Scn2a cKO mice exhibit auditory hypersensitivity without changes in peripheral function.
SCN2A is identified as a high-risk gene associated with ASD, wherein abnormal sensory processing including auditory hypersensitivity is frequently observed 9, 28. In our study, mice with Scn2a haploinsufficiency (Scn2a+/– mice)24, 29 exhibited a distinct and pronounced startle response to sudden, loud auditory stimuli (Supplementary Fig. 1A). This heightened startle response is consistent with the hypersensitivity commonly observed in individuals with ASD and in ASD mouse models24, 30. Intriguingly, we also observed myelin deficits within the auditory nervous system of these Scn2a+/– mice (Supplementary Fig. 1B-C). To investigate the cell-autonomous role of oligodendroglia Scn2a in the interplay between myelination and the auditory processing disorders, we utilized Scn2a conditional knockout (cKO) mice (PdgfraCreERT; Scn2afl/fl)27. First, we examined whether oligodendroglia specific deletion of Scn2a causes auditory processing abnormalities. In an acoustic startle reflex (ASR) test, Scn2a cKO mice displayed larger startle reflexes in response to stimuli above 95 dB SPL, comparing to control (n = 15 control vs n = 10 cKO mice, two-way ANOVA test, p = 0.0012, Fig. 1A). Scn2a cKO mice showed stronger startle at loud sound stimulation of 105 dB SPL (p = 0.0214) and 115 dB SPL (p = 0.0077, multiple comparison from two-way ANOVA with Bonferroni correction). The increased amplitude of the startle reflex in Scn2a cKO mice indicates heightened sensory perception, which is considered hypersensitivity. In addition, Scn2a cKO mice exhibited alterations in the pre-pulse inhibition (PPI) test. Typically, a weaker sensory stimulus (or pre-pulse) inhibits the reaction to a subsequent strong sensory event. In control, the inhibition rate drastically increased with the strength of pre-pulse stimulation. In contrast, Scn2a cKO mice demonstrated a diminished capacity to inhibit the startle reflex, particularly in response to the pre-pulse at 81 dB SPL sound stimulation (35.9 ± 3.22% PPI in n = 18 control vs 24.7 ± 3.19% PPI in n = 11 cKO mice, p = 0.0377, multiple comparison from two-way ANOVA with Bonferroni correction, Fig. 1B). The alteration in startle reflex to sound stimulation suggests that Scn2a cKO mice have an impaired sensory gating mechanism, a feature often seen in ASD 13, 24.
To ascertain the extent of auditory processing abnormalities in the sub-cortical regions of Scn2a cKO mice, we utilized in vivo auditory brainstem responses (ABRs) to assess the sum of evoked potential responding sound stimuli along the auditory pathway, particularly in the brainstem (Fig. 1C). There was no significant difference in ABR threshold (Fig. 1D) and the peak latencies between control and Scn2a cKO mice (data not shown). However, the amplitude of ABRs in response to 80 dB SPL (click) was significantly increased in Scn2a cKO mice compared to control (p < 0.0001, two-way ANOVA). The amplitude of peak II (2.44 ± 0.166 µV, n = 39 control vs 3.30 ± 0.231 µV, n = 26 cKO, p = 0.0002, multiple comparison from two-way ANOVA with Bonferroni correction) and peak III (2.09 ± 0.129 µV in control vs 2.70 ± 0.149 µV in cKO, p = 0.0191) were significantly larger in Scn2a cKO mice compared to those of control. There was a trend toward an increase, but no significant changes were observed in peak I (p = 0.352), peak IV (p = 0.3248) and peak V (p > 0.9999, Fig. 1E). The increased amplitude of ABRs suggested that Scn2a cKO mice exhibit the hypersensitivity to sound. To evaluate whether there were alterations in central gain, we calculated the ratio between peak amplitudes II to IV relative to peak I. Notably, our analysis indicated a prevalent trend where the ratios of peak II to peak I and peak III to peak I were larger in Scn2a cKO mice, despite the absence of statistical significance (p > 0.05, two-way ANOVA, Fig. 1F). We examined the possibility if the hypersensitivity could have originated from alterations in peripheral hearing sensitivity. There was no significant change in ABR threshold, peak I amplitude, and distortion product otoacoustic emissions (DPOAE, Fig. 1G), demonstrating intact peripheral hearing. Collectively, the results demonstrated robust behavioral phenotypes in Scn2a cKO mice, specifically auditory hypersensitivity and impaired sensory gating, mirroring sensory processing abnormalities in ASD.
Myelin related genes were significantly downregulated in mature OL from Scn2a cKO mice
To understand the impact of Scn2a deletion on OL development and myelination within the auditory brainstem circuitry, we characterized the single cell transcriptional profiles of mouse auditory brainstem using single nucleus RNA sequencing (snRNA-seq). After the exclusion of low-quality cells, we obtained a total of 3083 reliable nuclei for the analysis: 989 nuclei from the control and 2094 nuclei from Scn2a cKO mice. Shared Nearest Neighbor (SNN) clustering identified 16 distinct clusters on Uniform Manifold Approximation and Projection (UMAP) plot, including clusters of neurons (Rbfox3 positive), astrocytes (Aldh1l1 positive), microglia (Itgam positive), and OL lineage cells (Pdgfra and Mog positive, Supplementary Fig. 2). Specifically, OL lineage cells consisted of five clusters: Oligodendrocyte precursor cells (OPCs), Differentiating OL1 (Diff. OL1), Differentiating OL2 (Diff. OL2), Premature OL, and Mature OL. To identify the OL lineage, we analyzed OL-specific genes, such as Pdgfra, Mbp and Mog, expressed in OPCs and mature OLs, respectively (Fig. 2A, Supplementary Fig. 2). Given the dynamic nature of OL differentiation, we utilized trajectory analysis to examine this continuous differentiation. The result of trajectory analysis from those populations displayed a narrow differentiation path connecting from OPCs to mature OLs, that similar to patterns in a previously reported study25. The result indicates that OL development follows a clearly defined trajectory, which was outlined by genetic markers specific to the OL lineage.
Although there was no statistically significant difference in the proportion of OL lineage cells between control and Scn2a cKO mice, a trend towards a reduction in the mature OL population and a slight increase in OPCs and differentiating OLs was observed in the Scn2a cKO mice (Fig. 2B). When OL clusters from each genotype were visualized with different colors in the UMAP scatter plot, a distinct difference was observed in mature OLs (Fig. 2C). To quantify these differences further, we identified genes exhibiting significant differential expression between control and Scn2a cKO mice (adjusted p with Bonferroni correction < 0.05). Notably, 96 differentially expressed genes (DEGs) were detected in mature OLs between control and Scn2a cKO mice (Fig. 2C, Table S1-5). This result reveals significant alterations in gene expression patterns within mature OLs, suggesting the cascading implications of Scn2a deletion in the orchestration of gene expression within mature OLs. Furthermore, to examine the expression level of myelin-associated genes in mature OLs, we employed AmiGo2, a gene ontology data base, to pick genes for comparison. Among the myelin-related genes (total 391 genes), 33 DEGs were identified (p < 0.05). Notably, the volcano plot showed that Mobp, encoding myelin associated oligodendrocyte basic protein (MOBP), was significantly decreased in Scn2a cKO mice (Fig. 2D). The String analysis revealed that 30 genes out of 33 DEGs have a significant interconnection and can be functionally characterized into four main categories: cell skeletal structure, cell junction, membrane, and ion channel (Fig. 2E). In addition, the most significant down regulated genes; Mobp31, Ncam132, Ptn33, Gnao134, Mtmr235, Abca236, Arhgef1037 and Mpdz38 are known to be related to myelination, cell growth, and neurological diseases (Fig. 2F). Taken together, the transcriptional profiles highlight a pronounced differential expression of myelin-associated genes in mature OLs from Scn2a cKO mice.
Loss of Scn2a impairs myelination and alters axon caliber size in the auditory brainstem during postnatal development
Following the gene expression analysis, we assessed the structural aspects of myelinated axons using transmission electron microscopy (TEM, Fig. 3A). This evaluation focused on determining myelin thickness and axon caliber size within axon bundles located in the medical nucleus of trapezoid body (MNTB) in the auditory brainstem. Ultrastructural analysis revealed significant myelin deficits in Scn2a cKO mice, as evidenced by an elevated g-ratio providing critical insights into the intricate alterations occurring at the structural level. The g-ratio is an indicator of myelin thickness measured by the inner radius (r) divided by the outer radius (R) of axon in the corresponding axon bundles. To examine alteration in distribution patter of the g-ratio, linear regression analysis was used in which slope and y-intercept were compared between groups. While the regression slopes of control and Scn2a cKO were comparable, y-intercept was significantly higher in cKO, indicating a decreased myelin thickness in Scn2a cKO (Fig. 3B). Axon calibers showed a larger inner diameter in Scn2a cKO compared to control (1.39 ± 0.03 µm, n = 1154 axons in 10 controls vs 1.48 ± 0.03 µm, n = 786 axons in 4 Scn2a cKO mice, p < 0.0001, Mann-Whitney U-test, Fig. 3C). The right shift of the cumulative frequency curve indicates an increase in axon caliber size in Scn2a cKO compared to control. Intriguingly, the g-ratio was significantly increased in Scn2a cKO, indicating a decrease in myelin thickness of axons in the MNTB area (0.82 ± 0.002 vs 0.84 ± 0.002, p < 0.0001, Mann-Whitney U-test, Fig. 3D). In addition, there was also a right shift in the cumulative frequency distribution of g-ratio in Scn2a cKO, indicating g-ratio was larger in Scn2a cKO across examined axons. The results suggest that the targeted deletion of Scn2a in immature OLs detrimentally affects myelination, concurrently altering the axon caliber size within the auditory brainstem. Notably, the myelin deficits identified in Scn2a cKO mice were consistent with significant alterations in the expression of myelin-related genes in mature OLs from snRNA-seq data (Fig. 2). For further validation, we tested effect of Scn2a deletion on myelination in another OL-specific Scn2a cKO mice, in which Scn2a was removed from Sox10 expressing OLs, targeting all OL lineage cells (Sox10CreER; Scn2afl/fl)39, also displayed similar alterations in myelin and axon caliber size (Supplementary Fig. 3). The observed myelin deficits in Scn2a cKO mice can be attributed to both impaired OL development and alterations in myelin-associated genes. A reduction in the differentiation of OPCs to OLs, evidenced by a decreased mature OLs (CC1 + OLs) in Scn2a cKO mice, further supports these findings27. Together, these results collectively underscore the pivotal role of Scn2a in myelination and the maintenance of axonal integrity in the auditory brainstem.
OL-specific loss of Scn2a alters myelinated segments and Na+ channel expression at distal axons in the MNTB
Given the significant alterations in myelination, we postulated that myelin changes might manifest in broader structural and functional aberrations within the neural circuitry. During development, immature OLs play a pivotal role by interacting with axons, which in turn determines the length of myelinated axon segments and nodes5. To elucidate the consequences of Scn2a deletion in OLs on these interactions, we examined the long-myelinated axon extending from the cochlear nucleus to the MNTB and its terminal, the calyx of Held. Myelinated axon segments, the internode, can be visualized by myelin proteolipid protein (PLP) and caspr (a paranodal marker) flanked by two sodium channel clusters (Pan Nav) denoting nodes of Ranvier (Fig. 4A). Thus, we assessed the length of the internode by measuring the distance between two Nav clusters (Fig. 4B). A structural analysis of the calyx axons showed a shorter internode adjacent to the heminode at the distal axon in Scn2a cKO mice (42.01 ± 2.31 µm, n = 24 axons from 5 control vs 32.46 ± 1.83 µm, n = 46 axons from 5 cKO mice, p = 0.0024, unpaired t-test, Fig. 4C). However, no significant difference was observed in the length of subsequent internodes (p = 0.6703, unpaired t-test, Fig. 4C). The reduced length of the last myelinated segment in Scn2a cKO suggests that the myelinated segments near the nerve terminal may be underdeveloped or experience delayed development in these mice40. Furthermore, to determine if the thinner and shorter myelin alters ion channel distribution along the distal axon in Scn2a cKO mice, we examined the patterns of voltage-gated sodium channel (Nav) expression along the calyx of Held axon. Intriguingly, we found a dispersed Nav channel expression at heminodes in Scn2a cKO mice (2.51 ± 0.11 µm, n = 35 axons from 5 control vs 3.29 ± 0.17 µm, n = 67 axons from 7 cKO mice, p = 0.0018, unpaired t-test, Fig. 4D). Dispersed Nav channels at the heminode have also been previously observed in dysmyelinated axon terminals5. Nevertheless, Nav clusters at other nodes did not show any significant alteration (p = 0.1209 for node 1 and p = 0.616 for node 2, unpaired t-test, Fig. 4D). This result highlights the pivotal role of Scn2a in OL development, emphasizing its influence on axon-OL interactions and Nav channel distribution at the distal axons in synapse-rich areas of the auditory brainstem during postnatal development.
The intrinsic excitability at the nerve terminal was altered in Scn2a cKO mice
Alterations in ion channel distribution at the heminode influence the intrinsic properties of nerve terminal and presynaptic firing pattern. We therefore examined intrinsic excitability at the calyx terminal by recording presynaptic APs, evoked by depolarizing current injection. AP waveform analysis revealed that AP threshold was significantly higher in Scn2a cKO mice (-49.0 ± 0.76 mV, n = 32 cells from 17 control mice vs -46.9 ± 0.56 mV, n = 43 from 12 cKO mice, p = 0.0217, unpaired t-test, Fig. 5A, B), with no substantial changes in AP amplitude (p = 0.0912, unpaired t-test, Fig. 5B). The maximal dV/dt was lower in Scn2a cKO mice (468.1 ± 28.47 mV/ms, n = 32 control vs 411.8 ± 30.17 mV/ms, n = 43 Scn2a cKO, p = 0.0282, Mann-Whitney U-test), indicating a slower AP rise time in Scn2a cKO mice than control. Furthermore, the minimal dV/dt was significantly reduced in Scn2a cKO mice (-359.80 ± 22.60 mV/ms, n = 28 control vs -292.9 ± 21.75 mV/ms, n = 36 Scn2a ckO mice, p = 0.0061, Mann-Whitney U-test), indicating a slower repolarization. Other parameters in Scn2a cKO, despite not reaching statistical significance, showed trends of alteration: an increased rheobase current (p = 0.058, Mann-Whitney U-test) and a broadened AP width (p = 0.1437, unpaired t-test, Fig. 5B). An elevated threshold, a slower rising of the spike, and a slower repolarizing at the calyx terminal indicate that the nerve terminal has reduced excitability. To further support these findings, we also tested the intrinsic excitability by counting the number of APs evoked by incremental current injections. The input-output curves showed a distinct shift, demonstrating fewer APs per current step in Scn2a cKO (p = 0.0176, Two-way repeated measures ANOVA, Fig. 5C). Thus, the result suggests that alterations in Nav channel expression around myelinated segments have a profound impact on presynaptic excitability in Scn2a cKO mice. Intriguingly, another observation is that in ~ 20% of presynaptic recordings from Scn2a cKO (14 of 45 cells), the calyx terminal displayed aberrant spikes or/and spontaneous spikes when the resting membrane potential was around − 65 mV (Supplementary Fig. 4). However, spontaneous firing and aberrant spikes have not been observed before in ex vivo recording from WT with the same experimental setting. The results demonstrated that myelin deficits at the distal axon caused by OL-Scn2a loss can generate asynchronous and abnormal spikes along the distal axon.
Conduction failures impaired the reliability and fidelity of high frequency spikes at the nerve terminal in Scn2a cKO mice
To understand the implications of thinner myelin sheath, larger axon, and ion channel expression alterations at the distal axon observed in Scn2a cKO, we evaluated AP propagation along the distal axon and the fidelity of APs at the nerve terminal (Fig. 6A). In control, the calyx terminal efficiently followed high-frequency stimulation and demonstrated spikes without failures at both 50 Hz and 100 Hz. Conversely, the calyx terminal in Scn2a cKO mice exhibited AP failures at 100 Hz and had drastically more failures at higher frequencies (Fig. 6B, C). Among total recorded axons in response to 200 Hz stimulation, 28% of axons in the Scn2a cKO displayed failures compared to only 18% in control mice (data not shown). Furthermore, violin plots revealed a larger proportion of axons with over 80% failure at higher frequencies (300–500 Hz) in Scn2a cKO mice while no axons in control mice showed failure rate above 80% in the same frequency range (Fig. 6C). Thus, myelin defects and disruption of Nav channel clustering caused by Scn2a loss detrimentally affect the reliability and temporal fidelity of spikes at the nerve terminal.
To discern whether AP failures at the nerve terminal are caused by conduction failures throughout the distal axon or failure to evoke AP, we tested if increasing the stimulus intensity could recover AP failures. First, we determined the minimum stimulating intensity (threshold intensity, TI) necessary to trigger a single AP for each axon. We found that threshold intensity was significantly higher in Scn2a cKO than in control (0.55 ± 0.09 V, n = 11 from 3 control mice vs 1.43 ± 0.30 V, n = 9 from 4 cKO mice, p = 0.0048, Mann-Whitney U-test, Fig. 6D). Next, we quantified failure rate in AP train (200 Hz stimulation) with incremental intensities (Fig. 6E). The calyx terminal from Scn2a cKO showed significantly higher failure rates at the TI (10 ± 6.49%, n = 11 from 3 control mice vs 47.78 ± 7.95%, n = 9 from 4 cKO mice, p = 0.0117, Two-way repeated measures ANOVA with Sidak correction, Fig. 6F) which was recovered by increased stimulating intensity. Interestingly, all axons in control with failures at TI were able to recover by TI + 20% intensity, whereas such intensity could only recover ~ 56% of axons in Scn2a cKO. Most of the axons with failures in Scn2a cKO required at least TI + 40% to recover (p = 0.0012, Two-way repeated measures ANOVA, Fig. 6F). These findings were similarly observed in another OL-specific cKO mice (Sox10CreER; Scn2afl/fl, Supplementary Fig. 5). Taken together with EM and immunostaining analysis, these results suggest that a shortened distal internode and thinner myelin critically impair axon conduction throughout the distal axon and cause AP failures at the nerve terminal.
Alterations in structural and functional properties of the nerve terminal impact synaptic transmission and short-term plasticity.
Alterations in myelination and axonal properties at the nerve terminal critically impact synaptic plasticity4, 41. To investigate the effects of myelin plasticity changes on synaptic activities at distal axons during the critical period, we evaluated synaptic transmission and short-term plasticity at the calyx of Held-MNTB synapse using whole-cell patch-clamp recordings. In the analysis of miniature excitatory postsynaptic currents (mEPSCs), we found a significant reduction in the frequency of mEPSCs in Scn2a cKO mice (p = 0.0004 Mann-Whitney U-test) without changes in amplitude (p = 0.1989, Mann-Whitney U-test, Fig. 7A). The reduction in mEPSC frequency might be associated with changes in presynaptic properties. Furthermore, we recorded evoked EPSCs triggered by afferent fiber stimulation at the midline. There was no significant difference in the amplitude of a single EPSC between groups (6.56 ± 0.340 nA, n = 25 cells from 9 control, 6.45 ± 0.612 nA, n = 14 cells from 8 cKO mice, p = 0.9174, unpaired t-test, Fig. 7A). However, the paired-pulse ratio was significantly reduced in Scn2a cKO mice (0.852 ± 0.019, n = 25 vs 0.678 ± 0.060, n = 14, p = 0.0157, Mann-Whitney U-test, Fig. 7B), indicating an alteration in release probability (Pr). To further investigate this, we analyzed EPSC trains induced by 100 Hz fiber stimulation (50 stimuli, Fig. 7C), employing three established analysis methods for the calyx synapse42. Although trends were observed in Pr and RRP size, statistical significance varied (Supplementary Fig. 6). Across the train method analyses43, the vesicle number of RRPs was significantly reduced in Scn2a cKO mice (p = 0.0241, Mann-Whitney U-test), and Pr tended to be higher, albeit not significantly (p = 0.0737, Fig. 7D). Additionally, the replenishment rate was significantly lower in Scn2a cKO mice (p = 0.0107, Mann-Whitney U-test, Fig. 7E). These changes in the number of vesicles around the active zone, including the size of the readily releasable pool, and reduced replenishment could lead to enhanced short-term depression. Taken together, OL dysfunction led to myelin alterations at the distal axon, which in turn impacted synaptic plasticity at local synapses within the auditory brainstem circuitry.