Activation and inhibition of kidney CLC-K chloride channels by fenamates

Article abstract of DOI:10.1124/mol.105.017384

CLC-K Cl^- channels are selectively expressed in kidney and ear, where they are pivotal for salt homeostasis, and loss-offunction mutations of CLC-Kb produce Bartter's syndrome type III. The only ligand known for CLC-K channels is a derivative o

Full text of DOI:10.1124/mol.105.017384

0026-895X/06/6901-165–173$20.00
MOLECULAR PHARMACOLOGY
Copyright © 2006 The American Society for Pharmacology and Experimental Therapeutics
Mol Pharmacol 69:165–173, 2006
Vol. 69, No. 1
17384/3074760
Printed in U.S.A.
Activation and Inhibition of Kidney CLC-K Chloride Channels
by Fenamates
Antonella Liantonio, Alessandra Picollo, Elena Babini, Giuseppe Carbonara,
Giuseppe Fracchiolla, Fulvio Loiodice, Vincenzo Tortorella, Michael Pusch, and
Diana Conte Camerino
Unita` di Farmacologia, Dipartimento Farmacobiologico (A.L., D.C.C.) and Dipartimento Farmacochimico (G.C., G.F., F.L., V.T.),
Facolta` di Farmacia, Universita` di Bari, Bari, Italy; and Istituto di Biofisica, Consiglio Nazionale delle Ricerche, Genova, Italy
(A.P., E.B., M.P.)
Received July 28, 2005; accepted October 20, 2005
ABSTRACT
CLC-K ClϪ channels are selectively expressed in kidney and Flufenamic acid (FFA) derivatives or high doses of NFA pro-
ear, where they are pivotal for salt homeostasis, and loss-of- duced instead an inhibitory effect on CLC-Ka, but not on CLC-
function mutations of CLC-Kb produce Bartter’s syndrome Kb, and on blocker-insensitive CLC-Ka mutants, indicating that
type III. The only ligand known for CLC-K channels is a deriv- the activating binding site is distinct from the blocker site.
ative of the 2-p-chlorophenoxypropionic acid (CPP), 3-phenyl- Evaluation of the sensitivity of CLC-Ka to derivatives of NFA
CPP, which blocks CLC-Ka, but not CLC-Kb. Here we show and FFA together with a modeling study of these ligands allow
that in addition to this blocking site, CLC-K channels bear an us to conclude that one major characteristic of activating com-
activating binding site that controls channel opening. Using the pounds is the coplanarity of the two rings of the molecules,
voltage-clamp technique on channels expressed in Xenopus whereas block requires a noncoplanar configuration. These
laevis oocytes, we found that niflumic acid (NFA) increases molecules provide a starting point for identification of diuretics
CLC-Ka and CLC-Kb currents in the 10 to 1000 ␮M range. or drugs useful in the treatment of Bartter’s syndrome.
CLC-Ka and CLC-Kb chloride channels are expressed the countercurrent system of the inner medulla. Its loss
along the nephron from the thin ascending limb to the disrupts the entire system, as suggested by the nephro-
collecting duct, where they are pivotal for chloride exit
across the basolateral membrane (Jentsch et al., 2002;
Uchida and Sasaki, 2005). The significance of these chan-
nels for salt homeostasis is illustrated by genetic disorders
(Jentsch, 2005). Defects in the gene encoding CLC-Kb re-
duce channel activity, producing Bartter’s syndrome type
III, a disease characterized by severe salt wasting and
hypokalemia (Simon et al., 1997). CLC-Kb is implicated in
transepithelial NaCl transport in the thick ascending limb
and collecting ducts, whereas CLC-Ka-mediated ClϪ trans-
port in the thin ascending limb constitutes a component of
genic diabetes insipidus phenotype of CLC-K1 knock-out
mice (Akizuki et al., 2001).
CLC-K channels are also expressed in the inner ear, where
they are involved in endolymph secretion, a mechanism piv-
otal for sound signal transduction (Este´vez et al., 2001).
Furthermore, for a correct expression and function, CLC-K
channels require the presence of the barttin ␤ subunit (Es-
te´vez et al., 2001; Waldegger et al., 2002). Barttin produces
an increase of CLC-K channel expression at the plasma mem-
brane (Este´vez et al., 2001; Waldegger et al., 2002). Muta-
tions in barttin cause type IV Bartter’s syndrome, a disease
characterized by renal failure accompanied by sensorineural
deafness (Birkenha¨ger et al., 2001). In support of the hypoth-
esis that defects in barttin impair function of both CLC-K
channels, simultaneous CLC-Ka and CLC-Kb mutations re-
sult in a phenotype that mimics type IV Bartter’s syndrome
(Schlingmann et al., 2004).
This work was support financially by the Italian Research Ministry [Min-
istero dell’Istruzione, dell’Universita` e della Ricerca COFIN 2003051709002
(to D.C.C.) and Fondo per gli Investimenti della Ricerca di Base RBAU01PJMS
(to M.P.)] and Telethon Italy [grant GGP04018 (to M.P.)].
A.L. and A.P. contributed equally to this work.
D.C.C. and M.P. contributed equally to this work.
Article, publication date, and citation information can be found at
http://molpharm.aspetjournals.org.
It is noteworthy that a polymorphism of CLC-Kb has re-
cently been reported that confers a gain of function of the
doi:10.1124/mol.105.017384.
ABBREVIATIONS: CLC-K, kidney CLC chloride channels; CPP, 2-p-(chlorophenoxy)propionic acid; DIDS, 4,4Ј-diisothiocyanato-2,2Ј-stilbenedis-
ulfonic acid; NFA, niflumic acid; FFA, flufenamic acid; MFA, mefenamic acid; MCFA, meclofenamic acid; TFA, tolfenamic acid; DPC, 2-(phe-
nylamino)benzoic acid; WT, wild type.
165

166
Liantonio et al.
channel-activating CLC-Kb activity (Jeck et al., 2004a) in
vitro and predisposes to hypertension in vivo (Jeck et al.,
2004b).
Similar voltage-clamp pulse protocols for CLC-Ka and CLC-Kb,
with a longer pulse duration for CLC-Kb, were used. From a holding
potential of Ϫ30 mV, after a prepulse to 60 mV (or Ϫ100 mV) for 100
or 200 ms (for CLC-Ka and CLC-Kb, respectively), voltage was
stepped from Ϫ140 to 80 mV in 20-mV increments for 200 or 500 ms
(for CLC-Ka and CLC-Kb, respectively), followed by a final tail pulse
to Ϫ100 mV. To evaluate the on-set and washout of drug effects, a
pulse to 60 mV was applied every 2 s.
The involvement of these ClϪ channels in various physio-
pathological processes arouses a considerable interest to
identify specific high-affinity ligands as tools to explore
mechanisms of gating and permeation and as a starting point
for drug development.
As a control, we routinely applied a solution containing 100 mM IϪ
that blocks currents carried by CLC-K channels but not endogenous
currents (Pusch et al., 2000) and used the residual current in 100
In previous studies, we have identified compounds with
two chlorophenoxy groups capable of blocking rat CLC-K1
currents (Liantonio et al., 2002). Subsequently, we recog- mM IϪ to estimate the contribution of endogenous currents (Picollo
et al., 2004).
nized 3-phenyl-CPP as the minimal structure capable of
blocking CLC-K1 and CLC-Ka in a reversible and ClϪ ion-
dependent manner by interacting with the channel pore from
the extracellular side (Liantonio et al., 2004). The highly
homologous CLC-Kb channel was found to be 5-fold less
sensitive to 3-phenyl-CPP compared with CLC-Ka (Picollo et
al., 2004). We identified the amino acid residues that are
responsible for the difference in drug sensitivity and thus
probably form part of the drug-binding site. In particular,
asparagine 68 in CLC-Ka plays a pivotal role for the blocking
activity of 3-phenyl-CPP, because exchanging it with a neg-
atively charged aspartate (N68D), as found in CLC-Kb,
markedly reduces drug sensitivity. Furthermore, Asn68, to-
gether with Gly72, is also pivotal for the more pronounced
activity of the unrelated stilbene blocker DIDS on CLC-Ka
with respect to CLC-Kb (Picollo et al., 2004).
Apparent dissociation constants for drugs showing blocking activ-
ity, KD, were determined by calculating the ratio of the steady-state
current in the presence and in absence of the drug and fitting the
ratios to the equation I(c)/I(0) ϭ 1/(1 ϩ c/KD) where c is the concen-
tration. Errors in figures and in the text are indicated as S.E.M.
NFA Derivatives. The following drugs were purchased from Sig-
ma-Aldrich (Milano-Italy): NFA, FFA, mefenamic acid (MFA),
meclofenamic acid (MCFA), tolfenamic acid (TFA), and 2-(phe-
nylamino)benzoic acid (DPC).
Remaining compounds were synthesized according to previ-
ously reported procedures; MT-4, MT-6, and MT-7 were prepared
by condensation of 2-chloronicotinic acid with the appropriate
substituted amines (Sherlock et al., 1988). The phenoxypyridin-
ecarboxylic acids (Li-3, Li-10, Li-6, and Li-11) were synthesized by
condensing 2-chloronicotinic acid with the opportune sodium phe-
noxide using an excess of the phenol (Villani et al., 1975). For
Li-15, the condensation was modified using equimolar quantities
of the acid and the phenolate in DMF (Fujiwara and Kitagawa,
2000). EB-168 was synthesized by treating commercially available
NFA with an excess of borane methyl sulfide complex in dry
In this study, we demonstrate that in addition to this
blocking binding site, CLC-K channels also present an acti-
vating binding site that controls opening of the channels. We
show that niflumic acid (NFA), a drug belonging to a class of tetrahydrofuran. Compounds were daily prepared in dimethyl
sulfoxide stock solutions, and the final concentrations were ob-
tained by appropriate dilution with the solution used for the
electrophysiological recordings. (See Fig. 3 for structures of com-
pounds listed in this paragraph.)
Modeling Study. We searched the lowest energy conformers for
each drug by a systematic Merck Molecular Force Field analysis
neglecting solvent presence. A restricted number of conformers (Ta-
ble 1) were found for the fenamates in the range of 10 kcal/mol.
Results were similar to those reported previously (Dhanaraj and
Vijayan, 1988). 3-Phenyl-CPP, a more flexible molecule, showed a
fenamates usually used as nonsteroidal anti-inflammatory
drugs, produces an increase of currents carried by CLC-Ka
and CLC-Kb. In contrast to this activating effect, application
of flufenamic acid (FFA) derivatives or high doses of NFA
produced an inhibitory effect on CLC-Ka. The evaluation of
the sensitivity to fenamates of CLC-Kb and of CLC-K mu-
tants (Picollo et al., 2004), together with a molecular model-
ing study of the organic ligands, provides insight into the
mechanisms of activation and inhibition by fenamates.
This class of molecules represents until now the unique larger number of conformers (Table 1) in the same range of energy.
For FFA, MCFA, NFA, and 3-phenyl-CPP, the same conformational
analysis was also carried out in the presence of water, and results
were close to those without solvent.
In Table 1, we report the number of low energy conformers found
after the conformational analysis carried out with and without wa-
ter. When available, as in the case of DPC, FFA, MCFA, MFA, NFA,
and TFA molecules, the starting geometry parameters were derived
from X-ray data from Cambridge Crystallographic Data Center
(Cambridge Structural Database, ver 5.26).
tool able to open and to block renal CLC-K channels and thus
provides a starting point for the identification of drugs either
with diuretic action or useful in the treatment of type III
Bartter’s syndrome.
Materials and Methods
Expression in Xenopus laevis Oocytes and Voltage-Clamp
Analysis. WT CLC-Ka, CLC-Kb and their mutants, obtained as
described previously (Picollo et al., 2004), were coexpressed with the
activating mutant Y98A of human barttin (Este´vez et al., 2001).
Expression in oocytes and electrophysiological measurements were
performed as described previously (Pusch et al., 2000). In brief,
voltage-clamp data were acquired at room temperature (21–25°C)
using the Pulse program (HEKA, Lambrecht, Germany) or a custom
acquisition program (Gepulse) and a custom-built amplifier or a
TEC03 amplifier (NPI Electronic GmbH, Tamm, Germany). Cur-
rents were recorded in the standard solution containing 90 mM
NaCl, 10 mM CaCl2, 1 mM MgCl2, and 10 mM HEPES at pH 7.3. In
experiments with low extracellular [Ca2ϩ], 1.8 mM CaCl2 was used
and NaCl was increased to 106 mM to balance osmolarity and salt
strength.
For 3-phenyl-CPP, whose X-ray data were not available, the
molecule was first constructed by fragments, the molecular geom-
etry was optimized to DFT B3LYP/6-31G* level theory and then
TABLE 1
Number of low energy conformers obtained from the conformational
analysis carried out with and without water (see Materials and
Methods)
Without
H2O
With
H2O
Molecule
FFA
8
6
11
36
18
13
12
50
MCFA
NFA
3-Phenyl-CPP

Fenamates as CLC-K Channel Modulators
167
submitted for further calculations (Spartan ’04; Wavefunction,
Inc., Irvine, CA). Except for molecular mechanics and semiempiri-
cal models, the calculation methods used in Spartan ’04 have been
documented by Kong et al. (2000). We observed that the most
populated low energy conformer families are within the 3.0 kcal/
mol range; among them, the lowest energy conformer was selected
and used for the overlay shown in Fig. 9B. The conformers were
superimposed at the level of the aromatic ring not directly bound
to the carboxylic group, and of the heteroatoms, nitrogen or oxy-
gen, bridging the aromatic rings.
Results
Effect of Fenamates on CLC-Ka
Effect of NFA. In the first series of experiments, we
characterized the effect of NFA on the CLC-Ka channel.
CLC-Ka currents partially deactivate at positive potentials
and activate at negative potentials (Fig. 1A, left).
Application of 200 ␮M NFA induces an increase of CLC-Ka
currents both at negative and positive potentials (Fig. 1A).
The onset of the effect was rapid, and the return of the
currents to control value upon removal of drug was reached
within 7 to 9 min (Fig. 1B). The NFA-mediated current in-
All calculations were performed with the Spartan ’04 package
(Wavefunction Inc.). Graphical representations were performed by
DS Viewerpro 6.0 trial version (Accelrys Inc., San Diego, CA).
Fig. 1. Effect of NFA on CLC-Ka. A, voltage-
clamp traces of CLC-Ka currents before and dur-
ing application of 200 ␮M NFA. The pulse pro-
tocol is shown as inset. After a prepulse to Ϫ100
mV, voltage was stepped from Ϫ140 to 80 mV in
20-mV increments for 200 ms. B, time course of
CLC-Ka current at 60 mV. Lines indicate appli-
cation of NFA and wash solution. C, dose-re-
sponse relationship of NFA at 60 mV. The ratio
of the current in the presence and absence of
drug is plotted versus concentration.

168
Liantonio et al.
crease was practically voltage-independent (Idrug/I0 ϭ 2.38 Ϯ pretation of increased potency of NFA in low calcium is that
0.12 at 60 mV and Idrug/I0 ϭ 2.20 Ϯ 0.3 at Ϫ140 mV). At all in this condition, a larger population of channels is available
tested potentials, NFA induced an increase of the current for opening by NFA.
amplitude without producing kinetic modifications. Although
Structure-Activity Study. Searching to determine the
a current increase was seen for concentrations between 50 structural requisites to bind to and to modulate CLC-Ka
and 1000 ␮M, NFA blocked CLC-Ka currents at 2 and 2.5 activity, we evaluated the effect of a series of compounds,
mM (Fig. 1C). This biphasic dose-response relationship indi- modifying in different parts the NFA structure (Fig. 3). We
cates that the mechanism of action of NFA is complex and determined the potency of the various derivatives by compar-
probably involves two functionally different binding sites ing the effect produced by each drug at 200 ␮M, a concentra-
with opposite effects.
tion at which the NFA-mediated activating effect was evi-
dent.
Dependence of NFA Activation on Extracellular Cal-
cium Concentration. It is known that CLC-K channel ac-
tivity is strongly enhanced by extracellular calcium in the
millimolar range (Este´vez et al., 2001; Waldegger et al.,
2002). We examined whether calcium may interfere with the
Substitutions on the Phenyl Group. Derivative MT-7,
in which the CF3 group on the phenyl ring is shifted from the
meta to the para position with respect to the anilinic nitro-
gen, is still able to produce an increase of CLC-Ka currents,
albeit with a reduced potency (Idrug/I0 ratio of 1.89 Ϯ 0.04;
Fig. 4A). Substitution of the CF3 group with a chlorine atom
compromised the activating drug activity much more drasti-
cally. Indeed, application of derivatives MT-6 and MT-4,
which have a chlorine atom in the meta or para position,
respectively, only slightly increased or produced no effect,
respectively (Fig. 4A). These data indicate that an electro-
negative effect that reduces the density of the electron cloud
of this aromatic ring is not sufficient to confer drug activity.
Rather, the presence of the 3-fluoromethyl group seems to
confer a specific bulkiness in this part of the molecule suit-
able for the interaction with the activating binding site.
Elimination of the Carboxylic Group. The role of the
acidic function in the NFA molecule was investigated by
substituting the carboxylic group with a hydroxymethyl
group. Application of derivative EB-168 at 200 ␮M produced
no modification of CLC-Ka (Fig. 4B), suggesting a pivotal role
of the carboxylic group in mediating NFA-induced activating
effect.
activating effect of NFA. Lowering [Ca2ϩ
]
ext
from 10 to 1.8
mM reduces CLC-Ka currents (data not shown) and, impor-
tantly, the relative activating effect of 200 ␮M NFA was
significantly larger in low calcium (Fig. 2). A possible inter-
Isosteric Substitution of the Anilinic Nitrogen.
Changing the anilinic nitrogen that links the two rings abol-
ished activation of CLC-Ka, independent of other substitu-
tions in the phenyl ring as indicated by the fact that appli-
cation of each isosteric derivative (substances Li-15, Li-3,
Fig. 2. Effect of extracellular calcium on the activity of CLC-Ka mediated
by NFA. The Idrug/I0 ratio at 60 mV and at Ϫ140 mV measured in 10 mM Li-10, Li-6, and Li-11 in Fig. 3) at 200 ␮M produced no
and 1.8 mM extracellular Ca2ϩ are compared.
significant change in CLC-Ka currents (Fig. 5A). To exclude
Fig. 3. Chemical structures of niflumic acid (in the
inset above on the left) derivatives. A, compounds
with substitutions on the phenyl group. B, com-
pound with elimination of the carboxylic group. C,
drugs with isosteric substitution of the anilinic ni-
trogen. D, compounds with elimination of the pyri-
dinic ring. For comparison, the 3-phenyl-CPP mole-
cule is shown at lower right.

Fenamates as CLC-K Channel Modulators
169
the formal possibility that the lack of effect of substances an inhibitory binding site with some different structural
Li-15, Li-3, Li-10, Li-6, and Li-11 was due to a compensation requirements.
of activation and block, we evaluated them also at 50 ␮M. If
We also evaluated the effect of 3-phenyl-CPP, the CLC-K
block at 200 ␮M had masked an activating effect, an increase blocker with highest known affinity (Liantonio et al., 2004;
of currents would be expected at the lower concentration of Picollo et al., 2004). In agreement with a previous study
50 ␮M. However, none of these derivatives produced a sig- (Picollo et al., 2004), 3-phenyl-CPP inhibited CLC-Ka cur-
nificant effect at this concentration (data not shown).
rents with a potency reported in Table 2.
Conversion of the Pyridinic Ring into a Phenyl Ring.
Effect of Fenamates on CLC-Kb. Although highly ho-
In FFA, the pyridinic ring is substituted with a phenyl ring mologous (Kieferle et al., 1994), the two CLC-K isoforms
(Fig. 3). It is noteworthy that 200 ␮M FFA blocked CLC-Ka- show a different pharmacological profile (Picollo et al., 2004).
sustained outward and inward currents in a dose-dependent We evaluated the sensitivity of CLC-Kb to fenamates using
manner (Fig. 5B and 6A) with a Idrug/I0 ratio of 0.45 Ϯ 0.04 at NFA and FFA, the two lead compounds with activating and
60 mV (Fig. 6B). Onset of the effect as well as washout were inhibitory effect on CLC-Ka. At 200 ␮M, the effect of NFA on
quite rapid (Fig. 6B). The dose-response curve was well-fitted CLC-Kb is more pronounced compared with CLC-Ka, with a
by a simple titration curve at Ϫ140 and 60 mV, with appar- Idrug/I0 ratio of 3.5 Ϯ 1.0.
ent KD values reported in Table 2, suggesting 1:1 binding.
It is noteworthy that in contrast to what we observed with
Although the potentiation of CLC-Ka by NFA was depen- CLC-Ka, application of 200 ␮M FFA produced an increase of
dent on the presence of a CF3 group in meta position on the CLC-Kb currents with a Idrug/I0 ratio of 2.09 Ϯ 0.9 at 60 mV
phenyl group, a modification regarding this region of the (Fig. 7). Furthermore, even at high concentrations (Ͼ1 mM),
molecule on FFA led to an increase of the blocking activity. In NFA was incapable of blocking CLC-Kb sustained currents,
particular, the order of potency of FFA derivatives was producing a potentiationat all tested concentrations (data not
MCFA Ͼ MFA Ͼ TFA Ͼ FFA (Fig. 5A; Table 2). The elimi- shown).
nation of all substituents on this phenyl group (molecule
Effect of NFA and FFA on CLC-K Mutants. We have
DPC, Fig. 3) significantly reduced the drug inhibitory activ- previously identified Asn68 on CLC-Ka as a key amino acid
ity (Fig. 5A; Table 2). These results suggest that the substi- of the binding site for 3-phenyl-CPP and DIDS (Picollo et al.,
tution of the pyridinic group with a phenyl group completely 2004), inhibitors of CLC-K1 and CLC-Ka (Liantonio et al.,
shifted the affinity of NFA from an activating binding site to
Fig. 5. A, structure-activity relationship study: each point represents the
average of the Idrug/I0 ratio at 60 mV for each molecule (on the left)
obtained from n oocytes (on the right). B, current-voltage relationship
obtained in control condition, in presence of 200 ␮M NFA and 200 ␮M
FFA. Each point represents the average of determinations obtained from
four to six oocytes.
Fig. 4. Substitutions on the phenyl group. A, ratio of the current in the
presence and absence of compound is plotted versus time. The arrow
indicates application of compounds. B, voltage-clamp traces before and
during application of 200 ␮M EB-168.

170
Liantonio et al.
2004; Picollo et al., 2004). Most features of FFA-mediated both for CLC-Ka and CLC-Kb (Picollo et al., 2004). Each
inhibition resemble those of 3-phenyl-CPP, suggesting a com- mutant was sensitive to NFA with a potency comparable
mon inhibitory binding site shared by these two different
classes of CLC-Ka inhibitors. To assess this hypothesis, we
tested FFA and its derivatives on the CLC-Ka mutant N68D,
the mutant that was least sensitive to 3-phenyl-CPP block
(Picollo et al., 2004). FFA (200 ␮M) produced no significant
decrease of currents sustained by CLC-Ka N68D, suggesting
that this point mutation strongly reduced drug affinity.
MCFA, the most potent derivative among the compounds
tested here on CLC-Ka, showed a KD of 632 Ϯ 101 ␮M, a
decrease in affinity of almost 15-fold compared with WT. For
all other FFA derivatives, the apparent KD value was Ͼ1 mM
(Table 2).
with that observed in the related wild type (Fig. 8). However,
it is interesting that NFA produced a slightly smaller effect
on CLC-Ka N68D with respect to WT (Fig. 8), indicating that
the activating and inhibitory binding site may not be com-
pletely independent.
At high concentrations, such as 2 mM, NFA is still capable
of producing a block of CLC-Ka N68D mutant, although less
potently with respect to CLC-Ka wild type (Idrug/I0 ϭ 0.75 Ϯ
0.05 and 0.67 Ϯ 0.04 at 60 mV, respectively).
Modeling Study. Conformational search studies show
that all examined NFA derivatives exhibit a restricted num-
ber of low energy conformers characterized by the presence of
an intramolecular hydrogen bond between the amino group
and the oxygen atoms of carboxylic moiety. For all examined
molecules, the calculated low energy conformations include
the ones found in the X-ray crystal structures.
We next tested NFA on all mutants previously reported
The lowest energy conformer of NFA shows a quite planar
structure that is different from the corresponding conformers
of the remaining molecules in which the two aromatic rings
lie on different planes (Fig. 9A). The NFA planar structure is
the result of the intramolecular hydrogen bonding as well as
of the conjugation of the amino group lone pair with the
electron-withdrawing pyridine ring. This conjugation is less
effective in the FFA analogs because of the substitution of the
pyridine system with a phenyl system; at the same time, this
substitution introduces an additional hydrogen atom in the
ortho position to the amino group, causing a steric hindrance
in the resulting diphenyl system that forces the noncoplanar
arrangement of the aromatic rings (Dhanaraj and Vijayan,
1988). This behavior is enhanced in the compounds having
bulky substituents in the ortho positions of the aniline moi-
ety, such as MCFA, MFA, and TFA (Fig. 9A).
Regarding 3-phenyl-CPP, all lower energy conformer fam-
ilies exhibit the two aromatic rings lying on different planes
because of their molecular flexibility. An overlay of the lowest
energy conformers of the lead molecules indicates that all
TABLE 2
Isosteric derivatives of the pyridinic nitrogen
KD values are expressed as mean Ϯ S.E.M. calculated for each compound. n indicates
number of oocytes.
Inhibition (KD
)
Compound and Channel
n
60 mV
Ϫ140 mV
␮M
FFA
CLC-Ka
121 Ϯ 37
Ͼ5000
57 Ϯ 8
Ͼ1000
20
5
CLC-Ka N68D
TFA
CLC-Ka
CLC-Ka N68D
MFA
CLC-Ka
CLC-Ka N68D
MCFA
CLC-Ka
CLC-Ka N68D
DPC
CLC-Ka
148 Ϯ 22
Ͼ1000
195 Ϯ 47
Ͼ1000
5
4
103 Ϯ 24
Ͼ1000
107 Ϯ 11
Ͼ5000
4
4
43 Ϯ 6
632 Ϯ 101
39 Ϯ 8
172 Ϯ 39
5
4
Fig. 6. Effect of FFA on CLC-Ka. A, voltage-clamp traces of CLC-Ka
currents before and during application of 200 ␮M FFA. B, time course of
the CLC-Ka current blocked by 200 ␮M FFA. Arrow indicates the addi-
tion of drug. C, dose-response relationship of the block at 60 mV by FFA
for CLC-Ka. The line is drawn according to the equation I(c)/I(0) ϭ 1/(1 ϩ
c/KD).
502 Ϯ 135
Ͼ5000
Ͼ5000
Ͼ3000
3
3
CLC-Ka N68D
3-Phenyl-CPP
CLC-Ka
79 Ϯ 5
424 Ϯ 38
184 Ϯ 18
934 Ϯ 87
23
10
CLC-Ka N68D

Fenamates as CLC-K Channel Modulators
171
blocking compounds, such as FFA and 3-phenyl-CPP, dis- rings of FFA derivatives, increased the drug affinity toward
played nonplanar geometries in contrast to NFA (Fig. 9B).
the binding site.
The fact that NFA at high concentration is able to block
CLC-Ka and that FFA is able to open CLC-Kb indicates that
both drugs can bind the activating and blocking binding sites
but with different affinity. The presence of a pyridinic group
favors binding to the activating binding site, although it
cannot be excluded that a certain drug amount binds the
Discussion
In the present work, we demonstrated that in addition to a
blocking site (Picollo et al., 2004), CLC-K chloride channels
also have an activating binding site. Depending on the chem-
ical structure, fenamates are capable of blocking or opening inhibitory binding site at low concentrations. In support of
CLC-Ka. NFA is able to increase CLC-Ka currents at all this, the activating effect of 200 ␮M NFA was more evident
tested membrane potentials in the 10 to 1000 ␮M range with on CLC-Kb, the CLC-K isoform that is much less sensitive to
a rapid onset and a relative slow, but complete, recovery, blockers (Picollo et al., 2004). In parallel, the presence of two
suggesting binding to a site that is accessible from the extra- phenyl groups, as occurs in FFA derivatives, confers a larger
cellular side. The activating effect was markedly increased affinity toward the inhibitory binding site on CLC-Ka. How-
when the extracellular calcium concentration was lowered. ever, in this case, it seems that the same molecules are also
The opposite charge carried by NFA and calcium ions leads capable of interacting with the activating binding site, as
us to exclude a possible competition between these CLC-Ka evidenced by the FFA-induced increase of CLC-Kb currents.
activators for a common binding site. Thus, the increased
In a previous study, we mapped the binding site of 3-phe-
potency of NFA in low calcium is probably simply because nyl-CPP and DIDS on CLC-Ka, pinpointing the neutral
under these experimental conditions, more channels are amino acid Asn68 as crucial for 3-phenyl-CPP-mediated in-
available for NFA activity.
hibition, because mutating it to the negatively charged as-
Starting from NFA as the lead compound, and considering partate (N68D), as occurs in CLC-Kb, markedly reduced drug
all double-ring compounds shown in Fig. 3, the following sensitivity (Picollo et al., 2004). All FFA derivatives were
structural requisites for an efficient activation of CLC-Ka much less potent in blocking CLC-Ka N68D mutant with
emerged: the acidic carboxylic group, two aromatic rings (one respect to CLC-Ka wild type, indicating that Asn68 is pivotal
of which should be a pyridinic ring), a CF3 group in the meta for drug-blocking activity. Furthermore, we showed that FFA
position on the phenyl group, and an anilinic moiety connect- loses its blocking activity on CLC-Kb, producing a potentia-
ing the two rings. Combinations of these requisites enable tion of CLC-Kb. Thus, we speculate that, as in the case of
efficient blocking activity. These include the carboxylic group 3-phenyl-CPP (Picollo et al., 2004), the electrostatic interac-
and the two aromatic moieties linked by an electronegative tion between amino acids located in the external vestibule of
atom. All FFA derivatives produced a rapid and reversible the channel pore and the negatively charged group of FFA
block with a mechanism of action resembling that of 3-phe- derivatives might either permit (CLC-Ka) or impede (CLC-
nyl-CPP. It is noteworthy that the presence of these groups is Kb) the interaction of the inhibitors. At high concentrations,
pivotal for blocking activity also in the 3-phenyl-CPP struc- NFA was able to block CLC-Ka N68D mutant with only
ture. Furthermore, as indicated by the different potency slightly less potency with respect to wild type. This might
shown by the FFA derivatives, in line with what was ob- suggest that the inhibitory effect of NFA and FFA is medi-
served with 3-phenyl-CPP structure modifications (Liantonio ated by different blocking binding sites. On the other hand,
et al., 2004) the presence of an electronegative group, reduc- considering the possibility of a unique blocking binding site,
ing the charge the electronic cloud of one of the aromatic the presence of the pyridinic ring could account for the dif-
Fig. 7. Effect of NFA and FFA on CLC-
Kb. Voltage-clamp traces of CLC-Kb cur-
rents before and during application of 200
␮M NFA (A) and 200 ␮M FFA (B) and
immediately after washout.

172
Liantonio et al.
ferent behavior of NFA with respect to FFA on CLC-Ka N68D
One of the most important questions regarding the NFA
mutant. In fact, the reduced activity of FFA on mutant N68D action is whether the two opposite effects, potentiation and
is probably caused by an electrostatic repulsion (Picollo et al., block, are mediated by drug interaction with two different
2004). The carboxylic group of NFA may have a lower charge binding sites. In contrast to the FFA blocking effect, the
density caused by the protonation of the pyridinic ring, an NFA-induced increase of CLC-Ka currents was reproducible
event that can occur at the experimental pH. This could allow on all mutants described previously for the identification of
NFA to get closer to the blocking binding site than FFA in the 3-phenyl-CPP binding site (Picollo et al., 2004), although
mutant.
with a slightly smaller potency in the case of CLC-Ka N68D.
Overall, these data indicate the presence of two different
binding sites mediating the activating and blocking effects.
However, it should be kept in mind that, as we hypothesized
for the inhibitory binding site (Picollo et al., 2004), the effec-
tive activating binding site could be quite deep within the
pore. Thus, it should not be excluded that the two binding
sites are partially overlapping. The precise identification of
the activator site on CLC-K channels will be needed to con-
clusively resolve this issue.
We attempted to explain the different drug activity by
performing modeling investigations that allowed us to com-
pare the spatial geometry profiles associated at NFA, FFA,
and 3-phenyl-CPP structures. The most important conclu-
sion is that NFA and its analogs MT-4 and MT-7 that behave
as CLC-Ka openers show nearly planar conformations,
whereas FFA derivatives and 3-phenyl-CPP, exhibiting a
CLC-Ka blocking activity, are forced to assume a noncopla-
nar arrangement of the aromatic rings. Overlaying the low-
est energy conformers reveals that all blocking compounds
display nonplanar geometries compared with NFA and that
the increasing planarity distortion of the aromatic rings
ranging from DPC, FFA, MFA, and TFA to MCFA parallels
the observed increasing blocking activity in the same order.
The involvement of CLC-Kb in type III Bartter’ syndrome
and the parallel lack of drug treatment arouse a great inter-
est toward molecules able to open CLC-K channels. The use
of a compound similar to NFA for a treatment of Bartter’s
syndrome could improve diuresis in a direct manner through
an increase of CLC-K channel activity and in an indirect
manner through cyclooxygenase inhibition. Indeed, type III
Bartter’s syndrome patients show elevated prostaglandin ac-
tivity (Reinalter et al., 2002), and NFA belongs to the class of
nonsteroidal anti-inflammatory drugs. However, the activat-
ing effect of NFA on mutants of CLC-Kb has not been tested
in the current study.
Fig. 8. Effect of NFA on CLC-Ka, CLC-Kb and their mutants. Each bar
represents the Idrug/I0 ratio obtained as the ratio of the current in pres-
ence and absence of 200 ␮M NFA for CLC-Ka, CLC-Kb, and relative
mutants.
At the same time, considering the involvement of CLC-K
channels in the mechanisms of urine concentration, specific
inhibitors could represent a new class of drugs with diuretic
activity (Fong, 2004). From this regard, it noteworthy that
whereas CLC-K2/Kb is clearly restricted to basolateral mem-
branes of renal epithelial cells, the localization of CLC-K1/Ka
is still controversial. Both basolateral and apical membranes
(Uchida et al., 1995) or only basolateral localization (Vande-
walle et al., 1997) have been reported. Thus, CLC-K inhibi-
tors may have to reach their targets via the basolateral fluid.
This means that effective concentrations must be reached in
the renal interstitium in vivo for obtaining the diuretic effect,
and the relatively low affinity for the blockers may thus limit
the effectiveness. However, this alternative route could en-
sure drug activity also in some pathological conditions, such
as decreased renal blood flow or renal failure, in which the
therapeutic effectiveness of furosemide-like diuretics could
be seriously compromised.
Fig. 9. Modeling study of the lead compounds. A, lowest energy confor-
mation of MCFA, 3-phenyl-CPP, FFA, and NFA obtained as described
under Materials and Methods. B, overlay of 3-phenyl-CPP (blue), FFA
(green), and NFA (red) lowest energy conformations. The fitting of the
molecules was performed as described under Materials and Methods.
Besides, a polymorphism of CLC-Kb was recently reported

Fenamates as CLC-K Channel Modulators
173
Kokubo Y, Iwai N, Tago N, Inamoto N, Okayama A, Yamawaki H, Naraba H, and
Tomoike H (2005) Association analysis between hypertension and CYBA, CLC-
NKB and KCNMB1 functional polymorphisms in the Japanese population-the
Suita Study. Circ J 69:138–142.
Kong J, White CA, Krylov AI, Sherrill CD, Adamson RD, Furlani TR, Lee MS, Lee
AM, Gwaltney SR, Adams TR, et al. (2000) Q-Chem 2.0: a high-performance ab
initio electronic structure program package. J Computational Chem 21:1532–
1548.
Liantonio A, Accardi A, Carbonara G, Fracchiolla G, Loiodice F, Tortorella P, Tra-
verso S, Guida P, Pierno S, De Luca A, et al. (2002) Molecular requisites for drug
binding to muscle CLC-1 and renal CLC-K channel revealed by the use of phenoxy-
alkyl derivatives of 2-(p-chlorophenoxy)propionic acid. Mol Pharmacol 62:265–
271.
Liantonio A, Pusch M, Picollo A, Guida P, De Luca A, Pierno S, Fracchiolla G,
Loiodice F, Tortorella P, and Conte Camerino D (2004) Investigations of pharma-
cologic properties of the renal CLC-K1 chloride channel co-expressed with barttin
by the use of 2-(p-chlorophenoxy)propionic acid derivatives and other structurally
unrelated chloride channels blockers. J Am Soc Nephrol 15:13–20.
Picollo A, Liantonio A, Didonna MP, Elia L, Camerino DC, and Pusch M (2004)
Molecular determinants of differential pore blocking of kidney CLC-K chloride
channels. EMBO (Eur Mol Biol Organ) Rep 5:584–589.
(T481S) that, in contrast to mutations responsible of Bart-
ter’s syndrome, confers a gain of function. Indeed, T481S
mutant channels led to a dramatic increase in CLC-Kb cur-
rents when expressed in oocytes (Jeck et al., 2004a). This
polymorphism was associated with hypertension in one study
(Jeck et al., 2004b). However, an involvement of this poly-
morphism was not seen in other studies (Kokubo et al., 2005;
Speirs et al., 2005). Thus, a specific inhibition of CLC-K
channels as a treatment of such a condition remains specu-
lative. Although the CLC-K inhibitors identified up to now
showed a major affinity toward the CLC-Ka isoform, they
represent drugs with good therapeutic potential, permitting
a reduction of CLC-Ka activity for counteracting the exces-
sive CLC-Kb activity.
Pusch M, Liantonio A, Bertorello L, Accardi A, De Luca A, Pierno S, Tortorella V, and
Conte Camerino D (2000) Pharmacological characterization of the chloride chan-
nels belonging to the ClC family by the use of chiral clofibric acid derivatives. Mol
Pharmacol 58:498–507.
Reinalter SC, Jeck N, Brochhausen C, Watzer B, Nusing RM, Seyberth HW, and
Komhoff M (2002) Role of cyclooxygenase-2 in hyperprostaglandin E syndrome/
antenatal Bartter syndrome. Kidney Int 62:253–260.
Acknowledgments
We thank Thomas Jentsch for supplying all cDNA clones, Enrico
Gaggero for the construction of the voltage-clamp amplifier, and
Giacomo Gaggero for help in constructing the recording chambers.
Schlingmann KP, Konrad M, Jeck N, Waldegger P, Reinalter SC, Holder M, Sey-
berth HW, and Waldegger S (2004) Salt wasting and deafness resulting from
mutations in two chloride channels. N Engl J Med 350:1314–1319.
Sherlock MH, Kaminski JJ, Tom WC, Lee JF, Wong S-C, Kreutner W, Bryant RW,
and McPhail AT (1988) Antiallergy agents. 1. Substituted 1,8-naphthyridin-2(1H)-
ones as inhibitors of SRS-A release. J Med Chem 31:2108–2121.
Simon DB, Bindra RS, Mansfield TA, Nelson-Williams C, Mendonca E, Stone R,
Schurman S, Nayir A, Alay H, Bakkaloglu A, et al. (1997) Mutations in the
chloride channel gene, CLCNKB, cause Bartter’s syndrome type III. Nat Genet
17:171–178.
Speirs HJ, Wang WY, Benjafield AV, and Morris BJ (2005) No association with
hypertension of CLCNKB and TNFRSF1B polymorphisms at a hypertension locus
on chromosome 1p36. J Hyperten 23:1491–1496.
Uchida S and Sasaki S (2005) Function of chloride channels in the kidney. Annu
Rev Physiol 67:759–778.
Uchida S, Sasaki S, Nitta K, Uchida K, Horita S, Nihei H, and Marumo F (1995)
Localization and functional characterization of rat kidney-specific chloride chan-
nel, ClC-K1. J Clin Investig 95:104–113.
Vandewalle A, Cluzeaud F, Bens M, Kieferle S, Steinmeyer K, and Jentsch TJ (1997)
Localization and induction by dehydration of ClC-K chloride channels in the rat
kidney. Am J Physiol 272:F678–F688.
Villani FJ, Mann TA, Wefer EA, Hannon J, Larca LL, Landon MJ, Spivak W, Vashi
D, Tozzi S, Danko G, et al. (1975) Benzopyranopyridine derivatives. 1. Aminoalkyl
derivatives of the azaxanthenes as bronchodilating agents. J Med Chem 18:1–8.
Waldegger S, Jeck N, Barth P, Peters M, Vitzthum H, Wolf K, Kurtz A, Konrad M,
and Seyberth HW (2002) Barttin increases surface expression and changes current
properties of CLC-K channels. Pfleug Arch Eur J Phsyiol 444:411–418.
References
Akizuki N, Uchida S, Sasaki S, and Marumo F (2001) Impaired solute accumulation
in inner medulla of CLCNK1 Ϫ/Ϫ mice kidney. Am J Physiol 280:F79–F87.
Birkenha¨ger R, Otto E, Schurmann MJ, Vollmer M, Ruf E-M, Maier-Lutz I, Beek-
mann F, Fekete A, Omran H, Feldmann D, et al. (2001) Mutation of BSND causes
Bartter syndrome with sensorineural deafness and kidney failure. Nat Genet
29:310–314.
Dhanaraj V and Vijayan M (1988) Structural studies of analgesics and their inter-
actions. XII. Structure and interactions of anti-inflammatory fenamates. A con-
certed crystallographic and theoretical conformational study. Acta Crystallogr B
44:406–412.
Este´vez R, Boettger T, Stein V, Birkenha¨ger R, Otto E, Hildebrandt F, and Jentsch
TJ (2001) Barttin is a Cl channel beta-subunit crucial for renal Cl reabsorption and
inner ear K secretion. Nature (Lond) 414:558–561.
Fong P (2004) CLC-K channels: if the drug fits, use it. EMBO (Eur Mol Biol Organ)
Rep 5:565–566.
Fujiwara H and Kitagawa K (2000) Synthesis of pyridopyranoquinolines by the
Skraup reaction of amino-5H-benzopyrano[2,3-b]pyridine-5-ones. Heterocycles 53:
409–417.
Jeck N, Waldegger P, Doroszewicz J, Seyberth H, and Waldegger S (2004a) A
common sequence variation of the CLCNKB gene strongly activates ClC-Kb chlo-
ride channel activity. Kidney Int 65:190–197.
Jeck N, Waldegger S, Lampert A, Boehmer C, Waldegger P, Lang PA, Wissinger B,
Friedrich B, Risler T, Moehle R, et al. (2004b) Activating mutation of the renal
epithelial chloride channel ClC-Kb predisposing to hypertension. Hypertension
43:1175–1181.
Jentsch TJ (2005) Chloride transport in the kidney: lessons from human disease and
knockout mice. J Am Soc Nephrol 16:1549–1561.
Jentsch TJ, Stein V, Weinreich F, and Zdebik AA (2002) Molecular structure and
physiological function of chloride channels. Physiol Rev 82:503–568.
Kieferle S, Fong P, Bens M, Vandewalle A, and Jentsch TJ (1994) Two highly
homologous members of the ClC chloride channel family in both rat and human
kidney. Proc Natl Acad Sci USA 91:6943–6947.
Address correspondence to: Prof. Diana Conte Camerino, Section of Phar-
macology, Department of Pharmacobiology, Faculty of Pharmacy, University
of Bari, Via Orabona, 4, Campus, I-70125 Bari, Italy. E-mail: conte@
farmbiol.uniba.it