Bidirectional Modulation of the Voltage-Gated Sodium (Nav1.6) Channel by Rationally Designed Peptidomimetics

Disruption of protein:protein interactions (PPIs) that regulate the function of voltage-gated Na+ (Nav) channels leads to neural circuitry aberrations that have been implicated in numerous channelopathies. One example of this pathophysiology is mediated by dysfunction of the PPI between Nav1.6 and its regulatory protein fibroblast growth factor 14 (FGF14). Thus, peptides derived from FGF14 might exert modulatory actions on the FGF14:Nav1.6 complex that are functionally relevant. The tetrapeptide Glu-Tyr-Tyr-Val (EYYV) mimics surface residues of FGF14 at the β8–β9 loop, a structural region previously implicated in its binding to Nav1.6. Here, peptidomimetics derived from EYYV (6) were designed, synthesized, and pharmacologically evaluated to develop probes with improved potency. Addition of hydrophobic protective groups to 6 and truncation to a tripeptide (12) produced a potent inhibitor of FGF14:Nav1.6 complex assembly. Conversely, addition of hydrophobic protective groups to 6 followed by addition of an N-terminal benzoyl substituent (19) produced a potentiator of FGF14:Nav1.6 complex assembly. Subsequent functional evaluation using whole-cell patch-clamp electrophysiology confirmed their inverse activities, with 12 and 19 reducing and increasing Nav1.6-mediated transient current densities, respectively. Overall, we have identified a negative and positive allosteric modulator of Nav1.6, both of which could serve as scaffolds for the development of target-selective neurotherapeutics.


Introduction
In excitable cells, voltage-gated sodium (Nav) channels are the primary molecular determinants of the generation and conduction of action potentials [1]. Of the nine different pore-forming α-subunits that have been described (Nav1.1-1.9), the Nav1.1-1.3 and 1.6 isoforms are the primary Nav channels expressed in the central nervous system (CNS) [2]. Given their primacy in modulating neuronal excitability, it is unsurprising that aberrant activity of these isoforms has been implicated in the etiologies of numerous neurologic and neurodevelopmental disorders, including epilepsy [3][4][5][6], migraines [7][8][9], and autism [10][11][12]. Lending credence to the involvement of Nav channel aberrations in neuropsychiatric disorders, Nav channel blockers are also commonly used as adjunct therapies for the treatment of bipolar disorder [13,14], anxiety [15], and schizophrenia [16][17][18]. Given this primacy of Nav channel dysfunction in the etiology of virtually all CNS disorders, they have been the target of many drug discovery campaigns. Such efforts, however, have been of limited success, as developed lead compounds often fail to demonstrate appreciable isoform selectivity and resultantly confer undesirable off-target side effects [19].
Nav1.6 channels are abundantly expressed in medium spiny neurons (MSNs) of the nucleus accumbens (NAc), a neuroanatomical region that governs mesocorticolimbic circuitry and is involved in motivation [20], reward processing [21], learning [22], and locomotion [23][24][25]. With translational studies increasingly illuminating linkages between dysfunction of MSNs in the NAc and neuropsychiatric and neurological symptomologies [26][27][28][29], the development of chemical probes to modulate these neurons would enable further mechanistic elucidation of these neuropathophysiological processes. Given the abundant expression of Nav1.6 and its regulatory protein fibroblast growth factor 14 (FGF14) in these neurons, a feasible approach for developing such probes is through the rational design of small molecules targeting the FGF14:Nav1.6 protein:protein interaction (PPI) interface [30,31].
To that end, we previously identified two peptides capable of modulating FGF14:Nav1.6 complex assembly [32]. One of these peptides, Phe-Leu-Pro-Lys (FLPK; 1, Figure 1A), was derived from a four amino acid sequence located on the β12 sheet of FGF14 at the FGF14:Nav1.6 PPI interface, whereas the other, Glu-Tyr-Tyr-Val (EYYV; 5, Figure 1B), was derived from an amino acid sequence located on the exposed β8-β9 loop of FGF14 [32]. To improve the potency and drug-like properties of 1, we acetylated and aminated its N-terminus and C-terminus respectively (2, Figure 1A), and subsequently designed, synthesized, and pharmacologically evaluated peptidomimetics derived from the protected FLPK scaffold [30,31]. Briefly, altering the N-terminal substituent to Cbz and adding a Boc protective group to the constituent lysine residue of the protected FLPK scaffold produced ZL181 (4, Figure 1A), which inhibited FGF14:Nav1.6 complex assembly (half maximal inhibitory concentration (IC 50 ) = 63 µM) and decreased maximal and instantaneous firing frequencies in MSNs of the NAc [31]. Subsequent efforts to optimize 2 included altering the C-terminal substituent to methoxyl (OMe) and the addition of a Fmoc protective group to the constituent lysine residue of the protected FLPK scaffold, which produced ZL0177 (3, Figure 1A). Crucially, this analog displayed markedly improved potency, inhibiting FGF14:Nav1.6 complex assembly with an IC 50 value of 11 µM. Additionally, mechanism of action (MOA) studies using whole-cell patch-clamp electrophysiology revealed that ZL0177 significantly reduced Nav1.6-mediated peak current density and caused a depolarizing shift in voltage-dependence of Nav1.6 channel activation, suggesting that the peptidomimetic's activity is conferred by functioning as a partial FGF14 mimic [30].
To develop additional probes to further elucidate the intricate biophysical changes Nav1.6 channels undergo on account of their PPI with FGF14, we herein report our derivation and pharmacological evaluation of EYYV analogs (6, Figure 1B). To do so, all newly synthesized peptidomimetics were screened using an in-cell assay, which revealed that addition of hydrophobic protective groups to 6 followed by truncation to a tripeptide produced a potent inhibitor of FGF14:Nav1.6 complex assembly. Conversely, addition of hydrophobic protective groups to 6 followed by addition of an N-terminal benzoyl substituent produced a potentiator of the complex's assembly. After surface plasmon resonance (SPR) studies revealed that both peptidomimetics exhibited great affinity for Nav1.6, subsequent functional evaluation of the two analogs using whole-cell patch-clamp electrophysiology confirmed their inverse activities, with the inhibitor and potentiator of FGF14:Nav1.6 complex assembly reducing and increasing Nav1.6-mediated transient current densities, respectively. Overall, by employing this novel approach that combines in-cell screening with orthogonal validation measures, including SPR and whole-cell patch-clamp electrophysiology, we were able to identify both a negative and positive allosteric modulator of Nav1. 6, both of which could serve as ideal scaffolds for the development of targeted pharmacological probes and, with further chemical optimization, potential neurotherapeutic drug leads.

Chemistry
As outlined in Scheme 1, the diverse EYYV analogs were synthesized with modifications to both their Cand N-termini ( Figure 1B). The dipeptide analogs (10 and 11) were generated by condensation of commercially available compound 7 with compound 8 and 9, respectively. Deprotecting the Fmoc group of compounds 8 and 9, followed by coupling with compound 7, led to tripeptide compounds 12 and 13, respectively. Following a similar synthetic procedure to that of the preparation of compound 12, compounds 12 and 13 were converted into tetrapeptide compounds 15 and 16 in excellent yields. Compounds 17-24 were prepared by deprotection of compounds 15 and 16 and the subsequent introduction of the corresponding substituents, including acetyl, benzoyl, adamantanyl carbamic, adamantane-1-carbonyl, decanoyl, cyclohexanecarbonyl, and Cbz. Copies of 1 H and 13 C NMR spectra of all newly synthesized analogs are included in the Supplementary Materials.

In-Cell Testing of Analogs Using the Split Luciferase Complementation Assay (LCA)
We previously developed, optimized, and reported an in-cell LCA that allows for the reproducible identification of modulators of FGF14:Nav1.6 complex assembly [33]. Briefly, a HEK293 cell line stably expressing both CLuc-FGF14 and CD4-Nav1.6-NLuc recombinant proteins (hereafter referred to as Clone V cells) was developed. In this system, binding of FGF14 to the C-terminal tail of Nav1.6 facilitates the reconstitution of the luciferase enzyme, which in the presence of the substrate luciferin, produces luminescence. This assay was employed in the present investigation by reconstituting all the newly synthesized analogs derived from 6 in dimethyl sulfoxide (DMSO), bringing them to a preliminary screening concentration of 25 µM, administering them to Clone V cells seeded into 96-well plates, and subsequently normalizing the luminescence values observed in treatment wells to per plate controls (0.5% DMSO alone), the results of which are summarized in Figure 2A and Table 1.

Figure 2.
Screening of peptidomimetics against the FGF14:Nav1.6 complex using the split luciferase complementation assay (LCA). (A) Clone V cells were treated with compounds (25 µM; n = 8 wells per compound) in 96-well plates for 12 h prior to luminescence reading, and maximal luminescence for each well was normalized to per plate 0.5% DMSO controls (n = 32 per plate). (B) To assess effects of compounds on cell viability, the CellTiter Blue reagent was dispensed immediately following luminescence reading in (A). Fluorescence was detected after incubation for 18 h. Tamoxifen, known to be toxic to HEK293 cells [33], was used as a positive control. Individual replicate values with mean ± standard error of the mean (SEM) are shown. Significance was assessed using one-way ANOVA with post hoc Dunnett's multiple comparisons test. ** p < 0.0005. * p < 0.005.  Table 2. After characterizing the in-cell activity of the dipeptide and tripeptide analog, we subsequently sought to design, synthesize, and pharmacologically evaluate tetrapeptide analogs with improved cLogP values and stability compared to the parental EYYV peptide. To do so, hydrophobic protective groups were added to the protected EYYV scaffold, which produced 17. In-cell screening of 17 revealed that it markedly potentiated assembly of the FGF14:Nav1.6 complex ( Figure 2A and Table 1) and displayed low micromolar potency (EC 50 = 12.5 µM; Figure 3 and Table 2). Having identified a potentiator of FGF14:Nav1.6 complex assembly, further chemical interventions were employed to develop a more efficacious enhancer. To do so, we first replaced the NH 2 substituent at the C-terminus of compound 17 with a methoxyl (OMe) group, which produced 18. This R 1 modification to OMe resulted in a complete loss of in-cell activity, a recurring finding that was observed in the only other analog with a C-terminal OMe (16), suggesting that a C-terminal NH 2 substituent played a crucial role in conferring analogs with in-cell activity. To investigate how altering the hydrophobicity of peptidomimetics modulated their pharmacological activity, the acetyl group at the N-terminus of 17 was modified to various lipophilic groups, which produced 15, 19, 20, and 21. Substituting the Ac group with a Fmoc group (15) resulted in a significant loss of in-cell activity, a phenomenon that is likely attributable to the substituent conferring the analog with a highly steric structure and unfavorable cLogP value (cLogP = 7.45; Table 1). Further efforts to optimize 17 by introducing a N-terminal aryl group produced 19. Despite this analog displaying lessened potency relative to 17 (half maximal effective concentration (EC 50 ) = 24.5 ± 1.7 µM versus 12.5 ± 1.9 µM for 19 and 17, respectively; p < 0.005), its efficacy was significantly improved (E max = 192.4% ± 3.2% versus 153.4% ± 3.5% for 19 and 17, respectively; p < 0.0005; Figure 3 and Table 2). Whereas replacing the Ac group of 17 with an aryl group improved efficacy, modification of the R 2 locale to fused alkyl ring substituents (20 and 21) abrogated in-cell activity. Additional attempts to optimize 17 entailed altering its N-terminal substituent to a long-chain alkyl, cycloalkyl, and Cbz group, which produced 22, 23, and 24, respectively. Although 22 and 24 retained activity, 22 displayed lessened potency (EC 50 = 47.5 µM; Figure 3 and Table 2) relative to 17, and 24 s potentiation of FGF14:Nav1.6 complex assembly was comparably mild ( Figure 2A and Table 1). Overall, these SAR studies using the LCA identified three potential positive allosteric modulators (PAMs; 17, 19, and 22) and one negative allosteric modulator (NAM; 12) of FGF14:Nav1.6 complex assembly, the four of which were selected for protein:ligand binding studies using SPR.

Characterization of Peptidomimetic Interactions with FGF14 and Nav1.6
After SAR studies using the LCA-identified potential PAMs (17, 19, and 22) and a NAM (12), we next sought to examine their kinetic interactions with FGF14 and Nav1.6 C-terminal tail. SPR-based kinetic analysis of interactions between drug-like compounds and target proteins has become a key method for drug discovery [34,35]. The determined k on and k off rate constants provide important information about interaction mechanisms and compound properties, enabling investigation of SARs and the rational modification of compounds. Therefore, we used SPR to assess and compare the binding of peptidomimetics to FGF14 and Nav1.6 C-terminal tail protein. Proteins were purified as previously described [31,36], immobilized to CM5 sensor chips, and increasing concentrations of compounds (0.195-100 µM) were flown over the chips at 60 µL/min. The results are shown in Figure 4 and Tables 3 and 4. With the exception of 22, all compounds bound appreciably to FGF14 with affinities ranging from 2.9 to 14.3 µM (Table 3, left). While 12 and 19 bound strongly to Nav1.6, with affinities ranging from 2.3 to 9.7 µM, lower affinities were observed for 17 and 22 (Table 3, right).
Interestingly, the tripeptide (12) demonstrated two-fold greater binding affinity toward FGF14 compared to 19 (K D of 2.98 vs. 6.49 µM, respectively; p < 0.05). As similar dissociation kinetics were observed for the two compounds (k off = 1.86 × 10 −2 and 1.30 × 10 −2 s −1 for 12 and 19, respectively; p > 0.4), this difference was largely driven by the faster association rate of 12, which can also be observed by the leftward shift of the FGF14 steady-state saturation curve (red) for 12 relative to 19 ( Figure 4, right-top and lower middle). The affinities of the other two tetrapeptide analogs, 17 and 22, toward FGF14, were reduced by an even greater extent, collectively indicating that the presence of the glutamic acid residue, regardless of the R 2 substituent present, prevents high-affinity FGF14 binding. This may suggest that the binding region of FGF14 is topologically constrained such that it cannot readily accommodate the larger tetrapeptide analogs, or that the presence of the glutamic acid residue masks crucial interactions between the YYV motif and FGF14.
Whereas the presence of the glutamic acid residue reduced the affinity of analogs toward FGF14, its inclusion conferred tighter binding to Nav1.6. This finding was underscored by the tripeptide exhibiting the lowest detectable binding strength toward Nav1.6. This suggests that the region of Nav1.6 bound to by these analogs is less topologically constrained relative to FGF14. Additionally, compared to 12 and 17, 19 exhibited about 4-fold greater binding affinity toward Nav1.6 (p < 0.005 for 12 vs. 19; p < 0.05 for 12 vs. 19), a difference largely driven by its comparatively slow dissociation rate (k off = 1.5 × 10 −1 s −1 ). This difference can similarly be observed via the leftward shift of the Nav1.6 steady-state saturation curve ( Figure 4, right, blue curve).

Figure 4.
Determination of peptidomimetic binding kinetics to FGF14 and Nav1.6 C-terminal tail by SPR. Increasing concentrations of peptidomimetics (0.195-100 µM) were flown over purified FGF14 or Nav1.6 C-tail protein bound to CM5 chips using a flow rate of 60 µL/min. Kinetic analysis of each ligand/analyte interaction was obtained by fitting the response data to the simplest Langmuir 1:1 interaction model (K D = k off /k on ), and binding sensorgrams are shown for FGF14 (left) and Nav1.6 (middle), with fitted curves represented in black. Steady-state saturation plots (right) are shown for comparison of compound binding to FGF14 (red) and Nav1.6 (blue), with response units (RU) normalized to the maximal binding response. The resulting equilibrium dissociation constants (K D ), as well as kinetic association (k on ) and dissociation (k off ) rates, are provided in Table 3.
Consistent with the SAR studies using the LCA, which revealed that modification of 17 s R 2 substituent from Ac to benzoyl (19) improved efficacy (Table 2), this modification also resulted in an approximately 2-fold increase in binding to FGF14 and a 4-fold increase in binding to Nav1.6. Kinetically, this improved binding affinity was mediated via the benzoyl substitution conferring slower association and dissociation rates toward both proteins (as depicted by greater curvature in the SPR sensorgrams), a change likely attributable to the substituent's constitutive π bonds enabling additional interactions with both binding partners. Overall, given 19 s heightened efficacy and binding affinities toward both FGF14 and Nav1.6 relative to the other potential PAMs (17 and 22), it, along with the potential NAM (12), were selected for functional evaluation.  The equilibrium dissociation constants (K D ) were derived from data shown in Figure 4. The K D shown for each compound is an average of that calculated using the simplest Langmuir 1:1 interaction model ( where K D is the dissociation constant, and k off and k on are the first-order rate constants for the dissociation and association, respectively, of the protein:ligand complex) and the steady-state saturation model. b p < 0.005 compared to 17. c p < 0.05 compared to 19. d p < 0.05 compared to 12. e p < 0.05 compared to 17. f p < 0.05 compared to 19. g p < 0.005 compared to 12. While it is potentially somewhat surprising that 12, 17, and 19 bind to both FGF14 and the C-terminal tail of Nav1.6, this finding is consistent with a previous investigation that revealed both overlap and structural divergence between the FGF14:FGF14 homodimer and FGF14:Nav1.6 complex PPI interfaces [36]. That these peptidomimetics bind to both FGF14 and Nav1.6 is desirable in that it affords multiple mechanisms by which they could modulate the PPI between FGF14 and Nav1.6. In the case of the inhibitor (12), its disruption of the PPI between FGF14 and Nav1.6 could be conferred via direct binding to the FGF14 interaction site on the C-terminal tail of Nav1.6 or via binding to FGF14 and causing the protein to undergo a conformational change that makes it inaccessible to its native interaction site on the C-terminal tail of Nav1.6. Similarly, for the potentiator of FGF14:Nav1.6 complex assembly (19), its modulatory effects on the PPI could be conferred by both binding to the C-terminal tail of Nav1.6 that makes the FGF14 interaction site increasingly accessible to the regulatory protein, or, conversely, by binding to FGF14 and causing it to undergo a conformational change that affords it with increased accessibility to its interaction site on the C-terminal tail of Nav1.6. In the native system, it is expected that the protein:ligand interactions of 12 and 19 with FGF14 and the C-terminal tail of Nav1.6 will concomitantly occur and collectively enable functionally relevant modulation of FGF14:Nav1.6 complex assembly.

Electrophysiological Evaluation of Compounds 12 and 19
Heretofore, it had been shown that addition of a N-terminal benzoyl substituent (19) to tetrapeptide analogs produced an efficacious potentiator of FGF14:Nav1.6 complex assembly, whereas truncation to a tripeptide (12) yielded an inhibitor of the complex s assembly. Additionally, both 12 and 19 demonstrated promising protein:ligand binding interactions. As such, in an effort to further investigate their seemingly inverse activities, 12 and 19 were selected for functional evaluation as modulators of Nav1.6. To do so, HEK293 cells stably expressing Nav1.6 (HEK-Nav1.6) were incubated for 30 min with 0.1% DMSO (control), 20 µM 12, or 20 µM 19. After incubation, the modulatory effects of these peptidomimetics on Nav1.6-mediated currents were assessed using whole-cell patch-clamp electrophysiology ( Figure 5 and Table 4). The first kinetic property of Nav1.6 channels we investigated as potentially being modulated by 12 and 19 was peak current density. Investigation of how this property was modulated by these analogs revealed that treatment with 12 (-18.30 ± 2.02 picoampere per picofarad (pA/pF); n = 10) significantly reduced Nav1.6-mediated peak current density derived from transient Na+ current relative to treatment with 0.1% DMSO (-55.39 ± 2.16 pA/pF; n = 10), whereas treatment with 19 had the inverse effect and potentiated peak current density (-79.63 ± 3.55 pA/pF; n = 8) ( Figure 5B,C). The observed phenotypes could be attributable to the compounds altering channel availability by favoring or inhibiting steady-state (closed) inactivation, single channel conductance, or the number of channels at the plasma membrane. Such alterations could be mediated by 12 and 19 binding to the channel and inducing conformational changes that favor the channel adopting non-conductive (closed or inactivated) and conductive (open) states, respectively.
Lending further credence to 19 s potentiation of Nav1.6 channel activity, subsequent investigations revealed that treatment with 19 induced a -20.03 ± 0.85 mV (0.1% DMSO; n = 10) to -22.49 ± 0.72 mV hyperpolarizing shift (n = 8; p < 0.05) of V 1/2 of activation ( Figure 5E,F), which was not accompanied by a change in the V 1/2 of steady-state inactivation (n = 8; p = 0.984) ( Figure 5G,H). Conversely, treatment with 12 affected neither the V 1/2 of activation (n = 10; p = 0.756) nor the V 1/2 of steady-state inactivation (n = 6; p = 0.600) relative to treatment with 0.1% DMSO ( Figure 5E-H). Interestingly, both 12 and 19 were shown to significantly increase the long-term inactivation of Nav1.6 channels relative to the control ( Figure 5I,J). Overall, these findings suggest that both peptidomimetics have complex mechanisms of action mediated via modulation of multiple kinetic properties of Nav1.6 channels that collectively confer them with inverse activities.

Molecular Docking Studies of Compounds 12 and 19
With compounds 12 and 19 functionally validated as negative and positive allosteric modulators of Nav1.6 respectively, we next docked these compounds to the Nav1.6 C-terminal tail homology model [36] in silico to further elucidate their binding modes using Schrödinger Small-Molecule Drug Discovery Suite. Both compounds 12 and 19 docked well into the Nav1.6 C-terminal tail at a binding site identical to the EYYV motif (within the β8-β9 loop) of FGF14, enabling interactions with the same group of key residues, as shown in Figure 6. For compound 12 (as depicted in Figure 6A,B), the constituent oxygen atom of the backbone carboynl group between Y and V forms hydrogen bonds with Arg1891. The NH 2 group at the C-terminus of compound 12 forms two hydrogen bonds with residues Met1832 and Asp1833, whereas the Fmoc protecting group at the N-terminus of compound 12 interacts with Arg1866 via a π-cation interaction. The predicted docking model of compound 19 with the Nav1.6 C-terminal tail ( Figure 6C,D) is fundamentally different from that of compound 12, which may help elucidate the opposing actions of the two compounds on Nav1.6 activity. Whereas the C-terminus of 12 interacted strongly with Asp1833 and Met1832 via hydrogen bonds, compound 19 was inversely oriented, with its C-terminus engaging with Ser1859 and Asp1863. NH groups between the N-terminal benzoyl substituent and the glutamic acid residue of the backbone and between the two tyrosine residues of the backbone form hydrogen bonds with Ala1831 and Asp1858, respectively. Meanwhile, the constituent oxygen atom of the ester bond at the sidechain of the glutamic acid residue of the tetrapeptide backbone interacts with Arg1891 through hydrogen bonding. Like compound 12, the NH 2 group at the C-terminus of compound 19 also forms two hydrogen bonds, thus providing a structural rationale for why replacing the C-terminal NH 2 with OMe abrogated the in-cell activity of analogs. The N-terminal benzoyl substituent of compound 19 forms hydrophobic interactions with a small hydrophobic pocket in the Nav1.6 C-terminal tail that is surrounded by Ile1827, Glu1828, Ile1830, Ala1831, and Met1832, a result in agreement with our SAR studies demonstrating that introducing a benzene ring at the N-terminus is essential for heightened efficacy.
Additionally, an overlay of the FGF14:Nav1.6 complex with the highest scoring binding poses for compounds 12, 17, and 19 ( Figure 6E) exemplifies the drastic binding differences that result from the chemical changes, with the greatest disparity observed between 12 and 19. Importantly, the docking results for the tripeptide (12) point toward its binding interactions most resembling those of the native surface residues on the β8-β9 loop of FGF14 at its interface with Nav1.6 (shown as orange in Figure 6E). Theses similarities between the binding interactions of 12 and FGF14 with the Nav1.6 C-terminal tail suggest that 12 s truncated size relative to tetrapeptide analogs confers it with heightened mimicry of FGF14, thereby giving rise to its marked suppression of Nav1.6-mediated peak current density. Based upon theses structural analyses, and consistent with current models of Nav channel function [37][38][39], we propose that 12 and 19 exert opposite effects on Nav1.6-mediated currents as a result of differential interactions with the EF hand-like (EFL) and IQ domains of the C-terminal tail of the channel at sites that have established roles in channel trafficking and inactivation.

Discussion
Although it has been reported that aberrant firing of striatal neurons is implicated in a multitude of neuropsychiatric and neurological symptomologies [26][27][28][29]40,41], elucidation of how perturbations in their constituent PPIs contribute to these neuropathophysiological processes remains hampered by a lack of target-selective chemical probes. To that end, we sought to identify small molecule modulators of the PPI between FGF14 and Nav1.6, two proteins that are abundantly expressed in MSNs of the NAc and whose PPI, when perturbed, leads to neural circuitry aberrations [42,43]. To do so, we designed peptidomimetics derived from the EYYV parental tetrapeptide by employing a number of chemical interventions, including: (A) truncation of the tetrapeptide to a tripetide (YYV) and dipeptide (YV), (B) protection of the C-terminus of 6, and (C) modification of the N-terminus of 6 with various substituents. After these newly designed peptidomimetics were synthesized, their modulatory effects on FGF14:Nav1.6 complex assembly were assessed using the LCA. In-cell screening of these analogs revealed that a tripeptide (12) displayed low micromolar inhibitory activity against FGF14:Nav1.6 complex assembly (IC 50 = 23.7 µM), whereas addition of hydrophobic protective groups to 6 followed by addition of a N-terminal acetyl (17) or benzoyl (19) substituent produced potentiators of the complex's assembly. Subsequent dose response analyses using the LCA revealed that 19 was the most efficacious analog among the potentiators. Protein:ligand binding studies of the four most promising hits revealed that 12 showed strong binding affinities toward both FGF14 and Nav1.6 C-terminal tail proteins, with K D values of 2.88 and 9.86 µM, respectively. Additionally, 19 displayed appreciable binding affinities toward both of these proteins (K D = 6.3 and 2.25 µM for FGF14 and Nav1.6, respectively). Functional evaluation of these two analogs using whole-cell patch-clamp electrophysiology confirmed their inverse activities, with 12 and 19 reducing and increasing Nav1.6-mediated transient current densities, respectively. Lastly, molecular docking studies of compounds 12 and 19 with the Nav1.6 C-terminal tail homology model were performed, which indicated that these two compounds formed crucial hydrogen bonds and hydrophobic interactions with Nav1.6.
The findings of the present investigation are consistent with previous efforts to develop small molecule modulators of PPIs between Nav channels and their regulatory accessory proteins [30,31], although the results differ in subtle, but crucial ways. For example, we previously presented ZL0177 [30], a novel peptidomimetic derived from FLPK, a four amino acid sequence located on the β12 sheet of FGF14 at its interface with Nav1.6 [32]. Like the herein reported compound 12, ZL0177 suppressed Na v 1.6-mediated peak current density; however, it also accelerated the kinetics of current decay and induced a depolarizing shift in the voltage-dependence of Na v 1.6 channel activation [30]. That these phenotypes are not also induced by compound 12, despite 12 inducing an even stronger suppression of Na v 1.6-mediated peak current density relative to ZL0177 (−18.30 ± 2.02 pA/pF vs. -26.65 ± 6.3 pA/pF for 12 and ZL0177, respectively) likely stems from 12 being derived from EYYV, which, unlike FLPK, is located on the exposed β8-β9 loop of FGF14 at its interface with the C-terminal tail of Nav1.6. As such, it is expected that 12 and ZL0177 will differentially bind to the FGF14 interaction site on the C-terminal tail of Na v 1.6 and induce slightly divergent phenotypes.
Currently, these peptidomimetics are being employed as novel pharmacological probes to interrogate the biophysical properties of Nav channels. Given their high molecular weights, as well as their high clogP and tPSA values, chemical optimization of these peptidomimetics by replacing their natural amino acids with non-natural amino acids and substituting functional groups to confer improved water solubility is likely necessary to permit blood-brain permeability and will be the subject of future investigations [44]. By pursuing these chemical optimization strategies, we envision that the negative and positive allosteric modulator of Nav1.6 herein identified could be developed into CNS penetrant lead compounds with promising potential to be advanced into latter stages of preclinical testing as PPI-based neurotherapeutics.
To a solution of HNEt 2 (0.1 mL) and MeCN (0.4 mL), compound 11 (1.7 g, 3 mmol) was added and the solution was stirred at room temperature for 1 h. After the reaction was completed, the solution was concentrated to remove the solvent. Then, the residue was dissolved in 20 mL of dry CH 2 Cl 2 and Fmoc-Tyr(O t Bu)-OH (7) (1.3 g, 3 mmol) was added. The mixture solution was cooled to 0 • C with an ice bath. HOBt (459 mg, 3 mmol), HBTU (2.3 g, 6 mmol), and DIPEA (2 mL, 12 mmol) were added to the solution at 0 • C. Then, the ice bath was removed, and the mixture solution was stirred at room temperature overnight. After the reaction was completed (detected by TLC), the mixture was washed with 1 N NaHSO 4 , and saturated in NaHCO 3 and brine. After drying over anhydrous Na 2 SO 4 , the solution was concentrated and the residue was purified with silica gel column (CH 2 Cl 2 /MeOH = 50/1) to obtain compound 13 (1.7 g, 73%) as a white solid.
To a solution of HNEt 2 (0.1 mL) and MeCN (0.4 mL), compound 13 (2.4 g, 3 mmol) was added and the solution was stirred at room temperature for 1 h. After the reaction was completed, the solution was concentrated to remove the solvent. Then, the residue was dissolved in 20 mL of dry CH 2 Cl 2 and Fmoc-Glu(O t Bu)-OH (14) (1.3 g, 3 mmol) was added. The mixture solution was cooled to 0 • C with an ice bath. HOBt (459 mg, 3 mmol), HBTU (2.3 g, 6 mmol), and DIPEA (2 mL, 12 mmol) were added to the solution at 0 • C. Then, the ice bath was removed, and the mixture solution was stirred at room temperature overnight. After the reaction was completed (detected by TLC), the mixture was washed with 1 N NaHSO 4 , and saturated in NaHCO 3 and brine. After drying over anhydrous Na 2 SO 4 , the solution was concentrated and the residue was purified with silica gel column (CH 2 Cl 2 /MeOH = 50/1) to obtain compound 16 (2.2 g, 74%) as a white solid. 1 (17) To a solution of HNEt 2 (0.1 mL) and MeCN (0.4 mL), compound 15 (50 mg, 0.05 mmol) was added and the solution was stirred at room temperature for 1 h. After the reaction was completed, the solution was concentrated to remove the solvent. Then, the compound was dissolved in 5 mL of CH 2 Cl 2 and the solution was cooled to 0 • C with an ice bath. Then, Et 3 N (11 mg, 0.1 mmol) and acetyl chloride (7.9 mg, 0.1 mmol) were added. The mixture was stirred at room temperature overnight. The solution was diluted with 20 mL of CH 2 Cl 2 and washed with 1 N NaHSO 4 , saturated in NaHCO 3 and brine. After drying over anhydrous Na 2 SO 4, the solution was concentrated and the residue was purified with silica gel column (CH 2 Cl 2 /CH 3 OH = 50/1 to 20/1) to obtain compound 17 (26 mg, 66%) as a white solid. 1 13   To a solution of HNEt 2 (0.1 mL) and MeCN (0.4 mL), compound 16 (50 mg, 0.05 mmol) was added and the solution was stirred at room temperature for 1 h. After the reaction was completed, the solution was concentrated to remove the solvent. Then, the compound was dissolved in 5 mL of CH 2 Cl 2 and the solution was cooled to 0 • C with an ice bath. Then, Et 3 N (11 mg, 0.1 mmol) and acetyl chloride (7.9 mg, 0.1 mmol) were added. The mixture was stirred at room temperature overnight. The solution was diluted with 20 mL of CH 2 Cl 2 and washed with 1 N NaHSO 4 , saturated in NaHCO 3 and brine. After drying over anhydrous Na 2 SO 4 , the solution was concentrated and the residue was purified with silica gel column (CH 2 Cl 2 /CH 3 OH = 50/1 to 20/1) to obtain compound 18 (
Dose response curves were obtained using GraphPad Prism 8 by fitting the data with a non-linear regression: where x is log 10 of the compound concentration in M, x 0 is the inflection point (EC 50 or IC 50 ), A is the bottom plateau effect, B is the top plateau effect, and H is the Hill slope. Potency (IC 50 /EC 50 ) and efficacy (percent luminescence at the bottom or top plateau for inhibitors (I Min ) or enhancers (E Max ), respectively) were calculated from the dose response non-linear regression.

Cell Viability Assay
The CellTiter-Blue ® Cell Viability (CTB) assay (Promega, Madison, WI, USA) was used to counter-screen top compounds. Immediately following LCA luminescence reading from cells treated with experimental compounds or Tamoxifen (positive control with known toxicity toward HEK293 cells [33]), 30 µL of 1X CTB reagent was dispensed into 96-well plates, plates were incubated overnight (16 h) at 37 • C, and fluorescence was detected using the Synergy H1 reader (excitation λ = 560 nm, emission λ = 590 nm). Cell viability was expressed as percent of the mean fluorescent signal intensity of on-plate negative controls.

Electrophysiology Data Analysis
I Na was normalized to membrane capacitance to determine current density by dividing I Na amplitude by membrane capacitance. Current density was then plotted as a function of the holding potential to characterize current-voltage relationships. Tau of inactivation was calculated by fitting the decay phase of currents at −10 mV with a one-term exponential function. To assess voltage-dependence of Nav1.6 activation, conductance (G Na ) was first calculated using the following equation: where I Na is current amplitude at voltage V m , and E rev is the Na + reversal potential. Steady-state activation curves were then generated by plotting normalized G Na as a function of test potential. Plotted data was then fitted with the Boltzmann equation to determine V 1/2 of Nav1.6 activation values using the following equation: where G Na,Max is the maximum conductance, V a is the membrane potential of half-maximal activation, E m is the membrane voltage, and k is the slope factor. To assess steady-state inactivation, I Na normalized to max I Na (I Na /I Na,Max ) at the test potential was plotted as a function of pre-pulse potential. Data was then fitted with the Boltzmann function to determine V 1/2 using the following equation: where V h is the potential of half-maximal inactivation, E m is the membrane voltage, and k is the slope factor. For long-term inactivation, the peak I Na of depolarization cycles 2-4 was normalized to the peak I Na of depolarization cycle 1 (I Na /I Na,Cycle 1 ) and plotted as a function of depolarization cycle. Data analysis was performed using Clampfit 11.1 (Molecular Devices) and GraphPad Prism 8 (La Jolla, CA, USA) software. Results were expressed as mean ± standard error of the mean (SEM). Statistical significance was determined via unpaired t-tests comparing 0.1% DMSO to an experimental group, with p < 0.05 being considered statistically significant.

Molecular Docking Method
The molecular docking study was performed using Schrödinger Small-Molecule Drug Discovery Suite (Schrödinger, LLC, New York, NY, USA). FGF14:Nav1.6 homology model [36] was generated using the FGF13:Nav1.5:CaM ternary complex crystal structure (PDB code: 4DCK) as a template. The FGF14:Nav1.6 homology model was prepared with Schrödinger Protein Preparation Wizard using default settings. During this step, hydrogens were added, crystal waters were removed, and partial charges were assigned using the OPLS-2005 force field. The SiteMap (Schrödinger, LLC, New York, NY, USA) calculation was performed and a potential binding site was identified on the PPI of FGF14:Nav1.6. The chain of FGF14 was excluded from the model and the grid center was chosen on Nav1.6 C-terminal tail at the previously identified binding site generated from SiteMap results. The grid box was sized in 24 Å on each side to cover the PPI surface on Nav1.6. The 3D structure of ligands 12, 17, and 19 were created using Schrödinger Maestro and a low-energy conformation was calculated using LigPrep. Docking was employed with Glide using the SP protocol. Docked poses were incorporated into Schrödinger Maestro for a ligand-receptor interactions visualization. Top ranking of Glide GScore, as well as biological and chemical rationales were used for ligand docked pose evaluation and selection. The top docked poses of ligands 12, 17, and 19 were superimposed with FGF14:Nav16 complex homology model for an overlay analysis.