Comparative analysis of pyrimidine substituted aminoacyl-sulfamoyl nucleosides as potential inhibitors targeting class I aminoacyl-tRNA synthetases

Article abstract of DOI:10.1016/j.ejmech.2019.04.003

Aminoacyl-tRNA synthetases (aaRSs) catalyse the ATP-dependent coupling of an amino acid to its cognate tRNA. Being vital for protein translation aaRSs are considered a promising target for the development of novel antimicrobial agents. 5′-O-(N-aminoacyl)-

Full text of DOI:10.1016/j.ejmech.2019.04.003

Accepted Manuscript
Comparative analysis of pyrimidine substituted aminoacyl-sulfamoyl nucleosides as
potential inhibitors targeting class I aminoacyl-tRNA synthetases
Manesh Nautiyal, Steff De Graef, Luping Pang, Bharat Gadakh, Sergei V. Strelkov,
Stephen D. Weeks, Arthur Van Aerschot
PII:
S0223-5234(19)30302-2
DOI:
https://doi.org/10.1016/j.ejmech.2019.04.003
EJMECH 11238
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 29 January 2019
Revised Date: 1 April 2019
Accepted Date: 1 April 2019
Please cite this article as: M. Nautiyal, S. De Graef, L. Pang, B. Gadakh, S.V. Strelkov, S.D. Weeks,
A. Van Aerschot, Comparative analysis of pyrimidine substituted aminoacyl-sulfamoyl nucleosides
as potential inhibitors targeting class I aminoacyl-tRNA synthetases, European Journal of Medicinal
Chemistry (2019), doi: https://doi.org/10.1016/j.ejmech.2019.04.003.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Comparative analysis of pyrimidine substituted aminoacyl-sulfamoyl nucleosides as
potential inhibitors targeting class I aminoacyl-tRNA synthetases.
#
#
Manesh Nautiyal1 , Steff De Graef2 , Luping Pang1,2, Bharat Gadakh1, Sergei V. Strelkov2, Stephen
∞,
D. Weeks2,∞ and Arthur Van Aerschot1,
*
1
Medicinal Chemistry, Rega Institute for Medical Research, KU Leuven, Herestraat 49 box 1041,
B-3000 Leuven, Belgium
2
Laboratory for Biocrystallography, Department of Pharmaceutical and Pharmacological Sciences,
KU Leuven, Herestraat 49 box 822, B-3000 Leuven, Belgium
* To whom correspondence should be addressed. Tel: +32 16 37 26 24; Fax: +32 16 3 37340;
Email: arthur.vanaerschot@kuleuven.be
# These authors contributed equally to this work.
∞
Both authors equally supervised and gave guidance to this work.
Abstract
Aminoacyl-tRNA synthetases (aaRSs) catalyze the ATP-dependent coupling of an amino acid to its
cognate tRNA. Being vital for protein translation aaRSs are considered a promising target for the
development of novel antimicrobial agents. 5'-O-(N-aminoacyl)-sulfamoyl adenosine (aaSA) is a
non-hydrolysable analog of the aaRS reaction intermediate that has been shown to be a potent
inhibitor of this enzyme family but is prone to chemical instability and enzymatic modification. In
an attempt to improve the molecular properties of this scaffold we synthesized a series of base
substituted aaSA analogues comprising cytosine, uracil and N3-methyluracil targeting leucyl-,
tyrosyl- and isoleucyl-tRNA synthetases. In in vitro assays seven out of the nine inhibitors
demonstrated Kiapp values in the low nanomolar range. To complement the biochemical studies, X-
ray crystallographic structures of Neisseria gonorrhoeae leucyl-tRNA synthetase and Escherichia
coli tyrosyl-tRNA synthetase in complex with the newly synthesized compounds were determined.
These highlighted a subtle interplay between the base moiety and the target enzyme in defining
relative inhibitory activity. Encouraged by this data we investigated if the pyrimidine congeners
could escape a natural resistance mechanism, involving acetylation of the amine of the aminoacyl
group by the bacterial N-acetyltransferases RimL and YhhY. With RimL the pyrimidine congeners
were less susceptible to inactivation compared to the equivalent aaSA, whereas with YhhY the
converse was true. Combined the various insights resulting from this study will pave the way for the
further rational design of aaRS inhibitors.
1

ACCEPTED MANUSCRIPT
Graphical abstract
Keywords
Aminoacyl-tRNA synthetase; bisubstrate competitive inhibitor; X-ray crystallography; N-
acetyltransferase; structure-activity relationship
Highlights
1. Nine aminoacyl-sulfamoylated pyrimidine nucleosides were synthesized.
2. Several compounds displayed nanomolar activity in in vitro assays.
3. LeuRS and TyrRS ligand bound X-ray crystallographic structures determined.
4. Activity determined in cell extracts in the presence of N-acetyl transferases.
5. N-acetyl transferase activity affected by both the base and the amino acid.
1. INTRODUCTION
Aminoacyl-tRNA synthetases (aaRSs) are a kingdom-wide family of enzymes that play a pivotal
role in translating the genetic code. These proteins, all catalyze the two-step ligation of an amino
acid to its cognate tRNA. In the first step, the amino acid is activated by nucleophilic attack of the
α-carboxylate on the α-phosphate of adenosine triphosphate (ATP) generating aminoacyl-AMP (aa-
AMP, Fig. 1a) and pyrophosphate. The aa-AMP is then utilised in the second step whereby the
aaRS catalyses the transesterification of the 2′ or 3′ -hydroxyl of a tRNA terminal ribose, generating
the charged aa-tRNA species[1]. In a typical bacterium there are 20 different aaRSs, one for each
proteinogenic amino acid. Given their indispensable role in maintaining the fidelity of translation,
aaRSs have been recognised as suitable targets for the development of new antimicrobials[2, 3].
Screening of both natural and synthetic molecules for aaRS inhibitors has yielded diverse sets of
2

ACCEPTED MANUSCRIPT
chemical structures[1, 4-7]. However, for various reasons only a limited number of compounds
have reached the market. The best example is mupirocin (Fig. 1b), a competitive catalytic site
inhibitor of isoleucyl-tRNA synthetase (ileRS), which is used as a topical treatment of impetigo
resulting from Staphylococcus aureus (S. aureus) or Streptococcus pyogenes (S. pyogenes)
infection.
Fig. 1. The aaRS catalytic intermediate aa-AMP (a) and some well-known aaRS inhibitors, with the
clinically approved mupirocin (b), general structure of aaSA analogues (c), and compounds
developed at Cubist Pharmaceuticals (d). R in structures a and c corresponds to the side chain of a
proteinogenic amino acid.
The largest group of aaRS inhibitors described so far are the non-hydrolysable analogues of aa-
AMP, where the labile mixed anhydride linkage has been replaced with a hydrolytically stable
linkage[8, 9]. Among them, the 5'-O-(N-L-aminoacyl)-sulfamoyl adenosine (aaSAs) analogues
(Fig. 1c) have proven to be potent inhibitors in vitro[10]. Although having a strong affinity for the
target aaRS, these compounds lack in vivo efficacy due to their high polarity, have limited species
selectivity due to structural similarity with aa-AMP, and are prone to hydrolysis as a result of the
formation of a cyclic adenosine derivative[11]. Furthermore, studies have shown that such
compounds are prone to enzymatic degradation[12] or modification[13].
To address some of these issues, various efforts were made to modify the aaSA scaffold to obtain
analogues with a better activity profile. Such approaches include the modification or shortening of
the sulfur-based linker[14], substitution of the sugar ring[15] or the adenine base[16, 17]. In the
3

ACCEPTED MANUSCRIPT
latter case, Cubist Pharmaceuticals (now Merck) reported C-nucleosides where the adenine was
replaced with either a functionalized thiazole[18] or ethyl-linked tetrazole derivatives[19] (Fig. 1d).
The tetrazole analogues, designed to target isoleucyl-tRNA synthetase (ileRS), showed IC50 values
in the nanomolar range[20] with up to 2300 times higher selectivity towards the S. aureus and E.
coli enzyme compared to the human orthologue[19]. Unfortunately, these compounds did not prove
successful in vivo due to low cell permeability and strong binding to serum albumin[5, 19].
Therefore, mimicking the natural antibiotics albomycin and microcin C, attempts were made to
combine the tetrazole containing compounds with bacterial transport modules to improve uptake.
While the compounds were correctly processed, they did not enter bacteria suggesting that the base
in the natural compounds is necessary to traverse the bacterial cell wall[17].
In addition to in vivo efficacy, selectivity and hydrolysis issues, aaSA analogues are prone to
undergo enzymatic degradation or inactivation by the action of peptidases[21],
acetyltransferases[13, 22], and potentially by sulfatases[12]. In the case of acetyltransferases,
members of the GCN5-related N-acetyltransferase (GNAT) family have been shown to catalyse the
acetyl-coenzyme A (acetyl-coA) dependent modification of the α-amine group of aaSAs preventing
interaction of the compound with the target aaRS and therefore providing resistance to bacteria.
Intriguingly, a recently discovered microcin C analog containing a modified cytidine base was
shown to overcome GNAT mediated inactivation[23, 24].
Expanding on our previous study on ileRS inhibitors where adenosine was replaced with alternative
natural and modified bases[16], we have investigated whether a pyrimidine substitution strategy
could be extrapolated to other aaRSs. In particular, we have focused on ileRS, leucyl-tRNA
synthetase (leuRS) and tyrosyl-tRNA synthetase (tyrRS). All three enzymes belong to the class I
aaRS family, a subset with ten members sharing a catalytic domain based on a Rossmann fold[4].
We synthesized nine compounds [5-7(a-c)] substituting adenine for cytosine (C), uracil (U), or N3-
methyluracil (3meU). The synthesized congeners were evaluated for their ability to inhibit the
corresponding aaRS in in vitro assays with both the purified enzyme and S30 E. coli extract. In the
latter case the ability of the compounds to escape NAT catalysed inactivation was evaluated. To
complement the biochemical analyses, high-resolution X-ray crystallographic structures were solved
of the respective compounds bound to Neisseria gonorrhoeae (N. gonorrhoeae) leuRS and E. coli
tyrRS permitting the development of a structure-activity relationship model.
2. Results
Synthesis of pyrimidine aaSA analogues
4

ACCEPTED MANUSCRIPT
Synthesis of the desired compounds is straightforward and shown in Scheme 1, starting with
persilylation of cytidine (1a) and uridine (1b)[16]. To obtain the N3-methyluridine analog (2c),
persilylated uridine (2a) was treated for 20 h with MeI in the presence of K2CO3 using DMF:
acetone (1:1) as the solvent[25]. Selective desilylation at the 5'-O position was accomplished using
TFA: water: THF (4:1:1) to afford 3a-c[26] and reaction with in situ generated sulfamoyl chloride
lead to the formation of 5'-O-sulfamoylated analogues 4a-c. The 5'-O-sulfamoyl derivatives of
uridine (4b) and N3-methyluridine (4c) were condensed with the N-hydroxysuccinimide ester of
appropriately protected amino acids in the presence of DBU. Acidolysis of intermediates was
performed using TFA: H2O (5:2) to remove Boc and TBDMS protecting groups, which was
followed by hydrogenolysis for the tyrosine derivatives, leading to the desired 5'-O-(N-L-
aminoacyl)sulfamoylated cytidines (aaSC, 5a-6a), uridines (aaSU, 5b-7b), and N3-methyluridines
(aaS3meU, 5c-7c). Only 7a proved cumbersome with reduction of the cytosine base upon
hydrogenolysis. Hence, in this case the sidechain unprotected N-hydroxysuccinimide ester of Boc-
Tyr-OH was used for coupling to 5'-O-sulfamoyl cytidine in the presence of DBU.
In the remainder of the manuscript, we will designate both the synthesized compounds and the
aaSA analogues using a natural abbreviation, consisting of a letter (L, I or Y) indicating the amino
acid, followed by the letter S for sulfamoyl, and finally abbreviated designation of the base (A, U, C
or 3meU). This way, compound 5a becomes ISC etc.
O
O
B
O
B
B
HO
TBDMSO
(ii)
HO
(i)
HO
TBDMSO
OH
OTBDMS
TBDMSO
OTBDMS
2a, B= Cytosine
1a, B= Cytosine
1b, B= Uracil
3a, B= Cytosine
3b, B= Uracil
3c, B= N3-Me-Uracil
2b, B= Uracil
2c, B= N3-Me-Uracil
(iii)
(iv)
NH2
O
O
O
R
O
B
H
B
S
N
O
O
O
H2N
TBDMSO
4(a-c)
S
(v), (vi), (vii)
O
O
HO
OH
OTBDMS
5(a-c)-7(a-c)
5a-c :R = CH(CH3)CH2CH3:, 6a-c : R = CH2CH(CH3)2;7a-c : R = CH2C6H4-p-OH,
for 4-7, a: B = Cytosine, b: B = Uracil c: B = N3-Me-Uracil.
Scheme 1. Synthesis of 5'-O-(N-L-aminoacyl)-sulfamoyl nucleosides 5(a-c), 6(a-c) & 7(a-c).
Reagents and conditions: (i) TBDMSCl, imidazole, dry DMF, 50 oC, 20 h; 82-89% (ii) TFA: water:
5

ACCEPTED MANUSCRIPT
o
o
THF (4:1:1), 0 C, 6 h; 60-84% (iii) MeI, K2CO3, DMF: acetone (1:1), Reflux, 50 C, 20 h, 81%;
(iv) ClSO2NCO, formic acid, DMA, 66-72%; (v) Boc-aa-( OBn)-Osu (or Boc-Tyr-Osu for 7a),
DBU, DMF, 6-8 h; (vi) TFA: water (5:2), rt, 6 h; (vii) Pd/C, methanol, H2 atm. rt, overnight.
Cumulative yields from steps v-vii were: 5c-74%, 6a-58%, 6b-82%, 6c-65%, 7a-30%, 7b-62%, 7c-
60%.
In vitro assays with purified aaRSs
The inhibitory activity of the synthesized derivatives was tested by performing an in vitro
aminoacylation reaction and measuring the transfer of radiolabelled amino acid to tRNA. Reactions
were carried out with a total tRNA pool and the appropriate purified enzyme from E. coli. In view
of the strong inhibitory activity of the compounds under evaluation, dose-response data were fitted
using the Greco-Hakala equation[27] (Fig. 2) and data were compared to the respective aaSA
analogues (Table 1). While being weaker inhibitors than the aaSA analogues, all three uridine
derivatives ISU, LSU and YSU (5b-7b) showed Kiapp values in the nanomolar range against the
respective enzymes (Table 1). N3-methylation of this base (5c-7c) resulted in slightly weaker
inhibition of tyrRS and leuRS, but led to an almost 20-fold reduction in activity of ileS3meU
compared to ileSU, when evaluated against ileRS. The cytidine analogues (5a-7a) proved slightly
more potent against leuRS and ileRS compared to the uridine congeners (5b-7b), with Kiapp values
of 10 nM and 17 nM for LSC and ISC, respectively. Remarkably, the inhibitory activity of YSC
was strongly reduced resulting in a Kiapp of around 1.5 micromolar.
Fig. 2. Dose-response curves of leuRS, ileRS, and tyrRS from E. coli in the presence of serial
dilutions of the pyrimidine analogues. The reactions contained 2.5 nM leuRS, 10 nM ileRS and 0.5
nM tyrRS. Values were normalised to the control reaction containing no inhibitor. Points were fitted
with the Greco-Hakala equation. All reactions were performed in triplicate, the average value and
standard error are shown; red, green and black curves indicate cytidine, N3-methyluridine and
uridine derivatives respectively.
6

ACCEPTED MANUSCRIPT
Table 1. Enzyme inhibition values of the studied sulfonamides.
aaRS
Aa
U
C
3meU
IleRSb
1.9 ± 4.0
55.0 ± 7.6
20.8 ± 1.1
43.4 ± 3.4
17.2 ± 1.5
9.8 ± 0.4
1039 ± 122
29.5 ± 1.8
LeuRSb 0.14 ± 0.1
TyrRSb
2.9 ± 1.2
1484 ± 112 149 ± 12.0
Kiapp values of the aminoacylated sulfonamide nucleosides for the respective enzymes are given in
nM, for the adenosine (A), uridine (U), cytidine (C) and N3-methyluridine (3meU) analogues.
aThe Kiapp values for adenosine derivatives were taken from our prior work[28].
b10 nM ileRS, 2.5 nM leuRS, and 0.5 nM tyrRS were used in all reactions. The Kiapp values were
calculated utilizing Greco-Hakala equation, maintaining the input enzyme concentration as a
constant.
Crystallography of inhibitor-bound aaRS complexes
To obtain further insight into the relative inhibitory activity of the base-substituted intermediate
analogues, X-ray crystallographic studies of the compounds in complex with their respective target
aaRS were initiated. After extensive crystallization screening of recombinant leuRS and tyrRS from
multiple bacterial species and subsequent optimization, diffracting crystals were obtained for E. coli
tyrRS and N. gonorrhoeae leuRS. LeuRS from the latter gram-negative organism is a close
homologue to the E. coli enzyme, with an overall sequence identity of 57%. Further analysis shows
a 92% sequence identity (100% similarity) for the residues within 6 Å of LSA (see Supplementary
Fig. S1). Therefore we can extrapolate the information obtained from the solved structures to the
activity measurements obtained with E. coli leuRS.
Crystals of both aaRSs readily diffracted to 2.3 Å or better, following soaking with the appropriate
compounds 6a-c and 7a-c, or the analogous aminoacyl-sulfamoyl adenylate (see Supplementary
Table 1 and 2). The resulting calculated electron density maps unambiguously allowed for the
positioning of each compound. In the case of tyrRS, which presents itself as the natural homodimer
in the crystallographic asymmetric unit, both active sites are occupied with the soaked inhibitors
(see Supplementary Fig S2). Furthermore, for both enzymes, the refined structures gave an
excellent real-space correlation coefficient (RSCC) for the modelled ligand confirming their overall
quality (see Supplementary Table 2).
Structures of inhibitor-tyrRS complexes
7

ACCEPTED MANUSCRIPT
In each of the four inhibitor-tyrRS complexes solved, both active sites of the dimer are substantially
equivalent. For the four structures, superposition of the immediate ligand environment, defined as
residues within 6 Å of the compound, in chains A and B respectively reveal an all-atom RMSD
between 0.275 Å and 0.522 Å. Therefore, we will only focus on chain A of each structure for our
description of the differences between the compounds. Here, pairwise comparison of the ligand
environment in the YSA-bound structure with each of the pyrimidine-bound structures gives all-
atom RMSDs not exceeding 0.2 Å. Since this value is even lower than the obtained equivalent all-
atom RMSD between chains A and B, it indicates that the measured difference in Kiapp between
compounds 7a-c and YSA are the result of subtle structural differences. Reflecting the similarity of
amino acid positions in the active sites of the four ligand-bound structures, examination of the
inhibitor poses in the superposed models shows that the tyrosyl, sulfamoyl and ribose adopt nearly
equivalent confirmations. We will therefore, limit our further description to the region surrounding
the base moieties.
Within the perfect intermediate analog YSA, adenine makes a limited number of interactions with
the active site. It is stacked between Gly50 of the HIG50H-motif present in α-helix 3 and a non-
conserved branched amino acid, Leu227 (Fig. 3). The N6 and N1 atoms make backbone H-bond
interactions with the carbonyl and amino group of Ile228, a backbone interaction conserved in all
other class I aaRSs[28]. In comparison to YSA, the base of the three pyrimidine-substituted
congeners 7a-c are sitting in the same plane as adenine, but distinct differences are observed. For
YSC and YSU the respective bases adopt the same position as each other, sitting in the region that
would be occupied by the imidazole ring of the adenine (Fig. 3). Both make Van der Waals
interactions with Gly50, however, a likewise interaction with Leu227 is lost. In contrast to the
adenine which forms two direct H-bonds to the protein, only indirect H-bonding interactions via a
water molecule, known as a water bridge, are observed for the pyrimidine bases. Both the natural
bases make such a water bridge via the O2 atom, with the backbone O atom of Pro226. In the case
of YSC, an extra water bridge between the N3 and the backbone N atom of Ile228 is observed.
Modification of the N3 position with a methyl group changes the position of the base in the
YS3meU bound structure. While it still sits in the same plane as observed for the uracil analog
YSU, the methylated base is tilted 21.5° into the adenine binding pocket, rotating around the C1'-
O4 axis of the ribose, where the O4 atom shows a 2.3 Å shift relative to the equivalent atom in YSU
(see Supplementary Fig. S3A). The tilted position of the N3-methyluracil resembles more the
position of the adenine base, since the N1, C6 and C5 atoms of YS3meU are in the same position as
8

ACCEPTED MANUSCRIPT
N9, C8 and N7 of adenine and the methyl group lies in a position between the N1-C2 atoms of the
adenine base (Fig. 3). In comparison to the two other pyrimidine-based congeners, no water
molecules are observed in the proximity of the methyl-modified base. In addition Leu227 makes
Van der Waals interaction with the methyl group. The relative rotation of the N3-methyluracil is
further driven by stabilizing Van der Waals interactions of the methyl group with the side chain of
Val53 and the backbone of Pro226.
Fig. 3. Active site interactions of the tyrosyl sulfamoyl nucleoside congeners with E. coli
tyrRS. Structures are shown in order of decreasing activity of the compound from left to right. Top;
3D representation of the base mediated interactions. Each ligand and the interacting protein residues
are shown as a stick representation. The protein backbone is shown in green as a cartoon
representation. H-bonds are shown as black dashes and water molecules as red spheres. For the
three pyrimidine-based compounds the comparative position of the adenine base is shown as a
transparent cyan stick representation. Bottom; 2D representation of the protein-compound
interactions. H-bonds are shown as black dashes, stacking interactions as green dashes.
Structures of inhibitor-leuRS complexes
Our crystal structure of the N. gonorrhoeae leuRS with bound LSA reveals that the adenine base of
the inhibitor is stacked between Gly51 of the HIG51H-motif contain
α-helix and a branched
hydrophobic amino acid, in this case, Met582. This stacking is conserved in all class I aaRS and
also observed in our tyrRS bound structure with the respective analog. The base N6 amine interacts
with the protein backbone of Met635, which is part of the class I conserved KM635SKS loop.
9

ACCEPTED MANUSCRIPT
Furthermore, N1 and N6 are making H-bond interactions with the protein backbone of Val584, and
the N3 atom interacts with the sidechain of Gln580 (Fig. 4).
Pairwise superposition of the protein environment within 6 Å of the ligand in the LSA-bound
structure in each of the structures with pyrimidine-based inhibitors gives all-atom RMSD close to
1.0 Å. This indicates more substantial changes in the vicinity of the inhibitor compared to the tyrRS
complexes. Examination of the inhibitor poses in the superposed models shows that the leucyl,
sulfamoyl and ribose moieties adopt equivalent confirmations and that structural differences
between the inhibitors are base-related. We will therefore limit our further description to the base
moieties.
The increase in RMSD of atoms surrounding the inhibitors is the result of two distinct differences.
Firstly, none of the pyrimidine-substituted inhibitors make an H-bond with the backbone atoms of
Met635, this results in this residue and the surrounding KM635SKS loop being found 1.8 Å away
from the position observed in the LSA bound structure. Secondly, in all cases, the side chain of
Met582 is rotated 90° outward on the Cβ atom, reducing the stacking interaction with the different
pyrimidine congeners (see Supplementary Fig. S3B). The base position of LS3meU resembles the
position of the adenine base of LSA as the N1, C6 and C5 atoms of the former compound are in the
same position as the N9, C8 and C7 of LSA (Fig. 4). The cytosine and uracil bases are rotated
compared to the position of the bases of LS3meU and LSA. The cytosine base of LSC is rotated 24°
with respect to the C1’-O4 axis of LS3meU, resulting in a relative movement of the N4 atom by
2.3Å from the corresponding atom of LS3meU (see Supplementary Fig. S3C). The uracil base of
LSU shows a similar movement as LSC. The position of the uracil and cytosine base relative to the
ribose is different compared to the 3-methyluracil base as the dihedral angle is -131° compared to -
152° for LSC and LSU. To accommodate the position of the base of LSC and LSU, the sidechains
of His49 and His52, part of the class I conserved H49IGH52-motif, are repositioned. Relative to the
base position His52 is moved backwards 1.9 Å and, in the case of LSU, the imidazole ring is
additionally tilted 73°. The movement of His49 is more pronounced, with both the uracil and
cytidine congeners the residue backbone atoms are shifted closer to the base moiety (1.6 Å for the
Cα atom), consequently the imidazole ring is displaced 3 Å and rotated 72°, this rotation allows the
N4 atom of LSC to form a hydrogen bond with the ND1 atom of His49. This latter interaction is
nonexistent in the case of LSU as the imidazole ring is tilted 122° compared to the position when
LSC is present (see Supplementary Fig S3B). The space opened up by the rotation of the
10

ACCEPTED MANUSCRIPT
pyrimidine bases of LSC and LSU out of the pocket is filled by two water bridges. One is formed
between the N3 atom and the backbone nitrogen of Val583, and the other one links the O2 atom
with the backbone carbonyl group of Gly581 (Fig 4). Since the binding mode and the interactions
are similar, it is not surprising that the Kiapp for LSC and LSU are very close (Table 1).
Fig. 4. Active site interactions of the leucyl-sulfamoyl nucleoside congeners with N.
gonorrhoeae leuRS. Structures are shown in order of decreasing activity of the compound from left
to right. Top: 3D representation of the base-mediated interactions. Each ligand and the interacting
protein residues are shown as a stick representation. The protein backbone is shown in green as a
cartoon representation. H-bonds are shown as black dashes, and water molecules are shown as red
spheres. For the three pyrimidine-based compounds the comparative position of the adenine base is
shown as a transparent cyan stick representation. Bottom: 2D representation of the protein-
compound interactions. H-bonds are shown as black dashes, stacking interactions as green dashes.
Activity in S30 cell extracts
As the next step, we have validated the potency of the synthesised inhibitors in cellular extract. To
this end, respective aaRS activity was evaluated using an E. coli NCIB 8743 S30 extract pre-
incubated with the inhibitors. The respective aaSA analogue was taken each time as a positive
control (Fig 5A). These S30 extract experiments revealed a strong correlation with the inhibitory
activity observed in the purified enzyme assays. For leuRS and ileRS, it was confirmed that cytidine
11

ACCEPTED MANUSCRIPT
and uridine showed more potent activity (5-14% aa incorporation) compared to the N3-
methyluridine derivatives (17-34% aa incorporation). Likewise, in line with the enzymatic results,
YSU and YS3meU targeting tyrRS showed 37-62% tyrosine incorporation and thus proved less
inhibitory while YSC was devoid of inhibitory activity at the 250 nM concentration. The aaSA
analogues showed the most potent activity (0.05-1.8% aa incorporation), in full agreement with the
in vitro data.
Fig. 5. Evaluation of aaRS inhibitors in cellular extracts (A). Relative inhibitory activity of aaSA
analogues in E. coli NCIB 8743 cellular extracts. Activity was determined by measuring the transfer
of the appropriate 14C-labeled amino acid to tRNA in the presence of 250 nM of each compound
[aaSA analogues (black bars), aaSU (grey bars), aaSC (white bars) and the aaS3meU (vertical line
pattern)]. (B-D). Effect of different cellular extracts [5% overexpressed (RimL / RimI / YhhY /
SUMO) + 95% E. coli BL21 (DE3) cellular extract] on the inhibitory activity of adenine, uracil and
cytidine congeners targeting ileRS (B), leuRS (C) and tyrRS (D). The stated aaSA analogues were
incubated at 25°C, in the cell extract for 5 and 60 minutes, after which aliquots were taken and the
aminoacylation activity was measured. The relative activity was determined by comparing to values
12

ACCEPTED MANUSCRIPT
of the extract measured in the absence of any inhibitor and assuming 100% enzyme activity. The
results correspond to an average of three experiments, with SD presented as error bars.
Inhibitory results in cellulo in the presence of NATs
While our cell extract evaluations (Fig 3A) were in full agreement with the in vitro activity (Table
1), earlier in vitro data suggested ISU to be the best ileRS inhibitor, with ISC showing a similar
inhibitory activity as ISA[16]. These discrepancies could have been caused by several phenomena
which depend on the particular S30 cell extract and experimental conditions. The first potential
possibility is an off-target binding of the inhibitors to other enzymes than aaRSs in a particular
cellular assay. In particular, such off-target binding for ISA could have contributed to its lower
apparent activity in the previously published experiments. The second possibility, also resulting in a
changed inhibitory profile, is the loss of active compound due to chemical instability or their
biochemical modification, such as resulting from the activity of N-acetyltransferases (NATs).
To evaluate the stability of the various sulfamoyl congeners in cellular extracts, we incubated these
compounds with the S30 extract at 37 °C for 48 hours, taking samples at different time points and
assessing the ileRS activity in the radiotransfer assay. We did not observe any degradation of
compounds, as the level of ileRS activity remained unaltered during the course of the experiment,
where any decline correlated with a loss in activity of the enzyme as a result of extract instability
during the prolonged incubation (see Supplementary Figs. S4 & S5). These results contrast with
earlier studies of DSA and LSA, in a wild-type E. coli cellular extract that showed a clear and rapid
loss in inhibitor activity[13].
Next, acetylation of the α-amine of aaSAs and processed microcin C is known to be the principal
mechanism by which bacteria inactivate these compounds[13, 29]. The C-terminal domain of
MccE, the enzyme responsible for acetylation of microcin C, and the homologous bacterial N-
terminal acetyltransferases RimL and YhhY, have been reported to be responsible for the
acetylation of aaSA analogues requiring acetyl-CoA[13, 22]. We therefore hypothesized that upon
supplementing the E. coli NCIB 8743 S30 extract with acetyl-CoA we could mimic the assay
conditions used by Kazakov et al[13]. However, even upon addition of 0.5 mM of acetyl-CoA, we
did not detect any changes in the order of inhibitory activities (see Supplementary Fig. S6).
It has been described in the literature that NAT expression levels can vary dramatically depending
on the strain and culture conditions[30]. This, therefore, prompted us to investigate whether the
controlled overexpression of the aforementioned NATs affected the measured inhibitor activity. We
decided to focus here only on the compounds carrying the natural bases cytosine or uracil, as the
13

ACCEPTED MANUSCRIPT
best activity was obtained for the uridine analogues among the isoleucyl sulfamoyl derivatives[16]
and a recent paper reported that cytidine based compounds could also escape the GNAT based
modification[23, 24]. We overexpressed E. coli RimL, YhhY, RimI and the His-tagged small
ubiquitin-related modifier (SUMO) in the E. coli BL21(DE3) pLysS strain (see Supplementary Fig
S7). RimI, a NAT that has been previously shown not to acetylate aaSAs was used to evaluate
specificity. SUMO is a protein of similar molecular mass to the three NATs under study and was
used as negative control to ensure that any observed loss of inhibitory activity of the compounds
could not be associated with overexpression using the T7 RNA polymerase expression system or
the presence of the recombinant plasmid in the cells. To improve signal in the assay the extract from
the overexpression cell lines, which were grown to late stationary phase, were mixed with an extract
prepared from the untransformed BL21(DE3) E. coli pLysS cells harvested in mid log-phase
growth. In all experiments the reaction was supplemented with physiologically relevant amounts of
acetyl-CoA to drive the acetylation reaction.
Each tested inhibitor was preincubated at 25°C with the different extracts and a sample taken at 5
min and 60 min for measurement of target aaRS activity (Fig 5B-D). For both the RimI and the
SUMO extracts no significant effect on the activity of the inhibitor was observed. The relative
pattern of the two tested pyrimidines, compared to each other and the respective adenosine analog,
was the same for each target aaRS as was observed with the E. coli NCIB 8743 S30 extract (Fig
5A). For example, at the 250 nM tested concentration YSA shows potent inhibitory activity and is
better than YSU, while YSC shows no capacity to inhibit tyrRS as would be expected due to this
compound being at a concentration below its Kiapp (Fig 5A and 5D, Table 1). In contrast to the two
control experiments, all studied compounds showed a significant decrease in inhibitory activity due
to acetylation as a result of the presence of RimL in the cellular extract (Fig 5B-D). The isoleucyl
and leucyl derivatives showed a rapid loss of inhibitory activity after 5 minutes, which remained
constant over the following 55 minutes in the presence of RimL. Only ISC demonstrated slower
inactivation by RimL, yet after 60 minutes of pre-incubation the same level of ileRS inhibition was
obtained as observed for the two other base substituents (Fig 5B). For the isoleucyl derivatives
preincubated in the RimL S30 extract, comparison of the 5 min preincubation timepoints shows that
acetylation of ISA occurs more rapidly compared to the pyrimidine containing analogues. For ISU
and ISA the remaining ileRS activity after 5 minutes is similar; however the Kiapp for ileRS of ISA
is 29-fold higher compared to ISU (Table 1). Therefore, to obtain these values, the relative amount
of active non-acylated ISA molecules is lower compared to that of ISU after 5 minutes of
incubation.
14

ACCEPTED MANUSCRIPT
In the presence of YhhY, the isoleucyl derivatives showed apparent slower acetylation kinetics
when compared to the RimL extract results, as after 5 minutes of preincubation no loss of inhibitory
activity was observed. After 60 minutes of incubation a reduced inhibitory activity, yet better
compared to the final levels in the RimL extract, was observed for all tested compounds (Fig. 5B).
While the difference in kinetics can be attributed to a difference in expression levels between the
two NATs (see Supplementary Fig. S7), YhhY demonstrates a bias for the two pyrimidine
containing compounds as the remaining inhibitory activity is significantly less compared to ISA.
(Fig. 5B). In the presence of RimL, all leucyl derivatives were rapidly but still incompletely
acetylated (all tests showing around 50% inhibitory activity), and this level remained constant over
60 minutes. Therefore, in view of the 70-fold difference in Kiapp values (Table 1), it can be
concluded that the adenosine analogue is more readily acetylated than the cytidine congener. The
behavior of the leuRS targeting compounds in the presence of YhhY is overall similar to what was
seen with the ileRS compounds. In particular the remaining inhibitory levels of LSC and LSU are
lower compared to LSA after 60 minutes of preincubation in the cell extract. However, compared to
the isoleucyl congeners, YhhY acetylated the leucyl-based pyrimidine substitutions much faster, as
after 5 minutes of incubation almost a complete eradication of inhibitory activity is observed. (Fig
5C). Next, YSA and YSU analogues showed some increase in acetylation over 60 minutes upon
treatment with RimL extract, whereas YSC was inactive from the start as shown by the inactivity in
the RimI and SUMO extracts, and correlating with the fact that the Kiapp is above the used inhibitor
concentration (Fig. 5D). Most remarkable however, is the somewhat higher inhibitory effect of
YSU over YSA, after 5 minutes in the RimL extract, while being equally inhibitory after 1 hour.
This result again points to the fact that RimL more rapidly acetylates adenine-containing
compounds over pyrimidine congeners. On the contrary, YhhY seems to prefer the latter as no
reduction in inhibitory activity is observed for YSA while YSU shows a statistically significant
reduction in the capacity to inhibit tyrRS. Next to the base moiety the amino acid part on the
inhbitors also seems to influence the activity of the NAT, since in the presence of YhhY, the
inhibitory activity of YSU is higher compared to the isoleucyl and leucyl analogues (Fig 5D). The
tyrosyl moiety also affects the activity of RimL as after 5 minutes of incubation YSU is less
inactivated compared to the latter compounds. In summary, we established that RimL can
deactivate all tested compounds, with the enzyme showing a preference for adenine containing
congeners, while the corresponding pyrimidine derivatives are, relatively, less affected. However
the pyrimidine congeners suffer more from the presence of YhhY, as the reduction of inhibitory
15

ACCEPTED MANUSCRIPT
activity is observed faster and the final levels of remaining activity are less compared to the
equivalent adenine containing compounds.
3. DISCUSSION
Aminoacyl-tRNA synthetases have been recognized as promising targets for the development of
new antimicrobials[1, 4, 19]. This is most clearly evidenced by the clinical application of the natural
product mupirocin (Bactroban), a competitive inhibitor that specifically targets bacterial ileRS[31].
Despite this, attempts to target other members of this conserved family of enzymes have resulted in
limited success. Replicating the mode of action of mupirocin, bisubstrate inhibitors that typically
target the enzyme amino acid and ATP binding sites simultaneously, have proven to have the most
potent activity. Amongst the many scaffolds that have been described, aminoacyl-sulfamoyl
adenosines, non-hydrolysable mimetics of the enzymatic intermediate (Fig 1a and 1c) have family-
wide activity[1]. Previously it was shown that pyrimidine base-containing isoleucyl-sulfamoyl
nucleoside analogues had a strong ileRS inhibitory potency. The activity of these inhibitors in
whole cell lysates was even superior to that of the adenine-based compound[16]. The latter result
was somewhat surprising considering the reported specificity of aaRSs for ATP as a substrate over
other nucleoside triphosphates[32].
In the current study, we have systematically evaluated the activity of a series of pyrimidine-
substituted non-hydrolysable intermediate analogues against ileRS and two other class I enzymes,
leuRS and tyrRS. In total nine compounds, corresponding to three different pyrimidines for each
tested aaRS were successfully synthesized and their inhibitory activity was investigated in vitro.
Using the appropriate isolated enzyme from E. coli, seven of the inhibitors, the exceptions being
YSC and IS3meU, demonstrated a Kiapp above 150 nM, with the best inhibitor LSC having a value
below 10 nM (Fig. 2 and Table 1). Therefore, substitution of the adenine base with a pyrimidine is
not too destructive for inhibitors targeting class I aaRS members.
To obtain a molecular understanding of the inhibitory mechanism of the pyrimidine congeners and
to explain the pronounced differences in activity between them, crystallization trials for all targeted
aaRSs supplemented with the inhibitors were undertaken. High-resolution structures were obtained
for E. coli tyrRS and N. gonorrhoeae leuRS bound to the different synthesized inhibitors providing
insight into the specific enzyme-ligand interactions. Crystal structures for both enzymes in complex
with a non-hydrolysable intermediate analogue have been reported before[33-35]. The new
structures confirm the previously observed interactions with the amino acid and ribose, where the
16

ACCEPTED MANUSCRIPT
notable differences are related to residues in the active site that surround the base moiety and the
positioning of the different pyrimidines relative to each other and the adenine congener.
In all class I aaRSs, the canonical base is found sandwiched between the highly conserved HIGH-
motif containing α-helix and a hydrophobic residue, while making a number of H-bond interactions
with the protein backbone (Fig. 3 and 4)[28]. Importantly, our newly established structures reveal
that while the pyrimidine bases go to the same pocket, they do not form direct H-bonds to the
protein. Instead, H-bond interactions are only realized via ordered water molecules, i.e. water
bridges. This general observation is likely to explain the lesser potency of the pyrimidine-based
analogues compared to the original adenine-containing compounds.
In addition, the obtained aaRS structures with the respective pyrimidine-based compounds reveal
subtle structural differences in the binding mode of the base. At the same time, observation of these
structural features appears to have only a limited predictive power with respect to the observed
inhibitory activity. For the pyrimidine analogues that demonstrated the best inhibition, binding to
either aaRS is accompanied by the formation of one or two water bridges. However, the effect of
water recruitment on the potency of the compound seems to be different between the enzymes. In
the case of leuRS, the bases of LSU and LSC coordinate two water molecules and are more
effective compounds than LS3MeU where the additional methyl group prevents water binding. In
contrast, a similar number of bridging water molecules are observed in the tyrRS-YSC structure,
however the Kiapp for this compound is above 1 µM. This value is 34-fold higher than that of YSU,
which only coordinates one water molecule in the bound structure (Fig. 3). Next to the biophysical
parameters affecting ligand efficiency, the ability of the compounds to induce conformational
changes in the active site differs between leuRS and tyrRS. In the leuRS structure bound to LSA, an
additional stacking interaction mediated by Met582 is observed with the canonical purine group,
and the downstream conserved KM635SKS loop moves towards the base to interact with the
hydrogens of the N6 atom resulting in a more compact pocket. In contrast all pyrimidine congeners
targeting leuRS do not induce these changes (Fig. 4). In contrast, in the ligand bound tyrRS
structures the positioning of residues in the active site around the base are similar for both the
adenine and the three pyrimidines (Fig. 3).
The class I aaRS family consists of five structurally defined subclasses[32]. LeuRS and ileRS
belong to subclass IA alongside methionyl- and valyl-tRNA synthetase, while the dimeric tyrRS is a
member of subclass IC that also includes tryptophanyl-tRNA synthetase[36]. Since we observed
distinct differences in the order of inhibitory activity of the pyrimidine congeners between the
studied aaRS, that reflect this structural classification (Table 1), cross-class extrapolation of the
17

ACCEPTED MANUSCRIPT
present results should be limited to each aaRS subclass. The majority of the remaining aaRSs, such
as arginyl, glutaminyl- and glutamyl-tRNA synthetase (gluRS) are unable to catalyze the formation
of the aminoacyl-adenylate intermediate in the absence of tRNA, as a conformational change
induced by binding of the latter macromolecule is necessary for correct active site arrangement[37-
39]. Despite this requirement, non-hydrolysable intermediate analogues have been shown to inhibit
these enzymes[40-42]. In particular, the aminoalkyl-adenylate intermediate analog of Glu-AMP has
a Ki of 3 µM. Substitution of the adenine in this inhibitor scaffold with various pyrimidines resulted
in a three order of magnitude loss of inhibitory activity[43]. Therefore, there is a high probability
that for the remaining subclasses of class I the equivalent pyrimidine congeners will also
demonstrate a reduced activity compared to the adenine analog, although the relative response
between the different pyrimidines is difficult to predict, particularly in terms of the subtle
differences in recruiting water molecules for bridging interactions or the induction of
conformational changes within the active site.
Importantly, both the in vitro biochemical and structural data presented here are in agreement in that
all tested pyrimidines demonstrate lower inhibitor activity than the canonical aaSA. Moreover, for
all twelve compounds tested, cell extract experiments fully confirmed the relative potency observed
in experiments with isolated aaRSs.
Upon activity determination in whole cell lysates, some subtle differences with previous findings
remain[16]. Since a vast plethora of enzymes are present and active at different concentrations
within such an extract, the compounds could be prone to enzymatic modification or degradation to
varying degrees as seen for other antimicrobials[44-46]. Primary amines of antimicrobial agents are
known to be targets for enzymatic modification leading to resistance. Specifically, Kazakov et al.
previously reported that non-hydrolysable adenylates are substrates for the GNATs RimL and
YhhY[13]. As the levels of these host enzymes and their substrate acetyl-CoA can vary
considerably in cells, we overexpressed them in E. coli and evaluated their effect on the pyrimidine-
based aaRS inhibitors in cellular extracts. In line with our hypothesis, we observed that all
congeners can be deactivated by RimL and YhhY (Fig. 5B-D). However, with RimL adenine is
clearly preferred over the three pyrimidines. In this case acetylation of the α-amino group results in
a parity of the inhibitory activity of the adenine containing compounds with the pyrimidine
analogues in the bacterial extract following 60 minutes of incubation. In contrast, YhhY does not
seem to have a base preference, although there appears to be a larger discrepancy between the
attached amino acid, with leucine more prone to modification over its isomer and tyrosine. These
results are in agreement with previous studies that observed that a large number of aaSAs were
18

ACCEPTED MANUSCRIPT
inactivated by RimL independent of the attached amino acid, while only non-polar amino acids
were affected by YhhY[13].
In particular, we have studied the effect of the MccE enzyme that has been identified in the operon
responsible for the production of the Trojan-horse antibiotic microcin C7, a natural non-
hydrolysable intermediate analog targeting aspRS[22]. Similarly to the host GNATs, the C-terminal
domain of MccE has shown to acetylate the α-amine of the active component, providing self-
resistance to the producing strain. Structures of this domain in complex with non-hydrolysable
adenylate analogues have been solved, and show that the main determinant for binding is the
adenine base[47]. Specifically, the purine is sandwiched between two aromatic side chains resulting
in strong but non-specific
synthesized compounds partially replicates the adenine stacking interactions bound to the respective
aaRS enzyme (Figs. 3 and 4), it is likely that they also can form the same stacking interaction
π-π stacking interactions. Since the base moiety of the presently
π-π
in MccE, or the homologous RimL and YhhY enzymes. Interestingly a carboxymethylated cytidine-
based microcin C analog can escape MccE catalysed acetylation, retaining its antibacterial activity
while the equivalent adenine and cytosine congeners were inactivated. This suggests that
approptiate modification of the heterocyclic group can facilitate in avoiding this natural resistance
mechansim[23]. Understanding how the different GNATs bind the synthesised molecules,
providing knowledge of both the role of the base and the amino acid in recognition would be
beneficial in the further development of aaSA analogues as antimicrobials.
Combined, comparative biochemical evaluation and structural studies of the synthesized pyrimidine
congeners highlight the surprisingly limited number of interactions mediated by the class I aaRSs
with the base moiety and can help explain the promiscuity of this aaRS class with other competitive
inhibitors. For example, the ester-containing aliphatic chain of mupirocin effectively substitutes for
the adenine base, despite having no chemical resemblance (Fig. 1b). Similarly, Cubist
Pharmaceuticals reported compounds where the base was replaced by a tetrazole or triazole group,
targeting ileRS[19] and leuRS[18], respectively. Moreover, recently, Charlton et al. reported
modified N-leucinyl benzenesulfonamides targeting leuRS[15]. This is only a small overview of all
available compounds, yet what they have in common is strong inhibitory activity in spite of a large
variety of base modifications. Unfortunately, except for the mupirocin-ileRS complex[31], limited
structural information is available for these chemical entities. Our obtained crystallographic data
provide insight, for the first time, into the structure of class I aaRSs bound to base modified
inhibitors. While these protein-ligand structures can be employed for further drug development, the
current work also highlights the importance of evaluating the naturally occurring resistance
19

ACCEPTED MANUSCRIPT
mechanisms. The majority of reported aaRS inhibitors employ an amino acid to provide selectivity,
this though is likely the Achilles' heel of such entities and as such all compounds should be
evaluated against the NATs of the target organism.
4. EXPERIMENTAL SECTION
4.1. Materials and methods
Reagents and solvents were purchased from commercial suppliers and used as provided unless
indicated otherwise. DMF and THF were of analytical grade and were stored over 4 Å molecular
sieves. All other solvents used for reactions were analytical grade and used as provided. Reactions
were carried out in oven-dried glassware under a nitrogen atmosphere with stirring at room
temperature unless indicated otherwise. 14C-radiolabeled amino acids and scintillation liquid were
1
purchased from Perkin Elmer. H and 13C NMR spectra of the compounds dissolved in CDCl3,
CD3OD, DMSO-d6 or D2O have recorded on a Bruker Ultra Shield Avance 300 MHz, 500 MHz or
when needed at 600 MHz spectrometers. The chemical shifts are expressed as
δ
values in parts per
1
13
1
million (ppm), using the residual solvent peaks (CDCl3: H 7.26 ppm; C, 77.16 ppm; DMSO: H,
1
1
2.50 ppm; 13C, 39.52 ppm; HOD: H, 4.79 ppm; CD3OD: H, 3.31 ppm; 13C, 49.00 ppm) as a
reference. Coupling constants are given in Hertz (Hz). The peak patterns are indicated by the
following abbreviations: bs = broad singlet, d = doublet, m = multiplet, q = quadruplet, s = singlet
and t = triplet. High-resolution mass spectra were recorded on a quadrupole time-of-flight mass
spectrometer (Q-Tof-2, Micromass, Manchester, UK) equipped with a standard ESI interface;
samples were infused in 2-propanol/H2O (1:1) at 3 µL.min-1. For TLC, precoated aluminium sheets
were used (Merck, Silica gel 60 F254). The spots were visualized by UV light at 254 nm. Column
chromatography was performed on ICN silica gel 60A 60–200 µm. Final products were purified
using a C-18 110 Å column connected to a Shimadzu SPD-20A HPLC and Shimadzu SPD-20A
detector. Eluent compositions are expressed as v/v. Recordings were performed at 254 nm and 214
nm. Analytical data are only provided for all new compounds.
4.2. Chemical synthesis of the intermediates and final compounds
4.2.1. General procedure for synthesis of 2',3',5'-tri-O–TBDMS nucleosides (2a & 2b)
Following the protocol as described the persilylated nucleosides were obtained and analytical data
for the uridine and cytidine analogues corresponded to published literature data[16, 48, 49].
4.2.1.1 2',3',5'-tri-O-tert-butyldimethylsilyl-N3-Methyluridine (2c).
Following a described protocol[25], the persilylated uridine (1 equivalent) was suspended in a
mixture of DMF: acetone (1:1), along with K2CO3 (4 equivalent). MeI (1.5 equivalent) was added
20

ACCEPTED MANUSCRIPT
dropwise to the suspension, and the mixture was stirred overnight. After completion of reaction,
solvents were evaporated, and the residue was partitioned between ethyl acetate and water. The
layers were separated, and the aqueous layer was extracted with ethyl acetate. The combined
organic layers were washed with brine, dried over sodium sulfate, filtered and evaporated to yield
an oily residue which was purified by silica gel chromatography and the compound was eluted with
5-10% EtOAc: Hexane in 81% yield. 1H NMR (300 MHz, CDCl3 ) 0.06-0.12 (6s, 18H, CH3-Si),
0.88-0.93 (3s, 27H, tBu-CH3 ), 3.32 (s, 3H, N-CH3), 3.75 (d, 1H, J = 11.9 Hz, H-5'a ), 3.98 (d, 1H, J
= 11.9 Hz H-5'b ), 4.02-4.09 (m, 3H, H-4', H-3', H-2'), 5.71 (d, 1H, J = 8.1 Hz, H-5), 5.88 (bs, 1H,
H-1'), 7.98 (d, 1H, J = 8.2 Hz, H-6), 8.66 (s, 1H, NH). 13C NMR (75 MHz, CDCl3)
δ -5.9 (CH3 -Si),
-5.7 (CH3 -Si), -5.2 (CH3 -Si), -5.1 (CH3 -Si), -4.9 (CH3 -Si), -4.6 (CH3-Si), 17.6 (tBu C(CH3 )3),
17.7 (tBu C(CH3)3), 18.2 (tBu C(CH3)3), 25.4 (tBu CH3), 25.5 (tBu CH3), 25.69 (tBu CH3), 27.2
(CH3-N), 61.4 (C-5'), 70.3 (C-3'), 75.9 (C-2'), 83.9 (C-4'), 89.2 (C-1'), 100.9 (C-5), 137.7 (C-6),
150.7 (C-2), 162.7 (C-4). HRMS [ESI] m/z: calcd. for C28H57N2O6Si3 ([M+H]+): 601.3519, found:
601.3529.
4.2.2. General procedure for selective 5'-desilylation (3a–c)
The procedure reported in reference 15 was followed to obtain the 5’-desilylated compounds. The
uridine and cytidine analogues data corresponded to published literature data[16].
4.2.2.1. 2',3',-di-O- tert-butyldimethylsilyl-N3-methyluridine (3c).
The compound was eluted through column chromatography using 20-30% EtOAc: Hexane in 73%
yield. 1H NMR (300 MHz, CDCl3 )
δ 0.03 (s, 3H, CH3 -Si), 0.06 (s, 3H, CH3 -Si), 0.08 (s, 6H, CH3
-Si), 0.88 (s, 9H, tBu CH3 ), 0.91 (s, 9H, tBu CH3 ), 3.71 (d, 1H, J = 12.5 Hz H-5'a ), 3.94 (d, 1H, J
= 12.5 Hz H-5'b ), 4.08 (bs, 1H, H-3'), 4.15-4.18 (m, 3H, H-4'), 4.53-4.57 (m,H-2'), 5.46 (d, 1H, J
= 8.2 Hz, H-1’), 5.29 (d, 1H, J = 8.1 Hz, H-5), 7.63 (d, 1H, J = 8.1 Hz, H-6), 8.78 (bs, 1H, NH).
13C NMR (75 MHz, CDCl3)
δ -5.3 (CH3 -Si), -5.1 (CH3 -Si), -4.0 (CH3 -Si), -4.8 (CH3 -Si), 17.6
(tBu C(CH3)3 ), 17.7 (tBu C(CH3)3), 25.4 (tBu CH3), 25.5 (tBu CH3), 27.3 (CH3-N), 61.4 (C-5’), 71.5
(C-30), 73.0 (C-20), 86.0 (C-40), 84.5 (C-10), 101.2 (C-5), 140.6 (C-6), 150.1 (C-2), 162.5 (C-4).
HRMS [ESI] m/z: calcd. for C22H43N2O6Si2 ([M+H]+) 487.2654, found: 487.2653.
4.2.3. General procedure for synthesis of the respective 5'-O-sulfamoyl nucleosides (4a–c)[16]
The reported procedure was followed to get sulfamoylated compounds; data for uridine and cytidine
analogues corresponded to published literature data[16].Care should be taken to use a fresh bottle of
chlorosulfonyl isocyanate to avoid formation of the carbamoylated side product.
21

ACCEPTED MANUSCRIPT
4.2.3.1. 2',3'-di-O-tert-butyldimethylsilyl-5'-O-sulfamoyl-N3-methyluridine (4c).
The compound was eluted through column chromatography using 30-70% EtOAc: Hexane in 68%
1
yield. H NMR (300 MHz, CDCl3)
δ 0.08-0.09 (4s, 12H, CH3-Si), 0.88-0.90 (s, 18H, tBu-CH3),
3.30 (s, 3H, -N-Me), 4.03-4.07 (m, 1H, H-5’a), 4.23-4.27 (m, 2H, H-5’b, H-4’), 4.30-4.35 (m, 1H,
H-2’), 4.47–4.52 (m, 1H, H-3’), 5.59 (s, 2H, -SONH2), 5.69 (d, 1H, J = 3.5 Hz, H-1’), 5.82 (d, 1H,
J = 8.1 Hz, H-5), 7.58 (d, 1H, J = 8.2 Hz, H-6). 13C NMR (75 MHz, CDCl3)
δ -5.3 (CH3-Si), 5.2
(CH3-Si), -4.9 (CH3-Si), -4.6 (CH3-Si), 17.7 (tBu C(CH3)3), 25.4 (tBu CH3), 25.4 (tBu CH3), 67.9 (C-
5’), 70.4 (C-3’), 74.4 (C-2’), 80.9 (C-4’), 91.3 (C-1’), 101.2 (C-5), 138.0 (C-6), 150.6 (C-2), 162.9
(C-4). HRMS [ESI] m/z: calcd. for C22H44N3O8SSi2 ([M+H]+) 566.2382, found: 566.2394.
4.2.4. General procedure for synthesis of 5'-O-(N-L-aminoacyl)-sulfamoyl-nucleosides [5-7(a–
c)]
The coupling of various activated amino acids with sulfamoylated nucleosides was carried out as
described[8, 16]. The uridine and cytidine analogues (5a and 5b) data corresponded to published
literature data. Final product was eluted through column chromatography initially using MeOH:
DCM to get the intermediate and finally through HPLC using acetonitre:water to get desired
compounds as white fluffy solid in 30-82% yields (combined yields from coupling to final
deprotection).
4.2.4.1. 5'-O-(N-L-isoleucyl)-sulfamoyl-N3-methyluridine (5c)
The final compound was obtained in 74% yield. 1HNMR (300 MHz, D2O)
Ile- -CH3), 1.02 (d, 3H, J = 7.0 Hz, Ile- -CH3), 1.24–1.30 (m, 1H, Ile-
1H, Ile- -CH2 Hb), 2.03–2.04 (m, 1H, Ile- -CH), 3.28 (s, 3H, N3-Me), 3.79 (d,1H, J = 4.1 Hz, Ile-
-CH), 4.30–4.49 (m, 5H, H-5'a, H-5'b, H-4', H-3', H-2' ), 5.94 (d, 1H, J = 3.5 Hz, H-1') 6.00 (d,
1H, J = 8.1 Hz, H-5), 7.80 (d, 1H, J = 8.1 Hz, H-6). 13C NMR (75 MHz, D2O)
10.6 (Ile- -CH3),
14.2 (Ile- -CH3), 23.8 (Ile- -CH2), 27.4 (N3-Me), 36.1 (Ile- -CH), 59.5 (Ile- -CH2), 68.2 (C-5'),
δ
0.91 (t, 3H, J = 7.4 Hz,
δ
γ
γ-CH2 Ha), 1.43–1.52 (m,
γ
β
α
δ
δ
γ
δ
δ
β
α
68.9 (C-3'), 73.3 (C-2'), 81.1 (C-4'), 89.7 (C-1'), 101.4 (C-5), 139.0 (C-6), 151.7 (C-2), 165.1 (C-4),
173.7 (C=O, Ile). HRMS [ESI] m/z: calcd. for C16H25N4O9S ([M-H]-) 449.1347 found, 449.1355.
4.2.4.2. 5'-O-(N-L-leucyl)-sulfamoyl-cytidine (6a).
The final compound was obtained in 58% yield. 1HNMR (300 MHz, D2O)
-CH3), 1.66-1.81 (m, 3H, Leu- -CH2 & 1H-Leu- -CH), 3.79-3.84 (m, 1H, Leu-
δ 0.93-0.98(m, 6H, Leu-
δ
β
γ
α
-CH ), 4.28–4.50
22

ACCEPTED MANUSCRIPT
(m, 5H, H-5'a, H-5'b, H-4' H-3', H-2'), 5.97 (d, 1H, J = 3.9 Hz, H-1'), 6.12 (d, 1H, J= 7.6 Hz, H-5),
7.85 (d, 1H, J = 7.6 Hz, H-6). 13C NMR (75 MHz, D2O)
23.8 (Leu- -CH), 40.0 (Leu- -CH2), 54.0 (Leu- -CH), 67.7 (C-5'), 69.0 (C-3'), 73.7(C-2'), 81.0 (C-
δ 20.7 (Leu-δ-CH3), 21.6 (Leu-γ-CH3),
γ
β
α
4'), 89.5 (C-5), 96.1 (C-1'), 141.0 (C-6), 157.3 (C-2), 165.9 (C-4), 176.3 (C=O, Leu). HRMS [ESI]
m/z: calcd. for C15H24N5O8S ([M-H]-) 434.1351: found, 434.1351.
4.2.4.3. 5'-O-(N-L-leucyl)-sulfamoyl-uridine (6b).
The final compound was obtained in 82% yield. 1HNMR (300 MHz, D2O)
-CH3), 1.67-1.81 (m, 3H, Leu- -CH2 & 1H-Leu- -CH), 3.84-3.89 (m, 1H, Leu-
(m, 5H, H-5'a, H-5'b, H-4' H-3', H-2'), 5.94-5.97 (m, 2H, H-1', H-5), 7.82 (d, 1H, J = 7.6 Hz, H-6).
13C NMR (75? MHz, D2O)
20.6 (Leu- -CH3), 21.6 (Leu- -CH3), 23.7 (Leu- -CH2), 39.7 (Leu-
CH), 53.7 (Leu- -CH), 68.3 (C-5'), 69.1 (C-3'), 73.2 (C-2'), 81.3 (C-4'), 88.7 (C-1'), 102.2(C-5),
δ
0.94-0.98(m, 6H, Leu-
-CH ), 4.32–4.44
δ
β
γ
α
δ
δ
δ
γ
β-
α
141.2 (C-6), 151.3 (C-2), 165.8 (C-4), 175.1 (C=O, Leu). HRMS [ESI] m/z: calcd. for C15H24N4O9S
([M]+) 435.1191: found, 435.1184.
4.2.4.4. 5'-O-(N-L-leucyl)-sulfamoyl-N3-methyluridine (6c)
The final compound was obtained in 65% yield. 1HNMR (300 MHz, D2O)
-CH3), 1.66-1.81 (m, 3H, Leu- -CH2 & 1H-Leu- -CH), 3.28 (s, 3H, N3-Me), 3.79-3.84 (m, 1H,
Leu- -CH ), 4.28–4.50 (m, 5H, H-5'a, H-5'b, H-4' H-3', H-2'), 5.97 (d, 1H, J = 3.9 Hz, H-1'), 6.12
(d, 1H, J= 7.6 Hz, H-5), 7.85 (d, 1H, J = 7.6 Hz, H-6). 13C NMR (75 MHz, D2O)
20.7 (Leu-
CH3), 21.6 (Leu- -CH3), 23.7 (Leu- -CH2), 27.4 (N3-Me), 39.9 (Leu- -CH2), 53.9 (Leu- -CH),
δ 0.93-0.98(m, 6H, Leu-
δ
β
γ
α
δ
δ-
δ
γ
δ
β
α
67.7 (C-5'), 69.0 (C-3'), 73.4(C-2'), 81.3 (C-4'), 89.6 (C-1'), 101.4 (C-5), 139.0 (C-6), 151.8 (C-2),
165.1 (C-4), 176.0 (C=O, Leu). HRMS [ESI] m/z: calcd. for C16H25N4O9S ([M- H]-) 449.1348:
found, 449.1352.
4.2.4.5. 5'-O-(N-L-tyrosyl)-sulfamoyl-cytidine (7a).
For synthesis of N-Boc-Tyr-OSu, DCC (1.3 equivalent) was added to a solution of Boc-Tyr-OH (1
equivalent) and N-hydroxysuccinimide (1.2 equivalent) in dry THF (3 mL/mmol) at 0 °C. The
mixture was stirred for 0.5 h at 0°C, then for 2–6 h at room temperature under an argon atmosphere.
The resulting dicyclohexylurea (DCU) was removed by filtration, and the filtrate was concentrated
in vacuo to give the corresponding activated ester. This activated ester was used as such for
1
coupling. The final compound was obtained in 30% yield. HNMR (300 MHz, D2O)
δ 2.89 (d, 2H,
J = 6.7 Hz, Tyr-
β
-CH2 ), 3.56 (t, 1H, Tyr-α-CH ), 4.18–4.26 (m, 5H, H-5'a, H-5'b, H-4' H-3', H-2'),
23

ACCEPTED MANUSCRIPT
5.94-5.99 (m, 2H, H-5 and H-1'), 6.76 (d, 2H, J = 8.6 Hz, Tyr-Ar-H) and 7.11 (d, 2H, J = 8.3 Hz,
Tyr-Ar-H), 7.74 (d, 1H, J = 7.6 Hz, H-6). 13C NMR (75 MHz, D2O) 39.0 (Tyr-
β-CH2), 57.8 (Tyr-α-
CH), 67.1 (C-5'), 68.9 (C-3'), 73.8 (C-2'), 81.0 (C-4'), 89.3 (C-1'), 96.1 (C-5), 115.4 (Tyr-ortho-C),
127.9 (Tyr-ipso-C), 130.4 (Tyr-meta-C), 140.8 (C-6), 157.3 (C-2), 155.2 (Tyr-para-C), 165.8 (C-4),
181.9 (C=O, Tyr). HRMS [ESI] m/z: calcd. for C19H22N5O9S ([M-H]-) 484.1145: found, 484.1144.
4.2.4.6. 5'-O-(N-L- tyrosyl)-sulfamoyl-uridine (7b).
1
The final compound was obtained in 62% yield. HNMR (300 MHz, D2O)
δ 3.16 (d, 2H, J = 6.5
Hz, Tyr-
5.83 (d, 1H, J = 8.1 Hz, H-5), 5.93 (d, 1H, J = 4.5 Hz, H-1'), 6.84 (d, 2H, J = 8.4 Hz, Tyr-Ar-H)
and 7.16 (d, 2H, J = 8.4 Hz, Tyr-Ar-H), 7.72 (d, 1H, J = 8.1 Hz, H-6). 13C NMR (75 MHz, D2O)
35.5 (Tyr- -CH2), 56.2 (Tyr- -CH), 67.9 (C-5'), 69.1 (C-3'), 73.3 (C-2'), 81.3 (C-4'), 88.7 (C-1'),
β-CH2 ), 4.02-4.06 (m, 1H, Tyr- α -CH ), 4.21–4.36 (m, 5H, H-5'a, H-5'b, H-4' H-3', H-2'),
δ
β
α
102.1 (C-5), 115.4 (Tyr-ortho-C), 125.6 (Tyr-ipso-C), 130.6 (Tyr-meta-C), 141.1 (C-6), 151.3 (C-
2), 154.7 (Tyr-para-C), 165.8 (C-4), 172.2 (C=O, Tyr). HRMS [ESI] m/z: calcd. for C18H22N4O10S
([M-H]-) 485.0984: found, 485.0979.
4.2.4.7. 5'-O-(N-L- tyrosyl)-sulfamoyl-N3-methyluridine (7c).
1
The final compound was obtained in 60% yield. HNMR (300 MHz, D2O)
δ
3.15 (d, 2H, J = 3.1
Hz, Tyr-
β
-CH2 ), 3.25 (s, 3H, N3-Me), 4.03 (m, 1H, Tyr-
α
-CH ), 4.18–4.35 (m, 5H, H-5'a, H-5'b,
H-4' H-3', H-2'), 5.86 (d, 1H, J = 8.1 Hz, H-5), 5.93 (d, 1H, J = 4.0 Hz, H-1'), 6.80 (d, 2H, J = 8.5
Hz, Tyr-Ar-H) and 7.13 (d, 2H, J = 8.5 Hz, Tyr-Ar-H), 7.70 (d, 1H, J = 8.1 Hz, H-6). 13C NMR (75
MHz, D2O)
δ 27.4 (N3-Me), 35.5 (Tyr-β-CH2), 56.2 (Tyr-α-CH), 67.7 (C-5'), 68.8 (C-3'), 73.40(C-
2'), 81.1 (C-4'), 89.8 (C-1'), 101.3 (C-5), 115.3 (Tyr-ortho-C), 125.5 (Tyr-ipso-C), 130.6 (Tyr-meta-
C), 138.9 (C-6), 151.6 (C-2), 154.7 (Tyr-para-C), 165.0 (C-4), 174.2 (C=O, Tyr). HRMS [ESI] m/z:
calcd. for C19H23N4O10S ([M-H]-) 499.1140: found, 499.1132.
4.3. Inhibition assays with E. coli aaRS and with S30 whole cell extracts
Cloning, expression and purification of E.coli aminoacyl-tRNA synthetases and purification of
tRNA was performed as reported[28] (see supplementary file).
4.3.1. In vitro inhibitory activity determination with purified E. coli aaRS
To examine the inhibitory effect of the various compounds we performed a radiolabel transfer assay
using purified E. coli aaRS. Briefly, either 10 nM ileRS, 0.5 nM tyrRS, 2.5 nM leuRS, in 20 mM
Tris, 100 mM KCl, 10 mM MgCl2 , 5 mM β-mercaptoethanol, pH 7.5 was preincubated with the
compound, at different concentrations, at 37 °C in the presence of 50 µM of the appropriate 14C-
24

ACCEPTED MANUSCRIPT
labeled amino acid, 2 mg/mL tRNA and 0.5 mg/mL inorganic pyrophosphatase. After 10 min, pre-
warmed ATP was added to the mixture at a final concentration of 500 µM. The reaction was
quenched by addition of 0.2 M sodium acetate pH 4, 0.1% N-lauroylsarcosine and 5 mM unlabeled
amino acid (2 mM for tyrosine). 20 mL was spotted on 3MM Whatman paper. After thorough
washing with cold 10% TCA, the filters were washed twice with acetone and air dried. Addition of
scintillation liquid was followed by measurement of the radio activity using scintillation counter.
The linear zone of enzyme activity was determined for each aaRS. The quench time was picked
within this zone at which approximately 50% of total RNA is aminoacylated. The quench time was
4, 6 and 8 minutes for leuRS, ileRS and tyrRS respectively. Copeland pointed out that when the
Kiapp of the compound approaches or is lower than the used enzyme concentration the Michaelis-
Menten equation is no longer valid[27].
Therefore, the apparent inhibition constant (Kiapp or Ki*) is determined using the Greco-Hakala
equation where vo is the rate of reaction in the absence of inhibitor, vi is the rate of reaction in the
presence of inhibitor, Eo is the concentration of enzyme and Io is the concentration of the inhibitor.
4.3.2. Preparation of normal S30 cell extract
10 µl E. coli NCIB / E. coli wt (strain K-12 BW28357) / E. coli BL21 glycerol stock was used to
inoculate 5 mL of LB-medium, and it was shaken overnight at 180 rpm at 37 °C. Next morning 100
µl of this preculture was used to inoculate 180 mL LB medium and the culture was grown until the
optical density (at 600 nm) reached 1.2 OD, usually in 4 hours. Cells were then transferred to four
50 mL falcon tube and centrifuged at 15,000 rpm for 15min at 4°C, and the supernatant was
decanted. The cell pellet was resuspended in 50 mL buffer containing 20 mM Tris·HCl, 10 mM
MgCl2, 100 mM KCl, pH 8.0. The cell suspension was centrifuged again at 15,000 rpm for 15min at
4°C. This procedure was repeated twice. The pellet was then resuspended in 1 mL of the same
buffer, and 5 µL of 1 M DTT was added and kept at 0 °C. Subsequently, the cells were sonicated
for 30 s (at 80% amplitude, 0.5 cycle) and cooled on ice bath for 10 seconds; and this process is
repeated 15 times. The lysate was centrifuged (17,000 rpm) at 4 °C for 15 min using a bench-top
centrifuge and supernatant was taken and again centrifuged (17,000 rpm) at 4 °C for 15 min. The
supernatant obtained was used as S30 cell extract.
4.3.3. S30 cell extract for aminoacylation assay
25

ACCEPTED MANUSCRIPT
For the tRNA aminoacylation, a 20 µl reaction was prepared. First, 1 µl of the inhibitor (at a stock
concentration of 5 µM) or buffer was added to 3 µl of the E. coli S30 cell extract (10 times diluted
with buffer) and incubated for 5 min. Next, 16 µl of the aminoacylation mixture which was kept at
37°C was added and which contains Tris.HCl (30 mM, pH 8.0), DTT (1 mM), E. coli MRE 600
tRNA (5 g/L purchased from Sigma), ATP (3 mM), KCl (30 mM), MgCl2 (8 mM), and the
corresponding 14C-radiolabeled amino acid (15 µM, 28.59 µM, and 9.66 µM for isoleucine, leucine
and tyrosine respectively). The aminoacylation reaction was carried out at 37 °C. The reaction was
quenched after 1 minute by addition of 0.2 M sodium acetate pH 4, 0.1% N-lauroylsarcosine and 5
mM unlabeled amino acid (2 mM for tyrosine). Then 10 µL of the reaction mixture was spotted on
3MM Whatman paper, and it was transferred to 10% cold TCA solution. The papers were washed
thoroughly with 10% cold TCA (twice), then the papers were washed twice with acetone and later
dried in air. Dried papers were transferred to scintillation vial followed by the addition of
scintillation liquid (12 mL), the amount of radionuclide incorporation was determined using a Tri-
card 2300 TR liquid scintillation counter. The linear zone of enzyme activity present in the S30
extract was determined for each aaRS (see Supplementary Fig. S8). The quench time point was
picked within this zone and with approximately 50% of total RNA aminoacylation.
4.3.4. Prolonged incubation of compound with S30 cell extract to check the stability of the
compounds
For the stability testing of compounds, we used mixture of inhibitor (at a stock concentration of 5
µM): S30 extract (1:3) or the mixture of inhibitor (at a stock concentration of 5 µM): acetyl-CoA (at
stock concentration of 10 mM): S30 extract (1:1:3) which is incubated at 37 °C for specific period
of time. At the specified time intervals 4 µl (or 5 µl in case of acetyl-CoA mixture) of this inhibitor-
cellular extract mixture was taken and added to the 16 µl (or 15 µl in case of acetyl-CoA mixture)
of the aminoacylation mixture which was kept at 37 °C which contains Tris.HCl (30 mM, pH 8.0),
DTT (1 mM), E. coli MRE 600 tRNA (5 g/L purchased from Sigma), ATP (3 mM), KCl (30 mM),
MgCl2 (8 mM), and the corresponding 14C-radiolabeled amino acid (10 µM, 28.59 µM, and 9.66
µM for isoleucine, leucine and tyrosine respectively). This reaction was quenched after 1-minute by
addition of 0.2 M sodium acetate pH 4, 0.1% N-lauroylsarcosine and 5 mM cold isoleucine. And
later steps were similar as described in section 4.3.3.
4.3.5. Preparation of S30 cell extracts comprising overexpressed NATs
The coding sequences for RimI, RimL and YhhY were amplified from E. coli BL21 (DE3) genomic
DNA and cloned into pETCH vector[50] between the Nco I and Hind III restriction sites. The
SUMO protein was already available in the pETSUK vector, and in-house vector derived from
26

ACCEPTED MANUSCRIPT
pETHSUL[50]. Plasmids were transformed into BL21 (DE3) pLysS E. coli cells and grown in 2 L
LB at 37°C until OD 0.6, at which 0.1 mM IPTG was added. The cultures overexpressing RimL and
SUMO were incubated for another 3 hours at 37°C before harvesting. To increase the solubility of
RimI and YhhY, the cultures were incubated for 24 hours at 18 °C. Cells were harvested and
resuspended into 50 mM phosphate pH 7.5, 100 mM potassium acetate and 10 mM magnesium
acetate and stored at -80 °C until further workup. Cells were thawed and sonicated three times with
a 30 mins incubation on ice between cycles. The lysate was clarified by centrifugation at 20000 g at
4 °C for 60 mins. The supernatant was carefully removed, aliquots were flash frozen in liquid
nitrogen and stored at -80°C until used.
4.3.6. NAT activity determination using NAT overexpressed S30 cell extract
To determine NAT activity, a mixture of inhibitor (at a stock concentration of 5 µM), acetyl-CoA
(at a stock concentration of 1.25 mM), and respective S30 extract (1:1:3) is incubated at 37 °C for
the specified period. The addition of inhibitor and acetyl-CoA mixture to cellular extract is done at
timepoint zero and after 5 and 60 mins, respectively, 5 µl of this mixture is added to 15 µl of the
aminoacylation mixture which was kept at 37 °C and which contains phosphate (50 mM, pH 7.5),
DTT (1 mM), E. coli MRE 600 tRNA (5 g/L purchased from Sigma), ATP (3 mM), magnesium
acetate (10mM), potassium acetate (100mM), and the corresponding 14C-radiolabeled amino acid
(15 µM, 28.59 µM, and 9.66 µM for isoleucine, leucine and tyrosine respectively). The
aminoacylation reaction is quenched after 1-minute by addition of 0.2 M sodium acetate pH 4, 0.1%
N-lauroylsarcosine and 5 mM unlabeled amino acid (2 mM for tyrosine). All following steps are as
described in section 4.3.3.
4.4. Crystallization
Prior to crystallization, E. coli tyrRS was incubated in the presence of 10 mM pyrophosphate on ice
for 1 hour to remove the contaminating reaction intermediate. This was followed by buffer
exchange into 10 mM BIS-TRIS pH 6.5, 500 mM KCl, 5% v/v ethylene glycol and 2 mM DTT
using a Superdex 200 10/300 GL column (GE Healthcare Lifesciences). E. coli tyrRS crystals were
grown using the hanging drop vapor diffusion technique at 20°C by mixing 2 µL protein solution
concentrated to 10 mg/ml, 1.5 µL of crystallization solution consisting out of 10-15% PEG 3350, 20
mM glutamate pH 6, 10 mM BIS-TRIS pH 6.5 and 20% v/v ethylene glycol and a 0.5 µL seed
stock. Crystals were transferred to a soaking solution containing 15% w/v PEG-3350, 20 mM
glutamate pH 6, 10 mM BIS-TRIS pH 6.5, 20% ethylene glycol and 2 mM compound and left for 5
hours before harvesting in a cryo-loop and flash freezing in liquid nitrogen.
27

ACCEPTED MANUSCRIPT
N. gonorrhoeae leuRS was buffer exchanged into 10 mM Tris pH 7, 100 mM NaCl, 2.5 mM β-
mercaptoethanol using a Superdex 200 10/300 GL column and concentrated to 10 mg/mL with a
centrifugal concentrator device. Final crystals were grown using the hanging-drop vapor-diffusion
method from a 0.75:1.0:0.25 (v/v) mixture of protein and reservoir buffer containing 20% w/v PEG
3350, 100 mM bis-tris propane pH 8.5 and 100 mM MgCl2 with the addition of crystal seeds from
earlier optimizations in similar conditions. Suitable crystals were soaked with the corresponding
ligand diluted in an equivalent precipitant solution supplemented with 22 % v/v ethylene glycol for
2 hours. These crystals were subsequently mounted in cryo-loops and flash-cooled in liquid
nitrogen prior to data collection.
Datasets were collected at synchrotron ESRF (Grenoble, France) or synchrotron Soleil (Paris,
France). All datasets were processed using XDS[51], pointless[52], aimless[53] and ccp4[54],
within the automated processing pipeline of autoproc[55]. E. coli tyrRS structures were solved
using a molecular replacement with Phaser[56] using the previously published E. coli tyrRS
structure (PDB: 4OUD) as a search model[57]. The model was optimized by iterative rounds of
manual model building using Coot[58] and refined using BUSTER[59]. In the case of N.
gonorrhoeae leuRS, initial phases were determined using molecular replacement employing a
model generated from E. coli leuRS (PDB: 3ZJU)[60]. The initial solution was optimized by
iterative rounds of manual model building in Coot[58] and refinement using Phenix[61]. All ligand
restraints were generated by GRADE Web server. Enzyme-ligand interactions were analyzed using
Schrödinger Maestro Software.
All compound-bound structures have been deposited to PDB under accession codes 6Q89, 6Q8A,
6Q8C, 6Q8B, 6I5Y, 6HB5, 6HB6, 6HB7 and full reports were obtained.
Funding
This work was supported by the Research Foundation Flanders [Fonds voor Wetenschappelijk
Onderzoek, grants G077814N to S.V.S. and A.V.A., G0A4616N to S.D.W. and A.V.A., 12A3916N
to B.G., G077814N to M.N. 1S53516N to S. D. G.], the KU Leuven Research Fund [grant
3M14022 to S.D.W. and A.V.A.]; and the Chinese Scholarship Council to L.P.
Conflicts of interest
No conflicts of interest are declared.
28

ACCEPTED MANUSCRIPT
Acknowledgements
The authors thank the Belgian FWO for providing a SB Fellowship of the Research Foundation -
Flanders to Steff De Graef, the China Scholarship Council for providing scholarships to Luping
Pang and KU Leuven for financial support. Mass spectrometry was made possible by the support of
the Hercules Foundation of the Flemish Government [20100225E7]. We further greatly appreciate
the support from the beamline scientists at the PROXIMA 1 and PROXIMA 2A (Soleil
Synchrotron, France) and ID23-1, ID30A3 and ID29 (ESRF, France) beamlines in facilitating the
collection of X-ray scattering data. We are also indebted to Luc Baudemprez for help with NMR
spectra measurement and Jef Rozenski for providing MS analysis.
5. REFERENCES
[1] G.H.M. Vondenhoff, A. Van Aerschot, Aminoacyl-tRNA synthetase inhibitors as potential
antibiotics, Eur. J. Med. Chem., 46 (2011) 5227-5236.
[2] G. Raczniak, M. Ibba, D. Soll, Genomics-based identification of targets in pathogenic bacteria
for potential therapeutic and diagnostic use, Toxicol, 160 (2001) 181-189.
[3] J.G. Hurdle, A.J. O'Neill, I. Chopra, Prospects for Aminoacyl-tRNA Synthetase Inhibitors as New
Antimicrobial Agents, Antimicrob. Agents Chemother., 49 (2005) 4821-4833.
[4] B. Gadakh, A. Van Aerschot, Aminoacyl-tRNA synthetase inhibitors as antimicrobial agents: a
patent review from 2006 till present, Expert. Opin. Ther. Pat., 22 (2012) 1453-1465.
[5] J.S. Tao, P. Schimmel, Inhibitors of aminoacyl-tRNA synthetases as novel anti-infectives, Expert
opinion on investigational drugs, 9 (2000) 1767-1775.
[6] R.V.K. Cochrane, A.K. Norquay, J.C. Vederas, Natural products and their derivatives as tRNA
synthetase inhibitors and antimicrobial agents, Medchemcomm, 7 (2016) 1535-1545.
[7] P. Fang, M. Guo, Evolutionary limitation and opportunities for developing tRNA synthetase
inhibitors with 5-binding-mode classification, Life, 5 (2015) 1703-1725.
[8] B. Gadakh, S. Smaers, J. Rozenski, M. Froeyen, A. Van Aerschot, 5 '-(N-aminoacyl)-sulfonamido-
5 '-deoxyadenosine: Attempts for a stable alternative for aminoacyl-sulfamoyl adenosines as aaRS
inhibitors, Eur. J. Med. Chem., 93 (2015) 227-236.
[9] M. Serpi, V. Ferrari, F. Pertusati, Nucleoside Derived Antibiotics to Fight Microbial Drug
Resistance: New Utilities for an Established Class of Drugs?, J. Med. Chem., 59 (2016) 10343-
10382.
[10] M.J.B. Brown, L.M. Mensah, M.L. Doyle, N.J.P. Broom, N. Osbourne, A.K. Forrest, C.M.
Richardson, P.J. O'Hanlon, A.J. Pope, Rational design of femtomolar inhibitors of isoleucyl tRNA
synthetase from a binding model for pseudomonic acid-A, Biochemistry, 39 (2000) 6003-6011.
[11] B.P. Duckworth, T.W. Geders, D. Tiwari, H.I. Boshoff, P.A. Sibbald, C.E. Barry, 3rd, D.
Schnappinger, B.C. Finzel, C.C. Aldrich, Bisubstrate adenylation inhibitors of biotin protein ligase
from Mycobacterium tuberculosis, Chem. Biol., 18 (2011) 1432-1441.
[12] M.R. Bockman, C.A. Engelhart, S. Dawadi, P. Larson, D. Tiwari, D.M. Ferguson, D.
Schnappinger, C.C. Aldrich, Avoiding Antibiotic Inactivation in Mycobacterium tuberculosis by
Rv3406 through Strategic Nucleoside Modification, Acs Infectious Diseases, 4 (2018) 1102-1113.
[13] T. Kazakov, K. Kuznedelov, E. Semenova, D. Mukhamedyarov, K.A. Datsenko, A. Metlitskaya,
G.H. Vondenhoff, A. Tikhonov, V. Agarwal, S. Nair, A. Van Aerschot, K. Severinov, The RimL
29