CLIN. CHEM.37/3, 411-414(1991)
Use of Chicken Antibodies in Enzyme Immunoassays to Avoid Interference by Rheumatoid
Factors
Anders
Larseon,Alex Karlsson-Parra,1andJohn SjOqulst
Rheumatoid factor (RF) is a major source of interference
in many immunoassays. Most immunoassays use mammalian polyclonal or monoclonal antibodies, and RF can
react with lgG from mammalian species, thus causing
false-positive results. In this work we have studied RF
interference in a sandwich ELISA, where RF in the sample
may react with both the capture antibody and the detection antibody to give a false-positive reaction. We show
that rheumatoid factors do not react with chicken IgY; if
the capture antibody or detection antibody (or both) is of
avian origin, the interference of RF or other anti-lgG
antibodies in sandwich ELISA can be avoided.
AddItional Keyphrases: analytical error
immunoglobulins
avian vs mammalian antibodies
Various types of enzyme-linked immunosorbent assays (ELISA8) are widely used, one of the most common
being sandwich ELISA5, where one antibody is used to
capture the antigen and another, labeled, antibody is
used to detect the bound antigen (1).2 Both of these
antibodies are usually of mammalian origin.
Rheumatoid factor (RF) is an autoantibody that reacts
with the Fc portion of mammalian IgG (2). The disease
usually associated with RF is rheumatoid arthritis
(RA), but RF is also present in serum from patients with
many other diseases (3,4). The classical tests for RF are
agglutination
tests, in which the RF reacts with two (or
more if it is an IgM-RF) different mammalian IgG
molecules, thus causing agglutination. The same reaction will give a false-positive reaction in a sandwich
ELISA, owing to the binding of detection antibody to
capture antibody in the absence of the antigen that the
assay was designed to detect. More recently, ELISA8 have
been developed to detect RF. Whereas, the agglutination assays mainly detect IgM-RF, an ELISA can detect
different RF isotypes depending on the specificity of the
detection antibody. Fab fragments of mammalian antibodies (5) or chicken antibodies (6), used for detection to
avoid false-positive reactions, gave no reaction between
chicken IgG and RF because of the immunological
difference
between mammalian IgG and chicken IgG
Here we show that if either or both of the antibodies
(7).
Department
of Medical and Physiological Chemistry, Biomedical Center, Box 575, S-751 23 Uppsala, Sweden.
1Department of Clinical Bacteriology, University Hospital,
S-751 85 Uppsala, Sweden.
2NonsdjJ
abbreviations: ELISA, enzyme-linked immunosorbent assay; CRP, C-reactive protein; PBS, phosphate-buffered
saline; HA, rheumatoid arthritis; RF, rheumatoid factor(s); and
ALP, alkaline phosphatase.
Received August 24, 1990; accepted December 29, 1990.
in a sandwich EUSA is of avian origin,
from RF or other anti-mammalian
false reactions
IgG antibodies in the
samples can be avoided.
MaterIals and Methods
Antisera.
Gamma-globulin Kabi, 165 g/L for intramuscular use, was obtained from Kabi, Stockholm,
Sweden. Rabbit IgG was purified from rabbit serum by
affinity chromatography
on Protein A-Sepharose.
Chicken IgG, rabbit anti-chicken IgG, chicken antirabbit IgG, affinity-purified chicken anti-human C-reactive protein (CRP), and chicken anti-human IgG were
provided by Immunsystem IMS AB, Uppsala, Sweden.
Chicken IgG, human IgG, and rabbit IgG were conjugated with alkaline phosphatase (ALP, EC 3.1.3.1; Genzyme Diagnostics, Maidstone, U.K.) by a one-step glutaraldehydecouplingprocedure (8).
Affinity-purified
chicken anti-human CRP was biotinylated
with
N-hydroxysuccinimidobiotin
(Sigma
Chemical Co., St. Louis, MO) (9).
Serum samples. Sera from 31 patients with positive
latex-agglutination test results were selected from samples routinely
analyzed
at the Department
of Clinical
Bacteriology, University Hospital, Uppsala. Sera from
20 healthy blood donors were used as controls. Sera from
90 consecutive patients were analyzed for CRP with a
Hitachi 717 automated analyzer at the Department of
Clinical Chemistry, University
Hospital, Uppsala.
These sera were also analyzed by a CRP ELISA.
ELISA procedure. Flat-bottom MicroELISA plates (Dynatech M 129 A) were coated with 200 L of a 0.1 g/L
solution of human IgG or rabbit IgG in NaHCO3, 0.1
moIJL, pH 9.5, or with chicken IgG, 0.1 g/L in PBS
(phosphate-buffered saline: per liter, 20 mmol of
NaH2PO4, 0.15 mmol of NaCl, and 0.2 g of NaN3, pH
7.2), for 2 h at 37 #{176}C
or overnight at 4 #{176}C.
We then added
to the wells in duplicate 200 L of the samples at
various dilutions in PBS. After incubating the plates for
2 h at 37#{176}C,
we washed them three times with NaClTween (per liter: 9 g of NaCI, 0.5 mL of Tween 20, and
0.2 g of NaN3) in a Wellwash 4 washing machine
(Denley Instruments Ltd., Billingshurst,
U.K.). We
added to each well 200 L of the ALP conjugate (diluted
1000-fold), incubated the plates at 37 #{176}C
for 2 h, and
washed them with NaCI-Tween as above. Finally, we
pipetted into the wells 200 L of p-nitrophenyl phosphate, 1 gIL, in a solution of 1 mol of diethanolamine
and 0.5 mmol of MgC12 per liter, pH 9.8, and incubated
the plates for 30 nun at room temperature in darkness.
The reaction was stopped with 50 L of 5 molIL NaOH
reagent, and the absorbance of the plates was read in a
Titertek
Multiskan (Flow Laboratories Ltd., Irvine,
CLINICALCHEMISTRY, Vol. 37, No. 3, 1991 411
(A)
In these experiments an absorbance
of
0.05 at 405 nm was considered to be the lower limit of a
positive response.
For statistical analysis of the data we used Statgraphice (STSC, Inc., Rockville, MD).
Scotland).
(mean 0.002 A) or when chicken IgG-ALP was used as
detection antibody (mean 0.000 A) (Figure 2).
Tests with human IgG-coated plates and human IgGALP also gave a positive response, A = 0.480 (SE 0.086;
p <0.0000007). The combinations of human IgG-coated
plates with rabbit IgG-ALP and rabbit IgG-coated
Results
plates with
Effects of anti-IgG antibodies
on the ELISA. Anti-IgG
antibodies were added in fivefold dilutions to the plates
coated with chicken IgG, human IgG, or rabbit IgG.
Conjugates of nonspecific chicken IgG-ALP, human IgGALP, or rabbit IgG-ALP were used as detection antibodies. The anti-human IgG antibodies gave a positive
response with mammalian IgG-coated plates and rabbit
IgG-ALP but no reaction if either or both of the antibodies was chicken IgG (Figure 1). The same results
were obtained with human IgG-ALP (results not
shown). if we used a rabbit anti-chicken IgG antibody as
a control, the anti-chicken IgG antibodies gave a positive reaction when both the antibodies were chicken
IgG, but no reaction was observed if one or both of the
antibodies was of mammalian
origin (results not
shown).
Effects of RF-positive sera on the ELJSA. We added 10
,uL of RF-positive or -negative serum and 190 j.L of PBS
(i.e., 20-fold-diluted
samples) to plates coated with
chicken IgG, human IgG, or rabbit IgG. Conjugates of
nonspecific chicken IgG-ALP, human IgG-ALP, or rabbit IgG-ALP were the detection antibodies. We found a
significantly greater reaction with RF-positive sera [A
=
0.158 ± 0.033 (mean ± SE), P <0.00002] in plates
coated with rabbit IgG and with rabbit IgG-ALP as the
detection antibody than when chicken IgG-ALP was
used. Of the RF-positive sera, 70% gave a positive
response with ELISA plates coated with rabbit IgG and
with rabbit IgG-ALP as the detection antibody. No
response was observed with the RF-negative serum
increased reactions-0.080
A (SE 0.020,P <0.0003) and
0.154 A (SE 0.028, P <0.000002)-when
compared with
rabbit IgG-coated plates and chicken IgG-ALP. No reaction was seen if either or both of the antibodies were
chicken IgG.
CRP ELISA. In this assay, we added the sera to the
human IgG-ALP
also gave significantly
wells in serial fivefold dilutions. All sera and a control
were run in duplicate. The CRP ELISA was performed as
described above except that biotinylated anti-CRP (diluted 1000-fold) was used instead of the ALP conjugate.
Vectastain
ABC-ALP (Vector Laboratories, Inc.), diluted 1000-fold, was used to detect bound biotinylated
antibody. Incubation and washing were performed as
described in Materials and Methods.
We compared the results with those obtained with the
Hitachi 717 automated analyzer. Because the Hitachi
analyzer expressed all values between 0 and 10 mg/L as
<10 mg/L, we assigned, for the statistical analysis, the
value 5 mg/L to CRP results <10 mg/L in both analyses.
The results showed a good correlation
(r = 0.97).
Discussion
Waaler (10) and Rose et al. (11) first demonstrated
that rabbit antibody-sensitized
erythrocytes
from sheep
were agglutinated by most sera from subjects with HA.
These sera contain anti-immunoglobulins,
termed RF,
which can be of the IgA, IgG, or 1gM classes. Between
60% and 80% of patients with RA have RF activity in
their serum. HF have also been found in serum from
patients with other connective tissue diseases and in
many other diseases. Although RF react with the Fcpart of IgG, they display heterogeneity
in their reactivity to IgG of different species. This reactivity to IgG of
other species may produce false reactions in assays
involving antibodies.
A widely used approach for detecting antigens is to
E
C
‘I,
0
.
CD
CD
C.)
C
CD
.0
create a so-called antibody
“sandwich”:
immobilized
..u.
4
0
Ci,
E
.0
C
4
In
0
1:3,125
1:625
1:125
Dilution of antiserum
Fig. 1. Sandwich ELISA with plates coated with IgG from various
sourcesand samplescontaining anti-human 190 or anti-chicken IgG
in various concentrations
A rabbit IgG-alkaline phosphatase conjugatewas used as detectionantibody.
The 0 point on the abscissa denotes that no antibody was added.
0, plates coated with human igG and anti-chicken 190 added;
0, plates coated with human lgG and anti-human 190added;
0, plates coated with rabbit lgG and anti-chicken IgG added;
U, plates coated with rabbit 190 and anti-human lgG added;
& plates coated with chicken 190 and anti-chicken lgG added;
A, plates coated with chicken IgG and anti-human IgG added
412 CLINICAL CHEMISTRY, Vol. 37, No. 3, 1991
CD
CD 0.5
0
C
CD
.0
0
CD
.0
4
0
#{149}:
#{149}
#{149}-.#{149}
I
RFpos.
RF neg
RF pos. RFnPn
Fig. 2. Sandwich ELISAwith plates coated with rabbit igG
We added 10 pL of RF-positiveor -negative serum to the plates.Nonspecific
rabbit 190-alkaline phosphatase conjugateor chicken lgG-alkaline phosphatase conjugate was used as detection antibody
capture antibody, antigen, and directly or indirectly
labeled detection antibody. The samples are often serum
or plasma, and one of the problems in analyzing such
samples by sandwich ELISA is the interference by antiIgG antibodies in the samples. One approach to this
problem is the use of Fab fragments in the assay,
because the anti-IgG antibodies are usually directed
against the Fc part of the IgG molecule. However, the
digestion of IgG to produce pure Fab fragments is time
consuming and usually results in loss of antibody.
Instead, we have studied the possibility of using different antibodies in the test to avoid false-positive
reactions with HF. Because mammalian and chicken
IgG have no immunological cross-reactivity (12), and
because HF do not agglutinate latex particles coated
with chicken IgG (7), we have studied the possibility of
using chicken antibodies in ELISA to avoid the reactions
with anti-IgG antibodies. Here we show that if either or
both of the antibodies in the assay are avian, there will
be no false-positive reactions with anti-IgG antibodies
or HF. Because chickens have a different way of producing antibody diversity (13), this will probably also reduce interference from the anti-idiotype antibodies that
are ordinarily present in serum.
The treatment of IgG fractions or whole serum has
been used for many years in therapy of various diseases.
The antibody response in some of these cases is the
cause of “serum sickness.” The future will probably see
an increasing number of patients with anti-IgG antibodies other than the traditional HF. As monoclonal antibodies are used more frequently in treating cancer or
during organ transplantation, there will be an increasing number of patients with anti-monoclonal antibodies.
The monoclonal antibodies used in vivo have all been of
mammalian origin; this will probably continue in the
future as part of the effort to make the antibodies as
similar to human IgG as possible. In Sweden so far,
there is only one antibody registered as a therapeutic
agent, an anti-digoxin Fab fragment from sheep, but
there will probably be more such treatments available
in the future. Thus, anti-mammalian
IgG antibodies
will be found in patients with many different diseases,
and it will be difficult to predict which ones will lead to
interference from anti-IgG antibodies. For example,
occasionally a healthy blood donor will have high titers
of HF (3).
Very little is known about the normal occurrence of
anti-chicken antibodies in human serum. We have previously used chicken antibody-coated microtiter plates
to study circulating immune complexes (14). We used
the same system to study reactivity with nonspecific
chicken IgG, testing sera from ‘-‘450 patients and cerebrospinal fluid from ‘-‘.40patients; none of these samples
showed any antibody reactivity to chicken IgG. The
immunological difference between chicken and mammalian IgG means that any antibody present in the test
sample that reacts with antibodies from one of these
species will not recognize antibodies from the other
species. We thus suggest that the best choice for avoid-
ing false-positive results in sandwich ELISA is to use one
mammalian antibody and one chicken antibody; the
second best is to use both antibodies of avian origin.
Chicken antibodies are not very widely used. This is
probably due to tradition, because chicken antibodies
have advantages over mammalian antibodies in many
assays. For instance, chicken antibodies have good stability. We tested antibody preparations after five years
of storage at 4#{176}C
or after six months at room temperature; we detected no loss in antibody titer by immunodiffusion. Chickens have fragile veins, which makes it
difficult to obtain serum, but large amounts of antibodies can be found in egg yolk (one egg contains IgG
equivalent to 10 mL of serum or more). The chicken
antibodies
in this paper were all purified
from egg yolk.
Immunization of chickens is similar to immunization of
rabbits, and chicken antibodies can then be easily purified from egg yolk (15, 16). We obtained good antibody
response in chickens immunized with 25 or 100 p.g of
antigen under a procedure similar to the one we use for
rabbits. The amount of antibody produced in egg yolk
from a chicken exceeds the amount produced in a rabbit
during a similar time. Thus, there is no problem in
producing chicken antibodies for the laboratory, and
many chicken antibodies are also commercially available. The evolutionary difference between mammals
and chickens should mean that more numerous sites in
a mammalian antigen will evoke an antibody response.
Thus, a greater number of antibodies will bind to the
molecule, and the signal in an immunoassay will be
amplified (17).
In conclusion, there are many reports of false-positive
results caused by reaction with HF. This report shows
that chicken antibodies are a simple way to avoid this
problem in sandwich EUSA.
Excellent technical assistance was provided by Mrs. Monica
Ferm. This work was supported by the Swedish Medical Research
Council (project no. 13X-2518).
References
1. Tijssen P. Practice and theory of enzyme immunoassays. Lab
Tech Biochem Molec Biol 1985;15:329-84.
2. Hechemy KE, Michaelson EE. Latex particle assays in laboratory medicine. Part II. Lab Manage 1984; 22:26-35.
3. Christian CL. Rheumatoid arthritis. In: Samter M, Talmage
DW, Rose B, Sherman WB, Vaughan JH, eds. Immunological
diseases, 2nd ed. Boston: Little, Brown and Co., 1971:1014-28.
4. Johnson PM, Page Faulk W. Rheumatoid factor: its nature,
specificity and production in rheumatoid arthritis. Clin Immunol
Immunopathol
1976;6:416-30.
5. Teitsson I, Valdimarsson H. Use of monoclonal antibodies and
F(ab’)2 enzyme conjugates in ELISA for 1gM, IgA and IgG rheumatoid factors. J ImmunolMethods 1984;71:149-61.
6. Zrein M, WeissE, Van RegenmortelMHV. Detection of rheumatoid factors of different isotypes by ELISA using biotinylated
avian antibodies. Imniunol Invest 1988;17:165-81.
7. Larsson A, Sjaquist J. Chicken antibodies: a tool to avoid false
positive results by rheumatoid factor in latex fixation tests. J
Immunol Methods 1988;108:205-8.
8. Avranieas 5, Ternynck T, GuesdonJL. Coupling of enzymes to
antibodies and antigens. Scand
J
Immunol
1978;8(Suppl
9. Bayer EA, Skutelsky E, Wilchek M. The avidin-biotin
7):7-23.
complex
in affinity cytochemistry.
Methods Enzymol 1979;62:308-15.
10. Waaler E. On the occurrence of a factor in human serum
CLINICALCHEMISTRY,Vol. 37, No. 3, 1991 413
activating the specific agglutination of sheepblood corpuscles.
Acta Pathol Microbiol Scand 1940;17:172-88.
11. RoseHM, Ragan C, Pearce E, et al. Differential agglutination
of normal and sensitized erythrocytes by sera of patients with
rheumatoid arthritis. Proc Soc Exp Biol Med 1948;68:1-6.
12. Hadge D, Ambrosius H. Evolution of low molecular weight
immunoglobulins-IV. IgY-like immunoglobulinsof birds, reptiles
and amphibians, precursors of mammalian IgA. Mo! Immunol
1984;21:699-707.
13. Reynaud CA, Anquez V, Dahan A, Weill JC. A single reevent generatesmost of the chicken
light chain diversity. Cell 1985;40:283-91.
arrangement
immunoglobulin
414 CLINICALCHEMISTRY,Vol. 37, No. 3, 1991
14. Larsson A, JonssonL, SjoquistJ. Determination of circulating
immune complexesby chicken anti-human C3 and antihuman Clq
microELISA.
J Immunol
Methods 1988;113;93-9.
15. Jensenius JC, Andersen I, Hau J, Crone M, Koch C. Eggs:
conveniently packaged antibodies. Methods for purification of yolk
IgG.
J Immunol Methods 1981;46:63-8.
16. Losch U, Schranner I, Wanke R, Jurgens L. The chickenegg,
an antibody source. J Vet Med 1986;33:609-19.
17. Horton JJ, Holden CA, Ward PJ, MacDonald DM, Sanderson
AR. Exploitation of phylogenetic distance in cell surface immune
labeling;
1984;85:96-9.
studies with
beta2-microglobulin.
J
Invest
Dermatol