History of nitrocellulose

Nitrocellulose is a versatile polymer that has been broadly utilized since the 1800s.Reference Martin and Colombi ^{1 –} Reference Falkowitz and Piech ³ Also known as cellulose nitrate, nitrocellulose is created commercially by the reaction of cellulose (purified from plants, commonly wood pulp and cotton) with nitric acid, replacing the cellulose hydroxyl groups with nitrate groups.Reference Tonkinson and Stillman ⁴ Today, nitrocellulose membranes are created by phase inversion,Reference Ferry ^{5 –} Reference Zsigmondy and Bachmann ⁷ in which nitrocellulose is dissolved in an organic solvent that is evaporated in the presence of a nonsolvent, leaving a precipitated nitrocellulose membrane with high porosityReference Tonkinson and Stillman ^{4 ,} Reference Wong and Tse ⁸ ( Figure 1 a). The porosity and pore size of the membrane can be controlled by the solvents used, evaporation speed, temperature, and humidity.Reference Wong and Tse ⁸ The result is a material with the unique combination of tunable pore size, high surface-to-volume ratio, and very low cost.

Figure 1.

Nitrocellulose and nitrocellulose-based assays. (a) SEM image of a blade-cut edge of a porous nitrocellulose assay membrane. (b) Schematic of a traditional lateral flow assay. (c) Schematic of a multistep two-dimensional paper network assay.

Porous nitrocellulose membranes were first used to immobilize biomolecules in the 1960s.Reference Tonkinson and Stillman ⁴ Nygaard and Hall demonstrated in 1963 that RNA-DNA complexes adsorb onto nitrocellulose membranes, while free nucleic acid strands pass through.Reference Nygaard and Hall ⁹ Others then began immobilizing nucleic acids on nitrocellulose membranes to probe for interactions between a nucleic acid of interest and other biomolecules.Reference Gillespie and Spiegelman ^{10 ,} Reference Flavell, Birfelder, Sanders and Borst ¹¹ In 1975, Southern demonstrated the transfer of DNA from polyacrylamide gels to nitrocellulose.Reference Southern ¹² This groundbreaking technique, known as the “Southern blot,” allowed specific nucleic acid fragments to be captured for subsequent analysis. The Southern blot inspired the “Northern blot” for RNA transferReference Alwine, Kemp and Stark ¹³ and the “Western blot” for protein transfer to nitrocellulose.Reference Towbin, Staehelin and Gordon ^{14 ,} Reference Burnette ¹⁵ These blotting techniques have been widely employed in biological research and highlight the unique ability of nitrocellulose to interact with three of the most important classes of biomolecules (proteins, DNA, and RNA).Reference Tonkinson and Stillman ⁴

Stemming from the blotting assays, other assays were developed utilizing porous nitrocellulose as a solid support for diagnostic assays incorporating flow. Both flow-through assays and lateral flow (LF) tests feature capture molecules (e.g., target-specific antibodies) immobilized on nitrocellulose through which a sequence of sample and reagents is passed to detect the presence of a target molecule. The flow-through assay uses flow perpendicular to the membrane, driven by a large absorbent padReference Valkirs and Barton ¹⁶ or vacuum manifold.Reference Ijsselmuiden, Herbrink, Meddens, Tank, Stolz and Vaneijk ¹⁷ While the flow-through assay has the advantage of being multiplexable, the highly manual nature of the test typically relegates it to the laboratory. However, two examples of POC flow-through assays are the Multispot HIV testReference O’Connell, Agan, Anderson, Malia and Michael ¹⁸ and the DxBox.Reference Lafleur, Stevens, McKenzie, Ramachandran, Spicar-Mihalic, Singhal, Arjyal, Osborn, Kauffman, Yager and Lutz ¹⁹ For most POC applications, the LF test has become the assay of choice.

Lateral flow technology

Over the last 25 years, nitrocellulose has become a prominent material in POC diagnostics through its role in LF technology. Since the first LF test came to market in 1988,Reference Chard ²⁰ this popular technology has developed into a $2.1 billion market involving over 200 companies.Reference Wong and Tse ⁸ Because LF tests are inexpensive, rapid, portable, equipment-free, and easy to use, they are now used in many fields, including clinical, environmental, agricultural, veterinary, drug testing, and bio-defense.Reference Wong and Tse ⁸

The most common LF devices consist of a “strip” with four contacting porous components arrayed linearly: (1) sample pad, (2) conjugate pad containing dried, gold-conjugated detection antibodies specific to the target analyte, (3) nitrocellulose membrane with capture agents immobilized at control and test lines, and (4) wicking pad to draw the sample through the device (Figure 1b).Reference Wong and Tse ⁸ The nitrocellulose membrane is critical to the functioning of LF devices because analyte binding and detection occur within it. To begin the test, the user adds a microliter-volume of sample (e.g., blood, urine, saliva) to the sample pad. Results are typically available in less than 15 minutes and can be visually read based on the development of color at the test line, due to the accumulation of the gold-conjugated detection antibodies. The device’s simplicity permits easy manufacturing, and the dried reagents eliminate the need for refrigeration and extend shelf life.Reference Posthuma-Trumpie, Korf and van Amerongen ²¹

While LF devices have revolutionized low-cost testing, they have several drawbacks: they are usually qualitative, have lower sensitivity than gold standards (e.g., nucleic acid amplification tests, enzyme-linked immunoassays (ELISA), and bacterial/viral culture), require additional sample preparation, are rarely multiplexable, and cannot perform multiple automated steps.Reference Posthuma-Trumpie, Korf and van Amerongen ²¹ Currently, numerous innovations, such as two-dimensional paper networks (2DPNs),Reference Fu, Kauffman, Lutz and Yager ^{22 –} Reference Lutz, Trinh, Ball, Fu and Yager ²⁷ are being developed to overcome these disadvantages. Such innovations require an understanding of the scientific and engineering principles involved in assay design.

Physics of flow in porous media

Understanding flow in porous materials is essential when designing nitrocellulose-based assays. The principle of capillary flow describes the passive wicking of fluids that occurs in porous media. Capillary flow is affected by several characteristics of the porous medium: pore size, pore distribution, and porosity; by characteristics of the fluid (e.g., viscosity, surface tension); and by the interaction between the two: the contact angle of the fluid with the porous material.Reference Washburn ^{28 –} Reference Bird, Stewart and Lightfoot ³⁰

Though most porous media have some tortuosity (i.e., the “twistiness” that is defined by the ratio of the actual path length of a particle flowing through the material to the straight-line distance), they are most easily approximated as a bundle of straight pores. This approximation allows us to use conventional fluidic principles, such as the Reynolds number, to understand the behavior of fluid in these pores. In nitrocellulose membranes (where the most applicable Reynolds number is ~10^–3), flow is laminar. Thus, flow streamlines are parallel to the channel walls, and the primary mode of mixing between adjacent streams is diffusion.Reference Bird, Stewart and Lightfoot ^{30 ,} Reference Purcell ³¹

The “wet-out” process of a fluid wicking into a dry porous membrane is governed by the Washburn equation:Reference Washburn ²⁸

where L, the length of the fluid column in the porous material, depends on pore diameter (d_p ), fluid surface tension (γ), fluid viscosity (μ), and the cosine of the contact angle (θ). Taking the derivative of L with respect to t, we see that the wet-out linear velocity is inversely proportional to the fluid column length behind the wetting edge:

This slowing is due to the linear increase in viscous resistance as the fluid column lengthens.

From Equation 2, we see that the wet-out rate is inversely proportional to fluid viscosity (μ) and proportional to the membrane pore diameter (d_p ). Though the viscosity of a liquid is generally known, the pore diameter within a particular membrane is not. In nitrocellulose membranes, pore size distribution varies widely depending on the material and casting process used. The membrane wet-out rate is also proportional to the cosine of the contact angle (θ) and the surface tension (γ), which account for the liquid-air and liquid-solid interfacial energies. Surfactants are often added to nitrocellulose during manufacturing to decrease the contact angle, thereby improving wettability. ³² The surface tensions of common liquids are well known, though changes in temperature affect wet-out rates, and fluids used in real assays are never pure solvents.

Once a porous membrane has fully wetted out, flow is governed by Darcy’s equation for flow in porous media:Reference Darcy ²⁹

where the volumetric flow rate (Q) depends on the paper’s permeability (κ), cross-sectional paper area (wh), fluid viscosity (μ), and the pressure drop (∆P) that occurs over the length (l) of the paper network. This equation assumes that flow is fed by an infinite source, and, as the wicking pad becomes wetted, the capillary pressure it generates does not decrease. These assumptions are approximately correct as long as the wicking pad does not approach saturation.

The utility of both Washburn’s equation and Darcy’s law for modeling flow in paper devices has been repeatedly demonstrated.Reference Osborn, Lutz, Fu, Kauffman, Stevens and Yager ³³ Furthermore, methods such as pH-based electrochemical markingReference Kauffman, Fu, Lutz and Yager ³⁴ have been developed to quantify flow rates in porous devices to enable a systematic validation process for mathematical modeling.

Protein adsorption

The adsorption of biomolecules—typically proteins in commercial lateral flow tests—to the porous nitrocellulose membrane is another key assay consideration. While protein adsorption to nitrocellulose has been observed for years,Reference Brash and Horbett ^{35 ,} Reference Horbett and Brash ³⁶ the process is not well understood.Reference Tonkinson and Stillman ⁴ However, known principles should be considered when trying to achieve protein immobilization at a test line—or to prevent nonspecific adsorption elsewhere.

First, proteins are thermodynamically driven to adsorb to nitrocellulose.Reference Vanoss, Good and Chaudhury ³⁷ This adsorption is governed by noncovalent interactions between the two.Reference Horbett and Brash ^{36 ,} Reference Vanoss, Good and Chaudhury ³⁷ Some argue that this attraction is a hydrophobic interaction between the carbon-containing nitrocellulose and the hydrophobic portion of the protein,Reference Wong and Tse ⁸ while others describe it as a primarily electrostatic interaction between the dipoles of the nitrocellulose nitrate groups and dipoles within the protein. ³² Either way, it is well-established that proteins will adsorb to porous nitrocellulose membranes unless something is done to prevent this process.Reference Wong and Tse ⁸ In particular, nitrocellulose-based LF assays are often pre-treated with a protein-containing blocking solution to fill unused binding sites.Reference Wong and Tse ⁸ This blocking step prevents the sample analyte from nonspecifically adsorbing to the membrane before it can reach the capture line.

Second, the kinetics of protein adsorption and desorption to the membrane must be considered, especially for capture molecules immobilized at a test line. Contrary to classic adsorption studies in which protein adsorption—tested in simplified systems at equilibrium—is considered irreversible,Reference Horbett and Brash ³⁶ the complex systems of LF assays render protein adsorption a kinetic, two-way process. Specifically, the fluid flow velocity,Reference Washburn ²⁸ presence of potential stripping agents such as salts and detergents,Reference Hoffman and Jump ³⁸ and reagent drying timeReference Horbett and Brash ^{36 ,} Reference Vanoss, Good and Chaudhury ³⁷ are important considerations for assays that depend on protein immobilization.

The most commonly used tools for the deposition of proteins include piezoelectric spotters that can deposit small volumes into precise locations on a strip,Reference Fu, Liang, Spicar-Mihalic, Houghtaling, Ramachandran and Yager ^{26 ,} Reference Gonzalez-Macia, Morrin, Smyth and Killard ³⁹ high-volume stripers that “paint” a line of reagent across a large batch of strips,Reference Wong and Tse ⁸ and inkjet printers.Reference Lutz ^{40 ,} Reference Yu, Shang, Cheng, Paik, Kaplan, Andrade and Ratner ⁴¹ Each method has advantages and disadvantages; the choice depends on the required throughput, the need for patterning flexibility, and budget. These deposition tools are also important for patterning assay reagents so they may be stored directly on the device.

Dry reagent storage, rehydration, and patterning

When designing POC diagnostics, storing reagents in dry form directly on the device is most desirable. Using dry reagents reduces user steps, removes the need for continuous refrigeration (i.e., cold chain), provides a longer assay shelf life, and facilitates device automation.

Generally, proteins lose function at high temperatures or when dried in an uncontrolled manner.Reference Crowe, Carpenter and Crowe ⁴² However, a wide body of research has demonstrated that the presence of the non-reducing disaccharides trehaloseReference Crowe, Carpenter and Crowe ^{42 –} Reference Mazzobre, Buera and Chirife ⁴⁶ and (less frequently) sucroseReference Crowe, Carpenter and Crowe ^{42 ,} Reference Stevens, Petri, Osborn, Spicar-Mihalic, McKenzie and Yager ⁴⁷ can stabilize dried proteins for extended periods at elevated temperatures. It is believed that the high glass transition temperature of trehalose (106°CReference Elias and Elias ⁴⁸ ) ensures that molecular mobility of embedded proteins is kept low at elevated temperatures, thus protecting the protein against degradation.Reference Crowe, Carpenter and Crowe ^{42 ,} Reference Garcia, Kirkham, Hatch, Hawkins and Yager ^{43 ,} Reference Mazzobre, Buera and Chirife ⁴⁶ Furthermore, the sugar hydroxyl groups substitute for the waters of hydration of the protein, which are lost upon drying, helping to preserve the protein’s native conformation.Reference Crowe, Carpenter and Crowe ⁴²

Because dried reagents are important for POC diagnostics, several techniques have been developed for dry reagent storage and subsequent rehydration. Reagents are often stored dry within a conjugate pad placed in line with the fluidic pathways for rehydration at the time of use. Conjugate pads have been used in both LF immunoassaysReference Wong and Tse ⁸ and, more recently, 2DPNs.Reference Fu, Liang, Spicar-Mihalic, Houghtaling, Ramachandran and Yager ²⁶ Though these pads are effective for reagent storage, they offer minimal control over the spatial and temporal release of rehydrated reagents and require additional materials and manufacturing steps, which add cost. Recent developments have sought to address the latter of these concerns. Abe et al. showed the ability to rehydrate gold-anti-IgG antibody conjugate dried on blocked cellulose membranes and to use it downstream to detect IgG in buffer.Reference Abe, Kotera, Suzuki and Citterio ⁴⁹ In our group, Fridley et al. developed methods for controlled release of reagents printed directly onto nitrocellulose membranes. By patterning reagents and sugars into specific layouts, we controlled both the spatial and temporal distribution of reagents as they rehydrated on the membrane.Reference Fridley, Le, Fu and Yager ⁵⁰

Detection

In an ideal assay, the user would be able to detect the lowest clinically relevant concentration of an analyte in a patient sample. In reality, however, the limit of detection (LOD) of rapid tests often does not reach that ideal. Most commonly, LF tests use labels that can be interpreted by eye to eliminate the need for additional machinery. Three promising detection labels are colloidal gold nanoparticles, colored latex beads, and carbon black.Reference Wong and Tse ⁸ Signal amplification chemistries, such as metallic enhancement of colloidal gold and enzyme-based systems, can also greatly improve the assay LOD and dynamic range.

Numerous labels and chemistries have been developed and are being used in commercial assays.Reference Wong and Tse ⁸ See Chapter 5 of the Lateral Flow Immunoassay book for a thorough review of detection labels and amplification chemistries.Reference Wong and Tse ⁸