J Vet Diagn Invest 7:509-514 (1995)
Role of sodium azide in reducing nonspecific
color development in enzyme immunoassays
Parmesh K. Saini, Donald W. Webert, Joanne C. Judkins
Abstract. Improved enzyme immunoassay (EIA) procedures achieved by incorporating sodium azide during
predilution of serum samples in a solid-phase EIA for the detection of anti-Toxoplasma antibody in swine using
a peroxidase conjugate and in all washes of a bovine brucellosis rapid card test EIA using alkaline phosphatase
conjugate are reported. Without this modification, substantial background interference was encountered that
showed direct correlation with the degree of hemolysis of the serum samples. Anti- Toxoplasma gondii antibodynegative samples, separated by subjective groupings based on degree of hemolysis, into “clear,” “slight,” and
“gross/total” samples, had a mean ± standard deviation of 0.150 ± 0.072, 0.187 ± 0.105, and 0.232 ± 0.108,
respectively. The incorporation of sodium azide during the initial step of serum dilution dramatically eliminated
the background, giving a mean ± standard deviation of 0.079 ± 0.029, 0.076 ± 0.022, and 0.081 ± 0.029,
respectively. The level of endogenous peroxidase activity, a possible factor for this nonspecific interference, was
considerably elevated in some of the swine sera. The clear, slight, and gross/total categories had relative levels
of 1%, 2%, and 51% peroxidase activity compared to the conjugate peroxidase activity of 100%. Whereas sodium
azide could be used only in sample predilution in the swine toxoplasmosis peroxidase-conjugate test, in the
bovine brucellosis alkaline phosphatase-conjugate card test it could be used in all wash cycles. Many brucellosis
card test results were visually uninterpretable because of significant background color when the manufacturer’s
wash reagent was used. The substitution of a wash reagent containing sodium azide eliminated background
color, giving a visually unambiguous test.
The enzyme immunoassay (EIA) was introduced in
19716,23 and is now the most widely used testing methodology worldwide.1 Two of the most widely used enzyme conjugates in EIA methodologies are peroxidase
(PO) and alkaline phosphatase (AP). However, these
enzymes are also present in normal body tissues and
fluids. 14,17,19,21 These endogenous enzymes thus are a
potential source of nonspecific color development. In
a homogeneous EIA both the immunological reaction
and the detection of the extent of the reaction are carried out in a homogeneous solution without physical
separation of free and antibody-bound components.
The problem of interference by endogenous enzymes
is well recognized and enzyme markers such as ß-galactosidase and glucose oxidase, being absent endog22
enously, are used. In a heterogeneous EIA, such as
reported here, nonspecific color development resulting
from the presence of endogenous enzymes has not been
considered to be a serious problem because of the wash
cycles between each incubation step. Unlike homogeneous EIAs, all heterogeneous assays have at least 1
separation step that differentiates reacted from unreacted materia1.24
From the Serology Branch, Pathology and Serology Division, Science and Technology Program, Food Safety and Inspection Service,
U.S. Department of Agriculture, Bldg. 318-C, BARC-East, Beltsville,
MD 20705.
Received for publication April 7, 1994.
In a previous study this laboratory reported that
extensively hemolyzed serum samples adversely affected EIA test results in a methodology using PO conjugate.2 It was speculated that the interference was
caused by PO from white blood cells. The background
color could be eliminated by the addition of sodium
azide to the sample prior to being assayed.
In this article we report our experience in solving
substantial background-color problems encountered in
dealing with slightly to extensively hemolyzed serum
samples in assays using PO conjugate (detection of
swine anti- Toxoplasma antibodies using a microtiter
EIA procedure) and AP conjugate (detection of bovine
anti-Brucella antibodies using an onsite rapid card test
kit).
Materials and methods
Toxoplasmosis enzyme immunoassay
Antigen. Trophozoites of Toxoplasma gondiia (cell cultured, unfixed strain RH) with 2 x 109 organisms per ml
were centrifuged at 2,000 x g for 15 min after disruption by
freeze-thawing. The supematant, with a protein concentration of 3.3 mg/ml, was used as stock antigen and was stored
at -50 C.
Assay. An EIA was developed to detect anti-T. gondii
immunoglobulin G antibody in swine serum samples.6 Tests
were conducted in microtiter platesb with flat-bottom wells.
Plates were coated with 100 µl stock antigen diluted 1:400
in 0.03 M sodium bicarbonate buffer, pH 9.6, with 0.02%
509
510
Saini, Webert, Judkins
Figure 1. Comparison of mean ODs ± standard deviations and
P values of treated and untreated aliquots of 200 anti-T. gondii
antibody-negative samples. ODs were read at 405 nm and recorded
after a 5-minute substrate incubation period.
sodium azide (antimicrobial) and incubated overnight at 4
C. All test component additions to plate wells were 100 µl.
Incubation periods were 5 min or 8 min at 30 C, with shaking.c
Between incubation periods wells were washed six times with
phosphate-buffered saline, pH 7.4, containing 0.05% Tween
80d (PBST). To establish the role of sodium azide in reducing
nonspecific color development, each serum sample was divided into 2 aliquots. One aliquot was diluted 1:100 with
PBST (untreated) and the other was diluted 1:100 with PBST
containing 0.1% sodium azide (PBSTZ) (treated). Forty samples, in duplicate, plus 3 controls and a blank (4 wells each)
were tested in a single plate with sample placement to minimize well-position effect. 20 Anti-porcine IgG peroxidase
conjugatee was diluted 1:1,000 in PBST. The substrate solution was H2O2 and ABTSf diluted in distilled water. The
substrate reaction was stopped after 5 min incubation. Optical density (OD) was measured at 405 nm.
Serum samples. A total of 799 serum samples from 1 mo
of a 12-mo national survey for swine pseudorabies seroprevalence was used.4 The samples, when originally received
and stored in our serum bank, had been classified subjectively
into 4 groups based on degree of hemolysis: clear (no hemolysis), slight (tinge to light red), gross (slight to dark red),
and total (almost black).
Two hundred of the serum samples, anti-Toxoplasma antibody negative by a modified agglutination test (MAT) at a
dilution of 1:25 were used to develop the positive/negative
threshold for the toxoplasmosis enzyme immunoassay4,5 and
to conduct further analyses in the evaluation of the effect of
sodium azide in reducing nonspecific color development.
A second subgroup of 90 serum samples were anti-T. gondii positive by the reported EIA, based on the threshold
established in the analyses of the 200 MAT antibody-negative sera.
Measurement of endogenous PO activity. Endogenous PO
was demonstrated and its activity was measured by adding
substrate directly to diluted serum aliquots from pooled untreated samples representing clear, slight, and combined gross/
Figure 2. Comparison of mean ODs ±_ standard deviations and
P values of treated and untreated aliquots of 90 anti-T. gondii antibody-positive samples. ODs were read at 405 nm and recorded
after a 5-minute substrate incubation period.
total hemolysis categories and anti-porcine PO conjugate at
1:100, 1:200, 1:1,000, and 1:20,000 dilutions, respectively.
For reference, positive standards with known PO activityg
were run at 0.125, 0.25, and 0.5 units per ml. Expressed in
units, 1 unit of PO activity was equivalent to 1 µmol of ABTS
hydrolyzed per min.
Brucellosis card test
Brucellosis. Recently developed card tests for brucellosish
were used for the detection of anti-Brucella antibodies in
cattle serum samples. The test procedure was that of the
manufacturer, using the AP conjugate and other reagents
provided with the kit. In the comparative study to determine
the effect of sodium azide, serum samples were tested undiluted and were not pretreated with sodium azide. Tests
were run simultaneously on 2 aliquots of each serum sample.
The wash reagent used for 1 aliquot was that supplied with
the test kit (untreated) and for the second aliquot was PBSTZ
(treated). Test results were intrepreted after a 5-min substrate
incubation. Quantitative color readings were made using a
commercial chromameteri calibrated against a standard white
calibration plate. Results were reported as delta-E of the test
minus control values for both treated and untreated samples.
Serum samples. Anti-Brucella antibody-positive samplesj
were from cattle that culture-tested positive for Brucella
abortus field strain 19. All anti-Brucella antibody-negative
samplesk tested negative by the APHIS official rapid screening test, standard plate agglutination test, standard card test,
rivanol test, and complement fixation test.
Statistical methods. All statistical analyses of test results
were conducted using the paired t-test.
Results
Toxoplasmosis microtiter EIA
PBST vs. PBSTZ as serum diluent. In a comparison
of untreated and treated aliquots of the 200 serum
Sodium azide and background color reduction
Figure 3. PO activity in clear, slight, and gross/total serum samples and anti-porcine conjugate measured over time at 1:100, 1:200,
1:1,000, and 1:20,000 dilutions, respectively. All dilutions were made
with PBST. Calculated units of PO activity are per ml of undiluted
sample.
samples negative for anti-T. gondii antibody, the group
of untreated aliquots had an OD mean ± standard
deviation of 0.194 ± 0.105 (OD range of 0.032-0.66 1).
The treated aliquots had an OD mean ± standard
deviation of 0.078 ± 0.025 (OD range of 0.017-0.153).
The EIA analyses of treated and untreated aliquots of
the three hemolytic categories are shown in Fig. 1. The
mean OD readings of the untreated aliquots (no sodium azide) show a direct correlation with the degree
of hemolysis. The treated aliquots, however, show virtually identical mean OD readings.
In the comparison of untreated and treated aliquots
of the 90 serum samples positive for anti-T. gondii
antibody, the group of untreated aliquots had an OD
mean ± standard deviation of 0.274 ± 0.145 (OD
range of 0.078-0.640), whereas the treated aliquots
showed less variation, with an OD mean ± standard
deviation of 0.224 ± 0.116 (OD range of 0.142-0.824).
The EIA analyses of treated and untreated aliquots of
the three hemolytic categories are shown in Fig. 2. As
was evident with the negative samples, the untreated
aliquots of the positive samples showed progressively
higher ODs that correlated with the degree of hemolysis.
Compared to the negative samples, the percent reduction in enzyme activity in the treated aliquots of
the positive samples is much less, 19% vs. 60%. The
average of 60% reduction in enzyme activity in the
negative samples after PBSTZ treatment is composed
of 47%, 59%, and 65% in the clear, slight, and gross/
total categories, respectively. In the positive samples,
however, the average of 19% reduction in EIA activity
after treatment is broken down to 8% and 45% in the
slight and gross/total categories, respectively, whereas
the clear category showed an increase of 6% after treat-
511
Figure 4. Relative PO activity in untreated clear, slight, and
gross/total sera compared with the anti-porcine conjugate used during the actual assay procedure set at 100% activity. Calculated relative PO activity is based on sample and reagent dilutions used in
the assay (serum, 1:100; conjugate, 1:1,000).
ment. One sample in this group showed a 29% increase
in OD after treatment.
Endogenous PO activity
PO activity measured separately in the pooled clear,
slight, and gross/total hemolytic serum categories and
in anti-porcine conjugate is shown in Fig. 3 for untreated samples. PO activity in treated aliquots is not
indicated as OD readings were less than
those for
untreated aliquots at 1 minute incubation and remained at essentially the same level throughout the
30-minute evaluation period.
Figure 4 shows the relative endogenous PO activity
in pools of each of the serum hemolytic categories when
diluted 1:100 in PBST (untreated), the routine test dilution, compared to conjugate PO activity (equals
100%) when diluted 1:1,000 in PBST, the routine test
dilution.
Brucellosis card test
When the manufacturer’s protocol was followed, color differences between a test reaction and its respective
sample control were visually indistinguishable in many
of the untreated samples, whether anti-Brucella antibody negative or positive.
For anti-Brucella antibody-negative sera (Table 1),
the control and test reactions of the untreated aliquots
were not significantly different (P = 0.4651) nor were
they visually distinguishable. The same was true for
the treated aliquots (P = 0.5051). However, the mean
delta-E values of the treated aliquots were significantly
different than those of the untreated aliquots, both for
the control (P = 0.0001) and the test reactions (P =
0.0001).
For anti-Brucella antibody-positive sera (Table 2),
the control and test reactions of the untreated aliquots
were not significantly different (P = 0.1760) and were
visually indistinguishable. However, in the treated aliquots, the control and test reactions were significantly
different (P = 0.0001) and were visually unambiguous.
Discussion
One of the significant causes of background color
development in heterogenous enzyme immunoassays
is the nonspecific attachment of residual conjugates at
nonantigenic sites on the adsorbed protein and/or directly on the solid-phase surface.
The extreme sensitivity of EIA technology is due to
the high turnover of enzyme-labeled antibodies (conjugates). A single enzyme molecule can typically convert 103-106 molecules of substrate into products.9 On
the negative side, however, it will also amplify nonspecific reactions. In this study PBSTZ was crucial to
the elimination of the nonspecific interference present
in the test samples.
Blood PO, present primarily in leukocytes, is one of
the richest sources of PO activity,3,7,8 with some authors reporting its concentration to the extent of 5%
of the entire dry weight of the blood.18 Endogenous
PO, if immobilized on the solid-phase surface and/or
on preadsorbed proteins, could contribute significantly
to nonspecific color development as only 0.0075% PO
is required to catalyze a reaction.8
A previous study in this laboratory showed that
preassay treatment of hemolyzed sera with sodium
azide would eliminate the nonspecific reactions that
were speculated to be caused by endogenous peroxidase.2 However, the effect of the treatment by sodium
azide over a range of hemolysis and specific analyte
concentrations was not undertaken.
Sodium azide is a potent inhibitor of PO.12,13 It is
reported to inhibit catalytic PO activity irreversibly11
and would, therefore, render endogenous PO catalytically inactive during preassay serum dilution with
PBSTZ. We have shown that the simple step of prediluting serum samples with sodium azide was very
† Absolute average difference.
effective in eliminating an extensive background problem, especially in hemolyzed samples. If added to any
of the “wash” steps, sodium azide inactivated the conjugate PO.
Although this study utilized different degrees of hemolysis to separate and evaluate serum samples for
their varying contributions to nonspecific color development in the respective enzyme immunoassays, the
hemolysis acted only as a guide, and apparently a good
guide, to the potential damage done to other cellular
components of the blood.
The PO activity in untreated samples was demonstrated and was significant, especially in the gross/total
hemolytic category (Fig. 3). In this group endogenous
PO activity equaled
that of the conjugate PO when
both were diluted to routine test parameters (Fig. 4).
It is quite likely that 1 or more of the individual samples in the pool had endogenous peroxidase activity
equal to or greater than that normally present in the
working dilution of the conjugate.
In this report the demonstration of the magnitude
of reduction in OD in the treated aliquots of the different hemolytic sample categories (Figs. 1, 2) depended upon at least 3 primary factors: both the amount
of specific anti- Toxoplasma antibody and endogenous
enzyme in the test sample and the substrate incubation
time interval (5 minutes) when test results were recorded. There may be other factors involved as some
of the treated antibody-positive samples showed increased ODs after treatment.
When the ODs of the 200 anti- Toxoplasma antibody
negative samples diluted with PBSTZ (treated) are
matched with a corresponding aliquot diluted with
PBST (untreated), regardless of hemolytic category the
absence of specific anti- Toxoplasma antibody is clearly
demonstrated by the early reaction rate differences between the treated- and untreated-aliquot substrate re-
513
Sodium azide and background color reduction
actions at 5 minutes incubation. In each of the hemolytic categories the differences were statistically significant at this incubation time (Fig. 1). The mean OD
of all aliquots of the sodium azide-treated samples
(0.078) was less than V-2 that of the combined untreated
samples (0.194), with P < 0.0001. There was a direct
correlation between OD readings in the untreated samples and the hemolytic category groups.
When the 5-minute substrate incubation ODs of the
group of 90 anti- Toxoplasma antibody-positive samples are matched before and after sodium azide treatment, the reaction rate differences in the clear and
slight hemolytic categories are not significantly different at a 5-minute substrate incubation (Fig. 2). The
presence of specific anti- Toxoplasma antibody in these
samples significantly increases the reaction rates over
those seen in the respective antibody-negative samples.
Further, in these samples the endogenous PO is only
a small percentage when compared to the test PO itself,
and the reaction rates of the treated and untreated
aliquots are, therefore, similar in the shorter incubation period. However, in the gross/total hemolytic category, the endogenous PO is much greater and significantly increases the reaction rate, which is distinctly
discernible even at the 5-minute substrate incubation
interval (Fig. 2).
Although the sodium azide had a definite effect on
the test result, from the overall study it appears that
specific antigen/antibody interaction is not adversely
affected by the sodium azide treatment used in this
study.
Unlike in the toxoplasmosis enzyme immunoassay,
sodium azide could be used in all wash cycles of the
bovine brucellosis card test. APs, present ubiquitously
in body tissues and fluids in high concentration,14-16
are a family of enzymes with a wide range of characteristics. Intestinal AP is uniquely discerned from nonintestinal APs. Our results corroborate that sodium
azide has little or no effect on conjugate AP23 (which
is specified to be intestinal AP in most commercial
preparations).
However, it appears that sodium azide inhibits endogenous nonintestinal AP. We substituted levamisole, which preferentially inhibits nonintestinal AP,10
in lieu of PBSTZ in the wash reagent, with equal reduction in nonspecific color development. It is possible
that sodium azide, in relation to AP, functions similar
to levamisole, having no effect on conjugate AP, but
inhibiting endogenous nonintestinal AP, which we believe was a primary contributor to the nonspecific
background color of the brucellosis card test.
The results of a test system targeted for field use as
a screening tool must be as unambiguous as possible.
Too often users of a field test are required to make
judgements based on evaluations of ambiguous test
results. With the brucellosis card test used in the studies reported here, after a 5-minute substrate incubation, there was no visually distinguishable color in any
of the treated-control reactions nor in the treated antibody-negative test reactions. In treated antibodypositive test reactions, color development was generally readily visible within 15 seconds of the addition
of substrate. Hemolyzed samples did not interfere with
test result evaluations because the majority of the initial sample color was restricted from the reaction area
of the card during lateral flow of the serum sample
through a membrane. In the few cases when a tinge of
color reached the reaction area, the successive wash
cycles cleared it out.
In our test development procedures, but for the use
of sodium azide, the two test systems reported here
would have been discarded due to excessive nonspecific color development.
Acknowledgements
Drs. J. P. Dubey and W. C. Zimmermann (deceased) are
gratefully acknowledged for their assistance in providing invaluable serum samples from Toxoplasma-infected swine.
Dr. Dubey is further acknowledged for providing anti-Toxoplasma antibody MAT results on those sera used to develop
the threshold for the EIA test reported here. Thanks are due
to APHIS for providing the valuable serum samples from
the statistically based National Pseudorabies Prevalence
Study.
Sources and manufacturers
General Biometrics, Bountiful, UT.
Flow Laboratories, McLean, VA.
Bio-Tek Instruments, Winooski, VT.
Fisher Scientific, Fair Lawn, NJ.
Miles Scientific, Naperville, IL.
Kirkegaard and Perry Laboratories, Gaithersburg, MD.
Sigma Chemical Co., St. Louis, MO.
Brucellosis SeroCards™, Trinity Biotech, Inc., Irvine, CA.
Spectroscopy Systems, Inc., Silver Spring, MD.
Dr. P. Nicoletti, University of Florida, College of Veterinary
Medicine, Gainesville, FL.
Mr. R. Forde, Colorado State/Federal Brucellosis Laboratory,
Denver, CO.
Mention of a trade name, proprietary product, or specific equipment does not constitute a guarantee of warranty by the US Department of Agriculture and does not imply its approval to the
exclusion of other products that may be suitable.
References
Bates DL: 1987, Enzyme amplification diagnostics. Trends Biotechnol 5:204-209.
Berkowitz DB, Webert DW: 1986, Enzyme immunoassay-based
survey of prevalence of gentamicin in serum of marketed swine.
J Assoc Off Anal Chem 69:437-441.
Bos A, Wever R, Roos D: 1978, Characterization and quantification of the peroxidase in human monocytes. Biochim Biophys Acta 525:37-44.
Dubey JP, Leighty JC, Beal VC, et al.: 1991, National sero-
514
Saini, Webert, Judkins
prevalence of Toxoplasma gondii in pigs. J Parasitol 77:517521.
5. Dubey JP, Miller S, Desmonts G, et al.: 1986, Toxoplasma
gondii-induced abortion in dairy goats. J Am Vet Med Assoc
188:159-162.
6. Engvall E, Perlmann P: 1971, Enzyme-linked immunosorbent
assay (ELISA). Quantitative assay of immunoglobulin G. Immunochemistry 8:871-874.
7. Jain NC: 1967, Peroxidase activity in leukocytes of some animal species. Folia Haematol 88:297-304.
8. Kotkowski S, Machoy Z: 1968, Human blood peroxidase in
the process of oxidation of diluted indigocarmine solutions. Bull
Pol Med Sci Hist 11:68-69.
9. Maggio ET: 1980, Enzymes as immunochemical labels. In:
Enzyme-immunoassay, ed. Maggio ET, pp. 53-70. CRC Press,
Boca Raton, FL.
10. McComb RB, Bowers GN Jr, Posen S: 1979, Alkaline phosphatase. Plenum Press, New York, NY.
11. Ohno Y, Gallin JI: 1985, Diffusion of extracellular hydrogen
peroxide into intracellular compartments of human neutrophils.
Studies utilizing the inactivation of myeloperoxidase by hydrogen peroxide and azide. J Biol Chem 260:8438-8446.
12. Persijn JP, Jonker KM: 1978, A terminating reagent for the
peroxidase-labelled enzyme immunoassay. J Clin Chem Clin
Biochem 16:531-532.
13. Richardson TC, Chapman DV, Heyderman E: 1983, Immunoperoxidase techniques: the deleterious effect of sodium azide
on the activity of peroxidase conjugates. J Clin Pathol 36:411414.
14. Saini PK: 1970, Studies of alkaline phosphatases in membranes
of rat tissues. Ph.D. thesis, University of Sydney, Sydney, Australia.
15. Saini PK, Done J: 1972, The diversity of alkaline phosphatase
from rat intestine, isolation and purification of the enzyme(s).
Biochim Biophys Acta 258: 147-153.
16. Saini PK, Posen S: 1969, The origin of serum alkaline phosphatase in the rat. Biochim Biophys Acta 177:42-49.
17. Schrenk WJ, Kurzinger K: 1986, Purpose-made enzyme conjugates for quantitative immunoassays. Med Lab Sci 43:269276.
18. Schultz J, Kaminker K: 1962, Myeloperoxidase of the leukocyte
of normal human blood. 1. Content and localization. Arch
Biochim Biophys 96:465-467.
19. Steinmetz H, Pfreundschuh MG: 1987, Production of monoclonal antibodies against glucose oxidase, alkaline phosphatase
and peroxidase. Their application in a highly sensitive antigen
spot microassay. J Immunol Methods 101:251-259.
20. Stemshorn BW, Buckley DJ, Amour GS, et al.: 1983, A computer-interfaced photometer and systematic spacing of duplicates to control within-plate enzyme-immunoassay variation. J
Immunol Methods 61:367-375.
21. Straus W: 1971, Inhibition of peroxidase by methanol and by
methanol-nitroferricyanide for use in immunoperoxidase procedure. J Histochem Cytochem 19:682-688.
22. Ullman EF, Maggio ET: 1980, Principles of homogeneous enzyme-immunoassay. In: Enzyme-immunoassay, ed. Maggio ET,
pp. 105-134. CRC Press, Boca Raton, FL.
23. van Weemen BK, Schuurs AHWM: 1971, Immunoassay using
antigen-enzyme conjugates. FEBS Lett 15:232-236.
24. Voller A: 1980, Heterogeneous enzyme-immunoassays and their
applications. In: Enzyme-immunoassay, ed. Maggio ET, pp.
181-196. CRC Press, Boca Raton, FL.