Coupling of Slack and NaV1.6 sensitizes Slack to quinidine blockade and guides anti-seizure strategy development

Quinidine has been used as an anticonvulsant to treat patients with KCNT1-related epilepsy by targeting gain-of-function KCNT1 pathogenic mutant variants. However, the detailed mechanism underlying quinidine’s blockade against KCNT1 (Slack) remains elusive. Here, we report a functional and physical coupling of the voltage-gated sodium channel NaV1.6 and Slack. NaV1.6 binds to and highly sensitizes Slack to quinidine blockade. Homozygous knockout of NaV1.6 reduces the sensitivity of native sodium-activated potassium currents to quinidine blockade. NaV1.6-mediated sensitization requires the involvement of NaV1.6’s N- and C-termini binding to Slack’s C-terminus and is enhanced by transient sodium influx through NaV1.6. Moreover, disrupting the Slack-NaV1.6 interaction by viral expression of Slack’s C-terminus can protect against SlackG269S-induced seizures in mice. These insights about a Slack-NaV1.6 complex challenge the traditional view of ‘Slack as an isolated target’ for anti-epileptic drug discovery efforts and can guide the development of innovative therapeutic strategies for KCNT1-related epilepsy.

7 results support that NaV1.6 is required for the observed high sensitivity of native KNa 115 channels to quinidine blockade. 116 Transient sodium influx through NaV1.6 enhances NaV1.6-mediated sensitization 117 of Slack to quinidine blockade 118 We next investigated the biomolecular mechanism underlying NaV1.6-mediated 119 sensitization of Slack to quinidine blockade. Considering that Slack currents are 120 activated by sodium influx 22,29 , we initially assessed the effects of NaV1.6-mediated 121 sodium influx on sensitizing Slack to quinidine blockade. We used 100 nM tetrodotoxin 122 (TTX) to block NaV1.6-mediated sodium influx (Fig. S3A,B) 31 . In HEK293 cells 123 expressing Slack alone, 100 nM TTX did not affect Slack currents; nor did it affect 124 Slack's sensitivity to quinidine blockade (IC50 = 83.27 μM) ( Fig. 2A,B and Fig. S3C,D). 125 In contrast, upon co-expression of Slack and NaV1.6 in HEK293 cells, bath-application 126 of 100nM TTX significantly reduced the effects of NaV1.6 in sensitizing Slack to 127 quinidine blockade (IC50 = 25.04 μM) ( Fig. 2A,B). These findings support that sodium 128 influx through NaV1.6 contributes to NaV1.6-mediated sensitization of Slack to 129 quinidine blockade. 130 It is known that NaV1.6-mediated sodium influx involves a transient inward flux that 131 reaches a peak before subsequently decaying to the baseline within a few milliseconds; 132 this is termed a transient sodium current (INaT) 32 . A small fraction of NaV1.6 currents 133 are known to persist after the rapid decay of INaT, and these are termed persistent sodium 134 currents (INaP) 33 . We isolated INaT and INaP to explore their potential contributions in 135 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted March 18, 2023. ; https://doi.org/10.1101/2023.03.16.532982 doi: bioRxiv preprint 8 sensitizing Slack to quinidine blockade. We selectively inactivated INaT using a 136 depolarized prepulse of -40mV (Fig. S4A) and selectively blocked INaP by bath-137 application of 20 μM riluzole, which is a relatively specific INaP blocker that is known 138 to stabilize inactivated-state NaV channels and delay recovery from inactivation 34,35 . 139 Our findings ultimately confirmed that the 20 μM riluzole selectively blocked INaP 140 compared to INaT in HEK293 cells co-expressing Slack and NaV1.6 ( Fig. S4B) and that 141 20 μM riluzole had no effect on Slack currents when expressed alone (Fig. S4C,D). supporting that NaV1.6-mediated sodium influx activates Slack. Interestingly, 145 inactivating INaT resulted in a > 20-fold decrease in Slack's sensitization to quinidine 146 blockade (IC50 = 22.26 μM) (Fig. 2C,E). In contrast, blocking INaP had no effect on 147 Slack's sensitization to quinidine blockade (IC50 = 1.60 μM) (Fig. 2F,H). These 148 findings indicate that NaV1.6 sensitizes Slack to quinidine blockade via INaT but not 149

INaP. 150
Given that Slack current amplitudes are sensitive to sodium influx, and considering that 151 quinidine is a sodium channel blocker, we examined whether NaV1.6 has higher 152 sensitivity to quinidine blockade than other NaV channel subtypes, which could 153 plausibly explain the observed increased strength of sensitization. We used whole-cell 154 patch-clamping to assess the sensitivity of NaV1.1, NaV1.2, NaV1.3, NaV1.5, and NaV1.6 155 to quinidine blockade. These sodium channels exhibited similar levels of quinidine 156 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted March 18, 2023.  Table 3), all of which were at least 40-fold lower than the NaV1.6-mediated 158 sensitization of Slack to quinidine blockade (Fig. 1F). Additionally, co-expressing 159 Slack with NaV1.1, NaV1.2, NaV1.3, NaV1.5, or NaV1.6 in HEK293 cells did not change 160 the sensitivity of these NaV channel subtypes to quinidine blockade ( Fig. S5 and 161 Supplementary Table 3). Thus, differential quinidine affinity for specific NaV channel 162 subtypes cannot explain the large observed NaV1.6-mediated sensitization of Slack to 163 quinidine blockade. Moreover, it is clear that NaV1.6-mediated sensitization of Slack to 164 quinidine blockade is directly mediated by INaT, rather than through some secondary 165 effects related to NaV1.6's higher sensitivity to quinidine blockade. 166

Slack interacts with NaV1.6 in vitro and in vivo 167
We found that the specific voltage-gated sodium channel blocker TTX did not 168 completely abolish the effects of NaV1.6 on sensitizing Slack to quinidine blockade 169 ( Fig. 2B), so it appears that a sodium-influx-independent mechanism is involved in the 170 observed NaV1.6-mediated sensitization of Slack to quinidine blockade. We therefore 171 investigated a potential physical interaction between Slack and NaV1.6. We initially 172 assessed the cellular distribution of NaV1.2, NaV1.6, and Slack in the hippocampus and 173 the neocortex of mouse. Consistent with previous studies 36,37 , NaV1.2 and NaV1.6 were 174 localized to the axonal initial segment (AIS) of neurons, evident as the co-localization 175 of NaV and AnkG, a sodium channel-associated protein known to accumulate at the AIS 176 ( Fig. 3A). Slack channels were also localized to the AIS of these neurons (Fig. 3A), 177 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is indicating that Slack channels are located in close proximity to NaV1.6 channels, and 178 supporting their possible interaction in vivo. Moreover, NaV1.6 was co-179 immunoprecipitated with Slack in homogenates from mouse cortical and hippocampal 180 tissues and from HEK293T cells co-transfected with Slack and NaV1.6 ( Fig. 3B,C), 181 supporting that a physical interaction between Slack and NaV1.6 occurs in vivo. 182 To assess the interaction between Slack and NaV1.6 inside living cells, we performed a 183 FRET assay in transfected HEK293T cells 38 . Briefly, we genetically fused mTFP1 and 184 mVenus to the C-terminal regions of Slack and NaV1.6, respectively (Fig. 3D). Upon 185 imaging the emission spectra cells co-expressing NaV1.6-mVenus and Slack-mTFP1 186 (measured at the plasma membrane region) (Fig. 3E), we detected positive FRET 187 signals, indicating a Slack-NaV1.6 interaction (Fig. 3F,H). The plasma membrane 188 regions from HEK293T cells co-transfected with NaV1.6 and Slack showed FRET 189 efficiency values much larger than a negative control (in which standalone mVenus and 190 mTFP1 proteins were co-expressed) (Fig. 3G,H), indicating that Slack channels reside 191 in close spatial proximity (less than 10 nm) to NaV1.6 channels in membranes of living 192

cells. 193
We next characterized the consequences of the Slack-NaV1.6 interaction in HEK293 194 cells using whole-cell recordings. Slack increased the rate of recovery from fast 195 inactivation of NaV1.6 ( Fig. S6E), with no significant effects on the steady-state 196 activation, steady-state fast inactivation, or ramp currents (Fig. S6C,D,F and 197 Supplementary Table 3). Additionally, we found that NaV1.6 had no significant effects 198 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted March 18, 2023. ; https://doi.org/10.1101/2023.03.16.532982 doi: bioRxiv preprint on the activation rate or the current-voltage (I-V) relationship of Slack currents (Fig.  199 S6A,B). These results indicate that the physical interaction between Slack and NaV1.6 200 produces functional consequences. Taken together, these findings support functional 201 and physical coupling of Slack and NaV1.6 in vitro and in vivo. 202 NaV1.6's N-and C-termini bind to Slack's C-terminus and sensitize Slack to 203

NaV1.6 binds to and sensitizes epilepsy-related Slack mutant variants to quinidine 259
blockade 260 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted March 18, 2023. ; https://doi.org/10.1101/2023.03.16.532982 doi: bioRxiv preprint Over 50 mutations in KCNT1 (Slack) have been identified to related to seizure 261 disorders 11 . Having established that NaV1.6 can sensitize wild-type Slack to quinidine 262 blockade, we next investigated whether NaV1.6 also sensitizes epilepsy-related Slack 263 mutant variants to quinidine blockade. We chose 3 Slack pathogenic mutant variants 264 (K629N, R950Q, and K985N) initially detected in patients with KCNT1-related 265 epilepsy 15,39,40 . Considering that these 3 mutations are located in Slack's C-terminus, 266 and recalling that Slack's C-terminus interacts with NaV1.6 ( Fig. 5D), we first used co-267 immunoprecipitation assays and successfully confirmed that each of these Slack mutant 268 variants interacts with NaV1.6 in HEK293T cell lysates (Fig. 6A). 269 Subsequently, whole-cell recordings revealed that NaV1.6 significantly sensitized all of  Table 4). These results support 272 that NaV1.6 interacts with examined Slack mutant variants and sensitizes them to 273 quinidine blockade. It is plausible that the Slack-NaV1.6 interaction contributes to the 274 therapeutical role of quinidine in the treatment of KCNT1-related epilepsy. 275

Viral expression of Slack's C-terminus prevents Slack G269S -induced seizures 276
Having established that blocking NaV1.6-mediated sodium influx significantly reduced 277 Slack current amplitudes (Fig. 2D,G), we found that the heterozygous knockout of 278 NaV1.6 significantly reduced the afterhyperpolarization amplitude in murine 279 hippocampal neurons (Fig. S8), together indicating that NaV1.6 activates native Slack 280 through providing Na + . We therefore assumed that disruption of the Slack-NaV1.6 281 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is

KA. 300
We assessed a time course of KA-induced seizure stages at 10-min intervals and found 301 that viral expression of Slack G269S resulted in faster seizure progression in mice 302 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted March 18, 2023. ; https://doi.org/10.1101/2023.03.16.532982 doi: bioRxiv preprint compared to control GFP expression (Fig. 7E). We calculated the total seizure score 303 per mouse to assess seizure severity 46 . Slack G269S -expressing mice showed significantly 304 higher seizure severity than GFP-expressing mice (Fig. 7F). The percentage of mice 305 with stage VI~IX seizures also increased, from 9.1% in GFP-expressing control mice 306 to 58.3% in the Slack G269S -expressing mice (Fig. 7G). These results support that viral 307 expression of Slack G269S significantly increases seizure susceptibility in mice. 308 To evaluate the potential therapeutic effects of disrupting the Slack-NaV1.6 interaction, 309 we delivered two AAV9s (one for Slack G269S and one for Slack's C-terminus [residues 310 326-1238]) into the CA1 region of mice (Fig. 7D). Viral expression of Slack's C-311 terminus in Slack G269S -expressing mice significantly decreased seizure progression, 312 seizure severity, and the percentage of mice experiencing stage VI~IX seizures (Fig.  313 7E-G). These results support that viral expression of Slack's C-terminus can prevent 314 Slack G269S -induced seizures in mice, thus showcasing that using Slack's C-terminus to 315 disrupt the Slack-NaV1.6 interaction is a promising therapeutic intervention to treat 316 KCNT1-related epilepsy. 317

DISCUSSION 318
We here found that NaV1.6's Nand C-termini bind to Slack's C-terminus and sensitize 319 Slack to quinidine blockade via NaV1.6-mediated transient sodium currents. These 320 results suggest that the pharmacological blocking effects of a channel blocker are not 321 exclusively mediated by the channel per.se., but modulated by channel's interacting 322 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is  1,20 . 327 Slack is functionally coupled to sodium influx, which is known to be mediated by ion 328 channels and receptors, including NaV, AMPARs, and NMDARs 21-23,47 . Such Na + 329 sources can provide both the membrane depolarization and the Na + entry known to be 330 required for Slack activation, enabling Slack to contribute both to action potential 331 repolarization during neuronal high-frequency firing 7,48 and to regulating excitatory 332 postsynaptic potential (EPSP) at post-synaptic neurons 23 . Our results support that Slack 333 and NaV1.6 form a channel complex, while also implying that NaV1.6-mediated sodium 334 influx increases the Na + concentration in the close vicinity of Slack to activate Slack. 335 As a low threshold NaV channel subtype, NaV1.6 has been reported to dominate the  Therefore, it makes sense that disruption of the Slack-NaV1.6 interaction by 381 overexpressing Slack's C-terminus reduces the current amplitudes of gain-of-function 382 Slack mutant variants (Fig. 7A,B). Our successful demonstration that viral expression 383 of Slack's C-terminus prevents epilepsy-related Slack G269S -induced seizures in mice 384 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is   GPCR Extraction Reagent was from Pierce, NP40 lysis buffer was from Beyotime, 404 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is subcloned into the modified pcDNA3.1(+) vector using Gibson assembly. All 411 mutations and chimeras of ion channels were also constructed using Gibson assembly. 412 For GST pull down assay, the segments of Slack and NaV1.6 were subcloned into 413 pCDNA3.1(+) and pGEX-4T-1 vector, respectively. For FRET experiments, mVenus-414 tag was fused to the C-terminus of NaV1.6 sequence, and mTFP1-tag was also fused to 415 the C-terminus of Slack sequence. 416

Immunoprecipitation 417
The brain tissues or HEK293T cells co-expressing full-length or fragments of Slack 418 and NaV1.6 were homogenized and lysed in GPCR Extraction Reagent (Pierce) with 419 cocktail for 30 min at 4 °C. The homogenate was centrifuged for 20 min at 16 000 g 420 and 4 °C to remove cell debris and then supernatant was incubated with 5 μg Slack 421 antibody (NeuroMab) or Nav1.6 antibody (Alomone) for 12 h at 4 °C with constant 422 rotation. 40 μl of protein G Dynabeads (Invitrogen) was then added and the incubation 423 was continued until the next day. Beads were then washed three times with NP40 lysis. 424 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is Between washes, the beads were collected by DynaMag. The remaining proteins were 425 eluted from the beads by re-suspending the beads in 1×SDS-PAGE loading buffer and 426 incubating for 30 min at 37 °C. The resultant materials from immunoprecipitation or 427 lysates were then subjected to western blot analysis. 428

Western blot analysis 429
Proteins suspended in 1×SDS-PAGE loading buffer were denatured for 30 min at 37 °C.
where Y is the value of IQuinidine/IControl, Top is the maximum response, Bottom is the 498 minimum response, X is the lg of concentration, and IC50 is the drug concentration 499 producing the half-maximum response. Significance of fitted IC50 values compared to 500 control was analyzed using extra sum-of-squares F test. 501 For data presented in fig. S3, cells were excluded from analysis if series resistance > 5 502 MΩ and series resistance compensation was set to 70%~90%. 503 The time constants (τ) of activation were fitted with a single exponential equation: 504 CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is where I is the current amplitude, t is time, offset represents the asymptote of the fit, and 506 A represents the amplitude for the activation or inactivation. 507 Steady-state fast inactivation (I-V) and conductance-voltage (G-V) relationships were 508 fitting with Boltzmann equations: 509 where I is the peak current, G is conductance, Vm is the stimulus potential, V1/2 is the 513 midpoint voltage, ENa is the equilibrium potential, and k is the slope factor. Significance 514 of fitted V1/2 compared to control was analyzed using extra sum-of-squares F test. 515 Recovery from fast inactivation data were fitted with a single exponential equation: 516 Where I is the peak current of test pulse, Imax is the peak current of first pulse, A is the 518 proportional coefficient, t is the delay time between the two pulses, and τ is the time 519 constant of recovery from fast inactivation. 520

Fluorescence Imaging and FRET Quantification 563
The spectroscopic imaging was built upon a Nikon TE2000-U microscope. The 564 excitation light was generated by an Ar laser. The fluorescent protein mVenus fused to 565 NaV1.6 and mTFP1 fused to Slack were excited by laser line at 500 and 400-440 nm, 566 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is Using this approach, the spectral and positional information are well preserved, thus 581 allowing reliable quantification of FRET efficiency specifically from the cell 582 membrane. Spectra were corrected for background light, which was estimated from the 583 blank region of the same image. 584 FRET data was quantified in two ways. First, the FRET ratio was calculated from the 585 increase in mVenus emission due to energy transfer as described in the previous study. 59 586 Briefly, mTFP1 emission was separated from mVenus emission by fitting of standard 587 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is Regression analysis was used to estimate Ab in individual cells. From each cell, the 595 FRET ratioexp was experimentally determined. The predicted Ab value was then 596 computed by adjusting two parameters, FRET Ratiomax and apparent KD. Ab was in turn 597 used to give a predicted FRET ratiopredicted. By minimizing the squared errors (FRET 598 ratioexp -FRET ratiopredicted) 2 , KD was determined. 599 Second, apparent FRET efficiency was also calculated from the enhancement of 600 mVenus fluorescence emission due to energy transfer 59-62 using a method as previously 601 described. 63 Briefly, Ratio A0 and Ratio A were measured to calculate FRET efficiency. 602 Ratio A0 represents the ratio between tetramethylrhodamine maleimide emission 603 intensities (in the absence of fluorescein maleimide) upon excitation at the donor and 604 acceptor excitation wavelengths 62-64 , and was calculated in the present study at the 605 mVenus peak emission wavelength. A particular advantage of quantifying Ratio A0 for 606 FRET measurement is that changes in fluorescence intensity caused by many 607 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is

Adeno-associated virus construction and injection 631
The adeno-associated viruses (AAVs) and the negative GFP control were from 632 Shanghai GeneChem. Co., Ltd. The full-length SlackG269S sequence (1-1238aa) was 633 ligated into modified CV232 (CAG-MCS-HA-Poly A) adeno-associated viral vector. 634 The Slack's C-terminus sequence (326-1238aa) and the negative control were ligated 635 into GV634 (CAG-MCS-3×Flag-T2A-EGFP-SV40-Poly A) adeno-associated viral 636 vector. The viruses (>10 11 TU/ml) were used in the present study. 637 For dorsal CA1 viral injection, C57BL/6J mice aged 3 weeks were anesthetized with 638 isoflurane and placed in a stereotaxic apparatus (RWD Life Science Co., Ltd.). Using a 639 5 μL micro syringe (Hamilton) with a 30 gauge needle (RWD Life Science Co., Ltd.), 640 600 nL of the viruses was delivered at 10 nL/min by a micro-syringe pump (RWD Life 641 Science Co., Ltd.) at the following site in each of the bilateral CA1 regions, using the 642 stereotaxic coordinates: 2.5 mm (anterior-posterior) from bregma, 2 mm (medio-643 lateral), ± 1.5 mm (dorsal-ventral) 65 . The syringe was left in place for 5 min after each 644 injection and withdrawn slowly. The exposed skin was closed by surgical sutures and 645 returned to home cage for recovery. All the experiments were conducted after at least 646 3 weeks of recovery. All the mice were sacrificed after experiments to confirm the 647 injection sites and the viral trans-infection effects by checking EGFP under a 648 fluorescence microscope (ZEISS LSM 510 META NLO). 649 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted March 18, 2023. ; https://doi.org/10.1101/2023.03.16.532982 doi: bioRxiv preprint . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted March 18, 2023. ; https://doi.org/10.1101/2023.03.16.532982 doi: bioRxiv preprint . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted March 18, 2023. ; https://doi.org/10.1101/2023.03.16.532982 doi: bioRxiv preprint