Capillary Isoelectric Focusing of Physiologically Derived Proteins with

However, a major problem has always existed; separation performance is usually degraded by the presence of salts within the sample. Normally this requ...
0 downloads 0 Views 126KB Size
Download PDF
Anal. Chem. 1997, 69, 2786-2792

Capillary Isoelectric Focusing of Physiologically Derived Proteins with On-Line Desalting of Isotonic Salt Concentrations Nigel J. Clarke and Andy J. Tomlinson

Biomedical Mass Spectrometry Facility, Department of Biochemistry and Molecular Biology, Mayo Clinic, Rochester, Minnesota 55905 Gerhard Schomburg†

45470 Mu¨ lheim a.d. Ruhr, Stiftstrasse 39, Germany Stephen Naylor*

Biomedical Mass Spectrometry Facility, Department of Biochemistry and Molecular Biology, Department of Pharmacology and Clinical Pharmacology Unit, Mayo Clinic, Rochester, Minnesota 55905

Capillary isoelectric focusing within capillaries (cIEF) is a powerful and practical method for high-resolution separation of components within complex biological mixtures. However, a major problem has always existed; separation performance is usually degraded by the presence of salts within the sample. Normally this requires the removal of these components by some off-line sample cleanup method, prior to analyte separation by cIEF. In this study, we have shown it is possible to efficiently remove high salt levels from samples by on-line voltage ramping of the applied CE voltage. To allow this technique to be used effectively, a customized version of an existing method to internally coat a fused-silica capillary has been developed and examined for interexperimental reproducibility. We describe the systematic examination of the desalting process and its optimization through the use of model protein systems. Furthermore, we demonstrate the automated application of this on-line desalting cIEF scheme to studies of whole human blood and human cerebrospinal fluid which have undergone no manipulation or work up prior to cIEF analysis. A rapidly emerging alternative approach for the efficient separation of polypeptides and proteins, while overcoming the limitations of 2D gel electrophoresis, is capillary isoelectric focusing (cIEF).1-3 Isoelectric focusing within capillaries was first demonstrated by Hjerte´n.4 The basic process consists of initial focusing and separation of analytes based on their pI values, followed by mobilization via either pressure or electrophoretic means.5-7 Specifically, the capillary is filled with a mixture of

sample and ampholytes, which are zwitterionic compounds used to create a pH gradient. When a voltage is applied across the capillary, the proteins present in the mixture are focused into discrete zones, at positions in the capillary where the local pH is equal to their pI. These analyte bands are sharply defined and are characteristic of the IEF process, resulting in high-resolution separation of the analytes. Furthermore, analytes are concentrated by 50-100-fold due to efficient focusing. This permits the analysis of dilute analyte solutions. Disruption of the stationary pH gradient required throughout the focusing and separation stage adversely affects analyte resolution. A possible cause of gradient disruption is the electroosmotic flow (EOF) produced when an electrical charge is passed through a conducting solution within a bare fused-silica capillary.8 The silanol groups (SiO-) on the capillary wall are charged when the capillary is filled with an aqueous solution, allowing the formation of a moving band of positively charged molecules through the capillary from the anode toward the cathode (EOF). This directly affects the resolution possible in a cIEF experiment by acting to disrupt or destroy any pH gradient formed during focusing of the analytes. To overcome this problem, it is common to coat the interior surface of the capillary with a neutral compound which reduces or negates this EOF.9 Another factor that can disrupt the pH gradient is the presence of salts. Indeed, salt concentrations of >100 mM within sample solutions, such as physiological fluids, normally destroy the pH gradient completely.10 High salt contamination causes significant shrinkage of the pH gradient.11-13 This leads to very concentrated

† Professor Schomburg is the former head of the Department of Chromatography and Electrophoresis, Max-Planck-Institut fu ¨ r Kohlenforschung, 45470 Mu ¨ lheim a.d. Ruhr, Germany. (1) Liu, X.; Sosic, Z.; Krull, I. S. J. Chromatogr., A 1996, 735, 165-190. (2) Pritchett, T. J. Electrophoresis 1996, 17, 1195-1201. (3) Wu, J.; Pawliszyn, J. J. Chromatogr., B 1994, 657, 327-332. (4) Hjerte´n, S.; Zhu, M.-D. J. Chromatogr. 1985, 346, 255-270. (5) Weinberger, R. Practical Capillary Electrophoresis; Academic Press: San Diego, 1993; Chapter 4.

(6) Righetti, P. G.; Gelfi, C.; Chiari, M. Capillary Electrophoresis in Analytical Biotechnology; CRC Series in Analytical Biotechnology; Righetti, P. G., Ed.; CRC Press: Boca Raton, FL, 1996; Chapter 12. (7) Kila´r, F. In Handbook of Capillary Electrophoresis; Landers, J. P., Ed.; CRC Press: Boca Raton, FL, 1994; Chapter 4. (8) Tsuda, T. In Handbook of Capillary Electrophoreis; Landers, J. P., Ed.; CRC Press: Boca Raton, FL, 1994; Chapter 22. (9) Gilges, M.; Kleemiss, M. H.; Schomburg, G. Anal. Chem. 1994, 66, 20382046. (10) Wehr, T.; Zhu M.; Rodriguez, R.; Burke, D.; Duncan, K. Am. Biotechnol. Lab. 1990, 8, 22-29. (11) Liao, J.-L.; Zhang, R. J. Chromatogr., A 1994, 684, 143-148.

2786 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

S0003-2700(96)01283-8 CCC: $14.00

© 1997 American Chemical Society

protein bands which can ultimately precipitate out of solution.13 Furthermore, high salt concentrations afford elevated current values when voltage is applied and electrophoresis is used to mobilize analytes off the capillary. This results in excessive Joule heating with subsequent analyte diffusion and, hence, band broadening.7,10,14 To counteract these problems, it is typical to desalt the sample prior to subjecting it to cIEF. This approach was recently described by Wu and Pawliszyn,15 who constructed a sample dialysis tube to both remove salt and add ampholyte. An on-line desalting method has also been described by Liao and Zhang.11 They used analytes to substitute for the salts in a process that typically takes up to 15-30 min at 0.3-0.5 mol/L NaCl. We have found it is possible to achieve rapid on-line desalting with minimum sample handling by a modification to the standard cIEF procedure. This new protocol takes place within a single capillary that has been coated to reduce or negate the EOF and utilizes the much greater relative mobility of the salt anions and cations to allow their efficient removal while causing minimum perturbation of the remaining analytes. In this study we show automated, on-line desalting of solutions containing 200 mM salt. Further, we examine two types of biologically derived samples, human whole blood and cerebrospinal fluid (CSF), which have undergone no prior manipulation and therefore contain physiological levels of salts. EXPERIMENTAL SECTION Materials. Fused-silica capillaries, 50 µm I.D. × 360 µm o.d. (Polymicro Technologies Inc., Phoenix, AZ), were internally coated with poly(vinyl alcohol) (PVA) (Aldrich Chemical Co. Inc., Milwaukee, WI) to reduce or eliminate the electroosmotic flow. A nine-peptide standard was obtained from Bio-Rad (Hercules, CA), cytochrome c (bovine heart), myoglobin (equine skeletal muscle), carbonic anhydrase (bovine erythrocytes), and trypsin inhibitor (soy bean) were from Calbiochem (La Jolla, CA), and lysozyme (chicken egg) myoglobin (equine skeletal muscle), and servalytes were purchased from Sigma Chemical Co. (St. Louis, MO). Isolytes were obtained from ICN (Costa Mesa, CA), whole blood samples were collected from a volunteer, and cerebrospinal fluids were a gift from Dr. Noberto Guzman (RWJ Pharmaceutical Research Institute, Raritan, NJ) and Dr. Luis Hernandez (School of Medicine, Los Andes University, Merida, Venezuela). All reagents were analytical grade or better. Capillary Coating Protocol. A variation of the technique developed by Schomburg et al. was used to coat the capillaries.9 A solution of 6.5% PVA (w/w) in boiling water was mixed thoroughly and sonicated for 10 min. The solution was then centrifuged for 1 min to pellet undissolved PVA from the bulk solution. The capillary to be coated was washed with 10 column volumes of deionized water, and one end was placed with a vial containing the PVA solution inside a pressure bomb (Figure 1). A head pressure of 60 psi nitrogen gas was applied to the pressure bomb to force the PVA through the capillary, the free end of which was in a beaker of stirred water. Once the PVA solution was seen to be exiting the capillary, the pressure was reduced to 30 psi and the system left for 2 h. After this time, the capillary was withdrawn from the PVA vial and the excess PVA inside the (12) Weinberger, R. Am. Lab. 1996, 27, 28T-28U. (13) Zhu, M.; Rodriguez, R.; Wehr, T. J. Chromatogr. 1991, 63, 2852-2857. (14) Grushka, E.; McCormick, R. M.; Kirkland, J. J.; Anal. Chem. 1989, 61, 472479. (15) Wu. J.; Pawliszyn, J. Anal. Chem. 1995, 67, 2010-2014.

Figure 1. Schematic of the capillary coating apparatus.

capillary was slowly emptied by pressure. The capillary was then placed inside an oven and heated to 145 °C for a minimum of 3 h while being purged internally with nitrogen gas at 30 psi. The PVA polymerizes at this temperature, undergoing an increase in hydrophobicity, becoming insoluble in water, and forming a smooth coating to the inner capillary surface. cIEF-UV Conditions. All experiments were conducted on either a P/ACE 2100 CE instrument with single-wavelength detection or a P/ACE 5000 model with a diode array detector (Beckman Instruments, Fullerton CA); each uses a cartridge to hold the capillary and bathe it in coolant. A PVA-coated capillary was used in all experiments. Two types of ampholyte, isolytes, and servalytes were combined in a 50:50 ratio, since this mixture had been found, empirically, to give more reproducible results than either type individually, as previously suggested by Hjerte´n.16 The combined ampholytes (pH 3.0-10.0) were mixed into the sample to give a total ampholyte concentration of 2% by volume. The ampholyte/sample mix was then pressure injected onto the capillary until it was completely filled. Single-wavelength UV detection was at 280 nm to avoid UV responses from the ampholytes. The desalting protocol developed relies on the salt anions and cations being much smaller and more highly mobile than the analytes or ampholytes.12 By the application of a slow voltage ramp before focusing commences, it is possible to electrophoretically remove the majority of the salt ions without significantly disturbing the analytes or ampholytes. Once this is achieved, a conventional cIEF experiment is completed. The final voltage, voltage gradient, and ramp time required to properly desalt a solution are all a function of the salt concentration within the sample. A steep gradient or high final voltage can reduce the analysis time, but the likelihood of system perturbation is increased. Therefore, a balance between experimental run time and system perturbation was considered during method development. Typically, a voltage of 5-10 kV was applied as a linear gradient over 6 min. Great care was taken not to allow the current drawn to reach higher than 15 µA, since greater values tended to degrade analyte resolution. Focusing was accomplished by simply adjusting the applied voltage to the system at the end of the desalting process. (16) Hjerte´n, S. In Capillary Electrophoresis: Theory and Practice; Grossman, P. D., Colburn, J. C., Eds.; Academic Press: San Diego, 1992; Chapter 7.

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2787

Table 1. Determination of PVA-Coated Capillary Performancea migration time

myoglobin carbonic anhydrase trypsin inhibitor a

fwhh

(s) analysis 1

(s) analysis 21

diff (%)

analysis 1

analysis 21

diff(%)

302.6 426.0 695.0

305.0 428.8 698.0

0.79 0.65 0.57

28.2 32.0 28.0

29.5 31.2 27.5

4.4 2.5 1.8

Reproducibility of migration time and peak shape (fwhh, full width at half-height) after 21 consecutive analyses of the three-protein mixture.

Typically, a voltage of 15-20 kV was used for 5-10 min. The exact focusing time for each sample type was determined during the first experiment run with that sample by monitoring the expected exponential current drop as the analytes and ampholytes become focused and uncharged. The focus time for a sample was set as the time taken for the current to fall below 10% of the initial value and stabilize. The desalting and focusing steps both employed the same anolyte (50:49:1 methanol/water/acetic acid, v/v/v) and catholyte (50:49:1 methanol/water/ ammonium hydroxide, v/v/v). The presence of organic solvent in the electrolyte solutions was evaluated in order to have a catholyte solution that gave optimal performance in cIEF with both UV and ultimately mass spectrometric detection. Mobilization was initiated cathodically by replacement of the focus catholyte with a reservoir of solution identical to that used as the focus anolyte, namely, 50: 49:1 methanol/water/acetic acid (v/v/v). A constant voltage of 20 kV was then applied across the capillary, causing the focused anolyte bands to migrate sequentially past the UV detector situated 7 cm from the cathode. RESULTS AND DISCUSSION Capillary Coating. Initially it was very important to determine the efficient and reproducible performance of the newly coated capillaries. Several characteristics were examined, including suppression of EOF and inter-run reproducibility. Suppression of EOF was empirically examined through the use of a commonly used neutral marker compound, dimethylformamide (DMF). It was noted that DMF migration time through a 37 cm bare fusedsilica capillary was ∼8 min. This contrasted with a migration time of ∼50 min in a 37 cm newly coated PVA capillary. Reproducibility was examined with model protein mixtures dissolved in deionized water. The initial protein mixture consisted of myoglobin purchased from Calbiochem containing two isoforms at pI’s 7.0 and 6.8, carbonic anhydrase isoforms A (pI 5.4) and B (pI 5.9). A series of eight consecutive analyses was carried out using this protein mixture. The migration times for each protein were highly reproducible (n ) 8) with standard deviations of 2.78 (myoglobin pI ) 7.0), 3.79 (myoglobin pI ) 6.8), 2.87 (carbonic anhydrase A pI ) 5.4), and 2.87 s (carbonic anhydrase B pI ) 5.9). We subsequently studied the effect of capillary age on migration times and peak shape. A second protein mix consisting of myoglobin (pI 7.0), carbonic anhydrase B (pI 5.9), and trypsin inhibitor (pI 4.6) was dissolved in deionized water. It should be noted that this batch of myoglobin purchased from Sigma used in these analyses did not contain detectable amounts of the isoform found in myoglobin from Calbiochem used to assess migration time reproducibility described above. This mixture was examined via cIEF at the beginning and end of a series of 21 experiments. The results for these two experiments are detailed in Table 1. 2788

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

Little alteration of protein migration times or full width half-heights (fwhh) for this protein mix was observed with capillary age, giving confidence that this coating technique produced capillaries which were reproducible and robust for use in cIEF analyses. cIEF-UV in the Presence of High Salt Concentrations. The study of newly coated PVA capillaries showed them to be highly suitable for use with cIEF experiments. Therefore we subsequently examined the process of on-line sample desalting for cIEF-UV experiments using these PVA-coated capillaries. To facilitate this methodology development, a third protein standard mixture was prepared covering a wider range of pI’s and molecular masses. This enabled comparison of results recorded from desalted samples with results from salt-free control samples to determine whether discrimination on the basis of analyte molecular mass or pI had occurred. The protein mixture contained 50 µg/mL each of lysozyme (pI 8.9), cytochrome c (pI 9.6), myoglobin (Sigma) (pI 7.0), carbonic anhydrase (pI 5.9), and trypsin inhibitor (pI 4.6), all dissolved in deionized water. Desalting was effected by the electrophoretic removal of the salt ions using a linear voltage ramp. Since cations such as sodium and anions such as chloride are more highly mobile than any of the ampholytes, the former “migrate to their respective electrodes at high speed”.12 In all experiments described here, desalting conditions employed a linear gradient from 0 to 10 kV over 6 min. These ramping conditions were designed to cause minimum perturbation of the forming pH gradient and focusing analytes. During the desalting step, as the voltage increases linearly, there is a large initial increase in current, which attains a maximum at ∼4 min (see Figure 2). At this time the bulk of the salt cations and anions are being mobilized off the capillary. Subsequently, the current drops precipitously, signifying the removal of the majority of the salts off the capillary and indicating that focusing can now commence. The voltage is maintained at 10 kV for a further 2 min and then rapidly increased to 20 kV over 30 s. A concurrent, second increase in current is observed (Figure 2) which, upon reaching a maximum at ∼9 min, slowly decreases as a function of time as analytes are mobilized off the capillary. This second current profile is typically observed in cIEF experiments where electrophoretic mobilization has been used.5-7 Only this “second” current maximum is observed in the absence of salt on the capillary. In order to evaluate the efficacy of the desalting method, we initially took the five-protein mixture and subjected it to cIEF on a coated capillary in the absence of salt (see Figure 3A). The constituent components are all well resolved, although responses from the more acidic proteins are slightly broader than the basic migrating analytes. This is due to the former having longer residence time on the capillary leading to some analyte diffusion of the focused protein bands.

Figure 2. Example of current profile recorded during a standard desalting experiment. Desalting was completed using a linear voltage ramp to 10 kV reached over 6 min. The CE capillary used was 37 cm × 50 µm PVA-coated fused silica. Focusing and mobilization solutions: anolyte, 50:49:1 methanol/water/acetic acid; catholyte, 50: 49:1 methanol/water/ammonium hydroxide. Focusing occurred at 20 kV for ∼10 min. Mobilization was typically completed at 20 kV (not shown), and detection was by UV absorption at 280 nm.

Interestingly, the migration order of lysozyme and cytochrome c was found to be the reverse of that expected from their literature pl values (cytochrome c pl 9.6, lysozyme pl 8.9). Furthermore, due to the positioning of the UV detector 7 cm away from the cathodic capillary tip, any analytes with pl values greater than 9.0 should focus beyond the detector and not be observed. Clearly, this is not the case for cytochrome c in these results. Recently Tang et al.17,18 suggested that the greater mobility of acetate ions within the mobilization buffer compared to that of the ampholyte ions allowed the formation of a “diffuse ionic boundary” at the cathodic end of the separation capillary. Such a boundary would migrate toward the anode and act in a manner similar to the leading trailing buffer in a transient isotachophoresis experiment (tITP). Very basic analytes focusing close to the cathodic capillary tip would become negatively charged in the presence of such a zone and begin to migrate toward the anode. Such an effect would explain the movement of cytochrome c to the anodic side of the UV detector and, hence, its eventual detection. At the same time, the influx of the H+ ions at the anode will pass through the capillary toward the cathode acting as the trailing stacking buffer. Such an effect would setup tITP conditions near the cathodic capillary end, tending to cause cytochrome c and lysozyme to be found at a much closer proximity within the capillary than expected based on purely pl focusing. Individual analyte electrophoretic mobility would then become a much greater influence on observed migration order. In order to evaluate the role of electrophoretic mobilization on migration order, cytochrome c and lysozyme were focused on a 47 cm long PVA-coated capillary. They were subsequently mobilized by pressure (20 psi) rather than electrophoretically. The longer capillary was used to allow detection of cytochrome c in the absence of any tITP-like conditions that would occur on electrophoretic mobilization. Furthermore, a diode array detector was used in order to differentiate the two proteins since only the heme group of cytochrome c has a λmax ∼ 415 nm. Cytochrome c migrated past the detector first followed by lysozyme, indicating (17) Tang, Q.; Harrata, A. K.; Lee, C. S. Anal. Chem. 1995, 67, 3515-3519. (18) Tang, Q.; Harrata, A. K.; Lee, C. S. Anal. Chem. 1996, 68, 2482-2487.

Figure 3. Electropherograms recorded during mobilization of fiveprotein mix in the absence of salt. (a) was recorded in the absence of added salt while (b) was recorded from a solution originally containing 200 mM NaCl and diluted 1:1 with ampholyte to give a final salt concentration of 100 mM. There is a slight increase in the peak migration times of the salt-bearing sample which is probably due to a small shrinkage in the ampholyte zone. Focusing (not shown) and mobilization solutions: anolyte, 50:49:1 methanol/water/acetic acid; catholyte, 50:49:1 methanol/water/ammonium hydroxide. The desalting protocol used in (b) was a linear gradient to 10 kV reached over 6 min. Focusing, mobilization, and detection were as described in Figure 2.

that the former protein was focused at a more basic pI than lysozyme, as expected (data not shown). In order to determine the effect of electrophoretic mobility on migration order, we carried out a CZE separation of an equimolar mixture of the two proteins. We used the same PVA-coated capillary and mobilization solutions, now acting as electrolyte solutions, as previously used in cIEF separations. Lysozyme (average migration time 775 s; n ) 4) possessed a much greater electrophoretic mobility than cytochrome c (average migration time 846 s; n ) 4). Hence, it appears that, due to electrophoretic mobilization in cIEF, the migration order of cytochrome c and lysozyme is reversed. Subsequently, the protein mixture was subjected to cIEF in a solution containing 200 mM sodium chloride (Figure 3b). The analysis was conducted using the same capillary as for the previous salt-free experiment. Interestingly, the peak migration times were slightly longer for the analytes in the salt solution (Figure 3b) than were observed for the salt-free sample. These alterations in migration times appear to be the product of a small contraction in the pH gradient. This occurs during the desalting process when cations and anions of the salt are electrophoretically Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2789

Table 2. Resolution of Lysozyme and Cytochrome c in the Presence and Absence of Salt and Calculation by an Alternative Method to Full Width Half-Height calcd resolution protein identification

absence of salt

presence of salta

S1/S3 (for lysozyme) S2/S3 (for cytochrome c)

15.11 6.11

72.0 33.5

a in these experiments, the intensities of the major response for lysozyme and cytochrome c were taken as S1 and S3, respectively.

Figure 4. Theoretical partial resolution of idealized peaks, demonstrating values S1, S2, and S3 used in calculation of analyte separation efficiency for a cIEF-UV experiment.

mobilized from the capillary. Counterions from the buffer system pass onto the capillary, causing a small shrinkage in the pH gradient at both ends.13 Since the focused analyte zones are within this gradient, their relative distance to the detector will alter as this shrinkage occurs, affecting their migration times. Such contraction of the pH gradient may also induce a very small loss in separation efficiency. Often, however, the modest gain in resolution produced by bands being compressed during the gradient shrinkage compensates for this. Indeed, when the two electropherograms (Figure 3a,B) are compared, the peak widths (fwhh) for all five proteins are significantly improved (percent decrease in fwhh: lysozyme 32.4; cytochrome c 20.0; myoglobin 47.5; carbonic anhydrase 13.4; and trypsin inhibitor 31.6) after using the desalting step compared to conventional cIEF analysis of the mixture in the total absence of salt. Finally, the protein mixture containing 200 mM NaCl was subjected to cIEF without the slow voltage ramp. The voltage was increased rapidly over ∼30 s to 20 kV to effect focusing and subsequent electrophoretic mobilization. Due to the high ionic strength of the solution on application of the voltage, the CE current reached ∼45 µA and became unstable. It was not possible to obtain any satisfactory separation of proteins without first effecting removal of the NaCl using the voltage ramp. It should be noted that due to the analytes being separated spatially when focused within the capillary it is difficult to measure separation efficiency or resolution in cIEF in the same manner as calculated for other CE experiments. To allow such a comparison, a different protocol for measuring the resolution can be employed.19 Two analyte peaks are chosen which are not baseline resolved. The intensity of the two peaks are described as S1 and S2. The intensity at the lowest point of the interconnecting valley is described as S3 (see Figure 4 for diagram). The (19) Hjerte´n, S., Private communication, 1996.

2790 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

ratio of S1/S3 and S2/S3 will then be a measure of the resolution. The greater these ratios, the greater the resolution. Using the lysozyme/cytochrome c results gained in the salt-free and saltbearing experiments, such a calculation was completed and the results are reported in Table 2. This clearly shows an increase in analyte resolution after the sample has been desalted. This may be due to a small amount of analyte focusing occurring during the desalting process or possibly the small amount of salts remaining after desalting improving the focusing process by production of slightly higher currents. It has been reported that increasing the ionic strength of the buffer solution in CE experiments by the addition of small quantities of sodium chloride (90% of the responses have significant fwhh values and are not sharp spikes indicative of precipitated protein analytes. Finally, it was not possible to analyze the whole blood sample without voltage ramping to effect removal of the salt. As described previously, a high current led to instrument instability and focusing and electrophoretic mobilization were not possible to achieve. This experiment highlights the success of the desalting protocol for a physiologically derived fluid. It also draws attention to the very limited range of data that can be gathered using UV detection. However, with complete automation of data acquisition plus the speed and ease with which results can be gathered, such a system could find many uses in the examination of physiological

Figure 6. Human CSF examined by cIEF-UV. The CSF was mixed the ampholyte solution and desalted on-line. Focusing and mobilization solutions: anolyte, 50:49:1 methanol/water/acetic acid; catholyte, 50:49:1 methanol/water/ammonium hydroxide. A linear gradient to 10 kV over 6 min was used to desalt the sample. Focusing, mobilization, and detection were completed as described in Figure 2.

samples. One usage would be in the examination of the blood from diabetic patients, necessary in the control of the disease. By quantitatively examining the relative amounts of glycosylated to nonglycosylated hemoglobins, the level of control of the disease can be ascertained. Following the successful experiments with whole blood, a sample of human CSF was examined without prior manipulation or cleanup and a representative electropherogram is shown in Figure 6. The results gained are extremely complex in nature, and the lack of structural information provided by UV response data reduces the utility of these results. However, the desalting procedure has once more been proven to work successfully in the presence of a biologically derived sample matrix and the results so gained can still be very informative. It was not possible to analyze the CSF sample without effecting removal of the salt as described previously for the whole blood analysis. Alterations in the relative intensities of peaks may be noted if compared with a library of standard results allowing the identification of new or altered analytes present within the sample. This could provide an automated method of quickly and inexpensively screening large numbers of such complex physiological samples while requiring very small sample volumes and reducing the chance of contamination or analyte loss through cleanup procedures. CONCLUSIONS Here we have shown that it is possible to rapidly and effectively desalt on-line and within the separation capillary by the use of a well-chosen voltage profile. The procedure adds a small amount of time to the experiment, but this is negligible when compared with the extra complexity off-line desalting introduces. The reduction in sample handling and the inherent concentration of the technique allows small volumes of dilute samples to be examined with minimum sample loss or contamination. The entire process takes place within a single capillary, again reducing analyte losses to multiple surfaces and the use of commercially available capillary electrophoresis instruments allows for easy automation. Results have been shown to be very reproducible in both peak shape and migration times, and extremely complex physiological fluids have been examined directly with no pretreatment. Addition Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2791

of a more informative detection system should yield large amounts of structural data, and recently the use of capillary isoelectric focusing-mass spectrometry has been reported.17,18 Present studies in this laboratory are evaluating the use of cIEF with online desalting using mass spectrometry as a method of detection. ACKNOWLEDGMENT The authors thank the Mayo Foundation and Beckman Instruments for their financial support. The authors also thank Dr. Norberto Guzman and Dr. Luis Hernandez for their kind gift

2792

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

of human cerebrospinal fluid, and Prof. Stellan Hjerte´n for his discussion regarding calculation of resolution within a cIEF experiment.

Received for review December 18, 1996. Accepted May 9, 1997.X AC961283+ X

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