Articles Anal. Chem. 1994,66, 596-602
On-line Coupling of in Vivo Microdialysis Sampling with Capillary Electrophoresis Barry L. Hogant and Susan M. Lunte Center for Bioanalytical Research, University of Kansas, Lawrence, Kansas 66047 John F. Stobaugh Department of Pharmaceutical Chemistry and Center for Bioanalytical Research, University of Kansas, Lawrence, Kansas 66045 Craig E. Lunte' Department of Chemistry and Center for Bioanalytical Research, University of Kansas, Lawrence, Kansas 66045
Microdialysis sampling has become an important means of continuously monitoring reactions in vivo. This sampling technique places a constraint on the analysis method because of the very small sample volume provided. On the other hand, microdialysis provides the advantage of clean samples that do not require cleanup prior to analysis. An on-line coupling of microdialysis sampling to capillary electrophoretic (CE) analysis is described that uses the advantages of microcolumn separations to overcome the small volume limitation. An interface was designed which converts the continuous microdialysis sample stream into discrete 60-nL sample plugs and then injects a portion of this plug into the CE system. The on-line interface provided precision of 2.6% with minimal band broadening or peak height loss relative to off-line sampling. Using a high-speed micellar electrokinetic chromatography (MEKC) separation, resolution of the investigational antineoplastic SR 4233 from its main metabolite SR 4317 was achieved in less than 60 s. This allowed the on-line system to achieve a 90-s temporal resolution for determining the pharmacokinetics of SR 4233 in vivo. Dynamic chemical information about the processes occurring in living systems is of great interest in the biological sciences. Elucidation of the pathways and kinetics of absorption, biotransformation, and elimination is needed to fully assess the safety of both pharmaceutical and environmental compounds. Pharmacokinetic parameters for the parent compound and any metabolites are needed both systemically and at specific tissue sites. The time scale of these reactions is typically on the order of minutes to days. Neuropharmacological investigations, such as the release of neurotransPresent address: Department of Chemistry, Saint Louis University, St. Louis, MO 63103.
596
Analytical Chemistry, Vol. 66,No. 5, March 1, lg94
mitters in response to amphetamine' or cocaine? require acquisition of transient chemical information from specific brain regions but with temporal resolution of seconds to minutes. Current methods for obtaining this type of chemical information in tissues of living systems have involved either postmortem analysis at several time points or use of biosensors implanted in vivo. Theuse of postmortem analysis toconstruct a temporal plot of chemical events is difficult and often provides ambiguous results. While several compounds can be determined at each time point, each animal can provide data for only a single time point. It is obviously preferable in terms of both data quality and reduction in use of experimental animals to obtain the entire time course from a single animal. This can be accomplished through the use of biosensors but at the cost of less chemical information. Biosensorscan provide data during the entire reaction but are usually limited to monitoring a single chemical species. Microdialysis sampling is accomplished by implanting a small, semipermeable fiber at the site of interest. This fiber is slowly perfused with a sampling solution. Small molecules in the extracellular space diffuse into the fiber and are swept away to be collected for analysis. Microdialysis probes can be implanted in many tissues with minimal discomfort to the experimental animal. The introduction of microdialysis sampling has provided a technique which can continuously monitor chemical reactions in vivo. By use of the appropriate analytical method, several compounds can be determined simultaneously. Microdialysis therefore can provide both the temporal and chemical information needed to fully elucidate biochemical processes. (1) Adams, F.;Schwarting, R. K. W.;Boix, F.; Huston, J. P.Brain Res. 1991, 553, 318-322. (2) Kuczcnski, R.; Segal, D. J. Neurosci. 1989, 9, 2051-2065.
0003-2700/94/036&0596$04.50/0
0 1994 American Chemical Society
A limitation of microdialysis has been the temporal resolution that can be achieved. While microdialysis is a continuous sampling technique, it is typically coupled to separation methods that require discrete samples. The temporal resolution of the overall experiment is determined by a combination of the perfusion rate through the microdialysis probe and the sample volume requirement of the analytical technique. Microdialysis samples are by necessity aqueous and the analytes typically of small molecular weight and moderate to high water solubility. For these reasons, liquid chromatography (LC) has been the most popular analytical technique to couple to microdialysis sampling. Even with the use of 1-mm-i.d. microbore columns at least 1 pL of sample is needed. At a typical microdialysis perfusion rate of 1 pL/min, the maximum temporal resolution is only 1min. Higher perfusion rates can be used to generate larger sample volumes per unit time, but result in lower concentration recoveries, placing strain on the detection limits of the analytical method. Coupling microdialysis to a microseparation method such as capillary electrophoresis (CE) in which sample volumes of a few nanoliters are sufficient can obviously provide improved temporal resolution. Analysis of microdialysis samples by CE has been described.3-5 These initial studies used off-line collection of the microdialysis sample and subsequent analysis by CE. Although only 5 nL was ultimately injected into the CE, 5-pL samples of the dialysate were still collected. The reason for this was that quantitative handling of volumes less than 1 pL is too difficult in an off-line system. In particular, evaporation is a significant problem. An advantage of microdialysis sampling is that the dialysate is protein-free and amenable to direct injection into an LC or CE system. Therefore, microdialysis sampling can be coupled to these analytical techniques in an on-linefashion to avoid the problems of sample manipulation. On-line microdialysis sampling coupled to LC,687flow injection analysis,8and mass spectrometrygJOhas been described. However, the best temporal resolution reported to date has been 2 min. This report describes the on-line coupling of microdialysis to a micellar electrokinetic chromatography (MEKC) separation. CE has been coupled to LC in a multidimensional separation system" and to an FIA system.12 While similar in concept, the requirements of the microdialysis experiment and LC separation are different. The heart of the on-line system is the interface between the microdialysis apparatus and the CE. The interface must isolate the experimental animal from the high potential of the CE experiment and convert the microliter per minute microdialysis flow into discrete nanoliter volume samples for the CE while not adding
to the system's dispersion. This has been accomplished by using a commercially available rotary microinjection valve for LC with an injection apparatus designed in-house. CE analysis of dialysis samples has been accomplished in under 60 s. Therefore a 90-s time resolution for monitoring a biochemical process in vivo is possible.
EXPERIMENTAL SECTION Reagents. 3-Amino- 1,2,Cbenzotriazine 1,Cdi-N-oxide (SR 4233) and 3-amino-l,2,4-benzotriazine1-N-oxide (SR 43 17)weregiftsfromsterling Drug (Rochester,NY). Sodium dodecyl sulfate (SDS) was obtained from Aldrich (St. Louis, MO). All other chemicals were reagent grade or better and used as received. Capillary ElectrophoresisSystem. The CE system consisted of a CZE lOOOr high-voltage power supply (Spellman, Plainview, NY). Capillaries of 50 pm i.d., 360 pm o.d (PolymicroTechnologies, Pheonix, AR) having a total length of 40 cm and a length from injection to detection of 15 cm were used. A timer box constructed as described previously3 was used to make off-line electrokinetic injections. The injection end of the capillary was grounded and the detection end held at a large negative potential. The high potential end was housed in an isolation box. The MEKC running buffer consisted of 100 mM sodium phosphate buffer, pH 6.8, with 40 mM SDS as a micellar additive. Columns were flushed with 0.1 M NaOH prior to use. A run potential of -25 kV was used to achieve separation. Coupled CE columns were constructed to permit highspeed MEKC separations. A 25-cm section of polyethylene tubing (0.8 X 1 mm.) was glued onto the CE capillary 5 cm past the detection point. The polyethylene tubing was then filled with running buffer and inserted into the buffer reservoir in the high-voltage isolation box. High voltage can thus be applied across very short sections of capillary without a loss of voltage gradient. Detection was by laser-induced fluorescence (LIF) using a He/Cd laser (Model 4300, Liconix, Santa Clara, CA) with 5-7 mW of power at 442-nm for excitation. A 442 nm interference prefilter (Edmund Scientific, Barrington, NJ) was used for rejection of laser plasma. A 1-cm focal length biconvex lens (Melles Griot) was used to focus the beam within the capillary. A 20X microscope objective (Edmund Scientific) was used to collect the fluorescent image within the capillary. The image was filtered spatially through the use of an aperture placed at the image focal plane to reject scattered light cdlected from the capillary walls. The image was then filtered spectrally with a long-pass filter (520 nm, Schoeffel) and an interference filter (580 nm, Melles Griot). The (3) OShea, T. J.; Telting-Diaz, M.W.; Lunte, S.M.; Lunte, C. E.; Smyth, M. photomultiplier was an R1527 (Hamamatsu, Bridewater, NJ) R. Electroanal. 1992, 4, 463468. operated at -1000 V supplied by a regulated power supply (4) O'Shea,T. J.; Weber,P.L.;Bammel,B.P.;Lunte,C.E.;Lunte,S.M.;Smyth, M.R. J. Chromofogr. 1992,608, 189-195. (Pacific Photometric, Model 227). ( 5 ) Tcllez, S.;Forges, N.; Roussin, A.; Hernandez, L. J. Chromatogr. Biomed. Appl. 1992, 581, 257-266. Data collection was accomplished with an integrated (6) Damsma, G.; Westernik, B. H. C.; de Vries, J. B.; Van den Berg, C. J.; Horn, hardware/software package (Chemresearchversion 2.4, Isco, A S . J. Neurochem. 1987, 48, 1523-1528. (7) Church, W. H.; Justice, J. B., Jr. Anal. Chem. 1987, 59, 712-716. Lincoln, NE) running on a Tandon AT compatible micro(8) Boutelle, M. G.; Fellows, L. K.; Cook, C. Anal. Chem. 1992, 64, 1790-1794. computer. The 12-bit board collected data at 20 Hz and (9) Caprioli, R. M.;Lin, S.-N. Proc. Nafl. Acod. Sci. U.S.A. 1990, 24C-243. (10) Deterding, L. 3.; Dix, K.; Burka, L. T.; Tomcr, K. B. And. Chem. 1992,64, controlled the valve switching. Signal amplification, variable 2636-2641. offset and rc filtering (typically 300 ms) was accomplished (11) Bushey, M. M.; Jorgenson, J. W. And. Chem. 1990,62, 978-984. (12) Tsuda, T.; Zare, R. N. J . Chromotogr. 1991, 559, 103-110. with a locally constructed circuit. AnaiyticalChemistty, Vol. 68, No. 5, March 1, 1994
597
Syringe PumD #2 Transfer Line
A
Flgwe2. Schematicof the InjectionInterface.Components: A, transfer capillary; B, CE capillary;C, CE run buffer reservoir; D, CE run buffer; E, CE ground electrode; F, microscope slide: G, guide tubing. Flgwe 1. Sct"atic of the on-line microdlalysislcaplllary eleCh.0phoresls system.
MicrodialysisSystem. Microdialysiswas carried out using a CMA 100 microinfusion pump from Bioanalytical Systems, Inc./CMA (West Lafayette, IN) coupled to a flexible microdialysis probe. The microinfusion pump was operated at a flow rate of 1 pL/min. The construction of the flexible microdialysis probe has been described e1~ewhere.l~The dialysis membrane was a cellulose fiber with 232 pm i.d., 250 pm o.d., and a molecular weight cutoff of 5000 (Dow Chemical, Midland, MI). Connection from the microinjection pump to the microdialysis probe was made with 120-pm4.d. Teflon tubing. On-Line Microdialysis to CE Interface. The interface between the microdialysis system and the CE system must perform two functions. First, the continuous microdialysis samples must be converted to discrete samples for electrophoretic analysis. Second, the microdialysis flow rate must be converted to a rate compatible with CE. These functions are performed by different components of the interface. The dialysate is sampled using a microinjector for liquid chromatography while the flow conversion is accomplished by an injection interface designed in-house. The complete system is shown in Figure 1. The sampling valve was an electrically actuated internal sample loop HPLC valve (Model E2NI4W-06, Valco Instruments, Houston, TX) with a 60-nL internal sample loop. Transfer of the sample from the microinjection valve to the injection interface was accomplished using a second CMA 100 microinfusion pump. The transfer flow rate was varied between 1 and 10 pL/min with the optimum found to be 5 pL/min. A 26-cm length of 50-pm4.d. fused silica tubing was used as a transfer line from the microinjection valve to the injection interface. The injection interface consisted of a reservoir containing the electrophoretic run buffer into which fused silica capillaries entered from opposite sides (Figure 2). One of these was the CE capillary column and the other was the transfer line from the microinjection valve. The capillaries were aligned such that the out flow of the transfer capillary impinged directly on the inlet of the CE capillary. Reproducible alignment of the two capillaries in the injection interface was accomplished by the use of four guide tubes permanently secured in place following prealignment. Four 5-mm lengths of 0.5 X 1.5 mm polyethylene tubing (Alltech) were cut for use as alignment guides. These guides were placed over a 7-cm section of 360(13) Telting-Diaz, M. W.; Scott, D. 0.;Lunte, C. E. Anal. G e m . 1992,64,806810.
598 AnalytkalChemlstry, Vol. 66, No. 5, March 1, 1994
pm-0.d. capillary to align the guide tubes. The assembly was placed on a microscope slide with the center two guide tubes separated by 2 cm and the outer guide tubes separated by 4 cm. The guide tubes were permanently attached to the microscope slide with epoxy. After all of the guides were fixed in place, the length of capillary used for initial alignment was removed. Epoxy was then used to build the buffer reservoir with dimensionsof 2 cm X 2 cm and sufficient depth to cover the guide tubes and capillaries. The transfer line from the microinjectionvalve was brought into one side of the injection interface through two of the guide tubes and positioned so its orifice was at the midpoint of the buffer reservoir. Epoxy applied to the capillary behind the rear guide tube fixed its position. The CE capillary was brought into the interface through the remaining two guide tubes after having been aligned and fixed with respect to the detection optics. The alignment of the CE column in two dimensions with respect to the transfer capillary was controlled by the guide tubes. The gap between the capillaries was established manually with the use of an 8X magnifier (Radio Shack). The precise gap distance between the transfer capillary and the CE capillary can be measured optically with a measuring reticle. Estimation of the gap distance by comparison with the known diameter of the CE column produced good results in noncritical cases. Once correctly positioned, the CE capillary was permanently affixed to the outer guide tube with epoxy. Probe Characterization. It was necessary to determine the recovery characteristicsof the microdialysisprobe in order to determine the in vivo concentrationsof SR 4233. This was done by placing the dialysis probe in a stirred solution of known concentration of SR 4233 maintained at 37 OC. The microdialysis probe was perfused at 1 pL/min, and samples were collected in the loop of the sampling valve. These were injected automatically into the CE system, which had been previously calibrated with known concentrationsof standards by flow injection analysis. Recovery was calculated at several concentrations within the dosing range used and was found to be independent of concentration. The recovery value for a given probe was then used to calculate the sample concentrations from the dialysate concentrations for microdialysis sampling experiments. In Vivo Pharmacokinetics. Male Sprague-Dawley rats weighing 0.39-0.43 kg were used. The rats were anesthetized with a 3/ 1 mixture of ketamine/xylenol injected ip. Surgery for the insertion of the probe has been described previously,13 although in this instance the PE-50 dosing cannula was implanted into the jugular vein opposite the vein in which the
microdialysis probe was implanted. The implanted probe was perfused with Ringer’s solution at a flow rate of 1 pL/min. Dialysis samples were collected for intervals of 90-120 s. Blanks were collected from the animal for at least 1 h prior to dosing. The animals were then administered 4 mg/kg S R 4233 as an iv bolus in 0.5 mL of Ringers solution through the dosing cannula. Dialysis sampling was continued for at least 1.5 h after administration of SR 4233.
RESULTS AND DISCUSSION On-Line Interface Operation. The interface between the microdialysis sampling system and the CE analysis system must perform a variety of functions. It must convert the continuous sampling stream of the microdialysis system into discrete samples for CE analysis. The interface must also convert the microliter volumes of the microdialysis sample to nanoliter volumes for CE. Finally, the interface must shield the experimental animal from the high voltages of the CE system. The on-line interface described here performs all three functions. The microdialysis sample flows directly from the experimental animal into the sample loop of a microinjection valve. This flow is controlled by a microinfusion pump shown as pump 1 in Figure 1. The microdialysis flow rate of 1 pL/min is converted into a 60-nL plug by the microinjection valve. This sample plug is injected into the transfer line at timed intervals. The sample plug is transported to the injection interface by the flow of CE run buffer controlled by the transfer microinfusion pump (pump 2 in Figure 1). Pump 2 is continuously pumping CE run buffer through the transfer line into the CE run buffer reservoir. The sample plug exits the transfer line into the CE run buffer reservoir directly across from the injection end of the CE capillary. The run potential is constantly applied to the CE system such that buffer is continuously being drawn into the CE from the run buffer reservoir and by electrosmotic force. Injection of the sample plug into the CE system is accomplished because it is captured by the separation capillary as it passes near the orifice. Any sample not captured by the CE is swept away from the injection region by the run buffer pumped through the transfer line behind the sample plug. Proper operation of the injection interface requires careful positioning of the transfer capillary relative to the CE capillary. The open design of this interface allows for ready control and optical determination of positioning. In addition, the run buffer reservoir can be arbitrarily large. A relatively large reservoir volume provides enhanced CE run buffer stability and system reproducibility. Fresh CE run buffer is continually being pumped into the reservoir through the sample transfer line at a rate of 5 pL/min. With a 90-s interval between injections, 60 nL of sample is followed by 7.5 p L of buffer. Therefore, the injection interface is well-flushed between injections and the composition of the CE run buffer in the reservoir does not change as samples are injected. Interface Optimization. Several operational parameters were investigated in relationship to their effect on CE performance in terms of peak response and shape. These included the transfer capillary positioning, transfer flow rate, and transfer capillary inner diameter. In addition, operation of the microinjection valve was found to play a critical role
Table 1. CE Peak Parameters as a Functlon of Injection Gap
gap (pm) peak height (afuo) peak widthb (8) 50 100 300
1312 f 23 1416 f 29 690 f 118
2.82 f 0.08 2.86 f 0.07 5.1 f 0.63
peak symmetrp 0.77 f 0.03 0.78 f 0.03 0.80 f 0.04
a Arbitrary fluorescence units. Width at half-height. e Peak symmetry is defined as the ratio of area before the peak to the area after the peak.
Table 2. CE Peak Parameters as a Function of Transfer Flow Rate
flow rate (pLlmin) peak height (afu‘) 1.0 3.0 5.0 10.0
2532 f 186 1826 f 278 1437 f 134 1075 f 60
peak widthb (8)
peak symmetw
10.27 f 0.82 4.36 f 0.57 2.82 f 0.10 2.06 f 0.06
0.76 f 0.02 0.77 f 0.03 0.76 f 0.02 0.75 f 0.03
Arbitrary fluorescence units. Width at half-height. e Peak symmetry is defined as the ratio of area before the peak to the area after the peak.
in CE peak shape. The experimental antineoplastic SR 4233 was used as a test compound for these evaluations. Transfer Capillary Positioning. Thedesign of the injection interface using four guide tubes effectively aligns the transfer capillary with the CE capillary in two dimensions. The effect of varying the third dimension, the distance between the capillaries or interface gap, was studied at a transfer flow rate of 5 pL/min with the results as shown in Table 1. As the gap increases in distance, both peak height and peak shape deteriorate. Operation of this injection interface requires efficient transfer of the sample plug across the interface gap into the region being sampled electrosmotically by the CE. From the data of Table 1, this transfer is efficient with gaps less than 100 pm. At larger gap distances, the sample plug disperses into the run buffer reservoir and is sampled less efficiently, resulting in smaller CE peak heights. The CE peak is also much broader as the dispersed sample is less efficiently swept away from the interface by the following transfer buffer. An interface gap distance of 50 pm was used for all subsequent experiments. This distance is easily achieved and provides good operation. Smaller gap distances can provide marginally better operation but require more care in fabrication. Transfer Buffer Flow Rate. The second parameter investigated was the flow rate of the transfer buffer. The transfer buffer flow has two effects at the injection interface. The first effect is the transport of the sample plug into the CE electrosmotic injection region. The second effect is the removal of uninjected sample from the injection region. As seen from the data in Table 2, CE peak height decreases while peak shape improves as the transfer flow rate increases. Using a 50-pm interface gap, transfer from the transfer capillary into the CE injection zone appears to be extremely efficient even at the lowest transfer flow rates. In this case, the residence time of the sample plug in the injection zone determines the peak height rather than dispersion of the sample. The residence time of the sample plug also determines the CE peak shape. As the residence time increases, the peak width increases as a larger sample volume is injected into the CE AnalyticalChemistry, Vol. 66,No. 5, March 1, 1994
599
l
0
.
I
20
.
I
40
.
I
C
B
A
.
60
Tim (sec)
I
80
I
1
0
.
1
20
.
I
40
'
I
'
60
I
I
80
I m (sec)
I
0
'
"
20
'
I
40
.
'
60
Tim Qc)
I
I
80
0
"
.
20
'
"
40
'
I
I
60
80
Tim (sec)
Figure 3. Electropherogramsobtained usingdifferent peak clipping times of the microinjection valve. Electrophoreticconditions were as described in the text. The valve injection or peak Clipping times were A, 1 s; B, 3 s; C, 7 s; and D, 15 s.
system. A faster transfer flow more efficiently sweeps the sample plug past the injection zone. A compromise must therefore be made between C E peak height and peak shape. A transfer flow rate of 5 pL/min results in a decrease in CE peak height of less than 50% relative to a 1 pL/min transfer flow rate but an improvement in peak width of over 3-fold. Faster transfer flow rates result in only marginal improvements in peak shape but large decreases in peak height. A transfer flow rateof 5 pL/min was used in all subsequent experiments. Transfer Capillary Inner Diameter. Transfer capillaries of 25 and 50 pm were evaluated. The general performance of the interface was the same for both diameters except that the smaller diameter capillaries are less sensitive to the interface gapdistance. The 50-pm capillary showed decreased performance at 300 pm while the 25-pm capillary showed no decrease at this gap distance. However, the 25-pm transfer capillary was more prone to clogging during the extended operation required of this system. The pharmacokinetic experiments may last for several hours, during which time the system remains in continuous use. Any extended operational difficulty results in loss of the entire experiment; therefore, system reliability is of extreme importance. Transfer capillaries with inner diameters of 50 pm were chosen for optimal performance and reliability. Larger capillaries require smaller transfer gap distances with only marginal improvements in reliability while 25-pm capillaries provide little improvement in performance but are considerably less reliable. Microinjection Valve Operation. In studying the CE peak shape during the interface optimization, it was noted that all of the C E peaks exhibited considerable tailing. Optimization of the interface parameters improved the CE peak widths at half-height but did not decrease the tailing. It was ultimately determined that the peak tailing resulted from dead volumes associated with the microinjection valve. With a sample loop 600
AnalyticalChemistry, Vol. 66,No. 5,March 1, 1994
of only 60 nL, the exit port contributes a significant volume to the system. The commercial valve used in these experiments had an exit port diameter of 0,010 in., which results in a port volume of about 90 nL. This dead volume results in the tailing observed in initial experiments. To minimize the peak tailing, the operation of the microinjection valve was modified. Initially, the valve was allowed to remain in the inject position for a relatively long time (i.e., 15 s). By decreasing the time the valve is in the inject position, the effect of the exit port dead volume can be minimized. Figure 3 shows the effect of this peak "clipping" on the C E peak shape and peak height. Decreasing the injection time dramatically decreases the peak tailing with no effect on peak height until injection time is less than 3 s. A peakclipping time of 3 s was used in all subsequent experiments. It should be noted that specially drilled microinjection valves are now available which may eliminate much of the dead volume encountered with the current valve. Comparison to Off- Line Injection. The optimized on-line system was compared to off-line electrokinetic injection. Offline injections of 1-5 s at 25 kV were made. Comparing the peak height of samples using on-line injection to off-line injection showed a dilution factor of 4 for on-line injection. Comparison was made to off-line injections of 2-s duration as these gave efficiencies similar to on-line injections. System Response. Because this system is designed to study transient signals in vivo, the response time of the system is a critical parameter. The response of the system to a concentration change was first examined by flow injection directly from the dialysis pump (pump 1) into the microinjection valve. This was accomplished by replacing the microdialysis probe with a switching valve. Pump 1 can control up to three syringes simultaneously. Therefore, it was possible to instantaneously switch between solutions of two different concentrations of SR4233 and a blanksolution. This allowed thedetermination
I
0
600
1 2 0 0
1 8 0 0
T im e ( s e c o n d s )
Figure 4. Response of the interface system to a step change in concentrationusing flow injection analysis. Electrophoreticconditions were as described In the text. The concentrations of SR 4233 were 0 , 2 , and 10 bg/mL.
A
0
10
20
30
40
50
60
10
80
nme (seconds)
!I
Figure 8. Typical electropherograms of blood dialysates using the on-line microdiaiysWCE system. Electrophoretic conditions were as described in the text. Samples correspond to a blank prior to dosing (A) and 10 min after a 4 mg/kg ip dose of SR 4233 (B). Peak identities are peak 1, SR 4233 and peak 2, SR 4317.
I
I u I
I
I
I
I
I
0
200
400
600
800
1000
4
Time (seconds) Figure 5. Response of the interface system to a step change in concentrationusing microdialysissampling. Electrophoreticconditions were as described in the text. The concentrations of SR 4233 were 0 ,6 , and 10 bg/mL.
of the response time of the interface alone without contributions from microdialysis sampling. The resulting step function response is shown in Figure 4. With injections being made every 90 s, the response to an input concentration change is essentially instantaneous. This result was to be expected because, at the dialysis flow rate of 1 pL/min, the 60-nL sample loop is filled in less than 4 s. The peak height precision was 2.6% RSD. The response time was then investigated in the complete microdialysis sampling mode. In this case the microdialysis probe was used to sample a test solution. Three small beakers containing solutions of two different concentrations of SR 4233 and a blank were used. To create the concentration steps, the microdialysis probe was rapidly switched from one beaker to another. The step function response resulting from this experiment is shown in Figure 5. Again, an instantaneous response was obtained on the time scale of the CE step. The small peaks seen in Figure 5 after stepping back to the blank solution are an artifact of the manual switching of the microdialysis probe between beakers. The peak height precision was again found to be 2.6% RSD, indicating that the microdialysis sampling did not significantly affect system performance.
10
20
30
40
50
Time (minutes) Flgure 7. Typical pharmacokinetic curve from a 4 mg/kg ip dose of SR 4233 obtained using the on-line microdialysis/CE system. The bars represent the microdialysis data while the curve is the biexponential fit to the data.
Determinationof SR 4233 Pharmacokinetics. To evaluate the performance of this system for an actual pharmacokinetic experiment, the antineoplastic SR 4233 was investigated in the rat. Typical electropherograms of blood dialysates both before and after dosing with SR 4233 obtained using the online microdialysis sampling/CE detection system are shown in Figure 6. SR 4233 can be resolved from its main metabolite SR 43 17 in 60 s. No interferences are observed in the blank. Using this separation, microdialysis sampling could be continuously accomplished on-line with a 90-s sampling interval. A typical pharmacokinetic curve for a 4 mg/kg iv injection of SR 4233 obtained with this system is shown in Figure 7. In addition to the usual elimination phase, the rapid distribution phase is also observable. From a nonlinear least squares fit of the data, the half-life of elimination was determined to be 15.3 f 1.Omin and the half-life of distribution was determined to be 1.1 f 0.2 min ( n = 3 in both cases). It AnalyticalChemistry, Vol. 66, No. 5, March 1, 1994
601
is only because of the rapid sampling and analysis capabilities of this system that the distribution kinetics could be determined.
CONCLUSIONS Microdialysis sampling provides a powerful tool for monitoring biochemical reactions in vivo. The temporal resolution of the microdialysis experiment has been limited by the sample volume requirement and/or detection limit of the associated analysis technique. Refinement of the microdialysis experiment can be realized by moving the analysis step on-line to eliminate any sample manipulation. However, this limits the temporal resolution of the experiment to the speed of the analysis step. This work demonstrates the high temporal resolution that can be achieved using on-line capillary electrophoresis for the analysis of microdialysis samples due to the low sample volume requirement and high separation speed. The on-line coupling of C E to microdialysis sampling offers other advantages as well. The high ionic strength of the dialysate, necessary for biocompatibility, is well tolerated by the CE system in contrast to alternative on-line techniques such as mass spectrometry. CE also offers an alternative separation mode to LC. Many biologically important compounds are charged and therefore ideally suited to an electrophoretic separation. This system provides the capability of near real-time determination of multiple species in a conscious, freely moving animal. In essence, the on-line microdialysis/CE system represents a separation-based biosensor with multiple analyte capabilities. The delay between the event occurring in the
602
Analytical Chemistty, Vol. 66, No. 5, March 1, 1994
animal and the output of the analytical data is on the order of 5 min. Indeed most of this delay is the time needed to move the sample from the animal to the microinjection valve. This delay can be greatly reduced using anesthetized or restrained animals by positioning the microinjection valve closer to the implanted microdialysis probe. It should be noted that this delay does not represent a “time constant” for the system but a true delay as proper construction of the flow system results in minimal band broadening. The speed of separation used for this work does not approach the limit that can be achieved by CE. It was chosen primarily to assess theoperation of the interface at a pharmacokinetically useful sampling rate that could be readily achieved. As the separation speed improves to less than 30 s, the fundamental limitation of the microdialysis membrane response time and therefore the lower limit of temporal resolution for microdialysis sampling may be approached. The near real-time monitoring abilities of this system may prove useful in fundamental experiments by allowing feedback of information for experimental manipulation. An even more powerful use of such a system could lie in critical care monitoring of patients where rapid turnaround of analysis results is essential.
ACKNOWLEDGMENT This work was funded by an NCI training Grant, N I H grant R O 1GM44990, and the Kansas Technology Enterprise Corp. through the Center for Bioanalytical Research. Received for review June 7, 1993. 1993.”
Accepted December 3,
Abstract published in Aduance A C S Abstracts, February I , 1994.
