(CANCER RESEARCH 47, 56-61, January 1, 1987]

Antigenic Heterogeneity of Carcinoembryonic Antigen in the Circulation Defined byMonoclonal Antibodies against the Carbohydrate Moiety of CarcinoembryonicAntigen and Closely Related Antigens1

Akira Matsunaga, Masahide Kuroki, Hiroshi Higuchi, Fumiko Arakawa, Kyoko Takakura, Naomi Okamoto, andYuji Matsuoka2

First Department of Biochemistry, School of Medicine, Fukuoka University, 7-45-1 Nanakuma, Jonan-ku, Fukuoka SI 4-01¡A.M., M. K., F. A., K. T., N. O., Y. M.J andKyowa Hakko Technical Research Laboratory, 1-1 Kyowa-machi, Hofu-shi, Yamaguchi 747 [H. H], Japan

ABSTRACT

Six mouse monoclonal antibodies reactive with Carcinoembryonic antigen (CEA) were prepared and used for the analysis of the antigenicheterogeneity of CEA in patient sera. Their reaction specificity and thechemical nature of antigenic epitopes recognized by them were analyzedby radioimmunoassay on the basis of reactivities with different preparations of CEA, normal fecal antigen 2, and nonspecific cross-reactingantigen 2 before and after chemical and/or enzymatic treatment. Twoantibodies, F3-30 and F4-82, raised with CEA were reactive with differentpeptide epitopes on the antigen molecules and revealed a quite universa!reactivity with all CEA, normal fecal antigen 2, or nonspecific cross-reacting antigen 2 preparations tested. The serum CEA values obtainedwith these antibodies were highly correlated with those obtained withconventional radioimmunoassays for CEA. The other four antibodies (14-11 and F33-37 raised with CEA, F8-52 with normal fecal antigen 2, andF48-60 with nonspecific cross-reacting antigen 2) were found to recognizecarbohydrate epitopes with different specificities and revealed very heterogeneous reactivities. The serum CEA values estimated with these fourantibodies were highly variable depending on the antibody used, suggesting that the expression of carbohydrate epitopes on the CEA moleculesin patient sera was quite heterogeneous. The antigenic heterogeneity ofthe carbohydrate epitopes was detected even in a single patient serum byaffinity chromatography.

The causes that give rise to the difference in CEA values between theRoche and the Daiichi kits were analyzed on the basis of reactivities ofthree groups of patient sera, which showed extremely different ratios forthe Roche and Daiichi kits, with monoclonal anti-carbohydrate antibodies. The results obtained suggest that, at least in part, the diversity ofantigenic expression on carbohydrate chains on the CEA molecules inpatient sera and the variation in specificity or quantity of anti-carbohydrate antibodies in the polyclonal antibody preparations used for therespective assay systems may result in the differences in the estimatedCEA values.

INTRODUCTION

As the clinical usefulness of CEA3 is being widely accepted,

many assay systems for CEA are currently available. From aclinical viewpoint, however, it seems a troublesome problemthat the CEA value of a given cancer patient serum is highlyvariable depending on the assay systems used (1, 2). In aprevious study (3), we reported that the reaction intensity ofseveral purified CEA preparations with several assay systems ishighly variable depending on the match between the antigenpreparation and the assay system used. The variability in CEAassays must result from the antigenic heterogeneity of CEA,

Received 5/20/86; revised 9/5/86; accepted 9/10/86.The costs of publication of this article were defrayed in pan by the payment

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

1This work was supported in part by a Grant-in-Aid for Cancer Research fromthe Ministry of Education, Science and Culture, Japan.

2To whom requests for reprints should be addressed.3The abbreviations used are: CEA, Carcinoembryonic antigen; NFA-2, normal

fecal antigen 2; NCA-2, nonspecific cross-reacting antigen 2; RIA, radioimmunoassay; NCA, nonspecific cross-reacting antigen.

but the molecular basis of the heterogeneity of CEA in thecirculation remains to be elucidated.

It has been well documented that CEA is a glycoprotein witha molecular weight of about 180,000 containing 50-60% carbohydrate and that the amino acid composition of differentpurified preparations of CEA is very similar but the carbohydrate composition is fairly variable (4-7). Although the antigenic heterogeneity among purified CEA preparations washardly detectable upon immunodiffusion tests, their reactivityon RIA is highly variable depending on the antigens tested oron the antibodies used (3, 8, 9). Thus, minor antigenic heterogeneity may critically influence the very sensitive assay systems.Since a minor population of antibody-producing cells can beexpanded to the full by the hybrido ma technique (10), it couldbe expected that some monoclonal antibodies to CEA, especially those reactive with the sugar moiety of the CEA molecule,would be able to detect and analyze such a minor antigenicCEA molecule heterogeneity. Thus far, several studies usingmonoclonal antibodies have described the heterogeneity of theCEA molecule immunochemicaliy and histochemically (11-18), but only a few reports have investigated the heterogeneityof CEA in the circulation (19, 20).

In the present study, we prepared several monoclonal antibodies to CEA and closely related antigens, NFA-2 in normaladult feces (21, 22) and NCA-2 in meconium (23, 24), andfound some definite evidence for the antigenic heterogeneity onthe sugar moiety of the circulating CEA molecule, which givesrise to the variability of the estimated serum CEA levels.

MATERIALS AND METHODS

Antigens. Nine individual CEA preparations were highly purifiedfrom metastatic liver tumors from colon carcinomas as described previously (25, 26). A WHO international standard CEA preparation (73/601-2/22J) was provided by the National Institute for Biological Standards and Control, London, United Kingdom, and referred to asCEA(WHO). Highly purified preparations of NFA-2 and NCA-2 whichpossessed physicochemical and immunological properties very similarto those of CEA were obtained from perchloric acid extracts of fecesfrom a normal adult (22) and pooled meconium (24, 27), respectively.NCA, known to be partially cross-reactive with CEA (28, 29), wasprepared from pooled lungs (30). Eight partially purified NFA-2 and10 partially purified NCA-2 preparations were obtained individuallyfrom normal adult feces and meconium, respectively, by combinationof affinity chromatography on an immunoadsorbent and gel filtrationon a Sepharose 6B column. The initials of individuals from whomsource materials were obtained were indicated in parentheses for therespective preparations. The concentrations of purified antigens wereestimated from the extinction coefficient at 280 nm, which was determined on the basis of the dry weight of each antigen after completedeionization and lyophilization, as described previously (3). The concentrations of partially purified antigens were determined by the CEA-Daiichi kit (Daiichi Radioisotope Laboratories, Tokyo, Japan) usingpolyclonal anti-CEA antibodies.

on August 9, 2021. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANTIGENIC HETEROGENEITY OF CEA IN CIRCULATION

Radioiodination of Antigens. Purified CEA, NFA-2, and NCA-2 werelabeled with '"I by the chloramine-T method (31).

Chemical and Enzymatic Treatment of Antigen Preparations. In orderto destroy the carbohydrate structure of antigens, purified antigenpreparations were treated with neuraminidase and then with periodateaccording essentially to the method of Coligan and Todd (6). Vibriocholerae neuraminidase (1 unit/ml; Behringwerke AG, West Germany)was added in a ratio of 1 milliunit to 60 fig of each antigen. Periodateoxidation was performed with sodium metaperiodate in a final concentration of 5.33 HIMat room temperature in the dark for 40 h. Pepsindigestion of the antigens was described previously (26).

Monoclonal Antibodies. Four monoclonal antibodies (F3-30, F4-82,F4-11, and F33-37) were prepared from mice immunized with purifiedCEA by the procedure described previously (32). Two other monoclonalantibodies, F8-52 and F48-60, were induced from mice immunized withpurified NFA-2 and NCA-2, respectively. Monoclonal antibodies werecollected as ascites and used after salting out by adding ammoniumsulfate to 40% saturation for IgG or 50% saturation for IgM, respectively.

Isotypes and Affinity Constants of Monoclonal Antibodies. The classand subclass of monoclonal antibodies were determined by microscaleimmunodiffusion with class- and subclass-specific antisera (32). Theaffinity constants (A.,) of monoclonal antibodies to the respective immunizing antigens were determined by the Fair assay as describedpreviously (32) and calculated according to the method of Steward andPetty (33).

RIAs. The reactivities of monoclonal antibodies with the antigenpreparations and patient sera were estimated by sandwich-type solid-phase RIAs consisting of the respective monoclonal antibodies immobilized on polystyrene beads and an I25l-labeled polyclonal goat anti-

CEA antibody which was affinity purified with a CEA adsorbent andthen absorbed thoroughly with NCA (9). For another test for themodified antigens, a competitive inhibition assay was performed asdescribed previously (32). Two commercially available RIAs for CEA,the CEA-Daiichi kit and the CEA-Roche kit (Hoffman La Roche,Nutley, NJ), both of which utilize polyclonal anti-CEA antibodies (9,34), were used for a comparison of the CEA values.

Serum Samples. Sera from patients with various malignant or non-malignant diseases and from healthy individuals were collected andstored at -25*C until use.

Affinity Chromatography. The IgG fraction obtained from a polyclonal goat anti-CEA antiserum or from a monoclonal antibody F4-11was coupled to CNBr-activated Sepharose 4B (Pharmacia Fine Chem-

Table 1 Isotypes and affinity constants (K.) of 6 monoclonal antibodies

CloneImmunogenHeavy

chainLightchain/f.

(X \(PM~')F3-30CEAX2.4F4-82CEAyìK25F33-37CEAMX12F4-1ICEATiK6.8F8-52NFA-2•y2b1C0.4F48-60NCA-2K1.3

icals Co.), respectively, according to the manufacturer's instruction. A

CEA sample was extracted from a patient serum with the goat anti-CEA adsorbent and then affinity chromatographed by a column of 1-411-coupled Sepharose 4B.

RESULTS

Isotype and Affinity Constant (A:,) of the Monoclonal Antibodies. The classes and subclasses of monoclonal antibodies usedand their affinity constants (K,) to the respective immunogensare shown in Table 1. A remarkable difference in the affinityconstants was observed among them.

Reactivities of Monoclonal Antibodies with Individual Preparations of CEA and Related Antigens. The reactivities of 6monoclonal antibodies with purified or partially purified antigen preparations are summarized in Table 2. The representativereaction profiles of 2 antibodies with several antigen preparations are depicted in Fig. 1. Antibodies F3-30 and F4-82 reactedstrongly with all the antigen preparations examined. F4-11reacted strongly with the immunizing antigen CEA(TY) andexceptionally with a few other CEA preparations, but with noneof the preparations of NFA-2 and NCA-2. F33-37 reacted withall the antigen preparations with diverse reaction intensity. IS52 showed strong reactions with almost all of the NFA-2 andNCA-2 preparations but weak, faint, or negative reactions withthe CEA preparations. F48-60 reacted strongly with all theNCA-2 preparations but showed weak or faint reactivity with afew CEA or NFA-2 preparations. None of the 6 antibodiestested reacted with NCA purified from normal lungs (data notshown).

Reactivities with Enzymatically and/or Chemically ModifiedAntigens. The reactivity of all the preparations of neuraminidase- and periodate-treated antigens with a polyclonal goatanti-CEA antibody decreased slightly (Fig. 2A) as comparedwith the respective native antigens, while that with F3-30 (Fig.2B) or F4-82 (not shown) increased slightly but definitely. Incontrast, the reactivity of F4-11, F33-37, F8-52, or F48-60 withthus treated respective immunizing antigens was reduced toabout 1% of that of the native antigen (Fig. 2, C and D). Incompetitive inhibition assays, the asialo antigens treated withneuraminidase alone revealed the same inhibition profiles asthe native antigens in all assays (Fig. 3), suggesting that sialicacids in nonreducing termini of carbohydrate chains seem topossess no significant roles in the antigenicity of those antigenmolecules. In the assay with F4-82 (Fig. 3A) or F3-30 (notshown), no competitions with the pepsin digest of CEA were

Table 2 Reactivities of 6 monoclonal antibodies with purified or partially purified antigen preparationsNo. of samples exhibiting the reactivitiesindicated"Of

10 CEAsamples*AntibodyPeptide"F3-30F-82Carbohydrate"F4-11F33-37F8-52F48-60++10102400+002361±000321-006028Of9 NFA-2samples*++980081+010610±000302-009006Of 11 NCA-2samples*++9110101011+200110±000000-0011000

°The reactivity was estimated by the sandwich-type solid-phase RIA. The degree of reactivity was roughly classified into 3 grades. ++, strong reaction similar tothat of all the antigen preparations depicted in Fig. \A and that of NFA-2 and NCA-2 in Fig. \B; •.weak reaction similar to that of CEA(TT), CEA(MI), andCEA(TY) in Fig. !/(;*. faint reaction similar to that of the others in Fig. 1//: —,negative.

* All of 10 CEA samples, 1 of 9 NFA-2 samples, and 1 of 11 NCA-2 samples were highly purified preparations; other samples of NFA-2 and NCA-2 were partially

purified preparations.' Chemical nature of antigenic epitopes recognized by the respective monoclonal antibodies.

57

on August 9, 2021. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANTIGENIC HETEROGENEITY OF CEA IN CIRCULATION

\<f

I'04

«DIO3

10

10s B

50 300

I loia^oÅ’>

(n 103

IO210 50 300

Antigen (ng/ml) Antigen (ng/ml)

Fig. 1. Reactivities of monoclonal antibody F3-30 (Ai or F8-52 (li) immobilized on beads with CEAs and related antigens in the sandwich-type solid-phaseRIA. An affinity-purified '"I-labeled polyclonal anti-CEA antibody was used asthe tracer antibody. B, gross cpm at a given concentration of each antigen; !!„.cpm of the control in the absence of the antigen. Antigen: O, CEA(TY); 9,CEA(WHO); A, CEA(MI); A, CEA(HO); O, CEA(TT); LA.CEA(MY); A, NFA-2(MaK); •NCA-2 (pool).

03 IO-

l 3 10 50 300 1000Antigen (ng/ml)

3 10 50 300 1000Antigen (ng/ml)

10s

, K3'

10s

IO21 3 10 50 30O 10OO

Antigen (ng/ml)

10»

"

CDIO3

.10* 3 10 50 300 1000Antigen (ng/ml)

Fig. 2. Reactivities of neuraminidase- and periodate-treated antigens in thesandwich-type solid phase RIA using the beads coated with a polyclonal goatanti-CEA antibody (A) or monoclonal antibodies F4-82 (B), F4-11 and F33-37(Q, and F8-52 and F48-60 (D). An affinity-purified I25l-labeled polyclonal anti-CEA antibody was used as the tracer antibody. Antigen: •,untreated CEA(TY);O, neuraminidase- and periodate-treated CEA(TY); A, untreated NFA-2(MaK);A, neuraminidase- and periodate-treated NFA-2(MaK); • untreated NCA-2(pool); D, neuraminidase- and periodate-treated NCA-2 (pool). In C, since thesame antigen preparations were tested with 2 different antibodies, untreatedCEA(TY) (•with F4-11 and T with F33-37) and neuraminidase- and periodate-treated CEA(TY) (O with F4-11 and V with F33-37) were separately marked,respectively.

observed; but in the assay with F4-11 (Fig. 35), F8-52 (Fig.3O, F48-60 (Fig. 3D), or F33-37 (not shown), the pepsin digestof respective antigens retained their reactivity with each antibody. These results indicate that the antigenic epitopes recognized by F3-30 and F4-82 are of protein nature and that thoserecognized by F4-11, F33-37, F8-52, and F48-60 reside on thecarbohydrate moiety of the antigen molecules, although in ourprevious paper (32) we reported that the epitope recognized byF4-11 seemed to be of protein nature.

Serum CEA Values Estimated by Monoclonal Antibodies.Serum CEA values for patients with various malignant diseasesestimated by RIAs using respective monoclonal antibodies arelisted in Table 3, in which the values estimated with the Rocheand Daiichi kits are also included. The values obtained withboth F3-30 and F4-82 correlated very well with those obtained

.49 16 32 64 86 256 512 »24 2046

Inhibitor Anttgen (ng)46 16 32 64 126 256 512 »24 2046

inhibitor Antigen (ng)

*I50

I

t

150

48 16 32 64 128 256 51210242046inhibitor Antigen (ng)

46 16 92 64 128 256 512 1024 2046inhibitor Antigen (ng)

Fig. 3. Competitive inhibition of enzymatically modified antigens for thereaction between monoclonal antibody and labeled antigen. In the reaction withantibodies F4-82 (A) and F4-11 (B), CEA(TY) was used as labeled antigen, andin the cases of F8-52 (Q and F48-60 (D), NFA-2(MaK) and NCA-2(pool) wereused as labeled antigen, respectively. Inhibitor antigen: •,untreated CEA(TY);O O, neuramin¡dase-treated CEA(TY); O- •-O, pepsin-treated CEA(TY); A.untreated NFA-2(MaK); A A, neuraminidase-treated NFA-2(MaK); A- •-A,pepsin-treated NFA-2(MaK); •untreated NCA-2(pool); D D, neuraminidase-treated NCA-2(pool); G •••D, pepsin-treated NCA-2(pool). To a constant amountof monoclonal antibodies, increasing amounts of inhibitors and 1 ng each of I29I-

labeled antigens were simultaneously added. After 4 h incubation, antigens boundto antibodies were precipitated with ammonium sulfate. The amount of monoclonal antibodies used was that capable of precipitating 40 to 50% of "'I-labeled

antigens in the absence of inhibitor unlabeled antigen.

with the Roche or Daiichi kit. The correlation coefficiencies (r)between the values obtained with the Daiichi kit and F3-30 andthat between the Daiichi kit and F4-82 were 0.97 and 0.99,respectively, and that between F3-30 and F4-82 was 0.98,although the difference in the affinity constant between theseantibodies was more than 1 order of magnitude (Table 1). Onthe other hand, the values with the other 4 antibodies recognizing the carbohydrate epitopes were quite variable from case tocase, and no correlation with the values by the Roche or Daiichikit was observed. Furthermore, some patient sera reacted withall the 4 antibodies; but some sera reacted with 1, 2, or 3 ofthem; and some other sera did not react with any of them atall. Although the reactivity of F4-11 with the purified antigensseemed to be quite different from that of F33-37 (Table 2), theserum values obtained with these 2 antibodies correlated wellwith each other (r = 0.94) (Table 3). There were poor correlations between any other combinations of values obtained withthe 4 anti-carbohydrate antibodies. Mean values ±SD of 20sera from normal individuals obtained with the Daiichi kit, F3-30, and F4-82 were 1.8 ±0.5, 3.0 ±1.1, and 6.5 ±1.5,respectively, but all values with 4 anti-carbohydrate antibodieswere less than 1.0 except a few cases with the value of 1.0 to2.0.

In order to examine the homogeneity of carbohydrate epitopes on the CEA molecules in a single patient serum, a CEApreparation extracted from a patient serum (Table 3, Patient 3)was passed through an F4-11 antibody adsorbent column. Theresult obtained is depicted in Fig. 4 and indicates that only one-half of the CEA molecules in this patient was reactive with F4-11. Thus, the antigenic structure of carbohydrate chains of theCEA molecule is heterogeneous even in a single patient serum.

Analysis of the Causes That Give Rise to the Different CEAValues between the Roche Kit and the Daiichi Kit. As can beseen in Table 3, a fairly large difference in the estimated CEAvalues existed between the Roche kit and the Daiichi kit,

on August 9, 2021. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANTIGENIC HETEROGENEITY OF CEA IN CIRCULATION

Table 3 Serum CEA values obtained by 2 polyclonal antibodies and 6 monoclonal antibodies in patients with various cancers'

Kit(ng/ml)Patient12345678910111213141516171819202122232425262728293031323334353637383940414243444546DiagnosisColon

cancerColoncancerColoncancerColoncancerColoncancerColoncancerColoncancerColoncancerColoncancerColoncancerColoncancerColoncancerGastric

cancerGastriccancerGastriccancerGastriccancerGastriccancerGastriccancerGastriccancerGastriccancerGastriccancerGastriccancerGastriccancerGastriccancerLung

cancerLungcancerLungcancerLungcancerLungcancerLungcancerLungcancerLungcancerLungcancerBreast

cancerBreastcancerBreastcancerBreastcancerOvarian

cancerOvariancancerUterinecancerUterinecancerPancreaticcancerPancreaticcancerGallbladdercancerGallbladdercancerBladder

cancerRoche>2,500>2,500>2,500>2,5002,340>2,500>2,5006505405002256007,6007,700>2,500>2,5001,2506309354152801153512>2,5002,000>2,500800701,2203202203601,1306001812uoo1113353239>2,5001%860Daiichi5,3003,1003,1482,9501,2801,0406133943743082361827,2483,3411,0809407504202922611955520242,0501,6501,3001,2073032202147232650302541,0003669272171,180421453Monoclonal

antibodies(ng/ml)Peptide*F3-3014,41813,6524,8365,4304,5492,8632,1181,35093259155385013,1969,7305,3051,6962,6551,250773316418190125376,4714,5525,6592,5354257048631961491,41210352124,033881705669223,3723071,591F4-8213,23213,1704,0293,9513,2052,4551,3101,17985796450571311,7156,1254,3542,1842,0338501,19740629511765436,5132,6442,8687,4422855545051211361,4794937112,753861209436172,7977591,948F4-1127022164112<13<1<111991242<1<13<1<1<122,433178112<1412455<1<1<18<1<1<1121641Carbohydrate*F33-37795238613<17<1<14<13<11639<116<1<1<114<1<1<121,88676<111343082288<1<1<113<1<1<1<1126141<1F8-528001,709171,216194425224286432778,5077356727422222<1262968<121753<12<17<11<12,600<11426916814F48-6012223<11<17<1<13113632<1951<1<1<12184<1<1<14<1<1<15<1<1<12<1<1<1<1212<1<1

" Determined by the commercially available RIAs (the CEA-Daiichi kit or the CEA-Roche kit) or the sandwich-type RIAs consisted of polystyrene beads coatedwith the respective monoclonal antibodies and an affinity-purified '"¡-labeledpolyclonal anti-CEA antibody.

* Chemical nature of antigenic determinant of the CEA molecule recognized by the respective monoclonal antibodies.

although both values were correlated well with each other. Toelucidate the possibility that the variations in reactivity withcarbohydrate epitopes of the CEA molecule may result in thedifferent CEA values between the Roche and Daiichi kits, thefollowing analysis was performed. The CEA values of a largenumber of patient sera (770 samples) which contained morethan 50 ng of ( T A/ml estimated by the Roche kit were alsoestimated by the Daiichi kit. From these samples, the following3 extreme groups were selected (Table 4). Any sera of Group 1(101 samples) revealed much higher values for the Roche kitthan those for the Daiichi kit (more than 5 times). The Roche:Daiichi ratios of Group 2 (26 samples) fell between 1.0 and1.25. All the sera of Group 3 (28 samples) revealed lower valuesfor the Roche kit than for the Daiichi kit. The reactivities ofsera in each group with 3 anti-carbohydrate antibodies F33-37,F8-52, and F48-60 were examined by RIA. From the resultssummarized in Table 4, it is evident that the antigenic epitoperecognized by F33-37 is poorly expressed on the CEA moleculein the Group 1 samples but strongly expressed on those in the

Group 3 samples, and the epitope recognized by F8-52 isexpressed almost evenly on the CEA molecule among 3 groups,but the epitope recognized by F48-60 is scarcely in all groups.

DISCUSSION

To improve the clinical usefulness of the CEA assay, it seemsof great significance to make clear the molecular basis of theheterogeneity of the circulating CEA molecule, which has beensuggested by a few studies (19, 32). In the present study, wetried to analyze the antigenic heterogeneity of CEA in patientsera and to elucidate the causes that give rise to the variationsin CEA assays by using the monoclonal antibodies reactive witheither peptide or carbohydrate epitopes of the CEA molecule.

The peptide epitopes recognized by F3-30 and F4-82, whichhad been proved to be on different antigenic parts of the CEAmolecule, respectively (32), seem to be quite universally expressed on all the CEA molecule, because all CEA preparationsor patient sera tested reacted well with them (Tables 2 and 3).

59

on August 9, 2021. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANTIGENIC HETEROGENEITY OF CEA IN CIRCULATION

100

E

<LUO

Glycine-MCIBuffer. pH 2.3

. .10 21 25

Tube Number

30

Fig. 4. Elution profile of a CEA preparation (220 ng) from an adsorbentcolumn (O.S x 7 cm) of Sepharose 4B to which about 5 mg of F4-1 1 antibodywere bound. The CEA preparation was extracted from O.S ml of patient serum(Table 3, Patient 3) with an adsorbent of goat anti-CEA and then applied to thecolumn, and the antigen adsorbed to the column was eluted with 0.175 M glycine-HCI buffer, pH 2.3 (arrow). The effluent was collected in 1-ml portions. O, CEAvalue determined by the Daiichi kit using polyclonal anti-CEA antibodies; •.CEA values measured by a sandwich-type solid-phase RIA consisting of the beadscoated with F4-11 and an affinity-purified '"I-labeled polyclonal anti-CEA anti

body.

Table 4 Expression of carbohydrate epitopes on the CEA molecules in patientsera showing extremely different Roche.-Daiichi kit ratios

Group 1* Group 2 Group3No.

Roche-.Daiichi"Assay

systemRoche

DaiichiF33-37

F8-52F48-60101

26 28>S.O 1.25-1.0<1.0Mean

±SD(ng/ml)*2,582

±12,716 7,903 ±10,129 6,102 ±9,663334 ±1,277 6,978 ±8,995 7,910 ±10,4927

±14 151 ±222 202 ±271105 ±195 396 ±338 190 ±267

7±4 7±135±7Mean±SD/ 1,000 ng with Rochekit''F33-37

F8-S2F48-60Range

of F33-37/1,000 ng with

Rochekit'<1.0

1.0-9.910.0-99.9>100.08.5

±18.3 62.9 ±151.9 212.0 ±547.0144.4 ±355.2 141.8 ±188.9 101.3 ±134.5

10.2 ±27.9 2.0 ±3.3 7.4±18.1No.

of samples (%) revealing values in listed ranges'*26(25.7)

3(11.5) 4(14.3)56(55.4) 11(42.3) 6(21.3)17(16.8) 9(34.6) 11(39.3)2(2.0) 3(11.5) 7(25.0)

" From 770 sera containing more than 50 ng of (l'Ami estimated by the

Roche kit, the extreme groups of samples (155 sera in total) were selected on thebasis of RocherDaiichi ratios. The 615 sera that were not used in this analysisrevealed the ratios between 1.25 and 5.0.

* The values listed were obtained by the Roche kit, the Daiichi kit, and solid-phase sandwich RIAs using each of monoclonal antibodies (F33-37, F8-52, andF48-60) immobilized on beads and an affinity-purified '"I-labeled polyclonalanti-CEA antibody.

' The values of each sample obtained with monoclonal antibodies were nor

malized with 1,000 ng of Roche value and the mean values ±SD of thusnormalized values were listed below because there was a large difference in CEAcontents among groups.

" The values obtained with F33-37 antibody were normalized with 1,000 ng of

Roche value and divided into 4 ranges listed and the sample numbers whichshowed values in each range were counted. The percentage of numbers countedin each group was listed in parentheses.

However, the reactivity of these antibodies with the antigenpreparations was not completely uniform but slightly variable(Table 2; Fig. \A). This may result from the difference in thedegree or mode of masking of the peptide epitopes by carbohydrate chains as suggested by the results in Fig. IB. Since afew suggestive evidences for the difference in the peptide struc

ture of the CEA molecule have also been reported previously(14, 15, 35), neither possibility can be ruled out at present.

The CEA values estimated with F3-30 or F4-82 correlatedvery well to each other and to those obtained with the Rocheor the Daiichi kit (r> 0.97) (Table 3) and were generally higherthan those obtained with the Daiichi kit, although the sametracer of an '"I-labeled specific polyclonal goat anti-CEA an

tibody was used in these assays. As possible causes for this,undefined factors such as variations in affinity constant (K,) orcompetition between immobilized antibodies and the tracerantibody for antigen epitopes should be taken into consideration.

In contrast to the universal reactivity of antibodies to thepeptide epitopes, the reactivity of 4 antibodies directed to thecarbohydrate epitopes with the purified antigen preparationswas found to be quite heterogeneous or preferential to someantigen groups. Since F8-52 was reactive with all NFA-2 andNCA-2 preparations and F48-60 was almost exclusively reactive with all NCA-2 preparations from meconium, the epitopesrecognized by them must not be individual specific and thelatter antibody would recognize an epitope expressed specifically for fetal period. In fact, the epitope recognized by F48-60was expressed very poorly with few exceptions in patient sera(Table 3) or in normal sera. The reactivities of these 4 antibodieswith CEAs in sera from patients with various malignant diseaseswere surprisingly irregular (Table 3), indicating extensive variations in structure of carbohydrate chains of CEA molecules inpatient sera, even in a single patient serum (Fig. 4). No organ-specific expressions of carbohydrate epitopes were observedamong malignancies from various organ origins thus far tested(Table 3). Well correlated CEA values (r = 0.94) obtained withF4-11 and F33-37 suggest that they recognized a similar epitope

on carbohydrate chains, and an apparently wider reactivity ofF33-37 than that of F4-11 may be related to the IgM class ofF33-37.

The values obtained with anti-carbohydrate antibodies weregenerally very low as compared with those obtained with otherantibodies. This does not directly mean that a small portion ofthe CEA molecules possessed the corresponding epitope, because the factors lowering the values such as the low antigen-antibody affinity or some inhibitions by cross-reactive carbohydrate epitopes other than those on CEA molecules should betaken into consideration. In any case, the CEA values obtained,regardless of the assay system used, do not express the absoluteconcentration of CEA but do express the relative values compared with the standard of each assay system (2, 9, 36, 37).

It is probable that the polyclonal anti-CEA antibody preparations contained a small amount of antibodies reactive withcarbohydrate epitopes on the CEA molecule as can be deducedfrom the results in Fig. 2A. The results in Table 4 suggest thatthe antibody preparation of the Daiichi kit may contain antibody reactive with the carbohydrate epitope recognized by F33-37, whereas the Roche kit is scarcely reactive with it. Conversely, the Roche kit may be preferentially reactive with anepitope expressed more on the CEA molecules in sera of Group1 than on those in sera of Group 3, and then much highervalues of the Roche kit were observed in Group 1 (Table 4),although we did not specify any competent epitope in thisstudy. Thus, it is evident that the diversity of antigenic expression on the carbohydrate chains of the CEA molecule and thevariation in specificity and quantity of anti-carbohydrate antibodies in the antibody preparations used for the respectiveassay systems are responsible, at least in part, for the differencesin the CEA values among assay systems.

60

on August 9, 2021. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANT1GENIC HETEROGENEITY OF CEA IN CIRCULATION

Although there remains a possibility that cancer specificitymay reside on the very diverse carbohydrate chains of CEAmolecule, in order to improve the accuracy and the constancyof CEA assays, assay systems which are independent from thecarbohydrate epitopes should be established; such assay systemsare now under investigation using other monoclonal antibodies.

ACKNOWLEDGMENTS

The authors are grateful to Dr. A. Horie, School of Medicine,University of Occupational and Environmental Health, Kitakyushu,Japan; we are also grateful to Drs. M. KJkuchi and K. Shirakawa,School of Medicine, Fukuoka University, Fukuoka, Japan, for theirkind supply of cancerous and noncancerous materials. We extend ourthanks to WHO for kindly supplying the international standard preparation of CEA; to Dr. K. Nakajima, Otsuka Assay Laboratory, To-kushima, Japan, for supplying patient sera; and to Y. Hashimoto forsecretarial assistance.

REFERENCES

1. 0rjasaeter, H. Determination of carcinoembryonic antigen (CEA): comparison of two different assays. Acta Pathol. Microbio!. Scand. Sect. C, 86:265-275, 1978.

2. Wagener, C., and Breuer, H. Comparative studies on the radioimmunologicaldetermination of carcinoembryonic antigen in tumour tissue. J. Clin. Chem.Clin. Biochem., 16: 323-327, 1978.

3. Kuroki, M., Yamaguchi, A., Koga, Y., and Matsuoka, Y. Antigenic reactivities of purified preparations of carcinoembryonic antigen (CEA) and relatednormal antigens using four different radioimmunoassay systems for CEA. J.Immunol. Methods, 60: 221-233, 1983.

4. Krupey, J., Gold, P., and Freedman, S. O. Physicochemical studies of thecarcinoembryonic antigens of the human digestive system. J. Exp. Med., 128:387-398, 1968.

5. Westwood, J. H., Bessell, E. M., Bukhari, M. A., Thomas, P., and Walker,J. M. Studies of the structure of the carcinoembryonic antigen—I. Somedeductions on the basis of chemical degradations. Immunochemistry, 11:811-818, 1974.

6. Coligan, J. E., and Todd, C. W. Structural studies on carcinoembryonicantigen periodate oxidation. Biochemistry, 14: 805-810, 1975.

7. Hammarström, S., Engvall, E., Johansson, B. G., Svensson, S., Sundblad,G., and Goldstein, I. J. Nature of the tumor-associated determinant(s) ofcarcinoembryonic antigen. Proc. Nati. Acad. Sci. USA, 72:1528-1532,1975.

8. Vrba, R., Alpert, E., and Isselbacher, K. J. Carcinoembryonic antigen:evidence for multiple antigenic determinants and isoantigens. Proc. Nati.Acad. Sci. USA, 72: 4602-4606, 1975.

9. Matsuoka, Y., Kuroki, M., Koga, Y., Ogawa, H., Nakazawa, V. Tachibana,S., and Minamizawa, T. A new direct solid-phase radioimmunoassay forcarcinoembryonic antigen without pretreatment of serum samples. J. Immunol. Methods, 58: 31-47, 1983.

10. Köhler,G., and Milstein, C. Continuous cultures of fused cells secretingantibody of predefined specificity. Nature (Lond.), 256:495-497, 1975.

11. Kupchik, H. Z., Zurawski, V. R., Hurrell, J. G. R., Zamcheck, N.. and Black,P. H. Monoclonal antibodies to carcinoembryonic antigen produced bysomatic cell fusion. Cancer Res.. 41: 3306-3310, 1981.

12. Primus, F. J., Newell, K. D., Blue, A., and Goldenberg, D. M. Immunologia!heterogeneity of carcinoembryonic antigen: antigenic determinants on carcinoembryonic antigen distinguished by monoclonal antibodies. Cancer Res.,Â¥5:686-692,1983.

13. Wagener, C., Yang, Y. H. J., Crawford, F. G., and Shively, J. E. Monoclonalantibodies for carcinoembryonic antigen and related antigens as a modelsystem: a systematic approach for the determination of epitope specificitiesof monoclonal antibodies. J. Immunol., 130: 2308-2315, 1983.

14. Blaszczyk, M.. Pak, K. Y., Herlyn, M., Lindgren, J., Pessano, S., Steplewski,Z., and Koprowski, H. Characterization of gastrointestinal tumor-associatedcarcinoembryonic antigen-related antigens defined by monoclonal antibodies.Cancer Res., 44: 245-253, 1984.

15. Imai, K., Moriya, Y., Fujita, H., Tsujisaki, M., Kawaharada, M.. and Yachi,A. Immunologie characterization and molecular profile of carcinoembryonic

antigen detected by monoclonal antibodies. J. Immunol., 132: 2992-2997,1984.

16. Grünen,F., AhuI lar IciI. N., Schwarz, K., and von Kleist, S. Two CEA andthree NCA species, although distinguishable by monoclonal antibodies, havenearly identical peptide patterns. Int. J. Cancer, 36: 357-362,1985.

17. Nichols, E. J., Kannagi, R., Hakomori, S.-L, Kram/, M. J., and Fuks, A.Carbohydrate determinants associated with carcinoembryonic antigen (CEA).J. Immunol., 135:1911-1913, 1985.

18. Murani. 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 carcinoma irr.vuv normal adult tissues. Cancer Res., 45: 5769-5780,1985.

19. Rogers, G. T., Rawlings, G. A., Keep, P. A., Cooper, E. H., and Bagshawe,K. D. Application of monoclonal antibodies to purified CEA in clinicalradioimmunoassay of human serum. Br. J. Cancer, 44: 371-380, 1981.

20. Neumaier, M., Fenger, I .. and Wagener, C. Delineation of four carcinoembryonic antigen (CEA) related antigens in normal plasma by transblot studiesusing monoclonal anti-CEA antibodies with different epitope specificities.Mol. Immunol., 22:1273-1277,1985.

21. Matsuoka, Y., Hará,M., Takatsu, K., and Kitagawa, M. Presence of antigenrelated to the carcinoembryonic antigen in feces of normal adults. Gann, 64:203-206, 1973.

22. Kuroki, M., Koga, Y., and Matsuoka, Y. Purification and characterizationof carcinoembryonic antigen-related antigens in normal adult feces. CancerRes., 41: 713-720, 1981.

23. Burtin. P., Chavanel, G., and Hirsch-Marie, H. Characterization of a secondnormal antigen that cross-reacts with CEA. J. Immunol., Ill: 1926-1928,1973.

24. Matsuoka, Y., Kuroki, M., Koga, Y., Kuriyama, H., Mori, T., and Kosaki,G. Immunochemical differences among carcinoembryonic antigen in tumortissues and related antigens in meconium and adult feces. Cancer Res., 42:2012-2018, 1982.

25. Matsuoka, Y., Tsuru, E., and Sawada, H. Preparation and evaluation ofantisera directed against cancer specific moiety of antigenic determinants oncarcinoembryonic antigen. Immunochemistry, 12: 779-782, 1975.

26. Matsuoka, Y., Koga, Y., Manila. H., Yoshino, M., and Tsuru, E. Proteolyticrelease of antigenic fragments corresponding to normal fecal antigen andnon-specific cross-reacting antigen from carcinoembryonic antigen. Int. J.Cancer, 21:604-610, 1978.

27. Koga, Y., Kuroki, M., Matsunaga, A., Shinoda, T., Takayasu, T., andMatsuoka, Y. Further comparative studies on chemical properties of carcinoembryonic antigen in tumor tissues and closely related antigens in adultfeces and meconium. Mol. Immunol., 22:67-73, 1985.

28. von Kleist, S., Chavanel, G., and Burtin, P. Identification of an antigen fromnormal human tissue that crossreacts with carcinoembryonic antigen. Proc.Nati. Acad. Sci. USA, 69: 2492-2494, 1972.

29. Mach, .1.1'.. and Pusztaszeri, G. Carcinoembryonic antigen (CEA): demonstration of a partial identity between CEA and a normal glycoprotein.Immunochemistry, 9: 1031-1034, 1972.

30. Kuroki, M., Ichiki, S., Kuroki, M., and Matsuoka, Y. Coproduction ofcarcinoembryonic antigen and nonspecific cross-reacting antigen by a continuous cell line from a human pancreatic tumor. J. Nati. Cancer Inst., 69:401-408, 1982.

31. Greenwood, F. C., and Hunter, W. M. The preparation of '"I-labelled humangrowth hormone of high specific radioactivity. Biochem. J., 89: 114-123,1963.

32. Kuroki, M., Kuroki, M., Koga, Y., and Matsuoka, Y. Monoclonal antibodiesto carcinoembryonic antigen: a systematic analysis of antibody specificitiesby using related normal antigens and evidence for allotypic determinants oncarcinoembryonic antigen. J. Immunol., 133: 2090-2097, 1984.

33. Steward, M. W., and Petty, R. E. The use of ammonium sulphate globulinprecipitation for determination of affinity of anti-protein antibodies in mouseserum. Immunology, 22: 747-756, 1980.

34. Lo Gerfo, P., Krupey, J., and Hansen, H. J. Demonstration of an antigencommon to several varieties of neoplasia. Assay using zirconyl phosphategel. N. Engl. J. Med., 285:138-141, 1971.

35. Kuroki, M., Kuroki, M., Ichiki, S., and Matsuoka, Y. Identification andpartial characterization of the unglycosylated peptide of carcinoembryonicantigen synthesized by human tumor cell lines in the presence of tunicamycin.Mol. Immunol., 21: 743-746, 1984.

36. Persijn, J.-P., and Korsten, C. B. The development of a radioimmuno-assayfor carcino-embryonic antigen with some applications. Clinical evaluation ofcarcino-embryonic antigen, I. J. Clin. Chem. Clin. Biochem., 14: 377-387,1976.

37. Sizaret, P., and Estève,J. Relative potencies of four CEA preparationsassayed by three commercial radioimmunoassay kits. J. Immunol. Methods,34:79-89, 1980.

61

on August 9, 2021. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1987;47:56-61. Cancer Res   Akira Matsunaga, Masahide Kuroki, Hiroshi Higuchi, et al.   Related AntigensCarbohydrate Moiety of Carcinoembryonic Antigen and CloselyCirculation Defined by Monoclonal Antibodies against the Antigenic Heterogeneity of Carcinogembryonic Antigen in the

  Updated version

  http://cancerres.aacrjournals.org/content/47/1/56

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/47/1/56To request permission to re-use all or part of this article, use this link

on August 9, 2021. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from