Exploring 9-benzyl purines as inhibitors of glutamate racemase (MurI) in Gram-positive bacteria

Article abstract of DOI:10.1016/j.bmcl.2008.06.068

An early SAR study of a screening hit series has generated a series of 9-benzyl purines as inhibitors of bacterial glutamate racemase (MurI) with micromolar enzyme potency and improved physical properties. X-ray co-crystal EI structures were obtained.

Full text of DOI:10.1016/j.bmcl.2008.06.068

Bioorganic & Medicinal Chemistry Letters 18 (2008) 4368–4372
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Exploring 9-benzyl purines as inhibitors of glutamate racemase (MurI)
in Gram-positive bacteria
a
a
a
a
Bolin Geng a, , Gloria Breault , Janelle Comita-Prevoir , Randy Petrichko , Charles Eyermann ,
*
Tomas Lundqvist b, Peter Doig a, Elise Gorseth a, Brian Noonan a
a AstraZeneca R&D Boston, Infection Discovery, 35 Gatehouse Drive, Waltham, MA 02451, USA
b AstraZeneca R&D Mölndal, Global Structural Chemistry, Mölndal, SE-43183, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
An early SAR study of a screening hit series has generated a series of 9-benzyl purines as inhibitors of
bacterial glutamate racemase (MurI) with micromolar enzyme potency and improved physical proper-
ties. X-ray co-crystal EI structures were obtained.
Received 30 April 2008
Revised 18 June 2008
Accepted 19 June 2008
Available online 24 June 2008
Ó 2008 Elsevier Ltd. All rights reserved.
Keywords:
Glutamate racemase
MurI inhibitor
Benzyl purine
There has been a growing need for new antibacterials to fight
resistance, especially antibacterials with novel mechanism of
UDP-MurNAc-Ala
MurD
COOH
H
COOH
H
MurI
action (MOA) to combat ‘bad bugs’ such as MRSA and VRE.1,2
A
H2N
COOH
H2N
COOH
research program was initiated to develop a Gram-positive anti-
bacterial agent through inhibition of glutamate racemase (MurI),
an essential enzyme involved in bacterial peptidoglycan biosyn-
thesis (Scheme 1).3–5 The murI gene is conserved in all bacterial
species that synthesize peptidoglycan, and its essentiality has been
demonstrated in several bacterial species. In the literature, sub-
strate mimetic inhibitors have been designed.6 The rationale for
us to target this particular spectrum (Staphylococcus aureus, Entero-
coccus faecalis, Enterococcus faecium, etc.) was based on the high
degree of MurI sequence identity and biochemical similarity in
these selected Gram-positive pathogens. Phylogenic analyses dem-
onstrate that the MurI from Gram-positive bacteria cluster to a sin-
gle clade.7 High-throughput screening (HTS) of the AstraZeneca
compound collection identified several hit series with low micro-
molar potency against E. faecalis MurI. A 9-benzyl purine scaffold
was selected as one series to be advanced to the Lead Identification
phase. Here, we describe our Hit-to-Lead effort (Fig. 1).
L-Glutamate
D-Glutamate
UDP-MurNAc-Ala-
(D)Glu
peptidoglycan
Scheme 1. The role of glutamate racemase (MurI) in the peptidoglycan precursor
biosynthetic pathway.
cally vary substituents on position 2, 6, and 9 of the purine scaffold,
followed by purine core modification.
The synthesis of 9-benzyl-2-butylthio purines of type I, listed in
Table 1, as shown in Scheme 2, was readily achieved by benzyla-
O
N
NH
Based on the biological profile of the hits such as compound 1,
our objectives during the Lead Identification phase were to in-
crease MurI enzyme inhibitory potency, improve physical proper-
ties, especially solubility, and to expand the Gram-positive
spectrum. Our strategy to achieve these goals was to systemati-
62
N
7
5
N
N
1
HN
N
8
4
N
N
S
N
2
N
9
F
3
O
O2N
1
13
F
* Corresponding author. Tel.: +1 781 839 4297; fax: +1 781 839 4230.
E-mail address: bolin.geng@astrazeneca.com (B. Geng).
Figure 1. Representative HTS hit compound 1 and lead compound 13.
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.06.068

B. Geng et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4368–4372
4369
Table 1
To diversify substitution on positions 2 and 6, a route using aro-
matic displacement 6-Cl of 2,6-dichloro purine, followed by N-
alkylation, then second displacement of 2-Cl with R2-QH reagents
(Q@O, alcohols; Q@NH, amines) was employed as shown in
Scheme 3.9 The order of steps b and c can be switched to ease
the purification. Using this sequence the compounds of type II
listed in Tables 2 and 3 were obtained.
Four purine core modifications were performed as shown in
Schemes 4–7. Alkylation of commercially available purinyl-2-thiol
followed by benzylation gave the 6-des amino derivatives (type
III).10 The purin-6-one analogs (type IV) were easily accessed by
diazotation–hydrolysis through the Sandmeyer reaction.11 The
triazolopyrimidine (aza-purine) derivatives (type V) were prepared
R9 Benzyl variations (Type I, R6 = H)
n
R9
R2–Q
E.faa,b
M)
E.fma,b
M)
S.aua,b,c
M)
Solu.
M)
(l
(
l
(l
(l
1
n-Bu-S
9.4
17.7
>400
42
NO2
OMe
2
3
4
n-Bu-S
n-Bu–S
n-Bu-S
1.4
4.1
2.2
2.5
6.0
2.6
>400
>400
>400
25
50
25
NO2
F
F
F
Table 2
R2-Q chain variations (Type II, R6 = H)
n
R9
R2–Q
E.fa
E.fm
M)
S.au
M)
Solu.
M)
F
(l
M)
(
l
(
l
(
l
Cl
F
F
5
n-Bu-S
5. 4
6. 5
>400
>400
>400
25
5
n-Bu-S
5.4
6.5
>400
25
F
F
F
F
Me
F
Me
a
E.fa, Enterococcus faecalis; E.fm, Enterococcus faecium; S.au, Staphylococcus aur-
eus; solu., solubility (nephelometry).
6
7
n-Bu-O
9.8
27
50
25
b
Activity against MurI isozymes from several bacterial species was assessed by
IC50 determination for assay protocol see Ref. 7.
Me
F
c
Four hundred micromolar is the highest no-effect concentration tested. Same in
Tables 2–4 where 400 lM was used.
1-(C2F5)–n-Pr-O
4. 3
4.4
tion of a 2-butylthio-purine intermediate.8,10 This 2-butylthio-pur-
ine intermediate was obtained through the following sequence:
treatment of 4,6-diamino pyrimidine-2-thiol with sodium hydrox-
ide in methanol was followed by solvent evaporation and the
resultant solid was reacted with 1-bromobutane in DMF. Introduc-
tion of the nitroso group was achieved using sodium nitrite in
water. Subsequent catalytic hydrogenation gave the 4,5,6-triamine
intermediate, which was cyclized by refluxing the triamine in
formamide.
Me
F
8
9
n-Bu-NH
26.1
42.5
>400
>400
100
100
F
F
1-(AcNH)-Et-O
167
>400
F
Me
NH2
NH2
N
NH2
N
N
a
c
b
N
O
H2N
N
S
H2N
N
S
H2N
N
SH
NH2
N
NH2
N
NH2
N
H2N
H2N
e
N
N
N
d
N
H
N
S
N
S
N
S
R9
Type I
Scheme 2. General synthesis of 2-butylthio purines. Reagents and conditions: (a) NaOH, Methanol, n-BuBr, DMF; (b) NaNO2, water; (c) H2, PtO2, Ethanol; (d) HCONH2, reflux;
(e) BnBr or other R9 electrophiles, Cs2CO3, DMF.
R6
N
R6
N
R6
N
HN
N
HN
N
Cl
N
HN
N
a
c
b
N
N
N
N
N
N
N
R2
N
H
N
H
Q
Cl
Cl
Cl
R9
R9
Type II
Scheme 3. General synthetic route. Reagents and conditions: (a) ammonia or R6NH2, methanol; (b) BnBr or other R9 electrophiles, Cs2CO3, DMF; (c) butanol, NaOH, reflux.

4370
B. Geng et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4368–4372
Table 3
from modifications at this position. In subsequent rounds of opti-
R6 variations (Type II, R2–Q = n-Bu-O)
mization, the 2,6-difluoro substituents on benzyl moiety at R9
were kept constant while more polar solubilizing groups were
introduced at R2-Q on the 2-position of the molecule to reduce
lipophilicity of the molecule (Table 2). It was gratifying to discover
that the simple S to O switch of linking atom Q at 2-position (com-
pound 6) not only retained the enzyme potency, but also gained a
twofold increase in solubility. Further optimization along this
direction did not yield potency improvement. Polar groups were
not tolerated as exemplified in compound 9. Here the introduction
of the acetamide at the end of the R2 chain resulted in more than
15-fold potency loss in E.fa MurI inhibitory activity and the com-
pound became completely inactive against E.fm MurI. An O to NH
switch (compound 8) was attempted based on the expectation of
additional solubility increases, but again this resulted in potency
losses. Further SAR expansion using branched alkyls, cyclic alkyls,
benzyls, and aromatics generated only compounds with reduced
activity (data not shown). Oxidizing the linking atom sulfur to sul-
fone destroyed inhibitory activity. These data suggest that enzyme
potency for linkage Q follows the order: S ꢀ O > NH ꢁ SO2. We
concluded that linear 4–5 carbon chains provided ideal shape, size,
and lipophilicity at R2 position and hypothesized that, similarly to
R9, hydrophobic residues must surround the R2 region in the en-
zyme-binding pocket.
n
R9
R6
E.fa
M)
E.fm
S.au
M)
Solu.
M)
(
l
(
lM)
(l
(l
F
6
H
9.8
27
>400
>400
>400
50
F
Me
F
10
11
Methyl
Benzyl
5.5
13.9
100
F
F
>400
>400
3.1
F
Me
F
12
13
Pyridin-3-ylmethyl
2.0
2.7
>400
>400
18
F
F
Cl
F
2-(4-Morpholino)ethyl
5.8
5.5
200
The SAR of the 6-position was in sharp contrast to the 2- and 9-
position. Solubilizing groups were well tolerated. There is a striking
difference between benzylamino (compound 11) and pyridin-3-yl-
methyl amino (compound 12) analogs as shown in Table 3. The for-
mer compound was completely inactive and the latter was among
one of the most potent inhibitors of both E.fa and E.fm MurI. An
exciting example was the 2-(4-morpholino)ethyl amine group
(compound 13), as it not only retained all potency against both
E.fa and E.fm MurI enzymes, but also delivered a substantial boost
by reacting the product 4,5,6-triamine from step c in Scheme 212
with sodium nitrite. The 8-purinone analogs (type VI) were pre-
pared by hydrolysis of the corresponding 8-methoxy intermediate
according to literature.13
Data shown in Table 1 revealed the trend that ortho substitu-
ents on the benzyl ring tended to increase potency. However, vari-
ations of the substituent on the benzyl ring afforded, in general, no
significant potency improvements against E.fa and E.fm MurI en-
zyme and no inhibition of the S.au enzyme activity was observed.
Further benzyl replacement analogs containing small alkyl, bicyclic
heterocycles, and benzoyl substituents during an SAR expansion
study resulted in reduction of activity (data not shown). The fact
that the SAR at the 9-position is relatively unresponsive and that
only lipophilic groups were tolerated, meant that it would be diffi-
cult to obtain substantial improvements in physical properties
in solubility (equilibrium solubility 99 lM). This improved solubil-
ity allowed us to obtain an X-ray co-crystal structure using E.fa
MurI. This structure (1.65 Å) provided a working rationale for
understanding the potency discrepancy among MurI isozymes
studied, and laid the foundation for the project to potentially
advance to full Lead Optimization phase.
The inhibitor compound 1314 binds at each of two cryptic sym-
metrical allosteric hydrophobic pockets on the MurI homodimer
a
b
N
N
N
N
N
N
N
H
N
H
N
N
SH
N
S
N
S
R9
Type III
NH2
N
O
N
N
N
N
c
NH
N
N
O
S
O
R9
R9
Type IV
NH2
NH2
NH2
N
b
d
H2N
H2N
N
N
N
N
N
N
N
N
N
H
S
N
S
N
R9
Type V
NH2
N
NH2
N
NH2
N
H
N
e
f
N
N
N
Br
O
O
N
N
N
O
N
O
N
O
R9
R9
R9
Type VI
Schemes 4–7. Purine core modifications. Reagents and conditions: (a) n-BuBr, Et3N, DMF; (b) BnBr or other R9 electrophiles, Cs2CO3, DMF; (c) NaNO2, H2O; (d) NaNO2,
Ethanol; (e) NaOMe, Methanol; (f) HCl.

B. Geng et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4368–4372
4371
protein was replaced with Met153 in S.au. The Asn155 provides
weak polar interaction with purine core while Met153 certainly
does not.
Modification to the purine core gave varied results (Table 4).
The des-amino compound 14 showed an increase in enzyme po-
tency, while Type VI compound 8-Purinone 1713 was less active.
Compounds such as type IV 6-purinone 15 and type V triazolo-
pyrimidine (aza-purine) derivative 16 did not resulted in signif-
icant change in enzyme potency. Similar to the profiles observed
in the type I and type II scaffolds, all of these four core-modifi-
cation analogs still had the same spectrum issue. Finally, all of
the MurI inhibitors described in this Letter exhibit relatively
weak antibacterial activity (some compounds show MIC of 16–
Picture 1. (Left) Compound 13 binds at each of two allosteric sites of E. faecalis
MurI dimer. (Right) Detailed view at one allosteric binding pocket. C-terminal Trp
254 was displaced. Magenta shows the original position of Trp 254 in apo MurI. Asn
155 (below purine core) and Glu 153 (close to benzyl ring) were also shown.
64
pathogens.
From our perspective, these compounds represent viable lead
lg/ml, data not shown) against selected Gram-positive
Table 4
Core modifications (Type III–VI)
series for further optimization, particularly given the potential
for enzyme:inhibitor structure-enabled scaffold design17 to
increase intrinsic potency and continued improvement of physi-
cal properties. Issues to address in further optimization cycles
include antibacterial potency and desired antibacterial spec-
trum,16 as well as improvement in physical properties. In this
regard, the type II analogs with a handle at 6-position may
provide further optimizing flexibility, especially for physical
property improvement to reduce plasma protein binding (PPB)
and to enhance cell penetration.
n
Core structure
E.fa
M)
E.fm
M)
S.au
M)
Solu.
M)
(l
(l
(l
(l
N
N
N
N
S
14
1. 0
1. 8
>400
50
F
F
F
F
Cl
O
N
N
N
NH
In conclusion, we discovered a novel series of MurI inhibitors
against selected Gram-positive MurI isozymes, which have the po-
tential to be further optimized into an anti-Gram positive bacteria
agent.
15
16
17
11.5
4. 3
38.1
3.75
>400
100
S
S
O
F
Me
NH2
N
N
N
Acknowledgment
N
6. 2
>400
>400
50
N
The authors thank Microbiology Department at AstraZeneca
R&D Boston for MIC determination.
F
NH2
N
References and notes
H
N
N
O
F
1. Bush, K. Clin. Microbiol. Infect. 2004, 10, 10.
2. Payne, D. J.; Gwynnn, M. N.; Holmes, D. J.; Rosenberg, M. Methods Mol. Biol.
2004, 266, 231.
3. (a) Doublet, P.; van Heijenoort, J.; Mengin-Lecreulx, D. J. Bacteriol. 1992, 174,
5772; (b) Glavas, S.; Tanner, M. E. Biochemistry 2001, 40, 6199.
4. Doublet, P.; van Heijenoort, J.; Mengin-Lecreulx, D. Biochemistry 1994, 33, 5285.
5. Van Heijenoort, J. Nat. Prod. Rep. 2001, 18, 503.
97.8
200
N
F
6. For competitive inhibitors: (a) Glavas, S.; Tanner, M. E. Bioorg. Med. Chem. Lett
1997, 7, 2265; (b) Tanner, M. E.; Miao, S. Tetrahedron Lett. 1994, 35, 4073; (c) De
Dios, A.; Priesto, L.; Martin, J. A.; Rubio, A.; Ezquerra, J.; Tebbe, M.; Lopez de
Uralde, B.; Martin, J.; Sanchez, A.; LeTourneau, D.; McGee, J. E.; Boylan, C.; Parr,
T. R., Jr.; Smith, M. C. J. Med. Chem. 2002, 45, 4559; For peptide ligand, see: (d)
Kim, W. C.; Rhee, H. I.; Park, B. K.; Suk, K. H.; Cha, S. H. J. Biomol. Screen. 2000, 5,
435.
7. For investigation of structural and regulatory diversity of glutamate racemases,
see: Lundqvist, T.; Fisher, S. L.; Kern, G.; Folmer, R. H. A.; Xue, Y.; Newton, D. T.;
Keating, T. A.; Alm, R.; de Jonge, B. L. M. Nature 2007, 447, 817.
8. (a) Laxer, A.; Major, D. T.; Gottlieb, H. E.; Fischer, B. . J. Org. Chem. 2001, 66,
5463; (b) Temple,D. L. Jr.; US patent 4286093, 1981.; (c) Benedich, A.; Tinker, J.
F.; Brown, G. B. J. Am. Chem. Soc. 1948, 70, 3109.
9. (a) Komatsu, H.; Araki, T. Nucleosides, Nucleotides and Nucleic Acids 2005, 24,
1127; (b) Kurimoto, A.; Ogita, H. JP2005089334 A, Jpn. Kokai Tokkyo Koho,
2005.
10. Kikugawa, K.; Suehiro, H.; Aoki, A. Chem. Pharm. Bull. 1977, 25, 1811.
11. Kelley, J. L.; Schaeffer, H. J. J. Heterocycl. Chem. 1986, 23, 271.
12. Roblin, R. O., Jr.; Lampen, J. O.; English, J. P.; Cole, Q. P.; Vaughan, J. R., Jr. J. Am.
Chem. Soc. 1945, 67, 290.
(Picture 1). There seems to be a ‘hydrophobic collapse’,15 a term
often used in protein folding, between R2 and R9 to pre-organize
these two groups in the binding conformation (Picture 1, left and
right). Upon formation of the pocket, the R2 carbon chain displaced
the Trp254 indole ring out of original location and formed
p hydro-
phobic interactions. In subsequent attempts to explore this hydro-
phobic region, we observed that double bonds on the R2 chain
were tolerated, consistent with improved
p stacking interactions
with Trp254. As we expected, R9 benzyl interacts with hydropho-
bic side chains of Val14, Ile190, and Ile 250. One edge of the benzyl
group is solvent exposed and could be appropriately substituted to
improve interaction with Glu153. The region corresponding to R6
substituent pocket is close to the solvent front, which explained
its tolerance of polar groups and its intolerance of lipophilic sub-
stituents. Morpholino group is indeed positioned toward solvent
and is not forming significant interactions with the enzyme. A ten-
tative rationale for the striking E.fa and S.au isozyme potency pro-
file differences was proposed based on this co-crystal structure. By
comparing the binding pocket residues of apo enzyme of S.au and
above co-crystal structure of E.fa, it was observed that a key amino
acid at the binding site Asn155 present in the Enterococcus MurI
13. Kurimoto, A.; Ogino, T.; Ichii, S.; Isobe, Y.; Tobe, M.; Ogita, H.; Takaku, H.; Sajiki,
H.; Hirota, K.; Kawakami, H. Bioorg. Med. Chem. 2004, 12, 1091.
14. Beside compound 13, additional three X-ray co-crystal structures including
compound 12 were obtained and same binding mode was observed.
Interestingly, a small molecule X-ray crystal structure for compound 12
(data not shown), also shows a similar ‘hydrophobic collapse’ of R2 and R9 that
we observed in E–I co-crystal structures. All compounds were purified by
recrystallization, silica gel chromatograph or preparative Gilson HPLC

4372
B. Geng et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4368–4372
depending on initial purity of final step synthesis and characterized by LC-MS
and NMR techniques.Analyticals of compound 13: MS (ES+): 447 (MH+) for
22H28 F2 N6 O2 1H NMR (DMSO-d6) d: 1.00 (t, 3H); 1.48 (m, 2H); 1.72 (m, 2H);
2.60 (m, 4H); 3.64 (m, 6H); 4.27 (t, 2H); 5.43 (s, 2H); 7.21 (t, 2H); 7.54 (m, 1H);
7.64 (br s, 1H); 8.05 (s, 1H).
15. For hydrophobic collaps, see: Rich, D. H. Perspect. Med. Chem. 1993, 15.
16. For a recent review on challenges of target-based antibacterial discovery, see:
Payne, D. J.; Gwynn, M. N.; Holmes, D. J.; Pompliano, D. L. Nat. Rev. Drug Discov.
2007, 6, 29.
C
17. AutoDep EBI-36519 PDB ID code 2vvt for the coordinate entry.