Modulation of tactile feedback for the execution of dexterous movement

James M Conner; Andrew Bohannon; Masakazu Igarashi; James Taniguchi; Nicholas Baltar; Eiman Azim

doi:10.1126/science.abh1123

. Author manuscript; available in PMC: 2022 Apr 15.

Published in final edited form as: Science. 2021 Oct 14;374(6565):316–323. doi: 10.1126/science.abh1123

Modulation of tactile feedback for the execution of dexterous movement

James M Conner ¹, Andrew Bohannon ^1,^†, Masakazu Igarashi ^1,^†, James Taniguchi ¹, Nicholas Baltar ¹, Eiman Azim ^1,^*

PMCID: PMC8568303 NIHMSID: NIHMS1748970 PMID: 34648327

Abstract

While dexterity relies on the constant transmission of sensory information, unchecked feedback can be disruptive. Yet how somatosensory feedback from the hands is regulated and whether this modulation influences movement remain unclear. We found that mouse tactile afferents recruit neurons in the brainstem cuneate nucleus whose activity is modulated by distinct classes of local inhibitory neurons. Manipulation of these inhibitory circuits suppresses or enhances the transmission of tactile information, affecting manual behaviors. Top-down cortical pathways innervate cuneate in a complementary pattern, with somatosensory cortical neurons targeting the core tactile region of cuneate, and a large rostral cortical population driving feed-forward inhibition of tactile transmission through an inhibitory shell. These findings identify a circuit basis for tactile feedback modulation, enabling the effective execution of dexterous movement.

One-Sentence Summary:

Neural circuits regulate the transmission of tactile signals from the hands as they enter the brain, facilitating dexterous movement.

Much of our interaction with the world occurs through movements of the hand. This “organ of considerable virtuosity” (1) achieves its impressive dexterity through dynamic interactions between motor output and sensory feedback (2, 3). Not all feedback is treated equally, however. Signals can be disruptive to behavior when they are noisy, self-generated, or temporally delayed, implying circuit mechanisms for regulating the transmission of ascending information (3–5). Pathways responsible for feedback modulation throughout the nervous system have begun to be defined (6–12), revealing circuits that can regulate specific types of incoming signals to facilitate behavior. For somatosensation, injury of the pathways that carry ascending feedback from the limbs into the brain severely affects the smooth execution of dexterous behaviors across species, including rodents, cats, monkeys, and humans (13–15), highlighting critical roles for afferent signals from the skin and muscles in the control of movement (2). Yet, the functional organization of neural circuits that regulate the transmission of ascending somatosensory feedback and any influence this modulation has on dexterous motor output remain less clear.

The cuneate nucleus in the dorsal brainstem forms the major conduit for sensory signals from the hand ascending to the sensorimotor cortex (16, 17). Cuneate neurons receive forelimb sensory signals directly from the dorsal root ganglia, as well as indirectly through postsynaptic dorsal column pathway neurons in the cervical spinal cord (Fig. 1A, fig. S1A) (17). The core, or clusters, region of the middle cuneate (hereafter referred to as Cu) receives tactile input mainly from the distal forelimbs and is thought to specialize in processing discriminative touch signals from the hand (2, 18, 19). Cuneolemniscal neurons in Cu receive afferent input and project to several subcortical targets, most predominately the ventral posterolateral nucleus (VPL) of the thalamus, which then conveys sensory information to primary somatosensory cortex (Fig. 1A) (16, 19). The input and output connectivity and tactile specialization of this clusters region appears to be conserved across mammalian species, though the precise organization of hand somatotopy can differ (17). Cu receives descending input from corticofugal projections, and cortical stimulation experiments, primarily performed in cats and monkeys, have provided evidence for both excitation and inhibition of cuneate neurons (18–25). These findings suggest a mechanism for top-down modulation, in which the same cortical circuits that receive forelimb sensory feedback are responsible for regulating the flow of this incoming peripheral information to the brainstem (26, 27), potentially through the recruitment of local inhibitory neurons (27–30).

Fig. 1. — **(A)** Cu receives direct input from forelimb afferents in the dorsal root ganglia (DRG) and indirect input from ascending projections from the spinal cord (not shown). Cuneolemniscal (CL) neurons project to VPL thalamus, as well as other targets (17). MN, motor neuron. **(B)** Labeling direct cutaneous (left) and proprioceptive (right) projections by cholera toxin B subunit (CTB) injection into peripheral end organs. Cutaneous afferents from the glabrous pad of the hand innervate the ipsilateral core region of the middle cuneate (Cu) but avoid the external cuneate (ECu), whose major target is the cerebellum (17) (6 mice). Proprioceptive afferents from forelimb muscles (biceps and triceps brachii) rarely innervate Cu but target ipsilateral ECu (right) and other regions of the cuneate nucleus (5 mice, also see fig. S1B,C). **(C)** Retrograde labeling of CL neurons for slice recording by retrobead injection into contralateral VPL. **(D)** Resting membrane potential (RMP) of CL neurons (25 neurons in 11 mice; current-clamp, 0 pA holding current; all box-and whisker plots show median, 25^th and 75^th percentiles, and range). **(E)** Whole-cell recordings of a CL neuron (−70 mV) showing spontaneous events at baseline (black). Bath application of strychnine (blue) and bicuculline (red) progressively eliminated spontaneous inhibitory postsynaptic currents (IPSCs). **(F)** Sequential application of strychnine and bicuculline (left; 7 neurons in 4 mice; **P = 0.0015) or the reverse (right; 7 neurons in 3 mice; **P = 0.0040, *P = 0.0485) decreased the frequency of spontaneous IPSCs (Friedman repeated-measures test with Dunn’s multiple comparisons test). **(G)** Viral labeling of inhibitory cell bodies (red) and their synaptic terminals (green) in *VGAT-Cre* mice (2 mice). Inhibitory neurons (left; red cells, arrows) reside in the core and ventral shell (V shell) regions of Cu and gracile (Gr) nucleus and project into the core of Cu and Gr (middle; green), where they contact retrogradely labeled CL neurons (right; yellow). **(H)** Slice recording from CL neurons (top). Selective optogenetic activation of Cu inhibitory neurons (viral oChIEF expression in *VGAT-Cre* mice) produced large IPSCs that were eliminated by application of strychnine and bicuculline. Faint lines represent single trials, dark lines represent mean. **(I)** Cumulative reduction in the amplitude of light-evoked IPSCs (left, normalized to baseline) with an approximately equal mix of GABA and glycine components (right, error bars indicate SEM). (14 neurons in 7 mice; ***P = 0.0001; Wilcoxon two-tailed matched-pairs signed rank test; also see fig. S5C).

However, lack of circuit specific access has limited efforts to define the organization of these inhibitory neurons, the relationship they might have to descending cortical pathways, and any impact their putative feedback modulation might have on dexterous forelimb movement. It thus remains unclear which neural circuits regulate the incessant flow of sensory signals to ensure that only the appropriate and salient information is used to impact ongoing behavior.

Cuneate tactile and inhibitory inputs

We first localized the region of the middle cuneate that receives direct tactile input from the hand. Cutaneous afferents were targeted at their peripheral terminals by injection of cholera toxin B subunit (CTB) into the glabrous pad of the hand, revealing dense innervation of the ipsilateral Cu, which sends its most prominent output to contralateral VPL thalamus (Fig. 1B and fig. S1B) (16). Conversely, proprioceptive afferents were targeted by injection of CTB into either proximal or distal forelimb muscles, revealing minimal innervation of Cu, but some targeting of other cuneate regions and dense projections to the neighboring ipsilateral external cuneate nucleus (ECu), which mainly projects to the cerebellum (Fig. 1B and fig. S1B) (17). We confirmed these findings by genetically restricting our tracing to proprioceptors through conditional viral targeting of forelimb muscles in Pv-Cre mice (fig. S1C). These results establish that direct ascending cutaneous and proprioceptive afferents from the forelimb remain largely segregated at the level of the dorsal column nuclei complex in mice.

Sensory responses in Cu are subject to attenuation, potentially mediated by local inhibition (21, 27, 28, 30). Yet the detailed organization of local circuits that can affect tactile transmission, and any contribution they might have to movement are not well understood. Having localized the tactile recipient Cu region, we next sought to characterize the properties of cuneolemniscal neurons in Cu by recording from adult brainstem slices (Fig. 1C). Whole-cell recordings revealed that retrogradely labeled cuneolemniscal neurons had a relatively depolarized resting membrane potential (mean = −57.60 mV ± 1.22 mV SEM; Fig. 1D) that was on average slightly below the action potential threshold (mean = −49.67 mV ± 1.78 mV SEM; fig. S2A) (31). These neurons also received extensive spontaneous inhibitory input that was abolished by the combined application of strychnine and bicuculline (Fig. 1E,F and fig. S2B), revealing GABAergic and glycinergic inputs.

To identify the location of neurons that might provide this inhibition, we genetically restricted fluorophore expression to GABAergic and glycinergic neurons through conditional viral targeting of the Cu region of VGAT-Cre mice. We found inhibitory neurons within the core region of the Cu, where cuneolemniscal neurons reside, as well as throughout a shell region ventral to the Cu core, where cuneolemniscal neurons are absent (collectively referred to as Cu inhibitory neurons, Fig. 1G). Visualization of a synaptically tagged fluorophore revealed that axon terminals arising from these local inhibitory neurons project heavily into the Cu core region, where they provide dense synaptic input onto cuneolemniscal neurons (Fig. 1G). Genetic labeling of inhibitory subclasses revealed that glycinergic neurons are present throughout the Cu core and shell regions, whereas GABAergic neurons reside mostly in the ventral shell (fig. S3A,B). Monosynaptic retrograde rabies tracing originating specifically from Cu neurons that target VPL thalamus confirmed that both GABAergic and glycinergic neurons directly innervate cuneolemniscal neurons (fig. S3C,D).

To assess the functional connectivity of these circuits, we expressed the excitatory opsin oChIEF in Cu inhibitory neurons through conditional viral injection in VGAT-Cre mice and performed whole-cell recording from adult brainstem slices. Targeted inhibitory neurons showed robust responses to photoactivation (fig. S4A–C). These inhibitory neurons also received extensive spontaneous inhibitory inputs that were largely GABAergic (fig. S4D–F) (30), suggesting a means for local disinhibition. We next recorded postsynaptic responses from tagged cuneolemniscal neurons. Optogenetic activation of local Cu inhibitory neurons produced large inhibitory currents (Fig. 1H) with onset kinetics indicative of monosynaptic connectivity (fig. S5A,B). Combined application of strychnine and bicuculline abolished light-evoked responses (Fig. 1H,I), with each cuneolemniscal neuron showing a distinct mix of GABAergic and glycinergic inputs (Fig. 1I and fig. S5C), confirming functional connectivity between distinct classes of Cu inhibitory neurons and cuneolemniscal neurons.

Local inhibitory modulation of ascending tactile feedback

While these findings establish the anatomical and functional relationship between Cu inhibitory neurons and cuneolemniscal neurons, they do not demonstrate whether these circuits can modulate tactile signaling. We thus performed in vivo extracellular recordings from tactile-responsive neurons in the Cu of anesthetized mice while physical stimuli were applied to the glabrous pad of the ipsilateral hand by a rotating wheel (Fig. 2A), allowing us to identify neurons that are robustly sensitive to tactile input (Fig. 2B,C and fig. S6).

Fig. 2. — **(A)** *In vivo* extracellular recording in Cu of anesthetized mice with tactile stimuli applied to the ipsilateral glabrous pad of the hand. **(B)** Spike raster plots of tactile-evoked signals from an example Cu recording site across five trials with four stimuli per trial (20 total stimuli). Purple bars indicate periods when tactile stimulus was applied. **(C)** Total tactile-evoked spikes across all 20 stimuli for each recording site with minimal spontaneous activity during the inter-stimulus intervals (ISI). (51 recordings across 3 mice; ****P < 0.0001; Wilcoxon two-tailed matched-pairs signed rank test). **(D)** *In vivo* extracellular recording of tactile responsive units in Cu in anesthetized mice with optogenetic activation of local Cu inhibitory neurons (top, viral oChIEF expression in *VGAT-Cre* mice). Spike raster plots (middle) and mean spike number histograms (bottom; 0.1 sec bin) from an example recording site across ten interleaved trials (five light off, gray; five light on, blue; each with four tactile stimuli, purple bars). Lines in lower histogram indicate mean, shaded areas represent SEM. Also see fig. S6. **(E)** Mean number of spikes/sec during a single tactile stimulus (top; 0.1 sec bin) across all recordings (30 sites from 3 mice). Lines indicate mean, shaded areas represent SEM. Pairwise comparisons of the total number of spikes across five trials (20 tactile stimuli each for light off and light on) reveal suppression of tactile-evoked spikes (middle; ****P < 0.0001) and suppression of spontaneous activity during tactile ISI periods (bottom; ***P = 0.0004). (Wilcoxon two-tailed matched-pairs signed rank test). **(F)** *In vivo* extracellular recording with optogenetic inhibition of local Cu inhibitory neurons (top, viral stGtACR2 expression in *VGAT-Cre* mice). Spike raster plots from an example recording site (middle; as in (D); light off, gray; light on, red). Lines in histogram (bottom) indicate mean, shaded areas represent SEM. **(G)** Quantification (as in (E)) across all recordings (34 sites from 3 mice) shows an increase in tactile-evoked responses in Cu neurons during photoinhibition (top). Pairwise comparisons reveal an increase in spikes evoked by tactile stimuli (middle; ****P < 0.0001) and an increase in spontaneous activity during tactile ISI periods (bottom; ****P < 0.0001). (Wilcoxon two-tailed matched-pairs signed rank test).

We then asked how the transmission of tactile signals from the ipsilateral hand is affected by activation or inactivation of local Cu inhibitory circuits. First, conditional viral expression of oChIEF in VGAT-Cre mice was used to assess the impact of activating Cu inhibitory neurons on tactile-responsive neurons (Fig. 2D and fig. S6). During photostimulation, the number of tactile-evoked spikes was consistently suppressed (60.99% ± 3.71% SEM decrease in spike number), and this attenuation applied even to the rare spontaneous activity observed during tactile inter-stimulus periods (56.93% ± 15.40% SEM decrease in spike number) (Fig. 2D,E and fig. S6). Next, we used conditional viral expression of the inhibitory opsin stGtACR2 in VGAT-Cre mice to evaluate the impact of inactivating Cu inhibitory neurons (Fig. 2F). Photoinhibition of Cu inhibitory neurons amplified the number of tactile-evoked spikes (50.46% ± 8.87% SEM increase in spike number) and elicited a striking increase in inter-stimulus spiking activity (1,333.74% ± 354.81% SEM increase in spike number), suggesting that local inhibition prevents aberrant Cu neuronal firing and the transmission of spurious sensory information (Fig. 2F,G). Reasoning that this modulation might regulate signals ascending through the medial lemniscus pathway that facilitate sensory-guided behaviors, we recorded from tactile-responsive neurons in VPL thalamus while performing the same optogenetic perturbations of Cu inhibitory neurons. Activation produced an equivalent attenuation of tactile responses in VPL thalamus as it does in Cu (61.13% ± 4.64% SEM decrease in spike number) (fig S7A,B,E), while inhibition of Cu inhibitory neurons amplified responses in VPL thalamus, though to a more modest extent than in Cu (15.77% ± 5.32% SEM increase in tactile-evoked spikes; 517.22% ± 207.23% SEM increase in ISI spikes) (fig S7C–E).

Disrupting cuneate modulation perturbs dexterous movements

The regulation of feedback is a fundamental aspect of coordinated behavior (3, 5), and dexterous movements might be particularly dependent on the dynamic adjustment of sensory signaling (2). Fine movements are especially susceptible to the loss of somatosensory information (1, 13–15), but lack of temporal and circuit specificity in lesion experiments leave any role for feedback modulation unclear. Genetic access to Cu inhibitory circuits provided us the opportunity to answer a question that classical lesion studies cannot address – does the regulation of tactile feedback as it ascends into the brain have any impact on dexterous motor control?

As a first assay, mice were trained to perform a string pulling behavior that elicits the smooth alternation of left and right hands as the animal reaches, grasps, and pulls, mimicking many natural behaviors (32) (Fig. 3A and Movie S1). This assay could help to distinguish two possible scenarios: spinal tactile reflex circuits are sufficient for the smooth execution of string grasp and pull during this rhythmic behavior, or alternatively, ascending tactile signals need to be regulated appropriately in the cuneate for effective performance. To attenuate tactile signaling in the cuneate, we targeted viral delivery of oChIEF to Cu inhibitory neurons in VGAT-Cre mice (Fig. 3A). Photoactivation of these neurons affected the animals’ ability to coordinate ipsilateral grasping movements with string contact (Fig. 3B–D), indicating that spinal tactile circuits alone are not sufficient for coordinating grasping in this task. Automated tracking revealed that these prehension mistakes affected limb kinematics, causing a greater number of hand direction reversals after grasping errors and a reduction in movement path length as the animals made successive attempts to correctly time the grasp (Fig. 3D, Movies S1,S2). Mice did not miss contact with the string as the ipsilateral hand moved upward to grasp, indicating that the reach targeting component of the behavior was not affected by Cu inhibitory neuron activation (fig. S8A). Rather, we found a large increase in the latency between initial string contact and a successful grasp (Fig. 3D), and once a grasp was successful and a pull attempted, there was a decrease in the displacement of the hands as the load was transferred from the ipsilateral to contralateral hand (fig. S8A). These findings point to a primary deficit in the accurate timing of the grasp, followed by insufficient grip as the affected hand is less able to execute the downward pull before the string is transferred to the other hand. These prehension errors and kinematic deficits did not appear in the contralateral limb nor in control mice receiving photostimulation (Fig. 3B–D and fig. S8A,B). Conversely, to evaluate how an aberrant increase in tactile signaling affects behavior, we expressed the inhibitory opsin stGtACR2 in Cu inhibitory neurons of VGAT-Cre mice. During photoinhibition, in a subset of trials mice terminated or briefly paused string pulling movements at light onset (6 mice; mean 53.21% of trials ± 11.36% SEM), and in the more extreme cases, dropped the string and gripped the ipsilateral hand (Movie S3). The effects were variable, however, and in trials that continued, mice were no more likely to make string targeting or prehension mistakes during photoinhibition. These findings suggest that eliciting aberrant transmission through the cuneate, perhaps akin to paresthesia, can disrupt behavior, though mice can be resilient to spurious sensory signals when performing a reward-driven task.

Fig. 3. — **(A)** String pulling task (left). Local Cu inhibitory neurons were targeted for optogenetic activation through viral oChIEF expression in *VGAT-Cre* mice (right). **(B)** Example trajectories of ipsilateral (left) and contralateral (right) hands in horizontal (x) and vertical (y) dimensions over time. Light off (gray), photoactivation (blue). **(C)** Example y trajectories of ipsilateral (solid) and contralateral (dashed) hands. Light off (gray), photoactivation (blue). **(D)** Quantification of prehension errors and kinematics across 6 mice. Prehension errors of the ipsilateral, but not contralateral, hand increased during photoactivation (top left; % of grasp attempts in which an error was made across trials; ****P < 0.0001), as did the mean number of ipsilateral vertical (y) direction reversals (top right; mean number of direction reversals per trial; ****P < 0.0001, ***P = 0.0003). The mean y path length traversed by the ipsilateral hand between direction reversals decreased during photoactivation (bottom left; mean absolute distance between the peak and trough of a given path segment across trials; ****P < 0.0001, **P = 0.0016). There was also an increase in the mean latency between first string contact and successful grasp during photoactivation on the ipsilateral side (bottom right; ****P < 0.0001, ***P = 0.0002). (Two-way repeated measures ANOVA with Sidak multiple comparisons test). Also see fig. S8.

While the string pulling assay provides an ethologically relevant means to evaluate goal-directed limb movements, animals in this task are free to use any combination of tactile, proprioceptive, and visual feedback. We wanted to probe the tactile component of a dexterous movement more selectively. Human studies have shown that tactile acuity is surprisingly high during object manipulation when compared with psychometric detection thresholds (33), suggesting that a fine motor control assay could provide the most sensitive way to evaluate the impact of Cu feedback modulation. We therefore developed a quantitative tactile orienting assay for mice to isolate the contribution of salient cutaneous feedback to task execution (33). Head-fixed mice were trained to use their hand to turn a pedestal with an oriented texture cue to a defined target zone and hold in place (Fig. 4A–D, Materials and Methods). The pedestal was affixed to a motor-rotary encoder assembly, enabling the pedestal texture to be set to a random starting orientation at the beginning of each trial (fig. S9A). During 4-6 weeks of training, the task conditions gradually became more difficult in a closed-loop fashion, depending on each animal’s behavioral performance, until expert performance was achieved at the most stringent conditions (Fig. 4E, Movie S4, Materials and Methods).

Fig. 4. — **(A)** Head-fixed mouse performing the tactile orienting task (left). A pedestal (middle, red arrow) with parallel orientation ridges (right) was placed under the right hand. The pedestal was connected to a motor-rotary encoder assembly (also see fig. S9A). **(B)** The neutral angle of the ridges was defined as perpendicular to the body axis of the mouse (0°, dotted line), and the pedestal was passively mobile through a 180° range (blue, −30° to 150°). To receive water reward, the animal must turn the pedestal to align the ridges within a defined target range (red). **(C)** Task structure. After training, at the beginning of each trial, the pedestal is reset to −30° and then adjusted to a random starting orientation (0° ± 25°). The pedestal then becomes passively mobile, indicating the start of the trial. Water reward is delivered if the orientation of the ridges stays within the target zone (60° ± 20°) for > 0.6 sec. Otherwise, the trial is terminated without reward after 7 sec. The next trial begins after a 1 sec interval, and trials continue until either 105 rewards are delivered or 160 trials are completed. **(D)** Example of angular trajectory of pedestal during a single trial. **(E)** Example angular trajectories during training. Animals begin training with a small starting range and large target range (left). During training, the starting range gets larger and target zone gets smaller (middle). Mean success rate after training (right, 9 mice). **(F)** Ipsilateral Cu inhibitory neurons were targeted for optogenetic activation (as in Fig. 3A). Example angular trajectories (left, 70 trials) with light off (top; gray) and light on (bottom; blue), and plots (right) showing the proportion of successful trials at each time point (bars, 0.64 sec bins) and the cumulative distribution function (CDF) for only the successful trials (lines). **(G)** During optogenetic activation (blue), the overall success rate was unaffected (top), but the mean elapsed time to achieve success increased (bottom; 7 mice; **P = 0.0047; two-way mixed-effects model with Geisser-Greenhouse correction and Sidak multiple comparisons test). **(H)** Example angular trajectories (70 trials), binned successes, and CDF for a mouse performing the task with a smooth surface and the light off. **(I)** CDF across all three conditions shows an equivalent drop in performance from control conditions (texture with light off, gray line, 9 mice) when photoactivating Cu inhibitory neurons (blue line, 7 mice) or with texture removed (dashed gray line, 8 mice) (****P < 0.0001; Kruskal-Wallis test with Dunn’s multiple comparisons test). Also see fig. S9B,C.

To determine how disrupting the modulation of tactile feedback in the cuneate affects task performance, we expressed oChIEF in Cu inhibitory neurons (fig. S9B). Photoactivation did not affect overall task success in trained mice but resulted in an increase in the amount of time taken to reach the target, appearing as a rightward shift in the cumulative distribution of successes over time (Fig. 4F,G,I and fig. S9B,C). To determine how these optogenetically-induced deficits compare to normal performance when tactile information is more impoverished, we evaluated task performance using a smooth pedestal lacking tactile ridges, dissociating fine tactile cues from the cutaneous feedback used to detect contact with the pedestal. Unperturbed animals using a smooth pedestal also showed a rightward shift in the cumulative distribution of successes over time (Fig. 4H and fig. S9B,C) that was nearly identical to those using a textured platform during Cu inhibitory neuron activation (Fig. 4I). In all conditions, mice exhibited equivalent peak angular velocities when making turns, indicating their ability to contact and rotate the pedestal was not affected (fig. S9B). Yet, unperturbed mice using a textured platform were far more likely to achieve success rapidly, at the earliest stages of a trial, than mice undergoing Cu inhibitory neuron activation or mice using a smooth platform (fig. S9C). Behavioral performance was not affected in control animals receiving photostimulation (fig. S9D). These findings suggest that when tactile signals are attenuated or salient tactile cues are removed, mice can still perform the task, but they lose a tactile performance advantage and more frequently adopt incremental turn-and-wait strategies (Fig. 4F–I and Fig S9C).

Top-down corticofugal control of cuneate circuits

In a final set of experiments, we asked whether the modulatory circuits we identify can provide a basis for top-down control of tactile feedback. Ascending sensory transmission is attenuated before and during limb movement (34–36), potentially as a means to gate reafferent feedback caused by one’s own movement or increase the signal-to-noise of sensory information most relevant to task execution (5, 37). The primary somatosensory cortex (SSp) innervates the cuneate nucleus, at least in part through collaterals of corticospinal projection neurons (18, 22). Moreover, cortical activation can produce both excitatory and inhibitory effects in cat and monkey cuneate (18, 20–25), suggesting that sensory cortex provides top-down regulation of its afferent inputs. We first set out to determine whether this modulation is exclusive to SSp, or whether it might involve corticofugal neurons in other cortical regions. Using combinatorial genetic and viral tools, we broadly labeled cortical neurons that project to the cuneate, to cervical spinal cord, or to both. Corticospinal neurons throughout contralateral SSp send collateral projections to Cu, whereas corticospinal neurons in primary motor cortex (MOp) and secondary motor cortex (MOs) mostly avoid the core region of Cu (fig. S10A). Targeted anterograde labeling of corticospinal neurons revealed that while SSp densely innervates the core region of Cu, sparse MOp projections are mostly found in ventral cuneate regions (fig. S10B) (38). We also found corticofugal neurons in a broad, contralateral anterior cortical region, which we refer to as rostral sensorimotor cortex (rSM), that project to the Cu region but do not appear to project to the spinal cord (fig. S10A).

To explore the synaptic connectivity of these corticofugal projection neurons more directly, we performed monosynaptic retrograde rabies tracing (Fig. 5A). First, we found that cuneolemniscal neurons are directly targeted by corticofugal neurons in SSp as well as from the cervical dorsal root ganglia. We also found cortical inputs arising from contralateral supplemental somatosensory cortex (SSs), but sparse or absent labelling in MOp or the more anterior rSM region (Fig. 5A, left). In addition to directly exciting cuneolemniscal neurons, some descending projections might also inhibit tactile transmission by recruiting the Cu inhibitory circuits that we identified. To test this idea, we restricted rabies tracing to Cu inhibitory neurons in VGAT-Cre mice (Fig. 5A, right). We found that cortical neurons in SSp and SSs also innervate these inhibitory circuits, providing a potential explanation for previous findings that sensory cortex can both excite and inhibit cuneate neurons. Cervical dorsal root ganglia neurons also target these inhibitory neurons, suggesting that sensory pathways can drive feed-forward inhibition of tactile circuits, potentially as a means for temporal and spatial modulation (39, 40). We also found a much larger population of previously unidentified rSM neurons that directly innervate Cu inhibitory neurons, but do not target cuneolemniscal neurons (Fig. 5A).

Supporting these findings, dual anterograde labeling of corticofugal neurons in SSp and rSM in VGluT1-Cre mice showed essentially non-overlapping and complementary projection patterns; SSp heavily innervates the Cu core region, where cuneolemniscal neurons reside, while rSM innervates more ventral brainstem, including the Cu ventral shell region where inhibitory neurons are located, but completely avoids the Cu core (Fig. 5B). Additional combinatorial retrograde viral targeting from the cervical spinal cord confirmed that SSp projections to the Cu core are indeed collaterals of corticospinal projections (fig. S10C). rSM projections also descend through the pyramidal tract and decussate, but in contrast to SSp projections, instead innervate the Cu ventral shell region and do not send projections to forelimb regions of the spinal cord (fig. S10D). Finally, for a more comprehensive identification of regions that innervate Cu circuits, we used complementary rabies tracing approaches with a nuclear-localized fluorophore and automated serial two-photon tomography to quantify neuronal populations throughout the brain that innervate either cuneolemniscal or Cu inhibitory neurons. This labeling confirmed our retrograde and anterograde anatomical findings and also identified several subcortical regions as candidate modulators of signaling in the dorsal column nuclei (Tables S1,S2 and fig. S11A,B).

Corticofugal neurons in rSM project to a relatively large region of the brainstem (Fig. 5B and fig. S10D), leaving open the possibility that the inhibitory neurons they target (Fig. 5A, right) are not the same neurons that modulate tactile transmission within Cu. We therefore designed a disynaptic rabies tracing approach to determine whether rSM corticofugal neurons innervate inhibitory neurons that modulate cuneolemniscal neurons. As a control, we initiated monosynaptic rabies tracing selectively from cuneolemniscal neurons that target VPL thalamus, finding the expected inputs from SSp, cervical dorsal root ganglia, and local Cu inhibitory neurons (fig. S11C). In a separate group of mice, we again initiated monosynaptic tracing from cuneolemniscal neurons while introducing supplemental rabies virus glycoprotein (G) expression selectively in Cu inhibitory neurons in VGAT-Cre mice, thereby enabling rabies virus spread only through the inhibitory circuits that regulate the cuneolemniscal starter population. Using this approach, we found broad labeling in rSM as well as SSp, demonstrating a feed-forward inhibitory link from these corticofugal populations to cuneolemniscal neurons (Fig. 5C).

Finally, we wanted to determine whether this rSM corticocuneate projection can indeed inhibit tactile feedback within Cu. Seeking greater temporal control over tactile stimulation, and motivated by work in cats suggesting that descending modulation of tactile feedback is most apparent when inputs are not saturating (41), we developed an approach in which we could more flexibly adjust tactile stimulus timing and strength. Genetic expression of the excitatory opsin ChR2 in sensory afferents of VGluT1-Cre mice allowed us to activate cutaneous afferents by delivering brief light pulses to the glabrous pad of the hand. By adjusting the intensity of photostimulation, we could evoke near-threshold responses in Cu neurons (Fig. 5D). When the contralateral rSM was electrically stimulated just before hand afferent photoactivation, there was a consistent suppression of tactile-evoked spikes in Cu (38.56% ± 2.97% SEM reduction in spike number; Fig. 5D). Together, these results demonstrate that rSM corticofugal neurons can elicit inhibitory modulation of tactile feedback, revealing a newly defined, top-down cortical pathway for regulating ascending somatosensory information (Fig. 5E and fig. S12).

Discussion

Focusing on tactile signaling in the cuneate nucleus, we have established the functional connectivity of local GABAergic and glycinergic circuits that target cuneolemniscal neurons and enable bidirectional modulation of somatosensory feedback as it enters the brain. Perturbing the activity of these inhibitory circuits can suppress or enhance tactile responses in the cuneate, and a shift in the balance of ascending tactile transmission can severely impact the execution of behaviors that rely on sensory feedback from the hand. Distinct top-down neocortical circuits show complementary projection patterns that together provide circuit mechanisms for excitation or feed-forward inhibition of cuneolemniscal transmission to the thalamus (Fig. 5E and fig. S12). These results uncover new anatomical and functional circuit architecture for the adjustment of tactile feedback critical for the execution of dexterous movements, and provide insight into general circuit mechanisms, analogous to those identified for other sensory pathways (6–12, 42), that can attenuate disruptive feedback to facilitate successful behavior.

There are several reasons why suppression of tactile signals in the cuneate would be beneficial. Center-surround inhibition could sharpen receptive fields to augment resolution and acuity (26, 39, 40). In addition, feedback caused by movement should be attenuated to counteract disruptive feedback delays and to distinguish expected self-generated reafference from unexpected exafferent signals (5, 35, 37, 42, 43). Conversely, amplification of tactile feedback could, in principle, improve discrimination and select the inputs most relevant to a behavior (5, 24, 26, 37). Moreover, excitation could convey predictions of upcoming events during movement, preparing sensory circuits to process impending feedback through anticipatory modulation (2, 13, 18). Given the utility of bidirectional modulation, attenuation and augmentation are likely to occur simultaneously, as is seen with the coincident suppression of cutaneous and enhancement of proprioceptive feedback in the spinal cord during wrist movement in monkeys (10).

The circuits described here could provide the means to regulate distinct channels of tactile information as they ascend into the brain. The large majority of corticofugal neurons we identified target inhibitory cuneate neurons, suggesting that widespread and perhaps somewhat indiscriminate inhibition is desirable, or at least necessary, for coordinated behavior (36, 42). Indeed, broad suppression of somatosensory feedback and an increase in psychophysical detection thresholds before and during movement are well documented in several mammalian species (34–36, 43). One might speculate that preparatory activity in corticofugal neurons, including those across the large region of rostral cortex that we describe, could mediate this movement-related suppression. In contrast, direct innervation of cuneolemniscal neurons was found almost exclusively from excitatory neurons in primary somatosensory cortex, the cortical region that serves as the main recipient of cuneolemniscal signals and can directly influence motor output (44). Cortical connectivity to the cuneate might be somatotopically aligned (17, 19, 27), suggesting that in the face of broad suppression, specific cortical recipients might select and augment the feedback they receive to facilitate ongoing and future movements according to behavioral context (13, 24, 26, 45, 46). Indeed, this phenomenon of global inhibition alongside selective excitation has been observed for odor representation in rat olfactory cortex (47), and may represent a more general strategy for processing sensory stimuli. The behavioral implications of cortical inhibition and excitation of cuneate signaling remain difficult to define, in part because the corticofugal pathways we describe collateralize to many targets as they descend through the pyramidal tract (22). Reliable approaches for perturbing specific axon collaterals while leaving other targets unaffected are needed to address whether broad attenuation punctuated by selective activation characterizes tactile processing in the cuneate during movement. Moreover, top-down modulation of cuneate might have widespread effects. Cuneate neurons target many regions including the cerebellum, pontine nucleus, inferior olive, red nucleus, and superior colliculus (16, 17). Whether these distinct cuneate outputs are differentially modulated remains unknown, but this scenario could provide added flexibility, allowing cortical and subcortical areas to receive different versions of the same peripheral signals.

This cuneate circuit organization is consistent with a model in which top-down pathways convey sensory predictions of bottom-up sensory signals (48). One formulation of this predictive processing framework is that prediction error neurons sit at the interface of ascending and descending pathways and come in two flavors: positive prediction error neurons that signal unexpected inputs, and negative prediction error neurons that respond to the absence of a predicted input (49). A putative circuit implementation takes the form of a positive prediction error neuron receiving bottom-up excitation and top-down feed-forward inhibition, while a negative prediction error neuron receives the reverse – when excitation exceeds inhibition in either class of neuron, a corresponding error signal is sent to update an internal model and modify future predictions (49). These types of error predictions could emerge at any, or every, layer of the sensorimotor hierarchy. Indeed, at the first layer of tactile processing in the brain, we find bottom-up sensory and top-down cortical pathways forming all of the requisite excitatory and feed-forward inhibitory connections (Fig. 5E and fig. S12). A challenge will be to establish the resolution needed to determine whether each of the circuit elements match up in the appropriate combinations. Alongside advancing technologies for cuneate recording in behaving animals (45, 50), these approaches should help to resolve whether the connectivity and recruitment of neurons in the dorsal column pathway are reflective of hierarchical predictive processing.

Materials and Methods

Mice

Procedures performed in this study were conducted according to US National Institutes of Health guidelines for animal research and were approved by the Institutional Animal Care and Use Committee of The Salk Institute for Biological Studies. Approximately equal numbers of adult male and female mice were used for all experiments and data were combined because no sex differences were observed. Injections for tracing ascending sensory afferents from peripheral targets were performed on postnatal day (P)6-P9 pups, and tissue was collected after P56. All mice were maintained on a C57BL/6 background and housed on a 12:12 hour light cycle.

The following mouse lines were used: Wild-type (Figs. 1A–F, 2A–C, 5A, figs. S1B, S2, S8B, S9D, S10B–D, and S11C; The Jackson Laboratory and in-house colony); Avil-Cre (fig. S1A; B6.129P2-Avil^tm2(cre)Fawa/J; The Jackson Laboratory, 032536); PV-Cre (fig. S1C; B6.129P2-Pvalb^tm1(cre)Arbr/J; The Jackson Laboratory, 017320); VGAT-Cre (Figs. 1G–I, 2D–G, 3, 4, 5A,C, figs. S4, S5, S6, S7, S8A, S9B,C, S11B, Movies S1–S4, and Table S2; B6J.129S6(FVB)-^{Slc32a1tm2(cre)Lowl}/MwarJ; The Jackson Laboratory, 028862); GAD1-EGFP (52) (fig. S3A); GlyT2-EGFP (53) (fig. S3B); GAD2-FlpO (54) (fig. S3C; Hantman, Janelia Research Campus); GlyT2-FlpO (fig. S3D; Hantman, Janelia Research Campus); VGluT1-Cre (55) (Fig. 5B,D; Slc17a7-IRES-Cre; Hantman, Janelia Research Campus); VGluT2-Cre (fig. S11A and Table S1; B6J.129S6(FVB)-Slc17a6^tm2(cre)Lowl/MwarJ; The Jackson Laboratory, 028863); Rosa-LSL-tdTom (fig. S10A; Ai14; B6.Cg-Gt(ROSA)26Sor^{tm14(CAG-tdTomato)Hze}/J, The Jackson Laboratory, 007908); Rosa-FSF-tdTom (fig. S3C,D; Ai65F; B6.Cg-Gt(ROSA)26Sor^{tm65.2(CAG-tdTomato)Hze}/J; The Jackson Laboratory, 032864); Rosa-LSL-ChR2 (Fig. 5D; Ai32; B6;129S-Gt(ROSA)26Sor^{tm32(CAG-COP4*H134R/EYFP)Hze}/J, The Jackson Laboratory, 012569).

Viruses

The following adeno associated viruses (AAVs) were used, with serotype and titer (vg/ml) indicated: AAV5-EF1a-DIO-hChR2(H134R)-EYFP (fig. S1A; Penn Vector Core; 7.4 x 10¹²; Addgene plasmid #20298); AAV5-CMV-EGFP (figs. S1A, S8B, and S9D; Salk Vector Core; 2.08 x 10¹²; Addgene plasmid #32395); AAV9-FLEX-rev-ChR2-tdTomato (figs. S1C and S10B; Penn Vector Core; 1.4 x 10¹³ for muscle injections; else used at 2.8 x 10^12; Addgene plasmid #18917); AAV9-CAG-FLEX-EGFP (fig. S1C; Salk Vector Core; 1.4 x 10¹²; Addgene plasmid #51502); AAV1-hSyn-DIO-mRuby2-T2A-Synaptophysin-EGFP (56) (Fig. 1G; Vigene Biosciences; 2.5 x 10¹²; Lim, UCSD); AAV5-hSyn-DIO-oChIEF-Citrine (Figs. 1H,I, 2D,E, 3, 4, figs. S4A–C, S5, S6, S7A,B, S8A, S9B,C, and Movies S1,S2; Salk Vector Core; 2.5 x 10¹²; Addgene plasmid #50973); AAV1-hSyn-SIO-stGtACR2-FusionRed (Fig. 2F,G, fig. S7C,D, and Movie S3; Vigene Biosciences; 2.5 x 10¹²; Addgene plasmid #105677); AAV1-hSyn-DIO-TVA66T-tdTomato-CVS-N2cG (Fig. 5A, figs. S3C,D, S11A,B, and Tables S1,S2; Columbia Vector Core; 2.5 x 10¹²); AAV2-retro-EF1a-Cre (57) (Fig. 5A, figs. S3C,D and S10; Salk Vector Core; 1.0 x 10¹²; Addgene plasmid #55636); AAV1-SynP-DIO-splitTVA-EGFP-B19G (Fig. 5A; UNC Vector Core; 3.9 x 10¹²; Addgene plasmid #52473); AAV2-retro-CMV-EGFP (fig. S10A; Salk Vector Core; 3.7 x 10¹²; Addgene plasmid #32395); AAV1-EF1a-DIO-hChR2(H134R)-EYFP (Fig. 5B and fig. S10B–D; UNC Vector Core; 4.0 x 10¹² and Penn Vector Core; 4.25 x 10¹²; Addgene plasmid #20298); AAV1-EF1a-DIO-hChR2(H134R)-mCherry (Fig. 5B and fig. S10B; Salk Vector Core; 4.0 x 10¹²; Addgene plasmid #20297); AAV2-retro-Ef1a-FlpO (Fig. 5C and fig. S11C; Salk Vector Core; 2.5 x 10¹²; Addgene plasmid #55637); AAV8-CAG-fDIO-TC (TVA-mCherry) (Fig. 5C and fig. S11C; Salk Vector Core; 1.25 x 10¹²; Addgene plasmid #67827); AAV1-Syn-fDIO-N2cG-H2B-GFP (Fig. 5C and fig. S11C; Vigene Biosciences; 2.5 x 10¹²; Margrie, Sainsbury Wellcome Centre); AAV1-CAGGS-DIO-H2B-GFP-P2A-N2cG (Fig. 5C; Salk Vector Core; 2.5 x 10¹²; Addgene plasmid #73475). The following rabies viruses were used: EnvA-Rab-CVS-N2cΔG-EGFP (Fig. 5A and fig. S3C,D; Columbia Vector Core; 1.0 x 10⁹; Addgene plasmid #73461); EnvA-Rab-pSADΔG-mCherry (Fig. 5A, Salk Vector Core; 1.0 x 10⁹; Addgene plasmid #32636); EnvA-Rab-CVS-N2cΔG-H2B-EGFP (Fig. 5C, fig. S11, and Tables S1,S2; Salk Vector Core; 6.4 x 10⁹; Cetin, Allen Institute for Brain Science).

Antibodies and tracers

The following primary antibodies were used: goat anti-cholera toxin B subunit (1:2000; List Biological Laboratories, #703); rabbit anti-GFP (used for EGFP, citrine, EYFP; 1:1000; Thermo Fisher Scientific, A-11122); goat anti-GFP (used for EGFP, citrine, EYFP; 1:1000; Abcam, ab6673); rabbit Living Colors anti-DsRed (used for mCherry, tdTomato; 1:1000; Takara Bio, 632496); goat anti-RFP (used for mCherry, tdTomato; 1:1000; Sicgen, AB1140-100); rabbit anti-Fluorogold (1:500; Millipore Sigma, AB153-I). The following conjugated secondary antibodies were used at a concentration of 1:1000: donkey anti-goat-488 (Jackson ImmunoResearch Laboratories, 705-545-147); donkey anti-goat-555 (Thermo Fisher Scientific, A-21432); donkey anti-rabbit-488 (Jackson ImmunoResearch Laboratories, 711-545-152); donkey anti-rabbit-555 (Thermo Fisher Scientific, A-31572); donkey anti-rabbit-647 (Jackson ImmunoResearch Laboratories, 711-605-152); biotin-SP donkey anti-goat (Jackson ImmunoResearch Laboratories, 705-065-147); biotin-SP donkey anti-rabbit (Jackson ImmunoResearch Laboratories, 711-065-152). The following fluorophore conjugated streptavidin complexes were used: streptavidin-488 (Thermo Fisher Scientific, S11223); streptavidin-555 (Thermo Fisher Scientific, S32355); streptavidin-647 (Thermo Fisher Scientific, S21374). The following retrograde tracers were used: Fluorogold (4% solution in water; Fluorochrome); red retrobeads (Lumafluor); unconjugated cholera toxin B subunit (CTB; 1% in 0.1M phosphate buffer; List Biological Laboratories, #104). NeuroTrace 640/660 (1:100; Thermo Fisher Scientific, N21483) was used as a Nissl stain.

Immunohistochemistry and imaging

Animals used for histological purposes were perfused with 10 ml cold PBS and 20-25 ml cold 4% paraformaldehyde (PFA) in 0.1M phosphate buffer. Brains, spinal cords, and DRGs were removed and postfixed overnight at 4°C in 4% PFA in 0.1M phosphate buffer. Tissues were then transferred to a 30% sucrose solution for a minimum of 48 hours before being sectioned on a sliding microtome at 40 μm thickness. In most cases, native fluorescence was amplified by immunohistochemistry. For 2-way signal amplification, sections were washed in PBS, incubated for 20 min in PBS + 0.25% Triton X-100, and blocked in PBS + 5% donkey serum for 60 min. Sections were incubated for 1-3 days at 4°C in primary antibodies diluted in PBS + 0.25% Triton X-100 + 5% donkey serum at the concentrations listed above. Following primary antibody incubation, sections were washed 3-5 times in PBS and incubated overnight at 4°C with conjugated secondary antibodies at the concentrations listed above. When staining sections with NeuroTrace, tissue was incubated for 20 min. Sections were then washed 5 times in PBS, mounted onto glass slides, and coverslipped with Mowiol mounting media (Cold Spring Harbor Protocols). In cases where axonal fibers were being visualized or additional signal amplification was required, a modified tyramide signal amplification (TSA) protocol was used (58). For TSA amplification, sections were washed in PBS, incubated for 20 min in PBS + 0.25% Triton X-100, quenched for 30 min in PBS containing 0.6% H₂O₂, and blocked for 60 min in PBS + 5% donkey serum. Sections were then incubated for 2-3 days in primary antibody solution, which was diluted an additional 5-fold from the concentrations listed above. Sections were washed thoroughly in PBS and then incubated overnight with a solution containing biotinylated secondary antibodies at concentrations listed above. Sections were then washed and incubated sequentially with an avidin-biotin complex for 30 min (ABC kit; Vector Laboratories) and with biotinyl tyramide (58) (1:2500 dilution for 30 min). Sections were washed thoroughly and incubated overnight with conjugated streptavidin at the concentrations listed above. Sections were again washed thoroughly, mounted onto glass slides, and coverslipped with Mowiol mounting media. Processed sections were imaged on either a slide scanner (Olympus VS120) at 10X magnification or on a confocal microscope (Zeiss LSM 700; various magnifications). A motorized stage was used to facilitate the creation of montage images and z-axis image stacks. Images were post-processed in Photoshop and Illustrator (Adobe).

Surgical procedures

General:

Surgical procedures performed on adult mice were carried out under isoflurane anesthesia (1-3%). The surgical site was shaved and cleaned with betadine and alcohol and animals were given subcutaneous injections of carprofen (5 mg/kg) and bupivacaine (2 mg/kg). Throughout all surgical procedures, animals were kept on a heating pad to maintain body temperature and eye lubricant was applied. Viral injections were performed using pulled glass capillaries and a Nanoject III (Drummond Scientific) mounted to a stereotaxic manipulator. For cuneate targeting, animals were placed in a stereotaxic frame equipped with custom-made ear bars to allow the head to be dorsiflexed 30° from horizontal. From this position, an incision was made at the back of the head and the skin and overlying muscles were retracted to expose obex. In some cases, the most caudal portion of the occipital bone was removed to permit access to more rostral aspects of the cuneate nucleus. Coordinates were determined relative to obex and the dura was pierced with a 30½ gauge syringe to facilitate inserting the glass capillary. For cortical and thalamic injections, animals were placed in a stereotaxic frame in a flat skull position. A skin incision was made at the top of the head, coordinates were determined relative to bregma, and a small craniotomy was made with a dental drill in the skull overlying the injection target. Headplate fixation was performed in a similar fashion, leaving the skull intact. For spinal cord and DRG injections, animals were placed in a flat skull position and the spine was stabilized by securing the tail of the animal with a retractor positioned to produce slight tension along the antero-posterior axis. An incision was made over the cervical spinal cord, the skin and overlying muscles were retracted exposing the vertebrae, muscle was cleaned from the spinal cord using the back of a scalpel blade and fine cotton swabs, and a laminectomy of the cervical (C)6 or C7 vertebrae was performed to improve access to the spinal cord. Injections were performed by piercing the dura with a 30½ syringe to facilitate penetration of the glass capillary into the spinal cord tissue. For all surgical procedures, dorso-ventral coordinates were measured relative to the surface of the brain or spinal cord. All viruses were injected at a rate of 3-5 nl/sec. Following all surgical procedures, overlying muscle layers and skin were sutured (6-0 black braided silk, Ethicon), animals were given a subcutaneous injection of buprenex-SR (1 mg/kg), and were allowed to recover on a heating pad. All postoperative animals were housed individually until the end of the experiment.

Labeling sensory afferents:

a). DRG labeling:

To label sensory afferents through viral targeting of the DRG, adult Avil-Cre mice were anesthetized and the cervical spinal cord was exposed. After retracting overlying muscles, a laminectomy was performed to remove the C7 vertebra, extending the laminectomy laterally to partially remove the facet joint. A pulled glass capillary was loaded with AAV5-EF1a-DIO-hChR2(H134R)-EYFP and angled outward ~20°. The capillary was inserted under the remaining facet joint and guided under visual control into the DRG, and a total of 30 nl of virus was injected into the DRG. Overlying muscles and skin were sutured and animals were allowed to survive for 3-4 weeks before being perfused and processed for histological analysis. A second approach to label DRG afferents used injections of AAV5-CMV-EGFP into the DRG of adult wild-type mice. Similar results were observed with both approaches.

b). Afferent labeling:

To label sensory afferents from their peripheral targets, unconjugated CTB was injected as a tracer. To label cutaneous afferents, P6-P9 mice were anesthetized with isoflurane and 0.5-1.0 μl of 1% CTB solution was injected into the glabrous pad of the hand at a rate of 1 μl/min using a syringe pump (NE-300, New Era Pump Systems) and a Hamilton 10 μl syringe attached to a pulled glass capillary. To label proprioceptive afferents, P6-P9 pups were anesthetized with isoflurane, a small incision was made in the skin overlying the extensor digitorum lateralis, extensor digitorum communis, and extensor carpi radialis muscles (for distal injections) or the biceps and triceps brachii muscles (for proximal injections), and 0.5-1.0 μl of 1% CTB solution was injected into each muscle at a rate of 1 μl/min. Following muscle injections, the capillary was retracted and the skin incision was closed with Vetbond tissue adhesive (3M). For cutaneous and proprioceptive afferent labeling, Fast Green (1%; Thermo Fisher Scientific, ICN211922) was mixed 1:4 with the CTB to visualize the spread of the injection. Pups were placed on a heating pad until fully recovered before returning to their home cages with their mother. Animals matured to adulthood (~8 weeks) before being perfused and processed for histology.

c). Proprioceptor labeling:

To genetically restrict viral targeting to proprioceptive afferents (59, 60), P6-P9 PV-Cre pups were injected with 0.5-1.0 μl of AAV9-FLEX-rev-ChR2-tdTomato or AAV9-CAG-FLEX-EGFP. Virus was injected through a pulled glass capillary into biceps and triceps brachii (1 μl/min) following surgical procedures described above. Following injections, pups were placed on a heating pad until fully recovered before returning to their home cages with their mother. Animals matured to adulthood (~8 weeks) before being perfused and processed for histology.

Targeting local inhibitory circuits within the cuneate region:

a). Labeling local inhibitory neurons:

To characterize the location of inhibitory neurons targeting the core region of the cuneate, VGAT-Cre mice were injected with AAV1-hSyn-DIO-mRuby2-T2A-Synaptophysin-EGFP in the cuneate region at the following coordinates relative to obex: site 1: antero-posterior (A/P) 0.0 mm, medio-lateral (M/L) 0.6 mm, dorso-ventral (D/V) −0.25 mm; site 2: A/P 0.25 mm, M/L 0.8 mm, D/V −0.3 mm; site 3: A/P 0.5 mm, M/L; 1.1 mm, D/V −0.35 mm. A total of 40-50 nl of virus was injected at each site. In the same surgical session, cuneolemniscal neurons were labeled by injecting 4% Fluorogold into the contralateral ventral posterolateral (VPL) thalamus. VPL injections were made at the following locations relative to bregma: site 1: A/P −1.25 mm, M/L 1.75 mm, D/V −3.2 to −3.4 mm; site 2: A/P −1.50 mm, M/L 1.8 mm, D/V −3.3 to −3.5 mm; site 3: A/P −1.85 mm, M/L 1.9 mm, D/V −3.5 to −3.7 mm. A total of 140 nl of Fluorogold solution was injected at each site, with 70 nl injected at the most ventral D/V coordinate and 70 nl injected at the most dorsal coordinate. Animals were left for 3-4 weeks before being perfused and processed for histology.

b). Targeting local inhibitory neurons for electrophysiological recording and perturbation:

For slice whole-cell recording, in vivo extracellular recording, and behavioral experiments, inhibitory neurons in the cuneate region were targeted by injecting VGAT-Cre mice with the following viruses (40-50 nl per site at coordinates described above): AAV5-hSyn-DIO-oChIEF-Citrine and AAV1-hSyn-SIO-stGtACR2-FusionRed. For control behavioral experiments, wild-type mice were injected with AAV5-CMV-EGFP using the same approach. For behavioral experiments, a custom-made fiber optic cannula with a 200 μm core was implanted above the cuneate. Mice were anesthetized and placed in a stereotaxic frame in a flat skull position, and fibers were implanted at the following coordinates relative to lambda: A/P −3.15 mm, M/L 0.8 mm, D/V −4.88 mm. Fibers were cemented to the skull (Tetric EvoFlow, Ivoclar Vivadent) and the skin was sutured around the epoxy. Animals were left for 3-4 weeks before electrophysiological or behavioral experiments. For slice electrophysiology experiments, cuneolemniscal neurons were also labeled by injecting 50-60 nl of red retrobeads into contralateral VPL thalamus (at the sites described above) at least 1 week prior to recording.

c). Targeting GABAergic and glycinergic neurons:

GAD1-EGFP and GlyT2-EGFP mice were used to characterize the regional localization of GABAergic and glycinergic neurons within the cuneate core and ventral shell regions. In these mice, cuneolemniscal neurons were labeled by injecting 1% CTB or 4% Fluorogold into contralateral VPL thalamus (140 nl per site at coordinates described above).

Monosynaptic and disynaptic rabies tracing:

a). Labeling monosynaptic inputs to cuneolemniscal neurons:

To map inputs to cuneolemniscal neurons, a combinatorial viral strategy was used. First, AAV2-retro-EF1a-Cre was injected into contralateral VPL thalamus (50-60 nl per site at coordinates described above) of wild-type mice. Second, Cre-dependent rabies helper virus AAV1-SynP-DIO-splitTVA-EGFP-B19G was injected into the cuneate region (40-50 nl per site at coordinates described above). Animals were left for 3-4 weeks before undergoing an additional surgery where EnvA-Rab-pSADΔG-mCherry was injected into the cuneate region (40-50 nl per site at coordinates described above). Animals were left for an additional 10 days before being perfused and processed for histology.

To determine whether distinct subsets of local inhibitory neurons in the cuneate region (GABAergic and glycinergic) are monosynaptically connected to cuneolemniscal neurons, rabies tracing from cuneolemniscal neurons was carried out as described above in GAD2-FlpO x Rosa-FSF-tdTom mice or GlyT2-FlpO x Rosa-FSF-tdTom, enabling rabies labeled cells to be colocalized to fate mapped inhibitory neurons. For this purpose, the low affinity TVA rabies helper virus AAV1-hSyn-DIO-TVA66T-tdTomato-CVS-N2cG was targeted to cuneolemniscal neurons by retrograde viral infection with AAV2-retro-EF1a-Cre from the contralateral VPL thalamus. The TC66T strain of rabies helper contains a mutated low affinity TVA receptor that minimizes nonspecific expression, thus enabling reliable monosynaptic tracing of local connections (61). Animals were left for 3-4 weeks before undergoing an additional surgery where EnvA-Rab-CVS-N2cΔG-EGFP was injected into the cuneate. Animals were left for an additional 10 days before being perfused and processed for histology.

To confirm the results of these tracing experiments and to enable automated quantification through nuclear labeling, a second strategy for assessing monosynaptic inputs onto cuneolemniscal neurons was used. First, AAV1-hSyn-DIO-TVA66T-tdTomato-CVS-N2cG was injected into the cuneate region (40-50 nl per site at coordinates described above) of VGluT2-Cre mice to express rabies helper constructs only in cuneate excitatory neurons. Animals were then left for 7 weeks before undergoing a second surgery where EnvA-Rab-CVS-N2cΔG-H2B-EGFP (expressing nuclear-localized EGFP) was injected into contralateral VPL thalamus (60-70 nl per site at coordinates described above), thus targeting only the subset of cuneate excitatory neurons that give rise to cuneolemniscal projections. Animals were left for an additional 10 days before being perfused and processed for histology. Similar results were observed with both approaches. To determine the spatial localization and relative size of brain-wide cell populations providing monosynaptic inputs onto cuneolemniscal neurons, labeled brains were processed at the Whole Brain Microscopy Facility at the University of Texas Southwestern Medical Center using a TissueCyte 1000 system (TissueVision). Briefly, this system uses serial tissue sectioning and two-photon tomography to image the entire brain (62). An analysis pipeline was then used to implement supervised machine learning classifiers to generate a fluorescence probability map registered to a standardized 3-D atlas of the adult mouse brain (Common Coordinate Framework v3.0; Allen Institute for Brain Science) (63). Use of the nuclear localized fluorophore aided in segmentation by minimizing errors associated with labeled axons. Ipsilateral refers to the hemisphere containing the rabies virus starter cell population. Data collected includes: region volumes; mean raw intensity of fluorescence in each region across mice; mean intensity of fluorescence per cubic mm in each region; SEM of mean intensity of fluorescence per cubic mm in each region; the mean probability normalized to the highest whole-brain mean intensity per cubic mm; and the SEM of the mean probability normalized to the highest whole-brain mean intensity per cubic mm. Region abbreviations can be found at https://mouse.brain-map.org/static/atlas.

A third strategy for assessing monosynaptic inputs onto cuneolemniscal neurons used an approach similar to the first one but substituted AAV2-retro-Ef1a-FlpO injections into VPL thalamus and Flp-dependent rabies helper viruses (AAV8-CAG-fDIO-TC (TVA-mCherry) and AAV1-Syn-fDIO-N2cG-H2B-GFP) into the cuneate region (see details in control experiment of (c) below). All three monosynaptic tracing approaches labeled indistinguishable cortical populations.

b). Labeling monosynaptic inputs to cuneate inhibitory neurons:

To map inputs to cuneate inhibitory neurons, AAV1-hSyn-DIO-TVA66T-tdTomato-CVS-N2cG was injected into the cuneate region of VGAT-Cre mice (40-50 nl per site at coordinates described above). Animals were then left for 3-4 weeks before undergoing an additional surgery where EnvA-Rab-CVS-N2cΔG-EGFP was injected into the cuneate region (40-50 nl per site at coordinates described above). Animals were left for an additional 10 days before being perfused and processed for histology.

As above, to confirm the results of these tracing experiments and to enable automated quantification through nuclear labeling, a second strategy for assessing monosynaptic inputs onto cuneate inhibitory neurons was used. First, AAV1-hSyn-DIO-TVA66T-tdTomato-CVS-N2cG was injected into the cuneate region (40-50 nl per site at coordinates described above) of VGAT-Cre mice to express rabies helper constructs only in local inhibitory neurons. Animals were then left for 3-4 weeks before undergoing a second surgery where EnvA-Rab-CVS-N2cΔG-H2B-EGFP (expressing nuclear-localized EGFP) was injected into the cuneate region (40-50 nl per site at coordinates described above). Animals were left for an additional 10 days before being perfused and processed for histology. Similar results were observed with both approaches. Quantitative localization of brain-wide cell populations providing monosynaptic inputs onto cuneate inhibitory neurons was performed as described above.

c). Disynaptic tracing of inputs to cuneolemniscal neurons:

To determine whether the inhibitory neurons that modulate cuneolemniscal neurons are the same neuronal population targeted by rSM projections, we used a disynaptic rabies tracing approach. In this paradigm, G-deleted rabies virus tracing was initiated specifically from cuneolemniscal neurons, but local inhibitory cell populations were also selectively supplemented with G-protein (but not with TVA protein). Thus, while the starting cells for rabies virus tracing were restricted to cuneolemniscal neurons, the virus could propagate through local inhibitory cells that are synaptically connected to cuneolemniscal starter cells. First, AAV2-retro-Ef1a-FlpO was injected into the contralateral VPL thalamus (60-70 nl per site at coordinates described above) of VGAT-Cre mice. Second, Flp-dependent rabies helper viruses AAV8-CAG-fDIO-TC (TVA-mCherry) and AAV1-Syn-fDIO-N2cG-H2B-GFP were injected into the cuneate region (viruses were combined and 40-50 nl of virus solution was injected per site at coordinates described above). In the same surgery, supplemental G protein was delivered specifically to the local inhibitory neurons by injecting the Cre-dependent virus AAV1-CAGGS-DIO-H2B-GFP-P2A-N2cG into the cuneate region (40-50 nl per site at coordinates described above), enabling rabies to jump retrogradely from local inhibitory neurons that are presynaptic to targeted cuneolemniscal cells. Animals were left for 3-4 weeks before undergoing an additional surgery where EnvA-Rab-CVS-N2cΔG-H2B-EGFP (expressing nuclear-localized EGFP) was injected into contralateral VPL thalamus (60-70 nl per site at coordinates described above), selectively targeting cuneolemniscal neurons that express helper viruses. Animals were left for an additional 13 days (to allow for disynaptic spread) before being perfused for histology. We note that this approach could allow for local synaptic spread of rabies in the Cu inhibitory population, but any long-distance inputs identified will still ultimately be linked to cuneolemniscal neurons through these inhibitory circuits. As a control, the same procedures were performed in wild-type mice, omitting injection of supplemental G protein (AAV1-CAGGS-DIO-H2B-GFP-P2A-N2cG) into the cuneate region.

Labelling cortical axonal projections to the cuneate region:

a). Corticofugal projection labeling:

Initially we explored which types of corticofugal populations extend axon collateral inputs to the cuneate region. To address this question we used a dual retrograde labeling approach to separately identify corticospinal and corticocuneate populations. First, to label cervical-projecting corticospinal populations, AAV2-retro-EF1a-Cre was injected into the C6-C8 spinal segments at 5 equally spaced A/P locations in Rosa-LSL-tdTom reporter mice. At each location, the capillary was lowered to 1.5 mm ventral to the surface of the spinal cord, 25 nl of virus was injected, and the capillary was then raised in discrete 200 μm steps with 25 nl of virus being injected at each step. The most dorsal injection was made 250-300 μm ventral to the surface of the spinal cord. Second, in the same animals AAV2-retro-CMV-EGFP was injected into the cuneate region (40-50 nl per site at coordinates described above) to label all corticocuneate projections, including corticospinal collateral inputs. Animals were left for 3-4 weeks being perfused and processed for histology. A series of 40 μm slices throughout the full extent of the forebrain, spaced 120 μm apart, was immunostained and imaged.

b). SSp and MOp projection labeling:

To separately label corticospinal projection neurons in SSp and MOp that send collaterals to the cuneate region, wild-type mice were injected with AAV2-retro-EF1a-Cre into C6-C8 spinal segments, as described above. In half of these mice, AAV9-FLEX-rev-ChR2-tdTomato or AAV1-EF1a-DIO-hChR2(H134R)-mCherry was injected into the contralateral SSp and AAV1-EF1a-DIO-hChR2(H134R)-EYFP was injected into contralateral MOp. In the remaining mice the viruses were switched, with no change in results. The coordinates for each virus injection relative to bregma were as follows: SSp: A/P −0.1 / −0.2 / −0.3 mm, M/L 2.2 / 2.5 mm, D/V −0.6 mm; MOp: A/P 0.2 / 0.4 / 0.6 mm, M/L 1.0 / 1.2 mm, D/V −0.6 mm. At each site, 50 nl virus was injected. Animals were left for 3-4 weeks being perfused and processed for histology.

c). Dual labeling of SSp and rSM projections to the cuneate region:

To simultaneously visualize corticofugal projections arising from either SSp or rSM that target the cuneate region, VGluT1-Cre mice were injected with AAV1-EF1a-DIO-hChR2(H134R)-EYFP in SSp and AAV1-EF1a-DIO-hChR2(H134R)-mCherry into rSM. In some mice the viruses were switched, with no change in results. For SSp targeting, virus was injected into the cortex at 9-10 equally spaced sites spanning the following coordinates relative to bregma: A/P −1.5 → 0.5 mm, M/L 1.0 → 3.0 mm, D/V −0.4 → −0.7 mm. At each site, 50 nl of virus was injected, spread evenly across the D/V extent. For rSM targeting, virus was injected at the following 7 sites relative to bregma: site 1: A/P 1.7 mm, M/L 1.5 mm, D/V −0.8 mm; site 2: A/P 1.7 mm, M/L 2.0 mm, D/V −1.0 mm; site 3: A/P 1.7 mm, M/L 2.5 mm, D/V −0.8 to −2.25 mm; site 4: A/P 2.0 mm, M/L 1.5 mm, D/V −0.7 mm; site 5: A/P 2.0 mm, M/L 2.1 mm, D/V −0.7 to −2.25 mm; site 6: A/P 2.25 mm, M/L 1.5 mm, D/V −0.9; site 7: A/P 2.25 mm, M/L 2.0 mm, D/V −0.8 to −1.8 mm. At each site, the pipette was lowered to its most ventral location and 50 nl of virus was injected every 200 μm as the capillary was withdrawn along the D/V axis. In the same surgical session, cuneolemniscal neurons were also labeled by injecting 4% Fluorogold into contralateral VPL thalamus, as described above. Animals were left for 3-4 weeks being perfused and processed for histology.

d). Selective labeling of SSp corticospinal collaterals to the cuneate:

To label collaterals arising from corticospinal neurons in SSp, a combinatorial viral strategy was used. First, AAV2-retro-EF1a-Cre was injected into C6-C8 spinal segments, as described above. Second, in the same surgical session, AAV1-EF1a-DIO-hChR2(H134R)-EYFP was injected into contralateral SSp, as described above. Cuneolemniscal neurons were also labeled by injecting 4% Fluorogold into contralateral VPL thalamus, as described above. Animals were left for 3-4 weeks being perfused and processed for histology.

e). Selective labeling of rSM corticofugal collaterals to the cuneate region:

To label collaterals arising from corticofugal neurons in rSM, a combinatorial viral strategy was used. First, AAV2-retro-EF1a-Cre was injected into the cuneate region, as described above. Second, in the same surgical session, AAV1-EF1a-DIO-hChR2(H134R)-EYFP was injected into the contralateral rSM, as described above. In some cases, cuneolemniscal neurons were also labeled by injecting 4% Fluorogold into contralateral VPL thalamus, as described above. Animals were left for 3-4 weeks being perfused and processed for histology.

Slice electrophysiology

Slice preparation:

Acute brainstem slices containing the cuneate nucleus were prepared from adult (~8-12 week old) mice. In cases where cuneolemniscal neurons were targeted for whole-cell recordings, animals were injected with red retrobeads into contralateral VPL thalamus at least 1 week prior to recording, as described above. Brain slices were collected, as previously described (64). Briefly, animals were deeply anaesthetized and decapitated, and the brain was quickly removed and immediately placed in ice-cold cutting solution consisting of (in mM): 92 N-methyl-d-glucamine, 30 NaHCO3, 25 D-glucose, 20 HEPES, 1.25 KH2PO4, 5.0 ascorbate, 3.0 pyruvate, 2.0 thiourea, 2.5 KCl, 3.5 MgCl2, and 0.5 CaCl2. Coronal brain slices (250 μm thick) were sectioned using a vibratome (VT1000 S, Leica), and slices were transferred to cutting solution at 37°C for 5-10 min for recovery. Until recording, slices were stored in room temperature holding solution containing (in mM): 92 NaCl, 30 NaHCO3, 25 D-glucose, 20 HEPES, 1.25 KH2PO4, 5.0 ascorbate, 3.0 pyruvate, 2.0 thiourea, 2.5 KCl, 2.0 MgCl2, and 2.0 CaCl2. For recording, slices were transferred to room temperature artificial cerebrospinal fluid (aCSF) containing (in mM): 125 NaCl, 26 NaHCO3, 10 D-glucose, 1.25 NaH2PO4, 3.5 KCl, 2.0 CaCl2, and 2.0 MgCl2. All solutions were continuously oxygenated (95% O2, 5% CO2) at pH 7.4-7.5, with an osmolarity of 290-300 mOsm, adjusted with sucrose as necessary.

Whole-cell recording:

Individual slices were transferred to a submerged recording chamber mounted on the stage of a fixed-stage microscope (BX51WI, Olympus) equipped with differential interference contrast optics, a 40X water immersion lens, and infrared illumination to view neurons in the slices. The recording chamber was continuously perfused with oxygenated aCSF at room temperature (3 ml/min). Borosilicate patch electrodes were controlled by a motorized micromanipulator (MP-225, Sutter Instrument Company) that had an open tip resistance of 3-6 MΩ when filled with intracellular solution.

For experiments investigating inhibitory activity, the reversal potential of chloride was shifted to ~0 mV using an intracellular solution containing (in mM): 135 KCl, 10 Hepes, 10 creatine-PO4, 2 Mg-ATP, 0.2 Na-GTP, and 0.5 EGTA, with an adjusted pH of 7.3 and osmolarity of 280-290 mOsm, causing IPSCs to appear as inward currents. For all other experiments, the internal solution contained (in mM): 125 K-gluconate, 10 KCl, 10 Hepes, 10 creatine-PO4, 2 Mg-ATP, 0.2 Na-GTP, and 0.5 EGTA, with an adjusted pH of 7.3 and osmolarity of 280-290 mOsm. Tight seals of 1 GΩ or greater were obtained under visual guidance before breaking into whole-cell mode. Neurons expressing fluorescent markers were selected for patching based on their anatomical location. Neuronal health and patch quality were evaluated based on resting membrane potential, input resistance, and responses to depolarizing and hyperpolarizing pulses.

Data acquisition and analysis:

All results were obtained from cells recorded at least 3 weeks post-virus injection. Whole-cell recordings were obtained using a Multiclamp 700B amplifier (Molecular Devices). Signals were acquired using Clampex software with a Digidata 1550B interface (Molecular Devices). Evoked responses were digitized at 10 kHz, filtered at 2 kHz, and analyzed using Clampfit (version 10.0, Molecular Devices).

Spontaneous events were identified using the event detection feature of Clampfit. Spontaneous event frequency was calculated cumulatively as the total number of events detected over the entire recording period. The instantaneous inter-event interval for spontaneous events was defined as the duration between the onset of consecutive events. To generate the cumulative distribution function of inter-event intervals, all recorded events from all cells were pooled and sorted by increasing duration (x-axis). Interval durations were plotted against the probability (y-axis) that a given inter-event interval would have a nominal value at or below that duration. For evoked events, event latency was defined as the duration from stimulus onset to event onset and averaged over a minimum of 10 trials for each cell. Event jitter was defined as the standard deviation of trial-to-trial event latencies and calculated from a minimum of 10 trials for each cell.

Intrinsic cell properties, including resting membrane potential, input resistance, and action potential threshold, were calculated from traces generated under current-clamp configuration. Resting membrane potential was calculated from at least 5 sweeps as the average membrane potential over a 100 msec period with a 0 pA holding current. Input resistance was calculated with Ohm’s law, using the amplitude of the change in membrane potential in response to a 10-100 pA hyperpolarizing current. Action potential threshold was determined through phase plot of the rate of change of membrane potential versus membrane potential. Depolarizing current was injected incrementally in 10-50 pA steps, and the first evoked action potential was used to calculate action potential kinetics. Synaptic currents were recorded under voltage-clamp configuration to allow for evaluation of kinetics with minimal contamination due to spontaneous fluctuations in membrane potential. The responses of opsin-expressing cells to light pulses were recorded under both current-clamp and voltage-clamp configuration to confirm the presence of light-evoked action potentials and quantify photocurrent amplitudes, respectively.

Photostimulation:

Photostimulation of opsins was achieved through full-field illumination of the tissue via fluorescent light (X-Cite Turbo; Excelitas Technologies) passed through the microscope objective. Light was passed through a YFP filter (YFP-2427B-000, Semrock; excitation: 500/24-25, emission: 542/27-25) for activation of oChIEF. Unless otherwise noted, a 100 msec light pulse was used for activation with pulse timing triggered by Clampex software.

Drugs and drug application:

CNQX (10 μM, Abcam) and D-APV (20 μM, Abcam) were used to block AMPA and NMDA receptors, respectively, when quantifying inhibitory inputs. Bicuculline methiodide (10 μM, Abcam) was used to block GABA_A-mediated synaptic transmission. Strychnine (10 μM, Tocris) was used to block glycine-mediated transmission. All drugs were bath-applied for at least 10 min before assessing post-application responses, with each neuron serving as its own control.

In vivo electrophysiology

Animal preparation:

Adult mice were anesthetized with isoflurane and placed in a stereotaxic frame (Kopf Instruments) in a flat skull position. The skin and overlying fascia were cut and retracted and the skull was cleaned with cotton swabs and dried. A small burr hole was drilled above the cerebellar vermis and a 127 μm silver wire (A-M Systems) was lowered ~1 mm below the surface of the brain as the reference electrode for differential recordings. For optogenetic experiments, a second burr hole was drilled A/P −3.15 mm and M/L −0.8 mm relative to lambda and a custom made fiber optic cannula with a 200 μm core was lowered 4.88 mm ventral from the skull surface at lambda using a stereotaxic manipulator and cemented in place with epoxy (Tetric EvoFlow, Ivoclar Vivadent) along with the reference electrode. Next, a headplate was affixed to the skull immediately above bregma with epoxy.

For cuneate recordings, the animal was then repositioned in modified ear bars with the head dorsiflexed ~30° from horizontal and obex and the dorsal brainstem were exposed, as described above. The clusters region of the cuneate is even with or slightly rostral to obex and is usually obstructed by the cerebellum. Therefore, for recording experiments the caudal most aspect of the cerebellum was removed by aspiration to improve access to the cuneate, consistent with procedures performed in rats and cats (38, 41, 65, 66). For VPL thalamus recordings, a similar procedure was followed. With the animal in a flat-skull position in the stereotaxic frame, a modified headplate was affixed to the skull with epoxy such that bregma remained exposed. Next, a craniotomy was performed over an area covering A/P −2.0 to 0.5 mm and M/L 1.5 to 2.5 mm (relative to bregma) to permit access to VPL, and burr holes were drilled over the cerebellar vermis for the reference electrode and another for the optical fiber, as above. The animal was then removed from isoflurane, administered a cocktail containing 1.2 mg/kg urethane and 20 mg/kg xylazine (67), and transferred to a recording rig where the headplate was clamped with the head dorsiflexed and the cuneate accessible (for Cu recordings) or with the head in flat skull position (for VPL recordings). Animals were slightly suspended so that their forelimbs were held in a fixed position above the base of the recording platform. The preamplifier was grounded through an insulated copper strip lined with saline soaked gauze that was wrapped around the tail. For cuneate recordings, the dura and pia overlying the cuneate were removed and the surface of the brainstem was kept moist with a layer of sterile saline or paraffin oil placed in the space created by retracting the muscles. For thalamic recordings, the exposed cortical surface was kept moist by a layer of sterile saline or paraffin oil placed in a well created by the epoxy for the headplate and reference electrode. Throughout recordings, animals were kept on a DC-powered heating pad to maintain stable body temperature, and supplemental doses of urethane were given to maintain animals in an areflexic state, assessed by foot pinch.

Tactile stimuli:

For most experiments, tactile stimuli were applied with 4 small wooden sticks (2.5 mm diameter) that were oriented 90° from one another and rotated across the surface of the pad of the ipsilateral hand using a stepper motor (28BYJ-48m, MikroElektronika) controlled by a microcontroller board (Arduino UNO, Arduino). The board was triggered by a TTL pulse from a Cerebus neural signal processor (Blackrock Microsystems) and was programmed to rotate the stick-wheel at a rate of 60° per sec. There was a 7 sec interval between each complete revolution (1 revolution = 4 tactile stimuli). The duration of each individual stimulus was ~0.7-0.9 sec and the duration of a full revolution was 6 sec. For each recording site, a total of 20 tactile stimuli were applied (5 revolutions). For optogenetic experiments, 20 light off and 20 light on stimuli were delivered (5 revolutions each), interleaving conditions for each full rotation.

Cuneate and thalamus recording:

Extracellular recordings were performed using carbon fiber electrodes (~0.8 MOhm, Carbostar-1, Kation Scientific) mounted to a stereotaxic manipulator (Kopf Instruments) and connected to a Cereplex M headstage (Blackrock Microsystems), which interfaced with the Cerebus neural signal processor. Signals were sampled at 30 kHz and band-pass filtered between 250-5000 Hz. Tactile responsive neurons in the cuneate were targeted by placing the recording electrode at the following coordinates relative to obex: M/L 0.6 to 0.8 mm, A/P 0.0 to 0.3 mm. The electrode was then slowly lowered as tactile inputs were delivered to the ipsilateral hand and responses were monitored on an oscilloscope and through audio feedback. Robust tactile responses were typically found between 50 and 350 μm below the surface of the brainstem. Tactile responsive units in VPL thalamus were targeted by systematically lowering the recording electrode across a region spanning the following coordinates relative to bregma: A/P −1.25 to −1.85 mm and M/L 1.75 to 1.95 mm. Robust tactile responses were typically found between −3.2 to −3.7 mm below the surface of the brain. All data were collected within a 4-6 hour recording session and mice were perfused for histological analysis immediately following the experiment to verify viral expression and electrode placement. In a subset of experiments, the recording site was marked by iontophoretic injection of 2% Fluorogold (3 min with 7 sec on / 7 sec off cycle) at the same coordinates used for the recording electrode. Electrode sites were localized post mortem by cutting and mounting a continuous series of sections throughout the recording region. All data were analyzed offline by a second experimenter blinded to the experimental conditions. Data consisted of continuous raw data, spiking events (defined as crossing a threshold well above noise), and analog signals marking motor and optogenetic outputs. Cerebus Central Suite (Blackrock Microsystems) or Offline Sorter (Plexon) were used to set event thresholds, and NeuroExplorer (Plexon) was used to export recording files to MATLAB (MathWorks) for further analysis. Due to the high cellular density and anatomical arrangement of tactile responsive clusters within Cu, in some cases the recordings likely reflected multiple responsive units, despite conservative event thresholds (see fig. S6B). In all cases, comparison of responses in cuneate and thalamus was performed from the same recording site, and thus single- or multi-unit recordings from light off and light on conditions were always analyzed in a pairwise fashion.

Optogenetic manipulation of cuneate inhibitory neurons:

To evaluate the impact of activating local inhibitory circuits on tactile responses in the cuneate and thalamus, AAV5-hSyn-DIO-oChIEF-Citrine was injected into the cuneate region of adult VGAT-Cre mice, as described above. To evaluate the impact of suppressing local inhibitory circuits on tactile responses in the cuneate and thalamus, AAV1-hSyn-SIO-stGtACR2-FusionRed was injected into the cuneate region of adult VGAT-Cre mice, as described above. Following injections, mice were left for at least 3 weeks before electrophysiological recording. For both activation and inactivation, fiber optic cannulas were implanted, as described above. Light was delivered using a 470 nm fiber coupled LED (M470F3, Thorlabs) or a 473 nm laser (BL473T8, Shanghai Laser) with an output strength of ~3-10 mW. At each recording site, optogenetic activation (20 Hz, 15 msec pulse width) or inactivation (continuous) occurred on interleaved revolutions of the tactile wheel. Spike thresholds were manually set well above baseline noise during each recording using Cerebus Central Suite. Tactile motor and optogenetic trigger outputs from the Arduino board were sent as analog inputs to the Cerebus neural signal processor for alignment with recording data.

To quantify the effects of optogenetic manipulation, peri-event raster plots were generated and spikes were binned into 100 msec intervals, beginning at the initiation of motor revolution, and continuing for 6 sec, encompassing delivery of 4 tactile stimuli. The number of spikes generated at each recording site across all 20 tactile events (5 revolutions x 4 stimuli) for each condition (light on and light off) were then combined, and bin-averaged histograms with a 100 msec bin size were computed. To separate tactile evoked responses from tactile inter-stimulus interval (ISI) spiking, each 6 sec revolution was segmented into 4 individual 1.5 sec intervals corresponding to a quarter revolution of the stepper motor. Thus, a single tactile stimulus always occurred at some point within each 1.5 sec interval. Within each interval, we estimated the approximate window of tactile stimulation (0.7-0.9 sec) as the period when spike frequency rose to at least 200% above baseline activity, which was usually extremely low. Periods outside of this elevated spike activity were classified as ISI. The process was repeated for each recording site. All responses to optogenetic stimuli were subject to pairwise comparison across light off and light on conditions for each recording site.

Assessing rSM corticofugal modulation of tactile responses in the cuneate:

To assess the impact of activating descending rSM projections on tactile evoked responses in the cuneate, we developed an approach to deliver tactile stimuli by photostimulating the pad of the hand of VGluT1-Cre x Rosa-LSL-ChR2 mice with single pulses of 473 nm light (5 msec pulse width, BL473T8, Shanghai Laser). We used this optogenetic approach for two reasons: 1) we needed greater temporal precision than was possible with the stick-wheel in order to quantify the short latency effects of rSM stimulation; 2) previous work in cats has shown that when stimulating sensory cortex, modulation of tactile feedback is most apparent when tactile feedback is not saturating (41). An optogenetic approach offered more flexibility than the stick-wheel for varying the timing and strength of tactile stimulation, addressing both of these needs. Recordings were made in the cuneate and tactile responsive units were identified, as described above. The laser intensity was adjusted to just above the level required to evoke spiking in the cuneate at each recording site. In addition, bipolar stimulating electrodes, constructed from 125 μm teflon coated tungsten wire (California Fine Wire Company) spaced ~1 mm apart, were implanted into rSM (see coordinates above). The electrode pair was lowered ~500-600 μm below the cortical surface and affixed in place with epoxy. In accordance with prior studies (23, 68), a train of 6 pulses, 0.1 msec duration, at 150 μA was delivered at 333 Hz, with the first pulse delivered 30 msec before optogenetic stimulation of the hand, and 20 trials were performed for each recording site. To quantify the effects of rSM stimulation on tactile-evoked responses, signals were high-pass filtered at 250 Hz, with spike threshold at 133% above baseline noise (calculated using 3-Sigma Peak Heights detector, Offline Sorter, Plexon), and averaged across all recordings per unit or recording site. The total number of light-evoked spikes within 25 msec of the onset of hand optogenetic stimulation was then calculated. All responses were subject to pairwise comparison with and without rSM stimulation for each recording site.

String pulling behavioral assay

Assay:

The string pulling task was based on a previously developed assay (32). Mice were food restricted and maintained at ~85% of their original body weight. Pre-training for the task was carried out over 2-3 sessions before experimental trials. In the first pre-training session, mice were placed into a box similar to their home cage with 20 cotton strings, 2 mm in diameter, hanging through the top of the cage. Training strings ranged from 30-110 cm in length and were randomly spaced around the box perimeter. Half of the strings were baited with a peanut reward to increase motivation for performing the task. Animals were given 60 min to pull all 20 strings into their cage. If an animal failed to retrieve all 20 strings within the allotted time, the process was repeated the following day. By the end of the second day, all mice had successfully learned to pull the strings into the cage. Animals were then introduced into a 9 x 20 cm plexiglass testing chamber with a pulley system affixed to the top of the front panel. A 70 cm string was routed across the pulley and the end located outside the cage was affixed to a counterweight to encourage animals to pull the entire length of the string without interruption. Animals had to pull the string a total of ~40 cm to complete the trial, after which a peanut reward was manually dispensed. If the pulling bout was interrupted, the counterweight caused the string to retract to its initial position and mice had to restart the trial. The mice remained in the acquisition box until they successfully performed 4 complete pull trials, at which time they were considered acclimated to task conditions and ready to advance to the video acquisition trials.

For video acquisition trials, mice were placed in the testing chamber and allowed to acclimate before initiating the first trial. For each trial, mice had to retrieve the 70 cm, counterweighted string, after which a peanut reward was manually dispensed. Strings used for video acquisition were marked every 7 mm throughout their length to aid in the analysis of kinematics and pulling performance. Following each successful trial, the mice were given a 45 sec break before initiation of the next trial. Animals typically completed 10-15 trials in a single recording session. Video was acquired at 250 frames/sec by a camera (acA2040, Basler) placed 80 cm from the front panel of the testing chamber using Pylon software (Basler). The camera and testing box remained secured in place, such that the origin of the tracking system remained fixed to an arbitrary point in space.

Optogenetic manipulation of cuneate inhibitory neurons:

To evaluate the impact of activating local cuneate inhibitory circuits, AAV5-hSyn-DIO-oChIEF-Citrine was injected into the cuneate region of adult VGAT-Cre mice, as described above. To evaluate the impact of suppressing local cuneate inhibitory circuits, AAV1-hSyn-SIO-stGtACR2-FusionRed was injected into the cuneate region of adult VGAT-Cre mice, as described above. For control experiments, AAV5-CMV-EGFP was injected into the cuneate region of adult wild-type mice, as described above. After viral injections, animals were left for at least 3 weeks before undergoing an additional surgery to implant optical fibers targeting the cuneate, as described above. After fiber implant, animals were returned to their home cage and allowed at least 1 week to recover before behavioral testing.

Mice were briefly placed into an isoflurane induction chamber until they were sedated enough to enable the experimenter to attach a patch cable to the fiber optic cannula. Animals were then placed back in their home cage for 5 min to recover from anesthesia and acclimate to the tethered patch cable before being transferred to the testing chamber. Mice performed 8-14 trials, interleaving 2 light off trials with 2 light on trials. For all light on trials, a 473 nm laser with output calibrated to ~13 mW at the end of patch cable was triggered by a programmable Arduino board. For optogenetic activation (oChIEF) experiments, the laser was triggered shortly after the initiation of a pulling bout (10 Hz, 50 msec pulse width) and left on for the duration of the trial. Trial animals were tested at 10 Hz, 50 msec pulse width and 20 Hz, 50 msec pulse width. Both stimulation parameters produced a phenotype, so we opted for the lower frequency stimulation for subsequent oChIEF behavioral experiments. For optogenetic inactivation (stGtACR2) experiments, the laser was triggered shortly after the initiation of a pulling bout (continuous) and left on for the remainder of the trial. Periods of laser activation were marked for video analysis by a low intensity LED placed within the camera field of view but outside the view of the mouse.

Data analysis:

All manual string pull quantification was conducted by three experimenters blinded to the experimental conditions and results were averaged. Trials were first analyzed by adjusting video playback to 24% real-time, and manually counting: a) the total number of prehension mistakes; and b) the number of string misses during a grasp attempt across 5-7 trials per mouse, with a trial defined as a full string pull (~40 cm). Normally, mice alternate pulling the string with the left and right hands in a smooth and rhythmic fashion. Prehension mistakes were defined as any failed attempt to grasp the string on the first attempt after contact, causing a reversal in the direction of hand movement and a second attempt. Multiple grasping attempts in a single cycle were counted as a single prehension mistake. String misses were defined as a failure to make any contact with the string during the upward trajectory of the hand. For stGtACR2 experiments, we also quantified the proportion of trials when mice terminated or paused the behavior at light onset. Grasps associated with the initial engagement of the string on the first cycle or at the end of a trial when the string was falling into the cage were excluded from analysis. Rarely, mice grasped the string in their mouths as an alternate strategy for pulling the string downward. In these cases, the time when the string was in the mouth and the first grasp attempt immediately following string release from the mouth were excluded from analysis. After this first grasp attempt, behavior proceeded normally and subsequent prehension attempts were quantified. Blinded quantification (two experimenters) revealed that these biting attempts represented a mean of: 1.77% ± 1.07% SEM of all pull attempts in oChIEF light off trials and 3.78% ± 2.42% SEM of all pull attempts in oChIEF light on trials (6 mice; no significant difference between conditions; P = 0.2500; Wilcoxon two-tailed matched-pairs signed rank test). In control mice, biting attempts represented a mean of: 1.42% ± 0.54% SEM of all pull attempts in light off trials and 1.21% ± 0.94% SEM of all pull attempts in light on trials (6 mice; no significant difference between conditions; P = 0.8438; Wilcoxon two-tailed matched-pairs signed rank test).

To quantify string pull kinematics, DeepLabCut (69, 70) was used to automate tracking of the left and right hands. After tracking, the first ~7 continuous pulling segments per hand were selected from the longest pulling bout and were defined as a trial, resulting in approximately half of all the pulling bouts being used for kinematic quantification. Files containing x and y pixel coordinates for each hand were imported into MATLAB for kinematic analysis, and quantification was performed across 4-5 trials per mouse. All videos were manually reviewed to identify any outlier coordinates arising from labeling errors, which were infrequent. Any erroneously labeled frames were interpolated and all frame indices were converted to seconds by a conversion factor of 1/250. Next, pixels were converted into distance (mm) using an empirically determined conversion factor; the length of 10 consecutive markings on the string was measured to obtain the average distance between each mark, and pixel distances across each mark were then obtained from 2D image processing software (ImageJ) to compute the conversion factor. These steps were repeated five times and averaged, resulting in a factor of 0.34 mm per pixel. We used tracked videos to quantify: a) vertical direction reversals; b) vertical pathlength between direction reversals; c) latency between string contact and successful grasp; and d) the average vertical distance between the two hands as the load was transferred from the lower pulling hand to the upper grasping hand. To quantify vertical direction reversals (i.e., the number of times the animal reversed the direction of its hand movement in the y-axis), a peak-to-valley prominence threshold was defined through an empirical process that identified a threshold distance of 4.5 mm as sufficient for detecting nearly all direction reversals. Next, the number of vertical direction reversals exceeding threshold for each hand was calculated as the total number of peaks and troughs. To quantify the vertical pathlength (i.e., the average vertical distance traversed by the hand between direction reversals), the vertical distance between each peak and trough for every direction reversal was quantified and averaged. To quantify the latency between string contact and successful grasp, videos were analyzed frame by frame to index the following relevant events: contact was defined as the first occurrence of string contact by the hand on each pull cycle, and successful grasp was defined as the first occurrence of the hand firmly holding on to the string. To quantify the average vertical distance between the two hands during load transfer, videos were analyzed frame by frame to index the following relevant events: successful grasp with the top hand, as defined above, and release, defined as release of the string by the lower hand at the bottom of each pull cycle.

Tactile orienting behavioral assay

Assay:

To more selectively evaluate the contribution of tactile feedback in guiding dexterous motor output, we designed a novel head-fixed tactile orienting task for the mouse. The assay was loosely modelled after a task developed for use in humans to distinguish the acuity of tactile orienting from tactile perception (33). The task requires mice to use the glabrous pad of the hand to detect the orientation of grooves on a textured platform and use this tactile information to rotate a movable platform to a prescribed target angle. Throughout behavioral shaping, the task becomes progressively harder in a closed-loop fashion (i.e., introduction of a larger range of random starting orientations, smaller target zones, and longer hold periods). The complete design and code for this assay are available at http://www.github.com/azimlabsalk (51).

Tactile orientation cues were provided by a 3D printed (Form 2, Formlabs) pedestal lined with parallel ridges. The pedestal was 16 mm in diameter with 16 parallel and evenly spaced ridges at 1 mm intervals. In some trials the textured pedestal was replaced with a 3D printed smooth pedestal 16 mm in diameter. We reasoned that the removal of oriented ridges can be used to distinguish the use of fine tactile cues from the cutaneous feedback used to detect contact with the pedestal. The base of the pedestal was connected to a motor-encoder unit (precious metal brushes EBCCL, 3.2W, Maxon; Encoder MR type M, 256 counts per turn, Maxon) that controls and records the orientation of the pedestal. The motor-encoder unit is passively mobile through a 180° range (−30° to 150°, with 0° neutral angle defined as perpendicular to the body axis of the mouse), enabling mice to easily turn the pedestal when the motor is not actively engaged. During the task, the orientation of the pedestal ridges was actively manipulated through a closed-loop feedback system using an AutoPID library written for a microcontroller board (Arduino Mega 2560, Arduino) controlled through a custom-made PCB board. The PID regulates the pedestal angle by modulating motor torque based on feedback from the rotary encoder, including parameters such as current angle and angular velocity. Task related commands such as target angle, window play of target, task duration, and hold duration were controlled by a GUI task controller developed with the MATLAB data acquisition toolbox through a data acquisition device (NI DAQ USB-6002, National Instruments). Behavior was sometimes monitored using an infrared USB web camera (OV2710, OmniVision). The entire behavioral system was placed in a sound attenuation box (ENV-022MD-27, Med Associates Inc.), and training and experimental sessions were performed under red light.

Head-fixation:

The tactile orientation task was carried out in head-fixed mice that were water restricted and maintained at ~85% of their original body weight. The procedures for implanting the headpost and for post-operative care were based on previously established methods (71). Briefly, mice were anesthetized with isoflurane and placed into a stereotaxic frame, as described above. Custom-made headposts were lowered onto the skull using a stereotaxic manipulator designed to hold the headpost, centered 0.5 mm anterior from bregma along the midline, and secured with epoxy. After ~5 days of recovery, water restriction was initiated (1 ml/day, weight monitored daily), and mice were acclimated to head-fixation according to previously described methods (71, 72).

Task structure:

The final task was structured such that the trial was initiated when the pedestal was oriented to a reset position of −30° (30° aft of the axis perpendicular to the orientation of the mouse). Immediately after the reset, a randomly chosen starting orientation (0° ± 25°) was set, using the MATLAB randi function, which generates uniformly distributed pseudorandom integers. The trial was then initiated by rendering the motor passive, enabling the mouse to freely move the pedestal. In all cases, the behavioral apparatus forced animals to use their right hand to accomplish the task. The left hand was placed on a fixed surface of equal height. To receive a reward (8 μl water drop from a lick spout), the animal was required to orient the pedestal ridges to a defined target angle of 60° ± 20° by turning the platform clockwise and maintaining position within the target window for 600 msec. If the animal was unsuccessful in completing the task within 7 sec, the trial was terminated and the pedestal was reset to −30° without reward. The task continued until the mouse either performed 105 successful trials or until a total of 160 trials were attempted in a given day. If an animal became sated with water or if it struggled within the fixation frame, the task was terminated for the day.

Training period:

After 5 days of acclimating to head-fixation, training on the tactile orientation task was initiated. During the first week of training, task requirements were made easier by assigning a more restrictive starting window (0° ± 5°), a more lenient target window (60° ± 40°), and a shorter hold duration (100 msec) as reward criteria. After mice successfully learned to turn the pedestal under these lenient criteria, task difficulty was incrementally increased towards the final task structure described above. To maximize training efficiency, reward criteria were constantly adjusted via a computer controlled interface based on performance. For example, if the success rate over a given 10 trials was > 80%, the reward criteria for the next 10 trials was incrementally adjusted to the next difficulty. Conversely, if the success rate was < 70% over 10 trials, the reward criteria for the next 10 trials was eased. Mice were considered trained when they were able to obtain > 60% success with the most stringent criteria (target window = 60° ± 20°; random starting orientation = 0 ± 25°; hold duration = 600 ms), which typically occurred after 4-6 weeks of training. Mice unable to attain 60% success on the most stringent criteria were excluded from the study.

Optogenetic manipulation of cuneate inhibitory neurons:

To evaluate the impact of activating local cuneate inhibitory circuits, AAV5-hSyn-DIO-oChIEF-Citrine was injected into the cuneate region of adult VGAT-Cre mice, as described above. For control experiments, AAV5-CMV-EGFP was injected into the cuneate region of adult wild-type mice, as described above. Animals were left for at least 3 weeks before undergoing an additional surgery to implant optical fibers targeting the cuneate, as described above. Animals were returned to their home cage and allowed at least 1 week to recover before training was initiated. After mice were trained to perform the task, data acquisition trials were initiated. During the acquisition period, mice underwent 40 practice trials and then moved on to test trials in which they received photostimulation throughout the duration of the trial (473 nm laser, 10 Hz, 50 msec pulse width, ~13 mW at patch cable interface). Photostimulation trials were randomly selected and occurred at a probability of 30-40%.

Data analysis:

Data collected from each animal, including angular trajectories and time to complete trials, were concatenated for further analysis in MATLAB. The success rate was computed as the number of rewarded trials/total number of trials. To assess the efficiency of task execution specifically on successful trials, we excluded failures and computed the cumulative distribution functions (CDF) of the time to complete each trial. To compare CDF data between mice, we randomly selected a fixed number of samples from each animal (determined by the animal with the lowest number of successful trials). To quantify the peak angular velocity of pedestal rotation, sampling rates were standardized to the trial with the lowest rate in each recording session, and a low-pass filter (20 Hz cutoff frequency) was applied to minimize non-movement related noise (the filter threshold was determined through test recordings without a mouse in the apparatus). Instantaneous angular velocities were obtained by computing angular changes across all timestamps and the highest rate was assigned as the peak angular velocity for that trial. The mean peak angular velocity was then calculated for each mouse across all trials.

Statistics and data collection

Mice from each litter were randomly allocated to different groups for the electrophysiological and behavioral experiments. Group sizes were not pre-determined, but sample sizes are comparable to those commonly used for similar experiments and were selected such that appropriate statistical tests could be used. Data analysis for extracellular recording and manual quantification of string pulling behavior were blinded (see above). For string pull kinematics and the tactile orienting assay, data collection and analysis were automated, minimizing any potential influence by the experimenter. For in vivo electrophysiology, in all cases where good tactile-responsive recordings were obtained, the data were included in the analyses. Animals were only excluded from behavioral experiments if data collection could not be performed. These cases were: two mice in the stGtACR2 string pull experiment that did not reliably perform the behavior during training; one mouse in the stGtACR2 string pull experiment that had an injured right hand; two mice in the tactile orienting task that never reached behavioral training threshold; and one mouse in the tactile orienting task in which the optical fiber broke. Results are shown as box-and whisker plots indicating the median, 25^th and 75^th percentiles, and range, unless otherwise indicated. Normality was assessed using the Shapiro-Wilk test, and non-parametric tests were used for non-normally distributed data. All parametric and non-parametric tests used, as well as any multiple comparisons tests and corrections, are indicated in the figure legends, as are n values for each experiment. All statistical comparisons were two-tailed, when relevant. P < 0.05 was considered significant, * indicates P < 0.05, ** < 0.01, *** < 0.001, **** < 0.0001, and all significant P values are indicated in figure legends. Statistical analysis was performed in MATLAB or Prism (version 8.4.3, GraphPad).

Supplementary Material

Fig S1

Fig. S1. Termination of ascending forelimb sensory afferents in the dorsal column nuclei complex. (A) Labeling of sensory afferents through injection of AAV-DIO-hChR2-EYFP or AAV-CMV-EGFP into a single dorsal root ganglion (DRG, C6 or C7) of Avil-Cre mice reveals projections to the ipsilateral Cu and ECu (4 mice). The Avil-Cre mouse line drives expression selectively and broadly across sensory neuron classes (73). Fluorescence dorsal to Cu represents incoming fibers. (B) Labeling direct proprioceptive (left distal muscles; extensor digitorum lateralis, extensor digitorum communis, and extensor carpi radialis) and cutaneous (right glabrous pad) projections by cholera toxin B subunit (CTB) injection into peripheral end organs. As with proximal muscles (see Fig. 1B), proprioceptive afferents from distal forelimb muscles avoid Cu, but instead target the ipsilateral ECu and more caudal regions of the cuneate nucleus (left side of brainstem). Cutaneous afferents from the glabrous pad of the hand densely innervate Cu but avoid ECu (right side of brainstem) (6 mice). These results establish that direct ascending cutaneous and proprioceptive afferents remain largely segregated at the level of the dorsal column nuclei complex in mice (74, 75). (C) Selective labeling of proprioceptive neurons though injection of AAV-DIO-hChR2-tdTom or AAV-DIO-EGFP into proximal forelimb muscles (biceps and triceps brachii) of PV-Cre mice reveals minimal targeting of Cu, but extensive targeting of the ipsilateral ECu and some projections to other regions of the cuneate nucleus (not shown) (5 mice). The Pv-Cre mouse line drives sensory neuron expression that is largely specific to proprioceptors (76). Limb muscle schematics here and in Fig. 1B from (77).

NIHMS1748970-supplement-Fig_S1.jpg^{(28.7MB, jpg)}

Fig S2

Fig. S2. Cuneolemniscal neuron action potential threshold and spontaneous inhibitory input. (A) Whole-cell slice recordings from labeled cuneolemniscal (CL) neurons (as in Fig. 1C). Example phase plot (left) showing the rate of change of membrane potential versus membrane potential. The voltage threshold (V_threshold) for firing an action potential (AP) was calculated from the first sweep of current injection that produced an AP. Resting membrane potential (RMP, −59.06 mV ± 1.50 mV SEM) is slightly less than AP threshold (−49.67 mV ± 1.78 mV SEM) (right, 18 neurons from 10 mice; ****P < 0.0001; paired t test; also see Fig. 1D). (B) Cumulative probability distribution of the inter-event intervals of spontaneous postsynaptic currents recorded from CL neurons at baseline (black), followed by sequential bath application of strychnine (blue) then bicuculline (red) (left, 7 neurons in 4 mice), or bicuculline then strychnine (right, 8 neurons in 4 mice).

NIHMS1748970-supplement-Fig_S2.jpg^{(1.8MB, jpg)}

Fig S3

Fig. S3. GABAergic and glycinergic neurons are differentially distributed within the cuneate region and directly target cuneolemniscal neurons. (A) GAD1-EGFP mice were used to label GABAergic neurons (green), and cuneolemniscal (CL) neurons (red) were traced by injecting either CTB or Fluorogold into contralateral VPL thalamus (3 mice). Few GABAergic neurons are localized within the Cu core region, where CL neurons are found, but are instead located primarily within the ventral shell of the Cu (arrows). In no cases were GABAergic neurons co-labeled with CTB or Fluorogold. Overlay shown on right. Gr, gracile nucleus. (B) GlyT2-EGFP mice were used to label glycinergic neurons (green), and CL neurons (red) were traced by injecting either CTB or Fluorogold into contralateral VPL thalamus (3 mice). Many glycinergic neurons can be found within the Cu core region as well as in the ventral shell region (arrows). In no cases were glycinergic neurons co-labeled with CTB or Fluorogold. Overlay shown on right. It is possible that some of these inhibitory neurons are positive for both glycine and GABA (29). (C) Monosynaptic retrograde rabies tracing from CL neurons through injection of AAVretro-Cre into VPL thalamus and AAV-DIO-TVA-G into Cu, followed 3-4 weeks later by injection of EnvA-pseudotyped RabΔG-EGFP into Cu (2 mice). GABAergic cells (red) were labeled by crossing GAD2-FlpO mice with a Rosa-FSF-tdTom line. Yellow neurons (arrows) indicate GABAergic neurons, located largely in the Cu ventral shell, that provide monosynaptic inputs to CL neurons. Note that use of a TC66T strain of helper virus, containing a mutated, low-affinity form of the TVA receptor, limits nonspecific rabies transduction due to low levels of leaky TVA expression (61). (D) Monosynaptic retrograde rabies tracing from CL neurons (as in (C); 2 mice). Glycinergic cells (red) were labeled by crossing GlyT2-FlpO mice with a Rosa-FSF-tdTom line. Yellow neurons (arrows) indicate glycinergic neurons located in the Cu core and ventral shell that provide monosynaptic inputs to CL neurons.

NIHMS1748970-supplement-Fig_S3.jpg^{(20.5MB, jpg)}

Fig S4

Fig. S4. Optogenetic activation and spontaneous inhibition in cuneate inhibitory neurons. (A) In vitro slice recording from Cu inhibitory neurons optogenetically activated following injection of AAV-DIO-oChIEF-Citrine in the Cu core and ventral shell (V shell) regions of VGAT-Cre mice. Inhibitory neurons were targeted for recording by Citrine expression. (B) Example whole-cell recording from a labeled inhibitory neuron in current clamp (top) and voltage clamp (bottom), with a continuous 100 msec light pulse (left), or five 10 msec light pulses at 10 or 20 Hz (right). Light-evoked responses were observed under all conditions. (C) Quantification of light-evoked currents (100 msec light pulse, 7 neurons in 6 mice). (D) Example whole-cell recording from a Citrine-positive inhibitory neuron held at −70mV showing spontaneous events at baseline (black). Bath application of bicuculline (red) followed by strychnine (blue) progressively eliminated most spontaneous inhibitory post-synaptic currents (IPSCs). Pipette was filled with high chloride solution, causing IPSCs to appear as inward currents. (E) Sequential application of bicuculline and strychnine cumulatively decreased the frequency of spontaneous IPSCs in Cu inhibitory neurons (7 neurons in 5 mice, 4 neurons for strychnine; **P = 0.0063, *P = 0.0147; One-way mixed-effects model with Geisser-Greenhouse correction and Tukey’s multiple comparisons test). (F) Cumulative probability distribution of the inter-event intervals of spontaneous postsynaptic currents recorded from Cu inhibitory neurons at baseline (black), followed by sequential bath application of bicuculline (red) then strychnine (blue) (7 neurons in 5 mice, 4 neurons for strychnine). Subsequent application of strychnine minimally altered event probability, suggesting the majority of IPSCs are GABA_A receptor mediated.

NIHMS1748970-supplement-Fig_S4.jpg^{(3MB, jpg)}

Fig S5

Fig. S5. The relative ratio of GABAergic and glycinergic monosynaptic inhibition of cuneolemniscal neurons varies across cells. (A) In vitro slice recording from CL neurons retrogradely labeled by retrobead injection into contralateral VPL thalamus. Cu inhibitory neurons were optogenetically activated following injection of AAV-DIO-oChIEF-Citrine in the Cu of VGAT-Cre mice (as in Fig. 1H). (B) Onset kinetics of light-evoked IPSCs in CL neurons. Plot shows short latency and low jitter (standard deviation of trial-to-trial event latencies) of light-evoked IPSCs (error bars indicate SEM, 18 neurons in 8 mice). (C) Sequential application of bicuculline then strychnine or strychnine then bicuculline shows cumulative reduction in the amplitude of light-evoked IPSCs with an approximately equal mix of GABAergic and glycinergic components (see Fig. 1I). The ratio of GABAergic and glycinergic inhibition onto CL neurons varies across cells (14 neurons in 7 mice). In neurons where GABAergic inhibition is low, glycinergic inhibition is high, and vice versa. The order of strychnine and bicuculline application did not alter the combined effect. A previous model proposed that GABAergic cells inhibit cuneolemniscal neurons while glycinergic neurons disinhibit cuneolemniscal neurons by inhibiting GABAergic cells (27, 39). Our results expand upon and revise this model by demonstrating that both GABAergic and glycinergic cells elicit direct cuneolemniscal inhibition in varying combinations.

NIHMS1748970-supplement-Fig_S5.jpg^{(1MB, jpg)}

Fig S6

Fig. S6. Tactile-evoked responses in cuneate during optogenetic activation of cuneate inhibitory neurons. (A) In vivo extracellular recording in Cu of anesthetized mice with tactile stimuli applied to the ipsilateral glabrous pad of the hand and optogenetic activation of local Cu inhibitory neurons (viral oChIEF expression in VGAT-Cre mice). Typical Cu recording site shown on right, labeled by iontophoretic injection of Fluorogold (Materials and Methods). (B) Example raw voltage traces during one rotation of the stick-wheel (from recordings shown in Fig. 2D) with spike threshold, isolated spikes, and spike waveforms indicated (top, light off; bottom, light on).

NIHMS1748970-supplement-Fig_S6.jpg^{(6.3MB, jpg)}

Fig S7

Fig. S7. Cuneate inhibitory circuits regulate the transmission of tactile signals to the thalamus. (A) In vivo extracellular recording in VPL thalamus of anesthetized mice with tactile stimuli applied to the contralateral glabrous pad of the hand and optogenetic activation of local Cu inhibitory neurons (top, viral oChIEF expression in VGAT-Cre mice). Spike raster plots (middle) and mean spike number histograms (bottom; 0.1 sec bin) from an example recording site across ten interleaved trials (five light off, gray; five light on, blue; each with four tactile stimuli, purple bars). Lines in lower histogram indicate mean, shaded areas represent SEM. (B) Mean number of spikes/sec during a single tactile stimulus (top; 0.1 sec bin) across all recordings (31 sites from 3 mice). Lines indicate mean, shaded areas represent SEM. Pairwise comparisons of the total number of spikes across five trials (20 tactile stimuli each for light off and light on) reveal suppression of tactile-evoked spikes in VPL neurons (middle; 61.13% ± 4.64% SEM decrease in spike number; ****P < 0.0001) but no suppression of spontaneous activity during tactile ISI periods (bottom), suggesting that these thalamic neurons also receive convergent input from other locations or exhibit their own spontaneous activity. (Wilcoxon two-tailed matched-pairs signed rank test). (C) In vivo extracellular recording in VPL with optogenetic inhibition of local Cu inhibitory neurons (top, viral stGtACR2expression in VGAT-Cre mice). Spike raster plots from an example recording site (middle; as in (A); light off, gray; light on, red). Lines in histogram (bottom) indicate mean, shaded areas represent SEM. (D) Quantification (as in (B)) across all recordings (26 sites from 3 mice; top). Pairwise comparisons reveal an increase in spikes evoked by tactile stimuli (middle; 15.77% ± 5.32% SEM increase in spike number; **P = 0.0025) and an increase in spontaneous activity during tactile ISI periods (517.22% ± 207.23% SEM increase in spike number; **P = 0.0016). (Wilcoxon two-tailed matched-pairs signed rank test). (E) Typical VPL recording site, labeled by iontophoretic injection of Fluorogold (Materials and Methods).

NIHMS1748970-supplement-Fig_S7.jpg^{(9.5MB, jpg)}

Fig S8

Fig. S8. Perturbing cuneate inhibitory circuits disrupts string pulling behavior, while photostimulation in control mice has no effect. (A) String pulling task. Local Cu inhibitory neurons were targeted for optogenetic activation through viral oChIEF expression in VGAT-Cre mice. Photoactivation had no effect on how often the ipsilateral hand missed contact with the string during a reach-to-grasp attempt (left; % of all grasp attempts; Wilcoxon two-tailed matched-pairs signed rank test). However, once a grasp was successfully made, there was a decrease in the vertical distance between the hands as the load was transferred from the lower ipsilateral to the upper contralateral hand (right; ipsi/contra indicate lower hand during string release; ****P < 0.0001; no effect with contra on bottom; two-way repeated measures ANOVA with Sidak multiple comparisons test). (6 mice; also see Fig. 3). (B) As a control, wild-type mice received unilateral injection of AAV-EGFP into the Cu region followed by implantation of an optical fiber. Photostimulation had no effect on: ipsilateral or contralateral prehension mistakes (top left; % of grasp attempts in which an error was made across trials); the mean number of vertical (y) direction reversals (top middle; mean number of direction reversals per trial); the mean y path length traversed by either hand between direction reversals (top right; mean absolute distance between the peak and trough of a given path segment across trials); the mean latency between first string contact and a successful grasp (bottom left); the frequency of ipsilateral missed contacts with the string during a reach-to-grasp attempt (bottom middle; % of all grasp attempts); or the vertical distance between the hands during load transfer (bottom right; ipsi/contra indicate lower hand during string release). (6 mice; Wilcoxon two-tailed matched-pairs signed rank test for string misses; two-way repeated measures ANOVA with Sidak multiple comparisons test for all others).

NIHMS1748970-supplement-Fig_S8.jpg^{(2.9MB, jpg)}

Fig S9

Fig. S9. Aberrant activation of cuneate inhibitory circuits disrupts performance of a tactile orientation task. (A) Schematic of head-fixed, tactile orienting assay. A pedestal (16 mm diameter) with parallel orientation ridges (1 mm interval, evenly spaced) was placed under the right hand. A headpost was used to restrain the mouse. The pedestal was connected to a motor-rotary encoder assembly. A GUI task controller provides PID control of the pedestal, delivers task commands, and records performance (Materials and Methods). (B) Ipsilateral Cu inhibitory neurons were targeted for optogenetic activation (as in Fig. 4). During photoactivation (blue), the overall success rate (left) was unaffected in mice using either a textured (7 mice paired) or a smooth surface (8 mice with light off, 4 mice with light on; 4 paired). However, success rate was reduced when mice switched from a textured to a smooth surface with the light off (8 mice paired; *P = 0.0355) or light on (7 mice with texture, 4 mice with smooth; 2 paired; *P = 0.0245). The mean elapsed time to achieve success (middle) increased during photoactivation for mice using a textured surface (7 mice paired; (**P = 0.0047), and for mice that switched from a textured to a smooth surface with the light off (8 mice paired; *P = 0.0170). There was no difference in the mean peak angular velocity (deg/sec) when making turns across any of the conditions. (Two-way mixed-effects model with Geisser-Greenhouse correction and Sidak multiple comparisons test; also see Fig. 4F–H). Portions of this data are also presented in Fig. 4G. (C) The cumulative distribution function (CDF) for only the successful trials under each condition for individual mice and across all mice (right). A rightward shift of the CDF was seen during photoactivation in mice using a textured surface (7 mice; ****P < 0.0001), when mice switched from a textured to a smooth surface with the light off (8 mice; ****P < 0.0001), and when mice switched from a textured to a smooth surface with the light on (4 mice; ***P < 0.0007). (Kolmogorov-Smirnov test; also see Fig. 4I). The proportion of successful trials in the first time bin (0.6 – 1.24 sec) was reduced for mice using a textured surface during photoactivation (7 mice paired; **P = 0.0044), and for mice that switched from a textured to a smooth surface with the light off (8 mice paired; **P = 0.0065). (Two-way mixed-effects model with Geisser-Greenhouse correction and Sidak multiple comparisons test; also see Fig. 4F–H). (D) As a control, wild-type mice received unilateral injection of AAV-EGFP into the Cu region followed by implantation of an optical fiber. Photostimulation in mice using a textured surface did not affect the task success rate (4 mice; paired t test), the mean elapsed time to achieve success (paired t test), the mean peak angular velocity (paired t test), or the CDF of successful trials (Kolmogorov-Smirnov test).

NIHMS1748970-supplement-Fig_S9.jpg^{(7.8MB, jpg)}

Fig S10

Fig. S10. Identifying corticofugal and corticospinal inputs to the cuneate region. (A) Approach to broadly identify corticofugal and corticospinal inputs to the Cu region (left). AAVretro-Cre was injected into C6-C8 spinal segments in Rosa-LSL-tdTom mice, labeling corticospinal neurons (red). AAVretro-GFP was injected into the Cu to label descending corticocuneate projections (green). Corticospinal neurons that also target Cu are dual labeled (yellow). Corticospinal neurons (red) can be seen throughout contralateral primary (SSp) and supplemental (SSs) somatosensory cortices, as well as primary (MOp) and secondary motor cortices (MOs). Retrograde labeling from the Cu region showed considerable overlap with corticospinal projections (yellow), especially throughout SSp and SSs. Corticocuneate neurons (green) were found in contralateral rostral sensorimotor cortex (rSM), where very few corticospinal projection neurons are located. Corticocuneate neurons were rarely found within MOp or MOs. (3 mice). (B) Approach to distinguish projections from SSp and MOp cortices (left). To label corticospinal projections, AAVretro-Cre was injected into C6-C8 spinal segments, and AAV-DIO-hChR2-EYFP was injected into contralateral MOp and AAV-DIO-hChR2-tdTom (or AAV-DIO-hChR2-mCherry) was injected into contralateral SSp. Fluorophore expression was segregated to the two cortical regions (middle). The core regions of contralateral Cu and Gr (right) are heavily innervated by SSp (red), while only sparse innervation from MOp (green) can be found, mostly in the ventral shell region. In some mice the viruses were switched, with no change in results. (4 mice). (C) Approach to identify projections to the Cu region that specifically arise from corticospinal neurons (left). AAVretro-Cre was injected into C6-C8 spinal segments, and AAV-DIO-hChR2-EYFP was injected into contralateral SSp. CL neurons were retrogradely labeled by Fluorogold injection into contralateral VPL thalamus. Corticospinal neurons in SSp descend through the ipsilateral pyramidal tract (Py), decussate, and send collaterals that densely target the core regions of contralateral Cu and Gr, where CL neurons are located (also see Fig. 5B). (6 mice). (D) Approach to identify projections to the Cu region that specifically arise from rSM (left). AAVretro-Cre was injected into the Cu region and AAV-DIO-hChR2-EYFP was injected into contralateral rSM. rSM projections also descend through the ipsilateral Py and decussate. However, unlike SSp, rSM projections completely avoid the core regions of Cu and instead target more ventral brainstem regions, including the Cu ventral shell region where inhibitory neurons that target CL neurons are located (also see Fig. 5B). (5 mice).

NIHMS1748970-supplement-Fig_S10.jpg^{(20.7MB, jpg)}

Fig S11

Fig. S11. Identifying brain-wide inputs onto cuneolemniscal neurons and cuneate inhibitory neurons. (A) A complementary monosynaptic rabies tracing strategy was used to validate the first approach (see Fig. 5A) and deliver nuclear localized fluorophore to enable automated quantification of labeled neurons across the brain. AAV-DIO-TVA-G was injected into Cu of VGluT2-Cre mice (avoiding transduction of inhibitory neurons), followed 7 weeks later by injection of EnvA-pseudotyped RabΔG-H2B-EGFP into contralateral VPL thalamus (3 mice). As seen with the first approach, monosynaptic cortical inputs arise almost exclusively from contralateral primary (SSp; ii) and supplemental (SSs; iii) somatosensory cortices, but not primary motor cortex (MOp) or rostral sensorimotor cortex (rSM; i). Serial two-photon tomography was used to quantify other monosynaptic inputs throughout the brain (see Materials and Methods and Table S1). (B) Monosynaptic retrograde rabies tracing from Cu inhibitory neurons through injection of AAV-DIO-TVA-G into the Cu region of VGAT-Cre mice, followed 3-4 weeks later by injection of EnvA-pseudotyped RabΔG-H2B-EGFP into the Cu region (3 mice). Monosynaptic cortical inputs also arise from contralateral SSp (v) and SSs (vi), but also include a large population of corticofugal neurons throughout contralateral rSM (iv). Serial two-photon tomography was used to quantify other monosynaptic inputs throughout the brain (see Materials and Methods and Table S2). (C) Control experiment for disynaptic retrograde rabies tracing in Fig. 5C. For monosynaptic transport from CL neurons, AAVretro-FlpO was injected into the contralateral VPL thalamus and Flp-dependent rabies helper viruses AAV-fDIO-TVA and AAV-fDIO-G were injected into the Cu region. After 3-4 weeks, EnvA-pseudotyped RabΔG-H2B-EGFP (expressing nuclear-localized EGFP) was injected into contralateral VPL thalamus, selectively targeting CL neurons that had received helper viruses. To prevent disynaptic transport, supplemental G-protein was not provided to local Cu inhibitory neurons. Mirroring the results of monosynaptic tracing from CL neurons (Fig. 5A, left), cortical inputs arise almost exclusively from contralateral SSp and SSs, with little to no labeling present in rSM. As expected, labeling was also found in cervical DRG and in presumptive local Cu inhibitory neurons (not shown). (4 mice).

NIHMS1748970-supplement-Fig_S11.jpg^{(20MB, jpg)}

Fig S12

Fig. S12. Summary of local and long-distance cuneate circuits. The main cuneate nucleus is a major conduit of forelimb sensory information to supraspinal regions, including the neocortex, via cuneolemniscal projections (yellow) to the thalamus (VPL). The core region of the middle cuneate (Cu) receives direct input from cutaneous afferents (green) that innervate the glabrous pad of the hand and reside in the dorsal root ganglia (DRG). GABAergic neurons (orange) located largely in the cuneate ventral shell (V shell) and glycinergic neurons (red) located in the cuneate ventral shell and cuneate core directly inhibit cuneolemniscal neurons. These inhibitory neurons receive inhibitory input, potentially from local connections, and also receive direct input from ascending cutaneous afferents. The cuneate core region is heavily targeted by corticospinal neurons residing in primary somatosensory cortex (SSp; dark blue), and cuneolemniscal neurons that reside in this core region receive direct input from SSp projections. Inhibitory cuneate neurons also receive SSp input, but unlike cuneolemniscal neurons, are also targeted by corticofugal neurons in rostral sensorimotor cortex (rSM; light blue), which do not project to the cuneate core, but rather target the shell region ventral to the cuneate. Cuneate inhibitory circuits provide a means for bidirectional modulation of the transmission of tactile information through the cuneate core, and their activation or inhibition perturbs the execution of tactile-guided dexterous behaviors. ECu, external cuneate; Py, pyramidal tract.

NIHMS1748970-supplement-Fig_S12.jpg^{(1.6MB, jpg)}

Table S1

Table S1. Brain-wide monosynaptic inputs to cuneolemniscal neurons (separate file). Serial two-photon tomography was used to quantify monosynaptic inputs to cuneolemniscal neurons throughout the brain (see fig. S11A and Materials and Methods). Ipsilateral and contralateral quantification (relative to the injection site) is displayed in separate tabs for data collected from 3 mice. Columns display: region volumes; mean raw intensity of fluorescence in each region across mice; mean intensity of fluorescence per cubic mm in each region; SEM of mean intensity of fluorescence per cubic mm in each region; the mean probability normalized to the highest whole-brain mean intensity per cubic mm; and the SEM of the mean probability normalized to the highest whole-brain mean intensity per cubic mm. All data can be sorted by column values. Region abbreviations can be found at https://mouse.brain-map.org/static/atlas.

NIHMS1748970-supplement-Table_S1.xlsx^{(119.4KB, xlsx)}

Table S2

Table S2. Brain-wide monosynaptic inputs to cuneate inhibitory neurons (separate file). Serial two-photon tomography was used to quantify monosynaptic inputs to Cu inhibitory neurons throughout the brain (see fig. S11B and Materials and Methods). Ipsilateral and contralateral quantification (relative to the injection site) is displayed in separate tabs for data collected from 3 mice. Columns display: region volumes; mean raw intensity of fluorescence in each region across mice; mean intensity of fluorescence per cubic mm in each region; SEM of mean intensity of fluorescence per cubic mm in each region; the mean probability normalized to the highest whole-brain mean intensity per cubic mm; and the SEM of the mean probability normalized to the highest whole-brain mean intensity per cubic mm. All data can be sorted by column values. Region abbreviations can be found at https://mouse.brain-map.org/static/atlas.

NIHMS1748970-supplement-Table_S2.xlsx^{(118KB, xlsx)}

Movie S1

Movie S1. Normal execution of the string pull task. Cu inhibitory neurons were targeted for optogenetic activation by unilateral injection of AAV-DIO-oChIEF-Citrine into the Cu region of VGAT-Cre mice followed by implantation of an optical fiber. With the light off, behavioral performance was normal as both hands exhibited smooth cycles with uninterrupted pulling paths. Automated markerless tracking of the hands (69, 70) was used for kinematic quantification (see Fig. 3 and fig. S8).

Download video file^{(6.5MB, mov)}

Movie S2

Movie S2. Activation of cuneate inhibitory circuits disrupts string pull performance. In the same animal shown in Movie S1, photostimulation (473 nm, 10 Hz, 50 msec pulse width) resulted in frequent ipsilateral prehension mistakes (left hand) and multiple grasp attempts at the top of the string pull cycle (see Fig. 3 and fig. S8).

Download video file^{(8MB, mov)}

Movie S3

Movie S3. Suppression of cuneate inhibitory circuits can cause premature termination of string pull behavior. Cu inhibitory neurons were targeted for optogenetic inhibition by injecting AAV-SIO-stGtACR2-FusionRed into the Cu region of VGAT-Cre mice followed by implantation of an optical fiber. In a subset of trials, photoinhibition (473 nm, continuous) caused early termination of a pulling bout and clutching of the ipsilateral (right) hand.

Download video file^{(11.2MB, mov)}

Movie S4

Movie S4. Execution of the tactile orienting task. Head-fixed mouse successfully performing the tactile orienting task. A pedestal with parallel orientation ridges was placed under the right hand. The pedestal was connected to a motor-rotary encoder assembly. The left hand was placed on a fixed surface of equal height. The neutral angle of the ridges was defined as perpendicular to the body axis of the mouse (0°), and the pedestal was passively mobile through a 180° range (−30° to 150°). At the beginning of each trial, the pedestal was reset to −30° and then adjusted to a random starting orientation (0° ± 25°). The pedestal then became passively mobile, indicating the start of the trial. Water reward was delivered if the orientation of the ridges stayed within the target zone (60° ± 20°) for > 0.6 sec (see Fig. 4 and fig. S9).

Download video file^{(12.3MB, mov)}

Acknowledgements:

We are grateful to Phong Nguyen and Graham Salmun (Salk Institute) for assistance with mouse husbandry, histology, behavioral assays, and lab operations; Huijing Gao and Ayesha Thanawalla (Salk Institute) for assistance with string pull experiments; Byungkook Lim (University of California, San Diego) for the AAV-DIO-mRuby-T2A-Syp-EGFP plasmid; Troy Margrie (Sainsbury Wellcome Centre) for the AAV-fDIO-N2cG-H2B-GFP plasmid; Ali Cetin (Allen Institute for Brain Science) for the EnvA-RabΔG-H2B-EGFP plasmid; Adam Hantman (Janelia Research Campus) for the VGluT1-Cre, GAD2-Flp, and GlyT2-Flp mouse lines; Frank Cardone, Mark Stambaugh, Jeff Sandubrae (Qualcomm Institute, University of California, San Diego), and Dan Butler (Salk Institute) for assistance developing the tactile orienting assay; and Denise Ramirez (Whole Brain Microscopy Facility, University of Texas Southwestern Medical Center) for assistance with serial two-photon tomography and quantification. We thank Sho Aoki, Sliman Bensmaia, Andrew Fink, Kee Wui Huang, Denis Jabaudon, Kazuhiko Seki, John Tuthill, Michael Yartsev, and members of the Azim lab for valuable discussion and comments on the manuscript.

Funding:

Supported by a Salk Pioneer Fund Postdoctoral Scholar Award: AB; Uehara Memorial Foundation Postdoctoral Fellowship: MI; National Institutes of Health (R00NS088193, DP2NS105555, R01NS111479, and U19NS112959), the Searle Scholars Program, The Pew Charitable Trusts, and the McKnight Foundation: EA.

Footnotes

Publisher's Disclaimer: This manuscript has been accepted for publication in Science. This version has not undergone final editing. Please refer to the complete version of record at http://www.sciencemag.org/. The manuscript may not be reproduced or used in any manner that does not fall within the fair use provisions of the Copyright Act without the prior, written permission of AAAS.

Competing interests: The authors declare no competing interests.

Data and materials availability:

All data are available in the main text or the supplementary materials. Design and code for the tactile orienting assay are available at (51). www.github.com/azimlabsalk.

References and Notes

1.Mountcastle VB, The Sensory Hand : Neural Mechanisms of Somatic Sensation. (Harvard University Press, Cambridge, Mass., 2005), pp. xiv, 616 p. [Google Scholar]
2.Johansson RS, Flanagan JR, Coding and use of tactile signals from the fingertips in object manipulation tasks. Nat Rev Neurosci 10, 345–359 (2009). doi: 10.1038/nrn2621 [DOI] [PubMed] [Google Scholar]
3.Scott SH, A Functional Taxonomy of Bottom-Up Sensory Feedback Processing for Motor Actions. Trends Neurosci 39, 512–526 (2016). doi: 10.1016/j.tins.2016.06.001 [DOI] [PubMed] [Google Scholar]
4.Shadmehr R, Smith MA, Krakauer JW, Error correction, sensory prediction, and adaptation in motor control. Annu Rev Neurosci 33, 89–108 (2010). doi: 10.1146/annurev-neuro-060909-153135 [DOI] [PubMed] [Google Scholar]
5.Azim E, Seki K, Gain control in the sensorimotor system. Curr Opin Physiol 8, 177–187 (2019). doi: 10.1016/j.cophys.2019.03.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Sillito AM, Jones HE, Gerstein GL, West DC, Feature-linked synchronization of thalamic relay cell firing induced by feedback from the visual cortex. Nature 369, 479–482 (1994). doi: 10.1038/369479a0 [DOI] [PubMed] [Google Scholar]
7.Gilbert CD, Sigman M, Brain states: top-down influences in sensory processing. Neuron 54, 677–696 (2007). doi: 10.1016/j.neuron.2007.05.019 [DOI] [PubMed] [Google Scholar]
8.Lee S, Carvell GE, Simons DJ, Motor modulation of afferent somatosensory circuits. Nat Neurosci 11, 1430–1438 (2008). doi: 10.1038/nn.2227 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Fink AJ, Croce KR, Huang ZJ, Abbott LF, Jessell TM, Azim E, Presynaptic inhibition of spinal sensory feedback ensures smooth movement. Nature 509, 43–48 (2014). doi: 10.1038/nature13276 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Confais J, Kim G, Tomatsu S, Takei T, Seki K, Nerve-Specific Input Modulation to Spinal Neurons during a Motor Task in the Monkey. J Neurosci 37, 2612–2626 (2017). doi: 10.1523/JNEUROSCI.2561-16.2017 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Liu Y, Latremoliere A, Li X, Zhang Z, Chen M, Wang X, Fang C, Zhu J, Alexandre C, Gao Z, Chen B, Ding X, Zhou JY, Zhang Y, Chen C, Wang KH, Woolf CJ, He Z, Touch and tactile neuropathic pain sensitivity are set by corticospinal projections. Nature 561, 547–550 (2018). doi: 10.1038/s41586-018-0515-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Schneider DM, Sundararajan J, Mooney R, A cortical filter that learns to suppress the acoustic consequences of movement. Nature 561, 391–395 (2018). doi: 10.1038/s41586-018-0520-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Wall PD, The sensory and motor role of impulses travelling in the dorsal columns towards cerebral cortex. Brain 93, 505–524 (1970). doi: 10.1093/brain/93.3.505 [DOI] [PubMed] [Google Scholar]
14.Glendinning DS, Cooper BY, Vierck CJ Jr., Leonard CM, Altered precision grasping in stumptail macaques after fasciculus cuneatus lesions. Somatosens Mot Res 9, 61–73 (1992). doi: 10.3109/08990229209144763 [DOI] [PubMed] [Google Scholar]
15.Ballermann M, McKenna J, Whishaw IQ, A grasp-related deficit in tactile discrimination following dorsal column lesion in the rat. Brain Res Bull 54, 237–242 (2001). doi: 10.1016/s0361-9230(01)00431-2 [DOI] [PubMed] [Google Scholar]
16.Berkley KJ, Budell RJ, Blomqvist A, Bull M, Output systems of the dorsal column nuclei in the cat. Brain Res 396, 199–225 (1986). doi: 10.1016/0165-0173(86)90012-3 [DOI] [PubMed] [Google Scholar]
17.Loutit AJ, Vickery RM, Potas JR, Functional organization and connectivity of the dorsal column nuclei complex reveals a sensorimotor integration and distribution hub. J Comp Neurol, (2020). doi: 10.1002/cne.24942 [DOI] [PubMed] [Google Scholar]
18.Kuypers HG, Tuerk JD, The Distribution of the Cortical Fibres within the Nuclei Cuneatus and Gracilis in the Cat. J Anat 98, 143–162 (1964). doi: [PMC free article] [PubMed] [Google Scholar]
19.Cheema S, Whitsel BL, Rustioni A, The corticocuneate pathway in the cat: relations among terminal distribution patterns, cytoarchitecture, and single neuron functional properties. Somatosens Res 1, 169–205 (1983). doi: 10.3109/07367228309144547 [DOI] [PubMed] [Google Scholar]
20.Jabbur SJ, Towe AL, Effect of pyramidal tract activity on dorsal column nuclei. Science 132, 547–548 (1960). doi: 10.1126/science.132.3426.547 [DOI] [PubMed] [Google Scholar]
21.Andersen P, Eccles JC, Oshima T, Schmidt RF, Mechanisms of Synaptic Transmission in the Cuneate Nucleus. J Neurophysiol 27, 1096–1116 (1964). doi: 10.1152/jn.1964.27.6.1096 [DOI] [PubMed] [Google Scholar]
22.Rustioni A, Hayes NL, Corticospinal tract collaterals to the dorsal column nuclei of cats. An anatomical single and double retrograde tracer study. Exp Brain Res 43, 237–245 (1981). doi: 10.1007/BF00238364 [DOI] [PubMed] [Google Scholar]
23.Cole JD, Gordon G, Corticofugal actions on lemniscal neurons of the cuneate, gracile and lateral cervical nuclei of the cat. Exp Brain Res 90, 384–392 (1992). doi: 10.1007/BF00227252 [DOI] [PubMed] [Google Scholar]
24.Canedo A, Marino J, Aguilar J, Lemniscal recurrent and transcortical influences on cuneate neurons. Neuroscience 97, 317–334 (2000). doi: 10.1016/s0306-4522(00)00063-4 [DOI] [PubMed] [Google Scholar]
25.Harris F, Jabbur SJ, Morse RW, Towe AL, Influence of the cerebral cortex on the cuneate nucleus of the monkey. Nature 208, 1215–1216 (1965). doi: 10.1038/2081215a0 [DOI] [PubMed] [Google Scholar]
26.Canedo A, Primary motor cortex influences on the descending and ascending systems. Prog Neurobiol 51, 287–335 (1997). doi: 10.1016/s0301-0082(96)00058-5 [DOI] [PubMed] [Google Scholar]
27.Aguilar J, Rivadulla C, Soto C, Canedo A, New corticocuneate cellular mechanisms underlying the modulation of cutaneous ascending transmission in anesthetized cats. J Neurophysiol 89, 3328–3339 (2003). doi: 10.1152/jn.01085.2002 [DOI] [PubMed] [Google Scholar]
28.Rustioni A, Schmechel DE, Cheema S, Fitzpatrick D, Glutamic acid decarboxylase-containing neurons in the dorsal column nuclei of the cat. Somatosens Res 1, 329–357 (1984). doi: 10.3109/07367228409144554 [DOI] [PubMed] [Google Scholar]
29.Popratiloff A, Valtschanoff JG, Rustioni A, Weinberg RJ, Colocalization of GABA and glycine in the rat dorsal column nuclei. Brain Res 706, 308–312 (1996). doi: 10.1016/0006-8993(95)01280-x [DOI] [PubMed] [Google Scholar]
30.Lue JH, Jiang-Shieh YF, Shieh JY, Ling EA, Wen CY, Multiple inputs of GABA-immunoreactive neurons in the cuneate nucleus of the rat. Neurosci Res 27, 123–132 (1997). doi: 10.1016/s0168-0102(96)01139-x [DOI] [PubMed] [Google Scholar]
31.Bengtsson F, Brasselet R, Johansson RS, Arleo A, Jorntell H, Integration of sensory quanta in cuneate nucleus neurons in vivo. PLoS One 8, e56630 (2013). doi: 10.1371/journal.pone.0056630 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Blackwell AA, Banovetz MT, Qandeel IQ Whishaw DG Wallace, The structure of arm and hand movements in a spontaneous and food rewarded on-line string-pulling task by the mouse. Behav Brain Res 345, 49–58 (2018). doi: 10.1016/j.bbr.2018.02.025 [DOI] [PubMed] [Google Scholar]
33.Pruszynski JA, Flanagan JR, Johansson RS, Fast and accurate edge orientation processing during object manipulation. Elife 7, (2018). doi: 10.7554/eLife.31200 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Ghez C, Pisa M, Inhibition of afferent transmission in cuneate nucleus during voluntary movement in the cat. Brain Res 40, 145–155 (1972). doi: 10.1016/0006-8993(72)90120-5 [DOI] [PubMed] [Google Scholar]
35.Coulter JD, Sensory transmission through lemniscal pathway during voluntary movement in the cat. J Neurophysiol 37, 831–845 (1974). doi: 10.1152/jn.1974.37.5.831 [DOI] [PubMed] [Google Scholar]
36.Chapin JK, Woodward DJ, Modulation of sensory responsiveness of single somatosensory cortical cells during movement and arousal behaviors. Exp Neurol 72, 164–178 (1981). doi: 10.1016/0014-4886(81)90135-7 [DOI] [PubMed] [Google Scholar]
37.Chapman CE, Active versus passive touch: factors influencing the transmission of somatosensory signals to primary somatosensory cortex. Can J Physiol Pharmacol 72, 558–570 (1994). doi: 10.1139/y94-080 [DOI] [PubMed] [Google Scholar]
38.Shin HC, Chapin JK, Mapping the effects of motor cortex stimulation on single neurons in the dorsal column nuclei in the rat: direct responses and afferent modulation. Brain Res Bull 22, 245–252 (1989). doi: 10.1016/0361-9230(89)90049-x [DOI] [PubMed] [Google Scholar]
39.Soto C, Aguilar J, Martin-Cora F, Rivadulla C, Canedo A, Intracuneate mechanisms underlying primary afferent cutaneous processing in anaesthetized cats. Eur J Neurosci 19, 3006–3016 (2004). doi: 10.1111/j.0953-816X.2004.03432.x [DOI] [PubMed] [Google Scholar]
40.Witham CL, Baker SN, Modulation and transmission of peripheral inputs in monkey cuneate and external cuneate nuclei. J Neurophysiol 106, 2764–2775 (2011). doi: 10.1152/jn.00449.2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Towe AL, Jabbur SJ, Cortical inhibition of neurons in dorsal column nuclei of cat. J Neurophysiol 24, 488–498 (1961). doi: 10.1152/jn.1961.24.5.488 [DOI] [PubMed] [Google Scholar]
42.McComas AJ, Hypothesis: Hughlings Jackson and presynaptic inhibition: is there a big picture? J Neurophysiol 116, 41–50 (2016). doi: 10.1152/jn.00371.2015 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Blakemore SJ, Frith CD, Wolpert DM, Spatio-temporal prediction modulates the perception of self-produced stimuli. J Cogn Neurosci 11, 551–559 (1999). doi: 10.1162/089892999563607 [DOI] [PubMed] [Google Scholar]
44.Karadimas SK, Satkunendrarajah K, Laliberte AM, Ringuette D, Weisspapir I, Li L, Gosgnach S, Fehlings MG, Sensory cortical control of movement. Nat Neurosci 23, 75–84 (2020). doi: 10.1038/s41593-019-0536-7 [DOI] [PubMed] [Google Scholar]
45.Versteeg C, Rosenow JM, Bensmaia SJ, Miller LE, Encoding of limb state by single neurons in the cuneate nucleus of awake monkeys. J Neurophysiol, (2021). doi: 10.1152/jn.00568.2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Omlor W, Wahl AS, Sipila P, Lutcke H, Laurenczy B, Chen IW, Sumanovski LT, van ‘t Hoff M, Bethge P, Voigt FF, Schwab ME, Helmchen F, Context-dependent limb movement encoding in neuronal populations of motor cortex. Nat Commun 10, 4812 (2019). doi: 10.1038/s41467-019-12670-z [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Poo C, Isaacson JS, Odor representations in olfactory cortex: “sparse” coding, global inhibition, and oscillations. Neuron 62, 850–861 (2009). doi: 10.1016/j.neuron.2009.05.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Adams RA, Shipp S, Friston KJ, Predictions not commands: active inference in the motor system. Brain Struct Funct 218, 611–643 (2013). doi: 10.1007/s00429-012-0475-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Keller GB, Mrsic-Flogel TD, Predictive Processing: A Canonical Cortical Computation. Neuron 100, 424–435 (2018). doi: 10.1016/j.neuron.2018.10.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Suresh AK, Winberry JE, Versteeg C, Chowdhury R, Tomlinson T, Rosenow JM, Miller LE, Bensmaia SJ, Methodological considerations for a chronic neural interface with the cuneate nucleus of macaques. J Neurophysiol 118, 3271–3281 (2017). doi: 10.1152/jn.00436.2017 [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Azim E, azimlabsalk/Tactile-Orienting-Task: Tactile orienting assay for mice, Zenodo; (2021); doi: 10.5281/zenodo.5218877 [DOI] [Google Scholar]
52.Tamamaki N, Yanagawa Y, Tomioka R, Miyazaki J, Obata K, Kaneko T, Green fluorescent protein expression and colocalization with calretinin, parvalbumin, and somatostatin in the GAD67-GFP knock-in mouse. J Comp Neurol 467, 60–79 (2003). doi: 10.1002/cne.10905 [DOI] [PubMed] [Google Scholar]
53.Zeilhofer HU, Studler B, Arabadzisz D, Schweizer C, Ahmadi S, Layh B, Bosl MR, Fritschy JM, Glycinergic neurons expressing enhanced green fluorescent protein in bacterial artificial chromosome transgenic mice. J Comp Neurol 482, 123–141 (2005). doi: 10.1002/cne.20349 [DOI] [PubMed] [Google Scholar]
54.Alhadeff AL, Su Z, Hernandez E, Klima ML, Phillips SZ, Holland RA, Guo C, Hantman AW, De Jonghe BC, Betley JN, A Neural Circuit for the Suppression of Pain by a Competing Need State. Cell 173, 140–152 e115 (2018). doi: 10.1016/j.cell.2018.02.057 [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Huang CC, Sugino K, Shima Y, Guo C, Bai S, Mensh BD, Nelson SB, Hantman AW, Convergence of pontine and proprioceptive streams onto multimodal cerebellar granule cells. Elife 2, e00400 (2013). doi: 10.7554/eLife.00400 [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Knowland D, Lilascharoen V, Pacia CP, Shin S, Wang EH, Lim BK, Distinct Ventral Pallidal Neural Populations Mediate Separate Symptoms of Depression. Cell 170, 284–297 e218 (2017). doi: 10.1016/j.cell.2017.06.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Tervo DG, Hwang BY, Viswanathan S, Gaj T, Lavzin M, Ritola KD, Lindo S, Michael S, Kuleshova E, Ojala D, Huang CC, Gerfen CR, Schiller J, Dudman JT, Hantman AW, Looger LL, Schaffer DV, Karpova AY, A Designer AAV Variant Permits Efficient Retrograde Access to Projection Neurons. Neuron 92, 372–382 (2016). doi: 10.1016/j.neuron.2016.09.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Adams JC, Biotin amplification of biotin and horseradish peroxidase signals in histochemical stains. J Histochem Cytochem 40, 1457–1463 (1992). doi: 10.1177/40.10.1527370 [DOI] [PubMed] [Google Scholar]
59.Zhang JH, Morita Y, Hironaka T, Emson PC, Tohyama M, Ontological study of calbindin-D28k-like and parvalbumin-like immunoreactivities in rat spinal cord and dorsal root ganglia. J Comp Neurol 302, 715–728 (1990). doi: 10.1002/cne.903020404 [DOI] [PubMed] [Google Scholar]
60.de Nooij JC, Doobar S, Jessell TM, Etv1 inactivation reveals proprioceptor subclasses that reflect the level of NT3 expression in muscle targets. Neuron 77, 1055–1068 (2013). doi: 10.1016/j.neuron.2013.01.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Miyamichi K, Shlomai-Fuchs Y, Shu M, Weissbourd BC, Luo L, Mizrahi A, Dissecting local circuits: parvalbumin interneurons underlie broad feedback control of olfactory bulb output. Neuron 80, 1232–1245 (2013). doi: 10.1016/j.neuron.2013.08.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Ramirez DMO, Ajay AD, Goldberg MP, Meeks JP, in Multiphoton Microscopy. Neuromethods, Hartveit E, Ed. (Humana, New York, NY, 2019), vol. 148, pp. 195–224. [Google Scholar]
63.Poinsatte K, Betz D, Torres VO, Ajay AD, Mirza S, Selvaraj UM, Plautz EJ, Kong X, Gokhale S, Meeks JP, Ramirez DMO, Goldberg MP, Stowe AM, Visualization and Quantification of Post-stroke Neural Connectivity and Neuroinflammation Using Serial Two-Photon Tomography in the Whole Mouse Brain. Front Neurosci 13, 1055 (2019). doi: 10.3389/fnins.2019.01055 [DOI] [PMC free article] [PubMed] [Google Scholar]
64.Ting JT, Daigle TL, Chen Q, Feng G, Acute brain slice methods for adult and aging animals: application of targeted patch clamp analysis and optogenetics. Methods Mol Biol 1183, 221–242 (2014). doi: 10.1007/978-1-4939-1096-0_14 [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Giesler GJ Jr., Nahin RL, Madsen AM, Postsynaptic dorsal column pathway of the rat. I. Anatomical studies. J Neurophysiol 51, 260–275 (1984). doi: 10.1152/jn.1984.51.2.260 [DOI] [PubMed] [Google Scholar]
66.Andersen P, Etholm B, Gordon G, Presynaptic and post-synaptic inhibition elicited in the cat’s dorsal column nuclei by mechanical stimulation of skin. J Physiol 210, 433–455 (1970). doi: 10.1113/jphysiol.1970.sp009219 [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Lee C, Jones TA, Effects of Ketamine Compared with Urethane Anesthesia on Vestibular Sensory Evoked Potentials and Systemic Physiology in Mice. J Am Assoc Lab Anim Sci 57, 268–277 (2018). doi: [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Chambers WW, Liu CN, McCouch GP, Inhibition of the dorsal column nuclei. Exp Neurol 7, 13–23 (1963). doi: 10.1016/0014-4886(63)90090-6 [DOI] [PubMed] [Google Scholar]
69.Mathis A, Mamidanna P, Cury KM, Abe T, Murthy VN, Mathis MW, Bethge M, DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat Neurosci 21, 1281–1289 (2018). doi: 10.1038/s41593-018-0209-y [DOI] [PubMed] [Google Scholar]
70.Nath T, Mathis A, Chen AC, Patel A, Bethge M, Mathis MW, Using DeepLabCut for 3D markerless pose estimation across species and behaviors. Nat Protoc 14, 2152–2176 (2019). doi: 10.1038/s41596-019-0176-0 [DOI] [PubMed] [Google Scholar]
71.Guo ZV, Hires SA, Li N, O’Connor DH, Komiyama T, Ophir E, Huber D, Bonardi C, Morandell K, Gutnisky D, Peron S, Xu NL, Cox J, Svoboda K, Procedures for behavioral experiments in head-fixed mice. PLoS One 9, e88678 (2014). doi: 10.1371/journal.pone.0088678 [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Galinanes GL, Bonardi C, Huber D, Directional Reaching for Water as a Cortex-Dependent Behavioral Framework for Mice. Cell Rep 22, 2767–2783 (2018). doi: 10.1016/j.celrep.2018.02.042 [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Zhou X, Wang L, Hasegawa H, Amin P, Han BX, Kaneko S, He Y, Wang F, Deletion of PIK3C3/Vps34 in sensory neurons causes rapid neurodegeneration by disrupting the endosomal but not the autophagic pathway. Proc Natl Acad Sci U S A 107, 9424–9429 (2010). doi: 10.1073/pnas.0914725107 [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Hantman AW, Jessell TM, Clarke’s column neurons as the focus of a corticospinal corollary circuit. Nat Neurosci 13, 1233–1239 (2010). doi: 10.1038/nn.2637 [DOI] [PMC free article] [PubMed] [Google Scholar]
75.Niu J, Ding L, Li JJ, Kim H, Liu J, Li H, Moberly A, Badea TC, Duncan ID, Son YJ, Scherer SS, Luo W, Modality-based organization of ascending somatosensory axons in the direct dorsal column pathway. J Neurosci 33, 17691–17709 (2013). doi: 10.1523/JNEUROSCI.3429-13.2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
76.Hippenmeyer S, Vrieseling E, Sigrist M, Portmann T, Laengle C, Ladle DR, Arber S, A developmental switch in the response of DRG neurons to ETS transcription factor signaling. PLoS Biol 3, e159 (2005). doi: 10.1371/journal.pbio.0030159 [DOI] [PMC free article] [PubMed] [Google Scholar]
77.Delaurier A, Burton N, Bennett M, Baldock R, Davidson D, Mohun TJ, Logan MP, The Mouse Limb Anatomy Atlas: an interactive 3D tool for studying embryonic limb patterning. BMC Dev Biol 8, 83 (2008). doi: 10.1186/1471-213X-8-83 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Fig S1

NIHMS1748970-supplement-Fig_S1.jpg^{(28.7MB, jpg)}

Fig S2

NIHMS1748970-supplement-Fig_S2.jpg^{(1.8MB, jpg)}

Fig S3

NIHMS1748970-supplement-Fig_S3.jpg^{(20.5MB, jpg)}

Fig S4

NIHMS1748970-supplement-Fig_S4.jpg^{(3MB, jpg)}

Fig S5

NIHMS1748970-supplement-Fig_S5.jpg^{(1MB, jpg)}

Fig S6

NIHMS1748970-supplement-Fig_S6.jpg^{(6.3MB, jpg)}

Fig S7

NIHMS1748970-supplement-Fig_S7.jpg^{(9.5MB, jpg)}

Fig S8

NIHMS1748970-supplement-Fig_S8.jpg^{(2.9MB, jpg)}

Fig S9

NIHMS1748970-supplement-Fig_S9.jpg^{(7.8MB, jpg)}

Fig S10

NIHMS1748970-supplement-Fig_S10.jpg^{(20.7MB, jpg)}

Fig S11

NIHMS1748970-supplement-Fig_S11.jpg^{(20MB, jpg)}

Fig S12

NIHMS1748970-supplement-Fig_S12.jpg^{(1.6MB, jpg)}

Table S1

NIHMS1748970-supplement-Table_S1.xlsx^{(119.4KB, xlsx)}

Table S2

NIHMS1748970-supplement-Table_S2.xlsx^{(118KB, xlsx)}

Movie S1

Download video file^{(6.5MB, mov)}

Movie S2

Download video file^{(8MB, mov)}

Movie S3

Download video file^{(11.2MB, mov)}

Movie S4

Download video file^{(12.3MB, mov)}

Data Availability Statement

All data are available in the main text or the supplementary materials. Design and code for the tactile orienting assay are available at (51). www.github.com/azimlabsalk.

[R1] 1.Mountcastle VB, The Sensory Hand : Neural Mechanisms of Somatic Sensation. (Harvard University Press, Cambridge, Mass., 2005), pp. xiv, 616 p. [Google Scholar]

[R2] 2.Johansson RS, Flanagan JR, Coding and use of tactile signals from the fingertips in object manipulation tasks. Nat Rev Neurosci 10, 345–359 (2009). doi: 10.1038/nrn2621 [DOI] [PubMed] [Google Scholar]

[R3] 3.Scott SH, A Functional Taxonomy of Bottom-Up Sensory Feedback Processing for Motor Actions. Trends Neurosci 39, 512–526 (2016). doi: 10.1016/j.tins.2016.06.001 [DOI] [PubMed] [Google Scholar]

[R4] 4.Shadmehr R, Smith MA, Krakauer JW, Error correction, sensory prediction, and adaptation in motor control. Annu Rev Neurosci 33, 89–108 (2010). doi: 10.1146/annurev-neuro-060909-153135 [DOI] [PubMed] [Google Scholar]

[R5] 5.Azim E, Seki K, Gain control in the sensorimotor system. Curr Opin Physiol 8, 177–187 (2019). doi: 10.1016/j.cophys.2019.03.005 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] 6.Sillito AM, Jones HE, Gerstein GL, West DC, Feature-linked synchronization of thalamic relay cell firing induced by feedback from the visual cortex. Nature 369, 479–482 (1994). doi: 10.1038/369479a0 [DOI] [PubMed] [Google Scholar]

[R7] 7.Gilbert CD, Sigman M, Brain states: top-down influences in sensory processing. Neuron 54, 677–696 (2007). doi: 10.1016/j.neuron.2007.05.019 [DOI] [PubMed] [Google Scholar]

[R8] 8.Lee S, Carvell GE, Simons DJ, Motor modulation of afferent somatosensory circuits. Nat Neurosci 11, 1430–1438 (2008). doi: 10.1038/nn.2227 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.Fink AJ, Croce KR, Huang ZJ, Abbott LF, Jessell TM, Azim E, Presynaptic inhibition of spinal sensory feedback ensures smooth movement. Nature 509, 43–48 (2014). doi: 10.1038/nature13276 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] 10.Confais J, Kim G, Tomatsu S, Takei T, Seki K, Nerve-Specific Input Modulation to Spinal Neurons during a Motor Task in the Monkey. J Neurosci 37, 2612–2626 (2017). doi: 10.1523/JNEUROSCI.2561-16.2017 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R11] 11.Liu Y, Latremoliere A, Li X, Zhang Z, Chen M, Wang X, Fang C, Zhu J, Alexandre C, Gao Z, Chen B, Ding X, Zhou JY, Zhang Y, Chen C, Wang KH, Woolf CJ, He Z, Touch and tactile neuropathic pain sensitivity are set by corticospinal projections. Nature 561, 547–550 (2018). doi: 10.1038/s41586-018-0515-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Schneider DM, Sundararajan J, Mooney R, A cortical filter that learns to suppress the acoustic consequences of movement. Nature 561, 391–395 (2018). doi: 10.1038/s41586-018-0520-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Wall PD, The sensory and motor role of impulses travelling in the dorsal columns towards cerebral cortex. Brain 93, 505–524 (1970). doi: 10.1093/brain/93.3.505 [DOI] [PubMed] [Google Scholar]

[R14] 14.Glendinning DS, Cooper BY, Vierck CJ Jr., Leonard CM, Altered precision grasping in stumptail macaques after fasciculus cuneatus lesions. Somatosens Mot Res 9, 61–73 (1992). doi: 10.3109/08990229209144763 [DOI] [PubMed] [Google Scholar]

[R15] 15.Ballermann M, McKenna J, Whishaw IQ, A grasp-related deficit in tactile discrimination following dorsal column lesion in the rat. Brain Res Bull 54, 237–242 (2001). doi: 10.1016/s0361-9230(01)00431-2 [DOI] [PubMed] [Google Scholar]

[R16] 16.Berkley KJ, Budell RJ, Blomqvist A, Bull M, Output systems of the dorsal column nuclei in the cat. Brain Res 396, 199–225 (1986). doi: 10.1016/0165-0173(86)90012-3 [DOI] [PubMed] [Google Scholar]

[R17] 17.Loutit AJ, Vickery RM, Potas JR, Functional organization and connectivity of the dorsal column nuclei complex reveals a sensorimotor integration and distribution hub. J Comp Neurol, (2020). doi: 10.1002/cne.24942 [DOI] [PubMed] [Google Scholar]

[R18] 18.Kuypers HG, Tuerk JD, The Distribution of the Cortical Fibres within the Nuclei Cuneatus and Gracilis in the Cat. J Anat 98, 143–162 (1964). doi: [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Cheema S, Whitsel BL, Rustioni A, The corticocuneate pathway in the cat: relations among terminal distribution patterns, cytoarchitecture, and single neuron functional properties. Somatosens Res 1, 169–205 (1983). doi: 10.3109/07367228309144547 [DOI] [PubMed] [Google Scholar]

[R20] 20.Jabbur SJ, Towe AL, Effect of pyramidal tract activity on dorsal column nuclei. Science 132, 547–548 (1960). doi: 10.1126/science.132.3426.547 [DOI] [PubMed] [Google Scholar]

[R21] 21.Andersen P, Eccles JC, Oshima T, Schmidt RF, Mechanisms of Synaptic Transmission in the Cuneate Nucleus. J Neurophysiol 27, 1096–1116 (1964). doi: 10.1152/jn.1964.27.6.1096 [DOI] [PubMed] [Google Scholar]

[R22] 22.Rustioni A, Hayes NL, Corticospinal tract collaterals to the dorsal column nuclei of cats. An anatomical single and double retrograde tracer study. Exp Brain Res 43, 237–245 (1981). doi: 10.1007/BF00238364 [DOI] [PubMed] [Google Scholar]

[R23] 23.Cole JD, Gordon G, Corticofugal actions on lemniscal neurons of the cuneate, gracile and lateral cervical nuclei of the cat. Exp Brain Res 90, 384–392 (1992). doi: 10.1007/BF00227252 [DOI] [PubMed] [Google Scholar]

[R24] 24.Canedo A, Marino J, Aguilar J, Lemniscal recurrent and transcortical influences on cuneate neurons. Neuroscience 97, 317–334 (2000). doi: 10.1016/s0306-4522(00)00063-4 [DOI] [PubMed] [Google Scholar]

[R25] 25.Harris F, Jabbur SJ, Morse RW, Towe AL, Influence of the cerebral cortex on the cuneate nucleus of the monkey. Nature 208, 1215–1216 (1965). doi: 10.1038/2081215a0 [DOI] [PubMed] [Google Scholar]

[R26] 26.Canedo A, Primary motor cortex influences on the descending and ascending systems. Prog Neurobiol 51, 287–335 (1997). doi: 10.1016/s0301-0082(96)00058-5 [DOI] [PubMed] [Google Scholar]

[R27] 27.Aguilar J, Rivadulla C, Soto C, Canedo A, New corticocuneate cellular mechanisms underlying the modulation of cutaneous ascending transmission in anesthetized cats. J Neurophysiol 89, 3328–3339 (2003). doi: 10.1152/jn.01085.2002 [DOI] [PubMed] [Google Scholar]

[R28] 28.Rustioni A, Schmechel DE, Cheema S, Fitzpatrick D, Glutamic acid decarboxylase-containing neurons in the dorsal column nuclei of the cat. Somatosens Res 1, 329–357 (1984). doi: 10.3109/07367228409144554 [DOI] [PubMed] [Google Scholar]

[R29] 29.Popratiloff A, Valtschanoff JG, Rustioni A, Weinberg RJ, Colocalization of GABA and glycine in the rat dorsal column nuclei. Brain Res 706, 308–312 (1996). doi: 10.1016/0006-8993(95)01280-x [DOI] [PubMed] [Google Scholar]

[R30] 30.Lue JH, Jiang-Shieh YF, Shieh JY, Ling EA, Wen CY, Multiple inputs of GABA-immunoreactive neurons in the cuneate nucleus of the rat. Neurosci Res 27, 123–132 (1997). doi: 10.1016/s0168-0102(96)01139-x [DOI] [PubMed] [Google Scholar]

[R31] 31.Bengtsson F, Brasselet R, Johansson RS, Arleo A, Jorntell H, Integration of sensory quanta in cuneate nucleus neurons in vivo. PLoS One 8, e56630 (2013). doi: 10.1371/journal.pone.0056630 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] 32.Blackwell AA, Banovetz MT, Qandeel IQ Whishaw DG Wallace, The structure of arm and hand movements in a spontaneous and food rewarded on-line string-pulling task by the mouse. Behav Brain Res 345, 49–58 (2018). doi: 10.1016/j.bbr.2018.02.025 [DOI] [PubMed] [Google Scholar]

[R33] 33.Pruszynski JA, Flanagan JR, Johansson RS, Fast and accurate edge orientation processing during object manipulation. Elife 7, (2018). doi: 10.7554/eLife.31200 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Ghez C, Pisa M, Inhibition of afferent transmission in cuneate nucleus during voluntary movement in the cat. Brain Res 40, 145–155 (1972). doi: 10.1016/0006-8993(72)90120-5 [DOI] [PubMed] [Google Scholar]

[R35] 35.Coulter JD, Sensory transmission through lemniscal pathway during voluntary movement in the cat. J Neurophysiol 37, 831–845 (1974). doi: 10.1152/jn.1974.37.5.831 [DOI] [PubMed] [Google Scholar]

[R36] 36.Chapin JK, Woodward DJ, Modulation of sensory responsiveness of single somatosensory cortical cells during movement and arousal behaviors. Exp Neurol 72, 164–178 (1981). doi: 10.1016/0014-4886(81)90135-7 [DOI] [PubMed] [Google Scholar]

[R37] 37.Chapman CE, Active versus passive touch: factors influencing the transmission of somatosensory signals to primary somatosensory cortex. Can J Physiol Pharmacol 72, 558–570 (1994). doi: 10.1139/y94-080 [DOI] [PubMed] [Google Scholar]

[R38] 38.Shin HC, Chapin JK, Mapping the effects of motor cortex stimulation on single neurons in the dorsal column nuclei in the rat: direct responses and afferent modulation. Brain Res Bull 22, 245–252 (1989). doi: 10.1016/0361-9230(89)90049-x [DOI] [PubMed] [Google Scholar]

[R39] 39.Soto C, Aguilar J, Martin-Cora F, Rivadulla C, Canedo A, Intracuneate mechanisms underlying primary afferent cutaneous processing in anaesthetized cats. Eur J Neurosci 19, 3006–3016 (2004). doi: 10.1111/j.0953-816X.2004.03432.x [DOI] [PubMed] [Google Scholar]

[R40] 40.Witham CL, Baker SN, Modulation and transmission of peripheral inputs in monkey cuneate and external cuneate nuclei. J Neurophysiol 106, 2764–2775 (2011). doi: 10.1152/jn.00449.2011 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R41] 41.Towe AL, Jabbur SJ, Cortical inhibition of neurons in dorsal column nuclei of cat. J Neurophysiol 24, 488–498 (1961). doi: 10.1152/jn.1961.24.5.488 [DOI] [PubMed] [Google Scholar]

[R42] 42.McComas AJ, Hypothesis: Hughlings Jackson and presynaptic inhibition: is there a big picture? J Neurophysiol 116, 41–50 (2016). doi: 10.1152/jn.00371.2015 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R43] 43.Blakemore SJ, Frith CD, Wolpert DM, Spatio-temporal prediction modulates the perception of self-produced stimuli. J Cogn Neurosci 11, 551–559 (1999). doi: 10.1162/089892999563607 [DOI] [PubMed] [Google Scholar]

[R44] 44.Karadimas SK, Satkunendrarajah K, Laliberte AM, Ringuette D, Weisspapir I, Li L, Gosgnach S, Fehlings MG, Sensory cortical control of movement. Nat Neurosci 23, 75–84 (2020). doi: 10.1038/s41593-019-0536-7 [DOI] [PubMed] [Google Scholar]

[R45] 45.Versteeg C, Rosenow JM, Bensmaia SJ, Miller LE, Encoding of limb state by single neurons in the cuneate nucleus of awake monkeys. J Neurophysiol, (2021). doi: 10.1152/jn.00568.2020 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R46] 46.Omlor W, Wahl AS, Sipila P, Lutcke H, Laurenczy B, Chen IW, Sumanovski LT, van ‘t Hoff M, Bethge P, Voigt FF, Schwab ME, Helmchen F, Context-dependent limb movement encoding in neuronal populations of motor cortex. Nat Commun 10, 4812 (2019). doi: 10.1038/s41467-019-12670-z [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] 47.Poo C, Isaacson JS, Odor representations in olfactory cortex: “sparse” coding, global inhibition, and oscillations. Neuron 62, 850–861 (2009). doi: 10.1016/j.neuron.2009.05.022 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] 48.Adams RA, Shipp S, Friston KJ, Predictions not commands: active inference in the motor system. Brain Struct Funct 218, 611–643 (2013). doi: 10.1007/s00429-012-0475-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R49] 49.Keller GB, Mrsic-Flogel TD, Predictive Processing: A Canonical Cortical Computation. Neuron 100, 424–435 (2018). doi: 10.1016/j.neuron.2018.10.003 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] 50.Suresh AK, Winberry JE, Versteeg C, Chowdhury R, Tomlinson T, Rosenow JM, Miller LE, Bensmaia SJ, Methodological considerations for a chronic neural interface with the cuneate nucleus of macaques. J Neurophysiol 118, 3271–3281 (2017). doi: 10.1152/jn.00436.2017 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] 51.Azim E, azimlabsalk/Tactile-Orienting-Task: Tactile orienting assay for mice, Zenodo; (2021); doi: 10.5281/zenodo.5218877 [DOI] [Google Scholar]

[R52] 52.Tamamaki N, Yanagawa Y, Tomioka R, Miyazaki J, Obata K, Kaneko T, Green fluorescent protein expression and colocalization with calretinin, parvalbumin, and somatostatin in the GAD67-GFP knock-in mouse. J Comp Neurol 467, 60–79 (2003). doi: 10.1002/cne.10905 [DOI] [PubMed] [Google Scholar]

[R53] 53.Zeilhofer HU, Studler B, Arabadzisz D, Schweizer C, Ahmadi S, Layh B, Bosl MR, Fritschy JM, Glycinergic neurons expressing enhanced green fluorescent protein in bacterial artificial chromosome transgenic mice. J Comp Neurol 482, 123–141 (2005). doi: 10.1002/cne.20349 [DOI] [PubMed] [Google Scholar]

[R54] 54.Alhadeff AL, Su Z, Hernandez E, Klima ML, Phillips SZ, Holland RA, Guo C, Hantman AW, De Jonghe BC, Betley JN, A Neural Circuit for the Suppression of Pain by a Competing Need State. Cell 173, 140–152 e115 (2018). doi: 10.1016/j.cell.2018.02.057 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R55] 55.Huang CC, Sugino K, Shima Y, Guo C, Bai S, Mensh BD, Nelson SB, Hantman AW, Convergence of pontine and proprioceptive streams onto multimodal cerebellar granule cells. Elife 2, e00400 (2013). doi: 10.7554/eLife.00400 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R56] 56.Knowland D, Lilascharoen V, Pacia CP, Shin S, Wang EH, Lim BK, Distinct Ventral Pallidal Neural Populations Mediate Separate Symptoms of Depression. Cell 170, 284–297 e218 (2017). doi: 10.1016/j.cell.2017.06.015 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R57] 57.Tervo DG, Hwang BY, Viswanathan S, Gaj T, Lavzin M, Ritola KD, Lindo S, Michael S, Kuleshova E, Ojala D, Huang CC, Gerfen CR, Schiller J, Dudman JT, Hantman AW, Looger LL, Schaffer DV, Karpova AY, A Designer AAV Variant Permits Efficient Retrograde Access to Projection Neurons. Neuron 92, 372–382 (2016). doi: 10.1016/j.neuron.2016.09.021 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R58] 58.Adams JC, Biotin amplification of biotin and horseradish peroxidase signals in histochemical stains. J Histochem Cytochem 40, 1457–1463 (1992). doi: 10.1177/40.10.1527370 [DOI] [PubMed] [Google Scholar]

[R59] 59.Zhang JH, Morita Y, Hironaka T, Emson PC, Tohyama M, Ontological study of calbindin-D28k-like and parvalbumin-like immunoreactivities in rat spinal cord and dorsal root ganglia. J Comp Neurol 302, 715–728 (1990). doi: 10.1002/cne.903020404 [DOI] [PubMed] [Google Scholar]

[R60] 60.de Nooij JC, Doobar S, Jessell TM, Etv1 inactivation reveals proprioceptor subclasses that reflect the level of NT3 expression in muscle targets. Neuron 77, 1055–1068 (2013). doi: 10.1016/j.neuron.2013.01.015 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R61] 61.Miyamichi K, Shlomai-Fuchs Y, Shu M, Weissbourd BC, Luo L, Mizrahi A, Dissecting local circuits: parvalbumin interneurons underlie broad feedback control of olfactory bulb output. Neuron 80, 1232–1245 (2013). doi: 10.1016/j.neuron.2013.08.027 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R62] 62.Ramirez DMO, Ajay AD, Goldberg MP, Meeks JP, in Multiphoton Microscopy. Neuromethods, Hartveit E, Ed. (Humana, New York, NY, 2019), vol. 148, pp. 195–224. [Google Scholar]

[R63] 63.Poinsatte K, Betz D, Torres VO, Ajay AD, Mirza S, Selvaraj UM, Plautz EJ, Kong X, Gokhale S, Meeks JP, Ramirez DMO, Goldberg MP, Stowe AM, Visualization and Quantification of Post-stroke Neural Connectivity and Neuroinflammation Using Serial Two-Photon Tomography in the Whole Mouse Brain. Front Neurosci 13, 1055 (2019). doi: 10.3389/fnins.2019.01055 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R64] 64.Ting JT, Daigle TL, Chen Q, Feng G, Acute brain slice methods for adult and aging animals: application of targeted patch clamp analysis and optogenetics. Methods Mol Biol 1183, 221–242 (2014). doi: 10.1007/978-1-4939-1096-0_14 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R65] 65.Giesler GJ Jr., Nahin RL, Madsen AM, Postsynaptic dorsal column pathway of the rat. I. Anatomical studies. J Neurophysiol 51, 260–275 (1984). doi: 10.1152/jn.1984.51.2.260 [DOI] [PubMed] [Google Scholar]

[R66] 66.Andersen P, Etholm B, Gordon G, Presynaptic and post-synaptic inhibition elicited in the cat’s dorsal column nuclei by mechanical stimulation of skin. J Physiol 210, 433–455 (1970). doi: 10.1113/jphysiol.1970.sp009219 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R67] 67.Lee C, Jones TA, Effects of Ketamine Compared with Urethane Anesthesia on Vestibular Sensory Evoked Potentials and Systemic Physiology in Mice. J Am Assoc Lab Anim Sci 57, 268–277 (2018). doi: [DOI] [PMC free article] [PubMed] [Google Scholar]

[R68] 68.Chambers WW, Liu CN, McCouch GP, Inhibition of the dorsal column nuclei. Exp Neurol 7, 13–23 (1963). doi: 10.1016/0014-4886(63)90090-6 [DOI] [PubMed] [Google Scholar]

[R69] 69.Mathis A, Mamidanna P, Cury KM, Abe T, Murthy VN, Mathis MW, Bethge M, DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat Neurosci 21, 1281–1289 (2018). doi: 10.1038/s41593-018-0209-y [DOI] [PubMed] [Google Scholar]

[R70] 70.Nath T, Mathis A, Chen AC, Patel A, Bethge M, Mathis MW, Using DeepLabCut for 3D markerless pose estimation across species and behaviors. Nat Protoc 14, 2152–2176 (2019). doi: 10.1038/s41596-019-0176-0 [DOI] [PubMed] [Google Scholar]

[R71] 71.Guo ZV, Hires SA, Li N, O’Connor DH, Komiyama T, Ophir E, Huber D, Bonardi C, Morandell K, Gutnisky D, Peron S, Xu NL, Cox J, Svoboda K, Procedures for behavioral experiments in head-fixed mice. PLoS One 9, e88678 (2014). doi: 10.1371/journal.pone.0088678 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R72] 72.Galinanes GL, Bonardi C, Huber D, Directional Reaching for Water as a Cortex-Dependent Behavioral Framework for Mice. Cell Rep 22, 2767–2783 (2018). doi: 10.1016/j.celrep.2018.02.042 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R73] 73.Zhou X, Wang L, Hasegawa H, Amin P, Han BX, Kaneko S, He Y, Wang F, Deletion of PIK3C3/Vps34 in sensory neurons causes rapid neurodegeneration by disrupting the endosomal but not the autophagic pathway. Proc Natl Acad Sci U S A 107, 9424–9429 (2010). doi: 10.1073/pnas.0914725107 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R74] 74.Hantman AW, Jessell TM, Clarke’s column neurons as the focus of a corticospinal corollary circuit. Nat Neurosci 13, 1233–1239 (2010). doi: 10.1038/nn.2637 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R75] 75.Niu J, Ding L, Li JJ, Kim H, Liu J, Li H, Moberly A, Badea TC, Duncan ID, Son YJ, Scherer SS, Luo W, Modality-based organization of ascending somatosensory axons in the direct dorsal column pathway. J Neurosci 33, 17691–17709 (2013). doi: 10.1523/JNEUROSCI.3429-13.2013 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R76] 76.Hippenmeyer S, Vrieseling E, Sigrist M, Portmann T, Laengle C, Ladle DR, Arber S, A developmental switch in the response of DRG neurons to ETS transcription factor signaling. PLoS Biol 3, e159 (2005). doi: 10.1371/journal.pbio.0030159 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R77] 77.Delaurier A, Burton N, Bennett M, Baldock R, Davidson D, Mohun TJ, Logan MP, The Mouse Limb Anatomy Atlas: an interactive 3D tool for studying embryonic limb patterning. BMC Dev Biol 8, 83 (2008). doi: 10.1186/1471-213X-8-83 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Modulation of tactile feedback for the execution of dexterous movement

James M Conner

Andrew Bohannon

Masakazu Igarashi

James Taniguchi

Nicholas Baltar

Eiman Azim

Abstract

One-Sentence Summary:

Fig. 1. Cuneate tactile and inhibitory inputs.

Cuneate tactile and inhibitory inputs

Local inhibitory modulation of ascending tactile feedback

Fig. 2. Local inhibitory circuits bidirectionally modulate tactile responses.

Disrupting cuneate modulation perturbs dexterous movements

Fig. 3. Perturbing cuneate inhibitory circuits disrupts dexterous movements.

Fig. 4. Activation of cuneate inhibitory circuits compromises tactile-dependent behavior.

Top-down corticofugal control of cuneate circuits

Fig. 5. Distinct corticofugal pathways target cuneate circuits.

Discussion

Materials and Methods

Mice

Viruses

Antibodies and tracers

Immunohistochemistry and imaging

Surgical procedures

General:

Labeling sensory afferents:

a). DRG labeling:

b). Afferent labeling:

c). Proprioceptor labeling:

Targeting local inhibitory circuits within the cuneate region:

a). Labeling local inhibitory neurons:

b). Targeting local inhibitory neurons for electrophysiological recording and perturbation:

c). Targeting GABAergic and glycinergic neurons:

Monosynaptic and disynaptic rabies tracing:

a). Labeling monosynaptic inputs to cuneolemniscal neurons:

b). Labeling monosynaptic inputs to cuneate inhibitory neurons:

c). Disynaptic tracing of inputs to cuneolemniscal neurons:

Labelling cortical axonal projections to the cuneate region:

a). Corticofugal projection labeling:

b). SSp and MOp projection labeling:

c). Dual labeling of SSp and rSM projections to the cuneate region:

d). Selective labeling of SSp corticospinal collaterals to the cuneate:

e). Selective labeling of rSM corticofugal collaterals to the cuneate region:

Slice electrophysiology

Slice preparation:

Whole-cell recording:

Data acquisition and analysis:

Photostimulation:

Drugs and drug application:

In vivo electrophysiology

Animal preparation:

Tactile stimuli:

Cuneate and thalamus recording:

Optogenetic manipulation of cuneate inhibitory neurons:

Assessing rSM corticofugal modulation of tactile responses in the cuneate:

String pulling behavioral assay

Assay:

Optogenetic manipulation of cuneate inhibitory neurons:

Data analysis:

Tactile orienting behavioral assay

Assay:

Head-fixation:

Task structure:

Training period:

Optogenetic manipulation of cuneate inhibitory neurons:

Data analysis:

Statistics and data collection

Supplementary Material

Acknowledgements:

Funding:

Footnotes

Data and materials availability:

References and Notes

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK