MexAB-OprM specific efflux pump inhibitors in Pseudomonas aeruginosa. Part 3: Optimization of potency in the pyridopyrimidine series through the application of a pharmacophore model

Article abstract of DOI:10.1016/j.bmcl.2003.10.060

The addition of substituents to the pyridopyrimidine scaffold of MexAB-OprM specific efflux pump inhibitors was explored. As predicted by a pharmacophore model, the incorporation substituents at the 2-position improved potency. Piperidines were found to be optimal, and further introduction of polar groups without compromising the activity was shown to be feasible. Careful positioning of the essential acidic moiety of the pharmacophore relative to the scaffold led to the discovery of vinyl tetrazoles with still greater potency.

Full text of DOI:10.1016/j.bmcl.2003.10.060

Bioorganic & Medicinal Chemistry Letters 14 (2004) 475–479
MexAB-OprM specific efflux pump inhibitors in Pseudomonas
aeruginosa. Part 3: Optimization of potency in the
pyridopyrimidine series through the application of a
pharmacophore model
Kiyoshi Nakayama,a,* Haruko Kawato,a Jun Watanabe,a Masami Ohtsuka,a
Ken-ichi Yoshida,a Yoshihiro Yokomizo,a Atsunobu Sakamoto,a Noriko Kuru,a
Toshiharu Ohta,a Kazuki Hoshino,a Kumi Yoshida,a Hiroko Ishida,a Aesop Cho,b
Monica H. Palme,b Jason Z. Zhang,b Ving J. Leeb and William J. Watkinsb
aDaiichi Pharmaceutical Co., Ltd, 1-16-13, Kitakasai, Edogawa, Tokyo 134-8630, Japan
bEssential Therapeutics, Inc., 850 Maude Avenue, Mountain View CA 94043, USA
Received 6 October 2003; revised 8 October 2003; accepted 20 October 2003
Abstract—The addition of substituents to the pyridopyrimidine scaffold of MexAB-OprM specific efflux pump inhibitors was
explored. As predicted by a pharmacophore model, the incorporation substituents at the 2-position improved potency. Piperidines
were found to be optimal, and further introduction of polar groups without compromising the activity was shown to be feasible.
Careful positioning of the essential acidic moiety of the pharmacophore relative to the scaffold led to the discovery of vinyl tetra-
zoles with still greater potency.
# 2003 Elsevier Ltd. All rights reserved.
Peudomonas aeruginosa is an opportunistic pathogen
whose intrinsic resistance to a wide variety of anti-
microbial agents (including quinolones, b-lactams, ami-
noglycosides, macrolides and tetracyclines) is largely
attributable to the expression of several efflux pumps,
four of which have been identified to date: MexAB-
OprM, MexCD-OprJ, MexEF-OprN, and MexXY-
OprM.1À4 We recently reported the discovery and initial
optimization5,6 of agents that are specific for the inhibi-
tion of MexAB-OprM. Our initial efforts were directed
towards the improvement of physicochemical properties
of the initial lead 1 (Fig. 1) so as to achieve in vivo
activity. Two main strategies were explored: the reduc-
tion of high protein binding and increase in aqueous
solubility through the incorporation of polar groups
into the structure,5 or redesign of the molecule by
utilizing alternative scaffolds.6
The latter strategy culminated in the discovery of a ser-
ies of pyridopyrimidines (typified by 2) whose protein
binding was greatly improved and which were effica-
cious in the potentiation of the quinolones Levofloxacin
(LVFX) and Sitafloxacin (STFX) in vivo. The design of
2 was based upon the pharmocophore model depicted in
Figure 2.6 It is apparent from inspection of this model
that 1 bears a hydrophobic moiety that is not present in
2, and that the introduction of 2-substituents on the
pyridopyrimidine scaffold might furnish more potent
analogues. Herein, we report the results of SAR studies
on the pyridopyrimidine scaffold designed to exploit
Keywords: Efflux pump inhibitor; drug resistance; MexA,B-OprM
efflux pump.
* Corresponding author. Tel.: +81-3-5696-9079; fax:+81-3-5696-
8772; e-mail: nakaygoi@daiichipharm.co.jp
Figure 1. Structures of the first generation of MexAB-OprM specific
efflux pump inhibitors.
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.10.060

476
K. Nakayama et al. / Bioorg. Med. Chem. Lett. 14 (2004) 475–479
philic displacement was straightforward. More than 80
cyclic, acyclic, mono- and di-alkylamines were intro-
duced in this manner. Representative examples are
shown in Figure 3.
The activity of the efflux pump inhibitors (EPIs) was
assessed using PAM1723,10 an experimental strain of P.
aeruginosa in which the MexAB-OprM pump is over-
expressed and MexCD-OprJ and MexEF-OprN are
genetically disrupted. Data for the potentiation of two
MexAB-OprM substrates, LVFX and aztreonam
(AZT), in the presence or absence of human serum
albumin (HSA) is displayed in Table 1. To quantify the
activity of the inhibitors, we defined the term MPCn as
the minimum concentration (mg/mL) of inhibitor
required to reduce (potentiate) the activity of anti-
bacterial drug n-fold. The protein binding was measured
for those analogues that displayed reasonable activity
that was not dramatically affected by the addition of
HSA.
Figure 2. Pharmacophore model generated by CATALYSTTM7 and
overlaid structures of compound 1 (red) and 2 (green). Light blue
spheres are hydrophobic sites, and deep blue for acidic ionizable site.
the extra hydrophobic binding pocket, together with
optimization of the acidic moiety, leading to molecules
possessing high MexAB-OprM pump inhibitory activity.
In a brief initial study to explore the optimum position
of the acidic moiety relative to the scaffold,6 we dis-
covered that analogues in which the tetrazole in 2 was
replaced by an acrylic acid moiety displayed increased
potency (data not shown). Despite poor stability due to
extremely facile photoisomerization of the olefin8 the
potency benefit, coupled with ease of synthesis, drove
the decision to incorporate this change at the outset of
subsequent efforts to add extra hydrophobic sub-
stituents on the scaffold. The ethylene tether between
the aromatic entities in 2 was retained in this exercise in
order to keep the protein binding as low as possible.6
6-Membered cyclic amines furnished the most active
compounds (Table 1, 16–26); smaller (13–15) or larger
(28) rings resulted in less potent or inactive analogues.
Some acyclic alkylamino derivatives were active (6 and
12), but the potency was less than that of the cyclic
amines (e.g., 13, 16). Among the 6-membered cyclic
amines, piperidine (16, 18, 19, 21, 22, 25, 26) and mor-
pholine (17) derivatives provided the best results. Inter-
estingly, incorporation of neutral polar residues such as
hydroxyl (18, 19, 21, 22 versus 16) or carbamoyloxy
groups (25, 26 versus 16) on the 3- or 4-position of
piperidine did not affect the activity, although amino
substituents at these positions were not tolerated (23,
24). 3-Substituted piperidines were more soluble than
the 4-substituted isomers (data not shown). In general,
the protein binding of novel active analogues was simi-
lar to that of 2. Although its activity was moderate
compared to the piperidines, the lower protein binding
of the piperazine variant (27) was notable.
Scheme 1 depicts the route by which the unsaturated
acidic analogues were made.9 The use of activated
malonic acid derivatives in the generation of the pyr-
idopyrimidine was strategically convenient for two rea-
sons. Formylation at the 3-position was facilitated by
the presence of the 2-OH group, and after tosylation,
introduction of hydrophobic 2-substituents by nucleo-
Scheme 1. Synthesis of 2-substituted analogues.

K. Nakayama et al. / Bioorg. Med. Chem. Lett. 14 (2004) 475–479
477
Scheme 2. Synthesis of acrylic acid variants.
Table 1. In vitro potentiation activity of novel EPIsa
MPC8 (AZT) mg/mLb
MPC4 (LVFX)
mg/mLb
MPC8 (LVFX)
mg/mLb
Serum protein bindingd
(%)
Without HSA
With 0.125% HSA
1
2
3
4
5
6
7
8
2
16
16
16
32
64
128
32
>64
>64
>32
64
32
32
8
32
64
8
32
1
>16c
8
8
32
2
8
16
8
8
16
8
2
4
8
0.5
2
0.5
1
8
2
1
2
>16c
>32c
64
>98
81.5
—
82.1
—
—
—
—
—
—
—
—
—
—
69.3
—
92
82.1
—
—
91.5
—
—
—
91.2
>32c
64
128
32
>64c
>64c
>32c
64
32
16
8
32
128
4
16
8
16
64
16
>128c
16
>64c
>64c
>32c
64
128
64
8
32
128
4
16
8
8
64
16
8
128
>128c
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
16
16
64
16
16
16
>128c
128
32
128
128
16
16
32
8
32
1
2
8
16
16
128
16
32
128
67.8
—
128
8
a Versus PAM1723.10 All the analogues were devoid of intrinsic antibacterial activity (MIC >320 mg/mL).
b Values are for 100% growth inhibition.
c Compound precipitation observed.
d Rat serum was used.

478
K. Nakayama et al. / Bioorg. Med. Chem. Lett. 14 (2004) 475–479
Table 2. In vitro potentiation activity of acrylic acid surrogatesa
MPC8 (AZT); mg/mLb
MPC4 (LVFX); mg/mLb
MPC8 (LVFX); mg/mLb
Without HSA
With 0.125% HSA
17
18
29
30
31
16
8
32
16
>128c
4
16
2
16
8
>128c
4
16
0.5
32
0.5
2
>128c
4
8
a Versus PAM1723.10 All the analogues were devoid of intrinsic antibacterial activity (MIC >320 mg/mL).
b Values are for 100% growth inhibition.
c Compound precipitation observed.
to be much worse. Notable was the activity of the tet-
razole variant (30), which displayed very high potency
along with no apparent influence of the addition of
HSA. However, the solubility in aqueous media was
inferior to the carboxylic acid counterpart (18).
Of the novel analogues, we confirmed that 17 and 18,
like 2,6 potentiated LVFX in murine models of infection
with P. aeruginosa. Details of the in vivo experiments
will be reported elsewhere.
In conclusion, the in vitro activity of 2 was significantly
enhanced by the incorporation of a hydrophobic group
at the 2-position of the pyridopyrimidine scaffold, con-
sistent with the pharmacophore model for this series of
MexAB-OprM-selective EPIs. In addition, we have dis-
covered that hydrophilic substitution was feasible on a
piperidine moiety at the 2-position without compromis-
ing activity, providing a potential avenue for the
enhancement of aqueous solubility. The inclusion of an
olefin spacer between the tetrazole and pyridopyr-
imidine scaffold in 2 gave a further boost in potency,
although the structural motif was prone to photo-
chemical isomerization.8 Efforts to address this issue
will be reported in due course.
Figure 3. Representative amines introduced to 2-position of pyr-
idopyrimidine scaffold. The number under the structure represents the
compound number after attachment to the scaffold. Yields for the last
two steps (Scheme 1) are also shown below the structure.
References and notes
1. Poole, K.; Krebes, K.; McNally, C.; Neshat, S. J. Bacter-
iol. 1993, 175, 7363.
2. Poole, K.; Gotoh, N.; Tsujimoto, H.; Zhao, Q.; Wada, A.;
Yamasaki, T.; Neshat, S.; Yamagishi, J.; Li, X. Z. Nish-
ino, Mol. Microbiol 1996, 21, 713.
3. Koehler, T.; Michea-Hamzehpour, M.; Henze, U.;
Gotoh, N.; Curty, L. K.; Pechere, J. C. Mol. Microbiol.
1997, 23, 345.
4. Mine, T.; Morita, Y.; Kataoka, A.; Mizushima, T.; Tsu-
chiya, T. Antimicrob. Agents Chemother. 1999, 43, 415.
5. Nakayama, K., Ishida, Y., Ohtsuka, M., Kawato, H.,
Yoshida, K., Yokomizo, Y., Hosono, S., Ohta, T., Hos-
hino, K., Ishida, H., Yoshida, K., Renau, T. E., Leger,
R., Zhang, J. Z., Lee, V. J., Watkins, W. J. Bioorg. Med.
Chem. Lett., in press.
Next, further optimization of the acid surrogate, invol-
ving both exploration of the linker between the scaffold
and the acidic moiety itself, was addressed. The synthesis
of the derivatives is displayed in Scheme 2. Simple
hydrogenation of 18 provided the saturated variant (29).
The requisite tetrazole was synthesized via the cyano-
ethylamide, followed by treatment with triflic anhydride
and sodium azide and deprotection under basic condi-
tions. Finally, the acetylene-linked carboxylic acid was
generated utilizing a Colvin rearrangement11 and Pd-
mediated carbonylation12 as the key steps.
6. Nakayama, K., Ishida, Y., Ohtsuka, M., Kawato, H.,
Yoshida, K., Yokomizo, Y., Ohta, T., Hoshino, K.,
Otani, T., Kurosaka, Y., Yoshida, K., Ishida, H., Lee, V.
J., Renau, T. E., Watkins, W. J. Bioorg. Med. Chem.
Lett., in press.
7. CatalystTM, version 4.6; Accelrys, Inc: San Diego, CA,
2002.
The in vitro activity of the acrylic acid surrogates along
with the parent compounds (17 and 18) is displayed in
Table 2. Saturation of the double bond (29) resulted
in almost complete loss of activity, whereas the potency
of the acetylene variant was similar to that of the vinyl
parent (17 versus 31), although the stability was found

K. Nakayama et al. / Bioorg. Med. Chem. Lett. 14 (2004) 475–479
479
8. Exposure of solutions of E-isomers to house light for as
little as 2 h was sufficient to generate predominantly the
Z-isomer.
Renau, T.; Leger, R.; Hecker, S.; Watkins, W.; Hoshino,
K.; Ishida, H.; Lee, V. Antimicrob. Agents Chemother.
2001, 45, 105.
9. Hermecz, I.; Horvath, A. J. Heterocycl. Chem. 1992, 29,
559.
11. Feldman, K. S.; Cutarelli, T. D.; Florio, R. D. J. Org.
Chem. 2002, 24, 8528.
10. Lomovskaya, O.; Warren, M. S.; Lee, A.; Galazzo, J.;
Fronko, R.; Lee, M.; Blais, J.; Cho, D.; Chamberland, S.;
12. Vasilevsky, S. F.; Trofimov, B. A.; Mal’kina, A. G.;
Brandsma, L. Synth. Commun. 1994, 24, 85.