Functional Membrane-Implanted Lab-on-a-Chip for ... - ACS Publications

This could enable us to fabricate a miniaturized lab-on-a-chip device that totally performs various processes, for instance, pretreatment of a nanolit...
0 downloads 0 Views 516KB Size
Download PDF
Anal. Chem. 2005, 77, 7901-7907

Functional Membrane-Implanted Lab-on-a-Chip for Analysis of Percent HDL Cholesterol Joo-Eun Kim,† Joung-Hwan Cho,†,‡ and Se-Hwan Paek*,†,‡,§ 1Graduate

School of Biotechnology, Korea University, 1, 5-ka, Anam-dong, Sungbuk-ku, Seoul 136-701, Korea, BioDigit Laboratories Corp., Korea University, Biotechnology Building, Technology Incubation Center, 1, 5-ka, Anam-dong, Sungbuk-ku, Seoul 136-701, Korea, and Department of Biotechnology, Korea University, Jochiwon, Choongnam 339-800, Korea

A functional lab-on-a-chip has been developed for simultaneous quantitative analyses of high-density lipoprotein (HDL) cholesterol (HDL-C) and total cholesterol (totalC) in a submicroliter plasma sample. The analytical device was fabricated by placing commercial membranes, traditionally used for rapid diagnostics, within microfluidic channels engraved on the surface of a plastic chip. The concentration of HDL-C was measured using enzymatic reactions to produce a colorimetric signal after separation of the single plasma lipoprotein from a mixture. Two small pieces of different membrane pads were used to provide each group of reagents, for HDL separation and enzyme reactions, deposited within their tiny pores in a dry state. To maintain a connection toward the capillary action of the medium, the pads were arranged in a sequence within the fluidic channel that controlled the inlet and outlet of the flow. Upon the addition of a sample, the fluid was delivered through the pads of the chip and a color signal was subsequently generated in proportion to the concentration of HDL-C. The level of total-C was concurrently determined by following identical processes, except absent HDL separation. The two signals were simultaneously determined by employing optical detectors based on transmittance of a light. Such total analyses were completed within 2 min, and the sample sizes were able to be reduced to 0.4 µL for HDL-C and 0.1 µL for total-C, enough to cover the clinically required dynamic ranges. Rapid analytical devices based on chromatography using the lateral flow of medium through micropores present within the matrixes of membrane pads have been conventionally applied for the diagnoses of various diseases and symptoms.1-3 Despite their simplicity in use, one of the major drawbacks in routine, frequent application has been the induction of severe pain, in the case of using whole blood for specimens, because of a large amount of * To whom correspondence should be addressed. Tel: 82-2-3290-3438. Fax: 82-2-927-2797. E-mail: [email protected]. † Graduate School of Biotechnology. ‡ BioDigit Laboratories Corp. § Department of Biotechnology. (1) Paek, S. H..; Lee. S. H.; Cho, J. H.; Kim Y. S. Methods 2000, 22, 53-60. (2) Paek, S. H.; Jang, M. R.; Mok, R. S.; Kim, S. C.; Kim, H. B. Biotechnol. Bioeng. 1999, 62, 145-153. (3) Kasahara, Y.; Ashihara, Y. Clin. Chimi. Acta 1997, 267, 87-102. 10.1021/ac0510484 CCC: $30.25 Published on Web 11/12/2005

© 2005 American Chemical Society

sampling. To reduce the sample size, membrane pads can typically be cut smaller than 4 mm in width, which would make it difficult to hold them in a precise arrangement. This causes a low reproducibility of analysis and inaccuracy in detection. For a device utilizing a flow-through mode,4 the same problems must be addressed when membranes are of smaller sizes. These are probably the major reasons that products handling low-capacity samples have not yet appeared in the market. The sample volume required by current, commercially available rapid analytical devices is typically in the range of 15-200 µL.5 As a trend of recent development in analytical devices, a technology of microelectrical, mechanical systems6-8 has been used for the fabrication of microfluidic channels9,10 and microscopic structures11 on a variety of solid surfaces. This could enable us to fabricate a miniaturized lab-on-a-chip device that totally performs various processes, for instance, pretreatment of a nanoliter sample, physical separation of biomolecules, and generation of a signal in proportion to the analyte concentration.12-14 Such total analysis may be carried out on a 1 × 1 mm sized plastic chip or possibly one that is even smaller. However, since the present status of this technology remains undeveloped in some aspects, such as reproducibility in mass production of the chip, the time of its practical application appears considerably delayed.11 Both analytical resources mentioned, membranes used for rapid analysis and microfluidic channels enabling the miniaturization of a device, can be combined in order to achieve a practical lab-on-a-chip capable of handling quite a small sample. Many different commercially available membranes can perform various functions that may be needed for analyses, such as filtration, ion exchange, reagent release, laminar flow, and absorption.2,3 The (4) Chu, A. E. United States Patent US 6,284194 B1, 2001. (5) Tuˆdos, A. J.; Besslin, G. A. J.; Schasfoort, R. B. M. Lab Chip 2001, 1, 8395. (6) Ehrfeld, W. Electrochim. Acta 2003, 48, 2857-2868. (7) Manz, A.; Graber, N.; Widmer, H. M. Sens. Acuators, B 1990, 1, 244-248. (8) Erickson, D.; Li, D. Anal. Chimi. Acta 2004, 507, 11-26. (9) Guber, A. E.; Heckele, M.; Herrmann, D.; Muslija, A.; Saile, V.; Eichhorn, L. Chem. Eng. J. 2004, 101, 447-453. (10) Fujii, T. Microelectron. Eng. 2002, 61/62, 907-914. (11) Schueller, O. A.; Btittain, S. T.; Whitesides, G. M. Sens. Acuators, A 1999, 72, 125-139. (12) Sato, K.; Hibara, A.; Tokeshi, M.; Hisamoto, H.; Kitamori, T. Adv. Drug Delivery Rev. 2003, 55, 379-391. (13) Weigl, B. H.; Bardell, R. L.; Cabrera, C. R. Adv. Drug Delivery Rev. 2003, 55, 349-377. (14) Wang, J. Anal. Chim. Acta 2004, 507, 3-10.

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005 7901

membranes can be cut to widths of 1 mm or narrower and then installed within the channels of a plastic chip. This approach facilitates precise arrangement and assembly of the small pieces of membranes together for the fabrication of a functional lab-ona-chip. Such a novel device would offer three advantages in addition to sample reduction: (1) realization of variable functions by selecting appropriate membranes mentioned; (2) implantation of membranes as parts of a complete channel for total analysis; and (3) transfer of medium by capillary action without the assistance of an external force. In this study, we have demonstrated the concept by applying it for the quantitative measurements of high-density lipoprotein (HDL) cholesterol (HDL-C) and total plasma cholesterol (totalC). The level of HDL-C in plasma and, alternatively, the ratio of the two analyte concentrations have been reported to act as prognostic indicators of coronary heart disease,15,16 as well as cerebral apoplexy.17,18 To determine HDL-C, HDL must be separated from a mixture of plasma lipoproteins, after which the cholesterol content can be measured by sequential enzymatic reactions. Total-C can also be measured using the same process without HDL separation. We carried out both analytical processes within two different channels present on the surface of a plastic chip and installed with functional membrane pads inside. The cholesterol levels were measured based on transmittance of light through separate aqueous media colored in proportion to each analyte concentration. EXPERIMENTAL SECTION Materials. Poly(methyl methacrylate) (PMMA) was obtained from LG Chem (PMMA IF870, Seoul, Korea). Light-emitting diodes (LED; KA3-R2311T, 635 nm) and photodiodes (S178712) were purchased from Idea Electric (Seoul, Korea) and Hamamatsu (Hamamatsu, Japan), respectively. Human very lowdensity lipoproteins (VLDL; 439 mg of cholesterol/100 mL), lowdensity lipoproteins (LDL; 1,067 mg of cholesterol/100 mL), HDL (381 mg of cholesterol/100 mL), bovine lipoproteins (1950 mg of cholesterol/100 mL.), horseradish peroxidase (HRP; 251 units/ mg; EC 1.11.1.7), and delipidated serum were purchased from Calbiochem (La Jolla, CA). Cholesterol oxidase (CO; microbial product, 400 kunits/mg; EC 3.1.1.13) and cholesterol esterase (CE; from Nocardia species, 500 kunits/mg; EC 1.1.3.6) were obtained from Kikkoman (Noda, Japan). Cholesterol calibrators (100-400 mg/100 mL of bovine lipoprotein cholesterol), sodium cholate, sodium tungstate, dextran sulfate (sodium salt form, MW 50 000), magnesium chloride, 3,3′,5,5′-tetramethylbenzidine‚2HCl (TMB), Triton X-100, bovine serum albumin (BSA; heat-shocked, fraction V powder, minimum 98% purity), ethylenediaminetetraacetic acid (EDTA), trehalose, and 2-(N-morpholino)ethanesulfonic acid (MES) were obtained from Sigma-Aldrich (St. Louis, MO). Poly(vinyl acetate) (PVA)-bound and plain glass-microfiber membranes (F147-11 and Rapid 27Q), and anion exchange membrane (DE81) were purchased from Whatman (Singapore). Thin, adhesive polymer films were supplied by G&L Inc. (31401, San Jose, CA). (15) Barter, P.; Kastelein, J.; Nunn, A.; Hobbs, R. Atherosclerosis 2003, 1-17. (16) Despres, J.; Lemieux, I.; Dagenais, G.; Cantin, B.; Lamarche, B. Atherosclerosis 2000, 153, 263-272. (17) Frick, M.; Elo, O.; Happa, K.; Hcinonen, O.; Heinsalmi, P. N. Engl. J. Med. 1987, 317, 1237-1245. (18) Manninen, V.; Tekanen, L.; Koskinen, P.; Manttari, M.. Circulation 1992, 85, 37-45.

7902

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

Two different cholesterol analysis kits for HDL-C (HDL-Cholesterol EX, GN 20112) and for total-C (Cholesterol SP, GN 20125) and standards (GN 90999, Biocal-Multicalibrator) were obtained from Gernon (Barcelona, Spain). All other chemicals used were of analytical reagent grade. Fabrication of Functional Membrane Pads. (a) Blood Cell Filtration Pad. For removing blood cells in a specimen, a filtration pad was prepared using a PVA-bound glass-microfiber membrane (1 × 4 mm). The membrane was treated by immersion in 10 mM phosphate buffer, pH 7.4, (PB) containing 1% BSA, 0.1% Triton X-100, and 2 mM EDTA. It was then dried on a paper towel in an incubator maintained at 37 °C for 1 h. (b) HDL Separation Pad. To separate HDL from a mixture of plasma lipoproteins, a membrane pad conducting a chemical precipitation of negatively charged LDL and VLDL was fabricated. An anion exchange membrane (0.5 × 6.5 mm) was treated in PB containing BSA and Triton X-100 and dried on a paper towel at 37 °C for 1 h. One microliter of PB containing 0.3% dextran sulfate and 0.3 M magnesium chloride was added onto this pad, and the pad was dried. (c) Bypass Pad. In the case of total-C measurement, no separation of lipoproteins was necessary, and thus, a bypass pad was located in place of the HDL separation pad (see below for details). The pad was prepared using the same procedure as with Rapid 27Q membrane (0.5 × 6.5 mm), except EDTA was absent, as indicated in the fabrication of the blood cell filtration pad. (d) Signal Generation Pad. The concentration of lipoprotein cholesterol was measured by employing a consecutive reaction of enzymes. After pretreating a glass-microfiber membrane (2 × 2 mm) with methanol and drying for 30 min in the air, it was impregnated with 1.5 µL of PB containing 20% trehalose, 1% sodium tungstate, 5% sodium cholate, 300 units/mL CE, 300 units/ mL CO, and 500 units/mL HRP. After drying in the incubator at 37 °C for 1 h, 0.7 µL of 5 mg/mL TMB diluted in deionized water was loaded onto the center of the pad, which was then dried again under identical conditions. Construction of Lab-on-a-Chip for Cholesterol Analyses. Microchannels were engraved on the surfaces of PMMA, essentially enabling us to hold the functional membrane pads and deliver an aqueous medium (see Figure 1 for the overall structure). The depth of the channels was kept constant at 300 µm, except those for medium delivery were at a depth of 200 µm. Sample inlet ports and enzyme reaction cells were built by drilling holes 2 mm and 1 mm in diameter, respectively. Channels were made to install the membrane pads and deliver a medium from the enzyme pad to the reaction cell. The arrangement of the pads for HDL-C is shown in Figure 2, and an identical array was also used for total-C, except using the bypass pad rather than the HDL separation pad. The channels were then enclosed by covering with a thin adhesive film (16 × 12 mm). It is noted that the blood cell filtration pad was only used for demonstrating its function, as shown in Figure 2, and omitted in the construction of all other analytical devices. Analytical Procedures. (a) Lab-on-a-Chip. Using a pipet, a defined amount of sample (0.5 µL for HDL-C or 0.1 µL for totalC) was loaded onto the HDL separation pad for HDL-C, or the bypass pad for total-C, of the chip through the medium buffer inlet port. After absorption into the pad, 10 mM MES buffer (5

Figure 1. Plastic chips with microsized channels that can be installed with functional membrane pads (left) and a transmittance basedphotometric detector used for quantification of cholesterol in a sample (right). The chips were made of PMMA and used after spray-painting in black. The detector was constructed by aligning LED and photodiode on opposite sides.

Figure 2. Capillary migration paths of the plastic chip composed of different membrane pads and a microfluidic channel. Upon addition of a whole blood sample (1.5 µL), the separation of red blood cells using a lateral filtration membrane pad and migration of the filtrate were demonstrated. The chip without black-painting was shown for visualization.

µL), pH 5.3, was added into the medium buffer inlet port. A color signal was then measured at 100 s by placing the chip within a photometric detector or was continuously monitored in the case of time-response studies. The detector was constructed by alignment of the LED and photodiode on the opposite sides of the enzyme reaction cell of the chip (refer to Figure 1). LED was operated by furnishing 1.68 V from a regulated dc power supply (LP-303TP, Digital Electronics, Seoul, Korea). For signal processing, the signal detected on the photodiode was intensified on an amplifier (C2719, Hamamachu, Japan) and then converted to a digital signal using an analog-to-digital conversion chip (PCI-6024E, National Instrument, Austin, TX). It was finally recorded by software (LabView 6.0, National Instrument) installed in a Pentium III personal computer. (b) Gernon Products. Two different cholesterol analysis kits in the liquid phase, HDL-Cholesterol EX for HDL-C and Choles-

terol SP for total-C, were used for comparative studies. Each analysis (sample volume: 5 µL for HDL-C and 10 µL for total-C) was performed according to the guidelines of the manufacturer, and the final colorimetric signals were measured at the absorbance of 600 nm for HDL-C and 500 nm for total-C using a spectrophotometer (Clima Plus, RAL Technica para el Laboratorio, Barcelona, Spain). Prior to use, the meter was calibrated by employing standards according to the protocol provided by the manufacturer. (c) Bioscanner 2000. A portable biosensor device and test strips for HDL-C and total-C (Polymer Technology System, Indianapolis, IN) were used to compare analytical performance. The assay procedure followed the protocol provided by the manufacturer. In brief, the device was first calibrated by insertion of a chip supplied together with the test strips. After installation of a strip, a sample (15 µL) was added onto it for measurement. Performance Tests of the Lab-on-a-Chip. (a) HDL Separation. To test the effectiveness of the lab-on-a-chip toward HDL separation, two sets of standards for HDL-C concentration were prepared in a range of 0-100 mg/dL by serial dilution of a stock with delipidated serum. LDL was added into one set of the solutions to maintain a final, constant concentration of 200 mg/ dL cholesterol. The HDL-C standards with or without LDL were loaded into each lab-on-a-chip constructed and analyzed as presented in the analytical procedure above. Identical measurements in a group were repeated 10 times in the presence of LDL and 9 times in the absence of LDL. (b) Time Responses. Kinetic responses of the chip to different concentrations of lipoprotein cholesterol were obtained. Samples for HDL-C measurement were prepared by dilution of a HDL stock to standard concentrations with delipidated human serum, and a constant amount of LDL was added to maintain a final concentration of 200 mg/dL cholesterol. For total-C, bovine lipoproteins were used as samples by dilution with the serum. The same analytical protocol as indicated above was used for a continuous monitoring of the signal during a predetermined time interval. (c) Standard Curves. For analyzing samples, dose responses of the chip to each analyte concentration were obtained and plotted to prepare standard curves. By diluting a stock of human HDL with delipidated serum or bovine lipoproteins, standard samples were prepared in different concentration ranges of 25-85 mg/ Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

7903

dL for HDL-C and 110-350 mg/dL for total-C, respectively. These samples were analyzed according to the analytical protocol described above, and the signals were measured at 100 s after addition of the medium buffer. The same measurement at each analyte concentration was repeated six times, and the mean values were used for plotting. (d) Correlation Tests. Correlations between measured and estimated values for each analyte were examined to assess the accuracy and variations of the chips. Standard samples for HDL-C measurement were formulated by adding different classes of human lipoproteins in delipidated serum to contain 25-85 mg/ dL HDL-C, 20 mg/dL VLDL-C, and residual LDL-C to maintain a constant concentration of total cholesterol, 210 mg/dL. Standards for total-C were also prepared by dilution of a stock of bovine lipoproteins with the same medium. These samples were applied to the chip and analyzed for HDL-C and total-C, respectively, as explained above. Each analyte concentration was determined using the respective standard curves prepared as mentioned. The same experiment was repeated six times, and the mean values were used for plotting against the corresponding estimates. For comparison of the analytical performances, different test kits (Gernon products and Bioscanner 2000) were purchased from commercial sources and used for the analyses of the standard samples for both analytes as described above. The mean values of six repetitive measurements for each analyte concentration were finally plotted against the respective matching estimate. (e) Measurement of %HDL-C. Using a chip with dual channels for HDL-C and total-C, both analytes were simultaneously measured for the same sample. Standard samples were identical to the ones used for the correlation test of HDL-C measurements. The sample was loaded onto the chip and was simultaneously analyzed for both analytes. The concentrations of each analyte were determined using the standard curves prepared as mentioned. Four repetitions of the identical procedure were accomplished, the ratios of the two analyte concentrations in %HDL-C were obtained, and the mean value was plotted against the estimate. RESULTS AND DISCUSSION A plastic lab-on-a-chip installed with different membrane pads was constructed for simultaneous measurements of two different analytes, HDL-C and total-C, in this study, using a minimal amount of human serum (Figure 1, left). The membranes were placed within channels engraved on the chip surface by means of mechanical lithography and served as functional actors required by the device for filtration, ion exchange, reagent supply, and absorption. Differently from conventional media, such as beads made of various materials19-21,23,24 and filtering microstructures,22 membranes facilitated the transfer of aqueous solution by capillary action through the interstitial spaces without employment of an (19) Li, B.; Zhang, Z.; Zhao, L. Anal. Chim. Acta 2001, 445, 161-167. (20) Choi, J. W.; Ahn, C. H.; Bhansali, S.; Henderson, H. T. Sens. Acuators. B 2000, 68, 34-39. (21) Choi, J. W.; Liakopoulos, T. M.; Ahn, C. H. Biosens. Bioelectron. 2001, 16, 409-416. (22) Andersson, H.; Wijingaart, W.; Enoksson, P.; Stemme, G. Sens. Acuators. B 2000, 67, 203-208. (23) Lv, Y.; Zhang, Z.; Chen, F. Talanta 2003, 1-6. (24) Rosa, C. C.; Cruz, H. J.; Vidal, M.; Oliva, A. G. Biosens. Bioelectron. 2002, 17, 45-52.

7904

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

external driving force. For illustrating the utility of such a lab-ona-chip, we used it for point-of-care testing of %HDL-C as a prognostic indicator of atherosclerosis.25, 26 Analytical Concept Using a Membrane-Implanted Chip for %HDL-C. To determine %HDL-C, HDL-C and total-C have to be independently measured at the same time using an identical sample, such as whole blood. In the case of whole blood usage, removal of cellular components is necessary prior to quantification of the lipids. A further step of fractionation is carried out for isolating HDL from other plasma lipoproteins, and cholesterol species contained in the HDL fraction are quantified based on enzyme assays. Total concentration of cholesterol can be determined merely using the same procedure without separation of the lipoproteins. These processes were steered to be accomplished in a small scale, at which pains in drawing blood samples would be significantly alleviated. To this end, we designed a lab-on-achip such that two separate channels were allocated for each analyte, and functional membranes suitable for conducting the respective assignment were also selected among commercially available products. (a) Construction of an Analytical Chip. We constructed a chip composed of five different compartments for (1) cell filtration, (2) HDL separation, (3) enzyme reagent supply, (4) fluid delivery, and (5) enzyme reaction for signal generation. To carry out such processes in sequence, microsized channels were carved on a plastic chip, in which tiny membrane pads, as described on the respective functional mechanism below, were placed in each predetermined position (see Figure 2). The rest of the channels were used as a fluidic path of aqueous solution and a reaction cell in which a colorimetric signal proportional to the cholesterol dose was finally generated. If whole blood was used as the specimen, the installation of three distinct membrane pads was necessary. For excluding cells, we have chosen a pad filtering them in a lateral direction, which enabled us to deliver the filtrate successively to the contiguous pad (refer to the illustration in Figure 2). Among variable methods of HDL separation,27 mixed processes of ion exchange and chemical precipitation were adopted for rapid, efficient elimination of the other lipoprotein classes.28,29 It was possible to conduct the separation on an anion exchange membrane concomitantly with precipitation of LDL and VLDL by predeposited reagents (see below for details). For eventually producing a signal, a glassmicrofiber membrane was prepared by impregnating reagents, necessary for enzyme reactions (CE and CO) and consequent color generation (e.g., TMB), in a dry state. (b) Analytical Procedure. A constructed membrane-implanted plastic chip was utilized for the quantitation of cholesterol contained in a single class of lipoprotein such as HDL. Upon addition of whole blood into the sample inlet port, the sample was absorbed into the cell filtration pad by capillary action. This driving force transferred the filtrate into the next pad while cells remained (25) Gordon, D.; Probsfield, J.; Garrison, R.; Neaton, J.; Castelli, W.; Knoke, J. Circulation 1989, 79, 8-15. (26) Krauss, R.; Lindgren, F.; Williams, P.; Kelsey, S. Lancet 1987, 2, 62-66. (27) Rifai, N.; Warnick, G. R.; Dominiczak, M. H. Handbook of Lipoprotein Testing, 2nd ed.; AACC: New York, 2000; pp 220-244. (28) Warnick, G. R.; Benderson, J.; Albers, J. J. Clin. Chem. 1982, 28, 13791388. (29) Bairaktrari, E.; Elisaf, M.; Katsaraki, A.; Tsimihodimos, V.; Tselepis, A. D.; Siamopoulos, K. C.; Tsolas, O. Clin. Biochem. 1999, 32, 339-346.

Figure 3. Progressive signal generation (upper diagram) as a result of enzymatic decomposition of HDL cholesterol after HDL separation from the blood filtrate in Figure 2 and responses of the chip (lower diagram) to different concentrations of HDL-C. Chemicals needed for HDL separation and various reagents, including enzymes, for signal generation were already present on the respective membrane pad in a dry state. The chips were used without black-painting only for the purposes of photography.

stagnant by filtration (see the red zone in Figure 2). The plasma lipoproteins were then carried into the HDL separation pad by a medium flow from the inlet port by addition and were fractionated by the ionic interaction and precipitation process as described above, which separated only HDL from a mixture. This HDL fraction moved into the enzyme reagent supply pad and instantly dissolved the components (e.g., detergent, enzymes, and the chromogenic substrate, TMB) impregnated beforehand in a dry state (Figure 3, top). The lipoprotein particles were ruptured by detergent, and cholesterol species consequently released were decomposed by two enzymes, CE and CO. Hydrogen peroxide produced from these reactions acted as a substrate of HRP in the presence of TMB. This chromogen was oxidized as a result of the catalytic reaction of HRP, which produced a blue color as a signal. The mixture gradually migrated, via the fluid delivery channel, into the enzyme reaction cell where the signal generation was completed. The color increased at an initial duration of time and reached a steady state within a couple of minutes. The signal intensity was proportional to the concentration of HDL-C (Figure 3, bottom), which can immediately be quantitatively determined based on photometric methods, such as the measurement of transmittance of a light,30,31 as shown in Figure 1 (right). (30) Ceriotti, L.; Weible, K.; Rooij, N. F.; Verpoorte, E. Microelectron. Eng. 2003, 67/68, 865-871. (31) Du, W.; Fang, Q.; He, Q.; Fang, Z. Anal. Chem. 2005, 77, 1330-1337.

Determination of Optimal Analytical Performances. The main purpose of this study was to develop an analytical chip consuming a minimal sample. Reduction of the chip, however, could be limited by the minimum size of the membrane pad that can be precisely cut and physically manageable for insertion into the channels. Rather than striving for miniaturization, we have selected a pad width of 0.5 mm for the first demonstration of such an analytical device. It is also noted that a delipidated serum, instead of whole blood, was used as medium in preparing standard samples of cholesterol in the following studies. (a) HDL Separation. The lipoproteins with lower densities, LDL and VLDL, contain a unique, hugely negatively charged apolipoprotein, apo B-100, on the particle surface.27,32 A selective precipitation of the species can occur in the presence of divalent cations (e.g., magnesium chloride) and polyanions (e.g., dextran sulfate) in solution.28,29 Divalent ions used in excess bind to apo B-100 and link this protein to a polyanionic polymer by charge interactions. Multiple cross-linking of the lipoprotein particles causes precipitation, and these agglomerates are then caught on the surfaces of an anion exchange membrane while carried by a medium flow. The separation efficiency of HDL from a mixture was regulated by adjustment of the concentrations of the two ionic precipitating (32) Rifai, N.; Warnick, G. R.; Dominiczak, M. H Handbook of Lipoprotein Testing, 2nd ed.; AACC: New York, 2000; pp 1-29.

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

7905

Figure 4. Comparison of dose responses of the chip to HDL-C contained in two different sample preparations, i.e., a mixture of HDL and LDL, and HDL only. The two results showed a consistency regardless of the presence of LDL in the samples, which indicated that the conditions used for HDL separation were optimized. Coefficients of variation of repetitive measurements were within 8.8% with the mixture samples and 5.9% with those containing HDL only.

agents and the size of the membrane pad. To optimize these parameters, samples were prepared to contain HDL in different cholesterol concentrations, while LDL remained constant, and dose responses of the chip to HDL-C were obtained. The results were compared with those from comparative control runs with the same samples, but in the absence of LDL. The optimal conditions for the parameters were determined such that the two dose-response results were consistent (Figure 4). During such optimization, an excess of magnesium chloride was used for complete precipitation of the apo B-100-containing lipoproteins, which, however, adversely caused an inhibition of the enzyme reactions in the next step. This problem was eventually overcome by employing a chelating agent (e.g., EDTA27-29) in the same pad. (b) Signal Generation. For signal generation proportional to the amount of HDL-C, the HDL fraction separated as described above was carried by the medium into the enzyme reagent supply pad. The HDL separation was not applied in the case of measuring total-C. Since the dynamic ranges of both cholesterol measurements should correspond to the extent of respective clinical concentration, two variables, sample volume and enzyme concentrations, were adjusted to satisfy this requirement. For HDL-C, we kept the amounts of enzymes in excess and then determined a minimum sample size (e.g., 0.4 µL) that would yield a signal generation in the clinical range. To the contrary, for total-C, the amounts of the reagents were reduced to control the levels of the signal produced with the smallest sample size (e.g., 0.1 µL) manageable by pipetting. In addition, two stabilizers for TMB, trehalose and sodium tungstate, were supplemented.27,29

Under the optimal conditions, we derived time-response curves of the plastic chip for various cholesterol doses (Figure 5). The signals measured on the basis of light transmittance were initially transient along the time course of monitoring during approximately the first 50 s after addition of the medium buffer. In this time interval, they were even inconsistent in their relative positions of each signal. This may be caused by a variation in the speed and flow distribution of the medium coming into the enzyme reaction cell via the fluid delivery channel. Nonetheless, the responses eventually approached each steady state in an arranged manner according to the cholesterol concentration. The values at each state were able to be measured as signals at 100 s. The entire spans of the signal changes covering complete clinical dose ranges (0-100 mg/dL for HDL-C and 0-400 mg/dL for totalC)33 were approximately 2 V for HDL-C (Figure 5, left) and 2.5 V for total-C (right). Characterization of the Lab-on-a-Chip as an Analytical Tool. The signals measured by the lab-on-a-chip for various standard samples prepared from stocks of human HDL for HDL-C and bovine lipoproteins for total-C were correlated with the respective estimated value (Figure 6). Their correlations were quite high for HDL-C (left) when compared with the standard line, while some deviations were shown for total-C in an elevated dose range (right). To grade such performances of the chip, we selected two comparative, commercially available analytical systems, Gernon and Bioscanner 2000, which consumed relatively large amounts of sample, 100 and 15 µL, respectively. The former was a laboratory version product that manually carried the consecutive reactions in the liquid phase, and the latter was a point-of-care testing (POCT) version that used vertically stacked membrane pads. The liquid-phase system conducted the worst analyses of the three. This was probably due to errors in manual handling of the reagents (see Figure 6). The POCT version showed a slightly worse performance in measuring HDL-C than did the chip, but a better performance for total-C. From such comparisons, the analytical chip requiring a minimal sample appeared to have a capability to perform accurate measurements of different cholesterols. Since the analytical performances of the chip toward the respective analyte were quite satisfactory, we further tested its capability for simultaneous measurements of both analytes in the same sample. To prepare standards that closely imitated clinical specimens, different classes of lipoproteins from human sources were mixed to contain variable HDL-C between 25 and 85 mg/ dL, a fixed VLDL-C, and residual LDL-C for keeping total-C

Figure 5. Time-response curves derived from the chip at different concentrations of HDL-C (left, serum sample volume 0.4 µL) and total-C (right, sample volume 0.1 µL). After addition of the medium buffer, the color signal produced in the enzyme reaction cell began to be monitored by measurement of transmittance and reached a steady state within ∼100 s, although transient fluctuations were shown at the beginning. 7906

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

Figure 6. Comparison of the correlation curves of the lab-on-a-chip with those obtained from other commercially available analytical systems (Gernon and Bioscanner 2000). Using the same samples, their analytical performances were compared to each other in terms of accuracy and scatter.

Figure 7. Simultaneous measurements of HDL-C and total-C using the lab-on-a-chip and assessment of its accuracy in measuring the ratio of HDL-C over total-C, i.e., %HDL-C. Standard deviations calculated from four replicate measurements are shown and were within 8%.

constant (210 mg/dL). The same sample was applied separately on each channel of the chip for analyses, and the concentrations of HDL-C and total-C were eventually determined using the respective standard curve. The concentration ratios of HDL-C over total-C, i.e., %HDL-C, obtained from each sample were plotted against the estimated values (Figure 7). The results showed a high correlation (correlation coefficient 0.994) with the straight line with the slope of unity. In conclusion, a membrane-implanted lab-on-a-chip offering a minimal sample requirement and analytical functions necessary for simultaneously measuring HDL-C and total-C, yielding %HDL-C (33) Hainline, A.; Karon, J.; Lippel, K. Lipid Research Clinics program, and Lipoprotein Analysis, 2nd ed.; AACC: Bethesda, MD, 1982.

as a prognostic indicator, has been constructed. The chip led the sample flow through the channel merely by capillary action without using an external driving force, which would allow the use of the device for POCT. Since the device is a miniaturized version for sample reduction that would alleviate a refusal against a finger prick, it would be suitable for frequent testing of the blood circulatory system of human body that, particularly, needs management such as diet and exercise varying the concentration of HDL cholesterol. In parallel with such positive aspects, its analytical performances were quite acceptable under optimal conditions toward, for instance, accuracy, reproducibility, response time, and dynamic range compared to those of commercial products. Although some deviation was observed at a high concentration range of total cholesterol, probably, due to nonlinearity of the calibration curve, the chip would be a valuable tool for the prognosis of a circulatory illness for subjects, particularly, looking normal in total cholesterol concentration, typically, less than 200 mg/dL. In the next stage of development, the chip will be supplemented with additional functions of cell filtration and a single-time sample application for both analytes. ACKNOWLEDGMENT This study was supported by a grant (01-PJ1-PG4-01PT02-0009) from the Korea Health 21 R&D Project, Ministry of Health & Welfare, Republic of Korea. Received for review June 14, 2005. Accepted October 18, 2005. AC0510484

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

7907