Anal. Chem. 2004, 76, 6247-6253
Continuous Flow Microfluidic Device for Rapid Erythrocyte Lysis Palaniappan Sethu,† Melis Anahtar,† Lyle L. Moldawer,‡ Ronald G. Tompkins,† and Mehmet Toner*,†
Surgical Services and Center for Engineering in Medicine, Massachusetts General Hospital, Harvard Medical School and Shriners Hospital for Children, Boston, Massachusetts 02114, and Department of Surgery, University of Florida College of Medicine, Gainesville, Florida 32610
Leukocyte isolation from whole blood to study inflammation requires the removal of contaminating erythrocytes. Leukocytes, however, are sensitive to prolonged exposure to hyper/hypoosmotic solutions, temperature changes, mechanical manipulation, and gradient centrifugation. Even though care is taken to minimize leukocyte activation and cell loss during erythrocyte lysis, it is often not possible to completely avoid it. Most procedures for removal of contaminating erythrocytes from leukocyte preparations are designed for bulk processing of blood, where the sample is manipulated for longer periods of time than necessary at the single-cell level. Ammonium chloride-mediated lysis is the most commonly used method to obtain enriched leukocyte populations but has been shown to cause some activation and selective loss of certain cell types. The leukocyte yield and subsequent activation status of residual leukocytes after NH4Clmediated lysis have been shown to depend on the time of exposure to the lysis buffer. We have developed a microfluidic lysis device that deals with erythrocyte removal at nearly the single-cell level. We can achieve complete lysis of erythrocytes and ∼100% recovery of leukocytes where the cells are exposed to an isotonic lysis buffer for less than 40 s, after which the leukocytes are immediately returned to physiological conditions. Theoretically, this process can be made massively parallel to process several milliliterss of whole blood to obtain a pure leukocyte population in less than 15 min. Blood is a readily available resource rich in information about the inflammatory and immune state of the host and can be readily sampled. Rapid isolation of pure populations and subsets of leukocytes from whole blood is essential for immunological evaluation. However, the activation status of peripheral blood leukocytes is very sensitive to the isolation process, and hyperor hypoosmotic solutions, temperature changes, mechanical manipulation, and gradient centrifugation can alter the cell phenotype.1 The biggest challenge is to obtain an enriched * Corresponding author. E-mail:
[email protected]. Phone: (617) 371 4883. Fax: (617) 523-1684. † Harvard Medical School and Shriners Hospital for Children. ‡ University of Florida College of Medicine. (1) Pelegri, C.; Rodriguez-Palmero, M.; Morante, M. P.; Comas, J.; Castell, M. Franch. A. J. Immunol. Methods 1995, 187 (2), 265-71. 10.1021/ac049429p CCC: $27.50 Published on Web 09/30/2004
© 2004 American Chemical Society
population of leukocytes without the process activating or altering the cellular phenotype. This process generally requires the rapid depletion of erythrocytes and subsequent isolation of enriched leukocyte populations. There are two commonly used techniques to isolate enriched leukocyte populations from whole blood that are often used in tandem: (a) erythrocyte lysis to obtain a enriched leukocyte population using different osmotic lytic agents such as deionized water, NaCl buffer, NH4Cl-KHCO3 buffer,2 and commercially available lysing solutions such as Erythrolyse, FACSlyse, Zap-o-globin, Becton Dickenson FACS lysing solution, and Coulter-Q-Prep lysing solution3 or (b) density centrifugation techniques using Ficoll, Percoll, sucrose, or dextran, alone or in conjunction with antibody-based selection systems using red cell rosetting (RosetteSep) for lymphocyte and granulocyte isolation.4 In most cases, some combination of the two techniques is often required to obtain highly enriched leukocyte populations essentially devoid of erythrocyte contamination. NH4Cl-based lysis buffers are widely used and are superior to other lysis solutions and approaches; however, there have been few studies that have examined the actual process. Theoretically, the calculated lysis time for an erythrocyte exposed to an excess NH4Cl-based lysis solution is ∼0.1 s, but experimentally, the observed lysis time for a single erythrocyte is ∼20 s.2 However, in the macroscale environment, the lysis is diffusion-limited and depends on the concentration of erythrocytes in the lysis solution. During a typical lysis procedure, 1 mL of whole blood is incubated with 15 mL of NH4Cl lysis buffer for 5 min, after which the leukocytes are recovered by centrifugation for another 5 min. It has been shown that incubation with the lysis solution for greater than 10 min resulted in damage to granulocytess5,6 and incubation times of 15-20 min were fatal to all leukocytes.7 There is also evidence to suggest that reduced lysis time can have a favorable effect on cell yield and viability.7 Microfabrication technology can be used to construct devices where sample processing is addressed at the single-cell level. There have been several attempts to design devices to perform (2) Maren, T.; Wiley, C. W. Mol. Pharmacol. 1970, 6 (4), 430-40. (3) Vuorte, J.; Jansson, S. E.; Repo, H. Cytometry 2001, 43 (4), 290-6. (4) Macey, M. G.; McCarthy, D. A.; Vordermeier, S.; Newland, A. C.; Brown, D. A. J. Immunol. Methods 1995, 181 (2), 211-9. (5) Tait, J. F.; Smith, C.; Wood, B. L. Blood Cells, Mol., Dis. 1999, 25 (5), 271-8. (6) van Oss, C.; Bronson, P. M.; Dinolfo, E. A.; Chadha, K. C. Immunol. Commun. 1981, 10 (6), 549-55. (7) Kouoh, F.; Levert, H.; Gressier, B.; Luyckx, M.; Brunet, C.; Dine, T.; Ballester, L.; Cazin, M.; Cazin, J. C. APMIS 2000, 108 (6), 417-21.
Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 6247
Figure 1. Microfluidic lysis device: device design and construction to perform rapid lysis of whole blood to obtain pure leukocyte populations. Table 1. Composition of the NH4Cl Lysis Buffer
microfluidic lysis of single or few cells in microchannels for extraction of proteins and nucleic acids8-10 and for lysis of single erythrocytes using a minielectrophoresis apparatus,11 but to date there have been no reports on development of systems for rapid continuous lysis of erythrocytes. In the present report, we describe a microfluidic lysis device capable of addressing blood cells individually, fabricated using simple soft lithography techniques. We can achieve complete lysis of erythrocytes and nearly 100% recovery of leukocytes. Whole blood enters the device, and the cells are focused into a narrow stream. Every cell experiences contact with the lysis buffer for the minimum required time for complete lysis of erythrocytes following which the solution is immediately returned to physiological conditions. MATERIALS AND METHODS Blood Samples and Reagents. Approval for the collection of blood from rodents and healthy subjects was obtained from the Institutional Animal Care and Use Committee and Institutional Review Board, respectively, of Massachusetts General Hospital. Blood was drawn from Lewis outbred rats (Charles River Breeding Laboratory, Wilmington, MA) using a 20-gauge needle (Becton Dickinson, Franklin Lakes, NJ) and collected in a 2-mL Vacutainer tubes with heparin anticoagulant (Becton Dickinson). Human blood was drawn from healthy volunteers using a syringe with an 18-gauge needle, which contained a sodium citrate anticoagulant. The blood in both cases was immediately stored at 4 °C until the experiments were performed. NH4Cl lysis buffer was made using NH4Cl, KHCO3, and tetrasodium EDTA dihydrate (8) Schilling, E.; Kamholz, A. E.; Yager, P. Anal. Chem. 2002, 74 (8), 1798804. (9) He, Y.; Zhang, Y. H.; Yeung, E. S. J. Chromatogr., A 2001, 924 (1-2), 27184. (10) McClain, M. A.; Culbertson, C. T.; Jacolson, S. C.; Allbritton, N. L.; Sims, C. E.; Ramsey, J. M. Anal. Chem. 2003, 75 (21), 5646-55. (11) Chaiyasut, C.; Tsuda, T.; Khansuwan, U.; Siriwan, O.-c. Chromatography 2001, 23(1), 1-6.
6248 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004
reagent
final concn
stock
in 500 mL of H2O
NH4Cl EDTA KHCO3 Polymixin B H2O (endotoxin free)
8.29 M 0.037 M 500 mg 10 IU/mL n/a
n/aa n/a n/a 10000 IU/mL n/a
4.146 g 18.5 mg 500 mg 0.5 mL n/a
a
n/a, not applicable.
((ED4SS) Sigma-Aldrich Corp., St. Louis, MO), in ratios as shown in Table 1. The buffer is mixed well, 1.5-µm filter purified, and stored at 4 °C. PBS solution is used to dilute the lysing solution and return the cells to normal physiological conditions. Zap-oglobin (Beckman Coulter, Franklin Lakes, NJ), a commercially available lysing solution, was also used for lysis. Bulk Characterization. Bulk characterization of erythrocyte lysis in excess buffer conditions where diffusion is not rate limiting was performed using a standard hemocytometer. The hemocytometer was placed under a microscope, and then, 2 µL of whole blood was injected into the port using a pipet, following which 20 µL of NH4Cl lysis buffer was introduced. Erythrocyte lysis was recorded using a video recording system (Panasonic PV-VS4821), and the time taken for complete lysis of erythrocytes surrounded by lysis buffer was determined. Device Design and Fabrication. The device design is shown in Figure 1. There are three inlets; whole blood is introduced through the first inlet, and lysis buffer, which is introduced through the second inlet, branches out into two streams, one on either side of the stream of whole blood in the lysis channel. The arrangement allows for whole blood to be focused into a narrow stream flanked on both sides by lysis buffer. The width of the streams can be controlled by varying the flow rates of both blood and lysis buffer. Finally, the third inlet is used for restoration of
normal physiological conditions using phosphate-buffered saline (PBS) solution, which again branches out into two streams and flanks the main channel at the end of the lysis channel. The dimensions of the channels are 31 µm × 80 µm, and the devices were fabricated with varying channel lengths from 15 to 80 cm to provide sufficient residence time for the cells in the device at different flow rates. The device was fabricated using soft lithographic techniques. A silicon wafer is first treated with oxygen plasma in an asher (March Instruments, Concord, MA) and then spin coated with SU-8, a negative photoresist (SU-8 50, MicroChem, Newton, MA). Standard photolithography using a transparency mask (CAD ART Services Inc., Poway, CA) was generated using AutoCAD layout software (Autodesk, Inc., San Rafael, CA) and used to create negative replicas of the designed channel structures (Figure 1). A silicone elastomer, poly(dimethylsiloxane) (PDMS; Dow Corning, Midland, MI) was then mixed 10: 1 with a cross-linking agent, poured on top of the silicon wafer, and cured at 60 °C for 12 h in a Petri dish. The cured elastomer with the replicated channels was released, and access holes to the channels were punched using a 22-gauge syringe needle. The PDMS piece with the channels was then bonded irreversibly to a glass slide after treatment with oxygen plasma in the asher. Access tubing (Tygon, Small Parts Inc. Miami Lakes, Fl) with a slightly larger diameter than the holes was subsequently press fitted into the punched holes. Experimental Setup. The microscale device was mounted onto the stage of an inverted microscope (Nikon Eclipse TE 2000U) with a CCD camera and a video monitoring system (Panasonic PV-VS4821). The sample, lysis buffer, and PBS were introduced with three 1-mL syringes controlled by three separate syringe pumps (Pico pump, Harvard Apparatus, Holliston, MA). Whole blood prior to introduction was incubated with SYTO 13 dye (Molecular Probes Inc., Eugene, OR), a fluorescent nucleic acid stain, a marker for nucleated cells. A 1-mL syringe was filled with 0.1 mL of whole blood to deliver the cells to the device at a constant flow rate. The pumps were placed such that the syringes were vertical and there was no sedimentation of cells on the bottom of the syringe, but rather sedimentation occurred at the nozzle, which results in a high concentration of blood cells (blood cells make up 45% volume of blood, which results in ∼2 times original concentration) for the first 0.05 mL followed by an extremely low concentration of cells for the following 0.05 mL. All of the experiments were performed using the initial 0.05 mL of blood. The devices were designed to introduce blood cells in continuously flowing lysis buffer such that every cell was exposed to the uniform conditions in the channels and to minimize the time required for complete diffusion of lysis buffer across the stream of whole blood. This, however, was not possible because even small variations in flow rate due to the mechanically operated syringe pumps could result in cessation of the flow of blood cells into the lysis channel. In the syringe pump, the motion of the syringe is controlled by a rotating screw. When metering liquids at low flow rates, small variations in the rotation of the screw can result in large variations in the effective flow rate. To counter this, the cells were admixed at slightly higher flow rates, which resulted in the stream of blood being 2-3 cells instead of the intended stream length of 1 cell, and higher flow rates of blood resulted in streams of 5-10 cells.
Data Analysis. Experiments to determine the diffusion coefficient of lysis buffer across the blood stream were performed by initially using fluorescein dye in the lysis buffer to determine the diffusion coefficient of fluorescein in blood and then use that data to estimate the diffusion coefficients of sodium and chloride ions in blood. Fluorescence measurements, using the Metamorph software (Universal Imaging Corp., Downingtown, PA), were performed by analyzing line scan measurements across the width of the channel. The line scan measures the red, blue, and green emissions. This was done at different locations to determine the time required for complete diffusion of lysis buffer (fluorescein dye) across the stream of blood. Initially, the emission from the lysis buffer is high due to the presence of fluorescein in the lysis buffer while there is no emission from the blood stream. But as the lysis buffer diffuses, the emission across the channel changes slowly and becomes constant. To characterize erythrocyte lysis and leukocyte recovery for each of the experiments, erythrocytes in the channels were tracked using phase contrast imaging and the cells traveling through the channels were analyzed by taking snapshots (0.0001-s exposure) at different locations using the Metamorph software. It is assumed that complete lysis of erythrocytes occurs when greater than 99.5% of the cells have been lysed. The same software was used to count the number of cells by analyzing the image snapshots using the cell-counting feature. Cells stained with the SYTO 13 dye were excited using blue fluorescence, and the number of stained cells at the inlet and outlet were counted. The experiments were carried out using two independent samples of blood from rats and one sample from a healthy human subject. For each sample of blood, the experiments were carried out under different flow conditions. For each of the flow conditions, the experiment was repeated three times. For every snapshot taken, the data were analyzed at three different locations on the channel. Experiments were also carried out in a fashion similar to test lysis of rat blood with three different lysing reagents. All the results are expressed as means ( SEM, n ) 18 for rat samples and n ) 9 for human samples. The combined flow rate of blood and lysis buffer in all the experiments conducted was maintained at less than 3.5 µL/min, since higher flow rates did not allow for capture of clear images required for accurate counting of cells. RESULTS Bulk Characterization of NH4Cl-Mediated Lysis. Control experiments to carry out bulk characterization were performed on the samples of rat and human blood to determine the time required for lysis of single erythrocytes. The experiments were carried out under conditions of excess lysis buffer where diffusion was not rate limiting. The first experiment was carried out at 25 °C, and the observed time taken for complete lysis for rat blood was 25 s and for human blood was 27 s. The experiment was repeated at 4 °C, and the observed times were 28 s for rat blood and 29.5 s for human blood. For both the experiments, the sample size was five and the standard error of the mean was (3 s. From these results, a reasonable understanding of the expected time scale for lysis in microfluidic channels was determined. Diffusion Measurements. Laminar flow in microchannels due to low Reynold’s numbers ensures that mixing almost exclusively occurs due to diffusion. Total elimination of erythrocyte lysis requires complete diffusion of lysis buffer across the channel, Analytical Chemistry, Vol. 76, No. 21, November 1, 2004
6249
Figure 2. Diffusion of lysis buffer across microchannel: diffusion of lysis buffer mixed with fluorescein across the width of the channel determined by fluorescence microscopy. Also shown are the green, blue, and red emissions across the width of the channel at different locations. Experimental conditions: blood 0.2 µL/min and lysis buffer 1.5 µL/min.
which leads to the accumulation of NH4+ and Cl- ions inside all of the erythrocytes from the NH4Cl-KHCO3 buffer. Variation in the flow conditions leads to changes in the widths of the blood and lysis buffer streams. The blood stream is dense and has a high concentration of cells. Lysis buffer is initially in contact only with the cells at the interface. Diffusion leads to transport of the lysis buffer across the stream of blood cells, exposing the cells on the inside of the stream to the NH4Cl solution, which in turn leads to erythrocyte lysis. The time taken for complete diffusion of lysis buffer depends on the width of the stream of blood and the corresponding widths of the streams of lysis buffer. Diffusion of fluorescein across the channel through a stream of high concentration of whole blood was experimentally determined (Figure 2). The time taken for uniform diffusion of fluorescein across the width of the channel at a blood flow rate of 0.2 µL/ min and lysis buffer flow rate of 1.5 µL/min was 0.9 s. Based on 6250
Analytical Chemistry, Vol. 76, No. 21, November 1, 2004
the observed time taken, a diffusion coefficient (D) for diffusion of fluorescein across a stream of whole blood was observed to be 4.55 × 10-7 cm2/s. The diffusion coefficient of fluorescein in water calculated using the Hayduk and Laudie method was 5.8 × 10-6 cm2/ s. The difference in observed and calculated diffusion coefficients was due to the difference in viscosity between water and whole blood, which results in ∼1 × 10-1 cm2/s increase in the diffusion coefficient. The calculated diffusion coefficient of NH3+ and Clin water are 3.67 × 10-5 and 2.29 × 10-5 cm2/s. Due to the higher viscosity of whole blood, their respective diffusion coefficients will be 3.67 × 10-6 and 2.29 × 10-6 cm2/s. For widths of 22 and 33 µm, the time for complete diffusion were calculated to be 0.21 and 0.45 s, respectively. Diffusion is a rapid process, and even for large blood stream widths the time required is less than 1 s. Lysis in Microfluidic Channels. Once initial characterization of NH4Cl-mediated lysis times in the bulk system was established
Figure 3. Microfluidic lysis: snapshots illustrating lysis of erythrocytes in microfluidic channels. L denotes the distance traveled in the microchannels and % lysis denotes the percentage of lysed erythrocytes. The channel dimensions are 30 µm × 80 µm, and the flow rate is 1.8 µL/min
and the diffusion and flow conditions in the channels were determined, experiments were designed to characterize the system for lysis of erythrocytes in whole blood. The variables in these experiments were the concentration and type of blood used, the type of lysing solution used, and the flow rates at which the blood and lysing solution were admixed. The experiments were also repeated using a human blood sample. Other lysing solutions such as deionized water and Zap-o-globin were also used to compare their performance with the NH4Cl lysis buffer. All of the experiments were performed at room temperature (25 °C) to ensure uniformity. Figure 3 shows lysis of erythrocytes in microfluidic channels using the fabricated device. The effects of changing the flow rates of blood to achieve different blood/lysis buffer ratios on the time taken for lysis of erythrocytes were analyzed and compared with the observed bulk characterization results. The flow rate of lysis buffer was maintained constant, and the flow rate of blood was varied. Three experiments were conducted each using two samples of rat blood. The lysis buffer flow rate was maintained at 1.5 µL/min, and the range of blood flow rates was 0.2, 0.3, and 0.5 µL/min. The percentage lysis of erythrocytes as a function of distance traveled in the channel is plotted in Figure 4a. Based on the total flow rate and the length of the channel at which complete lysis was observed, the time taken for complete lysis of erythrocytes could be calculated. For a constant lysis buffer flow rate of 1.5 µL/min and blood flow rates of 0.2, 0.3, and 0.5 µL/min, the time taken for complete lysis was calculated to be 28.8, 30.3, and 38.2 s. It can be seen from the obtained results that, for a constant lysis buffer flow rate, the time required for lysis increases with an increase in the blood flow rate. The experiments were repeated with a human blood sample using the same flow conditions. Figure 4b shows the percentage of erythrocytes lysed as a function of distance traveled in the channel. Again, for a constant lysis buffer flow rate, the time required for complete lysis increased with an increase in the blood flow rate. The time required for complete lysis for the three flow conditions was calculated to be 30.8, 33.9, and 36.8 s. The increase in time required for lysis in these
Figure 4. Effect of variation of flow conditions on lysis of (a) rat blood and (b) human blood: lysis of erythrocytes plotted as a function of distance traveled across the channel. The flow rate of lysis buffer was maintained a constant, and the flow rate of blood was varied to observe the effect of different blood/lysis buffer ratios on the lysis. All the results are expressed as means (% of intact erythrocytes) ( SEM for n ) 12 for the rat sample and n ) 6 for the human sample.
experiments can be attributed to the fact that increases in the blood flow rate while maintaining a constant lysis buffer flow rate resulted in wider blood streams. The corresponding time required for complete diffusion of lysis buffer across the blood streams also increased. The time required for lysis is illustrated in Figure 5a, where the time taken for complete lysis of rat blood at each of the flow conditions and the corresponding width of the blood streams are shown. Experiments were also performed to determine whether the time required for complete erythrocyte lysis remained constant if the flow rates of blood and lysis buffer were increased while the ratio of the flow rates was maintained constant. Initially, the time required for complete erythrocyte lysis was determined when the blood and lysis buffer flow rates were fixed at 0.5 and 1.5 µL/ min, respectively, resulting in a blood to lysis buffer flow rate ratio of 0.33. Subsequent experiments were carried out at higher blood and lysis buffer flow rates while the ratio between them was maintained at 0.33. The time required for complete erythrocyte lysis for the three conditions was plotted along with the corresponding blood stream widths in Figure 5b. From the figure, it can be seen that the widths of the blood stream and the time for complete lysis were approximately the same for constant blood/ lysis buffer ratios. All the plots represented only the percentage of erythrocytes; leukocyte recovery was ∼100% for all of the experiments (Table 2).The time required for lysis varies between 28 and 40 s based on the flow conditions. Further lysis studies were also conducted with a sample of rat blood using other lysing agents and compared with the results obtained using NH4Cl buffer. The other lysing agents used were deionized water and Zap-o-globin, a commercially available lysing agent for rapid lysis of erythrocytes for performing flow cytometric Analytical Chemistry, Vol. 76, No. 21, November 1, 2004
6251
Figure 5. Time taken for complete lysis of erythrocytes at different flow conditions: (a) constant blood flow rate and variable lysis buffer flow rates and (b) constant ratio of lysis buffer to blood (3:1). The experiments were carried out using two samples of rat blood. The pictures above show the width of the blood stream for the corresponding flow condition. All the results are expressed as means (time for complete erythrocyte lysis) ( SEM for n ) 12.
Figure 6. Comparison of lysis using different lysing agents. Plots show the lysis of erythrocytes as a function of distance traveled across the channel. All the results are expressed as means (% of intact erythrocytes) ( SEM for n ) 12. Table 2. Rat and Human Blood Leukocyte Recovery after Complete Erythrocyte Lysis at Different Flow Conditionsa
sample rat
human
blood flow rate (µL/min)
lysis buffer flow rate (µL/min)
leukocyte recovery (%)
0.2 0.3 0.5 0.7 0.8 0.2 0.3 0.5
1.5 1.5 1.5 2.0 2.5 1.5 1.5 1.5
100.33 ( 3.8 101.32 ( 4.3 101.00 ( 1.1 100.63 ( 2.5 100.63 ( 2.5 101.00 ( 7.5 98.00 ( 10.1 99.12 ( 2.1
a All the results are expressed as means (% of leukocytes) ( SEM for n ) 12.
analyses. The results for erythrocyte depletion using the different lysing solutions are plotted in Figure 6. Deionized water is hypotonic and causes cellular swelling followed by immediate lysis of erythrocytes, with delayed lysis of leukocytes. Complete lysis of erythrocytes occurs in 15 s (compared to ∼30 s using NH4Cl 6252
Analytical Chemistry, Vol. 76, No. 21, November 1, 2004
buffer), and the leukocytes appeared to be intact at 15 s. However, at delayed intervals (>30 s), visual evidence demonstrated that there were few leukocytes exhibiting structural damage. The other lysing solution used was Zap-o-globin. The exact mechanism by which lysis occurs is unclear; however, it is thought to be due to destruction of the cytoskeleton of the cells. The leukocytes have a higher resistance to the cytoskeleton damage and are morphologically intact for ∼1-2 h after treatment. The lysing solution was diluted 1:100 in isotonic buffer as per manufacturer’s instructions and resulted in lysis times of ∼8 s. The lysis is extremely rapid; however, at 8 s there appears to be a 3-5% loss of leukocytes and more leukocytes are lost further downstream. Lysis using Zap-o-globin results in extremely fast (∼8 s) lysis of erythrocytes accompanied by a 5-10% loss of leukocytes. DISCUSSION When red blood cells are incubated in an NH4Cl buffer, the cells swell and lyse. The process of lysis begins as NH3 enters the cell by diffusion and is converted to NH4+ ions to maintain pH equilibrium. The need for counterions results in generation of HCO3- from CO2 inside the cell catalyzed by carbonic anhy-
drase (the rate-limiting step in this process)2 and is then exchanged for Cl- ions from outside through the band 3 anion translocator. The accumulation of NH4+ and Cl- ions in the cells leads to osmotic swelling and subsequent lysis. The shape of the erythrocytes is biconcave, but with the accumulation of NH4+ and Cl- within the cells, swelling causes the cells to become spherical, and once their volume exceeds 2.5 times the original volume, the cells begin to lyse. Theoretically, the calculated times for lysis of a single cell has been determined to be 0.1 s; however, experiments carried out with erythrocytes in excess lysis buffer conditions show that the time required for lysis is on the order of 20 s, which may be due to diffusion rates and resistance to membrane swelling.2 Macroscale lysis requires >5 min of incubation time in the lysis buffer to ensure 100% lysis of erythrocytes.1 Though the process of lysis itself only requires 20-30 s, the time required for complete diffusion of the lysis buffer (to ensure 100% lysis) is longer. There is evidence to suggest that reduced lysis times can have a significant effect on the yield of leukocytes and help minimize their activation.7 Microfluidics allows us to expose blood cells to the lysis solution at almost a single-cell level, minimizing the time taken for diffusion. Our system is designed to introduce cells into a channel in a narrow stream flanked by lysis buffer on both sides. The lysis buffer diffuses uniformly across the width of the channel, and the cells are exposed to the NH4Cl buffer. The total time required for complete lysis of erythrocytes with our devices varies between 28 and 40 s depending on the ratios of blood and lysis buffer, following which the cells are returned to normal isotonic conditions by immediate quenching with PBS solution. We have also found that leukocytes following microscale lysis are morphologically intact and the yield is ∼100%. Whole blood before introduction into the device is incubated with a nucleic acid dye SYTO 13, which stains only the cells that contain nucleic acids (leukocytes and reticulocytes); the mature erythrocytes, which lack nucleic acids, are not stained. Structural damage to the leukocytes will cause the nucleic acids to escape from the cells and adhere to the channel walls, but in our experiments, it can be seen from observation of the channels under the microscope that the leukocytes are structurally intact and that there is no membrane rupture. The time required for complete lysis depends on the time required for uniform diffusion, which in turn depends on the width of the blood stream and the corresponding widths of the lysis buffer streams. Using the estimated diffusion coefficients of NH3+ and Cl- in blood (3.67 × 10-6 and 2.29 × 10-6 cm2/s), the time taken for uniform diffusion across the channel was calculated for each of the flow conditions. We then carried out experiments using both rat and human blood to determine the time taken for complete lysis of erythrocytes at different lysis buffer and blood flow rates. The ratios of lysis buffer and blood were varied, which led to corresponding variations in the widths of the blood and lysis buffer streams. It was found for both human and rat blood that the time required for lysis, which was calculated based on the flow rates and location in the channels where complete lysis occurred (Figure 4), increases with increase in the blood/lysis buffer ratio. At higher blood/lysis buffer ratios, the blood stream is wider, resulting in increased time required for complete diffusion and a corresponding increase in lysis times.
Experiments were also carried out using other lysing solutions, deionized water, and a commercially available lysing agent called Zap-o-globin. Lysis with deionized water is entirely due to accumulation of water inside the cells resulting in swelling and lysis. Osmotic lysis affects all cell types including leukocytes; however, the resistance to osmotic shock is greater for leukocytes when compared to erythrocytes. Some clinical applications require rapid isolation of pure leukocyte populations from several milliliters of blood. To meet those requirements, this process can be scaled further. Theoretically, the process can be made massively parallel with multiple devices working in parallel and processing performed in longer and taller channels at higher flow rates. To determine the effect of an increase in flow rates of both the blood and lysis buffer, while maintaining the ratio of one to the other constant, experiments were performed by holding the ratio the flow rates of blood to lysis buffer at 0.33 and increasing both the flow rates, respectively. The results obtained show that the time required for lysis was approximately the same. These findings suggest that longer and taller channels can be constructed to operate at higher flow rates. Another parameter that can be changed is the width of the channel. Enlarging the width of the channel can also add to the increased flow rate for increased processing speed, but the time required for complete lysis will be slightly prolonged due to the increased diffusion time for the lysis buffer across the channel. With simple modifications of the device, several milliliters of whole blood can be processed in 10-15 min, while exposing the cells to lysis buffer for the minimum required time. In conclusion, we have developed a microfluidic device for the rapid lysis of erythrocytes where every cell is exposed individually to a lysis solution. We can achieve complete lysis of erythrocytes in ∼30 s due to significantly reduced diffusion times. After complete erythrocyte lysis, the leukocytes can be returned to physiological isotonic conditions. All cells are exposed to similar conditions and minimum time of exposure to the lysis buffer resulting in improved cell viability and yield. The output contains leukocytes and is erythrocyte-free (
