10.1002/cmdc.201900193
ChemMedChem
COMMUNICATION
1,3,5-triazines represent a promising DegS inhibitor class with
defined structure-activity relationships and potent inhibitory
potential.
disrupting and bactericidal agent, PmBN is considerably less
antibacterial, lacks bactericidal activity and mainly acts to
disorganize and permeabilize the outer membrane. In
accordance with our findings with 1, these assays revealed that
also the more potent DegS inhibitor 3 completely impaired DegS
reporter activity at 30 µM while the inactive control compound 2
had no effects (Fig. 3A). Accordingly, these results not only
demonstrate that 3 targets DegS in vivo but also indicates that
DegS inhibitors may act synergistically with outer membrane
perturbing antibiotics. Moreover, our chemical approach
supports previous genetic evidence indicating that DegS is
essential even under optimal growth conditions.[13]
Given the well-established implications of DegS in cell
envelope protein stress response, we tested the synergy of
DegS inhibition by pyrazolo[1,5-a]-1,3,5-triazines with the outer
membrane perturbing and last resort antibiotic Col. To this end,
we first determined the minimal inhibitory concentration of Col
for E. coli ATCC25922 as 0.25 µg/mL (Supporting Table 3).
Subsequently, the corresponding MIC for various pyrazolo[1,5-
a]-1,3,5-triazines either in absence or in presence of 0.03 µg/mL
Col (corresponding to 1/8 MIC of Col) was determined (Fig. 3B).
From these values, we calculated the fractional inhibitory
concentration (FIC) index as a measure of synergy between two
compounds. For all active DegS inhibitors, a FIC index ≤ 0.5 was
determined, indicating synergy. In addition, we tested synergy
for further Gram negative bacteria. While the combination of 1
and colistin was synergistic for Acinetobacter baumanni,
Enterobacter cloacae and Klebsiella pneumoniae, no synergy
was detected for Pseudomonas aeruginosa (Supporting Table
4); as 1 alone had no effect on Pseudomonas aeruginosa but on
the three other strains (Supporting Table 1), this finding further
supports the suggested synergy mechanism. Moreover, the
synergy between the DegS inhibitor and membrane-perturbing
antibiotic is selective. We have tested the combination of 1 with
other antibiotic classes with different mechanisms of action and
could not see any increase in antibiotic susceptibility (Supporting
Table 5). Overall, these studies demonstrate that impairment of
the outer membrane stress response via DegS acts
synergistically in combination with established antibiotics
inducing outer membrane stress.
To rationalize the structure-activity relationship of the
pyrazolo[1,5-a]-1,3,5-triazines at the molecular level, we
explored the binding of 1 (parent compound with intermediate
activity) and 4 (best binder) to DegS. To validate our model, we
also used the inactive compound 2 as a negative control. The
first step was to perform docking calculations of these
compounds to the PDZ domain of DegS. Binding to the protease
domain was discarded since the parent compound 1 is not able
to inhibit DegSΔPDZ (Supporting Fig. 4). We thus identified three
main binding poses (C1, C2 and C3) of 1, 2 and 4 to the PDZ
domain of DegS (Fig 1D). The analysis of C1 and C2 allows
rationalizing why chemical modifications of positions ‘B’ and ‘C’
failed to improve the IC50 of the compounds since in these
binding poses both positions are placed in PDZ binding pockets
that are limited in space. Next, we performed molecular
dynamics (MD) simulations of the three docked compounds with
C1 and C2 as starting positions. The MD simulations then
showed that both conformations lead to protein-ligand
complexes that were conserved during the simulations (see
Supporting Information for further discussion). In C1 and C2, all
compounds also engage in interactions with loop L3 as indicated
by the dynamic network analysis (Supporting Fig. 7-9).[13] To
initially test this model, we reasoned that binding of 1 to the PDZ
domain of DegS could be contested by allosteric activators. We
therefore examined inhibition of DegS by 1 in the presence of
two peptides that are known allosteric activators i.e.
DNRLGLVYQF and DNRLGLVYWF, the affinity of which is 53
µM and 3.2 µM, respectively.[6] Both peptides prevent inhibition
of DegS, suggesting competitive binding to the PDZ domain and
thus demonstrating feasibility of the proposed binding mode
(Supporting Fig. 10).
We continued to evaluate this class of inhibitors by
measuring DegS target engagement in living E. coli cells using 1
as the starting compound as well as 3 as an inhibitor with better
inhibitor potential (note that the most active compound 4 did not
display sufficient water solubility at higher concentrations to
perform these assays) and the inactive compound 2 as a
negative control. To this end, we investigated the impact of the
three compounds on the growth of the DegS WT E. coli strain
CAG16037 as well as the degS knockout strain CAG33315
(note that strain CAG33315 must contain so far unknown
suppressor mutations allowing it to grow in the absence of degS)
(Fig. 2). In these assays, the DegS inhibitors 1 and, in
accordance with its higher inhibitory potential, to a larger extent
3 displayed dose-dependent growth defects; more importantly,
stronger effects were observed for the WT than for the degS
knockout strain. In contrast, the inactive control compound 2
was insensitive to the different genetic backgrounds. These
results demonstrate that the employed compounds target and
inhibit DegS in living bacteria, leading to differential DegS-
dependent growth defects.
In summary, we have developed the first non-covalent
small molecule inhibitors of DegS. These inhibitors are based on
a pyrazolo[1,5-a]-1,3,5-triazine scaffold and display distinct
structure-activity relationships. These compounds also inhibit
DegS in living bacteria, thus allowing for the first time to
demonstrate that in vivo target engagement induces growth
inhibitory effects, in particular under stress conditions. In fact,
chemotherapeutical induction of outer membrane stress is
synergistic to DegS inhibition. Together, these findings validate
DegS and the E stress pathway as new targets for the
development of antibiotics with a novel mode-of-action, thus
opening promising avenues for rational antibiotic drug design.
These results encouraged us to further test in vivo target
engagement. Accordingly, we modified the DegS activity
reporter strain by employing the outer membrane perturbating
antibiotics Colistin (Col) and Polymyxin B nonapeptide (PmBN)
for DegS activation. While Col is a highly potent membrane-
Acknowledgements
Financial support by CRC1093, EH 100/16-1 and KA 2894/4-1
by Deutsche Forschungsgemeinschaft to M.E., E.S.-G. and M.K.
This article is protected by copyright. All rights reserved.