A sensitive micro-immunoassay using β-galactosidase/ anti-β-galactosidase complexes

Journal of Immunological Methods, 97 (1987) 19-27 Elsevier

19

JIM 04221

A sensitive micro-immunoassay using fl-galactosidase/anti-fl-galactosidase complexes H. D u r b i n and W.F. B o d m e r Imperial Cancer Research Fund, Lincoln's Inn Fields, London WC2A 3PX, U.K. (Received 11 June 1986, revised received 10 September 1986, accepted 17 October 1986)

This paper describes a new sensitive microELISA based on enzyme/anti-enzyme complexes following an unlabelled antibody bridging step. fl-Galactosidase/anti-fl-galactosidase complexes were made using a monoclonal antibody raised against bacterial (E. coli) fl-galactosidase and enzyme activity was quantified with a fluorogenic substrate. Because of its high sensitivity the assay is particularly suitable for the detection of limited amounts of antigen. One application illustrated is the analysis of Class I and Class II histocompatibility antigens on peripheral blood lymphocytes using 5000 cells/well in 60-well Terasaki or 96-well microtitre plates. Key words: ELISA, micro-; fl-Galactosidase/anti-fl-galactosidase antibody

Introduction

Enzyme-linked immunoassays are now widely used for the detection of solid-phase bound antigen (Voller et al., 1978; Avrameas et al., 1979; Kelly et al., 1979; Voller et al., 1981). However, for situations where antigen is limited, an assay offering increased sensitivity would be of use in a variety of applications. A few examples include looking for small changes in antigen expression after treatment of cells with agents which inCorrespondence to: H. Durbin, Imperial Cancer Research Fund, P.O. Box 123, Lincoln's Inn Fields, London WC2A 3PX, U.K. Abbreviations: ELISA, enzyme-linked immunosorbent assay; PBS, phosphate-buffered saline; FCS, foetal calf serum; RPMI 1640, Roswell Park Memorial Institute 1640 medium; 4MUG, 4-methylumbelhferyl-fl-D-galactosidase; GAG complex, anti-fl-galactosidase/fl-galactosidase complex; PBL, peripheral blood lymphocyte; EBV, Epstein-Barr virus; cLL, chronic lymphocytic leukaemia; RAM, rabbit anti-mouse immunoglobulin.

complex; Fluorogenic substrate; Increased sensitivity; Monoclonal

fluence differentiation, in gene mapping for the analysis of cell surface antigens in somatic cell hybrids where expression may be low and also for tissue typing of peripheral blood lymphocytes. Because of their fine specificity, monoclonal antibodies are increasingly being used to replace alloantisera but not all monoclonal antibodies fix complement so that standard cytotoxicity techniques as used in tissue typing are not generally suitable. Existing ELISA methods for the detection of binding of HLA monoclonal antibodies to peripheral blood lymphocytes are inadequate because the minimum cell number required for a detectable response is on the borders of practicability (Lansdorp et al., 1982). The unlabelled enzyme/anti-enzyme technique (Lansdorp et al., 1980) for the assay of monoclonal antibodies bound to whole cells offers the advantage of increased sensitivity over the enzyme-linked second antibody method (Douillard et al., 1980; Kennet et al., 1980) but a major drawback which affects sensitivity still exists.

0022-1759/87/$03.50 © 1987 Elsevier Science Publishers B.V. (Biomedical Division)

20 When using peroxidase/anti-peroxidase complexes high backgrounds due to endogenous cell peroxidase activity can be a problem (Strauss, 1979). This is only partially overcome by various blocking agents such as phenylhydrazine or methanol/hydrogen peroxide (Lansdorp et al., 1982) as positive as well as non-specific values may be substantially reduced by such treatment. We now describe a new sensitive microELISA which is based on E. colt fl-galactosidase. Unlike systems using horseradish peroxidase or alkaline phosphatase, it is possible to avoid endogenous target cell enzyme activity by selecting a p H at which bacterial enzyme activity is optimal and at which mammalian (including mouse) fl-galactosidase is inactive (H/Ssli et al., 1978). In this assay, which is based on the system first described by Sternberger (1979) for immunohistological tissue staining using horseradish peroxidase, mouse monoclonal anti-E, colt fl-galactosidase/fl-galactosidase complexes are used following an antimouse bridging step (Fig. 1). High sensitivity has been achieved by the use of a fluorogenic substrate for fl-galactosidase (Ishikawa et al., 1978). This has allowed miniaturisation of the ELISA and the use of as few as 5000 cells and test volumes of 5 or 10 /~1 in 60-well Terasaki microtitre plates. Fluorescence is detected and recorded using a Leitz MPV compact MT inverted microscope linked to an Apple II computer. The assay may also be carried out in standard 96-well microtitre plates, fluorescence being recorded using a Dynatek Microfluor apparatus. Alternatively, qualitative fluorescence may be simple recorded by photography over ultraviolet light.

2

3

Cell surface antigen

Fig. 1. Diagrammatic representation of enzyme/anti-enzyme system. 1: mouse antibody; 2: anti-mouse bridging antibody; 3: fl-galactosidase/anti-fl-galactosidase complex.

Materials and methods

(1) Monoclonal antibodies to E. colt fl-galactosidase (a) Production. B A L B / c mice were immunised with fl-galactosidase EC 3.2.1.23 from E. co6 (Sigma). Fusion of immune spleen cells with P3/NS1/1-Ag4-1 mouse myeloma cells was carfled out according to Galfr~ et al. (1977). Hybrids were selected in RPMI 1640 containing 20% FCS, 10 - 4 M hypoxanthine, 1.6 × 10 -s M thymidine, 10 -s M methotrexate, 100 U / m l penicillin and 100 U / m l streptomycin in the presence of mouse peritoneal macrophages. Hybrid supernatants were screened against fl-galactosidase bound to 96-well microtitre plates (Nunc) (100 /~1 of a 1 m g / m l solution for 1 h at room temperature) using an indirect horseradish peroxidase linked second antibody ELISA (Kennett, 1980). Selected hybrids were cloned by single cell picking. (b) Characterisation and purification. Class and subclass isotypes were determined by double diffusion (Ouchterlony, 1978) in 1% agarose using specific anti-mouse reagents (Miles). Ascites fluid was produced by i.p. injection of 3 × 10 6 hybrid cells into pristane (Aldrich) primed B A L B / c mice. Immunoglobulins were purified from this, IgG by affinity chromatography using Protein A-Sepharose 4B (Ey et al., 1978) and IgM by dialysis against distilled water. Purity was monitored by isoelectric focusing on agarose. Protein concentrations were determined by the method of Lowry et al. (1951) or by optical density measurements at 280 nm. Purified immunoglobulin was stored at 4 ° C with 0.02% NaN 3. (2) Assay to monitor enzyme activity of fl-galactosidase when bound to anti-fl-galactosidase monoclonal antibodies 50 btl protein A (Pharmacia) 0.1 m g / m l in PBS was allowed to dry overnight in 96-well polystyrene plates (Nunc). 50 /~1 rabbit anti-mouse serum (DAKO) diluted 1 : 100 in PBS was added and incubated for 45 min at room temperature. After washing four times with PBS/Tween 20, 50 /zl supernatant containing anti-fl-galactosidase monoclonal antibody were added and incubated for 1 h at room temperature. After washing as above, 50/xl of 100/~g/ml fl-galactosidase in PBS were added and incubated for a further 1 h at

21

room temperature. Hates were washed as above and substrate reaction determined either colorimetrically by adding 100 /~1 1 m g / m l onitrophenyl-fl-o-galactopyranoside (Sigma) in PBS containing 1.5 m M MgCl 2, 100 mM fl-mercaptoethanol and 0.2% Tween 20; or fluorometrically with 100 /xl of a saturated solution of 4 M U G (Sigma) in the same buffer. Incubation in each case was at 37°C for 30 min. Optical densities were recorded on a Titertek Multiscan at 405 nm. Fluorescence in wells was recorded by photographing the plate over an ultraviolet light box with Polaroid Land film 57 using an orange filter.

TABLE I CHARACTERISTICS OF THE MONOCLONAL BODIES USED IN THIS WORK Name

HLA specificity

CA2

Monomorphic D-region DQwl DQwl DRw52 DQw3 DR7+ + Monomorphic ABC ABC-B7 ABC-A2, Aw69

SDR1.2 SDR4.1 SDR8.1 2HB6 17.3.3 W6 32 BB7.1 BB7.2

ANTI-

Subclass

Reference

IgG2a IgG1 IgG2b IgG2b IgG1 IgG2a

Eckels et al., 1981 De Kretser et al., 1982 Bodmer et al., 1985 Bodmer et al., 1985 Shannon et al., 1985 Ozato et al., 1980

IgG2a IgG1 IgG2a

Barnstaple et al., 1978 Brodsky et al., 1979 Brodsky et al., 1979

(3) fl-galactosidase / anti-fl-galactosidase complexes 0.5 mg equivalent to 500 U of fl-galactosidase were mixed with 150/~g purified monoclonal antibody in 1 ml 0.05 M Tris buffer pH 7.4 and incubated at 4°C for 24 h. No further purification was necessary. Complexes were stored at 4 ° C with N a N 3 0.02% and were stable for at least 6 months.

(4) Cells WT46 (DRwl3), WT51 (DR4), MST (DR2) and Mann (DR7) (Brodsky et al., 1979) are B lymphoblastoid cell lines obtained by EBV transformation of peripheral blood lymphocytes. Normal PBL and cLL lymphocytes were separated from whole blood on Ficoll-Triosil (B/Syum et al., 1968). Cell lines were maintained in RPMI 1640 containing penicillin 100 U / m l and streptomycin 100 U / m l and 10% FCS. PBL were either used immediately after preparation or stored in liquid N 2 until required as were cLL cells.

(5) Monoclonal antibodies The monoclonal antibodies used are documented in Table I. Antibody dilutions were made from culture supernatant or purified immunoglobulin 1 mg/ml.

(6) GAG assay (a) Preparation of solid-phase test plates.

60well Terasaki plates (Falcon) were precoated with 0.1 m g / m l poly-L-lysine (Sigma) in PBS for 1 h. Cells, lymphoblastoid fines or PBL, from culture medium or frozen stocks were washed twice in

PBS and added to wells in densities ranging from 5 × 103 to 105 in 10 #1 suspension volumes in PBS. Plates were centrifuged (Sorvall GLC-2B) for 5 min at 1200 rpm and then gently flooded with 0.025% glutaraldehyde in PBS. After 15 min the plates were shaken empty, washed once with PBS and flooded with 0.1% gelatin in PBS. Plates were covered and stored at 4 ° C until required for up to 3 months. Solid-phase microtitre plates (Dynatek M129A) were prepared in the same way except that cells were added to wells in 50 btl volumes. (b) Assay procedure. After shaking empty, plates were washed twice by gently flooding and shaking out with PBS containing 0.2% casein (Oxoid L41) (Kenna et al., 1985). In Terasaki plates 5 or 10 /~1 antibody were added and incubated for 1 h at room temperature. Wells were aspirated and washed as above with PBS/casein containing 0.2% Tween 20. They were then incubated further with 10 /~1 rabbit anti-mouse immunoglobulin (DAKO Z259) diluted 1 : 100 in 200 m M Tris buffer p H 7.4 containing 10% normal human serum. After washing with P B S / c a s e i n / Tween, 10 /al G A G complex diluted 1:1000 in 200 mM Tris buffer pH 7.4 containing 10% normal human serum were added and incubated for 1 h at room temperature. After washing again with P B S / c a s e i n / T w e e n 10 #1 of a saturated solution of M U G were added and the plate incubated for 15 min. The reaction was stopped by adding 2 /~l/well cold glycine/NaOH, 0.5 M p H 10. Fluorescence was measured and recorded using a Leitz

22 inverted fluorescence microscope MPV compact M T with incident light excitation provided by a 75 W high pressure xenon lamp. An Apple II computer controlled the automatic plate screening process and quantitative print out. The assay procedure in 96-well plates was identical except that 25 /xl volumes were used at all stages apart from the additions of substrate and glycine where 100/~1 and 50 ffl were used respectively. 96-well plates were read on a Dynatek Microfluor reader. Fluorescence in some plates was recorded by photography as already described.

Results

were not as good as DC1 4C7 complexes used alone. Mixed complexes, namely two or three monoclonal antibodies bound to fl-galactosidase were also found to be inferior to the best single antibody complex used alone. Therefore DC1 4C7 complexes were selected and further experiments carried out to determine the optimal dilution for use in the assay.

Determination of optimal assay conditions Checkerboard experiments using a range of concentrations of RAM, 1 : 10, 1 : 20, 1 : 50, 1 : 100 and 1:200 and G A G , 1:100, 1:200, 1:500, 1 : 1000, 1 : 5000 were carried out using 104 MST cells/well with two polymorphic monoclonal antibodies SDR1.2 (positive) and 2HB6 (negative).

Monoclonal antibodies to fl-galactosidase

h4ST 104 PBS / Tween

Ten hybrids which showed reactivity to flgalactosidase were expanded in culture. It was necessary to establish that the activity of the enzyme was maintained when coupled to the antienzyme. On the basis of high levels of fl-galactosidase activity after enzyme binding to these monoclonal antibody-containing supernatants, four were selected for cloning and further study. Clones DC1 4C7, DC2 12H4, DC3 10D6 all IgG1 and DC4 7A4, an IgM, were grown as ascites in mice from which immunoglobulin was purified.

PBS / Casein

i 2000

'~

\

i

c3

150C

Determination of optimal monoclonal antibody/ enzyme complex A n t i b o d y / e n z y m e complexes were made with DC1 4C7, DC2 12H4, DC3 10D6 and DC4 7A4 as described. G A G assays were carried out on Mann B lymphoblastoid cells, 5 × 103/well, using supernatant of the monomorphic D region antibody CA2-11. Complexes were used in dilutions of 1:100, 1:200, 1:500, 1:1000, 1:2000 and 1 : 5000. DC1 4C7 gave the best response, namely greater than 10 times the non-specific antibody value at 1:5000. Similar values were given by DC2 12H4 at 1:1000. DC3 10D6 was less effective even at dilutions of 1:100 and DC4 7A4 complexes gave no response at 1:100. G A G assays were also carried out using mixtures of two or three complexes. When DC1 4C7 complexes were used mixed with DC2 12H4 or DC3 10D6 complexes, or using mixtures of all three, responses

1000s

8

2~

5oo! I/

"

I

L 1

2

~

5 6 7

1

2 3 4 5 6 7

Log 10 x ddutlon purified entibody

Fig. 2. Effect of PBS/casein washing steps to reduce nonspecific binding. MST (104 cells/well) were assayed with 4 anti-Class II monoclonal antibodies; SDR1.2, 0 0 (positive), and 2HB6, • •; SDR8,1, zx zx; 17.3.3, ~, • (negative). Each point represents the mean of triplicate values in one assay.

23

3000

PBL HLiAl'A2'w632B51,B8 BB71

BBZ2

105 0 5x104 zl 104 • 5x10 3 •

~

I

I

I

L

I

o :3

-.T-

I

I

PBL HLA A3, A23, B7, B41

©

BB71

3000

BB7.2

2000

10OO

I

I

I

I

I

Log 10xdilution

I

I

I

I

I

of purified antibody

Fig. 3. Cell numbers/well. Binding of anti-Class I monoclonal antibodies to PBL. a: HLA A1, A2, B51, B8 assayed with antibodies W6 32 and BB7.2 (positive) and BB7.1 (negative); b: HLA A3, A23, B71, B41 assayed with antibodies W6 32 and BB7.1 (positive) and BB7.2 (negative). Each point represents the mean of duplicate values in one assay.

24 Optimal concentrations of RAM and GAG, that is those which together showed the best positive reactivity with low negative values were 1 : 1 0 0 and 1:1000 for R A M and G A G respectively. When first antibody incubation times were increased up to 24 h, binding was not increased although negative antibody values were increased slightly. Lengthening substrate incubation times also increased negative as well as positive values.

Reduction of non-specific binding When cell numbers below 105/well were used, unacceptably high negative antibody values were obtained. This proved to be largely due to nonspecific binding of monoclonal antibody to the polystyrene support. It was possible to reduce this to some extent by increasing the concentration of the gelatin blocking reagent to 0.1%. However, if PBS containing 0.1% casein was used to wash the

Fig. 4. An example of one Terasaki plate, part of an early screen of supernatant fluids for specific monoclonal antibody secretion. GAG assay used 5 × 103 lymphoblastoidcells/well and duplicates of 5/zl of culture supernatant.

25

gelatin blocked plates before the addition of the monoclonal antibody, and PBS/casein/Tween was used for intermediate washing steps in the assay, non-specific binding was almost completely eliminated (Fig. 2). Using PBS/casein instead of gelatin as a primary blocking reagent was ineffective and resulted in high non-specific binding.

Cell numbers/well in GAG assay In order to determine the minimum cell number required, GAG assays were carried out over a range of cell densities for B lymphoblastoid cell lines, peripheral blood lymphocytes and cLL cells using HLA-ABC and HLA-D region monoclonal antibodies. Serial ten-fold dilutions of purified

Fig. 5. An example of the G A G assay using 96-well microtitre plates; part of an experiment to characterise a new set of monoclonal antibodies. Duplicate 25/~1 volumes of supernatant were assayed against 104 cells/well.

26 monoclonal antibody immunoglobulin or doubling dilutions of culture supernatants were used. G A G complexes were used diluted at 1 : 1000 and RAM diluted 1:100. Using lymphoblastoid cell lines, both class I and class II monomorphic and polymorphic antibodies gave very good responses with 5 × 103 cells/well. 103 cells/well also gave significant detectable reactivity. When PBL were analysed with class I antibodies 5 x 103 cells/well gave clear cut positive reactions (Fig. 3). When the predominantly B cell population of PBL from cLL patients were analysed with class II monoclonal antibodies, 104 cells/well gave good responses. It was also possible to analyse for class II antigens on whole PBL. Between 5 x 104 and 105 cells/well, of which the reactive B cell population would be approximately 10%, were required to give good polymorphic reactivity.

Monoclonal antibody screening The G A G assay has also been used for monoclonal antibody screening. Fig. 4 shows part of an early screen against a range of antigenic determinants on 5 x 103 whole lymphoblastoid cells in a Terasaki plate in an experiment to raise polymorphic HLA class II monoclonal antibodies. Fig. 5 is an example of part of an extensive multiple screen in 96-well microtitre plates to characterise a set of placental alkaline phosphatase monoclonal antibodies (to be reported elsewhere).

Discussion

We have shown that mouse monoclonal antifl-galactosidase enzyme complexes may be used as a reagent in a simple sensitive microELISA. Because of the high sensitivity achieved it has been possible to miniaturise the assay for use in 60-well Terasaki plates although standard 96-well microtitre plates may also be used. Non-specific binding mainly of monoclonal antibody to the plastic microtitre plates became increasingly apparent as cell numbers were reduced. In experiments to eliminate this, casein (possibly because of its different surface charge (Taborsky, 1974)) was found to be the most effective of the commonly used blocking agents compared. When used in both the prewashing and subsequent wash-

ing steps of the assay non-specific binding was almost completely eliminated. It has been suggested by Lansdorp et al. (1982) that the efficiency of binding in a bridging step may be quantitatively affected by the subclass of the monoclonal antibody in the enzyme/anti-enzyme complex. Their findings showed that when analyzing peripheral blood lymphocytes with Class I monoclonal antibodies, a polymorphic IgG1 antibody gave higher binding than a monomorphic IgG2a antibody. The subclass of the PAP complex monoclonal antibody used in this study was IgG1. When using the G A G ELISA, we were able to confirm previous binding patterns using a radioimmune binding assay (Parham and Bodmer, 1978) in which monomorphic antibodies for Class I and also for Class II, both IgG2a, gave higher binding than polymorphic antibodies, even though these were of the same subclass IgGI as the complex monoclonal antibody. Before complexes were made, experiments to learn more about the epitopes recognised by these antibodies were carried out using competition binding assays with pairs of antibodies one of which was labelled with 1251 (results not shown). These were generally difficult to interpret but did suggest that at least two antibodies (DC1 4C7 and DC3 10D6) recognised different sites on the /3galactosidase molecule. However, in practice, when used in the G A G assay mixed complexes gave lower binding than the best complex DC1 4C7 used alone. For the analysis of cell surface antigens, enzyme-linked immunoassays have already undergone many improvements and refinements. These include the attachment of cells to plates in advance of requirement (Stocker and Heusser, 1979), the use of the enzyme/anti-enzyme technique (Lansdorp, 1980), the use of enzyme systems in which endogenous enzyme activity does not contribute to background (Cobbold and Waldmann, 1981), the use of a fluorogenic substrate (Labrousse et al., 1982) and miniaturisation of the assay (Pateraki et al., 1981; Labrousse et al., 1982; Michaelides et al., 1983). This report describes an assay which combines these attributes with the use of a monoclonal antibody/fl-galactosidase complex to achieve significant overall improvements, not least in sensitivity.

27

References Avrameas, S.P., H~Ssli, P., Stanislawski, M., Rodriget, M. and Vogt, E. (1979) J. Immunol. 122, 648. Barnstaple, C.J., Bodmer, W.F., Brown, G., Galfr6, G., Milstein, C., Williams, A.F. and Ziegler, A. (1978) Cell 14, 9. Bodmer, J.G., Heyes, J.M. and Lindsay, J. (1985) In: E.D. Albert (Ed.), Histocompatability Testing 1984 (SpringerVerlag, Berlin) pp. 432-438. B/Syum, A. (1968) Stand. J. Clin. Lab. Invest. 21 (suppl.), 97. Brodsky, F.M., Parham, P., Barnstaple, C.J., Crumpton, M.J. and Bodmer, W.F. (1979) Immunol. Rev. 47, 3. Cobbold, S.P. and Waldmann, H. (1981) J. Immunol. Methods 44, 125. De Kretser, T., Crumpton, M.J., Bodmer, J.G. and Bodmer, W.F. (1982) Eur. J. Immunol. 12, 600. Douillard, J.Y., Hoffman, T. and Herberman, R.B. (1980) J. Immunol. Methods 39, 309. Eckels, D.D., Woody, J.N. and Hantsman, R.J. (1981) Hum. Immunol. 3, 133. Ey, P.L., Prowse, S.J. and Jenkin, C.R. (1978) Immunochemistry 15, 429. Galfr6, G., Howe, S.C., Milstein, C., Butcher, G.W. and Howard, J.C. (1977) Nature 266, 550. H~Ssli, P., Avrameas, S., Ullmann, A., Vogt, E. and Rodriget, M. (1978) Clin. Chem. 24, 1325. Ishikawa, E. and Kato, K. (1978) Scand. J. Immunol. 8 (suppl. 7), 43. Kelly, B., Levy, J. and Sikora, L. (1979) Immunology 37, 45. Kenna, J.G., Major, G.N. and Williams, R.S. (1985) J. Immunol. Methods 85, 409. Kennett, R.H. (1980) In: R.H. Kennett, T.J. McKearn and K.B. Bechtol (Eds.), Monoclonal Antibodies (Plenum, New York) p. 376.

Labrousse, J.-L., Guesdon, J., Ragimbeau, J. and Avrameas, S. (1982) J. Immunol. Methods 48, 133. Lansdorp, P.M., Astaldi, G.C.B., Oosterhof, F., Janssen, M.C. and Zeijlemaker, W.P. (1980) J. Immunol. Methods 39, 393. Lansdorp, P.M., Oosterhof, F., Astaldi, G.C.B. and Zeijlemaker, W.P. (1982) Tissue Antigens 19, 11. Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J. (1951) J. Biol. Chem. 193, 265. Michaelides, M.C., Sato, N. and Wallach, M.K. (1983) J. Immunol. Methods 58, 267. Ouchterlony, O. and Nilsson, C.A. (1978) In: D.M. Weir (Ed.), Handbook of Experimental Immunology (Blaekwell Scientific Pubfications, Oxford) p. 19. Ozato, K., Mayer, N. and Sachs, D.H. (1980) J. Immunol. 124, 533. Parham, P. and Bodmer, W.F. (1978) Nature 276, 397. Pateraki, E., Guesdon, J.-L., Serie, L. and Avrameas, S. (1981) J. Immunol. Methods 46, 361. Shannon, A.D., Rudd, C.E., Bodmer, J.G., Bodmer, W.F. and Crumpton, M.J. (1985) In: E.D. Albert (Ed.), Histocompatability Testing 1984 (Springer-Verlag, Berlin) pp. 439-442. Sternberger, L.A. (1979) Immunocytochemistry (Wiley, New York). Stocker, J.W. and Heusse, C.H. (1979) J. Immunol. Methods 26, 87. Strauss, W. (1979) J. Histochem. Cytochem. 27, 1349. Taborsky, G. (1974) Adv. Protein Chem. 28, 1. Voller, A. and Bidwell, D.E. (1981) ELISA, Vol. 2 (Microsysterns, Guernsey) pp. 1-126. Voller, A., Bartlett, A. and Bidwell, D.E. (1978) J. Clin. Pathol. 31,507.