Improving specificity vs bacterial thymidylate synthases through N-dansyl modulation of didansyltyrosine

Article abstract of DOI:10.1021/jm0491445

N,O-Didansyl-L-tyrosine (DDT) represented the starting lead for further development of novel non-substrate-like inhibitors of bacterial thymidylate synthase. The N-dansyl structure modulation led to a submicromolar inhibitor of Lactobacillus casei TS (LcT

Full text of DOI:10.1021/jm0491445

J. Med. Chem. 2005, 48, 913-916
913
Improving Specificity vs Bacterial
Thymidylate Synthases through
N-Dansyl Modulation of Didansyltyrosine
Donatella Tondi,*,† Alberto Venturelli,†
Stefania Ferrari,† Stefano Ghelli,‡ and
M. Paola Costi*,†
Dipartimento di Scienze Farmaceutiche,
Universita` degli Studi di Modena e Reggio Emilia,
via Campi 183, 41100 Modena, Italy, and SPINLAB srl,
via Tamagno 3, Rubiera (RE), Italy
Received October 26, 2004
Abstract: N,O-Didansyl-L-tyrosine (DDT) represented the
starting lead for further development of novel non-substrate-
like inhibitors of bacterial thymidylate synthase. The N-dansyl
structure modulation led to a submicromolar inhibitor of
Lactobacillus casei TS (LcTS), which is highly specific with
respect to human TS (hTS). Using molecular dynamics simu-
lation, a binding mode for DDT vs LcTS was predicted,
explaining activity and species-specificity along the series.
Thymidylate synthase (TS) (EC 2.1.1.45) is an es-
sential enzyme for most living organisms. It catalyzes
the formation of the DNA precursor, 2′-deoxythymidine-
5′-monophosphate, from the substrate 2′-deoxyuridine-
5′-monophosphate (dUMP) and the cofactor 5,10-meth-
ylene-5,6,7,8-tetrahydrofolate (mTHF). Because of its
critical function, efforts have been focused on the design
of TS inhibitors with anticancer or antimicrobial activ-
ity. However, the design of species-specific inhibitors
that are able to discriminate among the different species
of this highly conserved enzyme has been challenging.2-4
During previous research, a strategy to design novel
TS inhibitors successfully combined structure-based
virtual screening and in-parallel synthetic elaboration
of a small, focused library. The most potent inhibitor
in the library was N,O-didansyl-L-tyrosine (DDT) show-
Figure 1. A: DDT-LcTS predicted orientation. Carbon atoms
are colored magenta for ligand and cyan for dUMP; oxygen
atoms, red; nitrogen atoms, blue; sulfur atoms, yellow. B:
DDT-EcTS X-ray orientation. Key residues are shown. Carbon
atoms are colored gray for protein, cyan for dUMP and
magenta for ligand; oxygen atoms, red; nitrogen atoms, blue;
sulfur atoms, yellow.
domain region of 50 amino acids (residues lc-90-lc-139),
which are missing in EcTS and hTS.5-8
Using the docking program DOCK 3.5, an interaction
model of DDT with LcTS was predicted (Figure 1A).1
The X-ray structure of the ternary complex dUMP-
EcTS-DDT has recently been solved, showing that
the binding orientation of DDT to EcTS is quite unique
and different from the one DOCK predicted vs LcTS
(Figure 1B).9 Upon binding to EcTS, as seen in the X-ray
structure, DDT induces large rearrangements of resi-
dues both within and remote from the active site,
promoting the binding of DDT in an intramolecular
sandwiched conformation (folded conformation), which
is very different from the linear one DOCK predicted
(Figure 1A).
The observed crystallographic binding orientation for
DDT compared to the predicted one raises the question
of the reliability of our model vs LcTS and its ability to
support the design of new derivatives that are highly
specific for bacterial TS with respect to hTS. In the effort
to answer this question and study the molecular basis
for the recognition of DDT in LcTS, EcTS, and hTS
binding pockets, a new set of dansyl derivatives stem-
ming from the lead compound was designed.
ing the same affinity toward Lactobacillus casei TS
(LcTS) and Escherichia coli TS (EcTS) (KiLcTS 1.4 µM;
KiEcTS 1.8 µM), as a result being 35-fold more specific
against these two bacterial species compared to human
TS (KihTS 49 µM).
In particular, LcTS does not come from a pathogen
itself but represents a good model for TS from pathogens
such as Staphylococcus aureus, Enterococcus faecalis,
and Bacillus anthracis, in that they all share the small
* Corresponding authors: tondid@unimore.it, costimp@unimore.it,
phone +390592055134, fax+390592055131.
† Universita` degli Studi di Modena e Reggio Emilia.
‡ SPINLAB s.r.l.
10.1021/jm0491445 CCC: $30.25 © 2005 American Chemical Society
Published on Web 02/02/2005

914 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4
Letters
Scheme 1a
Table 1. DDT Derivatives and Their Biological Activity Profile
(SI ) specificity index)
a Reagents and conditions: a: NaHCO3, NaOH pH 10, r.t.; b:
H2, Pd/C.
a Ki hTS/Ki LcTS. b Ki hTS/Ki EcTS. c See ref 1.
showed a competitive pattern with respect to the folate
cofactor. Each experiment was repeated at least three
times and individual measurements did not differ from
the mean by more than 20%.
Substituting the N-dansyl fragment with different
moieties provided the opportunity to span the active site
of TS: the introduction of different chemical groups
allowed the improvement and modulation of DDT activ-
ity and specificity toward bacterial species rather than
hTS. Using molecular dynamics simulation, a new
binding mode vs LcTS was calculated to provide a better
explanation of activity and specificity along the series,
in comparison with hTS.
The DDT derivatives were synthesized as depicted in
Scheme 1. The dansyl derivatives were prepared by
reacting the dansyl-L-tyrosine (DT) with the appropriate
sulfonyl chloride prepared in-house, in NaHCO3 at room
temperature.
The compounds were screened for their activity and
specificity against LcTS, EcTS, and hTS. Apparent
inhibition constants (Ki) were determined, and the
specificity index (SI) against hTS (Table 1) was calcu-
lated.
LcTS, EcTS, and hTS were purified as described.10-12
Enzyme kinetic experiments were conducted under
standard conditions.7,13 The concentration of mTHF was
kept 10 times the Km (Km values of EcTS, LcTS, and
hTS are 5.3, 6.9, and 4.4 µM, respectively), and the
dUMP concentration was 100 µM. Stock solutions of the
inhibitors were freshly prepared in dimethyl sulfoxide
(DMSO). The DMSO never exceeded the concentration
of 5% in the reaction mixture. All the compounds
Molecular dynamics studies were undertaken for the
LcTS-dUMP-DDT complex. Two possible ternary com-
plexes of LcTS-dUMP-DDT were built: in each of
them a different binding conformation of the inhibitor
was considered. The protein coordinates were taken
from the 1LCA structure (LcTS-dUMP-CB3717) and
some modifications were introduced into the protein
conformation.5 Since the crystal structure of the
EcTS-dUMP-DDT complex shows that DDT induces
essential and unique rearrangements in EcTS, these
conformational changes were included in our model of
the LcTS-dUMP-DDT complex, assuming that this
compound could induce in LcTS the same rearrange-
ments as observed in EcTS.9 Therefore, based on the
1JG0 structure (EcTS-dUMP-DDT), the conformation
of two regions of LcTS (lc-Gly17-lc-Try29 and lc-Phe294-
lc-Val316) were modified to properly fit DDT into two
binding conformations: in the first one, DDT is in the
X-ray folded conformation; in the second one, the
N-dansyl-moiety orientation was taken from the docking
binding model.1 dUMP coordinates were taken from the
1JG0 structure. The complexes were solvated with 24
Å spherical caps of TIP3P (AMBER6.0). The complexes
were initially energy-minimized and then subjected to
400 ps of molecular dynamics (Sander_classic/AMBER

Letters
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4 915
6.0).14 All protein residues within 10 Å from the inhibi-
tor, and all the water molecules were considered in the
calculations. EcTS9 and hTS15 crystal structures were
used for comparative analysis. InsightII was used for
visual analysis.16
In a small subset of DDT derivatives the N-dansyl
moiety was substituted with chemical groups that have
different molecular properties; the biological profile of
the synthesized molecules was evaluated against LcTS,
EcTS, and hTS enzymes and the structural require-
ments necessary for the specific recognition of DDT and
derivatives vs LcTS compared to EcTS and hTS were
analyzed.
Molecular dynamics studies were undertaken and two
different models of LcTS-dUMP-DDT were calculated.
In the first binding mode, DDT assumes a “folded”
conformation, showing a pattern of interactions similar
to that observed in the EcTS crystal structure, except
for two hydrogen bonds formed with ec-Thr78 and
ec-Ser54. These residues correspond to lc-His80 and
lc-Leu56 in LcTS. lc-His80 interacts with the ligand but
with poor hydrogen bond geometry, whereas the apolar
side-chain of lc-Leu56 cannot form any hydrogen bonds.
In the second binding mode, DDT assumes a “semi-
folded” conformation and forms ring-ring interactions
with lc-His80 and lc-Phe228, and hydrophobic interac-
tions with lc-Ile81, lc-Trp82, lc-Leu195, and lc-Leu224.
The carboxyl group interacts with the side chain of
lc-Lys50, and hydrogen bonds are formed between the
ligand and water molecules present in the active site
(Figure 2A).
The residue ec-Thr78 is the key difference be-
tween LcTS and EcTS. In the folded X-ray conformation,
ec-Thr78 forms a hydrogen bond with the O69 ligand
atom, whereas the corresponding residue lc-His80 does
not. In the case of the semifolded bound conformation,
lc-His80 forms a ring-ring interaction with the tyrosine
ring of DDT, whereas ec-Thr78 cannot.
These results led us to the conclusion that the folded
conformation is the one preferred by EcTS, while the
semifolded conformation is preferred by LcTS. On the
basis of these results, the activity and specificity data
for some of the dansyl derivatives under study were
evaluated.
In hTS, in the folded conformation, DDT loses two
hydrogen bonds compared to EcTS. In fact, h-Gly83 and
h-Lys107 (corresponding to ec-Ser54 and ec-Thr78) do
not form hydrogen bonds with DDT. In the semifolded
conformation, h-Lys107 (corresponding to lc-His80) does
not form a ring-ring interaction with DDT. Moreover,
it is reasonable that there is a small change in the
enzyme conformation because of the bulkier residue
h-Met311 (lc-Val314). Therefore, in hTS, both of these
two binding conformations are less favored, in agree-
ment with the differences in Ki.
Compound 1 showed the most interesting profile,
being very specific vs LcTS with respect to EcTS and
hTS, improving the specificity already shown by DDT
for LcTS against hTS (SI from 35-fold to 130-fold,
respectively) (Table 1).
Figure 2. DDT (A) and compound 1 (B) in their semifolded
conformation in LcTS. Carbon atoms are colored green for
protein, cyan for dUMP, and magenta for ligand; oxygen
atoms, red; nitrogen atoms, blue; sulfur atoms, yellow. Figure
B is rotated compared to image A for a better representation
of the “Z” group interacting with the protein. Carbon atoms
are colored green for protein, cyan for dUMP, and blue for
ligand; oxygen atoms, red; nitrogen atoms, blue; sulfur atoms,
yellow.
group (“Z” group) of 1 to interact with the small domain
region in LcTS, whereas it is exposed to the solvent in
the case of EcTS and hTS, thus making 1 highly specific
toward LcTS. Compound 1 can orient the “Z” group
toward lc-Glu84, lc-Phe104, lc-Gly105, and lc-Tyr118
(Figure 2B). These residues are missing in EcTS and
hTS, except for lc-Glu84 corresponding to ec-Glu82 and
to h-Ala111. Therefore, in these enzymes the “Z” group
of 1 lies on the protein surface.
However, compound 2, a close derivative of 1, showed
a completely different activity profile. Moving the
functionality from position 5 to 4 removed the specificity
shown by compound 1 vs LcTS. Moreover, compound 2
(KihTS 11 µM) lost specificity vs bacterial species TS if
compared to DDT and compound 1 (Table 1).
Compounds 3, 4, and 5 were synthesized to evaluate
the possible role of different functionalities on the
N-dansyl and how these could interfere with activity
and specificity. Compounds 3 and 4 showed moderate
specificity against bacterial TS with respect to hTS (SI
within 14-28). Nevertheless, they showed sub-micro-
molar affinity for LcTS and EcTS. They showed the
same enzyme inhibition profiles even if the N-group
carries a nitro, lipophilic group in 3 and an amino,
hydrophilic one in 4 (Table 1).
For compound 4, we considered the same binding
modes as for DDT. Visual inspection of LcTS and EcTS
binding sites suggests that the amino group in posi-
tion 6 of the N-dansyl ring allows the formation of
additional hydrogen bonds with the enzyme residues
Because of the bulky substituent in position 5 of the
N-dansyl ring, we consider that compound 1 binds to
all the enzymes in the semifolded conformation. The
substitution in position 5 causes the benzyloxycarboxyl

916 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4
Letters
(3) Costi, M. P.; Tondi, D.; Rinaldi, M.; Barlocco, D.; Pecorari, P.
G.; Soragni, F.; Venturelli, A.; Stroud, R. M. Structure-based
studies on species-specific inhibition of thymidylate synthase.
Biochim. Biophys. Acta (BBA) - Mol. Basis Dis. 2002, 1587,
206-214.
(4) Costi, P. M.; 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 thymidy-
late synthase inhibition. J. Med. Chem. 1999, 42 (12), 2112-
2124.
and water molecules present in the binding sites, in this
way stabilizing the hydrogen bond networks. This
explains the increase in the affinity of 4 compared to
DDT.
Compound 5 seems able to discriminate between
LcTS and EcTS (KiLcTS 1.6 µM; KiEcTS 15 µM) showing
nine times lower affinity compared to EcTS (Table
1).
Preliminary evaluation of the antimicrobial activity
of DDT derivatives was performed: compound 1 showed
improved antibacterial activity toward Staphylococcus
aureus K23 with respect to DDT.19
(5) Finer-Moore, J.; Fauman, E. B.; Foster, P. G.; Perry, K. M.; Santi,
D. V.; Stroud, R. M. Refined structures of substrate-bound and
phosphate-bound thymidylate synthase from Lactobacillus casei.
J. Mol. Biol. 1993, 20, 1101-1116.
(6) Hardy, L. W.; Finer-Moore, J.; 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.
(7) Carreras, C. W.; Santi, D. V. The catalytic mechanism and
structure of thymidylate synthase. Annu. Rev. Biochem. 1995,
64, 721-762.
(8) Coque, T. M.; Singh, K. V.; Weinstock, G. M.; Murray, B. E.
Characterization of dihydrofolate reductase genes from tri-
methoprim-susceptible and trimethoprim-resistant strains of
Enterococcus faecalis. Antimicrob. Agents Chemother. 1999, 43,
141-147.
(9) Fritz, T. A.; Tondi, D.; Finer-Moore, J. S.; Costi, M. P.; Stroud,
R. M. Predicting and harnessing protein flexibility in the design
of species-specific inhibitors of thymidylate synthase. Chem. Biol.
2001, 8, 981-995.
(10) Climie, S.; Santi, D. V. Chemical synthesis of the thymidylate
synthase gene. Proc. Natl. Acad. Sci. 1990, 87, 633-637.
(11) Davisson, V. J.; Sirawaraporn, W.; Santi, D. V. Expression of
human thymidylate synthase in Escherichia coli. J. Biol. Chem.
1989, 264, 9145-9148.
We have described the synthesis, biological evalua-
tion, and molecular dynamics studies of a new series of
non-substrate-like antifolate inhibitors of TS. Among
the synthesized molecules, compound 1 showed high
specificity vs LcTS with respect to EcTS and hTS. It
binds to the bacterial enzyme LcTS at sub-micromolar
concentrations and has 130-fold lower activity toward
hTS. Molecular modeling studies for 1 vs LcTS, EcTS,
and hTS suggest that the N-dansyl ring interacts
directly with residues of TS that are peculiar of LcTS.
This inhibitor may lead to antimicrobial drugs targeting
TSs enzymes that resemble LcTS in their sequence.7,8
Furthermore, studies involving the design and syn-
thesis of structurally constrained dansyl derivatives in
a forthcoming paper will clarify conformation-mediated
enzyme specificity mechanisms in TS enzymes.20
(12) Davisson, V. J.; Sirawaraporn, W.; Santi, D. V. Expression of
human thymidylate synthase in Escherichia coli. J. Biol. Chem.
1994, 269, 30740.
Acknowledgment. This work was supported by
MURST-Costi-2002033121_002. We would like to thank
CIGS (Centro Interdipartimentale Grandi Strumenti)
and CICAIA (Centro Interdipartimentale di Calcolo
Elettronico) of the University of Modena and Reggio
Emilia for instruments and computing facilities. We
thank Prof. D. V. Santi and Prof. F. Maley for supplying
the enzyme strains, Mrs. F. Soragni for her excellent
technical assistance, and Prof. P. Pecorari for the critical
reading of the manuscript.
(13) Pogolotti, A. L.; Danenberg, P. V.; Santi, D. V. Kinetics and
mechanism of interaction of 10-propargyl-5,8-dideazafolate with
thymidylate synthase. J. Med. Chem. 1986, 29, 478-482.
(14) Case, D. A. P., D. A.; Caldwell, J. W.; Cheatham, T. E., III.; Ross,
W. S.; Simmerling, C. L.; Darden, T. A.; Merz, K. M.; Stanton,
R. V.; Cheng, A. L.; Vincent, J. J.; Crowley, M.; Tsui, V.; Radmer,
R. J.; Duan, Y.; Pitera, J.; Massova, I.; Seibel,; Singh, G. L.;
Weiner, U. C.; Kollman, P. K. AMBER 6. 1999: University of
California, San Francisco.
(15) Phan, J.; Koli, S.; Minor, W.; Dunlap, R. B.; Berger, S. H.;
Lebioda, L. Human thymidylate synthase is in the closed
conformation when complexed with dUMP and raltitrexed, an
antifolate drug. Biochemistry 2001, 40, 1897-1902
(16) InsightII, InsighII, Modeling Environment. 2000; http://www.
accelrys.com/.
(17) 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, 5, 1445-1450.
(18) Strynadka, N. C.; Eisenstein, M.; Katchalski-Katzir, E.; Shoichet,
B. K.; Kuntz, I. D.; Abagyan, R.; Totrov, M.; Janin, J.; Cherfils,
J.; Zimmerman, F.; Olson, A.; Duncan, B.; Rao, M.; Jackson, R.;
Sternberg, M.; James, M. N. Molecular docking programs
successfully predict the binding of a beta-lactamase inhibitory
protein to TEM-1 beta-lactamase. Nat. Struct. Biol. 1996, 3,
233-239.
Supporting Information Available: Mono- and bidi-
mensional 1H NMR data for all compounds. More detailed
information on the modeling methodologies applied and a
detailed description of the calculated model are included. This
material is available free of charge via the Internet at
http://pubs.acs.org.
References
(1) Tondi, D.; Slomczynska, U.; Costi, M. P.; Watterson, D. M.;
Ghelli, S.; Shoichet, B. K. Structure-based discovery and in-
parallel optimization of novel competitive inhibitors of thymidy-
late synthase. Chem. Biol. 1999, 6, 319-331.
(2) Jackson, R. C. Contributions of protein structure-based drug
design to cancer chemotherapy. Semin. Oncol. 1997, 24, 164-
172.
(19) Compound 1, MIC 10 µg/mL. Rossi T. Personal communication.
(20) Tondi, D.; Calo`, S.; Ferrari, S.; Ghelli, S.; Venturelli A.; Costi,
M. P. Constrained dansyl derivatives probing molecular recogni-
tion vs bacterial Thymidylate Synthases. In preparation.
JM0491445