Sandwich enzyme immunoassay using three monoclonal antibodies against different epitopes of carcinoembryonic antigen (CEA)

Immunology Letters, 5 (1982) 85-91

Elsevier Biomedical Press

SANDWICH ENZYME IMMUNOASSAY USING THREE MONOCLONAL ANTIBODIES A G A I N S T D I F F E R E N T EPITOPES OF C A R C I N O E M B R Y O N I C A N T I G E N (CEA) F. BUCHEGGER, C. METTRAUX, R. S. ACCOLLA, S. CARREL and J.-P. MACH Institute of Biochemistry, University of Lausanne and Ludwig Institute for Cancer Research, Lausanne Branch, 1066 Epalinges, Switzerland

(Received 25 May 1982) (Accepted 16 June 1982)

1. S ~ m ~ Purified monoclonal antibodies (Mab) produced by 3 hybridomas and reacting with 3 different epitopes of carcinoembryonic antigen (CEA) were used in a solid phase enzyme immunoassay. Two Mabs were physically adsorbed to polystyrene balls and the third Mab was coupled to alkaline phosphatase using the bifunctional reagent N-succinimidyl-3-(2-pyridyldithio)propionate. During a first incubation, CEA from heatextracted serum samples was immunoadsorbed to the antibody coated balls. After washing of the balls, bound CEA was detected by a second incubation with the enzyme coupled Mab. The sensitivity of the assay was 0.6 ng per ml of serum. A total of 196 serum samples from patients with various types of carcinoma, with liver cirrhosis, or from healthy blood donors with or without smoking habits, were tested. The results obtained with the monoclonal enzyme immunoassay (M-EIA) were compared with those obtained with perchloric acid extracts of the same serum samples tested by an inhibition radioimmunoassay using conventional goat anti-CEA antiserum. There was an excellent correlation between the two assays. In particular, the new M-EIA gave good results for the detection of tumor recurrences in the follow-up of colon carcinoma patients. However, despite the use of exclusively monoclonal antibodies the new assay detected a similar percentage of slightly elevated CEA values as the conventional assay in patients with nonKey words: carcinoembryonicantigen - monoclonal antibodies - enzyme immunoassay - colon carcinoma

malignant disease, suggesting that the CEA associated with non-malignant diseases is immunologically identical to the CEA released by colon carcinoma.

2. Introduction The different assays for measuring CEA in human sera have been the subject of much controversy. The first CEA assay was developed by the group of Gold as an inhibition radioimmunoassay (RIA) performed on perchloric acid extracts of serum. It was initially thought to detect elevated CEA levels exclusively in patients with digestive tract carcinomas [ 1]. A modification of this assay reported by Hansen [2] has become the most commonly used CEA test. This assay also detected elevated CEA levels in patients with carcinomas derived from non-digestive tract organs, as well as in patients with non-malignant inflammatory diseases and in healthy individuals with heavy smoking habits [2,3]. In parallel, several inhibition RIAs utilizing unextracted serum were developed [4,5]. They gave slightly higher CEA values in normal serum samples. The idea of extracting serum by heat was first proposed by MacSween et al. [6], and was also used by Hirai [7] who developed the first sandwich solid phase RIA for CEA. A convenient solid phase enzyme immunoassay (EIA) was developed on the same principle and used by Majolini et al. [8]. We described the first EIA using a single monoclonal antiCEA antibody and polyclonal goat anti-CEA antibodies [9]. The quality of this assay was satisfactory but it also detected CEA in non-malignant diseases, as has

0165-2478/82/0000-0000/$2.75 © 1982 Elsevier Biomedical Press

85

been the case for all previously mentioned CEA assays. The idea of developing a sandwich assay which uses exclusively monoclonal antibodies is attractive because it involves only reagents of perfect homogeneity, unlimited availability, and well defined specificity. The purpose of this report is to describe such an EIA using 3 monoclonal antibodies (M-EIA) reacting with 3 different epitopes of the CEA molecule and to compare its clinical value with a conventional RIA for CEA.

3. Materials and methods

3.1. Monoclonal anti-CEA antibodies Hybridomas secreting monoclonal antibodies (Mabs) against CEA [10] were obtained by fusion of spleen cells from CEA-immunized mice with cells from the myeloma line P3-NSI/1-Ag4 [11]. Mabs from 3 hybridomas were selected for the assay by the following criteria: (1) capacity of binding more than 60% of J2Sl-labeled purified CEA as determined in a soluble phase RIA; (2) specificity for CEA as determined in an inhibition RIA using previously described methods [10]. It was required that more than 2000 times higher amounts of normal lung glycoprotein (NGP or NCA) [12,13] were necessary to give the same binding inhibition of the Mabs to radiolabeled CEA as purified CEA; (3) reactivity with different epitopes on the CEA molecule, as shown by a reciprocal inhibition assay using immobilized CEA. In this test the binding of one radiolabeled Mab was not inhibited by an excess of the other two Mabs. Mabs were purified from ascitic fluid obtained from Pristane-primed mice injected intraperitoneally with 10-20 × 10 6 cloned hybrid cells. Ascitic fluid was precipitated with ammonium sulfate at 45% saturation at 4°C and Mabs were purified by DEAE 52 cellulose (Whatman, Balston Ltd., U.K) ion exchange chromatography. Thc Mabs were eluted by a gradient of slowly increasing molarity of phosphate buffer pH 8, starting with a molarity of O.O1 [14]. 3.2. Coupling o f alkaline phosphatase to anti-CEA Mab Coupling of purified Mab VII 23 with alkaline phosphatase was achieved with limited loss of antibody activity using the bifunctional reagent N-succinimidyl-3-(2-pyridyldithio)-propionate (SPDP, Pharmacia, Uppsala) [ 15]. Briefly, 5 mg of alkaline 86

phosphatase (Sigma, type VII, specific activity 900 U/mg, St. Louis, MO) were incubated with an 8 M excess of SPDP for 30 min at room temperature in a solution of 0.1 M NaCI and 0.1 M phosphate, pH 7.5 (phosphate-NaCl). Separately, 2 mg of purified Mab VII 23 was incubated in the same way with a 16 M excess of SPDP. Free SPDP was then removed from both the enzyme and Mab by a Sephadex G-25 filtration on 2 columns equilibrated with a solution of 0.1 M NaCI and 0.1 M acetate, pH 4.5, for phosphatase and with phosphate-NaCI, pH 7.5, for the Mab. It was calculated that, by this procedure, 2.6 M pyridyldisulfide residues were linked per molecule of phosphatase and 6.6 residues per molecule of Mab. Pyridyldisulfide residues on the phosphatase were then reduced by a 30 min incubation at room temperature with dithiothreitol (Merck, Darmstadt) at a final concentration of 50 raM. The phosphatase with activated thiol groups was rechromatographed on a Sephadex G-25 column equilibrated in phosphate-NaCl buffer, pH 7.5. The enzyme and Mab were then mixed and incubated for 6 h at room temperature. They were concentrated to 1 ml and chromatographed on a Sepharose 6B column equilibrated in PBS pH 7.4. The eluted fractions were analyzed for antibody activity with soluble ~2Sl-labeled CEA, for phosphatase activity with p-nitrophenylphosphate and for coupled enzyme-antibody conjugates by incubation with immobilized CEA and measurement of bound enzymatic activity. The most active enzyme-antibody conjugates were eluted from Sepharose 6B column with fractions having a molecular weight of 250,000500,000. These fractions were pooled and stored at -20°C or lyophilized and they retained full activity after 1 year.

3.3. Preparation o f Mab-coated balls Mabs Xlll 73 and XIII 47 from a recent fusion (previously unreported)were purified from hybridoma ascitic fluid and used mixed in equal part for coating 6.5 mm diameter polystyrene balls (Precision Plastic Ball Co., Chicago, IL). Batches of 5 0 - 3 0 0 balls were incubated for 2 h at room temperature in phosphatebuffered saline (PBS) (0.14 M NaC1 + 0.01 M phosphate, pH 7.4) containing 5/ag per ml purified Mabs. Testing of the remaining antibody in the PBS after this incubation showed that 90% of the antibody was bound to the balls, representing about 0.5 #g of anti-

body adsorbed per ball. The balls were then saturated by a further incubation for 2 h with PBS containing 1% bovine serum albumin (BSA) and were stored in PBS with 0.1% BSA (PBS-BSA) and 0.02% sodium azide. Balls were washed once before use with PBS-BSA. 3.4. Conventional inhibition RIA All sera tested in M-EIA had been previously assayed by a conventional inhibition RIA using a goat anti-CEA antiserum as originally described by Thomson et al. [I] and modified by Mach et al. [16]. For this method duplicate 1 ml serum samples were extracted with 1 M perchloric acid, dialyzed against deionized water, lyophilized and tested in RIA. 3.5. Selection o f serum samples Sera already tested by RIA and stored at -70°C were selected for comparison with M-EIA and classified as follows: group A, 63 sera from colorectal carcinoma patients with a range of CEA concentrations from 0 to 80 ng CEA per ml (as determined by RIA); group B, 18 sera from patients with cacinomas of the digestive system not derived from colon or rectum (9 pancreas, 4 gall bladder, 5 esophagus with elevations of serum CEA concentrations ranging from 4 to 80 ng per ml); group C, 42 sera from patients with breast (33) or lung (9) carcinoma with elevated CEA concentrations; group D, 45 sera from patients with liver cirrhosis (selected for having elevated CEA concentrations) and 32 sera from healthy blood donors with a history of smoking; group E, 28 sera from unselected healthy blood donors without smoking habits. 3.6. Standard curve and assay procedure A standard curve was prepared using as diluent large batches (2 liters) of normal bovine serum which had been enriched with 1 ng CEA per ml of serum in order to obtain the same background value as with the lowest normal human sera. Samples of bovine serum were then enriched with standard amounts (0.6-160 ng per ml) of CEA purified from liver metastases from colon carcinoma which had been shown to have the same antigenic activity as the British CEA Standard 73/601 [ 17]. Duplicates of 0.3 ml serum samples were diluted with 0.7 ml of 0.2 M acetate buffer, pH 5 and heated at 70°C for 15 rain. After centrifugation at 5000 X g

for 5 min, 0.6 ml of the supernatant was incubated overnight at 25°C with a single polystyrene ball coated with Mab in disposable plastic tubes of 5 ml capacity. Balls were then washed once with 2 ml of PBS-BSA and once with 0.02 M Tris-HC1, pH 7.4, containing 1 mg of BSA per ml (Tris-BSA). Balls were then incubated for 75 min at 37°C with alkaline phosphatase-coupled Mab at a dilution selected to give an optical density of 1 in the test for a concentration of 10 ng of CEA per ml of serum. This incubation was performed in Tris-BSA containing 1% normal mouse serum which had been previously heated for 30 min at 56°C to destroy alkaline phosphatase activity. The addition of heated normal mouse serum slightly decreased the background enzymatic activity non-specifically bound to the balls. After 3 washings with Tris-BSA, the enzymatic activity bound to the balls was revealed by a 2 h incubation at 37°C with 0.5 mg ofp-nitrophenylphosphate diluted in 0.5 ml diethanolamine buffer (10% diethanolamine, MgC12 0.5 mM, pH 9.8). The reaction was stopped with 0.1 ml of 3 M NaOH and the optical density read at 405 nm against the substrate in diethanolamine buffer to which 20% NaOH had been added.

4. Results 4.1. Standard curves A standard curve obtained with normal bovine serum enriched with standard quantities of CEA is shown in Fig. 1 (full line). The sensitivity of the M-EIA was 0.6 ng per ml serum. The second curve (dashed line) was obtained by measuring CEA in serial dilutions (1:6 to 1:384) of the serum from a patient with colon carcinoma which contained 800 ng CEA per ml as determined by the inhibition RIA. The complete parallelism observed indicated that our standard CEA has the same antigenicity as the patient's circulating CEA for the 3 Mab involved in the M-EIA. 4.2. Intra-assay and inter-assay variations Table 1 shows the results of an intra- and interassay variation test. Batches of normal bovine serum enriched with 3.5, 10 and 36 ng CEA per ml were analyzed. Twenty analyses from each concentration performed in 1 experiment resulted in standard deviations in the range of 3.5 to 7.1% of the mean value 87

/

2.6-

/

2.2-

E

0

g

/

1.81

qr >. (n zi&l ci

(= coefficient of variation). The same batches analyzed in duplicates on 20 different days gave coefficients of variation of about 10%.

1.4

0.60.20

6

oiel.is21s g -1'o 2.0 4'0 8'0 leo ng CEA PER

ml SERUM

3~;4 192 9'6 48 2~ 4'2 ~, REC,P.OCA" SERUM O"tmO.S Fig. 1. Standard curves of M-EIA obtained either with normal bovine serum enriched with CEA purified from metastases of a colon carcinoma (x - - -x) or with dilutions of the serum from a patient having a cecum carcinoma and 800 ng CEA/ml serum ( c - - - o ) . Optical densities at 405 nm higher than 1.6 were read at 1:2 dilutions.

Table 1 Variation of monoclonal enzyme immunoassay results at different CEA levels lntra-assay variation mean value in ng/ml standaid deviation coefficient of variation (%) number of tests (monoplicates) Inter-assay variation mean value in ng/ml standard deviation coefficient of variation (%) number of tests (duplicates) on different days

88

3.5 0.25 7.1 20

9.7 0.34 3.5 20

36.5 1.68 4.6 20

3.5 0.33 9.4

10.5 1.08 10.2

36.9 3.68 10.0

20

20

20

4.3. Comparison o f theM-EIA with the inhibition RIA A comparison of the results obtained with the conventional inhibition RIA (abscissa)with those obtained with the new M-EIA (ordinate) is presented in Fig. 2. Panel A shows the results obtained in the sera from 63 patients with colorectal carcinomas. The correlation coefficient between the two assays is r = 0.95 (r was calculated for the real CEA values in ng per rnl as measured in the two tests, whereas the regression lines shown in the figure were calculated for the logarithmic values utilized for presentation). Panel B shows the results obtained with the sera from 18 patients with carcinomas of digestive organs other than colon or rectum (pancreas, gall bladder, esophagus). The correlation coefficient is also r = 0.95. Panel C shows the results obtained with sera from 42 patients with breast or lung carcinomas. The correlation coefficient is r = 0.93. Panel D shows the results obtained in the sera of a group of individuals with no evidence of malignant disease. It includes the sera from 32 healthy blood donors with smoking habits and from 13 patients with liver cirrhosis. The correlation coefficient in this group is r = 0.88. No significant difference between the two tests was observed in the 4 groups, indicating that the M-EIA with the 3 Mabs and the conventional RIA with the polyclonal antiserum recognized the same CEA molecules in the 4 groups of patients tested. The results obtained with sera from 28 normal individuals without smoking habits also showed no significant difference between the two assays. The mean value and standard deviation for CEA in these normal people using M-EIA was 1.4 -+ 1.3, and for the inhibition RIA it was 1.5 + 1.1. 4.4. Follow-up studies Sequential serum samples from 4 patients in whom recurrent colorectal carcinoma had been predicted by an elevation of circulating CEA measured by the inhibition RIA were also analyzed by the M-EIA. Correlation of the two tests in these individual follow-up studies was excellent. Fig. 3 shows the example of a 56-year-old female patient who was reoperated 3 times because of recurrences of a right colon carcinoma

160

B 80

r 0.95

m

,( ,T,

4020-

~

10-

z

_~ ~

2.5 1.25 I

~)

I

I

l

I

I

I

I

I

!

I

I

160

D X~ uJ

80-

r 0.93

/

r 0.88

./

4020< 10-

Ae

o.

.<,

•

5-

0 2.5-

ee

1.25 1.25

215

;

1~0

2'0

I

40

80

--"~1

1.25

215

!

5

I

10

I

20

410

8'0

ng CEA PER ml AFTER HCI04 EXTRACTION DETERMINED BY RIA

Fig. 2. Comparison of CEA results obtained by the M-EIA (ordinate) and by the inhibition RIA (abscissa) in various groups of patients and of healthy individuals with smoking habits. The calculated correlation coefficients (r) between M-EIA and RIA results axe indicated on each Panel. A: 63 sera of patients with colorectal cancers. B: 18 sera of patients with carcinoma of pancreas (o), gall bladder (D) and esophagus (~). C: 42 sera of patients with breast (X) and lung (o) carcinoma. D: 13 sera of patients with cirrhoses (•) and 32 sera of healthy individuals with smoking habits (e). 89

24-

= e~ tu (i)

20-

"s M,

1e-

Ist OPERATION

2nd OPERATION

3 r d OPERATION

4 t h OPERATION o

! ! !

.1 L <

12-

0 lID i-

7

•, S~'**S

8~ e 0

4

8

I~

~..~t7,,.8: 12

1(1

210

o r /

24

2'8

312

' 3w6

410

4w4

4w8

5'2

MONTHS FOLLOWING FIRST OPERATION.

Fig. 3. Repeated CEA measurements by both RIA (o e) and M-EIA ( o - - - o ) performed on serum samples of a patient who, after removal of a primary colon adenocarcinoma Dukes C developed 3 recurrences and was 3 times reoperated. There is an excellent correlation of the two tests with a correlation coefficient ofr = 0.954.

Dukes C. A regional tumor recurrence was resected twice and the CEA returned to normal values, whereas the third time unresectable pelvic recurrence was found and the CEA progressively increased. The parallelism between the results obtained by RIA (solid line) and M-EIA (dashed line), in this case as well as in the other cases tested, indicate that the two assays have the same capacity to detect tumor recurrences.

5. D i s c u s s i o n

In most enzyme immunoassays (EIA), also called enzyme linked immunosorbent assay (ELISA), the antigen is first adsorbed to immobilized antibody and then detected with another antibody coupled to an enzyme. With polyclonal antibodies it is possible, especially at low concentrations of antigen, that the majority of epitopes may be bound by insolubilized antibody resulting in an inhibition of binding of the second antibody. The use of Mabs reacting with dif90

ferent distant epitopes on the antigenic molecule should avoid this inhibition. An EIA using Mabs can theoretically be performed using a single incubation step. In fact, Uotila et al. have succeeded in such a single incubation assay with two Mabs to measure ct-fetoprotein [18]. However, a single incubation might give false-negative results in the presence of an excess of antigen when both the insolubilized and the enzyme-coupled Mab can bind only a small percentage of the antigen present. In our assay, a distinct second incubation is essential, because our enzyme-coupled Mab 23 binds with more affinity to CEA in low molarity buffer (0.01-0.05 M) than in buffer of physiologic molarity (0.15 M). Thus, the two sequential incubations allowed us to select a buffer of low molarity after the second intermediate washing. Another advantage of the two sequential incubations is that one can add selectively to the enzyme-coupled Mab 1% of normal mouse serum representing an excess of mouse IgG as compared to the enzyme-coupled Mab. The mouse IgG can inhibit

the binding of the Mab to any anti-mouse IgG antibody of human serum origin which may have been immunoadsorbed to the Mab-coated polystyrene beads during the first incubation. We have tested this possibility by adding rabbit serum against mouse lgG in the M-EIA. Even after heat extraction some antimouse IgG remained in solution and was able to bridge the two mouse Mabs (insolubilized and enzyme coupled) leading to false-positive results. However, when we added 1% normal mouse serum to the enzymecoupled Mab these non-specific results were abolished. The reason for our use of two Mabs on the immunoadsorbent follows from the fact that Mabs normally recognize non-repetitive epitopes on a protein. Thus they bind the antigen with only one arm on a single epitope and the binding is not very stable. In the case of Mabs anti-CEA 73 and 47 immobilized separately on balls, the maximum binding of J251-1abeled CEA in physiologic PBS was 50 and 30% respectively. Maximum binding was, however, 85% when both Mabs were immobilized in equal amounts on the same ball. This increase of binding is not merely additive as observed in a separate experiment (data not shown) using Mabs of lower affinity. Two Mabs also reacting with different epitopes of CEA and with weak affinity were put separately or together on the balls. Respectively 7% or 5% of radiolabeled CEA was bound when the Mabs were used separately on the balls, but binding was 27% when the two Mabs were mixed together in equal amount. The present results obtained with the new M-EIA involving 3 distinct, well-defined Mabs, as well as preliminary results obtained with 3 additional anti-CEA Mabs (data not shown), did not allow the identification of a subpopulation o f CEA specific for colorectal carcinoma. These results argue strongly against the hypothesis [ 19] that each type of carcinoma or tissue would produce its own subpopulation of CEA with specific epitopes. Thus the initial goal o f increasing the carcinoma specificity of the CEA test by the use o f Mabs has not been fulfilled. However, we demonstrate here that a convenient and reproducible assay for CEA can be developed with the exclusive use o f Mabs. The new M-EIA gave excellent correlation with the results obtained by the inhibition RlA using polyclonal goat antiserum as well as with those obtained with our recently described EIA using both goat polyclonal antibodies and a single mouse Mab [91. Further-

more, the new M-EIA has been found useful for the detection of recurrences of colorectal carcinoma, and it has been used to our satisfaction for more than 2000 CEA tests over a period of one year in our clinical studies. Acknowledgements We thank Prof. Ch. Haskell for reviewing this manuscript and Esther Oriol for excellent technical help.

References [1] Thomson, D. M. P., Krupey, J., Freedman, S. O. and Gold, P. (1969) Proc. Natl. Acad. Sci. U.S.A. 64,161. [2] Hansen, H. J., Snyder, J. J., Miller, E., Vandervoorde, J. P., Miller, O. N., Hines, L. R. and Burns, J. J. (1974) J. ttum. Pathol. 5,139. [3] Zamcheck, N. (1974) Adv. Intern. Meal. 19,413. [4 ] Egan, M. L., Lautenschleger, J. T., CoLigan, J. E. and Todd, C. W. (1972) lmmunochemistry 9,289. [5] Meriadec, B., Martin, l.'., Gu~rin, J., Henry, R. and Klipping, C. (1973) Bull. Cancer 60,403. [6] MacSween, J. M., Warner, N. L. and Mackay, I. R. (1973) Clin. Immunol. Immunopathol. 1,330. [7] Hirai, H. (1977) Cancer Research 37, 2267. [8] Maiolini, R., Bagrel, A., Chavance, C., Krebs, B., Herbeth, B. and Masseyeff, R. (1980) Clin. Chem. 26, 1718. [9] Buchegger, F., Phan, M., Rivier, D., Carrel, S., Aecolla, R. S. and Mach, J.-P. (1982) J. lmmunol. Meth. 49, 129. [ 10] Accolla, R. S., Carrel, S. and Mach, J.-P. (1980) Proc. Nail. Acad. Sci. U.S.A. 77,563. [ 11 ] K6hler, G., Howe, S. C. and Milstein, C. (1976) Eur. J. Immunol. 6,292. [12] Mach, J.-P. and Pusztaszeri, G. (1972) lmmunochemistry 9, 1031. [13] yon Kleist, S., Chavanel, G. and Burtin, P. (1972) Proc. Natl. Acad. Sci. U.S.A. 69, 2492. [ 14] Buchegger, F., Accolla, R. S., Carrel, S., Carmagnola, A., Girardet, Ch. and Mach, J.-P. (1980) in: Protides of the Biological Fluids, Vol. 28 (Peeters, H. Ed.) p. 51, Pergamon Press, Oxford. [15] Carlsson, J., Drevin, 1t. and Axen, R. (1978) Biochem. J. 173,723. [ 16] Mach, J.-P., Jaeger, Ph., Bertholet, M. M., Ruegsegger, C. H., Loosli, R. M. and Pettavel, J. (1974) Lancet ii, 535. 117] Fritsch~, R. and Mach, J.-P. (1977) lmmunochemistry 14, 119. [18] Uotila, M., Ruoslahti, E. and EngvaU, E. (1981) J. lmmunol. Meth. 42, 11. [ 19] Fuks, A., Banjo, C., Shuster, J., Freedman, S. O. and Gold, P. (1974) Biochim. Biophys. Acta 417,123. 91