Discovery of a novel series of inhibitors of human cytomegalovirus primase

Article abstract of DOI:10.1016/j.bmcl.2006.06.066

Infection by human cytomegalovirus (hCMV) remains a potent threat to susceptible people throughout the world. We have discovered a series of imidazolyl-pyrimidine compounds, which were found to be irreversible inhibitors of the hCMV UL70 primase based on

Full text of DOI:10.1016/j.bmcl.2006.06.066

Bioorganic & Medicinal Chemistry Letters 16 (2006) 4879–4883
Discovery of a novel series of inhibitors of human
cytomegalovirus primase
T. D. Cushing,* J. Adrian, X. Chen, H. DiMaio, B. Doughan, J. Flygare, L. Liang,
V. Mayorga, S. Miao, H. Mellon, M. G. Peterson, J. P. Powers, F. Spector, C. Stein,
M. Wright, D. Xu, Q. Ye and J. Jaen
Amgen Inc., 1120 Veterans Boulevard, South San Francisco, CA 94080, USA
Received 18 May 2006; revised 16 June 2006; accepted 16 June 2006
Available online 30 June 2006
Abstract—Infection by human cytomegalovirus (hCMV) remains a potent threat to susceptible people throughout the world. We
have discovered a series of imidazolyl-pyrimidine compounds, which were found to be irreversible inhibitors of the hCMV UL70
primase based on results from radiolabeling and SAR studies. Two promising analogs are described that rival ganciclovir and
cidofovir in antiviral potency and possess improved cytotoxicity profiles.
ꢀ 2006 Elsevier Ltd. All rights reserved.
In spite of advances in anti-hCMV therapy and the
introduction of (HAART),1 hCMV infection remains a
significant health problem.2 Cytomegalovirus (CMV) is
a common virus of the herpes family with estimated
infection rates of 50–85% of the United States popula-
tion. In the countries of Asia, Africa, and South Amer-
ica the incidence is upwards of 95%. Generally the virus
is dormant but in immuno-compromised patients, such
as HIV carriers, the infection rate can be as high as
90–95%, with hCMV disease manifesting itself in
30–40% of those cases. Patients undergoing transplants
often develop hCMV disease that may lead to graft
rejection, organ dysfunction, and, in the case of bone
marrow or heart–lung transplantation, interstitial pneu-
monia.3 Additionally, hCMV is the most common form
of congenital infection, with 40,000 infants affected year-
ly in the US alone; 10% of these will develop CMV dis-
ease with symptoms such as microcephaly and mental
retardation.4 Current therapy is hampered by the diffi-
culties associated with drug delivery, drug-resistant
hCMV strains,5 and pronounced side effects. Most ther-
apeutic agents are administered intravenously or have
poor oral bioavailability.6 Side effects include neutrope-
nia, thrombocytopenia, and nephrotoxicity.7
In the course of high-throughput screening against
hCMV using an hCMV-luciferase replication assay8
we discovered the imidazolyl-pyrimidine 1. Medicinal
chemistry was initiated around the core of pyrimidine
1, varying the substituents at the 2-, 4-, 5-, and 6-posi-
tions in an effort to improve the potency and the physi-
co-chemical properties of the molecule.9 Figure 1
summarizes important aspects of the SAR results. The
anti-hCMV activity (IC50) of 1 was found to be 0.3
and 0.55 lM in the viral DNA replication (dot blot)
and luciferase reporter cell-based assays, respectively.10
In general, the potency of compounds in these two
Keywords: Cytomegalovirus; CMV; Ganciclovir; Cidofovir; Imidazo-
lyl-pyrimidine; UL70 primase; Irreversible inhibitor; Radiolabeling;
Tritiation.
*
Corresponding author. Tel.: +1 650 244 2406; fax: +1 650 837
9369; e-mail: tcushing@amgen.com
Figure 1. Summary of the early SAR.
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.06.066

4880
T. D. Cushing et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4879–4883
Table 1. Anti-hCMV activity of early analogs
assays was comparable, although the dot blot assay was
used for the initial evaluation of all compounds. In
addition, the cytotoxicity of all compounds was tested
against two different cell types (Jurkat cells and the
human foreskin fibroblasts (HFF) utilized for viral
infection).
The synthesis of the majority of the analogs is outlined
in Scheme 1. 2,4-Dichloro-6-methyl-5-nitropyrimidine
(3) was prepared in a modification of the known proce-
dure.11 We found that the yield of 3 could be enhanced
by the use of pre-dried tetraethyl ammonium chloride
and anhydrous acetonitrile. Compound 3 was hydro-
lyzed selectively with sodium acetate in acetic acid to
provide compound 4,12 which on treatment with various
amine or alcohol nucleophiles, followed by POCl3,
provided an intermediate chloro adduct 6. This material
was then subjected to a refluxing solution of the desired
imidazole to give the final product 7, usually in good
overall yield. Alternatively, the 2,4-dichloropyrimidine
3 was subjected to a controlled amount of an amine
and imidazole under refluxing conditions, to provide a
complex mixture of three main isomeric products, in
which the desired material was usually the minor
component.
cellular toxicity profiles also compared favorably to
those of GCV and CDV in Jurkat and HFF cells,15
and in the more sensitive bone marrow cell assay
(colony-forming unit of granulocyte/macrophage or
CFU-GM) providing a large window between cellular
toxicity and antiviral activity (Table 2).16
Using the chemistry outlined in Scheme 1, a small
library of compounds was constructed in order to
evaluate the lead and optimize the contribution of the
2-anilino group. Some of these SAR results are shown
in Table 1. Most analogs tested were worse or similar
in potency to 1 except for benzyl amines such as the
N-methylbenzyl amine 9 and the indane-containing ana-
log 12. They inhibited viral replication (dot blot assay)
with IC50 values of 0.06 lM and 0.05 lM, respectively.
Additionally, 9 and 12 were found to inhibit viral
growth, as measured by a one-step viral yield assay, with
IC99.9 value of 0.2 lM and 0.1 lM, respectively.13 Viral
plaque formation was also completely inhibited with
IC99.9 value of 0.5 lM.14
While 9 and 12 are more potent than the original com-
pound 1, and compare favorably to GCV and CDV in
terms of antiviral activity and cytotoxicity, their overall
physicochemical profiles are not optimal. They remain
poorly soluble and are metabolically labile. For exam-
ple, incubation of 9 with rat liver homogenates for 2 h
in the presence of NADPH led to 63% degradation
and formation of two adducts, one arising from the
demethylation of the N-methyl-benzyl moiety and
another arising from the hydroxylation of the phenyl
ring. Subsequent SAR studies were carried out without
the N-methyl group and with the oxidation sites
blocked. The stability of the cyclic analog 12 was much
greater than that of 9. After 2 h incubation with rat liver
homogenate, there was only a 10% loss of parent and a
single phenolic metabolite was formed.
These two compounds showed a significant improve-
ment in anti-CMV potency relative to the two marketed
drugs cidofovir (CDV) and ganciclovir (GCV). Their
Notwithstanding the improvement of antiviral potency,
a significant amount of our initial efforts were centered
on the imidazole/nitro group dyad. It was postulated
that the activity of these compounds might be the result
of covalent bond formation between a nucleophile on an
unknown viral protein and the pyrimidine ring by SNAr
displacement of the imidazole. Interestingly, most
changes in the 4-position of the pyrimidine had large
negative effects on potency, becoming more pronounced
as the substituent deviated further from the imidazole
core.
Scheme 1. Reagents and conditions: (a) Et4NCl, dimethylaniline,
POCl3, CH3CN, 87%; (b) 1–2 equiv 2-methylimidazole, 2–4 equiv
amine, THF, ꢀ78 ꢁC, 6–10%; (c) NaOAc, AcOH, EtOH, H2O, 79%;
(d) NaOAc, amine, EtOH, reflux (R1 = 1-aminoindane 66%); (e)
POCl3, 100 ꢁC (R1 = 1-aminoindane 51%); (f) R2 = various imidazoles,
CH3CN, reflux (R1 = 1-aminoindane, R2 = 2-methylimidazole, 95%);
(g) NaOEt, EtOH, reflux, R1 = OEt, 79%.
Table 3 contains analogs with various substituted imi-
dazoles, showing that activity decreases as the size of
the substituent at the 20-position is increased. The
20-phenyl imidazole analog 16 was 250 times weaker than
the imidazole-containing compound 13. Additionally,
the structurally similar but only weakly electrophilic

T. D. Cushing et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4879–4883
4881
Table 2. Antiviral activity, cellular toxicity and bone marrow cellular toxicity comparison of 1, 9, 12, GCV (ganciclovir), and CDV (cidofovir)
Drug
In vitro activity (IC99.9 (lM))
CMV yield Plaque reduction
Cellular toxicity (IC50 (lM))
HFF toxicity
Jurkat toxicity
CFU-GM
1
2.0
0.2
0.1
5.0
1.0
2.5
0.5
0.5
100
>100
>100
>100
>100
80
25
30
30
40
30
10
9
12
100
>100
100
GCV
CDV
10
2.5
Table 3. Anti-CMV activity of imidazole analogs
Table 4. Anti-CMV activity of variously substituted pyrimidines
Analog
R
Anti-hCMV
(Luc-activity)
IC50 (lM)
24
9
H
3.0
Me
Et
0.039
0.17
3.0
25
26
27
28
i-Pr
Cl
>30
6.0
removal or elongation beyond an ethyl group led to a
reduced effect on antiviral activity (Table 4).
Because the mode of action remained in doubt, a series
of biological experiments were designed to determine if
and where the inhibitors were covalently bound. Our
synthetic approach employed alkylation of 9 with
Eshenmocher’s salt, followed by quaternization of the
amine 28 and subsequent elimination to provide an
alkene product 29. This alkene was radiolabeled with
tritium via a controlled hydrogenation, taking advan-
tage of the more reactive primary alkene relative to
the hindered and less reactive nitro moiety. This proce-
dure provided a pure sample of tritiated (50 Ci/mmol)
[3H]2-25 after an HPLC purification to remove the
byproducts of over-reduction17 (Fig. 3).
benzimidazole 18 and the unreactive pyrrole 19 were
even less active. These results are consistent with the
postulate that this series of analogs act as covalent mod-
ifiers of a target molecule, presumably a protein. In
addition, there appears to be a steric limitation with
respect to the recognition of these molecules by the
viral target, as evidenced by the significantly decreased
potency of compound 17.
Compounds were synthesized in order to replace or
remove the nitro group, to further assess the electrophilic
requirements for this lead. Various alternate electron-
withdrawing groups were selected but they all proved to
be much weaker in potency, including nitrile 20 (Fig. 2).
The [3H]2-25 was delivered to HFF cells infected with
the rhCMVLuc wild type virus. A comparison of the
protein content of infected and uninfected cells on poly-
acrylamide gel under denaturing conditions showed one
unique band that was specific to the infected cells,
An effort was then undertaken to explore the SAR
around position 6 of the pyrimidine ring, in conjunction
with a radiolabeling study (see below). It appeared that
the methyl group was required for anti-CMV activity, as
Figure 2. Analogs of 1 that lack the nitro group (all with IC50 > 30 lM).

4882
T. D. Cushing et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4879–4883
tested that shows any sensitivity to these compounds,
typically with at least 10-fold decreased potency
compared to human CMV.
In summary, we have discovered a novel class of non-
nucleoside hCMV inhibitors, which bind in an irrevers-
ible manner to hCMV UL70 primase. Two compounds,
9 and 12, have been identified in this preliminary medic-
inal chemistry effort that possess optimized profiles of
antiviral potency and a promising therapeutic window
(antiviral activity vs cytotoxicity). Additional medicinal
chemistry will be reported on our efforts to improve the
overall PK profile of this series while retaining and
enhancing their potency.
Figure 3. Synthesis of [3H]2-25.
References and notes
appearing after 48 h postinfection (Fig. 4). From this
and subsequent work, we determined that the protein
band was part of the viral helicase:primase complex.
Of the three proteins in this complex, UL102, UL105,
and UL70, it was determined through immunoprecipita-
tion experiments that the binding protein was the UL70
primase. A cartoon of this binding event is shown below
(Fig. 5).18
1. (a) Highly active antiretroviral therapy for use against HIV;
(b) Goldberg, D. E.; Smithen, L. M.; Angelilli, A.; Freeman,
W. R. Retina—J. Retin. Vitr. Disease 2005, 25, 633; (c)
Holtgrave, D. R. Int. J. STD AIDS 2005, 16, 777; (d) Luber,
A. D. Expert Rev. Anti. Infect. Ther. 2005, 3, 815; (e)
Stebbing, J.; Hatzimichael, E.; Bower, M.; Winston, A. J.
Med. Chem. 2005, 1, 635; (f) Wood, E.; Hogg, R. S.;
Harrigan, P. R.; Montaner, J. S. G. Lancet Infect. Dis. 2005,
5, 407; (g) Taylor, G. H. Am. Fam. Physician 2003, 67, 519.
2. (a) Gilbert, C.; Boivin, G. Antimicrob. Agents Chemother.
2005, 49, 873; (b) Shehu-Xhilago, M.; Tachedjian, G.;
Crowe, S. M.; Kedzierska, K. Curr. Med. Chem. 2005, 12,
1705.
Interestingly, the group of analogs described here are
highly specific for hCMV. They have no effect on
the replication of other viruses, such as HSV, EBV,
HIV, HHV6, rat CMV, guinea pig CMV, or mouse
CMV. Rhesus CMV is the only other CMV virus
3. (a) Pereyra, F.; Rubin, R. H. Curr. Opin. Infect. Dis. 2004,
17, 357; (b) Holland, N. R.; Power, C.; Mathews, V. P.;
Glass, J. D.; Forman, M.; McArthur, J. C. Neurology
1994, 44, 507.
4. (a) Schleiss, M. Herpes 2005, 12, 66; (b) Vancikova, Z.;
Dvorak, P. Curr. Drug Targets Immune Endocr. Metabol.
Disord. 2001, 1, 170.
5. Baldanti, F.; Gerna, G. J. Antimicrob. Chemother. 2003,
52, 324.
6. Drew, W. L.; Ives, E.; Spector, S. A.; Benson, C. A.;
Friedberg, D. N.; Hubbard, L.; Stempion, M. J. N. Eng. J.
Med. 1995, 333, 615.
7. (a) Brosgart, C. L.; Louis, T. A.; Hillman, D. W.;
Craig, C. P.; Alston, B.; Fisher, E.; Abrams, D. I.;
Luskin-Hawk, R. L.; Sampson, J. H.; Ward, D. J.;
Thompson, M. A.; Torres, R. A. AIDS 1998, 12, 269;
(b) Lalezari, J. P.; Stagg, R. J.; Kuppermann, B. D.;
Holland, G. N.; Kramer, F.; Ives, D. V.; Youle, M.;
Robinson, M. R.; Drew, W. L.; Jaffe, H. S. Ann.
Intern. Med. 1997, 126, 257.
Figure 4. A phosphoimage of the SDS polyacrylamide gel showing the
covalent linkage between a 110 kDa protein (UL70 primase) and
3[H]2-25. hpi, hour postinfection.
8. The luciferase screen involved
a recombinant virus
(rhCMVLuc) containing a luciferase reporter gene, under
the control of hCMV UL99 promoter, inserted in the
CMV genome between US9 and US10. HFF cells
(104 cells/well in 96-well plates, incubated for 2 h) were
infected with rhCMVLuc, at a multiplicity of infection
(MOI) of 1 plaque forming unit/cell (pfu/cell). After 1 h,
the inoculum was removed and the test drugs (0.005–5 lM
or 0.05–50 lM) in fresh media were added and incubated
for an additional 48 h at 37 ꢁC. Luciferase activity was
measured with the addition of Mg(OAc)2, 30 mM Tricine
(pH 7.8), and 200 lM EDTA. The plates were read with a
luminocount machine and the data quantified with
Microsoft Excel. IC50 values represent the concentration
of test compound required to inhibit hCMV DNA
replication by 50%.
Figure 5. Nucleophilic attack of hCMV UL-70 on compound 25.

T. D. Cushing et al. / Bioorg. Med. Chem. Lett. 16 (2006) 4879–4883
4883
9. Some aspects of this work were presented at the 41st
Interscience conference on Antimicrobial Agents and
Chemotherapy, presentations 1672 and 1673. Chicago,
IL, December 18th 2001 (posters).
quantified as before and the IC99.9 determined to be the
concentration of drug required to inhibit the hCMV
replication by 99.9%.
14. The plaque reduction assay was performed by counting
plaques from infected cells (105 cells with 1000 pfu of wild
type, clinical isolates or resistant viruses) after a 10-day
incubation. IC99.9 values represent the amount of drug
required to inhibit plaque formation by 99.9%.
10. The CMV DNA replication assay (dot blot): The HFF cells
were handled exactly as in Ref. 8 except there were 2 · 104
cells/well, rhCMVLuc, clinical isolates, and wild type
viruses were used, and the additional incubation time was
72 h not 48 h. After 72 h, the media were removed and lysis
buffer (0.4 N NaOH, 0.5 M EDTA) was added. An aliquot
was applied to a zeta-probe membrane by microfiltration,
which was washed, dried (slightly), and subjected to UV
radiation (Stratalinker) to crosslink the DNA to the
membrane. The immobilized DNA was hybridized with a
hCMV-sequence-specific radiolabeled probe, washed
again, dried, and treated to phosphoimaging (plate and
imager) to detect 32P. The data were quantified with Mac
BAS and MS Excel to provide IC50 values.
11. (a) Pesson, M. Bull. Soc. Chim. Fr. 1951, 428; (b) Hanson,
J. C. PCT Int. Appl. WO 91/01310; See also: Hodgetts, K.
J.; Yoon, T.; Huang, J.; Gulianello, M.; Kieltyka, A.;
Primus, R.; Brodbeck, R.; De Lombaert, S.; Doller, D.
Bioorg. Med. Chem. Lett. 2003, 13, 2497.
12. Clark, J.; Ramsden, T. J. Chem. Soc. (C) 1971, 679.
13. The viral yield assay involves HFF cells treated exactly as
reference 10 except wild type, clinical isolates or drug-
resistant viruses were used and the incubation time was
5–7 days not 72 h. At this time when the cytopathic effect
(CPE) was observed, the supernatant in each well was
diluted (1:2) 16 times into fresh media, and used to infect
fresh HFF cells (104 cells/well of 96-well plates). After 1 h,
each inoculum was removed and fresh media were added
to each well. The newly infected cells were incubated for 5–
7 days when CPE was again apparent. The viral yield was
15. Thecytotoxicity assay in HFF orJurkatcells was performed
by seeding cells into 96-well plates, waiting 3 h, and then
adding compound. This was followed by Alamar blue
(10 lL) staining at various time points (0, 24, 48, and 72 h).
After incubating an additional 3 h, the fluorescence was
determined (PerSeptive Biosystems CytoFluor II) and
quantified (MS Excel) providing an IC50 value, concentra-
tion of compound that inhibited cell growth by 50%.
16. The cytotoxicity assay in bone marrow cells was per-
formed in the following way. Human bone marrow cells
were harvested, washed [Iscoves modified Dulbecco’s
medium (IMDM)–2% FBS], and the viable nucleated or
mononuclear cell number was determined. The cells were
diluted with IMDM–2% FBS until there were 106 cells/mL
at which point the drug was added. The cells were diluted
1:10 methocult (stem cell technologies), vortexed, and
plated onto tissue culture dishes. After incubating at 37 ꢁC
in a humid 5% CO2 atmosphere for 15 days, the CFU-GM
colonies were quantified by microscopic inspection (invert-
ed microscope). The concentration of drug that inhibited
colony formation by 50% (IC50) was calculated.
17. American Radiolabeled Chemicals Inc., St. Louis, MO,
performed the final tritiation step.
18. Details of this and other experiments that characterize the
exact cysteine residue on UL70 responsible for imidazole
displacement will be published in due course.