Abstract
An in vitro screening model was developed to determine the reactivity of acyl glucuronide metabolites from carboxylic drugs. This assay is composed of two phases. The first is a phase of biosynthesis of acyl glucuronides by human liver microsomes (HLM). The second, during which acyl glucuronides are incubated with human serum albumin (HSA), consists of assessing the reactivity of acyl glucuronides toward HSA. Both phases are performed successively in the same experiment. This model was validated using eight carboxylic drugs that were well known for their reactivity, their extent of covalent binding, and their immunological potential. These products were representative of the scale of reactivity. Each compound was incubated with HLM at 400 μM and metabolized into acyl glucuronide to different extents, ranging from 5.6% (tolmetin) to 89.4% (diclofenac). The first-order aglycone appearance rate constant and the extent of covalent binding to proteins were assayed during the incubation of acyl glucuronides formed with HSA for 24 h. Extensive isomerization phenomenon was observed for each acyl glucuronide between the two phases. An excellent correlation was observed (r2, 0.94) between the extent of drug covalent binding to albumin and the aglycone appearance constant weighted by the percentage of isomerization. This correlation represents an in vitro reactivity scale, which will be helpful in drug discovery support programs to predict the covalent binding potential of new chemical entities. This screening model will also allow the comparison of acyl glucuronide reactivity for related structure compounds.
Many acidic drugs with carboxylic acid functions are metabolized to reactive acyl glucuronides. These metabolites are unstable at physiological pH and can result in free aglycone by hydrolysis and lead to positional isomers by acyl migration. Acyl migration involves the transfer of the acyl group from the position 1β to the C-2, C-3, or C-4 position of the glucuronic acid ring, which results in the formation of isomeric acyl glucuronides (Faed, 1984; Spahn-Langguth and Benet, 1992).
Acyl-migrated glucuronide isomers were shown to bind covalently to proteins in vitro and in vivo causing potential toxicity (Spahn-Langguth et al., 1996). The glucuronide-mediated toxicity depends on the covalent binding of acyl glucuronides to specific target proteins located in specific tissues. The toxicological mechanisms are still unknown (Park et al., 1987; Riley and Leeder, 1995; Dansette et al., 1998).
However, data literature review provides much information regarding immunologically based and clinically relevant adverse reactions of several drugs that are probably related to the formation of highly reactive acyl glucuronides. These drugs include tolmetin, zomepirac, diclofenac, and diflunisal (Hasegawa et al., 1982; Hyneck et al., 1988a,b; Ojingwa et al., 1988, 1994; Smith et al., 1990, 1992;Kretz-Rommel and Ding et al., 1994; Boelsterli, 1994; Williams and Dickinson, 1994; Ebner et al., 1999).
Thus, models for predicting the glucuronide-mediated toxicity of new chemical entities are desirable particularly for all drug discovery support programs. Benet et al. (1993) found on six drugs a correlation between the extent of covalent binding to protein and the global degradation rate constant of β-1-O-acyl glucuronide. But, this correlation seems to be less evident when literature data are implemented in the model (Spahn-Langguth et al., 1996). The global degradation rate constant of β-1-O-acyl glucuronide may not be used as a direct measure of potential protein reactivity. As covalent binding to proteins primarily occurs via acyl-migrated isomers, a distinction between the degradation due to hydrolysis and the degradation due to acyl migration is certainly necessary.
On the other hand, studies assessing the reactivity of acyl glucuronides need a preliminary step for the chemical synthesis or the isolation from urine or microsomal incubation of the β-1-O-acyl glucuronide standard. This step is often the limiting factor for an early detection of highly reactive acyl glucuronides for new chemical entities.
The objective of this study was to develop a screening model for the reactivity of acyl glucuronides. This reactivity was assessed by the instability (hydrolysis + isomerization) of these acyl glucuronides and by their extent of covalent binding to human serum albumin. The model developed allowed in the same experiment, i) the production of acyl glucuronide by human liver microsomes, ii) the determination of its hydrolysis and isomerization rate constant, and iii) the determination of the extent of covalent binding with HSA.1 The model was evaluated with eight acidic drugs metabolized into acyl glucuronide (tolmetin, zomepirac, diclofenac, fenoprofen, ketoprofen, ibuprofen, suprofen, and furosemide). These drugs have been extensively studied and represent a large scale of reactive products (Smith et al., 1986; Hyneck et al., 1988a; Castillo and Smith, 1991; Volland et al., 1991; Hayball et al., 1992; Dubois et al., 1993, 1994; Smith and Liu, 1993; Castillo et al., 1995; Ebner et al., 1999; Mizuma et al., 1999). Tolmetin and zomepirac were withdrawn from the market because of hypersensitivity reactions; ibuprofen is considered to be the safest nonsteroidal anti-inflammatory drug, and furosemide shows very low levels of covalent binding.
Materials and Methods
The experiment was divided into two steps. The first phase (“biosynthesis” phase) was elaborated to synthesize acyl glucuronide by human liver microsomes. The second phase (“reactivity” phase) was dedicated to the determination of the hydrolysis and isomerization rate constant of 1-O-acyl glucuronide and the extent of covalent binding to HSA.
Chemicals.
Tolmetin, zomepirac, diclofenac, suprofen, fenoprofen, ketoprofen, ibuprofen, furosemide, HSA (fraction V), UDP-N-acetylglucosamine, UDP-glucuronic acid, and bovine β-glucuronidase were purchased from Sigma (l'Isle d'Abeau Chesnes, France). The pool of human liver microsomes (29 donors) used in this study was prepared by Biopredic International (Rennes, France). All other reagents and solvents were of analytical grade and obtained from Sigma or Merck (Darmstadt, Germany).
In Vitro Biosynthesis of Acyl Glucuronides.
Test compounds (tolmetin, zomepirac, fenoprofen, ketoprofen, ibuprofen, suprofen, diclofenac, and furosemide; Fig.1) were incubated at 400 μM in triplicate for 4 h at 37°C with human liver microsomes (3 mg/ml) in 100 mM Tris buffer, pH 7.4, containing 1% dimethyl sulfoxide, 5 mM MgCl2, 5 mM UDP-glucuronic acid, and 1 mM UDP-N-acetylglucosamine in a final volume of 4.8 ml. Two 400-μl aliquots were withdrawn at time points 0 and 4 h. The reaction was stopped by protein precipitation with the addition of 1 ml of 4% trifluoroacetic acid in acetonitrile (pH lowered to 3–4) and then centrifuged at 1500 rpm for 10 min. Supernatants were collected and stored at −80°C until analysis for the determination of the amount of acyl glucuronide synthesized.
Structures of the eight drugs for which in vitro potential of metabolism by human microsomes and the extent of covalent binding with HSA were investigated.
Reactivity of Acyl Glucuronides.
At the end of the 4-h incubation duration, the remaining mixture was centrifuged at 1500 rpm for 20 min (withdrawal of microsomes). Supernatants (3.0 ml) were then transferred into capped new tubes and incubated for various times with 0.5 mM HSA. At the sample times of 15 and 30 min and 1, 2, 4, 6, and 24 h, a 300-μl aliquot was withdrawn and transferred into a tube containing 1 ml of 4% trifluoroacetic acid in acetonitrile and then centrifuged at 1500 rpm for 10 min. Supernatants were collected and stored at −80°C. Pellets were washed with 1 ml of 5% aqueous trifluoroacetic acid (gentle shaking for 10 min and then centrifugation at 1500 rpm for 10 min) and then three times successively with 1 ml of methanol (gentle shaking for 10 min and then centrifugation at 1500 rpm for 10 min). Washed pellets and dry residues from supernatants from the last washing step only were stored at −80°C until analysis.
Controls were performed in parallel and treated identically except that no cofactor was added during the incubation with microsomes. One 400-μl aliquot was withdrawn after 0 and 4 h of incubation in the microsome mixture and treated like the aliquots of the biosynthesis phase. The residual mixture was centrifuged at 1500 rpm for 20 min; then supernatants (1.0 ml) were transferred into capped new tubes and underwent the reactivity phase (i.e., incubation for various times with 0.5 mM HSA). After an incubation period of 6 and 24 h, a 300-μl aliquot was withdrawn and treated like the other aliquots of the reactivity phase.
Analytical Method.
Incubation samples generated for each drug were analyzed by a generic liquid chromatography-tandem mass spectrometry method. The analytical column was a Hypersil BDS (125 × 4-mm i.d.; Thermoquest; Thermo Finnigan MAT; San Jose, CA). Separation of the different acyl glucuronide isomers from their aglycone was achieved using a gradient elution mode. The mobile phase was a mixture of acetonitrile/10 mM acetate ammonium buffer (70:30, v/v) + 0.5% acid acetic for solvent A and acetonitrile/10 mM acetate ammonium buffer (4:96, v/v) for solvent B. The gradient profile was adjusted for each compound with a flow rate of 1 ml/min and run time around 15 min. Detection and quantification were performed by tandem mass spectrometry using a turbo ion spray ion source (API 365; Applied Biosytems, Toronto, ON, Canada).
Concentrations of aglycone, 1-O-β-acyl glucuronide, and acyl glucuronide isomers were determined in supernatants of the biosynthesis and of reactivity phase. The principle of the assay is summed up below. Each sample was divided into three aliquots. In the first aliquot, free aglycone concentration (i) was determined. The second aliquot was incubated with 1000 units of bovine β-glucuronidase at 37°C for 2 h to cleave β1-conjugates and liberate the corresponding aglycone part. A positive control with phenolphthalein-1-O-glucuronide was examined to verify the β-glucuronidase enzyme activity. The aglycone concentration found (ii) minus the free aglycone concentration (i) determined earlier corresponded to the 1-O-β-acyl glucuronide concentration. In the same way, the third aliquot was submitted to alkaline hydrolysis (1 N KOH at 80°C for 3 h) to hydrolyze all acyl glucuronides present into their corresponding aglycone. The concentration of acyl glucuronide isomers was estimated as the difference between total aglycone concentration (iii) and aglycone concentration coming from the cleavage of β1-conjugates (ii). As mentioned earlier, two 400-μl aliquots were withdrawn at time points 0 and 4 h of the biosynthesis phase, and one of these aliquots was used to check the total hydrolysis of the 1-O-β-acyl glucuronide and the chemical stability of the acyl glucuronides isomers during the 2-h incubation period with β-glucuronidase. The total hydrolysis of the 1-O-β-acyl glucuronide and the chemical stability of the acyl glucuronides isomers were evaluated by direct graphic assessment of the peak's area, as shown in Fig.2. This preliminary step was performed before the analysis of all the samples.
Principle of assay.
➁, liquid chromatography/tandem mass spectrometry chromatogram obtained after drug incubation with microsomes or HSA; 1-O-acyl glucuronide, 2-, 3-, and 4-O-acyl glucuronide were separated and free aglycone drug was assayed. ➀, samples were hydrolyzed with β-glucuronidase to estimate 1-O-acyl glucuronide (obtained by the difference of ➀ − ➁). ③, the acyl glucuronide isomer concentration was obtained after alkaline hydrolysis (③ − ➀).
The extent of covalent binding yield to human serum albumin was evaluated for each drug. Extensively washed protein pellets obtained during the second incubation period were submitted to alkaline hydrolysis (1 N KOH at 80°C for 3 h). Moles of aglycone released by this procedure were considered as irreversibly bound to HSA. Protein pellets from control samples, performed without cofactors, were also submitted to alkaline hydrolysis. Although no aglycone release was expected, the detected levels were considered as background noise and were subtracted from the results obtained.
Samples for the determination of free aglycone and the 1-O-β-acyl glucuronide concentration were just diluted in mobile phase from 1/10 to 1/40, depending on the sensitivity of the compound in mass spectrometry before injection into the analytical system. Calibration curves, from 50 to 10,000 ng/ml, were prepared by spiking the adequate amount of standard (ketoprofen, diclofenac, suprofen, tolmetin, zomepirac, furosemide, fenoprofen, and ibuprofen) in the blank matrix. Samples for the determination of acyl glucuronide isomers and the covalent binding concentration were extracted by a solid-phase extraction method using Oasis HLB cartridges (Waters, Saint Quentin en Yvelines, France). Briefly, the cartridges were conditioned by 1 ml of methanol followed by 1 ml of water. Then, the samples were loaded onto the cartridges. The cartridges were washed with 1 ml of water. Elution was based on 1 ml of the mixture acetonitrile/10 mM ammonium acetate buffer (75:25, v/v) + 0.05% acetic acid. Calibration curves, from 5 to 1000 ng/ml, were prepared in the chromatographic mobile phase. Dry residues from the last washing fraction of protein pellets were redissolved in 1 ml of chromatographic mobile phase and analyzed to ensure the exhaustiveness of the washing procedure. Only traces of free aglycone or free acyl glucuronide should still remain.
Analyte peak areas (determined by mass spectrometry) were used for quantification together with the different calibration curve (external calibration). Quality control samples were performed for each phase of the model at three concentrations (15, 150, and 300 μM). Cofactors were added at the end of the incubation period for QCs of the biosynthesis phase. QCs of the reactivity phase were incubated with HSA for the selected duration and then treated as an experimental incubation. QCs were used to ensure accuracy and precision of the method. All QCs showed accuracy within 80 to 120%.
Data Analysis.
The degradation rate was defined as the initial loss of the 1-O-β-acyl glucuronide component. Hydrolysis was defined as the initial formation of the aglycone, and acyl migration (isomerization) was defined as the formation of positional isomers. According to Sidelmann et al. (1996), the hydrolysis rate was calculated as the degradation rate corrected for the formation of positional isomers, and the acyl migration rate was calculated as the degradation rate corrected for hydrolysis. Kinetic data of degradation of acyl glucuronides were calculated by nonlinear regression analysis of the measured data using the equation for first-order reaction kinetics, C =C(0)e−kt. In the same way, aglycone release kinetic data were analyzed by nonlinear regression analysis using the equation for first-order reaction kinetics, C =C(0)ekt .
Results
Biosynthesis of Acyl Glucuronides.
The first step consisted in acyl glucuronide synthesis by human liver microsomes in straight conditions (400 μM substrate; 3 mg/ml microsomal proteins; 4 h incubation). The conditions retained were able to produce acyl glucuronide for the eight compounds tested. The percentage of metabolism or the percentage of acyl glucuronides formed was determined by the quantification of parent drug depletion during the 4-h incubation periods. This percentage of metabolism ranged from 5.6% (tolmetin) to 89.4% (diclofenac) (Table1). Thus, this first step allowed an estimation of the capability of each drug to be metabolized into acyl glucuronide.
Percentages of drugs metabolized into acyl glucuronide and of acyl glucuronide isomerized between the biosynthesis phase and the reactivity phase
Reactivity Assessment with Human Serum Albumin.
Instability assessment
During the second step, the supernatant of the first step was incubated with 0.15 mM phosphate buffer containing 0.5 mM HSA for 24 h. The time-dependent degradation of 1-O-acyl glucuronide and the appearance of its isomers and hydrolyzed aglycone were monitored for each drug. An example of the time course observed for each species derived from diclofenac and ibuprofen acyl glucuronide is shown in Fig.3. The sum of the initial concentrations observed is lower than expected (400 μM). A binding (reversible or irreversible) on microsomes and a drug loss during centrifugation could explain this decrease on concentrations. This phenomenon should not effect the data interpretation because we always referred to the initial acyl glucuronide concentration of the second phase. 1-O-Acyl glucuronides were mostly expected to be detected at the beginning of the incubation with HSA. However, extensive acyl migration occurred during the process between the two steps. A majority of acyl glucuronide isomers were detected from early kinetic points. The analytical method developed allowed good separation between 1-O-acyl glucuronide and its isomers but not totally between the isomers themselves. This isomer resolution was time consuming and not compatible with a screening purpose. Therefore, high isomerization could not be seen since the levels of isomers remained constant. Only the time-dependent degradation of acyl glucuronide isomers by hydrolysis could be followed. 1-O-Acyl glucuronide levels remained low over the incubation period. This phenomenon observed for diclofenac and ibuprofen was also observed for the other compounds studied. The percentage of isomerization between the two steps is presented in Table 1 for each compound. The determination of the aglycone appearance and acyl glucuronide degradation rate was shown in Fig. 4. Apparent first-order degradation and the appearance constants for all compounds are listed in Table2.
Time courses of rearrangement and hydrolysis of diclofenac (A) and ibuprofen (B) acyl glucuronide in 0.15 mM phosphate buffer containing 0.5 mM HSA, pH 7.4, at 37°C.
Each point and vertical bar represent the mean ± S.D. of three independent series.
Nonlinear regression analysis of the degradation of the whole acyl glucuronide (top) and appearance of aglycone (bottom) during incubation at 37°C with HSA, pH 7.4.
♦, tolmetin; ▪, zomepirac; ▴, ketoprofen; ▵, fenoprofen; ●, suprofen; ○, furosemide; ×, diclofenac; ✳, ibuprofen.
Apparent first-order degradation of various acyl glucuronides and aglycone appearance constants measured during incubation phases with 0.5 mM HSA, pH 7.4, at 37°C
Irreversible binding to HSA.
The time dependence for irreversible binding to HSA of each acyl glucuronide studied was also investigated during this second step. The extent of covalent binding was expressed in millimoles of aglycone covalently bound per mole of protein. As shown in Fig.5, all acyl glucuronides rapidly produced a covalent adduct with HSA. The maximum yield was achieved after 4 to 6 h of incubation, except for fenoprofen for which maximum covalent binding was achieved after 24 h of incubation. The maximum extent of covalent binding varied from 0.43 to 8.20 mmol irreversibly bound/mol of protein (Fig.6A). However, the amount of drug irreversibly bound was obviously related to the amount of acyl glucuronide present at the beginning of the reactivity phase. Thus, the extent of covalent binding was normalized to protein content and expressed as the percentage of total acyl glucuronide present at the beginning of the reactivity phase. Percentages of covalent binding ranged from 5.7 to 0.34% (Fig. 6B). This expression of the extent of covalent binding changed the compound ranking. Tolmetin, which is known to be the most reactive acyl glucuronide, changed from a low intrinsic value of covalent binding (Fig. 6A) to the highest percentage of covalent binding related to a low amount of acyl glucuronide (Fig. 6B). On the other hand, furosemide remained in the same position no matter what expression system was chosen, thus showing the low covalent binding capacity of furosemide acyl glucuronide (Benet et al., 1993;Mizuma et al., 1999). The ranking according to the amount of drug irreversibly bound observed for the eight compounds studied was consistent with the literature data. Indeed, tolmetin, zomepirac, and diclofenac (Smith et al., 1986; Hyneck et al., 1988a; Munafo et al., 1989) are considered as reactive products, whereas ibuprofen, ketoprofen, and furosemide (Dubois et al., 1993; Castillo and Smith, 1995; Presle et al., 1996; Mizuma et al., 1999) are mentioned as safer products.
Time-dependent irreversible binding of several aglycone after incubation of their respective acyl glucuronide (produced by human liver microsomes; incubation of 400 μM aglycone) in human serum albumin solution 0.5 mM at 37°C, pH 7.4.
Data are the average of triplicate measurements. ♦, tolmetin; ▪, zomepirac; ▴, ketoprofen; ▵, fenoprofen; ●, suprofen; ○, furosemide; ×, diclofenac; ✳, ibuprofen.
The ranking of compounds according to their extent of covalent binding expressed in millimoles irreversibly bound per mole of protein and normalized by protein content (A) and expressed as the percentage of total acyl glucuronide present at the beginning of the reactivity phase (B).
Data are the average of triplicate measurements; bars indicate +S.D.
Analysis of the last washing fraction of protein pellets revealed that the washing procedure was sufficiently exhaustive for most of the compounds. The mean values of aglycone found were below the limit of quantification (<5 ng/ml) for ketoprofen, diclofenac, suprofen, tolmetin, and zomepirac and were equal to 77 ng/ml for furosemide, 63 ng/ml for fenoprofen, and 29 ng/ml for ibuprofen. The same results were obtained for the analysis of covalent binding in controls. Background noise levels found for furosemide, fenoprofen, and ibuprofen were subtracted from the results obtained.
Discussion
The objective of this work was to develop a screening model to assess the reactivity of acyl glucuronides. This screening model allowed us to assess the instability of the biosynthesized acyl glucuronides and to rank compounds according to their maximum covalent binding to HSA.
The first biosynthesis phase was designed with straightforward experimental conditions for standardization purpose. Being easy to work on, human liver microsomes were used for the production of acyl glucuronides. High protein concentration (3 mg/ml), use of glucuronosyl transferase activities activator (UDP-N-acetylglucosamine), high drug concentration (400 μM) of incubation, and long incubation duration (4 h) were chosen as standard conditions to maximize the biosynthesis of acyl glucuronide but were not specifically optimized for each case. Consequently, the linearity of acyl glucuronide formation could not been studied for each compound. However, the selected conditions should allow comparison of acyl glucuronidation potential between compounds from the same chemical family. In the same way, human serum albumin was chosen as a reference target protein to assess the extent of covalent binding during the reactivity phase. The extent of covalent binding was shown to vary greatly depending on the nature of albumin preparation used (Williams and Dickinson, 1994; Ebner et al., 1999). Thus, in vitro assays on covalent binding to proteins can be expected to be highly variable. Moreover, plasma proteins are not the only targets of covalent binding. Acyl glucuronides can irreversibly bind to several tissues or organ macromolecules. Unfortunately, irreversible binding on all proteins cannot be assessed in a screening process. Furthermore, HSA remains the protein most extensively studied. It is widely distributed in the plasma compartment and easily related to the immune system. The knowledge gathered on covalent binding of acyl glucuronide and HSA allowed the comparison and validation of the results achieved with this model. A product showing a significant covalent binding to HSA in our model will require specific attention during the development process. The constant values calculated for the disappearance of acyl glucuronides were lower than those previously published, especially for zomepirac and tolmetin (Hasegawa et al., 1982; Ojingwa et al., 1994). In fact, those reported values represented the global degradation of 1-O-acyl glucuronide (hydrolysis + isomerization), whereas the values presented in this study only represent the degradation of isomeric forms. In the conditions of analysis described here, the exact distribution of isomers could not be determined. Only the hydrolysis of acyl glucuronide isomers could be examined. Therefore, the greater the isomerization process was compared with hydrolysis, the lower the degradation constant was. Thus, the values observed in this study were obviously lower than the published values. The results achieved showed that the acyl glucuronide isomer degradation constant seemed to be less interesting because we could not distinguish whether the acyl glucuronide isomers were strongly isomerized or were very stable. This could be illustrated by the low constant value observed for furosemide. The constant determined for the rate of aglycone appearance could then be an alternative parameter for the assessment of acyl glucuronide instability. Indeed, the appearance of aglycone in the second incubation medium came from the hydrolysis of 1-O-acyl glucuronide and its isomers. The rate of aglycone release could be an indicator of the instability and therefore of the reactivity of acyl glucuronide.
The in vitro extent of covalent binding to HSA allowed the rank of acyl carboxylic drugs according to their reactivity potential. However, as the relation between covalent binding and toxicological effect is still unclear, we suggest extending the predictability of our model by integrating a second parameter, which is the instability of the acyl glucuronides. For this purpose, correlation between the amount of covalent binding observed and the instability of each acyl glucuronide was attempted.
First, we tried to reproduce the correlation described by Benet (Benet et al., 1993) between the moles of drug maximally bound irreversibly per mole of protein versus the degradation rate constant for the β-1-O-acyl glucuronide conjugates. The maximum covalent binding (normalized to protein content and expressed as the percentage of total acyl glucuronide present) observed was not correlated with the global degradation rate constant for acyl glucuronide isomers (Fig. 7A). Indeed, the global degradation rate constant of acyl glucuronide isomers was not a better parameter for reactivity prediction than the global acyl glucuronide degradation rate constant described by Benet. A correlation was also searched for between covalent binding and the hydrolysis rate. The aglycone appearance rate constant corresponded to the global acyl glucuronide hydrolysis rate. A satisfactory correlation was obtained for six of eight drugs (r2, 0.89). When zomepirac and tolmetin were taken into account, the correlation was less satisfactory (r2, 0.62) (Fig.7B). The hydrolysis rate constant was the same for both compounds, whereas the extent of covalent binding was higher for tolmetin. The rate of hydrolysis (i.e., the rate of aglycone appearance) does not discriminate sufficiently to explain the reactivity of all compounds. For most of the acyl glucuronides under study, isomerization was found to occur between the first and the second incubation. The extent of this phenomenon seemed to be more or less important for each acyl glucuronide (Table 1). The extent of this isomerization was certainly related to the instability of the 1-O-acyl glucuronide formed and its covalent binding capacity. Indeed, the formation of isomeric forms via acyl migration is a prerequisite for covalent binding to proteins by “imine” mechanism. Therefore, the aglycone appearance rate constant was weighted by the percentage of isomerization between the two incubation steps. An excellent correlation was then obtained between the maximal amount of bound drug, expressed as percentage of acyl glucuronide present in the incubation medium, and the aglycone appearance rate constant weighted by the percentage of isomerization (r2, 0.94) (Fig. 8). The results presented here showed that the extent of covalent binding could be predicted on the basis of acyl glucuronide hydrolysis rate combined with acyl migration propensity. The combination of these two parameters seemed to be more accurate for covalent binding prediction than the 1-O-acyl glucuronide degradation rate used by Benet. The correlation was still confirmed even when data from products like ibuprofen, suprofen, and diclofenac were added to the correlation. Thus, an in vitro reactivity scale was drawn up. The rank of the drugs tested was consistent with the literature. Tolmetin appeared as the most reactive, furosemide as the least. Ibuprofen and ketoprofen showed a similar low reactivity, whereas diclofenac, which has not been previously evaluated, showed a level of reactivity higher than zomepirac. In these conditions, suprofen appeared as a low reactive product. Smith showed that 0.75% of suprofen acyl glucuronide became covalently bound to HSA after 6 h of incubation (Smith and Liu, 1993). The 0.62% value found in this study was found in fairly good accordance with Smith's data.
Correlation between the extent of covalent binding (protein content normalized and expressed as the percentage of total acyl glucuronide present at the beginning of the reactivity phase) versus the degradation rate constant (h−1) (A) and the aglycone appearance rate constant (h−1) (B) for the in vitro incubation of various acyl glucuronide with HSA (0.5 mM).
The solid line represents correlation with eight drugs; the dotted line represents correlation with six drugs.
Correlation between the extent of covalent binding (protein content normalized and expressed as the percentage of total acyl glucuronide present at the beginning of the reactivity phase) versus the aglycone appearance rate constant weighted by the percentage of isomerization (between biosynthesis and reactivity phase) (h−1) during in vitro incubation of various acyl glucuronide with HSA (0.5 mM).
The structural relationship between acyl glucuronide degradation and covalent binding put forward by Benet (Benet et al., 1993) was also observed in this study. Acyl glucuronides of α-unsubstituted acetic acid derivatives such as tolmetin and zomepirac (Fig. 1) exhibited the highest covalent binding. Mono-α-substituted acetic acids (fenoprofen) showed intermediate levels of covalent binding. At last, fully substituted α-acetic acids, such as furosemide, led to the lowest irreversible binding. Additional compounds not tested by Benet complied with this structural relationship. Indeed, diclofenac, an α-unsubstituted acetic acid, demonstrated high levels of covalent binding, whereas ketoprofen, ibuprofen, and suprofen (mono-α-substituted acetic acids) showed intermediate levels of irreversible binding. This observation must be confirmed with more compounds and could be taken into account in the drug design process.
For the first time, a screening model allowing, in one experiment, the formation of acyl glucuronide metabolite by human microsomes and the assessment of its reactivity was presented. An excellent correlation (r2, 0.94) was found between the maximal amount of covalently bound drug (normalized to protein amount and expressed as percentage of total acyl glucuronide synthesized in the incubation medium) and the aglycone appearance rate constant weighted by the percentage of isomerization. This correlation represents an in vitro reactivity scale, which will help predict the covalent binding potential of new chemical entities.
Footnotes
-
This work has been presented in part at the 6th International Society for the Study of Xenobiotics (ISSX) meeting, in Munich on October 8–11, 2001.
- Abbreviations used are::
- HSA
- human serum albumin
- QCs
- quality controls..
- Received October 11, 2001.
- Accepted December 24, 2001.
- The American Society for Pharmacology and Experimental Therapeutics
