Analysis and application of substrate hydrolysis rates in indirect ELISA of a purified plant virus

Jourrlal of Virological Elsevier

Merhods.

19 (1988)

141

IJI-150

JVM 00684

Analysis and application of substrate hydrolysis rates in indirect ELISA of a purified plant virus V.L. Vilker’

and B.J.M.

Verduin*

‘Department of Chemical Engineering, University of California, Los Angeles, California, U.S.A.; ‘Department of Virology, Agricultural University, Wageningen, The Netherlands (Accepted

20 November

1987)

Summary

The transient calorimetric signal in a microtiter plate is used to quantify a purified plant virus, cowpea mosaic virus (CPMV), over five concentration decades in a single plate. The method involves the coating of the polystyrene microtiter plate wells directly with the CPMV antigen, followed by incubation with a rabbitderived CPMV-specific antibody, and lastly by incubation with a commercially available antibody against rabbit immunoglobulin which has been pre-labeled with alkaline phosphatase. The rate of p-nitrophenylphosphate hydrolysis, both nonspecific and that which was catalyzed by this enzyme, was measured spectrophotometrically at 405 nm. Enzyme-catalyzed hydrolysis rates followed first order kinetics at all antigen coating concentrations, and the 1” rate constants, which ranged from 2 X 10m6 min-’ to 1 X 1O-3 min-‘, were found to increase with increasing antigen concentration. Plant virus; ELISA; Hydrolysis rate

Introduction

Investigations about interactions of virus particles with biological material and other adsorbing substrates (in solution) require accurate measurement of virus concentration over a wide range. Enzyme-linked immunosorbent assay (ELISA) methods are among the most rapid and sensitive. The great sensitivity of the ELISA method is an important characteristic leading to its emerging applications for quantitative as well as qualitative antigen detection (Van Weeman, 1985). This Correspondence fo: V.L. Vilker, Angeles, CA 90024, U.S.A.

Department

of Chemical

Engineering,

University

of California,

Los

142

sensitivity depends on the combined effects of (a) selectively isolating the antigen on a solid adsorbent support using specific antibodies. and (b) employing a colorimetric detection system of sufficiently high sensitivity and discrimination to accurately measure antigen concentration. High sensitivity can only be obtained when there exists a high ratio of ‘signal’ (calorimetric change due only to antigen presence in adsorbed state) to noise’ (calorimetric change due to background effects not unique to antigen presence). Reagent and technique variables involved with each of the effects (a) and (b) are well reviewed by Clark (1981). For viral antigens which are highly purified, several investigators find that indirect ELISA on unprecoated plates (that is, virus bound directly to solid phase) gives improved sensitivity relative to the direct double antibody sandwich ELISA (Koenig, 1981; Koenig and Paul, 1982; Lommel et al., 1982; Ehlers and Paul. 1984). In this paper, we describe an indirect ELISA technique which we used to quantify changes in cowpea mosaic virus (CPMV) particle concentration during the interactions of this virus with oxide surfaces in electrolyte solutions. By this technique, we were able to measure up to lOO,OOO-fold changes in virus concentration. This appears to extend the range of concentration changes available by ELISA methods by a factor of 10-100 over that which has been previously achieved in other investigations of plant virus systems. This was achieved by analysis of the transient units) which occurred in the calorimetric signal (for Abosn,,, below 2.0 absorbance ELISA plate wells during the course of p-nitrophenylphosphate hydrolysis. The hydrolysis was catalyzed by antibody-bound alkaline phosphatase. The reaction rates are modeled as 1” order and the rate constants determined at each different CPMV coating concentration.

Materials Cowpea mosaic virus The CPMV used in these studies was the SB strain which causes severe yellowing in cowpea. The virus was cultured and purified according to the methods described in Van Kammen (1967). Virus was stored in 0.01 M phosphate buffer pH 7.0 at 4°C. The virus concentration of this stock preparation (determined spectrophotometrically using an extinction coefficient at 260 nm of 8.1 mg/ml) was 3.85 mg CPMV/ml and was unchanged over the eight month study. Antisera and immunoglobulin The antisera were prepared using rabbits which were given one milligram injections of CPMV-SB. A total of three injections were given at 10 day intervals. Immuno-gamma globulin (IgG) fractions were prepared from antisera using methods described by Clark and Adams (1977). Enzyme-labeled anti-&G conjugate The conjugate used was a commercially available goat anti-rabbit with alkaline phosphatase (Tago Inc., Burlingame, California, Product

IgG* labeled Code 6530).

143

Solutions and materials The following buffers were used: phosphate buffered saline (0.01 M phosphate, 0.15 M NaCl, pH 7.4) containing 0.5 ml/l Tween 20 (PBST); PBST containing polyvinylpyrrolidone (molecular mass 44 kDa) 20 g/l (antigen buffer): antigen buffer containing ovalbumin 2 g/l (conjugate buffer); carbonate-coating buffer (0.05 M carbonate, pH 9.6); diethanolamine in water 10% (v/v) adjusted to pH 9.8 with HCl (substrate buffer). Substrate solutions were prepared fresh before each determination by dissolving p-nitrophenylphosphate disodium (pNP, Sigma Chemical Co.) in substrate buffer to a concentration of 0.6 mg/ml.

Methods

Virus dilutions Standard solutions of CPMV in carbonate buffer were made up at six concentrations (10 p_g CPMVlml, 1 kg CPMViml, . . . . 0.0001 kg CPMV/ml) by diluting an aliquot of the stock virus preparation immediately before introduction of virus to the microtiter plates. These dilutions were carried out as rapidly as possible in 15 ml glass test tubes which were ‘silanized’ prior to use. This silane coating was found necessary in order to minimize virus binding to glass resulting in excessive and variable losses of virus from coating buffer prior to application of the standard solutions to the plates. In the first step of the silane-coating process, glassware which had been previously detergent washed, rinsed three times in deionized water and air-dried overnight. was oven dried at lOO-120°C for 2 h. After cooling to room temperature, the test tubes were then soaked for one hour in a solution of 2% dimethyldichlorosilane in l.l,l-trichloroethane (BDH Chemicals Ltd., Poole, U.K. Product No. 33164). The tubes were drained, oven dried at lOO-120°C for one hour, and finally rinsed three times with deionized water and air dried overnight. The microtiter plates were flat-bottomed polystyrene microELISA plates from Dynatech Laboratories, Inc. Plate coatings and setups Four layers of reactants were applied to microtiter plates in volumes of 0.2 ml/well in the following order: 1st layer: CPMV standards, unknowns from interaction with oxides (these studies not discussed here), or deionized water; 2nd layer: IgG in antigen buffer at 2 kg IgG/ml; 3rd layer: Enzyme conjugate in conjugate buffer at 0.5 pg/ml or 1 pgirnl; 4th layer: Substrate solution. Two different incubation protocols were used and both gave similar results; protocol 1: plate incubation for at least 12 h at 4°C after layer 1 application, incubation for 4-6 h at 37°C after layer 2, incubation for at least 12 h at 4°C after layer 3; protocol 2: plate incubation for 2-4 h at 37°C after layer 1 application, incubation for at least 12 h at 4°C after layer 2, incubation for 4-6 h at 37°C after layer 3. Plates were covered during each incubation in order to minimize temperature variations across the plate. After allowing them to equilibrate to room temperature the plates were washed as described below. Preliminary experiments revealed

144

some variation in calorimetric signal with the position which a sample occupied on the 96-well plates. Trends were not clear cut but suggested that increased reproducibility could be achieved by not placing samples in the outer wells of the plate. We also minimized the systematic variations associated with row and column positions (McLaughlin et al., 1981; Burrows et al., 1984) by pairing each standard sample in an inner well with a sample placed in a relatively more outer location. Four samples at each of the six standard CPMV concentrations were thus located on a single plate. Ultimately, the calorimetric signal corresponding to each concentration was the average reading of these four wells. Plate washing After each incubation. plate conditioning for the next reaction step was begun with a deionized water flush from a squirt bottle while the plate was held in inverted position. The plate wells were next filled with PBST buffer by applying the buffer gently from a squirt bottle while tilting the plate in such a manner that buffer runoff flowed down from the plate end containing the least concentrated antigen coating to the end with the most concentrated coating. The plate was allowed to soak in buffer for about 3 min before the water flush was repeated as above. These steps were repeated twice more before the next incubation layer was applied. Substrate application and plate development CPMV antigen concentrations which were in the decade of 1 kg CPMViml to 10 kg CPMV/ml were difficult to interpret because plate color development was rapid, and absorbance values exceeded the 2.0 A maximum before many time exposures could be obtained with an automated plate reader (Titertek Multiskan. Flow Laboratories). To facilitate the interpretation of these early time substrate hydrolysis rates at high antigen concentrations, it was necessary to standardize the procedures for handling substrate solutions just prior to application on the plates (layer 4), and to establish the earliest time at which the rate of color development in all four plate wells at the highest virus concentration had become equal. In order to achieve this. substrate solutions were prepared fresh from refrigerated substrate buffer and pNP which had been stored at -20°C. We found that the quality of these substrate solutions was the most important procedural variable for obtaining low ‘noise’. strong ‘signal’ calorimetric data during plate development. The solutions were stored on ice while plates were being washed following the layer 3 incubation period. Three-tenths milliliter of substrate solution was added to the wells in two passes over a plate with an g-tip microplate pipette applicator. A plate could be substrate loaded in 2 min. The earliest plate reading time was established by studying the variability of the rates at which the yellow color developed in the four wells at antigen concentration 10 pg CPMV/ml. Typically, at about ten minutes after substrate application, the range of absorbance measurements would be 0.3-0.5 A. At about 20 min this range narrowed to 0.1-0.3 A and remained constant. This constancy of signal range after 20 min was true for all experiments and therefore became the earliest time for analyzing calorimetric data from a plate.

145

Two types of controls were used on each plate in order to monitor the rate of non-specific color development. In the first type of control, some wells were filled with deionized water at the time of layering steps 1-3, but were otherwise treated the same way as the sample wells. In these controls, color development was always below 0.10 A for experiments lasting up to 810 min. In the second type of control. some wells were filled with deionized water during the antigen coating step (layer 1), but otherwise received the same layer 2, 3 and 4 solutions which were applied to the sample wells. Results reported here are only for experiments in which color development in these controls was below 0.15 A for the duration of plate development, up to 810 min. About one-third of all experiments attempted were discarded due to this last control being in excess of 0.5 A after 120 min. Plates were covered and maintained at room temperature during color development. Data analysis After the minimum time of about 20 min, absorbance measurements were made at appropriate intervals using the automated plate reader. Measurements from the type 2 control wells described above were averaged and used to correct absorbance measurements for wells containing virus-antibody-antibody bound alkaline phosphatase. These correction factors for non-enzyme related PNP hydrolysis were a significant fraction of the overall calorimetric response for wells coated with 0.001 kg CPMV/ml or less of the antigen. Absorbance data were recorded up to 2.0 A405nm units. Concentration of PNP product was calculated using an extinction coefficient of 1.85 l/(mmol-mm) (Bergmeyer, 1983). The fraction of p-nitrophenylphosphate unreacted at any time in the 10 mm cuvette was obtained from the relation

s

-ZZ so

S&,,/(1.85

x 10) S”

where So is the initial pNP concentration and A405 is the average ing for the four plate wells at each virus antigen concentration control absorbance measurement.

(I) absorbance readminus the type 2

Results and Discussion The microtiter plate wells were coated with CPMV antigen from carbonate coating buffer solutions in which the virus concentrations were varied from 0.0001 kg/ml-l0 kg/ml. The immunoglobulin concentration was constant throughout at 2 pg/ml. The concentration of alkaline phosphatase labeled conjugate was tested at two levels, 0.5 kg IgG*/ml and 1 kg IgG*/ml, without showing an effect. Substrate initial concentration was constant for all experiments at So = 0.6 mg pNPlm1. Fig. 1 shows a typical plate development pattern as A4r,5 increases as a function of time at the six CPMV coating concentrations. The double-headed arrows for

2or

Log

Fig. 1. Increasing absorbance of time and antigen (cowpea

(pg

CPMV/ml

1

due to p-nitrophenylphosphate conversion to p-nitrophenol, as function mosaic virus) concentration. Error bars at each reading represent range of four measurements (wells) from same plate.

each data point indicate the range of absorbance measurements for the four sample wells at each concentration. The lines are drawn through the arithmetic mean of the four measurements. The slope of these lines between each decade is an increasing function of time. Therefore, unknowns at the lower concentration values become more easily determinable at later times, so long as the background color development due to unspecific JJNP hydrolysis did not emerge faster than the slope development of the concentration decade 0.0001 P.g CPMViml to 0.001 Fg CPMViml. For the data shown in Fig. 1. the absorbance at 810 min for the sam-

b

c1

1_1

180 Time

(mln)

Time

(mln)

Fig. 2. Fraction of p-nitrophenylphosphate unreacted as function of time and CPMV antigen concentration, C,,. Initial pNP concentration S,, = 2.28 mM (0.6 mg pNP/ml). 2a; 0 C,, = 10 kg/ml, 0 C,, = 1 kg/ml. A C,, = 0.1 pgiml. 2b: n C,, = 0.01 pgiml. 4 C,, = 0.0001 kg/ml. A C,, = 0.0001 kg/ml.

137

ples at C, = 0.001 kg CPMViml is 0.31 while the control wells were at 0.16, indicating the importance of taking the ‘noise’ of the calorimetric data into account. Only one experiment provided statistically significant absorbance data at C,, = 0.0001 kg CPMViml. For each CPMV concentration, A&t) data were converted to fraction of substrate unconverted, s(t)/&, using equation 1. Total conversion was always low since at about 5% pNP conversion to p-nitrophenol. absorbance at 405 nm exceeds 2.0 A. The conversion was found to obey first order kinetics at all concentrations C,,. Curves like those shown in Figs. 2a and b were fit to a 1” order model for all experiments: ln(S/&,)

= -kt

(2)

and rate constants k determined at each concentration C,,. Table 1 gives a summary of these rate constant determinations. Fig. 3 shows that k increases with C,, with about a three-fourths power dependence. The pNP hydrolysis rate constant would be expected to depend on efficiency of antigen binding to the microELISA plate. IgG antisera concentration and IgG* conjugate concentration. In our experiments, IgG and conjugate IgG* particle concentrations were always at least six times larger than the adsorbed CPMV particle number when using 5.5 x lo6 for molecular weight (Van Kammen, 1967) and assuming that CPMV adsorption is 100%. For complete adsorption from a solution containing 10 Fg CPMV/ml, the fractional area of a well which is covered by virus is less than 1%. The coverage would be significantly less than this estimate if the more reasonable assumption is made that CPMV binding to the microELISA plate obeys a linear Henry’s law partitioning over the whole range of coating solution CPMV concentrations. In either case, it is clear that the overall hydrolysis kinetics of pNP is a dilute solution event with low substrate conversion. It is also clear that reproducible hydrolysis reaction rates are critically dependent on minimizing losses of antigen particles prior to coating, such as those losses we first observed to the glassware used in preparing antigen standards. McLaughlin et al. (1981) reported on pNP hydrolysis kinetics in their study of clover yellow vein virus (CYVV) detection by direct ELISA with an alkaline phosTABLE 1 First order tration.

rate constants

for p-nitrophenylphosphate

CU. wg CPMViml

No. experimental determinations

10 1 0.1 0.01 0.001 0.0001

6 9 9 9 9 1

hydrolysis as function of CPMV antigen concen-

k, min-’ (1.2 k (8 (3.0 t (6 (6

0.1) 2 2) 0.5) t 1) k 2) 2.4

x x x x x x

10-a lo-” 10-J 10m5 10mh 10-h

-7 -5

I -4

I -3

I -2

I -1

I 0

I 1

log Co, /IgCPMV/ml

Fig. 3. Variation

of hydrolysis number

rate constant with CPMV antigen concentration. of measurements at each CPMV concentration.

C,,. See Table

1 for

phatase labeled IgG* conjugate. They also reported first order kinetics at the single virus antigen concentration 1 pg CYVV/ml. Absorbance measurements were made at 400 nm. However, their analysis of the reaction rate constant is not directly comparable to the data we report in Table 1. In their work, equations 1 and 2 are combined to give a two-parameter equation; A = a{ 1-exp(-kt))

(3)

Both parameters (Yand k were determined by least squares curve fit of absorbance measurement extending to 12 h. The results were (Y = 2.266 and k = 0.296 h-‘. = 70. then k can be recalculated If the parameter (Yis first set equal to (~~~,,xlxS,,) from their data to be in the range 0.003-0.01 hh’. more in line with the value of Table 1 although the first order rate expression represents the CYVV data less satisfactorily. Although both studies employ alkaline phosphatase to catalyze pNP hydrolysis, the many other differences between the two studies may play a significant role in the observation of different overall reaction rates.

Acknowledgments We thank Dick Lohuis for preparation of the antisera and labeled IgG* conjugate. The support of the Netherlands Commission and the Agricultural University of Wageningen program to VLV is gratefully acknowledged. Partial support US National Science Foundation Grant CPE 8212227.

IgG and the enzyme American Fulbright Research Fellowship was provided by the

149

References Bergmeyer, H.U. (1983) Methods of Enzymatic Analysis. 3rd ed.. Vol. 2. 269. Burrows, P.M.. Scott, S.W.. Barnett. O.W. and McLaughlin. M.R. (1984) J. Viral. Methods 8, X7-216. Clark, M.F. (1981) Ann. Rev. Phytopathol. 19, 83-106. Clark, M.F. and Adams, A.N. (1977) J. Gen. Viral. 34. 475-483. Ehlers, U. and Paul. H.L. (1984) J. Virol. Methods 8. 717-224. Koenig. R. (1981) J. Gen. Viral. 55. 53-62. Koenig. R. and Paul. H.L. (1982) J. Virol. Methods 5. 113-125. Lommel. S.A., McCain. A.H. and Morris. T.J. (1982) Phytopathology 72. 1018-1032. McLaughlin. M.R.. Barnrtt. O.W.. Burrows, P.M. and Baum. R.H. (1981) J. Viral. Methods 3, l-F-25. Van Kammen, A. (1967) Virology 31, 633-642. Van Weeman. B.K. (1985) J. Viral. Methods IO. 371-378.