Variation in glucuronidation of lamotrigine in human liver microsomes

Lamotrigine (LTG), a diaminotriazine anti-epileptic, is principally metabolized at the 2-position of the triazine ring to form a quaternary ammonium glucuronide (LTGG) by uridine glucuronosyl transferease (UGT) 1A3 and UGT1A4. It has been hypothesized that glucuronidation of anti-epileptic drugs is spared with age, despite a known decrease in liver mass, based on older studies with benzodiazepines such as lorazepam. To examine this, the formation rates of LTGG formation were measured by liquid chromatography-mass spectrometry (LC-MS) in a bank of human liver microsomes (HLMs) obtained from younger and elderly donors at therapeutic concentrations. The formation rate of LTGG was not significantly different in HLMs obtained from younger and elderly subjects. A four- to five-fold variation for the formation of LTGG was observed within each microsomal bank obtained from elderly and younger donors, and the range of LTGG formation was observed to be 0.15–0.78 nmoles min⁻¹ mg⁻¹ of protein across the entire set of HLMs (n = 36, elderly and younger HLMs). UGT1A4 and UGT1A3 catalysed the formation of LTGG with an intrinsic clearances of 0.28 and 0.02 μl min⁻¹ mg⁻¹ protein, respectively. UGT2B7 and UGT2B4 showed no measurable activity. No correlation was observed across the HLM bank for glucuronidation of LTG and valproic acid (a substrate for multiple UGT isoforms including UGT1A4).

Keywords: Uridine glucuronosyl transfereases (UGTs); glucuronidation; lamotrigine; anti-epileptic; valproic acid

Epilepsy is a group of neurological disorders that leads to abrupt, unpredictable, and recurrent seizures or convulsions. 3,5-Diamino-6-[2,3-dichlorophenyl]-1,2,4 - triazine (LTG) (Lamictal®) is an anti-epileptic drug that was approved in 1994. It has become increasingly popular for the treatment of secondary generalized seizures and complex partial seizures ([21]) and is also being used more in pregnant women with epilepsy due to its lower risk of birth defects compared with the older standard agents. It is also approved for maintenance treatment of bipolar disorder. LTG is thought to inhibit sodium ion channels, preventing glutamate release and blocking the rapid firing of depolarized neurons ([7]). LTG is principally metabolized to a 2N-glucuronide (LTGG), a conjugation product that was first characterized in the authors' laboratory ([27]; [31]). LTG glucuronidation is subject to significant interspecies variation ([7]). In rats, LTG is oxidized by a CYP450-mediated pathway, followed by conjugation of the oxidative intermediate. In rat bile, a unstable glutathione conjugate has also been identified and is presumably formed from a reactive arene-oxide metabolite on the dichlorophenyl ring ([22]). In dogs, LTG is excreted primarily as a 5N-glucuronide conjugate along with a 5N-methylated product. A 2N-methylated metabolite has also been identified in dogs. In humans, however, LTGG is the major metabolite and about 90% of the administered dose of LTG is excreted into urine in this manner. A 5N-glucuronide of LTG is also formed to a less significant extent ([9]; [7]). Metabolism of LTG is illustrated in Figure 1. Glucuronidation of LTG has been reported to be carried out by UGT1A3 and UGT1A4 ([14], [15]; [13]; [32]).

Graph: Figure 1. Metabolism of LTG. In humans, LTG is primarily metabolized to its 2N-glucuronide (90% of the administered dose), which is excreted in the urine. UGT1A4 (expressed in the liver) and UGT1A3 (expressed in the liver and small intestine) catalyse this conjugation reaction, with UGT1A3 contributing to a lesser extent.

Graph: Figure 2. Comparative velocities of LTG glucuronidation in HLMs. The velocities of LTG glucuronidation were measured at two different substrate concentrations: 0.16 mM { bars} and 1.25 mM { bars}, which were in the linear range for LTGG formation. The mean velocities of triplicate measurements ± standard deviation (nmoles min−1 mg−1 of protein) at both substrate concentrations are depicted: (A) elderly HLMs (n = 18); and (B) younger HLMs (n = 18).

Graph: Figure 3. Michaelis–Menten kinetics for LTGG formation. Kinetic profiles are shown for LTGG production in pooled HLMs, pooled UGT1A3 and UGT1A4 Superomes®. Velocity (y-axis) is in nmol min−1 mg−1 of protein; and substrate concentration (x-axis) is in mM. UGT1A3 and UGT1A4 have been implicated to play a role in LTG glucuronidation. No activity was observed in UGT2B4 and UGT2B7 Supersomes. Data are the mean of triplicate measurements ± standard deviation (nmoles min−1 mg−1 of protein).

Graph: Figure 4. Correlation plot for metabolism of VPA (1 mM) and LTG (1.25 mM) in HLM bank (n = 36). The x-axis indicates LTGG formation in nmoles min−1 mg−1; and the y-axis indicates VPAG formation in μmoles min−1 mg−1 as a mean of triplicate measurements. Substrate concentrations for correlation plot with VPA and LTG were 1 mM and 1.25 mM, respectively.

UDP-glucuronosyltransferases or UGTs are the enzymes responsible for the conjugation of glucuronic acid moiety to various functional groups ([11]). Inside a cell, UDP-glucuronosyltransferases (UDPGTs or UGTs) are localized in the endoplasmic reticulum with their active site facing the lumen. Like the cytochrome P450 enzymes, UGTs make up a superfamily of enzymes that is comprised of three distinct gene sub-families: UGT1A (chromosome 2q37), UGT2A (chromosome 4q13) and UGT2B (chromosome 4q13). The UGT1A sub-family in humans consists of several enzymes that share an identical C-terminus (containing the UDPGA [Uridine diphosoglucuronic acid] binding pocket), encoded by four conserved exons. Twelve variable region exons known as cassette exons encoding the N-terminus region (containing the substrate-binding site) are spliced to the constant region exons to form each individual mRNA ([5]). It has been reported that the clearance of anti-epileptic drugs by glucuronidation remains unaffected with progression of age. Clearance of benzodiazepines such as oxazepam, temazepam and lorazepam was not substantially altered in young–old subjects (65–84 years of age) ([17], [19]; [8]). It has been hypothesized that glucuronidation is generally spared in the elderly population, despite a known decrease in liver mass ([19]). In elderly subjects, liver mass and liver blood flow have been documented to undergo age-related changes ([35]). However, these factors alone fail to elucidate the alteration of drug clearance ([10]). It was our objective to study the rates of in vitro glucuronidation of LTG in liver microsomes obtained from elderly and younger human donors. Furthermore, we aimed to gather evidence toward correlation of LTG and valproic acid (VPA) glucuronidation in vitro because recent studies show that both substrates might be metabolized by common UGT enzymes.

The objective of this study was to evaluate whether in vitro metabolism of LTG is different in elderly versus younger subjects, to obtain in vitro correlation for LTG-VPA glucuronidation in the HLM bank, and to determine the kinetic profile of LTG glucuronidation with UGT1A3 and UGT1A4. To study the variability in glucuronidation of LTG as a factor of age, in vitro HLMs in the bank were divided into two groups containing 18 HLMs each based on the age of the donor, that is, livers from young donors (2–56 years of age) and liver from elderly donors (65 years onwards). These HLMs were screened for their glucuronidation activity toward LTG. To ascertain the involvement of other UGTs in the metabolism of LTG, we carried out detailed kinetic studies with cloned expressed UGTs (Superomes®). Moreover, to compare the glucuronidation of VPA and LTG in vitro (both have been known to be metabolized by common UGT enzymes; [15]; [32]; [28]; [4]) the activities were correlated across the entire HLM bank (n = 36). The variability in LTG glucuronidation velocities at concentrations near the Km concentrations is reported in the present investigation.

LTG reference standard was obtained from Toronto Research Chemicals (North York, Ontario, Canada). LTGG standard was kindly provided by Glaxo (GSK). High-performance liquid chromatography (HPLC)-grade formic acid was obtained from Mallinckrodt Baker, Inc. (Phillipsburg, NJ, USA). HPLC-grade ammonium acetate, magnesium chloride, ethyl acetate, potassium chloride, dibasic and monobasic potassium phosphate, glycerol and methanol were purchased from Fischer Scientific Co. (Fair Lawn, NJ, USA), whereas glacial acetic acid, Trizma buffer, 1,4-saccharolactone, UDPGA, tetrasodium pyrophosphate, and DMSO were purchased from Sigma Aldrich Co. (St Louis, MO, USA). High-purity acetonitrile for MS was obtained from Honeywell Burdick & Jackson (Muskegon, MI, USA). Alamethicin was purchased from Fluka GmbH (Buchs, Switzerland). The bicinchonic acid (BCA) protein assay kit was purchased from Pierce Biotechnology, Inc. (Rockford, IL, USA). HLMs other than those prepared in our laboratory were bought from Cellz Direct, Inc. (Durham, NC, USA).

Most HLMs were prepared in the laboratory by further modification of standard reported methods ([33]; [23]; [20]). In brief, the frozen liver tissue was broken into chunks of 5–10 g. The weighed liver tissues were added to 3 × (v/w) of 1.15% KCl in 0.01 M potassium phosphate buffer (pH 7.4). The liver tissue samples were thawed and homogenized (in a graduated cylinder) with Polytron Homogenizer (Brinkman, Westbury, NJ, USA) in 30-s bursts on ice. The tubes were spun at 9,500g for 20 min, at 4°C (8,500 rpm) in a Beckman JL2 centrifuge. Supernatants were transferred into ultracentrifuge tubes and weighed for balancing. The tubes were then centrifuged at 100,000g for 65 min at 4°C (L-90K Beckman ultracentrifuge; 35,000 rpm). The microsomal pellet from the bottom of the tube was carefully separated from the surrounding pasty glycogen, pooled and then resuspended in 1 M sodium pyrophosphate buffer (pH 7.4). This suspension was further homogenized on ice in a Potter-Elvehjem homogenizer (three strokes) and centrifuged a second time at 100,000g for 65 min at 4°C. The pellet from this spin was resuspended in 0.1 M phosphate buffer containing 1 mM EDTA and 20% glycerol (one-third of the cytosolic volume) and homogenized on ice, aliquotted into labelled cryovials, and stored at −80°C. Protein quantitation was performed by a bicinchonic acid dye method (BCA assay; Pierce).

Before setting the specifications for in vitro incubations with LTG, linearity studies with respect to time and protein were performed. Linearity for formation of LTGG with protein was determined over a range of protein concentration (0.1–1.0 mg ml−1) and time (10–80 min) at 1.25 and 0.16 mM substrate, respectively (approximately equal to or lower than the Km), in pooled HLMs. Based on these studies, the microsomal incubations were carried out with a protein concentration of 0.5 mg ml−1 for 30 min. Glucuronidation incubations for each LTG concentration were carried out in triplicate, with some modification of the procedures developed earlier in this laboratory ([30]; [6]). Briefly, microsomes were pre-incubated with the 'buffer mix' containing Trizma buffer (0.1 M, pH 7.4 at 37°C), 5 mM MgCl2, 5 mM saccharolactone and alamethicin (approximately 100 μg mg−1 of protein) on ice for 15 min. Varying concentrations of LTG were then added. Stock solutions of LTG were prepared by dissolving LTG in DMSO (and 0.1 M acetic acid with the help of sonication) such that the final organic concentration in the incubation mixture did not exceed 1%. Addition of LTG solution prepared by the above procedure to the incubation mixture led to decrease in the pH of the incubation medium to 7.0. However, control experiments demonstrated that the glucuronidation activity in HLMs was not affected at the lower pH. The above incubation mixture was pre-incubated at 37°C for 2 min. The reaction was started by addition of freshly prepared 3 mM UDPGA. Incubations without UDPGA were employed as negative control. The final incubation volume of 200 μl was incubated for 30 min at 37°C at 30 rpm in a shaking water bath (Model BS-11, Lab Companion, Seoul, Korea). An equal volume of chilled acetonitrile was used to stop the reaction, followed by centrifugation at 13,000 rpm for 5 min. The supernatant was the centrifuged through 0.2 μ nylon spin filters and carried forward for LC-MS analysis. In vitro incubations with cloned expressed UGTs (Superomes®) were conducted at a protein concentration of 0.125 mg ml−1. The other incubation conditions for Superomes® and HLMs were identical. In vitro incubation conditions for carrying out VPA glucuronidation were similar to those described above with two exceptions. The microsomal protein concentration used for VPA studies was 0.29 mg ml−1 and the substrate concentration was 1 mM. This protocol has been described in detail by [4].

LTGG quantitation was carried out by a method developed previously in this laboratory ([6]; [29]). A brief summary of the assay is as follows: quantitation of the parent ion for LTGG ([M]+; m/z 432.2) was optimal at the following MS conditions. Spray voltage: 1.9 kV and detector voltage: 1.5 kV under ESI mode, CDL capillary temperature: 250°C, interface temperature: 250°C, heater block temperature: 200°C, and nebulizing gas flow rate: 0.18l min−1. Study samples were assayed for the presence of LTGG with the above instrumental settings. The analytical separation was carried out on a Higgins Analytical Haisil C-18 column ( 100 mm × 2. 1 mm, 5 μ) at room temperature. The gradient was comprised of 20–50% acetonitrile in 20 mM ammonium acetate buffer, pH 6.67, over a period of 8 min. The percentage of organic was ramped up to 80% over the next 2 min and held for a min, followed by a 4-min re-equilibration step. The flow rate for the analysis was 0.3 ml min−1 and the LTGG peak, which eluted at 2.2 min, was monitored by ultraviolet light at 254 nm as well. Data acquisition was achieved on LC-MS Solution Software® from Shimadzu, Inc. Standard curves for LTGG (added to blank pig liver microsomes) were developed with concentrations of 1 μM to 1 mM and showed an R2 ≥ 0.999. The coefficient of variation was < 10% at all concentrations of LTGG.

The LC-MS/MS method for detection and quantitation of VPAG has been described previously ([4]). In short, the formation of the parent ion VPAG ([M – 1]−; m/z 319.2) was optimal at the spray voltage of −3.25 kV and collision energy of 33 eV. The parent ion was fragmented with argon as the collision gas to yield a major daughter ion of m/z 143.2. Other conditions such as a capillary temperature of −110°C, a collision pressure of 1.0 mTorr, and a sheath gas pressure of 39 ml min−1 were maintained. The separation of VPAG was carried out on a Beta-Basic C-18 column ( 150 mm ×  1 mm, 3 μ, Thermo-Finnigan) at room temperature. A gradient comprising of 30–50% acetonitrile in 20 mM ammonium acetate buffer, pH 3.77, over a period of 6 min was used to separate VPA from VPAG. The concentration of acetonitrile was held at 50% acetonitrile for 4 min more, followed by a 4-min re-equilibration step. The flow rate for the analysis was 10 μl min−1. Data acquisition was achieved on Xcalibur Software® from Thermo-Finnigan Corp. Standard curves for VPAG were developed with VPAG concentrations of 3 μM to 3 mM and showed an R2 ≥ 0.999.

Kinetic analysis for LTGG formation in pooled HLMs and UGT Superomes® was performed with Sigma Plot (Systat Software, Inc., Richmond, CA, USA). The data from LC-MS analysis (HLMs and Superomes®) were best fit to a standard Michaelis–Menten kinetics model, as determined by analysis of residuals. Statistical tests to compare the variation in the formation of LTGG across the microsomal bank (Kolgomorov and Smirnov distribution test, Bartlett's test and Student's t-test) were carried out. The resultant rates of formation were employed to study the correlation in formation of LTGG and VPAG.

A bank of 18 HLMs obtained from elderly donors (greater than 65 years of age) was screened for glucuronidation activity at 0.16 and 1.25 mM substrate concentrations. Approximately four- to five-fold variation for formation of LTGG was observed within the microsomal bank. In the elderly HLM bank, the range for the velocity of LTGG formation extended from 0.15 to 0.72 nmoles min−1 mg−1 of protein at 1.25 mM substrate concentration. The average velocity at the same substrate concentration was 0.40 ± 0.03 nmoles min−1 mg−1 of protein, whereas that at 0.16 mM substrate concentration was 0.13 ± 0.04 nmoles min−1 mg−1 of protein, respectively. Similarly, within the bank of 18 HLMs obtained from younger donors (2–56 years of age) approximately four- to five-fold variation for the formation of LTGG was observed. In the younger HLM bank, the range for the velocity of LTGG formation extended from 0.16 to 0.78 nmoles min−1 mg−1 of protein at 1.25 mM substrate concentration. The average velocity at the same substrate concentration was 0.45 ± 0.08 nmoles min−1 mg−1 of protein, whereas that at 0.16 mM substrate concentration was 0.14 ± 0.03 nmoles min−1 mg−1 of protein (Figure 2).

Intrinsic clearances (Vmax/Km) for cloned expressed UGTs (Superomes®) – UGT1A3 and UGT1A4 – were determined (Figure 3). UGT2B7 and UGT2B4 were also screened for activity toward LTG glucuronidation. UGT1A4 catalysed the formation of LTGG with an intrinsic clearance of 0.28 μl min−1 mg−1 protein. UGT1A3 had approximately a ten-fold lower intrinsic clearance – 0.02 μl min−1 mg−1 protein – as compared with UGT1A4. UGT2B7 and UGT2B4 demonstrated no activity for LTGG formation, even at much lower substrate concentrations (10–100 μM). Rates of LTG glucuronidation as detected in Superomes® are represented in Table 1. Data from kinetic experiments in randomly chosen younger and elderly HLMs as well as pooled HLMs (six concentrations, each in triplicate) were fit to a Michaelis–Menten kinetic model. The resultant range of Vmax was 0.5–1.1 nmol min−1 mg−1 protein and that of Km was 0.7–5.6 mM across the chosen samples. The mean Vmax in pooled HLMs was 1.1 nmol min−1 mg−1 protein and the mean Km was 1.2 mM. No significant differences in intrinsic clearances (Vmax/Km) were observed in HLMs from elderly and younger population.

Table 1.  Intrinsic clearances for lamotrigine-2-N-glucuronide (LTGG) formation in uridine glucuronosyl transfereases (UGT) Superomes®.

2 In vitro kinetic experiments for glucuronidation of lamotrigine (LTG) performed in Supersomes revealed a maximal velocity (measured as the mean of three replicates) as illustrated. The data were fitted to Michaelis–Menten kinetics and intrinsic clearance values as shown were obtained. No activity was observed for UGT1B4 and UGT2B7 Supersomes.

A correlation plot was obtained for VPAG and LTGG formation in in vitro incubations in HLMs (n = 36). The r2 value was 0.0009, and no significant correlation was observed. This could be attributed to the involvement of multiple UGTs: UGT1A3, UGT1A4, UGT1A6, UGT1A8, UGT1A10 and UGT2B7 in glucuronidation of VPA as opposed to only UGT1A4 and UGT1A3 for LTG glucuronidation (Figure 4).

LTGG, a quaternary ammonium glucuronide is positively charged and generally elutes out at the solvent front in reversed-phase HPLC conditions. Previously, LTGG was quantitated in our laboratory by the use of sodium dodecyl sulfate as an ion-pairing agent ([30]). However, application of LC-MS as a detection technique prevented the use of sodium dodecyl sulfate as an ion pair. Experiments to increase the retention time of LTGG with volatile ion-pairing agents such as heptafluorobutyric acid and trifluoroacetic acid resulted in variability in retention time and/or ion suppression as well as a loss of assay precision. By employing a higher concentration of acetic acid–ammonium acetate buffer (20 mM, pH 6.67), we were able to obtain enough retention to separate out LTGG from the solvent front. The LC conditions resulted in a retention time of LTGG of 2.2 min. The percentage of organic (100% acetonitrile) was further ramped up to elute unreacted LTG from the incubation mixture. Mass spectrometer conditions such as capillary temperature, heater block temperature, spray voltage, detector voltage, etc. as discussed in the Materials and Methods section, were optimized to obtain maximum sensitivity for the quantitation of LTGG. LC-MS analysis involved a SIM of LTGG (m/z 432.2) in the positive mode by means of an ESI probe.

There was a four- to five-fold intra-bank variability in LTGG formation in vitro in HLMs. The rates of LTGG formation in HLMs from younger and elderly donors were not significantly different. It must be noted that the rates of LTG glucuronidation in the HLMs were calculated at two substrate concentrations bracketing the Km. These findings support the hypothesis that the intrinsic clearance of LTG is spared in elderly subjects, assuming that the general quality and enzyme stability are similar between the two banks of HLMs. It is known that liver mass and hepatic blood flow decrease in elderly subjects ([35]). Collectively these data indicate that the total in vivo clearance of antiepileptic drugs (AEDs) may decrease in the elderly population ([17], [16], [18], [19]; [8]; [10]). It must also be noted that the microsomes that constituted the HLM bank were obtained from a variety of sources and were prepared by different methods and stored in different buffer solutions. It is unclear at this point how these factors might affect the functional activity of HLMs. Full kinetic experiments were carried out in some younger and elderly HLMs as well as pooled HLMs to obtain a measure of LTGG formation in vitro. Data collected from the in vitro incubations were best fit to a Michaelis–Menten kinetic model. The velocity versus time profile afforded a range of Vmax (0.5–1.1 nmol min−1 mg−1 protein) and Km (0.7–5.6 mM) values across the samples. Similar studies in pooled HLMs yielded a mean Vmax of 1.1 nmol min−1 mg−1 protein and a mean Km of 1.2 mM. No significant differences in intrinsic clearances (Vmax/Km) were observed in HLMs from elderly and younger population. However, the results suggest that the intrinsic metabolism of AEDs is unaffected in elderly liver as compared with the younger subjects. An age-related depletion of liver mass would suggest that LTG clearance in vivo may decline with age. Unfortunately, these results could not be normalized to the level of UGT1A4 protein (a lack of a selective antibody) or UGT1A4 mRNA present in each sample (flash frozen tissue not available for all livers).

Pacifici and co-workers have compared the variability in glucuronidation across HLM banks for various UGT substrates. In the standard glucuronidation assay conditions employed by Pacifici and co-workers, mycophenolic acid and resveratrol had approximately five-fold variation across a HLM bank consisting of 50 and 10 HLMs, respectively ([34]). Zidovudine glucuronidation afforded ten-fold variability in bank of 20 HLMs ([25]). Similarly, ethinyl oestradiol and testosterone yielded approximately 20-fold variability in a bank of more than 100 HLMs each ([24]). The present investigation has shown that LTG glucuronidation resulted in about four- to five-fold variability in the HLM bank (n = 36, in the HLM bank not subdivided by age). However, each of these substrates is metabolized by multiple UGTs and the relative contribution of the different UGT isozymes for each substrate in vivo is not yet clear. It is unclear how the expression and polymorphisms of these enzymes might cumulatively affect the in vitro variability in glucuronidation.

It is understood that in vivo pharmacokinetics of LTG are significantly altered by co-administration of VPA. LTG metabolism was significantly inhibited by co- administration with VPA, resulting in an increase in half-life from 25 h ([26]) to 50–70 h ([1]), whereas co-medication with inducing AEDs such as carbamazepine, phenobarbital, or phenytoin resulted in an average half-life of approximately 12 h ([26]). Our laboratory previously found that VPA is converted to its acyl glucuronide in vitro by UGT1A4, UGT1A8 and UGT1A10 at lower velocities as compared with UGT2B7 ([2]; Argikar & Remmel 2009[4]) in addition to previously known catalysis by UGT1A3, UGT1A9 and UGT2B7 (data not shown). Based on inhibition experiments with zidovudine (AZT) and VPA, Miners and co-workers recently suggested UGT2B7 may be a low-capacity, low Km isoform for LTG glucuronidation ([28]). Miners and co-workers reported a Hill component due to UGT2B7 and a Michaelis–Menten component due to UGT1A4, for LTG glucuronidation in HLM incubations containing fatty acid-free albumin. In the present study kinetics of LTG glucuronidation were evaluated in four Superomes® – UGT1A3 UGT1A4, UGT2B4 and UGT2B7 – to evaluate the isozymes responsible for LTG glucuronidation and to determine their contribution to the LTG-VPA interaction. Intrinsic clearances (Vmax/Km) values for UGT1A4 and UGT1A3 for lamotrigine are reported in Table 1. Cloned, expressed UGT2B7 and UGT2B4 did not show any activity toward LTG glucuronidation ([2]). We could not confirm LTG glucuronidation by cloned expressed UGT2B7 isozyme even at low LTG concentrations (10–100 μM) by our sensitive LC-MS assay. The Ki values for inhibition of LTG glucuronidation by VPA, as reported by Miners and co-workers, were significantly different with (approximately 0.3 mM) or without bovine serum albumin (BSA) (2.5 mM). We did not use BSA in the experiments and it is unclear how this might affect LTG glucuronidation in vivo. It is hypothesized that long-chain free fatty acids are released during microsomal incubations. The fatty acids are substrates of several UGTs and result in competitive inhibition and altered Km values. The results with cloned, expressed enzymes indicate that UGT2B7 does not catalyse LTG glucuronidation in vitro. Recently, UGT2B10 has been shown to catalyse the N-glucuronidation of nicotine and tobacco nitrosamines, and it is not known if this enzyme contributes to LTG metabolism. Nevertheless, based on the literature reports and the results from the present investigation, UGT1A3 and UGT1A4 might be the only UGTs contributing toward the VPA-LTG interaction ([14]; [13]; [12]; [2]). We hypothesize that co-administration of VPA and LTG in vitro could lead to competition for the common UGT enzymes – UGT1A3 and UGT1A4 – resulting in reduction in clearance and increase in the half-life of LTG. Further studies will be needed to determine if other enzymes such as UGT2B10 contribute to the interaction. Additional information on the in vivo expression of UGTs through the use of quantitative proteomics, the concentrations of the drugs at the enzyme active site and free fraction of the drug in the liver will be needed to estimate the inhibition of in vivo clearance from in vitro data in HLMs or with cloned, expressed enzymes. As expected, there was no correlation between VPAG and LTGG formation rates from in vitro incubations in HLMs (n = 36) and thus failed to explain the LTG-VPA in vivo interaction. Involvement of multiple UGTs – UGT1A3, UGT1A4, UGT1A6, especially UGT2B7 (low affinity, high velocity) – in VPA glucuronidation ([3]) as opposed to only UGT1A4 and UGT1A3 for LTG glucuronidation may have lead to such an observation in vitro.

Approximately four- to five-fold intra-bank variation was observed for glucuronidation of LTG within the bank of elderly as well as younger HLMs. No significant inter-bank difference was observed between elderly and younger HLM banks. These results support the hypothesis that metabolism of LTG is similar in elderly and younger HLMs in vitro. However, a known decrease in liver mass and blood flow could lead to a reduction in in vivo clearance of AEDs in the elderly population. These results must be considered along with the information from in vivo studies to comprehend the overall consequence of aging on LTG metabolism. UGT1A3 and UGT1A4 turn over LTG and we have previously reported the glucuronidation activity of these UGT isozymes toward VPA ([3]). In vitro correlation studies failed to establish an association between LTG and VPA glucuronidation.

We thank Dr Timothy S Tracy for his guidance in preparation of the HLMs. We also thank Glaxo Inc. (GSK) for providing the LTGG reference standard. The assistance of Ms Bisi Abdul in carrying out microsomal incubations and that of Ms Falguni Gadkari in preparation of this manuscript is greatly appreciated.

Declaration of interest: The authors thank the National Institute of Health (NIH Grant NINDS P50 NS16308) for providing the financial support for this study.