Design and synthesis of novel benzofurans as a new class of antifungal agents targeting fungal N-myristoyltransferase. Part 2

Article abstract of DOI:10.1016/S0960-894X(01)00808-3

Modification of the C-2 position of a benzofuran derivative 6 (RO-09-4609), an N-myristoyltransferase (Nmt) inhibitor, has led us to discover antifungal agents that are active in a murine systemic candidiasis model. The drug design is based on the analysi

Full text of DOI:10.1016/S0960-894X(01)00808-3

Bioorganic & Medicinal Chemistry Letters 12 (2002) 607–610
Design and Synthesis of Novel Benzofurans as a
New Class of Antifungal Agents Targeting Fungal
N-Myristoyltransferase. Part 2
Hirosato Ebiike,a Miyako Masubuchi,a Pingli Liu,a Ken-ichi Kawasaki,a
Kenji Morikami,a Satoshi Sogabe,a Michiko Hayase,b Toshihiko Fujii,b
Kiyoaki Sakata,b Hidetoshi Shindoh,c Yasuhiko Shiratori,a Yuko Aoki,b
Tatsuo Ohtsukaa,* and Nobuo Shimmaa
aDepartment of Chemistry, Nippon Roche Research Center, 200 Kajiwara, Kamakura, Kanagawa 247-8530, Japan
bDepartment of Mycology, Nippon Roche Research Center, 200 Kajiwara, Kamakura, Kanagawa 247-8530, Japan
cDepartment of Preclinical Science, Nippon Roche Research Center, 200 Kajiwara, Kamakura, Kanagawa 247-8530, Japan
Received 29 August 2001; accepted 28 November 2001
Abstract—Modification of the C-2 position of a benzofuran derivative 6 (RO-09-4609), an N-myristoyltransferase (Nmt) inhibitor,
has led us to discover antifungal agents that are active in a murine systemic candidiasis model. The drug design is based on the
analysis of a crystal structure of a Candida Nmt complex with 2. The optimization has been guided by various biological evalua-
tions including a quasi in vivo assay and pharmacokinetic analysis. # 2002 Elsevier Science Ltd. All rights reserved.
N-Myristoyltransferase (Nmt) has been proven to be
essential for the viability of fungi, including medically
important pathogenic fungi, Candida albicans1 and
Cryptococcus neoformance,2 making it a possible target
for the development of fungicidal drugs with a novel
mode of action. An Nmt inhibitor could therefore be a
new type of antifungal agent. Because of its novel
mechanism of action, an Nmt inhibitor is expected to
have advantages over azole antifungal agents in terms
of activity against azole resistant fungal strains and lack
of the drug–drug interactions that are a drawback of the
azole agents.3 In our previous paper,4 we reported our
optimization of the C-3 and C-4 substituents of a novel
Nmt inhibitor, a benzofuran derivative 1, discovered by
random screening. The modification resulted in identifi-
cation of 6 (RO-09-4609) that showed improved activity
in both C. albicans Nmt (CaNmt) inhibitory activity
(IC50: 0.1 mM) and in vitro antifungal activity (IC50
against C. albicans CY1002: 1.6 mM) as compared with
1. This compound, however, did not show in vivo effi-
cacy in a murine systemic candidiasis model, probably
because of its weak antifungal activity and its short
elimination half-life (t1/2: 0.42 h) in mice. The ester
group of 6 was easily hydrolyzed by esterases in the
mice to give an inactive carboxylic acid metabolite. To
overcome these problems, we modified the C-2 sub-
stituent of compound 6. The inhibitors were designed
and optimized based on the analysis of the X-ray struc-
ture of a CaNmt complex with a benzofuran inhibitor 24
and the structure–activity relationship (SAR) analysis.
The modification was guided by the enzyme inhibitory
activity, the antifungal activity in YNBPB medium, the
antifungal activity in 80% calf serum (quasi in vivo
assay), and the pharmacokinetic (PK) properties. The
quasi in vivo assay could be more predictive for in vivo
efficacy than the conventional in vitro antifungal assay,
since the former assay system includes a factor of drug–
protein binding.
*Corresponding author. Tel.: +81-467-45-4356; fax: +81-467-45-
6824; e-mail: tatsuo.ohtsuka@roche.com
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00808-3

608
H. Ebiike et al. / Bioorg. Med. Chem. Lett. 12 (2002) 607–610
In this paper, we wish to report the design and synthesis
of metabolically more stable derivatives at the C-2
position of compound 6 and their biological activities
including antifungal activity in vivo.
with Boc followed by oxidation of the resulting alcohol
9 with manganese dioxide gave the aldehyde 10. Wittig
reaction of 10, hydrogenation of the resulting olefin,
and subsequent deprotection gave the phenethyl deri-
vative 11. The phenylether derivatives 12a–12k were
synthesized from alcohol 7 and an appropriate phenol
by Mitsunobu reaction. In most cases, a modified Mit-
sunobu reaction with 1,10-(azodicarbonyl)dipiperidine
(ADDP) and tributylphosphine in toluene5 gave better
yields (ca. 60%) than the standard Mitsunobu reaction
(DEAD-TPP/THF).
Synthesis
N-Phenylcarbamoyl derivative 5 was prepared from
ester 3 in a good yield as outlined in Scheme 1. Ester 3
was hydrolyzed by lithium hydroxide, and the resulting
acid was converted into its acid chloride by treatment
with thionyl chloride, which was immediately subjected
to the next anilide formation reaction. The resulting N-
phenylcarbamoyl compound 4 was heated to 70 ꢀC with
3-aminomethylpyridine in EtOH to give 5.
Results and Discussion
The binding mode of the compound 6 derivatives to
CaNmt and their SARs
As shown in Scheme 2, phenylthioether 8, a phenethyl
derivative 11, and phenylethers 12a–12k were synthe-
sized from alcohol 7 that was prepared by lithium alu-
minum hydride reduction of 6. The treatment of alcohol
7 with thiophenol and trifluoroacetic acid (TFA) in di-
chloromethane gave the thioether 8. N-Protection of 7
The crystal structure of a CaNmt complex with 2
revealed that the C-2 substituent of 2 was surrounded
by three phenylalanine residues, Phe 115, Phe 240, and
Phe 339.4 To strengthen the binding between the C-2
substituent and these phenylalanine residues by aro-
matic–aromatic interaction, we introduced a phenyl
group to the C-2 position via various linkers: –CONH–
(5), –CH2S– (8), –CH2CH2– (11), and –CH2O– (12a).
We also expected that the replacement of the C-2 ester
with the above-mentioned linkers would increase the
metabolic stability of the inhibitors. Both the CaNmt
inhibitory activity and the antifungal activity of these
compounds were strongly affected by the linker struc-
ture. Namely, the compound 12a having an ether linker
was the most potent among the C-2 derivatives 5, 8, 11,
and 12a (Table 1). This result suggested that the ether
oxygen atom in 12a and one of the ester oxygens in 6
were interacting with CaNmt through hydrogen bond-
ing. To further increase the binding affinity, we intro-
duced electron-withdrawing groups to the phenyl group
of 12a, because an aromatic ring having electron-with-
Scheme 1. Preparation of N-phenylcarbamoyl derivative 5.
Scheme 2. Preparation of benzofuran derivatives.

H. Ebiike et al. / Bioorg. Med. Chem. Lett. 12 (2002) 607–610
609
Table 1. Enzyme inhibitory activity, antifungal activity, and pharmacokinetic (PK) properties
Compd
Enzyme inhibition (mM)
Antifungal activity (mM)
PK parameterse
AUC (ngꢂh/ml)
CaNmta
HsNmtb
Serum(ꢁ)c
Serum(+)d
t1/2 (h)
5
6
2.2
0.1
0.62
1.2
0.072
0.0083
0.11
0.0052
0.0037
0.0075
0.0057
0.028
0.073
0.017
0.0094
NT
>480
>450
>470
12
12
1.6
11
13
NTf
NT
180
>240
NT
240
330
390
290
670
560
370
340
500
420
330
360
330
280
0.9
0.4
1.4
1.1
1.9
1.6
1.5
1.9
1.9
1.6
2.0
3.6
2.4
0.5
1.2
4.0
8
11
12a
12b
12c
12d
12e
12f
12g
12 h
12i
12j
12k
Fluconazole
77
0.37
>470
>110
67
0.04
0.11
0.021
0.08
0.03
0.035
0.074
0.92
1.2
0.6
3.3
0.7
0.57
0.34
0.33
1.3
6.6
3.3
62
>450
>430
>400
>110
19
>460
NT
0.16
0.72
1.4
0.5
340
19,600
aInhibitory activity against C. albicans Nmt (CaNmt) as assessed by IC50 using substrate peptide GLTISKLFRR-NH2 (0.5 mM) and myristoyl-CoA
at 0.5 mM.
bInhibitory activity against human Nmt (HsNmt) as assessed by IC50 using substrate peptide GNAASARR-NH2 (0.5 mM) and myristoyl-CoA at
0.5 mM.
cAntifungal activity against C. albicans CY1002 as assessed by IC50 in YNBPB medium (1% glucose, 0.25% K2HPO4, pH 7).
dAntifungal activity against C. albicans CY1002 as assessed by IC50 in 80% calf serum (80% calf serum supplemented with 10 mM FeCl3 6H2O,
.
10 mM deferoxamine, 2% dextrose).
eCassette dosing: five compounds were intravenously administered to a rat and the plasma concentration of each compound was measured by LC–
MS.9
fNT, not tested.
drawing groups is known to interact with another aro-
matic ring more strongly than does an unsubstituted
aromatic ring.6 We thus introduced fluoro, chloro,
bromo, and cyano groups to the phenyl ring of the
phenylether 12a. As we expected, most of the phenyl-
ether derivatives having electron-withdrawing groups
showed stronger inhibitory activity against CaNmt than
the unsubstituted phenylether 12a (Table 1). Among
them, the 2-fluoro 12b, 4-fluoro 12d, 2,3-difluoro 12e,
2,4-difluoro 12f, 2,3,4-trifluoro 12g, and 4-cyano 12k
derivatives showed very strong inhibitory activity
against CaNmt (IC50: 3.7–9.4 nM). In contrast, the 3-
fluoro derivative 12c was 10–30 times less active than
those derivatives. The variation of the inhibitory activ-
ity among the ether derivatives could be explained by
the electron density on the ether oxygen that was calcu-
lated by the electrostatic potentials around aromatic
rings of the compounds using a molecular orbital
method. The theoretical calculation and the interpreta-
tion of the SARs will be reported in detail separately.
A crystal structure (3.5 A) of a CaNmt complex with
12e was obtained by a soaking experiment and the
structure of the binding site is shown in Figure 1. The
positions of the fluorine atoms could not be well defined
because of their poor electron density. The data strongly
supported the supposition that the aromatic–aromatic
interaction between the aryl group at the C-2 position
and the hydrophobic pocket formed by Phe 115, Phe
240, and Phe 339 of CaNmt was very important in
gaining binding affinity. The aryl group at C-2 position
is exactly situated between two phenylalanine residues,
Phe 240 and Phe 339. The crystal structure also sug-
gested important hydrogen bonds between the amido-
NH2 of Asn 392 and the ether oxygen, and between one
of the ring nitrogens of His 227 and the ring oxygen of
the benzofuran. The importance of the latter hydrogen
bond may be supported by very weak inhibitory activity
(IC50: 8.6 mM) of the benzothiophene analogue of com-
pound 12f, [3-[2-(2,4-difluoro-phenoxymethyl)-3-methyl-
benzothiophene-4-yloxy]-propyl]-pyridin-3-ylmethyl-
amine. The lack of activity of a benzothiophene analo-
gue may be caused by the difference in hydrogen accep-
tor ability between the sulfur and oxygen atoms and/or
the geometrical difference between the thiophene and
furan rings. The phenylethers 12a–12k showed extre-
mely high selectivity to CaNmt over HsNmt (Table 1).
The quasi in vivo and in vivo antifungal activities
An in vitro antifungal assay containing serum was
thought to be more predictive for in vivo efficacy than
the conventional in vitro assay without serum. We
Figure 1. The crystal structure of a CaNmt complex with 12e.

610
H. Ebiike et al. / Bioorg. Med. Chem. Lett. 12 (2002) 607–610
adopted an in vitro assay system in 80% calf serum
.
gations, including the quasi in vivo assay and the cas-
sette dosing PK studies in rats.9 This ‘multi-dimensional
optimization’ enabled us to rapidly identify the com-
pounds that are active in vivo, starting from the enzyme
inhibitor 1 with only marginal antifungal activity in
vitro.
supplemented with 10 mM FeCl3 6H2O, 10 mM deferox-
amine, and 2% dextrose in our assay cascade (quasi in
vivo assay).7,8 The compounds 12b, 12d, 12e, 12f and
12g were as potent as fluconazole (IC50: 0.5 mM in this
assay protocol) in serum, although their antifungal
activity was reduced by the addition of calf serum.
Furthermore, all the phenylether derivatives in Table 1
showed longer t1/2 than did compound 6.
In conclusion, we have discovered a new class of anti-
fungal agents that selectively and strongly inhibited
CaNmt and exhibited in vivo efficacy. Because of their
unique mode of action, the Nmt inhibitors may over-
come the drawbacks of current drugs such as resistant
development and drug–drug interactions in azole anti-
fungal agents as reported.3 The identification of potent
and selective CaNmt inhibitors, 12f and 12g, having
enhanced in vivo efficacy warrants further modification
studies to develop a new class of fungicidal agents.
Since compounds 12f and 12g exhibited the highest
quasi in vivo antifungal activity and compound 12h
showed the longest t1/2 among the phenylether deriva-
tives, they were further evaluated in a rat systemic
candidiasis model. Fisher rats (n=5) were infected
intravenously with a lethal dose of C. albicans (CY1002)
and treated iv with multiple doses (three times per day
for 2 days) of the test compound. Efficacy of the com-
pounds was calculated as the effective dose (mg/kg) for
50% survival (ED50) on day 7. The fluoro derivatives
12f and 12g were active in this model with ED50s of
7.1 mg/kg. Although compound 12h had a longer t1/2
than did the other compounds, it showed much weaker
in vivo efficacy in this model (ED50: >30 mg/kg) than
12f or 12g. This in vivo result correlated with its weaker
antifungal activity in the quasi in vivo assay. Thus, the
in vivo efficacy was well predicted by the results of the
quasi in vivo assay and the PK study. Although these
compounds showed in vivo efficacy, they were still 10
times weaker than fluconazole in the rat systemic candi-
diasis model. Their weaker in vivo activity might be
caused by their poorer PK profiles than those of fluco-
nazole. Thus further modifications are required to
develop a clinically useful fungicidal agent. Since the
antifungal activity [serum (ꢁ)] against C. albicans of the
benzofurans showed a good correlation with the enzyme
inhibition activity against CaNmt, the growth inhibition
of C. albicans was considered to be caused by the inhib-
ition of the enzyme. The correlation coefficient (R) and F
values of the correlation between log IC50 values of anti-
fungal activity and those of enzyme inhibition activity
were calculated to be 0.892 and 7.92ꢃ10ꢁ6, respectively.
Acknowledgements
We thank Prof. Tetsuto Tsunoda of Tokushima Bunri
University for useful advice on improving the Mitsu-
nobu reaction conditions.
References and Notes
1. Weinberg, R. A.; McWherter, C. A.; Freeman, S. K.;
Wood, D. C.; Gordon, J. I.; Lee, S. C. Mol. Microbiol. 1995,
16, 241.
2. Lodge, J. K.; Jackson-Machelski, E.; Toffaletti, D. L.; Per-
fect, J. R.; Gordon, J. I. Proc. Natl. Acad. Sci. U.S.A. 1994,
91, 12008.
3. Benedetti, M. S.; Bani, M. Drug Metab. Rev. 1991, 31,
665.
4. Masubuchi, M.; Kawasaki, K.; Ebiike, H.; Ikeda, Y.; Tsu-
jii, S.; Sogabe, S.; Fujii, T.; Sakata, K.; Shiratori, Y.; Aoki, Y.;
Ohtsuka, T.; Shimma, N. Bioorg. Med. Chem. Lett. 2001, 11,
1833.
5. Tsunoda, T.; Yamamiya, Y.; Ito, S. Tetrahedron Lett. 1993,
34, 1639.
6. Hunter, C. A.; Sanders, J. K. M. J. Am. Chem. Soc. 1990,
112, 5525.
7. Masubuchi, K.; Okada, T.; Kohchi, M.; Sakaitani, M.;
Mizuguchi, E.; Shirai, H.; Aoki, M.; Watanabe, T.; Kondoh,
O.; Yamazaki, T.; Satoh, Y.; Kobayashi, K.; Inoue, T.;
Horii, I.; Shimma, N. Bioorg. Med. Chem. Lett. 2001, 11,
395.
8. Meunier, F.; Lambert, C.; Van der Auwera, P. J. Anti-
microb. Chemother. 1990, 25, 227.
9. Frick, L. W.; Adkison, K. K.; Wells-Knecht, K. J.; Wool-
lard, P.; Higton, D. M. Pharm. Sci. Technol. Today 1998, 1,
12.
In summary, the extensive modification of the C-2 sub-
stituent of a novel CaNmt inhibitor, 6 (RO-09-4609),
which showed moderate antifungal activity in vitro, led
to identification of the novel derivatives, 12f and 12g,
which were active in a rat systemic candidiasis model.
The modification work was carried out by the combi-
nation of rational drug design based on the crystal
structure of the binary complex of CaNmt with 2 and
the SAR analysis guided by various biological investi-