Experimental Procedures

Materials.

The free bases (R)-(+)- and (S)-(−)-ketotifen were donated by Novartis Pharma AG (Basel, Switzerland) and amitriptyline hydrochloride was donated by Bayer Vital (Leverkusen, Germany). A mixture of ketotifen N-glucuronide isomers was prepared as described (Mey et al., 1999), and amitriptylineN-glucuronide was purified from patient urine (Breyer-Pfaff et al., 1990). UDP-glucuronate was purchased from Roche (Mannheim, Germany); HPLC grade acetonitrile was obtained from Merck (Darmstadt, Germany). The establishment of a clonal HK293 cell line stably expressing UGT1A4 protein was reported previously (Green et al., 1995). These cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 10 mM HEPES, and 50 μg/ml of streptomycin. Addition of geneticin at 700 μg/ml in the first passage and 300 μg/ml in the second passage served for eliminating nontransfected cells. Cell cultures were grown at 37°C in a humidified atmosphere containing 5% CO₂. After four passages, confluent cultures were washed twice with cold saline and harvested with 1.5 ml/plate of 0.25 M sucrose containing 10 mM Tris-HCL (pH 7.4). After centrifugation for 6 min at 5000 rpm and 4°C, the sediment was suspended in 0.2 ml/plate of the same buffer and homogenized by 12 strokes by hand in a Dounce homogenizer; aliquots were stored at −80°C. HK293 cells transiently expressing UGT1A3 were grown as described previously (Green et al., 1998), and microsomes were prepared by a modification of the procedure ofBattaglia et al. (1994). Harvested cells were rinsed twice with PBS, suspended in 0.25 M sucrose containing 5 mM HEPES pH 7.4 and 0.5 mM dithiothreitol, and homogenized for 10 s using a tissue tearor. After sonication for 10 s, the homogenate was diluted 10-fold with the same buffer without dithiothreitol and homogenized again for 10 s. It was centrifuged for 20 min at 9000g, and microsomes were sedimented from the supernatant by centrifuging for 1 h at 100,000g. V79 cells stably expressing UGT1A9 (Fournel-Gigleux et al., 1990) were grown and harvested under the same conditions as the HK293 cells expressing UGT1A4 except for the use of 100 U/ml of penicillin and 100 μg/ml of streptomycin as antibiotics in the medium. Microsomes were prepared from human liver samples as described (Mey et al., 1999). Protein was measured (Lowry et al., 1951) using BSA as standard.

UGT Assays.

Cell homogenates containing UGT1A9 conjugated the standard substrate 4-methylumbelliferone (Lilienblum et al., 1982) at a rate of 2 to 3 nmol/min/mg of protein in accordance with the literature (Ebner and Burchell, 1993). Ketotifen N-glucuronidation in cell membranes expressing UGT1A3 or in homogenates of cells expressing UGT1A4 or UGT1A9 was measured in 1 ml of the medium used previously in experiments with human liver microsomes (57 mM Tris-HCl pH 8.0, 5 mM MgCl₂, and 2 mM UDP-glucuronate). The incubation time was 25 to 40 min at 37°C and the protein concentration was 1 to 1.5 mg/ml in experiments with UGT1A4 and UGT1A9 and 0.7 mg/ml with UGT1A3, such that N-glucuronide formation was linear with respect to time and protein concentration. Ten different concentrations of (R)- and (S)-ketotifen between 1 and 300 μM were incubated in duplicate. Conjugation of amitriptyline (1–500 μM) was studied in human liver microsomes at a protein concentration of 0.5 mg/ml for 30 min. For kinetic measurements with UGT1A4, 1 to 400 μM amitriptyline was incubated for 40 min with 0.9 mg/ml of cell homogenate protein. Reactions were stopped by substrate extraction with tert-butyl methyl ether, and the aqueous phase was prepared for HPLC analysis by extraction withn-hexane and adjustment to 0.2 N HClO₄(Mey et al., 1999).

HPLC Analysis.

The isomers of (R)- and (S)-ketotifenN-glucuronide were separated and quantified by the procedure described earlier (Mey et al., 1999). AmitriptylineN-glucuronide was measured in the same system consisting of a 4.6 × 20-mm clean-up column with C₁₈silica gel and a 4.6 × 250-mm Prodigy 5 μm ODS (3) 100 Å analytical column (Phenomenex, Hösbach, Germany). Samples of 0.5 ml were applied to the clean-up column by pumping water (1 ml/min) for 2 min and adsorbed substances transferred to the analytical column by running the eluent 10 mM perchloric acid adjusted to pH 2.5/acetonitrile (70:30, v/v) for 3 min in the reverse direction. The eluate was monitored at 240 nm along with external standards of amitriptyline N-glucuronide and data were processed by the MT2 integration program (Kontron, München, Germany). Means of duplicate incubations were used for calculations. The coefficients of variation within series of eight to nine different concentrations were between 5 and 21% (mean 10%).

Calculations.

K_M and V_maxvalues were calculated from kinetic data according to the Michaelis-Menten equations for one- and two-enzyme kinetics by nonlinear least-squares regression analysis (Fig. P; Biosoft, Cambridge, UK). Goodness of fit represented byr² was 0.985 or greater in measurements with liver microsomes or UGT1A4 and 0.97 to 0.98 in experiments with UGT1A3. Variations given are S.D.

Results

Expressed human UGT1A4 and UGT1A3 catalyzed quaternary ammonium-linked glucuronide formation from both (R)- and (S)-ketotifen. In contrast, no conjugates were detectable in incubates with expressed UGT1A9. As in human liver microsomes (Mey et al., 1999), two isomeric glucuronides were produced from each ketotifen enantiomer. When UGT1A4 was incubated with (R)-ketotifen, the contribution of the faster eluting N-glucuronide isomer decreased from about 65% at 1 to 5 μM substrate to about 35% at 200 μM; a similar change occurred with (S)-ketotifen, the fraction of the faster eluting isomer amounting to about 50% at 1 μM and to about 20% at 200 μM. In contrast, UGT1A3 produced the isomers at constant ratios independent of the substrate concentrations, the faster eluting compounds contributing 50% with (R)- and 25% with (S)-ketotifen. Addition of Triton X-100 at a concentration of 0.2 mg/mg protein did not affect theN-glucuronide quantity produced from 50 μM racemic ketotifen by UGT1A4.

For the sums of the N-glucuronide isomers, one-enzyme kinetics were applicable with UGT1A4 and UGT1A3 (Table1). Although maximal reaction rates with UGT1A4 were nearly identical for the two ketotifen enantiomers, enzyme efficienciesV_max/K_M were somewhat higher for the (R)- than for the (S)-isomer (2.2 versus 1.7 μl/min/mg of protein). In accordance with this, (R)- and (S)-ketotifenN-glucuronides were produced at a ratio of 1:0.7 on incubation of 50 μMracemic ketotifen. TheK_M values for the individual enantiomers were higher than or similar to the low-affinityK_M2 values derived from kinetic studies in native microsomes (Table 1, Fig. 1, A and B). Both isomers were poor substrates of expressed UGT1A3 (Table 1) with lower affinities than toward UGT1A4 and enzyme efficiencies of about 0.03 and 0.1 μl/min/mg of protein for (R)- and (S)-ketotifen, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Eadie-Hofstee plots of the kinetics of N-glucuronide formation from (R)-ketotifen (A), (S)-ketotifen (B), and amitriptyline (C) in human liver microsomes and in homogenates of HK293 cells expressing UGT1A4.

All measurements were carried out in the absence of detergents. The lines were calculated from the parameters for one- or two-enzyme Michaelis-Menten kinetics listed in Table 1.

Incubation of 1 to 200 μM amitriptyline with native human liver microsomes resulted in N-glucuronidation rates compatible with two-enzyme kinetics withK_M1 and K_M2values around 1 and 500 μM (Table 1, Fig. 1C). Inclusion of higher amitriptyline concentrations failed to result in saturation kinetics. Instead, reaction rates exceeded those calculated from the parameters shown in Table 1 and derived K_M2 andV_max2 would have been severalfold higher. UGT1A4 conjugated amitriptyline according to one-enzyme kinetics with aK_M value about 5-fold lower thanK_M2 in liver microsomes (Table 1, Fig. 1C).

Discussion

In this investigation, substrate affinities were to be compared between liver microsomal preparations and homogenates or membranes from cells expressing UGT enzymes. For a valid comparison, enzymes should be in the activated form. This was achieved by Triton addition to microsomes, whereas in the case of cells, homogenization seemed sufficient because Triton failed to affect the conjugation activity. Measurements in liver microsomes in the absence of detergent were included to make sure that biphasic conjugation kinetics were not confined to incubates with detergent.

The results on (R)- and (S)-ketotifenN-glucuronidation by HK293-expressed UGT1A4 and UGT1A3 confirmed earlier results with racemic ketotifen, including the fact that higher rates were achieved with UGT1A4 than UGT1A3 (Green et al., 1995, 1998). That UGT1A9 expressed in V79 cells did not catalyze the reactions was in accordance with findings by Ebner and Burchell (1993)with the racemic drug.

For (S)-ketotifen N-glucuronidation, the apparentK_M for expressed UGT1A4 (86 μM) was nearly identical with the K_M2 value in liver microsomes (92 μM), whereas it was 2-fold (160 μM) in the absence of Triton (Fig. 1B). Similar affinities were found toward UGT1A3 (Table 1). Kinetic analyses of (R)-ketotifenN-glucuronidation are hampered by the small difference between apparent K_M1 andK_M2 values in liver microsomes in the absence of Triton and the lack of clear-cut biphasic kinetics in its presence (Table 1). Because the K_M values of UGT1A4 and UGT1A3 even exceed K_M2 in liver microsomes, the low-affinity component of the conjugation kinetics in liver microsomes seems to be catalyzed by these enzymes. The same conclusion can be drawn with regard to amitriptylineN-glucuronidation. The K_M value of 100 μM by expressed UGT1A4 (which agreed fairly well with 128 and 170 μM found by Green et al., 1998) came far closer toK_M2 (about 310 μM) thanK_M1 (1.4 μM) values in liver microsomes with Triton X-100 added (Breyer-Pfaff et al., 1997). This also applied to data measured in its absence (Fig. 1C, Table 1). Kinetics without Triton were complicated inasmuch as reaction rates obeyed biphasic Michaelis-Menten kinetics up to 200 μM substrate, although larger increases occurred at 400 or 500 μM. In contrast, substrate inhibition had become apparent at 700 or 1000 μM amitriptyline when microsomes were incubated with Triton X-100 (Breyer-Pfaff et al., 1997). The possibility has to be considered that amitriptyline exerted a detergent effect on the microsomal membranes in the absence of an added detergent. In spite of these limitations, it seems justified to state that the low-affinity component of the conjugation kinetics in liver microsomes seems to be catalyzed by UGT1A4.

This leaves open the question as to the nature of the high-affinity UGT. Two possibilities have to be considered, one supposing the existence of one or more yet unidentified UGT(s) with high affinity for tertiary amine drugs, and the other the ability of UGT1A4 (and/or UGT1A3) to attain a high-affinity conformation in human liver microsomes, but not in HK293 cell membranes. This conformation may result from the association to homo- or heterooligomers. Evidence from radiation inactivation analysis pointed to differential catalytic roles for di- and tetrameric UGTs in liver microsomes (Peters et al., 1984;Gschaidmeier and Bock, 1994). Other experiments including chemical cross-linking, coimmunopurification, and site-directed mutagenesis indicated that dimerization is a prerequisite for catalytic activity and that heterodimer formation may lead to novel substrate specificity (reviewed by Radominska-Pandya et al., 1999). A change in substrate affinity in conjunction with a change in oligomeric state has apparently not been described, but it can probably not be dismissed. Such an effect would explain these results, because heterodimers or heterooligomers can be formed in human liver microsomes, but not in cell membranes in which a single UGT is expressed. Whether high-affinity tertiary amine N-glucuronidation requires a special UGT isozyme or a special conformation of a UGT, which in a different conformation catalyzes low-affinity conjugation, is a challenging issue because enzymatic efficiencies of the high-affinity component were far in excess of those of the low-affinity component with the drugs hitherto tested (Breyer-Pfaff et al., 1997; Mey et al., 1999).