CIinica Chimicn ktu, 205 f 1992) 213-222 0 I992 Elsevier Science Publishers B.V. All rights reserved. ~9-~9~1/9~~~~~.~~

213

CCA 05217

A competitive ELISA for lipoprotein(a) K.H.James

Yea, Trevor A. ~alrnsle~~ Maurice C. OWXI and Peter M. George

(Received 8 March 199i; revision received 19 November 1991: accepted I! Dewmbw 1991) Kq worh: Lipoprotein(a): Competitive ELISA; Biotin-strcptavidin compared; Coronary heart disease; LDL: Plasminogen

system; Commercial

IRMA

Summary

A competitive ELISA for fipoproteinfa) (Lp(a)) is described. The method uses a commercially available polyclonal anti-Lp(a) antibody and an IgG biotinstr~ptavidin-horseradish peroxidase detection system. The method is simple and robust with an assay sensitivity of 0.7 q/well (I .4 ,@I). The antibody &ross-reactivity was 0.14% against LDL and 0.70% against pfasminogen. The coefficients of variation obtained with control sera of 266 and 552 mg/l were: 5.0% and 4.6% (n = 6), respectively for the intraassay; and 10.8% and 9.5% (n = 16) respectively for the interassay. The method showed an excellent correlation with a commercial immunoradiometric assay (IRMA), y (ELBA) = 0.94x (IRMA) - 8, (r = 0.98). A recovery study in which a 200 mg/L standard and four plasma samples were diluted with different proportions of a low plasma sample, gave linear relationships and also confirmed the specificity of the antibody.

With the high prevalence of coronary heart disease (CHD) in Western Society interest in the diagnosis and prevention of this disease has increased [ 1,2]. One recently identified risk factor is a raised concentration of lipoprotein(a) (Lp(a)) in the plasma [3-51. Because of this, increasing attention is being paid to the measurement of this lipoprotein as an aid to the clinical management of patients [6,7]. ~~~~~~~~~~~~~~;,~~~, IO: K.H. James Yeo. Department Christchurch. New Zealand.

of Ciinicat Biochemistry, Christchurch

Hospital,

214

On ultracentrifugation of the plasma Lp(a) Boats between the density range of LDL and HDL ]8] and has a density of 1.057- 1.075 g/ml [9]. The protein moiety consists of one molecule each of ape(a) and apo Bioo, which are linked by a single disulphide bond [4]. Because of their sensitivity and specificity, immunological methods are considered to be the most appropriate means of measuring Lp(a) in the plasma. These methods include radial and electroimmunodiffusion [4,10,11], radioimmunoassay [ 121 and ELISA [9,10,13- 151. The immunodiffusion methods are not generally preferred because they lack sensitivity and consume large amounts of antibody. Although radioimmunoassay has a much improved sensitivity the radiation hazard. the need for expensive ~rnrn~-counters and the short shelf-life of the reagents are major disadvantages. On the other hand ELISA methodologies have equivalent or improved sensitivity [16], and use stable non-radioactive reagents. Most of the published ELISA techniques for Lp(a) use the ‘sandwich’ approach with antibodies directed to two antigenic sites. However, only one monoclonal antibody-based competitive ELISA has been documented and this was only used to study the cross-reactivity of Lp(a) with plasminogen [ 171. We now present a sensitive and robust polyclonal antibody-based competitive ELISA for routine measurement of Lp(a) in plasma. The method has an equivalent sensitivity to the “sandwich’ method, but uses only a tiny fraction of the reagents. This makes our simple and specific assay a very economical diagnostic method for clinical management of patients with CHD. Materials

and Methods

from Pharmacia Heparin-Sepharose 48 and an “‘1 IRMA kit were obtained Fine Chemicals (Uppsala, Sweden). In the “‘1 IRMA kit, the Lp(a) used in the preparation of the standards and controls consisted of an unknown mixture of Lp(a) isoforms, and the antiserum was prepared from unknown Lp(a) phenotypes (personal communication from Pharmacia Fine Chemicals). Human plasminogen and bovine serum albumin (BSA) were obtained from Sigma Chemical Co. (St. Louis, MO, USA). Flexible Falcon 3912 MicroTest plates were purchased from Becton Dickinson and Co, (Oxnard, CA, USA). A programmable ELISA Processor II machine was from Behring (Marburg, Germany).

Sheep anti-Lp(a) was purchased from Immuno AG (Vienna, Austria). The antigen used to prepare this antibody was pooled material with a mixture of F, Sl, S2, S3, S4 isoforms (personal communication from Immuno AG). Swine anti-goat IgG antibody biotin conjugate (Tago Code 6601) was bought from Tago Inc. (Burlingame, CA, USA). Streptavidin horseradish peroxidase (HRP) was from Amersham (Buckinghamshire, UK). Sheep anti-apo Bloo was purchased from Boehringer (Mannheim, Germany).

215

Buffers Coating bujfir; 50 mM Sodium bicarbonate-carbonate buffer, pH 9.6. ELISA wash buffer: 50 mM Sodium phosphate buffered saline, pH 7.4, containing 0.15 M sodium chloride and 0.1% Tween 20. ELISA diluent: ELISA wash buffer containing 0.1% gelatin (bacteriological grade, Difco, IL, USA) and 0.01% thiomersal. Substrate bufjhr; 25 mM Citric acid and 50 mM disodium hydrogen orthophosphate, pH 5.0. Freshly prepared substrate solution consisted of 2.2 mM ophenylene diamine dihydrochloride and 5.3 mM of 30% hydrogen peroxide. Stopping solution: 2.5 M Sulphuric acid.

Purification

ofLp(a)

Previously ultracentrifuged and analysed fractions of d > 1.063 g/ml from 100 EDTA plasma samples were pooled together to obtain a total volume of 120 ml. These plasma fractions had been previously spun in a Kontron 45.6 fixed angled rotor (Kontron AG, Zurich, Switzerland) at 140,000 X g at 4°C for 18 h. The density of the pooled fractions was adjusted to 1.100 g/ml by adding solid KBr and the Lp(a) was isolated by preparative ultracentrifugation in a Beckman 60Ti rotor (Beckman Instruments Inc., Palo Alto, CA, USA) at 150,000 x g at 10°C for 16 h. The lipoprotein was further purified by heparin-Sepharose 4B affinity purification according to the method of Fless et al. [18]. The resulting protein was identified by Ouchterlony plate using anti-Lp(a) and anti-apo B,,, and the purity was verified by agarose gel electrophoresis. The protein concentration was determined by the method of Lowry et al. [18] using BSA as the standard.

Isolation

of LDL

LDL was required for the micro Ouchterlony plate and anti-Lp(a) specificity study. A 10 ml EDTA plasma sample was obtained from a normal patient and a narrow range low density lipoprotein fraction of 1.030 < cl < 1.050 g/ml was collected after ultracentrifugation in the steps as described above. The density of the sample was adjusted to 1.030 or 1.050 g/ml prior to each step. Further purification of the LDL was performed according to the method of Fless et al. [18]. The identity was verified by an Ouchterlony plate using anti-apo B to0 and the protein concentration was determined by the method of Lowry et al. [19].

QC material for ELISA Control sera consisting of the Sl and S3 Lp(a) isoforms, with stated values of 266 and 552 mg/l of Lp(a) were purchased from Immuno AG (Vienna, Austria).

Pur$ed

Lp(a)

standards

A 25-mg/l primary standard was prepared by diluting the purified Lp(a) in an appropriate volume of the ELBA diluent. Standard solutions of range from 0.098 to 12.5 mg/l were made by diluting this primary standard serially with the ELISA diluent. Sample coIlection and preparation EDTA blood specimens were collected from all patients. After centrifugjng at 3,000 rev./min for 5 min at room temperature the pIasmas were collected and stored at -80°C until further processing. On the day of assay, all plasma samples (25 ~1) from patients and quality control sera were diluted 200-fold with the ELBA diluent (5 ml) and mixed by vortexing prior to dispensing into the plate. Microtitre plate coating procedure Each well of a flexible Falcon MicroTest plate was filled with 100 ~1 of a solution containing 0.75 p&ml purified Lp(a) in coating buffer. Following an overnight incubation at 4°C the plate was washed four times with ELISA wash buffer. One hundred and fifty ~1 of ELISA diluent was dispensed into each well to block unbound sites and minimise non-specific protein binding. The plate was incubated at room temperature for 1 h, and prior to assay the diluent was discarded. Coated microtitre plates could be stored in coating buffer at 4°C for up to 2 weeks. ELBA

procedure

Fifty ~1 of each standard and the diluted samples or controls were dispensed manually in duplicate into the coated microtitre plate. All the other reagent additions and washing steps were automatically performed by the ELISA Processor II machine. All incubations were performed at room temperature as described betow. Following the addition of 50 ~1 of an 8,000-fold dilution of sheep anti-Lp(a) in ELISA diluent the plate was incubated for 2 h. After washing ( x4) with ELISA wash buffer, 100 ~1 of 4,000-fold diluted swine anti-goat IgG biotin conjugate in ELISA diluent was dispensed into each well and incubated for 2 h. The plate was then washed ( x 4) and 100 ~1 of a 2,000-fold dilution of streptavidin-HRP in ELBA diluent was dispensed into each well. After a 45-min incubation, the plate was again washed (x 4) and 100 ~1 of substrate solution dispensed into each well. After a 20 min incubation in the dark, the chromogenic reaction was terminated by adding 100 ~1 of stopping solution and the absorbance was read at 492 nm with 650 nm reference. Recovery study A 200 mg/l standard and 4 plasma samples were individually mixed with a diluent plasma, which contained a very low concentration of Lp(a). The proportions of sample to diluent plasma used were: 2: 1, 1:I, I:2 and I :5. Following brief mixing the

217

diluted plasmas were incubated at 37°C for 15 min and analysed for Lp(a) as described above. Results The specificity of the commercially obtained anti-Lp(a) antiserum was excellent. An Ouchterlony plate analysis against purified LDL, purified Lp(a) and

0

313

625

1250

LPW Fig. 1. Standard

curve for the competitive

ELBA

2500

ms/L method

using purified

Lp(a) standards.

218

plasminogen showed only one precipitin arc between anti-Lp(a) and purified Lp(a). No precipitin arc was detected between anti-Lp(a) and either purified LDL or plasminogen. In addition. an Ouchterlony plate test using anti-apo Bt,, antiserum against purified LDL and purified Lp(a) yielded a precipitin arc between the antibody and both lipoproteins. This confirms that our purified Lp(a) contains apo B ,sO, being linked to apo (a). Agarose gel electrophoresis of a 10 pg sample confirmed a single band with pre-beta mobility [1 11, indicating that the purified Lp(a)

.

:

200

o

I

I:’ I

0

200

400

1000

.a” 800 --

600 --

400 -n 200 -- / o :’ 0

200

400

“Theoretical”

600

600

11 30

Lp(a) value, mg/L

Fig. 2. Effect of dilution of known plasma samples with a low plasma as diluent. Each sample was diluted 2:1, I: 1, 1:2, and 1:5 for each sample. The ‘theoretical’ values were calculated from values obtained for undiluted plasma and the dilution. The ‘experimental’ value was obtained by assaying each dilution with the competitive ELBA. The linear regression equations are as follows: A, 160 mg/l J‘ = 0.91s + 7, r = 0.99; 4, 200 mg/l (standard) y = 0.94.~ + 15. r = 0.99; 0. 615 mgil .v = 0.87.~ + 38, I = 0.99; V. 292 mg/l y = 1.04.~ + 4, I = 0.99; 0. 1.308 mgil y = 0.95.~ + 6, I = 0.99.

219

is free from other protein contamination. Further veri~cation on the specificity of this antibody was demonstrated by comparing the standard curves obtained with purified LDL and plasminogen against the purified Lpfa) standards. The crossreactivity for these species was 0.14% and 0.70%, respectively. The standard curve obtained using purified Lp(a) as a primary standard in our competitive ELBA method is shown in Fig. 1. The range of the standards was 4.9 ng to 625 ng/well, which corresponded to Lp(a) concentration of 19.6-2500 mg/l for a 200-fold dilution of plasma. The sensitivity of the assay, defined as the smallest amount of Lp(a) that is different from zero at the 95% confidence limit, was 0.7 ngiwell which corresponds to 14 ,ug/l or 2.8 mg/l in the undiluted plasma. The imprecision of the assay was determined by repeated analyses of two Immuno AC control sera (266 and 552 mgll). The coefficients of variation were 5.0% and 4.6% (intraassay, n = 6), respectively, and 10.8% and 9.5% (interassay, n = 16), respectively. Samples prepared by diluting known plasma samples with a low concentration plasma sample were also analysed. Four such samples, range 160- 1,308 mg/i, and a purified Lp(a) standard, 200 mg/l, were studied. For all these 5 samples, regression equations comparing the ‘theoretical’ values (calculated from the initial values and the dilution) with the measured values, gave correlation coefficients of >0.99, see Figs. 2a and 2b. Parallelism of the method was demonstrated by analysing dilutions of a medium and a high Lp(a) samples in the ELBA diluent. The results are shown by a log-log plot in Fig. 3. The regression equation for the medium and high samples are y = 0.96.x + 5, r = 0.99 and y = 0.97x + 5, I = 0.99, respectively. The inaccuracy of the method was confirmed by measuring the Lp(a) concentration in plasma samples from 12 patients with Lp(a) ranging from 12 to 1,070 mg/l,

Log (Dilutlon Fig. 3. Effects of dilutions

of a medium

Factor)

(0, y = 0.96x + 5, r = 0.99) and a high ([II, y = 0.97s

r = 0.99) Lp(a) samples

with the ELBA

diluent.

+ 5.

800 600

0

200

400

600

IRMA Lp(a), Fig. 4. Linear correlation

800

ELISA

method. .v (ELISA)

1; 30

mg/L

of 12 patient plasma Lp(a) concentrations

kit and the competitive

1000

measured by a commercial

= 0.94s (IRMA)

IRMA

- 8. r = 0.98.

The results obtained with the ELISA and the Pharmacia IRMA kit are shown in Fig. 4. Linear regression analysis of the results from these samples showed, _V (ELISA) = 0.94x (IRMA) - 8, P = 0.98.

There have been a number of reports of sandwich ELISA methods for estimation of Lp(a) in plasma [9,10,13- 151. However, the only brief and inadequately described competitive ELISA method of Guo et al. was used solely for a cross-reactivity study with plasminogen [19]. In most of the sandwich methods the microtitre plates were coated with either a monoclonal or polyclonal anti-Lp(a), whilst the probes used were peroxidase-labelled anti-Lp(a) 1141or rabbit Fab’ portion of anti-Lp(a) [lo]. Alternatively, Fless et al. used a goat anti-apoB antibody foliowed by a commercial rabbit anti-goat alkaline phosphatase conjugate 1131, whilst, Rainwater and Manis used a biotinylated anti-apoB antibody followed by adding the avidin-biotinylated alkaline phosphatase complex 1151. One could argue that the high cost of the commercially available Lp(a) and the large molecular weight of the lipoprotein, would make the selection of a competitive ELISA inappropriate, Similarly in a competitive assay, binding of the anti-Lp(a) might be biased towards either the solid-phase or free Lp(a) sample present in the well. However, in practice we have found that our polyclonal antibody-based competitive ELISA is robust and as sensitive as the reported sandwich methods. We have developed this competitive ELISA for Lp(a) for routine clinical use. The method is simple, sensitive, robust, and requires only a small volume of plasma for analysis. The standard curve ranges from 4.9 to 625 nglwell and corresponds to

221

plasma levels of 19.6-2,500 mg/l. The sensitivity of our method is 0.7 ngwell which is comparable to that reported in the sandwich ELISAs [ 10,151. Although sandwich ELISAs are sensitive methods for measuring Lp(a) there are some problems. A variety of approach has been reported for the analyses of plasma samples. These range from a simple analysis of each sample in singles [ 141, through to a complex procedure utilising a combined Ouchterlony plate test and sandwich ELBA of the sample at three different dilutions aliquoted in triplicates onto the microtitre plate 1131. In contrast we have found that our ELBA procedure allows the analysis of most samples using a single dilution and that analysis in duplicate provides an acceptable coefficient of variation. Purification of Lp(a) by the method of Fless et al. [IS] is quite simple, and as only 75 ng of the purified Lp(a) antigen is coated onto the surface of each well, a single preparation is sufficient for coating over 100 microtitre plates. If the purified Lp(a) is stored at 4°C in 6 mM barbital-O.15 M NaCl buffer (pH 7.4), containing 0.04% disodium EDTA and 0.02% sodium azide as preservatives, it is stable for up to 12 months. This purified Lp(a) can also be used as a standard for the assay. Alternatively the assay could be calibrated with a commercial Lp(a) standard. We have also found that the isolated Lp(a) is sufficiently stable at 4°C to allow storage in the coating buffer for up to 2 weeks without showing any loss of antibody binding activity. Judging from published reports it is also possible to use RID and Laurel1 rockets to estimate Lp(a) routinely [4, IO,1I]. However, this is expensive in using large quantities of primary antibody and gives an inferior assay sensitivity [ 161. Similarly in the reported sandwich ELISAs, which are generally very specific, the amount of antiLp(a) or anti-ape(a) coated to the well has varied from 200 ng to 1,300 ng [ 13,151. In contrast, our competitive ELISA uses only 5 ng/well of sheep anti-Lp(a). Although RIA and IRMA methods allow satisfactory alternative assays, the combined problems of tracer instability and radiation hazards make them inappropriate choices for many laboratories. Despite the remarkable homology of the kringle-4 domains of plasminogen and of Lp(a) [20,21], we found that the anti-Lp(a) antibody we used has excellent specificity. It exhibits a cross-reactivity of 0.14% and 0.70% against LDL and plasminogen respectively. These cross-reactivities are comparable to those obtained with the sandwich methods. The specificity of this assay was further confirmed by the parallelism study, shown in Fig. 3, where linear relationships are obtained at the extremes of sample dilution (i.e. from IOO-to 2,000-fold dilution) of a medium and a high plasma samples. If LDL, plasminogen, or both interfered in this assay the results obtained at a lower dilution would be expected to depart from linearity. In this assay we have used a sheep antibody to Lp(a) whilst the second antibody is a swine anti-goat IgG. This combination was selected because the porcine antigoat antibody cross-reacts strongly with the sheep antibody. We also used a biotinstreptavidin system, in which biotin is attached to the second antibody and streptavidin is conjugated to horseradish peroxidase. Thus our system requires several incubation steps. This probably increases the sensitivity of the assay. However, we would predict that the replacement with a simple second antibody-enzyme-labelled conjugate would afso work.

222

Although our method requires the dilution and manual dispensing of standards and samples into a microtitre plate, this only constitutes approximately 10% of the time required for processing the assay. The rest of the processing time is required for washing, reagent dispensing, incubation and reading of the plate. In our laboratory these steps have all been automated by a Behring ELISA Processor II. Thus. an assay requires only about 1 h of technician time to analyse 24 samples in duplicate. This factor combined with the small volumes of reagents used make this a very economical assay for routine use in clinical laboratories. References I

Houlston R, Quiney J, Mount familial hypercholesterolaemia.

2

Ghilan J-M, Parfonry A. Kozyreff V. Heller FR. Lipoprotein(a), cholesterol and coronary heart disease. Lancet ii 1988;963. Berg K, Dahlen G, Frick MH. Lp(a) lipoprotein and pre-beta-lipoproteins in patient with coronary heart disease. Clin Genet 1974;6:230-235.

3 4 5 6

1 8 9 10

11 12 13 14 15 16 17

18 19 20 21

J, Watts CF. Lewis B. Lipoprotein(a) Lancet ii 1988;405.

and coronary

heart disease in

Gaubatz JW, Heideman C, Gotto AM, Morrisett JD, Dahlen GH. Human plasma lipoprotein(a). J Biol Chem 1983;258:4582-4589. Rhoads GG, Dahlen G, Berg K, Morton NE, Dannenberg AL. Lp(a) lipoprotein as a risk factor for myocardial infarction. J Am Med Assoc 1986;256:2540-2544. Dahlen GH, Guyton JR, Arrar M, Farmer JA, Gotto Lp(a), plasma lipids and other lipoproteins with coronary Circulation 1987;84:758-765. Hoefler G. Harnoncourt F, Paschke E et al. Lipoprotein

AM. Association of levels of lipoprotein heart disease documented by angiography. Lp(a). A risk factor for myocardial

infarc-

tion. Arteriosclerosis 1988;8:398-401. Utermann G. Weber W. Protein composition of Lp(a) lipoprotein from human plasma. FEBS Lett 1983;154:357-361. Labeur C, Michiels G, Bury J. Usher DC, Rosseneu M. Lipoprotein(a) quantified by an enzymelinked immunosorbent assay with monoclonal antibodies. Clin Chem 1989~35: 1380- 1384. Abe A, Maeda S. Makino K et al. Enzyme-linked immunosorbent assay of lipoprotein(a) in serum and cord blood. Clin Chim Acta 1988; 177:31-40. Albers JJ, Hazzard WR. lmmunochemical quantification of human plasma Lp(a) lipoprotein. Lipids 1974;9: 15-26. Albers JJ, Adolphson JL, Hazzard WR. Radioimmunoassay of human J Lipid Res 1977;18:331-338. Fless GM, Snyder ML, Scanu AM. Enzyme-linked immunoassay

plasma for

Lp(a) lipoprotein.

Lp(a).

J Lipid

Res

1989;30:65 I-662. Wong WLT, Eaton DL, Berloul A. Fendly B. Hass PE. A monoclonal-antibody-based enzymelinked immunosorbent assay of lipoprotein(a). Clin Chem 1990;36:192-197. Rainwater DL, Manis GS. Immunochemical characterization and quantitation of lipoprotein(a) in baboons, Atherosclerosis 1988;73:23-31. Labeur C, Shepherd J, Rosseneu M. Immunological assays of lipoproteins in plasma: methods and instrumentation. Clin Chem 1990;36:591-597. Guo H-C, Armstrong VW, Luc G et al. Characterization of five mouse monoclonal antibodies to apolipoprotein(a) from human Lp(a): evidence for weak plasminogen reactivity. J Lipid Res 1989;30:23-37. Fless GM, Rolih CA, Scanu AM. Heterogeneity of human plasma lipoprotein(a). J Biol Chem 1984;259:11470-11478. Lowry OH, Rosebrough NJ, Farr AL, Randall RA. Protein measurement with the folin phenol reagent. J Biol Chem 1951;193:265-275, Eaton DL, Fless GM, Kohr WJ et al. Partial amino acid sequence of apolipoprotein(a) shows that it is homologous to plasminogen. Proc Nat1 Acad Sci U.S.A. 1987:84:3224-3228. McLean JW, Tomlinson JE, Kuang W-J et al. cDNA sequence of human apohpoprotein(a) is homologous to plasminogen. Nature 1987;330: 132- 137.

A competitive ELISA for lipoprotein(a).

A competitive ELISA for lipoprotein(a) (Lp(a)) is described. The method uses a commercially available polyclonal anti-Lp(a) antibody and an IgG biotin...
723KB Sizes 0 Downloads 0 Views