Application of Solid-Phase Microextraction in the Determination of

Xu Zhang , Ken D. Oakes , Md Ehsanul Hoque , Di Luong , Chris D. Metcalfe , Janusz ... John Z. Wen , Chris D. Metcalfe , Janusz Pawliszyn , and Mark R...
0 downloads 0 Views 86KB Size
Anal. Chem. 2001, 73, 4410-4416

Application of Solid-Phase Microextraction in the Determination of Diazepam Binding to Human Serum Albumin Haodan Yuan and Janusz Pawliszyn*

Department of Chemistry, University of Waterloo, Waterloo, Ontario, N2L 3G1, Canada

In this paper, protein-drug interactions were studied by solid-phase microextraction (SPME) using diazepam binding to human serum albumin as a model system. Since drug compounds are normally polar and nonvolatile by nature, direct SPME is used in this work. The SPME extraction is an equilibrium process among the concentrations of the analyte partitioned onto the SPME fiber, free and bound drug in the solution. A calibration curve was first constructed by employing the amount of the analytes partitioned on the fiber versus the free analyte concentration in the solution in the absence of protein. In method I, the extraction was performed in the protein solution with known diazepam concentration. In method II, diazepam was first loaded onto the fiber by extracting in solution with known diazepam concentration. This fiber was subsequently transferred into the protein solution for desorption. The amount of the analyte left on the fiber was analyzed after the system reached equilibrium. The free drug concentration was then obtained from the calibration curve for both methods. The Scatchard plot was finally employed to obtain the number of binding sites and the equilibrium binding constants. Since only a very small amount of the protein solution is required (150 µL for each extraction), method II is very useful for circumstances where the protein amount is very limited. The direct measurement method proposed in this paper does not need a GC response factor, which significantly decreases the experimental error. The only measurement needed is the area count change (ratio) of the fiber injections before and after the protein was introduced into the solution. The difference between the direct measurement method for method I and method II is discussed. The result illustrated that the SPME direct measurement method provided both theoretical accuracy and simplicity in such applications. The process of drug transformation and drug storage always involves the binding of a drug to the relative protein. It can be stated that all the important stages of the fate of a drug in the body are ruled by its protein interaction. The investigation of possible multiple binding interactions of a drug in the blood is important to understand the distribution and biotransformation * To whom all correspondence should be addressed. Tel: (519) 888-4641. Fax: (519) 746-0435. E-mail: [email protected].

4410 Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

pathways of a drug in the human body. It also has a pathological meaning since binding may be modified in disease states, due to changes in the nature or amount of protein, dehydration, or alteration of pH. The bound drug complex is of high molecular weight and unavailable for tissue membrane transport. Only the unbound drug is capable of diffusing across tissue membranes to reach the target site. Therefore, only the free drug is considered to be the pharmacologically active.1 Within the plasma proteins, serum albumin is undoubtedly the most important carrier for drugs and other small molecules. The attention of many scientists has been drawn to the phenomenon of the interactions between drug and the serum albumin. Because of the relative ease of the experiments, most standard binding assays use serum albumin. Many techniques have been employed for the drug-protein binding study. The more conventional techniques include equilibrium dialysis, ultrafiltration, and gel filtration. Although the equilibrium dialysis technique is very simple and convenient for multiple-samples analysis, a period of 12 h or more is required to attain the equilibrium, which may allow time for possible decomposition of unstable compounds or for the growth of bacteria. While these problems can be minimized by carrying out the experiments at lower temperatures, binding will thereby be altered. Normally the drug compounds will bind not only to the glass containers but also, to a greater degree, to the dialysis bag. Significant overestimation of the free fraction can result from even a slight leakage of protein into the dialysate. Ultrafiltration has been introduced widely for routine free drug monitoring in clinical laboratories.2 In comparison with equilibrium dialysis, this method offers significant advantages as represented by its short analysis time, simplicity, commercially available kits, and lack of dilution effects. However, it is perhaps not so readily adapted for large numbers of samples, and it uses a somewhat more expensive and complicated apparatus. Another important reservation is that since a portion of the aqueous phase is forced away from the protein, the latter solution becomes more concentrated, thereby tending to increase binding. Gel filtration is a very popular technique for measuring protein binding.2 In gel filtration, a solution of drug and protein passes through a column containing a dextran molecular exclusion gel. (1) Seydel, J. K.; Schaper, K.-J. Pharmacol. Ther. 1982, 15, 131. (2) La Du, B. N., Mandel, H. G., Way, E. L., Eds. Fundamentals of Drug Metabolism and Drug Disposition; The William & Wilkins Co.: Baltimore, MD, 1971; p 67. 10.1021/ac010227s CCC: $20.00

© 2001 American Chemical Society Published on Web 08/16/2001

The protein-bound drug is separated from the free drug by emerging before the free drug peak. However, if the equilibrium between bound and free drug is rapidly reversible, this approach would not be appropriate. One would expect that the molecules of bound drug should dissociate as the protein-drug complex begins to separate from the free drug and that the drug should emerge from the column as a smear rather than discrete peak. Despite the fact that chromatographic methods have been long used for the determination of drug-protein binding parameters, they have earned only limited attention. In recent years, the progress in chromatographic technology has led to the development of highly automated systems yielding high resolution on small columns, allowing shorter analytical times, consuming less chemicals, and avoiding the use of radiolabeled ligands. The most widely used chromatographic method is affinity chromatography, which provides the possibility of detecting very small differences in the binding affinity of ligands. However, such methods suffer from short column lifetime, experimental inconvenience, and timeconsuming column preparation. In this work, solid-phase microextraction (SPME) is used to determine the binding parameters of a drug binding to protein by employing diazepam binding to human serum albumin (HSA) as a model system. SPME is a relatively novel but well-known sampling and sample preparation technique. It was favored by its high extraction efficiency in volatile compound analysis, simple operation, and inexpensive apparatus.3 Recently, SPME has been applied to protein binding analysis. It has been successfully applied in the determination of the equilibrium constant of alkylbenzenes binding to bovine serum albumin (BSA) by headspace extraction.4 SPME yielded accurate results in those applications, providing that the data were properly interpreted. Diazepam (7-chloro-1,3-dihydro-1-methyl-5-phenyl-3H-1,4-benzodiazepin-2-one) is a member of the benzodiazepine drug family, which are widely used as tranquilizers, hypnotics, muscle relaxants, and anticonvulsants.5,6 It is well known that diazepam binds to HSA. Various authors7 proved the presence of one specific binding site. The binding constants, the thermodynamic parameters, and their variation with pH have also been determined.8,9 It is believed that diazepam binding to HSA occurs in binding site II, which is also called the indole and benzodiazepine binding site.10-12 This site binds several indole derivatives and benzodiazepines with a high degree of structural specificity.7,13-15 In fact, diazepam is one of the specific marker ligands for this site.12 (3) Pawliszyn, J. Solid-Phase Microextraction: Theory and Practice; Wiley-VCH: New York, 1997. (4) Yuan, H.; Ranatunga, R.; Carr, P.; Pawliszyn, J. Analyst 1999, 124, 1443. (5) Mule´, S. J.; Casella, G. A. J Anal. Toxicol. 1989, 13, 179. (6) Drouet-Coassolo, C.; Aubert, C.; Coassolo, P.; Cana, J. J. Chromatogr. 1989, 487, 295. (7) Mu ¨ ller, W. E.; Wollert, U. Naunyn-Schmiedeberg’s Arch. Pharmacol. 1973, 280, 229. (8) Mu ¨ ller, W. E.; Wollert, U. Naunyn-Schmiedeberg’s Arch. Pharmacol. 1974, 283, 67-82. (9) Coassolo, P.; Sarrazin, M.; Sari, J. C.; Briand, C. Biochem. Pharmacol. 1978, 27, 2787. (10) Mu ¨ ller, W. E.; Wollert, U. Pharmacology 1979, 19, 59. (11) Sudlow, G.; Birkett, D. J.; Wade, D. N. Mol. Pharmacol. 1976, 12, 1052. (12) Sjo¨holm, I.; Ekman, B.; Kober, A.; Ljungstedt-Pahlman, I.; Seiving, B.; Sjo¨din, T. Mol. Pharmacol. 1979, 16, 767. (13) Mu ¨ ller, W. E.; Wollert, U. Naunyn-Schmiedeberg’s Arch. Pharmacol. 1975, 288, 17.

Figure 1. Schematic of equilibrium in three-phase system (sample solution-dissolved protein-fiber coating).

THEORY Drug-protein binding is a reversible reaction between a drug molecule and a protein. K

[P] + [D] {\} [PD]

(1)

where [P] is the free protein concentration, [D] is the free drug concentration, and [PD] is drug-protein complex concentration. At equilibrium, it will have the equilibrium constant K:

K)

Cb [PD] ) [P][D] CpCs

(2)

where Cb, Cp, and Cs are the molar concentrations of bound drug (protein), free protein, and free drug, respectively. One of the predominant values measured in the protein binding study is the free drug concentration. Knowing the total concentration of the drug and protein, the percentage of the drug bound can be calculated and the K value can be determined. Since diazepam is a nonvolatile, relatively polar organic compound, direct SPME, where the extraction is performed directly in the sample solution instead of headspace, is used. The experiment configuration, where the headspace is totally eliminated, is illustrated in Figure 1. In the calibration system, where there is no protein in the sample solution, the system is a twophase system (buffer solution-fiber coating). In the measurement system, where the protein is present, it is a three-phase system (sample solution-dissolved protein-fiber coating). In the calibration system, the mass balance is

ntotal ) nf + ns

(3)

where ntotal, nf, and ns are the total amount of the analyte in the system, the amount partitioned on the fiber coating, and the amount freely dissolved in the buffer solution, respectively. Since (14) Mu ¨ ller, W. E.; Wollert, U. Mol. Pharmacol. 1975, 11, 52. (15) Sjo ¨din, T.; Roosdorp, N.; Sjo¨holm, I. Biochem. Pharmacol. 1978, 25, 2131.

Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

4411

ns ) CsVs, a few steps of simple derivations yield

Cs ) (ntotal - nf)/Vs

(4)

where Vs is the volume of the sample solution. If nf is negligible, Equation 4 can be simplified to

Cs ) C0 ) ntotal/Vs

(5)

This means that if the amount of the analyte on the fiber can be neglected, the calibration curve of initial concentration (C0), instead of free concentration (Cs), versus the amount of the analyte on the fiber can be employed to calculate the free analyte concentration in protein binding study. However, if the compound has a large partition coefficient toward the fiber, the amount of the analyte partitioned onto the fiber has to be taken into consideration. It can be measured from the response factor of diazepam on GC, which was determined from the syringe injection of a certain amount of the analyte into the GC injector. Once the free concentration Cs was known (from the calibration curve corresponding to the fiber injection amount nf), the Scatchard plot was then used to calculate the equilibrium constant and the number of binding sites. METHOD DESCRIPTIONS Method I. A 2-mL vial was used in this experiment. A stir bar was first put into the vial, and then 1.9 mL of buffer solution (in the calibration system) or protein buffer solution (in the measurement system) was added to completely fill the vial. This process leaves no headspace. The vial was mounted on the stirrer to start the agitation. The agitation speed was kept at 800 rpm with the vial capped. After the agitation was stable, a poly(dimethylsiloxane) (PDMS) fiber was inserted into the vial. The extraction time was optimized at 45 min. After the extraction, the fiber was transferred to the GC injector for analysis with a desorption time of 3 min. No carryover of analytes was observed. The extraction profile and calibration curve of diazepam were investigated in 1/15 M phosphate buffer solution (pH 7.4). For the protein binding study, 1 mg/mL HSA solution was prepared in the same buffer solution and used for the extraction. The free drug concentration was calculated from the GC response factor and the calibration curve. The Scatchard method was then applied to estimate the equilibrium constant. Method II. This method is designed for the situation where only a small volume of protein solution is available for analysis. The experimental setup is illustrated in Figure 2. A polyethylene insert (150 µL) positioned in a 2-mL vial was used in the experiment. The solution for analysis was put in this insert. The analyte was first loaded onto the fiber from a buffer solution with known analyte concentration. The exact amount of the analyte loaded on the fiber can be obtained from the GC response factor. This initial loading of the analyte was performed by the same procedure used in the calibration step in method I, so that the same calibration curves can be employed as for the free analyte concentration determination in the measurement system with protein present. Note that in this small-volume analysis, the amount of analyte partitioned on the fiber should not be neglected. 4412

Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

Figure 2. Experimental configurations for diazepam binding to HSA in small-volume analysis.

The loaded fiber was then inserted into the solution (with and without protein present) for equilibration. After the equilibrium had been reached, the fiber was withdrawn from the solution and analyzed by GC with fiber injection. The amount of the analyte left on the fiber can be determined by the GC response factor. The amount of the analyte initially loaded onto the fiber is the total amount of the analyte in this protein binding process. So it is termed as “n′total”, which satisfies

n′total ) n′b + n′s + n′f ) C′bVs + C′sVs + n′f

(6)

The symbol “′” stands for the measurement system with protein present. The amount of the analyte left on the fiber (n′f) can be determined from the GC response factor. The free dissolved drug concentration can be determined from the calibration curve of the loading solution. Since the volume of the protein (Vs) and the amount of the analyte loaded onto the fiber (n′total) are known parameters, the bound concentration (C′b) can be easily obtained. The equilibrium constant was finally obtained using the same Scatchard method described in method I. DIRECT MEASUREMENT OF THE EQUILIBRIUM CONSTANT This method was developed to calculate the equilibrium constant based on the change of the amount (GC area count ratio) of the analyte extracted by the fiber in the absence and presence of the protein. The only assumption was that both the GC and SPME are linear systems within the experiment range, which holds true in this application. Since this method avoids the determination of the GC response factor, the result tends to be simpler and more accurate. Direct Measurement for Method I. Since there is no headspace present, if the effect of fiber extraction amount is negligible, which is the case in method I, the mass balance for the system with protein present can be written as

n′total ) n′f + n′s + n′b ) C′s(Vf/Ksf + Vs + KCpVs) ≈ C′sVs(1 + KCp)

(7)

where Ksf is the partition coefficient of the analyte between the sample solution and fiber coating.

In the calibration system with the same total analyte concentration, the mass balance can be expressed as

0 and Combining eqs 11 and 7a, using the fact that ntotal ) ntotal 0 0 Cs/Cs ) A/A , we have

ntotal ) nf + ns

A0 - A 1 ) 1 + Vf/KsfVs A0

≈ CsVs

(8)

Since Cs/C′s ) A/A′, where A and A′ are GC area counts with protein absence and presence, and ntotal ) n′total, combining eqs 7 and 8 yields

A/A′ ) 1 + KCp

Substituting eq 12 into 9a, we get

A A0 - A KCp )1+ A′ A0

Cp ) Cp,total - Cb ) Cp,total - (C′total - C′s)

The only question left is how to calculate the free protein concentration Cp without using the GC response factor. The bound analyte concentration Cb can be determined by

C′b )

ntotal - n′f - C′s Vs

)

ntotal - n′f n′f Cs - Cs ntotal - nf nf

(10)

)

A0 - A′ A′ C s - Cs 0 A A -A

where C′total is the total diazepam concentration in the protein binding system. Therefore, Cp can be also obtained from the area count ratio of the fiber extraction before and after the protein was added. Direct Measurement for Method II. For the system described in method II, the amount of analyte partitioned in the fiber cannot be ignored. There are three steps in this method. The first is the analyte loading in buffer solution with analyte concentration C0s . The amount loaded on the fiber can be expressed as

)

A0(A - A′) Cs A(A0 - A)

)

A - A′ 0 Cs A0 - A

) Cp,total - (C′total - C′total(A′/A))

n0total ) (C0s /Ksf)Vf

ntotal ) nf + ns (7a)

In the last step, the loaded fiber was introduced into solution with protein present. The mass balance for this system is

n′total ) n′f + n′s + n′b ) C′s(Vf/Ksf + Vs + KCpVs) ) C′sVs(1 + Vf/KsfVs + KCp)

(8a)

Using the same procedure as in method I, from eqs 7a and 8a, the following formula can be derived:

A 1 KC )1+ A′ 1 + Vf/KsfVs p

(13)

Therefore, the free protein concentration is

Cp ) Cp,total -

A - A′ 0 Cs A0 - A

(10a)

(11)

In the second step, the loaded fiber was transferred into solution without protein being present. When equilibrium was reached, the mass balance for the system could be written as

) CsVs(1 + Vf/KsfVs)

(9b)

(9)

From eq 9, a straight line can be obtained when A/A′ is plotted against Cp. The value of the equilibrium constant K can be obtained from slope of this curve. Free protein concentration Cp can be determined as follows:

) Cp,total - C′total(1 - A′/A)

(12)

(9a)

EXPERIMENTAL SECTION Chemicals and Materials. Diazepam was purchased from Radian (Austin, TX) as a 1 mg/mL methanol solution. This solution was diluted with methanol into 0.1, 0.01, and 0.001 mg/ mL solutions for experimental convenience. All these stock solutions were stored at -10 °C. The human serum albumin (96% purity, no fatty acid) was purchased from Sigma (Mississauga, ON, Canada). SPME devices and fibers (100-µm PDMS) and all the vials used in the experiments were purchased from Supelco (Bellefonte, PA). For small-volume analysis (method II), a polyethylene insert, which has a volume of 150 µL, was positioned in a 2-mL vial for the analysis. The pH 7.4 buffer solution was prepared by combining 200 mM disodium hydrogen orthophosphate and 200 mM soldium dihydrogen orthophosphate solution at a certain ratio while monitoring with a pH meter. This buffer solution was diluted into 3 times to form 1/15 M pH 7.4 buffer solution. The calibration and protein binding measurements were performed in this buffer solution. Instrumentation and Analytical Conditions. All analyses were performed on a Varian (Sunnyvale, CA) GC 3500 gas chromatograph equipped with a 10 m × 0.25 mm i..d × 0.25 µm SPB-5 column (Supelco, Bellefonte, PA), a septum-equipped programmable injector (SPI) with SPME insert, and a FID. The Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

4413

Table 1. Summary of the Data for Diazepam Analysis in Buffer Solution (Method I)a Ctotal (ng/mL)

area count

ntotal (ng)

nf (ng)

nf/ntotal (%)

Cs (ng/mL)

100 200 300 400 500 600 700 800 900 1000

17 124 32 818 47 621 60 318 71 144 86 088 102 845 111 570 124 120 135 109

200 400 600 800 1000 1200 1400 1600 1800 2000

4.75 9.11 13.2 16.7 19.8 23.9 28.6 31.0 34.5 37.5

2.38 2.28 2.20 2.09 1.97 1.99 2.04 1.94 1.91 1.88

97.6 195 294 392 490 588 686 785 883 981

a

Figure 3. Extraction profile of diazepam by a PDMS 100-µm fiber in 0.067 M pH 7.4 phosphate buffer. Agitation speed was 800 rpm.

carrier gas was helium (25 psi head pressure). The column temperature program used for the fiber injection in the experiments was 120 °C, hold for 1 min, increase at 10 °C min-1 to 300 °C, and hold for 5 min. During the whole analysis, the injector and detector temperatures were 250 and 300 °C, respectively. The detector response factor was determined by a syringe injection of 1 mg/mL standard diazepam in methanol solution using the same column temperature program. The SPI injector was temperature programmed as follows: 50 °C hold for 0.5 min, increase at 250 °C/min to 250 °C, and hold for 22 min. Liquid CO2 was used to cool the injector before all injections. All the extractions were carried out at 23 °C. A 7 mm L × 2 mm D stir bar was used in each of the extraction vial (2 mL clear vial from Supelco). RESULTS AND DISCUSSION Decomposition of Diazepam. Like most of the benzodiazepines, diazepam is a thermolabile compound, which is likely to be decomposed in the GC system if the temperature is sufficiently high. It was found that diazepam was decomposed when a 30-m column was used, since the compound took more time to elute out. The problem was eliminated when a 10-m column was used. Extraction Profile. All extractions in this study were carried out at 23 °C. The extraction profile was first investigated to determine the equilibrium time. The equilibrium profile is presented in Figure 3. The change of the equilibrium time in the situation where protein was present was not observed. From the extraction profile, we can see that the equilibrium is reached after 35 min. Therefore, 45 min was used as the extraction for all experiments. GC Response Factor. The GC response factor was determined by syringe injection of 0.6 µL of a 1 mg/mL (total mass, 60 ng) diazepam methanol solution into the GC system. The “sandwich” method was used for the syringe injection. The response factor was determined as 3602 area counts/ng of diazepam. Calibration Curve. The calibration curve was performed in 1/ M phosphate buffer (pH 7.4) solution with diazepam concen15 trations varied from 0.25 to 10 µg/mL. A diazepam standard methanol solution was spiked into the buffer solution to obtain a sample solution of a given concentration. It was found that the trace amount of methanol could effect the precision of analysis by swelling the fiber coating. Therefore, during the calibration, 4414 Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

The response factor is 3062 area count units/ng of diazepam.

as well as during the protein binding analysis, methanol was added to the solution to keep the methanol concentration the same in each analytical vial for all the concentrations of diazepam. The amount of the analyte partitioned onto the fiber coating can be calculated from the response factor and the area count of the fiber injection. The results are shown in Table 1. Column 5 in Table 1 shows that the amount of the analyte extracted by the fiber is less than 3% of the total amount of the analyte in the solution. This amount is so small that it can be neglected in plotting the calibration curve. Therefore, the calibration curve of the total concentration Ctotal versus the area count can be used as the calibration curve to determine the free drug concentration in the protein binding study. The calibration curve is linear with regression equation of y ) 140.77x and a square regression coefficient (R2) of 0.9903. Drug Loading and Calibration in Method II. The same calibration curve was obtained as that obtained with method I. The amount of the drug loaded on the fiber is determined from this curve using the area count obtained by GC analysis and the GC response factor, which was determined from the previous analysis, 3062/ng. This calibration curve was also employed to determine the free analyte concentration. Determination of the Binding Parameters. Method I. The moles of drug bound per mole of protein (r), the molar free concentration ([D]), and the value of r/[D] have been calculated and are summarized in Table 2. A Scatchard plot is employed to calculate the equilibrium constant and the number of binding sites in this study. The concentration of HSA used in this study was 1 mg/mL, which is 1.45 × 10-5 M (molecular weight of HSA, 69 000). Since the analyte concentration loaded on SPME fiber only reached equilibrium with the free analyte concentration in the solution, the concentration obtained from the calibration curve is the free diazepam concentration in the solution. The amount of the analyted loaded on the SPME fiber can be calculated through the GC response factor. Therefore, the bound drug concentration in the solution could be easily obtained from eq 7. The Scatchard plot is presented in Figure 4. From the regression equation of the Scatchard plot, the slope was equal to 1.02 × 106, the y-intercept equal to 1.03 × 106, and the x-intercept equal to 1.0. Therefore, for the equilibrium constant K ) 1.02 × 106 L mol-1, the log K value was 6.01. The number of binding sites per protein molecule was 1.0.

Table 2. Summary of the Experimental Data of Diazepam Binding to HSA (Method I) Ctotal (ng/mL)

area count

Cs (ng/mL)

drug bound (%)

r

[D] (mol L-1)

r/[D] (L mol-1)

1000 2000 3000 4000 5000 6000

11 304 18 814 29 395 65 777 123 885 205805

80.3 134 209 467 880 1462

92.0 91.1 89.6 84.4 78.0 70.8

0.223 0.331 0.434 0.614 0.756 0.857

2.82 × 10-7 4.69 × 10-7 7.33 × 10-7 1.64 × 10-6 3.09 × 10-6 5.14 × 10-6

7.90 × 105 7.05 × 105 5.92 × 105 3.74 × 105 2.44 × 105 1.67 × 105

Table 3. Summary of the Experimental Data of Diazepam Binding to HSA in Small Volume Analysis (Method II)a

a

Cload (ng/mL)

ntotal (ng)

area count

nf (ng)

Cs (ng/mL)

nb (ng)

drug bound (%)

r

r/[D]b

500 1000 2000

19.8 37.5 68.4

8 970 18 566 37 042

4.2 7.9 14.4

62.3 128.9 257.1

6.28 10.27 15.45

40.2 34.7 15.5

0.203 0.332 0.499

926 773 733 083 552 710

The protein concentration was 0.05 mg/mL. b Molar concentration is used.

Figure 4. Scatchard plot for diazepam binding to HSA (method I).

The apparent equilibrium constant and the total binding constant were reported as 1.159 × 106 and 4.919 × 105 L mol-1.7,8 This is the only value that is published about diazepam binding to HSA. Compared with the result in this study, which is 1.02 × 106 L mol-1, the two results are very comparable with the relative difference of their log values less than 1%. Method II. During the protein binding analysis, three concentrations are investigated. They are loaded from 500, 1000, and 2000 ng/mL 2-mL buffer solutions, respectively. For each of the concentrations, three replicates are measured. The extraction time in the drug loading step and the desorption time in the protein solution are controlled at 45 min (Figure 3). The protein concentration in this study is 0.05 mg/mL, which is 7.25 × 10-7 M. Table 3 summarizes the experimental results. The Scatchard plot is presented in Figure 5. From the regression equation of the Scatchard plot, the slope equals 1.25 × 106, y-intercept equals 1.17 × 106, and x-intercept equals 0.93. Therefore, the equilibrium constant K is 1.25 × 106 M-1, and the log K value is 6.10. The number of binding sites per protein molecule is 0.93, which is close to 1. In this study, the protein concentration used was 7.25 × 10-7 M (50 ppm) instead of 1.45 × 10-5 M as in the last experiment. The total drug concentration introduced by the amount of the

Figure 5. Scatchard plot for diazepam-HSA binding analysis (method II).

analye loaded on the fiber is 4.53 × 10-7 M (128.9 ng/mL for the analye loaded from 1000 ng/mL buffer solution). Therefore, the molar concentrations of total drug and protein were in the same order of magnitude. This was an important consideration in the experimental design to minimize the error on both bound and free concentration calculations. Determination of the Equilibrium Constant without Calibration. Direct determination of the binding parameters from the ratio of the fiber extraction in the absence and presence of the protein was performed for both method I and method II analyses. Figure 6 shows the plot of A/A′ versus Cp obtained from method I. The equilibrium constant was obtained from the slope of the plot, which is 9.77 × 105 L mol-1. This value agreed with the value attained from previous Scatchard plot. Figure 7 shows the plot of A/A′ versus Cp obtained from method II. In this experiment, a protein concentration of 0.05 mg/ mL was used. From the plot, the slope was determined as 0.995 × 106 M-1. Noting that factor (A0 - A)/A0 ) 0.801, the diazepamto-HSA binding constant K was then calculated as 1.23 × 106 M-1, which agrees with the result from that of last section. Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

4415

Figure 6. Curves used for calculation of the equilibrium constant of diazepam binding to HSA without calibration (method I).

Figure 7. Curves used for calculation of the equilibrium constant of diazepam binding to HSA without calibration (method II).

CONCLUSION SPME is an equilibrium process between the amount of drug partitioned onto the SPME fiber, the concentration of free drug,

4416

Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

and the concentration of bound drug in the solution. The portion of the drug analyzed in a GC system through fiber injection represents the true amount of the drug that had reached equilibrium with the solution. Therefore, the equilibrium between the drugprotein binding process was not affected. In most cases, this value is so small that it could be neglected without influencing the equilibrium system. However, the verification should be always performed before this assumption is exercised. For large-volume analysis (method I), the analyte partitioned into the fiber is less than 3% of the free analyte amount in the solution. Therefore, it can be safely ignored. However, for small-volume analysis (method II), the analyte partitioned on the fiber is comparable to the free analyte in the solution (∼20%), and its contribution has to be considered. This work demonstrated that SPME is a valid method in the protein binding study by employing diazepam binding to HSA as a model system. The SPME method provides theoretical accuracy and simplicity in operation. By integrating sample extraction, loading, and transferring into a single step, this paper presents a novel method to use SPME to determine the drug-protein binding parameters for small-volume analysis (method II) by loading the drug compound first on the fiber and then desorption in the protein solution. This small-volume method is especially useful under the circumstance where only a small amount of protein is available. In this research, both the Scatchard plot and direct measurement were employed to determine diazepam-HSA binding parameters. In the direct measurement approach, no calibration and GC response factor were needed. The results were compared with the value from the literature. Unlike some conventional methods, SPME does not have any theoretical limitations in the determination of the free analyte concentration and proteinanalyte binding constant. It is very convenient for both data collecting and binding parameter calculation. Received for review February 26, 2001. Accepted June 29, 2001. AC010227S