Structure-based design of new DHFR-based antibacterial agents: 7-aryl-2,4-diaminoquinazolines

Article abstract of DOI:10.1016/j.bmcl.2011.07.059

Dihydrofolate reductase (DHFR) inhibitors such as trimethoprim (TMP) have long played a significant role in the treatment of bacterial infections. Not surprisingly, after decades of use there is now bacterial resistance to TMP and therefore a need to develop novel antibacterial agents with expanded spectrum including these resistant strains. In this study, we investigated the optimization of 2,4-diamnoquinazolines for antibacterial potency and selectivity. Using structure-based drug design, several 7-aryl-2,4- diaminoquinazolines were discovered that have excellent sub-100 picomolar potency against bacterial DHFR. These compounds have good antibacterial activity especially on gram-positive pathogens including TMP-resistant strains.

Full text of DOI:10.1016/j.bmcl.2011.07.059

Bioorganic & Medicinal Chemistry Letters 21 (2011) 5171–5176
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Structure-based design of new DHFR-based antibacterial agents:
7-aryl-2,4-diaminoquinazolines
Xiaoming Li, Mark Hilgers, Mark Cunningham, Zhiyong Chen, Michael Trzoss, Junhu Zhang,
Lucy Kohnen, Thanh Lam, Chris Creighton, Kedar GC, Kirk Nelson, Bryan Kwan, Mark Stidham,
⇑
Vickie Brown-Driver, Karen J. Shaw, John Finn
Trius Therapeutics, San Diego, CA 92121, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Dihydrofolate reductase (DHFR) inhibitors such as trimethoprim (TMP) have long played a significant role in
the treatment of bacterial infections. Not surprisingly, after decades of use there is now bacterial resistance to
TMP and therefore a need to develop novel antibacterial agents with expanded spectrum including these
resistant strains. In this study, we investigated the optimization of 2,4-diamnoquinazolines for antibacterial
potency and selectivity. Using structure-based drug design, several 7-aryl-2,4-diaminoquinazolines were
discovered that have excellent sub-100 picomolar potency against bacterial DHFR. These compounds have
good antibacterial activity especially on gram-positive pathogens including TMP-resistant strains.
Ó 2011 Elsevier Ltd. All rights reserved.
Received 6 May 2011
Revised 14 July 2011
Accepted 14 July 2011
Available online 23 July 2011
Keywords:
Antibacterial
DHFR
Staphylococcus aureus
Inhibitor
Selectivity
Dihydrofolate reductase (DHFR EC 1.5.1.3) catalyses the NADPH
dependent reduction of dihydrofolate to tetrahydrofolate and is
critical for the biosynthesis of purines, thymidylate and some
amino acids.1 DHFR is a well known target in several therapeutic
areas including cancer and antiinfectives where it is used to gener-
ate antibacterial, antifungal and antiparasitic agents.2
Historically, most antibacterial DHFR agents are based on the
5-benzyl-2,4-diaminopyrimidine scaffold due its excellent potency
and selectivity for bacterial versus mammalian DHFR. The most
widely used DHFR antibacterial agent is trimethoprim (TMP),
which has been used for decades to treat a variety of bacterial dis-
eases.3 Not surprisingly, bacterial resistance to TMP is significant in
both gram-positive and gram-negative pathogens.4 To expand cov-
erage to TMP-resistant pathogens, more potent 5-benzyl-2,4-
diaminopyrimidines have been discovered as exemplified by
iclaprim (ICL) a drug that has been clinically investigated through
Phase 3 trials.5 Due to their inherent selectivity and potency, the
5-benzyl-2,4-diamino-pyrimidine class continues to be an active
area of research.6–9
NH2
N
NH2
N
H3CO
H3CO
O
N
NH2
H3CO
N
NH2
OCH3
OCH3
trimethoprim (TMP)
iclaprim (ICL)
not useful antibacterial drugs due to poor selectivity. Compounds
based on the 2,4-diaminoquinazoline scaffold are also highly po-
tent DHFR inhibitors and have been investigated as antibacterial
agents.10–13 The majority of these efforts have focused on 6-substi-
tuted derivatives that explore the same SAR vectors as methotrex-
ate. While many quinazolines in this series had good antibacterial
activity on Staphylococcus aureus, these compounds had little selec-
tivity (10ꢀ) for bacterial versus the mammalian enzymes10,11 In
addition, in vivo efficacy in animal models was not demonstrated
and some toxicity was observed.
DHFR inhibitors based on bicyclic scaffolds especially the
2,4-diaminopyrido[2,3-d]pyrimidine ring of methotrexate (MTX)
and the 2,4-diaminopyrido[2,3-d]pyridine ring piritrexim (PTX)
are well known for their high enzymatic potency.6 MTX is used
as an anticancer agent and PTX is used as an antiparasitic agent.
Although these compounds have antibacterial activity, they are
CO2H
OCH3
NH2
N
O
HO2C
N
H
NH2
N
N
N
NH2
N
N
N
OCH3
⇑
Corresponding author.
E-mail address: jfinn@triusrx.com (J. Finn).
N
NH2
methotrexate (MTX)
piritrexim (PTX)
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.07.059

5172
X. Li et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5171–5176
Despite these negative findings, from a structural viewpoint, the
4ꢀ more potent DHFR inhibitors with a similar potency gain in
antibacterial activity compared to the 6-unsubstituted analog.
There was little difference in selectivity between compounds with
or without the 6-methyl group. Selectivity for simple members of
the series is modest (50–250ꢀ range) and far below that observed
for the 5-benzyl-2,4-diaminopyrimidines TMP (15,000ꢀ) and ICL
(9,000ꢀ).
2,4-diamino-quinazoline ring fills the deep binding pocket of DHFR
more completely than 5-benzyl-2,4-diaminopyrimidines like TMP.
The activity of the unsubstituted 2,4-diaminoquinazoline 1 sup-
ports this observation. As reported by Harris,11 compound 1 is a
good DHFR inhibitor with submicromolar potency on S. aureus
DHFR and displays surprising antibacterial activity for a small frag-
ment. For these reasons, we selected the 2,4-diaminoquinaoline
scaffold as a starting point for structure-guided optimization with
the goal of using information from co-crystal structures to engi-
neer in potency and selectivity. In this report, we explore the
SAR of 7-aryl substituted quinazolines.
Multiple co-crystal structures of different compounds in this
series with S. aureus DHFR were determined. One of the first struc-
tural themes noticed was that the torsional angle around the qui-
nazoline-aryl bond is dependent on the substituent in the ortho
position. Figure 1 shows the structure of the 23-S. aureus DHFR
complex overlayed with the TMP-S. aureus DHFR structure. This
structure shows that the aryl ring adopts a conformation that is
roughly on the same plane and fills the same pocket as the trimeth-
oxygroup of TMP. The ortho-ethoxy group of 20 fills a pocket close
to NADPH in a similar fashion to one of the methoxy groups of
TMP. Filling this pocket is energetically favorable and helps fix
the conformation of the 7-aryl ring. In contrast, analogs lacking
an ortho substituent adopt different conformations as shown by
the structure of the 12-S. aureus DHFR structure in Figure 2. Table
2 compiles the observed X-ray conformation of different com-
pounds in Table 1. Interestingly, compound 11, containing a smal-
NH2
MW 160
N
S. aureus DHFR Ki = 93 nM
S. aureus MIC = 32 µg/mL
N
NH2
1
Synthesis of 7-aryl-2,4-diaminoquinazolines was readily
accomplished by a two step process.14 A wide variety of biaryl
compounds 4 were prepared by Suzuki coupling of 4-bromobenzo-
nitriles 2 with boronic acids 3. The resulting products were con-
verted to multiple 2,4-diaminoquinazolines using guanidine
carbonate and base. Alternatively, the scheme can be reversed
and 7-bromo-2,4-diaminoquinazoline 29 prepared first and then
used in the Suzuki reaction.
ler ortho methoxy substituent, yielded
a diffraction pattern
indicating that the compound was bound in two conformations:
Route A
R6
Br
HO
CN
F
H2N
N
Guanidine
carbonate
CN
F
R6
R6
R
Pd[0]
2
N
+
DMF
K2CO3
microwave
DMF
NH2
OH
B
4
R
5-7, 9-17, 21-22, 25-28
R
3
Route B
H2N
H2N
Guanidine
R6
Br
CN
R6
Br
Pd[0]
R6
carbonate
N
N
NMP
K2CO3
microwave
DMF
NH2
N
NH2
F
N
2
29
+
R
HO
OH
B
7-8,18-20, 23-24
R
3
Compounds were tested for antibacterial activity against
one resembling the 23 conformation found for and one resembling
the 11 conformation. Figure 3 compares the 11 and TMP S. aureus
DHFR bound structures.
gram-positive and gram-negative bacteria using standard CLSI mic-
rodilution methods. These assays were conducted in Mueller Hin-
ton-cationic adjusted medium either with or without 20% mouse
serum.15 Enzymatic activity was determined by measuring the oxi-
dation of NADPH.16 Compounds were co-crystallized with the S. aur-
eus DHFR enzyme in space group P212121 and typically diffracted to
1.7 Å or better.17 Structures were solved by molecular replacement.
Table 1 compiles the enzymatic and antibacterial data for a variety of
7-aryl-2,4-diaminoquinazolines. Those compounds that were co-
crystallized with S. aureus DHFR are indicated in italic numbers.
As was expected from previous work in the quinazoline area,
simple 7-phenyl derivatives have reasonable potency and antibac-
terial activity but poor selectivity. For example, the 7-phenyl ana-
logs 5 and 6 are good DHFR inhibitors (Ki of 9.1 and 1.6 nM on
S .aureus DHFR) with good antibacterial activity (MIC of 8 and
Using the observation that an ortho-ethoxy phenyl ring posi-
tions the aromatic group in the same desirable space as the aryl
rings of trimethoprim and iclaprim, we prepared additional
ortho-ethoxyphenyl analogs containing substituents in the 5 posi-
tion. Small groups in the 5 position such as methyl, formyl, chloro
and bromo have modest potency (S. aureus DHFR Ki between
0.24–0.85 nM for the 6-methyl compounds 20–23), indicating a
small contribution from the 5-substituent. Likewise, the addition
of a pyridine ring system in the 5-postion provided a compound
with similar potency, as demonstrated by compound 25 (S. aureus
DHFR Ki 0.30 nM). In contrast, compound 26 containing a 30-sul-
fonamide biphenyl group is a much more potent DHFR inhibitor
(S. aureus DHFR Ki 0.020 nM). The 26-S. aureus DHFR co-crystal
structure (Fig. 4) shows that the sulfonamide group forms a hydro-
gen bond with the conserved residue Arg58. This Arg group is used
1
lg/mL on S. aureus). This pair of compounds demonstrated that
compounds containing the 6-methyl group are generally around

Table 1
7-Aryl-2,4-diaminoquinazoline Antibacterial and Enzymatic SAR
H2N
R6
R7
N
MIC (lg/mL)
Ki (nM)
NH2
N
Cmpd/Route
R7
R6
S. aureus
S.aureus + seruma
E. faecalis
H. influnezae
E. coli
S. aureus DHFR
H. influenzae DHFR
Human DHFR
Ratiob
TMP
ICL
5/A
6/A
7/B
—
—
Ph
Ph
—
—
H
Me
H
Me
H
Me
H
Me
Me
Me
Me
H
1
0.13
8
1
4
1
8
2
2
0.5
1
0.5
0.25
4
1
0.5
0.03
8
0.5
8
0.5
16
1
4
0.5
4
0.5
0.25
4
0.5
1
16
2
4
2
16
8
16
4
8
4
1
4
16
8
32
8
>64
32
32
8
16
8
1.24
0.081
9.1
1.6
2.7
0.89
5
1.2
0.19
0.007
0.52
0.0073
0.16
0.0018
0.22
19070
775
743
129
702
15000
9500
82
0.13
64
2
2
1
8
1
4
0.5
1
81
4-NH2C6H4
4-NH2C6H4
4-CH3SO2C6H4
4-CH3SO2C6H4
2-CH3OC6H4
3,4-(CH3O)2C6H4
2,5-(CH3O)2C6H4
2-CH3C6H4
2-CH3CH2C6H4
2-CH3CH2OC6H4
260
155
51
37
421
35
8/B
9/A
138
253
10/A
11/A
12/A
13/A
14/A
15/A
16/A
0.0023
0.21
44.1
1,138
12.2
27.5
95.6
41.1
85
2.7
0.35
0.66
0.77
0.19
3.1
0.0018
0.014
0.037
0.032
0.27
42
2
0.5
8
124
216
27
8
16
8
32
17/A
18/B
Me
H
0.25
4
1
8
0.5
4
8
16
32
0.13
4.4
0.037
0.310
20.9
323
161
73
CH(CH3)2
O
64
19/B
20/B
O
Me
Me
0.5
1
1
1
0.5
0.5
1
2
4
0.48
0.24
0.006
0.22
26.3
12
55
50
H
O
16
OCH2CH3
Cl
21/A
22/A
23/B
Me
Me
Me
1
1
1
2
4
8
0.5
0.5
0.5
8
8
0.38
0.85
0.76
0.023
0.034
0.044
11.1
22.2
11.1
29
26
15
OCH2CH3
OCH2CH3
16
32
16
32
Br
OCH2CH3
OCH2CH3
HN
24/B
25/A
Me
H
0.25
0.5
1
8
0.25
0.5
16
16
8
0.61
0.3
0.082
0.011
245
402
34
N
>16
10.1
OCH2CH3
(continued on next page)

5174
X. Li et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5171–5176
Figure 1. Overlay of the S. aureus DHFR bound structures of TMP (grey) and
compound 23 (orange).
Figure 2. Overlay of the S. aureus DHFR bound structures of TMP (grey) and
compound 12 (orange).
Table 2
Effect of ortho group on ring conformation
Cmpd Aryl in TMP plane? Yes/no S. aureus DHFR Ki (nM) Ortho group
8
No
0.89
2.7
None
OMe
None
None
OEt
OEt
OEt
OEt
11
12
19
20
23
26
27
Yes/no two conformations
No
No
Yes
Yes
Yes
Yes
0.35
0.48
0.24
0.76
0.020
0.026
to bind one of the acids of the glutamic acid part of dihydrofolate
and MTX. Figure 4 shows the S. aureus bound structures of 26 (or-
ange) overlayed with the MTX structure (white). The incorporation
of a group that binds to this Arg has been previously demonstrated
in the 5-benzyl-2,4-diaminopyrimidine series.18
An alternative strategy to highly potent inhibitors of bacterial
DHFR was to add hydrophobic groups in the 5 or the 5,6 position.
The
a-napthyl analog 27 uses this strategy and is a very potent
S. aureus DHFR inhibitor (S. aureus DHFR Ki 0.026 nM). The 27-
S. auerus DHFR structure shows that the new second ring sits in a
similar position as the dihydropyran ring of ICL and is surrounded
by hydrophobic residues Leu29, Val32 and Leu55. Figure 5 shows the
S. aureus bound structures of 27 (orange) overlayed with the ICL
structure (white). The 5-diethylamide, compound 28 was designed
to add hydrophobicity in this region and resulted in a highly potent
compound (S. aureus DHFR Ki 0.013 nM).
Overall, the enzymatic potency of the lead compounds (25–28) in
this series compare favorably to the standards TMP and ICL and are
significantly better than the quinazolines previously reported in

X. Li et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5171–5176
5175
the literature. In general, compounds in this series displayed good
enzymatic spectrum with typically enhanced activity on
Haemophilus influenzae DHFR compared to S. aureus DHFR. While
the design aspect focused on the S. aureus enzyme, some analogs
were surprisingly highly potent on H. influenzae DHFR. For example,
while the difference in enzymatic potency between the 6-methyl
and the 6-unsubstituted quinazoline is about 4ꢀ on S. aureus DHFR,
the differences in H. influenzae DHFR is often dramatic ranging from
50ꢀ to 116ꢀ (compare pairs 5–6, 7–8, 9–10, and 18–19).
Modeling the human enzyme with the bacterial enzyme sug-
gested that some selectivity would be gained by building into the
hydrophobic pocket of the 5 and 6 position of the ortho-ethoxy-
phenyl ring. Of the analogs in Table 1, only compound 28 showed
an improved selectivity (ratio of 1360) but this is still well below
the level seen with TMP and ICL. It is likely that there is some sig-
nificant conformational flexibility in the region of the human DHFR
enzyme surrounding the 7-aryl group, and this flexibility com-
bined with the multiple conformations available to 7-aryl-quina-
zolines yields potency in the sub-100 nM for the human DHFR
enzyme.
Figure 3. Overlay of the S. aureus DHFR bound structures of TMP (grey) and the two
conformations of compound 11 (orange).
Antibacterial activity for the 7-aryl series is generally good on
gram-positive strains but weaker on gram-negatives. While the
activity of the most potent compounds, for example 26–28, are
comparable to ICL on S. aureus, they are significantly weaker on
gram-negative pathogens such as H. influenzae and Escherichia coli.
While most compounds have the same MIC value on S. aureus with
or without serum, some of the more hydrophobic compounds
show higher MIC values in 20% mouse serum indicating high ser-
um binding for these compounds.
Due to their increased enzymatic potency on bacterial DHFR,
compounds in this series with potent antibacterial activity on wild
type S. aureus also have antibacterial activity on TMP-resistant
S. aureus strains comparable to ICL. Table 3 compares the enzy-
matic and antibacterial activity of 27 and 28 to TMP and ICl on both
wild type and the F99Y TMP-resistant strain. The F99Y mutation is
the most common form of TMP resistance in S. aureus. Similar to
ICL, the ratio of antibacterial potency on resistant versus wild type
strains parallels the enzymatic potency.
Figure 4. Overlay of the S. aureus DHFR bound structures of MTX (white) and
compound 26 (orange).
In conclusion, the SBDD optimization of the 7-aryl-2,4-diamino-
quinazoline series provided several highly potent inhibitors of
bacterial DHFR with good antibacterial activity especially on
gram-positive pathogens. While the selectivity of these com-
pounds on human DHFR was less than desired, the structural infor-
mation provides direction for additional optimization in
selectivity.
Acknowledgments
We wish to thank Mr. Grayson Hough for providing an assess-
ment of purity using HPCL for all the compounds in this program.
References and notes
1. Schnell, J.; Dyson, H. J.; Wright, P. Annu. Rev. Biophys. Biomol. Struct. 2004, 33,
119.
Figure 5. Overlay of the S. aureus DHFR bound structures of ICL (white) and
compound 27 (orange).
2. Kompis, I.; Islam, K.; Then, R. Chem. Rev. 2005, 105, 593.
Table 3
Activity of leads on wild type and TMP-resistant enzymes and S. aureus strains
DHFR Ki (nM)
MIC (lg/mL)
S. aureus wild type
S. aureus F99Y
90
0.9
2.4
6
S. aureus wild type
S. aureus F99Y
TMP
ICL
27
1.24
1
32
2
1
0.081
0.026
0.025
0.125
0.125
0.5
28
8

5176
X. Li et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5171–5176
3. Veyssier, P.; Bryskier, A. Antimicrobial Agents In Bryskier, A., Ed.; ASM Press:
Washington, DC, 2005; pp 941–963.
4. Huovinen, P. Clin. Infect. Dis. 2001, 32, 1608–1614.
5. Sorbera, L. A.; Castaner, J.; Rabasseda, X. Drugs Future 2004, 29, 220.
6. Then, R. J. Chemother. 1993, 5, 361.
7. Clark, C.; Ednie, L.; Lin, G.; Smith, K.; Kosowska-Shick, K.; McGhee, P.; Dewasse,
B.; Beachel, L.; Caspers, P.; Gaucher, B.; Mert, G.; Shapiro, S.; Appelbaum, P.
Antimicrob. Agents Chemother. 2009, 53, 1353.
(1–4 nM active site concentrations) were incubated with test compound in the
presence of NADPH (100 M) for 10 min prior to initiation of reaction by
addition of dihydrofolate (H2F) (100 M). Assay buffer consists of 50 mM
sodium phosphate pH 7.0, 25 mM potassium chloride, 1 mM EDTA, 10
b-mercaptoethanol and 20 g/ml BSA. Km values for H2F were determined at
fixed NADPH (100 M) and varying H2F. Inhibition modalities were confirmed
to be competitive with respect to H2F using standard procedures. Inhibition
constants (Ki) were determined using the Morrison tight-binding equation to
account for ligand depletion. All analysis is carried out using Graphpad Prism
4.0.
l
l
lM
l
l
8. Bourne, C.; Barrow, E.; Bunce, R.; Bourne, P.; Berlin, K.; Barrow, W. Antimicrob.
Agents Chemother. 2010, 54, 3825.
9. Beierlein, J.; Frey, K.; Bolstad, D.; Pelphrey, P.; Joska, T.; Smith, A.; Priestley, N.;
Wright, D.; Anderson, A. J. Med. Chem. 2008, 51, 7532.
17. The Wild type gene for DHFR from S. aureus was cloned and expressed in E. coli
BL21-AI using the pET28 expression vector. The protein was purified with
affinity chromatography, exploiting a vector-derived C-terminal 6xHis tag.
Compounds were co-crystallized by hanging drop vapor diffusion using a
mother liquor containing 15% (w/v) PEG 8000, 0.05 M Magnesium acetate, 7%
(v/v) MPD, and 0.1 M ADA, pH 6.2. The resulting crystals were of space group
P6122 and typically diffracted to 1.7 Å or better on a rotating anode X-ray
source. Structures were solved by molecular replacement. Coordinates for the
co-crystal structures of compounds 11, 12, 19, 20, 23, 26 and 27 have been
deposited in the Protein Data Bank (http://www.pdb.org) under accession
codes 3SQY, 3SR5, 3SRQ, 3SRR, 3SRS, 3SRU, and 3SRW, respectively.
18. Chan, D.; Fu, H.; Forsch, R.; Queener, S.; Rosowsky, A. J. Med. Chem. 2005, 48,
4420.
10. Harris, N.; Smith, C.; Bowden, K. J. Med. Chem. 1990, 33, 434.
11. Harris, N.; Smith, C.; Bowden, K. Eur. J. Med. Chem. 1992, 27, 7.
12. Hynes, J.; Pathak, A.; Panos, C.; Okeke, C. J. Heterocycl. Chem. 1988, 25, 1173.
13. Gangjee, A.; Adair, O.; Pagley, M.; Queener, S. J. Med. Chem. 2008, 51, 6195.
14. Hynes, J.; Tomazic, A.; Kumar, V. J. Heterocycl. Chem. 1991, 28, 1981.
15. Strains used in this study: S. aureus – Smith strain; ATCC 13709; E. faecalis –
NCCLS QC recommended strain ATCC 29212; H. influenzae
– Ampicillin
resistant; NCCLS QC recommended strain ATCC 49247; E. coli – NCCLS QC
recommended strain ATCC 35218; S. aureus F99Y mutant strain made in house.
16. DHFR activity was determined spectrophotometrically by monitoring the
decrease in absorbance at 340 nm resulting from the oxidation of NADPH
coupled to reduction of substrate dihydrofolate at room temperature. Enzymes