HYBRIDOMA Volume 9, Number 3, 1990 Mary Ann Liebert, Inc., Publishers

Effect of Cross-Reactivity of Alpha-Fetoprotein Monoclonal Antibody on Quantitation of Serum AFP and Radioimmunodetection of Hepatocellular Carcinoma REIKO MAKIDONO Department of Radiology, Faculty of Medicine, Kyushu University 3-1-1, Maidashi, Higashi-ku, Fukuoka 812, Japan

ABSTRACT

Murine hybridomas were generated by immunizing mice with purified alphafetoprotein (AFP) to determine the cross-reactivity of AFP antibodies with human serum albumin (HSA). Eighteen out of 23 hybridoma products were cross-reactive with HSA. All 23 monoclonal antibodies (MAbs) could be subgrouped into 3 groups by their binding affinity to AFP and HSA, and one MAb from each group, IIB3, IIIB1 and IIIC6, were chosen for this study. The affinity to AFP was highest for IIB3 followed by IIIC6 and IIIB1, and that to HSA was in the order IIIB1 > IIB3. Immunohistochemical staining revealed no cross-reactions of these antibodies with HSA at the tissue level. IIB3 was the most efficient in the quantitative assays, indicating high affinity and specificity (non-cross-reactivity) of the antibodies, both of which are prerequisites for in vitro assays. Scans with good tumor

localization were obtained from mice which received 131I-labeled IIB3 or IIIC6 after xenografting of the human hepatocellular carcinoma. The images obtained with IIB3 were inhibited in the presence of either AFP or HSA in the circulation. These findings indicated that the use of cross-reactive and/or high-affinity antibodies against AFP provides an inhibitory factor for tumor localization. MAb IIIC6, because of its high specificity and moderate affinity to AFP, seems to be the best candidate for immunoscintigraphy of the tumors in future studies.

INTRODUCTION

Alpha-fetoprotein (AFP) is a major serum protein synthesized by fetal liver cells, yolk sac cells, and in trace amounts by the fetal gastro-intestinal tract (1, 2). AFP reappears in adult serum in the presence of hepatocellular carcinoma (HCC) and germ cell tumors containing yolk sac cell elements and is, therefore, used to monitor these tumors (2-4). The hybridoma technique established by Koehler and Milstein allows the production of large quantities of homogeneous, monospecific antibody to AFP, which replaces conventional antisera in various immunological examinations (5). Such monoclonal antibodies are also useful in the in vivo detection of the tumors by radio-imaging (6).

257

The author has noticed for years the high incidence of cross-reactivity with human serum albumin (HSA) of AFP monoclonal antibodies from various sources including commercial sources (e.g. anti-human a -Fetetoprotein, MCA 165 Serotec, Oxford, England). This facilitated making our own monoclonal antibodies. Radioimmunoassay and other sensitive immunological assays place special requirements on the antibodies used, in terms of specificity and affinity. The crossreactivity of the antibody with HSA may particularly affect the quantitation of serum AFP and in vivo localization of the tumor, since 102-103 times more HSA than AFP is present in patients' sera. These 2 antigens are structurally very similar in size and amino-acid composition, and have a common ancestry (7). Immuno-cross-reactivity, if there is any, may originate in their closely related genes. To obtain specific antibodies as well as to determine the effect of antibody cross-reactivity with HSA on quantitative assays for AFP and radioimaging of HCC, murine hybridomas were generated by immunizing mice with highly purified AFP. A panel of monoclonal antibodies thus produced provided the opportunity to study the efficacy of an Individual antibody of known specificity, affinity, class and concentration in each examination. The results indicated that each of the in vitro and in vivo uses of the antibodies placed different requirements on the specificity and affinity of the antibodies. MATERIALS AND METHODS

Mice Female mice of the inbred BALB/C strain were obtained from and kept at the Central Animal Breeding Unit of Kyushu University under strictly aseptic conditions.

Cell Line The myeloma cell line, P3-X63-Ag81 653 (Flow Laboratories, Inc., McLean, Va., U.S.A), was used for hybridization.

Antigen Pure human AFP was used for the primary immunization and the booster injections of mice, as well as for the screening of culture supernatants of hybridomas by enzyme-linked immunoadsorbent assay (EIA). For the immunoprecipitation and SDS-polyacrylamide slab gel electrophoresis (SDS-PAGE) analyses, 50% pure human AFP was used. Both AFP preparations with reference data were a gift from Daiichi

Radioisotope Ltd., Tokyo, Japan. Immunization and Cell Fusion

Mice were immunized by subcutaneous injection of 50 /. g of AFP emulsified in complete Freund's adjuvant and boosted by intravenous injection of 100 p g of AFP in saline 2-3 times, at 3- to 4-week intervals. Two weeks after the final 100pg injection, 50 p g of AFP was given intravenously on 3 successive days. Three days later, the spleens of 3 mice were removed for cell fusion. The hybridization and subsequent culture and cloning of the hybridomas were performed according to the method of Herzenberg with some modifications (8). The cultures were tested for anti-AFP activity by the EIA described below. The selected antibodyproducing hybridomas were cloned by limiting dilution in microtiter plates, and then grown in culture or cryopreserved. Ascites fluid was produced by intraperitoneal injection of 106 hybridoma cells into pristane-primed BALB/C mice.

258

Identification of Isotypes (Heavy Chains) and Light Chains of Monoclonal Antibodies Antibodies were isolated by using an anti-mouse immunoglobulin-coated Sepharose 4B immunoadsorbent column or a Protein A Sepharose column (Pharmacia, Uppsala, Sweden). Monoclonal antibody heavy and light chains were identified by immunodiffusion using chain-specific antisera (Nordic Immunology, Tilburg, Netherlands ).

Immunoprecipitation Experiments

Fifty percent pure human AFP or pooled patient sera containing AFP were labeled with 12BI by Bolton Hunter reagents (Amersham, Tokyo, JAPAN). 12BI-AFP was then immunoprecipitated by using the MAbs and rabbit anti-mouse immunoglobulins (Nordic Immunology). To separate specific imraunoprecipitates of 12BI-AFP, a 10% (V/V) suspension of fixed and heat-killed staphylococcus aureus, Cowan I strain, was used as an Immunoadsorbent. The eluates were subjected to 12.5% SDSPAGE at 20 mA for 4-5 hr (9). Fluorographs of dried gels were obtained by exposure for 2-7 days at 70°C. Enzyme-Linked Immunoadsorbent Assay (EIA) (a) For the screening of hybridomas, wells in microtiter plates (Flow Laboratories) were coated with 1 p g of pure AFP per ml. Culture supernatants After the were incubated in the coated wells for 2 hr at room temperature (RT). plates were washed, rabbit antibody to mouse immunoglobulins (Nordic Immunology) labeled with peroxidase were added to the wells and incubation was continued for another 2 hr. Enzyme activity bound to the wells was then measured using ophenylendiamine as a chromogen. The adsorbance was measured at 410 nm with a Multiskan (Flow Laboratories), (b) Sandwich EIA was performed by using one of 3 MAbs (IIB3, IIIB1, or IIIC6) and a polyclonal rabbit anti-AFP antibody, obtained by immunizing animals with commercial human AFP (Human AFP standard serum, Behringwerke AG, Manburg W, Germany). Polystyrene beads (4-mm diameter) were

coated with monoclonal antibody at a concentration of 42 p g/ml in Tris HC1 buffer (pH 8.5). Various concentrations of AFP (as standards) or patients' sera containing unknown amounts of AFP were then applied to duplicate beads which were then incubated for 2 hr at 37°C. After the beads were washed, a properly diluted monoclonal or polyclonal anti-AFP antibody, labeled with horse-radish peroxidase (Sigma Chemical Co., St. Louis, Mo.) by using a maleimide compound (10), was added to each bead. The beads were further incubated for 2 hr at 37°C. Enzyme activity bound to the beads was then measured by using o-phenylendiamine as a

chromogen. Radioimmunoassay (RIA)

Radioimmunoassay

was performed by using a kit constructed by the same assay the EIA described above. 12BI-labeled polyclonal rabbit anti-AFP antibody as a second antibody was used for the assay.

system

as

Competitive Inhibition Studies Antibodies to be used were first titrated on plates coated with AFP to determine the concentration of antibody giving half-maximal binding. Antibody at this concentration was then incubated with various concentrations of the inhibitor, HSA, at 37°C for 3 hr. The plates were then processed as described above for routine EIA.

259

[mmunohi stochemical Staining

The binding of antibodies in the following tissues

investigated:

10 The yolk sac tumors were a gift from Dr. M. Sawada (11). Frozen 5-j_ra sections as well as paraffin sections of tissue blocks which were fixed in 10% buffered formaldehyde were prepared. For the localization of AFP and HSA reacting with specific antibodies, a modification of the avidin-biotin-peroxidase complex technique was used (12). Both kinds of sections were stained with MAbs to AFP (IIB3, IIIB1 and IIIC6), and a polyclonal (goat) anti-human albumin antibody (Cappel Laboratories, Cochranville, Pa.) by using the avidin-biotin-peroxidase complex (ABC reagent, Vector Laboratories, Burlingame, Calif.) following the manufacturer's instructions. The 3 MAbs to AFP were applied to tissue sections at a final concentration of 1 # g/section. For each tissue specimen, a negative control was included consisting of normal mouse IgG instead of the MAbs. Endogenous peroxidase was blocked during this staining by exposure to 0.5% H202 in methanol for 20 min. The staining solution was made by diluting 0.24% 3-amino-9-ethylcarbazole in dimethyl-formamide 1/10 in 0.2 M acetate buffer (pH 5.2), which contained 0.012% hydrogen peroxide. All tissue specimens were then counterstained with 1% methyl green. Since there was no difference in staining between frozen and paraffin sections, and frozen sections could not be prepared for all the cases, only the results with paraffin sections are presented.

HCCs, 11 cirrhotic livers, 10 normal livers and 6 yolk

was sac tumors.

Preparation of 131I-labeled Monoclonal Antibody for Radioimaging Monoclonal antibody was purified from the ascites fluid of mice in which the antibody-secreting hybridomas were growing, by binding to Protein A Sepharose-

columns. Each of the eluted antibodies 100 p g was labeled with 750 p C of 13:LI-Na by the chloramine T method with minor modifications (13). The labeling efficiency was on the average 80%. Immediately before use, the labeled antibody was spun in a microcentrifuge to remove aggregates.

Radioimaging

Scintiphotoscanning was performed with a gamma camera with a pinhole collimator (Toshiba GCA-192, Osaka, Japan). HCC obtained from a patient (M.I.) was used for this study. For radioimaging, 20 mice bearing tumors of nearly the same size (1.0 cm3) in their right flanks were divided into 3 groups leaving 2 control mice. The control mice received 131I-labeled normal mouse IgG. Two mice from each group were then injected intravenously with approximately 20 /_ g of 13:LI-labeled IIB3, IIIB1 or IIIC6. Another 2 sets of 2 mice from each group were injected intravenously with 5 p g of AFP or 50 pg of HSA, 1 hr before the injection of labeled antibodies. Serial scintiphotoscanning was performed 24, 48, 72, and 96 hr after the injection of labeled antibody or normal mouse IgG. To estimate the blood pools in the mice °°"Tc-HSA, 100 pCi, was injected intravenously after the completion of the 96-hr scanning and the mice were scanned 15 min later. RESULTS

I. Characterization of the MAbs A total of 23 anti-AFP antibody-producing hybridomas were derived from the fusions. Hybridoma products were screened for their ability to bind to AFP and HSA by using the EIA method. Eighteen of the 23 were cross-reactive with HSA. All antibodies were divided into 3 groups by their binding affinity to AFP and HSA, and representative MAbs from each group, 2 cross-reactive (IIIB1 and IIB3)

260

94K 67K 43K

.OK 12

3

4

5

6

7

^pH

Autoradiographs of SDS-PAGE Representing Monoclonal Antibody-Reactive Polypeptides. Partially purified human AFP (50% purity) was labeled with 12BI by using Bolton Hunter reagents. 12BI-labeled AFP was immunoprecipitated by using MAbs and rabbit anti-mouse immunoglobulins. The immunoprecipitates of _2BI-AFP were analysed by SDS-PAGE. Lane 1, 12SI-HSA; Lane 3, IIB3; Lane 4, FIGURE 1

IIIC6; Lane 5, IIIB1 and Lane 7, 12BI-AFP.

and 1 non-cross-reactive (IIIC6), were then chosen from each group for further analysis. These 3 antibodies were found to be IgG 1 (k) by immunodiffusion

analysis.

The molecular weight (M.W.) of the MAb-reactive polypeptides was determined by SDS-PAGE analysis. MAbs IIIC6 and IIB3 precipitated a polypeptide of 70,000 M.W. from a partially purified AFP sample as well as from pooled serum from hepatoma patients (10,000 ng of AFP per ml) under non-reducing conditions (Figs. 1 and 2). MAb IIIB1 also precipitated a polypeptide of 70,000 M.W. from the AFP sample, but a polypeptide of 67,000 M.W. from the hepatoma serum. This finding indicated higher affinity of MAb IIIB1 of HSA. The affinity of each antibody was assessed by the end point titer in a

FIGURE 2 Molecular Weight Analysis of the Serum Proteins Immunoprecipitated by the Anti-AFP MAbs. Pooled human serum from hepatocellular carcinoma patients was applied to IIB3, IIIB1 or IIIC6 MAb-coated immunoadsorbent columns. The proteins in the acid eluate from the columns were labeled with 12BI by using Bolton Hunter reagents, and then analyzed by SDS-PAGE. Lane 1, IIIB1; Lane 3, IIB3; Lane 5, IIIC6 and Lane 6, molecular weight standard.

261

0.8

1.6

Antibody

3.1 do««

6.3

(j_g/ml!

Binding of Monoclonal Antibodies to AFP Plates in an Enzyme-Linked Immunoadsorbent Assay (EIA). Wells in microtiter plates were coated with serially diluted AFP. MAbs at a 1 p g/ml concentration were then incubated in the coated wells to allow binding to AFP. FIGURE 3

dilution series by means of EIA (Fig. 3). A comparable end point titer could be determined by using the lowest binding that is reliably above the background noise. When 2 levels (20% and 30% of the maximum level of binding) were used as end point titers in this assay, they corresponded to the steep part of the doseresponse curve in Fig. 3. The maximum binding was defined as the plateau binding by the highest affinity antibody (IIB3). End point titers were reached at a lower concentration with high-affinity antibodies, and at much higher concentrations when the affinity was lower. Thus, the affinity of MAb IIB3 was determined to be the highest, followed by MAbs IIIC6 and IIIB1. The cross-reactivity of each MAb with the HSA sample analyzed is shown in Fig. 4. MAb IIIB1 was the most cross-reactive, showing the highest binding to HSA. MAb IIB3 also cross-reacted but in a lesser degree. This was further shown in inhibition assays by using HSA as the inhibitor (Fig. 5). The competitive effect of HSA on the binding of the antibodies to AFP was significantly high only for MAb IIIB1.

57«

ÏÏ7-2.. 5

4S.0

Antibody dos« (^q/.l)

FIGURE 4 Binding of Monoclonal Antibodies to HSA Plates in an EIA. microtiter plates were coated with a dilution series of human serum MAbs (1 /¿g/ml) were allowed to bind to HSA.

262

Wells in albumin.

Concentration of inhibitor, HSA( mg/ml )

FIGURE 5 Inhibition of the Binding of Monoclonal Antibodies to AFP by HSA. Antibodies given half maximal binding to AFP were incubated with the indicated concentrations of inhibitor for 3 hr at 37°C in order to allow the inhibitor to bind to antibodies, before being used in an EIA on AFP plates.

II. Efficacy of Cross-Reactive and Noncross-Reactive MAbs in the Quantitation of Serum AFP

By using these characterized MAbs, 2-step solid-phase immunoadsorbent assays for AFP were constructed, and the efficacy of each antibody in quantitation of serum AFP was determined. No binding of the second antibody was observed when the same antibody was used for coating and as an enzyme-labeled second antibody in the sandwich assays. However, when the 2 were combined, all the combinations resulted in substantial binding of the conjugate on the solid support. This finding indicated that they did not interfere with each other in the binding to AFP, but reacted with a different determinant on the AFP molecule (data not shown). When semiquantitive assays were carried out, they did not give a good dose response, probably because of insufficient affinity and/or specificity of the IIIB1 and IIIC6 MAbs. Each of the antibodies in combination with an enzymelabeled polyclonal rabbit anti-AFP antibody was then tested for its efficiency in the assays. The best dose response was obtained with MAb IIB3 (Fig. 6). A panel of samples representing the clinical applications of an AFP test were assayed in this sandwich assay system (EIA) using MAb IIB3 as the first antibody. The concentration of the AFP standard ranged from 12.5 to 800 ng/ml. The AFP standard curve was reproducibly linear in this range. Such a standard curve is shown in Fig. 6, where the addition of 0.5% HSA to the dilution buffer decreased the sensitivity of the assay, but the addition of 1% bovine serum albumin (BSA) increased it (Fig. 6). All the sera tested by EIA were also analyzed by RIA, and the two assay systems gave identical results. The cut-off value of both assays was 5 ng/ml. The absolute serum AFP concentrations in the samples from 99 hepatoma patients found by the 2 assay techniques showed an excellent correlation (Fig. 7). The values correlated with a Pearson-Bravais coefficient of 0.95. The regression line was Y=1.01 X-24.1. III.

Specificity of Binding of the MAbs

to Tissue AFP

Specificity of the MAbs against AFP was determined immunohistochemically by using the tissues of HCC, cirrhotic and normal livers and yolk sac tumors. The location of the AFP and HSA-positive cells in each tissue was identified by staining consecutive tissue specimens with specific antibodies against AFP (MAb IIIC6) and HSA (a goat antibody). AFP-positive cells were detected in most of

263

SO

100

Concetrition of AFP

200

400

BOO

(ng/ml)

Standard Curves with MAb IIB3 for Quantitation of AFP by the 2-Step Sandwich Method of EIA. AFP in the sample reacted with unsolubilized MAb IIB3 on polystyrene beads and then the enzyme-labeled polyclonal antibody (goat) was added to the antigen-antibody complexes. 0.5% HSA in the AFP •-• none; o-o dilution buffer; G-O 1.0% BSA in the AFP dilution buffer. FIGURE 6

,

,

,

Y

-

R

=

F6

1.-6-06 X

+

-S4.11.5

.9-19215 88S.891

FIGURE 7 Comparison of the AFP Values Obtained by Radioimmunoassay (RIA) and EIA with MAb IIB3. Serum samples from 99 hepatoma patients were examined. Each data point was generated by the concentration of AFP in the serum (ng/ml).

264

FIGURE 8 Immunohistochemical Staining of Normal Livers, a) Section stained with goat anti-HSA antibody. No staining was observed with any of the 3 MAbs to AFP in consecutive tissue specimens (data not shown), b) Sections stained with hematoxylin and eosin (HE). Original magnifications x 91.

the HCCs (9 of 10 cases), a cirrhotic liver (1 of 11 cases) and yolk sac tumors (6 of 6 cases), but none in normal livers (Figs. 8-11). HSA-positive cells were detected in all the tissues of HCC and cirrhotic as well as normal livers (Figs. 8-10), but not in most of the yolk sac tumors (positive in 1 of 6 cases). These immunohistochemical observations indicated that at any given time, only a fraction of the hepatocytes were engaged in the active production of albumin and that

FIGURE 9 Immunohistochemical Staining of a Cirrhotic Liver, a) Section stained with anti-HSA antibody, b) Section stained with a MAb against AFP, IIIC6. c) Section stained with HE. Original magnifications x 91.

265

FIGURE 10 Immunohistochemical Staining of a Hepatocellular Carcinoma, a) Section stained with anti-HSA antibody. b)-d) Sections stained with MAbs against AFP, IIIC6, IIB3 and IIIB1, respectively, e) Section stained with HE. Original magnifications x 91. cells were capable of producing both antigens simultaneously under certain conditions. When these tissues with a known distribution of AFP and HSA were stained with MAbs IIB3 and IIIB1, both gave staining patterns identical to those obtained with the specific antibody IIIC6 in the consecutive tissue specimens (Fig. 10 b-d). Thus, no cross-reactivity of these antibodies was demonstrated at the tissue level. IV. Efficacy of Cross-Reactive and Non-Cross-Reactive MAbs in Radioimmunodetec-

tion of HCC

Each of the 3 131I-labeled MAbs was intravenously injected into groups of mice and examined for its ability to detect the location of the HCC xenografted 266

FIGURE 11

AFP, IIIC6.

f) YST-6.

Immunohistochemical Staining of Yolk Sac Tumors with a MAb against Sections are a) YST-1, b) YST-2, c) YST-3, d) YST-4, e) YST-5 and Original magnifications x 91.

in nude mice. Scans with good tumor localization were obtained 24, 48, 72 and 96 hr for the mice which received 131I-labeied IIB3 and IIIC6 MAbs, in which AFP released from the tumor did not exceed 50 ng/ml in the circulation. Five micrograms of AFP or 50 ^g of HSA injected 1 hr before the injection of the labeled MAbs inhibited tumor images obtained by MAb IIB3, with virtually no effect on the accumulation of the labeled IIIC6 MAb (Fig. 12). The in vivo distribution of labeled MAbs in mouse tissues was examined immediately after completion of the 96-hr scans. Mice were exsanguinated and the radioactivity in the blood, tumors and tissues was assessed. The organs were weighed and, after correction for the physical decay rates of 131I, the counts per minute per gram were calculated for each tissue. Tissue tumor ratios were then obtained by dividing the specific counts per minute per gram of each tissue by the specific counts per minute per gram of tumor. The scanning results paralleled those obtained by direct measurement of the radioactivity (Figs. 1315). Tissue tumor ratios in mice which had received labeled IIIB1 or IIB3 MAb 267

Scintigraphic Images of a Mouse Injected with 131I-IIIC6 against AFP. a) Image of blood pool with SB-Tc-HSA 15 min after injection. b)-d) Images obtained 24, 72 and 96 hr respectively after 131I-IIIC6 Injection. FIGURE 12

with AFP and HSA suggested that images without the background subtraction would be suboptimal because of the high blood pool of radioactivity (Figs. 14 and 15). The great increase in radioactivity in the blood as well as the lung, spleen, and liver in such mice indicated the accumulation of immune complexes formed by the injection of the antibodies. The blood-tumor ratios were lower only in the mice which received MAb IIIC6 even with injections of AFP and HSA (Fig. 13).

DISCUSSION The fact that MAbs react with a single antigenic determinant in the corresponding antigens still brings up the possibility that unexpected crossreactivities could appear. Cross-reactivity of MAbs against AFP with HSA may originate in their closely related genes (7). These include nucleotide sequence

CA>'" i -mee

(s)1" i -race

*

so u> «sa

(c)1*' i -mc6

*

su« afp

FIGURE 13 Relative x31I-Radioactivity in Tissues 96 hr after injection of MAb IIIC6. The tissue tumor ratios were obtained by dividing the specific counts per minute per gram of each tissue by the specific counts per minute per gram of tumor. (A) 131I-IIIC6 alone; (B) with HSA; (C) with AFP.

268

(8)'" I -0 83

(A)'" I -UB

t

50 uj HSA

CO'" 1 -DB3

I

ui AFP

:-rd.

111

iL

FIGURE 14 Relative 131I-Radioactivity in Tissues 96 hr after injection of MAb IIB3. (A) 131I-IIB3 alone; (B) with HSA; (C) with AFP. See also legend for Figure 13.

homology and immunological cross-reactivity of the denatured peptides (14, 15). In this study, 99% pure AFP was used for immunizing mice to generate hybridomas. But still many of the anti-AFP antibodies, the hybridoma products, cross-reacted with HSA. This might supports the possibility of the existence of an evolutionarily related AFP/albumin gene family and structural homology among the epitopes on the proteins.

The effects of cross-reactivity of antibodies against AFP with HSA in various antibody assays has scarcely been discussed. However, it is assumed that (b)"'

i -niel

I -I

FIGURE 15 Relative 13XI-Radioactivity in Tissues 96 hr after injection of MAb IIIB1. (A) 131I-IIIB1 alone; (B) with HSA; (C) with AFP. See also legend for Figure 13. 269

any MAbs to AFP can be cross-reactive with HSA. A higher cross-reactivity with other antigen(s) equally means a reduced specificity of the antibodies. Besides, the extent of the cross-reactivity of the antibody with HSA may have a particular influence on the sensitivity of the solid-phase assays, since a high concentration of HSA is present in most serum samples to be assayed for AFP; undiluted patient sera contain 102-103 times more HSA than AFP. This was supported by

the finding that MAb IIIB1, highly cross-reactive with HSA, precipitates either

HSA or AFP, depending on their concentrations (Figs. 1 and 2). It was assumed that MAb IIIB1 can eventually be utilized in the solid-phase assays if higher dilutions of the serum are used for the assays. However, because of its low affinity to AFP, the sensitivity of the assays could not be increased (data not shown). The predominant binding of MAb IIIBI to soluble HSA was also evident in the assay for inhibition of binding to AFP using a solid-phase EIA (Fig. 5). Such cross-reactivity of antibody may also have caused an error in the molecular weight estimation for human AFP in early studies (16). Cellular and subcellular immuno-localization of AFP and albumin has been extensively studied to determine the nature, dynamic behavior and junctional interrelationship of these two protein-producing liver cell populations, and expression characteristic of the evolutionary related genes during ontogeny or carcinogenesis (17-20). Most investigators have used polyclonal antibodies, with only a few using MAbs. In such studies, cross-reactivity of the antibody against AFP with HSA may have yielded conflicting results, although there were no crossreactions of the antibodies (MAbs IIB3 and IIIBI) at the tissue level in this

study.

The lectin-binding molecular heterogeneity of AFP present in amniotic fluid, serum and patient serum has been reported (21-24). This has led to extensive investigations of the diagnostic usefulness of the molecular variants of human AFP. The molecular variants were found at least in AFP secreted by HCC and germ cell tumors. It was suggested that the heterogeneity in the reactivity of AFPs derived from different sources is due primarily to the structure of the carbohydrate moiety of each AFP (24, 25). A structural analysis of the AFP carbohydrates from HCC revealed the presence of biantennary sugar chains, while AFP from yolk sac tumors had a series of sialylated and nonsialylated biantennary complex type sugar chains with bisecting-ß -N-acetyl-glucosamine, none of which binds to concanavalin A-sepharose. Therefore, MAbs that distinguish between these AFPs of different origins may be highly useful. The antibodies used in this study, however, did not show any preferential reactivity with either AFP; both AFPs were stained in the tissue sections (Figs. 10 and 11), and reacted not only with serum AFP of HCC but also with that of yolk sac tumors and fetal origin (data not shown). These findings indicate that they do not recognize epitopes related to the sugar moieties on the AFP molecule. Since Pressman (26) and Bale (27) first succeeded in radio-imaging tumors by using antibodies against experimental tumors, numerous studies with different antibodies and radiolabels have confirmed the general efficacy of this method of non-invasive disclosure of known and occult tumors. Neither echography nor computed tomography has been demonstrated to be sufficiently sensitive to detect hepatic lesions smaller than 1 cm in diameter and they are not specific enough to discriminate HCC from liver metastasis. Therefore, there has been great hope for the radioimmunodetection of HCC using antibodies against AFP. The sensitivity of this method, however, did not exceed 50% in most studies (28). Several factors could explain this low localization rate. First, some 10-15% of HCCs escape detection as they do not produce any AFP (29). A vast majority of AFP-positive HCCs, on the other hand, produce large amounts ranging from 1000-1,000,000 ng/ml in the serum (30). This causes a failure to disclose the tumor as well. The injected antibody contributes to form immune complexes with circulating AFP, which affects the ability of the antibody to bind to the antigen target on the HCC. The same condition should occur after the injection of MAbs with any specificity, if they cross-react with HSA. This was shown by the Injection of MAbs IIB3 and IIIBI, resulting in forming immune complexes with circulating HSA (Figs. 14 and 15). MAb IIIC6 with moderate affinity accumulated better in the tumor than the high-affinity antibody IIB3. This finding indicates that the use of highaffinity antibodies provides an inhibitory factor for tumor localization if serum fetal

270

antigen levels are elevated. Some investigators have emphasized the phenomenon that good tumor images could be obtained even in the presence of circulating antigens (31). In such cases, however, a lower affinity of the antibodies used could have contributed to their tumor localization ability. The following observation could also be explained by the affinity of the antibodies. When tumor uptake of whole molecules of MAbs with high affinity were compared with that of their F(ab')2 or Fab fragments, the latter two showed higher tumor uptake in some studies (32, 33). This result has been explained by the absence of interaction with the Fc receptor, better penetration in poorly vascularized tumors and higher blood clearance of the fragments, leading to a decrease in background. However, it can also be related to their difference in affinity to the corresponding antigens. Reduced affinity of the monovalent Fab was shown by Beaumier et al. in their comparative study of tumor localization in vivo using intact MAbs and their F(ab')2 and Fab fragments (34). Thus, the affinity of antibodies has an implication in the clinical application of monoclonal antibodies. MAb IIIC6, because of Its high specificity and moderate affinity to AFP, seems to be the best candidate for immunoscintigraphy of the tumors in future studies. ACKNOWLEDGMENTS The author is grateful to Daiichi Radioisotope Laboratories, Ltd., for kindly providing purified AFP and to Dr. Masumi Sawada at the Research Institute for Microbial Diseases, Osaka University, for the yolk sac tumors. She also wishes to thank Mrs. Aya Oishi for her expert technical assistance and Miss Takako Kaneyuki for secretarial assistance. This work is supported in part by a Grant-in-Aid for Cancer Research from the Ministry of Education, Health and Welfare of Japan. REFERENCES

Perricelli, A. and Gitlin, G.M. (1972) Synthesis of alphafetoprotein by liver, yolk sac and gastrointestinal tract of the human

1. Gitlin, D.,

conceptus. Cancer Res. 32:979-982.

and fetal

2.

Ruoslahti, E. and Seppala, M. (1979) a -fetoprotein in development. Adv. Cancer Res. 29:275-346.

3.

Abelev, G.I., Perova, S.D., Khramkova, If.I., Postnikova, Z.A. and Irlin, I.S. (1963) Production of embryonal ß -globulin by transplantable mouse hepatomas. Transplantation 1:174-180.

cancer

4. Bellet, D.H., Wands, J.R., Isselbacher, K.J. and Bohuon, C. (1984) Serum fetoprotein levels in human disease. Perspective from a highly specific monoclonal radioimmunoassay. Proc. Nati. Acad. Sei. USA 81:3869-3873. 5. Köhler, G. and Milstein, C.

(1975) Continuous cultures of fused cells ing antibody of predefined specificity. Nature (Lond.) 256:495-497.

6.

a-

secret-

Goldenberg, D.M., Kim, E., Deland, F., Spremulli, E., Nelson, 0., Gockerman, J.P., Primus, F.J., Corgan, R.L. and Alpert, E. (1980) Clinical studies on the radioimmunodetection of tumors containing alpha-fetoprotein. Cancer 45:2500-2505.

7.

Law, S.W. and Dugaiczyk, A. (1981) Homology between the primary structure of a-fetoprotein, deduced from a complete cDNA sequence, and serum albumin. Nature 291:201-205.

8.

Herzenberg, L.A., Herzenberg, L.A. and Milstein, C. (1978) Cell hybrids of myelomas with antibody forming cells and T-lymphomas with T cells. Chapter 25, Hand Book of Experimental Immunology. Blackwell Scientific Publications, Oxford, London, Edinburgh, Melbourne. 271

9.

Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227:680-685.

10. Yoshitake, S., Imagawa, M., Ishikawa, E., Niitsu, Y., Urushizaki, I., Nishimura, M., Kanazawa, R., Kurosaki, H., Tachibana, S., Nakazawa, N. and Ogawa, H. (1982) Mild and efficient conjugation of rabbit Fab' and horseradish peroxidase using a maleimide compound and its use for enzyme immunoassay. J. Biochem. 92:1413-1424. 11.

Sawada, M., Matsui, Y., Hayakawa, K., Nishiura, H., Okudaira, Y. and Taki, I. (1982) Human gynecologic cancers heterotransplanted into athymic nude rats. Gynecologic Oncology 13:220-228.

12. Hsu, S.M.,

Raine, L. and Franger, H. (1981) A comparative study of the peroxidase-antiperoxidase method and on avidin-biotin-complex method for studying polypeptide hormones with radioimmunoassay antibodies. Am. J. Clin. Pathol. 75:734-738.

13. Hunter, W.M. and Greenwood, F.C. (1962) Preparation of iodine 131I-labeled human growth hormone of high specific activity. Nature 194:495-496. 14.

Ruoslahti, E. and Engvall, E. (1976) Immunological crossreaction between alpha-fetoprotein and albumin. Proc. Nati. Acad. Sei. USA 73:4641-4644.

15.

Morinaga, T., Sakai, M., Wegmann, T.G. and Tamaoki, T. (1983) Primary structures of human a-fetoprotein and its mRNA. Proc. Nati. Acad. Sei. USA 80:4604-4608.

16. Nishi, S. (1970) Isolation and characterization of a human fetal a-globulin from the sera of fetuses and a hepatoma patient. Cancer Res. 30:2507-2513. 17.

Guillouzo, A., Belanger, L., Beaumont, C., Valet, J.P., Briggs, R. and Chiu,

J.F. (1978) Cellular and subcellular immunolocalization of a-fetoprotein and albumin in rat liver. Réévaluation of various experimental conditions. J. Histochem. Cytochem. 26:948-959.

18.

Engelhardt, N.V., Goussev, A.I., Shippova, L.Jr. and Abelev, G.I. (1971) Immunofluorescent study of alpha-foetoprotein ( fp) in liver and liver tumours. I. Technique of fp localization in tissue sections. Int. J. Cancer 7:198-206.

19. Gleiberman, A.S., Kuprina-Khramkova, N.I., Rudinskaya-Beloshapkina, T.D. and Abelev, G.I. (1983) Alpha-fetoprotein synthesis in relation to structural peculiarities in postnatal and regenerating mouse liver. Int. J. Cancer 32:85-92. 20.

Higgins, P.J., Darzynkiewicz, Z. and Melamed, M.R. (1983) Secretion of albumin and alpha-foetoprotein by dimethylsulphoxide-stimulated hepatocellular carcinoma cells. Br. J. Cancer 48:485-493.

21.

Smith, C.J. and Kelleher, P.C. (1973) 1-a fetoprotein: Separation of two molecular variants by affinity chromatography with concanavalin A-agarose. Biochem. Biophys. Acta 317:231-235.

22.

Smith, C.J., Kelleher, P.C., Belanger, L. and Dallaire, L. (1979) Reactivity of amniotic fluid alpha-fetoprotein with concanavalin A in diagnosis of

neural tube defects. Br. Med. J. 1:920-921. 23.

Krusius, T. and Rouslahti, E. (1982) Carbohydrate structure of the concavalin A molecular variants of a-fetoprotein. J. Biol. Chem. 257:3453-3457.

24.

Yoshima, H., Mizuochi, T., Tshii, M. and Kobata, A. (1980) Structure of the

272

sugar chains of fluid. Cancer Res. 40:4276-4281.

asparagine-linked

a

-fetoprotein purified from human ascites

25. Yamashita, K., Hitoi, A., Tsuchida, Y., Nishi, S. and Kobata, A. (1983) Sugar chain of a-fetoprotein produced in human yolk sac tumor. Cancer Res. 43: 4691-4695. 26. Pressman, D. and

osteogenic

L. (1953) The in vivo localization of anti-Wagnerantibodies. Cancer 6:619-623.

Korngold,

sarcoma

27. Bale, W.F., Spar. F.L., Goodland, R.L. and Wolfe, D.E. (1955) In vivo and in vitro studies of labelled antibodies against rat kidney and Walker carcinoma. Proc. Soc. Exp. Biol. Med. 89:564-568. 28. Bergmann, J., Lumbroso, J., Manil, L., Saccavini, J., Rougier, P., Assicot, M., Mathieu, A., Bellet, D. and Bohuon C. (1987) Radiolabelled monoclonal antibodies against alpha-fetoprotein for in vivo localization of human hepatocellular carcinoma by immunotomoscintigraphy. Eur. J. Nucl. Med. 13:385-390. 29.

Vogel., CL., Primack, A., Mclntire, K.R., Carbosie, P.P. and Anthony, P.P. (1974) Serum alpha-fetoprotein in 184 Ugandan patients with hepatocellular carcinoma. Cancer 33:959-964.

30.

Ruoslahti, E., Salaspuro, M., Pihko, H., Anderson, L. and Seppala, M. (1974) Serum a-fetoprotein: Diagnostic significance in liver disease. Br. Med. J. 2:527-529.

31.

Goldenberg, D.M., DeLand, M.F., Kim, E.E., Benett, S., Primus, F.J., van Nageil, J.R., Esters, N., DeSimone, P. and Rayburn, P. (1978) Use of radiolabelled antibodies to carcinoembryonic antigen for the detection and localization of diverse cancers by external photoscanning. N. Engl. J. Med. 298: 1384-1388.

32.

Buchegger, F., Haskell, CM., Schreyer, M., Scazziga, B.R., Randin, S., Carrel, S. and Mach, J. (1983) Radiolabeled fragments of monoclonal antibodies against carcinoembryonic antigen for localization of human colon carcinoma grafted into nude mice. J. Exp. Med. 158:413-427.

33. Buchegger, F., Mach, J., Leonnard, P. and Carrel, S. (1986) Selective tumor localization of radiolabeled anti-human melanoma monoclonal antibody fragment demonstrated in the nude mouse model. Cancer 58:655-662. 34. Beaumier, P.L., Krohn, K.A., Carrasquillo, J.A., Eary, J., Hellstrom, I., Hellstrom, K.E., Nelp, W.B. and Larson, S.M. (1985) Melanoma localization in nude mice with monoclonal Fab against P97. J. Nucl. Med. 26:1172-1179.

Address reprint requests to:

Makidono, Dept. of Immunology, Faculty of Medicine, Kurume University, 670, Asahi-machi, Kurume-city, FUKUOKA 830, JAPAN Dr. Reiko

Received for

Accepted

publication January 26, 1989 April 3, 1990

after revisions

273

Effect of cross-reactivity of alpha-fetoprotein monoclonal antibody on quantitation of serum AFP and radioimmunodetection of hepatocellular carcinoma.

Murine hybridomas were generated by immunizing mice with purified alpha-fetoprotein (AFP) to determine the cross-reactivity of AFP antibodies with hum...
12MB Sizes 0 Downloads 0 Views