Journal of Virological Methods, 3 (1981) 155-165

Elsevier/North-Holland

ENZYME-LINKED

155

Biomedical Press

FLUORESCENCE

AND ANTIBODIES COVALENTLY

IMMUNOASSAYS

USING &GALACTOSIDASE

BOUND TO POLYSTYRENE

PLATES

A. ROBERT NEURATH and NATHAN STRICK

The Lindsley

F. Kimball Research Institute of the New York Blood Center, New York, NY 10021,

U.S.A.

(Accepted 15 May 1981)

A solid-phase enzyme-linked immunoassay using a fluorogenic substrate (4-methylumbelliferylP-D-gahactopyranoside) was developed. Antibodies were covalently linked to glutaraldehyde-activated
96-well aminopolystyrene plates. Antigens from test samples were adsorbed to the solid phase and
detected using antibodies conjugated with E. coli pgalactosidase. Glutaraldehyde, N-succinimidylor N-succinimidyl-6-(4-azido-2-nitrophenylamino)-hexanoate
were
3-(2-pyridyldithio)-propionate
used as linkers between antibodies and the enzyme. The measurement of fluorescence can be automated for rapid screening of many specimens. The sensitivity limit of the test for HBsAg is about
S-10 pg.

INTRODUCTION

Quantitative enzyme-labeled
immunoassays
(ELlSA), originally described by Engvall
and Perlmann (1971) and by Van Weemen and Schuurs (1971), are being used with
increasing frequency in clinical diagnosis and biomedical research. The advantages of
ELISA over radioimmunoassays,
i.e. stability of reagents, lack of potential health hazards
and need for less expensive equipment, are occasionally outweighed by the lower sensitivity for some antigen-antibody
systems. However, recent reports indicate that the use
of fluorogenic substrates may lead to an increase in sensitivity of ELISA tests. The detection limits of the corresponding enzyme-linked
fluorescence immunoassays (ELFA) may
surpass the limits of BIA tests (Kato et al., 1976; Ishikawa and Kato, 1978; Yolken and
Stopa, 1979; Yolken et al., 1980).
One of the limitations of solid-phase ELISA or RIA tests is the elution during each
step of the immunoassay procedure of antibodies or antigens originally adsorbed to the
solid support (Herrmann et al., 1979). The resulting decrease in sensitivity of immunoassays can be overcome by covalent linking of antibodies or antigens to the solid phase.
We describe here an immunoassay into which both features required for high sensitivity -- the use of a fluorogenic substrate and the covalent linking of antibodies to the
solid phase - have been incorporated. The application of the test to antigens specifically
associated with some forms of viral hepatitis in humans or in experimental animals is also
described.
016660934/81/0000-0000/$02.50

@ Elsevier/North-Holland

Biomedical Press

156

MATERIALS

AND METHODS

Antigens and antibodies

Sera containing

hepatitis

B surface antigen (HBsAg) and hepatitis B e antigen (HBeAg)

were obtained from humans chronically infected with hepatitis B virus. HBsAg was purified as described (Neurath et al., 1978). Sera of multiply transfused individuals and
hyperimmune sera prepared by immunization
of rabbits and goats with HBsAg served as
sources for preparation of anti-HBs. The corresponding goat serum was a gift from North
American Biologicals, Inc., Miami, FL.
IgG was prepared from the sera by chromatography
on DEAE-cellulose (Neurath,
1981). Affinity chromatography
on columns of HBsAg linked to Sepharose 4B was used
for immunochemical
purification
of anti-HBs. The methods described earlier (Prince et
al., 1975) were followed, except that the anti-HBs was eluted from the columns by 3
M NaSCN. The eluted antibodies were extensively dialyzed against the appropriate
buffer used for conjugation with /J-galactosidase.
Anti-HBe IgG was isolated from sera of human HBsAg carriers by chromatography
on DEAE-cellulose.
Woodchuck serum containing woodchuck hepatitis surface antigen (WHsAg) and goat
anti-WI-Is were obtained from Dr. J. Summers (Cancer Research Institute, Fox Chase,
Philadelphia, PA) and Dr. G. Tyler (Penrose Laboratory, Philadelphia Zoological Gardens).
IgG from sera containing anti-HBe and anti-WHs, respectively, was derivatized with
2,4dinitrophenyl
groups (DNP) as described (Neurath, 1981). Immunochemically
purified rabbit anti-DNP was prepared as reported (Neurath, 1981).
Covalent linking of antibodies to polystyrene microtiter plates
Several methods

of binding

antibodies

to solid supports

were compared.

These

methods are referred to in the Results section. Only the optimal procedure finally
adopted will be described here.
The wells of 96-well polystyrene tissue culture plates (Catalogue No. 3042, Falcon
Products, Oxnard, CA) were placed into a desiccator in a well ventilated hood and filled
with methanesulfonic
acid. After overnight incubation at 20C the wells were emptied,
extensively washed with distilled water, and dried. The plates were placed into a heating
block (40C) and tilled with a mixture of glacial acetic acid and fuming nitric acid
(1 : 1). After 4 h, the nitrated plates were extensively washed with water until the pH of
the wash was 6 or higher. The plates were put into the heating block (50C) and ftlled
with 0.5% sodium dithionite in 0.5 M NaOH to convert polynitrostyrene
into polyaminostyrene. After 1 h, the plates were extensively washed with HzO, filled with 1% glutaraldehyde in 0.05 M phosphate, pH 8.5, and left standing for 2 h at 2OC, followed by
- 16 h at 4C. The plates were washed and each well was filled with 300 4 of a solution

157

containing

IgG isolated

from antisera by DEAE-cellulose

ml in 0.05 M phosphate,
HBsAg testing.

pH 8.5). Human

After overnight

with 1 M glycine-0.1

incubation

chromatography

IgG was used routinely

at 4C

(100 pg IgG/

to coat plates for

the plates were washed and treated

M borate, pH 8.5, for 4 h at 20C. Finally, the plates were rinsed

with Tris-buffered saline (0.14 NaCl, 0.01 M Tris, 0.02% NaN3, pH 7.2; TS). For storage,
the wells were filled with TS and the plates were covered with covers supplied with the
plates. The coating solution containing IgG was not reused, since most of the IgG (- 10
pg/well) was linked to the derivatized polystyrene.
Conjugation of antibodies with &galactosidase

Immunochemically
purified antibodies (rabbit anti-HBs (subtype ad) and rabbit antiDNP) and Pgalactosidase (Boehringer Mannheim Biochemicals, Indianapolis, IN), extensively dialyzed against the buffer used for conjugation, were employed routinely. The
followmg reagents were selected for linking antibodies with the enzyme:
GlutaraMehyde. A solution containing 2 mg/ml each of enzyme and IgG in 0.1 M phosphate, pH 6.8, was mixed with glutaraldehyde
(final concentration
0.05%) for 3 h at
20C and then dialyzed overnight at 4C against the same buffer. A solution of 1 M

lysine (PH 8) was added to a final concentration

was chromatographed on Sepharose 6B CL.

of 0.1 M. Thirty min later, the mixture

N-succinimidyE6-(4-azido-2-niaophenyEamino)-hexanoate

(SANAH).
One mg of IgG
in 0.8 ml of 0.05 M phosphate, pH 8, was mixed with 30 ~1 of a solution of SANAH
(Pierce Chemical Company, Rockford, IL; 2 mg/ml in dimethylsulfoxide.
The mixture
was kept for 1 h at 23C protected from light, mixed with 100 d of 0.1 M lysine, pH 8,

and chromatographed
on a 0.7 X 20 cm column of Sephadex G-25 in darkness. The
eluant was 0.05 M phosphate, pH 8. The fractions containing protein were pooled, mixed
with 1 mg of Pgalactosidase and illuminated for 16 h with a 500 W lamp (Photo-Ect,
Sylvania, Danvers, MA), placed at a distance of 50 cm from the surface of the reaction
mixture, and cooled -with a fan.
N-succinimidyl-3-(2-pyridyldithio)-propionate

(SPDP). Five mg of IgG in - 1 ml of 0.1

M phosphate-O.1
M NaCl, pH 7.5 (PBS) were mixed with 25 ~1 of a solution of SPDP
(10 mg/ml in ethanol, Pharmacia Fine Chemicals AB, Uppsala, Sweden). After 30 min at
room temperature,
the mixture was chromatographed
on a 0.7 X 20 cm column of
Sephadex G-25 in 0.1 M acetate-O.1 M NaCl, pH 4.5. Fractions containing protein were
pooled and solid dithiothreitol
was added (7.7 mg/ml). After 20 min at room temperature, the mixture was chromatographed
on a 0.7 X 20 cm column of Sephadex G-25 in
PBS. Fractions containing the major portion of protein were pooled, mixed with 3 mg
of Pgalactosidase
at room temperature
overnight, and stored at 4C until chromatography on Sepharose 6B CL.

158

Performance of ELFA
Samples (200 4) containing

purified

HBsAg or other antigens in non-purified

form

were added to wells of the derivatized polystyrene plates. Normal human serum was used
as diluent and for control blanks. After standing overnight at room temperature, the
samples were removed by suction and the plates were washed with TS. For HBsAg tests, the
anti-HBs-/3-galactosidase conjugate was added (200 j.d of a 1 : 50 to 1 : 400 dilution of the
conjugate (eluted from Sepharose 6B CL columns) (Fig. 3) in TS containing 50 mg/ml
of bovine serum, ry 1 mg of normal human serum proteins/ml, 100 units of Trasylol/ml
and 1 mg of Tween 2O/ml). After 1 h at 45C, the plates were washed with TS, 200 I_tl
of the substrate (obtained from Sigma Chemical Company, St. Louis, MO) solution were
added and the enzyme attached to the plates was determined under conditions given for
Fig. 2. The fluorescence was determined from the following formula:
fluorescence

= 100 X

fluorometer

digital readout

fluorometer

sensitivity

range *

Fluorescence was measured in the Spectra/&o fluorometer (Gilson Medical Electronics,

Inc., Middleton, WI) equipped with filters used for determination of o-phtalaldehyde.
For determination
of HBeAg or WHsAg, plates after overnight incubation with the
specimens were washed and incubated with DNP-anti-HBe and DNP-anti-WHs, respectively, under conditions described before (Neurath and Strick, 1979; Neurath, 1981).
The plates were washed and incubated for 2 h at 37C with 200 fi of a 1 : 50 dilution of
an anti-DNP-/$galactosidase

conjugate.

Otherwise, the conditions

for ELFA were the same

as described for HBsAg.

Other methods. The activity of alkaline phosphatase
and of horseradish peroxidase (grade I, Boehringer
determined

spectrophotometric~y

diamine as substrates, respectively

using

(Type VII, Sigma, St. Louis, MO)

Mannheim, Indianapolis, IN) was

p-nitrophenyl

phosphate

and o-phenylene-

(Engvall, 1980).

RESULTSANDDISCUSSION
~eoretical sensitivity ~~rnitof ELFA
/3&galactopyranoside as substrate

using ~-galactosidase artd 4~met~yl~mbe~life~l-

One of the prerequisites for a sensitive ELFA is the high fluorescence yield of the
product of the enzymatic reaction and the negligible spontaneous degradation of the
substrate resulting in a low reagent blank. The substrate, ~me~ylurn~~feryl-~Dgalactopyranoside,
and the product of its hydrolysis, 4-~~ylumbe~ferone,
fulfill
both of these requirements.
Reagent blanks obtained with this substrate are approximately 50-fold lower than with 4-methylumbelliferyl
phosphate (Ishikawa and Kato,
1978; own unpublished data), the substrate for alkaline phosphatase, the enzyme most

159

frequently used in ELFA (Yolken and Stopa, 1979). The relationship between fluorescence and the level of 4-methylumbelliferone
is linear within a wide concentration
range
(Fig. 1). The sensitivity limit for Cmethylumbelliferone
detection is approximately
30 pg. In comparison, the sensitivity limit for detection of o-(p)nitrophenol,
the product
measured in ELBA tests with pgalactosidase,

is - 100 ng.

Another prerequisite for a sensitive ELFA is a high turnover number of the enzyme
linked to antibodies (antigens). o-n-Galactosidase
from E. coli meets this requirement
(Ishikawa and Kato, 1978). The sensitivity limit for the detection of /3-galactosidase
under conditions used by us was 0.2 pg, corresponding to 3.7 X lo-l9 mol of enzyme
(Fig. 2). In comparison, the limits for spectrophotometric determinations
of Bgalactosidase, alkaline phosphatase and horseradish peroxidase - the enzymes most frequently
used in ELISA (Avrameas et al., 1978) - were 3.5 X 10-16, 2 X lo-l6 and lo-l4 mol of
enzyme, respectively. Assuming that other conditions necessary for an ultrasensitive
ELFA have been fulffled (availability of purified antibodies with high avidity, optimal
conjugation
with the enzyme), the detection limit for antigens should have a similar
order of magnitude.
Covalent linking of antibodies to the solid support used in ELFA
In preliminary

experiments,

several methods

for the immobilization

of antibodies

IOZ-

8
z IO -

s
i

f5
3 tooL

10-l-

IO-Z0

I02
I03
4-METHYLUMBELLIFERONE

Fig. 1. Relationship between fluorescence

0.01 M phosphate, pH 7.0.

104

IO'

(PC+)

and the quantity

of 4-methylumbelliferone

in 200 ~1 of

160

to-2

to-1

too

10

I02

IO3

to4

/3-GALACTOSIDASE

105

to6

to

(pg)

Fig. 2. Calibration curve for determination of pgalactosidase. The diluent was 0.01 M phosphate,
pH 7.0. The initial concentration of substrate was 5 X IO- M. 4-Me~ylum~l~~erone
released within
30 min at 37C was measured in a total volume of 200 &

on solid supports were evaluated: binding to partly hydrolyzed nylon balls using carbodiimides or glutaraldehyde
(Hendry and Hermann, 1980); binding to diazotized polyaminostyrene
(Phillips et al., 1980); and binding to albumin-coated
beads using glutaraldehyde as cross-linker (Eshikawa and Kato, 1978). None of these methods offered any
advantages over non-covalent
adsorption of antibodies to either polystyrene beads or
polyvinyl plates (Catt and Tregear, 1967), when HBsAg was used as test antigen (data
not shown). The best results were obtained

with polystyrene

plates activated by a proce-

dure originally developed by Reimer et al. (1978) for latex spheres. The method was
further improved by pretreatment
of the plates with methanesulfonic
acid leading to
the eli~~ation
of all functional groups originally present in polystyrene and contributing to non-specific adsorption of proteins (Dr. 8. Kent, personal communication).
In order to demonstrate
that polystyrene activated by this method binds proteins
irreversibly, plates were coated with gelatin and radiolabeled hemoglobin, added as a
marker, and then exposed for 1 h at 37C to 1 M carbonate, pH 10.9, sodium dodecyl
sulfate (1 mg/ml) expected to elute non-cov~en~y
bound proteins. Only 0.7% of the
radiolabeled protein bound originally to polyaminostyrene
was ehrted, suggesting that the
major portion of protein was covalently linked to the plates. On the other hand, similar
treatment of polystyrene plates to which radiolabeled hemoglobin was adsorbed noncovalently resulted in the release of 75% of the labeled protein initially adsorbed.

161

The covalent linking of anti-HBs to polyaminostyrene

resulted in a 32-fold increase
in sensitivity for the detection of HBsAg as compared with an ELFA in which polystyrene plates with non-covalently

adsorbed anti-HBs were utilized.

Selection of methods for covalent linking of &galactosidase with antibodies

The one-step glutaraldehyde procedure, originally described by Avrameas (1979) was

used for initial attempts to bind antibodies (anti-HBs) to P-galactosidase following essentially the protocol reported by Cameron and Erlanger (1976). The recovery of enzymatic
activity after cross-linking was lower as compared with methods tested subsequently
(Fig. 3A-C). The separation of components
differing in size by molecular exclusion
chromatography
and the utilization in ELFA exclusively of conjugates eluted immediately behind the fractions corresponding
to the void volume of the Sepharose 6B CL
column were essential. The unfractionated
conjugate apparently contained IgG monomers and/or hopolymers which had an inhibitory effect in ELFA. The need to separate
the enzyme-antibody
conjugate from other components applied also to methods oflinking
antibodies to /I-galactosidase which are described below. Conjugation with glutaraldehyde
resulted also in the formation of high molecular weight aggregates which had to be
removed by centrifugation
before column chromatography.
Despite these drawbacks,
the enzyme-labeled
antibody could be utilized in ELFA which had the same sensitivity
limit for HBsAg detection as a commercial RIA (- 940 pg/ml; Ausria II, Abbott I-aboratories, North Chicago, IL).
In order to improve the linking of antibodies to P_galactosidase, the application of two
step cross-linking methods, expected to minimize the formation of hopolymers, was
explored. Conjugation
with either NJ-o-phenylenedimaleimide
(Kato et al., 1976;
Ishikawa and Kato, 1978), which involves reduction of disulfide bonds in antibodies,
or N-succinimidyl-6-(4-azido-2-nitrophenylamino)hexanoate,
which involves photochemically induced cross-linking,
failed to yield products better than the conjugate
obtained by linking with glutaraldehyde. The photochemical cross-linking appeared quite
inefficient (Fig. 3B). It results in an enzyme-antibody
of as little as 150 pg of HBsAg. The disadvantages
have been anticipated by Avrameas et al. (1978).
The best results were obtained

conjugate permitting the detection

of dimaleimides as coupling reagents

with the heterobifunctional

reagent N-succinimidyl3-

(2-pyridyldithio)propionate
(SPDP), originally synthesized by Carlsson et al. (1978)
which has become commercially available recently. All further work was carried out with
SPDP. Anti-HBs+galactosidase
conjugates prepared in this way were well suited for
ELFA tests, allowing the detection of as little as 5-10 pg of HBsAg (Fig. 4). Additional
tests were developed for hepatitis B e antigen (HBeAg) and woodchuck hepatitis surface
antigen (WHsAg) (Fig. 5). Anti-HBe and anti-WHs used in these tests were not immunochemically purified and corresponded to IgG prepared from serum by chromatography
on DEAE-cellulose. The sensitivity of ELFA for HBeAg was similar to that reported for
the corresponding RIA test (Neurath et al., 1979).

162

FRACTION

Fig.

3. Separation

molecular
fraction
protein
dase

of anti-HBs-pgalactosidase

exclusion
(2-4
after

NUMBER

chromatography

~1) were

tested

chromatography

conjugates

were

prepared

for @-galactosidase.
of purified
by the

succinimidyl-3-(2-pyridyldithio>propionate
amino)-hexanoate
fractions

(B). The eluant

were collected.

conjugates

on columns

from

Arrow

indicates

free IgG in a separate

one-step
(A),

free @-galactosidase

(1.5 X 30 cm) of Sepharose

glutaraldehyde
or

with

fraction

experiment.
procedure

and anti-HBs

by

6B CL. Aliquots

of each

with the highest

level of

The anti-HBs-pgalactosi(C), by linking

with N-

N-succinimidyl-6-(4-azido-2nitrophenyl-

was 0.14 M NaCl 0.01 M phosphate,

0.02% NaN,,

pH 7. One ml

20

60

HBsAg

100

diluent

anti-HBe

B e antigen

against

(DNP)

blanks

Serial two-fold

160

dilutions

( -)

against

corresponding

human

followed
to normal

respectively,

surface
serum

diluent

diluted

20-fold

A 400to normal

in Tris-buffered

anti-DNP

saline (pH

antibodies.

were quantitated

corresponding

serum were tested.

The antigens

blanks

with p-galactosidase-conjugated

(WHsAg) ( 04).

against

and sheep serum

by incubation

antigen

was measured

human

SERUM DllUTlON
(200 @I) of HBsAg in normal
was 6.
hepatitis

solution)

and woodchuck

the substrate

was used in the test. Fluorescence

and antiO-WHs,

(HBeAg)

140

of HBsAg.

120

conjugate

quantities

(Pg)

of the blank (measured

7.2) for HBeAg and WHsAg, respectively.

was measured

2,4-dinitrophenylated

of hepatitis

serum. The fluorescence

Fluorescence

using

60

curve for subnanogram

40

of a P-galactosidase-anti-HBs

Fig. 5. Determination

human

fold dilution

Fig. 4. Calibration

25

164

Although

the ELFA for HBsAg appeared about two orders of magnitude

tive than the corresponding

RIA test, the theoretical

sensitivity

more sensi-

limit expected

on the

basis of fluorometric measurements of /.%galactosidase activity (Fig. 2) was not achieved.

The reason for this was the high background fluorescence obtained with HBsAg-negative
samples.

We devoted

considerable

efforts

to overcome

this problem.

These efforts in-

cluded: changes in the composition of the diluent for the enzyme-antibody

conjugates;
using F(ab)* fragments instead of IgG to either coat the plates or prepare the conjugate;
and passing the conjugated antibodies through a column of insolubilized IgG. None of
these alterations resulted in an improved assay.
Our results demonstrate the universal potential of ELFA for ultrasensitive immunoassays, but at the same time indicate that novel approaches will be needed to accomplish
assay sensitivities

predicted by theoretical

considerations.

ACKNOWLEDGEMENTS

This study was supported by grants HL 09011-17 from The National Heart, Lung and
Blood Institute, and C 172100 from the New York State Health Research Council. We
thank Dr. J. Summers and Dr. G. Tyler for sera containing woodchuck hepatitis surface
antigen and the corresponding antibodies, respectively.

REFERENCES
Avrameas,

S., 1969, Immunochemistry

Avrameas,

S., T. Ternynck

Cameron,

D.J. and B.F. Erlanger,

Carlsson,
Catt,

Engvall,

E., 1980,

E. and P. Perlmann,

R.M. and J.E. Herrmann,

Kato,

E. and K. Kato,
A.R.,

and M.F. Collins,

1978, Stand.

in: Methods

and J.J. Langone

(Academic

A.R. and N. Strick,

Neurath,

A.R., A.M. Prince

Neurath,

A.R., N. Strick,

Phillips,

D.J., C.B. Reimer,

A.M., D. Chicot,

8,87 1.
Methods

J. Immunol.

in Enzymology,

8, Suppl.

10, 210.

7,43.

1976, J. Immunol.

116, 1554.

Vol. 70, Part B, eds. H. van Vunakis

and J.J. Lan-

(in press).

1979, Mol. Immunol.

and J. Giacalone,

W. Szmuness,

16, 625.

1978, Experientia

C.E. Stevens

J. Vnek,

34,414.

and E.J. Harley,

T.W. Wells and C.M. Black,

R. Wright,

35, 285.

1979, J. Clin. MicrobioL

and E. Ishikawa,

Press, New York)

Neurath,

Prince,

70A, eds. H. van Vunakis

1980, J. Immunol.

Y. Hamaguchi

1981,

gone (Academic

J. 173, 723.

1971, Immunochemistry

J.E., R.M. Hendry

K., H. Fukui,

8, SuppL 7, 7.

158, 1570.

in Enzymology,

Hendry,

Neurath,

J. Immunol.

116, 1313.

p. 4 19.

Engvall,
Herrmann,

1978, Biochem.

1967, Science

in: Methods

Press, New York)

1978, Stand.

1976, J. Immunol.

J., H. Drevin and R. Axen,

K. and G.W. Tregear,

Ishikawa,

5, 43.

and J.-L. Guesdon,

1979, J. Gen. Viral. 42, 493.

1980, J. Immunol.

A.R. Neurath,

A. Lippin,

Methods

34, 315.

S. Swiss and C. Ehrich,

1975,

VOX Sang. 29,319.

Relmer,

C.B.,

Staining

D.J. Phillips,

Techniques,

Press, Amsterdam)

CM.

Black

eds. W. Knapp,
p. 189.

and T.W. Wells, 1978,

K. Holubar

in: Immunofluorescence

and G. Wick (Elsevier/North-Holland

and Related
Biomedical

165

Van Weemen,

B.K. and A.W.H.M.

Yolken,

R.H. and P.J. Stopa,

Yolken,

R.H., V.M. Torsch,

Schuurs,

1971, FEBS Lett.

1979, J. Clin. Microbial.

R. Berg, B.R. Murphy

15,232.

10, 317.

and Y.C. Lee, 1980, J. Infect.

Dis. 142, 516.