Benzoflavone activators of the cystic fibrosis transmembrane conductance regulator: Towards a pharmacophore model for the nucleotide-binding domain

Article abstract of DOI:10.1016/S0968-0896(03)00435-8

Our previous screen of flavones and related heterocycles for the ability to activate the cystic fibrosis transmembrane conductance regulator (CFTR) chloride channel indicated that UCCF-029, a 7,8-benzoflavone, was a potent activator. In the present study, we describe the synthesis and evaluation, using cell-based assays, of a series of benzoflavone analogues to examine structure-activity relationships and to identify compounds having greater potency for activation of both wild type CFTR and a mutant CFTR (G551D-CFTR) that causes cystic fibrosis in some human subjects. Using UCCF-029 as a structural guide, a panel of 77 flavonoid analogues was prepared. Analysis of the panel in FRT cells indicated that benzannulation of the flavone A-ring at the 7,8-position greatly improved compound activity and potency for several flavonoids. Incorporation of a B-ring pyridyl nitrogen either at the 3- or 4-position also elevated CFTR activity, but the influence of this structural modification was not as uniform as the influence of benzannulation. The most potent new analogue, UCCF-339, activated wild-type CFTR with a K_d of 1.7 μM, which is more active than the previous most potent flavonoid activator of CFTR, apigenin. Several compounds in the benzoflavone panel also activated G551D-CFTR, but none were as active as apigenin. Pharmacophore modeling suggests a common binding mode for the flavones and other known CFTR activators at one of the nucleotide-binding sites, allowing for the rational development of more potent flavone analogues.

Full text of DOI:10.1016/S0968-0896(03)00435-8

Bioorganic & Medicinal Chemistry 11 (2003) 4113–4120
Benzoflavone Activators of the Cystic Fibrosis Transmembrane
Conductance Regulator: Towards a Pharmacophore Model
for the Nucleotide-Binding Domain
Mark F. Springsteel,a Luis J. V. Galietta,b Tonghui Ma,c Kolbot By,a Gideon O. Berger,a
Hong Yang,c Christopher W. Dicus,a Wonken Choung,a Chao Quan,a Anang A. Shelat,e
R. Kiplin Guy,d A. S. Verkman,c Mark J. Kurtha,* and Michael H. Nantza,*
aDepartment of Chemistry, University of California, Davis, CA 95616, USA
bLaboratorio di Genetica Molecolare, Istituto Giannina Gaslini, 16148 Genova, Italy
cDepartments of Medicine and Physiology, Cardiovascular Research Institute, University of California,
San Francisco, CA 94143, USA
dDepartments of Pharmaceutical Chemistry and Cellular and Molecular Pharmacology, University of California,
San Francisco, CA 94143, USA
eChemistry and Chemical Biology Program, University of California, San Francisco, CA 94143, USA
Received 4 February 2003; accepted 22 April 2003
Abstract—Our previous screen of flavones and related heterocycles for the ability to activate the cystic fibrosis transmembrane
conductance regulator (CFTR) chloride channel indicated that UCCF-029, a 7,8-benzoflavone, was a potent activator. In the present
study, we describe the synthesis and evaluation, using cell-based assays, of a series of benzoflavone analogues to examine
structure–activity relationships and to identify compounds having greater potency for activation of both wild type CFTR and a
mutant CFTR (G551D-CFTR) that causes cystic fibrosis in some human subjects. Using UCCF-029 as a structural guide, a panel
of 77 flavonoid analogues was prepared. Analysis of the panel in FRT cells indicated that benzannulation of the flavone A-ring at
the 7,8-position greatly improved compound activity and potency for several flavonoids. Incorporation of a B-ring pyridyl nitrogen
either at the 3- or 4-position also elevated CFTR activity, but the influence of this structural modification was not as uniform as the
influence of benzannulation. The most potent new analogue, UCCF-339, activated wild-type CFTR with a Kd of 1.7 mM, which is
more active than the previous most potent flavonoid activator of CFTR, apigenin. Several compounds in the benzoflavone panel
also activated G551D-CFTR, but none were as active as apigenin. Pharmacophore modeling suggests a common binding mode for
the flavones and other known CFTR activators at one of the nucleotide-binding sites, allowing for the rational development of
more potent flavone analogues.
# 2003 Elsevier Ltd. All rights reserved.
Introduction
patients,3,4 no clinically acceptable therapy has emerged.5
Consequently, the identification of small molecule
CFTR activators is an important goal of CF research,
reasoning that CFTR activators may correct the
impaired chloride transport in cells expressing mutant
CFTRs in CF patients.6,7
Cystic fibrosis (CF) is the most common lethal genetic
disease among Caucasians afflicting one in 2000–2500
Caucasians at birth.1 CF is caused by mutations in the
CF transmembrane conductance regulator (CFTR)
protein resulting in chloride-impermeable epithelial cells
in the airways, pancreas, intestine and other organs.2
Although several novel approaches including gene ther-
apy have attempted to improve CFTR activity in CF
We recently reported the synthesis and cell-based assay
of a series of analogues of known CFTR activators.8
Chloride conductance was measured using a fluores-
cence assay in which cells expressing a halide-sensitive
green fluorescent protein analogue together with CFTR
were treated with test compounds and subjected to halide
gradients.9 We found that the analogue UCCF-029, a
*Corresponding authors. Tel.: +1-530-752-6357; fax: +1-530-752-8995;
e-mail: mhnantz@ucdavis.edu (M. H. Nantz); Tel.: +1-530-752-8192;
fax: +1-520-752-8995 e-mail: mjkurth@ucdavis.edu (M. J. Kurth).
0968-0896/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0968-0896(03)00435-8

4114
M. F. Springsteel et al. / Bioorg. Med. Chem. 11 (2003) 4113–4120
7,8-benzoflavone, was ꢀ10-fold more potent than
existing flavones in activating wild type CFTR. Sec-
ondary analyses using UCCF-029 indicated that CFTR
activation occurred without elevation of cellular cAMP
or inhibition of cell phosphatase activity, suggesting a
direct interaction with CFTR.
Four of the eight aryl ketone building blocks used in
this study were not commercially available and
required synthesis. Aryl ketone 1-(5-hydroxy-2,3-dihydro-
benzo[1,4]dioxin-6-yl)-ethanone was prepared from
gallacetophenone using a straightforward alkylation
approach.13 A carboxylic acid-to-methyl ketone trans-
formation furnished 20-hydroxy-30-acetonaphthone from
commercially available 3-hydroxy-2-napthoic acid.14,15
Bromination of 10-hydroxy-20-acetonaphthone accord-
ing to a literature procedure provided the corresponding
40-bromo-10-hydroxy-20-acetonaphthone derivative.16,17
Following a literature procedure, reduction of p-naph-
thoquinone, acetylation and Lewis-acid induced
Fries rearrangement gave 40-ethoxy-10-hydroxy-20-aceto-
naphthone.18 All other aryl ketones used to prepare the
panel of analogues were purchased from the Aldrich
Chemical Company (Milwaukee, WI, USA).
The purpose of the present study was to establish
structure–activity relationships for the benzoflavone
class of flavonoids, with a goal of identifying com-
pounds having improved potency in the activation of
wild type and G551D-CFTR. We selected the G551D-
CFTR mutant since, to our knowledge, all reported
activators of G551D-CFTR are either flavone- or iso-
flavone-derived compounds. We systematically modified
the UCCF-029 structure by varying the position of the
A-ring benzannulation and the B-ring pyridyl-nitrogen
position. We also explored the influence of substitutent
groups at several A- and B-ring positions. Compounds
were screened by our previously reported fluorescence
cell-based assay, and the best activators were subjected
to dose–response analysis and short-circuit chloride
current measurements. The most active analogues were
then used to generate a four-point common features phar-
macophore model using the HipHop program (Accelrys,
Inc). Alignment of this model with other known CFTR
agonists suggests a common binding mode at one of the
nucleotide binding pockets of CFTR.
The 6-bromo-flavone analogues were further diversified
using the Suzuki coupling procedure.19 A representative
synthesis is given in Scheme 2.20,21 The wide variety of
commercially available arylboronic acids makes the
Suzuki approach a convenient route for synthesis of
novel aryl-substituted flavones.
Pharmacology
Tables 1 and 2 and Figures 1–3 summarize the data
obtained for the analogues examined in our study. Fla-
vones are referred to by UCCF-OOO (University of
California Cystic Fibrosis—compound number) desig-
nations when prepared and screened in our labs. Each
compound was screened for activation of CFTR in tri-
plicate at 50 mM together with a low concentration of
the cAMP agonist forskolin which itself produced little
CFTR activation. Apigenin and 3-isobutyl-1-methyl-
xanthine (IBMX) also were included as positive controls.
Apigenin22 and IBMX7,23 are well-recognized, potent
CFTR activators. Phosphate-buffered saline (PBS), api-
genin and IBMX CFTR-activating potencies were
expressed as slopes of fluorescence versus time curves. The
PBS, apigenin and IBMX controls gave maximal slopes
of fluorescence change of 11, 40 and 37, respectively.
Results and Discussion
Compound synthesis
All flavones were synthesized according to the Baker–
Venkataraman (B-V) synthesis.11,12 A representative
synthesis is depicted in Scheme 1. Typically, esterifica-
tion of o-hydroxyacetophenones (e.g., 1) is followed by
treatment with base to promote an intramolecular
mixed-Claisen rearrangement which furnishes a dike-
tone (e.g., 4). In the final step, the flavone C-ring is
established via an acid-induced cyclodehydration. All
flavones were submitted for testing as 1.0-mM stock
solutions in DMSO. Although some required initial
warming to dissolve, each compound was soluble at
room temperature at this concentration.
With UCCF-029 and its conjugate base UCCF-275 as
leads, we sought to determine the importance of the two
unique structural features: (i) the B-ring 40-nitrogen and
(ii) benzene ring annulation of the A-ring at its 7,8-
positions. Furthermore, analogues devoid of one or
both of these structural features (e.g., benzoflavones
without a B-ring nitrogen) were compared in the matrix.
Although UCCF-029 was originally prepared and eval-
uated as the pyridinium bisulfate salt, we have since
Scheme 1. Representative synthesis. Reagents and conditions: (a) 2 (1.2
equiv), pyridine, 0 ꢁC to rt, 2 h, 63%; (b) pulverized KOH(1.4 equiv),
pyridine, 40ꢁC, 1 h; (c) 4% H2SO4 in AcOH, reflux, 1.5 h, 38% (from 3).
Scheme 2. Representative Suzuki coupling. Reagents and conditions:
(a) PhB(OH)2 (2.1 equiv), 2 M Na2CO3 (6 equiv), Pd(Ph3P)4 (2.0
mol%), 1:1 EtOH/PhMe, 90 ꢁC, 2 h, 70%.

M. F. Springsteel et al. / Bioorg. Med. Chem. 11 (2003) 4113–4120
Table 1. Influence of pyridyl and benzo featuresa
4115
Position of benzo ring
No ring
X=Y=Z=CHX=N/Y=Z=CHY=N/X=Z=CHZ=N/X=Y=CH
Flavone
27
UCCF-315
17
UCCF-324
Toxic
UCCF-344
18
UCCF-343
UCCF-284
19
UCCF-345
10
23
UCCF-289
6.0
UCCF-301
16
UCCF-275b
39
5,6-Benzo
6,7-Benzo
7,8-Benzo
UCCF-304
22
UCCF-346
15
UCCF-303
22
UCCF-023
33
UCCF-347
21
UCCF-279
37
aEach cell contains the compound identity number and its corresponding CFTR activity data (reported as initial slope and derived as explained in
Experimental, Biology).
bConjugate base of UCCF-029.
determined that the free base (UCCF-275) has identical
CFTR activity. Consequently, all pyridyl analogues
were evaluated as neutral compounds except where
specifically derivatized as alkyl pyridinium salts.
analogues may be expected to influence their transport
across biological membranes, and this may account for
the lower activity relative to UCCF-029.
Table 2b features flavone and benzoflavone analogues
that do not have a pryidyl B-ring. With the exception of
the fluoro-isomers, annulation of the A-ring at the 7,8-
benzo ring appears to improve activity relative to the
benzo-free analogues. None of the non-pyridyl analo-
gues displayed activity comparable to apigenin. Inter-
estingly, the most active analogues in this series were the
30-F analogue (UCCF-017) and the 40-F analogues
(UCCF-018 and UCCF-349). Since fluorine is a H-bond
acceptor,25 these results are consistent with the obser-
vations that the presence of a H-bond acceptor—such as
a pyridyl nitrogen—in the vicinity of the 40-position
improves activity.22 When the 7,8-benzo ring is replaced
with a non-aromatic ethylenedioxo ring fused to the
same position, the activity is greatly reduced.
Results of the isomer study are summarized in Table 1.
Incorporation of a B-ring nitrogen appears to only
minimally activate CFTR by comparison to the corre-
sponding benzo isomer devoid of nitrogen. The influ-
ence of the benzannulation modification is more
pronounced. Within a given pyridyl series (columns),
the most active member of the series is clearly the 7,8-
benzo isomer. The greatest distinctions are noted for the
30- and 40-pyridyl series where the addition of a 7,8-
benzo ring dramatically improves activity. The activities
of these two compounds, UCCF-275 and-279, at the 50-
mM assay concentration (Table 1 conditions) are com-
parable to the potent CFTR activator apigenin, which is
roughly 10-fold more potent than genistein.24
Table 2a features additional pyridyl B-ring flavones.
The assay data once again reflect the greater influence of
the 7,8-benzannulation feature relative to the pyridyl
feature in CFTR activation. In this panel, the strongest
activators possess a 7,8-benzo ring. Pyridyl analogues
without the 7,8-benzo structural feature displayed only
minimal CFTR activation (i.e., maximal slope <20).
The most active compounds in this panel of analogues
contained a 2-furyl substituent at the 6-position. As
noted with other substitutent series, the activation
afforded by the 2-furyl group depends on the presence
of the 7,8-benzo ring as well as a pyridyl nitrogen in the
B-ring—omission of either of these two features results
in lower activities for the 6-(2-furyl) analogues (e.g.,
compare UCCF-339 vs UCCF-324). Two B-ring 40-
alkylpyridinium salts were prepared and tested to see if
a non-titratable positive charge at this position would
enhance activity. The data in Table 2a shows that
alkylation of the pyridyl ring greatly diminishes CFTR
activation (e.g., compare UCCF-029 vs UCCF-340 or
UCCF-341). The non-titratable positive charge in these
To confirm whether the principal hits in our initial 50
mM fluorescence assay were sufficiently positive to war-
rant further investigation, we conducted a dose–
response study on the most active compounds using
both the aforementioned fluorescence assay and an
electrophysiological assay (short-circuit current analy-
sis). We performed dose–response studies on apigenin
and analogues UCCF-275, -281, -339, -349 and -353.
UCCF-281, -349 and -353 had no significant activity at
the 10 mM dose (data not shown). However, UCCF-275
and UCCF-339 were active at the lower doses. For
example, the original plate reader traces for UCCF-339
are depicted in Figure 1a and the corresponding dose–
response curve is plotted in Figure 1b. The fitted Kd was
4.8 mM for UCCF-275, similar to that of apigenin, and
1.7 mM for UCCF-339 (Fig. 1c). Ussing chamber
experiments showed that UCCF-339 evoked large
transepithelial Clꢂ currents that could be abolished by
glibenclamide, a CFTR blocker (Fig. 2a). This benzo-
flavone was considerably more efficacious than apigenin
at similar concentrations (Fig. 2b). Interestingly, in

4116
M. F. Springsteel et al. / Bioorg. Med. Chem. 11 (2003) 4113–4120
Table 2. (a) Pyridyl-containing analogues; (b) nitrogen-free analogues
contrast to apigenin, UCCF-339 did not activate the
mutant G551D CFTR (Fig. 2c).
Common features pharmacophore models were derived
for all flavones and for the most active (top 10%) fla-
vones. The model generated from all flavones (Fig. 3a)
is a subset of the four-point model describing the most
active flavones (Fig. 3b). The four-point model has the
following features: a hydrogen bond acceptor (A), an
aromatic ring (B), and two hydrophobic groups (C and
D). The inter-atomic distances between the center-of-
mass coordinates of each pharmacophore point are as
follows: A–B (3.0 A), B–C (5.4 A), B–D (3.2 A). The
angles between the pharmacophore points are: A–B–C
(88.9ꢁ), A–B–D (159.3ꢁ), C–B–D (111.8ꢁ). The improper
torsion angle, A–B–D–C, is 180.0ꢁ, indicating a planar
geometry. The additional hydrophobic pharmacophore
feature present in the model derived from the most
active flavones (point D) reflects the increase in activity
due to benzannulation at the 7,8-position.
UCCF-
A-Ring substituent
B-Ring modification
Slopea
320
321
322
324
323
296
291
327
326
330
329
333
332
339
338
336
335
350
353
311
029
340
341
6-Phenyl
6-(2-Thienyl)
6-(3-Thienyl)
6-(2-Furyl)
6-(2-Benzofuryl)
Y=CH, Z=N
Y=CH, Z=N
Y=CH, Z=N
Y=CH, Z=N
Y=CH, Z=N
Y=CH, Z=N
Y=N, Z=CH24
Y=CH, Z=N
Y=N, Z=CH14
Y=CH, Z=N
Y=N, Z=CH29
Y=CH, Z=N
Y=N, Z=CH23
Y=CH, Z=N
Y=N, Z=CH33
Y=CH, Z=N
Y=N, Z=CH18
Y=CH, Z=N
Y=CH, Z=N
Y=CH, Z=N
15
17
15
19
15
16
7,8-Benzo-6-bromo
7,8-Benzo-6-bromo
7,8-Benzo-6-phenyl
7,8-Benzo-6-phenyl
7,8-Benzo-6-(2-thienyl)
7,8-Benzo-6-(2-thienyl)
7,8-Benzo-6-(3-thienyl)
7,8-Benzo-6-(3-thienyl)
7,8-Benzo-6-(2-furyl)
7,8-Benzo-6-(2-furyl)
7,8-Benzo-6-(2-benzofuryl)
7,8-Benzo-6-(2-benzofuryl)
7,8-Benzo-6-ethoxy
7,8-Benzo-6-hydroxy
7,8-Ethylenedioxo
7,8-Benzo
˚
˚
˚
12
22
18
38
16
16
22
Toxicb
38
14
We then applied our model to analysis of other known
CFTR agonists. Recently, Ma et al.10 reported a novel
set of CFTR agonists with diverse scaffolds. Three of
the 17 most active compounds were excluded from our
study because they appeared to activate CFTR indir-
ectly by elevating cellular cAMP. The remaining 14
strongly active compounds from the Ma report were
overlaid on the four point pharmacophore model and
11 of the 14 compounds aligned well to all four points
of the flavone pharmacophore (see Fig. 3c for examples
of the alignments). Three compounds, CFTRact-05,
CFTRact-09, and CFTRact-13, failed to map the correct
chemical functionalities to each of the four pharmaco-
phore points. Furthermore, the flavone pharmacophore
overlaps nicely with known CFTR agonists such as
adenosine and the xanthine CPX. Figure 3d shows the
alignment of these two compounds: the xanthine maps
well to all points, whereas the heterocylic ring and sugar
of adenosine occupy three of the four pharmacophore
points. These compounds have been shown to bind the
nucleotide-binding pocket of CFTR,26 suggesting that
the flavones and some of the previously identified
CFTR agonists share a common binding mode at this
site. Indeed, flavonoids have been shown to bind the
ATP-binding domains of other biologically important
molecules, including the cyclin-dependent kinase 2,27
mouse P-glycoprotein28 and a haematopoietic cell
kinase.29 Further understanding of the interactions at
the nucleotide-binding domain of CFTR could enable
design of small molecule effectors targeting pathophys-
iologic CFTR mutations.
.
Y=CH, Z=N H2SO4
Y=CH, Z=NCH3I
Y=CH, Z=NCH2CH3I
7,8-Benzo
7,8-Benzo
14
UCCF-
A-Ring substituent
R0
Slopea
016
017
018
313
314
290
316
044
268
040
317
318
300
319
352
025
282
281
271
355
269
293
294
292
297
325
328
331
337
334
351
348
349
307
308
309
310
305
306
None
None
None
None
None
5,6-Benzo
5,6-Benzo
5,6-Benzo
5,6-Benzo
5,6-Benzo
6,7-Benzo
6,7-Benzo
6,7-Benzo
6,7-Benzo
6,7-Benzo
7,8-Benzo
7,8-Benzo
20-F
30-F
40-F
25
31
33
14
18
19
19
19
18
17
Toxicb
13
17
40-Br
40-OCH3
20-F
30-F
40-F
40-Br
40-OCH3
20-F
30-F
40-F
40-Br
40-OCH3
20-F
14
Toxicb
25
26
30-F
40-F
7,8-Benzo
7,8-Benzo
7,8-Benzo
7,8-Benzo
23
20
40-Br
40-OH15
40-OCH3
20-F
24
18
14
17
7,8-Benzo-6-bromo
7,8-Benzo-6-bromo
7,8-Benzo-6-bromo
7,8-Benzo-6-bromo
7,8-Benzo-6-phenyl
7,8-Benzo-6-(2-thienyl)
7,8-Benzo-6-(3-thienyl)
7,8-Benzo-6-(2-furyl)
7,8-Benzo-6-(2-benzofuryl)
7,8-Benzo-6-ethoxy
7,8-Benzo-6-ethoxy
7,8-Benzo-6-hydroxy
7,8-Ethylenedioxo
7,8-Ethylenedioxo
7,8-Ethylenedioxo
7,8-Ethylenedioxo
7,8-Ethylenedioxo
7,8-Ethylenedioxo
40-F
40-OCH3
H21
20-F
22
29
17
27
16
14
16
33
24
23
19
14
17
20-F
20-F
20-F
20-F
30-F
Conclusion
40-F
40-F
We prepared and screened a panel of benzoflavones for
activation of wild type CFTR. The structures were
selected to probe the relevance of A-ring benzannula-
tion and a B-ring pyridyl-nitrogen, both structural fea-
tures that were identified by our previous screening of
MPB-07–genistein hybrid analogues. The data shows
that benzannulation at the 7,8-positions of the flavone
20-F
30-F
40-F
40-I
40-OCH3
H21
aInitial slope, derived as explained in Experimental, Biology.
bCells detached during the assay.

M. F. Springsteel et al. / Bioorg. Med. Chem. 11 (2003) 4113–4120
4117
Figure 1. CFTR activation by UCCF-339 and apigenin: (a) representative fluorescence traces showing CFTR-dependent Iꢂ influx by UCCF-339 at
different concentrations and by apigenin (at 10 mM); (b) corresponding dose–response relationships for UCCF-339 and apigenin, data fitted with the
Hill equation; (c) most active compounds with respective Kd in mM.
Experimental
scaffold significantly improves the activation of wild-
type CFTR. The presence of a rigid p-system in this
domain is uniformly superior to benzannulation at
other A-ring positions. The pyridyl B-ring appears less
influential in activating CFTR than the 7,8-benzannu-
lation. Furthermore, modifications of the flavone B-ring
with a pyridyl-nitrogen or fluorine substituent at posi-
tions 30 or 40 are equally well tolerated. Two benzo-
flavone analogues displayed activity comparable to
apigenin and better than IBMX at the 50-mM concen-
tration. Dose–response data in both fluorescence and
electrophysiological studies confirm that UCCF-339 has
4-fold higher maximal activity than UCCF-029 and
apigenin, the most potent flavonoid activators of wild-
type CFTR. Computational analysis of the SAR data
led to a consensus common features pharmacophore
model that provides an emerging picture of the binding
site for activators of CFTR.
1
General synthesis methods. Hand spectra were recor-
ded in CDCl3 on a Mercury (300 MHz) and an Inova
(400 MHz) spectrometer with TMS as an internal stan-
dard. Infrared spectra of neat samples were recorded on
a Mattson Genesis II FTIR. Melting points were deter-
mined using an Electrothermal digital melting point
apparatus and are uncorrected. Elemental analyses were
performed by Midwest Microlabs, Indianapolis, IN,
USA. Flash column chromatography was carried out
using EM Science silica gel 60 (230–400 mesh). TLC was
performed on Merk silica gel 60 F254 plates and visua-
lized under a UV lamp. THF was distilled from sodium
benzophenone ketyl immediately prior to use, CH2Cl2
was distilled from calcium hydride immediately prior to
use and all other reagents were used as purchased from
Aldrich, Acros and Fisher.

4118
M. F. Springsteel et al. / Bioorg. Med. Chem. 11 (2003) 4113–4120
Figure 2. Measurement of CFTR-dependent Clꢂ transport by short-circuit current experiments. (a,b) Effect of increasing concentrations of UCCF-
339 or apigenin on FRT cells expressing wild-type CFTR. Forskolin was applied at a low concentration to provide a minimal CFTR phosphoryl-
ation. (c) Stimulation of FRT cells expressing G551D-CFTR by high forskolin concentration followed by UCCF-339 and apigenin.
Representative B-V synthesis. Step 1: 2-acetyl-4-bromo-
naphthyl-pyridine-4-carboxylate (3). To a solution of 2-
acetyl-4-bromo-1-naphthol (2.02 g, 7.62 mmol) in pyri-
dine (40 mL) at 0 ꢁC was added isonicotinoyl chloride
HCl (1.72 g, 9.18 mmol) in one portion. After 5 min, the
reaction was warmed to room temperature and stirred
an additional 2 h. The reaction was quenched by addi-
tion of a 1:1 mixture of 5% aq HCl/crushed ice (40 mL).
The resultant precipitate was collected by filtration and
washed with ice cold water. The precipitate was then
dissolved in EtOAc (100 mL) and the solution was
washed with brine (20 mL) and dried (Na2SO4). The
solvents were removed and the residue was purified by
chromatography with gradient elution (10–75% EtOAc
in CH2Cl2) to obtain ester 3 (1.77 g, 63%) as a white
solid; mp 150–151 ꢁC; TLC, Rf (EtOAc) 0.46; IR 3041,
Steps 2+3: 6-bromo-2-(4-pyridyl)benzo[h]4H-chromen-
4-one (UCCF-296). To a solution of 2-acetyl-4-bromo-
naphthylpyridine-4-carboxylate (1.63 g, 4.39 mmol) in
pyridine (11 mL) at 40 ꢁC was added freshly pulverized
KOH(340 mg, 6.1 mmol) in one portion. After 1 h, the
reaction was quenched by addition of 5% aq HCl (7
mL). The resultant precipitate was collected by filtration
and washed with ice cold water. The precipitate was
dried and then dissolved in glacial acetic acid (22 mL).
Conc. sulfuric acid (0.9 mL) was added and the reaction
mixture was heated to reflux. After 1.5 h, the mixture was
cooled to room temperature and quenched by pouring
over 50 g crushed ice. The precipitate was filtered and
washed with ice cold water. The precipitate was dissolved
in a mixture of CH2Cl2 (100 mL) and MeOH(50 mL)
and then the solution was washed with sat. NaHCO3
(3ꢃ25 mL). The organic fraction was dried (Na2SO4),
concentrated by rotary evaporation, and the residue was
purified by chromatography, eluting with 5% MeOHin
CH2Cl2, to yield UCCF-296 (0.58 g, 38%) as an off-
white solid; mp 279–280 ꢁC; TLC, Rf 0.51 (5% MeOHin
1737, 1687 cmꢂ1
;
1HNMR (300 MHz)
d 8.93 (dd,
J=4.1, 1.8 Hz, 2H), 8.28 (d, J=8.8 Hz, 1H), 8.21 (s,
1H), 8.11 (dd, J=4.7, 1.8 Hz, 2H), 8.01 (d, J=8.2 Hz,
1H), 7.75 (ddd, J=7.6, 7.0, 1.2 Hz, 1H), 7.62 (ddd,
J=8.2, 7.0, 1.2 Hz, 1H), 2.62 (s, 3H); 13C NMR
(75 MHz) d 195.7, 163.6, 151.0, 145.5, 136.1, 134.6,
130.2, 128.9, 128.4, 128.3, 127.7, 126.4, 123.4, 123.2,
120.7, 29.8. Anal. calcd C18H12BrNO3: C, 58.40; H,
3.27; N, 3.78. Found: C, 58.29; H, 3.54; N, 3.56.
CH2Cl2). IR 3060, 3020, 1654 cmꢂ1
;
1HNMR
(300 MHz) d 8.90 (d, J=5.5 Hz, 2H), 8.60 (ddd, J=8.1,
1.6, 0.8 Hz, 1H), 8.45 (s, 1H), 8.36 (ddd, J=8.0, 1.6, 0.6
Hz, 1H), 7.89–7.78 (m, 4H), 7.05 (d, J=0.6, Hz, 1H);

M. F. Springsteel et al. / Bioorg. Med. Chem. 11 (2003) 4113–4120
4119
Figure 3. Common features pharmacophore modeling from the flavone series of CFTR activators and comparison to other channel activators. (a)
The consensus pharmacophore model derived from all flavones in the study, shown aligned to the parent flavone. Pharmacophore features are
shown as labeled, colored spheres: red (A, hydrogen bond acceptor), green (B, aromatic ring), orange (C, hydrophobic group). (b) The four-point
pharmacophore derived from the most active flavones (top 10%) in this study, shown aligned to UCCF-029. The hydrophobic feature corresponding
to point D reflects the importance of benzannulation at the 7,8-position for activity. (c) Alignment of CFTRact04 (red), CFTRact14 (green),
CFTRact12 (blue), and CFTRact17 (purple) to the four-point pharmacophore model from (b). (d) Alignment of adenosine (red) and the xanthines
CPX (purple) and DAX (blue) to the four-point model from (b).
13C NMR (100 MHz) d 176.7, 160.3, 153.0, 151.2, 139.2,
134.5, 131.0, 128.4, 128.3, 125.3, 124.4, 122.7, 121.1,
120.6, 119.9, 110.8. Anal. calcd C18H10BrNO2: C, 61.39;
H, 2.86; N, 3.98. Found: C, 61.11; H, 2.85; N, 3.93.
J=8.8, 2.3 Hz, 1H), 7.76 (d, J=5.3 Hz, 2H), 7.65 (dd,
J=4.1, 3.5 Hz, 2H), 7.62 (s, 1H), 7.49–7.36 (m, 3H),
6.91 (s, 1H); 13C NMR (75 MHz) d 117.9, 160.4, 155.4,
150.8, 139.0, 138.9, 138.7, 133.0, 129.0, 128.0, 127.1, 124.1,
123.5, 119.7, 118.7, 109.2. Anal. calcd C20H13NO2: C,
80.25; H, 4.38; N, 4.68. Found: C, 80.18; H, 4.37; N,
4.72.
Representative Suzuki coupling: 6-phenyl-2-(4-pyri-
dyl)4H-chromen-4-one (UCCF-320). To a degassed (Ar)
mixture of 6-bromo-2-(4-pyridyl)4H-chromen-4-one
(151 mg, 0.500 mmol), phenylboronic acid (125 mg, 1.03
mmol) and 2 M Na2CO3 (1.5 mL) in EtOH(2.5 mL)
and toluene (2.5 mL) at room temperature was added
Pd(Ph3P)4 (11.5 mg, 9.95 mM). After heating to reflux 2
h, the reaction was cooled to room temperature and
diluted with EtOAc (50 mL) and water (10 mL). The
layers were separated and the organic fraction was
washed with brine (1ꢃ25 mL) and dried (Na2SO4).
After removal of the solvents, the crude material
was purified via recrystallization from EtOAc to
yield UCCF-320 (122 mg, 70%) as white crystals;
mp 204–205 ꢁC, TLC, Rf (EtOAc) 0.26. IR 3047,
Computational chemistry
Common features pharmacophore models for all fla-
vones and the top 10% most active flavones were gen-
erated using the HipHop module in Catayst (Accelrys,
Inc.). Briefly, the HipHop algorithm accepts a collection
of conformational models for each molecule and
attempts to align the molecules based on a set of che-
mical features, such as hydrogen bond donor/acceptor,
aromatic ring, or hydrophobe. Pharmacophore features
are assigned where like chemical features map to the
same region in space, within a user-defined tolerance.
Prior to input into HipHop, each molecule was sub-
jected to geometry optimization using quantum
1636, 1611 cmꢂ1
.
J=5.3 Hz 2H), 8.41 (d, J=2.3 Hz, 1H), 7.96 (dd,
1HNMR (300 MHz)
d 8.81 (d,

4120
M. F. Springsteel et al. / Bioorg. Med. Chem. 11 (2003) 4113–4120
mechanics at the HF/6-31G* level of theory (Gaussian
98). Conformational models were generated using the
Confirm algorithm with the Best option and default
parameters (Accelrys, Inc.). Molecules were aligned to
the pharmacophore models and viewed using the Com-
pare/Fit module in Catalyst (Accelrys, Inc.). Alignments
were viewed and analyzed using the program MOE
(Chemical Computing Group, Inc).
References and Notes
1. Young, L. Y.; Koda-Kimble, M. A. Applied Therapeutics, The
Clinical Use of Drugs, 6th Ed.; Applied Theraputics: Vancouver,
NA, 1995; p 100.1.
2. Pilewski, J. M.; Frizzell, R. A. Physiol. Rev. 1999, 79, S215.
3. Liu, X.; Jiang, Q.; Mansfield, S. G.; Puttaraju, M.; Zhang,
Y.; Zhou, W.; Cohn, J. A.; Garcia-Blanco, M. A.; Mitchell,
L. G.; Engelhardt, J. F. Nat. Biotechnol. 2002, 20, 47.
4. Kitson, C.; Alton, E. Expert Opin. Invest. Drugs 2000, 9, 1523.
5. Marshal, E. Science 2000, 287, 565.
6. Zeitllin, P. L. J. Clin. Invest. 1999, 103, 447.
7. Schultz, B. D.; Singh, A. K.; Devor, D. C.; Bridges, R. J.
Physiol. Rev. 1999, 79, S109.
8. Galietta, L. J. V.; Springsteel, M. F.; Eda, M.; Niedzinski,
E. J.; By, K.; Haddadin, M. J.; Kurth, M. J.; Nantz, M. H.;
Verkman, A. S. J. Biol. Chem. 2001, 267, 19723.
9. Galietta, L. J. V.; Jayaraman, S.; Verkman, A. S. Am. J.
Physiol. Cell Physiol. 2002, 281, C1734.
Biology
Halide transport assay. Fisher rat thyroid (FRT) cells
stably co-expressing human wild-type CFTR and the
yellow fluorescent protein YFP-H148Q were cultured
and assayed as described previously.10 Briefly, cells were
plated in 96-well microplates and, after 24–48 h, washed
with PBS and incubated for 15 min with 40 mL of the
same solution containing 100 nM forskolin with or
without compounds to be tested or reference CFTR
activators (apigenin and IBMX) at various concentra-
tions. The microplates were then processed in a FluoStar
fluorescence microplate reader (BMG Lab Technolo-
gies) equipped excitation/emission filters for yellow
fluorescent protein (Chroma) and two syringe pumps
for liquid addition. Each well fluorescence was read for
14 s with a sampling time of 0.2 s. After 2 s of fluores-
cence reading, 165 mL of a PBS containing 137 mM NaI
instead of NaCl was added to measure CFTR-dependent
fluorescent quenching. The decay of fluorescence in each
well was fitted with a third-order polynomial to derive
the initial fluorescence slope at the time of iodide addi-
tion.
10. Ma, T.; Vetrivel, L.; Yang, H.; Pedemonte, N.; Zegarra-
Moran, O.; Galietta, L. J. V.; Verkman, A. S. J. Biol. Chem.
2002, 277, 37235.
11. Baker, W. J. Chem. Soc. 1933, 1381.
12. Mahal, H. S.; Venkataraman, K. J. Chem. Soc. 1934, 1767.
13. Dallacker, F.; Wersch, J. V. Chem. Ber. 1972, 105, 3301.
14. Brunner, H.; Schiebling, H. Bull. Soc. Chim. Belg. 1994,
103, 119.
15. Banu, H. S.; Pitchumani, K.; Srinivasan, C. Tetrahedron
1999, 55, 9601.
16. Furniss, B. S.; Hannaford, A. J.; Smith, P. W. G.; Tatchell,
A. R. Vogel’s Textbook of Practical Organic Chemistry, 5th
Ed.; John Wiley & Sons: New York, 1989; p 980.
17. Paranjape, M. V.; Wadodkar, K. N. Indian J. Chem. 1981,
20B, 808.
18. Boyer, J. L.; Krum, J. E.; Myers, M. C.; Fazal, A. N.;
Wigal, C. T. J. Org. Chem. 2000, 65, 4712.
19. Suzuki, A. Pure Appl. Chem. 1985, 57, 1749.
20. Wei, Z.-Y.; Brown, W.; Takasaki, B.; Plobeck, N.;
Delorme, D.; Zhou, F.; Yang, H.; Jones, P.; Gawell, L.; Gag-
non, H.; Schmidt, R.; Yue, S.-Y.; Walpole, C.; Payza, K.; St-
Onge, S.; Labarre, M.; Godbout, C.; Jakob, A.; Butterworth,
J.; Kamassah, A.; Morin, P.-E.; Projean, D.; Ducharme, J.;
Roberts, E. J. Med. Chem. 2000, 43, 3895.
21. Muller, D.; Fleury, J.-P. Tetrahedron Lett. 1991, 32, 2229.
22. Illek, B.; Lizarzaburu, M. E.; Lee, V.; Nantz, M. H.; Kurth,
M. J.; Fischer, H. Am. J. Physiol. Cell Physiol. 2000, 279, C1838.
23. Bulteau, L.; Derand, R.; Mettey, Y.; Metaye, T.; Morris,
M. R.; McNeilly, C. M.; Folli, C.; Galietta, L. J. V.; Zegarra-
Moran, O.; Pereira, M. M. C.; Jougla, C.; Dormer, R. L.;
Vierfond, J. M.; Joffre, M.; Becq, F. Am. J. Physiol. Cell
Physiol. 2000, 279, C1925.
Short-circuit current measurements. FRT cells expres-
sing wild-type or G551D-CFTR were plated at high
density on Snapwell (Corning-Costar) permeable inserts
as described previously.10 After 7–10 days, Snapwell
inserts were mounted in a modified Ussing chamber.
The basolateral side was filled with a Ringer solution
containing: 130 mM NaCl, 2.7 mM KCl, 1.5 mM
KH2PO4, 1 mM CaCl2, 0.5 mM MgCl2, 10 mM Na-
Hepes (pH 7.3), and 10 mM glucose. The apical side
contained a similar solution in which half of NaCl was
replaced with sodium gluconate and the CaCl2 increased
to 2 mM. The basolateral membrane of FRT cells was
permeabilized with 250 mg/mL of amphotericin B for 30
min. Both sides were connected to a voltage clamp
(World Precision Instruments) by means of voltage-
sensing and current-passing Ag/AgCl electrodes. Short-
circuit current data were converted in digital form and
stored on an Apple personal computer.
24. Illek, B.; Zhang, L.; Lewis, N. C.; Moss, R. B.; Dong,
J. Y.; Fischer, H. Am. J. Physiol. Cell Physiol. 1999, 277,
C833.
25. Diamond, J. M.; Wright, E. M. Proc. Royal Soc., Ser. B
1969, 172, 273.
26. Cohen, B. E.; Lee, G; Jacobson, K. A.; Kim, Y.-C.;
Huang, Z.; Sorscher, E. J.; Pollard, H. B. Biochemistry 1997,
36, 6455.
27. Azevedo, W. F. D., Jr.; Mueller-Dieckmann, H.-J.;
Schulze-Gahmen, U.; Worland, P. J.; Sausville, E.; Kim, S.-H.
Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 2735.
28. Conseil, G.; Baubichon-Cortay, H.; Dayan, G.; Jault, J.-
M.; Barron, D.; Di Pietro, A. Proc. Natl. Acad. Sci. U.S.A.
1998, 95, 9831.
Acknowledgements
This work was supported by a drug discovery grant
from the Cystic Fibrosis Foundation. We thank Mr.
Sung Hee Hwang for assistance with compound pre-
parations. A.A.S. is supported by the National Defense
Science and Engineering Graduate Fellowship.
29. Sicheri, F.; Moarefi, I.; Kuriyan, J. Nature 1997, 385, 602.