Design and optimisation of orally active TLR7 agonists for the treatment of hepatitis C virus infection

Article abstract of DOI:10.1016/j.bmcl.2011.02.092

The synthesis and structure-activity relationships of a series of novel interferon inducers are described. Pharmacokinetic studies and efficacy assessment of a series of 8-oxo-3-deazapurine analogues led to the identification of compound 33, a potent and selective agonist of the TLR7 receptor with an excellent in vivo efficacy profile in a mouse model.

Full text of DOI:10.1016/j.bmcl.2011.02.092

Bioorganic & Medicinal Chemistry Letters 21 (2011) 2389–2393
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design and optimisation of orally active TLR7 agonists for the treatment
of hepatitis C virus infection
Thien-Duc Tran a, , David C. Pryde a, Peter Jones a, Fiona M. Adam a, Neil Benson c, Gerwyn Bish a,
⇑
Frederick Calo a, Guiseppe Ciaramella b, Rachel Dixon d, Jonathan Duckworth c, David N. A. Fox a,
Duncan A. Hay a, James Hitchin d, Nigel Horscroft b, Martin Howard c, Iain Gardner c,
Hannah M. Jones c, Carl Laxton b, Tanya Parkinson b, Gemma Parsons a, Katie Proctor a,
Mya C. Smith a, Nicholas Smith a, Amy Thomas b
a Discovery Chemistry, Dynamics and Metabolism, Pfizer Global Research and Development, Sandwich Laboratories, Ramsgate Road, Sandwich, Kent, CT13 9NJ, UK
b Discovery Biology, Dynamics and Metabolism, Pfizer Global Research and Development, Sandwich Laboratories, Ramsgate Road, Sandwich, Kent, CT13 9NJ, UK
c Department of Pharmacokinetics, Dynamics and Metabolism, Pfizer Global Research and Development, Sandwich Laboratories, Ramsgate Road, Sandwich, Kent, CT13 9NJ, UK
d SRG, London Office, 16, St. Helen’s place, London, EC3A 6DF, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
The synthesis and structure–activity relationships of a series of novel interferon inducers are described.
Pharmacokinetic studies and efficacy assessment of a series of 8-oxo-3-deazapurine analogues led to the
identification of compound 33, a potent and selective agonist of the TLR7 receptor with an excellent
in vivo efficacy profile in a mouse model.
Received 12 January 2011
Revised 16 February 2011
Accepted 20 February 2011
Available online 16 March 2011
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
TLR7 agonist
HCV
Interferon inducer
Aldehyde oxidase
Hepatitis C (HCV) is a global health problem currently affecting
more than 190 million people worldwide and has been recognized
as one of the leading causes of liver impairment such as cirrhosis
and hepatocellular carcinoma in humans.1 The current standard
kines, including type 1 and type 2 interferons. Among these, TLR7
is of particular interest because small molecule ligands for this
receptor have been identified, including the guanosine analogue
isatoribine,5 the mixed TLR7/8 agonists Imiquimod6 and Resiqui-
mod,7 and the 8-oxo adenine SM-3603208 (Fig. 1).
of care for HCV is a combination of pegylated interferon-
a (IFN)
and ribavirin.2 Current therapy with IFN is associated with signifi-
cant side effects including fatigue, headache, myalgia, fever and
nausea. Ribavirin also causes haemolytic anemia in 10–20% of pa-
tients. Furthermore, the administration of IFN by intramuscular or
subcutaneous injection causes pain and irritation at the site of
injection. Therefore, the development of an oral IFN inducer, which
enhances the release of endogenous IFN by oral administration, is
very much desired.
Toll-like receptors (TLRs) are a family of pathogen recognition
receptors that mediate the innate immune response.3 To date,
there have been 11 different TLRs genes identified in humans4
and their stimulation shown to cause the release of multiple cyto-
As part of our research efforts to identify new therapeutic
agents for the treatment of HCV, we recently reported the discov-
ery of the deazapurine lead candidate compound 1 as the proto-
type compound from our TLR7 agonist programme.9
NH2
C6
H
N
N7
C8
N
1
O
EC50 269nM(10)
N
N9
F3C C2
Sol. 1µg/mL
Compound 1 was shown to be highly selective against a panel of
other TLRs and has an exciting profile with a 1–50 mg qd dose pre-
dicted to induce the same level of INFa2a required to achieve clin-
⇑
Corresponding author. Tel.: +44 1304 644548; fax: +44 1304 651819.
E-mail address: thien.tran@pfizer.com (T.-D. Tran).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.02.092

2390
T.-D. Tran et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2389–2393
O
N
ical efficacy in HCV patients. However, due to its low solubility, the
NH2
N
NH2
NH2
H
N
formulation of 1 for oral administration was complicated. This
Letter describes our attempts to identify a back-up compound with
improved potency and solubility compared to compound 1.
Our strategy to improve solubility was based primarily on
driving the lipophilicity down by adding polar substituents on
the N-9 position. Test compounds were conveniently prepared by
the general methodology described in Scheme 1.
S
N
HN
N
N
N
N
N
N
N
O
O
H2N
N
O
O
O
HO
HO
HO
OH
O
Isatoribine Imiquimod
Resiquimod
SM-360320
The synthesis of intermediate 2 was described in our previous
publication.11 Compound 2 was debenzylated in concentrated
Figure 1. Structure of known TLR7 agonists.
Cl
Cl
Cl
O
Cl
N+
NO2
NO2
N
NO2
NH2
N
N
O
(b)
F
N
(c)
(a)
F
F
F
F
F
R
O
F
NH
F
N
N
F
4
F
F
3
O
O
5
F
2
O
(d)
NH2
NH2
NH2
NH2
N
NO2
H
N
N
F
(f)
(e)
F
R
N
F
N
F
R
O
O
F
N
F
N
R
F
O
O
F
F
O
7
6
8-18
Scheme 1. Reagents and conditions: (a) concd H2SO4, rt, 100%; (b) EtO2CCl, NEt3, Me–THF, rt, 80%; (c) R-X, K2CO3, NaI, acetone, rt, 65–81%; (d) concd NH3, THF, rt, 88%;
(e) Pd/C, H2, EtOH, 40psi, 40 °C, 100%; (f) AcOH, 80 °C, 71%.
Table 1
Potency and selected data for N-9 analogues
NH2
H
N
N
O
F
F
N
R
F
a
Compound
R
TLR7 EC50 (nM)
log D (c log P)
Solubility (
l
g/mL)
HLM (RLM) (
l
L/min/kg)
Rat cytosol T1/2 (min)
1
269
833
3.3 (3.3)
1
<7 (<8.5)
ND
8
9
1.9 (1.9)
2.0 (1.8)
23
37
<8 (<11)
<7 (<8.5)
ND
ND
MeO
O
1440
10
50
2.3 (2.3)
22
<7 (<8.5)
46
N
N
N
11
12
13
95
328
411
2.2 (2.3)
2.3 (2.3)
1.1 (1.1)
3
<8 (<8.5)
<8 (27)
83
3
ND
ND
N
N
21
<8 (<8.5)

T.-D. Tran et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2389–2393
2391
Table 1 (continued)
a
Compound
R
TLR7 EC50 (nM)
log D (c log P)
Solubility (
l
g/mL)
HLM (RLM) (
<8 (<8.5)
lL/min/kg)
Rat cytosol T1/2 (min)
14
15
366
1.6 (1.4)
152
ND
ND
N
N
N
352
1.8 (1.4)
ND
<8 (<8.5)
N
N
16
17
18
103
79
2.1 (1.6)
2.3 (2.2)
1.7 (1.3)
4
<8 (<9)
Stable (>1000)
Stable (>1000)
Stable (>1000)
O
8
<7 (<10)
<8 (<10)
N
N
S
193
35
O
HLM = human liver microsomes; RLM = rat liver microsomes.
a
Values are means of at least two experiments; the TLR7 agonist potency assay is described in Ref. 12.
Table 2
Structure–activity relationships of selected C-2 analogues
NH2
N
N
N
O
R
N
a
R
TLR7 EC50 (nM)
log D (c log P)
Solubility (
l
g/mL)
HLM RLM (
9 (<8.5)
lL/min/kg)
Rat cytosol T1/2 (min)
A1 (A2a) IC50 (nM)
970 (840)
O
26
27
102
1.6 (1.4)
66
46
N
N
O
N
87
2.1 (1.7)
19
<7 (<8.5)
Stable (>1000)
1600 (1500)
N
N
28
29
51
1.6 (1.7)
2.5 (1.8)
123
45
<7 (<8.5)
<8 (<8.5)
ND
ND
1900 (2300)
2800 (3400)
N
N
633
O
30
31
>4000
>4000
0.2 (2.8)
1.8 (2.3)
ND
ND
<8 (<8.5)
<8 (<8.5)
ND
ND
ND (ND)
ND (ND)
N
O
MeO
O
O
32
33
474
99
1.8 (1.9)
2.1 (2.3)
ND
17
<8 (<8.5)
<8 (<8.5)
ND
ND (ND)
O
Stable (>1000)
Not active (not active)
O
HLM = human liver microsomes; RLM = rat liver microsomes.
a
Values are means of at least two experiments; the TLR7 agonist potency assay is described in Ref. 12.
sulphuric acid at room temperature with quantitative yield.
Intermediate 3 was then converted into the corresponding carba-
mate 4 by reaction with ethyl chloroformate. Reaction of 4 with
a range of electrophiles under basic conditions provided intermedi-
ates 5 generally in good yield. Chloride displacement with concen-
trated ammonia, followed by nitro reduction under pressurised
hydrogen provided intermediate 7, which was then cyclised under
acid conditions to provide target compounds 8–18 (Table 1).

2392
T.-D. Tran et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2389–2393
Table 3
A small set of five-membered heterocycles were then designed and
Rat pharmacokinetics (IV 1 mg/kg, po 0.2 mg/kg)
characterised with this strategy in mind. Both isoxazole 16 and
oxazole 17 analogues showed promising potency and were stable
in rat cytosol (Table 1), however, the improvement in solubility
was quite modest. The oxazole analogue 18 offered a 35-fold
improvement in solubility, but when dosed in vivo 18 was 4-fold
less potent than 1 and was not pursued.14
To further improve solubility and potency, we then turned our
attention to the C-2 position. Our strategy was to keep the potency
and solubility enhancing pyridine substituent at the N-9 position
identified in compound 10 and explore a range of more polar C-2
substituents.
Compound
10
33
Cl (mL/min/kg)
PPB (%)
Cmax (nM)
T1/2 (h)
F (%)
Fabs (%)
25
83
132
0.5
78
36
41
116
0.4
17
100
40
Cl = Clearance; PPB = rat Plasma protein binding; F = bioavail-
ability; Fabs = fraction of compound absorbed.
Test compounds were prepared by the general methodology
described in Scheme 2.
Numerous N-9 analogues were synthesised and representative
examples are summarised in Table 1. Initial attempts to incorpo-
rate alkoxy side chains such as 8 or cyclic ethers such as 9 in the
N-9 position resulted in a loss in potency. A range of aromatic
substitutions were then explored to look for a more polar expres-
sion of the benzyl moiety of compound 1. First, a small number
of N-9 substituted pyridine analogues were synthesised (com-
pounds 10–12) and screened, allowing the identification of com-
pound 10 as more potent and more soluble than 1. Attempts to
introduce a second heteroatom into the pyridine ring (compounds
13–15) to improve solubility resulted in a loss of potency. Com-
pound 10 was selected to be progressed into a rat pharmacokinetic
study (Table 3). In vivo, compound 10 was completely absorbed
when administered as a suspension but the T1/2 of this compound
was quite short by virtue of metabolism of the pyridine ring by
aldehyde oxidase, which represented the major metabolic route
in rat. This was confirmed by screening the compound in either
isolated cytosol or hepatocytes, in the presence or the absence of
Raloxifene, an aldehyde oxidase inhibitor. Scaling from rat to hu-
man gave a prediction in human of liver blood flow clearance for
10. It was decided to look for alternative polar N-9 substituents
with the goal of avoiding aldehyde oxidase recognition. Aldehyde
oxidase generally oxidises the sp2 carbon adjacent to the aromatic
nitrogen atom of heterocyclic systems, so a very simple strategy
was to block the C–H adjacent to this pyridine N aromatic atom.13
The nitration of the commercially available 2,6-dichloro or
2,6-dibromo-4-amino-pyridine 19 in concentrated sulphuric acid
provided intermediate 20, which was then converted to the corre-
sponding carbamate 21 with ethyl chloroformate. Reaction of
intermediate 21 with the favoured N-9 substituent 5-chloro-
methyl-2-methyl-pyridine provided intermediate 22, which was
then converted to its corresponding amino analogue 23 with con-
centrated ammonia. When X was bromide, a range of heterocyclic
systems were introduced via Negishi coupling reaction.15 The
resulting intermediate 24 was reduced and cyclised to provide tar-
get compounds 26 and 27. When X was chloride, a direct chloride
displacement was performed with a range of nucleophiles. The
resulting intermediate 25 was reduced and cyclised to provide tar-
get compounds 28–33. Numerous C-2 analogues were synthesised
and key examples are summarised in Table 2.
The oxazole analogue 26 retained good potency with improved
solubility over compound 1. In rat cytosol, the compound was not
stable, by virtue once again of metabolism of the pyridine ring by
aldehyde oxidase. When tested against a broad panel of enzymes,
receptors and ion channels, compound 26 inhibited both adenosine
receptors A1 and A2a (Table 2). The subsequent five-membered
heterocycles synthesised such as the methyl oxazole analogue
27, N-2 triazole analogue 28 and N-1 pyrazole analogue had
similar adenosine receptor polypharmacology issues and were
X
O
O
X
O
X
X
N+
N+
N+
(a)
(b)
(c)
N
O
N
O
N
O
O
N
X
NH2
X
NH
X
N
X
NH2
22
O
O
O
20
21
N
19
X = Cl or Br
(d)
NH2
O
NH2
N+
H
NH2
O
N
N
O
O
N+
N
O
(f)
N
O
O
(e)
N
N
Ar
Ar
X
N
O
N
N
O
N
24
26-27
23
(g)
O
NH2
N+
NH2
O
N
H
N
N
O
N
Nu
(h)
N
Nu
O
O
N
25
28-33
N
Scheme 2. Reagents and conditions: (a) concd H2SO4, concd HNO3, 0 °C to rt, 70%; (b) EtO2CCl, NEt3, Me–THF, rt, 79%; (c) 5-chloromethyl-2-methyl-pyridine, K2CO3, NaI,
acetone, rt, 61%; (d) concd NH3, THF, rt, 88%; (e), X = Br, Ar-ZnCl, PdCl2(PPh3)2, microwave, 60 °C, 15 min, 86%; (f) (i) Pd/C, H2, EtOH, 40psi, 40 °C; (ii) AcOH, 80 °C, 40%; (g)
X = Cl, nucleophile, K2CO3, MeCN, 65%; (h) (i) Pd/C, H2, EtOH, 40psi, 40 °C; (ii) AcOH, 80 °C, 40%.

T.-D. Tran et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2389–2393
2393
250
200
150
100
50
induced by 33 in the mouse. The relationship between 33 pharma-
cokinetics and IFN levels was modelled in mouse using an indirect
response model with 33 stimulating the production of IFN.16
Figure 2 shows the observed IFN levels versus those predicted
using the model. The data points are dispersed around the line of
identity indicating that the model described accurately the ob-
served IFN data.
The EC50 (468 nM, coefficient of variation = 21%, corresponding
to the production of ꢀ30 pg/ml/h of IFN) from this model was cou-
pled with a viral dynamic model17 and a physiologically based
pharmacokinetic model18 to estimate potential efficacious doses
in human. Compound 33 was predicted to have comparable effi-
cacy to exogenous IFN when administered to man at doses less
than 30 mg once daily.
In summary, a number of TLR7 agonists for the treatment of
HCV were synthesized. Evaluation of the potency and pharmacoki-
netic properties of these compounds led to the identification of
compound 33, a back-up candidate to compound 1 with similar po-
tency and improved solubility to achieve suitable oral absorption
for toxicology and human clinical studies.
0
0
50
100
150
200
250
predicted data
Figure 2. Observed versus predicted data from mouse IFN model fit. The observed
data represent the observed IFN levels in the study and predicted data represent the
predicted values from the model. All units are in pg/ml.
not pursued. However, it is interesting to note that introducing a
methyl substituent to the oxazole ring was enough to confer to
analogue 27 its stability in rat liver cytosol. Introducing a basic
centre at the C-2 position resulted in a loss of potency as illustrated
by analogue 30 and aliphatic ether side chains exemplified by ana-
logue 31 were not tolerated either. The tetrahydrofuran analogue
32 had promising potency, prompting us to explore further cyclic
ether rings. The tetrahydropyran 33 had good potency, reasonable
solubility, no chirality and it shows no activity when tested against
References and notes
1. World Health Organisation. Weekly Epidemiological Record, 1999, No 49.
2. Wang, Q. M.; Heinz, B. A. Prog. Drug Res. 2000, 55, 1.
3. (a) Janeway, C. A.; Medzhitoov, R. Annu. Rev. Immunol. 2002, 20, 197; (b)
Beutler, B. Nature 2004, 430, 257.
4. O’Neill, L. A. Science 2004, 303, 1481.
5. Lee, J.; Chuang, T.; Redecke, V.; She, L.; Pitha, P. M.; Carson, D. A.; Raz, E.;
Cottam, H. B. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 6646.
6. (a) Tran, H.; Moreno, G.; Shumack, S. Expert Opin. Pharmacother. 2004, 5, 427;
(b) Miller, R. L.; Meng, T. C.; Tomai, M. A. Drug News Perspect. 2008, 21, 69.
7. Dockrell, D. H.; Kinghorn, G. R. J. Antimicrob. Chemother. 2001, 48, 751.
8. Kurimoto, A.; Tobe, T.; Ichii, S.; Isobe, Y.; Tobe, M.; Ogita, H.; Takaku, H.; Sajiki,
H.; Hirota, K.; Kawakami, H. Bioorg. Med. Chem. 2004, 12, 1091.
9. Pryde, D. C.; Tran, T. D.; Jones, P.; Parsons, G. C.; Bish, G.; Adam, F. M.; Smith, M.
C.; Hay, D.; Paradowski, M.; Proctor, K. J. W.; Parkinson, T.; Laxton, C.; Fox, D. N.
A.; Horscroft, N. J.; Ciaramella, G.; Jones, H. M.; Duckworth, J.; Benson, N.;
Harrison, A.; Webster, R. Med. Chem. Commun. 2011, doi:10.1039/C0MD00197J.
11. Bish, G.; Adam, F. M.; Calo, F.; Carr, C.; Castro, N.; Hay, D.; Hodgson, P.; Jones, P.;
Knight, C.; Parsons, G.; Proctor, K. J. W.; Pryde, D. C.; Rota, F.; Smith, N.; Tran, T.
D. Org. Process Res. Dev., in press.
TLRs 2, 3, 4, 5, 8 and 9 at 20 lM, and highly selective when tested
against a broad panel of enzymes, receptors and ion channels. The
compound was selected for further profiling. Interestingly, like 27,
compound 33 was stable in rat liver cytosol and was not a sub-
strate for aldehyde oxidase. Turning off oxidation by aldehyde oxi-
dase by adding bulky substituents remotely to the vulnerable ring
has not been reported in the literature before.13 This mitigation
strategy could be applicable for other drug discovery programmes
with aldehyde oxidase mediated oxidation issues.
12. Thomas, A.; Laxton, C.; Rodman, J.; Myangar, N.; Horscroft, N.; Parkinson, T.
Antimicrob. Agents Chemother. 2007, 51, 2969.
Following an intravenous administration to the rat, compound
33 demonstrated a mean elimination half-life of 0.4 h due to a
blood clearance of 36 ml/min/kg and a volume of distribution of
1.3 L/kg. When dosed orally as a suspension, compound 33 exhib-
ited an oral bioavailability of 17% with the amount absorbed being
estimated as 40% (Table 3).
Human pharmacokinetic predictions have been made using
allometric scaling from the rat. Scaling to man from the rat gave
a predicted blood clearance of 16 ml/min/kg, with a volume of dis-
tribution (Vss) of 2.1 L/kg and a half-life of 1.5 h. A dose prediction
13. Pryde, D. C.; Dalvie, D.; Hu, Q.; Jones, P.; Obach, R. S.; Tran, T. D. J. Med. Chem.
2010, 53, 8441.
14. Data not shown.
15. Milne, J. E.; Buchwald, S. L. J. Am. Chem. Soc. 2004, 126, 13028.
16. Benson, N.; de Jongh, J.; Duckworth, J. D.; Jones, H. M.; Pertinez, H. E.; Rawal, J.
K.; van Steeg, T. J.; Van der Graaf, P. H. Antimicrob. Agents Chemother. 2010, 54,
1179.
17. Neumann, A. U.; Lam, N. P.; Dahari, H.; Gretch, D. R.; Wiley, T. E.; Layden, T. J.;
Perelson, A. S. Science 1998, 282, 103.
18. Jones, H. M.; Parrott, N.; Jorga, K.; Lave, T. Clin. Pharmacokinet. 2006, 45, 511.
Further reading
has been made to man using the murine IFN
pharmacokinetic parameters from the rat.
a model and the scaled
10. The potency discrepancy for compound 1 between this article and Ref. 9 is due
to the use of different assay formats. In this article, a co-culture assay in a 384-
well plate format was used compared to a 96-well plate format used in Ref. 9.
A PK/PD model was constructed based on the observed clinical
efficacy of IFN 2a in HCV patients and the measured IFN 2a
a
a