ACS Infectious Diseases
Letter
To assess these hypotheses, we now synthesized a series of
rationally designed structural analogues with side chains of
varying length (Table 1, 11e−11j) and steric bulk (11c and
11d). Two additional analogues would retain the overall 14-
carbon side chain but move the position of the ester from C8
(1) to C7 (11a) or C9 (11b), which would serve to manipulate
the macrocyclic conformation. Employing our previously
reported synthetic route,15 we began by installing an Evans
chiral auxiliary to commercially available n-eneoic acids
(Scheme 1). This enabled a stereoselective Davis oxidation
followed by TBS protection affording various fragments with
general structure 4, where n = 0, 1, or 2. Cross metathesis with
homoallylic alcohols (5), synthesized to accommodate a wide
range of alkyl substituents, followed by hydrogenation and
subsequent auxiliary removal afforded fragments with general
structure 8, where R corresponds to the alkyl tail of the final
analogues (Table 1). EDC-mediated esterification with
compound 9 (synthesis reported previously15) followed by
global deprotection afforded analogues 11a−11j, which were
obtained in overall yields ranging from 15 to 42%.
With 10 new analogues in hand, we evaluated their
inhibitory activity against the PA strains, PA14 and PAO1,
and the resistant PA strain selected for in our laboratory, RO5
(Table 1, Table S1). As we expected, the activity of the linear
alkyl chain analogues varied significantly with chain length.
Generally, those with truncated chains relative to promysalin
were less active (Table 1, 11e−11g), while extending the chain
by one or two carbons increased activity (11h, 11i). These
results indicate that the hydrophobic interactions with the alkyl
tail are important for binding affinity. However, if too many
carbons were added (11j), activity diminished, suggesting that
we had achieved optimal chain length in analogue 11h.
Alternatively, adding sterically hindered or rigid groups such as
phenyl (11c) or isopropyl (11d) to a truncated chain greatly
reduced activity, suggesting that the binding pocket does not
easily accommodate nonlinear tails. Finally, changing the
position of the ester and presumably disrupting the intra-
molecular hydrogen bonding network, decreased activity (11a,
11b), indicating that this structural feature is critical for an
optimized conformation within the binding pocket. Interest-
ingly, the activity of most analogues decreased only marginally
in the resistant strain (5−12-fold decrease) relative to
promysalin (∼50-fold decrease) (Table S1). However, other
analogues had a greater reduction in activity compared to
promysalin. Overall, a general trend was not observed between
chain length and differential inhibitory activity between
resistant and susceptible strains.
(Figure 3A, green). However, all analogues with increased tail
length were positioned such that their tail was oriented into a
previously unidentified binding cleft (Figure 3A, orange), as
the native orientation would unfavorably extend the hydro-
phobic side chain into a solvent-exposed region. We postulated
that this is the driving force for the longer analogues to occupy
this new region of the binding pocket.
We also identified a tryptophan residue (Trp161, Figure 3B,
blue) in proximity to the alkyl chain terminus in the new
binding orientation. We speculated that installing a terminal
aromatic group on the side chain would induce π-stacking
interactions with the protein resulting in a higher binding
affinity.31 Additional computational analysis revealed that an
ideal chain length would contain six carbons between the
aromatic moiety and the ester linker arriving at analogue 12
(Figure 3B, purple; Figure 3C). Analogue 12 is easily
accessible via our previously described synthetic sequence in
21% overall yield, requiring only minor modifications (Scheme
S1). We then examined its inhibitory activity against PA14.
These results revealed that 12 has an IC50 = 0.75 μM and an
IC90 = 1.8 μM. Although this is less potent than the parent
compound, it is >200 fold more active than the truncated aryl
analogue 11c. This indicates that this new binding cleft is more
amenable to changes, and this analogue provides a promising
starting point for further optimization.
In conclusion, we have synthesized 11 promysalin analogues
based on its computationally predicted binding mode in Sdh.
Biological investigation of these analogues revealed that
manipulating the hydrophobic character on the alkyl chain
has a direct effect on growth inhibition, resulting in two
analogues more potent than the natural product. Notably, a
new putative binding cleft was revealed through computational
modeling of Sdh, which has the ability to accommodate greater
structural diversity, including both linear and rigid aromatic
moieties, and leads to improved IC90 values. This provides the
basis for future work focused on optimization of binding
interactions in the newly discovered cleft combined with other
improvements previously reported by our group toward the
development of a potent PA-specific antimicrobial agent.
ASSOCIATED CONTENT
■
sı
* Supporting Information
The Supporting Information is available free of charge at
Synthetic procedures and biological assays (PDF)
Upon closer inspection of the growth curves, we noticed a
change in the slope for analogues with longer side chains
(Figure 2). For that reason, we calculated IC90 values for
inhibition against PA14 (Table 1), which are more clinically
relevant values due to the possibility that 50% inhibition may
not be sufficient for desired outcomes.28−30 Additionally, a
steeper inhibition curve indicates that small increases in
concentration result in large increases in activity, making
inhibition more attainable at physiologically relevant concen-
trations.
AUTHOR INFORMATION
■
Corresponding Author
William M. Wuest − Department of Chemistry and Emory
Antibiotic Resistance Center, Emory School of Medicine, Emory
University, Atlanta, Georgia 30322, United States;
Authors
In order to rationalize the observed biological activity, we
built computational models of these analogues in complexes
with Sdh. We were surprised to find that the positioning of the
alkyl chain in these models was not always the same as in
promysalin. All analogues with shorter alkyl chains resulted in
conformations very similar to that observed with promysalin
Savannah J. Post − Department of Chemistry, Emory University,
Atlanta, Georgia 30322, United States
Colleen E. Keohane − Department of Chemistry, Emory
University, Atlanta, Georgia 30322, United States
Lauren M. Rossiter − Department of Chemistry, Temple
University, Philadelphia, Pennsylvania 19122, United States
D
ACS Infect. Dis. XXXX, XXX, XXX−XXX
Post, Savannah J.
