J. Med. Chem. 2006, 49, 7215-7226
`
`7215
`
`Application of Phosphoramidate ProTide Technology Significantly Improves Antiviral Potency
`of Carbocyclic Adenosine Derivatives
`
`Christopher McGuigan,*,† Alshaimaa Hassan-Abdallah,† Sheila Srinivasan,† Yikang Wang,† Adam Siddiqui,† Susan M. Daluge,‡
`Kristjan S. Gudmundsson,‡ Huiqiang Zhou,‡ Ed W. McLean,‡ Jennifer P. Peckham,‡ Thimysta C. Burnette,‡ Harry Marr,‡
`Richard Hazen,‡ Lynn D. Condreay,‡ Lance Johnson,‡ and Jan Balzarini§
`Welsh School of Pharmacy, Cardiff UniVersity, King Edward VII AVenue, Cardiff CF10 3XF, UK, DiVision of Chemistry MV CEDD; Drug
`Metabolism and Pharmacokinetics and Virology Departments, GlaxoSmithKline, Research Triangle Park, North Carolina 27709, and Rega
`Institute for Medical Research, Katholieke UniVersiteit LeuVen, B-3000 LeuVen, Belgium
`ReceiVed June 30, 2006
`
`We report the application of phosphoramidate pronucleotide (ProTide) technology to the antiviral agent
`carbocyclic L-d4A (L-Cd4A). The phenyl methyl alaninyl parent ProTide of L-Cd4A was prepared by
`Grignard-mediated phosphorochloridate reaction and resulted in a compound with significantly improved
`anti-HIV (2600-fold) and HBV activity. We describe modifications of the aryl, ester, and amino acid regions
`of the ProTide and how these changes affect antiviral activity and metabolic stability. Separate and distinct
`SARs were noted for HIV and HBV. Additionally, ProTides were prepared from the D-nucleoside D-Cd4A
`and the dideoxy analogues L-CddA and D-CddA. These compounds showed more modest potency
`improvements over the parent drug. In conclusion, the ProTide approach is highly successful when applied
`to L-Cd4A with potency improvements in vitro as high as 9000-fold against HIV. With a view to preclinical
`candidate selection we carried out metabolic stability studies using cynomolgus monkey liver and intestinal
`S9 fractions.
`
`Introduction
`Nucleoside analogues continue to dominate antiviral therapy
`and also make a significant contribution to the chemotherapy
`of cancer, particularly leukemia. Without exception, nucleoside
`analogues with such activity require phosphorylation in vivo
`to their active nucleotide forms. In the case of antiviral
`nucleosides this is almost always the 5¢-triphosphate. Poor
`phosphorylation can be a major cause of poor activity, with
`several examples now known where nucleoside analogues are
`inactive, despite the corresponding triphosphates being inhibitors
`at their enzyme (polymerase, reverse transcriptase) target.1,2 The
`triphosphates themselves cannot be considered to be useful drugs
`due to their inherent hydrolytic instability and poor membrane
`permeation. However, it appears that in most cases the first
`phosphorylation to the 5¢-monophosphate is the rate-limiting
`step,3 leading to the consideration of the monophosphates as
`chemotherapeutic agents. In fact, nucleoside monophosphates
`suffer from similar qualitative problems as triphosphates;
`instability (in this case to phosphatases and nucleotidases) and
`poor membrane permeation. Given these problems, and the
`perceived advantage of bypassing the nucleoside kinase depen-
`dence of nucleoside analogues, many groups have worked on
`phosphate prodrug (“ProTide”) strategies.4-6 Since 1990, we
`have developed a phosphoramidate strategy; initial work was
`on anti-retroviral AZTa derivatized with alkyl phosphates
`carrying an esterified amino acid.7 Alanine quickly emerged as
`a most effective amino acid. Subsequently, we discovered aryl
`
`* To whom correspondence should be addressed. Tel: +44 29 20874537.
`Fax: +44 29 20874537. E-mail: mcguigan@cardiff.ac.uk.
`† Cardiff University.
`‡ GlaxoSmithKline.
`§ Katholieke Universiteit Leuven.
`a Abbreviations: HIV, human immunodeficiency virus; HBV, hepatitis
`B virus; AZT, 3¢-azido-3¢-deoxythymidine; d4T/d4A, 2 ¢,3¢-dideoxy-2¢,3¢-
`didehydrothymidine/adenosine; ddU/ddA, 2¢,3¢-dideoxyuridine/adenosine;
`3TC, L-3¢ -thia-2¢ ,3¢ -dideoxycytidine; BVDU, E-5-(2-bromovinyl)-2¢ -deox-
`yuridine; L-Cd4A, (1R,cis)-4-(6-amino-9H-purin-9-yl)-2-cyclopentene-1-
`methanol.
`
`phosphate analogues as potent, nucleoside kinase-independent
`antiretrovirals.8,9 Thus, phenyl methyl alanine phosphoramidates
`have emerged as general nucleotide delivery forms, known as
`aryloxy phosphoramidate ProTides. We have applied this motif
`successfully to d4T,10 ddU,11 3TC,12 ddA,13 and d4A.14 In the
`case of d4A, a 100-4000-fold boost in vitro antiviral activity
`was noted on application of phosphoramidate ProTide technol-
`ogy. Other labs have also utilized this methodology, notably
`Franchetti and co-workers15 on isoddA and 8-azaisoddA and
`Zemlicka et al.16 on alkene and related nucleosides. Applying
`our methodology to anti-herpetic BVDU gives unusual results;
`we found a decrease in antiviral action,17 while the NewBiotics
`group reported promising anticancer action for the same
`compounds.18 We have recently reported the enhancement of
`the in vitro profile of these agents by modifications in the
`phosphoramidate structure.19
`A further issue surrounding nucleosides as drugs is the lability
`of the glycoside (base-sugar) bond toward phosphorylase-
`induced cleavage. This frequently leads to inactivation of
`nucleoside drugs. Moreover, as in the case of 5-fluorouracil and
`E-5-(2-bromovinyl)arabinofuranosyluracil, for example, coad-
`ministration can lead to serious toxic events.20 Efforts to address
`this problem have largely led to carbocyclic nucleosides. The
`first of these to enter clinical use is the carbocyclic purine
`analogue abacavir (ABC) (1, Figure 1).21-23 We have recently
`reported the application of phosphoramidate ProTide methods
`to 1 and noted a ca. 50-fold boost in anti-HIV potency and
`correlated this directly with a similar increase in the intracellular
`levels of the bioactive carbovir triphosphate.24 A 10-20-fold
`boost was also noted in antihepatitis B activity for ProTides of
`1.
`
`Given the very high ProTide potentiation noted for adenines
`such as d4A,14 we were interested to examine the effect on
`carbocyclic adenines and particularly Cd4A.
`In fact,
`the
`“natural” D-form D-Cd4A (2) is approximately 3-fold less potent
`than (1) versus HIV.25 The enantiomer, L-Cd4A (3), has modest
`activity versus HBV (ca. 1 (cid:237)M) but is poorly active versus HIV
`
`10.1021/jm060776w CCC: $33.50 © 2006 American Chemical Society
`Published on Web 11/09/2006
`
`IPR2018-00123
`
`Page 1 of 12
`
`I-MAK 1008
`
`

`

`7216 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 24
`
`McGuigan et al.
`
`the effect of ester modification on the antiviral potency of
`phosphoramidate ProTides, with a clear preference for benzyl.31-2
`Indeed, our recent work on BVDU ProTides versus cancer
`indicated a >100-fold improvement of in vitro potency on
`replacement of the methyl ester present in NewBiotics’ lead
`thymectacin18 by a benzyl ester.19 Similarly, we have reported
`extensive SAR studies on the amino acid region, including
`natural amino acid variation,33 un-natural R,R-dialkyls,34 ster-
`eochemical variation,35 amino acid extensions,36 and replace-
`ments.37 In general, alanine and the un-natural amino acid R,R-
`dimethylglycine emerged as the amino acids of choice. Indeed,
`we recently noted that dimethylglycine was a particularly
`efficacious motif with regard to anti-HBV activity when applied
`to abacavir (1).24 Thus, using similar methods (Scheme 2) we
`prepared the glycine (4g), valine (4h), leucine (4i), isoleucine
`(4j), methionine (4k), methyl aspartate (4l), phenylalanine (4m),
`proline (4n), lysine (4o), tyrosine-O-tert-butyl ether (4p), and
`dimethylglycine (4r) analogues, each as the methyl ester. The
`tyrosine compound (4q) was prepared via TFA-mediated
`hydrolysis of (4p), it being notable that the phosphoramidate
`was stable to these conditions. As noted above, we have
`previously found D-alanine to be less effective than L-alanine.35
`However, this has not been extended to other amino acids and
`not on L-nucleosides. Thus, we prepared a small panel of
`D-amino acid analogues: D-alanine (4s), D-phenylalanine (4t),
`D-leucine (4u), D-valine (4v), D-tryptophan (4w), D-methyl
`aspartate (4x), D-proline (4y), and D-methionine (4z).
`As long ago as 1992 we noted the effects on in vitro potency
`of aryl substitution in phosphoramidate ProTides.8,9 We identi-
`fied p-halogen systems as particularly effective, including the
`p-chloro.38,39 Indeed, we subsequently published a rigorous
`QSAR analysis of this effect.40 The group of Uckun have very
`actively pursued the p-bromo derivative on d4T (“stampi-
`dine”).41
`Thus, by the above methodologies, and preparing the aryloxy
`phosphorochloridate from the appropriate phenol where it was
`not commercially available, we prepared methyl alanine ana-
`logues with aryl substitution as follows: p-chloro (4aa), p-nitro
`(4ab), p-CF3 (4ac), m-CF3 (4ad), 3,4-dichloro (4ae), p-CO2Me
`(4af), m-CO2Et (4ag), and o-CO2Et (4ah).
`Finally, for purposes of comparison, the parent phenyl methyl
`alanine derivatives were prepared from enantiomeric D-Cd4A-
`(2) and the corresponding L-CddA (6) and D-CddA(7) (com-
`pounds 5, 8, and 9, respectively).
`Antiviral Activity. All of the phosphoramidates described
`above (4a-ah), 5, 8, and 9 were tested in vitro against HIV-1,
`HIV-2, and HBV, with nucleosides 2, 3, 6, and 7 as controls.
`Cytotoxicity was also evaluated in MT4 and CEM cells. All of
`the data are presented in Tables 1 and 2 (in (cid:237)M). Thus, the
`parent phenyl methylalaninyl ProTide of L-Cd4A (4a) displayed
`a ca. 2700-fold boost in anti-HIV potency, being active at 30
`nM, vs 80 (cid:237)M for the parent. The ProTide was ca. >15 times
`more cytotoxic than the parent but still displays a selectivity
`index (SI ) CC50/EC50) of >200. As expected, no significant
`differences in potency were noted for HIV-2 vs HIV-1 and for
`MT4 vs CEM cells. Versus hepatitis-B virus (HBV), where 3
`is already quite active (EC50 ca. 1 (cid:237)M), 4a is ca. 60-times more
`potent at 17 nM and shows little toxicity (CC50 1280 (cid:237)M; SI
`ca. 75 000).
`As the ester was lengthened from methyl to ethyl (4b), there
`was no significant change in antiviral potency, while the pattern
`was variable for the secondary, ispropyl ester (4c) and tertiary
`tert-butyl ester (4d). In general, the tert-butyl ester was less
`active; this correlates with our previous conclusions29 and has
`
`Figure 1. Structures of some antiviral carbocyclic nucleoside ana-
`logues.
`
`(ca. 80 (cid:237)M).26 We wondered to what extent this difference might
`reflect the relative efficiency of phosphorylation, which might
`be bypassed by ProTide methodologies. This paper describes
`our initial attempts in this regard.
`
`Results and Discussion
`Chemistry. D-Cd4A (2) and L-Cd4A (3) were prepared as
`outlined in Scheme 1 following published procedures. Briefly,
`4-amino-2-cyclopentene-1-methanol (D) was prepared from
`commercially available azabicyclo[2.2.2]hept-5-en-3-one (A) as
`peviously described in the literature.27 Condensation of this
`(D) and 5-amino-4,6-
`4-amino-2-cyclopentene-1-methanol
`dichloropyrimidine (E) in butanol at elevated temperature
`resulted in the formation of 4-[(5-amino-6-chloro-4-pyrimidi-
`nyl)amino]-2-cyclopentene-1-methanol (F).28-29
`The carbocyclic chloropurine (G) was formed by treatment
`of F with triethylorthoformate in the presence of acid. Finally,
`treatment with liquid ammonia in a Parr bomb gave the desired
`carbocyclic L-Cd4A (3). Carbobocyclic L-ddA (6) was synthe-
`sized by reducing the cyclopentene using 5% Pd/C under 40
`psi of hydrogen.
`The D-analogues 2 and 7 were prepared in a similar manner
`as described for the L-analogues.
`We followed the standard phosphorochloridate approach to
`the synthesis of ProTides that we developed in the 1990s.9 This
`involved the preparation of an aryloxy phosphorodichloridate
`by reaction of an appropriate phenol with phosphoryl chloride,
`followed by condensation with an esterfied amino acid hydro-
`chloride to give the key phosphorochloridate reagent. Reaction
`of these phosphorochloridates with nucleosides such as Cd4A
`has two challenges. The first is poor solubility, and the second
`is regiochemistry. It is important to restrict the phosphorylation
`to the 5¢-hydroxyl group and eliminate any base (amino)
`phosphorylation. This was addressed very successfully by
`Uchiyama30 using Grignard reagents of strong bases to generate
`the 5¢-alkoxide, which gives preferential reaction with electro-
`philes. We have noted the efficacy of the Uchiyama method on
`abacavir.24 Thus, we employed the same general method here
`(Scheme 2).
`In the first instance, L-Cd4A (3) was converted into its phenyl
`methylalaninyl phosphoramidate (4a) in 81% yield. As noted
`for almost every nucleoside phosphoramidate ProTide, this was
`isolated as a roughly 1:1 mixture of phosphate diastereomers,
`as evidenced by two closely spaced 31P NMR signals ((cid:228)P 3.8,
`4.1). The isomeric mixture was also evident in the 1H NMR
`(e.g., OMe (cid:228)H 3.70, 3.72) and the 13C NMR (e.g., CH3-Ala, (cid:228)C
`19.8, 20.0). Similarly prepared were the alanine analogues with
`ester modification: ethyl (4b), isopropyl (4c), tert-butyl (4d),
`tert-butyl-CH2 (4e), and benzyl (4f). We have previously noted
`
`IPR2018-00123
`
`Page 2 of 12
`
`I-MAK 1008
`
`

`

`Phosphoramidate ProTide Technology
`
`Journal of Medicinal Chemistry, 2006, Vol. 49, No. 24 7217
`
`Scheme 1. The Synthetic Route to L-Cd4A (3) and L-CddA (6)
`
`Scheme 2. The Synthetic Route to ProTides of Carbocylic
`Nucleoside Analogues 2, 3, 6, and 7a
`
`a For details of the structures, see Table 1.
`
`been ascribed to the relative stability of tertiary esters to enzyme-
`mediated hydrolysis. However, the isopropyl showed a slight
`increase in potency versus HBV and variable results versus HIV
`dependent on the cell line and assay. The extended system with
`a tBuCH2 ester (4e) showed high activity versus both HIV and
`HBV, being ca. 6-12-fold more potent
`than 4a. Finally,
`regarding esters, the benzyl analogue 4f emerged as the most
`potent ester versus HIV, being active at 9 nM and thus ca. 9000
`times more active than 3. It was also rather nontoxic and
`displayed an SI of >28 000. It was also highly active versus
`HBV, with an EC50 of ca. 7 nM, although some cytotoxicity
`was noted in this assay (at 5 (cid:237)M).
`Turning now to amino acid variations, leaving the ester as
`methyl, we have previously noted a 60-70-fold reduction in
`anti-HIV potency for d4T ProTides on alanine to glycine
`replacement29 and a 20-40-fold reduction for the corresponding
`abacavir ProTides.24 In this study with L-Cd4A we again note
`a significant drop in anti-HIV potency on this substitution (4g),
`giving, depending on the cell line, a 10-200-fold reduction.
`However, HBV activity displays a different trend, showing only
`a modest 5-fold drop in potency, thus retaining a log more
`potency than the parent 3. Similarly, the valine compound 4h
`showed a 10-20-fold reduction in anti-HIV potency as com-
`pared to 4a but was equipotent to 4a against HBV, at 20 nM.
`The data for the isoleucine analogue 4j parallel that of the valine
`compound, as might be expected from their similar structure,
`while the leucine analogue (4i), with the amino acid branch
`one bond further out from the asymmetric center,
`is ap-
`
`proximately a log more active in each assay and thus rather
`similar to alanine. The methionine (4k) and methyl aspartate
`(4l) analogues were rather similar to the leucine compound,
`while the phenylalanine analogue 4m was slightly more active,
`particularly versus HBV, where it was the most potent amino
`acid to date at 4.5 nM. The proline compound 4n was the least
`active of the amino acids to date, an observation that we have
`made previously of this rather unique amino acid.29 We report
`in this paper our first successful ProTide example with lysine
`as the amino acid. This was isolated and tested as its TFA salt
`(4o) and found to be rather poorly active; in fact, it is rather
`similar to parent 3 in several assays and 2-10-fold less active
`than the proline analogue 4n. It is interesting to compare the
`methionine (4k) and lysine (4o) cases, as they have side chains
`with similar geometries. The lysine case is ca. 50-100 fold
`less potent versus both HIV and HBV. Partly,
`this may
`correspond with the higher polarity of the lysine compound
`(particularly when protonated) and diminished membrane
`permeability. The calculated ClogP values of 4k and 4o are
`0.9 and 0.46 (Chemdraw Ultra 9.0), but these figures may not
`fully reflect the likely protonation of the lysine side chain at
`physiological pH, further diminishing its lipophilicity.
`The tyrosine analogue 4q was prepared via its protected tert-
`butyl ether (4p), so we evaluated both the free and protected
`versions in vitro. In fact, they were both rather similar in
`antiviral profile and similar to valine. The boost in anti-HBV
`activity seen for the Phe analogue (4m) was not seen for the
`Tyr compounds.
`The dimethylglycine compound (4r) was not available for
`evaluation versus the whole panel of assays, but initial data
`indicate a slight reduction in potency vs HIV and slight increase
`in potency against HBV. Thus, 4r emerged as the most potent
`member of this series against HBV with activity at 2.5 nM,
`thus being almost 400 times more potent
`than the parent
`nucleoside.
`We have previously found D-alanine to be significantly less
`effective than natural L-alanine in phosphoramidate ProTides
`of d4T.35 We recently noted the same trend for activity of
`abacavir ProTides against HIV.24 In the present case the data
`are clear and marked; the D-alanine system (4s) is ca. 100-fold
`less potent than the L-alanine parent (4a) against HIV, but D
`and L are equipotent against HBV. Indeed, comparison of the
`D-alanine (4s) and dimethylglycine (4r) systems is instructive.
`
`IPR2018-00123
`
`Page 3 of 12
`
`I-MAK 1008
`
`

`

`7218 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 24
`
`McGuigan et al.
`
`Table 1. Anti-HIV Activity and Cytotoxicity Data for Nucleoside and Nucleotide Analogues
`MT4/(cid:237)M Rega
`HIV-1
`HIV-2
`
`Ar
`
`ester
`
`AA
`
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`Ph
`p-ClPh
`p-NO2Ph
`p-CF3Ph
`m-CF3Ph
`m,p-Cl2Ph
`p-CO2MePh
`m-CO2EtPh
`o-CO2EtPh
`-
`-
`Ph
`-
`Ph
`-
`Ph
`
`Me
`Et
`iPr
`tBu
`tBuCH2
`Bn
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`Me
`-
`-
`Me
`-
`Me
`-
`Me
`
`Ala
`Ala
`Ala
`Ala
`Ala
`Ala
`Gly
`Val
`Leu
`Ile
`Met
`MeAsp
`Phe
`Pro
`Lys(TFA)
`Tyr(OtBu)
`Tyr
`Me2Gly
`D-Ala
`D-Phe
`D-Leu
`D-Val
`D-Trp
`D-Asp(OMe)
`D-Pro
`D-Met
`Ala
`Ala
`Ala
`Ala
`Ala
`Ala
`Ala
`Ala
`-
`-
`Ala
`-
`Ala
`-
`Ala
`
`0.045
`ND
`0.15
`8.19
`0.043
`ND
`4.35
`1.09
`0.19
`0.96
`0.16
`0.32
`ND
`5.28
`15.2
`0.11
`0.58
`ND
`ND
`13.3
`1.24
`6.93
`13.3
`3.9
`117
`4.07
`0.009
`0.12
`0.035
`0.044
`0.12
`0.14
`0.053
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`
`0.043
`ND
`0.22
`3.08
`0.03
`ND
`5.73
`1.05
`0.13
`1.13
`0.21
`0.26
`ND
`21
`27.2
`0.13
`0.72
`ND
`ND
`5.93
`0.93
`14.2
`21.9
`3.71
`105
`3.48
`0.018
`0.17
`0.046
`0.04
`0.076
`0.056
`0.033
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`
`CEM/(cid:237)M Rega
`HIV-1
`HIV-2
`
`0.13
`ND
`0.3
`2.93
`0.067
`ND
`13.5
`1
`0.42
`2.1
`0.53
`1.1
`ND
`3.5
`25
`3
`5.33
`ND
`ND
`9
`1.6
`7.5
`20
`5
`110
`7.67
`0.015
`0.1
`0.06
`0.15
`0.09
`0.13
`0.065
`0.053
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`
`0.09
`ND
`0.25
`3.67
`0.083
`ND
`20
`4.5
`0.56
`3.5
`1
`3.5
`ND
`7.5
`150
`5.5
`7
`ND
`ND
`8.67
`3.5
`10
`25
`15
`105
`7.87
`0.047
`0.1
`0.06
`0.1
`0.11
`0.13
`0.08
`0.057
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`
`MT4/(cid:237)M GSK
`HIV-1
`CC50
`6.4
`>25
`100
`<25
`>10
`255
`.100
`<25
`>5
`>5
`>5
`>16
`16
`.10
`.25
`<40
`>32
`20
`>100
`>40
`<16
`>16
`>40
`.10
`ND
`>25
`2
`10
`ND
`4
`>4
`>4
`-
`<4
`>100
`>500
`3.7
`>10
`>12.5
`>125
`6.5
`
`0.03
`0.017
`0.425
`<0.26
`>4
`0.009
`7.3
`0.64
`0.64
`1.3
`1.4
`>1
`0.1
`10
`25
`,1
`0.95
`0.14
`2.05
`2.8
`<1
`2.3
`.16
`2.1
`ND
`2.5
`0.013
`0.18
`ND
`0.06
`0.077
`0.12
`-
`<0.26
`80
`15
`0.3
`>10
`1.8
`50
`0.20
`
`CC50
`18.1
`ND
`84
`130
`19.5
`ND
`g250
`167
`70
`118
`85.8
`>50
`ND
`92.1
`>250
`79.7
`110
`ND
`ND
`108
`88.5
`230
`93.4
`g250
`g250
`211
`3.11
`17.4
`6.4
`16.3
`13
`12.9
`8.39
`4.17
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`
`CC50
`16.4
`ND
`91.4
`84.9
`20.2
`ND
`>250
`152
`23.2
`99.6
`34.7
`53.5
`ND
`106
`>250
`19.8
`26.9
`ND
`ND
`71
`77.1
`101
`66.9
`g250
`227
`133
`3.43
`21.6
`5.87
`22.4
`15.6
`11.5
`4.39
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`ND
`
`compd
`4a
`4b
`4c
`4d
`4e
`4f
`4g
`4h
`4i
`4j
`4k
`4l
`4m
`4n
`4o
`4p
`4q
`4r
`4s
`4t
`4u
`4v
`4w
`4x
`4y
`4z
`4aa
`4ab
`4ac
`4ad
`4ae
`4af
`4ag
`4ah
`3
`2
`5
`6
`8
`7
`9
`
`The D-alanine analogue is ca. 10-fold less potent versus both
`HIV and HBV. However, it is also less cytotoxic (>5 to>20
`fold), leaving its selectivity index similar or slightly better.
`Comparing the potency of the alanine (4a), D-alanine (4s),
`and dimethylglycine (4r) systems indicates that substitution on
`the “D-face” of the amino acid is beneficial for HBV and
`detrimental for HIV. This may reflect cell to cell differences in
`processing of the different phosphoramidates. Similar L to D
`trends were observed for D-Phe (4t), D-Leu (4u), D-Val (4v),
`D-Trp (4w), D-MeOAsp (4x), D-Pro (4y), and D-Met (4z), giving
`reductions in anti-HIV potency for D-systems of 5-50-fold and
`retention or only slight reduction for HBV. Exceptions were
`the Val (4v) and Met (4z) cases, which did show significant
`reductions in anti-HBV potency. Thus, versus HIV, in conclu-
`sion alanine remained the most effective amino acid, although
`dimethylglycine was of similar potency, as were Met and Leu
`in some assays, and Val, Ile, and Tyr were also reasonably
`effective. Glycine, proline, and lysine were poorly effective, as
`were the D-amino acids in most cases. The amino acid could
`be varied considerably with little reduction in potency against
`HBV, and several amino acids were either equipotent or more
`potent than alanine, notably dimethylglycine and D-alanine but
`also phenylalanine.
`Finally, on the SAR of the L-Cd4A ProTides, we probed
`several phenyl modifications. In 1995, we had highlighted
`p-halogen substitution as a key area where activity could be
`
`boosted in d4T phosphoramidates39 and this has later been
`picked up by the group of Uckun with their development of
`stampidine, the p-bromo species.41 Thus, in the present case
`we first prepared the p-chloro analogue, which we favor over
`the p-bromo for toxicological reasons, given the mole for mole
`release of p-halophenol on ProTide activation. Indeed, the
`p-chlorophenyl methyl alanine compound (4aa) emerged as the
`most potent methyl alanine to date with a ca. 3-10-fold potency
`improvement against HIV as compared to 4a, with some cell
`to cell variation, and a 5-fold greater potency against HBV than
`4a. We have previously noted the poor anti-HIV efficacy of
`ProTides with phenyl groups containing strongly electron
`withdrawing substitution,40 and to some extent we found the
`same to be the case here, with the p-nitro analogue (4ab) being
`10-20 fold less active than the p-chloro lead (4aa). However,
`it was not significantly less active than the unsubstituted phenyl
`parent (4a) versus HIV. Moreover, 4ab was rather potent vs
`HBV, being equipotent with 4aa and thus slightly more active
`than 4a. This is in contrast to previous experience with d4T
`ProTides and HIV.40 The CF3 group is not as electron
`withdrawing as nitro and is more lipophilic; we previously noted
`it to be more effective in the case of d4T ProTides vs HIV,40
`and we noted the same trend here. Thus, 4ac was more active
`than 4ab, and in general than 4a also, versus HIV and HBV.
`Thus, 4ac is active versus HBV below 3 nM and is the most
`active of the methyl alanine ProTides herein reported. However,
`
`IPR2018-00123
`
`Page 4 of 12
`
`I-MAK 1008
`
`

`

`Phosphoramidate ProTide Technology
`
`Journal of Medicinal Chemistry, 2006, Vol. 49, No. 24 7219
`
`Table 2. Anti-HBV Activity, Cytotoxicity, and Stability Data for
`Nucleoside and Nucleotide Analogues
`
`HepG2-2.2.15
`GSK
`
`HBV
`
`0.017
`0.02
`0.006
`0.25
`0.003
`0.0075
`0.086
`0.02
`0.01
`0.06
`0.03
`0.05
`0.0045
`0.51
`2.0
`0.026
`0.05
`0.0025
`0.0265
`0.7
`0.13
`0.6
`0.2
`0.02
`>2
`0.22
`0.003
`0.007
`<0.0032
`<2
`0.005
`0.004
`0.004
`0.004
`0.98
`52
`4
`61
`0.60
`33
`5.0
`
`CC50
`1280
`22
`>2
`170
`13
`5
`>2
`120
`20
`76
`16
`>2
`9
`>2
`.100
`>2
`>2
`10
`>200
`200
`>2
`>200
`>2
`>2
`>2
`>2
`>2
`>2
`0.72
`>2
`>2
`>2
`>2
`12
`>200
`>200
`5.4
`>200
`>2
`>200
`38
`
`compd
`4a
`4b
`4c
`4d
`4e
`4f
`4g
`4h
`4i
`4j
`4k
`4l
`4m
`4n
`4o
`4p
`4q
`4r
`4s
`4t
`4u
`4v
`4w
`4x
`4y
`4z
`4aa
`4ab
`4ac
`4ad
`4ae
`4af
`4ag
`4ah
`3
`2
`5
`6
`8
`7
`9
`
`int
`
`74
`39
`90
`96
`57
`0
`85
`91
`75
`69
`66
`8
`-
`-
`17
`-
`0
`70
`94
`-
`-
`-
`-
`-
`-
`-
`-
`-
`-
`92
`-
`48
`42
`80
`-
`-
`-
`-
`-
`-
`-
`
`% S9
`remaining
`
`liver
`
`24
`15
`8
`1
`0
`0
`46
`2
`0
`0
`0
`4
`-
`-
`95
`-
`0
`16
`41
`-
`-
`-
`-
`-
`-
`-
`
`45
`-
`7
`19
`25
`-
`-
`-
`-
`-
`-
`-
`
`it is also notably toxic in the HBV assay, being the only ProTide
`in the family that is toxic at submicromolar concentrations.
`Interestingly, the meta analogue 4ad is rather less cytotoxic but
`also apparently slightly less active. Several other aryl substituted
`compounds (4ae-4ah) are also noted in Table 2. The meta-
`substituted ester is the most active against HIV, while several
`compounds are highly active vs HBV. As noted above, the anti-
`HBV activity appears less sensitive to aryl substitution than the
`anti-HIV activity.
`The D-enantiomer of 3, D-Cd4A (2), is slightly more active
`than 3 against HIV and rather poorly active vs HBV. It was
`interesting to see whether ProTides would have a similar impact
`here and we report in Table 1 data on the parent phenyl methyl
`alanine parent (5). Thus, a 50-fold boost in anti-HIV potency
`is noted. This is much less of an improvement (ca. 2600-fold)
`than noted above for the analogous compounds in the L-series
`(3 and 4a), and thus the D-ProTide (5) is about a log less potent
`than the L-ProTide analogue (4a). Similarly, versus HBV, the
`D-ProTide (5) is only 13-fold more potent than the nucleoside,
`whereas in the L-family the boost was 60-fold. More dramatic,
`the D-ProTide (5) is cytotoxic at its effective concentration, with
`a SI barely above unity, while the L-compound (4a) has an anti-
`HBV SI of >75 000.
`Finally, we briefly pursued the application of the technology
`to the dideoxy analogues L-CddA (6) and D-CddA (7) with the
`
`preparation of the ProTides 8 and 9. Thus, unlike several other
`dideoxynucleosides, the carbocyclic analogues 6 and 7 are both
`rather poorly active. Application of ProTides did give interesting
`boosts in potency in both cases. Indeed, while the L-system (6)
`was primarily enhanced (100-fold) versus HBV, the D-com-
`pound (7) was only slightly enhanced vs HBV but significantly
`so versus HIV (250-fold).
`In conclusion, ProTide methods have been shown to signifi-
`cantly enhance the antiviral profile of a series of carbocyclic
`nucleosides, primarily L-Cd4A but also its D-analogue, and the
`dd analogues in both L- and D-series. Very significant differences
`were noted for HIV and HBV, with separate leads emerging
`for each virus, with quite separate and distinct SARs noted for
`each. Effects were noted for variations in the ester, amino acid,
`and aryl regions. In general, the HBV system was more tolerant
`of structural modifications.
`Several nanomolar compounds emerged, representing an
`almost 4-log improvement in potency of several ProTides
`versus the parent L-Cd4A nucleoside. With this background
`we were keen to seek to perform some preclinical analysis
`of the most promising compounds that would allow rational
`choices regarding further evaluation. In particular, the stability
`of the ProTides to metabolic deactivation prior to reaching their
`target site was of particular interest, given our experience with
`abacavir ProTides.24 Thus, we employed a cynomolgus monkey
`liver and intestinal S9 stability assay to gain an initial
`understanding of the stability issues in this family and to probe
`any stability-structure correlations. Data are reported for
`selected compounds in Table 2. Under the conditions of the
`assay 4a underwent some decomposition (ca. 25% over 1 h) in
`the intestinal fraction, but the majority of ProTide remained.
`The position was reversed in the liver fraction, where only 30%
`remained. The most striking of the ester variations is the benzyl
`(4f), which showed complete disappearance in both assays over
`1 h. Thus, although 4f was rather potent in the in vitro antiviral
`assays, the S9 data may be predictive of a limited in vivo
`exposure. Interestingly, the tyrosine compound (4q) revealed a
`similar instability.
`In terms of ester variation; the ethyl ester (4b) was less stable
`than the methyl parent (4a) in both media, whereas branching
`to isopropyl (4c) and tBu (4d) stabilized it in intestine but
`destabilized it in liver. This would suggest that such esters may
`be beneficial to consider for delivery to the liver, in hepatitis
`or liver cancer.
`Several amino acid substitutions for alanine gave a similar
`profile of increased or maintained intestinal stability but
`diminished liver stability, e.g., Val (4h), Leu (4i), Ile (4j), and
`dimethylglycine (4r). The only compounds with enhanced liver
`stability over the parent 4a were the glycine (4g), Lys (4o) and
`D-alanine (4s) amino acid variants and the m-CF3 aryl substitu-
`tion (4ad). Interestingly, the lysine stabilization in liver was
`uniquely accompanied by a significant labilization in intestine.
`In terms of overall maximal stability, a property which might
`be anticipated to be useful with regard to systemic drug delivery,
`the D-alanine and glycine compounds emerge as the most
`notable. Both retained good antiviral potency in the HBV assays
`but were only moderate in the HIV assay. The m-CF3 compound
`(4ad) also looked rather stable in both S9 assays and showed
`good anti-HIV potency. This indicates the potential merit of
`extensive aryl modification to tune stability and improve
`pharmacokinetic properties; however, the relevance of the S9
`stability data to the disposition of these ProTides in monkey
`and/or across species has not been demonstrated.
`
`IPR2018-00123
`
`Page 5 of 12
`
`I-MAK 1008
`
`

`

`7220 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 24
`
`McGuigan et al.
`
`Conclusion
`
`A range of phosphoramidate ProTides of L-Cd4A were
`prepared and evaluated against HIV and HBV in vitro. Boosts
`in anti-HIV potency as high as 9000-fold were noted, with ca.
`250-fold enhancements for anti-HBV activity. Differing SARs
`emerged for each virus. The parent ProTide for enantiomeric
`D-Cd4A showed only modest potency boosts (10-50 fold) with
`parallel increases in cytotoxicity. Similar conclusions were
`reached with regard to the dideoxy analogues L-CddA and
`D-CddA. In conclusion, the ProTide approach is highly suc-
`cessful when applied to L-Cd4A with potency boosts in vitro as
`high as 9000-fold vs HIV. With a view to preclinical candidate
`selection, we carried out metabolic stability studies using
`cynomolgus monkey liver and intestinal S9 fractions;
`this
`revealed interesting differences in stability between the two
`enzyme systems, suggestive of a complex metabolic SAR.
`
`Experimental Section
`
`Chemistry. General Procedures. All experiments involving
`water-sensitive compounds were conducted under scrupulously dry
`conditions. Triethylamine was dried by refluxing over calcium
`hydride. Anhydrous tetrahydrofuran and dichloromethane were
`purchased from Aldrich. Nucleosides were dried by storage at
`elevated temperature over P2O5 in vacuo. Proton, carbon, and
`phosphorus nuclear magnetic resonance (1H, 13C, 31P NMR) spectra
`were recorded on a Bruker Avance DPX spectrometer operating at
`300, 75.5, and 121.5 MHz, respectively. All 13C and 31P spectra
`were recorded proton-decoupled. All NMR spectra were recorded
`in CDCl3 at room temperature (20 ( 3 (cid:176)C). Chemical shifts for 1H
`and 13C spectra are quoted in parts per million downfield from
`tetramethylsilane. Coupling constants are referred to as J values.
`Signal splitting patterns are described as singlet (s), doublet (d),
`triplet (t), quartet (q), or multiplet (m). Chemical shifts for 31P
`spectra are quoted in parts per million relative to an external
`phosphoric acid standard. Many proton and carbon NMR signals
`were split due to the presence of (phosphate) diastereoisomers in
`the samples. The mode of ionization for mass spectroscopy was
`fast atom bombardment (FAB) using MNOBA as matrix. Column
`chromatography refers to flash column chromatography carried out
`using Merck silica gel 60 (40-60 (cid:237)M) as stationary phase. HPLC
`(Shimadzu) was conducted on an SSODS2 reverse phase column
`using a water (containing 0.1% TFA)/acetonitrile (Fisher: HPLC
`grade) eluent. Method 1: 0% CH3CN (0 min), 80% CH3CN (35
`min), 80% CH3CN (45 min), 0% CH3CN (55 min); flow rate, 1
`mL/min; UV detection at 254 nm. Method 2: 0% CH3CN (0 min),
`80% CH3CN (15 min), 80% CH3CN (25 min), 0% CH3CN (35
`min); flow rate, 1 mL/min; UV detection at 254 nm. Final products
`showed purities exceeding 99% with undetectable levels (<0.02)
`of parent nucleosides in every case. UV absorptions were deter-
`mined using a Kontron Uvikon 860 UV spectrometer.
`Standard Procedure: Preparation of Amino Acid Ester Salts.
`Thionyl chloride (2.0 mol equiv) was added dropwise to a stirred
`solution of the appropriate alcohol (15.0 mol equiv) at 0 (cid:176)C under
`nitrogen. The mixture was stirred at 0 (cid:176) C for 30 min and then slowly
`allowed to warm to room temperature. The appropriate amino acid
`(1.0 mol equiv) was added and the mixture was heated at reflux
`