Mini-Reviews in Medicinal Chemistry, 2004, 4, 371-381
`
`371
`
`Aryloxy Phosphoramidate Triesters as Pro-Tides
`
`Dominique Cahard1, Christopher McGuigan*2 and Jan Balzarini3
`
`1UMR 6014 CNRS, Université de Rouen, 76821 Mont Saint Aignan cedex France
`2Welsh School of Pharmacy, Cardiff University, King Edward VII Avenue, Cardiff CF10 3XF, UK
`3Rega Institute for Medical Research, Katholieke Universiteit Leuven, B-3000 Leuven, Belgium
`Abstract: We herein describe the development of aryloxy phosphoramidate triesters as an effective pro-tide
`motif for the intracellular delivery of charged bio-active antiviral nucleoside monophosphates. The review
`covers the discovery of such aryl phosphoramidates, their mechanism of action and structure-activity
`relationships. The application of this strategy to a range of antiviral nucleosides is highlighted.
`Keywords: Nucleotide, Pro-drug, Phosphoramidate.
`
`SCOPE OF THIS WORK
`We will describe the discovery, in vitro evaluation,
`Structure-Activity Relationships (SARs) and Mechanism of
`Action (MoA) of phosphoramidate triesters of a range of
`antiviral nucleosides. We describe these compounds as
`triesters to emphasise the fact that all of the charges on the
`phosphate nucleus are blocked and to distinguish our
`compounds from the phosphoramidate diesters described by
`Wagner and Coworkers [1]. These latter compounds have
`been well reviewed [1,2], operated by a quite separate
`mechanism, and displayed distinct SARs to those
`compounds we will describe here. Therefore we will not
`include them below. We will thus describe the development
`of fully blocked phosphoramidates, culminating in our lead
`series, the aryloxy phosphoramidates.
`
`ALKYL AND HALOALKYL PHOSPHATE
`TRIESTERS
`Early work from our laboratories indicated that simple
`alkyl triesters of the antiviral agent araA (vidarabine) and the
`anti-neoplastic agent araC (cytarabine), of general formulas 1
`and 2, (Fig. 1) respectively displayed significant biological
`activity in tissue culture [3,4]. However, analogous dialkyl
`phosphate triesters of AZT (3) were devoid of significant
`anti-HIV activity, in marked contrast to the parent
`nucleoside analogue [5]. Similarly, whilst haloalkyl
`phosphate triesters of araA and araC (4, 5) had enhanced
`biological activity [6], the corresponding AZT derivatives
`(6) and also those of 2’,3’-dideoxycytidine (ddC) (7) were in
`general poorly active [7].
`Thus, although the bis(trifluoroethyl) analogue (6) was
`active at 0.4µM, and thus >200 times more active than
`compound (3), it was still 100-fold less potent than AZT
`itself [7]. Attempts to boost the potency of these haloalkyl
`phosphate triesters by changing the degree of halogenation
`were in general not successful [8].
`
`*Address correspondence to this author at the Welsh School of Pharmacy,
`Cardiff University, King Edward VII Avenue, Cardiff CF10 3XF, UK;
`Tel./ Fax: 44 2920 874537; E-mail: mcguigan@cardiff.ac.uk
`
`ALKYLOXY PHOSPHORAMIDATES
`The original rationale for preparing phosphoramidate-
`based pro-tides was the possibility that HIV aspartate
`protease [9] might cleave a suitable oligo-peptide from the
`phosphate moiety of a blocked nucleotide phosphoramidate.
`Simple model mono-amino acyl analogues were prepared
`and evaluated in the first instance and were of sufficient
`interest to pursue in their own right. Thus, a series of simple
`alkyloxy phosphoramidates of AZT were prepared with a
`small family of methyl esterified aminoacids (8) [10]. By
`comparison to earlier dialkyl phosphates of AZT (3, 6) the
`alkyloxy phosphoramidates (8) showed significant anti-HIV
`activity. A notable dependence of the antiviral activity on
`the aminoacid side-chain began to emerge; with alanine
`being most efficacious, and with leucine and, particularly,
`isoleucine being less active [10,11]. By contrast, the alkyl
`phosphate chain could be varied from C1 to C6 with no
`significant change in activity [11].
`In a subsequent study [12], α−aminoacids were
`compared to their β,γ derivatives etc (9). Anti-HIV activity
`was maximal for the parent α system (glycine) and
`diminished with increasing alkyl spacer length, being 10-
`fold less active for n=3 as compared to n=1 [12].
`Given the earlier improvements in antiviral activity noted
`for the haloalkyl phosphate parents, we wondered whether
`haloalkyl phosphoramidates might also be more potent.
`Therefore, a small series of compounds (10) was prepared
`[13]. For each of the aminoacids glycine, alanine and valine,
`the alkyl chain either was ethyl, trifluoroethyl or
`trichloroethyl. However, by contrast to earlier observations,
`we herein noted no enhancement in antiviral potency
`compared with the haloalkyl compounds, with one striking
`exception. The trichloroethyl alanine compound (10, X=Cl,
`R=Me) was active at 0.08µM and thus 50 times more potent
`than either the ethyl or trifuoroethyl analogues. Interestingly,
`this enhancement was only seen for the alanine series, and
`not for the glycine and valine systems [13]. Thus, alanine
`emerged as a preferred aminoacid, although the mechanistic
`origins of this preference were, and still largely remain,
`unknown. Much of the preceding literature from our labs and
`others has utilized alanine as the empirical aminoacid of
`choice, although as we will note below there are other
`aminoacids, which may usefully substitute for it.
`
`1389-5575/04 $45.00+.00
`
`© 2004 Bentham Science Publishers Ltd.
`
`IPR2018-00119
`
`Page 1 of 12
`
`I-MAK 1014
`
`

`

`372 Mini-Reviews in Medicinal Chemistry, 2004, Vol. 4, No. 4
`
`PHOSPHORODIAMIDATES
`Given the promising activity of alkyl phosphoramidates,
`particularly those related to alanine, we wondered whether
`diamidates might also be efficacious. Thus, several methyl
`esterified amino acyl phosphorodiamidates (11) were
`prepared and tested [14]. Non-amino acyl phosphoro-
`diamidates derived from simple primary and secondary
`amines were also prepared. Structure-activity relationships
`were noted that indicated a strong preference for aminoacids
`such as phenylalanine [14]. Thus a different aminoacid SAR
`emerged for these diamidates as compared to the earlier alkyl
`phosphoramidates. It is intriguing to note that this
`preference for aromatic side-chains was also seen by Wagner
`for the rather un-related phosphoramidate diesters [1].
`However, in general the diamidates appeared to offer no
`biological advantage over the amidates, and the chemical
`yields of the diamidates were significantly lower; hence they
`were not further pursued.
`
`LACTYL DERIVED SYSTEMS
`In an effort to establish the importance of the bridging
`aminoacid nitrogen atom for the biological activity of the
`phosphoramidates a small family of isosteric O-linked
`analogues derived from lactic and glycolic acid was prepared
`(12) [15]. In each case, lengthening of the alkyl phosphate
`chain (R”) leads to a reduction in potency. It was also
`notable that glycolyl systems (R’=H) were more active than
`lactyl (R’=Me) by a factor of ca. 20. This is in contrast to
`the earlier work on phosphoramidates noted above where
`alanine was preferred over glycine [10,11]. A brief hydrolytic
`stability study was undertaken on compounds 12, which
`revealed liberation of polar compounds and traces of AZT in
`biological media, but not in DMSO/water. Thus, enzyme-
`mediated activation was possible. However, since the anti-
`HIV activity of even the most active compound in the series
`was significantly (>10-fold) lower than AZT itself, these
`compounds were not further pursued.
`
`DIARYL PHOSPHATES OF AZT
`One of our major breakthroughs in phosphoramidate pro-
`tide research was made in 1992, when we noted the efficacy
`of aryloxy phosphates and phosphoramidates [16]. Thus,
`diaryl phosphates (13, Fig. 2 ) were prepared from AZT
`using simple phosphorochloridate chemistry. For the first
`time, the anti-HIV activity of these phosphate derivatives of
`AZT exceeded that of the parent nucleoside in some cases.
`Thus, the bis (p-nitrophenyl) phosphate was ca. 3-fold more
`potent than AZT vs. HIV-1 in C8166 cells, with an EC50 of
`3nM [16]. Moreover, whilst AZT was almost inactive (EC50
`100µM) in the JM cell line, the substituted diaryl phosphate
`was 10-times more active (EC50 10µM). At the time, it was
`considered that JM was AZT – insensitive due to poor
`phosphorylation [17]. It later emerged that an AZT-efflux
`pump was the source of this poor AZT sensitivity [18].
`However, the conclusion remains valid that the diaryl
`phosphate was more able to retain activity in the JM cell
`line, and that this may imply a (small) degree of intracellular
`phosphate delivery. The nitro group was implicated as vital
`to this activity, as the parent diphenyl phosphate was ca.
`100-fold less active (C8166 cells). The electron-withdrawing
`power of the p-nitro groups and putative enhancements in
`
`McGuigan et al.
`
`aryl leaving group ability were suggested as the major
`driving force of this SAR [16].
`Thus, a series of analogues of (13) were prepared, with
`various alternative para substituents (CN, SMe, CF3, I,
`OMe, H) [19]. A very clear correlation emerged between
`electron-withdrawing power of the para substituent and
`antiviral potency; the nitro and cyano substituted
`compounds being the most potent, the parent phenyl
`substituted compound intermediate in activity and the
`methoxy analogue least active, being 500-fold less active
`than the nitro compound. The effect of location of the
`electron withdrawing nitro group on the aryl rings was also
`briefly pursued, with symmetrical bis-ortho nitro and bis-
`meta nitro analogues being prepared [20]. In a study of both
`HIV-1 and HIV-2 in several cell lines it was found that the
`location of the nitro group had little effect on activity.
`However, for the first time we were able to assess the
`activity of the phosphate pro-drugs in the ‘true’ kinase-
`deficient cell line CEM-TK-. This was a clear but
`disappointing result, with all of the diaryl phosphates losing
`almost all their activity, alongside AZT, in the TK- cell line.
`This most likely implied poor intracellular phosphate
`delivery and that the diaryl phosphates were acting largely, if
`not entirely, as AZT pro-drugs, not as AZTMP pro-drugs as
`intended [20]. However, the earlier work using JM cells on
`phosphoramidates [16] had indicated that aryloxy
`phosphoramides may offer a chance for true phosphate
`delivery, and this became the main focus of our work.
`
`ARYLOXY PHOSPHORAMIDATES OF AZT
`Thus, a series of aryloxy phosphoramidates of AZT was
`prepared (14) with various p-aryl substituents and several
`aminoacids [21]. Compounds were only studied in the AZT-
`resistant JM cell line to probe potential (implied) AZTMP
`release, and the alanine phosphoramidate emerged as
`strikingly effective. In HIV-1 infected JM all cultures, AZT
`was inhibitory at 100µM, whilst the phenyl methoxy
`alaninyl phosphoramidate (14, R=Me, Ar=Ph) was active at
`0.8µM. This was taken as the first evidence of a successful
`nucleotide delivery. As had been noted by us previously in
`other series there was a marked preference for alanine over
`leucine (10-fold) and glycine (>100-fold). Moreover, whilst
`electron-withdrawing aryl substitution had been noted to be
`very effective in the diaryl systems [19], it was detrimental
`here. Para fluoro substitution had a slight adventitious
`effect, but not significantly so, whilst para-nitro substitution
`led to a 100-fold loss of activity. In a subsequent study [22]
`the range of aryl substituents was extended and compounds
`studied in true TK+ and TK- cell lines. None of the
`phosphoramidates retained the high (2-4 nM) potency of
`AZT in TK competent cell lines (CEM and MT-4) against
`either HIV-1 or HIV-2 [22]. However, whilst AZT lost all of
`its activity in the TK- deficient cell line CEM/TK-, most of
`the phosphoramidates retained antiviral activity, thus being
`ca >10-35-fold more active than AZT in this assay. Again,
`alanine emerged as an important component, with the
`glycine analogue being inactive in HIV-infected CEM/TK-
`all cultures. In this assay, leucine and phenylalanine were as
`effective as alanine, although they were less so in CEM/TK+
`assays. Thus,
`the parent phenyl methoxy alanyl
`phosphoramidate emerged as an important lead compound
`[22].
`
`IPR2018-00119
`
`Page 2 of 12
`
`I-MAK 1014
`
`

`

`Aryloxy Phosphoramidate Triesters as Pro-Tides
`
`Mini-Reviews in Medicinal Chemistry, 2004, Vol. 4, No. 4 373
`
`Me
`
`O
`
`HN
`
`O
`
`N
`
`O
`
`N3
`
`O
`
`PO
`
`OR'
`
`N
`H
`
`R
`
`CH
`
`CO
`
`MeO
`
`NH2
`
`N
`
`O
`
`HN
`
`Me
`
`NH2
`
`N
`
`O
`
`N
`
`O
`
`N
`
`O
`
`N
`
`NN
`
`NH2
`
`N
`
`N
`
`O
`
`O
`
`P
`
`RO
`
`OR
`
`O
`HO
`
`O
`
`O
`
`P
`
`RO
`
`OR
`
`O
`HO
`
`O
`
`O
`
`P
`
`RO
`
`OR
`
`OH
`
`OH
`
`O
`
`O
`
`P
`
`RO
`
`OR
`
`O
`
`O
`
`N3
`
`(1) R = Me, Et, Pr
`
`(2) R = Me, Et, Pr
`
`(3) R = Me, Et, Pr
`
`(4) R = CCl3 CH2,
`CF3CH2
`
`(5) R = CCl3CH2 ,
`CF3CH2
`
`(6) R = CCl3CH2 ,
`CF3CH2
`
`(7) R = CCl3CH2,
`CF3CH2
`
`(8) R' = Me, Et, Pr, Bu, Hex
` R = H, Me, iPr, CH2iPr, CHMeEt, Bn
`
`O
`
`CH
`
`O
`
`OP
`
`AZT
`
`R'
`
`OR'
`
`CO
`
`O
`
`NH)2
`
`OP
`
`AZT
`
`RO
`
`HC
`
`R
`
`CO
`
`O
`
`OP
`
`AZT
`
`(MeO
`
`O
`
`CX3
`
`N
`H
`
`CH
`
`R
`
`CO
`
`(10) X = H, F, Cl
` R = H, Me, iPr
`
`(11) R = H, Me iPr, CH2Ph,
`CH2CO2 Me
`
`(12) R = Me, Et; R' = H, Me;
`R'' = Me, nPr, nC12H25
`
`O
`
`(CH2 )n
`
`NH
`
`OP
`
`AZT
`
`MeO
`
`CO
`
`MeO
`
`OEt
`
`(9) n = 1-6
`
`O
`
`NH
`
`OP
`
`d4T
`
`HC
`
`CO
`
`O
`
`NH
`
`OP
`
`d4T
`
`RO
`
`HC
`
`CO
`
`MeO
`
`O
`
`O
`
`ArO
`
`OP
`
`AZT
`
`MeO
`
`R
`
`OR
`
`Me
`
`OPh
`
`(15) R = nPr, CCl3CH2
`
`(16) R = Me, Bn
`
`OAZT
`
`PO
`
`OAr
`
`CHC
`
`NH
`
`R
`
`(14) R = H, Me, CH2iPr, CH2Ph
`Ar = p-X-Ph
`X = H, F, NO2, Me, Et, nPr, nPnt, MeO
`
`OAr
`
`(13) Ar = Ph, p-XPh
`X = various
`
`O
`
`O
`
`RO
`
`CHC
`
`NH
`
`OP
`
`d4A
`
`Me
`
`OPh
`
`(21) R = Me, Bn
`
`O
`
`O
`
`MeO
`
`CHC
`
`NH
`
`OP
`
`ddA
`
`Me
`
`OPh
`
`(22)
`
`O
`
`NH
`
`OP
`
`Nuc
`
`OPh
`
`HC
`
`M e
`
`CO
`
`MeO
`
`(19) Nuc = ddU
`(20) Nuc = 3TC
`
`HO
`
`O
`
`O
`
`NH2
`
`N
`
`N
`
`X
`
`N
`
`N
`
`O
`
`O
`
`OP
`
`OPh
`
`N
`H
`
`HC
`
`Me
`
`CO
`
`MeO
`
`(17) X = N
`(18) X = CH
`
`N N
`
`O
`
`(31)
`
`HO
`
`Base
`
`HO
`
`Ade
`
`HC C CH
`
`HC CH
`
`(23) Base = adenine
`(24) Base = hypoxanthine
`
`HO
`
`Ade
`
`(25)
`
`O
`
`X
`
`HN
`
`O
`
`N
`
`HO
`
`O
`
`HO
`
`(29)
`
`(30) X = F, Cl, I
`
`Fig. (1). Structures of some nucleosides and nucleotides. All nucleotides are 5'-linked.
`
`OH
`
`(32)
`
`OH
`
`(33)
`
`HO
`
`Base
`CH
`
`Bas e
`
`CH
`
`(28) Bas e = Ade, Gua
`
`O
`
`HN
`
`C C H
`
`HO
`
`O
`
`N
`
`O
`HO
`
`HC CH
`
`MeO
`
`CHC
`
`NH
`
`OP
`
`Ade
`
`Me
`
`OPh
`
`(26)
`
`(27) Base = various
`
`Br
`
`C
`H
`
`HC
`
`O
`
`HN
`
`O
`
`N
`
`O
`
`NH
`
`N
`
`NH2
`
`HO
`
`O
`
`IPR2018-00119
`
`Page 3 of 12
`
`I-MAK 1014
`
`

`

`374 Mini-Reviews in Medicinal Chemistry, 2004, Vol. 4, No. 4
`
`APPLICATION TO OTHER NUCLEOSIDES
`As with other research groups reported in this
`compilation, we had sought to find a universal phosphate
`delivery motif that could be applied to a range of
`nucleosides. Indeed as early as 1993 we suggested that the
`phenyl alanyl phosphoramidate approach might be successful
`on a range of nucleosides (ddC, d4T) and phosphonates
`(PMEA) [22]. This has subsequently been confirmed to be
`the case with extensive application of the technology by
`others and us.
`Stavudine (d4T) was an early application of ours [23].
`This was a rational choice based on the known kinetics of
`the 2n d
`phosphorylation of d4T. Thus, whilst
`phosphorylation (AZTMP to AZTDP) but not the first
`phosphorylation (AZT to AZTMP) is regarded as rate
`limiting for AZT, the first step (d4T to d4TMP) is thought
`in general to be the slow step for d4T [24]. Thus, an
`intracellular (mono) nucleotide delivery should have
`maximal impact for d4T and similar nucleosides. In the first
`instance (halo)alkyloxy phosphoramidates of d4T (15) were
`prepared [23] and found to retain activity in d4T-resistant
`JM cells. The activity was dependent on the haloalkyl
`group; the parent propyl system being poorly active.
`Subsequent studies in HIV-infected CEM/TK- cell cultures
`[25] revealed the aryloxy phosphoramidates of d4T (16) to be
`highly effective and, notably, to retain their full activity in
`CEM/TK- cells. In this study the benzyl ester emerged as
`slightly more potent than the parent methyl compound,
`being almost 10-times more active than d4T in CEM/TK+
`assays and thus ca 300-500 fold more active than d4T, in
`CEM/TK- assays. Extensive studies followed on these
`promising d4T derivatives [26,27] which we will discuss
`later.
`In 1994 Franchetti and coworkers [28] applied the aryl
`phosphoramidate technology to 8-aza-isoddA (17) and
`isoddA (18). Very significant boosts in the antiviral potency
`of the parent nucleosides were noted; >25-fold for (17) and
`350-800 fold for (18). This was important an work, which
`demonstrated the power of the aryloxy phosphoramidate
`approach to greatly improve the biological profiles of poorly
`active nucleosides. Thus, (18) was transformed from 32µM
`activity versus HIV-2,
`to 40nM activity, on
`phosphoramidate formation. Subsequent analysis of these
`compounds by the Montpellier team [29] leads to the clear
`conclusion that they function as efficient intracellular
`phosphate delivery forms.
`To a large extent this could be regarded as an example of
`what in 1990 we termed ‘kinase bypass’, wherein an
`inactive, or moderately active and poorly phosphorylated
`nucleoside could be ‘activated’ or potentiated by suitable
`pro-tide modification [5,30]. A further example of this has
`emerged in our labs on application of the technology to ddU
`[31]. Thus, whilst dideoxyuridine (ddU) is almost inactive
`in C8166 cells,
`the
`( E C 5 0 200µM) vs. HIV-1
`phosphoramidate (19) was noted to be active at low µM
`levels and to retain potency in the AZT-resistant JM cell
`line. This activity was specific
`to
`the aryloxy
`phosphoramidate both with our lab [31] and the Montpellier
`group [32] noting poor activity for the alkyloxy
`phosphoramidates.
`
`McGuigan et al.
`
`Given the success of the phosphoramidate approach by
`the Franchetti lab when applied to iso nucleosides [28], we
`were interested to pursue other sugar modifications.
`Therefore, we applied the approach to 2’,3’-dideoxy-3’-
`thiacytidine (3TC) [33]. In fact, compound 20 was found to
`be less effective than 3-TC in deoxycytidine (dCyd) kinase
`competent HIV-1 and –2 infected cell assays, but assay in
`dCK deficient cells indicated far less of an impact on
`potency for the phosphoramidate than the parent CEM cells
`3TC (ca. 20-fold vs 2000-fold). Interestingly, both
`compounds were equally effective versus hepatitis B virus in
`hepatoma G2 cells indicating efficient pro-tide activation in
`these cells but not in the CEM cells used for the HIV assay
`[33]. This was amongst the first indications that the
`(relative) efficacy of phosphoramidates might be cell-line
`dependent.
`One of the most remarkable demonstrations of the
`effectiveness of the aryloxy phosphoramidate approach came
`from our application of
`the
`technology
`to
`the
`dideoxydidehydro purine d4A [34]. Compounds of the type
`21 were found to be exquisitely potent inhibitors of HIV-1
`and 2. Both the methyl and benzyl esters displayed EC50
`values of ca. 6-18nM thus being 1000-4000 times more
`potent than the parent nucleoside analogue d4A. Although
`the phosphoramidates (21) are more cytotoxic than d4A (ca.
`30-fold), their extraordinary potency enhancements still leave
`them with enhanced selectivities (50-150 fold) [34], and they
`are taken as a good example of nucleoside‘kinase’ (in this
`case adenosine) bypass. Subsequent application of the
`technology to dideoxyadenosine ddA (22) revealed a similar
`outcome; a >100-fold potency boost, with some increase in
`cytotoxicity [35].
`The Detroit-based lab a Jiri Zemlicka has pioneered the
`synthesis of highly modified nucleosides with alkene,
`alkyne, alkene, methylenecyclopropane, methylenecyclo-
`butane and spiropentane modifications and successfully
`applied our phosphoramidate technology. Indeed, they have
`recently reviewed these efforts [36].
`Thus, phenyl methoxy alaninyl phosphoramidates of the
`anti-HIV active adenallene (23) and the inactive hypoxallene
`(24) were prepared [37].
`A 10-20 fold boost in anti-HIV potency was noted on
`phosphoramidate formation from (23). Alkenyl adenine
`nucleosides such as (25) and (26) were similarly studied
`[38,39]. In these cases, both the Z (2 5) and E (2 6)
`nucleosides were inactive, whilst the phosphoramidate of
`(25) was active in the 1-10 µM range and non-toxic; the
`isomeric phosphoramidate (26) remained inactive. The
`hypoxanthine analogue of (25) was also poorly active
`[38,39].
`A study of methylenecyclopropane nucleoside
`phosphoramidates (27) was conducted by the Zemlicka
`group [40-43]. Besides these active Z-isomers, the inactive
`E-series were also phosphorylated and compounds evaluated
`against a very wide range of viruses (HCMV, HSV-1, HSV-
`2, HHV-6, EBV, VZV, HBV, HIV-1 and HIV-2). Amongst
`the conclusions were the following:
`-
`The Z-adenine compound is a potent inhibitor of a
`variety of viruses, but is cytotoxic.
`
`IPR2018-00119
`
`Page 4 of 12
`
`I-MAK 1014
`
`

`

`Aryloxy Phosphoramidate Triesters as Pro-Tides
`
`Mini-Reviews in Medicinal Chemistry, 2004, Vol. 4, No. 4 375
`
`-
`
`-
`
`The Z-guanine analogue is active against HCMV,
`HBV, EBV and VZV and is non-cytotoxic.
`The Z-diaminopurine is highly active against HBV
`and HIV-1, with lower activity against other viruses
`and is non-cytotoxic. This compound emerged as the
`best candidate for further development. Again, whilst
`the E-isomers of the parent nucleosides were either
`inactive or very poorly active, their phosphoramidates
`emerged as potent and selective antivirals, particularly
`the adenine compound.
`the
`The Zemlicka group has also applied
`phosphoramidate technology to methylene cyclobutane (28)
`[44] and spiropentane (29) [45] nucleosides with varying
`degrees of efficacy.
`Further pursuing the kinase bypass approach we prepared
`some inactive novel d4T derivatives with 5-halo substituents
`in place of the 5-methyl group (30) and converted them to
`their phosphoramidates [46]. Whilst all compounds (30)
`were poorly active/inactive at >10-50 µM,
`the
`phosphoramidates were all active. The phenyl methoxy
`alaninyl phosphoramidate of the 5-chloro compound (30, X
`= Cl) was active at sub µM concentrations, being >100-fold
`more potent than its nucleoside parent, and was also non-
`cytotoxic [46].
`that
`Inspired by Zemlicka, who had shown
`phosphoramidate formation could broaden the spectrum of
`activity of nucleoside analogues, we sought its application
`to a variety of nucleosides with different therapeutic targets.
`Thus, the anti-herpetic agents acyclovir (31), BVDU (32)
`and netivudine (3 3) were all converted to their
`phosphoramidates [47-49].
`In general terms, the approach failed for acyclovir (31),
`where the phosphoramidate was significantly less active than
`the parent versus HSV-2, and slightly more active versus
`HCMV [47]. Similarly, with BVDU (32) we noted a
`reduction in anti-VZV activity (5-25 fold) for the
`phosphoramidates, as well and very poor activity in TK-
`assays [48]. This was taken as evidence of a low degree of
`kinase bypass resulting from inefficient pro-tide activation in
`this case or, alternatively, fact dephosphorylation of the
`
`released nucleoside monophosphate. It is therefore surprising
`to note the apparent efficacy of these phosphoramidates of
`BVDU in anti-cancer assays reported by the NewBiotics
`Group [50]. It may be that the necessary pro-tide activating
`enzymes (see below) are absent in the cell lines used in our
`antiviral assays, but present in the (tumour) cell lines used
`by NewBiotics, or that the apparent anti-cancer activity of
`BVDU phosphoramidates does not arise via monophosphate
`release. However, the apparent clinical progression of these
`agents [51] would suggest that the anti-cancer arena may
`well be a fruitful area for future phosphoramidate studies.
`Finally, our application of the technology to the potent
`anti-VZV agent netivudine (3 3) [49] was again
`disappointing, with little activity being noted.
`Thus, in conclusion others and we have demonstrated
`that the phenol methoxyalaninyl group may significantly
`enhance the potency, selectivity and activity of spectrum of a
`range of nucleosides and by-pass their dependence on
`nucleoside-kinase mediated activation. The approach is very
`successful for the Zemlicka agents with highly modified
`sugar regions, and for ddA and d4A. It is also effective for
`d4T and AZT in nucleoside kinase deficient cells. However,
`it is clearly dependent on the nature of the parent nucleoside
`and the cell line/target studied. The example of BVDU
`highlights this final point [48,50].
`
`APPLICATION TO ACYCLIC NUCLEOSIDE
`PHOSPHONATES
`the published work on
`Until recently all of
`phosphoramidate pro-tides was on nucleoside analogues, as
`noted above. However, recently there have been a few reports
`of application of the technology to phosphonates, and in
`particular acyclic nucleoside phosphonates (ANPs). The
`Gilead group who has been active in the commercialization
`of ANPs have reported [52] that aryloxy phosphonamidates
`(34, Fig. 2 ) of PMPA (tenofovir) are highly active anti-
`retrovirals. We reached the same conclusion for PMPA and
`the closely related PMEA [53]. Boosts in antiviral potency
`of 30-100 fold were noted for PMPA and PMEA, with the
`usual preference for alanine as the aminoacid component.
`
`Ade
`
`O
`
`NH
`
`OP
`
`d4T
`
`OPh
`
`HC
`
`R
`
`CO
`
`O
`
`Me
`
`O
`
`MeO
`
`CC
`
`NH
`
`OP
`
`d4T
`
`MeO
`
`Me
`
`OPh
`
`O
`
`PO
`
`OPh
`
`CH
`
`NH
`
`R'
`
`CO
`
`RO
`
`(34) R = Me, iPr; R' = H, Me, Bn
`
`(35)
`
`(36) R = Et, nPr, nBu, Ph
`
`NN
`
`NH2
`
`N
`
`N
`
`HO
`
`(40)
`
`O
`
`O
`
`O
`
`O
`
`O
`
`(CH2)n
`
`NH
`
`OP
`
`d4T
`
`RO
`
`CHC
`
`O
`
`OP
`
`d4T
`
`MeO
`
`CHC
`
`OPh
`
`R'
`
`OPh
`
`Me
`
`N
`H
`
`OP
`
`d4T
`
`O
`
`X
`
`CO
`
`MeO
`
`(37) n = 1-6
`
`(38) R = Me, Et; R' = H, Me
`
`(39) X = various
`
`Fig. (2). Structures of some nucleotides.
`
`IPR2018-00119
`
`Page 5 of 12
`
`I-MAK 1014
`
`

`

`376 Mini-Reviews in Medicinal Chemistry, 2004, Vol. 4, No. 4
`
`Interestingly, we noted a significant preference for L-alanine
`over D-alanine (5-60 fold), whereas Gilead observed a
`preference for one phosphonate diastereoisomer over the
`other, observations we will further discuss later under
`‘stereochemistry’ matters. Gilead commenced clinical trials
`on their amidate GS7340 in 2002, most notably progressing
`with one phosphate diastereoisomer, afforded by an efficient
`large-scale synthesis and isomer separation [54].
`
`D4T Aryloxy Phosphoramidate SARs
`We have conducted fairly extensive structure-activity
`relationship studies of various
`regions of
`the
`phosphoramidate unit when attached to d4T, with over 250
`analogues prepared to date [26,27]. These will be discussed
`by the respective region of the phosphoramidate:
`Ester Region
`Some early work on AZT alkyloxy phosphoramidates
`revealed the importance of the carboxyl ester region for anti-
`HIV activity [55], which was subsequently confirmed for
`d4T aryloxy analogues [56]. Thus, a range of primary,
`secondary, tertiary, alkyl, benzyl and linear and branched
`esters related to (16) were prepared and evaluated against
`HIV-1 and –2 in thymidine kinase – competent and
`–deficient cell lines. Data are presented in Graph 1 as plot of
`potency (1/EC50) for a range of esters, versus HIV-1 in CEM
`TK+ cells, with d4T as control. A number of esters lead to
`potent activity, comparable with, or slightly more potent
`than, the methyl parent with compound. The benzyl, and
`
`McGuigan et al.
`
`naphthyl esters in particular were noted to be highly potent.
`The t-butyl ester analogue on the other hand was >10-times
`less potent than the methyl ester parent compound. As we
`will note below under Mechanism of Action studies, this
`correlates well with the poor esterase susceptibility of this
`particular ester. We later conducted a QSAR analysis using
`calculated physical properties (TSARTM) for 15 esters related
`to (16), which showed a good degree of correlation between
`predicted and measured activity, with a clear dependence on
`the shape and electronic distribution of the ester [57]. In
`particular, there emerged a strong dependence on the
`directional component of the ester group lipophilicity (the
`‘lipole moment’), indicating that electron withdrawing
`groups in the ester, but removed from the ester bond, should
`boost potency.
`Amino Acid
`As noted above, alanine had arisen as the aminoacid of
`choice based on
`limited studies with alkyloxy
`phosphoramidates of AZT. We now compared 13
`aminoacids related to (16, R=Me) [58]. Very clear SARs
`emerged from this study as presented in Graph 2, along with
`other highly amino-modified systems will be described later.
`Thus, versus HIV-1 in CEM TK+ cells, alanine remained the
`most effective aminoacid. However, several other aminoacids
`led to potencies, which were not significantly reduced,
`notably the un-natural, achiral α,α-dimethylglycine
`compound (35), which was slightly more potent than the
`alanine compound in CEM / TK- cells. In fact, all of the
`phosphoramidates retained full potency in the nucleoside
`
`Graph 1. The effect of ester changes on antiviral potency.
`
`IPR2018-00119
`
`Page 6 of 12
`
`I-MAK 1014
`
`

`

`Aryloxy Phosphoramidate Triesters as Pro-Tides
`
`Mini-Reviews in Medicinal Chemistry, 2004, Vol. 4, No. 4 377
`
`Graph 2. The effect of amino acid changes on antiviral potency.
`
`kinase deficient CEM cell assay, indicating action via
`intracellular d4TMP release. However, some aminoacids
`were less effective; proline in particular, leads to a
`compound 20-100 times less potent than (16); whether this
`relates to the importance of a free aminoacid NH, or steric or
`conformational issues relating particularly to proline is not
`entirely clear. However, valine and isoleucine were also
`poorly effective, indicating some steric restriction issues
`related to the side-chain. On the contrary, however, glycine
`is a striking example; the simple loss of the alanine methyl
`group resulting in a ca 60-70 fold reduction in potency.
`The potency of (35) was an important discovery, since
`for the first time it indicated that ‘un-natural’ (or less
`common) aminoacids could be utilized
`in
`the
`phosphoramidate approach. Indeed, we subsequently studied
`the α,α-diethyl- and -dipropyl analogues of (35), but found
`that these were poorly active (≥100-fold less potent than the
`dimethyl parent) [59]. On the other hand, considerable
`tolerance was allowed for un-natural, non-alkyl glycines, of
`the type (36). Thus, whilst α-ethylglycine was 10-fold less
`effective than alanine, the n-propyl and n-butyl analogues
`showed no subsequent losses in potency [59].
`These compounds are therefore all substantially more
`potent than the ‘natural’ glycine system. Replacement of the
`side-chain in (36) by a phenyl did however lead to a further
`
`loss of activity, to yield a material similar in potency to the
`glycine compound [58,59].
`Amino Acid Stereochemistry
`Given the importance of the aminoacid side-chain it
`became interesting to probe its stereochemical requirements.
`Thus, we prepared the isomeric, D- compound related to (16,
`R=Me) and found it to be 20-30 fold less effective than the
`L-parent [60]. Despite this reduction in potency, the D-
`compound did retain full potency in TK-deficient CEM
`assays, indicating its functioning entirely as a d4TMP
`delivery form, with little or no free d4T release. It is further
`interesting to note that the D-alanine compound is of similar
`(slightly higher) potency as the glycine analogue [58,60].
`This implies that a methyl group on the L-face of the
`aminoacid (as in L-alanine and α,α-dimethylglycine)
`contributes about a log in potency to the baseline of glycine,
`that the D-face-methyl group of D-alanine cannot substitute,
`and that the Pro-D-methyl group of α,α-dimethylglycine is
`neither advantageous nor detrimental to potency.
`Amino Acid Replacement
`In this section we briefly describe some large-scale
`aminoacid changes, which have a significant (largely
`negative) impact on potency. Thus, replacement of the
`aminoacid in (16) by a family of non-aminoacyl simple n-
`
`IPR2018-00119
`
`Page 7 of 12
`
`I-MAK 1014
`
`

`

`378 Mini-Reviews in Medicinal Chemistry, 2004, Vol. 4, No. 4
`
`alkylamines (C3 to C12) leads to a complete removal of
`activity [61]. The same result was observed for simple
`alkylamine analogues of the AZT phosphoramidates (14)
`[61]. Thus, an aminoacid appears a pre-requisite for a
`successful aryloxy phosphoramidate approach. This is in
`marked contrast to recent data from the Montpellier group
`who has prepared phosphoramidate – SATE hybrids of AZT,
`and found them to be highly potent and not to depend on an
`aminoacid type structure for potency [62]. As noted by
`Peyrottes et al. in this Compilation, this indicates a
`different mechanism of action for SATE – phosphoramidates
`as compared to aryloxy phosphoramidates (see below).
`We had previously noted that chain extended aminoacid-
`related alkyloxy phosphoramidates of type (9) were poorly
`effective [12]. It was of interest to extend this study to
`analogous beta- and other aminoacids, when applied to
`aryloxy derivatives of d4T. Thus, compounds (37) were
`prepared and found to be very poorly active; all extended
`compounds were ca 40-times less potent than the glycine
`parent compound [63]. Interestingly the β-alanine compound
`(37, n=2) was thus ca 2500 times less potent than its
`structural isomer, the alanine lead compound (16, R=Me). It
`is striking therefore, that the beta- and further