Repurposing Hsp90 inhibitors as antibiotics targeting histidine kinases

Article abstract of DOI:10.1016/j.bmcl.2017.10.036

To address the growing need for new antimicrobial agents, we explored whether inhibition of bacterial signaling machinery could inhibit bacterial growth. Because bacteria rely on two-component signaling systems to respond to environmental changes, and because these systems are both highly conserved and mediated by histidine kinases, inhibiting histidine kinases may provide broad spectrum antimicrobial activity. The histidine kinase ATP binding domain is conserved with the ATPase domain of eukaryotic Hsp90 molecular chaperones. To find a chemical scaffold for compounds that target histidine kinases, we leveraged this conservation. We screened ATP competitive Hsp90 inhibitors against CckA, an essential histidine kinase in Caulobacter crescentus that controls cell growth, and showed that the diaryl pyrazole is a promising scaffold for histidine kinase inhibition. We synthesized a panel of derivatives and found that they inhibit the histidine kinases C. crescentus CckA and Salmonella PhoQ but not C. crescentus DivJ; and they inhibit bacterial growth in both Gram-negative and Gram-positive bacterial strains.

Full text of DOI:10.1016/j.bmcl.2017.10.036

Accepted Manuscript
Repurposing Hsp90 inhibitors as antibiotics targeting histidine kinases
Chau D. Vo, Hanna L. Shebert, Shannon Zikovich, Rebecca A. Dryer, Tony P.
Huang, Lindsey J. Moran, Juno Cho, Douglas R. Wassarman, Bryn E. Falahee,
Peter D. Young, Garrick H. Gu, James F. Heinl, John W. Hammond, Taylor N.
Jackvony, Thomas E. Frederick, Jimmy A. Blair
PII:
S0960-894X(17)31030-2
https://doi.org/10.1016/j.bmcl.2017.10.036
BMCL 25365
DOI:
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
26 June 2017
5 October 2017
18 October 2017
Please cite this article as: Vo, C.D., Shebert, H.L., Zikovich, S., Dryer, R.A., Huang, T.P., Moran, L.J., Cho, J.,
Wassarman, D.R., Falahee, B.E., Young, P.D., Gu, G.H., Heinl, J.F., Hammond, J.W., Jackvony, T.N., Frederick,
T.E., Blair, J.A., Repurposing Hsp90 inhibitors as antibiotics targeting histidine kinases, Bioorganic & Medicinal
Chemistry Letters (2017), doi: https://doi.org/10.1016/j.bmcl.2017.10.036
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.

Graphical Abstract
To create your abstract, type over the instructions in the template box below.
Fonts or abstract dimensions should not be changed or altered.
Repurposing Hsp90 inhibitors as antibiotics
targeting histidine kinases
Leave this area blank for abstract info.
Chau D. Vo, Hanna L. Shebert, Shannon Zikovich, Rebecca A. Dryer, Tony P. Huang, Lindsey J. Moran, Juno Cho,
Douglas R. Wassarman, Bryn E. Falahee, Peter D. Young, Garrick H. Gu, James F. Heinl, John W. Hammond,
Taylor N. Jackvony, Thomas E. Frederick, and Jimmy A. Blair
Cl
HO
O
HO
N
N
H
5b
Diaryl pyrazole Hsp90 inhibitors
also have antimicrobial activity
Bergerat fold: conserved ATP binding domains

Bioorganic & Medicinal Chemistry Letters
Repurposing Hsp90 inhibitors as antibiotics targeting histidine kinases
Chau D. Voa,b, Hanna L. Sheberta,c, Shannon Zikovicha,d, Rebecca A. Dryera,e, Tony P. Huanga,f, Lindsey J.
Morana, Juno Choa,d, Douglas R. Wassarmana,g, Bryn E. Falaheea,h, Peter D. Younga,i, Garrick H. Gua,j,
James F. Heinla, John W. Hammonda, Taylor N. Jackvonya,k, Thomas E. Frederickl, and Jimmy A. Blaira,*
a Williams College, Department of Chemistry, 47 Lab Campus Drive, Williamstown, MA 01267, USA
b Present address: Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA
c Present address: Department of Chemistry and Chemical Biology, Cornell University, Ithaca, NY 14853, USA
d Present address: University of Michigan Medical School, 1301 Catherine Street, Ann Arbor, MI 48109, USA
e Present address: Emory University School of Medicine, 1648 Pierce Drive NE, Atlanta, GA 30307, USA
f Present address: Harvard University, Chemical Biology Program, Cambridge, MA 02138, USA
g Present address: Department of Cellular and Molecular Pharmacology, University of California, San Francisco, San Francisco, CA 94158, USA
h Present address: Harvard Medical School, Boston, MA 02115, USA
i Present address: Columbia University Internal Medicine Residency Program, New York, NY 10032, USA
j Present address: University of Massachusetts Medical School, 55 Lake Avenue North, Worcester, MA 01655, USA
k Present address: University of Connecticut School of Medicine, 263 Farmington Avenue, Farmington, CT 06032, USA
l Department of Biochemistry & Molecular Biophysics, Washington University School of Medicine, St. Louis, MO 63110, USA
ARTICLE INFO
ABSTRACT
Article history:
Received
Revised
Accepted
Available online
To address the growing need for new antimicrobial agents, we explored whether inhibition of
bacterial signaling machinery could inhibit bacterial growth. Because bacteria rely on two-
component signaling systems to respond to environmental changes, and because these systems
are both highly conserved and mediated by histidine kinases, inhibiting histidine kinases may
provide broad spectrum antimicrobial activity. The histidine kinase ATP binding domain is
conserved with the ATPase domain of eukaryotic Hsp90 molecular chaperones. To find a
chemical scaffold for compounds that target histidine kinases, we leveraged this conservation.
We screened ATP competitive Hsp90 inhibitors against CckA, an essential histidine kinase in
Caulobacter crescentus that controls cell growth, and showed that the diaryl pyrazole is a
promising scaffold for histidine kinase inhibition. We synthesized a panel of derivatives and
found that they inhibit the histidine kinases C. crescentus CckA and Salmonella PhoQ but not C.
crescentus DivJ; and they inhibit bacterial growth in both Gram-negative and Gram-positive
bacterial strains.
Keywords:
Antibacterial
3,4-pyrazoles
Histidine kinase
Two-component signaling
Bergerat fold
2009 Elsevier Ltd. All rights reserved.
———
* Corresponding author. Tel.: +1-413-597-4417; fax: +1-413-597-4150; e-mail: jab6@williams.edu .

With the worldwide rise in the frequency of drug-resistant bacterial infections, we face a critical clinical need for new therapies.10
One strategy to address this need involves the development of antimicrobial agents that target novel bacterial functions, and one such
class of targets is the two-component signaling (TCS) system, the signaling machinery employed by bacteria to respond to diverse
environmental signals such as nutrients, quorum signals, and ions (Figure 1A).12, 13 Bacteria rely on TCS pathways, which are mediated
by histidine kinases (HKs) and absent in humans,16 to respond to extracellular changes. This ability to respond and adapt is crucial for
survival in the harsh environments that bacteria encounter;18 hence several HKs are essential, including WalK (YycG) in
Staphylococcus aureus20 and CckA in Caulobacter crescentus.21 This suggests that disrupting TCS pathways will provide a novel
strategy for antibiotic drug discovery, because they play central roles in bacterial physiology.
Inhibition of HK function by targeting the conserved catalytic machinery of HKs is a promising strategy to disrupt TCS pathways.8,
12, 13, 17, 19 Upon activation, typically from its extracellular signaling domain interacting with its target ligand, HKs bind ATP at the
catalytic and ATP-binding (CA) domain and catalyze the transfer of the γ-phosphate of ATP to a conserved histidine residue on the HK
dimerization and histidine phosphotransfer (DHp) domain (Figure 1A).16, 22 HKs participate in a signaling cascade that transfers this
phosphate group from the conserved, phosphorylated histidine on a HK to a conserved aspartic acid in a response regulator domain
found on a response regulator protein (RR), the HK’s cognate signaling partner (Figure 1A).16, 22 Typical RRs are multi-domain
proteins harboring both a response regulator domain and a DNA-binding domain. Activation of the RR by phosphorylation triggers the
RR to bind to DNA and activate gene transcription, eliciting a genetic response to the extracellular stimulus (Figure 1A).16, 22 ATP
competitive inhibitors are one method for inhibiting these kinases, and this strategy has the potential to globally inhibit all bacterial
histidine kinases, thereby affording broad-spectrum antibacterial activity.12, 13 Indeed, recent drug discovery efforts in this area have
demonstrated that inhibitors of multiple HKs display growth inhibition against both Gram-positive and Gram-negative bacterial
strains.8, 17, 19, 23
To find a lead molecules, we asked whether inhibitors of eukaryotic Hsp90 chaperones could serve as potent lead molecules for
HKs. Histidine kinases are members of the GHKL superfamily that includes four families of functionally diverse proteins: the HKs, the
Hsp90 molecular chaperones, the MutL DNA-mismatch-repair enzymes, and the DNA-remodeling DNA gyrases. This superfamily is
defined by the Bergerat fold, a unique ATP-binding domain crucial for ATP binding and hydrolysis and one that is not found in other
protein families (Figure 1B).24 The shared structural and sequence homology in the Bergerat fold of HKs with Hsp90 molecular
chaperones suggested that Hsp90 inhibitors could also inhibit HKs (Supplemental Figure 1). Because cancer cells are often reliant on
the activity of Hsp90 to properly fold oncogenes, Hsp90 is an important drug target in its own right for the treatment of cancer, and
many inhibitors of Hsp90 have been developed, providing a selection of potential molecules for our screening efforts.25-27 This strategy
showed early success: Guarnieri and co-workers demonstrated that radicicol (Figure 2), a potent Hsp90 inhibitor, weakly bound to the
Salmonella PhoQ HK with a Kd = 715 µM,1 and in a follow- up study with a small library of Hsp90 inhibitors, they found one hit, JH-II-
126, that also bound weakly with a Kd = 391 µM.7 Neither of these studies, however, demonstrated antimicrobial activity of their hits.
Similarly, Boibessot et al. reported their rational design of HK inhibitors based on inhibitors of DNA gyrase. They found that their
inhibitors, which used a thiophene scaffold, inhibited both WalK and PhoR in vitro and demonstrated broad-spectrum antibacterial
activity.8 With these efforts as inspiration, we explored whether Hsp90 inhibitors could be repurposed as antibiotics by inhibiting
histidine kinases.
To this end, we purchased six commercially available Hsp90 inhibitors (Figure 2: NVP-AUY922,28 CCT018159,29 BIIB 021,30 PU
H71,31 and CUDC-30532) and tested them against the CckA histidine kinase from Caulobacter crescentus, which is a nonpathogenic,
Gram-negative bacterium that is a member of α-proteobacteria, and it serves as a model system for bacterial cell development and two-
component signal transduction.33, 34 Crucially, CckA is essential in C. crescentus, and its activity is required for controlling cell-cycle
progression as well as the transcription of more than 90 genes.21, 35-37 Moreover, CckA is conserved throughout α-proteobacteria,38
including the plant pathogen Agrobacterium tumefaciens39 and the zoonotic pathogen Brucella abortus.40 To determine which Hsp90
inhibitors interacted with CckA, we formulated the inhibitors in DMSO and measured the increase in thermal stability of CckA in a
ThermoFluor thermal shift assay at an inhibitor concentration of 30 µM (Table 1).41-43 This assay—alternatively called differential
scanning fluorimetry42, 44—has recently been used in a fragment-based discovery effort against the CA domain of HKs,17 in a screen for
small molecules that bind the HK extracellular receptor domain of the Staphylococcus aureus WalK ,45, 46 and in a screen to identify
amino acids that bind the extracellular domain of HK chemoreceptors in Pseudomonas syringae,47 highlighting its broad utility in HK
drug discovery and biochemistry. We used a construct of CckA (CckA-Δ3), comprised of amino acids 190–562 that includes its DHp
domain, its CA domain where ATP binds, and the first of its two Per-Arnt-Sim (PAS) domains that are involved in mediating
signaling.22, 48, 49 We chose this construct because of its ease of expression and purification from E. coli and its favorable thermal
denaturation properties in ThermoFluor assays. As a positive control for thermal stabilization of CckA-Δ3, we found a thermal shift of
3.5 ± 0.3 °C using 10 mM AMPPNP, a non-hydrolyzable ATP analog. We discovered that only two Hsp90 inhibitors—NVP-AUY922
and CCT018159—shifted the melting temperature of CckA-Δ3 greater than 1.5 °C relative to the DMSO-only control (Table 1),
suggesting that both these inhibitors bound tightly to CckA-Δ3.
Hsp90 inhibitors NVP-AUY922 and CCT018159 also inhibited CckA activity in vitro. We measured inhibition against CckA-ΔTM,
a construct of CckA spanning amino acids 70–691 that contains the entire cytosolic portion of CckA and includes an N-terminal His6-
tag for affinity purification,21, 50 using a continuous ATPase assay that couples the formation of ADP from ATP hydrolysis to NADH
oxidization using a pyruvate kinase/lactate dehydrogenase (PK/LDH) coupling system.51, 52 In this assay we followed the oxidation of
NADH by monitoring the decrease in absorbance of NADH at 340 nm as a function of time. This assay measures the activity of any
ATPase in the reaction mixture, and contaminant ATPase could lead to false positive activity. To ensure we were only measuring CckA
activity, we purified a kinase-dead variant—CckA-ΔTM H322A51—using identical purification methods as the wild-type variant, and
we demonstrated that its activity is negligible relative to the wild-type (Supplemental Figure 4). We screened all six Hsp90 inhibitors
against CckA-ΔTM and found that only NVP-AUY922 and CCT018159 significantly inhibited CckA-ΔTM below 50 µM, thereby
corroborating our ThermoFluor results (Table 1).

NVP-AUY922 and CCT018159 share structural similarity. Both feature a central five-membered heterocycle—NVP-AUY922
includes an isoxozole, whereas CCT018159 includes a pyrazole—that connects two aromatic rings, one of which is a resorcinol ring
that is essential for binding to Hsp90.29, 53 To explore the structure-activity relationship (SAR) of this scaffold, we synthesized a small
panel of CCT018159 analogs. We chose to focus on analogs of CCT018159 because these inhibitors could be readily synthesized in
three steps and because our thermal shift data suggested CCT018159 interacted with CckA more tightly than NVP-AUY922.
Furthermore, these pyrazole derivatives have been reported during the development of CCT018159 as an Hsp90 inhibitor.54-56 This
library was designed to allow us to explore the steric and electronic effects of the 4-position of the resorcinol ring, because substitutions
at this position were shown to be important for activity against Hsp90,29, 53 and we asked whether similar effects played a role in HK
inhibition. Likewise, we designed inhibitors with and without a 5-methyl pyrazole core to explore steric effects of the pyrazole ring.
Finally, we explored three different aromatic rings off the pyrazole 4-position to ask whether increased polar character and size affects
potency and solubility.

A
B
Signaling domain
HK~P
D
CA
H
H
P
H
D
H P
ADP
DHp
ATP
HK
monomer
P
D P
DNA
binding
domain
RR
RR~P
DNA
Figure 1. (A) A typical TCS pathway. The conserved histidine residue (H) is illustrated on the HK (cyan) in the DHp domain. Likewise, the conserved
aspartic acid residue (D) on the RR (yellow) is shown. (B) A structure-based alignment3, 4 of the CA domain of the CckA HK (PDB 5IDM5, cyan) and the
ATPase domain of yeast Hsp90 (PDB 1AMW11, magenta). Co-crystalized nucleotides (AMPPNP, CckA; ADP, Hsp90) are shown as ball-and-sticks. Core
structural features surrounding the nucleotides are highly conserved. These domains share 12% sequence identity. RMSD of aligned Cα atoms is 1.92 Å (87
residues).18
Figure 2. Hsp90 inhibitors. Guarnieri et al. demonstrated that radicicol1, 2
and JH-II-1267 bind the PhoQ HK.

We synthesized a small library of diaryl pyrazoles following synthetic schemes described by Cheung29 and Drysdale.53 Inhibitors
were prepared in two or three steps from commercially available resorcinols and phenylacetic acids. Scheme 1 shows the synthetic
route to 3,4-diarylpyrazoles 5a–e with a methyl group at the 5-position of the pyrazole ring. This three-step synthesis begins with
forming dihydroxyphenyl ketones 3a–e via Friedel-Crafts acylation of the respective resorcinols and substituted phenylacetic acids
using boron trifluoride diethyletherate (BF3•OEt2) as a Lewis acid catalyst. These products were then refluxed with acetic anhydride in
the presence of potassium carbonate to form alkylated isoflavones 4a–e, which are then converted to the final 3,4-diaryl pyrazoles 5a–e
by the heating the isoflavones with hydrazine hydrate. To synthesize inhibitors 7a–d, which lacked a 5-methyl group on the pyrazole
ring, we used a one-pot method57 to synthesize isoflavones 6a–d, followed by pyrazole ring formation. In Scheme 2 isoflavones 6a–d
are first formed by Friedel-Crafts acylation between resorcinols and substituted phenylacetic acids using BF3•OEt2, followed by
addition of freshly prepared N,N’-dimethyl(chloromethylene) ammonium chloride, the Vilsmeier reagent, which we generated in a
separate reaction between PCl5 and DMF. The second step of this scheme involves the formation of unalkylated pyrazoles 7a–d by
refluxing 6a–d with hydrazine hydrate in a manner similar to the final step of Scheme 1.
We then tested these inhibitors against CckA using both thermal shift binding assays and ATP hydrolysis assays (Table 1). Our
results show that inhibitors 5a, 5b, and 5c were the most potent out of this series. We found that the 4-chlororescorcinol group was
important for potency (compare 5b to 5d and 5e, Table 1), while the substitution pattern of the aromatic ring at the 4-position had less
of an impact on binding. The methyl substitution at the 5-position of the pyrazole ring also increased potency (compare 5b to 7b, Table
1). The IC50 values for inhibition of yeast Hsp90 ATPase activity in the presence of 370 µM54, 56 or 400 µM29 ATP have been reported
by others for six compounds in this study (radicicol, 0.26 µM;56 CCT018159, 7.1 µM;29 5a, 1.65 µM;29 5b, 1.7 µM;54 5c, 0.65 µM;54 and
5d, 36.9 µM54). Comparing our biochemical results for inhibition of CckA (Table 1) to those against Hsp90, we see that this series of
diaryl pyrazoles shows greater potency for yeast Hsp90 than for CckA by approximately a factor of ten. Radicicol, however, is very
potent against Hsp90 but very weakly potent against CckA, highlighting that despite the similarity of these molecular targets, there is
selectivity in inhibitor binding between the two proteins.
Scheme 1. (a) BF3•OEt2, 80 °C, overnight; (b) acetic anhydride, K2CO3, DMF, 160 °C, 4 h; (c) hydrazine hydrate, ethanol, 60 °C, 1 h.
Scheme 2. (a) one pot: (i) BF3•OEt2, 80 °C, overnight, (ii) PCl5 and DMF reacted independently, then added, rt, 1h; (b) hydrazine hydrate, ethanol, 60 °C, 1 h

Table 1. ThermoFluor and inhibition data against the CckA HK.
Compound
CckA-Δ3 thermal shifta
ΔTm ± SD (°C)
0.9 ± 0.4
CckA-ΔTM inhibitionc
ATPase IC50 / µM
184 (160–212)
Radicicol
NVP-AUY922
1.8 ± 0.2
7.3 (6.2–8.6)
CCT018159
4.0 ± 0.2
28.0 (20.6–38.1)
87% activity at 250 µMd
61% activity at 250 µMd
79% activity at 250 µMd
n.t.
BIIB 021
0.7 ± 0.5
PU H71
0.5 ± 0.4
CUDC-305
0.4 ± 0.5
3.5 ± 0.3b
AMPPNP
5a
5b
5c
5d
5e
7a
7b
7c
7d
4.5 ± 0.2
17.5 (16.1–19.0)
12.3 (10.8–14.0)
11.2 (9.7–13.0)
5.0 ± 0.5
5.3 ± 1.0
2.5 ± 0.1
46.0 (39.4–53.7)
18.6 (16.1–21.4)
33.6 (27.8–40.7)
32.7 (28.5–37.6)
32.3 (29.5–35.4)
56.9 (48.4–66.8)
4.1 ± 0.5
3.5 ± 0.6
4.2 ± 0.5
3.3 ± 0.3
3.5 ± 1.2
a Mean thermal shift of Tm, the midpoint of the thermal denaturation curve, relative to DMSO control; n = 4–6; conditions included
1.6 µM kinase and 30 µM of each compound; thermal melt curves are shown in Supplemental Figures 2–3; SD = standard deviation.
b Mean thermal shift of Tm, the midpoint of the thermal denaturation curve, relative to ultrapure water control; n = 4; conditions
included 1.6 µM kinase and 10 mM AMPPNP; thermal melt curves are shown in Supplemental Figure 2; SD = standard deviation.
c IC50 (95% confidence interval); n = 2; conditions included 5 µM kinase and 100 µM ATP; n.t. = not tested.
d Highest concentration of compound test was 250 µM; percent activity represents the residual ATPase activity measured at this
concentration.
Inhibition of multiple HKs may provide broad-spectrum antibacterial activity,12, 13 and recent drug discovery efforts8, 17, 19 against
HKs have focused on the development of pan-HK inhibitors. To test whether the compounds described here inhibit HKs other than
CckA, we measured the inhibition of two additional HKs, Salmonella typhimurium PhoQ, which is involved in Mg2+ sensing and is
required for virulence,58, 59 and Caulobacter crescentus DivJ, which regulates cellular differentiation.60, 61 We chose PhoQ and DivJ as
targets because of their diversity in the ATP-binding site: in the N-box of the HK catalytic core24, 62 sits a residue conserved as a Phe,
Tyr, His, or Ala, and the aromatic amino acids are unique to HKs among GHKL family members.6, 12, 63, 64 Residues at this position sit
above ATP, and aromatic sidechains—for example, Tyr393 in E. coli PhoQ (Supplemental Figures 1 and 5)6—make π–π stacking
interactions with the adenine ring. Wilke and Carlson hypothesized that these residues could be exploited to endow selectivity toward
HKs,12 and because PhoQ has Tyr, DivJ has Phe, and CckA has Ala at this position, we felt this suite of HKs could help define the role
this residue position played in inhibitor binding. We note that at this same position Hsp90 contains an Ala residue (Supplemental
Figure 1), as does CckA. We expressed and purified C. crescentus DivJ and Salmonella typhimurium PhoQ. Each construct included
the DHp-CA domains necessary for catalysis, with the DivJ construct49, 65 spanning residues 188–585 and including an N-terminal His6-
tag, and the PhoQ construct spanning residues 257–4871 and including an N-terminal His6-SUMO tag for affinity purification and
protein solubility.66 We used a radioactive phosphotransfer assay in a dot blot format49 that measures the transfer of the γ-32P-phosphate
from [γ-32P]-ATP to the HK. This assay has been used to quantify the in vitro activity of CckA and DivJ both in solution and on
liposomes.49 Initially, we screened all Hsp90 inhibitors and our synthesized diaryl pyrazoles at 500 µM against DivJ and PhoQ,
quantifying the percent autophosphorylation relative to a DMSO-only control. Inhibitors showed selectivity between these HKs:
treating with diaryl pyrazoles or NVP-AUY922 lowered PhoQ phosphotransfer activity to 9–87% of an untreated control but failed to
lower DivJ phosphotransfer activity, with only CCT018159 showing DivJ activity below 85% (70% activity, Table 2). Hsp90 inhibitors
BIIB 021, PU H71, and CUDC-305 showed little-to-no inhibition of either PhoQ or DivJ, which is consistent with our results with
CckA (Table 1). We then selected a subset of these inhibitors—NVP-AUY922, CCT018159, 5b, 5c, and 7c—to characterize their dose-
response activity against all three HKs in a phosphotransfer assay. Inhibition assays confirmed the selectivity these inhibitors show:
they inhibit CckA-ΔTM (IC50 = 14–44 µM) more potently than PhoQ (IC50 = 238–270 µM), and they show no measurable inhibition at
800 µM against DivJ. CCT018159 inhibited all activity of DivJ at 800 µM, which represented a very steep dose-response between 400
µM and 800 µM, suggesting a non-specific mechanism67 of CCT018159 against DivJ, which we attribute to aggregation of DivJ at high
CCT018159 concentrations (see below). These data suggest that the topology of the HK active site shows diversity that may be difficult

to globally exploit, and they highlight that diaryl pyrazoles may not effectively exploit the conserved aromatic rings as a selectivity
filter between HKs and Hsp90.
Table 2. Inhibitory data against multiple histidine kinases.
Compound
PhoQ
DivJ
PhoQ inhibitionb
IC50 / µM
DivJ
CckA-ΔTM
autophosphorylationa
autophosphorylationa
inhibitionb
inhibitionb
Mean % activity ± SD Mean % activity ± SD
IC50 / µM
n.t.
IC50 / µM
n.t.
Radicicol
86 ± 10
123 ± 13
109 ± 16
70 ± 8
n.t.
NVP-AUY922 73 ± 3
56% activity at 800 µMc
No inhibition
No inhibitiond 30 (21–45)
28 (21–37)
CCT018159
9 ± 7
261 (200–339)
BIIB 021
87 ± 8
103 ± 12
110 ± 23
45 ± 3
66 ± 8
35 ± 2
76 ± 2
36 ± 3
63 ± 10
62 ± 4
63 ± 4
87 ± 10
128 ± 2
122 ± 14
96 ± 17
115 ± 5
118 ± 20
113 ± 20
117 ± 9
118 ± 6
92 ± 12
112 ± 6
117 ± 14
85 ± 10
n.t.
n.t.
n.t.
PU H71
n.t.
n.t.
n.t.
CUDC-305
n.t.
n.t.
n.t.
5a
5b
5c
5d
5e
7a
7b
7c
7d
n.t.
n.t.
n.t.
238 (195–289)
No inhibition
14 (10–20)
28 (18–42)
n.t.
270 (220–332)
No inhibition
n.t.
n.t.
n.t.
n.t.
n.t.
n.t.
n.t.
n.t.
n.t.
n.t.
n.t.
48% activity at 800 µMc
No inhibition
n.t.
44 (30–65)
n.t.
n.t.
a Relative phosphotransfer as a percentage of DMSO-only control, which gives 100% activity; n = 3; conditions included 4.5 µM
kinase, 3 µM ATP, and 500 µM of each compound; SD = standard deviation.
b IC50 (95% confidence interval); n = 2; conditions included 4.5 µM kinase and 3 µM ATP; n.t. = not tested.
c Highest concentration of compound test was 800 µM; percent activity represents the residual ATPase activity measured at this
concentration.
d No inhibition observed at 400 µM CCT018159; at 800 µM, complete inhibition is observed, which we attribute to non-specific
aggregation.
First generation HK inhibitors were found to demonstrate non-specific inhibition through protein aggregation,63, 68 and recent HK
inhibitor discovery efforts have shown the utility of gel-based aggregation studies to identify aggregators.17, 19, 63 To assess whether the
inhibitors caused non-specific protein aggregation, we screened inhibitors against all four proteins to determine whether inhibitors
caused HK aggregation under the conditions we measured our biochemical assays. After incubation of compounds with each HK, we
separated each reaction by native polyacrylamide gel electrophoresis (native-PAGE) and visualized protein migration with colloidal
coomassie staining. We included negative controls of DMSO and ultrapure water, and we included Congo Red, a known protein
aggregator that displays nonspecific enzyme inhibition through aggregation,69 as a positive control. Inhibitors reported in this study did
not aggregate CckA at up to 333 µM (Figure 3), and a subset of these inhibitors screened against DivJ and PhoQ demonstrated no
aggregation at 400 µM, and only CCT018159 caused aggregation at 800 µM with DivJ (Figure 3). These results are consistent with our
phosphotransfer assay results for CCT018159 with DivJ, where we observed no inhibition at 400 µM and complete inhibition at 800
µM (Table 2), and we attribute this behavior to aggregation of DivJ by CCT018159. We note that CCT018159 does not aggregate
PhoQ at 800 µM, suggesting that native-PAGE aggregation studies have some target-based bias.

Figure 3. Gel-based aggression studies using Native-PAGE. Congo Red is a positive control for protein aggregation. (A) Conditions: 4.5 µM CckA-ΔTM,
333 µM compound, 5% (v/v) DMSO. (B) Conditions: 4.5 µM CckA-Δ3, 333 µM compound, 5% (v/v) DMSO. (C) Conditions: 4.7 µM PhoQ, 400 or 800 µM
compound, 10% (v/v) DMSO, 10% (v/v) glycerol. (D) Conditions: 4.5 µM DivJ, 400 or 800 µM compound, 10% (v/v) DMSO, 10% (v/v) glycerol. We denote
protein shifts indicative of protein aggregation (*).
In addition, first generation HK inhibitors also caused membrane damage and hemolysis,70 demonstrating a general cytotoxicity that
suggested those early chemical leads were poor starting points for antibiotic development. In an effort to define whether the inhibitors
described in this study demonstrate adverse effects on cell membranes, we screened inhibitors in a hemolysis assay for cell lysis. We
incubated 0.5 mM of each compound with citrated sheep red blood cells, and we determined the percentage of cell lysis by measuring
the absorbance of hemoglobin at 540 nm in cleared cell lysates in the presence of each compound compared to a positive control treated
with 1% (v/v) Triton X-100, which causes complete cell lysis and release of hemoglobin (Table 3). We found that the inhibitors
reported here show minimal hemolysis of sheep red blood cells with only CCT018159 (14%) and 5e (15%) showing hemolysis greater
than 3% among diaryl pyrazoles. Inhibitors CCT018159 and 5e are the only diaryl pyrazoles reported here that contain an ethyl
substituent at the 4-position of the resorcinol ring, suggesting that placing a nonpolar substituent at this position leads to undesirable
effects on membrane integrity.
To determine the mechanism of inhibition of diaryl pyrazoles against CckA, we measured the enzyme activity while varying both the
concentration of substrate ATP and inhibitor 5b. Given the structural similarity between HKs and Hsp90, we hypothesized that analogs
of CCT018159 would bind the HK ATP-binding site and therefore be competitive inhibitors of HKs because crystallographic studies of
Hsp90 with CCT018159 (PDB 2BRC)29 or NVP-AUY922 (PDB 2VCI)28 show that these inhibitors bind to the ATP-binding site of
Hsp90. Using our ATPase coupled enzyme assay with CckA-ΔTM, we measured the initial velocity at each inhibitor concentration and
substrate concentration (Figure 4). The data were fit to enzyme inhibition models with shared Vmax, Km, and Ki parameters.71 The data fit
best to a competitive model of inhibition, which is consistent with our hypothesis that these inhibitors bind to the ATP-binding site of
CckA. Given the close structural similarity among the diaryl pyrazoles, we predict that other diaryl pyrazoles inhibit HKs competitively
with ATP.

Figure 4. Inhibitor 5b inhibits the CckA HK competitively with ATP.
Inset, Lineweaver-Burk plot. Initial rates were measured in triplicate, and the
mean and standard deviation are shown. The best fit values (95% confidence
interval) for a competitive inhibition model are shown. 5 μM CckA-ΔTM was
used.
Because 5b inhibited CckA competitively, we asked which molecular interactions mediated binding to the ATP-binding site. To
predict these interactions, we performed molecular docking studies of 5a–e and 7a–d with both CckA and PhoQ. For these calculations,
we chose crystal structures of the CA domains of C. crescentus CckA (PDB 5IDM5) and E. coli PhoQ (PDB 1ID06) that each contained
the ATP-analog AMPPNP and a fully defined ATP lid, a flexible loop involved in nucleotide binding.24, 64 We prepared the targets using
the OEDocking toolkit72 by retaining a fragment of AMPPNP (AMPPNP trimmed back to 9-methyladenine) and two conserved water
molecules that make direct polar interactions with the adenine ring of AMPPNP. Using HYBRID,73 which uses a ligand-guided
approach to docking, we docked the diaryl pyrazole inhibitors, NVP-AUY922, and AMPPNP to each target with constraints for polar
interactions with both retained water molecules and a conserved Asp (Asp479 in CckA; Asp415 in E. coli PhoQ) in the D/G1-box that
hydrogen bonds to the N6 exocyclic amine of ATP. The top-scoring predicted binding modes of 5b are shown in Figure 5A-B, and
Supplementary Figures 6–7 show the remaining top-scoring docking results. The docking poses of AMPPNP compared closely
(RMSD <0.8 Å for 11 atoms) with identical atoms in the guide ligand, suggesting a successful re-docking of the ligand and validation
of this method (Supplementary Figures 6–7). The docking modes of 5b in both CckA and PhoQ were nearly identical (Figure 5A-B),
suggesting the key binding interactions between CckA and PhoQ are similar. Importantly, the predicted binding pose suggests that 5b
makes direct hydrogen-bonding interactions with the D/G1-box conserved Asp in the active site through both hydroxyl groups of the
resorcinol ring. This interaction is an important recognition element for the GHKL super family24: it is a key element in the binding of
ATP-competitive inhibitors against Hsp90 (for example, Figure 5C9), and this same interaction is seen in the PhoQ-radicicol structure
(PDB 3CGY), the sole crystal structure of an HK bound with a ligand other than an ATP analog at the ATP-binding site.1 These poses
help explain, in part, the SAR we observe, where the 4-chlororesorcinol derivatives 5a–c and 7a–c are more potent than the
ethylresorcinol 5e and resorcinol (5d, 7d) derivatives. Substituents at the 4-position of the resorcinol ring are predicted to occupy a
pocket largely composed of hydrophobic residues with a single, conserved Asn “lid” of the N-box (Leu446-Met472-Val386-Asn389 in
PhoQ; Leu510-Phe536-Val431-Asn434 in CckA). The predicted inhibitor binding modes suggest that larger aliphatic substituents such
as 4-ethyl in CCT018159 and 5e may not provide an optimal fit in to this pocket, while inhibitors lacking any substitution at 4-
resorcinol position do not complement this pocket. The 4-chlororesorcinol derivatives may inhibit best because they include a
substituent at the 4-resorcinol position that fills this pocket with enough polar character to limit unfavorable hydrophobic interactions
with the conserved Asn “lid”. The most common pose for inhibitors 5a–e, which contain a 5-methyl group on the pyrazole ring,
positions the 5-methyl group pointed out toward solvent in PhoQ and within 2.4 Å of Thr499 and 3.6 Å of Phe497 in CckA. We are
unable to explain the increased potency of this series relative to 7a–d, but we speculate that perhaps it makes favorable hydrophobic
contacts with the flexible ATP-lid. Taken together, these data suggest the active sites of two different HKs accommodate diaryl
pyrazoles via similar chemical interactions.
Moreover, the predicted poses of the diaryl pyrazole inhibitors illustrate a potential strategy for developing HK-selective inhibitors
from this series. Our PhoQ docking model for all the inhibitors described here show an important π–π stacking interaction with Tyr393
of the conserved N-box, which is a hydrophobic interaction observed between the adenine ring of AMPPNP and PhoQ (PDB 1ID06)
(Figure 5B, Supplementary Figure 7). This same π–π stacking interaction has been reported for similar docking studies with

compounds that share little structural similarity with the diaryl pyrazole scaffold, suggesting that this interaction is key for maintaining
productive ligand-protein interactions at the ATP-binding site of HKs.8, 17, 63 HKs conserve Ala, Tyr, Phe, and His at this site of the N-
box (Supplementary Figures 1 and 5),6, 12, 63 and because the aromatic residues Tyr, Phe, and His are conserved only in HKs among
the GHKL family members, Wilke and Carlson proposed exploiting this interaction as a selectivity filter between HKs and gyrase B,
MutL, and Hsp90.12 Our docking results support this hypothesis. A structure-based sequence alignment of the CA domain of HKs with
the Hsp90 ATPase domain shows that Hsp90 bears an Ala residue (Ala41, yeast; Ala55, human) at the end of the N-box where PhoQ
bears Tyr393 (Supplementary Figure 1). This smaller residue opens additional space within the N-box helix in Hsp90 compared to
PhoQ, and consequently, in crystals structures of diaryl pyrazoles bound to Hsp90—for example Hsp90 bound with VER-49009 (PDB
3OWB9), a close structural analog of 5b (Figure 5C)—the pyrazole ring pivots toward the N-box helix relative to the docking mode
calculated for PhoQ and CckA (Figure 5A–B). A structural alignment of the PhoQ-5b docking result with Hsp90-VER-49009
illustrates that the PhoQ HK is likely unable to accommodate the Hsp90-like binding mode because of direct steric interactions with
Tyr393 (Figure 5D). Therefore, we hypothesize that molecules that maintain the favorable π–π interactions with the N-box aromatic
residues and the hydrogen-bonding interactions with the D/G1-box Asp and ordered waters in the active site but are unable to pivot may
provide selectivity for HKs over Hsp90 inhibitors. What is striking is that CckA harbors Ala at this same position (Ala438), yet the
predicted binding mode adopts the PhoQ-like trajectory of the diaryl pyrazole rather than the Hsp90-like mode. It is tempting to suggest
that N-box aromatic residues are not strictly necessary to differentiate the HK active site from an Hsp90 active site, and we believe
structural evidence of a diaryl pyrazole bound to CckA bound would help resolve this outstanding question. We note that one possible
strategy to select for HK inhibition over Hsp90 inhibition could include exploiting the F-box,12, 13 a sequence motif that appears N-
terminal to the ATP-lid, that contains at least one Phe residue, and that is conserved in HKs but not other GHKL members.24 Our
docking results (Figure 5A–B) place this motif approximately 10.5 Å away from the C4-position of the pyrazole ring among our diaryl
pyrazoles, suggesting that derivatives at the pyrazole C4-position may be able to exploit this motif. Future work will focus on the design
of derivative molecules that bind the HK fold but have lost interaction with Hsp90 and other GHKL family members.
A
B
C
D
Figure 5. (A) Predicted binding mode of 5b with the CckA CA domain. The top-scoring pose of 5b with CckA (PDB 5IDM5) is shown. (B) Predicted
binding mode of 5b with the PhoQ CA domain. The top-scoring pose of 5b with PhoQ (PDB 1ID06) is shown. For (A) and (B) the HK is cyan with key
conserved HK elements highlighted (N-box, orange; D/G1-box, green; ATP lid, slate blue; G2-box, white/grey), amino acid sidechains within 5 Å of 5b are
shown as lines, 5b is yellow sticks, the conserved Asp that hydrogen bonds to the exocyclic ATP nitrogen is green sticks, and the conserved first Phe of the F-
box is shown as transparent, cyan spheres. (C) Co-crystal structure (PDB 3OWB9) of human Hsp90 with VER-49009, a close structural analog of 5b. Hsp90 is
magenta, and motifs conserved with HKs are highlighted as above. (D) Docked binding mode of 5b with PhoQ overlayed with VER-49009 from Hsp90. PhoQ
Tyr 393 is shown as transparent, orange spheres.
We next aimed to demonstrate that the inhibitory activity of CCT018159 analogs against HKs in vitro translated to an in vivo model.
We used the broth-dilution method to determine the minimum inhibitory concentration (MIC) of these compounds (Table 3).74 We
tested inhibitor susceptibility against model Gram-positive (B. subtilis YB88675) and Gram-negative strains (E. coli DC276, 77 and C.
crescentus NA100078). In terms of conservation of the HKs we assayed for biochemical inhibition among these three bacterial strains,
CckA and DivJ are only found in C. crescentus, and PhoQ is only found in E. coli. Because we did not know whether these inhibitors
would cross into the bacterial cytoplasm, where the ATP-binding domains of HKs reside, we used E. coli DC2, which is a strain
hypersensitive to antibiotics due to increased cell permeability.76, 77 Inhibitors were most potent against E. coli DC2, which we expected
based on the increased susceptibility to antibiotics this strain exhibits, and were least potent against C. crescentus, a Gram-negative
strain (Table 3). Inhibitor 5b, which effectively inhibited both CckA and PhoQ in vitro, was also effective at inhibiting bacterial growth
of each strain. Generally, the chlororesorcinol derivatives 5a–c and 7a–c were more potent than the ethylresorcinol 5e and resorcinol
(5d, 7d) derivatives. Comparing inhibitor 5b to our lead molecule CCT018159, we see that while 5b inhibits and stabilizes CckA better
than CCT018159 (Table 1), CCT018159 inhibits antimicrobial growth at lower concentrations for both C. crescentus and B. subtilis. Of
the Hsp90 inhibitors we originally screened (radicicol, NVP-AUY922, CCT018159, BIIB 021, PU H71, and CUDC-305), only the
diaryl pyrazole CCT018159 inhibited C. crescentus and B. subtilis growth at or below 300 µM, and radicicol, which Guarnieri
demonstrated bound weakly to Salmonella PhoQ HK1, 7 and which we found only weakly inhibited CckA, did not inhibit bacterial
growth. We note that CCT018159 demonstrated mild hemolysis effects at 0.5 mM and aggregated DivJ at 0.8 mM, suggesting that its
mechanism of action may be due in part to membrane disruption. Future experiments will explore the mechanisms of antimicrobial
action, including whether these inhibitors cause membrane disruption against bacteria, and whether they display bacteriostatic or
bactericidal effects.

The diaryl pyrazoles reported here demonstrate similar potency with recently described HK inhibitors (Figure 6). Recent drug
discovery efforts have utilized virtual screening (Liu,15 Lv,14 Velikova17), fragment-based screening (Velikova17), structure-based drug
design (Boibessot8), and high-throughput screening methods (Wilke19), and each method has yielded promising lead scaffolds for
antibiotics targeting HKs. Our method, which focuses on repurposing drug-like molecules that inhibit a conserved superfamily member,
demonstrates that lead molecules of comparable efficacy may be developed using rational design approaches. In biochemical assays for
either ATPase activity (this study, Liu,15 and Lv14) or phosphotransfer assays (this study, Boibessot,8 Velikova,17 and Wilke19), the IC50
values span roughly two orders of magnitude. For example Wilke 1119 has an IC50 of 1.21 µM against T. maritima HK853, and 5d has
an IC50 of 238 µM against S. typhimurium PhoQ (Figure 6). Compared to the best HK inhibitors described to date (Lv H2-8114 and Liu
H5-33,15 which have ATPase IC50 values of 24 µM against Staphylococcus epidermidis WalK and MIC values against Gram-positive S.
epidermidis of <2 µg mL–1), our inhibitors show similar biochemical activity but are less active at inhibiting bacterial growth. We have
not tested our inhibitors against WalK or S. epidermidis, so these differences may be target specific. Furthermore, only Wilke and co-
workers, who measured drug binding by displacement of a fluorescently labeled ATP-analog,19 and this study have demonstrated direct
competition of inhibitors with ATP; thus it remains possible that other reported HK inhibitors work by another mechanism that leads
them to be more potent in vivo. Because the biochemical assays use different ATP concentrations, and HKs have a wide range of Km
values for ATP (~ 2.4–300 µM, see Trajtenberg79), we note that direct comparisons between each study should be made with caution.
That said, two general trends appear. First, each inhibitor typically shows a range of potencies for HKs (Figure 6), suggesting one
pharmacophore may not be able to potently inhibit HKs universally. And second, the MIC values for Gram-negative bacteria typically
are higher than Gram-positive strains. For example, Boibessot 6d8 has an MIC that is 4.5× higher for E. coli than B. subtilis, while
Velikova and co-workers17 found that S1.13 did not affect E. coli growth at 500 µg mL–1 despite having an MIC of 8 µg mL–1 against S.
aureus (Figure 6). In this study, we found that 5b has a comparable MIC between C. crescentus and B. subtilis. These strain-related
differences may be attributed in part to membrane permeability: like Wilke and co-workers,19 we used E. coli DC2, a strain with a
compromised cell membrane,76, 77 and found higher potency of our diaryl pyrazoles with this strain compared to C. crescentus, which is
not hypersensitive to antibiotics. Recent work by Velikova and co-workers involving encapsulation of HK inhibitors in ε-poly-L-lysine
capped nanoparticles supports this hypothesis.23 For example, they demonstrated that capped nanoparticles harboring inhibitor Velikova
S1.13 had an MIC against E. coli CFT 073 of 25 µg mL–1, which is more than tenfold lower than Velikova S1.13, and these effects were
attributed to the ability of ε-poly-L-lysine capped nanoparticles to associate with the bacterial outer membrane, followed by release of
the HK inhibitor cargo inside the cell, thereby bypassing the protective outer membrane Gram-negative bacteria contain.23 In addition,
these effects may be explained by the fact that E. coli does not contain essential HKs,80, 81 whereas C. crescentus has four (CckA, CenK,
DivL, PetR),35, 37 and B. subtilis has one (YycG, which is also known as WalK or VicK).82, 83 Future experiments will focus on
distinguishing between these two potential modes of action.
Finally, we recognize that the diaryl pyrzoles, like many other HK inhibitors (Figure 6), inhibit growth only at relatively high
concentrations (≥150 µM), which suggests the antimicrobial effects may occur from either inhibiting multiple HKs, or by inhibiting off-
target enzymes or by damaging effects on the bacterial cell membrane. Given that 5b and 5c inhibit both CckA and PhoQ (Table 2) in
vitro, albeit with increased potency for CckA by an order of magnitude, the effects we see on C. crescentus may be due to inhibition of
multiple HKs. An outstanding question is whether these growth inhibition effects on C. crescentus are due to CckA inhibition, and we
are currently developing experiments to test this directly. At this time, we cannot rule out that antimicrobial activity occurs from off-
target inhibition or from membrane damage.
In summary we report that repurposing diaryl pyrazole-based, ATP-competitive Hsp90 inhibitors as antibiotics targeting HKs is a
promising strategy for the development of novel antibacterial agents. We synthesized and evaluated derivatives of CCT018159 and
discovered that the presence of a chlororesorcinol ring was the most important structural feature for increasing potency with this series.
Inhibitor 5b demonstrated favorable properties both for biochemical inhibition of CckA, an essential HK in C. crescentus, and PhoQ, a
HK in Salmonella that is essential for virulence, as well as inhibiting growth of three bacterial strains. Interestingly, despite the
conservation of the Bergerat fold ATP-binding domain between Hsp90 and HKs, we discovered that not all Hsp90 inhibitors are
effective HK inhibitors. This series represents a promising start toward the development of antibiotics that disrupt bacterial signaling
pathways. Because these diaryl pyrazole inhibitors also inhibit Hsp90, further development of this series will focus on both increasing
the antimicrobial activity and increasing HK inhibition while decreasing Hsp90 inhibition.
Acknowledgements
Selected structural analyses were performed with the UCSF Chimera package, which is developed by the Resource for
Biocomputing, Visualization, and Informatics at the University of California, San Francisco (supported by NIGMS P41-GM103311).
This work was supported in part by Williams College and The Hellman Fellows Fund (J.A.B.). We thank our colleagues Dr. Becky
Taurog for her technical support with protein purification and reading of this manuscript, Dr. Nathan Cook for his assistance with NMR
and MS analyses, and Prof. Chip Lovett for the Bacillus subtilis YB886 strain. We thank Prof. W. Seth Childers, University of
Pittsburgh, and members of his lab for the pAP433 plasmid, experimental advice, and thoughtful discussions on this project. We thank
Prof. Rui Zhao, University of Colorado, Denver, for plasmid phoQcyt; Prof. Patrick Loll, Drexel University, for vector pETHSUL; and
Prof. Gregory Bowman, Washington University School of Medicine, for discussions on molecular docking.

Table 3. Hemolysis data against sheep red blood cells and antimicrobial activities against three bacterial strains.
Compound
MIC values / µM (µg ml–1)b
Hemolysisa
E. coli
C. crescentus
B. subtilis
Mean % hemolysis ± SD
DC276, 77
NA100078
YB88675
Radicicol
0 ± 1
1 ± 1
14 ± 10
6 ± 8
1 ± 1
3 ± 1
1 ± 1
3 ± 2
2 ± 2
2 ± 2
15 ± 2
0 ± 1
0 ± 1
1 ± 1
3 ± 2
>300 (>109)
37.5–75 (17–35)
37.5–75 (13–26)
>300 (>95.6)
75–225 (38–115)
4.7–75 (2–33)
37.5–75 (11–23)
37.5–75 (12–25)
75 (27)
>300 (>109)
>300 (>140)
150–225 (53–79)
>300 (>96)
>300 (>154)
>300 (>133)
225 (68)
>300 (>109)
>300 (>140)
112.5 (40)
NVP-AUY922
CCT018159
BIIB 021
>300 (>96)
>300 (>154)
>300 (>133)
150 (45)
PU H71
CUDC-305
5a
5b
5c
5d
5e
7a
7b
7c
7d
225 (74)
150–225 (50–74)
112.5–150 (40–54)
>300 (>89)
112.5 (36)
225 (81)
75–150 (22–44)
75 (24)
>300 (>89)
225 (73)
37.5 (11)
150–225 (43–65)
225 (71)
225 (65)
37.5 (12)
150 (48)
37.5–75 (13–26)
37.5–112.5 (11–32)
150–225 (52–78)
>300 (>85)
112.5–225 (39–78)
>300 (>85)
a Mean relative hemolysis of citrated sheep red blood cells at 0.5 M compound, 10% (v/v) DMSO, reported as a percentage of the
positive control Triton X-100; n = 4; n = 8 for the controls; SD = standard deviation.
b Minimum inhibitory concentration measured using the broth-dilution method; (MIC in µg mL–1 units, calculated from MIC in
µM and MW of each compound); n = 2–3; ranges indicate the MIC values measured in replicate experiments; concentrations of 300
µM were the highest measured.

Figure 6. Comparison of 5b to previously reported HK inhibitors. Structures of HK inhibitors with selected biochemical and biological data from earlier
studies are shown, and compound names include the lead author surname and the compound name from each study (Boibessot8, Lv14, Liu15, Velikova17, Wilke19).
G+ = Gram positive strain; G– = Gram negative strain. MIC values in each study were reported either in units of µg mL–1 or µM. For ease of comparison, we
calculated each corresponding value using the compound MW, and we denoted (*) these calculated values.
References
1.
Guarnieri MT, Zhang L, Shen J, Zhao R. The Hsp90 inhibitor radicicol interacts with the ATP-binding pocket of bacterial sensor kinase
PhoQ. J Mol Biol. 2008;379(1): 82-93.
2.
3.
Silver LL. Challenges of antibacterial discovery. Clin Microbiol Rev. 2011;24(1): 71-109.
Meng EC, Pettersen EF, Couch GS, Huang CC, Ferrin TE. Tools for integrated sequence-structure analysis with UCSF Chimera. BMC
Bioinformatics. 2006;7(1): 339.
4.
Pettersen EF, Goddard TD, Huang CC, et al. UCSF Chimera--a visualization system for exploratory research and analysis. J Comput
Chem. 2004;25(13): 1605-1612.
5.
Dubey BN, Lori C, Ozaki S, et al. Cyclic di-GMP mediates a histidine kinase/phosphatase switch by noncovalent domain cross-linking.
Sci Adv. 2016;2(9): e1600823-e1600823.
6.
Marina A, Mott C, Auyzenberg A, Hendrickson WA, Waldburger CD. Structural and mutational analysis of the PhoQ histidine kinase
catalytic domain. Insight into the reaction mechanism. J Biol Chem. 2001;276(44): 41182-41190.
7.
Guarnieri MT, Blagg BSJ, Zhao R. A High-Throughput TNP-ATP Displacement Assay for Screening Inhibitors of ATP-Binding in
Bacterial Histidine Kinases. Assay Drug Dev Technol. 2011;9(2): 174-183.
8.
Boibessot T, Zschiedrich CP, Lebeau A, et al. The Rational Design, Synthesis, and Antimicrobial Properties of Thiophene Derivatives
That Inhibit Bacterial Histidine Kinases. J Med Chem. 2016;59(19): 8830-8847.
9.
Bruncko M, Tahir SK, Song X, et al. N-aryl-benzimidazolones as novel small molecule HSP90 inhibitors. Bioorg Med Chem Lett.
2010;20(24): 7503-7506.

10.
Laxminarayan R, Duse A, Wattal C, et al. Antibiotic resistance-the need for global solutions. Lancet Infect Dis. 2013;13(12): 1057-
1098.
11.
Prodromou C, Roe SM, O'Brien R, Ladbury JE, Piper PW, Pearl LH. Identification and structural characterization of the ATP/ADP-
binding site in the Hsp90 molecular chaperone. Cell. 1997;90(1): 65-75.
12.
13.
Wilke KE, Carlson EE. All signals lost. Sci Transl Med. 2013;5(203): 203ps212-203ps212.
Bem AE, Velikova N, Pellicer MT, Baarlen Pv, Marina A, Wells JM. Bacterial histidine kinases as novel antibacterial drug targets. ACS
Chem Biol. 2015;10(1): 213-224.
14.
Lv Z, Zhao D, Chang J, et al. Anti-bacterial and Anti-biofilm Evaluation of Thiazolopyrimidinone Derivatives Targeting the Histidine
Kinase YycG Protein of Staphylococcus epidermidis. Front Microbiol. 2017;8: 6293.
15.
Liu H, Zhao D, Chang J, et al. Efficacy of novel antibacterial compounds targeting histidine kinase YycG protein. Appl Microbiol
Biotechnol. 2014;98(13): 6003-6013.
16.
17.
Stock AM, Robinson VL, Goudreau PN. Two-component signal transduction. Annu Rev Biochem. 2000;69: 183-215.
Velikova N, Fulle S, Manso AS, et al. Putative histidine kinase inhibitors with antibacterial effect against multi-drug resistant clinical
isolates identified by in vitro and in silico screens. Sci Rep. 2016;6: 26085.
18.
Gotoh Y, Eguchi Y, Watanabe T, Okamoto S, Doi A, Utsumi R. Two-component signal transduction as potential drug targets in
pathogenic bacteria. Curr Opin Microbiol. 2010;13(2): 232-239.
19.
Wilke KE, Francis S, Carlson EE. Inactivation of multiple bacterial histidine kinases by targeting the ATP-binding domain. ACS Chem
Biol. 2015;10(1): 328-335.
20.
Martin PK, Li T, Sun D, Biek DP, Schmid MB. Role in cell permeability of an essential two-component system in Staphylococcus
aureus. J Bacteriol. 1999;181(12): 3666-3673.
21.
Jacobs C, Domian IJ, Maddock JR, Shapiro L. Cell cycle-dependent polar localization of an essential bacterial histidine kinase that
controls DNA replication and cell division. Cell. 1999;97(1): 111-120.
22.
23.
Gao R, Stock AM. Biological insights from structures of two-component proteins. Annu Rev Microbiol. 2009;63(1): 133-154.
Velikova N, Mas N, Miguel-Romero L, et al. Broadening the antibacterial spectrum of histidine kinase autophosphorylation
inhibitors via the use of ε-poly-L-lysine capped mesoporous silica-based nanoparticles. Nanomedicine. 2017;13(2): 569-581.
24.
Dutta R, Inouye M. GHKL, an emergent ATPase/kinase superfamily. Trends Biochem Sci. 2000;25(1): 24-28.
25.
Bhat R, Tummalapalli SR, Rotella DP. Progress in the discovery and development of heat shock protein 90 (Hsp90) inhibitors. J Med
Chem. 2014;57(21): 8718-8728.
26.
Shrestha L, Bolaender A, Patel HJ, Taldone T. Heat Shock Protein (HSP) Drug Discovery and Development: Targeting Heat Shock
Proteins in Disease. Curr Top Med Chem. 2016;16(25): 2753-2764.
27.
Wang M, Shen A, Zhang C, et al. Development of Heat Shock Protein (Hsp90) Inhibitors To Combat Resistance to Tyrosine Kinase
Inhibitors through Hsp90-Kinase Interactions. J Med Chem. 2016: acs.jmedchem.5b01106.
28.
Brough PA, Aherne W, Barril X, et al. 4,5-diarylisoxazole Hsp90 chaperone inhibitors: potential therapeutic agents for the treatment
of cancer. J Med Chem. 2008;51(2): 196-218.
29.
Cheung K-MJ, Matthews TP, James K, et al. The identification, synthesis, protein crystal structure and in vitro biochemical
evaluation of a new 3,4-diarylpyrazole class of Hsp90 inhibitors. Bioorg Med Chem Lett. 2005;15(14): 3338-3343.
30.
Kasibhatla SR, Hong K, Biamonte MA, et al. Rationally designed high-affinity 2-amino-6-halopurine heat shock protein 90 inhibitors
that exhibit potent antitumor activity. J Med Chem. 2007;50(12): 2767-2778.
31.
He H, Zatorska D, Kim J, et al. Identification of potent water soluble purine-scaffold inhibitors of the heat shock protein 90. J Med
Chem. 2006;49(1): 381-390.

32.
Bao R, Lai C-J, Qu H, et al. CUDC-305, a novel synthetic HSP90 inhibitor with unique pharmacologic properties for cancer therapy.
Clin Cancer Res. 2009;15(12): 4046-4057.
33.
34.
McAdams HH, Shapiro L. The architecture and conservation pattern of whole-cell control circuitry. J Mol Biol. 2011;409(1): 28-35.
Purcell EB, Boutte CC, Crosson S. Two-component signaling systems and cell cycle control in Caulobacter crescentus. Adv Exp Med
Biol. 2008;631: 122-130.
35.
Christen B, Abeliuk E, Collier JM, et al. The essential genome of a bacterium. Mol Syst Biol. 2011;7: 528.
36.
Jacobs C, Ausmees N, Cordwell SJ, Shapiro L, Laub MT. Functions of the CckA histidine kinase in Caulobacter cell cycle control. Mol
Microbiol. 2003;47(5): 1279-1290.
37.
Skerker JM, Prasol MS, Perchuk BS, Biondi EG, Laub MT. Two-Component Signal Transduction Pathways Regulating Growth and Cell
Cycle Progression in a Bacterium: A System-Level Analysis. PLoS Biol. 2005;3(10): e334.
38.
Brilli M, Fondi M, Fani R, et al. The diversity and evolution of cell cycle regulation in alpha-proteobacteria: a comparative genomic
analysis. BMC Syst Biol. 2010;4: 52.
39.
Kim J, Heindl JE, Fuqua C. Coordination of division and development influences complex multicellular behavior in Agrobacterium
tumefaciens. PLoS ONE. 2013;8(2): e56682.
40.
Willett JW, Herrou J, Briegel A, Rotskoff G, Crosson S. Structural asymmetry in a conserved signaling system that regulates division,
replication, and virulence of an intracellular pathogen. Proc Natl Acad Sci U S A. 2015;112(28): E3709-3718.
41.
Ericsson UB, Hallberg BM, DeTitta GT, Dekker N, Nordlund P. Thermofluor-based high-throughput stability optimization of proteins
for structural studies. Anal Biochem. 2006;357(2): 289-298.
42.
Niesen FH, Berglund H, Vedadi M. The use of differential scanning fluorimetry to detect ligand interactions that promote protein
stability. Nat Protoc. 2007;2(9): 2212-2221.
43.
Pantoliano MW, Petrella EC, Kwasnoski JD, et al. High-density miniaturized thermal shift assays as a general strategy for drug
discovery. J Biomol Screen. 2001;6(6): 429-440.
44.
Boivin S, Kozak S, Meijers R. Optimization of protein purification and characterization using Thermofluor screens. Protein Expr Purif.
2013;91(2): 192-206.
45.
Ji Q, Chen PJ, Qin G, et al. Structure and mechanism of the essential two-component signal-transduction system WalKR in
Staphylococcus aureus. Nature Commun. 2016;7: 11000.
46.
Monk IR, Howden BP, Seemann T, Stinear TP. Correspondence: Spontaneous secondary mutations confound analysis of the
essential two-component system WalKR in Staphylococcus aureus. Nature Commun. 2017;8: 14403.
47.
McKellar JLO, Minnell JJ, Gerth ML. A high-throughput screen for ligand binding reveals the specificities of three amino acid
chemoreceptors from Pseudomonas syringae pv. actinidiae. Mol Microbiol. 2015;96(4): 694-707.
48.
Lori C, Ozaki S, Steiner S, et al. Cyclic di-GMP acts as a cell cycle oscillator to drive chromosome replication. Nature. 2015;523(7559):
236-239.
49.
Mann TH, Seth Childers W, Blair JA, Eckart MR, Shapiro L. A cell cycle kinase with tandem sensory PAS domains integrates cell fate
cues. Nature Commun. 2016;7: 11454.
50.
Chen YE, Tsokos CG, Biondi EG, Perchuk BS, Laub MT. Dynamics of two Phosphorelays controlling cell cycle progression in
Caulobacter crescentus. J Bacteriol. 2009;191(24): 7417-7429.
51.
Blair JA, Xu Q, Childers WS, et al. Branched signal wiring of an essential bacterial cell-cycle phosphotransfer protein. Structure.
2013;21(9): 1590-1601.
52.
Kiianitsa K, Solinger JA, Heyer WD. NADH-coupled microplate photometric assay for kinetic studies of ATP-hydrolyzing enzymes with
low and high specific activities. Anal Biochem. 2003;321(2): 266-271.

53.
2007.
Drysdale MJ, Dymock BW, Barril-Alonso X, et al. 3,4-diarylpyrazoles and their use in the therapy of cancer. U.S. Patent 7 247 734;
Dymock BW, Barril X, Brough PA, et al. Novel, potent small-molecule inhibitors of the molecular chaperone Hsp90 discovered
54.
through structure-based design. J Med Chem. 2005;48(13): 4212-4215.
55.
Sharp SY, Boxall K, Rowlands M, et al. In vitro biological characterization of a novel, synthetic diaryl pyrazole resorcinol class of heat
shock protein 90 inhibitors. Cancer Res. 2007;67(5): 2206-2216.
56.
57.
58.
59.
Howes R, Barril X, Dymock BW, et al. A fluorescence polarization assay for inhibitors of Hsp90. Anal Biochem. 2006;350(2): 202-213.
Balasubramanian S, Nair MG. An Efficient “One Pot” Synthesis of Isoflavones. Synth Commun. 2000;30(3): 469-484.
Groisman EA. The pleiotropic two-component regulatory system PhoP-PhoQ. J Bacteriol. 2001;183(6): 1835-1842.
Miller SI, Kukral AM, Mekalanos JJ. A two-component regulatory system (phoP phoQ) controls Salmonella typhimurium virulence.
Proc Natl Acad Sci USA. 1989;86(13): 5054-5058.
60.
Curtis PD, Brun YV. Getting in the loop: regulation of development in Caulobacter crescentus. Microbiol Mol Biol Rev. 2010;74(1):
13-41.
61.
Ohta N, Lane T, Ninfa EG, Sommer JM, Newton A. A histidine protein kinase homologue required for regulation of bacterial cell
division and differentiation. Proc Natl Acad Sci USA. 1992;89(21): 10297-10301.
62.
63.
Grebe TW, Stock JB. The histidine protein kinase superfamily. Adv Microb Physiol. 1999;41: 139-227.
Francis S, Wilke KE, Brown DE, Carlson EE. Mechanistic insight into inhibition of two-component system signaling. MedChemComm.
2013;4(1): 269-277.
64.
Bilwes AM, Quezada CM, Croal LR, Crane BR, Simon MI. Nucleotide binding by the histidine kinase CheA. Nat Struct Biol. 2001;8(4):
353-360.
65.
Perez AM, Mann TH, Lasker K, Ahrens DG, Eckart MR, Shapiro L. A Localized Complex of Two Protein Oligomers Controls the
Orientation of Cell Polarity. mBio. 2017;8(1): e02238-02216.
66.
Weeks SD, Drinker M, Loll PJ. Ligation independent cloning vectors for expression of SUMO fusions. Protein Expr Purif. 2007;53(1):
40-50.
67.
68.
Shoichet BK. Interpreting steep dose-response curves in early inhibitor discovery. J Med Chem. 2006;49(25): 7274-7277.
Stephenson K, Yamaguchi Y, Hoch JA. The mechanism of action of inhibitors of bacterial two-component signal transduction
systems. J Biol Chem. 2000;275(49): 38900-38904.
69.
McGovern SL, Caselli E, Grigorieff N, Shoichet BK. A common mechanism underlying promiscuous inhibitors from virtual and high-
throughput screening. J Med Chem. 2002;45(8): 1712-1722.
70.
Hilliard JJ, Goldschmidt RM, Licata L, Baum EZ, Bush K. Multiple mechanisms of action for inhibitors of histidine protein kinases from
bacterial two-component systems. Antimicrob Agents Chemother. 1999;43(7): 1693-1699.
71.
72.
73.
74.
Copeland RA. Enzymes : a practical introduction to structure, mechanism, and data analysis. 2nd ed. New York: Wiley; 2000.
OEDOCKING. 3.2.0.2 ed. Santa Fe, NM: OpenEye Scientific Software http://www.eyesopen.com/.
McGann M. FRED and HYBRID docking performance on standardized datasets. J Comput Aided Mol Des. 2012;26(8): 897-906.
Wiegand I, Hilpert K, Hancock REW. Agar and broth dilution methods to determine the minimal inhibitory concentration (MIC) of
antimicrobial substances. Nat Protoc. 2008;3(2): 163-175.
75.
Yasbin RE, Fields PI, Andersen BJ. Properties of Bacillus subtilis 168 derivatives freed of their natural prophages. Gene. 1980;12(1-2):
155-159.

76.
Clark D. Novel Antibiotic Hypersensitive Mutants of Escherichia-Coli Genetic-Mapping and Chemical Characterization. FEMS
Microbiol Lett. 1984;21(2): 189-195.
77.
Clark DP. Mutant of Escherichia coli deficient in osmoregulation of periplasmic oligosaccharide synthesis. J Bacteriol. 1985;161(3):
1049-1053.
78.
Marks ME, Castro-Rojas CM, Teiling C, et al. The genetic basis of laboratory adaptation in Caulobacter crescentus. J Bacteriol.
2010;192(14): 3678-3688.
79.
Trajtenberg F, Graña M, Ruétalo N, Botti H, Buschiazzo A. Structural and enzymatic insights into the ATP binding and
autophosphorylation mechanism of a sensor histidine kinase. J Biol Chem. 2010;285(32): 24892-24903.
80.
Gerdes SY, Scholle MD, Campbell JW, et al. Experimental determination and system level analysis of essential genes in Escherichia
coli MG1655. J Bacteriol. 2003;185(19): 5673-5684.
81.
Yamamoto K, Hirao K, Oshima T, Aiba H, Utsumi R, Ishihama A. Functional characterization in vitro of all two-component signal
transduction systems from Escherichia coli. J Biol Chem. 2005;280(2): 1448-1456.
82.
Fabret C, Hoch JA. A two-component signal transduction system essential for growth of Bacillus subtilis: implications for anti-
infective therapy. J Bacteriol. 1998;180(23): 6375-6383.
83.
Kobayashi K, Ehrlich SD, Albertini A, et al. Essential Bacillus subtilis genes. Proc Natl Acad Sci USA. 2003;100(8): 4678-4683.