Efavirenz Metabolism: Influence of Polymorphic CYP2B6 Variants and Stereochemistry

Abstract

Efavirenz (more specifically the S-enantiomer) is a cornerstone antiretroviral therapy for treatment of HIV infection. The major primary metabolite is S-8-hydroxyefavirenz, which does not have antiretroviral activity but is neurotoxic. Cytochrome P450 2B6 (CYP2B6) is the major enzyme catalyzing S-8-hydroxyefavirenz formation. CYP2B6 genetics and drug interactions are major determinants of clinical efavirenz disposition and dose adjustment. In addition, as a prototypic CYP2B6 substrate, S-efavirenz and analogs can inform on the structure, activity, catalytic mechanisms, and stereoselectivity of CYP2B6. Metabolism of R-efavirenz by CYP2B6 remains unexplored. This investigation assessed S-efavirenz metabolism by clinically relevant CYP2B6 genetic variants. This investigation also evaluated R-efavirenz hydroxylation by wild-type CYP2B6.1 and CYP2B6 variants. S-Efavirenz 8-hydroxylation by wild-type CYP2B6.1 and variants exhibited positive cooperativity and apparent cooperative substrate inhibition. On the basis of Cl_max values, relative activities for S-efavirenz 8-hydroxylation were in the order CYP2B6.4 > CYP2B6.1 ≈ CYP2B6.5 ≈ CYP2B6.17 > CYP2B6.6 ≈ CYP2B6.7 ≈ CYP2B6.9 ≈ CYP2B6.19 ≈ CYP2B6.26; CYP2B6.16 and CYP2B6.18 showed minimal activity. Rates of R-efavirenz metabolism were approximately 1/10 those of S-efavirenz for wild-type CYP2B6.1 and variants. On the basis of Cl_max values, there was 14-fold enantioselectivity (S > R-efavirenz) for wild-type CYP2B6.1, and 5- to 22-fold differences for other CYP2B6 variants. These results show that both CYP2B6 516G > T (CYP2B6*6 and CYP2B6*9) and 983T > C (CYP2B6*16 and CYP2B6*18) polymorphisms cause canonical diminishment or loss-of-function variants for S-efavirenz 8-hydroxylation, provide a mechanistic basis for known clinical pharmacogenetic differences in efavirenz disposition, and may predict additional clinically important variant alleles. Efavirenz is the most stereoselective CYP2B6 drug substrate yet identified and may be a useful probe for the CYP2B6 active site and catalytic mechanisms.

SIGNIFICANCE STATEMENT

Clinical disposition of the antiretroviral S-efavirenz is affected by CYP2B6 polymorphisms. Expressed CYP2B6 with 516G>T (CYP2B6*6 and CYP2B6*9), and 983T>C (CYP2B6*16 and CYP2B6*18) polymorphisms had a diminishment or loss of function for efavirenz 8-hydroxylation. This provides a mechanistic basis for efavirenz clinical pharmacogenetics and may predict additional clinically important variant alleles. Efavirenz metabolism showed both cooperativity and cooperative substrate inhibition. With greater than 10-fold enantioselectivity (S- vs. R- metabolism), efavirenz is the most stereoselective CYP2B6 drug substrate yet identified. These findings may provide mechanistic insights.

Introduction

Efavirenz [(S)-6-chloro-4-(cyclopropylethynyl)-1,4-dihydro-4-(trifluoromethyl)-2H-3,1-benzoxazin-2-one] is a non-nucleoside reverse transcriptase inhibitor used as first-line therapy for human immunodeficiency virus (HIV) infection (Rakhmanina and van den Anker, 2010). One essential step in HIV replication is viral single-strand RNA conversion into double-strand DNA, catalyzed by viral reverse transcriptase, followed by viral DNA integration into the host genome. HIV reverse transcriptase has a catalytic p66 (66-kDa) subunit and a smaller p51 (55-kDa) subunit that functions mainly for structural support. The p66 subunit is further divided into N-terminal polymerase domain, which catalyzes complementary DNA polymerization from template RNA, and C-terminal RNase H domain, which digests viral RNA and removes RNA primers during DNA synthesis. Efavirenz binds to a hydrophobic pocket in the p66 polymerase domain about 10 Å from the active site and inhibits activity via an allosteric mechanism (Schauer et al., 2014). Shortly after synthesis of RS-6-chloro-4-(cyclopropylethynyl)-1,4-dihydro-4-(trifluoromethyl)-2H-3,1-benzoxazin-2-one, it was identified that reverse transcriptase inhibition was highly stereospecific, as the R-enantiomer (R)-6-chloro-4-(cyclopropylethynyl)-1,4-dihydro-4-(trifluoromethyl)-2H-3,1-benzoxazin-2-one (henceforth referred to as R-efavirenz) was inactive (Young et al., 1995), and all further drug development proceeded with the single S-enantiomer (henceforth referred to as S-efavirenz).

S-Efavirenz is extensively metabolized by cytochrome P450 enzymes (Scheme 1). The major primary metabolite is S-8-hydroxyefavirenz, both in vitro and in vivo, and a minor primary metabolite is S-7-hydroxyefavirenz (Ward et al., 2003; Desta et al., 2007). Secondary metabolites include 8,14-dihydroxyefavirenz and 7,8-dihydroxyefavirenz (Ogburn et al., 2010; Avery et al., 2013). These metabolites are devoid of significant pharmacologic activity toward HIV-1 (Avery et al., 2013). Nevertheless, they are not inert, as S-8-hydroxyefavirenz has been associated with clinical neurotoxicity (Decloedt et al., 2015) and was at least an order of magnitude more neurotoxic than S-efavirenz or S-7-hydroxyefavirenz in vitro (Tovar-y-Romo et al., 2012). Cytochrome P450 2B6 (CYP2B6) is the major enzyme catalyzing S-8-hydroxyefavirenz and thence 8,14-dihydroxyefavirenz formation, whereas CYP2A6 is responsible for 7-hydroxylation (Ward et al., 2003; Damle et al., 2008). CYP2B6 is a major determinant of clinical efavirenz metabolism and elimination; drug interactions resulting from CYP2B6 inhibition increase efavirenz exposure (Damle et al., 2008; Desta et al., 2016); and diminished CYP2B6 activity unmasks the influence of CYP2A6 on efavirenz exposure (di Iulio et al., 2009).

Scheme 1.

Metabolism of S-efavirenz catalyzed by CYP2B6.

The CYP2B6 gene is highly polymorphic (Zanger and Klein, 2013), with at least 38 allelic variants described (https://www.pharmvar.org/gene/CYP2B6), of which 25 are considered important and eight are common in at least one racial/ethnic population (Zhou et al., 2017). CYP2B6 metabolizes a broad range of substrates, constituting nearly 8% of marketed drugs (Nolan et al., 2006), although the relative contribution of CYP2B6 to total hepatic P450 content is small. In addition to efavirenz, clinically important CYP2B6 substrates include methadone, bupropion, ketamine, cyclophosphamide, and artemisinin.

The pharmacogenetics of efavirenz disposition has been comprehensively reviewed (Čolić et al., 2015; Sinxadi et al., 2015; Russo et al., 2016). The CYP2B6 516G>T polymorphism, alone constituting CYP2B6*9 or together with 785A>G constituting CYP2B6*6, is a canonical loss-of-function variant that was the first and most studied and is consistently associated with increased efavirenz exposure and reduced clearance and metabolism (Haas et al., 2004; Tsuchiya et al., 2004; Rotger et al., 2005). Efavirenz clearance is approximately 25% and 50% lower in 516GT and 516TT carriers, respectively (Čolić et al., 2015; Robarge et al., 2016). The less common CYP2B6 983T>C polymorphism, alone constituting CYP2B6*18 or together with 785A>G constituting CYP2B6*16, is also associated with increased efavirenz exposure (Wyen et al., 2008; Dhoro et al., 2015; Röhrich et al., 2016). The 516G>T, 785A>G, and 983T>C polymorphisms are more common in African than Caucasian populations, and the lattermost is considered essentially Africa-specific (Čolić et al., 2015; Russo et al., 2016). In Africans or African-Americans, CYP2B6*6/*6 and CYP2B6*6/*18 genotypes had the highest single-dose (Haas et al., 2009) or steady-state efavirenz concentrations (3- to 4-fold higher than CYP2B6*1/*1) (Maimbo et al., 2012). CYP2B6*6, *9, *16, and *18 constitute a poor-metabolizer phenotype (Čolić et al., 2015; Russo et al., 2016). Efavirenz adverse effects in general and adverse neurologic and neuropsychiatric effects in particular (e.g., neurocognitive impairment, depression, suicidality) have been associated with higher plasma efavirenz exposure, slow efavirenz metabolizer phenotype, or 516G>T and/or 983T>C polymorphisms (Haas et al., 2004; Rotger et al., 2005; Apostolova et al., 2015; Vo and Varghese Gupta, 2016; Gallien et al., 2017; Mollan et al., 2017; Chang et al., 2018). In contrast, 785A>G alone (CYP2B6*4) codes for a protein with increased efavirenz hydroxylation in vitro (Bumpus et al., 2006), but the clinical significance is ambiguous (Russo et al., 2016). CYP2B6 genetically guided efavirenz dosing has been evaluated and recommended (Gatanaga et al., 2007; Mukonzo et al., 2014; Vo and Varghese Gupta, 2016). Other CYP2B6 variant activity and clinical implications for efavirenz disposition are less well characterized. Furthermore, and surprisingly, excepting CYP2B6.4 and CYP2B6.6 (Bumpus et al., 2006; Ariyoshi et al., 2011; Zhang et al., 2011; Xu et al., 2012; Radloff et al., 2013), comparatively less is known about S-efavirenz metabolism by CYP2B6 variants in vitro than in vivo. Therefore, the first purpose of this investigation was to assess S-efavirenz metabolism by clinically relevant CYP2B6 variants, coexpressed with P450 oxidoreductase and cytochrome b₅ in a fully catalytically competent system.

CYP2B6 is pharmacologically and clinically relevant, and several CYP2B6 substrates are chiral, with varying degrees of enantioselective metabolism, and enantioselectivity may vary among different CYP2B6 variants (Wang et al., 2018). As a prototypic CYP2B6 substrate, S-efavirenz and analogs have been used to inform on the structure, activity, and catalytic mechanism of wild-type CYP2B6 (Bumpus and Hollenberg, 2008; Cox and Bumpus, 2014; Cox and Bumpus, 2016; Shah et al., 2018) and variants such as CYP2B6.4 (Bumpus et al., 2005). These compounds, together with molecular modeling, have provided insights into the active site configuration of CYP2B6. In this regard, the metabolism of R-efavirenz by CYP2B6, and by CYP2B6 variants, remains unexplored. Therefore, the second purpose of this investigation was to evaluate the metabolism of R-efavirenz by wild-type CYP2B6.1 and CYP2B6 variants, potentially to inform on CYP2B6 active-site character or activity.

Materials and Methods

Materials.

S-Efavirenz was purchased from TCI America (Portland, OR). R-Efavirenz was purchased from Carbosynth US (San Diego, CA). The standards of rac 7-hydroxyefavirenz-d4, rac-8-hydroxyefavirenz, rac-7-hydroxyefavirenz, and rac-8,14-dihydroxyefavirenz were purchased from Toronto Research Chemicals (TRC, Toronto, ON, Canada). Spodoptera frugiperda (Sf9) cells and Sf-900 III SFM culture media were purchased from ThermoFisher (Waltham, MA). Trichoplusia ni cells and ESF AF culture media were from Expression Systems (Davis, CA). β-NADP, glucose-6-phosphate, and glucose-6-phosphate dehydrogenase were purchased from Sigma Aldrich (St. Louis, MO).

Generation of Baculovirus Constructs.

The production of recombinant proteins of CYP2B6 variants (Table 1), wild-type P450 reductase (POR), and cytochrome b₅ were carried out as described previously (Wang et al., 2018). Briefly, the human genes of CYP2B6, POR, and b₅ were amplified from the Human Liver Quick-Clone cDNA library (Clonetech, Mountain View, CA) and inserted individually into the transfer vector pVL1393 using the In-Fusion HD Cloning system (Clontech). The plasmid carrying the gene of wild-type CYP2B6 was used as the template, and the polymorphic variants of CYP2B6 were generated by site-directed mutagenesis using Quik-Change II XL Site-Directed Mutagenesis Kit (Agilent, Santa Clara, CA), as described (Wang et al., 2018). BestBac 2.0 Baculovirus Cotransfection Kit (Expression Systems) was used for production of recombinant baculovirus. Sf9 insect cells were cotransfected with BestBac linearized DNA and the plasmid DNA of transfer vector carrying the gene of interest on a six-well plate to produce p0 generation of recombinant baculovirus. Sf9 cells in suspension culture were infected with p0 to make subsequent viral generation. The maximal viral generation is limited to p3 to minimized undesirable variants owing to successive amplification. All viral titers were determined using the BacPAK Baculovirus Rapid Titer Kit (Clontech).

TABLE 1.

CYP2B6 variants

CYP2B6 Allele	Variant	cDNA Sequence Mutation	Protein Sequence Mutationa	Allele Frequency (%)
CYP2B6*1		Wild-type	Wild-type
CYP2B6*4	rs2279343	785A>G	K262R	2-4 Ca
CYP2B6*5	rs3211371	1459C>T	R487C	12 Ca
CYP2B6*6	rs3745274, rs2279343	516G>T, 785A>G	Q172H/K262R	33 Af, 28 Ca
CYP2B6*7	rs3745274, rs2279343, rs3211371	515G>T, 785A>G, 1459C>T	Q172H/K262R/R487C	3 Ca
CYP2B6*9	rs3745274	516G>T	Q172H
CYP2B6*16	rs2279343, rs28399499	785A>G, 983T>C	K262R/I328T	6.9 Af
CYP2B6*17	rs33973337, rs33980385, rs33926104	76A>T, 83A>G, 85C>A, 86G>C	T26S/D28G/R29T	6.3 Af
CYP2B6*18	rs28399499	983T>C	I328T	9.4 Af
CYP2B6*19	rs34826503	516G>T, 785A>G, 1006C>T	Q172H/K262R/R336C	1.6 Af
CYP2B6*26	rs3826711, rs2279343, rs3745274	499C>G, 516G>T, 785A>G	P167A/Q172H/K262R	1.3 As

Af, African; As, Asian; Ca, Caucasian.

^aAll CYP2B6 variants result in missense mutations.

Expression of Recombinant Proteins in Insect Cells.

Protein expression was performed as described previously (Wang et al., 2018). All CYP2B6 variants were coexpressed with redox partners P450 reductase (POR) and cytochrome b₅ in insect cells by triple infection. Briefly, Trichoplusia ni cells in suspension culture in early log phase growth were infected with the recombinant baculoviruses carrying the genes of CYP2B6, POR and b₅, with the multiplicities of infection at ratio of 4:2:1 (CYP2B6/POR/b₅), in the presence of heme precursors 100 μM δ-aminolevulinic acid and 100 μM ferric citrate. After 48–72 hours growth post-infection, cells were harvested by centrifugation for 15 minutes at 3000g, and washed two times with phosphate-buffered saline followed by centrifugation in each wash step. The cell pellets were resuspended in 100 mM potassium phosphate buffer at pH 7.4, and stored frozen at −80°C. Frozen cells were thawed and lysed on ice in a Potter-Elvehjem tissue homogenizers. Full cell disruption was achieved by the combination of one freeze-thaw cycle and 10 strokes in the glass-Teflon Potter-Elvehjem pestle. Aliquots of 0.5 ml of homogenized cells were stored at −80°C.

P450 content, b₅ content, and POR activity were measured as described previously (Wang et al., 2018). Total protein concentrations were determined using Bio-Rad Protein Assay Dye Regent Concentrate, the basis of which is the Bradford method. P450 concentration was determined by difference spectrum of ferrous-carbon monoxide complex in a CO binding assay using an extinction coefficient Δε_{450–490 nm} of 91 mM⁻¹cm⁻¹. Cytochrome b₅ content was determined by difference spectrum of NADH-reduced and oxidized b₅ using an extinction coefficient Δε_{424–410 nm} of 185 mM⁻¹cm⁻¹. POR activity was measured in an NADPH-cytochrome c reductase activity assay, and the reaction rate was calculated using an extinction coefficient of ε_{550 nm} of 21 mM⁻¹cm⁻¹ for reduced cytochrome c. The POR activity was converted to POR content assuming that 3000 nmol of cytochrome c are reduced per minute per nanomoles POR at 23°C (Guengerich et al., 2009).

Efavirenz Metabolism.

Incubations were carried out in 96-well PCR plates with raised wells. The procedure used in this study was adapted with modifications from published protocols (Ward et al., 2003; Avery et al., 2013). Positive displacement pipettes were used in steps in which organic solvents were involved. Twenty-millimolar stock solutions of S- and R-efavirenz were prepared in 100% methanol. Substock solutions of efavirenz at 10 mM in 50% methanol, 2 mM in 25% methanol, and 200, 50, 10 μM in 10% methanol were prepared by dilution from the 20 mM stock. To a 96-well plate, efavirenz was added from the stock and substock solutions to 100 mM potassium phosphate buffer at pH7.4 containing CYP2B6/POR/b₅ proteins. The final CYP2B6 concentration was 25 pmol/ml. S-efavirenz concentrations were 0, 0.25, 0.5, 1.25, 2.5, 5, 10, 20, 40, 70, and 100 μM and R-efavirenz concentrations were 0.11–45 μM, owing to limited solubility and availability. The total reaction volume was 200 μl and the methanol concentration was controlled at 0.5% for every incubation. After preincubation for 5 minutes at 37°C, the reaction was initiated by adding an NADPH-regenerating system (final concentrations: 10 mM glucose 6-phosphate, 1 mM β-NADP, 1 IU/ml glucose-6-phosphate dehydrogenase, and 5 mM magnesium chloride, preincubated at 37°C for 10 minutes). The reaction was allowed to proceed for 20 minutes at 37°C, then terminated by withdrawing 100 μl of reaction mixture and mixing with 200 μl of ice-cold acetonitrile containing 32 ng/ml internal standard rac-7-hydroxyefavirenz-d4 in glass tubes (16 × 125 mm). The metabolite products were extracted using a liquid/liquid extraction method as described previously (Avery et al., 2013) with modifications. Three hundred microliters of 50 mM ammonium formate was added to the quenched reaction mixture, followed by extraction with 1.0 ml of hexane/ethyl acetate (1:1). All samples were vortex-mixed for 30 seconds and centrifuged at 2500 rpm for 5 minutes. Six hundred, twenty-five microliters of organic layer was transferred to another clean glass tube (13 × 100 mm) and evaporated under nitrogen to dryness at 30°C using Turbo Vap LV Evaporator (Zymark, Hopkinton, MA). For high-performance liquid chromatography/mass spectrometry (HPLC/MS) analysis, the residues of the samples were reconstituted in 200 μl of 0.05% formic acid in 50% acetonitrile.

Analysis of Efavirenz Metabolites by HPLC/Tandem Mass Spectrometry.

Calibration samples were prepared using standards of rac-8-hydroxyefavirenz, rac-7-hydroxyefavirenz, and rac-8,14-dihydroxyefavirenz, with all three analytes at identical concentrations of 0, 0.5, 1, 2.5, 5, 10, 25, 50, 100, 250, and 500 ng/ml in 100 mM potassium phosphate buffer (pH 7.4) containing 33% methanol. Calibrators were processed identically to incubation samples.

LC-MS/MS analysis was performed on a Shimadzu HPLC system composed of two LC-20AD XR pumps, DGU20A5R degasser, CBM-20A system controller, CTO-20C column oven, FCV-11AL solvent selection valve, and a SIL-20AC XR temperature-regulated autosampler. The LC system was coupled to an API6500 triple quadrupole tandem mass spectrometer (Applied Biosystems/MDS Sciex, Foster City, CA) operated with Analyst 1.6.2. software. MultiQuant 3.0.1 (AB Sciex) was used for peak integration, generation of calibration curves, and data analysis. Efavirenz metabolites were analyzed utilizing a Kinetex XB-C18 100A column (100 × 2.1 mm, 2.6 μm; Phenomenex) equipped with a Security Guard ULTRA Cartridge for C18 UHPLC (2 × 2.1 mm; Phenomenex). A 0.25-μm inline filter was additionally added prior to the sample entering the column. The column oven was at ambient temperature and the autosampler was at 4°C. The mobile phase A was 0.1% formic acid in Milli-Q water and mobile phase B was 0.1% formic acid in acetonitrile. Chromatographic separation was achieved using an isocratic condition of 50% mobile phase A and 50% mobile phase B and a flow rate of 0.15 ml/min. The injection volume was 5 μl and total run time is 12 minutes. Under these conditions, the approximate retention time was 6.5 minutes for 8-hydroxyefavirenz, 5.7 minutes for 7-hydroxyefavirenz, 5.6 minutes for 7-hydroxyefavirenz-d4, and 3.1 minutes for 7,8-dihydroxyefavirenz and 8,14-dihydroxyefavirenz. The mass spectrometer was operated with a turbo spray ion source in the negative mode with multiple reaction monitoring (MRM). Analytes were detected with the following MRM transitions: m/z 329.9 > 257.8 for 8-hydroxyefavirenz and 7-hydroxyefavirenz, m/z 346.0 > 274.0 for 7,8-dihydroxyefavirenz, and m/z 345.9 > 262.1 for 8,14-dihydroxyefavirenz.

Data Analysis.

Formation of 8-hydroxyefavirenz by enzyme variants at fixed concentrations was analyzed by analysis of variance with post-hoc Dunnett’s test (SigmaPlot 12.5; Systat). Results are the mean ± S.D.

8-Hydroxyefavirenz formation versus substrate concentration data were analyzed by nonlinear regression analysis. Results are the parameter estimate ± S.E. of the estimate. Metabolism of both efavirenz enantiomers at low concentrations (up to 40–45 μM) exhibited homotropic positive cooperativity. At higher concentrations of S-efavirenz there was evidence of substrate inhibition. Two approaches were used, depending on the substrate concentrations modeled.

Metabolism over the substrate concentration range 0.25–40 μM S-efavirenz and 0.11–45 μM R-efavirenz was analyzed using an allosteric model of the Hill equation (eq. 1), where [S] is the substrate efavirenz concentration, n is the Hill coefficient, and S₅₀ represents the substrate concentration at which the reaction reached half-maximal velocity.

(1)

R-efavirenz 8-hydroxylation by CYP2B6.19 showed weak substrate cooperativity. Therefore data were also analyzed using the Michaelis-Menten equation. R-efavirenz 8-hydroxylation by CYP2B6.19 also had high K_m and values relative to the substrate range. Thus it was also analyzed by linear regression. Specifically, when substrate concentrations are far below K_m, the observed rate versus [S] approaches a linear function and the Michaelis-Menten equation can be simplified to v = (V_max/K_m)*[S], and the slope of the v versus [S] plot represents an estimate of V_max/K_m. R-Efavirenz 8-hydroxylation by CYP2B6.4 showed substrate inhibition. Therefore data were also analyzed using the LiCata model (vide infra).

S-Efavirenz 8-hydroxylation over the substrate concentration range 0.25–100 μM exhibited both homotropic positive cooperativity and substrate inhibition for all active CYP2B6 variants. For R-efavirenz hydroxylation, only CYP2B6.4 showed substrate inhibition. Several models were evaluated for fitting positive cooperativity and substrate inhibition, including (i) a combination of the Hill equation with substrate inhibition (Müller et al., 2015), (ii) homotropic cooperativity with complete substrate inhibition and a single substrate molecule binding to an inhibitory site (a simplified version of the LiCata model, below) (Kapelyukh et al., 2008), (iii) cooperative catalysis and substrate inhibition (Pastra-Landis et al., 1978), (iv) substrate inhibition analogous to uncompetitive inhibition (Michaelis-Menten plus substrate inhibition) with facilitated sequential binding of two additional substrate (inhibitor) molecules (Bapiro et al., 2018), (v) uncompetitive inhibition (Michaelis-Menten plus substrate inhibition) assuming simultaneous binding of n molecules (determined from the data) of the substrate to the inhibitory site (Bapiro et al., 2018), and (vi) a modified Hill equation with cooperative substrate binding, substrate inhibition, and cooperative inhibitor binding (eq. 2) (LiCata and Allewell, 1997). In this model, K is the substrate dissociation constant, K_i is the inhibitor dissociation constant, V_i is the final velocity at infinite substrate concentration, and x represents a second Hill coefficient that allows for cooperativity of inhibitor substrate binding. To obtain convergence, the value of x must be fixed. The integer value for x that gave the best fit (x = 3) was determined empirically.

(2)

Analysis using models i–iv did not produce acceptable fits or did not converge at all. Model v gave some acceptable fits and parameters but some required constraint on K_m < K_i, and the model produced some unrealistic n and very high K_m and V_max values. Model vi achieved the best fits to the S-efavirenz metabolism data, on the basis of F values and parameter estimates and error variances, and was the final model chosen.

Because the in vitro intrinsic clearance parameter V_max/K_m is suitable only for reactions following Michaelis-Menten kinetics, data were further analyzed using Cl_max, the maximal clearance (eq. 3) (Houston and Kenworthy, 2000):

(3)

Results

8-Hydroxyefavirenz was the predominant metabolite of wild-type CYP2B6-catalyzed S-efavirenz metabolism, as expected (Fig. 1A). Very low amounts of 8,14-dihydroxyefavirenz were formed. Neither S-7-hydroxyefavirenz nor 7,8-dihydroxyefavirenz were detected. Two aspects of S-efavirenz 8-hydroxylation by CYP2B6.1 are notable. First, S-efavirenz hydroxylation was maximal at substrate concentrations of 20–40 μM, and higher substrate concentrations resulted in substantially less 8-hydroxyefavirenz formation (Fig. 1B). The diminished formation at high substrate concentrations was not the result of facile secondary metabolism of 8-hydroxyefavirenz to 8,14-dihydroxyefavirenz. Indeed, rates of secondary 14-hydroxylation of 8-hydroxyefavirenz were generally low compared with primary metabolism (Fig. 1; Table 2). However at low S-efavirenz concentrations (0.25–1.25 μM), 8,14-dihydroxyefavirenz formation was predominant, representing 70%–100% of total substrate metabolism. In contrast, at 20 μM S-efavirenz, 8-hydroxyefavirenz was 95% of product formation and only 5% was converted to 8,14-dihydroxyefavirenz. Thus at low S-efavirenz concentrations, secondary metabolism of 8-hydroxyefavirenz to 8,14-dihydroxyeavirenz was facile but was inhibited at higher substrate concentrations. These observations are consistent with substrate inhibition of both primary and secondary CYP2B6.1-catalyzed hydroxylation. The second notable aspect of S-efavirenz hydroxylation was the atypical kinetics. Metabolism by CYP2B6.1 at low substrate concentrations deviated from standard Michaelis-Menten hyperbolic kinetics and instead showed a sigmoidal pattern suggesting cooperativity consistent with multiple substrate binding sites (Fig. 1A). The Eadie-Hofstee plot showed curvature indicative of such cooperativity (Fig. 1A, inset). At high substrate concentrations there was substrate inhibition (Fig. 1B). The Eadie-Hofstee plot showed a circular pattern, consistent with both cooperativity and substrate inhibition.

Fig. 1.

Primary and secondary metabolism of S-efavirenz catalyzed by coexpressed wild-type CYP2B6.1, P450 oxidoreductase, and cytochrome b₅. Formation of 8-hydroxyefavirenz (8-OH-EFV) (●) and 8,14-dihydroxyefavirenz (8,14-diOH-EFV) (▲). Results are the mean ± S.D. of triplicate determinations. (A) Metabolism over the substrate range 0.25–40 μM. The solid line represents predicted concentrations on the basis of parameters from nonlinear regression using the Hill equation. The dotted line represents predicted concentrations on the basis of parameters from nonlinear regression using the Michaelis-Menten equation. The inset shows an Eadie-Hofstee plot for 8-hydroxyefavirenz formation. The solid line represents predicted values on the basis of parameters from nonlinear regression using the Hill equation. (B) Metabolism over the substrate range 0.25–100 μM, showing substrate inhibition. The solid line represents predicted concentrations on the basis of parameters from analysis using cooperative substrate binding and substrate inhibition with cooperativity in the inhibitory mode (LiCata model). The inset shows an Eadie-Hofstee plot for 8-hydroxyefavirenz formation. The solid line represents predicted concentrations on the basis of parameters from nonlinear regression using the LiCata model.

TABLE 2.

8-Hydroxyefavirenz and 8,14-dihydroxyefavirenz formation in low substrate concentration range

S-Efavirenz	8-Hydroxyefavirenz	8,14-diOH-efavirenz
µM	pmol/min per picomoles	pmol/min per picomoles	%	%
0.25	0	0.014 ± 0.010	0	100
0.5	0.001 ± 0.001	0.055 ± 0.046	2	98
1.25	0.055 ± 0.009	0.209 ± 0.033	21	79
2.5	0.332 ± 0.048	0.460 ± 0.182	42	58
5.0	1.46 ± 0.29	0.405 ± 0.078	78	22
10	2.43 ± 0.10	0.505 ± 0.144	83	17
20	4.03 ± 0.28	0.201 ± 0.095	95	5

S-Efavirenz 8-hydroxylation at therapeutic (5–10 μM steady-state) substrate concentrations catalyzed by coexpressed CYP2B6 (wild-type and variants), wild-type POR, and cytochrome b₅ is shown in Fig. 2. CYP2B6.6, CYP2B6.7, CYP2B6.9, CYP2B6.19, and CYP2B6.26 had diminished activity compared with CYP2B6.1, and CYP2B6.16 and CYP2B6.18 were essentially catalytically inactive. In contrast, CYP2B6.4 had higher activity than CYP2B6.1. Results at lower substrate concentrations showed comparatively little difference between variants, which may reflect differences in substrate cooperativity. At clinically relevant concentrations, hydroxylation rates were of the order CYP2B6.4 > CYP2B6.1 ≈ CYP2B6.5 ≈ CYP2B6.17 > CYP2B6.6 ≈ CYP2B6.7 ≈ CYP2B6.9 ≈ CYP2B6.19 ≈ CYP2B6.26 > > CYP2B6.16 and CYP2B6.18. For all CYP2B6 variants, 8,14-dihydroxefavirenz formation was less than by CYP2B6.1 (not shown).

Fig. 2.

Metabolism of S-efavirenz and R-efavirenz to 8-hydroxyefavirenz at therapeutic concentrations. Asterisks denote rates significantly different from wild-type (P < 0.05). Not shown are results for CYP2B6.16 and CYP2B6.18, which had negligible activity.

Concentration dependence of S-8-hydroxyefavirenz formation is shown in Fig. 3 for CYP2B6 variants, and kinetic parameters are provided in Table 3, for 0.25–40 μM S-efavirenz. Most variants had a sigmoidal curve indicating cooperativity, and data were analyzed by fitting the Hill equation. This was most apparent for CYP2B6.4 and CYP2B7.17, similar to wild-type CYP2B6.1. Hill coefficients (n), representing intensity of the cooperativity, varied from 1.4 to 2.5. Conversely, CYP2B6.9 data were hyperbolic, and regression analysis generated similar results from fitting either Hill or Michaelis-Menten equations, and n was close to 1, suggesting the absence of cooperative substrate binding. Differences in activity between CYP2B6 variants were the result of differences in V_max, which varied approximately 3-fold, and which varied approximately 2-fold. On the basis of the Cl_max values, relative activities for S-efavirenz 8-hydroxylation were in the order CYP2B6.4 > CYP2B6.1 ≈ CYP2B6.5 ≈ CYP2B6.17 > CYP2B6.6 ≈ CYP2B6.7 ≈ CYP2B6.9 ≈ CYP2B6.19 ≈ CYP2B6.26 > > CYP2B6.16 and CYP2B6.18.

Fig. 3.

8-Hydroxyefavirenz formation from S-efavirenz (●) and R-efavirenz (▲) catalyzed by CYP2B6 variants, POR.1, and cytochrome b₅. Solid lines represent predicted concentrations on the basis of parameters from nonlinear regression using the Hill equation. Parameter estimates are in Table 3.

TABLE 3.

Kinetic parameters for 8-hydroxyefavirenz formation from efavirenz enantiomers

Wild-type CYP2B6 and all variants were coexpressed with wild-type POR.1 and cytochrome b₅. Results (V_max and and n) are the parameter estimate and S.E. of the estimate, determined by nonlinear regression analysis of the Hill equation, over the substrate concentration range 0.25–40 μM S-efavirenz and 0.11–45 μM R-efavirenz.

CYP2B6 Variant	S-8-Hydroxyefavirenz Formation from S-Efavirenz				R-8-Hydroxyefavirenz Formation from R-Efavirenz
	Vmax		n	Clmax	Vmax		n	Clmax
	pmol/min per picomoles	mM		ml/min per nanomoles	pmol/min per picomoles	mM		ml/min per nanomoles
CYP2B6.1	4.2 ± 0.2	7.7 ± 0.6	2.1 ± 0.3	0.27	0.57 ± 0.09	16.1 ± 4.4	1.5 ± 0.3	0.019
CYP2B6.4a	4.5 ± 0.1	5.9 ± 0.3	2.5 ± 0.2	0.39	0.78 ± 0.06	7.0 ± 1.1	1.2 ± 0.1	0.071
CYP2B6.5	3.3 ± 0.1	7.3 ± 0.7	1.4 ± 0.1	0.25	0.44 ± 0.02	6.5 ± 0.8	1.8 ± 0.3	0.034
CYP2B6.6	3.4 ± 0.1	11.8 ± 0.5	1.8 ± 0.1	0.15	0.41 ± 0.05	14.5 ± 3.8	1.2 ± 0.2	0.018
CYP2B6.7	1.5 ± 0.1	5.0 ± 0.3	1.5 ± 0.1	0.16	0.14 ± 0.01	6.3 ± 1.3	1.6 ± 0.4	0.011
CYP2B6.9	1.7 ± 0.1	8.1 ± 1.0	1.2 ± 0.1	0.13	0.13 ± 0.03	13.5 ± 5.6	1.3 ± 0.3	0.006
CYP2B6.16b	0.19 ± 0.01				0.012 ± 0.002
CYP2B6.17	4.4 ± 0.2	9.3 ± 0.8	1.7 ± 0.2	0.24	0.38 ± 0.07	9.2 ± 4.30	1.2 ± 0.3	0.055
CYP2B6.18b	0.20 ± 0.01				0.009 ± 0.001
CYP2B6.19c	3.4 ± 0.1	13.4 ± 0.9	1.5 ± 0.1	0.13	0.65 ± 0.33	31 ± 28	1.2 ± 0.4	0.013
CYP2B6.26	1.5 ± 0.1	5.0 ± 0.3	1.8 ± 0.1	0.15	0.21 ± 0.01	4.2 ± 0.7	1.9 ± 0.4	0.025

^aCYP2B6.4 (only) showed substrate inhibition with R-efavirenz. Results in the table for CYP2B6.4and R-efavirenz were from the Hill equation and the substrate concentration range 0.11–18 μM. Data were also analyzed with the model for substrate inhibition (LiCata model, x = 3) over the substrate concentration range 0.11–45 μM, yielding V_max = 0.90 ± 0.18, K = 9.0 ± 3.5, n = 1.1 ± 0.1, Cl_max = 0.74, V_i = 0.09 ± 0.19, and K_i = 38 ± 11 μM.

^bCYP2B6.16 and CYP2B6.18 rates were measured at a fixed substrate concentration of 40 μM S- and R-efavirenz.

^cFor R-efavirenz hydroxylation by CYP2B6.19, an alternative Michaelis-Menten model of linear regression analysis at low substrate concentrations generated a ratio of V_max/K_m = 0.019.

Concentration dependence of S-8-hydroxyefavirenz formation is shown in Fig. 4 for CYP2B6 variants, and kinetic parameters are provided in Table 4 for 0–100 μM S-efavirenz. In addition to cooperativity, substrate inhibition is apparent, and data were analyzed by fitting a modified Hill equation with cooperative substrate binding, substrate inhibition, and cooperative binding of three inhibitor molecules. For CYP2B6.7 and CYP2B6.9 minimal substrate binding cooperativity was suggested by n values of 1.0 and 0.9, and thus Cl_max = Cl_int. Both V_max and K varied approximately 2-fold. K_i values were 3- to 7-fold greater than K. CYP2B6.19 showed minimal substrate inhibition, and the model incorporating substrate inhibition did not fit the data well. On the basis of the Cl_max values using the model for both cooperativity and substrate inhibition, the relative activities for S-efavirenz 8-hydroxylation were CYP2B6.4 > CYP2B6.1 ≈ CYP2B6.5 ≈ CYP2B6.17 > CYP2B6.6 ≈ CYP2B6.7 ≈ CYP2B6.9 ≈ CYP2B6.19 ≈ CYP2B6.26. This was similar to that using only limited substrate concentrations and cooperativity without inhibition.

Fig. 4.

S-Efavirenz hydroxylation by CYP2B6 variants, POR.1, and cytochrome b₅ showing positive cooperativity and substrate inhibition. Solid lines represent predicted concentrations on the basis of parameters from nonlinear regression using model of cooperative substrate binding with substrate inhibition and cooperative inhibitor binding. Parameter estimates are in Table 4.

TABLE 4.

Kinetic parameters for 8-hydroxyefavirenz formation from S-efavirenz

Results (V_max, K, n, K_i, and V_i) are the parameter estimates and S.E. of the estimate, determined by nonlinear regression analysis using a model of cooperative substrate binding with substrate inhibition and cooperative inhibitor binding over the substrate concentration range 0.25–100 μM.

	Vmax	K	n	Clmax	Ki	Vi
	pmol⋅min/pmol	µM			mM	pmol/min per picomoles
CYP2B6.1	7.4 ± 2.4	17 ± 8	1.4 ± 0.3	0.25	47 ± 10	0.0001 ± 0.42
CYP2B6.4	6.2 ± 1.0	9 ± 2	1.7 ± 0.3	0.35	53 ± 8	0.0002 ± 0.44
CYP2B6.5	4.1 ± 0.4	11 ± 2	1.2 ± 0.1	0.25	75 ± 10	5.8e−5 ± 0.42
CYP2B6.6	3.6 ± 0.2	13 ± 1	1.7 ± 0.1	0.14	84 ± 16	1.46 ± 0.43
CYP2B6.7	2.7 ± 1.0	16 ± 12	1.0 ± 0.2	0.17	46 ± 11	4.5e−5 ± 0.14
CYP2B6.9	2.8 ± 0.7	21 ± 10	0.9 ± 0.1	0.13	57 ± 9	4.4e−5 ± 0.15
CYP2B6.17	5.5 ± 0.5	13 ± 2	1.4 ± 0.1	0.23	71 ± 8	2.8e−5 ± 0.44
CYP2B6.19	3.5 ± 0.2	14 ± 2	1.5 ± 0.1	0.13	173 ± 448	7.1e−5 ± 21
CYP2B6.26	2.1 ± 0.4	9 ± 4	1.2 ± 0.3	0.15	56 ± 11	5.7e−5 ± 0.18

Evaluation of R-efavirenz metabolism by wild-type CYP2B6.1 showed that 8-hydroxyefavirenz was the only metabolite observed, and 7-hydroxyefavirenz, 8,14-dihydroxyefavirenz, and 7,8-dihydroxyefavirenz were not detected. Immediately apparent is that rates of R-efavirenz 8-hydroxylation were an order of magnitude less than those of S-efavirenz (Figs. 2 and 5). Metabolism of R-efavirenz by CYP2B6.1 showed a sigmoidal pattern suggesting cooperativity.

Fig. 5.

Stereoselectivity of efavirenz metabolism. Shown is formation of 8-hydroxyefavirenz from S-efavirenz (●) and R-efavirenz (▲) by CYP2B6.1. The solid line represents predicted concentrations on the basis of parameters from nonlinear regression using the Hill equation for 8-hydroxylation of S-efavirenz and R-efavirenz. The inset compares predicted concentrations for R-efavirenz hydroxylation on the basis of parameters from nonlinear regression using the Hill equation (solid line) and Michaelis-Menten equation (dotted line).

Metabolism of R-efavirenz by the various CYP2B6 variants was evaluated. Secondary metabolism to 8,14-dihydroxyefavirenz was not observed for any 2B6 variant. R-Efavirenz 8-hydroxylation at 2–9 μM substrate concentrations catalyzed by coexpressed CYP2B6 (wild-type and variants), wild-type POR, and cytochrome b₅ are shown in Fig. 2. Rates of R-efavirenz metabolism were approximately 1/10 those of S-efavirenz. Compared with CYP2B6.1, CYP2B6.7, and CYP2B6.9 had diminished activity, and CYP2B6.16 and CYP2B6.18 were essentially inactive, and CYP2B6.4, CYP2B6.5, and CYP2B7.17 had higher activity.

Concentration dependence of R-8-hydroxyefavirenz formation is shown in Fig. 3 for CYP2B6 variants, and kinetic parameters are provided in Table 3. The Hill equation was used to model the data. CYP2B6.4 had the highest Cl_max R-efavirenz 8-hydroxylation, similar to S-efavirenz. CYP2B6.4 was the only variant showing substrate inhibition. CYP2B6.4 data were analyzed using both the Hill equation over a limited concentration range, and the LiCata model over the broader concentration range. Both analyses afforded similar V_max, K, or , and Cl_max values. CYP2B6.19 fitting showed a high value relative to the substrate concentrations, and high S.E., suggesting uncertainty in the model parameters. Since cooperativity was minor (n = 1.2), data were also modeled using the Michaelis Menten equation, which yielded parameters of V_max = 0.88 pmol/min per picomoles, K_m = 54 ± 26 μM, and Cl_int (V_max/K_m) = 0.016 (not shown). Linear regression analysis was also performed in the linear range of 0.23–4.5 μM R-efavirenz and yielded V_max/K_m = slope = 0.019 (not shown). This is similar to the value obtained using the Hill (0.021) and Michaelis-Menten (0.016) equations. CYP2B6.19 was the only isoform with a high K or value. Differences in activity between CYP2B6 variants reflected differences in both V_max and K values. On the basis of Cl_max values, relative activities for R-efavirenz 8-hydroxylation were in the order CYP2B6.4 > CYP2B6.17 > CYP2B6.5 > CYP2B6.1 ≈ CYP2B6.6 ≈ CYP2B6.7 ≈ CYP2B6.19 ≈ CYP2B6.26 > CYP2B6.9 > > CYP2B6.16 and CYP2B6.18. Parameter estimates and reaction order should be interpreted cautiously, however, because of the low rates of metabolism.

Efavirenz 8-hydroxylation was stereoselective (S > R), and stereoselectivity was similar across all CYP2B6 variants. On the basis of Cl_max values, there was 14-fold enantioselectivity (Cl_{max, S-efavirenz}/Cl_{max, R-efavirenz} = 14) for wild-type CYP2B6.1 and 5- to 25-fold differences for other CYP2B6 variants (Table 3).

Discussion

CYP2B6-catalyzed 8-hydroxylation accounts for approximately 90% of S-efavirenz oxidative metabolic clearance (Ward et al., 2003). This investigation provides novel insight into the roles of CYP2B6 genetic polymorphisms in the metabolism of S-efavirenz, additional CYP2B6 variants probably of clinical significance, mechanisms of CYP2B6-catalyzed efavirenz metabolism, and expression system influences on CYP2B6 variants’ catalytic activity. In addition, results demonstrate the remarkable stereoselectivity of efavirenz metabolism by CYP2B6, and an unusual combination of cooperative metabolism and substrate inhibition, which may provide additional insights about this important P450 isoform.

The first major observation was that S-efavirenz 8-hydroxylation by CYP2B6 exhibited positive homotropic cooperativity, and that cooperativity was generally preserved across CYP2B6 variants. Cooperativity, rather than Michaelis-Menten kinetics, was evidenced by nonlinear Eadie-Hofstee plots. Cooperativity (n) was greatest at higher rates of S-efavirenz metabolism (CYPs 2B6.1 and 2B6.4) but occurred with all genetic variants. Previous studies reported that S-efavirenz 8-hydroxylation by human liver microsomes was cooperative (Ward et al., 2003) or followed single-site hyperbolic Michaelis-Menten kinetics (Ogburn et al., 2010) and was hyperbolic with baculovirus-expressed (Ward et al., 2003; Ogburn et al., 2010; Xu et al., 2012) and Escherichia coli-expressed wild-type CYP2B6 (Bumpus et al., 2006; Zhang et al., 2011). It is well known that some P450s exhibit allosteric regulation and cooperative behaviors (Denisov et al., 2009). For example, CYP3A4 has a large and flexible substrate binding pocket that allows simultaneous binding of multiple ligands, leading to cooperativity, but ligand binding to nearby allosteric sites could also be involved. CYP2B6 has a smaller substrate binding pocket that is only about 50% of the CYP3A4 active site volume (Gay et al., 2010). Nonetheless, CYP2B6 is still spacious relative to small molecules (Ekins et al., 1999; Lewis et al., 1999; Ekins et al., 2008; Wang et al., 2018) and can accommodate ligands of various geometries by movement of residues in the active site (Shah et al., 2018). Cooperativity among CYP2B6 substrates is relatively uncommon, having been described for CYP2B6.1 and 7-hydroxy-4-trifluoromethylcoumarin (n = 1.4) (Ekins et al., 1997), testosterone (n = 1.3) (Ekins et al., 1998), methadone (Totah et al., 2007), and S-efavirenz (n = 1.5) (Ward et al., 2003) but not several other substrates (Ekins et al., 1998; Liu et al., 2016), and also for CYP2B6.4 and 7-ethoxycoumarin (n = 1.7) (Ariyoshi et al., 2001). Interestingly, heteroactivation by efavirenz was recently reported, with enhanced midazolam hydroxylation by CYP3A4 via interaction at an allosteric site (Ichikawa et al., 2018).

In addition to positive cooperativity, S-efavirenz 8-hydroxylation showed apparent substrate inhibition. S-8-Hydroxyefavirenz formation by CYP2B6.1 was highest at 40 μM S-efavirenz and declined at higher concentrations. At 100 μM S-efavirenz, 8-hydroxyefavirenz formation was reduced to 4% of V_max for CYP2B6.1 and also for CYP2B6.4. Substrate inhibition was influenced by CYP2B6 polymorphism. Less inhibition was observed with CYP2B6.6. Efavirenz is a known mechanism-based CYP2B6.1 inhibitor (as is S-8-hydroxyefavirenz) (Bumpus et al., 2006). Substrate inhibition was observed previously with expressed CYP2B6.1 and efavirenz (Ward et al., 2003) and efavirenz analogs (Cox and Bumpus, 2016), but not always reported (Bumpus et al., 2006; Zhang et al., 2011; Xu et al., 2012), and not observed with human liver microsomes (Ward et al., 2003; Ogburn et al., 2010). It is interesting that CYP2B6.1 and CYP2B6.4 had the greatest intrinsic clearances, cooperativity (n > 2), and substrate inhibition. Further investigation is necessary to better understand the interactions of CYP2B6 with efavirenz, substrate binding cooperativity, and the influence on metabolism.

S-Efavirenz 8-hydroxylation data were best fit to a model with positive homotropic cooperativity of both metabolism and inhibitory substrate binding, and a second Hill coefficient of 3 for inhibitory substrate binding. Although homotropic cooperativity, multiple substrate binding to the active site, and substrate inhibition have often been reported with P450s (Denisov et al., 2009), multiple inhibitor binding (Bapiro et al., 2018) and concomitant catalytic and inhibitory cooperativity are relatively uncommon (Müller et al., 2015). Comprehensive modeling of both catalytic and inhibitory cooperativity resulted in many parameters relative to the number of experimental observations, with a concern for an over-parameterized model. Thus we included both this analysis and the analysis of the noninhibited data using the Hill equation alone over the uninhibited substrate concentrations. Both models afforded similar conclusions with respect to the relative activities of the CYP2B6 variants.

The second major observation was that CYP2B6 genetic variants had altered activity toward S-efavirenz. At the recommended adult efavirenz dose of 600 mg, therapeutic plasma concentrations are 1–4 μg/ml (3–13 μM) (Bednasz et al., 2017). At 10 μM S-efavirenz, relative activities were CYP2B6.4 > CYP2B6.1 > CYP2B6.5, CYP2B6.17 > CYP2B6.6, CYP2B6.7, CYP2B6.9, CYP2B6.19, and CYP2B6.26; CYP2B6.16 and CYP2B6.18 were relatively inactive. Rank order was different at lower substrate concentrations, owing in part to differing cooperativity for the variants. Cl_max values were CYP2B6.4 > CYP2B6.1 ≈ CYP2B6.5 ≈ CYP2B6.17 > CYP2B6.6 ≈ CYP2B6.7 ≈ CYP2B6.9 ≈ CYP2B6.19 ≈ CYP2B6.26 > > CYP2B6.16 and CYP2B6.18. Kinetic parameters for S-efavirenz 8-hydroxylation by CYP2B6 variants, mainly CYP2B6.1, CYP2B6.4, CYP2B6.6, and CYP2B6.9, have been reported (Table 5) (Bumpus et al., 2006; Ariyoshi et al., 2011; Zhang et al., 2011; Xu et al., 2012; Radloff et al., 2013; Watanabe et al., 2018). CYP2B6.4 (785G>T, K262R) activity was greater than wild-type when expressed in T. ni (144%, this investigation), Sf9 cells (142%) (Ariyoshi et al., 2011), and E. coli (170%) (Bumpus et al., 2006), or similar to wild-type in E. coli (96%) (Zhang et al., 2011). Greater CYP2B6.4 activity toward S-efavirenz in vitro is thus a relatively consistent observation.

TABLE 5.

Summary of reported 8-hydroxylation of S-efavirenz

Relative activities are shown as percentage of CYP2B6 variants Cl_int (Cl_max for this study) compared with the wild type, on the basis of data in Table 3.

	Bumpus et al., 2006	Ariyoshi et al., 2011	Zhang et al., 2011	Xu et al., 2012	Radloff et al., 2013	Watanabe et al., 2018	This Study (Clmax)
Expression system	E. coli	Sf9	E. coli	SF9	COS-1	HEK	T. ni
CYP2B6.1	100	100	100	100	100	100	100
CYP2B6.4	170	142	96			122	144
CYP2B6.5			138		83	109	92
CYP2B6.6		50	20	49	183	266	56
CYP2B6.7			156			166	59
CYP2B6.9			33			172	48
CYP2B6.17						85	89
CYP2B6.19						37	48
CYP2B6.26						183	56

More generally, CYP2B6 variant catalytic activity is variant-, substrate-, and expression system-dependent. With other substrates, CYP2B6.4 was more active than CYP2B6.1 toward methadone (Gadel et al., 2013, 2015) and artmether (Honda et al., 2011) but less active toward cyclophosphamide (Ariyoshi et al., 2011), ifosfamide (Calinski et al., 2015), bupropion (Zhang et al., 2011), and ketamine (Wang et al., 2018). CYP2B6.6 (516G>T, 785A>G, Q172H/K262R) had lesser activity toward S-efavirenz (53% of wild-type), consistent with most (20%–50%) (Ariyoshi et al., 2011; Zhang et al., 2011; Xu et al., 2012) but not all (Radloff et al., 2013; Watanabe et al., 2018) reports. CYP2B6.6 was also less active than CYP2B6.1 toward methadone (Gadel et al., 2013, 2015), ketamine (Wang et al., 2018), and bupropion (Zhang et al., 2011) but more active toward artmether and cyclophosphamide (Ariyoshi et al., 2011; Honda et al., 2011). CYP2B6.9 (516G>T, Q172H) had even lower activity toward S-efavirenz (38% of wild-type) than CYP2B6.6 in the current investigation. Likewise, CYP2B6.9 also had lower 8-hydroxylation activity (33%) in one investigation (Zhang et al., 2011) but not another (Watanabe et al., 2018), and lower activity than wild-type in metabolizing methadone (Gadel et al., 2013, 2015), bupropion (Zhang et al., 2011), and ketamine (Wang et al., 2018) but greater with ifosfamide (Calinski et al., 2015). Thus, the CYP2B6 516G>T polymorphism (coding for both CYP2B6*6 and CYP2B6*9) is a canonical loss-of-function polymorphism for efavirenz.

Some structure-activity information is available on the CYP2B6 variants. CYP2B6.6 and CYP2B6.9 have the common Q172H mutation. Several in vitro studies have evaluated these variants, yet it is not clear how Q172H remotely (Q172 is about 15 Šaway from the heme) affects the reaction in the active site. Moreover, effects of Q172H are moderated by K262R, and effects can be substrate-dependent (Ariyoshi et al., 2011). CYP2B6.16 and 2B6.18 share the I328T mutation in the J-helix, causing structural changes in the C and I helices that disrupt heme binding, alter ligand recognition, and reduce the ligand-binding pocket volume from 78 (CYP2B6.1) to 14 ų (CYP2B6.18) (Kobayashi et al., 2014; Wang et al., 2018).

While this report was in preparation, another investigation of efavirenz metabolism by CYP2B6 variants was published (Watanabe et al., 2018). Variants were expressed in human embryonic kidney (HEK)293 cells, without coexpression of P450 oxidoreductase or cytochrome b₅. As described previously (Wang et al., 2018), expression systems can influence P450 activity. Mammalian systems (e.g., monkey kidney COS, HEK cells) allow easy P450 expression and use native reductase and b₅, but P450 expression levels and protein integrity can vary widely. Some HEK results (Watanabe et al., 2018) differed substantially from previous reports (Table 5). Comparing kinetic parameters for efavirenz 8-hydroxylation using HEK versus our insect cell-expressed CYP2B6 shows that V_max with insect cell expression was higher than with HEK expression, for example, 12-fold for CYP2B6.1 (4.2 vs. 0.35 pmol/min per picomoles). Such activity differences may influence reported Cl_int values for the variants.

There is potential clinical significance to the genetic variability in S-efavirenz metabolism in vitro. CYPB6*6 (516G>T, 785A>G), CYPB6*9 (516G>T), CYPB6*16 (785A>G, 983T>C), and CYPB6*18 (983T>C) constitute a poor efavirenz metabolizer clinical phenotype, associated with reduced clearance and increased plasma concentrations (Čolić et al., 2015; Russo et al., 2016; Robarge et al., 2016), and with common general and neuropsychiatric side effects (Haas et al., 2004; Rotger et al., 2005; Apostolova et al., 2015; Vo and Varghese Gupta, 2016; Gallien et al., 2017; Mollan et al., 2017; Chang et al., 2018). The present results, showing moderately (CYP2B6.6, CYP2B6.9) or markedly (CYP2B6.16, CYP2B6.18) lower metabolism, provide a mechanistic explanation for the clinical observations that the CYP2B6 516G>T polymorphism (Haas et al., 2004; Tsuchiya et al., 2004; Rotger et al., 2005; Čolić et al., 2015; Dhoro et al., 2015; Robarge et al., 2016) and the 983T>C polymorphism (Haas et al., 2009; Maimbo et al., 2012; Dhoro et al., 2015; Röhrich et al., 2016) are associated with increased efavirenz exposure and reduced clearance and metabolism. Other CYP2B6 variants we tested with these two polymorphisms (CYP2B6.19, CYP2B6.26) also had diminished activity. Thus, both CYP2B6 516G>T and 983T>C are canonical loss-of-function variants for S-efavirenz 8-hydroxylation. Other variants with these polymorphisms (CYP2B6.13, CYP2B6.20, CYP2B6.29, CYP2B6.34, CYP2B6.36, CYP2B6.37, CYP2B6.38) would be predicted to also have diminished activity. With the consistent association between CYP2B6 516G>T or 983T>C and increased efavirenz exposure, these other alleles would also be expected to be phenotypic poor metabolizers. This has been reported for CYP2B6*20 (Čolić et al., 2015). Although the clinical significance of 785A>G alone (CYP2B6*4) for efavirenz disposition is ambiguous, (Russo et al., 2016) the above in vitro-in vivo correlations, together with increased efavirenz hydroxylation in vitro by CYP2B6.4, would predict lower plasma efavirenz exposures and suggest further clinical investigation. These findings further strengthen the rationale for patient genotyping (specifically for CYP2B6 516G>T, 785A>G, and 983T>C polymorphisms) and for CYP2B6 genetically-guided efavirenz dosing (Mukonzo et al., 2014; Vo and Varghese Gupta, 2016).

The third major result was the surprising observation that efavirenz 8-hydroxylation was highly stereoselective. In this novel evaluation of R-efavirenz, metabolism at specific concentration and Cl_max was generally at least 10-fold greater for S- versus R-efavirenz, for wild-type CYP2B6.1 and the active CYP2B6 variants. Differences in 8-hydroxylation were primarily owing to lower V_max, as substrate affinity (K) was not substantially different between enantiomers. In addition, whereas both the primary metabolite 8-hydroxyefavirenz and very low amounts of the secondary metabolite 8,14-dihydroxyefavirenz was observed with S-efavirenz, only 8-hydroxyefavirenz was detected from R-efavirenz. This may well relate, however, to the lower R-efavirenz turnover and assay sensitivity. The considerable stereoselectivity of efavirenz 8-hydroxylation is a novel observation and contrasts with other CYP2B6 substrates. For example, N-demethylation of individual enantiomers by CYP2B6.1 was 2-fold greater for R- versus S-methadone (Totah et al., 2007) and S- versus R-ketamine (Wang et al., 2018), and hydroxylation of S-bupropion was 3-fold greater than R-bupropion (Coles and Kharasch, 2008). Similar enantioselectivities occurred with methadone and ketamine with several CYP2B6 variants (Gadel et al., 2015; Wang et al., 2018). N-Dechloroethylation by CYP2B6.1 was approximately 1.5- to 2-fold greater for S- versus R-ifosfamide, although the difference between S- and R-ifosfamide was substantially greater for 4-hydroxylation (Roy et al., 1999). Although these other CYP2B6 substrates follow Michaelis-Menten kinetics and are characterized by Cl_int, and efavirenz was characterized by Cl_max, Cl_max from nonhyperbolic metabolism can be used as a substitute for Cl_int when assessing metabolism (Houston and Kenworthy, 2000). Thus, efavirenz appears to be the CYP2B6 substrate with the greatest metabolic enantioselectivity yet observed.

A crystal structure of CYP2B6 in complex with an efavirenz analog, with a methyl group replacing the carbonyl oxygen, has been reported (Shah et al., 2018). Docking was described as consistent with the major and minor efavirenz metabolites. The chlorine of the efavirenz analog formed a Cl-π bond with the aromatic side chain of the F108 phenylalanine residue. Regarding the chiral carbon, the cyclopropyl group on the 4-cyclopropylethynyl substituent was between the side chains of F206 phenylalanine and T302 threonine, whereas the trifluoromethyl substituent was near the I101 isoleucine and F115 phenylalanine side chains. The considerable influence of efavirenz chirality on 8-hydroxylation demonstrates the importance of these residues in the active site. In addition, the protein for crystallization contains engineered mutations of K262R (as in CYP2B6.4) and Y226H, for stability and solubility. The structure shows the side chain of arginine (similar to R262 of CYP2B6.4) in close contact with the side chains of residues threonine T255 and aspartic acid D266 to form hydrogen bonding. In wild-type CYP2B6.1, the side chain of lysine K262 is not able to form similar hydrogen bonds with the neighboring residues. This difference between CYP2B6.4 and CYP2B6.1 may influence their structures and functions and account for differences in the metabolism of efavirenz and other substrates by these variants. The highly substrate-specific effect of the K262R substitution further informs on CYP2B6. Catalytic data for S- and R-efavirenz metabolism by wild-type CYP2B6 and variants may be useful in future computational studies to better understand mechanisms of metabolism by this clinically important isoform.

Abbreviations

HEK

human embryonic kidney

HIV

human immunodeficiency virus

HPLC

high-performance liquid chromatography

MS

mass spectrometry

POR

NADPH cytochrome P450 oxidoreductase

P450

cytochrome P450

Authorship Contributions

Participated in research design: Wang, Kharasch.

Conducted experiments: Wang, Neiner.

Performed data analysis: Wang, Neiner.

Wrote or contributed to the writing of the manuscript: Wang, Kharasch.