Raman-Activated Cell Sorting Based on Dielectrophoretic Single-Cell ...

Article pubs.acs.org/ac

Raman-Activated Cell Sorting Based on Dielectrophoretic Single-Cell Trap and Release Peiran Zhang,†,‡ Lihui Ren,† Xu Zhang,† Yufei Shan,† Yun Wang,† Yuetong Ji,† Huabing Yin,∥ Wei E. Huang,†,§ Jian Xu,*,† and Bo Ma*,† †

Single Cell Center, CAS Key Laboratory of Biofuels and Shandong Key Laboratory of Energy Genetics, Qingdao Institute of Bioenergy and Bioprocess Technology, Chinese Academy of Sciences, Qingdao, Shandong 266101, China ‡ College of Marine Life Sciences, Ocean University of China, Qingdao, Shandong 266003, China § Department of Engineering Science, University of Oxford, Parks Road, Oxford OX1 3PJ, United Kingdom ∥ Division of Biomedical Engineering, School of Engineering, University of Glasgow, Glasgow G12 8LT, United Kingdom S Supporting Information *

ABSTRACT: Raman-activated cell sorting (RACS) is a promising single-cell technology that holds several significant advantages, as RACS is label-free, information-rich, and potentially in situ. To date, the ability of the technique to identify single cells in a high-speed flow has been limited by inherent weakness of the spontaneous Raman signal. Here we present an alternative pause-and-sort RACS microfluidic system that combines positive dielectrophoresis (pDEP) for single-cell trap and release with a solenoid-valve-suction-based switch for cell separation. This has allowed the integration of trapping, Raman identification, and automatic separation of individual cells in a high-speed flow. By exerting a periodical pDEP field, single cells were trapped, ordered, and positioned individually to the detection point for Raman measurement. As a proof-of-concept demonstration, a mixture of two cell strains containing carotenoid-producing yeast (9%) and non-carotenoid-producing Saccharomyces cerevisiae (91%) was sorted, which enriched the former to 73% on average and showed a fast Raman-activated cell sorting at the subsecond level.

R

off the substrate into a prepared container.10 In this case, the interrogation of SCRS is no longer a significant issue because the cells are kept static on the substrate. However, the relatively low sorting throughput (several cells per minute) still hampers its application to sort cells in aqueous samples.10 To achieve a high-throughput sorting, a flow-mode RACS (like FACS) is desired. The challenge of obtaining SCRS from fast-moving cells could potentially be addressed by improving the Raman detection sensitivity or by immobilizing a single cell at the laser detection spot for a period to allow an elongated acquisition time.11 Surface-enhanced Raman scattering (SERS) favors the former strategy and provides an ultrasensitive method for the detection, which could enhance the intensity of the Raman signal by 6−14 orders of magnitude via employing a colloidal suspension of metal nanoparticles or a patterned surface.12 This enables the interrogation of Raman spectra in a high-speed flow, and previous studies have demonstrated that cells could be identified in a high-throughput manner with the aid of SERS tags (nanometal particles bonded with SERS markers/antibodies).13−15 However, SERS is still impractical for RACS because it is usually difficult to directly

aman spectroscopy detects the inelastic scattered photons which correspond with the vibrational energy levels of the bonds in a sample molecule.1 By detecting vibrational signals from the molecules of a single cell, a typical single-cell Raman spectrum (SCRS) could provide an intrinsic chemical “fingerprint” of the cell, making it possible to directly interrogate cellular compositions,2−4 identify cell species,5−7 and distinguish physiological states8,9 in a label-free manner. On the basis of SCRS, Raman-activated cell sorting (RACS) holds several significant advantages, as RACS is label-free, information-rich, and potentially in situ compared to fluorescence-activated cell sorting (FACS).2 However, the spontaneous Raman signal is inherently weak, as only 1 in 106−108 photons incident on the sample will undergo Raman scattering, which usually requires a long interrogation time (several seconds to a few minutes, versus usually tens of microseconds needed for FACS) for measurement of an individual cell.2,9 The weak Raman signal has made it impractical to obtain an SCRS from a fast-moving cell for RACS, despite the fact that the optical module has been optimized and the acquisition time of SCRS has been reduced to 0.1 s or less in our previous work.10 A static version of RACS called the Raman-activated cell ejection (RACE) system has been developed for sorting the cells distributed on a CaF2 slide (coated with a light-adsorbing interlayer) by applying a pulsed laser to eject the selected cell © 2015 American Chemical Society

Received: October 24, 2014 Accepted: January 21, 2015 Published: January 21, 2015 2282

DOI: 10.1021/ac503974e Anal. Chem. 2015, 87, 2282−2289

Article

Analytical Chemistry

subsecond level. By exerting a periodical pDEP field (sine, 10 MHz, 14 Vpp) on the electrode array, single cells were trapped, ordered, and delivered edge-to-edge to the laser detection point to be measured by Raman spectroscopy one-by-one with adjustable trapping duration, ranging from 90 ms to minutes. By employing a fine-tuned hydrodynamic suction force generated by a solenoid valve,31 the selected cells were sucked into the storage chamber and temporarily trapped on the electrode array by pDEP during the sorting process. Homemade software, Qspec,32 was developed for integrating and synchronizing the RACS system. As a proof-of-concept demonstration for this pDEP-based RACS prototype, a yeast mixture of two strains containing carotenoid-producing yeast (9%) and non-carotenoid-producing S. cerevisiae (91%) was sorted, and the carotenoidproducing yeast cells were enriched to 73% on average. This system overcomes the inherent weak problem of SCRS and presents an alternative strategy for RACS which can sort cells at a subsecond level. It thus laid the foundation for highthroughput single-cell sorting based on the intrinsic chemical “fingerprints” of individual cells.

obtain reproducible and interpretable Raman spectra, particularly from single cells, which is of huge biochemical complexity.2 Laser tweezers Raman spectroscopy (LTRS) is a powerful label-free approach for manipulating and analyzing individual cells/particles in the aqueous solution.11 The technique uses a single tightly focused laser beam as an optical tweezers to trap cells/particles to elongate the acquisition time and also as an excitation source to stimulate the Raman signal from the trapped cell. The selected cell could then be moved out of the population (sorted) by optical tweezers (e.g., bacteria, yeast, and leukocytes) after measurement.16−20 However, the inability to capture and identify single cells under a high flowing speed is the key bottleneck for LTRS in high-speed RACS development due to the weak force of optical tweezers (100 aN to 100 pN).21,22 Dielectrophoresis (DEP) can provide a force ranging from 0.1 to 1 nN for particle manipulation, which is strong enough to trap single cells in a high-speed flow.22 DEP is the motion of a particle produced upon the interaction of a nonuniform electric field with the induced effective dipole moment of the particle.23 The induced moments interact with the nonuniform electric field, resulting in either an attractive force toward regions of high electric field density (positive dielectrophoresis, pDEP) or a repulsive force toward regions of low electric field density (negative dielectrophoresis, nDEP).23 For decades, DEP has been developed as a versatile tool for cell manipulation. For instance, mouse fibroblasts could be repulsed from the electrode edge and be easily cultivated during permanent field application.24 Fibroblast-like BHK-21 C13 cells could be attracted by an electrode array, and the doubling rate of cells exposed to DEP has been demonstrated to be not different from that of the control group.25 A single endothelial cell could be registered on a viscous surface by pDEP thereafter cultivated on that surface, and the difference between the proliferation rates of DEP-treated and non-DEP-treated cells has been proven unconspicuous.23 Single Saccharomyces cerevisiae cells could be caged in the center of octopole electrodes by nDEP, and they tolerated DEP treatment and reproduced well during the DEP caging.26 Studies also showed that nDEP was capable of concentrating pathogens from meat or the urinary tract at the laser detection spot for subsequent Raman identification.27,28 Single bacterial cells bonded with an antibody SERS tag could be captured and aligned with the laser detection spot by nDEP using quadrupole electrodes and thereafter measured by Raman.13 DEP has also been utilized for cell sorting. Lightinduced dielectrophoresis (ODEP) tweezers could capture and move single cells into the sorting outlet,29 and a mixture of cells could be sorted on the basis of intrinsic electric properties and concentrated by an nDEP trap for subsequent Raman measurement.30 However, there are still no reports about applications of DEP on RACS to address the need for orderly trapping and releasing single cells for Raman identification in a high-speed flow (>1 mm/s). Here, we propose an alternative strategy to overcome the key obstacle of SCRS acquisition in a high-speed flow by utilizing pDEP for single-cell trap/release, which aims to gain sufficient time for Raman spectrum acquisition by electronically trapping and holding the fast-moving single cells at the 532 nm laser detection spot repeatedly. On the basis of this strategy, an RACS system that integrates a pDEP single-cell trap/release unit and solenoid-valve-suction-based switch has been developed for trapping, identification, and automatic separation of individual cells, with the fastest sorting speed at the

■

EQUIPMENT Raman Setup. Single-cell Raman spectra were acquired by an optimized LabRam HR micro-Raman setup (Wellsens Biotech Ltd., China) described in our previous paper.10 A Nd:YAG 532 nm laser emitter (Ventus, Laser Quantum Ltd., United Kingdom) was used as the excitation light source. A 50× air objective (NA = 0.90, Olympus, United Kingdom) was employed to focus the laser beam on the sample. Raman signals were collected by an electron-multiplying charge-coupled device (EMCCD) (Newton DU970N-BV, Andor, United Kingdom) utilizing a 1600 × 200 array of 16 μm pixels with thermoelectric cooling to −70 °C. A 600 lines/mm grating was applied, and the spectral resolution was ∼1 cm−1 with 1581 data points, ranging from 600 to 2100 cm−1 for SCRS acquisition.10 In the optical setup, the Raman light path was shortened and a higher incident power was applied, which resulted in 300-fold enhancement of the Raman detection efficiency (e.g., the acquisition time of a single Escherichia coli cell under the improved Raman setup was 0.1 s; SNR (signalto-noise ratio) = 3.92).10 Dielectrophoresis Equipment. Alternating current (ac) voltages between 5 and 20 Vpp at frequencies of 1−10 MHz for dielectrophoretic cell trapping were generated by an arbitrary function generator (DG4620, RIGOL Ltd., China). The outputs of the function generator were connected to two pads of the electrodes on the microfluidic chip via conductive tapes. A sinusoid wave (output CH1) was modulated by a pulsed wave (output CH2) to generate a periodical ac field (Figure S1, Supporting Information) to trap/release cells automatically. A normally on relay (Q3F-1Z-12VDC, Zhengqi Ltd., China) was connected between the function generator and the electrodes of the microchip and was activated by a digital I/O unit (DIO-1616LX-USB, CONTEC Ltd., United States) to trigger immediate interruptions on pDEP for the ondemand release of cells. Sorting Switch. An individual cell that satisfies the Raman selection criteria was sorted by a solenoid-valve-suction-based switch described in our previous work.31 Briefly, the switch comprised a three-way solenoid valve (p01451, Cole-Parmer, Illinois) and a related controlling circuit, which includes a digital I/O unit (DIO-1616LX-USB, CONTEC) and a 2283

DOI: 10.1021/ac503974e Anal. Chem. 2015, 87, 2282−2289

Article

Analytical Chemistry

Inc., China) with a 5 μL syringe (injection rate of 3 μL/h). (2) To initialize the RACS system and start sorting: the parameters (as shown in Table 1) were set on the panel of Qspec software

transistor relay (pc817, SHARP Co., China) for controlling the power supplied to the solenoid valve. The “Comm.” port of the solenoid valve was connected to the suction actuation channel of the microfluidic chip through poly(ether ether ketone) (PEEK) tubing (o.d. = 1/16 in., i.d. = 508 μm; Upchurch Scientific, Washington). A piece of fused-silica capillary (length 13 cm, o.d. = 360 μm, i.d. = 150 μm; Labsmith Inc., Livermore, CA) was inserted into the tubing between the “Comm.” port and the chip; the “N.O.” and “N.C.” ports were connected to static reservoirs through PEEK tubing (o.d. = 1/16 in., i.d. = 508 μm; Upchurch Scientific). Control Module. Software for controlling the RACS system, Qspec, was developed for synchronizing the electronics (EMCCD, solenoid valve, and function generator, etc.) and adjusting the system parameters (e.g., spectrometer, XYZ stage motor, laser device of the RACS system). The sorting criteria, acquisition time, electric parameters, and energizing duration of the solenoid valve were set via the panel of the “RTD MODE” in Qspec software (Figure S2, Supporting Information).

Table 1. Parameter Settings for Sorting of CarotenoidProducing Yeast on the RACS Systema parameter

value

acquisition time optical parameters trap/release frequency release duration trapping waveform suction duration sorting criterion

30 ms 532 nm, 100 mW, 10-fold degradation 5 Hz 8 ms sine, 10 MHz, 14 Vpp 23 ms 1510 cm−1 > 410 AU

a The Raman background intensity at 1510 cm−1 is 155 AU, and the noise is 60 AU.

(Figure S2, Supporting Information), and then the sorting process was activated. (3) To stop sorting and validate the sorting efficiency: the sorted cells trapped in the storage chamber were identified by Raman to determine the ratio of carotenoid-producing yeast cells among the sorted cell population. (4) To export the sorted cells: the pDEP trap was switched off, and the stored cells were flushed to the sampling outlet by increasing the relative height of the solenoid valve.

■

METHODS Chip Fabrication and Hydrophilic Pretreatment. The cell sorting chip was composed of an indium tin oxide (ITO)− glass substrate (Shenzhen Nanbo Ltd., China) with 180 nm ITO transparent electrodes and a poly(dimethylsiloxane) (PDMS) slab with channels prepared by soft lithography. The detailed process of chip fabrication is illustrated by Figure S3, Supporting Information. The substrate and the PDMS slab were treated with a plasma cleaner (PLASMA-PREEN II-862, Plasmatic Systems, Inc., United States; 100 mW) for 30 s and aligned under a microscope with the lubrication of methanol. The aligned chip was baked at 80 °C for 24 h to achieve an irreversible bonding. A fused-silica capillary coated with polyacrylamide (Labsmith, i.d. = 50 μm, o.d. = 360 μm) was inserted into the sample inlet of the bonded chip with the lubrication of acetone and was sealed with premixed PDMS for sustainable and stable cell loading. The microchannels were flushed with 5% Pluronoic-F127 (Sigma-Aldrich Co. LLC) for 10 min and washed with deionized (DI) water for 1 min prior to operation. Cell Culture and Sample Preparation. The carotenoidproducing and non-carotenoid-producing S. cerevisiae strains were cultured in the standard liquid YPD medium at 30 °C and stored at 4 °C. The harvested cells were washed three times with deionized water and subjected to 5000g centrifugation to remove impurities. The medium and water used in this study were sterilized by being passed through 0.2 μm filters before use. The optical density (OD) of a 100 μL sample was measured at 600 nm by a plate reader (Synergy HT SIAFRTD, BioTek Ltd., United States). The two kinds of cells were diluted to the same OD value, typically 0.60. Then the diluted samples were mixed to achieve a specific cell ratio. The OD of the cell sample for loading was adjusted to 0.20. The actual cell ratio was later identified by sampling 100 or more cells using the Raman Points Mapping mode. Cell Sorting Procedure for RACS. The step-by-step cell sorting procedure is as follows: (1) To set up the microfluidic chip: the microchip was connected to reservoirs through PEEK tubing (o.d. = 1/16 in., i.d. = 508 μm; Upchurch Scientific), the heights of the buffer containers and the solenoid valve were adjusted to 200, 100, and 50 mm relative to the upper surface of the chip, and the cell mixture was loaded into the microchannel gently using a syringe pump (LSP01-2A, Longer

■

RESULTS AND DISCUSSION Chip Design. To perform Raman-activated cell sorting in a high-speed flow, two requirements need to be satisfied. One is to measure and identify the fast-moving cells one-by-one by Raman microspectroscopy, and the other is to separate the selected cells from the body flow precisely. In this work, positive DEP was employed to assist the Raman detection by trapping high-speed single cells to the laser detection spot with positional repeatability and adjustable trapping time, and the solenoid-valve-suction-based sorting switch was modified to separate the selected cells from the population ones. The microchip (Figure 1a−c) consisted of four functional units, including (1) a converging zone, (2) a single-cell trapping unit, (3) a buffering zone, and (4) a sorting switch. To realize the function of each unit, a system (Figure 1d,e) that integrates Raman microspectroscopy, DEP equipment, the solenoid-valvesuction-based sorting setup, and the related controlling software Qspec was developed (Figure S2, Supporting Information). The cell suspension was injected into the sample inlet of the microchip horizontally through a fused-silica capillary (Figure 1b) and was first converged to a single-cell stream by buffer 1 (Figure 1a) in the converging zone. Then cells flowed into the adjacent 10 μm wide microchannel crossed by an electrode array (single-cell trapping unit), traveled through a periodical pDEP field generated by applying an ac voltage (sine, 10 MHz, 14 Vpp, trap/release frequency 5 Hz) on the electrode array on the surface of the channel, and finally were captured by an electrode. Thereafter the trapped single cells were delivered to the 532 nm laser detection spot edge-to-edge and were measured one-by-one. Once the Raman spectrum of an individual cell satisfied the sorting criteria, the software Qspec would trigger an interruption of the pDEP to release the selected cell immediately and activate the sorting switch after a specific delay (depends on the fluidic velocity, typically set as 2284

DOI: 10.1021/ac503974e Anal. Chem. 2015, 87, 2282−2289

Article

Analytical Chemistry

Figure 1. RACS system overview. (a) Schematics of the RACS chip. The red spheres represent target cells, and the green triangle indicates the position of the 532 nm laser beam. Key: 1, converging zone; 2, single-cell trapping unit; 3, buffering zone; 4, sorting switch. (b) Picture of the RACS chip. The cell suspension was loaded through a fused-silica capillary, and the function generator outputs were connected to the pads of the electrodes on the microchip via conductive tapes. (c) Microstructure of the RACS chip. The laser detection spot was aligned with the middle point of the electrode edge (11 electrodes, width 25 μm, gap 25 μm). Schematic (d) and picture (e) of the RACS setup.

33 ms). Then the selected cell would be released from the trapped electrode and would fleet through the buffering zone (Figure 1a, buffer 2), which was designated to eliminate the hydrodynamic fluctuation caused by the on/off of the downstream sorting switch. The cell sorting was achieved by a solenoid valve connected to the suction pipe (Figure 1a, 100 μm width), aiming to separate the target cell accurately by the hydrodynamic suction. The suction channel was crossed by an electrode array (12 electrodes, gap 25 μm, width 25 μm), which serves as a pDEP trap for cell storage during sorting and validating. After sorting, by applying gentle pressure on the suction port, the sorted cells can be exported to the sampling outlet. Dielectrophoretic Single-Cell Trap/Release. To realize single-cell trapping in a high-speed flow for Raman identification, the following four factors are key: (1) capturing a fast-moving cell efficiently, (2) releasing the trapped cell stably once measured, (3) aligning the trapped cell with the 532 nm laser detection spot precisely, and (4) maintaining a stable cell stream by preventing the sedimentation of cells in the chip−tubing interface. In our preliminary trials, microfluidic chips with a single electrode pair have been used for DEP single-cell trapping and release experiments. As shown in Figure 2, by periodically applying positive DEP on a single electrode pair, the converged single cells can be trapped stably and be

aligned with the laser detection spot with positional repeatability for the subsequent Raman detection. However, the sticky cells might adhere to the edge of the electrodes and thus fail to be released immediately once the applied pDEP field is removed, which makes this strategy impractical. Several phenomena could occur upon adhesion: (1) measurement of upstream cells could be impossible when the sticky cells attach on the laser detection spot; (2) the alignment between the trapped cell and the laser detection spot could be disturbed, resulting in a declined throughput; (3) the sorting accuracy could deteriorate due to unexpected cell release. Surface hydrophilic pretreatment was capable of removing the majority of the adhered cells and could stay effective to sustain a clear surface for several minutes during sorting (Figure S4, Supporting Information; sample OD = 0.2, fluidic velocity 1 mm/s). However, the sporadic and unexpected irreversible adhesion would cause the sorting process to fail despite the employment of hydrophilic pretreatment. Enlarging the hydrodynamic release force by increasing the fluidic velocity could indeed remove the adhered cells, but the trapping efficiency would be reduced to an unacceptable low level ( 1510 cm−1 > 410 AU), the ratio of intensities (e.g., 10 > 1510 cm−1/1700 cm−1 > 3), or the difference of intensities (e.g., 500 > 1510 cm−1 − 1700 cm−1 > 200). Steps: 1, DEP-ON to trap a cell; 2, DEP-OFF to release the cell; 3, DEP-ON again to trap the next cell (note that the workflows in green and blue are independent of each other); 4, Raman spectrum acquisition; 5, comparison of the acquired spectrum with the sorting criteria; 6, DEP-OFF to release the criterion-satisfied cell; 7, activation of the sorting switch to separate the selected cell; 8, 9, acquisition of the next Raman spectrum.

Basically, the pDEP trap was turned on and off to trap and release cells to acquire SCRS periodically (step 1 → step 2 → step 3, Figure 6). If a spectrum met the sorting criteria, the program would trigger an instant interruption upon the periodical pDEP trap to release the selected cell forcedly (step 4 → step 5 → step 6, Figure 6) and then turn on the periodical pDEP trap again. After a specific delay, the program would activate the suction switch, and the released cell would be separated into the suction channel (step 6 → step 7 → step 8, Figure 6). If the spectrum failed to meet the criteria, the measured cell would be released and flow into the waste reservoir (step 4 → step 5 → step 9, Figure 6). 2287

DOI: 10.1021/ac503974e Anal. Chem. 2015, 87, 2282−2289

Article

Analytical Chemistry

Figure 7. Validation of the sorting efficiency by sorting a mixture of the carotenoid-producing yeast (9% of the population) and the non-carotenoidproducing S. cerevisiae (91% of the population). Raman spectra of the carotenoid-producing yeast cell (a, 30 ms acquisition time, 100 mW, 532 nm, on chip) and the non-carotenoid-producing S. cerevisiae cell (b, 100 ms acquisition time, 100 mW, 532 nm, on CaF2 slides) exhibit significant differences and could be distinguished at 1105, 1155, and 1510 cm−1 (highlighted with red squares), respectively, but not by the single-cell microscopic pictures. (c) Raman spectra of the 52 cells stored in the validation chamber. (d) Microscopic picture of the sorted cells in the storage chamber: 52 cells were trapped on the electrode array by pDEP (sine, 10 MHz, 5 Vpp) for subsequent validation. (e) Performance of the RACS prototype. “Unsorted” bar (left): 9% of the cells in the sample were carotenoid-producing yeast. “Sorted” bar (right): after sorting, the carotenoidproducing yeast cells were enriched on average to 73% (STD = 8%, n = 3), suggesting an 8-fold enrichment.

■

CONCLUSION AND OUTLOOK This study has demonstrated the prototype of a fast RACS system, which combines a pDEP unit with a solenoid-valvesuction-based switch for single-cell trapping and sorting based on intrinsic SCRS. This system allows label-free cell sorting at an average accuracy of 73%, which demonstrates rapid Ramanactivated cell sorting at the subsecond level. It has paved the way for the development of high-throughput RACS. The RACS chip exploited in this work was made of ITO− glass and PDMS, which has a strong Raman background and would interfere with the Raman spectra from single cells that do not contain strong Raman-active molecules. This would limit the application of this chip to sort other types of cells. However, this pausing and sorting style of pDEP-RACS should greatly expand the application of RACS by overcoming the challenge of inherent weak Raman signals. Meanwhile, development of a low Raman background pDEP-RACS chip made of pure quartz is ongoing in our laboratory. This chip is compatible with the current chip design and has low background in the “biofingerprint” region for sorting different kinds of cells, including tumor cells, stem cells, and microbial cells. This strategy also has the potential to provide a source of cells with specific phenotypes directly for downstream singlecell sequencing or subsequent cultivation.

Carotenoids have wide applications, which include food coloring agents, precursors of vitamin A in animal feed, additives in cosmetics, and antioxidants to reduce cellular and tissue damage, etc.33 Carotenoid-producing yeasts are an important workhorse for carotenoid production. Screening of carotenoid-producing yeast strains with high yield thus is crucial for improving the production. The difference in the carotenoid level in yeast cells is not easy to distinguish at the single-cell level under normal fluorescence microscopes, while the single-cell Raman spectrum can reveal the difference because of the strong Raman signals of carotenoids.34 Here a carotenoid-producing yeast (S. cerevisiae) was employed to test the performance of the RACS prototype. Raman spectra could be acquired continuously in the RACS system (Video S6, Supporting Information). As shown in Figure 7a,b, the carotenoid-producing yeast cell had three discriminable spectral feature bands (1105, 1155, and 1510 cm−1) compared to the non-carotenoid-producing cell, and the significant difference between the Raman spectra of carotenoid-producing and noncarotenoid-producing yeast cells can be clearly distinguished. Typically, the intensity of the peak around 1510 cm−1 was selected as the sorting criterion for the identification of the carotenoid-producing yeast cell. Once the signal satisfied the sorting criterion, the program would activate the sorting switch and separate the released cell into the suction channel. As elucidated by Figure 7c, 52 cells were sorted into the suction channel, among which 40 cells were carotenoid-producing yeast cells, and the ratio of the carotenoid-producing yeast was elevated to 77% (original sample 9%) over a duration of 9 min. At least 577 cells were trapped and identified during the sorting period of 540 s, which demonstrated a fast Raman-activated cell sorting at the subsecond level. The sorting performance of the RACS prototype was further validated by two additional cell sorting experiments. Figure 7e shows that an 8-fold enrichment could be obtained on this RACS prototype: on average, after sorting, the carotenoid-producing yeast cells were enriched to 73% (STD = 8%, n = 3). The throughput could be further improved by optimizing the RACS system. First, the throughput could be increased by reducing the channel height of the single-cell trapping unit (e.g., to 10 μm). Second, the throughput could also be improved by increasing the fluidic velocity and enlarging the electric voltage for DEP trapping.

■

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 Authors

*E-mail: [email protected]. Phone: + +86-0532-80662653. *E-mail: [email protected]. Phone: +86-0532-80662657. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was financially supported by a General Grant and Basic Research in Scientific Instrument Grant of the National Natural Science Foundation of China (31327001, 31470220) and a Scientific Instrument Development Grant and Key 2288

DOI: 10.1021/ac503974e Anal. Chem. 2015, 87, 2282−2289

Article

Analytical Chemistry

(27) Schröder, U. C.; Ramoji, A.; Glaser, U.; Sachse, S.; Leiterer, C.; Csaki, A.; Hübner, U.; Fritzche, W.; Pfister, W.; Bauer, M.; Popp, J.; Neugebauer, U. Anal. Chem. 2013, 85, 10717−10724. (28) Meisel, S.; Stöckel, S.; Rösch, P.; Popp, J. Food Microbiol. 2014, 38, 36−43. (29) Huang, K. W.; Wu, Y. C.; Lee, J. A.; Chiou, P. Y. Lab Chip 2013, 13, 3721−3727. (30) Chrimes, A. F.; Khoshmanesh, K.; Tang, S. Y.; Wood, B. R.; Stoddart, P. R.; Collins, S. S.; Mitchell, A.; Kalantar-Zadeh, K. Biosens. Bioelectron. 2013, 49, 536−541. (31) Zhang, Q.; Zhang, P.; Su, Y.; Mou, C.; Zhou, T.; Yang, M.; Zhou, T.; Xu, J.; Ma, B. Lab Chip 2014, 14, 4599−4603. (32) Ren, L.; Su, X.; Wang, Y.; Xu, J.; Ning, K. PeerJ 2014, 2, e436. (33) Aksu, Z.; Eren, A. T. Process Biochem. 2005, 40, 2985−2991. (34) Li, M.; Canniffe, D. P.; Jackson, P. J.; Davison, P. A.; FitzGerald, S.; Dickman, M. J.; Burgess, J. G.; Hunter, N.; Huang, W. E. ISME J. 2011, 6, 875−885.

Deployment Grant on Modern Agriculture from the Chinese Academy of Sciences (YZ201236, KSZD-EW-Z-021-1-5). We are grateful to Prof. Haihang Cui for his generous support of the simulation work. We also thank Franck Dhalenne, Jing Shen, and Xia Huang for technical support. We are also indebted to Zhibin Zhang and Shurong Liu for providing professional aid.

■

REFERENCES

(1) Ferraro, J. R. Introductory Raman Spectroscopy, 2nd ed.; Academic Press: Amsterdam, 2003; pp 1−6. (2) Huang, W. E.; Li, M.; Jarvis, R. M.; Goodacre, R.; Banwart, S. A. Adv. Appl. Microbiol. 2010, 70, 153−186. (3) Wang, T.; Ji, Y.; Wang, Y.; Jia, J.; Li, J.; Huang, S.; Han, D.; Hu, Q.; Huang, W. E.; Xu, J. Biotechnol. Biofuels 2014, 7, 58−70. (4) Ji, Y.; He, Y.; Cui, Y.; Wang, T.; Wang, Y.; Li, Y.; Huang, W. E.; Xu, J. Biotechnol. J. 2014, 9, 1512−1518. (5) Münchberg, U.; Rösch, P.; Bauer, M.; Popp, J. Anal. Bioanal. Chem. 2014, 406, 3041−3050. (6) Stockel, S.; Meisel, S.; Elschner, M.; Rosch, P.; Popp, J. Anal. Chem. 2012, 84, 9873−9880. (7) Li, M.; Ashok, P. C.; Dholakia, K.; Huang, W. E. J. Phys. Chem. A 2012, 116, 6560−6563. (8) Schie, I. W.; Alber, L.; Gryshuk, A. L.; Chan, J. W. Analyst 2014, 139, 2726−2733. (9) Huang, W. E.; Griffiths, R. I.; Thompson, I. P.; Bailey, M. J.; Whiteley, A. S. Anal. Chem. 2004, 76, 4452−4458. (10) Wang, Y.; Ji, Y.; Wharfe, E. S.; Meadows, R. S.; March, P.; Goodacre, R.; Xu, J.; Huang, W. E. Anal. Chem. 2013, 85, 10697− 10701. (11) Lau, A. Y.; Lee, L. P.; Chan, J. W. Lab Chip 2008, 8, 1116− 1120. (12) Jarvis, R. M.; Goodacre, R. Chem. Soc. Rev. 2008, 37, 931−936. (13) Lin, H. Y.; Huang, C. H.; Hsieh, W. H.; Liu, L. H.; Lin, Y. C.; Chu, C. C.; Wang, S. T.; Kuo, I. T.; Chau, L. K.; Yang, C. Y. Small 2014, 10, 4700−4710. (14) Nolan, J. P.; Duggan, E.; Liu, E.; Condello, D.; Dave, I.; Stoner, S. A. Methods 2012, 57, 272−279. (15) Watson, D. A.; Brown, L. O.; Gaskill, D. F.; Naivar, M.; Graves, S. W.; Doorn, S. K.; Nolan, J. P. Cytometry, Part A 2008, 73, 119−128. (16) Xie, C.; Chen, D.; Li, Y. Q. Opt. Lett. 2005, 30, 1800−1802. (17) Huang, W. E.; Ward, A. D.; Whiteley, A. S. Environ. Microbiol. Rep. 2009, 1, 44−49. (18) Krafft, C.; Dochow, S.; Popp, J. Combination of optical micromanipulation with Raman spectroscopy for cell sorting. European Conference on Lasers and Electro-Optics and International Quantum Electronics Conference (CLEO/EUROPE-IQEC), Munich, Germany, May 12−16, 2013; IEEE: Piscataway, NJ, 2013. (19) Pilát, Z.; Ježek, J.; Kaňka, J.; Zemánek, P. SPIE BiOS 2014, 89471M−89471M-9. (20) Berry, D.; Mader, E.; Woebken, D.; Wang, Y.; Zhu, D.; Lee, T. K.; Palatinszky, M.; Schintlmeister, A.; Schmid, M.; Shterzer, N.; Mizrahi, I.; Rauch, I.; Decker, T.; Bocklitzg, T.; Popp, J.; Gibson, C. M.; Fowler, P. W.; Huang, W. E. Proc. Natl. Acad. Sci. U.S.A. 2014, 112, 194−203. (21) Huang, N. T.; Zhang, H. L.; Chung, M. T.; Seo, J. H.; Kurabayashi, K. Lab Chip 2014, 14, 1230−1245. (22) Morgan, H.; Green, N. G. AC Electrokinetics: Colloids and Nanoparticles; Research Studies Press: Hertfordshire, U.K., 2003; pp 1−5. (23) Gray, D. S.; Tan, J. L.; Voldman, J.; Chen, C. S. Biosens. Bioelectron. 2004, 19, 771−780. (24) Fuhr, G.; Glasser, H.; Müller, T.; Schnelle, T. Biochim. Biophys. Acta, Gen. Subj. 1994, 1201, 353−360. (25) Archer, S.; Li, T. T.; Evans, A. T.; Britland, S. T.; Morgan, H. Biochem. Biophys. Res. Commun. 1999, 257, 687−698. (26) Jaeger, M. S.; Uhlig, K.; Schnelle, T.; Mueller, T. J. Phys. D: Appl. Phys. 2008, 41, 175502. 2289

DOI: 10.1021/ac503974e Anal. Chem. 2015, 87, 2282−2289