RESEARCH ARTICLE
`
`Deuterium isotope effects in drug
`pharmacokinetics II: Substrate-dependence of
`the reaction mechanism influences outcome
`for cytochrome P450 cleared drugs
`
`Hao Sun(cid:13), David W. PiotrowskiID(cid:13), Suvi T. M. Orr, Joseph S. Warmus, Angela C. Wolford,
`Steven B. Coffey, Kentaro FutatsugiID, Yinsheng Zhang, Alfin D. N. Vaz(cid:13)
`
`Medicine Design, Pfizer Global Research and Development, Groton, Connecticut, United States of America
`
`* david.w.piotrowski@pfizer.com (DWP); hsun@seagen.com (HS); vazadn@gmail.com (ADNV)
`
`Abstract
`
`Two chemotypes were examined in vitro with CYPs 3A4 and 2C19 by molecular docking,
`metabolic profiles, and intrinsic clearance deuterium isotope effects with specifically deuter-
`ated form to assess the potential for enhancement of pharmacokinetic parameters. The
`results show the complexity of deuteration as an approach for pharmacokinetic enhance-
`ment when CYP enzymes are involved in metabolic clearance. With CYP3A4 the rate limit-
`ing step was chemotype-dependent. With one chemotype no intrinsic clearance deuterium
`isotope effect was observed with any deuterated form, whereas with the other chemotype
`the rate limiting step was isotopically sensitive, and the magnitude of the intrinsic clearance
`isotope effect was dependent on the position(s) and extent of deuteration. Molecular dock-
`ing and metabolic profiles aided in identifying sites for deuteration and predicted the possibil-
`ity for metabolic switching. However, the potential for an isotope effect on the intrinsic
`clearance cannot be predicted and must be established by examining select deuterated ver-
`sions of the chemotypes. The results show how in a deuteration strategy molecular docking,
`in-vitro metabolic profiles, and intrinsic clearance assessments with select deuterated ver-
`sions of new chemical entities can be applied to determine the potential for pharmacokinetic
`enhancement in a discovery setting. They also help explain the substantial failures reported
`in the literature of deuterated versions of drugs to elicit a systemic enhancement on pharma-
`cokinetic parameters.
`
`Introduction
`Because of the potential to enhance pharmacokinetic properties or decrease toxicity by virtue
`of a kinetic deuterium isotope effect, the replacement of hydrogen by deuterium at non-
`exchangeable carbon-hydrogen bonds of drug molecules has received extensive attention as
`indicated by an exponential increase over the past decade in patent applications for deuterated
`versions of existing pharmaceuticals and new chemical entities [1,2]. As reported previously
`
`a1111111111
`a1111111111
`a1111111111
`a1111111111
`a1111111111
`
`(cid:50)(cid:51)(cid:40)(cid:49) (cid:36)(cid:38)(cid:38)(cid:40)(cid:54)(cid:54)
`
`Citation: Sun H, Piotrowski DW, Orr STM, Warmus
`JS, Wolford AC, Coffey SB, et al. (2018) Deuterium
`isotope effects in drug pharmacokinetics II:
`Substrate-dependence of the reaction mechanism
`influences outcome for cytochrome P450 cleared
`drugs. PLoS ONE 13(11): e0206279. https://doi.
`org/10.1371/journal.pone.0206279
`
`Editor: Markos Leggas, University of Kentucky,
`UNITED STATES
`
`Received: June 14, 2018
`
`Accepted: October 10, 2018
`
`Published: November 14, 2018
`Copyright: (cid:139) 2018 Sun et al. This is an open
`access article distributed under the terms of the
`Creative Commons Attribution License, which
`permits unrestricted use, distribution, and
`reproduction in any medium, provided the original
`author and source are credited.
`
`Data Availability Statement: All relevant data are
`within the paper and its Supporting Information
`files.
`
`Funding: This work was funded by Pfizer. All
`authors were employed by Pfizer Inc at the time
`this work was done. Pfizer provided support in the
`form of salaries for all authors, but did not have
`any additional role in the study design, data
`collection and analysis, decision to publish, or
`preparation of the manuscript. The specific roles of
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`1 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 1
`
`

`

`these authors are articulated in the ‘author
`contributions’ section.
`
`Competing interests: All authors were paid
`employees of Pfizer Inc at the time this work was
`done. Pfizer Inc provided all funding for this
`research. Some of the authors are owners of PFE
`stocks or shares and are named inventors on Pfizer
`patents. This does not alter our adherence to PLOS
`ONE policies on sharing data and materials.
`
`KDIE CYP 3A4
`
`by us for aldehyde oxidase-cleared drugs, successful application of a deuteration strategy
`requires a clear understanding of the metabolic and systemic clearance mechanisms, and spe-
`cies differences in metabolic pathways [3].
`Cytochrome P450 enzymes (CYP) are responsible for over 90% of all metabolic clearance of
`drugs and xenobiotics, and three quarters of these reactions are attributable to five CYP iso-
`forms (1A2, 2C9, 2C19, 2D6, and 3A4) with CYP3A4 contributing approximately 27% to the
`metabolism of all marketed drugs [4]. Thus, any deuteration strategy must consider the com-
`plex reaction mechanisms of these enzymes that can confound a deuteration strategy leading
`to a failure to achieve significant systemic pharmacokinetic gain even when metabolism by
`these enzymes may be rate-limiting in systemic clearance [5–8]. Examples of such mechanistic
`complexity include: a) Differences in reaction mechanisms of C-H bond cleavages such as the
`N- and O- dealkylation reactions, where single electron transfer and hydrogen atom abstrac-
`tion mechanisms can have substantial differences in the magnitude of their intrinsic deute-
`rium isotope effect [9,10]; b) Deuterium-induced metabolic switching to proximal or distal
`non-deuterated sites [11,12] which is possibly due to multiple binding orientations of a mole-
`cule within an active site, or freedom for a bound molecule to “tumble” within an active site
`because of the large active site cavity of some CYP enzymes, such that oxidation at a non-deu-
`terated site compensates for decreased metabolism at the deuterated site resulting in loss of
`an isotope effect on the intrinsic clearance and a redistribution of the relative abundance of
`metabolites; and c) A rate limiting release of product resulting in masking of the intrinsic deu-
`terium isotope effect (Hk/Dk) on the intrinsic clearance (HVm/Km / DVm/Km) [13,14].
`In this study we examined two structurally distinct chemo-types (Fig 1, 1a and 2a) where
`in-vitro clearance predictions with hepatic microsomes and hepatocytes suggested a blood
`flow-limited, CYP-mediated oxidative metabolic clearance. Using virtual molecular docking
`with CYPs 3A4 and 2C19, metabolic profiles and intrinsic clearance isotope effects with
`human liver microsomes and recombinant CYPs 3A4 and 2C19 with deuterated versions of 1a
`and 2a, we demonstrate the mechanistic complexities of CYP-catalyzed reactions where the
`rate limiting step may be determined by the substrate under consideration. The two chemo-
`types examined also provide an understanding of how to address a deuteration strategy for
`new chemical entities, and helps explain the numerous reports where deuteration has been
`largely ineffective in substantially altering the in-vivo pharmacokinetics of some CYP-cleared
`compounds [5–8].
`
`Materials and methods
`The synthesis and characterization of chemotypes 1a and 2a have been previously reported
`[15,16]. Synthesis procedures for 1a and analytical data for deuterated analogs of 1a and 2a are
`presented in Supporting information (S1 File). The identities of primary metabolites from 1a
`and 2a were established from their mass spectral fragment patterns and are presented in Sup-
`porting information (S2 File).
`
`Molecular docking
`Structures of 1a and 2a were constructed using ChemBioOffice (PerkinElmer Inc. Waltham,
`MA) and stored in SD format. These structures were then modified with a customized script
`written in Python: the explicit hydrogen atoms were added, formal charge was calculated,
`and the structures were transformed into PDB format, with the integration of the OEChem
`Toolkit (OpenEye Scientific Software Inc., Santa Fe, NM). The derived molecular structures
`were further optimized with a DFT/B3LYP (Becke three-parameter Lee-Yang-Parr) approach
`using a 6-31G⇤⇤ basis set in Gaussian 09 (Gaussian, Inc., Wallingford, CT). The energetically
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`2 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 2
`
`

`

`KDIE CYP 3A4
`
`Fig 1. Structures of chemotypes 1a and 2a and their respective deuterated forms examined in this study.
`
`https://doi.org/10.1371/journal.pone.0206279.g001
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`3 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 3
`
`

`

`KDIE CYP 3A4
`
`minimized structures of 1a and 2a were then modified by AutoDockTools (The Scripps
`Research Institute, La Jolla, CA) with flexible torsions defined and Gasteiger atomic charges
`assigned, as the final ligand input files for docking. To prepare the protein templates, three-
`dimensional coordinates of CYP3A4 and CYP2C19 structures were collected from both
`Protein Data Bank and Pfizer’s Protein Structure Database. The protein templates were
`selected and customized specifically for 1a and 2a with previous findings [17]. Specifically, for
`CYP3A4, the core template was 3NXU, a crystal structure determined at resolution of 2.0 Å
`with the inhibitor ritonavir bound [18]; for CYP2C19, 4GQS, a crystal structure determined at
`resolution of 2.9 Å complexed with the inhibitor (2-methyl-1-benzofuran-3-yl)-(4-hydroxy-
`3,5-dimethylphenyl)methanone was used as the template [19]. These templates were further
`modified by AutoDockTools to add polar hydrogen atoms, Kollman partial charges, and solva-
`tion parameters. The partial charge of the iron (Fe) was assigned as 0.262 and the proximal
`oxygen (O) as -0.342, with the compound I Fe-O length assigned as 1.6 Å, according to previ-
`ously quantum mechanically derived heme parameters [20]. The active site space of CYP3A4
`and CYP2C19 was defined by AutoGrid 4.0 (The Scripps Research Institute), which pre-calcu-
`lates the van der Waals, hydrogen bonding, electrostatics, torsional, and solvation interactions
`between protein and studied compounds. Docking procedures were accomplished with Auto-
`Dock 4.0 (The Scripps Research Institute) on Pfizer’s high performance computing Linux
`clusters. The globally optimized conformation and orientation of compound 1a and 2a were
`searched using a Lamarckian generic algorithm, a hybrid of generic algorithms and an adap-
`tive local search method. The derived 100 docking poses for each compound were clustered
`according to RMSD (root-mean-square deviation). The binding poses with the lowest binding
`energies and within 5Å to the heme iron-oxo were automatically chosen by customized scripts
`for further analysis, and visualized using PyMOL (Schro¨dinger, LLC, New York, NY).
`
`Reactions with human liver microsomes and recombinant CYP enzymes
`for the assessment of metabolic profiles, relative CYP isoform activities,
`and intrinsic clearance isotope effects
`Microsomal and recombinant CYP reactions were conducted at 37 ˚C in final volumes of 1.0
`mL (for first order substrate depletion rate constant assessment at substrate concentrations of
`1.0 M), and 2.0 mL (for metabolic profiles at substrate concentrations of 10 M). Each reac-
`tion contained 100 mM potassium phosphate buffer pH 7.4, either 0.5 mg/mL human liver
`microsomal protein or 10 pmol/mL rCYP isoform co-expressed with cytochrome P450 oxi-
`do-reductase in insect cell membranes. Reactions were initiated by the addition of 3.0 mM
`NADPH or an NADPH regenerating system (0.3mM NADP+, 1 mM isocitrate, 0.5 mM
`MgCl2 and 1.0 unit isocitrate dehydrogenase).
`For metabolic profiles, reactions were incubated for 30 minutes at 37 ˚C then quenched by
`adding 5.0 mL of acetonitrile. After mixing, the samples were centrifuged at 1800 x g for 20
`minutes and the supernatants were decanted and dried at room temperature under reduced
`pressure in a vacuum centrifuge. The residues were re-suspended in 200 L of acetonitrile:
`DMSO:water (5:20:75), centrifuged as above to remove insoluble material and a 50 to 100 L
`aliquot of the supernatant was analyzed by LC/MS as described below.
`For relative activities of CYP isoforms and intrinsic clearance, substrate depletion at 1.0 M
`was the method of choice for assessment of the depletion rate constants as each substrate pro-
`duced multiple metabolites, and an estimation of Km for substrates (1a and 2a) by the sub-
`strate depletion method showed that their respective Km’s were greater than 5 M [21,22].
`Relative CYP activities were determined from the ratio of the depletion rate constant for each
`isoform relative to that for CYP3A4. Intrinsic clearance isotope effects were determined from
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`4 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 4
`
`

`

`KDIE CYP 3A4
`
`competitive reactions using 1:1 mixtures of the protio and appropriate deuterio form of the
`substrate at 0.5 M each. Eight 100 L aliquots each were removed over a period of 70 to 90
`minutes and added to 100 L of a 0.1 M solution of an internal standard in methanol to
`quench the enzymatic reaction. The samples were filtered through a high protein-binding
`capacity filter membrane in a 96-well format. The filtrates were evaporated in a vacuum
`centrifuge to near dryness, diluted with 125 L of water, and a 99 L aliquot was analyzed by
`reversed phase chromatography/mass spectrometry using selected reaction monitoring. Rate
`constants were determined from the semi-logarithmic plots of the time versus ratio of the area
`under the peak for specific transitions of the various substrates and the internal standard. The
`intrinsic clearance isotope effect was determined from the ratio of the rate constants for the
`protio- and respective deuterio- forms.
`
`LC-MS methods
`An integrated Thermo-Finnigan LC/MS system consisting of a Surveyor Autosampler, LC
`pump, diode array detector and either an Orbitrap or LCQ mass spectrometer auto-tuned
`with the protio-form of the compound of interest were used in all analytical work. Two chro-
`matographic conditions were used for analysis. For rate measurements, a steep linear gradient
`from 30 to 95% acetonitrile was used with a Phenomenex Luna C18, 5 m 2 x 50 mm column.
`For the identification of metabolites, a shallow gradient from 5 to 95% acetonitrile at a linear
`rate of 2.25% per minute was used with a Phenomenex Luna C-18, 3 m, 4.6 x 150 mm col-
`umn. The gradients used are shown in Tables 1 and 2.
`Metabolites were identified by standard techniques that include: extraction of ion masses
`from the total ion current corresponding to known metabolic transformations; identifying
`drug-derived substances by extracting ion masses from the MS2 and MS3 ion chromatograms
`that are common to the parents in their MS2 and MS3 spectra; and, examining fragment pat-
`terns of ions in the total ion current spectrum to determine if they are drug-related, in regions
`where UV (250–400 nm) absorbing peaks occurred.
`
`Results
`Molecular modeling
`Molecular docking studies of compound 1a with CYP3A4 showed two energetically favored
`binding clusters (clustered at RMSD 2.0 Å) where either the pyrimidino-piperidine ring (Fig 2
`Panel A) or the terminal N-ethyl moiety (Fig 2 Panel B) of compound 1a are in proximity to
`the heme iron of CYP3A4 for aliphatic hydroxylations at the pyrimidino-piperidine ring and
`N-deethylation reactions, respectively. Other binding clusters that were also energetically
`favored (1–2 kcal/mol within the lowest-energy binding cluster) but not in an orientation for a
`typical P450-catalyzed reaction, were not included for analysis, an approach we previously
`reported [23–27]. For example, the binding pose with the trifluoromethyl moiety of 1a closest
`
`Table 1. Steep gradient for LC/MS analysis.
`
`Time (min)
`0
`2.0
`5.0
`6.0
`6.5
`8.0
`
`0.1% formic acid in water
`70
`70
`5
`5
`95
`95
`
`Acetonitrile
`30
`30
`95
`95
`5
`5
`
`Flow rate ( L/min)
`500
`500
`500
`500
`500
`500
`
`https://doi.org/10.1371/journal.pone.0206279.t001
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`5 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 5
`
`

`

`KDIE CYP 3A4
`
`Table 2. Shallow gradient for LC/MS analysis.
`
`Time (min)
`0
`5
`45
`50
`52
`60
`
`0.1% formic acid in water
`95
`95
`5
`5
`95
`95
`
`Acetonitrile
`5
`5
`95
`95
`5
`5
`
`Flow rate ( L/min)
`500
`500
`500
`500
`500
`500
`
`https://doi.org/10.1371/journal.pone.0206279.t002
`
`to the heme may also be energetically favored, but no common P450-catalyzed reactions are
`expected. For 1a, importantly, the three saturated carbon atoms on the piperidine ring are at a
`comparable distance to the putative CYP heme iron oxo species (3–4 Å, only one representa-
`tive pose from each binding cluster was selected for Fig 2). These two molecular orientations
`of 1a within the active site allow for either pyrimidino-piperidine ring hydroxylations or N-
`deethylation, predicting these to be major metabolic products from 1a. Due to side chain
`
`Fig 2. Snapshots of docked 1a to the active site of CYP3A4 (Panels A and B), and 2a to the active sites of CYP 3A4
`and CYP 2C19 (Panels C and D respectively). Panels A and B show two discrete binding modes for compound 1a in
`the active site of CYP3A4 with either the pyrimidino-piperidine ring (Panel A), or the N-ethyl (Panel B) in proximity
`to the putative heme iron-oxo species. The arrows in Panels C and D show the N- and O-methyl groups of 2a. Panel C
`shows the binding of 2a to the active site of CYP3A4 with both methyl groups in similar proximity to the putative
`oxidant in CYP3A4, and Panel D shows the preferred binding mode with CYP2C19, where the O-methyl group of 2a
`is in proximity to the putative oxidant.
`
`https://doi.org/10.1371/journal.pone.0206279.g002
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`6 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 6
`
`

`

`KDIE CYP 3A4
`
`interactions within the active site, these binding modes are discreet and cannot interconvert
`by “tumbling” of 1a within the active site. Thus, any shift from N-deethylation to pyrimidino-
`piperidine ring hydroxylation requires substrate dissociation from the enzyme and appropriate
`rebinding at the heme center.
`With 2a, molecular docking studies showed that both N-methyl and O-methyl moieties of
`compound 2a equally access the heme iron-oxo of CYP3A4 in an energetically favored orien-
`tation (Fig 2 Panel C), with the distance to the iron-oxo of 3.2 and 2.9Å, respectively. Anionic
`and polar active site residues including E308 and S312 of helix I, and Q484 of the C-terminal,
`together with other hydrophobic residues especially from the B-C and K-(cid:268) loops, might play
`important roles to juxtapose the N-methyl and O-methyl moieties of compound 2a in proxim-
`ity to the heme iron for catalysis, and indicates that the hydrogen atoms from either methyl
`groups have an equal access to the active oxidant of CYP3A4. Thus, the ratio of products (N-
`demethylation vs. O-demethylation) is more likely determined by the intrinsic reactivity and
`reaction mechanism of these two types of methyl groups (electron transfer/proton loss for N-
`demethylation or hydrogen atom abstraction for O-demethylation). The proximity of both
`methyl groups to the active oxidant of CYP3A4 predicts metabolic switching. Molecular
`modeling of compound 2a with CYP2C19 presents a different picture. Binding of 2a to
`CYP2C19 within its active site space is similarly confined by the I helix, F-G, B-C, and K-(cid:268)
`loops, but the binding pocket of CYP2C19 is much more restricted than that of CYP3A4,
`preventing both methyl groups from positioning at equal distances to the active oxidant of
`CYP2C19. The most favored binding pose for compound 2a was with the O-methyl moiety
`over the heme (3.0Å, Fig 2 Panel D). This was mainly driven by some hydrophobic interactions
`including those between F100 of the B-C loop, F476 of the C-terminal and the pyrazole and
`cyclopropyl moieties of compound 2a. This preferred binding mode predicts O-demethylation
`to be the primary route of metabolism of this substrate with CYP2C19 and if the rate limiting
`step for this substrate is hydrogen atom abstraction, an isotope effect should be expected pri-
`marily when the O-methyl group is deuterated (2b and 2d) and little to no effect when only
`the N-methyl group is deuterated (2c).
`
`Identity of metabolites from 1a and 2a
`Mass spectral data for the characterization of primary oxidation metabolites of 1a and 2a are
`presented in Supporting Information (S2 File). Fourteen metabolites were identified from 1a
`in human liver microsomes (Fig 3, panel HLM). Metabolites M1a1-M1a6 result from primary
`oxidative reactions, and metabolites M1a7-M1a14 result from secondary oxidation of the pri-
`mary oxidative products. A scheme for the oxidative metabolism of 1a in human liver micro-
`somes is shown in (Fig 4). Metabolite profiles with recombinant CYP isoforms showed that
`CYP3A4 formed the six primary metabolites, M1a1-M1a6 (Fig 3, panel CYP3A4). CYPs 2C9,
`2C19, 2D6 and 1A2 formed the N-desethyl metabolite (M1a1) that was visible as a UV peak in
`the chromatograms (Fig 3 panels CYP2C9, CYP2C19, CYP2D6, and CYP1A2). Of six recom-
`binant forms of CYP-enzymes examined, CYP 3A4 was the most active (Table 3). Although
`CYP 2C9 and 2C19 appear to be more active than CYP 3A4 in the N-deethylation of 1a based
`on the UV signal (Fig 3, Panels CYP2C9 and CYP2C19), their contributions to overall clear-
`ance is minor when assessed by disappearance of substrate (Table 3) and when normalized to
`their levels in human liver microsomes relative to CYP3A4.
`Three metabolites were identified from 2a in human liver microsomes. The primary oxida-
`tive demethylation products, N-desmethyl-2a (M2a1), the O-desmethyl-2a (M2a2), and the
`secondary metabolic product (M2a3) derived from either primary metabolite by further
`oxidative demethylation (Fig 5, panel HLM). Metabolite profiling with r-CYPs showed that
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`7 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 7
`
`

`

`KDIE CYP 3A4
`
`Fig 3. UV (300–400 nm) chromatograms for the metabolism of 1a in human liver microsomes and recombinant
`cytochromes 3A4, 2C9, 2C19, 2D6, and 1A2.
`
`https://doi.org/10.1371/journal.pone.0206279.g003
`
`CYP3A4 formed both primary metabolites with higher activity towards forming M2a1 than
`M2a2 (Fig 5, panel CYP3A4), whereas CYP2C19 was more active in forming M2a2 than M2a1
`(Fig 5, panel CYP2C19). CYPs 2C9 and 2C8 produced only trace amounts of M2a2. Fig 6
`shows a scheme for the metabolism of 2a. As shown in Table 3, CYP3A4 was the most active
`of the six recombinant CYP-isoforms examined. CYP2C19 showed 24% of the activity of
`CYP3A4 and CYPs 2C9, 2D6, and 2C8, were essentially inactive. When normalized to their
`
`Fig 4. Scheme for metabolites of 1a observed with human liver microsomes.
`
`https://doi.org/10.1371/journal.pone.0206279.g004
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`8 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 8
`
`

`

`KDIE CYP 3A4
`
`Table 3. Activities of recombinant human cytochrome P450 isoforms towards compounds 1a and 2a.
`
`CYP isoform
`CYP3A4
`CYP1A2
`CYP2C9
`CYP2C19
`CYP2D6
`CYP2C8
`
`kobs(min-1)
`0.663
`0.009
`0.050
`0.071
`0.015
`0.001
`
`https://doi.org/10.1371/journal.pone.0206279.t003
`
`1a
`
`2a
`
`As % of CYP3A4
`100
`1
`7
`11
`2
`0
`
`kobs(min-1)
`0.059
`<0.001
`<0.001
`0.014
`<0.001
`<0.001
`
`As % of CYP3A4
`100
`0
`0
`24
`0
`0
`
`respective levels in liver microsomes, CYP3A4 is the primary isoform responsible for micro-
`somal clearance of 2a.
`Accordingly, any strategy to address metabolic clearance by deuterium substitution in
`either chemotype requires a mechanistic consideration of the CYP3A4 reaction. As CYP2C19
`showed 24% of the activity relative to CYP3A4 for 2a, with the same metabolic profile but
`opposite preferences for the two metabolites, this isoform was also examined with 2a and its
`deuterated forms to compare mechanistic similarities and differences from CYP3A4.
`
`Intrinsic clearance kinetic deuterium isotope effects and metabolite
`profiles with 1b-1f and 2b-2d
`Results with chemotypes 1a, 2a and their deuterated forms (1b–1f and 2b–2d) show how the
`chemo-type influences the course of reactions with CYP isoforms and their rate limiting steps.
`
`Fig 5. UV (300–400 nm) chromatograms for the metabolism of 2a in human liver microsomes and recombinant cytochromes 3A4, 2C9,
`2C19, 2D6, and 1A2.
`
`https://doi.org/10.1371/journal.pone.0206279.g005
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`9 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 9
`
`

`

`KDIE CYP 3A4
`
`Fig 6. Scheme for metabolites of 2a observed with human liver microsomes.
`
`https://doi.org/10.1371/journal.pone.0206279.g006
`
`Table 4 summarizes the intrinsic clearance kinetic deuterium isotope effects observed for deu-
`terated forms of 1a (1b–1f) with human liver microsomes and recombinant CYP3A4. The lack
`of a deuterium isotope effect on the intrinsic clearance for any of the deuterated forms of 1a
`with either enzyme system, indicates that there is no overall decrease in metabolism of this
`chemotype as a function of deuteration. Fig 7 shows the extracted ion chromatographic pro-
`files for the identified metabolites from 1a-1f and Table 5 shows their percentages relative to
`total metabolites. For each deuterated form, the relative abundance of each metabolite is an
`indication of the extent to which deuterium substitution has influenced metabolic switching.
`The pattern that emerges is complex. When deuterium is present at the methylene carbon of
`
`Table 4. Kinetic deuterium isotope effect on intrinsic clearance for deuterated forms of 1a with human liver
`microsomes and recombinant CYP3A4.
`
`HCLint / DCLint
`
`Compound
`(1b)
`(1c)
`(1d)
`(1e)
`(1f)
`
`HLM
`1.0
`1.0
`1.0
`1.0
`1.1
`
`CYP3A4
`1.0
`1.1
`1.2
`1.0
`1.2
`
`https://doi.org/10.1371/journal.pone.0206279.t004
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`10 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 10
`
`

`

`KDIE CYP 3A4
`
`Fig 7. Extracted ion chromatograms for primary oxidative metabolites of 1a and its various deuterated forms
`with r-CYP3A4.
`
`https://doi.org/10.1371/journal.pone.0206279.g007
`
`the N-ethyl group (1b, 1c, 1d, and, 1f), N-deethylation is decreased between 1.4 and 2.5 fold.
`This decrease depends on the pattern of deuterium substitutions at other sites. With 1b and 1d
`where deuterium is present at the methylene of the ethyl group and additionally in 1d at the
`methylenes adjacent to the piperidine ring nitrogen, the decrease in N-deethylation is compa-
`rable but metabolic switching occurs to hydroxylation sites at the pyrimidino-piperidine ring
`of 1d. For 1b switching is primarily to the benzylic methylene carbon adjacent to the piperidi-
`nyl nitrogen (M1b3), whereas for 1d it is to the deuterated methylene carbon adjacent to the
`piperidine nitrogen (M1d2). With 1c (where the N-ethyl group is per-deuterated) and 1f
`(where the methylene of the N-ethyl is deuterated and the piperidine ring is per-deuterated),
`
`Table 5. Percentage of primary oxidative metabolites from deuterated forms of 1a with r-CYP3A4.
`
`(1a)
`
`(1b)
`
`(1c)
`
`(1d)
`
`(1e)
`
`(1f)
`
`Substrate
`
`Metabolite
`M1x11
`M1x2
`M1x3
`M1x4
`M1x5
`M1x6
`
`18
`16
`13
`3
`25
`26
`
`1 “x” refers to the form of substrate.
`
`https://doi.org/10.1371/journal.pone.0206279.t005
`
`9
`16
`23
`2
`26
`25
`
`Metabolite as % of total metabolites
`13
`7
`13
`33
`22
`12
`2
`4
`30
`31
`19
`13
`
`20
`21
`5
`4
`14
`36
`
`13
`18
`11
`3
`35
`20
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`11 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 11
`
`

`

`KDIE CYP 3A4
`
`the decreases in N-deethylation are comparable but are less than with 1b and 1d. The metabolic
`switching is also distinct. For 1c the metabolic switch occurs to the benzylic positions M1c3
`and M1c5, and for 1f the major metabolic switch is to the benzylic methylene M1f5 (deuter-
`ated). With 1e where deuterium substitution is only at the methylenes adjacent to the piperi-
`dine nitrogen, the N-deethylation is only slightly increased, whereas the hydroxylation profile
`is substantially altered. Large decreases in M1e3 (a deuterated site) and M1e5 (a non-deuter-
`ated site) and increases in M1e2 (a deuterated site) and M1e6 (an isotope insensitive site) are
`observed. These results are interpreted in light of the two discrete binding modes revealed by
`molecular modeling (Fig 2 Panel A and Fig 2 Panel B). N-Deethylation occurs only from the
`binding mode shown in Fig 2 Panel B where the methylene of the N-ethyl is proximal to the
`activated heme iron-oxo complex. Deuteration of the methylene results in a decreased N-
`deethylation and increased oxidations on the pyrimidino-piperidine ring, consistent with met-
`abolic switching between the two binding modes. By contrast, in the other binding cluster (Fig
`2 Panel A) where the pyrimidino-piperidine ring is above the heme iron-oxo complex, the dif-
`ferences in metabolic product distribution observed for the various deuterated forms suggests
`subtle differences in orientation of the pyrimidino-piperidine ring as a consequence of deute-
`rium in the different deuterated forms. In some instances, increases are observed in metabolites
`where C-D bonds are broken (M1e2 from 1e, and M1d2/M1d5 from 1d) which suggests higher
`constraints on substrates when bound in this binding cluster (Fig 2 Panel A). More impor-
`tantly, the results suggest a high degree of ‘commitment to catalysis’ [27] in this binding cluster
`such that oxidation occurs at those sites that are most proximal to the active oxidant irrespec-
`tive of them being deuterated, and the rate limiting step is past the catalytic event.
`Since CYP2C19 showed 24% of the activity of CYP3A4 with chemotype 2a (Table 3), and it
`has an active site substantially smaller than that of CYP3A4, isotope effects and product pro-
`files were examined for comparison with both isoforms. Table 6 summarizes the intrinsic
`clearance deuterium isotope effect observed for 2b–2d, with human liver microsomes, r-
`CYP3A4, and r-CYP2C19. With 2d, where both methyl groups are deuterated, the isotope
`effect with all enzymatic systems is the largest. This indicates that the rate limiting step for this
`chemotype is C-H bond cleavage of either or both methyl groups. Changes in the magnitude
`of the isotope effect with r-CYPs 2C19 and 3A4 for substrates 2b and 2c where either methyl
`group is deuterated, indicates the complexity of isoform-dependent CYP mechanisms. For r-
`CYP 2C19 where O-demethylation is the major metabolic route (Fig 8, Panel 2a), the isotope
`effect on intrinsic clearance is 4.0 with 2b where only the O-methyl group is deuterated and
`4.5 with 2d where both methyl groups are deuterated (Table 6), whereas the isotope effect is
`lost when only the N-methyl group is deuterated (2c, Table 6). The isotope effects are mirrored
`in the metabolic profiles shown in Fig 8. When only the O-methyl group is deuterated (2b) the
`N-des-methyl metabolite (M2b1) level is slightly higher than that from the non-deuterated
`form (2a, Fig 8 Panels 2a and 2b), suggesting metabolic switching. Whereas the N-des-methyl
`metabolites (M2c1 and M2d1) are essentially lost when the N-methyl groups in 2c and 2d
`are deuterated (Fig 8, Panels 2c and 2d). These metabolic profiles and isotope effects are
`
`Table 6. Kinetic deuterium isotope effect on intrinsic clearance for deuterated forms of 2a with human liver
`microsomes, r-CYP3A4 and r-CYP2C19.
`
`Compound
`(2b)
`
`(2c)
`
`(2d)
`
`HLM
`1.3
`
`1.5
`
`3
`
`https://doi.org/10.1371/journal.pone.0206279.t006
`
`HK / DK
`CYP3A4
`1.2
`
`2.9
`
`8.3
`
`CYP2C19
`4.0
`
`1.0
`
`4.5
`
`PLOS ONE | https://doi.org/10.1371/journal.pone.0206279 November 14, 2018
`
`12 / 17
`
`Auspex Exhibit 2002
`Apotex v. Auspex
`IPR2021-01507
`Page 12
`
`

`

`KDIE CYP 3A4
`
`Fig 8. Extracted ion chromatograms for primary oxidative metabolites of 2a and its various deuterated forms
`(shown in Fig 1) with r-CYP2C19.
`
`https://doi.org/10.1371/journal.pone.0206279.g008
`
`consistent with the low contribution of the M2a1 metabolic pathway to the overall metabolism
`by CYP2C19.
`For CYP3A4 where N-demethylation (M2a1) is the preferred metabolic route (Fig 5 Panel
`3A4, and Fig 9 Panel 2a), deuteration of only the O-methyl (2b), or N-methyl (2c) groups
`result in a decrease in the intrinsic clearance isotope effect when compared to deuteration at
`
`Fig 9. Extracted ion chromatograms for p