Anti-calmodulins and Tricyclic Adjuvants in Pain Therapy

Block the TRPV1 Channel

Zoltán Oláh1,2,3*, Katalin Jósvay1, László Pecze1, Tamás Letoha2,4, Norbert Babai5, Dénes Budai6, Ferenc Ötvös1, Sándor Szalma3, Csaba Vizler1

1 Institute of Biochemistry, Biological Research Center of the Hungarian Academy of Sciences, Szeged, Hungary, 2 Acheuron Hungary Ltd., Szeged,
Hungary, 3 Acheuron Pharmaceuticals Inc., San Diego, California, United States of America, 4 Department of Medical Chemistry, Faculty of General
Medicine, University of Szeged, Szeged, Hungary, 5 Department of Experimental Zoology and Neurobiology, University of Pécs, Pécs, Hungary,
6 Department of Biology, Gyula Juh{sz Faculty of Education, University of Szeged, Szeged, Hungary

Ca2+-loaded calmodulin normally inhibits multiple Ca2+-channels upon dangerous elevation of intracellular Ca2+ and protects
cells from Ca2+-cytotoxicity, so blocking of calmodulin should theoretically lead to uncontrolled elevation of intracellular Ca2+.
Paradoxically, classical anti-psychotic, anti-calmodulin drugs were noted here to inhibit Ca2+-uptake via the vanilloid inducible
Ca2+-channel/inflamatory pain receptor 1 (TRPV1), which suggests that calmodulin inhibitors may block pore formation and
Ca2+ entry. Functional assays on TRPV1 expressing cells support direct, dose-dependent inhibition of vanilloid-induced 45Ca2+-
uptake at mM concentrations: calmidazolium (broad range)$trifluoperazine (narrow range).chlorpromazine/amitriptyline.-
fluphenazine..W-7 and W-13 (only partially). Most likely a short acidic domain at the pore loop of the channel orifice
functions as binding site either for Ca2+ or anti-calmodulin drugs. Camstatin, a selective peptide blocker of calmodulin, inhibits
vanilloid-induced Ca2+-uptake in intact TRPV1+ cells, and suggests an extracellular site of inhibition. TRPV1+, inflammatory
pain-conferring nociceptive neurons from sensory ganglia, were blocked by various anti-psychotic and anti-calmodulin drugs.
Among them, calmidazolium, the most effective calmodulin agonist, blocked Ca2+-entry by a non-competitive kinetics,
affecting the TRPV1 at a different site than the vanilloid binding pocket. Data suggest that various calmodulin antagonists
dock to an extracellular site, not found in other Ca2+-channels. Calmodulin antagonist-evoked inhibition of TRPV1 and NMDA
receptors/Ca2+-channels was validated by microiontophoresis of calmidazolium to laminectomised rat monitored with
extracellular single unit recordings in vivo. These unexpected findings may explain empirically noted efficacy of clinical pain
adjuvant therapy that justify efforts to develop hits into painkillers, selective to sensory Ca2+-channels but not affecting
motoneurons.
Citation: Oláh Z, Jósvay K, Pecze L, Letoha T, Babai N, et al (2007) Anti-calmodulins and Tricyclic Adjuvants in Pain Therapy Block the TRPV1
Channel. PLoS ONE 2(6): e545. doi:10.1371/journal.pone.0000545

INTRODUCTION the cytosol resulting in elevation of intracellular free calcium

Recently, we and several other groups noted that TRPV1, the [Ca2+]i can turn on/off different enzymes and ion channels.
vanilloid ligand gated member of the TRP (transient receptor Camstatin, a recently found conserved polypeptide motif in PEP-
potential) super family, localizes both in the plasma membrane 19, neurogranin and neuromodulin, has been noted to enhance
(TRPV1PM) and endoplasmatic reticulum (TRPV1ER) membranes dissociation of Ca2+ from calmodulin [20] and disable interaction
[1–4], and upon ligand binding they release Ca2+ to the cytosol. with down-stream targets [21]. These observations suggest that
These pools are inducible with: i) exo-, or endovanilloid ligands one of the major functions of calmodulin would be to buffer and/
[5,6], ii) proton [7,8], iii) phosphorylation via intracellular signaling or neutralize the rapid increase of [Ca2+]i, thus to prevent
by protein kinases [9–11] and iv) heat (42–49uC) [12,13]. excitotoxicity. Serving as a shut off valve on broad-spectra of Ca2+-
Dynamics of TRP cation channel opening and closing has channels, Ca2+- calmodulin protects Ca2+-overload-induced cell
remained largely unknown, although various mechanisms have death, either due to necrotic or apoptotic mechanisms. The
been proposed [11,14]. Due to lack of purification and specific intracellular sites has been identified but the exact
crystallization protocol the three-dimensional (3D) structure mechanism of calmodulin binding is still debated [22,23].
information is scarce, the initial conformation change and
subsequent steps leading to pore opening/closing has yet to be
elucidated. Better understanding of transmembrane proteins is Academic Editor: Leslie Vosshall, The Rockefeller University, United States of
America
hampered by process development such as high scale production,
solubilization, and purification, which preserves the native state Received March 7, 2007; Accepted May 24, 2007; Published June 20, 2007
and function. As in case of many transmembrane domain
Copyright: ß 2007 Olah et al. This is an open-access article distributed under the
proteins/ion channels, 3D structure of TRPV1 is still subject of terms of the Creative Commons Attribution License, which permits unrestricted
intense research [15,16]. use, distribution, and reproduction in any medium, provided the original author
Calmodulin is the most well-known calcium binding protein, and source are credited.
which is ubiquitous and preserved in the eukaryotic cell, either Funding: ZO was supported by MC-IRG030854-PAINKILLER; Anyos Jedlik Program
human, animal, fungal or plant. Ca2+, among others, selectively NKFP-1-00019/2005; GVOP-3.3.1-05/1.-2005-05-0057/3.0. DB was also supported
interacts with the so-called ‘‘EF-hand’’/Ca2+-binding protein by GVOP, CV was also supported by grants the National Office for Research and
Technology (OM-00051/2005 and OMFB-01575/2006) and the Hungarian Ministry
motifs located either intracellularly or extracellularly [17]. Dose- of Health (552/2006).
dependent interaction of Ca2+ with calmodulin elicits a robust
conformational change that exposes hidden hydrophobic domains Competing Interests: The authors have declared that no competing interests
exist.
required for subsequent effects on down-stream protein targets
[18,19]. Ca2+- calmodulin complex, formed upon entry of Ca2+ to * To whom correspondence should be addressed. E-mail: [email protected]

PLoS ONE | www.plosone.org 1 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

The Ca2+- calmodulin mediated feedback due to increased TRPV1 expressing cells and neurons. Extracellularly added
[Ca2+]i, has recently been elucidated in detail in case of the TRP3 camstatin peptide, an antagonist of calmodulin, also blocked
channel. It has been noted that upon Ca2+-depletion, IP3R, capsaicin–induced Ca2+-uptake in intact cells. Studies carried out
a sensor of Ca2+-load of ER, directly interacts and props the at the cellular levels and in animal pain models were in concert
TRP3PM channel open by the so-called ‘‘store operated Ca2+- with previously noted analgesic actions of calmodulin-antagonists
entry’’ mechanism. Indeed, one or two specific domains of IP3R determined in vivo [32], but we gave here new evidence that
can interact with cognate sites of TRP3PM and contribute to calmodulin antagonists can directly block ion channel function of
opening of the pore. However, both Ca2+- calmodulin and TRPV1. We suggest that the analgesic effect of these calmodulin
cytoplasmic domain of IP3RER competes for an overlapping site antagonists is independent of conventional intracellular targets and
and either open or close the given TRP channel, respectively and we propose a specific site of action located extracellularly [23].
the preference only depends on the levels of [Ca2+]i. In fact, Pharmacophores of various calmodulin inhibitors studied here
calmodulin, upon saturation with Ca2+ displaces IP3RER, which may yield discovery of novel analgesics in the near future [33].
leads to termination of store operated Ca2+ entry. However, Ca2+-
calmodulin can be displaced by excess synthetic peptides, derived RESULTS
either from the competitive IP3R motif or from the heterologous To better understand potential effect of Ca2+- calmodulin
myosin light chain kinase. The former is known to block IP3RER inhibitors on nociception, activity of various phenothiazines were
binding to TRP3PM by direct competition, whereas, cognate studied at the molecular levels in TRPV1-NIH 3T3 cells, in which
domain from myosin light chain kinase, as well as calmidazolium, the pain signal was mimicked with capsaicin-induced 45Ca2+-
inhibit the interaction indirectly, due to prevention of Ca2+ loading uptake (Fig. 1). Assays were carried out in 261025 M extracellular
of calmodulin. It is conceivable that either mechanism can serve as Ca2+ in 96 well plate formats at room temperature (22uC) for
a shut off valve of TRP3PM. In general, either disruption of a TRP- 10 min with robotic liquid handling. Briefly, cells were co-
Ca2+-channel interaction or block of Ca2+-feedback by anti- incubated with capsaicin for vanilloid-induced opening of TRPV1.
calmodulin agents can deregulate store operated Ca2+ entry and Various Ca2+- calmodulin inhibitors were co-incubated in pro-
cause eventually excitotoxicity and cell death by Ca2+-overload gressively increasing concentration in the 2 mM capsaicin
[24,25]. Indeed, application of calmidazolium to HL-60 cells has supplemented incubation medium. Increased Ca2+-uptake was
recently been shown to increase [Ca2+]i, which is consistent with expected, due to a severe damage of shut off valve mechanism
disrupted Ca2+- calmodulin feedback regulation [26]. Ca2+- caused by the presence of calmodulin antagonists. Contrary to this
calmodulin-mediated termination of Ca2+-entry is not confined to theory, detailed referring to other TRP channels in the in-
TRP channels only [27], rise of [Ca2+]i also shuts off M-, and L-type troduction, we noted that the tested calmodulin inhibitors
voltage-gated channels. Opening of Ca2+-activated, small conduc- inhibited the capsaicin-induced Ca2+-uptake in TRPV1-NIH3T3
tance K+ (SK) channels is elicited by calmodulin binding to the C- cells with varying potency: calmidazolium (IC50 = 7 mM).trifluo-
terminus. Further elevation of [Ca2+]i and saturation of calmodulin perazine (IC50 = 9 mM).chlorpromazine (IC50 = 70 mM).W-7
with Ca2+ inactivate the pore opening mechanism [28,29]. (IC50,200 mM).fluphenazine (IC50,250 mM), and W-13
Consistent with this Ca2+-feedback theory, anti- calmodulin (IC50.300 mM), and [Ca2+]i did not accumulate in these cells at
agents, such as calmidazolium are expected to suspend the even higher concentrations of calmodulin inhibitors. The same
feedback on Ca2+-entry, and thus potentiate elevation of [Ca2+]i to agents were also tested in the absence of capsaicin, where none of
toxic levels [30]. Classical antipsychotic drugs, such as tri- them induced Ca2+ influx, i.e., they did not activate the vanilloid
fluoperazine, chlorpromazine and fluphenazine also target cal- receptor. For the sake of clarity these curves were omitted from
modulin, so they were also hypothesised to deregulate agonist- Fig 1. and the following figures also (Fig 2–6). Efficacy of
induced Ca2+ increase. To test this hypothesis and better calmidazolium in the mM scale was also validated in a human
understand channel function of TRPV1, we carried out experi- TRPV1-HaCaT keratinocyte line (data not shown). Interestingly,
ments with tricyclic anti-psychotic calmodulin antagonist drugs fluphenazine, but less pronouncedly both W13 and W7 increased
and other selective inhibitors of calmodulin. Our new data, in lieu Ca2+-uptake in the presence of 2 mM capsaicin at low concentra-
of expectations, suggest that bona fide calmodulin antagonists blocks tions. Similar cooperative effect was noted previously between
and do not promote Ca2+-transport via TRPV1. [3H] resiniferatoxin and different antipsychotic and antidepressant
To address paradoxical effects of anti- calmodulin agents, we drugs [34]. We hypothesized that calmodulin antagonists and
hypothesized that these compounds might not enter the cell, but some registered drugs may exert their blocking effect with another
inhibit the pore opening of pain/TRPV1 channel, most likely mechanisms on TRPV1 and not necessarily on their conventional
directly at the cation filter site, located at the extra-cellular orifice target. Acting extracellularly, the inhibitors prevent TRPV1
of the Ca2+-channel. In fact, sequence comparisons of various activation, therefore the intracellular calmodulin might be
TRP channels suggest that a short acidic amino acid stretch at the secondary and has little if any effect on [Ca2+]i signal induced
pore loop of TRPV1 resembles to ‘‘EF-hand’’ and ‘‘Excalibur’’ by inflammatory pain agonists either exo-, or endogenous.
motifs, both identified previously in calmodulin and calmodulin- To better define the quantitative structure-activity relationship
like proteins, respectively [31]. Since direct sequence homology (qSAR), other miconazole compounds, analogs of the commer-
was not that obvious, structure and function of TRPV1 was cially available calmidazolium, were functionally assayed in cells
probed with various anti-calmodulin agents such as antipsychotic expressing TRPV1. A small dedicated library of miconazoles,
drugs and homologues with basic pharmacophore similar to that clotrimazole and N-benzylimidazole compounds, referred to as the
identified in calmidazolium. Functional assays were carried out in M-set (see M1–M7 in Fig. 2) were tried. All of these analogues of
TRPV1-NIH3T3 cells permanently expressing the receptor ectop- calmidazolium originally were synthesized to develop inhibitors of
ically, and in primary cultures from embryonic rat dorsal root L-type Ca2+-channels [26]. Dose-response analysis of capsaicin-
ganglia (DRG), enriched in TRPV1+ nociceptor/pain neurons. induced Ca2+-transport carried out in TRPV1-NIH 3T3 cells
Consistent with our hypothesis, but contrary to the Ca2+- eliminated N-benzylimidazole and clotrimazole compounds form
calmodulin feedback theory, a set of anti-calmodulin agents further studies, since both groups showed either only partial
acutely inhibited the vanilloid/capsaicin–induced Ca2+-uptake in inhibition or less favorable IC50 than calmidazolium and other

PLoS ONE | www.plosone.org 2 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

Figure 1. Antagonist activity of Ca2+-calmodulin inhibitors in TRPV1-NIH3T3 cells. 45Ca2+-uptake experiments were carried out in 96 well plates
with robotic liquid handling. Cells were co-incubated with capsaicin (capsaicin, ED200 = 2 mM) in the presence of different concentrations of Ca2+-
calmodulin inhibitors. Calmidazolium (CMZ).trifluoperazine (TFP).chlorpromazine (CPZ) were identified as full antagonists of capsaicin-induced
Ca2+-uptake, in the micromolar range, while fluphenazine (FluPhe), W7 and W13 were determined as partial or weak inhibitors. Similar efficacy order
was determined in two additional experiments, carried out in duplicate samples.
doi:10.1371/journal.pone.0000545.g001

Figure 2. Screening of other calmidazolium analogs (M-set). Among miconazoles, clotrimazole, and N-benzylimidazole compounds (M1–M7)
calmidazolium was the most potent inhibitor in TRPV1-NIH3T3 cells. Each point on the graph is the average of triplicate determinations. Chemical
structure of miconazoles, clotrimazole, and N-benzylimidazole compounds used in these studies are indicated. Experiments were repeated two
additional times in triplicate with similar results.
doi:10.1371/journal.pone.0000545.g002

PLoS ONE | www.plosone.org 3 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

Figure 3. Kinetics of Ca2+-transport inhibition by calmidazolium (CMZ), chlorpromazine (CPZ) and capsazepine (CAZ) in TRPV1-NIH3T3 cells,
induced by increasing concentrations of capsaicin. Increasing doses of calmidazolium decreased Vmax of Ca2+-transport, however, affinity of TRPV1
to capsaicin remained constant, as indicated. Likewise, chlorpromazine showed distinctive, non-competitive inhibition kinetics, similar to that
determined to calmidazolium, consistent with a channel blocking mechanism on TRPV1. In contrast to calmidazolium and chlorpromazine,
capsazepine, a bona fide vanilloid antagonist shifted the capsaicin dose-response curves right, however, above 1 mM behaved as a mixed kinetics
inhibitor, also decreased Vmax. Experiments were repeated two additional times in duplicates with similar results.
doi:10.1371/journal.pone.0000545.g003

miconazoles (i.e. M1/M2 = 8 mM.M3,75 mM) homologues calmodulin peptide was coincubated with capsaicin in the Ca2+-
(Fig. 2). uptake medium. Likewise conventional calmodulin inhibitors,
Further kinetic analysis of calmidazolium addressed the camstatin inhibited capsaicin -induced Ca2+-transport (IC50 =
potential mechanism of inhibition in TRPV1 expressing cells. A 11 mM) within 10 min of the assay in TRPV1-NIH3T3 cells.
dramatic decrease in the maximal velocity (Vmax) of capsaicin- Prompt inhibition of inducible Ca2+-uptake suggested a rapid
induced Ca2+-uptake, but not the ED50 of capsaicin (0.3 mM of interaction with a potential extracellular docking site(s) of TRPV1
capsaicin) was determined. With progressively increasing concen- (Fig. 4). As a control, CAMKII kinase inhibitor peptide, similar in
trations of calmidazolium the Vmax gradually decreased, which size to camstatin but distinct in biological activity, did not block
also showed a characteristic maxima instead of a plateau due to capsaicin-induced Ca2+-uptake (data not shown). Experiments
a potential interaction with capsaicin on TRPV1. Above with peptides suggested that camstatin and other anti-calmodulin
a calmidazolium concentration of = 2.5 mM, Ca2+-transport was agents indeed recognize an extracellular domain of TRPV1.
almost completely abolished (Fig. 3a). Distinctive drop in Vmax To make initial structure activity relationship studies of TRPV1
were prominent at low concentrations of calmidazolium, which is inhibitors more complete and check efficacy of a clinically tried
rather consistent with a channel blocking mechanism, than drugs, amitriptyline, a known antidepressant in human [36],
competition for the capsaicin-binding site. Likewise calmidazo- gabapentine, a recently commercialized painkiller drug, and
lium, chlorpromazine showed distinctive, non-competitive in- carbamazepine, an amitriptyline analog prescribed for patients
hibition kinetics of Vmax (Fig. 3b). with chronic back pain were tried in capsaicin-induced Ca2+-
Capsazepine, a known competitive inhibitor at the intracellular uptake experiments. Among these substances amitriptyline, a struc-
vanilloid binding site, typically shifts the dose-response curve right ture analogue of phenothiazines, was determined the best inhibitor
(Fig. 3c), but at low concentrations does not affect Vmax, similar to (IC50,60 mM), however, its effect was significantly less prominent
that determined previously by a non-penetrating ligand of TRPV1 than that of trifluoperazine (IC50,10 mM). Carbamazepine, with
[35]. Inhibition kinetics of calmidazolium and chlorpromazine shorter side chain than that in amitriptyline was completely
were markedly differed from that determined to capsazepine, inactive in this assay, while gabapentine showed partial inhibition
a bona-fide vanilloid mimetic antagonist. A decline in Vmax, relative (Fig. 5). SAR of these drugs highlights the role of the aliphatic
to a plateau might be a sign of allosteric interaction of calmida- extension tethering the tricyclic ring of these types of analogs.
zolium and chlorpromazine with other domain(s) (Fig. 3a and 3b) To test selectivity and efficacy in an in vivo target, calmidazo-
that may affect access of capsaicin to TRPV1. lium, the best channel blocker of capsaicin-induced Ca2+-uptake in
From these experiments we hypothesized that maybe a calmod- TRPV1-NIH3T3 cell was tried in sensory neurons, expressing
ulin-like motif of TRPV1 is recognized extracellularly, therefore, TRPV1 endogenously. Primary cultures were prepared from
a calmodulin antagonist with poor membrane permeability can embryonic (E14) rat DRGs. Along with calmidazolium other
exert inhibition without entering the cell. To address extracellular polycyclic compounds were included in the vanilloid-induced
45
targeting of TRPV1, camstatin, a recently identified, selective anti- Ca2+-transport assays. SB 290157, a selective, high affinity,

PLoS ONE | www.plosone.org 4 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

Figure 5. Activity of other drugs, similar to chlorpromazine and

trifluoperazine, on TRPV1. Amitriptyline (AMI), carbamazepine and
gabapentine were tried in cell based capsaicin-induced Ca2+-uptake
experiments. Among these substances amitriptyline, a tricyclic ana-
logue of phenothiazines, was the best inhibitor, however, carbamaze-
pine with shorter side chain than amitriptyline was determined inactive.
Gabapentine showed only partial inhibition on TRPV1. Similar results
Figure 4. Effect of a non-membrane permeable inhibitor of were obtained in two additional experiments carried out in triplicate.
calmodulin on TRPV1-mediated Ca2+-transport. Camstatin, a recently doi:10.1371/journal.pone.0000545.g005
identified, selective, 25-mer polypeptide blocker of calmodulin, was
employed, which inhibited capsaicin-induced Ca2+-uptake with com-
parable activity (IC50 = 20 mM) determined previously to other, more (Fig. 7). The calmidazolium evoked inhibition reached its
conventional antagonists, of calmodulin in TRPV1-NIH3T3 cells. Results maximum within 10 to 15 min after ejection and lasted
with camstatin suggests calmodulin-like structure at the extracellular throughout the entire recording periods (about 2 hrs) of the
domains of TRPV1. Experiments were repeated two additional times in
experiment. The monitored cells had excitatory receptive fields
triplicate with similar results.
doi:10.1371/journal.pone.0000545.g004 located on one of the hind paws. These cells usually had little
spontaneous activity, located between 100–600 mm from the
surface of the dorsal horn, as estimated by microdrive readings.
competitive antagonist of the anaphylatoxin C3a receptor, CGP-
Pain signaling, evoked by noxious heat to cutaneous receptive field
37157, a cell-permeable benzodiazepine inhibitor of mitochondri-
of the paw were most likely transduced by TRPV1 expressing
al Na+/Ca2+ exchanger, and flunarizine, a piperazine derivative
nociceptive neurons but occurrence of other heat sensors, such as
blocker of T-type cardiac Ca2+ channels were included in assays in
TRPV2, the super heat receptor, could not be rule out. Likewise to
DRG cultures. All of them were far less potent than calmidazo- that determined previously in TRPV1-NIH3T3 cells, TRPV1-
lium. Moreover, calmidazolium inhibited capsaicin-induced Ca2+- HaCaT keratinocyte lines and primary culture of nociceptive
uptake in sensory neuron cultures with a higher potency than that neurons, calmidazolium inhibited the ratemeter recorded signals
was determined in TRPV1-NIH3T3 cells (i.e. 0.6 mM vs. ,2 mM, of radiation heat pain in the rat (Fig. 7). Iontophoretically ejected
respectively) (Fig. 1 and Fig. 6). Although CGP-37157 and calmidazolium to the site of recordings validated potent inhibitor
flunarizine were determined partially active, in a relative broad activity of a calmodulin antagonist in vivo, in the spinal dorsal horn
concentration range, calmidazolium was determined far more neurons that noted previously in TRPV1 expressing cells.
potent (,3 fold) inhibitor in DRG cultures enriched in nociceptive In another set of experiments, effects of calmidazolium were
neurons with TRPV1 immunoreactivity (data not shown) (Fig. 6). investigated on the responses of dorsal horn neurons evoked by
To validate in vivo efficacy of calmidazolium, experiments were NMDA and kainic acid agonists of different glutamate receptors,
carried out in a total of 22 nociceptive specific (NS) or wide well characterized transducers of pain signals, more centrally.
dynamic range (WDR) spinal dorsal horn neurons in 16 lightly Neuronal firing responses to iontophoresed NMDA or kainic acid
anesthetized rats. Effects of calmidazolium were studied on the were typically fast in onset and short in duration (Fig. 8). Ejection
heat evoked responses of dorsal horn neurons using noxious currents for these excitatory agents were delivered alternately in
peripheral heat stimuli delivered in every 3 min in the cutaneous every 2 min and were selected to produce a maximal peak height
receptive field of the hind paw. After establishing stable baseline of 50–70 spikes/s which corresponded to 300–400 action
control responses, calmidazolium was iontophoresed (i.e. with 100 potentials per stimulation epoch. Iontophoretic application of
nA) from one of the barrels of the compound electrode. Heat- calmidazolium produced significant reduction in responses to both
evoked responses of dorsal horn neurons were significantly NMDA and kainic acid. Maximal inhibition of NMDA- or kainic
reduced the responses to peripheral noxious heat to a mean of acid -evoked responses were reached 15–20 min after calmidazo-
34623% (p,0.01, n = 6) of pre- calmidazolium baseline control lium application and remained so during the rest of the 2 hour

PLoS ONE | www.plosone.org 5 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

Figure 6. Efficacy of Ca2+- calmodulin inhibitors in rat primary DRG cultures. Neuron cultures derived from embryonic rat DRGs readily show
inducible (10 fold over base line) activation with vanilloids due to endogenous expression of TRPV1. As with the TRPV1-NIH 3T3 cells, calmidazolium
was determined the most effective inhibitor of capsaicin-induced Ca2+-uptake. As controls of specificity, flunarizine, SB 290157, and CGP 37157
showed only partial blocking activity in DRG neurons. Similar results were obtained in two additional experiments carried out in triplicate.
doi:10.1371/journal.pone.0000545.g006

long experiment. Responses to NMDA were more pronounced, screening of drugs available now (Figs. 10 and 11) or new drugs
reduced the initial pain signals to 13612% (p,0.01, n = 16), can reach the clinic sooner. For better specificity, however, we
however, responses to kainic acid were moderate, inhibited the need to redirect drugs from the pleiotropic calmodulin target to
pre- calmidazolium baseline control responses to 47620% docking site at the pore loop of TRPV1 (Figs. 10 and 11).
(p,0.01, n = 16). There was a significant difference between the According to our results obtained in our TRPV1-NIH3T3/
decreases of responses to NMDA versus kainic acid in the presence HaCaT and DRG cell-based functional assays, as well as, in
of calmidazolium (13612% versus 47620%, respectively, laminectomised rat models, among clinically used pain adjuvant
p,0.05) (Fig. 9). medications, either trifluoperazine (i.e. TerfluzineH) or chlorprom-
azine (i.e. ThorazineH) would be a better choice than amitripty-
DISCUSSION line. Amitriptyline and fluphenazine showed almost an order of
Of important finding of these studies is that classical phenothia- magnitude less blocking efficacy among the registered drugs on
zines/antipsychotic drugs, all of them calmodulin inhibitors, also TRPV1. TerfluzineH would be an adjuvant of choice in therapeutic
can serve as blocker of current via the Na+/Ca2+-channel of protocols to potentiate either efficacy of opioids or NSAIDs,
TRPV1, the specific transducer of heat/vanilloid-induced in- although, some other factors, for example inadvertent side effects,
flammatory pain. Among these drugs amitriptyline (i.e. ElavilH/ specificity to calmodulin and ADMETox characteristics can modify
TeperinH) has clinically been tried in adjuvant therapy, and noted optimal selection. Calmidazolium, a miconazole type calmodulin
effective in post-herpetic neuralgia and painful diabetic neurop- antagonist was ranked to the 1st place. We showed in laminectomised
athy [37–39]. Although previously was validated in clinical pain, rat models that calmidazolium inhibits pain response to noxious
amitriptyline (IC50 = 60 mM) not ranked among the most potent heat, moreover firing evoked by NMDA and kainic acid. In the
inhibitors of TRPV1 channel in our cell-based assay. spinal dorsal horn calmidazolium inhibits other channels than the
Migraine might be treated with rationally chosen calmodulin heat and inflammatory pain channel and it is active on NMDA and
antagonist drugs currently available. Novel drug candidates such kainic acid receptors that usually prolong the duration of pain state.
as calmidazolium analogs (M-set) may even more promising in this Calmidazolium, a not yet registered drug candidate, has previously
respect. Even when a patient is not clinically depressed, been administered intrathecally in rat pain models and noted indeed
antipsychotic drugs and tricyclic antidepressants could be analgesic [32]. Although was effective on the intrathecal route, some
administered to fight with various forms of neuropathic pain. inadvertent action on locomotion has also been revealed. For human
Posttraumatic sympathetic dystrophy, postmastectomy pain, post- use, calmidazolium needs additional structural modifications, both
herpetic neuralgia, some forms of cancer pain, and other variants with computational drug discovery and medical chemistry means
of neuropathic pain syndromes maybe addressed with novel [40].
adjuvant treatment protocols, planned on our cell-based TRPV1 Calmodulin kinase II (CaMKII) mediated intracellular signaling
assays, which may help to select among a number of empirically has recently been recognized to contribute to inflammation, pain
employed drugs (30+) and rationalize adjuvants’ selection. and hyperalgesia [41] and activate TRPV1 in primary afferent
Additional lead optimization may be enhanced with in silico neurons at the levels of spinal cord [11]. Although chemically

PLoS ONE | www.plosone.org 6 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

afferents sensory neurons. Coupling between Ca2+ induced Ca2+

release and store operated Ca2+ entry is now a well accepted
hypothesis, therefore, [Ca2+]o is an ideal test-bead for searching
specific inhibitors affecting either one or both the TRPV1PM and
TRPV1ER pools. Contrary to that noted with other TRP channels,
calmodulin antagonists do not eliminate Ca2+-feedback. Even
more, an initial observation with the camstatin peptide prompted
us to hypothesise that antipsychotic inhibitors of calmodulin
eliminate vanilloid-induced Ca2+-uptake due to binding to an
extracellular domain of TRPV1. Likewise, calmidazolium may
block noxious heat-, NMDA- and kainic acid -induced pain
signaling by a mechanism that does not need internalization of the
compound in laminectomised rat. Of importance, calmidazolium
has explicit positive charge, and electrostatic interactions may be
required for specific docking at the pore loop of TRPV1 (Figs. 10
and 11). Classical antipsychotic drugs, calmidazolium and
camstatin may interact with TRPV1 on folds common in these
Ca2+-binding proteins (i.e., at the Ca2+-filter/binding region).
By comparative homology search we identified a highly acidic
domain near to the orifice of the Na+/Ca2+ channel. The
646
DLEFTENYD acidic tetrad of TRPV1 is unique and more
extended stretch than that occurs in other TRPs in the TRPV1
(Fig. 10a). This motif is similar to the EF-hand (DXDXDGXXDXXE)
and ‘‘Excalibur’’ (DXDXDXXXXE), both of them Ca2+ binding
consensus domains, identified previously in calmodulin and bacterial
calmodulin-like proteins, respectively [31]. Indeed, this acidic tetrad
is targeted by other positively charged inhibitors such as the R4W2
Figure 7. Heat-induced pain signal inhibition of the best Ca2+- peptide and ruthenium red. Both are well characterized inhibitors of
calmodulin inhibitor in the rat. Ratemeter recordings showing the capsaicin-induced Ca2+-current via TRPV1 channel. Our theory is
inhibitory effects of calmidazolium (CMZ) on responses of a spinal that a short segment of the pore loop may mimic some feature of
dorsal horn neuron to noxious heat delivered to its cutaneous receptive Ca2+-binding proteins where anti-calmodulin drugs dock and inhibit
field on the paw. Panel A: Heat-evoked responses evoked in every 3 min
during the control period.. Panel B: Effects of iontophoresed CMZ on down-stream signaling (Fig. 10). Ruthenium red is widely used as
the heat-evoked responses. Similar results were obtained in five a channel blocker in electrophysiology of sensory neurons, but only
additional experiments. A representative recording is shown. just recently has been mapped to bind at the acidic tetrad of TRPV1
doi:10.1371/journal.pone.0000545.g007 (Fig. 10). Camstatin, a highly basic and lysine rich polypeptide may
recognize this calmodulin-like region and block the TRPV1 channel
different, calmidazolium and W-7, equally potent antagonists of without internalization. This particular domain of the pore loop may
calmodulin inhibit both phases of the formalin-induced abdominal also serve as a cation filter and anti-calmodulin compounds can bind
pain and heat-induced tail-flick latency [32]. These in vivo after Ca2+ binds first to this TRPV1 structure (Fig. 10), thereby
observations also contradict to the Ca2+- calmodulin feedback preventing pore opening [44,45]. Moreover, point mutatation
theory of action of these agents. Animal studies, adjuvant therapy in studies established docking both ruthenium red and R4W2 to this
the clinical practice and studies in this paper suggest that structure. Conceivably, the most efficacious TRPV1 blockers have
antipsychotic calmodulin antagonist drugs rather elicit analgesia basic chemical character due to nitrogen heteroatoms in the
instead of hyperalgesia and block Ca2+-entry in peripheral structure that fit to the exposed acidic tetrad (Figs. 10 and 11)
nociceptive neurons. calmodulin inhibitors showed non-competitive [45]. Experiments with the above mentioned mutants of Garcia-
inhibitor kinetics with increasing dose of capsaicin. Our screening Martinez et al. and replacement experiment with radiolabeled
technology employs vanilloid/capsaicin-induced Ca2+-uptake assays R4W2 will provide the ultimate evidence for our claim based on the
in TRPV1 expressing cell lines, in the first round, then hits are current experimental results and in silico docking studies.
further tested and validated in primary neuron cultures from rat Calmodulin, due to interaction with Ca2+, undergoes a robust
embryonic DRG, enriched in TRPV1+ nociceptors. conformational change and exposes hidden hydrophobic docking site
A number of the Ca2+-transport experiments were carried out in for trifluoperazine and other conventional calmodulin antagonists
‘‘nominally zero Ca2+’’ buffer (261025 M extracellular Ca2+ [18,19,46]. By analogy, calmodulin-like mouth of the TRPV1 pore
([Ca2+]o) that reveals agonist-induced Ca2+-curents originating may react just like that described to other EF-hand structures [31]
from either outside sources or the ER, then further amplified by such as the S-100 Ca2+-binding proteins and ‘‘Excalibur’’. Likewise,
Ca2+ induced Ca2+ release mechanism, has recently noted to occur camstatin, an R/K-rich polypeptide may inhibit TRPV1 with
in TRPV1 immunopositive nociceptor neurons. By Ca2+ induced a mechanism similar to other calmodulin antagonists of basic nature.
Ca2+ release both TRPV1PM and TRPV1ER pools are actively The bulkiness of camstatin in fact has prompted us to hypothesize
involved in pain signaling [42,43]. Ca2+ induced Ca2+ release is extracellular targeting of TRPV1 by anti-calmodulin drugs.
usually followed by a currently recognized mechanism, so called Moreover, recently identified pore blockers from venom of a spider
store operated Ca2+ entry, which replenish the ER store emptied has further supported our TRPV1 channel blocker theory [46].
by different Ca2+-channels, including the Ca2+/vanilloid-induced From selectivity point of view, calmidazolium, trifluoperazine,
TRPV1ER. The usual function of store operated Ca2+ entry is to and W-7 have been determined inactive on intermediate Ca2+-
maintain Ca2+-homeostasis, around 1 mM in the ER, and activated K+ channels (IKCa) of human erythrocytes, which
[Ca2+]i = 1027 M, both essential for pain signaling in primary indicates that these compounds can distinguish among different

PLoS ONE | www.plosone.org 7 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

Figure 8. Calmidazolium (CMZ) differentially bocks NMDA-, and kainic acid (KA)-induced pain-responses. NMDA and kainic acid were
iontophoresed, sequentially, 2 min apart using 285 nA and 230 nA, respectively, then calmidazolium was ejected as shown, which markedly
inhibited the initial pain signals at the levels of dorsal horn neurons. Note the differential effects of calmidazolium on NMDA responses versus kainic
acid responses. Panel A: Responses to NMDA and kainic acid iontophoresed alternately in every 2 min. Panel B: Effects of iontophoresed
calmidazolium on the responses to these excitatory amino acids during the control period. Panel B: Differential effects of calmidazolium on the
NMDA- versus the kainic acid-evoked responses. Similar results were obtained in fifteen additional experiments. A representative recording is shown.
doi:10.1371/journal.pone.0000545.g008

cation channels [47]. Although structurally distinct, both calmi- other TRP channels and Ca2+-binding polypeptides mimicking
dazolium and W-7 are considered selective and potent inhibitors this region [46]. Better characterization of the extracellular
of calmodulin, yet they act extremely differently on the TRPV1 calmodulin-like structure by molecular modeling and in silico
target, which is an important issue in future drug discovery. Both screening might be exploited to design new generations of
calmidazolium and W-7 were noted analgesic intrathecally in both painkillers, thereby the novel but irreversible inflammatory pain
phases of formalin tests in the rat [32]. Oral bioavailability of neuron ablation technology [42,48–50] can be supplemented with
calmidazolium and additional derivatives, however, has yet to be reversible channel blocker drugs both acting via TRPV1.
evaluated in various pain models. Antipsychotic drugs selected on
empirical bases can be ranked for rational adjuvants selection in
various pain indications. Our data suggest that application of
MATERIALS AND METHODS
neuroleptics in severe cancer pain may enhance and complement Agents and abbreviations:
efficacy of conventional painkillers by direct inhibition of TRPV1 CGP-37157, 7-Chloro - 5 - (2 - chlorophenyl) - 1,5 - dihydro - 4,1 –
Ca2+-channel, an alternative target not used by opiates and benzothiazepin - 2(3H) - one; KA, Kainate, [2S-[2a,3b,4b-
NSAIDs. Our data in this paper consistent with specific action of (1Z,3E,5R)]]-2-Carboxy-4-(5-carboxy-1-methyl-1,3-hexadienyl)-3-
calmidazolium and other anti-calmodulin analogues on a short pyrrolidineacetic acid; M1, Miconazole 1; M2, Miconazole 2; M3,
acidic domain of TRPV1, at the cation filter region not found in Miconazole 3; M4, Clotrimazole 4, M5, Clotrimazole 5; M6,

PLoS ONE | www.plosone.org 8 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

embryonic DRG, seeded in poly-D lysine/collagen coated 96 well

plates. The ED100 for capsaicin-induction was determined to be
,1 mM in TRPV1-NIH3T3 cells at pH 7.5. Therefore antagonist-
screening assays were carried out with 2 mM capsaicin, an excess
amount of agonist, which fully activates TRPV1 either in DRG
neurons or in permanent cell lines ectopically expressing the vanilloid
receptor, but does not cause Ca2+-cytotoxicity during the 10 min
incubation period. To probe the TRPV1 specificity of capsaicin-
induced 45Ca2+-uptake, experiments were carried out in the parental
NIH 3T3 and HACAT cell lines and yielded negative results (data
not shown). The previously established cell lines and standardized
DRG cultures were extensively characterized for functional cell-
based TRPV1 studies [1–4]. Immediately before the transport assay,
cells were adapted to room temperature (20uC) for 5 min in HCM.
45
Ca2+-uptake was performed for 10 min at 20uC in HCM using
0.1 mCi 45Ca2+ as radioactive tracer in a 100 ml final volume. To
stop 45Ca2+-uptake, cells were rapidly changed back into 0.2 ml
HCM, washed 4 additional times with 0.2 ml HCM, and then lysed
in 80 ml/well RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM
NaCl, 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS, 5 mM
EDTA) for 30 min [1–4]. Aliquots of the solubilized cell extracts
Figure 9. Summary of the effects of calmidazolium (CMZ) on the were mixed with 120 ml Microscint-40 and counted in a 96 well plate
responses to heat or to iontophoresed excitatory amino acids of spinal liquid scintillation counter (TopCount-NXT, Packard).
dorsal horn neurons of nociceptive specific (NS) or wide dynamic range
(WDR) types. Note the decrease of the responses in the presence of
calmidazolium. Asterisks denote significant differences by ANOVA Anesthesia and surgery
(**p,0.05, ***p,0.01) as compared to their respective controls or the Extracellular single-neuron recordings were made in chloral
significant difference between the decrease of NMDA- and kainic acid - hydrate-anesthetized (4 g/kg initial dose, i.p., with supplemental
evoked responses in the presence of calmidazolium. See details in the text.
doi:10.1371/journal.pone.0000545.g009
doses as required) male Wistar rats weighing between 350–450 g.
The head of the animal was mounted in a stereotaxic frame. The
lumbar enlargement (L3–L5 segments) was exposed by a lami-
Clotrimazole 6; M7, N-Benzyl Imidazole; SB 290157, N2-[(2,2-
nectomy and the spinal cord was covered with a pool of warmed
Diphenylethoxy)acetyl]-L-arginine; W-7, N- 6 – aminohexyl – 5 –
mineral oil to prevent drying. Rectal temperature as well as the
chloro - 1 naphtalenesulfonamide; W-13, N - (4 - Aminobutyl) – 5
temperature of the mineral oil were monitored with temperature
- chloro -2- naphthalenesulfonamide.
probes and kept at 37uC by a warm water-heated blanket beneath
the rat and an infrared heat lamp from above. The level of
Cell lines expressing TRPV1 ectopically anesthesia was maintained so that the rats showed no sign of
TRPV1-NIH3T3 mouse fibroblast line, ectopically expressing the rat discomfort but the tail flick reflex could be evoked by application
TRPV1 was established previously. For validation of some data in of noxious heat (43–45uC) or mechanical stimulus (squeeze) to the
human cells, a permanent TRPV1-HaCaT keratinocyte line was tail. Recordings were only commenced at least 1 h after surgery.
prepared by drug selection, similar to that described previously [1–4]. All animal experiments were performed in accordance with
institutional animal welfare guidelines.
Primary cultures expressing TRPV1 endogenously
Primary DRG cultures were prepared from E16 embryonic rats as Extracellular recordings and iontophoresis
described previously then plated in 15 cm Petri dishes for 4 h to Compound recording/iontophoresis electrodes were made of
remove the most adherent fibroblasts. For Ca2+-uptake studies a seven barreled array of thin wall borosilicate glass capillary
primary afferents neurons cultured in 96 well plates (36103 cells/ tubings (Kation Scientific, Minneapolis, MN) The recording barrel
well), coated with poly-D lysine/collagen, which yielded both contained a low impedance (1 MV) 7-mm carbon fiber and drugs
axons and dendrites bearing neurons due to neuronal growth were delivered from the surrounding barrels. Single unit
factor supplementation in the culture medium [1–4]. The extracellular recordings were made from selected dorsal horn
dendrites are large processes branching from the cell body that neurons characterized by their responses to both innocuous (brush,
taper in diameter further from the cell body. The axons are pressure) and noxious (pinch, squeeze that was felt as painful by
a constant diameter and project over long distances from the cell the experimenter) stimuli applied to the excitatory receptive fields
body. Before the ion transport experiment, culture medium was of the hind paw. Mechanical stimuli of increasing strength were
removed and cells were washed three times. 45Ca-uptake assays delivered by a small brush and three serrated clamps with
were carried out in ‘‘nominally zero Ca2+’’ (i.e. 261025 M a graduated force. The experimenter released clamps so that force
extracellular Ca2+ ([Ca2+]o) in 25 mM Tris-HCl buffered Hanks’ was standard for each mechanical stimulus delivered over a period
balanced salt solution (pH 7.5), supplemented with 10 mM CaCl2 of 3 s, and about 1 min were allowed to pass between trials.
and 0.8 mM MgCl2medium (HCM) [1–4]. Neurons were characterized as low threshold (LT), nociceptive
specific (NS) or wide dynamic range (WDR) by their responses to
45
Vanilloid-induced Ca-uptake mechanical stimuli of increasing strength. Extracellular signals
Vanilloid agonist inducible calcium transport was assayed in from dorsal horn neurons were amplified and filtered (300 Hz to
adherent cell lines ectopically expressing the C-terminally e-tagged 8 KHz) using an ExAmp-20K equipment and preamplifier
rat TRPV1 (36104 cells/well) and primary cultures prepared from (Kation Scientific). Neuronal activity was displayed on an

PLoS ONE | www.plosone.org 9 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

Figure 10. (a) Homologous portions of various TRP channels, near the border of the pore loop and the 6th transmembrane domain were aligned
with a validated R4W2 peptide similar in biochemical character to ruthenium red. An acidic tetrad motif DXEXXEXXD which can bind the positively
charged peptides in human and rat TRPV1, is absent in TRPV2/VRL1 and TRPV3, (both close homologues of TRPV1) as well as in distantly related TRPs
and bKcsA, a bacterial cation channel. An acidic sequence, partially similar to the heat sensitive TRPs, is present in the cold responsive TRPM8/CMR1.
Distant TRPV homologues do not share the acidic tetrad motif either, such as g/mOTRPC4 and hOSM, nonselective cation channel orthologues from
Gallus gallus (chicken), mouse, and human, respectively that confer sensitivity to extracellular osmolarity, mTRP12, another osmotically activated TRP
channel from mouse; mGFRCC, mouse growth factor receptor coupled channel; hVOC, Homo sapiens Kv4.3 potassium channel; dSha12, a ‘‘shaker-
like’’ potassium channel from Drosophila melanogaster. (b) The TM5-pore loop-TM6 region of TRPV1 is analogous to the ‘‘inverted teepee’’, pore-
forming domain of bKcsA. Side-view of the TRPV1 tetramer channel depicts the hypothetical pore at the middle. Arrows point to the putative
ruthenium red/R4W2 binding site in each TRPV1 subunits of the tetramer. (c) To better represent the simulated quaternary structure of TRPV1, a view
perpendicular to the plasma membrane is generated with the homo-tetrameric TRPV1 domain fragments. The position of acidic domain is noted by
an arrow in a single subunit in this view of the model.
doi:10.1371/journal.pone.0000545.g010

oscilloscope and through an audio analyzer and detected with Temperature ramps were generated from a holding temperature
a WD-2 window discriminator (Dagan, Minneapolis, MN). The of 30uC to a peak of 50uC at a rate of about 2uC/s. The heat
number of action potentials per second was counted by the stimulus was turned off when skin temperature reached 50uC.
computer and the resulting peristimulus time histograms were
displayed. Iontophoretic drug delivery and collection of experi- Data analysis
mental data were performed by a multifunction data acquisition Statistical evaluations were made using the total number of spikes
board (PCI-1200, National Instruments, Austin, TX) placed in minus background activity evoked during each epoch of excitation by
a computer and programmed in LabVIEW 7 (National Instru- heat stimuli or iontophoretic application of an excitatory compound.
ments). Delivery of drugs by microiontophoresis was performed The background neuronal discharge was calculated by averaging
using Union-36 constant current source units (Kation Scientific). a 15 s period of ongoing activity preceding each epoch of excitation
Drug barrels of the combined electrodes contained one of the and this value was subtracted from the total number of evoked spikes.
following freshly made solutions: 100 mM N-methyl-D-aspartate Differences in magnitude between different response epochs of
Na (NMDA) in 100 mM NaCl (pH 8.0), 20 mM kainic acid in a single cell were confirmed by one-factor analysis of variance (with
100 mM NaCl (pH 8.0), 5 mM calmidazolium chloride in Student-Newman-Keuls test for post-hoc analysis) by comparing the
100 mM NaCl (pH 7.2) was prepared using a 100 mM calmida- total number of spikes per excitation period. To make data from
zolium in ethanol stock solution. NMDA and kainic acid were different experiments more comparable, analysis of pooled data was
ejected with negative iontophoretic currents ranging from 10 to done after normalizing the baseline stimulus-evoked response to
100 nA for 5 s in every 2 min. calmidazolium was ejected with 100%. Means6SD of a number (n) of observations are given
positive currents at 100 nA for 120 s. Retaining currents of throughout. A P value of ,0.05 was considered significant in all cases.
opposite directions between 3–10 nA were used for all drugs.

Molecular modeling of TM5-pore loop-TM6 domain

Heat stimulation
Single unit extracellular recordings were made from selected of TRPV1 and docking calmidazolium and
dorsal horn neurons responding to noxious heat delivered by chlorpromazine to the extracellular portion
a projector lamp focused on the blackened surface of the hind paw. The preliminary channel domain model was used in a fixed
Skin temperature was monitored using an infrared thermometer. protein flexible ligand docking protocol using simulated annealing

PLoS ONE | www.plosone.org 10 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

Figure 11. A docking of CMZ to pore loop domain of TRPV1. Symmetrical arrangement of the 5 and 6 TM helices (magenta) in the tetrameric TRPV1
receptor complex is shown. Aromatic residues within the 20 Å´ proximity to the inhibitor are shown by sticks, and the Asp and Glu residues as well as
the ligand by spacefilled models. Color codes for the atoms: grey, C; cyan, H; blue, N; red, O; green, Cl. The pore loop, localized in between
transmembrane domain 5 and 6 of TRPV1 is shown with the specific docking site of the positively charged CMZ. (a) side-view and (b) a view
perpendicular to the cell membrane were generated after docking of CMZ. Negatively charged ‘‘acidic domain’’ of TRPV1 in the homotetramer may
serve as ideal nest for channel blockers such as CMZ.trifluoperazine.chlorpromazine/amitriptyline, as well as, ruthenium red and R4W2, all charged
oppositely. The rule is that more basic is an anti- calmodulin substance that more attracted to the acidic moieties of the ‘‘nest’’ by electrostatic forces
near to the entrance (i.e. ion filter) of the pore. It is a tendency that longer is the hydrophobic side chain tethered to the tricyclic core better is the fit
inside the pore, such as determined with amitriptyline and carbamazepine.
doi:10.1371/journal.pone.0000545.g011

as it is implemented in the MOE 2004.03 program package from package (litSYBYL molecular modeling software, version 7.1,
Chemical Computing Group. The homotetramer of the TM5- Tripos Associates Inc., 1699 Hanley Rd, St. Louis, MO 63144-
pore loop-TM6 domain of TRPV1 was built using the Swiss- 2913.) using the general Tripos force field for the ligands
PdbViewer molecular graphics and the SWISS-MODEL com- containing chemical bonds unusual in biological macromolecular
parative homology modeling program package [51–53]. The modeling. The structure of the receptor-ligand complex was
Q8NER1 sequence from TrEMBL [54] was used as a target approximated by minimizing with the prepositioned ligand. The
protein. Transmembrane prediction using the TMHMM 2.0 backbone of the residues of the receptor complex outside the 20
server [55] was carried out to delineate the transmembrane angstrom boundary from the inhibitor molecule were constrained
segments. The predicted extracellular pore loop together with the in a fixed position allowing larger movement of the neighboring
fifth and sixth transmembrane helices were extracted from the full amino acids only. The calmidazolium, holding an inherent
length protein sequence and aligned with the sequence of the positive charge, was fixed by closely surrounding Asp and Glu
bacterial K+ channel KcsA [56]. The X-ray crystal structure of the residues. No aromatic-aromatic interactions were observed in-
M1-pore loop-M2 of the KcsA was used (PDB accession code dicating the dominance of charged residues in the channel
1K4C) as a template for homology modeling. Optimization of the blocking process.
loop regions was carried out by applying extensive constraint
minimization using the GROMOS96 [57] force-field implemen-
ted in Swiss-PdbViewer. The homotetramer structures were built ACKNOWLEDGMENTS
manually using the non-crystallographic symmetry operators
derived from the template. The final depictions of the S5-pore Author Contributions
loop-S6 segment of TRPV1 model were produced using the VMD Conceived and designed the experiments: Cv ZO. Performed the
program [58]. The possible interactions between the channel experiments: ZO KJ LP TL NB DB FO SS. Analyzed the data: ZO DB
region of the tetrameric TRPV1 receptor complex and the FO SS. Wrote the paper: Cv ZO.
inhibitors were calculated with the Sybyl molecular modeling

REFERENCES
1. Olah Z, Szabo T, Karai L, Hough C, Fields RD, et al. (2001) Ligand-induced 4. Karai L, Russell JT, Iadarola MJ, Olah Z (2004) Vanilloid receptor 1 regulates
Dynamic Membrane Changes and Cell Deletion Conferred by Vanilloid multiple calcium compartments and contributes to Ca2+-induced Ca2+-release
Receptor 1. J Biol Chem 276: 11021–11030. in sensory neuronsn. J Biol Chem.
2. Liu M, Liu MC, Magoulas C, Priestley JV, Willmott NJ (2003) Versatile 5. Zygmunt PM, Petersson J, Andersson DA, Chuang H, Sorgard M, et al. (1999)
regulation of cytosolic Ca2+ by vanilloid receptor I in rat dorsal root ganglion Vanilloid receptors on sensory nerves mediate the vasodilator action of
neurons. J Biol Chem 278: 5462–5472. anandamide. Nature 400: 452–457.
3. Marshall IC, Owen DE, Cripps TV, Davis JB, McNulty S, et al. (2003) 6. Olah Z, Karai L, Iadarola MJ (2001) Anandamide Activates Vanilloid Receptor
Activation of vanilloid receptor 1 by resiniferatoxin mobilizes calcium from 1 (VR1) at Acidic pH in Dorsal Root Ganglia Neurons and Cells Ectopically
inositol 1,4,5-trisphosphate-sensitive stores. Br J Pharmacol 138: 172–176. Expressing VR1. J Biol Chem 276: 31163–31170.

PLoS ONE | www.plosone.org 11 June 2007 | Issue 6 | e545

 Anti-CAM Drugs Inhibit TRPV1

7. Jordt SE, Tominaga M, Julius D (2000) Acid potentiation of the capsaicin 33. Williams M, Kowaluk EA, Arneric SP (1999) Emerging molecular approaches to
receptor determined by a key extracellular site. Proc Natl Acad Sci U S A 97: pain therapy. J Med Chem 42: 1481–1500.
8134–8139. 34. Acs G, Palkovits M, Blumberg PM (1995) Trifluoperazine modulates
8. Vellani V, Mapplebeck S, Moriondo A, Davis JB, McNaughton PA (2001) [3H]resiniferatoxin binding by human and rat vanilloid (capsaicin) receptors
Protein kinase C activation potentiates gating of the vanilloid receptor VR1 by and affects 45Ca uptake by adult rat dorsal root ganglion neurones. J Pharmacol
capsaicin, protons, heat and anandamide. J Physiol 534: 813–825. Exp Ther 274: 1090–1098.
9. Premkumar LS, Ahern GP (2000) Induction of vanilloid receptor channel 35. Jung J, Hwang SW, Kwak J, Lee SY, Kang CJ, et al. (1999) Capsaicin binds to
activity by protein kinase C. Nature 408: 985–990. the intracellular domain of the capsaicin-activated ion channel. J Neurosci 19:
10. Olah Z, Karai L, Iadarola MJ (2002) Protein Kinase Calpha Is Required for 529–538.
Vanilloid Receptor 1 Activation. EVIDENCE FOR MULTIPLE SIGNALING 36. Robinson LR, Czerniecki JM, Ehde DM, Edwards WT, Judish DA, et al. (2004)
PATHWAYS. J Biol Chem 277: 35752–35759. Trial of amitriptyline for relief of pain in amputees: results of a randomized
11. Jung J, Shin JS, Lee SY, Hwang SW, Koo J, et al. (2003) Phosphorylation of controlled study. Arch Phys Med Rehabil 85: 1–6.
vanilloid receptor 1 by Ca+2/calmodulin-dependent kinase II regulates its 37. Max MB (1994) Treatment of post-herpetic neuralgia: antidepressants. Ann
vanilloid binding. J Biol Chem. Neurol 35 Suppl: S50–53.
12. Tominaga M, Caterina MJ, Malmberg AB, Rosen TA, Gilbert H, et al. (1998) 38. Kieburtz K, Simpson D, Yiannoutsos C, Max MB, Hall CD, et al. (1998) A
The cloned capsaicin receptor integrates multiple pain-producing stimuli. randomized trial of amitriptyline and mexiletine for painful neuropathy in HIV
Neuron 21: 531–543.
infection. AIDS Clinical Trial Group 242 Protocol Team. Neurology 51:
13. Caterina MJ, Schumacher MA, Tominaga M, Rosen TA, Levine JD, et al.
1682–1688.
(1997) The capsaicin receptor: a heat-activated ion channel in the pain pathway.
39. Khoromi S, Cui L, Nackers L, Max MB (2006) Morphine, nortriptyline and
Nature 389: 816–824.
their combination vs. placebo in patients with chronic lumbar root pain. Pain.
14. Bhave G, Hu HJ, Glauner KS, Zhu W, Wang H, et al. (2003) Protein kinase C
phosphorylation sensitizes but does not activate the capsaicin receptor transient 40. Jorgensen WL (2004) The many roles of computation in drug discovery. Science
receptor potential vanilloid 1 (TRPV1). Proc Natl Acad Sci U S A 100: 303: 1813–1818.
12480–12485. 41. Fang L, Wu J, Lin Q, Willis WD (2002) Calcium-calmodulin-dependent protein
15. Jordt SE, Julius D (2002) Molecular basis for species-specific sensitivity to ‘‘hot’’ kinase II contributes to spinal cord central sensitization. J Neurosci 22:
chili peppers. Cell 108: 421–430. 4196–4204.
16. Kedei N, Szabo t, Lile JD, Treanor JJ, Olah Z, et al. (2001) Analysis of the native 42. Olah Z, Szabo T, Karai L, Hough C, Fields RD, et al. (2001) Ligand-induced
quaternary structure of vanilloid receptor 1. J Biol Chem 276: 28613–28619. dynamic membrane changes and cell deletion conferred by vanilloid receptor 1.
17. Donato R (1999) Functional roles of S100 proteins, calcium-binding proteins of J Biol Chem 276: 11021–11030.
the EF-hand type. Biochim Biophys Acta 1450: 191–231. 43. Karai LJ, Russell JT, Iadarola MJ, Olah Z (2004) Vanilloid receptor 1 regulates
18. Likic VA, Gooley PR, Speed TP, Strehler EE (2005) A statistical approach to the multiple calcium compartments and contributes to Ca2+-induced Ca2+ release
interpretation of molecular dynamics simulations of calmodulin equilibrium in sensory neurons. J Biol Chem 279: 16377–16387.
dynamics. Protein Sci 14: 2955–2963. 44. Mohapatra DP, Wang SY, Wang GK, Nau C (2003) A tyrosine residue in TM6
19. Douglass PM, Salins LL, Dikici E, Daunert S (2002) Class-selective drug of the Vanilloid Receptor TRPV1 involved in desensitization and calcium
detection: fluorescently-labeled calmodulin as the biorecognition element for permeability of capsaicin-activated currents. Mol Cell Neurosci 23: 314–324.
phenothiazines and tricyclic antidepressants. Bioconjug Chem 13: 1186–1192. 45. Garcia-Martinez C, Morenilla-Palao C, Planells-Cases R, Merino JM, Ferrer-
20. Slemmon JR, Morgan JI, Fullerton SM, Danho W, Hilbush BS, et al. (1996) Montiel A (2000) Identification of an aspartic residue in the P-loop of the
Camstatins are peptide antagonists of calmodulin based upon a conserved vanilloid receptor that modulates pore properties. J Biol Chem 275:
structural motif in PEP-19, neurogranin, and neuromodulin. J Biol Chem 271: 32552–32558.
15911–15917. 46. Kitaguchi T, Swartz KJ (2005) An inhibitor of TRPV1 channels isolated from
21. Johanson RA, Sarau HM, Foley JJ, Slemmon JR (2000) Calmodulin-binding funnel Web spider venom. Biochemistry 44: 15544–15549.
peptide PEP-19 modulates activation of calmodulin kinase II In situ. J Neurosci 47. Del Carlo B, Pellegrini M, Pellegrino M (2002) Calmodulin antagonists do not
20: 2860–2866. inhibit IK(Ca) channels of human erythrocytes. Biochim Biophys Acta 1558:
22. Rosenbaum T, Gordon-Shaag A, Munari M, Gordon SE (2004) Ca2+/ 133–141.
Calmodulin Modulates TRPV1 Activation by Capsaicin. J Gen Physiol 123: 48. Brown DC, Iadarola MJ, Perkowski SZ, Erin H, Shofer F, et al. (2005)
53–62. Physiologic and antinociceptive effects of intrathecal resiniferatoxin in a canine
23. Numazaki M, Tominaga T, Takeuchi K, Murayama N, Toyooka H, et al. bone cancer model. Anesthesiology 103: 1052–1059.
(2003) Structural determinant of TRPV1 desensitization interacts with 49. Karai L, Brown DC, Mannes AJ, Connelly ST, Brown J, et al. (2004) Deletion of
calmodulin. Proc Natl Acad Sci U S A 100: 8002–8006. vanilloid receptor 1-expressing primary afferent neurons for pain control. J Clin
24. Zhang Z, Tang J, Tikunova S, Johnson JD, Chen Z, et al. (2001) Activation of Invest 113: 1344–1352.
Trp3 by inositol 1,4,5-trisphosphate receptors through displacement of in- 50. Tender GC, Walbridge S, Olah Z, Karai L, Iadarola M, et al. (2005) Selective
hibitory calmodulin from a common binding domain. Proc Natl Acad Sci U S A ablation of nociceptive neurons for elimination of hyperalgesia and neurogenic
98: 3168–3173. inflammation. J Neurosurg 102: 522–525.
25. Tang J, Lin Y, Zhang Z, Tikunova S, Birnbaumer L, et al. (2001) Identification
51. Peitsch MC (1995) Protein modeling by e-mail. Bio/Technology 13: 658–660.
of common binding sites for calmodulin and inositol 1,4,5-trisphosphate
52. Guex N, Peitsch MC (1997) SWISS-MODEL and the Swiss-PdbViewer: an
receptors on the carboxyl termini of trp channels. J Biol Chem 276:
environment for comparative protein modeling. Electrophoresis 18: 2714–2723.
21303–21310.
26. Harper JL, Daly JW (2000) Effect of calmidazolium analogs on calcium influx in 53. Schwede T, Kopp J, Guex N, Peitsch MC (2003) SWISS-MODEL: An
HL-60 cells. Biochem Pharmacol 60: 317–324. automated protein homology-modeling server. Nucleic Acids Res 31:
27. Gamper N, Shapiro MS (2003) Calmodulin mediates Ca2+-dependent 3381–3385.
modulation of M-type K+ channels. J Gen Physiol 122: 17–31. 54. Boeckmann B, Bairoch A, Apweiler R, Blatter MC, Estreicher A, et al. (2003)
28. Bruening-Wright A, Schumacher MA, Adelman JP, Maylie J (2002) Localization The SWISS-PROT protein knowledgebase and its supplement TrEMBL in
of the activation gate for small conductance Ca2+-activated K+ channels. 2003. Nucleic Acids Res 31: 365–370.
J Neurosci 22: 6499–6506. 55. Krogh A, Larsson B, von Heijne G, Sonnhammer EL (2001) Predicting
29. Soldatov NM (2003) Ca2+ channel moving tail: link between Ca2+-induced transmembrane protein topology with a hidden Markov model: application to
inactivation and Ca2+ signal transduction. Trends Pharmacol Sci 24: 167–171. complete genomes. J Mol Biol 305: 567–580.
30. Boulay G (2002) Ca(2+)-calmodulin regulates receptor-operated Ca(2+) entry 56. Zhou Y, Morais-Cabral JH, Kaufman A, MacKinnon R (2001) Chemistry of ion
activity of TRPC6 in HEK-293 cells. Cell Calcium 32: 201–207. coordination and hydration revealed by a K+ channel-Fab complex at 2.0 A
31. Rigden DJ, Jedrzejas MJ, Galperin MY (2003) An extracellular calcium-binding resolution. Nature 414: 43–48.
domain in bacteria with a distant relationship to EF-hands. FEMS Microbiol 57. van Gunsteren WF (1996) Biomolecular simulation: the GROMOS96 manual
Lett 221: 103–110. and user guide. Vdf Hochschulverlag, ETHZ.
32. Menendez L, Perez-Vallina JR, Cantabrana B, Hidalgo A, Baamonde A (1996) 58. Humphrey W, Dalke A, Schulten K (1996) VMD: visual molecular dynamics.
Calmodulin inhibitors induce spinal analgesia in rats. Brain Res 731: 114–121. J Mol Graph 14: 33–38, 27–38.

PLoS ONE | www.plosone.org 12 June 2007 | Issue 6 | e545