Degradation of MAC13243 and studies of the interaction of resulting thiourea compounds with the lipoprotein targeting chaperone LolA

Article abstract of DOI:10.1016/j.bmcl.2013.02.005

The discovery of novel small molecules that function as antibacterial agents or cellular probes of biology is hindered by our limited understanding of bacterial physiology and our ability to assign mechanism of action. We previously employed a chemical genomic strategy to identify a novel small molecule, MAC13243, as a likely inhibitor of the bacterial lipoprotein targeting chaperone, LolA. Here, we report on the degradation of MAC13243 into the active species, S-(4-chlorobenzyl)isothiourea. Analogs of this compound (e.g., A22) have previously been characterized as inhibitors of the bacterial actin-like protein, MreB. Herein, we demonstrate that the antibacterial activity of MAC13243 and the thiourea compounds are similar; these activities are suppressed or sensitized in response to increases or decreases of LolA copy number, respectively. We provide STD NMR data which confirms a physical interaction between LolA and the thiourea degradation product of MAC13243, with a K _d of ～150 μM. Taken together, we conclude that the thiourea series of compounds share a similar cellular mechanism that includes interaction with LolA in addition to the well-characterized target MreB.

Full text of DOI:10.1016/j.bmcl.2013.02.005

Bioorganic & Medicinal Chemistry Letters 23 (2013) 2426–2431
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Degradation of MAC13243 and studies of the interaction of resulting
thiourea compounds with the lipoprotein targeting chaperone LolA
Courtney A. Barker a, Sarah E. Allison a, Soumaya Zlitni a, Nick Duc Nguyen b, Rahul Das b,
Giuseppe Melacini b, Alfredo A. Capretta b, Eric D. Brown a,
⇑
a Department of Biochemistry and Biomedical Sciences, McMaster University, 1280 Main Street West, Hamilton, ON, Canada L8S 2K1
b Department of Chemistry and Chemical Biology, McMaster University, 1280 Main Street West, Hamilton, ON, Canada L8S 4M1
a r t i c l e i n f o
a b s t r a c t
Article history:
The discovery of novel small molecules that function as antibacterial agents or cellular probes of biology
is hindered by our limited understanding of bacterial physiology and our ability to assign mechanism of
action. We previously employed a chemical genomic strategy to identify a novel small molecule,
MAC13243, as a likely inhibitor of the bacterial lipoprotein targeting chaperone, LolA. Here, we report
on the degradation of MAC13243 into the active species, S-(4-chlorobenzyl)isothiourea. Analogs of this
compound (e.g., A22) have previously been characterized as inhibitors of the bacterial actin-like protein,
MreB. Herein, we demonstrate that the antibacterial activity of MAC13243 and the thiourea compounds
are similar; these activities are suppressed or sensitized in response to increases or decreases of LolA copy
number, respectively. We provide STD NMR data which confirms a physical interaction between LolA and
Received 8 December 2012
Revised 29 January 2013
Accepted 1 February 2013
Available online 13 February 2013
Keywords:
Chemical biology
Structure–activity relationship
Lipoprotein trafficking
LolA
the thiourea degradation product of MAC13243, with a Kd of ꢀ150
lM. Taken together, we conclude that
the thiourea series of compounds share a similar cellular mechanism that includes interaction with LolA
MreB
Chemical–genetic interactions
in addition to the well-characterized target MreB.
Ó 2013 Elsevier Ltd. All rights reserved.
Small molecules have a proven track record not only as antibiot-
ics but also as probes of bacterial physiology.1 Thus the discovery of
new antibacterial molecules is a promising approach to the creation
of new tools to understand bacterial systems. While connecting
cellular phenotype of a small molecule and its molecular target(s)
remains a significant challenge, chemical genomic approaches are
finding broad application in defining mechanisms of action of new
chemical probes.2–6 Our group previously employed such a chemical
genomic strategy to identify cellular targets of novel small mole-
cules discovered in screens for growth inhibition of the model
microbe Escherichia coli. Of interest, the inhibitory action of a novel
compound, MAC13243, was suppressed when the lipoprotein
chaperone, lolA, was expressed at high copy.4 LolA, part of the five-
membered Lol (localization of lipoproteins) system, is responsible
for the sorting and transport of lipoproteins to the outer membrane
in the majority of Gram-negative bacteria.7 Further biochemical and
physiological experiments confirmed that MAC13243 interacted
with LolA to inhibit the localization of lipoproteins to the outer
membrane.
to 3,4-dimethoxyphenethylamine (2) and S-(4-chlorobenzyl)isot-
hiourea (3a) (Fig. 1). The latter we suggest is the chemical entity
responsible for the antibacterial activity associated with 1. Com-
pound 3a is a structural analog of A22 (S-(3,4-dichlorobenzyl)isot-
hiourea), a small molecule previously reported to inhibit the
bacterial actin-like protein MreB. A22 was originally identified in
a screen for inhibitors of chromosome partitioning based upon
its ability to induce a morphological change from rod-shaped to
spherical.8 Mutations conferring resistance to A22 isolated in both
Caulobacter crescentus and Escherichia coli mapped to the ATP-bind-
ing pocket of MreB.9,10 Further biochemical characterization sug-
gested that A22 competitively inhibits the binding of ATP to
MreB, leading to the disassembly of actin filaments.11 With the dis-
covery that 1 could break down to an active species that was a
close analogue of A22, we have further explored the chemical–ge-
netic interactions of lolA with A22 and related thiourea-containing
compounds.
In order to investigate the in vitro stability of 1, samples
(250 lM) were incubated under acidic, neutral or basic conditions
Since the original identification of MAC13243 as a probe of lipo-
protein targeting, we have discovered that it is unstable in aqueous
solution and its triazine ring is slowly hydrolyzed. Herein, we re-
port on the breakdown of MAC13243 (compound 1 in this work)
for up to 48 hours at room temperature and analyzed using reverse
phase HPLC. At time zero, compound 1 elutes as a single peak at
the 8 minute mark (Fig. 2A, DMSO). However, after incubation in
weakly acidic aqueous solution, the appearance of two additional
peaks is observed (Fig. 2A, pH 4). Quantification of these peaks over
time revealed that 1 is indeed unstable in aqueous solution and
that its half-life is pH dependent (Fig. 2B). Under acidic conditions
⇑
Corresponding author. Tel.: +1 905 5259140x22454; fax: +1 905 522 9033.
E-mail address: ebrown@mcmaster.ca (E.D. Brown).
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.02.005

C. A. Barker et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2426–2431
2427
Figure 1. Breakdown scheme of MAC13243. MAC13243 (1) is hydrolyzed at the triazine ring generating one molecule of 3,4-dimethoxyphenethylamine (2), one molecule of
S-(4-chlorobenzyl)isothiourea (3a) and two molecules of formaldehyde. Also shown is the structure of A22 a close analog of 3a.
1 was shown to have a half-life of ꢀ4 hours, whereas under neutral
or basic conditions the compound has a half-life of 13 and 59
hours, respectively. The stability of compound 1 in the organic sol-
vent DMSO, used routinely to dissolve and store frozen working
stocks of the compound, was also tested. In this solvent, 1 was
found to be relatively stable, with a half-life of 257 hours at room
temperature. These findings support (acid-catalyzed) hydrolysis of
the central triazine ring of compound 1. As such, hydrolysis of one
molecule of 1 would generate one molecule of 3,4-dimethoxy-
phenethylamine (2), one molecule of S-(4-chlorobenzyl)isothiou-
rea (3a), and two molecules of formaldehyde.
Subsequent analysis of the biological activity of each degrada-
tion product revealed that, of the breakdown products 2 and 3a,
the latter compound was the active species (Fig. 3). Based on these
results, we hypothesized that degradation into the compound 3a
could be responsible for the biological activity associated with
compound 1. Previous structure–activity relationship (SAR) data
for 1 supported this hypothesis, as major structural changes on
the left-hand side of the molecule, corresponding to the inactive
degradation product 2, did not significantly affect biological
activity.4 The same was not true for the right-hand side of the
molecule, where major structural modifications were not
Figure 2. Analysis and kinetics of compound degradation. To investigate the
stability of compound 1, samples of the compound at a final concentration of
250 lM were incubated in DMSO, H20, 50 mM acetate buffer (pH 4) or 50 mM
citrate buffer (pH 11) and examined using reverse phase HPLC. Chromatography
was on a Symmetry C18 (4.6 ꢁ 150 mm; 3.5
lM) column from Waters (Mississauga,
Ontario, Canada). Samples were eluted from the column with a 10 min linear
gradient from 95% buffer A (0.1% trifluoroacetic acid (TFA) in H2O) to 97% buffer B
(0.1% TFA in acetonitrile). Analytes were visualized at 270 nm. (A) HPLC trace for
MAC13243 in DMSO at time zero (solid line), overlaid with the trace of MAC13243
incubated at pH 4 for 4 h (dotted line). Peak identities were determined through
mass spectrometry. LC-ESI-MS data were obtained by using an Agilent 1100 Series
LC system (Agilent Technologies Canada, Inc.) and a QTRAP LC/MS/MS System
(Applied Biosystems/MDS Sciex). (B) Degradation profiles of 1 under each condi-
tion: (d) DMSO, (4) H2O, (s) 50 mM acetate buffer (pH 4) and (N) 50 mM citrate
buffer (pH 11). Peak areas corresponding to 1 were plotted against time and fit to an
exponential decay curve.
Figure 3. Compound 3a is the active breakdown component. The activity of
compound 1 (d) and the major breakdown products, 2 (s) and 3a (.), were tested
against E. coli MG1655. Cells were grown to an OD600 of 0.4 and subsequently
diluted 1:100 000 into fresh Luria-Bertani (LB) containing serially diluted com-
pound. After 16 h of incubation at 37 °C the absorbance at 600 nm was recorded.

2428
C. A. Barker et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2426–2431
Figure 4. LolA at high copy suppresses the action of the thiourea compounds 3a
and A22. E. coli strain AG1 harbouring plasmids pCA24N and pCA24N-LolA were
from the ASKA collection32 Overnight cultures grown directly from frozen stocks in
Figure 5. LolA depletion sensitizes E. coli to compound
1 and the thiourea
LB containing chloramphenicol (CM) were diluted 1:100 into 200
lL of fresh media
compounds 3a and A22. The growth inhibitory effects compounds 1, 3a, A22
MAC13243, and a variety of antibiotics of diverse mechanism and chemical class
were examined in a strain of E. coli (EB2120) where the expression of gene lolA is
under the control of an arabinose-induced promoter. Primers lolA-F and lolA-R were
used with Vent polymerase to PCR-amplify lolA from E. coli MG1655 chromosomal
DNA. The PCR product, which placed a consensus ribosome binding site upstream of
lolA, was cloned into the PmeI site of pBS-araBADKan. The resulting plasmid, pBS-
araBADlolAKan, was cut with KpnI and SacI and the 2.6 kb fragment containing lolA
was transformed into MG1655-pKD46. To screen for integration of lolA at the
araBAD locus, chromosomal DNA was isolated from KanR colonies and used as a
template in a PCR reaction with primers araC-F and lolA-R. A strain positive for lolA
and subsequently grown to an OD600 of 0.3. Cells were induced with 0.1 mM
Isopropyl b-
D-1-thiogalactopyranoside (IPTG) for 2 h. Cultures were further diluted
1:500 and 2
lL of the diluted culture was spotted in duplicate on LB CM agar plates
with or without 0.1 mM IPTG and containing 0, 2ꢁ, 4ꢁ, 8ꢁ and 16ꢁ the MIC of
respective compounds. Plates were incubated at 37 °C for 24 h.
tolerated. Furthermore, 3a was found to be fourfold more potent
than MAC13243 (Fig. 3). The interpretation of this finding was
compounded by the realization that 3a is a structural analog of
S-(3,4-dichlorobenzyl)isothiourea, known as A22 in the literature,
which has been reported to inhibit the bacterial cytoskeleton pro-
tein, MreB.9–11 To shed light on this paradox, we set out to explore
the existence of a possible chemical genetic interaction between
the thiourea series of compounds and lolA.
integration was termed EB2119. To create
a conditionally complemented lolA
deletion strain, EB2120, a double crossover PCR strategy was used to produce a
linear fragment for transformation of EB2119. Primers lolA-A and lolA-B, lolA-C and
lolA-D, and CM-F and CM-R were used with Vent polymerase to amplify MG1655
chromosomal DNA or p34S-Cm DNA in the latter case. The PCR products were
isolated and used as templates in a final reaction with primers lolA-A and lolA-D.
The resulting product contained a chloramphenicol resistance cassette with its
transcriptional promoter, but not its terminator, flanked by 500 bp homology to the
regions upstream and downstream of lolA. The 1.4 kb product was transformed into
EB2119, and replacement of native lolA was confirmed by PCR using primers lolA-E
and CM-F and lolA-F and CM-R. Strain EB2120 was grown O/N at 37 °C in the
presence of 0.2% arabinose and subcultured to a final OD600 of 0.001 into LB media.
We originally identified a chemical genetic interaction between
the lipoprotein chaperone, LolA, and compound 1 using an ordered
high-copy expression array.4 The growth inhibitory action of the
small molecule was suppressed by the presence of the protein tar-
get at high copy. Using an inducible expression system, we exam-
ined whether LolA at high copy could also suppress the inhibitory
actions of compound 3a and A22. Cultures overexpressing lolA
(pCA24N-lolA) or containing empty expression vector (pCA24N)
were spotted on LB agar plates containing increasing concentra-
tions of 1, 3a or A22. Overexpression of lolA increased the MIC
16-fold for each of the three compounds (Fig. 4). The ability of LolA
at high-copy to suppress the action of compound 3a and A22 sug-
gests that these breakdown products share a similar cellular mech-
anism to the parent compound, 1. Further, these data provide the
first evidence of a chemical genetic link between A22 and lolA.
Next we examined the effect of reduction of LolA copy number
in cells. Reducing expression of a cellular target typically enhances
the growth inhibition phenotype caused by a small molecule.5,12
Using a conditional lolA strain, where the chromosomal copy of lolA
was deleted and a complementing copy placed under arabinose
control at the araBAD locus (see Table 2 for a list of the plasmids
and primers used to make the conditional lolA strain), MIC values
were determined for a panel of antibiotics including 1, 3a and
A22. The MIC analysis was performed at multiple arabinose con-
centrations (0–0.2%), allowing for different cellular levels of LolA
to be examined. Upon depletion of LolA (0.00002% and 0% arabi-
nose) a 32-fold sensitization to compounds 1, 3a and A22 was ob-
served (Fig. 5). In this experiment we also looked for interactions
with known antibiotics of varying mechanism and chemical class
and none of these demonstrated an enhancement close to this
magnitude, apart from the cell wall-active compound fosfomycin
which registered as much as a 16-fold sensitization. Several outer
Cells were grown to an OD600 of ꢀ0.4–0.5 and 5
added to microwell plates containing serially diluted antibiotics ranging in
concentration from 1 to 256 g/mL in a final volume of 200 L. MIC values were
lL of culture was subsequently
l
l
determined in the presence of (0.2%, 0.02%, 0.00002% and 0%) arabinose. Plates were
incubated at 37 °C for 18 h. Fold sensitization was calculated based on MIC values
determined under the LolA depletion condition in comparison to the fully
complemented 0.2% arabinose condition. Depletion of LolA results in a 32-fold
sensitization to MAC13243, TU-1 and A22.
membrane lipoproteins have been found to impact the structural
integrity of the cell.13–15 As such, mislocalization of lipoproteins,
due to the absence of LolA, could well compromise the cell wall,
resulting in the observed sensitization. Together these experiments
provide further evidence of a profound chemical–genetic interac-
tion between lolA and the thiourea containing compounds.
Using the conditional LolA depletion strain, we were also inter-
ested in investigating the effects of reduced lolA expression on cel-
lular morphology. It has been well established that treatment with
A22 results in a morphological transformation from rod-shaped
cells to spherical8 and as such, we wished to examine whether LolA
depletion could elicit a similar phenotypic response. Phase contrast
light microscopy images were taken from samples expressing
(Fig. 6B) or depleted of LolA (Fig. 6C) and samples treated with
A22 (Fig. 6E) or 1 (Fig. 6F). Indeed, depletion of LolA resulted in
the change of the cell shape from rod to sphere, the same pheno-
typic transition shared by treatment with 1 or A22. Interestingly,
this phenotype is also seen on depletion of MreB depletion.16
Table 1 provides a detailed SAR study of compound 3a, where
the antibacterial activity of 22 structural analogs were studied

C. A. Barker et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2426–2431
2429
Figure 6. LolA depletion and compound treatment leads to the formation of spherical cells. E.coli strains EB68 (wild type MG1655) and EB2120 (conditionally expresses gene
lolA in response to arabinose; see legend to Fig. 5) were grown overnight at 37 °C in LB or LB media containing 0.2% arabinose. The cells were subcultured 1:100 into 5 mL LB
media containing 0.1%
arabinose or 0.1% glucose. EB68 was resuspended in LB + A22 (10
D
-fucose and grown until an OD600 ꢀ0.5 was reached. The cells were washed once with fresh media. EB2120 was resuspended into LB containing 0.2%
l
g/mL) or LB + compound 1 (10 g/mL). Cells were then applied to poly-lysine treated microscope slides and
l
examined by light microscopy at 1 h intervals. Images were captured by a Q-colour 3 camera (Olympus, Mississauga, ON, Canada).
Table 1
Structure–activity relationships of S-(4-chlorobenzyl)isothiourea derivativesa
Cl
Cl
Cl
HN
S
R
NH2
R
R
Compound Series 3
Compound Series 4
Compound Series 5
Compd
R
Organism
MIC (lg/mL)
Compd
R
Organism
MIC (lg/mL)
Cl
E. coli
B. subtilis
8
256
E. coli
B. subtilis
>256
>256
1
n/a
3k
P. aeruginosa
32
P. aeruginosa
32
E. coli
B. subtilis
P. aeruginosa
>256
nd
nd
E. coli
B. subtilis
P. aeruginosa
32
>256
64
2
n/a
3l
Cl
Cl
E. coli
B. subtilis
P. aeruginosa
2
>256
16
E. coli
B. subtilis
P. aeruginosa
>256
>256
128
3a
3m
Cl
Br
Cl
Cl
Cl
F
E. coli
B. subtilis
P. aeruginosa
64
>256
>256
E. coli
B. subtilis
P. aeruginosa
>256
>256
128
3b
3n
I
E. coli
16
E. coli
256
B. subtilis
P. aeruginosa
>256
64
B. subtilis
P. aeruginosa
>256
128
3c
4a
4b
4c
O
E. coli
B. subtilis
P. aeruginosa
2
>256
16
E. coli
B. subtilis
P. aeruginosa
>256
>256
128
3d (A22)
NO2
E. coli
B. subtilis
P. aeruginosa
1
>256
32
E. coli
B. subtilis
P. aeruginosa
32
>256
32
3e
3f
Cl
Cl
H
E. coli
B. subtilis
P. aeruginosa
4
>256
16
E. coli
B. subtilis
P. aeruginosa
>256
4
64
HN
HN
N
4d
4e
NH2
O
HN
E. coli
16
E. coli
>256
NH2
NH2
3g
B. subtilis
P. aeruginosa
>256
16
B. subtilis
P. aeruginosa
4
64
HN
S
NH2
NH
HN
E. coli
B. subtilis
P. aeruginosa
16
>256
32
E. coli
B. subtilis
P. aeruginosa
>256
>256
128
3h
3i
4f
NH2
H
N
E. coli
B. subtilis
P. aeruginosa
64
>256
32
E. coli
B. subtilis
P. aeruginosa
256
>256
256
S
N
S
5a
H2N
a
Minimum inhibitory concentrations (MIC) were determined in liquid LB media. Overnight cultures were diluted 1:100 into fresh media and grown to an OD600 of 0.4 and
then diluted 1:100,000. Diluted cells (100 L) were added to 96 well microplates containing serially diluted compound to a concentration of 1–256 g/mL in a final volume of
200 L and incubated at 37 °C for 16 h without shaking, except for B. subtilis which required aeration at 250 rpm. The lowest concentration at which the OD600 was below 0.05
was deemed MIC. Bacterial strains were E. coli MG1655, B. subtilis 168 and P. aeruginosa PAO1 (nd, not determined).
l
l
l

2430
C. A. Barker et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2426–2431
Table 2
Strains, plasmids and oligonucleotides used in this study
Strain
Description
Ref.
28
EB68
EB2119
EB2120
MG1655 (Fꢂ kꢂ)
MG1655 (Fꢂ kꢂ) araBAD::lolA(optimal RBS)kan
MG1655 (Fꢂ kꢂ) araBAD::lolA(optimal RBS)kan
This work
This work
D
lolA
Plasmid
pKD46
pBS-araBADkan
pBS-araBADlolAkan
p34S-CM
29
30
Red recombinase expression plasmid for transformation of linear DNA in E. coli
pBluescript with KanR cassette inserted between araBAD flanking sequence
pBluescript with lolA (optimal RBS) and KanR cassette inserted between araBAD flanking sequence
A source for a chloramphenicol resistance cassette
This work
31
Oligonucleotide
lolA-Fa
lolA-R
araC-F
lolA-A
50-AAGGAGGAATAATGATGAAAAAAATT-30
50-CTACTTACGTTGATCATCTAC-30
50-CGCCAGCAGCTCCGAATAGCGCCC-30
50-GTCCATGGTGCTTTTGTTCGC-30
lolA-B
lolA-C
lolA-D
lolA-E
lolA-F
50-TCGACCTGCAGGCATGCAAGCCATTATTCCTCAAATTACGTCACT-30
50-GAGTGGCAGGGCGGGGCGTAACGTAAGTAGAGGCACCTGAGT-30
50-CTCAGGGATTTCAACAGATAG-30
50-TCTTCCATCGCCTGAGTTAG-30
50-GGATATGCTCTACTCTGGGCC-30
50-AGTGACGTAATTTGAGGAATAATGGCTTGCATGCCTGCAGGTCGA-30
50-ACTCAGGTGCCTCTACTTACGTTACGCCCCGCCCTGCCACTC-30
CM-F
CM-R
a
Bold face sequence indicates optimal ribosome binding site.
using lab strains of E. coli (MG1655), Bacillus subtilis (168) and
Pseuomonas aeruginosa (PA01). Similar to compound 1, the thio-
urea-based compounds (series 3 in Table 1) were active against
both Gram-negative species, but had no significant effect (MIC
antibacterial activity. Lastly, the addition of a methylene spacer be-
tween the sulfur and the amino groups (4f), also completely abol-
ished the antibacterial activity. In summary, substitutions on the
benzyl ring were the most tolerated, whereas other modifications
largely resulted in a significant decrease in activity.
We previously characterized a physical interaction between 1
and LolA using saturation transfer double difference NMR (STDD
NMR).4 As such, we employed this technique to investigate interac-
tions between 3a and A22 with LolA in vitro. A sample of pure
>256
Additionally, we looked at the effect that lolA overexpression had
on each analog with a MIC <16 g/mL in E. coli (data not shown).
lg/mL) when tested against the Gram-positive B. subtilis.
l
Each active structural analog was suppressed by LolA at high copy,
indicative of a similar cellular mechanism. These results were con-
sistent with the idea that compound 3a and analogs work through
the inhibition the lipoprotein chaperone LolA, as Gram-positive
bacteria do not have the lipoprotein transport system due to the
absence of an outer membrane. On the other hand, Gram-positive
bacteria have been reported to have multiple homologs of MreB.17
Two homologs, Mbl and MreBH, have been identified in B. subtilis
and their functional redundancy could account for the inactivity
of the thiourea compounds within these species.18
We explored the effects of substitutions on the benzyl group of
the S-benzylisothiourea (series 3 in Table 1) as has been previously
reported by Iwai et al.19 We observed similar trends, where most
substitutions could be tolerated, only the presence of the strongly
electron donating methoxy substituent (3k) completely abolished
activity. However, potency was affected by the size and position
of the substituent, with the moderately sized halogens, Cl and Br
at position 4, being the most favourable (3a and 3f); suggesting
that an optimal size is required to fit within a proposed drug bind-
ing pocket. Dichloro subsititions were also very favourable, where
the 2,4 dichloro substitution (3e) was the most potent compound
in the series (Table 1).
Retaining the 4-chlorobenzyl moiety, we explored some new
chemical matter by turning our attention to the effects that
increasing (3m, 3n) or decreasing (4d) the chain length between
the sulfur and the benzyl ring might have on activity. These mod-
ifications, as well as complete removal of the benzyl ring (not
shown) rendered the compounds inactive, supporting the impor-
tance of this benzyl ring and its placement in relation to the sulfur
atom. When the sulfur was replaced by nitrogen (4a), carbon (4b)
or removed altogether (4e) activity is also lost. However, oxygen at
this position retained activity (4c), suggesting some tolerance in
modifications at this position. Further, an analog of A22 was cre-
ated which replaced the hydrogen from both amines with
methyl groups (5a). This modification completely abolished the
recombinant LolA (15
centrations ranging from 0 to 1000
l
M) was incubated with 3a or A22 at con-
M. The resulting STDD NMR
l
spectra revealed considerable intermolecular magnetization trans-
fer from LolA to 3a (Fig. 7A), a process that is dependent upon a di-
rect interaction between protein and ligand. This spectrum is
depicted in comparison to the one-dimensional reference proton
NMR spectrum of 3a (Fig. 7B). Correspondingly, no STD NMR spec-
trum could be obtained for compound 4b, a structural analog with
no antibacterial activity (data not shown). To further characterize
the interaction, peak areas corresponding to the aromatic protons
of 3a and A22 were used to determine the normalized STDaf and
plotted against compound concentration (Fig. 7C and D). These
data were fit to the best hyperbolic curve and as such the Kd for
the interaction of LolA with 3a was found to be 150 50
(Fig. 7C), while the Kd for the interaction with A22 was comparable
at 200 50 M (Fig. 7D). These Kd values are greater than that re-
ported for compound 1 and LolA (ꢀ8
M)4 or A22 and MreB
(ꢀ1.3
M)11 but indicate nevertheless that these compounds show
lM
l
l
l
a bona fide interaction with LolA in vitro. It remains unclear why
the binding affinity of the parent compound 1 to LolA is higher
than that for 3a or A22. One possible explanation is that the 3,4-
dimethoxyphenethylamine portion of compound 1, while not
essential for antibacterial activity, could nonetheless be involved
in additional interactions with LolA that increase its overall bind-
ing affinity.
In conclusion, we report here on the (acid-catalyzed) degrada-
tion of MAC13243 (1), a novel probe of lipoprotein targeting, into
the active species S-(4-chlorobenzyl)isothiourea (3a). Accordingly,
we chose to investigate whether 3a and its structural analog A22
share a similar cellular mechanism to that reported for 1. Our study
demonstrated a strong high-copy suppression (16-fold MIC) of
the action of both 3a and A22, as demonstrated for compound
1 previously.4 To control for possible artifacts associated with

C. A. Barker et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2426–2431
2431
synthesis and maintaining cellular morphology.21 While the exact
role of MreB remains enigmatic a large body of work implicates
MreB in cell wall peptidoglycan synthesis.22 Lipoproteins too are
well known to interact with cell wall peptidoglycan in Gram-neg-
ative bacteria23 and have recently been linked to the cell wall bio-
synthetic machinery.24,25 Accordingly, a cell wall biosynthetic
network that links MreB and LolA function could provide a com-
posite target for A22 and related thiourea compounds.
Taken together, the results from this investigation are neverthe-
less most consistent with the conclusion that compound 1 and the
thiourea compound series represented by 3a and A22 act through a
similar mechanism. Further we have characterized new chemical
genetic and biochemical interactions with the LolA protein for
the compound A22, thought to be a specific probe of the bacterial
actin-like protein MreB. While strong biochemical and genetic evi-
dence supports MreB as the target of this compound, and presum-
ably also related compounds studied here,9,11,26 the work
presented here suggests that A22 and related thiourea compounds
may inhibit the function of both MreB and LolA. Such a mechanism
would put A22 in the company of several well-known antibiotics,
including b-lactams, fluoroquinolones, D-cycloserine and fosfomy-
cin, that are known to have multiple cellular targets.27
Acknowledgments
We thank Hirotada Mori of the Nara Institute for providing the
lolA expression clone used in these studies and Tomasz Czarny for
assisting in the preparation of figures. This work was supported by
an operating grant from the Canadian Institutes of Health Research
(MOP-81330) and by a Canada Research Chair award to E.D.B.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2013.02.
005.
References and notes
Figure 7. Kd Determination using saturation transfer double difference (STDD)
NMR. The interaction of compounds 3a and A22 with LolA was examined using
saturation transfer double difference NMR as described previously.4 (A) STDD NMR
spectrum of LolA/TU-1 interaction generated by selectively saturating the methyl
region of LolA and obtaining the difference between the spectra of the ligand alone
and the spectra of the ligand and protein complex. (B) The 1D NMR reference
spectrum for 3a. The dissociation constants (Kd) for the interaction of 3a with LolA
interaction (C) and A22 with LolA (D) were obtained by plotting the calculated STDaf
against increasing concentrations of compound. The 1D-STD titration was carried
1. Falconer, S. B.; Czarny, T. L.; Brown, E. D. Nat. Chem. Biol. 2011, 7, 415.
2. Burdine, L.; Kodadek, T. Chem. Biol. 2004, 11, 593.
3. Parsons, A. B. et al Cell 2006, 126, 611.
4. Pathania, R. et al Nat. Chem. Biol. 2009, 5, 849.
5. Giaever, G. et al Nat. Genet. 1999, 21, 278.
6. Li, X. et al Chem. Biol. 2004, 11, 1423.
7. Narita, S.; Matsuyama, S.; Tokuda, H. Arch. Microbiol. 2004, 182, 1.
8. Iwai, N.; Nagai, K.; Wachi, M. Biosci. Biotechnol. Biochem. 2002, 66, 2658.
9. Gitai, Z. et al Cell 2005, 120, 329.
out by titrating 15
or 1000 M.
lM LolA with total ligand concentrations of 25, 50, 100, 200, 600
l
10. Kruse, T. et al Genes Dev. 2006, 20, 113.
11. Bean, G. J. et al Biochemistry 2009, 48, 4852.
12. Donald, R. G. et al Chem. Biol. 2009, 16, 826.
13. Mizushima, S. Mol. Cell Biochem. 1984, 60, 5.
14. Cascales, E. et al J. Bacteriol. 2002, 184, 754.
15. Ohara, M. et al J. Bacteriol. 1999, 181, 4318.
16. Wachi, M. et al J. Bacteriol. 1987, 169, 4935.
17. Schirner, K.; Errington, J. Microbiology 2009, 155, 3611.
18. Kawai, Y.; Asai, K.; Errington, J. Mol. Microbiol. 2009, 73, 719.
19. Iwai, N. et al Biosci. Biotechnol. Biochem. 2007, 71, 246.
20. Butland, G. et al Nature 2005, 433, 531.
21. Cabeen, M. T.; Jacobs-Wagner, C. Nat. Rev. Microbiol. 2005, 3, 601.
22. White, C. L.; Gober, J. W. Trends Microbiol. 2012, 20, 74.
23. Braun, V.; Wu, H. C. Lipoproteins, Structure, Function, Biosythesis and Model
for Protein Export In New Comprehensive biochEmistry; Ghuysen, J. M.,
Hakenbeck, R., Eds.; Elsevier Science: Amsterdam, 1994; Vol. 27, pp 319–344.
24. Paradis-Bleau, C. et al Cell 2010, 143, 1110.
25. Typas, A. et al Cell 2010, 143, 1097.
26. Kruse, T. et al EMBO J. 2003, 22, 5283.
27. Hopkins, A. L. Nat. Chem. Biol. 2008, 4, 682.
28. Link, A. J.; Phillips, D.; Church, G. M. J. Bacteriol. 1997, 179, 6228.
29. Datsenko, K. A.; Wanner, B. L. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 6640.
30. Campbell, T. L.; Brown, E. D. J. Bacteriol. 2002, 184, 5609.
31. Dennis, J. J.; Zylstra, G. J. Appl. Environ. Microbiol. 1998, 64, 2710.
32. Kitagawa, M. et al DNA Res. 2005, 12, 291.
overexpression, we also conducted LolA depletion experiments and
demonstrated specific enhancement (32-fold sensitization) of the
growth inhibition by 1, 3a and A22. Interestingly, 1, 3a and A22
all induce a morphological transition from rod to round cells that
is shared by depletion of the LolA protein. Finally, through STD
NMR we characterized a physical interaction between LolA and
both 3a and A22, as we have seen for compound 1 previously.4
Notwithstanding the biochemical interactions recorded here for
A22 and the thiourea compound 3a with LolA, it remains a formal
possibility that a yet uncharacterized interaction between MreB
function in the bacterial cytoskeleton and the OM lipoprotein
transport machinery is responsible for the observed chemical ge-
netic interactions. The bacterial cell is increasingly recognized as
a composition of a highly dense and interconnected genes and pro-
teins, where we could envision inhibition of LolA or MreB having
effects on the function of the other. MreB forms one of the most
highly connected protein nodes within the cell,20 involved in
critical cell processes such as chromosome segregation, cell wall