Pharmacogenomics of human P450 oxidoreductase

Cytochrome P450 oxidoreductase (POR) supports reactions of microsomal cytochrome P450 which metabolize drugs and steroid hormones. Mutations in POR cause disorders of sexual development. P450 oxidoreductase deficiency (PORD) was initially identified in patients with Antley–Bixler syndrome (ABS) but now it has been established as a separate disorder of sexual development (DSD). Here we are summarizing the work on variations in POR related to metabolism of drugs and xenobiotics. We have compiled mutation data on reported cases of PORD from clinical studies. Mutations found in patients with defective steroid profiles impact metabolism of steroid hormones as well as drugs. Some trends are emerging that establish certain founder mutations in distinct populations, with Japanese (R457H), Caucasian (A287P), and Turkish (399–401) populations showing repeated findings of similar mutations. Most other mutations are found as single occurrences. A large number of different variants in POR gene with more than 130 amino acid changes are now listed in databases. Among the polymorphisms, the A503V is found in about 30% of all alleles but there are some differences across different population groups.


INTRODUCTION
Cytochrome P450 reductase (POR) has a major role in metabolism of drugs and steroids Pandey and Flück, 2013;Riddick et al., 2013) (Figure 1). All microsomal cytochrome P450 get electrons from nicotinamide adenine dinucleotide phosphate (NADPH) through POR   (Figure 2). A single gene of 71,754 bp located on chromosome 7 (locus 7q11.23) encodes human POR ( Table 1). The human POR gene has one non-coding exon and 15 protein-coding exons and encodes a 680 amino acid membrane bound protein (RefSeq protein: NP_000932, UniProt P16435). Disruption of POR affects all microsomal P450 enzyme activities Pandey and Flück, 2013). POR also supplies electrons to many other interaction partners. POR knockout mice are embryonic lethal (Shen et al., 2002;Otto et al., 2003). The liver-specific knockout of POR produces mice with lower metabolism and lipid accumulation (Gu et al., 2003;Henderson et al., 2003;Porter et al., 2011).
Cytochrome P450 oxidoreductase deficiency (PORD); (OMIM: 613571 and OMIM: 201750) is a form of congenital adrenal hyperplasia Miller et al., 2004;Pandey et al., 2004;Flück and Pandey, 2013). Clinical symptoms of PORD were described in a 46,XY patient with disorder of sexual development (DSD) (Peterson et al., 1985). Sequencing of the POR gene in a Japanese 46,XX girl with a metabolic profile of steroid deficiency and symptoms of Antley-Bixler Syndrome (ABS) led to the characterization of PORD . Skeletal malformation in many ABS patients with ambiguous genitalia and defective steroid metabolism have now been linked to PORD (Huang et al., 2005). PORD is now listed as a separate metabolic disorder (Huang et al., 2005). An overview of reported cases shows that PORD disrupts steroid biosynthesis in adrenal gland and gonads (Huang et al., 2005;Dhir et al., 2007;Pandey et al., 2007;Flück et al., 2011a). This often leads to genital ambiguity at birth in both male and female. The absence of testosterone (T)/dihydrotestosterone (DHT) leads to development of female external genitalia (Flück and Pandey, in press). The presence of T/DHT triggers the formation of male genitalia. In humans the testosterone is mainly produced in testes by a process that starts by the conversion of pregnenolone to 17-hydroxypregnenolone (17-OHPreg) and dehydroepiandrosterone (DHEA) by the enzyme CYP17A1 in the adrenal (Miller and Auchus, 2011) (Figure 3). Like other microsomal P450 CYP17A1 requires POR (Miller and Auchus, 2011;Pandey and Flück, 2013). Aromatase converts androgens to estrogens and also requires POR Flück et al., 2011a;Bouchoucha et al., 2014). Thus, a defective POR affects sexual development (Flück and Pandey, 2011;Fukami et al., 2013) in both male and female. Many cases still do not have final diagnosis and other genes may be involved (Flück and Pandey, in press). Recently we have reported the first human mutations in aldo-keto reductases AKR1C2 (OMIM: 614279) and AKR1C4 (OMIM: 600451) (Flück et al., 2011b). This established two distinct pathways for the biosynthesis of androgens in fetal testis for control of sexual differentiation between male and female (Flück et al., 2011b;Biason-Lauber et al., 2013a,b;Flück and Pandey, in press). POR also plays a role in this alternate pathway (Biason-Lauber et al., 2013a;Pandey and Flück, 2013) (Figure 3). A large number of POR variants containing more than 130 amino acid changes in the POR protein have now been reported (Figure 4) by multiple researchers, and a list of variants in human POR is available at CYP allele website (www.cypalleles.ki.se/por.htm) Sim et al., 2009;Flück et al., 2011a;Tomkova et al., 2012).
This review describes current status of information and research on disease related as well as naturally occurring variants of the POR gene. We have collected data on reports of PORD to summarize the current information on effects of POR mutations on enzymatic activities of redox partners. In addition, we have compiled the list of naturally occurring variants in POR gene that have been reported by researchers specifically FIGURE 1 | Role of P450 oxidoreductase in biochemical pathways. POR is required for multiple metabolic processes especially the microsomal P450 enzymes involved in metabolism of xenobiotics.
sequencing the POR gene in distinct populations. We have also included information on POR variants available in genomic databases from large scale sequencing studies. Recently many POR variants have been studied for effects on drug metabolizing enzymes. We have also summarized studies describing the impact of individual POR variants on in vitro as well as in vivo drug metabolism cytochrome P450 activities. We have labeled the POR variants found in patients with disorders of steroid metabolism as "mutants" and variants identified from normal human population as "polymorphisms" (Figure 4). List of POR variants was compiled from publication of PORD case reports, POR sequencing studies, and human genome sequencing projects available through NCBI databases (http://www.ncbi. nlm.nih.gov).

POR AND STEROID METABOLISM
PORD was identified as a DSD resulting from disruption of steroid metabolizing cytochrome P450 activities Pandey et al., 2004). After the identification of POR variants in normal human population, early reports analyzed the impact of POR variants on three microsomal cytochrome P450 involved in steroid metabolism (CYP17A1, CYP19A1, and CYP21A2) Huang et al., 2005;Dhir et al., 2007;Pandey et al., 2007). A large number of POR mutations or polymorphisms (Huang et al., 2005;Flück et al., 2011a) have been studied and analyzed based on known structures of POR (Xia et al., 2011;Pandey and Flück, 2013). Many mutations in POR found in patients are in co-factor binding sites and cause severe POR deficiency. The p.A115V variation first found in patients seems to be a polymorphism  and shows similar to wild type activity in CYP17A1 assays. The p.T142A mutation FIGURE 2 | Electron transfer from NADPH to redox partners of POR. NADPH binds to POR located into the endoplasmic reticulum, and donates electrons which are received by FAD. Electron transfer to FAD causes a conformational change that brings together the FAD and FMN domains and electrons are transferred from FAD to FMN. The FMN domain of POR interacts with the P450s and other redox partners and completes the final step of electron transfer.

POR AND DRUG METABOLISM
After the discovery of POR mutations in patients and identification several variants of POR in healthy individuals we suggested that POR variants may also impact drug metabolizing P450 enzymes since POR is necessary for activities of all microsomal P450 Pandey et al., 2004;Huang et al., 2005). Variations in POR may affect many different redox partners. Many recent studies reported effects of POR variants on several drug metabolizing P450s Hart et al., 2008;Kranendonk et al., 2008;Gomes et al., 2009;Miller et al., 2009;Oneda et al., 2009;Flück et al., 2010;Marohnic et al., 2010;Nicolo et al., 2010;Tomalik-Scharte et al., 2010) and other metabolic processes Flück et al., 2010;Marohnic et al., 2010;Nicolo et al., 2010;Pandey et al., 2010;Sandee et al., 2010;Zhang et al., 2011). Here we are describing the reports on human studies, an overview of metabolic activities in POR mouse models by laboratories of Roland Wolf and Xin-Xin Ding has been reviewed earlier (Riddick et al., 2013). Cytochrome P450 3A4 (CYP3A4) metabolizes a wide range of drugs and xenobiotics (Klein and Zanger, 2013;Meyer et al., 2013;Zanger and Schwab, 2013). POR mutations affect CYP3A4 activities in vitro (Nicolo et al., 2010). The POR mutant p.A287P (common mutation found in patients from "European" population) showed 75% loss of CYP3A4 activity (Table 2 and Figure 6) Nicolo et al., 2010). The p.R316W and p.G413S variants had close to WT activities . The mutations disrupting FAD binding (p.R457H, p.Y459H and p.V492E) resulted in complete loss of CYP3A4 activity . POR polymorphisms p.A503V and p.G504R showed close to normal activities ( Table 3) . POR

FIGURE 5 | Impact of POR variants on CYP17A1 activity. A pie chart
showing the impact of POR variants on CYP17A1 17,20 lyase activity. The variants which had <5% of WT activity are grouped as red, variants with 5-25% of WT activity are in orange, variants with 25-50% of activity are in yellow, and variants with >50% activity are in green. A large number of recently identified variants from large scale sequencing studies have not been tested in enzymatic assays and these are grouped in gray color pie. mutations p.C569Y and p.V608F resulted in 65-85% loss of CYP3A4 and p.R616X lost all activity ( Table 2) . The POR mutant p.Y181D resulted in almost total loss of CYP3A4 activity, while variant p.Q153R was normal and variant p.T142A had 75-80% activity . A larger study of POR variants with CYP3A4 suggests POR may change activities of CYP3A4 in patients Nicolo et al., 2010). There may be substrate specific effects of POR variants on CYP3A4 activities. Agrawal et al. (2010) found that POR variants p.A287P and p.R457H had lower activity with all substrates. The p.Q153R had 76-94% of WT activity with midazolam and erythromycin, but showed 129-150% activity with testosterone and quinidine . The polymorphism p.A503V had 20-40% reduction in CYP3A4 activity with testosterone and midazolam. With quinidine and erythromycin the p.A503V has no significant change in activities . Studies from us and others have reported normal CYP3A4 activity with POR p.G413S Moutinho et al., 2012). POR polymorphisms p.P228L, p.R316W, p.G413S, p.A503V, and p.G504R have 40-100% activity in most assays ( Table 3). The p.A115V had less than 40-60% of activity, and p.V631I had 23-76% of wild-type activity (Huang et al., 2005). Agrawal et al. (2008) found POR mutations p.A287P and p.R457H resulted in almost complete loss of CYP1A2 and CYP2C19 activities ( Table 2). The polymorphism p.A503V had 85% of WT activity with CYP1A2 and 113% of WT activity with CYP2C19 ( Table 3). The p.Q153R, a disease-linked mutation increased activity of CYP1A2 to 144% and CYP2C19 activity to 284% of WT.
In addition to CYP3A4 the CYP2D6 is another major drug metabolizing enzyme . POR mutants p.A287P and p.R457H have no CYP2D6 activity with synthetic substrate EOMCC (Sandee et al., 2010). The p.A287P mutant retained 25% of WT activity with dextromethorphan and bufuralol as CYP2D6 substrates. The p.Q153R had increased activity with CYP2D6. The p.A503V resulted in activities of 85% with EOMCC, 62% with dextromethorphan, and 53% with bufuralol (Sandee et al., 2010).
There may also be differences in effects of POR variants due to presence of P450 isoforms but, this has not been explored in detail. Subramanian et al. (2012) studied the polymorphic forms of CYP2C9 (CYP2C9.1, CYP2C9.2, and CYP2C9.3) with POR variants. POR variants p.A287P and p.R457H resulted in reduced activities with all three variants of CYP2C9. The disease related variant p.R153Q had higher activities with all variants of CYP2C9 (Subramanian et al., 2012).

POR VARIANTS IN DIFFERENT GENETIC POPULATIONS
A study on sequencing of POR gene in 842 healthy people  detected 140 SNPs, and 43 of those were in ≥1% of alleles. 32 variants were in protein coding regions, 2 in untranslated exon 1U, 94 in the introns and 12 in 5 flanking DNA. Among the 32 variations in protein coding region, 13 new variants resulted in amino acid change. The new variants were p.delE53, p.P55L, p.D211N, p.G213E, p.P284L, p.P284T, p.R406H, p.P452L, p.A462T, p.V472M, p.A485T, p.R600W, and p.Y607C. The POR polymorphism p.A503V (rs1057868) (Huang et al., 2005), was present in about 25% of all alleles . Most of the polymorphic variants had >40% of WT activities in different in vitro assays.
Some researchers have attempted to sequence POR gene in human liver samples to study the impact of POR variant in enzymatic studies in liver microsomes. A study of POR sequence in human liver samples by Xiao-Bo Zhong added 3 novel amino acid variations (p.K49N, p.L420M, and p.L577P) (Hart et al., 2008). The POR variant p.L577P had reduced activities in many drug metabolizing P450 assays (Hart et al., 2008). Another study of POR variation analysis by Ulli Zanger in 150 human liver samples found 43 variations with 19 SNPs in exonic regions (Gomes et al., 2009). They also found two combinatorial alleles of (p.P228L + p.A503V and p.A503V + p.V631I) and similar to previous reports, the p.A503V allele was the most common variant (Gomes et al., 2009).
Several new studies have sequenced POR gene in specific population groups. Sequencing of POR gene in 235 Japanese individuals (Saito et al., 2011) found 4 new amino acid changes (p.T29M, p.R550W, p.R570C, and p.A659T). Here p.A503V was present at a somewhat higher frequency of 0.434. A recent study sequenced POR variants in distinct Jewish populations and identified 6 novel amino acid variants, p.S102P, p.V164M, p.V191M, p.D344N, p.E398A, and p.D648N (Tomkova et al., 2012). The average minor allele frequency (MAF) of p.A503V in different reports have ranged from 0.25 to 0.45. Analysis of genomic sequence data available at NCBI SNP database shows that there

Frontiers in Pharmacology | Pharmacogenetics and Pharmacogenomics
May 2014 | Volume 5 | Article 103 | 6 Activities are shown as percentage of the wild-type POR. POR, P450 oxidoreductase. Data were compiled from previous reports (Flück et al., , 2011aPandey et al., 2004Pandey et al., , 2007Pandey et al., , 2010Huang et al., 2005Huang et al., , 2008Agrawal et al., 2008Agrawal et al., , 2010Nicolo et al., 2010;Sandee et al., 2010). # DBSNP id not available, these were identified in a sequencing study by Bioventures Inc. (Solus et al., 2004) and are represented as Bioventure ID numbers BVxxxxx. are some differences among different genetic groups (Figure 7). In the Caucasian and Hispanic populations the p.A503V allele frequency is 31%, in Pacific Islanders 34%, and in Asian populations the p.A503V allele is present at 38%. In the Japanese population the p.A503V frequency is highest at 40% while the African Americans group has the lowest presence at less than 15% of all alleles. Homozygous presence of p.G5G (rs10262966, nucleotide change GGA to GGG) silent mutation has been associated with increased risk in African Americans (Haiman et al., 2007). This variant is not present in European and Asian populations and has a MAF of 0.023 in Hispanic population (Figure 8). In Sub-Saharan Africans it is present in 30% of alleles (MAF 0.297) and African Americans have the highest frequency with 40% of alleles (MAF 0.400) carrying the p.G5G variant of POR. The p.A50V variant was also tested for association with increased breast cancer risk in same study but no linkage was observed (Haiman et al., 2007).

IN VIVO DRUG CLEARANCE STUDIES EXAMINING THE IMPACT OF POR VARIANTS
Several studies have now examined the role of POR variants on in vivo drug clearance using many different standard drugs as markers. One report has described in vivo drug metabolism activities in a POR p.A287P homozygous patient and her heterozygous mother (Tomalik-Scharte et al., 2010). Genotyping of P450 enzymes predicted normal to high P450 activities in both the patient and her mother. The in vivo results showed subnormal activities of CYP1A2, CYP2C9, CYP2D6, and CYP3A4 in the patient. The heterozygous mother had reduced CYP1A2 and CYP2C9 activities. These data are in agreement with disruption of in vitro CYP3A4 activities by p.A287P variant of POR reported by us Nicolo et al., 2010). In patients with PORD, the in vitro and in vivo activities of POR dependent P450 may be taken into account to modify drug dosages and supplementation with steroids. POR variant p.A503V (POR * 28) had higher levels of CYP3A activities based on midazolam clearance (Oneda et al., 2009). Yang et al have studied the effect of POR variant p.A503V on in vivo CYP3A activity in healthy Chinese men (Yang et al., 2011). Out of 73 subjects (CC 21, TT11; CT 41) the midazolam metabolite ratio was greater in the TT group compared with carriers of the C allele. The p.A503V variant of POR had increased 1hydroxylation of midazolam. Further analysis showed association with hepatic but not intestinal CYP3A activities (Yang et al., 2011). There was no significant differences in combined hepatic plus intestinal CYP3A activity.
A study tested the impact of POR p.A503V on metabolism of tacrolimus (de Jonge et al., 2011). They found lower levels of tacrolimus in carriers of p.A503V allele but in the individuals with CYP3A5 * 3/ * 3 allele, the p.A503V POR had no effect (de Jonge et al., 2011). Another study measured the impact of p.A503V (rs1057868, DNA change GCC to GTC) allele of POR on metabolism of tacrolimus in healthy Chinese men . There was no difference in tacrolimus metabolism between two POR alleles. Further analysis showed no significant differences in tacrolimus pharmacokinetics in CYP3A5 non-expressers (CYP3A5 * 3/ * 3). The mean tacrolimus exposure for the p.A503V CC homozygotes in CYP3A5 expressers (CYP3A5 * 1/ * 1 or * 1/ * 3) were much higher than the p.A503V CT heterozygotes. These studies suggest a role for p.A503V allele of POR in the variability of tacrolimus exposure levels. In vitro studies of CYP3A5 activities comparing normal and p.A503V variants of POR may confirm this hypothesis.
A CYP1A2 genotyping study in smokers also checked POR polymorphic variants for potential linkage to metabolic defects (Dobrinas et al., 2012). During smoking there was no impact of any POR polymorphisms on CYP1A2 activity. CYP1A2 activity increased after smoking cessation in POR rs2302429A and rs1057868T (p.A503V) carriers. The carriers of POR rs2286823A and of rs17148944G-rs10239977C-rs3815455C-rs2286823A-rs2302429G-rs1057868C (503A) had decreased CYP1A2 activity.

CONCLUSIONS
PORD is more complex than any single enzyme or protein deficiency (Janner et al., 2006;Flück et al., 2011c;Camats et al., 2012). It affects drug and xenobiotic metabolism pathways (Henderson et al., 2006;Flück et al., 2010;Nicolo et al., 2010). Variants of POR have different interactions with different redox partners. Effect of POR variants on drug metabolizing P450 enzymes needs detailed studies. Deposition of heme in liver lowers cytochrome P450 levels and enzymatic activities (Pandey et al., 1995(Pandey et al., , 1996.  (rs10262966) is associated with increased risk of breast cancer when present in homozygous form in African Americans (Haiman et al., 2007). Genomic data from various sequencing projects show a clear pattern for the presence of rs10262966 in different populations. In the Caucasian and Asian populations rs10262966 is not present, while in Hispanic population it is present with a minor allele frequency of 0.023. In the Sub-Saharan African population the rs10262966 frequency is higher at 0.297 while in African Americans it is present in 40% of all alleles. Data were compiled from information available in NCBI SNP database as of April 2014 (http://www.ncbi.nlm.nih.gov/SNP/snp_ref.cgi?rs=10262966).
POR variants with lower activities may complicate these symptoms and result in a more severe disease. Heme oxygenase plays a major role in limiting the severity of malaria and sepsis infections (Pamplona et al., 2007;Larsen et al., 2010;Pandey et al., 2010;Ferreira et al., 2011). There may be potential links between the common POR variants and pathogenesis of infections. External flavin supplementation reverts loss of activity of some POR variants (Nicolo et al., 2010). An interesting possibility to explore in future experiments is whether external FMN can receive electrons from bound NADPH in POR as reported for a B. subtilis reductase (Deller et al., 2006). Whether flavin treatment may help patients with POR deficiency remains untested in clinical settings.