Using Bioinspired Thermally Triggered Liposomes for High-Efficiency

Anal. Chem. 2003, 75, 6906-6911

Using Bioinspired Thermally Triggered Liposomes for High-Efficiency Mixing and Reagent Delivery in Microfluidic Devices Wyatt N. Vreeland and Laurie E. Locascio*

Analytical Chemistry Division, National Institute of Standards and Technology, 100 Bureau Drive, MS 8394, Gaithersburg, Maryland 20899-8394

High-efficiency mixing is of fundamental importance for the successful development and application of lab-on-achip devices. In this report, we present the use of bioinspired thermally triggered liposomes for the controlled delivery and subsequent rapid mixing of reagents in a microfluidic device. In this technique, reagents are encapsulated inside the aqueous interior of liposomes that are dispersed evenly throughout a microfluidic system. Mixing of the encapsulated reagent and reaction do not occur until the reagent is released by a thermal trigger. This approach takes advantage of the dramatically increased lipid membrane permeability of liposomes near the gel-to-liquid phase transition temperature (Tm) to deliver reagents at a precise location in the microfluidic device through the modulation of temperature. Implementation of this technique requires the encapsulation of the desired reagent in a liposome whose formulation has an appropriate Tm, as well as accurate spatial control of the temperature in the microfluidic device. As the liposomes are uniformly dispersed through the microfluidic channel, mixing occurs quite rapidly upon the release of the reagent. We demonstrate this technique by using several formulations of thermally triggered liposomes to release the hydrophilic fluorescent dyes at controlled locations in a polycarbonate microfluidic device. Additionally, we demonstrate a DNA labeling reaction using liposomes in a capillary-based microfluidic device. Under the conditions studied here, mixing and reaction are complete in ∼200 µm of channel length. We believe this approach holds great promise for the performance of rapid high-throughput assays and in particular for biological analytes whose native environment is mimicked by the liposome. The implementation of microfluidic devices in chemical and biological assays promises to increase both the speed and ease with which scientists can obtain information. These advantages stem from the miniaturized geometries that comprise these devices, allowing for less reagent consumption and more efficient analytical separation and analysis. The ultimate goal of this technology is to fully automate a particular analysis so that manual * To whom orrespondence should be addressed: (phone) (301) 975-3130; (fax) (301) 977-0587; (e-mail) [email protected].

6906 Analytical Chemistry, Vol. 75, No. 24, December 15, 2003

interaction with the sample to be analyzed is minimized or even eliminated. Similar to traditional benchtop chemical analysis techniques, the successful employment of lab-on-a-chip devices requires efficient and effective mixing and metering of reagents. Methods for sample and reagent metering have been well established, and ultimately, it is mixing that limits device performance in many cases. Mixing is perhaps the most essential step in any chemical or biochemical reaction, as it is the basis by which two analytes interact, the subsequent products or results of which are of interest to the biologist or chemist. In a microchannel, mixing is most often accomplished by joining two or more channels and allowing for diffusion to mix the two species. However, microfluidic devices, owing to their micrometer-scale geometries, operate in the low Reynolds number regimes (Re ≈ 0.1-100) where fluid flow is purely laminar. Thus, diffusive mixing is a rather slow process and can require a substantial amount of time and space to acheive complete mixing. For devices that operate at high flow rates or include soluble species with low diffusion coefficients such as proteins and cells, this approach may not be suitable. Most other micromixing strategies reported to date require alteration or augmentation of the microfluidic channel itself. A brief list of these strategies will be enumerated. As the channels in microfluidic devices are too small to incorporate traditional mixing devices, several researchers have explored modification of the microfluidic channel walls as a means to achieve passive micromixing. Passive micromixers include flow-splitting or multilamina devices that split the fluid streams into several smaller streams, thereby increasing the interfacial area for mixing and reducing the diffusion distance.1 Earlier work in our laboraotry2,3 and by others4,5 involved fabrication of surface features in the microfluidic channel to facilitate transverse fluid movement to enhance fluid mixing. A 3-D serpentine channel was decribed by Liu et al. to facilitate mixing through chaotic advection.6 Other researchers have investigated improved mixing in microchannels (1) He, B.; Burke, B. J.; Zhang, X.; Zhang, R.; Regnier, F. E. Anal. Chem. 2001, 73, 1942-1947. (2) Johnson, T. J.; Ross, D.; Locascio, L. E. Anal. Chem. 2002, 74, 45-51. (3) Johnson, T. J.; Locascio, L. E. Lab Chip 2002, 2, 135-140. (4) Stroock, A. D.; Dertinger, S. K. W.; Ajdari, A.; Mezic, I.; Stone, H. A.; Whitesides, G. M. Science 2002, 295, 647-651. (5) Stroock, A. D.; Dertinger, S. K.; Whitesides, G. M.; Ajdari, A. Anal. Chem. 2002, 74, 5306-5312. (6) Liu, R. H.; Stremler, M. A.; Sharp, K. V.; Olsen, M. G.; Santiago, J. G.; Adrian, R. J.; Aref, H.; Beebe, D. J. J. Microelectromech. Syst. 2000, 9, 190-197. 10.1021/ac034850j Not subject to U.S. Copyright. Publ. 2003 Am. Chem. Soc.

Published on Web 11/11/2003

with multipatterned surface charges,7 porous polymeric monoliths,8 catalytic microbeads,9 flapper valves,10 and 3-D microvascular networks.11 Another general approach to micromixing involves the application of an external force to actively mix the fluid streams. Examples of active micromixers include ultrasonically actuated mixers,12,13 acoustic bubble mixers,14 optically actuated rotators,15 electrowetting-based droplet mixers,16 magnetic microstirrers,17 and electrokinetic instability micromixers.18 Biology has an elegant solution to the challenge of mixing and reagent delivery that is nearly ubiquitously employed in the microfluidic environment of the biological cellsthat is, to place the reagents in tiny packages called vesicles. These vesicles are then directed to a precise desired location where at a controlled time they become permeable and release their contents for mixing and reaction. Vesicle permeabilization is the motif that nature uses for a variety of control, signal transduction, response signaling, and material transport in nerve, muscle, endothelial, and epithelial cells and that we have adapted for use in microfluidic systems. In this report, we present the use of bioinspired vesicles called liposomes for the controlled delivery and rapid mixing of hydrophilic reagents in a microfluidic environment. The liposome is a spherical structure composed of a phospholipid bilayer membrane (similar to a cell membrane) that encapsulates a volume of intravesicular aqueous solution. Liposomes are bathed in an external aqueous solution and range in size from tens of nanometers to micrometers in diameter. The ability to encapsulate a species of interest inside liposomes isolates that species and renders it inert to chemicals residing outside of the membrane. Environmentally responsive liposomes can be engineered such that they are permeabilized by a range of external triggers including temperature, pH, ionic strength, and light. In this study, we have developed a number of thermally triggered liposome formulations for controlled release and mixing in a microfluidic system. Thermal permeabilization takes advantage of the dramatically increased bilayer permeability near the lipid chain melting transition temperature (Tm). At temperatures below the Tm, the bilayer exists in the gel (Lβ) phase and has a low permeability to ionic species; above the Tm, the bilayer is in the fluid (LR) phase and is also of relatively low permeability. However, at temperatures near the Tm, the bilayer is characterized by markedly increased (7) Stroock, A. D.; Weck, M.; Chiu, D. T.; Huck, W. T. S.; Kenis, P. J. A.; Ismagilov, R. F.; Whitesides, G. M. Phys. Rev. Lett. 2001, 86, 6050. (8) Rohr, T.; Yu, C.; Davey, M. H.; Svec, F.; Frechet, J. M. J. Electrophoresis 2001, 22, 3959-3967. (9) Seong, G. H.; Crooks, R. M. J. Am. Chem. Soc. 2002, 124, 13360-13361. (10) Voldman, J.; Gray, M. L.; Schmidt, M. A. J. Microelectromech. Syst. 2000, 9, 295-302. (11) Therriault, D.; White, S. R.; Lewis, J. A. Nat. Mater. 2003, 2, 265-271. (12) Yang, Z.; Matsumoto, S.; Goto, H.; Matsumoto, M.; Maeda, R. Sens. Actuators, A 2001, 93, 266-272. (13) Rife, J. C.; Bell, M. I.; Horwitz, J. S.; Kabler, M. N.; Auyeung, R. C. Y.; Kim, W. J. Sens. Actuators, A 2000, 86, 135-140. (14) Liu, R. H.; Yang, J. N.; Pindera, M. Z.; Athavale, M.; Grodzinski, P. Lab Chip 2002, 2, 151-157. (15) Ukita, H.; Kanehira, M. IEEE J. Sel. Top. Quantum Electron. 2002, 8, 111117. (16) Paik, P.; Pamula, V. K.; Pollack, M. G.; Fair, R. B. Lab Chip 2003, 3, 2833. (17) Lu, L. H.; Ryu, K. S.; Liu, C. J. Microelectromech. Syst. 2002, 11, 462-469. (18) Oddy, M. H.; Santiago, J. G.; Mikkelsen, J. C. Anal. Chem. 2001, 73, 58225832.

permeability. This is believed to be due to the highly permeable interfacial regions between coexisting gel and fluid bilayer phases. Further, it is well known that the Tm is related to the number of carbons in the acyl tail of the phospholipid. Thus, by altering the phospholipid composition of the liposome membrane, the Tm can be tailored to a desired temperature. In this work, liposomes loaded with species of interest are introduced into the microfluidic environment. The liposomes are then maneuvered to a desired location, and the intravesicular contents are released into the local environment through modulation of temperature in the microfluidic channel. Since the liposomes are uniformly distributed throughout the channel, upon permeabilization, mixing is inherently quite rapid as the reagent has only to diffuse the average distance between liposomes rather than across the entire microfluidic channel. EXPERIMENTAL SECTION19 Reagents. 1,2-Dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), and cholesterol were from Avanti Polar Lipids (Alabaster, AL). 5- and 6-carboxyfluorescein was obtained from Molecular Probes (Eugene, OR), Sulforhodamine B (SRB), tris(hydroxymethyl)aminomethane (Tris), ethidium bromide, calf thymus DNA, boric acid, and ethylenediaminetetraacetic acid were from Sigma (St. Louis, MO). Polycarbonate sheets were from McMaster-Carr (New Brunswick, NJ). Fused-silica capillaries were obtained from Polymicro Technologies (Phoenix, AZ). Liposome Preparation. Liposomes were prepared by the hydration method, which is known to result in multilamellar liposomes. First, 4 µmol of the desired lipid mixture in 30 µL of chloroform was dried on the bottom of a small glass test tube in a vacuum desiccator. The dried lipid film was then hydrated with 50 µL of the analyte-buffer solution (to be encapsulated) while thermostated in a water bath at 75 °C for 30 min. To improve the encapsulation efficiency, after lipid hydration, the liposome suspension was flash frozen on liquid nitrogen and thawed in 75 °C water bath. This procedure was repeated for a total of five freeze-thaw cycles. The liposomes were then allowed to anneal at room temperature for ∼1 h. To remove unencapsulated analyte, the liposome suspension was passed over a Sephadex G-50 gel filtration column. After gel filtration, the liposome preparation was employed directly in the microfluidic mixing experiment. Liposome sizes were determined by light scattering using a Coulter N4 MD19 submicrometer particle analyzer collecting scattered light at 90°. Liposome Preparation for Dequenching Assays. Self-quenched fluorescent dye solutions of carboxyfluorescein and sulforhodamine B were prepared in 0.5 M Tris buffer for encapsulation in liposomes. A 200 mM concentration of carboxyfluorescein was prepared by dissolving carboxyfluorescein at 400 mM in 1 M Tris and then diluting with an equal volume of water (final pH ∼7.9). Sulforhodamine B was dissolved directly in 0.5 M Tris at 100 mM, and the solution was then titrated with hydrochloric acid until the pH matched the pH of the carboxyfluorescein solution. For (19) Certain commercial equipment, instruments, or materials are identified in this report to specify adequately the experimental procedure. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose.

Analytical Chemistry, Vol. 75, No. 24, December 15, 2003

6907

buffer exchange on the gel filtration column, 0.5 M Tris was titrated with 12 M aqueous hydrochloric acid to a pH of ∼7.9. All buffers were prepared with 0.2-µm filtered 18 MΩ‚cm water. Liposome Preparation for DNA Labeling Assay. Ethidium bromide was encapsulated inside liposomes by hydrating the dried lipid in a solution of 2.5 M ethidium bromide in 1× TBE (89 mM Tris, 89 mM boric acid, 2 mM ethylenediaminetetraacetic acid). Unencapsulated ethidium bromide was separated from liposome suspension by gel filtration using 1× TBE for buffer exchange. Calf thymus DNA was dissolved at 0.01 unit/mL in 1× TBE (note that calf thymus DNA was not encapsulated inside the liposomes). Microfluidic Device Fabrication. Microfluidic channels were fabricated in 100-µm-thick polycarbonate sheets using methods similar to those previously published.20 Briefly, a polycarbonate sheet was centered on top of a silicon template and placed under 9.2 × 106 Pa (1300 psi) of pressure at 165 °C for ∼5 min. The temperature was then lowered to 125 °C, and the pressure was released. The imprinted polycarbonate sheet was then carefully removed from the silicon template and placed on another piece of polycarbonate in which holes were cut to access the microfluidic channels. The two sheets were then clamped together and bonded at 165 °C for 15 min. The resulting microchannel was of a trapezoidal cross section and 75 µm wide at the base, 25 µm wide at the top, and 30 µm deep. To mobilize fluid through the channel, the access port was mated with a commercially available microfluidic fitting (Nanoport, Upchurch Scientific, Oak Harbor, WA).19 A fused-silica capillary was used to connect the fitting to a syringe. A syringe pump was then used to control the fluid flow in the microfluidic device. DNA labeling experiments were conducted in fused-silica capillaries rather than polycarbonate microchannels, due to the low transmittance of polycarbonate to UV light necessary to excite efficiently DNA-ethidium bromide complexes. The polyimide coating was removed from ∼.5 mm of the 100-µm-inner diameter, 360-µm-outer diameter fused-silica capillary to facilitate optical interrogation of the channel. Temperature Gradient. The microfluidic chip or fused-silica capillary was mounted on an apparatus that contained two copper blocks, one with an electrical resistance element (for heating) and the other with a recirculating water bath (for cooling), separated by a distance of 2 mm. The water bath was thermostated to 20 °C, and the electric resistance heater was set to a higher temperature (that will be detailed later) to create a lateral thermal gradient along the length of the channel. The temperature as a function of lateral position along the channel was determined using the method of Ross et al.21 Data Acquisition. The microfluidic channel was imaged with a fluorescence microscope fitted with a long working distance 10× objective and a three-chip color CCD camera (DAGE-MTI, Michigan City, IN) 19 in which the autogain was disabled. Images were acquired with a CG-7 frame grabber card and analyzed with Scion Image software (both from Scion Corp., Frederick, MD).19 RESULTS AND DISCUSSION In this study, we use liposomes to encapsulate a chemical or biological species of interest and sequester it from the microfluidic (20) Ross, D.; Locascio, L. E. Anal. Chem. 2002, 74, 2556-2564. (21) Ross, D.; Gaitan, M.; Locascio, L. E. Anal. Chem. 2001, 73, 4117-4123.

6908 Analytical Chemistry, Vol. 75, No. 24, December 15, 2003

Figure 1. (a) Fluorescence micrograph showing a solution of a mole fraction of 95% DPPC and 5% cholesterol liposomes in 0.5 M Tris buffer encapsulating self-quenched 200 mM carboxyfluorescein flowing through a polycarbonate microfluidic channel under an applied temperature gradient of 20-64 °C over a 2-mm distance at a flow rate of 100 µL/h. The image shows only 1.7 mm of the microfluidic channel. (b) The same image as (a), but the fluorescence intensity is shown in false color, (c) Plot of normalized fluorescence intensity and temperature of the image in (a) as a function of position in the microfluidic channel.

environment. Then, using temperature as a trigger, the internal contents are released into the microfluidic environment where they can interact rapidly with other species residing in the extravesicular space. Dequenching Assays. The first series of experiments were performed to demonstrate the ability to encapsulate a model analyte inside a population of liposomes and then trigger its release by programming the temperature in the microchannel to match, at a specific point, the Tm of the liposomes. It is important that the liposome formulation be compatible with the analyte to be encapsulated to minimize spontaneous leakage from the liposomes. Programmed Carboxyfluorescein Release. For these experiments, 200 mM carboxyfluorescein (CF) was encapsulated in liposomes prepared from 5 mol % cholesterol and 95 mol % DPPC. At this CF concentration, the encapsulated dye was selfquenched and the liposomes exhibited a very low background fluorescence. CF encapsulated with this liposome formulation showed no detectable leakage over a several-hour period; however, after storage at room temperature for 48 h, there was substantial spontaneous CF release from the liposomes. This preparation of liposomes was measured by light scattering to have an average diameter of 2.67 µm with a relative standard deviation of 22.5%. The ends of the microchannel were thermostated to 20 and 64 °C, and a linear temperature gradient was generated between the two termini. The volumetric flow rate was 100 µL/h of liposome suspension, which corresponds to an average fluid velocity of 1.8 cm/s. Figure 1a presents the controlled release of encapsulated self-quenched carboxyfluorescein into the microfluidic environment. The fluorescence signal associated with the flowing liposomes was initially very low, but as the liposomes moved down the channel (from left to right in the figure), they reached the Tm at which point they were permeable to the dye. The CF quickly diffused out of the liposomes and into the extravesicular microfluidic environment, which lowered the local dye concentration

Figure 3. Normalized fluorescence intensity as a function of temperature in of the micrographs in Figure 2c-i.

Figure 2. (a) False color micrographs of the fluorescence intensity of a solution of a mole fraction of 97% DPPC and 3% cholesterol liposomes encapsulating self-quenched 100 mM sulforhodamine B in 0.5 M Tris buffer flowing through a polycarbonate microfluidic channel under an applied temperature gradient of 20-40 °C over 2-mm distance at a flow rate of 100 µL/h. Channel was the same as in Figure 1. (b) Same as (a) but temperature gradient was 20-45, (c) 20-50, (d) 20-55, (e) 20-60, (f) 20-65, (g) 20-70, (h) 20-75, and (i) 20-80 °C.

(to an unquenched concentration), leading to a dramatic increase in the fluorescence intensity. Figure 1b presents the same image with false color to show more clearly the increase in fluorescence intensity, and Figure 1c shows a plot of normalized average fluorescence intensity and temperature as a function of lateral position in the microfluidic channel. As can be seen, the liposome suspension in the microchannel had a relatively constant and low fluorescence intensity at temperatures below ∼34 °C (corresponding to a distance ∼1 mm downstream from the low-temperature terminus) with an average pixel intensity of 37.6 and a relative standard deviation of 4.1%. As the liposomes moved downstream to higher temperatures, CF was released and the signal increased to a relatively high and constant value of 80.5 with a relative standard deviation of 1.8%. Further, this transition occurred over a distance of ∼200 µm, with its midpoint centered at ∼37 °C. Thus, the dye went from the encapsulated “unmixed” self-quenched state to a “mixed” highly fluorescent state in a distance of 200 µm, which corresponds to an average mixing time of 110 ms. This mixing time is much faster than the roughly 5 s it would require for mixing to occur through diffusion alone. Because the liposomes were uniformly dispersed through the microfluidic channel, mixing of the fluorescent dye upon release from the liposomes was inherently quite fast. This stemmed from the fact that the dye only had to diffuse the relatively small distance between neighboring liposomes to “mix” rather than diffuse across the entire diameter of the microfluidic channel. Controlling Reagent Release Spatially. Figure 2 shows the false color images demonstrating the release of self-quenched sulforhodamine B into the microfluidic environment at a several controlled, lateral positions along the microfluidic channel. For

these experiments, 100 mM sulforhodamine B was encapsulated in liposomes prepared from a mole fraction of 3% cholesterol and 97% DPPC. Sulforhodamine B, owing to its extreme hydrophilicity, showed no detectable spontaneous loss after storage at room temperature over a period of several weeks. This liposome preparation had an average diameter of 3.35 µm with a relative standard deviation of 26.8%. Measurement of the concentration of SRB after triggered release and the magnitude of self-quenching while still in the liposomes showed an internal concentration of ∼10 mM, corresponding to an encapsulation efficiency of 10%. The low-temperature terminus of the microfluidic channel was thermostated to 20 °C, and the high-temperature terminus was thermostated to temperatures ranging from 40 to 80 °C in 5 °C increments. In all cases, the volumetric flow rate of the liposome suspension was 100 µL/h, again corresponding to a linear velocity of 1.8 cm/s. The Tm of this liposome population was measured using differential scanning calorimetry (DSC) to be 36.1 °C, with the transistion beginnig at 34.8 °C and complete at 38.1 °C (data not shown). In the first panel (Figure 2a), the temperature along the microchannel did not pass through the Tm of this liposome formulation in this segment of the microchannel and the channel had uniform low fluorescence intensity. The low fluorescence intensity was because the liposomes retained the quenched sulforhodamine B in the intravesicular space without apparent sign of leakage. In the subsequent panels (Figure 2b-i), the liposomes passed through their Tm at some point in the microchannel dictated by the temperature profile (gradient) imposed in that experiment. As might be expected, as the magnitude of the temperature gradient increased, the lateral position in the microchannel that matched the Tm of the liposomes shifted upstream (to the left in these images). The fluorescence intensity dramatically increased where the temperature in the microchannel coincided with Tm of the liposomes. In Figure 2b, the liposome passed out of the field of view before the release of sulforhodamine B was complete, leading to a lower maximum fluorescence in the image. Figure 3 shows the average normalized fluorescence profile as a function of temperature in the channel (not position) under each of the temperature gradients investigated here. As can be seen in Figure 3, the fluorescence intensity increased at approximately the same temperature in the microfluidic channel under each different applied temperature gradient. The midpoint of the transition occurred at an average temperature of 32.5 °C with an relative standard deviation of 9.1% (compared to 36.1 °C by DSC), which is within the reported accuracy of the temperature Analytical Chemistry, Vol. 75, No. 24, December 15, 2003

6909

Figure 4. (a) Fluorescence micrograph showing a solution of 90 mol % DPPC and 10 mol % cholesterol liposomes encapsulating selfquenched 200 mM carboxyfluorescein and 75 mol % DPPC, 15 mol % DSPC, and 10 mol % cholesterol in 0.5 M Tris buffer flowing through a polycarbonate microfluidic channel under an applied temperature gradient of 20-44 °C over a 2-mm distance at a flow rate of 100 µL/h. (b) Normalized fluorescence intensity of the image in (a) as a function of position in the microfluidic channel. The solid line (s) is the green channel, λ ) 500-540 nm, and the dashed line (- - -) is the red channel, λ ) 570-625 nm.

measurement method used here (3.5 °C).21 Additionally, when the sulforhodamine B flowed to the high-temperature terminus of the channel, the fluorescence intensity decreased due to the decreased quantum yield of sulforhodamine B at elevated temperatures.21 As the temperature of the high-temperature terminus of the microchannel increased, this effect becomes more pronounced. Engineering Liposomes for Controlled Release of Different Species. In Figure 4a, two formulations of liposomes with different Tm were introduced into the microfluidic channel. Liposomes having a lower Tm were prepared from 10 mol % cholesterol and 90 mol % DPPC and contained encapsulated 200 mM carboxyfluorescein. Higher Tm liposomes encapsulating 100 mM sulforhodamine B were prepared from 75 mol % DPPC, 15 mol % DSPC, and 10 mol % cholesterol. The Tm of the CFcontaining liposomes was measured with differential scanning calorimetry to occur over the range of 38.6 to 46.4 °C centered at 42 °C. Similarly, the SRB-containing liposomes transitioned from 36.9 to 50.3 °C with the maximum enthalply at 45 °C,. The ends of the microchannel were thermostated to 20 and 47 °C and the solution flow rate was set to 100 µL/h. As can be seen, the channel fluorescence was initially low as both liposome formulations contained their respective dyes at a quenched concentration. As the liposomes moved down the channel, they first passed through the Tm of the CF liposome formulation and the dye diffused into the extravesicular fluid causing the microchannel to become visibly green. This led to a marked increase in the green channel of the CCD camera that can be seen in Figure 4b at a location 0.8 mm from the low-temperature channel terminus. As the liposomes continued to migrate downstream in the channel, they passed through the Tm of the SRB liposome formulation and SRB was released into the extravesicular space causing the channel to become visibly red. This increase can be seen in the red channel of the color CCD camera in Figure 4b at ∼1.6 mm from the lowtemperature channel terminus. As the SRB was released into the extravesicular microfluidic environment, it quenched the fluores6910 Analytical Chemistry, Vol. 75, No. 24, December 15, 2003

Figure 5. (a) Fluorescence micrograph showing a solution of 100 mol % DPPC liposomes encapsulating 1 mg/mL ethidium bromide suspended in 1× TBE buffer containing 0.1 unit/mL calf thymus DNA flowing through a 100-µm fused-silica capillary under an applied temperature gradient of 20-80 °C over a 2-mm distance at a flow rate of 100 µL/h. (b) Normalized fluorescence intensity of the image in (a) as a function of position in the fused-silica capillary.

cence of the CF, as can be seen by the decrease in the fluorescence intensity of the green CCD camera channel at the position in the channel where the SRB solution was released. This is to be expected since the excitation of SRB closely matches the emission of CF. Assuming an encapsulation efficiency of 10%, mixing of the contents in the liposomes would result in an average distance between fluorophores of 8 nm, which is well within the range for fluorescence resonance energy transfer. The increased temperature did not account for the loss of green emission, as the fluorescence yield of CF is very weakly temperature dependent. Moreover, this figure shows the sequential, controlled release of two different species and their interaction at a controlled point in the microfluidic channel, as would be necessary if one were to implement this system for multispecies activation and mixing in a microfluidic device. Controlled Reaction of Biological Species. Often important in a variety of DNA analysis techniques is the ability to label DNA fluorescently, thus enabling their sensitive detection. Typically, this labeling procedure must be performed after a biological reaction, since the labeling procedure can alter the behavior of the DNA in the biological assay. Figure 5a demonstrates the use of thermally triggered liposomes for the controlled fluorescent labeling of DNA molecules which could be implemented after an on-chip biochemical reaction. In this experiment, a solution of 100% DPPC liposomes containing an intercalating DNA dye, ethidium bromide, was mixed with a solution of calf thymus DNA and introduced into the microfluidic device under a temperature gradient of 20-80 °C. (It should be noted that many intercalating DNA dyes were attempted using this procedure; however, the hydrophobic character common in these dyes prevented their effective use due to excessive nontriggered leakage from the liposome.) Initially, the ethidium bromide and DNA were separated from one another by the liposome membrane that was impermeable to both species; therefore, there was no detectable fluorescent signal. However, as the solution migrated across the microchannel to a higher temperature, the liposomes released the intravesicular ethidium bromide at a temperature of ∼32 °C, allowing it to interact with the DNA and form a fluorescent

complex. This effect can be seen in Figure 5b, which presents the normalized fluorescence profile versus position. At ∼0.4 mm from the low-temperature-channel terminus, the fluorescence intensity increased and reached a relatively constant value for the remainder of the microchannel, indicating the controlled release of the ethidium bromide and its intercalation with the DNA. Again, this “mixing” of ethidium bromide and DNA occurred over a distance of ∼200 µm; thus, with this technique only a small portion of the microfluidic channel was required to complete the labeling procedure as is demonstrated in this example. Both passive and active micromixers described earlier operate optimally at certain flow rates, owing to their fluid mechanical design. Mixing with thermally triggered liposomes, however, is controlled at first order by the temperature distribution in a microfluidic system; thus, its mixing efficiency is likely to be less impacted by operational flow rates than microfabricated mixers. Further, mixing with this technique obviates the need to fabricate specialized topographical features in the microfluidic channel. However, it does require the microfluidic device be operated under temperatures compatible with the particular liposome formulation used. Also, the chemical species to be mixed with this technique must be of sufficient hydrophilicity that they do not spontaneously diffuse across the lipid bilayer of the liposome over the time period of the analysis. Biological species, such as DNA, RNA, and proteins, are particularly well suited for this mixing approach. CONCLUSION We have demonstrated the initial segregation and subsequent controlled release of reagents in microfluidic devices using a system of bioinspired thermally triggered liposomes. This system takes advantage of the dramatically increased permeability of the lipid bilayer of a liposome near its gel-to-liquid phase transition temperature (Tm). Thus, components isolated from the extrave-

sicular environment by the liposome are released when the local temperature in a microfluidic device is brought to the Tm of the liposome. As the Tm is a function of liposome formulation, a mixed population of liposomes can be used to control a series of sequential automated reactions where the temperature in the channel determines the reaction timing and sequence. While there are other triggers than can permeabilize a liposome (i.e., pH, ionic strength, or electric field to name a few) we believe that temperature is a well-suited actuator for liposome release in microfluidic systems due to the ease with which temperature can be modulated in these devices. Since the liposomes that encapsulate the species of interest are dispersed throughout the channel, upon release from the liposome, “mixing” is inherently quite rapid and is accomplished within ∼200 µm of channel distance, corresponding to an average mixing time of 110 ms. Additionally, because this “mixing” technique is controlled by the Tm of the liposomes, rather than the fluid mechanical properties of the channel, this technique should be less sensitive to different operational flow rates than traditional microfabricated mixers. ACKNOWLEDGMENT W.N.V. acknowledges financial support of the National Research Council/National Institute of Standards and Technology (NIST) Postdoctoral Research Program. The authors also acknowledge Dr. Michael Gaitan (NIST) and Dr. Nicole Morgan (NIST) for the fabrication of the silicon templates used in these studies. W.N.V. acknowledges numerous helpful conversations with Dr. David Ross (NIST).

Received for review July 25, 2003. Accepted October 3, 2003. AC034850J

Analytical Chemistry, Vol. 75, No. 24, December 15, 2003

6911