The Tarantula Toxin ω-Avsp1a Specifically Inhibits Human CaV3.1 and CaV3.3 via the Extracellular S3-S4 Loop of the Domain 1 Voltage-Sensor

Inhibition of T-type calcium channels (CaV3) prevents development of diseases related to cardiovascular and nerve systems. Further, knockout animal studies have revealed that some diseases are mediated by specific subtypes of CaV3. However, subtype-specific CaV3 inhibitors for therapeutic purposes or for studying the physiological roles of CaV3 subtypes are missing. To bridge this gap, we employed our spider venom library and uncovered that Avicularia spec. (“Amazonas Purple”, Peru) tarantula venom inhibited specific T-type CaV channel subtypes. By using chromatographic and mass-spectrometric techniques, we isolated and sequenced the active toxin ω-Avsp1a, a C-terminally amidated 36 residue peptide with a molecular weight of 4224.91 Da, which comprised the major peak in the venom. Both native (4.1 μM) and synthetic ω-Avsp1a (10 μM) inhibited 90% of CaV3.1 and CaV3.3, but only 25% of CaV3.2 currents. In order to investigate the toxin binding site, we generated a range of chimeric channels from the less sensitive CaV3.2 and more sensitive CaV3.3. Our results suggest that domain-1 of CaV3.3 is important for the inhibitory effect of ω-Avsp1a on T-type calcium channels. Further studies revealed that a leucine of T-type calcium channels is crucial for the inhibitory effect of ω-Avsp1a.


Introduction
Dynamic changes in the intracellular calcium concentration ([Ca] i ) are used to regulate various biological processes [1], with abnormal homeostasis of [Ca] i leading to a diverse spectrum of diseases [2]. Both high-voltage-activated calcium channels (Ca V 1 and Ca V 2 subtypes) and low-voltage-activated calcium channels (Ca V 3 subtypes) are involved in regulation of [Ca] i in both excitable and non-excitable cells. Ca V 3 channels (also known as T-type channels) are expressed in a variety of tissues, including the heart, nervous system, smooth muscle, kidney, sperm, and endocrine organs. The differences in biophysical

Venoms
All arachnid venoms used for this study (for details see Supplementary Table S1) were isolated by light electrical stimulation as previously described [21], then lyophilised and stored at −20 • C prior to use.

Mass Spectrometry
All molecular masses were determined by using matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry (MALDI-TOF MS) using a Model 4700 Triple TOF mass spectrometer (Applied Biosystems, Foster City, CA, USA). Toxin samples were mixed 1:1 (v:v) with α-cyano-4-hydroxy-cinnamic acid matrix (7 mg/mL in 50/50 acetonitrile/H 2 O with 5% formic acid) and MALDI-TOF spectra were acquired in positive reflector mode. All reported molecular masses unless otherwise stated refer to the monoisotopic uncharged peptide.

N-Terminal Edman Sequencing
ω-Avsp1a was solubilised in 25 mM ammonium bicarbonate, reduced using dithiothreitol (25 mM) at 56 • C for 30 min, then alkylated using iodoacetamide (55 mM) at room temperature for 30 min. Fully reduced and alkylated Avsp1a was then purified via RP-HPLC using a Zorbax 300SB-C18 column (3 Å, 150 × 3 mm). The purified reduced/alkylated Avsp1a was then loaded onto a precycled Biobrene disc and N-terminal sequencing via Edman degradation was performed by the Australian Proteome Analysis Facility (Sydney, NSW, Australia) using an Applied Biosystems 494 Procise Protein Sequencer.

C-Terminal Sequencing Using Carboxypeptidase Y
The C-terminal sequence of native Avsp1a was determined by incubation with carboxypeptidase Y (CPY) according to the method previously described [22]. In short, 4.4 µg of native Avsp1a was digested with CPY (2 ng) for various time points (1,2,5,15,30,60,120, and 240 min), before the reaction was quenched by adding 1 µL of 1% formic acid. The resulting fragments were then analysed via MALDI-TOF MS as described in Section 2.3.
The purified reduced peptide was oxidatively folded in 0.33 M ammonium acetate, 2 M guanidine hydrochloride (pH 8.0), in the presence of 10-or 100-fold molar equivalents of oxidised and reduced glutathione respectively, stirring at 4 • C for 3 days. The single major product was isolated by preparative RP-HPLC, then co-eluted with native ω-Avsp1a on analytical RP-HPLC to confirm that the correct fold had been achieved.

Cell Culture and Transient Expression
Human embryonic kidney (HEK) 293 cells were grown in Dulbecco's modified Eagle's medium (DMEM) containing 5% fetal bovine serum and penicillin/streptomycin at 37 • C in a 5% CO 2 incubator. Cells were transiently transfected with 3 µg plasmid DNA using the calcium-phosphate method. Plasmids of Cav3.1, Cav3.2 or Cav3.3 were separately transfected into three individual plates of HEK-293 cells. After transfection, cells were incubated at 37 • C for 48 h prior to whole-cell voltage-clamp recordings.

Electrophysiology
Patch pipettes with a tip resistance of 2.8~3.5 MΩ were pulled from borosilicate glass capillary tubes (Warner Instruments, Holliston, MA, USA) using a P-97 Flaming-Brown type micropipette puller (Sutter Instrument, Novato, CA, USA). Ionic currents were recorded using an Axon Multiclamp 700B microelectrode amplifier (Molecular Devices, San Jose, CA, USA). The sampling frequency for acquisition was 50 kHz and data were low-pass filtered at 2 kHz. Voltage and current commands and digitisation of membrane voltages and currents were controlled using a Digidata 1440A interfaced with Clampex 10.4 (Molecular Devices, San Jose, CA, USA). Data were analysed using pCLAMP10.4 software (Molecular Devices, San Jose, CA, USA).
To measure Ca V 3 currents, cells were incubated with the following bath solution (in mM): 145 TEA-Cl, 5 CaCl 2 , 3 CsCl, 1 MgCl 2 , 5 glucose, and 10 HEPES, adjusted to pH 7.4 with TEA-OH, osmolarity~300 mOsm. The pipette solution was as follows (in mM): 130 CsCl, 20 HEPES, 10 EGTA, 5 MgCl 2 , 3 Mg-ATP, and 0.3 Tris-GTP, adjusted to pH 7.3 with CsOH, osmolarity~310 mOsm. All recordings were performed at room temperature (21-24 • C). For Cav3.1-or Cav3.2-transfected cells, their membrane potentials were initially held at −90 mV for 20 ms and then depolarized to −30 mV for 150 ms. For Cav3.3-transfected cells, the membrane potentials were initially held at −90 mV for 20 ms and then depolarized to −20 mV for 150 ms. Application of venom fractions or synthetic peptide was via a Picospritzer III dispenser (Parker Hannifin, Hollis, NH, USA). Puffed pressure and duration were 8-10 psi and 60 s, respectively. The peak values of Cav3 currents were recorded before and after addition of venom fractions or synthetic peptide. The toxin-reduced peak amplitudes represent the changes of peak values before or after addition of the toxin. The inhibition was calculated by dividing the toxin-reduced peak amplitudes by the peak amplitudes before additions of the toxin.

Construction and Mutagenesis of Plasmid cDNA
All plasmid constructs were made using a QuikChange site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA, USA). After mutagenesis, constructs were verified by sequencing. Chimeric constructs were combined by using the GeneArt Seamless Cloning and Assembly Enzyme Mix (Thermo Fisher Scientific, Waltham, MA, USA).

Determination of Three-Dimensional (3D) Structure of ω-Avsp1a by NMR
The 3D structure of ω-Avsp1a was deterg 2D NMR spectroscopy. Lyophilisedmined usin synthetic peptide was reconstituted to a final concentration of 1 mM in buffer containing 20 mM sodium phosphate at pH 6 and 5% D 2 O. All NMR spectra were recorded at 25 • C on a 600 MHz Bruker Advance II spectrometer (Billerica, MA, USA) equipped with a cryogenic probe. A combination of 2D spectra including natural abundance 1 H-15 N HSQC, 1 H-13 C HSQC, 1 H-1 H TOCSY and 1 H-1 H NOESY were acquired for resonance assignments. The spectra were processed using Topspin 3.2 (Bruker) and analysed using CcpNmr Analysis 2.4.1 [23]. Chemical shift assignments were deposited in BioMagResBank (BMRB) under the accession number 30872. The 2D 1 H-1 H NOESY acquired with a mixing time of 250 ms provided interproton distance restraints for the structure calculations. The spectrum was manually peak-picked and CYANA (v.3.97) was employed to automatically assign the peak list and extract distance restraints [24]. Dihedral-angle restraints were derived from chemical shifts using TALOS-N [25], with the restraint ranges for structure calculations set to twice the standard deviation. CYANA was used to calculate 200 structures based on 565 interproton distance restraints, 52 dihedral-angle restraints and 9 disulfide-bond restraints. The 30 structures with the lowest final target-function values were selected and were subjected to MolProbity analysis [26]. The 20 structures with the best stereochemical properties as judged by MolProbity were chosen to represent the structure of ω-Avsp1a.
The atomic coordinates of the final structure ensemble have been deposited in the Protein Data Bank (PDB) under accession code 7LVN. A summary of the restraints used in the structure calculation and the final structure statistics are presented in Table 1.

Toxin Isolation and Purification
An initial screen of a panel of arachnid venoms (Supplementary Table S1) revealed that crude venom (200 ng/µL) from the tarantula Avicularia spec. ("Amazonas Purple", Peru) specifically inhibited Ca V 3.1 and Ca V 3.3 subtypes. Avicularia spec. venom was then fractionated via RP-HPLC, with the largest peak (highlighted in red Figure 1A) reproducing the Ca V 3 inhibition seen with crude venom. Further purification of the active fraction by HILIC-HPLC ( Figure 1B) yielded a pure peptide with a molecular mass of 4224.91 Da.
N-terminal Edman sequencing of the purified active peptide yielded the sequence GDCHKFLGWCRGEPDPCCEHLSCSRKHGWC with the predicted molecular mass being 785.50 Da less than that observed for the native toxin. We therefore performed further C-terminal sequencing using digestion with CPY. These experiments revealed a C-terminal sequence of VWDWTV-NH 2 ( Figure 2). Together with the N-terminal sequence obtained by Edman (GDCHKFLGWCRGEPDPCCEHLSCSRKHGWC), the resulting complete sequence can be deduced as GDCHKFLGWCRGEPDPCCEHLSCSRKHGWCVWDWTV-NH 2 . This sequence has a calculated monoisotopic molecular mass of 4224.80 Da, only 0.11 Da less than the observed mass of the native peptide. We therefore conclude that this is the complete sequence of the peptide. Based on the rules for the rational nomenclature of venom peptides [27] and its activity on Ca V channels, we propose the following toxin name: ω-theraphotoxin-Avsp1a (or ω-TRTX-Avsp1a as the official abbreviation). Henceforth, for simplification, we refer to this toxin as ω-Avsp1a. N-terminal Edman sequencing of the purified active peptide yielded the sequence GDCHKFLGWCRGEPDPCCEHLSCSRKHGWC with the predicted molecular mass being 785.50 Da less than that observed for the native toxin. We therefore performed further C-terminal sequencing using digestion with CPY. These experiments revealed a C-terminal sequence of VWDWTV-NH2 ( Figure 2). Together with the N-terminal sequence obtained by Edman (GDCHKFLGWCRGEPDPCCEHLSCSRKHGWC), the resulting complete sequence can be deduced as GDCHKFLGWCRGEPDPC-CEHLSCSRKHGWCVWDWTV-NH2. This sequence has a calculated monoisotopic molecular mass of 4224.80 Da, only 0.11 Da less than the observed mass of the native peptide. We therefore conclude that this is the complete sequence of the peptide. Based on the rules for the rational nomenclature of venom peptides [27] and its activity on CaV channels, we propose the following toxin name: ω-theraphotoxin-Avsp1a (or ω-TRTX-Avsp1a as the official abbreviation). Henceforth, for simplification, we refer to this toxin as ω-Avsp1a.

Chemical Synthesis of ω-Avsp1a
ω-Avsp1a was synthesised using solid-phase peptide synthesis, and oxidatively folded to yield the correct disulfide-bond isomer as confirmed by co-elution with native ω-Avsp1a ( Figure 3).

Chemical Synthesis of ω-Avsp1a
ω-Avsp1a was synthesised using solid-phase peptide synthesis, and oxidatively folded to yield the correct disulfide-bond isomer as confirmed by co-elution with native ω-Avsp1a ( Figure 3).

Sub-Type Specific Inhibition of T-Type Calcium Channels by ω-Avsp1a
In order to determine the Ca V 3 selectivity of ω-Avsp1a, we tested native ω-Avsp1a isolated from A. spec. venom on Ca V 3.1, Ca V 3.2 and Ca V 3.3 expressed in HEK-293 cells.

Identification of T-Type Calcium Channel α1-Subunit Domains Involved in ω-Avsp1a Activity
We assume that the domain structure and amino acid sequence differences between the various Ca V 3 subtypes underlie their differential sensitivity to ω-Avsp1a. The poreforming α1-subunit of all Ca V channels, including Ca V 3.1, Ca V 3.2 and Ca V 3.3, is comprised of four homologous domains (D1-D4), with each domain containing six transmembrane segments denoted S1-S6 ( Figure 5A). In order to identify the amino acid(s) critical for channel inhibition caused by ω-Avsp1a, we generated chimeric channels from the less inhibited Ca V 3.2 and more affected Ca  Figure 5B). We found that the replacement of domains D1-D3 of Ca V 3.2 with the homologous domains of Ca V 3.3 increased the sensitivity of Ca V 3.2 to ω-Avsp1a to that of Ca V 3.3 ( Figure 5B). We further reduced the number of replaced domains to two (Ca V 3.3-D1-D2/Ca V 3.2-D3-D4) or one (Ca V 3.3-D1/Ca V 3.2-D2-D4) and found that replacing just D1 of Ca V 3.2 with the equivalent domain from Ca V 3.3 is sufficient to enhance the inhibitory effect of ω-Avsp1a on Ca V 3.2 ( Figure 5B). We conclude that domain D1 of Ca V 3.3 is critical for the inhibitory effect of ω-Avsp1a.
Biomedicines 2022, 10, x FOR PEER REVIEW 9 of 16 We assume that the domain structure and amino acid sequence differences between the various CaV3 subtypes underlie their differential sensitivity to ω-Avsp1a. The poreforming α1-subunit of all CaV channels, including CaV3.1, CaV3.2 and CaV3.3, is comprised of four homologous domains (D1-D4), with each domain containing six transmembrane segments denoted S1-S6 ( Figure 5A). In order to identify the amino acid(s) critical for channel inhibition caused by ω-Avsp1a, we generated chimeric channels from the less inhibited CaV3.2 and more affected CaV3  Figure 5B). We conclude that domain D1 of CaV3.3 is critical for the inhibitory effect of ω-Avsp1a.

Identification of ω-Avsp1a Binding Site on T-Type Calcium Channels
To narrow down the region within the D1 domain of Ca V 3.3 required for ω-Avsp1a inhibition, we generated additional chimeric channels. Each of the four homologous domains of Ca V channels contains six transmembrane helices (S1-S6) connected by two cytoplasmic and three extracellular loops (S1-S2, S3-S4 and S5-S6) as shown in Figure 5A. Replacement of D1-S1-S5 of Ca V 3.2 with the homologous S1-S5 helices of Ca V 3.3 D1 increased the inhibition induced by ω-Avsp1a from 25% to 71% (see Ca V 3.3-D1-S1-S5/Ca V 3.2 in Figure 5C). However, replacement of D1 S1-S3 of Ca V 3.2 with the equivalent region from Ca V 3.3 only increased the level of inhibition to 38% (see Ca V 3.3-D1-S1-S3/Ca V 3.2 of Figure 5C). These data suggest that the S3-S6 region of Ca V 3.3 is important for inhibition by ω-Avsp1a. Interestingly, it was previously shown that the D1 S3-S4 extracellular loop of Ca V 3.2 is important for sensing nickel and reducing agents [28,29]. We therefore replaced the D1 S3-S4 loop of Ca V 3.2 with the homologous extracellular loop from Ca V 3.3 and found that it is sufficient to make Ca V 3.2 as sensitive to ω-Avsp1a as Ca V 3.3 (see Ca V 3.3-D1-S3-S4/Ca V 3.2 in Figure 5C). These results suggest that amino acid residues in the D1 S3-S4 loop of Ca V 3.3 are critical for the inhibitory effect of ω-Avsp1a.
Next, we tried to pinpoint residues within the D1 S3-S4 loop involved in interaction with ω-Avsp1a. Alignment of the D1 S3-S4 loops of Ca V 3.1-3.3 revealed a high level of similarity, but notably a hydrophobic branched-chain leucine residue at position 169 in Ca V 3.3 and position 171 in Ca V .3.1 is replaced by a small glycine residue at the equivalent position in Ca V 3.2 ( Figure 6A). To examine the role of this leucine residue in the interaction with ω-Avsp1a, we mutated G190 in Ca V 3.2 to leucine and found that this single mutation markedly increased the toxin sensitivity of Ca V 3.2 ( Figure 6B,D). In contrast, when L171 in Ca V 3.1 or L169 in Ca V 3.3 was replaced by glycine, ω-Avsp1a failed to effectively inhibit the mutant channels ( Figure 6B,D). Using synthetic rather than native ω-Avsp1a recapitulated these effects ( Figure 6C,D). Conversely, the neighbouring glutamine in position 172 does not seem to be involved in toxin binding (Supplementary Figure S2). We therefore conclude that ω-Avsp1a differentially inhibits Ca V 3 isoforms by making a crucial interaction with a leucine residue in the extracellular D1 S3-S4 loop. Moreover, since ω-Avsp1a modulates the activity of Ca V 3.1/Ca V 3.3 by binding to the voltage sensor domain, it is clearly an allosteric modulator of channel activity rather than a pore blocker.

Quantification of the Potency of Synthetic ω-Avsp1a against Ca V 3.1 and Ca V 3.3
We examined the inhibitory potency of a range of ω-Avsp1a concentrations (10, 5, 1, and 0.1 µM) on Ca V 3.1 ( Figure 7A) and Ca V 3.3 ( Figure 7B) in order to quantify the toxin's potency. Here, only the synthetic ω-Avsp1a was used because of the difficulty of obtaining native toxin from the spider. The calculated IC 50 values for Ca V 3.1 and Ca V 3.3 were 5.06 ± 0.16 and 4.32 ± 1.21 µM, respectively. Because 10 µM ω-Avsp1a inhibited Ca V 3.2 by only 19%, we increased the concentration to 25 µM, but still only observed 31% inhibition. We were unable to further increase the working concentration of ω-Avsp1a because of limited solubility. Thus, the IC 50 for ω-Avsp1a inhibition Ca V 3.2 remains undetermined, but it is >25 µM.

Quantification of the Potency of Synthetic ω-Avsp1a against CaV3.1 and CaV3.3
We examined the inhibitory potency of a range of ω-Avsp1a concentrations (1 and 0.1 μM) on CaV3.1 ( Figure 7A) and CaV3.3 ( Figure 7B) in order to quantify the t potency. Here, only the synthetic ω-Avsp1a was used because of the difficulty of o ing native toxin from the spider. The calculated IC50 values for CaV3.1 and CaV3.3 5.06 ± 0.16 and 4.32 ± 1.21 μM, respectively. Because 10 μM ω-Avsp1a inhibited CaV only 19%, we increased the concentration to 25 μM, but still only observed 31% inhi We were unable to further increase the working concentration of ω-Avsp1a beca limited solubility. Thus, the IC50 for ω-Avsp1a inhibition CaV3.2 remains undeterm but it is >25 μM.

The 3D Structure of ω-Avsp1a
The solution structure of ω-Avsp1a was determined using NMR spectroscop obtained chemical shift assignments for 99.5% of the proton resonances, and the str was calculated based on 565 interproton-distance restraints, 52 dihedral-angle res and 9 disulfide-bond restraints. The final ensemble of 20 structures is highly precise a root-mean-square deviation of 0.16 ± 0.06 Å over the backbone atoms of residues which excludes the flexible N-and C-terminal regions. The peptide conforms to the itor cystine knot (ICK) motif which is commonly found in spider-venom peptides [ This motif is recognisable by three disulfide bonds with C1-C4, C2-C5, C3-C6 conn ity [32] that form a "pseudo-knot" architecture in which two of the disulfide bond the intervening sections of the peptide backbone form a ring that is pierced by the disulfide bond (Figure 8). In addition to the core cystine knot, the ICK motif is char ised by an antiparallel β sheet which in ω-Avsp1a is formed by residues L21-S24 (β W29-W32 (β2). Interestingly, there is a highly ordered network of hydrophobic re adjacent to β2 which includes all of the aromatic (F6, W9, W29, W32, W34) and le (L7, L21) residues in the peptide. These hydrophobic residues stack like the rung ladder on one face of the peptide, whereas the charged residues, in contrast, are evenly distributed over the surface of the peptide.

The 3D Structure of ω-Avsp1a
The solution structure of ω-Avsp1a was determined using NMR spectroscopy. We obtained chemical shift assignments for 99.5% of the proton resonances, and the structure was calculated based on 565 interproton-distance restraints, 52 dihedral-angle restraints and 9 disulfide-bond restraints. The final ensemble of 20 structures is highly precise, with a root-mean-square deviation of 0.16 ± 0.06 Å over the backbone atoms of residues 3-34, which excludes the flexible N-and C-terminal regions. The peptide conforms to the inhibitor cystine knot (ICK) motif which is commonly found in spider-venom peptides [30,31]. This motif is recognisable by three disulfide bonds with C1-C4, C2-C5, C3-C6 connectivity [32] that form a "pseudo-knot" architecture in which two of the disulfide bonds and the intervening sections of the peptide backbone form a ring that is pierced by the third disulfide bond (Figure 8). In addition to the core cystine knot, the ICK motif is characterised by an antiparallel β sheet which in ω-Avsp1a is formed by residues L21-S24 (β1) and W29-W32 (β2). Interestingly, there is a highly ordered network of hydrophobic residues adjacent to β2 which includes all of the aromatic (F6, W9, W29, W32, W34) and leucine (L7, L21) residues in the peptide. These hydrophobic residues stack like the rungs of a ladder on one face of the peptide, whereas the charged residues, in contrast, are more evenly distributed over the surface of the peptide. Biomedicines 2022, 10, x FOR PEER REVIEW 13 of 16

Discussion
ω-Avsp1a, the dominant component of Avicularia spec. ("Amazonas Purple", Peru) venom, was initially isolated from the venom based on its inhibition of CaV3. Edman sequencing revealed only the N-terminal 30 residues of the peptide, while the remaining six C-terminal residues were determined via MALDI-TOF MS following digestion of the peptide with CPY. Together, the Edman sequencing and MS data suggested the following amino acid sequence for ω-Avsp1a: GDCHKFLGWCRGEPDPC-CEHLSCSRKHGWCVWDWTV-NH2. Given that (i) synthetic ω-Avsp1a corresponding to this sequence co-elutes with native ω-Avsp1a on RP-HPLC; (ii) the molecular mass predicted from this sequence matches that of native ω-Avsp1a; and (iii) the activity of synthetic ω-Avsp1a on the various CaV3 subtypes recapitulates that of the native peptide, we are confident that the sequence we determined for ω-Avsp1a is correct.
NMR analysis revealed that the structure of ω-Avsp1a conforms to the ICK motif, which is commonly found in spider-venom peptides. The ICK architecture endows peptides with resistance to proteases, harsh chemicals, and high temperatures [33], which generally provides them with high levels of stability in human serum [34][35][36]. This property makes this class of peptide an attractive source of peptide-drug leads. The structure of ω-Avsp1a revealed a narrow hydrophobic strip composed of a stack of aromatic and leucine residues. Such hydrophobic patches are commonly found in spider toxins that

Discussion
ω-Avsp1a, the dominant component of Avicularia spec. ("Amazonas Purple", Peru) venom, was initially isolated from the venom based on its inhibition of Ca V 3. Edman sequencing revealed only the N-terminal 30 residues of the peptide, while the remaining six C-terminal residues were determined via MALDI-TOF MS following digestion of the peptide with CPY. Together, the Edman sequencing and MS data suggested the following amino acid sequence for ω-Avsp1a: GDCHKFLGWCRGEPDPCCEHLSCSRKHGWCVWDWTV-NH 2 . Given that (i) synthetic ω-Avsp1a corresponding to this sequence co-elutes with native ω-Avsp1a on RP-HPLC; (ii) the molecular mass predicted from this sequence matches that of native ω-Avsp1a; and (iii) the activity of synthetic ω-Avsp1a on the various Ca V 3 subtypes recapitulates that of the native peptide, we are confident that the sequence we determined for ω-Avsp1a is correct.
NMR analysis revealed that the structure of ω-Avsp1a conforms to the ICK motif, which is commonly found in spider-venom peptides. The ICK architecture endows peptides with resistance to proteases, harsh chemicals, and high temperatures [33], which generally provides them with high levels of stability in human serum [34][35][36]. This property makes this class of peptide an attractive source of peptide-drug leads. The structure of ω-Avsp1a revealed a narrow hydrophobic strip composed of a stack of aromatic and leucine residues. Such hydrophobic patches are commonly found in spider toxins that modulate ion channel activities via interaction with one or more of the voltage-sensing domains of the channel.
These hydrophobic patches have been proposed to facilitate interaction of the toxin with the plasma membrane so that it can diffuse laterally within the lipid bilayer to interact with the membrane-embedded voltage-sensor domain [37,38].
It is well documented that the three Ca V 3 subtypes participate differentially in a variety of physiological and pathological conditions. Selective pharmacological inhibition is therefore important for dissecting the pathophysiological functions of Ca V 3 channel subtypes and for the development of therapeutically useful subtype-selective inhibitors. Ca V 3 inhibitors such as ethosuximide and zonisamide are used clinically as anti-epileptic drugs, but they are not subtype-specific and not effective in all patients [40]. Preclinical data suggest that Ca V 3.2 might be a viable analgesic target, but none of the Ca V 3 blockers that have advanced to human clinical trials have been successful [41]. Thus, there is urgent need to identify new subtype-selective Ca V 3 inhibitors. In this study, we showed that ω-Avsp1a has 10-fold or higher selectivity for Ca V 3.1/Ca V 3.3 over Ca V 3.2, suggesting that it might be a useful pharmacological tool for dissecting out the role of Ca V 3 channels in various human pathophysiologies and for developing selective inhibitors of Ca V 3.1/Ca V 3.3.
Our chimeric channel experiments clearly demonstrate that ω-Avsp1a interacts with the extracellular S3-S4 loop of Ca V 3 and that Leu171 in this loop is essential for its inhibitory activity. Replacement of this Leu residue by Gly, the amino acid in the homologous position of Ca V 3.2, massively diminished the inhibitory activity of the peptide. The structure of ω-Avsp1a determined using NMR spectroscopy revealed a narrow hydrophobic strip, which is likely to be the region of the toxin that mediates its interaction with Leu171 in Ca V 3.1 and Leu169 in Ca V 3.3. In the Cryo-EM structure of Ca V 3.1 [42], an α-helix formed by the sequence around Leu171 of Ca V 3.1 is located in the most extracellular portion of the channel, providing a possible site for interaction with ω-Avsp1a. Our data suggest that targeting the D1 S3-S4 loop of Ca V 3 channels might be a good strategy for developing subtype-selective inhibitors of these channels.

Conclusions
A bioassay-guided fractionation strategy was used to isolate peptide ω-Avsp1a from venom of the Peruvian tarantula Avicularia spec. ("Amazonas Purple"). ω-Avsp1a was found to be a selective inhibitor of Ca V 3.1 and Ca V 3.3, with at least five-fold weaker activity on Ca V 3.2. By engineering chimeric Ca V channel constructs, we narrowed the ω-Avsp1a binding site on Ca V 3.3 to the D1 S3-S4 extracellular loop and showed that Leu171 in this loop is crucial for the toxin-channel interaction. Despite the low potency of ω-Avsp1a, its selectivity for Ca V 3.1/Ca V 3.3 over Ca V 3.2 makes it a useful pharmacological tool for dissecting out the role of these different Ca V 3 subtypes in disorders such as epilepsy or pain.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/biomedicines10051066/s1. Table S1: Crude arachnid venoms screened by patch-clamp electrophysiology for inhibitory activity (indicated in percent inhibition) against three Ca V 3 subtypes. n.d. = not determined. The venom of the Avicularia spec. that was used for the isolation of ω-Avsp1a is highlighted in red. Figure S1: Screening for Ca V 3 channel subtype specific inhibitors from arachnid venoms. Figure S2: Q172 of Ca V 3.1 is not important for inhibition function of ω-Avsp1a. Ca V 3.1/Q172 was mutated to histidine as the homologous residue in Ca V 3.2. There is no statistically significant difference between Ca V 3.1 and Ca V 3.1/Q172H mutant (Ca V 3.1 vs Ca V 3.1/Q172H, P = 0.70 according to two-tailed T-test). n = 5, 3, 5, 3, for Ca V 3.1, Ca V 3.1/Q172H, Ca V 3.2, Ca V 3.2/H191Q, respectively.