[CANCER RESEARCH 51. 3657-3662. July 15. 1991]

Transduction and Expression of the Human Carcinoembryonic Antigen Gene in aMurine Colon Carcinoma Cell LinePaul F. Robbins,1 Judith A. Kantor, Michael Salgaller, Patricia H. Hand, Philip D. Fernsten, and Jeffrey Schlom

Laboratory of Tumor Immunology and Biology, \alional Cancer Institute, .\IH, Bethesda, Maryland 20892

ABSTRACT

A cell line derived from the mouse colon adenocarcinoma, MC-38, hasbeen transduced with a retroviral construct containing complementaryDNA encoding the human Carcinoembryonic antigen (CEA) gene. MC-38, which forms tumors in syngeneic C57BL/6 mice, has been extensivelystudied as a target for active immunotherapy. Individual transducedclones that express high levels of cell surface CEA were isolated, andtwo clones, termed MC-38-ceal and MC-38-cea2, were extensively characterized. The levels of CEA found on the surface of these clones wereconsiderably higher than that found in a moderately differentiated humancolon carcinoma cell line (VViDr)and were comparable to those found onthe human colon carcinoma cell lines GEO and CBS (among the highestCEA-expressing cells reported). Further analysis demonstrated that theCEA expressed in the MC-38-ceal clone had a similar molecular weightto native CEA (M, 180,000), but the MC-38-cea2 cell line expressed asingle M, 70,000 glycosylated immunoreactive product. Seven anti-CEAmonoclonal antibodies were found to react with both clones. The CEAgene present in the MC-38-cea2 clone was partially sequenced and wasfound to contain a deletion of two of the three repeated domains presentin CEA. These results provide a basis for future studies to map imiiiu-nodominant epitopes of CEA and to develop a syngeneic model systemthat may aid in the design of reagents and protocols to study active andpassive immunotherapy directed against a carcinoma expressing humanCEA.

INTRODUCTION

TAAs2 have been used as targets for tumor diagnosis and

therapy. The ability of conjugated and unconjugated MAbs totarget and kill tumors in vivo has been evaluated primarily inmodel systems using human tumor xenografts in athymic mice.These experiments have led to clinical trials of tumor targetingwith MAbs, as well as clinical trials testing the efficacy ofunconjugated and conjugated MAbs as therapeutic agents.

TAAs also represent a potentially important target for activeimmunotherapy. For example, a system has recently been developed to evaluate the efficacy of active immunotherapy directed against the human melanoma antigen p97 (1). Initially,a cDNA clone encoding the p97 antigen was transfected intothe K1735 mouse melanoma cell line, which forms pulmonarymétastaseswhen injected i.v. into syngeneic C3H/HeN mice.The p97 cDNA clone was then introduced into vaccinia virus,and infection of Chinese hamster ovary cells with this recombinant was found to result in the expression of high levels ofp97 on the cell surface (2). Mice were found to be protectedfrom challenge with melanoma cells expressing p97 followingimmunization with the p97 vaccinia recombinant (1). Phase I

Received 1/29/91; accepted 4/29/91.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 L'.S.C. Section 1734 solely to indicate this fact.

' To whom requests for reprints should be addressed, at Laboratory of Tumor

Immunology and Biology. National Cancer Institute. N1H. Building 10. Room5B46, Bethesda, MD 20892.

2The abbreviations used are: TAA. tumor-associated antigen: CEA. Carcinoembryonic antigen; MAb. monoclonal antibody; LTR. long-terminal repeat: CMV,cytomegalovirus; NCA. nonspecific cross-reactive antigen: cDNA. complementary DNA; p97, M, 97,000 protein; IgG, immunoglobulin G: IgM, immunoglob-ulin M; i.s., intrasplenic; PVDF, polyvinylidene difluoride. PCR. polymerasechain reaction.

clinical trials are now being done, performed with the recombinant p97 vaccinia recombinant.

A potential target for active immunotherapy of gastrointestinal carcinomas is CEA, which is a M, 180,000 glycoproteinexpressed at high levels on the surface of nearly all tumors ofthe gastrointestinal tract (3). Model studies involving humanCEA as a target of active immunotherapy cannot be carriedout, however, since CEA is not expressed in rodent tumors,nor, to our knowledge, in any other animal model. In order todevelop an active anti-CEA tumor therapy model system, wehave transduced the MC-38 mouse colon adenocarcinoma cellline with CEA. In syngeneic C57BL/6 mice, i.v. injection ofMC-38 results in pulmonary tumors (4), and i.s. injectionresults in hepatic tumors (5). The MC-38 tumor has previouslybeen used in immunotherapy studies demonstrating the efficacyof lymphokine-activated killer cells in suppressing tumor cellgrowth (4). Tumor-infiltrating lymphocytes isolated from MC-38 tumors were effective at reducing the pulmonary microme-tastases (6) as well as advanced pulmonary and hepatic métastases (7).

In this study, we present the development and analysis of twoclones of transduced MC-38 cells that express high levels ofhuman CEA on their cell surface. One of the MC-38 cloneswas found to contain a truncated CEA gene product thatexpressed a number of distinct CEA epitopes. The nature ofthe mutation resulting in the altered glycoprotein was examinedby isolating and sequencing a portion of the CEA gene presentin this clone.

MATERIALS AND METHODS

Cell Lines. The MC-38 B/6 mouse colonie adenocarcinoma cell line(8) was obtained from Dr. Bernard Fox in the laboratory of Dr. StevenRosenberg (National Cancer Institute, NIH, Bethesda, MD). This cellline was established by continuous in vitro passage of MC-38 tumorcells (5). The ecotropic retroviral packaging cell line i/-2 (9) and theamphotropic packaging cell line PA317 (10) were obtained from Dr.Robert Bassin (National Cancer Institute, NIH, Bethesda, MD). Thehighly differentiated human colon carcinoma cell lines GEO and CBS,as well as the moderately differentiated human colon carcinoma cellline WiDr, have been previously described (11). All cell lines weremaintained in Dulbecco's modified Eagle's medium supplemented with

2 HIMglutamine, 0.1 HIMnonessential amino acids, O.I DIMsodiumpyruvate, and 10% newborn calf serum.

Isolation and Partial Sequencing of CEA cDNA. A cDNA library wasprepared in the A-orf8 cloning vector (12) from polyadenylate-contain-ing mRNA isolated from the human colon tumor cell line GEO, whichproduces high levels of CEA. A 534-base pair Pst\ fragment of the CEAcDNA clone pCEAl (13), kindly supplied by Dr. Jack Shively (City ofHope, Duarte, CA), was labeled with "P using the random primer

labeling method. Fifteen clones that hybridized to the Pstl pCEAlfragment were then screened with an oligomer complementary to aportion of the CEA leader region (5'-CAGGGGATGCAC-CATCTGTGGGGAGGGGCCGAG-3') (PR1), as well as an oligomer

complementary to a portion of the C-terminal hydrophobic regionofCEA(5'-AACCAGCACTCCAATCATGATGCCGACAGT-3')

(PR2). The oligomer PR2 hybridized to 8 of 15 clones, and PR1hybridized with 5 clones, but none of the clones hybridized with both

3657

Research. on September 7, 2020. © 1991 American Association for Cancercancerres.aacrjournals.org Downloaded from

CEA-TRANSDUCED CARCINOMA CELLS

oligomers. One of the inserts (NC-1) identified by hybridization withthe PR1 was subcloned and partially sequeneed using the dideoxynu-cleotide chain termination technique (14). The NC-1 insert was foundto be identical to the published CEA sequence (15), starting 85 basepairs upstream of the translation start site, but only extended to theinternal BamHl site present within the first repeated domain of CEA(Fig. la, Domain I). One of the inserts identified by hybridization withPR2 (NC-2) was subcloned, sequeneed, and found to contain theinternal BamHl site, the downstream coding region of CEA, along with340 base pairs of 3' untranslated region. The absence of full-length

clones may have resulted from restriction at the internal BamHl siteduring preparation of the library. The inserts NC-1 and NC-2 werethen ligated at the internal BamHl site, and the sequence of the resultantproduct, extending for approximately 200 bases 5' or 3' from this site,

was obtained. The sequence differed from the published sequence onlyat the junction of the second and third repeated domains (Fig. la.Domains 11 and III). At this position, a deletion of 3 base pairs wasfound, which would yield a protein lacking an alanine residue at thisposition. This residue has been shown to be encoded by nucleotidespresent within the sixth and seventh exons of the CEA gene (15); thus,this product may be the result of a splicing error.

Retroviral Expression Vector. The retroviral expression vector pBNC,provided by Dr. W. F. Anderson (National Heart, Lung, and BloodInstitute, NIH, Bethesda, MD), contains the Moloney leukemia virusLTR regions. This vector was derived from the B2 vector (16) byinsertion of a 1.5-kilobase EcoRI fragment of the neomycin phospho-transferase gene, isolated from the N2 vector (17), into the unique Sphlsite present in B2. The B2 vector was then digested with Sa/I andligated to an 816-base pair fragment of the CMV promoter/enhancer(18). The CEA insert was isolated by digestion with Smal and wasligated into the unique ßamHIcloning site in pBNC, which had beenconverted into a blunt end by the Klenow fragment of DNA polymerase.This construct was designated PB-CEA.

The structure of the retroviral vector containing the CEA insert isshown graphically in Fig. la. Expression of the CEA gene is drivenfrom the CMV promoter, and the neomycin phosphotransferase geneis driven by the retroviral LTR. The mature CEA glycoprotein containsan A'-terminal region, followed by a set of 3 very similar repeated

domains (designated I. II, and III), and a short hydrophobic C-terminalregion.

Transfection and Transduction of DNA. The pB-CEA plasmid wastransfected into the PA317 packaging cell line using Lipofectin (Bethesda Research Laboratories, Gaithersburg, MD) according to themanufacturer's instructions. One wk following selection in 0.5 mg/ml

of G418, the medium was changed, and the supernatant was collectedfor 24 h. Transduction was carried out by incubating 1 x IO6\/-2 cellsin a 6-cm dish with 1 ml of supernatant for 30 min at 37°C.Three ml

of medium were then added, and 24 h later cells were removed bytrypsinization and plated in 15-cm dishes. Twenty-four h later, G418was added (0.5 mg/ml), and after selection for 2 wks in mediumcontaining G418, cells were cloned by limiting dilution. Clones of \¡/-2cells containing the full-length CEA insert were identified by probing

5'LTRNeoCMVN1IIIII3-LTH

MC-38 cea-1

b.

1 KbMC-38 cea-2

Fig. 1. Structure of relroviral vector and CEA cDNA insert. In a, a full-lengthcDNA-encoding CEA was inserted into the unique Hom\\\ site of the vectorpBNC. The mature CEA glycoprotein contains an .V-tcrminal region (A'), 3repeated copies of an immunoglobulin-like domain (/. //. ///). and a hydrophobicC-terminal domain (C). The locations of the EcoKl (E) and BamHl (B) restrictionenzyme sites arc indicated, b, the proposed structure for the CEA insert in theMC-38-cca2 clone.

Southern blots of DNA isolated from those clones with the pCEAlPstl probe. MC-38 cells could not be transduced with the recombinantecotropic virus; therefore, PA317 cells were transduced with virusobtained from ^-2 clones shown to contain the CEA insert. Followingselection of transduced PA317 cells in 0.5 mg/ml of G418 for 3 wk,retroviral supernatants were obtained and used to transduce MC-38cells. Following transduction, MC-38 cells were cloned by limitingdilution.

Analysis of CEA Glycoprotein Expression. Immunofluorescence assays were carried out to analyze cell surface expression of CEA asdescribed (11). The anti-CEA monoclonal antibodies COL-1 (IgG2a),COL-4 (IgG2a), COL-6 (IgGl), COL-8 (IgGl), COL-9 (IgG2b), COL-12 (IgGl) (19), and Bl.l (IgG2a) (20), as well as the negative controlmurine myeloma MAb UPC-10 (IgG2a), were used as primary antibodies in this assay. The MAb B6.2 (IgGl) (21) does not react with CEAbut reacts with a closely related antigen, NCA (22, 23), and was alsoutilized as primary antibody. Fluorescein-labeled goat anti-mouse IgG+ IgM (Kirkegaard and Perry, Gaithersburg, MD) was used as thesecondary antibody. Analysis was performed with a FACScan (Becton-Dickinson Mountain View, CA) equipped with a blue laser excitationof 15 milliwatts at 488 nm. Each figure represents data obtained fromanalysis of 10,000 cells. Cell extracts were prepared by lysis of mono-layers in nonionic detergent (24). For Western blot analysis, proteinswere separated on 4 to 12% gradient sodium dodecyl sulfate-polyacryl-amide gels (Novex, Encinido, CA). Proteins were transferred to a PVDFmembrane (0.45-iim pore size) (Millipore Corporation, Bedford, MA)using a semidry blotter (Bio-Rad, Richmond. CA) in 20% methanol,0.039 M glycine, and 0.048 M Tris for l h at 1 watt/30 cm2. Followingtransfer, membranes were incubated for l h with the MAb COL-1 (1ut! ml). Bound antibody was detected using biotinylated goat anti-mouse antibody, followed by streptavidin-alkaline phosphatase suppliedwith the IMMUNOSELECT immunoblotting system, obtained fromBethesda Research Laboratories. Quantitation of CEA in cell extractswas carried out using a CEA enzyme ¡mmunoassaykit obtained fromRoche Diagnostic Systems, Inc. (Montclair, NJ), as well as a CEAradioimmunoassay kit, obtained from Abbott Laboratories, Inc. (Chicago, IL).

Radioimmunoprecipitations were carried out essentially as described(24). Briefly, 1x10" cells were seeded in a T25 flask, and the following

day the medium was removed. Labeling was carried out using theRPMI-1640 Select-Aminékit, obtained from GIBCO (Grand Island,NY). The flasks then received either 5 ml of medium without unlabeledleucine but with 0.5 mCi of ['H]leucine (40 to 60 Ci/mmol) (New

England Nuclear, Boston, MA) or medium with 50 Mg/ml of unlabeledglucose and 0.5 mCi of [3H]glucosamine (30 to 60 Ci/mmol) (New

England Nuclear) for 18 h. Cell extracts were then prepared as describedabove, and immunoprecipitations were carried out with 1 ml of tissueculture supernatant containing either the anti-CEA MAb COL-1 or thenegative control MAb RPC-5 (IgG2a). The immunoprecipitates wereanalyzed on a 10% sodium dodecyl sulfate-polyacrylamide gel, whichwas stained with 0.25% Coomassie blue, destained, and incubated withENHANCE (New England Nuclear) according to the manufacturer's

instructions. The gels were then dried and subjected to autoradiography.Amplification of DNA by PCR. Two primers were used to amplify

the human CEA gene from the genomic DNA of the transduced MC-38 clones. The first (5'-AACTCAAGCTCTTCTCCACA) was locatedin the 5'-untranslated region of the CEA gene, and the second (5'-

AAGCAATTTTAGACTGTAG) was complementary to a portion ofthe 3'-untranslated region. The PCR was carried out using l ¿igof

genomic DNA and 500 ng of each of the two primers in PCR buffer[10 mM Tris(pH 8.3)-50 itiM KC1-1.5 mM MgCl2] with 5 units of TagDNA polymerase (Bethesda Research Laboratories). Primers were annealed for 2 min at 55°C,extension was carried out for 3 min at 70°C,and denaturation was carried out for 2 min at 94°C.Amplification was

carried out for 30 cycles.

RESULTS

CEA Glycoprotein Expression in Transduced MC-38 Cells.Following selection in G418, transduced MC-38 cells were

3658

Research. on September 7, 2020. © 1991 American Association for Cancercancerres.aacrjournals.org Downloaded from

CEA-TRANSDUCED CARCINOMA CELLS

cloned by limiting dilution, and the expression of CEA on thesurface of transduced clones was then analyzed in an immuno-fluorescence assay. Nine clones that expressed high levels ofCEA were identified, and two (MC-38-ceal, MC-38-cea2) werechosen for more extensive analysis. The levels of cell surfaceexpression of CEA found on clones MC-38-ceal and MC-38-cea2 (Fig. 2, a and b) were comparable to those found on theGEO and CBS human colon carcinoma cell lines (Fig. 2, d ande). GEO and CBS are two human colon tumor cell lines thatexpress levels of CEA which are substantially higher than thosefound in most human colon carcinoma cell lines (11). Cellsurface expression of CEA on the MC-38-ceal and MC-38-cea2 cell lines was considerably higher than that found on thesurface of WiDr, a human colon carcinoma cell line expressingmoderate levels of CEA (Fig. If}.

Quantitation of CEA expression in cell extracts was thencarried out using the Roche enzyme immunoassay, which is adouble-determinant assay utilizing two anti-CEA MAbs. Levelsof CEA in MC-38-ceal and MC-38-cea2 cell extracts werehigher than that found in an extract prepared from the WiDrcell line (Table 1). Much higher levels were found in proteinextracts prepared from the GEO and CBS cell lines (Table 1).CEA levels were then quantitated using the Abbott radio-immunoassay, which is a double-determinant assay containingtwo additional anti-CEA MAbs. When the MC-38-ceal extractwas assayed with the Abbott assay, values similar to that obtained with the Roche assay were obtained, with the exceptionof the MC-38-cea2 extract, which was negative in the Abbottassay. This indicates that at least one of the epitopes presenton the native CEA molecule is not expressed on the transducedgene product expressed in MC-38-cea2 (to be discussed below).

Expression of CEA Epitopes. The expression of additionalCEA epitopes on the MC-38-ceal and MC-38-cea2 cell lineswas then examined by carrying out a cell surface immunofluo-rescence assay using a number of MAbs (Fig. 3). The epitopesdefined by the MAbs COL-1, 6, 8, and 9 are found on CEA butnot on NCA, whereas those defined by the MAbs COL-4, 12,and B 1.1 are found on both CEA and NCA (25). All of theanti-CEA MAbs reacted with the two transduced clones; however, the level of expression of these epitopes appeared to behigher on the MC-38-ceal clone than on the MC-38-cea2 clone.The ratio of the mean fluorescence intensity found on MC-38-ceal versus that found on MC-38-cea2 varied from 1.3 withCOL-9 to 2.5 with COL-8 (Fig. 3), further indicating that MC-38-cea2 may express an altered CEA gene product. The MAbB6.2, which reacts with NCA but not CEA (21), failed to reactwith either of the transduced clones.

Characterization of Transduced CEA Gene Products. The sizeof the CEA gene products expressed by the transduced cellswas then examined in a Western blot assay. The MC-38-cealcell line expressed the expected M, 180,000 glycoprotein (Fig.4, Lane 2), similar to that found in the WiDr cell line (Lane1). The immunoreactive product present in the MC-38-cea2

cell line, however, had a molecular weight of approximately70,000 (Lane 3). Examination of seven additional clones indicated that the size of the CEA gene products frequently variedin clones of transduced cells. Four clones expressed a single A/r70,000 immunoreactive product, two expressed a single MT140,000 product, and one expressed two products with molecular weights of 180,000 and 140,000 (data not shown).

The molecular weight of the MC-38-cea2 product was verysimilar to that of the predicted polypeptide backbone of CEA(M, 72,000) and could, therefore, have resulted from the lack

m

FLUORESCENCE INTENSITY

Fig. 2. Cell surface expression of CEA in transduced cell lines. The expressionof CEA was examined by carrying out immunofluorescence assays with the anti-CEA MAb COL-1 ( ) or the control antibody UPC-10 ( ). The meanfluorescence intensities obtained with the control MAb UPC-10 and the anti-CEA MAb COL-1. respectively, are given in parentheses following the celldesignations. The fluorescence intensity was plotted on a linear scale. The cellswhich were analyzed are as follows: A. MC-38-ceal (30.5. 95.2); B. MC-38-cea2(21.1, 100); C MC-38 (32.2. 29.2); D, GEO (43.4. 102); E. CBS (60.8. 109); andF. WiDr (51.2. 70.7).

Table 1 CEA expression in cell extracts

CelllineGEOCBSWiDrA375MC-38-cealMC-38-cea2MC-38DescriptionHuman

colonadenocarcinomaHumancolonadenocarcinomaHumancolonadenocarcinomaHuman

melanomaMousecolonadenocarcinoma(A/,

180,000CEA)Mousecolonadenocarcinoma(M,

70.000CEA)Mousecolon adenocarcinomaCEA

(ng/mg)Assay

1" Assay2°8400

44001300320017

9Negative*Negative230

ISO57

NegativeNegative

Negative" Quantitation of CEA was carried out using the Roche enzyme immunoassay

kit (Assay 1) and the Abbott radioimmunoassay kit (Assay 2).* Less than 2.5 ng/mg.

KLUCD

O

FLUORESCENCE INTENSITY

Fig. 3. Expression of CEA epitopes in transduced cell lines. A cell surfaceimmunofluorescence assay was carried out on the MC-38-ceal cell line ( )and the MC-38-cea2 cell line ( ) with the MAbs COL-1 (A)( 180. 106). COL-4 (B) (320. 201). COL-6 (C) (250. 152). COL-8 (D) (164. 65). COL-9 (E) (165.123). COL-12 (F) (151. 69). Bl.l (G) (253. 124), B6.2 (//) (29. 43). and UPC-10 (/) (26. 38). The numbers in parentheses represent the mean fluorescenceintensity obtained when using these MAbs to stain the MC-38-ceal and MC-38-cea2 clones, respectively. The fluorescence intensity was plotted on a logarithmicscale.

3659

Research. on September 7, 2020. © 1991 American Association for Cancercancerres.aacrjournals.org Downloaded from

CEA-TRANSDUCED CARCINOMA CELLS

1234

200 —

97 —

68 —

43 —

29 —

Fig. 4. Western blot analysis of CEA gene products. A Western blot wascarried out on protein extracts of the WiDr cell line (Lane I), MC-38-ceal (Lane2). MC-38-cea2 (Lane 3), and MC-38 (Lane 4).

200 —

97.4 —

68 —

43 —

25.4 —

18.4 —

Fig. 5. Analysis of radiolabeled CEA gene products. Immunoprecipitationswere carried out on extracts of the MC-38-cea2 cell line (Lanes I to 4) and GEOcell line (Lanes S to 8). Extracts in Lanes I. 2, 5. and 6 were prepared from cellslabeled in the presence of |3H]leucine, and those in Lanes 3, 4, 7, and * wereprepared from cells labeled in the presence of [3H]glucosamine. Immunoprecipi-tates in even-numbered lanes were prepared with MAb COL-1, while those inodd-numbered lanes were prepared with the control MAb RPC-5.

of glycosylation of the CEA gene product. To determine if thetransduced product of MC-38-cea2 was glycosylated, metaboliclabeling was carried out with either ['H]glucosamine or [3H]-leucine, followed by immunoprecipitation with the COL-1antibody. A M, 180,000 product was precipitated from both['Hjleucine- and ['Hjglucosamine-labeled extracts of the GEOcell line (Fig. 5, Lanes 6 and 8). In the MC-38-cea2 line, a M,70,000 immunoreactive product was strongly labeled with['Hjglucosamine, indicating that this gene product is highly

glycosylated (Lane 4). As shown in Fig. 5, the A/r 70,000transduced product was also labeled with ['Hjleucine (Lane 2).

The CEA gene product expressed in human colon carcinomacells has been shown to be inserted in the membrane throughthe use of a glycosylphospholipid anchor (26). Treatment withphospholipase C was found to remove the majority of the CEAfrom the surface of the MC-38-cea2 cell line (data not shown),indicating that the CEA gene product in transduced MC-38cells is also anchored to the membrane through a lipid linkage.

Analysis of Integrated CEA Gene. In order to characterize themutation present in MC-38-cea2, a Southern blot was carriedout with DNA isolated from this clone, as well as from theMC-38-ceal clone. Digestion with BamHl should release twofragments that hybridize to the CEA probe: a 1-kilobase fragment containing the 5' end of CEA as well as a unique fragment

whose size is dependent on the integration site of the vector(Fig. la). Unique CEA-hybridizing fragments of 4.5 and 6.0kilobases were seen following BamHl digestion of DNA fromMC-38-ceal and MC-38-cea2, respectively (Fig. 6, Lanes 2 and3). The 1-kilobase fragment, however, was present in theBamHl digest of MC-38-ceal but not MC-38-cea2 DNA, in

dicating that a portion of the CEA gene had been deleted inMC-38-cea2. Additional analysis has indicated that the high-molecular-weight band seen in the BamHl digest of MC-38-ceal DNA is present in digests of MC-38 and MC-38-cea2DNA (data not shown) and may therefore result from hybridization of this probe with a homologous mouse gene (27). Thus,only one copy of the human CEA gene appears to be presentin each of the two cloned cell lines. In order to determine thesize of this deletion, DNA isolated from the MC-38-cea2 clonewas digested with EcoRl, which should release the entire CEAinsert on a single 4.4-kilobase fragment. A 4.4-kilobase bandthat hybridized to the CEA probe was found following digestionof MC-38-ceal DNA with EcoRl (Lane 5), whereas a 3.3-kilobase band was seen with the EcoRl digest of MC-38-cea2DNA (Lane 6). The high-molecular-weight bands seen in theEcoRl digest appear to be shared with the parental MC-38 cellline and, thus, again may result from hybridization of the CEAprobe to homologous mouse genes. To further analyze thedeletion that had taken place, digestion was carried out withEcoRl plus BamHl, which should release the CEA insert ontwo fragments of 1.5 and 0.9 kilobases. Digestion of DNA from

123456789

23-

9.4-6.6-

4.4-

2.2-2.0-

Fig. 6. Southern blot analysis of DNA isolated from transduced clones.Southern blot analysis was carried out using DNA isolated from the MC-38(Lanes 1. 4, 7). MC-38-ceal (Lanes 2, 5. 8). and MC-cea2 (Lanes 3, 6, 9) celllines. The DNA samples were digested with either BamHl (Lanes I to 3), EcoRl(Lanes 4 to 6), or fcoRI plus BamHl (Lanes 7 to 9). and hybridization wascarried out with the pCEA 1 probe.

3660

Research. on September 7, 2020. © 1991 American Association for Cancercancerres.aacrjournals.org Downloaded from

CEA-TRANSDUCED CARCINOMA CELLS

the MC-38-ceal cell line with these enzymes gave rise to twobands of the expected size (Lane 8), whereas digestion of MC-38-cea2 DNA yielded a single 1.4-kilobase band that hybridizedwith the CEA probe (Lane 9). These results indicated that adeletion of approximately 1.0 kilobase had occurred within theCEA gene present in the MC-38-cea2 cell line.

The CEA gene present in MC-38-cea2 was then isolated byusing the polymerase chain reaction to amplify genomic DNAisolated from this clone. A 1.1-kilobase fragment was obtainedfrom the MC-38-cea2 cell line using oligonucleotide primersthat flank the CEA coding region, and this product was sub-cloned and partially sequenced. The sequence of approximately150 nucleotides at the 5' and 3' ends of the amplified product

corresponded to the published sequence (15). indicating that nodeletions had taken place at either end of the integrated CEAgene. The sequence of the MC-38-cea2 clone in the regionfollowing the A'-terminal domain corresponded to that of the

first of the repeated domains of CEA (Domain I) for 87 nucleotides (Fig. 7). Over the next 48 bases, however, 7 differenceswere found between the sequence of the MC-38-cea2 clone andthe Domain I sequence. The sequence in this region was,however, identical to that found in the corresponding region ofDomain III. This indicates that a recombinational event hastaken place between the first 87 nucleotides of Domains I andIII, resulting in a product containing one rather than threecopies of the repeated domain of CEA (Fig. 16). These findingsare consistent with the Southern blotting results discussedabove, which indicated that a 1-kilobase deletion had takenplace within the integrated human CEA gene in MC-38-cea2.

DISCUSSION

Studies have demonstrated expression of the CEA gene in L-cells, as well as Chinese hamster ovary cells (28), but the studiesreported here represent the first example of expression of thehuman CEA gene product in a mouse tumor cell line, and inparticular a colonie epithelial tumor cell. Following transduc-tion of a mouse colon adenocarcinoma cell line (MC-38) withthe CEA gene, clones were obtained that express levels of cellsurface CEA comparable to those found in GEO and CBS cells,two human colon tumor cell lines that express high levels ofCEA. However, the levels of CEA found in total cell extractsof the transduced clones were substantially lower than the levelsfound in the GEO and CBS tumor cell lines. The failure ofGEO and CBS to express significantly higher levels on theircell surface may reflect increased turnover of CEA in these

CEA I CCGGAGCTGCCCAAGCCCTCCATCTCCAGCAACAACTCCAAACCCGTGGAGGACAAGGAT 60CEA II G-A CA A T A C TG

CEA III GMC-38-cea2 —

CEA I GCTGTGGCCTTCACCTGTGAACCTGAGACTCAGGACGCAACCTACCTGTGGTGGGTAAAC 120CEA II A A -T A—A T

CEA III G A--A TMC-38-cea2 G A--A T

CEA ICEAi:CEA III GGMC-38-cea2 GG

AATCAGAGCCTCCCG 135

Fig. 7. Sequence of MC-38-cea2 gene product. The CEA gene was cloned fromthe MC-38-cea2 cell line as described in "Materials and Methods," and partial

sequencing was carried out on the isolated product. The sequence of the first 195bases of the MC-38-cea2 product which follow the A'-terminal domain is here

compared with the published sequence of the 3 repeated domains of CEA (26). Adash indicates identity with the sequence of CEA present in the CEA Domain I.

cells. Alternatively, since CEA is linked to the cell membranethrough a glycosyl phosphatidylinositol group (26), the availability of sites on the surface of these cells for linkage to CEAmay limit the cell surface expression of CEA in GEO and CBS.

The transduced product observed in the MC-38-cea2 cell lineis a glycosylated protein with a molecular weight of approximately 70,000, whereas the normal CEA gene product has amolecular weight of 180,000. Southern blot analysis indicatedthat a deletion of the CEA sequence had occurred, and sequencing of the CEA gene isolated from the MC-38-cea2 by thepolymerase chain reaction indicated that this product containedonly one copy of the domain present in three copies in thenormal CEA gene product. In spite of the loss of two domains,the MC-38-cea2 product expressed the epitopes recognized byall of the anti-CEA MAbs tested, with the exception of at leastone of the MAbs in the Abbott CEA radioimmunoassay kit.Antibodies that react with the MC-38-cea2 product include fourMAbs which recognize epitopes present on CEA but not onNCA, as well as three MAbs that react with epitopes sharedbetween CEA and NCA (25). Some of these epitopes may berepresented more than once on the normal CEA gene product,since approximately 70% of the amino acid sequence is conserved among the three repeated domains. The loss of one ortwo of the repeated domains would not be expected to affectrecognition by these antibodies. Some of these epitopes mayalso be present within the A'-terminal domain of CEA, whichappears to be unaltered in the MC-38-cea2 product.

In terms of appropriateness for an in vivo model for a CEA-expressing tumor, there are three points to consider: totalcellular CEA content, expression of CEA on the cell surface,and amount of CEA shed. For comparison, we have used humancolon cancer cell lines expressing high (GEO) and moderate(WiDr) levels of CEA. As shown in this study, the MC-38-cealand MC-38-cea2 cells express as much cell surface CEA as

GEO, and they express significantly more cell surface CEAthan the WiDr cell line. Additional experiments have shownthat transduced MC-38 cells shed relatively low amounts ofCEA, comparable to the amount shed by the WiDr cell line,but considerably less than that shed by the GEO cell line (datanot shown). One could speculate that the turnover of CEA inMC-38 cells may be comparable to that seen in moderately

expressing cell lines such as WiDr. Therefore, in comparisonto most human carcinomas, we believe the transduced MC-38cells express substantial amounts of cell surface CEA and shedlow levels of CEA, which should make this an appropriatetarget for immunotherapy.

The generation of a mouse tumor cell line which expresseshuman CEA should now allow development of an in vivo modelsystem to test anti-CEA therapies. Using these cells, activeimmunotherapy, as well as antibody-mediated therapy directedagainst CEA, can be evaluated in an animal model of disseminated disease, since MC-38 cells form pulmonary tumors following i.v. injection (4) and hepatic tumors following i.s. injection (5). Furthermore, the target tumor will have a matchedcontrol tumor, the only difference being the expression of thehuman CEA gene. Studies currently under way indicate thatthe CEA-transduced cells grow and express CEA in athymicmice, X-irradiated C57BL/6 mice and immunologically intact

C57BL/6 mice. Detailed studies are also in progress or plannedto compare growth rates of the MC-cea-1 and -cea-2 clones ineach of these hosts, as well as quantitation (if any) of hostimmune T- and/or B-cell responses.

3661

Research. on September 7, 2020. © 1991 American Association for Cancercancerres.aacrjournals.org Downloaded from

CEA-TRANSDUCED CARCINOMA CELLS

ACKNOWLEDGMENTS

We would like to thank Dianne Poole for her excellent technicalassistance.

REFERENCES

1. Estin. C. D., Stevenson, U. S.. Plowman, G. D., Hu, S-H., Sridhar. P..Hellstrom. I., Brown. J. P., and Hellstrom, K. E. Recombinant vaccinia virusvaccine against the human melanoma antigen p97 for use in immunotherapy.Proc. Nati. Acad. Sci. USA, «5:1052-1056. 1988.

2. Hu, S-L., Plowman, G. D., Sridhar. P.. Stevenson, U. S., Brown, J. P., andEstin, C. D. Characterization of a recombinant vaccinia virus expressinghuman melanoma-associated antigen p97. J. Virol., 62: 176-180, 1988.

3. Sikorska, H., Shuster, J., and Gold, P. Clinical applications of carcinoem-bryonic antigen. Cancer Detect. Prev., 12: 321-355, 1988.

4. Mule, J. J., Shu, S., Schwarz, S. L., and Rosenberg. S. A. Successful adoptiveimmunotherapy of established pulmonary métastaseswith LAK cells andrecombinant IL-2. Science (Washington DC), 225: 1487-1489, 1984.

5. Lafreniere. R., and Rosenberg, S. A. A novel approach to the generation andidentification of experimental hepatic métastasesin a murine model. J. Nati.Cancer Inst., 76: 309-322, 1986.

6. Rosenberg. S. A., Spiess, P. J., and Lafreniere. R. A new approach to theadoptive immunotherapy of cancer with tumor-infiltrating lymphocytes. Science (Washington DC). 233: 1318-1321. 1986.

7. Spiess, P. J., Yang, J., and Rosenberg, S. A. In vivo antitumor activity oftumor infiltrating lymphocytes expanded in recombinant IL-2. J. Nat!. Cancer Inst.. 79: 1067-1075. 1987.

8. Fox, B. A., Spiess, P. F., Kasid, A., Puri, R.. Mule, J. J., Weber, J. S., andRosenberg, S. A. In vitro and in vivo antitumor properties of a T-cell clonegenerated from murine tumor-infiltrating lymphocytes. J. Biol. ResponseModif., 9:499-511. 1990.

9. Mann, R., Mulligan. R. C, and Baltimore, D. Construction of a retroviruspackaging mutant and its use to produce helper-free defective retrovirus.Cell, .H: 153-159, 1983.

10. Miller, A. D., Law, M-F., and Verma, I. M. Generation of helper-freeamphotropic retroviruses that transduce a dominant-acting, methotrexate-resistant dihydrofolate reducíasegene. Mol. Cell Biol., 5: 431-437, 1985.

11. Guadagni, F., Witt, P. L., Robbins, P. F., Schlom, J.. and Greiner, J. W.Regulation of carcinoembryonic antigen expression in different human co-lorectal tumor cells by -y-interferon. Cancer Res.. 50:6248-6255, 1990.

12. Meissner, P. S., Sisk, W. P., and Berman, M. L. Bacteriophage X cloningsystem for the construction of directional cDNA libraries. Proc. Nati. Acad.Sci. USA, 84: 4171-4175. 1987.

13. Zimmermann, W., Ortlieb, B., Friedrich, R., and von Kleist, S. Isolation andcharacterization of cDNA clones encoding the human carcinoembryonicantigen reveal a highly conserved repeating structure. Proc. Nati. Acad. Sci.USA, 84: 2960-2964, 1987.

14. Sanger, F.. Nicklen, S., and Coulson. A. R. DNA sequencing with chainterminating inhibitors. Proc. Nati. Acad. Sci. USA, 74: 5463-5467, 1977.

15. Schrewe, H., Thompson, J., Bona, M.. Hefta, L. J. F.. Maruya, A.. Hassauer.M., Shively, J. E., Von Kleist, S., and Zimmermann. W. Cloning of the

complete gene for carcinoembryonic antigen: analysis of its promoter indicates a region conveying cell type-specific expression. Mol. Cell Biol., 10:2738-2748, 1990.

16. Zwiebel. J. A., Freeman, S. M., Kantoff, P. W., Cornetta, K., Ryan, U. S.,and Anderson, W. F. High-level recombinant gene expression in rabbitendothelial cells transduced by retroviral vectors. Science (Washington DC),243: 220-221, 1989.

17. Eglitis. M. A.. Kantoff, P., Gilboa, E., and Anderson, W. F. Gene expressionin mice after high efficiency retroviral-mediated gene transfer. Science (Washington DC). 230: 1395-1398. 1985.

18. Boshart, M., Weber, F., Jahn, G., Dorsch-Haesler, K.. Fleckenstein, B., andSchaffner, W. A very strong enhancer is located upstream of an immediateearly gene of human cytomegalovirus. Cell. 41: 521-530, 1985.

19. Muraro, R., Wunderlich. D., Thor. A., Lundy, J., Noguchi, P.. Cunningham,R., and Schlom, J. Definition by monoclonal antibodies of a repertoire ofepitopes on carcinoembryonic antigen differentially expressed in humancolon carcinomas versus normal adult tissues. Cancer Res., 45: 5769-5780,1985.

20. Herlyn, M.. Blaszczyk, M.. Sears, H. F., Verrill, H., Lindgren, J., Colcher,D.. Steplewski, Z., Schlom, J.. and Koprowski, H. Detection of carcinoembryonic antigen and related antigens in sera of patients with gastrointestinaltumors using monoclonal antibodies in double-determinant radioimmuno-assays. Hybridoma, 2: 329-339. 1983.

21. Hand, P. H., Nuti, M., Colcher, D.. and Schlom, J. Definition of antigenicheterogeneity and modulation among human mammary carcinoma cell populations using monoclonal antibodies to tumor-associated antigens. CancerRes., 43: 728-735, 1983.

22. Von Kleist, S., Chavanel, G., and Burtin, P. Identification of an antigen fromnormal human tissue that cross-reacts with the carcinoembryonic antigen.Proc. Nati. Acad. Sci. USA, 69: 2492-2494, 1972.

23. Tawaragi. Y., Oikawa, S., Matsuoka. Y.. Kosaki, G., and Nakazato, H.Primary structure of nonspecific cross-reacting antigen (NCA). a member ofthe carcinoembryonic antigen (CEA) gene family, deduced from cDNAsequence. Biochem. Biophys. Res. Commun., 150: 89-96, 1988.

24. Fernsten. P. D., Pekny, K. W„Reisfeld, R. A., and Walker, L. E. Biosynthesisand glycosylation of the carcinoma-associated antigen recognized by monoclonal antibody KS1/4. Cancer Res., 50: 4656-4663, 1990.

25. Kuroki. M.. Greiner, J. W., Simpson, J. F.. Primus, F. J., Guadagni, F., andSchlom, J. Serologie mapping and biochemical characterization of the carcinoembryonic antigen epitopes using fourteen distinct monoclonal antibodies. Int. J. Cancer. 44: 208-218. 1989.

26. Hefta. S. A., Hefta, L. J. F., Lee, T. D., Paxton, R. J., and Shively, J. E.Carcinoembryonic antigen is anchored to membranes by covalent attachmentto a glycosylphosphatidylinositol moiety: identification of the ethanolaminelinkage site. Proc. Nati. Acad. Sci. USA, «5:4648-4652, 1988.

27. Beauchemin, N., Turbide, C., Afar, D., Bell, J., Raymond, M.. Stanners, C.P., and Fuks, A. A mouse analogue of the human carcinoembryonic antigen.Cancer Res., 49: 2017-2021, 1989.

28. Hefta, L. J. F., Schrewe, H.. Thompson, J. A., Oikawa, S., Nakazato, H.,and Shively, J. E. Expression of complementary DNA and genomic clonesfor carcinoembryonic antigen and nonspecific cross-reacting antigen inChinese hamster ovary and mouse fibroblast cells and characterization of themembrane-expressed products. Cancer Res.. 50: 2397-2403, 1990.

3662

Research. on September 7, 2020. © 1991 American Association for Cancercancerres.aacrjournals.org Downloaded from

1991;51:3657-3662. Cancer Res   Paul F. Robbins, Judith A. Kantor, Michael Salgaller, et al.   Antigen Gene in a Murine Colon Carcinoma Cell LineTransduction and Expression of the Human Carcinoembryonic

  Updated version

  http://cancerres.aacrjournals.org/content/51/14/3657

Access the most recent version of this article at:

   

   

   

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

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

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

  Permissions

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

.http://cancerres.aacrjournals.org/content/51/14/3657To request permission to re-use all or part of this article, use this link

Research. on September 7, 2020. © 1991 American Association for Cancercancerres.aacrjournals.org Downloaded from