Abstract
Acyl-migrated isomers of drug β-1-O-acyl glucuronides have been implicated in drug toxicity because they can bind to proteins. The acyl migration and hydrolysis ofS-naproxen-β-1-O-acyl glucuronide (S-nap-g) was followed by dynamic stopped-flow HPLC-1H NMR and HPLC methods. Nine first order rate constants in the chemical equilibrium between six species (S-nap-g, its α/β-2-O-acyl, α/β-3-O-acyl, α/β-4-O-acyl, and α-1-O-acyl-migration isomers, andS-naproxen aglycone) were determined by HPLC-UV studies in 25 mM potassium phosphate buffer, pH 7.40, 25 mM potassium phosphate buffer in D2O pD 7.40, and 25 mM potassium phosphate buffer in D2O pD 7.40/MeCN 80:20 v/v (HPLC-1H NMR mobile phase). In the 25 mM potassium phosphate buffer (pH 7.40) the acyl-migration rate constants (h−1) were 0.18 (S-nap-g–α/β-2-O-acyl isomer), 0.23 (α/β-2-O-acyl–α-1-O-acyl), 2.6 (α-1-O-acyl–α/β-2-O-acyl), 0.12 (α/β-2-O-acyl–α/β-3-O-acyl), 0.048 (α/β-3-O-acyl–α/β-2-O-acyl), 0.059 (α/β-3-O-acyl–α/β-4-O-acyl), and 0.085 (α/β-4-O-acyl–α/β-3-O-acyl). The hydrolysis rate constants (h−1) were 0.025 (hydrolysis ofS-nap-g) and 0.0058 (hydrolysis of all acyl-migrated isomers). D2O and MeCN decreased the magnitude of all nine kinetic rate constants by up to 80%. The kinetic rate constants for the degradation of S-nap-g in the mobile phase used for HPLC-1H NMR determined using HPLC-UV could predict the results obtained by the dynamic stopped-flow HPLC-1H NMR experiments of the individual acyl-migrated isomers. It is therefore recommended that β-1-O-acyl glucuronide degradation kinetics be investigated by HPLC-UV methods once the identification and elution order of the isomers have been established by HPLC-1H NMR.
β-1-O-Acyl glucuronides are common phase II metabolites of drugs with a carboxylic acid functionality (Spahn-Langguth and Benet, 1992), including the widely used non steroidal anti-inflammatory drugS-naproxen (S)-6-methoxy-α-methyl-naphthalene acetic acid (Upton et al., 1980; Wilson and Ismail, 1986; Vree et al., 1992).
β-1-O-Acyl glucuronides have been shown to be reactive metabolites that will hydrolyze, acyl migrate, and bind covalently to proteins causing potential toxicity (Spahn-Langguth and Benet, 1992; Boelsterli et al., 1995). This toxicity may be caused by formation of immuno-reactive glucuronide protein adducts (Worrall and Dickinson, 1995), but toxicity mechanisms involving modification of active sites of enzymes (Terrier et al., 1999) and interaction with structural proteins (Bailey et al., 1998) has also been suggested. No common toxicity mechanism for β-1-O-acyl glucuronides has been established, as the β-1-O-acyl glucuronides of different drugs bind covalently to different proteins (Bailey and Dickinson, 1996) and on different sites on specific proteins (Qiu et al., 1998). The covalent binding to proteins proceeds predominantly via the acyl-migrated isomers (Dickinson and King, 1991; Ding et al., 1995;Liu et al., 1998; Qiu et al., 1998).
The stability and protein reactivity ofS-naproxen-β-1-O-acyl glucuronide (S-nap-g2; Fig.1) have been studied by several groups (Vree et al., 1992; Bischer et al., 1995; Iwaki et al., 1999). The generalized scheme of acyl migration is shown in Fig. 1, where the initial acyl-migration step from the β-1-O-acyl glucuronide to the β-2-O-acyl isomer (which can then mutarotate to the α-2-O-acyl isomer) is considered to be irreversible, whereas the rearrangement between the α/β-2-, α/β-3-, and α/β-4-O-acyl isomers is reversible (Spahn-Langguth and Benet, 1992).
Most investigators have mainly been concerned with the total degradation rate of the β-1-O-acyl glucuronide (Dickinson et al., 1994; Bischer et al., 1995; Castillo and Smith, 1995). The interest in the degradation of the β-1-O-acyl glucuronide was focused by a study in which the in vitro degradation rate gave a linear correlation with in vitro covalent protein binding for the nine different β-1-O-acyl glucuronides studied (Benet et al., 1993). The β-1-O-acyl glucuronides of S- andR-naproxen also fit this linear correlation (Bischer et al., 1995). However, as the covalent protein binding predominantly proceeds via the acyl-migrated isomers, the degradation rate of the β-1-O-acyl glucuronide per se may not be the most relevant bioactivity parameter. If the degradation is mainly due to hydrolysis, only small amounts of acyl-migrated isomers will be formed, and thus covalent binding to proteins will only be a minor reaction. It is consequently desirable to elucidate the complex kinetics of the acyl-migration rearrangement scheme (Sidelmann et al., 1996; Akira et al., 1998). A distinction between hydrolysis (kβ1-D) and acyl migration (kβ1-2) at the very least is necessary to evaluate the relative formation of acyl-migrated isomers compared with hydrolysis.
HPLC-1H NMR has been used to identify the individual isomers in mixtures of acyl-migrated 2-, 3-, and 4-fluorobenzoic acid glucuronides (Sidelmann et al., 1995a,b) andR- and S-phenylpropionic acid glucuronides (Akira et al., 2000). This procedure eliminates the need for preparative HPLC of the unstable isomers and subsequent off-line NMR spectroscopy to assign the chromatographic peaks in mixtures of acyl-migrated glucuronide isomers (Hansen-Møller et al., 1988; Bradow et al., 1989).
The interconversion kinetics of 2-fluorobenzoic acid glucuronide isomers was followed by HPLC-1H NMR in dynamic stopped-flow mode (Sidelmann et al., 1996). The chromatographic run was stopped with the peak(s) of interest in the NMR flow probe, and the degradation was followed over time. With this approach physical collection and purification of the individual isomers as done for the α/β-2-O-acyl isomer of S-nap-g (Iwaki et al., 1999) and the α/β-2-, α/β-3-, and α/β-4-O-acyl isomers of diflunisal-β-1-O-acyl glucuronide (Dickinson and King, 1991) is not necessary.
The reaction medium in dynamic HPLC-1H NMR studies is the HPLC mobile phase containing D2O to minimize solvent suppression artifacts in the1H NMR spectrum and acetonitrile for the chromatographic separation. These modifications compared with the normal in vitro conditions (pH 7.40 with no D2O or acetonitrile) must be considered when comparing kinetics obtained by other methods.
Recently the acyl-migrated isomers of S-nap-g were assigned in two chromatographic systems using stopped-flow HPLC-1H NMR in which the α-1-O-acyl isomer was determined (Mortensen et al., 2001). In the present paper HPLC based on UV-detection (HPLC-UV) and the previously determined elution order is applied to elucidate the interconversion kinetics between the acyl-migration isomers of S-nap-g, and the results are compared with dynamic stopped-flow HPLC-1H NMR methods. Also, the influence of D2O and increasing concentrations of acetonitrile is studied to clarify the effect on the kinetics.
Materials and Methods
All chemicals used were of analytical reagent grade or higher from commercial suppliers. Water was purified using a Milli-Q Plus water purification apparatus (Millipore A/S, Glostrup, Denmark). Deuterium oxide for HPLC-1H NMR was from Goss Scientific Instruments Ltd. (Essex, England). Acetonitrile for the HPLC-1H NMR experiments was NMR CHROMASOLV grade from Riedel de Haën (Sigma-Aldrich UK, Dorset, England). Biosynthetic S-nap-g was isolated from human urine using solid phase extraction and preparative HPLC as previously described (Mortensen et al., 2001).
Reaction Media.
The degradation kinetics of S-nap-g at 37°C was monitored by HPLC-UV in the following reaction media: 25 mM potassium phosphate buffer pH 7.40 (1); 25 mM potassium phosphate buffer in D2O pD 7.40 (2); and 25 mM potassium phosphate buffer in D2O pD 7.40/acetonitrile 80:20 v/v (3). Reaction medium 1 was prepared by adjusting a 25 mM KH2PO4 solution with 1 M KOH to pH 7.40. Reaction medium 2 was prepared by adjusting a 25 mM KH2PO4 solution in D2O with 1 M KOH in D2O to a pH meter reading of 7.40, using normal water-based calibration standards with pH 7.00 and 10.00 to calibrate the pH meter. Reaction medium 3 identical to the HPLC-1H NMR mobile phase was prepared by mixing reaction medium 2 with acetonitrile in the ratio 80:20 v/v. The effect of acetonitrile was studied in mixtures of reaction medium 1 and acetonitrile in the following ratios: 90:10 v/v, 80:20 v/v (identical to the mobile phase used for the degradation studies monitored by HPLC-UV), 70:30 v/v, 60:40 v/v, and 50:50 v/v.
Kinetic Experiments by HPLC-UV.
A HP 1100 series chromatographic system with HP ChemStation software was used (Agilent Technologies Denmark A/S, Birkerød, Denmark). This system was isocratic with a single-wavelength UV/visible-detector. The column was a Hibar LiChrospher 100 RP-C18 column with a 5-μm particle size and 250 mm × 4-mm i.d. (Merck, Darmstadt, Germany) operated at ambient temperature. The flow rate was 1 ml/min, and the detection wavelength was 272 nm. The mobile phase consisted of reaction medium 1 mixed with acetonitrile in a ratio of 80:20 v/v.
S-nap-g (0.7 mg) was dissolved in 1.0 ml of reaction medium and incubated at 37°C for 48 h. Aliquots of 50 μl were withdrawn at regular intervals and immediately stabilized by mixing with 100 μl of cold formic acid (5% v/v) to prevent acyl migration and hydrolysis. Stabilized samples were stored at 5°C for no more than 24 h before analysis by HPLC-UV. The amounts ofS-nap-g, the individual acyl-migrated isomers, andS-naproxen were unchanged for at least 48 h in the cold acidified samples as monitored by HPLC-UV. The samples were injected in the HPLC system using a Rheodyne injection valve fitted with a 20-μl loop (Rheodyne, Cotati, CA). S-Naproxen,S-nap-g, and its acyl-migrated isomers were assigned in the chromatograms as previously described (Mortensen et al., 2001). The area of the naproxen-related peaks was normalized to a total response of 100% before kinetic analysis.
Directly-Coupled 600-MHz HPLC-1H NMR Spectroscopy.
The LiChrospher 100 RP-C18 column was used, with the following hardware: a Bruker LC-22 pump (Bruker, Rheinstetten, Germany), a Bruker photodiode array detector (J & M Analytische Mess- und Regeltechnik GmbH, Aalen, Germany), a Bruker column oven-operated at 25°C, and a Bruker BPSU-36 flow control unit. The outlet of the detector was connected to the HPLC-1H NMR flow probe via an inert polyether-ether ketone capillary (3.1 m × 0.25-mm i.d.). HPLC-NMR-mass spectroscopy software (Hystar v1.1, Bruker) controlled the flow dynamics of the system and stored the chromatographic data.
The mobile phase buffer for HPLC-1H NMR was reaction medium 3 and was thus identical to the mobile phase used for HPLC-UV kinetic studies except for the substitution of H2O with D2O. This substitution did not affect the chromatographic elution order of the glucuronide isomers (Mortensen et al., 2001). S-nap-g (2.8 mg) was dissolved in 1.0 ml of reaction medium 1 and incubated at 37°C. The rearrangement reactions were terminated by the addition of 100 μl of cold 10% formic acid to stabilize the 1.0-ml sample. One sample stabilized at t = 6 h contained maximum amounts of the α/β-2-O-acyl isomer and was used for the dynamic stopped-flow HPLC-1H NMR degradation experiment with this isomer. A second sample stabilized att = 24 h contained maximum amounts of the α/β-3- and the α/β-4-O-acyl isomers and was used for the dynamic stopped-flow HPLC-1H NMR degradation experiments with these isomers. Samples were stored at 5°C for no more than 72 h before analysis by HPLC-1H NMR. The stabilized sample (100 μl) containing a total of 250 μg of glucuronide isomers was injected into the HPLC system for each stopped-flow experiment, and the flow was stopped on the peak of interest (Sidelmann et al., 1996).
NMR Spectroscopy.
The 1H NMR spectra were acquired using a Bruker AVANCE600 spectrometer operating at 600.13-MHz1H frequency equipped with a1H-13C inverse detection Z-gradient HPLC flow probe containing a 4-mm i.d., 120-μl cell.1H NMR spectra of the individual acyl-migrated glucuronide isomers were obtained in stopped-flow mode at 600.13 MHz and 37°C probe temperature. Dual solvent suppression of the acetonitrile and residual HDO signals was achieved using the standard one-dimensional nuclear Overhauser effect spectroscopy presaturation pulse sequence (Bruker) with relaxation and mixing delays of 2.0 and 0.1 s, respectively. Next, 1024 free induction decays were collected into 64 K computer data points with a spectral width of 20 ppm, corresponding to an acquisition time of 2.73 s. Each dynamic stopped-flow HPLC-1H NMR degradation experiment ran for 11 to 16 h with a time resolution of 90 min. Before Fourier transformation, an exponential apodization function was applied to the free induction decays, corresponding to a line broadening of 2 Hz. Chemical shifts were referenced to the acetonitrile signal at δ 2.0 and thus may differ slightly from signals referenced to HDO at δ 4.7.
The dynamic NMR degradation experiment of S-nap-g in 25 mM potassium phosphate buffer in D2O pD 7.40 was performed by injecting a 2.8-mg/ml solution directly into the flow-probe using the same NMR conditions as described above. Every hour for 7 h, 256 scans were acquired.
Isomer Concentration Determination by NMR.
Selected diagnostic NMR peaks from the individual glucuronide isomers were integrated relative to those of the aromatic region of the spectra. The aromatic region NMR peak integral was set to six hydrogens, and the amounts of the individual isomers were determined from the relative areas of their selected diagnostic peaks. The selected NMR signals with assignments are given in Table1. As the 2-, 3-, and 4-O-acyl isomers all exist as α/β-anomers in an equilibrium on the NMR time scale (Sidelmann et al., 1996), diagnostic signals for the individual α/β-anomers of each positional O-acyl isomer were integrated and then added to get the total concentration of eachO-acyl isomer. For S-naproxen aglycone and the α-1-O-acyl isomer a well resolved doublet from theS-naproxen methyl group at δ 1.5 was chosen as the diagnostic signal. Because a methyl group gives a three-proton signal, those integrals were divided by 3 to give the correct relative amount. The multiplet at δ 5.11 used for the α-3-O-acyl isomer quantitation is a superposition of the signals from the 1′ and 3′ protons, and the integrated signal is thus divided by 2 to give the correct relative amount. All other characteristic signals chosen corresponded to individual protons and thus needed no correction. Before kinetic analysis the responses were corrected to give the percentage of each compound.
Data Fitting.
Kinetic rate constants for the acyl-migration reactions were fitted for the individual experiments using the program Gepasi 3.21 (Mendes, 1993,1997; Mendes and Kell, 1998) using the kinetic scheme depicted in Fig.1 with first order mass-action reaction kinetics. The hydrolysis rate constants for the acyl-migrated isomers were fitted as one parameterkX-D (kX-D =kα1-D =k2-D = k3-D =k4-D). It is not possible to distinguish between those hydrolysis rate constants when the concentrations of the isomers are in the same order of magnitude because the product for all the reactions is the aglycone S-naproxen.
Results
The kinetic profiles of the degradation of S-nap-g in reaction media 1, 2, and 3 determined by HPLC-UV are given in Fig.2. Points are experimental data, and lines are the simulated curves using the fitted kinetic rate constants given in Table2. The effect of substituting the aqueous phosphate buffer (reaction medium 1, Fig. 2A) with phosphate buffer made up in D2O (reaction medium 2, Fig. 2B) was an overall decrease in reactivity, both for acyl migration (kβ1-2) as the maximum concentration of the α/β-2-O-acyl isomer is reached later and for hydrolysis (kβ1-D) as the formation rate of S-naproxen was decreased. Adding acetonitrile to the phosphate buffer made up in D2O in 80:20 (v/v) buffer/acetonitrile (reaction medium 3, Fig. 2C) further slowed down the rearrangement kinetics.
The kinetic rate constants for the individual experiments as determined by data fitting with the kinetic simulation program Gepasi 3.21 are shown in Table 2 for the acyl-migration rearrangement scheme in Fig. 1. The observed changes in reactivity shown in Fig. 2 are reflected in the fitted kinetic rate constants. The ratio am/h is the ratio between acyl migration and hydrolysis of S-nap-g. The pseudoequilibrium concentration ratios between the α-1-, α/β-2-, α/β-3-, and α/β-4-O-acyl isomers are calculated by the ratio of the reversible acyl-migration first order rate constants in the equilibrium.
The degradation kinetics of S-nap-g in reaction medium 2 determined by dynamic 1H NMR is shown in Fig. 3A. The degradation kinetic profiles of the individual acyl-migrated isomers in the HPLC-1H NMR mobile phase, reaction medium 3, determined by dynamic stopped-flow NMR are shown in Fig. 3, B (α/β-2-O-acyl isomer), C (α/β-3-O-acyl isomer), and D (α/β-4-O-acyl isomer). In Fig. 3, A, B, C, and D individual data points are experimental data from the dynamic NMR experiments while the lines are the predicted values using the kinetics derived from the degradation experiments as monitored by HPLC-UV (Fig. 2, B and C).
The degradation of S-nap-g in reaction medium 1 mixed with increasing ratios of acetonitrile was also monitored by HPLC-UV. The fitted first order kinetic rate constants for hydrolysis (kβ1-D) and acyl migration (kβ1-2) of the β-1-O-acyl glucuronide are given in Table3. As the ratio of phosphate buffer/acetonitrile in reaction medium 1 was varied from 100:0 to 50:50 (v/v) the magnitude of the rate constants for hydrolysis and acyl migration decreased by equal amounts, and thus no effect on the ratio between acyl migration and hydrolysis of S-nap-g (ratio am/h in Table 3) was observed. All nine first order kinetic rate constants in the overall migration scheme (Fig. 1) were decreased by acetonitrile as well (data not shown).
Discussion
The Effects of D2O and Acetonitrile on the Degradation Kinetics of S-nap-g.
The possible kinetic effect of acetonitrile and D2O in the HPLC-1H NMR mobile phase should be considered when performing dynamic stopped-flow HPLC-1H NMR kinetic analysis of the degradation of β-1-O-acyl glucuronides (Sidelmann et al., 1996). Because the addition of acetonitrile as well as the substitution of H2O with D2O is a necessary prerequisite for the stopped-flow HPLC-1H NMR experiment, the study of those modifications is not possible by HPLC-1H NMR. Thus, in the present degradation study HPLC-UV experiments were used to monitor the effects of those modifications of the reaction medium on the degradation kinetics ofS-nap-g. Table 2 shows that all rate constants are slowed down by D2O as well as 20% (v/v) acetonitrile. Table 3 shows the effect of increasing the acetonitrile content in reaction medium 1. Acetonitrile is seen to decrease the hydrolysis as well as the acyl-migration reaction rates. In addition, the ratio between the major acyl-migration reaction and the minor hydrolysis reaction is unchanged on addition of 0 to 50% acetonitrile (v/v). Thus, acetonitrile slows down the reaction rates but does not affect the ratio between the two parallel reactions.
Comparison of HPLC-UV and Dynamic NMR Methods in the Study of Acyl-Migration Kinetics.
In the present study of reversible kinetics the initial concentrations of the involved compounds in the NMR-probe are not important because the rate constants in the overall kinetic scheme are the same irrespective of which compound is the starting material. Thus, it is possible by kinetic analysis to derive the rate constants for the overall scheme by incubating the biosynthetic β-1-O-acyl glucuronide; therefore, dynamic stopped-flow HPLC-1H NMR experiments of the individual isomers are not necessary since no new kinetic information is obtained with these experiments.
Although dynamic stopped-flow HPLC-1H NMR is not necessary to elucidate the interconversion kinetics ofS-nap-g isomers, kinetic analysis by HPLC-UV would not be possible without the prior assignment of the chromatographic peaks by stopped-flow HPLC-1H NMR, and so the two approaches are complementary.
As shown in Fig. 3A, the kinetics determined from the HPLC-UV degradation experiment in reaction medium 2, phosphate buffer made up in D2O, predicts the degradation ofS-nap-g in the same reaction medium when determined by dynamic 1H NMR, as only the formation of the 4-O-acyl isomer is slightly underestimated by the model. This underestimation is probably attributable to the poor signal/noise ratio of the NMR spectrometer at low concentrations.
This experiment shows that the NMR magnetic field has no effect on the degradation kinetics. A comparison of the kinetics obtained by HPLC-UV and those obtained by dynamic stopped-flow HPLC-1H NMR using the same degradation medium should then be possible.
Figure 3, B, C, and D shows that the rate constants determined from the HPLC-UV degradation experiment with reaction medium 3, the HPLC-1H NMR mobile phase, can predict the kinetics when the individual acyl-migrated isomers are stopped in the flow probe and the degradation is studied. The lack of fit forS-naproxen in the degradation experiment of the 4-O-acyl isomer in Fig. 3D may be due to the hydrolysis rate constant k4-D being slightly overestimated by the common rate constant kX-D determined in the fitting procedure. The lack of fit for the 2-O-acyl isomer in the same experiment is probably attributable to the poor signal/noise ratio of the NMR spectrometer at low concentrations.
The magnitude of the individual kinetic rate constants in the overall acyl-migration rearrangement scheme is noteworthy. In all cases (Tables2 and 3) the initial acyl migration (kβ1-2) of the β-1-O-acyl glucuronide is faster than its hydrolysis (kβ1-D) by a factor of 6 to 16. This indicates for S-nap-g that the acyl migration is the dominating degradation reaction and is favored over hydrolysis under the experimental conditions investigated. Similar results showing that acyl migration of S-nap-g is favored over hydrolysis have previously been presented (Iwaki et al., 1999). The pseudoequilibrium between the individual 2-, 3-, and 4-O-acyl isomers favors the 3-O-acyl isomer as the pseudoequilibrium ratios in Table 2 indicate. This corresponds with a dynamic 13C NMR study ofS-ketoprofen-β-1-O-acyl glucuronide (Akira et al., 1998) in which the 3-O-acyl isomer was found to be the most stable. However, for the acyl-migrated isomers of (2-fluorobenzoyl)-d-glucopyranuronic acid the 4-O-acyl isomer was found to be the most stable (Sidelmann et al., 1996), so this is not a general trend.
The overall stability of the individual acyl-migrated isomers of β-1-O-acyl glucuronides probably reflects their reactivity toward proteins, as the most unstable acyl-migrated isomers will be the most reactive. However, the major value of the kinetic modeling described in the present study is the ability to determine the acyl-migration rate constants (kβ1-2) and the hydrolysis rate constants (kβ1-D) of the β-1-O-acyl glucuronides themselves and thus predict the potential protein binding.
Kinetic Results Regarding the Formation of the α-1-O-Acyl Isomer.
The α-1-O-acyl isomer was found to be the least stable species in the overall reaction scheme, as the acyl-migration rate from the α-1-O-acyl isomer to the α/β-2-O-acyl isomer (kα1-2) was the highest in the overall equilibrium irrespective of the reaction medium (Table2). The rapid equilibrium between the α-1-O-acyl isomer and the α/β-2-O-acyl isomer can also be observed in Fig.2A, in which the concentration of the α-1-O-acyl isomer closely follows the α/β-2-O-acyl isomer concentration, whereas the concentration of the α/β-3-O-acyl isomer lags behind.
The α-1-O-acyl isomer was formed in all reaction media, and its formation must be considered a general mechanism in the rearrangement scheme of S-nap-g regardless of the incubation conditions.
A dynamic stopped-flow HPLC-1H NMR experiment with the reactive α-1-O-acyl isomer as starting material was attempted. It was so reactive that the first time-point, which was an average over 24 min (256 scans), after initial temperature-equilibration time of about 10 min, consisted of more than 50% α/β-2-O-acyl isomer. This confirmed that the α-1-O-acyl-isomer migrates to the α/β-2-O-acyl isomer and not back to the β-1-O-acyl glucuronide, thus corroborating the overall reaction scheme in Fig. 1.
The good fit of the experimental data to the kinetic model (Σs2 in Table 2) does not in itself confirm the rearrangement scheme in Fig. 1, as a good mathematical fit (small Σs2) of a model to experimental data does not necessarily guarantee that the model actually describes the chemical relationship between the species modeled. However, the good fit combined with the observation that only α/β-2-O-acyl isomer was formed from the α-1-O-acyl isomer as well as the rapid equilibrium between these two species as observed in Fig. 2A shows that the reaction scheme depicted in Fig. 1 initially based on mechanistic considerations must be correct.
Dynamic stopped-flow HPLC-1H NMR experiments (Sidelmann et al., 1996) demonstrated that the α-anomers of the 2-, 3-, and 4-O-acyl isomers of (2-fluorobenzoyl)-d-glucopyranuronic acid were less stable than the corresponding β-anomers. Acyl-migration rates were faster between the α-anomers than between the β-anomers, and also the equilibrium between the individual α/β-anomers favored the β-anomer. In the present study there was no direct chemical equilibrium between the β-1-O-acyl glucuronide and the α-1-O-acyl isomer, as the α-1-O-acyl isomer was in an equilibrium with the α-2-O-acyl isomer, which is formed via the β-2-O-acyl isomer arising from the β-1-O-acyl glucuronide.
A direct anomerization reaction between the β-1-O-acyl glucuronide and the α-1-O-acyl isomer is not possible because the ring opening necessary for anomerization requires rearrangement of the 1′ hydroxy group to its aldehyde form, which is only possible for a free alcohol. However, it is clear that the α-1-O-acyl isomer is less stable than the β-1-O-acyl glucuronide in terms of acyl migration (kα1-2 is consistently higher than kβ1-2 in all reaction media) and thus follows the pattern from the α/β-2-, α/β-3-, and α/β-4-O-acyl isomers. Thus, the α-1-O-acyl isomer may be a significant species in terms of bioreactivity and toxicity, even though the concentration is low compared with the other acyl-migrated isomers. However, reactions between α-1-O-acyl glucuronide isomers and protein models have not been studied to date.
Although the initial acyl migration from the β-1-O-acyl glucuronide to the β-2-O-acyl isomer is widely thought to be irreversible or insignificant (Spahn-Langguth and Benet, 1992), Fig.2A shows that the concentration of S-nap-g att = 24 and 48 h does not follow the fitted irreversible first order degradation kinetics. As the concentration ofS-nap-g was higher than predicted by the model at these late time points, it is possible that an equilibrium exists between the β-1-O-acyl glucuronide and its β-2-O-acyl isomer. One study by HPLC indicated that diflunisal-β-1-O-acyl glucuronide exists in an equilibrium with the corresponding β-2-O-acyl isomer (Hansen-Møller et al., 1988), as it was re-formed from a mixture of the acyl-migrated isomers. Given that the α-1-O-acyl glucuronide exists in a rapid equilibrium favoring the α-2-O-acyl isomer as shown in the present study, the β-1-O-acyl glucuronide may exist in a similar equilibrium with the equilibrium strongly favoring the β-2-O-acyl isomer also. The back-formation of β-1-O-acyl glucuronide is then so minor that it has not been observed in the majority of studies of acyl-glucuronide rearrangement kinetics.
Footnotes
-
Send reprint requests to: Rasmus W. Mortensen, Pharmaceutical Stability Testing, Leo Pharmaceutical Products, 55, Industriparken, DK-2750 Ballerup, Denmark. E-mail:rasmus.mortensen{at}leo-pharma.com
-
↵1 Current address: Pharmaceutical Stability Testing, Leo Pharmaceutical Products, 55, Industriparken, DK-2750 Ballerup, Denmark.
-
This work was supported by the European Union Biomed 2 program “Hyphenated Analytical Techniques”, Grant BMH4-CT97–2533 (DG 12-SSMI).
- Abbreviations used are::
- S-nap-g
- S-naproxen-β-1-O-acyl glucuronide
- HPLC
- high-performance liquid chromatography
- pD
- minus the log10of the deuterium ion concentration
- Received September 28, 2000.
- Accepted December 14, 2000.
- The American Society for Pharmacology and Experimental Therapeutics