Medical Device Link.

Originally Published IVD Technology November/December 2005

MOLECULAR DIAGNOSTICS

Approaches to autoimmune antibody profiling using protein arrays

Multiplexing technology holds the potential to become the standard for laboratory analyses.

Craig S. Hixson and Steven R. Binder

Scanning electron micrograph of magnetic microparticles used in the Bio-Rad Laboratories Inc. (Hercules, CA) BioPlex 2200 immunoassay system, an instrument with assay multiplexing capability.

Autoimmune diseases are chronic degenerative or inflammatory conditions that result from abnormal immune reactions. Different autoimmune diseases affect the body in different ways. For example, in multiple sclerosis the autoimmune reaction is directed against the central nervous system, while the intestinal tract is afflicted in Crohn’s disease. Further, the tissues and organs affected may vary among individuals with the same disease.

The severity of an autoimmune disease depends on the suffering individual’s immune system. Such diseases occur in at least 3% of the population, with women and the elderly affected disproportionately.¹ Inflammation is a common symptom with many of these diseases, as is dizziness, fatigue, malaise, and low-grade fever. Organ-specific autoimmune diseases are destructive toward the targeted organ or tissue, leading to impaired function.²

Autoimmune diseases can be difficult to diagnose, particularly early in their development. Testing of healthy individuals for specific autoantibodies indicates that the prevalence of a positive result ranges from near 0% to more than 10%, depending on the disease. Most of the antibodies found in healthy people are of low titer and of no consequence to their health.

Autoantibodies are not always specific for a single rheumatic condition. As an example, antibodies reactive with native double-stranded DNA (dsDNA) are often associated with a diagnosis of systemic lupus erythematosus (SLE). However, anti-dsDNA antibodies are also found in some individuals with rheumatoid arthritis, Sjögren’s syndrome, scleroderma, drug-induced lupus, chronic active hepatitis, Graves’ disease, and other conditions. Anti-dsDNA antibodies appear in people with these conditions with a frequency of less than 5% typically. The presence of anti-dsDNA alone therefore is not diagnostic of SLE.

In a few cases, autoantibodies are quite disease specific. Autoantibody targets for several of the more common rheumatic diseases have been identified (see Table I).

This article reviews the technology for autoantibody testing, beginning with established methods and ending with a consideration of future trends. The core and focus of the discussion is multiplexing protein analyses. The article examines a variety of multiplexed autoimmune assays and describes challenges facing developers of such assays.

Current Testing Methods

The first tests available for autoimmunity-related antibodies were based on Ouchterlony double diffusion in agar gel, but these have long been replaced by faster, more sensitive semiquantitative methods.

Detection of autoantibodies in serum is frequently performed by means of such laboratory tests as immunoassays (usually enzyme-based), indirect immunofluorescence microscopy (IFA), immunoblotting, and immunoprecipitation. Only the first of these is semiquantitative; the rest are qualitative. Screening by IFA is labor-intensive and requires visual interpretation of the microscopically obtained staining pattern by a trained technician. This method, which lacks reliable standardization, is dependent on the skill of the observer. An IFA positive result does not define the antigen. Therefore, it is of limited value in suggesting a specific treatment regimen.

The most significant limitation of these methods is that they provide only a single result for each analysis. Often, multiple tests are necessary in order to be able to report a complete repertoire of required autoantibody results.

The long-sought objective of simultaneously testing for analytes that are routinely grouped—autoantibodies, allergens, and thyroid function tests are examples—was achieved with the advent of multiplexing technology. When assays are multiplexed, a single sample can be used to produce many reportable results from the same reaction vessel. Multiplexing is a technology well suited for diagnosing disease that is multifactorial and whose diagnosis requires extensive laboratory testing. Assaying rationally defined multiple analytes in a profile mode—multiplexing—provides many advantages:

• Cost savings in terms of reagents, lab consumables, and, especially, labor.
• The acquisition of considerable information from a small patient specimen, which is important particularly in pediatrics.
• An ability to test simultaneously for such diverse analytes as nucleic acids, antigens, antibodies, and drugs.
• The inclusion of internal quality controls to ensure the accuracy of test results.
• Increased sample throughput.
• An ability to identify patterns of analyte concentrations.

In addition, multiplexing addresses such current industry and market trends as labor shortages, the rise of automation, process standardization, and laboratory consolidation.

Array Technology

A recently developed technology is the multiplexing of protein analyses. This array technology involves primarily two approaches, one planar and the other bead-based.

Figure 1. A schematic diagram of the layout of a two-dimensional array used in one study that spotted 14 autoantigens (A–N), positive (+) and negative (–) controls, and IgG calibrators increasing in concentration twofold from 1 mg/ml (C1) to 16 mg/ml (C5) in discrete locations.

Planar Microarrays. One technique employs a two-dimensional microchip that has defined reaction loci for individual analyses. Flat-surface microarrays allow binding assays of several immobilized proteins in a complex mixture of proteins, metabolites, and other molecules, with ligands present in the soluble phase.

Several years ago, a group of researchers constructed autoantigen arrays that contained 1152 reaction features.³ The planar arrays used poly-L-lysine-coated glass microscope slides as the solid support. Ordered arrays were generated by spotting 196 distinct putative autoantigens in either four- or eight-replicate sets. Proteins bound to the solid substrate included 36 recombinant or purified proteins, 6 nucleic acid–based antigens, and 154 overlapping and immunodominant synthetic peptides derived from small nuclear ribonucleoproteins, Smith antigen proteins, poly(ADP-ribose) polymerase, and H1, H2A, H3, and H4 histones.² Included in the array were tests for antibodies directed against phosphorylation-activatable modified proteins. Goat anti-human IgG labeled with the fluorochrome Cy3 was used as a common label for the arrayed tests.

Antibodies could be detected at nanogram-per-milliliter concentrations, and assays were linear over a three-log range. Analytical sensitivities were found to be generally better than for enzyme-linked immunosorbent assay (ELISA) analogs.

One technical problem that this study identified was that several Smith antigen and histone proteins were inactivated as a result of the spotting technique. Presumably, epitopes were lost as a result of the binding of the proteins to the solid surface. The researchers noted that this problem might be corrected through optimization of the binding chemistry or improvement in the surface coating. In addition, some antigens bound to glass slides are affected negatively by high evaporation rates.

Another set of investigators conducted a study in which they spotted 14 autoantigens in duplicate in polystyrene wells, along with controls and calibrators (see Figure 1).⁴ They used horseradish peroxidase–labeled anti-human IgG in conjunction with an activatable chemiluminescent substrate, and recorded light signals by means of a chip reader based on a charge-coupled device (CCD) camera. Antibodies were quantified using human IgG calibrators. This approach integrated the most commonly ordered clinical tests in a simple miniaturized format. The analytical results achieved with the plastic array were comparable to those obtained through use of commercial assay kit products.

Figure 2. Bead map of a 5 X 5 red/orange 25-bead-set array in the Luminex system. In this liquid bead-array map, individual bead sets have different amounts of red and/or infrared fluorescent dyes such that each is identifiable by a unique fluorescence signature.

The Randox Evidence is an automated platform designed for use in clinical laboratories and biomedical research. It uses a protein biochip array technology; detection is based upon light emissions produced by chemiluminescent reactions, which are detected by a CCD camera. The performance of this instrument for cytokine markets has been recently described.⁵ No reagents for autoantibody measurement are currently available.

Bead-Based Microarrays. The second method of multiplexing immunoassays for proteins employs a suspension of microsphere sets in which each set represents an individual analytical test. This system is sometimes referred to as a liquid array. The microspheres—plastic beads approximately 5 µm in size—are internally encoded by fluorescent dyes and individually addressed to a specific test in the multiplexed analysis. The beads are detected through the application of flow spectrofluorometry.

Autoimmune multiplexing with microsphere suspensions is generally based on technology developed at Luminex Corp. (Austin, TX). With the company’s xMAP technology, fluorophore-coded microbeads function as bar codes for individual analyses.6 Each bead set can be coated with a reagent specific to a particular bioassay, the analyte being defined by the bead set’s discrete red and near-infrared fluorescent-dye concentrations. This enables specific analytes from a sample to be captured and detected (see Figure 2). Within the analyzer, one laser excites the beads’ internal dyes, which identify each microsphere by type, while a second laser excites reporter dyes (most commonly B-phycoerythrin) captured during the assay. Tens to hundreds of readings are made for each bead set.

The Luminex 100 flow instrument classifies each microsphere according to its predefined map region by means of digital signal processing. The surface of each microsphere contains multiple carboxyl groups that serve as sites for the covalent attachment of proteins. Quantitation of the reporter on the bead surface is determined with the use of a series of calibrators.

Microspheres, with their small size and large surface area, allow for better reaction kinetics than do flat-surface, solid-phase arrays. Also, the mobility of microspheres in suspension compares favorably with the static planar-array surface in this respect. Success in distributing a sample evenly to all planar-array addresses is especially critical, but it is an irrelevant matter in connection with suspension arrays.

Several reports of the use of Luminex technology in the study of autoimmunity have been published. One investigation examined the simultaneous detection of five extractable nuclear autoantigens and showed excellent correlation to predicate single-test ELISA methods.⁷ Another group studied antinuclear antigens (ANAs) and extractable nuclear antigens in the sera of 37 women with Sjögren’s syndrome.⁸ They found antibodies to Sjögren’s syndrome antigen A and antigen B (SSA and SSB) in 84% and 76% of the sera, respectively.

The BioPlex 2200 ANA screen from Bio-Rad Laboratories Inc. (Hercules, CA) is a Luminex-based fully automated system that simultaneously determines levels for 13 different autoimmune antibodies. A clinical study showed that the specificity of this ANA screen is comparable to that of an ELISA ANA screening test.⁹ Like the ELISA, the BioPlex system has a lower positive rate than indirect immunofluorescence for autoantibody screening.

Advantages and Disadvantages. Both of the principal multiplexing approaches—suspension and planar arrays—have advantages and shortcomings. Two-dimensional microchip technology excels particularly in high-density screening (involving more than 100 tests) of proteins or nucleic acids. Planar chips are very well suited for the simultaneous analysis of an entire genome or proteome of an organism, from bacteria to human, in a single experiment. Microsphere arrays, for their part, are particularly advantageous in the clinical laboratory where tested analytes are well defined and are measured in a limited-size panel.

Microchips sometimes lack the reproducibility necessary for high-throughput clinical applications.¹⁰ This is attributed largely to the sequential production process planar arrays require for preparation. The high-volume production of microsphere lots, by contrast, allows for an assay reproducibility that flat arrays cannot as easily provide.

Also, planar arrays do not enjoy the same level of redundancy and statistical scrutiny as do microsphere arrays. This is because hundreds of microspheres of each assay set are detected in a typical multiplexed bead reaction, each of them representing an individual immunoassay, and outliers in the data set are excluded before the results are calculated.

Suspension arrays help the customized coupling of ligands to each bead set. Ligands bound to discrete bead sets can react with individualized binding chemistries and reaction conditions. Each ligand may be as distinct in purity as, for example, cell lysate or recombinant proteins. Unique linkers between the bead sets and the ligands for each bead set can be used. And finally, different postcoating, or blocking, procedures can be employed in the manufacture of each bead set.

Microbead arrays offer a notable advantage in that each microsphere set is manufactured separately in an optimal manner, with the various bead sets thereafter pooled to prepare the final multiplexed bead reagent.

The large number of analytes that can be accommodated in microarray testing allows the use of multiple internal controls to ensure expected assay system performance. A variety of controls might be included in a multiplexed analysis of autoantibodies (see Table II).

Table II. Types of beads for internal controls that might be included in a multiplexed array for autoimmune antibodies.

Relative fluorescence is the usual signal output for microarrayed methods. A fluorescent internal standard (a discrete bead set or microchip address) can be used to normalize all of the assay signals. Normalization compensates for fluctuations in the illumination and detection systems. Other internal testing for autoimmune assays could be designed for the proper type of clinical specimen (serum, for example) and proper volume of dispensed sample.¹¹ Such internal controls should be a feature of every multiassay.

Other internal controls that could also be considered in multiplex testing are specific for one analyte or a family of analytes. For instance, an internal assay for rheumatoid factor (RF) could test for the potential interference by RF in IgM autoantibody analyses. A test for total IgG should prove useful for ensuring that IgG hypergammaglobulinemic specimens do not interfere in an IgM analysis owing to competitive inhibition.

The appropriate employment of internal controls requires an understanding of the technical issues related to the analyte, the matrix in which the assay is performed, and the performance characteristics of the instrumentation employed. Their use makes possible greater assay precision and accuracy, and the rejection of otherwise-unknown discrepant results.¹⁰

Challenges

Figure 3. Comparison of antibody indices for five representative tests in a panel of 16 analytes in otherwise-identical uniplex and multiplex assays. Analytes were measured in duplicate using a multicomponent positive control with either a single-bead (blue) or 16-bead (red) set. The antibody index is the fluorescence reading for the specimen divided by the fluorescence reading for the calibrator.

To become widely accepted, multiplexed autoimmune assays must overcome a number of technical hurdles. Protein-based microarrays are more challenging to develop than their nucleic acid analogs. This is a result of proteins being very heterogeneous in size, charge, shape, hydrophobicity, type and degree of posttranslational modification, and quaternary structure. And some proteins have great difficulty maintaining specificity and activity when tethered to a solid phase.

The simultaneous testing of autoantibodies, as with all immunoassays, requires well-defined, reproducible, and scalable antigens. With arrays for autoimmune antigens, challenges arise. These involve obtaining functional protein expression for array construction, developing methods to couple and maintain active protein on surfaces, and achieving both analytical sensitivity and a wide dynamic range with the chemistry and detection systems. Assay standardization, data interpretation, and storage can also introduce issues. And performance validation is required for each analyte. Furthermore, proteome chemistry is complicated by the existence of frequent and varied posttranslational modifications.

Reproducibility. The multiplexing of analyses can raise an impediment to satisfactory reproducibility of assay results. Precision is achieved through extensive optimization of all reagents and processes, as well as by providing redundant testing. With planar arrays, for each analyte two or more replicates may be tested, statistically tested for variance acceptability, and averaged. With bead suspension arrays, several hundred individual beads, or assays, for each analyte typically are analyzed. There is usually a direct relationship between the number of beads interrogated and the precision of the assay.

Crosstalk. Crosstalk between addresses in arrays can be a serious problem. This signal interference can occur with planar arrays, but it is theoretically more problematic with a microbead array. In these microsphere suspension arrays, assay reaction centers do not separate into discrete spatial addresses; the individual bead sets have the opportunity to interact for a long period in the liquid phase before they are put to use.

Proteins reacted with microspheres are coupled predominantly via covalent bonds, but, commonly, some material is loosely bound even after thorough washing. Noncovalently attached protein bound to a bead set can slowly dissociate from the solid surface and competitively inhibit the reaction desired of a liquid-phase reactant; the result may be falsely negative. The dissociated material may also migrate to another bead set and bind to its surface. This may lead to a false-positive result.

Thorough stability studies should be conducted to ensure that ligand-bound proteins are bound stably to their solid-phase substrate for the shelf life of the product. During the multiassay validation process, the results of multiplexed analyses must be found to be comparable to their respective uniplex results. Ideally, the stability study should be performed at the end of the claimed shelf-life period, when the problem is most likely to be manifest. Only when the multiplexed and uniplexed results for all analytes are nearly equivalent can an investigator conclude that no interference has resulted from mixing the bead sets (see Figure 3).

The number of possible interactions in a multiplexed assay is a function of the number of assays in a microsphere array (see Table III). The potential for unwanted interactions between analyses multiplies rapidly with increased array complexity. Such interference problems must be remedied prior to manufacturing.

Sensitivity and Specificity. Achieving clinical sensitivity and specificity with autoantibody arrays can be a daunting challenge. Because autoimmune antigens often are poorly characterized, it is highly difficult to obtain equivalent results between methods for which the key reaction components are unequal in terms of purity, concentration, and activity. Autoantibody assays are often conducted under conditions of nonequilibrium. Differences in assay kinetics and thermodynamics between different assay methods can lead to different assay results, particularly for antibodies that are intermediate in concentration.

Multiplexed assays by their nature provide a plethora of information. The most valuable feature of an array for autoantibodies is the generation of a pattern of self-antibodies. It is desirable that the results of a multiplexed array be analyzed with the aid of computer software that uses pattern recognition to make a differential disease diagnosis. Such software may improve diagnostic accuracy and also possibly throughput and cost efficiency. In addition, computer programs may discover complex autoantibody patterns that have not yet been recognized.

Many approaches to the construction of expert systems and interpretative algorithms exist. An appealing feature of pattern recognition technology is its independence of assumptions and of any need for a subjective human expert to interpret each result. Pattern recognition does not require that the analytes being measured be named or characterized.

Broad-Spectrum Approaches

In a study investigating whether an autoantibody pattern could predict resistance or susceptibility to the development of type 1 diabetes in mice, a panel of 266 different antigens were spotted onto glass chips. The antigens included peptides derived from heat-shock proteins; tissue antigens; immune system components; structural antigens; hormones; enzymes; plasma proteins; synthetic oligonucleotides; and antigens obtained from bacteria. From the original panel of 266 antigens, a combination of 27 antigens was found to provide a differential profile distinguishing mice resistant to a chemically accelerated form of diabetes (cyclophosphamide-accelerated diabetes; CAD) from those susceptible to it.¹²

Table III. The number of potential interactions in a multiplexed assay as a function of the number of assays in the array.

The investigators found that mice susceptible to CAD could be distinguished from resistant mice by their patterns of autoantibodies, even before the acceleration chemical was administered. Also discovered was a pattern of autoantibodies characteristic of healthy versus diabetic mice post-CAD; this pattern was distinct from that found for the pre-CAD antigen set. IgG reactivities to some antigens therefore may be predictive of future susceptibility to CAD, but not CAD itself, once diabetes emerges. On the other hand, some IgG reactivities may signify the disease but not susceptibility. Thus, prediction of future disease and diagnosis of present disease may depend on different titer patterns of autoantibodies.

This study illustrates broad-spectrum approaches to defining autoantibody profiles in various rheumatic diseases. In the future, novel antibody patterns may be found that are predictive of the course of autoimmune diseases or suggestive of a preferred course of therapy.

Future Applications

Multiplexing with autoantibody arrays promises several important future research and clinical applications. The most obvious is screening with a large number of antigens potentially reflective of a specific autoimmune disease for the purposes of diagnosis and treatment. Other possible future uses are the determination of the involved human IgG isotypes in assessing potential pathogenicity, the characterization of epitope spreading of autoimmunity responses, and direction of the course of specific antigen-pattern-defined therapy.3 Multiplexed assays will be particularly useful in the discovery of new autoantigens in as-yet-unrecognized autoimmune disease or subsets of diseases.

They will find use in other areas of medicine as well, including analyses of allergens, therapeutic and abused drugs, serologic antibodies, and cardiac disease and cancer protein markers.

Craig S. Hixson, PhD, (left) is a research manager and Steven R. Binder is director of technology development in the clinical diagnostic group at Bio-Rad Laboratories Inc. (Hercules, CA). The authors can be reached at craig_hixson@bio-rad.com and steve_ binder@bio-rad.com, respectively.

High-numeric microarrays—those involving more than 100 analytes—will prove to be most useful in the research setting. Multiplexing of hundreds or thousands of tests allows a researcher to identify a limited constellation of autoimmunity biomarkers out of a large number that are clinically irrelevant for a given disease. After the discovery phase, proteins in much more limited array densities of fewer than 100 analytes will be used in the clinical arena.

Conclusion

Full automation of multiplexed assays for autoantibodies and other analytes is now available, with one computer-controlled immunoassay open-system analyzer offering random and continuous access to specimens and panels of reagents, full integration with laboratory information systems, primary-tube sampling, and STAT capability The technical, economic, and work-flow advantages of such sophisticated products are so great that devices based on this new technology will offer an attractive alternative to traditional ELISA, immunoblot, and IFA techniques.

References

Copyright ©2005 IVD Technology