Two-pore channels provide insight into the evolution of voltage-gated Ca2+ and Na+ channels

Abstract

Four-domain voltage-gated Ca(2+) and Na(+) channels (CaV, NaV) underpin nervous system function and likely emerged upon intragenic duplication of a primordial two-domain precursor. To investigate if two-pore channels (TPCs) may represent an intermediate in this evolutionary transition, we performed molecular docking simulations with a homology model of TPC1, which suggested that the pore region could bind antagonists of CaV or NaV. CaV or NaV antagonists blocked NAADP (nicotinic acid adenine dinucleotide phosphate)-evoked Ca(2+) signals in sea urchin egg preparations and in intact cells that overexpressed TPC1. By sequence analysis and inspection of the model, we predicted a noncanonical selectivity filter in animal TPCs in which the carbonyl groups of conserved asparagine residues are positioned to coordinate cations. In contrast, a distinct clade of TPCs [TPCR (for TPC-related)] in several unicellular species had ion selectivity filters with acidic residues more akin to CaV. TPCRs were predicted to interact strongly with CaV antagonists. Our data suggest that acquisition of a "blueprint" pharmacological profile and changes in ion selectivity within four-domain voltage-gated ion channels may have predated intragenic duplication of an ancient two-domain ancestor.
Copyright © 2014, American Association for the Advancement of Science.

Full text

Two-pore channels provide insight into the evolution of voltage-
gated Ca2+ and Na+ channels
Taufiq Rahman1,*, Xinjiang Cai2, G. Cristina Brailoiu3, Mary E. Abood4, Eugen Brailoiu5,
and Sandip Patel2,*
1Department of Pharmacology, Cambridge University, Cambridge CB2 1PD, UK
2Department of Cell Developmental Biology, University College London, London, WC1E 6BT, UK
3Department of Pharmaceutical Sciences, Thomas Jefferson University School of Pharmacy,
Philadelphia, Pennsylvania 19107, USA
4Department of Anatomy and Cell Biology, and Center for Substance Abuse Research, Temple
University, Philadelphia, PA 19140, USA
5Department of Pharmacology and Center for Substance Abuse Research, Temple University,
Philadelphia, PA 19140, USA
Abstract
Four-domain voltage-gated Ca2+ and Na+ channels (CaV, NaV) underpin nervous system function
and likely emerged upon intragenic duplication of a primordial two-domain precursor. To
investigate if two-pore channels (TPCs) may represent an intermediate in this evolutionary
transition, we performed molecular docking simulations with a homology model of TPC1, which
suggested that the pore region could bind antagonists of CaV or NaV. CaV or NaV antagonists
blocked NAADP-evoked Ca2+ signals in sea urchin egg preparations and in intact cells that
overexpressed TPC1. By sequence analysis and inspection of the model, we predicted a
noncanonical selectivity filter in animal TPCs in which the carbonyl groups of conserved
asparagine residues are positioned to coordinate cations. In contrast, a distinct clade of TPCs
(TPCR) in several unicellular species had ion selectivity filters with acidic residues more akin to
CaV. TPCRs were predicted to interact strongly with CaV antagonists. Our data suggest that
acquisition of a “blueprint” pharmacological profile and changes in ion selectivity within four-
domain voltage-gated ion channels may have predated intragenic duplication of an ancient two-
domain ancestor.
Introduction
The voltage-gated ion channel superfamily is critical for a plethora of cellular processes in
both excitable and nonexcitable cells (1). Voltage-gated Ca2+ channels (CaVs) drive
*Joint corresponding authors (mtur2@cam.ac.uk or patel.s@ucl.ac.uk).
Author contributions: TR performed the modeling and docking. XC performed the phylogenetic analysis. GCB, MEA, and EB
performed the microinjection and Ca2+ imaging experiments. SP assisted with the alignments, performed the Ca2+ measurements in
egg homogenates, and wrote the paper with input from TR and XC.
Data and materials availability: Models are provided in the Supplementary Materials.
NIH Public Access
Author Manuscript
Sci Signal. Author manuscript; available in PMC 2015 November 18.
Published in final edited form as:
Sci Signal. ; 7(352): ra109. doi:10.1126/scisignal.2005450.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

neurotransmitter release at nerve terminals, thereby sustaining action potentials propagated
by voltage-gated Na+ channels (NaVs) (2). Mutations in the genes encoding these proteins
underlie many diseases (“channelopathies”), including migraine, arrhythmias and epilepsy
(3). Both CaV and NaV are composed of four homologous domains (DI-DIV) that assemble
in a pseudotetrameric arrangement (Fig. 1A). Each domain comprises six membrane-
spanning regions (S1–S6), which form the voltage sensor (S1–S4) and pore (S5–S6) (1).
CaV and NaV are targeted clinically by various drugs– including antihypertensives,
antiarrhythmics, anticonvulsants, and local anaesthetics– primarily through interactions with
S6 (4, 5). The reentrant pore loops (p-loops) located between S5 and S6 control ion
selectivity. Much is known about the structure-function relationships for these proteins,
including atomic insight into prokaryotic NaV homologs (6–9).
Two-pore channels (TPCs) are less well characterized voltage-gated ion channels (10).
Similar to CaV and NaV, TPCs are modular proteins that are also likely to be
pseudotetrameric (11–13) but they possess a distinctive structure comprising only two
homologous channel domains (Fig. 1A). Additionally, they are not present on the plasma
membrane but rather on acidic intracellular organelles that function as Ca2+ stores (14). In
plants, they localize to the vacuole and correspond to SV channels that underlie the “slow
vacuolar current” in response to Ca2+ and voltage (15). In animals, TPCs localize to the
analogous vesicles within the endosomal and lysosomal system, but their function is not
entirely clear. Multiple studies suggest TPCs are the target for the Ca2+-mobilizing
messenger NAADP (16–18). NAADP mobilizes Ca2+ from acidic organelles to regulate a
multitude of events in various cells, including sea urchin eggs where its Ca2+-mobilizing
activity was first described (19, 20). Gating of TPCs by NAADP, however, is complex and
likely involves accessory proteins and co-activation by the phosphoinositide PI(3,5)P2 (21–
23). The pharmacology of TPCs is ill-defined and the permeability properties of TPCs vary
substantially between studies with reports concluding that TPCs are nonselective (24, 25) or
selective for Ca2+ (26), Na+ (22), or H+ (27).
NaVs are thought to have evolved from CaVs (2), and their emergence can be traced to
unicellular organisms that were the ancestors of fungi and animals (28, 29). CaV-like
selectivity filters in NaVs from basal lineages support the CaV to NaV transition (28), as do
functional analyses demonstrating Ca2+ permeability of select NaV homologs from
Nematostella vectensis, a simple Eumetazoan (30). NaVs are also present in bacteria (31),
but bacterial NaVs are single-domain proteins that likely acquired Na+ selectivity
independently of animal NaVs (32). The topological similarities between the four domains
of animal CaV and NaV with the single domain of voltage-gated K+ channels, which
assemble as tetramers, has led to the proposal that four-domain channels evolved following
two rounds of intragenic duplication of a primordial, single-domain precursor (2). The
duplicated domain organization of TPCs makes them possible descendants of the putative
evolutionary two-domain intermediate (33). TPCs may, therefore, hold a key position in the
evolution of the voltage-gated ion channel superfamily. Definition of their properties is thus
important for reconstructing ancestry of ion channel attributes. Here, we provide evidence
that the structural determinants underlying channel blockade by pharmacological antagonists
and ion selectivity may have been acquired prior to intragenic duplication of primeval two-
Rahman et al. Page 2
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

domain channels. Key features of extant voltage-gated ion channels thus emerged early in
their evolution.
Results
Phylogenetic analysis of voltage-gated ion channel domains suggests a common ancestry
for TPCs, CaVs, and NaVs
The duplicated domain organization of TPCs is consistent with the relationship of TPCs to
an ancient two-domain precursor that underwent intragenic duplication to give rise to four-
domain channels (2). The phylogeny of extant two-domain and four-domain channels,
however, is ill-defined and difficult to establish given sequence divergence. We divided
TPCs, Cav, and Nav members of the voltage-gated ion channel superfamily (table S1) into
their individual channel domains (S1–S6). Ancestry was then probed through an “all-
domain” phylogenetic analysis using the DI – DIV segments of Nav, and Cav, and the DI –
DII segments of TPCs (Fig. 1B). For Cav and Nav, DI grouped with DIII; whereas DII
grouped with DIV, consistent with an ancestral duplication event (34). Importantly, TPC
domains appear phylogenetically closer to individual domains of CaV and NaV, than to each
other (Fig. 1B). DI of TPCs grouped with DI and DIII of CaVs and NaVs (green); whereas
DII of TPC grouped with DII and DIV of CaVs and NaVs (purple). Each domain of TPCs
may, therefore, be related to counterparts in CaV and NaV supporting a common ancestry.
CaV antagonists target the pore region of TPCs
Ca2+ and Na+ channel antagonists are thought to bind to similar regions within the pore of
their respective four-domain channels. Overlap in the molecular determinants underlying
drug binding and channel regulation suggests a similar structural motif for pharmacological
block (35). Because TPCs, CaVs, and NaVs are likely descended from a common ancestor,
all of these voltage-gated ion channels may contain a similar structural motif within their
pore regions. To test this, we generated homology models for the pore regions (S5–S6) of
TPC, CaV, and NaV for molecular docking analyses. The pore regions of TPCs and the
single-domain prokaryotic Na+ channels exhibit high sequence similarity (Fig. 2A),
especially within S5, S6, and the first pore-helix (PH1). Therefore, we used the crystal
structure of NaV from Arcobacter butzleri (PDB: 3RVY) (6) as a template for homology
modeling. In the structural model, TPC adopted the characteristic pyramidal structure in
which four inner helices (S6) cross to form a closed bundle near the cytosolic entrance (Fig.
2B).
To test whether TPCs could bind CaV blockers, we performed molecular docking studies
with dihydropyridines, antagonists of L-type CaV (Fig. 3A, Table 1). We successfully
docked a series of dihydropyridines to the pore region of TPC (Fig. 3A). Analogs were
tightly clustered and centrally placed in the inner cavity. The predicted free energy (ΔG)
values were −4.5 to −6.5 kcal/mol (Table 1). We also docked the dihydropyridines onto
models of CaV and NaV and calculated ΔG values for each antagonist (Table 1) As
expected, dihydropyridines docked to CaV with in a tight cluster biased towards the DIII–
DIV interface (Fig. 3B). Although dihydropyridines docked to NaV, the poses were less
clustered than for TPC or CaV, segregating into either the inner cavity or the DII–DIII
Rahman et al. Page 3
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

interface of NaV (Fig. 3C). This lack of consensus was most apparent from an overlay of the
volume occupied by the analogs in the three channel types (Fig. 3D). The ΔG values were
generally higher (less negative) for NaV (−3.1 to −6 kcal/mol) than TPC and CaV (Fig. 3D,
Table 1) predicting a rank order of potency for dihydropyridines of CaV > TPC > NaV (Fig.
3D). Additionally, verapamil (a phenylyalkamine class of L-type CaV antagonist) and
diltiazem (a benzothiazepine class of CaV antagonist) also docked to TPC (Fig. 3E, Table 1).
Thus, the molecular docking simulations suggested that TPCs can bind CaV antagonists.
Multiple studies, although not all, indicate that TPCs are activated by NAADP (16, 17, 22,
23). Several CaV antagonists block NAADP-evoked Ca2+ signals in sea urchin egg
homogenates (36) and TPC-dependent Na+ currents in cells overexpressing TPC (22, 37).
Using the sea urchin egg homogenate preparation, we showed that nifedipine, isradipine,
verapamil, and diltiazem all inhibited NAADP-induced Ca2+ release (Fig. 3F). Thus, the
molecular docking simulations and the NAADP-stimulated, Ca2+-release assay suggest that
CaV antagonists target and block the TPC pore.
NaV antagonists target the pore region of TPCs
To test whether TPCs are targeted by NaV antagonists, we performed a similar analysis
using a series of 10 drugs, typified by the local anaesthetic lidocaine. As with the Cav
antagonists, the NaV antagonists docked in a tight cluster within the TPC pore (Fig. 4A) and
bound with ΔG values that were similar to those calculated for these antagonists with the
NaV model (Table 2, fig. S1). Lidocaine blocked NAADP-evoked Ca2+ signals in the sea
urchin egg homogenate preparation but had little effect on Ca2+ released in response to
cyclic ADP-ribose (Fig. 4B), which is mediated by ryanodine receptors (38). This was
expected because, although ryanodine receptors are also intracellular channels that are
permeable to Ca2+, they are structurally distinct from TPCs (38). Indeed, molecular docking
indicated that ryanodine interacted weakly with the TPC pore (ΔG = −3.03 kcal/mol).
Bupivacaine, a local anaesthetic and anti-arrhythmic agent, also selectively inhibited
NAADP responses (Fig. 4B–C). The effects of lidocaine and bupivacaine on NAADP
responses were concentration-dependent with half-maximal inhibitory concentrations (IC50)
of 1.1 ± 0.2 and 0.1 ± 0.02 mM (n=3), respectively (Fig. 4C).
CaV and NaV antagonists target the pore region of TPCs through common sites
Although CaV and NaV antagonists docked at similar positions within the TPC pore (Fig. 3–
4), the poses obtained differed from those in models of their cognate four-domain channels,
thereby ruling out template bias during model building. For example, nicardipine, which the
simulations indicated interacted strongly with both TPC (ΔG = −6.5 kcal/mol) and CaV, (ΔG
= −7.6 kcal/mol), adopted an elongated pose approximately parallel to the axis of ion
conduction in TPC and a pose perpendicular to ion conduction in Cav (Fig. 5A). The
projection of nicardipine into the DIII–DIV interface of CaV is similar to previously
reported docking models for dihydropyridines (39).
We observed similar disparities upon close inspection of the poses for local anesthetics
docked in TPC or in NaV. The poses for several drugs were perpendicular in TPC compared
with their poses in NaV (Fig. 5B). The poses in NaV, exemplified by mepivacaine are
Rahman et al. Page 4
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

consistent with the pose reported for etidocaine docked in Nav (40). Intriguingly, we noted
congruence in poses for a range of CaV (nicardapine, verapamil and diltiazem) and NaV
(mepivacaine) antagonists with regard to their docked positions at the cytosolic end of the
TPC pore (Fig. 5C). The identified interacting residues in TPC for these antagonists mapped
to S6 in each of the domains (Fig. 5D). We identify Leu315 (DI), Val318 (DI), and Val675
(DII) as potential determinants for both CaV and NaV antagonist action at TPCs (Fig. 5D,
arrowheads). Leu315 in DI of TPC aligns with Val1165 in DIII of Ca 1.2 and Ile1469 in DIII
of NaV1.2 (Fig. 5D, bold residues). These residues are amongst those identified by site-
directed mutagenesis in mediating the effects of phenylalkamines in CaV (41) and local
anaesthetics and anticonvulsants in NaV (42) (Fig. 5B, underlined residues). These data
indicated that CaV and NaV antagonists likely interact at a common site in TPCs.
To test that the modeled docking data correlated with the inhibitory effects of these CaV and
NaV antagonists on NAADP-stimulated Ca2+ release, we compared predicted ΔG from the
docking analyses to IC from the sea urchin egg Ca2+ release assays (Fig. 5E). The in silico
experiments and the functional assays were positively correlated (Fig. 5E). These data
support the hypothesis that TPCs are targets for both CaV and NaV antagonists.
Despite the correlation between “dry” and “wet” pharmacology, some of the antagonists of
CaVs or NaVs are not entirely selective for their target four-domain channels, leaving open
the possibility that our docking and functional analyses are unrelated. To bolster our
hypothesis, we analyzed veratridine, which is considered a selective NaV agonist (43).
Similar to the NaV antagonists, veratridine inhibited NAADP-, but not cyclic ADP-ribose-
induced Ca2+ release in a concentration-dependent manner (Fig. 5F–G). This inhibition is
similar to the reported inhibitory effects of the CaV agonist BayK 8644 on NAADP-
stimulated Ca2+ responses (36). TPCs thus appear to possess a pharmacological site that is
related to, but clearly distinct from, both CaV and NaV.
CaV and NaV modifiers inhibit recombinant TPC1
To further investigate the functional effects of CaV and NaV antagonists on TPC activity, we
characterized NAADP-evoked Ca2+ signals in intact SKBR3 cells expressing TPC1 (Fig. 6).
Consistent with our previous analysis (44), microinjection of NAADP evoked robust Ca2+
signals in TPC1-expressing cells compared to mock-transfected cells (Fig. 6A). Bath
exposure to nifedipine inhibited NAADP-evoked response in a concentration-dependent
manner (Fig. 6B–C) with an IC50 (4 μM) similar to that observed for blockade of
endogenous NAADP responses in sea urchin egg homogenates (27 μM, Fig. 5E). Both
lidocaine and veratridine also blocked Ca2+ signals stimulated by NAADP in TPC1-
expressing cells (Fig. 6D).
TPCs possess noncanonical selectivity filters
CaVs and NaVs are highly selective for their respective ions, with well-defined selectivity
filters comprising conserved residues in each of the four domains (typically “EEEE” in CaVs
and “DEKA” in NaVs) (1). The ion selectivity of TPCs is not clear and no structural
information is available. Inspection of the model indicated that the central cavity of TPC is
constricted at the luminal end by short helices in DI (residues 278–282) and DII (637–640)
Rahman et al. Page 5
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

corresponding to the selectivity filter of AbuNaV (Fig. 2A, yellow region). There was a
notable absence of conserved charged residues in this region, consistent with studies
demonstrating poor ion selectivity of TPC2 (24, 25). The alanine residue in DIV of
selectivity filters in NaV, however, was present in the TPC model, although the lysine of
DIII, thought to be critical for Na+ selectivity through electrostatic repulsion of Ca2+ (45),
was not. We identified asparagine residues that were highly conserved in animal TPCs (Fig.
2A, residues highlighted in yellow), and the carbonyl groups of the side-chains projected
into the pore and therefore could be appropriately positioned to coordinate cations (Fig. 7A).
This arrangement is similar to the selectivity filters of Ca2+-permeable glutamate receptors
of the N-methyl-D-aspartate type (46) but differs from the selectivity filters of CaV and NaV.
Indeed, the position of these putative coordinating residues is not obvious from linear
sequence alignment with CaV and NaV (Fig. 2A) (33). Thus, the modeling data suggest that
TPCs possess noncanonical selectivity filters compared with the four-domain voltage-gated
ion channels.
Unicellular organisms have a distinct clade of TPCs
The modeling data indicated that selectivity filters of TPCs are divergent from CaV and
NaV. When did these structural changes occur? To gain insight into the evolution of ion
selectivity, we inspected the genomes of unicellular organisms that hold key positions in
animal evolution for TPC homologs (Fig. 7B, table S2). TPCs are present in the
choanoflagellates Monosiga brevicollis and Salpingoeca rosetta, which represent close
unicellular relatives of animals (47). TPCs are also present in Capsaspora owczarzaki, a
holozoan, and Thecamonas trahens, an apusozoan protist, both of which are more distantly
related to animals. Thecamonas trahens belongs to the putative unicellular sister group to
Opisthokonta (animals, fungi, and related protists) (48). Three of these unicellular species
possess multiple genes encoding TPCs (Fig. 7B). We used TPC sequences from these
species together with those from human and sea urchin to perform a phylogenetic
classification (Fig. 7C). We identified homologs of the three-member TPC family
characteristic of deuterostome animals (44). However, the remaining isoforms grouped as a
previously unreported clade, which we term TPCR (for TPC-related; Fig. 7C).
The selectivity filters of TPCRs are distinct
Unicellular TPC1-3 homologs possessed putative selectivity filters similar to the asparagine-
lined filters predicted for animal TPCs (Fig. 7C). The residues that were predicted to
function as the selectivity filters of TPCR were divergent. For example, in DI, the putative
coordinating asparagine residue was conserved but the preceding residue was acidic and
aligned with the coordinating acidic residues in CaV (Fig. 7C). The presence of acidic or
polar serine residues was also noted in DII (Fig. 7C). The identification of such residues in
the putative selectivity filter of TPCRs suggested similarities with CaVs. Together, these
analyses indicate that multiplication of the TPC gene may have occurred before the animal-
fungal split and was followed by changes in ion selectivity within a select TPC cohort
corresponding to the ancestors of CaV and NaV. Acquisition of ion selection thus may have
occurred prior to the intragenic duplication event that gave rise to four-domain channels.
Rahman et al. Page 6
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

TPCRs interact strongly with CaV antagonists
Because the putative selectivity filter of the TPCR family appeared more similar to the Cavs,
we generated homology models for select members of TPCR and TPC from Salpingoeca
rosetta (for which sequences of the pore appear complete) and compared docking of
dihydropyridines. The molecular docking simulations showed that all but one of the
dihydropyridines docked in a tight cluster at a domain interface in TPCR (Fig. 7D), which is
similar to the interaction of dihydropyridines with the DIII/DIV interface in animal CaV
(Fig. 3B). In contrast, dihydropyridines poses in the unicellular TPC models were less
clustered and more central (Fig. 7D), thus resembling data from animal TPC (Fig. 3A).
Importantly, the predicted ΔG values indicated that the interaction of dihydropyridines was
stronger for TPCR than for TPC Table 3). These data provide further support that TPCR are
similar to CaVs.
Discussion
Here, we identified a putative common binding site within TPCs for CaV and NaV
antagonists together with a relatively “loose” pharmacological profile. These data suggested
that that this motif was acquired prior to the domain duplication event(s) that led to four-
domain channels. The concentration range over which TPCs were blocked by NaV
antagonists is similar to that for blockade of NaV (49). This raises the possibility that off-
target effects of NaV antagonists on TPCs may contribute to their efficacy.
We propose that this pharmacological site underwent diversification, accounting for the
relatively selective actions of CaV and NaV modifiers in the four-domain lineage. Evolution
of this site is unlikely to be driven directly by selection pressure at the pharmacological
level, although the existence of endogenous modifiers that manipulate ion channel activity
cannot be ruled out. Perhaps more likely is that emergence and subsequent divergence of the
site is a secondary consequence of the selection pressures associated with an evolving pore –
a key functional domain.
Our analyses also identified a novel clade of TPCs that show similarities to CaV with respect
to both putative pharmacology and ion selection. These data suggested that acidification of
selectivity filters, likely reflecting acquisition of ion selectivity again may have occurred
prior to the putative domain duplication event(s) that led to four-domain channels.
Molecular cloning and functional characterization of these TPCs is urgently required.
Together, our analyses predict a potential trajectory for the evolution of voltage-gated ion
channels and provide a framework for probing evolutionary relevant structural attributes of
these ancient ion channels.
Materials and Methods
Sequence and phylogenetic analyses
Multiple sequence alignment and phylogenetic reconstruction were performed essentially as
described (28, 29). Briefly, nonredundant protein sequences were aligned using either
MAFFT or T-Coffee and columns containing more than 50% gaps were subsequently
Rahman et al. Page 7
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

removed from the sequence alignment using GapStreeze (Gap Strip/Squeeze v 2.1.0).
Unambiguous sequence alignments were then converted to PHYLIP or NEXUS format.
Identifiers for metazoan and unicellular TPCs are listed in table S1 and table S2,
respectively. ProtTest (50) was utilized to select the best-fit evolution model and parameter
estimates for the phylogenetic analyses. Maximum likelihood phylogeny was performed
using the GARLI web service (51) with the LG amino acid substitution model, empirical
frequency estimation, and the four-category discrete gamma model (LG + G + F) selected
by ProtTest and 100 bootstrap replicates. Consensus trees were obtained using the
CONSENSE program from the PHYLIP package (Version 3.69).
Homology modeling
Homology models for the pore regions (S5–S6) of Stronglylocentrotus purpuratus TPC1
(accession D1J6X7) (44), Salpingoeca rosetta TPC1a (accession EGD80440) and TPCRb
(accession EGD78654), and human CaV1.2 (accession Q13936) and NaV1.2 (accession
Q99250) were generated in Modeller 9.11 (http://salilab.org/modeller/) using the 2.7 Å
crystal structure of voltage-gated Na+ channel from Arcobacter butzleri (NaVAb, pdb
3RVY) as a template (6). Sequences of the pore regions from each of the domains were
aligned against the cognate template region using ClustalW2 (http://www.ebi.ac.uk/
Tools/msa/clustalw2/). The alignments used for TPCs are shown in Fig 2A and fig. S2.
Alignments for CaV and NaV corresponded to those reported in (40). We used the
“straightforward” alignment for CaV, and the adjusted alignment (around pore-helix 2) for
NaV (40). The large loops between S5 and pore-helix 1 and between pore-helix 2 and S6 in
CaV and NaV were excluded. For each domain, 100 models were initially generated and the
best model chosen based on the discrete optimized protein energy (DOPE) score
implemented within Modeller (52). Loops in TPCs were modeled using Robetta (http://
robetta.bakerlab.org/) (53). Domains were assembled as tetramers using the NaVAb
structure as template. A head to tail (trans) arrangement for TPCs was assumed (13). Pores
were further refined with KoBaMIN (http://csb.stanford.edu/kobamin) (54). The overall 3D
quality of the individual domains was assessed using Molprobity (http://
molprobity.biochem.duke.edu/). Boundaries and the proportion of Ramachandran-favored
residues for each domain are listed in table S3.
Docking
Docking of ligands was carried out as described previously (55). Briefly, CaV and NaV
antagonists (obtained from the ZINC database; zinc.docking.org) were docked using
AutoDock 4.2 (http://autodock.scripps.edu/). We adopted a blind docking approach using a
Grid map of 80×66×84 points in 0.375 Å spacings that encompassed the entire pore cavity.
Polar hydrogens and the Gasteiger partial atomic charges were added to the protein and
ligands using AutoDockTools™ (http://autodock.scripps.edu/resources/adt). Only the top-
ranked poses (those with the lowest free energy of interaction) of the drugs were considered.
For detailed inspection and analyses of the docked poses, ligand interaction diagrams were
derived using LigPlot+ (http://www.ebi.ac.uk/thornton-srv/software/LigPlus). All structural
representations were prepared using the PyMOL Molecular Graphics System. PDB files for
channel models and docked ligands are available in folders S1–S5.
Rahman et al. Page 8
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Ca2+ measurements
Sea urchin egg homogenates were prepared and loaded with Ca2+ as described previously
(56). Ca2+ release was measured by cuvette-based fluorimetry using the Ca2+ indicators
fluo-3 or fluo-4 (3 μM, Invitrogen). SKBR3 cells were cultured as described previously (16)
and transfected with TurboFectin 8 (Origene) using a 3:1 ratio of reagent:plasmid. The C-
terminally GFP-tagged TPC1 construct from Stronglylcentrotus purpuratus was described in
(44). Ca2+ imaging using Fura-2 and NAADP microinjection (pipette concentration 1 μM)
were performed as described in (23). Fluorescence intensity (Fluo dyes) and ratio (Fura-2)
values were normalized to values prior to stimulation and presented as F/F0 or R/R0,
respectively. All agonists and drugs were from Sigma.
Supplementary Material
Refer to Web version on PubMed Central for supplementary material.
Acknowledgments
We thank Alisdair Gibb, Bethan Kilpatrick, Christopher J. Penny and Martin Stocker for useful discussion.
Funding: This work was supported by grants RG65196 and RG69132 from the Royal Society (to TR), BB/
G013721/1 from the Biotechnology and Biological Sciences Research Council (to SP) and DA035926, DA023204
and P30 DA13429 from the National Institutes of Health (to MEA, EB and the Center for Substance Abuse, Temple
University School of Medicine). TR is a Royal Society University Research Fellow. XC is an Honorary Senior
Research Fellow at UCL.
References
1. Yu FH, Yarov-Yarovoy V, Gutman GA, Catterall WA. Pharmacol Rev. 2005; 57:387. [PubMed:
16382097]
2. Hille, B. Ionic channels of excitable membranes. 1. Sinauer Associates; Sunderland, Massachusets:
1984.
3. Ryan DP, Ptacek LJ. Neuron. 2010; 68:282. [PubMed: 20955935]
4. Savio-Galimberti E, Gollob MH, Darbar D. Front Pharmacol. 2012; 3:124. [PubMed: 22798951]
5. Cain SM, Snutch TP. Biofactors. 2011; 37:197. [PubMed: 21698699]
6. Payandeh J, Scheuer T, Zheng N, Catterall WA. Nature. 2011; 475:353. [PubMed: 21743477]
7. Zhang X, et al. Nature. 2012; 486:130. [PubMed: 22678295]
8. Payandeh J, Gamal El-Din TM, Scheuer T, Zheng N, Catterall WA. Nature. 2012; 486:135.
[PubMed: 22678296]
9. McCusker EC, et al. Nat Commun. 2012; 3:1102. [PubMed: 23033078]
10. Clapham DE, Garbers DL. Pharmacol Rev. 2005; 57:451. [PubMed: 16382101]
11. Hooper R, Churamani D, Brailoiu E, Taylor CW, Patel S. J Biol Chem. 2011; 286:9141. [PubMed:
21173144]
12. Churamani D, Hooper R, Brailoiu E, Patel S. Biochem J. 2012; 441:317. [PubMed: 21992073]
13. Rietdorf K, et al. J Biol Chem. 2011; 286:37058. [PubMed: 21903581]
14. Patel S, Docampo R. Trends Cell Biol. 2010; 20:277. [PubMed: 20303271]
15. Peiter E, et al. Nature. 2005; 434:404. [PubMed: 15772667]
16. Brailoiu E, et al. J Cell Biol. 2009; 186:201. [PubMed: 19620632]
17. Calcraft PJ, et al. Nature. 2009; 459:596. [PubMed: 19387438]
18. Hooper R, Patel S. Adv Exp Med Biol. 2012; 740:325. [PubMed: 22453949]
19. Churchill GC, et al. Cell. 2002; 111:703. [PubMed: 12464181]
Rahman et al. Page 9
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

20. Lee HC, Aarhus R. J Biol Chem. 1995; 270:2152. [PubMed: 7836444]
21. Lin-Moshier Y, et al. J Biol Chem. 2012; 287:2296. [PubMed: 22117075]
22. Wang X, et al. Cell. 2012; 151:372. [PubMed: 23063126]
23. Jha A, Ahuja M, Patel S, Brailoiu E, Muallem S. EMBO J. 2014; 33:501. [PubMed: 24502975]
24. Rybalchenko V, et al. J Biol Chem. 2012; 287:20407. [PubMed: 22500018]
25. Pitt SJ, et al. J Biol Chem. 2010; 285:24925. [PubMed: 20547763]
26. Schieder M, Rotzer K, Bruggemann A, Biel M, Wahl-Schott CA. J Biol Chem. 2010; 285:21219.
[PubMed: 20495006]
27. Pitt SJ, Lam AK, Rietdorf K, Galione A, Sitsapesan R. Sci Signal. 2014; 7:ra46. [PubMed:
24847115]
28. Liebeskind BJ, Hillis DM, Zakon HH. Proc Natl Acad Sci U S A. 2011; 108:9154. [PubMed:
21576472]
29. Cai X. J Membr Biol. 2012; 245:117. [PubMed: 22258316]
30. Gur BM, et al. Cell Rep. 2012; 2:242. [PubMed: 22854023]
31. Ren D, et al. Science. 2001; 294:2372. [PubMed: 11743207]
32. Liebeskind BJ, Hillis DM, Zakon HH. Curr Biol. 2013; 23:R948. [PubMed: 24200318]
33. Ishibashi K, Suzuki M, Imai M. Biochem Biophys Res Commun. 2000; 270:370. [PubMed:
10753632]
34. Strong M, Chandy KG, Gutman GA. Mol Biol Evol. 1993; 10:221. [PubMed: 7680747]
35. Hockerman GH, Peterson BZ, Johnson BD, Catterall WA. Annu Rev Pharmacol Toxicol. 1997;
37:361. [PubMed: 9131258]
36. Genazzani AA, Empson RM, Galione A. J Biol Chem. 1996; 271:11599–11602. [PubMed:
8662773]
37. Cang C, Bekele B, Ren D. Nat Chem Biol. 2014; 10:463. [PubMed: 24776928]
38. Lee HC. J Biol Chem. 2012; 287:31633. [PubMed: 22822066]
39. Tikhonov DB, Zhorov BS. J Biol Chem. 2009; 284:19006. [PubMed: 19416978]
40. Tikhonov DB, Zhorov BS. Mol Pharmacol. 2012; 82:97. [PubMed: 22505150]
41. Hockerman GH, Johnson BD, Abbott MR, Scheuer T, Catterall WA. J Biol Chem. 1997;
272:18759. [PubMed: 9228049]
42. Yarov-Yarovoy V, et al. J Biol Chem. 2001; 276:20. [PubMed: 11024055]
43. Catterall WA. J Biol Chem. 1975; 250:4053. [PubMed: 1168643]
44. Brailoiu E, et al. J Biol Chem. 2010; 285:2897. [PubMed: 19940116]
45. Heinemann SH, Terlau H, Stuhmer W, Imoto K, Numa S. Nature. 1992; 356:441. [PubMed:
1313551]
46. Burnashev N, et al. Science. 1992; 257:1415. [PubMed: 1382314]
47. King N. Curr Biol. 2005; 15:R113. [PubMed: 15723775]
48. Cavalier-Smith T, Chao EE. Protist. 2010; 161:549. [PubMed: 20537943]
49. Catterall WA, Striessnig J. Trends Pharmacol Sci. 1992; 13:256. [PubMed: 1321525]
50. Darriba D, Taboada GL, Doallo R, Posada D. Bioinformatics. 2011; 27:1164. [PubMed:
21335321]
51. Bazinet AL, Zwickl DJ, Cummings MP. Syst Biol. 2014; 63:812. [PubMed: 24789072]
52. Eswar N, et al. Curr Protoc Protein Sci. 2007; Chapter 2 Unit.
53. Kim DE, Chivian D, Baker D. Nucleic Acids Res. 2004; 32:W526. [PubMed: 15215442]
54. Rodrigues JP, Levitt M, Chopra G. Nucleic Acids Res. 2012; 40:W323. [PubMed: 22564897]
55. O’Reilly AO, et al. PLoS ONE. 2012; 7:e41667. [PubMed: 22848561]
56. Dickinson GD, Patel S. Biochem J. 2003; 375:805. [PubMed: 12914540]
Rahman et al. Page 10
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Fig. 1. Domain phylogeny of multidomain, voltage-gated ion channels
A, Schematic showing architecture of four-domain CaVs and Navs (top) and two-domain
TPCs (bottom). Each domain (DI-DIV) comprises six transmembrane regions (S1–S6,
numbered). S1–S4 form the voltage-sensing domain (VSD) and S5–S6 form the pore (P).
Arrows depict the direction of ion flow into the cytosol from either the extracellular space
(CaV and NaV) or from the lumen of acidic organelles (TPC). B, Unrooted maximum
likelihood tree constructed using sequences of individual domains of TPCs, CaVs, and NaVs
from representative members of the chordate, cephalachordate, echinoderm, and cnidarian
Rahman et al. Page 11
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

phyla. Species used were Homo sapiens (Hsa), the sea squirt Ciona intestinalis (Cin), the sea
urchin Stronglylocentrotus purpuratus (Spu), and the starlet sea anemone Nematostella
vectensis (Nve). Accession numbers are listed in Table S1. Bootstrap values at the basal
branches are shown. All other values were 40–100. Similar domains inferred from the
phylogenetic relationships are shaded green (I and III) and pink (II and IV) in B, and
connected by the colored lines in A.
Rahman et al. Page 12
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Fig. 2. A structural model of the TPC pore
A, Structure-based alignment of the pore regions of human (Hsa) and sea urchin (Spu) TPCs
(outlined with dashes in schematic) with prokaryotic NaVs from Arcobacter butzleri (Abu)
and Rickettsiales sp. HIMB114 (Rhi). Positions of S5 and S6 in NaVs are indicated with the
purple bars, the intervening turret region with a grey bar, the first (PH1) and second (PH2)
pore helices with green bars, and the selectivity filter (SF) with a yellow bar. Large
insertions within the corresponding turret regions of TPCs (ˆ) were omitted from the
alignment for clarity. Residues that coordinate cations in NaV are shaded cyan. Conserved
asparagine residues in TPCs within the selectivity filter are yellow. Black shading indicates
Rahman et al. Page 13
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

sequence identity and gray indicates conserved substitution. B, Homology model of the pore
of sea urchin TPC1 in side (left), cytosolic (middle), and luminal (right) orientations. Side
views are depicted in an upright (as opposed to inverted) “tepee” fashion to reflect their
organellar (as opposed to plasma membrane) subcellular localization.
Rahman et al. Page 14
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Fig. 3. Interaction of CaV antagonists with TPC
(A–C) Docking of a series of 8 dihydropyridines (DHP) to the pore of TPC (A), CaV (B),
and NaV (C) depicted in either side (left) or cytosolic (middle) orientations. Gray arrows
depict the direction of ion flow. Right panel shows poses for all ligands represented by the
gray mesh (cytosolic view) with the indicated select ligands highlighted in yellow or green.
White arrows mark the interfaces between DIII–DIV (in Cav) and DII–DIII (Nav). D,
Overlay (upper panel) of dihydropyridine poses in mesh representation for TPC, CaV, and
NaV. Plot (lower panel) shows ΔG values for docking of ligands to the three channel types.
Rahman et al. Page 15
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

The closed symbols are values for nifedipine. (E). Docking of verapamil and diltiazem to
the TPC pore (F). Ca2+ signals recorded from sea urchin egg homogenates stimulated with 1
μM NAADP in the absence (black traces) or presence (colored traces) of 100 μM nifedipine
(Nif.), isradipine (Isra.), verapamil (Vera.), or diltiazem (Dil.). Data are representative of 3
experiments.
Rahman et al. Page 16
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Fig. 4. Interaction of NaV antagonists with TPC
A, Docking of a series of 10 local anaesthetics (LA) to the pore of TPC depicted in either
side (left) or cytosolic (middle) orientation. Arrow depicts the direction of ion flow. Right
panel shows poses for all ligands represented by the grey mesh (cytosolic view), with
lidocaine highlighted in yellow. B, Representative Ca2+ signals recorded from sea urchin
egg homogenates stimulated with 1 μM NAADP or 5 μM cyclic ADP-ribose in the absence
(black traces) or presence of 3 mM lidocaine (blue traces) or 1 mM bupivacaine (gray
traces). C, Pooled data (mean ± s.e.m. of 3 independent experiments) quantifying the effect
of NaV antagonists on NAADP- and cADPR-induced Ca2+ release. D, Inhibition curve
showing concentration-dependent block of NAADP responses by lidocaine (blue) and
bupivacaine (gray).
Rahman et al. Page 17
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Fig. 5. Comparison of CaV and NaV antagonist docking
A–B, Zoomed views comparing docking of CaV (A) and NaV (B) antagonists to TPC (grey
ribbons) and their cognate four domain channels (white ribbons). Ligands are colored yellow
for docking to TPCs and green for docking to CaV and NaV. Arrows depict the direction of
ion flow. Nica., nicardipine; Mepi., mepivacaine; Etido., etidocaine; Prilo., prilocaine; Bupi.,
bupivacaine; Lido., lidocaine; Trime., trimecaine. C, Overlay of poses for the indicated CaV
and NaV antagonists docked to TPC. Dashed box highlights congruent nature of poses.
Arrow depicts the direction of ion flow through TPC. D, Interacting residues within the S6
Rahman et al. Page 18
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

regions of DI and DII of TPC for the ligands in C (same color code). Arrowheads highlight
residues implicated in interaction of both CaV and NaV antagonists. Known molecular
determinants for interactions of phenylalkylamines with rat CaV2.1 and local anaesthetics
and anticonvulsants with rat Nav2.1 are underlined in the corresponding S6 sequences of
DIII (RnoCaV, RnoNaV). E, Inhibition curves (left) showing concentration-dependent bock
of NAADP-induced Ca2+ release from sea urchin egg homogenates by the indicated ligands.
Plot (right) showing correlation of the half-maximal inhibitory concentrations (IC50) in
Ca2+-release assays for CaV and NaV antagonists with their predicted ΔG values for
docking. F, Representative Ca2+ signals recorded from sea urchin egg homogenates
stimulated with 1 μM NAADP or 5 μM cyclic ADP-ribose in the absence (black traces) or
presence of 100 μM veratridine (Verat., blue traces). G, Inhibition curve showing
concentration-dependent block of NAADP responses by veratridine (IC50 = 52 μM, n=2).
Rahman et al. Page 19
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Fig. 6. Effect of CaV and NaV modifiers on recombinant TPC1
Cytosolic Ca2+ signals from individual fura-2-loaded SKBR3 cells that were microinjected
with NAADP. A, Responses in mock-transfected cells (blue trace) or cells transiently
expressing recombinant sea urchin TPC1 (black traces). B, Responses in TPC1-expressing
cells pre-incubated for 1 h with increasing concentrations nifedipine (Nif.). Concentrations
used (from top to bottom) were 0.1, 1, 10, and 100 μM. C, Inhibition curve showing
concentration-dependent block of NAADP responses by nifedipine. D, Responses in TPC1-
expressing cells pre-incubated for 1 h with lidocaine (Lido. 100 μM) or veratridine (Verat.
100 μM). Results are means ± s.e.m. of 6 cells from 3 independent transfections.
Rahman et al. Page 20
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

Fig. 7. Properties of TPCs from unicellular organisms
A, Zoomed view of the TPC pore showing the presence of conserved asparagine residues
positioned within the putative selectivity filter to coordinate cations. B, Evolutionary
relationships of unicellular organisms in the lineages leading to metazoans (animals). The
number of identified TPCs in each of the species is shown in the boxes. C, Cladogram of
TPC sequences of representative metazoans (Hsa, human; Spu, sea urchin; shaded),
choanoflagellates (Mbr, Monosiga brevicollis; Sro, Salpingoeca rosetta), and basal species
(Cow, Capsaspora owczarzaki; Ttr, Thecamonas trahens). Accession numbers are listed in
table S2. CaV was used as the out-group (accession EGD78396.1). Bootstrap values were
81–100 except where indicated (*23–78). A previously unreported grouping (TPCR) is
Rahman et al. Page 21
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

highlighted by the dashed box. Sequences of the putative selectivity filters are shown to the
right. Acidic (Asp, Glu; yellow) and polar (Ser; cyan) residues are shaded. D, Interaction of
dihydropyridines with TPCR. Docking of a series of 8 dihydropyridines to the pore of TPCR
(top) and TPC (bottom) from Salpingoeca rosetta depicted in cytosolic orientations. White
arrow marks an interface between DI and DII.
Rahman et al. Page 22
Sci Signal. Author manuscript; available in PMC 2015 November 18.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript

NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
Rahman et al. Page 23
Table 1
Cav antagonist docking to metazoan channels. ΔG values for interaction of the indicated CaV antagonist with
TPC, CaV and NaV. nd, not determined.
Ligand ΔG TPC
(kcal/mol) ΔG CaV
(kcal/mol) ΔG NaV
(kcal/mol)
Amlodipine −5.3 −6.2 −5.1
Felodipine −5.8 −7.1 −5.5
Isradipine −5.9 −7.4 −5.9
Nicardipine −6.5 −7.6 −6.0
Nifedipine −6.1 −6.8 −3.1
Nimodipine −5.2 −6.7 −4.4
Nitrendipine −6.0 −6.9 −4.9
Oxodipine −4.7 −6.9 −4.9
Verapamil −5.1 nd nd
Diltiazem −5.8 nd nd
Sci Signal. Author manuscript; available in PMC 2015 November 18.

NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
Rahman et al. Page 24
Table 2
Nav antagonist docking to metazoan channels. ΔG values for interaction of local anaesthetics with TPC and
NaV.
Ligand ΔG TPC
(kcal/mol) ΔG NaV
(kcal/mol)
Benzocaine −4.9 −4.4
Bupivacaine −5.2 −6.9
Dibucaine −6.2 −6.3
Etidocaine −6.0 −5.7
Lidocaine −4.3 −5.7
Mepivacaine −6.7 −6.8
Prilocaine −5.8 −5.1
Procaine −5.0 −4.9
Tetracaine −5.1 −5.0
Trimecaine −6.1 −6.2
Sci Signal. Author manuscript; available in PMC 2015 November 18.

NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
Rahman et al. Page 25
Table 3
Cav antagonist docking to unicellular TPCs. ΔG values for interaction of dihydropyridines with TPC and
TPCR from Salpingoeca rosetta.
Ligand ΔG TPC
(kcal/mol) ΔG TPCR
(kcal/mol)
Amlodipine −4.9 −6.1
Felodipine −5.0 −6.4
Isradipine −5.3 −6.9
Nicardipine −5.5 −7.1
Nifedipine −5.1 −6.0
Nimodipine −4.9 −5.1
Nitrendipine −5.8 −5.9
Oxodipine −5.4 −6.6
Sci Signal. Author manuscript; available in PMC 2015 November 18.