Abstract
In the Toxoplasma gondii immunoglobulin M (IgM) capture fluorometric enzyme immunoassay used as a model, nonspecific responses due to the binding of human IgM to horseradish peroxidase (HRP) conjugates were observed despite the removal of the Fc portion of the immunoglobulin. This interaction may be mediated through the binding of human IgM to the HRP moiety of the conjugate. Addition of polymerized HRP into the reaction mixture reduced nonspecific signals in the majority of low false-positive serum reactions. Other plausible sites of interaction are conserved epitopes of mouse immunoglobulins presenting antigenic similarities with the allotopes of other species. Fragmentation of mouse antimicrobial IgG to Fab′ and selection of proper conjugation procedure improved assay specificity.
Solid-phase capture anti-immunoglobulin M (IgM) immunoassays were developed to overcome the problems with indirect immunoassays caused by rheumatoid factor (RF) and competition between specific IgG and IgM for antigen binding sites on the solid phase (3, 4, 7, 8). However, capture type assays also show immunological interferences. For example, bound to the anti-IgM capture antibodies, RF can produce false-positive signals by reacting with antimicrobial-labelled IgGs (3, 7, 8) through their Fc fragments. The second type of immunological interference involves specific human IgGs bound to the captured RF via their Fc fragment which are detected through the microbial antigen bound to the specific antibody (8, 9). The third type of immunological interference involves complex reactions of antinuclear antibodies (ANA) described for immunofluorescence tests (8). The fourth mechanism of immunological interference which may occur in both indirect and capture types of immunoassays is mediated through so-called “naturally occurring antibodies” or “natural autoantibodies,” which are of the IgG and IgM classes and exhibit a broad range of reactivities (1, 2, 10, 11). Below we describe two additional interference mechanisms that have not been previously reported.
The model.
Toxoplasma gondii IgM capture fluorometric enzyme immunoassay (FEIA) (5) was used as a model. Briefly, streptavidin-coated microplates (Labsystems, Helsinki, Finland) were used as a solid phase. Two microliters of each plasma sample was added to a mixture containing 150 μl of biotinylated polyclonal sheep anti-human IgM antibody in 0.01 M phosphate-buffered saline (PBS) (pH 7.4)–Tween 20–1% bovine serum albumin. After incubation, the microplates were washed, and 150 μl of sonicated T. gondii tachyzoites of RH strain which had been premixed with an anti-T. gondii horseradish peroxidase (HRP)-labelled mouse monoclonal antibody was added as an antigen (these tachyzoites were also used for immunoblotting and indirect enzyme immunoassay [EIA]). The fluorogenic 3-p-hydroxyphenylpropionic acid substrate (150 μl/well) reaction was performed for 30 min. The enzymatic reaction was stopped, and the signals were measured with a Fluoroskan II microplate fluorometer (Labsystems). Alternatively, instead of the anti-T. gondii conjugate with an antigen, the respective conjugate was used alone or was replaced by a variety of HRP conjugates (Table 1). All conjugates were prepared according to Ishikawa et al. (6) by optimized techniques. The proper preparation of the conjugates was confirmed by the molar ratio of HRP/IgG based on the spectrophotometric measurements at optical densities at 403 nm (OD403) and 280 nm (OD280) from each fraction. Interferences were also studied by using another model where a monoclonal anti-Toxoplasma gondii antibody was used intact (1 μg/ml) together with specific antigen. The attachment of intact antibody was detected by sequential addition of rabbit anti-mouse HRP-labelled IgG (Dako, Glostrup, Denmark). The reactivities of samples with the rabbit anti-mouse IgG-HRP conjugate alone were also studied.
TABLE 1.
Conjugates for the study of nonspecific reactions
| Conjugate specificitya | Antibody | Immunoglobulin or fragment | Concn of conjugate per HRP (μg/ml) | Cross-linker(s) (source of conjugate) |
|---|---|---|---|---|
| F(ab′)2-S-AMSA-SMCC-HRP, anti-T. gondii | Mouse monoclonal | F(ab′)2 | 0.2 | S-AMSA-SMCC (Labsystems) |
| F(ab′)2-S-AMSA-SPDP-HRP, anti-T. gondii | Mouse monoclonal | F(ab′)2 | 0.2 | S-AMSA-SPDP (Labsystems) |
| Fab′-SPDP-HRP, anti-T. gondii | Mouse monoclonal | Fab′ | 0.2 | SPDP (Labsystems) |
| IgG-HRP, anti-mouse | Rabbit polyclonal | IgG | 1:1,500b | Glutardialdehyde (Dako) |
| IgG-S-AMSA-SMCC-HRP, anti-TSH (αβ) | Mouse monoclonal | IgG | 8 | S-AMSA-SMCC (Labsystems) |
| F(ab′)2-S-AMSA-SMCC-HRP, anti-HBc | Mouse monoclonal | F(ab′)2 | 1.2 | S-AMSA-SMCC (Labsystems) |
| F(ab′)2-S-AMSA-SMCC-HRP, anti-HSV-1 | Mouse monoclonal | F(ab′)2 | 2 | S-AMSA-SMCC (Labsystems) |
| Fab′-SPDP-HRP, anti-HBs | Mouse monoclonal | Fab′ | 0.5 | SPDP (Labsystems) |
| Fab′-SPDP-HRP, anti-HBs | Sheep polyclonal | Fab′ | 1 | SPDP (Labsystems) |
| IgG-HRP, anti-fluorescein | Sheep polyclonal | IgG | 1:200-1:400b | —c (DuPont, Foster City, Calif.) |
TSH, thyroid-stimulating hormone; HBc, hepatitis B core antigen; HSV-1, herpes simplex virus type 1; HBs, hepatitis B surface antigen.
The original concentration was not known; optimized dilution was used.
—, conjugation method not indicated.
For all experiments the same controls were used in every run. Negative control serum was from a staff member, and positive and low-positive control sera were from Antibody Systems LTD (Bedford, Texas). The latter were proven to be true positives by an indirect T. gondii IgM EIA (Labsystems). The borderline control was artificially prepared by diluting (1:16) the positive control with the negative control, resulting in a signal that was approximately threefold greater than that of the negative control (5). To interpret the reactivity of samples with each conjugate tested, signals derived from each individual sample were compared to the signal of the borderline control.
Sixteen T. gondii IgM false-positive plasma samples were selected after screening several hundred adult specimens from the Arhangelsk Blood Bank (Arhangelsk, Russia). A T. gondii IgM capture EIA with F(ab)2-S-acetyl mercaptosuccinic anhydride (S-AMSA)-N-succinimidyl-4-(N-maleimidomethyl)cyclohexane-1-carboxylate (SMCC)-HRP (Table 1) as a conjugate was used for screening. To test for the persistency of interfering antibodies a new sample after 1 year was requested. A new sample was obtained from only 6 of 16 individuals. All samples were divided into aliquots and kept frozen at −20°C. Repeated freezing and thawing of the aliquots was difficult to avoid but the samples did not show signs of bacterial growth or any other visible contamination. All samples were tested with routinely used methods for the absence of ANA and RF in the Aurora Hospital (Helsinki, Finland) and were found to be negative. Additionally, one extraordinarily reactive false-positive serum sample was from a pregnant woman from France, and due to its small amount it was used only in some experiments (with Boehringer blockers [see Table 3] and in dilution experiments and reactions with HRP-coated plates). Also, one serum sample (Ang) from a staff member was T. gondii IgM borderline reactive and presented as nonspecific. The conclusion that the tested samples were indeed false positive was based on (i) the negative results of immunoblot analysis where possible, (ii) the patterns of reactivity with anti-T. gondii HRP conjugates in the presence and absence of specific antigen in the T. gondii IgM, IgG, and IgA FEIAs (see Fig. 1), (iii) nonreactivity in an indirect T. gondii IgM EIA (Labsystems), (iv) discrepant data (samples 11 and 8) from Platelia Toxo IgM (Sanofi Diagnostics Pasteur, Marnes la Coquette, France) and EIAGEN Toxoplasmosis IgM (CloneSystems, Casaleccio di Reno, Italy) assays, and (v) the reactivity of some samples with the commercial blocker polyPOD (Boehringer, Mannheim, Germany). To exclude interference in our test model by other autoimmune antibodies, human sera containing nucleolar, mitochondrial, histone, ANA-RF, and microsomal antibodies (Biomedical Resources, Hatboro, Pa.) were tested too.
TABLE 3.
Commercial blockers from Boehringer (Mannheim, Germany) for elimination of nonspecific reactions
| Blocker (catalog no.) | Recommended working concn (trial concn) | Possible mechanism of elimination of nonspecificity | Blocking effect |
|---|---|---|---|
| MAK 33 (1200 941) | 50–1,000 μg/ml (1,000 μg/ml) | Monoclonal antibody against Fc IgG1 for elimination of anti-mouse antibodies | None |
| Poly MAK 33 (1368 338) | 0.5–50 μg/ml (50 μg/ml) | Polymeric monoclonal antibody against Fc and Fab IgG1 for elimination of anti-mouse antibodies | None |
| Poly MAK 2b/2a (368 338) | 0.5–50 μg/ml (50 μg/ml) | Polymeric monoclonal antibody against Fc and Fab IgG2 for elimination of anti-mouse antibodies | None |
| H-IgG/Fab-Poly (BMO 9000 041) | 0.5–500 μg/ml (500 μg/ml) | Polymeric polyclonal human IgG for elimination of anti-human IgG-directed interferences | None |
| B-IgG (1293 621) | 0.5–2.5 mg/ml (2.5 mg/ml) | Bovine IgG | None |
| Poly BSA I (1 539 302) | 0.1–50 mg/ml (5 mg/ml) | Elimination of endogenous nonspecified interferences | None |
| Poly BSA II (1 816 438) | 0.1–50 mg/ml (5 mg/ml) | Elimination of endogenous nonspecified interferences | None |
| Poly BPLA I (BMO 9000 045) | 0.1–50 mg/ml (10 mg/ml) | Bovine plasma albumin II for elimination of endogenous nonspecified interferences | None |
| Poly BPLA IV (BMO 9000 044) | 0.1–50 mg/ml (10 mg/ml) | Bovine plasma albumin II for elimination of endogenous nonspecified interferences | None |
| Poly Pod, inactivated (BMO 9000 026) | 0.5–100 μg/ml (tested up to a concn of 1 mg/ml) | Polymerized horseradish peroxidase | Reduction of nonspecificity at 0.5–1 mg/ml; however, no complete abolishment in some samples |
FIG. 1.
Reactivities of samples 1 to 16 from Arhangelsk (Russia) and sample Ang from a laboratory member by capture T. gondii FEIA. The same reaction conditions were employed throughout the experiments. (A) Reactivities of samples with anti-human IgG capture antibody. (B) Reactivities of samples with anti-human IgA capture antibody. (C to E) Reactivities of samples with anti-human IgM capture antibody. The conjugates and concentrations are indicated in Table 1. The samples were tested in the presence (solid bars) and in the absence (open bars) of T. gondii antigen. Sample 5 was not available for all experiments. The negative, borderline, low-positive, and positive T. gondii IgM controls are marked N, B, L, and P, respectively.
To determine the class of interfering antibodies, biotinylated polyclonal sheep anti-IgM antibody was replaced by anti-IgG and anti-IgA for IgG and IgA determinations, respectively. The concentration of all capture antibodies used was the same (2 μg/ml). These capture antibodies were used in the assay with specific F(ab′)2-S-AMSA-SMCC-HRP (Table 1) conjugate with or without antigen (see Fig. 1).
To inhibit nonspecific reactions different blockers from Boehringer (Mannheim, Germany) were tested (see Table 3). Also, heat-inactivated (63°C, 30 min) non-T. gondii mouse F(ab′)2 fragments were added to the sample diluent and to the antigen-conjugate mixture in concentrations ranging from 0.5 to 100 μg/ml.
To study the reactivity of samples with HRP or anti-T. gondii F(ab′)2 fragments, polystyrene microplates (Labsystems) were coated overnight at 4°C with HRP or with anti-T. gondii F(ab′)2 fragments in concentrations of 1.25, 2.5, and 5 μg/ml in 0.01 M PBS, pH 7.4. Plasma samples from false-positive Arhangelsk blood donors, false-positive serum from France, and T. gondii IgM-positive and -negative samples were added at concentrations of 2 and 10 μl/well. The reaction of heterophilic IgM antibodies to either HRP or F(ab′)2 was detected by polyclonal antibody against human IgM labelled with alkaline phosphatase.
Mechanisms of interference.
The isotypes of the interfering antibodies and the effects of conjugation on the intensity of nonspecific signals are presented in Fig. 1. The immunoreactivity of each sample is expressed as a double bar on the histogram, and the signals are compared to the respective signal from the borderline control. The histograms show that signals exceeding the borderline level are observed only for the reaction with anti-IgM-coated microplates (Fig. 1C to E). The Ang sample showed slight reactivity in the immunoblot assay and high reactivity by the T. gondii IgG indirect method (data not shown), and it was slightly reactive in the capture IgG assay. Thus, this sample represented a complex immunoresponse of being IgM true borderline-low positive and nonspecific (Fig. 1C, lane Ang). Treatment of all tested samples with sheep anti-human IgG antibody and subsequent centrifugation did not remove nonspecific responses (data not shown), suggesting that IgG was not involved in the reaction mechanism. As seen from the comparison of the reactivities of three anti-T. gondii specific conjugates (Fig. 1C to E), the conjugation procedure and fragmentation of antibody to Fab′ instead of F(ab′)2 had a clear effect on the reduction of nonspecific signals. Figure 1D shows that the nonspecific responses of all samples with F(ab′)2-S-AMSA-N-succinimidyl-3(2-pyridyldithio)propionate (SPDP)-HRP were lower than those with F(ab′)2-S-AMSA-SMCC-HRP (Fig. 1C), and at withdrawal of the antigen the majority, except samples 14 and 16, showed no response. Fragmentation of F(ab′)2 to Fab′ further improved the specificity, without affecting the signals from the true-positive samples. It is noteworthy that sample 14 also showed some reactivity with the Fab′-HRP conjugate in the absence of antigen. This sample was also reactive with other HRP conjugates (Table 1) and with free HRP (Table 2). Long-term persistence of interfering antibodies interacting with F(ab′)2-S-AMSA-SMCC-HRP was confirmed for six individuals (samples 2, 4, 6, 9, 10, and 14) from whom paired plasma specimens taken 1 year apart were available. For those individuals IgM nonspecific antibodies were detectable, with signals from the first and the second (1 year later) samples being equal (data not shown). The triple sandwich assay in which mouse monoclonal antibody was used intact with a subsequent detection via rabbit polyclonal antibody revealed lower but clear reactivities of samples 5, 8, 11, 14, and 15 (data not shown). As in previous experiments, sample 14 also showed reactivity in the absence of antigen. Reactivity with an HRP-labelled polyclonal rabbit antibody alone was observed only for sample 14.
TABLE 2.
Characterization of nonspecific samples
| Sample group | Sample(s) | Observations | Plausible interference mechanisms |
|---|---|---|---|
| I | 1, 2, 3, 4, 5, 6, 7, 9, 10, 12, 13, 15, 16 | Reactivity with HRP conjugates of different specificities and utilizing different conjugation procedures (see Table 1); microbial antigen is not needed to produce nonspecific signals; negative in immunoblot assay; negative in indirect IgM EIA; variable reactivity with coated HRP; abolishment or drastic reduction of signals when polymerized HRP is added; intensities of nonspecific responses are variable | Polyreactive (anti-immunoglobulin and anti-HRP) heterophilic antibodies of IgM class |
| II | 8, 11 | Positive in our test model in the presence of specific microbial antigen and in some commercial IgM capture methods; negative in immunoblot assay and in some other commercial IgM capture methods; negative in indirect IgM EIA | Polyreactive heterophilic IgM antibodies of very low affinity; some uncertainty remains if sample 8 is a true positive |
| III | 14 | Strong reactivity with all tested conjugates and with free HRP in the capture method; reaction with solid phase coated with HRP; drastic reduction of signals when polymerized HRP is added | Predominantly anti-HRP IgM class antibodies |
| IV | Ang | Positive in indirect T. gondii IgG and low positive in immunoblot assays; other characteristics as in group I | Specific anti-T. gondii IgM response combined with nonspecificity by polyreactive heterophilic antibodies |
Among commercial blockers tested (Table 3) only polyPOD (polymerized HRP) had a clear effect on the reduction of nonspecific reactions; however, complete abolishment of all nonspecific reactions was not achieved even at a concentration of 1 mg/ml (the range of recommended concentrations is from 0.5 to 100 μg/ml) (data not shown). The addition of heat-inactivated non-T. gondii F(ab′)2 fragments to the reaction mixture in concentrations up to 100 μg/ml had no effect on specific positive and false-positive reactions (data not shown). Addition to the sample diluent of anti-T. gondii F(ab′)2 fragments at a concentration of 100 μg/ml showed that no competition of these fragments with the corresponding HRP conjugate was observed in false-positive reactions (data not shown). A comparison of dilution curves for the true- and false-positive samples giving extremely high signals showed that the shapes of the dilution curves were similar, and thus a dilution approach cannot be used to discriminate the true- from the false-positive signals. Testing of commercial sera from patients with autoimmune diseases did not reveal nonspecificity (data not shown), providing further and stronger evidence that nucleolar, mitochondrial, histone, microsomal, and most importantly, ANA-RF autoantibodies were not involved in the described interference mechanisms. The reactivity patterns of nonspecific samples in a variety of test systems allowed us to combine these samples into groups (Table 2).
Our experiments showed that when interference is mediated through the bridging between captured antibody and a conjugate, the method of preparation of the latter has a clear effect on the reduction of nonspecific signals (cf. Fig 1C, D, and E). Lower nonspecific signals with the F(ab′)2-S-AMSA-SPDP-HRP conjugate (Fig. 1D) compared with those with the F(ab′)2-S-AMSA-SMCC-HRP conjugate (Fig. 1C) may be accidental for two reasons. First, HRP in both conjugate preparations was linked to amino groups of immunoglobulin in an unpredictable fashion. Second, the degree of conjugation as well as the presentation of immunoreactive epitopes on immunoglobulins conjugated by both chemistries was only a matter of chance. Conversely, in the conjugation of Fab′ to HRP (via SPDP), HRP is linked to the sulfhydryl group of the hinge in a predictable and reproducible way with a fairly constant Fab′/HRP ratio equal to one. Probably, reduction of the epitope density on the conjugate diminishes binding sites for heterophilic human IgM antibodies. The lower content of HRP per conjugate molecule probably also contributed to the lower false-positive reactions with Fab′-HRP conjugates, although true-positive samples produced responses that were as high as those with the F(ab′)2-HRP conjugates for the same concentration (per HRP) (compare reactivities of the borderline, low-positive, and positive control samples in Fig. 1). Because false-positive reactions were observed with a variety of HRP conjugates with different idiotopes and of different origins, it can be speculated that heterophilic antibodies bind to allotopes on the constant domain of Fab′ fragments. These allotopes might represent conserved amino acid sequences in immunoglobulins of different species. Nonreactivity of samples in the indirect EIA in which the polystyrene surface was coated with F(ab′)2 fragments does not exclude interactions through F(ab′)2 because the attachment of F(ab′)2 fragments to the polystyrene surface could have modified or covered epitopes. If it is assumed that heterophilic antibodies possessing multiple reactivities reacted through F(ab′)2, conjugation to HRP might result in the “stretching” of some amino acid sequences from the globular structure of immunoglobulin, thus making them “approachable.” Because nonspecific reactions were observed with conjugates employing different linkage chemistries, we assume that heterophilic IgM was not elicited against linkers.
Another mechanism of interference, clearly observed in this study, is through the binding to the HRP moiety of the conjugate. Full elimination or attenuation of false signals by polymerized HRP, indirect reaction with HRP-coated surface, and multiple reactions with HRP-containing conjugates suggests that some individuals have anti-HRP antibodies. All available samples tested for the presence of interfering antibodies showed no decline of the response after 1 year, indicating persistence of the antigenic stimulus. The individuals might become immunized by HRP, e.g., through the alimentary pathway. If the latter interference mechanism predominates, it becomes clear why the addition of mouse serum had practically no effect on the decrease of nonspecific reactions. The reactivity merely through HRP may also explain why in some nonspecific samples anti-T. gondii F(ab′)2 fragments added to the sample diluent in concentrations drastically (200-fold) exceeding those of the conjugate did not compete with the relevant conjugate.
Acknowledgments
We thank Roger Eaton from the New England Regional Newborn Screening Program, Massachusetts State Laboratory Institute, Jamaica Plain, Mass., and Eskild Petersen from the Department of Parasitology, Statens Serum Institut, Copenhagen, Denmark, for critical comments and valuable suggestions.
REFERENCES
- 1.Arvameas S. Natural autoantibodies: from “horror autotoxicus” to “gnothi seauton.”. Rev Immunol Today. 1991;12:154–159. doi: 10.1016/S0167-5699(05)80045-3. [DOI] [PubMed] [Google Scholar]
- 2.Arvameas S, Ternynck T. The natural autoantibodies system: between hypothesis and facts. Rev Mol Immunol. 1993;30:1133–1142. doi: 10.1016/0161-5890(93)90160-d. [DOI] [PubMed] [Google Scholar]
- 3.Briantais M-J, Grangeot-Keros L, Pillot J. Specificity and sensitivity of the IgM capture immunoassay: studies of possible factors including false positive or false negative results. J Virol Methods. 1984;9:15–26. doi: 10.1016/0166-0934(84)90079-x. [DOI] [PubMed] [Google Scholar]
- 4.Champsaur H, Fattal-German M, Arranhado R. Sensitivity and specificity of viral immunoglobulin M determination by indirect enzyme-linked immunosorbent assay. J Clin Microbiol. 1988;26:328–332. doi: 10.1128/jcm.26.2.328-332.1988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Eaton R B, Petersen E, Seppänen H, Tuuminen T. Multicenter evaluation of a fluorometric enzyme immunocapture assay to detect toxoplasma-specific immunoglobulin M in dried blood filter paper specimens from newborns. J Clin Microbiol. 1996;34:3147–3150. doi: 10.1128/jcm.34.12.3147-3150.1996. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Ishikawa E, Imagawa M, Hashida S, Yoshitake S, Hamaguchi Y, Ueno T. Enzyme-labeling of antibodies and their fragments for enzyme immunoassays and immunohistochemical staining. J Immunoassay. 1983;4:209–327. doi: 10.1080/15321818308057011. [DOI] [PubMed] [Google Scholar]
- 7.Meurman O. Detection of antiviral IgM antibodies and its problems—a review. Curr Top Microbiol Immunol. 1983;104:101–131. doi: 10.1007/978-3-642-68949-9_7. [DOI] [PubMed] [Google Scholar]
- 8.Naot Y N, Remington J S. An enzyme-linked immunosorbent assay for detection of IgM antibodies to Toxoplasma gondii: use for diagnosis of acute acquired toxoplasmosis. J Infect Dis. 1980;142:757–766. doi: 10.1093/infdis/142.5.757. [DOI] [PubMed] [Google Scholar]
- 9.Naot Y N, Desmonts G, Remington J S. IgM enzyme-linked immunosorbent assay test for the diagnosis of congenital toxoplasma infection. J Pediatr. 1981;98:32–36. doi: 10.1016/s0022-3476(81)80528-8. [DOI] [PubMed] [Google Scholar]
- 10.Potasman I, Araujo F G, Remington J S. Toxoplasma antigens recognized by naturally occurring human antibodies. J Clin Microbiol. 1986;24:1050–1054. doi: 10.1128/jcm.24.6.1050-1054.1986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Verhofstede C, Van Renterghem L, Plum J. Evaluation of immunoblotting for the detection of Toxoplasma gondii immunoglobulin M antibodies. Eur J Clin Microbiol Infect Dis. 1990;9:835–837. doi: 10.1007/BF01967387. [DOI] [PubMed] [Google Scholar]