The number of membrane-based rapid immunochromatographic devices on the market is continuing to increase at a very quick pace. Major factors that are contributing to this growth include improvements in conjugate technology and a growing understanding among product developers of the general design principles involved.
Although today's immunochromatographic devices come in a wide variety of designs with a diverse assortment of housings, most commercially available tests are based on one of two simple formats. The most common format is the lateral-flow or dipstick design, which has become familiar through its use in physician-office assays as well as in over-the-counter tests (e.g., Unipath's Clear Blue pregnancy test). A less widespread format is the flow-through or transverse-flow design, which requires greater operator skill and is therefore usually restricted to professional use (e.g., Medmira's rapid HIV screen).
Figure 1. Achieving a crisp, clear test result, such as in the samples shown here, depends on correct binding of the capture reagent to the membrane.
Regardless of the format being used, achieving a sensitive and reproducible test requires the manufacturer to have an efficient procedure for applying the capture-line reagent. Companies involved in the rapid diagnostic industry have been active in publishing information about how to optimize capture-line application.1–5 This article offers further aid to product developers, discussing the basic principles involved in applying protein capture lines to nitrocellulose membranes, and highlighting some of the common problems that can be encountered during the development of an immunochromatographic assay. Because the problems associated with protein binding are more prevalent in lateral-flow assays, this article will focus especially on issues relating to such systems.
The Importance of Protein Binding
In immunochromatographic assays, the primary function of a protein applied to a membrane is to act as a capture reagent for the target analyte in a sample. Because the test result is totally dependent upon achieving a good binding of the capture reagent to the membrane, the importance of achieving a high and consistent level of protein binding cannot be overstressed (see Figure 1).
Despite the considerable amount of research that has been conducted since nitrocellulose was first used as a protein-binding membrane, the exact mechanism of that binding remains unknown.6 It is known that a number of forces are at work—specifically, hydrophobic interactions, hydrogen bonding, and electrostatic interactions—but a clear understanding of the exact effect and significance of each force has remained elusive. Two reasonable models have been proposed. The first model suggests that proteins are initially attracted to a membrane surface by electrostatic interaction, while long-term attachment is accomplished by a combination of hydrogen bonding and hydrophobic interactions. Although extremely difficult to prove, this model of the interaction fits the published experimental data and is often the accepted mode of interaction.1, 7–11
Figure 2. Problems with protein binding are typically visible in the capture line of an assay's test result, as in these examples.
A second model suggests that the initial attachment of the protein is caused by hydrophobic interactions, with long-term binding accomplished by electrostatic forces. This model also agrees with much of the published data. However, the electrostatic partition mechanism may not provide a full explanation for the long-term stability conferred on protein attachment by drying or the use of an alcohol fixation step.3,6
Whatever the balance of forces responsible for protein binding, it is widely agreed that product developers should consider all such forces when they are seeking to optimize the binding of proteins to a particular membrane. Such considerations will inevitably have implications for both the selection of materials to be used, and the ways that they will be processed. For instance, if the product developer selects a buffer that too greatly reduces either hydrophobic or electrostatic interactions, the level of protein binding could be dramatically reduced. Similarly, it is widely recognized that adequate drying of the membrane after protein application is an important practice for ensuring the long-term stability of the protein–membrane bond.1–4, 6
Figure 3. A weak capture line indicates that the amount of protein bound to the membrane is too low.
The manufacturer's selection of materials can have an effect on the binding of proteins to nitrocellulose membranes. Materials that interfere with protein binding can be divided into three general types: nonspecific proteins, materials that interfere with electrostatic interactions, and materials that interfere with hydrophobic interactions. Commonly used materials that reduce protein attachment include those that compete for binding sites, such as the classic bulking proteins (e.g., BSA, animal sera), as well as those that interfere with hydrogen bonding (e.g., formamide, urea) and those that interfere with hydrophobic bonding (e.g., Tween, Triton, or Brij). Man-made polymers such as polyvinyl alcohol (PVA), polyethylene glycol (PEG), and polyvinyl pyrrolidone (PVP) can also interfere with protein binding. Their mode of action may be a combination of effects that inhibit one or more of the forces essential to protein–membrane binding.
If an insufficient amount of protein binds to the membrane, or if the protein does not bond to the membrane with the necessary strength, some significant problems can arise. These problems are typically visible in the capture line of an assay's test result (see Figure 2). If the amount of protein bound to the membrane is too low, the resulting capture line will be weak and test sensitivity will be reduced (see Figure 3). If binding is inefficient, the protein can diffuse before finally becoming immobilized on the membrane. The resulting capture line will be broad and weak instead of crisp and clear, making test results difficult to interpret. In extreme cases where the physical attachment of the protein to the membrane is too weak, the passage of analyte proteins and surfactant solutions can actually wash capture reagents off the membrane. In such cases the assay will display a broad line—or no clear line at all—again making it difficult to interpret the test results (see Figure 4).
Figure 4. A diffuse capture line can result when the capture reagent is washed away by the passage of analyte proteins and surfactant solutions.
Problems such as these are regularly seen by product developers in the IVD industry, and can significantly slow the development of a successful immunochromatographic assay. To understand how to go about resolving such problems, developers should first have a firm grasp on the various factors that can influence protein–membrane binding, including those inherent in the materials and processing used for their tests. These elements will be discussed in the first installment of this article. Typical techniques for solving such problems will be given in the second installment, which will appear in a future issue of IVD Technology.
Factors That Influence Protein Binding
When investigating the binding of protein capture reagents to nitrocellulose membranes, product developers should consider each of the following five critical areas that can have an effect on the binding mechanism.
Although many development labs do a good job of studying and characterizing the application buffers and membranes used in their tests, they are less likely to fully investigate or optimize the capture reagents and application systems they employ. Such an omission is often due to the fact that the latter elements are frequently considered set even before the beginning of the development process, leaving little opportunity for changes to be made. With those factors out of consideration, product developers often have no choice but to focus on optimizing the other elements that are still within their discretion.
Capture Reagents. The proteins used as capture reagents vary from test to test. However subtle their differences, no single capture reagent is absolutely identical to another. Perhaps more important, different proteins exhibit varying levels of attachment to different membranes (see Figure 5).5 The process of optimizing binding is most straightforward with a monoclonal antibody, where the protein is a homogeneous material. Optimization is more difficult in the case of polyclonal antibodies because there are a variety of epitopes present, and ideally each requires slightly different binding conditions. Species such as IgA or IgM can present an even greater challenge because of the potential for structural or steric problems. Other proteins such as BSA, protein A, or protein G can cause significant difficulties due either to their chemistry or their size (large molecules are more likely to remain attached to a solid phase than smaller ones).
Application Equipment. Although systems used to apply capture reagents can also present problems, most commercially available equipment has both advantages and disadvantages. Variables can include the ability or inability to dispense measured volumes; capacity to handle strips, sheets, or membranes; speed of application; and postapplication handling of strips. The best solution is for the manufacturer to find an application system that satisfies the most significant practical issues, such as raw material limitations and system capacity. Other factors can then be optimized for that particular application system.
Figure 5. Comparative binding of IgG and albumin to a range of nitrocellulose membranes from different manufacturers. To replicate actual test conditions, data were generated using a flow-through system where the sample was applied to the membrane surface and pulled through the membrane by vacuum.16 Although the more common test method is to incubate the membrane with the protein solution, that method permits protein molecules to stack up within the pores of the membrane, resulting in the formation of a protein multilayer instead of a protein monolayer.1 The traditional test method thus results in artificially high levels of protein binding that are unrepresentative of actual use in rapid immunochromatographic tests.
Ambient Humidity. The humidity at the time that the capture line is applied can have a significant effect on the quality of the line, especially when spray systems are used. If atmospheric humidity is low a static charge can collect on the membrane, which can result in satellite spots when the protein is sprayed onto the membrane surface. Low humidity can also cause the development of hydrophobic patches on the membrane surface. By contrast, extremely high humidity can result in very rapid wicking of the applied protein, causing wide or diffuse capture lines. In general, the optimal humidity in which to apply proteins is between 45 and 65% RH. To ensure even properties throughout the feedstock, the membrane should be allowed to equilibrate with the atmosphere before application. The optimal equilibration time should be determined by experimental investigation.
Optimizing the Application Buffer
Because protein capture reagents vary, maximizing the binding of a given protein may also require buffer conditions that differ from those appropriate to another protein. There are two important factors that need to be optimized through modifications to the application buffer.
To ensure that sufficient protein is available in the applied capture line, it is first essential that the capture protein be soluble in the application buffer. In order to confer enough solubility to enable the protein to be dissolved, it is necessary to have some ions present in the application buffer. Although the ionic strength of the buffer can help to control the pH of the capture reagent, it also interferes with electrostatic interactions essential to protein binding. It is therefore important to determine the lowest possible ion level for the buffer that will result in a sufficient concentration of capture protein in solution.
If the molecules of a given protein concentration are stable in solution, they will tend to remain so. But if it is energetically favorable for the protein to partition onto the solid phase, then a greater proportion of protein will attach than if the protein is stable in solution. Such an energy state can be induced through the use of destabilizing or coprecipitating agents. However, too much correction in this direction can cause other problems. If the protein precipitates before it can be applied to the membrane, for instance, the entire system will become highly unstable and almost totally irreproducible. The amount of dissolved protein remaining for attachment to the membrane will thus be dramatically reduced. Precipitates may also cause problems by blocking the application equipment or clogging the pores of the membrane. There are some cases in which obtaining a reasonable level of binding may make it necessary to cause the protein to precipitate during application, but these are exceptions to the general rule.
As the above analysis suggests, protein binding can be altered by adjusting the properties of the application buffer (see Figure 6). Key properties that can be usefully modified include the buffer's ionic strength and acidity, and the level of coprecipitating agents employed.
Ionic Strength. Within a defined range of ionic strength, the solubility of a typical protein increases in direct proportion to the salt content of the application buffer. Because it is desirable to minimize the molecular stability of a capture protein in solution, the ionic strength of the solution should be kept as low as possible. Doing so will increase the speed of protein binding. Developers should also be aware that high salt concentrations can cause precipitation of proteins, and that the presence of large quantities of salt during drying can interfere with the sensitivity and stability of the test.
Acidity. The pH level of an application buffer can have a significant effect on its properties. The solubility of a typical protein is at its minimum at its isoelectric point. Since developers are aiming to minimize the molecular stability of the capture protein in solution, the ideal pH of the application buffer should therefore be at about the isoelectric point of the capture protein being used.
Coprecipitating Agents. When modifying an application buffer, developers may choose to add a destabilizing or coprecipitating agent in order to reduce the stability of the protein molecules in solution. The action of such coprecipitating agents relies on the differing stability that the fc and f(ab) regions of the IgG molecule have toward the agents used.11 The structure of the fc region is far more likely to be degraded by the action of coprecipitating agents. Partial destabilization of the fc regions leads to the exposure of more-hydrophobic groups that are normally hidden within the protein structure. Thus, regardless of which mechanism is accepted for the binding of proteins to nitrocellulose, the increase in protein hydrophobicity resulting from the use of such coprecipitating agents will improve protein binding.
Figure 6. Varied results from capture lines of 1mg/ml mouse IgG applied using different buffers: (a) 10 mmol phosphate, pH 7.2; (b) 10 mmol phosphate + 3% methanol, pH 7.2; (c) 10 mmol phosphate + 150 mmol NaCl + 3% methanol, pH 7.2; (d) 50 mmol phosphate + 150 mmol NaCl + 1% BSA, pH 7.2; (e) 50 mmol phosphate + 150 mmol NaCl, pH 7.2; (f) 50 mmol phosphate + 150 mmol NaCl, pH 6.0. All samples were detected by a 40 nmol gold-conjugated goat antimouse IgG antibody.
The most commonly used coprecipitating agent is alcohol, which can be recommended for a number of reasons. The presence of alcohol helps to rewet the membrane, reduces any static charge it may have, and has a destabilizing effect on the protein in solution. Levels of between 3 and 5% methanol can give considerable improvement in the performance of a membrane used for an immunoassay.3
The use of alcohol to improve protein binding to a solid phase has been known for several years in the production of ELISA plates, and is now regarded as a standard protocol.11,12 The influence of aliphatic alcohols on binding in nitrocellulose membranes was first reported in 1980, while a 1% isopropanol solution is widely used as a fixing solution in protein blotting experiments.6,13
Although other materials such as diethylaminoethyl or ammonium sulfate can sometimes have beneficial effects when used as coprecipitating agents, they are generally less desirable than alcohol. Even small variations in the concentration of these types of materials can have severe effects on the degree of protein precipitation. For this reason, precipitating agents other than alcohols should generally not be used.
Considering the points outlined above, a buffer comprised of 10 mmol phosphate +3% methanol pH 7 is suggested for initial development studies. Although such a buffer will not prove optimal for all applications, it offers a very good starting point for the development process.
The membrane itself has a significant effect on the protein binding observed in a rapid assay, with three key factors affecting membrane performance:
Because of the wide range of potential capture reagents, no single membrane will work optimally for every assay. The level of protein binding can vary dramatically among different types of membranes (see Figure 5). Unfortunately, this means that product developers must reinvestigate and optimize their membrane selection for each assay they develop. However, the potential improvement in test performance and assay reproducibility is sufficient compensation for the additional work involved.
Pore Size. Developers of lateral-flow immunoassays should treat supplier references to pore size with caution. The actual pore size of a membrane depends on the method used to measure it, and since different manufacturers use different measuring techniques, any two membranes with the same nominal pore size could differ significantly if measured by a constant technique (see Figure 7).
Figure 7. Pore size data for nitrocellulose membranes based on data from a Coulter porometer.
Pore sizes are usually measured in the filtration direction, that is, through thickness of the membrane. But the size and shape of pores in the filtration direction may have no relation to the size and shape of pores in the lateral direction (that is, along the length of the membrane). For a lateral-flow assay therefore, the conventional method of quoting pore size is not really relevant. Moreover, if a plastic cast membrane is used, measuring pore size in the filtration direction is physically impossible because of the presence of the film backing. In such cases, the pore sizes quoted by suppliers are often no better than best estimates based on lateral wicking data.
Product developers can use nominal pore size—cautiously—to differentiate membranes from a single manufacturer. But it is not recommended that such information be used to specify pore size for membranes from another manufacturer. Nominal pore size generally has no standard meaning in terms of protein binding, particularly for lateral-flow assays, and developers are better advised to screen a range of membranes when they begin the development of a new assay.
Although nominal pore size has little real importance, the lateral pore size and structure of a membrane does have a significant effect on its suitability for use in lateral-flow assays. Within any range of nitrocellulose membranes, as pore size decreases the protein binding capacity of the membrane increases because of the related increase in available membrane surface area.1 The approximate surface area for membranes of different pore sizes can be estimated by looking at the surface area ratio (SAR) for each material.3 The SAR represents the ratio of actual available surface in the pores of the membrane to the area of membrane used (see Table I).14 Another phenomenon of importance is that as a membrane's pore size decreases, the lateral wicking rate of the membrane also decreases (see Table II).15 A slower wicking rate increases the effective sensitivity of a test because it permits reagents to spend a longer time in the capture zone.
Figure 8. Water present during the application of posttreatments can make sections of the membrane hydrophobic, resulting in striations or intensity variations in the capture line.
The combined effect of these two phenomena is that greater relative sensitivity is achievable by using membranes with a smaller pore size. Thus, as a general guideline, a developer who is most concerned with the ultimate sensitivity of an assay should select a membrane with the smallest possible pore size; while a developer who is primarily concerned with the wicking speed of an assay should select a membrane with a larger pore size. Whatever their needs, developers can best find the optimal membrane for their tests by evaluating a variety of possibilities during the early stages of product development.
Posttreatments. Following manufacture, nitrocellulose membranes routinely receive posttreatment to remove dust (unincorporated polymer left on the surface of the membrane after manufacture) or to modify their rewetting characteristics. In either case, there is the possibility that such posttreatment may introduce trace chemicals or other substances that are not nitrocellulose, and that may have an effect on the performance of the finished test device.
In general terms, the manufacturer should always know what additional substances may be present in the membrane, in what concentration they are present, and how to measure their levels. Depending on what additional materials are present, significant effects may be observed in the level of protein binding, the flow rate of the membrane, and in the effects of aging on the membrane.
1. MA Harvey, Optimization of Nitrocellulose Membranes Based Immunoassays (Keene, NH: Schleicher & Schuell, 1991).
2. Guide to Building Molecular and Immunodiagnostic Device Platforms (Keene, NH: Schleicher & Schuell, 1997).
3. Short Guide for Developing Immunochromatographic Test Strips (Bedford, MA: Millipore Corp., 1996).
4. KD Jones, Technical Application Notes, No. 1–3 (Maidstone, UK: Whatman International, 1997–1998).
5. The Latex Course, 1994 (Fishers, IN: Bangs Laboratories, 1994).
6. E Harlow and D Lane, Antibodies: A Laboratory Manual (Cold Spring Harbor, NY: PUBLISHER NAME TK, 1988).
7. C Wallis, JL Melnick, and CP Gerba, “Concentration of Viruses from Water by Membrane Chromatography,” Annual Review of Microbiology 33 (1979): 413–437.
8. WG Presswood, Membrane Filtration: Applications and Problems, (New York: Marcel Dekker, 1981).
9. SR Farrah, DO Shah, and LO Ingram, “ARTICLE TITLE TK,” Proceedings of the National Academy of Science USA VOLUME NUMBER TK (1978): 1229–1236.
10. B Batteiger, V Newhall, and RB Jones, “The Use of Tween 20 as a Blocking Agent in the Immunological Detection of Proteins Transferred to Nitrocellulose Membranes,” Journal of Immunology Methods 55 (1982): 297–307.
11. P Tijssen, Practice and Theory of Immunoassays, 8th ed, (Amsterdam: Elsevier, 1993).
12. HC Wood and TG Wreghitt, “Techniques,” in ELISA in the Clinical Microbiological Laboratory, ed TG Wreghitt and P Morgan-Capner (London: PHLS, 1990), pp 6–21.
13. Z Schneider, “Aliphatic Alcohols Improve the Adsorptive Performance of Cellulose Nitrate Membranes—Application in Chromatography and Enzyme Assays,” Annals of Biochemistry 108 (1980): 96–103.
14. R Bowen, private communication with author, Swansea, UK, DATE TK.
15. Technical Data: Nitrocellulose Membranes (Maidstone, UK: Whatman International, 1996).
16. KD Jones and AK Hopkins, “Protein Binding in Nitrocellulose Membranes 0.2 to 12 µm: A Comparison of Commercially Available Membranes for a Novel Flow-Through Immunoassay,” poster no. 21, presented at the 1998 Annual Meeting of the American Association for Clinical Chemistry, Chicago, August 2–6, 1998.
17. Unpublished results (Maidstone, UK: Whatman International, YEAR TK).
18. AM Campbell, Monoclonal Antibody and Immunosensor Technology (Amsterdam: Elsevier, 1992).
19. KD Jones and AK Hopkins, “Evaluation of the Efficiency of a Range of Membrane Blocking Agents for Nitrocellulose Membrane Based In Vitro Diagnostic Devices,” poster no. 3, presented at the 1998 Annual Meeting of the American Association for Clinical Chemistry, Chicago, August 2–6, 1998.