Optimisation of assays: Interference in immunoassays - Recognize and avoid
Dr. Peter Rauch, Angela Zellmer, CANDOR Bioscience GmbH;
Dipl.-Chem. Nico Dankbar, Institute of analytical chemistry, university of Münster;
Dr. Christoph Specht, PARA Bioscience GmbH;
Dr. Detlef Sperling, MACHEREY-NAGEL GmbH & Co. KG
Immunoassays are analytical laboratory methods for the detection of different substances using antibodies. They are used commonly in bioanalytical and biochemical laboratories of the Life Science industry. The methods are Enzyme-linked immunosorbent assays (ELISA) and Enzyme immunoassays (EIA), Western blotting, Radioimmunoassays (RIA), Protein-Arrays, Immunohistochemistry or also the immunopolymerase chain reaction (Immuno-PCR). In all kinds of immunoassays the detection of the analytes can be influenced by interference. Very frequently cross reactivities, unspecific binding and matrix effects occur. Interfering substances are present in more or less significant concentrations in real specimen and interact with the analytes or with the capture respectively the detector antibodies directly. By application of a new buffer (LowCross-Buffer®) instead of the usually used buffers most interference effects can be avoided. Through this exchange the quality of the assays and the efficiency of the assay development is improved.
Key Words: LowCross-Buffer®, antibodies, interference factors in immunoassays, ELISA, Western blotting, protein arrays, immunohistochemistry, HAMA reactivity, rheumatoid factors
Niels Kaj Jerne, who received in 1984 the medicine Nobel Prize for his work on the specific construction and the control of the immune system, mentioned in his award lecture that every antibody is multi-specific. He related this statement to the antibodies building during the early phase of an immune response
1,2. Antibodies apparently characterized well with a high affinity to the target analytes show occasionally surprising results: During immunological detection on western blotting membranes unwanted bands are stained, on protein arrays one gets fluorescence signals for spots at the wrong positions from immobilized capture antibodies as well as a high background signal for blank samples. For ELISAs one gets a high background for the negative control or false negative signals in measurements. Misstatements due to the interference effects can lead to subsequent costs and also to false diagnosis3. All immunoassays are characterized by a binding reaction between the target analyte and antibody. The problem of these methods, which is solved only insufficiently till now, is the regularly occurring interferences which lead to faulty measurements. Typical interferences are unspecific binding which lead to a poor signal to noise ratio and a high background, cross reactivities and matrix effects. Expressed in simplified terms, most of these effects are based on a direct interaction of the analytes, the capture antibody or the detector antibody with external substances or surfaces. A simplified scheme of typical interference effects is represented in illustration 1. Known interfering factors, which are regularly described in the technical literature, are for example heterophilic antibodies as well as HAMAs (humane anti mouse antibody)2 or rheumatism factors, albumins, complement factors, lysozyme4 and others.
Interference by immunoassay label
It is common for immunoassays to mark the detector antibody - in the case of competitive assays: the standard analyte - with a label. Frequently used labels are enzymes (very often alkaline phosphatase or (horseradish)-peroxidase), fluorescence dyes, radioactive isotopes or also DNA (for immuno-PCR.) The unwanted effects appear here too. The danger, using fluorescence dyes as a label, is often that hydrophobic dyes change the binding qualities of the detector antibody causing undesirable binding by the dye itself and reducing the solubility of the labelled protein. In addition the antigen to antibody binding can become weaker5. For example these unspecific effects can cause increasing binding of the antibody to the surfaces (fig. 1 A and B), to extrinsic proteins in the sample (fig. 1 C) or to the capture antibody (fig. 1 D).In these cases false-positive measurement in absence of the analyte will occur or the entire assay suffers under a high background. For protein arrays increased background fluorescence for single spots is observed or altogether the signal-to-noise ratio deteriorates. Likewise proteins or antibodies from serum samples can bind to fluorescence dyes and reduce or even switch off the fluorescence of the dye. Therefore some researchers already recommend to abandon fluorescence dyes as a label for protein arrays or to choose other labels respectively6. The reactions on protein arrays are very complex since a variety of different capture and labelled detector antibodies are used simultaneously in one reaction volume. Therefore the probability of unspecific binding of proteins from the sample or labelled antibodies with single spots as well as the danger of interference effects of components from the sample with the antibodies is increased7.
Interference by cross reactions
Cross reactions are the ability of the antibody to bind also to other structures than the target analytes (fig. 1, I-K). Frequently these structures show a great similarity to the analytes. Examples therefore are metabolites or chemical substances with a similar molecular structure. Also proteins with a coincidental similarity or with homologies of the amino acid sequence can cross react. Especially in competitive assays cross reactions play a greater role since only one antibody is applied3,4. Very often possible cross reacting substances and the cross reactivity need to be quantified for validation of these kinds of assays8. Cross reactivities can play also a significant role at the detection of proteins on western blotting or at immunohistochemical applications. This becomes apparent in the staining of additional bands or cell structures, although one doesn't know the exact molecular causes for these unwanted bindings in every case. In Western blotting one assumes in many cases simply degradation products of the right protein, which result naturally or from the methodical approach. But in some cases that's not the truth and one has to consider cross reactivities of the primary or secondary antibodies.
Interference by unspecific binding
Closely related to the cross reactivities are unspecific or non-specific bindings. However the causes on a molecular level differ. In daily lab routine these differences are rarely noticeable. One deals with a cross reactivity, if the cross reactand is known and its cross reactivity can be measured for example with a competing concentration of the cross reactand8. In the case of unspecific binding the binding deals with substances, which are in far excess of the target analyte (i.e. unspecific binding to albumines or immuno globulines) or with surfaces (i.e. surfaces of ELISA-wells or of Western blotting membranes) or with spots of immobilised antibodies in protein arrays7.
Matrix effects are the sum of all interference effects of all components, which appear in a specimen and influence the measurement of a target analyte9. If the true molecular cause for an influence is not determined, but one knows that it comes from the specimen, then you call it in general a "matrix effect". The borders from one kind of effect to the other kind are floating and sometimes unclear. Some matrix effects are derived from "anti-animal-antibodies" others from heterophilic antibodies, from endogenous interferers or just from viscosity, pH-value or simply the salt concentration.
Interference by "anti-animal-antibodies"
Human anti-animal-antibodies (HAAA) can be of the IgG-, IgA, IgM- or IgE-type. They are part of the immunesystem's answer to contacts with immuneglobulines from animal origin. HAAAs are well known interferers in many diagnostical immunoassays and can be part of sometimes 80% of all clinical specimen – depending on the study cited. The concentrations of HAAAs achieve ranges up to milligrams per milliliter10. Human-anti-mouse-antibodies (HAMA) are the most popular under the interfering antibodies in immunoassays. HAMAs are human antibodies, binding with significant specifity and sometimes noticeable affinity mouse-antibodies. Reason for the build-up of these antibodies in patients are often therapeutic antibodies, which are given as drugs for cancer therapies. After medication the immune system of patient reacts to these extrinsic antibodies by forming own antibodies against the therapeutic mouse-antibodies. HAMAs therefore interfere with immunological assays, which make use of mouse-antibodies as assay-reagents. In sandwich assays with mouse monoclonals this can result in a direct brigde binding between capture and detector without any analyte (fig. 1 F). This leads to false-positive signals. There are similarities in the sequences between antibodies derived from different species. That means, that HAMA-containing sera can also make problems in assays which make use of antibodies derived from other species than mouse with the same results of false signals. HAMAs are not only derived from antibody-drug therapies. Long exposure to domestic animals and pets can also end up in the formation of antibodies against these animals. One has found in patients sera antibodies against rabbits, mouse, dogs, hamsters just to name a few. These anti-animal-antibodies interfere with some assay-antibodies with different affinities and different kinds of problems. Some interfering antibodies are not only directed against the Fc-fragment, but also against the Fab-fragments of assay-antibodies. This can result in a reduction or total hindrance of correct binding, resulting in false negative signals (fig. 1 G and H). If the HAAAs bind to the Fc-fragment they are called anti-isotypical interferers. If they bind to the highly variable Fab-fragment they are called anti-idiotypical interferers10.
Interference by heterophilic antibodies
Taber's Medical Dictionary defines heterophilic antibodies as „antibodies which bind other antigens than the specific antigen"11. Heterophilic antibodies can be of the IgG, IgA, IgM or IgE type. Particularly the IgM type plays a particular role in sera of rheumatic patients. These sera contain so-called rheumatism factors in high concentration. Rheumatism factors are IgM antibodies that bind to the Fc sections of humane antibodies and who can bind therefore species independent also to Fc sections of the antibodies used in the assay. Therefore rheumatic sera connect capture with detector antibodies with the consequence of false positive signals. This is at the same time the general interfering mechanism of the heterophilic antibodies. The effect of the rheumatoid sera resembles the effect of the HAAAs. The difference compared with the HAAAs is the origin of the heterophilic antibodies: These are not build on contact with animal immune globulins, but they are multi-specific antibodies of the early immune response or interfering antibodies with unknown immunological history of origins2, Interference by HAAAs or by heterophilic antibodies are known already for more than 30 years. The interfering antibodies are general weakly binding antibodies2 from animal origin and disturb predominantly assays that - due to the low concentration of the analytes - need to get along with a low dilution of serum or plasma samples12. Additions of blocking substances to the sample buffer – commonly non-specific sera, antibody fragments or high concentrations of animal immune globulins – are able to reduce the interference effects of the HAAAs or heterophilic antibodies by competition but don't circumvent at all times10.
Interference by endogenous components of the specimen
Even naturally occurring proteins of specimen can interfere with immunoassays. Well known interfering substances in human sera are for example albumins, complement factors, lysozymes and fibrinogen4. Analytes of low molecular weight can bind to albumin. This makes the accessibility of an antibody to the analyte difficult. Numerous hormones are bound to transportation proteins, what can lead to difficulties. Moreover, many proteins have the ability to bind other substances and proteins of course. This binding ability is often a substantial part of the biological function of the respective protein. Albumin, complement and C-reactive protein (CRP) are natural receptors for many substances. Unspecific binding or even cross reactivity – like with antibodies – are therefore possible, which complicates the recognition of certain analytes in an assay. Endogenous proteins can bind as an interfering factor to antibodies (fig. 1 C, I-K) or mask the target analyte (fig. 1 L). For example lysozymes bind non-specifically to proteins with a low isoelectric point. Therefore antibodies which have an isoelectric point of about 5 can be bound and build a bridge formation between capture and detector antibody4. An important aspect, which should be mentioned, is the interference by strongly lipid containing specimen, because some analytes are fat-soluble and as the case may be the binding between antibody and analyte can be affected by lipids.
The Hook effect
The Hook effect leads to false negative determinations which, however unlike the other interference effects, do not arise through interactions with interfering factors. A Hook effect can appear in assays, where the specimen is mixed directly with the assay antibodies and the analyte appears in very high concentrations. In this case the high concentration of the analyte, when it exceeds the concentration of the assay antibodies, capture and detector antibodies can be saturated. Thus high concentrations simulate far lower concentration in the assay and lead to a significant underestimation of the true concentration in the assay4. The Hook effect can be avoided in the practice by using higher concentrations of the assay antibodies or by dilution of the specimen. Alternatively a systematic dilution of the test has to assure, that the measured value is not subject to a Hook effect. Known clinical parameters, which can be subject to a Hook effect, are CRP, AFP, CA 125, PSA, Ferritin, Prolactin and TSH for example4.
Prevention of interference by LowCross-Buffer®
The reasons for the described interference effects are similar. There are unwanted low up to middle affine interactions of the interfering factors with the antibodies or the analytes. And there are low or middle affine binding of labeled antibodies to other proteins or surfaces as well as low to middle affine cross reactivities of the antibodies to structure related substances. These interference effects have something in common which the newly developed LowCross-Buffer® makes use of: The interference reaction is weaker than the specific binding of the real analytes. Of course there are rarely exceptions since very high affine cross reactivities can occur that achieve the same qualities as the real specific binding. In such cases one then must speak about a specific binding and one has in principle an antibody which is aimed at two different substances. Therefore such an antibody is not usable at all for specific assays. The LowCross-Buffer® was developed specifically to eliminate generally weak and middle affine binding, but not negatively affecting high affinity binding with high specificity in any way. Figures 2 to 5 show different examples of typical interference effects in immunoassays that are prevented by the use of LowCross-Buffer®. Figure 2 shows a protein array application at which LowCross-Buffer® reduces a high background and improves the signal to noise ratio from 3.4 to 17.3. In this experiment different polyclonal anti EPIL antibodies (EPIL = early placenta insulin like growth factor) were tested for their suitability.
The cleaned antibodies were immobilized on aminosilane-functionalised micro-array slides by means of a spotter (GMS 417) with a concentration of 500 μg/ml and a volume of 1.8 nl/spot. Afterwards 2 ml of medium of an EPIL-overexpressing cell line (SKBR3) were mixed with the dye Oyster-650 (Luminartis GmbH) and all proteins in this set were labeled. The incubation on the slide was carried out at a dilution of the medium in a relationship of 1:20 with LowCross-Buffer® in comparison with PBS (medium:buffer). After washing of the slides these were read with a fluorescence-scanner (GMS 418) and the data were evaluated with ImaGene (Biodiscovery Inc.). By the use of LowCross-Buffer® a clear reduction of the background signal could be reached which made the distinction of single antibodies regarding their suitability for the EPIL possible. Figure 3 shows the immunological detection of a western blotting of the cytokeratines 4, 5 and 6 from liver cells and from HeLa cells, where further bands than the proper bands were detected by cross reactivities in combination with unspecific binding. The protein samples were separated electrophoretically on a 12.5% SDS-poly acrylamide gel and then blotted on a nitro cellulose membrane. The immunoblot detection for cytokeratines in the protein samples was carried out with a modified standard protocol13 as well as with LowCross-Buffer. Bound anti-cytokeratin antibodies (polyclonal rabbit anti cytokeratin, Biomeda, Foster City, USA), diluted 1:2500 in LowCross-Buffer® or in TTBS (Tween-Tris buffered saline), were detected with an alkaline phosphatase coupled to a secondary antibody (Goat anti rabbit IgG, abcam, Cambridge, UK) (diluted 1:1250 in Low Cross-Buffer® or in TTBS), detected and visualized by means of BCIP/NBT. It is very difficult to uncover the exact molecular reason for the unwanted binding reaction, which made detection without LowCross-Buffer® impossible. Possibly cross reactivities are visible due to other cytokeratines and their cleavage products or unspecific binding with different proteins from the cell disruption of the liver or the HeLa cells takes place. By simple exchange from the TTBS buffer, where the primary and secondary antibodies were diluted, to LowCross-Buffer®, the unwanted binding could be reduced to the extend that cytokeratines are stained exclusively in the correct molecular weight area of 56 to 60 kDa. An example how a matrix effect can impact an ELISA is shown in illustration 4. With this model assay (developed by CANDOR Bioscience) a matrix effect was induced systematically. Rabbit serum was used as a matrix and spiked in defined concentrations with a humane C reactive protein (CRP, Biotrend). As capture antibody Clone C2 (Biotrend, 1 μg/ml Coating concentration in PBS), as a detector biotinylated Clone C6 (Biotrend, concentration 2 μg/ml) was used. The spiked serum samples were diluted either with a PBS-BSA buffer or with LowCross-Buffer® 1:2 and measured by ELISA. Detection was carried out via NeutrAvidinTM-Horseradish peroxidase conjugated (Pierce, concentration 0.05 μg/ml in PBS-BSA buffer) with ImmunoPure® TMB-substrate (Pierce). A matrix effect, whose exact molecular reason is not known, leads to a calibration curve with a bad sensitivity. Due to it's physiological function CRP is able to bind many proteins and substances (scavenger function of the CRP) which probably reduces significantly the accessibility of the epitope. Presumably an interference effect takes place as shown schematically in figure 1 (L) although interference effects as shown in figure 1 (I-K) cannot be excluded. LowCross-Buffer® prevents the binding of the CRP to endogenous substances of the rabbit serum and thus improves the sensitivity of the calibration curve by the factor of 3. Figure 5 is an example of an ELISA against immune globuline from guinea pig (developed by PARA Bioscience, Gronau) which is used for immune toxicological studies in guinea pigs. In this assay false positive bindings in the specifity control (row of A1-A12) as well as at the blank value (H1-H12) spoiled the interpretation and evaluation. The use of LowCross-Buffer® prevented the false positive signals and moreover made the detection of the concentrations possible in the rows of B to G 1-6 or B to G 7-12. Goat-anti-guinea-pig IgG F(ab‘)2 as a capture antibody and goat-anti-guinea pig IgG (Fcγ ) biotinylated as a detector (both Jackson Immunoresearch Laboratories Inc., concentration range of 0.31 to 10 μg each/ml in PBS) were used. The guinea pig IgG was either diluted in LowCross-Buffer® or PBS (range 1-6 50 ng/ml, range 7-12 10 ng/ml). PBS-BSA buffer was used as a blocking buffer. The detection was carried out with Streptavidin-peroxidase (Sigma) and ortho-Phenylendiamine (Sigma).
The phenomenon of interference in immunoassays is as old as the methodical use of antibodies for bioanalytical and diagnostical purposes. Numerous molecular causes were found in the course of the last 30 years and their interference mechanisms were evaluated which led to the development of prevention strategies. At today's level of the technological development many interference effects can be minimized and LowCross-Buffer® makes an essential contribution to it. It's new, that different interference effects with different molecular principles can be minimized with the same strategy. LowCross-Buffer is applicable for different immunoassays. The results shown here are a clipping from different interference effects in different methods, which could be minimized or avoided with LowCross-Buffer®. Furthermore LowCross-Buffer® could prevent interference effects by HAMAs and rheumatism factors, unspecific bindings in immunohistochemical applications as well as false positive bands in immuno-PCR. Altogether, the time consuming and costly effort for optimization strategies can be reduced and simplified significantly whereas improvements in reliability are reached simultaneously.
 Jerne, N.K., Science 229 (1985), 1057-1059
 Kaplan, I.V., Levinson, S.S., Clinical Chemistry 45:5 (1999), 616-618
 Miller, J.J., Clinical Laboratory International 28, 2 (2004), 14-17
 Selby, C., Ann Clin Biochem 36 (1999), 704-721
 Patton, W.F., Electrophoresis 21 (2000), 1123-1144
 MacBeath, G., Nat. Genet. 32 (2002), 526-532
 Kusnezow, W., Hoheisel, J.D., J. Mol. Recognit. 16 (2003), 165-176
 Miller, J.J., Valdes, R.Jr., J Clin Immunoassays 15 (1992), 97-107
 Wood, W.G., Scand. J. Clin. Lab. Invest. Suppl. 205 (1991), 105-112
 Kricka, L.J., Clinical Chemistry 45:7 (1999), 942-956
 Span, P.N., Grebenchtchikov N., Geurts-Moespot, J., Sweep, C.G.J., Clinical Chemistry 49:10 (2003), 1708-1709
 Gallagher, S., Immunoblot Detection, in: Current Protocols in Protein Science, John E. C. et al., eds., 1995, John Wiley & Sons
Dr. Peter Rauch
CANDOR Bioscience GmbH,