Identification of Small-Molecule Modulators of Diguanylate Cyclase by FRET-Based High-Throughput Screening

DOI: 10.1002/cbic.201800593

Full Papers

Identification of Small-Molecule Modulators of

Diguanylate Cyclase by FRET-Based High-Throughput

Screening

Matthias Christen,*^[a]Cassandra Kamischke,^[b]Hemantha D. Kulasekara,^[b]Kathleen C. Olivas,^[c]

Bridget R. Kulasekara,^[d]Beat Christen,^[e]Toni Kline,^[f]and Samuel I. Miller^[b]

The bacterial second messenger cyclic diguanosine monophos-

phate (c-di-GMP) is a key regulator of cellular motility, the cell

cycle, and biofilm formation with its resultant antibiotic toler-

ance, which can make chronic infections difficult to treat.

Therefore, diguanylate cyclases, which regulate the spatiotem-

poral production of c-di-GMP, might be attractive drug targets

for control of biofilm formation that is part of chronic infec-

tions. We present a FRET-based biochemical high-throughput

screening approach coupled with detailed structure–activity

studies to identify synthetic small-molecule modulators of the

diguanylate cyclase DgcA from Caulobacter crescentus. We

identified a set of seven small molecules that regulate DgcA

enzymatic activity in the low-micromolar range. Subsequent

structure–activity studies on selected scaffolds revealed a re-

markable diversity of modulatory behavior, including slight

chemical substitutions that reverse the effects from allosteric

enzyme inhibition to activation. The compounds identified rep-

resent new chemotypes and are potentially developable into

chemical genetic tools for the dissection of c-di-GMP signaling

networks and alteration of c-di-GMP-associated phenotypes. In

sum, our studies underline the importance of detailed mecha-

nism-of-action studies for inhibitors of c-di-GMP signaling and

demonstrate the complex interplay between synthetic small

molecules and the regulatory mechanisms that control the ac-

tivity of diguanylate cyclases.

Introduction

The second messenger cyclic dimeric guanosine monophos-

phate (c-di-GMP) mediates diverse bacterial cellular processes

including antibiotic resistance, biofilm formation, extracellular

carbohydrate and adhesin production, pilus- and flagellum-

based motility, and cell cycle progression.^[1–5]Signal integration

into c-di-GMP networks is, in part, controlled by diguanylate

cyclases (DGCs) that convert two molecules of GTP into

c-di-GMP. The enzymatic activity of DGCs resides within a con-

served domain consisting of the amino acid sequence glycine-

glycine-aspartate-glutamate-phenylalanine (GGDEF) that forms

the enzymatic active site. The GGDEF domain shares similarity

to the PALM 4 domain found in other classes of nucleotide cy-

clases.^[6–10]The apparent role of c-di-GMP in the cell cycle and

the existence of many paralogous DGC enzymes controlling di-

verse cellular functions indicate that there is likely tight spatial

and temporal regulation of c-di-GMP.^[10–12]Bacterial genomes

encode multiple GGDEF domains in proteins with signal-sens-

ing domains.^[13]However, the presence of multiple paralogues

makes it difficult to study the signaling processing properties

of c-di-GMP signaling networks by conventional genetic tech-

niques. Therefore, chemical genetic approaches, inactivating

each segment of the c-di-GMP signaling network, might be an

attractive approach to studying the overall biological function

of c-di-GMP. In pathogenic bacteria, cellular production of

c-di-GMP is essential to maintain biofilm formation, especially

under stressed conditions such as induction by aminoglycoside

antibiotics.^[1]Small molecules that effectively inhibit DGC activi-

ty have the potential to prevent biofilm formation, thus

making DGCs interesting targets for the development of new

classes of antimicrobial agents.

[a] Dr. M. Christen

Eidgençssische Technische Hochschule Zꢀrich, Department of Biology

Auguste-Piccard-Hof 1, 8093 Zꢀrich (Switzerland)

E-mail: matthias.christen@imsb.biol.ethz.ch

[b] C. Kamischke, Dr. H. D. Kulasekara, Prof. Dr. S. I. Miller

University of Washington, Department of Microbiology

1959 NE Pacific St., Box 357710, Seattle, WA 98195 (USA)

[c] Dr. K. C. Olivas

Seattle Genetics, Inc.

21823 30th Drive SE, Bothell, WA 98021 (USA)

[d] Dr. B. R. Kulasekara

University of Washington, Department of Genome Sciences

1959 NE Pacific St., Box 357710, Seattle, WA 98195 (USA)

[e] Prof. Dr. B. Christen

Eidgençssische Technische Hochschule Zꢀrich, Department of Biology

Otto-Stern-Weg-3, 8093 Zꢀrich (Switzerland)

We developed a sensitive and robust FRET-based DGC activi-

ty assay and performed high-throughput (HT) screening on a

comprehensive compound library to evaluate 27502 small

molecules for inhibition of the Caulobacter crescentus DGC

DgcA (CC3285). DgcA has been extensively characterized and

therefore serves as a model enzyme for study of c-di-GMP-

[f] Dr. T. Kline

Sutro Biopharma

310 Utah Avenue, Suite 150, San Francisco, CA 94080 (USA)

Supporting information and the ORCID identification numbers for the

authors of this article can be found under https://doi.org/10.1002/

cbic.201800593.

ChemBioChem 2018, 19, 1 – 15

1

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

&

These are not the final page numbers! ÞÞ

Full Papers

related phenotypic effects.^[9,11,12,14]Similarly to the majority of

DGC enzymes, DgcA is subject to high-affinity binding of c-di-

GMP to an allosteric site (I site), which efficiently blocks enzy-

matic activity in a noncompetitive manner and is located dis-

tant from the catalytic pocket (A site). Mutational analysis of

DgcA has provided convincing evidence that c-di-GMP binding

to several conserved charged amino acids at the I site is a key

mechanism for allosteric regulation of DGCs.^[9]

Results

Development of a FRET-based biochemical high-throughput

screen for monitoring c-di-GMP production

To identify small-molecule inhibitors of DgcA, we established a

sensitive FRET-based activity assay. We have previously report-

ed a FRET-based c-di-GMP biosensor, which we refer to here as

the biosensor.^[10]This biosensor consists of the c-di-GMP bind-

ing domain of the PilZ protein YcgR, C- and N-terminally

tagged with mCYPet and mYPet fluorescent proteins. Binding

of c-di-GMP to the YcgR domain induces a conformational

change that alters the relative orientation of the external fluo-

rescent subunit protein reporters, leading to a decrease in

FRET efficiency. Thus, the fluorescence properties (535/470 nm

emission ratio) of the biosensor reflect c-di-GMP levels. The

biosensor exhibits a change in net FRET (nFRET) of ꢀ60.6%,

with a binding constant of 198 nm for c-di-GMP and no detect-

able response to cyclic adenosine 3,5-monophosphate, cyclic

GMP, or guanosine 5-triphosphate.^[10]

Figure 1. FRET assay for c-di-GMP. A) Kinetics of fluorescence emission ratio

change (527/480 nm) measured in a 384-well plate format. Affinity-purified

YcgR FRET biosensor was added in the presence of 20 nm DgcA enzyme

and 20 mm GTP substrate (closed circles) or in the absence of GTP (open cir-

cles). B) Corresponding increase in c-di-GMP concentration derived from the

change in fluorescence emission ration (535/470 nm). Above 800 nm c-di-

GMP, allosteric product inhibition decreases the DGC activity of DgcA. Each

graph shows the average of three independent measurements. The reaction

volume per well was 20 mL in a 384-well Corning low-volume flat-bottomed

plate. C) Histogram of the fluorescence emission ratio after 3 h incubation of

20 nm DgcA with 20 mm GTP in the presence of 50 mgmL^ꢀ1compounds ( )

&

and 1% DMSO. Wells without added GTP substrate were used as positive

To measure the suitability and performance of the biosensor

for HT-capable DGC assay, we recorded kinetics of fluorescence

emission ratio change in 384-well plates in a 20 mL reaction

volume with 20 nm DgcA enzyme and 50 nm biosensor and

followed the change in fluorescence at intervals of 1 min (Fig-

ure 1A). Addition of 50 nm DgcA in the absence of GTP does

&

controls ( ); wells with exogenous c-di-GMP (5 mm) were used as negative

&

controls ( ). D) Plate uniformity of the HT screening. For every well, back-

ground fluorescence signal of the compound was measured prior to the ad-

dition of the FRET biosensor. Compounds with autofluorescence exceeding

7% of the FRET biosensor signal (535/470 nm) were discarded.

*

not alter FRET efficiency (Figure 1A, ) whereas injection of

maintained at the Institute of Chemical Biology (ICCB) at Har-

vard University, Boston, MA, funded by the Regional Centers of

Excellence in BioDefense and Emerging Infectious Disease

(NSRB/NERCE). During screening, we first preincubated the

DgcA enzyme with 200 nL compound (5 mgmL^ꢀ1) in DMSO

and measured fluorescence emission ratio before and after

addition of the biosensor to monitor fluorescence of chemical

compounds and FRET at 470 and 535 nm. This second mea-

surement will also detect compounds that mimic c-di-GMP

agonists. After addition of 20 mm GTP, FRET was measured 3 h

after incubation as an endpoint measurement (Figure 1C, D;

see also the Experimental Section). For a compound to be con-

sidered a hit, the following criteria had to be met: 1) com-

pounds with c-di-GMP production below 300 nm after 3 h of

incubation with DgcA and 20 mm GTP were considered strong

hits, 2) duplicates were consistent with a difference in final c-

di-GMP levels <100 nm, and 3) background fluorescence prior

to addition of the FRET biosensor had not to exceed 7% of the

FRET signal in CFP and YFP emission channels. We used total

fluorescence as a quality control criterion to identify inherently

fluorescent compounds. Of the 27502 compounds screened in

duplicate, 1029 (3.7%) were removed from the analysis due to

20 mm GTP induces a rapid change in FRET efficiency (Fig-

ure 1A, ). From the FRET ratio for the free and the c-di-GMP-

*

saturated biosensor, we determined the corresponding in-

crease in c-di-GMP (Figure 1B). Our sensitive and robust FRET

assay detects c-di-GMP production in the nm range and per-

mits determination of initial rate kinetics at levels below the

occurrence of allosteric c-di-GMP inhibition.

FRET-based HT screening and discovery of primary hits for

diguanylate cyclase inhibition

A major challenge of biochemical HT screening is to achieve

significance on large-scale experimental datasets for quality

control and hit selection. Through prior screening, we deter-

mined plate uniformity, signal variability, and repeatability

assessment assays for coefficient of variation (CV) and Z factor

calculation. The average Z factor for three plates read on two

consecutive days was 0.71, and the CV readings for high,

medium, and low signal were well below 5%. The chemical

library screened encompasses 27505 commercially available

compounds derived from small-molecule chemical libraries

&

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

2

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ÝÝ These are not the final page numbers!

Full Papers

autofluorescence. We identified 49 compounds (0.18%) as

strong hits, with c-di-GMP levels below 300 nm after 3 h incu-

bation with DgcA. An additional 241 compounds (0.87%) were

classified as moderate inhibitors, with c-di-GMP levels below

400 nm after incubation with DgcA.

for DgcA (Figure 2C), with 31.5% of DGC activity remaining in

the presence of an enzyme-saturating inhibitor concentration.

This indicates that upon binding of 2 the DgcA enzyme is con-

verted into a modified but still partially functional enzyme·sub-

strate·inhibitor complex, thus raising the possibility that 2

binds to a site distinct from the substrate binding site. For sub-

sequent structure–activity studies, we chemically synthesized a

set of 24 derivatives of compound 1 and 16 compounds with

structural similarities to scaffold 2 (Experimental Section).

Selection of compound scaffolds for secondary assays and

structure–activity relationship (SAR) studies

In selecting compounds for SAR studies, we placed a priority

on compounds likely to be functioning through our target

mechanism of action. We rejected cytotoxic or cytostatic com-

pounds and favored those compounds that exhibited synthetic

tractability and suitability for chemotype expansion. By these

criteria, we selected 18 compounds out of 49 initial hits for

confirmatory and secondary assays. We found that 38.9%

(seven out of 18) had IC₅₀values smaller than 50 mm, whereas

22.2% (four out of 18) also had IC₅₀values smaller than 10 mm

(Figure 2A). The results from our secondary assays led us to

focus further structure–activity studies on the two scaffolds [2-

oxo-2-(2-oxopyrrolidin-1-yl)ethyl] 1,3-benzothiazole-6-carboxyl-

ate (1) and 4-(2,5-dimethylphenoxy)-N-(4-morpholin-4-ylphe-

nyl)butanamide (2), with IC₅₀values of 4.0 and 6.4 mm, respec-

tively (Figure 2B and C). The concentration–response curve of

compound 1 revealed complete inhibition of DgcA activity

(Figure 2B). In contrast, compound 2 acts as a partial inhibitor

Characterization of SAR of scaffold 1

Structure–activity studies can be used to probe steric con-

straints imposed in the binding site and to reveal the molecu-

lar relationship between biological activity and structural prop-

erties of a ligand. Scaffold 1 contains a 2-oxo-2-(2-oxopyrroli-

din-1-yl)ethyl component linked to a 1,3-benzothiazole-6-car-

bonyloxy unit (Figure 2A). We individually probed the 2-oxo-

pyrrolidin-1-yl ring, the linker region, and the nucleobase-like

1,3-benzothiazole-6-carbonyloxy moiety for steric demand and

biological activities. We tested the binding affinities of ana-

logues of 1 by replacing the 2-oxopyrrolidin-1-yl ring by sub-

stituents R¹(Table 1) and keeping the 2-oxoethyl linker and the

1,3-benzothiazole-6-carbonyloxy motif constant. Our selection

of derivatives emphasized the pyrrolidin-1-yl group (as in 1a),

the 2-methylpropanamido group (as in 1b), the cyclohexane-

carboxamido group (as in 1c), the benzamido group (as in

1d), the 4-methylanilino group (as in 1e) and the 2-methoxy-

anilino group (as in 1 f) replacing the 2-oxopyrrolidin-1-yl ring.

We selected these residues for their steric demand and to

probe for hydrogen bonding, hydrophobic interactions, and p

stacking contacts.

The inhibitory potencies of the six analogues featuring 2-ox-

opyrrolidin-1-yl ring replacement are shown in Table 1. Many

of the derivatives retained, but did not exceed, a level of po-

tency comparable with that of 1. For example, the replacement

of the 2-oxopyrrolidin-1-yl moiety with a pyrrolidin-1-yl system

in 1a had only marginal effect on binding affinity, with a slight

increase in IC₅₀from 4 to 17.1 mm (Table 1). These results sug-

gest that the presence of an H-bond acceptor in the 2-oxopyr-

rolidin-1-yl component in 1 is not a prerequisite for mediating

binding affinity to DgcA. Substitution of the 2-oxopyrrolidin-1-

yl component with a 2-methylpropanamido system in 1b does

not affect binding affinity (IC₅₀of 3.6 versus 4.0 mm). However,

introduction of substituents with increased steric demand,

such as the cyclohexanecarboxamido motif in 1c or the benz-

amido group in 1d, in each case shows a tenfold reduction in

binding affinity, to 47.2 and 48.5 mm, respectively. To test

whether the decrease in binding affinity is caused by steric

hindrance at the binding site, we reduced the steric load and

introduced the 4- and 2-methoxyanilino analogues 1e and 1 f,

respectively. As expected, we observed increases in affinity, to

8.5 and 10.7 mm, respectively.

Figure 2. Inhibitors of the DGC enzyme DgcA. A) Structures of the small-

molecule inhibitors 1–7 with IC₅₀values below 50 mm identified in the

FRET screening. Out of these, the four compounds 1–4 exhibit IC₅₀values

<10 mm. Scaffolds 1 and 2 were selected for subsequent SAR studies. B) The

concentration–response curve of compound 1 (IC₅₀=4 mm) reveals complete

inhibition against DgcA. C) Compound 2 (IC₅₀=6.4 mm) acts as partial inhibi-

tor for DgcA. The corresponding concentration–response curve of 1 is

shown as a dashed line.

Thus, it seems that the 2-oxopyrrolidin-1-yl ring of scaffold 1

can be replaced by alternative substituents without dramatical-

ly altering binding affinity. However, although several substitu-

tions of the 2-oxopyrrolidin-1-yl ring did not affect binding

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

3

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

&

These are not the final page numbers! ÞÞ

Full Papers

Effects of benzothiazole substitutions in scaffold 1 on

binding affinity and modulation of DgcA

Table 1. Exploration of 2-oxopyrrolidin-1-yl ring substitutions in the 1

scaffold.

Next, we replaced the 1,3-benzothiazole-6-carbonyloxy moiety

on 1 and chemically synthesized 12 derivatives containing

alternative heterocycle and phenyl substituents (Experimental

Section). We introduced various heterocycles with both 5- and

6-substituted (benzothiazole numbering) carboxylato groups.

In addition, we altered the positions and types of heteroatoms

present in the heterocycles. For every derivative, we measured

dose–response curves and maximum inhibition values with the

FRET-based DGC assay. The inhibitory activity data and maxi-

mum inhibition values are summarized in Table 2. Whereas

compound 1i, in which the 1,3-benzothiazole-6-carbonyloxy

component was substituted with a 2-methyl-1,3-benzothia-

zole-5-carbonyloxy motif, exhibited decreased inhibitory effects

(IC₅₀=52.7 mm), we observed almost identical IC₅₀values in the

cases of the 1,3-benzoxazole-6-carbonyloxy derivative 1g and

the 2,3-dihydrobenzofuran 5-carbonyloxy derivative 1j, with

3.1 and 4.4 mm, respectively (Table 2). However, 1g and 1j

failed to block DGC activity effectively, and DgcA remained

partial active in the presence of the inhibitors (87.7% and

92.2%, respectively). These findings suggest that 1g and 1j

both act as silent allosteric modulators (SAMs) of DgcA

(Table 2). Furthermore, we found that substitution with the 1H-

indole-5-carbonyloxy group (as in 1k), the 1-methyl-1H-indole-

5-carbonyloxy group (as in 1l), the 1-methyl-1H-benzimida-

zole-5-carbonyloxy group (as in 1m), the isoquinoline-6-car-

bonyloxy group (as in 1n), and the quinoxaline-6-carbonyloxy

group (as in 1o) resulted in strongly reduced binding affinities

towards DgcA but maintained those analogues’ function as

negative modulators. Similarly, alteration of the carbonyloxy

side chain to a 1,3-benzothiazole-5-carbonyloxy unit in the

case of 1p had a dramatic effect and decreased the binding

affinity from 4 to 626 mm. These findings underline the isosteric

importance of the 1,3-benzothiazole-6-carbonyloxy component

in mediating the inhibitor potency of 1.

Cmpd R¹

IC₅₀ꢁSD [mm]^[a]Hill coeff.^[b]Activity ESI [%]^[c]

1

4.0ꢁ0.2

2.34ꢁ0.19

2.57ꢁ0.40

1.40ꢁ0.1

1a

1b

17.1ꢁ1.2

6.04ꢁ0.46

3.6ꢁ0.6

1.65ꢁ0.38 61.74ꢁ4.31

1c

1d

1e

47.2ꢁ8.4

1.46ꢁ0.26

9.01ꢁ0.78

48.5ꢁ3.8

8.5ꢁ0.7

2.89ꢁ0.32 14.64ꢁ0.74

2.93ꢁ0.48

3.09ꢁ0.25

1 f

10.7ꢁ2.5

1.48ꢁ0.13 59.42ꢁ1.61

[a] All experiments to determine IC₅₀values were run with at least dupli-

cates at each compound dilution. IC₅₀values were averaged when deter-

mined in two or more independent experiments. [b] Hill coefficients ob-

tained upon fitting the concentration–response curves. [c] DGC activities

of the enzyme·substrate·inhibitor complexes.

affinity, they had dramatic effects on the residual activities of

the enzyme·inhibitor·substrate complexes (Table 1). For exam-

ple, replacement of the 2-oxopyrrolidin-1-yl moiety of 1 (IC₅₀of

4 mm) with the 2-methylpropanamido system in 1b (IC₅₀of

3.6 mm) shifts the activity of the enzyme·inhibitor·substrate

complex from 1.4% to 61.7% (Table 1). Similar effects were ob-

served for the 4- and 2-methoxyanilino derivatives 1e and 1 f,

with IC₅₀values of 8.5 and 10.7 mm, respectively. Compound 1e

exhibits almost complete inhibition with only 3.1% DGC activi-

ty upon inhibitor saturation, whereas 1 f functions as a partial

inhibitor for DgcA, with 59.4% activity remaining in the en-

zyme·substrate·inhibitor complex. These results suggest that

1 f binding converts the DgcA enzyme into a modified but still

functional enzyme·substrate·inhibitor complex. Simultaneous

binding of substrate and inhibitor indicates that 1 and its de-

rivatives likely bind to a site distinct from the substrate binding

site of DgcA.

We next introduced benzoyloxy, 4-(dimethylamino)benzoy-

loxy, and 4-nitrobenzoyloxy moieties in 1h, 1q, and 1r, respec-

tively (Table 2). Surprisingly, whereas the introduction of these

substituents resulted in rather moderate binding affinities, we

observed that DgcA activity increased in the presence of these

compounds to activities higher than in their absence (118, 176,

and 156%, respectively), thus suggesting that these com-

pounds act as positive modulators of the DGC enzyme.

In the last group of derivatives, we investigated effects of

linker modifications. The parent compound 1 contains an ester

bond, which harbors the intrinsic risk of cleavage by cellular

esterases. A nonhydrolyzable amide analogue should provide

much better stability towards proteolysis by esterases. There-

fore, we synthesized the nonhydrolyzable amide derivative 1s

(Table S1 in the Supporting Information) and tested its inhibi-

tory properties against DgcA. We found that replacement of

the ester bond with an amide increased the IC₅₀from 4 to

31.6 mm and resulted in a partial inhibitor with 48% residual

activity of the enzyme·inhibitor·substrate complex. This sug-

gests that differences in hydrogen-bond-forming properties or

&

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

4

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ÝÝ These are not the final page numbers!

Full Papers

tween the 2-oxopyrrolidin-1-yl and the benzothiazole moieties

are critical determinants of binding affinity for DgcA.

Table 2. Substitution of the 1,3-benzothiazole component in scaffold 1

with alternative heterocycle and aryl substituents.

Although the parent compound 1 exhibits characteristics of

a complete inhibitor, we created a panel of derivatives of 1

with diverse effects on DgcA activity. Several analogues of the

scaffold 1 exhibit partial inhibition, whereas some act as silent

modulators or even function as positive modulators of DgcA.

Thus, it seems likely that, rather than binding in a competitive

mode, compounds of the scaffold 1 interact with an allosteric

site of DgcA. The fact that these structurally very similar com-

pounds exert diverse effects on the activity of the enzyme·inhi-

bitor·substrate complexes suggests a model in which an allo-

steric binding site of DgcA is highly responsive to slight con-

formational changes. Substitution, for example, of a single

sulfur atom in the benzothiazole moiety of 1 with oxygen in

1g preserves binding affinity (IC₅₀values of 4 and 3.1 mm, re-

spectively) but increases the activity of the enzyme·inhibitor

complex more than 60-fold from 1.4 to 87.7%. Similar effects

are observed with 1 and 1b: substitution of the 2-oxopyrroli-

din-1-yl system with 2-methylpropanamido group does not

affect IC₅₀values but increases the activity of the enzyme·inhi-

bitor complex from 1.4 to 62%.

Cmpd R²

IC₅₀AC₅₀ꢁSD

[mm]^[a]

Hill

Activity ESI

[%]^[c]

coeff.^[b]

Negative allosteric modulators

1i

52.9ꢁ10.7

2.72ꢁ1.28 68.1ꢁ8.8

1k

323.5ꢁ39

300.5ꢁ37.7

82.2ꢁ14.6

2.37ꢁ0.30

6.3ꢁ0.4

1l

1.65ꢁ0.38 61.74ꢁ4.31

1m

1.52ꢁ0.28

9.01ꢁ0.78

1n

1o

179.4ꢁ10.5

344.0ꢁ19.9

2.06ꢁ0.15

2.22ꢁ0.16

0.0ꢁ2.0

0.0ꢁ0.7

In conclusion, our structure–activity studies on scaffold 1

suggest a model in which subtle substitutions within distal 2-

oxopyrrolidin-1-yl and benzothiazole moieties both affect affin-

ity and determine whether derivatives of the scaffold 1 act as

negative, silent, or positive allosteric modulators of DgcA activ-

ity.

1p

626.5ꢁ49

1.48ꢁ0.13 59.42ꢁ1.61

Silent allosteric modulators

1g

3.1ꢁ0.7

4.4ꢁ2.2

2.35ꢁ0.82 87.7ꢁ4.2

2.05ꢁ0.12 92.2ꢁ0.6

Structure–activity relationships of scaffold 2

Similarly to several of the derivatives of 1, the second scaf-

fold—4-(2,5-dimethylphenoxy)-N-[4-(morpholin-4-yl)phenyl]bu-

tanamide (2)—also acts as a partial inhibitor. Thus, we hy-

pothesized that 2 might bind through a mechanism similar to

that observed for 1 and its derivatives 1a–u. If both scaffolds

bind to the same regulatory site of DgcA, it should be feasible

to alter their modulatory effect by exchanging the positions

and modifications of the aryl substituents in 2 to generate de-

rivatives with negative, silent, or even positive allosteric modu-

latory effects. We tested a set of 14 analogues of 2 by altering

the aryl substituents on R¹and R²and keeping the butanamide

linker constant (Table 3).

1j

Positive allosteric modulators

1h

238.1ꢁ72.0

2.02ꢁ0.19 117.9ꢁ1.4

2.51ꢁ0.68 176.1ꢁ3.3

1q

222.4ꢁ85.6

55.5ꢁ13.6

1r

2.01ꢁ0.18 156.4ꢁ1.7

The parent scaffold 2 harbors two methyl substituents as R¹

and one 4-morpholin-4-yl substituent as R²(Table 3). We found

that removal of all aryl substituents, in 2c, does not significant-

ly alter the binding affinity (10.6 vs. 6.4 mm) or the activity of

the enzyme-modulator-bound complex (38.6 vs. 31.4%). For

compound 2d, with a single 4-methyl substituent, we ob-

served a binding affinity of 207 mm. For the 2-methyl-substitut-

ed derivative 2e, we determined an IC₅₀of 2.2 mm. The steric

load of a single methyl group at the para position barely ac-

counts for a 100-fold difference in binding affinity between 2d

and 2e. Similar results were observed for alterations in the

methyl substitution pattern in compound 2a. Movement of a

methyl group from the 5- to the 4-position as R¹preserved

affinity at 11.7 mm but meanwhile increased the activity of the

[a] All experiments to determine IC₅₀and AC₅₀values were run at least in

duplicate at each compound dilution. IC₅₀and AC₅₀values were averaged

when determined in two or more independent experiments. [b] Hill co-

efficients obtained by fitting the concentration–response curves. [c] DGC

activities of the enzyme·substrate·inhibitor complexes.

increased structural flexibility of the CꢀOꢀC bond might be im-

portant determinants for the affinity and potency of the scaf-

fold 1. Similarly, the introduction of a methyl group in the

linker region, as in 1t, or extension of the linker space be-

tween the 2-oxopyrrolidin-1-yl and the 1,3-benzothiazole-6-

carbonyloxy components, as in 1u, completely abolished the

binding affinity, thus suggesting that both the composition of

the linker region and the relative orientation and distance be-

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

5

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

&

These are not the final page numbers! ÞÞ

Full Papers

Table 3. Exploration of aryl substituents in 2 and their modulatory effects

on DGC activity.

Cmpd R¹

R²

IC₅₀/AC₅₀

ꢁ

Hill

coeff.^[b]

Activity

ESI [%]^[c]

SD [mm]^[a]

Negative allosteric modulators

2

2-CH₃, 5-CH₃morpholin-4-yl

6.4ꢁ0.5 1.36ꢁ0.13 31.4ꢁ1.3

2a

2c

2d

2e

2 f

2g

2h

2i

2-CH₃, 4-CH₃morpholin-4-yl 11.7ꢁ1.8 2.01ꢁ0.08 69.8ꢁ0.7

H

10.6ꢁ0.8 2.45ꢁ0.34 38.6ꢁ2.1

4-CH₃

2-CH₃

morpholin-4-yl 207.0ꢁ77.8 2.42ꢁ1.35 42.2ꢁ9.3

morpholin-4-yl

2.2ꢁ0.5 2.34ꢁ1.04 63.3ꢁ8.5

127.3ꢁ0.9 1.29ꢁ0.10 5.9ꢁ0.4

2-CH₃, 5-CH₃3-COCH₃

H

4-NHCOCH₃

620.3ꢁ77.2 1.19ꢁ0.10 8.3ꢁ2.5

4-OCH₃

H

4-NHCO(C₃H₅) 271.1ꢁ13.9 0.96ꢁ0.07 1.2ꢁ0.0

4-N(CH₃)₂

4.4ꢁ2.2 2.05ꢁ0.12 92.2ꢁ0.6

12.1ꢁ5.2 1.99ꢁ0.04 95.5ꢁ0.1

Silent allosteric modulator

2j 4-morpholine

Positive allosteric modulators

H

2k

2l

2m

H

2-morpholine

3-sulfonyl-

morpholine

2-hydroxy,

5-sulfonyl-

morpholine

2-OCH₃,

69.9ꢁ5.8 2.87ꢁ0.61 172.6ꢁ2.4

46.8ꢁ3.4 2.84ꢁ0.46 167.3ꢁ2.2

25.0ꢁ5.3 2.06ꢁ0.15 142.0ꢁ1.4

4-OCH₃

H

2n

2b

4-tBu

25.4ꢁ3.5 2.08ꢁ0.51 146.5ꢁ4.4

Figure 3. Examples of silent, negative, and positive allosteric modulators

(SAMs, NAMs, and PAMs, respectively) of the DGC enzyme DgcA. A) Struc-

ture of and concentration–response curve for the 1,3-benzoxazole-6-carbon-

yloxy derivative 1g, a silent allosteric regulator of DgcA. B) Representative

example of an allosteric activator of the scaffold 1. The 4-nitrobenzoic acid

derivative of 1h (1r) acts as a positive allosteric regulator with AC₅₀of

55.5 mm. The concentration–response curves for the parent 1,3-benzothia-

zole-6-carbonyloxy 1 are shown as dashed lines in (A) and (B). C) Structure

of and concentration–response curve for 2a. Alteration of the two methyl

substituents from a 2,5 to a 2,4 arrangement in the 2 scaffold increases the

residual activity of the enzyme·substrate·inhibitor complex from 31.4% in 2

H

12.5ꢁ1.0 1.62ꢁ0.04 147.8ꢁ3.7

5-sulfonyl-

morpholine

[a] All experiments to determine IC₅₀and AC₅₀values were run at least in

duplicate at each compound dilution. IC₅₀and AC₅₀values were averaged

when determined in two or more independent experiments. [b] Hill coef-

ficients obtained upon fitting the concentration–response curves. [c] DGC

activities of the enzyme·substrate·inhibitor complexes.

*

(a) to 69.8% in 2a ( ). D) The 2-methoxy-5-(morpholine-4-sulfonyloxy)

derivative 2b is a positive allosteric regulator with AC₅₀of 12.5 mm. Concen-

tration–response curves for the parent 2 are shown as dashed lines in (C)

and (D).

enzyme·inhibitor complex from 31.4 to 69.8%. However, re-

moving either one of the methyl substituents in 2a resulted in

derivatives with differences in binding affinity.

We further expanded the spectrum of negative allosteric

modulators by introducing 3-acetyl substitution, in 2 f, 4-acet-

amido substitution, in 2g, 4-cyclopropanecarboxamido substi-

tution, in 2h, and 4-dimethylamino substitution, in 2i, at R².

With the exception of the dimethylamino derivative 2i, which

retained 71.1% DGC activity, all compounds had inhibitor-

bound activities smaller than 8.5%. Substitutions patterns in

the group of negative allosteric modulators of the scaffold 2

differ with respect to steric demand and hydrogen-bonding

properties; however, most of them contain para-substituents

at R². We investigated the modulatory effects of ortho- and

meta-substitution at R²and introduced either morpholin-2-yl at

the ortho-position, in 2k and 2l, or 3-sulfonyl-morpholinyl sub-

stitution at the meta-position, in 2m. Upon measuring dose–

response curves and maximum inhibition values by use of the

FRET based DGC assay, we observed increasing DGC activity,

between 142 and 172%, for all meta- and ortho-substituted de-

rivatives 2b and 2k–n, thus suggesting that these compounds

act as positive modulators of the DGC enzyme (Figure 3D and

Table 3).

Mechanism of inhibition for 1 and 2

Our results suggest that both scaffolds 1 and 2 are able to

bind to the DgcA enzyme at a site distinct from the GTP bind-

ing pocket. Although it might be possible for such mixed-type

modulators to bind at the active site, allosteric modulation

generally results from compounds binding to different sites.

An attractive candidate for such an allosteric binding site is the

product inhibition site (I site) of DgcA. Product inhibition by c-

di-GMP is a general regulatory mechanism of many DGC en-

zymes and requires a RXXD motif that forms a turn at the end

of a short five-amino-acid linker that directly connects the I site

with the conserved catalytic A site residues GG(D/E)EF.^[9]Previ-

ous molecular simulations on the DGC PleD in the presence

and in the absence of c-di-GMP revealed a significant decrease

in mobility for both I and A site residues upon complex forma-

tion.^[9]This suggests that dynamic coupling between the two

sites through the short connecting b-strand and a balance-like

movement could be responsible for direct information transfer.

To investigate whether allosteric binding of scaffolds 1 and 2

&

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

6

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ÝÝ These are not the final page numbers!

Full Papers

affects substrate binding or maximum reaction rate of DgcA at

the active site, we performed DGC inhibition assays in the

presence of increasing concentrations of substrate GTP

(Figure 4, Tables S2 and S3). For 1 we observed a decrease in

ples. In the future, similar FRET-based biosensors could provide

interesting alternatives to current existing assays. Often, recep-

tor proteins are known to recognize their cognate signaling

molecule with high specificity. Insertions of these receptor do-

mains between FRET pairs of fluorescent proteins should lead

to new biosensor designs that detect their cognate signaling

molecules with high specificity.

Several strategies for the chemical synthesis of c-di-GMP de-

rivatives as potential inhibitors targeting DGC have been re-

ported,^[18,19]and a number of small-molecule inhibitors of DGC

activity have been described previously.^[20–24]Although these

compounds were valuable research tools, their potencies and

pharmaceutical properties were insufficient to determine the

impact of DGC inhibition in vivo. Originally our HT screen was

designed to discover inhibitors of diguanylate cyclases. Howev-

er, our subsequent structure–activity studies unraveled a panel

of 44 compounds that exert surprising variations in modulato-

ry effects on DgcA activity. Among them is one complete neg-

ative allosteric modulator (1, with IC₅₀of 4 mm), two partial

negative regulators (1b and 1c, with 3.6 and 2.2 mm), two

silent allosteric modulators (1g and 1j, with IC₅₀values of 3.1

and 4.4 mm), and one positive allosteric modulator of DgcA ac-

tivity (2b, with AC₅₀of 12.5 mm).

In terms of activities, we discovered remarkable diversity be-

tween compounds, with even small chemical substitutions on

side chains reversing effects between inhibition and DGC acti-

vation. These results emphasize the importance of dynamic

conformational changes rather than static three-dimensional

structure for the biological function of DGC modulators. In

scaffold 1, for example, substitution of a single sulfur atom in

benzothiazole with oxygen increases the activity of the modu-

lator bound complex over 60-fold while maintaining binding

affinity at 4 mm. In the 2 scaffold, shifting a single para-sub-

stituent to the ortho-position converted the silent modulator

2j into the activator 2k. Whereas the I site has previously been

reported as an inhibitory site,^[8,10]our analysis reveals that DGC

enzymes are subjected to additional layers of allosteric regula-

tion. We demonstrate that several synthetic small molecules

are able to activate DgcA allosterically, rather than inhibiting. It

is interesting to speculate that other endogenous molecules or

protein domains might adopt a similar mechanism of action

and function as endogenous allosteric activators of DGC

enzymes coupling DGC activity to distinct cellular signaling

inputs. In fact, several GGDEF-domain-containing proteins, de-

spite harboring all critical conserved active-site residues, fail to

exhibit DGC activity in biochemical assays.^[25,26]Such enzymes

might exhibit conditional DGC activity in the presence of spe-

cific endogenous activators.

Figure 4. Effects of 1 and 2 on substrate binding and maximum reaction

rate of DgcA. Corresponding plot of the initial velocity versus GTP concen-

tration. A) The presence of 1 decreases affinity for substrate binding and

maximum rate of DGC reaction (Table S3). B) In contrast to that of 1, binding

of 2 affects only the K_mfor GTP but does not affect maximum velocity

(Table S3).

affinity for substrate binding as well as a reduction in maxi-

mum rate of the DGC reaction (Figure 4A, B and Table S2).

Thus, binding of 1 likely destabilizes the divalent Mg²⁺carbox-

ylate complex that is part of the transition state, as well as af-

fecting the conformation of additional residues at the guanine

binding pocket, leading to a reduction in binding affinity for

the substrate GTP. In contrast, we found that binding of 2

affects only the K_mfor GTP but does not impair maximum ve-

locity of enzyme catalysis (Figure 4A, B and Table S3). This sug-

gests that binding of 2 at the allosteric site likely induces con-

formational changes at the active site, resulting in reduced af-

finity for substrate binding, but probably maintains the three-

dimensional conformation of catalytic residues during the tran-

sition state.

Discussion

Thanks to their importance in governing amounts and spatio-

temporal distribution of c-di-GMP, DGC enzymes are consid-

ered important targets for control of biofilm formation and an-

tibiotic persistence phenotypes. Our results demonstrate that

fluorescence-protein-based FRET biosensors can be employed

as powerful analytical tools to detect c-di-GMP in the nanomo-

lar range and to monitor performance of diguanylate cyclase

enzyme reactions in a 384-well format in the low nanomolar

range. With increasing numbers of available assays that moni-

tor fluctuations in cellular c-di-GMP levels,^[10,15–17]we can begin

to identify effective approaches by which to discover potent

small-molecule modulators of c-di-GMP signaling and to moni-

tor their phenotypic effects and influence on cellular c-di-GMP

signaling patterns. In addition to c-di-GMP, many other cyclic

nucleotide second messengers, such as cAMP, cGMP, c-di-AMP,

and cGAMP, currently lack high-throughput-screening (HTS)-ca-

pable analytical methods that allow in situ detection of their

levels during enzymatic assays or in complex biological sam-

Conclusion

Our results demonstrate the complex interplay between syn-

thetic small molecules and the regulatory mechanism that

drives DGC activation. In many prokaryotic model systems, c-

di-GMP-mediated biofilm formation is regulated by multiple

DGC enzymes in parallel. Thus our studies underline the impor-

tance of in-detail mechanism-of-action studies of candidate

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

7

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

&

These are not the final page numbers! ÞÞ

Full Papers

Protein expression: Escherichia coli BL21 (DE3) strain BC1058^[9]

containing the constitutive active C. crescentus DGC (CC3285)

cloned into pET42b plasmid was used for recombinant expression

of the DGC enzyme. DgcA protein was overexpressed and His-tag

purified according to ref. [9]. E. coli BL21 (DE3) strain containing

pET15b::mCPet-12AA-mYPet plasmid was used for recombinant ex-

pression of the c-di-GMP biosensor. The expression vector used for

protein purification of c-di-GMP biosensors was a modified variant

of the pET15b::mCPet-12AA-mYPet^[28]containing an N-terminal

polyhistidine tag for affinity purification, monomerizing mutations

in CYPet (A206K) and YPet (A206K), and a multiple cloning site in

its 12-amino-acid linker region between the mYPet and mCYPet

FRET pair. Two restriction sites (SpeI and KpnI) were introduced by

PCR onto the 5’- and 3’-ends of a codon-optimized version of Sal-

monella enterica serovar Typhimurium ycgR (STM1798) gene, and

the products were cloned in-frame into the SpeI and KpnI sites of

pET15b::mCPet-12AA-mYPet.^[10]YcgR biosenso was overexpressed

and His-tag purified according to ref. [9].

DGC inhibitors prior to their phenotypic evaluation. In the

future, it will be interesting to investigate whether distinct

fragments of the compounds identified will act as general DGC

modulators or exhibit specificity towards distinct DGC sub-

types. Compounds that act specifically on distinct enzymes

might serve as exciting chemical genetics tools for the dissec-

tion of DGC networks, whereas inhibitors with generalized

effects of DGC activity might have the potential to serve as

lead compounds for the development of biofilm modulators

and other c-di-GMP-mediated phenotypes involved in bacterial

pathogenesis.

Experimental Section

Compound library: Compounds for screening were provided by

the national screening laboratory NSRB, in collaboration with the

ICCB-Longwood screening facility located at Harvard Medical

School. The chemical library screened is a subset of a compilation

of commercially available small-molecule chemical libraries main-

tained at the ICCB. The compounds have been prescreened for

drug-like properties (Lipinski’s rules).^[27]Stock libraries were stored

at a concentration of 5 mgmL^ꢀ1in DMSO at ꢀ808C in 384-well

format, with at least two empty columns on each plate to allow

for on-plate controls. Libraries (number of molecules) from the

following companies were evaluated: Asinex Corporation (12377,

www.asinex.com, Winston-Salem, NC, USA), Enamine, Ltd. (11264,

www.enamine.net, Kiev, Ukraine), and Life Chemicals (3893, Burling-

ton, Canada).

Storage of proteins and reagents: All proteins used in our assays

were purified by affinity chromatography followed by gel filtration.

These enzymes were kept at a low concentration (<0.1 mgmL^ꢀ1to

prevent aggregation) in a solution containing glycerol (20%), pro-

tease inhibitors, and b-mercaptoethanol. The DGC enzyme is active

for at least two weeks at 48C and >90% active for at least 48 h at

room temperature. The YcgR biosensor can be stably kept for

weeks in the dark at 48C. For long-term storage the biosensor was

stored at ꢀ808C, in small aliquots containing glycerol (20%), tris

buffer (pH 8, 10 mm), and NaCl (250 mm). We have tested biosen-

sor that has been frozen for more than one year for stability and it

did not show any significant loss of activity. C-di-GMP is kept at

ꢀ208C for long-term storage as a lyophilized powder or solution.

Our current stocks are made in house biosynthetically, by using a

constitutive active derivative of DgcA enzyme,^[9]and purified by

HPLC. GTP stock solutions are prepared just before the assay or

screening because it will hydrolyze over time.

Compounds used for dose–response (IC₅₀) assays: For confirma-

tory assays and dose–response (IC₅₀) assays the following com-

pounds were purchased: from Enamine:

1 (T5267517), 1a

(T5382986), 1b (T5423966), 1c (T5354248), 1d (T5283214), 1e

(T5290508), 1 f (T5841885), 2g (T5257569), 2h (T6309238), 2i

(T5401340), 2k (T5558380), 2l (T5555070), 2m (T5577173), 2n

(T5553390), 2b (T5577172), 2o (T6283377), 2p (T0520-5955), 3

Diguanylate cyclase assay: DGC reactions were performed at

258C with purified hexahistidine-tagged DgcA (20 nm) and hexa-

histidine-tagged mCPet-YcgR-mYPet biosensor (50 nm) to monitor

change in fluorescence upon c-di-GMP production. The DGC reac-

tion buffer contained NaCl (250 mm), Tris·Cl (pH 8.0, 25 mm), and

MgCl₂(20 mm). For screening assays, the reaction mixture was pre-

incubated with compound (200 nL, 5 mgmL^ꢀ1) in DMSO, and fluo-

rescence emission was measured prior to addition of YcgR biosen-

sor, then again after addition of GTP (t=0 min), and 3 h post GTP

addition as the endpoint measurement. For IC₅₀inhibition assays,

the protein was preincubated with different concentrations of

compounds dissolved in DMSO (1–100 mm, 1% final DMSO concen-

tration) for 2 min at 258C, after which GTP substrate was added to

a final concentration of 20 mm. During screening, the fluorescence

emission ratio was measured prior to addition of the YcgR biosen-

sor to monitor background fluorescence at 470 and 535 nm. Plates

were directly measured after addition of GTP (20 mm, t=0 min)

and then after 3 h incubation at RT. During secondary assays and

IC₅₀inhibition assays change in fluorescence ration was measured

at time intervals of 1 min. We determined the plate uniformity,

signal variability, and repeatability assessment by use of NCGC

guidelines. The data were analyzed by using the Excel file supplied

by the NCGC assay validation website. We conducted the “plate

uniformity assay” recommended by NCGC for CV and Z factor cal-

culation for measuring plate-to-plate variation as well as day-to-

day variation. The average Z factor for three plates, read on two

(T5580955),

4 (T5411955); from Asinex: 2 (BAS 05165102), 5

(BAS 01811924), 7 (BAS 14051565), 2c (BAS 01833594), 2d (BAS

03310955), 2e (BAS 03079336), 2j (BAS 01835812); from ABI Chem

(www.abichem.com, Munich, Germany);

6

(AC1N893M), 2a

(AC1LV267); from Molport Corporation (www.molport.com, Riga,

Latvia); 2 f (MolPort-002–121–569).

Screening equipment: Multiwell liquid dispensing for setting up

assays was performed with the liquid handler PlateMate Plus (Ham-

ilton) and Microflo plate dispenser (Biotek). HTS, DGC enzyme in-

hibition reactions, and dose–response (IC₅₀) assays were performed

in Corning Low Volume 384 Well Black Flat Bottom Polystyrene Mi-

croplates (Product #3821BC) in a final volume of 20 mL per well.

Change in fluorescence ratio (535/470 nm) was measured with an

Envision 2104 fluorescence plate reader (PerkinElmer) to monitor

for c-di-GMP production. The plate reader was equipped with a

CFP excitation filter (l_ex=430 nm, BW 24 nm, T_min90%) a dichroic

mirror module (D450/D505 nm), emission filters for YFP (CWL

535 nm, BW 30 nm, T_min90%) and CFP (CWL 470 nm, BW 24 nm,

T_min=90%), and dual PMT detectors for recording CFP and YFP sig-

nals simultaneously. Measurement mode was bi-directional by

rows (excitation and emission light set to 100% and number of

flashes to 10) with use of dual detector mode with a PMT gain of

383 (535 nm) and 572 (470 nm) and a measurement height of

7 mm above the plate height.

&

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

8

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ÝÝ These are not the final page numbers!

Full Papers

h

consecutive days, was 0.71 and the CV for high, medium, and low

signal were well below 5%.

½ðA_ESꢀ A_ESIÞ½Iꢂꢂ

AðIÞ ¼ A_ES

ꢀ

ð5Þ

h

IC₅₀þ ½Iꢂ

Data analysis: Data analysis was performed in Excel 97. All primary

screening was performed at 50 mgmL^ꢀ1compound concentrations.

The fluorescence emission ratio (535/470 nm) of the YcgR biosen-

sor in the absence of c-di-GMP was used as the positive control for

DgcA inhibition; biosensor in the presence of c-di-GMP (5 mm)

served as the negative control. For a compound to be considered

a hit, the following criteria had to be met: 1) compounds with c-di-

GMP levels below 300 nm after 3 h incubation with DgcA and GTP

(20 mm) were considered strong hits, 2) duplicates had to be con-

sistent, with difference in final c-di-GMP levels <100 nm, and

3) background fluorescence prior to addition of the FRET biosensor

could not exceed 7% of the FRET signal in CFP and YFP emission

channels. Total fluorescence was used as a quality control criterion

to identify inherently fluorescent molecules. Of the 27502 com-

pounds screened in duplicate, 1029 (3.7%) were removed from the

analysis due to autofluorescence, and 49 compounds (0.19%)

exhibited c-di-GMP levels below 300 nm after 3 h incubation with

DgcA enzyme. Of these, 18 compounds were tested in secondary

assays. Seven had potencies below 50 mm and three compounds

below 10 mm.

where A(I)=relative DGC enzymatic activity as a function of inhibi-

tor concentration, A_ES=relative activity in absence of inhibitor,

A_ESI=relative residual activity under inhibitor saturating conditions,

IC₅₀=half-maximum inhibitory concentration of inhibitor, [I] con-

centration of inhibitor, and h=Hill coefficient.

AC₅₀constants were fitted with the following function [Eq. (6)]:

h

½ðA_ESIꢀ A_ESÞ½Iꢂꢂ

AðIÞ ¼ A_ES

þ

ð6Þ

h

AC₅₀þ ½Iꢂ

where A(I)=relative DGC enzymatic activity as a function of activa-

tor concentration, A_ES=relative activity in absence of activator,

A_ESI=relative activity under activator saturating conditions, AC₅₀=

half-maximum activating concentration of activator, [I]=concentra-

tion of activator, and h=Hill coefficient.

All hit validation and HTS procedures were carried out according

to the general guidelines as published on the U.S. National Chemi-

cal Genomics Center web site (NCGC Assay Guidance Manual and

High-Throughput Assay Guidance criteria, http://www.ncbi.nlm.-

nih.gov/books/NBK53196/).

The FRET fluorescence emission ratio was defined as [Eq. (1)]:

Chemistry

Em_{527 nm}

Em_{470 nm}

ð1Þ

FRET ¼

General methods: All reactions were run under dry nitrogen. Re-

agents and solvents were obtained in the highest available purity

1

and used without further purification unless indicated. H NMR and

The following formula involving the means of the FRET ratios of

positive and negative controls (m₊, m ) and their standard devia-

¹³C NMR spectra were obtained with a 500 MHz (Bruker AV500) in-

strument. Identities of the compounds were confirmed by mass

spectrometry. A solution of each compound was infused into the

electrospray ionization source operating in positive- or negative-

ion mode. Low-resolution spectra were obtained with an Esquire

LC ion trap mass spectrometer (Bruker Daltonics, Billerica, MA). Ac-

curate mass measurements were performed with an APEX Qe 47

Fourier transform ion cyclotron resonance mass spectrometer

(Bruker Daltonics). Normal-phase silica gel purifications were per-

formed with a Biotage SP4 instrument and the cartridges supplied

by Biotage. RP-HPLC was carried out with a Varian instrument

equipped with a diode array ultraviolet detector. For preparative

reversed-phase chromatography a 10ꢁ250 mm C₁₈5m column at a

flow rate of 4.6 mLmin^ꢀ1was used; for analytical reversed-phase

chromatography a 4.6ꢁ250 mm C₁₈5m column at a flow rate of

1 mLmin^ꢀ1was used. Ultraviolet detection was at 215 and either

254 or 360 nm. Unless otherwise specified, buffer A was trifluoro-

acetic acid (TFA) in H₂O (0.05%), whereas buffer B was TFA in ace-

tonitrile (0.05%). Thin-layer chromatography was performed with

0.2 mm polygram SIL G/UV plates (Alltech, Deerfield, IL), developed

with mobile phases of varying compositions of ethyl acetate/

hexane, MeOH/CH₂Cl₂, or MeOH/CHCl₃, and visualized by UV light

supplemented by vanillin, ninhydrin, and other solution stains

where appropriate.

ꢀ

tions (s₊, s ) was used to calculate the Z’ factor of the FRET-based

ꢀ

HTS assay [Eq. (2)]:

3s_þþ 3s

ꢀ

Z⁰factor ¼ 1 ꢀ

ð2Þ

jm ꢀ m j

þ

ꢀ

V, the fraction of YcgR biosensor in the c-di-GMP bound state, was

calculated according to [Eq. (3)]:

FRET_freeꢀ FRET_assay

ð3Þ

V ¼

FRET_freeꢀ FRET_bound

where FRET_freeis the fluorescence emission ratio (535/470 nm) of

the free biosensor, FRET_boundis the fluorescence emission ratio

(535/470 nm) of YcgR with c-di-GMP bound, and FRET_assayis the

fluorescence emission ratio (535/470 nm) of the YcgR biosensor

during the assay.

The c-di-GMP concentration (in nm) was calculated according to

[Eq. (4)]:

rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

2

VK_D

ð4Þ

½c-di-GMPꢂ ¼

1 ꢀ V

1-[2-(Benzyloxy)acetyl]pyrrolidin-2-one (1w; Scheme 1): 1-(Trime-

thylsilanyl)pyrrolidin-2-one (5.9 mL, 36.95 mmol) in diethyl ether

(55 mL) was added dropwise at 08C to a solution of benzyloxyace-

tyl chloride (5.3 mL, 33.59 mmol) in diethyl ether (55 mL). The ice

bath was removed, and the reaction mixture was stirred at room

temperature 1.5 h and then concentrated in vacuo. The crude

product was purified by silica gel chromatography with a gradient

of ethyl acetate in hexanes (30 to 60%) to give 1w (7.425 g,

with K_D=198 nm, c-di-GMP dissociation constant of the YcgR bio-

sensor at 258C.

Initial velocity (V₀) was determined by plotting c-di-GMP concentra-

tion versus time and by fitting the curve according to Michaelis–

Menten kinetics. The dose–response curves of percentage activity

were fit in ProFit 5.6.7.

1

IC₅₀inhibition constants were fitted by using the following func-

tion [Eq. (5)]:

31.87 mmol). H NMR (500 MHz, CDCl₃): d=1.92–2.08 (m, 2H), 2.48

(td, J=8.0, 3.6 Hz, 2H), 3.75 (td, J=7.0, 3.7 Hz, 2H), 4.61 (s, 4H),

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

9

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

&

These are not the final page numbers! ÞÞ

Full Papers

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl benzo[d]oxazole-6-carboxyl-

ate (1g): Compound 1g was prepared according to the procedure

in General Method A from 1v and benzo[d]oxazole-6-carboxylic

acid on a 0.30 mmol scale. The crude

product was purified five times by

silica gel chromatography with

a

gradient of MeOH in CH₂Cl₂(0 to

5%) to give 1g (25 mg, 0.087 mmol).

¹H NMR (500 MHz, CDCl₃): d=2.08–

Scheme 1. Synthesis of 1v. a) Benzyloxyacetyl chloride, diethyl ether, 08C–

RT, 1.5 h. b) H₂, Pd-C, EtOH, DMF.

2.26 (m, 2H), 2.66 (t, J=8.1 Hz, 2H),

3.87 (t, J=7.2 Hz, 2H), 5.45 (s, 2H),

7.86 (d, J=8.3 Hz, 1H), 8.19 (d, J=

8.3 Hz, 1H), 8.26 (s, 1H), 8.38 ppm (s, 1H); ¹³C NMR (500 MHz,

CDCl₃): d=17.92, 32.93, 44.89, 65.13, 113.21, 120.40, 126.67, 127.15,

144.15, 149.65, 155.06, 165.56, 167.86, 176.13 ppm; MS: m/z: 311.2

[M+Na]⁺; HRMS: m/z calcd for C₁₄H₁₂N₂NaO₅[M+Na]⁺: 311.0638;

found: 311.0644.

7.15–7.28 (m, 1H), 7.29 (t, J=6.6 Hz, 2H), 7.32–7.49 ppm (m, 2H);

¹³C NMR (500 MHz, CDCl₃): d=17.74, 33.00, 44.78, 70.83, 73.21,

127.79, 127.94, 128.37, 137.53, 171.16, 175.70 ppm; MS: m/z: 256.2

[M+Na]⁺.

1-(2-Hydroxyacetyl)pyrrolidin-2-one (1v): In a flame-dried flask,

EtOH (200 mL) was bubbled with argon for 10 min. Palladium on

carbon (10%, 7.425 g) was added, and the suspension was bub-

bled with argon for 10 min, 1w (7.425 g, 31.87 mmol) in EtOH

(100 mL) and DMF (33 mL) was then added, and the system was

bubbled with argon for 10 min and then with H₂until completion.

The solution was bubbled with argon for a further 10 min, filtered

through celite, and concentrated in vacuo to give 1v (4.436 g,

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl

benzofuran-5-carboxylate

(1j): Compound 1j was prepared according to the procedure in

General Method A from 1v and benzofuran-5-carboxylic acid on a

0.38 mmol scale. The crude product

was purified by silica gel chromatog-

raphy with a gradient of MeOH in

1

31.02 mmol). H NMR (500 MHz, CDCl₃): d=2.05–2.20 (m, 2H), 2.58

(t, J=8.1 Hz, 2H), 3.83 (t, J=7.2 Hz, 2H), 4.62 ppm (s, 2H); ¹³C NMR

(500 MHz, CDCl₃): d=17.91, 32.88, 44.93, 64.17, 174.30,

175.92 ppm; MS: m/z: 166.1 [M+Na]⁺; HRMS: m/z calcd for

C₇H₁₃NNaO₄[M+Na+CH₃OH]⁺¹: 198.0737; found: 198.0736.

CH₂Cl₂(0 to 5%) to give 1j (23.6 mg,

0.082 mmol).

¹H NMR

(500 MHz,

CDCl₃): d=2.09–2.27 (m, 2H), 2.66 (t,

J=8.1 Hz, 2H), 3.87 (t, J=7.2 Hz, 2H),

5.45 (s, 2H), 6.87 (d, J=1.2 Hz, 1H),

7.57 (d, J=8.7 Hz, 1H), 7.72 (d, J=

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl 2-methylbenzo[d]thiazole-6-

carboxylate (1i)—General Method A (Scheme 2): N’-(3-Dimethyl-

2.0 Hz, 1H), 8.12 (dd, J=8.7, 1.4 Hz, 1H), 8.45 ppm (s, 1H); ¹³C NMR

(500 MHz, CDCl₃): d=17.92, 32.97, 44.90, 64.87, 107.17, 111.37,

124.25, 124.43, 126.44, 127.49, 146.31, 157.71, 166.30, 168.17,

176.09 ppm; MS: m/z: 310.3 [M+Na]⁺; HRMS: m/z calcd for

C₁₅H₁₃NNaO₅[M+Na]⁺: 310.0686; found: 310.0690.

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl 1H-indole-5-carboxylate (1k):

Compond 1k was prepared according to the procedure in General

Method A from 1v and 1H-indole-5-carboxylic acid on a 0.29 mmol

scale. The crude product was purified four times by silica gel chro-

Scheme 2. General method A. a) 2-Methylbenzo[d]thiazole-5-carboxylic acid,

EDCI, DMAP, CH₂Cl₂, DMF, RT, overnight.

aminopropyl)-N-ethylcarbodiimide (EDCI, 131 mg, 0.44 mmol), 4-di-

methylaminopridine (DMAP, 3 mg, 0.027 mmol), and 2-methyl-1,3-

benzothiazole-6-carboxylic acid (55 mg, 0.31 mmol) were added to

a solution of 1v (49 mg, 0.34 mmol) in CH₂Cl₂(2 mL) and DMF

(0.5 mL). The reaction mixture was stirred at room temperature

overnight and then partitioned between CH₂Cl₂and saturated

NaHCO₃solution. The organic layer

matography with a gradient of ethyl acetate in hexanes (10 to

50%) to give 1k (8.2 mg, 0.029 mmol). ¹H NMR (500 MHz,

[D₇]DMF): d=2.26–2.45 (m, 2H), 2.85 (t, J=8.1 Hz, 2H), 3.97 (t, J=

7.2 Hz, 2H), 5.59 (s, 2H), 6.88 (s, 1H), 7.70–7.89 (m, 2H), 8.04 (dd,

J=8.6, 1.4 Hz, 1H), 8.60 (s, 1H), 11.77 ppm (s, 1H); ¹³C NMR

(500 MHz, [D₇]DMF): d=17.87, 32.82, 45.05, 64.50, 103.02, 111.61,

120.70, 122.60, 123.51, 127.52, 127.98, 139.46, 166.96, 168.32,

176.78 ppm; MS: m/z: 309.2 [M+Na]⁺; HRMS: m/z calcd for

C₁₅H₁₄N₂NaO₄[M+Na]⁺: 309.0846; found: 309.0837.

was washed with brine, dried over

Na₂SO₄, filtered, and concentrated in

vacuo. The crude product was puri-

fied by silica gel chromatography

with a gradient of MeOH in CH₂Cl₂(0

to 3%) to give 1i (62 mg,

0.19 mmol).

¹H NMR

(500 MHz,

CDCl₃): d=2.08–2.21 (m, 2H), 2.62 (t,

J=8.1 Hz, 2H), 2.86 (s, 3H), 3.83 (t, J=7.2 Hz, 2H), 5.42 (s, 2H), 7.98

(d, J=8.6 Hz, 1H), 8.18 (dd, J=8.6, 1.7 Hz, 1H), 8.62 ppm (d, J=

1.3 Hz, 1H); ¹³C NMR (500 MHz, CDCl₃): d=17.89, 20.46, 32.91,

44.87, 65.01, 122.16, 124.06, 125.84, 127.55, 135.65, 156.58, 165.70,

167.92, 171.06, 176.09 ppm; MS: m/z: 341.5 [M+Na]⁺; HRMS: m/z

calcd for C₁₅H₁₄N₂NaO₄S [M+Na]⁺: 341.0566; found: 341.0569.

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl 1-methyl-1H-indole-5-carbox-

ylate (1l): Compound 1l was prepared according to the procedure

in General Method A from 1v and 1-methyl-1H-indole-5-carboxylic

acid on a 0.39 mmol scale. The crude product was purified three

times by silica gel chromatography with a gradient of MeOH in

CH₂Cl₂(0 to 3%) to give 1l (6.3 mg, 0.021 mmol). ¹H NMR

&

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

10

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ÝÝ These are not the final page numbers!

Full Papers

(m, 2H), 2.67 (t, J=7.6 Hz, 2H), 3.88

(t, J=6.2 Hz, 2H), 5.51 (s, 2H), 8.20

(d, J=8.4 Hz, 1H), 8.44 (d, J=8.1 Hz,

1H), 8.86–9.06 ppm (m, 3H); ¹³C NMR

(500 MHz, CDCl₃): d=17.94, 32.92,

44.90, 65.35, 129.75, 129.93, 130.77,

132.80, 142.30, 145.15, 146.01,

146.75, 165.24, 167.64, 176.14 ppm; MS: m/z: 322.3 [M+Na]⁺;

HRMS: m/z calcd for C₁₅H₁₃N₃NaO₄[M+Na]⁺: 322.0798; found:

322.0790.

(500 MHz, CDCl₃): d=2.10–2.26 (m, 2H), 2.67 (t, J=8.1 Hz, 2H),

3.86 (s, 3H), 3.89 (t, J=7.2 Hz, 2H), 5.44 (s, 2H), 6.62 (d, J=3.0 Hz,

1H), 7.14 (d, J=3.2 Hz, 1H), 7.37 (d, J=8.7 Hz, 1H), 8.02 (dd, J=

8.7, 1.4 Hz, 1H), 8.51 ppm (s, 1H); ¹³C NMR (500 MHz, CDCl₃): d=

17.93, 33.02, 33.07, 44.91, 64.64, 102.78, 108.90, 120.53, 123.31,

124.51, 128.00, 130.26, 139.36, 167.25, 168.55, 176.04 ppm; MS:

m/z: 323.2 [M+Na]⁺; HRMS: m/z calcd for C₁₇H₂₀N₂NaO₅

[M+Na+CH₃OH]⁺¹: 355.1264; found: 355.1258.

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl benzothiazole-5-carboxylate

(1p): Compound 1p was prepared according to the procedure in

General Method A from 1v and benzothiazole-5-carboxylic acid on

a 0.36 mmol scale. The crude prod-

uct was purified by silica gel chroma-

tography with a gradient of MeOH in

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl 1-methyl-1H-benzo[d]imida-

zole-5-carboxylate (1m): Compound 1m was prepared according

CH₂Cl₂(0 to 3%) to give 1p

(43.7 mg,

0.14 mmol).

¹H NMR

(500 MHz, CDCl₃): d=2.06–2.25 (m,

2H), 2.64 (t, J=8.1 Hz, 2H), 3.85 (t,

J=7.2 Hz, 2H), 5.46 (s, 2H), 8.03 (d,

J=8.4 Hz, 1H), 8.18 (d, J=7.6 Hz,

1H), 8.88 (s, 1H), 9.09 ppm (s, 1H); ¹³C NMR (500 MHz, CDCl₃): d=

17.91, 32.93, 44.89, 65.11, 121.93, 125.64, 126.36, 127.86, 138.91,

153.13, 155.39, 165.85, 167.86, 176.13 ppm; MS: m/z: 327.3

[M+Na]⁺; HRMS: m/z calcd for C₁₄H₁₂N₂NaO₄S [M+Na]⁺: 327.0410;

found: 327.0407.

to the procedure in General Method A from 1v and 1-methyl-1H-

benzimidazole-5-carboxylic acid on a 0.36 mmol scale under argon.

The crude product was purified twice by silica gel chromatography

with a gradient of MeOH in CH₂Cl₂(0 to 15%) to give 1m (4 mg,

1-(2-Bromoacetyl)pyrrolidin-2-one (1x; Scheme 3): Carbon tetra-

bromide (126 mg, 0.38 mmol) was added at 08C to a solution of

1v (50 mg, 0.35 mmol) in CH₂Cl₂(1 mL). A solution of triphenyl-

phosphine (100 mg, 0.38 mmol) in CH₂Cl₂(0.5 mL) was added drop-

1

0.013 mmol). H NMR (500 MHz, CDCl₃): d=2.11–2.28 (m, 2H), 2.68

(t, J=8.1 Hz, 2H), 3.81–4.03 (m, 5H), 5.47 (s, 2H), 7.46 (d, J=8.5 Hz,

1H), 7.99 (s, 1H), 8.15 (dd, J=8.5, 1.3 Hz, 1H), 8.64 ppm (s, 1H);

¹³C NMR (500 MHz, CDCl₃): d=17.94, 31.28, 33.00, 44.91, 64.88,

109.15, 123.35, 123.65, 124.93, 138.01, 143.40, 145.29, 166.64,

168.25, 176.05 ppm; MS: m/z: 302.2 [M+H]⁺; HRMS: m/z calcd for

C₁₅H₁₆N₃O₄[M+H]⁺: 302.1135; found: 302.1142.

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl

isoquinoline-6-carboxylate

(1n): Compound 1n was prepared according to the procedure in

General Method A from 1v and isoquinoline-6-carboxylic acid on a

0.36 mmol scale. The crude product was purified four times by

silica gel chromatography with a gradient of MeOH in CH₂Cl₂(0 to

Scheme 3. Synthesis of 1x. a) CBr₄, triphenylphosphine, CH₂Cl₂, RT, over-

night.

wise and the reaction mixture was allowed to warm to room tem-

perature overnight. The crude reaction mixture was concentrated

in vacuo and purified by silica gel chromatography with a gradient

of MeOH in CH₂Cl₂(0 to 5%) to give 1x (48 mg, 0.23 mmol).

¹H NMR (500 MHz, CDCl₃): d=2.05–2.17 (m, 2H), 2.65 (t, J=8.1 Hz,

2H), 3.86 (t, J=7.3 Hz, 2H), 4.50 ppm (s, 2H); ¹³C NMR (500 MHz,

CDCl₃): d=17.35, 30.09, 33.11, 45.82, 166.61, 175.38 ppm; MS: m/z:

10%)

to

give

1n

(6.4 mg,

0.021 mmol). ¹H NMR (500 MHz,

CDCl₃): d=2.13–2.31 (m, 2H), 2.69

(t, J=8.1 Hz, 2H), 3.90 (t, J=7.1 Hz,

2H), 5.52 (s, 2H), 7.80 (d, J=5.3 Hz,

1H), 8.09 (d, J=8.5 Hz, 1H), 8.27 (d,

J=8.4 Hz, 1H), 8.65 (d, J=5.3 Hz,

1H), 8.70 (s, 1H), 9.39 ppm (s, 1H);

¹³C NMR (500 MHz, CDCl₃): d=17.95,

230.2

[M+Na]⁺;

HRMS:

m/z

calcd

for

C₇HBrNNaO₃

[M+Na+CH₃OH]⁺¹: 259.9893; found: 259.9892.

32.94, 44.91, 65.27, 121.45, 126.98,

Benzyl-(2-oxo-2-(2-oxopyrrolidin-1-yl)ethyl)carbamate

(1y;

128.04, 129.96, 130.90, 135.02, 143.91, 152.57, 165.55, 167.74,

176.15 ppm; MS: m/z: 321.3 [M+Na]⁺; HRMS: m/z calcd for

C₁₆H₁₅N₂O₄[M+H]⁺: 299.1026; found: 299.1027.

Scheme 4): PCl₅(458 mg, 2.2 mmol) was added at ꢀ158C to a solu-

tion of Cbz-Gly (418 mg, 2.0 mmol) in diethyl ether (2.5 mL). The

reaction mixture was stirred for 3 h at 08C, and then concentrated

in vacuo. The crude acid chloride was taken up in diethyl ether

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl

quinoxaline-6-carboxylate

(3.5 mL),

and

1-(trimethylsilanyl)pyrrolidin-2-one

(0.35 mL,

(1o): Compound 1o was prepared according to the procedure in

General Method A from 1v and quinoxaline-6-carboxylic acid on a

0.37 mmol scale. The crude product was purified by silica gel chro-

matography with a gradient of MeOH in CH₂Cl₂(0 to 10%) to give

2.2 mmol) in diethyl ether (3.5 mL) was added dropwise at 08C.

The ice bath was removed and the reaction mixture was stirred at

room temperature for 1.25 h and then concentrated in vacuo. The

crude product was partitioned between ethyl acetate and NaHCO₃

1

1o (20.6 mg, 0.069 mmol). H NMR (500 MHz, CDCl₃): d=2.08–2.29

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

11

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

&

These are not the final page numbers! ÞÞ

Full Papers

45.67, 121.99, 123.42, 124.98, 131.42, 134.03, 154.79, 157.08,

166.87, 170.01, 175.71 ppm; MS: m/z: 304.4 [M+H]⁺; HRMS: m/z

calcd for C₁₄H₁₄N₃O₃S [M+H]⁺: 304.0750; found: 304.0751.

1-(2-Bromopropanoyl)pyrrolidin-2-one (1aa; Scheme 5): A solu-

tion of 1-(trimethylsilanyl)pyrrolidin-2-one (168 mL, 1.05 mmol) in

diethyl ether (2 mL) was added dropwise and quickly at 08C to a

Scheme 5. Synthesis of 1t. a) 2-bromopropionyl bromide, diethyl ether,

08C–RT, 1.25 h; b) Benzothiazole-6-carboxylic acid, Cs₂CO₃, DMF, RT, over-

night.

Scheme 4. Synthesis of 1s. a) PCl₅, diethyl ether, ꢀ15 to 08C, 3 h; b) 1-(trime-

thylsilanyl)pyrrolidin-2-one, diethyl ether, 08C–RT, 1.25 h; c) TFA, thioanisole,

TMSBr, RT, 1.5 h; d) benzothiazole-6-carboxylic acid, HBTU, HOAt, DIEA, DMF,

CH₂Cl₂, RT, overnight.

solution of 2-bromopropionyl bromide (100 mL, 0.95 mmol) in di-

ethyl ether (2 mL). The ice bath was removed and the reaction mix-

ture was stirred at room temperature for 1.25 h and then concen-

trated in vacuo. The residue was partitioned between ethyl acetate

and NaHCO₃(10%). The aqueous layer was reextracted three times

with ethyl acetate, and the combined organic layers were washed

with brine, dried over Na₂SO₄, filtered, and concentrated in vacuo

to give 1aa (191.7 mg, 0.87 mmol), which was used without further

purification. ¹H NMR (500 MHz, CDCl₃): d=1.76 (d, J=6.8 Hz, 3H),

1.99–2.10 (m, 2H), 2.53–2.67 (m, 2H), 3.71–3.89 (m, 2H), 5.63 ppm

(q, J=6.8 Hz, 1H); ¹³C NMR (500 MHz, CDCl₃): d=17.05, 20.62,

33.51, 40.44, 46.03, 170.29, 174.92 ppm; MS: m/z: 244.0 [M+Na]⁺.

solution (10%). The aqueous fraction was reextracted three times

with ethyl acetate, and the combined organic extracts were

washed with brine, dried over Na₂SO₄, filtered, and concentrated in

vacuo. The crude product was purified by silica gel chromatogra-

phy with a gradient of MeOH in CH₂Cl₂(0 to 10%) to give 1y

(369 mg, 1.34 mmol). ¹H NMR (500 MHz, CDCl₃): d=1.85–2.07 (m,

2H), 2.47 (t, J=8.1 Hz, 2H), 3.68 (t, J=7.2 Hz, 2H), 4.42 (d, J=

5.6 Hz, 2H), 5.05 (s, 2H), 5.88 (brt, J=5.2 Hz, 1H), 7.15–7.50 ppm

(m, 5H); ¹³C NMR (500 MHz, CDCl₃): d=17.52, 33.01, 45.10, 46.33,

66.72, 127.96, 128.29, 136.56, 156.56, 170.39, 175.85 ppm; MS: m/z:

299.3 [M+Na]⁺.

1-Methyl-2-oxo-2-(2-oxopyrrolidin-1-yl)ethyl

benzothiazole-6-

carboxylate (1t): Benzothiazole-6-carboxylic acid (156 mg,

0.87 mmol) and cesium carbonate (311 mg, 0.95 mmol) were

added to a solution of 1aa (192 mg, 0.87 mmol) in DMF (4 mL).

The reaction mixture was stirred overnight to completion and was

partitioned between H₂O and CH₂Cl₂. The aqueous layer was re-

extracted twice with CH₂Cl₂, and the combined organic layers were

washed with brine, dried over Na₂SO₄, filtered, and concentrated in

vacuo. The crude product was purified by silica gel chromatogra-

phy with a gradient of EtOAc in hexanes (40 to 60%) to give 1t

(184 mg, 0.58 mmol). ¹H NMR (500 MHz, CDCl₃): d=1.64 (d, J=

6.8 Hz, 3H), 1.96–2.17 (m, 2H), 2.48–2.77 (m, 2H), 3.67–3.98 (m,

2H), 6.20 (q, J=6.8 Hz, 1H), 8.09–8.27 (m, 2H), 8.71 (d, J=1.1 Hz,

1H), 9.14 ppm (s, 1H); ¹³C NMR (500 MHz, CDCl₃): d=16.63, 17.48,

33.35, 45.45, 70.89, 123.39, 124.52, 126.78, 127.50, 133.74, 156.20,

157.61, 165.56, 171.70, 175.30 ppm; MS: m/z: 341.2 [M+Na]⁺;

HRMS: m/z calcd for C₁₅H₁₅N₂O₄S [M+H]⁺: 319.0747; found:

319.0743.

1-(2-Aminoacetyl)pyrrolidin-2-one (1z): Thioanisole (4.0 mL,

33.88 mmol) and TMSBr (0.89 mL, 6.78 mmol) were added to a

solution of 1y (187 mg, 0.68 mmol) in TFA (13.5 mL). The reaction

mixture was stirred at room temperature for 1.5 h and then con-

centrated in vacuo. The crude product was purified by silica gel

chromatography with a gradient of MeOH in CH₂Cl₂(0 to 25%) to

give 1z, which was used without further purification. ¹H NMR

(500 MHz, MeOD): d=2.08–2.25 (m, 2H), 2.33 (t, J=8.1 Hz, 1H),

2.66 (t, J=8.1 Hz, 1H), 3.43 (t, J=7.0 Hz, 1H), 3.77–3.94 (m, 3H),

4.35 ppm (brs, 2H); ¹³C NMR (500 MHz, MeOD): d=17.23, 32.31,

42.30, 44.88, 167.37, 177.17 ppm; MS: m/z: 143.2 [M+H]⁺.

N-(2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl)benzo[d]thiazole-6-car-

boxamide (1s):

A solution of benzothiazole-6-carboxylic acid

(122 mg, 0.68 mmol), N-[(1H-benzotriazole-1-yl)(dimethylamino)-

methylene]-N-methylmethanaminium hexafluorophosphate N-

oxide (HBTU, 774 mg, 2.04 mmol), and 1-hydroxy-7-azabenzotria-

zole (HOAt, 278 mg, 2.04 mmol) in CH₂Cl₂(5 mL) and DMF (1 mL)

was stirred at room temperature for 1 h, 1z (96 mg, 0.68 mmol)

and N,N’-diisopropylethylamine (DIEA, 0.53 mL, 3.04 mmol) in DMF

(2 mL) were then added, and the reaction mixture was stirred over-

night. The reaction mixture was diluted with CH₂Cl₂and washed

with water and saturated NaHCO₃solution, dried over Na₂SO₄, fil-

tered, and concentrated in vacuo. The crude product was purified

by reversed-phase HPLC with a gradient from 10 to 95% B in A

over 30 min to give 1s (20.5 mg, 0.068 mmol). ¹H NMR (500 MHz,

CDCl₃): d=2.10–2.25 (m, 2H), 2.67 (t, J=8.1 Hz, 2H), 3.89 (t, J=

7.2 Hz, 2H), 4.84 (d, J=5.1 Hz, 2H), 7.25 (brs, 1H), 7.97 (dd, J=8.5,

1.6 Hz, 1H), 8.18 (d, J=8.5 Hz, 1H), 8.53 (d, J=1.4 Hz, 1H),

9.18 ppm (s, 1H); ¹³C NMR (500 MHz, CDCl₃): d=17.67, 33.09, 45.21,

3-tert-Butoxy-3-oxopropyl benzo[d]thiazole-6-carboxylate (1ab;

Scheme 6): Compound 1ab was prepared according to the proce-

dure in General Method A from tert-butyl 3-hydroxypropionate and

benzothiazole-6-carboxylic acid on a 1.0 mmol scale. The crude

product was purified by silica gel chromatography with a gradient

of MeOH in CH₂Cl₂(0 to 5%) to give 1ab (227 mg, 0.74 mmol).

¹H NMR (500 MHz, CDCl₃): d=1.40 (s, 9H), 2.69 (t, J=6.3 Hz, 2H),

4.56 (t, J=6.3 Hz, 2H), 8.09 (s, 2H), 8.60 (s, 1H), 9.11 ppm (s, 1H);

¹³C NMR (500 MHz, CDCl₃): d=28.02, 35.25, 61.07, 81.04, 123.33,

124.21, 127.20, 127.26, 133.73, 156.02, 157.43, 165.63, 169.80 ppm;

MS: m/z: 330.3 [M+Na]⁺.

3-(Benzo[d]thiazole-6-carbonyloxy)propanoic acid (1ac): Com-

pound 1ab (227 mg, 0.74 mmol) was dissolved in TFA (3 mL) and

CH₂Cl₂(3 mL), and the mixture was stirred at room temperature for

&

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

12

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ÝÝ These are not the final page numbers!

Full Papers

General Method A from 1v and 4-(dimethylamino)benzoic acid on

a 0.35 mmol scale. The crude product was purified by silica gel

chromatography with a gradient of MeOH in CH₂Cl₂(0 to 10%) to

give 1q (3.1 mg, 0.011 mmol). ¹H NMR (500 MHz, CDCl): d=2.09–

2.23 (m, 2H), 2.65 (t, J=8.1 Hz, 2H), 3.07 (s, 6H), 3.87 (t, J=7.2 Hz,

2H), 5.36 (s, 2H), 6.69 (d, J=9.0 Hz, 2H), 8.00 ppm (d, J=9.0 Hz,

2H); ¹³C NMR (500 MHz, CDCl₃): d=17.90, 32.99, 40.12, 44.88,

64.32, 110.69, 116.05, 131.74, 153.56, 166.39, 168.68, 175.99 ppm;

MS: m/z: 291.1 [M+H]⁺; HRMS: m/z calcd for C₁₅H₁₉N₂O₄[M+H]⁺:

291.1345; found: 291.1348.

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl 4-nitrobenzoate (1r): Com-

pound 1r was prepared according to the procedure in General

Method A from 1v and 4-nitrobenzoic acid on a 0.35 mmol scale.

Scheme 6. Synthesis of 1u. a) tert-Butyl 3-hydroxypropionate, EDCI, DMAP,

CH₂Cl₂, DMF, RT, overnight; b) H₂, Pd-C, EtOH, DMF, TFA, CH₂Cl₂; c) benzothia-

zole-6-carboxylic acid, EDCI, DMAP, CH₂Cl₂, DMF, RT, overnight.

2.5 h and concentrated in vacuo to give 1ac (171 mg, 0.68 mmol),

which was used without further purification. ¹H NMR (500 MHz,

CDCl₃): d=2.94 (t, J=6.0 Hz, 2H), 4.70 (t, J=6.0 Hz, 2H), 8.21 (s,

2H), 8.73 (s, 1H), 9.22 ppm (s, 1H); MS: m/z: 252.3 [M+H]⁺.

3-Oxo-3-(2-oxopyrrolidin-1-yl)propyl benzo[d]thiazole-6-carbox-

ylate (1u): Compound 1ac (170 mg, 0.68 mmol) was taken up in

thionyl chloride (5 mL) with a drop of DMF, stirred at room temper-

ature overnight, and then concentrated in vacuo. The crude acid

chloride was taken up in THF (7 mL) and chilled to 08C in an ice

bath. 1-(Trimethylsilanyl)pyrrolidin-2-one (0.12 mL, 0.75 mmol) in

THF (1 mL) was added dropwise quickly, the ice bath was removed,

and the reaction mixture was stirred at room temperature for 4 h

before concentration in vacuo. The crude product was purified by

silica gel chromatography with a gradient of MeOH in CH₂Cl₂(0 to

The crude product was purified by silica gel chromatography with

a gradient of MeOH in CH₂Cl₂(0 to 3%) to give 1r (50.9 mg,

0.17 mmol). ¹H NMR (500 MHz, CDCl₃): d=2.02–2.25 (m, 2H), 2.64

(t, J=8.1 Hz, 2H), 3.84 (t, J=7.2 Hz, 2H), 5.43 (s, 2H), 8.05–

8.36 ppm (m, 4H); ¹³C NMR (500 MHz, CDCl₃): d=17.91, 32.85,

44.87, 65.40, 123.57, 131.06, 134.85, 150.72, 164.29, 167.34,

176.20 ppm; MS: m/z: 315.2 [M+Na]⁺; HRMS: m/z calcd for

C₁₃H₁₂O₆N₂Na [M+Na]⁺: 315.0588; found: 315.0594.

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl

quinoline-6-carboxylate

1

5%) to give 1u (108 mg, 0.34 mmol). H NMR (500 MHz, CDCl₃): d=

(1ad): Compound 1ad was prepared according to the procedure

in General Method A from 1v and quinolone-6-carboxylic acid on a

1.95–2.13 (m, 2H), 2.59 (t, J=8.0 Hz, 2H), 3.40 (t, J=6.2 Hz, 2H),

3.81 (t, J=7.1 Hz, 2H), 4.69 (t, J=6.2 Hz, 2H), 8.04–8.22 (m, 2H),

8.63 (s, 1H), 9.13 ppm (s, 1H); ¹³C NMR (500 MHz, CDCl₃): d=17.20,

33.50, 36.38, 45.34, 60.21, 123.31, 124.27, 127.32, 127.38, 133.70,

156.01, 157.42, 165.79, 170.91, 175.62 ppm; MS: m/z: 341.3

[M+Na]⁺; HRMS: m/z calcd for C₁₅H₁₄N₂NaO₄S [M+Na]⁺: 341.0566;

found: 341.0562.

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl benzoate (1h): Compound

1h was prepared according to the procedure in General Method A

from 1v and benzoic acid on

0.35 mmol scale. The crude product

was purified by silica gel chromatogra-

a

0.36 mmol scale. The crude product was purified by silica gel chro-

matography with a gradient of MeOH in CH₂Cl₂(0 to 10%) to give

1ad (26.6 mg, 0.089 mmol). ¹H NMR (500 MHz, CDCl₃): d=2.09–

2.32 (m, 2H), 2.66 (t, J=8.1 Hz, 2H), 3.88 (t, J=7.2 Hz, 2H), 5.49 (s,

2H), 7.42–7.63 (m, 1H), 8.18 (d, J=8.8 Hz, 1H), 8.29 (d, J=7.6 Hz,

1H), 8.38 (dd, J=8.8, 1.8 Hz, 1H), 8.70 (d, J=1.4 Hz, 1H), 8.97–

9.16 ppm (m, 1H); ¹³C NMR (500 MHz, CDCl₃): d=17.91, 32.91,

44.88, 65.10, 121.87, 127.42, 127.50, 129.13, 129.90, 131.52, 137.38,

150.28, 152.63, 165.64, 167.85, 176.04 ppm; MS: m/z: 299.2

[M+H]⁺; HRMS: m/z calcd for C₁₆H₁₄N₂NaO₄[M+Na]⁺: 321.0846;

found: 321.0846.

phy with a gradient of MeOH in CH₂Cl₂

(0 to 3%) to give 1h (52.1 mg,

0.21 mmol). ¹H NMR (500 MHz, CDCl₃):

d=2.02–2.18 (m, 2H), 2.61 (t, J=8.1 Hz,

2H), 3.83 (t, J=7.2 Hz, 2H), 5.40 (s, 2H),

7.46 (t, J=7.7 Hz, 2H), 7.58 (t, J=

7.4 Hz, 1H), 8.11 ppm (d, J=8.3 Hz, 2H); ¹³C NMR (500 MHz, CDCl₃):

d=17.87, 32.90, 44.86, 64.82, 128.40, 129.48, 129.92, 133.28, 166.11,

167.96, 176.05 ppm; MS: m/z: 270.3 [M+Na]⁺; HRMS: m/z calcd for

C₁₃H₁₃NNaO₄[M+Na]⁺: 270.0742; found: 270.0741.

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl

biphenyl-4-carboxylate

(1ae): Compound 1ae was prepared according to the procedure

in General Method A from 1v and biphenyl-4-carboxylic acid on a

2-Oxo-2-(2-oxopyrrolidin-1-yl)ethyl 4-(dimethylamino)benzoate

(1q): Compound 1q was prepared according to the procedure in

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

13

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

&

These are not the final page numbers! ÞÞ

Full Papers

[7] P. Aldridge, R. Paul, P. Goymer, P. Rainey, U. Jenal, Mol. Microbiol. 2003 ,

47 , 169 5 –1 708.

0.36 mmol scale. The crude product was purified by silica gel chro-

matography with a gradient of MeOH in CH₂Cl₂(0 to 3%) to give

1ae (49.2 mg, 0.15 mmol). H NMR (500 MHz, CDCl₃): d=1.82–1.99

(m, 2H), 2.40 (t, J=8.1 Hz, 2H), 3.62 (t, J=7.2 Hz, 2H), 5.07 (s, 2H),

7.17 (t, J=7.3 Hz, 1H), 7.25 (t, J=7.5 Hz, 2H), 7.41 (d, J=7.4 Hz,

2H), 7.46 (d, J=8.3 Hz, 2H), 7.96 ppm (d, J=8.3 Hz, 2H); ¹³C NMR

(500 MHz, CDCl₃): d=18.13, 33.17, 45.11, 65.10, 127.32, 127.54,

128.44, 129.18, 130.72, 140.18, 146.21, 166.23, 168.23, 176.33 ppm;

MS: m/z: 346.3 [M+Na]⁺; HRMS: m/z calcd for C₁₉H₁₇NNaO₄

[M+Na]⁺: 346.1050; found: 346.1052.

1

[8] C. Chan, R. Paul, D. Samoray, N. C. Amiot, B. Giese, U. Jenal, T. Schirmer,

Proc. Natl. Acad. Sci. USA 2004 , 101 , 1708 4 –1 7089.

[9] B. Christen, M. Christen, R. Paul, F. Schmid, M. Folcher, P. Jenoe, M.

Meuwly, U. Jenal, J. Biol. Chem. 2006 , 281 , 3201 5 –3 2024.

[10] M. Christen, H. D. Kulasekara, B. Christen, B. R. Kulasekara, L. R. Hoffman,

S. I. Miller, Science 2010 , 328 , 129 5 –1 297.

[11] B. R. Kulasekara, C. Kamischke, H. D. Kulasekara, M. Christen, P. A. Wig-

gins, S. I. Miller, eLife 2013, 2, e01402.

[12] E. Mills, E. Petersen, B. R. Kulasekara, S. I. Miller, Sci. Signaling 2015 , 8 ,

ra57.

[13] R. Paul, S. Weiser, N. C. Amiot, C. Chan, T. Schirmer, B. Giese, U. Jenal,

Genes Dev. 2004 , 18 , 71 5 –7 27.

[14] M. Folcher, S. Oesterle, K. Zwicky, T. Thekkottil, J. Heymoz, M. Hohmann,

M. Christen, M. Daoud El-Baba, P. Buchmann, M. Fussenegger, Nat.

Commun. 2014, 5, 5392.

[15] I. S. Pultz, M. Christen, H. D. Kulasekara, A. Kennard, B. Kulasekara, S. I.

Miller, Mol. Microbiol. 2012 , 86 , 142 4 –1 440.

[16] C. A. Kellenberger, S. C. Wilson, J. Sales-Lee, M. C. Hammond, J. Am.

Chem. Soc. 2013 , 135 , 490 6 –4 909.

[17] C. A. Kellenberger, J. Sales-Lee, Y. Pan, M. M. Gassaway, A. E. Herr, M. C.

Hammond, RNA Biol. 2015 , 12 , 118 9 –1 197.

[18] S. M. Ching, W. J. Tan, K. L. Chua, Y. Lam, Bioorg. Med. Chem. 2010 , 18 ,

665 7 –6 665.

[19] T. Kline, S. R. Jackson, W. Deng, C. L. M. J. Verlinde, S. I. Miller, Nucleosides

Nucleotides Nucleic Acids 2008 , 27 , 128 2 –1 300.

[20] D. L. Caly, D. Bellini, M. A. Walsh, J. M. Dow, R. P. Ryan, Curr. Pharm. Des.

2015, 21, 12–24.

Acknowledgements

Compounds for the HTS were provided by the National Screening

Laboratory for the regional centers of excellence in biodefense

and emerging infectious diseases (NIAID U54 AI057159). We

thank Su Chaing and the staff at the NERCE National Screening

Laboratory (NSRB) for assistance in performing the HTS and post-

screen analysis, as well as Jonathan Venetz for comments on the

manuscript. This research was supported by the ETH Zꢀrich and

by the National Institute of Allergy and Infectious Diseases (NIH,

1R21NS067579-01 and U54AI057141 to S.I.M, 1R21NS067579-01

to H.D.K, NSF Graduate Research Fellowship to B.R.K), the Swiss

National Foundation (PA00P3-124140 and PBBSA-120489 to M.C.

and PA00P3-126243 to B.C.), and the Novartis Foundation to M.C.

[21] P. Ohana, D. P. Delmer, R. W. Carlson, J. Glushka, P. Azadi, T. Bacic, M.

Benziman, Plant Cell Physiol. 1998 , 39 , 14 4 –1 52.

[22] O. J. Lieberman, M. W. Orr, Y. Wang, V. T. Lee, ACS Chem. Biol. 2014 , 9 ,

18 3 –1 92.

[23] K. Sambanthamoorthy, R. E. Sloup, V. Parashar, J. M. Smith, E. E. Kim,

M. F. Semmelhack, M. B. Neiditch, C. M. Waters, Antimicrob. Agents Che-

mother. 2012 , 56 , 520 2 –5 211.

Conflict of Interest

The authors declare no conflict of interest.

[24] K. Sambanthamoorthy, C. Luo, N. Pattabiraman, X. Feng, B. Koestler,

C. M. Waters, T. J. Palys, Biofouling 2014 , 30 , 1 7 –2 8.

[25] H. Weber, C. Pesavento, A. Possling, G. Tischendorf, R. Hengge, Mol. Mi-

crobiol. 2006 , 62 , 101 4 –1 034.

[26] R. R. Spurbeck, R. J. Tarrien, H. L. T. Mobley, mBio 2012, 3, e00307-12-

e00307-12.

[27] C. A. Lipinski, J. Pharmacol. Toxicol. Methods 2000 , 44 , 23 5 –2 49.

[28] T. Ohashi, S. D. Galiacy, G. Briscoe, H. P. Erickson, Protein Sci. 2007 , 16 ,

142 9 –1 438.

Keywords: c-di-GMP · diguanylate cyclase inhibitors · FRET ·

high-throughput screening · structure–activity relationships

[1] L. R. Hoffman, D. A. D’Argenio, M. J. MacCoss, Z. Zhang, R. A. Jones, S. I.

Miller, Nature 2005 , 436 , 117 1 –1 175.

[2] E. Mills, I. S. Pultz, H. D. Kulasekara, S. I. Miller, Cell. Microbiol. 2011 , 13 ,

112 2 –1 129.

[3] D.-G. Ha, G. A. O’Toole, Microbiol. Spectrum 2015, 3, MB-0003–2014.

[4] M. J. Bush, N. Tschowri, S. Schlimpert, K. Flꢂrdh, M. J. Buttner, Nat. Rev.

Microbiol. 2015 , 13 , 749.

[5] S. Ozaki, A. Schalch-Moser, L. Zumthor, P. Manfredi, A. Ebbensgaard, T.

Schirmer, U. Jenal, Mol. Microbiol. 2014 , 94 , 58 0 –5 94.

[6] N. Ausmees, R. Mayer, H. Weinhouse, G. Volman, D. Amikam, M. Benzi-

man, M. Lindberg, FEMS Microbiol. Lett. 2001 , 204 , 16 3 –1 67.

Manuscript received: October 4, 2018

Accepted manuscript online: November 5, 2018

Version of record online: && &&, 0000

&

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

14

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ÝÝ These are not the final page numbers!

FULL PAPERS

Controling biofilm formation: Diguany-

late cyclases (DGCs) are attractive anti-

invective drug targets for control of bio-

film formation. We report a FRET-based

HTS assay coupled with detailed struc-

ture–activity studies to identify modula-

tors of DGCs. Detailed mechanism-of-

action studies demonstrate the complex

interplay between the synthetic com-

pounds and the regulatory mechanisms

that control the activity of DGCs.

M. Christen,* C. Kamischke,

H. D. Kulasekara, K. C. Olivas,

B. R. Kulasekara, B. Christen, T. Kline,

S. I. Miller

&& – &&

Identification of Small-Molecule

Modulators of Diguanylate Cyclase by

FRET-Based High-Throughput

Screening

ChemBioChem 2018, 19, 1 – 15

www.chembiochem.org

15

ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

&

These are not the final page numbers! ÞÞ

Article Doi

Identification of Small-Molecule Modulators of Diguanylate Cyclase by FRET-Based High-Throughput Screening

DOI: 10.1002/cbic.201800593

Source and publish data:

Authors:

Article abstract of DOI:10.1002/cbic.201800593

Full text of DOI:10.1002/cbic.201800593

Products guided by the article

R&D Labs maybe for 4296-47-3

Relevant to this article

Hot Product