Interactions of Disulfide-deficient Selenocysteine Analogs of μ-Conotoxin BuIIIB with the Voltage-Gated Sodium Channel Na_V1.3

Abstract

Inhibitors of the neuronal voltage-gated sodium channel subtype Na_V1.3 are of interest as pharmacological tools for the study of neuropathic pain associated with spinal cord injury and have potential therapeutic applications. The recently described μ-conotoxin BuIIIB from Conus bullatus (μ-BuIIIB) was shown to block Na_V1.3 with sub-micromolar potency (K_d = 0.2 μM), making it one of the most potent peptidic inhibitors of this subtype described to date. However, oxidative folding of μ-BuIIIB results in numerous folding isoforms, making it difficult to obtain sufficient quantities of the active form of the peptide for detailed structure-activity studies. Here we report the synthesis and characterization of μ-BuIIIB analogs incorporating a disulfide-deficient, diselenide-containing scaffold designed to simplify synthesis and facilitate structure-activity studies directed at identifying amino acid residues involved in Na_V1.3 blockade. Our results indicate that, like other μ-conotoxins, the C-terminal residues (Trp16, Arg18 and His20) are most crucial for Na_V1 block. At the N-terminus, replacement of Glu3 by Ala resulted in an analog with increased potency for Na_V1.3 (K_d = 0.07 μM), implicating this position as a potential site for modification for increased potency and/or selectivity. Further examination of this position showed that increased negative charge, through γ-carboxyglutamate replacement, decreased potency (K_d = 0.33 μM), while replacement with positively-charged 2,4-diamonobutyric acid increased potency (K_d = 0.036 μM). These results provide a foundation for the design and synthesis of μ-BuIIIB-based analogs with increased potency against Na_V1.3.

Introduction

The μ-conotoxins are a family of venom-derived peptides that block voltage-gated sodium channels (Na_V1), several of which are implicated in various pain pathways. These peptides act by binding at neurotoxin receptor site 1 on the outer vestibule of the Na⁺ conductance pore [1]. The μ-conotoxins are characterized by a six-cysteine framework (C_IC_II(X_n)C_III(X_n)C_IV(X_n)C_VC_VI) cross-linked by three disulfide bridges. The earliest reports identified μ-conotoxins, specifically μ-GIIIA and μ-PIIIA, that were selective for the skeletal muscle subtype (Na_V1.4) [2, 3]. More recent studies have focused on μ-conotoxins such as μ-KIIIA and μ-SIIIA, which preferentially block neuronal subtypes (Na_V1.2) over skeletal (Na_V1.4) and cardiac (Na_V1.5) muscle subtypes. These peptides have attracted considerable interest because of their potent analgesic activity in animal models of pain [4], although non-selective block of other Na_V1 subtypes has hindered their development as therapeutics. In an attempt to engineer in subtype selectivity, structure-activity relationship (SAR) studies were conducted to determine critical amino acid residues and subsequently identify potential sites for chemical modification. To date, detailed SAR studies have been carried out on a limited number of these peptides, including μ-GIIIA, μ-PIIIA, μ-KIIIA and μ-SIIIA [3-7].

Recently, several new μ-conotoxins were identified in the venom of Conus bullatus [8]. μ-BuIIIB exhibited significant differences from previously described μ-conotoxins, particularly with respect to the primary sequence at the N-terminus (Figure 1, Panel A). Initial studies revealed that μ-BuIIIB almost completely blocked sodium current in the Na_V1.4 (skeletal muscle) subtype, with near irreversibility [8]. More recently, Wilson and co-workers showed that μ-BuIIIB also blocked the Na_V1.3 subtype with relatively high potency (K_d = 0.2 μM), whereas μ-KIIIA and μ-SIIIA exhibited only modest block of this subtype (IC₅₀ = 8 and 11 μM, respectively) [9]. The importance of Na_V1.3 with respect to neuropathic pain lies in its increased expression following axotomy, where expression of other subtypes such as Na_V1.8 and Na_V1.9 is decreased [10, 11]. To further examine the role of Na_V1.3 in neuropathic pain, potent and selective inhibitors of this channel are needed. However, such ligands are currently not available.

Figure 1.

(A) Sequence alignment of μ-BuIIIB with other μ-conotoxins. These toxins are characterized by a shared disulfide-framework and the ability to block VGSC subtypes. μ-BuIIIB is distinguished from other members of this class by an extended N-terminus, in addition to increased length of inter-cysteine loop 1. (B) Description of the ddSecBuIIIB scaffold. Solution structure of μ-BuIIIB, with disulfide connectivity, is shown (pdb ID 2LO9) (19). Diselenide replacement of the Cys5-Cys17 bridge (blue) facilitated cyclization between these two residues independent of disulfide bridge formation. Removal of the Cys6-Cys23 bridge (red), by replacement with Ala, reduced the number of possible folding species to three for each ddBuIIIB analog shown. The BuIIIB[C5U,C17U,C6A,C23A] analog (i.e. ddSecBuIIIB) was selected as a framework to carry out Ala-replacement studies based on the ‘native-like’ block of Na_V1.3 (K_d = 0.2 μM). The ddSecBuIIIB scaffold was employed to assess the importance of each non-cysteine residue within the primary sequence against Na_V1.3 and Na_V1.4.

Here, we describe the development of a disulfide-deficient, diselenide-bridge containing analog of μ-conotoxin BuIIIB (ddSecBuIIIB) as the basis for an alanine scan to identify residues that contribute to block of Na_V1.3 (Figure 1, Panel B). The advantages of this approach over a more traditional Ala-scan using the peptide containing three disulfide bridges are simplification of the oxidative folding pathway and unambiguous assignment of the disulfide connectivity. μ-BuIIIB was chosen as a model system for this study because of the particular challenges faced by oxidative folding of this peptide. By reducing the number of potential folding intermediates through diselenide replacement of the Cys5-Cys17 pairing, and elimination of the Cys6-Cys23 disulfide bridge, we created a peptide scaffold in which the individual contributions of non-cysteine amino acid residues could be studied. Disulfide deletion and diselenide-bridge incorporation have previously been used, either individually or in combination, to facilitate oxidative folding studies of other conotoxins such as µ-SIIIA, µ-KIIIA and ω-GVIA [12-14], although these strategies have yet to be employed to facilitate SAR studies of disulfide-rich conotoxins. This approach could also be employed to rapidly determine amino acid residues critical for biological activity in other cysteine-rich peptides.

Results

Chemical Synthesis

The synthesis, purification, and oxidative folding steps are often the greatest impediments to SAR studies of disulfide-rich conotoxins. Oxidation of such conotoxins can result in multiple folding isoforms, yielding limited quantities of the properly folded peptide for pharmacological characterization. Furthermore, determination of the disulfide connectivities of the active isoform can be labor-intensive and often requires sophisticated analysis methods such as selective reduction-alkylation, direct MS fragmentation, or NMR methods [15, 16]. Before embarking on SAR studies of μ-BuIIIB, we first determined the contributions of individual disulfide bridges in μ-BuIIIB to Na_V1.3 inhibition.

An initial series of disulfide-deficient analogs was constructed where we systematically replaced the pair of cysteines in each disulfide with a pair of alanines (Table 1). Removal of the Cys5-Cys17 bridge led to a slight increase in k_on (0.17 versus 0.085 μM^-1min^-1) and a much greater increase in k_off as compared to the unmodified peptide (2.6 versus 0.07 min^-1). These values translated to a significant decrease in potency for BuIIIB[C5A,C17A] (K_d = 15.7 versus 0.2 μM), indicating that this bridge is important for Na_V1.3 inhibition. The Cys5-Cys17 bridge was therefore selected for replacement by a redox-favored diselenide bridge. Removal of either the Cys6-Cys23 or Cys13-Cys24 bridges yielded similar results, with k_on values of 0.23 and 0.34 μM^-1·min^-1, respectively. The k_off values were also similar, and closer to that of the unmodified peptide, 0.059 and 0.038 min^-1, respectively. These values translated to potencies in blocking Na_V1.3 equal to or greater than that of wild-type μ-BuIIIB (K_d of 0.26 and 0.11 μM for BuIIIB[C6A,C23A] and BuIIIB[C13A,C24A], respectively) (Table 1). The solution structure of µ-BuIIIB shows that the N-terminal helix is anchored to the core of the molecule by the Cys5-Cys17 and Cys6-Cys23 disulfide bridges [16]. Removal of either the Cys6-Cys23 or Cys13-Cys24 bridges would be expected to have structural consequences, although it appears that those changes do not significantly affect the peptide’s activity against Na_V1.3.

Table 1.

Structural and pharmacological characterization of ddBuIIIB and disulfide-deficient, selenocysteine-containing BuIIIB scaffolds.

Peptide	MW (Calc./Obsv.)	Correct Isomer	kon (μM·min)-1	koff (min)-1	Kd (μM)	ΔKd
μ-BuIIIB	2761.20/2761.19a	16%	0.085 ± 0.01	0.017 ± 0.007	0.20 ± 0.02
BuIIIB[C5A,C17A]	2700.25/2700.24a	32%	0.16 ± 0.06	2.6 ± 0.86	15.7 ± 2.6	79
BuIIIB[C6A,C23A]	2700.25/2700.24a	37%	0.23 ± 0.02	0.059 ± 0.004	0.26 ± 0.03	1.3
BuIIIB[C13A,C24A]	2700.25/2700.80a	35%	0.34 ± 0.03	0.038 ± 0.003	0.11 ± 0.013	0.55
BuIIIB[C5U,C17U,C6A,C23A]	2796.13/2796.13b	51%	0.12 ± 0.01	0.026 ± 0.004	0.2 ± 0.04
BuIIIB[C5U,C17U,C13A,C24A]	2796.13/2796.14b	43%	0.19 ± 0.02	0.08 ± 0.005	0.42 ± 0.05	2.1

^aMasses were calculated by MALDI-TOF or electrospray mass spectrometry.

^bAnalogs were screened against r-Na_V1.3. U denotes selenocysteine. ΔK_d denotes the ratio of each analog to wild-type μ-BuIIIB.

Using the information obtained from disulfide-deficient BuIIIB analogs, two additional analogs, BuIIIB[C5U,C17U,C6A,C23A] and BuIIIB[C5U,C17U,C13A,C24A], were constructed to identify the optimal disulfide-deficient, selenoconotoxin scaffold as a basis for SAR studies. Slight differences in biological activity were observed between the disulfide-deficient analogs and the ddSecBuIIIB scaffolds. BuIIIB[C5U,C17U,C6A,C23A] actually exhibited a faster k_on and k_off than BuIIIB[C5U,C17U,C13A,C24A], resulting in a scaffold that exhibited ‘native-like’ potency for Na_V1.3 (K_d = 0.2 μM) (Table 1).

Oxidative folding

Crude peptide was removed from the solid support resin using modified Reagent K cleavage mixture containing 2,2’-dithiobis(5-nitropyridine) (DTNP) as previously described [17, 18]. Briefly, the cleavage mixture was supplemented with 1.3 equivalents DTNP in order to remove the p-methoxybenzyl protecting group from selenocysteine, resulting in a selenocysteine – 2-thio-5-nitropyridine adduct. This adduct was then removed by treatment for 2 h with 50 mM DTT, leading to spontaneous formation of the diselenide bridge between residues 5 and 17 (Sec5-Sec17) [19] Deprotection and cyclization steps were performed directly on the crude mixture to minimize losses from additional purification steps. As such, quantitative yields of deprotection were not determined. However, analysis of the crude, Se-Se containing BuIIIB[C5U,C17U,C6A,C23A] analog revealed a single major product comprising 22% of the total mixture based on analytical HPLC peak area. The diselenide-containing peptides were then purified in parallel by C₁₈ solid-phase extraction. Finally, formation of either the Cys13-Cys24 or Cys6-Cys23 bridge was achieved using the solid support oxidant CLEAR-Ox^®. For each of the ddSecBuIIIB analogs, a well-separated major peak corresponding to the fully folded peptide was observed (Figure 2). Oxidative folding resulted in greater accumulation of the desired products, as determined by peak area using analytical HPLC, for both BuIIIB[C5U,C17U,C6A,C23A] (51%) and BuIIIB[C5U,C17U,C13A,C24A] (43%) compared to 16% for the unmodified μ-BuIIIB (Table 1, Figure 2 Panels B and E).

Figure 2.

Comparison of wild type μ-BuIIIB with ddSecBuIIIB. (A) Sequence of the unmodified peptide showing experimentally-determined disulfide connectivity. (B) HPLC folding profiles of linear peptide (top) and folding mixture (bottom), asterisks indicate peaks tested for functional activity. (C) Representative time course of block of rNa_V1.3 by 10 μM peptide (bar indicates when peptide was present). Inset shows current traces obtained before (green) and during (red) exposure of the peptide. (D) Sequence of the ddSecBuIIIB scaffold showing positions of selenocysteine (U) and alanine replacements. (E) HPLC profile of linear peptide (top) and folding mixture (bottom), asterisks indicate peak tested for functional activity. The sample profile illustrates the efficiency of the folding pathway. (F) Representative time course of block of Na_V1.3 (see Table 1 for comparison of kinetic constants). Inset shows current traces obtained before (green) and during (red) exposure to peptide.

¹H NMR of ddSecBuIIIB

The conformation of BuIIIB[C5U,C17U,C6A,C23A] was compared to that of the native peptide using NMR spectroscopy. Following optimization of the pH and temperature (Figure 3), 1D and 2D ¹H NMR spectra were acquired at pH 3.0 and 35 °C, ¹H chemical shifts were determined by standard sequential assignment (Table S1 and Table S2, BMRB ID #19923), and plots of deviations from random coil chemical shifts [20] for ddSecBuIIIB and μ-BuIIIB were constructed. The 1D ¹H spectrum for ddSecBuIIIB was less well dispersed than that of μ-BuIIIB (Figure 4 Panel A). H^α chemical shift deviations from random coil (Figure 4) suggested that the N-terminus of ddSecBuIIIB was largely unstructured, with some degree of helicity near the C-terminus between residues Gly14 and His20. H^α chemical shifts were generally similar to μ-BuIIIB, while the H^N shifts showed greater differences compared to the native peptide (Figure 4 Panel B). The solution structure of native μ-BuIIIB contains helices at both the N- and C-termini [16], and it is clear that the C-terminal helix is partially preserved in ddSecBuIIIB but the N-terminal region is less structured. These results are of interest since it was reported previously that functionally important residues in μ-KIIIA reside in the C-terminal helical region of that peptide [21].

Figure 3.

Optimization of NMR conditions for ddSecBuIIIB. (A) 1D ¹H NMR spectra of BuIIIB[C5U,C17U,C13A,C24A] and BuIIIB[C5U,C17U,C6A,C23A] versus μ-BuIIIB. Peptide samples were 160 μM in 10 mM phosphate buffer containing 5% ²H₂O, pH 5.5. Water signal was suppressed by excitation sculpting. Spectra were recorded at 22 °C over 128 scans at 600 MHz at a spectral width of 11 ppm. Amide/aromatic region has been shown to illustrate the decreased spectral dispersion of the disulfide-deficient, diselenide bridge-containing analogs compared to the wild-type μ-BuIIIB. (B) 1D ¹H NMR spectra of BuIIIB[C5U,C17U,C6A,C23A] (ddSecBuIIIB) at different temperatures in 5% ²H₂O. Sample pH was 4.8 and spectra were acquired using a Bruker 600 MHz spectrometer using 128 scans at a spectral width of 12 ppm. (C) pH titration of ddSecBuIIIB. 1D ¹H NMR spectra were collected in 100% ²H ₂O, pH was adjusted to with NaO²H and spectra were acquired at 25 °C. Spectra were collected using 32 scans at a spectral width of 16 ppm. Amide/aromatic region is shown.

Figure 4.

Structural comparisons of μ-BuIIIB and ddSecBuIIIB. (A) 1D ¹H-NMR spectra of ddSecBuIIIB scaffold (red) and wild-type μ-BuIIIB (black) at 35 °C, pH 3.0 and 600 MHz. (B) Deviation from random coil shifts of the H^α, H^β, and H^N resonances of μ-BuIIIB (black) and ddSecBuIIIB (red) at 35 °C and pH 3.0 [20]. Random coil shift values for oxidized Cys were used to estimate chemical shift deviation from random coil for selenocysteines. H^N and H^α chemical shift differences between ddSecBuIIIB and μ-BuIIIB are also plotted.

Structure-activity studies on ddSecBuIIIB

Although NMR suggested conformational differences between ddSecBuIIIB and μ-BuIIIB, structure-activity studies were carried out to identify amino acid residues in ddSecBuIIIB that contributed to Na_V1.3 blockade. These studies identified a number of residues, particularly near the C-terminus, that were important for Na_V1.3 potency (Table 2). Specifically, Ala-replacement of Trp16 or His 20 resulted in slow on-rate kinetics, which prevented steady-state binding from being achieved within the experimental time window at peptide concentrations near the expected K_d or IC₅₀. These analogs had an estimated 150-times lower potency compared with μ-BuIIIB or ddSecBuIIIB (K_d > 30 μM). Ala-replacement of Arg15, Arg18 or Arg22 also led to deleterious effects on Na_V1.3 blockade (K_d = 1.84, 15.7 and 3.81 μM, respectively) (Table 2; Figure 5). These results were interesting since earlier studies identified a crucial role for basic residues at the equivalent position to Arg15 in μ-BuIIIB [5]. Although mutations at this position decreased potency, it is clear that the other positively-charged residues in this region play more pronounced roles in Na_V1.3 inhibition. These results were consistent with those of McArthur et al. [22], which showed three basic residues (Arg12, Arg14 and Lys17) near the C-terminus of μ-PIIIA contributing to the block of Na_V1.2 despite sharing a high degree of sequence homology with Na_V1.4-selective μ-GIIIA [22]. Our results highlight the importance of basic residues at the C-terminus of ddSecBuIIIB for block of Na_V1.3 (Table 2; Figure 5).

Table 2.

Summary of the ability of μ-BuIIIB and Alanine-scan analogs of ddSecBuIIIB to block rNa_V1.3 and rNa_V1.4 expressed in X. laevis oocytes, as determined by two-electrode voltage-clamp measurements. Rate constants were calculated from at least three independent experiments using Prism software. K_d = k_off/k_on.

Analog	NaV1.3			NaV1.4
	kon (μM·min)-1	koff (min)-1	Kd (μM)	kon (μM·min)-1	koff (min)-1	Kd (μM)
μ-BuIIIB	0.085 ± 0.01	0.017 ± 0.007	0.2 ± 0.086	4.20 ± 0.760	0.015 ± 0.005	0.004 ± 0.001
ddSecBuIIIB	0.13 ± 0.07	0.026 ± 0.002	0.2 ± 0.02	7.15 ± 0.036	0.009 ±0.002	0.001 ± 0.000
[V1A] †	0.17 ± 0.015	0.038 ± 0.001	0.22 ± 0.02	0.940 ± 0.011	0.026 ± 0.002	0.027 ± 0.002
[G2A] †	0.087 ± 0.013	0.042 ± 0.003	0.48 ± 0.08	1.64 ± 0.258	0.054 ± 0.007	0.033 ± 0.007
[E3A] †	0.29 ± 0.047	0.02 ± 0.001	0.07 ± 0.012	41.6 ± 2.973	0.030 ± 0.002	0.001 ± 0.000
[R4A] †	0.026 ± 0.005	0.028 ± 0.003	1.1 ± 0.23	0.78 ± 0.081	0.022 ± 0.001	0.028 ± 0.003
[K7A] †	0.07 ± 0.005	0.05 ± 0.005	0.71 ± 0.088	1.73 ± 0.187	0.010 ± 0.001	0.006 ± 0.001
[N8A] †	0.062 ± 0.007	0.016 ± 0.001	0.26 ± 0.032	1.45 ± 0.220	0.005 ± 0.000	0.003 ± 0.001
[G9A] †	0.068 ± 0.013	0.027 ± 0.001	0.4 ± 0.08	1.03 ± 0.100	0.012 ± 0.001	0.012 ± 0.001
[K10A] †	0.050 ± 0.004	0.051 ± 0.004	1.02 ± 0.11	2.05 ± 0.276	0.008 ± 0.001	0.004 ± 0.001
[R11A] †	0.033 ± 0.002	0.037 ± 0.005	1.12 ± 0.15	1.11 ± 0.190	0.010 ± 0.000	0.009 ± 0.002
[G12A] †	0.098 ± 0.019	0.033 ± 0.004	0.34 ± 0.08	5.12 ± 0.068	0.008 ± 0.001	0.002 ± 0.000
[G14A] †	0.190 ± 0.002	0.070 ± 0.006	0.230 ± 0.107	5.53 ± 0.262	0.014 ± 0.001	0.003 ± 0.000
[R15A] †	0.060 ± 0.008	0.240 ± 0.012	1.84 ± 0.385	0.410 ± 0.035	0.105 ± 0.005	0.260 ± 0.026
[W16A] †	N/A	1.69 ± 0.180	>30	N/A	1.68 ± 0.123	2.07 ± 0.284
[R18A] †	0.06 ± 0.008	0.78 ± 0.075	15.7 ± 0.296	1.26 ± 0.124	0.147 ± 0.008	0.13 ± 0.009
[D19A] †	0.22 ± 0.012	0.73 ± 0.073	0.54 ± 0.027	22.5 ± 3.005	0.144 ± 0.009	0.009 ± 0.001
[H20A] †	N/A	1.01 ± 0.119	>30	0.03 ± 0.004	0.219 ± 0.023	10.3 ± 2.448
[S21A] †	0.09 ±0.003	0.02 ±0.003	0.37 ± 0.059	3.41 ± 0.078	0.018 ± 0.002	0.005 ± 0.001
[R22A] †	0.06 ±0.002	0.24 ±0.012	3.81 ± 0.194	N/A	0.174 ± 0.007	0.28 ± 0.005

^†Analogs possessing the ddSecBuIIIB scaffold (BuIIIB[C5U,C17U,C6A,C23A]). N/A denotes K_d value was not available due to slow kinetics, which precluded steady state from being achieved at concentrations tested within the experimental time frame. Kinetic data for μ-BuIIIB are from [10].

Figure 5.

Effects of Ala-substitution of non-cysteine residues in the ddSecBuIIIB scaffold. Bar graphs compare the potencies (ΔpK_ds) of ddSecBuIIIB Ala-walk analogs with that of wild-type μ-BuIIIB (ΔpK_d = 0) against Na_V1.3 (A) and Na_V1.4 (B). Substitutions that increased potency are shown in blue, those that dramatically decreased potency are in red, and those that produced mild to moderate effects are in black. (C) Sequence alignment of the P-loop regions of Na_V1.3 and Na_V1.4, which is the site of interaction of all μ-conotoxins including μ-BuIIIB. Amino acid differences between subtypes in this region are listed in red; residues comprising the selectivity filter (D, E, K and A) are underlined.

Replacement of Val1, Asn8, Gly9, Gly12, Gly14 or Ser21 with Ala had little effect on Na_V1.3 potency (Table 2). These positions in the sequence are therefore potential sites of modification for peptide engineering to improve the physicochemical and/or pharmacological properties of μ-BuIIIB. Similar approaches were applied previously, resulting in potent and/or selective analogs of the μ-conotoxins KIIIA and SIIIA [3, 23]. In the context of the ddSecBuIIIB scaffold, mutations made to the basic residues in loop 1 (Lys7, Lys10, and Arg11) led to moderate effects on Na_V1.3 inhibition (K_d = 0.71, 1.02, and 1.12 μM, respectively) and little or no effect on Na_V1.4 inhibition (Figure 5).

The most striking outcome of these studies was the improved potency obtained through substitution of Glu3 with Ala. This replacement actually improved Na_V1.3 potency by nearly three-fold (K_d = 0.07 μM) (Table 2; Figure 5). To further investigate the effects of amino acid replacement of the acidic residue, a positional scan of Glu3 was conducted.

Positional scanning Glu3

Removal of the negatively-charged residue at position 3 led to increased potency against Na_V1.3. This was consistent with findings by Ekberg et al., who showed that the charge state of μ-conotoxins was important for interaction with the negatively-charged VGSC pore region [24]. To further explore the effects of charge at this position, analogs were constructed where Glu3 was replaced with either γ-carboxyglutamate (Gla; -2 charge) or L-2,4-diaminobutyric acid (Dab; +1 charge) (Figure 6). Increasing the negative charge at position 3 reduced Na_V1.3 potency by over 1.5-fold (K_d = 0.33 μM), whereas reversal of the negative charge increased Na_V1.3 potency (K_d = 0.038 μM) beyond that of the (neutral) Ala substitution (Figure 6, Panel A and Panel B)

Figure 6.

Exploration of charge effects at position 3 using the ddSecBuIIIB scaffold. (A) Representative electrophysiology traces of ddSecBuIIIB analogs that reveal the consequences of different charge at position 3. (B) Table of kinetics of block of Na_V1.3 by ddSecBuIIIB analogs. Each value represents the average from at least three different peptide concentrations tested in triplicate. BuIIIB[C5U,C17U,C6A,C23A,E3Dab] was nearly three-fold more potent than μ-BuIIIB in blocking Na_V1.3. Current traces obtained before (green) and during (red) exposure of the peptide.

The steric effects of basic residues at this position were also examined (Figure 7, Panel A and Panel B). A series of positively-charged analogs was synthesized and tested that replaced Glu3 with L-2,3-diaminopropionic acid (Dap), ornithine (Orn), lysine (Lys), arginine (Arg) or histidine (His). All analogs containing basic residues at position 3 exhibited improved Na_V1.3 potency. These studies revealed an optimal size for the side chain at this position, with groups smaller (e.g. Dap) or larger (e.g. ornithine, lysine, arginine and histidine) than Dab resulting in decreased potency as compared to BuIIIB[C5U,C17U,C6A,C23A,E3Dab](Figure 7). Similar to what was seen in experiments using Na_V1.3, the [E3Dab] mutant showed increased potency against Na_V1.4, with k_on = 32.1 μM·min^-1, k_off = 0.017 ± 0.001 min^-1 and K_d = 0.0005 ± 0.0001 μM (data not shown). The SAR data suggested that any changes in Na_V1.3 potency were closely mirrored by changes in potency against Na_V1.4. As such, the other analogs were assayed against Na_V1.3 only, as our focus was on potency against Na_V1.3 rather than subtype selectivity.

Figure 7.

Exploration of steric effects at position 3 in the ddSecBuIIIB scaffold. (A) Representative electrophysiology traces of ddSecBuIIIB analogs that reveal the consequences of increased steric bulk of residue side-chains at position 3. (B) Table of kinetics of block of Na_V1.3 by ddSecBuIIIB analogs. Each value represents the average from at least three experiments tested in triplicate. Analogs with increased or decreased steric bulk, relative to that of BuIIIB[C5U,C17U,C6A,C23A,E3Dab] had greater potency than μ-BuIIIB in blocking Na_V1.3, but not to the extent that BuIIIB[C5U,C17U,C6A,C23A,E3Dab] did. Current traces obtained before (green) and during (red) exposure of the peptide.

Discussion

We have created an analog of the μ-conotoxin BuIIIB, ddSecBuIIIB, which folds efficiently during oxidative folding to a single major product with well-defined disulfide/diselenide connectivity. This stands in marked contrast to the native peptide, where, under optimized glutathione-assisted folding conditions, the biologically-active isoform comprised only 16% of the total folding mixture (Table 1; Figure 2).

¹H NMR spectra of the ddSecBuIIIB scaffold showed reduced amide peak dispersion compared with μ-BuIIIB. The differences in structure presumably resulted from greater flexibility of the molecule caused by removal of the Cys6-Cys23 disulfide bridge. It is intriguing, therefore, that this analog retained native-like potency for Na_V1.3. These results are not, however, without precedent as Tietze et al. showed that less-structured forms of the conotoxin μ-PIIIA also exhibited activity against the muscle subtype Na_V1.4, albeit with lower potency [25]. Closer examination of chemical shift plots (Figure 4) shows that ddSecBuIIIB retains some degree of α-helical content near the C-terminus, which is known to be important for Na_V1-subtype blockade. In light of the conformational differences between the two peptides in solution, we suggest that the ddSecBuIIIB scaffold should be considered as a new blocker of this channel rather than simply as a proxy for native μ-BuIIIB. This analog therefore represents a useful new pharmacological tool for studies of Na_V1 channels, one that, importantly, refolds very efficiently to a well-defined product.

Structure-activity studies using ddSecBuIIIB identified the critical roles of aromatic and basic residues near the C-terminus of ddSecBuIIIB (Figure 5). These results were largely consistent with previous SAR studies of other μ-conotoxins, presumably due to the high degree of homology amongst the neuronal-subtype preferring µ-conotoxins at the C-terminus. This region is thought to contribute to general VGSC target specificity, whereas structural diversity at the N-termini is thought to aid in discrimination amongst VGSC subtypes [26]. μ-BuIIIB differs from previously described μ-conotoxins in having a short, helical extension of the N-terminus. The ddSecBuIIIB [V1A] and [G2A] analogs exhibited native-like potency for Na_V1.3 (K_d = 0.2 and 0.48 μM, respectively). This differs slightly from the results for N-terminally substituted analogs of μ-BuIIIB described by Kuang et al. [16], which showed a modest decrease in potency against Na_V1.3 upon Ala-replacement of Val1 (K_d = 0.71 ± 0.3 μM) [16]. The differences between Na_V1.3 potencies for ddSecBuIIIB and μ-BuIIIB likely arise from greater flexibility in the N-terminus resulting from removal of the stabilizing effects of the Cys6-Cys23 bridge. In further support of this, previous NMR studies showed that the conformation of the N-terminus of μ-BuIIIB could be further constrained through replacement of Gly2 with D-Ala, resulting in increased Na_V1.3 potency above that of wild-type μ-BuIIIB [16]. Similar improvement to Na_V1.3 blockade was not, however, seen with the [G2DA] replacement on the ddSecBuIIIB background where results more closely reflected those of unmodified ddSecBuIIIB (data not shown).

In addition to the N-terminal extension, another distinguishing feature of µ-BuIIIB, as well as other μ-conotoxins from C. bullatus described to date (μ-BuIIIA and μ-BuIIIC), is an extended inter-cysteine loop between the second and third Cys ranging between five and eight residues (Figure 1, Table 3). This loop includes a characteristic Lys10-Arg11 dipeptide that is not observed in the other neuronal Na_V1-preferring μ-conotoxins. μ-SmIIIA also possesses a basic dipeptide at the equivalent position (Arg7-Arg8), and exhibits potent block of Na_V1.3, which may make loop 1 of interest for designing subtype-preferring peptide analogs [9]. Additionally, there appear to be differences in both the contributions and distribution of residues that confer mild to moderate effects on Na_V1 inhibition within this region. Where previous studies of μ-conotoxins showed little to no effect of Ala-substitution of residues near the N-terminus, μ-BuIIIB appears more sensitive to modification in this region. This is illustrated by an increased number of residues for which a three- to five-fold decrease in Na_V1 potency was observed following Ala-replacement (Table 3).

Table 3.

Contributions of individual amino acids to activity in µ-conotoxins possessing three disulfide bridges. This table highlights the relative importance of residues near the N-terminus of µ-BuIIIB; other μ-conotoxins appeared insensitive to changes in this region. Z, pyroglutamic acid; O, hydroxyproline. All peptides are amidated at the C-terminus. Residues are categorized by the effect of neutral amino acid replacement at each position (red, significant reduction in Na_V1 potency; pink, moderate reduction in Na_V1 potency; green, increased Na_V1 potency).

Peptide		NaV1-subtype	Reference
μ-TIIIA		NaV1.2	(21)
μ-KIIIA		NaV1.2	(24)
μ-SIIIA		NaV1.2	(22)
μ-BuIIIB		NaV1.3	This Work
μ-BuIIIB		NaV1.4	This Work
μ-PIIIA		NaV1.4	(27)

The structural basis for μ-conotoxin blockade of Na_V1 channels has been investigated extensively, and models of these interactions have been constructed [22, 27, 28]. Because of the large amount of available SAR data, efforts have focused primarily on interactions between μ-conotoxin GIIIA and Na_V1.4. The selectivity of μ-conotoxins for Na_V1.3 or other Na_V1 subtypes is most likely caused by interactions with the turret region (Figure 5 Panel C), for which there are no similar sequences in databases, making modeling a challenging task (S. Kuyucak, personal communication).

Crystal structures of bacterial sodium channels have been reported recently [29, 30]. Unfortunately, the lack of tetrameric symmetry, coupled with differences in amino acid composition and selectivity filter loop size of the bacterial channels, have limited their use as proxies for mammalian Na_V1 subtypes. Until accurate models of the mammalian channel are developed, μ-conotoxin analogs such as those described here will prove useful as molecular tools to test and refine models of Na_V1.3.

The ddSec-strategy has proven valuable in facilitating rapid identification of amino acid residues critical for Na_V1.3 blockade, and those amenable to modification for potential improved pharmacological activity. It is also likely that the strategy employed here will be more broadly applicable to other disulfide-rich peptides for which inefficient oxidative folding precludes detailed SAR studies.

Experimental procedures

Ethics statement

Use of animals in this study followed protocols approved by the University of Utah’s Animal Care and Use Committee that conform to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Peptide synthesis

All analogs were synthesized at 30 μmol scale using an Apex 396 automated peptide synthesizer (AAPPTec, Louisville, KY) employing standard Fmoc-protocols. Peptides were constructed using pre-loaded Fmoc-Cys(Trt) Rink Amide MBHA resin (Substitution = 0.32 meq·g^-1). Fmoc-removal was accomplished by 20 min deprotection using 20% (vol/vol) piperidine (PIP) in dimethylformamide (DMF). Standard amino acids were purchased from AAPPTec and used without further purification. Fmoc-_L-Sec(pMeOBzl)-OH was purchased from Chem-Impex International, Inc. Amino acid coupling was accomplished using 1 equivalent 0.22 M benzotriazol-1-yl-oxytripyrrolidino-phosphonium hexafluorophosphate (PyBOP) and 2 equivalents 2 M diisopropylethyl amine (DIPEA) in N-methyl-2-pyrrolidone (NMP). Standard amino acids were coupled for 60 min in 10-fold excess and Fmoc-selenocysteine was coupled for 90 min in 3-fold excess. Upon completion of peptide synthesis, crude peptides were cleaved from solid-support resin by 5 h treatment with enriched reagent K (trifluoroacetic acid (TFA)/thioanisole/phenol/nH₂O (90:2.5:7.5:2.5, vol/vol/vol/vol) containing 1.3 equivalents 2,2’-dithiobis(5-nitropyridine) (DTNP). DTNP was added to the mixture to facilitate removal of Mob-protecting groups from selenocysteine side-chains. Crude peptides were removed from resin by vacuum filtration and were precipitated overnight at -20 °C in chilled methyl-tert butyl ether (MTBE). Precipitate was washed repeatedly with chilled MTBE. 5-thionitropyridyl derivatives were removed from Sec side chains, with concomitant diselenide-bridge formation by 2 h treatment with 50 mM dithiothreitol (DTT), 0.1 M Tris-HCl (tris(hydroxymethyl)aminomethane), 1 mM EDTA (ethylenediaminetetraacetic acid), pH 7.5, at room temperature.

RP-HPLC purification and oxidative folding

Crude peptide analogs containing the pre-formed diselenide bridge were purified, in parallel, by solid phase extraction (SPE) using a Supelco Visiprep vacuum manifold. Briefly, 3 mL Supelco Discovery DSC-18 SPE columns were conditioned with five column volumes of solvent B₉₀ (90% (vol/vol) acetonitrile (ACN); 10% (vol/vol) H₂O; 0.1% (vol/vol) TFA) followed by equilibration with five column volumes of solvent A (0.1% (vol/vol) TFA; nH₂O). A gradient of increasing organic solvent concentration was applied to each column ranging from 5 - 35% solvent B₉₀, and flow-through was checked for purity by analytical RP-HPLC equipped with a C₁₈ column (Vydac Cat# 218TP54) over a linear gradient ranging from 10 - 50% solvent B₉₀ with a 10 min pre-equilibration at initial conditions. Clean fractions were pooled and final yields were calculated by UV absorbance at 280 nm.

Formation of the Cys13-Cys24 disulfide bridge was performed in parallel using 48 molar equivalent excess (1 equiv = 2.768 μg resin/nmol peptide/number of disulfides to be oxidized) of the solid-support oxidant CLEAR-Ox^® (Peptides International, Inc.). Resin was prepared by first swelling in 500 μL dichloromethane (DCM) for 30 min at room temperature. Excess DCM was removed by vacuum filtration and the resin was washed once with 500 μL of each of the following: DMF, MeOH, 50% (vol/vol) ACN in H₂O, and 0.05 M Tris-HCl; 75% (vol/vol) ACN, pH 8.7. Peptide was dissolved in 0.05 M Tris-HCl; 75% (vol/vol) ACN, pH 8.7 to a final concentration of 3 mM and added directly to CLEAR-Ox resin. Oxidation was allowed to proceed for 2 h at room temperature, after which folding was quenched by 100-fold dilution with solvent A. Folded analogs were purified by preparative RP-HPLC over a linear gradient ranging from 10 - 50% solvent B₉₀ in 40 min with a 10 min pre-equilibration at initial conditions. The folded peptide analogs were quantified by UV absorbance (λ = 280 nm). Molecular masses of the folded disulfide-depleted, diselenide-containing peptides were confirmed by electrospray ionization-MS (Table 1).

Two-electrode voltage clamp electrophysiology

Xenopus laevis oocytes were injected either with 2.5 ng/μL Na_V1.3 or 94 ng/μL Na_V1.4 cRNA. Although μ-conopeptide activity can be sensitive to co-expression of Na_Vβ-subunits [31], neither Na_V1 was co-expressed with any Na_Vβ-subunits so that results could be compared directly with those obtained previously [9]. Two-electrode voltage clamping was performed using microelectrodes containing 3 M KCl (<0.5 MΩ). Sodium currents were acquired using a holding potential of -80 mV and stepping to 0 mV for 50 ms every 20 s. Peptides were applied to oocytes under static bath conditions. Off-rates (k_off) were calculated from single exponential fits of the time course of recovery from block following washout. On-rates (k_on) were calculated by linear regression of the slopes of plots of k_obs versus peptide concentration. Dissociation constants (K_d) were determined from the ratio of k_off/k_on. Experiments were all conducted in triplicate (n = 3 oocytes) for each data point at room temperature.

NMR spectroscopy

Peptide samples were dried by lyophilization and dissolved to a final concentration of 500 µM (ddSecBuIIIB) or 80 μM (μ-BuIIIB) in 95% (vol/vol) H₂O, 5% (vol/vol) ²H₂O, and the pH was adjusted to 3.0. All spectra were acquired on a Bruker 600 MHz NMR spectrometer. For one-dimensional ¹H NMR the water signal was suppressed using excitation sculpting [32, 33]. Experimental conditions were established through a series of temperature dependence and pH titration experiments which showed better dispersion at elevated temperature and acidic pH (Figure 3). Chemical shifts were referenced to water. Two-dimensional ¹H NMR spectra were acquired with a mixing time of 200 ms (NOESY) and spin-lock times of 30 and 70 ms (TOCSY). Data were processed with NMR Pipe software and ¹H chemical shifts were determined by standard sequential assignment of the NOESY and TOCSY spectra in CcpNmr Analysis [34].

Supplementary Material

Supp Table S1-S2. Supporting Information.

Table S1: Chemical shifts for wild-type μ-BuIIIB at 35 °C, pH 3.0

Table S2. Chemical shifts for BuIIIB[C5U,C17U,C6A,C23A] at 35 °C, pH 3.0

Abbreviations

BuIIIB

μ-conotoxin BuIIIB

ddSecBuIIIB

disulfide-deficient selenocysteine-containing BuIIIB

HPLC

high performance liquid chromatography

Na_V1.3, Na_V1.4 etc.

the α-subunit of the voltage-gated sodium channel subtype 1.3, 1.4, etc. cloned from rat

NMR

nuclear magnetic resonance

SAR

structure-activity relationship

The abbreviations for the common amino acids (L-isomers unless indicated otherwise) are in accordance with the recommendations of the IUPAC-IUB Joint Commission on Biochemical Nomenclature (Eur. J. Biochem. 1984, 138 9-37.