Mycophenolic acid (MPA), the active metabolite of the prodrug mycophenolate mofetil (MMF), is a standard immunosuppressive drug used after solid organ transplant and in several autoimmune diseases. MMF is currently being used for the prophylaxis of acute rejection in liver, kidney, heart, and lung transplantation (Shaw et al., 1998). Its mechanism of action involves the direct inhibition of inosine monophosphate dehydrogenase, thus preventing the proliferation of B and T lymphocytes (Lipsky, 1996). Several studies in various transplant subpopulations have recognized the large variability in the pharmacokinetic parameters of the drug, thus emphasizing the need to individualize MMF dosing (Bullingham et al., 1998; Ensom et al., 2002). Potential causes of such variability include genetic variations in the metabolic pathways of the drug. To explore this hypothesis, we need to accumulate data on the enzymes involved in the metabolism of MPA. Its metabolism of MPA is for the most part mediated by UDP-glucuronosyltransferases (UGTs) (Bullingham et al., 1998)

In vitro metabolic studies with microsomes prepared from human tissues predict that the formation of MPAG occurs in the kidney and the intestine but predominantly in the liver (Bowalgaha and Miners, 2001; Shipkova et al., 2001). However, the identity of specific UGT enzymes involved in the formation of MPA glucuronide derivatives is still not well defined since literature reports are inconsistent, and there have been no systematic studies of all known functional UGTs. To date, 16 functional proteins have been identified in humans and are classified in distinct families and subfamilies based on their sequence homology. The UGT1A subfamily includes nine functional isoforms (UGT1A1 and 1A3–10), while the UGT2B subfamily comprises seven isoforms (UGT2B4, 2B7, 2B10, 2B11, 2B15, 2B17, and 2B28) (Guillemette, 2003). A first study, using thin-layer chromatography to detect the formation of MPA glucuronide metabolites in enzymatic assays with ¹⁴C-UDP-glucuronic acid, indicated UGT1A8, 1A9, and 1A10 as potent isoforms toward MPA, whereas other isoenzymes, namely UGT1A1, 1A3, 1A4, 1A6, 2B7, and 2B10, were found to be totally inactive (Mackenzie, 2000). In contrast, a second study, supported by a high-performance liquid chromatography protocol with UV detection, found that UGT1A1, 1A3, 1A4, 1A6, 1A7, 1A9, 1A10, 2B4, 2B7, and 2B15 had similar capacities for the formation of MPAG, 1A7 and 2B4 being most efficient (Shipkova et al., 2001). Not all UGTs were tested in the same experimental conditions in both of these studies, and a systematic approach to identify the UGT(s) responsible for the formation of the main metabolite MPAG was still missing.

The main objective of this study was to test all 16 UGT1A and UGT2B human recombinant UGT proteins known to date for their capacity to form MPAG in the same experimental conditions. For the determination of MPA and MPAG, we have first developed a simple, rapid, and sensitive high-performance liquid chromatography coupled with mass spectrometry (LC/MS) method. Experiments were performed in parallel to assess the conjugation activities of liver, kidney, and intestine microsomes, and those were compared with recombinant UGTs. Finally, the effects of common polymorphism of UGT allozymes involved in the metabolism of MPA were investigated to gain information on the potential role of genetic factors to the variability in pharmacokinetic parameters of MMF therapy.

Materials and Methods

Chemicals and Reagents. UDP-glucuronic acid and MPA were obtained from Sigma Diagnostics Canada (Mississauga, ON, Canada). MPAG was a generous gift from Hoffman-La Roche (Mississauga, ON, CA). All other chemicals and reagents were of the highest grade and commercially available.

Human UGT1A and UGT2B Alleles and Tissues. UGT-HEK293 cells including 1A1, 1A7, and 1A9 polymorphic allozymes have been described previously (Gagne et al., 2002; Villeneuve et al., 2003). Before enzymatic assays, Western blot analyses and catalytic activities of known substrates were performed for each preparation. In addition, preparations of UGT1A1, 1A9 (BD Gentest, Woburn, MA), 1A7, and 1A10 (PanVera Corp., Madison, WI) were obtained from commercial sources to validate results obtained with UGT-HEK293 cells. Two pools of human liver (n = 5 subjects), two human kidney samples (white male/head trauma and white female/stroke) and two pools of human intestine (jejunum, n = 10; ileum, n = 5) were obtained from Tissue Transformation Technologies (Edison, NJ). Four additional human liver samples obtained from Human Cell Culture Center Inc. (Laurel, MD) were also studied. UGT1A8 and 1A10 variants (Huang et al., 2002; Elahi et al., 2003) were generated by polymerase chain reaction site-directed mutagenesis using pcDNA3 vector on UGT^*1 constructions, kindly provided by Dr. Thomas Tephly (University of Iowa, Iowa City, IA) (Cheng et al., 1998). Primers used for 1A8 site-directed mutagenesis were 5′-CGCCAGGGGAATAGCTTGCCACTATCTTG-3′ (forward) and 5′-CAAGATAGTGGCAAGCTATTCCCCTGGCG-3′ (reverse) for 1A8^*2 (A¹⁷³G) and 5′-GGTGGTATCAACTACCATCAGGGAAAGCC-3′ (forward) and 5′-GGCTTTCCCTGATGGTAGTTGATACCACC-3′ (reverse) for 1A8^*3 (C²⁷⁷Y). Primers used for 1A10^*2 (E¹³⁹K) site-directed mutagenesis were 5′-GTAGAATACTTAAAGAAGAGTTCTTTTGATGCA-3′ (forward) and 5′-TGCATCAAAAGAACTCTTCTTTAAGTATTCTAC-3′ (reverse). Stable HEK293 cell transfection with variant pcDNA3-UGT expression plasmids, preparation of microsomes by differential centrifugation, and determination of UGT protein levels by Western blot have been previously described (Guillemette et al., 2000; Gagne et al., 2002; Villeneuve et al., 2003). Relative to the UGT1A9^*1 microsomes, UGT protein expression values were 13.8, 16.2, 1.3, 1.0, and 44.6 for UGT1A1, UGT1A7, UGT1A8, UGT1A9, and UGT1A10, respectively. Variant allozyme expression values were 1.1, 0.3, 0.9, and 0.4 for UGT1A8^*2, UGT1A8^*3, UGT1A9^*2, and UGT1A9^*3, respectively.

MPA and MPAG Analysis by LC/MS. Detection of MPA and MPAG was supported by a LC/MS protocol. The analysis system consisted of a high-performance liquid chromatography module (Alliance model 2695; Waters, Milford, MA) and a LCQ Advantage quadrupole ion trap mass spectrometer using an electrospray ion source (Thermo Finnigan, San Jose, CA) controlled by Xcalibur software (Thermo Finnigan). Acidified assays were centrifuged for 6 min at 14,000g and 250 μl of supernatant were collected. Samples (10 μl), maintained at 4°C, were injected on a 4.6 mm × 75 mm (3.5-μm diameter) Symmetry C-18 reversed-phase column (Waters). The mobile phase consisted of solution A [water with 0.001% formic acid (v/v) and 5 mM ammonium formate] and solution B [methanol with 0.001% formic acid (v/v) and 5 mM ammonium formate] with the following gradient: 35 to 95% B (0–8 min); 95% B (9 min); 95 to 35% B (9–9.01 min); 35% B (12 min). The flow rate was 0.8 ml/min. Injections were carried out at a spray voltage of 2.0 kV, a capillary voltage of -13.0 V, and a capillary temperature of 250°C. MS detection of MPA was followed in the negative ion mode with full MS base peak at m/z 319.1, whereas detection of MPAG was followed with full MS base peak at m/z 495.1 and its identification was supported with a scan of full MS/MS base peak at 319.1. Under these conditions, retention times for MPAG and MPA were 3.70 and 6.96 min, respectively. The signals were found to be linear from 0.1 to 25 μg/ml for both MPAG and MPA. The limit of quantification was 0.1 μg/ml using a signal-to-noise ratio of 3. The within-day precision was <5.0% and the between-day precision was <10%.

Enzymatic Assays. To determine which UGTs might be involved in the formation of MPAG, initial experiments with UGT1A and UGT2B microsomes including commercial preparations consisted of 4- and 16-h incubations at 37°C with 0.2 mM MPA, 10 to 50 μg of UGT membrane protein, 50 mM Tris-HCl, 10 mM MgCl₂, 1 mM UDP-glucuronic acid, pepstatin, and phosphatidylcholine, yielding a pH of 6.8, in a standard procedure performed in our laboratory (Gagne et al., 2002). Detergent was omitted in the enzymatic assays since a previous study demonstrated no beneficial activation with common detergents (Shipkova et al., 2001). Determination of V_max and K_m was then performed for UGTs with a significant glucuronidation activity for MPA (>5% glucuronide formation). UGTs and human liver, kidney, and intestine microsomes were incubated for 1 h with MPA (1–1000 μM). Kinetic parameters were also determined for allelic variants UGT1A8^*1, ^*2 (A¹⁷³G), and ^*3 (C²⁷⁷Y), and for UGT1A9^*1, ^*2 (C³Y), and ^*3 (M³³T). For the low-activity UGTs, namely 1A1, 1A7, and 1A10, activities were determined using 1 mM MPA. Visual inspection of fitted functions (V as a function of [S]) and Eadie-Hofstee plots (V as a function of V/[S]) was used to select the best-fit enzyme kinetic model. Kinetic parameter calculations were performed with Sigma Plot 8.0 software assisted by Enzyme Kinetics 1.1 software (SPSS Inc., Chicago, IL). Values were expressed as the mean of at least two experiments performed in duplicate. p value calculations were performed with JMP 4.0.2 software (SAS Institute, Cary, NC) using Student's t test.

Results and Discussion

The present in vitro incubation studies using human liver, intestine, and kidney microsomes and cDNA-expressed human UGTs clearly demonstrate that the metabolism of MPA occurs at multiple sites and is mediated primarily by two UGT isoforms, the extrahepatic UGT1A8 and the hepatic UGT1A9. Specific coding region polymorphic alleles of the high MPA-glucuronidating activity UGT1A8 and UGT1A9, as well as in the low-activity UGTs, namely UGT1A1, 1A7, and 1A10, are predicted to have an impact on the metabolism of MPA.

MPAG Formation by Liver, Kidney, and Intestine Microsomes. Previous studies proposed the liver to be the main organ involved in the metabolism of MPA, whereas the kidney demonstrated one of the highest efficiencies to generate MPAG of all tissues tested (Bowalgaha and Miners, 2001; Shipkova et al., 2001). In this study, the two pools of human liver microsomes and individual samples of human kidney displayed similar capacities to generate MPAG (Cl_int from 42 to 102 μl/min/mg protein) (Table 1). A considerable variability in kinetic parameters was observed between individual liver samples as well as between pools, especially in the values of velocity, ranging from 4822 to 13,790 pmol/min/mg protein. The highest affinity was observed using pooled liver samples with K_m values of 95 to 135 μM, whereas the highest rates were observed with kidney microsomes, with velocities ranging from 23 to 36 nmol/min/mg, 3-fold higher in mean than liver microsomes. On the other hand, pooled kidney microsomes demonstrated the lowest affinity, with K_m values ranging from 553 to 887 μM. Interestingly, HL-P2, which demonstrated a superior catalytic efficiency for MPA glucuronidation, was also found to have the highest UGT1A9 protein expression compared with other liver samples (data not shown). Additional liver samples from four unrelated subjects were also investigated, and the affinity for the formation of MPAG was in the same range as the K_m observed for pooled livers, ranging from 151 to 272 μM (Table 1). Three liver samples, HL-S1, HL-S2, and HL-S3, were found to have similar capacities to generate MPAG with Cl_int ranging from 12 to 18 μl/min/mg protein. The fourth sample, HL-S4, had a 2-fold higher catalytic efficiency for MPAG formation compared with the other samples. As for the human liver pools, catalytic efficiencies of all four liver samples were found to be associated with the level of UGT protein content assessed by Western blot analysis (data not shown).

The two pools of jejunum and ileum microsomes had a significantly lower activity toward MPA, with 2- to 7-fold reduced catalytic efficiencies compared with the other tissues tested. In a previous study, it was reported that MPA glucuronidation activity is similar in jejunum and ileum (Shipkova et al., 2001), consistent with a contribution of the gastrointestinal tract to the first-pass metabolism of MPA. The present study has found ileum microsomes to have a lower capacity to generate MPAG compared with jejunum. This observation is in agreement with previous studies reporting that UGT activity is the highest in the jejunum within the small intestine (Tukey and Strassburg, 2000).

Identification of Human UGTs Involved in the Formation of MPAG. For the first time, all 16 known human UGTs were tested in the same experimental conditions. A sensitive LC/MS protocol was used for the detection of MPA and MPAG with authentic standards, and results were further confirmed using commercial preparations of UGTs. Using 200 μM MPA for 4- and 16-h incubation periods, the formation of MPAG was observed in the presence of a limited number of UGT microsomes. Only UGT1A1, UGT1A7, UGT1A8, UGT1A9, and UGT1A10 were shown to generate a significant amount of MPAG, whereas all other UGTs demonstrated barely detectable or no ability to form MPAG, corresponding to less than 5% conversion of MPA into MPAG. In addition, compared with HEK293-UGT microsomes, the commercial sources of UGT1A1, 1A7, 1A9, and 1A10 shared similar MPA conjugation activities in these conditions after normalization for UGT expression levels (data not shown).

Kinetic analyses were then performed with the most reactive UGTs, namely, UGT1A1, 1A7, 1A8, 1A9, and 1A10 (Table 2). Upon normalization for UGT expression levels, data point to UGT1A8 and UGT1A9 as the main enzymes involved in the formation of MPAG. Both enzymes demonstrated the highest catalytic efficiency for MPAG formation, at least 30-fold higher than any other active UGTs, with a Cl_int of 43 μl/min/mg for 1A8, followed by 1A9, with a Cl_int of 29 μl/min/mg. In contrast, the low capacities of 1A1, 1A7, and 1A10 suggest that the contribution of these isoforms is likely to be minor. Our in vitro metabolic studies are mostly in agreement with the previous study of Mackenzie et al. (2000), although in their study, the detailed kinetics was not presented and the capacity of UGTs in family 1 and 2 was incomplete. On the other hand, in contrast to the study of Shipkova et al. (2001), our findings point to a limited number of UGTs involved in the glucuronidation of MPA. One of the potential reasons for the differences is likely to be related to the conditions of the assays and to the fact that kinetic parameters were not adjusted according to protein expression, as was done in the present study. In support of the main conclusions of our metabolic investigation using in-house HEK293-UGT preparations, results were further corroborated with commercial preparations of UGTs, and similar results were obtained.

Given that UGT1A9 demonstrates the highest capacity to form MPAG compared with 1A1, also present in the liver, a central role of this isoenzyme in the hepatic clearance of MPA is proposed. The expression of 1A8 and 1A10 is limited to the gastrointestinal tract, whereas 1A1 and 1A9 are also found in these tissues and would play a role in the enterohepatic metabolism of MPA, a process that is substantial for this drug in vivo (Shaw and Nowak, 1995). In turn, UGT1A7 appears be a higher-affinity, but much lower-capacity enzyme compared with all other active UGTs, despite the fact that this enzyme is not present in the liver, the gastrointestinal tract and the kidney (Tukey and Strassburg, 2000), suggesting an unlikely role of this specific UGT in the in vivo glucuronidation of MPA. Furthermore, only UGT1A9 was reported to be expressed in the kidney, a tissue that exhibits a significant glucuronidation activity for MPA (Bowalgaha and Miners, 2001; Shipkova et al., 2001).

MPAG Formation by UGT Variants. Genetic polymorphisms that affect the metabolic activity or expression of UGT biotransformation enzymes, especially UGT1A8 and UGT1A9, may be important contributors to interindividual differences in MPA pharmacokinetics. In line with this hypothesis, we have investigated several known polymorphisms of the UGT1A8 and UGT1A9 genes (Table 2). UGT1A8^*1 and ^*2 (A¹⁷³G) allelic variants possess similar K_m values, 298 ± 82 and 317 ± 69 μM, respectively. In turn, the K_m of the UGT1A8^*3 (C²⁷⁷Y) allele was not significantly lower (118 ± 56 μM; p = 0.12). After normalization for the level of UGT-expressed proteins, the relative velocity of the UGT1A8^*3 allozyme (C²⁷⁷Y) (5371 pmol/min/mg) was drastically decreased by 5-fold, compared with the UGT1A8^*1 protein (25,252 pmol/min/mg) (p = 0.03). Conversely, compared with the UGT1A9^*1 protein, the velocities of the ^*2 and ^*3 allozymes remain unchanged. In turn, compared with the UGT1A9^*1 (K_m = 291 ± 31 μM) and UGT1A9^*2 (K_m = 224 ± 32 μM) proteins, the affinity of the UGT1A9^*3 (M³³T) allozyme was significantly decreased (K_m = 720 ± 250 μM; p = 0.05) (Table 2). Cl_int of the UGT1A9^*3 allele was 1.7-fold lower than that of the UGT1A9^*1 protein. In previous studies, several mutations in the UGT1A1 gene were identified that induced nonsynonymous amino acid alterations, including G⁷¹R (^*6), Y⁴⁸⁶D (^*7), P²²⁹Q(^*27), and L²³³R (^*35) (Guillemette, 2003). All variant UGT1A1 proteins were shown to have a substantial decrease (reduced by 86–100%) toward the formation of MPAG compared with the 1A1^*1 protein. Since both UGT1A7 and UGT1A10 demonstrated detectable MPA glucuronidation activity, all nine 1A7 and the two 1A10 variant proteins described previously were also tested (Elahi et al., 2003; Villeneuve et al., 2003). In the analysis of UGT1A7 allozymes, the highest MPA glucuronidation activity was observed for UGT1A7^*1, ^*2 (K¹²⁹K¹³¹), ^*6 (D¹³⁹), and ^*7 (K¹²⁹K¹³¹D¹³⁹). The previously described low-activity variant proteins, namely ^*3 (K¹²⁹K¹³¹R²⁰⁸), ^*4 (R²⁰⁸), ^*5 (S¹¹⁵), ^*8 (K¹²⁹K¹³¹D¹³⁹R²⁰⁸), and ^*9 (S¹¹⁵K¹²⁹K¹³¹), presented lower rates of MPAG formation compared with ^*1 (G¹¹⁵N¹²⁹R¹³¹E¹³⁹W²⁰⁸), similar to what was observed for every other substrate tested so far (Gagne et al., 2002; Villeneuve and Guillemette, 2003). Furthermore, a 1.5-fold reduced MPA glucuronidation activity was also observed for the UGT1A10^*2 (E¹³⁹K) protein compared with the ^*1 protein (data not shown).

In conclusion, the UGT1A8^*3 (C²⁷⁷Y) allozyme demonstrated a significantly altered catalytic efficiency toward MPA. A similar impact of these variations was reported with other substrates of the UGT1A8 protein (Huang et al., 2002). Since UGT1A8 is not present in the liver, the effect of the ^*3 allele may be limited to the metabolism of MPA in the gastrointestinal tract and could have a potential role in the toxicities associated with MMF in this tissue (Shaw et al., 1998). Similarly, the UGT1A9^*3 (M³³T) protein demonstrated a lower Cl_int compared with the ^*1 and suggests a potential impact of this variation to the in vivo metabolism of the drug, particularly in the liver, and to the interindividual variability in the pharmacokinetics of MMF. However, the low frequency of the UGT1A8^*3 and UGT1A9^*3 alleles, present in less than 5% of the population, would indicate a limited impact overall (Huang et al., 2002; Villeneuve et al., 2003). Based on the results of our studies, a unique involvement of the hepatic UGT1A9 enzyme in MPAG formation in the liver is shown and is predicted to impact the in vivo variability of the drug in this tissue. Further studies on genetic factors, mainly of the UGT1A9 gene, associated with altered glucuronidation profiles of MPA are needed.