- Email: [email protected]
The importance of antibody quality in sandwich ELISA systems
Journal oflmmunologicalMethods,
87 (1986) 29-33
29
Elsevier JIM03820
The importance of antibody quality in sandwich ELISA systems Evaluation of selected commercial reagents J.G. Shields * a n d M.W. T u r n e r Department of Immunology, Institute of Child Health, 30 Guilford Street, London WC1N 1EH, U.K.
(Accepted 6 September1985)
The detection of low levels of immunoglobulin in cell culture supernatants etc. by ELISA procedures demands high sensitivity and absolute specificity. We have used commercially available antibodies prepared by positive affinity purification and immunoglobulin fractions obtained by ion-exchange chromatography in various combinations in ELISA systems for such quantitative analyses. These demonstrate that when affinity-purified reagents lacking unwanted anti-species cross-reactivity are used as both the capture antibody and the indicator antibody the results are markedly superior (high sensitivity, low backgrounds and a wide working range) compared to those obtained with all other possible combinations. Key words: EL1SA; Affinity-purified antibodies," Human lgG." Culture supernatants
Introduction
Among the many attributes of the enzyme-linked immunosorbent assay (Engvall and Perlmann, 1971; Voller et al., 1978) is its potential for detecting specific classes or subclasses of immunoglobulins at very low concentrations in a variety of biological fluids or cell culture supernatants (Butler et al., 1982). Although the choice of antisera is crucial, there have been relatively few investigations into the level of purity required for successful assays. Several companies have recently introduced highly specific affinity-purified anti-immunoglobulin preparations which make it possible for most laboratories to establish sensitive ELISAs without needing to undertake the difficult and laborious task of preparing and labelling affinitypurified antisera. We report here our experiences using a variety of commercially available anti-human IgG reagents as capture a n d / o r indicator * Correspondence to: J.G. Shields.
antibodies in high sensitivity ELISA systems. Some of the pitfalls which may be encountered when developing these assays will be highlighted.
Materials and methods A n tisera
The antisera used in this study were all purchased from commercial suppliers and are listed in Table I. The IgG fractions had been prepared by salt precipitation and DEAE ion-exchange chromatography. Affinity-purified antisera had been positively selected by binding to and elution from an insolubilized human IgG preparation. Indicator antibody was labelled directly with horseradish peroxidase (Nordic and Tago) or indirectly by reacting with univalent fragments of affinity-purified sheep anti-rabbit IgG labelled with HRPO just before use (Miab). All antisera were stored according to the manufacturer's recommendation.
0022-1759/86/$03.50 © 1986 Elsevier SciencePublishers B.V. (BiomedicalDivision)
30 TABLE I CHARACTERISTICS USED Nature
OF THE
ANTI-IgG
ANTISERA
Species Source
Fig. 2 ref. no.
Capture antibody Affinity purified Rabbit Miab (Uppsala, Sweden) Affinity purified Goat Tago (TCS, Slough, U.K.) IgG fraction Sheep Unipath (Bedford, U.K.) Indicator antibody a Affinity purified Rabbit Miab (Uppsala, Sweden) 4 Affinity purified Goat Sigma (Poole, U.K.) 5 Nordic (Maidenhead,U.K.) 6 lgG fraction Goat Labelled with horseradish peroxidase.
Standard IgG preparations Both whole serum and an IgG-rich fraction prepared from serum by Sephadex G-200 gel filtration and DEAE ion-exchange chromatography were used as standards. The two preparations were standardized using a nephelometric technique against the Calibration Standard for Specific Pro-
tein Assay (SPS-01, batch no. 851) supplied by the Sheffield Protein Reference Unit. There were no observable differences using these two preparations and only results obtained with the unfractionated normal human serum will be presented.
Plates Round-bottomed, polystyrene microtiter plates (Cooke) from Sterilin (cat. no. M24A) were used throughout.
ELISA protocol The ELISA procedure is outlined diagrammatically in Fig. 1. The plates were coated with the capture antibody by incubating overnight (at room temperature) with a solution of antibody in carbonate buffer pH 9.6 (80/zl/well of 10/~g/ml (Unipath) or 1 ttg/ml (Tago and Miab)). The optimum concentrations of capture antibody were determined by titration. After two washes with carbonate buffer, 100 /zl of 5% serum was added to each well for 1 h to reduce non-specific binding during subsequent steps. The serum used in this step and in the diluent for subsequent steps was
SENSITIVE ELISA SYSTEM FOR IgG DETERMINATION (1) Capture antibody added to plate Incubate overnight Wash (2) Blocking serum added Incubate 1 h Wash
Y
Y
Y
. Y..Y..Y.
(3) Add standards (1 3000 ng/ml) diluted in PBS/Tween/5% serum Incubate 90 min Wash (4) Add HRPO-anti-IgG diluted in PBS/Tween/5% serum at recommended dilution Incubate 90 rain Wash (5) Add substrate (OPD/H202) diluted in phosphate/citrate buffer Incubate 20 min (6) Read absorbance at 492 nm on Titertek plate reader
Fig. 1. Flow diagram of steps involved in ELISA system for IgG determination (see text for further details).
31
from the same species of animal as the indicator antibody, i.e., goat serum when Sigma and Nordic preparations were used and rabbit serum when
6.0-
Miab reagents were used. The plates could be stored at 4°C either before or after this blocking stage for at least 4 weeks without any appreciable
Ca)
(b) 6.0.
/ S
tn
4.0-
.~ 4.0c
P~
<-
2.0
2.0" A
10
100 IgG ( ng / ml )
10'00
5000
10 100 IgG ( n g / m l )
1.5 15
1000
5000
(d)
(c)
~ 1.0 10
g
<~
0.5
0.5
&
10 IgG
100 ( ng/ml)
1000
5000
10
100 ~gG ( ng /rnl )
1000
5000
Fig. 2. Absorbance curves obtained with IgG standards (1-3000 ng/ml) and various combinations of capture and indicator antibodies. A, each symbol represents the mean of duplicate determinations (the standard deviations were all less than 10%; o, mean of duplicate determination of a 'zero' standard (buffer alone at step 3 in Fig. 1). a: Capture antibody: IgG fraction (no. 3 in table)-Indicator antibody: lgG fraction (no. 6 in table) b: Capture antibody: affinity purified (no. 2 in table)-Indicator antibody: IgG fraction (no. 6 in table) c: Capture antibody: IgG fraction (no. 3 in table)-Indicator antibody: affinity purified (no. 5 in table) d: Capture antibody: affinity purified (no. 1 in table)-Indicator antibody: affinity purified (no. 4 in table) Adding the 3000 ng/ml standard to an uncoated well in each system gave absorbance values of 0.18, 0.19, 0.01 and 0.01 respectively in systems a - d .
32 loss of activity. Two washes with 0.05 M PBS (pH 7.2)/0.05% Tween 20 were followed by addition of the standards (80 /~l/well) diluted in P B S / Tween/5% serum. After 90 min incubation at room temperature the plates were washed 5 times with PBS/Tween before the addition of indicator antibody (80 ~l/well) at a dilution of 1 : 1000 in P B S / T w e e n / 5 % serum. A further 90 rain incubation followed, then the plates were washed 5 times and freshly prepared substrate o-phenylenediamine (0.5 m g / m l , 80 /~l/well) in 0.05 M citrate/0.1 M phosphate buffer, pH 5.0 + 0.015% H~O~ was added. After 15 min the reaction was stopped by the addition of 2 M H2SO 4, (40 7~l/well) and absorbance readings measured on a Titertek plate reader at a wavelength of 492 nm. Standards were assayed in duplicate over the range 1 to 3000 n g / m l , and standard curves plotted of absorbance versus log IgG concentration.
Results
All four possible combinations of capture and indicator antibody (i.e., IgG fraction or affinitypurified sandwiched with IgG fraction or affinity purified) are presented in Fig. 2. When IgG fractions were used on either side of the sandwich very high backgrounds were obtained and no discernible curve was apparent (Fig. 2a). Although this type of antibody preparation can be used successfully, we have illustrated here one possible problem where one antibody preparation is recognising the other. In this case we believe it is the goat enzyme-labelled antibody which recognises the sheep capture antibody since the excess goat serum present in the diluent does not reduce the background values. When an affinity-purified capture antibody was used the background was appreciably reduced and a standard curve was obtained (Fig. 2b). Most workers would, however, still find the background too high and the working range (2 60 n g / m l approx.) limited. When the enzymelabelled affinity-purified antibody was used in conjunction with the IgG fraction as capture antibody, the background was much lower but the sensitivity of the assay was now around 100 200 n g / m l and again the range was restricted (Fig. 2c). When affinity-purified reagents were used
throughout the background remained low, the sensitivity was much improved and the working range was wide (2 500 n g / m l ) (Fig. 2d). The interassay coefficient of variation for this final combination of reagents was 9.9% (10 consecutive assays).
Discussion
Theoretically, the use of positively selected affinity-purified antibodies should result in highly specific and extremely sensitive immunoassays. However, either for reasons of economy or lack of the appropriate reagent many investigators resort to using less pure preparations. Often, these will also contain antibodies that cross-react with other components in the system. We have found this to be the case when using IgG fractions of anti-human IgG antisera: unwanted anti-species cross-reactivity was present in IgG fractions but was not apparent when affinity-purified antisera were employed. Nevertheless when developing any ELISA system it is prudent, even with affinity-purified reagents, to control for unwanted cross-reactions. On those occasions when IgG fractions of antisera did give low background values (Fig. 2c) it was found that the sensitivity of the assay was 50 100-fold less than when only affinity-purified preparations were used (Fig. 2d). The blocking agent used in these high sensitivity assays is important. Our final choice was based on the assumption that, if either the capture antibody or the bound human IgG was going to cross-react with the indicator antibody, the presence of an excess of serum from the same species as that from which the indicator antibody was derived would block this unwanted binding of enzyme-labelled antibody. Initially, we used human serum albumin to block non-specific binding but the small amounts of IgG and IgA present in HSA interfered with the assay and made this an unsuitable choice. H u m a n serum albumin was, however, suitable for use in IgE assays. Similarly, we did not use bovine serum albumin since this also contains traces of bovine IgG which theoretically could have interfered in the assay even when affinity-purified antisera were used. Although we have only presented results for
33
measuring IgG levels, we have also developed systems for measuring low levels of IgA, IgE and lgM using the appropriate affinity-purified antisera. The techniques for detecting lgA and IgM are very similar to that described for IgG and are sensitive down to 3 n g / m l and 1 n g / m l respectively. Even lower levels of IgE can be measured ( > 0.4 n g / m l ) but this requires longer incubation times. Affinity-purified antisera can also be used as indicator antibodies to detect either subclass-specific or antigen-specific antibodies in serum or cell-culture supernatants'and have proved more sensitive
in our hands than IgG fractions in detecting antibodies to a wide variety of food antigens.
References Butler, M., D.J. Atherton and R.J. Levinsky, 1982, Clin. Exp. Immunol. 50, 92. Engvall, E. and P. Perlmann, 1971, Immunochemistry 8, 871. Voller, A., A. Bartlett and D.E. Bidwell, 1978, J. Clin. Pathol. 31,507.
87 (1986) 29-33
29
Elsevier JIM03820
The importance of antibody quality in sandwich ELISA systems Evaluation of selected commercial reagents J.G. Shields * a n d M.W. T u r n e r Department of Immunology, Institute of Child Health, 30 Guilford Street, London WC1N 1EH, U.K.
(Accepted 6 September1985)
The detection of low levels of immunoglobulin in cell culture supernatants etc. by ELISA procedures demands high sensitivity and absolute specificity. We have used commercially available antibodies prepared by positive affinity purification and immunoglobulin fractions obtained by ion-exchange chromatography in various combinations in ELISA systems for such quantitative analyses. These demonstrate that when affinity-purified reagents lacking unwanted anti-species cross-reactivity are used as both the capture antibody and the indicator antibody the results are markedly superior (high sensitivity, low backgrounds and a wide working range) compared to those obtained with all other possible combinations. Key words: EL1SA; Affinity-purified antibodies," Human lgG." Culture supernatants
Introduction
Among the many attributes of the enzyme-linked immunosorbent assay (Engvall and Perlmann, 1971; Voller et al., 1978) is its potential for detecting specific classes or subclasses of immunoglobulins at very low concentrations in a variety of biological fluids or cell culture supernatants (Butler et al., 1982). Although the choice of antisera is crucial, there have been relatively few investigations into the level of purity required for successful assays. Several companies have recently introduced highly specific affinity-purified anti-immunoglobulin preparations which make it possible for most laboratories to establish sensitive ELISAs without needing to undertake the difficult and laborious task of preparing and labelling affinitypurified antisera. We report here our experiences using a variety of commercially available anti-human IgG reagents as capture a n d / o r indicator * Correspondence to: J.G. Shields.
antibodies in high sensitivity ELISA systems. Some of the pitfalls which may be encountered when developing these assays will be highlighted.
Materials and methods A n tisera
The antisera used in this study were all purchased from commercial suppliers and are listed in Table I. The IgG fractions had been prepared by salt precipitation and DEAE ion-exchange chromatography. Affinity-purified antisera had been positively selected by binding to and elution from an insolubilized human IgG preparation. Indicator antibody was labelled directly with horseradish peroxidase (Nordic and Tago) or indirectly by reacting with univalent fragments of affinity-purified sheep anti-rabbit IgG labelled with HRPO just before use (Miab). All antisera were stored according to the manufacturer's recommendation.
0022-1759/86/$03.50 © 1986 Elsevier SciencePublishers B.V. (BiomedicalDivision)
30 TABLE I CHARACTERISTICS USED Nature
OF THE
ANTI-IgG
ANTISERA
Species Source
Fig. 2 ref. no.
Capture antibody Affinity purified Rabbit Miab (Uppsala, Sweden) Affinity purified Goat Tago (TCS, Slough, U.K.) IgG fraction Sheep Unipath (Bedford, U.K.) Indicator antibody a Affinity purified Rabbit Miab (Uppsala, Sweden) 4 Affinity purified Goat Sigma (Poole, U.K.) 5 Nordic (Maidenhead,U.K.) 6 lgG fraction Goat Labelled with horseradish peroxidase.
Standard IgG preparations Both whole serum and an IgG-rich fraction prepared from serum by Sephadex G-200 gel filtration and DEAE ion-exchange chromatography were used as standards. The two preparations were standardized using a nephelometric technique against the Calibration Standard for Specific Pro-
tein Assay (SPS-01, batch no. 851) supplied by the Sheffield Protein Reference Unit. There were no observable differences using these two preparations and only results obtained with the unfractionated normal human serum will be presented.
Plates Round-bottomed, polystyrene microtiter plates (Cooke) from Sterilin (cat. no. M24A) were used throughout.
ELISA protocol The ELISA procedure is outlined diagrammatically in Fig. 1. The plates were coated with the capture antibody by incubating overnight (at room temperature) with a solution of antibody in carbonate buffer pH 9.6 (80/zl/well of 10/~g/ml (Unipath) or 1 ttg/ml (Tago and Miab)). The optimum concentrations of capture antibody were determined by titration. After two washes with carbonate buffer, 100 /zl of 5% serum was added to each well for 1 h to reduce non-specific binding during subsequent steps. The serum used in this step and in the diluent for subsequent steps was
SENSITIVE ELISA SYSTEM FOR IgG DETERMINATION (1) Capture antibody added to plate Incubate overnight Wash (2) Blocking serum added Incubate 1 h Wash
Y
Y
Y
. Y..Y..Y.
(3) Add standards (1 3000 ng/ml) diluted in PBS/Tween/5% serum Incubate 90 min Wash (4) Add HRPO-anti-IgG diluted in PBS/Tween/5% serum at recommended dilution Incubate 90 rain Wash (5) Add substrate (OPD/H202) diluted in phosphate/citrate buffer Incubate 20 min (6) Read absorbance at 492 nm on Titertek plate reader
Fig. 1. Flow diagram of steps involved in ELISA system for IgG determination (see text for further details).
31
from the same species of animal as the indicator antibody, i.e., goat serum when Sigma and Nordic preparations were used and rabbit serum when
6.0-
Miab reagents were used. The plates could be stored at 4°C either before or after this blocking stage for at least 4 weeks without any appreciable
Ca)
(b) 6.0.
/ S
tn
4.0-
.~ 4.0c
P~
<-
2.0
2.0" A
10
100 IgG ( ng / ml )
10'00
5000
10 100 IgG ( n g / m l )
1.5 15
1000
5000
(d)
(c)
~ 1.0 10
g
<~
0.5
0.5
&
10 IgG
100 ( ng/ml)
1000
5000
10
100 ~gG ( ng /rnl )
1000
5000
Fig. 2. Absorbance curves obtained with IgG standards (1-3000 ng/ml) and various combinations of capture and indicator antibodies. A, each symbol represents the mean of duplicate determinations (the standard deviations were all less than 10%; o, mean of duplicate determination of a 'zero' standard (buffer alone at step 3 in Fig. 1). a: Capture antibody: IgG fraction (no. 3 in table)-Indicator antibody: lgG fraction (no. 6 in table) b: Capture antibody: affinity purified (no. 2 in table)-Indicator antibody: IgG fraction (no. 6 in table) c: Capture antibody: IgG fraction (no. 3 in table)-Indicator antibody: affinity purified (no. 5 in table) d: Capture antibody: affinity purified (no. 1 in table)-Indicator antibody: affinity purified (no. 4 in table) Adding the 3000 ng/ml standard to an uncoated well in each system gave absorbance values of 0.18, 0.19, 0.01 and 0.01 respectively in systems a - d .
32 loss of activity. Two washes with 0.05 M PBS (pH 7.2)/0.05% Tween 20 were followed by addition of the standards (80 /~l/well) diluted in P B S / Tween/5% serum. After 90 min incubation at room temperature the plates were washed 5 times with PBS/Tween before the addition of indicator antibody (80 ~l/well) at a dilution of 1 : 1000 in P B S / T w e e n / 5 % serum. A further 90 rain incubation followed, then the plates were washed 5 times and freshly prepared substrate o-phenylenediamine (0.5 m g / m l , 80 /~l/well) in 0.05 M citrate/0.1 M phosphate buffer, pH 5.0 + 0.015% H~O~ was added. After 15 min the reaction was stopped by the addition of 2 M H2SO 4, (40 7~l/well) and absorbance readings measured on a Titertek plate reader at a wavelength of 492 nm. Standards were assayed in duplicate over the range 1 to 3000 n g / m l , and standard curves plotted of absorbance versus log IgG concentration.
Results
All four possible combinations of capture and indicator antibody (i.e., IgG fraction or affinitypurified sandwiched with IgG fraction or affinity purified) are presented in Fig. 2. When IgG fractions were used on either side of the sandwich very high backgrounds were obtained and no discernible curve was apparent (Fig. 2a). Although this type of antibody preparation can be used successfully, we have illustrated here one possible problem where one antibody preparation is recognising the other. In this case we believe it is the goat enzyme-labelled antibody which recognises the sheep capture antibody since the excess goat serum present in the diluent does not reduce the background values. When an affinity-purified capture antibody was used the background was appreciably reduced and a standard curve was obtained (Fig. 2b). Most workers would, however, still find the background too high and the working range (2 60 n g / m l approx.) limited. When the enzymelabelled affinity-purified antibody was used in conjunction with the IgG fraction as capture antibody, the background was much lower but the sensitivity of the assay was now around 100 200 n g / m l and again the range was restricted (Fig. 2c). When affinity-purified reagents were used
throughout the background remained low, the sensitivity was much improved and the working range was wide (2 500 n g / m l ) (Fig. 2d). The interassay coefficient of variation for this final combination of reagents was 9.9% (10 consecutive assays).
Discussion
Theoretically, the use of positively selected affinity-purified antibodies should result in highly specific and extremely sensitive immunoassays. However, either for reasons of economy or lack of the appropriate reagent many investigators resort to using less pure preparations. Often, these will also contain antibodies that cross-react with other components in the system. We have found this to be the case when using IgG fractions of anti-human IgG antisera: unwanted anti-species cross-reactivity was present in IgG fractions but was not apparent when affinity-purified antisera were employed. Nevertheless when developing any ELISA system it is prudent, even with affinity-purified reagents, to control for unwanted cross-reactions. On those occasions when IgG fractions of antisera did give low background values (Fig. 2c) it was found that the sensitivity of the assay was 50 100-fold less than when only affinity-purified preparations were used (Fig. 2d). The blocking agent used in these high sensitivity assays is important. Our final choice was based on the assumption that, if either the capture antibody or the bound human IgG was going to cross-react with the indicator antibody, the presence of an excess of serum from the same species as that from which the indicator antibody was derived would block this unwanted binding of enzyme-labelled antibody. Initially, we used human serum albumin to block non-specific binding but the small amounts of IgG and IgA present in HSA interfered with the assay and made this an unsuitable choice. H u m a n serum albumin was, however, suitable for use in IgE assays. Similarly, we did not use bovine serum albumin since this also contains traces of bovine IgG which theoretically could have interfered in the assay even when affinity-purified antisera were used. Although we have only presented results for
33
measuring IgG levels, we have also developed systems for measuring low levels of IgA, IgE and lgM using the appropriate affinity-purified antisera. The techniques for detecting lgA and IgM are very similar to that described for IgG and are sensitive down to 3 n g / m l and 1 n g / m l respectively. Even lower levels of IgE can be measured ( > 0.4 n g / m l ) but this requires longer incubation times. Affinity-purified antisera can also be used as indicator antibodies to detect either subclass-specific or antigen-specific antibodies in serum or cell-culture supernatants'and have proved more sensitive
in our hands than IgG fractions in detecting antibodies to a wide variety of food antigens.
References Butler, M., D.J. Atherton and R.J. Levinsky, 1982, Clin. Exp. Immunol. 50, 92. Engvall, E. and P. Perlmann, 1971, Immunochemistry 8, 871. Voller, A., A. Bartlett and D.E. Bidwell, 1978, J. Clin. Pathol. 31,507.