Biotransformation of daclatasvir in vitro and in nonclinical species: Formation of the main metabolite by pyrrolidine d-Oxidation and Rearrangements

Article abstract of DOI:10.1124/dmd.115.068866

Daclatasvir is a first-in-class, potent, and selective inhibitor of the hepatitis C virus nonstructural protein 5A replication complex. In support of nonclinical studies during discovery and exploratory development, liquid chromatography-tandem mass spectrometry and nuclear magnetic resonance were used in connection with synthetic and radiosynthetic approaches to investigate the biotransformation of daclatasvir in vitro and in cynomolgus monkeys, dogs, mice, and rats. The results of these studies indicated that disposition of daclatasvir was accomplished mainly by the release of unchanged daclatasvir into bile and feces and, secondarily, by oxidative metabolism. Cytochrome P450s were the main enzymes involved in the metabolism of daclatasvir. Oxidative pathways included d-oxidation of the pyrrolidine moiety, resulting in ring opening to an aminoaldehyde intermediate followed by an intramolecular reaction between the aldehyde and the proximal imidazole nitrogen atom. Despite robust formation of the resulting metabolite in multiple systems, rates of covalent binding to protein associated with metabolism of daclatasvir were modest (55.2-67.8 pmol/mg/h) in nicotinamide adenine dinucleotide phosphate (reduced form)-supplemented liver microsomes (human, monkey, rat), suggesting that intramolecular rearrangement was favored over intermolecular binding in the formation of this metabolite. This biotransformation profile supported the continued development of daclatasvir, which is now marketed for the treatment of chronic hepatitis C virus infection.

Full text of DOI:10.1124/dmd.115.068866

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Biotransformation of Daclatasvir in Vitro and in Nonclinical Species: Formation of
the Main Metabolite by Pyrrolidine -Oxidation and Rearrangement
Wenying Li, Weiping Zhao, Xiaohong Liu, Xiaohua Huang, Omar D. Lopez, John E.
Leet, R. Marcus Fancher, Van Nguyen, Jason Goodrich, John Easter, Yang Hong, Janet
Caceres-Cortes, Shu Y. Chang, Li Ma, Makonen Belema, Lawrence G. Hamann, Min
Gao, Mingshe Zhu, Yue-Zhong Shu, W. Griffith Humphreys, and Benjamin M. Johnson
Departments of Pharmaceutical Candidate Optimization (W.L., W.Z., X.L., X.H., R.M.F.,
J.C.-C., S.Y.C., L.M., M.Z., Y.-Z.S., W.G.H., B.M.J.); Synthesis and Analysis Technology
Team (J.E.L.); Virology (M.G.) and Chemistry (O.D.L., V.N., J.G., J.E., Y.H., M.B.,
L.G.H.); Bristol-Myers Squibb Company, Wallingford, Connecticut
1

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Running title: Biotransformation of daclatasvir
Address correspondence to: Dr. Benjamin M. Johnson, Pharmaceutical Candidate
Optimization, Bristol-Myers Squibb Company, 5 Research Parkway, Wallingford, CT
0
6492. email: benjamin.m.johnson@bms.com.
Number of text pages: 53
Number of tables: 6
Number of figures: 6
Number of references: 37
Number of words in the Abstract: 198
Number of words in the Introduction: 671
Number of words in the Discussion: 1482
Nonstandard abbreviations: 1D, one-dimensional; 2D, two-dimensional; BDC, bile-
duct cannulated; CID, collision-induced dissociation; CV, collision voltage; DP,
declustering potential; FID, free induction decay; GSH, glutathione; HCV, hepatitis C
n
virus; HPLC, high-performance liquid chromatography; INF- , interferon- ; LC-MS ,
liquid chromatography-tandem mass spectrometry; m/z, mass-to-charge ratio; MS, mass
spectrometry; NADPH, nicotinamide adenine dinucleotide phosphate (reduced form);
NMR, nuclear magnetic resonance; NOE, nuclear Overhauser enhancement; NS5A,
nonstructural protein 5A; PEG-INF- , pegylated interferon- ; RNA, ribonucleic acid;
TRA, total radioactivity analysis; VAP, vascular access port
2

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Abstract
Daclatasvir is a first-in-class, potent and selective inhibitor of the hepatitis C virus
nonstructural protein 5A replication complex. In support of nonclinical studies during
discovery and exploratory development, liquid chromatography–tandem mass
spectrometry and NMR were used in connection with synthetic and radiosynthetic
approaches to investigate the biotransformation of daclatasvir in vitro and in the
cynomolgus monkey, dog, mouse, and rat. The results of these studies indicated that
disposition of daclatasvir was accomplished mainly by release of unchanged daclatasvir
into bile and feces, and, secondarily, by oxidative metabolism. Cytochromes P450 were
the main enzymes involved in the metabolism of daclatasvir. Oxidative pathways
included -oxidation of the pyrrolidine moiety, resulting in ring opening to an
aminoaldehyde intermediate followed by an intramolecular reaction between the
aldehyde and the proximal imidazole nitrogen atom. Despite robust formation of the
resulting metabolite in multiple systems, rates of covalent binding to protein associated
with metabolism of daclatasvir were modest (55.2 – 67.8 pmol/mg/h) in NADPH-
supplemented liver microsomes (human, monkey, rat) suggesting that intramolecular
rearrangement was favored over intermolecular binding in the formation of this
metabolite. This biotransformation profile supported the continued development of
daclatasvir, which is now marketed for the treatment of chronic HCV infection.
3

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Introduction
Hepatitis C virus (HCV) infection is a global health problem that affects an
estimated 130 – 170 million people, representing 2.2 – 3.0% of the world’s population
Lavanchy, 2009). The rise of HCV infection to this level of prevalence occurred in the
0th century as a result of person-to-person transmission via parenteral exposure to
(
2
contaminated blood and medical equipment. Developed countries in North America and
Europe have maintained controls since the 1980s that have limited the transmission of
HCV during blood transfusions. However, medically related modes of transmission have
been a problem in developing countries and have contributed to even higher infection
rates in Africa, Southeastern Europe and the Far East. Chronic HCV infections may
progress to cirrhosis, hepatocellular carcinoma and/or death, with mortality rates differing
among industrialized and developing countries (The Global Burden of Hepatitis C
Working Group, 2004).
In the past several years, new options to treat chronic HCV infection have
emerged that are replacing the need for therapies based on pegylated interferon- (PEG-
IFN- ), which are accompanied by unwanted side effects (Rodriguez-Torres et al., 2013).
For example, insights into the workings of the HCV replication complex have resulted in
direct-acting antivirals against the multiple viral targets. For example, the HCV NS3
protease is one such target that has been the subject of drug discovery efforts (McPhee et
al., 2012). New medicines targeting the HCV NS5B polymerase have also been
developed, including nucleotide-based inhibitors such as sofosbuvir (Sofia et al., 2010)
and non-nucleotide analogs such as dasabuvir that target one of several allosteric sites on
the NS5B protein (Lesburg et al., 1999; Bressanelli et al., 1999). Combinations of such
4

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
medicines have been the subject of clinical studies to determine which combinations are
most effective against the different genotypes of HCV, and which might be appropriate
for use in patient populations that are difficult to treat. Inhibitors of the non-structural
protein 5A (NS5A) represent another important class of direct-acting antivirals that is
emerging as a key option in the treatment of such patients (Piroth et al., 2015). While the
functionality of NS5A is not fully understood, it is clearly an important component of the
HCV RNA replicase complex (Blight et al., 2000). Subgenomic, non-infectious replicon
systems enabling the study of this complex in a cellular environment have been
developed (Lohmann et al., 1999). NS5A has been characterized as a protein with three
discrete structural domains and an amphipathic -helix that tethers the protein to the
membrane bilayer (Penin et al., 2004). Of these domains, domain I is the best
characterized and appears to be the most directly involved in RNA replication. This
domain includes a zinc-binding motif that must be coordinated to a zinc atom for RNA
replication to occur, and a basic groove – possibly for RNA binding (Tellinghuisen et al.,
2
004). Crystal structures of this domain have been reported, providing complementary
information on its structure (Tellinghuisen et al., 2005; Love et al., 2009; Lambert et al.,
014). It has also been postulated that a higher order of NS5A structure exists, explaining
2
both the exquisite potency of NS5A inhibitors and resensitization of resistant NS5A
variants by NS5A synergists (Sun et al., 2015). As a result of its demonstrated critical
role in HCV replication, NS5A represents an attractive target for anti-HCV therapy.
Daclatasvir (dimethyl (2S,2'S)-1,1'-((2S,2'S)-2,2'-(4,4'-(biphenyl-4,4'-diyl)bis(1H-
imidazole-4,2-diyl))bis(pyrrolidine-2,1-diyl))bis(3-methyl-1-oxobutane-2,1-
diyl)dicarbamate; Fig. 1) is a first-in-class, selective inhibitor of the NS5A replication
5

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
complex, which evolved from a series of dimeric compounds possessing C
Belema et al., 2014). It is among the most potent anti-HCV drugs reported thus far, with
2
symmetry
(
EC50s against the wild-type 1a and 1b replicons of 50 pM and 9 pM, respectively (Gao et
al., 2010). Combinations of direct-acting antivirals including daclatasvir have
demonstrated high cure rates in the clinic, even among difficult-to-treat populations
including those infected with genotype 3, patients with cirrhosis or fibrosis, previous non-
responders to IFN/ribavirin-based regimens and patients coinfected with HIV (Nelson et
al., 2015; Lok et al., 2012; Chayama et al., 2012; Wyles et al., 2015). Here we report the
preclinical biotransformation of daclatasvir, including a full structural characterization of
key metabolites.
6

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Materials and Methods
3
3
Substances and Enzymes Daclatasvir (BMS-790052; 1), [ H]-daclatasvir {[ H]-
14
14
1
}, [ C]-daclatasvir {[ C]-1} (Easter et al., submitted), and compounds 2 (methyl ((1S)-
1
-(((2S)-2-(4-(4'-(2-((2S)-1-((2S,3R)-4-hydroxy-2-((methoxycarbonyl)amino)-3-
methylbutanoyl)-2-pyrrolidinyl)-1H-imidazol-4-yl)-4-biphenylyl)-1H-imidazol-2-yl)-1-
pyrrolidinyl)carbonyl)-2-methylpropyl)carbamate), 3 (methyl ((1S)-1-(((8S)-5-hydroxy-
2
1
-(4'-(2-((2S)-1-((2S)-2-((methoxycarbonyl)amino)-3-methylbutanoyl)-2-pyrrolidinyl)-
H-imidazol-4-yl)-4-biphenylyl)-5,6,7,8-tetrahydroimidazo[1,2-a]pyridin-8-
yl)carbamoyl)-2-methylpropyl)carbamate), 4 (methyl (S)-1-((S)-2-(5-(4'-(2-((S)-1-((S)-2-
amino-3-methylbutanoyl)pyrrolidin-2-yl)-1H-imidazol-5-yl)biphenyl-4-yl)-1H-imidazol-
2
-yl)pyrrolidin-1-yl)-3-methyl-1-oxobutan-2-ylcarbamate), 6 (two diastereomers: methyl
N-[1-(2-{5-[4-(4-{2-[(2S,4S)-4-hydroxy-1-[(2S)-2-[(methoxycarbonyl)amino]-3-
methylbutanoyl]pyrrolidin-2-yl]-1H-imidazol-5-yl}phenyl)phenyl]-1H-imidazol-2-
yl}pyrrolidin-1-yl)-3-methyl-1-oxobutan-2-yl]carbamate and methyl N-[1-(2-{5-[4-(4-
{2-[(2S,4R)-4-hydroxy-1-[(2S)-2-[(methoxycarbonyl)amino]-3-
methylbutanoyl]pyrrolidin-2-yl]-1H-imidazol-5-yl}phenyl)phenyl]-1H-imidazol-2-
yl}pyrrolidin-1-yl)-3-methyl-1-oxobutan-2-yl]carbamate), 7 (two diastereomers: methyl
N-[(2S)-1-[(2R,3S)-3-hydroxy-2-{4-[4-(4-{2-[(2S)-1-[(2S)-2-[(methoxycarbonyl)amino]-
3
-methylbutanoyl]pyrrolidin-2-yl]-1H-imidazol-4-yl}phenyl)phenyl]-1H-imidazol-2-
yl}pyrrolidin-1-yl]-3-methyl-1-oxobutan-2-yl]carbamate and methyl N-[(2S)-1-[(2R,3R)-
-hydroxy-2-{4-[4-(4-{2-[(2S)-1-[(2S)-2-[(methoxycarbonyl)amino]-3-
3
methylbutanoyl]pyrrolidin-2-yl]-1H-imidazol-4-yl}phenyl)phenyl]-1H-imidazol-2-
yl}pyrrolidin-1-yl]-3-methyl-1-oxobutan-2-yl]carbamate), and 8 (2-(4'-(2-((S)-1-
7

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
(methoxycarbonyl)-L-valyl)pyrrolidin-2-yl)-1H-imidazol-5-yl)-[1,1'-biphenyl]-4-yl)-2-
(
oxoacetic acid) were synthesized chemically at Bristol-Myers Squibb Co. (Wallingford,
CT). The remaining metabolites 5, 9 - 23, and 25 - 26 were not matched against
standards. Chemical reagents including glutathione (GSH), NADPH, amodiaquine,
diclofenac, dextromethorphan, midazolam, bucetine, clotrimazol, quinidine, desethyl-
amodiaquine, 4’-hydroxydiclofenac, dextrorphan, 6 -hydroxytestosterone, and 1’-
hydroxymidazolam were purchased from Sigma-Aldrich (St. Louis, MO) unless noted
otherwise. Tacrine, buproprion, S-mephenytoin, hydroxybupropion, fluvoxamine, and
2
[
(
H ]-N-desethylamodiaquine were purchased from Toronto Research Chemicals
5
2
2
Toronto, ON). 4’-Hydroxymephenytoin, [ H
6
]-hydroxybupropion, [ H ]-4’-
3
2
hydroxymephenytoin, and [ H
3
]-dextrorphan were purchased from BD Biosciences
2
(
Woburn, MA). [ H
3
]-6 -Hydroxytestosterone was procured from Cerilliant (Round
Rock, TX), and 1’-hydroxytacrine was obtained from TLC PHARMECHEM
Mississauga, ON). All of the solvents [high-performance liquid chromatography (HPLC)
(
grade] were purchased from Mallinckrodt Baker (Phillipsburg, NJ). Pooled liver
microsomes (male cynomolgus monkey, lot 23791, n = 150; male beagle dog, lot 00269,
n = 7; male Sprague Dawley rat, lot 85157, n = 164; male human, lot 382898, n = 150)
were obtained from BD Gentest (Woburn, MA). Pooled liver S9 (male balb B/C mouse,
lot 0810018, n = 200; male beagle dog, lot 90958, n = 6) was purchased from BD
Gentest. Pooled liver S9 (cynomolgus monkey, lot 0910273, n = 12; human, lot 0910396,
n = 50) was obtained from XenoTech (Lenexa, KS). Pooled liver S9 (Aroclor-1254-
induced male Sprague Dawley rat, lot 2572, n = 35) was purchased from Molecular
Toxicology, Inc (Boone, NC). Pooled cryopreserved hepatocytes (human, lots GMK,
8

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
KKO, TNM; rat, lot QSV, n = 20; ICR/CD-1 mouse, lots ONJ, SFZ, AFZ, n = 3; beagle
dog, lot OES, n = 5; male cynomolgus monkey, lots OVW, VCE, OAY, n = 3) were
14
obtained from In Vitro Technologies (Baltimore, MD). Compound [ C]-1 was prepared
3
as described (Easter et al., submitted). Compound [ H]-1 was prepared by dissolving 1
into acetic acid and treating with 4 equivalents of bromine (1M) in acetic acid. The
reaction was stirred overnight and concentrated under vacuum. The resulting dibromo
3
analog was converted to [ H]-1 via debromination with carrier-free tritium gas. The final
3
product [ H]-1 was isolated by HPLC to a purity of 99.3% and a specific radioactivity of
2
4.5 Ci/mmol.
Organic Synthesis of Key Metabolites Procedures outlining the synthesis of
metabolites are provided as supplemental material.
14
Incubations in Liver Microsomes and Liver S9 Compound [ C]-1 (10 μM)
was incubated together with glutathione (GSH; 5 mM) and either liver microsomes (1 mg
protein/mL) or liver S9 (2 mg protein/mL) in potassium phosphate buffer (0.1 M, pH 7.4,
1
mL total volume) for 1 h at 37 °C with constant shaking. Alternatively, 1 (10 μM) was
incubated together with potassium cyanide (1 mM) and liver microsomes (1 mg
protein/mL) in potassium phosphate buffer for 30 min at 37 °C with constant shaking.
The samples were preincubated for 3 min, and then reactions were started by addition of
NADPH (1 mM). Control incubations without NADPH were done in parallel.
Incubations in Hepatocytes Cryopreserved hepatocytes were thawed as
recommended by the vendor and diluted to 1.4 to 3.1 million viable cells/mL in a Krebs-
Henseleit buffer (pH 7.4) containing sodium hydrogen carbonate (2.2 g/L), HEPES (22.6
9

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
14
mM), and calcium chloride (1 mM). Compound [ C]-1 (10 μM) was added to the
suspensions and incubated for 4 h at 37 °C under an atmosphere of 5% carbon dioxide.
Sample Processing Protein was precipitated by adding an equal volume of ice-
cold acetonitrile, followed by vortexing and centrifugation. Supernatants were dried to
approximately one-half to one-third of the original volume under nitrogen, and aliquots
(90 μL) were analyzed by LC-MS/MS with fraction collection and offline scintillation
counting.
3
Measurement of Covalent Binding to Protein Compound [ H]-1 (10 μM) was
incubated in triplicate in 96-well plates with pooled rat, dog, monkey or human liver
microsomes (1 mg protein/mL) in potassium phosphate buffer (0.1 M, pH 7.4).
Incubations were performed both in the presence and absence of GSH (5 mM) and were
3
supplemented with MgCl
2
(3 mM), NADPH (1 mM). Incubations of [ H]-imipramine (10
μM, Sigma-Aldrich, St. Louis, MO) were also carried out under identical conditions as a
positive control. These mixtures were incubated for 1 h at 37 °C, quenched with 4
volumes of acetone, and then filtered using a Brandel Harvester (Brandel Inc.,
Gaithersburg, MD). Proteins that precipitated on the filter paper were washed 4 times
with 80% methanol. After washing, the filters were placed into scintillation vials, treated
with SDS solution (7.5%) and incubated overnight in a water bath at 55 °C. The
concentration of protein in each vial was determined using the BCA protein assay, and
the rate of tritium decay in each vial was measured over 5 min using a liquid scintillation
counter.
Inhibition of P450 Enzymes A study was carried out to measure the time-
dependent inhibition of P450 enzymes in human liver microsomes by 1. Assay conditions
1
0

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
are summarized in Table 1. Assays were performed in 384-well microplates in a total
volume of 30 μL. Automated liquid handling equipment (Tecan Genesis 150, Tecan
Group Ltd, Hombrechtikon, Switzerland; Velocity 11 BenchCel, Velocity 11, Palo Alto,
CA; LabCyte ECHO 550, LabCyte, Sunnyvale, CA; Multidrop Combi, Thermo Fisher
Inc., San Jose, CA) was used in all steps of sample preparation. Compound 1 was tested
in two independent experiments at final concentrations of 40, 13.3, 4.4, 1.48, 0.49, 0.165,
0
.055, 0.018, 0.006, and 0.002 μM; positive controls (e.g. 60 µM clotrimazole, 40 µM
quinidine, or 10 µM fluvoxamine) were also included in each experiment. Test
compounds in DMSO (60 nL) were mixed with the probe substrate in assay buffer (15
μL; 100 mM potassium phosphate, pH 7.4, 2 mM MgCl , 1 mM NADPH), then treated
2
with human liver microsomes (15 μL) in assay buffer and incubated at 37°C. Reactions
were terminated by adding quench buffer (30 μL; 94% water, 5% acetonitrile, 1% formic
acid) containing reaction-specific internal standards. To assess time-dependent inhibition,
test articles in DMSO (60 nL) were pre-incubated with human liver microsomes (15 μL)
in assay buffer for 30 min at 37°C. Mixtures were then treated with probe substrate in
assay buffer (15 μL), and the reaction continued. At the end of the incubation, reactions
were terminated by adding quench buffer (30 μL).
The concentration of specific probe metabolite produced in each well was
determined by RapidFire™-Mass Spectrometry (solid phase extraction-MS/MS) and LC-
MS/MS (for the CYP3A4-testosterone 6β-hydroxylation reactions). Sample was loaded
on-line onto a C4 micro-scale solid phase extraction column (BIOCIUS Life Sciences,
Woburn, MA) using water containing 0.01% trifluoroacetic acid and 0.09% formic acid
(for CYP1A2, 2C8, 2C9 and 2D6 assays) or water (for CYP2B6 and 2C19 assays) and
1
1

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
eluted with acetonitrile containing 0.01% trifluoroacetic acid and 0.09% formic acid (for
CYP1A2, 2C8, 2C9 and 2D6 assays) or methanol/water (4:1 v/v) (for the CYP2B6 and
2
4
C19 assays). The eluent was directed to a RapidFire™ 200 HT System and an AB Sciex
000 QTRAP™ mass spectrometer. For the CYP3A4 assay, 6 -hydroxytestosterone was
analyzed using an Agilent Zorbax SB-C18 (150 × 2.1 mm, 5 µm) column (Santa Clara,
CA) and a mobile phase consisting of Solvent A (0.1% formic acid in water) and Solvent
B (0.1% formic acid in acetonitrile) with a linear gradient delivered at 0.5 mL/min as
follows: 20% B for 0.1 min, 20 to 51% B over 3.5 min, 51 to 90% B over 0.1 min, and
9
0% B isocratic for 0.5 min. The column was then reequilibrated to initial conditions
over 0.8 min. Selected reaction monitoring was used for analysis of metabolites and
internal standards. The MS/MS data was processed with RapidFire™ peak integration
software, and the results were exported into an Excel file for IC calculation.
5
0
For analysis of CYP3A4 activity using testosterone as the probe substrate, an
aliquot of the reaction mixture (120 μL) was transferred onto a 96-well MultiScreen
deep-well filtration plate equipped with 0.45 μm hydrophobic polytetrafluoroethylene
membrane (Millipore Co., Billerica, USA). The filtration plate was stacked atop 2 mL
9
6-well receiver plates, vortexed for 30 sec, and all mixtures were passed through the
membrane bycentrifugation for 5 min at 2000g. Filtrates were analyzed by LC-MS/MS.
The signal intensity of the MS/MS peak corresponding to each metabolite was
normalized to the signal of internal standard. Each sample signal ratio was then
normalized to signal ratios of the reactions performed in the absence of the test substance
(solvent control, 0% inhibition) and in the presence of Inhibitor Cocktail, (background,
1
00% inhibition). Results were expressed as: percent inhibition = 100(1- (S-B)/(T-B)),
1
2

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
where T = average solvent controls, B = average background, S = sample. The results
were then imported into CurveMaster curve fitting software using MathIQ (ID Business
Solutions, Ltd., UK) to determine the IC50s for each compound.
Studies Using Animals All methods were in compliance with the Guide for the
Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, 1996) as
adopted by the U.S. National Institutes of Health and approved by the Bristol-Myers
Squibb Animal Care and Use Committee. In-life portions of studies using bile-duct-
cannulated (BDC) rats and intact animals (mouse, rat, monkey) were completed at
Charles River Laboratories (Shrewsbury, MA).
14
Intact Mice Following a fast of approximately 5 to 6.5 h, [ C]-1 was
administered orally by gavage (50 mg/kg, 166.7 Ci/kg) in a vehicle of H PO buffer (0.1
3
4
M, pH 3)/PEG 400/Povidone K-30/Vitamin E TPGS (75/15/5/5) to 30 male rasH2 mice
in Groups 1 and 9 male rasH2 mice in Group 2. The rasH2 strain was selected so that
metabolism could be evaluated in the same strain that was used in the carcinogenicity
study of 1. Food was returned to the animals approximately 4 h after dosing. Terminal
blood samples were collected into tubes containing K
2
EDTA from mice in Group 1
(5 mice per time point) at 1, 4, 8, 12, 24, and 48 h post-dose. Urine and feces were
collected from mice in Group 2 (3 mice per cage) prior to dosing. After dosing, urine was
collected over the intervals 0–12, 12–24 h, and then in 24-h intervals through 168 h after
dosing. Feces was collected in 24-h intervals through 168 h.
14
Intact Rats Following a fasting period of 11.5 to 12.5 h, [ C]-1 was administered
orally by gavage (30 mg/kg, 100 Ci/kg) in a vehicle of H PO buffer (0.1 M,
pH 3)/PEG 400/Povidone K-30/Vitamin E TPGS (75/15/5/5) to 27 male SD rats in
3
4
1
3

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Group 1, and 3 male SD rats per group in Groups 2 and 3. Food was returned to the
animals approximately 4 h after dosing. Terminal blood samples for metabolite profiling
were collected into tubes containing K EDTA from animals in Group 1 (3 rats per time
2
point) prior to dosing and at 1, 2, 4, 8, 12, 24, 48, and 96 h post-dose. Blood samples for
total radioactivity analysis (TRA) were collected from each animal in Group 3 pre-dose
and at 15 time points through 168 h. Urine and feces were collected from the animals in
Group 2 pre-dose and over regular intervals through 168 h.
14
Intact Monkeys Following a fasting period of approximately 12.5 h, [ C]-1 was
administered orally (30 mg/kg, 40 Ci/kg) in a vehicle of H PO buffer (0.1 M,
3
4
pH 3)/PEG 400/Povidone K-30/Vitamin E TPGS (75/15/5/5) to 3 male cynomolgus
monkeys. Food was returned to the animals approximately 4 h after dosing. Blood
samples for TRA were collected into tubes containing K EDTA from each animal pre-
2
dose and at 15 time points through 168 h after dosing. Additional blood samples were
collected for metabolite profiling pre-dose and at 1, 4, 8, 12, 24, and 48 h. Urine and
feces were collected pre-dose and over regular intervals through 168 h.
14
BDC Rats Following a fasting period of approximately 9 h, [ C]-1 was
administered orally (30 mg/kg, 100 Ci/kg) in a vehicle of H PO buffer (0.1 M,
3
4
pH 3)/PEG 400/Povidone K-30/Vitamin E TPGS (75/15/5/5) to 3 BDC male Sprague
Dawley rats. Food was returned to the animals approximately 4 h after dosing. Blood
samples were collected into tubes containing K
, 4, 12, 24, and 48 h post-dose. Bile, urine, and feces were collected pre-dose and over
regular intervals through 48 h.
2
EDTA from each animal pre-dose and at
1
1
4

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Catheterization of Dog and Monkey Bile Ducts The procedure for
cateterization of dog and monkey bile ducts is provided as supplemental material.
14
BDC Dogs Following a fasting period of approximately 8 h, [ C]-1 was
administered orally (50 mg/kg, 20 Ci/kg) in a vehicle of H PO buffer (0.1 M,
3
4
pH 3)/PEG 400/Povidone K-30/Vitamin E TPGS (75/15/5/5) to 3 BDC male beagle dogs.
Food was returned to the animals approximately 4 h after dosing. Blood samples were
collected into tubes containing K EDTA pre-dose and at 0.5, 1.5, 4, 8, 24, 48, and 72 h
2
post-dose. Bile and urine samples were collected over the intervals of 0-8, 8-24, 24-48,
and 48-72 h post-dose. Similarly, feces samples were collected over the intervals of 0-24,
2
4-48, and 48-72 h post-dose.
Placement of Vascular Access Ports in Monkeys The procedure for the
placement of vascular access ports in monkeys is provided as supplemental material.
14
BDC Monkeys Following a fasting period of approximately 8 h, [ C]-1 was
administered orally (100 mg/kg, 40 Ci/kg) in a vehicle of H PO buffer (0.1 M,
3
4
pH 3)/PEG 400/Povidone K-30/Vitamin E TPGS (75/15/5/5) to 3 BDC male cynomolgus
monkeys. Food was returned to the animals approximately 4 h after dosing. Blood
2
samples were collected into tubes containing K EDTA pre-dose and at 0.5, 2, 4, 8, 24, 48,
and 72 h post-dose. Bile and urine samples were collected over the intervals of 0-8, 8-24,
2
4
4-48, and 48-72 h post-dose. Feces were collected over the intervals of 0-24, 24-48, and
8-72 h post-dose.
Processing of Animal Samples Upon collection, blood samples were placed on
ice and centrifuged for 15 min at 1000 g to separate plasma. Equal volumes of plasma
from each animal in a particular study were combined to create a pooled plasma sample
1
5

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
for each time point. Pooled plasma samples were treated with 2 volumes of
methanol/acetonitrile (50:50 v/v) and centrifuged at 3500 g for 30 min. Supernatants
were collected, and the pellet was extracted twice with 2 volumes of
methanol:acetonitrile:water (25:25:50 v/v/v). The supernatants from each extraction
were combined and evaporated to small volume under nitrogen. Any insoluble material
was removed by centrifugation at 19,000 g for 15 min, and the concentrated extracts (20-
9
0 μL) were analyzed by LC-MS/MS. Feces samples were homogenized (4 mL/g) with
acetonitrile/water (1/1) during studies using BDC rats and intact animals (mouse, rat,
monkey). Feces were diluted by 3 mL/g in the BDC monkey and BDC dog studies.
Pooled feces homogenates in each study were prepared by mixing a constant percentage
(2-30%) of feces by weight (intact mouse, intact rat, BDC rat, and intact monkey) or by
volume (for BDC monkey and BDC dog) from each time interval, exclusive of the pre-
dose samples. Pooled fecal homogenates were extracted with 2 volumes of
acetonitrile/methanol (1/1), vortexed (3500 rpm) for 30 min, sonicated for 5 min, and
centrifuged at 19,000 g for 15 min. After the supernatant was removed, the extraction
was repeated twice with 2 volumes of methanol/acetonitrile/water (1/1/2). The
supernatants from each extraction step were combined and evaporated to small volume
under nitrogen, and aliquots (20-90 μL) were analyzed by LC-MS/MS. In studies using
BDC animals, pooled bile samples were prepared in similar fashion to the pooled fecal
homogenates. Pooled bile samples were centrifuged at 13,000 rpm for 5 min, and aliquots
(25-90 μL) of the supernatant was analyzed by LC-MS/MS.
TRA Measurement Aliquots of the bile, plasma, or urine were mixed with liquid
scintillation cocktail (Ecolite or Permafluor E+), and the amount of radioactivity was
1
6

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
measured using either a Packard Tri-Carb 2250 liquid scintillation counter (PerkinElmer
Life and Analytical Sciences) or a Model LS 6000 liquid scintillation counter (Beckman
Instruments, Inc., Fullerton, California). Fecal homogenates from intact animals (mouse,
rat, monkey) and BDC rats were dried at ambient temperature for at least 4 h and burned
in a Model 307 sample oxidizer (PerkinElmer, Inc., Boston, Massachusetts). The
1
4
resulting CO
fluid, and analyzed by liquid scintillation counting. Portions of feces homogenates (0.05
0.15 g) from BDC monkeys and BDC dogs were mixed with Soluene 350 in a 25-mL
2
was trapped in Carbo-Sorb E (PerkinElmer, Inc.), mixed with scintillation
–
glass scintillation vial and incubated overnight at 60 °C with gentle shaking. After
cooling, 30% hydrogen peroxide (0.2 mL) and Ecolite (15 mL) were added, and the
resulting sample was analyzed by liquid scintillation counting.
LC-MS/MS and Radiochromatographic Analysis Chromatographic separations
were carried out using an Agilent or Shimadzu HPLC system (binary pump, autosampler
and a photodiode array UV detector) and a Phenomenex (Torrance, CA) Luna C18
column (4.6 × 150 mm, 5 μm) that was maintained at ambient temperature. The column
temperature was ambient. The mobile phase of 10-mM ammonium acetate (pH 5) in
water (solvent A) and acetonitrile (solvent B) was delivered at a flow rate of 1.0 mL/min
with gradient conditions as follows: 5% to 32% B over 18 min, 32% B for 27 min, 32%
to 55% B over 15 min, 55% to 80% B over 5 min, and 80% B for 3 min. The column was
then reequilibrated to initial conditions over 6 min. The HPLC eluent was directed
through a splitter so that approximately 20% of the eluate was analyzed by positive-ion
electrospray MS and MS/MS using a Thermo (San Jose, CA) LTQ or LTQ Orbitrap mass
spectrometer. Metabolites were identified based on HPLC retention time, elution order,
1
7

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
n
and LC/MS fragmentation patterns compared to those of 1 and other available reference
standards. Compound 3 was also isolated for NMR analysis. For quantification of
radioactivity in samples, the remaining 80% of the HPLC eluate was collected in regular
TM
intervals (12 s) into Deepwell LumaPlate 96-well plates (PerkinElmer Life and
Analytical Sciences, Boston, MA) using a Gilson Model FC 204 fraction collector
(Gilson Medical Electronics, Middletown, WI). Fractions were dried in an Automatic
Environmental Speed Vac (Savant, Holbrook, NY) and counted for radioactivity for 10
TM
min using a Packard TopCount NXT Scintillation and Luminescence Counter (PE Life
Sciences, Downers Grove, IL). A background of 2 to 3 counts per min was subtracted
from the value of each fraction, and metabolite profiles (radiochromatograms) were
®
drawn by plotting the resulting net counts-per-minute values vs. time using Microsoft
Excel. The relative abundance of each metabolite was calculated by summing the
radioactivity in the fractions corresponding to the metabolite and dividing by the net
(background-subtracted) radioactivity collected during the HPLC run. In certain cases
minor metabolites corresponding to small peaks in the radiochromatogram could not be
identified. In these cases, the reported sum of radioactive metabolites is less than 100% of
the total radioactivity in the sample.
Isolation of 3 from Rat Bile A composite sample of BDC rat bile (55 mL) from
two animals was extracted twice with equal volumes of ethyl acetate. The ethyl acetate
phase was collected and evaporated to dryness under nitrogen at room temperature. The
ethyl acetate extract was re-dissolved in 0.5 mL methanol, centrifuged to remove
insoluble matter, and the supernatant was concentrated under nitrogen. This methanol
treatment procedure was repeated two additional times. The resulting clear supernatant
1
8

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
100 μL concentrate) was processed by semi-preparative HPLC in multiple injections (30
(
μL aliquots) using an Agilent HP-1100 HPLC workstation and a Waters Sunfire C-18 (5
µm) 4.6 × 150 mm column. The mobile phase consisted of 10-mM ammonium acetate
(pH 5) and 5% acetonitrile (solvent A) and acetonitrile (solvent B) with a linear gradient
as follows: 15 to 85% B over 25 min, 85% B isocratic for 5 min, then a return to initial
conditions over 2 min. The flow rate was 1.2 mL/min. Eluent from the column was
routed in-line to an Agilent photodiode array detector (254 nm) and then to an Agilent
MSD operated in positive ion electrospray mode. Fractions were collected in parallel
using a Gilson 215 fraction collector (time-based protocol, left 80 format) into a 96-deep-
well Beckman polypropylene block (0.4 min/fraction for 32 min). The metabolite of
interest, 3 (m/z 755), eluted at 10.4 minutes under these chromatographic conditions.
Fractions containing 3 were pooled and evaporated overnight under nitrogen at room
temperature. The isolated metabolite was then re-dissolved in 50 L of DMSO-d and
6
transferred to a 1.7-mm capillary tube.
NMR Spectroscopy. The isolated metabolite 3 was analyzed using a Varian
Inova 600 MHz NMR spectrometer equipped with a 5-mm triple-resonance cold probe.
The temperature of the probe was maintained at 25 ºC during all experiments unless
noted otherwise. A proton spectrum was acquired, with a spectral width of 9607.3 Hz,
consisting of 23,643 data points and 64 scans. Before the Fourier transformation, the
dataset was zero-filled to 65,536 (64K) points and a line broadening of 0.5 Hz was
1
13
applied. The 2-dimensional H– C correlation [heteronuclear single quantum correlation
1
13
(
HSQC)] experiment was performed with 9,611 × 25,632 ( H × C) Hz and 2,048 × 96
points. For each of the 96 free-induction decay cycles (FIDs), 256 scans were used.
1
9

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Before Fourier transformation, squared cosine window functions for each dimension and
1
3
a linear prediction for the indirect C dimension were applied. The final spectral size was
1
1
2
048 × 2048 points. H– H correlation spectroscopy (COSY), total correlation
spectroscopy (TOCSY) and transverse rotating frame Overhauser enhancement
spectroscopy (TROESY) experiments were acquired with 9,612 Hz in both dimensions
and 2048 × 96 points. During the acquisition of these data, residual water was suppressed
with low-power radiation during the relaxation delay. For each FID in these datasets, 128
scans were used. The final sizes of these spectra were 2048 × 2048. Linear predictions
and cosine window functions were applied before the Fourier transformation. To assist in
the interpretation of NMR spectra, ACD/HNMR Predictor and ACD/CNMR Predictor
(Advanced Chemistry Development, Toronto, ON) were used.
2
0

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Results
Compound [ C]-1 was incubated with liver microsomes (dog, human, monkey,
1
4
rat), liver S9 (dog, human, monkey, mouse, Aroclor-induced rat), and hepatocytes (dog,
human, monkey, mouse, rat). It was also administered p.o. to intact animals (mice,
monkeys, rats) and BDC animals (dogs, monkeys, rats). Data that was recorded during
positive-ion electrospray LC-MS/MS analysis of 1 and its metabolites is summarized in
Table 2. Proposed biotransformation pathways of 1 are illustrated in Fig. 2.
1
4
The in vitro metabolite profiles of [ C]-1 were similar across species, with few
differences observed among the hepatic matrices that were used to generate metabolites
14
(
Table 3). Radiochromatograms that were acquired following the incubation of [ C]-1
(10 μM) with liver microsomes, liver S9, and suspended hepatocytes are available as
supplemental materials. In summary, most metabolites observed during in vitro studies
were formed via oxidative processes, and the main metabolic pathways that occurred in
hepatocytes were also observed in liver S9 and microsomes. Compounds 2, 3, and 4 were
among the most abundant metabolites and were present in all species and matrices. Most
metabolites were formed in NADPH-dependent fashion, although 4 was also detected in
incubations without NADPH at levels that were approximately 70% lower across species
(data not shown). This observation demonstrated a degree of involvement by P450 in
catalyzing the formation of 4. Whereas 2 was relatively abundant in human liver
incubations in vitro, it did not turn out to be a major metabolite in humans (Farley, 2015).
No GSH conjugates were found in any matrix, and no cyano adducts were found in
incubations with liver microsomes supplemented with potassium cyanide. Additionally,
2
1

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
prominent metabolites such as 3 were observed in the presence of trapping agents (e.g.
GSH, cyanide) in the incubations with liver microsomes.
Covalent Binding to Protein Also, a study was performed to measure the
3
3
amount of radioactivity covalently bound to protein after incubations of [ H]-1 and [ H]-
imipramine (positive control) with liver microsomes (human, monkey, rat) and NADPH.
The results of this study, summarized in Table 4, indicated that levels of irreversible
binding to protein across species were low and below those of the positive control.
Although rates were attenuated in the presence of GSH in certain species (mouse,
monkey), no GSH conjugates were detected. The absence of such conjugates is
interpreted to mean that the differences in rates, with and without GSH, were small
enough to preclude detection of a GSH conjugate that might have accounted for these
differences. Hence, while the results provide evidence that GSH conjugates might have
formed at very low levels, none were detected directly by LC-MS. Rates of binding by
3
[
H]-imipramine measured in the study were comparable to reported values (Evans et al.,
2
004).
Inhibition of P450 Enzymes Compound 1 was evaluated as a time-dependent
inhibitor of P450 enzymes in human liver microsomes. The results of these experiments
demonstrated that 1 did not inhibit CYP1A2, CYP2B6, CYP2C8, CYP2C9, CYP2C19,
or CYP2D6, either with or without a 30-min preincubation (all IC50s > 40 µM). In
contrast, 1 was found to be a weak to moderate inhibitor of CYP3A4, as evidenced by
time-dependent inhibition of midazolam 1’-hydroxylation (IC50 = 31.8 µM without
preincubation and IC50 = 13.5 µM with preincubation) and inhibition of testosterone 6 -
2
2

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
hydroxylation (IC50 = 11.0 µM without preincubation and IC50 = 8.9 µM with
preincubation).
Recovery of Radioactivity from Animals The mass-balance of radioactivity
1
4
after oral administration of [ C]-1 to intact animals (mice, monkeys, rats) and BDC
animals (dogs, monkeys, rats) is provided as supplementary material. In intact animals,
most of the administered radioactivity was recovered (90-96% of dose), with the majority
of radioactivity excreted in the feces (69-91%) and only a small portion in the urine (<
2
%) or remaining in the carcass (< 2%). The relatively low recovery of radioactivity in
feces from intact monkeys (72%) was attributed to soft feces that could not be completely
collected, since approximately 20% of the dose was recovered in cage debris and was of
probable fecal origin. Total recovery of radioactivity was lower in BDC animals (63-84%
of dose) because of the shorter collection time. Bile and feces both represented major
routes for the excretion of radioactivity in BDC animals, and together they accounted for
the majority (55-81%) of radioactivity administered.
14
Metabolites in Plasma The distribution of radioactivity among [ C]-1 and its
metabolites in plasma, following oral administration to animals, is summarized in Table
5
. Radiochromatograms are also provided as supplemental materials. Plasma metabolite
14
profiles were qualitatively similar across species. Unchanged [ C]-1 comprised the
majority (74-94%) of circulating radioactivity. Compound 3 was a major metabolite in
monkey plasma and was also detected in plasma from all other animals. All other
metabolites were minor, with each representing < 10% of circulating radioactivity.
Metabolites in Excreta Metabolite profiles of pooled feces and bile following
1
4
oral administration of [ C]-1 as a single dose to animals are summarized in Table 6, and
2
3

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
corresponding radiochromatograms are provided as supplemental materials. Metabolite
14
profiles in feces were qualitatively similar across the species. Compound [ C]-1 was the
main drug-related component recovered in feces (all species) and was also the main
compound in bile (rats, dogs). In contrast, 3 was the most abundant drug-related
compound in bile from monkeys. Compound 4, which was formed in vitro across species,
was a common component of excreta with levels ranging from 0.9 to 5.9% in feces and
0
.2 to 2.2% in bile. Compound 8, a minor metabolite in vitro, was relatively abundant in
rodent feces with levels reaching 7.3% (mouse) and 9.4% (rat) over an extended
collection period.
Elucidation of the Structures of Metabolites. Positive-ion electrospray MS/MS
product ions of 1 were generated via CID and analyzed using a Thermo LCQ Deca XP
ion trap mass spectrometer. Since 1 possesses a C axis of symmetry, most fragmentation
2
pathways could be attributed to either of two equivalent sites. The monoisotopic mass-to-
charge ratios and proposed sites of fragmentation leading to these product ions are
illustrated in Fig. 3A. Since the C symmetry of 1 was disrupted during
2
biotransformation, the MS/MS product-ion spectra of metabolites were characterized by
pairs of ions (with the intervening space equal to the overall mass shift) in instances
where an operative fragmentation plane lay between the symmetry axis and the site of
modification. These observations were used to assist in elucidating the structures of
metabolites of 1.
As examples of how tandem mass spectrometry was used to assign the structures
of metabolites, product-ion MS/MS spectra of the metabolites 2, 3, and 4 are shown in
Fig. 3B-D. The MS/MS spectrum of 2 included a pair of ions of m/z 592 (which was also
2
4

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
observed in the MS/MS spectrum of 1) and 598 (not observed in the MS/MS spectrum of
) that differed by the mass of a single oxygen atom introduced during metabolism.
1
These ions were formed by neutral losses of the two methyl carbamate valine moieties,
one of which bore the oxygen atom and the other of which did not. The pair of ions thus
served as a signal that the site of metabolism lay outward from the fragmentation plane
with respect to the C symmetry axis. Since the most likely soft spots for oxidation
2
leading to the formation of 2 were inferred to be on the valine moiety, standards of 2 and
its diastereomer 39 were synthesized chemically (synthetic procedures are provided as
supplemental materials). The MS/MS spectra of both 2 and 39 resembled the spectrum of
the metabolite detected in vitro. However, these standards were separable by HPLC, and
only the retention time of the standard of 2 matched that of the metabolite.
In the MS/MS product ion spectra of 3, the two most intense peaks of m/z 580 and
5
81 were interpreted to result from neutral losses of water (18 u) and ammonia (17 u),
respectively, from the “devalinated” product ion of m/z 598 that was formed via amide-
bond cleavage (N7–C17). These data suggested monooxidation of the pyrrolidine ring.
Although the product ion spectrum of 3 was similar to those of 6 or 7, the LC retention
time of 3 differed from those of these standards in either of their diastereomeric forms
(not shown), precluding the possibility that either C9 or C10 (or C41 or C42) were sites
of oxidation leading to the formation of 3 and instead implying the site of oxidation was
C6 or C8. The structure of 3 was thus further elucidated by NMR on a sample isolated
from rat bile (see next section).
Compound 4 formed a protonated molecule of m/z 681 during positive-ion
electrospray MS, suggesting removal of the methyl carbamate. Product ions that were
2
5

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
observed during MS/MS were also consistent with this assignment. The MS/MS
spectrum exhibited a pair of ions of m/z 524 and 582, equal to the separation between the
masses of 1 and 4 and signaling heteroatom dealkylation associated with the valine
moiety. As noted above, the formation of 4 was at least partially dependent on NADPH.
Such reactions have been reported previously as potentially mediated by P450 enzymes
(Berndt et al., 1995; Ghosal et al., 2011).
Structures of the remaining metabolites of 1 were proposed, again following the
logic described above around the spectral interpretation of metabolites of symmetrical
compounds. The data supporting these assignments are summarized in Table 2. In
addition to the interpretation of MS/MS results shown here, LC-MS/MS analysis of
synthetic standards of 2, 3, 4, 6, 7 and 8 helped to verify the identity of these metabolites
in biological samples. Structures of the remaining metabolites were proposed based on
LC-MS/MS data alone.
Characterization of 3 by NMR. Although the isolated metabolite 3 eluted as a
1
13
single component using the HPLC method described, there were two discrete H and C
resonances for C5, C8, H8, H25, and NH7. Each represented one-half of the theoretical
integration, suggesting the presence of a pair of conformers or diastereomers. Compared
1
to the H NMR spectrum of 1, the spectrum of 3 exhibited an additional pair of
resonances at 5.52 ppm and 5.48 ppm (H8) (Fig. 4). These two resonances correlated
with carbinol carbons (C8) of 76.6 ppm and 77.6 ppm, respectively, as illustrated in the
HSQC spectra in Fig. 5. A number of diagnostic through-bond and through-space
correlations are summarized in Fig. 6. Among them, the carbinol methine (H8)
1
1
demonstrated a 2-bond H- H correlation with the methylene H9 (2.1 ppm) in the COSY
2
6

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
1
1
spectrum and a 5-bond H- H correlation to the methine H6 (4.9 ppm) in the TOCSY
spectrum. H6 also correlated to an exchangeable proton H7 (8.3 ppm) in the COSY
spectrum. These observations suggested that 3 retained the intact H6/C6–C2 linkage from
1
and that H8, H9, H6 and H7 belonged to the same substructure (Fig. 6A).
Additional NMR data were collected subsequently on a synthetic standard of 3.
1
1
13
The H (Fig. 4) and H- C HSQC (Fig. 5) spectra of the standard were found to match
those of the isolated metabolite, indicating that the synthetic product was the same
compound isolated from the rat bile. The greater quantity of the synthetic material
enabled additional observations of 3 using NMR. For example, the TROESY data
showed extensive through-space correlations (NOEs) for a network of protons and
enabled the characterization of a tetrahydroimidazolo[1,2-a]pyridine partial structure
(
Fig. 6B). Finally, the gHMBC spectra revealed multiple key long-range through-bond
correlations from H5, H6, and H8 to C2 (Fig. 6C), confirming the proposed structure for
(Fig. 2).
Finally, a doubling of H and C resonances of H6/C6, H25 and NH7 was
3
1
13
apparent. In order to investigate this observation further, additional experiments were
1
performed including the acquisition of a ROESY spectrum and H spectra at various
temperatures (10, 20, 30 and 40°C). In the ROESY spectrum (not shown), no exchange
1
1
cross-peaks were observed between these H pairs. In the H spectra, no narrowing or
1
broadening of the H pairs was observed as a function of temperature.
2
7

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
Discussion
Compound 1 is a first-in-class, selective inhibitor of the NS5A replication
complex that is marketed as a treatment for hepatitis C virus infection. This compound
represents the product of a medicinal chemistry campaign described previously (Romine
et al., 2011; St. Laurent et al., 2012; Lopez et al., 2013; St. Laurent et al., 2013; Belema
et al., 2013; Belema et al., 2014a; Belema et al., 2014b). The present studies were
completed during the discovery and early development of 1 in support of its candidacy
for clinical development.
During the in vivo biotransformation experiments, doses in excess of the clinical
dose were studied. These doses were selected to mimic toxicologic doses and help
determine which metabolites might be represented during nonclinical safety studies.
14
Following oral administration of [ C]-1 as a single dose to intact animals (mouse,
monkeys, rats), the majority of radioactivity was excreted in the feces. Studies in BDC
animals (dogs, monkeys, rats) also demonstrated important contributions from biliary
excretion in the disposition of 1, whereas urinary excretion of radioactivity was minor.
Unchanged 1 was a prominent component in feces (across species) and bile (rat, dog).
Qualitatively similar metabolite profiles were observed in the feces and bile across
species. The only major metabolite that represented > 10% of total excreted radioactivity
(monkey, BDC rat) was 3. Compound 8, which represented 7.3% (mouse) and 9.4% (rat)
of the drug-related material excreted in feces, was more abundant in vivo than might have
been predicted based on the in vitro data, although it is not known whether metabolic
processes inside the gut might have contributed to its formation. In contrast, metabolites
such as 2 and 4 that were detected in vitro across species turned out to be minor products
2
8

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
14
in excreta following the administration of [ C]-1 in animals. Since 3 was also the main
metabolite of 1 in human liver microsomes, and results between in vitro and in vivo
models were generally in good agreement, the formation of 3 was also anticipated to be
the main pathway for the metabolism of 1 in humans.
Since the MS/MS spectrum of 3 (Fig. 2C) did not provide sufficient information
to allow the regiochemistry of mono-oxidation to be assigned, 3 was isolated from rat
bile for additional structural characterization using NMR. Despite the low sample
concentration and spectral complexity, the spectra acquired were interpretable and
1
provided evidence for the site of oxidation. In comparison to the NMR data of 1, the H
chemical shift (5.48 and 5.52 ppm) of the new methine proton in the isolated metabolite 3
were consistent with those of a CH group bound to both oxygen and nitrogen. Since the
mono-oxidation of methylene or aromatic sites other than C8 would not have resulted in
1
H resonances at 5.5 ppm, it was thought initially that 3 may have formed by simple
hydroxylation of C8. The two discrete resonances that were ascribed to H8/C8 were
thought to represent two diastereomers that might have been formed by rebound of
oxygen from the P450 active site onto either face of the pyrrolidine ring. The possibility
that these resonances could alternatively be attributed to tautomers, representing open and
closed forms of the pyrrolidine ring that interchanged slowly with respect to the NMR
1
1
time scale, was another option that was considered. However, the observation of a H- H
correlation between H6 (4.9 ppm) and an exchangeable proton H7 (8.3 ppm) in the
COSY spectrum (correlations illustrated in Fig. 6) of 3 provided the first evidence that
precluded the possibility of an intact pyrrolidine ring, since an exchangeable proton
would not have been assignable in this structure. Since the small amount of 3 that was
2
9

DMD Fast Forward. Published on March 30, 2016 as DOI: 10.1124/dmd.115.068866
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 68866
isolated from the rat bile sample precluded additional experiments to fully investigate the
structure of 3 and the cause of the doubling of H8/C8 resonances, preparation of a
synthetic standard of 3 was undertaken.
A synthetic pathway to 3 was devised with the intention of introducing a hydroxyl
group onto C8, in keeping with the structure that was proposed following NMR analysis
1
of the isolated metabolite. Subsequent analysis of the synthetic material gave H-NMR
spectra that were of higher quality, and LC-MS/MS data that were consistent with those
obtained previously during analysis of the isolated material, suggesting that the isolated
and synthetic compounds were the same. Additional 2D-NMR spectral data from
TROESY and gHMBC experiments enabled the continued characterization of the
tetrahydroimidazolo[1,2-a]pyridine moiety (Fig. 6) and complete structure of 3. The
formation of 3 was rationalized as a result of P450-mediated /C8-oxidation of the
pyrrolidine moiety and ring-opening to an aminoaldehyde intermediate, followed by an
intramolecular reaction between the aldehyde and a proximal imidazole nitrogen N1. The
1
3
structure of 3 also explained the C chemical shift differences between C6 and C38.
1
13
Specifically, using ACD software for the prediction of H and C chemical shifts based
13
on NMR data of similar sub-structures, the C chemical shift of C38 (54 ppm) was
shown to be consistent with its presence in a 5-membered (i.e. pyrrolidine) ring, whereas
13
the C chemical shift of C6 (44 ppm) was consistent with its inclusion in a 6-membered
13
(
piperidine) ring. Similarly, the observed C chemical shift of C8 (77 ppm) was closer to
the predicted chemical shift in the revised structure containing a piperidine ring (77 ppm)
than in the preliminary structure containing a pyrrolidine ring (80 ppm). The NMR data
of the synthetic standard of 3 also suggested the presence of two isomers of equal
3
0