Anal. Chem. 2005, 77, 7282-7287
Analysis of Lipoproteins by Microchip Electrophoresis with High Speed and High Reproducibility Guichen Ping,*,†,‡,§ Bingmei Zhu,‡,§ Mohammad Jabasini,‡,§ Feng Xu,‡ Hiroaki Oka,† Hirokazu Sugihara,† and Yoshinobu Baba‡,§,|
Nanotechnology Laboratory, Advanced Technology Research Laboratories, Matsushita Electric Industrial Co. Ltd., 3-4 Hikari-dai, Seika, Soraku, Kyoto 619-0237, Japan, Department of Molecular and Pharmaceutical Biotechnology, Graduate School of Pharmaceutical Sciences, The University of Tokushima, 1-78 Shomachi, Tokushima 770-8505, Japan, Department of Applied Chemistry, Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan, and Single-molecule Bioanalysis Laboratory, National Institute of Advanced Industrial Science and Technology (AIST), 2217-14 Hayashicho, Takamatsu 761-0395, Japan
A method for the fast analysis of lipoproteins by microchip electrophoresis with light-emitting diode confocal fluorescence detection has been developed. Lipoproteins labeled with BODIPY FL C5-ceramide are found to strongly adsorb on the bare surface of a poly(methyl methacrylate) (PMMA) microchip. Sodium dodecyl sulfate and cetyltrimethylammonium bromide were therefore utilized to alter lipoproteins and channel surface to make them bear the same type of charge. After modification, the peak shape of lipoproteins was greatly improved, demonstrating lipoprotein adsorption on a PMMA chip dramatically reduced due to electrostatic repulsion. In addition, polymers were added into the running buffer to suppress electroosmotic flow and to serve as a sieving matrix. As a result, lipoprotein separation was manipulated by both electrophoretic mobilities and particle sizes. Various separation parameters including surfactant concentration, buffer pH, and polymer concentration as well as on-line concentration were investigated systematically. Under optimal conditions, two baseline separations of standard lipoproteins including high-density lipoprotein, low-density lipoprotein, and very low-density lipoprotein were achieved with different selectivity. This method affords high separation speed (within 100 s) and high reproducibility. The intraassay and interassay RSDs of lipoprotein migration times were in the range of 0.901.9%, indicating this method is highly reliable. Lipoproteins, nanometer-sized particles, are macromolecular complexes of lipids and globular proteins held together by hydrophobic interaction and electrostatic attraction.1 The primary function of lipoproteins is to transport lipids through vascular and * Corresponding author. E-mail:
[email protected]. Tel/Fax: (52) 789-4666. † Matsushita Electric Co. Ltd. ‡ The University of Tokushima. § Nagoya University. | AIST. (1) Daykin, C. A.; Corcoran, O.; Hansen, S. H.; Børndottir, I.; Cornett, C.; Connor, S. C.; Lindon, J. C.; Nicholson, J. K. Anal. Chem. 2001, 73, 1084-1090.
7282 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
extravascular body fluids. Disorders in lipoprotein metabolism are critical to the development of atherosclersis, coronary heart artery disease, liver dysfunction, and cancer. Therefore, lipoprotein analysis in serum has been one of the most important and ubiquitous clinical measurements. In general, lipoproteins are classified into high-density lipoprotein (HDL), low-density lipoprotein (LDL), and very low-density lipoprotein (VLDL) based on isolation by ultracentrifugation. At present, commonly used methods for lipoprotein analysis include sequential and gradient ultracentrifugation,2-5 specific enzyme assay,6 selective precipitation,7,8 and size exclusion chromatography.9 However, all these methods are labor-intensive, time-consuming, or both. With the advent of capillary electrophoresis (CE), it has been demonstrated to be a powerful tool for the analysis of biopolymers.10,11 Capillary zone electrophoresis (CZE)12-14 and capillary isotachophoresis15-20 have been utilized to profile lipoproteins, (2) Gofman, J. W.; Lindgren, F. T.; Elliot, H. J. Biol. Chem. 1949, 179, 973979. (3) Groot, P. H. E.; Scheek, L. M.; Havekes, L.; van Noort, W. L.; van’t Hoof, F. M. J. Lipid Res. 1982, 23, 1342-1353. (4) Patsch, J. R.; Patsch, W. Methods Enzymol. 1986, 129, 3-26. (5) Kahlon, T. S.; Glines, L. A.; Lindgren, F. T. Methods Enzymol. 1986, 129, 26-45. (6) Friedewald, W. T.; Levy, R. I.; Fredrison, D. S. Clin. Chem. 1972, 18, 499502. (7) Burstein, M.; Scholnick, H. R.; Morfin, R. J. Lipid Res. 1970, 11, 583-595. (8) Lopes-Virella, M. F.; Stone, P.; Ellis, S.; Colwell, J. F. Clin. Chem. 1977, 23, 882-884. (9) Weiller, B. H.; Ceriotti, L.; Shibata, T.; Rein, D.; Roberts, M. A.; Lichtenberg, J.; German, J. B.; de Rooij, N. F.; Verpoorte, E. Anal. Chem. 2002, 74, 17021711. (10) Hu, S.; Dovichi, N. J. Anal. Chem. 2002, 74, 2833-2850. (11) Quigley, W. W. C.; Dovichi, N. J. Anal. Chem. 2004, 76, 4645-4658. (12) Hu, A. Z.; Gruzado, I. D.; Hill, J. W.; McNeal, C. J.; Macfarlane, R. D. J. Chromatogr., A 1995, 717, 33-39. (13) Stocks, J.; Miller, N. E. J. Lipid Res. 1998, 39, 1305-1309. (14) Macfarlane, R. D.; Bondarenko, P. V.; Cockrill, S. L.; Cruzado, I. D.; Koss, W.; NcNeal, C. J.; Spiekerman, A. M.; Watkins, L. K. Electrophoresis 1997, 18, 1796-1806. (15) Bo ¨ttcher, A.; Schlosser, J.; Kronenber, F.; Dielinger, H.; Knipping, G.; Lackner, K. J.; Schmitz, G. J. Lipid Res. 2000, 41, 905-915. (16) Zorn, U.; Haug, C.; Celik, E.; Wennauer, R.; Schimid-Kotsas, A.; Bachem, M. G.; Gru ¨ nert, A. Electrophoresis 2001, 22, 1143-1149. (17) Zorn, U.; Wofl, C.; Wennauer, R.; Bachem, M. G.; Gru ¨ nert, A. Electrophoresis 1999, 20, 1619-1626. 10.1021/ac050896w CCC: $30.25
© 2005 American Chemical Society Published on Web 10/21/2005
respectively. Stocks and Miller demonstrated the analysis of LDL by CZE using methylglucamine as a dynamic coating.13 Furthermore, Watkins et al. added 0.5 mM sodium dodecyl sulfate (SDS) into both sample solution and running buffer to measure single lipoprotein samples such as HDL, LDL, and VLDL in CZE.14 Schmitz et al. developed a capillary isotachophoresis procedure for lipoprotein separation and identified 14 subpopulations in lipoprotein sample.18-20 However, up to nine spacers are required to be added into running, leading, and terminating buffers, making this method complicated and less practical. Microchip electrophoresis is a rapidly developing analytical technology that has attracted great attention in recent years due to its substantial advantages over conventional analytical technologies including high speed, low consumption of sample and buffer, easy integration, and miniaturization. Separation performance of microchip electrophoresis is basically similar to that of conventional CE, but microchip electrophoresis can be completed on a time scale of seconds.21-23 Recently Verpoorte et al. utilized microchip electrophoresis to separate HDL and LDL.9,24 The authors tried to transfer reported separation condition to microchip electrophoresis.13 Unfortunately, HDL and LDL were not baseline resolved and the method suffered from poor reproducibility, which shows, as a coating material, methylglucamine cannot effectively suppress lipoprotein adsorption on a separation channel. Hence, it is necessary to find a more suitable method for lipoprotein analysis with enough resolution and high reproducibility. In this paper, we choose SDS and cetyltrimethylammonium bromide (CTAB) as anionic and cationic surfactants to simultaneously modify lipoproteins and the channel surface. As surfactant alkyl chains adsorb on the inner surface of the poly(methyl methacrylate) (PMMA) chip and the lipoproteins form complexes with the surfactants, both the lipoproteins and channel surface are either negatively or positively charged. As a result, strong lipoprotein adsorption on the PMMA chip would largely reduce due to electrostatic repulsion. Furthermore, taking into account the fact that the lipoproteins differ in size, polymers are added into the running buffer to serve as sieving matrix and to regulate separation selectivity. EXPERIMENTAL SECTION Reagents and Buffer Solutions. The solutions of HDL, LDL, and VLDL were obtained from Sigma (St. Louis, MO). According to product specifications, the original concentrations of HDL, LDL, and VLDL were 8.70, 3.67, and 1.68 mg/mL, respectively. BODIPY FL C5-ceramide was ordered from Molecular Probes (Eugene, OR). Methylcellulose (MC), hydroxypropylmethylcellulose (HPMC) (viscosity of 2% aqueous solution at 20 °C, 4000 cP), 2-(4morpholino)ethanesulfonic acid (MES), tricine, 3-morpholinopropanesulfonic acid (MOPS), ethylene glycol, and methylglucamine were purchased from Sigma. Dimethyl sulfoxide (DMSO) and (18) Schmitz, G.; Mo ¨llers, C.; Richter, V. Electrophoresis 1997, 18, 1807-1813. (19) Schmitz, G.; Mo ¨llers, C. Electrophoresis 1994, 15, 31-39. (20) Schmitz, G.; Borgman, U.; Assman, G. J. Chromatogr. 1985, 320, 253262. (21) Reyes, D. R.; Iossifidis, D.; Auroux, P. A.; Manz, A. Anal. Chem. 2002, 74, 2623-2636. (22) Auroux, P. A.; Reyes, D. R.; Iossifidis, D.; Manz, A. Anal. Chem. 2002, 74, 2637-2652. (23) Vikner, T.; Janasek, D.; Manz, A. Anal. Chem. 2004, 76, 3373-3386. (24) Ceriotti, L.; Shibata, T.; Folmer, B.; Weiller, B. H.; Roberts, M. A.; de Rooij, N. F.; Verpoorte, E. Electrophoresis 2002, 23, 3615-3622.
SDS were obtained from Fluka (Steinheim, Germany). CTAB was acquired from Calbiochem (Darmstadt, Germany). Poly(ethylene oxide) (PEO) (Mv ∼4 000 000) was bought from Aldrich (Milwaukee, MI). Running buffers with polymer and surfactant additives were prepared by adding polymer into buffer salt solution containing various amounts of either SDS or CTAB and stirring slowly until the solution appeared homogeneous and transparent. Subsequently, buffers were adjusted to the desired pH using sodium hydroxide or hydrochloride. The specific concentrations of polymer and surfactant for each measurement are given in the caption of the corresponding figures. Double-deionized water was used for buffer and sample preparation. Derivatization Procedure. Prior to separation, the lipoproteins were derivatized with BODIPY FL C5-ceramide, a lipophilic fluorescence dye. Fluorescence stock solution was prepared by dissolving 250 µg of BODIPY FL C5-ceramide in 50 µL of DMSO and diluting with 450 µL of ethylene glycol. The derivatization procedure was as follows: 10 µL of lipoprotein was diluted with 70 µL of water, then 20 µL of fluorescence stock solution was added, followed by vortexing for 1 min; finally, the above solution was diluted to 800 µL with water. Sample solution was made by mixing the same aliquot of each lipoprotein derivatization solution. The final concentrations of HDL, LDL, and VLDL in the sample solution were 36.3, 15.3, and 7.00 µg/mL, respectively. Instruments. All experiments were carried out on a Hitachi SV 1100 microchip electrophoresis instrument with a light-emitting diode (LED) confocal fluorescence detector and an external power supply capable of providing voltage ranging from 0 to 5000 V (Hitachi Electronics Engineering, Tokyo, Japan). Data acquisition and analysis were performed using the software supplied with the system. A blue LED with a median excitation wavelength of 470 nm was used as excitation source. The fluorescence was collected with condensing lens, spectrally filtered by a beam splitter (transmission >530 nm) and an emission filter (transmission >580 nm), and then detected by an avalanche photodiode (Hamamatsu Photonics, Hamamatsu, Japan). The electrophoresis chip specification has been demonstrated in our previous papers.25,26 For the separation of the lipoproteins incorporating with SDS, the sample was loaded by applying 300 V to the sample waste reservoir for 30 s while grounding the other reservoirs. During separation, the buffer waste reservoir was applied to 750 V and the buffer reservoir was grounded. Meanwhile, 130 V was applied to both sample and sample waste reservoirs. Since the lipoproteins incorporating with CTAB were positively charged, the aforementioned analysis procedure cannot be applied to the separation of such lipoproteins. Separation strategy was regulated as follows: first, the sample was loaded into the injection channel by applying vacuum to the sample waste reservoir over a period of 15 s by means of a syringe while leaving other reservoirs at ambient pressure; then separation started by applying 750 V to the buffer reservoir and grounding buffer waste reservoir. In the meantime, 590 V was applied to both sample and sample waste reservoirs. RESULTS AND DISCUSSION Surface Modification. One of the difficulties of lipoprotein analysis in CE is the high propensity to adsorb on the separation (25) Zhang, L.; Dang, F.; Baba, Y. Electrophoresis 2002, 23, 2341-2346. (26) Dang, F.; Zhang, L.; Hagiwara, Y.; Baba, Y. Electrophoresis 2003, 24, 714721.
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7283
Figure 1. Electrophoregrams of HDL: (A) 40 mM tricine, 40 mM methylglucamine pH 9.0; (B), 0.50% (w/w) HPMC in 20 mM borate, pH 9.0; (C), 0.50% (w/w) MC in 20 mM borate, pH 9.0. E ) 168 V/cm, Leff ) 3.0 cm.
channel, leading to peak broadening or even loss of peak. Methylglucamine has been utilized as coating material to diminish lipoprotein adsorption on the bare fused-silica capillary13 and borosilicate glass microchip.9,24 At the beginning of our experiment, we tested the applicability of using this reagent as coating material on the PMMA chip. As depicted in Figure 1, the HDL peak appears to be very broad even if there is 40 mM methylglucamine in the running buffer, which shows methylglucamine cannot effectively suppress HDL adsorption on a bare PMMA chip. Based on methylglucamine structure, it is a highly hydrophilic molecule; therefore, its interaction with a hydrophobic PMMA chip is too weak to reduce lipoprotein adsorption. In our previous paper, a series of coating materials including amines, surfactants, and neutral polymers were studied to minimize the strong adsorption of oligosaccharide labeled with 8-aminopyrene-1,3,6trisulfonate on a bare PMMA chip.26 Among these coating materials, MC and HPMC were found to be capable of efficiently suppressing the adsorption of the above analytes. Whereas, Figure 1 indicates the presence of as high as 0.50% MC or HPMC in the running buffer could not improve HDL peak shape, which further confirms the strong lipoprotein adsorption on a bare PMMA chip. The above results show the published experiment conditions cannot be transferred to our separation system directly. To suppress such strong analyte-wall interaction, we start to use either anionic or cationic surfactant to modify both lipoproteins and inner channel surface. In the case of SDS, it is well known that SDS molecules can easily form complexes with proteins. It is reasonable to deduce that SDS molecules can form complexes with lipoproteins due to the presence of apolipoproteins on the lipoprotein surface. The charge number of the lipoproteins increases as a result of such complexation. Meanwhile, PMMA is a relative hydrophobic material compared to the fused-silica capillary; thus, the alkyl chain of SDS can readily adsorb on the PMMA inner surface, leaving the sulfate group outside.27 Therefore, both lipoproteins and the inner channel surface are negatively charged. On the contrary, in the case of CTAB as coating material, (27) Nagata, H.; Tabuchi, M.; Hirano, K.; Baba, Y. Electrophoresis 2005, 26, 2247-2253.
7284 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
Figure 2. Separation of HDL and LDL. Sample solution contains 0.2 mM SDS; BGE, 0.050% HPMC and 0.3 mM SDS in 20 mM MOPS, pH 6.93.
the lipoproteins and the inner channel surface would bear positive charges. The adsorption could suppress due to electrostatic repulsion in both cases. However, since the lipoproteins incorporating with surfactants and the PMMA channel have the same type of charge, the direction of lipoprotein electrophoresis is opposite to that of electroosmotic flow (EOF). In our experiment, no peak was observed when the sample was analyzed directly, indicating the magnitude of EOF is comparable to or higher than that of lipoprotein electrophoretic velocity. To overcome this problem, it is necessary to suppress EOF in this system. It has been welldocumented that the addition of polymers into the running buffer can dramatically reduce EOF. After addition of 0.050% (w/w) polymer such as HPMC, MC, and PEO, the EOF marker was not detected within 20-min (maximum separation time permitted by the software), which demonstrates EOF has been substantially suppressed. Figure 2 shows HDL and LDL are readily baseline separated within 70 s with good peak shape when using SDS as the coating material and using HPMC to suppress EOF. The above separation result proves it is successful to diminish lipoprotein adsorption and to suppress EOF by utilizing our experiment strategy. Whereas, when we use CTAB, a cationic surfactant, to modify the lipoproteins and chip channel, multiple peaks were found for a single lipoprotein. For instance, Figure 3 shows HDL has as many as three peaks, which will make further lipoprotein separation complicated. Therefore, in subsequent studies, we will focus on utilizing SDS to modify the lipoproteins and chip channel. Effect of SDS Concentration. The effect of SDS concentration in running buffer on lipoprotein separation was investigated. The electrophoretic mobility of HDL was found to slightly increase with SDS concentration in the running buffer while those of LDL and VLDL kept nearly constant. It is largely because more SDS molecules incorporated into HDL when the SDS concentration was increased. This result is consistent with the fact that the apolipoprotein weight percentage of HDL is highest among these three lipoproteins; therefore, SDS becomes most highly prone to forming complexes with HDL. Moreover, any peak split of LDL and VLDL was not observed as shown in Figure 4 while HDL peak number increased to three (H1, H2, H3) as SDS concentration increased from 0.2 to 1.0 mM. We do not think such changes
Figure 3. Electrophoregrams of HDL, LDL, and VLDL. Sample solution contains 0.2 mM CTAB; BGE, 0.050% HPMC and 0.3 mM CTAB in 20 mM MES, pH 6.02, E ) 163V/cm.
Figure 4. Effect of SDS concentration in the running buffer on lipoprotein separation. Sample solution contains 0.3 mM SDS; BGE, 0.32% PEO in 40 mM MOPS, pH 6.86.
are due to delipidation since HDL is most resistant to delipidation among these three lipoproteins.14 It is most likely that these peaks correspond to HDL subpopulations. Verpoorte et al. observed a focusing effect for LDL when the sample solution contained 0.3 mM SDS while the running buffer was void of SDS.24 In our experiment, we did observe the same focusing effect for LDL when the SDS concentration in the running buffer was lower than or even close to that in the sample solution. However, with further increase in SDS concentration in the running buffer, such a focusing effect started to gradually disappear. Interestingly, we did not find such a focusing effect for HDL and VLDL. This focusing effect mechanism has not been clear until now and needs further investigation. We speculated that at least field-amplified sample stacking was involved in this focusing process. Watkins et al. observed the significant changes in lipoprotein density when 0.5 mM SDS was present in the sample solution by density analysis,14 while Verpoorte et al. confirmed that the presence of 0.3 mM SDS in the sample solution did not disrupt
Figure 5. Influence of buffer pH on lipoprotein separation. 0.2 mM SDS in sample solution; BGE, 0.050% HPMC and 0.3 mM SDS. Buffer salt concentration is 20 mM.
lipoprotein structure by laser light scattering analysis.24 Taking into account the fact that the higher the SDS concentration in the running buffer, the higher the likelihood of lipoprotein delipidation, hence, 0.3 mM SDS was chosen in the following separation. Additionally, lipoprotein peak height appears to decrease with the increase in SDS concentration in the running buffer. It might be because some fluorescence dyes incorporating with the lipoproteins are displaced with SDS molecules. Hence, keeping the SDS concentration in the running buffer low helps to maintain high detection sensitivity and lipoprotein micellike structure. Effect of pH Value. In general, buffer pH affects not only the magnitude of EOF mobility but also that of electrophoretic mobility of an analyte with ionizable groups. In this separation system, the effect of buffer pH on the magnitude of EOF mobility can be neglected since the EOF has been suppressed completely due to the presence of polymer in the running buffer. Therefore, lipoprotein migration is solely driven by the electrophoretic mobilities. As indicated in Figure 5, the elution order is HDL, VLDL, and LDL, which is different from that in the literature, where LDL was first eluted followed by HDL.9 This was because HDL and LDL were driven by both electrophoretic and EOF mobilities under the condition in the literature. Meanwhile, the direction of lipoprotein electrophoresis was against that of EOF and the magnitude of EOF mobility was larger than that of lipoprotein electrophoretic mobilities. Accordingly, HDL and LDL eventually migrated with the EOF. The elution order in the literature shows the electrophoretic mobility of HDL was higher than that of LDL, which agrees well with our results. Richter et al.18 presented the same lipoprotein elution order as ours in an EOF-free system. In addition, Figure 5 shows the variation in buffer pH does not have a significant effect on HDL migration time, while the migration times of LDL and VLDL increase with buffer pH, meaning LDL and VLDL electrophoretic mobilities decrease with an increase in pH. Generally, the extent of apolipoprotein dissociation increases with buffer pH; thus, the mobilities of LDL and VDL are expected to increase correspondingly. However, the Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7285
Figure 6. Baseline separation of lipoproteins. BGE, 0.27% PEO and 0.3 mM SDS in 50 mM MOPS, pH 6.93.
experiment result above is contradictory to our expectation. In our separation system, lipoprotein mobilities are actually affected by both dissociation degree of the apolipoproteins and the amount of SDS molecules incorporating into the lipoproteins. The latter is related to apolipoprotein weight percentage and buffer pH. Although apolipoprotein dissociation degree decreases with the reduction in buffer pH, it is most likely that the lipoproteins simultaneously incorporate with more SDS molecules. These additional SDS molecules may offset the loss of charges due to the reduction in apolipoprotein dissociation and even may increase the net charge number of the lipoproteins. As a consequence, the mobilities of LDL and VLDL eventually increase with the decrease in buffer pH. Effect of Polymer Concentration. When polymer concentration is relatively low in the running buffer, the separation is mainly based on lipoprotein electrophoretic mobilities. Figure 6 illustrates three lipoproteins are baseline resolved within 100 s when using 0.27% PEO in the running buffer. Here the main role of PEO is to suppress EOF. As we know, one of the distinguished differences in these lipoproteins is particle size. Generally, the LDL particle is about 2 times larger than HDL, and the VLDL particle is 15 times larger than LDL.14 With a gradual increase in polymer concentration in the running buffer, the sieving mechanism starts to play a role in the separation and VLDL migration will be greatly retarded by the polymer network. Figure 7 demonstrates the dramatic change in lipoprotein separation with increasing polymer concentration and the separation mechanism undergo variation from electrophoretic mobility dominated to polymer sieving dominated. In the electrophoretic mobility dominated region, the elution order is HDL, VLDL, and LDL. Further increase in polymer concentration makes the polymer entangled and the larger lipoprotein is more retarded with respect to the smaller one. As a result, it eventually leads to the reversal of elution order of LDL and VLDL, and the final elution order becomes HDL, LDL, and VLDL. When the HPMC concentration reaches 0.80% (w/w), the second baseline separation is achieved within 320 s. Under this condition, the sieving matrix plays a dominant role in the separation. To our best knowledge, it is the first time that separation of the lipoproteins in polymer solution on the basis of their different particle sizes and baseline separation by microchip electrophoresis have been achieved. The above results demon7286 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
Figure 7. Effect of HPMC concentration on lipoprotein separation. 0.2 mM SDS in sample solution; BGE, 0.3 mM SDS in 20 mM MOPS, pH, 6.93 with various concentrations of HPMC.
Figure 8. Ferguson plots for HDL and LDL.
strate the addition of polymers in the running buffer provides an additional factor to tailor resolution and selectivity of lipoprotein separation. As shown in Figure 8, the logarithm of lipoprotein mobilities reduces linearly with HPMC concentration within the range from 0.05 to 0.60%, which depicts that migration behaviors of HDL and LDL are subject to the Ogston model under this condition, whereas with the further increase in polymer concentration, the plot slopes change twice. Meanwhile, Figure 8 also shows the difference in HDL and LDL mobilities is enhanced with polymer concentration. Hence, the resolution of HDL and LDL can be regulated flexibly by altering polymer concentration. However, it should be pointed out VLDL is found to be a broad peak at high polymer concentration. It is because the mesh size of polymer network decreases with an increase in polymer concentration. As a result, it becomes more difficult for VLDL, the largest lipoprotein among these three lipoproteins, to go through the pore of entangled polymer network. On-Line Concentration. In our previous papers, we have demonstrated high on-line concentration capability of the sample stacking technique in an electroseparation mode.28,29 In this
Table 1. Precision of Lipoprotein Migration Time (% RSD, n ) 10) first separationa
intraassay interassay
second separationa
HDL
LDL
VLDL
HDL
LDL
0.92 1.7
0.76 1.5
1.0 1.8
0.59 1.9
1.0 1.9
a First and second separations were presented in Figures 6 and 7, respectively. For the second separation, 0.80% HPMC was used.
Figure 9. On-line concentration of lipoproteins. 0.2 mM SDS in sample solution. BGE, 0.27% PEO, 0.3 mM SDS, MOPS at various concentrations, pH 6.93.
method, a sample is dissolved into a solution with lower ionic strength than that of the running buffer. Electric field strength in the sample plug is thus higher than that in the buffer zone. Accordingly, charged analytes move faster in the sample plug. Once the analytes move into the buffer zone, migration velocity of the analytes dramatically decreases and they are hence concentrated at the boundary of the sample plug and the buffer zone. The ease of operation of this technique makes it very practical in improving detection sensitivity. It can be seen from Figure 9 that the detection sensitivity of lipoproteins could be improved by a factor of 5 in terms of peak height of VLDL when buffer salt concentration is increased from 20 to 50 mM. However, the peak height of lipoproteins dramatically decreases when buffer concentration reaches 60 mM, demonstrating sample stacking capability reaches a maximum in this system at 50 mM buffer concentration. In addition, lipoprotein peak shape was found to be improved with the increase in buffer concentration. It is because the increase in buffer concentration is helpful to further reduce analyte-wall interaction. Reproducibility. Reproducibility is a very important factor during method development. As shown in Table 1, intraassay and interassay reproducibilities are found to be less than 1.0 and 1.9% (28) Ping, G.; Zhang, Y.; Zhang, W.; Zhang, L.; Zhang, L.; Schmitt-Kopplin, P.; Kettrup, A. Electrophoresis 2004, 25, 421-427. (29) Zhang, Y.; Zhu, J.; Zhang, L.; Zhang, W. Anal. Chem. 2000, 72, 57445747.
RSD (n ) 10) for both separations. Since it is difficult to determine the exact migration time of VLDL in the second separation, the RSD of VLDL migration time for the second separation is not given. These results are notably improved compared to those in the literature,4 which proves this method is highly reliable. It is also feasible and successful to use surfactants to modify both lipoproteins and a separation channel to minimize strong lipoprotein adsorption and to use polymers as a sieving matrix to regulate the selectivity of lipoprotein separation. Further studies on blood sample pretreatment, quantitation curve establishment, blood sample analysis, and integration of all experimental steps into a chip are under investigation in our group. CONCLUSIONS Two baseline separations of three lipoproteins were realized by microchip electrophoresis with high speed and high reproducibility, using surfactants to modify lipoproteins a and separation channel as well as using polymers to suppress EOF and serve as a sieving matrix. Lipoprotein adsorption was efficiently suppressed due to electrostatic repulsion, and peak shape of the lipoproteins was thus greatly improved. Furthermore, with the change in separation mechanism from electrophoretic mobility dominated to polymer sieving dominated, the migration order of LDL and VLDL reversed. It shows this method has the ability to regulate the resolution and selectivity of lipoprotein separation. In addition, detection sensitivity was readily improved 5 times in terms of VLDL peak height by simply increasing the ionic strength of the running buffer. ACKNOWLEDGMENT This work was financially supported by New Energy and Industrial Technology Development Organization (NEDO) of the Ministry of Economy, Trade and Industry, Japan. G.P. thanks Dr. Tabuchi for her kind help. Received for review May 21, 2005. Accepted September 25, 2005. AC050896W
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7287
