Two for the Price of One: Heterobivalent Ligand Design Targeting Two Binding Sites on Voltage-Gated Sodium Channels Slows Ligand Dissociation and Enhances Potency

Voltage-gated sodium (NaV) channels are pore-forming transmembrane proteins that play essential roles in excitable cells, and they are key targets for antiepileptic, antiarrhythmic, and analgesic drugs. We implemented a heterobivalent design strategy to modulate the potency, selectivity, and binding kinetics of NaV channel ligands. We conjugated μ-conotoxin KIIIA, which occludes the pore of the NaV channels, to an analogue of huwentoxin-IV, a spider-venom peptide that allosterically modulates channel gating. Bioorthogonal hydrazide and copper-assisted azide–alkyne cycloaddition conjugation chemistries were employed to generate heterobivalent ligands using polyethylene glycol linkers spanning 40–120 Å. The ligand with an 80 Å linker had the most pronounced bivalent effects, with a significantly slower dissociation rate and 4–24-fold higher potency compared to those of the monovalent peptides for the human NaV1.4 channel. This study highlights the power of heterobivalent ligand design and expands the repertoire of pharmacological probes for exploring the function of NaV channels.


■ INTRODUCTION
Voltage-gated sodium (Na V ) channels are fundamental for the generation and propagation of action potentials in excitable cells, and they are important therapeutic targets for antiepileptic, antiarrhythmic, and analgesic drugs. 1−3 Humans have nine Na V channel subtypes denoted Na V 1.1−Na V 1.9. Na V 1.1−Na V 1.3 and Na V 1.6 are expressed in both the central nervous system (CNS) and the peripheral nervous system (PNS), while Na V 1.7−Na V 1.9 are found primarily in peripheral sensory neurons. 2 Na V 1.4 and Na V 1.5 are predominantly located in skeletal and cardiac muscles, respectively, where they play critical roles in muscle contraction. 2 Na V channels are large transmembrane proteins composed of a pore-forming α-subunit in complex with one or two auxiliary β-subunits that modulate their expression, localization, gating, kinetics, and pharmacology ( Figure 1). 3,4 The α-subunit (∼260 kDa) folds into four homologous but nonidentical domains (denoted D I −D IV ) joined by intracellular linkers, with each domain containing six transmembrane segments (S1−S6). The S1−S4 segments within each domain form a voltage-sensing domain (VSD), while the S5 and S6 segments from each domain come together in a circular fashion to form the central pore of the channel ( Figure  1A). 1,2,4 The VSDs allow the channel to respond to changes in the membrane electrical potential, causing it to cycle (or "gate") among three distinct states: a closed/resting state in which the channel can be activated by membrane depolariza-tion, an open ion-conducting state, and a nonconducting inactivated state. 1,2,4 Although Na V channels are important drug targets, their therapeutic potential is far from fulfilled. Many venom peptides from arachnids, 4,5 cone snails, 6−8 sea anemones, 9 and other venomous animals target Na V channels with high potency and selectivity, and consequently have attracted interest both as pharmacological tools and as lead compounds for new analgesic, antiepileptic, and antiarrhythmic drugs. 10−14 These peptides can be divided into two broad classes based on their mechanism of action: (i) pore blockers that bind to the outer vestibule of the channel, thereby sterically preventing the entry of Na + into the channel pore, and (ii) allosteric modulators known as "gating modifiers" that interact with one or more of the VSDs and alter the gating and kinetics of the channel. 15 A new era in Na V channel research began with the determination of the first three-dimensional structures of vertebrate Na V channels, namely, Na V 1.4 from both electric eel 16 and humans, 17 and human Na V 1.7. 18 The muscle-specific Na V 1.4 channel has been the subject of extensive functional and mechanistic studies, and mutations in this channel have been linked with muscle channelopathies such as paramyotonia congenita and hyperkalemic periodic paralysis. 19,20 Na V 1.7 is of particular interest as a potential analgesic target because of its strong genetic association with pain. Loss-of-function mutations in the gene encoding Na V 1.7 lead to a congenital insensitivity to pain, whereas gain-of-function mutations underlie disorders such as erythromelalgia and paroxysmal extreme pain disorder that are characterized by severe episodic pain. 21,22 As part of our ongoing attempts to develop new pharmacological probes and therapeutic leads for human Na V channels, 10,23−26 we devised and tested in this study a bivalent linker design with a focus on Na V 1.4 and Na V 1.7 due to their (patho)physiological relevance and experimentally determined structures. Our strategy was to covalently link a pore blocker toxin with a gating modifier toxin using variable-length polyethylene glycol (PEG) linkers to simultaneously target two binding sites of the channel, thereby potentially enhancing binding kinetics, potency, and subtype selectivity ( Figure  1B,C). We show that joining monovalent ligands with an optimal-length PEG linker leads to a bivalent ligand with significantly enhanced potency at Na V 1.4 due to a greatly reduced rate of dissociation from the channel.
■ RESULTS Bivalent Ligand Design. For the pore blocker, we chose μ-conotoxin KIIIA (hereafter μ-KIIIA), a peptide isolated from venom of the marine cone snail Conus kinoshitai, with well- Figure 1. Na V channel architecture and overview of the bivalent inhibitor strategy. (A) Topology of Na V channel αand β-subunits. The α-subunit comprises four domains (denoted I−IV), with each domain containing six transmembrane segments (S1−S6). Segments S1−S4 in each domain form a voltage-sensing domain (VSD, gray), while S5, S6, and the membrane-penetrant pore loops (P-loops) form the pore domain (white). 1,4 (B) Schematic of the bivalent ligand strategy. Initial binding of either a gating modifier peptide (green) or a pore-blocking peptide (magenta) should bring the other peptide close to the channel, thereby enhancing binding kinetics (red arrows) and potency compared to those of monovalent ligands. The dotted line illustrates the spatial limit of the local concentration effect of the conjugated gating modifier when the pore blocker is bound. (C) Cryo-electron microscopy structure of hNa V 1.7-β1 in the presence of HwTx-IV and μ-KIIIA. The hNa V 1.7-β1 structure was used to determine the distance between the two peptides as this channel is our target of interest and because this structure contains HwTx-IV. A triplemutant variant of HwTx-IV (E1G, E4G, Y33W; m 3 -HwTx-IV) was placed in the HwTx-IV density in a random orientation due to the unknown interaction sites with the channel. The distance between the center of the m 3 -HwTx-IV density and the N-terminus of μ-KIIIA is ∼50 Å in a direct line (dotted line) (i.e., if steric overlap is ignored) and ∼80 Å considering the length of a half-circle (solid line) that comfortably avoids steric overlap with the channel (PDB entries 5T3M, 6J8E, and 6J8G and EMD entry 9781). 18 established pharmacology at Na V 1.4 and Na V 1.7, and extensive structure−activity relationship (SAR) information. 28 μ-KIIIA is a 16-residue peptide with an α-helical core stabilized by three disulfide bonds with Cys I −Cys V , Cys II −Cys IV , and Cys III − Cys VI connectivity ( Figure 2). 29 It preferentially blocks rat (r) Na V 1.2 (IC 50 = 5 nM) and rNa V 1.4 (IC 50 = 48 nM) over rNa V 1.7 (IC 50 = 147 nM). 30,31 Cryo-electron microscopy (cryo-EM) studies of μ-KIIIA bound to hNa V 1.2 27 in combination with SAR studies 8,28 revealed that residues K7, W8, R10, D11, and R14 are functionally important for the pore blocking of Na V channels, while the N-terminus can be modified without abrogating binding. Thus, we added an εazido-L-lysine to the N-terminus of μ-KIIIA (AzK-KIIIA) to make it suitable for bioorthogonal copper-catalyzed azide− alkyne cycloaddition (CuAAC) chemistry, 32 while retaining the free N-terminal α-amino group as it might affect the peptide's binding kinetics. 31 For the gating modifier, we chose an optimized analogue of μ-theraphotoxin-Hs2a [HwTx-IV; optimized analogue m 3 -HwTx-IV ( Figure 2)] originally identified in the venom of the tarantula Cyriopagopus schmidti (formerly Haplopelma schmidti). 33 m 3 -HwTx-IV has three mutations relative to the native toxin (E1G, E4G, Y33W), which makes it an exceptionally potent inhibitor of human (h) Na V 1.7 (IC 50 = 0.4 nM). 34 m 3 -HwTx-IV is a 35-residue peptide containing an inhibitor cystine knot (ICK) motif 35 in which a doublestranded antiparallel β-sheet is stabilized by three disulfide bonds with Cys I −Cys IV , Cys II −Cys V , and Cys III −Cys VI connectivity, with a three-dimensional (3D) structure highly similar to that of native HwTx-IV ( Figure 2). 33,36 m 3 -HwTx-IV also inhibits Na V 1.1−Na V 1.3 and Na V 1.6 with low nanomolar potency and is a moderately potent inhibitor of hNa V 1.4 (IC 50 = 370 nM). 36 Mutational and cryo-EM structural studies show that HwTx-IV binds to the D II VSD domain of Na V 1.7. 18 Residues W30 and K32 are critical for its activity, and while the N-terminus can be extended with polar or nonpolar residues without a loss of potency, the C-terminal amide is essential for potent inhibition of Na V 1.7. 34 On the basis of this information, we introduced an N-terminal serine residue (S-m 3 -HwTx-IV) that can be selectively converted into an aldehyde, thereby making it suitable for bioorthogonal hydrazone ligation. 37 We selected linker lengths for toxin conjugation that would allow simultaneous binding of the two peptides to their respective Na V channel binding sites based on the cryo-EM structures of hNa V 1.7 in complex with HwTx-IV 18,38 and hNa V 1.2 bound to μ-KIIIA 27 ( Figure S1). Ideally, binding of either peptide to its binding site should bring the second peptide into the proximity of its binding site, resulting in enhanced potency and altered binding kinetics and subtype selectivity. The distance between the two binding sites was estimated to be ∼50 Å in a direct line and ∼80 Å considering a half-circle ( Figure 1C). We thus decided on a systematic series of PEG linkers ranging in length from 40 to 120 Å. We included a shorter 40 Å linker, two linkers spanning the distance predicted from the cryo-EM structures (60 and 80 Å), and a longer linker of 120 Å. The linker lengths of 40−120 Å are approximate values determined using Avogadro software. 39 We hypothesized that the shorter 40 Å linker would not provide any bivalent effects as it does not span the two toxin binding sites, that the longer 120 Å linker might provide less optimal bivalent binding effects due to being too dynamic, and that the 60 and 80 Å linkers should yield pronounced and observable bivalent binding effects because they are within the optimal length to span the two toxin binding sites. 40 Heterobifunctionalized PEG linkers were designed to tether the two toxins together using a hydrazide function compatible with bioorthogonal hydrazone ligation and an alkyne function compatible with bioorthogonal CuAAC chemistry. 32,37,41 PEG is a nontoxic amphiphilic polymer that is monodisperse at the lengths employed here. PEG has good aqueous solubility and has been successfully used as a linker in many applications. 42−45 Synthesis, Folding, and Bioactivity of Unconjugated Pore Blocker and Gating Modifier Peptides. μ-KIIIA and AzK-KIIIA were assembled using manual 9-fluorenylmethyloxycarbonyl (Fmoc) solid-phase peptide synthesis (SPPS), 46 followed by oxidative folding. This yielded two distinct isomers with identical masses for both μ-KIIIA (observed monoisotopic mass, 1882.62 Da; calculated, 1882.64 Da), as reported previously, 29 and AzK-KIIIA (observed monoisotopic mass, 2036.70 Da; calculated, 2036.71 Da) ( Figure S2). The ability of each analogue to inhibit hNa V 1.7 was determined by whole-cell patch-clamp electrophysiology; the major isomer from oxidative folding of AzK-KIIIA potently inhibited the channel (IC 50 = 96 nM), whereas the minor isomer did not (IC 50 = 934 nM). We therefore selected the major isomer for bioorthogonal conjugation (Table S1). S-m 3 -HwTx-IV was assembled using automated microwave-assisted Fmoc-SPPS, after which oxidative folding yielded a single isomer (observed monoisotopic mass, 4070.27 Da; calculated, 4069.91 Da) ( Figure S3).
We compared the inhibitory potency of AzK-KIIIA and Sm 3 -HwTx-IV on both hNa V 1.4 and hNa V 1.7 to assess the impact of the modifications (Table 1). μ-KIIIA inhibited hNa V 1.7 with a potency [IC 50 = 132 ± 37 nM ( Table 1)] similar to what was previously reported for inhibition of rNa v 1.7 (IC 50 = 147 nM). 31 Addition of the AzK residue to μ-KIIIA slightly enhanced the potency against both channels (from 132 ± 37 to 96 ± 41 nM on hNa V 1.7 and from 48 ± 6 to 32 ± 10 nM on hNa V 1.4), confirming that bioactivity was HwTx-IV 34 17 n is the number of cells, with each cell considered an independent experiment. b IC 50 values were determined on rat Na V channel. c The IC 50 value was determined for the peptide with no C-terminal amidation. Legend: n.a., not available; ↑, fold increase in IC 50 relative to ligand 1; ↓, fold decrease in IC 50 relative to ligand 1.  47 Unreacted sites were capped with methanol. Repeated couplings of Fmoc-protected PEG 4 (2× PEG 4 for PEG40, 3× PEG 4 for PEG60, 4× PEG 4 for PEG80, and 6× PEG 4 for PEG120) were carried out using standard Fmoc-SPPS protocols. 46 Fmoc-Lpropargylglycine was used as the final amino acid to incorporate the alkyne moiety. The PEG linkers were cleaved with TFA and purified using RP-HPLC. The linker is illustrated as PEG[Å], where Å indicates the linker length estimated using Avogadro software. 39 retained upon N-terminal modification. Addition of the Nterminal serine residue to m 3 -HwTx-IV was also well tolerated; this change improved potency at hNa V 1.4 (IC 50 decreased from 369 ± 196 to 212 ± 20 nM) but reduced potency at hNa V 1.7, although it still exhibited excellent potency on this subtype (IC 50 increased from 0.4 ± 0.1 to 4 ± 0.3 nM). Linker Synthesis and Heterobivalent Ligand Assembly. Four PEG linkers ranging in length from 40 to 120 Å (PEG40/60/80/120) with N-terminal alkyne and C-terminal hydrazide functionalities were synthesized manually on solid support (Scheme 1 and Figures S4 and S5).
To conjugate these linkers to the peptides, the N-terminal Ser in m 3 -HwTx IV was first oxidized with sodium periodate (1.5 equiv) in sodium phosphate buffer (10 mM, pH 7.0) for 2 min at 25°C. The individual PEG linkers were then ligated to the N-terminal aldehyde of m 3 -HwTx-IV in sodium citrate buffer (100 mM, pH 4.5) for 24 h at −20°C. 37 Under these low-temperature conditions, slow-growing ice crystals produce locally high concentrations of reactants, which favors hydrazone bond formation. 48 AzK-KIIIA was then conjugated to the alkyne moiety of the linker via CuAAC chemistry 32 with a 70/30 (v/v) H 2 O/ t BuOH mixture, copper sulfate (1.4

Scheme 2. Bivalent Ligand Assembly Strategy for [m3-HwTx-IV]-[PEG[Å]]-[K-KIIIA] Constructs a a
Selective oxidation of the N-terminal serine of S-m 3 -HwTx-IV to an aldehyde via sodium periodate treatment, followed by hydrazone ligation (colored blue). Final conjugation of AzK-KIIIA to the alkyne moiety of the linkers via copper-catalyzed azide−alkyne cycloaddition to form a triazole-linked conjugate (colored red). KIIIA is colored magenta, and m 3 -HwTx-IV is colored green. Disulfide bonds are shown as yellow sticks. Journal of Medicinal Chemistry pubs.acs.org/jmc Article equiv), and ascorbic acid (5 equiv) for 1 h at 25°C, yielding a triazole linkage (Scheme 2 and Figure S6). Structural Ligand Integrity of Bivalent Constructs and Their Precursors. One-dimensional (1D) 1 H nuclear magnetic resonance (NMR) spectra were recorded to examine the structural integrity of AzK-KIIIA, S-m 3 -HwTx-IV, and the PEG-linked conjugates. Secondary Hα chemical shifts of AzK-KIIIA aligned well with published values for μ-KIIIA, 29 except near the N-terminus where the AzK residue was added ( Figure  3A). The negative secondary Hα shifts for residues 8−13 of AzK-KIIIA confirmed the presence of an α-helix in this region that is part of the toxin pharmacophore. 27,28 The fingerprint regions of the 1D 1 H NMR spectra of the PEG conjugate [m 3 -HwTx-IV]-[PEG80]-[K-KIIIA] overlapped well with the corresponding spectra of the S-m 3 -HwTx-IV and AzK-KIIIA precursors, indicating that the individual toxins retained their disulfide-stabilized 3D structures after PEG ligation ( Figure  3B).
Inhibition of hNa V 1.4 and hNa V 1.7 by Bivalent Ligands. We compared the inhibitory potency of the bivalent ligands and the monovalent precursors (individually and as equimolar mix) at hNa V 1.4 and hNa V 1.7 using patch-clamp electrophysiology to reveal any observable bivalent effects in terms of potency and selectivity ( Figure 4 and Table 1).
At hNa V 1. We also studied the potency impact of the PEG80 linker when attached to AzK-KIIIA or S-m 3 -HwTx-IV to exclude the possibility of the linker being responsible for the observed effects. Linker attachment caused a 4.5-fold decrease in inhibitory potency on Na V 1.   Figure 4A and Table 1). This observation prompted us to examine the binding kinetics of the monovalent and bivalent ligands at hNa V 1.4 and hNa V 1.7. The bivalent ligands had reduced subtype selectivity, because ligand binding was primarily determined by the most potent ligand for each Na V subtype (μ-KIIIA for hNa V 1.4 and m 3 -HwTx-IV for hNa V 1.7). While m 3 -HwTx-IV had an     Figure  4A,C and Table 1).
Ligand Binding Kinetics at hNa V 1.4 and hNa V 1.7. Ligand binding affinity is characterized by the equilibrium dissociation constant (K d ) and is determined from the ratio of kinetic rate constants that reflect formation of the ligand− receptor complex (association rate constant, k on ) and its dissociation (dissociation rate constant, k off ), with the equation K d = k off /k on . Experimentally, we determined k on and k off using ligand wash-in and washout periods, described by the time constant τ, using the formulas k on = (1/τ on − k off )/[ligand] and k off = 1/τ off . 23,50 k off could not be determined accurately for some ligands due to the poor reversibility of binding, and k on was calculated as the observed k on (k on* ), described by the equation k on* = 1/τ on . Kinetic data for precursors and bivalent ligands were determined at concentrations 10-fold higher than their respective IC 50 values using patch-clamp electrophysiology to identify potential bivalent effects ( Figure 5 and Table  2).
At hNa V 1.  Table 2). This can be explained by the washout results revealing that the monovalent ligand S-m 3 -HwTx-IV is already a nearly irreversible binder at this channel [k off = (8.12 ± 0.04) × 10 −7 s −1 ], leaving little room for improvement in terms of the dissociation rate for the bivalent ligand ( Figure 5B,C). Although the k off and K d values for this bivalent ligand remain to be calculated for an accurate comparison to the monovalent ligands, the comparison of the remaining hNa V 1.7 currents at the end of the washout period revealed that S-m 3  AzK-KIIIA had recoveries of 2.2 ± 0.8%, 3.6 ± 0.3%, and 12.6 ± 1.2%, respectively, normalized to I max at the end of the washout period ( Figure 5C and Table 2).

■ DISCUSSION
Conjugation of ligands that target the same ion channel via distinct modulatory mechanisms and binding sites is an innovative strategy for expanding the pharmacological toolbox available to study these channels. Bivalent or multivalent ligands often increase the effective concentration in the vicinity of the target, which can translate into various observable multivalent effects, including enhanced potency and binding kinetics. 51,52 For example, an engineered homobivalent protein kinase inhibitor had 100-fold higher potency for a particular subgroup of kinases, 53 and a homobivalent agonist targeting oxytocin receptor homodimers displayed potency that was ∼1000-fold greater than that of its monovalent counterpart. 54 Heterobivalent and multivalent ligands with improved potency have also been developed against the 5-HT 3 receptor 55 and the nicotinic acetylcholine receptor, 56 respectively.
Here, we conjugated the pore-blocking conotoxin μ-KIIIA to the optimized gating modifier spider toxin m 3 -HwTx-IV via bioorthogonal ligation with different length PEG linkers (40− 120 Å) and characterized the inhibitory potency, subtype selectivity, and binding kinetics of the bivalent and monovalent ligands at hNa V 1.4 and hNa V 1.7. Both venom peptides in the bivalent ligand [m 3 -HwTx-IV]-[PEG80]-[K-KIIIA] retained their overall 3D structure ( Figure 3B), which was reflected in their bioactivity ( Table 1). The dependence of bivalent effects on linker length was consistent with the structural model used to design the bivalent ligands ( Figure 1C). The bivalent ligand with the PEG80 linker produced the most pronounced bivalent effects, reflecting the measured half-circle length of 80 Å, an important finding that informs the appropriate linker lengths for future design strategies. The bivalent ligands with shorter (60 Å) and longer (120 Å) linkers displayed bivalent effects that were less pronounced than those of the 80 Å linker, and as predicted, the bivalent ligand with a 40 Å linker did not display any bivalent effects as it should not be able to span the two targeted binding sites.
The , respectively. This improvement in potency seems to be driven by a greatly reduced dissociation rate of the bivalent ligand (<5% current recovered after a 25 min washout period) when compared to those of the monovalent constituents (40− 55% recovered), while having similar on-rates (k on* ) despite the larger size of the bivalent ligand (  (Figures 4 and 5). We did not observe any bivalent effects in terms of potency or binding kinetics for [m 3 -HwTx-IV]-[PEG80]-[K-KIIIA] at hNa V 1.7, which can be explained by the binding kinetics of the bivalent and monovalent ligands. At hNa V 1.4, bivalency enhanced potency by slowing dissociation. At hNa V 1.7, this is not possible, because monovalent S-m 3 -HwTx-IV is already a nearly irreversible binder (k off of 8.12 × 10 −7 s −1 compared to a value of >10 −3 s −1 at hNa V 1.4). This hypothesis is supported by a recent study that investigated a similar heterobivalent ligand design comprising μ-KIIIA enzymatically ligated via a different linker to spider-venom peptide PaurTx3 (also known as β-TRTX-Ps1a). 57 PaurTx3 is a reversible binder at hNa V 1.7 (in contrast to S-m 3 -HwTx-IV), and therefore, in this case, the heterobivalent ligand yielded improved potency along with slower dissociation compared to those for the monovalent ligands. It is important to note that [m 3 -HwTx-IV]-[PEG80]-[K-KIIIA] might still have therapeutically beneficial bivalent effects at hNa V 1.7, which could not be observed with the washout period that we used but could become apparent in vivo, for example, through longer analgesic effects due to slower k off rates compared to that of m 3 -HwTx-IV.
In terms of selectivity, [m 3 -HwTx-IV]-[PEG80]-[K-KIIIA] was nearly equipotent at both channels (IC 50 values of 9 nM for hNa V 1.4 and 6 nM for hNa V 1.7), because binding was driven by the most potent ligand subunit for each channel (KIIIA for hNa V 1.4 and m 3 -HwTx-IV for hNa V 1.7). This might be of interest for molecular probe development where such modulation of selectivity could be an advantage of devising new pharmacological tools to study the effects of multiple subtypes simultaneously. It also highlights that ligand selection is critical, particularly for heterobivalent drug development, because reduced selectivity can translate into undesirable off-target effects.
Our results highlight the importance of investigating potency at the level of k on and k off rates in the design and engineering of bivalent ligands. Ligand binding kinetics are particularly important for therapeutic development because they define the target interaction, length of effects, dosing, and therapeutic window. 58,59 Ligands with slow dissociation rates, especially peptides with high selectivity, are often preferred drug leads because this translates into an increased target residence time, extended therapeutic effects, and improved patient compliance due to a lower frequency of drug administration. 60−62 The design of such long-acting ligands, however, remains challenging, and targeting two binding sites on a single channel via bivalent ligand design, as demonstrated in this work, represents an elegant strategy for delivering such longacting therapeutic leads.

■ CONCLUSION
In summary, we report the design, synthesis, and pharmacological characterization of a series of heterobivalent peptide ligands targeting hNa V 1.4 and hNa V 1.7. We developed a synthetic strategy that employed bioorthogonal ligation chemistry to conjugate a pore-blocking peptide to a gating modifier peptide using a panel of different length PEG linkers. We identified a heterobivalent ligand with improved potency, a switch from reversible to nearly irreversible binding, and new channel selectivity. This work highlights the power of heterobivalent ligand design to decrease the ligand−channel dissociation rate, which can translate into more potent and longer-lasting therapeutic effects. It furthermore provides important insights for future bivalent design strategies, including ligand-and linker-length selection. The strategy described here is expected to be broadly applicable to other ligands and ion channels, adding to the chemical repertoire of ion channel probes and drug leads. Oxidative Folding of Peptides. Oxidative folding of KIIIA peptides was performed as described previously. 29 Oxidative folding of S-m 3 -huwentoxin-IV was accomplished by glutathione-assisted folding at 25°C overnight under the following conditions: 15 μM  Serine to Aldehyde Conversion of S-m 3 -HwTx-IV. The Nterminal serine of S-m 3 -HwTx IV was oxidized with sodium periodate to an N-terminal aldehyde moiety in 10 mM sodium phosphate (pH 7). The sodium periodate stock solution was freshly prepared at 100 mM in H 2 O. The reaction was performed with 0.5 mM peptide and a 1.5-fold molar excess of sodium periodate (0.75 mM). The periodate solution was incubated for 2 min at 25°C in the dark. Oxidation was terminated by the addition of N-α-Fmoc-L-serine (Iris Biotech GmbH) to a final concentration of 5 mM.
Hydrazone Ligation. Ligation of the hydrazide-PEG linker with the aldehyde moiety of m 3 -HwTx-IV was performed with 100 mM sodium citrate (pH 4.5) using a peptide concentration of 1 mg/mL (130 μM) and a 2-fold molar excess of hydrazide-PEG linker (260 μM). The reaction was allowed to proceed at −20°C in the dark for 24 h. The product was confirmed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) on a TOF/TOF 5800 mass spectrometer ( Mass Spectrometry. The mass and purity of peptides and bioorthogonal reaction products were determined using liquid chromatography-coupled MS (LC-MS) using a high-resolution API Qstar Pulsar mass spectrometer (PerkinElmer Sciex, Foster City, CA) or a high-resolution TripleTOF 5600 mass spectrometer system (AB Sciex). LC with the API Qstar MS system was performed with an Atlantis T3-C 18 column (2.1 mm × 100 mm, 3 μm; Waters), and LC with the TripleTOF 5600 MS system was carried out with a Zorbax RRHD 300 SB-C 18 1 H− 1 H total correlated spectroscopy (TOCSY), and 2D 1 H− 1 H nuclear Overhauser effect spectroscopy (NOESY) spectra were recorded on a Bruker Avance 600 MHz NMR spectrometer equipped with a cryogenically cooled probe (cryoprobe) at 25°C. Spectra were processed using TopSpin (Bruker), and sequencespecific resonance assignments were made using CCPNMR Analysis 2.4.1. 63 Cell Culture. Cell culture reagents were from Life Technologies Corp. unless otherwise stated. Human embryonic kidney (HEK) 293 cells co-expressing either hNa V 1.4 or hNa V 1.7 and the β1 auxiliary subunit (SB Drug Discovery, Glasgow, U.K.) were maintained at 37°C in a humidified 5% CO 2 incubator in Minimal Essential Medium (Sigma-Aldrich) supplemented with 10% FBS, 100 units/mL penicillin, 100 μg/mL streptomycin, 2 mM L-glutamine, and variable concentrations of blasticidin, geneticin, and zeocin according to the manufacturer's protocols. Replicating cells were subcultured every 3− 4 days in a 1/5 ratio using 0.25% trypsin/EDTA.
Patch-Clamp Electrophysiology. Sodium currents were recorded using an automated whole-cell patch-clamp system (QPatch 16X; Sophion Bioscience, Ballerup, Denmark) as described previously. 23 The extracellular solution comprised 1 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, 3 mM KCl, 140 mM NaCl, and 20 mM TEA-HCl at pH 7.3 (320 mOsm), and the intracellular solution comprised 140 mM CsF, 1 mM EGTA/5 mM CsOH, 10 mM HEPES, and 10 mM NaCl at pH 7.3 (320 mOsm). The elicited currents were sampled at 25 kHz and filtered at 4 kHz. Cells were maintained at a holding potential −80 mV, and Na + currents elicited by 20 ms voltage steps to 0 mV from a −120 mV conditioning pulse applied for 200 ms. To obtain concentration−response curves, cells were incubated for 5 min with increasing concentrations of precursor peptides or bivalent ligands. This incubation period should be sufficient to obtain accurate IC 50 values for even the most potent ligands described in this study, as it has been used previously to study exceptionally potent inhibitors of Na V 1.7 (IC 50 < 1 nM). 23 However, Journal of Medicinal Chemistry pubs.acs.org/jmc Article we cannot exclude the possibility that the IC 50 might be overestimated for the most potent ligands, but this would change none of our conclusions. For on-rate experiments, Na + currents were measured at 15 s intervals over 15 min immediately following addition of peptide at a concentration equivalent to 10 times its IC 50 for the Na V subtype being analyzed. For k off measurements, cells were incubated with peptide for 10 min at a concentration equivalent to 10 times its IC 50 for the Na V subtype being analyzed, and Na + currents were assessed at 10 s intervals during 25 min saline washes. The k on , k off , and K d values were calculated using the equation K d = k off /k on (nM), where k off = 1/ τ off (s −1 ) and k on = (1/τ on − k off )/[ligand] (nM −1 s −1 ). Data were analyzed using Assay software (Sophion Biosciences), and Na + currents (I Na ) plotted as I/I max .
Data Analysis. For the in vitro electrophysiological recordings, curve fitting was performed using GraphPad Prism version 10 (GraphPad Software, San Diego, CA) using nonlinear regression with log inhibitor versus normalized response and variable Hill slope for dose−responses and IC 50 determination, and exponential one-phase association and dissociation for on-and off-rate analysis, respectively. Data are means ± SEM.
Additional figures illustrating cryo-electron microscopy structures for determination of linker length; characterization of peptides, linkers, and bivalent ligands; details of reaction monitoring for bivalent assembly; and inhibition data of μ-KIIIA analogue peptides (PDF) SMILES data (CSV)