Altered cortical processing of sensory input in Huntington disease mouse models

Huntington disease (HD), a hereditary neurodegenerative disorder, manifests as progressively impaired movement and cognition. Although early abnormalities of neuronal activity in striatum are well established in HD models, there are fewer in vivo studies of the cortex. Here, we record local field potentials (LFPs) in YAC128 HD model mice versus wild-type mice. In multiple cortical areas, limb sensory stimulation evokes a greater change in LFP power in YAC128 mice. Mesoscopic imaging using voltage-sensitive dyes reveal more extensive spread of evoked sensory signals across the cortical surface in YAC128 mice. YAC128 layer 2/3 sensory cortical neurons ex vivo show increased excitatory events, which could contribute to enhanced sensory responses in vivo. Cortical LFP responses to limb stimulation, visual and auditory input are also significantly increased in zQ175 HD mice. Results presented here extend knowledge of HD beyond ex vivo studies of individual neurons to the intact cortical network.


Introduction
Circuit changes and synaptic dysfunction precede neurodegeneration in several adult onset disorders of movement and cognition, including Alzheimer, Parkinson and Huntington disease (HD; reviewed by [1][2][3][4] ). The most common inherited adult-onset neurodegenerative disorder, HD is a progressive disorder of movement, mood and cognition caused by a CAG triplet repeat expansion greater than 35 in exon 1 of the HTT gene, which encodes the protein huntingtin (HTT) 5 . The monogenic inheritance facilitates generation of mouse models with high construct and face validity 6 , and ability to identify gene-expansion carriers in the prodromal stage to enable therapeutic intervention to delay onset of clinical symptoms. Although genetic approaches to lower brain levels of HTT are under investigation in early-stage HD 7 , these may not restore synaptic and circuit function 8 . A better understanding of the mutant HTT (mHTT)-induced mechanisms underlying early alterations in synaptic and circuit function is needed to develop effective treatment for these changes The earliest neuropathological changes of HD occur in the striatum and in the motor, limbic and associative regions of cortex, which project glutamatergic afferents to the striatum (reviewed by 9,10 ), providing input to the basal ganglia-thalamic-cortical loop that selects motor actions and regulates emotional and cognitive behaviours 11 . Abnormalities in cortical-striatal communication that occur prior to neurodegeneration are well documented in mouse models of HD [10][11][12][13][14] , and may, in part, explain early motor incoordination and chorea. Notably, selective knock-down of mHTT in cortical pyramidal neurons ameliorates behavioral phenotypes and improves cortico-striatal synaptic function [15][16][17][18] . However, intra-cortical network connectivity and processing in early HD is less well studied.
Although disorders of movement, impairments in learning, and emergence of behaviors associated with depression and anxiety are the subject of numerous studies in HD mice models, less is known about sensory processing. Many patients with HD experience symptoms associated with reduced awareness of their body in space 19 , and posterior cortical regions associated with visuo-spatial processing often show early thinning on MRI 20 . Auditory processing, especially sound source localization and understanding speech in the context of ambient noise, show deficits in HD patients 21 , while older studies suggest sensory perceptual changes and abnormalities of sensory-evoked potentials 22,23 .
To begin to investigate cortical signaling networks in HD, we use sensory stimulation to evoke cortical responses measured on a mesoscale level, using electrophysiological recording from multiple brain regions as well as voltage-sensitive dye imaging in vivo, in two different HD mouse models. We find that both YAC128 and zQ175 mice demonstrate enhanced power in sensory responses compared to WT littermates, and ex vivo patch clamp recordings suggest increased NMDA receptor-mediated excitatory cortical activity contributes to this enhanced sensory response.

Results
Cortical sensory processing is not well-studied in HD mouse models; however, changes in cortical regions involved in sensory processing have been reported on structural MRI from prodromal HD gene-expansion carriers 24 . We are interested in mapping cortical network activity on a wide scale, and began by measuring the cortical response to a sensory stimulus because it is time-locked and can be analyzed with fast techniques such as in vivo electrophysiology and voltage-sensitive dye imaging to compare genotype responses with precision.

Response to limb sensory input in YAC128
To evaluate sensory input in YAC128 HD mice, local field potentials (LFP) were recorded in the primary forelimb sensory cortex (FLS1), barrel sensory cortex (BCS1) and motor cortex (M) of 6 month-old mice anesthetized with isoflurane to reduce activity related to voluntary behavior and movement artifacts. Depth of anesthesia was determined by burst suppression due to isoflurane and evaluated for each experiment to ensure equivalence between genotypes (Supplemental

VSDI response to limb sensory input in YAC128
To better determine the spatial extent of the cortex activated by limb stimulation, we used mesoscale voltage-sensitive dye (VSD) imaging through a large craniotomy exposing the cortical hemisphere contralateral to the stimulation. As previously described 25 , hindlimb stimulation resulted initially in discrete regional depolarization of primary (HLS1) and secondary (HLS2A and HLS2B) hindlimb sensory cortex in WT with a small expansion to mostly midline cortical areas ( Figure 3A). HLS1, FLS1 and secondary sensory areas were functionally determined by the center of activation and used to estimate the position of other cortical areas based on coordinates from the Allen Brain Mouse Reference Atlas. The spread of hindlimb-evoked sensory responses across the cortical surface was markedly more extensive in YAC128 mice compared to WT. In general this manifested as a large non-uniform expansion of the signal into additional areas such as primary BCS along with a longer lasting depolarization ( Figure 3A and B). Contralateral hindlimb stimulation in YAC128 mice elicited a transient wave of depolarization encompassing 21.87 ± 2.17 mm 2 (n=5) of the cortical surface (defined as pixels with a ΔF/F response at least 5x baseline RMS noise) compared to a significantly smaller 6.85 ± 2.61 mm 2 (n=4) response in WT mice (p=0.0029, t=4.473, df=7 by 2-tailed unpaired t-test).
The spatiotemporal spread of VSD signals in mesoscale imaging makes these data amenable to optical flow analysis 26 . This approach calculates velocity vector fields to quantify the speed and direction of motion 27 . In the context of our data, we can measure the trajectory, direction and speed of the spread of neural activity across the cortex, represented by changes in the brightness of pixels over time and space. We quantified VSD cortical dynamics with the Optical Flow Analysis Toolbox (OFAMM) 27 (available at http://lethbridgebraindynamics.com/ofamm/) which revealed an increase in trajectory length (HLS1: p=0.0247, K-S D=0.2198; FLS1: p=0.0238, K-S D=0.1773) and temporal speed (HLS1: p<0.0001, K-S D=0.4648; FLS1: p-0.0041, K-S D=0.2475) in YAC128 from both HLS1 and FLS1 in response to hindlimb stimulation ( Figure 3C). Secondary HL and FL areas, as well as primary and secondary barrel cortex also showed increased trajectory In contrast to hindlimb stimulation, responses to forelimb stimulation resulted in widespread cortical depolarization that did not significantly differ between genotypes: (18.78 ± 5.35 mm 2 (n=5) and 21.68 ± 1.66 mm 2 (n=4) in YAC128 and WT mice respectively; p=0.6554, t=0.4660, df=7 by 2-tailed unpaired t-test). However, HLS1, FLS1 and BCS1 depolarization in YAC128 all showed higher correlation with other cortical areas during the 500ms following hindlimb stimulation compared to WT as shown by seed-pixel correlation maps and regional correlation matrices (Supplemental Figure 4A). In addition, correlation between cortical areas following whisker stimulation was abnormally high in YAC128 compared to WT, suggesting that other types of sensory stimulation are augmented in YAC128 mice (Supplemental Figure 4C).

Synaptic events ex vivo in YAC128
In HD models, the cortex and striatum have an imbalance in excitation and inhibition with increased glutamate signaling at extrasynaptic NMDA-type glutamate receptors (NMDAR; 4,13 ).
Furthermore, HD stage-dependent changes in inhibitory and excitatory input to cortical pyramidal neurons have been reported in acute brain slice recordings from R6/2 and YAC128 HD mice 28,29 .
The augmented sensory response in 6 month-old YAC128 mice shown here could result from increased excitation or decreased inhibition in cortical circuits. To investigate these possibilities in acute brain slices, we conducted whole-cell voltage clamp experiments to measure excitatory and inhibitory synaptic responses in the sensory cortex of 6 month-old YAC128 mice. The cell capacitance and membrane resistance of layer 2/3 pyramidal neurons was similar in WT (n=24(7), capacitance 102 +/-8.07 pF; resistance 146.1 +/-18.14 MΩ) and YAC128 (n=19(6), capacitance 101.6 +/-11.13 pF, NS p=0.97, t=0.03, df=41 by unpaired t-test; resistance 188.5 +/-31.08 MΩ; NS p=0.23, t=1.23, df=41). Excitatory postsynaptic currents (EPSCs) in layer 2/3 pyramidal neurons were evoked by a short train of 10 stimulations at 20Hz with a microelectrode placed 300μm ventral. NMDAR-mediated EPSC were isolated by holding the cells at +30mV while blocking AMPAR with CNQX and GABAAR with PTX. There was no difference between genotypes in the amplitude of EPSCs evoked by 20Hz train stimulation (Supplemental Figure 5, p = 0.15 unpaired t test, t=1.11, df=9), but on some trials electrogenic events occurred that were > 5x larger in amplitude and obscured the baseline EPSC ( Figure 4A). In order to compare evoked EPSCs without large amplitude events, the lowest stimulation intensity required to evoke synaptic responses was used (not different between genotypes). However, more YAC128 neurons than WT neurons showed these large amplitude events that have previously been described as "NMDAR spikes" 30-32 (*p=0.04, Chi-square=4.295, df=1, n=10(6) YAC128 and n=11(7) WT) following minimal stimulation and they also occurred spontaneously (Figure 4 B).
In contrast to the difference in excitatory events seen in YAC128 layer 2/3 pyramidal neurons, spontaneous inhibitory postsynaptic currents (sIPSCs) frequency and amplitude were similar in these neurons from both WT and YAC128 mice recorded at +10mV in the sensory cortex ( Figure   4B). We also recorded miniature IPSCs in layer 5 pyramidal cells and found no significant difference between WT and YAC128 in either the mean frequency (WT 14.86Hz +/-1.28 vs 13.73Hz +/-1.68 YAC128, p=0.59, t=0.54, df=22, by unpaired two tailed t-test with n=13(5) and n = 11(5) (respectively) or amplitude (WT 54.13 pA +/-5.34 vs YAC128 43.67pA +/-7.34, p=0.25, t=1.176, df=22). Together, these data suggest an increase in excitatory input rather than altered inhibition to cortical layer 2/3 pyramidal neurons contributes to enhanced cortical spread of sensory responses in YAC128 mice.

Limb sensory input in zQ175
To determine if the altered sensory response observed in YAC128 mice was found in other HD mouse models, we measured LFP power in zQ175 HD mice. The level of anesthesia for these experiments was lower than that used in YAC128 mice (Supplemental Figure1

Auditory and Visual sensory input in zQ175
Unlike YAC128 mice that are on a visually impaired FVB/N background 33 , zQ175 mice on a C57/Bl6 background could be tested with a visual stimulus. As shown in Figure 6   Overall, the significantly increased sensory response in LFP power in zQ175 is not reflected in MU activity suggesting that local neuronal post-synaptic supra-threshold responses are not the primary driver of widespread cortical activation. Indeed, the similarity of responses in LFP power and VSD imaging correlation between cortical areas is consistent with increased connectivity and a more global sensory response in both YAC128 and zQ175 mice.

Discussion
Previously published studies suggest altered sensory processing in patients with Huntington disease 19,21-23 , but these changes and their underlying mechanisms have not been fully investigated. Aberrant cortical processing of sensory input could impact accuracy of movement and also impair cognition, key areas of clinical decline in patients with HD. Here we report that cortical responses to sensory stimulation, as measured by in vivo brain imaging or electrophysiological approaches, are augmented in two HD mouse models, YAC128 and zQ175, compared to WT littermates. Multiple cortical areas were depolarized in YAC128 in contrast to the discrete spatial response in WT mice shown here by VSDI. Results for hindlimb stimulation as observed by VSDI and LFP recordings were remarkably complementary, with both methods showing augmentation of the sensory response by cortical dynamics in the HD model mice.
The spread of sensory-evoked signals across the cortical surface (sensory-spread), measured with VSDI or other methods, has been documented in numerous mammalian species ranging from rodents to primates [34][35][36] . This phenomenon involves the coordinated activity of thousands of cortical neurons and is thought to be subserved by a diffuse network of inter-cortical projections, which extend radially from individual cortical columns in all directions 37,38 . The physiological role(s) of sensory-spread are incompletely understood [39][40][41] . However, the widespread signals across the cortical surface, often irrespective of function boundaries, suggests diverse roles in cortical integration. Although we have examined cortical activity exclusively in anaesthetized animals, sensory spread is observed in awake animals 26,42 .
In WT animals the extent of the sensory spread (measured with VSDI) varied considerably with the modality tested. In contrast, YAC128 mice show consistently large areas activated with the maximal hindlimb sensory-spread more than triple that of WT. Consistent with this, coherence between cortical areas was greater in YAC128 than WT mice not only following hindlimb stimulation, but also whisker stimulation. Similarly, both YAC128 and zQ175 mice had greater responses to multiple sensory modalities compared to WT in motor and sensory cortical areas, including forelimb stimulation when responses were measured by recording local field potentials, which are more sensitive to high frequency oscillations.
At the level of cortical synapses, impaired balance of inhibition and excitation could contribute to the increased sensory spread in YAC128. Sensory cortical spontaneous IPSCs are reduced in older symptomatic R6/2 mice and EPSCs are more frequent 28,43 . However, the same study showed an increase in IPSC frequency in the CAG140 and YAC128 models. Previous studies of neurons in the motor cortex showed a decrease in IPSC amplitude with an increase in frequency in 12 month-old zQ175 mice 29 . A decrease in inhibition in layer 2/3 of the motor cortex is also shown by in vivo calcium imaging and ex vivo immunohistochemistry in HD mouse models and patients 44,45 . These studies have shown conflicting results as to how inhibition is affected in the cortex in HD, and likely differ depending on the region studied. Here, in HD mice that exhibit aberrant sensory stimulus-induced activation of both the sensory and the motor cortex in vivo, our ex vivo experiments show no difference between WT and YAC128 mice in the amplitude or frequency of sIPSCs recorded from sensory cortex pyramidal neurons in layer 2/3 or of mIPSCs in layer 5. Although YAC128 mice at 6 months of age do show an HD-like phenotype and synaptic deficits, it is possible that a change in IPSCs occurs in older YAC128 mice.
Interestingly, cortical neurons in R6/2 mice show enhanced spontaneous, large amplitude "complex" events that increase in frequency and duration compared to WT as the mice age 28,29 .
Here we report that YAC128 neurons display larger amplitude events in response to stimulation of excitatory afferents than WT. Although the experimental conditions were different in the R6/2 study, it is striking that both HD models exhibit unlooked-for large amplitude excitatory events.
We and others have previously shown increased excitatory transmission and extrasynaptic NMDAR function in the striatum in HD models 12,13,[46][47][48] . The large amplitude events shown here are consistent with augmented extrasynaptic NMDAR-mediated events in the cortex of YAC128, since they increase with glutamate spillover and occur in the presence of AMPA and GABAA receptor blockers. It is possible that these events occur due to network bursting and synaptic integration 30 or by astrocytic release of glutamate following stimulation 49 . Astrocytes also contribute to hyperexcitability in the striatum in HD models, although fewer studies in HD models look at astrocytes in the cortex 50 . Future studies will investigate the mechanism of these events and their relation to sensory responses in vivo.
In addition to the increased spread of sensory-evoked VSD responses in YAC128 cortex, the optical flow analysis revealed a greater maximum temporal speed of signal propagation. Previous work has shown that sensory cortical neurons of R6/1 and R6/2 mice have increased input resistance, decreased cell capacitance, and a depolarized membrane potential at symptomatic ages 28,51 . Cortical pyramidal neurons from YAC128 and CAG140 mice also exhibit increased input resistance starting at 6 or 12 months of age, but normal resting membrane potential and cell capacitance 28 . Although those previously published data suggest that changes to membrane properties of cortical neurons in the HD brain could explain the observed increase in propagation speed, our data show no significant difference in membrane capacitance or resistance in layer 2/3 cortical pyramidal neurons from 6 month-old YAC128 vs. WT mice.
Subthreshold signals in dendritic processes, which represent the majority of the neuronal surface area, appear to predominantly mediate the sensory-spread 34,52 . Based on this, it's perhaps not surprising that the enhanced sensory-spread in zQ175 mice was not associated with increased cortical multi-unit firing probability. Regardless of the underlying mechanisms, it seems that the aberrant cortical sensory processing is more dependent on alterations to network synchrony than single-neuron bursting activity. However, recordings with higher density electrodes would be required to further examine the sensory-evoked activity of individual cortical neurons in relation to their relative anatomical positions. It is noteworthy to mention that we noticed a region-and stimulus-dependent variability in the spiking activity between animals that could result from the relative locations of cortical point spreads to the recording sites 53 .
Given that a correlated noise over large cortical areas can decrease stimulus acuity 54 , the augmented cortical sensory response in HD mice could be detrimental to performance of sensorymotor tasks and contribute to impaired motor learning. In fact, inhibition in the sensory cortex is important for hand grasping in humans 55 , and HD patients typically show deficits in reaching and grasping movements 56 . Human EEG studies show that suppression of gamma power in the visual cortex modulates reactions to sensory input 57 , suggesting that increased gamma power in other cortical areas could also impair reactions to sensory input. It is interesting to note that in YAC128 mice, hindlimb sensory responses of higher frequency LFP in the beta and gamma range were more augmented than theta frequency power. Similarly, aberrantly increased gamma oscillations in awake behaving mice have been shown in the cortex and striatum of the R6/2 mouse model of HD 43 .  59 ) and their wild-type C57/B6 littermates were implanted with electrodes as below and allowed to recover for 1 to 4 weeks before experiments.

Electrode Implant surgery
Mice were anesthetized with isoflurane at 3% for induction then reduced to 1.5 to 2 % for stereotaxic surgery. The eyes were covered with eye lubricant (Lacrilube; www.well.ca) and body temperature was maintained at 37⁰C using a heating pad with a feedback thermistor. A skin flap extending over the dorsal cortex was cut and removed. Fascia or connective tissue was lightly scraped away from the skull and small (< 1 mm diameter) holes were drilled through the skull, using a high-speed dental drill with sterile bit, over the cortex. Twisted tungsten wire tetrodes Miniature connectors (2 x 2 x 2 mm) were cemented to the skull (with dental adhesive). Ground and reference electrodes (silver wire) were fixed onto the surface of the posterior skull. Prior to implantation, tetrodes were painted with fluorescent 1,1-dioctadecyl-3,3,3,3tetramethylindocarbocyanine perchlorate (DiI, ~10% in dimethylfuran, Molecular Probes, Eugene, OR) and the solvent allowed to evaporate. For histology, immediately following the experiment, animals were decapitated and the brain fixed in 4% paraformaldehyde. The brains were sliced on a vibratome and diI labeling counter-stained with DAPI was used to identify the tetrode tract and confirm the approximate cortical location.

In vivo electrophysiology
Mice were anesthetized with 1 -2% Isoflurane and body temperature maintained at 37⁰C with For multi-unit spike analysis, the raw signals were band-pass-filtered (300 to 3000 Hz), after which spike detection was performed in MATLAB (R2019a). The threshold for spike detection was set to 3.5-fold of the SD of a two-second spike-free window of each recorded signal. As a quality control of the isolated multi-units, we inspected the shape of spike waveforms, and only the units with a clear negative deflection in the spike waveform were extracted. The probability distributions of the spike times around the stimulation (2s before to 3s after the stimulation) with a binning of 0.1 second were calculated and averaged for each experiment (10-20 trials per animal), and are represented as multi-unit probability (MU Probability).

Surgery
Six month-old YAC128 Line 53 on an FVB background and wild-type FVB mice underwent a craniotomy. Mice were anesthetized with isoflurane at 3-5% for induction and maintained at 1.0-1.5% during imaging. Mice were placed on a metal plate that could be mounted on the stage of an upright microscope and the skull was fastened to a steel plate. A 7x6 mm unilateral craniotomy (bregma 2.5 to -4.5 mm, lateral 0 to 6 mm) was made and underlying dura removed as described previously 26,27 . Body temperature was maintained at 37°C with a heating pad and feedback thermistor.

VSD Imaging
VSD imaging was performed as described previously 25,26 . Briefly, the dye RH1692 (Optical Imaging, New York, NY) was dissolved in HEPES-buffered saline solution (1 mg ml -1 ) and applied to the exposed cortex for 60-90 min for each mouse, staining neocortical layers. VSD imaging began ~30 minutes following washing unbound VSD. The brain was covered with 1.5% agarose made in HEPES-buffered saline to minimize movement artifacts, and sealed with a glass 10-45 trials of stimulus presentation were averaged to reduce the effects of on-going spontaneous activity. There was a 10 s interval between stimulation trials, and non-stimulation trials were collected and used for normalization of stimulated data.

Analysis
All VSD responses were expressed as percentage relative to baseline VSD responses ((F-F0)/F0)*100, where F0 is the baseline at the start of the trial, to reduce regional biases in VSD signal caused by uneven loading of the dye (calculated using MATLAB).
For region-based analyses, the coordinates of the hindlimb primary sensory area were determined by centering a 5x5 pixel ROI over the initial point of response to contralateral hindlimb stimulation for each animal. Similarly, forelimb sensory areas and barrel cortex sensory areas were experimentally mapped. The coordinates for other brain areas of interest were determined based on relative position to the hindlimb primary sensory area and stereotaxic coordinates as described previously 25,26 .
Sensory stimulations typically initiated waves of VSDI-measured activity which spread from modality appropriate sensory areas across the cortical surface. We first employed a threshold approach in Fiji-ImageJ to compare the extent of this sensory-mediated activity spread between genotypes. To do so, the exposed cortical surface in each ΔF/F VSDI movie was manually traced and the whole cortex area stored as a region of interest (ROI). The root mean square (RMS) noise of ΔF/F magnitude at each pixel within the ROI was measured during the initial 200 ms (baseline) of a movie prior to sensory stimulation. The peak ΔF/F value following sensory stimulation was identified at each pixel with a maximum signal projection of the entire movie and this value divided by a given pixel's baseline RMS noise. The cortical area activated following sensory stimulations was determined by pixels active over a 5x baseline RMS noise threshold and compared between animals (Similar genotype differences were seen when this threshold was varied between 3 -10x RMS noise).
Notably, the cortical area activated following sensory stimulation increased with stimulation intensity, but in the cases of forelimb and hindlimb stimulations typically plateaued at 0.5 -1.0 mA intensities. This maximum (plateau) value was specifically examined to facilitate meaningful comparisons between animals and genotypes. Although the above analysis proved useful for quantifying the spatial extent of evoked signal spread, it could be biased by differences in baseline RMS noise between animals and only examined each pixel's maximum ΔF/F value following stimulation. Therefore we also examined, and compared between genotypes, spatially averaged ΔF/F time-courses at select 5 x 5 pixel ROIs (outlined above) during plateau amplitude sensory stimulations.

Optical Flow Analysis
Optical Flow analysis of the VSD data was performed using MATLAB and Graphpad Prism. The Optical-Flow Analysis Toolbox (OFAMM) for MATLAB (http://lethbridgebraindynamics.com/ofamm/) was used to analyze the spatiotemporal dynamics of the VSD data 27 . This toolbox allows us to estimate the spatiotemporal dynamics, such as trajectory and speed, of pixels in our VSD recordings, using a variety of optical flow analysis methods. We used the Horn-Schuck (HS) method 27,61 for our optical flow analysis, and repeated select analyses with the Combined Local-Global (CLG) method, which combines the HS method with the Lucas-Kanade (LK) method 27,62,63 . The main difference between these methods is that the LK method assumes a pixel's motion is constant relative to neighboring pixels, whereas the HS method does not make this assumption 27 .
Optical flow analysis of 9x9 pixel (603x603 μm) regions of interest for hindlimb, forelimb, and barrel cortex primary and secondary sensory areas, as well as motor barrel cortex, was performed, to determine the trajectory and temporal speed of activity originating or flowing through that region in response to sensory stimulation. Regions were determined functionally through sensory stimulation or in relation to bregma.

Whole-cell voltage clamp in acute cortical slice experiments
Mice were anesthetized with isoflurane, decapitated and the brain rapidly removed. Sagittal brain slices (300µm) with sensory cortex were cut on a vibratome (VT1000 Leica) in ice cold artificial QX-314Cl) to record miniature IPSCs.

Statistics
Statistical analysis was calculated with GraphPad Prism. LFP power was compared by 2-way ANOVA of area under the curve of normalized power for 1.5 s following stimulus and Šídák's multiple comparisons test (Figures 1, 2, 5 and 6). Unpaired, two-tailed Student's t-test was used to compare the VSDI area activated with hindlimb stimulation (Figure 3B), layer 5 mIPSC frequency and amplitude (results text), layer 2/3 membrane properties, sIPSC frequency, and amplitude (results text and Figure 4B) and isoflurane burst suppression (Supplemental Figure   1B,C,E and F). Number of cells with NMDAR events were compared by Chi-square ( Figure 4A) and TBOA effects on frequency and amplitude by Kruskal-Wallis ANOVA with Dunn's multiple comparisons post hoc ( Figure 4E). P values less than 0.05 were considered significant. In compared using the two-sample Kolmogorov-Smirnov test ( Figure 3C and Supplemental Figure   4).

Data and Code availability
The data that support the findings of this study and the code used for the analysis are available from the corresponding author upon request.   Log Power (db) 11 Primary forelimb sensory cortex