INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Over the past decade, capillary electrophoresis (CE) has been developed into a standard technique in biotechnology. CE is used for analytical separation of charged solutes, such as carbohydrates, proteins, DNA, pharmaceuticals, herbicides, and pesticides, as well as lower-molecular-weight ions (1, 19, 27). CE is a powerful clinical diagnostic tool for profiling, screening, and detecting drugs, carbohydrates, lipids, enzymes, proteins, and nucleic acids (46). Variants of CE, known as micellar electrokinetic capillary chromatography (27) and capillary electrochromatography (36), allow solutions of nonionic analytes to be resolved.

Recently, CE has been applied to suspensions involving particles of biological interest, including viruses (19), bacteria (9, 19), silica sols (31), and polystyrene nanospheres (45). An advantage of electrophoretic separation compared with filtration or centrifugation is that electrophoresis is characteristically gentle and suitable for labile compounds and microorganisms. For example, Ebersole and McCormick (9) found that more than 90% of the bacteria that they tested remained viable after CE, and other workers (13, 35) have obtained similar viability results for cells separated dielectrophoretically in electric fields having comparable strengths (ca. 100 V/cm).

While the capabilities of CE to separate compounds and particles are well-established (for reviews see references 2, 26, and 27), the method has not been widely used to measure electrophoretic mobilities. Accurate measurements of electrophoretic mobilities are important in many biological and environmental sciences and technologies, ranging from clinical diagnostics (40, 41) to biocolloid adhesion (8, 10, 12, 17, 33, 39, 43, 44). Tobacco mosaic virus (6, 15) and polymer latex mobilities (6, 15, 23) have been characterized by CE, but Zhu and Chen (47) found that for human erythrocytes, the electrophoretic mobility measured by CE differed by 21% from previously published values. As there have been no direct comparisons between CE and other techniques, we determined electrophoretic mobilities by CE and by microelectrophoresis (ME), and we report the results obtained here. Our objectives were essentially to examine whether CE could be used to measure the electrophoretic mobility of larger particles (diameter, ca. 1 µm) of biological interest and to assess the accuracy of the CE method compared with ME (8, 29, 42). We present results for the electrophoretic mobilities of three strains of bacteria and three chemically distinct microspheres (used in biocompatibility determinations) at two pHs and three ionic strengths. The method that we used was based on the work done with tobacco mosaic virus by Grossman and Soane (15). Differences between CE and ME mobility measurements were not statistically significant, and the CE data typically showed lower variances than the ME data. In addition, it was possible by using CE to detect electrophoretic heterogeneities in two of the bacterial species, as well as aggregation in one of the latex suspensions. To our knowledge, this was the first use of CE to resolve mobility variations within a monoclonal population.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and preparation. The bacterial strains investigated in this study were A1264, a gram-negative, motile, ellipsoidal bacterium (ratio of length to width, 2.5) that was isolated from the Savannah River deep subsurface environment, and two strains isolated from the Department of Energy site at Oyster, Va. The Oyster groundwater isolates were CD1 (a gram-negative, nonmotile, ellipsoidal bacterium with a ratio of length to width of 1.5) and PL2W31 (a gram-positive, nonmotile, ellipsoidal bacterium with a ratio of length to width of 2.5). The equivalent spherical radii of the bacteria used, based on their projected areas, were all approximately 0.5 µm. Cultures were grown to the early stationary phase (absorbance at 600 nm, approximately 1.5) in 250-ml Erlenmeyer flasks containing 100 ml of 10% PTYG (0.25 g of peptone per liter, 0.25 g of tryptone per liter, 0.50 g of yeast extract per liter, 0.6 g of MgSO₄ · 7H₂O per liter, 0.07 g of CaCl₂ · 2H₂O per liter) at room temperature (22 to 24°C) with constant agitation (150 rpm). The cells were separated by centrifugation (20 min, 10⁴ × g) and resuspended in 100 ml of filter-sterilized 3-(N-morpholino)propanesulfonic acid (MOPS) buffer containing 4.186 g of MOPS (Sigma; titrated to pH 7.02 with 1 N NaOH; conductivity, 760 µS/cm; ionic strength, 10 mM) per liter. The washing procedure was repeated to produce each final suspension (concentration, 10⁹ to 10¹⁰ cells/ml), which was then incubated at room temperature for 18 h. (The bacterial species and stage of growth were predicated on the results of a related bacterial attachment study (3); the CE technique which we describe is robust and adaptable to species at all stages of growth.) At that point, cell density was measured by acridine orange direct counting (20). Throughout the process, culture purity was repeatedly verified by direct visual observation (approximately 10⁴ cells were observed), plating, and comparisons of colony morphology. Direct visual observation was also used to confirm that the cells were monodisperse at the time of the experiments.

Microspheres and preparation. The following three types of microspheres were studied: 0.6-µm-diameter, surfactant-free, hydrophobic amidine polystyrene latex microspheres (Interfacial Dynamics Corp., Portland, Oreg.); 0.22-µm-diameter, surfactant-free, hydrophilic carboxylate-modified polystyrene latex (CML) microspheres (Interfacial Dynamics Corp.); and 0.24-µm-diameter, hydrophobic polymethyl methacrylate (PMMA) microspheres (NASA Space Science Lab, Marshall Space Flight Center, Huntsville, Ala.). The hydrophobicities reported below are the hydrophobicities under the test conditions used in this study. The polystyrene and PMMA spheres were supplied at stock concentrations equivalent to roughly 4 and 20% solids, respectively. The stock solutions were centrifuged for 4 to 16 min, depending on the size of the microspheres, resuspended in the appropriate carrier buffers, and sonicated for at least 10 min to produce monodisperse suspensions. The carrier buffers used were 100 mM borate (Beckman borate buffer kit; pH 8.39; conductivity, 1.60 mS/cm) and 2 mM borate (pH 8.35; conductivity, 41.3 µS/cm). The 2 mM borate buffer was produced by diluting a 100 mM buffer solution with ultrapure water (Ultra-Pure Water Research Group, Department of Chemical and Environmental Engineering, The University of Arizona). The buffers were filter sterilized. The washing procedure was repeated three times for the CML and amidine particles and 6 to 10 times for the PMMA particles since the PMMA stock solutions contained residual impurities from the emulsion polymerization process. The suspensions were diluted to final volume fractions of 0.01 to 0.001 and stored overnight at 4°C to ensure that the spheres were in equilibrium with the carrier buffers. The amidine and CML particle measurements were made in the 2 and 100 mM borate buffers. The PMMA measurements were made in the 2 mM borate buffer. Prior to measurements, the suspensions were inspected for aggregation with a light microscope.

CE measurements. A Beckman P/ACE System 2100 capillary zone electrophoresis unit, controlled by Beckman System Gold software, was used to measure electrophoretic mobilities. A schematic diagram of the instrument is shown in Fig. 1. Application of an electric field down the axis of the capillary drove particle transport by the following two mechanisms: electrophoresis (translation due to the particles' native charge) and electroosmosis (bulk convection of the carrier electrolyte in which the particles were immersed). M, the electrophoretic mobility of a particle population (or subpopulation), follows from the translation rate compared with the translation rate of a neutral dopant, which moves from the inlet past the UV detector at the electroosmotic velocity of the buffer (21, 28): M = (L_totL_d/V) (1/t_p  1/t_ref), where L_d is the distance from the inlet reservoir to the detector, V is the voltage difference applied across the length of the capillary (L_tot), and t_ref and t_p are the mean migration times of the reference marker and the particle population (or subpopulation), respectively (27).

View larger version (28K): [in this window] [in a new window]   FIG. 1.   Schematic diagram of the CE apparatus.

12

ME measurements. ME electrophoretic mobilities were measured with a Lazer Zee model 501 meter (PenKem, Inc., Bedford Hills, N.Y.). Measurements were made as recommended in the user's manual, at room temperature. The model 501 meter did not have a thermostat and reported the data as a  potential, which could be converted to electrophoretic mobility by using the Smoluchowski equation (37): M = (₀/µ), where µ is the buffer viscosity,  is the relative permittivity of the buffer, and ₀ is the permittivity of a vacuum (8.854 × 10¹² C²/N · m²). The parameter values internal to the model 501 meter were reference values for water at 20°C. Measurements were repeated at least 10 times for each particle.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacteria. Capillary electropherograms for the carrier buffer and the bacterial filtrate exhibited no electrophoretic peaks. The strength of the electroosmotic flow (EOF) was 7.13 ± 0.55 µm · cm/V · s (n = 36). The marker did not affect the magnitude or distribution of the electrophoretic mobilities for the bacteria. Preliminary experiments showed that the results were also independent of the injection technique employed.

View larger version (12K): [in this window] [in a new window]   FIG. 2.   Typical electropherograms for bacterial cells. Measurements were obtained in 10 mM MOPS buffer with an injection interval of 5 s. Unless indicated otherwise, an electric field of 175 V/cm was applied over a 57-cm column. (a) Neutral marker peak (2.2 min) and two bacterial peaks (2.8 and 3.3 min) for strain A1264. An electric field of 426 V/cm was applied over a 47-cm column. (b) Neutral marker peak (8.9 min) and two bacterial peaks (12.5 and 13.6 min) for CD1. The column length was 77 cm. (c) Neutral marker peak (7.1 min) and a single bacterial peak (7.7 min) for PL2W31.

Microspheres. Light microscopy of the microspheres revealed monodisperse suspensions except for the PMMA particles, which exhibited significant aggregation. No electrophoretic peaks were observed with the baseline. The strength of the EOF was 10.77 ± 0.17 µm · cm/V · s (n = 9) in 2 mM borate buffer and 6.60 ± 0.18 µm · cm/V · s (n = 8) in 100 mM borate buffer. The marker had no effect on the magnitude or distribution of the electrophoretic mobilities for the latex spheres. Typical electropherograms for amidine and CML particles are shown in Fig. 3. The results of the CE and ME experiments are summarized in Table 2.

View larger version (12K): [in this window] [in a new window]   FIG. 3.   Typical microsphere electropherograms. Measurements were obtained in 100 mM borate buffer with an applied electric field of 350 V/cm over a 57-cm column. (a) Neutral marker peak (2.8 min) and a microsphere peak (3.5 min) for the amidine polystyrene microspheres. The injection interval was 10 s. (b) Neutral marker peak (2.8 min) and a microsphere peak (5.8 min) for the CML microspheres. The injection interval was 1 s.

View larger version (17K): [in this window] [in a new window]   FIG. 4.   Electropherograms for the PMMA suspension. Measurements were obtained in 2 mM borate buffer with an applied electric field of 175 V/cm over a 57-cm column. The injection interval was 5 s. (a) Neutral marker peak (4.2 min) and two microsphere peaks (7.0 and 7.7 min) for the polydisperse PMMA particles (prior to sonication). (b) Evolution of bimodal electrophoretic mobility in PMMA suspensions. Electropherograms based on suspensions were obtained immediately after sonication (dashed line) and 2.5 h after sonication (solid line).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The general agreement between the data obtained with the CE and ME mobility measurement techniques is indicated in Fig. 5. Statistical differences in the mean electrophoretic mobilities determined by CE and ME were investigated by using a paired t test (P = 0.40) (7), a sign test (P = 0.50), and a Wilcoxon signed rank test (P = 0.47) (14); differences between the means were insignificant. Systematic differences associated with the choice of colloid, pH, or ionic strength were not observed.

View larger version (21K): [in this window] [in a new window]   FIG. 5.   Comparison of the electrophoretic mobilities determined by CE (MCE) and ME (MME). The error bars are 2 standard deviations wide. The diagonal is a 1:1 line representing complete data agreement. The data are labeled according to sample type.

The finding that CE results agree with ME results is not surprising. Electrokinetic theory shows that, as long as the particle size is no more than a few percent of the capillary diameter, the influence of hydrodynamic interactions on mobility measurements is negligible (25). Colloidal (e.g., van der Waals) and/or hydrophobic interactions between the analyte and the capillary wall can also bias mobility measurements. Such interactions are well-documented for proteins and other organic analytes (18, 24, 30), which tend to adsorb to quartz capillaries. Similar effects have been observed for 0.2-µm-diameter hydrophobic polystyrene and PMMA particles in capillaries with inside diameters of 25 µm (6). Analyte-wall interactions can be diminished by using capillaries with larger cross-sectional areas (5, 6), by modifying the carrier buffer, or by using coated capillaries (34). Coatings that reduce wall interactions, however, often reduce EOF and increase analysis time.

Based on similar numbers of replicate measurements, the standard deviations for CE results were typically much lower than the standard deviations for ME measurements (Table 2). This was probably because a single CE mobility measurement was based on (i) analysis of a relatively large number of particles (approximately 10⁴ particles) and (ii) the time required for the analyte to move distances on the order of 0.1 m in an electric field stronger than 100 V/cm. By contrast, ME and related measurement techniques (22) detect particle translation over short distances (typically less than 100 µm) in weak fields (10 V/cm or less) and often rely on observation of far fewer particles.

Two of the bacteria used, A1264 and CD1, and one of the microsphere types used, PMMA, exhibited multimodal behavior during CE measurements. Similar distributions were not obtained with ME. CE is a sensitive analytical separation technique tailored to resolve mixtures of charged solutes on the basis of their electrophoretic mobilities. The sensitivity stems from the strength of the applied field and the uniformity of the particle velocities over the capillary cross section. Thus, when a large number of particles is sampled, an electrophoretic histogram and a more detailed picture are obtained.

The results of related studies on bacterial sorption (3) suggest that the bimodal electrophoretic mobilities of the CD1 and A1264 populations were due to intrapopulation heterogeneities in surface charge rather than aggregation, contamination, or artifactual peak splitting. Bimodal mobilities of cells have been observed in other contexts, and in these cases the bimodality has been attributed to variations in surface structure (the presence or absence of appendages or capsular material) (8) and surface antigens (38). Ebersole and McCormick (9) found that differences in chain morphology can cause multimodal behavior in bacterial electrophoretic mobility. Visual inspection confirmed that the cells used here were monodisperse. Spread plating revealed no major contaminant. The detection limit for bacteria was approximately 2,500 cells or roughly 1/10th the standard number of cells injected. The presence of a significant contaminant (observable with CE as a separate peak) should have been recognized by direct visual observation or by differences in colony morphology following growth on solid media. The multiple peaks were in no way due to the carrier buffer, cell exudates, or mesityl oxide. Ermakov et al. (11) concluded (based on both theory and experiments) that artifactual peak splitting is avoided by buffering solutions at least 1 pH unit from the particle isoelectric point. Since bacterial isoelectric points are generally in the range from 2 to 4, depending on species and growth stage (16), and since the pH of the MOPS buffer was 7.02, artifactual peak splitting is an unlikely source of multimodal mobilities observed.

The PMMA microspheres also exhibited an electrophoretic mobility distribution. Direct visual observation revealed both individual particles and aggregated clumps in suspensions used for the original CE and ME measurements. Sonication subsequently produced a monodisperse suspension of PMMA microspheres that yielded a single peak in the corresponding electropherograms. Subsequent CE measurements obtained with the same suspension revealed the time-dependent evolution of a second peak. Within 2.5 h, the suspension was comprised of mainly aggregated particles, as indicated by visual inspection. Over the same period the area of the peak containing the monodisperse microspheres decreased, while the second peak grew, presumably due to aggregation.

Despite the noticeable difference in the areas under the peaks in their respective electropherograms, the numbers of amidine and CML (Fig. 3) particles injected were actually similar. A likely explanation for this phenomenon is the differences in the specific absorbances of the particles, as the area of an electrophoretic peak is a function of both the number of particles injected and the specific absorbance of the particles. Since both colloid volume fractions (less than 0.01 for both microorganisms and microspheres) and UV absorbance (typically less than 0.1 absorbance unit) were low, detection artifacts associated with high particle concentrations are unlikely.

In the CE measurements, electroosmotic flow was characterized by doping the samples with a neutral marker, mesityl oxide. EOF measurements were reproducible for all of the carrier buffers tested. The standard deviation of the EOF in 10 mM MOPS buffer (pH 7.02) was roughly three times the standard deviation of the values obtained at both ionic strengths in the borate buffer (pH 8.35). This probably was due to the neutral pH of MOPS solutions, since electroosmosis in uncoated capillaries is extremely sensitive to pH changes in the range from pH 4 to 8 (27).

CD1 samples were injected both electrokinetically and by high pressure. Electrokinetic injections have been known to bias samples (27), since analytes with lower (more negative) electrophoretic mobilities tend to remain in the sample vial. No difference in either the distribution or the magnitude of the electrophoretic mobilities exhibited by CD1 was observed in response to the change in injection technique (data not shown).

CD1 exhibited multimodal behavior in electrophoretic mobility at relatively low field strengths (175 V/cm). Higher applied electric fields were required for A1264 because at low field strengths, microbial motility distorted the CE peaks. Attempts to uncover multimodal electrophoretic behavior in bacterial strain PL2W31 by increasing the applied electric field to approximately 350 V/cm were unsuccessful.

The reproducibility of the CE data allowed batch-to-batch variations to be studied in the bacteria. A1264 and CD1 showed variations in the distributions of their populations between the respective mobility modes, although this had only a minor effect on the average mobilities of the populations. PL2W31 exhibited batch-to-batch variability in its (unimodal) mean electrophoretic mobility. Investigators should be able to use CE to help assess the influence of such variability on biophysical processes (e.g., attachment to surfaces) or, alternatively, to help evaluate connections between variations in surface charge density and outer membrane structure and composition.

In summary, CE measurements of electrophoretic mobility were not distinguishable from measurements of electrophoretic mobility obtained by ME. Based on similar numbers of replicate measurements, the standard deviations for CE results were typically much lower than the standard deviations for ME measurements. Distributed mobilities, which were readily observed with CE for two bacterial species and one of the polymer latices, could not be detected with ME.