A 15-s Protein Separation Employing Hydrodynamic Force on a

Jun 28, 2003 - Department of Medicinal Chemistry, Faculty of Pharmaceutical Sciences, The University of Tokushima, Tokushima 770-8505, Japan and CREST...
82 downloads 12 Views 422KB Size
Download PDF
Anal. Chem. 2003, 75, 3799-3805

A 15-s Protein Separation Employing Hydrodynamic Force on a Microchip Mari Tabuchi,*,† Yasuhiro Kuramitsu,‡ Kazuyuki Nakamura,‡,§ and Yoshinobu Baba†,|

Department of Medicinal Chemistry, Faculty of Pharmaceutical Sciences, The University of Tokushima, Tokushima 770-8505, Japan and CREST, Japan Science and Technology Corporation, JST, Japan, Department of Biochemistry & Biomolecular Recognition and Central Laboratory for Biomedical Research and Education, Yamaguchi University School of Medicine, Ube 755-8505, Japan, and Single-molecule Bioanalysis Laboratory, National Institute of Advanced Industrial Science and Technology (AIST), Hayashi-cho 2217-14, Takamatsu 761-0395, Japan

We report here a novel pressurization technique for microchip electrophoresis that enables 15-s separation of protein mixtures extracted from biological samples. Although pressure-driven flow is usually parabolic flow, pressurization prior to electrophoresis separation produced a plug flow and achieved a dramatic migration time reduction without compromising resolution. Sample plugs were pushed forward by pressurization after loading the sample but before electrophoresis separation, in the absence of an electric potential. Higher pressures enabled higher speed separation; furthermore, the resolution could be easily controlled using an optimal pressure. In addition, the slow medium-pressurization technique enabled 2-D separation in only a single channel on a microchip. Utilizing this technique, 12 samples of complex protein mixture extracted from a human T lymphoblastic cell line, Jurkat cells, were separated within 15 s in a single run using a 12-microchannel array. In addition, target proteins from Jurkat cells were detected within this time. This novel pressurization technique on a microchip will offer enormous advantages for proteome analysis over commonly used 2-D electrophoresis.

which require only electric potentials. Geometric design,1-4 electrokinetic focusing,5-9 and electroosmotically induced hydraulic pumping10-12 are used to drive fluid flow. Recently there have been reports investigating pressure-driven flow13-16 produced by external pumps on µ-CE, but pressure-driven flow was not found to provide adequate separation14 because of parabolic flow,15 which seems to mix rather than separate.16 Nevertheless, we still believed that hydrodynamic force is required to manipulate the sample mass. We speculated that sample focusing would be achieved if the pressure were independent of the electric potential, with pressure driving the sample before separation by electrophoresis, where plug flow would be produced. Moreover, we speculated that such a pressure would shorten the effective length in the cross-type of microchannel (Figure 1) and achieve higher speed separation. Herein, we focused on hydrodynamic force in µ-CE and developed two pressurized sample-loading methods, the VP method and the PP method. We also applied this novel technique to HTS systems and 2-D separation for proteome analysis and were able to easily detect target proteins from Jurkat cells. To our knowledge, no such method has ever been proposed, and thus, this paper is the presentation of a proof of concept.

Microchip electrophoresis (µ-CE) has recently received a great deal of attention for proteome analysis in terms of short analysis times, small amounts of samples and reagents, and the capability of construction of high-throughput screening (HTS) systems. However, in practice, traditional two-dimensional electrophoresis (2-DE) or SDS-PAGE is primarily used, since the separation capability of µ-CE is sometimes unsatisfactory. First, to improve the resolution of µ-CE, we turned our attention to hydrodynamic force. Traditional CE uses both hydrodynamic and electrokinetic force for sample injection. On the other hand, µ-CE relies almost exclusively on electrokinetic injections,1-12

EXPERIMENTAL PROCEDURES Samples and Reagents. Bovine insulin (5.7 kDa), lysozyme (14.4 kDa), myoglobin (17.0 kDa), trypsin inhibitor (21.5 kDa), trypsin (23.0 kDa), carbonic anhydrase (31.0 kDa), ovalbumin

* Corresponding author: (e-mail) [email protected]. † University of Tokushima, CREST, JST. ‡ Department of Biochemistry & Biomolecular Recognition, Yamaguchi University School of Medicine. § Central Laboratory for Biomedical Research and Education, Yamaguchi University School of Medicine. | National Institute of Advanced Industrial Science and Technology. (1) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science 1993, 261, 895-897. (2) Kutter, J. P.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1997, 69, 5165-5171. 10.1021/ac030051p CCC: $25.00 Published on Web 06/28/2003

© 2003 American Chemical Society

(3) Paegel, B. M.; Hutt, L. D.; Simpson, P. C.; Mathies, R. A. Anal. Chem. 2000, 72, 3030-3037. (4) Griffiths, S. K.; Nilson, R. H. Anal. Chem. 2000, 72, 5473-5482. (5) von Heeren, F.; Verpoorte, E.; Manz, A.; Thormann, W. Anal. Chem. 1996, 68, 2044-2053. (6) Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1997, 69, 3212-3217. (7) Jacobson, S. C.; Ermakov, S. V.; Ramsey J. M. Anal. Chem. 1999, 71, 3273-3276. (8) Haab, B. B.; Mathies, R. A. Anal. Chem. 1999, 71, 5137-5145. (9) Ocvirk, G.; Munroe, M.; Tang, T.; Oleschuk, R.; Westra, K.; Harrison, D. J. Electrophoresis 2000, 21, 107-115. (10) Culbertson, C. T.; Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 2000, 72, 2285-2291. (11) McKnight, T. E.; Culbertson, C. T.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 2001, 73, 4045-4049. (12) Santiago, J. G. Anal. Chem. 2001, 73, 2353-2365. (13) Dutta, P.; Beskok, A. Anal. Chem. 2001, 73, 1979-1986. (14) Cao, P.; Moini, M. Electrophoresis 1998, 19, 2200-2206. (15) Ross, D.; Johnson, T. J.; Locascio, L. E. Anal. Chem. 2001, 73, 2509-2515. (16) Stroock, A. D.; et al. Science 2002, 295, 647-651.

Analytical Chemistry, Vol. 75, No. 15, August 1, 2003 3799

Figure 1. (A) (a) Diagram of the pressurization technique procedure. The first step is sample loading by voltage or pressure. (B) For the second step, the sample plug at the cross section was advanced by pressurization. (C) Electropherograms of several µ-CE conditions. (a) VLVH (loading, squeezing voltage: 100-300 V, 500 V); (b) VHVL (500 V, 100-300 V); (c) VLPL (loading voltage, pressure applied to separation channel: 100-300 V, 1.0-2.0 kPa); (d) VLPM (100-300 V, 2.5-5.0 kPa); (e) PMPH (pressure sample injection, pressure applied to the separation channel: 2.5-5.0, 5.5-8.5 kPa). . (D) Peak identifications were performed by adding an additional 1 mg/mL of each component to the amounts of (Ce). (a) 1, lysozyme (Mr 14.4 kDa); (b) 2, trypsin inhibitor (21.5 kDa); (c) 3, carbonic anhydrase (31 kDa); (d) 4, ovalbumin (45 kDa); (e) 5, serum albumin (66.2 kDa); (f) 6, phosphorylase B (97 kDa); (g) 7, β-galactosidase (116 kDa); and (h) 8, myosin (200 kDa) were added. (E) Effect of the pressurization technique. Samples were injected under low voltage (100-300 V), and the following pressure conditions were applied to the separation channel. (a) PL, PM, and PH for bovine insulin; (b) PL for bovine insulin, myoglobin, and both bovine insulin and myoglobin; (c) PL for bovine insulin, myosin, and PM for both bovine insulin and myosin. Peak identification: 1, bovine insulin (5.7 kDa); 2, myoglobin (17.0 kDa); and 3. myosin (200.0 kDa). Pressure conditions are shown in (C).

(45.0 kDa), serum albumin (66.2 kDa), bovine serum albumin (66.5 kDa), phosphorylase B (97.0 kDa), β-galactosidase (116.0 3800 Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

kDa), and myosin (200.0 kDa) (Bio-Rad Laboratories, Richmond, CA, and Sigma Chemical, St. Louis, MO) were used as standard

Figure 2. Simulated images of the sample mass before electrophoresis (top row) and after electrophoresis, 0.3 s later (bottom row), for (A) the standard method, (B) VPL method, (C) VPM method, and (D) PMPH method. Colors indicate the distribution of the density of the sample plug; red represents high density and blue represents low density. The cross-type of channel was used for the simulation. The following calculation parameters were used: loading channel (horizontal channel) L1 ) 500 µm; separation channel length (vertical channel length) L2 ) 200 µm; each channel width W ) 50 µm, loading voltage (V), pressure (P), separation and squeezing voltages (V ′, V ′′): (A) 18, P2 ) 0, 18, 12; (B) 18, P2 ) 1e-6, 18, 9; (C) 18, P2 ) 2e-6; 18, 9; (D) P1 ) 2e-6; P2 ) 1e-5, 18, 9.

proteins. Proteins were dissolved in 0.1% sodium dodecyl sulfate (SDS; Wako Pure Chemical Industries, Tokyo, Japan)/Milli-Q water (ICN Biomedicals, Aurora, OH), and the final concentrations were adjusted to between 7.5 µg/mL and 0.3 mg/mL. Protein mixture extracted from human Jurkat cells, which are T lymphoblastic cells, was used as a biological sample. Jurkat cells were cultured at 37 or 51 °C for 30 min in RPMI 1640 medium (Nissui, Tokyo, Japan) with 10% fetal bovine serum (Equitech-Bio, Kerrville, TX).17 The cultured Jurkat cells were lysed in 1 mM EDTA (pH 7.4). The cell lysate was centrifuged at 15000g for 30 min, and the supernatant was stored at -80 °C until use. The protein concentration of the supernatant was determined by Lowry’s method.18 The extracts were subjected to 2-DE and µ-CE. A borate buffer, 0.05 M, pH 9.3 (adjusted with sodium hydroxide), and 3% polydimethylacrylamide were used as the run buffer for µ-CE. A 1-µL volume of Sypro Orange (Bio-Rad Laboratories, excitation/ emission; 485/590 nm) or another fluorescent reagent (Agilent Technologies No. 5065-4430, 650/680 nm) reacted with 100 µL of each protein. Coomassie Brillant Blue (Nacalai tesque, Kyoto, Japan) was used for 2-DE staining. The µ-CE System. A µ-CE system (cosmo-i; SV1100 or SV1210, Hitachi Electronics Engineering, Hitachi, Japan), equipped with an LED emitting at 470 nm or a diode laser emitting at 635 nm, and a microchip (i-chip 3 (Figure 1A) or i-chip12 (Figure 3A); Hitachi Chemical, Hitachi, Japan), made of poly(methyl methacrylate) and having channels 100 µm wide and 30 µm deep made by injection molding, were used for the separations. A 2-10-µL volume of buffer or sample was loaded from each reservoir. Simultaneous detection of 12 channels of the i-chip12 was performed at 42 mm downstream from each cross section. (17) Fujimoto, M.; Nagasaka, Y.; Tanaka, T.; Nakamura, K. Electrophoresis 1998, 19, 2515-2520. (18) Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall, R. J. J. Biol. Chem. 1951, 193, 265-275.

µ-CE Conditions. After the microchannels and the reservoirs (Figure 1A) were filled with buffer or sample, loading voltages of 100-500 V were applied to the loading channel. Then, squeezing voltages of 100-500 V at the cross section and a separation voltage of 890 V were applied to the separation channel. In one of the new methods, after filling the channels with buffer and sample, voltages of 100-300 V were applied to the loading channel. After pressurization was applied to the separation channel (Figure 1B), the sample was separated with 890 V, and in this paper we refer to this method as the VP method. In the other new method, the samples were injected via pressure on the loading channel, instead of by electrokinetic sample injection. Pressure was continuously applied to the separation channel in the absence of any electric potential, and samples were then separated at 890 V for SV1100 and 1300 V for SV1210. In this paper, we refer to this method as the PP method. Pressures of PL ) 1.0-2.0 kPa, PM ) 2.5-5.0 kPa, and PH ) 5.5-8.5 kPa were applied for 1 s for each method using a syringe. Pressures are applied at the inlet well while outlet well is open. Computer Simulation. Injection studies of the new methods were conducted using computer simulations based on mathematical theory.5,12-14 We used Coventor Ware software (Coventor, Inc.) for the simulation. The detailed conditions of the simulation are described in the caption to Figure 2. RESULTS AND DISCUSSION Proposal for Pressurization Immediately Prior to Electrophoresis. We speculated that pressurization applied to the separation channel before electrophoresis would shorten the effective channel length (Figure 1B) and achieve high-speed separation and that higher pressure would make separation even faster. We investigated these points. When we used our methods to separate a sample protein mixture (Figure 1C), it clearly demonstrated that pressurization Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

3801

Table 1. Pressure Effect on Resolution conditions peak no.a methodb

VLPL VLPM methodc PMPH methodd

1/2

2/3

3/4

9.3 3.3 -e 2.5 -e

16.0 12.5 1.7 -

2.5 8.0 + 1.5 -

resolution 4/5 5/6 1.3 4.0 + 1.0 -

1.0 17.5 + 1.0 -

6/7

7/8

6.0 20.0 + 1.3 -

4.0 8.0 + 1.5 -

a Figure 1C. b Figure 1Cc. c Figure 1Cd. d Figure 1Ce. e Increase (+) or decrease (-) in resolution against (b).

is effective, since all protein peaks were observed when pressurization was applied (Figure 1Cc-e), whereas some peaks were missing in this time course when it was not (Figure 1Ca and b). In addition, Figure 1C clearly shows that higher pressurization produced higher speed separation. At low pressure, using electrokinetic sample loading and pressurization of the separation channel (the VP method), eight protein peaks appeared (Figure 1Cc), and more pressurization reduced the migration time and broadened the peak interval (Figure 1Cd). With the other new method, in which samples were loaded by pressure and pressurization of the separation channel (the PP method), high pressure applied to the separation channel resulted in high-speed separation within 15 s (Figure 1Ce). Peaks belonging to trypsin inhibitor, ovalbumin, phosphorylase B, and myosin were low due to the concentration of the preparation. Each peak was further identified by adding an additional excess (1 mg/mL) of each component to the amounts for Figure 1C (Figure 1D). The reproducibility of the calculated size of each peak was within 5.0%. Pressurization Technique Can Reduce Migration Time. The effect of the pressurization technique was further confirmed for separation of a single protein (Figure 1Ea) and a mixture of two proteins (Figure 1Eb and c) in detail. Figure 1Ea clearly demonstrates that the application of higher pressure to the separation channel reduces the migration time. The application of medium pressure in the separation of the two-protein mixture also resulted in reduced migration times compared to their lowpressure peaks (Figure 1Ec). On the other hand, the two lowpressure peaks are located at the same migration times as their corresponding low-pressure peaks (Figure 1Eb). This indicates that pressurization force relates to migration time. Based on these results, we concluded that pressure application to the separation channel before separation by electrophoresis reduced migration time and the degree of migration time reduction depended on the degree of pressurization. Free Resolution Improvement by the Pressurization Technique. Since we observed a broader interval between adjacent peaks (Figure 1Cd), we decided to calculate the resolution of each peak. Table 1 clearly shows that application of medium pressure to the separation channel improved resolution, since almost all peak resolutions under medium pressure (VLPM method) were better than those under low or high pressure. On the other hand, although the resolutions of almost all peaks were decreased when the PP method was used, all were greater than 1.0, which are acceptable values. We investigated the influence of the pressure on the height equivalent to a theoretical plate (HETP; H) using the equation H 3802 Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

) 1/N (m/plate), N ) 5.54/(tr/W0.5)2, where W0.5 is the peak width at half-height, tr is the retention time, and a Gaussian peak profile is assumed. The values are (0.17-2.2) × 10-6 m/plate for the VLPL method, (0.17-4.4) × 10-6 m/plate for the VLPM method, and (8.5-1.5) × 10-6 m/plate for the PMPH method. Although the HETP decreased as the pressurization increased, the values are acceptable. High separation voltages usually greatly reduce migration times, but with the high-voltage method, some loss of resolution due to thermal convection is unavoidable.19 Moreover, since the high-voltage method hastens the migration time only in parallel or it causes the interval between the peaks to narrow, no improvement in resolution can be expected in the standard method. Since the decreased resolution is due to short separation channels, many researchers have attempted to solve this problem by using a longer separation channel, resulting in longer analysis time.20 Our findings, however, indicate that the pressurization technique may facilitate migration time reduction without compromising resolution, even with a short separation channel. In addition, suitable pressure conditions, especially medium pressure, can produce high-resolution separation. In Figure 1Cd, we obtained broadly separated peaks in which the last peak was not accelerated, although the first peak was. This phenomenon can be explained as follows. First, the sample plug was compressed by pressurization, where hydrodynamic resistance exists because of the narrow exit structure of the microchannel. By increasing pressurization, the sample plug will advance forward and the first peak will be accelerated. However, the sample plug will be back after release of pressure, resulting in broadly separated peaks, where the last peak will not be accelerated. More pressurization, however, will compress the plug again, where all peaks will be accelerated with a narrow interval between the peaks. Therefore, we think that this phenomenon represents the unique hydrodynamics in the microchannel and thus depends on microchannel size or shape, buffer viscosity, and pressure force. Computer Simulation of the Pressurization Technique. We further confirmed these phenomena using computer simulation, since computer simulations have been a useful tool for analysis of sample transport in microfabricated devices.10-16,21-24 Good agreement between the simulated and experimental data in previous studies confirmed that mathematical models are accurate and can be used for analysis and prediction of experimental results.12,16,21-23 Figure 2 clearly demonstrates that the pressurization technique shortens effective channel length, since the sample plugs were situated downstream of the separation channel after applying pressurization (before electrophoresis, top row) and the situation depended on the pressure force. This is consistent with the diagrams illustrating our hypothesis (Figure 1B) and our experimental data (Figure 1C and E). This result (19) Yao, S.; Anex, D. S.; Caldwell, W. B.; Arnold, D. W.; Smith, K. B.; Schultz, P. G. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 5372-5377. (20) Palmer, J.; Burgi, D. S.; Munro, N. J.; Landers, J. P. Anal. Chem. 2001, 73, 725-731. (21) Patankar, N. A.; Hu, H. H. Anal. Chem. 1998, 70, 1870-1881. (22) Bianchi, F.; Ferrigno, R.; Girault, H. H. Anal. Chem. 2000, 72, 1987-1993. (23) Ermakov, S. V.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1998, 70, 4494-4504. (24) Ermakov, S. V.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 2000, 72, 3512-3517.

from the simulation indicates that pressure manipulations can deliver the sample plug to the point we desired. Moreover, the pressurization technique can produce higher density sample plugs even downstream of the separation channel, since red in this simulation represents high mass density, and the pressurized method showed red even after electrophoresis separation (Figure 2B-D, bottom row). By contrast, in the same electrophoresis conditions, the standard method showed light green, which represents low density (Figure 2A, bottom row). In addition, medium pressurization created a broader sample band with high mass density (Figure 2C) that enabled broadening the peak interval without decreasing the intensity. The condensation by stacking effect may have occurred in the microchannel by the pressurization technique. In addition, pressure-driven flow usually produces a parabolic flow profile;15 our pressurization methods clearly produced plug flows, which also enabled the production of sharp peaks without the decreasing the resolution. Protein separation has the major shortcoming of a low concentration detection limit, which seems to be attributable to diffusion and adsorption to the wall. To overcome this limitation, a large-volume sample plug was introduced.25 Nevertheless, increasing sample plug size decreases the resolution of µ-CE in the standard method.2 Ross et al.15 imaged electroosmotic flow (EOF) in plastic microchannels and reported that the EOF also produces increases in sample dispersion. This is consistent with our simulation, since the standard method showed the low-density plug after electrophoresis separation (Figure 2A). To improve the low resolution due to diffusion, injection plug length can be conversely minimized by fabricating narrow channel dimensions for the injection valve26 and confining the sample volume within the cross section with an electric field.1,5-9 However, even if the sample plug is pinched at the cross section, the axial dispersion downstream of the separation channel is unavoidable with this method. In addition, since the injection plug length is minimized, the peak intensity is reduced. On the other hand, both our simulation and experimental results indicate that our novel pressurization technique easily pinched the sample plug even in the axial direction without compromising mass density. Since higher sensitivity is required to detect and characterize the low concentrations of target proteins, the method we developed is promising for proteome analysis. Comparison between Other Pressure-Driven Methods and Our Pressurization Methods. Cao and Moini14 used pressure assistance and pressure programming to analyze peptides and protein digests, but their method did not provide adequate separation for more complicated mixtures, such as protein digests. Ross et al.15 imaged pressure-driven flow in plastic microchannels and showed that pressure-driven flow produced a parabolic flow profile. We also recognized unplugged flow in our simulation when the electric potential and pressure were simultaneously applied. By contrast, our data in Figure 2 clearly demonstrate that the VP and PP methods are capable of producing plug flow, since our pressure was applied before electrophoresis separation, not simultaneously, thereby producing sharp peaks. However, when the pressure force was extremely low, line flow was recognized (25) Zhu, L.; Lee, H. K. Anal. Chem. 2001, 73, 3065-3072. (26) Zhang, C.-X.; Manz, A. Anal. Chem. 2001, 73, 2656-2662.

in our simulation. Optimized pressure force is required for pressurization technique. Pressurization Effect in HTS Systems. We investigated this pressurization effect in HTS systems using a 12-microchannel arrayed microchip (Figure 3A). Figure 3B clearly shows successful parallel separation of 12 channels, since all channels show good separation for the same standard protein mixture. We also confirmed this pressurization effect for a real biological protein mixture, the intracellular soluble proteins extracted from Jurkat cells, which are human T lymphoblastic cells. The electropherograms of channels 1-5 in Figure 3C show the sample from Jurkat cells, cultured at 37 °C; channels 6-10 show samples heat-treated at 51 °C for 30 min. Channels 11 and 12 show the standard proteins (protein ladder). Apparently, the simultaneous separation of 12 samples is successful because the reproducibility of each 5 channels is good (less than 5%), especially for the target proteins that are marked by solid reversed triangles. Cellular responses to stress signals, especially heat shock signals, have been extensively studied, and studies on the expression of heat shock proteins and their functional regulation are topics of whole research fields of cell biology.27 Jurkat cells possess target proteins, which increase after heat treatment at 51 °C for 30 min (Figure 3F). It has been reported that stathmin is a ubiquitous protein having multiple phosphorylation sites that are targets for protein kinases activated by extracellular signals.28 In 2-DE shown in Figure 3F, spot I, thymosin-β4 (Mr 7 kDa), and spot II′, phosphorylated stathmin (Mr 18 kDa), increased significantly after heat treatment. The increase of these peaks was also clearly detected in a 12-microchannel array employing our pressurization technique (Figure 3C and Db). In addition, we found certain heat-treated peaks were decreased in our µ-CE method (III (37 kDa) and IV (92 kDa)), which could not be detected as clear spots by 2-DE analysis (Figure 3F). Moreover, the peaks at 92 kDa were completely separated by medium pressurization in the VP method (Figure 3Ea and b). The calculated molecular size and concentration of each peak are summarized in Table 2. These target proteins were detected within only 15 s by the new method, while several hours were required for conventional 2-DE analysis. Yao et al.19 separated a six-protein mixture (9-116 kDa) within 35 s using a 4.5-cm separation channel for µ-CE, but the resolution was slightly diminished. Bousse et al.29 designed a protein sizing assay for 10 samples within 30 min for practical use, but buffer and sample loading take 1 min, and separation takes 45 s for a sample. By contrast, our system analyzed 12 samples of a complex mixture of biological proteins (7-200 kDa) in a single run within 15 s and buffer and sample loading are achieved easily in ∼1 s. Therefore, our results indicate that HTS systems for proteome analysis are possible using our newly developed pressurization technique. Pressurization Effect in 2-D Separation. The peak at 18 kDa, which is composed of stathmin (pI 5.9; II) and phosphorylated stathmin (pI 5.6; II′), was completely separated by the slow-pressurization technique in the PP method (5.0 kPa/3 s; Figure 3Ec). This result suggests that it is possible to perform (27) Beretta, L.; Dubois, M.-F.; Sobel, A.; Bensaude, O. Eur. J. Biochem. 1995, 227, 388-395. (28) Sobel, A. Trends Biochem. Sci. 1991, 16, 301-305. (29) Bousse, L.; Mouradian, S.; Minalla, A.; Yee, H.; Williams, K.; Dubrow, R. Anal. Chem. 2001, 73, 1207-1212.

Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

3803

Figure 3. (A) Illustration of the 12-microchannel array (i-chip12, Hitachi Chemical). (B) Standard protein separation by the PP method on a 12-microchannel array using the µ-CE system (cosmo-i SV1210, Hitachi Electronics Engineering) at 1300 V. Lysozyme (14.4 kDa) and myosin (200.0 kDa) were used as a lower or upper marker. Peak numbers were same as the proteins in Figure 1. (C) Proteins extracted from Jurkat cells were separated; channels 1-5 are samples from the cells cultured at 37 °C, channels 6-10 are samples from the cells, heat-treated at 51 °C for 30 min, and channels 11 and 12 show the ladder proteins. (D) Comparison of µ-CE analysis between culture conditions of (a) 37 and (b) 51 °C, which are the same as in (C). (E) The resolution improvement by medium pressurization for the specified low-resolution peak of 92 kDa in (D); (a) 37 and (b) 51 °C. (c) 2-D separation on a microchip using the slow medium-pressurization technique (5.0 kPa/3 s). Acidic condition of the sample (pH 5.0) was introduced into the alkali condition buffer (pH 9.7). (F) The 2-DE patterns of (a) 37 and (b) 51 °C. 3804 Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

Table 2. Peak Identification (ID) and Concentrations migration mark in conc (%)a peak time size Figure no. (s) (kDa) 37 °C 51 °C 3D,F ID 1 2

0.6 1.5

7.0 18.0

5.5 3.2

20.5 12.7

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

2.6 3.4 3.8 4.9 5.2 6.2 6.9 7.5 7.9 8.5 9.2 9.9 10.5 11.0 12.1 13.2 15.0

20.6 31.1 34.5 37.0 43.2 48.5 52.2 59.6 64.4 73.2 84.9 92.1 115.8 127.0

1.9 8.3 3.8 7.0 25.6 1.7 3.2 2.4 3.0 5.5 4.1 19.6

1.4 7.3 8.3 0.6 25.4 1.8 2.8

0 0

3.9 1.2

a

200.0

3.2 7.9 0.3 2.4

I II II′

thymosin-β4 stathmin phosphorylated

III

IV

/

upper marker

Percent of concentration.

2-D separation using the pressurization technique, using the first dimension for size separation and the second dimension for electric focusing separation, in only a single microchannel on a microchip. When an acidic sample was introduced into the alkaline conditions of the buffer, a pH gradient of pH was produced by the slow medium pressurization, in which we could separate proteins having different pIs. Since medium pressurization is capable of producing broader intervals between specific adjacent peaks, as mentioned above, many proteins with different pIs could be analyzed in this interval. In this way, we achieved 2-D separation in only a single separation channel on a microchip. To our knowledge, no such 2-D separation in a single separation channel on a microchip utilizing a pressurization technique has ever been proposed. In addition, utilizing the 12-microchannel array, more powerful 2-D separation could be achieved on the microchip. Finally, the µ-CE method employing the pressurization technique we developed seems to be more effective than conventional

2-DE, which has been the most useful technique for proteome analysis, in terms of rapid detection of target proteins. These systems represent an industrial revolution for proteome research. CONCLUSIONS In summary, we have demonstrated a novel pressurization technique combined with a microchannel array that offers enormous advantages in terms of HTS systems for proteome analyses. This is the first time, to our knowledge, that pressurization was effectively utilized in a microchannel just prior to electrophoresis separation. This paper is the presentation of a proof of concept. In conclusion, hydrodynamic force seems to be more effective than electric field supply alone as a means of achieving high-speed, high-resolution, and high-intensity protein separation and constructing 2-D and HTS systems by employing pressure prior to electrophoresis separation. Utilizing this technique, we separated target proteins from Jurkat cells within 15 s. We propose that our novel technique enables proteins extracted from cells to be analyzed, and especially small amounts of marker proteins, those which are idiotypic for a specific physiologic or pathologic state of cells or tissues, can be quickly identified using this technique instead of conventional 2-DE. ACKNOWLEDGMENT The present work was partially supported by a Grant of Core Research for Evolutional Science and Technology (CREST) from the Japan Science and Technology Corporation (JST), a Grant from the New Energy and Industrial Technology Development Organization (NEDO) of the Ministry of Economy, Trade and Industry, Japan, a Grant in-Aid for Scientific Research from the Ministry of Health, Labour, and Welfare, Japan, a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Technology, Japan, and Single-molecule Bioanalysis Laboratory, National Institute of Advanced Industrial Science and Technology (AIST), Japan.

Received for review February 4, 2003. Accepted April 18, 2003. AC030051P

Analytical Chemistry, Vol. 75, No. 15, August 1, 2003

3805