Anal. Chem. 2001, 73, 3776-3780
Technical Notes
Fluidics Cube for Biosensor Miniaturization James M. Dodson, Mark J. Feldstein, Dan M. Leatzow,† Layla K. Flack,‡ Joel P. Golden, and Frances S. Ligler*
Center for Bio/Molecular Science & Engineering, Code 6900, Naval Research Laboratory, Washington D.C. 20375-5348
To create a small, portable, fully automated biosensor, a compact means of fluid handling is required. We designed, manufactured, and tested a “fluidics cube” for such a purpose. This cube, made of thermoplastic, contains reservoirs and channels for liquid samples and reagents and operates without the use of any internal valves or meters; it is a passive fluid circuit that relies on pressure relief vents to control fluid movement. We demonstrate the ability of pressure relief vents to control fluid movement and show how to simply manufacture or modify the cube. Combined with the planar array biosensor developed at the Naval Research Laboratory, it brings us one step closer to realizing our goal of a handheld biosensor capable of analyzing multiple samples for multiple analytes. Fluid control is necessary for many systems capable of automated chemical and biochemical analysis. These systems typically require liquid samples, reagents, and buffers to be dispensed in a controlled manner. Making these analysis systems portable presents unique demands on fluidics systems that are not successfully met by currently available technology. These demands stem from the combined requirements of automation, compact size, and compatibility with unprocessed samples, especially for field operations or point-of-care applications. For laboratory-scale devices, there is an assortment of mechanical valves suitable for fluid handling and control. However, the size of these components makes them impractical for portable analysis systems. While small valves of analogous design have been developed and are commercially available, as the valve size is reduced, clogging by the components of complex sample matrixes becomes a limiting factor. The need for intermediate-scale fluid-handling systems has been identified.1 Among the developments in this area are pneumatic diaphragm valves integrated directly into the device’s fluidics channels. This approach provides fluid regulation while * Corresponding author: (tel) 202-404-6002; (fax) 202-404-6009; (e-mail)
[email protected]. † Current address: Department of Chemical Engineering, Washington State University, Pullman, WA 99164. ‡ Hoffstra University, Hempstead, NY 11549. Current address: The College of William & Mary, Williamsburg VA 23186. (1) VerLee, D.; Alcock, A.; Clark, G.; Huang, T. M.; Kantor, S.; Nemcek, T.; Norlie, J.; Pan, J.; Walsworth, F.; Wong, S. T. Technical Digest. Solid-State Sensor and Actuator Workshop, 1996; pp 9-14.
3776 Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
adding only slightly to the overall size of the system. However, diaphragm-based valves can suffer from sticking, clogging, and performance loss with diaphragm aging. Valveless fluid control has also been developed, thus eliminating the problem of valve clogging by suspended contaminants. For example, pressure control and pressure differentials can switch fluid flow between microchannels.2 This method of fluid control is based on the application and regulation of differential pressures to each fluid channel and is only applicable in the low Reynolds number regime. The regulation of differential pressures makes the design inherently complex, and further, the requirement for a pressure source and regulators limits the feasibility of this method for portable instrumentation. The limitation to the low Reynolds numbers regime makes the method impractical for the control of aqueous fluids in channels greater than ∼100 µm.3,4 Although valves may not be clogged in these approaches, the fluid channels themselves are likely to be clogged by suspended contaminants. Electrokinetic pumping and switching have also accomplished valveless fluid control in micrometer-scale devices.5,6 Similarly, however, such designs are limited to the low Reynolds number regime, where micrometer-scale channels are prone to clogging. Further, these methods require large driving potentials, typically on the order of 1 kV, and fluid flow can be drastically affected by sample components adhering to the wall of the channel. To develop portable, handheld biochemical analysis devices, which operate reliably with environmental and clinical samples, a compact means of fluid control needs to be developed. Fluid control that does not require passage of complex fluids through valves would be the ideal method for analytical devices in the regime between benchtop laboratory systems and total microanalysis systems. Our efforts, focused in this intermediate regime, have produced a novel fluid-handling and delivery device: the fluidics cube. We have coupled this to a planar array7,8 (2) Brody, J. P. U.S. Patent 5,726,404, 1998. (3) Brody, J. P.; Yager, P. Technical Digest. Solid-State Sensor and Actuator Workshop, 1996; pp 105-108. (4) Brody, J. P.; Yager, P.; Goldstein, R. E.; Austin, R. H. Biophys. J. 1996, 71, 3430-3441. (5) Manz, A.; Verpoorte, E.; Raymond, D. E.; Effenhauser, C. S.; Burggraf, N.; Widmer, H. M. In Micro Total Analysis Systems; van den Berg, A., Bergveld, P., Eds.; Kluwer Academic Publishers: Dordrecht, 1995; pp 5-27. (6) Manz, A.; et al. Adv. Chromatogr. 1993, 33, 1-67. (7) Wadkins, R. M.; Golden, J. P.; Pritsiolas, L. M.; Ligler, F. S. Biosens. Bioelectron. 1998, 13, 407-415. 10.1021/ac010168i CCC: $20.00
© 2001 American Chemical Society Published on Web 06/30/2001
less power is required. Furthermore, the thermoplastic is milled easily to CAD specifications and these designs are readily modified for retesting for different applications or improvements.
Figure 1. Fluidics cube design and component assembly. (A) Simple concept of the cube design: dedicated reservoirs for samples/ reagents on each side, channels and a common exit tube between the two sides. Our design contained six reservoirs per side in a staggered, close-packed arrangement to minimize cube size. (B) Fluidics module after assembly. The prepared waveguide was inserted in a cassette composed of an aluminum bottom bracket, PDMS gasket (not visible in photo), and black, plexiglass flow manifold. The cube was secured to the flow manifold using nylon screws that secured the aluminum bracket, protruded through the plexiglass, and entered the cube from below.
in order to automate sample analysis in the portable multianalyte sensor. The fluidics cube is a compact cluster of reservoirs and channels in a block of thermoplastic (poly(methyl methacrylate), PMMA). A pair of air vents, one on either side of the cube, regulates the flow of fluids from the reservoirs, through the channels, and onto the sensing surface of the detector. Figure 1 shows a schematic of only four reservoirs, but it demonstrates the concept. The fluid-filled reservoirs are sealed off and negative pressure is applied to the exit ports at the bottom of the cube. By opening an air vent, which couples to one side of reservoirs or the other, fluid will begin flowing from that side. The cube is made from stacked layers of thermoplastic, which when aligned and annealed or bonded into a single unit create a network of reservoirs and fluid channels. Using the cube creates numerous benefits. It eliminates tubing and interconnects that can leak, and it also eliminates fluid valves that can clog. The small reservoirs and short fluid path mean that smaller pumps can be used and (8) Feldstein, M. J.; Golden, J. P.; Rowe, C. A.; MacCraith, B. D.; Ligler, F. S. J. Biomed. Microdevices 1999, 1:2, 139-153.
MATERIALS AND METHODS Antibodies and Toxin. Staphylococcal enterotoxin B (SEB) and anti-SEB antibodies were obtained from Toxin Technologies (Sarasota, FL). To generate capture antibodies, a long-chain derivative of biotin, N-hydroxysuccinimidyl ester (EZ-Link NHSLC-Biotin; Pierce, Rockford, IL) was attached to the anti-SEB at a 10:1 biotin/protein ratio as recommended by the manufacturer. Labeled protein was separated from unincorporated biotin using a Bio-Gel P10 column, (Bio-Rad, Hercules, CA). Fluorescent tracer antibodies were prepared by labeling anti-SEB antibodies with Cy5 bisfunctional reactive dye (λex ) 649 nm, λem ) 670 nm, Amersham Life Science Products, Arlington Heights, IL) according to the manufacturer’s instructions. Dye-to-protein ratios ranged from 2.5 to 4.0. Fluidics Cube Design and Production. Cube layers were designed using MasterCam 8.0 software (CNC Software Inc., Tolland, CT) and were manufactured from 0.25-in. clear cast acrylic (AtoHaas North America, Inc., Philadelphia, PA) using a three-axis servorouter (Techno-Isel, Hyde Park, NY). Each layer of acrylic was milled to contain a hole, groove, or both. When the layers were aligned, the holes and grooves combined to form a three-dimensional network of channels and reservoirs. The cube was designed to contain a bank of sample reservoirs on one side and reagent reservoirs on the other with channels between the reservoirs (Figure 1A). Other features that were milled into the layers formed holes for alignment pins and holes that were used to attach the cube to the flow manifold. To form a solid cube, the layers were secured with stainless steel pins, then lightly clamped in a vise, and heated to 140 °C for 3 h. After cooling to room temperature, stainless steel tubing was inserted into the 12 exit holes to create exit ports. The tubing was secured with a small amount of 5 Minute Epoxy (Devcon, Inc., Danvers, MA). After the epoxy had set, the tubing was cut to the desired length using a variable-speed rotary tool equipped with a cutoff wheel (Dremel, Inc., Racine, WI). Alternatively, the cube was created by applying Weld-On 3, an acrylic solvent cement, (IPS Corp., Gardena, CA) to a layer, then carefully placing the next layer on top of it with light manual pressure, and allowing the cement to dry. Layers were built up in this manner until the entire cube was created. After the layers were cemented into a cube, it was placed in a vise under light pressure and heated to 140 °C for 3 h. Each cube was tested for proper fluid flow and also checked for leaks between reservoirs and channels or to the exterior. Flow Manifold Design and Production. A flow manifold containing six channels and entry/exit holes for fluid passage was designed using MasterCam 8.0 software (CNC Software Inc.). The flow manifolds were manufactured from 0.25-in. clear cast acrylic (AtoHaas North America, Inc.) or black Lucite L cast acrylic (ICI Acrylics, Wilmington, DE) using a 3-axis servo router (TechnoIsel). In the case of the manifold containing the PDMS gasket,9,10 the flow channels were 2.74 mm wide × 38.1 mm long and 2.54 (9) Rowe, C. A.; Scruggs, S. B.; Feldstein, M. J.; Golden, J. P.; Ligler, F. S. Anal. Chem. 1999, 71, 433-439. (10) Leatzow, D. M.; Dodson, J. M.; Golden, J. P.; Ligler, F. S. Biosens. Bioelectron, in press.
Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
3777
mm deep. The PDMS barrier separation between each channel measured ∼1.1 mm. Component Assembly. The flow manifold with the PDMS gasket was attached to the glass waveguide through compression in a cassette assembly.10 The assembly included the acrylic flow manifold with integrated PDMS gasket, the glass waveguide, a bottom aluminum mounting bracket, and nylon mounting screws (Figure 1B). The waveguide was held in place between the flow manifold and the mounting bracket by tightening the mounting screws. The cube was attached to the top of the flow manifold by a pair of the nylon mounting screws, which extended above the top surface of the manifold and entered the cube from below. Assay Protocol. Custom-manufactured silver-clad waveguides8 (Opticoat Associates, Protected Silver) were cleaned, coated with NeutrAvidin (Pierce, Rockford, IL), and patterned with antibodies as previously described.8,9 Following component assembly, the assay module (cube, flow manifold, and waveguide) was placed on the detector. To verify the system’s integrity and block nonspecific binding, the cube’s reservoirs were filled with PBSTB, which was drawn through the flow manifold with negative pressure from a downstream peristaltic pump. During the PBSTB flow-through, images of the waveguide were captured to check for flow and leaks. To assess nonspecific binding, 200 µL of 10 µg/mL Cy5-anti-SEB antibody solution were loaded into one bank of reservoirs and drawn through the system. The system was flushed with 250 µL of PBSTB per reservoir and an image was captured. The image showed negligible binding of Cy5-anti-SEB antibody to the waveguide or to the edges of the flow manifold touching the waveguide. Once the system checks were completed, 250 µL of each dilution of SEB (0-50 ng/mL) was loaded into one bank of reservoirs. The other bank of reservoirs was loaded with 250 µL of Cy5-anti-SEB, fluorescent tracer antibody, at 10 µg/mL. By opening and closing the appropriate air vents, the reservoirs containing the tracer antibodies were closed and the reservoirs containing the samples were opened, allowing only the sample reservoirs to flow. An off-board peristaltic pump at a flow rate of ∼0.35 mL/min was used to create negative pressure downstream of the assay module. After 5 min, the sample reservoirs had been drained and the vents were then closed. The tracer antibody reservoirs were then opened, and flow was confirmed by capturing an image during the flow-through. After 5 min, the antibody reservoirs had been drained. The antibody reservoirs were filled with PBSTB, and the buffer was flushed through the flow manifold. A final image, demonstrating detection of various concentrations of SEB, was captured and analyzed. RESULTS Regulation of Fluid Flow Using a Pressure Relief Vent. Fluid selection from one or more fluid reservoirs was provided through actuation of a pressure relief vent associated with each reservoir. The pressure relief vent, which is crucial to the cube’s utility, was demonstrated in a two-reservoir system (Figure 2). Each reservoir, R1 and R2, had a fluid output connected through a T-junction into a fluorescence detector. R1 contained water and R2 contained a 60 nM aqueous solution of fluorescent dye Cy5. Each reservoir was sealed to the atmosphere except that they were connected to relief microvents (LFAA12034, The Lee Co.), V1 and V2, respectively. The default closed position of a valve 3778 Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
Figure 2. Demonstration of pressure relief vent with a two-reservoir system. (A) Reservoirs, R1 and R2, contained water and the fluorescent dye Cy5, respectively, with output through a T-junction into a fluorescent detector. Relief vents, V1 and V2, were actuated in an alternating mode to open each reservoir to atmospheric pressure. (B) Fluorescent signal was measured as the valves periodically switched open and closed.
caused the given reservoir to be sealed from the atmosphere. The relief vents could be individually actuated (via a 12-V signal) to open a given reservoir to atmospheric pressure. A peristaltic pump, running at 1.5 mL/min, was used to draw fluid from the reservoirs and through the detector to waste. In this configuration and as described above, when V1 was open the fluid in R1 (water) would be drawn through the detector by the pump. The fluid in R2 (Cy5) would not flow because of greater resistance to flow resulting from the inability of air to backfill that reservoir. The fluid in R2 would flow, exclusively, when V1 was closed and V2 opened. When V1 was opened, water was pulled through the system and the fluorescence detector recorded a signal level of zero. However, when V1 was closed and V2 was opened, the fluorescence signal rose sharply since the Cy5 solution in R2 was drawn through the system and detected. The slight delay of the signal rise as compared to the opening of V2 was due to the finite distance that the fluid needed to flow from the T-junction to the detector. The tailing of the signal level to zero when V1 was open was attributed to the detection of residual Cy5 in the fluid channels being washed out by the water in R1. Cube Design and Function. While the test of the pressure relief vent worked well, premature mixing of sample and reagent upstream of the waveguide surface occurred. The cause of this problem was the shared exit tube for the fluids and capillary action drawing the sample into the exit channels from the closed
Figure 4. SEB assay using the fluidics cube. The fluorescent spots were quantitated using Scion Image and the adjacent background fluorescence subtracted.10 Mean fluorescence intensity increased as a function of SEB concentration. Six readings were taken at each concentration and the mean (( SD) was plotted. Figure 3. Cube design variations. Cross-sectional schematics showing variations in the cube’s exit port design. The original design failed to keep fluids from mixing before being drawn out to the flow manifold. To address this problem, while maintaining a single exit port, the intermediate design was created and tested. It, too, failed. Thus, the final design with two exit ports was created. This required modification of the flow manifolds to accept two rows of ports.
reservoirs of fluorescent reagent. A redesign of the cube was the simplest approach to solving this problem. The cube was redesigned so that the sample and reagent did not flow through a common channel until further downstream (Figure 3). The intermediate design was also unsatisfactory since early mixing still occurred. The final design completely separated the sample and reagent channels, resulting in two exit ports per reservoir pair (Figure 3). This modification required a change in the flow manifold to accommodate the dual exit ports, but the final design eliminated the problems with mixing of the sample and fluorescent reagent. SEB Detection. Dilutions of SEB were loaded into the reservoirs of the cube and assayed in our detector system. The fluorescent signal of the spots were determined by subtracting the mean fluorescent intensity of the adjacent regions with no capture antibody (nonspecific binding) from the fluorescent intensity of the region including the capture antibody. The net fluorescence for each capture antibody spot was plotted as a function of SEB concentration. Values were proportional to SEB concentration (Figure 4) and standard deviations comparable to those obtained using large, conventional pumps and valves (data not shown). The system was able to detect concentrations of SEB from 5 to 50 ng/mL in a 200-µL sample, i.e., 1-10 ng (36-360 fmol) of SEB. Six samples were analyzed simultaneously with six assay replicates of each sample (i.e. six separate assay spots) in under 20 min. DISCUSSION One of the challenges in our pursuit of a handheld, multianalyte detector has been handling the fluid samples and reagents. Our approach was to create a small, easily modified and manufactured cube, which could adequately deliver sample and reagent to a
Figure 5. Multichannel analysis system and negative pressure sources. An example of an alternative embodiment of the cube system. The schematic shows two reagents, in reservoirs R1 and R2, drawn to the sensor surface.
detector surface, while avoiding the typical problems associated with small-scale fluid-handling systems. There are no examples in the literature of a similar approach to this common fluidhandling problem. Our combination of pressure relief vents and reservoirs in a small cube resulted in highly responsive control of fluid flow. Since multiple relief vents can be used to control multiple banks of reservoirs, a single negative pressure source (e.g., peristaltic pump) may be all that is required to draw fluids through the cube. If more than one of the relief vents is open to allow gas backfill, then fluid may be drawn from multiple reservoirs simultaneously. Also, a systemwide relief vent can be incorporated to disable fluid flow by providing centralized depressurization of the fluid circuit. Complex arrangements where a larger number of fluids can be selectively dispensed without passing through any valves are also possible. For example, to analyze three samples simultaneously in a multichannel bio/chemical sensor where each sample requires processing with two reagents, an embodiment as shown in Figure 5 is possible. In this configuration, samples 1, 2, and 3 (S1, S2, S3) can be drawn into a multichannel analysis system by opening a single pressure relief vent that is connected to all three sample reservoirs. When this relief vent is closed and the relief vent on reagent reservoir 1 or 2 (R1, R2) is open, the sample flow will cease and the appropriate reagent will be dispensed into each of the three samples reservoirs and then into each of the three Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
3779
analysis channels. To achieve parallel dispensing, as illustrated by this figure, each channel needs an independent negative pressure source. This configuration could be scaled to control any number of samples and reagent reservoirs. Our work on the cube and flow cell was guided by our idea of a disposable assay module. We envision mass production of an injection-molded component that includes both the cube and flow manifold. The waveguides, coated with the appropriate capture antibodies, would be inserted or attached to this combined component prior to the assay. To reduce the work for the device operator, thus reducing chances for error, the cube could contain lyophilized fluorescent tracer antibody that could be reconstituted in buffer at the time of use. The need for a simple, handheld multianalyte detector for field use or point-of-care use is real.
3780
Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
Small, lightweight, disposable components that require minimal operator manipulation will be important in the development of such a device. ACKNOWLEDGMENT The Office of Naval Research and NASA funded this work. J.M.D. was supported by an Associateship from the National Research Council. The views expressed here are those of the authors and do not represent those of the U.S. Navy, the U.S. Department of Defense, or the U.S. government. Received for review February 8, 2001. Accepted May 8, 2001. AC010168I
