Capillary Isoelectric Focusing of Proteins with Liquid Core Waveguide

Anal. Chem. 2003, 75, 4887-4894

Capillary Isoelectric Focusing of Proteins with Liquid Core Waveguide Laser-Induced Fluorescence Whole Column Imaging Detection Zhen Liu and Janusz Pawliszyn*

Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1

A capillary isoelectric focusing (CIEF) system with liquid core waveguide (LCW) laser-induced fluorescence whole column imaging detection was developed in this study. A Teflon AF 2400 capillary was used as both the separation channel and the axially illuminated LCW. The excitation light was introduced at one end of the capillary, and propagated forward within the capillary. As the Teflon AF 2400 capillary has a refractive index (n ) 1.29-1.31) lower than that of water (n ) 1.33), total internal reflection was very apparent. The employment of the Teflon AF 2400 capillary avoided the use of high refractive index additives such as glycerol, accommodating the system to wider applications. Due to its inert chemical properties, the capillary exhibited limited protein adsorption and electroosmotic flow; thus, the need for capillary preconditioning with polymeric solution and the addition of polymeric additives into the sample mixture can be avoided. Three types of proteins, naturally fluorescent proteins, covalently labeled proteins, and noncovalently labeled proteins, were examined using this method. CIEF under denaturing conditions was also explored, and several advantages over the native mode were found. When compared to a commercially available instrument with UV detection, the separation efficiency and peak capacity were similar while the detection sensitivity was enhanced by 3-5 orders of magnitude. Capillary isoelectric focusing (CIEF) is a high-resolution capillary electrophoresis (CE) technique for the separation of proteins and other zwitterionic biomolecules.1-5 Analytes are separated based on their differences in isoelectric points (pI). Generally, CIEF provides a resolving power (or ∆pI, representing the minimum difference in isoelectric points to resolve two protein) of 0.02 using pH 3-10 carrier ampholytes. Recently, ∆pI as small as 0.004 has been reported.6 CIEF is a unique technique in which the sample components within the whole separation capillary are focused into narrow bands, allowing it to act as a * Corresponding author. E-mail: [email protected]. (1) Hjerte´n, S.; Zhu, M. D. J. Chromatogr. 1985, 346, 265-270. (2) Liu, X.; Sosic, Z.; Krull, I. S. J. Chromatogr., A 1996, 735, 165-190. (3) Righetti, P. G.; Gelfi, C.; Conti, M. J. Chromatogr., B 1997, 699, 91-104. (4) Shen, Y.; Smith, R. D. Electrophoresis 2002, 23, 3106-3124. (5) Shimura, K. Electrophoresis 2002, 23, 3847-3857. (6) Shen, Y.; Xiang, F.; Veenstra, T. D.; Fung, E. N.; Smith, R. D. Anal. Chem. 1999, 71, 5348-5353. 10.1021/ac034587m CCC: $25.00 Published on Web 08/08/2003

© 2003 American Chemical Society

sample concentration technique as well. For this reason, in twodimensional separations, CIEF is widely employed as a firstdimension separation coupled with a second-dimension separation such as capillary zone electrophoresis (CZE),7 microcolumn highperformance liquid chromatography (µ-HPLC),8 mass spectrometry,9,10 or capillary gel electrophoresis.11 CIEF continues to play an increasingly important role in the rapidly developing field of proteomics. As opposed to other CE modes such as CZE where the analytes migrate past a stationary detector, CIEF is a steady-state electrophoretic technique and the focused sample bands are still inside the capillary at the completion of focusing if no outer force is applied. Thus, the detection in CIEF is different. Three detection schemes in CIEF exist. The first scheme is single-point detection, which needs a mobilization step to chemically,12-15 electroosmotically,16-19 or hydrodynamically20,21 mobilize the focused sample bands past the detection point. However, the mobilization step requires extra time and might distort the focused bands. The second scheme is whole column scanning detection,12,22 in which the entire column is transported past a single detection point. This method, in principle, obviates the need for mobilization of the focused zones, but inhomogeneity of the capillary wall may cause relatively high noise level and band broadening may occur, as the voltage is not maintained during detection. The third scheme is whole column imaging detection (WCID).23-25 By using a short (7) Mohan, D.; Lee, C. S. Electrophoresis 2002, 23, 3160-3167. (8) Chen, J.; Lee, C. S.; Shen, Y.; Smith, R. D.; Baehrecke, E. H. Electrophoresis 2002, 23, 3160-3167. (9) Tang, Q.; Harrata, A. K.; Lee, C. S. Anal. Chem. 1995, 67, 3515-3519. (10) Yang, L.; Lee, C. S.; Hofstadler, S. A.; Smith, R. D. Anal. Chem. 1998, 70, 4945-4950. (11) Yang, C.; Liu, H.; Yang, Q.; Zhang, L.; Zhang, W.; Zhang, Y. Anal. Chem. 2003, 75, 215-218. (12) Hjerte´n, S.; Liao, J.-L.; Yao, K. J. Chromatogr. 1987, 387, 127-138. (13) Zhu, M. D.; Hansen, D. L.; Burd, S.; Gannon F. J. Chromatogr. 1989, 480, 311-320. (14) Zhu, M. D.; Rodriguez, R.; Wehr, T. J. Chromatogr. 1991, 559, 479-488. (15) Zhu, M.; Wehr, T.; Levi, V.; Rodriguez, R.; Shiffer, K.; Cao, Z. A. J. Chromatogr. 1993, 652, 119-129. (16) Mazzeo, J. R.; Krull, I. S. Anal. Chem. 1991, 63, 2852-2857. (17) Mazzeo, J. R.; Krull, I. S. J. Chromatogr. 1992, 606, 291-296. (18) Mazzeo, J. R.; Krull, I. S. Biotechnology 1991, 10, 638-645. (19) Molteni, S.; Frischknecht, H.; Thorman, W. Electrophoresis 1994, 15, 2230. (20) Chen, S. M.; Wiktorowicz, J. E. Anal. Biochem. 1991, 206, 84-90. (21) Hempe, J. M.; Craver, R. D. J. Clin. Chem. 1994, 40, 2288-2295. (22) Wang, T.; Hartwick, R. A. Anal. Chem. 1992, 64, 1745-1747. (23) Wu, J.; Pawliszyn, J. Anal. Chem. 1992, 64, 224-227. (24) Wu, J.; Pawliszyn, J. Anal. Chem. 1992, 64, 2934-2941.

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003 4887

separation column (5 cm long or shorter) and a charge-coupled device (CCD) camera, the focused sample bands within the whole column are perpendicularly imaged, eliminating the need for a mobilization step. Using WCID, the total separation time is greatly reduced, usually less than 5 min. The WCID scheme has been accomplished with concentration gradient,23-26 UV adsorption,25,27 and laser-induced fluorescence (LIF)28 detection, and instruments specially designed for CIEF-UV-WCID have been commercialized.29-31 Due to its speed, CIEF-WCID has also been employed as a second-dimension separation coupled with a first-dimension separation such as HPLC,32 CZE, or micellar electrokinetic chromatography.33 The sensitivity in CIEF is poor when UV detection is used (usually 10-6-10-7 M for protein), due to the strong absorption of the carrier ampholyte in the UV absorption range of proteins. A problem associated with UV detection in CIEF is protein precipitation. Proteins tend to precipitate when they reach their pI values. The precipitation problem is more serious at higher protein concentration, particularly in single-point detection configuration because of substantial delay between focusing and detection. To improve the detection sensitivity and reduce precipitation effects, attempts have been made to couple CIEF with LIF detection,34-42 most of which employ single-point detection. Recently, Huang and Pawliszyn41 proposed an axially illuminated LIF-WCID for CIEF. In this method, the incident laser beam was introduced into the separation capillary and propagated inside a capillary by means of total internal reflection. Total internal reflection was realized through the use of a Teflon capillary with a refractive index of 1.38 and the addition (20% v/v) of an additive of high refractive index (glycerol, n ) 1.47) to the sample solution. A limit of detection (LOD) of 10-13 M was obtained for the naturally fluorescent protein R-phycoerythrin. This method has also been applied to noncovalently labeled protein, in which Sypro red-labeled bovine serum albumin (BSA) was focused.42 Unfortunately, the enhancement in detection sensitivity was not significant, only 1 order of magnitude improvement over UV detection. Compared to transversely illuminated WCID, the axial illumination configuration exhibits great advantages, including ease in light alignment and reduction in background noise. However, the use of glycerol required for this method is undesirable in some cases. For example, the maximum tolerable (25) Wu, J.; Pawliszyn, J. Anal. Chem. 1994, 66, 867-873. (26) Wu, J.; Pawliszyn, J. Anal. Chim. Acta 1995, 299, 337-342. (27) Wu, J.; Pawliszyn, J. Analyst 1995, 120, 1567-1571. (28) Wu, X.-Z.; Wu, J.; Pawliszyn, J. Electrophoresis 1995, 16, 1474-1978. (29) Wu, J.; Watson, A. J. Chromatogr., B 1998, 714, 113-118. (30) Wu, J.; Li, S.; Watson, A. J. Chromatogr., A 1998, 817, 163-171. (31) Wu, J.; Watson, A. H.; Torres, A. R. Am. Biotech. Lab. 1999, 17, 24-26. (32) Tragas, C.; Pawliszyn, J. Electrophoresis 2000, 21, 227-237. (33) Sheng, L.; Pawliszyn, J. Analyst 2002, 127, 1159-1163. (34) Shimura, K.; Kasai, K. Electrophoresis 1995, 16, 1479-1484. (35) Lillard, S. J.; Yeung, E. S. J. Chromatogr., B 1996, 687, 363-369. (36) Cruickshank, K. A.; Olvera, J.; Muller, U. R. J. Chromatogr., A 1998, 817, 41-47. (37) Shimura, K.; Matsumoto, H.; Kasai, K. Electrophoresis 1998, 19, 22962300. (38) Verbeck, G. F.; Beale, S. C. J. Microcolumn Sep. 1999, 11, 708-715. (39) Horka´, M.; Willimann, T.; Blum, M.; Nording, P.; Friedl, Z.; Sˇ lais, K. J. Chromatogr., A 2001, 916, 65-71. (40) Sˇ lais, K.; Horka´, M.; Nova´ckova´, J.; Friedl, Z. Electrophoresis 2002, 23, 16821688. (41) Huang, T.; Pawliszyn, J. Analyst 2000, 125, 1231-1233. (42) Sze, N. S. K.; Huang, T.; Pawliszyn, J. J. Sep. Sci. 2002, 25, 15-17.

4888

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

concentration of glycerol in the case of proteins noncovalently labeled with NanoOrange is only 10%.43 In addition, interaction between glycerol and fluorescein isothiocyanate-labeled proteins was found when the glycerol content was higher than 20%, giving rise to changes in pI values (see later discussion). Clearly, other strategies to realize total internal reflection may facilitate wider applications of CIEF with the axially illuminated LIF-WCID. Fortunately, the Dupont patented Teflon AF is a desired material for this purpose. Teflon AF is an amorphous copolymer of tetrafluoroethylene and 2,2-bis(trifluoromethyl)-4,5difluoro-1,3-dioxole, which is an excellent material for spectroscopic applications due to its unique optical properties. The Teflon AF family has two grades, AF 1600 and AF 2400, differing in their glass transition temperature. Teflon AF 2400, the more popular grade in spectroscopic studies, has a low refractive index (n ) 1.29-1.31) and is transparent from UV to IR range (transmission >95%). It showed excellent performance in liquid core waveguide (LCW)-HPLC44,45 and CE.46 In this study, a CIEF system with LCW-LIF-WCID was built using a Teflon AF 2400 capillary. The use of the Teflon AF 2400 capillary facilitates the realization of total internal reflection and thus avoids the addition of glycerol. Due to the inertness of the Teflon AF 2400 capillary, CIEF can be operated without the employment of a polymeric additive that is used to suppress electroosmotic flow (EOF) and protein adsorption. CIEF of naturally fluorescent proteins, covalently labeled proteins, and noncovalently labeled proteins was explored. Denaturing mode was also developed, and some advantages over the native mode were observed. EXPERIMENTAL SECTION Apparatus. A schematic of the instrument setup is illustrated in Figure 1. The system was composed of three units: laser illumination unit, separation unit, and detection unit. In the laser illumination unit, the laser beam produced by an air-cooled argon ion laser (Cyonics, San Jose, CA) was filtered with a 488-nm bandpass filter and then focused onto a 100-µm core-size optical fiber with a convex lens with a focal length of 15 cm. The optical fiber was held by an adapter. A baffle was placed in the front of the laser and was open only during imaging, to avoid photodecomposition of the sample. Both the separation unit and detection unit were covered with a black box. The separation unit was a cartridge supporting the separation capillary and two electrolyte reservoirs, built on a 7.5-cm-long glass plate. The separation capillary was a 5.5-cm-long Teflon AF 2400 tube. The anodic end was connected with a piece of Teflon AF 2400 capillary and the cathodic end with a piece of fused-silica capillary, through two small pieces of microporous hollow fiber by gluing. Two polystyrene vials were glued on the ends of the separation capillary as electrolyte reservoirs, and the effective detection length of the capillary was 5 cm. The microporous hollow fibers functioned as physical barriers to prevent the macromolecule analyte from entering the electrolyte reservoirs but allow small ions, like protons and (43) http://www.probes.com/media/pis/mp06666.pdf. (44) Marquardt, B. J.; Vahey, P. G.; Synovec, R. E.; Burgess, L. W. Anal. Chem. 1999, 71, 4808-4814. (45) Gooijer, C.; Hoornweg, G. P.; de Beer, T.; Bader, A.; van Iperen, D. J.; Brinkman, U. A. T. J. Chromatogr., A 1998, 824, 1-5. (46) Wang, S.-L.; Huang, X.-J.; Fang, Z.-L.; Dasgupta, P. K. Anal. Chem. 2001, 73, 4545-4549.

Figure 1. Schematic diagram of the instrumental setup for CIEF-LCW-LIF-WCID.

hydroxyl ions, to enter the separation capillary from the electrolyte reservoirs. It also was used to couple light from the end of the optical fiber to the separation channel. The cartridge was mounted onto an x-y translational stage to facilitate adjustment of the cartridge to keep the separation capillary horizontal and parallel to the plane of the CCD camera. High voltage was provided with a RE-3002B regulated high-voltage supply (Northeast Scientific Co., Cambridge, MA). The optical fiber was inserted into the connecting capillary at the anodic end, guiding the laser beam into the separation capillary. The detection unit was composed of two subunits, i.e., a fluorescence imaging detection part and an absorption detection part. The fluorescence imaging detection part included two components, a UV-sensitive CCD camera (Princeton Instrument, Trenton, NJ) and a 530-nm long-pass filter. The CCD camera was coupled with an A4869 50-mm UV lens (Hamamatsu Inc., Hamamatsu, Japan) and controlled by a ST-130 controller (Princeton Instrument). It was thermoelectrically cooled to -20 °C. The exposure time of the CCD camera ranged from 5 ms to 5 s, depending on the fluorescence intensity of the sample. Data processing was implemented with the software WinView (Princeton Instrument) on a personal computer. The absorption detection included three components, a S2506 photodiode (Hamamatsu Corp., Bridgewater, NJ), an OPA-111 operational amplifier (Burr Brown, Tucson, AZ), and a common multimeter. The photodiode was facing the injection end of the separation channel, with its plane vertical to the connecting capillary at the injection end. Compared to the instrument setup previously used,41 the main differences included the modification of the laser illumination unit, the utilization of a Teflon AF 2400 capillary as the separation channel, and the addition of the absorption detector. For comparison of sensitivity, several experiments were performed on an iCE 280 instrument (Convergent Bioscience Ltd., Toronto, Canada) with UV detection at 280 nm and a cartridge of 5 cm × 100 µm i.d. internally fluorocarbon-coated fused-silica capillary. Reagents and Materials. R-Phycoerythrin was purchased from Calbiochem-Novabiochem Corp. (La Jolla, CA), at a concentration of 20 mg/mL. Green fluorescent protein (GFP), at 0.5 mg/ mL, was donated by Convergent Bioscience Ltd. (Toronto, Canada). Methyl cellulose (MC, average molecular weight ∼86 000 and viscosity of 2% aqueous solution 4000 cP) was purchased from

Aldrich (Milwaukee, WI). Polyvinylperrolidone (PVP, average molecular weight ∼360 000 and intrinsic viscosity 80-100 K), glycerol, fluorescein isothiocyanate (FITC)-insulin, FITC-labeled human serum albumin (FITC-HSA), bovine albumin-R-D-galactopyranosylphenyl isothiocyanate fluorescein isothiocyanate (R-Dglactosylated FITC-albumin), and BSA were obtained from Sigma (St. Louis, MO). Carrier ampholytes (Pharmalytes, pH 3-10 and pH 4-6.5) were obtained from Sigma. NanoOrange protein quantitation kit was purchased from Molecular Probes (Eugene, OR). Anolyte and catholyte were 100 mM phosphoric acid and 100 mM sodium hydroxide, respectively. Water was purified with an ultrapure water system (Barnstead/Thermolyne, Dubuque, IA). Optical fiber with a 100-µm core was purchased from Polymicro Technologies Inc (Phoenix, AZ). Microporous hollow fiber of 0.03µm pore size and 383-µm i.d. was obtained from Hoechst Celanese (Frankfurt, Germany). Teflon AF 2400 capillaries of 250-µm i.d. and 500-µm o.d. were obtained from Biogeneral Inc. (San Diego, CA), and the Teflon AF 2400 capillaries of 167-µm i.d. and 364µm o.d. were purchased from Random Technologies (San Diego, CA). Noncovalent Protein Labeling. NanoOrange working solution (1×) was prepared according to the vendor instructions. The labeling procedure was modified based on the description in the instructions. Specifically, 10 µL of 0.1 mg/mL BSA aqueous solution was mixed with a 990 µL of 1× NanoOrange working solution. The mixture was then incubated at 90-96 °C for 10 min and cooled naturally to room temperature for 20 min, protected from light. CIEF. For experiments with a polymeric additive, the separation capillary was conditioned with an aqueous solution containing 0.25% MC or 0.5% PVP for 30 min. For experiments using a bare capillary, the capillary was rinsed with water for 10 min. Samples were injected with a syringe. A 1-3-kV voltage was then applied to the electrolyte reservoirs to begin the focusing. RESULTS AND DISCUSSION Optical Considerations. In a previous axially illuminated LIFWCID study,41 where a common Teflon tube was used, the addition of glycerol to the sample solution was critical. To reach total internal reflection, at least 20% glycerol had to be added. Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

4889

Figure 2. Comparison of CIEF of FITC-insulin in the absence of (A) and in the presence of (B) glycerol in the sample solution. Sample, 2 µg/mL FITC-insulin containing 0.25% MC/4% carrier ampholyte (pH 3-10) without glycerol (A) and with 21% glycerol (B); separation capillary, 5 cm × 250 µm i.d. Teflon AF 2400 tube; applied voltage, 1 kV; excitation wavelength, 488 nm; detection filter, 530 nm longpass filter.

Comparative studies of CIEF of R-phycoerythrin, with and without the addition of glycerol, indicate that the presence of glycerol in the sample mixture generally does not influence the focused profiles of R-phycoerythrin, other than reducing the focusing speed

due to the high viscosity of glycerol (data not shown). However, it was observed that the presence of glycerol in the sample solution is detrimental in some cases. For example, the addition of glycerol at a concentration higher than 20% was unfavorable for CIEF of FITC-labeled insulin. As shown in Figure 2, when 21% of glycerol was added to the sample solution, a substantial shift in the peak position of one of the two peaks observed in the absence of glycerol was observed. In addition, the presence of glycerol in CIEF of GFP resulted in reduced resolution (data not shown). All these findings may be attributed to the interaction between glycerol and the tested protein. In contrast, using the Teflon AF 2400 capillary, total internal reflection could be easily reached without the addition of glycerol at all. A prerequisite for LIF detection is that the analyte must have absorption at the excitation wavelength. However, strong absorption will give rise to noticeable attenuation of the laser intensity along the separation channel, which is unfavorable for quantitative analysis. For a given fluorescent protein, the absorption is related to the concentration of the sample. In this study, the distribution of the laser intensity inside the separation channel was evaluated in terms of the fluorescence distribution imaged with a sample injected into the separation channel and without applying voltage. The total absorption of the sample inside the separation channel and the connecting capillary was evaluated by comparison of the response signal of the photodiode with that when water, instead of sample, was injected. As shown in Figure 3, the intensity attenuation and light adsorption depended on the concentration of the sample. At lower concentrations, light loss and absorption were less significant. When 0.2 µg/mL R-phocoerythrin was injected, only 6% of the laser intensity was lost after the laser beam had traveled 5 cm, and only 20% of the incident laser was absorbed after the laser beam had traveled 7 cm. As a comparison, when 20 µg/mL was injected, more serious light attenuation and absorption occurred, which resulted in a loss of 25% of original intensity within a distance of 5 cm and an absorption of 40% within a distance of 7 cm. The data shown in Figure 3 were different for fluorescence and absorption, but they were in good agreement if one considered that the sample within the connecting capillary contributed extra absorption. On the basis of the discussion above,

Figure 3. Fluorescence intensity distribution along the separation channel and light absorption of the sample at different concentrations of R-phycoerythrin. FE and FS represent the fluorescence intensity at the ending and starting point of the separation channel, respectively. I and I0 represent the laser intensity detected by the photodiode when the sample and water were injected, respectively. Conditions were identical to those described in Figure 2 except no voltage was applied. 4890

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

sample solution should be prepared as dilute as possible in order to avoid significant laser intensity attenuation. For a new sample, a decision would need to be made according to the absorption level measured by the photodiode. As a rule of thumb, if the intensity loss was higher than 30%, a more dilute sample was required. Adsorption of protein onto the capillary wall was also observed to be a factor that may influence the total internal reflection. Protein adsorption might produce a protein layer of high refractive index; thus, loss of the laser intensity will occur at the point where protein adsorption occurred. If the laser shot was controlled to be as short as possible, such a phenomenon was seldom observed. However, if the sample solution was continuously exposed to laser radiation during focusing, it was often observed. In such a case, it is most likely that protein fragments generated by photodecomposition were more adsorptive to the capillary wall than the intact protein was. To protect samples from photochemical decomposition, a baffle was arranged in the front of the laser, which was open only at the time of imaging data acquisition. CIEF under Native Conditions. Native CIEF was explored with three types of proteins, including naturally fluorescent protein, covalently labeled protein, and noncovalently labeled protein. Representative electropherograms are shown in Figure 4. R-Phycoerythrin is a naturally fluorescent protein with maximum excitation wavelength of 565 nm (second maximum 480 nm) and maximum emission wavelength of 578 nm. It was focused within 10 min, exhibiting a single peak (Figure 4A). The maximum excitation wavelength and maximum emission wavelength of FITC-labeled proteins are 488 and 520 nm, respectively. R-DGlactosylated FITC-albumin was focused within 12 min, and two broad peaks were observed (Figure 4B). The peak focused at higher pH was likely a protein degradation product produced during extended storage. NanoOrange is a typical noncovalent labeling dye, allowing accurate detection of proteins in solution ranging from 10 ng/mL to 10 µg/mL.43 When bound to proteins, this dye has a broad excitation peak centered at ∼470 nm and a broad emission peak centered at ∼570 nm. The focusing of NanoOrange-labeled BSA was accomplished within 15 min, and a very sharp peak was observed (Figure 4C). Clearly, native CIEF exhibited good separation efficiency for the naturally fluorescent protein and noncovalently labeled protein tested but rather poor separation efficiency for the covalently labeled proteins (Figures 2A and 4B). The broad peaks of R-D-glactosylated FITC-albumin were probably due to multiple labeling. According to the vendor, the FITC content in R-D-glactosylated FITC-albumin was 2.6 mol/ mol of protein. In the case of FITC-insulin characterized by the stoichiometric ratio (FITC/insulin) of 1.2:1 (Figure 2A), the poor peak efficiency observed was most likely due to the interaction of the protein with the capillary wall. CIEF under Denaturing Conditions. Although native mode is more common in CIEF, it might not always be as effective, such as in Figures 2A and 4B. As an alternative, a denaturing mode with the utilization of urea may offer some distinct advantages. The presence of urea can effectively solubilize the protein and reduce the interactions between the protein and the capillary wall, thus avoiding protein precipitation and improving the peak shape. More importantly, the denaturing mode may be used to indicate structural changes of the protein due to denaturation. We applied

Figure 4. Native CIEF: (A) naturally fluorescent protein, (B) covalently labeled protein, and (C) noncovalently labeled protein. Sample: (A) 2 µg/mL R-phycoerythrin containing 0.25% MC/4% carrier ampholyte (pH 3-10), (B) 2 µg/mL R-D-glactosylated FITCalbumin containing 0.25% MC/4% carrier ampholyte (pH 3-10), and (C) 200 ng/mL NanoOrange-labeled BSA containing 2% carrier ampholyte (pH 3-10). Conditions for (A) and (B) identical to those described in Figure 2. Conditions for (C): separation capillary, 5 cm × 167 µm i.d. Teflon AF 2400 tube; applied voltage, 0-5 min, 1 kV, after 5 min, 3 kV; other conditions identical to those described in Figure 2.

the denaturing mode exclusively to the naturally fluorescent proteins and the FITC-labeled proteins. As the presence of urea breaks hydrophobic interaction, the maximum tolerable concentration of urea for NanoOrange-labeled proteins is only 1 M.43 Thus, NanoOrange-labeled proteins were excluded from the Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

4891

Table 1. Comparison of Detection Sensitivity between the CIEF with LCW-LIF-WCID and the ICE 280 CIEF Instrument with UV-WCID LOD (S/N ) 3) (M) protein

a

UV 10-7

R-phycoerythrin GFP HSA

8.3 × 6.5 × 10-7 1.0 × 10-6

BSA

9.1 × 10-8

LIF

improvement

6.4 × 3.6 × 10-11 2.2 × 10-10 (FITC labeled)a 1.2 × 10-10 (FITC labeled)b 8.1 × 10-12 (NanoOrange labeled)

130 000 18 000 4 500 8 300 11 000

10-12

Native mode. b Denaturing mode.

denaturing mode. The sample solutions under denaturing mode contained 6 M urea while other constituents including the carrier ampholyte and polymer additive were the same as in native mode. The prepared samples were kept in the dark for at least 1 h before use. The focusing process was finished in 15 min. The apparent difference under denaturing mode between the two types of protein was that the fluorescence signal of the naturally fluorescent proteins became much weaker, whereas the FITC-labeled proteins maintained a consistent fluorescence signal to that observed in native mode. The reduced fluorescence observed for the naturally fluorescent proteins may be attributed to the destruction of some fluorophores through the denaturation process, whereas the covalent labeling dye FITC was robust enough to withstand the presence of urea. Representative electropherograms are shown in Figure 5. R-Phycoerythrin is made up of at least three different subunits and each subunit has at least one fluorophore,47 so the presence of at least three peaks was expected with denaturation. The results shown in Figure 5A confirm the presence of several peaks. Insulin is composed of two different subunits that are crosslinked by two disulfide bridges. As urea usually cannot break the disulfide bonds, only one peak was expected. However, at least three peaks were observed under denaturing mode, as shown in Figure 5B. The reason for this conflicting result is not clear at the present time. On the other hand, comparing Figure 5 with Figures 2A and 4A clearly suggests that the peak shape for R-phycoerythrin and FITC-insulin was greatly improved under denaturing conditions. The focused peaks for FITC-insulin showed high efficiency, with a peak width of ∼2.5 mm. For R-Dglactosylated FITC-albumin and FITC-HSA, the improvement of the peak shape was less significant and is probably due to the multiple labeling of these proteins. With greater peak efficiency, the detection sensitivity for FITC-HSA under denaturing conditions was improved by a factor of almost 2 (see Table 1). Performance of Bare Capillary. EOF and protein adsorption strongly influence the stability and efficiency of the focused bands in CIEF-WCID. Usually, capillary preconditioning with a polymer solution such as MC or PVP and addition of the same polymeric additive to the sample mixture are necessary to suppress EOF and protein adsorption. In this study, the likelihood of direct employment of bare Teflon AF 2400 capillary without use of polymeric additive at all was investigated. An example of good peak efficiency in a bare capillary has been illustrated in Figure 4C. Using R-phycoerythrin and GFP as test proteins, the perfor(47) http://www.martekbio.com/Fluorescent_Products/Technical%20Bulletin% 201.pdf.

4892 Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

Figure 5. Denaturing CIEF: (A) R-phycoerythrin and (B) FITCinsulin. Sample: (A) 2 µg/mL R-phycoerythrin containing 6 M urea/ 0.5% PVP/2% carrier ampholyte (pH 3-10); (B) 2 µg/mL FITC-insulin containing 6 M urea/0.5% PVP/2% carrier ampholyte (pH 3-10); applied voltage: 0-5 min, 1 kV, after 5 min, 3 kV; other conditions identical to those described in Figure 2.

mance of the bare Teflon AF 2400 tube was further investigated. Eight consecutive runs were carried out with no loss in efficiency, indicating that Teflon AF 2400 capillary was inert for the proteins tested and that protein adsorption was very small. This is not surprising if it is noted that fluorocarbon, which has chemical composition and properties similar to those of Teflon, has also been used as a coating for eliminating protein adsorption in CE for a long time. Once the focusing completed, the focused peak was found to be very stable inside the capillary. The average peak shifting was 0.6 pixels/min or 0.03 mm/min under 3 kV, being comparable to that with preconditioning and addition of MC or

Figure 7. Peak broadening and resolution enhancement due to extracolumn volume. Sample, 500 ng/mL GFP containing 2% carrier ampholyte (pH 4-6.5) with 0.5% PVP; separation capillary, 5 cm × 44 µm i.d. Teflon AF 2400 tube; applied voltage, 3 kV; other conditions identical to those described in Figure 2.

Figure 6. Comparison of CIEF of GFP in the presence of (A) and in absence of (B) PVP inside the separation channel. Sample, 500 ng/mL GFP containing 2% carrier ampholyte (pH 4-6.5) with 0.5% PVP (A) and without the addition of PVP (B); separation capillary, 5 cm × 167 µm i.d. Teflon AF 2400 tube; applied voltage, 0-5 min, 1 kV, after 5 min, 3 kV; other conditions identical to those described in Figure 2.

PVP to the sample mixture. This result indicates that the Teflon AF 2400 tubing has a very small EOF under electric field. Electropherograms of GFP with and without the use of PVP are shown in Figure 6A and B, respectively. Clearly, the efficiency, resolution, and peak position using both sets of conditions were almost the same. Based on above results, it can be concluded that the Teflon AF 2400 capillary can be directly used as the separation channel without addition of polymeric additive, simplifying the experimental procedure and sample preparation. Peak Broadening and Increased Resolution due to Extracolumn Effect. It was found that the volume inside the connecting points played a critical role in the formation of the pH gradient. Since the pH gradient is formed within the total space between the two microporpus hollow fibers, the pH gradient will be generated predominantly within the connecting points instead of the separation channel, if the volume inside the connecting points is much larger than the volume of the separation channel. As the volume of the connecting point is void, it is referred to as extracolumn volume, a term taken from chromatographic nomenclature. The effect of extracolumn volume causes a narrower pH gradient across the separation channel and thus results in peak broadening and increased resolution (the effect on resolution is different from that in chromatography). The extracolumn effect

becomes more serious when a capillary of a narrower bore size is used. If the allowable extracolumn volume for each connecting point is 5% of the total volume of the separation channel, when the inner diameter of the microporous hollow fiber is 380 µm, the allowable maximum length for the connecting point is 30 and 480 µm for 44- and 167-µm-i.d. capillaries, respectively. In practice, a very small distance (such as 30 µm) is rather difficult to control by gluing. Figure 7 shows a CIEF electropherogram of GFP in a cartridge of a 44-µm-i.d capillary with connecting point lengths of 50-80 µm (estimated by naked eye under microscope). The peaks were broader while the resolution was higher as compared with Figure 6. Further, the length of the connecting points must be the same at the anodic and cathodic ends. Otherwise, the fluxes of proton and hydroxyl ion are different, and the focused peak will shift to the end with shorter length (data not shown). In future studies, capillaries of smaller outer diameter and microporous hollow fibers of smaller inner diameter should be employed to limit the dead volume of connecting points. Efficiency, Peak Capacity, Resolution, and Reproducibility. The best separation efficiency observed in this study was the focused peak of NanoOrange-labeled BSA, giving a peak width of 0.15 mm and a peak capacity of 33. It was observed that the separation efficiency and peak capacity in this study are low. However, it should be noted that the peak efficiency in CIEF depends on the titration curve and the homogeneity of the protein under study. CIEF of GFP with LCW-LIF-WCID was compared with that with UV-WCID on the commercial instrument iCE 280. As shown in Figure 8, the peak efficiency and peak capacity of the CIEF with LCW-LIF-WCID were very similar to those obtained on the instrument iCE 280 with UV-WCID, which can generate a peak capacity of at least 100. Due to lack of fluorescent pI markers, evaluation of the resolving power of the system was not performed. The system built in this study displayed good reproducibility. The standard deviation for run-to-run reproducibility of peak position was 1 and 0.8 mm (n ) 8) for the PVP-conditioned capillary and bare capillary, respectively. Detection Sensitivity. Detection sensitivity of the CIEF-LCWLIF-WCID system was evaluated and compared with that obtained on the iCE 280 system with UV-WCID. The LODs were calculated Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

4893

much lower concentration, and thus protein precipitation was effectively eliminated and not observed throughout this study. It should be emphasized that there is still space for further improvements of the detection sensitivity, since the optics of the detection unit in this work was not optimized. The working temperature of the CCD camera was set at -20 °C because no ventilation was employed in the box holding the experimental setup. In open air, it can be as low as -40 °C. The CCD camera and lens used in this study were all optimized for use in UV. Moreover, only a small portion of the fluorescence emission was collected by the camera ( ∼2%).

Figure 8. CIEF of GFP with LCW-LIF-WCID (A) and UV-WCID (B). Sample: (A) 500 ng/mL GFP containing 0.25% MC/4% carrier ampholyte (pH 3-10); (B) 50 µg/mL GFP containing 0.25% MC/4% carrier ampholyte (pH 3-10). Conditions for (A) identical to those described in Figure 2. Conditions for (B): separation capillary, 5 cm × 100 µm i.d. internally fluorocarbon coated fused-silica capillary; detection wavelength, 280 nm; applied voltage, 3 kV.

in terms of peak height and are listed in Table 1. The LODs for UV-WCID ranged from 10-6 to 10-8 M. In contrast, the LODs for LIF-WCID ranged from 10-10 to 10-12 M, illustrating improved detection by a factor of 4500-130 000, depending on the protein tested. Due to their high quantum yields, naturally fluorescent proteins showed very high detection sensitivities, as did NanoOrange-labeled BSA. The relatively poor detection sensitivity of FITC-HSA was attributed to the long-pass filter used (cutoff at 530 nm), which excluded the maximum emission at 520 nm of the protein. This filter is also not ideal for GFP detection, as its maximum emission occurs at 515 nm. Since lower LODs were achieved using the LCW-LIF-WCID, proteins were present at

4894

Analytical Chemistry, Vol. 75, No. 18, September 15, 2003

CONCLUSIONS By virtue of the unique optical properties of Teflon AF 2400 tubing, a CIEF system with LCW-LIF-WCID has been developed. Due to its lower refractive index compared to water, total internal reflection was achieved, avoiding the addition of high refractive index additives. The Teflon AF 2400 capillary also showed limited protein adsorption and EOF under an electric field; thus, the use of polymers such as MC or PVP was not necessary. These properties not only simplified experimental procedure and sample preparation but also eliminated problems associated with the presence of polymeric additive. Performance of the system was evaluated using naturally fluorescent proteins, covalently labeled proteins, and noncovalently labeled proteins under native and denaturing modes. The system was shown to be stable, reproducible, and sensitive. In comparison with the commercialized CIEFUV-WCID instrument, the CIEF-LCW-LIF-WCID system studied exhibited similar separation efficiency and peak capacity, while the sensitivity was improved by 3-5 orders of magnitude without optimization of the optical system. The photodiode can be used to detect the light intensity loss in the channel to determine the need for more diluted samples. ACKNOWLEDGMENT The authors gratefully acknowledge the financial support from Convergent Bioscience Ltd. and the Natural Sciences and Engineering Research Council of Canada (NSERC) and the donated sample of Teflon AF 2400 capillary from Dr. Ilia Koev at Biogeneral Inc. Z.L. gratefully thanks Dr. Jiaqi Wu and Mr. Tom Kerr at Convergent Bioscience Ltd. for their technical support and suggestions, Dr. Tiemin Huang for his invaluable technical assistance, and Dr. Carolyn Goodridge and Ms. Jennifer Cunliffe for their editorial assistance. Received for review May 31, 2003. Accepted July 11, 2003. AC034587M