NaV1.1 is essential for proprioceptive signaling and motor behaviors

The voltage-gated sodium channel (NaV), NaV1.1, is well-studied in the central nervous system; conversely, its contribution to peripheral sensory neuron function is more enigmatic. Here, we identify a new role for NaV1.1 in mammalian proprioception. RNAscope analysis and in vitro patch-clamp recordings in genetically identified mouse proprioceptors show ubiquitous channel expression and significant contributions to intrinsic excitability. Notably, genetic deletion of NaV1.1 in sensory neurons caused profound and visible motor coordination deficits in conditional knockout mice of both sexes, similar to conditional Piezo2-knockout animals, suggesting that this channel is a major contributor to sensory proprioceptive transmission. Ex vivo muscle afferent recordings from conditional knockout mice found that loss of NaV1.1 leads to inconsistent and unreliable proprioceptor firing characterized by action potential failures during static muscle stretch; conversely, afferent responses to dynamic vibrations were unaffected. This suggests that while a combination of Piezo2 and other NaV isoforms is sufficient to elicit activity in response to transient stimuli, NaV1.1 is required for transmission of receptor potentials generated during sustained muscle stretch. Impressively, recordings from afferents of heterozygous conditional knockout animals were similarly impaired, and heterozygous conditional knockout mice also exhibited motor behavioral deficits. Thus, NaV1.1 haploinsufficiency in sensory neurons impairs both proprioceptor function and motor behaviors. Importantly, human patients harboring NaV1.1 loss-of-function mutations often present with motor delays and ataxia; therefore, our data suggest that sensory neuron dysfunction contributes to the clinical manifestations of neurological disorders in which NaV1.1 function is compromised. Collectively, we present the first evidence that NaV1.1 is essential for mammalian proprioceptive signaling and behaviors.

while a combination of Piezo2 and other NaV isoforms are sufficient to elicit activity in response 48 to transient stimuli, NaV1.1 is required for transmission of receptor potentials generated during 49 sustained muscle stretch. Impressively, recordings from afferents of heterozygous conditional 50 knockout animals were similarly impaired, and heterozygous conditional knockout mice also 51 exhibited motor behavioral deficits. Thus, NaV1.1 haploinsufficiency in sensory neurons impairs 52 both proprioceptor function and motor behaviors. Importantly, human patients harboring NaV1.1 53 loss-of-function mutations often present with motor delays and ataxia; therefore, our data suggest 54 sensory neuron dysfunction contributes to the clinical manifestations of neurological disorders in 55 which NaV1.1 function is compromised. Collectively, we present the first evidence that NaV1.1 is 56 essential for mammalian proprioceptive signaling and behaviors. 112 labeling, express Scn1a transcripts (Fig 1A). RNA-sequencing datasets have consistently 113 identified NaV1.1 expression in proprioceptors; thus, we next analyzed NaV1.1 expression in these 114 cells using a Parvalbumin Cre ;Rosa26 Ai14 reporter line (Pvalb Ai14 ) and found 100% of genetically 115 identified proprioceptors were positive for Scn1a message (Fig 1B). This contrasted with low

134
In addition to NaV1.1, NaV1.6 and NaV1.7 expression has also been reported in proprioceptors 135 (Carrasco et al., 2017). As with Scn1a, mRNA for Scn8a and Scn9a is also found in 100% of genetically 136 identified proprioceptors (Fig 2A-B). Cumulative distribution plots of Scn8a and Scn9a integrated 137 fluorescence density measurements showed higher variability as compared to NaV1.1 (Fig 2C-D, Fig   138   1E). This was quantified using the coefficient of variation, a relative measure of the extent of variations 139 within data. The coefficient of variation for Scn1a transcript expression was calculated to be 75.6, 140 whereas this value increased to 97.3 and 88.1 for Scn8a and Scn9a, respectively. This indicates that 141 while all three isoforms are ubiquitously expressed in proprioceptors, the relative levels differ, with NaV1.1 142 having the most consistent level of expression across neurons analyzed. Furthermore, the average 143 integrated density of the NaV1.1 signal for a given proprioceptive DRG neuron was significantly higher 144 than both NaV1.6 and NaV1.7 (Fig 2E, p < 0.0001).

161
Collectively, these data suggest that in proprioceptors NaV1.1 is a dominant functional NaV 162 subtype.

171
Statistical significance was determined using a Kruskal-Wallis test with Dunn's post hoc comparisons. F. Top, 172 experimental workflow of serial pharmacological blockade of NaV channels expressed in proprioceptors. We first elicited 173 a whole-cell sodium current in the absence of drug. We next bath applied 500nM of ICA 121431 to block current carried 174 by NaV1.1. Subsequently, we bath applied AH-TTX (300nM) to block NaV1.6 mediated current, and PF-05089771 175 (25nM) to block the NaV1.7 mediated current. Finally, TTX (300nM) was used to block residual current and to confirm 176 there was no contribution of TTX-resistant NaVs in proprioceptors. Bottom, representative current traces following 177 application of NaV selective inhibitors. All drugs were applied for 1 minute. (G) Quantification of the average percentage 178 of the whole-cell sodium current that was sensitive to the individual drugs used (n=8). n = cells. ****p<0.0001.

187
We next determined the biophysical features of the whole-cell sodium current (INa) in 188 proprioceptors, which has not been previously reported (Fig 3A-D). The current-voltage 189 relationship shows the first detectable current appeared at voltages near -50 mV and was maximal 190 at voltages near -30 mV when evoked from a holding potential of -90 mV (Fig 3B). Voltage 191 dependence of peak conductance was best fit to a single Boltzmann function and the voltage for 192 half maximal activation was -38.7 mV (Fig 3C). The voltage-dependence of inactivation was 193 determined with 40 ms prepulse steps ranging from -120 mV to +10 mV. The midpoint of the 194 inactivation curve was -64.5 mV and was best fit to a single Boltzmann function (Fig 3D). To 195 analyze recovery from fast inactivation, TdTomato+ neurons were depolarized to -20 mV, followed 196 by a series of recovery periods ranging from 0.5 ms to 10 ms before a second test step to -20 mV 197 was given to assess sodium channel availability. INa recovery was rapid (t = 0.54 ms), with greater 198 than 50% of INa recovered after 0.5 ms (Fig 3E). Finally, entry intro slow inactivation was 199 determined. Cells were held at 0 mV during conditioning voltage steps ranging from 10 ms to 200 2000 ms, separated by two 2-ms pulses to -20 mV to compare channel availability before and after the conditioning pulse (Fig 3F). The tau for entry into slow inactivation was 928.6 ms, with 202 more than 50% of channels available after a 2000 ms conditioning pulse.  ms test pulse to -20mV from -100mV was followed by conditioning pulses at 0 mV for varying durations before a third 218 test step to -20mV. 12 ms recovery periods after the first test pulse and before the second were included to remove 219 fast inactivation. Bottom right, representative current trace elicited before and after a 2000 ms conditioning pulse. n = 220 cells.

222
We next asked how blocking NaV1.1 channels affects proprioceptor function in vitro.  found an average tau of 0.4 ms and an average current density of -96 pA/pF (Fig 4E). Of note,

229
there was a wide distribution of current densities for the ICA-sensitive component, ranging from ~-28 pA/pF to ~-263 pA/pF, suggesting some variability in the contribution of NaV1.1 to 231 proprioceptor excitability that may be proprioceptor subtype dependent. We next used current 232 clamp experiments to determine the effect of ICA on proprioceptor intrinsic excitability (Fig 4F).

233
Pharmacological inhibition of NaV1.1 significantly reduced the number of evoked action potentials 234 in most genetically identified proprioceptors (Fig 4G); however, 5 of the cells recorded had low 235 firing rates that were not inhibited by ICA. This further suggests NaV1.1 is important for repetitive 236 firing in most proprioceptors, but some subtypes with lower intrinsic excitability instead rely on a 237 combination of NaV1.6 and NaV1.7. Action potential amplitude (Fig 4H,

254
The Wilcoxon matched-pairs signed rank test was used to determine statistical significance.

255
It is important to note that ICA 121431 also blocks NaV1.3 channels, which could be

268
To clarify the importance of NaV1.1 to proprioceptor function and avoid the caveats 269 associated with in vitro pharmacological studies, we took an in vivo approach and analyzed motor 270 behaviors in Scn1a-cKO mice of both sexes. We were precluded from using a Pvalb cre driver line 271 to directly interrogate a role for NaV1.1 in proprioceptors, as loss of NaV1.1 in Pvalb-expressing 272 brain interneurons produces an epilepsy phenotype that prevents behavioral analyses in adult 273 animals (Ogiwara et al., 2007). Consistent with in vitro data, Scn1a-cKO animals of both sexes 274 displayed profound and visible motor abnormalities. These abnormalities include ataxic-like tremors when suspended in the air (Fig 5 -video 1), abnormal limb positioning (Fig 5 -videos 2-276 3), and paw clasping, which are absent in Scn1a-floxed littermate controls and heterozygous 277 animals (Pirt Cre ;Scn1a fl/+ , Scn1a-Het, respectively, Fig 5A). We first ran animals in the open field 278 test for ten minutes each to quantify spontaneous locomotor behaviors (Fig 5B). We found that 279 Scn1a-cKO animals traveled significantly less (Fig 5C) and slower (Fig 5D) than Scn1a-floxed 280 littermate controls (p = 0.0077 and 0.0057, respectively). Surprisingly, Scn1a-Het mice also 281 displayed motor abnormalities in the open field test, performing similarly to Scn1a-cKO animals 282 (Fig 5B-D), demonstrating NaV1.1 haploinsufficiency in sensory neurons for motor behaviors. No 283 genotype-dependent differences were observed in the amount of time spent moving, suggesting 284 gross motor function was intact (Fig 5E). Additionally, the amount of time spent in the center of 285 the open field chamber was also independent of genotype ( Fig 5F). We next used the rotarod 286 assay to investigate differences in motor coordination. Mice were assayed on three consecutive 287 days and latency-to-fall and RPM were quantified. Unlike in the open field assay, both Scn1a-288 floxed and Scn1a-Het mice performed at similar levels during the three-day period (Fig 5G-H).

289
Conversely, Scn1a-cKO animals performed significantly worse. By day 3, on average they were 290 only able to maintain their position on the rotarod for 41 s, falling over 50% faster Scn1a-floxed 291 and Scn1a-Het mice. We did not observe any sex dependent differences in performance in the 292 open field or rotarod tests (Fig 5 -figure supplement 3). We confirmed that our mouse model

394
We next asked whether the motor deficits of Scn1a-cKO mice are due to altered synaptic

410
Images were acquired with a 63X, 1.4 NA water-immersion objective. Sections were stained using immunochemistry with VGLUT1 (yellow) and ChAT (magenta

417
We next asked whether proprioceptor electrical signaling is altered in Scn1a-cKO mice.

418
While in vitro patch-clamp electrophysiology can assess NaV function at DRG somata and provide 419 insight as to how they contribute to intrinsic excitability, the physiological contributions of ion

426
Almost all afferents from Scn1a-floxed mice could fire consistently throughout the entire 4s hold 427 phase (Fig 8A), but loss of one or both copies of NaV1.1 led to either firing only near the beginning 428 of stretch or inconsistent firing in a high percentage of afferents lacking NaV1.1 (Fig 8B,C). We

454
We next examined the requirement of NaV1.1 for proprioceptor afferent responses to 455 sinusoidal vibration, which is a measure of dynamic sensitivity, and found no differences with loss of NaV1.1 (Fig 9A-C, Tables 2-4). We characterized a unit as having entrained to vibration if it 457 fired at approximately the same time every cycle of the 9s vibration. In most cases, afferents 458 lacking NaV1.1 were equally likely to entrain to vibration than Scn1a-floxed afferents (Fig 9D-F).

459
Indeed, Scn1a-cKO afferents were able to maintain firing during the entire 9 s sinusoidal vibration,

460
in contrast to their inability to maintain consistent firing during 4 s of static stretch. There were no 461 significant differences in firing rate during vibration between Scn1a-floxed, Scn1a-Het, and 462 Scn1a-cKO afferents (Fig 9D-F). Taken together, our ex vivo recordings suggest that behavioral

469
Representative traces from afferents that were able to entrain to a 50Hz, 100µm vibration as well as graphs with the

489
The critical role for NaV1.1 in various brain disorders has overshadowed the potential

502
pharmacological inhibition of NaV1.1 is sufficient to significantly attenuate action potential firing is most proprioceptors; yet, it should be noted that 25% of the neurons we recorded fired action 504 potentials that were insensitive to ICA application. This suggests that some proprioceptor 505 subtypes rely more heavily on NaV1.6 and NaV1.7 for electrical activity. Interestingly, the 506 proprioceptors that were insensitive to ICA application also were less intrinsic excitable, firing only  Hz, suggesting a specific role for NaV1.1 in proprioceptors responses to static muscle movement.

515
Finally, we found that loss of NaV1.1 in sensory neurons had no effect on proprioceptor

555
The loss of consistent firing we observed during static stretch in Scn1a-cKO and Scn1a-

556
Het animals is functionally similar to deletion of NaV1.1 in other brain cell types. Indeed, loss of a 557 single copy of NaV1.1 is sufficient to attenuate sustained action potential firing in parvalbumin-

567
Loss of NaV1.1 notably impacted proprioceptor afferent static sensitivity during ramp-and-568 hold stretch, but not dynamic sensitivity as measured by entrainment to sinusoidal vibrations using 569 ex vivo muscle-nerve recordings. Afferents from Scn1a-cKO animals were more likely to have action potential failures and thus were largely unable to fire consistently throughout the 4 s of 571 stretch, which was accompanied by a higher coefficient of variability in the ISI. This indicates that 572 NaV1.1 has a critical role in transmitting static stretch information to the central nervous system.

573
Interestingly, however, dynamic sensitivity in these afferents appears to be unimpaired. Both

574
Scn1a-cKO and Scn1a-Het afferents were able to entrain throughout the entire 9 s vibration; 575 therefore, NaV1.1 does appear to have a generalized role in maintaining high frequency firing, but 576 a more specific deficit in static sensitivity. NaV1.1 has been localized to muscle spindle afferent 577 endings and has been hypothesized to help amplify receptor current (Carrasco et al., 2017) . Our 578 results support a model whereby current from Piezo2 and potentially other mechanically sensitive 579 ion channels at the start of stretch produces a sufficient receptor potential to generate firing at the 580 heminode, but that amplification of the receptor potential by NaV1.1 is necessary to maintain firing 581 during held stretch.

582
A similar deficit in static but not dynamic sensitivity was seen following loss of synaptic-

585
This may indicate that glutamate plays a more general role maintaining excitability, whereas 586 NaV1.1 is required for reliable action potential generation at heminodes during static stimuli.

587
Alternatively, or in addition, NaV1.1 expressed along the axon could be essential for sustained

641
If a similar mechanism is at play for NaV1.1 in proprioceptors, reduced neurotransmitter release 642 from Scn1a-Het afferent terminals could be sufficient to produce quantifiable, albeit more subtle,

767
A series of nine 4 s ramp-and-hold stretches were given to 3 different stretch lengths repeated 3 768 times each (2.5%, 5%, and 7.5% Lo; ramp speed 40% Lo/s). A series of twelve 9 s sinusoidal 769 vibrations were given (25, 50, and 100 µm amplitude; 10, 25, 50, and 100 Hz frequency). A one-770 minute rest was given between each length change. Firing rates during a 10 s baseline before 771 stretch (resting discharge or RD) and the maximal firing rate during the ramp up phase of stretch 772 (dynamic peak or DP) were calculated for all animals. We determined whether the response to

804
Experimental design and statistical analysis. Summary data are presented as mean ± SEM, 806 from n cells or afferents, or N animals. For quantitative analysis of in situ hybridization data, at 807 least 3 biological replicates per condition were used and the investigator was blinded to genotype 808 for analysis. Behavioral experiments and analysis were also performed genotype-blind. Statistical 809 differences were determine using parametric tests for normally distributed data and non-810 parametric tests for data that did not conform to Gaussian distributions or had different variances.

811
Statistical tests are listed in Results and/or figure legends. Statistical significance in each case is 812 denoted as follows: *p <0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Statistical tests and 813 curve fits were performed using Prism 9.0 (GraphPad Software). All data generated or analyzed