and Two-Dimensional Miniaturized Electrophoresis of Proteins with

2-D gel electrophoresis sorts proteins in two discrete steps. First, proteins are separated according to their isoelectric points (pI) via isoelectric...
0 downloads 0 Views 301KB Size
Download PDF
Anal. Chem. 2004, 76, 1359-1365

One- and Two-Dimensional Miniaturized Electrophoresis of Proteins with Native Fluorescence Detection Chanan Sluszny and Edward S. Yeung*

Ames Laboratory-USDOE and Department of Chemistry, Iowa State University, Ames, Iowa 50011

Miniaturized electrophoresis was successfully coupled with native fluorescence detection for direct analysis of proteins in one- and two-dimensional separations. The detection setup was based on direct observation of the UV-induced fluorescence of proteins using a CCD camera and a Hg (Xe) lamp for sample excitation. Protein mixtures were readily separated by size on a 1-cm segment of the one-dimensional gel in 8 min, and a detection limit of 0.04 ng per band was achieved. The dynamic range of the system was larger than 2 orders of magnitude. Miniaturized slab gel electrophoresis was performed on a special holder designed to couple isoelectric focusing with SDS-PAGE. Two-dimensional separation, including rehydration of IEF strip and fluorescence detection was completed in 2.5 h. Approximately 200 protein spots from Escherichia coli were detected on a 1 cm2 area. A detection limit of 0.1 µg of total protein was achieved. The operation should be amenable to total automation. Proteomics, namely the large-scale screening of proteins of a cell, organism, or biological fluid, was given its name in the mid 1990s but had actually originated over 20 years ago when the separation of proteins from total cell extracts was accomplished.1,2 While numerous methods have been devised to improve the speed and throughput for the analysis of nucleic acids,3,4 an integral aspect of proteomics, namely protein separation, at present still relies primarily on two-dimensional (2-D) slab gel electrophoresis, which was originally described by O’Farrell some 30 years ago.2,5 2-D gel electrophoresis sorts proteins in two discrete steps. First, proteins are separated according to their isoelectric points (pI) via isoelectric focusing (IEF). Proteins are then separated in the second dimension according to their molecular weights.2 For the latter, the anionic detergent sodium dodecyl sulfate (SDS), which is uniformly bound to proteins, is utilized. In both steps, polyacrylamide (PAA) gel is used. The technique is therefore termed 2-D polyacrylamide gel electrophoresis (2-D PAGE), or SDSPAGE for the simple case of one-dimensional separations. (1) O′Farrell, P. H. J. Biol. Chem. 1975, 250, 4007-4021. (2) Rabilloud, T. Proteome Research: Two-Dimensional Gel Electrophoresis and Identification Methods, Springer-Verlag: Berlin, 2000. (3) Kambara, H.; Takahashi, S. Nature 1993, 361, 565-566. (4) Ueno, K.; Yeung, E. S. Anal. Chem. 1994, 66, 1424-1431. (5) Pennington, S. R.; Dunn, M. J. Proteomics, From Protein Sequence to Function Springer-Verlag: New York, 2001. 10.1021/ac035336g CCC: $27.50 Published on Web 01/31/2004

© 2004 American Chemical Society

Gels are usually cast into 10-20-cm slabs. Minigels (6-8 cm), which are commercially available, enable easy handling and faster separations. Moreover, various miniaturized systems, referred to as microelectrophoresis techniques, have been developed.6-14 Electrophoresis in those cases can be run in capillaries for IEF or slab gels for 1-D and 2-D separations. Microgels are usually 2-3.5 cm in length.8 Microelectrophoresis offers the obvious advantage of faster operation in addition to much lower sample demands. Indeed, in the mid 1960s, Matioli et al.6 developed a microelectrophoresis technique for the analysis of hemoglobin in single erythrocytes, albeit with fairly poor resolution. Despite this, the practical advantages of microelectrophoresis were hardly recognized, and the method is rarely used. This is probably due to the requirement of higher level of delicacy in the implementation and the lack of suitable detection schemes, as compared to conventional slab-gel designs. For example, early experiments with capillary gels (2 cm × 0.25 mm) and 1.5-cm-square 2-D gels failed to produce adequate separation resolution.9 Subsequently, IEF in 1-2-cm gels have been improved by using ultrathin gradient gels10 with a detectability of 10-15 ng of protein. Detection of proteins on 1-D and 2-D gels is usually achieved by staining them with dyes or metals.2 Coomassie Brilliant Blue dyes are the most common protein staining materials.2 The detection limit of this method is 0.1-1 µg per spot. Silver staining has a limit of detection of ∼1 ng per spot. The latter, however, is dependent on water quality and requires freshly made solutions and rigidly controlled reactions.2 Moreover, the dynamic range of silver staining is low, and the measured response is nonlinear.2 After silver staining, DNA fragments separated in one dimension (14 mm) have been detected at the several nanogram range by thermal lens spectrometry.12 Fluorescence labeling is closely related to staining with dyes; fluorescent molecules bind to proteins, covalently or otherwise, and enable their detection.2 Ethidium bromide-labeled DNA fragments have been separated and detected in 5 mm × 1 mm tube gels.13 The dynamic range of (6) Matioli, G. T.; Niewisch, H. B. Science 1965, 150, 1824-1826. (7) Rainer Maurer, H.; Dati, F. A. Anal. Biochem. 1972, 46, 19-32. (8) Neuhoff, V. Micromethods in Molecular Biology, Springer-Verlag: New York, 1973. (9) Ruchel, R. J. Chromatogr. 1977, 132, 451-468. (10) Kinzkofer, A.; Radola, B. J. Electrophoresis 1981, 2, 174-183. (11) Poehling, H. M.; Neuhoff, V. Electrophoresis 1980, 1, 90-102. (12) Zheng, J.; Odake, T.; Kitamori, T.; Sawada, T. Anal. Chem. 1999, 71, 50035008. (13) Heller, M. J.; Tullis, R. H. Electrophoresis 1992, 13, 512-520. (14) Quentin, C.-D.; Neuhoff, V. Intern. J. Neuroscience 1972, 4, 17-24.

Analytical Chemistry, Vol. 76, No. 5, March 1, 2004 1359

such fluorescence-based staining methods may cover 5 orders of magnitude. Sensitivity of these methods, however, is currently inferior when compared to silver staining2 due to unfavorable binding and a high background at low protein concentrations. New dyes and fluorescence scanners are now commercially available to facilitate detection based on fluorescence staining. In all cases, a subsequent destaining step is required to reduce the background. For the special case of catalyzed tetrazolium formation, lactate dehydrogenase isoforms have been separated and detected in 1 cm × 0.45 mm tube gels by IEF.14 Detection of proteins by radioisotope labeling offers the best sensitivity, but the long exposure times and safety concerns are drawbacks.2 Direct detection of protein spots from a thin polyacrylamide gel by mass spectrometry has been demonstrated at the 10-ng range.15 Most protein mass spectrometry, however, is performed after digesting the spots to produce peptide fragments for protein identification. In these cases, certain staining procedures cannot be employed. Detection could, however, be based on direct observation of the proteins rather than staining them. Several such techniques, which are based on the direct measurement of the UV absorbance or native fluorescence of proteins, have been suggested before.16-20 For example, Yamamoto et al.17 measured the 280-nm UV absorbance of proteins in PAA gels. The sensitivity was in the microgram range. Kazmin et al.19 reached a limit of detection (LOD) of 1 µg per protein band by using a chemical modification of tryptophan, which yields products that emit in the visible region. Hogan et al.16 obtained a similar LOD by using an indirect fluorometric method, and Koutny et al.18 used UV absorbance and native fluorescence with microgram sensitivities. More recently, Roegener et al. utilized a high-power UV laser as the excitation source for direct detection of proteins in PAA gels.20 The researchers used a frequency-tripled Ti/sapphire quasicontinuous laser operating at 280 nm to irradiate 1-cm2 segments of the 6 × 7 cm gel at 35 mW/cm2. Sequential images were acquired for the full slab gel, and a detection limit of 1-5 ng for various protein bands was obtained. The combination of a high excitation intensity and a sensitive charge-coupled device camera (CCD) (QE > 65%) was the main reason for the improved sensitivity in this case. Almost all proteins contain some aromatic amino acids, in particular tryptophan, that make them naturally fluorescent. Here, we demonstrate the successful coupling of microelectrophoresis with native fluorescence detection. 1-D and 2-D gel electrophoresis, including detection, are completed in 1 and 2.5 h, respectively. The detection limits achieved are superior to those of silver staining. In addition, the staining and destaining steps were eliminated altogether. EXPERIMENTAL SECTION Chemicals. SDS-PAGE. Acrylamide, bisacrylamide, 1.5 M Tris-HCl solution (pH 8.8), and 0.5 M Tris-HCl solution (pH (15) Ogorzalek Loo, R. R.; Stevenson, T. I.; Mitchell, C.; Loo, J. A.; Andrews, P. C. Anal. Chem. 1996, 68, 1910-1917. (16) Hogan, B. L.; Yeung, E. S. Appl. Spectrosc. 1989, 43, 349-350. (17) Yamamoto, H.; Nakatani, M.; Shinya, K.; Kim, B. H.; Kakuano, T. Anal. Biochem. 1990, 191, 58-64. (18) Koutny, L. B.; Yeung, E. S. Anal. Chem. 1993, 65, 183-187. (19) Kazmin, D.; Edwards, R. A.; Turner, R. J.; Larson, E.; Starkey, J. Anal. Biochem. 2002, 301, 91-96. (20) Roegener, J.; Lutter, P.; Reinhardt, R.; Bluggel, M.; Meyer, E. M.; Anselmetti, D. Anal. Chem. 2003, 75, 157-159.

1360

Analytical Chemistry, Vol. 76, No. 5, March 1, 2004

6.8) were purchased from Bio-Rad (Hercules, CA). SDS, TEMED, ammonium persulfate, and glycerol were obtained from Sigma (St. Louis, MO). The running buffer, purchased from Bio-Rad, contained 25 mM Tris, 192 mM glycine, and 0.1% (w/v) SDS. The sample buffer for 1-D electrophoresis contained 62.5 mM TrisHCl at a pH of 6.8, 25% glycerol, 2% SDS, and 0.01% bromophenol blue (Bio-Rad). β-Mercaptoethanol was obtained from Sigma. Coomassie G250 was purchased from Bio-Rad. Agarose IEF was purchased from Amersham Biosciences (Piscataway, NJ). IEF. Immobilized pH gradient (IPG) strips, 7 cm long, pH range of 3-10 nonlinear, were purchased from Amersham Biosciences. DTT, urea, CHAPS, IPG buffer, iodoacetamide, bromophenol blue, and dry IEF gels were purchased from Amersham Biosciences. Dry IEF gels were cut into narrow strips (1 × 22 mm) with a paper trimmer. Carrier ampholytes for the micro 2-D separations were Bio-Lyte 3/10 and Bio-Lyte 5/7 from Bio-Rad. Protein Standards. For the 1-D experiments, standard proteins of R-lactalbumin (MW 16 000), carbonic anhydrase (MW 29 000), bovine serum albumin (MW 60 000), conalbumin (MW 77 000), phosphorylase b (MW 97 000), and β-galactosidase (MW 116 000) were purchased from Sigma. Proteins were dissolved in water to a final concentration of 2 µg/µL and stored at -20 °C until use. For the 2-D experiments, sample proteins were from Escherichia coli (Geno Tech, St. Louis, MO and Bio-Rad). The protein sample was dissolved in 8 M urea, 2 mM DTT, and 2% CHAPS to a final concentration of 1-2 µg/µL and stored at -80 °C until use. Double-deionized water was used in all cases for sample preparations. Procedures. Gels. In all cases, slab-gels were made up of a 4-8% PAA stacking layer and a 12% PAA resolving gel. Stacking gels for the 1-D separations were composed of 8% T and 2.6% C at a pH of 6.8, and those for the 2-D experiments were 6% T, 3.3% C at a pH of 8.8. The resolving gel in both cases was 12% T and 3.3% C at a pH of 8.8. The stacking gel for the micro 2-D separations was 4% T, 3.3% C at a pH of 6.8, and the resolving gel was 12% T, 3.3% C at a pH of 8.8. Slab gels for the 1-D separations were cast using the MiniProtean II clamp assembly (Bio-Rad). The final size of the gels was adjusted to 3.5 × 8 cm by using a set of shortened casting glass plates made from the original (7 × 8 cm). Five holes of 0.5-mm diameter were drilled in the center of the upper casting plate for the preparation of injection wells and for sample delivery into the gel. The holes were 1.9 mm apart and were located 8 mm from the upper edge of the plate. First, the resolving gel was poured onto the casting plates. The gel was leveled 3 mm from the lower base of the drilled holes and polymerized for 30 min. Next, the holes were covered with Parafilm to prevent leakage of gel during preparation of the injection wells. Fused-silica capillaries (1.5-cm long, 360-µm o.d., and 250-µm i.d.) (Polymicro Technologies, Tempe, AZ) were inserted through the film into the holes. The stacking gel was then added. Following polymerization, rounded injection wells were formed. Gels for 2-D separations were made without injection wells, but were similar in dimensions to those for the 1-D separations. The dimension of the resolving gel for the micro 2-D separations was approximately 9 × 20 mm. Electrophoresis. For 1-D SDS-PAGE, protein mixtures of various concentrations were prepared, mixed with sample buffer

Figure 1. Experimental setup for micro 2-D electrophoresis. Dimensions of the holder were 5 × 6.5 cm. Volume of the buffer containers was 2 mL each. A resolving gel of 0.9 cm length was cast into a 0.75 mm depression of dimensions 2 × 1.5 cm in the center.

and 5% beta-mercaptoethanol, and heated to 100 °C for 5 min. Samples were then cooled and injected. Prior to injection, the gel was placed on a flat surface to enable easy sample delivery. A T-connection split-flow apparatus was used to inject 0.1-µL samples into each well by using a 100-µL syringe. Two capillary tubes of 75 µm and 250 µm i.d. were used for sample delivery and for discharge of excess solution, respectively. This design resulted in a factor of ∼100:1 for the volumes flowing through the large- and small-diameter capillaries, respectively. This is in agreement with theoretical values.21 Next, the injection holes were carefully sealed with agarose (0.5% w/v) to prevent sample leakage. Gels were run vertically in a conventional Mini-Protean electrophoresis cell. A constant voltage of 200 V was applied for 8 min. Following electrophoresis, the 1-cm-long separation area, defined by the injection wells and the bromophenol marker, was cut, and the gels were fixed in 10% acetic acid, 30% methanol, and 60% water for 30 min and rinsed three times in water for 5 min. Gels were then placed on a fusedsilica plate (Suprasil2, Heraeus, Germany) for fluorescence analysis. Five independent SDS-PAGE separations were run for each sample. Isoelectric focusing of the 7-cm IPG strips was performed with an Ettan IPGphor system (Amersham Biosciences) according to the manufacturer’s instructions. The rehydration solution contained 8 M urea, 2% CHAPS, 0.5% IPG buffer, 0.3% (w/v) DTT, 0.002% bromophenol blue, and the protein sample. IPG strips were rehydrated inside the holder for 12 h at a constant temperature of 20 °C, then high voltage was applied until the system reached a total of 30 000 V. This step usually took 5-6 h for its completion. Subsequently, IPG strips were equilibrated for 15 min in solution containing 50 mM Tris-HCl (pH 8.8), 6 M urea, 30% (v/v) glycerol, 2% SDS, 1% DTT (w/v), and 0.002% bromophenol blue. To improve the results, a second equilibration step was applied. IPG strips were equilibrated for 15 min in the SDS solution, in which DTT was replaced with 2.5% (w/v) iodoacetamide. The second dimension was run on gels similar to those of the 1-D experiments. Electrophoresis was run at a constant 200 V and (21) Roy, D. N. Applied Fluid Mechanics Wiley and Sons: New York, 1988.

was completed in 16 min. Gels were then rinsed and analyzed. 2-D separations were also performed on standard minigels (6 cm length). For these, electrophoresis was run for 60 min at a constant 200 V. Gels were stained with coomassie dye, rinsed, and imaged using a scanner (Image Scanner, Amersham Biosciences). The experimental setup for the 2-D microelectrophoresis experiments is shown in Figure 1. Isoelectric focusing and SDSPAGE were performed directly on the same plate. The holder was ceramic (Mica Ceramics), which served as an excellent electrical insulator and, in addition, allowed efficient cooling during electrophoresis. The holder was placed on a thermoelectric cooling plate (model TCP-30, Advanced Thermoelectric, NH) to maintain a temperature of 17-20 °C throughout the experiments. The slab gel layer was formed inside a 0.75-mm depression (see Figure 1). Prior to gel casting, the depression was covered with a glass slide and sealed with 1% agarose gel. The resolving and stacking gels were ∼9 mm and 3 mm in length, respectively. A 1-mm gap was maintained between the upper stacking gel layer and the IEF strip. This gap was necessary in order to prevent sample leakage into the slab gel during isoelectric focusing. The gap was filled with 0.5% agarose prior to running the second dimension. The procedure for running the micro 2-D separations was as follows: First, the IEF strip was rehydrated for 1 h in a solution containing 8 M urea, 2% CHAPS, 3% carrier ampholytes (composed of a 60% Bio-Lyte 5/7 and 40% Bio-Lyte 3/10), 2.5% mercaptoethanol, 0.005% bromophenol blue, and the protein sample. The dry IEF strip absorbed ∼14 µL of the hydration solution. This was consistent throughout the experiments. The hydrated strip was then positioned on the holder (see Figure 1) for isoelectric focusing. Running conditions for IEF were 10 min at 60 V, 5 min at 120 V, 5 min at 200 V, and 40 min at 300 V. The IEF gel was equilibrated in situ with SDS-containing solution. A 100-µL aliquot was sufficient for complete coverage of the strip with the equilibration solution. Next, the equilibration solution was carefully removed, and the 1 mm gap that had isolated the slab gel from the IEF strip was filled with a 0.5% agarose solution. Running buffer was then added, and the system was run at 200 V for 3 min, during which the bromophenol marker had reached the Analytical Chemistry, Vol. 76, No. 5, March 1, 2004

1361

Figure 2. Experimental setup for native fluorescence detection of proteins in miniaturized gels.

anodic buffer reservoir. Subsequently, the slab gel was removed from the holder, rinsed, and imaged. Detection Setup. The detection setup is presented in Figure 2. A 500 W Hg (Xe) lamp (Oriel, Stamford, CT) was used as the excitation source. This light source emits intense UV radiation. In particular, intense bands are located at 270-320 nm. A twoelement condenser (4.8 cm diameter) collimated the light from the arc lamp to the gel. The optical components were protected from damaging IR light by a water filter positioned at the output of the condenser. This liquid filter had an external chamber for cooling water circulation. A set of two matched filters (a color glass and a band-pass filter) and a UV mirror were placed in the optical path such that relatively broadband (30 nm) UV radiation reached the gel sample. The color glass filter (UG-5, Schott Glass, Duryea, PA) transmitted 80% of the light between 250 and 380 nm (85% transmittance at 280 nm). A custom-designed band-pass filter (285AESP) was purchased from Omega Filters (Brattleboro, VT). This interference filter had a center wavelength of 280 nm with 20% transmittance, and the fwhm was 35 nm. The optical density for wavelengths longer than 290 nm was >4. The mirror (50.8 mm, Y4-2037-0, CVI laser, Albuquerque NM) was designed to reflect 99% of the UV light between 250 and 290 nm and only ∼1% outside of this specified spectral range. The average radiation power of the system was 1 mW/cm2. To reduce background effects, the setup was contained in a closed box. In addition, special care was taken to reduce scattered light within the enclosed box. The irradiated gel was imaged by a UV camera lens (Nikon; f.l. ) 105 mm, f ) 4.5) onto a back-illuminated, 16-bit CCD camera (TE/CCD-512-TKB, Princeton Instruments, Princeton, NJ). This CCD had an ∼40% quantum efficiency for UV light. A long-pass color glass filter (WG-305, Schott Glass) and a band-pass filter (330WB80, Omega Filters) were used to block the reflected UV radiation. The CCD exposures were set at 1-4 min. Image analysis was performed with the Pdquest software (BioRad). Quantification was based on the integrated fluorescence intensity of each protein spot. RESULTS AND DISCUSSION Experimental Development and Design. Rapid analysis of small samples is probably the main advantage of microelectrophoresis over the traditional macroscaled techniques. Due to the 1362

Analytical Chemistry, Vol. 76, No. 5, March 1, 2004

reduced dimensions of gels, however, microelectrophoresis requires a higher level of precision than that needed in the corresponding macroscaled approaches. Consider the initial step of running a 1-D gel, that is, sample introduction. Samples are typically delivered into relatively large injection wells (∼3 mm). These become significantly smaller (typically 0.7-1 mm width) for the micro design.8 Various approaches have been suggested for sample introduction into ultrathin slab gels. For example, Zheng et al.12 applied a specially designed ceramic comb to form rectangularly shaped wells, which enabled the use of a regular micropipets for sample delivery. Alternatively, Hietpas et al.22 used a capillary to inject samples into ultrathin gels. Samples were delivered by moving the injection capillary along the gel by a stepping motor. In this work, we used a simple method for sample delivery into 1-D gels based on sample introduction through the upper glass casting plate into the underlying wells. Sample introduction thus became straightforward, as the microinjection tip (a fusedsilica capillary) was aligned exactly into the sample well by the supporting glass opening. Protein detection becomes a critical issue in microelectrophoresis due to the smaller amounts of sample and the smaller areas probed. Furthermore, the small dimensions of microgels makes them more difficult to handle during sample loading and postrun staining. Finally, even though microgels offer the advantage of fast separation,8,9 analysis time is typically not reduced due to the staining and destaining steps undertaken. Thus, a direct detection method would be advantageous. Indeed, recently,20 laserinduced fluorescence was applied for slab-gel analysis. However, at present, extensive application of UV lasers for gel analysis is unrealistic because of their high cost. Rastering of a laser beam also takes time and requires precise movements. Intensity fluctuations in these lasers and gel inhomogeneity further limit detection performance. Here, we used a UV lamp as an alternative, cheaper excitation source for direct analysis of gel-separated proteins. Lamps are more convenient, more stable, and more available than UV lasers, but they do not produce the same high radiant power as the corresponding lasers. Detection was therefore based on exposing the gels to broadband UV-radiation and collection of the emitted fluorescence by using a sensitive CCD. Coupling to microgels becomes advantageous in this case, since the radiation can be focused onto a small area, thus increasing the power density. As shown below, 1-4 min of UV exposure was sufficient for complete visualization of all separated proteins in the 1-D and 2-D gels. One-Dimensional SDS-PAGE. We first examined onedimensional separations on microgels. Solutions containing various concentrations of proteins of diverse molecular weights were analyzed. The amounts loaded varied between 0.1 and 20 ng per protein. Electrophoresis was completed in 8 min, in which the bromophenol blue marker migrated ∼12 mm toward the anode. This is ∼7 times faster than running a 6-cm minigel. The relevant separation area, bounded by the injection wells and the marker band, was cut, rinsed, and analyzed. The imaged size was set to 11 mm square. (22) Hietpas, P. B.; Bullard, K. M.; Gutman, D. A.; Ewing, A. G. Anal. Chem. 1997, 69, 2292-2298.

Results of the 1-D SDS-PAGE separation of the six standard proteins are presented in Figure 3. In all cases, images were inverted; thus, proteins appear as dark bands on the relatively nonfluorescent gel background. Figure 3 emphasizes the superior detection capabilities of the microelectrophoresis technique, as compared to those of conventional gels and standard staining methods. The signal-to-noise (S/N) ratio of the 0.1-ng protein spots was ∼8; therefore, the absolute limit of detection (LOD) for this method is 0.04 ng, which is 2000-fold and 20-fold better than that of coomassie and silver staining, respectively.2 The corresponding protein concentrations were 6 × 10-8 M for the smallest protein (R-lactalbumin) and 8 × 10-9 M for the largest one (β-galactosidase). This is equivalent to ∼0.02 µg for conventional gels (assuming that 20 µL sample is loaded). It is worth noting the rounded shapes of the 116-kDa protein bands in Figure 3. These spots are related to the confinement of samples into the round injection wells. Because of the fast separation and the short length of the gels, lateral spreading effects due to diffusion23 are minimized. The faster migrating bands are gradually narrowed due to partial stacking. In Figure

3a and b, the inset at the right shows the original data displayed in gray scale for the highest concentration sample. Because of the limited dynamic range of the gray scale display, the lowest concentration samples would be masked by the substantial background fluorescence and scattering from the gel. On implementing background offset and scale expansion, the gel images on the left were obtained. Naturally, the highest-concentration samples in each gel then produced saturated images. The LOD is determined by fluctuations in the background levels around the bands (fluorescence plus scattering), not by the absolute levels of the background. In principle, very long exposures would allow S/N to increase with the square root of the exposure time, up to the limit of photobleaching. The electron well depth of the CCD is limited, but the CCD can be read often to avoid saturation, whereas the final accumulated counts are used for image analysis. Conversely, to extend the useful detection range to high concentrations of proteins, the exposure time should be reduced to maintain the integrity of the signal. In that case, background offset would not have to be applied. Therefore, by using a very long exposure time but also utilizing several intermediate data sets, a wide dynamic range can be achieved. Figure 4 presents the linear calibration plots for the 1-D SDSPAGE separations. The low-range sample, 0.1-2 ng, corresponds to Figure 3a, and the high-range sample, 2-20 ng, corresponds to Figure 3b. To maintain a maximum count of 70% saturation of the CCD detector, exposure times were set to 1 and 4 min for the higher and lower range, respectively. It was found that the fluorescence intensity of proteins in PAA gel for the 4-min exposure increased by only a factor of 3, as compared to the 1-min exposure. This is probably due to bleaching of proteins during UV irradiation. Figure 4 shows that the system exhibits a large and linear dynamic range, covering the picogram region to an upper limit of 20 ng. This is better than the corresponding dynamic range obtained by silver staining2 and coomassie dyes.24 Two-Dimensional Separations. Traditionally, 2-D electrophoresis involves a number of steps and the use of various devices to complete a single run. Clearly, the size of the gels and the use of post-electrophoresis staining methods complicate any attempt of coupling all the steps into a single device. Here, we present a simple and rapid procedure for coupling the major steps involved in 2-D separation into a single, miniaturized device. First, the feasibility of miniature gels to separate a complex protein mixture and the sensitivity of native fluorescence detection for 2-D gels were evaluated. Isoelectric focusing was performed in IPG strips, and short slab gels (3 cm length) were used for the second dimension. To display the full size of the 2-D gel, three sequential fluorescence images (3-cm squares) were acquired. These were reconstructed into a single (inverted) image, which is shown in Figure 5a. Figure 5b presents results obtained under similar conditions by using conventional mini slab gels (6 cm length) and coomassie stain for detection. In both cases, ∼300 protein spots were detected. The patterns show a high degree of similarity. For the second-dimension separation above, a total of 3 h, which includes a 60-min separation and 2 h for staining and destaining, was needed for the standard minigel in Figure 5b. However, for the smaller, 3-cm gel of Figure 5a, separation was completed in 16 min, and an additional 45 min was used for the

(23) Tong, J.; Anderson, J. L. Biophys. J. 1996, 70, 1505-1513.

(24) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.

Figure 3. Native fluorescence detection in one-dimensional miniaturized gels. Inverted images are presented. Gel dimensions were 11 × 11 mm. Samples were divided into two groups: (a) low range, 0.1-2 ng per protein band, and (b) high range, 2-20 ng per protein band. Exposure time was set at 4 min for the low range and 1 min for the high range. Contrast of images was adjusted for proper visualization. Insets on the right correspond to the highest loaded amount as it appeared on gels without contrast adjustments.

Analytical Chemistry, Vol. 76, No. 5, March 1, 2004

1363

Figure 4. Calibration plots for SDS-PAGE separations in microgels. Each point represents the average of five replicates.

Figure 5. (a) Native fluorescence detection of 2-D electrophoresis of E. coli extract in a 3-cm PAA gel. Inverted images are presented. Three images of contiguous sections of the gel were acquired and reconstructed by using Photoshop software. The loaded protein amount was 10 µg. Each image was exposed for 2 min. (b) 2-D electrophoresis of E. coli extract in a 6-cm minigel. Detection was based on coomassie staining. The loaded protein amount was 20 µg. The images were resized to facilitate comparison with each other.

rinsing of the gel and fluorescence detection. In other experiments (data not shown), the rinsing time was reduced to 25 min with little effect on the S/N ratio. Nonetheless, resolution of these techniques was comparable. 1364

Analytical Chemistry, Vol. 76, No. 5, March 1, 2004

Following the successful coupling of miniature slab gels with native fluorescence detection, the next step was the fabrication of a micro 2-D system, in which the majority of steps are performed on a single device with even smaller dimensions. The system that was assembled (Figure 1) in our first attempt toward integration in 2-D electrophoresis. To simplify electrophoresis, a horizontal configuration was employed. Protein loading was combined with rehydration of the dry IEF strip and was done separately in a narrow strip holder. From that stage on, all steps associated with running 2-D electrophoresis were performed on the microholder. These included running IEF, equilibration of the strip with SDS-containing solution, and running the seconddimension separation. The 2 × 0.9-cm slab gel was subsequently removed from the holder for fluorescence imaging. Complete two-dimensional separation, including rehydration of the IEF strip and fluorescence detection, was completed in ∼2.5 h. The results of these 2-D experiments for various protein amounts loaded are presented in Figure 6. A zoom-in view of a specific region of these gels is also given. In addition, a coomassiestained image of an IEF strip is shown as the top inset. A low protein loading of 0.1 µg was sufficient for the detection of a large number of protein spots. On the basis of these results, the LOD of the micro 2-D technique is between 0.1 and 0.25 µg of total protein loaded. The number of protein spots detected with the Pdquest software was 220, 195, and 115 for the 2.5-, 0.25-, and 0.1-µg samples, respectively, for separation on a gel area of 0.9 × 1.5 cm. Note that the majority of proteins in this case were actually localized on ∼75% of the slab gel. Whereas 300 proteins were detected on the larger-format separations of Figure 5, 200 were detected on the microgels of Figure 6. However, for the large gels, an IPG strip with an extended pH gradient (IPG strip 3-10 nonlinear) was used. For the microgels, though, carrier ampholytes were used for IEF, so the higher spot count for the large gels is actually due to a wider separation range for the first, IEF, dimension. Although the number of spots detected with the image analysis software (Figure 6) was larger for the sample with higher amounts of protein because of S/N, separation appeared better for the gels loaded with smaller amounts of protein. Therefore, in contrast to minigels in which a relatively high amount of sample is employed, in microelectrophoresis protein loading should not

short exposure, noise spots appeared in the images while proteins could hardly be seen. The second image can then be superimposed on top of the first image to identify the noise spots. Distinguishing between protein spots and noise spots thus becomes relatively simple.

Figure 6. Two-dimensional miniaturized gel electrophoresis of E. Coli. Proteins were detected by native UV-fluorescence. Exposure was set at 2 min. Inverted images are presented. The microgels were 15 × 9 mm. Loaded total protein amounts ranged between 0.1 and 2.5 µg. Insets are enlargements of the areas bordered by the dashed lines. The IEF strip at the top left shows separation of a 10 µg protein sample stained with coomassie dye.

exceed 1 µg. The smaller wells in 1-D SDS-PAGE (Figure 3) and the smaller IEF gels in 2-D electrophoresis (Figure 6) also mean that smaller sample volumes are required. Careful inspection indicates the presence of noise in the fluorescence gel images. These appear as distinct black dots on the microgels (Figure 6) and as fainter dots in the larger gels (Figure 5). These dots are presumably due to dust and imperfections in the gels and can be differentiated from the larger protein bands on the basis of their size and shape. However, since 2-D gels are often analyzed with imaging software, erroneous detection may result. In using fluorescence detection, screening of these noise spots is straightforward. It was found that the fluorescence emission or scattering from such noise spots was considerably larger than that of proteins. Therefore, a simple procedure was devised for their identification. First, the gel was imaged for an extended period in order to maximize the sensitivity for proteins. Next, the gel was exposed for a short, 1-s period. During this

CONCLUSIONS This study demonstrates the successful coupling of microelectrophoresis with native fluorescence detection. The method was applied to the one- and two-dimensional separations of proteins. In one-dimensional sizing, a mixture containing six proteins of diverse molecular weights was resolved on a 1-cm-long gel. Separation was completed in 8 min, and a detection limit of 0.04 ng was reached. A simple holder was designed for coupling IEF with SDS-PAGE for 2-D electrophoresis. Reproducible results were obtained for various protein loadings. Two-dimensional electrophoresis was completed in ∼2.5 h, and 200 protein spots were detected on a 1-cm-square area of the slab gel. The detection limit of this format was lower than 0.25 µg of total protein. Although microelectrophoresis has not been recognized as an alternative to conventional slab-gel techniques, the clear advantages of high speed, high sensitivity, elimination of pre- or postlabeling, and compatibility with subsequent mass spectrometric analysis should be of great importance for the developing field of proteomics. In this research, the improved design of sample introduction into 1-D microgels and the successful employment of UV lamps and CCD are important steps in the development of high-throughput methods for fast and convenient analysis of complex protein mixtures. Quantitative results can be derived directly from the images without the need for a densitometer. Furthermore, significant progress toward sensitivity enhancement and integration of 2-D electrophoresis was made. Although manual operation of the system is presented here, the individual steps are ready for full automation. For example, microfluidic channels can be built into a single device in which an IEF strip and a microslab gel have been incorporated for the consecutive operations of sample loading onto the IEF strip, rinsing, IEF separation, equilibration with SDS, second-dimension electrophoresis (size separation), and on-chip native fluorescence detection. Indeed, onchip detection through a fused-silica coverslip on top of the miniaturized gel would reduce noise spots due to dust particles. ACKNOWLEDGMENT E.S.Y. thanks the Robert Allen Wright Endowment for Excellence for support. The Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University under Contract No. W-7405-Eng-82. This work was supported by the Director of Science, Office of Basic Energy Sciences, Division of Chemical Sciences. Received for review November 12, 2003. Accepted December 15, 2003. AC035336G

Analytical Chemistry, Vol. 76, No. 5, March 1, 2004

1365