3132
S. Abdeen et al. / Bioorg. Med. Chem. Lett. 26 (2016) 3127–3134
might be due to efflux of the molecules, we tested against an
That many of the most potent GroEL/ES biochemical inhibitors
were ineffective against the MC4100 acrB E. coli cells could be
MC4100
D
acrB E. coli strain, which has one of the central compo-
D
nents of the AcrA/AcrB/TolC efflux pump removed.45,46 Compounds
due to the presence of other efflux pumps that were still functional,
as it is known that E. coli contains several classes of efflux
pumps.47–50 However, another possibility is that the molecules
were not able to traverse the highly impermeable lipopolysaccha-
ride (LPS) outer membrane characteristic of Gram-negative bacte-
ria. To probe this, we tested the compounds against a mutant
SM101 lpxA2 E. coli strain, which has a temperature sensitive lpxA
allele leading to compromised LPS biosynthesis at non-permissive
8 and 18 were the most potent inhibitors of this efflux-compromised
E. coli strain (EC50 = 2.3 and 21 lM, respectively), with the remainder
of the compounds being inactive.
(A)
temperatures, and consequently
a greater permeability to
molecules.51,52 We found that 10 compounds inhibited the growth
100
18
of this E. coli strain, with compounds 8 and 18 still proving to be
the most effective (EC50 values of 0.33 and 3.3 lM, respectively).
These results were further supported by the ability of many com-
pounds to inhibit the growth of a Gram-positive bacterium, Bacillus
subtilis (Table 2 and Fig. 4B), which does not contain an LPS outer
membrane. That more compounds failed to inhibit either the
SM101 lpxA2 E. coli or B. subtilis bacteria is putatively because of
the presence of efflux pumps that were still intact for these strains,
and/or the continued impermeability of compounds across the cell
membranes.
While E. coli and B. subtilis were good model systems for initial
proof-of-principle studies, we wanted to elucidate the antibiotic
potential of our GroEL/ES inhibitors against a panel of more clini-
cally relevant bacteria, in particular the ESKAPE pathogens. We
adapted the general bacterial proliferation assay to test molecules
against E. faecium, S. aureus (plus an MRSA strain), K. pneumoniae,
A. baumannii, P. aeruginosa, and E. cloacae (see Supporting informa-
tion for detailed protocols). A summary of the EC50 values for com-
pounds against these bacteria is presented in Table 3 (and
graphically in Fig. 4B), with a comparison against several common
antibiotics. As four of the ESKAPE pathogens are Gram-negative (K.
pneumoniae, A. baumannii, P. aeruginosa, and E. cloacae), it is not
surprising that the GroEL/ES inhibitors were largely ineffective
against them. The remaining two ESKAPE pathogens, E. faecium
and S. aureus, are Gram-positive bacteria. Compound 8 was very
8
10
1
0.1
0.1
1
10
100
GroEL/ES-dMDH Refolding IC50 (μM)
(B)
100
18
potent at inhibiting E. faecium growth (EC50 = 0.15
pound 32 was moderately potent (EC50 = 15 M). Somewhat sur-
prisingly, compound 18, which emerged as lead inhibitor
lM), and com-
l
a
against the E. coli and B. subtilis cells, was inactive against E. fae-
cium. The remaining compounds were ineffective against E. fae-
cium, which again supports the notion that the presence of an
LPS membrane is not the sole determinant leading to compound
inactivity. While compounds 8 and 18 emerged as the lead inhibi-
8
10
1
tors of S. aureus growth (EC50 = 0.20
four other compounds were also moderately effective with EC50
values between 10 and 50 M (5, 11, 19, and 28). To determine if
lM and 1.8 lM, respectively),
l
they were bactericidal or bacteriostatic, we analyzed lead com-
pounds 8 and 18 against the methicillin susceptible S. aureus strain
(ATCC 25923) and found that both were acting as bactericidal inhi-
bitors (refer to Fig. S1 in Supporting information). We further
tested compounds against an MRSA strain (ATCC #BAA-44:
HPV107 strain, SCCmec Type I, Iberian PFGE Type) and found a cor-
relation with the methicillin susceptible S. aureus strain (Table 3).
While we have identified numerous inhibitors of GroEL/ES bio-
chemical function, several of which we now know inhibit the
growth of pathogenic bacteria (in particular S. aureus and MRSA),
there remained the possibility that these compounds could be
toxic to host cells owing to targeting of HSP60/10. To account for
this possibility, in vitro counter-screens were carried out in analo-
gous chaperonin-mediated dMDH refolding and ATPase biochemi-
cal assays employing HSP60/10 (Table 4). As indicated in Figure 5A,
there was a correlation observed for inhibiting both E. coli GroEL/ES
and human HSP60/10; however, compounds were generally more
potent at inhibiting E. coli GroEL/ES. Notably, compounds 1 and
0.1
0.1
1
10
100
HSP60/10-dMDH RefoldingIC50 (μM)
Liver (THLE-3)
Kidney (HEK-293)
Figure 5. (A) Compounds inhibit both E. coli GroEL/ES and human HSP60/10
chaperonin systems, but are generally more selective for the bacterial homolog.
Compound 18 is inactive against human HSP60/10, whereas compound 8 exhibits
low selectivity for GroEL/ES. (B) Even though compounds can inhibit HSP60/10
biochemical function, many exhibit low or no cytotoxicity to human liver and
kidney cells. Compound 18 exhibits no cytotoxicity, whereas compound 8 exhibits
moderate or low toxicity. Data plotted in the grey zones represent results beyond
the assay detection limits (i.e., >100 lM).