Genetic Variants of Pregnane X Receptor (PXR) and CYP2B6 Affect the Induction of Bupropion Hydroxylation by Sodium Ferulate

This study investigated the effects of pregnane X receptor (PXR/NR1I2) and CYP2B6 genetic variants on sodium ferulate (SF)-mediated induction of bupropion hydroxylation. The pharmacokinetics of bupropion and hydroxybupropion were evaluated after an oral dose of bupropion (150 mg) administered with and without SF pretreatment for 14 days in 33 healthy subjects. The area under the time-concentration curve (AUC) ratio of AUC_hyd (AUC(0-∞) of hydroxybupropion)/AUC_bup (AUC(0-∞) of bupropion) represents the CYP2B6 hydroxylation activity, which was significantly lower in CYP2B6*6 carriers (NR1I2 TGT noncarriers or carriers) than in noncarriers in both the basal and SF-induced states (p-value<0.05). AUC ratio and AUC_hyd of NR1I2 -24113AA variant were markedly lower than GA and GG genotypes (7.5±2.1 versus 14.5±3.3 and 20.6±1.1, and 8873±1431 versus 14,504±2218 and 17,586±1046) in the induced states. However, -24020(-)/(-) variant didn't show significant difference in the induction of CYP2B6 hydroxylation activity by SF compared with other -24020[GAGAAG]/(-) genotypes. NR1I2 TGT haplotype (-25385T+g.7635G+g.8055T) carriers exhibited a significantly decreased AUC ratio, compared with TGT noncarriers, in the basal states (7.6±1.0 versus 9.7±1.0), while this result wasn't observed in CYP2B6*6 noncarriers. Moreover, individuals with complete mutation-type [CYP2B6*6/*6+NR1I2 TGT+ -24113AA+ -24020 (-)/(-)] showed even lower percent difference of AUC ratio (8.7±1.2 versus 39.5±8.2) than those with complete wild-type. In conclusion, it is suggested that NR1I2 variants decrease the bupropion hydroxylation induced by SF treatment, particularly in CYP2B6*6 carriers. Trial Registration ChiCTR.org ChiCTR-TRC-11001285

Sodium ferulate (3-methoxy-4-hydroxy-cinnamate sodium, C 10 H 9 NaO 4 , SF) is the sodium salt of ferulic acid (FA), which is widely distributed in herbs and Chinese formulas such as Ligusticum, Chuanxiong and Chaihu-Sugan-San [21,22]. It is usually used as food supplements or herbal medicine in countries or areas accepting the theory of Traditional Chinese Medicine (TCM). With the anti-oxidant, anti-atherogenic, anti-platelet clotting, anti-inflammatory, lipid-lowering, cholesterol biosynthesis inhibitory and analgesic effects [23][24][25][26], FA has the potential to be developed into an effective pure compound for prevention and treatment of cardiovascular diseases. Presently, SF has been approved by State Food and Drugs Administration of China (SFDA) as a clinical therapy for cardiovascular and cerebrovascular diseases [27,28].
Combination of SF and CYP2B6 substrate drugs has the potential possibility of drug-drug interactions and may lead to undesirable and harmful clinical consequences. Although human CYP2B6 represents approximately 1% of total hepatic CYP content, it shows a relative contribution of 2% to 10% in total hepatic CYP activity, and participates in the metabolism of a variety of substances including bupropion, selegiline, valproic acid, cyclophosphamide, ifosfamide, nevirapine, efavirenz, propofol, ketamine, and synthetic opioid methadone [29]. In particular, the metabolic pathway of hydroxybupropion is almost exclusively catalyzed by CYP2B6 that is a standard model for studies of drugdrug interactions of CYP2B6 substrate drugs [30][31][32][33]. Recently, we verified that bupropion hydroxylation metabolism (represents CYP2B6 metabolism activity) was induced by a 14 days pretreatment of 150 mg SF in healthy volunteers [34]. Our previous investigations in HepG2 cells suggested that FA may increase 67% transcriptional expression of CYP2B6 through PXR activation compared with control group (unpublished data). Due to frequently combinational use, especially in China, of SF and other drugs during the treatment of clinical diseases, it is important to be aware of the possibility of interactions of combination of SF and CYP2B6 substrate drugs, and to prevent harmful clinical toxicity.
Effects of CYP2B6 variations on protein expression levels and enzyme activities may cause up to hundreds fold of interindividual difference in exposure to drugs [35]. Early studies indicated CYP2B6*6 carriers contributed to the interindividual difference in CYP2B6 substrate drugs disposition [36,37]. The low-activity and high-frequency of allele CYP2B6*6 was very prevalent in African Americans (32.8%), Papua New Guineans (62%) and Asians (21%) [35]. It can be simply speculated that the number of abnormal population of CYP2B6 variants will be marked, when other CYP2B6 functional variants are added. Therefore, it is very critical to investigate whether CYP2B6 variants effect the induction of bupropion hydroxylation by SF.
Based on above studies, SF-induced CYP2B6 activity is suggested to be associated with both CYP2B6*6 polymorphisms and other factors such as NR1I2 genetic variants. Furthermore, both NR1I2 and CYP2B6 variants may be associated with the clinical pharmacokinetics and/or interactions of SF and bupropion. However, clinical pharmacogenetics study data of NR1I2 and CYP2B6 variants is still scarce. Further studies are still necessary to investigate the functions of NR1I2 and CYP2B6 genetic variants in clinic. The purpose of this paper is to evaluate the effect of NR1I2 and CYP2B6 genetic variants, and to demonstrate the relationship between these genetic variants and metabolic induction of bupropion hydroxylation by SF administration in Chinese individuals.

Materials and Methods
The protocol for this trial and supporting CONSORT checklist are available as supporting information; see Checklist S1 and Protocol S1.

Genotyping
Genomic DNA was isolated from peripheral blood samples using SQ Blood DNA KitII(Omega Bio-Tic, Georgia, USA). CYP2B6. The wild-type allele CYP2B6*1 was defined as 516G/ 785A, and CYP2B6*6 was defined as 516T/785G. CYP2B6*6 was detected in a haplotype assay using a two-step allele-specific PCR as described previously [38]. The validity of the method was confirmed by sequencing.

Subjects
To detect NR1I2 and CYP2B6 genetic polymorphisms in the Chinese population, a total of 152 individual samples (from the DNA bank, Hunan Key laboratory of Pharmacogenetics, Central South University) were genotyped. Thirty-four healthy male volunteers (eighteen CYP2B6*1/*1, nine CYP2B6*1/*6 and seven CYP2B6*6/*6) were enrolled in the clinical trial with informed consent form signed (aged 20 to 24 years; weight range: 52-77 kg; body mass index range: 18-25 kg/m 2 ), and divided into CYP2B6*6 noncarriers (CYP2B6*1/*1, wild-type) and CYP2B6*6 (CYP2B6*1/*6+CYP2B6*6/*6) carriers. The health status of subjects were ascertained by checking medical history and taking a full clinical examination, drug screening, and standard hematologic and blood chemical laboratory tests. Standardized protein-rich diets with no vegetables, fruits or cereals were provided for subjects for 2 weeks prior to study and during the whole study, in order to exclude the influence of food-originated FA. Drugs, alcohol, soft drinks, tobaccos, vitamins and caffeinecontaining beverages, any nutritional supplements were refrained for 2 weeks before study commencement and throughout the study. Regular heavy drinkers, smokers, users of glucocorticoids and those with body weight exceeded their ideal weight by 20% were excluded. Finally, 33 of recruited subjects finished the trial (the date range of subject enrollment was from April 24, 2011 to July 25, 2011).

Study Design
This study was carried out in a two-phase, randomized, crossover manner with a 2-week washout period between phases. In each phase, after an overnight fast, subjects were given pretreatment with or without three 50-mg SF tablets (one tablet, three times a day) of the same batch (Lot No.: 100810; HengDa ShengKang Pharmaceutical Co., Sichuan, China) for fourteen days. The signature of subjects, supervision of investigators and detection of plasma concentrations of FA were carried out to assure subject compliance to treatment. On day 15, after an overnight fast, a single dose of 150 mg bupropion (two tablets of 75 mg Zyban SR; WanTe, Hainan, China) was given to each subject by oral administration with 200 ml water at 8:00 a.m. Subjects fasted for another 4 h after drug administration, except water drinking 2 h after dosing. Standard meals were provided for all of the participants. Serial blood samples for PK analysis (5 ml) were collected using a forearm in-dwelling venous catheter (anticoagulation with sodium heparin) before dosing and at 0.5, 1, 2, 3, 4, 5, 6, 8, 10, 12, 24, 36, 48, 60 and 72 h after bupropion ingestion.

Drug Concentration Analysis
The plasma concentrations of bupropion and hydroxybupropion were examined by liquid chromatography-mass spectrometry using Waters Micromass Quattro Micro API LC/MS/MS instrument (Milford, MA, USA). An Angel kromasil C18 (5 mm, 15062.1 mm) and a mobile phase (acetonitrile:0.1% formic acid:20 mM ammonium formate = 4:3:3) at a flow rate of 0.2 ml/min were applied. Propranolol was used as the internal standard. The ion transitions monitored were as follows: m/z 240 to 184 for bupropion,m/z 256 to 238 for hydroxybupropion and m/z 260 to 183 for propranolol. These transitions represent the product ions of the [M+H] + ions. The lower limits of detection for bupropion and hydroxybupropion were 0.25 ng/ml and 1.168 ng/ml, and the assay ranges used were 0.42-430.1 ng/ml and 2.344-1200 ng/ml, respectively. The linear correlation coefficient for bupropion calibration curves was 0.998, and for hydroxybupropion was 0.996. The highest bupropion and hydroxybupropion plasma concentration measured were 323.70 ng/ml and 615.66 ng/ml. The mean extraction recovery and precision of bupropion and hydroxybupropion was assessed by determining quality control (QC) plasma samples at three concentration levels (concentrations of bupropion and hydroxybupropion were 2.344, 37.5, 600 ng/ml and 0.84, 13.438, 430 ng/ml, respectively). The recovery of bupropion and hydroxybupropion, determined at each concentration level was 92.167.

Pharmacokinetic Analysis
The maximum plasma concentration (C max ) and the time to C max (T max ) were obtained by inspection of the concentration-time data. The AUC to the last quantifiable concentration AUC 0-t was determined by use of the linear trapezoidal rule. ke is the elimination rate constant determined from the terminal slope of the log concentration-time plot. The elimination half-life (t 1/2 ) was calculated as 0.693/ke. The area under the concentration-time curve extrapolated to infinity AUC 0-' was calculated as AUC 0-' = AUC 0-72 +C72/ke, where C72 is the plasma concentration measured 72 h after drug administration. The oral clearance (CL/F) of bupropion was calculated by dividing the bupropion dose by the AUC of bupropion and the subject's weight.

Statistical Analysis
Study sample sizes were estimated based on prior bupropion PK data. The planned sample size, statistical power, and alpha level were performed using the NCSSV2007 program. For instance, if three subjects in TGT carriers for CYP2B6*6 noncarriers group were enrolled, the power (AUC ratio) is up to 0.92371 (a = 0.05, b = 0.2) according to prior bupropion PK data. Likewise, in TGT carriers for CYP2B6*6 carriers group, if five subjects participated in this study, the power will add to 0.93165 (a = 0.05, b = 0.2) (calculation methods as described NCSS2007 instruction). The bioequivalence approach was used to determine clinically relevant interactions [39]. AUC ratio, namely AUC_hyd (AUC (0-') of hydroxybupropion) was divided by AUC_bup (AUC (0-') of bupropion) for each period, representing CYP2B6 activity. The percent differences in the PK parameters between the basal and SF-treated states were calculated as an absolute of 1006(inducedbasal)/basal. WinNonlin (version 5.2; Pharsight, Mountain View, CA) was used for the PK analysis. The paired two-tailed t-tests were used to determine the difference between basal and induced states, and logarithmic transformation was used for the nonnormally distributed data before analysis. The differences in PK parameters between noncarriers and carriers groups of NR1I2 TGT haplotype and CYP2B6*6 genotypes were obtained using the Wilcoxon rank-sum test. The difference among NR1I2 -25385C .T (CC, CT and TT) and -24113G.A (GG, GA and AA) genotype groups was obtained by use of the Kruskal-Wallis test. The Fisher's exact test was used to detect difference of genotype distributions between CYP2B6*1/*1 and CYP2B6*1/*6+CYP2B6*6/*6. Results were expressed as mean 6 standard deviation in the text and tables, and as mean 6 standard error in the figures. Linkage disequilibrium (LD) analysis and haplotype construction were performed using the Haploview 4.

Results of the Clinical Study
According to the ICH-GCP guideline [40], no serious drugrelated adverse event was observed from the 33 subjects during the course of this study. No clinically significant alterations were observed in heart rate, blood pressure or body temperature. No significant difference was shown in the demographic characteristics of the volunteers or in the distributions of CYP2B6 genotypes among NR1I2 haplotypes (see Table 1).
As shown in Fig. 1, individual plots of CYP2B6 and NR1I2 variants indicated that the combination of NR1I2 TGT haplotype and CYP2B6*6 affected the AUC ratio in both the basal and induced states. The concentration-time profiles of hydroxybupropion were very different for CYP2B6*6+NR1I2 TGT carriers from other groups, with the lowest values in both the basal and induced states (see Fig. 2). The Cmax values of hydroxybupropion in the basal and induced states showed no significant difference between NR1I2 TGT carriers and noncarriers (unpublished data). However, the hydroxybupropion Cmax in CYP2B6*6+NR1I2 TGT carriers was significantly lower than CYP2B6*6+NR1I2 TGT noncarriers in both the basal and induced states (284.3640.8 versus 395.2640.5, and 332.9627.6 versus 424.5632.8, pvalue,0.05).
Further haplotypes analysis showed that compared with the Korean, the composition and frequencies of NR1I2 TGT haplotypes in Chinese people were completely different. Eight haplotypes were inferred based on SNPs in positions -25385C.T, g.7635A.G, and g.8055, and the frequency distribution of TGT haplotype was slightly lower (0.136) than that of Korean (0.199). Furthermore, linkage manners of NR1I2 SNPs were inconsistent with the Korean. A slight but not complete LD among -25385C.T, -24113G.A, and -24020[GAGAAG].(-) and between g.7635A.G and g.8055C.T was observed [20]. These results suggest that the NR1I2 gene has unique characteristics of high polymorphism and significant interethnic variants.
As shown in Table 4, we found in the clinical investigation that in TGT noncarriers (TGT noncarriers, n = 14), the overall pharmacokinetic parameters of bupropion and hydroxybupropion (AUC_bup, AUC_hyd, and AUC ratio) indicated the strongest effects including basal activities, induced activities and their percent differences. However, with the emergence of TGT and CYP2B6*6 variants (TGT carriers, n = 4 and TGT noncarriers, Pharmacogenetics of PXR and CYP2B6 PLOS ONE | www.plosone.org n = 9, respectively), the strongest effects of the overall pharmacokinetic parameters of bupropion and hydroxybupropion became weaker and smaller, until TGT variant existed in CYP2B6*6 carriers (TGT carriers, n = 6). Moreover, in each step of the attenuation effects of the pharmacokinetic parameters, TGT and CYP2B6*6 carriers always showed smaller effects than noncarriers. Therefore, this result suggests that the decreased metabolism of bupropion with SF treatment is affected by both NR1I2 TGT and CYP2B6*6 variants. Also, similar findings were obtained from -24113G.A, -24020 [GAGAAG]. (-), and complete wild/ mutation-type individuals (see Table 2 and 3). In short, our data strongly support the hypothesis that NR1I2 TGT haplotype, -24113AA, and CYP2B6*6 variants play very important roles in bupropion disposition. Interestingly, in previous study, NR1I2 TGT carriers slightly manifested stronger effects on some pharmacokinetics parameters of bupropion and hydroxybupropion than the corresponding noncarrier groups (p-value.0.05). Conversely, CYP2B6*6 carriers showed smaller effects on AUC_hyd, and AUC ratio than the noncarriers (p-value,0.05) [20]. Reports also showed that CYP2B6 expression increased in the basal state while decreased in the induced state when treated with rifampin in PXR.2 cells [44]. However, these results were not observed in our study. Rifampin is a known selective human PXR activator with little cross-interaction with other receptors, such as small heterodimer Table 2. Effects of complete wild-type and mutation-type individuals on SF-mediated metabolic induction of bupropion hydroxylation.

Complete wild-types (n = 6)
Complete mutation-types (n = 2)  Table 3. Effects of NR1I2 SNP polymorphisms on SF-mediated metabolic induction of bupropion hydroxylation.  partner (SHP) and hepatocyte nuclear factor-4a(HNF-4a) [45,46]. Early study indicated that interaction of PXR with HNF-4aand its coactivators, peroxisome proliferator-activated receptor-c-coactivator-1a(PGC-1a) contributed to the strong induction of CYP3A4 by rifampin, whereas gene expression of SHP was simultaneously inhibited by PXR, which weakened inhibitory effect of SHP on CYP3A4 expression and strengthened the HNF-4ainducibility of CYP3A4 [45,47]. Therefore, we may assume that under the situation that some interfering factors of SHP gene expression exceeded the effects of PXR variants (PXR function variant led to increased SHP gene expression), the activity of CYP did not decline but increased. More attention should be paid to the study of SHP gene expression and regulation in the near future. Moreover, Owen's group reported that genetic variability in constitutive androstane receptor (CAR) was involved in the metabolism and disposition of CYP2B6 substrate drugs recently [48]. In our study, further analysis showed that TGT carriers, only in the basal states, had significantly lower AUC ratio and percent differences (7.661.0 versus 9.761.0, and 17.169.5 versus 23.267.0) than TGT noncarriers. However, CYP2B6*6 carriers exhibited significant differences in the most of pharmacokinetic parameter values of bupropion and hydroxybupropion (AUC_bup, AUC_hyd, and AUC ratio) compared with CYP2B6*6 noncarriers in both the basal and induced states (see Table 4). In addition, from Fig. 1 and Fig. 2, a tenuous distinction existed in the concentration (conc.) -time curves of hydroxybupropion and AUC ratio when TGT carriers appeared; while the curves and AUC ratio values quickly dropped when CYP2B6*6 carriers came. This result suggests that CYP2B6*6 variants had stronger reduced metabolic capacity than NR1I2 TGT haplotype. Interestingly, the complete mutation-type [CYP2B6*6/*6+NR1I2 TGT+ -24113AA+-24020(-)/(-)] individuals indicated even lower metabolism activities (8.761.2 versus 39.568.2) than the complete wildtypes (see Table 2). Therefore, we tentatively conclude that reduced metabolic capacity is more significant in individuals including CYP2B6*6 mutations, NR1I2 TGT haplotype, and other NR1I2 variants with reduced functions.
To date, the verified sites and positions of the SNPs (or haplotypes) of NR1I2 functional variants were as follows:  [18,20,42,43,49,50]. However, some investigations of clinical pharmacogenetics of the NR1I2 functional variants were not consistent with their findings in vitro. For instance, NR1I2 -25385C.T, -24113G.A, 7635A.G, or 8055C.T was reported to be associated with higher magnitude of induction of intestinal CYP3A by rifampin in vitro [18], but recently, the subjects with -25385C.T or TGT (-25385T+g.7635G+g.8055T) carriers were verified to have decreased CYP2B6 activity (AUC ratio) induced by rifampin in Korean, and NR1I2*1B (8055C.T+2654T.C) haplotype was strongly associated with its downstream target genes of MDR1 in Asian breast cancer patients [42]. Moreover, the result for our subjects with -24113AA showed the lowest percent differences of AUC ratio (11.367.9) after SF induction compared with wild genotypes (see Table 3). The specific mechanism of this difference in vitro and in vivo is yet unknown. However, we must admit that the result in vitro was easily interfered by diverse uncontrollable factors. These results highlighted the important role of NR1I2 pharmacogenetics in the disposition of putative drug substrates. It can be assumed that NR1I2 genetic polymorphisms will play an essential role in affecting interethnic variations in drug disposition.
Previous researches reported that oxysterol, 24(S), 25-epoxycholesterol (LXR agonists), glucocorticoid (GR agonists), and vitamin D (VDR agonists) could induce expression of CYP2B6 through the binding co-activators of the corresponding ligands and PXR [51][52][53]. Moreover, gender factor also affected the results of clinical trials [54,55]. It is worth mentioning that these interference factors were well balanced through our strict subject exclusion criteria and good clinical trial control. However, there are also some deficiencies in our study. For instance, we paid more attention to the NR1I2 variants, which had the functions reported in vitro or in vivo such as -25385C.T, -24113G.A, -24020[GA-GAAG]/(-), 7635A.G, and 8055C.T [18,19,20,42,49,50,[56][57][58]. The rarely reported or less concerned NR1I2 variants were not included in this research. Future clinical pharmacogenetics research of NR1I2 variants should be focused on the reported functional variants with lower distribution frequencies, which have potential possibility to play more important roles than the star variants. In addition, lower concomitance mutation frequencies and strict subject exclusion criteria also affected our subjects enrollment. Relatively small and uneven numbers of individuals with various genotypes were investigated in our study, which were the limitations of drawing of conclusions based upon this sample sizes. Further clinical studies of the NR1I2 variants pharmacoge- netics should be operated in larger groups, and even different ethnic populations [59].

Conclusions
As the data indicated, NR1I2 TGT haplotype, -24113AA, CYP2B6*6, and the complete mutation-type [CYP2B6*6/ *6+NR1I2 TGT+ -24113AA+-24020 (-)/(-)] individuals have strong evidence to show the ability to reduce the metabolic capacity of CYP2B6 after SF administration in Chinese individuals. Individuals/ethnic populations with different genetic backgrounds may show significant differences in drug metabolism and efficacy, sometimes even manifested as severe adverse drug reactions or no efficacy. Our findings provide an important reference for carrying out the gene oriented individual/interethnic therapy of CYP2B6 substrate drugs, and avoiding the adverse effects of SF and CYP2B6 substrate drugs combination. Whether other NR1I2 and regulator variants also have impact on the disposition of CYP2B6 substrate drugs by SF requires further exploration. Large-scale population pharmacokinetic, pharmacodynamic, and nosazontology analysis using pharmacogenomics method are needed to clarify the role of NR1I2 and CYP2B6*6 variants in the efficacy, safety, and drug interactions of CYP2B6 substrate drugs, even disease susceptibility between individuals.

Supporting Information
Checklist S1 CONSORT checklist.