Inhibition of mycobacterium tuberculosis methionine aminopeptidases by bengamide derivatives

Article abstract of DOI:10.1002/cmdc.201100003

Methionine aminopeptidase (MetAP) carries out an essential function of protein N-terminal processing in many bacteria and is a promising target for the development of novel antitubercular agents. Natural bengamides potently inhibit the proliferation of ma

Full text of DOI:10.1002/cmdc.201100003

DOI: 10.1002/cmdc.201100003
Inhibition of Mycobacterium tuberculosis Methionine
Aminopeptidases by Bengamide Derivatives
Jing-Ping Lu,[a] Xiu-Hua Yuan,[a] Hai Yuan,[a] Wen-Long Wang,[a] Baojie Wan,[b]
Scott G. Franzblau,[b] and Qi-Zhuang Ye*[a]
Methionine aminopeptidase (MetAP) carries out an essential
function of protein N-terminal processing in many bacteria and
is a promising target for the development of novel antituber-
cular agents. Natural bengamides potently inhibit the prolifera-
tion of mammalian cells by targeting MetAP enzymes, and the
X-ray crystal structure of human type 2 MetAP in complex with
a bengamide derivative reveals the key interactions at the
active site. By preserving the interactions with the conserved
residues inside the binding pocket while exploring the differ-
ences between bacterial and human MetAPs around the bind-
ing pocket, seven bengamide derivatives were synthesized and
evaluated for inhibition of MtMetAP1a and MtMetAP1c in dif-
ferent metalloforms, inhibition of M. tuberculosis growth in rep-
licating and non-replicating states, and inhibition of human
K562 cell growth. Potent inhibition of MtMetAP1a and
MtMetAP1c and modest growth inhibition of M. tuberculosis
were observed for some of these derivatives. Crystal structures
of MtMetAP1c in complex with two of the derivatives provided
valuable structural information for improvement of these in-
hibitors for potency and selectivity.
Introduction
Methionine aminopeptidase (MetAP) is present in every cell
and carries out co-translational modification by removing the
N-terminal methionine residue from many nascent proteins.[1]
MetAP is divided into two subtypes, namely types 1 and 2. Eu-
karyotic cells have both subtypes, and prokaryotic cells have
only one. Deletion of either of the two MetAP genes in Saccha-
romyces cerevisiae rendered a slow growth phenotype, and le-
thality was observed if both genes were deleted.[2] Human
type 1 and type 2 MetAPs play important roles in cell prolifera-
tion as well. For example, bengamides, a class of natural prod-
ucts that were isolated from marine sponge,[3] show nanomolar
potency against cancer cell lines (compounds 1 and 2 in
Figure 1),[4,5] and by a proteomics approach, human MetAPs
were identified as the cellular targets.[6] Bengamides arrest cells
at the G1 and G2M phases of the cell cycle,[4,7] and inhibition of
MetAPs led to regulation of c-Src non-receptor tyrosine kinase
activity.[8] An anticancer clinical trial was carried out with a syn-
thetic bengamide derivative, LAF389 (3).[9]
In contrast to eukaryotic cells, most bacteria have a single
MetAP gene, which codes for a type 1 MetAP. Deletion of this
gene is lethal in Escherichia coli[10] and Salmonella typhimuri-
um,[11] suggesting that MetAP is an essential enzyme and a
promising target for the development of broad-spectrum anti-
biotics.[12] We recently demonstrated that a few MetAP inhibi-
tors display significant antibacterial activity toward E. coli and
Bacillus subtilis through MetAP inhibition.[13] Multiple MetAPs
are not common in bacteria, but with more genomic sequen-
ces reported, two or more putative MetAP genes have been
identified in a small number of bacteria. So far, 20 genomes of
mycobacteria have been sequenced, and putative MetAP pro-
teins in each mycobacterial genome, ranging from two to four,
have been identified by sequence analysis. For example, Myco-
bacterium tuberculosis has two MetAP genes (mapA and mapB
in the H37Rv genome and map_1 and map_2 in the CDC1551
genome), both of which belong to type 1 MetAPs with high
homology to E. coli MetAP.
Little is known about the biochemical properties of these
putative MetAPs beyond their sequences. The protein from the
mapB gene of M. tuberculosis, named MtMetAP1c, was purified,
and its structures in both the apo-form and in complex with
methionine were reported.[14] Structural analysis revealed an
SH3 binding motif in its N terminus, and potential ribosome in-
teraction through this motif was proposed to facilitate co-
translational methionine excision.[14] We recently further char-
acterized this enzyme for metal activation and inhibition, and
described three X-ray crystal structures of MtMetAP1c with dif-
ferent inhibitors bound.[15] Interestingly, the other MetAP (from
the mapA gene) of M. tuberculosis, named MtMetAP1a, is short-
er at the N terminus and has no such SH3 binding motif. When
purified, MtMetAP1a could be activated by divalent metal ions
and inhibited by small molecules.[16] The mRNA transcripts of
both MtMetAP1a and MtMetAP1c have been analyzed, and
they show different levels in log phase and stationary phase,
leading to the conclusion that the two MetAPs may perform
[a] Dr. J.-P. Lu, Dr. X.-H. Yuan, Dr. H. Yuan, Dr. W.-L. Wang, Prof. Q.-Z. Ye
Department of Biochemistry and Molecular Biology
Indiana University School of Medicine
635 Barnhill Drive, Indianapolis, IN, 46202 (USA)
Fax: (+1)317-274-4686
E-mail: yeq@iupui.edu
[b] B. Wan, Prof. S. G. Franzblau
Institute for Tuberculosis Research
College of Pharmacy, University of Illinois at Chicago
Chicago, IL, 60612 (USA)
ChemMedChem 2011, 6, 1041 – 1048
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1041

Q.-Z. Ye et al.
MED
X-ray crystal structures of
MtMetAP1c in complex with two
such inhibitors reveal the bind-
ing mode of these bengamide
derivatives and provide the
structural basis for further im-
provement of MetAP inhibitors.
Chemistry
Design of bengamide deriva-
tives
MetAP is
a metal-dependent
enzyme and requires a divalent
metal ion for catalysis such as
CoII, MnII, NiII, or FeII.[21,22] The
active site is
a shallow and
mostly hydrophobic pocket that
accommodates the terminal me-
thionine residue with two metal
ions sitting at the bottom, as re-
vealed in several X-ray struc-
tures.[23] Bengamides are inhibi-
tors of both human type 1 and
type 2 MetAPs with micromolar
Figure 1. Chemical structures of natural bengamides 1 and 2 and their synthetic derivative 3. Compounds 4–10
are the newly designed and synthesized bengamide derivatives.
important functions during different growth phases of M. tu-
berculosis.[17] The MtMetAP1a gene is expressed more in log
phase, whereas the MtMetAP1c gene shows higher levels in
the stationary phase. The high level of MtMetAP1a in log
phase is consistent with the recent studies demonstrating that
knockdown of the MtMetAP1a gene by antisense RNAs results
in decreased viability of M. tuberculosis, whereas knockdown of
the MtMetAP1c gene shows little effect on growth.[18] Although
MtMetAP1c may not play a major role during growth phase, its
high level in the stationary phase suggests its potential role
during dormancy. Based on a comparison of mycobacterial ge-
nomes, it was predicted that both MtMetAP1a and MtMetAP1c
are essential for M. tuberculosis survival in vivo and pathogenic-
ity.[19] M. tuberculosis is unique from most prokaryotes in that it
must evade and propagate within human macrophages. There-
fore, the question is whether MtMetAP1c plays a role in the
survival of M. tuberculosis within mammalian cells.[18] The spe-
cial characteristics of the mycobacterial life cycle may require
more than one MetAP to carry out important co-translational
modifications.
potency,[8] and their bound conformation at the active site was
illustrated by the only reported X-ray structure of the benga-
mide derivative LAF153 (11) in complex with human type 2
MetAP (Figure 2A and Figure 3A, PDB code 1QZY).[6] In the di-
metalated structure, the triol moiety of 11 coordinates with
two CoII ions to form two tetrahedral coordination geometries,
reminiscent of the binding of transition-state inhibitors such as
methionine phosphonate (13, Figure 2B, PDB code 1C23)[24]
and the bestatin analogue (14, PDB code 3MAT).[25] It is possi-
ble that the spatial arrangement of three hydroxy groups in 11
uniquely satisfies the coordination requirement and confers
high-affinity binding. On one side of the triol moiety, the tert-
butylalkene substituent occupies the site reserved for the ter-
minal methionine, and on the other side, the caprolactam ring
beyond the amide bond interacts with residues toward the
opening of the active site pocket (Figure 3A).
Most residues in the active site pockets of MetAPs of both
prokaryotic and eukaryotic origin are conserved, making the
interactions of an inhibitor with the residues and with the
metal ions more predictable. However, variation around the
opening of the pocket is significant because of the various
lengths and residues in the N-terminal extension of human
and bacterial MetAPs.[23] The caprolactam moiety of benga-
mides was believed to be essential for their potency at human
MetAPs, and synthetic efforts in anticancer therapy have been
directed toward modification of this moiety by introducing dif-
ferent functional groups on the ring while maintaining this
seven-membered ring structure.[4,5,26] Our goal is the design of
bengamide derivatives that show strong inhibition of tubercu-
lar MetAPs and weak or no inhibition of human MetAPs. There-
fore, we kept the triol moiety and the tert-butylalkene substitu-
Tuberculosis (TB) is a deadly disease caused by mycobacteri-
al infection, and M. tuberculosis is the major TB pathogen in
humans. Cases of multidrug-resistant and extensively drug-re-
sistant TB are currently happening at an alarming rate.[20] To
overcome such drug resistance, new antibiotics with new
mechanisms of action are urgently required. MetAP is a prom-
ising target for the development of novel drugs against TB-
causing drug-resistant bacteria.[12] Herein we report several
newly designed MetAP inhibitors based on natural benga-
mides which not only inhibit purified tubercular MetAP en-
zymes, but also show initial antitubercular activity. Additionally,
1042
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1041 – 1048

Bengamide MetAP Inhibitors
Synthesis of bengamide deriva-
tives
Commercially available 15 was
converted into the common in-
termediate 16 in several steps
according to a reported synthet-
ic procedure, with minor modifi-
cations (Scheme 1).[27] Coupling
of lactone 16 with various
amines was accomplished by
holding at reflux in isopropanol.
Final deprotection produced the
target compounds 4–10 via in-
termediate 17 in yields ranging
from 9 to 25% for the two steps.
Results and Discussion
Inhibition of MtMetAP1a and
MtMetAP1c
Figure 2. Design of antitubercular bengamides. A) Coordination of 11 with two metal ions inside the active site
pocket of human type 2 MetAP. Color scheme: carbon, yellow (inhibitor) or grey (protein); oxygen, red; nitrogen,
blue. Two CoII ions are shown as green spheres. Coordination between the metal ions and the heteroatoms of the
inhibitor or protein residues is shown as dashed lines. B) Chemical structures of 11, methionine (12), methionine
phosphonate (MetP, 13), and the bestatin analogue 14. C) Stereo view of the binding of newly synthesized 9
(carbon, magenta) and 10 (carbon, cyan) at the active site, in comparison with 11 (carbon, yellow); CoII (green)
and MnII (orange) ions are shown as spheres.
Both
MtMetAP1a
and
MtMetAP1c belong to type 1
MetAPs and share high se-
quence homology.[16] When puri-
fied as apo-enzymes, both can
be activated by divalent metal
ions including CoII, MnII, NiII, and
ent as the bengamide core structure for interaction with the
two metal ions and with the hydrophobic methionine binding
pocket. At the same time, we removed the caprolactam
moiety and replaced it with various amide moieties attached
to the core structure to explore the additional interactions
near the opening of the active site pocket. Initially, seven such
derivatives (4–10, Figure 1) were designed, synthesized, and
evaluated as potential antitubercular agents.
FeII.[15,16] Because inhibitors of a metalloenzyme often interact
directly with the metal ions at the active site, inhibitors can
display marked variation in inhibitory potency toward different
metalloforms of the enzyme, as we have shown for E. coli
MetAP.[28] For therapeutic application, it is critical that these
compounds can effectively inhibit the cellular enzyme in its
native metalloform. We have shown that E. coli MetAP uses FeII
for catalysis in E. coli cells, and inhibitors of E. coli MetAP with
selectivity for the FeII form display antibacterial activity.[29] We
Figure 3. Binding of bengamide derivatives at the active site pocket of human type 2 MetAP and MtMetAP1c. A) Inhibitor 11 (carbon, yellow) with human
type 2 MetAP. B) Inhibitor 9 (carbon, magenta) with MtMetAP1c. C) Inhibitor 10 (carbon, cyan) with MtMetAP1c. CoII (green) and MnII (orange) ions are shown
as spheres. Color Scheme for enzyme surface: carbon, grey; oxygen, red; and nitrogen, blue.
ChemMedChem 2011, 6, 1041 – 1048
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1043

Q.-Z. Ye et al.
MED
Inhibitors of MtMetAP1a and MtMetAP1c were reported only
recently.[15,16,18] Some of these bengamide derivatives with mi-
cromolar potency are promising lead compounds with a new
scaffold for further structural modifications. It is not known
which metal(s) MtMetAP1a and MtMetAP1c use as their native
metal ion. Characterization of the metalloform-selective inhibi-
tion of these bengamides and discovery of metalloform-selec-
tive inhibitors will provide the necessary research tools to
reveal the metal requirement for catalysis of MetAP in M. tuber-
culosis cells. Clearly, therapeutic MetAP inhibitors need to effec-
tively block the function of cellular MetAPs with their native
metal in place.
Scheme 1. Synthesis of target compounds 4–10 via intermediate 17:
a) iPrOH, R-NH2, reflux, 10 h; b) CH3OH, 1n HCl, RT, 4 h. Yield range: 9–25%
over two steps.
X-ray structures of MtMetAP1c in complex with 9 and 10
The previous structure of human type 2 MetAP revealed the
binding mode of 11.[6] To confirm that the newly synthesized
bengamide derivatives bind to tubercular MetAPs in the same
way as designed, we obtained two X-ray crystal structures of
MtMetAP1c in complex with either 9 or 10 at high resolution
(both at 1.25 ꢁ, Figure 2C and Figure 3B and C). The overall
fold of the new MtMetAP1c structures was the same as our
previously obtained structures of
demonstrated that MtMetAP1c uses FeII in E. coli cells,[15] but
the native metalloform for MtMetAP1a and MtMetAP1c in
M. tuberculosis remains unknown. Therefore, we tested the
bengamide derivatives for their inhibitory activity toward
MtMetAP1a and MtMetAP1c activated by each of the four
metal ions (Table 1).
the same enzyme in complex
with other inhibitors,[15] and two
MnII ions and the new benga-
mide inhibitors bound at the
active site pocket. Similar to 11,
both the bengamides 9 and 10
used their triol moieties to coor-
dinate with the two active site
metal ions, and the tert-butylal-
kene chain to occupy the hydro-
phobic S1 site.
Table 1. Inhibition of MetAPs in different metalloforms and growth inhibition of M. tuberculosis and human
K562 cells.
Compd
IC50 [mm]
M. tub. MIC [mm][a]
K562 IC50 [mm]
MtMetAP1a
MtMetAP1c
MABA[b]
LORA[c]
Co
45
Mn
Ni
Fe
67
Co
Mn
Ni
Fe
4
5
6
7
8
68
141
>250 110
>250
>250 0.96
39
52
>250 0.76
50
75
1.3
>250
>250
150
>250
>250
120
179
>250
61
68
201
3.7
0%
0%
15%
48%
29%
122
0%
0%
8%
12%
28%
0%
>333
>333
79.6
88 187
6.0
8.1
31
21
7.9
11
12
11
49
6.9
26
80
140
114
33
5.5
6.7
18
14
4.5
0.40
0.20
96.5
>333
>333
37.8
9
10
0.62
0.54
>250
42
53.9
106
Our strategy in designing the
bengamide derivatives for selec-
tivity was to substitute the cap-
rolactam ring in 11 with various
amide moieties to explore differ-
[a] Minimum inhibitory concentration or percent inhibition at 128 mm. [b] Microplate Alamar Blue assay. [c] Low
oxygen recovery assay. All values in this table are averages of at least triplicate determinations.
All seven derivatives 4–10 showed micromolar or sub-micro-
molar potency at the MnII form of MtMetAP1c. Considering
that the IC50 values are close to the concentration of
MtMetAP1c used (0.5 mm), which was limited by assay sensitivi-
ty, the actual potency for some of them could be higher. All
compounds showed almost no activity against the NiII form of
MtMetAP1c, and their potency at the CoII or FeII forms was also
weak, with compounds 6, 7, 9, and 10 showing relatively
higher potency. Although these synthetic bengamides are
clearly selective for the MnII form of MtMetAP1c, the metallo-
form selectivity for MtMetAP1a was not as noticeable. The in-
hibition of the NiII form of MtMetAP1a was also weaker for all
of the bengamides, but 6, 7, and 10 all showed considerable
potency toward the CoII, MnII, and FeII forms of MtMetAP1a.
Compounds 6, 7, 8, and 10 all have aromatic rings in their
amide moiety; it is possible that these substituents can be ac-
commodated at the active site, and that they provide addition-
al binding interactions.
ent interactions at the opening of the active site pocket.
Indeed, the significant difference in binding was observed at
the amide moiety. While the amide moiety in 9 is shorter and
takes a position similar to that of the caprolactam ring in 11
(Figure 3B), the trimethylphenyl group in 10 makes a sharp
turn, reaches the shallow cavity at the opening unique to
MtMetAP1c, and fits snugly in the cavity (Figure 3C). With this
structural information on binding, additional substitutions on
the phenyl ring could be introduced to explore the interac-
tions with different residues in the cavity in order to improve
both potency and selectivity.
Growth inhibition of M. tuberculosis at replicating state and
non-replicating state
All seven newly synthesized bengamide derivatives were evalu-
ated for their antitubercular activity against M. tuberculosis
strain H37Rv (replicating phenotype, R-TB) by using a micro-
1044
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1041 – 1048

Bengamide MetAP Inhibitors
plate Alamar Blue assay,[30,31] and against M. tuberculosis strain
H37Rv-CA-luxAB (non-replicating persistent phenotype, NRP-TB)
by using a low oxygen recovery assay.[32] Among the com-
pounds tested (Table 1), compound 10 exhibited the best anti-
tubercular activity against both replicating M. tuberculosis
(MIC=50.6 mm; 0.08 mm for rifampin and 0.24 mm for isoniazid)
and non-replicating M. tuberculosis (MIC=107.4 mm; 1.96 mm
for rifampin and >128 mm for isoniazid). Notably, 10 was also
one of the best inhibitors of MtMetAP1a and MtMetAP1c. It is
interesting that the bengamides that displayed antitubercular
activity in the microplate Alamar Blue assay also displayed ac-
tivity in the low oxygen recovery assay (Table 1), possibly due
to inhibition of both MtMetAP1a and MtMetAP1c by these
bengamides in cells.
Experimental Section
General Methods. All chemicals were reagent grade and were
used without further purification. Solvents were of analytical grade.
1H (500 MHz) and 13C (125 MHz) NMR data were obtained on a
Bruker Avance II 500 instrument. Chemical shifts (d) are reported in
ppm, using d=7.26 ppm (CDCl3, 1H NMR) and d=77.23 ppm
(CDCl3, 13C NMR), or d=2.50 ppm ([D6]DMSO, 1H NMR) and d=
40.60 ppm ([D6]DMSO, 13C NMR) as internal standards. Signals are
reported as intervals in cases of multiplets. Signals were abbreviat-
ed as: s, singlet; d, doublet; m, multiplet. All tested compounds
were at least 95% pure on the basis of HPLC–MS using an LTQ Or-
bitrap mass spectrometer (Thermo-Fisher Scientific) coupled to an
Accela HPLC system (Thermo-Fisher Scientific). High-resolution MS
data were obtained on either a Thermo Finnigan MAT-95 XP or a
Waters/Micromass LCT instrument. All reactions were routinely
checked by thin-layer chromatography (TLC) on silica gel 60 F254
(Merck) and visualized with 5% ethanolic phosphomolybdic acid
solution and heat. Organic solutions were dried over anhydrous
MgSO4, filtered, and concentrated with a Bꢂchi rotary evaporator
at reduced pressure. Yields are of purified product and were not
optimized.
Growth inhibition of human K562 cells
In the development of antitubercular bengamide derivatives,
we want to minimize inhibition of human MetAPs to decrease
potential toxicity. To evaluate the antiproliferative effect of
these derivatives, we tested their effect on the growth of
human K562 (leukemia-derived) cells (Table 1). Inhibitors 4, 5,
8, and 9 showed no or weak activity at 333 mm, the highest
concentration tested, and inhibitors 6, 7, and 10 showed low
and reproducible activity. Interestingly, 10 was the best growth
inhibitor of M. tuberculosis, and it was also the most active
against the growth of human cells. Although the initial results
from the limited number of bengamide derivatives are encour-
aging, significant improvement on potency and selectivity is
required.
General procedure for the preparation of compounds 4–10. A
solution of 16 (0.1 g, 0.35 mmol) and an appropriate amine
(0.9 mmol) in iPrOH (5 mL) was stirred at reflux for 10 h. When the
amine was in the form of a HCl salt, 1.4 mmol of 2-ethylhexanoate
was added.[33] The solvent was removed, and the residue was dis-
solved in EtOAc (20 mL), washed with 10% aq. citric acid (2ꢃ
15 mL) and aq. NaHCO3 (20 mL) successively. The organic layer was
dried over MgSO4, filtered, concentrated to give crude acetonide
intermediate 17. The intermediate 17 was dissolved in CH3OH
(3 mL); 3n aq. HCl in CH3OH (1 mL) was added dropwise, and the
mixture was stirred at RT for 4 h. The solution was adjusted to
pH 8–9 with aq. NaHCO3, and the organic solvent was removed.
The residue was diluted with H2O (10 mL) and extracted with
EtOAc (2ꢃ15 mL). The combined organic layers were dried over
MgSO4, concentrated, and purified by column chromatography
(hexane/EtOAc) to give the desired product 4–10.
Conclusions
(2R,3R,4S,5R,E)-3,4,5-Trihydroxy-N-[(1-isobutyrylpiperidin-3-yl)-
methyl]-2-methoxy-8,8-dimethylnon-6-enamide (4). Yield for two
steps: 9%; H NMR (CDCl3, 500 MHz): d=1.06 (s, 9H), 1.13 (m, 6H),
1.53–1.41 (m, 2H), 1.90–1.67 (m, 4H), 2.84–2.75 (m, 2H), 3.46–3.11
(m, 3H), 3.56 (s, 3H), 3.80–3.62 (m, 3H), 3.88–3.86 (m, 1H), 4.20–
4.17 (m, 1H), 5.50–5.42 (m, 1H), 5.86–5.82 (m, 1H), 7.23 ppm (s,
1H); 13C NMR (CDCl3, 125 MHz): d=19.49, 25.13, 31.11, 31.29,
33.02, 35.72, 41.35, 46.37, 60.40, 72.01, 72.52, 74.55, 82.05, 82.60,
99.99, 123.34, 145.63, 172.18, 175.84 ppm; HRMS (ESI-TOF): m/z
451.2784 (calcd for C22H40N2O6Na: 451.2784).
Natural bengamides with demonstrated antiproliferative po-
tency on mammalian cells provide a unique scaffold to devel-
op a new class of MetAP inhibitors for antibacterial and antitu-
bercular therapeutics. Based on analysis of the binding mode
of a bengamide derivative 11 on human type 2 MetAP, we re-
placed the caprolactam moiety in natural bengamides with
various amide moieties to take advantage of differences at the
inhibitor binding pocket, aiming to obtain inhibitors with high
potency toward M. tuberculosis MetAPs and little or no potency
toward human MetAPs. Some of these newly designed and
synthesized bengamide derivatives showed modest but repro-
ducible activity against both replicating and non-replicating
M. tuberculosis. Evaluation of these derivatives on human K562
cells revealed their ability to inhibit growth, indicating their ac-
tivity at mammalian MetAPs has not been completely eliminat-
ed. Importantly, the crystal structures of two such inhibitors
(compounds 9 and 10) in complex with MtMetAP1c have pro-
vided valuable structural information regarding the interaction
of these inhibitors with the active site and will guide the devel-
opment of other bengamide derivatives with better potency
and selectivity toward tubercular MetAPs.
1
(2R,3R,4S,5R,E)-N-{[1-(Cyclopropanecarbonyl)piperidin-3-yl]meth-
yl}-3,4,5-trihydroxy-2-methoxy-8,8-dimethylnon-6-enamide (5).
1
Yield for two steps: 18%; H NMR (CDCl3, 500 MHz): d=0.78–0.77
(m, 2H), 0.97–0.95 (m, 2H), 1.06 (s, 9H), 1.57–1.33 (m, 2H), 1.87–
1.74 (m, 7H), 3.47–3.26 (m, 2H), 3.51 (s, 3H), 3.87–3.78 (m, 2H),
4.30–4.16 (m, 2H), 5.48–5.40 (m, 1H), 5.86–5.85 (m, 1H), 7.29 ppm
(s, 1H); 13C NMR (CDCl3, 125 MHz): d=7.40, 10.92, 24.63, 28.40,
29.42, 33.03, 35.66, 41.35, 45.52, 46.16, 46.55, 60.20, 72.25, 74.66,
82.3, 99.99, 123.29, 145.68, 172.12, 172.27 ppm; HRMS (ESI-TOF):
m/z calcd for C22H38N2O6Na: 449.2628; found: 449.2606.
(2R,3R,4S,5R,E)-N-[3-(Furan-2-yl)-3-phenylpropyl]-3,4,5-trihy-
droxy-2-methoxy-8,8-dimethylnon-6-enamide (6). Yield for two
steps: 16%; 1H NMR (CDCl3, 500 MHz): d=1.01 (s, 9H), 2.15–2.14
(m, 1H), 2.37–2.35 (m, 1H), 3.33–3.27 (m, 2H), 3.47 (s, 3H), 3.57–
ChemMedChem 2011, 6, 1041 – 1048
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1045

Q.-Z. Ye et al.
MED
3.52 (m, 2H), 3.69–3.67 (m, 1H), 3.78–3.77 (m, 1H), 3.99–3.98 (m,
1H), 4.21–4.20 (m, 2H), 5.44–5.39 (dd, 1H), 5.85–5.81 (d, 1H), 6.07
(s, 1H), 6.29 (s, 1H), 6.86 (s, 1H), 7.25–7.22 (m, 3H), 7.33–7.30 ppm
(m, 3H); 13C NMR (CDCl3, 125 MHz): d=14.16, 29.46, 29.70, 31.62,
33.03, 34.25, 37.67, 43.17, 59.89, 72.42, 74.57, 81.13, 105.80, 110.16,
123.20, 127.03, 127.70, 128.76, 141.57, 145.91, 156.68, 172.53 ppm;
HRMS (ESI-TOF): m/z calcd for C25H35NO6Na: 468.2362; found:
468.2357.
10 mm ascorbic acid). For inhibition of MtMetAP1c, each well con-
tained 80 mL assay mixture with 50 mm MOPS (pH 7.5), 100 mm
Met-AMC, apo-enzyme (500 nm), and metal ions (50 mm CoCl2,
250 mm MnCl2, 50 mm NiCl2, or 10 mm FeCl2 plus 10 mm ascorbic
acid). The inhibitors were tested at six or more serially diluted con-
centrations. IC50 values were calculated from nonlinear regression
curve fitting of percent inhibition values as a function of inhibitor
concentration.
(2R,3R,4S,5R,E)-N-[3-(4-Fluorophenyl)-3-(furan-2-yl)propyl]-3,4,5-
trihydroxy-2-methoxy-8,8-dimethylnon-6-enamide (7). Yield for
two steps: 10%; H NMR (CDCl3, 500 MHz): d=1.07 (s, 9H), 2.16–
M. tuberculosis MABA and LORA assays. Minimum inhibitory con-
centrations (MICs) against replicating and non-replicating cultures
of M. tuberculosis were determined by microplate Alamar Blue
assay (MABA)[30,31] and low oxygen recovery assay (LORA),[32] re-
spectively. The former was determined against M. tuberculosis
H37Rv ATCC 27294 (American Type Culture Collection) following
7 days incubation with test samples. The latter was determined
against low oxygen adapted M. tuberculosis H37Rv luxAB carrying a
luciferase reporter gene following 10 days incubation under low
oxygen, followed by 28 h normoxic recovery. Both assays were
conducted in microplate format in 7H12 medium.[30] MIC values are
defined as the lowest concentration effecting a decrease of ꢀ0%
in fluorescence (MABA) or luminescence (LORA) relative to untreat-
ed controls.
1
2.13 (m, 1H), 2.37–2.35 (m, 1H), 3.04 (s, 1H), 3.35–3.27 (m, 2H),
3.52–3.51 (d, 1.9 Hz, 3H), 3.61–3.59 (m, 1H), 3.72–3.70 (dd, 1H),
3.82–3.96 (m, 1H), 4.02–3.99 (m, 1H), 4.17–4.15 (m, 1H), 4.24–4.22
(m, 1H), 5.47–5.42 (dd, 1H), 5.88–5.84 (d, 1H), 6.10–6.09 (m, 1H),
6.33–6.32 (m, 1H), 6.90–6.80 (m, 1H), 7.04–7.01 (m, 2H), 7.24–7.23
(m, 2H), 7.35 ppm (d, 1H); 13C NMR (CDCl3, 125 MHz): d=29.42,
33.04, 34.38, 37.58, 42.38, 60.20, 72.38, 81.19, 99.99, 105.88, 110.21,
115.50, 123.18, 129.16, 141.77, 145.99, 160.83, 162.78, 172.52 ppm;
HRMS (ESI-TOF): m/z calcd for C25H34FNO6Na: 486.2268; found:
486.2253.
(2R,3R,4S,5R,E)-N-(2,3-Dihydro-1H-inden-2-yl)-3,4,5-trihydroxy-2-
methoxy-8,8-dimethylnon-6-enamide (8). Yield for two steps:
Crystallization and data collection. Crystals of the enzyme–inhibi-
tor complexes were obtained independently by a hanging-drop
vapor-diffusion method at room temperature. Each of the inhibi-
tors (9 or 10; 100 mm in DMSO) was added to concentrated meta-
lated enzyme (10 mgmLÀ1, 0.32 mm protein; 2 mm metal) in
50 mm Tris, pH 8.0, 150 mm NaCl, and the molar ratio of inhibitor
to MtMetAP1c was 5:1 or 10:1. The enzyme–inhibitor mixture was
mixed with a reservoir buffer at a 1:1 ratio. The reservoir buffer
was 100 mm Bis-Tris (pH 5.5), 1.3m (NH4)2SO4, and 10% glycerol.
Diffraction data were collected at the Advanced Photon Source, Ar-
gonne National Laboratory (beamlines 19ID and 19BM) and were
processed with HKL3000.[34] Both crystals belong to space group
P63. One molecule is in the asymmetric unit.
1
25%; H NMR (CDCl3, 500 MHz): d=1.05 (s, 9H), 2.89–2.83 (m, 2H),
3.41–3.36 (m, 2H), 3.49 (s, 3H), 3.61–3.60 (d, 1H), 3.75–3.74 (d, 1H),
3.83–3.82 (d, 1H), 4.26–4.23 (m, 1H), 4.79–4.77 (m, 1H), 5.46–5.42
(dd, 1H), 5.88–5.84 (d, 1H), 7.08–7.07 (d, 1H), 7.24–7.21 ppm (m,
4H); 13C NMR (CDCl3, 125 MHz): d=29.42, 33.05, 40.00, 50.14,
59.79, 72.36, 72.44, 74.59, 81.19, 123.19, 124.80, 126.97, 140.40,
145.98, 172.29 ppm; HRMS (ESI-TOF): m/z calcd for C21H31NO5Na:
400.2100; found: 400.2082.
(2R,3R,4S,5R,E)-N-(2-Amino-2-oxoethyl)-3,4,5-trihydroxy-2-me-
thoxy-8,8-dimethylnon-6-enamide (9). Yield for two steps: 14%;
1H NMR (CDCl3, 500 MHz): d=1.01 (s, 9H), 3.51 (s, 2H), 3.79 (s,
3H), 3.83–3.92 (m, 2H), 4.16–4.21 (m, 1H), 4.25–4.30 (m, 1H), 5.37–
5.45 (m, 1H), 5.83–6.01 ppm (m, 1H); 13C NMR (CDCl3, 125 MHz):
d=29.28, 42.72, 60.00, 66.29, 73.00, 74.30, 81.80, 99.71, 121.27,
145.62, 169.25, 170.71 ppm; HRMS (ESI-TOF) m/z calcd for
C14H26N2O6Na: 341.1689; found: 341.1689.
Structural solution and refinement. Structures were solved by
molecular replacement with MolRep[35] in CCP4[36] with CCP4i inter-
face,[37] using our previously published MtMetAP1c structure (PDB
code 3IU7)[15] as the search model. The structure was refined with
REFMAC5[38] with iterative model building using WinCoot.[39] The re-
finement was monitored with 5% of the reflections set aside for
Rfree factor analysis throughout the entire refinement process. Elec-
tron density was clear for almost all residues, and residues from
the second (P2) in the native protein to the end (L285) were mod-
eled. Comparison of structures and generation of structural draw-
ings were carried out by using PyMOL.[40] Statistical parameters in
data collection and structural refinement are listed in Table 2.
Atomic coordinates and structure factors for the two structures
have been deposited in the RCSB Protein Data Bank.
(2R,3R,4S,5R,E)-3,4,5-Trihydroxy-N-[2-(mesityloxy)ethyl]-2-me-
thoxy-8,8-dimethylnon-6-enamide (10). Yield for two steps: 15%;
1H NMR (CDCl3, 500 MHz): d=1.06 (s, 9H), 2.22 (s, 9H), 3.59 (s,
3H), 3.63–3.88 (m, 7H), 4.25–4.28 (m, 1H), 5.44–5.48 (dd, 1H), 5.85–
5.88 (d, 1H), 6.85 (s, 2H), 7.53 ppm (s, 1H); 13C NMR (CDCl3,
125 MHz): d=16.15, 20.65, 29.44, 33.04, 39.46, 60.15, 70.10, 72.39,
72.67, 74.59, 80.89, 123.21, 129.61, 130.15, 133.66, 145.88, 152.74,
172.81 ppm; HRMS (ESI-TOF): m/z calcd for C23H37NO6Na: 446.2519;
found: 446.2505.
MtMetAP1a and MtMetAP1c inhibition assays. MtMetAP1a and
MtMetAP1c were expressed in E. coli and purified as apo-enzymes
as previously described.[15,16] Both can be activated by CoII, MnII, NiII,
and FeII instantly. Enzymatic activity was monitored by fluorescence
(lex =360 nm, lem =460 nm) on a Spectramax Gemini XPS plate
reader (Molecular Devices, Sunnyvale, CA, USA) following hydroly-
sis of the fluorogenic substrate, methionyl aminomethylcoumarin
(Met-AMC), at room temperature as described.[22] All kinetics ex-
periments were carried out on 384-well plates. For inhibition of
MtMetAP1a, each well contained 80 mL assay mixture with 50 mm
MOPS (pH 7.5), 100 mm Met-AMC, apo-enzyme and metal ions
(50 nm enzyme, 50 mm CoCl2; 200 nm enzyme, 250 mm MnCl2;
12.5 nm enzyme, 50 mm NiCl2; or 50 nm enzyme, 10 mm FeCl2,
Human K562 growth inhibition assay. K562 cells from ATCC were
cultured as suspension in RPMI 1640 medium, containing 10%
newborn calf serum. Growth inhibition assays were carried out in a
total volume of 120 mL in each well on 96-well white plates. Com-
pounds were serially (twofold) diluted to 12 concentrations, and
cells were seeded at 12000 cells per well by dispensing suspended
K562 cells in growth medium with an eight-channel automated
MultiDrop liquid dispenser. The plates were incubated at 378C in a
humidified 5% CO2 incubator for 48 h. A modified luciferase activi-
ty assay was used to monitor cell growth. A mixture of luciferase,
luciferin, Triton X-100, and MgCl2 in a total volume of 80 mL was
added to the cells in each well, and luminescence was determined
1046
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1041 – 1048

Bengamide MetAP Inhibitors
[4] F. R. Kinder, Jr., R. W. Versace, K. W. Bair, J. M. Bontempo, D. Cesarz, S.
Chen, P. Crews, A. M. Czuchta, C. T. Jagoe, Y. Mou, R. Nemzek, P. E. Phil-
lips, L. D. Tran, R. M. Wang, S. Weltchek, S. Zabludoff, J. Med. Chem.
2001, 44, 3692–3699.
Table 2. X-ray crystallographic data collection and refinement statistics.
Parameter
9
10
[5] Z. Thale, F. R. Kinder, K. W. Bair, J. Bontempo, A. M. Czuchta, R. W. Ver-
sace, P. E. Phillips, M. L. Sanders, S. Wattanasin, P. Crews, J. Org. Chem.
2001, 66, 1733–1741.
[6] H. Towbin, K. W. Bair, J. A. DeCaprio, M. J. Eck, S. Kim, F. R. Kinder, A. Mo-
rollo, D. R. Mueller, P. Schindler, H. K. Song, J. van Oostrum, R. W. Ver-
sace, H. Voshol, J. Wood, S. Zabludoff, P. E. Phillips, J. Biol. Chem. 2003,
278, 52964–52971.
Space group:
a [ꢁ]
b [ꢁ]
c [ꢁ]
a [8]
P63
P63
106.0
106.0
50.3
90
90
120
106.0
106.0
50.4
90
90
120
b [8]
g [8]
Data Collection
Resolution range [ꢁ]
Collected reflections
Unique reflections
Completeness [%]
I/s (I)
[7] P. E. Phillips, K. W. Bair, J. Bontempo, P. Crews, Czuchta, A. M. , F. R.
Kinder, A. Vattay, R. W. Versace, B. Wang, J. Wang, A. Wood, S. Zabludoff,
Proc. Am. Assoc. Cancer Res. 2000, 41, 59.
[8] X. Hu, Y. Dang, K. Tenney, P. Crews, C. W. Tsai, K. M. Sixt, P. A. Cole, J. O.
Liu, Chem. Biol. 2007, 14, 764–774.
50–1.25 (1.27–1.25)[a]
559494
85942
97.0 (73.9)
40.0 (1.7)
50–1.25 (1.27–1.25)
633834
89260
100 (100)
36.8 (3.3)
5.7 (52.5)
[9] H. Dumez, H. Gall, R. Capdeville, C. Dutreix, A. T. van Oosterom, G. Giac-
cone, Anticancer Drugs 2007, 18, 219–225.
Rmerge [%]
5.3 (47.5)
Refinement Statistics
R [%]
Rfree [%]
RMSD bonds [ꢁ]
RMSD angles [8]
No. solvent molecules
B protein [ꢁ2]
B inhibitor [ꢁ2]
B water [ꢁ2]
[10] S. Y. Chang, E. C. McGary, S. Chang, J. Bacteriol. 1989, 171, 4071–4072.
[11] C. G. Miller, A. M. Kukral, J. L. Miller, N. R. Movva, J. Bacteriol. 1989, 171,
5215–5217.
[12] M. D. Vaughan, P. B. Sampson, J. F. Honek, Curr. Med. Chem. 2002, 9,
385–409.
[13] a) S. C. Chai, Q. Z. Ye, Bioorg. Med. Chem. Lett. 2010, 20, 2129–2132;
b) W. L. Wang, S. C. Chai, Q. Z. Ye, Bioorg. Med. Chem. Lett. 2009, 19,
1080–1083; c) W. L. Wang, S. C. Chai, M. Huang, H. Z. He, T. D. Hurley,
Q. Z. Ye, J. Med. Chem. 2008, 51, 6110–6120.
16.8
18.1
0.032
2.49
211
13.1
11.7
19.4
16.9
18.8
0.033
2.51
223
12.2
10.5
19.0
[14] A. Addlagatta, M. L. Quillin, O. Omotoso, J. O. Liu, B. W. Matthews, Bio-
chemistry 2005, 44, 7166–7174.
[a] Values given in parentheses correspond to the outer shell of data.
[15] J. P. Lu, S. C. Chai, Q. Z. Ye, J. Med. Chem. 2010, 53, 1329–1337.
[16] J. P. Lu, Q. Z. Ye, Bioorg. Med. Chem. Lett. 2010, 20, 2776–2779.
[17] X. Zhang, S. Chen, Z. Hu, L. Zhang, H. Wang, Curr. Microbiol. 2009, 59,
520–525.
[18] O. Olaleye, T. R. Raghunand, S. Bhat, J. He, S. Tyagi, G. Lamichhane, P.
Gu, J. Zhou, Y. Zhang, J. Grosset, W. R. Bishai, J. O. Liu, Chem. Biol. 2010,
17, 86–97.
on a Spectramax Gemini or Spectramax M5 plate reader. The IC50
values were calculated from nonlinear regression curve fitting of
percent inhibition values as a function of inhibitor concentration.
Abbreviations: MetAP, methionine aminopeptidase; MtMetAP1a,
M. tuberculosis methionine aminopeptidase type 1a; MtMetAP1c,
M. tuberculosis methionine aminopeptidase type 1c; Met-AMC, me-
thionyl aminomethylcoumarin; TB, tuberculosis.
[19] M. L. Ribeiro-Guimaraes, M. C. Pessolani, Microb. Pathog. 2007, 43, 173–
178.
[20] A. S. Fauci, J. Inf. Dis. 2008, 197, 1493–1498.
[21] V. M. D’Souza, R. C. Holz, Biochemistry 1999, 38, 11079–11085.
[22] J. Y. Li, L. L. Chen, Y. M. Cui, Q. L. Luo, J. Li, F. J. Nan, Q. Z. Ye, Biochem.
Biophys. Res. Commun. 2003, 307, 172–179.
[23] W. T. Lowther, B. W. Matthews, Biochim. Biophys. Acta 2000, 1477, 157–
167.
PDB ID codes: Coordinates and structure factors for MtMetAP1c
complexed to compounds 9 and 10 have been deposited in the
RCSB Protein Data Bank (PDB) under access codes 3PKB and 3PKA,
respectively.
[24] W. T. Lowther, Y. Zhang, P. B. Sampson, J. F. Honek, B. W. Matthews, Bio-
chemistry 1999, 38, 14810–14819.
[25] W. T. Lowther, A. M. Orville, D. T. Madden, S. Lim, D. H. Rich, B. W. Mat-
thews, Biochemistry 1999, 38, 7678–7688.
Acknowledgements
[26] a) G. Liu, Y. M. Ma, W. Y. Tai, C. M. Xie, Y. L. Li, J. Li, F. J. Nan, ChemMed-
Chem 2008, 3, 74–78; b) F. R. Kinder, Jr., S. Wattanasin, R. W. Versace,
K. W. Bair, J. Bontempo, M. A. Green, Y. J. Lu, H. R. Marepalli, P. E. Phillips,
D. Roche, L. D. Tran, R. Wang, L. Waykole, D. D. Xu, S. Zabludoff, J. Org.
Chem. 2001, 66, 2118–2122.
[27] D. D. Xu, L. Waykole, J. V. Calienni, L. Ciszewski, G. T. Lee, W. Liu, J.
Szewczyk, K. Vargas, K. Prasad, O. Repic, T. J. Blacklock, Org. Process Res.
Dev. 2003, 7, 856–865.
[28] Q. Z. Ye, S. X. Xie, M. Huang, W. J. Huang, J. P. Lu, Z. Q. Ma, J. Am. Chem.
Soc. 2004, 126, 13940–13941.
[29] S. C. Chai, W. L. Wang, Q. Z. Ye, J. Biol. Chem. 2008, 283, 26879–26885.
[30] K. Falzari, Z. Zhu, D. Pan, H. Liu, P. Hongmanee, S. G. Franzblau, Antimi-
crob. Agents Chemother. 2005, 49, 1447–1454.
This work was supported by National Institutes of Health Re-
search Grants R01 AI065898 and R56 AI065898, and by a Re-
search Support Fund Grant and a Biomedical Research Grant
from Indiana University (to Q.-Z.Y.). We thank the staffs at the
Structural Biology Center of the Advanced Photon Source, Ar-
gonne National Laboratory (beamlines 19ID and 19BM) for assis-
tance with X-ray diffraction data collection.
Keywords: antibiotics
·
drug discovery
·
inhibitors
·
[31] L. Collins, S. G. Franzblau, Antimicrob. Agents Chemother. 1997, 41,
1004–1009.
hydrolases · metalloenzymes
[32] S. H. Cho, S. Warit, B. Wan, C. H. Hwang, G. F. Pauli, S. G. Franzblau, Anti-
microb. Agents Chemother. 2007, 51, 1380–1385.
[33] W. Liu, D. D. Xu, O. Repic, T. J. Blacklock, Tetrahedron Lett. 2001, 42,
2439–2441.
[1] R. A. Bradshaw, W. W. Brickey, K. W. Walker, Trends Biochem. Sci. 1998,
23, 263–267.
[34] W. Minor, M. Cymborowski, Z. Otwinowski, M. Chruszcz, Acta Crystallogr.
Sect. D 2006, 62, 859–866.
[35] A. Vagin, A. Teplyakov, J. Appl. Crystallogr. 1997, 30, 1022–1025.
[2] X. Li, Y. H. Chang, Proc. Natl. Acad. Sci. USA 1995, 92, 12357–12361.
[3] E. Quinoa, M. Adamczeski, P. Crews, G. J. Bakus, J. Org. Chem. 1986, 51,
4494–4497.
ChemMedChem 2011, 6, 1041 – 1048
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1047

Q.-Z. Ye et al.
MED
[36] Collaborative Computational Project Number 4, Acta Crystallogr. 1994,
D50, 760–763.
[40] The PyMOL Molecular Graphics System, Version 1.3, Schrçdinger, LLC
(2002): http://www.pymol.org (accessed March 14, 2011).
[37] E. Potterton, P. Briggs, M. Turkenburg, E. Dodson, Acta Crystallogr. D
Biol. Crystallogr. 2003, 59, 1131–1137.
Received: January 7, 2011
Revised: March 1, 2011
[38] G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallogr. Sect. D
1997, 53, 240–255.
[39] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D 2004, 60, 2126–2132.
Published online on April 4, 2011
1048
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1041 – 1048