7202 J. Med. Chem. 2010, 53, 7202–7218
`DOI: 10.1021/jm100863x
`
`0
`-r-fluoro-2
`Discovery of a β-D-2
`-Deoxy-2
`Treatment of Hepatitis C Virus
`
`0
`
`0
`
`-β-C-methyluridine Nucleotide Prodrug (PSI-7977) for the
`
`Michael J. Sofia,* Donghui Bao, Wonsuk Chang, Jinfa Du, Dhanapalan Nagarathnam, Suguna Rachakonda,
`P. Ganapati Reddy, Bruce S. Ross, Peiyuan Wang, Hai-Ren Zhang, Shalini Bansal, Christine Espiritu, Meg Keilman,
`Angela M. Lam, Holly M. Micolochick Steuer, Congrong Niu, Michael J. Otto, and Phillip A. Furman
`
`Pharmasset, Inc., 303A College Road East, Princeton, New Jersey 08540
`
`Received July 10, 2010
`
`Hepatitis C virus (HCV) is a global health problem requiring novel approaches for effective treatment
`of this disease. The HCV NS5B polymerase has been demonstrated to be a viable target for the
`0
`0
`0
`-R-fluoro-2
`development of HCV therapies. β-D-2
`-β-C-methyl nucleosides are selective
`-Deoxy-2
`inhibitors of the HCV NS5B polymerase and have demonstrated potent activity in the clinic.
`0
`0
`0
`0
`-R-fluoro-2
`-phosphate derivative of the β-D-2
`-β-C-
`Phosphoramidate prodrugs of the 5
`-deoxy-2
`methyluridine nucleoside were prepared and showed significant potency in the HCV subgenomic
`replicon assay (<1 μM) and produced high levels of triphosphate 6 in primary hepatocytes and in the
`livers of rats, dogs, and monkeys when administered in vivo. The single diastereomer 51 of diastereomeric
`mixture 14 was crystallized, and an X-ray structure was determined establishing the phosphoramidate
`stereochemistry as Sp, thus correlating for the first time the stereochemistry of a phosphoramidate
`prodrug with biological activity. 51 (PSI-7977) was selected as a clinical development candidate.
`
`Introduction
`The hepatitis C virus (HCVa) presents a global health
`problem with approximately 180 million individuals infected
`worldwide with 80% of those progressing to chronic HCV
`infection.1 Of those chronically infected individuals, approxi-
`mately 30% will develop liver cirrhosis and 10% will go on to
`develop hepatocellular carcinoma.2 The current standard of
`care (SOC) for HCV infected patients consists of regular
`injections of pegylated interferon (IFN) and oral ribavirin
`(RBV) administration. However, SOC has proven to be
`effective in producing a sustained virological response in only
`40-60% of patients treated, dependent on viral genotype and
`other predictors of host immune responsiveness. In addition,
`drug discontinuations may be high because of adverse side
`effects associated with the SOC treatment regimen.3,4 Conse-
`quently, the development of alternative treatment options is
`greatly needed. The search for novel therapies for the treat-
`ment of HCV infection has focused on the development of
`direct acting antiviral agents (DAAs).5,6
`
`*To whom correspondence should be addressed. Phone: (609) 613-
`4123. Fax: 609-613-4150. E-mail: michael.sofia@pharmasset.com.
`a Abbreviations: HCV, hepatitis C virus; SOC, standard of care; IFN,
`interferon; RBV, ribavirin; DAA, direct acting antiviral; RdRp, RNA
`dependent RNA polymerase; b.i.d., twice daily; q.d., once daily; YMPK,
`uridine-cytidine monophosphate kinase; NDPK, nucleoside diphos-
`phate kinase; SAR, structure-activity relationship; NMI, N-methyl-
`imidazole; DCM, dichloromethane; EC90, compound concentration that
`returns 90% of inhibition; SGF, simulated gastric fluid; SIF, simulated
`intestinal fluid; NTP, nucleoside triphosphate; PK, pharmacokinetic;
`Cmax, maximum concentration; AUC, area under the curve; tmax, time at
`maximum concentration; PAMPA, parallel artificial membrane perme-
`ability assay; IC50, compound concentration that returns 50% of
`inhibition; NOAEL, no observed adverse effect level; TFA, trifluoro-
`acetic acid, THF, tetrahydrofuran; IPE,
`isopropyl ether; DMSO,
`dimethylsulfoxide.
`
`pubs.acs.org/jmc
`
`Published on Web 09/16/2010
`
`HCV is a plus strand RNA virus of the Flaviviridea family
`with a 9.6 kb genome encoding for 10 proteins: three structural
`proteins and seven nonstructural proteins. The nonstructural
`proteins, which include the NS5B RNA dependent RNA
`polymerase (RdRp), provide several attractive targets for
`the development of anti-HCV therapy.7,8 The HCV RdRp is
`part of a membrane associated replication complex that is
`composed of other viral proteins, viral RNA, and altered
`cellular membranes.3 The NS5B polymerase is responsible for
`replicating the viral RNA genome and thus is absolutely
`required for HCV replication. As in the case of other viral
`polymerases, two approaches have been pursued to identify
`small molecule HCV NS5B polymerase inhibitors. These
`approaches include the identification of nucleoside analogues
`that function as alternative substrate inhibitors that induce a
`chain termination event and non-nucleoside inhibitors that
`bind to allosteric sites on the polymerase leading to a non-
`functional enzyme.5,6
`Several nucleoside classes have been or continue to be in
`development as inhibitors of HCV. These classes include the
`0
`0
`0
`0
`-R-F-2
`β-D-2
`-β-C-methylribose, the β-D-2
`-β-
`-deoxy-2
`0
`-azidoribose classes.9-13 These classes
`methylribose, and the 4
`are represented by the clinical candidates 1 (RG7128), 2 (NM-
`283), and 3 (R1626), respectively (Figure 1).
`0
`0
`Compound 1 is a 3
`,5
`-diisobutyrate ester prodrug of the
`cytidine nucleoside 4. In clinical studies when administered at
`1000 mg b.i.d., 1 demonstrated efficacy in genotype 1 infected
`patients (reduction of HCV RNA levels) in a 14-day mono-
`therapy study (-2.7 log10 decrease in HCV RNA) and
`produced a 88% RVR in a 4-week combination study with
`SOC-pegylated interferon plus ribavirin.14,15 In addition, 1
`was shown to be efficacious in HCV genotype 2,3 patients who
`had not responded to prior therapy, the first direct-acting
`
`r 2010 American Chemical Society
`
`Gilead 2004
`I-MAK v. Gilead
`IPR2018-00125
`
`

`

`Article
`
`Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19
`
`7203
`
`antiviral to show multiple genotype coverage in the clinic.16
`However, even with the positive clinical attributes of 1, we
`were interested in investigating second generation agents with
`improved potency, enhanced pharmacokinetic properties
`(i.e., q.d. dosing), and the potential for generating high concen-
`trations of the active triphosphate in the liver to enable low
`doses and potentially fixed-dose combinations of DAAs. To
`0
`0
`0
`-R-F-2
`achieve this objective, we focused on β-D-2
`-deoxy-2
`-
`
`Figure 1. Structures of HCV nucleoside inhibitors 1, 2, 3, and 4.
`
`β-C-methyluridine (5) (Figure 2). Earlier studies had shown
`that while 5 was inactive in the HCV replicon assay, its
`triphosphate was a potent inhibitor of HCV NS5B with a
`Ki of 0.42 μM.17-19 In addition, metabolism studies with 4
`showed that the monophosphate of 4 can be deaminated to
`the uridine monophosphate derivative and subsequently ana-
`bolized to the triphosphate 6 by uridine-cytidine mono-
`phosphate kinase (YMPK) and nucleoside diphosphate kinase
`(NDPK) (Figure 2). This uridine triphosphate was shown to
`have an intracellular half-life of 38 h.17-19 Therefore, in order
`to leverage the desired attributes of the uridine derivative, we
`needed to deliver the monophosphate of uridine nucleoside 5.
`To accomplish this, we required a monophosphate prodrug
`that would bypass the nonproductive phosphorylation step
`and that would potentially accomplish our other objective: the
`delivery of high liver concentrations of the desired triphos-
`phate 6.
`Phosphoramidate prodrug strategies had been shown
`to enhance nucleoside potency in cell culture presumably
`by increasing intracellular concentrations of the active
`nucleotide.20-22 However, at the time we began our work
`there was no example where phosphoramidate prodrug tech-
`nology was applied to the inhibition of HCV. We speculated
`that application of the phosphoramidate prodrug method
`would be an ideal approach for delivering the desired uridine
`monophosphate to hepatocytes in an in vivo setting (Figure 3).
`We hoped to take advantage of first pass metabolism where
`the liver enzymes would hydrolyze the terminal carboxylic acid
`
`Figure 2. Metabolism of 4 leads to both the active triphosphate and the inactive nucleoside 5. The monophosphate metabolite of 4 is also
`metabolized to the uridine monophosphate derivative which is then further phosphorylated to the active uridine triphosphate 6.
`
`Figure 3. First pass metabolism of the phosphoramidate prodrug derivative of the monophosphate of 5 releases the monophosphate in the
`liver at the desired site of action.
`
`

`

`7204 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19
`
`Sofia et al.
`
`Scheme 1a
`
`a (a) 70% aqueous acetic acid, 100 °C; (b) 25% methanolic ammonia, 0-15 °C.
`
`Scheme 2a
`
`a NMI, DCM, -5 to 5 °C; (b) NMI, DCM, 5-25 °C.
`
`ester of the phosphoramidate moiety triggering a cascade of
`chemical and enzymatic events that would produce the
`desired uridine monophosphate at the desired site of action,
`the liver. Subsequently, several reports demonstrated that
`phosphoramidates of several other anti-HCV nucleosides
`were able to improve potency, but these reports did not
`translate this improved in vitro potency into a clinical
`development candidate.23-25 Here we describe the discov-
`0
`0
`0
`-R-F-2
`-β-
`-deoxy-2
`ery of phosphoramidate prodrugs of 2
`0
`C-methyluridine 5
`-monophosphate and the selection of 14
`and ultimately of its single isomer 51 as clinical develop-
`ment candidates.
`
`Results and Discussion
`
`The development of phosphoramidate prodrugs of 5 began
`with the investigation of the anti-HCV SAR around the
`phosphoramidate portion of the molecule. The synthesis of
`0
`0
`0
`-R-F-2
`-β-C-methyluridine phosphoramidates
`-deoxy-2
`the 2
`began with the preparation of the uridine nucleoside 5. To
`obtain the uridine nucleoside 5, we started with the benzoyl
`0
`0
`0
`-R-F-2
`-β-C-methylcytidine (7). We
`-deoxy-2
`protected 2
`recently reported on the efficient synthesis of 7.26 From 7,
`preparation of the uridine nucleoside was efficiently accom-
`plished by a two-step process (Scheme 1). The benzoyl
`cytidine 7 was heated with 80% acetic acid overnight to afford
`the protected uridine 8, which was then treated at room
`temperature with methanolic ammonia to provide 5 in 78%
`yield.
`The phosphoramidate derivatives of the uridine nucleoside
`were prepared as shown in Scheme 2. The phosphoramidate
`
`moiety was appended by reacting 5 with a freshly prepared
`chlorophosphoramidate reagent 11 in the presence of NMI.20
`Each chlorophosphoramidate reagent was prepared by stir-
`ring an amino acid ester 9 with the appropriate phosphoro-
`dichloridate reagent 10 in the presence of an amine base (Et3N
`or NMI) in either THF or dichloromethane. The reaction
`0
`-phosphoramidates 12-49 as the major
`provided the desired 5
`0
`0
`0
`- and 3
`,5
`-phosphor-
`product with lesser amounts of the 3
`0
`-phosphoramidates were
`amidate derivatives. The desired 5
`purified by chromatography as a 1:1 mixture of diastereomers
`at the phosphorus center.
`The anti-HCV activity of these prodrugs was assessed using
`the clone A replicon and a quantitative real time PCR assay.12
`Each compound was simultaneously evaluated for cytotox-
`icity by assessing for the levels of cellular rRNA.12 The
`objective was to identify phosphoramidate prodrugs that
`exhibited submicromolar activity with the hope that this
`increased activity would translate into reduced drug load in
`the clinic relative to 1. A survey of the terminal carboxylic acid
`ester of the phosphoramidate moiety in which the amino acid
`was alanine and the phosphate ester was simply phenyl
`showed that small simple alkyl and branched alkyl groups
`provided the desired submicromolar activity; however, in the
`case of the n-butyl (15), 2-butyl (16), and n-pentyl (17) esters,
`cytotoxicity was observed (Table 1). Small cycloalkyl (18) and
`benzyl (22, 23) esters were also compatible; however, phenyl
`(21) and halogenated alkyl groups (19, 20) did not provide the
`desired potency enhancement.
`A survey of the phosphoramidate phosphate ester sub-
`stituent (Table 2) demonstrated that the 1-naphthyl ester 29
`
`

`

`Article
`
`Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19
`
`7205
`
`Table 1. HCV Replicon Activity of Phosphoramidate Prodrugs 12-23: Modification of the Phosphoramidate Ester Moiety
`
`compd
`
`R2
`
`EC90 cloneA (μM)a
`3.9
`1
`0.91
`Me
`12
`0.98
`Et
`13
`i-Pr
`0.52
`14
`n-Bu
`0.09
`15
`0.06
`2-Bu
`16
`n-Pen
`>50
`17
`c-Hex
`0.25
`18
`FCH2CH2
`1.72
`19
`F2CHCH2
`6.80
`20
`18.50
`Ph
`21
`0.13
`Bn
`22
`0.24
`4-F-Bn
`23
`a Each value is a result of n = 2 determinations. b Clone A cells.
`
`inhibition of cellular rRNA replication at 50 μM (%)b
`
`0
`0
`36.9
`25.9
`79.6
`93.8
`92
`61
`43.8
`38.3
`0
`74.3
`0
`
`Table 2. HCV Replicon Activity of Phosphoramidate Prodrugs 12 and 24-30: Modification of the Phosphoramidate Phenolic Ester Substituent
`
`compd
`
`R3
`
`EC90 cloneA (μM)a
`0.91
`Ph
`12
`0.69
`4-F-Ph
`24
`0.58
`4-Cl-Ph
`25
`2.11
`4-Br-Ph
`26
`0.45
`3,4-Cl-Ph
`27
`0.69
`2,4-Cl-Ph
`28
`0.09
`1-Napth
`29
`>50
`Et
`30
`a Each value is a result of n = 2 determinations. b Clone A cells.
`
`inhibition of cellular rRNA replication at 50 μM (%)b
`
`0.0
`16.8
`62.8
`30.8
`63.7
`10.9
`95.4
`16.8
`
`provided the greatest potency and that mono- and dihalo-
`genated phenolic esters also gave inhibitors with submicromolar
`potency. The derivative with a simple alkyl phosphate ester
`(30) was not active against HCV. Although the 1-naphthol
`ester substitution produced the most potent HCV inhibitor,
`this substitution also led to substantial cytotoxicity and was
`therefore not considered a viable substituent.
`Study of the amino acid side chain demonstrated that a
`small alkyl group (12, 32) was accommodated, but R-sub-
`stitution larger than ethyl showed substantial reduction in
`potency (Table 3). R-Disubstituted amino acids that included
`an R-cyclopropanylamino acid derivative 39 did not provide
`the target submicromolar potency (Table 4). Additionally, it
`was shown that the natural L-amino acid was required for
`activity, since the D-alanine derivative 40 was inactive
`(Table 4).
`Results of the phosphoramidate moiety single substituent
`modifications showed that L-alanine was the preferred amino
`
`acid moiety, that methyl, ethyl, isopropyl, or cyclohexyl
`carboxylate esters provided the desired potency enhancement,
`and that the phosphate ester accommodated simple phenyl or
`halogenated phenyl substituents. Subsequently, select combi-
`nations of these preferred substitutions were prepared in
`which only the phenyl or para-halogenated phenyl phosphate
`ester analogues were examined. Polyhalogenated phosphate
`esters were excluded from further evaluation in order to
`preempt any potential toxicity issues that may arise from the
`release of polyhalogenated phenols upon conversion of the
`phosphoramidate to the desired nucleoside monophosphate
`(Table 5).27-29 The most dramatic difference observed in the
`SAR for the phosphoramidate substituent combinations was
`associated with the terminal carboxylic acid ester substituent
`where the cyclohexyl ester derivatives (18, 47-49) showed as
`much as a 10-fold improvement in potency relative to their
`methyl, ethyl, or isopropyl analogues. On the basis of replicon
`potency, initial cytotoxicity profile, and structural diversity,
`
`

`

`7206 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19
`
`Sofia et al.
`
`Table 3. HCV Replicon Activity of Phosphoramidate Prodrugs 12 and 31-37: Modification of the Amino Acid Side Chain
`
`compd
`
`R1
`
`EC90 clone A (μM)a
`22.11
`H
`31
`0.91
`Me
`12
`1.61
`Et
`32
`Me2CH
`>50
`33
`Me2CHCH2
`5.4
`34
`MeSCH2CH2
`60.13
`35
`PhCH2
`57.65
`36
`15.6
`indole-3-CH2
`37
`a Each value is a result of n = 2 determinations. b Clone A cells.
`
`inhibition of cellular rRNA replication at 50 μM (%)b
`
`0.0
`0.0
`0.0
`0.0
`0.0
`24.1
`20.6
`68.4
`
`Table 4. HCV Replicon Activity of Phosphoramidate Prodrugs 24 and 38-40: R-Disubstituted Amino Acid Side Chains
`
`compd
`
`EC90 clone A (μM)a
`0.69
`24
`2.20
`38
`>50
`39
`>50
`40
`a Each value is a result of n = 2 determinations. b Clone A cells.
`
`inhibition of cellular rRNA replication at 50 μM (%)b
`
`16.8
`0.0
`0.0
`0.09
`
`a select set of seven compounds (12, 14, 18, 41, 44, 45, and 47)
`was chosen for further evaluation.
`To achieve the objective of identifying a phosphoramidate
`0
`0
`-R-F-2
`-β-C-methyluridine monophosphate
`prodrug of the 2
`suitable for clinical studies as a treatment for HCV, the
`prodrug moiety would need to survive exposure in the gastro-
`intestinal tract and preferentially release the nucleotide mono-
`phosphate in the liver. Consequently, compounds 12, 14, 18,
`41, 44, 45, and 47 were further evaluated for gastrointestinal
`stability using simulated gastric fluid (SGF) and simulated
`intestinal fluid (SIF). In addition, stability in human plasma
`and stability on exposure to human liver S9 fraction were
`also evaluated (Table 6). The human liver S9 fraction was
`chosen as a surrogate in vitro model to test for liver stability.
`The ultimate objective was to select compounds that showed
`
`improved potency relative to 4 in the HCV replicon and
`stability in SGF, SIF, and plasma but showed a short half-
`life in liver S9, which could indicate rapid release in hepato-
`cytes. Table 6 shows the in vitro stability data for the key
`compounds selected from combinations of the preferred
`phosphoramidate substituents. Each of these compounds
`exhibited prolonged stability (t1/2>15 h) in SGF, SIF, and
`human plasma but decomposed quickly to the monophos-
`phate when incubated with human liver S9 fraction.
`Having the desired target activity and stability profile,
`compounds 12, 14, 18, 41, 44, 45, and 47 were evaluated in
`vivo to determine liver levels of the active uridine triphosphate
`6 after oral administration. Since HCV replicates in liver cells,
`measurable levels of nucleoside triphosphate (NTP) should be
`a strong indication of in vivo efficacy, the assumption being
`
`

`

`Article
`
`Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19
`
`7207
`
`Table 5. HCV Replicon Activity of Phosphoramidate Prodrugs: Simultaneous Carboxylate and Phenolic Ester Modification of the
`Phosphoramidate Moiety
`
`compd
`
`R2
`
`R3
`
`Ph
`Me
`12
`4-F-Ph
`Me
`24
`4-Cl-Ph
`Me
`25
`4-Br-Ph
`Me
`26
`Ph
`Et
`13
`4-F-Ph
`Et
`41
`4-Cl-Ph
`Et
`42
`4-Br-Ph
`Et
`43
`i-Pr
`Ph
`14
`i-Pr
`4-F-Ph
`44
`i-Pr
`4-Cl-Ph
`45
`i-Pr
`4-Br-Ph
`46
`c-Hex
`Ph
`18
`c-Hex
`4-F-Ph
`47
`c-Hex
`4-Cl-Ph
`48
`c-Hex
`4-Br-Ph
`49
`a Each value is a result of n = 2 determinations. b Clone A cells.
`
`EC90 cloneA (μM)a
`1.62
`0.69
`0.58
`2.11
`0.98
`0.76
`0.39
`0.36
`0.52
`0.77
`0.42
`0.57
`0.25
`0.04
`0.054
`0.039
`
`inhibition of cellular rRNA replication at 50 μM (%)b
`
`0.0
`16.8
`62.8
`30.8
`36.9
`55.3
`0.0
`80.5
`25.9
`0.0
`0.0
`0.0
`61.1
`52.1
`66.9
`91.5
`
`Table 6. Stability Assessment in SGF, SIF, Human Plasma, and
`Human Liver S9 Fraction for Compounds 12, 14, 18, 41, 44, 45, and 47
`
`compd
`
`SGFa
`
`SIFb
`
`stability t1/2 (h)
`human plasmac
`
`human S9d
`
`0.18
`16.7
`>20
`15.5
`12
`0.57
`>24
`>24
`22
`14
`1.4
`>24
`>20
`17
`18
`0.23
`>8
`>20
`17
`41
`0.42
`>24
`>20
`>20
`44
`0.35
`>24
`>20
`>20
`45
`0.18
`>24
`>20
`20
`47
`a SGF = simulated gastric fluid, pH 1.2, 50 μg/mL concentration,
`37 °C, 20 h. b SIF = simulated intestinal fluid, pH 7.5, 50 μg/mL concen-
`tration, 37 °C, 20 h. c 100 μM, 37 °C, 24 h. d 100 μM, 37 °C, 24 h, pH 7.4.
`
`the larger the amount of NTP, the greater the potential
`efficacy. Therefore, each of the seven key compounds was
`evaluated in a screening rat PK study where each compound
`was administered as a single 50 mg/kg oral dose, livers were
`removed, and liver extracts were assayed for levels of the
`0
`0
`0
`-R-F-2
`-β-C-methyluridine triphosphate. The rela-
`-deoxy-2
`2
`tive levels of triphosphate found in the liver samples would be
`an indication of anticipated in vivo potency and would be
`used to select which of the key compounds would progress
`further. Table 7 shows the PK parameters for each of the key
`compounds. Among the seven compounds, compounds 12,
`14, and 47 produced the highest Cmax and AUC values. These
`results strongly suggest that each of these compounds was able
`to traverse the GI tract, remain intact during the absorption
`phase, and arrive intact at the target organ, ultimately result-
`ing in high drug exposure in the liver.
`Since little is known about which species is predictive
`of human exposure for nucleoside phosphoramidates, addi-
`tional in vivo PK assessment was undertaken in dog and
`cynomolgus monkeys to provide a cross-species comparison
`
`Table 7. PK Parameters of Uridine Triphosphate 6 in Rat Liver after an
`Oral Dose of 50 mg/kg Phosphoramidate Prodrugs 12, 14, 18, 41, 44, 45,
`and 47a
`
`compd
`
`Cmax
`(ng/g)
`
`tmax
`(h)
`
`AUC(0-t)
`AUC(inf)
`(ng 3 h/g)
`(ng 3 h/g)
`18968
`14206
`6.00
`1985
`12
`18080
`16796
`4.00
`1934
`14
`8831
`6487
`2.00
`557
`18
`5423
`4191
`4.00
`291
`41
`7375
`6140
`6.00
`519
`44
`8468
`5143
`1.00
`339
`45
`9888
`8937
`4.00
`716
`47
`a Livers were removed at time points 0.5, 1, 2, 4, 6, and 12 h postdose.
`
`of liver exposure to determine whether a dramatic species
`difference in PK behavior existed among compounds 12, 14,
`and 47. Comparison of the in vivo PK characteristic in dogs
`and cynomolgus monkeys was accomplished by evaluating
`both plasma and liver exposures upon oral q.d. dosing over
`4 days with a 50 mg/kg daily dose. Plasma samples were taken
`on day 3, and liver samples were taken on day 4. Liver levels
`were determined at a single time point 4 h postdose on day 4
`and therefore reflect a single concentration at a single point in
`time. Plasma and liver levels of phosphoramidate prodrug and
`liver triphosphate levels were analyzed by LC/MS/MS. In the
`dog study, compound 14 showed a 16- and 110-fold higher
`overall plasma exposure (AUC) of the parent prodrug than
`compounds 12 and 47, respectively (Table 8). In the monkey
`compound 14 also provided greater (>3-fold) plasma expo-
`sure than did compounds 12 and 47 (Table 9). Similarly, liver
`exposures in dog and monkey of the parent prodrugs were
`higher for compound 14 than for compounds 12 and 47.
`Triphosphate levels in the liver demonstrated the same rela-
`tive trends. In dog and monkey compound 14 produced
`higher triphosphate levels relative to compounds 12 and 47.
`
`

`

`7208 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19
`
`Sofia et al.
`
`Table 8. Dog Plasma and Liver PK Profile after Oral Administration of Compounds 12, 14, and 47
`plasma (prodrug)b
`
`liverc
`
`compda
`
`dose (mg/kg)
`
`Cmax (ng/mL)
`
`Tmax (h)
`
`AUC(0-t)
`AUC(inf)
`(ng 3 h/mL)
`(ng 3 h/mL)
`4960
`5.24
`418
`420
`1.00
`317
`50
`12
`10560
`612
`6894
`6903
`0.50
`6179
`50
`14
`476
`8.72
`54
`62
`0.25
`36
`50
`47
`a Dose of 50 mg/kg for 4 consecutive days. b Blood samples collected at 1, 2, 4, 6, 12, and 24 h postdose on day 3. c Livers removed at 4 h time point
`postdose on day 4.
`
`prodrug
`(ng/g liver)
`
`uridine triphosphate 6
`(ng/g liver)
`
`Table 9. Cynomolgus Monkey Plasma and Liver PK Profile after Oral Administration of Compounds 12, 14, and 47
`plasma (prodrug)b
`
`liverc
`
`compda
`
`dose (mg/kg)
`
`Cmax (ng/mL)
`
`Tmax (h)
`
`AUC(0-t)
`AUC(inf)
`(ng 3 h/mL)
`(ng 3 h/mL)
`26
`4.66
`27
`34
`0.25
`19
`50
`12
`57
`177
`86
`170
`1.00
`33
`50
`14
`NA
`13
`NA
`NA
`6.00
`1.8
`50
`47
`a Dose of 50 mg/kg for 4 consecutive days. b Blood samples collected at 1, 2, 4, 6, 12, and 24 h postdose on day 3. c Livers removed at 4 h time point
`postdose on day 4.
`
`prodrug
`(ng/g liver)
`
`uridine triphosphate 6
`(ng/g liver)
`
`Table 10. AlogP and PAMPA Profile of Compounds 12, 14, and 47
`
`compd
`
`AlogP a
`
`PAMPA permeability (nm/s)b
`
`0.19
`12
`0.92
`14
`2.26
`47
`a Reference 31. b pH 7.4, incubated for 12-24 h.
`
`0.07
`0.46
`4.88
`
`generated by cells when incubated with cytidine nucleoside 4.
`These enhanced triphosphate levels in clone A cells are
`consistent with the increased potency observed for com-
`0
`0
`-C-methyl-
`-F,2
`pounds 12, 14, and 47 when compared to 4. 2
`0
`-triphosphate levels in primary hepatocytes of
`uridine 5
`human, rat, dog, and monkey incubated with compounds 12,
`14, and 47 were also shown to be high, whereas in each case
`the triphosphate levels following incubation with 4 were
`shown to be below the limit of detection. Compound 47
`consistently showed the lowest triphosphate levels of the three
`phosphoramidate analogues. Compound 14 demonstrated 3-
`to 4-fold higher triphosphate levels in human primary hepa-
`tocytes relative to compounds 12 and 47. Compounds 12 and
`14 produced comparable amounts of triphosphate in monkey
`and rat hepatocytes, whereas in dog hepatocytes compound
`12 showed 2-fold higher levels than 14. The interspecies
`differences observed regarding conversion of phosphor-
`amidates 12, 14, and 47 to the active triphosphate 6 may be
`attributed to either cross-species differences in the level of
`enzymes needed to convert the phosphoramidates to the
`intermediate monophosphate or to differences in cell penetra-
`tion. Similarly, since each of the phosphoramidates 12, 14, and
`47 is converted to the same triphosphate metabolite, intra-
`species differences could be attributed to the selectivity of
`processing enzymes or to the ability to penetrate the species
`specific hepatocytes by each of the different analogues. To
`assess relative membrane permeability characteristics, both
`calculated AlogP values and parallel artificial membrane
`permeability (PAMPA) were evaluated. Both AlogP and
`PAMPA results indicate that the predicted order of cell
`permeability would be 47 > 14 >12, in line with lipophilic
`character (Table 10).31,32 This relative order tracks well with
`the relative replicon EC90 values; however, the order does not
`necessarily correlate to triphosphate levels observed in hepato-
`cytes. Therefore, it is possible that other factors such as
`
`Figure 4. Uridine triphosphate (6) levels in human, rat, dog, and
`monkey primary hepatocytes when treated with 4 and uridine
`phosphoramidates 12, 14, and 47.
`
`However, it should be noted that the liver triphosphate data
`were only used for an intraspecies comparison of compounds.
`One should not compare the relative dog to monkey liver
`triphosphate levels because of the liver single time point
`sampling protocol.
`Since the generation of triphosphate levels is critical to
`predicting in vivo potency of nucleos(t)ide analogues, in vitro
`triphosphate production in primary human hepatocytes was
`examined and compared to in vitro triphosphate levels deter-
`mined in primary hepatocytes from rat, dog, and monkey
`when incubated with compounds 12, 14, and 47. An analysis
`of uridine triphosphate (6) production in these primary
`hepatocyte studies would provide a comparison of the inher-
`ent capacity of each nucleotide phosphoramidate to both
`enter hepatocytes and be converted to the target triphosphate.
`In vitro analysis of triphosphate production was accom-
`plished by incubating each compound for 48 h with primary
`hepatocytes from rat, dog, monkey, and human and then
`extracting the cells and analyzing the extracts by HPLC.19,30
`The triphosphate levels generated in the primary hepatocytes
`were compared to triphosphate levels generated in clone A
`replicon cells and to the triphosphate levels generated on
`exposure to a known clinically efficacious agent, 4. Figure 4
`0
`0
`-C-methylur-
`-F,2
`shows that for compounds 12, 14, and 47 2
`0
`-triphosphate levels in clone A cells were 6- to 16-fold
`idine 5
`0
`0
`0
`-C-methylcytidine 5
`-F,2
`-triphosphate levels
`greater than 2
`
`

`

`Article
`
`Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19
`
`7209
`
`selectivity for processing enzymes or relative levels of proces-
`sing enzymes in hepatocytes from different species are con-
`tributing to the relative difference in cellular triphosphate
`levels observed.
`Nucleos(t)ide toxicity can be correlated to several cellular
`effects. Mitochondrial toxicity has been reported to be asso-
`ciated with the long-term use of certain nucleos(t)ide ana-
`logues resulting in myopathy, peripheral neuropathy, and
`pancreatitis.33,34 Hematotoxicity, which can lead to neutro-
`penia, severe anemia, and thrombocytopenia, has also been
`associated with several nucleoside analogues including the
`HCV polymerase inhibitor 3.35,36 One cause of hematotoxicity
`is bone marrow toxicity.35,36 To examine for these possible
`toxicities, compounds 12, 14, and 47 were evaluated for
`general cytotoxicity against an expanded cell panel and
`evaluated in vitro for both mitochondrial toxicity and toxicity
`to bone marrow progenitor cells. In an expanded cell panel
`that included two human hepatocyte cell lines, Huh7 and
`HepG2, a human pancreatic cell line, BxBC3, and a human T
`lymphoblast cell line, CEM, compounds 12, 14, and 47 were
`found to show no cytotoxicity up to 100 μM, the highest
`concentration tested. Compounds 12, 14, and 47 were assessed
`for mitochondrial toxicity in both CEM and HepG2 cells by
`incubating each compound for 14 days. Percentage inhibition
`of mitochondrial DNA production was determined relative to
`a no drug control. No inhibition of mitochondrial DNA
`synthesis up to the highest concentration tested (50 μM) was
`observed for the three compounds. To evaluate the potential
`for bone marrow toxicity, compounds 12, 14, and 47 were
`screened from 0.1 to 50 μM for their effect on human
`erythroid and myeloid progenitor cell colony proliferation.
`Differentiation of hematopoietic progenitors into erythroid or
`granulocyte-myeloid cell lineages over 14 days was measured.
`The results showed that compounds 12 and 14 had IC50 values
`of >50 μM for both erythroid and myeloid progenitor cells.
`Erythroid and myeloid progenitor cells were more sensitive to
`compound 47 which showed IC50 values of 37 ( 5 μM and
`30 ( 5 μM for erythroid and myeloid progenitor cells, res-
`pectively. Since each phosphoramidate derivative leads to the
`same metabolic intermediates, it is unlikely that the metabolic
`intermediates leading to the triphosphate active metabolite
`contribute to the effects on erythroid or myeloid progenitor
`cells. Therefore, since the only difference between compounds
`12, 14, and 47 resides in the nature of the carboxylic acid ester
`and the phenolic ester substituents, it would appear that the
`character of the carboxylate and/or phenolic esters contributes
`significantly to the effect on erythroid and myeloid progenitor
`cells and therefore the potential for bone marrow toxicity. The
`effect on progenitor cells observed for compound 47 could be a
`result of the phosphoramidate itself or from the released ester
`moieties after prodrug metabolism.
`A comparison of in vivo acute toxicity of compounds 12, 14,
`and 47 was undertaken by single dose oral administration in rats
`with a 14-day postdose observation period. Rats (three males
`and three females) were dosed with each compound at doses of
`50, 300, and 1800 mg/kg. Fourteen days after dose administra-
`tion, all rats were euthanized and daily clinical observations,
`body weights, macroscopic pathology including kidney and
`liver weights were assessed. For each of the compounds, no test-
`article-related mortality, clinical signs of toxicity, body weight
`changes, macroscopic pathology, or organ weight changes for
`liver and kidney were observed in any of the treatment groups.
`Consequently, for each of the compounds studied the NOAEL
`was established at >1800 mg/kg.
`
`Figure 5. X-ray crystal structure of 51 crystallized from CH2Cl2.
`
`In vitro hepatocyte triphosphate levels and in vivo PK
`profiles and effects on bone marrow progenitor cells were
`factors that differentiated compounds 12, 14, and 47. Com-
`pound 14 consistently produced high levels of triphosphate
`across all species and in particular showed the greatest triphos-
`phate levels in primary human hepatocytes by as much as
`3-fold over compounds 12 and 47. The in vivo PK assessment
`showed that compound 14 consistently demonstrated the
`highest liver triphosphate levels across species. Bone marrow
`toxicity studies showed that no difference in safety profile was
`observed between compounds 12 and 14; however, bone
`marrow progenitor cells were more sensitive to the presence
`of compound 47. On the basis of the overall profile, com-
`pound 14 (PSI-7851) was selected for further development.37
`14 is a 1:1 mixture of diastereomers at the phosphorus
`center of the phosphoramidate moiety and is a low melting
`point (mp = 66-75 °C) amorphous solid. The diastereomers
`of 14 were separated by HPLC chromatography to give the
`two pure diastereomers 50 (fast moving isomer) and 51 (slow
`moving isomer). In the clone A replicon assay, compounds 50
`and 51 produced anti-HCV activity with EC90 values of 7.5
`and 0.42 μM, respectively, thus demonstrating a >10-fold
`difference in activity between the two isomers. Diastereomer
`51 was subsequently crystallized using methylene chloride as
`the solvent, and a single crystal X-ray structure of 51 was
`obtained, definitively establishing the configuration of the
`phosphorus center as Sp and by corollary the configuration of
`50 as Rp (Figures 5 and 6). This is the first demonstrated
`crystallization and X-ray structure determination of a phos-
`phoramidate nucleotide prodrug and the first example where
`stereochemistry at phosphorus could be correlated unequiv-
`ocally to nucleotide phosphoramidate activity.
`Each diastereomer was also evaluated against repli