Quinazolinone fungal efflux pump inhibitors. Part 2: In vitro structure-activity relationships of (N-methyl-piperazinyl)-containing derivatives

Article abstract of DOI:10.1016/j.bmcl.2004.07.071

Structure-activity relationships of a novel series of fungal efflux pump inhibitors with respect to potentiation of the activity of fluconazole against strains of Candida albicans and Candida glabrata over-expressing ABC-type efflux pumps are systematical

Full text of DOI:10.1016/j.bmcl.2004.07.071

Bioorganic & Medicinal Chemistry Letters 14 (2004) 5133–5137
Quinazolinone fungal efflux pump inhibitors.
Part 2: In vitro structure–activity relationships of
(N-methyl-piperazinyl)-containing derivatives
a
a
a
a
William J. Watkins,a, Remy C. Lemoine, Lee Chong, Aesop Cho, Thomas E. Renau,
*
´
Bonnie Kuo,a Vickie Wong,a Maria Ludwikow,a Negar Garizi,a Nadeem Iqbal,b
John Barnard,b Renata Jankowska,b Rajeshwar Singh,b Deidre Madsen,a Karen Lolans,a
Olga Lomovskaya,a Uma Ozaa and Michael N. Dudleya
aEssential Therapeutics, Inc., 850 Maude Ave., Mountain View, CA 94043, USA
bNAEJA Pharmaceutical Inc., #2 4290-91 A St., Edmonton, Alberta, Canada T6E 5V2
Received 18 May 2004; accepted 28 July 2004
Available online 20 August 2004
Abstract—Structure–activity relationships of a novel series of fungal efflux pump inhibitors with respect to potentiation of the activ-
ity of fluconazole against strains of Candida albicans and Candida glabrata over-expressing ABC-type efflux pumps are systemati-
cally explored.
ꢀ 2004 Elsevier Ltd. All rights reserved.
1. Introduction
O
N
N
N
N
The design and execution of a screening strategy for the
discovery of inhibitors of efflux pumps in Candida albi-
cans has recently been described.1 From these efforts
two main chemical series emerged: natural products of
the milbemycin type,2,3 and synthetic compounds based
upon a quinazolinone core. Screening hits in the latter
series showed exquisite selectivity for the inhibition of
an ABC-type pump in C. albicans (CDR1); the lack of
activity against azole-resistant Candida spp. (particu-
larly C. glabrata), and the very limited aqueous solubil-
ity gave cause for concern about the viability of the
series as a starting point for optimization. However,
exploratory chemistry showed that the introduction of
a basic substituent gave a dramatic improvement in
aqueous solubility and increased the spectrum of inhibi-
tion to include Candida spp. other than C. albicans4
These studies culminated in the discovery of MC-5805
(1, Fig. 1) and stimulated a more thorough exploration
H
N
N
O
Cl
O
1
O
Figure 1. MC-5805 (1).
of the structural requirements for activity elsewhere in
the molecule.
We now report the structure–activity relationships for
in vitro potentiation of fluconazole (FLU) versus C. albi-
cans and C. glabrata for 3-(N0-methyl-piperazinyl)quin-
azolinones of this type.
2. Chemistry
New compounds were initially synthesized as racemates
using routes analogous to that used for 1, which
is shown in Scheme 1. The enantiomers of 2 were
*
Corresponding author at present address: Gilead Sciences, 333
Lakeside Drive, Foster City, CA 94404, USA. Tel.: +1 650 522 5384;
fax: +1 650 522 5899; e-mail: w.watkins@gilead.com
0960-894X/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.07.071

5134
W. J. Watkins et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5133–5137
Table 1. Variants of the 3-chlorophenyl urea moiety in 1
O
O
N
O
N
(a), (b)
OH
NH2
N
N
N
N
N
N
N
O
(c), (d)
R
LINK
O
No. Link
R
MPC8 (lg/mL)
C. albicans C. glabrata
O
N
N
O
N
N
2
N
N
N
N
(e)
1
CONH
H
(3-Cl)Ph
––
iPr
1
4
H
N
2
>32
>32
2
>32
>32
1
N
O
Cl
NH
O
3
CONH
CONH
CONH
CONH
CONH
CONH
CONH
CONH
CONH
O
4
(4-Cl)Ph
(2-Cl)Ph
Ph
O
O
1
5
>32
16
4
>32
16
4
6
Scheme 1. Racemic synthesis. Reagents and conditions: (a) (EtCO)2O,
70ꢁC, forming the isatoic anhydride; (b) 1-amino-4-methylpiperazine,
AcOH, 70ꢁC; (c) Br2, AcOH, NaOAc; (d) 2,4-dimethoxyaniline,
N-methylpyrrolidinone, 80ꢁC; (e) 3-chlorophenylisocyanate, 1,2-
dichloroethane.
7
(4-F)Ph
(4-Me)Ph
8
16
16
32
16
4
9
(4-OMe)Ph
(3-CN)Ph
(3-Cl)-6-
2
10
11
32
8
Pyridazinyl
2-Benzthiazolyl
(4-Cl)Ph
12
13
14
15
16
17
CONH
CO2
SO2
2
16
32
separated by chiral HPLC, facilitating the preparation
of enantiomerically pure final products.
32
32
>32
4
(4-Cl)Ph
>32
>32
>32
>32
CONHSO2 (4-Cl)Ph
CO
CO
(4-Cl)Ph
(2-Cl)Ph
The synthesis of all variants of the quinazolinone core
started from the requisite anthranilic acid. Displacement
of the bromide by 2,4-dimethoxyaniline leading to 2 re-
quired unusually forcing conditions, indicative of signif-
icant steric hindrance. Although the yield of 2 was
satisfactory (85%), it was lower for less nucleophilic rea-
gents (e.g., 2,4-difluoroaniline, 48%) and the reaction
failed altogether for (3-trifluoromethylsulfonyl)aniline.
Variants of the (3-chlorophenyl)urea were made by reac-
tion of the aniline 2 with the appropriate electrophile.
Thus for the carbamate 13, sulfonamide 14 and the
amide 16 the aniline was treated with (4-chloro)phenyl-
chloroformate, (4-chloro)phenylsulfonyl chloride and
(4-chloro)benzoyl chloride, all in the presence of trieth-
ylamine; for the acylsulfonamide 15, treatment with (4-
chloro)phenylsulfonyl isocyanate in the absence of base
provided the desired material.
4
conformation, forcing the aryl ring out of co-planarity
with the urea linkage. Overall, the 3- or 4-halo-substi-
tuted compounds 1, 4 and 7 show the best balance of po-
tency and activity against both Candida spp.
It proved difficult to find a heterocyclic replacement for
the phenyl ring: even the relatively conservative pyrid-
azine isostere was significantly less active (cf. 11 vs 4),
and although other more lipophilic alternatives such as
the benzthiazole 12 showed good activity against C. alb-
icans, the C. glabrata potency was reduced.
A brief survey of alternative linkers demonstrated that
the urea was preferred. A carbamate retained the bal-
anced activity versus C. albicans and C. glabrata but
was significantly less potent (cf. 13 vs 4), and neither sul-
fonamide 14 nor the acylsulfonamide 15 proved an ade-
quate surrogate. Amide 16 gave good C. albicans
potency, but was inactive against C. glabrata. In con-
trast to the SAR of the ureas, the 2-chlorophenyl amide
17 is as active as the 4-chloro isomer 16. This is, in fact,
unsurprising, as the conformational restriction of terti-
ary benzamides such as these (wherein the aromatic ring
is forced out of the plane of the amide for steric reasons,
even the parent phenyl case) is well known;5 this merely
reinforces the argument that a conformational change
accounts for the disparate activity of the urea 5. None-
theless, the overall shape of analogues containing an
amide link clearly differs significantly from the ureas,
and the fact that C. albicans activity is seen in both indi-
cates either that the binding site of this particular pump
is flexible or that more than one binding site may exist.
Similarly, the divergent SAR across Candida spp. indi-
3. Structure–activity relationships
The relative activity of analogues was conveniently
expressed as the minimum potentiation concentration
(MPC8), which is the lowest concentration achieving
an 8-fold reduction in FLU MIC. Initially, variants of
the 3-chlorophenyl urea moiety were examined, and
Table 1 displays the results for representative examples.
The lack of activity of 2 confirmed the requirement for
some sort of derivative of the free aniline, and the
inactivity of the alkyl urea 3 illustrates the need for an
aromatic substituent. A clear preference for meta- or
para- over ortho-substitution is evident among the
chloro-substituted phenyl ureas 1, 4 and 5; the latter is
even less active than the unsubstituted analogue 6, and
this observation leads us to presume that the presence
of the 2-substituent might drive a change in the local

W. J. Watkins et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5133–5137
Table 2. Variants of the 2,4-dimethoxyphenyl moiety in 1 Table 3. Variants of the quinazolinone core in 1
5135
O
N
O
N
5
N
N
N
N
R
N
R
N
N
H
N
8
H
N
N
O
O
O
O
F
F
No.
R
MPC8 (lg/mL)
No.
R
MPC8 (lg/mL)
C. albicans
C. glabrata
C. albicans
C. glabrata
7
(2,4-diOMe)Ph
Me
Ph
4
4
7
H
4
4
18
19
20
21
22
23
24
25
26
27
28
29
>32
>32
>32
8
>32
>32
>32
32
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
5-F
6-F
7-F
8
8
1
2
(2-OMe)Ph
(4-OMe)Ph
(3-OMe)Ph
0.5
0.125
0.25
0.125
0.5
0.5
8
1
6,7-diF
6-Cl
7-Cl
0.5
2
16
>32
>32
8
>32
>32
>32
>32
32
(2,4,6-triOMe)Ph
(3,4-diOMe)Ph
(3,4,5-triOMe)Ph
(2,4-diF)Ph
0.5
4
6-Me
6-OMe
8-OMe
6-OH
2
8
8
(2,4-diMe)Ph
(2,4-diOCH2cPr)Pha
3-(2,6-diOMe)pyridyl
8
>32
2
16
4
1
4
16
6-OCOCH2CH3
6-OCH2CONH2
6-NHSO2CH3
6-NHCOCH3
7-CN
4
32
>32
16
>32
32
>32
2
a cPr = cyclopropyl.
>32
0.5
7-CO2M
7-CO2H
e
32
2
cates either that the binding sites of the pumps differ or
that the mode of binding of the inhibitors is different (or
both). Overall, we regarded the (4-fluoro)phenyl urea 7
as the preferred substituent for this region of the lead,
based upon its potency, balanced spectrum and reason-
able lipophilicity.
>32
32
>32
32
7-Aza
8-Aza
>32
>32
We next examined the requirements for activity around
the dimethoxyphenyl residue. The results for key ana-
logues are displayed in Table 2.
rather, we assume that the affinity is driven by a p-stack-
ing interaction that requires a lipophilic, electron-rich
aromatic ring, together with possible hydrogen bond(s)
to the substituents.
The structural requirements for activity at this position
are stringent. Alkyl substituents such as 18 were com-
pletely devoid of activity, as was the simple phenyl ana-
logue 19. Among methoxy substituted phenyl
derivatives, there is a clear and very strong preference
for the parent 2,4-disubstitution pattern; removal of
either of these (20, 21) leads to dramatic loss in activity.
Similarly, all other mono- or tri-substituted variants
(22–25) are much less active. Not only is the 2,4-disub-
stitution pattern important for activity, but the need
for alkoxy substituents is highlighted by the relative
inactivity of 26 and 27. However, the size of the alkoxy
groups is also important; although 28 potentiated FLU
satisfactorily in the strains shown, it exhibited mild
intrinsic antifungal activity versus C. albicans (MIC
32lg/mL) and S. cerevisiae (data not shown). Finally,
it proved possible to retain modest activity by replacing
the phenyl ring with a suitably substituted pyridyl surro-
gate 29, although we regard this as a rather conservative
change, given the inductive effect of the flanking substit-
uents on the heteroatom.
Having established requirements for activity in the two
isolated phenyl rings, we turned our attention to
the quinazolinone itself, for which results are shown in
Table 3.
The presence of a halo substituent at the 6- or 7-posi-
tions of the quinazolinone (30–35) increased potency
against both yeasts, and small alkyl or alkoxy substitu-
ents at the 6-position (36, 37) also improved activity,
particularly versus C. albicans; the 8-methoxy analogue
38, however, was slightly less active. Wishing to reduce
the lipophilicity of the lead as much as possible, we ex-
plored more polar substituents at the 6-position and
were gratified to find some residual activity in phenol
39. The less polar propionate derivative 40 restored
the potency, but the a-acetamide 41 was not active.
The weak activity of sulfonylanilide 42 and the lack of
activity of the (more polar) acetanilide 43 provide fur-
ther evidence of this trend.
The 7-cyano derivative 44 proved quite active, but its
isolation was complicated by a marked tendency to-
wards hydrolysis of the quinazolinone amide bond.
Ester 45 was considerably less active, and unsurpris-
ingly, acid 46 showed no activity at all.
Given the extreme rigidity of this portion of the lead4
(the aryl ring is forced out of the plane of the urea by
the adjacent quinazolinone), it is unlikely that the ob-
served SAR is due to any change in conformational bias;

5136
W. J. Watkins et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5133–5137
Table 4. In vitro activity of enantiomers of 32
O
N
N
O
N
N
N
N
N
N
F
F
H
N
H
N
N
O
N
O
O
O
O
O
F
F
50
49
No.
Concn required for 50% pump inhibition in S. cerevisiae (lg/mL)a
MPC8 (lg/mL)
C. glabrata
CDR1
CDR2
CgCDR1
CgCDR2
C. albicans
49
50
0.5
>32
16
>32
2
>32
32
>32
0.25
>32
0.5
>32
a See Materials and methods.
In keeping with the general requirement for lipophilicity
in the quinazolinone portion of the pharmacophore, two
analogues in which the carbacyclic portion of the bicycle
were replaced by more polar heterocycles (47, 48) were
inactive.
methanesulfonic acid salt; no precipitation was observed
at the test concentrations employed for assessment of
antifungal activity. The structural identity of each com-
1
pound was confirmed by H NMR and MS.
The enantiomers of 2 were isolated by chromatography
using a Chiralpack AD 250 · 4.6mm column, eluting
with 30% 2-propanol in hexanes at 1mL/min. Absolute
stereochemistry was assigned by analogy with relative
retention times of the enantiomers of 1.4
The activity of one of the most potent racemic analogues
above (32) was shown to reside in the (S)-enantiomer 49.
Like 1, 49 is selective for the inhibition of CDR1 in
C. albicans and CgCDR1 in C. glabrata (Table 4).
5.2. In vitro potentiation of fluconazole
4. Summary and conclusions
FLU MICs for C. albicans strain YEM15 (over-express-
ing both CDR1 and CDR2) and C. glabrata strain
YEM19 (over-expressing both CgDR1 and CgDR2)
were measured in the presence and absence of varying
concentrations of efflux pump inhibitor (checkerboard
format). Drugs were tested in RPMI 1640 according to
NCCLS reference methods.6 The in vitro potency of
the inhibitor is expressed as the lowest concentration
achieving an 8-fold reduction in FLU MIC (minimum
potentiation concentration, or MPC8). All the com-
pounds reported had no intrinsic antifungal activity
(MIC > 32lg/mL) with the exception of 28 (see text).
Not all of the analogues reported in this study were as-
sessed in full checkerboard format; an abbreviated
method of evaluation involved testing varying concen-
trations of efflux pump inhibitor in the presence of fixed
concentrations of FLU equivalent to 1/8 and 1/32 MIC.
This method was validated on a set of eight analogues
by comparison with the MPC8 values derived from full
checkerboard studies; the results were identical.
In this and the previous study,4 the absolute structural
requirements for satisfactory efflux pump inhibition
in vitro by 3-(N0-methyl-piperazinyl)quinazolinones such
as 1 have been outlined. All three aromatic entities are
necessary, and the optimal linkers are shown to be those
present in the original lead. A halo substituent at the 3-
or 4-position of the phenyl urea is preferred, and there
is a strong requirement for 2,4-dialkoxy substitution on
the other phenyl ring. Significant enhancements in po-
tency were achieved through the introduction of halo
substituents on the quinazolinone ring. It proved impos-
sible to incorporate significantly more polar substituents
anywhere on the three aromatic residues without signifi-
cantly compromising efflux pump inhibitory activity.
Future publications will describe the SAR for fungal
efflux pump inhibition of variants of the N-methyl pip-
erazine residue, and the means by which the pharmaco-
kinetics of the series were optimized.
The inhibition of specific ABC transporters from C. alb-
icans and C. glabrata was assessed using defined strains
of S. cerevisiae in which the various pumps from Can-
dida spp. are overexpressed under the control of a com-
mon promoter: YEM139 (PDR5 deleted; FLU MIC
0.5lg/mL), YEM172 (CDR1 over-expressed; FLU
MIC 128lg/mL), YEM171 (CDR2 over-expressed;
FLU MIC 8lg/mL), Y170 (CgCDR1 over-expressed;
FLU MIC 128mug/mL) and Y218 (CgCDR2 over-
5. Materials and methods
5.1. Chemicals
All analogues were purified by reverse-phase MPLC
using a Phenomenex Synergy Hydro-RP 50 · 21.2 mm
column, eluting with MeCN/water (10–100%, ramped
over 20min) at 20mL/min and tested as the TFA or

W. J. Watkins et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5133–5137
5137
expressed; FLU MIC 2lg/mL). An inoculum of 104 cells
of each strain was incubated at 30ꢁC in YPD in the pres-
ence of varying concentrations of inhibitor and a single
concentration of FLU corresponding to the geometric
mean of the MIC of the test strain and the MIC of
the parent pump-deleted strain, for example, YEM172
was incubated with 8lg/ml FLU, which is the geometric
mean of 128 (MIC of YEM 172) and 0.5 (MIC of
YEM139). MICs were determined as 80% inhibition of
growth after 48h.
Agents and Chemotherapy, Toronto, ON, September
2000, Abstract 1500.
2. Lemoine, R.; Yang, J.; Staley, A.; Mathias, K.; Lolans, K.;
Vijayam, U.; Shao, L.; Price, M. E.; Watkins, W. J.;
Dudley, M. N. In 41st Interscience Conference on Antimi-
crobial Agents and Chemotherapy, Chicago, IL, December
2001, Abstract 2160.
3. Lemoine, R. C.; Yang, J.; Lolans, K.; Oza, U.; Blais, J.;
Staley, A.; Lee, M.; Lomovskaya, O.; Watkins, W. J.
Bioorg. Med. Chem. Lett., submitted for publication.
4. Lemoine, R. C.; Glinka, T. W.; Watkins, W. J.; Cho, A.;
Yang, J.; Iqbal, N.; Singh, R.; Madsen, D.; Lolans, K.;
Lomovskaya, O.; Oza, U.; Dudley, M. N. Bioorg. Med.
Chem. Lett. 2004, 14, preceding paper in this issue.
doi:10.1016/j.bmcl.2004.07.070.
References and notes
5. Clayden, J.; Johnson, P.; Pink, J. H.; Helliwell, M. J. Org.
Chem. 2000, 65, 7033.
6. NCCLS. 1997. Reference method for broth dilution anti-
fungal susceptibility testing of yeasts: approved standard.
NCCLS document M27-A.
1. Mistry, A.; Warren, M. S.; Blais, J.; Staley, A. L.; Galazzo,
J. L.; Fuernkranz, H.; Chamberland, S.; Lee, M. D.;
Watkins, W. J.; Lomovskaya, O.; Miller, G. H.; Sanglard,
D. In 40th Interscience Conference on Antimicrobial