Boronic ester-linked macrocyclic lipopeptides as serine protease inhibitors targeting Escherichia coli type I signal peptidase

Article abstract of DOI:10.1016/j.ejmech.2018.08.086

Type I signal peptidase, with its vital role in bacterial viability, is a promising but underexploited antibacterial drug target. In the light of steadily increasing rates of antimicrobial resistance, we have developed novel macrocyclic lipopeptides, linking P2 and P1′ by a boronic ester warhead, capable of inhibiting Escherichia coli type I signal peptidase (EcLepB) and exhibiting good antibacterial activity. Structural modifications of the macrocyclic ring, the peptide sequence and the lipophilic tail led us to 14 novel macrocyclic boronic esters. It could be shown that macrocyclization is well tolerated in terms of EcLepB inhibition and antibacterial activity. Among the synthesized macrocycles, potent enzyme inhibitors in the low nanomolar range (e.g. compound 42f, EcLepB IC₅₀ = 29 nM) were identified also showing good antimicrobial activity (e.g. compound 42b, E. coli WT MIC = 16 μg/mL). The unique macrocyclic boronic esters described here were based on previously published linear lipopeptidic EcLepB inhibitors in an attempt to address cytotoxicity and hemolysis. We show herein that structural changes to the macrocyclic ring influence both the cytotoxicity and hemolytic activity suggesting that the P2 to P1’ linker provide means for optimizing off-target effects. However, for the present set of compounds we were not able to separate the antibacterial activity and cytotoxic effect.

Full text of DOI:10.1016/j.ejmech.2018.08.086

Accepted Manuscript
Boronic ester-linked macrocyclic lipopeptides as serine protease inhibitors targeting
Escherichia coli type I signal peptidase
Natalia Szałaj, Lu Lu, Andrea Benediktsdottir, Edouard Zamaratski, Sha Cao, Gustav
Olanders, Charles Hedgecock, Anders Karlén, Máté Erdélyi, Diarmaid Hughes,
Sherry L. Mowbray, Peter Brandt
PII:
S0223-5234(18)30758-X
10.1016/j.ejmech.2018.08.086
EJMECH 10697
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 9 July 2018
Revised Date: 22 August 2018
Accepted Date: 29 August 2018
Please cite this article as: N. Szałaj, L. Lu, A. Benediktsdottir, E. Zamaratski, S. Cao, G. Olanders,
C. Hedgecock, A. Karlén, Máé. Erdélyi, D. Hughes, S.L. Mowbray, P. Brandt, Boronic ester-linked
macrocyclic lipopeptides as serine protease inhibitors targeting Escherichia coli type I signal peptidase,
European Journal of Medicinal Chemistry (2018), doi: 10.1016/j.ejmech.2018.08.086.
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
Boronic ester-linked macrocyclic lipopeptides as serine
protease inhibitors targeting Escherichia coli type I signal
peptidase.
a
a
c
Natalia Szałaj a, †, Lu Lu b, †, Andrea Benediktsdottir , Edouard Zamaratski , Sha Cao , Gustav
Olanders a, Charles Hedgecock , Anders Karlén , Máté Erdélyi , Diarmaid Hughes , Sherry L.
a
a
d
c
Mowbray b, *, Peter Brandt a, *
a Department of Medicinal Chemistry, Uppsala University, Box 574, SE-751 23 Uppsala, Sweden.
b Department of Cell and Molecular Biology, Uppsala University, Box 596, SE-751 24 Uppsala,
Sweden.
c Department of Medical Biochemistry and Microbiology, Uppsala University, Box 582, SE-75 123
Uppsala, Sweden.
d Department of Chemistry – BMC, Uppsala University, SE-751 23 Uppsala, Sweden.
† N.S. and L.L. have contributed equally to this work and should both be considered as first authors.
* Corresponding authors. Tel.: +46 18 471 49 90. E-mail address: sherry.mowbray@icm.uu.se (S. L.
Mowbray). Tel.: +46 18 471 43 94. E-mail address: peter.brandt@ilk.uu.se (P. Brandt).
ABSTRACT
Type I signal peptidase, with its vital role in bacterial viability, is a promising but
underexploited antibacterial drug target. In the light of steadily increasing rates of
antimicrobial resistance, we have developed novel macrocyclic lipopeptides, linking P2 and
P1’ by a boronic ester warhead, capable of inhibiting Escherichia coli type I signal peptidase
(EcLepB) and exhibiting good antibacterial activity. Structural modifications of the
macrocyclic ring, the peptide sequence and the lipophilic tail led us to 14 novel macrocyclic
boronic esters. It could be shown that macrocyclization is well tolerated in terms of EcLepB
inhibition and antibacterial activity. Among the synthesized macrocycles, potent enzyme
inhibitors in the low nanomolar range (e.g. compound 42f, EcLepB IC50 = 29 nM) were
identified also showing good antimicrobial activity (e.g. compound 42b, E. coli WT MIC =
16 µg/mL). The unique macrocyclic boronic esters described here were based on previously
published linear lipopeptidic EcLepB inhibitors in an attempt to address cytotoxicity and
1

ACCEPTED MANUSCRIPT
hemolysis. We show herein that structural changes to the macrocyclic ring influence both the
cytotoxicity and hemolytic activity suggesting that the P2 to P1’ linker provide means for
optimizing off-target effects. However, for the present set of compounds we were not able to
separate the antibacterial activity and cytotoxic effect.
GRAPHICAL ABSTRACT
Keywords
P2–P1’ Boronic ester-linked macrocycles
Antibacterial lipopeptides
Bacterial type I signal peptidase
Escherichia coli type I signal peptidase (EcLepB)
Highlights
•
•
•
•
•
First-in-class P2–P1’ boronic ester-linked macrocyclic lipopeptides.
Compounds are potent inhibitors of bacterial type I signal peptidase (LepB).
Good antibacterial activity was shown for macrocycles.
The antibacterial effect and cytotoxicity could not be separated.
The macrocyclic boronic esters can be used for inhibition of serine proteases.
Abbreviations
2-CTC 2-chlorotritylchloride
AMR antimicrobial resistance
2

ACCEPTED MANUSCRIPT
CuAAC copper(I)-catalyzed azide-alkyne cycloaddition
DCM dichloromethane
DIEA N, N-diisopropylethylamine
DMF dimethylformamide
DMSO dimethyl sulfoxide
E. coli Escherichia coli
EcLepB Escherichia coli type I signal peptidase
FMCA fluorometric microculture cytotoxicity assay
Fmoc 9-fluorenylmethyloxycarbonyl
HBTU N, N, N′,N′-tetramethyl-O-(1H-benzotriazol-1-yl)uronium hexafluorophosphate
HFIP 1,1,1,3,3,3-hexafluoro-2-propanol
LiOH lithium hydroxide
MeOH methanol
NMO 4-methylmorpholine N-oxide
TBAF tetra-n-butylammonium fluoride
SPase I bacterial type I signal peptidase
TFA trifluoroacetic acid
1. Introduction
The high frequency of antimicrobial resistance (AMR) presents a global health emergency
according to a recent report by the World Health Organization [1]. Largely due to widespread
antibiotic misuse, resistance has made formerly treatable diseases deadly again [2]. Currently,
the number of deaths from bacterial infections is estimated at 2 million people a year, and
projected to quintuple by 2050 [3]. These facts show that antimicrobial resistance places a
considerable burden on society and healthcare systems, and highlight the importance of
developing new antibiotics to replace those that are no longer useful. The current WHO
3

ACCEPTED MANUSCRIPT
priority pathogens list has a strong focus on Gram-negative bacteria, including Escherichia
coli (E. coli), in its highest-priority (“critical”) category for which new antibiotics are needed
[4].
Bacterial type I signal peptidase (EC 3.4.21.89, SPase I), a membrane-bound serine
endopeptidase, recognizes and cleaves the N-terminal signal peptide found in the majority of
pre-proteins that are translocated across membranes via the Sec and Tat pathways. This
allows the secretion of a large number of mature proteins that are critical for the survival and
virulence of bacteria [5,6]. In contrast to classical Ser/His/Asp proteases, and to type I signal
peptidases from the eukaryotic endoplasmic reticulum that act through a Ser/His dyad,
bacterial type I signal peptidases act through a unique Ser/Lys catalytic dyad, which is
conserved in both Gram-positive and Gram-negative species [7]. Their role in bacterial
viability, together with the afore mentioned differences in the mode of action and the
relatively accessible position of its active site on the outer surface of the cytoplasmic
membrane, make this enzyme a promising but underexploited antibacterial drug target [8].
Only a few distinct groups of signal peptidase inhibitors have been reported, including the
macrocyclic arylomycin family (lipopeptides) [9–11], krisynomycin [12], 5S-penems (β-
lactams) [13–16], oligopeptides [17,18] and a beta-aminoketone [19,20]. Many opportunities
thus remain for antibacterial drug discovery.
To be effective, new classes of antibiotics must overcome the general problems of
entry into the cell, susceptibility to active efflux, and possible toxicity caused by the lack of
selectivity between bacterial and mammalian cells [21]. Periplasmic enzymes, such as SPase
I, have advantages when developing antibiotics that target Gram-negative bacteria, as such
drugs need only to pass the outer membrane of the double-membrane structure. Macrocycles,
which have been predominantly developed for treatment of infectious disease and for use in
oncology [22], have several features that can help overcome the problems listed above. Due to
their structural preorganization combined with some residual remaining flexibility,
macrocycles can interact both with shallow binding sites and dynamic protein targets. In
addition macrocyclization also has the potential to increase both binding affinity and
selectivity [23]. It has been shown that macrocyclization can be designed for improved
membrane permeability [24–26], metabolic stability [27,28] and pharmacokinetics [29]. In
addition, the number of cyclic peptide natural products with demonstrated oral bioavailability
suggests that mid-sized peptidic macrocycles may display favorable physicochemical
properties beyond the rule of five [30]. Thus, medicinal chemists have become increasingly
interested in macrocyclic compounds for expanding druggable space.
4

ACCEPTED MANUSCRIPT
In our own work, we decided to develop macrocyclic oligopeptides, aiming at
inhibition of the E. coli SPase I (EcLepB) and having antibacterial activity. Here we report the
design, synthesis and biological evaluation of first-in-class P2–P1’ boronic ester-linked
macrocyclic lipopeptides. We assessed their inhibition of EcLepB, as well as antibacterial
activity against a panel of Gram-negative strains Further, tests of cytotoxicity and hemolysis
.
were carried out as a preliminary investigation of safety.
2. Chemistry
2.1 Design
Recently, our group reported on linear lipopeptides [31] based on the decanoyl-
PTANA-aldehyde EcLepB inhibitor published by Buzder-Lantos et al. [18]. We showed that
replacing the aldehyde by a boronic ester improved the EcLepB IC50 50-fold, and that boronic
esters and acids were equipotent. However, modest antibacterial activity combined with
cytotoxicity limited their potential as leads for drugs. Since extensive optimizations of similar
lipopeptide EcLepB inhibitors, including arylomycin-like macrocycles, have been published
in recent patent literature [32,33], there was an interest to explore novel chemical space to
address the toxicity issues. Therefore, we searched for alternative sites for macrocyclization
allowing for chemistry different from previous inhibitors. As arylomycin is cyclized between
P2 and P4 and the side-chains of P1 and P3 are buried in hydrophobic pockets in the enzyme,
we focused on a novel alternative cyclization between P2 and P1’ (see Figure 1). The P1’
position corresponds to the diol attached to the boron in the warhead. Macrocyclic boronic
esters, usually containing more than one boron atom, have mostly been studied in
supramolecular chemistry [34]. To the best of our knowledge, this is the first report of boronic
ester-linked macrocycles being used as protease inhibitors.
5

ACCEPTED MANUSCRIPT
Figure 1. Design of the boronic ester macrocyclic oligopeptides (B) based on previously
reported linear EcLepB inhibitor (A).
We started by investigating the conformational preference of the peptidic backbone
upon P2–P1’ macrocyclization, to rule out the possibility that the cyclization would
destabilize the bioactive conformation of the peptide or the boronic ester. The initial
calculations were performed using a 3,4-dihydroxypyrrolidine boronic ester linked to a P2
cysteine, aspartic acid or glutamic acid, with amides, thioethers and disulfides as linkers.
However, as the original hit had an asparagine in P2, it made sense to keep this in the cyclized
analogue. To facilitate the synthesis, the amide was replaced by a triazole, an excellent
peptide-bond isostere that can easily be formed by the ‘Click’ reaction both in solution and on
a solid support [35]. Modelling different chain lengths between the P2 substituent and the
boron-capturing diol suggested that variants including 2–5 atoms all gave reasonable
structures. By means of density functional theory calculations, the macrocyclizations were
estimated to be energetically favored; see Supplementary Material for details. The final design
featured three unique boronic ester macrocyclic scaffolds; a cis-pyrrolidine-3,4-diol boronic
ester with either an ethylene or a propylene linker, and an all-cis cyclopentane-1,2,4-triol
boronic ester with an ethylene linker as shown in Figure 1. We also decided to slightly modify
the oligopeptide sequence at positions P4 and P5, and to investigate changes to the lipophilic
tails at the N-terminus.
As a positive charge at P5 is crucial for MIC activity in this series [31], we used commercial
ornithine instead of the 4-aminomethyl proline to simplify the synthesis without
compromising the potency (for examples, see 31g and 31h in Table 1).
2.1.1. Structure-based design
The three selected macrocyclic structures, truncated at the N-terminus, were further
investigated by means of conformational sampling of the covalent protein-macrocycle
complexes. It was concluded that all three macrocycles could fit the EcLepB binding site with
backbone conformations identical to arylomycin and a non-cyclized analogue, as shown in
6

ACCEPTED MANUSCRIPT
Figure 2. Overlays of arylomycin A2, the linear analogue, and a pyrrolidine-based macrocycle
in the predicted protein-bound conformations are shown in Figure 3. The different linker
lengths did not disturb the backbone conformation of the macrocyclic compounds compared
to the linear analogue (see Figures 2 and 3).
HO
OH
OH
O
O
O
O
HN
H
N
H
N
N
O
N
H
N
O
O
HO
O
O
Arylomycin A2, Pink
NH2
OH
O
H
N
O
O
H
H
N
N
B
O
N
H
N
H
H
O
O
O
Yellow
+H3N
NH+
N
N
N
O
OH
O
O
B
H
N
H
N
N
H
N
N
H
H
H
O
O
O
Ac-31a, Green
+H3N
Figure 2. Comparison of the binding conformation of arylomycin A2 (left pane, tail not
shown) as observed in the crystal structure (PDB entry: 3IIQ,[36] pink), the modelled N-
acetylated analogue of the macrocyclic compound 31a (green), and an N-acetylated linear
analogue of compound 31a with an ethylene glycol boronic ester (yellow).
7

ACCEPTED MANUSCRIPT
Figure 3. A) Two views of modeled binding conformations of the three designed macrocycles
covalently attached to S91 in EcLepB (PDB entry: 3IIQ) (the numbering system for the
protein has been adjusted to reflect the corrected sequence reported by Paetzel et al. [37]).
2.2 Chemistry
Tetrapeptides 3a-d were synthesized by manual solid-phase peptide synthesis (SPPS)
on 2-chlorotritylchloride (2-CTC) polymer resin using standard protocols for N-terminal 9-
fluorenylmethyloxycarbonyl-protected amino acids (Fmoc-AA-OH) solid-phase peptide
synthesis as described in the experimental chemistry section (Scheme 1).
O
O
O
R2 R1
N
O
H
b,c
a
FmocHN
FmocHN
N
OH
O
N
O
FmocHN
H
O
O
OtBu
1
2
3a-d
R1
H
R2
Cmpd
3a
NHBoc
NHBoc
3b
Me
H
H
N
NHPbf
NH
3c
N
NTrt
3d
H
8

ACCEPTED MANUSCRIPT
Scheme 1. Solid-phase synthesis of tetrapeptides 3a-d. Reagents and conditions: a) 2-CTC,
DIEA, DCM, r.t., 2 h, then MeOH, r.t., 0.5 h; b) 20% piperidine in DMF, twice, 5 min + 20
min, r.t.; c) Fmoc-AA-OH, HBTU, DIEA, DMF, r.t., 4 h or 2 h + 2 h double-coupling or
overnight.
Synthesis of heteroaromatic tails is depicted in Scheme 2. Compound 6 was prepared
via Suzuki coupling of methyl 6-bromo-nicotinate with (4-butylphenyl)boronic acid, followed
by hydrolysis of the ester. The N-aryl pyridone tail 9 was prepared from the corresponding 2-
pyridone derivative and (4-hexylphenyl)boronic acid using a process similar to that reported
earlier [38]. The octynylnicotinic acid tail 12 was synthesized using standard Sonogashira
coupling of ethyl 5-bromonicotinate with 1-octyne, followed by alkaline hydrolysis of the
ester. The benzimidazole tail 15 was synthesized by formylation of hexylbenzene using
hexamethylenetetramine in trifluoroacetic acid (TFA). In the second step, the resulting crude
aldehyde was condensed with 3,4-diaminobenzoic acid in the presence of sodium pyrosulfite
to afford the final compound. Compound 18 was made via iodination of methyl 1H-indole-5-
carboxylate with iodine in the presence of potassium carbonate, followed by Boc protection
under standard conditions. Suzuki coupling with (4-hexylphenyl)boronic acid and subsequent
lithium hydroxide (LiOH) hydrolysis afforded the required indole tail 20. The
biphenylpropiolic acid tail 23 was synthesized by coupling of ethyl propiolate to biphenyl
boronic acid in the presence of copper(I) iodide and silver(I) oxide as described earlier [39],
followed by ester hydrolysis.
9

ACCEPTED MANUSCRIPT
Scheme 2. Synthesis of the heteroaromatic tails. Reagents and conditions: a) (4-
butylphenyl)boronic, Pd(PPh3)4, Na2CO3, toluene/EtOH/H2O 3:1:2.5, 110 °C, microwave
irradiation, 40 min; b) NaOH, THF/H2O 5:1, reflux, 2 h; c) (4-hexylphenyl)boronic acid,
Cu(OAc)2, Py, mol. sieves, DCM, r.t., 20 h; d) LiOH, THF/MeOH/ H2O 3:2:1, r.t., overnight;
10

ACCEPTED MANUSCRIPT
e) 1-octyne, bis(triphenylphosphine) palladium(II) dichloride, copper(I) iodide and
diethylamine in DMF; 50 °C, 15 min; f) LiOH·H2O in dioxane/H2O, r.t., overnight; g)
hexamethylenetetramine, TFA, 75 °C, overnight; h) 3,4-diaminobenzoic acid, sodium
pyrosulfite, DMF, 100 °C, 16 h; i) I2, K2CO3, DMF, r.t., overnight; j) Boc2O, DMAP, THF,
r.t., 1 h; k) (4-hexylphenyl)boronic acid, K2CO3, toluene/MeOH 1:1, Pd(PPh3)4, 75 °C, 40
min; l) LiOH·H2O, THF/MeOH/H2O 3:2:1, r.t., overnight; m) ethyl propiolate, CuI, Ag2O,
CsCO3, DCE, 80 °C, 24 h; n) LiOH·H2O, THF/MeOH/H2O 3:2:1, r.t., 1 h.
Coupling of the tail at the N-terminus of the synthesized tetrapeptides 3a-d was
performed using 4’-hexyl-biphenyl-carboxylic acid or 6-(4-butylphenyl)nicotinic acid 6,
giving intermediates 24a-e (Scheme 3). The final oligopeptides 25a-e were cleaved from the
solid support using a solution of 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP)/DCM (1:4).
O
O
O
O
R2 R1
N
O
R2 R1
N
O
H
N
H
a
b
N
3a-d
R3
R3
N
N
N
O
N
OH
H
H
H
H
O
O
O
O
OtBu
OtBu
24a-e
25a-e
Cmpd
25a
R1
H
R2
R3
NHBoc
NHBoc
I
25b
Me
H
I
H
N
NHPbf
NH
R3
=
25c
I
N
NTrt
25d
H
I
N
C6H13
C4H9
NHBoc
25e
H
II
I
II
Scheme 3. Synthesis of oligopeptides 23a-e and 24a. Reagents and conditions: a) RCOOH,
HBTU, DIEA, DMF, r.t., 18 h; b) HFIP/DCM 1:4, r.t., 1 h.
To synthesize 28a and 28b required for the Click reaction with propargylic peptides 25a-e,
cis-pyrrolidine-3,4-diol was protected using a solution of p-toluenesulfonic acid in
benzaldehyde dimethyl acetal (Scheme 4) [40]. The resulting acetal was N-alkylated with 1-
bromo-3-chloro-propane or 1-bromo-2-chloro-ethane in the presence of potassium carbonate.
11

ACCEPTED MANUSCRIPT
HO
HO
O
Cl
O
O
a
b
HCl
NH
NH
N X
O
X
= CH2-CH2 or CH2
26
27
28a-b
Scheme 4. Synthesis of the chloroalkyl derivatives of cis-pyrrolidine-3,4-diol. Reagents and
conditions: a) benzaldehyde dimethyl acetal, p-toluenesulfonic acid, r.t., overnight; b) 1-
bromo-3-chloro-propane or 1-bromo-2-chloro-ethane, K2CO3, MeCN, 21 h.
The copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC) [41] between the
propargylic peptides 25a-e and azides, generated in situ by reaction of sodium azide with the
corresponding alkyl halides 28a and 28b, afforded the triazole derivatives 29a-f (Scheme 5).
Next, coupling of (R)-boroAla-(+)-pinanediol at the C-terminus, followed by complete
deprotection with TFA/DCM 1:1 mixture at elevated temperature resulted in
transesterification of pinanediol with cis-pyrrolidine-3,4-diol to give the final cyclic boronic
ester peptides 31a-e. Structure elucidation studies by NMR revealed partial epimerization
during the (R)-boroAla-(+)-pinanediol coupling step. Such a racemization occurring via a
5(4H)-oxazolone intermediate is a well-established side effect of carboxyl activation [42,43].
The resulting epimers could not be separated by any HPLC method used and so the
compounds were submitted for testing as 1:2 mixtures.
12

ACCEPTED MANUSCRIPT
Scheme 5. Synthesis of peptides 31a-f with the cyclic boronic ester warheads. Reagents and
conditions: a) NaN3, DMF, 60 °C, overnight; b) 25a-e in H2O/DMF, Na2CO3, copper(II)
sulfate pentahydrate, sodium ascorbate, 50 °C, 24 h and r.t., 24 h or 30 °C, 48 h; c) (R)-
boroAla-(+)-pinanediol hydrochloride, HATU, DIEA, DCM, 0 °C to r.t., 1 h; d) TFA/DCM
1:1, 50 °C, 5 h.
For the preparation of the required cyclopentane-1,2,4-triol moiety 37, commercially
available cyclopentenol 32 was protected using tert-butyldiphenylsilyl chloride and sodium
hydride (Scheme 6). The resulting olefin was converted to the corresponding diol as described
earlier [44] using catalytic amounts of OsO4 and NMO as re-oxidant via the Upjohn
dihydroxylation protocol [45]. In agreement with the previous report, the ratio between anti-
and syn-isomers was approximately 8:1 and the isomers could be separated by column
chromatography. The diol was converted into the ketal with 1,3-diphenyl-2-propanone and
catalytic amounts of tosylic acid. Subsequent deprotection with TBAF gave the free alcohol
34. Mitsunobu inversion [46] followed by ester hydrolysis with lithium hydroxide afforded
13

ACCEPTED MANUSCRIPT
the corresponding syn-alcohol 35. This was alkylated with tert-butyl bromoacetate using
sodium hydride as a base and then reduced with LiAlH4 in THF to give pentatriol moiety 36.
Tosylation under standard conditions followed by a nucleophilic substitution with NaN3 led to
the required azide 37.
HO
O
a, b
e, f
c, d
HO
O
Si
O
OH
tBu
OH
32
33
34
HO
O
O
HO
N3
O
O
O
i, j
g, h
O
O
O
35
36
37
Scheme 6. Synthesis of the protected syn-1,2,4-cyclopentanetriol moiety with the azidoethyl
linker. Reagents and conditions: a) NaH, dry THF, TBDPSCl, 16 h; b) OsO4, NMO,
tBuOH/THF/H2O 3:2:1, r.t., 2 h; c) diphenylacetone, TsOH, DCM, r.t., overnight; d) TBAF,
THF, r.t., 1 h; e) p-nitrobenzoic acid, THF, DEAD, r.t., 16 h; f) LiOH, THF/MeOH/water
3:2:1, r.t., 3 h; g) NaH, DMF, tert-butyl 2-bromoacetate, r.t., 14 h; h) LiAlH4,THF, r.t., 3 h; i)
TsCl, Py, DCM, r.t., 24 h; j) NaN3, MeCN, 80 °C, 48 h.
Coupling of the commercially available acids (biphenyl-carboxylic acid, 4’-hexyl-biphenyl-
carboxylic acid and 6-oxo-1-phenyl-1,6-dihydropyridine-3-carboxylic acid) or the synthesized
acids (compounds 9, 12, 15, 20 and 23) at the N-terminus of the synthesized tetrapeptide 3a
was followed by a solid-phase click reaction between the propargylic group and azide 37.
Cleavage from the solid support gave peptides 40a-h. The resulting free acids were coupled
with (R)-boroAla-(+)-pinanediol at the C-terminus; followed by simultaneous deprotection of
the side chains and the diol to give the final boronic esters 42a-h. The lipopeptide 43 used in
the NMR studies was prepared using the same protocols as for 42a-h. Compound 44 was
formed from 43 by air-oxidation similar to that described by Snyder et al.[47]
14

ACCEPTED MANUSCRIPT
Scheme 7. Synthesis of cyclic boronic ester peptides 42a-h. Reagents and conditions: a)
RCOOH, HBTU, DIPEA, DMF, r.t., overnight; or RC(O)OC(O)OtBu, DIPEA, DMF, r.t.,
overnight; b) azide 36, copper(II) sulfate pentahydrate, sodium ascorbate, THF/water 3:1, r.t.,
15

ACCEPTED MANUSCRIPT
overnight; c) HFIP in DCM (1:4), r.t., 2 h; d) (R)-boroAla-(+)-pinanediol hydrochloride,
HATU, DIPEA, DCM, 0 °C, 1h; e) TFA/DCM 1:1, 45 °C, 2h.
3. Results and discussion
3.1. In vitro biology
To test in vitro potency against EcLepB, we used a previously described FRET-based
assay [31]. The measured IC50 values (nM) for compounds 31a-f (esters with a cis-
pyrrolidine-3,4-diol moiety), 31g-h (linear reference compounds, see Supplementary data for
details) and 42a-h (esters with all-cis cyclopentane-1,2,4-triol moiety) are given in Tables 1
and 2, respectively.
In vitro antibacterial activity was determined against seven Gram-negative bacterial
strains, including three strains of Escherichia coli (WT, efflux-defective and drug-
hypersensitive), two strains of Pseudomonas aeruginosa (WT and efflux-defective),
Klebsiella pneumoniae and Acinetobacter baumannii. Staphylococcus aureus was used as a
Gram-positive reference. The results are given as MICs (µg/mL) in Tables 1 and 2.
All of the synthesized compounds were also evaluated on the HepG2 cell line for
cytotoxicity using a fluorometric microculture cytotoxicity assay (FMCA) [48] and for
hemolytic activity on human blood (see Tables 1 and 2).
3.1.1. Effects of macrocyclization on enzyme inhibition
Comparing the homologous macrocycles 31a, 31f, and 42b (EcLepB IC50s = 105, 104,
and 175 nM, respectively) with the linear reference analogue 31h having an asparagine in P2
and a boronic acid pinanediol ester (EcLepB IC50 = 17 nM), it can be seen that the
macrocyclization slightly decreases the potency of enzyme inhibition. If this is an effect of the
macrocyclization per se or a consequence of the difference in P1’ and/or P2 is difficult to say.
However, as was expected from modeling, the minor variations in the macrocycle in positions
facing solvent had no dramatic impact on enzyme inhibition. Thus, this part of the molecule
can be used for modulating bacterial penetration and ADMET properties.
3.1.2. Arginine and histidine in P5 and N-methylation in P4 to reduce cytotoxicity
As both 31a and 31f showed cytotoxicity against HepG2 cells (Table 1), we
considered changes in P5 to alleviate this issue. Sosic et al. [49] have reported that replacing
an aminoethyl group with an guanidinoethyl group could provide a reduction of non-specific
cytotoxicity. We therefore synthesized two variants in P5 carrying arginine (31c) and histidine
16

ACCEPTED MANUSCRIPT
(31d) instead of ornithine. As was expected from the modeling, since the P5 side chain is
solvent-exposed, both compounds were potent inhibitors of LepB with IC50 values similar to
those of 31a and 31f. Compound 31c with arginine in P5 was clearly best, in terms of
inhibition (EcLepB IC50 = 41 nM) and antibacterial activities, but unfortunately, it was worse
in terms of cytotoxicity and hemolysis. The less basic P5 histidine substitution (31d),
although active on the enzyme, was completely inactive in the MIC assay, and did not fare
well in the cytotoxicity and hemolysis assays.
N-methylation can modulate the properties of peptides, and so a P4 N-methylated
compound (31b) was also synthesized. Although this was tolerated from an enzyme inhibition
perspective, there was a significant loss in terms of MIC. In addition, the methylation did not
have any dramatic effect on either cytotoxicity or hemolysis.
3.1.3. Replacing cis-pyrrolidine-3,4-diol by all-cis cyclopentane-1,2,4-triol to reduce charge
A major challenge with compounds 31a-d and the smaller macrocyclic inhibitor 31f
was clearly their hemolytic properties. It is known that reducing the number of positive
charges on amphipathic compounds can reduce hemolysis [50]. Therefore, we designed a new
macrocyclic scaffold where the basic pyrrolidine was replaced by a neutral cyclopentyl group
linked to P2 by an ether linkage. Although the resulting macrocycle 42b showed a two-fold
reduction in EcLepB inhibition (IC50 = 175 nM) compared to the two other macrocyclic
variants (compounds 31a and 31f), it was twice as good in terms of its antibacterial activity.
Compared to the equal-sized pyrrolidine based compound 31a the level of hemolysis was
halved for the ether analogue 42b (10.6%, see Table 2) and the cytotoxicity was reduced.
3.1.4. Increasing polarity of the tails to reduce cytotoxicity
Inspired by the work on polymyxin analogues by Magee et al. [51], where more polar
tails resulted in reduced cytotoxicity, a number of different tails, including truncated versions,
were introduced (compounds 31e and 42c – 42g, see Tables 1 and 2). From the results, it can
be concluded that the aliphatic part of the lipophilic tail is not the major driver for enzyme
inhibition (Table 2). Comparing, for example, 42a (biphenyl, IC50 = 72 nM) with 42b
(hexylbiphenyl; IC50 = 175 nM), and 42c (1‐phenylpyridin‐2‐one; IC50 = 1660 nM) with 42d
(1‐(4‐hexylphenyl)pyridin‐2‐one; IC50 = 495 nM), it is clear that the hexyl chain only
marginally affects the enzyme inhibition. However, the length/lipophilicity of the tail is
crucial for the antibacterial activity, as shown by a lack of MIC activity for the truncated
analogues 42a and 42c.
17

ACCEPTED MANUSCRIPT
With regard to the heteroaromatic tails (see Tables 1 and 2), the lipophilicity of these
was in general significantly reduced, with the exception of 42f and 42g, which were
synthesized to incorporate hydrogen-bond donors. Two of the less hydrophobic tails, 6-(4-
butylphenyl)nicotinyl (31e) and the elongated 5-(oct-1-yn-1-yl)nicotine (42e) were well-
tolerated in terms of enzyme inhibition, whereas the bent pyridin‐2‐one based compounds
(42c and 42d) were less potent. Compound 42g, geometrically similar to the latter, was less
potent in terms of enzyme inhibition. In the case of 2-(4-hexylphenyl)-1H-benzo[d]imidazole
(42f), we observed an improvement in the EcLepB inhibition, with this compound being the
most potent inhibitor in the current study (IC50 = 29 nM).
Although all of these compounds were active in terms of enzyme inhibition, all
changes to the lipophilic tail resulted in increased MIC values. Only 42g, the indole
derivative, had any appreciable antibacterial properties, but still was appreciably worse than
42b. In terms of cytotoxicity and hemolytic properties, the less lipophilic compounds, 31e,
42a, 42c, 42d, and 42e were, however, superior to 42b. Clearly, the separation of antibacterial
properties and toxicity is a significant challenge for lipopeptides targeting EcLepB.
18

ACCEPTED MANUSCRIPT
Table 1
Inhibition potency against EcLepB, antibacterial activity, cytotoxicity and hemolysis for macrocyclic boronic esters 31a-h with a cis-pyrrolidine-3,4-diol moiety.
MIC (µg/mL)
EcLepB IC50
Cytotoxicity
IC50 (µM)j
Hemolysis
(%)
cmpd
structure
E. c.b
E. c.c
E. c.d
P. a.e
> 64
P. a.f
64
K. p.g
A. b.h
S. a.i
32
(nM)a
105
31a
32
8
8
32
16
4.0
8.5
17.4 ± 1.1
9.6 ± 1.8
(68-162)
98
31b
31c
64
16
16
32
> 64
> 64
64
32
32
16
64
16
32
16
(59-159)
41
(31-52)
4
8
2.4
21.0 ± 2.9
107
31d
31e
> 64
> 64
> 64
64
64
> 64
> 64
> 64
> 64
> 64
> 64
> 64
> 64
> 64
5.4
4.1 ± 0.2
7.3 ± 0.5
(84-136)
42
32
32
29.1
(34-51)
104
31f
64
8
16
1
32
> 64
> 64
64
2
32
4
32
9.5
10.1 ± 1.1
4.8 ± 0.4
(77-141)
52
31g
0.5
64
16
0.025
10.1
(38-71)
19

ACCEPTED MANUSCRIPT
17
31h
8
2
0.5
64
32
2
4
0.025
11.7
11.7 ± 0.8
(16-18)
a95% Confidence interval is shown in parentheses; bE. coli ATCC 25922 WT; cE. coli ∆tolC (CH3130, efflux-defective mutant isogenic to ATCC 25922); dE. coli D22 (CGSG 5163, lps mutant, drug-hypersensitive); eP. aeruginosa PAO1
WT; fP. aeruginosa efflux-defective PAO750 (isogenic to PAO1); gK. pneumoniae ATCC 13883 WT; hA. baumanii ATCC 19606 WT; iS. aureus ATCC 29213 (Gram-positive reference); jcytotoxicity of HepG2 cells.
Table 2
Inhibition potency against EcLepB, antibacterial activity, cytotoxicity and hemolysis for macrocyclic boronic esters 42a-h with a cyclopentane-1,2,4-triol moiety.
MIC (µg/mL)
EcLepB IC50
(nM)a
Cytotoxicity
IC50 (µM)i
cmpd
R
E. c.a
> 64
16
E. c.b
> 64
4
E. c.c
> 64
4
P. a.d
> 64
64
P. a.e
> 64
32
K. p.f
> 64
8
A. b.g
> 64
8
S. a.h
> 64
2
Hemolysis (%)
0.2 ± 0.3
72
42a
> 64
8.0
(56-93)
175
42b
42c
10.6 ± 1.4
(144-213)
1660
> 64
> 64
> 64
> 64
> 64
> 64
> 64
> 64
> 64
0.04 ± 0.01
(1060-2590)
495
(359-683)
41
42d
42e
42f
> 64
> 64
> 64
32
> 64
64
> 64
> 64
64
> 64
> 64
> 64
64
> 64
> 64
64
> 64
> 64
64
> 64
> 64
64
64
32
4
> 64
> 64
24.4
23.7
0.03 ± 0.15
0.01 ± 0.01
0.5 ± 0.2
(33-52)
29
16
(25-34)
225
42g
32
16
64
64
64
16
2.8 ± 0.6
(195-261)
20

ACCEPTED MANUSCRIPT
227
42h
> 64
> 64
> 64
> 64
> 64
> 64
> 64
32
28.2
0.01 ± 0.02
(168-305)
a95% Confidence interval is shown in parentheses; bE. coli ATCC 25922 WT; cE. coli ∆tolC (CH3130, efflux-defective mutant isogenic to ATCC 25922); dE. coli D22 (CGSC 5163, lps mutant, drug-hypersensitive); eP. aeruginosa PAO1
WT; fP. aeruginosa efflux-defective PAO750 (isogenic to PAO1); gK. pneumoniae ATCC 13883 WT; hA. baumanii ATCC 19606 WT; iS. aureus ATCC 29213 (Gram-positive reference); jcytotoxicity of HepG2 cells.
21

ACCEPTED MANUSCRIPT
3.2. Confirmation of macrocyclization
Macrocyclization of compounds 42a-h was corroborated by mass spectrometric and
NMR analyses of compound 43 (Figure 4). This polyalanine analogue has a simplified
structure that enabled unambiguous and quick spectral analyses.
1H NMR assignment was performed using a NOESY-TOCSY backbone walk, and
further corroborated by the cross peaks observed in gCOSY, 1H,13C- and 1H,15N-gHSQC, and
1H,13C- and 1H,15N-gHMBC spectra. The spectra were obtained at several temperatures within
the 25 °C-80 °C range to support interpretation of overlapping signals (see supplementary
data for details). The observed molecular mass [M+H]+ 729.3558 (calcd 729.3532 for
C36H45BN8O8), along with the key NOE correlations shown in Figure 4, confirmed ring
closure. The latter NOEs were in agreement with the computed cyclic geometry of 43 (Table
S1), and none corresponded to its hydrolyzed, ring-opened analogue (Figure 4). The 1H NMR
signals of 43 were doubled, and did not coalesce at high temperatures, which revealed it to be
a 2:1 mixture of epimers due to partial racemization of the P2 amino acid during the final
coupling of the boronic amino acid. It is well known that activation of the C-terminal
carboxylic acid of the peptide chain is associated with oxazolone formation that leads to the
formation of peptide epimers [42].
The largest chemical shift differences of the epimers were observed for the nuclei of
the macrocycle, suggesting that the point of epimerization is H-1’ (Figure 4). The amide
temperature coefficients of all amide protons of 43 were in the range of 4.0-4.6 ppb/K,
suggesting them to be in equilibrium between intramolecularly hydrogen-bonded and solvent-
exposed states, apart from those of Ala-1 and P2, whose values of 9.2 ppb/K and 8.1 ppb/K
revealed them to be solvent exposed [52,53]. The latter are most likely unavailable for
intramolecular hydrogen bonding due to their participation in the macrocycle. Upon
hydrolysis to 44, the amide temperature coefficient of P2 decreased to 3.8 ppb/K, and hence
this group became available for intramolecular hydrogen bonding during the opening of the
macrocycle and the consequent major structural change.
Furthermore, the computed cyclization free energies given in Table S1 predict that the
ring-closed forms of the macrocycles are stable towards hydrolysis in DCM, DMSO and
water.
22

ACCEPTED MANUSCRIPT
Figure 4. The cyclic peptide 43 (IC50 18 nM; MIC > 64 µg/mL on all strains) and the linear
peptide 44 used for NMR confirmation of macrocyclization, with the key NOE information
being shown as red arrows.
4. Summary and conclusions
This project was initiated in response to the observed cytotoxic and hemolytic
liabilities among previously studied linear lipopeptide boronic acids and esters[31]. We have
designed and synthesized first-in-class P2–P1’ boronic ester-linked macrocycles. This extends
the chemical space available for designing EcLepB inhibitors, allowing more options for
tuning the properties of the boronic ester warhead, and potentially leading to potent EcLepB
inhibitors with improved antibacterial activity and in vitro toxicity profiles.
Ultimately, 14 novel macrocyclic lipopeptides differing in the macrocyclic linker, the
peptide sequence and in the lipophilic tail were synthesized to investigate the structural
requirements for enzyme inhibition, antibacterial activity and toxicity. MS and NMR analyses
provided the macrocycle to exist in the ring-closed form. Good enzyme inhibition and MIC
data showed that the macrocyclization was indeed tolerated by the enzyme. Macrocyclization
via an ether linkage, as in 42b, resulted in a compound active against all strains in the
antibacterial panel, including wild types.
23

ACCEPTED MANUSCRIPT
Cytotoxicity and hemolysis remained highly correlated with antibacterial activity. For
the macrocyclic EcLepB inhibitors having MIC activities, cytotoxicity and hemolysis
remained at the levels observed for the linear analogues. Thus, striking the right balance
between the total charge of the peptide, the composition and length of the hydrophobic tail,
the overall hydrophobicity and amphipathicity resulting in more effective and safe EcLepB
inhibitors remains an important challenge when targeting this enzyme.
To conclude, we have demonstrated that the P2 to P1’ linker can provide a means for
future optimization of the properties of EcLepB inhibitors. We also envision that P2–P1’
boronic ester-linked macrocycles can be of general use against other proteases.
5. Experimental section
5.1. Chemistry
5.1.1. General methods
Analytical thin layer chromatography (TLC) was performed using Merck aluminum sheets
precoated with silica gel 60 F254. Column chromatography was performed on Merck silica gel
60 (40-63 µm). The microwave reactions were performed in a Biotage Initiator producing
controlled irradiation at 2450 MHz with a power of 0–300 W. 1H and 13C NMR spectra were
recorded on Varian Mercury Plus instruments; 1H at 399.9 MHz and 13C at 100.6 MHz at 25
°C. Exact molecular masses were determined on Micromass Q-Tof2 mass spectrometer
equipped with an electrospray ion source. Analytical RP-HPLC-MS was performed on a
Gilson RP-HPLC system with a Finnigan AQA quadrupole low-resolution mass spectrometer
in positive or negative ESI mode using a Onyx Monolithic C18 4.6 × 50 mm 5
µm
(Phenomenex) with MeCN in 0.05% aqueous HCOOH as mobile phase at a flow rate of 4
mL/min. Preparative RP-HPLC was performed on a system equipped with a Nucleodur C18
HTec 5 µm column (150 × 21.2 mm) or a Phenomenex C8 5 µm column (150 × 21.2 mm),
using a H2O/CH3CN gradient with 0.1% TFA or H2O/CH3CN gradient with 20 mM TEAA, in
both cases using UV detection at 220 nm and 254 nm. The purity of the peptides was
determined by RP-HPLC using the columns: ACE 5 C18 (50 × 4.6 mm) and Nucleodur C18
HTec 5 µm (4.6 × 50 mm), H2O/CH3CN gradient with 0.1% TFA or H2O/CH3CN gradient
with 20 mM TEAA and UV detection at 220 nm and 254 nm. All peptides showed purity
above 95%.
24

ACCEPTED MANUSCRIPT
5.1.2. General method for solid phase synthesis of oligopeptides
Tetrapeptides 3a-d were synthesized by manual solid-phase peptide synthesis (SPPS) using 2-
chlorotritylchloride polymer resin (2-CTC, 1.63 mmol/g) in a 10-mL disposable syringe fitted
with a porous polyethylene filter and N-terminal fluorenylmethyloxycarbonyl-protected
amino acids. Coupling of the first amino acid: Fmoc-propargyl-Gly-OH (0.8 eq.) to the 2-
CTC resin (1.0 eq.) was performed in anhydrous DCM (7.5 mL) in the presence of DIEA (3.0
eq.) with shaking at room temperature for 2 h. Then, MeOH (2 mL) was added to cap the
unreacted resin and the mixture was shaken for 30 min. The resin was filtered, washed with
several portions of DCM, DMF and again DCM and dried under high vacuum, overnight.
Loading of the starting material was calculated via increase in dry weight. Next, the resin
swelling was performed in DMF for 20 min, Fmoc protecting group was removed by
treatment with 20% piperidine in DMF (2 × 3 mL, 5 + 20 min) and the polymer was washed
with DMF (5 × 3 mL, 5 × 2 min). Then, coupling of the appropriate amino acid: Fmoc-AA-
OH (4.0 eq.) was performed in DMF (3 mL) using HBTU (4.0 eq.), in the presence of DIEA
(8.0 eq.) with shaking at room temperature. In the case of Fmoc-Ala-OH for 4 h, with Fmoc-
Orn(Boc)-OH, Fmoc-Thr(tBu)-OH and Fmoc-N-Me-Thr(tBu)-OH overnight and with Fmoc-
Arg(Pbf)-OH and Fmoc-His(Trt)-OH the double coupling was performed (after 2 h the resin
was washed with DMF and treated with a freshly made solution of Fmoc-AA-OH, HBTU and
DIEA in DMF and shaken for 2h). After each coupling, the resin was washed with DMF (4 ×
4 mL) and subsequently de-protected and washed as described above. At the end of the
coupling cycle the resin was washed with several portions of DMF, MeOH and DCM and
dried under high vacuum, overnight. Method (A) for coupling of the tailDMF via HBTU
activation. The coupling of the tails at the N-terminus of the synthesized tetrapeptides was
performed using the acid (4 eq.) in DMF (2 mL), in the presence of DIEA (8 eq.) and HBTU
(4 eq.) at room temperature, shaking overnight. The resin was then washed extensively with
several portions of DMF, MeOH and DCM and dried under high vacuum. Method (B) for
coupling of the tail via mixed anhydride. Step 1. Heterocyclic acid derivative (0.520 mmol, 1
eq.) was reacted with Boc anhydride (2.110 mmol, 4 eq.) in dry tetrahydrofuran in presence of
pyridine (2.640 mmol, 5 eq.) at r.t. overnight. After full consumption of starting material
volatiles were removed in vacuo. Aqueous 0.5M hydrochloric acid solution was added and the
product was extracted with ethyl acetate. The organic layer was washed with water, separated
and dried over sodium sulfate. Volatiles were removed in vacuo yielding a crude product as
yellowish oil, which was used in the next step without further purification. Step 2. The 2-CTC
resin containing 3a (0.130 mmol, 1 eq.) was suspended in DMF (3 mL) containing mixed
25

ACCEPTED MANUSCRIPT
anhydride prepared in Step 1 (presumed 0.52 mmol, 4 eq.) and DIPEA (1.820 mmol, 14 eq.).
The reaction mixture was shaken at r.t. overnight. The resin was filtered, washed with DMF
(5 x 2 mL), DCM (5 x 2 mL) and THF (5 x 4 mL). General method for cleavage from the
solid support. 2-CTC resin containing oligopeptides was added into a mixture of HFIP/DCM
1:4 (4 mL) and the reaction mixture was shaken at room temperature for 1.5 h. The resin was
filtered off and washed with DCM and THF. The volatiles were evaporated in vacuo.
Purification. The crude peptides were dissolved in THF (2.5 mL) and H2O/MeCN (1:1, 2.5
mL), filtered through a 0.45 µm nylon membrane filter and purified in 1-3 runs by preparative
RP-HPLC. The reported yields are based on calculated loading of the oligopeptide on 2-CTC
resin.
5.1.3. General method for the copper(I)-catalyzed azide-alkyne cycloaddition in solution
To a solution of the alkyl chloride 28a or 28b (1.3 eq. or 1.5 eq.) in DMF, sodium azide (2.6
eq. or 3.0 eq.) was added. The reaction mixture was stirred at 60 °C overnight. Then, peptide
25a-e (1.0 eq.) in H2O/DMF 1:3, Na2CO3 (3 eq.), sodium ascorbate (0.8 eq.) and
CuSO4·5H2O (0.4 eq.) were added and stirred at 50 °C or 60 °C for 24 h and at r.t. for 24 h or
at 30 °C for 48 h. The reaction mixture was filtered through a 0.45 µM nylon membrane,
diluted with H2O/MeCN (1:1) and purified by preparative RP-HPLC.
5.1.4. General method for the copper(I)-catalyzed azide-alkyne cycloaddition on the solid
phase
2-CTC resin containing bound oligopeptides 38a-h (0.130 mmol, 1 eq.) was mixed with THF
(1.8 mL) containing azide 37 (0.140 mmol, 1.1 eq.). In a separate vial sodium ascorbate
(0.140 mmol, 1.1 eq.) was dissolved in water (0.900 mL). To this solution copper sulfate was
added (0.060 mmol, 2.3 eq.), which quickly generated brown solution. The resulting solution
was immediately taken into the reactor containing oligopeptide and azide. The reaction
mixture was shaken at r.t. overnight. Then, the resin was filtered, washed with DMF (3 x 5
mL), H2O (3 x 5 mL), THF (3 x 5 mL) and DCM (3 x 5 mL).
5.1.5. General method for the coupling of oligopeptides with R-(+)-boro-Ala-pinanediol and
for side chain deprotection and cyclization
To a solution of the peptide (1 eq.) in anhydrous DCM (10 mL) HATU (2 eq.) and DIEA (3
eq.) were added under nitrogen atmosphere at 0 °C and the reaction mixture was stirred at 0
°C for 6 min, then R-(+)-boro-Ala-pinanediol (2 eq.) was added and the reaction mixture was
26

ACCEPTED MANUSCRIPT
stirred at room temperature for 1h. To the DCM solution of crude peptide generated in R-(+)-
boro-Ala-pinanediol coupling step (1 eq.) TFA was added to a final ratio of TFA/DCM 1:1.
The resulting mixture was stirred at 50 °C for 5 h. The solvents were removed under reduced
pressure and the residue was purified by preparative RP-HPLC.
5.2. Computational details
All calculations were performed within the Schrödinger Small-Molecule Drug Discovery
Suite 2017-1 using the OPLS3 force field.[54] The crystal structure of LepB from E. coli was
downloaded from the Protein Data Bank [55] (PDB entry: 3IIQ) and thereafter prepared using
the Protein Preparation Wizard [56] implemented in Maestro [57] using default settings. This
included addition of hydrogens, hydrogen-bond network optimization, removal of water
molecules forming less than three hydrogen bonds to non-waters and a restrained
minimization. To evaluate if macrocycles cyclized between the boronic acid and P2 could fit
into the LepB binding pocket and interact with the protein in a similar way as arylomycin, the
macrocyclic compounds were built into the active site manually using the protein-bound
conformation of arylomycin as a template. This was done by adding an (R)-boro-alanine to
arylomycin which was attached covalently to S91 in the protein, exchanging the side-chains
of arylomycin to an Orn-TANA sequence, deletion of the remaining amino acids, truncating
the lipophilic tail to an acetyl group and converting the boronic acid to an ethylene glycol
boronic ester. The resulting complex was refined using 1 000 steps of MCMM conformational
sampling in a GB/SA continuum solvation model for water. The boronate ester and protein
side chains within 5 Å from the boronic-peptide were sampled. Protein backbone and amino
acids more than 5 Å away from the boronate ester were kept frozen. The resulting binding
mode was consistent with the previously published modeling of a similar linear peptide
boronic acid [31]. Using the refined structure, seven macrocycles with linker lengths varying
between 2-5 and 2-4 carbon atoms between amine-tetrazole and ether-tetrazole were built,
respectively. To evaluate if the peptides still could adopt the arylomycin backbone
conformation after macrocyclization, the protein-macrocycle complexes were refined using
5 000 steps MCMM conformational sampling and a GB/SA continuum solvation model for
water. Side chains within 5 Å of the macrocycles were free to move but not sampled. The
protein backbone and side chains more than 5 Å away from the macrocyclic compound were
kept frozen.
27

ACCEPTED MANUSCRIPT
5.3. EcLepB inhibition assay
In vitro potency against EcLepB was determined using a FRET-based assay as described
previously. Briefly, the peptide substrate included Dabcyl as the fluorescence acceptor, and
EDANS
as
the
donor,
attached
to
the
termini
(i.e.
Dabcyl-
VGGTATA
↓GAFSRPGLE(EDANS)-OH). After pre-incubation of EcLepB (estimated 50
nM) with synthesized compounds for 10 min, the reaction was initiated with substrate (final
concentration 8 M; Km 20 M), and the increasing fluorescence intensity was monitored
µ
µ
continuously at 535 nm at 22 °C for 2 h. Cleavage rates were plotted as a function of inhibitor
concentration, and non-linear regression analysis of the sigmoidal dose-response curves was
employed to calculate half-maximal inhibitory concentration (IC50) in CDD
(https://www.collaborativedrug.com).
5.4. Minimal inhibitory concentration (MIC) assay
Compounds were evaluated against a panel of Gram-negative bacteria, including: E. coli
ATCC 25922, wild-type; CH3130, an efflux-defective
∆tolC mutant isogenic to ATCC
25922; D22, a drug-hypersensitive lpxC mutant of E. coli; P. aeruginosa PAO1, wild-type;
PAO750, an efflux-defective mutant isogenic to PAO1; K. pneumoniae ATCC 13833, wild-
type; A. baumannii ATCC 19606, wild-type; and the Gram-positive S. aureus ATCC 29213,
wild-type. In vitro antimicrobial activity of each compound was determined by measuring the
MIC value using the broth micro-dilution technique in cation-adjusted Mueller-Hinton II
medium (Becton-Dickinson, 212322) according to EUCAST and CLSI guidelines. Compound
prepared in MHII medium was dispensed into a 96-well round-bottomed microtiter plate to
give final assay concentrations from 64 µg/mL down to 0.25 µg/mL (two-fold dilution series
in 10 wells, plus two control wells: medium control with no bacteria or compound, and
growth control with bacteria added but no compound). Bacteria prepared from fresh colonies
(grown on non-selective agar, incubation 18-24 h at 35 °C ± 2 °C) were suspended in saline to
0.5 McFarland (
10 mL of MHII broth to give a final bacterial concentration
range 3 L of bacterial suspension was pipetted into each well (except
7 × 105 CFU/mL). 50
the medium control well, where 50 L MHII was pipetted). Final volume in each well was
100 L. Plates were covered and incubated without shaking for 16-20 h at 35 °C ± 2 °C. MIC
≅
1.5 × 108 CFU/mL). 50
µL of this bacterial suspension was transferred to
≅
5 × 105 CFU/mL (acceptable
‒
µ
µ
µ
28

ACCEPTED MANUSCRIPT
was read visually, as complete inhibition of growth by the unaided eye, using the medium-
only wells as the control.
5.5. Cytotoxicity assay
Cytotoxicity was assessed in HepG2 cells using fluorometric microculture cytotoxicity assay
(FMCA) [48]. The cell line HepG2 was obtained from ATCC, and cultured in Eagle's
Minimum Essential Medium (ATCC) supplemented with 10% fetal bovine serum,
penicillin/streptomycin (100 U/100 µg/mL) and L-glutamine 2 mM (all from Sigma St Louis,
MO, USA). HepG2 cells were passaged 2 times/week and used maximally for 20 passages.
Cells were finally seeded in Nunc 384-well assay plates at a density of 1000 cells/well. The
FMCA is based on measurement of fluorescence generated from hydrolysis of fluorescein
diacetate (FDA) to fluorescein by cells with intact plasma membranes. HepG2 cells
(1000/well) were seeded in 384-well plates using the pipetting robot Biomek 4000 (Beckman
Coulter, Fullerton CA, USA) and cultured overnight before inhibitors were added by acoustic
dispensing (Echo 550 from Labcyte Inc, CA, USA) directly from 384-well source plates
(Labcyte) containing 1 or 10 mM compound stock solutions in DMSO. Each compound was
dispensed for dose-response, with eight two-fold dilutions in duplicate from highest
concentrations of 64 or 32 µM, to generate IC50s. Bortezomib (2 µM) was used as positive
control, and dose response testing of doxorubicin (highest concentration 2 µM) was
performed repeatedly to follow assay performance over time. After 72 h incubation with
drugs in a culture chamber (at 37 °C, humidity 95%, 5% CO2) the 384-well plates were
centrifuged (200 × g, 60 sec), the culture medium was removed by a Biotech ELX washer
(Biotek, Winooski, VT, USA) and 70 µL phosphate buffered saline (PBS, Sigma) was added
to the wells. This procedure was repeated once. Subsequently, the plates were centrifuged
(200 × g, 30 sec) and the PBS was removed by the ELX washer after which 50 µL/well of
assay buffer (HEPES-buffered saline with 0.5 mM MgCl2 and 0.5 mM CaCl2, pH 7.4) was
added to the plates using a Multidrop 384 (Thermo Fischer Scientific, NY, USA). Finally, 1
µL of fluorescein diacetate (0.5 mg/mL in DMSO) was added to each well using the pipetting
robot Biomek NX (Beckman Coulter, Fullerton CA, USA). The plates were then incubated
for 50 min in a Cytomat (37 °C, humidity 95%, 5% CO2, Thermo Fischer Scientific) before
being analyzed in a Fluostar Omega microplate reader (BMG Technologies, Germany) with
wavelengths set at 485 nm (excitation) and 530 nm (emission). Cytotoxicity was assessed
after 72 h with cell survival presented as survival index (SI, %) defined as fluorescence in test
wells in percent of control cultures with blank values subtracted. Criteria for a successful
29