Novel Dense CO2 Technique for β-Galactosidase Immobilization in

Biomacromolecules 2008, 9, 1027–1034

1027

Novel Dense CO2 Technique for β-Galactosidase Immobilization in Polystyrene Microchannels Jeffrey LeClair Ellis,† David L. Tomasko,*,† and Fariba Dehghani‡ Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, Ohio 43210, and School of Chemical and Bimolecular Engineering, University of Sydney, Sydney, 2006, Australia Received December 5, 2007; Revised Manuscript Received January 9, 2008

In this study we design new fabrication techniques and demonstrate the potential of using dense CO2 for facilitating crucial steps in the fabrication of polymeric lab-on-a-chip microdevices by embedding biomolecules at temperatures well below the polymer’s glass transition temperature (Tg). These new techniques are environmentally friendly and done without the use of a clean room. Carbon dioxide at 40 °C and between 4.48 and 6.89 MPa was used to immobilize the biologically active molecule, β-galactosidase (β-gal), on the surface of polystyrene microchannels. To our knowledge, this is the first time dense CO2 has been used to directly immobilize an enzyme in a microchannel. β-gal activity was maintained and shown via a fluorescent reaction product, after enzyme immobilization and microchannel capping by the designed fabrication steps at 40 °C and pressures up to 6.89 MPa.

Introduction Lab-on-a-Chip Applications. In recent years researchers have decreased the size of diagnostic biomedical devices. One category of these devices is lab-on-a-chip (LOC), which have channels for fluid flow with a diameter on the order of micrometers down to nanometers. LOC microdevices are gaining popularity due to their wide variety of diagnostic uses, fast sampling time, and ease of use. The miniaturization of these fluidic devices has reduced the required quantity of sample for analysis, the amount of human labor, and the time for analysis. The devices are designed to execute sampling, sample pretreatment, mixing, separation, and detection all on a single chip.1 Combined with portable electronic equipment LOCs have diverse applications such as detecting explosives,2 monitoring nutrients in agricultural water, controlling quality of food production, controlling processes in the chemical industry,3 point-of-care analysis of bodily fluids,4 analyzing forensic evidence,5 and tracking pollution in environmental water or wastewater.6 The global demand for these types of devices is tremendous; a single hospital can do as many as hundreds of thousands of point-of-care tests on patients each year.7 A major incentive to improve further upon the current technology is to help global human health, especially in developing countries, by diagnostic testing using chemically and physically robust LOCs.8 The first devices were fabricated from polycrystalline silicon which is expensive and requires high processing costs. More recently, low-price polymers have been introduced as the material for fabricating these devices, thus decreasing their cost. Polymers are made with a variety of properties to fit the need of the device being designed, including physical, chemical, and thermal resistance, optical characteristics, and biocompatibilities. Some LOCs contain immobilized enzymes to perform a specific reaction on a substrate to produce a more easily measurable * Corresponding author: telephone, (614) 292-4249; fax, (614) 292-3769; e-mail, [email protected]. † The Ohio State University. ‡ University of Sydney.

product or to create an electrical charge.9 Immobilizing an enzyme can increase its robustness;10–14 however the high temperatures or organic solvents used during polymer processing are still lethal for these fragile molecules. Immobilization Techniques. A variety of immobilization and impregnation techniques have been reported (e.g., adsorption,14–16 cross-linking,17–21 covalent bonding,12,22–28 entrapment/encapsulation,29–32 and dense gas processing),33–38 each with intrinsic advantages and disadvantages for a variety of applications. Dense CO2 has been used previously to impregnate thermally labile compounds into polymer matrices; Kazarian et al. immobilized ibuprofen into poly(vinylpyrrolidone),34 Sproule et al. immobilized an immunoglobin into poly(methylmethacylate),37 Powell et al. immobilized paclitaxel into polylactic acid,39 and Kayrak-Talay et al. immobilized glucose oxidase into polypyrrole and polyurethane/polypyrrole composite foam.38 Immobilized enzymes/proteins within a LOC can be located in different places, such as on an electrode, on porous media placed in a microchannel, or on the walls of the channel itself. Our focus was to design a technique to immobilize enzyme in accordance with the latter of the three. Other strategies16,20,21,23–28 for immobilization in microchannels are briefly reviewed here. Holden et al.23 developed a method to photopattern patches of enzyme on the inside surfaces of an enclosed microchannel. They first adsorbed fibrinogen to the inner walls, next attached a photobleached free radical of biotin-4-flourescein, and finally attached streptavidin or avidin to the desired enzyme, subsequently to the modified biotin, and ultimately the enzyme was immobilized on the wall. This method was used to pattern three different enzymes in microchannels including glucose oxidase, horseradish peroxidase, and alkaline phosphastase. Markov et al.24 also used streptavidin to immobilize an enzyme on a wall of a microchannel, although the order of attachment was reversed from the method used by Holden et al. They irreversibly immobilized fluorescently labeled biotin and reversibly immobilized human immunoglobin G (IgG) Fc onto the streptavidin. Markov et al. demonstrated that after the activity of the

10.1021/bm701343m CCC: $40.75  2008 American Chemical Society Published on Web 02/23/2008

1028 Biomacromolecules, Vol. 9, No. 3, 2008

human IgG Fc had depleted, it could be easily removed and replaced, meaning that the same device could be reused for an extended period of time. Park et al.16 immobilized β-galactosidase (β-gal) exclusively on the glass wall of their microchannel by first adsorbing a nitrocellulose membrane to the glass and then adsorbing the enzyme to the membrane. Yang et al.26 deposited small unilamellar vesicles on the inner walls of their microchannel, which self-arranged to form supported bilayers. These bilayers contained immobilized dinitrophenyl conjugated lipids which were used for binding antidinitrophenyl IgG antibodies. Jang et al.25 were able to attach two different types of IgG to the inner wall of their microchannel. In this method, first an organosilane layer and then the IgG were attached to the wall. Hashimoto et al.20 were able to immobilize glucose oxidase on only one wall of their microchannel via a sol–gel (organosilane) transfer technique, whereas other researchers27 using an organosilane layer were not able to spatially control the immobilization. Xiong et al.27 also used an organosilane to immobilize β-gal on all walls of a borosilicate glass microchannel. The techniques used by Jang et al. and Hashimoto et al. were timeconsuming, at least 20 h for enzyme immobilization. Koh et al.21 used a hydrogel, poly(ethylene glycol), to immobilize alkaline phosphatase on the wall of a microchannel for a micro total analysis application. The enzyme immobilization was accomplished when the hydrogel was cross-linked by photoinitiation after its precursors were flowed into the channel. This technique used a solution containing the organic solvent, perfluorooctane, to enhance the adhesion of the hydrogel inside the channels. Wu et al.28 immobilized trypsin on poly(dimethylsiloxane) (PDMS) microchannels by the following method. The walls were first treated with an aqueous solution of NaIO4, benzyl alcohol, and acrylic acid to expose the PDMS carbonyl groups near the surface. Two different chemistries were then used to immobilize the enzyme: the first used carbodiimide chemistry to covalently attach a linker molecule to the exposed carbonyl groups and then attach the trypsin to the linker; the second method used an electrostatic interaction between the negatively charged carbonyl groups and the positively charged linker molecule. In both cases the immobilized trypsin was efficient for protein digestion. While the complete digestion of proteins took hours and minutes in macro batch reactors and other microreactors, respectively, it only took a matter of seconds for immobilized trypsin prepared by this technique. Acquiring fast results is a commonality for microdevices due to the low sample volume, high local concentration ratio of enzyme to reactive substrate, and short diffusion path lengths for the substrate and product. All of the reviewed techniques for immobilizing enzymes in a microchannel are multistep and require many different reagents, some involving organic solvents. Immobilization using dense CO2 is a one-step technique that requires nothing more than the enzyme solution and the CO2. In this study we demonstrate the feasibility of developing a one-step distinctive physical, rather than a chemical, process for immobilization of a biologically active compound into a polymeric microchannel. Microchip Fabrication. Bonding the caps on polymeric labon-a-chip microchannels has been accomplished in a variety of different ways, such as using organic solvents,40,41 hot embossing at temperatures above the glass transition temperature (Tg),1 low-temperature plasma activation,42 localized heating,43 and using dense CO2.44,45 All except for the two latter processing techniques are detrimental to biologically active molecules.

Ellis et al.

Another potential problem during bonding is that as the features on the chips are reduced in size, bonding without geometric deformation becomes a more prominent issue, but Yang et al. has eliminated this dilemma by bonding at moderate conditions with CO2, even down to the nanometer scale.45–47 All of the aforementioned research for immobilizing enzymes/ proteins in a microchannel used either PDMS, glass, or both. Researchers fabricated LOCs out of PDMS by first using photolithography to make a silicon wafer master template, with the desired microfeatures. Next, they mixed silicone base with a curing agent and poured it onto the silicon wafer master template, allowed it to cure, and then lifted it off. Although photolithography is expensive, the master template can be reused many times making the average cost per PDMS LOC device inexpensive. Multiple research groups21,23–26 have used glass to cap their PDMS channels following an oxidizing plasma treatment of the adjoining surfaces, thus creating a hermetic microfluidic device. However, plasma surface treatments destroy covalent bonds in order to expose specific molecules and make the surface more reactive. This type of bonding would denature immobilized biologically active molecules, so it is not a viable option in this fabrication technique. One of the objectives of this study was to develop a lowtemperature, solvent-free, non-clean-room technique for bonding an enclosing cap on an enzyme activated microchannels fabricated from an inexpensive polymer, while retaining enzymatic activity. We used dense CO2 at moderate temperatures and pressures for immobilization of the enzyme and capping of the microchannels. Polystyrene was used as the polymer matrix to immobilize β-gal. We have eliminated foam formation, a major issue in previous dense gas immobilization,37 and engineered a nonfoaming microstructure by meticulously governing the depressurization rate.

Experimental Section Chemicals. Polystyrene (PS) (CX5197; Mn ) 86000, Mw ) 211700, Mz ) 436100) from Total Petrochemicals (formerly Atofina) was used as received. β-D-Galactoside galactohydrolase derived from the microorganism Aspergillus oryzae (Enzeco Fungal Lactase Concentrate) donated by Enzyme Development Corporation in New York, NY, which will be referred to as β-gal in this paper, was used. Lactochem from Friesland Borculo Domo Ingredients (R-monohydrate pharmaceutical lactose, USP) was used as an enzymatic substrate. Acetate buffer was prepared from glacial acetic acid, sodium hydroxide, sodium acetate trihydrate all supplied by APS Ajax Finechem (Australia). Resorufin β-D-galactopyranoside was purchased from Marker Gene Technologies, Inc., in Eugene, OR. Carbon dioxide (food grade, 99.9% purity) was purchased from BOC. FluoReporter Oregon Green 488 Protein Labeling Kit (F-6153) was purchased from Molecular Probes. 2-Nitrophenylβ-D-galactopyranoside (ONPG) was purchased from Sigma-Aldrich. Compression Molding PS. A Geo. E. & Son hot press was used to prepare a sheet of PS, and the temperature used was well above the Tg of the polymer. Sixty-five grams of PS was placed in a stainless steel mold measuring 25 × 25 cm. Polyetherimide film was placed on the top and bottom of the PS and used as a mold release. The filled mold was placed on the bottom platen of the hot press and was heated to 200 °C to melt the PS. The PS was held at 200 °C and 0.7 MPa of compressive pressure for 1 h. Then the heating of the bottom platen was turned off and the PS was allowed to cool back to room temperature after ∼4 h. A 1 mm thick sheet of PS was produced. The sheet was cut into smaller pieces in order to make samples for lap shear bond strength testing and for fabrication of microfluidic chips. Hot Embossing Microchannels. A PDMS stamp was used to emboss microchannels into the PS substrate. Photolithography was used to pattern a negative tone resist (AZPN114) coated on a Si wafer in a

Enzyme Immobilization by Dense CO2 Technique

Biomacromolecules, Vol. 9, No. 3, 2008 1029

Figure 1. Schematic of hot embossing setup. The size of the sample is magnified. Figure 3. CO2 pressure vs depressurization time in the pressure cell after immobilization or bonding at 6.89 MPa. The volumetric flow rates given are the rate at which the volume of the pressurized system was set to increase in order to decrease pressure.

Figure 2. Schematic of enzyme immobilization setup. The size of the sample and the pressure vessel have been magnified.

clean room. The PDMS stamp with an inverse pattern was then prepared from cross-linking the silicone elastomer (Sylgard 184, Dow Corning) and lifting it off the Si wafer (aka the master template). Multiple stamps can be made from one master template, and numerous chips can be fabricated from one stamp all outside of the clean room, making this process mass producible. The hot embossing of microchannels was started by heating a hot plate up to 205 °C and placing a flat piece of polished stainless steel on it, then a piece of Teflon film, and the PS on top, see Figure 1. The PS became malleable within 5 min and the PDMS stamp was placed on it. Another piece of Teflon film and a piece of stainless steel were placed on top of the stamp, and finally a compressive force of 130 N was applied by placing a 13 kg weight on the assembly. There were also two stainless steel spacers placed between the Teflon so that the thickness of the PS would not be compromised during embossing. Once the weight was placed on top, the hot plate heater was shut off. After 10 min the system was quickly cooled down below the polymer Tg by pouring water on it. The weight and then the PDMS stamp were carefully removed, leaving microchannels in the PS. The dimensions of microchannels were 50 µm in depth, 100 µm in width, and a 75 µm space between each channel. This procedure was similar to that used by Yang et al.46 Immobilization of β-gal into PS Microchannels. The physical setup for enzyme immobilization, Figure 2, was assembled by placing a PS substrate complete with microchannels in a high-pressure vessel (Jerguson Site Gauge) and then placing a piece of porous glass (Ace Glass Inc. filter disk with maximum pore diameter of 4–8 µm) on top of it. One hundred microliters of 5 mg/mL β-gal in distilled water was dripped onto the porous glass. The β-gal solution percolated through the porous glass and came into contact with the PS substrate. The highpressure vessel was then sealed airtight and submerged in a 40 °C water bath. A minimum of 15 min was allowed to ensure thermal equilibrium, and then an ISCO syringe pump (500D) was used to pressurize it with CO2 (4.48–6.89 MPa). The temperature and pressure were held constant for 2.5 h during which the immobilization of the β-gal into the PS took place. The system was then depressurized over nearly 3 h to avoid foaming. The depressurization profile is represented graphically in Figure 3. The ISCO pump was used to increase the system volume by 1 mL/min for the first hour, 5 mL/min the second hour, and then 25 mL/min until completely depressurized. During the third hour the pump reached its maximum volume twice, so while holding it at constant pressure by purging off excess CO2, the volume of the pump was decreased. Once the pump reached its maximum volume for a third time, less than 0.3 MPa of pressure remained, which was expelled by slowly opening the

purge valve. After depressurization the enzyme immobilized PS sample was sonicated (Unisonics ultrasonic cleaner (FXP8)) in deionized water at room temperature for 10 min to remove the free β-gal from the surface. The PS with immobilized enzyme was then sealed in a plastic bag and stored at 5 °C until future application and characterization. Capping Microchannels. In this study we use a benign subcritical CO2 at the low temperature of 40 °C to bond the cap on the microchannels to preserve both enzyme activity and microchannel features. Polystyrene substrates (1.0 × 1.0 × 0.1 cm) with microchannels were capped by another piece of PS measuring the same dimensions. Prior to bonding, a 0.6 mm diameter drill bit was used to bore holes in the PS cap for fluid injection/extraction ports. During the process the two separate PS substrates were sandwiched by Teflon film and then surrounded by polished stainless steel plates. Two 19 mm foldback clips (standard office binder clips) were used to apply the compressive force, which measured 46 ( 3.0 N over the entire bonded surface area (Honeywell load cell model 13). The samples were placed in a Thar Technologies Reactor (R100W) and heated up to 40 °C. An ISCO 500D syringe pump was used to pressurize the system with CO2 to a desired pressure of 4.48-6.89 MPa. The vessel was kept at these conditions for a period of time that varied between 1 and 2.5 h. The system was then depressurized slowly using the syringe pump using the profile shown in Figure 3 to avoid foaming. Characterization Techniques. Immobilized β-gal Confirmation by Fluorescence. The presence of immobilized β-gal after dense CO2 processing was confirmed by confocal fluorescent microscopy on uncapped PS microchannels. β-gal was labeled with Oregon Green 488 dye, via the manufacturer’s instructions (Molecular Probes), which binds to the primary amines of the protein and has an absorption/emission light spectrum at 496/524 nm, respectively. Next, the labeled enzyme was immobilized using the previously described dense CO2 processing technique and again an ultrasonic cleaner was used to remove free enzyme from the microchannel surfaces. An Olympus IX81 confocal microscope equipped with a Hamamatsu C9100 CCD camera, a Yokogawa csu22 spin disk confocal scanning unit, a Cobolt Calypso 488 nm laser, and an Olympus 20× objective assembled by VisiTech International was used to image the immobilized enzyme. ActiVity of β-gal by Glucose Production. The activity of the β-gal was tested quantitatively using lactose as the substrate to conduct the enzymatic reaction. This reaction was run in a test tube using aqueous solutions of the enzyme and substrate, no PS was present. During the reaction, β-gal stimulated the hydrolysis of the glycosidic bond of lactose11 into two monosaccharides, one of them being glucose. A Roche/Hitachi Gluco-quant Glucose/HK was used to measure glucose concentration. The optimum activity for the enzyme is between pH 4.5 and 5.0 at 40 °C. The reaction was conducted at 40 °C with 2.5 mL of 55.5 mg/mL lactose in a 0.1 M acetate buffer solution at pH 4.5 and 100 µL of the β-gal solution in deionized water. While the reaction was running, aliquots (100 µL) of the solution were removed at set

1030 Biomacromolecules, Vol. 9, No. 3, 2008

Ellis et al.

Figure 4. Schematic of lap shear bonding sample.

time intervals (every 2 min) and mixed with 0.3 M NaOH in equal volumes to quench the reaction. These solutions from the corresponding time intervals were analyzed for their glucose concentration. ActiVity of Immobilized β-gal by Fluorescent Microscopy. The activity of the β-gal after immobilization into the PS was assessed in the capped microchannels qualitatively using a fluorescent substrate, resorufin β-D-galactopyranoside. This molecule is composed of a galactose sugar unit with one of the oxygen atoms bonded to resorufin. The resorufin can be fluoresced via laser excitation after being separated from the sugar. β-gal cleaves the aforementioned bond, leaving the resorufin by itself and easily identifiable by fluorescent microscopy. If the β-gal is no longer active, the bond cleavage will not occur, thus leaving the original molecule intact and nonfluorescent. A dilute solution (5 × 10-6 M) of resorufin β-D-galactopyranoside in dimethyl sulfoxide (DMSO) was dripped onto a PS substrate with immobilized β-gal. The fluorescent molecule was excited by a laser at 543 nm and monitored at 580–650 nm by a Leica DMIRE2 inverted stand fluorescent confocal microscope, equipped with a Leica TCS SPII confocal, a multiphoton system, and a titanium sapphire laser manufactured by Coherent Inc. The objective used was a 20× NA 0.5 Plan FLUOTAR lens. The transmitted light images were illuminated using a 633 nm laser. This fluorescent enzymatic substrate solution was used to qualitatively verify the activity of the β-gal after the microchannels had been capped by injecting the solution into the channels. Lap Shear Bonding Samples. A study of the lap shear bond force was conducted to quantify how well the cap was bonded on to the microchannels. An Instron 5567 equipped with a 10 kN load cell and a cross head speed of 1 mm/min was used to extensionally pull three samples at each condition to test them for their lap shear bond strength. Pieces of PS for lap shear bonding samples, Figure 4, were cut from the compression molded PS sheet and microchannels were hot embossed into one of the pieces to be bonded. Each single piece measured 3.0 × 1.0 × 0.1 cm. Two samples, one with microchannels, were overlapped by 1.0 cm making the total bonded surface area 1.0 cm2. Once the two pieces of PS were bonded together, tabs were glued on. The tabs were needed to make sure the extensional force during breaking was applied at the bonded interface and in a perpendicular geometry. Samples were all bonded at 40 °C and with the same compressive force applied, but the CO2 pressure and bonding time were varied. Two 19 mm foldback clips (standard office binder clips) were used to apply the compressive force during bonding, just as the bonding during the capping of the microchannels. At each condition three samples were tested to provide sufficient information for statistical analysis of the data and provide reliable conclusions.

Results and Discussions Enzyme Immobilization and Activity. Dense CO2 was used to immobilize the enzyme, β-gal, on the walls of a PS channel. Carbon dioxide is inexpensive, readily available, environmentally benign, nontoxic, nonflammable, noncorrosive, a tunable plasticizer,36 and has good sterilization properties.48 Dense CO2 is efficient as a polymer processing agent due to its gas-like transport properties and liquid-like densities. As a small linear molecule it readily diffuses into polymer matrices causing swelling, increased void space, enhanced polymer chain mobility,47 and glass transition temperature depression.49 The mech-

Figure 5. β-gal immobilization at 40 °C, 2.5 h, and CO2 pressure of 0.1 MPa (top) and 6.89 MPa (bottom).

anism for enzyme immobilization in this process is thought to be analogous with that of the microvoid model by Von Schnitzler and Eggers.50 Their model can be summarized as follows, when a polymer is exposed to dense CO2, it swells, thus increasing the free volume between polymer chains. At this point the enzyme diffuses into the free volume, known as microvoids, of the polymer by a concentration gradient. The size of the microvoids is unknown but is thought to increase as the Tg is approached; the size of the folded β-gal is 17.5 × 13.5 × 9.0 nm.51 The CO2 pressure is then released, the polymer relaxes back to it original size, and the enzyme is physically trapped within the polymer matrix. A disadvantage to using this technique is that if the enzyme is immobilized too deeply into the polymer matrix, it will not have its active site available for reacting. We demonstrate that the enzyme is immobilized in the PS microchannels using a fluorescence imaging technique. A reconstructed stack of two-dimensional confocal images to form one three-dimensional image of the fluorescently labeled immobilized enzyme in the PS microchannels is presented in Figure 5. The control sample, with no immobilized enzyme (not shown), showed no fluorescence. The relative intensity of the imaged fluorophore increased 88% between the two experimental conditions, based on the average of three samples at each. Similar florescent intensities were measured on samples that had been processed at 4.48, 5.52, and 6.89 MPa. This significant intensity increase of the processed sample proves that dense CO2 dramatically increases the quantity of enzyme immobilized into the microchannels over low pressure (0.1 MPa) CO2. The conditions used in this work have been shown to successfully immobilize the enzyme without causing the sample to foam but they have not been optimized for enzyme loading or processing time. To confirm that the activity of the enzyme was preserved, glucose production in an aqueous solution containing lactose

Enzyme Immobilization by Dense CO2 Technique

Biomacromolecules, Vol. 9, No. 3, 2008 1031

Figure 6. β-gal activity before being in the presence of CO2 (9), after immobilization conditions (2), and after immobilization and bonding conditions (0).

and free (nonimmobilized) β-gal was monitored as outlined above. The results presented in Figure 6 underline a reduction, 40% of its original concentration, in the production of glucose by an aqueous solution of β-gal (5 mg/mL in deionized water) exposed to CO2 at 40 °C and 6.89 MPa for a total period of 5 h, plus decreased pressures for another nearly 6 h. The aqueous β-gal solution was treated to the same conditions as a sample undergoing first immobilization and then capping, including the nearly 3 h of depressurization for each step. This is a worse case scenario since the highest pressure and longest time were used for both steps. This reduction in β-gal activity may result from a decrease in pH of the aqueous solution. Carbon dioxide dissolves in aqueous solutions at high pressures and reduces the pH of the solution down to as low as 3 when the pressure increases to 5.0 MPa.52 β-gal has been shown to denature at a pH lower than 4.0 (Enzyme Development Corp. Tech Data Sheet). Another possible explanation for the decease in enzymatic activity is the formation of a carbamate by the reaction between CO2 and the β-gal terminal amine groups. The activity of the β-gal does decrease, but there is still a sufficient active amount to run an enzymatic reaction. In the future using a buffered solution instead of water could possibly help retain the activity of the β-gal. The same tests were run with N2 replacing CO2, and β-gal showed no decrease in activity from its original value. The aforementioned data confirmed that there is still active free enzyme left after exposing the β-gal to the conditions used for processing the microchip. We were able to corroborate that the immobilized enzyme was also still active. The data acquired from an activity assay using lactose precursor on the actual channel surface did show an increase in glucose over time but was not reliable due to the small surface area of the microchannels for immobilization and, subsequently, the low concentration of glucose produced, which was near the minimum limit of detection of the Roche/Hitachi Gluco-quant Glucose/ HK. In addition, the activity of the enzyme may have also decreased as a result of the interaction with the polymer matrix due to being immobilized too deeply and therefore physically inaccessible for the reaction. Alternatively, fluorescent microscopy with higher resolution and a lower limit of detection was used successfully with a reactive substrate concentration 7 orders of magnitude less than that of the lactose substrate. The results of the Leica fluorescent confocal microscopy analysis demonstrate that β-gal maintained activity after immobilization and capping of the microchannel. Both the control test (a and b)

Figure 7. (a) Optical image through hole in cap of chip without immobilized enzyme. (b) Fluorescent image of sample shows mild autofluorescence when resorufin β-D-galactopyranoside is added. (c) Optical image through cap of chip with immobilized enzyme. (d) Fluorescent image of sample shows intense fluorescence in wells filled with resorufin confirming β-gal is active.

and also the validation of active β-gal after CO2 processing (c and d) are depicted in Figure 7. In Figure 7a we present an optical image of a microfluidic chip with a cap bonded on and no immobilized enzyme. The bonding conditions for capping the channels were T ) 40 °C, PCO2 ) 6.89 MPa, and t ) 1.0 h. The channels could not be imaged through the cap due to this particular piece of PS being slightly opaque, so the image was taken through part of a fluid injection/extraction port. Resorufin β-D-galactopyranoside in DMSO (5 × 10-6 M) was injected through the port and incubated for 30 min at room temperature. Figure 7b shows limited autofluorescence after being excited at 543 nm. An optical image of a microfluidic chip with immobilized enzyme is shown in Figure 7c. The immobilization conditions were T ) 40 °C, PCO2 ) 6.89 MPa, and t ) 2.5 h and the bonding conditions were T ) 40 °C, PCO2 ) 6.89 MPa, and t ) 1.0 h. This image was taken through the cap because it was transparent and the microchannels can be seen horizontally. Resorufin β-D-galactopyranoside in DMSO (5 × 10-6 M) was injected through the ports in the cap and was incubated for 30 min at room temperature. The laser was set to the same conditions as the control sample and the fluid in the microchannels intensely fluoresced as can be observed in Figure 7d, corroborating that the β-gal is active after CO2 processing of enzyme immobilization and capping of the channels. Bonding a Cap on Microchannels. Different bonding conditions were studied via measuring the lap shear bond strength using an Instron. There were five different CO2 bonding conditions used including 4.48 MPa for 1.0 and 2.5 h, 5.52 MPa for 1.75 h, and 6.89 MPa for 1.0 and 2.5 h. The attempt to bond the cap on microchannels at CO2 pressures below 5.52 MPa and 40 °C was unsuccessful for practical use. The bond was broken either before or during loading them into the Instron clamps, so quantitative data could not be obtained. However, bonding did occur at pressures g5.52 MPa and the average bond strength was increased 3-fold when the CO2pressure was raised from 5.52 to 6.89 MPa, Figure 8. Moreover, samples prepared at 6.89 MPa were successfully bonded, and the average bond

1032 Biomacromolecules, Vol. 9, No. 3, 2008

Ellis et al.

Figure 8. Lap shear bond test results (40 °C).

strength was enhanced slightly when the pressurization period was increased from 1 to 2.5 h. The experimental conditions for the lap shear bond samples were chosen by using a 2 × 2 factorial design with a center point and with three replicates at each point. An analysis of variance test with R ) 0.05 was used to analyze the bond strength data for the effects of both CO2 pressure and time on the bond strength. The analysis of variance test showed that CO2 pressure (p ) 0.0073) had a significant effect and that time (p ) 0.3278) did not have a significant effect on bond strength. All of the samples broke at the bond interface. Since the samples bonded at 4.52 MPa broke before testing and were handled the same as the rest, it can be assumed that their bond strength was less than that of the other samples. The bond strength plotted in Figure 8 is the average, and the error bars are the confidence interval (3σ) of three samples tested at each condition. This high-pressure bonding was also attempted by using N2 in place of CO2. Even after 24 h of 6.89 MPa of N2 at 50 °C and the same compressive force as the previous tests, no bonding was observed, thus proving that it is not viable to use for this fabrication even though it has no negative effect on the enzymatic activity. It is critical in the fabrication of LOCs to retain the original microstructure dimensions and geometry during processing. The ideal case would be high bond strength with no deformation of the microstructures. For all samples fabricated in our study, the microchannel structure was transferred to the bonded cap to some extent, but none of the samples deformed enough to block the channels from fluid flow; hence all samples were still usable as microfluidic devices. Optical images, taken by an Olympus BX60 microscope equipped with an Olympus DP10 digital camera and a 20× lens, of the cross section of a bonded microchip sample fabricated from PS by our dense gas process at bonding conditions of T ) 40 °C, PCO2 ) 6.89 MPa, and t ) 2.5 h are shown in Figure 9; the cap is on top of the PS containing the microchannels. Both images are from the same cross section at different locations. The sample was first scored on both the top and bottom using a stainless steel surgical knife and then broken at the score line all at room temperature. The fracture left a rough edge, so it was milled until smooth. As shown, some places bonded with no deformation (top) and others bonded with substantial deformation (bottom), but all channels are still usable, thus proving the validity of this bonding technique for LOC devices.

Conclusions The new LOC fabrication technique developed in this study possesses distinct advantages over its predecessors. The primary advantage is the simplicity of the technique, as both enzyme immobilization and bonding of the cap on the microchannels were performed by using dense CO2 near biological temperature, without

Figure 9. Optical images of the microchannels hot embossed into PS after being capped at T ) 40 °C, PCO2 ) 6.89 MPa, and t ) 2.5 h with no deformation (top) and substantial deformation (bottom).

using multistep chemical reaction techniques. In addition, no volatile organic compounds were used in this environmentally friendly process. A recyclable thermoplastic, polystyrene, was used to fabricate the LOC microdevice instead of unrecyclable crosslinked PDMS. The novel process opens an avenue for engineering LOCs from other specialty or biomedical polymers with a reasonably low Tg that can be plasticized by CO2. We were successful in both immobilizing an enzyme in microchannels and bonding a cap on the channels using only environmentally benign dense CO2. β-Galactosidase was immobilized in polystyrene microchannels at 40 °C using CO2 at pressures from 4.48 to 6.89 MPa in a process that took 5 h to complete. We have shown by confocal fluorescent microscopy that the immobilization of the enzyme occurs to an appreciable extent only when dense CO2 is used and not when low pressure is applied. The next step in this fabrication technique was to cap the biologically activated microchannels with a second piece of PS by using CO2 at 40 °C and pressures of either 5.52 or 6.89 MPa in a process that took up to 5 h. Both immobilization and capping processes were gentle enough to keep the fragile β-gal active, as shown by identifying the fluorescent product, resorufin, of a specific enzymatic reaction in the sealed microchannels. The results of this study and other previous impregnation/immobilization studies of proteins demonstrate the potential of dense CO2 for immobilization of various robust biomolecules into polymeric matrices at moderate temperatures. The depressurization profiles designed in this study eliminated the issue of foam formation in polymers with large CO2 solubility. The process we have developed in this paper can be considered “green” due to the use of only nontoxic, noncorrosive, environmentally benign chemicals, and it can be applied for the non-clean-room fabrication of lab-on-a-chip devices.

Enzyme Immobilization by Dense CO2 Technique

Acknowledgment. This project was supported by the National Science Foundation (EEC-0425626). We thank Dr. Yong Yang for his insightful suggestions and guidance. We also thank Renee Whan of the Electron Microscope Unit at the University of Sydney for her help with fluorescent microscopy and Orin Hemminger for his help also with fluorescent microscopy at the NanoTech West Laboratories at OSU. A final thank you to Aldric S. Tumilar for his bonding experiments using N2 as the bonding agent.

References and Notes (1) Lee, G. B.; Chen, S. H.; Huang, G. R.; Sung, W. C.; Lin, Y. H. Microfabricated plastic chips by hot embossing methods and their applications for DNA separation and detection. Sens. Actuators, B 2001, B75 (1–2), 142–148. (2) Wang, J.; Tian, B.; Sahlin, E. Micromachined electrophoresis chips with thick-film electrochemical detectors. Anal. Chem. 1999, 71 (23), 5436–5440. (3) Gardeniers, H.; Van Den Berg, A. Micro- and nanofluidic devices for environmental and biomedical applications. Int. J. EnViron. Anal. Chem. 2004, 84 (11), 809–819. (4) Vrouwe, E. X.; Luttge, R.; Vermes, I.; van den Berg, A. Microchip capillary electrophoresis for point-of-care analysis of lithium. Clin. Chem. (Washington, DC, U.S.) 2007, 53 (1), 117–123. (5) Verpoorte, E. Microfluidic chips for clinical and forensic analysis. Electrophoresis 2002, 23 (5), 677–712. (6) Bowden, M.; Diamond, D. The determination of phosphorus in a microfluidic manifold demonstrating long-term reagent lifetime and chemical stability utilizing a colorimetric method. Sens. Actuators, B 2003, B90 (1–3), 170–174. (7) Park, J. Y.; Kricka, L. J. Prospects for nano- and microtechnologies in clinical point-of-care testing. Lab Chip 2007, 7 (5), 547–549. (8) Chin, C. D.; Linder, V.; Sia, S. K. Lab-on-a-chip devices for global health: Past studies and future opportunities. Lab Chip 2007, 7 (1), 41–57. (9) Kok, F. N.; Bozoglu, F.; Hasirci, V. Construction of an acetylcholinesterase-choline oxidase biosensor for aldicarb determination. Biosens. Bioelectron. 2002, 17 (6–7), 531–539. (10) Fortier, G.; Belanger, D. Characterization of the biochemical behavior of glucose oxidase entrapped in a polypyrrole film. Biotechnol. Bioeng. 1991, 37 (9), 854–858. (11) Quinn, C. P.; Pathak, C. P.; Heller, A.; Hubbell, J. A. Photo-crosslinked copolymers of 2-hydroxyethyl methacrylate, poly(ethylene glycol) tetra-acrylate and ethylene dimethacrylate for improving biocompatibility of biosensors. Biomaterials 1995, 16 (5), 389–96. (12) Rejikumar, S.; Devi, S. Immobilization of β-galactosidase onto polymeric supports. J. Appl. Polym. Sci. 1995, 55 (6), 871–878. (13) Sonawat, H. M.; Phadke, R. S.; Govil, G. Covalent immobilization of FAD and glucose oxidase on carbon electrodes. Biotechnol. Bioeng. 1984, 26 (9), 1066–1070. (14) Yavuz, H.; Bayramoglu, G.; Kacar, Y.; Denizli, A.; Yakup Arica, M. Congo Red attached monosize poly(HEMA-co-MMA) microspheres for use in reversible enzyme immobilization. Biochem. Eng. J. 2002, 10 (1), 1–8. (15) Ding, H.-M.; Shao, L.; Liu, R.-J.; Xiao, Q.-G.; Chen, J.-F. Silica nanotubes for lysozyme immobilization. J. Colloid Interface Sci. 2005, 290 (1), 102–106. (16) Park, S.-S.; Cho, S. I.; Kim, M.-S.; Kim, Y.-K.; Kim, B.-G. Integration of on-column immobilized enzyme reactor in microchip electrophoresis. Electrophoresis 2003, 24 (1-2), 200–206. (17) Delvaux, M.; Demoustier-Champagne, S. Immobilization of glucose oxidase within metallic nanotubes arrays for application to enzyme biosensors. Biosens. Bioelectron. 2003, 18 (7), 943–951. (18) Li, X.; Zhao, K.; Men, Y.; Tu, W. Immobilization and characteristics of β-galactosidase on polymer resin. Bull. Soc. Chim. Belg. 1996, 105 (5), 217–222. (19) Zhou, Q. Z. K.; Chen, X. D. Effects of temperature and pH on the catalytic activity of the immobilized β-galactosidase from KluyVeromyces lactis. Biochem. Eng. J. 2001, 9 (1), 33–40. (20) Hashimoto, M.; Upadhyay, S.; Suzuki, H. Precise enzyme immobilization at the bottom of a micro flow channel and its application to a sensing system. Chem. Sens. 2004, 20 (Suppl. A), 85–87. (21) Koh, W.-G.; Pishko, M. Immobilization of multi-enzyme microreactors inside microfluidic devices. Sens. Actuators, B 2005, B106 (1), 335– 342.

Biomacromolecules, Vol. 9, No. 3, 2008 1033 (22) Bautista, F. M.; Campelo, J. M.; Garcia, A.; Jurado, A.; Luna, D.; Marinas, J. M.; Romero, A. A. Properties of a glucose oxidase covalently immobilized on amorphous AlPO4 support. J. Mol. Catal. B: Enzym. 2001, 11 (4-6), 567–577. (23) Holden, M. A.; Jung, S.-Y.; Cremer, P. S. Patterning Enzymes Inside Microfluidic Channels via Photoattachment Chemistry. Anal. Chem. 2004, 76 (7), 1838–1843. (24) Markov, D. A.; Swinney, K.; Bornhop, D. J. Label-Free Molecular Interaction Determinations with Nanoscale Interferometry. J. Am. Chem. Soc. 2004, 126 (50), 16659–16664. (25) Jang, Y.; Oh, S. Y.; Park, J.-K. In situ electrochemical enzyme immunoassay on a microchip with surface-functionalized poly(dimethylsiloxane) channel. Enzyme Microb. Technol. 2006, 39 (5), 1122– 1127. (26) Yang, T.; Jung, S.-y.; Mao, H.; Cremer, P. S. Fabrication of phospholipid bilayer-coated microchannels for on-chip immunoassays. Anal. Chem. 2001, 73 (2), 165–169. (27) Xiong, L.; Regnier, F. E. Channel-specific coatings on microfabricated chips. J. Chromatogr., A 2001, 924 (1–2), 165–176. (28) Wu, H.; Zhai, J.; Tian, Y.; Lu, H.; Wang, X.; Jia, W.; Liu, B.; Yang, P.; Xu, Y.; Wang, H. Microfluidic enzymatic-reactors for peptide mapping: strategy, characterization, and performance. Lab Chip 2004, 4 (6), 588–597. (29) Lin, T.-Y.; Wu, C.-H.; Brennan, J. D. Entrapment of horseradish peroxidase in sugar-modified silica monoliths: Toward the development of a biocatalytic sensor. Biosens. Bioelectron. 2007, 22 (9–10), 1861– 1867. (30) Podual, K.; Doyle, F. J.; Peppas, N. A. Glucose-sensitivity of glucose oxidase-containing cationic copolymer hydrogels having poly(ethylene glycol) grafts. J. Controlled Release 2000, 67 (1), 9–17. (31) Retama, J. R.; Lopez-Ruiz, B.; Lopez-Cabarcos, E. Microstructural modifications induced by the entrapped glucose oxidase in crosslinked polyacrylamide microgels used as glucose sensors. Biomaterials 2003, 24 (17), 2965–2973. (32) Sahney, R.; Anand, S.; Puri, B. K.; Srivastava, A. K. A comparative study of immobilization techniques for urease on glass-pH-electrode and its application in urea detection in blood serum. Anal. Chim. Acta 2006, 578 (2), 156–161. (33) Kazarian, S. G.; Brantley, N. H.; West, B. L.; Vincent, M. F.; Eckert, C. A. In situ spectroscopy of polymers subjected to supercritical CO2: plasticization and dye impregnation. Appl. Spectrosc. 1997, 51 (4), 491–494. (34) Kazarian, S. G.; Martirosyan, G. G. Spectroscopy of polymer/drug formulations processed with supercritical fluids: in situ ATR-IR and Raman study of impregnation of ibuprofen into PVP. Int. J. Pharm. 2002, 232 (1–2), 81–90. (35) Muth, O.; Hirth, T.; Vogel, H. Polymer modification by supercritical impregnation. J. Supercrit. Fluids 2000, 17 (1), 65–72. (36) Nikitin, L. N.; Gallyamov, M. O.; Vinokur, R. A.; Nikolaec, A. Y.; Said-Galiyev, E. E.; Khokhlov, A. R.; Jespersen, H. T.; Schaumburg, K. Swelling and impregnation of polystyrene using supercritical carbon dioxide. J. Supercrit. Fluids 2003, 26 (3), 263–273. (37) Sproule, T. L.; Lee, J. A.; Li, H.; Lannutti, J. J.; Tomasko, D. L. Bioactive polymer surfaces via supercritical fluids. J. Supercrit. Fluids 2004, 28 (2–3), 241–248. (38) Kayrak-Talay, D.; Akman, U.; Hortacsu, O. Glucose oxidase immobilization on conducting polymers in supercritical CO2 environment: An exploratory study. J. Supercrit. Fluids 2007, 42 (2), 273–281. (39) Powell, H. M.; Ayodeji, O.; Summerfield, T. L.; Powell, D. M.; Kniss, D. A.; Tomasko, D. L.; Lannutti, J. J. Chemotherapeutic implants via subcritical CO2 modification. Biomaterials 2007, 28 (36), 5562–5569. (40) Ryu, W.; Min, S. W.; Hammerick, K. E.; Vyakarnam, M.; Greco, R. S.; Prinz, F. B.; Fasching, R. J. The construction of threedimensional micro-fluidic scaffolds of biodegradable polymers by solvent vapor based bonding of micro-molded layers. Biomaterials 2006, 28 (6), 1174–1184. (41) Brown, L.; Koerner, T.; Horton, J. H.; Oleschuk, R. D. Fabrication and characterization of poly(methyl methacrylate) micro fluidic devices bonded using surface modifications and solvents. Lab Chip 2006, 6 (1), 66–73. (42) Kettner, P.; Pelzer, R. L.; Glinsner, T.; Farrens, S. New results on plasma activated bonding of imprinted polymer features for Bio MEMS applications. J. Phys.: Conf. Ser. 2006, 34, 65–71. (43) Su, Y.-C.; Lin, L. Localized bonding processes for assembly and packaging of polymeric MEMS. IEEE Trans. AdV. Packag. 2005, 28 (4), 635–642. (44) Lu, C.; Xie, Y.; Yang, Y.; Cheng, M. M. C.; Koh, C.-G.; Bai, Y.; Lee, L. J.; Juang, Y.-J. New valve and bonding designs for microfluidic biochips containing proteins. Anal. Chem. 2007, 79 (3), 994–1001.

1034 Biomacromolecules, Vol. 9, No. 3, 2008 (45) Yang, Y.; Lee, L. J.; Lu, W. Subcritical carbon dioxide assisted polymer nanofabrication at low temperatures. J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.-Process., Meas., Phenom. 2005, 23 (6), 3202–3204. (46) Yang, Y.; Basu, S.; Tomasko, D. L.; Lee, L. J.; Yang, S.-T. Fabrication of well-defined PLGA scaffolds using novel micro-embossing and carbon dioxide bonding. Biomaterials 2005, 26 (15), 2585–2594. (47) Yang, Y.; Zeng, C.; Lee, L. J. Three-dimensional assembly of polymer microstructures at low temperatures. AdV. Mater. (Weinheim, Ger.) 2004, 16 (6), 560–564. (48) Dillow, A. K.; Dehghani, F.; Hrkach, J. S.; Foster, N. R.; Langer, R. Bacterial inactivation by using near- and supercritical carbon dioxide. Proc. Natl. Acad. Sci. U.S.A. 1999, 96 (18), 10344–10348.

Ellis et al. (49) Keddie, J. L.; Jones, R. A. L.; Cory, R. A. Size-dependent depression of the glass transition temperature in polymer films. Europhys. Lett. 1994, 27 (1), 59–64. (50) Von Schnitzler, J.; Eggers, R. Mass transfer in polymers in a supercritical CO2 atm. J. Supercrit. Fluids 1999, 16 (1), 81–92. (51) Jacobson, R. H.; Zhang, X. J.; DuBose, R. F.; Matthews, B. W. Threedimensional structure of beta -galactosidase from E. coli. Nature (London, U.K.) 1994, 369 (6483), 761–766. (52) Hofland, G. W.; de Rijke, A.; Thiering, R.; van der Wielen, L. A. M.; Witkamp, G. J. Isoelectric precipitation of soybean protein using carbon dioxide as a volatile acid. J. Chromatogr., B: Biomed. Sci. Appl. 2000, 743 (1 + 2), 357–368.

BM701343M