Journal of lmmunological Methods, 132 (1990) 73-80 Elsevier
73
JIM05659
ELISA for quantitation of the extracellular domain of p185 HER2 in biological fluids Patricia E. Sias 1, Claire E. Kotts 2, David Vetterlein 3, Mike Shepard 4 and Wai Lee T. W o n g 1 J Department of Immunology Research and Assay Technologies, 2 Department of Medicinal and Analytical Chemistry, 3 Department of Process Recovery, and ~ Department of Developmental Biology, Genentech Inc., So. San Francisco, CA 94080, U.S.A.
(Received 7 February 1990, revised received 30 April 1990, accepted l May 1990)
The HER2/neu proto-oncogene encodes a receptor that belong to the tyrosine-specific protein kinase family. Amplification of the HER2 gene in patients with breast and ovarian cancer has been shown to predict poorer survival rates. In order to understand the role of HER2 in malignant and normal cells, it is necessary to devise assays that can quantitate expression levels of the HER2 gene product (p185 HER2) in production samples, biopsy specimens and biological fluids. We have developed a simple, quantitative ELISA that uses two monoclonal antibodies directed against the extracellular domain of the HER2 gene product, p185 HER2 (HER2 ECD). The assay has a detection range of 0.25-120 ng/ml, is precise and sensitive. The ability of this assay to detect biologically active rHER2 ECD is demonstrated by its correlation to a growth inhibitory bioassay (r = 0.92). The sandwich ELISA can also accurately quantitate rHER2 ECD in mouse and monkey serum. This assay should be useful for quantitating low levels of circulating rHER2 ECD in animals in which rHER2 ECD is being used as antigen for immunotherapy and in patients which 'shed' receptor. Key words: HER2/neu; c-erbB-2; ELISA; Bioassay
Introduction
HER2 (c-erbB-2, neu) is a proto-oncogene which encodes a glycoprotein with extensive homology to the human epidermal growth factor receptor (Coussens et al., 1985; King et al., 1985; Schechter et al., 1985; Semba et al., 1985; Yamarnoto et al., 1986). It is encoded by a 4.8 kb
Correspondence to: W.L.T. Wong, Immunology Research and Assay Technologies, Genentech Inc., 460 Point San Bruno Boulevard, South San Francisco, CA 94080, U.S.A. Abbreviations: ELISA, enzyme-linked immunosorbent assay; mAb, monoelonal antibody; ECD, extracellular domain; HRPO, horseradish peroxidase, DEAE, diethylaminoethyl; BSA, bovine serum albumin; SDS, sodium dodecyl sulfate.
messenger RNA that is expressed in a variety of normal and malignant tissues (Kearney et al., 1979; King et al., 1985; Semba et al., 1985; Fukushige et al., 1986; Yamamato et al., 1986; Van de Vijver et al., 1987). The gene is found on q21 of chromosome 17 in humans (Coussens et al., 1985). The gene product, p185 HER2, is a transmembrane protein and possesses intrinsic tyrosine kinase activity (Yarden et al., 1988). It is believed to be a receptor although a iigand for the receptor has not as yet been identified. HER2 amplification has been correlated with a negative prognosis and high probability of relapse in patients with primary breast adenocarcinoma ( S i m o n et al., 1987, 1989). Overexpression of the related rat protooncogene, neu, in transgenic mice results in tumor development in 100% of males and females
0022-1759/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)
74 (Muller et al., 1988). In addition, a monoclonal antibody against the extracellular domain of p185 tiER2 has been shown to inhibit the in vitro proliferation of human breast tumor cells which overexpress p185 HER2 (Hudziak et al., 1989). The present study describes the development of a sensitive and specific enzyme-linked immunosorbent assay (ELISA) for quantitating the extracellular domain of the gene product of HER2 (HER2 ECD) in production samples and biological fluids. A sensitive ELISA will permit the dilution of serum to eliminate potential interfering substances and thus can be used to detect HER2 ECD in serum samples from animals participating in pharmacology studies.
Material and methods
Anti-HER2 ECD antibodies and enzyme-conjugated antibodies Monoclonal antibodies (mAb) directed against the extracellular domain of p185 aER2 were prepared by immunizing mice with N I H 3 T 3 / H E R 2 3400 cells expressing high levels of p185 HER2 (Hudziak et al., 1988; Fendly et al., 1989). The mAbs were purified from ascites fluid using protein A-Sepharose (Repligen Corp., Cambridge MA.) following established procedures (Ey et al., 1978; Goding, 1978) and stored sterile in 0.01 M sodium phosphate, 0.15 M sodium chloride, pH 7.2 (PBS) at 4°C. Purified mAbs were conjugated to horseradish peroxidase (HRPO) (Nakane and Kawaoi, 1974) and stored at - 2 0 ° C in PBS containing 5 m g / m l bovine serum albumin (BSA). ELISA assay standards rHER2 ECD (residues 1-624) has been cloned and stably expressed in transfected C H O cells (Hudziak et al., in preparation). Harvest fluid containing rHER2 ECD was first concentrated by ultrafiltration and then purified by immunoaffinity chromatography. Trace contaminants were removed by DEAE ion exchange chromatography. The purified rHER2 ECD amino acid composition matched what was theoretically expected for the cloned gene product and was homogeneous based on analysis of SDS gel. The concentration of rHER2 ECD in a purified preparation was
established by quantitative amino acid analysis. Dilutions of this sample in PBS containing per liter, 5 g of BSA, 0.5 ml of Tween 20 and 0.1 g of thimerosal (assay buffer) were then used as standards in the buffer based ELISA. For serum based ELISA, the. dilutions were prepared in the corresponding serum.
Samples Samples containing various concentrations of rHER2 ECD were obtained at different stages of purification including harvested tissue culture supernatants. Samples with high concentrations of rHER2 ECD were diluted further with assay buffer and re-quantified. ELISA assay procedure 96-well microtiter plates (Maxisorb, Nunc, Kamstrup, Denmark) were coated with mAb 7F3 by incubating overnight at 4°C with 100/~l/well of antibody at 5 ~tg/ml in 50 mmol/1 of sodium carbonate buffer, pH 9.6 (coat buffer). After removal of the coating solution, the coated plates were blocked with 150 ~tl/well of 5 g/1 of BSA in PBS for 1 h at room temperature, and washed six times with 0.5 g/1 of Tween 20 in PBS (wash buffer). Standards and samples were dispensed onto the coated wells (100/d/well). Plates were sealed and incubated at room temperature for 2 h with gentle agitation. Plates were washed six times with wash buffer. Horseradish peroxidase (HRPO)-labelled mAb 4D5 was then added (100 /~l/well). After further incubation for 1 h at room temperature, the plates were washed six times with wash buffer. Freshly prepared substrate solution (0.4 g of ophenylenediamine dihydrochloride in 1 liter of PBS plus 0.4 ml of 30% hydrogen peroxide) was added to the plates (100/~l/well) and incubation carried out in the dark for 15 min at room temperature. The reaction was stopped by the addition of 100/~1 of 2.25 mol/1 sulfuric acid and absorbance at 490 nm determined on a Vmax plate reader (Molecular Devices, Menlo Park, CA). A standard curve was generated by plotting absorbance vs. log of rHER2 ECD concentration, using a 4-parameter nonlinear regression curve fitting program (developed at Genentech). Sample concentrations
75
were obtained by interpolation sorbance on the standard curve.
of
their
ab-
cients of variation were 5.8% and 13.3%, respectively.
Bioassay method The bioassay used here is a modification of that described for assessing growth inhibitory effects of m A b s to H E R 2 E C D (Hudziak et al., 1989). In this case, we used the bioassay to measure the ability of H E R 2 E C D to block the inhibitory effects of one m A b , 4D5-. H u m a n breast carcinoma cells, SK-BR-3 (ATCC, Rockville, M D ) were maintained in a 1 : 1 mixture of Dulbecco's modified Eagle's medium (high glucose) and H a m ' s F12 medium containing 2 m M glutamine, 100 U penicillin/ml, 100 #g s t r e p t o m y c i n / m l and 10% fetal bovine serum (FBS, Gibco, G r a n d Island, NY). All incubations were carried out at 37°C, 5% CO 2 and 100% humidity. Cells (0.1 ml) were plated onto 96-well microtiter plates (Falcon) at a density of 5 x 10 4 c e l l s / c m 2 in medium containing 2% FBS (assay medium). After 2 h of attachment, 0.1 ml of appropriate test medium was applied to each well and incubation carried out for 4 days. The assay was terminated by gently rinsing cells twice with 0.2 ml PBS and staining with 0.05 ml 0.5% crystal violet ( v / v , in methanol) for 15 rain at ambient temperature. Stained cell monolayers were then rinsed with water and dried. Methanol (0.1 ml) was added to each well to dissolve the dye and absorbance quantitated at 540 nm using an automated plate reader (Bio-Tek Instruments, Model EL309 Burlington, VT). Test media containing various concentrations of H E R 2 E C D were prepared by serially diluting stock H E R 2 E C D in assay medium containing 2 ttg of m A b 4 D 5 / m l . Since 0.1 ml of test medium was added to 0.1 ml of cells, the final levels of medium constituents were one-half that of test medium. Controls received H E R 2 E C D in assay medium without added 4D5. Results are expressed as percent control, r H E R 2 E C D concentrations in unknown samples were quantitated by interpolation of responses on a standard curve of a reference r H E R 2 ECD. All samples were assayed at several dilutions in triplicate, only those dilutions whose responses fell in the linear range of the standard curve were used for quantitation. The intra- and interassay coeffi-
Results
Selection of coat and conjugated antibody Ten mAbs to H E R 2 E C D have been characterized and described elsewhere (Hudziak et al., 1989; Fendly et al., 1990). Based on their reactivities with immobilized p185 HER2, immunoprecipitation patterns, and ability to bind to SK-BR-3 cells, four of them (4D5, 7F3, 2 H l l , 2C4) with varying degrees of reactivities and immunoprecipitation patterns but high SK-BR-3 binding ability were conjugated to H R P O . All ten mAbs were then tested as coat antibody in a sandwich ELISA using these four conjugates. The pair that gave the most sensitive assay for r H E R 2 E C D were 7F3 as coat and 4D5 as H R P O conjugated antibody (data not shown). Consequently these were chosen for ELISA assay development.
Assay range and sensitioity The assay range for the m A b ELISA was determined to be 0.25 to 120 n g / m l (Fig. 1). The lower limit of detection (sensitivity) as defined by Rodbard (Rodbard et al., 1978) is equivalent to the concentration corresponding to the mean absorbance of the zero plus twice the standard devia-
2.0
E t-o 03
o E
1.0
o ..Q < 0.0
...... .1
I 1
........
I 10
........
I
........
100
t
1000
rHEN2 EOD (ng/mL) Fig. 1. Comparison of ELISA standard curves for r H E R 2 E C D in buffer ( o o), mouse serum (n D) and monkey serum (O e).
76 1oo
TABLE I ._I
CROSS-REACTIVITY Protein tested
rhGH-binding protein rCD4 rgpl20
~E Concentration
Cross
Tested
Measured
reactivity "
(#g/ml)
(ng/mi)
(%)
10.0 10.0 5.0
< 0.1 < 0.1 < 0.1
< 0.01 < 0.01 < 0.01
a Percent cross-reactivity was calculated as concentration meas u r e d / a m o u n t tested × 100.
80
,
..o czl 0 w r~r w
40
20
0
,
0
tion and was determined to be 0.14 n g / m l in buffer.
Assay specificity The specificity of this assay was tested with two other soluble receptors (rCD4 and rhGH-binding protein) and an unrelated protein produced in C H O cells (HIV envelope protein, gpl20). All three proteins were obtained from Genentech. The result (Table I) showed that the assay has less than 0.01% cross-reactivity with these proteins.
Samples with low, medium, or high rHER2 ECD protein concentrations in buffer were analyzed in 20 replicates for the assessment of intra-assay precision (Table IIA). Interassay precision was determined by measuring samples of low, medium or high rHER2 ECD concentrations in ten separate experiments (Table liB). The coefficients of intra-assay variation at all three levels
TABLE II PRECISION O F T H E mAb-BASED ELISA Sample 1
Sample 2
Sample 3
20 16.38 0.54 3.3
20 66.06 4.89 7.4
10 21.47 0.75 3.5
10 79.85 5.43 6.8
(,4) Intra-assay precision 20 3.32 0.08 2.4
(B) Interassay precision Replicates Mean ( n g / m l ) SD ( n g / m l )
c v (%)
i
20
,
,
i
40
,
,
i
60
.
,
i
80
,
,
i
1 O0
Theoretical calculation (ng/mL) Fig. 2. Plot of theoretical rHER2 ECD concentration based on dilution of known concentration vs. rHER2 ECD concentration measured by buffer based ELISA for tissue culture supernatant (O) ( r = 0.999; y = 0.992x +0.564), and purified sample (©) (r =1.00; y = 0.988x +0.956).
ranged from 2.4-7.4%, while the coefficients of interassay variation ranged from 3.5%-6.8%.
Linearity of the assay
Assay precision
Replicates Mean ( n g / m l ) SD ( n g / m l ) CV (%)
,
10 4.22 0.17
4.0
The linearity of the assay was determined by making serial two-fold dilutions of two samples, one a purified sample in assay buffer and the other a tissue culture supematant, and measuring the concentration of rHER2 ECD. Results were examined by correlating the observed concentration determined in the ELISA with the calculated concentration obtained by multiplying dilution factor with the concentration of the undiluted sample determined in the ELISA. Linear regression analysis for both samples resulted in correlation coefficients of 0.99 or greater (Fig. 2), indicating that the assay is linear between the assay range, i.e., 0.25-120 n g / m l .
Spike recovery Purified rHER2 ECD at a concentration of 37.5 n g / m l in assay buffer was added in equal volume to media samples diluted 1 / 5 or 1 / 1 0 in assay buffer. These samples were assayed in the ELISA. The theoretical concentration was calculated for each mixed sample and was used to calculate percent recovery. The data in Table III show an average recovery of 98.8% with a range of
77 TABLE llI BUFFERED-BASED ELISA: A C C U R A C Y a Concentration ( n g / m l )
Recovery
Purified sample b
Media sample b
Expected ¢
Observed
(%)
37.5
203.0 in 1 / 5 media 37,5 1.7
120.3 37.5 19.6
126.8 36.3 17.0
105.4 96.8 86.7
37.5
119.0 in 1 / 1 0 media 7.1 0.7
78.3 22.3 19.1
75.6 22.7 20.1
96.6 101.8 105.2
Average = 98.8% Range = 86.7-105.4% a Equal volumes of two purified samples (column no. 1) were each added to based samples in varying concentration of media (column no. 2) and assayed in ELISA. b rHER2 ECD concentrations had been determined previously by ELISA. c (spiked + based)/2.
86.7-105.4%, demonstrating the accuracy of the assay.
shows results obtained by measuring 18 different rHER2 ECD samples in both assays.
Correlation of E L I S A and bioassay To determine the ability of the rnAb based ELISA to quantitate active rHER2 ECD, eighteen samples were assayed in the ELISA and results compared with those obtained by bioassay. The bioassay used here measures the ability of added HER2 ECD preparations to compete with native HER2 receptor on SK-BR-3 cells for mAb 4D5 binding. When no HER2 ECD is present in culture medium, 4D5 greatly inhibits cell growth (Fig. 3). Maximal inhibition occurred at a concentration of 1 /~g 4 D 5 / m l . Consequently, our HER2 ECD blocking studies were performed with a background level of 1/~g 4DS/ml. The effect of adding various levels of rHER2 ECD to cells exposed to 4D5 is shown in Fig. 4. The resultant curve exhibits linearity over a narrow range of HER2 ECD concentrations (37.5-300 ng/ml). In order to quantitate HER2 ECD in unknown samples, it was necessary to assay several dilutions of each sample and use only those dilutions which resulted in blocking activities in the linear portion of the curve (from 75% to 95% controls, Fig. 4). The correlation of bioassay values for HER2 ECD level with those obtained by ELISA was highly significant ( r = 0 . 9 2 , P < 0 . 0 0 1 ) . Fig. 5
Serum standard curves The feasibility of this ELISA to quantitate HER2 ECD in mouse and monkey sera was tested by spiking rHER2 ECD into these sera and generating the standard curves (Fig. 1). The curve in mouse serum was indistinguishable from that in buffer resulting in a range of 0.25-120 n g / m l . The curve in monkey serum was shifted suggesting 110
-
100
-
90-
L)
80-
70-
n 60-
50-
'
0
I
2
'
I
]
3
4
'
I
I
5
6
mAb (~g/mL) Fig. 3. Effect of mAbs on growth of SK-BR-3 human breast cancerous cells. • *, mAb 6E9 (non-neutralizing); o o , mAb 4D5 (neutralizing). Results are expressed as % control (no mAb).
78 110
~"
100
100 -
E 100
90
o
80
8o
8o -
J LIJ
60
60
C) (5 IJJ rcq r
40
40
20
20 •
.
73
JIM05659
ELISA for quantitation of the extracellular domain of p185 HER2 in biological fluids Patricia E. Sias 1, Claire E. Kotts 2, David Vetterlein 3, Mike Shepard 4 and Wai Lee T. W o n g 1 J Department of Immunology Research and Assay Technologies, 2 Department of Medicinal and Analytical Chemistry, 3 Department of Process Recovery, and ~ Department of Developmental Biology, Genentech Inc., So. San Francisco, CA 94080, U.S.A.
(Received 7 February 1990, revised received 30 April 1990, accepted l May 1990)
The HER2/neu proto-oncogene encodes a receptor that belong to the tyrosine-specific protein kinase family. Amplification of the HER2 gene in patients with breast and ovarian cancer has been shown to predict poorer survival rates. In order to understand the role of HER2 in malignant and normal cells, it is necessary to devise assays that can quantitate expression levels of the HER2 gene product (p185 HER2) in production samples, biopsy specimens and biological fluids. We have developed a simple, quantitative ELISA that uses two monoclonal antibodies directed against the extracellular domain of the HER2 gene product, p185 HER2 (HER2 ECD). The assay has a detection range of 0.25-120 ng/ml, is precise and sensitive. The ability of this assay to detect biologically active rHER2 ECD is demonstrated by its correlation to a growth inhibitory bioassay (r = 0.92). The sandwich ELISA can also accurately quantitate rHER2 ECD in mouse and monkey serum. This assay should be useful for quantitating low levels of circulating rHER2 ECD in animals in which rHER2 ECD is being used as antigen for immunotherapy and in patients which 'shed' receptor. Key words: HER2/neu; c-erbB-2; ELISA; Bioassay
Introduction
HER2 (c-erbB-2, neu) is a proto-oncogene which encodes a glycoprotein with extensive homology to the human epidermal growth factor receptor (Coussens et al., 1985; King et al., 1985; Schechter et al., 1985; Semba et al., 1985; Yamarnoto et al., 1986). It is encoded by a 4.8 kb
Correspondence to: W.L.T. Wong, Immunology Research and Assay Technologies, Genentech Inc., 460 Point San Bruno Boulevard, South San Francisco, CA 94080, U.S.A. Abbreviations: ELISA, enzyme-linked immunosorbent assay; mAb, monoelonal antibody; ECD, extracellular domain; HRPO, horseradish peroxidase, DEAE, diethylaminoethyl; BSA, bovine serum albumin; SDS, sodium dodecyl sulfate.
messenger RNA that is expressed in a variety of normal and malignant tissues (Kearney et al., 1979; King et al., 1985; Semba et al., 1985; Fukushige et al., 1986; Yamamato et al., 1986; Van de Vijver et al., 1987). The gene is found on q21 of chromosome 17 in humans (Coussens et al., 1985). The gene product, p185 HER2, is a transmembrane protein and possesses intrinsic tyrosine kinase activity (Yarden et al., 1988). It is believed to be a receptor although a iigand for the receptor has not as yet been identified. HER2 amplification has been correlated with a negative prognosis and high probability of relapse in patients with primary breast adenocarcinoma ( S i m o n et al., 1987, 1989). Overexpression of the related rat protooncogene, neu, in transgenic mice results in tumor development in 100% of males and females
0022-1759/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)
74 (Muller et al., 1988). In addition, a monoclonal antibody against the extracellular domain of p185 tiER2 has been shown to inhibit the in vitro proliferation of human breast tumor cells which overexpress p185 HER2 (Hudziak et al., 1989). The present study describes the development of a sensitive and specific enzyme-linked immunosorbent assay (ELISA) for quantitating the extracellular domain of the gene product of HER2 (HER2 ECD) in production samples and biological fluids. A sensitive ELISA will permit the dilution of serum to eliminate potential interfering substances and thus can be used to detect HER2 ECD in serum samples from animals participating in pharmacology studies.
Material and methods
Anti-HER2 ECD antibodies and enzyme-conjugated antibodies Monoclonal antibodies (mAb) directed against the extracellular domain of p185 aER2 were prepared by immunizing mice with N I H 3 T 3 / H E R 2 3400 cells expressing high levels of p185 HER2 (Hudziak et al., 1988; Fendly et al., 1989). The mAbs were purified from ascites fluid using protein A-Sepharose (Repligen Corp., Cambridge MA.) following established procedures (Ey et al., 1978; Goding, 1978) and stored sterile in 0.01 M sodium phosphate, 0.15 M sodium chloride, pH 7.2 (PBS) at 4°C. Purified mAbs were conjugated to horseradish peroxidase (HRPO) (Nakane and Kawaoi, 1974) and stored at - 2 0 ° C in PBS containing 5 m g / m l bovine serum albumin (BSA). ELISA assay standards rHER2 ECD (residues 1-624) has been cloned and stably expressed in transfected C H O cells (Hudziak et al., in preparation). Harvest fluid containing rHER2 ECD was first concentrated by ultrafiltration and then purified by immunoaffinity chromatography. Trace contaminants were removed by DEAE ion exchange chromatography. The purified rHER2 ECD amino acid composition matched what was theoretically expected for the cloned gene product and was homogeneous based on analysis of SDS gel. The concentration of rHER2 ECD in a purified preparation was
established by quantitative amino acid analysis. Dilutions of this sample in PBS containing per liter, 5 g of BSA, 0.5 ml of Tween 20 and 0.1 g of thimerosal (assay buffer) were then used as standards in the buffer based ELISA. For serum based ELISA, the. dilutions were prepared in the corresponding serum.
Samples Samples containing various concentrations of rHER2 ECD were obtained at different stages of purification including harvested tissue culture supernatants. Samples with high concentrations of rHER2 ECD were diluted further with assay buffer and re-quantified. ELISA assay procedure 96-well microtiter plates (Maxisorb, Nunc, Kamstrup, Denmark) were coated with mAb 7F3 by incubating overnight at 4°C with 100/~l/well of antibody at 5 ~tg/ml in 50 mmol/1 of sodium carbonate buffer, pH 9.6 (coat buffer). After removal of the coating solution, the coated plates were blocked with 150 ~tl/well of 5 g/1 of BSA in PBS for 1 h at room temperature, and washed six times with 0.5 g/1 of Tween 20 in PBS (wash buffer). Standards and samples were dispensed onto the coated wells (100/d/well). Plates were sealed and incubated at room temperature for 2 h with gentle agitation. Plates were washed six times with wash buffer. Horseradish peroxidase (HRPO)-labelled mAb 4D5 was then added (100 /~l/well). After further incubation for 1 h at room temperature, the plates were washed six times with wash buffer. Freshly prepared substrate solution (0.4 g of ophenylenediamine dihydrochloride in 1 liter of PBS plus 0.4 ml of 30% hydrogen peroxide) was added to the plates (100/~l/well) and incubation carried out in the dark for 15 min at room temperature. The reaction was stopped by the addition of 100/~1 of 2.25 mol/1 sulfuric acid and absorbance at 490 nm determined on a Vmax plate reader (Molecular Devices, Menlo Park, CA). A standard curve was generated by plotting absorbance vs. log of rHER2 ECD concentration, using a 4-parameter nonlinear regression curve fitting program (developed at Genentech). Sample concentrations
75
were obtained by interpolation sorbance on the standard curve.
of
their
ab-
cients of variation were 5.8% and 13.3%, respectively.
Bioassay method The bioassay used here is a modification of that described for assessing growth inhibitory effects of m A b s to H E R 2 E C D (Hudziak et al., 1989). In this case, we used the bioassay to measure the ability of H E R 2 E C D to block the inhibitory effects of one m A b , 4D5-. H u m a n breast carcinoma cells, SK-BR-3 (ATCC, Rockville, M D ) were maintained in a 1 : 1 mixture of Dulbecco's modified Eagle's medium (high glucose) and H a m ' s F12 medium containing 2 m M glutamine, 100 U penicillin/ml, 100 #g s t r e p t o m y c i n / m l and 10% fetal bovine serum (FBS, Gibco, G r a n d Island, NY). All incubations were carried out at 37°C, 5% CO 2 and 100% humidity. Cells (0.1 ml) were plated onto 96-well microtiter plates (Falcon) at a density of 5 x 10 4 c e l l s / c m 2 in medium containing 2% FBS (assay medium). After 2 h of attachment, 0.1 ml of appropriate test medium was applied to each well and incubation carried out for 4 days. The assay was terminated by gently rinsing cells twice with 0.2 ml PBS and staining with 0.05 ml 0.5% crystal violet ( v / v , in methanol) for 15 rain at ambient temperature. Stained cell monolayers were then rinsed with water and dried. Methanol (0.1 ml) was added to each well to dissolve the dye and absorbance quantitated at 540 nm using an automated plate reader (Bio-Tek Instruments, Model EL309 Burlington, VT). Test media containing various concentrations of H E R 2 E C D were prepared by serially diluting stock H E R 2 E C D in assay medium containing 2 ttg of m A b 4 D 5 / m l . Since 0.1 ml of test medium was added to 0.1 ml of cells, the final levels of medium constituents were one-half that of test medium. Controls received H E R 2 E C D in assay medium without added 4D5. Results are expressed as percent control, r H E R 2 E C D concentrations in unknown samples were quantitated by interpolation of responses on a standard curve of a reference r H E R 2 ECD. All samples were assayed at several dilutions in triplicate, only those dilutions whose responses fell in the linear range of the standard curve were used for quantitation. The intra- and interassay coeffi-
Results
Selection of coat and conjugated antibody Ten mAbs to H E R 2 E C D have been characterized and described elsewhere (Hudziak et al., 1989; Fendly et al., 1990). Based on their reactivities with immobilized p185 HER2, immunoprecipitation patterns, and ability to bind to SK-BR-3 cells, four of them (4D5, 7F3, 2 H l l , 2C4) with varying degrees of reactivities and immunoprecipitation patterns but high SK-BR-3 binding ability were conjugated to H R P O . All ten mAbs were then tested as coat antibody in a sandwich ELISA using these four conjugates. The pair that gave the most sensitive assay for r H E R 2 E C D were 7F3 as coat and 4D5 as H R P O conjugated antibody (data not shown). Consequently these were chosen for ELISA assay development.
Assay range and sensitioity The assay range for the m A b ELISA was determined to be 0.25 to 120 n g / m l (Fig. 1). The lower limit of detection (sensitivity) as defined by Rodbard (Rodbard et al., 1978) is equivalent to the concentration corresponding to the mean absorbance of the zero plus twice the standard devia-
2.0
E t-o 03
o E
1.0
o ..Q < 0.0
...... .1
I 1
........
I 10
........
I
........
100
t
1000
rHEN2 EOD (ng/mL) Fig. 1. Comparison of ELISA standard curves for r H E R 2 E C D in buffer ( o o), mouse serum (n D) and monkey serum (O e).
76 1oo
TABLE I ._I
CROSS-REACTIVITY Protein tested
rhGH-binding protein rCD4 rgpl20
~E Concentration
Cross
Tested
Measured
reactivity "
(#g/ml)
(ng/mi)
(%)
10.0 10.0 5.0
< 0.1 < 0.1 < 0.1
< 0.01 < 0.01 < 0.01
a Percent cross-reactivity was calculated as concentration meas u r e d / a m o u n t tested × 100.
80
,
..o czl 0 w r~r w
40
20
0
,
0
tion and was determined to be 0.14 n g / m l in buffer.
Assay specificity The specificity of this assay was tested with two other soluble receptors (rCD4 and rhGH-binding protein) and an unrelated protein produced in C H O cells (HIV envelope protein, gpl20). All three proteins were obtained from Genentech. The result (Table I) showed that the assay has less than 0.01% cross-reactivity with these proteins.
Samples with low, medium, or high rHER2 ECD protein concentrations in buffer were analyzed in 20 replicates for the assessment of intra-assay precision (Table IIA). Interassay precision was determined by measuring samples of low, medium or high rHER2 ECD concentrations in ten separate experiments (Table liB). The coefficients of intra-assay variation at all three levels
TABLE II PRECISION O F T H E mAb-BASED ELISA Sample 1
Sample 2
Sample 3
20 16.38 0.54 3.3
20 66.06 4.89 7.4
10 21.47 0.75 3.5
10 79.85 5.43 6.8
(,4) Intra-assay precision 20 3.32 0.08 2.4
(B) Interassay precision Replicates Mean ( n g / m l ) SD ( n g / m l )
c v (%)
i
20
,
,
i
40
,
,
i
60
.
,
i
80
,
,
i
1 O0
Theoretical calculation (ng/mL) Fig. 2. Plot of theoretical rHER2 ECD concentration based on dilution of known concentration vs. rHER2 ECD concentration measured by buffer based ELISA for tissue culture supernatant (O) ( r = 0.999; y = 0.992x +0.564), and purified sample (©) (r =1.00; y = 0.988x +0.956).
ranged from 2.4-7.4%, while the coefficients of interassay variation ranged from 3.5%-6.8%.
Linearity of the assay
Assay precision
Replicates Mean ( n g / m l ) SD ( n g / m l ) CV (%)
,
10 4.22 0.17
4.0
The linearity of the assay was determined by making serial two-fold dilutions of two samples, one a purified sample in assay buffer and the other a tissue culture supematant, and measuring the concentration of rHER2 ECD. Results were examined by correlating the observed concentration determined in the ELISA with the calculated concentration obtained by multiplying dilution factor with the concentration of the undiluted sample determined in the ELISA. Linear regression analysis for both samples resulted in correlation coefficients of 0.99 or greater (Fig. 2), indicating that the assay is linear between the assay range, i.e., 0.25-120 n g / m l .
Spike recovery Purified rHER2 ECD at a concentration of 37.5 n g / m l in assay buffer was added in equal volume to media samples diluted 1 / 5 or 1 / 1 0 in assay buffer. These samples were assayed in the ELISA. The theoretical concentration was calculated for each mixed sample and was used to calculate percent recovery. The data in Table III show an average recovery of 98.8% with a range of
77 TABLE llI BUFFERED-BASED ELISA: A C C U R A C Y a Concentration ( n g / m l )
Recovery
Purified sample b
Media sample b
Expected ¢
Observed
(%)
37.5
203.0 in 1 / 5 media 37,5 1.7
120.3 37.5 19.6
126.8 36.3 17.0
105.4 96.8 86.7
37.5
119.0 in 1 / 1 0 media 7.1 0.7
78.3 22.3 19.1
75.6 22.7 20.1
96.6 101.8 105.2
Average = 98.8% Range = 86.7-105.4% a Equal volumes of two purified samples (column no. 1) were each added to based samples in varying concentration of media (column no. 2) and assayed in ELISA. b rHER2 ECD concentrations had been determined previously by ELISA. c (spiked + based)/2.
86.7-105.4%, demonstrating the accuracy of the assay.
shows results obtained by measuring 18 different rHER2 ECD samples in both assays.
Correlation of E L I S A and bioassay To determine the ability of the rnAb based ELISA to quantitate active rHER2 ECD, eighteen samples were assayed in the ELISA and results compared with those obtained by bioassay. The bioassay used here measures the ability of added HER2 ECD preparations to compete with native HER2 receptor on SK-BR-3 cells for mAb 4D5 binding. When no HER2 ECD is present in culture medium, 4D5 greatly inhibits cell growth (Fig. 3). Maximal inhibition occurred at a concentration of 1 /~g 4 D 5 / m l . Consequently, our HER2 ECD blocking studies were performed with a background level of 1/~g 4DS/ml. The effect of adding various levels of rHER2 ECD to cells exposed to 4D5 is shown in Fig. 4. The resultant curve exhibits linearity over a narrow range of HER2 ECD concentrations (37.5-300 ng/ml). In order to quantitate HER2 ECD in unknown samples, it was necessary to assay several dilutions of each sample and use only those dilutions which resulted in blocking activities in the linear portion of the curve (from 75% to 95% controls, Fig. 4). The correlation of bioassay values for HER2 ECD level with those obtained by ELISA was highly significant ( r = 0 . 9 2 , P < 0 . 0 0 1 ) . Fig. 5
Serum standard curves The feasibility of this ELISA to quantitate HER2 ECD in mouse and monkey sera was tested by spiking rHER2 ECD into these sera and generating the standard curves (Fig. 1). The curve in mouse serum was indistinguishable from that in buffer resulting in a range of 0.25-120 n g / m l . The curve in monkey serum was shifted suggesting 110
-
100
-
90-
L)
80-
70-
n 60-
50-
'
0
I
2
'
I
]
3
4
'
I
I
5
6
mAb (~g/mL) Fig. 3. Effect of mAbs on growth of SK-BR-3 human breast cancerous cells. • *, mAb 6E9 (non-neutralizing); o o , mAb 4D5 (neutralizing). Results are expressed as % control (no mAb).
78 110
~"
100
100 -
E 100
90
o
80
8o
8o -
J LIJ
60
60
C) (5 IJJ rcq r
40
40
20
20 •
.
