Trifunctional Saxitoxin Conjugates for Covalent Labeling of Voltage-Gated Sodium Channels

Abstract

Voltage-gated sodium ion channels (Na_Vs) are integral membrane protein complexes responsible for electrical signal conduction in excitable cells. Methods that enable selective labeling of Na_Vs hold potential value for understanding how channel regulation and post-translational modification are influenced during development and in response to diseases and disorders of the nervous system. We have developed chemical reagents patterned after (+)-saxitoxin (STX)—a potent and reversible inhibitor of multiple Na_V isoforms—and affixed with a reactive electrophile and either a biotin cofactor, fluorophore, or ‘click’ functional group for labeling wild-type channels. Our studies reveal enigmatic structural effects of the probes on the potency and efficiency of covalent protein modification. Among the compounds analyzed, a STX-maleimide-coumarin derivative is most effective at irreversibly blocking Na⁺ conductance when applied to recombinant Na_Vs and endogenous channels expressed in hippo-campal neurons. Mechanistic analysis supports the conclusion that high-affinity toxin binding is a prerequisite for covalent protein modification. Results from these studies are guiding the development of next-generation tool compounds for selective modification of Na_Vs expressed in the plasma membranes of cells.

Graphical Abstract

Trifunctional chemical probes derived from the potent shellfish poison, (+)-saxitoxin (STX), irreversibly inhibit wild-type voltage-gated sodium channels (Na_Vs). Saxitoxin derivatives decorated with a maleimide electrophile and either biotin, a fluorescent dye, or biorthogonal-reactive group were synthesized and evaluated using whole-cell, voltage-clamp electrophysiology.

Introduction

The initiation and propagation of action potentials in electrically excitable cells relies on the coordinated actions of ion channels, among which voltage-gate sodium channels are exemplary.^[1] These large, heteromeric protein complexes, comprising a pore-forming α-subunit and one or two ß-auxiliary proteins, gate the flow of sodium ions in response to changes in membrane potential.^[2] Proper transmission of action potentials requires stringent control of protein spatial distributions, expression levels, and activities, which occurs through transcriptional, translational and post-translational controls. The influence of post-translational protein modifications (PTMs) on channel function is of particular interest to our lab and has motivated the development of tool compounds for selectively capturing membrane-inserted Na_Vs.^[3] Isolation of Na_Vs by immunoprecipitation with Na_V-specific Abs is possible;^[4] however, this method does not discriminate between channels inserted in the membrane of cells and those held in intracellular stores. The inability to differentiate between these two limiting ‘populations’ of channels complicates efforts to identify changes in PTMs that occur in rapid response to cellular cues to alter electrical signaling (e.g., acute cellular injury). The lack of specificity of Na_V Abs presents further challenges. Accordingly, we envisioned developing Na_V-selective, affinity-based ligands that would enable irreversible alkylation of Na_Vs present exclusively in the outer membrane of cells. Such reagents would be cell-impermeable and engineered for efficient labeling and isolation of membrane Na_Vs.

Guanidinium toxins, which target the outer vestibule of Na_Vs to block ion permeation, present an optimal scaffold from which to design Na_V-selective tool compounds.^[5] These hydrophilic compounds act reversibly with fast on and fast off binding kinetics and have measured IC₅₀ values of ≤ 10 nM against six of nine rat Na_V subtypes (i.e., rNa_V1.1–1.4, 1.6, 1.7). In our original plan, we conceived of introducing a cysteine (Cys) mutation into one of the four p-loops that form the outer vestibule of the channel. In turn, a Cys-reactive electrophile along with an affinity purification tag (e.g., biotin, ‘click’ functional group) would be conjoined to STX. This type of covalent, chemogenetic pair would, in principle, irreversibly label only the Cys-mutant channel, thus enabling isolation and characterization of a single channel subtype expressly located in the outer membrane.^[6] In our efforts to develop such tools, we found that STX-maleimide 1, a first-generation Cys-reactive toxin conjugate, irreversibly inhibited wild-type Na_V1.2 and 1.4 (Figure 1).^[7] These findings were entirely serendipitous and remain surprising, as there is no obvious candidate Cys group within ~20 Å of the toxin binding site (estimated from cryo-EM structures of Na_Vs, Figure 1).^[8] Herein, we describe results from a comprehensive study to measure the inhibitory behavior of a collection of STX-maleimide conjugates. The aim of this research was to understand how maleimide substitution and ligand concentration alter the efficiency of irreversible block and to determine if ligand binding precedes protein modification. Our findings evidence unexpected structure-activity relationships between different maleimide conjugates and identify a coumarin derivative, 15, that is a potent and effective irreversible inhibitor of Na_Vs expressed in heterologous cells and present in hippocampal neurons.

Figure 1A.

Saxitoxin and a cysteine-reactive derivative. 1B. Top and side view of Na_V1.4 showing STX (green space filling) docked in the outer vestibule. Cys723 and Cy725 are highlighted (yellow space filling). Other cysteine residues are colored in yellow. Domain I = orange; Domain II = red; Domain III = gray; Domain IV = cyan.

Results and Discussion

Design and Synthesis of Substituted Maleimide STX Conjugates

In our original study, maleimide was selected over other Cys-reactive electrophiles such as acetylhalides, acrylamides, and mixed disulfides for its intrinsic reactivity and ease of modification. Design of effective STX-maleimide inhibitors was predicated upon minimizing disruption of toxin binding, assuming that this step would precede the protein alkylation event. Prior studies^[7] revealed that C2’-substituted maleimide STX conjugates containing pendant alkyl chains retained irreversible binding activity against wild type-Na_Vs and was therefore considered an obvious location to introduce a ‘click’ group or affinity tag. Direct incorporation of biotin^[9] or a fluorogenic dye^[5a,10] was desired as no additional synthetic operations would be required to isolate or visualize toxin-labeled channels. Toxin-maleimide-‘click’ reactive probes, by virtue of the smaller size of the prosthetic groups, might retain higher affinity for binding to the channel, but would necessitate an extra reaction step following protein alkylation.^[11]

A synthetic route (Scheme 1) was devised to permit facile attachment of a reporter group to a C2’-substituted maleimide through amide bond formation using N-hydroxysuccinimide (NHS) ester intermediate 6. This chemistry was designed to enable facile preparation of maleimide derivatives to test the effects of the substituent groups on toxin binding and on the rate of protein alkylation.

Scheme 1.

Synthesis of C2’-substituted maleimides. Conditions: (a) O₃, PPh₃, CH₂Cl₂, MeOH, –78 °C, 94%; (b) glyoxylic acid, morpholine HCl, dioxane/H₂O, 100 °C; (c) allyl bromide, Aliquat 336, NaHCO₃, CH₂Cl₂, H₂O, 32% (2 steps); (d) Dess-Martin periodinane, CH₂Cl₂, 79%; (e) β-alanine, AcOH, 110 °C, 78%; (f) Boc₂O, DMAP, t-BuOH, 61%; (g) 5 mol% Pd(PPh₃)₄, morpholine, THF, MeOH, 94%; (h) N-hydroxysuccinimide, N,N’-dicyclohexylcarbodiimide, CH₂Cl₂, 88%; (i) i-Pr₂NH, DMF, RNH₂, 49–88%; (j) CF₃CO₂H, CH₂Cl₂; (k) N-hydroxysuccin-imide, EDC, CH₂Cl₂, 60–72% (2 steps).

Synthesis of C2’-substituted maleimides commenced with ozonolysis of undecylenic acid 2, which proceeded in 94% yield, followed by acidic condensation with glyoxylic acid to form γ-hydroxybutenolide 3 (Scheme 1). Subsequent allylation of the carboxylic acid under Schotten-Baumann conditions, oxidation to the maleic anhydride with Dess-Martin periodinane, and condensation with β-alanine in acetic acid furnished the desired disubstituted maleimide 4. Carboxylic acid protection as the tert-butyl ester and Pd-catalyzed cleavage of allyl ester afforded 5. Interestingly, extended reaction times for the latter reaction resulted in quantitative isomerization of the maleimide to the exocyclic alkene. By limiting reaction times to < 1 h, acid 5 could be exclusively and quantitatively obtained. Activation of 5 as the corresponding NHS ester 6 was followed by coupling to a 1° amine derived with a prosthetic group of choice.

Fully elaborated trifunctional STX conjugates can be prepared through NHS ester coupling reactions to aminoethyl-modified STX.^[7] As shown in Figure 2, a diverse set of reporter groups was successfully affixed to the STX maleimide scaffold including biotin (9) and biorthogonal-reactive alkyne (10), ketone^[12] (11–12), norbornene^[13] (13) and cyclopropene^[14] (14) units. In addition to the biotin- and ‘click’-based probes, a 7-(diethylamino)coumarin conjugate (15) was prepared to visualize by fluorescence imaging covalently-labeled Na_Vs in cells and/or on western blots. The coumarin dye was selected for its optical properties, small size, and neutral charge.^[15]

Figure 2.

Preparation of trifunctional STX maleimide conjugates.

Electrophysiological Characterization of Trifunctional STX Maleimide Probes

Whole cell patch-clamp electrophysiology recordings were initially performed against wild-type rat brain channel, rNa_V1.2, expressed in Chinese hamster ovary (CHO) cells to assess the inhibitory activity of the trifunctional STX maleimide conjugates. The extent of irreversible current inhibition was determined by comparing the amount of Na⁺ current (I) recovered after toxin wash-out relative to initial peak current levels (I₀) prior to toxin application. Following a six-minute application of a saturating concentration (10 μM) of 15, only partial restoration of initial peak current levels was observed (I/I₀ = 43 ± 3%) despite extensive wash-off with toxin-free buffer solution (Figure 3B and 3C). This finding stands in marked contrast to the fully reversible binding behavior of the parent toxin, STX. Decreasing the application time of 15 to one or three minutes resulted in concomitant increase in the extent of current recovery (81 ± 5% and 63 ± 2%, respectively; Figure S2A and S2B), a trend that is consistent with a time-dependent alkylation event and fewer channels being irreversibly labeled. Application of 1 μM 15 for six minutes, however, resulted in almost complete restoration of peak current (I/I₀ = 90 ± 1%), indicating that the extent of covalent modification is concentration dependent (Figure S2C). Unfortunately, evaluation of application times beyond six minutes posed a technical challenge due to difficulties with maintaining a high resistance seal on the patched cell for the duration of the experiment. Altogether, these findings suggest that 15 is a competent irreversible inhibitor of Na_V1.2, thus validating our probe design.

Figure 3A.

Structure of maleimide (15) and succinimide (16) coumarin STX conjugates. 3B. Current recordings after a 6 min application and wash-off of 10 μM maleimide 15 to CHO cells expressing rNa_V1.2. 3C. Representative time course following application and wash-off of 15. 3D. Current recordings after a 6 min application and wash-off of 10 μM succinimide 16. 3E. Representative time course following application and wash-off of 16. Average current recovery upon wash out was 43 ± 3% for 15 compared to 95 ± 10% for 16.

Succinimide coumarin derivative 16, a saturated analog of 15 incapable of protein alkylation, was employed as a control compound to confirm that irreversible current block of rNa_V1.2 is dependent on the electrophilic maleimide. As with 15, incubating 10 μM 16 with cells expressing rNa_V1.2 resulted in complete block of sodium current (Figure 3D and 3E). However, in stark contrast to 15, peak current (I/I₀ = 95 ± 10%) was restored upon toxin wash-out, behavior mirroring that of STX. The divergent activity between maleimide 15 and succinimide 16 conjugates strongly implicates protein alkylation by the maleimide electrophile as the mechanism for irreversible current block of rNa_V1.2.

We evaluated biotin 9, α-ketoamide 12, norbornene 13, and cyclopropene 14 conjugates to determine if and to what extent the choice of prosthetic group altered the extent of irreversible current block.^[7] Surprisingly, these derivatives showed markedly different propensities to irreversibly inhibit rNa_V1.2 compared to 15. Incubation of biotin 9 (10 μM, 6 min) resulted in full restoration of peak current following wash-off, indicating that no reaction with the channel occurred. ‘Click’-based probes, 12 and 13, fared better than 9, but were notably less efficient than 15 (Figure S3A–C) and cyclopropene 14 demonstrated similar binding behavior to its succinimide counterpart. These results underscore the challenges in designing, optimizing, and advancing effective trifunctional covalent inhibitors of wild-type Na_Vs.

IC₅₀ values were measured for trifunctional succinimide analogs to gauge the relative affinity of these compounds for Na_V1.2 and to potentially gain insight into the disparate results between the maleimide conjugates tested (Figure S4). An IC₅₀ value of 79 ± 8 nM was determined for coumarin 16 against rNa_V1.2, only a two-fold decrease in binding affinity relative to the unsubstituted STX-succinimide conjugate (IC₅₀ = 38 ± 5 nM).^[7] Concentration-response measurements for succinimide derivatives 17–19 yielded IC₅₀ values of 257 ± 9 nM, 126 ± 15 nM, and 105 ± 6 nM, respectively, and 324 ± 48 nM for the reversible maleimide, 9. Collectively, these results reveal that structural changes to both the linker and reporter groups alter rather substantially ligand potency and the rate of wild-type rNa_V1.2 alkylation.

We performed additional characterization of coumarin 15, the most effective irreversible inhibitor of rNa_V1.2, against other Na_V isoforms and Na_Vs expressed in primary cells. Inhibitory constants for succinimide 16 (Figure S5A) were measured using rNa_V1.4 (IC₅₀ = 105 ± 14 nM, CHO cells) and rat embryonic hippocampal neurons (IC₅₀ = 99 ± 6 nM), the latter of which endogenously express multiple STX-sensitive Na_V isoforms in complex with β subunits.^[16] As expected, wash-off of 16 was entirely reversible. These data show marginal differences in the binding affinities and kinetics for 16 towards STX-sensitive isoforms and cell types. With 15, irreversible inhibition was confirmed (10 μM, 6 min incubation time) against both rNa_V1.4 (I/I₀ = 50 ± 4%, Figure S5B) and rat embryonic hippocampal neurons (I/I₀ = 44 ± 7%, Figure S5C). The magnitude of irreversible block is equivalent to that measured with rNa_V1.2 and demonstrates that this phenomenon is neither cell-specific nor noticeably affected by the presence of β subunits.

Investigating the Mechanism of Irreversible Na_V Inhibition

We conducted a series of experiments to further understand the mechanism of covalent inhibition of wild-type Na_Vs by coumarin 15. Electrophysiology experiments using 20, a maleimide coumarin conjugate derived from β-saxitoxinol, a micromolar inhibitor of STX-sensitive channels, support the view that high-affinity toxin binding to Na_V is a prerequisite for irreversible current block (Figure 4A). The IC₅₀ of β-saxitoxinol is ~10,000-fold less than that of STX.^[17] Accordingly, a six-minute application of 10 μM 20 to CHO cells expressing rNa_V1.2 afforded only partial current block and complete current restoration (I/I₀ = 103 ± 5%) following wash-out (Figure 4B). That 20 bears a reactive maleimide moiety and does not irreversibly alter peak current suggests that protein alkylation by 15 is mediated through a specific interaction between toxin conjugate and channel.

Figure 4A.

Structure of β-saxitoxinol maleimide coumarin 20. 4B. Representative wash-off time course of 10 μM 20 against rNa_V1.2. 4C. Representative wash-off time course of 10 μM 15 against rNa_V1.4 E758D. Average current recovery following wash-off was 81 ± 6% (B), and 103 ± 4% (C).

We employed protein mutagenesis to provide additional support that high affinity toxin binding to the channel pore is required for protein alkylation. In STX-resistant clams, E758D (rNa_V1.4 numbering) is a naturally occurring amino acid variation in the outer vestibule of Na_Vs that significantly destabilizes toxin binding (IC₅₀ = 9.8 ± 0.18 μM).^[18,19] Accordingly, the binding behavior of coumarin 15 was evaluated against the rNa_V1.4 E758D mutant. Application of 10 μM 15 to cells expressing rNa_V1.4 E758D reduced peak current by ~40%, consistent with the destabilizing effect of this mutation on toxin binding. Toxin wash-off, however, resulted in complete restoration of I₀ (I/I₀ = 103 ± 4%, Figure 4C). These observations argue that protein alkylation is dependent on specific and potent toxin binding within the outer channel pore.

The potency of STX towards Na_Vs is influenced, albeit modestly, by the frequency of the electrophysiology stimulation protocol (i.e., use-dependence).^[20] As shown in Figure 5A, use-dependent effects for STX against rNa_V1.2 manifest as differences in IC₅₀ values, 1.2 ± 0.1 nM and 2.9 ± 0.1 nM, for recordings at high (f =1/2 Hz) versus low (f = 1/30 Hz) frequency, respectively.^[20a] We examined use-dependent binding with succinimide 16 to determine the extent to which, if any, irreversible block of current was altered by the stimulus protocol. Like STX, 16, shows a modest (< 3-fold) change in IC₅₀ at lower stimulating frequencies (79 ± 8 nM vs 214 ± 52 nM, respectively).

Figure 5A.

Use-dependent concentration-response measurements of STX and succinimide 16 against rNa_V1.2 at high (f = 1/2 Hz) and low (f = 1/30 Hz) frequency stimulation. 5B. Representative time course experiment demonstrating use-dependent irreversible inhibition of rNa_V1.2 with 10 μM 15.

To investigate the frequency dependence of irreversible inhibition, time course wash-off experiments were performed with coumarin 15 (Figure 5B). Following application of 15 (10 μM, 6 min) at low frequency stimulus (f = 1/30 Hz), wash-off with toxin-free buffer solution led to the recovery of 66 ± 8% of initial current. This value compares to I/I₀ = 43 ± 3%, which was obtained for 15 using a high-frequency stimulation protocol. These findings provide additional support that tight binding of toxin to its receptor is necessary for efficient protein alkylation.

Efforts to identify the protein residue(s) that undergoes reaction with the maleimide electrophile have yielded inconclusive results. Although prior work to implicate a Cys as the nucleophilic residue were unsuccessful,^[7] we revisited this possibility following a report that μOξ-conotoxin GVIIJ, a peptidic cone snail toxin, formed a disulfide bond with a Cys residue in the receptor site of wild-type rNa_V1.2.^[21] This conserved residue (C723, rNa_V1.4 numbering) and a neighboring Cys (C725) were each successfully mutated to leucine and alanine in rNa_V1.4, respectively (see Figure 1). When transfected in CHO cells, these mutant channels produced macroscopic current levels sufficient in magnitude for electrophysiological evaluation of toxin binding. Application of parent maleimide 1 (5 μM, 6 min) resulted in the irreversible inhibition of rNa_V1.4 C723L (I/I₀ = 19 ± 2%) and C725A (I/I₀ = 17 ± 9%) mutant channels at levels that mirror results with wild-type rNa_V1.4.^[7a] Based on these experiments, we conclude that C723 and C725 do not engage the maleimide. At this time, we cannot rule out the possibility that a nucleophilic residue(s) other than Cys is covalently modified.

Conclusion

We prepared seven unique trifunctional STX-maleimide conjugates and performed whole-cell electrophysiology experiments to characterize the binding activity and efficacy of these compounds as covalent modifiers of wild-type Na_V isoforms. Among this collection of reagents, a coumarin derivative was identified as optimal in terms of potency and efficiency of irreversible channel block. Mechanistic studies with succinimide-based analogues and mutant sodium channels establish that high-affinity toxin binding to the channel is a prerequisite for successful protein modification. Protein mutagenesis experiments aimed at identifying the nucleophilic amino acid residue(s) that engages the maleimide electrophile were, however, inconclusive. Efforts to further improve these tool compounds for affinity labeling and proteomics analyses of Na_Vs are ongoing.