Generation of Femtoliter Reactor Arrays within a Microfluidic Channel

Jul 9, 2012 - Japan Science and Technology Agency (JST), 4-6-1, Komaba, Meguro-ku, Tokyo 153-8505, Japan. Anal. Chem. , 2012, 84 (15), pp 6346–6350 ...
0 downloads 12 Views 2MB Size
Download PDF
Technical Note pubs.acs.org/ac

Generation of Femtoliter Reactor Arrays within a Microfluidic Channel for Biochemical Analysis Sadao Ota,† Hiroaki Kitagawa,† and Shoji Takeuchi*,†,‡,§ †

Institute of Industrial Science, The University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo, 153-8505, Japan Exploratory Research for Advanced Technology (ERATO), 4-6-1, Komaba, Meguro-ku, Tokyo 153-8505, Japan § Japan Science and Technology Agency (JST), 4-6-1, Komaba, Meguro-ku, Tokyo 153-8505, Japan ‡

S Supporting Information *

ABSTRACT: We present a simple microfluidic method to generate highdensity femotoliter-sized microreactor arrays within microfluidic channels. In general, we designed a main channel with many small chambers built into its walls. After sequentially infusing aqueous solution and organic solvent from a single tube into the device, aqueous droplets are confined in the chambers by the solvent flow. The generated reactors are small and stable enough for carrying out ultrasensitive biochemical assays at single molecule levels. As a demonstration, in this paper, we optically observed hydrolysis activity of βgalactosidase enzymatic molecules in the reactor arrays at single molecule levels. Further, this method has the following advantages: (1) the droplets are observable immediately after formation and (2) its simple procedure is sufficiently robust such that even handy infusion of the preloaded solutions is reproducible. We believe our method provides a platform attractive to a variety of single molecule studies and sensing applications such as clinical diagnostics. first step to this integration is to generate the stable femtoliter reactor arrays within the microchannel. Various microfluidic systems have been developed to perform experiments with the reactors on the scales of pico- or nanoliters but not yet on that of the femtoliters. For example, there exists a field of digitalmicrofluidics, wherein monodisperse droplets in an immiscible solvent are generated and dynamically manipulated through microfluidic circuits.14−26 Typically, however, the droplets are not spatiotemporally stable and the size remains relatively large (>1 pL) to prevent coalescence. We also note that efforts have been made to trap pregenerated droplets into microfabricated static chambers.19−26 Still, the size of droplets is relatively large mainly because the required energy to hold the droplets may easily break them up, and the chamber density is limited due to the required chamber geometry. Here, we developed a method that utilizes an immiscible liquid phase to instantly enclose thousands of uniform femtoliter droplets within microchambers built in microchannel walls. The generated reactors are small and stable enough to study enzymatic activities at single molecule levels. The generation process is semiautomatic, thus free from any technical difficulty. The droplet arrays are observable immediately after formation, allowing more efficient experiments. In addition, this process is sufficiently simple and robust such that even handy infusion of the preloaded solutions could reproducibly generate uniform droplet arrays. As a demon-

T

his paper describes a simple microfluidic method to generate high-density microreactor arrays that are small (∼100 fL) and stable (>1 h) enough for carrying out ultrasensitive bioassays at single molecule levels. Microfluidic technologies are promising to minimize the labor cost and time in laboratory operations and also to improve efficiency of the experiments via miniaturization of the sample volume and precise control of the experimental conditions.1−4 Microreactors, with discrete uniform volumes of reagents typically segregated by microfabricated solid structures or immiscible liquid phases, are fundamental components in the microfluidic systems.5−26 These small containers, ranging from nano- to femtoliters, not only reduce the reagent volume used but also act as highly sensitive platforms to detect reactions occurring inside: a minute quantity of the reaction product increases its concentration in the microreactors much more rapidly than in bulk, generating detectable signals in a sufficiently short time. This feature was first made useful by the Noji group by isolating enzymes in the reactor arrays to detect their catalytic activities, even at single-molecule levels when the reactor volume was as small as femtoliters.5,6,8−12 Later, this approach was further employed for highly sensitive detection of serum protein molecules at subfemtomolar concentrations.13 While these femtoliter microreactor arrays for single molecule studies have been proven powerful, integration of such reactors with microfluidics is still missing. Facilitating direct contact between the femtoreactors and a microfluidic channel will let us fully utilize the potential of microfluidics, such as precise control of concentrations of chemicals, dynamic exchange of solutions and automation of experiments.14 The © XXXX American Chemical Society

Received: May 5, 2012 Accepted: July 9, 2012

A

dx.doi.org/10.1021/ac301204v | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Technical Note

stration, by enclosing β-galactosidase enzymatic molecules and fluorecein-di-β-D-galactopuranoside in the reactor arrays, we optically observed the hydrolysis activity of the biomolecules at single molecule levels.

They were also permanently bonded to form a closed channel structure. When preloading aqueous solution and organic solvent in the single tube before the experiments, no air bubbles remaining at the interface of water/oil was allowed to form clean droplets in the end. During the fluidic process, the solutions were continuously infused into the channel at a constant rate of 2 nL/min using a syringe pump (KD Scientific). For preventing water absorption into porous PDMS walls, the fabricated device was immersed and degassed in a water bath overnight to saturate the PDMS with water. In the case of Figure 1a, we performed experiments in the PDMS device immediately after taking it out from the water bath. On the other hand, in cases of Figure 1b, we modified the PDMS surface to be highly hydrophobic. For this modification, we infused (tridecafluoro1,1,2,2-tetrahydrooctyl)-1-trichlorosilane (Sigma Aldrich) diluted in ethanol at 20% concentration in volume into the device, followed by baking at 120 °C for 3 min, rigorous washing with clean ethanol, and flushing away the excess solution from the channel by nitrogen gas. Fluorescent beads were purchased from Invitrogen and imaged under a normal fluorescence microscope. A model enzyme, β-galactosidase, was purchased from Roche, and its molecular weight was 540 kDa. The experiments were conducted in a 100 mM phosphate buffer at pH 7.5, in the presence of 0.05 to 10 mg/mL BSA, 1 mM MgCl2, and 2 mg/ mL mercaptoethanol. The concentration of fluorescein di-β-Dgalactopyranoside (FDG) was 200 μM . The enzymes were diluted down to 8.0 ng/mL, corresponding to the concentration of 1.07 enzymes per 120 fL chamber.

■

EXPERIMENTAL SECTION The microfluidic device made of polydimethylsiloxane (PDMS) has a main channel with a number of small chambers built into either its bottom- or sidewalls, as depicted in parts a and b of Figure 1, respectively. A single microfuidic process results in

Figure 1. 3D schematic of fluid−fluid shear-induced generation of uniform water droplets in microchamber arrays, either in the bottomor side-walls of the PDMS channels. In part a, the device is composed of two layers of PDMS: one has a microchannel and the other has dense microchamber arrays. These two layers are bonded while the channel and chambers are aligned to face each other. In part b, a single layer of PDMS has a channel with a number of small chambers built into its walls. This layer is bonded to a glass coverslip to form an enclosed channel system. In the experiments, an immiscible organic solvent and a sample solution are sequentially loaded into a single tube and infused into the device in reverse order, resulting in instantaneous formation of femtoliter droplet arrays.

■

RESULTS Microreactor Generations in the Two Different Configurations. In the configuration shown in Figure 1a, we are easily able to scale down the chamber size to femtoliter orders while increasing their density. To confirm that there was no leakage from the generated microreactors, we enclosed fluorescent beads (200 nm) as shown in Figure 2a,b. When diluting the concentration of the beads, we could observe the Brownian motion of individual beads and confirm their localization within each reactor, showing that a stable liquid phase was present (insert of Figure 2b and movie in the Supporting Information). We never observed any bead escape from the chambers, proving no possibility of connecting water channels. The configuration shown in Figure 1b limits the number of chambers observable in a single field of view. Instead, we are easily able to observe how the immicible phase of water and oil as well as of water and PDMS walls are formed since it is perpendicular to the imaging planes. As an example, when we modified the surface of the channels to be highly hydrophobic, the surface repelled water and the formed droplets became spherical (Figure 2c,d and movie in the Supporting Information). Furthermore, this configuration let us observe the water absorption into porous PDMS walls more in detail (data not shown). In the case of the PDMS devices without water saturation, the wall absorbed water relatively fast and droplets disappeared less than after 1 h. When we treated the device by immersing and degassing it in water overnight, little absorption was observed over several hours. Further, when the device was kept immersed inside water (with a homemade water sink) during experiments, the absorption could be perfectly prevented. In addition, the advantage that we can

instantaneous droplet generations. In advance, a fluorocarbon oil (3M Performance Fluid PF-5060 from Sumitomo 3 M Limited, Japan) and an aqueous solution that later constitutes the content of the droplets are sequentially loaded into a single tube. In the process, these solutions are infused into the device at a constant rate, in reverse order. The buffer solution first fills the device, while pushing residual air bubbles out through the porous PDMS walls.27,28 The infused immiscible oil then sweeps away the first solution from the channel but confines the residual aqueous solution within the chambers, instantly generating static droplet arrays. All the microfluidic devices used in this paper were fabricated using standard soft lithography. The device configuration shown in Figure 1a is composed of two plates of PDMS: one has a microchannel and the other has a dense microchamber array. In this work, we fabricated the femtoliter-sized chamber array (5 μm in diameter and 6 μm in depth, corresponding to ∼120 fL in volume) and a channel (150 μm in width and 20 μm in height). These two layers were permanently bonded in a way that the channel and chambers were aligned to face each other. This bonding was performed after applying oxygen plasma to the PDMS surfaces, followed by baking at 80 °C for 1 h for dehydration. On the other hand, the configuration shown in Figure 1b is composed of a glass coverslip and a plate of PDMS that has a microchannel with a series of microchambers in its sidewalls. The height and width of the channel was 20 and 30 μm, respectively, and the depth of the chambers was 20 μm. B

dx.doi.org/10.1021/ac301204v | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Technical Note

aim here was to directly observe the individual activities of isolated single enzymes. We confirmed this idea using βgalactosidase (β-Gal) and fluorecein-di-β-D-galactopuranoside (FDG) as model enzymes and substrates.5,12 A mixture of the enzyme and substrate was preloaded in the tube and infused into the microchannels, followed by the organic solvent for closing. The confined β-gal hydrolyzed the FDG to fluorescein over time, increasing the fluorescent intensity in the chambers. For this experiment, we used the configuration shown in Figure 1a, high-density 120 fL microreactor arrays, to simultaneously measure the fluorescent intensity change of ∼300 chambers in a single view field. When diluting the enzyme solution down to a ratio close to 1:1 enzyme per chamber, the distribution of brightness in the array became inhomogeneous (Figure 3a,b). We took the intensity differences between two images, taken at an interval of 2 min. These differences, which were proportional to the enzymatic activities in each reactor, showed a clear quantization (Figure 3c). The histogram in Figure 3d, plotting the intensity increments in the 2 min, showed four peaks. These peaks were equally spaced, indicating that the increment value was corresponding to the activity of the single enzymes. We attributed these peaks to the presence of zero, one, two, or three enzymes in the corresponding chambers, respectively. The area under each peak then corresponded to the number of chambers that contained the quantized number of enzymes. The resultant histogram shown in the inset of Figure 3d showed consistency with the Poisson statistics, which is expected for a random distribution of molecules over the reactors.5,12 By fitting this distribution with the theoretical Poisson distribution and comparing it with the concentration of enzymes in the solution used in the experiment, we could estimate that ∼88% of the enzymes remained active. We believe that these results confirmed the usefulness of our femtoliter-

Figure 2. Generation of water droplets in the microchamber array and simultaneous particle encapsulation. Panels a and b are fluorescent images of the device, operated as depicted in Figure 1a. In part a, water containing fluorescent 200 nm diameter beads filled the whole device (scale bar is 50 μm). After oil flushed away the water from the channel, a uniform water droplet array was generated within each chamber and the beads were simultaneously encapsulated inside as seen in part b. The insert in part b is a magnified fluorescent image of the fluorescent beads with low white light illumination (scale bar is 10 μm). Panels c and d are bright-field images of the device, operated as depicted in Figure 1b (scale bar is 20 μm). In parts c and d, the inner walls of the PDMS devices were surface-modified with silane-coupling reactions to become highly hydrophobic. Consequently, the generated droplets were formed in a spherical shape within the rectangular chambers.

observe liquid interfaces has been proven beneficial when using it as a platform for forming lipid bilayer membranes.27,28 Performing Enzymatic Assays at Single Molecule Levels. In the case of enzymes, the microarray traps both the proteins and their substrates/products in the same reactor. Again, the extremely small volume lets a minute quantity of enzymatic products reach a detectable concentration, and our

Figure 3. Digitized increase in fluorescence intensity of individual chambers confirmed detection of the activity of confined enzymes, β-galactosidase, at single molecule levels. The β-gal was dispersed and confined in an arrayed 120 fL microchamber at the low concentration of 8.0 ng/mL, theoretically corresponding to 1.07 enzymes per chamber. This confined β-gal then hydrolyzed fluorecein-di-β-D-galactopuranoside to fluorescein over time. Panels a, b, and c are fluorescent images of the enzymatic activity in the chamber arrays (scale bar is 50 μm). Images started to be taken just after the oil swept away the enzyme solution from the channel to generate the droplet array of the same solution (left panel, a), and 2 min later (middle panel, b). The right panel (c) shows the intensity difference between these two images. In the following statistic analysis, we ignored the far left and far right droplet columns in a channel, wherein the droplet enclosure seemed less controlled due to hydrodynamic drag by the channel sidewalls. In part d, the histogram shows the number of chambers versus the increase in fluorescence intensity over 2 min from part c. We attributed the four peaks found in part d, fit with a sum of Gaussians, to occupancy of 0, 1, 2, or 3 enzymes in each chamber. The insert shows the occupancy distribution in the experiment. We summed up the number of chambers below each Gaussian peak to obtain the ratio (x) of chambers with an occupancy of N enzymes (N = 0, 1, 2, 3). This ratio was shown as bars in the inset and fit with the statistical Poisson distribution x = μN e−μ/N! (spots), wherein μ, the average number of active enzymes per chamber, was estimated using the least-squares method. The good fit and quantitatively consistent μ value of 0.945 confirmed the previous attribution and indicated 88.3% of the confined enzymes (1.07 enzymes/chamber) remained active. C

dx.doi.org/10.1021/ac301204v | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Technical Note

several advantages to the device. For instance, the generated reactors were very stable and, to our knowledge, the smallest among the ones integrated in a fluidic system. Also, since the generation is instantaneous such that the flow-rate can be quasistatistic, this simple method was robust enough that even handy infusion of the solutions could reproducibly form uniform droplet arrays. Moreover, our femtoreactor array integrated with more complex microfluidic systems is attractive to a variety of sensing applications such as clinical diagnostics. In a recent work, for example, magnetic beads were used as substrates to capture enzymes at minute concentrations and confined in the femtoliter-sized microreactors segregated by solid structures.13 The sensitivity of the system reached femtomolar levels, proving significant potential of the femtoreactors for sensing purposes. If applying our fluidic system, it may more efficiently trap the beads while further reducing the reagent volume and experimental steps. Additionally, many functions of microfluidic systems such as washing of beads, concentrators, and mixers of molecules can be potentially integrated into these assays, further improving the efficiency and sensitivity of the assays. In general, with adequate modifications, a variety of assays can be performed in the generated reactors, including chemical synthesis,29 catalyst discovery, amplification of nucleic acids, microbial detection,30 immunoassays,13 single cell analysis,31 and DNA sequencing.15

sized microreactor arrays for studying enzymatic hydrolysis activities at single molecule levels.

■

DISCUSSION Unique feature of our reactors, compared to those formed by existing techniques, is that they face the fluid flow in the microchannel. By just preloading several plugs of aqueous solutions in the tube, the droplets of the previous solution could be quickly washed away by the next solution, which then became the next droplets. We were therefore able to repeatedly replace the whole reactor contents from one to another in the same channel. We also note that we can more precisely replace or control the contents of droplets by designing more channels connected to the chamber area. The remaining challenge is to partially exchange the contents of individual chambers. We believe that the idea of making ducts under the generated droplets, proposed by Du et al., may be applied to solve this issue in the near future.25 On the other hand, when the size of the reactors becomes very small and the surface to volume ratio subsequently becomes large, the surface effects on molecules could presumably become significant such as surface denaturation of the biomolecules. In the present experiment of detecting enzymatic activity, we coated the PDMS surfaces with an excess amount of BSA to minimize the adverse effect.5 Further, we inspected the system for still possible surface denaturation effects. To do so, we again took a series of fluorescent images of the same reactor arrays over 5 min at intervals of 1 min. The fluorescent intensity of some representative chambers was plotted over time (Supporting Information). This measurement of the enzymatic activities in each chamber also revealed obvious inhomogeneity in the rate of fluorescent increase, confirming that the each reactor contains only a few molecules. Over this time period, the product concentration continuously increased in each reactor, showing that activities were maintained. Thus, there was no significant adverse surface effect on molecules such as permanent inactivations. We note that there is still a possible issue of temporal inactivations on biomolecules due to their nonspecific binding to the PDMS surfaces. Although the estimated activity of the confined enzymes was as high as that in previous works, this issue might cause less discrete distribution in Figure 3d.5 We believe that we can solve this issue via more careful surface modifications of the PDMS surfaces. For example, when the walls are more hydrophobic, the generated droplets become spherical, as shown in Figure 2d, minimizing the physical contacts between the solid walls and molecules in the reactors.

■

■

ASSOCIATED CONTENT

* Supporting Information S

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Fax: +81 03 5452 6649. Phone: +81 03 5452 6650. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS We thank Dr. Wei-Heong Tan and Daisuke Kiriya for helpful discussions. REFERENCES

(1) Beebe, D. J.; Mensing, G.; Walker, G. M. Annu. Rev. Biomed. Eng. 2002, 4, 261−286. (2) DeMello, A. J. Nature 2006, 442, 394−402. (3) Whitesides, G. M. Nature 2006, 442, 368−373. (4) Dittrich, P. S.; Manz, A. Nat. Rev. Drug Discovery 2006, 5, 210− 218. (5) Rondelez, Y.; Tresset, G.; Tabata, K. V.; Arata, H.; Fujita, H.; Takeuchi, S.; Noji, H. Nat. Biotechnol. 2005, 23, 361−365. (6) Rondelez, Y.; Tresset, G.; Nakashima, T. Nature 2005, 433, 773− 777. (7) Tan, W.; Yeung, E. S. Anal. Chem. 1997, 69, 4242−4248. (8) Rissin, D. M.; Walt, D. R. Nano Lett. 2006, 6, 520−523. (9) Rissin, D. M.; Walt, D. R. J. Am. Chem. Soc. 2006, 128, 6286− 6287. (10) Rissin, D. M.; Gorris, H. H.; Walt, D. R. J. Am. Chem. Soc. 2008, 130, 5349−5353. (11) Gorris, H. H.; Rissin, D. M.; Walt, D. R. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 17680−17685. (12) Sakakihara, S.; Araki, S.; Iino, R.; Noji, H. Lab Chip 2010, 10, 3355−3363.

CONCLUSIONS

We developed a method of generating the microreactor array that is compatible with a dynamic fluidic system. This reactor array is femtoliter-sized, spatiotemporally stable, and highdensity. Using this array, we have demonstrated the detection of enzymatic activities at single molecule levels. This method utilizes a continuous flow of an organic solvent in the channel for enclosing water droplets in the chambers. Because of this fluidic access to the droplets, we could exchange the droplets of one solution with those of another by preloading multiplugs in a tube. Also, in a fixed configuration, enclosing efficiency of the droplets simply depends on the flow rate of the fluidic flow, which is corresponding to pressure in the fluidic channel. This flow rate is typically easy to control and this feature provides D

dx.doi.org/10.1021/ac301204v | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Technical Note

(13) Rissin, D. M.; Kan, C. W.; Campbell, T. G.; Howes, S. C.; Fournier, D. R.; Song, L.; Piech, T.; Patel, P. P.; Chang, L.; Rivnak, A. J.; et al. Nat. Biotechnol. 2010, 28, 595−599. (14) Teh, S. Y.; Lin, R.; Hung, L. H.; Lee, A. P. Lab Chip 2008, 8, 198−220. (15) Sims, P. A.; Greenleaf, W. J.; Duan, H.; Xie, X. S Nat. Methods 2011, 8, 575−580. (16) Takinoue, M; Onoe, H.; Takeuchi, S. Small 2010, 6, 2347− 2377. (17) Brouzes, E.; Medkova, M.; Savenelli, N.; Marran, D.; Twardowski, M.; Hutchison, J. B.; Rothberg, J. M.; Link, D. R.; Perrimon, N.; Samuels, M. L. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 14195−14200. (18) Song, H.; Chen, D. L.; Ismagilov, R. F. Angew. Chem., Int. Ed. 2006, 45, 7336−7356. (19) Pompano, R. R.; Liu, W.; Du, W.; Ismagilov, R. F. Annu. Rev. Anal. Chem. (Palo Alto, Calif.) 2011, 4, 59−81. (20) Boukellal, H.; Selimović, S.; Jia, Y.; Cristobal, G.; Fraden, S. Lab Chip 2009, 9, 331−338. (21) Edd, J. F.; Humphry, K. J.; Irimia, D.; Weitz, D. A; Toner, M. Lab Chip 2009, 9, 1859−1865. (22) Shi, W.; Qin, J.; Ye, N.; Lin, B. Lab Chip 2008, 8, 1432−1435. (23) Casadevall, I; Solvas, X.; deMello, A. Chem. Commun. 2011, 47, 1936−1942. (24) Li, L.; Ismagilov, R. F. Annu. Rev. Biophys. 2010, 39, 139−158. (25) Du, W.; Li, L.; Nichols, K. P.; Ismagilov, R. F. Lab Chip 2009, 9, 2286−2292. (26) Liu, W.; Chen, D.; Du, W.; Nichols, K. P.; Ismagilov, R. F. Anal. Chem. 2010, 82, 3276−3282. (27) Ota, S.; Suzuki, H.; Takeuchi, S. Lab Chip 2011, 2485−2487. (28) Ota, S.; Yoshizawa, S.; Takeuchi, S. Angew. Chem., Int. Ed. 2009, 48, 6533−7. (29) Hatakeyama., T; Chen, D. L.; Ismagilov, R. F. J. Am. Chem. Soc. 2006, 128, 2518−2519. (30) Boedicker, J. Q.; Li, L.; Kline, T. R.; Ismagilov, R. F. Lab Chip 2008, 8, 1265−1272. (31) Taniguchi, Y.; Choi, P. J.; Li, G. W.; Chen, H.; Babu, M.; Hearn, J.; Emili, A.; Xie, X. S. Science 2010, 329, 533−538.

E

dx.doi.org/10.1021/ac301204v | Anal. Chem. XXXX, XXX, XXX−XXX