The hydroxypyridinone chelator deferiprone is commonly used in Europe and Asia to treat life-threatening iron overload disorders in patients with thalassemia major (Galanello, 2007). Post-transfusional iron overload is associated with high rates of morbidity and mortality, especially due to cardiac disease (Borgna-Pignatti et al., 2004). Deferiprone therapy markedly improves the prognosis of individuals requiring repetitive blood transfusions (Borgna-Pignatti et al., 2006). As opposed to deferoxamine, which must be administered by subcutaneous infusion, the oral formulation of deferiprone simplifies treatment, improves patients' quality of life, and improves adherence (Tricta et al., 1996; Franchini and Veneri, 2004; Delea et al., 2007). Although the long-term safety profile of deferiprone is well defined, the mechanism of and factors involved in its adverse effects, such as gastrointestinal disturbances, arthralgia, neutropenia, and agranulocytosis, are unclear (Agarwal et al., 1990; Taher et al., 2001; Ceci et al., 2002; Cohen et al., 2003; Franchini and Veneri, 2004), thus emphasizing the need to decipher the underlying mechanisms and factors involved in these adverse effects. Deferiprone has good bioavailability, but its clearance is accelerated by rapid biotransformation: approximately 85% of the drug is metabolized to a nonchelating (inactive) 3-O-glucuronide conjugate (Huang et al., 2006) by UDP-glucuronosyltransferases (UGTs). Of all of the known human UGTs, 12 were previously evaluated for deferiprone glucuronidation in a single study, which revealed that only UGT1A6 was capable of catalyzing the glucuronidation of deferiprone (Haverfield et al., 2005). However, a systematic evaluation of all known human UGTs remains to be assessed, and the influence of common genetic variants on deferiprone glucuronidation has yet to be addressed.

Human UGTs are divided into two distinct families: UGT1 and UGT2. The UGT1A gene encodes nine known functional proteins, whereas the UGT2B gene encodes seven, and the UGT2A gene encodes three. In addition, a splicing mechanism involving the use of an alternative exon 5 in the UGT1A locus that has recently been reported generates nine additional UGT1A proteins named isoform 2 (i2). These UGT1A_i2 proteins, although enzymatically inactive, seem to act as modulators of their corresponding UGT1A_i1 classic forms (Girard et al., 2007; Lévesque et al., 2007b).

Genetic variability in drug-metabolizing enzymes is one putative causal factor associated with interpatient variability in responses to drug therapy. Indeed, a large number of genetic variants, especially single nucleotide polymorphisms have been reported in human UGT genes (Guillemette, 2003; Nagar and Remmel, 2006). Of the numerous genetic variations in human UGTs that have been shown to be functional in vitro, several have an impact on in vivo metabolism. For instance, the variant allele UGT1A1*28, associated with reduced expression of the gene (Bosma et al., 1995), increases the risk of toxicity upon treatment with irinotecan, a cytotoxic drug commonly used for the treatment of colon cancer (Ando et al., 2000). Two variations (–2152C>T and –275T>A) in the promoter region of UGT1A9 have been found to be associated with increased expression of the gene (Girard et al., 2004) and decreased exposure in vivo to mycophenolic acid, a commonly used immunosuppressive agent that is metabolized by UGT1A9 (Kuypers et al., 2005; Lévesque et al., 2007a). Reports on the variability in deferiprone glucuronidation and the impact of UGT genetic variants have not yet been published.

Microsomal binding has been shown previously to modify enzyme kinetics and sometimes lead to underestimation of the K_m (Obach, 1997; McLure et al., 2000; Uchaipichat et al., 2006). Basic substrates seem to be especially prone to nonspecific microsomal binding (McLure et al., 2000), but the degree of binding of deferiprone, a rather neutral compound with a very slight basic tendency, to the microsomal membrane, has never been assessed.

The aim of this study was to identify human UGTs involved in deferiprone glucuronidation and to compare their activities in the principal drug-metabolizing tissues. An analytical method based on high-performance liquid chromatography coupled with mass spectrometry (LC/MS) was developed to measure deferiprone 3-O-glucuronide (DG) formation. Binding of deferiprone to the microsomal preparations was also evaluated. We assessed the influence of UGT variant allozymes and splice variants on the kinetics of deferiprone glucuronidation.

Materials and Methods

Chemicals and Reagents. UDP-glucuronic acid was obtained from Sigma Diagnostics Canada (Mississauga, ON, Canada), geneticin (G418) was from Wisent (St. Bruno, QC, Canada), blasticidin was from Invitrogen (Carlsbad, CA), and Lipofectin Reagent was from Stratagene (La Jolla, CA). Protein assay reagents were obtained from Bio-Rad (Hercules, CA). Human embryonic kidney (HEK)293 cells were obtained from the American Type Culture Collection (Manassas, VA). Microsomal preparations from tissues (liver, kidney, ileum, and jejunum) and commercial Supersomes (UGT1A1, 1A3, 1A4, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B15, and 2B17 Supersomes) were purchased from Tissue Transformation Technology (Edison, NJ). Anti-calnexin antibody was obtained from Nventa Biopharmaceuticals (San Diego, CA). Donkey anti-rabbit IgG antibody conjugated with the horseradish peroxidase was purchased from GE Healthcare (Little Chalfont, Buckinghamshire, UK). Liver tissue from eight individuals was kindly provided by Dr. Ted T. Inaba from University of Toronto (Sumida et al., 1999). Deferiprone (APO-066 or L1) (Kontoghiorghes et al., 1990), deferiprone glucuronide, and a deferiprone analog used as an internal standard (1-ethyl-2-methyl-3-hydroxy-4-pyridone) were provided by ApoPharma, Inc. (Toronto, ON, Canada). The substrates 4-nitrophenol and serotonin were purchased from Sigma-Aldrich (St. Louis, MO). Single-use 96-well equilibrium dialyzer plates were purchased from Harvard Apparatus (Ville St. Laurent, QC, Canada). All other chemicals and reagents were of the highest grade commercially available.

Microsomal Preparation from HEK293 Cells. Stable transfection of HEK293 cells with UGT variants (in expression plasmid pcDNA3) and preparation of microsomes by differential centrifugation have been described previously (Villeneuve et al., 2003). The UGT protein levels were determined by Western blotting. In brief, microsomal proteins (20 μg), separated by 10% SDS-polyacrylamide gel electrophoresis and transferred onto nitrocellulose membranes were probed with the anti-human UGT1A antiserum RC71 (1:1000 dilution) (Gagné et al., 2002). To verify equal sample loading, blots were also probed with anti-calnexin antibody (1:2000 dilution). Donkey anti-rabbit IgG antibody conjugated with horseradish peroxidase was used as the secondary antibody (1:10,000 dilution). Visualization and expression level quantification were performed as published previously (Villeneuve et al., 2003). The UGT expression of each microsomal preparation (UGT1A1, 1A3, 1A4, 1A5, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B10, 2B11, 2B15, 2B17, and 2B28) was measured, and enzyme kinetics were adjusted for their relative expression. UGT2A family members were not tested. For stable expression of UGT1A6 isoforms 1 (UGT1A6_i1) and 2 (UGT1A6_i2), HEK293 cells were cotransfected with pcDNA6/V5-HisA-UGT1A6_i1 and pcDNA3/UGT1A6_i2 expression plasmids (derived from human UGT1A6*1) as described previously, using geneticin (1 mg/ml) and/or blasticidin (10 μg/ml) (Guillemette et al., 2000). Several clones expressing both UGT1A proteins were isolated from the initial pool to obtain different expression levels of these two proteins in the same cell and to ensure stable expression over time. Clone 2 was used in these studies and presents a relative expression ratio of i1:i2 = 10:1, which was thought to represent mRNA expression in liver and kidney (Girard et al., 2007).

Enzymatic Assays. To determine which UGTs might be involved in the formation of DG, deferiprone glucuronidation activity of UGT isoforms was measured at 0.2 and 20 mM deferiprone using all commercially available recombinant human UGT isoforms and in-house microsomal preparations from HEK293-UGT clones. UGT1A6_i2 microsomes and UGT1A6_i1/UGT1A6_i2 microsomes were also evaluated. The experiments were conducted at 37°C with ∼30 μg of UGT membrane protein, 50 mM Tris-HCl (pH 7.5), 10 mM MgCl, 2 mM UDP-glucuronic acid, pepstatin, and phosphatidylcholine, as described previously (Bernard et al., 2006). The reactions were terminated by adding 100 μl of methanol + 0.1% tetrafluoroacetic acid, and the mixtures were centrifuged at 14,000g for 10 min before analysis. Detailed kinetic parameters were determined for UGTs exhibiting significant glucuronidation activity for deferiprone (>5% glucuronide formation) upon validation of the assays conditions. Additional enzyme kinetic studies were performed with a fixed UDP-glucuronic acid (2 mM) concentration but varying concentrations of deferiprone (0.1–20 mM), using microsomes from eight liver samples, a pool of liver samples, kidney samples, jejunum samples, ileum samples, and recombinant human UGT1A6*1, *2 (Ala⁷Ala¹⁸¹Ser¹⁸⁴), *3 (Ala⁷), *4 (Ala⁷Ser¹⁸⁴), and *5 (Ala¹⁸¹). Enzyme kinetic studies were also performed with varying concentrations of 4-nitrophenol (0.25–10 mM; 30-min incubation) and serotonin (0.5–40 mM; 60-min incubation) with UGT1A6*1 isoform 1 (UGT1A6_i1) and UGT1A6_i1/UGT1A6_i2 microsomes. These assays were terminated with 100 μl of MeOH (for 4-nitrophenol) or 100 μl of acetonitrile + 0.000002% (v/v) butylated hydroxytoluene (for serotonin). Statistical evaluations of the best-fit model were used to select the enzyme kinetic model and confirmed by a visual inspection of fitted functions (V as a function of [S]) and Eadie-Hofstee plots (V as a function of V/[S]). Kinetic parameter calculations were performed with Sigma Plot 8.0 software assisted by Enzyme Kinetics 1.1 software (SPSS Inc., Chicago, IL). Values are expressed as the mean of at least two experiments performed in triplicate.

Binding of Deferiprone, Serotonin, and 4-Nitrophenol to Microsomal Preparations. Binding of deferiprone, serotonin, and 4-nitrophenol to the microsomal membrane was investigated using a method based on that described by McLure et al. (2000). In brief, 100 μl of assay mix, as prepared for the previously described enzymatic assays (but containing no UDP-glucuronic acid, deferiprone, or microsomes) was loaded on one side of the dialysis plate. The same buffer containing the microsomal preparation (liver, ileum, HEK293-UGT1A6_i1 (*1), or HEK293-UGT1A6_i1 + _i2 for deferiprone; UGT1A6*1 for 4-nitrophenol and serotonin) at the concentration used in the enzymatic assays, and varying concentrations of substrate, as used in the enzymatic assays (0.1–20 mM for deferiprone, 0.25–10 mM for 4-nitrophenol, and 0.5–40 mM for serotonin) was loaded on the other side of the membranes. The plate was rotated at 12 rpm for 3.5 h in an incubator set at 37°C. Control experiments with buffer or microsomes on both sides of the dialysis well were performed to ensure that the equilibrium was reached. Moreover, amitriptyline, previously shown to be highly bound to microsomes (McLure et al., 2000), was used as a positive control at 1, 20, and 60 μM. The contents of each side of the well was pipet-mixed, and 80 μl was collected from each side; 80 μl of methanol + 0.1% tetrafluoroacetic acid (for deferiprone), methanol (for 4-nitrophenol), acetonitrile (for serotonin), or acetonitrile + 0.2% formic acid (amitriptyline) was added and vortex-mixed. The mixtures were then centrifuged at 14,000g for 10 min before analysis by high-performance liquid chromatography-tandem mass spectrometry.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Deferiprone glucuronidation activity of UGT isoforms expressed in HEK293 cells or available as commercial Supersomes. Top, deferiprone glucuronidation activity in expressed UGTs and commercially available Supersomes measured at a 20 mM substrate concentration. Bottom, relative UGT content was assessed by Western blot for each isoform using anti-UGT1A and anti-UGT2B antibodies. (UGT1A1 microsomes, UGT1A10 Supersomes, and UGT2B7 microsomes and Supersomes were arbitrarily set to 1.0, respectively, for UGT1A and UGT2B family members.)

LC/MS Analysis of Enzymatic and Microsomal Binding Assays. 4-Nitrophenol glucuronide and serotonin glucuronide were detected using LC/MS, as described elsewhere (Nguyen et al., 2008). Detection of DG was supported by the following LC/MS protocol. The analytic system consisted of a high-performance liquid chromatography module (Alliance model 2690; Waters, Milford, MA) and an API 3000 TurboIon spray quadrupole mass spectrometer with an electrospray ion source (Thermo Fisher Scientific, Waltham, MA). Assay mixtures were centrifuged for 10 min at 14,000g, and the supernatant was collected. Samples (5 μl) were injected on to a Gemini C₁₈ 10 cm × 4.6 mm, 4-μm column (Phenomenex, Torrance, CA). The mobile phase consisted of 30% solution A (acetonitrile) and 70% solution B (0.5% heptafluorobutyric acid in 0.01% trifluoroacetic acid in ultrapure water). The flow rate was 1.1 ml/min. Ionization was performed at 550°C in the positive ion mode with an ionization voltage of 5 kV, an orifice voltage of 50 V, and collision energy of –26 V. DG, deferiprone, and the APO-066 analog (internal standard) were detected using the following mass transitions: 315.9 → 140.2 (DG), 140.1 → 96.0 (deferiprone), and 154.1 → 126.1 (internal standard). Under these conditions, retention times for DG, deferiprone, and the internal standard were 1.08, 1.32, and 1.54 min, respectively. The range of detection spanned from 0.05 to 150 μg/ml. The limit of quantification was 50 ng/ml using a signal/noise ratio of 10. A validation study of this method demonstrated good intra- and interday precision (coefficient of variation <10%) and accuracy (bias <5%). To further separate the different glucuronides found to arise from deferiprone, the mobile phase was modified to contain 15% solution A and 85% B. β-Glucuronidase treatment was performed to confirm the glucuronide nature of the peaks observed upon high-performance liquid chromatography-tandem mass spectrometry analysis.

Data Analysis. Statistical analyses were performed with JMP 4.0.2 software (SAS Institute, Cary, NC). Data are expressed as mean ± S.D. Normality of distribution was assessed using the Shapiro-Wilk W test. P < 0.05 with analysis of variance was considered statistically significant.

Results

Identification of Human UGTs Involved in DG Formation. We assayed 16 human UGT isoforms of the UGT1A and UGT2B families, overexpressed in HEK293 cells, and 12 commercial isoforms for deferiprone glucuronidation activity. Among the HEK293-UGTs assayed at 0.2 and 20 mM substrate concentrations, UGT1A6 exhibited the highest deferiprone glucuronidation activity. Six other isoforms (UGT1A7, 1A8, 1A9, 1A10, 2B7, and 2B15) had measurable, although significantly lower activities (<3% of the activity observed with UGT1A6 at 20 mM deferiprone) (Fig. 1). Commercial sources of UGTs (Supersomes) usually had lower activities than the corresponding HEK293-UGTs under these conditions, but UGT1A6 remained the most effective enzyme in conjugating deferiprone.

Detection of a Second Minor Glucuronide. The formation of a second glucuronide with the same mass transition (315.9 → 140.2) as the 3-O-glucuronide was detected in human tissue samples and with some of the cloned UGT enzymes. The second glucuronide was most abundant in incubations with UGT1A8, 1A9 and 1A10 microsomes, representing 41 to 86% of the total glucuronides formed in these preparations using 20 mM substrate. In tissue samples and UGT1A6 microsomes, its formation was scant, ranging from 4.1 to 7.9% of the glucuronides formed (Table 1). The chromatography conditions were modified to separate the two glucuronides: a representative chromatogram from each protocol is illustrated in Fig. 2. Under the modified conditions, the retention times of the most and least abundant forms were 1.7 and 2.0 min, respectively, whereas the retention times for deferiprone and the internal standard were 3.1 and 4.3 min. Both glucuronides had identical mass transitions (315.9 → 140.2); moreover, both metabolite peaks disappeared upon treatment with β-glucuronidase, confirming their glucuronide nature.

Kinetics of Tissue Microsomes and Recombinant UGT1A6. The kinetic parameters of UGTs present in tissue samples and UGT1A6 expressed in HEK293 cells were evaluated. The Michaelis-Menten kinetic model adequately described the kinetic data for liver tissue and UGT1A6-HEK293, whereas the Hill equation better described the observed kinetics for ileum, jejunum, and kidney tissues (Fig. 3). Table 2 reports the kinetic estimates. For liver microsome preparations (n = 8 individuals), V_max values varied from 7.9 to 19.2 nmol/min/mg protein, whereas the apparent K_m values were less variable (7.6–10.0 mM). Liver and kidney microsomes had similar drug clearance rates, which were significantly higher than those for jejunum and ileum.

Kinetics of UGT1A6 Allelic Variants. The enzymatic activity of several UGT1A6 allelic variants expressed in HEK293 cells was assessed. Kinetic parameter values for the most common UGT1A6 variants, UGT1A6*2 (Ala⁷Ala¹⁸¹Ser¹⁸⁴), *3 (Ala⁷), *4 (Ala⁷Ser¹⁸⁴), and *5 (Ala¹⁸¹) were determined (Table 3). The Michaelis-Menten kinetic model adequately described the kinetic data of all variants. Compared with UGT1A6*1 (the reference UGT1A6 allelic variant; V_max = 25.6 nmol/min/mg protein; K_m = 10.0 mM), the relative maximum velocity of all evaluated variants was significantly reduced (V_max = 7.5–19.1 nmol/min/mg), but the affinity remained in a similar range, as reflected by similar K_m values (K_m = 7.4–8.2 mM) (Fig. 4). UGT1A6*4 also showed a significant (71%) decrease in the estimated intrinsic clearance (V_max/K_m) compared with UGT1A6*1.

The UGT1A6 Isoform 2 Splice Variant. Enzymatic assays using microsomal fractions of HEK293 cells stably expressing the UGT1A6_i2 protein demonstrated that this protein was enzymatically inactive when UDP-glucuronic acid was used as a cosubstrate (data not shown). Because UGT1A6_i2 was shown to be coexpressed with UGT1A6_i1 in samples from liver and kidney (Girard et al., 2007) and based on previous observations with other UGT1 proteins (Girard et al., 2007; Lévesque et al., 2007b), we sought to determine whether UGT1A6_i2 protein potentially influenced UGT1A6_i1 enzymatic activity. In the presence of UGT1A6_i2 at a relative expression ratio of approximately 0.1:1 (UGT1A6_i2/UGT1A6_i1) for the clonal cells used, the transferase reaction of UGT1A6-expressing cells was slightly modified with a reduction in glucuronidation activity in the presence of UGT1A6_i2 compared with UGT1A6_i1 (*1) activity alone (Table 3). No significant changes could be detected for either 4-nitrophenol or serotonin in the presence of this low level of i2 protein (Table 3).

Binding of Deferiprone, Serotonin, and 4-Nitrophenol to the Microsomal Membrane. To evaluate whether the observed variations in kinetic data could be due to differential binding of the substrate to the different microsomal preparations, microsomal binding of deferiprone to liver and ileum microsomal preparations and to HEK293-UGT1A6*1 and HEK293-UGT1A6_i1 + i2 preparations were performed. However, the unbound fraction for each of these experiments was 99 to 100%, meaning there was no significant binding of deferiprone to any of these microsomal preparations at the concentrations used in the kinetics experiments. Binding of serotonin and 4-nitrophenol indicates that the free fraction for serotonin was approximately 100% and approximately 85% for 4-nitrophenol.

Discussion

Deferiprone was the first orally active iron chelator approved for treatment of iron overload and has been estimated to have been used in more than 7500 patients, in some cases daily for more than 14 years (Kontoghiorghes et al., 2003). Despite the long-term use of deferiprone, the mechanisms underlying variations in the patients' response to therapy and the appearance of side effects have been little studied to date (Galanello, 2007). Given that genetic variability in drug-metabolizing enzymes could be a relevant factor, the objective of this study was to assess deferiprone metabolism by UGTs and the possible impact of genotypic variation on glucuronidation in vitro.

Our findings indicate that UGT1A6 is the major enzyme responsible for deferiprone glucuronidation; several other UGTs (1A7, 1A8, 1A9, 1A10, 2B7, and 2B15) make a minor contribution (less than 3% of UGT1A6 activity). Therefore, they are expected to account for little of the in vivo metabolic clearance of deferiprone. This statement is supported by comparable kinetics in human tissues, especially for the liver. Data therefore suggest that deferiprone represents a reasonable UGT1A6-specific substrate (Fig. 1). These findings are similar to data previously reported by Haverfield et al. (2005).

UGT1A6 has been demonstrated to have a particular affinity for substrates containing a nucleophile attached to an aromatic ring (Sorich et al., 2004), which is consistent with the chemical structure of deferiprone (Fig. 2). UGT substrates typically contain electrophilic acceptor groups such as aliphatic alcohols and phenols, amines, and sulfhydryl or carbonyl groups (Dutton, 1980; Shipkova and Wieland, 2005). UGT1A6 preferentially metabolizes phenols (Harding et al., 1988; Orzechowski et al., 1994; Tukey and Strassburg, 2000), favoring formation of the 3-O-glucuronide, which was identified and measured in earlier studies (Kontoghiorghes et al., 1990; Huang et al., 2006). The second deferiprone glucuronide form, detected in glucuronidation assays with specific UGTs (UGT1A6, 1A8, 1A9, and 1A10) and certain human tissues, has not been reported previously. The in vitro production rate of this second glucuronide form in liver, kidney, jejunum, and ileum microsomes (4.1–7.9% of total glucuronide formed) suggests that the amount formed in vivo is likely to be minimal in relation to that of the main metabolite. The molecular structure of this second glucuronide could not be determined precisely and its biological activity is unknown.

Of the tissues examined in vitro, liver and kidney cleared deferiprone most effectively, indicating that these are likely to be major contributors to deferiprone glucuronidation, followed by the intestines. Indeed, jejunum and ileum had much lower clearance rates compared with the other organs, possibly due to the lower relative abundance of UGT1A6 in these tissues. The fact that the K_m in the liver was comparable to that for UGT1A6 is consistent with the involvement of a single enzyme in this organ. Modifications in kinetic models have previously been shown to be due to unspecific binding of the substrate to the microsomal membrane, leading to underestimation of the K_m (Obach, 1997; McLure et al., 2000; Uchaipichat et al., 2006). Such binding to the microsomal membrane has been shown to be substrate-dependent, with some, such as amitriptyline, having a high level of binding, some, such as trifluoperazine (Obach, 1997), having a concentration-dependent binding level (Uchaipichat et al., 2006), and others, such as naproxen, having no measurable binding (McLure et al., 2000). Based on our results, deferiprone falls in the latter category, possibly because of its rather neutral nature. Therefore, the difference in the kinetic models (Michaelis-Menten in the liver and Hill in kidneys and intestines) suggests the potential involvement of other enzymes in these extrahepatic tissues, for example, UGT1A8, which is not expressed in the liver but is expressed in the other organs, or could also be caused by the contribution of UGT2A enzymes, which were not tested here.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Detection of a second glucuronide generated by some UGT isoforms. A, representative chromatogram of deferiprone and deferiprone glucuronide; retention times for deferiprone glucuronide, deferiprone, and the internal standard were 1.08, 1.32, and 1.54 min, respectively. Right panel, MS scans confirm the nature of the observed peaks. B, representative chromatogram from modified chromatographic conditions to observe the presence of two deferiprone glucuronides. Under these conditions, the retention times of the most and least abundant forms were 1.7 and 2.0 min, respectively, whereas the retention times were 3.1 and 4.3 min for deferiprone and the internal standard. C, molecular structure of deferiprone and its 3-O-glucuronide.

In the present study, all common variant UGT1A6 allozymes demonstrated significantly lower velocities compared with the reference UGT1A6*1 enzyme, although their affinities for deferiprone were similar. The frequencies of UGT1A6*2 (Ala⁷Ala¹⁸¹Ser¹⁸⁴), *3 (Ala⁷), *4 (Ala⁷Ser¹⁸⁴), and *5 (Ala¹⁸¹) allelic variants were estimated to be 31, 6, 1, and 1%, respectively, in a group of Americans composed mainly of Caucasian individuals (Krishnaswamy et al., 2005a). These frequencies are in line with those provided by the Pharmacogenetics and Pharmacogenomics Knowledge Base (http://www.pharmgkb.org). It is therefore expected that individuals carrying *2, *3, *4, or *5 variations (therefore, approximately 40% of the population) may present reduced glucuronidation of deferiprone in vivo. Previous in vitro studies indicated that although the codon 181 (541 A>G) and 184 (552 A>C) variations, present in alleles *2, *4, and *5, were generally associated with increased activity for several substrates (serotonin, 5-hydroxytryptophol, 4-nitrophenol, acetaminophen, and valproic acid) (Nagar et al., 2004; Krishnaswamy et al., 2005b), decreased activity was reported with 4-nitrophenol, 4-tert-butylphenol, 3-ethylphenol/4-ethylphenol, 4-hydroxycoumarin, butylated hydroxyanisole, and butylated hydroxytoluene (Ciotti et al., 1997). The codon 7 variation (19 T>G), present in allele *2, *3 and *4 and also found in complete linkage disequilibrium with promoter polymorphisms (–1710 C>G, –1310 del 5, and –652 G>A), was associated with lower UGT1A6 mRNA levels in human liver. However, no significant impact on protein level and glucuronidation activity was confirmed (Krishnaswamy et al., 2005a). In another study, a 2-fold increase in glucuronidation was observed for the UGT1A6*2 protein with the substrates serotonin, 5-hydroxytryptophol, 4-nitrophenol, acetaminophen, and valproic acid. Moreover, homozygous *2/*2 human liver microsomes demonstrated higher K_m values compared with *1/*1 microsomes (Krishnaswamy et al., 2005b). Rapid in vivo metabolism has been reported in UGT1A6*2 carriers for two substrates, sodium valproate (Sun et al., 2007) and salicylic acid (Chen et al., 2007). Taken together, the data support a substrate-dependent effect of these amino acid variations in UGT1A6 protein. A recent in vivo study of the effect of UGT1A6 polymorphisms on deferiprone pharmacokinetics failed to demonstrate a significant impact of the *2 variant compared with the *1 variant (Limenta et al., 2008). However, this study was limited to 22 individuals, and the impact of other variants (*3, *4, and *5) was not assessed.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Kinetics from liver followed the Michaelis-Menten model (A), which differed from the kidney (B), jejunum (C), and ileum (D), where the Hill equation better described the kinetics.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Comparison of relative V_max for UGT1A6 genetic variants generated in HEK293 cells. The V_max values for UGT1A6*2, *3, *4, and *5 are significantly reduced compared with that of *1. A reduction in velocity is also observed with UGT1A6_i1 + i2. Protein expression of each variant was analyzed by Western blot, and their respective relative quantification values are shown. *, significantly different from UGT1A6*1 at P ≤ 0.05.

Using the recent discovery of an isoform 2 splice variant of UGT1 proteins (Girard et al., 2007; Lévesque et al., 2007b), we showed that the presence of the UGT1A6_i2 isoform reduced the glucuronidation of deferiprone by the classic UGT1A6_i1 enzyme. This finding is in agreement with our previous findings with other UGT1A proteins, in which the isoform 2 acted as a negative modulator of isoform 1 proteins (Girard et al., 2007). This observation provides a potential source of variation in the in vivo metabolism of deferiprone because the mRNA of UGT1A6_i2 was detected in all liver and kidney samples tested, although in variable amounts (Girard et al., 2007). These organs are most likely to be responsible for deferiprone glucuronidation based on in vitro kinetic data. In contrast, the lower rates of DG formation in the intestinal microsomes are more likely to have been due to intrinsically lower UGT1A6 activity in the intestine, as our preliminary studies suggest a very low level of expression in the intestines (Girard et al., 2007). Differential expression of this second isoform 2 among individuals might therefore modify the glucuronidation rate of the drug. Moreover, the influence of the isoform 2 protein at this low level of expression compared with that of isoform 1 (in a ratio of 10/1 for i1/i2) appears to be substrate-specific because no significant differences were observed for 4-nitrophenol and serotonin glucuronidation. Additional investigations are required. However, as there was no significant binding of deferiprone to either the UGT1A6*1 (i1) or the i1+ i2 microsomal preparations, the observed differences could not be explained by differential binding to the microsomal membrane.

We conclude that the glucuronidation of deferiprone depends almost exclusively on UGT1A6 and that common genetic variants are likely to be associated with reduced deferiprone glucuronide formation. The presence of the UGT1A6_i2 splice variant protein may also represent a potential source of variation in deferiprone glucuronidation by UGT1A6.