Inter-isoform Hetero-dimerization of Human UDP-Glucuronosyltransferases (UGTs) 1A1, 1A9, and 2B7 and Impacts on Glucuronidation Activity

Human UDP-glucuronosyltransferases (UGTs) play a pivotal role in phase II metabolism by catalyzing the glucuronidation of endobiotics and xenobiotics. The catalytic activities of UGTs are highly impacted by both genetic polymorphisms and oligomerization. The present study aimed to assess the inter-isoform hetero-dimerization of UGT1A1, 1A9, and 2B7, including the wild type (1A1*1, 1A9*1, and 2B7*1) and the naturally occurring (1A1*1b, 1A9*2/*3/*5, and 2B7*71S/*2/*5) variants. The related enzymes were double expressed in Bac-to-Bac systems. The fluorescence resonance energy transfer (FRET) technique and co-immunoprecipitation (Co-IP) revealed stable hetero-dimerization of UGT1A1, 1A9, and 2B7 allozymes. Variable FRET efficiencies and donor-acceptor distances suggested that genetic polymorphisms resulted in altered affinities to the target protein. In addition, the metabolic activities of UGTs were differentially altered upon hetero-dimerization via double expression systems. Moreover, protein interactions also changed the regioselectivity of UGT1A9 for querectin glucuronidation. These findings provide in-depth understanding of human UGT dimerization as well as clues for complicated UGT dependent metabolism in humans.

Human UDP-glucuronosyltransferases (UGTs) are key phase II metabolism related enzymes, which conjugate various endobiotics and xenobiotics with the glucuronic acid group of uridine diphosphoglucuronic acid (UDPGA). Given their high polarity, glucuronides are excreted in the urine or bile 1,2 . Thereby, glucuronidation catalyzed by human UGTs represents one of the most important detoxifying processes. To date, nineteen functionally active human UGTs have been identified, and are divided into three subfamilies (UGT1A, 2A, and 2B) based on evolutionary divergence 3 . Of the human UGT members, UGT1A1, 1A9 and 2B7 possess significant catalyzing capacities, and play key roles in the metabolism of therapeutic and environmental compounds, as well as endobiotics such as steroids, fatty acids, and bilirubin 2 . With respect to substrate selectivity, UGT enzymes display distinct, but overlapping functions. Representative substrates of UGT1A1 include bilirubin, ethynylestradiol, raloxifene, and irinotecan [4][5][6] . Meanwhile, estrone, mycophenolic acid, raloxifene and retinoic acid are typical substrates for UGT1A9 5,[7][8][9] . UGT2B7 is known as the most important human UGT enzyme responsible for phase II metabolism of most clinically used drugs, including morphine, zidovudine, and carbamazepine [10][11][12][13] .
As previously reported, human UGTs are coded by polymorphic genes, and the polymorphism effects on metabolism have been comprehensively investigated [14][15][16] . UGT1A1, 1A9 and 2B7 are considered highly polymorphic enzymes. UGT1A1 is the only identified UGT enzyme with bilirubin biotransformation activity. A slightly active variant of UGT1A1 causes bilirubin accumulation, resulting in fatal hyperbilirubinemia 4 . UGT1A9 acts as the predominant enzyme to terminate the biological activity of mycophenolic acid, which is the active metabolite of mycophenolate mofetil, a commonly used immunosuppressant and antiproliferative agent for transplantation patients 17 . The metabolic rates of UGT1A9 allozymes toward mycophenolic acid display a 3-fold variability 18 . UGT2B7 polymorphism significantly alters clearance and outcomes of therapeutic drugs 19,20 . Moreover, individuals harboring the UGT2B7*2 allele have an increased risk of certain cancers, e.g. bladder cancer 21 .
In addition to polymorphisms, oligomerization is another feature of human UGTs. Direct evidences of human UGTs forming various homo-and hetero-oligomers were obtained by co-immunoprecipitation (Co-IP) and fluorescence resonance energy transfer (FRET) techniques. UGT2B1 is known to form complexes with UGT1A proteins 22 . UGT1A members are capable of forming dimers within the enzyme family 23 . Fujiwara et al. confirmed the existence of protein-protein interactions between wild type UGT2B7 and several UGT1A members, using Co-IP 24 . Homo-and hetero-dimerization of UGT1A1 and UGT1A9 allozymes 25 , as well as UGT2B7 allozymes 12 , have recently been reported by our research team. Furthermore, increasing evidence suggests that oligomerization remarkably affects the catalytic activity of individual enzymes involved in the oligomerization system. Dimerization of two different inactive mutants of UGT2B1 results in enzymatic activity recovery 26 . Double expressed wild type UGT2B7 with its allelic variants show significant differences in kinetic behaviors for zidovudine glucuronidation, compared with the single expressed form 12 . Therefore, protein-protein interactions of UGTs play key roles in glucuronidation activity regulation.
The current study aimed to assess inter-isoform hetero-dimerization among UGT1A1, 1A9 and 2B7. The studied enzymes included wild types (*1), one splice mutant of UGT1A1 (1A1*1b), three allelic variants of UGT1A9 (1A9*2, *3, and *5), and three UGT2B7 mutants (2B7*71S, *2, and *5). The Bac-to-Bac system has been used for single or dual expression of target UGT enzymes. We applied quantitative FRET and Co-IP for assessing hetero-dimerization among UGT1A1, 1A9, and 2B7. To assess the relationships between dimerization and glucuronidation activities, quercetin and zidovudine were selected as probe substrates to characterize the catalytic patterns of heterodimers.

FRET analysis.
Hetero-dimerization of UGT1A1*N/1A9*N, UGT1A1*N/2B7*N, and UGT1A9*N/2B7*N was analyzed by quantitative FRET, a reliable method for assessing intra-and inter-molecular interactions [27][28][29] . CFP-tagged UGT1A1*N and 1A9*N, and YFP-tagged UGT1A9*N and 2B7*N enzymes were individually detected with both CFP and YFP channels, and fluorescence tagged UGTs were stably expressed in the endoplasmic reticulum (ER) membrane (Fig. S1). Consequently, recombinant UGT enzymes were double expressed in Sf9 cells as demonstrated by fluorescence microscopy. Sf9 cells co-infected with CFP/YFP baculovirus and those infected with CFP-linker-YFP baculovirus served as negative and positive controls, respectively. We employed the acceptor photobleaching method to determine the FRET phenomenon. After photobleaching the acceptor YFP, no increase of CFP signals in the negative control ( Fig. S2A) was observed; meanwhile, a significant increase occurred in the positive control (Fig. S2B). Except UGT1A1*1b CFP /2B7*5 YFP , all the tested combinations of UGT1A1*N, 1A9*N and 2B7*N exhibited the FRET phenomenon in the co-expression system (Fig. 1); these findings suggested that the two enzymes resided in a close distance and formed protein-protein interactions.
The affinity among UGT isoforms can be quantitatively evaluated by FRET efficiency (E%) and donor-acceptor distance (r). E% of the dimers varied from − 0.74% to 24.3% (Table 1), while r values ranged from 6.38 nm to more than 10 nm. The higher FRET efficiency and shorter donor-acceptor distance both indicated higher interaction ability with the target protein. Comparing the obtained E% and r values, UGT1A1*1 CFP and 1A9*3 YFP showed the highest affinity with each other, while UGT1A1*1b CFP and 2B7*5 YFP displayed the lowest. Compared with wild type UGT1A1 involved dimeric complexes, FRET efficiencies of dimers formed with the splice mutant UGT1A1*1b were dramatically decreased, implying weaker interactions. Based on highly altered FRET efficiencies, the mutation sites in UGT1A9 (*2, *3, and *5) and UGT2B7 (*71S, *2, and *5) variants all influenced the protein-protein interactions. Co-immunoprecipitation. In addition to FRET, hetero-dimerization of UGTs was confirmed by conventional co-immunoprecipitation (Co-IP). Fusion proteins, including UGT1A1*N CFP , UGT1A9*N CFP , UGT1A9*N HA and UGT2B7*N HA , were used for Co-IP assays (Fig. S3). Anti-HA beads were used to immunoprecipitate HA-tagged UGT1A9 and UGT2B7 allozymes. Co-IP experiments were performed with mixtures of two different UGT single expression systems, and alternatively, with UGT double expression systems. For the former operation, cell lysates from UGT1A1*N CFP (81 kDa) or UGT1A9*N CFP (81 kDa) single expression system were mixed with those of UGT1A9*N HA (56 kDa) or UGT2B7*N HA (56 kDa). Subsequently, the mixtures were immunoprecipitated with anti-HA beads and detected by Western blot. As shown in Fig. S5, the bands corresponding to UGT1A1*N CFP or UGT1A9*N CFP did not appear after Co-IP. Only HA-tagged UGT1A9 and UGT2B7 allozymes were detectable. Meanwhile, in the second Co-IP procedure, cell lysates from CFP-tagged and HA-tagged UGTs double expression systems were processed for anti-HA bead immunoprecipitation and Western blot analysis. Both bands of CFP-tagged (81 kDa) and HA-tagged (56 kDa) UGTs were found in Western blot experiments (Fig. 2). These findings indicated that HA-tagged UGT enzymes were immunoprecipitated with CFP-tagged proteins, and hetero-dimers existed. Notably, without a significant FRET phenomena (FRET efficiency was − 0.74%, the donor-acceptor distance > 10 nm.), hetero-dimerization of UGT1A1*1b and 2B7*5 remained undefined. Nevertheless, based on Co-IP analysis, the two proteins still formed a complex during co-expression. The obtained results implied that co-expression, or intracellular environment was an essential condition for the hetero-dimerization of UGT1A1, 1A9 and 2B7.

Discussion
Accumulating evidence demonstrates that human UGTs function in oligomeric forms 26,[31][32][33][34][35] . Oligomerization of UGT1A subfamily members and UGT2B7 isoform, as well as their influences on metabolic activities, have continuously attracted interest. Recently, our lab reported homo-and hetero-dimerization of UGT2B7 allozymes 12 , as well as UGT1A1 and UGT1A9 allozymes 25 . In the current study, FRET and Co-IP were employed to determine inter-isoform interactions among UGT1A1, 1A9 and 2B7 enzymes, further establishing the relationship between protein interactions and alterations in catalytic properties using a panel of double expression systems. FRET is a frequently used technique for assessing macromolecular interactions within 10 nm 27,36 . However, when the distance exceeds the valid amount (> 10 nm), FRET is not capable of determining oligomerization, and Co-IP provides a complementary biochemical evidence for protein oligomerization. Employing quantitative FRET, we confirmed the existence of most hetero-dimers among UGT1A1, 1A9 and 2B7 allozymes. However, without significant FRET phenomena, the UGT1A1*1b and UGT2B7*5 dimer was finally detected by Co-IP, indicating that Co-IP might be more sensitive for identifying relatively weak interactions. FRET and Co-IP results  Table 4.
obtained in this study revealed the promiscuous nature of UGT1A1, 1A9 and 2B7 allozymes to hetero-dimerize with each other, which is also supported by related literatures 23,24 . Compared with the homo-dimerization of UGT1A1, UGT1A9 and UGT2B7 allozymes 12,25 , inter-isoform hetero-dimerizing interactions were obviously  Table 5. weaker. This suggests that the homo-dimers of UGT1A1, UGT1A9 and UGT2B7 might be more predominant than hetero-dimers in vivo.
An array of reports have suggested the importance of oligomerization in biological functions of human UGTs 24,32,34 . In this study, the enzymatic features of target dimers were characterized using double expression systems. Distinct alterations of kinetics were observed compared to the corresponding single enzymes. Although 1A9*N or 1A9*N/2B7*N K m (μM) V max (pmol/min/unit) CL int (μL/min/unit) % of 1A9*N CFP  Table 4. Kinetic parameters of M1 (quercetin-7-glucuronide) formation by single expressed UGT1A9*N CFP and double expressed UGT1A9*N CFP /2B7*N HA systems. Asterisks indicate statistically significant differences compared with UGT1A9*N CFP single expression systems (***P < 0.0001, **P < 0.005, *P < 0.05). Data are mean ± SD of three independent determinations.  Table 5. Kinetic parameters of M2 (quercetin-3-glucuronide) formation by single expressed UGT1A9*N CFP and double expressed UGT1A9*N CFP /2B7*N HA systems. Asterisks indicate statistically significant differences compared with UGT1A9*N CFP single expression systems (***P < 0.0001, **P < 0.005, *P < 0.05). Data are mean ± SD of three independent determinations.
Without the three-dimensional (3D) structures of UGT protein-protein complexes, it is difficult to predict the accurate interaction mechanism of the proteins. However, combination of quantitative FRET data and changes of kinetics parameters provide several hints for the dimerization mode. Based on calculated donor-acceptor distances and FRET efficiencies, the affinities of C-terminal truncated (absence of residues 445-530) mutant UGT1A1*1b to UGT1A9 and UGT2B7 allozymes were notably decreased compared with wild type UGT1A1. Nevertheless, UGT1A1*1b could still form dimers. These findings suggested that C-terminal residues 445-530 belong to the UGT1A1 protein interaction region. The constructed UGT1A1 3D-structural model by Laakkonen and Finel 37 showed that residues 445-533 fold into two C-terminal envelop α -helixes (env1 and env2) and a transmembrane helix (TM-helix) (Fig. S14). UGT1A subfamily members share an identical C-terminal domain; thus, it is likely that the corresponding part of UGT1A9 (residues 442-530) is also involved in protein interactions (Fig. S14). In terms of varying affinities to certain target proteins, C3Y (*2), M33T (*3), and D256N (*5) mutations of UGT1A9, as well as A71S (*71S), H268Y (*2) and D398N (*5) mutations of UGT2B7 may locate in the interface of the two molecules and interrupt protein interactions. The constructed UGT1A9 structural model showed that residues C3, M33 and D256 are in the N-terminal domain, in the signal peptide, loop between the Nβ 1 sheet and Nα 1 helix, and loop between the Nα 9 helix and Cα 0, respectively (Fig. S14) 38 . UGT1A9 and UGT1A1 share a relatively high sequence identity (66%) at the N-terminus. Probably, the N-terminus of UGT1A enzymes facilitates hetero-dimerization. The crystal structure of the C-terminal domain of UGT2B7 clearly revealed that residue D398 resides at Cα 5, which is directly exposed to the solvent 39 . By multiple sequence alignment with UGT1A1 and UGT1A9, A71 and H268 are both considered N-terminal residues. Taken together, we speculate that hetero-dimerization of UGT1A1, 1A9, and 2B7 occurs via both N-and C-terminal domains.
The changes of kinetics also support our hypothesis regarding the UGTs interaction mode. It is commonly assumed that conformational changes arising from protein interactions play a critical role in catalytic activity alterations. The K m value reflects the affinity between substrate and enzyme. Analyzing the kinetic parameters obtained for single and double expressed UGT1A1, 1A9 and 2B7, we found that most K m values were markedly changed by protein interactions. Therefore, hetero-dimerization induced conformational changes of the substrate  Table 6.  Table 7. binding site, resulting in altered K m value. Based on constructed 3D-structure models of human UGTs and the crystal structure of UGT2B7 C-terminal domain [37][38][39][40][41] , we propose that the substrate-binding site is located at the variable N-terminal region, and is involved in protein-protein interactions. Representing the affinity of a substrate to the enzyme, the K m value directly correlates with clearance efficiency, which is evaluated by CL int . However, in many hetero-dimers tested in this study, sharply reduced K m values did not lead to increased CL int , e.g. UGT1A9*2/2B7*N dimers for the catalysis of M1 formation. Conversely, with lower substrate affinity to the enzyme (higher K m values), clearance efficiency of the co-expressed enzymes was even higher, e.g. the UGT1A1*1/2B7*1 dimer for catalyzing zidovudine conversion. Based on these observations, we hypothesize that protein dimerization may also change the spatial arrangement of the co-factor's binding site at the highly conserved C-terminus, and compensate the effects on substrate binding sites. Combined with the above findings, protein-protein interactions in UGT hetero-dimers might extensively include both N-and C-terminal domains of UGTs.
Integrated with K m and V max , clearance efficiency (CL int ) is a parameter reflecting the overall metabolic activity of UGT enzymes. With respect to CL int values, heterodimerization with various allozymes resulted in various alterations of catalytic activities. It should be mentioned that association with the inactive mutant UGT1A1*1b mostly facilitated a dominant negative effect on the glucuronidation activity of UGT1A9 and 2B7. Binding with UGT1A1*1b similarly inhibited the catalytic activity of wild type UGT1A1 25 . We speculate that UGT1A1*1b may serve as a negative modulator for the entire human UGT family rather than just the UGT1A1 isoform. Previously, we reported that most of hetero-dimerizations between UGT1A9 allozymes decreased quercetin glucuronidation activity 25 . Conversely, with a few exceptions, interactions with UGT2B7 allozymes enhance UGT1A9 dependent quercetin metabolism. For UGT2B7 allozymes, interactions with wild type UGT1A1 or UGT1A9 significantly elevated zidovudine catalytic activity, corroborating previous findings 24 . In the present study, UGT2B7 allozymes complexes with UGT1A9 mutants all displayed reduced glucuronidation activities. However, as reported by Fujiwara et al., the dimer of wild type UGT2B7 with UGT1A9*2 mutant exhibits a drastically higher activity 2B7*N or 1A1*N/2B7*N K m (μM) V max (pmol/min/unit) CL int (nL/min/unit) % of 2B7*N HA  Table 6. Kinetic parameters of zidovudine O-glucuronidation by single expressed UGT2B7*N HA and double expressed UGT1A1*N CFP /2B7*N HA systems. Asterisks indicate statistically significant differences compared with UGT2B7*N HA single expression systems (***P < 0.0001, **P < 0.005, *P < 0.05). Data are mean ± SD of three independent determinations.  Table 7. Kinetic parameters of zidovudine O-glucuronidation by single expressed UGT2B7*N HA and double expressed UGT1A9*N CFP /2B7*N HA systems. Asterisks indicate statistically significant differences compared with UGT2B7*N HA single expression systems (***P < 0.0001, **P < 0.005, *P < 0.05). Data are mean ± SD of three independent determinations. than the single expressed UGT2B7 24 , contrasting our data. Differential expression system and protein tags may contribute to this discrepancy. In addition, we previously reported that dimerization affects the regioselectivity of UGT1A9 enzymes 25 . This phenomenon also occurs in a number of UGT1A9*N/2B7*N dimers. Concerning K m values, UGT1A9*2 and *5 both prefer the 7-glucuronidation site of quercetin, while UGT1A9*2 dimers with UGT2B7*1, *71 or *5, and UGT1A9*5 with UGT2B7*1, *71 or *2 change to 3-glucuronidation site. UGT1A9*3 possesses a higher affinity to the 3-glucuronidation site than the 7-glucuronidation site. However, its complexes with UGT2B7*1 and *71S display a preference to the 7-glucuronidation site. The alterations of regioselectivity also suggest the conformational changes of substrate binding site at the N-terminal domain. Considering the complicated catalytic behaviors of UGT mutants and dimers, the associations of structure and function need to be fully established. Moreover, UGTs are key enzymes in drug metabolism. The enzymes terminate the biological actions of their substrates and enhance excretion. It is well known that UGTs form dimers in vivo; however, how dimeric interactions affect UGT activity remains undefined. We established a recombinant system to assess the activity of target hetero-dimers of UGT1A1, 1A9 and 2B7. This enables us to study the enzymes without disturbance from other homologues, still in an in vivo system. The obtained catalytic parameters of the dimers may serve as a reference for a more accurate prediction of drug metabolism in vivo as well as determination of individual drug administration. In conclusion, utilizing quantitative FRET and Co-IP, the current study demonstrated the inter-isoform hetero-dimerization of UGT1A1, 1A9, and 2B7 allozymes. Enzymatic activities were altered by dimerization, to varying degrees. It is conceivable that protein dimeric interactions induce conformational changes and result in altered catalytic properties. The obtained results provide key information for the combined effects of polymorphisms and dimerization on UGT1A1, 1A9 and 2B7 catalytic activity in vivo, and allow in-depth understanding of human UGTs. Heterologous Expression of UGT1A1, 1A9, and 2B7 in Sf9 cells. The pFastBac1-UGT1A1*N-CFP (UGT1A1*1, UGT1A1*1b), pFastBac1-UGT1A9*N-CFP, pFastBac1-UGT1A9*N-YFP and pFast-Bac1-UGT1A9*N-HA (UGT1A9*1, UGT1A9*2, UGT1A9*3, and UGT1A9*5), pFastBac1-UGT2B7*N-YFP and pFastBac1-UGT2B7*N-HA (UGT2B7*1, UGT2B7*71S, UGT2B7*2, and UGT2B7*5) plasmids were previously constructed. Heterologous expressions of UGT1A1 1A9, and 2B7 with Bac-to-Bac system were carried out as the method described by our lab 25 . After transformation of recombinant plasmids into the competent E. coli DH10Bac cells, recombinant bacmid-UGT1A1*N-CFP, bacmid-UGT1A9*N-CFP, bacmid-UGT1A9*N-YFP, bacmid-UGT1A9*N-HA bacmid-UGT2B7*N-YFP and bacmid-UGT2B7*N-HA were separately purified and confirmed. Subsequently, Sf9 cells were transfected with recombinant bacmids and cultured at 27 °C in Sf900II SFM medium. The first passage baculovirus was collected after 72 h of transfection. For protein production, a high titer of recombinant baculovirus was obtained by repeating the amplification. measured with an Olympus BX61W1-FV1000 confocal microscope (Olympus, Tokyo, Japan). The acceptor photobleaching method was employed for the determination of FRET efficiency 42 . Images of fixed Sf9 cells were taken at both CFP (excitation: 405 nm, emitter: 476 nm) and YFP (excitation: 515 nm, emitter: 527 nm) channels. Following was the bleaching procedure. Avoiding detectable photobleaching, the laser intensity was adjusted to less than 5% of the laser power at both of CFP (405 nm) and YFP (515 nm) channels. To achieve the best possible dynamic range, the crosstalk between CFP and YFP channel was eliminated by adjusting the gain of the photomultiplier tube (PMT). Lastly, a selected Sf9 cell was bleached at 515 nm with 100% laser intensity. Two images before photobleaching and eight images after photobleaching were collected at both CFP and YFP channels. The collected data were processed by OLYMPUS FLUOVIEW FV1000 software. The equation E = (D a − D b )/D a was used for the calculation of FRET efficiency (E), where D a and D b are the donor fluorescence intensities after and before photobleaching, respectively. The distance (r) between the donor and acceptor was determined by the equation Co-immunoprecipitation. Sf9 cells with a density of 2 × 10 6 cells/well were seeded and co-infected with recombinant UGT1A1*N-CFP/1A9*N-HA, UGT1A1*N-CFP/2B7*N-HA, and UGT1A9*N-CFP/2B7*N-HA baculovirus. The cells were collected after 72 h, and lysed with 1 mL lysis buffer (0.05 M Tris-HCl, pH 7.4, 0.15 M NaCl, 0.25% deoxycholic acid, 1% Nonidet P-40, 1 mM EDTA) containing 1% protease inhibitor cocktail (Sigma Chemical Co., USA). After centrifugation, the supernatant was mixed with 30 μ L of anti-HA beads (Roche Applied Science, Germany) and incubated overnight at 4 °C. After washing with cold lysis buffer, anti-HA beads were loaded on SDS-PAGE followed by Western blot analysis with anti-UGT1A and anti-UGT2B7 antibodies.

Western blot analysis.
After separation by SDS-PAGE, co-immunoprecipitation samples were transferred onto a polyvinylidene fluoride (PVDF) membrane (Millipore, Bedford, MA). PBST (1 × PBS, 0.1% Tween-20) containing 10% nonfat milk was used for membrane blocking. Subsequently, the PVDF membrane was incubated with rabbit anti-human UGT1A and UGT2B7 polyclonal antibodies diluted 1:2000 for 2 h at room temperature. Horseradish peroxidase conjugated goat anti-rabbit secondary antibodies diluted 1:5000 were then added for 1 h at room temperature. After PBST wash, the membrane was visualized using SuperSignal West Pico (Pierce) and exposed on an X-ray film (Kodak).
For assessing the relative expression levels of CFP-tagged UGT1A1 and UGT1A9, and HA-tagged UGT1A9 and UGT2B7 allozymes in single or double expression systems, total cell homogenates (0.8-16 μ g) were separated by 10% SDS-PAGE and transferred onto a nitrocellulose membrane. Western blot analysis was carried out with anti-UGT1A and anti-UGT2B7 antibodies; band intensities were analyzed on an Odyssey infrared imaging system (LI-COR Biosciences). UGT1A1*1 HA and UGT2B7*1 HA single expression levels were set as 100% for the standard curve. By comparing Western blot densities to those of the calibration curve, the relative expression levels of UGT1A1*N CFP , UGT1A9*N CFP , UGT1A9*N HA and UGT2B7*N HA allozymes were quantified and used for kinetic experiments.
Enzyme activity assay. For determination of relative quercetin glucuronidation activities of UGT1A1*N CFP / 1A9*N HA and UGT1A1*N CFP /2B7*N HA double expression systems, 1 mg/mL total cell homogenates, 0.1 mM quercetin, and alamethicin (50 mg/mg of protein) were incubated in 100 μ L Tris-HCl buffer (0.1 M, pH = 7.4). After 5 min of pre-incubation, the reaction was initiated by adding UDPGA at 37 °C, and was terminated by adding 300 μ L leuteolin in 30 min. Following centrifugation (13,000 × g, 10 min), the supernatant was subjected into high performance liquid chromatography (HPLC) system. Kinetic parameters of quercetin conversion by UGT1A1*N CFP /1A9*N HA and UGT1A9*N CFP /2B7*N HA double expression systems were measured with a serial concentrations of quercetin (5.011-150.330 μ M). A total volume of 100 μ L potassium phosphate buffer (50 mM, pH 7.4) containing 1.5 mg/mL total cell homogenates, 5 mM MgCl 2 , 10 mM UDPGA, alamethicin (50 mg/mg of protein), and zidovudine (75.12-2403.84 mM) was used for zidovudine glucuronidation assay of the co-expressed UGT2B7*N HA with UGT1A1*N CFP or UGT1A9*N CFP . The assay system was pre-incubated for 5 min. As the formation of zidovudine O-glucuronide was linear with time-dependent increase within 120 min, the reaction time was set as 120 min at 37 °C. Adding 10 mL of 30% perchloric acid, the mixture was centrifugated at 13,000 × g for 10 min. The supernatant was analyzed by HPLC. An Agilent 1200 system, equipped with an Agilent ExtendTM C18 (250 mm × 4.6 mm, 5 μ m) column, was used for the analysis of glucuronidation. The mobile phase for separation of quercetin and its metabolites consisted of methanol and 0.02 M pH 2.0 phosphoric acid (46:54, v/v) with a flow rate of 1 mL/min. The detection wavelength was 368 nm. Zidovudine and its glucuronide were analyzed under the following conditions: 16% methanol and 84% KH 2 PO 4 (10 mM, pH 2.2), 1.0 mL/min flow, and 265 nm detection wavelength.
Kinetic analyses of glucuronidation. The software GraphPad Prism, version 5.0 (GraphPad Software Inc., San Diego, USA) was employed for kinetic analysis. The equations, including Michaelis-Menten equation, v = V max · S/(K m + S), the substrate inhibition model, v = V max /(1 + (K m /S) + (S/K si ), and CL int (intrinsic clearance) = V max /K m , were used for the kinetic data calculation, where v is the velocity of the reaction, V max is the maximum velocity, S is the substrate concentration, K m is the Michaelis-Menten constant, K si is the constant describing the substrate inhibition interaction.