First-Derivative Spectroscopic Determination of Binding Characteristics of Rifampicin to Human Albumin and Serum

First-Derivative Spectroscopic Determination of Binding Characteristics of Rifampicin to Human Albumin and Serum

First-Derivative Spectroscopic Determination of Binding Characteristics of Rifampicin to Human Albumin and Serum S.LACHAU', M. A. ROCHAS*,A. E. TUFENK...

356KB Sizes 0 Downloads 19 Views

First-Derivative Spectroscopic Determination of Binding Characteristics of Rifampicin to Human Albumin and Serum S.LACHAU', M. A. ROCHAS*,A. E. TUFENKJI*, N. MARTIN*, P.LEVILLAIN~,AND G. HOUIN*' Received March 5,1990, from the 'Labratoire de Pharmacocin6tique, facult6 des Sciences Pharmaceutiques, 35 Chemin des Maraichers, 31062~ToulouseCedex et Centre lnterdisciplinaire dEtudes Pharmacocin6tiques de Toulouse (CINET), and the Labratoire de Chimie Analytique, facult6 des Sciences Pharmaceutiques, Boulevard Tonne14 37000 Tours, France. Accepted for publication April 8,1991.

*

Abstract 0 A first-derivativespectroscopic method for the simultaneous determination of bound and unbound drug in human serum and serum albumin solution was developed. As an example, the binding characteristics of rifampicin were studied. In serum albumin solution, the rifampicin bound and unbound fractions were determined at 473.5and 475.8 nm, respectively. For human serum, the unbound fraction was determined at 479.4 nm. The results obtained by the first-derivative spectroscopic method were in agreement with those obtained by equilibrium dialysis. The proposed method is very simple and accurate when applied to measurements of drug:protein binding. Moreover, it allows a direct measurement of the bound and unbound forms without any physical separation.

It is generally accepted that only the unbound fraction of a drug can diffuse in the tissues and is responsible for the pharmacological activity of the drug. Moreover, the pharmacokinetic disposition of the drug may be modified by its binding to circulating proteins. The study of drug:protein binding is always based on a physical separation of the bound and unbound fractions. The difference in molecular weights between the two fractions is actually used. During this physical separation, errors may occur; these errors are the basis of different results obtained with the same drug when comparing the various available techniques.' Our aim was to develop a technique, avoiding this initial separation, that is based on the variations of a physical characteristic of the drug, using the electronic spectrum which is modified when binding occurs through ionic or covalence links. As the observed variations are usually small, the classical spectrophotometric methods are inadequate. However, this problem can be solved by using the zero-crossing method in derivative spectroscopy.2~3 This technique was previously used in many different fields (for an extended bibliography, see Morellid).In the present study, derivative spectrophotometry is used for the first time to characterize drug:protein binding. Rifampicin was chosen as a test molecule because its electronic spectrum is well differentiated from that of plasma proteins (see chemical structure). The binding of rifampicin in a human albumin solution and in human serum was studied, and the results are compared with those previously obtained using the standard technique of equilibrium dialysis.

Experimental Section Apparatus and Reagents-Spectrophotometer-Measurements were made using a Hewlett-Packard spectrophotometer (98155 A type) with a model 7470 A plotter. The cell paths used were in quartz, thermostated (37"C), and of 2-,5-,and 10-mm lengths. Calculation of derivates was computerized, with the possibility to change the integration step. Reagents-Rifampicin was supplied by Laboratoires Mere11 (Paris, France), and the purity was ~99.5%based on two different TLC systems. Purified fatty acid-free serum albumin was purchased from Behring Institute (Paris, France) and drug-free human serum was Oo22-3549/9~0300-O287$02.50/0 0 1992. American Pharmaceutical Association

CH3

CH3

RIFAMPICIN from the Centre de Transfusion (Toulouse, France). All other reagents were of analytical grade and were obtained from Prolabo (Paris, France). P r o c e d u r e d t a n d c r r d Solutions-Rifampicin was dissolved a t a concentration of 1.82 mh4 in a pH 7.4 phosphate buffer (0.067 M Na2HP04* 12 H,O; 0.067 M KH2P04). T w o solutions of human albumin in the same buffer were prepared at concentrations of 580 and 1160 pM. Calibration Curves for Bound and Unbound Concentrations-In order to determine the free and bound concentrations of rifampicin in the studied solutions, three calibration curves were prepared in different media between 12 and 72.9 pM (six concentrationsfor each). For unbound rifampicin, the experiments were conducted in the pH 7.4 phosphate buffer. For serum albumin-bound rifampicin, the zero-crossing wavelength was estimated with a n excess of protein (80 g/L, 1160 pM).The first-derivative absorption spectra were measured between 464 and 484 nm, with a window of 6 points. To estimate the serum-bound rifampicin concentrations, the calibration curve was prepared in serum overloaded with human albumin (1160pM). The first-derivative absorption spectra were measured between 450 and 500 nm, with a window of 9 points to minimize background noise. Serum Albumin Binding Stdy-The binding of rifampicin was measured in a 40-g/L (580pM) human albumin solution at concentrations between 24.3 and 486.6 pM (n = 111, in six separate experiments. Human Serum Binding S t u d p T h e binding of rifampicin was measured in pure serum with concentrations between 18.2 and 182.2 pM (n = lo), in six separate experiments. Calculations-The exact wavelengths of the zero c m i n g were graphically determined, as shown in Figure 1. From the derivative values obtained with the mixture of the two forms (Figure lB),the unbound and bound concentrations were estimated with the two calibration curves that were previously obtained. The binding parameters were calculated from the fitting of the experimental c w e a of the bound form as a function of the freeform, using a nonlinear least-squares regression analysis based on a Gauas-Newton algorithm.5 Journal of Pharmaceutical Sciences I 287 Vol. 81, No. 3, March 1992

Similarly, at 475.8 nm, the obtained value depended only on the free concentration. Using these experimental data, an excellent linearity was observed for the two calibration curves of free (eq 1)and bound (eq 2) rifampicin:

I

B

/

I

R

v

h '0

.'02

i

t

I

C = 540230 - 0.027

r = 0.999

(1)

C = 584460 - 1.8

r = 0.999

(2)

where C is the free or bound rifampicin concentration and D is the derivative value of absorbance (Abs/nm). Under our experimental conditions, rifampicin binding to human albumin (580 pM) appeared to follow a saturable process that was associated with a nondurable one, as shown in Figure 2. The corresponding number of sites and affinities is given in Table I. Human Serum Binding-Using the same experimental conditions, we studied the binding of rifampicin to human serum. The zero-crossingwavelength of the bound form (479.4nm)was determined with serum overloaded with human albumin (80 The choice of albumin was based on previous results showing that rifampicin is essentially bound to this protein in human serum.' Moreover, our studies did not show any displacement of the maximum wavelength of rifampicin when the serum was supplemented with a-1-glycoprotein. An excellent linearity was observed for the two calibration curves of free (eq 3) and bound (eq 4)rifampicin:

a).

-m02.

I

i

I

\

C = 143270 - 0.513

r

C = 190500 - 0.649

r = 0.999

=

0.999

(3) (4)

In this study, we were unable to see the saturable phenomenon previously observed with serum albumin as shown in Figure 2. Therefore, only the total binding capacities were estimated, and they are reported in Table I. Rifampicin Recovery-In each experiment, irrespective of the concentration of rifampicin and the nature of the solution, we compared the recovery of rifampicin as estimated from the ratio of the sum of bound and free concentrations to the initial concentration. No statistically significant differences were observed, the slope of the corresponding regression line being 1.007 2 0.004(r = 0.987,p < 0.001).

-"04k ' ' uo 472

.

416 .

Wavelen,gth, nm Flgure 1-First-derivative spectra used to obtain regression curves: (A) free rifampicin at 475.8 nm; (B) mixture of the two forms in human albumin solution (580 f l ;rifampicin concentrations of 24.3 to 486.6 f l ) (C) ; bound rifampicin at 473.5 nm (rifampicin concentrationsof 12 to 72.9 pM).

Results Human Albumin Binding-The first-derivative curves for two series of rifampicin solutions at increasing concentrations are shown in Figures 1A and 1C. The first curve (1A) was determined in phosphate buffer containing only free rifampicin. The other curve (1C) used a high concentration of human albumin (80g/L)and corresponded to the bound form. Higher albumin concentrations did not change this wavelength, which was therefore considered as representative of the bound fraction. In each case, the curves crossed a common wavelength value (zero crossing) which corresponded to 473.5nm for free rifampicin and 475.8 nm for bound rifampicin. In a mixture containing increasing amounts of rifampicin and a constant concentration of human albumin (40 g/L; Figure lB), a series of curves was obtained which never crossed at a common wavelength value, as observed on the original expanded graphs. As previously demonstrated,3.6 a t 473.5 nm, the derivative value depended only on the concentration of the bound form. 288 I Journal of Pharmaceutical Sciences Vol. 81, No. 3, March 1992

m

ae

40

0

100

zoo

300

400

SO0

Protein Conc: pMlL

Flgure 2-Percentage of rifampicin binding to either (V) human serum (HS) or (0)human serum albumin (HSA) solutions (145 and 580 pM, respectively;mean ? SD, n = 6 for each dose)as a function of rifampicin concentrations (fl/L).

Table Cinteractlon of Rlfamplcln with Human Serum Albumin (HSA) and Human Serum (HS) as Measured by First-Derivative and Equillbrlum Dialysis Methods’

Solution HSA HS

Binding, M-’

Concentration, ClM 580 493

First-Derivative Method

Equilibrium

4587.8 2 368.1

4035 2 24 4.6 f 0.3

6.342 1.03

* Results are expressed as mean

2

Dialysis

SD (n = 5).

Discussion Examination of the derivative profiles shows that the two wavelength maxima are different by 2.3 nm. In a classical spectrum, it would be impossible to distinguish such a small difference. By using the first derivative, our study showed a very good differentiation, thereby allowing the simultaneous estimation of the concentration of the two forms (Figure 1).As the zero-crossingwavelengths were different in serum and in pure serum albumin, other serum proteins must bind rifampicin. Lipoproteins and a-1-glycoprotein binding could be involved, as previously demonstrated by the equilibrium dialysis method.8 We examined different pH 7.4 buffers to estimate the free zero-crossing point (Ciba-Geigy table). With the exception of one buffer, borax:H,PO, from Kolmoff, all buffers had the same zero-crossing wavelength for free rifampicin. The phosphate buffer was used as it corresponds to the reference buffer used in equilibrium dialysis. Bolt and Remmerg recommended the use of a reducing agent, ascorbic acid, to prevent the oxidation of rifampicin to its quinone derivative. However, in our study, the short time (<0.25 h) necessary for the experimentation makes its use unnecessary. This would not be the case with equilibrium dialysis. We have chosen first-derivative spectroscopy because it provides the highest sensitivity with satisfactory selectivity. Higher order derivatives would have increased the selectivity, but the sensitivity would have decreased too much since the ratio of the calculated parameter to the background noise decreases rapidly. To lessen this phenomenon, we calculated a mean derivative value for several experimental points around the wavelength of interest. The specificityof the curve increased with the number of points taken in account. To maintain the maximum resolution of the curve, six to nine points were found to be optimum, with a 2-nm step.3f3 Equilibrium dialysis, ultrafiltration, and ultracentrifugation remain the most commonly used methods. Generally, equilibrium dialysis is used as the research reference method. For clinical purposes, however, ultrafiltration is often preferred since it is rapid and simple. Ultracentrifugation is also used when the two other methods fail, as recently shown for determining unbound cyclosporin in plasma.10 However, compared with derivative spectroscopy,all three of these methods have certain important disadvantages. Despite the use of membranes exhibiting a cut-off at 10 000 Da, dialysis takes between 2 and 4 h to reach equilibrium. In derivative spectroscopic method, equilibrium is almost instantaneous and simulates in vivo conditions. With substances exhibiting ex vivo metabolism, such as bilirubin or esters, the spectroscopic method has a definite advantage. The adsorption of the substance being studied to dialysis or ultrafiltration membranes and/or to the walls of the systems may be of critical importance.11 For equilibrium dialysis, this

-

interfering binding can be estimated and taken in account, if
Conclusions The proposed method enables the rapid and accurate determination of the binding of drugs in either pure protein solutions or in serum and is an alternative method to established procedures. The preliminary results obtained are in good accordance with those using equilibrium dialysis as the reference method. However, the study of the binding of other drugs, under different conditions (drug and protein concentrations), is needed to further document the validity of this method.

References and Notes 1. Tillement, J. P.; Houin, G.; Zini, R.; Urien, S.; Albengres, E.; Barre, J.; Sebille, B. In Pharrnacologie Clinique, Bases de la Thkra uti ue, 2nd ed.; Giroux, J. P.; Mathe, G.; Meyniel, G., Eds.; g p . Zci. Fr.: Paris, 1988; pp 15-24. 2. O’Haver, T. C. Clin. Chem. 1979,25, 1546-1553. 3. Levillain, P.; Fompeydie D. Analysis 1986, 14, 1-20. 4. Morelli, B.J . Pharrn. Sci. 1988, 12, 1042. 5. Hamberger, C.; Barre, J.; Zini, R.; Taiclet, A.; Houin, G.; Tillement, J. P. Znt. Clin. Pharm. Res. 1986, 6, 441-449. 6. Levillain, P.; Fompeydie, D. Anal. Chem. 1985,57, 2561-2563. 7. Boman, G.;Ringerger, V. A. Eur. J . Clin. Phnrrnacol. 1974, 7, 369-373. 8. Chauvelot, L. Ph.D. Thesis; Paris University, Paris, France, 1978. 9. Bolt, H. M.; Remmer, H. Xenobiotica 1976, 6, 21-32. 10. Legg, B.; Rowland, M. J . Phnrm. Pharmucol. 1987,39,599-603. 11. Tillement, J. P. In Advances in Drug Research, Testa, B., Ed.; Academic Press: London, 1984; Vol. 13, pp 59-94. Wilensky, A. J.; Friel, P. N. In Applied Pharrnaco12. Levy, R. H.; kinetics, Princi les of Therapeutic Drug Monitoring, 2nd ed.; Evans, W. E.; &hentag, J. J.; Jusko, W. J., Eds.; Applied Therapeutics: Spokane, WA, 1986; pp 540-569. 13. Winter, M. E.; Tozer, T. N. In Applied Pharmucokinetics, Principles of Therapeutic Drug Monitoring, 2nd ed.; Evans, W. E.; Schenta , J. J.; Jusko, W. J., Eds.; Applied Therapeutics: Spokane, 1986; pp 493-539.

d4,

Journal of Pharmaceutical Sciences I 289 Vol. 81, No. 3, March 1992