Anal. Chem. 1994,66, 3782-3787
Reductive Electrochemical Detection for Capillary Electrophoresis Michael A. Malone,? Paul L. Weber,t Malcolm R. Smyth,* and Susan M. Lunte'gt Center for Bioanalytical Research, Lawrence, Kansas 66047, and School of Chemical Sciences, Dublin City University, Dublin, Ireland
A system has been developed for reductive electrochemical detection with capillary electrophoresis. The instrumentalsetup is much simpler than that of reductive liquid chromatography/ electrochemistry (LCEC). The new method also possesses major advantages over LC-based methods in terms of a less complicated deoxygenationsystem and shorter deoxygenation times. Concentration detection limits were similar to those reported for reductive LCEC. Mass detection limits for anthraquinone-2-carboxylic acid and (dinitropheny1)-yaminobutyric acid were 1.3 and 1.6 fmol, respectively, at a S/N of 2. The reproducibility of the signal for a 9.8-fmol anthraquinone-2-carboxylic acid injection yielded a relative standard deviation of 6.2%. The high selectivity of reductive electrochemicaldetection was demonstrated by the detection of mitomycinC directly in human serum without prior extraction procedures. The majority of electrochemical investigations to date have dealt with reducible compounds. This is primarily due to the early use of classical mercury electrodes. However, liquid chromatography/electrochemistry (LCEC) has been very popular because of its applicability to oxidizable compounds. The use of reductive electrochemical detection in LC, although quite extensively investigated, has been compromised by the high background currents which occur due to dissolved oxygen and trace metals in the system. In 1978, MacCrehan and Durstl used a gold/mercury amalgam electrode with differential pulse reductive electrochemical detection for the determination of organomercury cations in biological samples after liquid chromatographic separation. Reductive LCEC was further advanced by Bratin and Kissinger,2 who evaluated various electrode substrates, including carbon and gold/ mercury amalgams. They applied their system to the determination of several electrochemically reducible compounds, including explosive^^,^ and nitro-containing pharmaceuticals.' Reductive LCEC was later investigated for the detection of nitroaromatic compoundsG8 and for the determination of polynuclear aromatic hydrocarbons in diesel exhaust^.^ All of these applications required lengthy deoxygenation procedures, including overnight heating of the mobile for Bioanalytical Research. Dublin City University. (1) MacCrehan, W. A.; Durst, R. A. Anal. Chem. 1978,50, 2108. (2) Bratin, K. Ph.D. Dissertation, Purdue University, West Lafayette, IN, 1981. (3) Bratin, K.; Briner, R. C.; Kissinger, P. T.; Bruntlett, C. S.Anal. Chim. Acta 1981, 130. 295. (4) Bratin, K.; Briner, R. C. Curr. Sep. 1980, 2 , 1. ( 5 ) Bratin, K.; Kissinger, P. T.Curr. Sep. 1982, 4, 4. (6) Jacobs, W. A. Ph.D. Dissertation, Purdue University, West Lafayette, IN, t Center 4
1983.
(7) Jacobs, W. A.; Kissinger, P. T. J. Liq. Chromatogr. 1982, 5, 881. (8) Jacobs, W. A.; Kissinger, P. T. J. Liq. Chromatogr. 1982, 5, 669.
3782
Analyfical Chemistty, Vol. 66, No. 21, November 1, 1994
phase with nitrogen sparging. In addition, the chromatographic system needed to be plumbed completely with stainless steel tubing to prevent diffusion of oxygen into the system. Despite these drawbacks, reductive LCEC has been employed for the analysis of various types of compounds including antitumor agents,lOJ antirheumatic compounds,12 benzodia z e p i n e ~ and , ~ ~ vitamins.I4 Electrode substrates employed have included mercury,lOJ1,13 gold/mercury amalgam,12and glassy carbon.14 Two other techniques used to circumvent the high background characteristic of reductive LCEC include reverse pulsed amperometric detection'' and dual electrode detection.16 Reversed pulse amperometry is based on the application of an unsymmetrical square wave with a large negative deposition potential followed by a positive potential pulse. This technique has been used for the detection of metals following chromatographic separation. Dual electrode detection involves the employment of a generator electrode poised at a potential more negative than the half-wave potential of the analyte of interest and a detector electrode at a potential suitable for the oxidation of the product generated by the first electrode. Because the actual detection occurs in the oxidative mode, there is no interference from oxygen and trace metals. However, this approach is applicable only to chemically reversible c o m p ~ u n d s . ~ * ~ J ~ - ~ ~ Capillary electrophoresis has several advantages over LC, particularly in terms of its high efficiency and low volume requirements. Electrochemical detection is ideally suited to CE due to the fact that it can usually be miniaturized without a loss of sensitivity. The utility of capillary electrophoresis/ electrochemistry (CEEC) has been demonstrated in the past.2b25 However, the majority of applications have dealt (9) Rappaport, S.M.; Jin, Z. L.; Xu, X. 9. J. Chromatogr. 1982, 240, 145. (10) Tjaden, U. R.; Langenberg, J. P.; Ensing, K.;Van Bennekom, W. P.; De Bruijin, E. A.; Van Oosterom, A. T. J. Chromatop. 1982, 232, 355. (11) Treskes, M.; De Long, J.; Lceuwenkamp, 0. R.; Van der Vijgh, W. J. F. J. Liq. Chromatogr. 1990, 13, 1321. (12) Joyce, D. A.; Wade, D. N. J . Chromatogr. 1988, 430, 319. (13) Lloyd, I. B. F.; Parry, D. A. J. Chromatogr. 1988, 449, 281. (14) Smith, M. T.; Fluck, D. S.;Eastmond, D. A,; Rappaport, S.M. Life Chem. Rep. 1985, 3, 250. (15) Maitoza, P.; Johnson, D. C. Anal. Chim. Acta 1980, 118, 233. (16) Roston, D. A.; Shoup, R. E.; Kissinger, P. T. Anal. Chem. 1982,54. 1417A. (17) Radzik, D. M.; Brodbelt, J. S.;Kissinger, P. T. Anal. Chem. 1984,56, 2927. (18) Bergens, A. J . Chromatogr. 1987, 410, 437. (19) Haroon, Y.;Schubert, C. A. W.; Hauschka, P. V. J. Chromatogr. Sci. 1984, 22, 89. (20) Wallingford, R. A.; Ewing, A. G. Anal. Chem. 1989, 61, 98. (21) Curry, P. D.; Engstrom-Silverman, C. E.; Ewing, A. G. Electroanalysis 1991, 3, 358. (22) OShea, T. J.; Lunte, S.M. Anal. Chem. 1993, 65, 247. (23) Lu, W.; Cassidy, R. M. Anal. Chem. 1993, 65, 1649. (24) Lu, W.; Cassidy, R. M.;Baranski, A. S.J . Chromotogr. 1993, 640, 433. (25) O'Shea, T. J.; Greenhagen, R. D.; Lunte, S.M.; Lunte, C. E.; Smyth, M. R.; Radzik, D. M.; Watanabe, N. J . Chromafogr.1992, 593, 305. 0003-2700/94/0366-3782$04.50/0
0 1994 American Chemical Society
with oxidative detection. The exception to this is the work of Cassidy et al.,23,24who investigated the use of gold/mercury electrodes for the reductive end column detection of metal ions. However, in this case, the system was not deoxygenated, and there was a significant background current due to reduction of oxygen. The authors reported limited stability of the gold/ mercury disk electrode in this paper, which could be due to the oxidation of Hg by oxygen present in the run buffer. In the present work we set out to demonstrate the feasibility of reductive CEEC for organic compounds, including nitroaromatics and quinones. We have shown that the system possesses major advantages over reductive LCEC, including very short deoxygenation times and the fact that stainless steel plumbing is not necessary. We have applied the system to the separation and detection of dinitrophenyl (DNP)derivatized amino acids and a series of anthraquinones. Finally, the system was applied to the determination of an antitumor compound, mitomycin C, in human serum. To the best of our knowledge, this is the first report of reductive CEEC for the determination of organic compounds without oxygen interference.
EXPERIMENTAL SECTION Reagents and Materials. The (dinitropheny1)-L-amino acid derivatives, mitomycin C, 2-(N-morpholino)ethanesulfonic acid (MES), and human serum were all purchased from Sigma (St. Louis, MO). Anthraquinone-2-carboxylic acid and anthraquinone-2-sulfonicacid were obtained from Aldrich (Milwaukee, WI). (Sulfo butyl ether)-0-cyclodextrin was supplied by the Center for Drug Delivery Research, University of Kansas, Lawrence, KS. All other chemicals were analytical reagent grade, and all chemicals were used as received. Solutions were prepard in NANOpure water (SybronBarnstead, Boston, MA) and passed through a membrane filter (0.2-pm pore size) before use. The separation buffer consisted of 10mM MES containing 1 mM Na2EDTA, which was adjusted to pH 7 with sodium hydroxide. This was used as the electrolyte throughout the study. All stock solutions of the DNP-amino acids, anthraquinones, and mitomycin C were prepared daily in run buffer and stored at 4 OC. Ultrahigh-purity nitrogen was used for all deoxygenations throughout the study. Apparatus. The basic CEEC system has been described.2s Electrophoresis in the capillary was driven by a high-voltage supply (Spellman Electronics Corp., Plainview, NY). Polyimide-coated fused silica capillary columns (0.d. 360 pm, i.d. 50 pm) were obtained from Polymicro Technologies (Phoenix, AZ), and effective capillary lengths between 60 and 80 cm were used. Sample introduction was accomplished using a laboratory-built pressure injection system. The injection volume was calculated to be 9.8 nL when the continuous fill mode was used by recording the time required for the sample to reach the detector. The high separation current produced in the capillary was isolated from the detector using a Nafion joint.2s Briefly, a capillary cutter (Supelco, Bellfonte, PA) was used to score the polyimide coating end approximately 2 cm from the detection end of the capillary column. A l-cm length of Nafion tubing (i.d. 360 pm, 0.d. 510 pm) (Perma Pure Products, Tom’s River, NJ) was then carefully threaded over the score mark. Both ends of the Nafion tubing were sealed to the capillary using UV-Cure-glue (UVEXS, Sunny-
vale, CA) and mounted on a glass microscope slide using deposits of the same glue. The whole assembly was cured for ca. 20 min. Gentle pressure was applied to the cured Nafion tubing, causing the capillary to fracture at the score. The Nafion tubing holds the capillary joint in place and ensures correct alignment. The Nafion joint assembly was placed in a small plastic beaker that served as the cathodic reservoir. The detection end of the capillary was then inserted into the electrochemical detection cell. The cathodic buffer reservoir (ca. 2-mL volume) was filled with deoxygenated buffer and sealed with a rubber septum before the working electrode was inserted in the detection end of the capillary. Cylindrical carbon fiber microelectrodes were constructed using 33-pm-diameter fibers (AVCO Specialty Products, Lowell, MA) bonded to a length of copper wire using silver epoxy (Ted Pella, Inc., Redding, CA). Capillary tubes were pulled to a narrow tip with a Liste-Medical (Greenvale, NY) Model 3A vertical pipet puller. The microelectrode was then inserted through the capillary until it protruded approximately 0.5 cm from the tip. UV-Cure-glue was applied to the tip at the junction of the capillary and the carbon fiber. The fiber was drawn back until the desired length (150-300 pm) protruded, and the glue was cured under an ultraviolet light source. The opposite end of the capillary was then sealed with Thermogrip glue (Black & Decker) to fix the copper connecting wire in place. System Deoxygenation. Some modifications of the CEEC system were necessary for the reductive electrochemical detection (Figure 1). A laboratory-built system was employed which consistedof one nitrogen inlet tube and pressure adjustor leading to a set of four adjustable valves, each of which could be controlled separately to regulate the flow through its respective nitrogen delivery tube. Complete deoxygenation of the anodic buffer, the cathodic buffer, and the electrolyte was carried out off-line for 15 min at ambient temperature. Thecathodic cell surrounding the Nafion joint was filled with deoxygenated buffer and sealed with a rubber septum, through which the cathodic platinum electrode was introduced. Following this, the microelectrode was inserted to a depth of approximately 100 pm into the end of the capillary using an X-Y-Z micromanipulator with the aid of a microscope for visualization. The electrochemical cell was filled with deoxygenated electrolyte (buffer). Both the cathodic reservior and the electrochemical cell were housed in a box which could be loosely sealed. This box was lined with a series of nitrogen delivery tubes derived from inlet tube 1 (see Figure 1). Once the cathodic reservior and the electrochemical cell were filled with deoxygenated buffer, the box was closed and the nitrogen inlet tubes were switched on. This created a nitrogen atmosphere and a positive pressure inside the box, ensuring the expulsionof any atmospheric oxygen present. This positive pressure was maintained throughout the experiment, preventing the entry of oxygen. The anodic reservoir was then filled with deoxygenated buffer and sealed in position. Using an 18-gauge needle, a nitrogen delivery tube (tube 2, Figure 1) was introduced through the rubber septum which was used to seal the reservoir. A second needle inserted in the septum allowed a continuous outward flow of nitrogen. In this way a nitrogen blanket was maintained over the anodic buffer throughout the experiment. Once the high-voltage source Analytical Chemistry, Vol. 66, No. 21, November 1, 1994
3783
1
1I
I
I I -
Plexiglas
+safety box
tr viais Schematic representation of the reductive CEEC system. Lines 1-4 represent nitrogen delivery tubes, as described in the text. Flgure 1.
was switched on, the system stabilized within ca. 5 min at negative applied potentials. Samples were deoxygenated using nitrogen lines 3 and/or 4 for 10 min prior to analysis. The whole procedure of deoxygenation and setup took no more than 30 min, and once operational, the system could be used for 24 h. Cyclic Voltammetry. All cyclic voltammetry experiments were carried out on deoxygenated solutions using a Model CySy- 1 computerized electrochemical analyzer (Cypress Systems, Lawrence, KS). A three-electrode cell was used which consisted of a carbon fiber working electrode, a Ag/ AgCl reference electrode, and a platinum auxiliary electrode. All cyclic voltammograms were obtained in 10 mM MES pH 7 buffer using a scan rate of 100 mV/s. Hydrodynamic Voltammetry. Hydrodynamic voltammograms were obtained by repeated injections of standards while the working potential was stepped between -200 and -1 000 mV and the resultant peak currents were measured at 100mV intervals. The separation was performed in 10 mM pH 7 MES buffer with an applied voltage of 25 kV. The effective capillary length was 60 cm. Sample Preparation. Aliquots of 1 mL of serum were spiked with the appropriate amount of mitomycin C stock solution to achieve the desired final concentration. Dilution (1:4) of the serum was carried out using 10 mM sodium borate (pH 9) buffer. Final solutions were filtered through a membrane filter (0.2-km pore size) before analysis. It was necessary to deoxygenate the samplevery slowly to avoid excessive frothing of the serum. Complete deoxygenation was not essential since the analyte peak was resolved from the oxygen response. Samples were injected in the same manner described for standard solutions. RESULTS AND DISCUSSION Cyclic Voltammetry. The initial study involved cyclic voltammetric investigation of the behavior of the DNP amino acid derivatives on carbon fiber electrodes. Voltammograms were run in MES buffer at pH 7 since good separation of the amino acid derivatives was obtained at this pH by CEUV. 3784
Analytical Chemistty, Vol. 66,No. 21, November 1, 1994
C
i
/
/
800
380
40
-460
-880
-1300
Potential (mv) va, AglAgCI
Flgure 2. Cyclic voltammograms on a bare carbon fiber electrode of 2 X lo4 M DNPglycine In 10 mM MES buffer (pH 7). Curves A and C are voltammograms of the blank buffer before and after electrochemical pretreatment of the carbon fiber electrode, respectively. Curves B and D show voltammograms of 2 X lo-' M DNPglycine before and after electrochemical pretreatment of the carbon fiber.
The use of low-pH electrolytes is advantageous for reductive electrochemical processes, since the half-wave potentials of proton-dependent reductions will move to less negative potentials, introducing an improvement in selectivity. However, since the initial reduction of each nitro group is a 4H+/ 4e- process, there should be only a 15 mV/pH shift in reduction potential. Therefore, for this particular class of analytes, a substantial drop in pH would be necessary to have any significant effect on the half-wave potential. The use of low pH presents a problem in CE since the electroosmotic flow is severely attenuated at pH values below 7, resulting in excessively long migration times for the analytes. For this reason, a pH 7 buffer was used throughout. Figure 2 shows the cyclic voltammetric behavior of DNPglycine on a bare carbon fiber electrode before and after electrochemical pretreatment. It is obvious from these voltammograms that electrode activation is necessary. In all
further studies, the electrode was electrochemically activated before each injection using a 50-Hz square-wave pulse of 2 V amplitude for 30 s.26 The voltammogram of DNP-glycine shows two reduction processes at ca. -650 and -800 mVversus Ag/AgCl, respectively, which is in agreement with behavior reported for polynitroaromatic compounds using macroelectr~des.~,? HydrodynamicVoltammetry. Hydrodynamic voltammetry (HDV) using capillary electrophoresis and reductive electrochemical detection was performed for several DNP amino acids and anthraquinones and for the pH 7 MES background electrolyte. Figure 3 represents plots of the currents normalized against the maximum current response in each HDV versus the applied potential for (A) background electrolyte and anthraquinone-2-carboxylic acid, respectively, and (B) DNP-glycine. The HDV behavior of the DNP-glycine was in excellent agreement with the data obtained using CV. It can be seen from Figure 3A that at potentials lower than-600 mV there is a significant increase in the background current due to the reduction of residual oxygen. The maximum background current measured (1 1.1 nA) was observed at an applied potential of -1000 mV. At the potentials chosen as the working potentials for anthraquinones (-700 mV) and DNP amino acids (-800 mV), the background currents were 2.7 and 5.9 nA, respectively. The substantial decrease in background current due to deoxygenation is evident when these values are compared to the background currents measured in a non-deoxygenated buffer at the same potentials (72 and 100 nA). Without deoxygenation there was also a base line drift of ca. 1.4 nA/min, which made high-sensitivity measurements impossible. Under deoxygenated conditions this drift was less than 70 pA/min. We believe that this is the first report of the use of carbon fiber electrodes at such negative potentials. Carbon fiber electrodes are preferable to glassy carbon electrodes for use in CE due to their much lower background currents. A carbon fiber was chosen over a gold/mercury amalgam electrode for this work because it is more rugged and easier to insert. Figure 3A shows the HDV of the anthraquinone. The 2H+/2e- reduction occurs at a potential less extreme than that of the derivatized amino acids. On the basis of these studies, working potentials of -800 and -700 mV were used for subsequent reductive CEEC studies of the DNP amino acids and anthraquinones, respectively. DNP Amino Acids. The separation and reductive detection of DNP amino acids were carried out to demonstrate the utility of the system. Figure 4 shows the separation of a mixture of 20 pM each of DNP derivatives of glycine, GABA, serine, phenylalanine, and tyrosine. It is evident that even at -800 mV a reasonable base line is obtained, enabling the detection of these compounds. The change in base line at ca. 6.5 min is believed to be due to a discrepancy in 02 concentration between the sample and the run buffer. The linearity of this method was evaluated for a series of standards ranging from 1 to 10 pM (equivalent to 9.8-98 fmol injected) DNP-GABA. The slope (sensitivity) was 0.41 pM/nA with a regression coefficient r = 0.998 (n = 5 ) . The limit of detection was calculated to be 1.6 fmol at a S/N of 2, which is 3 orders of (26) Gonon, F. G.; Fombarlet, C. M.;Buda, M. J.; Pujol, J.-F. Anal. Chem. 1981, 53, 1386.
A
1
100
-
a0
-
8 c
a
60-
g,
2
E
40-
0
2
20
-
0.
-Id00
#
'
-800
,
'
,
'
-400
-600
I
-200
Working Potential (mV)
B 100
a0
4 -
8 60-
3 -0
w
z
40-
0 z
20
-
0'
,
-1000
*
,
,
,
-600
-800
'
,
-400
*
I
-200
Working Potential (mV) Flgure 3. Normalized hydrodynamic voltammograms using reductbe CEEC of (A) background buffer (A),10 mM MES (pH 7), and 10 pM anthraquinone-2-carboxyiic acid and (B)50 pM DNPglyclne (0). Separation voltage was 25 kV. See text for procedures.
m
magnitude lower than that reported by Jacobs and Kissinger, who used reductive LCEC.6 The concentration detection limit for DNP-GABA (0.16 pM) was comparable to that reported for the LC measurements. The reproducibility of the detector was studied by making several injections of a 98-fmol DNP-GABA standard and measuring the corresponding peak height of the reduction response. This yielded a relative standard deviation of 6.2% (n = 3). At higher concentrations (196 fmol), the reproducibility was better (ca. 4%, n = 7). The reproducibility could be further improved by employing an internal standard to minimize variations in injection volume, one of the precisionlimiting factors in CE.27 (27) Williams, S.J.; Goodall, D. M.; Evans, K. P. J. Chromorogr. 1993,629,379.
Analytical Chemlstty, Vol. 66, No. 21, November 1, 1994
3781
1
0.005 AUFS
0 2 4 6 8 l o l 8 l 4 I 6
Migration time ( m i d Flgure 4. Electropherogram of a mixture of 10 pM each of five DNP amino acids in 10 mM MES (pH 7) buffer. Peak identities: (1)tyrosine; (2) phenylalanine; (3) serine; (4) GABA; and (5) glycine. Separation voltage was 20 kV. An applied working potential of -800 mV was used.
B
2
/-----
A
9 L
8
.
.
0.5 nA
I
.
12
I
16
20
Time (min)
Flgure 6. Electropherograms of blank serum using (A) reductive
0
2
4
6
8
1 0 1 2
14
1 6 1 8 2 0
23
Migration time ( m i d Figure 5. Electropherogram of 10 pM (98 fmol injected) each of (1) anthraquinone-2-carboxylicacid and (2) anthraquinone-2-sulfonic acid. Separation voltage was 15 kV. An applied working potentialof -700 mV was used.
Anthraquinones. The reduction of the quinone moiety to hydroquinone occurs more readily than reduction of the nitro group, so the less extreme potential of -700 mV was employed. More stable base lines were obtained at this working potential. Figure 5 shows the reductive electrochemical detection of a mixture of 10 pM anthraquinone-2-carboxylic acid and anthraquinone-2-sulfonic acid. The electropherogram demonstrates a good resolution and a stable base line, reinforcing the effectiveness of the system for reductive detection without interference from oxygen. The linearity of response was 3786
Analytical Chemistty, Vol. 66,No. 21, November 1, 1994
electrochemicaldetectionat an applied working potentialof -700 mV and (6) UV detection at a wavelength of 216 nm. A sodium borate buffer (pH 9) containing 5 mM (sulfo butyl ether)-&cyciodextrin was used. Separation voltage was 15 kV.
evaluated for a concentration range of 1-10 pLM (equivalent to 9.8-98 fmol injected) anthraquinone-2-carboxylic acid and yielded a slope of 0.293 pM/nA with a regression coefficient r = 0.999 (n = 6). The linearity was also evaluated for anthraquinone-2-sulfonic acid over a wider concentration range between 5 and 200 pM (equivalent to 0.05-1.2 pmol injected). In this case, the regression coefficient was r = 0.995 (n = 10). The reproducibility of the detector response for a 1 pM solution of anthraquinone-2-carboxylic acid was 6.7% (n = 3). The mass limit of detection was determined to be 1.3 fmol of anthraquinone-2-carboxylic acid based on S/N = 2. The concentration limit of detection for the same compound was calculated as 0.13 pM. These results demonstrate that the sensitivity of this system is as good as or better than that of reductive LCEC.
H
0
12
8
16
20
Time (min)
Figure 7. Electropherogram of human serum containing 10 pg/mL mkomycin C using electrochemical detectlon at an applied working potentlaiof -700 mV. A sodium borate buffer (pH 9) containing 5 mM (sulfo butyl ether)-&cyclodextrin was used. Separation voltage was 15 kV.
Mitomycin C Determination in Human Serum. Mitomycin C, an antineoplastic agent produced by Streptomyces caespitosus, contains the quinone moiety within its structure. A good response was obtained for this compound when CEEC was used at an applied potential of -700 mV. The response was linear between 9.8 and 98 fmol injected with a regression coefficient of r = 0.999 ( n = 5) and a sensitivity of 0.223 pM/nA. The mass limit of detection under the same conditions was found to be 0.77 fmol based on S / N = 2. For this compound, the concentration limit of detection was about 1 order of magnitude lower than that obtained using reductive LCEC with a static mercury drop electrode.1° However, the mass detection limit was 3 orders of magnitude lower. In initial studies using pH 7 buffer, mitomycin C eluted around the same time as the system peak due to oxygen. This peak is more pronounced in serum because of the difficulty of sample deoxygenation. To enhance the separation, (sulfo butyl ether)P-cyclodextrin was added to the run buffer to increase the migration time of mitomycin C relative to the system peak. This highly soluble form of cyclodextrin has been shown previously to complex with aromatic compounds.28 The final separation was achieved using a sodium borate (pH 9) buffer containing 5 mM (sulfo butyl ether)-0-cyclodextrin. Figure 6 compares the selectivity of (A) reductive electrochemical detection at an applied working potential of -700 mV with (B) UV detection at a wavelength of 216 nm under the same conditions for a blank serum sample. It is obvious that the strong absorbance of endogenous serum compounds at 216 nm makes selectivedetectionof mitomycin C impossible. In contrast, the blank serum shows little response when reductive electrochemical detection is employed. The peak seen at approximately 12 min in Figure 6 A is endogenous to the serum samples. The large negative peak seen at approximately 13 min in the same figure is due to the discrepancy ~~~
~~~
~
~~
~~
~~
~~
(28) Tait, R.; Skanchy, D. J.; Thompson, D. P.; Chetwyn, N. C.; Dunshw, D. A,; Rajewski, R. A.; Stella, V.J.; Stobaugh, J. F. J . Pharm. Biomed. Anal. 1992, IO, 615.
between the amount of residual oxygen present in the run buffer and that in the sample. Figure 7 shows an electropherogram of a human serum sample containing 10 pg/mL of mitomycin C, which is readily detected by CEEC without interference from endogenous compounds. The same level could not be detected by CEUV at 216 nm because of endogenous interferences. Spectroscopic detection of 10 pg/ mL of mitomycin C at 360 nm was also attempted because this approach has been shown to be more selective than detection at 216 nm. However, due to the low molar absorptivity of the compoundat that wavelengthand thesmall path length characteristic of the fused silica capillary, mitomycin C could not be detected a t that level. The selectivity and sensitivity of reductive electrochemical detection in biological samples is convincingly demonstrated by this example. In addition, the electrode showed no significant loss in response after eight consecutive injections of serum samples. For the serum sample containing 10 pg/ mL mitomycin C, the peak height reproducibility was 7.8% ( n = 3). In the future, we hope to apply this technique to the study of the pharmacokinetics of mitomycin C using microdialysis sampling. CONCLUSIONS We have shown the utility of CE with reductive electrochemical detection and have outlined its major advantages over LC with reductive detection, particularly in terms of rapid and simple setup and ease of operation. The deoxygenation time is substantially shorter for CEEC (15 min) compared to LCEC (8 h), and there is no need to replumb the CEEC system with stainless steel or to reflux the mobile phase. In addition, the concentration detection limits are very close to those obtainable by LCEC, making CEEC an extremely sensitive and selective technique. CEEC provides additional advantages in those cases where one is sample-limited, such as microdialysis ampl ling^^.^^ or analysis of single cells.31J2 With CE, it is also possible to analyze a single 1-pL sample several times using different separation and detection conditions without a noticeable loss of sample volume. This is in contrast to LC, where sample volumes of 1-20 pL are typically required for a single analysis. Through a comparison of CEEC with CEUV, we have demonstrated the sensitivity and selectivity of the system for the analysis of a reducible compound in a biological sample. We are confident that this technique will be applicable to many other compounds of biological or environmental interest in the future. ACKNOWLEDGMENT The authors acknowledge Bioanalytical Systems, West Lafayette, IN, for supplying the LC4C detector used in this work. The authors also thank Nancy Harmony for assistance in the preparation of this manuscript and Dr. Thomas J. OShea for helpful discussions. Support for this work was provided by the Center for Bioanalytical Research and The Kansas Technology Enterprise Corp. Recelved for review April 18, 1994. Accepted July 28, 1994. (29) O’Shea, T. J.; Weber, P. L.; Bammell, B. P.; Smyth,M.R.; Lunte, C. E.; Lunte, S . M. J . Chromarogr. 1992, 602, 189. (30) OShea, T. J.; Telting-Diaz, M.;Lunte, C. E.; Lunte, S. M.Electroanalysis 1992, 4, 463. (31) Hogan, B. L.; Yeung, E. S . Anal. Chem. 1992, 64, 2841. (32) Ewing, A. G. J. Neurosci. Methods 1993, 48, 215. e Abstract published in Advance ACS Absrracrs, September 15, 1994.
Analytical Chemistry, Vol. 66,No. 21, November 1, 1994
3707
