Halogen Substitution Influences Ketamine Metabolism by Cytochrome P450 2B6: In Vitro and Computational Approaches

Abstract

Ketamine is analgesic at anesthetic and subanesthetic doses, and used recently to treat depression. Biotransformation mediates ketamine effects, influencing both systemic elimination and bioactivation. CYP2B6 is the major catalyst of hepatic ketamine N-demethylation and metabolism at clinically relevant concentrations. Numerous CYP2B6 substrates contain halogens. CYP2B6 readily forms halogen-protein (particularly Cl-π) bonds, which influence substrate selectivity and active site orientation. Ketamine is chlorinated, but little is known about metabolism of halogenated analogs. This investigation evaluated halogen substitution effects on CYP2B6-catalyzed ketamine analogs N-demethylation in vitro, and modeled interactions with CYP2B6 using various computational approaches. Ortho phenyl ring halogen substituent changes caused substantial (18-fold) differences in K_m, on the order Br (bromoketamine, 10 μM)< Cl <F <H (deschloroketamine, 184 μM). In contrast, V_max varied minimally (83–103 pmol/min/pmol CYP). Thus apparent substrate binding affinity was the major consequence of halogen substitution and the major determinant of N-demethylation. Docking poses of ketamine and analogs were similar, sharing a π-stack with F297. Libdock scores were deschloroketamine < bromoketamine < ketamine < fluoroketamine. A Bayesian log K_m model generated with Assay Central had a ROC of 0.86. The probability of activity at 15μM for ketamine and analogs was predicted with this model. Deschloroketamine scores corresponded to the experimental K_m, but the model was unable to predict activity with fluoroketamine. The binding pocket of CYP2B6 also suggested a hydrophobic component to substrate docking, based on a strong linear correlation (R²=0.92) between lipophilicity (AlogP) and metabolism (log K_m) of ketamine and analogs. This property may be the simplest design criteria to use when considering similar compounds and CYP2B6 affinity.

Ketamine is a dissociative anesthetic, in clinical use since 1970, and is listed as a World Health Organization Essential Medicine. Ketamine is analgesic at anesthetic and subanesthetic doses, and causes hypnosis, amnesia and sedation with minimal respiratory system depression.¹

Ketamine was initially synthesized as a derivative of phencyclidine (PCP, 1-(1-phenylcyclohexyl)piperidine).^2–4 Phencyclidine was developed in the 1950s by Parke Davis as a general anesthetic,⁵ but was discontinued in 1965 due to undesirable side effects including hallucination and neurotoxicity. A series of related arylcyclohexylamine compounds was synthesized in the search for a safer alternative.^{3, 4} One of the compounds, 2-(2-chlorophenyl)-2-(methylamino)cyclohexan-1-one, was an excellent anesthetic (scheme 1). It was named ketamine for its two functional groups (ketone and amine). Ketamine has three structural differences compared to phencyclidine: the piperidino changed to a methylamino group, introduction of a ketone on the cyclohexyl ring, and chlorine addition to the phenyl ring. Ketamine binds to the activated open ion channel of the N-methyl-D-aspartate receptor (NMDAR), causing uncompetitive blockade.⁶ Therapeutic applications for ketamine have recently expanded, to include severe and treatment-resistant major depression, with rapid and sustained response.^7–10

Scheme 1.

Ketamine was initially synthesized as a derivative of phencyclidine.

Biotransformation plays a seminal role in the clinical effects of ketamine, influencing both systemic elimination and bioactivation. Ketamine undergoes extensive hepatic N-demethylation to the predominant metabolite norketamine.^{1, 11} Norketamine and the other primary metabolite hydroxyketamine are both further metabolized by hydroxylation and N-demethylation, respectively, to hydroxynorketamine. Norketamine is also further metabolized to dehydronorketamine. Drug interactions which alter hepatic ketamine metabolism greatly influence systemic ketamine concentrations.^12–14 Oral ketamine undergoes considerable first-pass metabolism, and is consequently only 15% bioavailable.¹ Metabolism also influences the molecular pharmacology of ketamine. Norketamine and hydroxynorketamine are pharmacologically active and may contribute to the antidepressant and/or analgesic effects of ketamine, possibly through NMDA or α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) receptors.^15–18 Ketamine N-demethylation is catalyzed by both CYP2B6 and CYP3A4, however these are high-affinity (low Km) and low-affinity (high Km) isoforms, respectively, and CYP2B6 is the major catalyst in vitro of hepatic ketamine N-demethylation, norketamine hydroxylation, and ketamine metabolism overall, at clinically relevant concentrations,^19–24 as well as clinically.¹³ Ketamine N-demethylation by CYP2B6 is also the first step in bioactivation to a putative active metabolite, hydroxynorketamine.²⁵

Halogen substitution is strategically used in drug design and development, typically to improve drug-target binding affinity and/or selectivity, as well as decrease metabolism.^{26, 27} The most common organohalogen amongst marketed drugs is chlorine, because heavy organohalogens form hydrogen bonds while organofluorines generally do not. Nevertheless, a recent survey found that in addition to influencing receptor pharmacology, 19% of halogen-protein interactions involved those mediating absorption, distribution, metabolism, and excretion, including P450-catalyzed metabolism and detoxification.²⁷ Indeed, halogenation was suggested as an underused strategy for tuning metabolism and molecular optimization,²⁷ particularly exploiting halogen-π interactions between ligand and protein.²⁸ Because numerous CYP2B6 substrates contain halogens, and this enzyme readily forms halogen-protein (particularly Cl-π) bonds,²⁹ the role of halogen-π bonds in substrate selectivity and active site orientation interactions was recently investigated, using the halogenated monoterpene bornyl bromide and the nonhalogenated analog bornane. These results suggested that such interactions may be unique to the CYP2B family of enzymes within the CYP2 family.²⁸

The availability of halogenated and non-halogenated ketamine analogs (Scheme 2) afforded the opportunity to investigate further the influence of halogen substitution on CYP2B6-catalyzed metabolism. The purpose of the current investigation was to directly evaluate halogen substitution effects on ketamine N-demethylation catalyzed by CYP2B6, and attempt to predict the interaction of these variously halogenated analogs with CYP2B6 using various computational approaches.

Scheme 2.

N-demethylation of ketamine analogs catalyzed by CYP2B6.

Materials and Methods

Materials

Formaldehyde, acetylacetone and S-ketamine were purchased from Sigma-Aldrich (St Louis, MO). R-ketamine was purchased from Cayman Chemical (Ann Arbor, MI). Racemic RS-ketamine was from Spectrum Chemical Corp (New Brunswick, NJ). Norketamine (d0 and d4) were from Cerilliant (Round Rock, TX). Spodoptera frugiperda (Sf9) cells and Sf-900 III SFM culture media were purchased from ThermoFisher (Waltham, MA). Trichoplusia ni (Tni) cells, ESF AF culture media, and BestBac 2.0 Baculovirus Cotransfection Kits were from Expression Systems (Davis, CA). BacPAK Baculovirus Rapid Titer Kit was from Clontech (Mountain View, CA). Strata-X 33u (30 mg) solid-phase extraction plates were from Phenomenex (Torrance, CA). Other reagents were from SigmaAldrich. Ketamine analogs deschloroketamine (2-phenyl-2-(methylamino)cyclohexan-1-one), fluoroketamine (2-(2-fluorophenyl)-2-(methylamino)cyclohexan-1-one, and bromoketamine (2-(2-bromophenyl)-2-(methylamino)cyclohexan-1-one were synthesized using reported methods.

Generation of recombinant baculoviruses

Production of recombinant human proteins CYP2B6, P450 reductase (POR) and cytochrome b₅ was performed as previously described.³⁰ In brief, CYP2B6, POR and b₅ genes were amplified from the Human Liver Quick-Clone cDNA library (Clonetech), and inserted individually into the transfer vector pVL1393 using the In-Fusion HD Cloning system (Clontech). Recombinant baculovirus was produced with BestBac 2.0 Baculovirus Cotransfection Kit (Expression Systems). Sf9 insect cells were cotransfected with BestBac linearized DNA and the plasmid DNA of transfer vector carrying the gene of interest on a 6-well plate to produce p0 generation of recombinant baculovirus. Sf9 cells in suspension culture were infected with p0 to make subsequent viral generation. Viral titers were determined using the BacPAK Baculovirus Rapid Titer Kit (Clontech).

Expression of recombinant proteins in insect cells

Protein expression was performed as described previously.³⁰ Briefly, Tni cells in suspension culture in early log phase growth were infected with the recombinant baculoviruses carrying CYP2B6, POR and b₅ genes, with multiplicities of infection of 4:2:1 (CYP2B6:POR:b₅), in the presence of heme precursors δ-aminolevulinic acid and ferric citrate. After 48 to 72 hr growth post infection, cells were harvested by centrifugation, washed twice with phosphate-buffered saline, and resuspended in potassium phosphate buffer pH 7.4 and stored frozen at −80°C. Frozen cells were thawed and lysed on ice in a Potter-Elvehjem tissue homogenizer and aliquots stored at −80 °C. P450 content, b₅ content and POR activity were measured as described previously.³⁰

Ketamine demethylation measured by formaldehyde formation

Ketamine is converted to norketamine and formaldehyde in the primary metabolism step by CYP2B6. The product formaldehyde was measured using Nash reagent for ketamine analog metabolism, while both formaldehyde and norketamine were measured for metabolisms of R-, S-, and RS-ketamine. All incubations were carried out in 96-well PCR plates with raised wells. Ketamine analogs and RS-ketamine with concentrations of 0, 1, 2.5, 5, 12.5, 25, 50, 100, 200, 400, 700, and 1000 μM was mixed with the enzyme CYP2B6/POR/b₅, and preincubated at 37 °C for 5 min. Substrate concentration range of 0–500 μM was used for enantiomers R- or S-ketamine. Final CYP2B6 concentration was 5 pmol/ml, and total reaction volume was 200 μl in 100 mM potassium phosphate pH7.4. The reaction was initiated by adding an NADPH regenerating system (final concentrations: 10 mM glucose 6-phosphate, 1 mM ß-NADP, 1 U/ml glucose-6-phosphate dehydrogenase, and 5 mM magnesium chloride, preincubated at 37°C for 10 min). After incubation for 10 min, the reaction was quenched by adding 40 μl of 15% zinc sulfate, and the plate was centrifuged at 2500 rpm for 5 min to remove precipitated proteins and zinc hydroxide. Formaldehyde was measured using a previously described method,³¹ with minor modifications. A stock solution of 38.8 mM acetylacetone was prepared in 30% (W/V) ammonium acetate. Reaction supernatant (160 μL) was transferred to a new plate, and mixed with 80 μL of 38.8 mM acetylacetone, and incubated at 60 °C for 20 min. The plate was centrifuged at 2500 rpm for 5 min and 160 μL supernatant was transferred to a Corning 96-well black polystyrene plate with clear flat bottom, and mixed with 20 μL of 6% (W/V) bovine serum albumin solution. After incubation for 5 minutes, fluorescence intensity was measured at excitation and emission wavelength of 415nm, and 480 nm, respectively, using the Biotek Synergy MX Microplate Reader. For calculation of formaldehyde concentrations, a standard curve was generated using standardized formaldehyde sample. A stock solution of 33 mM formaldehyde was prepared from 37% formaldehyde solution (Sigma Aldrich) and standardized by following a previously described procedure of pH titration,³² Formaldehyde was reacted with excess sodium sulfite by adding 25 mL of 33 mM formaldehyde into 25 mL 0.1 M sodium sulfite, and pH was recorded prior to and after addition of formaldehyde (sodium hydroxide released in the reaction caused pH increase in the range of pH 9–11). Then the mixture was titrated back to the original pH using 0.1 N HCl solution which was prepared from standard 1 N HCl solution (Fisher SA48–1), and the volume of HCl consumption was used to calculate the standard formaldehyde concentration. A series of formaldehyde solutions with concentrations 0.2, 0.5, 1, 2, 4, 6, 8 and 10 μM were prepared from the standard stock solution. The standard solutions were transferred to the 96-well plate that also carries samples after CYP2B6 incubation. Reaction with acetylacetone and fluorescence measurement were carried out under same conditions as described above.

Ketamine demethylation measured by norketamine formation

Norketamine formation from R-, S-, and RS-ketamine was measured by HPLC/MSMS.³⁰ Substrate concentrations were 0.125–500 μM R- or S-ketamine, and 0.25–1000 μM RS-ketamine. Incubations with CYP2B6 were carried out under the same conditions as described above, except that the reactions were quenched by adding 40 μL 20% trichloroacetic acid containing internal standard norketamine-d4 (final concentration 90 ng/ml). Norketamine concentrations were determined by enantioselective HPLC tandem mass spectrometry using solid-phase extraction, as originally implemented for plasma analysis.^{11, 30} Calibration standards contained 0, 2, 5, 10, 20, 50, 100, 200, 500, 1000 and 2000 ng/ml racemic norketamine and were processed identically to incubation samples. Stereoselective analysis was performed using a ChiralPak AGP analytical column (100 × 2.0mm, 5μm) and AGP guard column (10 × 2.0mm) (Chiral Technologies, West Chester, PA), coupled to an API 4000 QTrap LC-MS/MS linear ion trap triple quadrupole tandem mass spectrometer (Applied Biosystems/MDS Sciex, Foster City, CA), as described previously.³⁰ Interday coefficients of variation were 5% or less at 5, 50, and 500 ng/ml norketamine, and accuracy was within 5%.

Data Analysis

Results are reported as the mean ± SD of triplicates. Product formation (formaldehyde or norketamine) versus substrate concentration data were analyzed by nonlinear regression analysis (SigmaPlot 12.5, Systat, San Jose, CA) using single enzyme Michaelis-Menten kinetics.

Docking

Ketamine and all analogs were drawn in XMDS without stereochemistry. Ketamine and all ketamine analogs were prepared using Discovery Studio (Biovia, San Diego, USA) using the default settings. Two stereoisomers were created for each ligand as well as a charged and neutral variant. The crystal structure of cytochrome P4502B6 (3IBD) with inhibitor 4-(4-chlorophenyl)imidazole³³ was prepared in Discovery Studio using the default settings (waters removed). A default sized binding pocket was created by selecting the crystallized ligand 4-(4-chlorophenyl)imidazole as an initiation point and creating an input site sphere of 23.6216, 13.8463, 25.7039, 15Å. The docking software Libdock within Discovery Studio was then used to dock the ligands within these constraints. The highest scoring poses were saved as .pdb files and visualized in pymol (Schrodinger,1.8.2.3, New York, NY).

CYP2B6 Bayesian model

We curated a total of 58 unique molecules from multiple sources and experimental groups with log K_m data for CYP2B6,³⁴ and 4 additional compounds (efavirenz and 3 analogs³⁵) were added from ChEMBL (www.ebi.ac.uk/chembl/). Of note is that this dataset contained both R- and S-ketamine with calculated K_m values of 74 and 44μM, respectively, obtained using human liver microsomes.²⁰ This varies significantly from the current data with K_m values of 10–15μM. These compounds were therefore removed from the model in order to reduce an activity bias for the ketamine analogs. A Bayesian model was then built with Assay Central.^36–38 The model was built with a threshold of 15μM using ECFP6 descriptors, which generated 11 active compounds. Five-fold cross validation was performed. In addition, using data from ChEMBL CYP2B6 inhibitor models were built where the IC₅₀/EC₅₀ measurements (198 compounds, 78 actives) were calculated with a threshold of 15.22 μM and the K_i (224 compounds, 83 actives) was calculated with a threshold of 16.27 μM.³⁹

Results

Enantioselective HPLC MS/MS methods have been developed for analysis of ketamine and metabolites, using metabolite reference standards.^{11, 40} The method was used successfully for recent investigation of stereoselective ketamine N-demethylation to norketamine by CYP2B6 genetic variants.³⁰ However, for the present analysis of the metabolism of ketamine analogs, the reference standards bromo-norketamine and fluoro-norketamine were not available. Therefore, in the current investigation a formaldehyde assay was used as an alternative method to evaluate N-demethylation of ketamine analogs.

The single enantiomers, S- and R-ketamine, and racemic RS-ketamine were incubated with CYP2B6 in the presence of POR and cytochrome b₅. The two products of N-demethylation, norketamine and formaldehyde, were measured by HPLC MS/MS and formaldehyde assay, respectively (Figure 1 and Table 1). Apparent K_m values from the formaldehyde analysis were somewhat higher than from the norketamine analysis, V_max values were generally in good agreement, and the in vitro intrinsic clearance (Cl_int, V_max / K_m) values, particularly for the racemate, were in excellent agreement. Our previous studies showed that conversion of norketamine to secondary metabolites was not detectable under the experimental conditions used.³⁰ Thus, formaldehyde formation was an acceptable surrogate for norketamine formation from RS-ketamine.

Fig 1.

Comparison of ketamine N-demethylation measured by norketamine formation using HPLC MS/MS analysis and by formaldehyde formation. (A) Norketamine formation from ketamine analyzed by HPLC MS/MS. S-norketamine from S-ketamine substrate (●), R-norketamine from R-ketamine substrate (▲), S-norketamine from racemic RS-ketamine substrate (○), R-norketamine from racemic RS-ketamine substrate (△). (B) Formaldehyde formation from S-ketamine substrate. (C) Formaldehyde formation from R-ketamine substrate. (D) Formaldehyde formation from RS-ketamine substrate. Each data point represents the mean ± SD of triplicates. Lines represent velocities predicted from nonlinear regression analysis of data using single enzyme Michaelis-Menten equation.

Table 1.

Kinetic parameters for ketamine demethylation measured from norketamine and formaldehyde formation

Substrate	Product	Km (μM)	Vmax (pmol・min−1・pmol CYP−1)	Clint (ml・min–1・nmol P450–1)
S-ketamine	S-norketamine	10 ± 1	106 ± 1	10.4
S-ketamine	formaldehyde	15 ± 1	106 ± 1	7.2
R-ketamine	R-norketamine	11 ± 1	65 ± 1	5.7
R-ketamine	formaldehyde	15 ± 1	92 ± 2	6.2
RS-ketamine	S-norketamine	5 ± 1	45 ± 1	9.0
RS-ketamine	R-norketamine	4 ± 1	38 ± 1	9.2
RS-ketamine	formaldehyde	17 ± 1	98 ± 1	5.7

N-demethylation of ketamine analogs by CYP2B6 was evaluated by formaldehyde analysis (Figure 2), and the kinetic parameters are provided in Table 2. With substituent changes at the ortho position on the phenyl ring from hydrogen to fluorine, chlorine, and bromine, substantial (18-fold) differences in K_m values were observed, from 184 μM for deschloroketamine to 10 μM for bromoketamine. In contrast, V_max values showed relatively minor differences (range 83 to 103 pmol/min/pmol CYP).

Fig 2.

N-demethylation of (A) deschloroketamine, (B) fluoroketamine, and (C) bromoketamine measured from formaldehyde formation. Each data point represents the mean ± S.D. of triplicates. Lines represent velocities predicted from nonlinear regression analysis of data using single enzyme Michaelis-Menten equation.

Table 2.

Kinetic parameters for demethylation of ketamine analogs measured by formaldehyde formation

Ketamine analog	Halogen substitution	Km (μM)	Vmax (pmol・min−1・pmol CYP−1)	Clint (ml・min–1・nmol P450–1)
RS-deschloroketamine	H	184 ± 18	83 ± 3	0.45
RS-fluoroketamine	F	40 ± 3	103 ± 2	2.6
RS-ketamine	Cl	17 ± 1	98 ± 1	5.7
RS-bromoketamine	Br	10 ± 1	94 ± 2	9.6

Docking

To understand the influence of halogen substitution on ketamine metabolism by CYP2B6, we carried out molecular docking simulations to study CYP2B6-ligand interactions in the active site and the structural basis of ligand binding affinity and substrate selectivity associated with halogen substitution.

For deschloroketamine, a docking pose was chosen not only based on a high score, but also based on the logical positioning of the N-methyl group of the molecule in proximity to the P450 heme, because of the substantial size of the CYP2B6 binding site. The docking poses of ketamine and all the analogs were very similar, sharing a π-stack with F297. The libdock scores had the following relationship: deschloroketamine < bromoketamine < ketamine < fluoroketamine (Figure S1).

Bayesian models

The Bayesian K_m model generated with literature data using Assay Central resulted in an ROC of 0.86 and other statistical measures after 5-fold cross-validation (Figure 3) which were generally acceptable for such a small training set. The probability of activity at 15μM for ketamine and analogs was predicted based on this model to produce an output score and applicability value (Table 3). The deschloroketamine score strongly suggested inactivity, which corresponded to the experimental K_m value of 184μM (Tables 2 and 3). The model was unable to accurately predict the activity of fluoroketamine, which was predicted to be active, (although the observed experimental K_m, 40μM, was relatively high (Tables 2 and 3). This was similar to the lack of accuracy of the docking scores. Surprisingly, when we expanded this model to include the updated K_m values for R- and S-ketamine of < 15μM, deschloroketamine was not well predicted to be active as compared to the other analogs, suggesting the importance of the halogen for predicted activity. We also scored the inhibitory potential of these ketamine analogs against Bayesian models built from publically accessible K_i and EC₅₀ or IC₅₀ data for CYP2B6 from ChEMBL. The K_i model had an ROC of 0.81 and the EC₅₀ or IC₅₀ model had an ROC of 0.84 (Figure S2). Each of the ketamine compounds scored as probable inhibitors of CYP2B6 with a probability score of approximately 0.5 (Table 3).

Figure 3:

ROC plot for the CYP2B6 K_m model and predictions for ketamine and analogs in Assay Central.

Table 3.

Ketamine and analogs Assay Central K_m, K_i, IC₅₀/EC₅₀ models with applicability scores and some molecular properties.

Molecule	Molecular Structure	MW	AlogP	Assay Central Probability Score (Km model)	Assay Central Applicability Score (Km model)	Assay Central Probability Score (Ki model)	Assay Central Applicability Score (Ki model)	Assay Central Probability Score (IC50/EC50 model)	Assay Central Applicability Score (IC50/EC50 model)
deschloroketamine		204.3	0.985	−3.614	0.579	0.527	0.474	0.595	0.553
fluoroketamine		222.3	1.19	15.231	0.535	0.510	0.442	0.530	0.422
ketamine		238.7	1.649	8.554	0.488	0.564	0.465	0.595	0.488
bromoketamine		283.2	1.733	8.144	0.419	0.584	0.395	0.568	0.419

Higher probability scores relate to increased affinity. Applicability scores address overlap of fragments of a molecule compared to the training set for each model. The higher the score the more similar to compounds in the training set.

Assessing the binding pocket of CYP2B6 suggests a hydrophobic component to substrate docking. There are several hydrophobic residues, including I101, I114, L363, V367, and V477 in the substrate binding pocket. Comparison of AlogP (calculated in Discovery Studio) and log K_m of ketamine and ketamine analogs showed a strong linear correlation (R²=0.92, Figure 4). This calculated chemical property successfully predicted the experimental log K_m, with an increase in hydrophobicity correlating with an increase in affinity.

Figure 4.

Correlation of log Km and AlogP for N-demethylation of halogenated and non-halogenated ketamine analogs

Discussion

The main finding of this investigation was that the halogen substituent at the ortho position on the phenyl ring of racemic ketamine analogs markedly influenced the rate of N-demethylation. Metabolism, based on Cl_int, was on the order Br > Cl > F > H. V_max values were relatively unchanged, while K_m was on the order Br < Cl < F < H. Thus, apparent substrate binding affinity was the major consequence of halogen substitution on ketamine analogs, and the major determinant of N-demethylation.

Incorporation of halogen atoms into a drug molecule can have profound pharmacological implications. Halogen substitution has been investigated extensively in drug research for optimization of drug adsorption, distribution, recognition and selectivity. It is estimated that approximately 20% of approved drugs are halogenated, as are a similar portion in other stages of discovery and development.²⁷ Halogen effects on interactions of small molecules with proteins can be mediated through classical halogen properties of reactivity, lipophilicity, and electronegativity, and the newly characterized halogen bond.⁴¹ Halogen bonding is analogous to hydrogen bonding, driving short, stabilizing electrostatic interactions with electron-rich atoms, with the halogen serving as a halogen bond donor to an electron rich acceptor – typically the carbonyl oxygen of the peptide bond in proteins.⁴¹ The strength of halogen bonding correlates with polarizability of the halogen atoms and changes in the order of F<Cl<Br<I.²⁶ Similarly, the hydrophobicity of halogenated substrates is also affected by polarizability, and lowest for fluorine and highest for iodine.

Major determinants of substrate binding affinity to P450s depend on the specific isoform. In general, determinants for binding include hydrophobic interactions and hydrogen bonding features, while electronic factors determine reaction rates.^{42, 43} For CYP2B6 more specifically, halogen-π interactions between substrates and phenylalanine side chains in the hydrophobic active site are considered to drive substrate selectivity and affinity.²⁸ There are however few systematic evaluations, of the effects on metabolism or affinity more generally, of halogen substitution on a particular substrate scaffold.²⁶

Based on the results of this investigation, the major determinant of halogen substituent effects on ketamine analogs binding affinity to CYP2B6 is likely hydrophobicity. This conclusion is based on the excellent correlation between the calculated property AlogP and log K_m (R²=0.92, Figure 4). Thus hydrophobicity changes due to halogen substitution may be the simplest design criteria to use when considering similar compounds and the affinity for CYP2B6. There is also species evidence that proposes a hydrophobic element to CYP2B substrate selection, with variation in the hydrophobic active sites found between orthologs.²⁸ An unexpected truism is that halogens are generally hydrophobic atoms, with Br having the same hydrophobic effect as a methyl group.^{44, 45} This has been shown experimentally, where halogen substitutions had a drastic effect on the logP(octanol-water) of compounds and the larger the halogen the more dramatic the effect.⁴⁶ It has also been previously suggested that hydrophobic interactions are a major contribution in the interface between human protein kinase CK2α and a series of halogen-substituted benzotriazoles, and was directly linked with the hydrophobicity of the halogen substituent.⁴⁷ Inhibition increased as the electronegativity of the halide substituent decreased (IC₅₀: F<Cl<Br). Molecular modeling suggested that the substituted halogen forms direct interactions with hydrophobic residues within the binding pocket in a similar fashion to those found in the crystal structure of human pCDK2. These interactions are thought to stabilize the complex by protecting the hydrophobic halogen atoms from solvent accessibility.

The present results can be compared with some previous observations. Many other CYP2B6 substrates are also halogenated, including the drugs bupropion, efavirenz, ticlopidine, and amlodipine, and various environmental chemicals.⁴⁸ There has also been evidence that suggests halogen-π interactions may also play a role in CYP2B6-substrate interactions, with the crystal structure of CYP2B6 and its halogenated monoterpene bornyl bromide demonstrating bromine-π bonding with the aromatic side chain of phenylalanine 108.²⁸ Similar computational modeling of myrtenyl bromide suggested bromine-π bonding to the aromatic side chain of phenylalanine 297.²⁸ These results were interpreted to suggest that halogen-π interactions may be unique to CYP2B enzymes, which possesses a sufficiently malleable active site to enable halogen interactions with multiple phenylalanine residues.²⁸ A kinetic investigation probed the influence of halogenation of CYP2B6-catalyzed metabolism.⁴⁹ Using a small series of 4-trifluoromethyl-7-ethoxycoumarins to probe metabolism (O-dealkylation) by CYP2B6, it was found that 2-ethoxy chlorination, and more so bromination, moderately decreased the K_m, while k_cat was markedly reduced (by more than 90%) by either bromination or chlorination, resulting in net reduction (by 85–90%) in catalytic efficiency (k_cat/K_m).⁴⁹ These turnover results are in contrast to the present results with ketamine analogs. Nevertheless, together these combined observations may implicate the role of halogen bonding in the selectivity of ligand binding to CYP2B enzymes.

There is little other catalytic, and no clinical information, about metabolism of halogenated ketamine analogs to our knowledge. It has been suggested, albeit without any accompanying data, that deschloroketamine, a black market designer drug, be more potent and longer lasting than ketamine.⁵⁰ The results of the present study do suggest that slower metabolism of deschloroketamine might mean diminished systemic clearance. This presents potential opportunities for deriving new analogs in the future with improved pharmacokinetic properties. For example, analogs with reduced metabolism may have improved oral availability and/or greater duration of effect. Avid ketamine metabolism and low bioavailability may underlie the lack of analgesia of oral ketamine.⁵¹ Conversely, since ketamine antidepressant effects have been attributed to one or more active metabolites, such as norketamine and hydroxynorketamine,^{16, 17, 25} greater metabolite formation may enhance antidepressant effects. It is not apparent how chloro- (or more generally halogen) substitution alone influences interactions between ketamine and its drug targets.³

There are recognized limitations with the various computational approaches. The Libdock program used for docking only considers halogen H-bonding/π-stacking and did not include a halogen bond scoring function. Halogen bonding is very complex with many caveats, including directionality. The anisotropic electron distribution of halogens coupled with an aromatic ring adds an additional level of complexity, partially because it is impacted by the other electron withdrawing groups found on the ring. The binding pocket of CYP2B6 is large relative to a small molecule like ketamine making the docking very challenging.

CYP2B6 metabolizes 3–10% of therapeutic drugs, but is still less well characterized than many other CYP isoforms.^{34, 52–56} Monoterpenes have also been suggested as ideal active site probes for CYP2B6.^{28, 29, 57} The present results suggest that ketamine analogs may also be very useful probes, particularly for representatives of drugs as well as environmental contaminants such as pesticides and nerve agents and mitigating the toxicity of such molecules.⁵⁸ These ketamine analogs suggest unmet needs and opportunities to better understand substrate-CYP2B6 interactions and enable reliable future computational predictions.