the ceacam1-l ser503 residue is crucial for inhibition of colon cancer cell tumorigenicity

The CEACAM1-L Ser503 residue is crucial for inhibition of colon cancercell tumorigenicity

BeÂ neÂ dicte FourneÁ s1,5, Svetlana Sadekova1,5, Claire Turbide1, SteÂ phanie LeÂ tourneau1 andNicole Beauchemin*,1,2,3,4

1McGill Cancer Centre, McGill University, Montreal, QueÂbec, Canada H3G 1Y6; 2Department of Biochemistry, McGill University,Montreal, QueÂbec, Canada H3G 1Y6; 3Department of Medicine, McGill University, Montreal, QueÂbec, Canada H3G 1Y6;4Department of Oncology, McGill University, Montreal, QueÂbec, Canada H3G 1Y6

CEACAM1 (also known as biliary glycoprotein, C-CAMor CD66a) is a cell adhesion molecule of the immunoglo-bulin family behaving as a tumor inhibitory protein incolon, prostate, liver, endometrial and breast cancers.Inhibition of tumor development is dependent upon thepresence of the long 71 ± 73 amino acid cytoplasmicdomain of the CEACAM1 protein (CEACAM1-L). Wehave recently de®ned a number of cis-acting motifs withinthe long cytoplasmic domain participating in tumor cellgrowth inhibition. These are Tyr488, corresponding to anImmunoreceptor Tyrosine-based Inhibition Motif, as wellas the three terminal lysine residues of the protein. In thisstudy, we provide evidence that treatment with phorbolesters leads to increased phosphorylation of in vivo 32P-labeled CEACAM1-L in mouse CT51 carcinoma cells,in the mouse 1MEA 7R.1 liver carcinoma cells and in 293human embryonic kidney cells transfected with theCeacam1-L cDNA. Basal level Ser phosphorylation wasabrogated by treatment with the staurosporine inhibitor,but not by the protein kinase C-speci®c inhibitorcalphostin C or other inhibitors such as H7 orsphingosine. Speci®c inhibitors of protein kinase A orcalmodulin kinase had only minimal e�ects on the levelsof basal or PMA-induced Ser phosphorylation. Further-more, PMA treatment of the CT51 cells induced cellspreading and cellular relocalization of the CEACAM1-Lprotein. Since Ser503 has been described as a PMA-induced phosphorylation site in other cell systems, weinvestigated whether Ser503 was involved in theseresponses in mouse intestinal cells. No di�erences werenoticed in the basal or the PMA-induced phosphorylationlevels, kinase inhibitor sensitivity or the PMA-inducedrelocalization of the protein between the wild-type and theSer503Ala mutant CEACAM1-L. However, we provideevidence that Ser503 participates in CEACAM1-L-mediated tumor inhibition as its mutation to an Ala ledto in vivo tumor development, contrary to the tumorinhibitory phenotype observed with the wild-type CEA-CAM1-L protein. Oncogene (2001) 20, 219 ± 230.

Keywords: CEACAM1; CEA; BGP; CD66a; C-CAM;tumor suppressor; colon cancer; Ser phosphorylation;PMA

Introduction

Control of a number of cellular processes, includingepithelial proliferation and di�erentiation, have beenshown to be highly dependent upon phosphorylationevents (Hunter, 2000). Several kinases participating inmany signal transduction pathways have been identi-®ed as key players in the phosphorylation of many cellsurface receptors or cell adhesion molecules (Gumbi-ner, 2000). Additionally, several cell adhesion mole-cules such as the cadherins or N-CAM play animportant role in preventing the development ordissemination of colon cancer (Kinzler and Vogelstein,1996; Gumbiner, 2000). Biliary glycoprotein (Bgp) is amember of the carcinoembryonic antigen (CEA)family. The nomenclature of this large gene familyhas recently been revised (Beauchemin et al., 1999)and Bgp is now referred to as CEACAM1.CEACAM1 is an immunoglobulin (Ig)-like proteinbearing either two or four Ig-like extracellulardomains. It is anchored to the cell membrane andassociates tightly with the actin cytoskeleton (Sadeko-va et al., 2000). It is expressed with two variants of itscytoplasmic domain, known as either the short(CEACAM1-S: 10 amino acids) or long tail(CEACAM1-L: 71 ± 73 amino acids). These di�erentisoforms result from alternative splicing events of theCEACAM1 gene and are well conserved in di�erentspecies (Barnett et al., 1989; Najjar et al., 1993;NeÂ dellec et al., 1995). CEACAM1 is generallyexpressed in epithelial cells of the gastro-intestinaltract, in hepatocytes as well as in endothelial andhemopoietic cells (Beauchemin and Lin, 1998).

CEACAM1 is down-regulated in microadenoma andadenoma phases of tumorigenesis (Neumaier et al.,1993; Nollau et al., 1997; Ilantzis et al., 1997). We haveprovided evidence that introduction of CEACAM1-Linto a tumorigenic mouse colon carcinoma cell line(CT51), lacking in its expression, led to abrogation ofin vivo tumor development (Kunath et al., 1995). Thisphenotype was entirely dependent upon the presence ofthe CEACAM1 protein with the longer cytoplasmic

Oncogene (2001) 20, 219 ± 230ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00

www.nature.com/onc

*Correspondence: N Beauchemin, McGill Cancer Centre, McGillUniversity, 3655 Promenade Sir William Osler, Montreal, QueÂ bec,Canada H3G 1Y65These authors contributed equally to the workReceived 31 July 2000; revised 10 October 2000; accepted 26October 2000

tail, as cells expressing CEACAM1-S behaved essen-tially as the wild-type cells. Similar results have beenobtained using prostate and breast carcinoma models(Hsieh et al., 1995; Luo et al., 1997). A number ofresidues within the longer domain participate in thisphenotype: in particular, a single point mutation ofTyr488, conforming to an Immuno-receptor TyrosineInhibition Motif (ITIM), was su�cient to reverse the invivo tumor cell growth inhibition. In addition, threelysines in the C-terminal region were essential for theCEACAM1-L-mediated tumor inhibitory phenotype(Izzi et al., 1999).

CEACAM1 performs other cellular functions. Itwas shown to function as an adhesion molecule(Ocklind and Obrink, 1982; Rojas et al., 1990, 1996;McCuaig et al., 1992). Formisano et al. (1995)demonstrated that rat CEACAM1 (also known aspp120 or HA4) is involved in internalization of theinsulin receptor. In the mouse, CEACAM1 acts as themouse hepatitis virus receptor (Dveksler et al., 1991),whereas it serves as a receptor for a number ofNeisseria gonoccoci in human neutrophils (Virji et al.,1996; Gray-Owen et al., 1997). Moreover, we haverecently shown that CEACAM1 participates in signaltransduction events mediated by Tyr phosphorylationof its cytoplasmic domain, particularly at residueTyr488 (Beauchemin et al., 1997). Tyr488 phosphor-ylation is induced by Src-like kinases or the insulinreceptor (Rees-Jones and Taylor, 1985; Brummer etal., 1995) and is controlled in part by residues locatedat the carboxy-terminal region of the CEACAM1cytoplasmic domain (Huber et al., 1999). EnhancedTyr phosphorylation of CEACAM1-L results inrecruitment of the cytosolic Tyr phosphatases SHP-1and SHP-2 to the cell surface where they bind to anddephosphorylate CEACAM1-L (Beauchemin et al.,1997; Huber et al., 1999). Hauck et al. (1998) havealso determined that activation of CEACAM1 inneutrophils by a recombinant Neisseria opa52 proteininduced its Tyr phosphorylation by the Fgr and HckTyr kinases leading to activation of Rac1, PAK andthe Jun-N-terminal kinase.

Many Ser/Thr kinase consensus sites are presentwithin the CEACAM1 cytoplasmic domain such asthose for cAMP-dependent protein kinase, calmodulinkinase, p34cdc2 and glycogen synthase kinase 3(Edlund et al., 1998). In addition, 13 out of the 17Ser and Thr in CEACAM1-L represent potential PKCphosphorylation consensus sequences (Edlund et al.,1998). Odin et al. (1986) have demonstrated thatCEACAM1 was basally phosphorylated on Ser andThr residues. More recently, in vitro Ser phosphoryla-tion of the cytoplasmic domain of CEACAM1mediated by the protein kinase C (PKC) b2 isoformhas been reported. In this study, the use of syntheticpeptides mimicking the Ser449 environment hasidenti®ed this residue as the PKC phosphorylatedresidue (Edlund et al., 1998).

Relatively little is known about the Ser/Thr kinasescontrolling CEACAM1 functions. CEACAM1-L-de-pendent internalization of the insulin receptor

requires phosphorylation of Tyr488 and Ser503 onthe cytoplasmic domain of CEACAM1-L (Formisanoet al., 1995). In addition, CEACAM1-L-mediated bileacid e�ux requires Ser503 phosphorylation and isregulated by Tyr488 (Sippel et al., 1994). As Ser503has been described as functionally important in manycellular or functional contexts, we questioned whetherit played a signi®cant role in inhibition of tumordevelopment. Our results show that Ser503 is crucialfor this phenotype. We also demonstrate that thewild-type CEACAM1-L is basally phosphorylatedand that the phosphorylation is enhanced in vivoafter phorbol ester treatment of either mouse CT51colon carcinoma cells, mouse 1MEA 7R.1 livercarcinoma cells or human 293 embryonic kidneycells. Basal or PMA-induced CEACAM1-L Serphosphorylation is not dependent upon proteinkinase C activity as treatment with calphostin C, aspeci®c PKC inhibitor, did not change the phosphor-ylation of the protein. Other inhibitors to di�erentSer/Thr kinases similarly had no e�ect on overalllevels of CEACAM1-L Ser phosphorylation. TheSer503Ala mutation does not dramatically alter thebasal or PMA-induced CEACAM1-L phosphoryla-tion. Furthermore, PMA treatment of CT51 cellscaused cell spreading and relocalization of wild-typeCEACAM1-L, but mutation at Ser503 did notchange this behavior.

Results

Ser503 is essential for CEACAM1-dependent inhibitionof tumor cell development

We have previously demonstrated that the longcytoplasmic domain of the CEACAM1 protein iscrucial for inhibition of mouse colon tumor celldevelopment (Kunath et al., 1995). Similar resultshave also been obtained in prostate, breast andbladder carcinoma models (Hsieh et al., 1995; Kleiner-man et al., 1995; Luo et al., 1997). Ser503 has beenshown to play a crucial role in insulin receptorinternalization (Formisano et al., 1995) and in bilesalt extrusion (Sippel et al., 1994). We thereforeinvestigated the role of Ser503 in CEACAM1-L-mediated tumor inhibition. The Ser503 residue wasmutated to an Ala and, after retroviral-mediatedinfections with the CEACAM1-L Ser503 vector,CT51 cells expressing this mutant protein werederived. We veri®ed that the protein was expressedat the surface of CT51 cells by staining with anti-CEACAM1 antibodies and determined the levels ofexpression over background by FACS analyses(Figure 1b). We have previously demonstrated thatin vivo inhibition of tumor development required`physiological' levels of CEACAM1 expression (Tur-bide et al., 1997) and therefore CT51 cell populationsexpressing CEACAM1 at approximately 2 ± 4-foldover background were selected for these assays. Twoindependent tumor assays were conducted with

CEACAM1 in tumor inhibitionB FourneÁs et al

220

Oncogene

transfected CT51 cell populations injected intosyngeneic BALB/c mice (minimum of 10 mice/experiment). As shown in Table 1, the CT51 parentalcells, the control transfected cells (neo) or cellsexpressing a natural variant of the CEACAMprotein containing a short cytoplasmic domain(CEACAM1-S) developed a substantial number oftumors during the 42 ± 56 days of the in vivo tumorassays. No signi®cant inhibition of tumor developmentwas seen relative to the parental cells with either theneo control cells or cells expressing the CEACAM1-Snatural variant (P40.05), as reported previously

(Kunath et al., 1995). On the other hand, only 31%(P=0.002) of the mice injected with CT51 cellpopulation expressing the wild-type CEACAM1-Lprotein developed tumors (Table 1) (Kunath et al.,1995; Izzi et al., 1999). When CT51 cell populationsexpressing the Ser503 point mutant were injected intomice, 18/20 mice developed tumors (Table 1, Ser503,P40.05). This result is comparable to that of theparental, control transfected cells or those carrying theCEACAM1-S protein. Therefore, the reversion of thetumor inhibitory phenotype with the mutantCEACAM1-L-Ser503 is indicative that the Ser503

Figure 1 Sequence of mutants used and expression of CEACAM1 at the cell surface. (a) Graphic representation of CEACAM1-Lpoint and deletion mutants. Stars represent the stop codon inserted into the cDNAs at positions corresponding to the given aminoacid. (b) CEACAM1-transfected CT51 cells were processed for cell surface analyses by FACS using the AgB10 monoclonalantibody and FITC-labeled goat anti-rat IgGs. Numbers under the names indicate the median ¯uorescence of the cells labeled withboth primary and secondary antibodies relative to that of cells labeled with secondary antibody alone. Solid lines correspond toaddition of primary and secondary antibodies. Dashed lines represent pro®les of cells incubated with the secondary ¯uorescently-labeled antibody alone

Oncogene


221

residue is important in the regulation of tumor cellgrowth mediated by CEACAM1-L.

Phosphorylation of the CEACAM1-L protein inepithelial cells

CEACAM1-L is basally phosphorylated on Ser andThr residues in rat hepatocytes and this phosphoryla-tion is enhanced in PMA-treated transfected COS cells(Odin et al., 1986; McEntire et al., 1989; Culic et al.,1992; Sippel et al., 1994). However, little is knownabout the regulation of CEACAM1-L phosphorylationin epithelial cells. We therefore investigated the e�ectsof various stimuli and protein kinase inhibitors on Ser/Thr phosphorylation of CEACAM1-L in a number ofepithelial cell lines. Human embryonic kidney cells(HEK293), mouse CT51 colon carcinoma cells ormouse BNL 1MEA 7R.1 liver carcinoma cellstransiently or stably expressing mouse CEACAM1-Lwere metabolically-labeled with 32P-orthophosphateand treated with the phorbol ester PMA. Cells werelysed and the CEACAM1-L protein subjected toimmunoprecipitation with the CC1 anti-CEACAM1monoclonal antibody or the rabbit polyclonal anti-CEACAM1 antibodies (Ab 231 or 2456). Transfectionof CEACAM1-L into these cell lines revealed animmunoprecipitated 32P-labeled CEACAM1-L protein(Figure 2a,b lanes 2; 2c, lane 3) migrating as a basally-phosphorylated 110 ± 120 kDa protein. This basalphosphorylation was also observed in serum-starvedcells (data not shown). Treatment of the transfectedcells with PMA for 10 min enhanced phosphorylationby a factor of 2 ± 5-fold (Figure 2a,b, lanes 3; 2c, lane4). Maximal enhancement of phosphorylation wasnoticed at PMA concentrations greater than 250 nM,although a net increase was also observed at 25 nM(data not shown). Therefore, PMA induced phosphor-ylation of the mouse CEACAM1-L protein in humanor mouse epithelial cells.

Effect of Ser/Thr protein kinase inhibitors onCEACAM1-L phosphorylation

It has previously been suggested that either proteinkinase C or protein kinase A might be involved in invivo or in vitro CEACAM1-L phosphorylation (Lin

and Guidotti, 1989; Sippel et al., 1994). Edlund et al.(1998) demonstrated that a membrane-proximal site(SRKS/T), present in the CEACAM1 short cytoplas-mic domain, was phosphorylated in vitro afterincubation with the PKC b2 isoform. In our system,we have also observed the stimulation of CEACAM1-Lphosphorylation by PMA treatment, known to activate

Table 1 Tumorigenicity assays with the CEACAM1-L Ser503 mutant

Cell line Experiment Latency (days) Incidence Total number of tumors Inhibition (%) P value relative CT51

CT51 1 42 7/102 53 9/10 16/20 20 1.0

neo 1 42 7/102 53 8/10 15/20 25 40.05

CEACAM1-S 1 42 7/102 56 9/10 16/20 20 40.05

CEACAM1-L-WT 1 42 3/102 56 6/19 9/29 69 50.05

CEACAM1-L-S503A 1 42 10/102 56 8/10 18/20 10 40.05

Figure 2 PMA enhances CEACAM1-L phosphorylation inepithelial cells. HEK293 human embryonic kidney cells (a),CT51 mouse colon carcinoma cells (b) or BNL 1MEA7R.1mouse liver carcinoma cells (c) were either transiently or stablytransfected with either the empty vector or a vector encoding themouse Ceacam1-L cDNA. Cell populations or clones weremetabolically labeled with 32P-orthophosphate and treated withthe phorbol ester PMA at concentrations of 200 or 500 nM for10 min. Cells were lysed and CEACAM1-L was collected from500 mg of total cell lysate proteins by immunoprecipitation withthe monoclonal CC1 antibody. Immune complexes were collectedand the proteins were separated by SDS±PAGE electrophoresis.Dried gels exposed to X-ray ®lms. 1MEA7R.1-neo cellscorrespond to cells transfected with the empty vector


222

Oncogene

protein kinase C (Newton, 1994). Consequently, we®rst investigated the e�ect of protein kinase Cinhibitors on in vivo CEACAM1-L phosphorylation.Cells were metabolically labeled for 2 h with 32P-orthophosphate. At the end of this period, eitherstaurosporine (1 or 5 mM) or the protein kinase C-speci®c inhibitor calphostin C (1 mM) (Kobayashi etal., 1989), was added to the media for 15 min prior tothe 10 min treatment with PMA (250 or 500 nM).

When CT51 transfected cells were incubated withstaurosporine, both untreated and PMA-treatedcells demonstrated a signi®cant reduction of theCEACAM1-L Ser phosphorylation levels (Figure 3a,lanes 3 and 4 versus lanes 1 and 2). However, theprotein kinase C-speci®c inhibitor calphostin C, didnot lead to reduction of the Ser phosphorylation.Similarly, the H7 and sphingosine inhibitors whichhave been shown to have inhibitory e�ects towardsprotein kinase C (Casnellie, 1991), did not a�ect eitherbasal or PMA-induced CEACAM1-L phosphorylation(data not shown). The levels of CEACAM1-Lexpressed were identical as can be judged in theimmunoblot (Figure 3b). Therefore, the protein kinaseresponsible for CEACAM1-L phosphorylation wassensitive to staurosporine and resistant to calphostin Cand other protein kinase C inhibitors.

As 13 Ser/Thr residues on the mouse CEACAM1-Lcytoplasmic domain are positioned within consensussequences for Ser/Thr phosphorylation (Kennely andKrebs, 1991; Edlund et al., 1998), we examined anumber of potential kinases known to be sensitive tostaurosporine (Gschwendt et al., 1994). There are twobinding sites for calmodulin on the CEACAM1-Lcytoplasmic domain (Edlund et al., 1996) and there-

fore, the KN-62 inhibitor of calmodulin kinase wastested. No decrease of basal or PMA-induced Serphosphorylation was noticed (Figure 3a, lanes 7 and8). Neither were changes observed when the RP-cAMP,a speci®c inhibitor of protein kinase A was used(Figure 3a, lanes 9 and 10). Similar results wereobtained with the BNL 1MEA 7R.1 mouse livercarcinoma cells stably transduced with theCEACAM1-L cDNA construct (data not shown).Therefore, CEACAM1-L phosphorylation is sensitiveto staurosporine; however, the staurosporine-sensitivekinase(s) using CEACAM1-L as a substrate remains tobe identi®ed.

Is the CEACAM1-L Ser503 residue a site ofSer phosphorylation in vivo?

Ser503 in the rat CEACAM1-L cytoplasmic domain isknown to be phosphorylated and was shown toparticipate in either bile salt extrusion or internaliza-tion of the insulin receptor (Sippel et al., 1994;Formisano et al., 1995). Hence, we investigatedwhether Ser503 was a major phosphorylation site inmouse CEACAM1-L. The Ser503 residue was con-verted to a non-phosphorylatable Ala residue and thismutant was stably transduced into CT51 cells and anumber of cell clones were isolated and selected forCEACAM1-L/Ser503 expression by FACS analysis.The level of basal or PMA-induced phosphorylationwas evaluated in the resultant cell lines (Figure 4).PMA treatment of a representative cell clone expres-sing the wild-type CEACAM1-L increased its levels ofphosphorylation by threefold (Figure 4a,b, lanes 1 and2; 4c, lanes 1 and 4). When a CT51 cell line expressingthe Ser503 mutant was treated with PMA, an increasein Ser/Thr phosphorylation levels was also observed(Figure 4a,b, lanes 3 and 4; 4c, lanes 7 and 9), althoughthis increase was generally by a factor of 1.5 ± 2-fold.This result suggests that the mouse Ser503 does notconstitute the unique site of Ser phosphorylation by thePMA-inducible kinase(s).

As previously noticed, staurosporine completelyinhibited basal Ser phosphorylation levels of the wild-type CEACAM1-L protein (Figure 4c, lane 2) and ofthe Ser503 mutant (Figure 4c, lane 8). Calphostin Cdid not inhibit CEACAM1-L phosphorylation of boththe wild-type or mutated CEACAM1-L protein (Figure4c, lanes 3 and 6, and 11 and 12). Similar results wereobtained in HEK293 transiently transfected cells (datanot shown).

In vitro phosphorylation of CEACAM1-L

As our in vivo data di�ered from the in vitro resultswith the rat CEACAM1-L (Edlund et al., 1998), weproceeded to verify whether the mouse CEACAM1-Lcytoplasmic domain could indeed be phosphorylated invitro by protein kinase C and if so, which sites werepreferentially used by this kinase. A number ofglutathione S-transferase fusion proteins expressingeither the short- or full-length, mutated or deleted

Figure 3 Staurosporine decreases CEACAM1-L phosphoryla-tion. A CEACAM1-L-transfected CT51 cell clone (2L12) wasmetabolically labeled with 32P-orthophosphate and treated withthe indicated inhibitors for 15 min as described in Materials andmethods. The cells were then treated with the phorbol ester PMAat 500 nM for 10 min, lysed and CEACAM1-L was collected from500 mg of total cell lysate proteins by immunoprecipitation withthe monoclonal CC1 antibody. Immune complexes were collected,the proteins were separated by SDS±PAGE electrophoresis,transferred to Immobilon membranes and immunoblotted withthe polyclonal 2456 anti-CEACAM1 antibody. The membraneswere subsequently exposed to X-ray ®lms and quanti®ed using aFuji BioImager. (a) autoradiogram. (b) immunoblot. Stau.,staurosporine

Oncogene


223

long CEACAM1 cytoplasmic domains (Figure 1a)were subjected to in vitro kinase assays using puri®edprotein kinase C a, b and g isoforms. As seen in Figure5a, all GST fusion proteins expressing eitherCEACAM1 cytoplasmic domain were phosphorylatedby the puri®ed protein kinase C isoforms, whereasglutathione S-transferase remained unphosphorylated.Even the shortest CEACAM1-L deletion mutant,terminating at amino acid 461 (Figure 1a, D461), wasadequately phosphorylated in these conditions, suggest-ing that minimally, the membrane-proximal site(SRKS) was recognized by protein kinase C. Tocon®rm whether other protein kinase C sites in thelong cytoplasmic domain were also used in vitro,tryptic maps of the in vitro 32P-labeled GST fusionproteins expressing the CEACAM1 mutants wereperformed. In all cases, the same pattern showingtwo 32P-labeled peptides was observed (Figure 5b). Todetermine that these two peptides were identical in allcases, they were collected from all thin layerchromatography plates, eluted from the matrix andthe resulting mix of peptides re-run on a TLC plate.This gave rise to the same chromatogram shown inFigure 5b. These results indicate that the membrane-proximal SRKS site is phosphorylated in vitro byprotein kinase C. In addition, these results con®rm thatthis site constitutes the unique in vitro protein kinase Csite in the long cytoplasmic domain of CEACAM1-L.

PMA treatment of CT51 cells leads to CEACAM1-Lrelocalization

It has previously been shown that CEACAM1-L'smembrane localization and its association with theactin cytoskeleton correlated with its tumor inhibitoryphenotype (Sadekova et al., 2000). It has also beenshown that PMA provokes diverse phenotypic e�ectson cell morphology, spreading and motility andrelocalization of E-cadherin in MDCK cells (Imamuraet al., 1998; Kamei et al., 1999). We questionedwhether PMA treatment of CT51 cells would changethe localization or solubility of CEACAM1-L or itsSer503 mutant derivative. As shown in Figure 6a, inuntreated CT51 cells expressing wild-type CEACAM1-L, the protein was found at the cell ± cell contacts(Figure 6a, panel 1). We have previously demonstratedthat it is then in tight association with cortical actin

Figure 4 Response of the Ser503 CEACAM1-L mutant to PMAstimulation and inhibitors. Wild-type (2L12) or Ser503-mutated(Ser66) CEACAM1-L-transfected CT51 cell clones were metabo-lically labeled with 32P-orthophosphate. The cells were thentreated with the phorbol ester PMA at 500 nM for 10 min, lysedand CEACAM1-L was collected from 500 mg of total cell lysateproteins by immunoprecipitation with the monoclonal CC1antibody. Immune complexes were collected, the proteins wereseparated by SDS±PAGE electrophoresis and transferred toImmobilon membranes. The membranes were subsequentlyexposed to X-ray ®lms and the protein bands were quanti®edusing a Fuji BioImager. (a) Autoradiogram. (b) Immunoblot. (c)Metabolically 32P-labeled cells were ®rst treated with the indicatedinhibitors for 15 min as described in Materials and methods, priorto treatment with PMA. Autoradiogram of immunoprecipitatedproteins. Stauro., staurosporine

Figure 5 In vitro PKC phosphorylation. (a) Samples of 10 ng ofthe puri®ed PKC a, b and g isoforms were used in in vitro kinaseassays to phosphorylate 100 ng of the short or long CEACAM1-L cytoplasmic domains or point and deletion mutants (identity inFigure 1a) produced as GST fusion proteins in bacteria asdescribed in Materials and methods. Resulting proteins wereseparated on 10% SDS±PAGE gels and either exposed directlyto X-ray ®lms (top) or immunoblotted with an anti-GST antibody(bottom). (b) Labeled proteins were extracted from the gels andsubjected to trypsin cleavage. Resulting peptides were separatedon TLC plates in two-dimensions according to van der Geer etal., 1994


224

Oncogene

Figure 6 Treatment of cells with PMA leads to CEACAM1-L relocalization. (a) CT51 cells transfected with either the wild-type(panels 1 ± 6) or Ser503 mutant (panels 7 and 8) were either untreated (panels 1, 2 and 7) or treated with PMA (panels 3 ± 6 and 8) for15 min. Cells were also treated with the staurosporine (stauro.) or calphostin C inhibitors for 15 min prior to addition of PMA (panels 5and 6). The cells were then processed for immuno¯uorescence as described in Materials and methods. PMA led to the relocalization ofthe CEACAM1-L protein from the cell contacts. This was inhibited by staurosporine, but not by calphostin C. (b) Confocal sectioningwas performed on CEACAM1-L-transfected CT51 cells to examine localization of CEACAM1-L after PMA treatment. The protein ispresent in a ring-like structure at the top of the cells. (c) CEACAM1-L solubility after PMA treatment. A wild-type CEACAM1-L-expressing clone (2L12) was either untreated (lanes 1 and 2) or treated for a short (lanes 3 and 4) or prolonged (lanes 5 and 6) period with500 nM of PMA. Cells were collected in a 1% Triton X-100-containing bu�er as previously described (Sadekova et al., 2000).CEACAM1-L was collected by immunoprecipitation with the CC1 Ab from the soluble and insoluble fractions. The proteins wereelectrophoresed on 8% SDS±PAGE gels and revealed using an anti-CEACAM1 polyclonal Ab

Oncogene


225

(Sadekova et al., 2000) (Figure 6a, panel 2). Theexpression of CEACAM1-L at cell ± cell contacts wascon®rmed by confocal Z-sectioning through a repre-sentative cell cluster (Figure 6b, panels 1 ± 6). Incomparison, no signi®cant di�erences in cellularlocalization of the Ser503 mutant protein were noticedrelative to that of the wild-type CEACAM1-L (Figure6a, panels 7 and 8 compared to panels 1 and 3).However, treatment of CT51 transfected cells withPMA for 15 min led to a dramatic ¯attening andspreading of the cells (Figure 6a, panel 3). In addition,CEACAM1-L was no longer present at cell contacts,but appeared di�use as speckles condensed in aperinuclear ring as seen by confocal Z-sectioning(Figure 6b, panels 7 ± 12: magni®cation identical tothat of panels 1 ± 6). Importantly, the actin cytoskele-ton was also a�ected by the PMA treatment asphalloidin staining was co-incident with that ofCEACAM1-L (Figure 6a, panel 4). To determinewhether CEACAM1-L was localized intracellularly,PMA-treated CEACAM1-L-expressing CT51 cells wereimmunostained before or after permeabilization of thecells. With or without cell permeabilization,CEACAM1-L was detected in the ring-like structureshown in Figure 6b. This result suggests thatCEACAM1-L remains at the cell surface but reloca-lizes from the cell ± cell contacts to a ring-like structurecondensed at the top of the cell (Figure 6b, panels7 ± 12). Staurosporine completely abolished the e�ect ofPMA on cells expressing either wild-type (Figure 6a,panel 5) or Ser503 mutant CEACAM1-L (data notshown), whereas calphostin C treatment, prior toaddition of PMA, did not alter the cell spreading orCEACAM1-L relocalization mediated by PMA (Figure6a, panel 6). Therefore, these results show that PMAtreatment of mouse colon tumor cells expressingCEACAM1-L induced cell spreading and disappear-ance of CEACAM1-L from cell contacts.

We have previously shown that CEACAM1-L istightly associated with the actin cytoskeleton and wasfound in a 1% Triton X-100 insoluble fraction(Sadekova et al., 2000). We then questioned whetherthe relocalization of CEACAM1-L changed its associa-tion to the cytoskeleton. As shown in Figure 6c, short(lanes 3 and 4) or prolonged (lanes 5 and 6) PMAtreatment of cells did not modify the CEACAM1-Lsolubility, which remained in the 1% Triton X-100insoluble fraction.

Discussion

CEACAM1 is an important member of the carcinoem-bryonic antigen (CEA) family (Beauchemin et al.,1999). CEACAM1 plays a role in the development ofmany types of cancers. Its expression is dramaticallyreduced in the early phases of intestinal cancers(Neumaier et al., 1993; Rosenberg et al., 1993; Nollauet al., 1997; Ilantzis et al., 1997). It is also down-regulated in prostate cancer and is absent fromhyperproliferative prostate tissue (Hsieh et al., 1995).

Furthermore, CEACAM1's expression is absent in liverand endometrial carcinomas and in 30% of breastcancers (Tanaka et al., 1997; Bamberger et al., 1998;Huang et al., 1998). CEACAM1 expresses twocytoplasmic variants; a short 10 amino acid cytoplas-mic domain and a longer variant of 71 ± 73 amino acids(Beauchemin et al., 1998). We and others havedemonstrated that expression in tumorigenic cells ofthe longer CEACAM1 variant resulted in in vivogrowth inhibition of colon, prostate and breast cancers(Kunath et al., 1995; Hsieh et al., 1995; Luo et al.,1997). This phenotype was valid providing that theexpression levels of the CEACAM1-L protein werecomparable to that found in normal tissue (Turbide etal., 1997). We have also shown that the dynamics ofCEACAM1-L Tyr phosphorylation and dephosphor-ylation are crucial for inhibition of tumor cell growthsince mutation of Tyr488 resulted in tumor develop-ment (Izzi et al., 1999). In addition, deletion of thethree terminal Lys residues, involved in the regulationof CEACAM1-L Tyr phosphorylation (Huber et al.,1999), produced a similar phenotype (Izzi et al., 1999).This suggested that either intramolecular connectionsbetween di�erent regions of the cytoplasmic domainwere necessary for e�cient folding of the protein orthat intermolecular associations with di�erent keyplayers in signaling cascades depended on the presenceand phosphorylation of the Tyr488 residue.

There are 17 Ser/Thr in the cytoplasmic domain,many of these residues being located in consensus sitesfor a variety of Ser/Thr protein kinases. Ser503 hasbeen shown to play a crucial role in CEACAM1-L-mediated processes such as insulin receptor internaliza-tion (Formisano et al., 1995) or bile salt extrusion(Sippel et al., 1994). For this reason, we focused ourinvestigation on the role played by Ser503 inCEACAM1-L-mediated tumor growth inhibition. Ourresults demonstrate that mutation of the Ser503 residueto an Ala a�ected the in vivo inhibition of tumordevelopment usually observed with cells expressing thewild-type CEACAM1-L.

We then considered potential mechanisms involvedin the Ser503-directed reversal of tumor growthinhibition and ®rst focused on Ser phosphorylation.CEACAM1-L is basally phosphorylated on Ser/Thrresidues in epithelial cells. Similar results had beenobtained in ®broblast and epithelial models (Odin etal., 1986; McEntire et al., 1989; Culic et al., 1992;Sippel et al., 1994; Najjar et al., 1995). We demon-strated that phorbol ester-mediated activation ofkinase(s) led to increased phosphorylation of theCEACAM1-L protein. As PMA is a known activatorof protein kinase C (Newton, 1994), we examinedwhether this kinase was indeed responsible for thephosphorylation of the long cytoplasmic domain. Ourresults with the PKC inhibitors calphostin C, H7 andsphingosine demonstrate that PKC had no obviouse�ect on in vivo phosphorylation of CEACAM1-Lexpressed in mouse cells. This was in spite of the factthat a CEACAM1-L membrane-proximal PKC site isphosphorylated in vitro by puri®ed isoforms of PKC.


226

Oncogene

In vivo, however, di�erences in overall Ser phosphor-ylation levels might not be apparent when so manysites are available for phosphorylation. On the otherhand, this discrepancy between the in vitro and in vivophosphorylation data might also result from theunavailability or masking of the PKC membrane-proximal site. Instead of PKC, it is likely that di�erentstaurosporine-sensitive Ser/Thr kinases, other thancalmodulin kinase and protein kinase A, are likelye�ective. In vivo experiments revealed that bothCEACAM1-L and its Ser503 mutant responded toPMA treatment. Proteolytic maps of the in vivo 32P-labeled wild-type and Ser mutant CEACAM1-Lproteins will be needed to ascertain which sites aretargets for the various kinases. Hence, the preciseCEACAM1-L Ser phosphorylation site(s) or speci®ckinase(s) provoking PMA-mediated increasedCEACAM1-L phosphorylation remain to be identi®ed.

Interestingly, both Tyr488 and Ser503 are needed forinsulin receptor internalization (Formisano et al., 1995).In this context, mitogenic responses to insulin wereenhanced in cells transfected with either the Tyr488 orSer503 mutants. This is in agreement with the in vivotumor assays with CT51 cells expressing either of thesemutants (Izzi et al., 1999 and this study) and it appearsthat a delicate balance is established between these tworesidues to regulate tumor cell growth. One model toexplain our tumor inhibition data would be that theconformation of the cytoplasmic domain might varydepending on its binding to a number of proteincomplexes. Steric hindrance may need to be consideredi.e. occupancy of one site by a protein complex mightrestrict accessibility of the other site to its own signalingcomplex. We have shown that Tyr488 is required forbinding of the Tyr phosphatases SHP-1 and SHP-2(Beauchemin et al., 1997; Huber et al., 1999). It has alsobeen shown for instance that the 35 distal amino acids(including Ser503) of the rat CEACAM1-L cytoplasmicdomain associate with an 80 kDa protein (CAP-80).Ser503 may be crucial for the binding of this proteinsince a deletion mutant truncating the protein atLys499, thereby eliminating Ser503, no longer had thecapability of binding the CAP-80 protein. Although itsidentity is still unknown, CAP-80 appears important inthe CEACAM1-L Ser503-dependent growth suppres-sive e�ects (Luo et al., 1998). In addition, tightassociation of CEACAM1-L with the actin cytoskele-ton is mediated by the same 35 distal amino acids(Sadekova et al., 2000).

A second potential mechanism to explain theinvolvement of Ser503 in the CEACAM1-L-mediatedtumor suppressive phenotype is the CEACAM1-Lcellular localization. We have not seen any di�erencesin localization between the wild-type and the Ser503-expressing CEACAM1-L. However, the localization ofCEACAM1-L is not static. In normal colon tissue,CEACAM1 is generally concentrated at the luminalglycocalyx and fuzzy coat of epithelial cells (FraÈ ngs-myr et al., 1995). The protein is also localized tothe lateral borders of enterocytes (Hansson et al.,1989). Additionally, in serum-starved ®broblasts,

CEACAM1-L is predominantly found in the cyto-plasm (Sadekova et al., 2000). In these cells, itundergoes translocation to the cell membrane whereit engages in cell adhesion upon activation of Rac1,RhoA or Cdc42. This process requires association ofCEACAM1-L with polymerized actin and reorganiza-tion of the actin cytoskeleton provoked relocalizationof CEACAM1-L (Sadekova et al., 2000). We haverecently observed similar di�erences in localization ofCEACAM1 depending on activation of the Rho-likeGTPases in MDCK cells (FourneÁ s et al., in prepara-tion). As shown in Figure 6, CT51 CEACAM1-L-transfected cells also undergo cell spreading andrelocalization of CEACAM1-L in response to PMAtreatment. In these conditions, CEACAM1-L is tightlyassociated with cortical actin and its relocalization iscorrelated with that of the actin cytoskeleton. AsPMA is also known to provoke a number of cellularmodi®cations (Newton, 1994), there remains a possi-bility that the CEACAM1-L-associated phenotypesmay be due to a bystander e�ect. Not much is knownabout directed targeting of cell adhesion molecules.Recent data on co-localization of E-cadherin and thec-Met receptor have indicated that PMA treatment ofMDCK cells caused cell spreading leading to theconcomitant disappearance of these proteins from thesites of cell ± cell adhesion. After a prolonged PMAtreatment, they are found within the late endosomes(Kamei et al., 1999). This process is linked tocytoskeletal reorganizations via activation or inhibi-tion of a number of Rho-like GTPases (Imamura etal., 1998; Kamei et al., 1999). Thus, a dynamic tra�cestablished between cytoplasmic or cell surfacelocalization of CEACAM1-L is becoming moreapparent than had previously been observed. WhetherPMA treatment ®rst a�ects the CEACAM1-L phos-phorylation or whether it directly or indirectlyin¯uences cytoskeletal plasticity resulting in alterationsin cell shape will require further investigations.CEACAM1's dynamic transport to and from the cellmembrane with disruption and reorganization ofcytoskeletal components will also need to be con-sidered.

Materials and methods

Cell culture

Human embryonic kidney (HEK293) cells were obtainedfrom the American Tissue Culture Collection (ATCCCRL1573) and grown in a-minimal essential mediumsupplemented with 2 mM glutamine and 10% heat-inacti-vated fetal bovine serum at 378C in a humidi®ed atmosphereof 5% CO2. HEK293 cells (56105 cells per 100 mm dish)were transfected with 5 mg/100 mm dish of the pXM139-CEACAM1-L-expressing plasmid DNA using calciumphosphate co-precipitation (Southern and Berg, 1982).Growth of CT51 mouse colon carcinoma cells, generouslyprovided by Dr Michael Brattain (Medical College of Ohio,Toledo, Ohio, USA) and insertion of the CEACAM1-L ormutant cDNA constructs by retroviral-mediated infection

Oncogene


227

have previously been described (Kunath et al., 1995). Cellclones were manually selected from G418-resistant popula-tions (G418: 1.5 mg/ml) and CEACAM1-L-positive cloneswere identi®ed by FACS analyses and immunoblotting aspreviously described (Kunath et al., 1995). The mouse livercarcinoma cell line, BNL1MEA.7R.1 (ATCC: TIB 75), is amethylcholanthracene epoxide-transformed mouse liver cellline derived from the normal BALB/c embryonic liver cellline BNL CL.2 (ATCC: TIB 73). This cell line wastransfected with the CEACAM1-L cDNA construct byretroviral-mediated transductions using a protocol similarto that reported in Kunath et al. (1995). Metabolic labelingof HEK293, CT51 cells or BNL 1MEA cells was performedfor 2 h with 200 mCi of 32P-orthophosphate (NEN Dupont)/100 mm dishes in 2 ml of phosphate-free medium containing2.0% dialysed FBS (Gibco). PMA, or inhibitor treatments(all purchased from Sigma) were performed at respectiveconcentrations of: 25 ± 1000 nM of PMA for 10 min, 1 or5 mM of staurosporine, 1 mM of calphostin C, 73 mM of H7,10 mM of sphingosine, respectively for 15 min. Cells werealso treated with cell permeable protein kinase inhibitors ofcalmodulin kinase (KN-62, 2 mM for 15 min) and of proteinkinase A (RP-cAMP, 150 mM for 15 min). After treatment,cells were scraped from the dishes in the presence of lysisbu�er (600 ml/100 mm) containing 100 mM Tris-HCl(pH 8.0), 10 mM EDTA, 2% NP-40, 20 mg/ml of proteaseinhibitors (leupeptin, aprotinin, phenylmethylsulfonyl ¯uor-ide, N-a-p-tosyl-1-lysine chloromethyl ketone, N-tosyl-1-phenylalanine chloromethyl ketone), 100 mM sodium ¯uor-ide and 2 mM sodium orthovanadate. After centrifugation(1500 r.p.m.) for 5 min at 48C to remove cellular debris,protein content of cell lysates was measured using the BCAprotein assay kit (Pierce Chemicals). Three separateexperiments were performed using at least two cell clonesof CT51- and MEA-transduced cells.

Site-directed mutagenesis and preparation of bacterial fusionproteins

Deletion and point mutants used in this study have beenpreviously described (Huber et al., 1999). The CEACAM1mutated constructs were reintroduced into the pLXSN orpXM139 vectors used in stable or transient transfectionassays. The cytoplasmic domains of CEACAM1 mutantswere ampli®ed by PCR and cloned in frame into thepGEX2T vector. All mutated constructs were subjected toDNA sequence analysis to con®rm their integrity. Fusionprotein constructs were introduced into E. coli BL21 bacteriaand induction of expression and puri®cation of the GSTproteins were performed according to standard procedures(Beauchemin et al., 1997).

In vitro phosphorylation reactions

In vitro phosphorylation was performed by incubating 100 ngof the CEACAM1 cytoplasmic domain GST fusion proteinswith 10 ng of a mixture of puri®ed a, b, and g proteinkinase C isoforms (Calbiochem) in 20 ml of bu�er containing20 mM Tris-HCl, pH 7.5, 0.1 mM CaCl2, 5 mM MgCl2,0.03% Triton X-100 and 1 ml of g-labeled 32P-ATP(6000 Ci/mmole, Amersham Corp.) for 30 min at 378C.Reactions were terminated by addition of 2 ml of loadingdye bu�er and boiling for 10 min. The labeled proteins wereseparated on 10% polyacrylamide gels and subjected toautoradiography. The amount of GST fusion proteins used ineach reaction was controlled by immunoblotting with an anti-GST antibody (Pharmacia).

Tryptic maps of GST fusion-CEACAM1

GST fusion proteins were labeled as described above andsubjected to trypsin cleavage according to van der Geer et al.(1994). Digested peptides were spotted on thin layerchromatography plates, and electrophoresed in the ®rstdimension in a pH 1.9 bu�er. After drying, the peptideswere separated in the second dimension by chromatographyas previously described (van der Geer et al., 1994). Driedplates were then exposed to BioMax ®lms (Kodak).

Immunoprecipitation and immunoblotting

A polyclonal anti-mouse CEACAM1 antibody (Ab 231 or2456), speci®c to the ®rst extracytoplasmic Ig domains ofCEACAM1 (Daniels et al., 1996) or a monoclonal Ab (CC1)speci®c to the ®rst Ig domain of this protein (Dveksler et al.,1993), have previously been described. Total cell lysateproteins (500 mg ± 1 mg) were immunoprecipitated with in-dicated antibodies at 48C for 2 h and immune complexeswere adsorbed onto protein A or G Sepharose beads.Proteins were resolved on 8% SDS±PAGE gels. Forimmunoblotting, proteins were transferred to Immobilonmembranes. After blocking unspeci®c sites for 30 min at208C with 5% milk/TBS, membranes were incubated with a1 : 1000 dilution of anti-CEACAM1 antibodies. Immunecomplexes were visualized by incubating the membranes withan HRP-coupled anti-rabbit antibody and ECL detection(Amersham). Alternatively, 32P-labeled proteins were exposedto X-ray ®lms.

Solubility assays

To determine whether CEACAM1-L was associated with thesoluble or insoluble fractions, cells were treated as previouslydescribed (Sadekova et al., 2000).

Immunofluorescence

Immuno¯uorescence detection was done essentially asdescribed in Lamarche et al. (1996). In brief, cells wererinsed in PBS, ®xed in 1.8% paraformaldehyde for 10 minand permeabilized in 0.2% Triton X-100/PBS for 5 min.Coverslips were incubated in 10% BSA for 45 min, thenfurther incubated for 2 h with the anti-CEACAM1 primaryantibody (Ab 2456, 1 : 800 dilution) diluted in PBS/5% BSA,washed with PBS/2% Triton X-100 and transferred to theappropriate secondary antibody (Cy3-conjugated goat anti-rabbit, 1 : 300 dilution) for 1 h. Coverslips were mounted withmoviol containing p-phenylenediamine (1 mg/ml). The cover-slips were examined with either a Zeiss Axiophot ¯uorescencemicroscope or a BioRad confocal microscope.

Quantification

Quanti®cation of the radioactive bands was performed usinga Fuji BioAnalyzing System 2000.

Tumor proliferation assays

In vivo tumor development was monitored for a period of42 ± 56 days after injection of 46106 CT51 cell populationsinto the ¯ank of 6-week-old BALB/c female mice. Ten micewere used per conditions tested. The weight and dimensionsof the tumors were recorded. Statistical analyses wereperformed as previously described (Kunath et al., 1995).


228

Oncogene

AbbreviationsN-CAM, neural cell adhesion molecule; CEA, carcinoem-bryonic antigen; Bgp or BGP, biliary glycoprotein;CEACAM1; carcinoembryonic antigen cell adhesion mole-cule 1; TPA or PMA, phorbol 12-myristate 13-acetate;PKC, protein kinase C; MDCK, Mardin-Darby caninekidney cells; VEGF, vascular endothelial growth factor;PCR, polymerase chain reaction.

AcknowledgmentsWe are greatly indebted to Philippe Grondin for prelimin-ary experiments, to Anne-Claude Gingras (McGill Uni-versity) and Dr AndreÂ Veillette (Institut de RecherchesCliniques de MontreÂ al) for advice during the course of thiswork and critical reading of the manuscript. We also thankDr Kathryn V Holmes (Department of Microbiology,University of Colorado) for the CC1 antibody. Thisresearch was supported by the Cancer Research SocietyInc. B FourneÁ s is funded by the `Association pour laRecherche sur le Cancer', N Beauchemin is supported by a`Chercheur National' program from the `Fonds de laRecherche en SanteÂ du QueÂ bec'.

References

Bamberger AM, Riethdorf L, Nollau P, Naumann M,Erdmann I, GoÈ tze J, BruÈ mmer J, Schulte HM, WagenerC and LoÈ ning T. (1998). Am. J. Pathol., 152, 1401 ± 1406.

Barnett T, Kretschmer A, Austen DA, Goebel SJ, Hart JT,Elting JJ and Kamarck ME. (1989). J. Cell. Biol., 108,267 ± 276.

Beauchemin N, Kumath T, Robitaille J, Chow B, Turbide C,Daniels E and Veillette A. (1997). Oncogene, 14, 783 ± 790.

Beauchemin N and Lin SH. (1998). In Cell Adhesion andCommunication: Role of C-CAM as a tumour suppressor.Stanners CP (ed). Harwood Academic Publishers, Am-sterdam, pp. 155 ± 175.

Beauchemin N, Chen T, Draber P, Dveksler G, Gold P,Gray-Owen S, Grunert F, HammarstroÈ m S, Holmes KV,Karlson A, Kuroki M, Lin SH, Lucka L, Najjar SM,Neumaier M, OÈ brink B, Shively JE, Skubitz KM, StannersCP, Thomas P, Thompson JA, Virji M, von Kleist S,Wagener C, Watt S and Zimmermann W. (1999). Exp.Cell. Res., 252, 243 ± 249.

BruÈ mmer J, Neumaier M, GoÈ pfert C and Wagener C. (1995).Oncogene, 11, 1649 ± 1655.

Casnellie JE. (1991). Adv. Pharmacol., 22, 167 ± 205.Culic O, Huang QH, Flanagan D, Hixson D and Lin SH.

(1992). Biochem. J., 285, 47 ± 53.Daniels E, LeÂ tourneau S, Turbide C, Kuprina N, Rudins-

kaya T, Yazova AC, Holmes KV, Dveksler GS andBeauchemin N. (1996). Dev. Dynamics, 206, 272 ± 290.

Dveksler GS, Pensiero MN, Cardellichio CB, Williams RK,Jiang GS, Holmes KV and Die�enbach CW. (1991). J.Virol., 65, 6881 ± 6891.

Dveksler GS, Pensiero MN, Die�enbach CW, CardellichioCB, Basile AA, Elia PE and Holmes KV. (1993). Proc.Natl. Acad. Sci. USA, 90, 1716 ± 1720.

Edlund M, Blikstad I and OÈ brink B. (1996). J. Biol. Chem.,271, 1393 ± 1399.

Edlund M, Wisktrom K, Toomik R, Ek P and Obrink B.(1998). FEBS Lett., 425, 166 ± 170.

Formisano P, Najjar SM, Gross CN, Philippe N, Oriente F,Kern-Buell CL, Accili D and Gorden P. (1995). J. Biol.Chem., 270, 24073 ± 24077.

FraÈ ngsmyr I, Baranov V, Prall F, Yeung MM-W, Wagener Cand HammarstroÈ m S. (1995). Cancer Res., 55, 2963 ± 2967.

Gray-Owen SD, Dehio C, Haude A, Grunert F and MeyerTF. (1997). EMBO J., 16, 3435 ± 3445.

Gschwendt M, Kittstein W and Marks F. (1994). FEBSLett., 338, 85 ± 88.

Gumbiner BM. (2000). J. Cell. Biol., 148, 399 ± 404.Hansson M, Blikstad I and OÈ brink B. (1989). Exp. Cell. Res.,

181, 63 ± 74.

Hauck CR, Meyer TF, Lang F and Gulbins E. (1998).EMBO J., 17, 443 ± 454.

Hsieh J-T, Luo W, Song W, Wang Y, Kleinerman DI, VanNT and Lin S-H. (1995). Cancer Res., 55, 190 ± 197.

Huang J, Simpson JF, Glackin C, Riethorf L, Wagener Cand Shively JE. (1998). Anticancer Res., 18, 3203 ± 3212.

Huber M, Izzi L, Grondin P, Houde C, Kunath T, Veillette Aand Beauchemin N. (1999). J. Biol. Chem., 274, 335 ± 344.

Hunter T, (2000). Cell, 100, 113 ± 127.Ilantzis C, Jothy S, Alpert LC, Draber P and Stanners CP.

(1997). Lab. Invest., 76, 703 ± 716.Izzi L, Turbide C, Houde C, Kunath T and Beauchemin N.

(1999). Oncogene, 18, 5563 ± 5572.Imamura H, Takaishi K, Nakano K, Kodama A, Oishi H,

Shiozaki H, Monden M, Sasaki T and Takai Y. (1998).Mol. Biol. Cell., 9, 2561 ± 2575.

Kamei T, Matozaki T, Sakisaka T, Kodama A, YokohamaS, Peng YF, Nakano K, Takaishi K and Takai Y. (1999).Oncogene, 18, 6776 ± 6784.

Kennely PJ and Krebs EW. (1991). J. Biol. Chem., 266,15555 ± 15558.

Kinzler KW and Vogelstein B. (1996). Cell, 87, 159 ± 170.Kleinerman DI, Troncoso P, Lin SH, Pisters LL, Sherwood

ER, Brooks T, von Eschenbach AC and Hsieh JT. (1995).Cancer Res., 55, 1215 ± 1220.

Kobayashi E, Nakano H, Morimoto M and Tamaoki T.(1989). Biochem. Biophys. Res. Comm., 159, 548 ± 553.

Kunath T, OrdonÄ ez-Garcia C, Turbide C and BeaucheminN. (1995). Oncogene, 11, 2375 ± 2382.

Lamarche N, Tapon N, Stowers L, Burbelo PD, AspenstroÈ mP, Bridges T, Chant J and Hall A. (1996). Cell, 87, 519 ±529.

Lin SH and Guidotti G. (1989). J. Biol. Chem., 264, 14408 ±14414.

Luo W, Wood CG, Early K, Hung MC and Lin SH. (1997).Oncogene, 14, 1697 ± 1704.

Luo W, Early K, Tantingco V, Hixson DC, Liang TC andLin SH. (1998). Oncogene, 16, 1141 ± 1147.

McCuaig K, Turbide C and Beauchemin N. (1992). Cell,Growth Di�., 3, 165 ± 174.

McEntire KD, Mowery J and Hixson DC. (1989). CancerRes., 49, 6795 ± 6802.

Najjar SM, Accili D, Philippe N, Jernberg J, Margolis R andTaylor SI. (1993). J. Biol. Chem., 268, 1201 ± 1206.

Najjar SM, Philippe N, Suzuki Y, Ignacio GA, Formisano P,Accili D and Taylor SI. (1995). Biochemistry, 34, 9341 ±9349.

NeÂ dellec P, Turbide C and Beauchemin N. (1995). Eur. J.Biochem., 231, 104 ± 114.

Oncogene


229

Neumaier M, Paululat S, Chan A, Matthaes P and WagenerC. (1993). Proc. Natl. Acad. Sci. USA, 90, 10744 ± 10748.

Newton AC. (1994). J. Biol. Chem., 270, 28495 ± 28498.Nollau P, Scheller H, Kona-Horstmann M, Rohde S,

Hagenmuller F, Wagener C and Neumaier M. (1997).Cancer Res., 57, 2354 ± 2357.

Ocklind C and OÈ brink B. (1982). J. Biol. Chem., 257, 6788 ±6795.

Odin P, TingstroÈ m A and OÈ brink B. (1986). Biochem. J., 236,559 ± 568.

Rees-Jones RW and Taylor SI. (1985). J. Biol. Chem., 260,4461 ± 4467.

Rojas M, Fuks A and Stanners CP. (1990). Cell Growth Di�.,1, 527 ± 533.

Rojas M, DeMarte L, Screaton RA and Stanners CP. (1996).Cell Growth Di�., 7, 655 ± 662.

Rosenberg M, NeÂ dellec P, Jothy S, Fleiszer D, Turbide Cand Beauchemin N. (1993). Cancer Res., 53, 4938 ± 4945.

Sadekova S, Lamarche-Vane N, Xiaodong L and Beauche-min N. (2000). Mol. Biol. Cell., 11, 65 ± 77.

Sippel CJ, Fallon RJ and Perlmutter DH. (1994). J. Biol.Chem., 269, 19539 ± 19545.

Southern PJ and Berg P. (1982). J. Mol. Appl. Genet., 1,314 ± 327.

Tanaka K, Hinoda Y, Takahashi H, Sakamoto H, NakajimaY and Imai K. (1997). Int. J. Cancer, 74, 15 ± 19.

Turbide C, Kunath T, Daniels E and Beauchemin N. (1997).Cancer Res., 57, 2781 ± 2788.

van der Geer P, Luo K, Sefton BM and Hunter T. (1994). InCell Biology: A Laboratory Handbook: PhosphopeptideMapping and Phosphoamino Acid Analysis on CelluloseThin-Layer Plates. Academic Press, pp. 422 ± 447.

Virji M, Makepeace K, Ferguson DJP and Watt SM. (1996).Mol. Microbiol., 22, 941 ± 950.


230

Oncogene

the ceacam1-l ser503 residue is crucial for inhibition of colon cancer cell tumorigenicity

Documents