Seven novel modulators of the analgesic target Na_V 1.7 uncovered using a high-throughput venom-based discovery approach

Abstract

Background and Purpose

Chronic pain is a serious worldwide health issue, with current analgesics having limited efficacy and dose-limiting side effects. Humans with loss-of-function mutations in the voltage-gated sodium channel Na_V1.7 (hNa_V1.7) are indifferent to pain, making hNa_V1.7 a promising target for analgesic development. Since spider venoms are replete with Na_V channel modulators, we examined their potential as a source of hNa_V1.7 inhibitors.

Experimental Approach

We developed a high-throughput fluorescent-based assay to screen spider venoms against hNa_V1.7 and isolate ‘hit’ peptides. To examine the binding site of these peptides, we constructed a panel of chimeric channels in which the S3b-S4 paddle motif from each voltage sensor domain of hNa_V1.7 was transplanted into the homotetrameric K_V2.1 channel.

Key Results

We screened 205 spider venoms and found that 40% contain at least one inhibitor of hNa_V1.7. By deconvoluting ‘hit’ venoms, we discovered seven novel members of the NaSpTx family 1. One of these peptides, Hd1a (peptide μ-TRTX-Hd1a from venom of the spider Haplopelma doriae), inhibited hNa_V1.7 with a high level of selectivity over all other subtypes, except hNa_V1.1. We showed that Hd1a is a gating modifier that inhibits hNa_V1.7 by interacting with the S3b-S4 paddle motif in channel domain II. The structure of Hd1a, determined using heteronuclear NMR, contains an inhibitor cystine knot motif that is likely to confer high levels of chemical, thermal and biological stability.

Conclusion and Implications

Our data indicate that spider venoms are a rich natural source of hNa_V1.7 inhibitors that might be useful leads for the development of novel analgesics.

Tables of Links

TARGETS
Cav channels	Nav1.2 channel	Nav1.6 channel
Kv2.1 channel	Nav1.3 channel	Nav1.7 channel
Kv11.1 (ERG) channel	Nav1.4 channel	Nav1.8 channel
Nav1.1 channel	Nav1.5 channel	Nav1.9 channel

LIGANDS
Glutathione
Tetrodotoxin (TTX)
Veratridine

These Tables list key protein targets and ligands in this article which are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Pawson et al., 2014) and are permanently archived in the Concise Guide to PHARMACOLOGY 2013/14 (Alexander et al., 2013).

Introduction

Chronic pain is a major health issue worldwide that affects ∼15% of the adult population (Gaskin and Richard, 2012). The annual economic burden of chronic pain in the United States is ∼$600 billion, which exceeds the combined economic cost of cancer, diabetes and stroke (Gaskin and Richard, 2012). There is an unmet clinical need for more effective analgesics to treat chronic pain as most currently available drugs have limited efficacy and dose-limiting side effects.

Voltage-gated sodium (Na_V) channels are transmembrane proteins that regulate the electrical properties of cells. There are nine mammalian subtypes denoted Na_V1.1–Na_V1.9 (Catterall et al., 2005). Several remarkable genetic studies led to the emergence of human Na_V1.7 (hNa_V1.7) as an analgesic target. Gain-of-function mutations in the SNC9A gene encoding hNa_V1.7 cause painful inherited neuropathies (Yang et al., 2004; Fertleman et al., 2006; Estacion et al., 2008; Cheng et al., 2011; Theile et al., 2011), whereas loss-of-function mutations result in congenital indifference to all forms of pain (Cox et al., 2006). Moreover, single nucleotide polymorphisms in SCN9A are associated with differences in pain sensitivity (Reimann et al., 2010; Duan et al., 2013; Reeder et al., 2013). Thus, the combined genetic data suggest that subtype-selective inhibitors of hNa_V1.7 are likely to be useful analgesics for treating a broad range of pain conditions (King and Vetter, 2014).

Spider venoms are a rich source of peptidic Na_V channel modulators (Herzig et al., 2011; Gilchrist et al., 2014), which are classified into 12 families of Na_V channel toxins (NaSpTxs) based upon their primary structure and disulfide framework (Klint et al., 2012). Most NaSpTxs are gating modifiers that perturb channel function by stabilizing one or more voltage sensors in a particular state (Gilchrist et al., 2014). Since the voltage sensors are less conserved than the pore region of Na_V channels, these toxins have the potential to selectively modulate particular Na_V channel subtypes.

Spiders are the most successful group of venomous animals, with >45 000 extant species. Their venoms contain hundreds to thousands of peptides (Escoubas et al., 2006), with even a conservative estimate of 200 peptides per venom, leading to a total of 9 million spider-venom peptides. Thus far, only ∼0.01% of this vast pharmacological landscape has been explored, providing massive scope for discovery of novel Na_V channel modulators from these arachnids. Here, we describe a high-throughput pipeline for the discovery and characterization of hNa_V1.7 modulators from spider venoms. We report seven novel peptides from NaSpTx family 1 (NaSpTx-F1) discovered using this approach, and use one of these peptides to exemplify high-throughput methods we developed for determining their structure and Na_V channel binding site. We anticipate that development of detailed structure–function relationships for members of NaSpTx-F1, including the seven peptides described here, will facilitate the engineering of selective inhibitors of hNa_V1.7 that might be useful analgesics.

Methods

High-throughput screen of spider venoms against hNa_V1.7

Mild electrical currents were applied to the chelicerae of spiders to stimulate venom secretion. Venom was collected in polypropylene tubes, lyophilized and stored at −20°C until required. Venoms and venom fractions were assessed against hNa_V1.7 endogenously expressed in SH-SY5Y neuroblastoma cells (Vetter et al., 2012). Briefly, SH-SY5Y cells were maintained at 37°C/5% CO₂ in RPMI containing 15% FBS and 2 mM L-glutamine and seeded on black-walled 96- or 384-well imaging plates at a density of 30 000 or 150 000 cells per well. After 48 h, cells were loaded for 30 min at 37°C with Calcium 4 No-Wash dye (Molecular Devices, Sunnyvale, CA, USA) diluted in physiological salt solution [composition (in mM) NaCl 140, glucose 11.5, KCl 5.9, MgCl₂ 1.4, NaH₂PO₄ 1.2, NaHCO₃ 5, CaCl₂ 1.8, HEPES 10]. Ca²⁺ responses following addition of veratridine (50 μM) were measured using a fluorescence imaging plate reader (FLIPR^TETRA; Molecular Devices) (excitation 470–495 nm, emission 515–575 nm) after 5 min pre-incubation with venom or venom fractions. Under these conditions, most of the veratridine-induced response is mediated by hNa_V1.7, with smaller contributions from hNa_V1.2 and hNa_V1.3 (Vetter et al., 2012).

Isolation of Na_V1.7 inhibitors

Venom (1 mg) was diluted with H₂O, filtered (Ultrafree-MC Centrifugal Filter, 0.22 μm; Merck Millipore, Bayswater, Vic, Australia) and then loaded onto an analytical C₁₈ reverse-phase (RP) HPLC column (Vydac 4.6 × 250 mm, 5 μm; Grace, Columbia, MD, USA) attached to a Prominence HPLC system (Shimadzu, Rydalmere, NSW, Australia). Components were eluted at 1 mL·min⁻¹ with solvent A [99.5% H₂O, 0.05% trifluoroacetic acid (TFA)] and solvent B (90% CH₃CN, 0.05% TFA in H₂O) using isocratic elution at 5% solvent B for 5 min, followed by a gradient of 5–20% solvent B over 5 min, then 20–40% solvent B over 40 min and then 40–80% solvent B over 5 min (or minor variations of this gradient). hNa_V1.7-active fractions were further fractionated using a polysulfoethyl cation exchange column (4.6 × 100 mm, 3 μm, 300 Å pore size) and eluted at 1 mL·min⁻¹ with solvent A [KH₂PO₄, 20% (v/v) CH₃CN, pH 2.7] and solvent B (1 M KCl in solvent A) using a gradient of 0–50% solvent B over 50 min. Absorbance was measured at 214 and 280 nm using a Shimadzu Prominence SPD-20A detector.

Sequencing of Na_V1.7-active peptides

Peptide masses were determined using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) MS using a Model 4700 Proteomics Analyser (Applied Biosystems, Foster City, CA, USA). HPLC fractions were spotted with α-cyano-4-hydroxycinnamic acid (5 mg·mL⁻¹ in 50% CH₃CN). Peptides were reduced and alkylated before being sequenced. Disulfide bonds were reduced by incubating peptides for 15 min at 65^oC in 150 mM Tris (pH 8), 1 mM EDTA and 5 mM DTT. Thiol groups were pyridylethylated using 5 μL of 95% 4-vinylpyridine and 20 μL of CH₃CN; the reaction was allowed to proceed for 2 h in darkness at room temperature (RT). Both reactions were performed under N₂. Samples were desalted using RP-HPLC (C₁₈ column, 2.1 × 100 mm, 5 μm; Thermo Scientific, Waltham, MA, USA) and eluted in solvent B (5% for 15 min, 5–40% over 35 min, 40–80% over 10 min, 0.25 mL·min⁻¹). Reaction progress was monitored using MS. N-terminal sequencing was performed by the Australian Proteome Analysis Facility.

Production of recombinant Hd1a

A synthetic gene encoding Hd1a (peptide μ-TRTX-Hd1a from the venom of the spider Haplopelma doriae), with codons optimized for Escherichia coli expression, was cloned into the pLic-MBP (maltose binding protein) vector (Cabrita et al., 2006). This plasmid, which enables periplasmic production of a His₆-MBP-Hd1a fusion protein, was transformed into E. coli strain BL21(λDE3) for Hd1a production. Cultures were grown in Luria-Bertani medium at 37°C with shaking at 180 r.p.m. When the OD₆₀₀ reached 0.8–1.0, the culture was cooled to 16°C and induced with 1 mM isopropyl b-d-1-thiogalactopyranoside (IPTG). Cells were harvested 12–14 h later by centrifugation for 15 min at 8000× g. For production of uniformly ¹³C/¹⁵N-labelled Hd1a, cultures were grown in minimal medium supplemented with ¹³C₆-glucose and ¹⁵NH₄Cl as the sole carbon and nitrogen sources respectively. Cells were disrupted under constant pressure at 30 kPa (TS Series Cell Disrupter, Constant Systems, Northants, UK), then the His₆-MBP-toxin fusion protein was captured on Ni-NTA Superflow resin (Qiagen, Valencia, CA, USA). The resin was washed with 15 mM imidazole to remove non-specifically bound proteins and then fusion protein was eluted with 500 mM imidazole. After buffer exchange to remove imidazole, reduced and oxidized glutathione were added to 0.6 and 0.4 mM, respectively, to activate TEV protease and promote peptide folding. After the addition of His₆-tagged TEV protease (∼40 μg·mg⁻¹ Hd1a), cleavage was allowed to proceed at RT for 12 h. His₆-MBP and His₆-TEV protease were precipitated with TFA (1%), then Hd1a was purified from the supernatant using RP-HPLC (C₄ column; 250 × 10 mm, 5 μm; flow rate 3 mL·min⁻¹; Phenomenex, Torrance, CA, USA) with a gradient of 10–60% solvent B (0.43% TFA in 90% acetonitrile) in solvent A (0.5% TFA in water) over 50 min. If required, a final purification was performed (Aquasil C₁₈ column, 150 × 4.6 mm, 5 μm; flow rate 1 mL·min⁻¹) using a gradient of 25–30% solvent B over 20 min.

Construction of hNa_V1.7/K_V2.1 chimeras

Channel chimeras were generated using sequential PCR with K_V2.1Δ7 (Frech et al., 1989; Swartz and MacKinnon, 1997) and hNa_V1.7 (Origene Technologies, Rockville, MD, USA) as templates. The K_V2.1Δ7 construct contains seven-point mutations in the outer vestibule that render the channel sensitive to agitoxin-2, a pore-blocking scorpion toxin (Garcia et al., 1994). cRNA was synthesized using T7 polymerase (mMessage mMachine kit; Life Technologies, Federick, MD, USA) after linearizing DNA with appropriate restriction enzymes.

Electrophysiology

Xenopus laevis oocytes were injected with cRNA encoding hNa_V α and β1 subunits, K_V2.1, K_V2.1/hNa_V1.7 chimeras, or K_V2.1/hNa_V1.9 chimeras. Two-electrode voltage-clamp electrophysiology (OC-725C, Warner Instruments, Hamden, CT, USA; 150 μL recording chamber) was used to measure currents 1–4 days after cRNA injection and incubation at 17°C in ND96 that contained (in mM) 96 NaCl, 2 KCl, 5 HEPES, 1 MgCl₂, 1.8 CaCl₂ and 50 μg·mL⁻¹ gentamycin, pH 7.6. Data were filtered at 4 kHz and digitized at 20 kHz using pClamp software (Molecular Devices). Microelectrode resistances were 0.5–1 MΩ when filled with 3 M KCl. For K_V channel experiments, the external recording solution contained (in mM) 50 KCl, 50 NaCl, 5 HEPES, 1 MgCl₂, 0.3 CaCl₂, pH 7.6 with NaOH. For Na_V channel experiments, the external recording solution contained (in mM) 96 NaCl, 2 KCl, 5 HEPES, 1 MgCl₂, 1.8 CaCl₂, pH 7.6 with NaOH. All experiments were performed at RT (∼22 °C). Toxin samples were diluted in recording solution with 0.1% BSA. Leak and background conductance, identified by blocking channels with agitoxin-2 or tetrodotoxin (TTX), were subtracted for all K_V or hNa_V1.7 currents respectively.

Voltage–activation relationships were obtained by measuring tail currents for K_V channels or by monitoring steady-state currents and calculating conductance for Na_V channels. Occupancy of closed or resting channels by toxins was examined using negative holding voltages where the open probability was very low, and the fraction of unbound channels was estimated using depolarizations too weak to open toxin-bound channels. After the addition of toxin to the recording chamber, equilibration between toxin and channel was monitored using weak depolarizations elicited at 5–10 s intervals. For all channels, voltage–activation relationships were recorded in the absence and presence of toxin. Off-line data analysis was performed using Clampfit (Molecular Devices) and Origin 7.5 (Originlab, Northampton, MA, USA).

Determination of Hd1a structure

¹³C/¹⁵N-labelled Hd1a (500 μM in 20 mM sodium acetate buffer containing 5% D₂O, pH 4.9) was filtered (Ultrafree-MC 0.22 μm centrifugal filter; Merck Millipore) and then 300 μL was added to a susceptibility-matched microtube (Shigemi, Allison Park, PA, USA). Data were acquired at 25°C on a 900 MHz NMR spectrometer (Bruker Biospin, Ettlingen, Germany) equipped with a cryogenic probe. Spectra used for resonance assignments were acquired using non-uniform sampling (NUS) (Mobli et al., 2010). NUS data were processed using the Rowland NMR toolkit with automatic selection of maximum entropy parameters (Mobli et al., 2007). ¹³C- and ¹⁵N-edited HSQC-NOESY experiments (mixing time 200 ms) were acquired with uniform sampling. Dihedral angles were derived using TALOS+ (Shen et al., 2009); restraint ranges for structure calculations were set to twice the estimated SD. NOESY spectra were manually peak picked and integrated, then peak lists were automatically assigned, distance restraints extracted and structures calculated using CYANA 3.0 (Güntert, 2004).

Results

Screening spider venoms for hNa_V1.7 modulators

For this assay, we utilized a human cell line (SH-SY5Y) that expresses high levels of hNa_V1.7 (Vetter et al., 2012). Activation of hNa_V1.7 with veratridine leads to membrane depolarization, which activates endogenous voltage-gated calcium (Ca_V) channels (Sousa et al., 2013). The resultant influx of calcium ions induces a fluorescence response when they bind internalized Calcium 4 dye (see control wells 1A–1H and 12A–12H in Figure 1A). Due to the large cellular Ca²⁺ gradient and the comparatively high sensitivity of calcium dyes, this method yields better signal-to-noise ratio than alternative approaches, including membrane potential or Na⁺-specific dyes. We utilized a two-addition protocol in which spider venoms (0.4–20.0 μg) and subsequently veratridine were added to the wells of a 96-well plate, and used this assay to screen for hNa_V1.7 modulators. Venoms with putative Na_V1.7 inhibitors caused a loss of the veratridine-induced fluorescent signal (e.g., wells 2A, 3E, 5F in Figure 1A). In contrast, putative activators were characterized by a spontaneous increase in fluorescence prior to addition of veratridine (e.g., wells 2G and 7B). Some venoms appear to contain both activators and inhibitors of Na_V1.7 as they induced a fluorescent signal prior to veratridine addition and also blocked the veratridine-induced response (e.g., well 3G). Since spider venoms are also a rich source of Ca_V channel modulators (King, 2007; King et al., 2008a), some venoms might contain molecules that abrogate the fluorescent response due block of Ca_V rather than Na_V channels. Modulation of hNa_V1.7 by crude venoms could also be due to the synergistic effect of multiple compounds, an effect that might be lost upon venom fractionation.

Figure 1.

High-throughput screen of spider venoms against hNa_V1.7. (A) Example of a 96-well plate from a FLIPR-based screen of spider venoms against hNa_v1.7 endogenously expressed in SH-SY5Y cells. Wells 1A–1H and 12A–12H (pale green) were buffer controls showing the fluorescent response obtained when hNa_V1.7 was activated by veratridine. All other wells contained spider venoms. Wells containing venoms that inhibited the veratridine-induced fluorescent response are highlighted in blue, whereas wells containing venoms that elicited a fluorescence response before the addition of veratridine and thus putatively contain activators of hNa_V1.7 are highlighted in red. (B) Statistics of venom screen. Eighty-two of the 205 spider venoms inhibited the fluorescent response and therefore putatively contain inhibitors of hNa_V1.7, whereas 30 venoms caused a spontaneous increase in fluorescence, suggesting that they contain activators of hNa_V1.7. Some venoms caused a spontaneous increase in fluorescence as well as a subsequent reduction in the veratridine-induced fluorescent response (e.g. well G3), and hence they putatively contain both activators and inhibitors of hNa_V1.7. For the statistical analysis, these venoms were counted only as inhibitors; hence, the number of activators might be higher than indicated here.

We used this assay to screen venoms from 205 spiders belonging to 13 different taxonomic families (Supporting Information Table S1). Eighty-two venoms (40%) inhibited the fluorescent response and therefore putatively contain at least one inhibitor of hNa_V1.7, whereas 30 venoms (14.6%) caused a spontaneous increase in fluorescence and therefore putatively contain at least one activator of hNa_V1.7 (Figure 1B). Crude venoms that potently inhibited hNa_V1.7 were fractionated using RP-HPLC and then individual fractions were tested in the FLIPR assay. Active fractions were purified to single components (peptides) using an orthogonal ion-exchange chromatography step. These peptides were reduced, alkylated and sequenced using Edman degradation. Using this approach, 15 novel peptides were isolated from eight venoms. Seven peptides were found to be members of NaSpTx-F1 and they will be described here.

Novel members of NaSpTx-F1

As for all previously described NaSpTx-F1 members, the seven new peptides were isolated from theraphosid (tarantula) venoms. Toxins were named based upon the rational nomenclature proposed for spider-venom peptides (King et al., 2008b), with the μ prefix denoting inhibition of Na_V channels. Although it is possible that some of these toxins might inhibit Ca_V rather than Na_V channels, this seems unlikely since all NaSpTx-F1 toxins characterized to date exclusively target Na_V channels (Klint et al., 2012).

Figure 2 outlines purification of the seven peptides, while their amino acid sequences are provided in Table 1. Each peptide comprises 32–35 residues with six cysteine residues that form three disulfide bonds based upon mass of the reduced and oxidized peptides. We did not experimentally determine their disulfide connectivity since, based upon sequence homology, it is presumably the same as previously characterized NaSpTx-F1 members that all contain an inhibitor cystine knot (ICK) with cysteine connectivity C1–C4, C2–C5, C3–C6 (Pallaghy et al., 1994). As shown in Table 1, there is a good agreement between the molecular mass of the native toxins determined using MALDI-TOF MS and the mass calculated from the amino acid sequence. Thus, none of the peptides are C-terminally amidated or contain any other post-translational modification aside from disulfide bonds. Despite the fact that all seven peptides belong to NaSpTx-F1, their sequence similarity is only moderate (38–62% sequence identity with respect to Hd1a; see Table 1). μ-TRTX-Ccy1a and μ-TRTX-Ccy1b are paralogues isolated from the same venom that differ only at position 6 (Ile6 in Ccy1a, Phe6 in Ccy1b).

Figure 2.

Isolation of novel Na_V1.7 inhibitors from spider venom. Each panel shows a chromatogram resulting from fractionation of crude venom using RP-HPLC; the acetonitrile gradient is indicated by the dotted line. Coloured fractions are those that putatively contain inhibitors of Na_V1.7. These fractions were further purified using ion exchange chromatography to obtain pure, active peptides. The top inset in each panel shows a photo of the spider from which venom was obtained. The bottom inset in each panel is a MALDI-TOF MS spectrum showing the M + H⁺ ion for the purified active peptide. These six panels show isolation of (A) μ-TRTX-Hd1a from the venom of Haplopelma doriae; (B) μ-TRTX-Se1a from the venom of Selenocosmia effera; (C) μ-TRTX-Osp1a from the venom of Orphnaecus species 1; (D) μ-TRTX-Osp1b from the venom of Orphnaecus species 2; (E) μ-TRTX-Ep1a from the venom of Euathlus pulcherrimaklaasi; (F) μ-TRTX-Ccy1a and μ-TRTX-Ccy1b from the venom of Chromatopelma cyaneopubescens.

Table 1.

Amino acid sequences of hNa_V1.7-active peptides discovered in this study

Venom	Toxin name	Sequence	Observed mass (Da)	Calculated mass (Da)	Identity (%)
Haplopelma doriae	μ-TRTX-Hd1a	-ACLGFGKSCNPSNDQCCKSSSLACSTKHKWCKYEL	3819.9	3819.7	100
Chromatopelma cyaneopubescens	μ-TRTX-Ccy1a	DDCLGIFKSCNPDNDKCCES--YKCSRRDKWCKYVL	4027.0	4027.8	62
Chromatopelma cyaneopubescens	μ-TRTX-Ccy1b	DDCLGFFKSCNPDNDKCCEN--YKCNRRDKWCKYVL	4115.6	4115.8	59
Orphnaecus species 11	μ-TRTX-Osp1a	-GCKGFGKACKYGADECCKN--LVCSKKHKWCKYTL	3692.5	3692.8	60
Orphnaecus species 21	μ-TRTX-Osp1b	-ECLGWMKGCEPKNNKCC--SSYVCTYKYPWCRYDL	3963.5	3963.7	56
Euathlus pulcherrimaklaasi	μ-TRTX-Ep1a	-DCLKFGWKCNPRNDKCC--SGLKCGSNHNWCKLHL	3800.8	3800.7	50
Selenocosmia effera	μ-TRTX-Se1a	-DCLGWMAGCDFNDNKCC--AGYVCK-KHPWCRYDL	3707.9	3707.5	38

Masses are monoisotopic masses, either determined experimentally using MALDI-TOF MS or calculated using PeptideMass (http://web.expasy.org/peptide_mass/). % identity was calculated relative to Hd1a^.

¹Orphnaecus species 1 and 2 were collected from Sibaliw and Maanghit Caves, respectively, on Panai Islands in the Philippines. We have tentatively assumed that these are the same species, and the toxins have been named accordingly.

The seven spider-venom peptides are similar to several previously described members of NaSpTx-F1. μ-TRTX-Cc1a/b, μ-TRTX-Se1a and μ-TRTX-Osp1b are most similar to κ-TRTX-Gr4a (82, 55 and 55% identity, respectively) and β-TRTX-Cm1a (76, 52 and 58% identity, respectively). κ-TRTX-Gr4a (VSTX3) binds to the voltage sensor domain of the archaebacterial voltage-gated potassium channel K_VAP (Ruta and MacKinnon, 2004) but it has not been tested against Na_V channels. β-TRTX-Cm1a (ceratotoxin-1) was isolated based upon its ability to inhibit mammalian Na_V channels; it potently inhibits Na_V1.1, Na_V1.2, Na_V1.4 and Na_V1.5 with IC₅₀ values ranging from 3 to 890 nM (Bosmans et al., 2006). μ-TRTX-Hd1a (Hd1a), μ-TRTX-Osp1a and μ-TRTX-Ep1a are most similar to μ-TRTX-Hhn1b (hainantoxin-IV, Hhn1b) (82, 62 and 51% identity, respectively), a potent inhibitor of TTX-sensitive Na_V currents in dorsal root ganglion (DRG) neurons (Liu et al., 2012).

Recombinant Hd1a preferentially inhibits hNa_V1.7

In order to validate the ability of the FLIPR assay to identify inhibitors of hNa_V1.7, we undertook detailed structure–function characterization of Hd1a. Hd1a was produced in the periplasm of E. coli to facilitate disulfide bond formation (Klint et al., 2013). The His₆-MBP-Hd1a fusion protein, which was the major cellular protein produced after IPTG induction (Figure 3A, inset), was purified using nickel affinity chromatography, then Hd1a was liberated from the fusion protein with TEV protease and purified using RP-HPLC (Figure 3A). Hd1a was purified to >98% homogeneity as judged by RP-HPLC (Figure 3B) and MS (Figure 3C). In order to facilitate TEV protease cleavage of the His₆-MBP-Hd1a fusion protein, Hd1a was produced with an additional N-terminal glycine residue, the preferred P1′ residue in the TEV protease recognition site (Kapust et al., 2002; Renicke et al., 2013); all remaining work was completed with this recombinant toxin (rHd1a).

Figure 3.

Expression and purification of recombinant Hd1a. (A) RP-HPLC chromatogram showing purification of recombinant Hd1a after cleavage from the His₆-MBP fusion tag by TEV protease. The peak corresponding to Hd1a is highlighted with an asterisk. Inset: SDS-PAGE gel showing Escherichia coli cells before (lane 1) and after (lane 2) induction of Hd1a expression with IPTG. Lane 3 contains molecular mass standards, with masses indicated in kDa on the right of the gel. The arrow indicates the running position of the MBP-toxin fusion protein. (B) RP-HPLC chromatogram of recombinant Hd1a after purification, showing a single uniform peak. (C) MALDI-TOF MS spectrum showing the M + H⁺ ion for purified recombinant Hd1a (observed, 3877.77 Da; calculated, 3877.71 Da).

Application of 1 μM rHd1a caused almost complete inhibition of hNa_V1.7-mediated currents recorded from oocytes (Figure 4A). Fitting the Hill equation to concentration–response data (Figure 4B) yielded an IC₅₀ of 111 ± 7 nM and a Hill coefficient of 0.95 ± 0.06, indicative of a single toxin binding site. Conductance–voltage (G-V) relationships obtained before and after addition of a non-saturating concentration of rHd1a (50 nM) revealed no significant shift in activation (Figure 4C), and inhibition was not voltage-dependent over the range of −25 to 0 mV (Supporting Information Fig. S1A). While this suggests a pore-blocking mechanism of action, additional experiments designed to determine the Hd1a binding site (described below) indicate this is not the case. Hd1a had no effect on steady-state inactivation (Figure 4D) or recovery from inactivation (Figure 4E), and it caused only a minor increase in inactivation time constants (Supporting Information Fig. S2). Inhibition by Hd1a is reversible as hNa_V1.7 currents recovered quickly following washout in toxin-free solution (Figure 4F).

Figure 4.

Effect of rHd1a on hNa_V1.7/β1. Currents were recorded from Xenopus oocytes using the two-electrode voltage-clamp technique. (A) Sodium currents before (black) and after (red) addition of 1 μM rHd1a. (B) Current–voltage relationship before (black) and after (red) addition of 50 nM rHd1a. Currents were evoked from a holding potential of −90 mV, stepping from −70 to −10 mV in 5 mV increments. (C) I/I₀ as a function of rHd1a concentration (n = 3–4; error bars represent SEM). Fitting the Hill equation to the data yielded an IC₅₀ of 111 ± 7 nM and Hill coefficient of 0.95 ± 0.06. (D) Steady-state fast inactivation before (black) and after (red) addition of 1 μM rHd1a. Currents were evoked from a holding potential of −90 mV stepping to −15 mV in 5 mV increments. (E) Recovery from fast inactivation before (black) and after (red) addition of 1 μM rHd1a. Currents were evoked from a holding potential of −90 mV by applying a −15 mV pulse followed by repolarization to−90 mV. A second pulse to −15 mV was applied after a period ranging from 0 to 50 ms, increasing in 1 ms increments. (F) hNa_V1.7 currents (I/I_max) following application of 1 μM rHd1a for 5 min followed by washout for 10 min. Depolarizing pulses were applied from −90 to −20 mV every 80 ms. (G) Selectivity of rHd1a. hNa_V1 currents (I/I₀) in the presence of 1 μM rHd1a (n = 3–4; error bars denote SEM). Orange and green bars denote TTX-sensitive and TTX-resistant hNa_V subtypes respectively. All currents were evoked by a depolarization to −20 mV from a holding potential of −90 mV.

The subtype selectivity of rHd1a was assessed by examining its ability to inhibit a complete panel of hNa_V channel subtypes expressed in oocytes, including the key off-target subtypes hNa_V1.4, hNa_V1.5 and hNa_V1.6 (Figure 4G). At a concentration of 1 μM, Hd1a was highly selective with no inhibition of hNa_V1.5 or hNa_V1.8, 23–31% inhibition of hNa_V1.3, hNa_V1.4 and hNa_V1.6, moderate inhibition of hNa_V1.2 (55%), and robust inhibition of hNa_V1.1 (87%) and hNa_V1.7 (87%) (Figure 4D). Moreover, this selectivity pattern was maintained over a large voltage range (from −25 to 0 mV; Supporting Information Fig. S1B). Hd1a was inactive against hERG (K_v11.1 channels; Supporting Information Fig. S3).

Hd1a interacts with domain II voltage sensor of hNa_V1.7

The Na_V channel binding site has been examined for only one member of NaSpTx-F1, namely huwentoxin-IV (μ-TRTX-Hh2a, Hh2a), which binds to the S3b-S4 helix-turn-helix motif in the domain II voltage sensor (Xiao et al., 2008). This so-called paddle motif (Alabi et al., 2007; Bosmans et al., 2008) is targeted by a variety of animal toxins. In order to determine whether rHd1a also binds to the paddle region, we transplanted the S3b-S4 paddle motif from each of the four voltage-sensor domains of hNa_V1.7 into the homotetrameric K_V2.1 channel (Figure 5). For domains (D) I, III and IV of hNa_V1.7, we generated functional chimeras using previously described paddle-motif boundaries (Bosmans et al., 2008; 2011) (Figure 5B and C). In the case of DII, we could only record ionic currents when four N-terminal residues were not transferred (Figure 5B and C). Nonetheless, the transferred region in all functional chimeras contained the crucial basic residues that contribute to gating charge movement in K_V channels (Aggarwal and MacKinnon, 1996; Seoh et al., 1996), suggesting that the four voltage-sensing domains in hNa_V1.7 contain paddle motifs that are involved in sensing membrane voltage changes.

Figure 5.

Sensitivity of hNa_V1.7/rK_V2.1 paddle chimeras to rHd1a. (A) Schematic of the hNa_V1.7 channel showing the voltage-sensor paddle motifs (coloured segments) that were transferred from hNa_V1.7 to rat K_V2.1 (rK_V2.1). The paddles from domains I, II, III and IV are coloured purple, red, cyan and green respectively. (B) Sequence alignment of the paddle region of rK_V2.1 with the paddle motifs (S3b-S4 region) from each domain of hNa_V1.7. For the domain II transplant, only the chimera highlighted with an asterisk was functional. (C) Families of potassium currents for each of the hNa_V1.7/rKv2.1 chimeras. Scale bars: 1 μA (ordinate axis) and 50 ms (abscissa). (D) Tail current voltage-activation relationships for each of the hNa_V1.7/rK_V2.1 chimeras (n = 5; error bars denote SEM). The holding voltage was −90 mV (−120 mV for DIII) and the tail voltage was −60 mV (or −90 mV for DIII). (E) Mean time constants (τ) from single exponential fits to channel activation (filled circles) and deactivation (open circles) plotted as a function of the voltage at which the current was recorded (n = 3–5; error bars denote SEM). (F) Potassium currents elicited by depolarizations to 70 mV. Currents are shown before (black) and in the presence (coloured traces) of 1 μM rHd1a. Colours denote the domain from which the hNa_V1.7 paddle motif was transferred: DI paddle (purple); DII paddle (red); DIII paddle (blue); DIV paddle (green). Scale bars: 50 ms (abscissa), 1 μA (DII, DIII) and 2 μA (DI, DIV, K_V2.1) (ordinate axis).

Examination of the G-V relationships for the hNa_V1.7/K_V2.1 chimeras revealed that each voltage-sensor paddle has a distinct effect on K_V2.1 gating; the midpoints of the G-V relations are ≥50, 44 ± 1, −16 ± 2 and 9 ± 1 mV for the DI, II, III and IV constructs respectively (Figure 5D). The voltage sensor in DIV plays a crucial role in fast inactivation, possibly due to its slower response to changes in membrane voltage compared with DI–III and its coupling to the inactivation gate (Sheets et al., 1999; Horn et al., 2000; Chanda and Bezanilla, 2002). The DIV paddle motifs within rNa_V1.2 and rNa_V1.4 delay channel opening when transplanted into K_V channels, suggesting that this region helps determine the slower kinetics of DIV voltage-sensor activation (Bosmans et al., 2008). To explore whether the DIV paddle motif serves a similar role in hNa_V1.7, we measured the kinetics of activation and deactivation of the four chimeras in response to membrane depolarization and repolarization respectively (Figure 5E). As observed for previously studied chimeric channels, the kinetics for the DIV hNa_V1.7/K_V2.1 chimera were slower compared with other hNa_V1.7/K_V2.1 chimeras, suggesting that the paddle motif in domain IV contributes to fast inactivation in hNa_V1.7.

We next examined the effect of rHd1a on K_V2.1 and the hNa_V1.7/K_V2.1 paddle chimeras. One micromolar rHd1a had no effect on wild-type K_V2.1 or the DI, DIII and DIV chimeras, but it strongly inhibited K⁺ currents mediated by the DII hNa_V1.7/K_V2.1 chimera (Figure 5F). We conclude that rHd1a inhibits hNa_V1.7 primarily by interacting with the S3b-S4 paddle in DII.

Since hNa_V1.9 has proven recalcitrant to heterologous expression, we also examined the effect of rHd1a on a panel of previously described hNa_V1.9/K_V2.1 paddle chimeras in which the S3b-S4 paddle motifs of hNa_V1.9 were transplanted into K_V2.1 (Bosmans et al., 2011). rHd1a had no effect on currents mediated by any of these chimeras (Supporting Information Fig. S4). Since the transplanted S3b-S4 paddle region encompasses the binding site for rHd1a on hNa_V1.7, we tentatively concluded that rHd1a has no effect on hNa_V1.9.

Solution structure of rHd1a

The bacterial expression system described earlier allowed production of uniformly ¹³C/¹⁵N-labelled Hd1a for structure determination using NMR. ¹H_N, ¹⁵N, ¹³C′, ¹³Cα, ¹³Cβ resonance assignments for Hd1a were obtained by analysing amide-proton strips in three-dimensional (3D) HNCACB, CBCA(CO)NH and HNCO spectra. Additional side chain chemical shifts were obtained from ¹⁵N/¹³C-NOESY spectra and a four-dimensional (4D) HCC(CO)NH-TOCSY experiment that provides side chain ¹H-¹³C connectivities (Mobli et al., 2010). An assigned two-dimensional (2D) ¹H-¹⁵N HSQC spectrum of rHd1a is shown in Supporting Information Fig. S5, and complete chemical shifts are available from BioMagResBank (accession number 19998). CYANA 3.0 was used to automatically assign NOESY spectra and calculate 200 structures from random starting conformations; the 20 structures with best stereochemical quality as judged by MolProbity (Chen et al., 2010) were selected to represent the structure of rHd1a (PDB ID 2MPQ; Figure 6).

Figure 6.

Ensemble of rHd1a structures. Stereoview of the ensemble of 20 rHd1a structures (PDB 2MPQ) overlaid over the backbone atoms of residues 2–35. β strands are shown in blue and the three disulfide bonds are highlighted in yellow. The four inter-cystine loops and the N- and C-termini are labelled.

Statistics highlighting the high precision and stereochemical quality of the ensemble of rHd1a structures are shown in Table 2. The ensemble is highly precise with backbone and heavy-atom root-mean-square deviation values of 0.05 ± 0.01 and 0.41 ± 0.04 Å, respectively, and it ranks as ‘high resolution’ based upon measures of precision and stereochemical quality (Kwan et al., 2011).

Table 2.

Structural statistics for the ensemble of Hd1a structures

Experimental restraints
Interproton distance restraints
Intra-residue	178
Sequential	168
Medium range (i–j < 5)	142
Long range (i–j ≥ 5)	214
Disulfide bond restraints	9
Dihedral angle restraints (ϕ, ψ)	57
Total number of restraints per residue	21.3
RMSD to mean coordinate structure (Å)
Backbone atoms	0.05 ± 0.01
All heavy atoms	0.41 ± 0.04
Stereochemical quality
Residues in most favoured Ramachandran region (%)	87.9 ± 0.9
Ramachandran outliers (%)	0.0 ± 0.0
Unfavourable side chain rotamers (%)	11.1 ± 2.2
Residues with bad angles	0.0 ± 0.0
Clashscore, all atoms	0.0 ± 0.0
Overall MolProbity score (percentile)	1.89 ± 0.07 (81 ± 3)

Root-mean-square deviation (RMSD) values were calculated over the well-defined regions of the structures (residues 2–35). Measures of stereochemical quality are from MolProbity (http://helix.research.duhs.duke.edu). Clashscore is the number of steric overlaps >0.4 Å per 103 atoms. All statistics are given as mean ± SD.

Figure 6 shows an overlay of the ensemble of 20 Hd1a structures. As expected, Hd1a contains an ICK motif in which the Cys17–Cys32 disulfide bond pierces a 12-residue ring formed by the other two disulfides (Cys3–Cys18, Cys10–Cys25) and the intervening sections of peptide backbone. Thus, Hd1a contains four inter-cystine loops (numbered 1–4 in Figure 6). The dominant secondary structure is a β hairpin comprising strands β1 (residues 23–26) and β2 (residues 31–34) connected by a short hairpin loop (residues 27–30; loop 4).

Discussion

High-throughput venom screens

In contrast to the diversity of Na_V channels in mammals, insects contain a single isoform. Interference with this channel has lethal consequences, making it an ideal target for insect predators such as spiders, scorpions and centipedes (King et al., 2008a). As a consequence, the venoms of these arthropods are rich in Na_V channel modulators (Rodriguez De La Vega and Possani, 2005; Gurevitz et al., 2007; King et al., 2008a; Yang et al., 2012; Gilchrist et al., 2014). Since there is no evolutionary selection pressure to prevent these toxins acting on vertebrates, many spider-venom peptides also modulate the activity of mammalian Na_V channels (Nicholson and Little, 2005; Escoubas and Bosmans, 2007). However, since the level of sequence identity between insect and human Na_V channels is only 55–60% (King et al., 2008a), it is difficult to predict the effect of these peptides on specific hNa_V subtypes. We therefore developed a high-throughput approach for screening spider-venom peptides directly against hNa_V1.7.

We took advantage of the observation that the neuroblastoma cell line SH-SY5Y expresses high levels of Na_V1.7 to the exclusion of most other subtypes (Vetter et al., 2012). This formed the basis of a high-throughput FLIPR assay in which venoms were monitored for their ability to prevent a calcium-dependent fluorescent response induced by activation of the endogenous hNa_V1.7 population with veratridine, a non-specific Na_V channel agonist. This assay enabled ∼90 venoms to be screened in 96-well plate format; if we conservatively estimate that each venom contains 200 peptides, then this corresponds to a rapid screen of 18 000 peptides. Scaling to 384-well plates would enable rapid screening of >350 venoms (>70 000 peptides) and minimize the amount of venom required.

Using this high-throughput FLIPR assay, we demonstrated that spider venoms are an incredibly rich source of hNa_V1.7 modulators with the potential to serve as leads for analgesic development. A remarkable 40% of all venoms screened putatively contain at least one inhibitor of hNa_V1.7. Moreover, deconvolution of ‘hits’ revealed that some spider venoms contain more than one hNa_V1.7 inhibitor (e.g. see Figure 2F).

High-throughput structure–function studies

Using our hNa_V1.7 screen, we discovered seven novel members of NaSpTx-F1 and focused on one of these (Hd1a) in order to develop robust methods for structure–function characterization. rHd1a potently inhibited hNa_V1.7 with an IC₅₀ of ∼110 nM, with modest or zero effect on all other subtypes except Na_V1.1. Since most spider-venom-derived Na_V1.7 modulators are gating modifiers that bind to one or more of the voltage sensors, we developed an approach that allows rapid determination of which voltage-sensor is involved in toxin binding. Competition binding experiments can be difficult to interpret because some spider-venom peptides interact with more than one voltage sensor (Bosmans et al., 2008); therefore, we developed an alternative approach in which a panel of channel chimeras was created by transplanting the voltage-sensor paddles from each of the four domains of hNa_V1.7 into rK_V2.1 (Figure 5). This enabled us to ask whether transplantation of the hNa_V1.7 paddles into rK_V2.1 sensitizes this K_V channel to the effects of a Na_V channel toxin. Using this approach, we demonstrated that rHd1a binds exclusively to the voltage-sensor paddle from hNa_V1.7 domain II (Figure 5F). Hd1a joins an extensive list of animal toxins that target this paddle motif to prevent Na_v channel opening (Cestele et al., 1998; Catterall et al., 2007; Bosmans et al., 2008; Bosmans and Swartz, 2010; Xiao et al., 2011).

High-throughput screens such as the one described here are increasingly being used in venom-based drug discovery pipelines (Vetter et al., 2011). This results in more lead molecules becoming available very early in the discovery pipeline than in traditional discovery programmes focussed on one or several venoms (Bosmans et al., 2006). Hence, there is a need to rapidly determine the structure of several venom peptides to allow timely development of structure–activity relationships (SAR). Although NMR is the dominant approach for determining the structure of polypeptides <10 kDa (Kwan et al., 2011), it suffers from low-throughput. Thus, we integrated several approaches to speed up venom-peptide structure determination. Firstly, we acquired all scalar-coupling experiments using NUS in order to expedite data acquisition and increase resolution in the indirect frequency dimensions. This enables a complete 3D/4D data set to be acquired in <5 days, and the increased digital resolution makes the spectra highly amenable to automated assignment. Secondly, we used automated NOESY assignment and structure calculation (Güntert, 2004). Together, these approaches enable high-resolution structures of excellent stereochemical quality to be determined in a highly automated fashion in 1 week, as exemplified by the structure of rHd1a reported here (Figure 6).

The structure of rHd1a is similar to that of Hhn1b and Hh2a (Figure 7A). A positively charged patch on the surface of Hhn1b is important for its inhibition of TTX-sensitive Na_V currents in DRG neurons, with mutation to alanine of residues K27, R29 and K32 resulting in >100-fold decrease in potency (Liu et al., 2012). These three cationic residues are conserved in Hd1a (corresponding residues are K28, K30 and K33; Figure 7B). Extensive mutagenesis of Hh2a, which is 68% identical to Hd1a, also revealed that these charged residues are important for hNa_V1.7 inhibition, with a K32A mutation causing almost complete loss of activity (Minassian et al., 2013; Revell et al., 2013). Hydrophobic interactions with the channel are also important, as a W30A mutation caused complete loss of hNa_V1.7 inhibition, whereas F6A and Y33A mutations triggered major reductions in potency (Minassian et al., 2013; Revell et al., 2013). Interestingly, W30 and K32 are two of the few residues that are completely conserved in NaSpTx-F1 (Klint et al., 2012). The mutational data are mapped onto the structure of rHd1a in Figure 7C. The functionally important hydrophobic and cationic residues form adjacent clusters that together create a contiguous surface epitope that likely represents the peptide's pharmacophore for hNa_V1.7.

Figure 7.

Structural comparison of rHd1a with characterized NaSpTx-F1 members. (A) Ribbon representation of the 3D structures of Hd1a (green, PDB 2MPQ), μ-TRTX-Hhn1b (cyan, PDB 1NIY) and μ-TRTX-Hh2a (pink, PDB 1MB6). The C-termini are labelled. (B) Alignment of the amino sequences of the three peptides. Red denotes residues that cause a reduction in inhibition of Na_V1.7 when mutated to alanine. Sequence numbering is with respect to μ-TRTX-Hh2a and μ-TRTX-Hhn1b. (C) Solvent accessible surface of Hd1a. Hydrophobic and basic residues are coloured green and blue, respectively, and are labelled. This view is related to that in panel (A) by a 180° rotation around the long axis of the molecule.

Most of the residues important for interaction of NaSpTx-F1 peptides with hNa_V1.7 are located in the structurally well-conserved C-terminal region and loop 4 (Figure 7A). There are subtle sequence variations in these regions (see Table 1), and detailed SAR studies will be required to unravel the molecular determinants that underlie differences in subtype selectivity within NaSpTx-F1. The recently described tethered-toxin screening approach, which allows alanine scans to be performed without the need to produce and purify individual mutants (Gui et al., 2014), should expedite such studies.

In summary, we developed a high-throughput screen for discovery of hNa_V1.7 modulators and used this screen to demonstrate that spider venoms are perhaps the best natural source of inhibitors of this target analgesic. We discovered seven novel members of NaSpTx-F1 using this screen. One of these toxins, rHd1a, potently inhibited hNa_V1.7 with a high level of selectivity over other hNa_V subtypes In addition, we engineered a panel of hNa_V1.7/rKv2.1 channel chimeras that can be used to rapidly assess whether hNa_V1.7 toxins target any of the channel's voltage-sensor paddles. The venom-based drug discovery pipeline described here provides a paradigm for high-throughput screening of animal venoms against other therapeutic ion channel targets.

Glossary

Abbreviations

Ca_V channel

voltage-gated calcium channel

FLIPR

fluorescence imaging plate reader

Hd1a

peptide μ-TRTX-Hd1a from venom of the spider Haplopelma doriae

hNa_V1.7

human voltage-gated sodium channel subtype 1.7

ICK

inhibitor cystine knot

K_V channel

voltage-gated potassium channel

MALDI-TOF

matrix-assisted laser desorption/ionization time-of-flight

MBP

maltose binding protein

NaSpTx

spider toxin that targets voltage-gated sodium channels

NaSpTx-F1

NaSpTx family 1

Na_V channel

voltage-gated sodium channel

RMSD

root-mean-square deviation

SAR

structure–activity relationship

TTX

tetrodotoxin

Author contributions

G. F. K., R. J. L. and F. B. directed this program of research. G. F. K., R. L., I. V., V. H., J. K. K. and F. B. conceived and designed the experiments. J. K. K., J. J. S., D. R., S. Y. R., S. S., I. V., M. M., V. H. and F. B. performed the experiments: J. K. K., J. J. S., D. R., S. Y. R., S. S., M. M., F. B. and G. F. K. analysed the data. J. K. K., F. B. and G. F. K. wrote the paper.

Conflict of interest

The authors have no conflicts of interest.

Supporting Information

Additional Supporting Information may be found in the online version of this article at the publisher's web-site: