Structural specificity requirements in the binding of beta lactam antibiotics to human serum albumin

Chemico-Biological Interactions 104 (1997) 179 – 202

Structural specificity requirements in the binding of beta lactam antibiotics to human serum albumin Bibiana Nerli, Diana Romanini, Guillermo Pico´ * Chemical-Physics Department, Faculty of Biochemical and Pharmaceutical Sciences, CIUNR and CONICET, National Uni6ersity of Rosario, Suipacha 531, (2000) Rosario, Argentina Received 10 January 1997; received in revised form 19 March 1997; accepted 20 March 1997

Abstract The binding of some cephalosporins of pharmacological interest, to human serum albumin was studied using ultrafiltration method. The identification of the binding sites in albumin was also performed using probes for the so-called sites I, II, bilirubin and fatty acids binding sites. Cephalosporins were classified into three groups according to their affinity for albumin: low affinity (K=10–102 M − 1), medium affinity (K =103 M − 1) and high affinity (K =104 M − 1). Cephalosporin binding to albumin produced a perturbation of several basic amino acids of the protein such as histidine and lysine. It was found that only cefuroxime, ceftazidime and cefoperazone interact slightly with site I on serum albumin, while site II possesses capacity to bind: cephradine, cephalexin, ceftazidime, ceftriaxone, cefoperazone, cefaclor and cefsulodin. The bilirubin binding site showed capacity to interact with a great number of cephalosporins: ceftriaxone, cefazolin, cephaloglycin, cefamandole, cefotaxime, cefoxitin, cefuroxime, cefoperazone and cefadroxil. Ceftriaxone showed capacity to bind to

Abbre6iations: HSA, human serum albumin; LYS, lysyl; TYR, tyrosyl, TRP, tryptophanyl; HIS, histidyl; BL, bilirubin; RA, retinoic acid; DSS, dansylsarcosine; DSNA, dansylamide; CFC, cephalosporin C; CFLE, cephalexin; CFCl, cefaclor; CFDO, cefsulodin; CFRI, cephaloridine; CFGL, cephaloglycin; CFLO, cephalothin; CFOL, cefamandole; CFTA, cefotaxime; CFDR, cefadroxil; CFPI, cephapirin; CFU, cefuroxime; CFOP, cefoperazone; CFNE, ceftriaxone; CFZO, cefazolin; CFXI, cefoxitin; CFRA, cephradine; CFT, ceftazidime. * Corresponding author. Current address: Facultad de Ciencias Bioquı´mivas y Farmace´uticas, Suipacha 570, (2000) Rosario, Argentina. Fax: + 54 (041) 240010. 0009-2797/97/$17.00 © 1997 Elsevier Science Ireland Ltd. All rights reserved. PII S 0 0 0 9 - 2 7 9 7 ( 9 7 ) 0 0 0 2 4 - 0

180

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

the fatty acid binding site on HSA. No relation was found between the displacement of the marker and the chemical nature of the substituents at R1 and R2. Cephalosporins interact with HSA at the binding region that involves: tyrosyl 411, histidyl 146 and lysyls 195, 199, 225, 240 and 525 residues. The chemical modification of specific amino acids showed that the interaction of these amino acids with beta lactam antibiotics is not carried out to the same extent for all the cephalosporins tested. The results obtained revealed that the binding sites for cephalosporins on albumin are structurally heterogeneous, having different amino acids in the vicinity of the ligand molecule. © 1997 Elsevier Science Ireland Ltd. Keywords: Albumin; Cephalosporins; Binding; Amino acids; Beta lactam; Antibiotics

1. Introduction The binding of drugs to serum albumin has important pharmacokinetic consequences because it influences their distribution, excretion and pharmacological effect in the body. Albumin binding is of central pharmacological interest, because it has been demonstrated that the therapeutic effect is related in a direct manner with the free concentration of drugs in plasma [1]. The family of the beta lactam antibiotics has a great number of derivatives which are widely used in clinical therapy. Previous reports [2,3] found that these antibiotics were bound to serum albumin with diverse affinity and their therapeutic efficacy was therefore influenced by their protein binding. Several reports on the cephalosporins binding to serum albumin were performed by dialysis equilibrium [2,3], however, published results are not consistent. Furthermore, the identity of the cephalosporin binding sites in albumin is only partially known. Besides, there is very little information available about the molecular mechanism by which the interaction albumin-cephalosporin takes place, in contrast to other drugs such as benzodiazepine and nonsteroidal anti-inflammatory ones, whose binding sites in albumin are known. The identity of the amino acids which participate in the cephalosporin-albumin binding in the albumin molecule also remains unknown. As a continuation of previous studies [4,5] about the binding of beta lactam antibiotics to human serum albumin, the purpose of the present study was to elucidate the molecular mechanism of the cephalosporin-albumin complex formation and to identify the binding sites in albumin for these ligands. A comparative study of the binding of some cephalosporins with different substituents at positions R1 and R2 in their molecule was also carried out.

2. Materials and methods Chemical: Crystallized human serum albumin (HSA) essentially fatty acids free ( B 0.005%), fluorescein isothiocyanate, p-nitrophenyl acetate, dansyl chloride, di-

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

181

ethyl pyrocarbonate, 4-nitrophenylanthranilate, sulfamerazine, ibuprofen, 7amino-4-methyl-coumarin, bilirubin (BL), retinoic acid (RA), dansylsarcosine (DSS), dansylamide (DNSA) and the following cephalosporins: cephalosporin C (CFC); cephalexin (CFLE); cefaclor (CFCl); cefsulodin (CFDO); cephaloridine (CFRI); cefotaxime (CFTA); cephaloglycin (CFGL); cephalothin (CFLO); cefamandole (CFOL); ceftazidime (CFT); cefadroxil (CFDR); cephapirin (CFPI); cefuroxime (CFU); cefoperazone (CFOP); ceftriaxone (CFNE); cefazolin (CFZO); cefoxitin (CFXI) and cephradine (CFRA) were purchased from Sigma and used without further purification. Ceftazidime (CFT) was donated by Glaxo (UK). Cephalosporin solutions (CF) were prepared immediately before use. The solutions of BL and RA were prepared according to [14,15] and their concentrations were determined by absorbance measurements. All the other reagents were of analytical quality. The fluorescence measurements were made using a Jasco FP 770 spectrofluorometer, the absorbance measurements, using a Shimadzu 240-02 double beam spectrophotometer and the pH measurements were performed on an Orion SA 720 pH meter.

2.1. Measurement of the HSA-CF binding It was performed by ultrafiltration using a MICROSEP from Filtron Technology, with a molecular weight cut off of 10 000 Da. Concentration of free cephalosporin (c) in the ultra filtrate was determined by UV spectrophotometry. By comparison with experiments carried out either with protein or with CF alone, it was verified that neither CF or HSA was adsorbed on the membrane.

2.2. Chemical modification of the amino acids in HSA Histidyl residues were modified by reaction with diethyl pyrocarbonate [6], the number of residues modified was determined measuring the change in absorbance of the protein at 242 nm. TYR 411 was labeled by reaction with p-nitrophenylanthranilate [7]. Lysyl residues were labeled with the following reagents: dansyl chloride (LYS 195) [8], 2,4-dinitrofluorobenzene (LYS 240) [9], acetylsalicylic acid (LYS 199) [10], LYS 525 by glycosylation with glucose [11] and fluorescein isothiocianate (LYS 225) [12]. The number of molecules of covalent marker bound to the protein per HSA molecule was determined as published in the references below.

2.3. Identification of the CF binding sites on HSA To identify the CF binding sites on HSA the following specific markers of albumin binding sites were used: DNSA and DSS as markers of sites I and II

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

182

respectively [13]; BL and RA, to examine the effect of CF on the bilirubin and fatty acids binding sites respectively [14,15]. The effect of CF on the specific markers binding sites bound to albumin was assayed measuring the fluorescence change of bound marker: DF = 100 (Fi -F°)/F°, where F° and Fi are the fluorescence of the marker bound to albumin in absence and presence of i concentration of cephalosporin.

2.4. Determination of the pK of the CF The direct titration of the CF dissolved in water, was performed by addition of chloride acid until the pH medium was about the unity. Then the medium was titrated with sodium hydroxide solutions of known concentration and the pH of the medium was measured. Non-lineal curve fitting of the data was used to calculate the pK values.

2.5. Acid-base titration of the HSA An approximately 6% solution of HSA was deionized by passing the solution repeatedly over a mixed bed ion-exchange column (Amberlite IRA 400 and IRA 120) until the solution reached a constant conductance. The acid base titration of HSA was carried out as described in [16]. All titrations were made at 20°C, at ionic strength of 0.16, and the protein concentration was 150 mM. Albumin was titrated with NaOH in presence of a constant concentration of CF, and the results expressed as the mol number of protons bound per mol of protein as a function of pH.

2.6. Effect of CF on the esterase acti6ity of HSA: Hydrolysis of p-nitrophenyl acetate by HSA was studied using the method of Kuromo et al. [17]. The p-nitrophenol formed, was measured by absorbance at 370 nm. HSA concentration was tenfold greater than the substrate concentration (5 mM) in order to generate a pseudo first order reaction of p-nitrophenyl acetate. The apparent rate constant, k, for the substrate hydrolysis by albumin was determined from a plot of log(A -At) against time, where A and At are the absorbances at the completion of the reaction and at time t respectively. Linear regression analysis was used to estimate the k value.

2.7. Analysis of the binding cur6es The binding isotherms were adjusted by a non-lineal regression analysis, with a multiparametric curve-fitting program, using the equation: r=%

ni Ki c (1 + Ki c)

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

183

where r is the binding ratio (the bound ligand concentration over the total protein concentration); ni and Ki are the number of binding sites and the affinity constant of i th type of site; and c is the free ligand concentration.

3. Results

3.1. Structure-binding relationship of beta lactam antibiotics to HSA Table 1 shows the position and type of substituents at R1 and R2 of the CF studied in this work. The binding data for each CF to albumin were analyzed by means of Scatchard plots as illustrated in Fig. 1(A, B). The plots for CFZO, CFNE and CFOP are curvilinear suggesting the presence of at least two independent types of binding sites, while for the other CF, linear plots were obtained (only shown for CFU, CFTA and CFDO). The calculated binding parameters for the cephalosporins studied are listed in Table 2. A comparison of the affinity constants values of the CF assayed offers the following picture: A group of CF (number 1–9 in Table 2) with very weak affinity for HSA (K around 10–100/M), a second group of CF (numbers 10 – 15) having only moderate affinity for HSA (K around 103 M − 1) and finally, strong binding was observed for CFOP, CFNE and CFZO (16 – 18), with two classes of binding sites on albumin. Except for CFXI and CFC, which bind a great number of molecules per protein molecule, the rest of CF assayed yielded a number of binding sites in HSA between 1–5. The effect of the temperature on the CF-HSA complex formation was also analyzed, by determining the enthalpic (DH°) and the entropic (DS°) changes. The DH° was calculated from the van’t Hoff equation:dlnK/dT = DH/RT 2, assuming an independence of the enthalpic change with the temperature variation, while DS° was calculated from the relation DS°= (−DG°+ DH°)/T being DG°=-RTlnK. For all the CF assayed, positive values of both thermodynamic parameters were obtained (see Table 2), suggesting the participation of a ‘hydrophobic effect’ component in the interaction. The effect of the variation of pH between 6 and 8.5 on the CF binding to albumin was also assayed. We have selected the following CF: CFLO (pKa = 3.07), without acid or basic groups at R1 and R2; CFXI (pKa = 3.50, pKa = 7.00), with an amino group at R2; CFGL (pKa = 3.53, pKa 7.00), CFLE (pKa = 5.30, pKa 7.38) and CFZO (pKa =2.1, pKa = 6.92), with basic groups at R1. Fig. 2(A–C) shows the dependence of the affinity constant on the pH, it can be seen that the K values for the CF assayed decrease continuously when the pH raises from 6–8.5. A slightly different behaviour is observed for CFXI, which shows a very slight increase in the affinity value at pH above 8.

3.2. Effect of the CF on the acid-base equilibrium of HSA Fig. 3 shows the effect of CF selected, on the acid-base equilibrium of HSA, monitored by the pH titration technique as it was previously described [16]. DZH +

184 Table 1

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202 Table 1 (continued)

185

186

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

Fig. 1. Scatchard plots for the CF binding to HSA. HSA concentration 250 mM. Medium sodium · ), CFZO (“) and phosphate 20 mM, pH 7.4. Temperature 20°C CFU (), CFTA (Õ), CFDO ( CFNE ().

represents the difference in the mol number of proton bound (calculated per mol of protein) between the protein-drug complex and the free protein, both at the same pH. These measurements were performed at a constant ratio drug/albumin. Positive DZH + values were found for all the CF assayed, suggesting that the protein-drug complex has more protons bound than the free protein does.

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

187

3.3. Identification of the CF binding sites on HSA 3.3.1. Inhibition of the fluorescence of specific fluorescent markers bound to albumin Two specific binding sites have been established on the human serum albumin: site I (so-called warfarin site) and site II (so-called benzodiazepine site) which were characterized by Sudlow et al. [13] and Sjo¨holm et al. [18]. The bilirubin and fatty acid binding sites have also been reported [14,15]. To identify these binding sites, competitive studies were performed using specific markers of binding sites such as Table 2

(i) Binding parameters for the CF studied to HSA CF

n

K(M−1)

1. CFC 2. CFXI 3. CFCl 4. CFDO 5. CFRI 6. CFT 7. CFRA 8. CFGl 9. CFLO 10. CFOL 11. CFTA 12. CFDR 13. CFPI 14. CFU 15. CFLE 16. CFNE

26 29 4.46 3.46 1.31 2.04 2.38 1.54 5.12 2.53 1.98 1.01 2.08 1.76 2.06 0.76 1.32 1.20 1.15 1.06 5.47

(3.90 90.10) (4.30 90.50) (1.30 90.10) (3.80 9 0.90) (4.70 90.30) (4.76 9 0.06) (5.53 90.01) (8.10 90.01) (9.6090.01) (1.21 9 0.50) (1.85 90.14) (1.90 90.50) (2.50 90.20) (3.50 90.20) (4.30 9 0.60) (3.16 90.13) (3.00 91.00) (5.60 90.32) (3.11 9 0.10) (1.28 90.60) (1.00 90.50)

17. CFOP 18 CFZO

101 101 102 102 102 102 102 102 102 103 103 103 103 103 103 104 103 104 102 105 102

(ii) Thermodynamic parameters for the CF-HSA binding CF

dH° (Kcal/mol)

dS° (e. u.)

CFOL CFDO CFOP

13.79 1.2 4.690.6 4.69 0.8 1.39 0.2 5.59 0.4 7.0 90.6

60.0 9 4.2 27.4 95.6 37.3 9 7.3 15.9 9 3.4 39.29 5.3 39.7 9 2.8

CFNE

The values are the mean of four independent measurements. Medium sodium phosphate buffer 100 mM. pH 7.4. Temperature 20°C.

188

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

DNSA (for site I), DSS (for site II), bilirubin and retinoic acid for bilirubin and fatty acid binding sites respectively [15].

Fig. 2. pH effect on the binding of CF to HSA. The results were expressed as dependence of the affinity constant with the pH (see Section 2). Medium sodium phosphate buffer, HSA total concentration 250 mM. [CF]/[HSA] =5. CFZO (“), CFXI (), CFLE (2), CFLO ( ), CFGL (").

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

189

Fig. 2. (continued)

No inhibition of the fluorescence of the marker for Site I (DSNA) was observed, except for CFU, CFT and CFOP where a very poor decrease of the fluorescence of DSNA was observed, suggesting that these CF may interact with site I on HSA. Some of the other CF assayed produced an enhacement of the fluorescence of the bound DSNA (see Fig. 4A) suggesting that these derivatives produce a perturbation at the binding site of this fluorophore. CFRA, CFLE, CFCL, CFT, CFNE, CFDO and CFOP produced a substantial inhibition in the fluorescence of DSS (marker for site II) bound to HSA, (Fig. 4B), suggesting that these CF bind to site II in HSA, the other CF assayed did not produce any change on the fluorescence of this marker. Fig. 4C shows that increasing concentrations of the following cephalosporins: CFZO, CFXI, CFNE, CFGL, CFDR, CFU, CFTA, CFOP and CFOL produced a significant decrease in the fluorescence of BL bound to HSA, suggesting that these CF are displacing BL from its binding site in HSA. Only CFT, CFOP and CFNE produced an inhibition of the fluorescence of the fatty acids site marker bound to HSA as it is shown in Fig. 4D. Other CF assayed produced an enhacement of the fluorescence, as CFCL did, which produced a very significant increase of the fluorescence of RA. It was found that all the CF assayed did not change the fluorescence of BL and of RA in the absence of albumin.

3.3.2. Displacement of CF from HSA by drugs which bind to sites I and II The displacement effect of ligands which bind to sites I and II in HSA on the CFHSA binding was assayed. Two groups of CF were selected: A first group of CF, which produced the displacement of the Site II marker (CFRA, CFDO, CFOP and

190

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

Fig. 3. Difference in proton bound (DZH + ) by HSA-CF complex and albumin alone as a function of pH. Protein concentration 150 mM, ionic strength 0.16 CFZO (“), CFXI (), CFLO ( ). CFLE (2) and CFGL (").

CFLE), and a second group of CF which produced the displacement of the bilirubin from its binding site on HSA (CFTA and CFGL), CFU, which showed inhibition effect on the site I marker, was also included in this group because of the relation that exists between this site and that of the bilirubin. The dependence of the bound fraction with the increasing concentration of ibuprofen (which binds to site II), was assayed on the first and second CF group and the results are shown in Fig. 5A and 5B respectively. In both cases a significant decrease of the bound CF fraction of the CF was observed at increasing concentrations of ibuprofen. These results confirm the above finding that CFRA, CFDO and CFLE bind to the site II in HSA. On the other hand, the displacement effect of the site II marker on binding of CF which bind to the bilirubin site and site I, is suggesting a proximity between these three binding sites in the HSA molecule. The effect of ligands that bind to site I on HSA, such as 7-amino-4-methylcoumarin (a warfarin derivative) and sulfamerazine [18] was assayed. A net displacement by the coumarin is observed for the group of CF which bind to site II (see Fig. 5C) while a slight displacement was observed for the CF which bind to BL binding site on HSA (CFTA and CFGL)(Fig. 5D). On the other hand, CFU, which binds to site I, was displaced in a significant manner by coumarin. Sulfamerazine did not produce any effect on both groups of CF assayed (Fig. not shown).

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

Fig. 4.

191

192

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

Fig. 4. (continued)

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

193

3.3.3. Effect of CF on the hydrolysis of p-nitrophenyl acetate Fig. 6 shows the results obtained, where the k° and ki on the ordinate are the hydrolysis rate constants in absence and presence of i concentration of CF. Examination of the CF presence effect on the esterase activity of albumin, indicates that the CF studied may be divided into three clear subgroups because of the pattern of the curves obtained. A first group of CF (CFDO, CFCL and CFOL; Fig. 6A) produced a significant inhibition of the esterase activity at high [CF]/[HSA] ratio (about a 60% decrease of the initial value). According to Kuromo et al. [17], ligands that produce this inhibition type (R-T-type inhibition), are those which bind to site II and to another unknown site on HSA. A second group of CF (CFGL and CFDR) (Fig. 6B), which produced a slight inhibition of the esterase activity (about 20% decrease of the initial activity at high ratio [CF]/[HSA]). The above mentioned authors have suggested that these ligands are bound only to site II on albumin (R-type inhibition). At last, a third group of CF (CFRA and CFOP) (Fig. 6C), which produced no inhibition effect produce at low [CF]/[HSA] ratio, while an importart decrease was observed at higher [CF]/[HSA] ratios. The ligands that produce this inhibition type (U-R type inibition) bind to site I at low concentrations, and to site II, when the ligand concentration is increased. 3.3.4. Effect of the modification of albumin amino acid residues on the CF– HSA interaction The results were expressed as a percentage of the variation of the affinity constant of the CF, for the chemically modified HSA with respect to unmodified HSA, are shown in Table 3. For this purpose two groups of CF were selected: one with high affinity for HSA (CFOP, which binds to the sites I, II and to bilirubin and fatty acids binding sites and CFNE, which binds to site II and to bilirubin and fatty acid binding sites) and the other, having medium affinity (CFDO which binds to site II and CFOL which bind to bilirubin binding site). HIS 146: the treatment of HSA with diethylpyrocarbonate at a ratio 1:1, labeled the HIS 146. A slight decrease was observed on the affinity for CFOL, while an important decrease was observed for CFOP, CFDO and CFNE. TYR 411: this very reactive tyrosine, was labeled with p-nitrophenylanthranilate as it has been previously reported [9]. As it is shown in Table 3 the binding of Fig. 4. (A) Effect of CF on the bound DSNA fluorescence. The fluorescence was measured at 480 nm when the fluorophore was excited at 340 nm and expressed as percent of variation respect to HSA-DSNA fluorescence in absence of CF. HSA concentration was 9 mM. DSNA concentration 3 mM . ), CFDR ( . ), CFOP (), CFT () and CFU (). (B) Effect of CF on the fluorescence of CFOL ( the site II marker bound to HSA. HSA concentration 15 mM, fluorophore concentration 3 mM. Medium . ), CFT (.), CFOP (), · ), CFCL (2 phosphate buffer 100 mM, pH 7.4. Temperature 20°C. CFDO ( CFLE (2), CFRA ( ) and CFNE (). (C) Effect of CF on the fluorescence of BL bound to HSA. The fluorescence was measured at 530 nm when the BL was excited at 460 nm and expressed as percent of variation respect to HSA-BL fluorescence in absence of CF. HSA concentration 20 mM. BL concentra. ), CFTA (Õ) and tion 15 mM. CFRA ( ), CFLE (2), CFZO (“), CFXI (), CFNE (), CFDR ( . .). (D) Effect of CF on the fluorescence of the fatty acid marker bound to HSA. HSA CFOL ( concentration 15 mM, RA concentration 3 mM. Medium phosphate buffer 100 mM, pH 7.4. Temperature 20°C. CFRA ( ), CFZO (“), CFT (), CFOP () and CFNE ().

194

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

CFDO and CFOP decreased in a significant manner, while the interaction with CFNE was completely abolished and CFOL did not show any change. These results suggests an important interaction between CFNE and the TYR 411 of HSA.

Fig. 5. Displacement of CF by the increasing concentration of drugs: (A) and (B) ibuprofen, (C) and (D) 7-amino-4-methyl-coumarin. [HSA] 180 mM. CFU (), CFGL ("), CFTA (Õ), CFLE (2), CFOP · ) and CFRA ( ). (), CFDO (

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

195

Fig. 5. (continued)

LYS 199: it has been reported that LYS 199 is present at the site II in HSA [10], the reaction of this amino acid with acetylsalicylic acid produced a decrease of the affinity for CFOP, CFNE and CFOL, while CFDO showed a slight enhancement of the affinity for the protein.

196

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

LYS 195: It has been reported that dansylation of albumin at low [reagent]/ [protein] ratio leads to the modification of LYS 195, resulting in a blockage of the benzodiazepine binding site [10]. This HSA labelling produced an important

Fig. 6. (A–C) Effect of CF on the hydrolysis of NPA. Sodium phosphate 100 mM, pH 7.4, Temperature 20°C. HSA 50 mM and NPA 10 mM. All the points are the mean of three measurements. CFDO (.), . .), CFOL ( . .), CFGL ("), CFDR ( . .), CFRA ( ) and CFOP ( · ). CFCL (2

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

197

Fig. 6. (continued)

decrease in the affinity of CFOL and CFOP, the binding of CFDO was completely abolished, while no change was observed for CFNE binding. LYS 225: Anderson et al. [12] have labeled the LYS 220 of bovine serum albumin with fluorescein isothiocianate. In HSA, LYS 220 is not present, but it has a LYS residue at position 225. It is probable that this residue labels with fluorescein isotiocianate. For the binding of the modified HSA, a decrease in the affinity of CFDO and CFNE was observed, the affinity for CFOP was totally abolished, while a slight increase was observed for CFOL. Table 3 Interaction of CF with the native and modified albumins Modified amino acid

HIS 146 TYR 411 LYS 195 LYS 199 LYS 240 LYS 225 LYS 525

Relative change in Cephalosporin affinity for chemical modified HSA (%) CFDO

CFOL

CFNE

CFOP

76 78 0 144 4 17 18

85 122 7 10 30 200 7

32 0 118 490 300 6 120

18 12 3 6 49 0 0

The results were expressed as the percentage of the modification of the affinity constant of the CF binding to the chemical modified HSA with respect to the unmodified protein. Average values resulting from three independent measurements. HSA employed was 200 mM.

198

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

LYS 240: this amino acid was labeled with 2,4-dinitrofluorobenzene. A decrease in the binding of CFDO, CFOP and CFOL was observed, while an increase in the CFNE affinity was observed. LYS 525: the glycosilation in vitro of HSA, is the reaction of glucose with lysyl residues, the labelling resulting in a greater proportion of the LYS 525 [11]. A significant decrease in the interaction was observed for the CF assayed except CFNE, which showed a slight increase in the affinity for the modified protein.

4. Discussion After analysing the results, it is evident that the presence of rich electronic density substituents at R1 and R2 position increase in a significant manner the affinity for albumin. Further information about the importance of chemical structure characteristics of both substituents on the binding, can be obtained by comparing CF with similar substituents. CFC, CFGL, CFLO, CFTA and CFPI have at R2 a group with a low electronic density: a methyl acetoxi group, HSA showing a very low or only moderate affinity for them. CFC is the derivative with the lowest affinity, probably because it is the only CF that has an aliphatic group at R1. For the other CF of this group it is observed that the affinity values increase in a direct manner with the R1 electronic density. By comparing CFRA, CFDR and CFLE with a common methyl group at R2, the two first derivatives showed only moderate affinity for HSA, while CFRA with a non-aromatic cycle at R1 showed very low affinity. In the same manner, CFXI and CFU both with an acetamide group at R2, the presence of N-acyl at R1 in CFU, produce an increase in the affinity with respect to CFXI. The above results suggest that if R2 is a group without an electrical charge and with low electronic density, the interaction of the CF with HSA may vary from very poor to moderate, depending on the R1 electronic density. On the other hand, CFOL and CFOP have a non-charged group at R2, but with high electronic density. For these derivatives, the affinities are moderate and high, respectively, depending on the R1 electronic density in a direct manner. A different behaviour showed CFRI, CFDO and CFT, which possess a substituent cationic group at R2, and groups with different characteristics at R1, all these derivatives showing low affinity for HSA. This fact suggests that in this case R1 substituent does not influence the affinity, i.e. CFT has an heterocycle group at R1 with high electronic density, however, this CF showed a very low HSA-binding extent. The above results suggest that the binding of CF to HSA is driven mainly by the R2 chemical properties. It can be concluded that: 1. the presence of positive charge at r2 produce a decrease of the affinity, 2. if R2 is a group without charge, the affinity for the HSA increases in a direct manner with the electronic density of this group,

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

199

3. the presence at R1 of substituents with high electronic density, favours the CF-HSA interaction, but this fact is not a sufficient condition to possess high affinity values. The dependence of the CF-HSA interaction with the pH variation of the medium, is suggesting that electrostatic forces are involved in CF-HSA interaction. CFLO has neither acid nor basic group at R1 and R2, however, the interaction of this derivative with albumin decreases when the pH raises from 6 to 8.5, suggesting an interaction between some prototropic groups positively charged on the protein and the carboxylic group of this CF. The other CF assayed, showed the same behaviour, except for CFXI. This CF has an amino group at R2 with pKa 7.00, and it previously was discussed, that the presence of cationic groups at R2, does not favor the interaction, then it is possible that the increase of CFXI affinity for HSA at pH higher than eight might be caused by the loss of its positive electrical charge. Moreover, these results are consistent with those obtained from the acid-base titration of HSA in the presence of CF, demonstrating the involvement of protons in the CF-HSA complex formation. In the alkaline pH region, DZH + is clearly positive for all the CF assayed, which may be caused by an increase in the pKa of the positively charged group on the protein, suggesting that imidazole groups from histidines (pKa 6.9) may be involved in the interaction CF-HSA. Elsewhere, the increase of DZH + at pH 8 – 8.5 suggests that e-amino groups from lysines (pKa 8.8) in HSA may also be perturbed [19]. The pH dependence of the binding constant is related with the difference in bound protons associated to the binding process, by the Wyman relation [20]: dlnK/dpH = − DH + , where DH + is approximately equal to the value of DZH + . As it was previously discussed, the affinity values decrease with increasing pH, which implies that dInK/dpH is negative, while DZH + is positive for all the CF assayed. This fact demonstrates the fulfillment of Wyman expression and confirms that the results from Fig. 2 are in agreement with those from Fig. 3. Examination of drug-induced changes in the fluorescence of bound probe indicates that only CFU, CFT and CFOP interacts slightly with site I on HSA, while site II possesses capacity to bind: CFRA, CFLE, CFCL, CFT, CFNE, CFOP and CFDO. The bilirubin binding site showed the capacity to interact with a great number of CF, such as: CFZO, CFXI, CFNE, CFGL, CFDR, CFU, CFTA, CFOP and CFOL. Finally only CFNE showed capacity to bind to the fatty acid binding sites on HSA. On the basis of the esterase activity inhibition pattern proposed by Kuromo et al. [17], the CF assayed can be divided into three groups: (R-T): CFOL, CFCL and CFDO bind to site II and then to an unknown site; (R): CFGL and CFDR bind to site II exclusively; (U-R): CFRA and CFOP bind to site I and then to site II. By comparing with the results from fluorescence marker displacement, it is evident that there is not a complete correspondence. In general the CF that produce a displacement of the BL site marker corresponds to the R group proposed by Kuromo et al. [17] while those that produce the DSS displacement correspond to the R-T group.

200

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

For CFOP both the competition results and those of the esterase activity inhibition are in agreement, because it displaces both fluorescent markers (DSNA and DSS) and it corresponds to the U-R group of Kuromo et al. [17]. Our results indicate that the CF bind to site II and bilirubin binding site rather than site I and they are in opposition with those obtained by Tawara et al. [3] which suggested that CF bound to the warfarin site (site I) in HSA. These authors assayed only the effect of increasing concentration of warfarin and phenylbutazone on the CF – HSA binding. We consider that only one direct displacement experience is insufficient to determine the identity of a ligand-binding site in HSA. A double competition experiment is necessary to confirm the identity of a binding site, first, testing the displacement of a specific fluorescent marker (as it was proposed by Sudlow [13]) at increasing concentrations of CF and then, assaying the displacement of the bound CF to HSA by other ligands whose binding site on albumin is known. However, in spite of the incomplete results obtained by Tawara et al. [3], some of their results are in agreement with ours, because we also found that 7-amino-4-methyl coumarin displaced the CF assayed. However, sulphamerazide, another site I specific ligand, produced no effect on CF binding. We also found that either ibuprofen or deoxicholate, site II specific ligands, displaced all the CF assayed from the HSA. These results may be explained by a cascade mechanism such as McNamara et al. postulated [2] for CFNE behaviour. They found that CFNE displaced warfarin from HSA, but they also found that probenecid and diazepan (site II markers), were displaced from albumin by CFNE. They postulated that CFNE binds to site I on albumin and its displacement by probenecid and diazepan was due to a conformational change in albumin which subsequently caused a change in the affinity of the warfarin (CFNE) binding site. The extent of the binding of CFNE, CFDO and CFOP was remarkably reduced by the modification of the HIS 146, and a similar behaviour is observed for these CF when TYR 411 is modified, while a slight, or no effect is observed for CFOL in both amino acids. These results are consistent with fluorescent markers displacement experiments, because the three CF mentioned above displaced DSS, a site II marker, from albumin and it has been reported [21] that either HIS 146 or TYR 411 are involved in this site. From the acid – base titration curve of the HSA–CF complex it was found that basic amino acids such as histidine and lisine were perturbed during the protein complex formation. Jacobsen and Faerch [8] found that LYS 195 and 225 are present at the bilirubin binding site. However, Farruggia et al. [21] found that LYS 225 and 240 were perturbed in bile salts binding to HSA (site II). Besides, the dansylation of LYS 195 resulted in a blockage of benzodiazepine binding sites (site II) [22]. Our finding confirmed that LYS 195 and 240 are present at the bilirubin binding site, because the blockage of these residues inhibited significantly the binding of CFOL, which according to the fluorescence markers displacement experiments, only displaced the BL from HSA. On the other hand, it was observed that CFDO displaced DSS from site II in HSA and the fact that the labelling of LYS 195, 225 and 240 inhibited CFDO binding to HSA, demonstrated that these residues are

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

201

perturbed in the binding to site II as Farruggia et al. [21] had found, and this indicates that these residues might be located near each other. Moreover, for CFOP, which displaced either DSS or BL from its sites on albumin, an important decrease was observed in its binding to the three modified albumins (LYS 195, 225 and 240). Based on competitive inhibition studies, LYS 199 interacts with drugs that bind to site I [17]. This is in agreement with our finding that CFOP, which decreased DSNA (site I marker) fluorescence, presented weak interaction with LYS 199 modified albumin. We found that acetylation of this LYS inhibited the interaction between HSA and CFOL, that displaced only BL from its site. This behaviour may be attributed to the reactivity of this LYS residue with an unusually low pKa of 7.9, which can be rationalized based on the close interaction with HIS 242. Okabe and Hashizime [23] reported that the non-enzymatic glycosylation of HSA results in a glycosylation of LYS 525 and LYS 199 as primary and secondary amino acid modified respectively. Our finding suggests that LYS 525 is in site II, based on the decrease of the interaction of CFOP and CFDO with modified glycosylated albumin. CFOL also show a decrease in its interaction with this modified albumin. Taking into account that a similar behaviour was observed for this CF and the LYS 199 modified albumin, we confirm that both LYS 199 and LYS 525 are in site II, but we postulate an interaction between this residue and the BL site as it was previously reported [21].

Acknowledgements This work was supported by Grant: 98/93 from Secretaria de Ciencia y Te´cnica (U.N.R). We thank Mrs A. Robson for the language correction of the manuscript.

References [1] C. Briand, M. Sarrazin, R. Peyrot, M. Bourdeaux, J. Sari, Study of the interaction between human serum albumin and some cephalosporins, Mol. Pharmacol. 21 (1982) 92 – 99. [2] P. McNamara, V. Trueb, K. Stoeckel, Ceftriaxone binding to human serum albumin. Indirect displacement by probenecid and diazepam, Biochemical Pharmacology 40 (1990) 1247 – 1253. [3] S. Tawara, S. Matsumoto, Y. Matsumoto, T. Kamimura, S. Goto, Structure-binding relationship and binding sites of cephalosporins in human serum albumin, J. Antibiotics 45 (1992) 1346 – 1357. [4] B. Nerli, G. Pico´, Evidence of human serum albumin beta-lactamase activity, Biochemistry and Molecular Biology International 32 (1994) 789 – 795. [5] B. Nerli, G. Pico´, Identification of the cephalosporin-human serum albumin binding sites, Pharmacology and Toxicology 73 (1993) 297 – 300. [6] N. Carrillo, R. Vallejos, Essential histidyl residues of ferredoxin-NADP + oxidoreductase revealed by diethyl pyrocarbonate inactivation, Biochemistry 25 (1983) 5889 – 5896. [7] B. Desfosses, N. Cittanova, W. Urbach, M. Waks, Ligand bidnign at membrane mimetic interfaces, Eur. J. Biochem. 199 (1991) 79 – 87. [8] J. Jacobsen, T. Faerch, Kinetics of the neutral transition of human serum albumin, Biochimica et Biophysica Acta 670 (1981) 124 – 128.

202

B. Nerli et al. / Chemico-Biological Interactions 104 (1997) 179–202

[9] U. Lind, U. Kragh-Hansen, J.V. Moller, Binding of bromohenol blue by chemical modifications of human serum albumin, Biochim. Biophys. Acta 371 (1974) 451 – 461. [10] J. Walker, Lysine 199 of human serum albumin is modified by acetylsalicylic acids, FEBS Letters 66 (1976) 73–175. [11] N. Iberg, R. Flu¨ckiger, Nonenzimatic glycosylation of albumin in vivo, J. Biolog. Chem. 261 (1986) 13542–13545. [12] L.O. Andersson, The heterogeneity of bovine serum albumin Biochim, Biophys. Acta 117 (1966) 115–133. [13] G. Sudlow, D. Birkett, D. Wade, The characterization of two specific drug binding sites on human serum albumin, Molecular Pharmacology 11 (1975) 824 – 832. [14] A. Lamola, J. Flores, Effect of buffer viscosity on the fluorescence of bilirubin bound to human serum albumin, J. Am. Chem. Soc. 104 (1982) 2530 – 2534. [15] B. Nerli, G. Pico´, Use of retinoic acid as a fluorescent probe to study conformational change of the albumin molecule, Biochemistry and Molecular Biology International 32 (1994) 781 – 788. [16] L. Janssen, M. Wilgenburg, Experimental observation of ionic forces in drug-protein interactions as illustrated by the binding of sulfaethidole by human serum albumin, Molecular Pharmacology 14 (1978) 884–889. [17] Y. Kurono, I. Kushida, H. Tanaka, K. Ikeda, Esterase-like activity of human serum albumin. VIII Reaction with amino acid p-nitrophenyl esters, Chem. Pharm. Bull. 40 (1992) 2169 – 2172. [18] I. Sjo¨holm, B. Ekman, A. Kober, I. Ljungsted-Pahlman, B. Seiving, T. Sjo¨din, Binding of drugs to human serum albumin. The specificity of three binding sites as studied with albumin inmobilized in microparticles, Molecular Pharmacology 16 (1979) 767 – 777. [19] C. Tanford, Physical Chemistry of Macromolecules, Wiley, New York, 1961, pp. 234 – 403. [20] J. Wyman and S. Gill, Binding and linkage. Functional chemistry of biological macromolecules. University Science Books, California, 1990, pp. 63 – 163. [21] B. Farruggia, F. Garcı´a, G. Pico´, Structural features of the hydroxy- and keto- disubstituted bile salts: human serum albumin binding, Biochimica et Biophysica Acta 1252 (1995) 59 – 68. [22] H. Fehske, E.W. Muller, U. Wollert, The location of drugs binding sites in human serum albumin, Biochem. Pharmacol. 30 (1981) 687 – 692. [23] N. Okabe, N. Hahsizume, Drug binding properties of glycosylated human serum albumin as measured by fluorescence and circular dichroism, Biol Pharm. Bull 17 (1994) 16 – 21.

.