Identification of the binding modes of N-phenylphthalimides inhibiting bacterial thymidylate synthase through X-ray crystallography screening

Article abstract of DOI:10.1021/jm2005018

To identify specific bacterial thymidylate synthase (TS) inhibitors, we exploited phenolphthalein (PTH), which inhibits both bacterial and human enzymes. The X-ray crystal structure of Lactobacillus casei TS (LcTS) that binds PTH showed multiple binding modes of the inhibitor, which prevented a classical structure-based drug design approach. To overcome this issue, we synthesized two phthalimidic libraries that were tested against TS enzymes and then we performed X-ray crystallographic screening of the active compounds. Compounds 6A, 8A, and 12A showed 40-fold higher affinity for bacterial TS than human TS. The X-ray crystallographic screening characterized the binding mode of six inhibitors in complexes with LcTS. Of these, 20A, 23A, and 24A showed a common unique binding mode, whereas 8A showed a different, unique binding mode. A comparative analysis of the LcTS X-ray complexes that were obtained with the pathogenic TS enabled the selection of compounds 8A and 23A as specific compounds and starting points to be exploited for the specific inhibition of pathogen enzymes.

Full text of DOI:10.1021/jm2005018

ARTICLE
pubs.acs.org/jmc
Identification of the Binding Modes of N-Phenylphthalimides
Inhibiting Bacterial Thymidylate Synthase through X-Ray
Crystallography Screening
Stefano Mangani,*,‡ Laura Cancian,† Rosalida Leone,‡,§ Cecilia Pozzi,‡ Sandra Lazzari,† Rosaria Luciani,†
Stefania Ferrari,*,† and M. Paola Costi*,†
†Dipartimento di Scienze Farmaceutiche, Universitꢀa degli Studi di Modena e Reggio Emilia, Via Campi 183, 41126 Modena, Italy
‡Dipartimento di Chimica, Universitꢀa degli Studi di Siena, Via Aldo Moro 2, 53100 Siena, Italy
S Supporting Information
b
ABSTRACT: To identify specific bacterial thymidylate synthase
(TS) inhibitors, we exploited phenolphthalein (PTH), which
inhibits both bacterial and human enzymes. The X-ray crystal
structure of Lactobacillus casei TS (LcTS) that binds PTH
showed multiple binding modes of the inhibitor, which pre-
vented a classical structure-based drug design approach. To
overcome this issue, we synthesized two phthalimidic libraries
that were tested against TS enzymes and then we performed X-ray crystallographic screening of the active compounds. Compounds
6A, 8A, and 12A showed 40-fold higher affinity for bacterial TS than human TS. The X-ray crystallographic screening characterized
the binding mode of six inhibitors in complexes with LcTS. Of these, 20A, 23A, and 24A showed a common unique binding mode,
whereas 8A showed a different, unique binding mode. A comparative analysis of the LcTS X-ray complexes that were obtained with
the pathogenic TS enabled the selection of compounds 8A and 23A as specific compounds and starting points to be exploited for the
specific inhibition of pathogen enzymes.
’ INTRODUCTION
synthesis provided specific compounds with moderate affinity for
bacterial TS. X-ray crystallographic screening of the compound
library yielded four crystallographic LcTSꢀinhibitor complexes
showing unique binding modes. Information gained through the
crystallographic studies on LcTS suggests structural guidelines
that would prospectively lead to more specific, higher-affinity
inhibitors toward TS from pathogenic organisms.
Thymidylate synthase (TS) (EC 2.1.1.45) catalyzes the re-
ductive methylation of 20-deoxyuridine-50-monophosphate (dUMP)
to form 20-deoxythymidine-50-monophosphate (dTMP), using
the cofactor N5,N10-methylene tetrahydrofolate (MTHF). Be-
cause TS represents the only means of synthesizing dTMP in
human cells and in many pathogens, it is typically a target in
designing chemotherapeutic agents such as anticancer drugs and,
more recently, antibacterial drugs.1ꢀ10
’ RESULTS AND DISCUSSION
In previous work, we obtained three X-ray crystal structures of
Lactobacillus casei TS (LcTS) in a complex with phenolphthalein
(PTH)3,4 or naphthalein derivatives (R156, PDB ID 1TSL;
MR20, PDB ID 1TSM) and one X-ray crystal structure of
Escherichia coli TS (EcTS) in complex with a naphthalein
derivative (GA9, PDB ID 2A9W)5,9 (Supporting Information,
Figure S1). Those results showed that when ligands of this class
of compound entered into the active site, they adopted multiple
binding modes; therefore, it was not possible to identify the key
interactions made by the compounds within the binding site.
Consequently, structure-based drug design is difficult to perform
and the results might be inconclusive.
To develop high-affinity inhibitors targeting bacterial TS, we
needed to identify specific inhibitors that presented a unique
binding mode. Therefore, starting from PTH, we designed and
synthesized a new class of phthalimidic compounds using a
ligand-based approach with two scaffolds. The first round of
Scaffold Identification and Validation. PTH has received
little attention as a TS inhibitor, with the exception of a limited
structureꢀactivity relationship study that investigated phenol-
substituted analogues.7,8,10 We further explored PTH through
a retrosynthetic approach11 to identify fragments that could be
further elaborated (Figure 1). Two fragments were identified:
phenol (1) and 2-benzofuran-1,3-dione (2). The phenol was not
considered useful because of its small size and its expected low
binding specificity. By contrast, the 2-benzofuran-1,3-dione was
more suitable for multiple derivatizations; therefore, we used this
core structure to examine the ChemFinder database.12 A number
of compounds were retrieved, and we identified eight potential
scaffolds, compounds 2ꢀ9 (Figure 2), for further synthetic
Received: April 25, 2011
Published: June 22, 2011
r
2011 American Chemical Society
5454
dx.doi.org/10.1021/jm2005018 J. Med. Chem. 2011, 54, 5454–5467
|

Journal of Medicinal Chemistry
ARTICLE
elaboration. In a preliminary chemical modification study, we
applied different synthetic strategies. Compounds 3 (4-methyl-2-
benzofuran-1,3-dione) and 4 (5-methyl-2-benzofuran-1,3-dione)
were the most promising molecules because, with an amine library,
we could selectively derivatize either the phthalic oxygen and/or
positions 4 or 5, with a preliminary bromination. With respect to the
bromination of the methyl group, compound 3 was the most active
and compound 4 did not react properly. Compound 3 was selected
as scaffold 1 for the subsequent production of the substituted
phthalimide libraries. A second scaffold was derived from com-
pound 3, where the anhydride functionality was aminomethylated
and the methyl in position 4 was modified to a bromo-methylene
group to obtain scaffold 2 (i.e., compound 11) (Scheme 1).
Design and Synthesis of Libraries A and B. A fragment
virtual library based on molecular diversity was designed with
ChemFinder software.12 Of the 23130 available amines, we
discarded molecules (fragments) with a molecular weight greater
than 200 Da, those that were structurally complex, and those
with cross-reacting groups. The remaining 303 molecules were
further filtered. Their molecular properties were calculated
(ACD Laboratories 6.0),13 and the final number was reduced
to 47 candidates according to Lipinsky’s rules,14 molecular diversity,
and commercial availability.
The fragments used to synthesize libraries A and B were
chosen to ensure that all the expected compounds possessed
satisfactory drug-like properties and high molecular diversity
(Supporting Information Charts S1 and S2, Table S1, Figure S2).
The compounds of library A possessed molecular masses ranging
from 238 to 451 Da, exhibited predicted water solubility between
3.6 ꢁ 10ꢀ7 and 1.34 mol/L, and had computed octanol/
water partition coefficients (clogD values) from ꢀ2.21 to 5.52
(pH 7.4). The compounds of library B had molecular masses
ranging from 243 to 370 Da, had predicted water solubilities from
2.6 ꢁ 10ꢀ6 to 0.14 mol/L, and displayed clogD values from
ꢀ3.80 to 4.46 (pH 7.4).
Library A derivatives were prepared with two different meth-
ods. In method 1, a small excess of the appropriate amine,
previously dissolved in ethanol, was added dropwise to a solution
of scaffold 1 in toluene, and the reaction mixture was refluxed for
15 h. The solvent was removed under reduced pressure, and the
residue was washed with ethanol to eliminate excess amine. In
method 2, the opportune amine and scaffold 1 were mixed at an
equimolar ratio in acetic acid, and then the reaction mixture was
refluxed for 5 h. After precipitation with water, the product was
isolated by filtration. Compounds 1A, 4A, 6A, 11Aꢀ12A, 15A,
18Aꢀ19A, and 21Aꢀ23A were synthesized using method 1, and
compounds 5A, 8A, 17A, 20A, and 24A were synthesized using
method 2. The synthesis of library B commenced with the
bromination of 2,4-dimethyl-1H-isoindole-1,3(2H)-dione (10)
with N-bromosuccinimide and benzoyl peroxide under UV light
to produce a good yield of 4-(bromomethyl)-2-methyl-1H-
isoindole-1,3(2H)-dione (11). Treatment of 11 with the suitable
amine in DMF at 75 °C for 15 h produced compounds 3B, 4B,
7B, 9B, 11B, 13B, 15B, 16B, 22B, and 23B. A total of 26
compounds were obtained with at least 95% purity.
Figure 1. Retrosynthetic analysis of phenolphthalein (PTH).
Figure 2. Compounds 2ꢀ9 were identified for further synthetic
elaboration through ChemFinder database exploration based on 2-ben-
zofuran-1,3-dione (2).
Biological Evaluation of Libraries A and B. The 26 synthe-
sized compounds were tested against TS enzyme from humans,
Scheme 1a
a Reagents and conditions: (a) method 1: the opportune amine, toluene, EtOH, reflux, 15 h; method 2: the opportune amine, acetic acid, reflux, 5 h.
(b) NH2CH3, toluene, EtOH, yield 47%; (c) NBS, benzoyl peroxide, CCl4, hν, yield 64%; (d) the opportune amine, DMF, 75 °C, 15 h.
5455
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
Table 1. Biological Activity of Libraries A and B Derivativesa
code
R156c
PTHd
3 (scaffold 1)
11 (scaffold 2)
1A
Ki hTSb
Ki EcTSb
Ki EfTSb
Ki LcTSb
30 ( 5
0.60 ( 0.12
0.70 ( 0.15
4.7 ( 0.9
ni
1.2 ( 0.2
1.4 ( 0.3
ni
49 ( 10
ni
ni
ni
ni
95 ( 22
20 ( 4
44 ( 6
14 ( 2
23 ( 5
22 ( 5
34 ( 7
3.0 ( 0.6
ni
ni
ni
ni
4A
ni
ni
ni
5A
ni
18 ( 2
111 ( 25
31 ( 6
ni
60 ( 12
6A
ni
7.0 ( 1.4
8A
ni
8.0 ( 1.6
11A
12A
15A
17A
18A
19A
20A
21A
22A
23A
24A
3B
206 ( 41
33 ( 7
81 ( 16
3.0 ( 0.6
ni
2.0 ( 0.35
ni
11 ( 2
60 ( 12
42 ( 8
ni
11 ( 2
ni
ni
61 ( 17
93 ( 19
69 ( 14
14 ( 4
25 ( 5
36 ( 7
6.0 ( 1.2
9.0 ( 1.6
27 ( 5
56 ( 11
41 ( 8
ni
30 ( 6
ni
ni
43 ( 9
36 ( 7
ni
ni
ni
26 ( 5
ni
33 ( 7
41 ( 8
ni
ni
ni
ni
ni
ni
ni
ni
ni
4B
ni
ni
109 ( 22
7B
48 ( 10
ni
183 ( 37
9B
ni
21 ( 4
ni
61 ( 12
42 ( 8
82 ( 16
ni
ni
ni
ni
ni
ni
ni
ni
11B
13B
15B
16B
22B
23B
33 ( 7
14 ( 3
51 ( 10
10 ( 2
ni
ni
ni
92 ( 18
14 ( 3
ni
ni
ni
38 ( 8
a Ki for the enzymes tested are reported (Ki hTS, Ki EcTS, Ki LcTS, Ki
LcTS). b ni = no detectable inhibition at 50 μM. c Already reported in refs
1,2, and 6. d Already reported in ref 5.
Figure 3. (A) Superposition of the two crystal structures of LcTS-
dUMPꢀ6A. (B) Superimposition of the two crystal structures of LcTS-
dUMPꢀ12A. Compound 12A was modeled in two possible bound
conformations in one crystal structure (PDB ID 3BYX; the two
conformations are shown with C in pink in the picture). (C) Crystal
structure of LcTS-dUMPꢀ8A. The protein is represented as a ribbon;
ligands are represented as sticks that are color-coded according to the
atom (N in blue, O in red, C in pink, yellow, or orange).
from the pathogenic bacteria Escherichia coli and Entercoccus
faecalis, and from the model bacteria Lactobacillus casei with a
rapid screening assay (Table 1 and Supporting Information
Figure S3). All active compounds showed competitive inhibition
with respect to the MTHF cofactor.
Compound 3 was tested against TS enzymes and showed
moderate affinity for EcTS (20% inhibition at 50 μM) but did not
show affinity for human TS (hTS) (no activity at 50 μM).
Compounds 11A, 17A, 19A, and 20A that inhibited hTS
were deemed less likely candidates than those such as 5A, 6A,
8A, 15A, and 21A that had no activity against hTS but had Ki
values below 30 μM for EcTS or Entercoccus faecalis TS (EfTS).
Compound 12A had activity against hTS but even greater
activity against EcTS and EfTS, so the compound displayed
selectivity.
Within library B, only one compound (22B) inhibited EcTS
with a Ki value lower than 30 μM; none of the compounds show
Ki values lower than 30 μM against EfTS. Only two compounds,
7B and 11B, showed some inhibitory activity against hTS with Ki
values of 48 and 21 μM, respectively. All the other candidates in
library B showed no detectable inhibitory activity against hTS at
50 μM concentration.
All compounds were also tested against LcTS because it is
often used in TS crystallographic studies, even though it is from a
nonpathogenic bacteria. All but four of the compounds (15A,
17A, 9B, and 22B) were active against LcTS, with Ki values
ranging from 3 to 93 μM.
X-Ray Crystallographic Screening. We performed X-ray
crystallographic screening to gain the necessary structural in-
formation on the binding modes of these compounds. Compared
to library B, library A had a higher number of and more active and
specific compounds that were easier to synthesize. Nevertheless,
our attempts to crystallize library A compounds in complex with
TS from the different pathogens (EcTS and EfTS) failed; crystal
structures were obtained only with a TS from a nonpathogenic
bacteria, LcTS. Despite not being able to rule out that the same
compounds could show different binding modes in other bacterial
5456
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
5457
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
Figure 4. Binding modes of compounds as observed in the crystal structures. (A) 6A bound in crystal structure 3C0A or (B) in crystal structure 3C06.
(C and D) Compound 12A is shown in its (C) first binding mode and (D) second binding mode as modeled in crystal structure 3BYX. (E) Compound
12A is shown in its binding mode in the 3BZ0 crystal structure; (F) compound 8A bound in the 3BNZ crystal structure.
enzymes, we could use proven binding modes as a guide to design
better derivatives taking into account the structural differences
among the different enzymes.
in catalysis included the nucleophilic Cys198, Ser219, His259,
and Asn229, which could interact with the dUMP ribose and
uridine moieties, as well as Ile81, Trp82, Trp85, Leu224, Phe228,
and Val314, which defined a large hydrophobic cavity capable of
hosting the aromatic moiety of MTHF. After the determination
and refinement of the crystal structure (see Experimental Section),
the Fo ꢀ Fc difference Fourier maps (ΔF) were calculated with
phases from the refined LcTS model; these ΔF maps were
inspected to identify the bound inhibitor molecule. The ΔF
maps showed a large positive electron density (between 2σ
and 7σ) in the active sites due to the presence of dUMP and
the different ligands. Although the ΔF maps showed a low or
incomplete electron density for the inhibitor molecule, we
were able to confidently position the main features of the
ligand molecules. The dUMP substrate occupied the usual
dUMP site, which extended from the phosphate binding site
(Arg23, Arg218, and Ser219 from one subunit and Arg1780
and Arg1790 from the symmetry-related subunit), located at
the interface between the two subunits, to the pyrimidine
binding site located deeper in the cavity between the backbone
of Asp221 and the side chain of Asn229.16
Eight crystals of LcTS bound to six different ligands (6A, 8A,
12A, 20A, 23A, and 24A) were successfully obtained and
analyzed (Table 2). The LcTS structure consisted of a homo-
dimer with the two subunits related by a 2-fold crystallographic
axis of the hexagonal P6122 space group; all LcTS-complex
crystals showed this symmetry. Each LcTS subunit had two
domains. The larger domain (residues 1ꢀ69 and 141ꢀ316) had
an R/β structure with six R-helices, one β-hairpin, a two-stranded
β-sheet, and a five-stranded β-sheet. The second domain, termed
the small domain (SD, residues 70ꢀ139), consisted of four
R-helices and a long interconnecting loop.15 This region con-
sistently showed low electron density as a result of structural
flexibility, especially between residues 95 and 120, which could
not be modeled in some of the structures. The active site cavity
was embedded in the large domain, but one of its walls contained
two Arg residues (Arg1780 and Arg1790) from the symmetry-
related subunit; these Arg residues contributed to the binding site
of the dUMP phosphate moiety. The active site residues involved
5458
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
Figure 5. (A) Superposition of the crystal structures for LcTS-dUMPꢀ20A, LcTS-dUMPꢀ23A, and LcTS-dUMPꢀ24A. (B) Detailed interactions of
the phenyl moiety of ligand 20A with LcTS. (C) Detailed interaction of the phenyl moiety of ligand 23A with LcTS. (D) Detailed interaction of the
phenyl moiety of ligand 24A with LcTS. All structures are color-coordinated, with oxygen in red, nitrogen in blue, sulfur in yellow, and phosphate in cyan.
Carbon atoms are sky blue, purple, and green in LcTS-dUMPꢀ20A, LcTS-dUMPꢀ23A, and LcTS-dUMPꢀ24A, respectively. Hydrogen bonds are
indicated with black lines.
For two ligands (6A and 12A; parts A and B of Figure 3,
respectively), the X-ray crystallographic screen was used to
identify different binding modes. For each ligand, two structures
were obtained from independent crystallization and diffraction
experiments that showed the molecule bound in distinct orienta-
tions to the active site of LcTS. In one crystal structure (PDB ID
C03A), the inhibitor 6A formed stacking interactions with
Phe228, Trp82, and Trp85, a hydrogen bond with Glu60, and
contact interactions with Ile81, Glu84, Leu195, Leu224, and Gly225
(Figure 4A). In the other crystal structure (PDB ID 3C06), 6A
formed stacking interactions with Trp82, Trp85, and the pyrimidine
ring of dUMP as well as contact interactions with Ile81, Leu224, and
Gly225 (Figure 4B). A comparison of the two structures showed
that, in 3C0A, the phthalimidic ring of 6A was rotated by ∼90° with
respect to its position in 3C06 (Figure 3A). The observed binding
modes of 6A in the two complexes displayed three important features:
(i) the 6A pyridine ring was located close to Trp82 and Trp85,
(ii) the chlorine (Cl) atom interacted with the indole Nε1 atom
of Trp85, and (iii) the phthalimidic moiety of 6A was involved in
the aromatic stacking interactions with dUMP or with Phe228.
In the case of compound 12A, one structure (PDB ID 3BZ0),
even though it was viewed at a relatively low resolution (2.70 Å),
allowed us to model the ligand at full occupancy. By contrast, the
other structure (PDB ID 3BYX; resolution 2.40 Å) had a weaker
electron density; thus, the ligand was modeled in two possible
bound conformations, and partial occupancy was assigned to
each conformation (Figure 4C, 4D). In the first binding mode
observed for molecule 12A (PDB ID 3BYX, conformation A), no
hydrogen bond was formed, but stacking interactions were
observed with Phe228 and Trp82 and contacts were made with
Ile81, Trp85, Leu224, Gly225, Asn229, and dUMP (Figure 4C).
In the second binding mode observed for molecule 12A (PDB ID
3BYX, conformation B), a hydrogen bond was formed with
Trp85, stacking interactions were made with Trp85 and dUMP,
and contact interactions were made with Arg23, Ala194, Leu195,
Asp221, Gly225, Val314, Ala315, and Arg1790 (Figure 4D). In
the last binding mode observed for molecule 12A (PDB ID
3BZ0), a hydrogen bond was formed with Trp85, stacking
interactions were made with Trp82, Trp85, Phe228, and dUMP,
and contact interactions were made with Ile81, Glu84, Leu195,
and Leu224 (Figure 4E).
Compounds 6A and 12A show the best outcome from this
study based on having the best affinity and specificity profiles;
however, their multiple binding modes prevent the possibility of
applying a structure-based approach for further structural
development.
In the case of the LcTS-dUMPꢀ8A complex (PDB ID
3BNZ), the ΔF map showed positive electron density in the
active site that was clearly attributable to dUMP. Less density was
present in the hydrophobic pocket, where the inhibitor was
5459
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
Figure 6. (A) Details of the LcTSꢀ8A complex. Colored in magenta is the Pro53-Lys92 sequence. (B) Structure comparison of LcTSꢀ20A
(in magenta) and LcTSꢀ23A (in cyan) complexes. For clarity purposes, only Arg23 residue and Pro21-Gly27 and Pro105-Glu113 trace sequences of
LcTSꢀ23A complex are shown to highlight the conformational differences between these two complexes. (C) Protein sequence alignment of EfTS, hTS
and LcTS; written in white on a black background are identical residues among the three sequences; written in white on a gray background are identical
residues between two out of three sequences. The magenta rectangle highlights the sequence Pro53-Lys92 (LcTS sequences number) that is more
conserved between the two bacterial TS with respect to hTS; the green circles indicate the residues interacting with compound 8A.
expected to bind. Compound 8A interacted with LcTS by
forming stacking interactions with Phe228 and dUMP and by
making contacts with Glu60, Ile81, Glu84, Leu224, Gly225, and
Asn229 (Figure 3C). Thus compound 8A with its single binding
mode represents a good initial candidate for EfTS inhibition.
However, the EfTS crystallographic structure is not available in
the Protein Data Bank for a three-dimensional comparison of the
different TS isoforms. Sequence alignment (Figure 6C) showed
that all the residues interacting with compound 8A are conserved
between LcTS and EfTS. Comparison with hTS showed that of
these residues, only Glu84 is not conserved (Ala in hTS); the
whole sequence (Pro53-Glu88) surrounding the interacting
residues forming the side wall of the 8A binding site is more
conserved between LcTS and EfTS on one side than between
LcTS and hTS on the other side (Figure 6A, 6C). These
differences could account for the observed specificity.
Three ligands (20A, 23A, and 24A) showed a common
binding pattern, where the phthalimidic core of these molecules
formed (i) stacking interactions with Trp82, Phe228, and the
pyrimidine ring of the substrate, dUMP, (ii) a hydrogen bond
with Asp221 through a conserved water molecule, and (iii)
contacts with Glu60, Ile81, Leu224, and Gly225 (Figure 5A).
In all cases, because of the substituent linked at the phthalimidic
nitrogen, the phenyl ring on the substituent interacted with the
uridine ring of dUMP, with Trp85 (stacking interactions) and
with Leu195 (contact). Small differences were observed in the
hydrogen-bonding network due to the different functional groups
that were present on the phenyl ring. In the case of 20A, the
5460
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
Chart 1. Library A, Compounds 1Aꢀ24A
hydroxyl group established a hydrogen bond with the ribose ring
of dUMP, and the carboxyl group formed hydrogen bonds
mediated by 1 or 2 water molecules with Tyr146, Pro196, and
Arg218 (Figure 5B). In the case of compound 23A, the acetyl
group interacted with the ribose moiety of dUMP through a
water molecule bridge, with the phosphate moiety of dUMP and
with Arg23 (Figure 5C). In compound 24A, the amidine group
formed only a hydrogen bond with the ribose moiety of dUMP
(Figure 5D).
Initially, we obtained compounds that showed unique binding
modes to LcTS, such as 8A, 20A, 23A, and 24A. This is the first
time that inhibitors structurally unrelated to the folate cofactor
crystallized with the TS protein in a unique binding mode.
However, it was necessary to understand how to use the
information obtained on a nonpathogenic TS, such as LcTS, to
design effective antibacterial agents against EcTS and EfTS
targets. With this aim, we critically analyzed each crystallographic
complex and then selected the most suitable complexes for
further studies.
Compounds 20A, 23A, and 24A showed different specificity
profiles: 20A is active toward LcTS, EcTS, and hTS, and thus, it is
not specific; 23A is active toward LcTS and EcTS, and thus, it is
moderately specific; 24A is active only toward LcTS, and thus, it
is of low interest. Comparison of LcTSꢀ20A and LcTSꢀ23A
complexes shows that among the residues interacting with the
two ligands, there is a conformational change in residue Arg23.
This residue moves toward the ligand in the case of 23A (Figure 6B).
Analyzing the whole protein trace reveals that conformational
changes between the two structures involve the whole sequence
(Pro21-Gly27) as well as the adjacent small domain sequence
from Gly105 to Glu113 (Figure 6B). Previous computational
studies17,18 described how these two regions move in a coordi-
nated manner during the flexibile motion of the protein high-
lighting major differences between bacterial TS and hTS. We
suggest that the differences in the dynamics of LcTS vs hTS could
be fundamental to the specific profile of compound 23A.
Structure-BasedꢀActivity Relationships. All eight struc-
tures showed that all the ligands bound in the active site of the
enzyme where the MTHF cofactor and the classical folate
analogue inhibitors usually bind. This accounted for the compe-
titive inhibition observed in the enzyme inhibition assays, in
which competition for binding with the folate substrate was
observed. Although all these compounds (6A, 8A, 12A, 20A,
23A, and 24A) could bind to the same site, they showed different
binding modes depending on the substituent present on the
phthalimidic nitrogen; the phthalimidic core contributed to the
interaction with the enzyme, but this core alone was unable to
recognize a specific binding site in the enzyme active site cavity
(Chart 1).
Three pairs of compounds from the two libraries have the
same or similar substituents R/R0 (R on position 2 of scaffold 1
and R0 on position 4 of scaffold 2): 4Aꢀ9B, 8Aꢀ15B, and
23Aꢀ4B. The compounds within the first pair show different
specificity profiles. Compound 4A is active only toward LcTS
(Ki = 44 μM), while 9B is inactive against LcTS and active only
toward EcTS (Ki = 61 μM); no structural data was obtained with
these compounds. In the pair 8Aꢀ15B, compound 8A is active
toward all the bacterial enzymes, while compound 15B is active
5461
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
Chart 2. Library B, Compounds 1Bꢀ23B
only against LcTS. The analysis of the crystallographic complex
of LcTS and 8A (Ki of 22 μM) suggested that the binding site of
8A can be conserved also for 15B (Ki of 51 μM); 15B can bind
with slight motion of the enzyme to accommodate the larger R0
group in position 4 of scaffold 2 (Chart 2). Considering the case
in which the main stacking interaction with Phe228 and dUMP
are conserved, the 15B ligand could, by flipping around the
pthalimidic plane with respect to 8A, cause its larger substituent
in position 4 to interact with Trp82 and Phe64. Both compounds
23Aꢀ4B, in which the two substituents are quite similar, show
that they are both inactive toward hTS and active toward LcTS,
even if 4B is approximately 10 times less active than 23A (6 and
56 μM, respectively); however, they differ in the activity toward
the other bacterial enzymes: 23A is active toward EcTS, while 4B
is active toward EfTS. The analysis of the crystallographic
complex of LcTS and 23A suggested that the binding site of
23A can be conserved also in the case of 4B, with a slight rotation
of the ligand to accommodate the bigger R0 group in position 4 of
scaffold 2. In this way it is possible that the main stacking
interaction with Trp82 and Phe228 are retained and that this
larger R0 substituent could face the inner pocket of the active site
(particularly toward Phe64 and Tyr146) or toward the outer pocket.
Derivatization of scaffold 1 gave active compounds in library A,
which could be further crystallized with LcTS. The best com-
pounds from this new series had Ki values in the same range as
that of PTH toward bacterial TS but they showed lower or no
activity against human TS, therefore they are selective for the
pathogenic enzymes. X-ray crystallographic screening was ap-
plied and provided eight structures of LcTS complex with six
compounds out of 26 compounds analyzed. Only compounds
20A, 23A, and 24A, revealed a unique binding mode and a
common binding pattern as shown by the crystallographic
complexes, while compound 8A showed a different but unique
mode of binding. Even if our study was limited by having crystal
structures from a non pathogenic TS, i.e. LcTS, it demonstrates
that the suggested approach can successfully characterize the
binding mode of the library compounds and identify unique
binding modes even in cases of such a large and hydrophobic
active site as that of TS.
In conclusion, our library design and subsequent crystallo-
graphic screening represent a strategic tool for providing leads
with unique binding modes which are thus suitable starting
points for further medicinal chemistry elaboration. In our case,
further studies will take advantage of information derived from
crystallographic analysis on compounds 8A and 23A to address
the structure-based optimization of the phthalimidic core to
develop other derivatives that retain their binding profile but
have improved activity and specificity.
’ CONCLUSION
In searching for new specific inhibitors of bacterial TS
enzymes, we combined a ligand-based design study and a X-ray
crystallographic screening of the synthesized libraries to char-
acterize the binding modes of the active compounds.
’ EXPERIMENTAL SECTION
From PTH, a lead inhibitor unsuitable for structure-based
drug design, we derived two potential scaffolds that we then used
to design and synthesize candidate compound libraries.
Synthesis and Characterization of Libraries A and B.
Reactions were performed using B€uchi’s Syncore parallel synthesizer,
which allowed the simultaneous production of 24 reactions. Reaction
5462
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
progresses were monitored by TLC on precoated silica gel 60 F254 plates
(Fluka), and visualization was accomplished with UV light (254 nm).
The purity of all synthesized compounds was determined by elemental
analyses performed on a Perkin-Elmer 240C instrument, and the results
were within (0.4% of the theoretical values. All the compounds tested
through a rapid screening assay show >95% purity. Yields of these
reactions referred to purified products, and not optimized, were:
compounds A method 1 in the range of 20ꢀ50%, method 2 in the
range 35ꢀ55% (except 24A and 8A, whose yields were 13% and 73%
respectively), compounds B method 3 in the range 20ꢀ45%. Melting
points were determined on a Stuart SMP3 capillary melting point
apparatus and are uncorrected. All compounds were characterized by
1H NMR on a Bruker AC200 and Bruker MX400 WB instruments
(CIGS, University of Modena e Reggio Emilia). Two dimensional NMR
techniques (heteronuclear single quantum coherence and heteronuclear
multiple bond correlation) were used to aid the assignment of signals in
H-8). Anal. Calcd for C17H11NO6: C, 62.77; H, 3.41; N, 4.31. Found: C,
62.90; H, 3.40; N, 4.29.
4-Methyl-2-(2-thiophen-2-ylethyl)-1H-isoindole-1,3(2H)-dione (4A).
Compound 4A was synthesized in a similar manner to the method 1 from
2-thiophen-2-ethylamine. The reaction was heated at 90 °C for 5 h. After
the evaporation of the solvent, the crude residue was crystallized from
methanol/H2O, affording the white solid product; mp 58ꢀ60 °C. MS(M+
H+) m/z 272. 1H NMR (CDCl3) δ 2.70 (s, 3H, H-4), 3.23 (t, 2H, J = 5.8
Hz, H-6), 3.96 (t, 2H, J= 5.8 Hz, H-5), 6.91 (m, 2H, H-7 and H-8), 7.15 (m,
1H, H-9), 7.46 (d, 1H, J = 7.8 Hz, H-3), 7.58 (t, 1H, J = 7.8 Hz, H-2), 7.69
(d, 1H, J=7.8Hz, H-1). Anal. CalcdforC15H13NO2S: C, 66.40; H, 4.83; N,
5.16. Found: C, 66.28; H, 4.92; N, 4.99.
2-(2-Chloropyridin-4-yl)-4-methyl-1H-isoindole-1,3(2H)-dione (6A).
Compound 6A was synthesized in a similar manner to the method 1 from
4-amino-2-chloropyridine. The reaction was heated at 80 °C for 16 h. After
the evaporation of the solvent, the residue was treated with diethyl ether and
cooled. The precipitate, the unreacted anhydride, was filtered off while the
filtrate was concentrated and the residue was washed with diethyl ether,
affording the beige solid product; mp 155 °C. MS (M + H+) m/z 273. 1H
NMR (CDCl3) δ2.78 (s, 3H, H-4), 7.61 (dd, 1H, J= 5.4, 1.8 Hz, H-7), 7.67
(d, 1H, J = 6.6 Hz, H-3), 7.70 (t, 1H, J = 6.6 Hz, H-2), 7.75 (d, 1H, J = 1.8
Hz, H-5), 7.83 (d, 1H, J = 6.6 Hz, H-1), 8.51 (d, 1H, J = 5.4 Hz, H-6). Anal.
Calcd for C14H9ClN2O2: C, 61.66; H, 3.33; N 10.27. Found: C, 61.27; H,
3.08; N, 10.14.
1
1H spectra. Compound numbering for H NMR peaks assignment is
reported in Supporting Information (Table S2). Spectra were recorded
in DMSO-d6 or CDCl3. Chemicals shifts are reported in ppm from
tetramethylsilane as an internal standard. When peak multiplicities are
given, the following abbreviations are used: s, singlet; d, doublet; t,
triplet; q, quartet; m, multiplet; br, broadened signal. Mass spectra were
determined on a Finnigan MAT SSQ 710A mass spectrometer (CIGS,
University of Modena and Reggio Emilia).
Of the 47 expected compounds of libraries A and B (24 and 23
molecules respectively), only 26 compounds were successfully obtained
(16 and 10 molecules, respectively). The other selected fragments did
not react well, affording complex mixture of undesired products.
2,4-Dimethyl-1H-isoindole-1,3(2H)-dione (10). A solution of
methylamine (35%) in H2O (1.9 mL, 22 mmol) dissolved in ethanol
(20 mL) was added dropwise to a solution of 4-methyl-2-benzofuran-
1,3-dione (3) (3 g, 18 mmol) in toluene (25 mL), and the mixture was
heated at 80 °C for 4 h. After evaporation of the solvent, the oil residue
was extracted four times with diethyl ether. Yield 1.52 g, 47%. 1H NMR
(CDCl3) δ 2.71 (s, 3H, H-4), 3.16 (s, 3H, H-5), 7.45 (d, 1H, J = 7.6 Hz,
H-3), 7.56 (t, 1H, J = 7.6 Hz, H-2), 7.59 (d, 1H, J = 7.6 Hz, H-1). Anal.
Calcd for C10H9NO2: C, 68.56; H, 5.18; N, 8.00. Found: C, 68.94; H,
5.21; N 7.77.
2-[4-(1H-Imidazol-1-yl)phenyl]-4-methyl-1H-isoindole-1,3(2H)-dione
(11A). Compound 11A was synthesized in a similar manner to the
method 1 from 4-(1H-imidazol-1-yl)aniline, and it was isolated as a
1
beige solid. mp 279ꢀ280 °C. H NMR (DMSO-d6) δ 2.68 (s, 3H,
H-4), 7.13 (s, 1H, H-10), 7.58 (d, 2H, J = 9.0 Hz, H-6 and H-7), 7.70
(t, 1H, J = 6.3 Hz, H-2), 7.78 (m, 2H, H-1 and H-3), 7.79 (s, 1H, H-9),
7.80 (d, 2H, J = 9.0 Hz, H-5 and H-8), 8.30 (s, 1H, H-11). Anal. Calcd for
C18H13N3O2: C, 71.28; H, 4.32; N, 13.85. Found: C, 71.15; H, 4.31;
N, 13.82.
2-(4-Hydroxybiphenyl-3-yl)-4-methyl-1H-isoindole-1,3(2H)-dione
(12A). Compound 12A was synthesized in a similar manner to the
method 1 from 3-aminobiphenyl-4-ol dissolved in a solution ethanol/
methanol 1:1. The reaction was heated at 80 °C for 2 h. The brown solid
product was washed with diethyl ether; mp 160ꢀ165 °C. MS (M + H+)
m/z 330. 1H NMR (CDCl3) δ 2.78 (s, 3H, H-4), 5.92 (br, 1H, H-13),
7.18 (d, 1H, J = 8.2 Hz, H-5), 7.27 (s, 1H, H-7), 7.32ꢀ7.46 (m, 2H, H-6
and H-10), 7.55ꢀ7.59 (m, 4H, H-8, H-9, H-11 and H-12), 7.55 (d, 1H,
J = 7.4 Hz, H-3), 7.68 (t, 1H, J = 7.4 Hz, H-2), 7.82 (d, 1H, J = 7.4 Hz,
H-1). Anal. Calcd for C21H15NO3: C, 76.58; H, 4.59; N, 4.25. Found: C,
76.60; H, 4.50; N, 4.00.
4-Methyl-2-(1,3-thiazol-2-yl)-1H-isoindole-1,3(2H)-dione (15A).
Compound 15A was synthesized in a similar manner to the method 1
from 2-aminothiazol, and it was isolated as a beige solid; mp 96ꢀ99 °C.
1H NMR (CDCl3) δ 2.78 (s, 3H, H-4), 7.32 (s, 1H, H-6), 7.68 (d, 1H,
J = 7.4 Hz, H-3), 7.69 (t, 1H, J = 7.4 Hz, H-2), 7.81 (s, 1H, H-5), 7.83
(d, 1H, J = 7.4 Hz, H-1). Anal. Calcd for C12H8N2O2S: C, 59.00; H, 3.30;
N, 11.47. Found: C, 58.83; H, 3.31; N, 11.49.
4-(Bromomethyl)-2-methyl-1H-isoindole-1,3(2H)-dione
(11). N-Bromosuccinimide (1.52 g, 8.6 mmol) was added to a solution
of 10 (1.5 g, 8.6 mmol) in CCl4 (20 mL). The mixture was refluxed and
irradiated with an UV lamp, and then the catalyst benzoyl peroxide
(0.208 g, 0.86 mmol) in 6 mL of CCl4 was added. The mixture was
cooled with ice, and the resulting precipitate was removed by filtration.
The filtrate was concentrated in vacuum, and the oil residue was
crystallized several times with CH2Cl2/n-hexane, affording the product
as a white solid. Yield 1.39 g, 64%; mp 138ꢀ140 °C. MS (M + H+) m/z
1
254. H NMR (CDCl3) δ 3.19 (s, 3H, H-5), 4.99 (s, 2H, H-4), 7.70
(d, 1H, J = 6.4 Hz, H-1), 7.75 (t, 1H, J = 6.4 Hz, H-2), 7.80 (d, 1H, J =
6.4 Hz, H-3). Anal. Calcd for C10H8BrNO2: C, 47.27; H, 3.17; N, 5.51.
Found: C, 46.94; H, 3.20; N, 5.46.
2-(3-Hydroxyphenyl)-4-methyl-1H-isoindole-1,3(2H)-dione (18A).
Compound 18A was synthesized in a similar manner to the method 1
from 3-aminophenol, and it was isolated as a beige solid; mp 195ꢀ
197 °C. 1H NMR (DMSO-d6) δ 2.65 (s, 3H, H-4), 6.82 (m, 3H, H-5,
H-6 and H8), 7.29 (m, 1H, H-7), 7.68 (t, 1H, J = 7.2 Hz, H-2), 7.74
(m, 2H, H-1 and H-3), 9.68 (br, 1H, H-9). Anal. Calcd for C15H11NO3:
C, 71.14; H, 4.38; N, 5.53. Found: C, 70.90; H, 4.38; N, 5.51.
3-(4-Methyl-1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)benzoic acid
(19A). Compound 19A was synthesized in a similar manner to the
method 1 from 3-aminobenzoic acid. The product was isolated as an oil.
1H NMR (DMSO-d6) δ 2.67 (s, 3H, H-4), 7.65 (t, 1H, J = 7.4 Hz, H-2),
7.76 (m, 2H, H-1 and H-3), 7.77 (m, 2H, H-6 and H-7), 7.99 (d, 1H,
J = 7.5 Hz, H-8), 8.35 (s, 1H, H-5), 13.00 (br, 1H, H-9). Anal. Calcd for
General Synthetic Method 1 for Library A Compounds 1A,
4A, 6A, 11ꢀ12A, 15A, 18ꢀ19A, 21ꢀ23A. The appropriate
ammine (0.37 mmol) in ethanol (2 mL) was added dropwise to a
solution of 4-methyl-2-benzofuran-1,3-dione (3) (0.05 g, 0.31 mmol) in
toluene (2 mL), and the reaction mixture was refluxed for 15 h. The
solvent was removed under reduced pressure, and the residue was
washed with ethanol (3 ꢁ 2 mL).
5-(4-Methyl-1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)benzene-1,3-
dicarboxylic Acid (1A). Compound 1A was synthesized in a similar
manner to the method 1 from 5-amino-isophthalic acid, and it was
isolated as a beige solid; mp 250ꢀ254 °C. 1H NMR (DMSO-d6) δ 2.68
(s, 3H, H-4), 7.70 (m, 1H, H-2), 7.78 (m, 2H, H-1 and H-3), 8.27 (t, 1H,
J = 1.6 Hz, H-6), 8.49 (d, 2H, J = 1.6 Hz, H-5 and H-7), 12.95 (br, 2H,
5463
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
C16H11NO4: C, 68.32; H, 3.94; N, 4.98. Found: C, 68.45; H, 3.93;
N, 4.96.
manner to the method 2 from 4-amino-2-hydroxybenzoic acid, and it
was isolated as a beige solid; mp 289ꢀ291 °C. MS (M + H+) m/z 298.
1H NMR (DMSO-d6) δ 2.66 (s, 3H, H-4), 7.37 (d, 1H, J = 8.2 Hz, H-6),
7.48 (dd, 1H, J = 8.2, 1.6 Hz, H-7), 7.55 (d, 1H, J = 1.6 Hz, H-5), 7.69 (t,
1H, J = 7.3 Hz, H-2), 7.77 (m, 2H, H-1 and H-3), 10.23 (br, 1H, H-8),
13.0 (br, 1H, H-9). Anal. Calcd for C16H11NO5: C, 64.65; H, 3.73; N,
4.71. Found: C, 64.59; H, 3.99; N, 4.87.
2-(1H-Indol-4-yl)-4-methyl-1H-isoindole-1,3(2H)-dione (21A). Com-
pound 21A was synthesized in a similar manner to the method 1 from
4-aminoindole. The product was isolated as a black solid; mp 266ꢀ270 °C.
1H NMR (DMSO-d6) δ 2.67 (s, 3H, H-4), 6.25 (d, 1H, J = 5.4 Hz, H-8),
6.99 (d, 1H, J = 7.2 Hz, H-7), 7.19 (t, 1H, J = 7.2 Hz, H-6), 7.36 (d, 1H, J =
5.4 Hz, H-9), 7.50 (d, 1H, J = 7.2 Hz, H-5), 7.69 (t, 1H, J = 7.4 Hz, H-2),
7.78 (m, 2H, H-1 and H-3), 11.31 (br, 1H, H-10). Anal. Calcd for
C17H12N2O2: C, 73.90; H, 4.38; N, 10.14. Found: C, 74.11; H, 4.37;
N, 10.13.
2-[4-(5-Chloro-1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)-3-methylphenyl]-
4-methyl-1H-isoindole-1,3(2H)-dione (22A). Compound 22A was
synthesized in a similar manner to the method 1 from 2-(4-amino-2-
methylphenyl)-5-chloro-1H-isoindole-1,3(2H)-dione. The reaction was
heated at 90 °C. After 24 h, 2 mL of acetic acid were added and the
mixture was heated at 110 °C for other 4 h. The product was isolated
through filtration as a white solid; mp 274ꢀ276 °C. MS (M + H+) m/z
431. 1H NMR (CDCl3) δ 2.27 (s, 3H, H-8), 2.78 (s, 3H, H-4), 7.33 (d,
1H, J = 8.4 Hz, H-7), 7.43 (d, 1H, J = 8.0 Hz, H-9), 7.51 (s, 1H, H-5),
7.53 (d, 1H, J = 7.6 Hz, H-3), 7.67 (t, 1H, J = 7.5 Hz, H-2), 7.78 (d, 1H,
J = 8.4 Hz, H-6), 7.82 (d, 1H, J = 7.6 Hz, H-1), 7.90 (s, 1H, H-11), 7.94
(m, 1H, H-10). Anal. Calcd for C24H15ClN2O4: C, 66.91; H, 3.51; N,
6.50. Found: C, 66.69; H, 3.73; N, 6.43.
2-(4-Acetylphenyl)-4-methyl-1H-isoindole-1,3(2H)-dione (23A).
Compound 23A was synthesized in a similar manner to the method 1
from 40-aminoacetophenone, and it was isolated as a white solid; mp
201ꢀ203 °C. 1H NMR (CDCl3) δ 2.65 (s, 3H, H-4), 2.77 (s, 3H, H-9),
7.56 (d, 1H, J = 7.6 Hz, H-3), 7.63 (d, 2H, J = 8.8 Hz, H-5 and H-8), 7.67
(t, 1H, J = 7.6 Hz, H-2), 7.81 (d, 1H, J = 7.6 Hz, H-1), 8.10 (d, 2H, J =
8.8 Hz, H-6 and H-7). Anal. Calcd for C17H13NO3: C, 73.11; H, 4.69; N,
5.02. Found: C, 73.23; H, 4.70; N, 5.00.
General Synthetic Method 2 for Library A Compounds 5A,
8A, 17A, 20A, and 24A. A mixture of amine (6.2 mmol) and 3 (1 g,
6.2 mmol) in acetic acid (10 mL) was stirred and refluxed for 5 h; the
product was precipitated by addition of water to the cooled solution. The
solid was filtrated and washed with water.
4-Methyl-2-[4-(trifluoromethyl)phenyl]-1H-isoindole-1,3(2H)-dione
(5A). Compound 5A was synthesized in a similar manner to the method
2 from 4-(trifluoromethyl)aniline and it was isolated as a white solid; mp
151ꢀ153 °C. MS (M + H+) m/z 306. 1H NMR (CDCl3) δ 2.77 (s, 3H,
H-4), 7.57 (d, 1H, J = 7.6 Hz, H-3), 7.62 (t, 1H, J = 7.6 Hz, H-2), 7.67 (d,
1H, J = 7.6 Hz, H-1), 7.74 (d, 2H, J = 8.0 Hz, H-6 and H-7), 7.82 (d, 2H,
J = 8.0 Hz, H-5 and H-8). Anal. Calcd for C16H10F3NO2: C, 62.96; H,
3.30; N, 4.59. Found: C, 63.00; H, 3.12; N 4.78.
4-(4-Methyl-1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)benzonitrile (8A).
Compound 8A was synthesized in a similar manner to the method 2 from
4-aminobenzonitrile, and it was isolated as a white solid; mp 245ꢀ250 °C.
MS (M + H+) m/z 263. 1H NMR (CDCl3) δ 2.77 (s, 3H, H-4), 7.58 (d,
1H, J = 7.2 Hz, H-3), 7.69 (t, 1H, J = 7.2 Hz, H-2), 7.70 (d, 2H, J = 9.0 Hz,
H-5 and H-8), 7.81 (d, 1H, J = 7.2 Hz, H-1), 7.82 (d, 2H, J = 9.0 Hz, H-6
and H-7). Anal. Calcd for C16H10N2O2: C, 73.27; H, 3.64; N, 10.68. Found:
C, 73.22; H, 3.45; N, 10.79.
({[4-(4-Methyl-1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)phenyl]-
carbonyl}amino)acetic Acid (17A). Compound 17A was synthesized
in a similar manner to the method 2 from {[(4-aminophenyl)carbonyl]-
amino}acetic acid, and it was isolated as a beige solid; mp 211ꢀ213 °C.
MS (M + H+) m/z 339. 1H NMR (DMSO-d6) δ 2.67 (s, 3H, H-4), 3.96
(s, 2H, H-10), 7.57 (d, 2H, J = 8.6 Hz, H-5 and H-8), 7.70 (t, 1H, J = 7.6
Hz, H-2), 7.78 (m, 2H, H-1 and H-3), 7.99 (d, 2H, J = 8.6 Hz, H-6 and
H-7), 8.91 (br, 1H, H-9), 12.42 (br, 1H, H-11). Anal. Calcd for
C18H14N2O5: C, 63.90; H, 4.17; N, 8.28. Found: C, 64.21; H, 4.02; N, 8.58.
2-Hydroxy-4-(4-methyl-1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)-
benzoic Acid (20A). Compound 20A was synthesized in a similar
4-(4-Methyl-1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)benzenecarbox-
imidamide Hydrochloride (24A). Compound 24A was synthesized in a
similar manner to the method 2 from 4-aminobenzenecarboximidamide
hydrochloride. The reaction was refluxed for 20 h. The mixture was
cooled, and the formed precipitate was filtered off. The filtrate was
evaporated, and the residue was triturated with acetone, affording the
1
beige solid product; mp 234ꢀ235 °C. MS (M ꢀ Cl)+ m/z 280. H
NMR (DMSO-d6) δ 2.68 (s, 3H, H-4), 7.71 (d, 2H, J = 8.8 Hz, H-5 and
H-8), 7.73 (t, 1H, J = 7.6 Hz, H-2), 7.79 (m, 2H, H-1 and H-3), 7.96 (d,
2H, J = 8.8 Hz, H-6 and H-7), 9.30 (br, 3H, H-9). Anal. Calcd for
C16H14ClN3O2: C, 60.86; H, 4.47; N, 15.49. Found: C, 60.74; H, 4.75;
N, 15.13.
General Synthetic Method 3 for Library B Compounds. A
mixture of 11 (0.05 g, 0.19 mmol) in DMF (2 mL) and an opportune
amine (0.3 mmol) in DMF (1 mL) was heated at 75 °C for 15 h. Then
5 mL of H2O were added to the cooled solution.
6-{[(2-Methyl-1,3-dioxo-2,3-dihydro-1H-isoindol-4-yl)methyl]amino}-
pyridine-3-carboxylic Acid (3B). Compound 3B was synthesized in a
similar manner to the method 3 from 5-aminopyridine-2-carboxylic acid,
and it was isolated as a beige solid; mp 240ꢀ245 °C. 1H NMR (DMSO-
d6) δ 3.09 (s, 3H, H-5), 5.96 (s, 2H, H-4), 7.16 (d, 1H, J = 9.2 Hz, H-8),
7.80 (m, 3H, H-1, H-2 and H-3), 8.28 (d, 1H, J = 9.2 Hz, H-7), 8.80 (s,
1H, H-6). Anal. Calcd for C15H11N3O4: C, 60.61; H, 3.73; N, 14.14.
Found: C, 60.45; H, 3.72; N, 14.11.
Ethyl 4-{[(2-Methyl-1,3-dioxo-2,3-dihydro-1H-isoindol-4-yl)methyl]-
amino}benzoate (4B). Compound 4B was synthesized in a similar
manner to the method 3 from benzocaine, and it was isolated as a white
solid; mp 198ꢀ201 °C. 1H NMR (CDCl3) δ 1.38 (t, 3H, J = 7.1 Hz,
H-11), 3.21 (s, 3H, H-5), 4.34 (q, 2H, J = 7.1 Hz, H-10), 4.96 (s, 2H,
H-4), 6.66 (d, 2H, J = 9.0 Hz, H-6 and H-9), 7.66 (m, 2H, H-1 and H-3),
7.78 (t, 1H, J = 7.4 Hz, H-2), 7.86 (d, 2H, J = 9.0 Hz, H-7 and H-8). Anal.
Calcd for C17H15N3O4: C, 62.76; H, 4.65; N, 12.92. Found: C, 62.62; H,
4.64; N, 12.96.
4-{[(3-Methoxypropyl)amino]methyl}-2-methyl-1H-isoindole-1,3(2H)-
dione (7B). Compound 7B was synthesized in a similar manner to the
method 3 from 3-methoxypropylamine, and it was isolated as a beige
solid; mp 170ꢀ175 °C. 1H NMR (CDCl3) δ 1.62 (m, 2H, H-8), 2.55
(m, 2H, H-7), 3.19 (s, 3H, H-5), 3.22 (s, 3H, H-10), 3.41 (t, 2H, J = 7.1
Hz, H-9), 4.20 (s, 2H, H-4), 7.78 (m, 3H, H-1, H-2 and H-3). Anal.
Calcd for C13H16N2O3: C, 62.89; H, 6.50; N, 11.28. Found: C, 63.01; H,
6.49; N, 11.27.
2-Methyl-4-{[(2-thiophen-2-ylethyl)amino]methyl}-1H-isoindole-
1,3(2H)-dione (9B). Compound 9B was synthesized in a similar manner
to the method 3 from 2-thiopheneethylamine, and it was isolated as a
beige solid; mp 150ꢀ155 °C. 1H NMR (CDCl3) δ 3.19 (s, 3H, H-5),
3.10 (t, 2H, J = 7.4 Hz, H-8), 3.34 (m, 2H, H-7), 4.61 (br, 2H, H-4), 6.91
(m, 2H, H-9 and H-10), 7.18 (m, 1H, H-11), 7.68 (m, 2H, H-1 and H-3),
7.80 (t, 1H, J = 7.6 Hz, H-2). Anal. Calcd for C15H14N2O2S: C, 62.92; H,
4.93; N, 9.78. Found: C, 62.74; H, 4.94; N, 9.75.
4-{[(2-Methyl-1,3-dioxo-2,3-dihydro-1H-isoindol-4-yl)methyl]-
amino}benzenesulfonic Acid (11B). Compound 11B was synthesized
in a similar manner to the method 3 from sulfanilic acid, and it was
isolated as a beige solid; mp 90ꢀ95 °C. 1H NMR (DMSO-d6) δ 3.05 (s,
3H, H-5), 4.95 (s, 2H, H-4), 6.45 (d, 2H, J = 8.8 Hz, H-6 and H-7), 7.28
(d, 2H, J = 8.8 Hz, H-8 and H-9), 7.74 (m, 2H, H-1 and H-3), 7.78 (t,
1H, J = 7.6 Hz, H-2). Anal. Calcd for C15H12N2O5S: C, 54.21; H, 3.64;
N, 8.43. Found: C, 54.33; H, 3.65; N, 8.41.
5464
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
2-Methyl-4-({[4-(phenylcarbonyl)phenyl]amino}methyl)-1H-iso-
indole-1,3(2H)-dione (13B). Compound 13B was synthesized in a
similar manner to the method 3 from 4-aminobenzophenone, and it
was isolated as a beige solid; mp 109ꢀ115 °C. 1H NMR (CDCl3) δ 3.20
(s, 3H, H-5), 4.89 (s, 2H, H-4), 6.75 (d, 2H, J = 8.8 Hz, H-6 and H-8),
7.31 (m, 3H, H-11, H-12 and H-13), 7.66 (m, 2H, H-1 and H-3), 7.50 (d,
2H, J = 8.8 Hz, H-7 and H-9), 7.70 (m, 2H, H-10 and H-14), 7.78 (t, 1H,
J = 7.6 Hz, H-2). Anal. Calcd for C22H16N2O3: C, 74.15; H, 4.53; N,
7.86. Found: C, 74.29; H, 4.55; N, 7.84.
4-{[(2-Methyl-1,3-dioxo-2,3-dihydro-1H-isoindol-4-yl)methyl]amino}-
benzonitrile (15B). Compound 15B was synthesized in a similar manner
to the method 3 from 4-aminobenzonitrile, and it was isolated as a white
solid; mp 179ꢀ188 °C. 1H NMR (CDCl3) δ 3.20 (s, 3H, H-5), 4.81 (s,
2H, H-4), 6.61 (d, 2H, J = 8.4 Hz, H-6 and H-9), 7.40 (d, 2H, J = 8.4 Hz,
H-7 and H-8), 7.66 (m, 2H, H-1 and H-3), 7.78 (t, 1H, J = 7.3 Hz, H-2).
Anal. Calcd for C16H11N3O2: C, 69.31; H, 4.00; N, 15.15. Found: C,
69.25; H, 3.99; N, 15.13.
mixture was the same as in the TS standard assay. Stock solutions of each
inhibitor were freshly prepared in DMSO and stored at ꢀ80 °C. In the
reaction mixture, the concentration of DMSO never exceeded 5%; the
effect of increasing DMSO concentration in the TS assay mixture was
studied, and it was observed that no change in TS activity was seen at
concentrations up to 8% DMSO.24 To obtain Ki values, the inhibitors
were tested at concentrations ranging from 5 to 500 μM depending
on the solubility of the compound studied. When the inhibition of
hTS was studied, the concentration of the inhibitor was tested up
to 50 μM.
Crystallization of Thymidylate Synthase Complexes. Crys-
tals of LcTS complexes were grown over a few days using the sitting/
hanging drop vapor diffusion technique at 20 °C from a 4 mg/mL LcTS
solution (buffer: 25 mM KH2PO4/K2HPO4, pH 6.8ꢀ7.4, 1 mM
EDTA) containing 40 mM dUMP. A 2 mM solution of each inhibitor
in DMSO was then added in a 1:10 volume ratio to the protein/dUMP
solution.15,25,26
2-Methyl-4-{[(4-nitronaphthalen-1-yl)amino]methyl}-1H-isoindole-
1,3(2H)-dione (16B). Compound 16B was synthesized in a similar
manner to the method 3 from 4-nitro-1-naphthylamine, and it was
isolated as a brown solid; mp 160ꢀ163 °C. 1H NMR (CDCl3) δ 3.19 (s,
3H, H-5), 5.03 (s, 2H, H-4), 6.70 (d, 1H, J = 8.4 Hz, H-6), 7.59 (m, 4H,
H-8, H-9, H-10 and H-11), 7.74 (m, 2H, H-1 and H-3), 7.82 (t, 1H, J =
7.5 Hz, H-2), 8.39 (d, 1H, J = 8.4 Hz, H-7). Anal. Calcd for
C19H13N3O4: C, 65.70; H, 3.77; N, 12.10. Found: C, 65.76; H, 3.76;
N, 12.12.
(2-{[(2-Methyl-1,3-dioxo-2,3-dihydro-1H-isoindol-4-yl)methyl]-
amino}ethyl)phosphonic Acid (22B). Compound 22B was synthesized
in a similar manner to the method 3 from 2-aminoethylphosphonic acid,
and it was isolated as a beige solid; mp 110ꢀ113 °C. 1H NMR (DMSO-
d6) δ 2.54 (m, 4H, H-7 and H-8), 3.04 (s, 3H, H-5), 4.95 (s, 2H, H-4),
7.71 (d, 1H, J = 7.4 Hz, H-3), 7.79 (t, 1H, J = 7.4 Hz, H-2), 7.90 (d, 1H,
J = 7.4 Hz, H-1), 8.13 (br, 1H, H-6). Anal. Calcd for C11H13N2O5P: C,
46.49; H, 4.61; N, 9.86. Found: C, 46.43; H, 4.62; N, 9.89.
In the case of complexes LcTS-dUMPꢀ20A and LcTS-dUMPꢀ23A
complexes, drops were prepared by mixing 3 μL of the ternary complex
solutions with 1 μL of precipitant solution made of 10% saturated
ammonium sulfate and 1 mM DTT. The drops were then inverted over
wells containing 20 mM K2HPO4 (pH 6.8) and 1 mM DTT. In the case
of all the other complexes, drops were made by mixing equal volumes
(2ꢀ3 μL) of the LcTS ternary complex solution with a precipitant
solution consisting of (NH4)H2PO4/(NH4)2HPO4 100 mM, 5% v/v
PEG 400, and 20 mM Tris-HCl buffer at pH 7.4 and equilibrated against
the same precipitant solution.
Crystals of the complexes appear in 5ꢀ6 days as hexagonal
bipyramids and grow to the final dimensions of about 150 μm ꢁ
150 μm ꢁ 400 μm in about 2 weeks. Before the data collection,
crystals were transferred to a cryoprotectant solution composed of
the precipitant solution with added 30% PEG 400 and flash frozen in
liquid nitrogen. The data collection and refinement statistics are
reported in Table 2.
3-{[(2-Methyl-1,3-dioxo-2,3-dihydro-1H-isoindol-4-yl) methyl]amino}
benzenesulfonamide (23B). Compound 23B was synthesized in a
similar manner to the method 3 from 3-aminobenzenesulfonamide,
and it was isolated as a white solid; mp 230ꢀ235 °C. 1H NMR (DMSO-
d6) δ 3.06 (s, 3H, H-5), 4.77 (s, 2H, H-4), 6.71 (d, 1H, J = 7.2 Hz, H-6),
6.81 (br, H-10), 6.99 (d, 1H, J = 7.2 Hz, H-8), 7.06 (s, 1H, H-9), 7.16 (br,
2H, H-11), 7.21 (t, 1H, J = 7.2 Hz, H-7), 7.71 (m, 3H, H-1, H-2 and
H-3). Anal. Calcd for C15H13N3O4S: C, 54.37; H, 3.95; N, 12.68.
Found: C, 54.34; H, 3.94; N, 12.65.
Enzymology. The compounds were screened for their activity and
their species-specificity against EcTS, EfTS, and hTS. Because crystal-
lographic studies were performed also against LcTS, the activity of these
compounds toward LcTS enzyme was assayed as well. LcTS, EcTS, and
hTS were purified as described.19ꢀ21 EfTS was purified following the
procedure used for EcTS. The EfTS strain was a gift of Prof. Robert M.
Stroud, University of California San Francisco (CA, USA). Enzyme
kinetics experiments were conducted by chromogenic assay under
standard conditions.22 The Km values, by curve-fitting to hyperbolic
curves, of EcTS, EfTS, LcTS, and hTS for MTHF were, respectively, 5.8,
13.6, 5.7, and 6.9 μM. The Ki ( standard errors were determined from
the nonlinear regression analysis of data from at least two independent
experiments performed in triplicate.
Data Collection and Structure Determination. Data sets
related to LcTS-dUMPꢀ6A_1, LcTS-dUMPꢀ8A, LcTS-dUMPꢀ12A_1
complexes were collected using synchrotron radiation from the
European Synchrotron Radiation Facility, Grenoble, France (ESRF)
ID14ꢀ1 with wavelength of 0.930 Å with the rotation technique
(Δj = 1°) and an ADSC Quantum 4 detector. The data set of the
LcTS-dUMPꢀ6A_2 complex was collected using synchrotron radiation
from the European Synchrotron Radiation Facility, Grenoble, France
(ESRF) ID23ꢀ1 with wavelength of 1.008 Å, the rotation technique
(Δj = 0.75°), and a MARresearch CCD 225 detector. The data set of
the LcTS-dUMPꢀ12A_2 complex was collected using synchrotron
radiation of European Molecular Biology Laboratory (EMBL) PX
beamline BW7B at the DORIS storage ring, DESY, Hamburg, Germany,
with wavelength of 1.078 Å, the rotation technique (Δj = 1°),
and detector MARresearch CCD 225 detector. The data sets of the
LcTS-dUMPꢀ20A, LcTS-dUMPꢀ23A, and LcTS-dUMPꢀ24A com-
plexes were collected using synchrotron radiation from the ESRF
beamline ID23ꢀ1 with wavelength of 0.983 Å, the rotation technique
(Δj = 0.5°), and a ADSC Quantum Q315r detector.
All the data sets were processed using MOSFLM 6.227,28 and scaled
with SCALA.29
The structure solutions of all complexes were performed by the
molecular replacement method using the MOLREP software,28,30 where
one subunit of LcTS (PDB entry 1LCA) was used as a model for the
rotation and translation functions. All structures were refined and
water molecules were added using REFMAC29,31 and ARP/WARP,31
respectively.
The inhibition patterns for all the compounds were determined by
steady-state kinetic analysis of the dependence of enzyme activity on
MTHF concentration at varying inhibitor concentrations. All the
compounds showed competitive inhibition with respect to MTHF.
The apparent inhibition constant (Ki app, simply defined as Ki within
the text) values were obtained from the linear least-squares fit of the
residual activity as a function of inhibitor concentration, using suitable
equations for competitive inhibition.23 In the inhibition assays, the reaction
The program Xtalview/Xfit33 and Coot32 were used for manual
rebuilding of the molecule, modeling of the inhibitor molecules into
the electron density map and visualization for all structures.
5465
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
Figures were generated using CCP4MG from the CCP4 suite28 and
Chimera34 program.
’ REFERENCES
(1) Carreras, C. W.; Santi, D. V. The catalytic mechanism and
structure of thymidylate synthase. Annu. Rev. Biochem. 1995, 64, 721–762.
(2) Chu, E.; Callender, M. A.; Farrell, M. P.; Schmitz, J. C. Thymi-
dylate synthase inhibitors as anticancer agents: from bench to bedside.
Cancer Chemother. Pharmacol. 2003, 52 (Suppl1), S80–S89.
(3) Shoichet, B. K; Stroud, R. M.; Santi, D. V.; Kuntz, I. D.; Perry,
K. M. Structure-based discovery of inhibitors of thymidylate synthase.
Science 1993, 259, 1445–1450.
(4) Finer-Moore, J. S., Blaney, J., Stroud, R. M., Computational and
Structural Approaches to Drug Discovery; Stroud, R. M. and Finer-Moore,
J. S., Eds; RSC: Cambridge, UK, 2008. p 9.
(5) Stout, T. J.; Tondi, D.; Rinaldi, M.; Barlocco, D.; Pecorari, P.;
Santi, D. V.; Kuntz, I. D.; Stroud, R. M.; Shoichet, B. K.; Costi, M. P.
Structure-based design of inhibitors specific for bacterial thymidylate
synthase. Biochemistry 1999, 38, 1607–1617.
(6) Costi, M. P.; Rinaldi, M.; Tondi, D.; Pecorari, P.; Barlocco, D.;
Ghelli, S.; Stroud, R. M.; Santi, D. V.; Stout, T. J.; Musiu, C.; Marangiu,
E. M.; Pani, A.; Congiu, D.; Loi, G. A.; La Colla, P. Phthalein derivatives
as a new tool for selectivity in thymidylate synthase inhibition. J. Med.
Chem. 1999, 42, 2112–2124.
(7) Ghelli, S.; Rinaldi, M.; Barlocco, D.; Gelain, A.; Pecorari, P.;
Tondi, D.; Rastelli, G.; Costi, M. P. ortho-Halogen naphthaleins as
specific inhibitors of Lactobacillus casei thymidylate synthase. Conforma-
tional properties and biological activity. Bioorg. Med. Chem. 2003,
11, 951–963.
’ ASSOCIATED CONTENT
S
Supporting Information. Binding modes of known
b
phthalein and naphthalein compounds (PTH; R156, MR20,
GA9); designed libraries A and B; calculated molecular proper-
ties for libraries A and B; atom numbering in compounds
characterized through NMR; plot of the enzyme kinetic inhibi-
tion profiles (Ki) for libraries A and B; contact distances between
substrates and LcTS residues. This material is available free of
charge via the Internet at http://pubs.acs.org.
Accession Codes
The X-ray coordinates of compounds 6A, 8A, 12A, 20A, 23A,
and 24A bound to LcTS-dUMP have been deposited in the
Protein Data Bank with accession numbers 3C06 (6A), 3C0A
(6A), 3BNZ (8A), 3BYX (12A), 3BZ0 (12A), 3IK1 (20A), 3IJZ
(23A), and 3IK0 (24A).
’ AUTHOR INFORMATION
Corresponding Author
(8) Calꢀo, S.; Tondi, D.; Venturelli, A.; Ferrari, S.; Pecorari, P.;
Rinaldi, M.; Ghelli, S.; Costi, M. P. A step further in the discovery of
phthalein derivatives as thymidylate synthase inhibitors. ARKIVOC
2004, 382–396.
(9) Finer-Moore, J. S.; Anderson, A. C.; O’Neil, R. H.; Costi, M. P.;
Ferrari, S.; Krucinski, J.; Stroud, R. The structure of Cryptococcus
neoformans thymidylate synthase suggests strategies for using target
dynamics for species-specific inhibition. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 2005, D61, 1320–1334.
(10) Costi, M. P.; Gelain, A.; Barlocco, D.; Ghelli, S.; Soragni, F.;
Raniero, F.; Rossi, T.; Ruberto, A.; Guillou, C.; Cavazzuti, A.; Casolari,
C.; Ferrari, S. Anti-bacterial agent discovery using thymidylate synthase
bio-library screening. J. Med. Chem. 2006, 49, 5958–5968.
(11) Babaoglu, K.; Shoichet, B. K. Deconstructing fragment-based
inhibitor discovery. Nature Chem. Biol. 2006, 2, 720–723.
(12) ChemFinder Ultra Database 7.0; Cambridge Soft Corporation,
htpp://www.cambridgesoft.com.
*For M.P.C.: phone, 0039-059-205-5125; fax, 0039-059-205-
5131; E-mail, mariapaola.costi@unimore.it. For S.M.: phone,
0039-0577-234255; fax, 0039-0577-234233; E-mail, mangani@
unisi.it. For S.F.: phone, 0039-059-205-5331; fax, 0039-059-205-
5131; E-mail, stefania.ferrari@unimore.it.
Present Addresses
§Roche Pharma (Schweiz) AG, Sch€onmattstrasse 2, 4153 Reinach
(BL), Switzerland.
’ ACKNOWLEDGMENT
This work was financially supported by MIUR-PRIN 2006-
2006030430_004 CSTMPL. We thank the Cassa di Risparmio di
Modena (CDRM) Foundation. We are grateful to Merck Eprova
AG, Switzerland, an affiliate of Merck KGaA, Darmstadt,
Germany, for providing the (6R)-5,10-methylenetetrahydrofolic
acid sodium salt. We also acknowledge ESRF and the EMBL
Hamburg outstation for the use of their beamlines for data
collection. We thank Professor B. K. Shoichet for providing the
pdb file of the crystallographic complex LcTS-PTH, Professor
Marcella Rinaldi for assistance with the synthetic work, and
Fabrizia Soragni for her technical assistance.
(13) ACDlab 6.0 Software; Cambridge Soft Corporation, htpp://
www.cambridgesoft.com.
(14) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J.
Experimental and computational approaches to estimate solubility and
permeability in drug discovery and development settings. Adv. Drug
Delivery Rev. 2001, 46, 3–26.
(15) Hardy, L. W.; Finer-Moore, J. S.; Montfort, W. R.; Jones, M. O;
Santi, D. V.; Stroud, R. M. Atomic structure of thymidylate synthase:
target for rational drug design. Science 1987, 235, 448–455.
(16) Finer-Moore, J.; Fauman, E. B.; Foster, P. G.; Perry, K. M.;
Santi, D. V.; Stroud, R. Refined structures of substrate-bound and
phosphate-bound thymidylate synthase from Lactobacillus casei. J. Mol.
Biol. 1993, 232, 1101–1116.
’ ABBREVIATIONS USED
(17) Ferrari, S.; Losasso, V.; Costi, M. P. Sequence-Based identifica-
tion of specific drug target regions in the thymidylate synthase enzyme
family. ChemMedChem 2008, 3, 392–401.
CDCl3, deuterated trichloromethane; DMF, dimethylforma-
mide; DMSO, dimethyl sulfoxide; dTMP, 20-deoxythymidine-
50-monophosphate; DTT, dithiotreitol; dUMP, 20-deoxyuridine-
50-monophosphate; EcTS, Escherichia coli thymidylate synthase;
EDTA, ethylenediaminetetraacetic acid; EfTS, Enterococcus fae-
calis thymidylate synthase; hTS, human thymidylate synthase; Ki,
inhibition constant; LcTS, Lactobacillus casei thymidylate synthase;
MTHF, N5,N10-methylene tetrahydrofolate; PEG 400, polyethy-
lene glycol, MW 400;PTH, phenolphthalein;SD, small domain;TS,
thymidylate synthase
(18) Ferrari, S.; Costi, M. P.; Wade, R. Inhibitor specificity via
protein dynamics: insights from the design of antibacterial agents
targeted against thymidylate synthase. Chem. Biol. 2003, 10, 1183–1193.
(19) Kealey, J. T.; Santi, D. V. Purification methods for recombinant
Lactobacillus casei thymidylate synthase and mutants: a general, auto-
mated procedure. Protein Expression Purif. 1992, 4, 380–385.
(20) Maley, G. F.; Maley, F. Properties of a defined mutant of
Escherichia coli thymidylate synthase. J. Biol. Chem. 1988, 263, 7620–7627.
5466
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467

Journal of Medicinal Chemistry
ARTICLE
(21) Pedersen-Lane, J.; Maley, G. F.; Chu, E.; Maley, F. High-level
expression of human thymidylate synthase. Protein Expression Purif.
1997, 10, 256–262.
(22) Pogolotti, A. L., Jr.; Danenberg, P. V.; Santi, D. V. Kinetics and
mechanism of interaction of 10-propargyl-5,8-dideazafolate with thymi-
dylate synthase. J. Med. Chem. 1986, 29, 478–482.
(23) Segel, I. H. Enzyme Kinetics. Behaviour and Analysis of Rapid
Equilibrium and Steady-State Enzyme Systems; John Wiley and Sons:
New York, 1975, p 105.
(24) Cavazzuti, A; Paglietti, G; Hunter, W. N.; Gamarro, F; Piras, S;
Loriga, M; Alleca, S; Corona, P; McLuskey, K; Tolloch, L.; Gibellini, F.;
Ferrari, S.; Costi, M. P. Discovery of potent pteridine reductase
inhibitors ot guide antiparasite drug development. Proc. Natl. Acad.
Sci. U.S.A. 2008, 105, 1448–1453.
(25) Jaspard, E. Role of proteinꢀsolvent interactions in refolding:
effects of cosolvent additives on the renaturation of porcine pancreatic
elastase at various pHs. Arch. Biochem. Biophys. 2000, 15, 220–228.
(26) Bhattacharjya, S.; Balaram, P. Effects of organic solvents on
protein structures: observation of a structured helical core in hen egg-
white lysozyme in aqueous dimethylsulfoxide. Proteins 1997, 29, 492–507.
(27) Leslie, A. G. W. Crystallographic Computing V; Moras, D.,
Podjarny, A. D., Thierry, J.-C., Eds.; Oxford University Press: Oxford,
1991; pp50ꢀ61.
(28) Collaborative Computational Project, No. 4. The CCP4 suite:
programs for protein crystallography. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 1994, 50, 760ꢀ763.
(29) Evans, P. R. SCALA, continuous scaling program. Joint CCP4
ESF-EACBM Newsletter 1997, 33, 22ꢀ24.
(30) Vagin, A; Teplyakov, A. MOLREP: an automated program for
molecular replacement. J. Appl. Crystallogr. 1997, 30, 1022–1025.
(31) Murshudov, G. N.; Vagin, A. A.; Dodson, E. J. Refinement of
macromolecular structures by the maximum-likelihood method. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 1997, 53, 240–255.
(32) Perrakis, A.; Morris, R. J. H.; Lamzin, V. S. Automated protein
model building combined with iterative structure refinement. Nature
Struct. Biol. 1999, 6, 458–463.
(33) McRee, D. E. XtalView/Xfit—a versatile program for manip-
ulating atomic coordinates and electron density. J. Struct. Biol. 1999,
125, 156–165.
(34) UCSF Chimera, Production Version 1 Build 2038 2004/10/18,
2004.
5467
dx.doi.org/10.1021/jm2005018 |J. Med. Chem. 2011, 54, 5454–5467