Medical Device Link.

Originally Published IVD Technology March 2003

Assay Development

Developing microscale immunoassays for parallel analysis of multiple analytes.

John L. Tonkinson, Dan S. Osborn, Brett A. Stillman, and Weiwen Zhao

Faster, smaller, and more powerful. This mantra is applicable not only to microprocessors and the high-tech industry, but also to the in vitro diagnostics and biotechnology industries. For example, during the past five years, the DNA microarray has evolved from being a mysterious beast used only by technology-driven labs into a technique used as routinely as any high-throughput system.^1,2 

Point-of-care diagnostic tests such as lateral-flow pregnancy tests, drugs of abuse tests, and infectious-disease tests have also continued to become cheaper and more powerful. Along these same lines, microscale protein immunoassays represent a common ground between high-throughput research tools, such as microarrays, and the diagnostics industry’s desire to produce more-efficient tests.

The theory behind microscale assays did not begin with DNA microarrays. More than a dozen years ago, researchers first postulated that reformatting standard enzyme-linked immunosorbent assays (ELISA) and radioimmunoassay-type assays by using microspots would result in increased sensitivity, dynamic range, and the opportunity to conduct parallel analysis of antigens.³

The first practical use of microspot technology eventually came in the form of DNA microarrays for the basic research and genomics arena.¹ The vast amount of information generated by sequencing the genome as well as concomitant developments in manufacturing and analyzing microassays gave researchers the opportunity to look at the expression levels of thousands of genes simultaneously.

However, for a number of reasons, the development of microscale protein immunoassays has lagged behind DNA-based assays. One major reason is the relative complexity and diversity of protein binding reactions compared with nucleic acid hybridization reactions. Each protein binding reaction is unique, involving discrete chemical entities, while hybridization reactions involve the same chemical components that differ only in sequence. Thus, creating a single platform to study multiple reactions is much more difficult for proteins than nucleic acids.

It is also acknowledged that, depending on the application, protein arrays may differ from DNA arrays by having a smaller number of

 elements. In addition, quantitative labeling of complex protein mixtures is much more difficult than labeling complex nucleic acid mixtures. For DNA arrays, the probe is labeled during a reverse transcription reaction, so it is relatively easy under those conditions to ensure quantitative labeling of all species in the mix. Labeling of complex protein mixtures, while chemically trivial, remains difficult to control. 

Combined with the commercial availability of antibody pairs for a large number of relevant antigens, these factors have made it likely that commercial protein microarrays will emerge as a sandwich-based format similar to ELISA. The use of antibody pairs in this format obviates the need to label complex protein mixtures, and these protein microspot assays may be similar to DNA microarrays in spot size (150 mm diam) and detection methodology (e.g., fluorescence). This article examines efforts to develop multianalyte microimmunoassays, some of which have resulted in commercial products.

Protein Microarray 
Literature Review

The term protein microarray encompasses a large number of different assays. The basic research literature has several reports of early attempts to use protein microarrays for expression analysis, similar to how DNA expression arrays are commonly used.^4,5 These efforts are in their infancy, with many focused on developing model systems. Perhaps the biggest hurdle to overcome in this type of analysis is the efficient and consistent labeling of complex protein mixtures. An equally difficult issue is obtaining the thousands of binding specificities that need to be arrayed in order to make an expression analysis experiment meaningful.

Figure 1. Digital images showing quantitative retention of an arrayed antibody on nitrocellulose.  (Click to enlarge)

Other protein microarrays that have been discussed in the literature are cell or tissue arrays.⁶ These arrays have been viewed as potential diagnostic and prognostic tools that are derived from common techniques such as flow cytometry (e.g., immunophenotyping) and immunohistochemistry. The primary advantage of a microarray format for these tests is throughput, such that multiple patient or cell culture samples can be tested simultaneously. This allows easier correlations with specific drug treatments or disease states by using less sample than traditional methods.

From a clinical and research diagnostics standpoint, it is perhaps easiest to see traditional ELISA tests as the most ideal method for developing protein microspot assays. In an ELISA, one specificity is immobilized, and the other is used as a detection element in a solution. Thus, it is easy to conceptualize how this format could be miniaturized, allowing multiple specificities to be arrayed in microspots and then using a cocktail of detection specificities to identify which of the capture antibodies has in fact captured an antigen.

Figure 2. Log-transformed data from a quantitative retention of an arrayed antibody.  (Click to enlarge)

Much of this conceptualization has already been applied in developing lateral-flow immunoassays. In lateral-flow tests, the capture antibody is immobilized on a nitrocellulose membrane. The analyte interacts with an antibody conjugated to a labeled microparticle, and this complex flows with the mobile phase through the nitrocellulose until it encounters the capture antibody. The microspot concept merely expands the number of reactions occurring simultaneously.

Several reports have also been published detailing the use of antibody pairs to develop antibody arrays.^7–11 Although many of these arrays have been macroscale arrays performed on membranes and visualized by chemiluminescence, they demonstrate the possibility of combining multiple immunological reactions. In addition, almost all of these assays were developed to study cytokines and growth factor levels. The primary reason for this focus on cytokines is the relevance of cytokines to basic research in disease states such as cancer as well as clinical and pharmaceutical research. Another reason is the availability of quality antibody pairs for these antigens, which makes the transition from a bulk assay to an array format easy to conceptualize.

Microspot Immunoassay Theory

The traditional ELISA sandwich assay, in which one antibody or specificity is immobilized in bulk in a polystyrene well and a second antibody or specificity is used to detect the antigen bound to the first one, is a powerful tool. Using two antibodies introduces a high degree of specificity, provides a simple method of separating bound antigens from free ones, and enables the output signal to be directly proportional to the antigen concentration, meaning that the output can be compared linearly to a standard reference curve.

Initial efforts to develop microspot immunoassays were driven by attempts to achieve the greatest possible sensitivity from an assay in the shortest possible time. Researchers proposed that an ambient analyte assay using “vanishingly small” antibody concentrations would provide extremely high sensitivity.^3,12 Using this theory, a small concentration of antibody (<0.01/K, where K = equilibrium constant of the reaction, M-1) is immobilized in a microspot that is 100 µm2 or less. The amount of antibody immobilized is low enough to render the amount of bound antigen negligible relative to the total concentration. Furthermore, because of the incredibly small size of these spots (r = 6 µm), a large number of spots can be accommodated per unit of surface area.

Other researchers demonstrated that a microspot immunoassay with a spot radius of 100 µm is capable of mass-sensing.¹³ That is, the microspot acted as a sample concentrator so that the maximum perturbation of the analyte concentration can be achieved upon binding. Using this method, sensitivities equivalent to ELISA were achieved, but it was used only on a fraction of the capture antibodies required by ELISA.

Most efforts at developing microspot immunoassays have relied on standard robotic instrumentation to create the microspots. These spots are typically 100–200 µm in diameter and are made from source-plate antibody concentrations that are about 1 mg/ml, with 350 pl to 1 nl spotted. Depending on surface retention, this results in approximately 1–4 ¥ 109 immunoglobulin G molecules per spot, which is within the mass-sensing range posed by researchers.¹³ 

Developing an Antibody Microarray: From Bulk ELISA to Microspot

Choosing a Surface. 

Assay development efforts by Schleicher & Schuell BioScience Inc. (Keene, NH) focused on using a three-dimensional microporous nitrocellulose surface as an immobilization substrate. In order to be useful as an immobilization medium for quantitative analysis, a surface must immobilize in a near-quantitative fashion, allow for long-term storage of the immobilized protein, and allow solution phase probes to interact with the molecules that have been immobilized. Nitrocellulose meets all of these requirements.

Nitrocellulose has been utilized extensively as a surface for immobilizing proteins in research techniques such as Western blotting and lateral-flow immunodiagnostics. Microporosity and nitrocellulose offer many benefits for microspot immunoassays as opposed to two-dimensional surfaces such as polystyrene and derivatized glass. These benefits include high binding capacity, noncovalent attachment of proteins, a stable long-term immobilization environment, and a milieu conducive to consistent binding.

Figure 3. A comparison of standard curves for the detection of IL1b using a microspot assay and ELISA. Standard curves from a microspot assay for IL1b using four source plate concentrations (0.25, 0.5, 1, and 2 mg/ml) of monoclonal antibody (a), and a standard curve generated by ELISA (b). The data from the two methods were normalized and transformed so they could be compared linearly on the same graph (c).  (Click to enlarge)

A nitrocellulose-coated glass slide provides approximately 2.6 ¥ 105 µm3 of binding volume per microspot. Within this volume, there is an enormous amount of surface area due to the microporous nature of the polymer. While the binding capacity of a nitrocellulose membrane has been estimated at 70 µg/cm2 of surface area, this value for a nitrocellulose thin film used in microarraying would be approximately 14 µg/cm2, about 625 times greater than the amount of protein deposited from a typical arrayer. Thus, with a microporous surface, every molecule of an antibody that is arrayed will be immobilized. Furthermore, data indicate that when an antibody is deposited on nitrocellulose in an appropriate buffer, the binding activity of that antibody is stable for at least 771 days.
In contrast, two-dimensional surfaces are limited in their binding capacity to the surface area being utilized. In an ELISA, this surface area includes the bottom and sides of a well, while in a microspot assay, this surface area is limited to approximately 0.018 mm2, a fraction of the volume associated with a three-dimensional surface.

Comparing Sensitivity and Dynamic Range. 

For a microspot immunoassay to be accepted by the research and diagnostics communities, it must be empirically determined that it maintains all of the features of ELISA, the industry-accepted standard. Once equivalence with ELISA has been demonstrated, then the features of the microformat can be exploited. 

Microspot assay development can be approached by looking at the signal response as a function of the amount of capture antibody arrayed. A donkey anti-goat capture antibody was arrayed onto nitrocellulose-coated slides from a source plate containing serial dilutions of the antibody (10–300 µg/ml), and each concentration was arrayed in triplicate. Piezoelectric noncontact dispensing was used so that a defined amount of material (350 pl) was dispensed on each spot. The array was then incubated with a series of biotinylated goat polyclonal antibodies at 200 ng/ml, followed by streptavidin-linked Cy5. Digital images were generated by confocal laser-based scanning at a 10 micron pixel resolution (see Figure 1).

Figure 4. A map of an anti-cytokine antibody array on a slide with eight nitrocellulose coated areas. Each antibody was arrayed in triplicate.  (Click to enlarge)

The log-transformed data from this experiment demonstrated that a linear relationship exists between the amount of antibody arrayed and the amount of signal generated (see Figure 2). It is important to note that the log-log slopes of this function are close to being the same, which implies that over a wide concentration range, the relationship between the amount of deposited antibody and active antibody is constant. 
The next step was to compare the results from a capture microassay with a traditional ELISA. One anti-IL1b monoclonal antibody sample was arrayed onto nitrocellulose-coated slides at the indicated concentrations, while another was coated onto the sides and bottoms of polystyrene microtiter wells at 4 µg/ml. After the appropriate blocking steps, both immobilized antibody samples were incubated with serial dilutions of purified recombinant human IL1b, followed by a detection antibody linked to biotin. Finally, SA-Cy5 was used to generate a signal in the microarray, while SA-HRP followed by color detection was used in the ELISA.

The shape of the dose response curves generated by both methods is the same, indicating that the antibody pairs are functioning identically in both techniques (see Figures 3a and 3b). In order to compare the data more easily, the plots were transformed and normalized, demonstrating a linear relationship with identical slopes (see Figure 3c). The only difference was that the microspot format on the nitrocellulose was 1-log more sensitive than the traditional ELISA. This observation may be a function of the method (microspot versus bulk), the detection system (fluorescence versus absorbance), or a combination of both.

Figure 5. Digital images of two arrays. The array on the left was incubated with lysates from THP-1 cells (a), while the array on the right was incubated with lysates from THP-1 cells stimulated with lipopolysaccharide (b).  (Click to enlarge)

The log-transform slopes, the limits of detection, and the concentration range over which the dose response was linear were determined for 17 cytokines measured by the microspot assay and traditional ELISA (see Table I). In all cases, the standard curves were generated using a purified recombinant human antigen diluted in buffer. Besides two exceptions (IP10 and TNFr2), the microspot assay resulted in standard curves that were steeper than ELISA. In addition, the microspot assay was typically at least as sensitive as ELISA and was linear for 2–3 orders of magnitude, whereas ELISA was only linear for 1.5 to 2 orders of magnitude. These data provided empirical evidence that a microspot sandwich immunoassay was appropriate for measuring cytokines.

Use with Biological Samples and Quantification. In order to be a useful diagnostic tool, protein microspot immunoassays must be compatible with biological samples and must be able to quantify antigens as an ELISA does. To examine these issues, human THP-1 lymphocytes were cultured in an RPMI media containing 10% fetal bovine serum. These cells either were kept untreated or were stimulated with low concentrations of lipopolysaccharide (LPS). At various times after stimulation, the media were harvested as well as the cells. Cell extracts were made by lysing the cells in a phosphate-buffered saline containing IGEPAL detergent. 

By using a slide with eight areas of nitrocellulose, an array of 16 monoclonal antihuman cytokine antibodies was created, each arrayed in triplicate, with six positive controls above and below the array (see Figure 4). Each area on the slide contained the same array configuration, and by using a multiwell chamber, all eight arrays could be processed simultaneously. Thus, on one slide, 480 data points could be collected, corresponding to 128 answers (in triplicate) plus the controls. Also, up to eight different samples could be analyzed at once, leaving several arrays free to generate a standard curve. 

Both the conditioned media mentioned above as well as the crude cell lysates were incubated with the microspot arrays as undiluted samples

Table I. A comparison of standard curve data measured by a microspot assay and traditional ELISA. (Click to enlarge)

 so that the levels of various cytokines could be assessed. Detection was accomplished through the use of biotinylated detection antibodies and streptavidin Cy5.

A qualitative comparison of arrays incubated with THP-1 crude lysates from LPS treated and untreated cells was generated (see Figure 5). By referring to Figure 4, it is evident that there is an increase in the level of ICAM-1, IL1b, and IL8 inside the THP-1 cells in response to being treated and untreated cells was generated (see Figure 5). By referring to Figure 4, it is evident that there is an increase in the level of ICAM-1, IL1b, and IL8 inside the THP-1 cells in response to being treated with LPS. The signal intensity of the positive control spots (the top and bottom 6) did not change significantly. 

To examine further the power of this technique, a four-point standard curve was generated for IL8 using four arrays on a slide. By using a commercially available recombinant human IL8 weight standard, samples of lysate or media from treated and untreated THP-1 cells were incubated in the four wells of the slide. In addition, the results of the microspot assay were compared with a traditional ELISA using a seven-point standard curve. For all four samples, the level of IL8 was determined to be virtually identical in both methods (see Table II). 

Conclusion

Table I. A comparison of standard curve data measured by a microspot assay and traditional ELISA. (Click to enlarge)

During the past few years, there has been a substantial amount of hype concerning the miniaturization of assays. At the same time, developing and using reliable DNA microarrays has been a significant milestone toward miniaturized diagnostic tests that provide real-time quantitative answers. However, due to a variety of reasons, the transition from nucleic acid–based tests to protein immunoassays is a substantial hurdle. Proteins are a much more chemically diverse set of molecules than nucleic acids. In addition, the long-term stability of proteins is difficult to achieve relative to nucleic acids, and labeling complex protein mixtures is difficult, if not impossible, necessitating the use of detection antibodies or specificities. Furthermore, from an information standpoint in clinical as well as nonclinical diagnostics, the ability to make a chip with thousands of elements is less important for the miniaturization of immunoassays than reproducibility and quantification factors. 

Efforts toward miniaturizing assays have centered on developing microspot immunoassays relative to the currently accepted ELISA tests. Whether an ELISA is used by the hundreds in a clinical trial to screen patient sera for drug response effects, or is used individually in research labs to follow cell cultures, they are reliable tests that provide a quantitative output. However, a significant drawback to ELISA is the amount of sample required to obtain just one answer. One approach has been to develop microspot immunoassays that provide information in a manner equally reliable as ELISA, but in a multiplexed format so that dozens of events can be observed simultaneously.

Basic research on protein microarrays can potentially offer the diagnostics community significant technological advances in assay development. However, both technological research as well as applications research will need to find a common ground. As protein microarrays continue to be developed, researchers will need to proceed with an assay-development frame of reference. Developing an interesting research tool without regard for stability, reproducibility, dynamic range, and quantification will result in an interesting research tool and not much more. Likewise, in their search for technology that provides answers more efficiently than current tests, IVD manufacturers will need to accept an assay format that isn’t a 96-well polystyrene microtiter plate. If both fields recognize the potential power of miniaturization to assay development, then protein microarrays have a promising future.

References 

1. MD Schena et al., “Quantitative Monitoring of Gene Expression Patterns with a Complementary DNA Microarray,” Science 270 (1995): 467–470.

2. A Butte, “The Use and Analysis of Microarray Data,” Nature Reviews Drug Discovery 12 (2002): 951–960.

3. RP Ekins, F Chu, and E Biggart, “The Development of Microspot Multianalyte Radiometric Immunoassay Using Dual Fluorescent-labeled Antibodies,” Analytica Chimica Acta 227 (1990): 73–90.

4. G Walter et al., “Protein Arrays for Gene Expression and Molecular Interaction Screening,” Current Opinion in Microbiology 3 (2000): 298–302.

5. BB Haab, “Advances in Protein Microarray Technology for Protein Expression and Interaction Profiling,” Current Opinion in Drug Discovery and Development 4, no. 1 (2001): 116–123.

6. L Belov et al., “Immunophenotyping of Leukemias Using a Cluster of Differentiation Antibody Microarray,” Cancer Research 61 (2001): 4483–4489.

7. JX Huang et al., “High-Throughput Genomic and Proteomic Analysis Using Microarray Technology,” Clinical Chemistry 47, no. 10 (2001): 1912–1916.

8. RP Huang et al., “Simultaneous Detection of Multiple Cytokines from Conditioned Media and Patient’s Sera by an Antibody-Based Protein Array System,” Analytical Biochemistry 294 (2001): 55–62.

9. MD Moody et al., “Array-based ELISAs for High-Throughput Analysis of Human Cytokines,” BioTechniques 31, no. 1 (2001): 186–194.

10. R Wiese et al., “Simultaneous multianalyte ELISA Performed on a Microarray Platform,” Clinical Chemistry 47, no. 8 (2001): 1451–1457.

11. R Huang et al., “Connexin 43 Suppresses Human Glioblastoma Cell Growth by Down-Regulation of Monocyte Chemotactic Protein 1, as Discovered Using Protein Array Technology,” Cancer Research 62 (2002): 2806–2812.

12. RP Ekins and FW Chu, “Multianalyte Microspot Immunoassay–Microanalytical ‘Compact Disk’ of the Future,” Clinical Chemistry 37, no. 11 (1991): 1955–1967.

13. JW Silzel et al., “Mass-Sensing, Multianalyte Microarray Immunoassay with Imaging Detection,” Clinical Chemistry 44, no. 9 (1998): 2036–2043. 

John L. Tonkinson, PhD, is a senior scientist, Dan S. Osborn is a research associate, Brett A. Stillman, PhD, is a staff scientist, and Weiwen Zhao, PhD, is a staff scientist at Schleicher & Schuell BioScience Inc. (Keene, NH). They can be reached via jtonk@s-and-s.com,    dan_osborn@s-and-s.com,  brett_stillman@s-and-s.com, and weiwen_ zhao@s-and-s.com, respectively.

Copyright ©2003 IVD Technology