Suppression of Electroosmotic Flow and Prevention of Wall Adsorption

Sep 1, 1997 - Addition of 2 mM CAS U to the electrophoretic buffer prevents adsorption of cationic proteins lysozyme, α-chymotrypsinogen A, cytochrom...
0 downloads 5 Views 194KB Size
Download PDF
Anal. Chem. 1997, 69, 3435-3441

Suppression of Electroosmotic Flow and Prevention of Wall Adsorption in Capillary Zone Electrophoresis Using Zwitterionic Surfactants Ken K.-C. Yeung and Charles A. Lucy*

Department of Chemistry, The University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada T2N 1N4

Addition of zwitterionic surfactants such as dodecyldimethyl(3-sulfopropyl)ammonium hydroxide, hexadecyldimethyl(3-sulfopropyl)ammonium hydroxide, and coco (amidopropyl)hydroxyldimethylsulfobetaine (Rewoteric AM CAS U) to an electrophoretic buffer suppress the electroosmotic flow by 50-90%. Onset of suppression occurs at approximately the critical micelle concentration of the surfactant. CAS U effectively suppresses the electroosmotic flow over the pH range 3-12. Addition of 2 mM CAS U to the electrophoretic buffer prevents adsorption of cationic proteins lysozyme, r-chymotrypsinogen A, cytochrome c, and ribonuclease A. Migration time reproducibility for these proteins is ∼1% RSD within 1 day and 2-5% from day to day. Efficiencies in excess of 750 000 plates/m and recoveries of >80% were observed for protein injections from distilled water. Alternatively if 2 mM CAS U is added to samples, recoveries were quantitative, although efficiencies decreased to 325 000600 000 plates/m. The natural electroosmotic flow of the capillaries is regenerated simply by rinsing with sodium hydroxide. Capillary zone electrophoresis (CZE) is a powerful tool for the separation of solutes ranging from small molecules and ions to biopolymers such as proteins, DNA, and carbohydrates.1 However, lack of control of the electroosmotic flow (EOF) and adsorption of solutes, particularly cationic proteins, onto the charged capillary walls plague CZE. A number of approaches for suppressing the EOF and/or minimizing wall adsorption have been explored. These include the following: adjustment of buffer pH;2-4 addition of high concentrations of ionic salts5 or zwitterions;6 addition of buffer modifiers;7 and permanent wall coatings.8-12 Some of the permanent coatings have demonstrated remarkable stability and migra* Author to whom correspondence should be addressed. Facsimile: 403289-9488. Electronic mail: [email protected]. (1) Landers, J. P., Ed. Handbook of Capillary Electrophoresis; CRC Press: Boca Raton, FL, 1994. (2) McCormick, R. M. Anal. Chem. 1988, 60, 2322. (3) Lauer, H. H.; McManigill, D. Anal. Chem. 1986, 58, 166. (4) Walbroehl, Y.; Jorgenson, J. W. J. Microcolumn Sep. 1989, 1, 41. (5) Green, J. S.; Jorgenson, J. W. J. Chromatogr. 1989, 478, 63. (6) Bushey, M. M.; Jorgenson, J. W. J. Chromatogr. 1989, 480, 301. (7) Stover, F. S.; Haymore, B. L.; McBeath, R. J. J. Chromatogr. 1989, 470, 241. (8) Hjerte´n, S. J. Chromatogr. 1985, 347, 191. (9) Tsuji, K.; Little, R. J. J. Chromatogr. 1992, 594, 317. (10) Schmalzing, D.; Piggee, C. A.; Foret, F.; Carrilho, E.; Karger, B. L. J. Chromatogr., A 1993, 652, 149. (11) Chiari, M.; Dell’Orto, N.; Gelain, A. Anal. Chem. 1996, 68, 2731. (12) Huang, M.; Bigelow, M.; Byers, M. Am. Lab. 1996, 28 (Oct), 32. S0003-2700(96)01231-0 CCC: $14.00

© 1997 American Chemical Society

tion time reproducibility (e.g., ref 10). Nonetheless, dynamic coatings formed by buffer additives remain attractive because of their low cost, simplicity, and versatility. Previous work has shown that cationic fluorosurfactants reduce wall adsorption of cationic proteins but also cause a strong reversed EOF.13,14 This paper investigates the use of low concentrations of zwitterionic surfactants to suppress the EOF and prevent adsorption of cationic proteins on the walls of the capillary. BACKGROUND Wall Adsorption. Theoretical studies,15,16 of wall adsorption in CZE have shown that models based on plate height theory agree with more sophisticated models and with computer simulations. Thus, only a short discussion of the simple plate height theory is given herein. The plate height in CZE is:15

H)

k′2

2D

r2υ

+ υ

+ (4k′ + 1)2 D

∑H

i

(1)

i

The first term in eq 1 is the longitudinal diffusion broadening which is dependent upon the diffusion coefficient of the solute (D) and the net (electrophoretic plus electroosmotic) migration velocity (υ). The second term is the resistance to mass transfer in the mobile phase term that arises from the slow radial diffusion to the retentive capillary wall. This broadening term depends upon the capacity factor k′ as defined in chromatography and the inner radius of the capillary (r), in addition to the diffusion coefficient and velocity. In eq 1, band broadening due to slow kinetic desorption from the stationary phase (Hs) is considered negligible. The last term in eq 1 lumps together all other sources of band broadening (e.g., Joule heating, extracolumn broadening, siphon effects). A few observations regarding broadening due to wall adsorption can be made.15 First, for slow diffusing solutes such as the proteins studied herein, efficiency losses can be severe even for slightly retained compounds. Indeed, plate heights increase 20fold for proteins with k′ values as small as 0.05.5 Second, while eq 1 suggests that decreasing the capillary radius decreases the effects of wall adsorption, no significant improvement is observed.15 The increase in capacity factor (k′) caused by the (13) Emmer, A° .; Jansson, M.; Roeraade, J. J. High Resolut. Chromatogr. 1991, 14, 738. (14) Emmer, A° .; Jansson, M.; Roeraade, J. J. Chromatogr. 1991, 547, 544. (15) Schure, M. R.; Lenhoff, A. M. Anal. Chem. 1993, 65, 3024. (16) Ermakov, S. V.; Zhukov, M. Yu.; Capelli, L.; Righetti, P. G. J. Chromatogr., A 1995, 699, 297.

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997 3435

Figure 1. Zwitterionic surfactants used in this work: (A) dodecyldimethyl(3-sulfopropyl)ammonium hydroxide (C12N3SO3); (B) hexadecyldimethyl(3-sulfopropyl)ammonium hydroxide (C16N3SO3); and (C) coco (amidopropyl)hydroxyldimethylsulfobetaine (CAS U). R ) C8-C18.

increase in surface area to volume upon lowering the capillary radius largely offsets the explicit effect of capillary radius. Several factors need to be considered in evaluate coatings in CZE:17 (a) separation efficiency, in terms of plates per meter. In theory, this should approach 1-2 million/m; (b) recovery of protein, which should approach 100% in the ideal case; (c) reproducibility of migration time from run to run, as well as for longer time periods, such as day to day; and (d) retention of EOF so that both cationic and anionic proteins may be separated during the same run. Adsorption of solutes onto a bare capillary wall arises primarily from electrostatic attraction. This is most clearly illustrated by the reduction of protein adsorption using the high concentration zwitterionic buffers developed by Bushey and Jorgenson6 and later commercialized by Waters as Z1 Methyl Additive. Secondary retention due to hydrophobic and hydrogen-bonding interactions between adsorbed and free solute can also occur.16 Furthermore, given the strong inverse dependence between diffusion coefficient and band broadening due to wall adsorption (eq 1), wall adsorption affects are most severe for cationic macromolecules. Mazzeo and Krull17 recommended the use of cationic proteins such as lysozyme and cytochrome c as test standards. Items a and b indicate the effectiveness of the coating at shielding the silanol groups on the capillary wall, thereby preventing adsorption. Item c indicates the stability of permanent coatings, and in our studies will indicate the reproducibility of the dynamic coating. These criteria17 were developed to evaluate permanent coatings. Nonetheless, they provide a rigorous framework from which to assess the efficacy of the dynamic coatings formed by zwitterionic surfactants. Zwitterionic Surfactants. Zwitterionic surfactants consist of a hydrophobic tail and a hydrophilic head group possessing both cationic and anionic functionalities. These characteristics are evident in the example sulfobetaine zwitterionic surfactants presented in Figure 1. Zwitterionic surfactants display critical micelle concentrations (cmc) higher than nonionic surfactants and lower than ionic (17) Mazzeo, J. R.; Krull, I. S. In Handbook of Capillary Electrophoresis; Landers, J. P., Ed.; CRC Press: Boca Raton, FL, 1994; Chapter 18.

3436 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

surfactants of similar hydrophobic chain length. They form spherical or near-spherical micelles with aggregation numbers and surface charge densities similar to ionic surfactants.18-20 Unlike typical anionic or cationic surfactants, ionic strength has little effect on the cmc of zwitterionic surfactants. The cmc of C12N3SO3 decreases only 16% when the ionic strength increases from 0 to 0.2 M.21 Fourier transform infrared and nuclear magnetic resonance spectroscopy studies indicate that no intramolecular ion pairing occurs within these surfactants. That is, the head group does not wrap around to form an ion pair with itself.18 Similarly binding of counterions from solution is very weak and occurs only well above the cmc.20 This low ion binding results from the high delocalization of the formal charges within the head group. For instance, quantum mechanics calculations of the carboxylate analog (betaine) indicate that the charge on the quaternary nitrogen is +0.038 e-, much below the formal charge of +1 e-.18 Most of the charge is located on the methyl and methylene groups attached to the nitrogen. The charge on the carboxylate of betaine is -0.776 e-. EXPERIMENTAL SECTION Apparatus. A P/ACE 2100 (Beckman Instruments, Fullerton, CA) with a direct UV absorbance detector was used for all experiments. Untreated silica capillaries (PolymicroTechnologies, Phoenix, AZ) with an inner diameter of 75 µm, outer diameter of 365 µm, and a total length of 47 cm (40 cm to the detector) were used unless otherwise noted. Data acquisition (5 Hz) and control was performed using System Gold software (Beckman) on a 386based microcomputer. Chemicals. All solutions were prepared in Nanopure ultrapure water (Barnstead). Buffers were prepared from reagent grade orthophosphoric acid (BDH) and sodium hydroxide (BDH). Three zwitterionic surfactants, dodecyldimethyl(3-sulfopropyl) ammonium hydroxide (C12N3SO3; Aldrich), hexadecyldimethyl(3-sulfopropyl)ammonium hydroxide (C16N3SO3; Aldrich), and coco (amidopropyl)hydroxyldimethylsulfobetaine (Rewoteric AM CAS U; Witco) were used as received. The structures of these surfactants are given in Figure 1. The R group in Rewoteric AM CAS U represents an alkyl chain with variable length from C8 to C18. An average molecular weight of 450 was used. The ionic strength of 10 mM phosphate buffer solutions was adjusted using sodium chloride (BDH). A 5 mM mesityl oxide (Aldrich) solution in water was used as the EOF marker. Lysozyme (chicken egg white), R-chymotrypsinogen A (bovine pancreas), ribonuclease A (bovine heart), and cytochrome c (bovine heart) were used as received from Sigma. EOF Measurements. New capillaries were used for each zwitterionic surfactant to avoid hysteresis effects. Each new capillary was pretreated with 0.1 M NaOH for 10 min. Before each run, the capillary was rinsed at high pressure (20 psi) with NaOH for 1 min, H2O for 1 min, and buffer for 2 min. Mesityl oxide was introduced as an EOF marker onto the capillary using low pressure (0.5 psi) hydrodynamic injection for 1 s. Detection was at 254 nm. Mesityl oxide had previously been found to be (18) Weers, J. G.; Rathman, J. F.; Axe, F. U.; Crichlow, C. A.; Foland, L. D.; Scheuing, D. R.; Wiersema, R. J.; Zielske, A. G. Langmuir 1991, 7, 854. (19) Kamenka, N. Chevalier, Y. Zana, R. Langmuir 1995, 11, 3351. (20) Kamenka, N.; Chorro, M.; Chevalier, Y.; Levy, H.; Zana, R. Langmuir 1995, 11, 4234. (21) Herrmann, K. W. J. Colloid Interface Sci. 1966, 22, 352.

an appropriate EOF marker in studies of EOF reversal using cetyltrimethylammonium bromide (CTAB).22 Also, the EOF measured herein was found to be the same whether mesityl oxide or methanol was used as the marker. The capillary was thermostated to 25 °C, and a constant voltage of 15 kV was applied. In the ionic strength study, 10 kV was used to avoid Joule heating as the solution ionic strength increased. Measurements of the electroosmotic mobility (µ) were performed in two ways. When the magnitude of the EOF was high (>2.5 × 10-4 cm2/V‚s), the migration time of mesityl oxide under a constant voltage was used:

µ ) LtLd/tmV

(2)

where Lt and Ld refer to the total length (47 cm) and the length to detector (40 cm) of the capillary, tm is the time for sample to migrate to the detector under the electric field, and V is the applied voltage. When the electroosmotic mobility is low (>cmc), the EOF plateaus at ∼2.5 × 10-4 cm2/V‚s (46% of the EOF in the absence of C12N3SO3). This plateau indicates the adsorption of C12N3SO3 onto the capillary wall has reached a saturated level. Effect of C16N3SO3 Concentration and pH on EOF. Figure 3 shows the effect of C16N3SO3 concentration on the EOF. The behavior observed for C16N3SO3 at pH 8 is similar to that observed in Figure 2, allowing for the difference in the cmc of the two surfactants. The cmc of C16N3SO3 is 0.028-0.07 mM.18,21,32 The asymptotic EOF observed at pH 8 for high C16N3SO3 concentrations is at ∼2.0 × 10-4 cm2/V‚s. This value is comparable to that observed for C12N3SO3 (2.5 × 10-4 cm2/V‚s) under the same pH and ionic strength conditions. This demonstrates that the alkyl chain length of the zwitterionic surfactants has little effect on the degree of EOF suppression. Rather, the hydrophobic chain length only affects the cmc of the surfactant. This in turn determines the surfactant concentration at which EOF suppression begins. Studies of EOF reversal with cationic surfactants yielded a similar conclusion.22 Figure 3 also illustrates the effect of pH on the EOF in the presence of C16N3SO3. All experiments were performed at constant ionic strength. Sodium chloride was added to the phosphate buffers to give a constant ionic strength of 50 mM. Figure 3 shows that the pH of the buffer has a remarkable effect on the EOF in the presence of C16N3SO3. The EOF suppression was observed only under mild pH conditions (pH 6 and 8). At low pH (pH 3), the intrinsic electroosmotic mobility is already very low due to the protonation of the silanol groups on the capillary surface. No effect of C16N3SO3 is evident under such (32) Mukerjee, P.; Mysels, K. J. Critical Micelle Concetrations of Aqueous Surfactant Systems; National Bureau of Standard, U.S. Government Printing Office: Washington, DC, 1971.

3438 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

Figure 4. Effect of Rewoteric AM CAS U (coco (amidopropyl)hydroxyldimethylsulfobetaine-CAS U) concentration on the EOF at the following pH: 3, 3.0; ], 6.0; O, 8.0; +, 10.0; 4, 12.0. Experimental conditions, as in Figure 2.

conditions. At high buffer pH (g10), the effect of C16N3SO3 on EOF is minimal, even at surfactant concentrations well above the cmc. It is believed that the adsorption of zwitterionic surfactant occurs when the cationic group of the surfactant interacts favorably with the anionic capillary wall. This interaction must be sufficiently strong to override the repulsion between the anionic wall and the anionic group of the surfactant. As the pH increases, the degree of deprotonation of the silanol groups increases, resulting in a high charge density on the capillary wall. This high wall charge increases the repulsion between the wall and the anionic group of the zwitterionic surfactants. Hence, adsorption of the zwitterionic surfactant onto the capillary wall becomes increasingly disfavored as the pH (and thus the wall charge) increases. In bare capillaries, the EOF decreases with the square root of the ionic strength.33 The effect of ionic strength on the EOF in the presence of C16N3SO3 above its cmc was studied at pH 7.33 over the ionic strength range 0.02-0.12 M. The relationship observed (not shown) between the observed EOF and the ionic strength was consistent (within the 95% confidence limits) with that observed in a bare capillary. Thus, ionic strength only alters the EOF by shrinking the electrical double layer. Effect of CAS U Concentration and pH on EOF. Figure 4 shows the effect of CAS U concentration on the EOF measured at various pH. A number of observations can be made from this figure. First, EOF suppression occurs even at the lowest CAS U concentrations studied (10-5 M). This suggests that the cmc of CAS U is very low. Second, CAS U is much more effective at EOF suppression than C12N3SO3 and C16N3SO3. For example, at pH 8 the EOF at high CAS U concentration is down to ∼0.3 × 10-4 cm2/V‚s, which is only 6% of the EOF obtained in the absence of CAS U. C12N3SO3 and C16N3SO3 only reduced the EOF to 45 and 35% of its natural value, respectively. Third, the EOF is largely suppressed over the entire pH range studied (pH 3-12). An additional EOF measurement performed for 4 mM CAS U at pH 13 and 10 kV still reduced the EOF to ∼40% of its initial value. This result is not included in Figure 4 because the ionic strength of the pH 13 buffer exceeds the ionic strength (50 mM) used in the studies shown in this figure. (33) Tsuda, T. In Handbook of Capillary Electrophoresis; Landers, J. P., Ed.; CRC Press: Boca Raton, FL, 1994; Chapter 22.

Figure 5. Proposed hemimicelle model. See text for details.

Proposed Hemimicelle Model. It is proposed that zwitterionic surfactants adsorb onto the capillary in a hemimicellar or bilayer fashion, as is depicted in Figure 5. This structure results from the adsorption of zwitterionic micelles onto the capillary walls. The localized charge density at the surface of the hemimicelle will be a function of the electronic repulsion and attraction between the head groups. However, the net surface charge would be zero. Silanols (with a free counterion) would still be present on the capillary wall, but would be shielded from the bulk solution by the adsorbed surfactant. It is believed that the EOF observed above results from the migration of the cations associated with the silanols. However, in the presence of the adsorbed surfactant, these cations (e.g., Na+) are migrating through an ordered or viscous media localized at the capillary wall surfaces. Thus the magnitude of the EOF is greatly reduced. This hemimicelle model is supported by a number of factors. First, for C12N3SO3 and C16N3SO3, a good correlation was observed between the cmc of the surfactant and the onset of EOF suppression. Second, such a hemimicelle structure was reported for adsorption of zwitterionic surfactants onto silicon nitride34 and silica.35 Similarly, hemimicelle formation has previously been used to explain cationic surfactants onto capillary walls in CZE.22 Third, based on “capillary rise” within wetted capillaries, capillaries in contact with zwitterionic surfactant solutions display hydrophilic surface properties. Fourth, addition of very high concentrations (up to 2 M) of nonsurface active zwitterions did not significantly alter the EOF.6 Thus, the surface-active nature of the buffer additives used herein must be critical to their behavior. CAS U displays the greatest effectiveness and pH stability for EOF suppression. The only major structural differences between CAS U and C16N3SO3 are the additional hydroxyl and amide functionalities in CAS U and the variable hydrophobic chain lengths in CAS U. As surfactant behavior is extremely complex, it is not possible to state which of these features is responsible for the enhanced performance of CAS U. Suffice to say that CAS U generates a denser or more ordered surface coating, which more effectively suppresses the EOF. Protein Adsorption. Addition of zwitterionic surfactant to the electrophoretic buffer would shield the silanol groups on the capillary wall but expose a zwitterionic surface to the solution (Figure 5). Thus, it was not obvious whether the use of zwitterionic surfactant buffer additives would prevent wall adsorption of cationic proteins. In the absence of CAS U in the electrophoretic buffer, no peaks were observed upon injection of (34) Ducker, W. A.; Clarke, D. R. Colloids Surf. A 1994, 94, 275. (35) Chorro, M.; Kamenka, N.; Faucompre, B.; Partyka, S.; Lindheimer, M.; Zana, R. Colloids Surf. A 1996, 110, 249.

Figure 6. Basic protein separation in the presence of 2 mM CAS U. Peaks: (1) lysozyme; (2) R-chymotrypsinogen. Conditions as in Experimental Section.

lysozyme and R-chymotrypsinogen A. Addition of 2 mM CAS U to the pH 7.2 phosphate buffer resulted in sharp peaks for both cationic proteins, as shown in Figure 6. Table 1 summarizes the figures of merit for this coating, along with values for some literature procedures for preventing wall adsorption. Migration time reproducibility was assessed by performing five replicate runs on three successive days. The run-to-run migration time reproducibility achieved is comparable to those achieved using other approaches. However, some drift in migration times was evident during a day’s experiments. The day-to-day migration time reproducibility is comparable to that achieved using a dynamic cationic fluorosurfactant coating14 and somewhat poorer than that achieved using permanent coatings (most notably ref 10, where the day-to-day reproducibility was determined over 600 runs!). Nonetheless it is reasonable. Protein migration can be altered by adsorption of surfactant.28,29 However this effect is relatively minor. For instance, the mobility of lysozyme only changed from 1.1 × 10-4 to 1.5 × 10-4 cm2/v‚s in the extreme case of replacing the 2 mM CAS U with 2 mM of the cationic surfactant CTAB. Further, no change in lysozyme mobility was observed in CAS U/CTAB surfactant mixtures up to ratios of 1:1. The efficiencies observed for both lysozyme and R-chymotrypsinogen A are excellent, in excess of 750 000 plates/m. This is superior to those achieved previously (seen Table 1 in ref 17 for further comparisons). However, it is still somewhat shy of the ideal 1-2 million plates/m, suggesting that some wall adsorption may be occurring. The initial alkaline conditioning and regeneration between runs was critical to achieving good efficiencies, particularly for R-chymotrypsinogen A. NMR studies of hemimicelles of cationic surfactants have found that the top layer of the structure free exchanges with solution surfactant while the bottom layer is static.36 It is therefore believed that it is the bottom layer of the hemimicelle structure that is responsible for shielding of the silanols and prevention of wall adsorption. The generation of a high surface silanol concentration would be essential to ensuring that there is a high population of surfactant in the bottom layer of the hemimicelle structure to shield the wall. Protein recoveries are difficult to determine since the small quantity of protein injected precludes the use of fraction collection (36) So ¨derlind, E.; Stilbs, P. Langmuir 1993, 9, 2024. (37) Swedberg, S. A. Anal. Biochem. 1990, 185, 51.

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3439

Table 1. Efficiency and Migration Time Reproducibility for Basic Proteins Separated by CZE % RSD of migration time approach 2 mM CAS

Ua 134b

0.14 mM FC neutral coatingc neutral coatingd zwitterionic

saltse

nonionic surfactants on C18 coatingf

protein

within day

day to day

recovery (%)

efficiency (plates/m)

lysozyme R-chymotrypsinogen A lysozyme

0.6 1.3 0.7 1-4 0.3 0.4

2.3 4.5% 2.4 7 0.3 0.4

97 80

94

770 000 790 000 413 000 500 000-600 000 385 000 664 000 200 000 290 000 154 000g

95

184 000g

lysozyme R-chymotrypsinogen A lysozyme R-chymotrypsinogen A lysozyme

0.8 R-chymotrypsinogen A

1.3

a Experimental conditions, as in Figure 6. b Reference 14: Fluororad FC134, CF (CF ) SO NH(CH ) N+(CH ) I-. Experimental conditions: 3 2 7 2 2 3 3 3 buffer, 0.05 M phosphate (pH 7); applied field, 300 V/cm. c Reference 37: coating prepared using (3-aminopropyl)trimethoxysilane as coupling agent, followed by reaction with pentafluorobenzoyl chloride to give the final coating. Experimental conditions: buffer, pH 7; applied field, 250 V/cm. Test solutes: lysozyme, ribonuclease, trypsinogen, whale myoglobin, horse myoglobin, human carbonic anhydrase, and bovine carbonic anhydrase B. d Reference 10. Two-layer coating consisting of a highly cross-linked hydrophobic polymeric siloxane sublayer onto which was coated a mixture of polysiloxanediol and silane cross-linker. Experimental conditions: buffer, 20 mM 6-aminocaproic acid/acetic acid (pH 4.4); applied field, 526 V/cm. Number of injections performed on one capillary, 600. e Reference 6. Protein adsorption prevented by high concentrations of zwitterionic salts within the buffer. Experimental conditions: buffer, 0.04 M phosphate-2 M betaine-0.1 M K2SO4 (pH 7.6); applied field, 400 V/cm. f Reference 24. A C18-coated capillary was rinsed with 0.5% (w/w) solution of a nonionic surfactant (Tween-20 or Brij-35) for 2 h. Experimental conditions: 0.01 M phosphate (pH 7.0) with 0.001% Brij-35; applied field, 300 V/cm. Test solutes: lysozyme, cytochrome c, ribonuclease A, chymotrypsinogen A, and myoglobin. g In a Note Added in Proof in ref 24, the observation was made that separation efficiencies increased drastically if the concentration of Brij-35 in the run buffer was increased to 0.01%. Unfortunately no explicit efficiencies were reported.

with off-line quantification. The only previous assessment of recoveries was that of Towns and Regnier using two on-capillary detectors positioned 50 cm apart.24 The loss of protein was determined by the difference in peak area measured by the two detectors. As our instrument has only a single detector, recoveries were determined herein by performing the protein separation using two capillary lengths at a constant applied potential (319 V/cm). The recoveries based on triplicate injections at each capillary length were 97% for lysozyme and 80% for R-chymotrypsinogen A. The recovery for lysozyme is near quantitative and comparable to that of Towns and Regnier, the only previous reported recoveries.24 The lower recoveries for R-chymotrypsinogen A are consistent with the greater difficulties experienced in achieving high efficiencies with this protein. Paradoxically, studies of high concentrations of zwitterionic salts experienced more difficulties with lysozyme than R-chymotrypsinogen A.6 The discussion above and the results in Table 1 indicate that zwitterionic surfactant coatings rival the effectiveness of permanent coatings. However, one must not forget that the coating is actually dynamic in nature. This is best evidenced by a small modification in the recovery study performed abovessimply that 2 mM CAS U was added to the protein sample. The resultant recoveries, based on five replicate injections at each capillary length, were 100 ( 1.3 and 100 ( 2.5% for lysozyme and R-chymotrypsinogen, respectively. The wall coating is a micellar phenomenon. Thus, if surfactant is not present in the injection plug, the hemimicelles on the wall will begin to disassemble. This disintegration of the hemimicelle allows protein greater access to the silanols on the capillary wall. However, the improved recoveries by addition of zwitterionic surfactant to the samples is not without cost. The efficiencies were reduced to 325 000 and 600 000 plates/m for lysozyme and R-chymotrypsinogen, respectively, upon adding 2 mM CAS U to the sample. Presumably, addition of the surfactant to the sample reduces stacking effects. To evaluate the versatility of the CAS U dynamic coating, additional cationic proteins and pH buffers were investigated. 3440 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

Figure 7. Separation of cationic proteins over a range of pH values. Solutes: (1) lysozyme; (2) cytochrome c; (3) ribonuclease A; (4) R-chymotrypsinogen A. Experimental conditions: applied voltage, 8.6 kV; temperature, 25 °C; capillary length, 27 cm (20 cm to detector); buffer, 10 mM phosphate and 2 mM CAS U buffer adjusted to 50 mM ionic strength with NaCl; sample, 0.1 mg/mL of each protein.

Figure 7 shows separations of lysozyme, cytochrome c, ribonuclease A, and R-chymotrypsinogen A in buffers ranging from pH 4.0 to 7.0. In all cases sharp peaks were observed, indicating that the dynamic coating is effective at preventing protein wall

adsorption under these pH conditions. Higher pH values were not studied since such pH values would exceed the pI of the proteins. Wall adsorption would not be expected under such conditions. Mazzeo and Krull suggested that retention of EOF is desirable so that both anionic and cationic proteins might be determined in a single run.17 To this end, 2.0 mM C16N3SO3 was added to the pH 7.2 phosphate buffer, in place of CAS U (but not to the sample). Unfortunately, no peaks were observed under these conditions for lysozyme or R-chymotrypsinogen A. Therefore, it appears that there is a correlation between a zwitterionic surfactant’s effectiveness at suppressing the EOF and its ability to prevent wall adsorption of cationic proteins.

ACKNOWLEDGMENT This work has supported by the Natural Science and Engineering Research Council of Canada and by The University of Calgary. K.K.-C.Y. thanks the Killam Foundation for support of his studies through an Izaak Walton Killam Fellowship. Thanks also to Susan Kutay and Dr. Laurie Schramm of the Petroleum Recovery Institute of Calgary and Dr. Jed Harrison of the University of Alberta for their helpful comments and suggestions. Received for review December 4, 1996 Accepted June 6, 1997.X AC961231K X

Abstract published in Advance ACS Abstracts, July 15, 1997.

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3441