Backscattering particle immunoassays in wire-guide droplet manipulations
© Yoon and You; licensee BioMed Central Ltd. 2008
Received: 20 March 2008
Accepted: 17 November 2008
Published: 17 November 2008
Skip to main content
© Yoon and You; licensee BioMed Central Ltd. 2008
Received: 20 March 2008
Accepted: 17 November 2008
Published: 17 November 2008
A simpler way for manipulating droplets on a flat surface was demonstrated, eliminating the complications in the existing methods of open-surface digital microfluidics. Programmed and motorized movements of 10 μL droplets were demonstrated using stepper motors and microcontrollers, including merging, complicated movement along the programmed path, and rapid mixing. Latex immunoagglutination assays for mouse immunoglobulin G, bovine viral diarrhea virus and Escherichia coli were demonstrated by merging two droplets on a superhydrophobic surface (contact angle = 155 ± 2°) and using subsequent back light scattering detection, with detection limits of 50 pg mL-1, 2.5 TCID50 mL-1 and 85 CFU mL-1, respectively, all significantly lower than the other immunoassay demonstrations in conventional microfluidics (~1 ng mL-1 for proteins, ~100 TCID50 mL-1 for viruses and ~100 CFU mL-1 for bacteria). Advantages of this system over conventional microfluidics or microwell plate assays include: (1) minimized biofouling and repeated use (>100 times) of a platform; (2) possibility of nanoliter droplet manipulation; (3) reprogrammability with a computer or a game pad interface.
Whether we are able to manipulate a droplet or not depends primarily on the surface tension of the droplet, which is closely associated with its contact angle. Figure 1 in the bottom right graphically defines the liquid contact angle θ. In most cases, the liquid is water or other aqueous solution. A liquid drop sitting on a certain surface maintains its shape due to the equilibrium in its surface tensions, specifically at its three-phase borderline. At this line, three different surface tensions are in exact balance, γ SL (solid-liquid), γ SV (solid-vapor) and γ LV (liquid-vapor). The latter two are often abbreviated as γ S and γ L since vapor can be approximated as vacuum (hence no subscript). As shown in Figure 1, the balance of three surface tension vectors should provide the following Young equation :
γ SL =γ S -γ L cos θ.
The Dupré equation describes the interaction between two different materials. For a solid surface and a liquid drop, the free energy of the work of adhesion (W a ; the adhesive energy of a liquid drop to a solid surface) is :
W a = γ S + γ L - γ SL .
Combining the Young and Dupré equations yields the following definition on the work of adhesion (Young-Dupré equation):
W a = γ L (1 + cos θ).
Equation (3) tells us how much work should be provided to overcome W a of a droplet to a surface, which is primarily a function of θ (γ L is constant if the liquid is mostly water).
In magnetofluidics, it is difficult to calculate the exact magnetic energy between paramagnetic particles and a magnetic bar shown in Figure 1 in the left. Since magnetofluidics has not been successful with conventional plastic surfaces, we can assume such magnetic energy is generally too weak to overcome W a for many conventional surfaces. Therefore, a superhydrophobic surface (whose water contact angle θ is 150° or higher) is needed in order to minimize both the contact area and W a for a droplet (the latter is often defined as "frictional force").
In both cases, external electric or magnetic fields may affect the behavior of biomolecules, thus affecting subsequent bioanalysis and biorecognition. In fact, optical detection was never demonstrated for magnetofluidics. A simpler way for manipulating droplets on a flat surface is needed to minimize the complications described above. A clean, metal wire (water contact angle θ < 10°) may be inserted into a droplet to guide its movement on a surface. However, the W a of a droplet to a metal wire is simply too small to overcome the W a to a flat surface due to the small contact area between the droplet and wire. Since W a = γ L (1 + cos θ), where γ L is the liquid surface tension, a very large θ may make this movement possible. In this work, we used a superhydrophobic surface with θ = 155° for droplet manipulations. Linear movements and subsequent merging of two droplets were attempted.
We demonstrated this droplet merging for particle immunoassays (more specifically, latex immunoagglutination assays). One droplet contained antibody-conjugated latex particles and the other contained target antigens. Three different target antigens were tested: mouse immunoglobulin G (mIgG; model protein), bovine viral diarrhea virus (BVDV; model virus) and Escherichia coli (E. coli; model bacterium). Antibody-antigen binding caused the latex particles to agglutinate, leading to the increased extent of light scattering, which was used for detection .
Light scattering detection is the most appropriate sensing modality for latex immunoagglutination, as there is no fluorescent dye in the system. Incident beam of light is scattered to all directions by latex particles, which is usually elastic (i.e. the wavelength of incident light is the same as that of scattered light). Incident and scattered light can be distinguished by locating a light detector not in parallel with the light source but at a certain angle (15°, 30°, 45°, 90° and 180° are commonly used; we used 180° which is back scattering). We used microparticles (920 nm in diameter) whose light scattering roughly follows the Mie theory . In this regime, light scattering intensity is a strong function of the particle size rather than the particle number; hence latex immunoagglutination leads to larger extent of light scattering . We maintained the intensity of a light source as constant, i.e. static light scattering.
Figure 2 shows the merging of two 10 μL droplets of deionized water (from Millipore Simplicity, Molsheim, France) on a superhydrophobic surface, clearly demonstrating two basic droplet manipulations: moving and merging. Movements were repeatable over the same line more than 10 times, regardless of the content of droplets. Movements were also successful for 5 and 20 μL droplets. Droplets were removed simply by tilting the surface and no further cleaning/rinsing was performed. Similar experiments were performed using polystyrene surfaces (plastic Petri dishes from Fisher Scientific; Pittsburg, PA, USA), but movements were not successful.
(73 mN/m) (1 + cos 155°) (2.0 mm2) = 14 nJ.
Similar analysis can be made for the metal wires. (Note that a resistor was used merely as a metal wire; no voltage was applied.) Since the diameter of the metal wire was 0.5 mm and the insertion depth was 2 mm, the contact area between the metal wire and the droplet was 0.39 mm2. Since most clean metal surfaces have a water contact angle of 10° , W a can be estimated as:
(73 mN/m) (1 + cos 10°) (0.39 mm2) = 57 nJ,
which is larger than that of a droplet to a superhydrophobic surface, consequently enabling droplet movement.
The water contact angles on the polystyrene surfaces were 91 ± 5° and the contact area of 10 μL droplets were 10.5 ± 0.4 mm2 (again using FTÅ200). The W a of 10 μL droplet to a polystyrene surfaces is:
(73 mN/m) (1 + cos 91°) (10.5 mm2) = 750 nJ,
indicating a droplet cannot be moved with a metal wire on a polystyrene surface.
Figures 3, 4 and 5 show the maximum light intensities taken from the merged droplets. All results are the averages of three different experiments (i.e. each taken from different merged droplets). Error bars indicate standard deviations. Paired, two-tailed t-tests were performed by comparing each dilution with a blank (10 mM PBS). Dilutions with significant differences from the blank are indicated by the grey color. The detection limit for mouse immunoglobulin G (mIgG) was 50 pg mL-1, equivalent to 0.5 pg of mIgG in a 10 μL target droplet.
All three figures follow the so-called Heidelberger-Kendall curve , characterized by initial increase at lower target (antigen) concentrations, followed by a decrease at higher concentrations. The left-hand side (initial increase) can serve as a calibration curve, as well as providing a linear range of immunoassay. The right-hand side (decrease at higher concentrations) represents the antigen-excess region (i.e. too much target antigens for a fixed amount of antibodies bound to the particles), which inhibits antibody-antigen binding. The apparent advantage of this droplet manipulation is that the liquids are in minimal contact with solid surfaces. The bottom superhydrophobic surface consists largely of air pockets and the contact area of a top wire is extremely small. This minimized contact reduces biomolecular adsorption (biofouling), enabling repeated use of a platform. In fact, a single superhydrophobic surface could be repeatedly used for > 100 times for all droplets used in this experiments (blank and target solutions as well as antibody-conjugated particle suspensions). In this sense, we believe this setup can replace the standard microwell plate assays, where liquids are in direct contact with surfaces (thus resulting in more biofouling). Liquid volume is smaller than that of microwell plate assays, which may be further lowered to nanoliter scale (presumably in oil immersion to prevent rapid evaporation).
We also demonstrated the rapid mixing of a merged droplet, by mechanically vibrating the wire with a vibration motor. Figure 7 shows the snapshots of this movie. Complete movie is also available as Additional file 1. The total cost of materials and supplies were less than $230 (excluding the experimenter's labor), which demonstrates the feasibility of this new droplet manipulation. Additionally, smaller stepper motors and integrated circuits (all commercially available) could greatly reduce the overall size of this setup.
To summarize, a proof-of-concept was demonstrated for the wire-guide droplet manipulations and subsequent bioanalysis. Detection limits were extremely low compared to the other immunoassay demonstrations in conventional microfluidics; > 1 ng mL-1 for proteins [13–15], > 100 TCID50 mL-1 for viruses [16, 17], > 100 CFU mL-1 for bacteria [18, 19]. We hope this new method would substitute magnetofluidics by minimizing or even eliminating their complications, such as biomolecular adsorption, possible interferences by external electrical or magnetic field. The simplicity of this wire-guide manipulation is also demonstrated with automated, programmed movements of droplets.
Immunoglobulin G from murine serum, or mouse immunoglobulin G (mIgG), was chosen as a model protein (from Sigma-Aldrich; St. Louis, MO, USA). It was dissolved in 10 mM, pH 7.4 phosphate buffered saline (PBS) and various dilutions were made from a single stock solution. Bovine viral diarrhea virus (BVDV) was selected as a model virus pathogen. As its name indicates, BVDV causes diarrhea in cattle, leading to productivity loss and death. BVDV was cultured in Madin-Darby bovine kidney cells (MDBK) with appropriate tissue culture media (containing 5–10% fetal calf serum), followed by cell denaturation and centrifugal washing (from NVRQS; National Veterinary Research and Quarantine Service; Anyang, South Korea). Various dilutions were made from this stock solution; the tissue culture infectious dose 50 (TCID50) value was provided by the manufacturer (NVRQS). Finally, E. coli was selected as a model bacterial pathogen. Lyophilized E. coli K-12 powder was purchased from Sigma-Aldrich and cultured in brain heart infusion broth (from Remel; Lenexa, KS, USA) at 37°C for 20 h. The colony forming units (CFU) was evaluated by plating some of the above dilutions on eosin methylene blue agar (from DIFCO; Lawrence, KS, USA), incubating at 37°C for 20 h and counting the number of colonies with a light microscope (from Nikon; Tokyo, Japan). Serial dilutions from the stock solution were made using 10 mM phosphate buffered saline (PBS).
Latex particles were purchased from Bangs Laboratories (Fishers, IN, USA). These particles were highly carboxylated, with 10.3 Å2 parking area per carboxyl surface group, and a mean diameter of 920 nm according to the manufacturer's specifications. These particles should be mixed faster with target solution through their higher diffusivity, without using any surfactants, as we have recently demonstrated in microfluidic platforms . Maximum signals were obtained at 2 min, and all data points were taken at 2 min after merging two droplets. No significant evaporation was observed for this time frame. The particles were conjugated with three different polyclonal antibodies: anti-mIgG from Sigma-Aldrich, anti-BVDV from Jeno Biotech (Chuncheon, South Korea) and anti-E. coli from Abcam (Cambridge, MA, USA). These antibodies were conjugated to the particles by physical adsorption following the same protocol published previously . Antibody-conjugated particles were centrifuged twice, until no antibodies could be found in supernatants with the absorbance measurements at 280 nm. The surface coverage of antibodies was set to approximately 33% of its maximum possible value . The solid content of antibody-conjugated particles was 0.02% w/v.
Superhydrophobic surfaces made from nanocoatings of fluoropolymer on standard glass microscope slides were purchased from Surface Innovations (Durham, England). These nanocoatings create air pockets and put a microdrop in metastable Fakir state .
The Arduino board outputs simultaneously to three stepper motor driver boards, one per axis, through digital outputs integrated on the board. An original Nintendo controller wired into the digital inputs on the Arduino board allows the user to control the wire's position on-the-fly. The directional pad controls the X and Y axis movements, and the A and B buttons control the Z axis movement (i.e. insertion/retraction of a wire into a droplet). In conjunction with the controller, the system can be easily programmed to complete certain movements precisely and with consistency each time. The stage can be adjusted to ensure level surface conditions when working with superhydrophobic surfaces.
The authors are grateful to Mr. Jin-Hee Han at the University of Arizona for their help in antibody conjugation and E. coli cell counting, and Dr. Jae-Young Song at National Veterinary Research and Quarantine Service (NVRQS) in South Korea for helpful discussion in BVDV assay. Funding for this work was provided by NVRQS, award no. C-AD14-2006-11-00.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.