, and intercellular adhesion molecules such as N-CAM [28] and MAG [19]. A key functional feature of the immunoglobulin gene superfamily proteins is their role in basic cell surface recognition events. The striking homology between CEA and other members of immunoglobulin gene superfamily has led to recent studies that suggest that CEA may play a role in homotypic cellular adhesion (Ca2+-independent aggregation of the cells through the interaction of CEA molecules on their surfaces). Benchimol et al. [2] reported the homotypic aggregation of both cultured human colorectal adenocarcinoma cells, and of rodent cells transfected with a functional CEA cDNA. Hostetter et al. [ll, 121 reported the binding of the colorectal cancer cell KM-12-C to CEA immobilized on plastic; two other colorectal cancer lines did not bind. These investigators related this binding to increased liver metastases by KM12-C in athymic nude mice after systemic injection of CEA. Updyke and Nicolson [27] have demonstrated that increased homotypic intercellular adhesion may play an important role in invasion and metastasis of tumors. Cell aggregates may have a higher probability than single cells of breaking away from a primary tumor, or have a better chance of survival in the circulation and anchoring in
other organs [27]. It is interesting to note that another member of CEA gene family, nonspecific cross-reacting antigen (NCA), expressed on the surface of hamster cells mediates adhesion of cells by homotypic interactions [ 181. In the present work, adhesion between highly purified CEA immobilized on plastic, and tumor cells bearing membrane-bound CEA was studied. Radiolabeled cells were used to quantitate the adherence. All CEA expressing cell lines studied specifically adhered to CEA. The extent of adhesion was reduced by enzymatic removal of CEA from the cell membrane, and blocked by co-incubation with an anti-CEA monoclonal antibody, consistent with CEA to CEA homotypic adhesion. Homotypic adhesion was further supported by a correlation between extent of adhesion and cell membrane CEA expression determined by FACS. No effect was seen by co-incubation with the calcium/magnesium chelating agent EDTA and the metabolic inhibitor sodium azide. Preliminary results of this work have been presented in [ 161. Materials and Methods Materials Highly purified CEA was isolated from cultured LS174T cells by affinity chromatography with an anti-CEA (clone 10) column. The purity was verified by analytical HPLC and SDS-PAGE. This CEA was a gift from Abbott Labs (Abbott Park, IL). Human fibronectin was from Biomedical Technologies, Inc. (Stoughton, MA). Collagen type I, laminin (from EHS mouse sarcoma cells), and alphar-acid glycoprotein were obtained from Sigma (St. Louis, MO). The murine monoclonal anti-CEA Cl10 antibody (IgGl) was from Abbott Labs. [ ‘251]CEA was obtained by iodination of 20 pg of CEA with 1 mCi of Na’251 in presence of iodobeads (Pierce, Rockford, IL) with subsequent gel filtration on Sephadex G-50. 51Cr for cell labeling was purchased from NEN-DuPont (Billerica, MA). Anti-transferrin receptor murine monoclonal
145
antibody 7D3 (IgGl) was a gift from Dr. V. Raso. The 48-well virgin styrene microtiter plates that had not been treated with trichloroacetic acid to induce cell adhesion were used for cell adhesion assays. The plates were obtained by a special order from Costar (Cambridge, MA). Human colorectal cell lines LoVo, HT29 [4], and LS174T were obtained from American Type Culture Collection (Rockville, MD). The human mesothelioma cell line, H-Meso-1, was derived from a surgical biopsy, and has been previosly described [21]. LS174T-C was a subclone of LS174T selected by serial passage in vivo for an increased ability to form lung metastasis in nude mice (S. Shah et al., pers. commun.). Phosphatidylinositolspecific phospholipase C (PI-PLC) from Bat. thuringiensis was a kind gift from Dr. Stanley Udenfriend (Roche Institute, Nutley, NJ). Fluoresceinated horse antimurine antibody was purchased from Flow Laboratories. MOPC antibody was from Sigma. Preparation of the microtiter wells Dilutions of affinity-purified CEA in PBS (200 PI/well) were used to coat the wells at 37OC for 2 h. At the end of the incubation, the liquid in the wells was aspirated and the wells were gently washed with PBS. The 200 ~1 of 4% BSA, which had been denatured at 80°C for 3 min, was added and the plate was incubated for another hour at 37OC to block non-specific binding sites. The wells were then washed twice with PBS to remove unabsorbed protein. In order to quantitate CEA that adhered to the wells, trace amounts of lz51labeled CEA (217 000 cpm/well) were added to concentrations of unlabeled CEA from 0.5 to 50 pg/ml and incubated for 2 h at 37OC. At the end of the incubation, the wells were washed with PBS 2 times and the adsorbed protein was solubilized (as in [20]) by rinsing the wells three times with 200 ~1 of 1% SDS in 0.5 N NaOH and counted in a gamma counter. For the cell adhesion experiments, CEA concentration used for well coating was 20 - 50 &ml.
Labeling the cultured cells with 5’Cr Cell monolayers were detached with trypsinEDTA, and were washed three times with medium containing 1 mg/ml heat-denatured BSA instead of fetal calf serum (this solution was used for all subsequent cell treatments). The cells were then incubated with Nazs1Cr04 (200 $i/ml) for 2 h at 37OC, pelleted and washed three times with the same medium. Two washes removed essentially all unbound radioactivity. The amount of radiolabel incorporated into the cells was sufficient for use in the cell adhesion experiments (usually not less than 150 000 cpm for 5 x 10’ cells). Cell adhesion assays Microtiter wells were coated in quadruplicate with 200 ~1 of 20 pg/ml solutions of proteins needed for the experiment in PBS and with two concentrations of CEA (20 pg/ml and 50 pg/ml) with subsequent blocking with BSA as described above. Wells coated with BSA, alphal-acid glycoprotein, collagen type I, laminin and fibronectin at 20 pg/ml were used as the controls. An aiiquot of 200 ~1 of 51Cr-labeled cell suspension (5 x lo5 cells/well) was added to each well and incubated for 2 h at 37OC. Afterwards, non-adherent cells were removed by washing the cells 3 times with the same medium as in cell labeling. Adherent cells were solubilized with SDS/NaOH solution and counted in a gamma counter. Adherent cells were expressed as the percentage of radioactivity bound per microtiter well. Final concentrations of 2 mM EDTA and 0.02% NaN3 per well were used where needed in the experiments. FACS analysis of CEA binding by the cells Cells were detached from T-flasks as described, put on ice (approx. 2 x lo6 cells/tube) and incubated for 30 min with excess of either anti-CEA antibody or MOPC antibody and were washed three times with the cold medium. After this, the cells were incubated with excess fluorescein isothiocyanate-conjugated horse anti-murine anti-
146
body at 4OC for 30 min, washed three times and fixed in 4% formaldehyde. For flow cytometry, cells were resuspended in 1 ml medium and analyzed in an Ortho flow cytometer . Incubation of LoVo cells with PI-PLC LoVo cells were seeded into a regular 24 well microtiter plate at 5 X lo5 cells/well for 24 h and washed twice with PBS and once with release buffer (25 mM Tris, pH 7.5; 0.25 mM sucrose; 10 mM glucose; 100 PM leupeptin; 100 PM TLCK and 5 mM iodoacetic acid). Seven units of PI-PLC in 300 ~1 of release buffer were added to the wells (in triplicate) and the plate was incubated at 37OC for 1 h 45 min. Control cells were treated with release buffer alone. Thereafter, the cells were washed twice with PBS and 1.1 pg of CEA (with trace amounts of [1251]CEA, approx. 4.8 x lo5 cpm/well) was added to each well in 300 ~1 of PBS and incubated at 37OC for 1 h 45 min. The cells were then washed twice with 1% BSA in PBS, solubilized and counted as described. Hocking the L.oVo cell adhesion to CEAcoated wells by anti-CEA antibody Aliquots of concentrated anti-CEA antibody solution (in order not to increase 200 ~1 volume in the wells appreciably) were added to CEA-coated wells together with the labeled cells to a final antibody concentrations of 10 and 100 pg/well and cells were allowed to adhere for 2 h. The same concentrations of the 7D3 antibody were used as a control. Also, LoVo cells which had been precoated with anti-CEA antibody at concentrations 100 and 500 pg/ml at 37OC for 30 min and washed twice with the medium, were added. BSAcoated wells with antibody coated LoVo cells served as controls. After a 2 h incubation the ceils were processed and radioactivity counted in a regular manner.
Immobilization of CEA on plastic In order to assess the coating of the bottoms
of microtiter wells with CEA after the incubation with CEA solution, the wells were inwith variable simultaneously cubated concentrations of cold CEA and a constant trace amount of radioactive [‘251]CEA per concentration point for 2 h, followed by removal of unbound CEA by washing, solubilization of bound CEA, and gamma counting. The amount of radioactivity bound to the wells with different concentrations of added CEA (in triplicates) is shown in Fig. 1. The amount of CEA bound to the wells was calculated using percent of bound radioactive CEA, which assumes that binding of cold and [ 1251]CEA was non-competitive. As can be seen from Fig. 1, the use of radiolabeled CEA antibody makes it possible to detect CEA coating in microtiter wells. At the range of CEA concentrations used there is a correlation between amounts of CEA added and CEA adsorbed in the wells. Specific adhesion of l.oVo cells to CEA-coated wells The experiments were performed to determine whether CEA-expressing tumor cells would adhere, in a specific manner, to CEA immobilized on plastic. Two cell lines were
0
1
2
3
4 CEA
Fig.1. Adherence
5 ADDED
6
7
8
9
10
(&
of fluid-phase CEA to CEA-coated microtiter wells. Variable amounts of unlabeled CEA and a fixed amount of [ 1’25]CEA were incubated in the wells for 2 h, as indicated in Methods and Materials. Values, mean of triplicate determinations * S.D. The straight line was fitted by using the method of the least squares.
al-acld ,l~YcOPrOtei”
Fig. 2.
WA Control
comqen type
1
Specific adhesion
fNbro-
,Urn,“,”
nect,n
CEA
CEA
%
1wig
of LoVo cells to CEA-coated
wells. ( Cl ) LoVo cells; ( n ) H-Meso-l cells. For details see Materials and Methods. of these experiments, Statistical significance was determined by two way analysis of variance. If effects were significant, StudentNewman-Keuls Multiple Comparision Procedure (MCP) were used for pairwise comparisions. The significance level was set at 5%. Values, mean of quadruplicate determinations f S.D. lP < 0.05 vs. BSA control for LoVo and H-Meso-l cells.
used: LoVo, a moderate CEA producer [5], and H-Meso-1, a mesothelioma cell line which does not produce CEA (unpublished data). For coating the wells, in addition to CEA (20 and 50 pg/ml) , BSA and or-acid glycoprotein (which has a similar to CEA carbohydrate content but is non-adhesive) were used. Also, the well known adhesive proteins derived from the extracellular matrix such as collagen type I,
Table 1. Adhesion
of LoVo cells in presence
Treatment
BSA CEA CEA CEA BSA CEA
Control + NaN, + EDTA Control (23’C) (23OC)
“See Materials and Methods
cpm/well (Mean f 7194 241158 238176 217075 3938 212554 for details.
fibronectin, and laminin served as controls for a membrane adhesion function of tumor cells. The extent of coating with extracellular matrix proteins was not quantitated, but the conditions were similar to those described in the literature [3,20]. It can be seen from Fig. 2 that cell adherence in control wells (with only BSA added) was very small (1.4 f 1.34% for LoVo and 0.9 f 0.75 for H-Meso-1). LoVo cells attached to CEA to a greater degree (29% and 43% at 4 and 10 pg added, respectively), than the CEA-devoid H-Meso-l cells (attachment of H-Meso-l for both concentrations is lower than 0.6%). Both cell lines adhered to collagen type I, fibronectin, and laminin, though the degree of adhesion of each cell type to a particular adhesion protein varied. These proteins were used as positive controls of adhesion. Most immunoglobulin supergene family members have demonstrated Ca*+-independent adhesion in tests of their intercellular adhesion properties. Lack of Ca*+ dependence for CEA adhesion was also reported by Benchimol et al. [2] for CEA. In contrast, the human biliary glycoprotein of the CEA superfamily and a mouse homolog have demonstrated Ca2+ and temperature dependence of homotypic adhesion [26]. We found in our system that Ca*’ was not required for adhesion, since EDTA addition did not appreciably diminish adhesion (Table I). Also, an active metabolic state of the cell is probably not important for the homotypic adhesion, because pretreatment of the cells with sodium
of EDTA. sodium
azide and at 23’C”. Percentage
S.D.) zt 1368 + 21710 f 12533 f 29302 ztz 2458 zt 32795
2.2 72.6 71.7 65.4 1.2 64
f + f f f f
0.4 6.5 3.8 8.8 0.7 9.9
of bound cells
148 Table
II.
Effect of phosphatidylinositol-specific
phospholipase
pre-treatment
on binding
of [ 1251]CEA by LoVo
cells?. Cell line
Test
LOVO
Control
LOVO
PI-PLC Pre-treatment
“The details of the experiment
(release buffer)
are described
S.D.)
11135
1605
6706
f
-
GIZ224
60.3
ZIZ2%
in Materials and Methods.
azide was without effect (see Table I). The adhesion of LoVo cells to immobilized CEA did not appear to be a temperature-dependent process. Binding of [‘25r]CEA by PI-PLC-digested LoVo cells Comparison of [ ‘251]CEA binding to LoVo cell monolayers treated with PI-PLC with those untreated showed that enzymatic digestion greatly decreased CEA binding by treated cells (see Table II). Cells treated with PI-PLC for 1 h 45 min bound only 60.3 * 2% fluid phase CEA as compared to untreated LoVo monolayers. Since it is known that CEA is anchored to cell membranes via glycosylphosphatidylinositol moiety [lo] and that the CEA is released from cell membranes after digestion with PIPLC [22], these data strongly suggest that adhesion occurs through homotypic interactions
Table 111. Inhibition of adhesion
% Control
cpm/well (Mean f
between solution.
CEA molecules
Inhibition
of CEA-to-CEA
on the cells and in
adhesion
by anti-
CEA antibody The inhibition of adhesion of LoVo cells to the CEA-coated wells by the anti-CEA murine monoclonal antibody Cl10 is shown in Table III. The control (7D3) antibody had no effect. The absence of increased inhibition with the higher concentration of specific antibody after a simultaneous addition, may be due to the use of intact biyalent antibody, which permits bridging of solid phase and cell-associated CEA at the higher antibody concentrations. Comparison
of
adhesion
of
various
CEACEA Binding of the various cell lines to the CEAcoated wells was compared. BSA- and expressing
of LoVo cells to CEA-coated
cells to immobilized
wells by anti-CEA
antibodya.
Treatment
Antibody concentration
cpm/well (mean * S.D.)
% Adherent cells
BSA Control
0 pg/well 0 pg/well 10 pg/well 100 wg/well 10 pg/well 100 pg/well 100 pg/ml 500 pg/ml 100 pg/ml 500 rg/ml
1598 110090 109362 100329 31478 28630 9632 5188 ‘71999 49938
1.4 100 99.3 91.1 28.6 26.0 8.7 5.0 63.4 45.5
CEA Control Simultaneous antibody
addition
Pretreatment with Cl10
of LoV0 cells antibody
of
7D3 Control Cl10 Anti-CEA BSA-coated Well control CEAW-coated Well
‘Details of this study are described in Materials and Methods. ‘P < 0.05 vs. the CEA control by Student Newman-Keuls MCP
f 435 f 11411 f 1323 f 7902 + 7902 f 3367 f 2381 ZIZ733 f 17574 •t 8681
f
0.4
f f f + + f f f
120 5.8’ 7.2’ 3.1’ 2.2 0.7 15.6’ 7.9’
149 100 90
fT?zzzq LOVO
-
Fj
HT29
m hxq
LS174T LS174T-C
80 H-MESO g tj g 9 = 8
E
T
7Q605040 -
F? 30 I? 20 10 -
colorectal cancer cells to immobilized CEA. The experimental Fig. 3. Comparison of adhesion of CEA-expressing design is described in Materials and Methods. Values, mean of quadruplicate determinations * S.D. P < 0.05 for (1) adhesion of all cell lines to CEA vs. BSA control and (2) adhesion of 7 LoVo, LS174T. LS174T-C and HT 29 to CEA vs. adhesion of H-Meso-l to CEA.
laminin-coated wells were used as negative and positive controls of adhesion, respectively. As can be seen from Fig. 3, all cell lines used adhered to laminin, while adherence to BSAcoated plates did not exceed 1% (only HT29 had 2.7% adherence). Among known CEA producers, HT29 had 4.58% adherence to Table IV. Cell line
H-MESOLS174T-C HT-29 LS174T LOVO
Comparison
of binding of colorectal cpm/well (mean f 3477 1824 6594 71431 92738
immobilized CEA, LoVo 14.1%, LS174T 12.4% and the subclone LS174T-C (2.7%). The results of flow cytometry analysis of these cell lines reacted with anti-CEA antibody are presented in Table IV. These data agree well with the results of adhesion of these cells to CEA coated wells indicating that the extent of
cancer cell lines to CEA and CEA surface expressions % Bound
FACS median
2.4 2.7 4.6 12.4 14.1
14 17 35 67 75
S.D.)
zt 1162 ztz 507 * 912 ztz 4762 zt 3006
“See Fig. 3 and Materials and Methods
for details
+ 0.8 * 0.8 z!z 0.6 * 0.8 f 4.6
channel
by FACS”
150
cellular adhesion to CEA is related to the extent of CEA expression on the cell membrane.
Discussion This work demonstrates that CEA expressing cell lines adhere to fluid-phase CEA immobilized on plastic. The results strongly suggest that CEA-expressing cells adhere in a specific manner to immobilized CEA via CEA molecules on their cell membrane, since, in contrast to LoVo cells, CEA-devoid H-Meso-l ceils did not adhere to the CEA-coated wells. On the other hand, very few LoVo cells bound to the wells coated with either BSA or alphai-acid glycoprotein (which also contains about 60% carbohydrate). The adherence of LoVo cells to CEA-coated wells was Ca2+-independent and occurred with azidetreated cells. Anti-CEA (but not control) monoclonal antibody inhibited this binding, probably by masking the CEA domains responsible for adhesion. The highest degree of the inhibition of adhesion was observed when the anti-CEA antibody was added to CEAcoated wells simultaneously with the suspension of LoVo cells. A domain model for CEA has been proposed, which consists of a 108 - 110 residue Nterminal domain which is relatively free of carbohydrates, three repeating loop internal domains of .178 residues, and a short 26 residue hydrophobic C-terminus that may be lost when the molecule is secreted. The IV-terminal domain resembles the immunoglobulin variable region [28,29] and therefore is an excellent candidate as the site of cell-to-cell interaction. The outcome of the experiments with antiCEA antibody blocking of homotypic CEA adhesion (Table III) suggests participation of the N-terminal domain of CEA, since the antibody used is specific to an epitope in this domain (S. Stroupe and G. Summerdon, pers. commun.). Inhibition experiments with other monoclonal antibodies specific for defined epitopes of CEA may shed light on the role of the different domains in homotypic adhesion. Since cross-linking of CEA molecules on the
cells with those on plastic might occur at high concentrations of the bivalent antibody, the use of Fab fragments of antibody also may be of interest. Diminished adherence of CEA to the PI-PLC-digested LoVo cells is in agreement with the results of Sack et al. [22] who found that a 3 h incubation with PI-PLC of the colorectal cancer cell line, LS174T, released 59.4% of CEA into the medium. The results of PI-PLC treatment and the binding of the various tumor cell lines (Table IV) strengthen the hypothesis that cell adhesion to immobilized CEA occurs via CEA on their surface, since the extent of adhesion correlated with the membrane CEA expression. Alternatively, CEA expression may reflect a differentiated cellular phenotype with increased propensity to intercellular adhesion [3]. Studies of CEA to CEA interaction in artificial model systems such as CEA-bearing liposomes or with CEA transfection experiments may be required to establish the homotypic nature of this binding. One of the observations made in this work was that LoVo cells did not adhere to BSAcoated wells by more than 2% while the adherence of these cells to immobilized CEA was consistently and significantly higher but did vary among experiments. Cells used in the experiments had been cultured from 2 to 9 days. Drewinko et al [4] reported that total cellular content of CEA in LoVo cells differed depending upon the number of days in culture. This may partly account for the variability in our experiments, but may not be the sole factor. In addition, detaching cells from the flasks with trypsin may produce cells with slightly altered membrane characteristics. Trypsin should not affect CEA at these conditions, since this molecule is extremely resistant to proteolytic digestion [23,24] but obviously, it may greatly affect the integrity of many other membranes and matrix proteins on LoVo cells. Trypsin treatment had no direct effect on the adhesion, but it might affect molecular environment of membrane bound CEA from experiment to experiment rendering a specific domain(s) on CEA responsible for homotypic adhesion more or less accessible to the interaction with the im-
151
mobilized CEA, thus changing the percent of cell adhesion. Recently, Benchimol et al. [2] have shown that CEA can function as an intercellular adhesion molecule, by mediating homotypic cultured human colon aggregation of adenocarcinoma cells (LS-180) as well as rodent cells transfected with functional CEA cDNA. In these studies, aggregation was determined by visual inspection in a hemocytometer of the number of single cells. Specificity was shown by the use of the cDNA construct in the antisense orientation as the control. Oikawa et al. [18] also reported that hamster cells transfected with CEA or NCA genes aggregated in homotypic and heterotypic manthe aggregation of the ner, although NCA-expressing cells was weaker. Lisowska et al. [17] have reported that the CEA molecules formed non-covalent dimers in solution, consistent with homotypic interaction. In the work of Hostetter et al. [ 11,12,14] it was suggested that adhesion properties of CEA may be involved in the ability of tumor cells to form organ-selective metastases. Of the three colorectal cell lines with comparable CEA content tested, one (KM-12C) increased its metastatic spread to the liver after simultaneous injection into nude mice with fluid phase CEA. This behavior was associated with increased adhesion of KM-12C to CEA-coated plastic (13.3%) in comparison to that of the other two colorectal cancer cell lines (4.7 and 3.7%), which was not different from control. In summary, this study adds further support to the concept that CEA functions as an intercellular adhesion molecule for colorectal cancer This interaction appears cells. homotypic, and may involve participation of the N-terminal domain of the CEA molecule. This may represent an oversimplification of the actual biological function of CEA: Williams and Barclay [28] h ave proposed that the functional evolution of interactions within the immunoglobulin superfamily has been from a primordial function involving a single domain interacting with itself between opposed membranes to heterotypic interactions and antigen
recognition. Other members of this superfamily have many functions other than cellular adhesion: participation in antigen recognition (TCR/CD3, surface Ig) , interaction with membrane ligands on lymphocytes (CD,), and signal transduction (PDGFR) . Further study on the biologic function of this important tumor marker is clearly warranted. Acknowledgement This study was supported grant ROI CA39748.
in part by PHS
References Barnett, T.R., Kretschmer, A., Austen, D.A., Goebel, S.J., Hart, J.T.. Eking, J.J. and Kamarck, M.E. (1989) Carcinoembryonic antigens: alternative splicing accounts for the multiple mRNAs that code for novel members of the
carcinoembryonic antigen family. J. Cell Biol., 267 - 276. Benchimol, S., Fuks, A., Jothy, S.. Beauchemin. Shirota, K. and Stanners, C. (1989) Carcinoembryonic
108, N.. an-
tigen, a human tumor marker, functions as an intercellular adhesion molecule. Cell, 57, 327 -334. Daneker, G.W., Piazza, A.J., Steel. Jr., G.D. and Mercurio, A.M. (1989) Relationship between extracellular matrix interactions and degree of differentiation in human colon carcinoma cell lines. Cancer Res., 49, 681- 686. Drewinko, B. and Yang, L.Y. (1980) Observations on the synthesis of carcinoembryonic antigen by an established human colonic carcinoma cell line. Oncology, 37. 336 - 342. Drewinko, B., Romsdahl, M.M.. Yang, L.Y., Ahearn. M.J. and Trujillo, J.M. (1976) Establishment of a human carcinoembryonic antigen-producing cinema cell line. Cancer Res., 36, 467 Gold, P. and Friedman, S.O. (1965) tumor-specific antigens in human colonic
colon adenocar-475. Demonstration of carcinoma by im-
munological tolerance and adsorption techniques. J. Exp.. Med., 121, 439-462. Goldenberg. D. (1988) Targeting of cancer with radiolabeled antibodies: prospects for imaging and therapy. Arch. Pathol., 112. 580 - 586. Griffin, T.W., Stevens. S., Brill, A.B., Collins, J-A., Bokhari, F., Stochl, M.C., Gionet, M., Rusckowski, M., Stroupe. S.D., Keifer. H.C., Sumerdon, G.A.. Johnson. D.K. and Hnatowich, D.J. (1991) Initial clinical study of indium-111 labeled clone 110 anti-CEA antibody in patients with colorectal cancer. J. Clin. Oncol., 9. 149 - 159. Griffin, T.W.. Haynes, L.R. and DeMartino, J.A. (1982) Selective
cytotoxicity
carcinoembryonic
of
a
ricin
A-chain-anti-
antigen antibody conjugate for a human
152 colorectal adenocarcinoma 69, 799-805. 10
11
12
13
14
15
16
17
18
19
cell line. J. Natl. Cancer
Hefta, S.A., Hefta, L., Lee, T., Paxton. R. and Shively, J. (1988) Carcinoembryonic antigen is anchored to membranes by covalent attachment to a glycosylphosphatidylinositol moiety: Identification of the ethanolamine linkage site. Proc. Natl. Hostetter, R.B., C.A., Thomas, bryonic antigen
oltgodendrocyte
Inst., 20
21
Acad. Sci., U.S.A., 85, 4648 -4652. Campbell, D.E., Chi, K., Fan, D., Toth, P. and Jessup, J.M. (1990) Carcinoemenhances metastatic potential of human
colorectal carcinoma. Arch. Surg., 125, 300 - 304. Hostetter, R.B., Augustus, L.B., Mankarious, R., Chi, K., Fan, D., Toth, C., Thomas, P. and Jessup, J.M. (1990) Carcinoembryonic antigen as a selective enhancer of col-
22
orectal cancer metastasis. J. Natl. Cancer Inst., 82, 380 - 385. ho, T., Qiu, H., Collins, J.A., Brtll, A.B., Johnson, D.K. and Griffin, T.W. (1991) Preclinical assessments of q-
23
labeled Cl 10 anti-CEA immunotoxin: a potent therapeutic immunoconjugate for human colon cancer. Cancer Res., 51, 255-260. Jessup, J.M. and Thomas, P. (1989) Carcinoembryonic antigen: Function in metastasis by human colorectal carcinoma. Cancer Metastas. Rev., 8, 263-280. Levin, L.V., Griffin, T.W., Childs, L.R., Davis, S. and Haagenson, D.E. (1987) Multiple anti-CEA immunotoxins active against human adenocarcinoma cells. Cancer Immunol. Immunother., 24, 202 - 206. Levin, L.V. and Griffin, T.W. (1991) Adhesion of carcinoembryonic antigen (CEA)-bearing human tumor cells to immobilized CEA. Proc. Am. Assoc. Cancer Res., 22, 23. Lisowska, E., Krop-Watorek, A. and Sedlaczek, P. (1987) Intermolecular interactions in CEA and NCA. Tumor Biol., 8, 337-338. Oikawa, S.H., Inuzuka, C.H., Kuroki, N., Matsuoka, Y., Kosaki, G. and Nakazato (1989) Cell adhesion activity of non-specific cross-reacting antigen (NCA) and carcinoembryonic antigen (CEA) expressed on Cl10 cell surface: homophiltc and heterophilic adhesion. Biochem. Biophys. Res. Commun., 164, 39-45. Poltorak, M., Sadoul, R., Keilhauer, G., Landa, C., Fahrig, T. and Schachner, M. (1987) Myelin-associated glycoprotein, a member of the LZ/HNK-1 family of neural in neuroncell adhesion molecules, is involved
24
25
26
27
28
29
and
oligodendrocyte-oligodencrocyte
interaction. J. Cell Blol., 105, 1893- 1899. Rauvala, H., Carter, W. and Hakomori, S-l. (1981) Studies of cell adhesion and recognition, J. Cell Biol., 88, 127 - 137. Reale, F.R., Griffin, T.W., Comptom, J-M., Graham, S., Townes, P.L. and Bogden, A.E. (1987) Characterization of a human malignant mesothelioma cell line. A biphasic solid and ascitic tumor model. Cancer Res., 47, 3328 - 3336. Sack, T.L., Gum, J., Low, M. and Kim, Y. (1988) Release of carcinoembryonic antigen from human colon cancer cells by phosphatidylinositol-specific phospholipase C. 3. Clin. Invest., 82, 586- 593. Shively, J.E., Kessler, M.J. and Todd, C.W. (1978) Amino-terminal sequences of the major tryptic peptides obtained from carcinoembryonic antigen by digestion with trypsin in the presence of Triton X-100. Cancer Res., 38, 2199-2208. Sundblad, G., Jornvall, H. and Hammarstrom. S. (1979) Structural and antigenic properties of carcinoembryonic antigen (CEA). Mol. Immunol., 16, 335- 339. Thompson, J.A., Pande, H., Paxton, R., Shively, L., Padma, A., Simmer, R.L., Todd, C.W., Riggs, A.D. and Shively, J.E. (1987) Molecular cloning of a gene belonging to the carcinoembryonic antigen gene family and discussion of a domain model. Proc. Nat). Acad. Sci. U.S.A., 84, 2965 - 2969. Turbide, C., Rojas, M., Stanners, C.P. and Beauchemin, N. (1991) A mouse carcinoembryonic antigen gene family member is a calcium-dependent cell adhesion molecule. J. Biol. Chem., 266, 1, 309-315. Updyke, T.V. and Nicolson, G.L. (1986) Malignant melanoma cell lines selected in vitro for increased homotypic adhesion properties have increased experimental metastatic potential. Clin. Exp. Met., 4, 273-284. Williams, A.F. and Barclay, A.N. (1988) The imsuperfamily-domains for cell surface munoglobulin recognition. Annu. Rev. Immunol., 6, 381-405. Zimmermann, W., Ortlieb, B., Friederich, R. and Von Kleist, S. (1987) Isolation and characterization of cDNA clones encoding the human carcinoembryonic antigen reveal a highly conserved repeating structure. Proc. Natl. Acad. Sci. U.S.A., 84. 2960-2964.
143
Specific adhesion of carcinoembryonic antigen-bearing colorectal cancer cells to immobilized carcinoembryonic Larissa
V. Levin
Dioision 01655
and Thomas
ojHematology/Oncology,
antigen
W. Griffin
University
of Massachusetts
Medical School,
55 Lake Avenue
North,
Worcester,
MA
(U.S.A.)
(Received
9 July
(Accepted
9 August
1991) 1991)
Summary
CEA-producing cell lines (LoVo, HT29, LSl74T and LS174-S), correlated with membrane CEA expression as determined by FACS analysis. The results of these experiments add support to the concept that CEA may function as a specific homotypic cellular adhesion molecule for colorectal cancer cells.
Recent characterization of the genomic structure of carcinoembryonic antigen (CEA) is consistent with that of a cellular adhesion molecule. To examine this function in colorectal cancer, the adherence of cell lines to microtiter wells coated with CEA and welldescribed adhesive molecules was determined. The CEA-positive cell line LoVo and the CEAdevoid cell line H-Meso-l did not differ in adherence to the extracellular matrix proteins laminin, collagen and fibronectin, whereas LoVo cells adhered to CEA (10 pg/well) in a specific manner (43% bound cells vs. 1.5 % bound cells with BSA or cY-acidglycoprotein controls, P < 0.01) while H-MESOshowed no adhesion to CEA (C 0.6% bound cells). This adhesion of LoVo cells to CEA was not affected by co-incubation of cells with EDTA, sodium azide, or at 23OC. Howeuer, the CEA to CEA adhesive interaction was inhibited by a monoclonal antibody directed against an epitope in the N-terminal domain of the CEA molecule, and decreased by enzymatic removal of CEA from the LoVo cell membrane. The extent of adhesion to immobilized CEA by.four Correspondence to: Hematology/Oncology,
Larissa V. Levin, Division of University of Massachusetts Medical
School, 55 Lake Avenue,
North Worcester,
0304.3835/91/$03,50 Published and Printed
MA 01655,
Keywords: carcinoembryonic adhesion; colorectal cancer
cell
Introduction Carcinoembryonic antigen (CEA) , originally discovered by Gold and Friedman [6], is a human tumor marker widely used in clinical oncology. Clinical immunoassays of CEA in sera of patients are used for surveillance after colorectal cancer surgery and to monitor response to treatment. In recent years, the presence of CEA on colorectal cancer cells has been used for the immunolocalization of primary tumors and their metastasis [7,8], and as a target antigen for radioimmunotherapy [7,13] and therapy with immunotoxins [9,15]. CEA is a complex immunoreactive glycoprotein with a molecular weight of 180 000 Da containing 60% carbohydrate. The antigen is produced during human embryogenesis and is reexpressed in tumors of epithelial origin. The
U.S.A.
0 1991 Elsevier Scientific Publishers in Ireland
antigen;
Ireland
Ltd
144
highest tissue concentrations are found in adenocarcinomas of digestive system, as well as tumors of breast and lung. CEA is present on the surface of tumor cells and is also released into the extracellular fluid. CEA membrane expression is perhaps the most prevalent of phenotypic changes documented for human epithelial cancer cells. Although it is 25 years since its discovery, the biological function of CEA is still not welldefined. The high degree of glycosylation of CEA has made its characterization difficult. Recently, cDNA sequencing of the gene for the protein core of CEA has revealed a highly significant homology with members of the immunoglobulin gene superfamily [ 1,25,29]. These proteins include immunoglobulins, Tcell antigens, growth factor receptors, and intercellular adhesion molecules such as N-CAM [28] and MAG [19]. A key functional feature of the immunoglobulin gene superfamily proteins is their role in basic cell surface recognition events. The striking homology between CEA and other members of immunoglobulin gene superfamily has led to recent studies that suggest that CEA may play a role in homotypic cellular adhesion (Ca2+-independent aggregation of the cells through the interaction of CEA molecules on their surfaces). Benchimol et al. [2] reported the homotypic aggregation of both cultured human colorectal adenocarcinoma cells, and of rodent cells transfected with a functional CEA cDNA. Hostetter et al. [ll, 121 reported the binding of the colorectal cancer cell KM-12-C to CEA immobilized on plastic; two other colorectal cancer lines did not bind. These investigators related this binding to increased liver metastases by KM12-C in athymic nude mice after systemic injection of CEA. Updyke and Nicolson [27] have demonstrated that increased homotypic intercellular adhesion may play an important role in invasion and metastasis of tumors. Cell aggregates may have a higher probability than single cells of breaking away from a primary tumor, or have a better chance of survival in the circulation and anchoring in
other organs [27]. It is interesting to note that another member of CEA gene family, nonspecific cross-reacting antigen (NCA), expressed on the surface of hamster cells mediates adhesion of cells by homotypic interactions [ 181. In the present work, adhesion between highly purified CEA immobilized on plastic, and tumor cells bearing membrane-bound CEA was studied. Radiolabeled cells were used to quantitate the adherence. All CEA expressing cell lines studied specifically adhered to CEA. The extent of adhesion was reduced by enzymatic removal of CEA from the cell membrane, and blocked by co-incubation with an anti-CEA monoclonal antibody, consistent with CEA to CEA homotypic adhesion. Homotypic adhesion was further supported by a correlation between extent of adhesion and cell membrane CEA expression determined by FACS. No effect was seen by co-incubation with the calcium/magnesium chelating agent EDTA and the metabolic inhibitor sodium azide. Preliminary results of this work have been presented in [ 161. Materials and Methods Materials Highly purified CEA was isolated from cultured LS174T cells by affinity chromatography with an anti-CEA (clone 10) column. The purity was verified by analytical HPLC and SDS-PAGE. This CEA was a gift from Abbott Labs (Abbott Park, IL). Human fibronectin was from Biomedical Technologies, Inc. (Stoughton, MA). Collagen type I, laminin (from EHS mouse sarcoma cells), and alphar-acid glycoprotein were obtained from Sigma (St. Louis, MO). The murine monoclonal anti-CEA Cl10 antibody (IgGl) was from Abbott Labs. [ ‘251]CEA was obtained by iodination of 20 pg of CEA with 1 mCi of Na’251 in presence of iodobeads (Pierce, Rockford, IL) with subsequent gel filtration on Sephadex G-50. 51Cr for cell labeling was purchased from NEN-DuPont (Billerica, MA). Anti-transferrin receptor murine monoclonal
145
antibody 7D3 (IgGl) was a gift from Dr. V. Raso. The 48-well virgin styrene microtiter plates that had not been treated with trichloroacetic acid to induce cell adhesion were used for cell adhesion assays. The plates were obtained by a special order from Costar (Cambridge, MA). Human colorectal cell lines LoVo, HT29 [4], and LS174T were obtained from American Type Culture Collection (Rockville, MD). The human mesothelioma cell line, H-Meso-1, was derived from a surgical biopsy, and has been previosly described [21]. LS174T-C was a subclone of LS174T selected by serial passage in vivo for an increased ability to form lung metastasis in nude mice (S. Shah et al., pers. commun.). Phosphatidylinositolspecific phospholipase C (PI-PLC) from Bat. thuringiensis was a kind gift from Dr. Stanley Udenfriend (Roche Institute, Nutley, NJ). Fluoresceinated horse antimurine antibody was purchased from Flow Laboratories. MOPC antibody was from Sigma. Preparation of the microtiter wells Dilutions of affinity-purified CEA in PBS (200 PI/well) were used to coat the wells at 37OC for 2 h. At the end of the incubation, the liquid in the wells was aspirated and the wells were gently washed with PBS. The 200 ~1 of 4% BSA, which had been denatured at 80°C for 3 min, was added and the plate was incubated for another hour at 37OC to block non-specific binding sites. The wells were then washed twice with PBS to remove unabsorbed protein. In order to quantitate CEA that adhered to the wells, trace amounts of lz51labeled CEA (217 000 cpm/well) were added to concentrations of unlabeled CEA from 0.5 to 50 pg/ml and incubated for 2 h at 37OC. At the end of the incubation, the wells were washed with PBS 2 times and the adsorbed protein was solubilized (as in [20]) by rinsing the wells three times with 200 ~1 of 1% SDS in 0.5 N NaOH and counted in a gamma counter. For the cell adhesion experiments, CEA concentration used for well coating was 20 - 50 &ml.
Labeling the cultured cells with 5’Cr Cell monolayers were detached with trypsinEDTA, and were washed three times with medium containing 1 mg/ml heat-denatured BSA instead of fetal calf serum (this solution was used for all subsequent cell treatments). The cells were then incubated with Nazs1Cr04 (200 $i/ml) for 2 h at 37OC, pelleted and washed three times with the same medium. Two washes removed essentially all unbound radioactivity. The amount of radiolabel incorporated into the cells was sufficient for use in the cell adhesion experiments (usually not less than 150 000 cpm for 5 x 10’ cells). Cell adhesion assays Microtiter wells were coated in quadruplicate with 200 ~1 of 20 pg/ml solutions of proteins needed for the experiment in PBS and with two concentrations of CEA (20 pg/ml and 50 pg/ml) with subsequent blocking with BSA as described above. Wells coated with BSA, alphal-acid glycoprotein, collagen type I, laminin and fibronectin at 20 pg/ml were used as the controls. An aiiquot of 200 ~1 of 51Cr-labeled cell suspension (5 x lo5 cells/well) was added to each well and incubated for 2 h at 37OC. Afterwards, non-adherent cells were removed by washing the cells 3 times with the same medium as in cell labeling. Adherent cells were solubilized with SDS/NaOH solution and counted in a gamma counter. Adherent cells were expressed as the percentage of radioactivity bound per microtiter well. Final concentrations of 2 mM EDTA and 0.02% NaN3 per well were used where needed in the experiments. FACS analysis of CEA binding by the cells Cells were detached from T-flasks as described, put on ice (approx. 2 x lo6 cells/tube) and incubated for 30 min with excess of either anti-CEA antibody or MOPC antibody and were washed three times with the cold medium. After this, the cells were incubated with excess fluorescein isothiocyanate-conjugated horse anti-murine anti-
146
body at 4OC for 30 min, washed three times and fixed in 4% formaldehyde. For flow cytometry, cells were resuspended in 1 ml medium and analyzed in an Ortho flow cytometer . Incubation of LoVo cells with PI-PLC LoVo cells were seeded into a regular 24 well microtiter plate at 5 X lo5 cells/well for 24 h and washed twice with PBS and once with release buffer (25 mM Tris, pH 7.5; 0.25 mM sucrose; 10 mM glucose; 100 PM leupeptin; 100 PM TLCK and 5 mM iodoacetic acid). Seven units of PI-PLC in 300 ~1 of release buffer were added to the wells (in triplicate) and the plate was incubated at 37OC for 1 h 45 min. Control cells were treated with release buffer alone. Thereafter, the cells were washed twice with PBS and 1.1 pg of CEA (with trace amounts of [1251]CEA, approx. 4.8 x lo5 cpm/well) was added to each well in 300 ~1 of PBS and incubated at 37OC for 1 h 45 min. The cells were then washed twice with 1% BSA in PBS, solubilized and counted as described. Hocking the L.oVo cell adhesion to CEAcoated wells by anti-CEA antibody Aliquots of concentrated anti-CEA antibody solution (in order not to increase 200 ~1 volume in the wells appreciably) were added to CEA-coated wells together with the labeled cells to a final antibody concentrations of 10 and 100 pg/well and cells were allowed to adhere for 2 h. The same concentrations of the 7D3 antibody were used as a control. Also, LoVo cells which had been precoated with anti-CEA antibody at concentrations 100 and 500 pg/ml at 37OC for 30 min and washed twice with the medium, were added. BSAcoated wells with antibody coated LoVo cells served as controls. After a 2 h incubation the ceils were processed and radioactivity counted in a regular manner.
Immobilization of CEA on plastic In order to assess the coating of the bottoms
of microtiter wells with CEA after the incubation with CEA solution, the wells were inwith variable simultaneously cubated concentrations of cold CEA and a constant trace amount of radioactive [‘251]CEA per concentration point for 2 h, followed by removal of unbound CEA by washing, solubilization of bound CEA, and gamma counting. The amount of radioactivity bound to the wells with different concentrations of added CEA (in triplicates) is shown in Fig. 1. The amount of CEA bound to the wells was calculated using percent of bound radioactive CEA, which assumes that binding of cold and [ 1251]CEA was non-competitive. As can be seen from Fig. 1, the use of radiolabeled CEA antibody makes it possible to detect CEA coating in microtiter wells. At the range of CEA concentrations used there is a correlation between amounts of CEA added and CEA adsorbed in the wells. Specific adhesion of l.oVo cells to CEA-coated wells The experiments were performed to determine whether CEA-expressing tumor cells would adhere, in a specific manner, to CEA immobilized on plastic. Two cell lines were
0
1
2
3
4 CEA
Fig.1. Adherence
5 ADDED
6
7
8
9
10
(&
of fluid-phase CEA to CEA-coated microtiter wells. Variable amounts of unlabeled CEA and a fixed amount of [ 1’25]CEA were incubated in the wells for 2 h, as indicated in Methods and Materials. Values, mean of triplicate determinations * S.D. The straight line was fitted by using the method of the least squares.
al-acld ,l~YcOPrOtei”
Fig. 2.
WA Control
comqen type
1
Specific adhesion
fNbro-
,Urn,“,”
nect,n
CEA
CEA
%
1wig
of LoVo cells to CEA-coated
wells. ( Cl ) LoVo cells; ( n ) H-Meso-l cells. For details see Materials and Methods. of these experiments, Statistical significance was determined by two way analysis of variance. If effects were significant, StudentNewman-Keuls Multiple Comparision Procedure (MCP) were used for pairwise comparisions. The significance level was set at 5%. Values, mean of quadruplicate determinations f S.D. lP < 0.05 vs. BSA control for LoVo and H-Meso-l cells.
used: LoVo, a moderate CEA producer [5], and H-Meso-1, a mesothelioma cell line which does not produce CEA (unpublished data). For coating the wells, in addition to CEA (20 and 50 pg/ml) , BSA and or-acid glycoprotein (which has a similar to CEA carbohydrate content but is non-adhesive) were used. Also, the well known adhesive proteins derived from the extracellular matrix such as collagen type I,
Table 1. Adhesion
of LoVo cells in presence
Treatment
BSA CEA CEA CEA BSA CEA
Control + NaN, + EDTA Control (23’C) (23OC)
“See Materials and Methods
cpm/well (Mean f 7194 241158 238176 217075 3938 212554 for details.
fibronectin, and laminin served as controls for a membrane adhesion function of tumor cells. The extent of coating with extracellular matrix proteins was not quantitated, but the conditions were similar to those described in the literature [3,20]. It can be seen from Fig. 2 that cell adherence in control wells (with only BSA added) was very small (1.4 f 1.34% for LoVo and 0.9 f 0.75 for H-Meso-1). LoVo cells attached to CEA to a greater degree (29% and 43% at 4 and 10 pg added, respectively), than the CEA-devoid H-Meso-l cells (attachment of H-Meso-l for both concentrations is lower than 0.6%). Both cell lines adhered to collagen type I, fibronectin, and laminin, though the degree of adhesion of each cell type to a particular adhesion protein varied. These proteins were used as positive controls of adhesion. Most immunoglobulin supergene family members have demonstrated Ca*+-independent adhesion in tests of their intercellular adhesion properties. Lack of Ca*+ dependence for CEA adhesion was also reported by Benchimol et al. [2] for CEA. In contrast, the human biliary glycoprotein of the CEA superfamily and a mouse homolog have demonstrated Ca2+ and temperature dependence of homotypic adhesion [26]. We found in our system that Ca*’ was not required for adhesion, since EDTA addition did not appreciably diminish adhesion (Table I). Also, an active metabolic state of the cell is probably not important for the homotypic adhesion, because pretreatment of the cells with sodium
of EDTA. sodium
azide and at 23’C”. Percentage
S.D.) zt 1368 + 21710 f 12533 f 29302 ztz 2458 zt 32795
2.2 72.6 71.7 65.4 1.2 64
f + f f f f
0.4 6.5 3.8 8.8 0.7 9.9
of bound cells
148 Table
II.
Effect of phosphatidylinositol-specific
phospholipase
pre-treatment
on binding
of [ 1251]CEA by LoVo
cells?. Cell line
Test
LOVO
Control
LOVO
PI-PLC Pre-treatment
“The details of the experiment
(release buffer)
are described
S.D.)
11135
1605
6706
f
-
GIZ224
60.3
ZIZ2%
in Materials and Methods.
azide was without effect (see Table I). The adhesion of LoVo cells to immobilized CEA did not appear to be a temperature-dependent process. Binding of [‘25r]CEA by PI-PLC-digested LoVo cells Comparison of [ ‘251]CEA binding to LoVo cell monolayers treated with PI-PLC with those untreated showed that enzymatic digestion greatly decreased CEA binding by treated cells (see Table II). Cells treated with PI-PLC for 1 h 45 min bound only 60.3 * 2% fluid phase CEA as compared to untreated LoVo monolayers. Since it is known that CEA is anchored to cell membranes via glycosylphosphatidylinositol moiety [lo] and that the CEA is released from cell membranes after digestion with PIPLC [22], these data strongly suggest that adhesion occurs through homotypic interactions
Table 111. Inhibition of adhesion
% Control
cpm/well (Mean f
between solution.
CEA molecules
Inhibition
of CEA-to-CEA
on the cells and in
adhesion
by anti-
CEA antibody The inhibition of adhesion of LoVo cells to the CEA-coated wells by the anti-CEA murine monoclonal antibody Cl10 is shown in Table III. The control (7D3) antibody had no effect. The absence of increased inhibition with the higher concentration of specific antibody after a simultaneous addition, may be due to the use of intact biyalent antibody, which permits bridging of solid phase and cell-associated CEA at the higher antibody concentrations. Comparison
of
adhesion
of
various
CEACEA Binding of the various cell lines to the CEAcoated wells was compared. BSA- and expressing
of LoVo cells to CEA-coated
cells to immobilized
wells by anti-CEA
antibodya.
Treatment
Antibody concentration
cpm/well (mean * S.D.)
% Adherent cells
BSA Control
0 pg/well 0 pg/well 10 pg/well 100 wg/well 10 pg/well 100 pg/well 100 pg/ml 500 pg/ml 100 pg/ml 500 rg/ml
1598 110090 109362 100329 31478 28630 9632 5188 ‘71999 49938
1.4 100 99.3 91.1 28.6 26.0 8.7 5.0 63.4 45.5
CEA Control Simultaneous antibody
addition
Pretreatment with Cl10
of LoV0 cells antibody
of
7D3 Control Cl10 Anti-CEA BSA-coated Well control CEAW-coated Well
‘Details of this study are described in Materials and Methods. ‘P < 0.05 vs. the CEA control by Student Newman-Keuls MCP
f 435 f 11411 f 1323 f 7902 + 7902 f 3367 f 2381 ZIZ733 f 17574 •t 8681
f
0.4
f f f + + f f f
120 5.8’ 7.2’ 3.1’ 2.2 0.7 15.6’ 7.9’
149 100 90
fT?zzzq LOVO
-
Fj
HT29
m hxq
LS174T LS174T-C
80 H-MESO g tj g 9 = 8
E
T
7Q605040 -
F? 30 I? 20 10 -
colorectal cancer cells to immobilized CEA. The experimental Fig. 3. Comparison of adhesion of CEA-expressing design is described in Materials and Methods. Values, mean of quadruplicate determinations * S.D. P < 0.05 for (1) adhesion of all cell lines to CEA vs. BSA control and (2) adhesion of 7 LoVo, LS174T. LS174T-C and HT 29 to CEA vs. adhesion of H-Meso-l to CEA.
laminin-coated wells were used as negative and positive controls of adhesion, respectively. As can be seen from Fig. 3, all cell lines used adhered to laminin, while adherence to BSAcoated plates did not exceed 1% (only HT29 had 2.7% adherence). Among known CEA producers, HT29 had 4.58% adherence to Table IV. Cell line
H-MESOLS174T-C HT-29 LS174T LOVO
Comparison
of binding of colorectal cpm/well (mean f 3477 1824 6594 71431 92738
immobilized CEA, LoVo 14.1%, LS174T 12.4% and the subclone LS174T-C (2.7%). The results of flow cytometry analysis of these cell lines reacted with anti-CEA antibody are presented in Table IV. These data agree well with the results of adhesion of these cells to CEA coated wells indicating that the extent of
cancer cell lines to CEA and CEA surface expressions % Bound
FACS median
2.4 2.7 4.6 12.4 14.1
14 17 35 67 75
S.D.)
zt 1162 ztz 507 * 912 ztz 4762 zt 3006
“See Fig. 3 and Materials and Methods
for details
+ 0.8 * 0.8 z!z 0.6 * 0.8 f 4.6
channel
by FACS”
150
cellular adhesion to CEA is related to the extent of CEA expression on the cell membrane.
Discussion This work demonstrates that CEA expressing cell lines adhere to fluid-phase CEA immobilized on plastic. The results strongly suggest that CEA-expressing cells adhere in a specific manner to immobilized CEA via CEA molecules on their cell membrane, since, in contrast to LoVo cells, CEA-devoid H-Meso-l ceils did not adhere to the CEA-coated wells. On the other hand, very few LoVo cells bound to the wells coated with either BSA or alphai-acid glycoprotein (which also contains about 60% carbohydrate). The adherence of LoVo cells to CEA-coated wells was Ca2+-independent and occurred with azidetreated cells. Anti-CEA (but not control) monoclonal antibody inhibited this binding, probably by masking the CEA domains responsible for adhesion. The highest degree of the inhibition of adhesion was observed when the anti-CEA antibody was added to CEAcoated wells simultaneously with the suspension of LoVo cells. A domain model for CEA has been proposed, which consists of a 108 - 110 residue Nterminal domain which is relatively free of carbohydrates, three repeating loop internal domains of .178 residues, and a short 26 residue hydrophobic C-terminus that may be lost when the molecule is secreted. The IV-terminal domain resembles the immunoglobulin variable region [28,29] and therefore is an excellent candidate as the site of cell-to-cell interaction. The outcome of the experiments with antiCEA antibody blocking of homotypic CEA adhesion (Table III) suggests participation of the N-terminal domain of CEA, since the antibody used is specific to an epitope in this domain (S. Stroupe and G. Summerdon, pers. commun.). Inhibition experiments with other monoclonal antibodies specific for defined epitopes of CEA may shed light on the role of the different domains in homotypic adhesion. Since cross-linking of CEA molecules on the
cells with those on plastic might occur at high concentrations of the bivalent antibody, the use of Fab fragments of antibody also may be of interest. Diminished adherence of CEA to the PI-PLC-digested LoVo cells is in agreement with the results of Sack et al. [22] who found that a 3 h incubation with PI-PLC of the colorectal cancer cell line, LS174T, released 59.4% of CEA into the medium. The results of PI-PLC treatment and the binding of the various tumor cell lines (Table IV) strengthen the hypothesis that cell adhesion to immobilized CEA occurs via CEA on their surface, since the extent of adhesion correlated with the membrane CEA expression. Alternatively, CEA expression may reflect a differentiated cellular phenotype with increased propensity to intercellular adhesion [3]. Studies of CEA to CEA interaction in artificial model systems such as CEA-bearing liposomes or with CEA transfection experiments may be required to establish the homotypic nature of this binding. One of the observations made in this work was that LoVo cells did not adhere to BSAcoated wells by more than 2% while the adherence of these cells to immobilized CEA was consistently and significantly higher but did vary among experiments. Cells used in the experiments had been cultured from 2 to 9 days. Drewinko et al [4] reported that total cellular content of CEA in LoVo cells differed depending upon the number of days in culture. This may partly account for the variability in our experiments, but may not be the sole factor. In addition, detaching cells from the flasks with trypsin may produce cells with slightly altered membrane characteristics. Trypsin should not affect CEA at these conditions, since this molecule is extremely resistant to proteolytic digestion [23,24] but obviously, it may greatly affect the integrity of many other membranes and matrix proteins on LoVo cells. Trypsin treatment had no direct effect on the adhesion, but it might affect molecular environment of membrane bound CEA from experiment to experiment rendering a specific domain(s) on CEA responsible for homotypic adhesion more or less accessible to the interaction with the im-
151
mobilized CEA, thus changing the percent of cell adhesion. Recently, Benchimol et al. [2] have shown that CEA can function as an intercellular adhesion molecule, by mediating homotypic cultured human colon aggregation of adenocarcinoma cells (LS-180) as well as rodent cells transfected with functional CEA cDNA. In these studies, aggregation was determined by visual inspection in a hemocytometer of the number of single cells. Specificity was shown by the use of the cDNA construct in the antisense orientation as the control. Oikawa et al. [18] also reported that hamster cells transfected with CEA or NCA genes aggregated in homotypic and heterotypic manthe aggregation of the ner, although NCA-expressing cells was weaker. Lisowska et al. [17] have reported that the CEA molecules formed non-covalent dimers in solution, consistent with homotypic interaction. In the work of Hostetter et al. [ 11,12,14] it was suggested that adhesion properties of CEA may be involved in the ability of tumor cells to form organ-selective metastases. Of the three colorectal cell lines with comparable CEA content tested, one (KM-12C) increased its metastatic spread to the liver after simultaneous injection into nude mice with fluid phase CEA. This behavior was associated with increased adhesion of KM-12C to CEA-coated plastic (13.3%) in comparison to that of the other two colorectal cancer cell lines (4.7 and 3.7%), which was not different from control. In summary, this study adds further support to the concept that CEA functions as an intercellular adhesion molecule for colorectal cancer This interaction appears cells. homotypic, and may involve participation of the N-terminal domain of the CEA molecule. This may represent an oversimplification of the actual biological function of CEA: Williams and Barclay [28] h ave proposed that the functional evolution of interactions within the immunoglobulin superfamily has been from a primordial function involving a single domain interacting with itself between opposed membranes to heterotypic interactions and antigen
recognition. Other members of this superfamily have many functions other than cellular adhesion: participation in antigen recognition (TCR/CD3, surface Ig) , interaction with membrane ligands on lymphocytes (CD,), and signal transduction (PDGFR) . Further study on the biologic function of this important tumor marker is clearly warranted. Acknowledgement This study was supported grant ROI CA39748.
in part by PHS
References Barnett, T.R., Kretschmer, A., Austen, D.A., Goebel, S.J., Hart, J.T.. Eking, J.J. and Kamarck, M.E. (1989) Carcinoembryonic antigens: alternative splicing accounts for the multiple mRNAs that code for novel members of the
carcinoembryonic antigen family. J. Cell Biol., 267 - 276. Benchimol, S., Fuks, A., Jothy, S.. Beauchemin. Shirota, K. and Stanners, C. (1989) Carcinoembryonic
108, N.. an-
tigen, a human tumor marker, functions as an intercellular adhesion molecule. Cell, 57, 327 -334. Daneker, G.W., Piazza, A.J., Steel. Jr., G.D. and Mercurio, A.M. (1989) Relationship between extracellular matrix interactions and degree of differentiation in human colon carcinoma cell lines. Cancer Res., 49, 681- 686. Drewinko, B. and Yang, L.Y. (1980) Observations on the synthesis of carcinoembryonic antigen by an established human colonic carcinoma cell line. Oncology, 37. 336 - 342. Drewinko, B., Romsdahl, M.M.. Yang, L.Y., Ahearn. M.J. and Trujillo, J.M. (1976) Establishment of a human carcinoembryonic antigen-producing cinema cell line. Cancer Res., 36, 467 Gold, P. and Friedman, S.O. (1965) tumor-specific antigens in human colonic
colon adenocar-475. Demonstration of carcinoma by im-
munological tolerance and adsorption techniques. J. Exp.. Med., 121, 439-462. Goldenberg. D. (1988) Targeting of cancer with radiolabeled antibodies: prospects for imaging and therapy. Arch. Pathol., 112. 580 - 586. Griffin, T.W., Stevens. S., Brill, A.B., Collins, J-A., Bokhari, F., Stochl, M.C., Gionet, M., Rusckowski, M., Stroupe. S.D., Keifer. H.C., Sumerdon, G.A.. Johnson. D.K. and Hnatowich, D.J. (1991) Initial clinical study of indium-111 labeled clone 110 anti-CEA antibody in patients with colorectal cancer. J. Clin. Oncol., 9. 149 - 159. Griffin, T.W.. Haynes, L.R. and DeMartino, J.A. (1982) Selective
cytotoxicity
carcinoembryonic
of
a
ricin
A-chain-anti-
antigen antibody conjugate for a human
152 colorectal adenocarcinoma 69, 799-805. 10
11
12
13
14
15
16
17
18
19
cell line. J. Natl. Cancer
Hefta, S.A., Hefta, L., Lee, T., Paxton. R. and Shively, J. (1988) Carcinoembryonic antigen is anchored to membranes by covalent attachment to a glycosylphosphatidylinositol moiety: Identification of the ethanolamine linkage site. Proc. Natl. Hostetter, R.B., C.A., Thomas, bryonic antigen
oltgodendrocyte
Inst., 20
21
Acad. Sci., U.S.A., 85, 4648 -4652. Campbell, D.E., Chi, K., Fan, D., Toth, P. and Jessup, J.M. (1990) Carcinoemenhances metastatic potential of human
colorectal carcinoma. Arch. Surg., 125, 300 - 304. Hostetter, R.B., Augustus, L.B., Mankarious, R., Chi, K., Fan, D., Toth, C., Thomas, P. and Jessup, J.M. (1990) Carcinoembryonic antigen as a selective enhancer of col-
22
orectal cancer metastasis. J. Natl. Cancer Inst., 82, 380 - 385. ho, T., Qiu, H., Collins, J.A., Brtll, A.B., Johnson, D.K. and Griffin, T.W. (1991) Preclinical assessments of q-
23
labeled Cl 10 anti-CEA immunotoxin: a potent therapeutic immunoconjugate for human colon cancer. Cancer Res., 51, 255-260. Jessup, J.M. and Thomas, P. (1989) Carcinoembryonic antigen: Function in metastasis by human colorectal carcinoma. Cancer Metastas. Rev., 8, 263-280. Levin, L.V., Griffin, T.W., Childs, L.R., Davis, S. and Haagenson, D.E. (1987) Multiple anti-CEA immunotoxins active against human adenocarcinoma cells. Cancer Immunol. Immunother., 24, 202 - 206. Levin, L.V. and Griffin, T.W. (1991) Adhesion of carcinoembryonic antigen (CEA)-bearing human tumor cells to immobilized CEA. Proc. Am. Assoc. Cancer Res., 22, 23. Lisowska, E., Krop-Watorek, A. and Sedlaczek, P. (1987) Intermolecular interactions in CEA and NCA. Tumor Biol., 8, 337-338. Oikawa, S.H., Inuzuka, C.H., Kuroki, N., Matsuoka, Y., Kosaki, G. and Nakazato (1989) Cell adhesion activity of non-specific cross-reacting antigen (NCA) and carcinoembryonic antigen (CEA) expressed on Cl10 cell surface: homophiltc and heterophilic adhesion. Biochem. Biophys. Res. Commun., 164, 39-45. Poltorak, M., Sadoul, R., Keilhauer, G., Landa, C., Fahrig, T. and Schachner, M. (1987) Myelin-associated glycoprotein, a member of the LZ/HNK-1 family of neural in neuroncell adhesion molecules, is involved
24
25
26
27
28
29
and
oligodendrocyte-oligodencrocyte
interaction. J. Cell Blol., 105, 1893- 1899. Rauvala, H., Carter, W. and Hakomori, S-l. (1981) Studies of cell adhesion and recognition, J. Cell Biol., 88, 127 - 137. Reale, F.R., Griffin, T.W., Comptom, J-M., Graham, S., Townes, P.L. and Bogden, A.E. (1987) Characterization of a human malignant mesothelioma cell line. A biphasic solid and ascitic tumor model. Cancer Res., 47, 3328 - 3336. Sack, T.L., Gum, J., Low, M. and Kim, Y. (1988) Release of carcinoembryonic antigen from human colon cancer cells by phosphatidylinositol-specific phospholipase C. 3. Clin. Invest., 82, 586- 593. Shively, J.E., Kessler, M.J. and Todd, C.W. (1978) Amino-terminal sequences of the major tryptic peptides obtained from carcinoembryonic antigen by digestion with trypsin in the presence of Triton X-100. Cancer Res., 38, 2199-2208. Sundblad, G., Jornvall, H. and Hammarstrom. S. (1979) Structural and antigenic properties of carcinoembryonic antigen (CEA). Mol. Immunol., 16, 335- 339. Thompson, J.A., Pande, H., Paxton, R., Shively, L., Padma, A., Simmer, R.L., Todd, C.W., Riggs, A.D. and Shively, J.E. (1987) Molecular cloning of a gene belonging to the carcinoembryonic antigen gene family and discussion of a domain model. Proc. Nat). Acad. Sci. U.S.A., 84, 2965 - 2969. Turbide, C., Rojas, M., Stanners, C.P. and Beauchemin, N. (1991) A mouse carcinoembryonic antigen gene family member is a calcium-dependent cell adhesion molecule. J. Biol. Chem., 266, 1, 309-315. Updyke, T.V. and Nicolson, G.L. (1986) Malignant melanoma cell lines selected in vitro for increased homotypic adhesion properties have increased experimental metastatic potential. Clin. Exp. Met., 4, 273-284. Williams, A.F. and Barclay, A.N. (1988) The imsuperfamily-domains for cell surface munoglobulin recognition. Annu. Rev. Immunol., 6, 381-405. Zimmermann, W., Ortlieb, B., Friederich, R. and Von Kleist, S. (1987) Isolation and characterization of cDNA clones encoding the human carcinoembryonic antigen reveal a highly conserved repeating structure. Proc. Natl. Acad. Sci. U.S.A., 84. 2960-2964.
