The classic hyperbolic Michaelis-Menten equation has long been used to characterize the in vitro kinetics of enzyme-catalyzed reactions. More recently, cytochrome P450 (P450)-catalyzed enzymatic reactions that exhibit sigmoidal kinetics have been described in the literature. Notable examples of this type of kinetic behavior include P450 3A4-catalyzed oxidation of testosterone (Ueng et al., 1997) and carbamazepine (Korzekwa et al., 1998). Rate versus substrate concentration profiles for sigmoidal kinetics have a characteristic initial lag at lower concentrations and result in characteristic curved Eadie-Hofstee plots. Such kinetic behavior is often attributed to autoactivation of enzymes that occurs with increasing substrate concentrations and is often described empirically using the Hill equation (Houston and Kenworthy, 2000). Models involving the binding of multiple substrate molecules to the enzyme active site have been proposed to provide a mechanistic description of autoactivation kinetics (Ueng et al., 1997; Korzekwa et al., 1998)

More recently, examples of similar kinetic phenomena have been reported for UDP-glucuronosyltransferases (UGTs). The formation of estradiol-3-glucuronide, a UGT1A1-selective reaction, was observed to have a better fit to the Hill equation in both human liver microsomes and recombinant UGT1A1 (Fisher et al., 2000; Soars et al., 2003). Similar autoactivation kinetics has also been observed for the formation of the glucuronic acid conjugate of acetaminophen in human liver microsomes and in recombinant UGT1A1 and UGT1A9 (Fischer et al., 2000; Court et al., 2001). The mechanism of autoactivation kinetics observed for UGTs is currently unknown.

Valproic acid (2-propylpentanoic acid, VPA) is a broad-spectrum anticonvulsant with a unique branched-chain fatty acid structure (Davis et al., 1994). Previous studies in sheep demonstrated that VPA was primarily eliminated in this species via glucuronidation with ∼70 to 80% of an administered intravenous dose recovered in urine as VPA glucuronide (VPAG) (Wong et al., 2001). The appearance of VPAG in sheep urine appeared to be formation rate-limited rather than excretion rate-limited. Based upon these properties, we were able to assess the in vivo apparent V_max and K_m of VPA glucuronidation in sheep using either the disappearance of the parent compound from plasma or the excretion rate of VPAG in urine (Wong et al., 2001). In our preliminary studies examining VPAG formation in sheep liver microsomes, characteristics of autoactivation kinetics were observed (S. Kumar and H. Wong, unpublished results). As mentioned, sigmoidal kinetics has been previously observed in vitro for both P450- and UGT-catalyzed reactions. However, the in vivo relevance of sigmoidal/autoactivation kinetics has never been clearly demonstrated. Valproic acid is unique in that high doses of the compound can be administered such that saturation of its elimination (glucuronidation) occurs in vivo, thus allowing for full characterization of kinetic parameters for its metabolism in vivo. This ability to characterize VPA glucuronidation kinetics in vivo provided a unique opportunity for us to investigate whether the autoactivation kinetics for VPA glucuronidation can be observed in vitro in liver microsomes and, if so, whether this would translate to the in vivo situation. Thus, the objective of the studies described in this manuscript was to investigate the occurrence of sigmoidal/autoactivation kinetics for VPA glucuronidation in vitro and in vivo in sheep.

Materials and Methods

In Vivo Animal Experiments. Five Dorset Suffolk cross-bred ewes, with a body weight of 61.9 ± 7.3 kg (mean ± S.D.) were surgically prepared with a minimum of 3 days before experimentation. Polyvinyl catheters (Dow Corning, Midland, MI) were implanted in a femoral artery and vein (catheter i.d. 1.02 mm and o.d. 2.16 mm) as described by Kumar et al. (1999). Briefly, animals were administered an i.v. bolus of VPA (sodium valproate; Sigma-Aldrich, St. Louis, MO) at a dose of 100 mg/kg b.wt. over 1 min via the femoral vein. Blood samples (∼3 ml) were collected via the femoral artery at 5, 15, 30, and 45 min and 1, 2, 4, 6, 9, 12, 15, 24, 36, 48, 60, and 72 h after drug administration. Blood samples were placed into heparinized Vacutainer tubes (Becton-Dickinson, Rutherford, NJ) and centrifuged at 2000g for 10 min. The plasma supernatant was harvested and placed into clean borosilicate test tubes with polytetrafluoroethylene-lined caps. Cumulative urine samples were collected via a Foley bladder catheter for the following time intervals: 0 to 2, 2 to 4, 4 to 6, 6 to 9, 9 to 12, 12 to 15, 15 to 24, 24 to 36, 36 to 48, 48 to 60, and 60 to 72 h postdose. Plasma and urine samples were stored frozen at –20°C until the time of analysis. All studies were approved by the University of British Columbia Animal Care Committee, and the procedures performed on sheep conformed to the guidelines of the Canadian Council on Animal Care.

Unbound plasma concentrations of VPA were determined ex vivo in all adult sheep plasma samples using an ultrafiltration procedure as described by Wong et al. (2001). Concentrations of VPA and its metabolites in plasma ultrafiltrate and urine were measured simultaneously using an established gas chromatographic-mass spectrometric analytical method (Yu et al., 1995). The variability and bias of all analytes measured using this analytical method were determined to be <15% in earlier assay validation studies (Yu et al., 1995). VPA and metabolite calibration and quality control standards as well as control (blank) biological fluid samples were run with each batch of study samples. Concentrations of the VPAG in urine were determined as described by Wong et al. (2001).

In Vitro Valproic Acid Glucuronidation Kinetics. The glucuronidation kinetics of valproic acid were examined in pooled sheep liver microsomes (pool of 10) (XenoTech LLC, Lenexa, KS). The kinetics of VPA glucuronidation was determined under the following incubation conditions: VPA (5–640 μM), 0.5 mg/ml microsomal protein, 2.5 mM MgCl₂, 5 mM saccharolactone, alamethicin (10 μg/mg protein), and 3 mM UDP-glucuronic acid (Sigma-Aldrich) in 100 mM phosphate buffer (pH 7.5). The total incubation volume was 200 μl. All contents of each incubation, with the exception of UDP-glucuronic acid, were preincubated for 5 min at 37°C. After the preincubation period, reactions were initiated by the addition of UDP-glucuronic acid. Reactions were terminated at 20 min by the addition of 200 μl of ice-cold acetonitrile containing the internal standard (diclofenac at 2 μM final concentration; Sigma-Aldrich) followed by thorough vortex mixing. The resulting samples were centrifuged at 2000g for 10 min at 4°C, and 10 μlofthe supernatant was injected into a liquid chromatograph coupled with a tandem mass spectrometer for quantitation of VPAG. All reactions were performed in triplicate. In preliminary experiments, reactions performed under the described conditions were linear with respect to both microsomal protein concentration and incubation time (data not shown).

VPAG was quantitated using a modification of a liquid chromatographytandem mass spectrometry method described previously by Tong et al. (2005). The instrument consisted of a Shimadzu LC-10 ADvp liquid chromatograph (Shimadzu Corporation, Columbia, MD) interfaced with a Finnigan TSQ Quantum triple quadrupole mass spectrometer (Thermo Electron Corporation, San Jose, CA). Samples were injected by a Shimadzu SIL-HTc autosampler (4°C) onto a Waters Symmetry C₁₈ column (50 mm × 2 mm i.d.; Waters, Milford, MA). The mobile phase consisted of (A) 95% water and 5% acetonitrile with 5 mM ammonium acetate and (B) 95% acetonitrile and 5% water with 5 mM ammonium acetate. The liquid chromatograph pumps were programmed to pump 25% B from 0 to 1 min, followed by an increase to 90% B from 1 to 2 min, a hold at 90% B from 2 to 4 min, and a return to 25% B for column reequilibration. The high-performance liquid chromatography flow rate was 0.2 ml/min, and the run time was 6 min. The mass spectrometer was operated in negative ion mode with multiple reaction monitoring (VPAG: 319.1 > 143.2, internal standard: 293.9 > 250.0) at a collision energy of 17 eV and mass spectrometer dwell time of 0.2 s. Concentrations of VPAG were determined using a calibration curve prepared with authentic standard. Calibration curves (0.39–50 μM) for all assays performed were linear with r² values >0.99 and coefficients of variation <20% at the limit of quantification and <15% at all other concentrations.

Determination of Unbound Valproic Acid in Sheep Liver Microsomes. The unbound fraction of VPA in pooled sheep liver microsomes under in vitro incubation conditions was determined by ultracentrifugation using [¹⁴C]VPA (valproic acid, [carboxyl-¹⁴C], sodium salt; American Radiolabeled Chemicals, St. Louis, MO). Experiments were performed at 5 and 700 μM VPA to span the full range of concentrations used in the in vitro studies examining VPA glucuronidation kinetics. The determination of the VPA unbound fraction in sheep liver microsomes was performed under the following incubation conditions: VPA (5 and 700 μM), 0.5 mg/ml microsomal protein, 2.5 mM MgCl₂, alamethicin (10 μg/mg microsomal protein), and 5 mM saccharolactone in 100 mM phosphate buffer (pH 7.4). The total incubation volume was 4 ml. After an incubation period of 30 min at 37°C, 1-ml aliquots of each sample were transferred in triplicate to centrifuge tubes (11 × 34 mm polycarbonate tubes; Beckman Coulter, Fullerton, CA), and samples were centrifuged at 100,000 rpm for 3 h using an Optima TLX ultracentrifuge (Beckman Coulter). Radioactivity content was determined in the supernatant from the samples after centrifugation and in the original microsomal incubation before centrifugation using a Tri-Carb 3100 TR liquid scintillation analyzer (PerkinElmer Life and Analytical Sciences, Boston, MA). The fraction unbound was calculated as the ratio of the average radioactivity measured in the supernatant layers (unbound fraction) to that in the initial microsomal incubate before centrifugation.

Pharmacokinetic and Enzyme Kinetic Analyses. In vivo estimates of apparent Hill (V_max^app, S₅₀^app, and n) parameters for overall VPA elimination from plasma were obtained through fitting of individual unbound plasma concentration-time profiles using SAAM II V1.2 (The SAAM Institute Inc., Seattle, WA). Unbound plasma concentration-time profiles were fit to a one- or two-compartment model with elimination described by either the Michaelis-Menten (eq. 1) or Hill equation (eq. 2), where v is the rate of elimination, V_max^app is the apparent maximum rate of elimination, C is the substrate concentration, K_m^app/S₅₀^app is the apparent substrate concentration resulting in an elimination rate equal to 50% of V_max^app, and n is the Hill coefficient. A two-compartment model with elimination described by the Hill equation provided the best “fit” of the data from adult animals compared with a simpler one-compartment model with similar elimination characteristics. Model selection was based upon a lower Aikake's information criterion (Wagner, 1993). Mean estimates for parameters are presented as a mean ± S.D.

In vivo estimates of kinetic parameters for VPA glucuronidation from urinary excretion data were determined by fitting v (urinary excretion rate of the glucuronide metabolite) versus C_mid (unbound VPA plasma concentration at the midpoint of the urine collection interval) data to the Hill equation [v = C_midⁿ × V_max^app/(S₅₀^{app n} + C_midⁿ)]. C_mid, v, and n are as defined above, V_max^app is the apparent maximal formation rate of VPAG, and S₅₀^app is the substrate concentration resulting in 50% of V_max^app. The fitting to the Hill equation rather than to the classic Michaelis-Menten equation was determined on the basis of the distinct shape of the Eadie-Hofstee plots and by minimization of the sum of squares of residuals and the standard error of parameter estimates when data were fitted to the Hill equation. Datasets were individually fit for each animal using GraphPad Prism V4.02 (GraphPad Software Inc., San Diego, CA).

Rates of VPAG formation versus VPA incubation concentrations from in vitro enzyme kinetic experiments using pooled sheep liver microsomes were fit to the Hill equation using GraphPad Prism V4.02 as described above for the in vivo urine data. The VPA incubation concentration substituted for C_mid when data from the in vitro experiment were fitted.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Valproic acid plasma concentration-time profiles fitted to a two-compartment model with Hill equation elimination for animals A to E.

Simulations. Simulations of the two-compartment model with elimination described by the Hill or Michaelis-Menten equation were performed using SAAM II V1.2. Values of V_max^app (3.09 μmol/min/kg), K_m^app, or S₅₀^app (208 μM) used in simulations were calculated from estimates previously reported by Wong et al. (2001). Volume of distribution (V_d) (0.272 l/kg) and rate constants from the peripheral to the central compartment (k₂₁: 1.90 1/h) and from the central compartment to the peripheral compartment (k₁₂: 1.40 1/h) in a two-compartment model were mean estimates obtained from animals in the current study. Dose and n were arbitrarily assigned values of 100 mg/kg and 2.00, respectively, to illustrate differences between elimination governed by the Michaelis-Menten and Hill equations.

Results

Urinary Recovery of VPA Dose. The recovery of VPA and its metabolites in urine after a 100 mg/kg i.v. dose is presented in Table 1. Essentially, the entire VPA dose was recovered in urine from all five animals. VPAG was the main metabolite found in urine, accounting for approximately 80% of the entire dose. Following the glucuronide metabolite, unchanged VPA accounted for the second largest portion of the administered dose recovered in urine (∼7%). All other metabolites combined, on average, accounted for <10% of the administered dose (Table 1).

In Vivo Estimation of V_max^app, S₅₀^app, and n of Overall VPA Elimination from Plasma. Unbound plasma concentration-time profiles were fitted to a two-compartment model with elimination from the central compartment governed by the Hill equation. Unbound concentration-time profiles from individual animals along with their model-predicted plasma profiles are presented in Fig. 1. The in vivo estimates of apparent V_max^app, S₅₀^app, and n of overall VPA elimination from plasma are presented in Table 2. These resulting parameters are hybrid constants that are largely reflective of overall metabolic elimination. These estimates are comparable to estimates generated using VPAG urinary excretion data (Table 2) consistent with glucuronidation being the primary path of VPA elimination in sheep. An attempt was made to fit the unbound plasma concentration-time profiles to a one- or two-compartment model with elimination governed by the Michaelis-Menten equation. However, parameter estimates could not be obtained as these models failed to converge.

In Vivo Estimation of Apparent V_max^app, S₅₀^app, and n of VPA Glucuronidation from Urinary Excretion Data. A condition for estimating the apparent kinetic parameters of VPA glucuronidation using urine data is that the appearance of VPAG in urine is formation rate-limited, as opposed to elimination rate-limited (Gibaldi and Perrier, 1982). Metabolites demonstrating formation rate-limited urinary excretion exhibit plasma concentration versus time profiles that decline in parallel to those of the parent compound (Houston, 1986). In previous investigations with VPA using pregnant sheep, we attempted to measure the plasma concentrations of the VPAG by subjecting the samples to hydrolysis and subtracting the concentrations of the unconjugated VPA. However, it was evident from these attempts that the plasma concentrations of the VPA glucuronide conjugate were substantially lower than those of the parent compound and could not be reliably measured by our difference-based analytical approach (S. Kumar, unpublished results). This conclusion is consistent with the likely formation rate-limited urinary excretion of the VPA glucuronide. Furthermore, in the absence of a plasma profile, a plot of urinary excretion rate of VPAG versus t_mid (i.e., time at the midpoint of the urine collection interval) can be used to estimate the slope of the terminal decline/half-life of the VPAG in plasma (Gibaldi and Perrier, 1982). Figure 2 is a plot of mean urinary excretion rate of VPAG versus t_mid, together with a mean unbound VPA plasma concentration-time profile. The parallel decline of these two plots is also consistent with the urinary excretion of VPAG being formation rate-limited. This finding is consistent with our previous observation of formation rate-limited urinary excretion of VPAG at i.v. doses of 250 mg/kg (Wong et al., 2001).

Eadie-Hostee plots of v versus v/C_mid for individual animals are shown in Fig. 3. The plots from all animals were characteristically curved, consistent with autoactivation kinetics described by the Hill equation. Plots of v versus C_mid from individual animals (Fig. 3) were fitted to the Hill equation and the resulting estimates of V_max^app, S₅₀^app, and n of VPA glucuronidation are shown in Table 2.

Enzyme Kinetics of VPA Glucuronidation in Sheep Liver Microsomes.Figure 4 shows a plot of v versus the VPA incubation concentration and the corresponding Eadie-Hofstee plot in pooled sheep liver microsomes. The Eadie-Hofstee plot possessed a distinctive “hooked” profile characteristic of autoactivation kinetics. Estimates of V_max^app, S₅₀^app, and n generated from fitting the data to the Hill equation are presented in Fig. 4. The in vitro estimates of S₅₀^app and n were comparable with those estimated in vivo (Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Unbound VPA plasma-concentration time profile together with plot of VPAG urinary excretion rate versus t_mid.

Unbound fractions of VPA in sheep liver microsomes at VPA concentrations of 5 and 700 μM were 93 and 98%, respectively, suggesting negligible nonspecific binding in microsomes over the VPA concentration range used in our in vitro studies (5–640 μM). Hence, no correction was made for the binding of VPA to liver microsomal protein.

Discussion

The in vitro kinetics of enzyme-catalyzed reactions is commonly characterized by the hyperbolic Michaelis-Menten equation. Sigmoidal or autoactivation kinetics are distinguished from Michaelis-Menten kinetics by rate-substrate concentration profiles with initial lags at lower concentrations and result in characteristic curved Eadie-Hofstee plots. Sigmoidal kinetics is commonly described empirically using the Hill equation (Houston and Kenworthy, 2000). The exact mechanistic explanation for various aspects of sigmoidal kinetics has yet to be fully elucidated. Early observations of sigmoidal kinetics for xenobiotic metabolizing enzymes were mostly in human liver microsomal and cDNA-expressed systems (Ueng et al., 1997; Korzekwa et al., 1998). Later work using dextromethorphan by Witherow and Houston (1999) demonstrated that sigmoidal kinetics could also be observed in hepatocytes. More recently, a similar kinetic phenomenon has been reported for UGT in both human liver microsomes and recombinant systems (Fischer et al., 2000; Court et al., 2001; Soars et al., 2003). However, the in vivo relevance of autoactivation kinetics has never been clearly demonstrated.

Valproic acid presents a unique tool to examine the relevance of sigmoidal or autoactivation kinetics in vivo. Studies using rat hepatocytes over a wide concentration range (100 nM–1.8 mM) indicate that the uptake of VPA into hepatocytes is linear and rapid and does not involve carrier-mediated transport (Booth et al., 1996). We have demonstrated that the excretion of VPAG in urine is formation rate-limited rather then excretion rate-limited in the current study and in a previous study (Wong et al., 2001). On the basis of these two properties, the in vivo excretion kinetics of VPAG reflects enzymatic processes involved in its formation rather than transport and/or distribution phenomena. In a previous study, VPAG urinary excretion data from multiple in vivo experiments were pooled to generate a single rate versus substrate concentration plot (VPAG urinary excretion rate versus C_mid) (Wong et al., 2001). This profile was fitted to the Michaelis-Menten equation to provide in vivo estimates of apparent V_max and K_m for VPA glucuronidation. The sigmoidal features of rate-substrate concentration plots in our previous study were probably masked as a result of the data pooling process. Eadie-Hofstee plots from individual sheep are presented for the current study (Fig. 3) and clearly display the “curved” shape that is characteristic of sigmoidal or autoactivation kinetics.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Plots of urinary excretion rate of valproic acid-glucuronide (V) versus unbound valproic acid plasma concentration (C_mid) and corresponding Eadie-Hofstee plots for animal A (A and B), animal B (C and D), animal C (E and F), animal D (G and H), and animal E (I and J). Fit to the Hill equation is indicated by the dashed lines.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Rate of valproic acid-glucuronide formation versus valproic acid incubation concentration (A) and corresponding Eadie-Hofstee plot (B) in pooled (n = 10) sheep liver microsomes. Fit to the Hill equation is indicated by the dashed line. Kinetic parameter estimates are presented as estimate ± S.E. V is defined as the rate of valproic acid-glucuronide formation and S is defined as substrate (VPA) concentration in the figure.

Because VPA is almost entirely eliminated via glucuronidation (Table 1), the disappearance of the unbound parent compound from plasma should be reflective of kinetic processes governing glucuronidation. VPA is a low clearance drug in sheep such that its total body clearance approximates f_u (the unbound fraction of VPA) × Cl_int (intrinsic clearance) (Wilkinson and Shand, 1975). Thus, the clearance of unbound VPA is equal to its Cl_int. Cl_int is described by eq. 3 for Michaelis-Menten kinetics and eq. 4 for Hill kinetics: For compounds exhibiting Michaelis-Menten kinetics, Cl_int is linear and at its maximal value at concentrations that are ≪K_m, where Cl_int approximates V_max/K_m. Cl_int decreases as substrate concentrations approach and exceed K_m. In contrast, the Cl_int of compounds exhibiting autoactivation kinetics are not at their maximum at low substrate concentrations (≪S₅₀). With increasing concentrations, the value of Cl_int increases to a maximum, followed by an eventual decrease due to saturation of metabolic enzymes (Houston and Kenworthy, 2000). Figure 5 is a simulation to illustrate differences between elimination from a two-compartment model governed by the Michaelis-Menten and Hill equations. The distribution phase for both simulation profiles occurs between 0 and 2 h postdose. From ∼2 to 5 h, the shape of the unbound plasma concentration-time profile is slightly convex in both profiles, characteristic of saturation of elimination processes. The difference between the two profiles becomes more obvious at lower concentrations after 6 h. Elimination governed by the Michaelis-Menten equation results in a steeper terminal slope (i.e., shorter terminal t_1/2) in the unbound plasma concentration-time profile because the Cl_int goes from being saturated at higher VPA concentrations to becoming first order and at its maximum at lower concentrations. In contrast, elimination governed by autoactivation kinetics results in a shallower terminal slope (i.e., longer terminal t_1/2) because Cl_int is not at its maximum at low substrate concentrations. The degree of difference between the 2 equations is dependent on the value of n, providing all other parameters are kept constant. The unbound plasma concentration-time profiles presented in Fig. 1 are similar in shape to the simulation performed for the Hill equation displayed in Fig. 5. As VPA is eliminated almost entirely by glucuronidation in sheep, V_max^app, S₅₀^app, and n estimates from fitting unbound plasma concentration-time profiles were comparable with estimates derived from the VPAG urinary excretion data (Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Simulation of two-compartment model with elimination described by either the Michaelis-Menten or Hill equation. The following values were used for this simulation: dose (100 mg/kg), V_d (0.272 l/kg), k₂₁ (1.90 1/h), k₁₂ (1.40 1/h), V_max^app (3.09 μmol/min/kg), K_m^app or S₅₀^app (208 μM), and n (2.00).

The described differences in shape of the unbound plasma concentration-time profile provide an explanation for our failed attempt to characterize the unbound VPA plasma-concentration time profiles using Michaelis-Menten elimination. Previously, we had assessed in vivo apparent V_max and K_m values for VPA overall elimination using unbound plasma-concentration time profiles in sheep. Characterization of the nonlinear elimination of VPA using Michaelis-Menten kinetics required the simultaneous fitting of unbound plasma concentration-time profiles from three doses (50, 100, and 250 mg/kg) (Wong et al., 2001) in contrast to the single dose data (100 mg/kg) being fitted in the current study. The doses used in the previous study spanned a range over which metabolic saturation was obvious from the observed decreases in unbound VPA clearance with increases in dose (Wong et al., 2001).

Our in vitro experiment in pooled sheep liver microsomes demonstrates that the glucuronidation of valproic acid is characterized by sigmoidal or autoactivation kinetics and provides an explanation for our in vivo observations. To our knowledge, this is the first report of in vitro VPA glucuronidation exhibiting autoactivation kinetics. Sigmoidal kinetics that are observed in vitro may result as a consequence of in vitro incubation conditions under which substrate depletion occurs due to nonspecific binding to an incubation matrix or an overabundance of enzyme. At the lowest concentration of VPA tested in our study (5 μM), ∼2% of the substrate was converted to the glucuronide conjugate. In addition, nonspecific VPA microsomal binding appeared to be negligible in the concentration range over which VPA glucuronidation kinetics was assessed. Based upon these conditions, our observations of sigmoidal kinetics in vitro are probably not an experimental artifact and are related to characteristic enzyme-substrate interactions.

Reports of enhanced rates of glucuronidation in humans at high doses of VPA (1000 mg) have been previously described in the literature (Granneman et al., 1984). These enhanced rates could not be accounted for entirely by increases in unbound concentrations of VPA that occurred with increasing dose. The reported enhancement of VPA glucuronidation is consistent with autoactivation kinetic behavior and our in vivo observations in sheep. The in vivo kinetics of VPA glucuronidation would be more difficult to fully characterize in humans as doses normally administered are far less than those used in our current study.

To our knowledge, the current study is the first clear demonstration of the occurrence of autoactivation kinetics in vivo in any species. Because the uptake of VPA into hepatocytes has been shown to be rapid and linear over a wide concentration range (Booth et al., 1996) and the excretion of VPAG in sheep urine is formation rate-limited, our observations of sigmoidal kinetics in vivo are most likely reflective of the enzymatic processes involved in VPA glucuronidation. In agreement with this hypothesis, our in vitro studies examining VPA glucuronidation using sheep liver microsomes clearly displayed the features of autoactivation kinetics.