Sensitive fluorogenic enzyme immunoassay on nitrocellulose membranes for quantitation of virus

Journal of Virological Methods, 22 (1988) 149-164 Elsevier

149

JVM 00789

Sensitive fluorogenic enzyme immunoassay on nitrocellulose membranes for quantitation of virus Roberta E. F&on,

Jonathan

P. Wong, Yunus M. Siddiqui and May-S. Tso

Biomedical Defence Section, Defence Research Establishment Sufjield, Ralston, Alberta, Canada (Accepted

24 May 1988)

Summary

A highly sensitive fluorogenic enzyme-linked immunosorbent assay (FELISA), which utilizes nitrocellulose membranes as solid phase support, has been developed for the detection and identification of virus in clinical samples. Reagents were standardized and, using purified Newcastle disease virus (NDV) as a model, the theoretical lower limits of test sensitivity of the FELISA were compared, in both “sandwich” and “indirect” formats, to those of a comparable chromogenic enzyme-linked immunosorbent assay (CELISA). Of the systems evaluated, the “sandwich” FELISA exhibited maximum sensitivity and detected 10 fg of purified virus protein per milliliter of test sample (500 ag per test volume). Specificity of the “sandwich” FELISA was evaluated by challenging the system with heterologous strains of NDV and with other serologically related and unrelated viruses. In a clinical trial in which fecal materials from chickens undergoing vaccination with NDV were assayed directly by FELISA, the virus was detected from the first to approximately the tenth day post-vaccination. The test is simple to perform and results can be obtained in approximately 4 h. Fluorescence immunoassay; Fluorogenic enzyme-linked immunosorbent assay; Nitrocellulose; Newcastle disease virus; Diagnosis; Detection and identification

Correspondence tot R.E. F&on, Biomedical field, Ralston, Alberta, Canada TOJ 2N0.

Defence

Section,

Defence

Research

Establishment

Suf-

Introduction

A variety of solid phase enzyme immunoassay procedures has been applied to the direct detection and quantitation of microbial antigens in body fluids (Voller et al., 1980; Yolken, 1982). However, conventional methods of solid phase enzyme immunoassay are, in general, insufficiently sensitive for the measurement of microbial antigens which occur in body fluids in low concentration (Chao et al., 1979; Pronovost et al., 1981; Harmon and Pawlik, 1982). One factor which influences the degree of sensitivity achievable is the physical properties of the solid phase selected. Although polystyrene is the most commonly used solid phase support for attachment of proteins in enzyme immunoassays (Voller et al., 1980), its effective binding capacity for proteins is low (Solonen and Vaheri, 1979). This limitation is particularly serious in assays which utilize antibodies attached to the solid phase to capture antigens, since the amount of antigen which can be bound is proportional to the concentration of antibody which can be adsorbed to the solid phase (Yolken, 1982). Another factor influencing sensitivity in enzyme immunoassays is the extent of detectability of the substrate products. Although colored products of enzyme-substrate reactions are the most commonly used indicator systems in conventional enzyme immunoassays, fluorescent end products of enzyme reactions are detectable at lower concentrations (Fernley and Walker, 1965; Kato et al., 1976). Indeed, the sensitivity of enzyme immunoassays which utilize fluorogenic substrates has been demonstrated to be higher than that of assays which utilize the same enzyme with a chromogenic substrate (Yolken and Stopa, 1979). In this paper, we describe a microplate enzyme immunoassay with enhanced sensitivity for the detection and quantitation of virus directly in body fluids. Using NDV as a model, we have optimized test sensitivity by combining the high protein-binding capacity of nitrocellulose membrane as solid phase with the high sensitivity achievable by use of a high-energy fluorogenic substrate. The sensitivity of this new enzyme immunoassay system is approximately two million times greater than that of conventional microplate immunoassays which utilize polystyrene as solid phase support and a chromogenic substrate (R.E. Fulton, unpublished data).

Materials

and Methods

Antibodies

Guinea pigs (albino, Hartley outbred) (Charles River Ltd., St. Constant, Que.) were immunized with NDV strains NJ-La Sota and Bl, respectively, by intramuscular (IM) administration of purified virus and Freund’s incomplete adjuvant (l:l), containing 100 kg of virus protein. Animals were subsequently boosted at two-week intervals for a period of four weeks with a similar 1:1 virus-emulsion mixture containing 50 pg of purified virus protein. Blood samples were collected by heart puncture and the antibody response to NDV was ascertained by hemagglutination inhibition (HAI) test. Leghorn chickens (Shaver) (La Poulet Poultry Farm, Medicine Hat, Alta.) were

151

immunized initially with a single dose of NDV vaccine (NJ-La Sota strain) (Salsbury Laboratories, Kitchener, Ont.), administered in the drinking water, according the manufacturer’s instructions. Chickens were subsequently boosted at weekly intervals for a period of four weeks by IM administration of purified virus and incomplete Freund’s adjuvant (l:l), containing 100 pg virus protein. Blood samples were collected from the medial wing vein and assayed by HA1 test for antibody to NDV. IgG was purified from NDV-immune guinea pig and chicken sera by salt fractionation and ion exchange chromatography, performed under contract by Jackson ImmunoResearch Laboratories (Avondale, PA). Hyperimmune ascitic fluids to NJ-La Sota and Bl strains of NDV were prepared in Swiss Webster mice (COBS outbred) (Charles River Ltd.) by a modification of procedures described by Russell et al. (1970) and Chiewsilp and McCown (1972). On day 1, mice were inoculated with purified virus (200 &ml): 0.2 ml intraperitoneally (IP), 0.2 ml subcutaneously (SC), 0.1 ml IM and 0.1 ml of complete Freund’s adjuvant at the IP site. This schedule was repeated on day 3 with the exception that no adjuvant was given. A third and fourth series of injections of purified NDV were given, IP, on days 24 and 30, respectively, and on day 30 each mouse also received, IP, 0.5 ml of freshly harvested mouse sarcoma cells (S-180) (American Type Culture Collection, Rockville, MD) at a concentration of 2 x 10’ cells per ml. On day 43, ascitic fluids were harvested by paracentesis and tested by HA1 for antibody to NDV. Immunoglobulins were precipitated by a procedure described by Volk et al. (1982). Normal rabbit serum was purchased from Miles Scientific (Naperville, IL), Alkaline phosphatase-labelled affinity purified goat anti-guinea pig (Kirkegaard and Perry) and anti-mouse IgG antibodies were purchased from Mandel Scientific (Edmonton, Alta.) and Bio-Rad Laboratories (Mississauga, Ont.), respectively. Viruses

NDV (NJ-La Sota, Bl and NJ-Roakin), parainfluenza - 1 (Sendai), 2 and 3 (HA1) - and influenza (A/PR/8/34) were purchased from the American Type Culture Collection and were cultivated by established methods (Hawkes, 1979) in the allantoic cavity of embryonated hens’ eggs. Harvested allantoic fluids were assayed for virus titer by hemagglutination (HA) test. NJ-La Sota and Bl strains of NDV were purified from allantoic fluids by two cycles of differential centrifugation (6400 X g for 15 min; 39,800 x g for 1 h) followed by discontinuous density gradient centrifugation (113,000 x g for 4 h) in 20, 40 and 60% (w/v) sucrose in 0.01 M Tris-HCl buffer, pH 7.4, containing 0.1 M NaCl and 1 mM EDTA. Virus was further purified by centrifugation (113,000 x g for 4 h) on preformed continuous sucrose gradients ranging in concentration from 17 to 53% (w/v). Fractions containing virus, as indicated by peaks in absorbance (260 and 280 nm) and HA activity, were pooled.

152

Enzyme immunoassays FELISA and CELISA were evaluated and compared both in a “sandwich” format, in which capture antibody (CAb) was adsorbed to the solid phase to capture the antigen and in an “indirect” format, in which the antigen to be detected was applied directly to the solid phase. In both methods, antigen was subsequently detected by unlabelled detector antibody (DAb) and indicated by enzyme-labelled indicator antibody (IAb). All FELISA and CELISA reactions were carried out in 96-well MillititerTM HA filtration plates (Millipore Corp., Mississauga, Ont.) in which 0.45 km nitrocellulose membranes formed the bottom surface. Removal of unbound reagents and washing steps were achieved by vacuum filtration in a Millititer filtration system (Millipore Corp.). Immediately prior to use, wells were washed three times with phosphate buffered saline, pH 7.4 (PBS). Unless otherwise indicated, incubation was carried out at 37°C. A volume of 200 ~1 of washing solution per cycle was used in washing steps. Optimal working dilutions of DAb and IAb were determined by checkerboard titration. Conditions for blocking, incubation and washing which yielded the highest ratio of positive to background fluorescence or absorbance (FELISA and CELISA, respectively), were adopted for routine use. FELISA For the “sandwich” method, wells were sensitized with the optimal concentration of CAb by the addition of 50 ~1 of mouse anti-NDV IgG (20 pgiml) in 0.05 M carbonate-bicarbonate buffer, pH 9.6. After overnight incubation at 4°C wells were washed three times with PBS and unoccupied sites blocked by incubating for 1 h with 200 ~1 of PBS containing 2% bovine serum albumin (BSA) and 0.1% Tween 20 (T) (PBS-BSA-T). Wells were again washed with PBS and the blocking step was repeated twice, followed by an additional three washes with PBS. Test NDV (50 Al), diluted in PBS-BSA-T, was then added. Plates were incubated for 1 h, then washed five times with PBS. Fifty microliters of the optimal dilution (1:2500 in PBS-BSA-T) of DAb (guinea pig anti-NDV IgG) was then added and incubation carried out for 1 h, followed by five cycles of washing with PBS containing 0.05% Tween 20 (PBS-T). Fifty microliters of the optimal dilution (1:2500 in PBS-BSA-T) of IAb (phosphatase-labelled goat anti-guinea pig IgG) was then added and the plates incubated a further hour. After six cycles of washing with PBS-T, 200 ~1 of lop4 M 4-methylumbelliferyl phosphate (4-MUP) (Sigma Chemical Co., St. Louis, MO) in 10% diethanolamine buffer, pH 9.8, was added and the plates incubated 15 min at room temperature in the dark. The fluorescent product resulting from hydrolysis of the substrate by bound enzyme-labelled antibody was measured fluorometrically. For the “indirect” method, wells were sensitized with 50 ~1 of purified NDV diluted in 0.05 M carbonate-bicarbonate buffer, pH 9.6. Plates were incubated overnight at 4°C then washed three times with PBS. Blocking and all subsequent steps were performed as described for the “sandwich” method.

153

CELISA

The CELISA “sandwich” and “indirect” methods were performed as described for the FELISA “sandwich” and “indirect” methods, respectively, with the exception that, following the final wash, 200 l_dof p-nitrophenyl phosphate (PNP) (Sigma Chemical Co.), 1 mg/ml in 10% diethanolamine buffer, pH 9.8, was added to the wells and incubated 30 min at room temperature in the dark. The colored product resulting from hydrolysis of the substrate by bound enzyme-labelled antibody was measured calorimetrically. Data acquisition and analysis

For the FELISA, relative fluorescence of the enzyme-substrate reaction product was measured directly on Millititer HA plates by a microFLUOR fluorometer (Dynatech Laboratories, Alexandria, VA) fitted with 365 nm and 450 nm filters for excitation and emission, respectively. For the CELISA, reation products were transferred from Millititer plates to flat bottomed 96-well Immulon II microwell plates (Dynatech Laboratories) and absorbances read at a wavelength of 405 nm with a Titertekm Multiscan microplate calorimeter (Flow Laboratories, Mississauga, Ont.). Appropriate controls were included for each test. Results were considered positive if the mean fluorescence and absorbance readings by FELISA and CELISA, respectively, were equal to or greater than two standard deviations (SD) above the mean readings of respective control wells. Clinical evaluation

Fecal materials were collected at La Poulet Poultry Farm from chickens undergoing vaccination for NDV. Chickens (Babcock leghorn), 5 weeks of age, were administered NDV (Bl) and infectious bronchitis virus (IBV) (Mass. and Conn.) combined vaccine (Salsbury Laboratories) in the drinking water, according to manufacturer’s protocol. Fecal materials, collected one day before and at regular intervals over a two-week period following vaccination, were suspended in PBS-BSA-T (1 g/10 ml) and mixed thoroughly on a vortex mixer. Solid components were removed by centrifugation (5000 x g/l5 min) and 50 u,l of varying dilutions of the supernatants were tested for NDV (Bl) by “sandwich” FELISA, competitive inhibition “sandwich” FELISA, HA and HAT tests. Fecal materials from unvaccinated chickens served as negative controls. HA and HAI assay

HA and HA1 tests were performed with 0.75% rooster erythrocytes (Institute Armand Frappier, Laval, Que.) by standard techniques (Grist et al., 1974). Protein estimation

Protein concentrations of immunoglobulin and of purified virus preparations were estimated by a micro-Lowry procedure (Layne, 1957).

154

3600I-

2800

,-

w I

I

I

-9

I

1

I

LOG MOUSE

ANTI-NDV

I

-3

-6 IgG IglmLI

Fig. 1. Optimization of CAh. Fifty microliter volumes of varying concentrations (1 @ml to 100 pgiml) of mouse anti-NDV IgG, diluted in 0.05 M carbonate-bicarbonate buffer, pH 9.6, were immobiIized on nitrocellulose and titrated fluorometrically with the optimal dilution (1:2500) of phosphakase-labelled anti-mouse IgG diluted in PBS-BSA-T. Data points represent the mean of triplicate determinations on a single plate. Error bars represent SD of the mean.

Results

Optimization of CAb The optimal concentration of CAb for the FELISA and CELISA “sandwich” procedures was determined fluorometrically by titration with the optimal dilution of enzyme-labelled anti-species antibody. From the data obtained, a curve relating CAb concentration and fluorescence count was constructed (Fig. 1). Fluorescence count increased with increasing additions of CAb to a concentration maximum, beyond which, further addition of CAb had little effect on the fluorescence count. The optimal concentration of CAb, defined as the lowest concentration yielding the highest fluorescence count, was determined from this curve to be 20 pgiml (1 ~glwell).

of FELISA and CELISA

Sensitivity

The lower limits of test sensitivity of the FELISA and CELISA “sandwich” and “indirect” methods, respectively, were determined by titration with varying concentrations of purified NDV (NJ-La Sota). The FELISA “sandwich” method detected virus at a concentration of 10 fgknl (500 agiwell), whereas the detection limit by the CELISA “sandwich” method was 10 ng/ml (500 pg!well) (Fig. 2). By the FELISA “indirect” method, the lowest concentration of virus detectable was 100 pg/ml (5 pg/well), while the limit of sensitivity by the CELISA “indirect” method was 10 ng/ml (500 pg/well) (Fig. 3).

1400-

1200 -

O-O A--d o---O A---A

FELlSA CELISA F&ISA CONTROL +2 SD CELISA CONTROL i2SD

l.‘ooo

2 u ifi 8

800

0

D

5

poo 400

200

--__o---(j

--*--_-~_~~&

~_-___d_~_&_ ----__

_)

-6

-9 LOG NDV CONCENTRATION

-12

-1! 5

IgfmLt

Comparison of sensitivities of FELISA and CELISA by the “sandwich” method. Varying concentrations of purified NDV (10 p&ml to I fg/mI) were titrated by FELISA and CELISA and fluorescence count and absorbance, respectively, determined. Data points are the mean of triplicate determinations on a single plate. Error bars represent SD of the mean.

Fig. 2.

156

! 0A-A

*-_-A

1.0

FELISA CELISA FELISA CONTROL +2 SD CELISA CONTROL +2 SD

0.8

0.2

1

I

I

-6

I

I

I

I

-9 LOG NW

CONCENTRATION

I

-12 (g/mL)

Fig. 3. Comparison of sensitivities of FELISA and CELISA by the “indirect” method. Varying concentrations of purified NDV (10 kg/ml to 1 pgiml) were titrated by FELISA and CELISA and fluorescence count and absorbance, respectively, determined. Data points are the mean of triplicate determinations on a single plate. Error bars represent SD of the mean.

Specificity of FELISA

The specificity of the “sandwich” FELISA for the NJ-La Sota strain of NDV was investigated by challenging the assay system with the homologous and several heterologous strains of NDV and with other serologically related and unrelated viruses. Allantoic fluids containing the equivalent of one HA unit, respectively, of NJ-La Sota (homologous), Bl and NJ-Roakin (heterologous) strains of NDV, parainfluenza 1, 2 and 3, and influenza AlPRlSl34 were titrated by FELISA and the resulting respective titers compared (Fig. 4). All three strains of NDV were positive in the assay although the homologous strain (NJ-La Sota) reacted most strongly, followed in order of respective reactivity by the heterologous strains, Bl and NJ-Roakin. Other members of the paramyxovirus group, parainfluenza 1, 2 and 3 were also positive but at a comparatively reduced fluorescence count. Influ-

2400

--NDV NJ-LA SOTA

61

_____( NJ-ROAKIN

I

PARAINFLUENZA 1

2

----I 3

,NFLUENZA

CONTROL

AIPRi813.9

Fig. 4. Specificity of FELISA. Each infected allantoic fluid was diluted in PBS to contain one HA unit as follows: NDV NJ-La Sota (1:32), NDV Bl (1:32), NDV NJ-Roakin (1:512), parainfluenza 1 (1:512), parainfluenza 2 (1:4), parainfluenza 3 (1:4) and influenza AIPR8134 (1:32). Serial log dilutions of each unit HA stock, prepared in PBS-BSA-T, were tested by FELISA. FELISA control consisted of uninfected allantoic fluid. Numerals indicate dilutions of infected and uninfected allantoic fluids. FELISA titers are the mean of triplicate determinations.

enza AlPR/8/34, a member of the orthomyxovirus line control level.

group, reacted only at the base-

Clinical evaluation The sensitivity of the “sandwich” FELISA for detection of NDV in Sensitivity. crude clinical specimens was evaluated by challenging the assay system with fecal materials collected from chickens, at fixed time intervals, following vaccination with a combined vaccine containing the Bl strain of NDV and IBV. Each sample was titrated both by FELISA and by HA and, from the data obtained, curves relating fluorescence count and HA titer as a function of days post-vaccination, were constructed (Fig. 5). Both fluorescence count and HA titer increased over time from a negative baseline in samples taken one day prior to vaccination (negative control), to a peak concentration at day five, after which both fluorescence count and HA titer declined, reaching baseline level again at approximately day 10. From a standard curve relating fluorescence count and concentration of purified virus antigen, it was estimated that the FELISA detected NDV in fecal materials as early as one-half day and as late as 10 days, post-vaccination, at a concentration of approximately 50 ng/ml (2.5 @well). Maximum concentration of virus in fecal samples was reached at day five post-vaccination, at a concentration of approximately

158

2 c 1: 9 0

I

I

-1

I

I

I

I

I

I

I

1

3

5

7

9

11

13

DAYS POST-VACCINATION

Fig. 5. Clinical evaluation: sensitivity. Fecal materials collected one day prior and at regular intervals following vaccination with NDV (Bl) were tested by “sandwich” FELISA (C-O) and by HA (A-A) for the presence of NDV Bl antigen. FELISA data points are the mean of triplicate determinations; error bars represent SD of the mean. FELISA negative control consisted of fecal materials taken one day prior to vaccination. HA data points are the mean of duplicate determinations. The insert represents a portion of a FELISA standard curve for concentration. O--O, FELISA control plus 2 SD.

1 pg/ml (50 ngiwell). The assumption that HA activity was due to the presence of NDV (Bl) was confirmed by the finding that HA activity was inhibited in the HAI test by antisera specific to the Bl strain. Specificity. The specificity of the “sandwich” FELISA for NDV in fecal samples was confirmed by competitive inhibition tests in which immunoglobulins specific for NDV (Bl), or control sera containing no antibodies to NDV, were used to compete with the DAb for combining sites on antigens present in fecal materials from vaccinated or unvaccinated chickens. From the data obtained, curves relating fluorescence count and concentration of competing immunoglobulin were constructed (Fig. 6). When fecal materials from vaccinated chickens were used as antigen, the addition of increasing concentrations of specific competing immunoglobulin resulted in decreasing fluorescence counts, reaching a concentration plateau, after which further addition of competing immunoglobulin had little effect on the fluorescence count. In contrast, with increasing concentrations of control sera, the fluorescence count remained at a relatively constant, but elevated, level. When fecal materials from unvaccinated chickens were used as antigen, increasing concentrations of specific competing immunoglobulin had little or no ef-

159

feet on the fluorescence count and the level of fluorescence activity remained at the baseline level. These findings imply that specific competing immunoglobulin blocked combining sites for the DAb in fecal materials from vaccinated chickens, thus supporting the assumption that the identity of the antigen in fecal samples was NDV.

Discussion

The use of nitrocellulose as an adsorptive material for proteins, is well established (Kuno and Kihara, 1967; Gilman, 1970; Newman and Wilson, 1980). The now standard “Western blot” technique (Towbin et al., 1979) utilizes nitrocellulose membranes to adsorb proteins from polyacrylamide gels by an electrophoretic transfer procedure. In the so-called “dot-blot” procedure, first described by Hawkes (1982), it was demonstrated that proteins could be efficiently detected when spot-

1800

!i 2 8 Y

1600

x s Y 8 3 Y

1400

1200 1

I

-9 LOG CONCENTRATION

I

I

1

I

-6 COMPETING

I

,

-3 IMMUNOGLOBULIN

(6hLj

Fig. 6. Clinical evaluation: specificity. A competitive inhibition “sandwich” FELISA was performed on fecal materials collected on day five post-vaccination and on control feces from unvaccinated chickens. After incubation with the antigen and subsequent washing steps, and prior to the addition of DAb and subsequent reactants, wells were exposed for 1 h to 50 ~1 of chicken anti-NDV IgG, or control normal rabbit serum, followed by five cycles of washing. Fecal materials from vaccinated chickens reacted with chicken anti-NDV IgG (C-O) and normal rabbit serum (O-O); control fecal materials reacted with chicken anti-NDV IgG (A-A).

160

ted directly on nitrocellulose membranes. Similarly, nitrocellulose has been used as solid phase support in the detection and identification of a variety of different antigens by immunoenzymatic means (Palfree and Elliott, 1982; HoTejif and Hilgert, 1983; Bode et al., 1984; Davis et al., 1984; Berger et al., 1985). The major advantage in the use of nitrocellulose as solid phase in immunoenzymatic techniques is its high absorptive capacity for proteins. Close to 100% of the applied protein sample has been shown to bind to nitrocellulose membranes, even in the presence of detergent; by contrast, the conventional plastic microtiter wells are inefficient at binding protein (less than 8% of that applied) (Palfree and Elliott, 1982). The enhanced binding of proteins to nitrocellulose can, in part, be attributed to the large surface area available for adsorption. Whereas adsorption to standard polystyrene microtiter wells occurs on the surface only, adsorption on nitrocellulose membranes occurs both on and within the membrane pore matrix, thus providing an extremely large surface area for attachment. The use of fluorogenic substrates for detection of virus antigens by immunoassay techniques has also been previously documented. Yolken and Stopa (1979) reported a sensitive enzyme-linked fluorescence assay for the detection of rotavirus in stool specimens. Based on similar principles, fluorogenic immunoassay systems have been described for several other viruses, including herpes simplex virus (Shekareki et al., 1985), La Crosse arbovirus (Hildreth et al., 1982) and hepatitis B virus antigens (Ishikawa and Kato, 1978; Neurath and Strick, 1981). That fluorometric methods are more sensitive than calorimetric methods was recognized early (Lowry and Oliver, 1948) and it is now generally accepted that a theoretical lOO- to lOOO-fold increase in sensitivity can be achieved using fluorometric rather than calorimetric detection methods (Ishikawa and Kato, 1978; Shalev et al., 1980; Clark and Engvall, 1985). The concentration of CAb required for saturation of nitrocellulose membranes for the “sandwich” FELISA and CELISA was determined experimentally to be 1 pg/well. Given a measured well diameter of 0.7 cm, the surface area of nitrocellulose exposed to CAb during the coating process can be calculated, using the formula for area of a planar circle (A = IT?), to be 0.38 cm2. Assuming an adsorption efficiency of 85-100% (Anonymous, 1983; Palfree and Elliott, 1982), the concentration of protein adsorbed per cm’ was 2.2-2.6 kg. This binding capacity is approximately 15-17 times greater than that which was reported achievable using polystyrene as solid phase support (1 Fg16.5 cm’) (0.15 kg/cm2) (Contarero et al., 1980). The binding characteristics of the solid phase for the capture reagent directly influence the amount of antigen which can be bound and, ultimately, the test sensitivity. The fact that significantly more CAb was adsorbed to nitrocellulose Millititer wells than is typically possible using conventional polystyrene wells, undoubtedly contributed to the high test sensitivity achieved in our model. The lower limit of test sensitivity of the “sandwich” FELISA (10 fgiml) was 10,000 times greater than that of the “indirect” FELISA (100 pg/ml). The enhanced sensitivity of the “sandwich” method may be ascribed to an amplification of fluorescence signal occurring as a result of the use of an additional layer (CAb) on the solid phase. Enhanced sensitivity of “sandwich” over “indirect” methods

161

was, however, not observed in the CELISA: the lower limit of test sensitivity of both the “sandwich” and the “indirect” CELISA was 10 ng/ml. This observation could be accounted for if the lower limit of detection had already been reached in the “indirect” method at a concentration of 10 ng/ml of virus protein. The sensitivity of the “sandwich” FELISA was one million times greater than that of the “sandwich” CELISA; similarly, the sensitivity of the “indirect” FELISA was one hundred times greater than that of the “indirect” CELISA. The enhanced test sensitivity of the FELISA over the CELISA, whether by the “sandwich” or “indirect” method, may be attributed to the higher sensitivity of fluorogenic over chromogenic methods. In this study, the presence of specific antigen was indicated by reaction of alkaline phosphatase enzyme-labelled IAb with the fluorogenic substrate 4-MUP in the FELISA, or, the chromogenic substrate PNP in the CELISA. The sensitivity of immunoassays is affected by the turnover number or specific activity of the enzyme and the coefficient of molar extinction of the colored or fluorescent product (Clark and Engvall, 1985). Although PNP is hydrolyzed to p-nitrophenol by alkaline phosphatase at twice the rate of conversion of 4-MUP to 4-methylumbelliferone, the latter is detectable at a concentration of 10M9M compared with 10e5 M for PNP (Ishikawa and Kato, 1978). A number of other workers have reported enhanced sensitivity for the detection of certain macromolecular antigens by the use of fluorogenic substrates (Ishikawa and Kato, 1978; Yolken and Stopa, 1979; Shaliv et al., 1980; Yolken and Leister, 1982). In experiments designed to evaluate the specificity of FELISA for the NJ-La Sota strain of NDV, antisera to NJ-La Sota (homologous) strain cross-reacted with other (heterologous) strains of NDV (Bl and NJ-Roakin) and with other serotypes of the paramyxovirus group (parainfluenza 1, 2 and 3) but did not cross-react with influenza AIPRl8134. This finding was expected, since, although NDV shares common structural antigens with other members of the paramyxovirus group, it is antigenically and serologically distinct from members of the orthomyxovirus group (Rhodes and van Rooyen, 1968; Kingsbury, 1985). In addition, antisera to the NJLa Sota strain of NDV detected one HA unit of the homologous strain at a higher8 fluorescence count than it did a comparable concentration of several heterologous strains (Bl and NJ-Roakin) or other serotypes of paramyxovirus (parainfluenza 1, 2 and 3). That FELISA discriminated between the homologous strain of NDV and other paramyxoviruses is not surprising, since appropriate conventional serological tests (neutralization, HA1 and/or complement fixation) have similarly been used to distinguish members of the paramyxovirus group (Chanock and Parrott, 1965; Lancaster and Alexander, 1975). FELISA, however, appeared also to distinguish among the three strains of NDV. This observation is of particular interest since strains of NDV are not distinguishable by the usual conventional serological means, but rather are characterized by pathogenicity (mean death time in eggs) (Lancaster and Alexander, 1975). Evidence that FELISA was specific for the detection of NDV (Bl) in fecal samples collected from chickens undergoing vaccination with a combined vaccine containing NDV (Bl) and IBV, was provided by the finding that HA activity, which increased over time with increasing fluorescence count, was inhibited by antibody

162

specific to the Bl strain of NDV. That FELISA was specific for the Bl strain of NDV was further confirmed by competitive inhibition tests. In these experiments, anti-NDV immunoglobulin, prepared in an animal species different from that used to prepare the DAb, specifically blocked attachment of the DAb to the antigen, thereby inhibiting the fluorescence signal. Conversely, control sera, lacking specific antibodies, could not block attachment of the DAb to the antigen and fluorescence activity was maintained at a relatively constant level. NDV, in our studies, was used as a model for the development of a new immunoassay system with enhanced sensitivity capabilities. However, the developed assay may have diagnostic application in the agricultural industry. Chickens and other poultry excrete NDV in the feces following natural infection or vaccination (Hanson, 1973; Uterback and Schwartz, 1973); however, methods for detection of the virus in feces are inefficient, typically involving culture in embryonated eggs (Lancaster and Alexander, 1975). The FELISA may provide a rapid and sensitive alternative for identification of virus directly in fecal materials collected from clinically normal, inapparently infected birds, or, from birds exhibiting overt infection. A highly sensitive and specific fluorogenic enzyme immunoassay, utilizing nitrocellulose membrane as solid phase support, has been developed for the rapid detection and identification of virus in clinical samples. The technique can detect as little as 10 fg of purified NDV protein per ml of test sample (500 agitest well). This sensitivity is approximately two million times greater than that of conventional solid phase enzyme immunoassays which utilize polystyrene microtiter wells and a chromogenic substrate (R.E. Fulton, unpublished data). Plates may be used directly after sensitization or can be pre-sensitized and stored frozen until used (data not shown). Assay procedures are simple to perform and, once CAb has been adsorbed to the solid phase, may be completed within 4 h. Tests are performed in plates of 96-well format, hence a large number of samples may conveniently be assayed at one time. Readings, in the form of hard copy data, are obtained directly from a microprocessor-controlled microfluorometer, thus, operator bias is eliminated. Tests can easily be adapted to further automation for rapid processing of multiple samples with computer analysis of results.

Acknowledgements

The Defence Research Establishment Suffield acknowledges La Poulet Poultry Farm, Medicine Hat, Alta., for their cooperation and kind assistance in carrying out the clinical evaluation for this study. References Anonymous (1982) Immobilization of protein to type HA MillititerTM plates. In: Millititer Millipore Corp. Bedford, Ma. Berger, P.H., Thornbury, D.W. and Pirone, T.P. (1985) Detection of picogram quantities ruses using a dot blot immunobinding assay. J. Virol. Methods 12, 31-39.

Methods. of potyvi-

163 Bode, L., Beutin, L. and Kohler, H. (1984) Nitrocellulose enzyme-linked immunosorbent assay (NCELISA): a sensitive technique for the rapid visual detection of both viral antigens and antibodies. J. Viral. Methods 8, 111-121. Cantarero, L.A., Butler, J.E. and Osborne, J.W. (1980) The absorptive characteristics of proteins for polystyrene and their significance in solid-phase immunoassays. Anal. Biochem. 105, 375-382. Chanock, R.M. and Parrott, R.H. (1965) Para-influenza viruses. In: F.L. Horsfall and I. Tamm (Eds.), Viral and Rickettsial Infections of Man, 4th edit., pp. 741-752. Lippincott, Toronto. Chao, R.K., Fishaut, M., Schwartzman. J.D. and McIntosh, K. (1979) Detection of respiratory syncytial virus in nasal secretions from infants by enzyme-linked immunosorbent assay. J. Infect. Dis. 139, 483-486. Chiewsilp, D. and McCown, J.M. (1972) Elimination of repeated clot formation in mouse ascitic fluid containing arbovirus antibodies. Appl. Microbial. 24, 288-289. Clark, B.R. and Engvall, E. (1985) Enzyme-linked immunoassay (ELISA): theoretical and practical aspects. In: E.T. Maggio (Ed.), Enzyme-Immunoassay, pp. 167-179. CRC Press, Boca Raton. Davis, J.W., Angel, J.M. and Bowen, J.M. (1984) A quantitative immunobinding radioimmunoassay for antigens attached to nitrocellulose paper. J. Immunol. Methods 67, 271-278. Fernley, H.N. and Walker, P.G. (1965) Kinetic behaviour of calf-intestinal alkaline phosphatase with 4-methylumbelliferyl phosphate. Biochem. J. 97, 95-103. Gilman, A. (1970) A protein binding assay for adenosine 3’:5’-cyclic monophosphate. Proc. Natl. Acad. Sci. USA 67, 30.5-312. Grist. N.R., Ross, C.A. and Bell, E.J. (1974) Hemagglutination and hemagglutination inhibition tests. In: Diagnostic Methods in Clinical Virology, 2nd edit., pp. 103-111. Blackwell Scientific Publications, London. Hanson, R.P. (1973) Research support for eradication of velogenic viscerotropic Newcastle disease and other diseases. J. Am. Vet. Med. Assoc. 163, 1094-1096. Harmon, M.W. and Pawlik, K.M. (1982) Enzyme immunoassay for direct detection of influenza type A and adenovirus antigens in clinical specimens. J. Clin. Microbial. 15, 5-11. Hawkes, R., Niday, E. and Gordon, J. (1982) A dot-immunobinding assay for monoclonal and other antibodies. Analyt. Biochem. 119, 142-147. Hildreth, SW., Beaty, B.J., Meegan, J.M., Fazier, C.L. and Shope, R.E. (1982) Detection of La Crosse arbovirus antigen in mosquito pools: application of chromogenic and fluorogenic enzyme immunoassay systems. J. Clin. Microbial. 15, 879-884. Ho;ej$ V. and Hilgert, I. (1983) Nitrocellulose membrane as an antigen or antibody carrier for screening hybridoma cultures. J. Immunol. Methods 62, 325-329. Ishikawa, E. and Kato, K. (1978) Ultrasensitive enzyme immunoassay. Stand. J. Immunol. 8, 43-55. Kato, K.Y., Hamaguchi, Y., Fukui, H. and Ishikawa, E. (1976) Enzyme-linked immunoassay. Con-, jugation of rabbit anti-(human immunoglobulin G) antibody with beta-D-galactosidase from Escherichiu coli and its use for human immunoglobulin G assay. Eur. J. Biochem. 62, 285-292. Kingsbury, D.W. (1985) Orthomyxoand paramyxoviruses and their replication. In: B.N. Fields, D.M. Knipe, J.L. Melnick, R.M. Chanock, B. Roizman and R.E. Shope (Eds), Virology, pp. 1157-1178. Raven Press, New York. Kuno, H. and Kihara, H.K. (1967) Simple microassay of protein with membrane filter. Nature (London) 215, 974-975. Lancaster, J.E. and Alexander, D.J. (1975) Newcastle disease virus and spread. Monograph 11. Canada Dept. of Agriculture, Ottawa. Layne, E. (1957) Spectrophotometric and turbidimetric methods for measuring proteins. In: S.P. Colowich and N.O. Kaplan (Eds), Methods in Enzymology, pp. 447-454. Academic Press, New York. Lowry, O.H. and Oliver, H. (1948) A microphotofluorometer. J. Biol. Chem. 173, 677-682. Neurath, A.R. and Strick, N. (1981) Enzyme-linked fluorescence immunoassays using B-galactosidase and antibodies covalently bound to polystyrene plates. J. Virol. Methods 3, 155-165. Newman. M. and Wilson, T.H. (1980) Solubilization and reconstitution of the lactose transport system in Escherichio coli. J. Biol. Chem. 225, 10583-10586. Palfree, R.G.E. and Elliott, B.E. (1982) An enzyme-linked immunosorbent assay (ELISA) for detergent solublized Ia glycoproteins using nitrocellulose membrane discs. J. Immunol. Methods 52, 395-408.

164 Pronovost, A.D., Baumgarten, A. and Hsiung, G.D. (1981) Sensitive chemiluminescence enzyme-linked immunosorbent assay for quantitation of human immunoglobulin G and detection of herpes simplex virus. J. Clin. Microbial. 13, 97-101. Rhodes, A.J. and van Rooyen, C.E. (1968) Ribonucleic acid-containing viruses (riboviruses): myxoviruses, paramyxoviruses and other riboviruses. In: Textbook of Virology. 5th edit,, pp. 417-512. Williams and Wilkins Co., Baltimore. Russell, P.K., Chiewsilp, D. and Brandt, W.E. (1970) Immunoprecipitation analysis of soluble complement-fixing antigens of dengue viruses. J. Immunol. 105, 838845. Solonen, E.M. and Vaheri, A. (1979) Immobilization of viral and mycoplasmal antigens and of immunoglobulins on polystyrene surface for immunoassays. J. Immunol. Methods 30, 209-218. Shalev, A., Greenberg, A.H. and McAlpine, P.J. (1980) Detection of attograms of antigen by a high sensitivity enzyme-linked immunosorbent assay (HS-ELISA) using a fluorogenic substrate. J. Immunol. Methods 38, 125-139. Shekareki, I.C., Sever, J.L., Nerurkar, L. and Fuccillo, D. (1985) Comparison of enzyme-linked immunosorbent assay with enzyme-linked fluorescence assay with automated readers for detection of rubella virus antibody and herpes simplex virus. J. Clin. Microbial. 21, 92-96. Towbin, H., Staehalin, T. and Gordon. J. (1979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl. Acad. Sci. USA 76, 43504354. Utterback, W.W. and Schwartz, J.H. (1973) Epizootiology of velogenic viscerotropic Newcastle disease in Southern California, 1971-1973. J. Am. Vet. Med. Assoc. 163, 1080-1088. Volk, W.A., Snyder, R.M., Benjamin, D.C. and Wagner, R.R. (1982) Monoclonal antibodies to the glycoprotein of vesicular stomatitis virus: comparative neutralizing activity. J. Viral. 42, 220-227. Voller, A., Bidwell, D. and Bartlett, A. (1980) Enzyme-linked immunosorbent assay. In: N.R. Rose and A. Friedman (Eds), Manual of Clinical Immunology, 2nd edit. pp. 359-371. American Society Microbiology, Washington, D.C. Yolken, R.H. and Stopa, P.J. (1979) Enzyme-linked fluorescence assay: ultrasensitive solid-phase assay for detection of human rotavirus. J. Clin. Microbial. 10, 317-321. Yolken, R.H. (1982) Enzyme immunoassays for the detection of infectious antigens in body fluids: current limitations and future prospects. Rev. Infect. Dis. 4, 35-68. Yolken, R.H. and Leister, F.J. (1982) Comparison of fluorescent and colorigenic substrates for enzyme immunoassays. J. Clin. Microbial. 15. 757-760.