Journal of Immunological Methods, 19 (1978) 87--93 G Elsevier/North-Holland Biomedical Press
87
AN INDIRECT ANTIBODY ASSAY USING H A P T E N A T E D ANTIGEN AND '25I-LABELLED ANTI-HAPTEN ANTIBODY
R.C. AALBERSE Central Laboratory of the Netherlands Red Cross Blood Transfusion Service, Amsterdam, the Netherlands and the Laboratory of Experimental and Clinical Immunology of the University of Amsterdam, the Netherlands (Received 6 May 1977, accepted 30 June 1977)
Hapten (trinitrophenyl) was coupled to antigen (ovalbumin). The haptenated antigen was bound by anti-ovalbumin antibody and binding was quantitated with t2SI-labelled anti-hapten antibodies. Thus, with a single radioactive reagent antibodies against a variety of antigens can be detected, whilst the problems inherent in a labelled antiglobulin binding test are avoided. In the ovalbumin system, the haptenated antigen binding test proved to be approximately 20 times as sensitive as the iodinated ovalbumin binding test.
INTRODUCTION
Quantitative antibody assays are often based on the binding of either '2sIlabelled antigen or ~2SI-labelled antiglobulin. In the antiglobulin binding reaction, only a single labelled reagent is used; it is theoretically the most sensitive assay. However, non-specific uptake of immunoglobulins often sets the practical limit o f sensitivity. The use of '2SI-labelled antigen provides greater specificity and in practice higher sensitivity, but when antibodies against a n u mb er of different antigens are to be measured, all the antigens have to be labelled. In the present c om m uni c a t i on, an indirect labelled antigen binding technique is described, using a single radioactive reagent by which antibodies against different antigens can be measured. The antigen is labelled with a hapten, e.g. the trinitrophenyl (TNP)-group; binding of the haptenated antigen is quantitated with '25I-labelled anti-hapten antibodies. This test combines the sensitivity of the '2SI-labelled antigen binding tests with the versatility of the antiglobulin-binding test. An assay for antibodies to ovalbumin is described as an example. MATERIALS AND METHODS H a p t e n a t i o n o f p r o t e i n antigens
Trinitrophenylation of bovine serum albumin (BSA, crystallized, Sigma) and ovalbumin (OVA, 5X crystallized, Calbiochem} was perform ed with
88 trinitrobenzene sulfonic acid (TNBS, Baker) as described by Little and Eisen {1967). Fifty mg of TNBS was added to 500 mg protein in 50 ml 0.02 M sodium borate buffer pH 8.6. After 1 h reaction at room temperature the solution was dialyzed against phosphate buffered saline (PBS) for three days at 4°C. The optical absorbance at 345 nm of the dialyzed solution (calculated protein concentration 8 mg/ml) was 33.6 for TNP-BSA and 19.0 for TNPOVA, corresponding to TNP/protein ratios of 15.9 and 6.5 respectively.
Preparation of anti-hapten antibodies TNP-Sepharose was prepared by reacting 1 ~mol TNBS with 1 ml diaminoethane (DAE)-Sepharose at pH 8.6. DAE-Sepharose was prepared by reacting CNBr-activated Sepharose (Pharmacia) with an excess of diaminoethane (Merck) at pH 9; subsequently the excess of diaminoethane was removed. Anti-TNP antibody was raised in a rabbit by immunization with 0.5 mg TNP-BSA in complete Freund's adjuvant, followed after 5 weeks by a second injection with 0.05 mg TNP-BSA in incomplete Freund's adjuvant. Seven days later citrated plasma was obtained. To isolate anti-TNP antibodies, the plasma was defibrinated and inactivated by heating for 1 h at 56°C. Five ml defibrinated plasma was passed through a column consisting of 1 ml TNPSepharose mixed with approximately 5 ml Sephadex-G25, layered on approximately 35 ml Sephadex-G25 (total column dimensions 55 × 1 cm). The column was washed with 200 ml PBS and with 50 ml glycine buffer (0.1 M glycine--HC! in 0.15 M NaC1, pH 2.5); antibodies were eluted with glycine buffer--dioxan 9 : 1 (v/v) (Hill, 1972).
Labelling with ~2Sl 0.5 ml of the protein-containing neutral column fractions (A280 = 0.09) was labelled with 1 mCi ;25I (Amersham) using 20 ~g chloramine T and 1 min reaction time at room temperature. The iodination reaction was stopped by addition of 48 pg Na-metabisulphite and 100 ~l normal rabbit serum. This solution was gel-filtered over a column of 30 ml Sephadex-G25 layered on 30 ml ACA-44 Ultrogel, equilibrated with diluent (PBS containing 3 mg/ml BSA and 1 mg/ml NaN3 pH 7.2). A single protein peak was obtained, apart from the free iodine peak. A stock tracer solution was prepared by diluting the peak material to 30 ml. For the assay, 0.5 ml of this stock solution was diluted with 10 ml diluent: 50 ~l of this solution was used per test; in the experiments reported here, this corresponded to 40,000 cpm in a Wallac gammaspectrometer.
Preparation of Sepharose-coupled ovalbumin OVA-Sepharose was prepared by reacting 0.2 mg OVA with 100 mg CNBrSepharose according to manufacturer's instructions. OVA-Sepharose was suspended in 100 ml diluent to which 0.01 M EDTA had been added.
89 RESULTS
Binding of ~25I-anti-TNP to TNP-DAE-Sepharose In order to det ect picomole amounts of haptenated antigen, the anti° hapten antibodies should be of high avidity. To test the feasibility of the approach, an upper limit estimate of the lower limit of detection of the antihapten antibodies was obtained. With the anti-TNP reagent used TNP in less than 1 ~l o f a suspension of 1 , m o l e TNP coupled to 1 ml CNBr-activated Sepharose suspended in 100 ml could be detected; thus the detection range for directly coupled TNP is in the picomole range. Half maximal binding was found with 10 gl of this suspension in 0.5 ml which corresponds to a molar TNP concentration of 2 )~ 10 -v M. As mentioned before, these estimates represent upper limits, since it is assumed that all TNP is bound and is available for the antibodies. Therefore, this anti-hapten reagent should in principle be useful for the detection of picomole or fem t om ol e amounts of haptenated antigens.
Haptenated AnUgen Bind=ng Assay
Antibody
Sepharose -
bound
+
+
(patient'S serum)
antigen
+
,,t;
- > ',
125 I- labelled + anti-hapten antibody
Fig. 1. Schematic representation of the principle of the test.
+
haptenated antigen
i f " :4 :
--
--)
90
Anti-OVA assay (fig. 1) This test is a modification of system 4 as described by Wide (1970) and Aalberse et al. {1974}. Up to 500 ul sample was incubated with 0.5 ml OVASepharose for 1--4 h at room temperature. After washing the Sepharose, 0.5 ml diluent containing in addition 0.01 M EDTA and 0.3% Tween-20 (incubation medium) was added, followed by the addition of 50 pl of 1 ug/ml TNPOVA (7 pmole TNP). The tubes were incubated for 1 h or overnight and washed. 0.5 ml incubation medium containing in addition 10% normal bovine serum and 1% normal rabbit serum was added, followed by the addition of 50 ~l ~:sI anti-TNP dilution. After another overnight incubation, the Sepharose was washed again and Sepharose-bound radioactivity was measured. All incubations were performed in rotating tubes. The dose-response curve for a rabbit anti-OVA antiserum containing 2.46 mg/ml precipitable antibody, is shown in fig. 2. When 100 ~l o f a 1 : 120,000 dilution of this antiserum was tested i.e. 2 ng of antibody with a theoretical binding capacity in this test of 14 fmole of TNP-OVA or 89 fmole of TNP, 6.3% of the added 40,000 cpm radioactivity was bound. When a negative serum was tested, 1.3% of radioactivity was bound. When similar tests were performed with directly l:sIlabelled OVA, using similarly 50 gl of 1 ~g/ml I:SI-OVA which yielded
Anti-ovalbumin assay using OVA-TNP and 1251-anti-TNP of added radioactivity 6O ~z
j / 40
20
I
0
L
I
L
L
I
50 ~1 rabbit anti-ovalbumin diluted 1:500
I
I
I
100
Fig. 2. Dose-response curve o f a rabbit a n t i - o v a l b u m i n a n t i s e r u m . On the o r d i n a t e is indic a t e d the p e r c e n t a g e o f radioactivity b o u n d to the S e p h a r o s e particles.
9] 13,000 cpm, 100 pl of a 1 : 5000 dilution of the rabbit antiserum (i.e. 49 ng antibody) was needed to obtain 5.9% binding of radioactivity; when a negative serum was tested, 1.4% of the radioactivity was bound. Thus, in the OVA system the haptenated antigen binding assay is at least an order of magnitude more sensitive than the iodinated antigen binding assay. The haptenated antigen binding assay was compared with a 12SI-labelled anti-IgG binding assay, using as control tests in which 50 ug/ml soluble antigen was included in the incubation medium. Serum samples from 20 laboratory workers were so tested; six were negative in both tests, and nine positive in both tests, 5 were positive only in the haptenated antigen binding assay. No antibodies were found with 12SI-labelled anti-IgA or anti-IgM. DISCUSSION The test described involves one more reaction step than either the ~2sIlabelled antigen binding test, or the labelled antiglobulin binding test; what then are the advantages? Compared with the 12SI-labelled antigen binding test, the most obvious advantage is that the use of labelled antigen is circumvented. This is clearly economical when a large number of antigens has to be tested over prolonged periods. Furthermore, efficient iodination of some antigens is difficult, or even impossible; ovalbumin for example is notoriously difficult to iodinate to high specific activity, and in fact, in a comparison of the haptenated antigen binding assay with an iodinated antigen binding assay, the former appeared to be more than 20-fold as sensitive when ovalbumin was tested as antigen. TNP-groups can easily be introduced into many antigen molecules, and other haptens may be even more generally applicable. Experiments using diazotized haptens are in progress. The specificity problems of the labelled antiglobulin binding test are not always fully appreciated. Non-specific binding to unsubstituted carrier is often different from that to antigen-substituted carrier; moreover, considerable differences in non-specific binding among different sera, not obviously related to their immunoglobulin content, are often encountered. When iodinated anti-IgG of high specific activity is used, the maximal volume of serum that can be used safely without specificity problems is less than 0.1 pl per test. Although the iodinated antiglobulin binding test is extremely sensitive with regard to the a m o u n t of antibody that can be detected, for detection of antibody in unfractionated serum non-specific binding of non-antibody immunoglobulin is the limiting factor. We have performed labelled antigen binding tests with a number of antigens: apart from ovalbumin, we have tested the milk proteins casein, a-lactalbumin and fl-lactoglobulin; a-gliadin from wheat; tetanus and diphtheria toxoid; and human IgG, IgA, and IgE. In none of these tests was non-specific binding of labelled antigen to Sepharose-bound antigen induced by control sera, even when 500 ul undiluted serum was used for testing. This low non-
92 specific binding is a major advantage of this type of test. In almost all antibody assays, the antigen is modified in some way: the antigen is either coupled to a large particle, or to a marker molecule or atom. It is often difficult to exclude the possibility that this modification reaction selectively inactivates some antigenic determinant(s). This limitation obviously holds also with respect to the hapten-labelling procedure. However, in double diffusion experiments a reaction of identity was found between TNP-OVA and unsubstituted OVA with the rabbit anti-OVA antiserum. Furthermore, no human sera were found to be positive in the antiglobulin-binding reaction, but negative in the TNP-OVA-binding reaction. In the haptenated antigen binding test described, the antibodies were extracted from the test sample with Sepharose-bound antigen. The antibodies can be extracted also by other procedures, e.g. with antiglobulin Sepharose or with protein A-Sepharose. The extraction with antigen-Sepharose has limited application in that only effectively polyvalent antibodies will be detected. In order to prevent bivalent antibodies combining with two antigen molecules on the same Sepharose bead, the antigen density on the beads should be kept low. Experimentally, a decreased binding of the labelled antigen was indeed observed when the antigen density on the Sepharose was increased. Extraction with antigen-Sepharose has the advantage that only antibody molecules are extracted; with antiglobulin-Sepharose, e.g. rabbit anti-human IgG antibodies coupled to Sepharose, all IgG molecules of the test sample are extracted. Thus, non-antibody IgG may interfere in the latter procedure. The practical difference between the procedures lies in the volume of serum from which the antibody can be extracted by a given a m o u n t of Sepharose. When the extraction is complete, washing before addition of the labelled antigen is not necessary. One practical advantage of the non-radioactive labelling procedure is not yet mentioned. When a component within a crude antigen extract is to be labelled, it may be much more efficient to label the whole extract, and to purify afterwards. During the purification procedure the antigen can be measured easily using Sepharose-coupled antibodies to the antigen and '2sIlabelled anti-hapten antibodies as reagents. Non-labelled proteins can be introduced for stabilization or for marking purposes. In this way wasteful concentration steps can often be avoided, whereas no manipulation with prohibitively large amounts of radioactivity is needed. ACKNOWLEDGEMENT The rabbit anti-ovalbumin antiserum was prepared by Miss B. de Lange, who also determined the concentration of precipitable antibody. '25I labelled antiglobulins were kindly provided by T.A. Out. This investigation was supported by the Netherlands Asthma Foundation.
93 REFERENCES Aalberse, R.C., E. van Loghem, P.J.J. van Munster and J.H.J. Nadorp, 1974, J. Lab. Clin. Med. 83,831. Hill, R.J., 1972, J. Immunol. Methods 1,231. Little, J.R. and H.N. Eisen, 1967, in: Methods in Immunology and Immunochemistry, eds. C.A. Williams and M.W. Chase {Academic Press, New York, London) vol. 1, p. 128. Wide, L., 1970, in: Radioimmunoassay methods, eds. K.E. Kirkham and W.M. Hunter (E. and S. Livingstone, Edinburgh) p. 405.
87
AN INDIRECT ANTIBODY ASSAY USING H A P T E N A T E D ANTIGEN AND '25I-LABELLED ANTI-HAPTEN ANTIBODY
R.C. AALBERSE Central Laboratory of the Netherlands Red Cross Blood Transfusion Service, Amsterdam, the Netherlands and the Laboratory of Experimental and Clinical Immunology of the University of Amsterdam, the Netherlands (Received 6 May 1977, accepted 30 June 1977)
Hapten (trinitrophenyl) was coupled to antigen (ovalbumin). The haptenated antigen was bound by anti-ovalbumin antibody and binding was quantitated with t2SI-labelled anti-hapten antibodies. Thus, with a single radioactive reagent antibodies against a variety of antigens can be detected, whilst the problems inherent in a labelled antiglobulin binding test are avoided. In the ovalbumin system, the haptenated antigen binding test proved to be approximately 20 times as sensitive as the iodinated ovalbumin binding test.
INTRODUCTION
Quantitative antibody assays are often based on the binding of either '2sIlabelled antigen or ~2SI-labelled antiglobulin. In the antiglobulin binding reaction, only a single labelled reagent is used; it is theoretically the most sensitive assay. However, non-specific uptake of immunoglobulins often sets the practical limit o f sensitivity. The use of '2SI-labelled antigen provides greater specificity and in practice higher sensitivity, but when antibodies against a n u mb er of different antigens are to be measured, all the antigens have to be labelled. In the present c om m uni c a t i on, an indirect labelled antigen binding technique is described, using a single radioactive reagent by which antibodies against different antigens can be measured. The antigen is labelled with a hapten, e.g. the trinitrophenyl (TNP)-group; binding of the haptenated antigen is quantitated with '25I-labelled anti-hapten antibodies. This test combines the sensitivity of the '2SI-labelled antigen binding tests with the versatility of the antiglobulin-binding test. An assay for antibodies to ovalbumin is described as an example. MATERIALS AND METHODS H a p t e n a t i o n o f p r o t e i n antigens
Trinitrophenylation of bovine serum albumin (BSA, crystallized, Sigma) and ovalbumin (OVA, 5X crystallized, Calbiochem} was perform ed with
88 trinitrobenzene sulfonic acid (TNBS, Baker) as described by Little and Eisen {1967). Fifty mg of TNBS was added to 500 mg protein in 50 ml 0.02 M sodium borate buffer pH 8.6. After 1 h reaction at room temperature the solution was dialyzed against phosphate buffered saline (PBS) for three days at 4°C. The optical absorbance at 345 nm of the dialyzed solution (calculated protein concentration 8 mg/ml) was 33.6 for TNP-BSA and 19.0 for TNPOVA, corresponding to TNP/protein ratios of 15.9 and 6.5 respectively.
Preparation of anti-hapten antibodies TNP-Sepharose was prepared by reacting 1 ~mol TNBS with 1 ml diaminoethane (DAE)-Sepharose at pH 8.6. DAE-Sepharose was prepared by reacting CNBr-activated Sepharose (Pharmacia) with an excess of diaminoethane (Merck) at pH 9; subsequently the excess of diaminoethane was removed. Anti-TNP antibody was raised in a rabbit by immunization with 0.5 mg TNP-BSA in complete Freund's adjuvant, followed after 5 weeks by a second injection with 0.05 mg TNP-BSA in incomplete Freund's adjuvant. Seven days later citrated plasma was obtained. To isolate anti-TNP antibodies, the plasma was defibrinated and inactivated by heating for 1 h at 56°C. Five ml defibrinated plasma was passed through a column consisting of 1 ml TNPSepharose mixed with approximately 5 ml Sephadex-G25, layered on approximately 35 ml Sephadex-G25 (total column dimensions 55 × 1 cm). The column was washed with 200 ml PBS and with 50 ml glycine buffer (0.1 M glycine--HC! in 0.15 M NaC1, pH 2.5); antibodies were eluted with glycine buffer--dioxan 9 : 1 (v/v) (Hill, 1972).
Labelling with ~2Sl 0.5 ml of the protein-containing neutral column fractions (A280 = 0.09) was labelled with 1 mCi ;25I (Amersham) using 20 ~g chloramine T and 1 min reaction time at room temperature. The iodination reaction was stopped by addition of 48 pg Na-metabisulphite and 100 ~l normal rabbit serum. This solution was gel-filtered over a column of 30 ml Sephadex-G25 layered on 30 ml ACA-44 Ultrogel, equilibrated with diluent (PBS containing 3 mg/ml BSA and 1 mg/ml NaN3 pH 7.2). A single protein peak was obtained, apart from the free iodine peak. A stock tracer solution was prepared by diluting the peak material to 30 ml. For the assay, 0.5 ml of this stock solution was diluted with 10 ml diluent: 50 ~l of this solution was used per test; in the experiments reported here, this corresponded to 40,000 cpm in a Wallac gammaspectrometer.
Preparation of Sepharose-coupled ovalbumin OVA-Sepharose was prepared by reacting 0.2 mg OVA with 100 mg CNBrSepharose according to manufacturer's instructions. OVA-Sepharose was suspended in 100 ml diluent to which 0.01 M EDTA had been added.
89 RESULTS
Binding of ~25I-anti-TNP to TNP-DAE-Sepharose In order to det ect picomole amounts of haptenated antigen, the anti° hapten antibodies should be of high avidity. To test the feasibility of the approach, an upper limit estimate of the lower limit of detection of the antihapten antibodies was obtained. With the anti-TNP reagent used TNP in less than 1 ~l o f a suspension of 1 , m o l e TNP coupled to 1 ml CNBr-activated Sepharose suspended in 100 ml could be detected; thus the detection range for directly coupled TNP is in the picomole range. Half maximal binding was found with 10 gl of this suspension in 0.5 ml which corresponds to a molar TNP concentration of 2 )~ 10 -v M. As mentioned before, these estimates represent upper limits, since it is assumed that all TNP is bound and is available for the antibodies. Therefore, this anti-hapten reagent should in principle be useful for the detection of picomole or fem t om ol e amounts of haptenated antigens.
Haptenated AnUgen Bind=ng Assay
Antibody
Sepharose -
bound
+
+
(patient'S serum)
antigen
+
,,t;
- > ',
125 I- labelled + anti-hapten antibody
Fig. 1. Schematic representation of the principle of the test.
+
haptenated antigen
i f " :4 :
--
--)
90
Anti-OVA assay (fig. 1) This test is a modification of system 4 as described by Wide (1970) and Aalberse et al. {1974}. Up to 500 ul sample was incubated with 0.5 ml OVASepharose for 1--4 h at room temperature. After washing the Sepharose, 0.5 ml diluent containing in addition 0.01 M EDTA and 0.3% Tween-20 (incubation medium) was added, followed by the addition of 50 pl of 1 ug/ml TNPOVA (7 pmole TNP). The tubes were incubated for 1 h or overnight and washed. 0.5 ml incubation medium containing in addition 10% normal bovine serum and 1% normal rabbit serum was added, followed by the addition of 50 ~l ~:sI anti-TNP dilution. After another overnight incubation, the Sepharose was washed again and Sepharose-bound radioactivity was measured. All incubations were performed in rotating tubes. The dose-response curve for a rabbit anti-OVA antiserum containing 2.46 mg/ml precipitable antibody, is shown in fig. 2. When 100 ~l o f a 1 : 120,000 dilution of this antiserum was tested i.e. 2 ng of antibody with a theoretical binding capacity in this test of 14 fmole of TNP-OVA or 89 fmole of TNP, 6.3% of the added 40,000 cpm radioactivity was bound. When a negative serum was tested, 1.3% of radioactivity was bound. When similar tests were performed with directly l:sIlabelled OVA, using similarly 50 gl of 1 ~g/ml I:SI-OVA which yielded
Anti-ovalbumin assay using OVA-TNP and 1251-anti-TNP of added radioactivity 6O ~z
j / 40
20
I
0
L
I
L
L
I
50 ~1 rabbit anti-ovalbumin diluted 1:500
I
I
I
100
Fig. 2. Dose-response curve o f a rabbit a n t i - o v a l b u m i n a n t i s e r u m . On the o r d i n a t e is indic a t e d the p e r c e n t a g e o f radioactivity b o u n d to the S e p h a r o s e particles.
9] 13,000 cpm, 100 pl of a 1 : 5000 dilution of the rabbit antiserum (i.e. 49 ng antibody) was needed to obtain 5.9% binding of radioactivity; when a negative serum was tested, 1.4% of the radioactivity was bound. Thus, in the OVA system the haptenated antigen binding assay is at least an order of magnitude more sensitive than the iodinated antigen binding assay. The haptenated antigen binding assay was compared with a 12SI-labelled anti-IgG binding assay, using as control tests in which 50 ug/ml soluble antigen was included in the incubation medium. Serum samples from 20 laboratory workers were so tested; six were negative in both tests, and nine positive in both tests, 5 were positive only in the haptenated antigen binding assay. No antibodies were found with 12SI-labelled anti-IgA or anti-IgM. DISCUSSION The test described involves one more reaction step than either the ~2sIlabelled antigen binding test, or the labelled antiglobulin binding test; what then are the advantages? Compared with the 12SI-labelled antigen binding test, the most obvious advantage is that the use of labelled antigen is circumvented. This is clearly economical when a large number of antigens has to be tested over prolonged periods. Furthermore, efficient iodination of some antigens is difficult, or even impossible; ovalbumin for example is notoriously difficult to iodinate to high specific activity, and in fact, in a comparison of the haptenated antigen binding assay with an iodinated antigen binding assay, the former appeared to be more than 20-fold as sensitive when ovalbumin was tested as antigen. TNP-groups can easily be introduced into many antigen molecules, and other haptens may be even more generally applicable. Experiments using diazotized haptens are in progress. The specificity problems of the labelled antiglobulin binding test are not always fully appreciated. Non-specific binding to unsubstituted carrier is often different from that to antigen-substituted carrier; moreover, considerable differences in non-specific binding among different sera, not obviously related to their immunoglobulin content, are often encountered. When iodinated anti-IgG of high specific activity is used, the maximal volume of serum that can be used safely without specificity problems is less than 0.1 pl per test. Although the iodinated antiglobulin binding test is extremely sensitive with regard to the a m o u n t of antibody that can be detected, for detection of antibody in unfractionated serum non-specific binding of non-antibody immunoglobulin is the limiting factor. We have performed labelled antigen binding tests with a number of antigens: apart from ovalbumin, we have tested the milk proteins casein, a-lactalbumin and fl-lactoglobulin; a-gliadin from wheat; tetanus and diphtheria toxoid; and human IgG, IgA, and IgE. In none of these tests was non-specific binding of labelled antigen to Sepharose-bound antigen induced by control sera, even when 500 ul undiluted serum was used for testing. This low non-
92 specific binding is a major advantage of this type of test. In almost all antibody assays, the antigen is modified in some way: the antigen is either coupled to a large particle, or to a marker molecule or atom. It is often difficult to exclude the possibility that this modification reaction selectively inactivates some antigenic determinant(s). This limitation obviously holds also with respect to the hapten-labelling procedure. However, in double diffusion experiments a reaction of identity was found between TNP-OVA and unsubstituted OVA with the rabbit anti-OVA antiserum. Furthermore, no human sera were found to be positive in the antiglobulin-binding reaction, but negative in the TNP-OVA-binding reaction. In the haptenated antigen binding test described, the antibodies were extracted from the test sample with Sepharose-bound antigen. The antibodies can be extracted also by other procedures, e.g. with antiglobulin Sepharose or with protein A-Sepharose. The extraction with antigen-Sepharose has limited application in that only effectively polyvalent antibodies will be detected. In order to prevent bivalent antibodies combining with two antigen molecules on the same Sepharose bead, the antigen density on the beads should be kept low. Experimentally, a decreased binding of the labelled antigen was indeed observed when the antigen density on the Sepharose was increased. Extraction with antigen-Sepharose has the advantage that only antibody molecules are extracted; with antiglobulin-Sepharose, e.g. rabbit anti-human IgG antibodies coupled to Sepharose, all IgG molecules of the test sample are extracted. Thus, non-antibody IgG may interfere in the latter procedure. The practical difference between the procedures lies in the volume of serum from which the antibody can be extracted by a given a m o u n t of Sepharose. When the extraction is complete, washing before addition of the labelled antigen is not necessary. One practical advantage of the non-radioactive labelling procedure is not yet mentioned. When a component within a crude antigen extract is to be labelled, it may be much more efficient to label the whole extract, and to purify afterwards. During the purification procedure the antigen can be measured easily using Sepharose-coupled antibodies to the antigen and '2sIlabelled anti-hapten antibodies as reagents. Non-labelled proteins can be introduced for stabilization or for marking purposes. In this way wasteful concentration steps can often be avoided, whereas no manipulation with prohibitively large amounts of radioactivity is needed. ACKNOWLEDGEMENT The rabbit anti-ovalbumin antiserum was prepared by Miss B. de Lange, who also determined the concentration of precipitable antibody. '25I labelled antiglobulins were kindly provided by T.A. Out. This investigation was supported by the Netherlands Asthma Foundation.
93 REFERENCES Aalberse, R.C., E. van Loghem, P.J.J. van Munster and J.H.J. Nadorp, 1974, J. Lab. Clin. Med. 83,831. Hill, R.J., 1972, J. Immunol. Methods 1,231. Little, J.R. and H.N. Eisen, 1967, in: Methods in Immunology and Immunochemistry, eds. C.A. Williams and M.W. Chase {Academic Press, New York, London) vol. 1, p. 128. Wide, L., 1970, in: Radioimmunoassay methods, eds. K.E. Kirkham and W.M. Hunter (E. and S. Livingstone, Edinburgh) p. 405.
