Among the >120 modified ribonucleosides in the prokaryotic epitranscriptome, many tRNA modifications are critical to bacterial survival, which makes their synthetic enzymes ideal targets for antibiotic development. Here we performed a structure-based design of inhibitors of tRNA-(N1G37) methyltransferase, TrmD, which is an essential enzyme in many bacterial pathogens. On the basis of crystal structures of TrmDs from Pseudomonas aeruginosa and Mycobacterium tuberculosis, we synthesized a series of thienopyrimidinone derivatives with nanomolar potency against TrmD in vitro and discovered a novel active site conformational change triggered by inhibitor binding. This tyrosine-flipping mechanism is uniquely found in P. aeruginosa TrmD and renders the enzyme inaccessible to the cofactor S-adenosyl-l-methionine (SAM) and probably to the substrate tRNA. Biophysical and biochemical structure-activity relationship studies provided insights into the mechanisms underlying the potency of thienopyrimidinones as TrmD inhibitors, with several derivatives found to be active against Gram-positive and mycobacterial pathogens. These results lay a foundation for further development of TrmD inhibitors as antimicrobial agents.
Among the >120 modified ribonucleosides in the prokaryotic epitranscriptome, many tRNA modifications are critical to bacterial survival, which makes their synthetic enzymes ideal targets for antibiotic development. Here we performed a structure-based design of inhibitors of tRNA-(N1G37) methyltransferase, TrmD, which is an essential enzyme in many bacterial pathogens. On the basis of crystal structures of TrmDs from Pseudomonas aeruginosa and Mycobacterium tuberculosis, we synthesized a series of thienopyrimidinone derivatives with nanomolar potency against TrmD in vitro and discovered a novel active site conformational change triggered by inhibitor binding. Thistyrosine-flipping mechanism is uniquely found in P. aeruginosa TrmD and renders the enzyme inaccessible to the cofactor S-adenosyl-l-methionine (SAM) and probably to the substrate tRNA. Biophysical and biochemical structure-activity relationship studies provided insights into the mechanisms underlying the potency of thienopyrimidinones as TrmD inhibitors, with several derivatives found to be active against Gram-positive and mycobacterial pathogens. These results lay a foundation for further development of TrmD inhibitors as antimicrobial agents.
The emergence of antibiotic
resistance in bacterial pathogens on
a global scale and the lack of new antibiotics represent a crisis
with significant societal and economic impact. This is illustrated
by >2 million annual cases multidrug resistant bacterial infections,
which result in 23000 deaths.[1,2]Mycobacterium
tuberculosis (Mtb) and Pseudomonas
aeruginosa (Pa) are examples of major human pathogens
for which drug resistance is emerging as a serious public health problem:[1] multidrug resistant Mtb and Pa cause >10000
and
∼6700 deaths per year, respectively. Pa is the cause of the
most common hospital-acquired infection among the immunocompromised,
the elderly, the chronically ill, and patients with in-dwelling medical
devices such as catheters, nasogastric tubes, and drains. These two
pathogens illustrate the point that new antibiotics, particularly
those that avoid resistance mechanisms and are aimed at novel targets,
are urgently needed to alleviate the current antibiotic crisis.Post-transcriptional ribonucleotide modifications of RNA, especially
tRNA, play critical roles in translation in all organisms.[3−7] In addition to the essentiality of some of the enzymes catalyzing
these modifications for growth, studies with bacteria,[4−8] yeast,[3,5] and parasites[9] have demonstrated that many tRNA modifications are critical in the
cell stress response by facilitating selective translation of proteins
critical to surviving the stress. Loss of the ability to synthesize
these tRNA modifications renders bacteria susceptible to killing by
the immune response and other environmental stresses.[4,8] Given their role in bacterial cell survival, these critical tRNA
modification synthesis enzymes constitute attractive targets for antibiotic
development.The bacterial tRNA (guanine37-N1)-methyltransferase
(EC2.1.1.228; TrmD) catalyzes methyl transfer from S-adenosyl-l-methionine (SAM) to the guanine N1 at nucleotide position 37 in a subset of bacterial tRNA isoacceptors (Supporting
Information, Figure S1) and has proven
to be an essential enzyme in most bacterial species.[10−14] While of central importance to mammalian cell health,[13] Trm5, the functional homologue of TrmD in eukaryotes,
has dissimilar active sites and different binding modes of SAM than
TrmD.[15,16] These observations suggest that TrmD is
an attractive antibiotic target, a conclusion supported by the efforts
of several groups to develop TrmD inhibitors.[15,17,18] We recently developed a radioactivity-free
bioluminescence-based high-throughput screening (HTS) assay that identified
a series of novel TrmD inhibitors.[17] In
2013, Hill et al.[15] used X-ray crystallography-guided
fragment screening to develop a series of SAM-competitive TrmD inhibitors
against a broad range of bacterial isozymes. These inhibitors showed
minimal antibacterial activity, presumably due to poor cell permeability.[15] Interestingly, one of these inhibitors, compound 51 (AZ51), showed remarkable inhibitory potency toward all
the tested TrmD isozymes and subnanomolar activity against P. aeruginosa TrmD (PaTrmD). Despite
this potent TrmD inhibitor activity, AZ51 lacked antibacterial activity.Building on this work and the TrmD inhibitor scaffolds identified
in aforementioned HTS,[17] we sought here
to understand the structural basis for the potent TrmD inhibition
caused by the thienopyrimidinone compounds, with the goal of refining
the structures for antibiotic activity. Crystal structures of TrmD
from Pa and Mtb in complex with
AZ51 revealed conformational changes unique to the Gram-negative bacterial
TrmD. On the basis of these structures, we then used the thienopyrimidinone
scaffold (Figure )
to design and synthesize a series of 33 derivatives with the goal
of improved potency and antibacterial activity. Structure–activity
relationship (SAR) studies defined critical features of the thienopyrimidinone
that drive enzyme inhibition potency as well as antibacterial activity.
Figure 1
Structure
of TrmD inhibitors based on the thienopyrimidinone scaffold
(A) and their O6-derivatives (B).
Structure
of TrmD inhibitors based on the thienopyrimidinone scaffold
(A) and their O6-derivatives (B).
Results
AZ51 Has
Broad-Spectrum TrmD Inhibition Activity
Previously,
Hill et al. discovered an interesting inhibition mechanism where one
of the thienopyrimidinone derivatives (compound 38)[15] ordered the position of the lid domain of Haemophilus influenzae TrmD (HiTrmD)
via its terminal imidazole substituent. The lid domain covered the
active site of TrmD and was essential for tRNA binding.[19,20] AZ51 is another analogue derived from the thienopyrimidinone scaffold
(Figure A; Table ) and is generally
more potent than other thienopyrimidinone inhibitors across a variety
of Gram-negative and Gram-positive isozymes,[15] in particular to PaTrmD. Thus, it raised the question
about the potential unique inhibition mechanism of AZ51. We confirmed
that it inhibited PaTrmD with nanomolar potency (IC50 180 ± 20 nM; Figure A, Table ) using a different, bioluminescence-based assay.[17] In addition, we determined the binding affinities of AZ51
to three bacterial TrmDs by surface plasmon resonance (SPR; Table ). Interestingly,
AZ51 has much stronger binding affinity with Gram-negative bacterial
TrmDs than mycobacterial and Gram-positive bacterial TrmDs (100- to
300-fold lower Kd; Table ). For PaTrmD and Staphylococus aureus TrmD (SaTrmD),
this relationship paralleled differences in TrmD inhibition potency
by AZ51: PaTrmD Kd 91
nM and IC50180 nM here, 39 nM Hill et al.;[15] and SaTrmD Kd 27500 nM and IC50 1200 nM.[15] Because the basis for these striking differences in AZ51 binding
energetics and inhibition potency for the different bacterial classes
may lie in different structures of the AZ51–enzyme complexes,
we undertook to solve the crystal structures of AZ51 with PaTrmD and MtbTrmD.
Table 1
Structures and IC50 Values
for Thienopyrimidinone PaTrmD Inhibitorsa
Biochemical potency
is represented
as IC50 value (μM); values are mean ± SD for
at least two independent experiments done in duplicate.
The listed pharmacophores replace
the group N(R1)(R2).
Figure 2
Crystal structures
of AZ51 with PaTrmD and MtbTrmD.
(A) The IC50 values determined for AZ51
at tRNA concentrations of 1.5 μM (●), 4.5 μM (□),
and 15 μM (▲), respectively, are 0.18 ± 0.03, 0.33
± 0.04, and 0.84 ± 0.09 μM. Each data point represents
an average of duplicate experiments with error bars indicated as SD.
(B,D) Close-up views of catalytic sites of PaTrmD–AZ51
(B; PDB 6JOE) and MtbTrmD–AZ51 (D; PDB 6JOF) showing the binding
modes of inhibitor AZ51, respectively. Polypeptide chains are shown
as cartoons (chain A in gray and chain B in cyan), whereas key interacting
residues are shown as sticks. AZ51 are shown as sticks with an 2Fo −Fc electron
density (gray) map contoured at 1.0σ. Potential interactions
involved in AZ51 binding are indicated by broken lines (black, hydrogen
bonds; green, stacking interactions). (C) Inhibitor AZ51 binding at
catalytic site induces wall-loop conformational change and thereafter
blocks substrate tRNA (G37) binding. SAM-bound PaTrmD (PDB 5WYQ) and tRNA-bound H. influenzae TrmD (PDB 4YVI) were superimposed onto AZ51-bound PaTrmD, respectively.
Polypeptide chains of PaTrmD–AZ51 are shown
as surface, while the bound AZ51 (yellow) and side chain flipped residue
Tyr120 (gray) are shown as sticks. PaTrmD–SAM
structure (light pink) displays SAM and residue Tyr120 in sticks representation.
G37of tRNA substrate in HiTrmD–tRNA structure
is shown as sticks in orange. Polypeptide chains of PaTrmD–SAM and HiTrmD–tRNA are omitted
for clarity. (E) The superposition of catalytic sites of PaTrmD–AZ51 and MtbTrmD–AZ51. The inhibitor
AZ51 and residues in different positions from PaTrmD
and MtbTrmD are indicated as sticks.
Table 2
Binding Kinetics of TrmD Ligandsa
ligand
KD (μM)
kon(M–1 s–1)
koff (s–1)
PaTrmDc
SAHb
5.9
55500 ± 479
0.326 ± 0.002
AZ51
0.091
128000 ± 318
0.0117 ± 0.0000
15
0.0016
13500000 ± 87800
0.0218 ± 0.0001
MtbTrmDd
SAH
10.1
7700 ± 14
0.0775 ± 0. 0001
AZ51
10.4
20100 ± 80
0.209 ± 0.0004
15
0.19
645000 ± 3030
0.120 ± 0.0003
SaTrmDe
SAH
3.3
279000 ± 36000
0.92 ± 0.03
AZ51
27.5
59900 ± 566
1.65 ± 0.01
15
3.1
2940000 ± 241000
9.15 ± 0.73
Binding data determined
by surface
plasmon resonance (SPR) as described in Experimental Procedures.
S-Adenosylhomocysteine;
Gram-negative bacterial TrmD;
Mycobacterial TrmD.
Gram-positive bacterial TrmD.
Biochemical potency
is represented
as IC50 value (μM); values are mean ± SD for
at least two independent experiments done in duplicate.The listed pharmacophores replace
the group N(R1)(R2).Crystal structures
of AZ51 with PaTrmD and MtbTrmD.
(A) The IC50 values determined for AZ51
at tRNA concentrations of 1.5 μM (●), 4.5 μM (□),
and 15 μM (▲), respectively, are 0.18 ± 0.03, 0.33
± 0.04, and 0.84 ± 0.09 μM. Each data point represents
an average of duplicate experiments with error bars indicated as SD.
(B,D) Close-up views of catalytic sites of PaTrmD–AZ51
(B; PDB 6JOE) and MtbTrmD–AZ51 (D; PDB 6JOF) showing the binding
modes of inhibitor AZ51, respectively. Polypeptide chains are shown
as cartoons (chain A in gray and chain B in cyan), whereas key interacting
residues are shown as sticks. AZ51 are shown as sticks with an 2Fo −Fc electron
density (gray) map contoured at 1.0σ. Potential interactions
involved in AZ51 binding are indicated by broken lines (black, hydrogen
bonds; green, stacking interactions). (C) Inhibitor AZ51 binding at
catalytic site induces wall-loop conformational change and thereafter
blocks substrate tRNA (G37) binding. SAM-bound PaTrmD (PDB 5WYQ) and tRNA-bound H. influenzae TrmD (PDB 4YVI) were superimposed onto AZ51-bound PaTrmD, respectively.
Polypeptide chains of PaTrmD–AZ51 are shown
as surface, while the bound AZ51 (yellow) and side chain flipped residue
Tyr120 (gray) are shown as sticks. PaTrmD–SAM
structure (light pink) displays SAM and residue Tyr120 in sticks representation.
G37of tRNA substrate in HiTrmD–tRNA structure
is shown as sticks in orange. Polypeptide chains of PaTrmD–SAM and HiTrmD–tRNA are omitted
for clarity. (E) The superposition of catalytic sites of PaTrmD–AZ51 and MtbTrmD–AZ51. The inhibitor
AZ51 and residues in different positions from PaTrmD
and MtbTrmD are indicated as sticks.Binding data determined
by surface
plasmon resonance (SPR) as described in Experimental Procedures.S-Adenosylhomocysteine;Gram-negative bacterial TrmD;Mycobacterial TrmD.Gram-positive bacterial TrmD.
AZ51 Induces a Unique Conformational Change
of a Wall-Loop Residue
in PaTrmD
To gain insight into the molecular mechanism used
by AZ51 for PaTrmD inhibition, we soaked a PaTrmD–SAM cocrystal with AZ51 and determined the
structure at a resolution of 2.21 Å using a crystal form having
a complete PaTrmD dimer in the asymmetric unit (Table ). The electron density
of AZ51 is clearly visible at one active site of the PaTrmD, whereas the copurifying SAM cofactor still occupies the other
active site (Figure B). The observation that cocrystal soaking with the competitive inhibitor
led to the displacement of SAM at only one of the two active sites
suggests that either the binding affinity to SAM or the rigidity of
binding sites is different for the two active sites within a TrmD
biological dimer. However, as the difference in AZ51 binding of two
active sites are found in the protein crystal by the soaking method
where the structural dynamics of the crystal form are restricted,
we cannot exclude the possibility that the difference in AZ51 binding
is simply due to the nature of the method. Similar to SAM binding
(Supporting Information, Figure S2), inhibitor
AZ51 is mainly stabilized by three active-site loops in PaTrmD (named the “cover loop”, “bottom loop”,
and “wall loop”) (Figure B): the thienopyrimidinone ring is tightly bound in
the adenine pocket, is hydrogen-bonded to residues Ile138, Tyr141,
and Leu143, and also forms stacking interactions with residues Pro94
and Leu143. The phenyl ring is locked in place by stacking interactions
with residues Pro94 and Tyr120, while the piperidine ring forms stacking
interactions with residue Tyr120 and is potentially hydrogen-bonded
to residue Asp182 by its protonated nitrogen. The terminal amine (i.e.,
the −NH2 hydrogen bond donor incorporated onto the
piperidine ring) interacts with the carboxylate of Glu121, a highly
conserved residue among TrmD isozymes (Supporting Information, Figure S3).
Table 3
Data Collection and
Refinement Statistics
for TrmD Atructuresa
PaTrmD–AZ51
PaTrmD–11
PaTrmD–15
ApoMtb–TrmD
MtbTrmD–SAH
MtbTrmD–AZ51
MtbTrmD–12
MtbTrmD–15
PDB ID
6JOE
5ZHM
5ZHN
5ZHI
5ZHJ
6JOF
5ZHK
5ZHL
data collection
space group
P3221
P3221
P3221
P21
C121
C121
C121
C121
cell dimensions
a, b, c (Å)
85.50, 85.50,
147.54
84.50, 84.50, 147.27
84.67, 84.67,
148.56
44.17, 113.07, 44.21
72.96, 50.76,
53.31
73.07, 51.38, 57.95
73.09, 50.80,
58.081
73.69, 50.23, 57.94
α, β,
γ (deg)
90.00, 90.00, 120.00
90.00,
90.00, 120.00
90.00, 90.00, 120.00
90.00,
110.75, 90.00
90.00, 95.10, 90.00
90.00,
90.18, 90.00
90.00, 90.56, 90.00
90.00,
90.95, 90.00
solvent content (%)
52
51
52
38
35
41
40
40
resolution (Å)
42.75–2.21
49.09–2.76
42.33–2.65
41.30–2.20
53.10–1.75
42.03–2.20
58.08–2.30
41.50–2.25
no. of reflns
267240 (21374)
167650 (24471)
201645 (27032)
72052 (5534)
55961 (8132)
44682 (3655)
23534 (3432)
32518 (4380)
no. of unique reflns
32130 (2724)
16240 (2335)
18516 (2392)
19717 (1588)
18952 (2704)
10831 (917)
8972 (1287)
9936 (1397)
Wilson B-factor (Å2)
56.2
70.2
82.0
16.8
16.2
20.5
19.7
24.3
Rmerge (%)
5.1 (84.3)
8.8 (99.3)
5.7 (131.3)
5.5 (22.6)
9.0 (40.8)
4.8 (17.0)
11.5 (43.8)
5.8 (23.8)
CC(1/2)
0.999 (0.874)
0.998 (0.917)
0.999 (0.865)
0.997 (0.929)
0.994 (0.721)
0.998 (0.945)
0.989 (0.733)
0.997 (0.942)
I/σI
21.1 (3.0)
15.2 (2.7)
22.0 (2.5)
16.1 (5.2)
6.9 (2.5)
18.9 (6.7)
6.6 (2.3)
13.9 (5.8)
completeness
(%)
99.8 (98.8)
99.9 (100.0)
100.0 (100.0)
96.1 (91.8)
96.7 (94.6)
97.9 (94.8)
94.2 (92.8)
97.8 (95.3)
multiplicity
8.3 (7.8)
10.3 (10.5)
11.3 (10.9)
3.7 (3.5)
3. 0 (3.0)
4.1 (4.0)
2.6 (2.7)
3.3 (3.1)
Refinement
monomers in ASU
2
2
2
2
1
1
1
1
no. reflns
31929
16202
18474
18708
18930
10818
8972
9900
Rwork/Rfree
0.2090/0.2455
0.2150/0.2751
0.2079/0.2543
0.1596/0.2001
0.1768/0.2021
0.2182/0.2619
0.1733/0.2265
0.1583/0.2110
no. of non-hydrogen
atoms
protein
3826
3786
3789
3253
1634
1610
1629
1633
water
191
123
50
87
145
81
122
159
ligands
60
53
60
NA
32
28
30
30
average B-factor (Å2)
protein
69.20
85.03
102.90
19.95
19.79
27.09
27.85
29.05
water
76.24
73.95
92.18
18.25
39.32
32.23
38.95
42.66
ligands
62.95
72.81
98.33
NA
25.31
31.11
46.36
39.25
inhibitor
62.82
70.75
98.33
NA
25.32
31.11
46.36
39.25
RMS deviations
bond lengths (Å)
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
bond angles (deg)
1.01
1.11
1.08
1.30
0.99
1.04
1.07
1.07
Ramachandran plots
favored (%)
97.1
96.0
97.1
98.6
99.1
97.5
99.0
98.1
allowed (%)
99.8
100
100
100
100
100
100
100
no. of
outliers
1
0
0
0
0
0
0
0
Values in parentheses are for the
highest resolution shell; NA, not applicable.
Values in parentheses are for the
highest resolution shell; NA, not applicable.In contrast to the SAM-bound structure, AZ51 binding
induces conformational
changes of the wall loop, whereupon the side chain of aromatic ring
of Tyr120 flips about 180° and forms stacking interactions with
both the phenyl and piperidine rings of AZ51 (Figure C). This feature appears unique to AZ51 and PaTrmD and was not observed in other reported TrmD inhibitor
structures.[15,17] In addition, because residue
Tyr120 is strictly conserved among TrmD orthologues (Supporting Information, Figure S3), AZ51 can presumably also trigger
a comparable side chain flip in other TrmD isozymes. The conformational
rearrangement of protein upon ligand binding generally could link
to complex binding kinetics.[21−23] Compared to SAH binding to PaTrmD, SPR reveals a faster (∼2-fold) on-rate (association
rate) and slower (∼30-fold) off-rate (dissociation rate) for
AZ51 binding (Table ).To further understand the biophysical parameters governing
AZ51
binding, the thermal stability of PaTrmD in the presence
of AZ51 or the SAM byproduct S-adenosyl-l-homocysteine (SAH) were determined (Figure D; Supporting Information, Figure S4). Both active-site ligands decrease the thermal
vibration (ΔTm = 4–29 °C),
suggesting that the stability of PaTrmD is enhanced
by ligand binding. In particular, AZ51 is able to significantly increase
the Tm to 79 °C, which is 17 °C
higher than thermostabilization conferred by SAH. Therefore, formation
of a highly stable complex of PaTrmD with AZ51 binding
possibly causes a conformational change at the active site, which
locks the structure in a very stable state. We also showed that AZ51
interacts with many other bacterial TrmDs to increase the enzymes’
thermal stability (Supporting Information, Figure S5). This further confirms that AZ51 has broad-spectrum TrmD-binding
activity.
Figure 3
Interactions of compound 15 with TrmDs. (A,B) The
polypeptide chains are shown as cartoons, whereas the residues involved
in inhibitor binding are shown as sticks. Ligands are shown as sticks
with 2Fo – Fc electron density maps (gray) contoured at 1σ. (A) Detailed
interactions between 15 and PaTrmD (chain
A) (PDB 5ZHN). (B) The interactions between 15 and MtbTrmD (PDB 5ZHL). (C) In PaTrmD, compound 15 induces
a conformational change of wall-loop residues, which could block substrate
tRNA (G37) binding. SAM-bound PaTrmD and
tRNA-bound H. influenzae TrmD (PDB 4YVI) were superimposed
onto 15-bound PaTrmD, respectively.
(D) The thermal shift of PaTrmD and MtbTrmD in the presence of SAH (2 mM), AZ51 (0.1 mM), or 15 (0.1 mM) was analyzed. Error bars represent mean ± SD. (E)
A superposition of PaTrmD–15 (subunit
1 and 2), MtbTrmD–15, and HiTrmD–tRNA (PDB 4YVI) showing the close contacts between 15 and interdomain linker. The interdomain linker of HiTrmD (residues 157–174), and tRNA are shown as
cartons, whereas guanosine is shown as stick. (F) Dose-dependent inhibition
of PaTrmD at varying tRNA titrations. The IC50 values determined for compound 15 at tRNA concentrations
of 1.5 μM (●), 4.5 μM (□), and 15 μM
(▲), respectively, are 0.02 ± 0.01, 0.06 ± 0.01,
and 0.26 ± 0.02 μM, respectively. Each data point represents
an average of duplicate experiments with error bars indicated as SD.
Interactions of compound 15 with TrmDs. (A,B) The
polypeptide chains are shown as cartoons, whereas the residues involved
in inhibitor binding are shown as sticks. Ligands are shown as sticks
with 2Fo – Fc electron density maps (gray) contoured at 1σ. (A) Detailed
interactions between 15 and PaTrmD (chain
A) (PDB 5ZHN). (B) The interactions between 15 and MtbTrmD (PDB 5ZHL). (C) In PaTrmD, compound 15 induces
a conformational change of wall-loop residues, which could block substrate
tRNA (G37) binding. SAM-bound PaTrmD and
tRNA-bound H. influenzae TrmD (PDB 4YVI) were superimposed
onto 15-bound PaTrmD, respectively.
(D) The thermal shift of PaTrmD and MtbTrmD in the presence of SAH (2 mM), AZ51 (0.1 mM), or 15 (0.1 mM) was analyzed. Error bars represent mean ± SD. (E)
A superposition of PaTrmD–15 (subunit
1 and 2), MtbTrmD–15, and HiTrmD–tRNA (PDB 4YVI) showing the close contacts between 15 and interdomain linker. The interdomain linker of HiTrmD (residues 157–174), and tRNA are shown as
cartons, whereas guanosine is shown as stick. (F) Dose-dependent inhibition
of PaTrmD at varying tRNA titrations. The IC50 values determined for compound 15 at tRNA concentrations
of 1.5 μM (●), 4.5 μM (□), and 15 μM
(▲), respectively, are 0.02 ± 0.01, 0.06 ± 0.01,
and 0.26 ± 0.02 μM, respectively. Each data point represents
an average of duplicate experiments with error bars indicated as SD.The protein structural changes induced by AZ51
binding to PaTrmD also raised questions about the
effect of the changes
on tRNA binding. To test this idea, we modeled the tRNA substrate
into the AZ51-bound structure by superimposing the structures of PaTrmD–AZ51 (this work) and HiTrmD
bound to tRNA (PDB 4YVI) (Figure C). Interestingly,
we found that the flipped side chain of Tyr120 overlaps with G37 of
the tRNA substrate, indicating that the AZ51-bound structure becomes
unsuitable for tRNA binding. This is consistent with the results of
the tRNA competition assay (Figure A). Taken together, we conclude that, in the PaTrmD structure, AZ51 not only mimics substrate SAM and
competitively occupies the SAM binding pocket but also induces conformational
changes in the active-site wall loop that probably could prevent the
binding of tRNA. This hypothesis of “dual functions”
of AZ51 may account for the high potency of this molecule in enzyme
inhibitory activity for PaTrmD. Given the striking
differences in AZ51 binding energetics with PaTrmD
and MtbTrmD, these results with PaTrmD raised questions about AZ51-induced structural changes in MtbTrmD.
AZ51 Binds to MtbTrmD without Inducing Conformational
Changes
As noted earlier, compared with its binding activity
with Gram-negative
TrmDs, AZ51 binds with much lower affinity to mycobacterial and Gram-positive[15] bacterial TrmDs (Table ). In addition, AZ51 increases the Tm of PaTrmD by 29 °C,
while the corresponding ΔTm values
of mycobacterial and Gram-positive bacterial TrmDs range from 1 to
13 °C (Figure D; Supporting Information, Figures S4, S5). These observations suggest that AZ51 may use a different mechanism
for binding to mycobacterial and Gram-positive bacterial TrmDs in
spite of the fact that the overall protein sequence similarity is
high among bacterial TrmDs (Supporting Information, Figure S3, Table S1). To gain insight
into the inhibitory mechanism on MtbTrmD, we crystallized
its free form and determined the structure at 2.20 Å resolution.
We then optimized the crystal soaking conditions and successfully
soaked SAH and AZ51 into the MtbTrmD crystals, which
allowed determination of the structures at high resolution (Table ).As shown
in Supporting Information, Figure S2D, MtbTrmD forms a biological dimer with the active site located
at the dimer interface, as with PaTrmD. Indeed, MtbTrmD and PaTrmD are highly similar in
structure with RMS deviations of 1.3 Å for all C-α atoms
after superposition. The electron density of SAH is clearly identified
at the active site in MtbTrmD–SAH structure,
and SAH is positioned by three active-site loops (Supporting Information, Figure S2D). The adenine base is stacked between
Pro85 and Leu138 and forms hydrogen bonds with main chains atoms of
Ile133, Gly134, Tyr136, and Leu138. The 2′ hydroxyl group of
the ribose makes a hydrogen bond to the carbonyl oxygen of Pro83.
The amino acid tail interacts with side chain of Thr84 and carbonyl
oxygen of Glu112 by forming hydrogen bonds. A protein sequence alignment
of multiple bacterial TrmDs shows that the residues involved in adenine
base interactions are strictly conserved but more diverse for those
interacting with the ribose and tail moiety of SAH (Supporting Information, Figure S3). Compared to MtbTrmD,
SAM in PaTrmD has four additional hydrogen bonds
with side chains of residues Tyr91, Gln95, Asp182, and His185, where
the corresponding residues in MtbTrmD (Val82, Ala86,
Glu180, and Ser183) do not interact with SAH (Supporting Information, Figure S2E). Thus, these additional interactions
observed at active site in PaTrmD could account for
the higher binding affinity for SAM and SAH, which is consistent with
SPR data (Table ).The crystal structure of MtbTrmD–AZ51 complex
was determined at a resolution of 2.20 Å (Table ). Similar to the AZ51 binding mode in PaTrmD (Figure B), AZ51 is positioned at the SAH binding site by interacting
with three active-site loops of MtbTrmD (Figure D). Residues involved
in AZ51 interactions are strictly conserved in other bacterial TrmDs
including PaTrmD (Supporting Information, Figure S3). However, the binding of AZ51 does
not induce the side chain flip of Tyr111 in MtbTrmD,
while the corresponding residue Tyr120 in PaTrmD
turned 180° to form stacking interactions with the phenyl ring
of the inhibitor (Figure E), suggesting a more rigid active site in MtbTrmD compared to PaTrmD. In the absence of the conformational
changes at the wall loop, the piperidine ring of AZ51 in MtbTrmD is positioned differently than in PaTrmD, resulting
in the loss of interactions between the terminal amine and the side
chain of Glu112. Even though the binding mode of the thienopyrimidinone
ring is highly similar in MtbTrmD and PaTrmD, the conformational changes induced by AZ51 in PaTrmD lead to additional interactions (hydrophobic and hydrogen-bonding)
and eventually higher binding affinity. Thus, the structural flexibility
of the wall loop in Gram-negative bacterial TrmDs appears to play
a unique role in the binding of AZ51. Exploiting this unique mechanism
of AZ51 inhibition of TrmD, we next designed and synthesized a series
of derivatives with the goal of improving enzyme inhibition and antibacterial
activity.
Design and Synthesis of Thienopyrimidinone Derivatives
The PaTrmD crystal structure was used to design
a series of inhibitors based on the thienopyrimidinone molecular scaffold
shown in Figure .
Inspired by the unique mechanism leading to a potential tRNA competition
instead of exclusively designing substituents that make contacts with
residues Glu121, Asp182, and Tyr120 as found with AZ51, our idea driving
the inhibitor design process was that alkyl and aryl substitutions
at R1 and R2 would allow full access to the
free space outside the SAM binding site, with the hydrophobicity of
the substituent and the charge distribution around the centered ammonium
ion enhancing enzyme inhibitory activity and thus antibacterial potency.On the basis of these design principles, we synthesized a series
of thienopyrimidinone analogues. The key aldehyde 8 was
synthesized using a reaction sequence and procedures similar to those
reported by Hill et al.:[15] reaction of
methyl 2-oxopropanoate (1) and ethyl 2-cyanoacetate (2) in the presence of sulfur in DMF afforded the thiophene
amino diester (3), which upon condensation with formamidine
acetate in ethanol under reflux yielded thienopyrimidinone (4). Trans amidation of 4 with
benzylamine derivative (7), which was synthesized from
4-formylbenzonitrile (5) followed by treatment with trifluoracetic
acid, afforded the key aldehyde 8 (Scheme ). We then modified the procedure of Hill
et al.[15] for reductive amination of aldehyde 8 with various amines. We found that the reductive amination
with titanium isopropoxide (Ti(OiPr)4),
sodium triacetoxyborohydride (NaBH(OAc)3) was sluggish
and needed long reaction times. Thus, we optimized the same transformation
using sodium cyanoborohydride (NaCNBH3) in methanol with
catalytic amount of acetic acid at ambient temperature for 18 h, which
afforded compounds 9–26. Various
Boc-protected diamines were coupled using the above-mentioned procedure
to afford compounds 27–30, which,
on treating with 4 N HCl in dioxane, afforded compounds 31–34.
Scheme 1
Synthesis of TrmD Inhibitors Based on the
Thienopyrimidinone Scaffold;
Synthesis of Compounds 9–26 and 31–34 Using Reductive Amination Method
Conditions: (a) sulfur, triethylamine
DMF, 50 °C; (b) formamidine acetate, ethanol, reflux; (c) ethylene
glycol, pyridinium-p-toluenesulfonate, Dean–Stark;
(d) LiAlH4, THF, 0 °C to room temparature; (e) Et3N, EtOH, reflux; TFA, DCM, 0 °C to ambient temperature;
(f) amine, NaCNBH3, cat. acetic acid, MeOH; (g) 4 N HCl
in dioxane.
Synthesis of TrmD Inhibitors Based on the
Thienopyrimidinone Scaffold;
Synthesis of Compounds 9–26 and 31–34 Using Reductive Amination Method
Conditions: (a) sulfur, triethylamineDMF, 50 °C; (b) formamidine acetate, ethanol, reflux; (c) ethylene
glycol, pyridinium-p-toluenesulfonate, Dean–Stark;
(d) LiAlH4, THF, 0 °C to room temparature; (e) Et3N, EtOH, reflux; TFA, DCM, 0 °C to ambient temperature;
(f) amine, NaCNBH3, cat. acetic acid, MeOH; (g) 4 N HCl
in dioxane.
Biochemical Structure–Activity
Relationships (SAR) for
TrmD Inhibitors
To explore the relationship between inhibitor
structure and the potency of TrmD inhibition, we quantified the concentrations
of compounds producing 50% inhibition of TrmD methylation of a tRNA
substrate. Compared with AZ51, inhibitory activity was significantly
reduced (∼6–16-fold) when the R1 and R2 positions are present in a cyclic structure (without the
terminal amine) shown in compounds 9 and 10 (Table ). This observation
is consistent with our structural data (explored next), where the
terminal amine group as a hydrogen bond donor forms the interaction
with PaTrmD. Furthermore, compounds 14, 15, 16, and 17 explore the
optimal bulk structure of a single substitution, with benzyl (14), cyclohexyl (16), and adamantyl (17) groups all showing submicromolar activity. However, the n-octyl substitution in 15 confers 5–30-fold
more potent inhibitory activity (IC50 ∼ 24 nM),
which suggests that the long straight-chain alkyl group could play
a role in PaTrmD binding. This observation prompted
us to quantify the effect of alkyl chain length on enzyme activity.
Accordingly, we synthesized compounds 21–24 with 4-, 6-, 10-, and 12-methylene group alkyl chain lengths.
This series of analogues of 15 shows a clear optimal
chain length of 8–10 carbons for the most potent TrmD inhibitory
activity (Supporting Information, Figure S6). Substitutions at the R2 position, however, generally
had a detrimental effect on TrmD inhibitory activity (11–13, 18, 19, and 20) when the substituent at R1 position is also
bulky, which suggests steric limitations beyond the SAM binding pocket.
For example, when −H at R2 position of compound 15 is substituted with −Et where compound 19 is generated, the enzyme inhibitory activity is significantly decreased
from 24 to 370 nM (Table ). Take compound 20 as another example, when
substituents at both R1 and R2 positions are
bulk structures (n-octyl and benzyl groups), its
enzyme inhibitory activity is nearly abolished.
Thienopyrimidinone
Inhibitors Induce the Same Conformational
Change at Active Site as AZ51 in PaTrmD
To further investigate
the inhibitory mechanism of the thienopyrimidinone compounds, we solved
the crystal structures of PaTrmD in complex with 11 and 15 (Table ). The interactions of PaTrmD with 15 are illustrated in Figure A, and with 11 in Figure A. Similar to the AZ51 binding mode (Figure A), these thienopyrimidinone
inhibitors are mainly stabilized by the three active-site loops (cover,
bottom, and wall loops). At the cover loop, the thienopyrimidinone
ring remains tightly bound in the adenine pocket, with hydrogen bonds
to residues Ile138, Tyr141, and Leu143, and stacking interactions
with residues Pro94 and Leu143. The phenyl ring of 11 and 15 is locked in place by stacking interactions
with residues Pro94 and Tyr120 in the bottom loop. Electron densities
for the thienopyrimidinone and phenyl rings are clearly visible in
the structures of all thienopyrimidinone inhibitors. However, the
electron densities for the flexible tails (substituents) of the thienopyrimidinone
inhibitors are poorly visible and may possess multiple conformations.
Similar to the AZ51 crystal structure, the binding of thienopyrimidinone
inhibitors to PaTrmD induces a conformational change
of the wall loop, in which the side chain of residue Tyr120 flips
∼180° and forms stacking interactions with the phenyl
ring (Figure A; Figure A). Compound 15 also highly thermostabilizes PaTrmD (ΔTm = 12 °C) (Figure D; Supporting Information, Figure S4) and other bacterial TrmDs (Supporting Information, Figure S5). It is noteworthy that the correlation
is not strong between enzyme inhibition IC50 values (Table ) and thermal stability Tm (Supporting Information, Figure S4), indicating that the substituents play additional
roles in enzyme inhibition other than forming a thermally stable complex
with the protein, such as tRNA competition. We will elaborate this
point using compound 15 as an example next.
Figure 4
Crystal structures
of compound 11 bound to PaTrmD (A; PDB 5ZHM) and compound 12 bound to MtbTrmD (B; PDB 5ZHK), respectively.
The polypeptide chains are shown as cartoons, whereas
the residues involved in inhibitor binding are shown as sticks. Ligands
are shown as sticks with 2Fo – Fc electron density maps (gray) contoured at
1.0σ. Potential interactions involved in inhibitor binding are
indicated by broken lines (black, hydrogen bonds; green, hydrophobic
contacts including stacking interactions).
Crystal structures
of compound 11 bound to PaTrmD (A; PDB 5ZHM) and compound 12 bound to MtbTrmD (B; PDB 5ZHK), respectively.
The polypeptide chains are shown as cartoons, whereas
the residues involved in inhibitor binding are shown as sticks. Ligands
are shown as sticks with 2Fo – Fc electron density maps (gray) contoured at
1.0σ. Potential interactions involved in inhibitor binding are
indicated by broken lines (black, hydrogen bonds; green, hydrophobic
contacts including stacking interactions).To investigate the potential role of thienopyrimidinone inhibitors
in tRNA binding, we then modeled the tRNA substrate with the PaTrmD–15 structure by superimposing
the HiTrmD–tRNA (PDB 4YVI) structure (Figure C). Similar to the
mechanism inferred for AZ51, we found that the flipped side chain
of Tyr120 also overlapped with G37 of tRNA substrate, indicating that
the 15-bound structure was unavailable for tRNA binding.
The role of the flexible alkyl chain of 15 was explored
below.
Crystal Structures of MtbTrmD in Complex with Thienopyrimidinone
Inhibitors
To further define the structural differences between PaTrmD and MtbTrmD, we soaked 12 (Figure B) and 15 (Figure B) into MtbTrmD crystals and solved the structures
at high resolution (Table ; 2.30 and 2.25 Å, respectively). Similar to the AZ51
binding mode in MtbTrmD, 12 (Figure B) and 15 (Figure B), are
positioned at the SAH binding site by interacting with the three active-site
loops of MtbTrmD, but the inhibitors did not induce
the side chain flip of Tyr111 (Figure A). Inhibitor 15 shows over 100-fold lower
affinity for MtbTrmD compared to the PaTrmD binding affinity (Table ), further indicating that the flip of the Tyr side chain
could play an important role in binding to the TrmD protein. Furthermore,
similar to the PaTrmD–15 structure,
the location of the substituent of 15 is defined by relatively
weak electron density in MtbTrmD. Taken together,
we found that the thienopyrimidinone-based inhibitors adopt similar
binding modes in both MtbTrmD and PaTrmD by interacting with conserved residues, except for the ability
to induce a conformational change in the active-site wall loop.
Thienopyrimidinone Substituent Flexibility Is Critical for Potent
TrmD Inhibition
The structure–activity studies noted
earlier identified optimal R1 and R2 substituents
on the thienopyrimidinone scaffold, with a single alkyl chain of 8–10
carbons producing 8–50-fold stronger affinity to TrmDs compared
with AZ51 (Table ).
We next explored the mechanistic basis for this potency. In crystal
soaking experiments with PaTrmD, most thienopyrimidinone
compounds replaced SAM at only one of the two active sites in the
dimer, presumably due to the stronger affinity of SAM for binding
to the second active site and/or different dynamics of two active
sites. Unexpectedly, we were able to identify 15 at both
active sites of the dimer (Supporting Information. Figure S7). We assume that the successful replacement of SAM
at both active sites resulted from a higher affinity of 15 relative to other compounds. The electron densities of the alkyl
substituent in 15 at the two active sites were poorly
resolved, indicating flexibility and multiple orientations. Interestingly,
the unique side chain flip of residue Tyr120 induced by 15 only occurs at one of the active sites, which indicates that, in
the crystal soaking condition, the two active sites have different
structural environments that require conformational flexibility in 15 for binding. To further investigate the role of the alkyl
chain of 15 in enzyme inhibition, we superimposed the
published structure of tRNA-bound HiTrmD[20] onto the PaTrmD–15 structure and found a potentially close contact between
the flexible 15-alkyl chain and the interdomain linker,
which was also observed in MtbTrmD (Figure E). Moreover, the hydrophobicity
of the alkyl chain probably favors the wall-loop conformational change
leading to the formation of a hydrophobic core and facilitating ligand
binding. In Table , 15 shows much higher binding on-rate (kon) to PaTrmD compared with AZ51 which
leads to ∼50-fold higher binding affinity (KD). This observation could be due to the role of the hydrophobic
alkyl chain in forming the hydrophobic core at the active site. To
test this hypothesis, we replaced the long alkyl group with diethylene
glycol to reduce the hydrophobicity of the substituent but still retaining
a certain level of structural flexibility (compounds 25 and 26 in Table ). Clearly, the ethylene glycol substituents reduce the enzyme
inhibitory activity by 4–8-fold, compared with the alkyl chain,
further suggesting that hydrophobicity or lipophilicity of the substituent
facilitate TrmD inhibitory activity.Taken together, we show
that 15 occupies the SAM binding site and induces a Tyr120
side-chain flip. From the crystal structures, we hypothesize that
the flipped side chain and the flexible alkyl chain could interfere
with tRNA binding. To further investigate this hypothesis, we have
performed a tRNA competition assay, in which the IC50 value
was observed to increase by 13-fold from 0.02 to 0.26 μM when
the tRNA concentration was increased (Figure F).
O6-Substituted Thienopyrimidinones
To extend
the biochemical SAR study, we assessed the effect of O6-subtituents on the TrmD inhibitory activity of thienopyrimidinones
(Figure B). We synthesized
a series of O6-substituted analogues with alkyl substituents
of varying lengths by treating 37, 38, and 10 with alkyl/arylalkyl halides in the presence of K2CO3 in DMF to afford O6-alkylated analogues 39–47 (Scheme ). PaTrmD IC50 values were then determined for the compounds using the in vitro
assay described earlier (Table ). The result shows that these compounds do not inhibit PaTrmD, with the exception of 47, which is
likely to undergo facile degradation in which the aryl substituent
is released to form 15. The basis for the lack of activity
of the O6-substituted thienopyrimidinone derivatives likely
results from the limited space available to accommodate the O6-group in the adenine-base binding site (Figure B). In addition to lower enzyme
inhibitory activity compared to the other thienopyrimidinone, the
O6-substituted analogues also only show a minor impact
on protein thermal stabilization (ΔTm = 1–2 °C), suggesting weak interactions with PaTrmD (Supporting Information, Figure S4).
Scheme 2
Synthesis of O6-Substituted Thienopyrimidinones
Conditions: (a) Et3N, EtOH, reflux; (b) alkyl bromide,
K2CO3,
DMF; (c) 4 N HCl in dioxane.
Table 4
Structures and IC50 TrmD
Inhibition Values for O6-Subsituted Thienopyrimidinone
Derivatives
NA = not active
up to 50 μM.
Synthesis of O6-Substituted Thienopyrimidinones
Conditions: (a) Et3N, EtOH, reflux; (b) alkyl bromide,
K2CO3,
DMF; (c) 4 N HCl in dioxane.NA = not active
up to 50 μM.
Thienopyrimidinone
Analogues with Long Alkyl Substituents Possess
Antibacterial Activity
On the basis of the biochemical activity
of the thienopyrimidinone analogues as TrmD inhibitors, we next quantified
the antibacterial activity of the compounds. The first analysis involved
a single-dose (100 μM) growth inhibition assay against a collection
of Gram-positive and Gram-negative bacteria. Of the compounds listed
in Table , only 15, 23, and 24 showed growth inhibitory
activity at 100 μM against Gram-positive bacteria and mycobacteria.
We then used these compounds to determine minimal inhibitory (MIC)
(Table ) and minimal
bactericidal (MBC) concentrations (Table ). The discovery of the activity of the 15 series analogues against Gram-positive bacteria and mycobacteria
is in agreement with the broad-spectrum binding affinity of 15 to three TrmDs (Table ; Supporting Information, Figure S5). It is worth noting that 15 and 24 also exhibit antibacterial activity against the Gram-negative pathogens Acinetobacter baumannii and Salmonella
enteritidis with high MIC50/MIC90 values. Thus, 15, 23, and 24 show signs of broad-spectrum antibacterial activity, possibly due
to their multiple TrmD targets. In an attempt to extend and improve
the antibacterial activity to Gram-negative bacteria, we either added
primary amines[24] to 15 and
its series analogues (Scheme ), or conjugated with siderophores[25,26] (Supporting Information, Scheme S1),
where we synthesized compounds 31–34, 53, and 57, respectively (Table ). These compounds retained
submicromolar TrmD inhibitory activity, although they did not show
activity against Gram-negative bacteria and even lost the activity
to Gram-positive bacteria (data not shown).
Table 5
Antibacterial
Activities (μM)
for Selected Thienopyrimidinone Analogues
15
23
24
pathogen
MIC50a
MIC90a
MBC99.9b
MIC50
MIC90
MBC99.9
MIC50
MIC90
MBC99.9
Gram-negative
P. aeruginosa
NAc
NA
NA
NA
NA
NA
NA
NA
NA
A. baumannii
62
95
>100
>100
>100
>100
75
99
>100
K. pneumoniae
NA
NA
NA
NA
NA
NA
NA
NA
NA
S. enteritidis
NA
NA
NA
NA
NA
NA
>100
>100
>100
E. coli
NA
NA
NA
NA
NA
NA
NA
NA
NA
Gram-positive
S. aureus
31
36
54
6
8.5
34
5
8.5
34
E. faecalis
NA
NA
NA
NA
NA
NA
NA
NA
NA
S. pneumoniae
60
73
>100
34
56
>100
62
74
>100
mycobacteria
M. smegmatis
58
72
>100
13
18
36
17
23
69
M. tuberculosis
78
>100d
>100
22
43
75
44
88
nde
MIC50 and MIC90 represent hit concentrations inhibiting growth
by 50% and 90%, respectively.
MBC99.9 represents hit
concentrations kill bacteria by 99.9%.
NA = no bacterial activity.
Denotes detectable growth inhibition
or bactericidal activity but less than 50% for MIC50, less
than 90% for MIC90, and less than 99.9% for MBC99.9.
nd = not determined.
MIC50 and MIC90 represent hit concentrations inhibiting growth
by 50% and 90%, respectively.MBC99.9 represents hit
concentrations kill bacteria by 99.9%.NA = no bacterial activity.Denotes detectable growth inhibition
or bactericidal activity but less than 50% for MIC50, less
than 90% for MIC90, and less than 99.9% for MBC99.9.nd = not determined.As with the TrmD inhibitors
developed by Hill et al.,[15] there is little
relationship between the potency
of TrmD inhibition by the new thienopyrimidinone compounds developed
here and their antibacterial activity. The most likely explanation
is poor cell permeability, although it is possible that inhibition
of the otherwise essential TrmD by the compounds is incomplete in
the cellular environment. That bacterial uptake is responsible for
the observed antibacterial activity is supported by the observation
of stronger activity against two Gram-positive pathogens (S. aureus and S. pneumoniae) compared to the single Gram-negative pathogen, A.
baumannii, with its additional outer cell membrane
(Table ). However,
mycobacteria have a complex mycolic acid-rich cell wall that presents
a drug permeability problem similar to Gram-negative bacteria, yet M. smegmatis and M. tuberculosis show sensitivity to TrmD inhibitors similar to Gram-positive S. aureus (Table ). This idiosyncratic activity could result from mechanisms
of antibacterial activity other than TrmD inhibition, drug efflux
pumps, or compound degradation. The strong SAR for TrmD inhibition
by thienopyrimidinone compounds established here provides a foundation
for pursuing antibacterial SAR.
Hemolytic Activity of the
Thienopyrimidinone Compounds
To further explore the behavior
of the thienopyrimidinone analogues,
we assessed the ability of the compounds to rupture red blood cells
as an index of membrane disrupting potential. The hemolytic activity
of all compounds is shown in Supporting Information, Table S2. In general, most of the tested compounds show no
or weak hemolytic activity at the highest tested concentration (100
μM).
Discussion and Conclusions
Elaborating
on a thienopyrimidinone scaffold, we prepared and analyzed
a series of TrmD inhibitors, which revealed a novel SAM-competitive,
active site Tyr-flipping inhibition mechanism that distinguished Gram-negative
TrmDs from Gram-positive and mycobacterial counterparts. Several of
these compounds showed nanomolar TrmD inhibition, tRNA-competitive
binding, and micromolar antimicrobial activity against Gram-positive
bacteria and, in some instances, Gram-negatives and mycobacteria.
Experimental Section
Protein Expression and
Purification
Production of PaTrmD protein
was described previously.[14] Briefly, a
truncated fragment (spanning residues Leu5-Asp250)
of PaTrmD was cloned into the vector pNIC28-Bsa4
and transformed into chemically competent Escherichia.
coli BL21 (DE3) Rosetta T1R cells. For M. tuberculosis TrmD, S. aureus TrmD, E. faecalis TrmD, and S. pneumoniae TrmD, the full-length genes were inserted
into vector pYUB28b-cHIS6 and transformed into chemically
competent BL21 (DE3). The following protein expression and purification
procedures were applied to all TrmD proteins. E. coli cells harboring a plasmid were cultivated in Luria–Bertani
(LB) medium at 37 °C to OD600 0.6–0.8. The
proteins were then overexpressed by the addition of 0.5 mM IPTG at
16 °C for 24 h. Cells were pelleted and stored at −80
°C.All enzyme purification was performed on ÄKTA
(GE Healthcare) chromatography system at 4 °C. The thawed pellets
from 1 L cell cultures were resuspended in 30 mL of lysis buffer (20
mM Na-HEPES, pH 7.5, 0.3 M NaCl, 5% glycerol, 0.5 mM TCEP), sonicated
in the presence of one tablet of EDTA-free protease inhibitors (Roche),
and the lysates cleared by centrifugation at 22000 rpm for 45 min
in a JA-25.50 rotor. The proteins were purified by immobilized metal
affinity chromatography (IMAC) using HisTrap HP IMAC column (GE Healthcare)
and by size exclusion on a HiLoad Superdex 200 16/60 column (GE Healthcare).
Purified proteins were concentrated to 20 mg/mL by ultrafiltration
in 20 mM Na-HEPES, pH 7.5, 0.3 M NaCl, 10% (v/v) glycerol, 0.5 mM
TCEP, and were stored at −80 °C.
tRNA in Vitro Synthesis
and Purification
The preparation
of tRNA substrates has been reported previously.[14] A PaTrmD substrate, tRNALeu(CAG), was synthesized using in vitro transcription by MEGAshortscrip
T7 Transcription Kit (cat. no. AM1354, Thermo Scientific) according
to the manufacturer’s instructions. DNA template and primers
(Supporting Information, Table S3) were
obtained from Integrated DNA Technologies, Inc. (IDT). The transcribed
tRNA was desalted in RNase-free water and purified on an Agilent HPLC
1200 equipped with an Agilent Bio SEC-3 size-exclusion column (300
Å pore size, 3 μm particle size, 7.8 mm i.d.) operated
at 60 °C eluted with 100 mM ammonium acetate buffer. The purity
of the tRNA product was further assessed on an Agilent 2100 Bioanalyzer
with a small RNA chip, and the tRNA was quantified using a Nanodrop
spectrophotometer. The final product was desalted in RNase-free water,
concentrated to 20 mg/mL, and frozen at −20 °C.
TrmD Inhibitor
Assay
The tRNA methyltransferase activity
of TrmD was measured by a bioluminescence-based assay using the MTase-Glo
Assay Kit (Promega, cat. no. V7601) in a 384-well format. Generation
of SAH was measured by its conversion to ADP by the MTase-Glo reagent,
with MTase-Glo detection solution applied to convert ADP to ATP, which
is then used by luciferase to generate a luminescent signal proportional
to the coupled TrmD methyltransferase activity. Luminescence was measured
using a plate-reading luminometer (BioTek Synergy 4 microplate reader)
and was correlated to SAH concentration using an SAH standard curve
generated according to manufacturer’s instructions. The TrmD
concentration used in this assay is 8 nM. Thus, the estimated limit
of IC50 detection for a substrate competitive inhibitor
is about 4 nM, which is half of the molar concentration of the enzyme.
The detailed TrmD activity assay and kinetic studies have been described
in our recent work.[14,17]The TrmD methyltransferase
reaction was performed at 37 °C in 5 μL reaction mixtures
containing 8 nM TrmD enzyme, 50 mM Tris-HCl buffer, pH 8.0, 80 mM
NaCl, 1 mM MgCl2, 0.5 mM DTT, 0.01% (v/v) Triton X-100,
2 μM SAM, and 0.5 μM tRNALeu(CAG), and 1×
MTase-Glo reagent. TrmD enzyme in 1× assay buffer (50 mM Tris-HCl
buffer, pH 8.0, 80 mM NaCl, 1 mM MgCl2, 0.5 mM DTT, 0.01%
v/v Triton X-100) was preincubated with an inhibitor in a serial dilution
(0–500 μM) at ambient temperature for 15 min. The negative-control
mix was prepared in an identical manner except that 1× assay
buffer was used in place of the inhibitor solution. The reaction was
initiated by adding the TrmD enzyme mixture to a final concentration
of 0.25 μg/mL. The plate was sealed and incubate at 37 °C
for 45 min. After reaction, the mixture was mixed with 5 μL
of MTase-Glo detection solution and further incubated for 30 min at
ambient temperature in the dark. The luminescent signal was read in
the microplate luminometer and background subtracted from all sample
and control reads before analyzing the net signal intensity in Prism
GraphPad. The rate for each activity assay with a series of compound
titrations was expressed as a percentage of the control assay and
analyzed using nonlinear regression fit in Prism GraphPad to estimate
IC50 values for each compound. The assay was done in two
independent experiments performed in technical duplicates.
Surface
Plasmon Resonance (SPR)
SPR analysis was performed
using a BIACORE T200 instrument (GE Healthcare) equipped with a CM5-S
sensor chip. The surfaces of all flow cells were activated for 7 min
with a 1:1 mixture of 0.1 M NHS and 0.1 M EDC at a flow rate of 10
μL/min. Anti-His antibody was captured on all surfaces to ∼12000
RU. Surfaces were then blocked with a 7 min injection of 1 M ethanolamine,
pH 8.0. The protein TrmD, at a concentration of 50 μg/mL in
50 mM Tris-HCl pH8, 150 mM NaCl, 1 mM MgCl2, 1 mM DTT,
and 0.05% Tween-20, was captured on the anti-His antibody surface
to a density of ∼2200 RU. Flow cell no. 1 was left blank to
serve as a reference surface. Different concentrations (2-fold dilutions)
of the test analyte were injected from lowest to highest concentration
in the same buffer over the flow cell at 30 μL/min for a contact
time of 30 s and dissociation time of 240 s. Analysis of inhibitor
binding affinity to TrmD was quantified using 2-fold dilutions. Data
were collected at a rate of 10 Hz, and at a temperature of 25 °C.
The KD value was calculated using steady-state
affinity fit.
Crystallization and Data Collection: PaTrmD
The crystal
screening and optimization for PaTrmD has been described
in our recent work.[14] Briefly, Crystals
of PaTrmD were grown by mixing equal volumes of 20
mg mL–1 protein and precipitant solution containing
0.1 M Tris-HCl, pH 8.6–8.8, 20% (v/v) MPD, 20% (w/v) PEG 1000,
and 5% (w/v) PEG200 and incubating at 20 °C using vapor diffusion
method (hanging-drop). The crystals were soaked with 1–5 mM
inhibitor in the precipitating solution supplemented with 20% (v/v)
glycerolat 20 °C for at least 4 h and rapidly frozen in liquid
nitrogen. X-ray diffraction data were collected at the PXIII beamline
in SLS (Villigen, Switzerland) for the PaTrmD–AZ51
crystal. For both PaTrmD–11 and PaTrmD–15 crystals, X-ray diffraction
data were collected at the MX2 beamline in Australian Synchrotron.
Crystallization and Data Collection: MtbTrmD
Crystallization
conditions were screened at 20 °C with protein concentration
at 10 mg/mL using the vapor diffusion method and commercial crystallization
screens in Intelli 96-3 wells sitting-drop plates. Three precipitant:protein
ratios (1:1, 1:2, and 2:1) were tested using the mosquito crystallization
robot (Art Robbins Instruments) and drop volumes of 0.2 μL.
Optimized crystals of the MtbTrmD were obtained using
the vapor diffusion method (sitting drops) in 24-well trays by mixing
a volume of 2 μL of protein with 1 μL of a precipitant
solution containing 100 mM Bis-Tris propane at pH 6.5, 20% (w/v) PEG3350,
and 0.1 M ammonium acetate. Prior to data collection, crystals were
soaked with 1–5 mM inhibitor (for obtaining inhibitor-bound
structure) or 5 mM SAH in their respective precipitating solution
supplemented with 20% (v/v) glycerol and rapidly frozen in liquid
nitrogen. X-ray diffraction data were collected at the MX1 beamline
in the Australian Synchrotron for apo MtbTrmD crystal,
at the PXIII beamline in SLS (Villigen, Switzerland) for the MtbTrmD–SAH and MtbTrmD–12 crystals, and at the MX1 and MX2 beamlines in the Australian
Synchrotron for MtbTrmD–AZ51 andMtbTrmD–15 crystals, respectively.
Structure Determination
and Refinement: PaTrmD
Diffraction
intensities were reduced with XDS,[27] scaled,
merged, and truncated with SCALA/TRUNCATE.[28,29] The structures were determined by molecular replacement using Phaser,[30] with the structure from PaTrmD–SAM
(PDB 5WYQ) as
search probe. Model for the 3D structure of PaTrmD
was built iteratively at the computer graphics using COOT[31] and refined using Autobuster.[32] Waters and ligands were added to the model after a few
runs of refinement. The geometrical parameters for inhibitors were
generated using program PRODRG.[33]
Structure
Determination and Refinement: MtbTrmD
Data
reduction is the same as described above for PaTrmD.
The structures were determined by molecular replacement using Phaser,[30] with the MtbTrmD structure
modeled from I-TASSER server[34] as search
probe. A model for the 3D structure of MtbTrmD was
built using COOT[31] and refined using Autobuster,[32] except that apo MtbTrmD structure
was refined by Refmac.[35] The geometrical
parameters for inhibitors were generated using program PRODRG.[33]The quality of the structures was assessed
using the MOLPROBITY server[36] (molprobity.biochem.duke.edu),
and figures were generated using the Pymol software.[37] Data collection and structure refinement parameters are
summarized in Table .
Thermal Stability Assay
The thermal stability analysis
of PaTrmD was performed in a 96-well PCR plate (Bio-Rad)
with 50 μL per reaction containing 5× SYPRO Orange dye
(Invitrogen), 4 μM test protein, and the test ligand(s) at various
concentrations. The assay buffer (50 mM Tris-HCl pH8, 150 mM NaCl,
1 mM MgCl2, 1 mM DTT, and 0.05% Tween-20) was added instead of the
test ligand as a negative control. The temperature was increased from
25 to 95 °C in an i-Cycler iQ5 real-time PCR (Bio-Rad). The thermal
stability curve and the temperature midpoint Tm for the protein-unfolding transition was analyzed using the
Bio-Rad iQ5 software.
Antibacterial Activity Assay
The
synthetic compounds
were first screened at a fixed concentration of 100 μM against
a series of bacteria strains by monitoring the bacterial growth in
broth at 96-well format. Bacteria P. aeruginosa PA14 (−), A. baumannii ATCC
17961 (−), K. pneumoniae ATCC
13883 (−), S. enteritidis ATCC
13076 (−), E. coli BW25113 (−),
and S. aureus ATCC 43300 (methicillin-resistant)
(+) were tested in cation-adjusted Mueller–Hinton (MH) broth
at 37 °C. The E. faecalis (+)
ATCC 51299 strain was tested in brain heart infusion broth. S. pneumoniae ATCC 49619 (+) was tested in MH medium
supplemented with 3% defibrinated sheep blood at 37 °C and 5%
CO2. M. smegmatis mc2155 (+) was cultured in the complete Middlebrook 7H9 medium.[4] Briefly, four single colonies of each bacterial
strain were picked from a freshly streaked plate and grown overnight
at 37 °C. The overnight culture was subcultured to fresh medium
and further grown at 37 °C to obtain log-phase bacteria. The
bacterial concentration was further adjusted to approximately 5 ×
105 cells per mL of each well in the presence or absence
of 100 μM of the test compound. The total culture volume of
each well was 100 μL. Ampicillin and kanamycin were testing
in parallel as positive controls. DMSO control and blank control were
applied. The culture was allowed to grow for 24 h at 37 °C before
cell density was measured by OD600 using a BioTek Synergy
4 microplate reader. The percent bacterial growth (percentage survival)
was expressed as the OD600 value of the test well as a
percentage of that from the untreated control wells. For S. pneumoniae, the cells were spin down and the supernatant
was measured at OD450. The percentage survival was expressed
as the OD450 value of the test well as a percentage of
that from the untreated control wells.Minimum inhibitory concentration
(MIC) of the hits from the screening were determined by serial broth
microdilution method and following the procedures described above,
where cell suspensions were incubated with serial dilutions of test
compound. The experiment was done in two independent experiments performed
in duplicate. MIC50 and MIC90 represent the
concentration of compound that inhibit 50% and 90% bacterial growth
respectively as compared to drug-free controls. Experiments were performed
in biological duplicates. The MBC was determined by CFU enumeration
on MH agar plates (except BHI agar plates for E. faecalis and TSAagar plates with 5% sheep blood for S. pneumoniae) after exposure to a given concentration of test compound. Bacterial
cultures were grown to mid log phase, adjusted to a final cell density
of 5 × 105 CFU/mL, and then treated with concentrations
equivalent to 1×, 2×, and 4× MIC90 of test
compound for 24 h at 37 °C. Drug-free cultures were plated at
the start of the experiment to determine the bacterial load of the
inoculum. After incubation for 24 h, the compound-treated cultures
were plated to determine CFU. MBC99.9 are defined as the
concentration of the test compound that caused at least 1000-fold
reduction in CFU as compared to the untreated inoculum at time point
zero.Bacterial culture of M. tuberculosis and its MIC determination was performed in BSL-3 laboratory in NUS
Singapore, and the details were described previously.[38,39] Briefly, M. tuberculosis (ATCC 27294)
was maintained in complete Middlebrook 7H9 medium supplemented with
0.05% (v/v) Tween 80, 0.5% (v/v) glycerol, and 10% (v/v) Middlebrook
albumin-dextrose-catalase at 37 °C in PETG sterile square bottles
with shaking at 80 rpm in 10 mL volumes. MICs are determined by the
broth dilution method as described. Mid log phase cultures (OD600 = 0.3–0.6) are spun down by centrifugation, resuspended
in fresh medium, and adjusted to an OD600 = 0.1. Cell suspension
(100 μL) was added into wells containing 100 μL of 2-fold
serially diluted compound (highest concentration 100 μM) in
transparent flat-bottomed 96-well plates, sealed with Breath-Easy
membranes. Isoniazid (Sigma-Aldrich) was used as a positive control.
The inoculated plates were incubated for 7 d with shaking at 80 rpm
at 37 °C. Cultures were manually resuspended, and absorbance
at 600 nm was measured using a Tecan Infinite M200 Pro spectrophotometer.
MIC50 and MIC90 represent, the concentration
of compound that inhibit 50% and 90% bacterial growth, respectively,
as compared to drug-free controls. Experiments were performed in biological
duplicates. The MBC was determined by CFU enumeration on complete
7H10 agar plates after exposure to a given concentration of test compound.
Bacterial (M. tuberculosis H37Rv) cultures
were grown to mid log phase, adjusted to a final OD600 =
0.05, and then treated with concentrations equivalent to 1×,
2×, and 4×x MIC90 of the test compounds for 7
d at 37 °C with shaking at 80 rpm. Drug-free cultures were plated
at the start of the experiment to determine the bacterial load of
the inoculum. After incubation for 7 days, the compound-treated cultures
were plated to determine CFU. MBC99.9 are defined as the
concentration of the test compound that caused 1000-fold reduction
in CFU as compared to the untreated inoculum at time point zero.
Hemolysis Assay
Fresh human red blood cells (RBCs)
were washed with PBS until the supernatant was clear after centrifugation.
The pellet was resuspended in PBS to an OD600 of 24 and
added (100 μL) to each well of a 96-well U-bottom plate. Candidate
compounds were serially diluted in PBS and added (100 μL) to
the wells. Triton X-100 was used as a positive control. After one
hour of incubation at 37 °C without shaking, cells were centrifuged
at 1000g for 15 min. The supernatant was diluted
and OD450 measured using a BioTek Synergy 4 microplate
reader. The experiment was done in triplicate.
Synthetic
and Analytical Chemistry
All reagents, starting
materials, and solvents (including dry solvents) were obtained from
commercial suppliers and used as such without further purification.
Reactions were carried out in oven-dried glassware under a positive
pressure of argon unless otherwise mentioned. Air-sensitive reagents
and solutions were transferred via syringe or cannula and were introduced
to the apparatus via rubber septa. Reactions were monitored by thin-layer
chromatography (TLC) with 0.25 mm precoated silica gel plates (60
F254). Visualization was accomplished with either UV light, iodine
adsorbed on silica gel, or by immersion in an ethanolic solution of
phosphomolybdic acid (PMA), p-anisaldehyde, or KMnO4, followed by heating with a heat gun for ∼15 s. Column
chromatography was performed on silica gel (100–200 or 230–400
mesh size). All final compounds have purity ≥95 as determined
by Shimadzu Prominence UFLC using a reverse phase column [Phenomenex
C18 column (5 μm, 250 mm × 4.60 mm)] and a solvent gradient
of A (0.05% TFA in H2O) and solvent B (0.039% TFA in 90%
ACN and 10% H2O). Method A: Solvent B, 0–100% in
25 min. Method B: Solvent B, 0–60% in 10 min to 60–100%
in 10 min. Deuterated solvents for NMR spectroscopic analyses were
used as received. All 1H NMR and 13C NMR spectra
were obtained using a 400 MHz spectrometer. Coupling constants were
measured in hertz. All chemical shifts were quoted in ppm, relative
to TMS, using the residual solvent peak as a reference standard. HRMS
(ESI) were recorded on a 6520 QTOF mass spectrometer equipped with
a dual-spray electrospray ionization source (Agilent Technologies,
Santa Clara, CA). Chemical nomenclature was generated using Chem Bio
Draw Ultra 14.0.
General Procedure for Reductive Amination
Compound 8 (100 mg, 0.32 mmol) and corresponding amine
(0.64 mmol)
were suspended in MeOH (5.0 mL), and a catalytic amount of acetic
acid was added with stirring at ambient temperature for 30 min. NaCNBH3 (0.96 mmol) was added in portions over 5 min. The reaction
mixture was stirred at room temperature for 24 h. Upon completion
of the reaction, the solution was diluted with 10% aqueous sodium
bicarbonate (5.0 mL) and extracted with ethyl acetate (3 × 5.0
mL) or 10% MeOH in dichloromethane (4 × 10 mL). Organic layer
was washed with brine (5.0 mL) and dried over anhydrous Na2SO4. The product was purified using column chromatography
(5–10% methanol in chloroform) to get the desired compound
as off-white solids.
General Procedure for Transamidation
To a solution
of amines 35 and 36 (1.5 equiv) and ester 4 (1.0 equiv) in anhydrous ethanol was added trimethylamine
(3.0 equiv), and the resulting reaction mixture was stirred under
vigorous reflux for 24 h, and the reaction mixture was cooled, evaporated,
and purified by column chromatography using 5–15% MeOH in dichloromethane.
For Boc-Protected Secondary Amines
Products obtained
above were dissolved in dry dichloromethane and treated with 20–30
equiv of 4 N HCl in dioxane at 0 °C and stirred at ambient temperature
for 4–6 h. After complete consumption of starting material,
the reaction mass was evaporated and washed with diethyl ether to
afford an off-white solid as their HClsalts.
General Procedure for the
Synthesis of O6-Substituted
Thienopyrimidinone Analogues
Under argon atmosphere alkyl
or aryl alkyl bromide (2.0 equiv) was added to a suspension of compounds
(37, 38, and 10, 1.0 equiv)
and K2CO3 (3.0 equiv) in anhydrous DMF at 0
°C, with the resulting reaction mixture was stirred at ambient
temperature for 3–4 h. The progress of the reaction was monitored
by TLC and was eventually quenched with saturated aqueous NH4Cl and extracted with EtOAc (3 × 10 mL). The combined organic
layers were washed with brine (10 mL), dried over anhydrous Na2SO4, and evaporated under reduced pressure. The
crude material was purified by column chromatography (silica gel 200–400
mesh) using 5–10% MeOH in dichloromethane to afford product
as off-white solids. O6-Substituted products obtained above
were dissolved in dry dichloromethane and treated with 20–30
equiv of 4 N HCl in dioxane at 0 °C and stirred at ambient temperature
for 4–6 h after complete consumption of starting material the
reaction mass was evaporated and washed with diethyl ether to afford
off-white solid as their HClsalts (39–44, 47).
Authors: K Kobayashi; S D Ehrlich; A Albertini; G Amati; K K Andersen; M Arnaud; K Asai; S Ashikaga; S Aymerich; P Bessieres; F Boland; S C Brignell; S Bron; K Bunai; J Chapuis; L C Christiansen; A Danchin; M Débarbouille; E Dervyn; E Deuerling; K Devine; S K Devine; O Dreesen; J Errington; S Fillinger; S J Foster; Y Fujita; A Galizzi; R Gardan; C Eschevins; T Fukushima; K Haga; C R Harwood; M Hecker; D Hosoya; M F Hullo; H Kakeshita; D Karamata; Y Kasahara; F Kawamura; K Koga; P Koski; R Kuwana; D Imamura; M Ishimaru; S Ishikawa; I Ishio; D Le Coq; A Masson; C Mauël; R Meima; R P Mellado; A Moir; S Moriya; E Nagakawa; H Nanamiya; S Nakai; P Nygaard; M Ogura; T Ohanan; M O'Reilly; M O'Rourke; Z Pragai; H M Pooley; G Rapoport; J P Rawlins; L A Rivas; C Rivolta; A Sadaie; Y Sadaie; M Sarvas; T Sato; H H Saxild; E Scanlan; W Schumann; J F M L Seegers; J Sekiguchi; A Sekowska; S J Séror; M Simon; P Stragier; R Studer; H Takamatsu; T Tanaka; M Takeuchi; H B Thomaides; V Vagner; J M van Dijl; K Watabe; A Wipat; H Yamamoto; M Yamamoto; Y Yamamoto; K Yamane; K Yata; K Yoshida; H Yoshikawa; U Zuber; N Ogasawara Journal: Proc Natl Acad Sci U S A Date: 2003-04-07 Impact factor: 11.205
Authors: Karen O'Dwyer; Joseph M Watts; Sanjoy Biswas; Jennifer Ambrad; Michael Barber; Hervé Brulé; Chantal Petit; David J Holmes; Magdalena Zalacain; Walter M Holmes Journal: J Bacteriol Date: 2004-04 Impact factor: 3.490
Authors: Ulrike Begley; Madhu Dyavaiah; Ashish Patil; John P Rooney; Dan DiRenzo; Christine M Young; Douglas S Conklin; Richard S Zitomer; Thomas J Begley Journal: Mol Cell Date: 2007-12-14 Impact factor: 17.970
Authors: Ian W Davis; Andrew Leaver-Fay; Vincent B Chen; Jeremy N Block; Gary J Kapral; Xueyi Wang; Laura W Murray; W Bryan Arendall; Jack Snoeyink; Jane S Richardson; David C Richardson Journal: Nucleic Acids Res Date: 2007-04-22 Impact factor: 16.971
Authors: Tim R Fischer; Laurenz Meidner; Marvin Schwickert; Marlies Weber; Robert A Zimmermann; Christian Kersten; Tanja Schirmeister; Mark Helm Journal: Nucleic Acids Res Date: 2022-05-06 Impact factor: 19.160
Figure 1
Figure 2
Figure 3
Scheme 1
Figure 4
Scheme 2