Synchrony within Aβ mechanoreceptor subtypes governs signal propagation to primary somatosensory cortex

Abstract

Fast-conducting Aβ slowly adapting type I low threshold mechanoreceptors (SAI-LTMRs) and Aβ rapidly adapting type I (RAI-) LTMRs are critical for discriminative touch from glabrous skin. Recent studies used mouse genetic loss-of-function and optogenetic gain-of-function manipulations to determine that signals from these Aβ LTMRs are integrated subcortically to build the cortical representation of touch. However, the precise influence of each subtype on downstream responses is unclear, in part because previous manipulations lacked the capacity to reproduce physiological firing rates and patterns in individual Aβ LTMR subtypes. Here, we took advantage of a fast variant of channelrhodopsin, CatCh, which enabled us to generate irregular spike trains with mean rates of 20 Hz in either subtype while monitoring responses in mouse primary somatosensory cortex (S1). Doing so revealed that propagation of steady-state signals from sustained responses of Aβ SAI-LTMRs to neural activity (spiking and local field potential) in S1 is dependent on synchronous activation of multiple Aβ SAI-LTMRs. Asynchronous activation of the same Aβ SAI-LTMRs rarely produced sustained responses in S1 measured at the level of single unit spiking as well as local field potentials. This suggests that the irregular firing patterns of Aβ SAI-LTMRs during static indentations contribute to the preponderance of transient cortical responses. By contrast, both synchronous and asynchronous activation of multiple Aβ RAI-LTMRs resulted in robust S1 responses. Overall, the temporal patterning and synchrony of activity within Aβ LTMR subtypes govern how well their signals propagate through the ascending somatosensory system.

Abstract Figure Legend

We used an optogenetic gain-of-function manipulation to investigate how sensory signals from two subtypes of Aβ LTMRs contribute to the cortical representation of touch. CatCh, a sensitive and fast channelrhodopsin variant, enables pulsed-light generation of high-frequency action potential trains in Aβ LTMR subtypes. Synchronous activation of Aβ LTMRs generated larger responses, especially when considering sustained activity, than asynchronous activation. This suggests that the temporal patterning and synchrony of activity within Aβ LTMR subtypes govern how well their signals propagate through the ascending somatosensory system.

Graphical Abstract

Introduction

Our sense of touch is one of the primary ways we interact with the external world (Reed et al., 1996; de Haan & Dijkerman, 2020). Physical stimuli are transduced into electrical impulses by a diverse array of mechanically sensitive peripheral nerve endings (Handler & Ginty, 2021). In mammalian glabrous (non-hairy) skin, Aβ low-threshold mechanoreceptors (LTMRs) are the main transducers of gentle and discriminative touch (Johnson, 2001; Walcher et al., 2018). Two major subtypes of Aβ LTMRs are the Aβ rapidly adapting type I (RAI)-LTMRs that innervate Meissner corpuscles and the Aβ slowly adapting type I (SAI)-LTMRs that are associated with Merkel cells (Johnson, 2001; Handler & Ginty, 2021). Aβ RAI-LTMRs produce action potentials at the onset and offset of ramp-and-hold mechanical stimuli and are especially sensitive to moving stimuli and vibrations at frequencies between 30 and 100 Hz (Johansson & Vallbo, 1979; LaMotte & Whitehouse, 1986; Vega-Bermudez & Johnson, 1999; Neubarth et al., 2020). Aβ SAI-LTMRs, by contrast, produce a burst of action potentials at touch onset and maintain spiking with irregular interspike intervals during the static phase of sustained touch (Iggo & Muir, 1969; Knibestol & Vallbo, 1980; Woodbury & Koerber, 2007; Wellnitz et al., 2010; Maksimovic et al., 2014; Al-Basha & Prescott, 2019).

Sensory inputs from these Aβ LTMRs and other mechanoreceptors are relayed to the primary somatosensory cortex (S1) through the spinal cord, dorsal column nuclei, and the thalamus, where integration and transformation of the signals can occur at each stage (Abraira & Ginty, 2013; Emanuel et al., 2021; Suresh et al., 2021; Chirila et al., 2022). The majority of touch-responsive S1 neurons generate a transient response that resembles the ON/OFF pattern of Aβ RAI-LTMRs in response to sustained touch (Sur et al., 1984; Pei et al., 2009; Callier et al., 2019; Emanuel et al., 2021). In rodents, selective, developmental ablation of glabrous-skin Aβ RAI-LTMRs reduces the amplitude and sensitivity of the cortical response to touch but does not change its transient ON/OFF firing pattern (Emanuel et al., 2021). It remains unclear how inputs from only Aβ SA-LTMRs are transformed from a sustained firing pattern into this transient ON/OFF response pattern. The slow recovery time of the ReaChR opsin we previously used to activate Aβ LTMR subtypes only allowed for the generation of action potentials at low rates (< 5 Hz) and limited our ability to infer the impact of Aβ LTMR firing at physiological rates (Emanuel et al., 2021). Therefore, we expressed an opsin with fast kinetics, CatCh (Kleinlogel et al., 2011), in Aβ RAI-LTMRs or Aβ SAI-LTMRs to determine how trains of action potentials evoked at physiological rates and in physiological patterns in sensory neuron subtypes influence S1 activity.

Methods

Ethical approval

All experimental procedures were approved by Emory University Institutional Animal Care and Use Committee (Protocol 202200102) and were performed in compliance with the Guide for Animal Care and Use of Laboratory Animals. The investigators understand the ethical principles under which the journal operates and this work complies with the animal ethics checklist.

Mice

Mice were housed in temperature- and humidity-controlled facilities in a 12h:12h light:dark cycle with ad libitum access to food and water. To selectively express the CatCh opsin (fused to EYFP) in Aβ SAI-LTMRs and Aβ RAI-LTMRs we generated mice with the genotypes TrkC^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh} and TrkB^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh}, respectively. All alleles were previously described: TrkC^CreER: (Bai et al., 2015), RRID:IMSR_JAX:030291; TrkB^CreER: (Rutlin et al., 2014), RRID:IMSR_JAX:027214 ; Avil^FlpO: (Choi et al., 2020), MGI:6740440; R26^{FSF-LSL-CatCh} [Ai80]: (Daigle et al., 2018); RRID:IMSR_JAX:025109. TrkB^CreER was generously provided by the Garraway lab at Emory. All other alleles were imported to Emory from the Ginty laboratory at Harvard Medical School. To selectively induce Cre recombination in Aβ SAI-LTMRs, we administered 2–3 mg of tamoxifen (dissolved in sunflower seed oil) to the pregnant dam at embryonic day 12.5 (E12.5) via oral gavage. To selectively induce Cre recombination in Aβ RAI-LTMRs, we administered 0.5 mg tamoxifen (i.p.; dissolved in sunflower seed oil) at postnatal day 2–3 (P2–3). All mice were from mixed C57Bl/6J and CD1 backgrounds. A total of 12 TrkC^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh} mice (5 males and 7 females, age: 155.8 ± 170.0 days, weight: 34.6 ± 8.7 g) and 13 TrkB^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh} mice (5 males and 7 females, age: 90.7 ± 31.6 days, weight: 25.9 ± 6.6 g) were included.

In vivo dorsal root ganglion electrophysiology

We administered dexamethasone (2 mg/kg) to adult (>P56) mice 4–12 hours prior to surgery to minimise swelling. Anaesthesia was introduced with urethane (1.5 g/kg body weight i.p.) and maintained using isoflurane (1–2% in inhaled 100% oxygen). After reaching a surgical plane of anaesthesia, which we assessed and monitored through toe-pinch responses, we shaved the back of the mouse and made a midline incision over the lumbar or cervical vertebrae. For recordings from L4 DRG, we applied a spinal clamp (Kopf) to the L3 vertebra to secure the spinal column. We dissected the paravertebral muscles over the L4 vertebra and removed the bone covering the L4 dorsal root ganglion (DRG) with a fine-tipped rongeur. For recordings from C7 DRG, we suspended the T2 vertebrae with a suture to secure the spinal column, dissected the paravertebral muscles over the C6-T1 vertebrae, and removed the bone covering the C7 DRG with a fine-tipped scissor and forceps. The surface of the L4 or C7 DRG was cleaned and then submerged in HEPES-buffered saline.

A 32-channel silicon probe (ASSY-37 H7b, Cambridge NeuroTech) was inserted into the L4 or C7 DRG. We gently brushed the mouse’s hindpaw or forepaw with a paintbrush or a cotton-tipped applicator to identify neurons that respond to mechanical stimulation of the plantar surface. Raw signals were digitised and amplified through an Intan RHD 32 channel headstage (Intan Technologies) and an Open Ephys acquisition board (Siegle et al., 2017). Data acquisition was controlled with Open Ephys Software (version 0.6.7).

For optical stimulation of Aβ LTMRs expressing CatCh, pulses of light were generated by a 300 mW, 445-nm laser (CST-H-445–300, Ultralasers). Light was directed to different locations on the paw through two scanning galvanometer mirrors (GVS002, Thorlabs) and an Fθ lens (FTH160–1064-M39, Thorlabs). The intensity of the laser measured at the skin was 20–22 mW and the beam was focused to a spot size of 5,105 μm². To measure the spot size, we attenuated the laser with a 4.0 neutral density filter and illuminated a lens-less CMOS sensor with an exposure time of 5 μs. Area was calculated from the resulting image binarised at 1/e² maximum intensity.

Laser pulses (0.3 ms in duration; 6.0–6.6 μJ) were delivered to the glabrous skin region of the hindpaw or forepaw in an 8 × 6 mm grid (0.2 mm spacing) to map optical receptive fields of CatCh-expressing neurons. To assess how reliably action potentials follow the optical stimulation, we delivered two laser stimulus sets to the receptive field of that neuron: (1) a series of increasing frequencies ranging from 2 to 40 Hz, and (2) the irregular temporal pattern used for the synchronous stimulation (10 pulses in 0.5 s; see below).

After the DRG recording, mice were euthanised by cervical dislocation. The brain and paws were extracted and preserved with post-fixation overnight (4% paraformaldehyde for brain and Zamboni fixative for paws). The tissue samples were then washed with PBS, cryoprotected with 30% sucrose, and frozen embedded in optimal cutting temperature (OCT) compound.

Histology

We sliced glabrous skin pedal pads into 25 μm sections using a cryostat (Leica) and collected sections on glass slides (Fisher Superfrost Plus). We stained sections by washing twice for 5 min each with 1x PBS containing 0.1% Triton X-100 (0.1% PBST), then incubating with blocking solution (5% normal donkey serum [Jackson ImmunoResearch Labs 017–000-121] in 0.3% PBST) for 1 h at room temperature and then incubating with primary antibodies in blocking solutions at 4 °C overnight. Primary antibodies included chicken polyclonal anti-green fluorescent protein (Aves Labs GFP-1020; RRID: AB_10000240; 1:500), rabbit polyclonal anti-S100β (ProteinTech 15146–1-AP; RRID: AB_2254244; 1:500), and rat monoclonal anti-Troma1 (DSHB AB_531826; 1:200). We then then washed the tissue four times (15 min) with 0.1% PBST and subsequently incubated with secondary antibodies (donkey anti-chicken conjugated to Alexa 488, Invitrogen A78948; RRID: AB_2921070; donkey anti-rabbit conjugated to Alexa 546, Invitrogen A10040; RRID: AB_2534016; donkey anti-rat conjugated to Alexa 647, Invitrogen A48272; RRID: AB_2893138; all 1:500) in blocking solution for 2 h at room temperature. We washed three times (15 min) with 0.1% PBST and included DAPI solution (5 μg/mL; AAT Bioquest) in the second wash. Finally, we washed with 1x PBS for 15 min and mounted with Fluoromount-G (Southern Biotech). We imaged tissue with a Keyence BZX-7000 automated fluorescence microscope.

To visualise the DiI electrode tract, we cryosectioned brains into 40-μm floating sections and mounted them onto glass slides. We stained the sections with DAPI (5 μg/mL in PBS) for 15 min, washed with PBS for 5 min, and mounted with Fluoromount-G. We imaged tissue with a Keyence BZX-7000 automated fluorescence microscope. Electrode tracts were manually aligned to the atlas (Allen Mouse Common Coordinate Framework ([CCFv3–2017, Wang et al., 2020] in the EBRAINS interactive atlas viewer) to determine the anterior-posterior coordinate relative to bregma. The medial-lateral coordinate was measured by from the histological image and normalised by multiplying the measured value by the ratio of the Atlas brain width (8.4 mm) to the actual brain width. We were only able to recover a subset of electrode tracts (11 of 23 forepaw S1 sessions and 8 of 9 hindpaw S1 sessions) because they either had insufficient DiI staining or were too close to other tracks to resolve.

Surgery and craniotomy

We administered dexamethasone (2 mg/kg) to adult (>P56) mice 4–12 hours prior to surgery to minimise swelling. We used dual-mode analgesia to minimise pain and discomfort associated with surgery. Immediately before the procedure, we administered sustained-release meloxicam (2–4 mg/kg sq; Wedgewood Pharmacy). During surgery, mice were anaesthetised with 1.5–2% isoflurane and locally at the incision with Lidocaine (1%, 0.03 mL). We removed the scalp and periosteum, dried the skull, and mounted a titanium headplate to the skull with dental acrylic (Metabond). We made a circular craniotomy (approximately 1 mm in diameter) above forepaw S1 (right hemisphere: 2.10 mm lateral, 0.00 mm posterior to bregma) or hindpaw S1 (right hemisphere: 1.65 mm lateral, 0.60 mm posterior to bregma. We sealed the craniotomy with Kwik-Sil or Kwik-Cast (WPI) and cemented an aluminium ring onto the headplate to provide a well for a recording bath solution. Extended-release buprenorphine (0.5–1 mg/kg sq; Wedgewood Pharmacy) was administered post-operatively for analgesia. Mice recovered overnight before the first recording session.

In vivo cortex electrophysiology

All S1 recordings were performed in awake, head-fixed mice. Prior to each recording, the mouse was habituated to head fixation for 10 minutes. We removed the craniotomy cap and submerged the craniotomy in a HEPES-buffered saline solution (150 mM NaCl, 2.5 mM KCl, and 10 mM HEPES; pH = 7.4). We inserted a Neuropixels 1.0 probe coated with DiI (D3911, Thermo Fisher) into forepaw S1 or hindpaw S1 of the right hemisphere and advanced the tip of the probe to approximately 1,600 μm below the dura. To verify S1 placement, we gently brushed the mouse’s skin on the contralateral side (left) with a fine paintbrush. We monitored brush-evoked spikes from multiple recording channels to verify the approximate location of the receptive field. If the receptive field was not on the forepaw or hindpaw pedal pads, we removed the probe from the brain and reinserted it into a new location. We restrained the left forepaw or the left hindpaw with vinyl tape in a natural position over a rectangular aperture cut into an acrylic platform supporting the mouse.

We acquired data using the Neuropixels PXIe control system (IMEC) and the Open Ephys GUI software (Siegle et al., 2017). Raw signals were digitised, amplified, filtered, and saved into two bands: AP band, which was digitised at 30 kHz, amplified 500x, and bandpass-filtered at 300–10,000 Hz, and LFP band, which was digitised at 2.5 kHz, amplified 250x, bandpass-filtered at 0.5–500 Hz.

Optical skin stimulation during S1 recordings

In the first experiment, we delivered all light stimuli (generated with the same stimulator described for DRG recordings) to the centre of the forepaw contralateral to the recording site. Stimulation patterns included regularly spaced trains of laser pulses at 5, 20, and 40 Hz over 0.5 s as well as a train of laser pulses that emulated the previously recorded (Emanuel et al., 2021) response of an Aβ SAI-LTMR to a 40-mN, 0.5-s step indentation (“physiological” laser stimulation). Each laser pattern was repeated 50 times with 5-s inter-trial intervals. For this experiment, each stimulus was presented in a separate block of trials.

In a separate group of mice, we delivered the laser pulses across the 5 pedal pads on the mouse’s paw. The x and y coordinates of the centre of the 5 pedal pads on the left forepaw or hindpaw were calibrated. In each trial, trains of laser pulses were delivered either synchronously or asynchronously to each of 5 locations equally spaced on a circle (0.6-mm diameter) centred on one pad. Each train included 10 laser pulses within 0.5 s. For synchronous stimulation, pulses were delivered to each of the five locations in rapid succession, with a 0.4-ms delay between adjacent pulses (the minimum interval possible given the response time of the galvanometer mirrors). For asynchronous stimulation, we delivered pulses to the five locations with inter-pulse intervals of at least 3 ms, preventing synchronous activation. For each of the 5 pads, both the synchronous stimulation and asynchronous stimulation trials were repeated for 30 trials, resulting in a total of 300 trials per session. For this experiment, the trials were interleaved and their order randomised.

Spike sorting

We concatenated all AP band (300–10,000 Hz) binary files from a session for spike sorting, which we performed with Kilosort 2.5 (Pachitariu et al., 2024) and then manually curated with the Phy2 GUI. We classified clusters as putative single units if they exhibited (1) large spike amplitudes relative to baseline noise, (2) waveforms that were distinct from neighbouring clusters, and (3) a clear refractory period (>1 ms) evident in the autocorrelogram. We only included single units in subsequent analyses and excluded all action potentials from other clusters.

Laminar and cell-type identification

We determined the laminar locations of individual S1 units and LFP channels based on the spatial distribution of waveforms along the probe, which we registered to physiological indicators of cortical layers, as described and validated previously (Emanuel et al., 2021). We defined the centre of layer IV as the depth corresponding to the earliest current sink in the local field potential current source density (CSD) at the onset of skin indentation. To do so, we low-pass filtered voltage waveforms from the LFP band (0.5–500 Hz) at 250 Hz with a third-order Butterworth filter and then calculated the second derivative of this signal across laminar locations on the probe. We rigidly adjusted the depth of S1 units and LFP channels so that the centre of layer IV identified in the CSD plot was assigned to 476 μm below the cortical surface. S1 units and LFP channels were classified into cortical layers according to the following layer depths: layer II/III: 119–416.5 μm; layer IV: 416.5–535.5 μm; layer V: 535.5–952 μm; layer VI: 952–1200 μm. We classified S1 units as regular-spiking (RS) or fast-spiking (FS) based on the trough-to-peak times of the spike waveform on the channel with the largest amplitude (Bartho et al., 2004; Niell & Stryker, 2008). RS units were designated as those with a trough-to-peak time >0.55 ms and FS units were designated as those with a trough-to-peak time ≤0.55 ms.

Analysis of optically evoked spiking activity

For optical stimulation sessions, we calculated z-scored firing rates in 20-ms bins using the mean and standard deviation in the baseline time window (0.8 s before the first laser pulse onset in each trial). We used a parameter-free ZETA test (Montijn et al., 2021) to identify whether each unit showed a time-locked modulation of spiking activity in response to optical stimulation with a p-value cutoff of 0.05. To identify sustained responses, we applied the ZETA test to the sustained time window (200–500 ms) for certain stimulus patterns.

Analysis of local field potential response to optical stimulation

We bandpass-filtered the LFP waveform at 1–250 Hz with a third-order Butterworth filter, aligned LFP waveforms to the onset of optical stimulation of each trial, and computed the mean LFP waveforms for each pad for synchronous and asynchronous stimulation conditions. For each recording session, we calculated the mean LFP waveform across all channels in Layer IV. We then measured the magnitude of waveforms during the stimulus onset window (0–60 ms) and sustained window (200–500 ms). When more than one session was conducted in the same animal, we took the mean of the LFP responses across sessions prior to running statistical tests.

Statistics

We report values as mean ± SD. Percentages of responsive units are reported as weighted mean ± weighted SD, where both the mean and SD are weighted by the number of units from each session.

We used Student’s t-tests to compare the fractions of hindpaw or forepaw units responsive to optical stimulation in TrkC and TrkB animals.

We used Levene’s test to evaluate differences in variability of individual unit responses between activation of Aβ SAI-LTMRs (TrkC) and Aβ RAI-LTMRs (TrkB).

We used two-way ANOVAs with stimulus condition (repeated measure) and genotype as factors as a statistical test for LFP analyses. Summary tables included in the appendix.

We used linear mixed effects models (LMEs) for additional statistical comparisons. Our experiments were designed primarily to detect differences between synchronous and asynchronous stimulus conditions and secondarily to detect differences between activation of Aβ RAI-LTMRs and Aβ SAI-LTMRs. We did not design our experiments to detect differences but did include both male and female animals. Therefore, we used three fixed effects (genotype, sex, and one of stimulus condition, layer, and cell type). We used animal ID as a random effect to account for the hierarchical nature of our dataset (multiple units recorded from the same animal). We calculated p-values of fixed effects and interactions according to Satterthwaite’s method. We acknowledge that the random-effects variance may be estimated imprecisely with fewer than 10 animals per group. We used the lme4 and lmerTest packages in R (version 4.5.3) to fit LMEs (Bates et al., 2015; Kuznetsova et al., 2017). Post hoc paired t-tests with Bonferroni corrections for multiple comparisons were conducted at both the animal and unit levels if there was a significant interaction between the fixed effects.

List of LMEs (LME summary tables included in the appendix):

• Predict number of pads that generated S1 forepaw unit responses with fixed effects of stimulus condition (synchronous or asynchronous), genotype (TrkC or TrkB), and sex (male or female) and random effects of animal ID and unit ID (accounts for repeated measures). Slopes within animals were not fixed because they may vary due to variability in the exact probe location within the somatotopic representation of forepaw S1 and the proportion of Aβ SAI-LTMRs or Aβ RAI-LTMRs that express CatCh.

• Predict number of pads that generated S1 hindpaw unit responses with fixed effects of stimulus condition (synchronous or asynchronous), genotype (TrkC or TrkB), and sex (male or female) and random effect of animal ID and unit ID. Slopes within animals were not fixed because they may vary due to variability in the exact probe location within the somatotopic representation of hindpaw S1 and the proportion of Aβ SAI-LTMRs or Aβ RAI-LTMRs that express CatCh.

• Predict Z-scored firing rates (onset [ON] or sustained [SUS] period) of S1 forepaw units with fixed effects of stimulus condition (synchronous or asynchronous), genotype (TrkC or TrkB), and sex (male or female) and random effect of animal ID and unit ID. Slopes within animals were not fixed because they may vary due to variability in the exact probe location within the somatotopic representation of forepaw S1 and the proportion of Aβ SAI-LTMRs or Aβ RAI-LTMRs that express CatCh.

• Predict Z-scored firing rates (ON or SUS period) of S1 hindpaw units with fixed effects of stimulus condition (synchronous or asynchronous), genotype (TrkC or TrkB), and sex (male or female) and random effect of animal ID and unit ID. Slopes within animals were not fixed because they may vary due to variability in the exact probe location within the somatotopic representation of forepaw S1 and the proportion of Aβ SAI-LTMRs or Aβ RAI-LTMRs that express CatCh.

• Predict difference of Z-scored firing rates (ON or SUS period) between synchronous and asynchronous conditions of forepaw S1 units with fixed effects of layer (layers II/III, IV, V, or VI), genotype (TrkC or TrkB), and sex (male or female) and random effect of animal ID.

• Predict difference of Z-scored firing rates (ON or SUS period) between synchronous and asynchronous conditions of hindpaw S1 units with fixed effects of layer (layers II/III, IV, V, or VI), genotype (TrkC or TrkB), and sex (male or female) and random effect of animal ID.

• Predict difference of Z-scored firing rates (ON or SUS period) between synchronous and asynchronous conditions of forepaw S1 units with fixed effects of cell type (RS or FS) , genotype (TrkC or TrkB), and sex (male or female) and random effect of animal ID.

• Predict difference of Z-scored firing rates (ON or SUS period) between synchronous and asynchronous conditions of hindpaw S1 units with fixed effects of cell type (RS or FS) , genotype (TrkC or TrkB), and sex (male or female) and random effect of animal ID.

Data associated with this study are available on the DANDI Archive (RRID:SCR_017571): https://dandiarchive.org/dandiset/001788 (Liu et al., 2026).

Results

Activation of the opsin CatCh generates high-frequency firing in Aβ LTMRs

For both mouse lines (TrkC^CreER;Avil^FlpO;R26^{LSL-FSF-Catch} and TrkB^CreER;Avil^FlpO;R26^{LSL-FSF-Catch}), only axons of the large-diameter sensory neurons of interest expressed CatCh-EYFP within forepaw glabrous skin (Fig. 1A). Merkel cells, which we identified with an antibody against Troma1, were innervated by a CatCh-EYFP-expressing axon (labelled with an antibody against GFP) but Meissner corpuscles, which we identified with an S100b antibody that labels lamellar cells at the apex of dermal papillae, were not innervated in TrkC^CreER;Avil^FlpO;R26^{LSL-FSF-Catch} mice. Meissner corpuscles, but not Merkel cells, were associated with CatCh-EYFP axons in TrkB^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh} mice. We first validated that the kinetics of the CatCh opsin were sufficiently fast to enable Aβ LTMRs to follow high-frequency trains of laser stimulation. To do so, we recorded from Aβ LTMRs in L4 or C7 DRGs with multiunit electrode arrays while optically stimulating the glabrous skin of the mouse’s ipsilateral hindpaw or forepaw. We mapped the regions of the skin that generated an action potential when illuminated (i.e., optical receptive fields) by directing pulse trains of 10 spikes in 0.25 s (rate of 40 Hz) to spots on a grid centred on the suspected receptive field hot spot (example in Fig. 1B). When the stimulus was directed to the most distal part of the optical receptive field, which corresponded to the skin location responsive to brushing, action potentials followed the laser patterns reliably. Less reliable responses were achieved when the laser was directed to more proximal sites (likely activating the axon within the nerve rather than at its terminal) in that an action potential only fired in response to the first laser pulse.

Figure 1. CatCh expression enables high-frequency responses to pulsed optical stimulation of Aβ LTMR terminal.

(A) Aβ LTMR subtypes selectively expressed CatCh-EYFP in immunostained forepaw sections of a TrkC^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh-EYFP} mouse (left) and a TrkB^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh-EYFP} mouse (right). Scale bar: 50 μm. Arrows indicate Troma1+ Merkel cells (head and shaft; at the base of the epidermis) and S100+ Meissner corpuscles (head only; in the dermal papillae) that were associated with GFP+ (GFP antibody binds EYFP) fibres (red) or not associated with GFP+ fibres (white). (B) The optical receptive field of an Aβ RAI-LTMR measured by delivering 10-pulse trains (at 40 Hz) to multiple locations on the glabrous skin of a mouse’s hindpaw. The colour indicates the probability of laser-evoked action potentials. Raster plots show action potentials generated by the stimulus when directed at the axon terminal (top) or the nerve (bottom). DRG recording schematic created with BioRender.com. (C) Raster plot showing action potentials generated by a series of laser pulses in ascending frequencies (2–40 Hz). (D) As in (C) for a train of 10 laser pulses with variable interpulse intervals (mean pulse rate of ~20 Hz). (E) The probability of evoked action potentials for each stimulus in four L4 Aβ RAI-LTMRs and five C7 Aβ RAI-LTMRs. Black lines and error bars indicate the mean and SD across all LTMRs.

Then, we stimulated the region of the optical receptive field that most reliably followed repeated illumination with trains of laser pulses that varied in frequency and regularity. In the example neuron, the probability of action potentials after each laser pulse was 100% for trains at frequencies between 2 and 20 Hz and was 92.5% for trains at a frequency of 40 Hz (Fig. 1C). The probability of action potentials after each laser pulse was also 100% for a train of pulses at 20 Hz with irregular pulse intervals (Fig. 1D). The latency of responses to laser pulses in the L4 DRG was 6.09 ± 0.83 ms (mean ± SD) and the latency in C7 DRG was 4.87 ± 0.55 ms. Similar responses were evoked in the majority of CatCh-expressing Aβ LTMRs tested, including those sampled from both L4 or C7 DRGs. The probability of action potentials after each laser pulse was greater than 95% for 2–20 Hz pulse trains and was 74.9% ± 37.3% for the 40 Hz and 86.3% ± 22.5% the irregular-pulse train (L4 DRG: 4 LTMRs from 4 animals [3F]; C7 DRG: 5 LTMRs from 4 animals [2F]; Fig. 1E). This demonstrates that CatCh expression enables LTMRs to follow pulse trains at frequencies up to 40 Hz.

Selective activation of Aβ SAI-LTMRs drives S1 neurons

We next examined how S1 neurons respond to optogenetically evoked spike trains in Aβ SAI-LTMRs. In TrkC^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh} mice (n = 4 [3 female]) that expressed the opsin CatCh in Aβ SAI-LTMRs, optical stimulation directed to the centre of the forepaw generated robust S1 responses. Light-responsive units (n = 273 from 8 recording sessions) are shown in Fig. 2. At a low stimulation frequency (5 Hz), units generated a response to each laser pulse. At higher frequencies, the spike patterns in the sustained phase of 32.2% ± 18.8% (weighted mean ± weighted SD; 88 of 273 total units) and 26.0% ± 13.0% (70 of 273) of units differed from baseline periods at 20 and 40 Hz, respectively. A similar pattern was observed in response to a physiological pulse train built from a recorded spiking response of a TrkC⁺ Aβ SAI-LTMR during a 0.5-s, 40-mN skin indentation (data from Emanuel et al., 2021); the spiking pattern during the sustained phase differed from that at baseline in 25.6% ± 22.5% of units (71 of 273).

Figure 2. S1 responses to high-frequency activation of Aβ SAI-LTMRs.

(A) Top: An illustration of the optical stimulation and S1 recording setup. Created with BioRender.com. Bottom: Temporal patterns of optical stimulation, which were delivered to the centre of the left forepaw. (B) Raster plots of one example S1 unit in response to each stimulus pattern. (C) Heatmaps of z-scored firing rate responses of S1 units during optical stimulation at the paw centre with each pattern. Units are sorted by depth from the cortex surface with dashed lines indicating layer boundaries. (D) Z-scored firing rates (grand mean ± SD) of S1 units (n = 273 units from 8 sessions in 4 animals).

The proportion of sustained responses to 20–40 Hz optical stimuli are not similar to previously reported responses to mechanical stimuli; evoked activity during the sustained phase was rarely observed in forepaw S1 in response to 20-mN indentations (only in 5 of 834 units; Emanuel et al., 2021). One potential explanation for this difference is that the pulsed light activates multiple Aβ SAI-LTMRs synchronously whereas static indentation generates irregular spiking patterns in Aβ SAI-LTMRs (Iggo & Muir, 1969; Wellnitz et al., 2010). The irregularity of Aβ SAI-LTMR spiking during static indentation likely results in desynchronisation of their population-level firing pattern. Our artificially imposed synchrony may generate signals that propagate to S1, including during the sustained period.

Synchronous activation of Aβ LTMRs generates sustained S1 activity

Therefore, we next directly tested whether the synchronicity of activation of Aβ SAI-LTMRs governs the propagation of their signals to S1 using another cohort of TrkC^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh} mice (TrkC; n = 8 [4 female]). For comparison, we applied the same stimuli to TrkB^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh} mice (TrkB; n = 7 [4 female]) to measure S1 unit responses to synchronous and asynchronous optical stimulation of Aβ RAI-LTMRs. We designed optical stimulation patterns to emulate synchronous or asynchronous recruitment of Aβ LTMRs with irregular pulses at mean rates of 20 Hz, which corresponds to the sustained firing rate of Aβ SA-LTMRs at moderate forces (~20 mN; Emanuel et al., 2021). On each trial, we delivered laser pulses to five discrete locations within one of five pedal pads of the mouse’s forepaw either synchronously or asynchronously (Fig. 3A).

Figure 3. Forepaw S1 responses to selective synchronous or asynchronous activation of Aβ SAI-LTMRs or Aβ RAI-LTMRs.

(A) On each trial, synchronous or asynchronous stimulation patterns were delivered to five different locations on one of five forepaw pedal pads. (B) Raster plot example (top) and heatmaps (bottom) of S1 unit responses to optical stimulation of Aβ SAI-LTMRs (TrkC) with synchronous and asynchronous patterns at each of the five pads on the mouse left forepaw. Raster plot sorted by stimulus location, which is indicated by colour. Responsive units grouped by recording sessions (n = 695 units from 13 sessions from 8 mice). Within each recording session, units sorted by depth from the cortex surface. (C) As in (B) for responses to optical stimulation of Aβ RAI-LTMRs (TrkB; n = 773 responsive units from 10 sessions from 7 mice). (D) The number of responsive pads during synchronous and asynchronous activation at the animal (left) and individual unit (right) level. Left: Connected markers indicate means within each animal while the black lines and error bar indicate the mean and SD across animals. Linear mixed effects model (LME) p values indicated for fixed effects and their interactions, *p < 0.05. Circle: female, triangle: male. Right: Bars represent mean across units and error bars indicate the SD across units. (E) Left: Example image of an electrode track used to register recording sites. Magenta: DiI; Blue: DAPI. Scale bar: 500 μm. Right: Schematic of recording sites. Each dot represents one recording site. Unit: mm. Paw and brain schematics created with BioRender.com.

Many S1 units (47.3% ± 12.5%; 695 of 1469 units collected in 13 recording sessions from 8 mice [4 female]) responded to optical activation of Aβ SAI-LTMRs with a change in firing pattern, typically an increase in firing rate (Fig. 3B). A similar fraction (60.3% ± 23.5%; 773 of 1282 units collected in 10 recording sessions from 7 mice [4 female]; t(13) = 1.80, p = 0.0867) responded to optical activation of Aβ RAI-LTMRs (Fig. 3C). For subsequent analyses, we only considered responsive neurons.

We quantified the number of pads that recruited in each S1 unit as a proxy for the size of the receptive field and used a linear mixed effects model (LME) with fixed effects of stimulus condition (asynchronous or synchronous) and genotype (TrkC or TrkB) and a random effect of animal ID to compare across conditions while accounting for the hierarchical data structure (Fig. 3D). This revealed that the number of pads recruited with activation of Aβ LTMRs was greater with synchronous stimulation than with asynchronous stimulation and that the number of pads recruited with stimulation of Aβ SAI-LTMRs was smaller than the number of pads recruited with Aβ RAI-LTMRs (Fig. 3D). We did not detect an interaction between these fixed effects. Histological examination of the electrode tracks revealed that recording locations across sessions were tightly clustered near the targeted coordinates (Fig. 3E).

We also delivered the same optical stimulation to the pedal pads of the mouse’s hindpaw while measuring S1 unit responses in TrkC^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh} mice (n = 4 [2 female]) and TrkB^CreER;Avil^FlpO;R26^{LSL-FSF-CatCh} mice (n = 5 [2 female]; Fig. 4A). 38.2% ± 10.6% of hindpaw S1 units (174 of 456 units collected in 4 recording sessions from 4 mice) responded to optical activation of Aβ SAI-LTMRs (Fig. 4B). A statistically indistinguishable fraction of hindpaw S1 units (56.3% ± 14.6%; 284 of 504 units collected in 5 recording sessions from 5 mice; t(7) = 1.672, p = 0.138) responded to optical activation of Aβ RAI-LTMRs (Fig. 4C).

Figure 4. Hindpaw S1 responses to selective synchronous or asynchronous activation of Aβ SAI-LTMRs or Aβ RAI-LTMRs.

(A) On each trial, synchronous or asynchronous stimulation patterns were delivered to five different locations on one of five forepaw pedal pads. (B) Raster plot example (top) and heatmaps (bottom) of S1 unit responses to optical stimulation of Aβ SAI-LTMRs (TrkC) with synchronous and asynchronous patterns at each of the five pads on the mouse left hindpaw. Responsive units grouped by recording sessions (n = 174 units from 4 sessions from 4 mice). Within each recording session, units sorted by depth from the cortex surface. (C) As in (B) for responses to optical stimulation of Aβ RAI-LTMRs (TrkB; n = 284 units from 5 sessions from 5 mice). (D) The number of responsive pads during synchronous and asynchronous activation at the animal (left) and individual unit (right) level. Left: Connected markers indicate means within each animal while the black lines and error bar indicate the mean and SD across animals. LME p values indicated for fixed effects and their interactions. *p < 0.05. Circle: female, triangle: male. Right: Bars represent mean across units and error bars indicate the SD across units. (E) Left: Example image of an electrode track used to register recording locations. Magenta: DiI; Blue: DAPI. Scale bar: 500 μm. Right: Schematic of recording locations. Each dot represents one recording location. Unit: mm. Paw and brain schematics created with BioRender.com.

As seen in forepaw S1, activation of Aβ LTMRs with synchronous stimulation recruited hindpaw S1 responses from more pads than with asynchronous stimulation and stimulation of Aβ SAI-LTMRs recruited responses from fewer pads than stimulation of Aβ RAI-LTMRs (significant fixed effects of stimulus condition and genotype with no interaction; Fig. 4D). Post hoc histological examination of the electrode tracks revealed that recording locations across sessions were tightly clustered near the targeted coordinates (Fig. 4E).

To facilitate comparisons of recordings across neurons, recording sessions, and animals, we identified the stimulus condition that generated the largest response for each responsive S1 neuron and compared synchronous and asynchronous stimulation of that pad (Fig. 5). We used LMEs with fixed effects of stimulus condition (asynchronous or synchronous) and genotype (TrkC or TrkB) and animal ID as a random effect to predict z-scored firing rates at the onset (0 to 60 ms after first pulse) or sustained (200 to 500 ms after first pulse) phases of the stimulus.

Figure 5. S1 responses to synchronous and asynchronous activation of Aβ SAI-LTMRs and Aβ RAI-LTMRs.

(A) Heatmaps of forepaw (top) and hindpaw (bottom) S1 units in response to optical stimulation of Aβ SAI-LTMRs (TrkC) or Aβ RAI-LTMRs (TrkB) with synchronous and asynchronous patterns. Each row represents the z-score firing rate of one unit when synchronous and asynchronous stimuli were applied to the pad that generated the largest response. Units sorted by depth from the cortex surface with dashed lines indicating layer boundaries. (B) Scatterplots showing the baseline-subtracted firing rate onset (left) and sustained (right) responses of individual units for synchronous and asynchronous activation of Aβ SAI-LTMRs (top) or Aβ RAI-LTMRs (bottom). The large points and error bars indicate mean ± SD firing rates of all fast-spiking (FS) or regular-spiking (RS) units (forepaw TrkC: 695 units [248 FS; 447 RS] from 13 recordings in 8 animals [4 female]; forepaw TrkB: 773 units [257 FS; 516 RS] from 10 recording in 7 animals [4 females]; hindpaw TrkC: 174 units [63 FS; 111 RS] from 4 recordings in 4 animals; hindpaw TrkB: 284 units [84 FS; 200 RS] from 5 recordings in 5 animals).

We found a significant interaction (Fig. 6A) between the fixed effects of stimulus condition and genotype when predicting the onset response and post hoc paired t-tests revealed that synchronous stimulation of Aβ SAI-LTMRs produced larger onset responses than asynchronous stimulation, but the same was not true of stimulation of Aβ RAI-LTMRs. There was no interaction (Fig. 6B) between the fixed effects when predicting the sustained response and both fixed effects of stimulus condition and genotype were significant, indicating that synchronous stimulation of either Aβ LTMR subtype results in larger sustained responses and that responses after activation of Aβ RAI-LTMRs were larger than those after activation of Aβ SAI-LTMRs.

Figure 6. Synchronous activation of Aβ SAI-LTMRs and Aβ RAI-LTMRs generates greater sustained S1 response across layers and cell types.

(A) Onset and (B) sustained z-scored firing rate of forepaw S1 units at the animal (top) and individual unit level (bottom) in response to synchronous (sync) and asynchronous (async) stimulation of Aβ SAI-LTMRs (TrkC) or Aβ RAI-LTMRs (TrkB). Connected markers indicate mean of unit responses within each animal while the black lines and error bar indicate the mean and SD across animals. Circle: female, triangle: male. The boxes show the quartiles of the z-scored firing rates (FR) while the whiskers extend to 1.5 inter-quartile range of the lower and upper quartile. (C) Difference in z-scored firing rate (synchronous - asynchronous) for onset responses of forepaw S1 units across all layers. (D) as in (C) for sustained responses. (E) Difference in z-scored firing rate (synchronous - asynchronous) for onset responses of regular-spiking (RS) and fast-spiking (FS) units. (F) as in (E) for sustained responses. (G)-(L) as in (A)-(F) for hindpaw S1 units in response to hindpaw optical stimulation. LME p values indicated for fixed effects and their interactions, *p < 0.05.

The effect of stimulus condition (synchronous versus asynchronous) was similar in both regular-spiking (RS) and fast-spiking (FS) units and across layers. There were no interactions between layer and genotype or fixed effects of layer when we used LMEs to predict the difference in z-scored firing rates between responses to synchronous and asynchronous stimulation at onset or sustained periods (Figs. 6C-D). The fixed effect of genotype was significant at the onset response (Fig. 6C), reflecting the interaction reported in Fig. 6A. Similarly, there were no interactions between cell type and genotype, or fixed effects of cell type when we used separate LMEs to predict the difference in z-scored firing rates between responses to synchronous and asynchronous stimulation at onset or sustained periods (Figs. 6E-F).

We used the same approach to compare hindpaw S1 responses to synchronous and asynchronous optical activation of Aβ SAI-LTMRs or Aβ RAI-LTMRs that innervate the hindpaw (Figs. 6G-L). For hindpaw S1, we did not find a significant fixed effect of stimulus condition or genotype on the onset response (Fig. 6G). For the sustained response, fixed effects of stimulus condition and genotype were both significant, indicating that synchronous stimulation of Aβ LTMRs resulted in stronger sustained responses than asynchronous stimulation of Aβ LTMRs and that Aβ SAI-LTMRs resulted in smaller sustained responses than Aβ RAI-LTMRs (Fig. 6H). As for forepaw S1, the effect of stimulus condition (synchronous versus asynchronous) was similar in both regular-spiking (RS) and fast-spiking (FS) hindpaw S1 units and across layers. There were no interactions between cell type and genotype or fixed effects of layer when we used LMEs to predict the difference in z-scored firing rates between responses to synchronous and asynchronous stimulation at onset or sustained periods (Figs. 6I-J). Similarly, there were no interactions between cell type and genotype or fixed effects of layer when we used separate LMEs to predict the difference in z-scored firing rates between responses to synchronous and asynchronous stimulation at onset or sustained periods (Figs. 6K-L).

In both forepaw S1 and hindpaw S1, stimulation of Aβ RAI-LTMRs, whether synchronous or asynchronous, generated larger variability in onset and sustained responses between individual units than stimulation of Aβ SAI-LTMRs (S1 forepaw: ON: synchronous: F(1, 1466) = 4.492, p = 0.0342; asynchronous: F(1, 1466) = 16.158, p < 0.001; sustained: synchronous: F(1, 1466) = 35.548, p < 0.001; asynchronous: F(1, 1466) = 32.764, p < 0.001; S1 hindpaw: ON: synchronous: F(1, 456) = 8.062, p = 0.00472; asynchronous: F (1, 456) = 9.225, p = 0.00252; sustained: synchronous: F(1, 456) = 28.249, p < 0.001; asynchronous: F(1, 456) = 29.587, p < 0.001; Levene’s test).

Synchronous activation of Aβ SAI-LTMRs generates stronger sustained local field potentials in forepaw S1

To evaluate the summed activity of S1 neuron populations, we extracted the local field potentials (LFPs) during synchronous and asynchronous activation of Aβ LTMRs. Trial-averaged LFPs revealed responses to optical activation of either cell type that were largest in layer IV (Fig. 7A). To quantify, we therefore calculated the mean layer IV LFP responses from the pedal pad that evoked the largest response. The magnitude of the onset response did not differ between stimulus conditions or genotypes for both forepaw S1 and hindpaw S1 (Fig. 7B). For forepaw S1, synchronous activation of both Aβ LTMR subtypes generated stronger sustained LFP activity than asynchronous activation (Fig. 7B). S1 hindpaw LFP responses were smaller than S1 forepaw LFP responses, the responses evoked by Aβ RAI-LTMRs were larger than those evoked by Aβ SAI-LTMRs and we did not detect any significant differences between synchronous and asynchronous activation (Fig. 7B).

Figure 7. S1 local field potentials during synchronous and asynchronous activation of Aβ LTMR subtypes.

(A) Trial-averaged LFPs (mean ± SD) during synchronous and asynchronous activation of Aβ SAI-LTMRs across S1 layers from an example session. (B) Mean Layer IV LFP magnitude during synchronous and asynchronous activation of Aβ SAI-LTMRs (TrkC; S1 forepaw: n = 8 animals [4 female]; S1 hindpaw: n = 4 animals [2 female]) and Aβ RAI-LTMRs (TrkB; S1 forepaw: n = 7 animals [4 female]; S1 hindpaw: n = 5 animals [2 female]). Black lines and error bars indicate the mean and SD across all animals. P values: main effects of condition (cond), genotype (geno), and interaction (interact) between condition and genotype from two-way ANOVAs.

Discussion

In this study, we used an opsin with fast kinetics (CatCh) to assess how selective activation of Aβ LTMR subtypes at physiological rates contributes to S1 responses. Using in vivo DRG recordings, we validated that pulsed optical stimulation of the axon terminals of CatCh-expressing Aβ LTMRs can reliably produce action potentials, even at rates as high as 40 Hz. Then, we investigated how stimulation of Aβ LTMR subtypes with distinct temporal patterns contributes to the S1 cortical response. We found that selectively activating either Aβ SAI-LTMRs or Aβ RAI-LTMRs generates responses in many S1 neurons, consistent with previous observations (Emanuel et al., 2021). Unexpectedly, activation of Aβ SAI-LTMRs with physiological temporal patterns generated prolonged increased firing in a subset of S1 neurons, perhaps because stimulation in the centre of the paw tended to recruit multiple Aβ SAI-LTMRs synchronously (Fig. 2). Synchronous activation of Aβ SAI-LTMRs and Aβ RAI-LTMRs generated stronger responses in S1 than asynchronous activation, especially when considering sustained activity.

While selective activation of CatCh-expressing Aβ SAI-LTMRs or of Aβ RAI-LTMRs generated responses in S1, their response properties were not identical. Selective activation of Aβ RAI-LTMRs tended to produce larger responses in S1 than selective activation of Aβ SAI-LTMRs, consistent with responses to selective activation of ReaChR-expressing Aβ LTMRs (Emanuel et al., 2021). Furthermore, while activation of Aβ SAI-LTMRs or Aβ RAI-LTMRs both generated reliable onset responses in S1, activation of Aβ RAI-LTMRs generated stronger sustained activity that showed larger variability between individual units. Despite the subcortical convergence and integration of inputs from Aβ LTMR subtypes (Emanuel et al., 2021; Suresh et al., 2021; Chirila et al., 2022), Aβ RAI-LTMRs and Aβ SAI-LTMRs appear to differentially recruit subcortical circuits that regulate S1 response properties.

The characterisation of DRG responses revealed that functional CatCh was expressed along the axon fibres of sensory neurons. Thus, optical stimulation of the skin will activate the axon terminal as well as some fibres. However, only the axon terminal consistently followed high frequency laser pulse trains. Therefore, stimulating Aβ LTMRs with high frequency laser pulses principally recruits the axon terminals of the LTMRs. The LMEs revealed substantial variation in responses at the animal-to-animal level, which likely arose due to stochastic labelling of Aβ LTMRs as well as the variability in the precise location of the recording electrode relative to the S1 somatotopic map. Our asynchronous and synchronous stimulation patterns were targeted to five spots within each pedal pad without knowledge of the actual location of CatCh-expressing axon terminals. This, along with the propensity of skin to scatter light, makes it difficult to estimate the number of neurons recruited with each laser pulse. Nevertheless, the differential effects of synchronous versus asynchronous activation across both subtypes indicate that the spatial pattern of illumination recruited multiple distinct neurons during the sustained phase of stimulation.

Our results demonstrate that synchronous activation of Aβ SAI-LTMRs or Aβ RAI-LTMRs generates signals that propagate through the ascending pathways and drive cortex more effectively than asynchronous activation. This is consistent with predictions from neural network models, in which synchronous and asynchronous action potentials regulate the propagation of signals and multiplex rate coding and temporal coding in the central nervous system (Kumar et al., 2008; Ratte et al., 2013). Synchronous spiking facilitates stable signal propagation through neural networks (Kumar et al., 2008; Zhou et al., 2010) whereas asynchronous spiking and its associated variability can stabilise communication and increase information coding (Padmanabhan & Urban, 2010; Gast et al., 2024; Terada & Toyoizumi, 2024).

Distinct stimuli are likely to differentially drive synchronous and asynchronous firing in Aβ LTMRs. As noted above, irregular spiking patterns are signatures of Aβ SAI-LTMR static responses and are expected to produce asynchronous spiking patterns during sustained indentations. However, Aβ SAI-LTMRs produce larger responses with more regularity during dynamic stimuli (Iggo & Muir, 1969), which are likely to be synchronised across Aβ SAI-LTMRs with nearby receptive fields. The resultant enhanced propagation may allow for efficient cortical encoding of stimulus shapes and contours with high spatial contrast. A similar multiplexed representation has been observed in the olfactory bulb and S1 (Friedrich et al., 2004; Lankarany et al., 2019). With multiplexing of Aβ SAI-LTMRs signals, different components of the response could be used to influence separate downstream processes. Salient, high-contrast features from dynamic sampling may efficiently propagate through the system for perception through synchronous spiking. Maintained stimulus contact, however, that is encoded through asynchronous action potentials may be used in the spinal cord or the brainstem to modulate signals from other sensory neurons to, for example, regulate pain (Gautam et al., 2024). Non-stimulus factors, such as brain and body rhythms may also regulate synchronisation within the somatosensory system, which could contribute to previously reported dependencies of cortical sensory evoked potentials on the phase of the heartbeat or alpha rhythms in the nervous system (Al et al., 2020; Makeig et al., 2002). Conversely, the degree of synchronisation of Aβ LTMR signals may be a critical factor for determining whether sensory input entrains or resets the phases of ongoing brain rhythms (Makeig et al., 2002).

Enhanced synchrony in the peripheral somatosensory system is associated with nerve injury and pain perception (Devor & Wall, 1990; Kim et al., 2016; Zheng et al., 2022; Chen et al., 2023). The transmission of asynchronous spikes in dorsal column axons to the somatosensory cortex is blocked by feedforward inhibition and fails to produce paraesthesia (Sagalajev et al., 2024). Our results support the idea that synchronous peripheral neuron activation and convergence of multiple sensory neurons onto downstream targets facilitate propagation through the ascending somatosensory system. Developing ways to control the extent to which skin or peripheral nerve stimulation synchronises or desynchronises activity within and across sensory-neuron populations could enhance the capacity to modulate brain activity for therapeutics and better provide haptic feedback for prostheses (Graczyk et al., 2016; Valle et al., 2018; Pfeifer et al., 2021; Valle et al., 2024).

Key Points.

• CatCh, a sensitive and fast channelrhodopsin variant, enables pulsed-light generation of physiological spiking patterns in Aβ low threshold mechanoreceptor (LTMR) subtypes.

• The synchrony of multiple mechanoreceptors within a subtype controls the extent to which their signals influence activity in primary somatosensory cortex.

• Aβ rapidly adapting type I (RAI-) LTMRs drive stronger cortical responses than Aβ slowly adapting type I (SAI-) LTMRs, especially when considering asynchronous activation of Aβ LTMRs within each subtype.

Appendix: Full Summaries of LME and ANOVA Tests

Fig. 3D

lm_npads_fp <- lmer(npads ~ cond + geno + sex + cond:geno + cond:sex + (cond|animalID) + (1|animalID:unitID), data=npads_fp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	1.392	0.999	1.783	6.631	10.629	0.00004
fixed		condsync	0.493	0.211	0.776	3.263	10.997	0.00755
fixed		genoTrkC	−0.623	−1.094	−0.151	−2.469	11.220	0.03078
fixed		sexM	−0.200	−0.676	0.273	−0.789	11.448	0.44629
fixed		condsync:genoTrkC	0.173	−0.169	0.514	0.948	11.714	0.36233
fixed		condsync:sexM	−0.024	−0.369	0.320	−0.132	12.021	0.89747
ran_pars	animalID:unitID	sd__(Intercept)	0.979
ran_pars	animalID	sd__(Intercept)	0.462
ran_pars	animalID	sd__condsync	0.329
ran_pars	animalID	cor__(Intercept).condsync	−0.360
ran_pars	Residual	sd__Observation	0.695

CI: confidence interval

Intra-class correlation (ICC) = 0.708

Fig. 4D

lm_npads_hp <- lmer(npads ~ cond + geno + sex + cond:geno + cond:sex + (cond|animalID) + (1|animalID:unitID), data=npads_hp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	1.063	0.577	1.528	3.992	5.550	0.00842
fixed		condsync	0.658	0.266	1.037	3.076	5.647	0.02353
fixed		genoTrkC	−0.782	−1.324	−0.253	−2.613	5.683	0.04203
fixed		sexM	0.171	−0.354	0.713	0.573	5.630	0.58847
fixed		condsync:genoTrkC	−0.025	−0.444	0.426	−0.103	5.807	0.92127
fixed		condsync:sexM	0.058	−0.386	0.480	0.242	5.743	0.81729
ran_pars	animalID:unitID	sd__(Intercept)	0.904
ran_pars	animalID	sd__(Intercept)	0.409
ran_pars	animalID	sd__condsync	0.323
ran_pars	animalID	cor__(Intercept).condsync	0.786
ran_pars	Residual	sd__Observation	0.715

ICC = 0.690

Fig. 6A

lm_fp_ON <- lmer(mfrON ~ cond + geno + sex + cond:geno + cond:sex + (cond|animalID) + (1|animalID:unitID), data=mfr_fp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	1.029	0.840	1.220	10.155	10.012	0.00000
fixed		condsync	−0.067	−0.172	0.039	−1.196	10.054	0.25931
fixed		genoTrkC	−0.295	−0.534	−0.061	−2.347	11.216	0.03828
fixed		sexM	−0.080	−0.320	0.159	−0.625	12.023	0.54343
fixed		condsync:genoTrkC	0.303	0.171	0.432	4.361	11.162	0.00110
fixed		condsync:sexM	0.059	−0.072	0.192	0.838	11.872	0.41849
ran_pars	animalID:unitID	sd__(Intercept)	1.205
ran_pars	animalID	sd__(Intercept)	0.194
ran_pars	animalID	sd__condsync	0.112
ran_pars	animalID	cor__(Intercept).condsync	−0.766
ran_pars	Residual	sd__Observation	0.455

ICC = 0.877

Fig. 6B

lm_fp_SUS <- lmer(mfrSUS ~ cond + geno + sex + cond:geno + cond:sex + (cond|animalID) + (1|animalID:unitID), data=mfr_fp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.221	0.151	0.291	5.956	10.815	0.00010
fixed		condsync	0.286	0.185	0.386	5.327	10.108	0.00032
fixed		genoTrkC	−0.113	−0.198	−0.024	−2.450	12.253	0.03024
fixed		sexM	−0.053	−0.143	0.034	−1.135	13.191	0.27674
fixed		condsync:genoTrkC	−0.098	−0.221	0.025	−1.491	11.135	0.16370
fixed		condsync:sexM	−0.067	−0.190	0.059	−0.996	11.741	0.33927
ran_pars	animalID:unitID	sd__(Intercept)	0.304
ran_pars	animalID	sd__(Intercept)	0.069
ran_pars	animalID	sd__condsync	0.108
ran_pars	animalID	cor__(Intercept).condsync	0.660
ran_pars	Residual	sd__Observation	0.417

ICC = 0.384

Fig. 6C

lm_diff_layer_fp_ON <- lmer(diffON ~ layer+geno+sex + layer:geno+ layer:sex + (1|animalID), data=diff_fp2)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.019	−0.134	0.172	0.232	38.750	0.81744
fixed		layerl4	−0.058	−0.263	0.147	−0.551	1435.980	0.58204
fixed		layerl5	−0.117	−0.260	0.024	−1.606	1435.698	0.10847
fixed		layerl6	−0.126	−0.296	0.045	−1.443	1430.612	0.14928
fixed		genoTrkC	0.310	0.117	0.502	3.019	46.587	0.00410
fixed		sexM	0.012	−0.188	0.210	0.112	52.921	0.91154
fixed		layerl4:genoTrkC	−0.064	−0.315	0.188	−0.500	1433.684	0.61719
fixed		layerl5:genoTrkC	0.023	−0.160	0.206	0.246	1434.792	0.80592
fixed		layerl6:genoTrkC	−0.056	−0.291	0.175	−0.470	1433.494	0.63848
fixed		layerl4:sexM	−0.096	−0.356	0.164	−0.722	1431.264	0.47041
fixed		layerl5:sexM	0.066	−0.128	0.256	0.670	1434.239	0.50324
fixed		layerl6:sexM	0.189	−0.061	0.428	1.511	1434.925	0.13100
ran_pars	animalID	sd__(Intercept)	0.115
ran_pars	Residual	sd__Observation	0.646

ICC = 0.031

ANOVA

	Sum Sq	Mean Sq	NumDF	DenDF	F value	Pr(>F)
layer	1.861	0.620	3	1433.184	1.487	0.21618
geno	6.112	6.112	1	13.129	14.656	0.00206
sex	0.192	0.192	1	13.937	0.461	0.50842
layer:geno	0.446	0.149	3	1432.579	0.357	0.78432
layer:sex	1.991	0.664	3	1432.682	1.592	0.18952

Fig. 6D

lm_diff_layer_fp_SUS <- lmer(diffSUS ~ layer+geno+sex + layer:geno+ layer:sex + (1|animalID), data=diff_fp2)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.302	0.158	0.446	3.929	37.857	0.00035
fixed		layerl4	0.006	−0.181	0.197	0.061	1435.866	0.95160
fixed		layerl5	−0.017	−0.146	0.117	−0.248	1435.928	0.80408
fixed		layerl6	0.015	−0.141	0.174	0.190	1430.499	0.84962
fixed		genoTrkC	−0.125	−0.306	0.055	−1.297	45.298	0.20107
fixed		sexM	−0.117	−0.303	0.070	−1.184	51.269	0.24193
fixed		layerl4:genoTrkC	0.029	−0.204	0.259	0.245	1433.388	0.80629
fixed		layerl5:genoTrkC	0.024	−0.147	0.190	0.277	1434.506	0.78169
fixed		layerl6:genoTrkC	−0.003	−0.219	0.210	−0.026	1433.266	0.97961
fixed		layerl4:sexM	−0.003	−0.243	0.235	−0.025	1431.083	0.98003
fixed		layerl5:sexM	0.059	−0.119	0.235	0.654	1434.000	0.51291
fixed		layerl6:sexM	0.118	−0.108	0.341	1.027	1434.722	0.30442
ran_pars	animalID	sd__(Intercept)	0.111
ran_pars	Residual	sd__Observation	0.595

ICC = 0.034

ANOVA

	Sum Sq	Mean Sq	NumDF	DenDF	F value	Pr(>F)
layer	0.607	0.202	3	1433.020	0.572	0.63334
geno	0.885	0.885	1	13.529	2.502	0.13677
sex	0.370	0.370	1	14.328	1.046	0.32329
layer:geno	0.055	0.018	3	1432.393	0.052	0.98455
layer:sex	0.502	0.167	3	1432.532	0.473	0.70128

Fig. 6E

lm_diff_celltype_fp_ON <- lmer(diffON ~ cellType+geno+sex + cellType:geno + cellType:sex + (1|animalID), data=diff_fp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	−0.094	−0.220	0.032	−1.395	20.389	0.17794
fixed		cellTypeRS	0.035	−0.074	0.142	0.635	1455.790	0.52534
fixed		genoTrkC	0.322	0.164	0.479	3.815	23.371	0.00087
fixed		sexM	0.066	−0.097	0.229	0.764	26.336	0.45138
fixed		cellTypeRS:genoTrkC	−0.014	−0.152	0.127	−0.194	1457.189	0.84585
fixed		cellTypeRS:sexM	−0.013	−0.159	0.135	−0.171	1457.806	0.86386
ran_pars	animalID	sd__(Intercept)	0.112
ran_pars	Residual	sd__Observation	0.644

ICC = 0.030

Fig. 6F

lm_diff_celltype_fp_SUS <- lmer(diffSUS ~ cellType+geno+sex + cellType:geno + cellType:sex + (1|animalID), data=diff_fp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.330	0.212	0.448	5.203	19.992	0.00004
fixed		cellTypeRS	−0.057	−0.155	0.042	−1.126	1455.568	0.26038
fixed		genoTrkC	−0.135	−0.283	0.012	−1.711	22.879	0.10057
fixed		sexM	−0.108	−0.260	0.044	−1.329	25.686	0.19557
fixed		cellTypeRS:genoTrkC	0.044	−0.083	0.172	0.673	1456.941	0.50119
fixed		cellTypeRS:sexM	0.062	−0.074	0.196	0.895	1457.552	0.37068
ran_pars	animalID	sd__(Intercept)	0.107
ran_pars	Residual	sd__Observation	0.590

ICC = 0.032

Fig. 6G

lm_hp_ON <- lmer(mfrON ~ cond + geno + sex + cond:enoGroup + cond:sex + (cond|animalID) + (1|animalID:unitID), data=mfr_hp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.580	0.429	0.738	7.500	13.200	0.00000
fixed		condsync	0.149	0.035	0.265	2.292	3.943	0.08462
fixed		genoTrkC	−0.145	−0.327	0.031	−1.614	15.313	0.12697
fixed		sexM	0.197	0.020	0.373	2.229	14.232	0.04241
fixed		condsync:genoTrkC	0.019	−0.116	0.147	0.251	4.184	0.81338
fixed		condsync:sexM	0.002	−0.126	0.134	0.026	4.077	0.98069
ran_pars	animalID:unitID	sd__(Intercept)	0.809
ran_pars	animalID	sd__(Intercept)	0.045
ran_pars	animalID	sd__condsync	0.086
ran_pars	animalID	cor__(Intercept).condsync	1.000
ran_pars	Residual	sd__Observation	0.325

Random effect variances not available.

Fig. 6H

lm_npads_hp <- lmer(npads ~ cond + geno + sex + cond:geno + cond:sex + (cond|animalID) + (1|animalID:unitID), data=npads_hp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.151	0.100	0.205	5.775	239.410	0.00000
fixed		condsync	0.202	0.104	0.292	3.879	5.914	0.00842
fixed		genoTrkC	−0.130	−0.191	−0.069	−4.231	269.647	0.00003
fixed		sexM	−0.012	−0.071	0.047	−0.400	254.487	0.68929
fixed		condsync:genoTrkC	−0.101	−0.204	0.009	−1.704	6.289	0.13705
fixed		condsync:sexM	0.060	−0.048	0.168	1.012	6.122	0.34990
ran_pars	animalID:unitID	sd__(Intercept)	0.172
ran_pars	animalID	sd__(Intercept)	0.003
ran_pars	animalID	sd__condsync	0.068
ran_pars	animalID	cor__(Intercept).condsync	−1.000
ran_pars	Residual	sd__Observation	0.265

Random effect variances not available.

Fig. 6I

lm_diff_layer_hp_ON <- lmer(diffON ~ layer+geno+sex + layer:geno+ layer:sex + (1|animalID), data=diff_hp2)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.012	−0.189	0.209	0.108	27.360	0.91476
fixed		layerl4	0.204	−0.062	0.468	1.496	440.809	0.13532
fixed		layerl5	0.157	−0.059	0.371	1.413	440.276	0.15841
fixed		layerl6	0.153	−0.116	0.424	1.087	427.975	0.27782
fixed		genoTrkC	0.182	−0.016	0.393	1.564	20.199	0.13336
fixed		sexM	0.218	0.002	0.421	1.830	21.930	0.08080
fixed		layerl4:genoTrkC	−0.114	−0.396	0.185	−0.763	440.555	0.44567
fixed		layerl5:genoTrkC	−0.237	−0.465	−0.012	−2.036	440.872	0.04235
fixed		layerl6:genoTrkC	−0.298	−0.587	0.002	−1.960	442.638	0.05066
fixed		layerl4:sexM	−0.544	−0.839	−0.256	−3.629	440.012	0.00032
fixed		layerl5:sexM	−0.208	−0.431	0.027	−1.755	442.911	0.07998
fixed		layerl6:sexM	−0.151	−0.449	0.144	−0.984	441.023	0.32579
ran_pars	animalID	sd__(Intercept)	0.094
ran_pars	Residual	sd__Observation	0.452

ICC = 0.042

ANOVA

	Sum Sq	Mean Sq	NumDF	DenDF	F value	Pr(>F)
layer	0.585	0.195	3	441.956	0.953	0.41475
geno	0.012	0.012	1	4.947	0.059	0.81759
sex	0.002	0.002	1	4.897	0.010	0.92600
layer:geno	1.139	0.380	3	442.007	1.855	0.13650
layer:sex	2.799	0.933	3	442.412	4.559	0.00370

Fig. 6J

lm_diff_layer_hp_SUS <- lmer(diffSUS ~ layer+geno+sex + layer:geno+ layer:sex + (1|animalID), data=diff_hp2)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.219	0.050	0.387	2.375	37.400	0.02278
fixed		layerl4	−0.031	−0.244	0.191	−0.273	441.418	0.78463
fixed		layerl5	−0.016	−0.188	0.166	−0.176	441.224	0.86021
fixed		layerl6	−0.008	−0.221	0.228	−0.068	432.792	0.94574
fixed		genoTrkC	−0.196	−0.370	−0.022	−2.047	27.840	0.05015
fixed		sexM	0.234	0.058	0.414	2.398	30.161	0.02289
fixed		layerl4:genoTrkC	0.091	−0.149	0.325	0.742	441.238	0.45844
fixed		layerl5:genoTrkC	0.094	−0.095	0.276	0.986	441.462	0.32487
fixed		layerl6:genoTrkC	0.170	−0.079	0.406	1.367	442.779	0.17233
fixed		layerl4:sexM	−0.112	−0.350	0.127	−0.908	440.858	0.36433
fixed		layerl5:sexM	−0.209	−0.400	−0.024	−2.159	442.955	0.03140
fixed		layerl6:sexM	−0.345	−0.592	−0.105	−2.746	441.712	0.00628
ran_pars	animalID	sd__(Intercept)	0.078
ran_pars	Residual	sd__Observation	0.371

ICC = 0.042

ANOVA

	Sum Sq	Mean Sq	NumDF	DenDF	F value	Pr(>F)
layer	0.427	0.142	3	442.237	1.037	0.37607
geno	0.345	0.345	1	6.999	2.508	0.15729
sex	0.138	0.138	1	6.929	1.005	0.34980
layer:geno	0.269	0.090	3	442.277	0.653	0.58114
layer:sex	1.179	0.393	3	442.572	2.861	0.03656

Fig. 6K

lm_diff_celltype_hp_ON <- lmer(diffON ~ cellType+geno+sex + cellType:geno + cellType:sex + (1|animalID), data=diff_hp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.197	0.044	0.345	2.226	12.298	0.04544
fixed		cellTypeRS	−0.070	−0.230	0.100	−0.825	452.000	0.40969
fixed		genoTrkC	0.011	−0.153	0.183	0.110	11.700	0.91395
fixed		sexM	0.029	−0.141	0.195	0.297	11.962	0.77124
fixed		cellTypeRS:genoTrkC	0.002	−0.180	0.191	0.022	450.086	0.98250
fixed		cellTypeRS:sexM	−0.037	−0.229	0.142	−0.390	451.402	0.69656
ran_pars	animalID	sd__(Intercept)	0.088
ran_pars	Residual	sd__Observation	0.459

ICC = 0.036

Fig. 6L

lm_diff_celltype_hp_SUS <- lmer(diffSUS ~ cellType+geno+sex + cellType:geno + cellType:sex + (1|animalID), data=diff_hp)

effect	group	term	estimate	CI 2.5	CI 97.5	statistic	df	p.value
fixed		(Intercept)	0.228	0.102	0.356	3.248	19.250	0.00418
fixed		cellTypeRS	−0.039	−0.169	0.102	−0.570	451.937	0.56929
fixed		genoTrkC	−0.147	−0.289	−0.007	−1.889	18.186	0.07501
fixed		sexM	0.067	−0.074	0.209	0.851	18.648	0.40577
fixed		cellTypeRS:genoTrkC	0.063	−0.091	0.212	0.814	450.983	0.41605
fixed		cellTypeRS:sexM	0.000	−0.154	0.148	0.003	451.792	0.99734
ran_pars	animalID	sd__(Intercept)	0.067
ran_pars	Residual	sd__Observation	0.375

ICC = 0.036

Fig. 7B

Forepaw S1 ON

ANOVA

Source	SS	DF1	DF2	MS	F	p
genotype	14531.742	1	13	14531.742	2.555	0.13395
cond	90.936	1	13	90.936	0.102	0.75399
Interaction	708.774	1	13	708.774	0.799	0.38777

Forepaw S1 sustained

ANOVA

Source	SS	DF1	DF2	MS	F	p
genotype	259.499	1	13	259.499	0.744	0.40401
cond	2852.066	1	13	2852.066	29.395	0.00012
Interaction	25.480	1	13	25.480	0.263	0.61693

Hindpaw S1 ON

ANOVA

Source	SS	DF1	DF2	MS	F	p
genotype	273.404	1	7	273.404	0.142	0.71756
cond	135.776	1	7	135.776	0.365	0.56487
Interaction	1773.571	1	7	1773.571	4.766	0.06534

Hindpaw S1 sustained

ANOVA

Source	SS	DF1	DF2	MS	F	p
genotype	622.429	1	7	622.429	35.851	0.00055
cond	346.618	1	7	346.618	5.142	0.05768
Interaction	10.531	1	7	10.531	0.156	0.70443

Data Availability

Data are publicly available on the DANDI Archive (RRID:SCR_017571) at the following URL: https://dandiarchive.org/dandiset/001788.

Code Availability

Analysis code is publicly available in GitHub at https://github.com/EmanuelLab/Liu_Worthy_Emanuel_2026.