Krause corpuscles are genital vibrotactile sensors for sexual behaviours

Mouse lines

Mice were handled according to protocols approved by the Harvard Medical Area Standing Committee on Animals and are in accordance with federal guidelines. Female and male adult mice were used for the experiments. Mice were housed in a temperature-controlled and humidity-controlled facility, maintained under a 12 h–12 h light–dark cycle, and were given food and water ad libitum. All of the mice used in this study have been previously described, including TrkB^creER (JAX, 027214)¹³, Ret^creER (MGI, 4437245)¹⁴, Advillin^cre (JAX, 032536)⁵¹, Advillin^FlpO (ref. ⁵²), Ret^CFP (MGI, 3777555)⁵³, TrkB^flox54, Brn3a^cKOAP (JAX, 010558)⁵⁵, Tau^FSF-iAP (ref. ⁵⁶), Mrgprd^GFP (ref. ⁵⁷), Th^2A-creER (ref. ⁵²), Mrgrpb4^cre (ref. ⁵⁸), PLP^eGFP (JAX, 033357)⁵⁹, Piezo2^smFP-Flag (ref. ³¹), R26^{FSF-LSL-ReaChR-mCitrine} (JAX, 024846), Ai80 (R26^{FSF-LSL-CatCh}; JAX, 025109), Ai14 (R26^{LSL‐tdTomato}; JAX, 007914)⁶⁰, Ai65 (R26^{FSF‐LSL‐tdTomato}; JAX, 021875)⁶¹, Ai96 (R26^LSL-GCaMP6s)⁶¹ and Ai148 (TIGRE^{LSL-GCaMP6f-tTA2})⁶². All lines were kept on a mixed background, while Advillin^cre and TrkB^flox mice were bred from a mixed background to the C57Bl/6 background for two generations for mating behaviour testing. For Advillin^cre;TrkB^flox/flox mice, non-neuronal recombination in the ear or tail was routinely detected, probably due to the leaky expression of Advillin^cre. However, the observed non-neuronal recombination of the TrkB^flox allele was not a result of germline deletion of TrkB, which is lethal⁶³.

Tamoxifen treatments

Tamoxifen was dissolved in 100% ethanol, diluted 1:2 in sunflower seed oil and vacuum-centrifuged for 30 min. The TrkB⁺ population was densely labelled for immunohistochemistry and physiology with an intraperitoneal injection of 0.5 mg of tamoxifen at P5 and sparsely labelled for AP and immunohistochemistry with 0.002–0.005 mg of tamoxifen. In some cases, tamoxifen (3 mg) was administrated to the pregnant mother at E14.5 or E15.5 through oral gavage, to label the same populations¹⁵. The Ret⁺ population was densely labelled with 3 mg of tamoxifen delivered to the pregnant mother through oral gavage at E11.5 or E12.5 and sparsely with 0.5 mg at E12.5.

Perfusion and post-fixation

Mice were anaesthetized with isoflurane and transcardially perfused with approximately 15 ml of 1× PBS with heparin (10 U ml⁻¹) and fixed with approximately 15 ml of 4% paraformaldehyde (PFA) in 1× PBS. Once perfused, the vertebral column and brain, if needed, were removed and post-fixed in 4% PFA in PBS at 4 °C overnight. In male mice, genital tissue, including the external prepuce, was removed at the base of the penis after an incision in the scrotal area. In female mice, the entire perineal area was removed from above the protrusion of hairy skin (homologous to the “external prepuce” in males) to below the vaginal opening. The prostatic urethra, from the bladder to the base of the penis, in male mice, and the vaginal canal in female mice were roughly dissected. Genital tissue was post-fixed in Zamboni solution (phosphate-buffered picric acid-formaldehyde) at 4 °C overnight, and tissues were washed in 1× PBS the next day.

Fixation of the erection state

The penis was collected after the mouse was freshly perfused with 1× PBS with heparin and the penis was isolated from the prepuces. One suture was fastened around the proximal base of the penis (proximal corpus cavernosum, not the corpus cavernosum glans) and the other was loosely wrapped around the base of the glans, just distal to the 90° bend. A 30 G needle attached to a 3 ml syringe containing Zamboni solution or 2% PFA was inserted between the sutures. Care was taken to ensure that the needle stayed inside the glans penis. Gentle pressure was applied to the syringe until the penis assumed a cup shape and, while maintaining pressure on the syringe, the distal suture was fastened. As the syringe was removed, the sutures were further tightened, and the tissue was post-fixed in Zamboni solution at 4 °C overnight. Subsequently, the tissue was transferred to 30% sucrose at 4 °C until it sank to the bottom, followed by embedding in OCT for cryosectioning.

Immunohistochemistry

Spinal cords and DRGs were isolated with forceps after removal of the dorsal and ventral vertebral column, and nodose ganglia were dissected after removal of the overlying submandibular glands and musculature. Penile tissue was isolated from the preputial glands and external and internal prepuces, and the proximal portion of the penis was discarded. Working proximal to distal, the clitoris was isolated from perineal tissue by removal of remaining pelvic musculature and preputial glands, then by fine dissection of the clitoris and urethra from above the vaginal canal and inside of the external prepuce. Glabrous digit-tips were cut from the paws for staining.

Tissue was cryoprotected in 30% sucrose in 1× PBS at 4 °C overnight, embedded in Neg-50 and frozen at −80 °C. Spinal cord and DRG samples were sectioned transversely using a cryostat (Leica) at 30 µm and placed onto slides. Genital tissue and glabrous digit-tips were sectioned sagittally at 30 µm thickness and placed onto slides or at 200 µm and placed into a well plate with 1× PBS. Thin cryosections were allowed to dry overnight at room temperature. The sections were rehydrated with 1× PBS, blocked with 5% normal donkey serum in 0.1% PBST (0.1% Triton X-100 in 1× PBS) for 2 h at room temperature, and incubated with primary antibodies in blocking solution for 2 days at 4 °C. The slides were rinsed in PBST three times for 10 min and incubated with secondary antibodies in blocking solution overnight at 4 °C. The slides were then rinsed in PBST 3 times for 10 min and mounted with Fluoromount-G. The sections were imaged on the Zeiss LSM 700 or 900 confocal microscope.

Immunohistochemistry of thick sections was performed using previously described protocols for whole-mount tissue⁵⁶. Sections were washed in 0.3% PBST five times for 1 h and incubated in primary antibody in blocking solution (5% normal donkey serum, 75% PBST, 20% DMSO) at room temperature for 3–5 days under gentle agitation. After five 1 h washes with 0.3% PBST, secondary antibodies in blocking solution were added for 2–4 days at room temperature. The sections were then washed twice in 0.3% PBST for 1 h, stained with DAPI (1 µg ml⁻¹, 5 min) to visualize nuclei, further washed three times in 0.3% PBST for 1 h, and dehydrated in serial methanol concentrations (50%, 75%, 100%, 1 h each) and then overnight in 100% methanol. Tissue was cleared in BABB (1:2, benzyl alcohol: benzyl benzoate) and imaged on the Zeiss LSM 700 or 900 confocal microscope or AxioZoom stereoscope while fully submerged. Dehydrated tissue was stored in methanol at 4 °C.

The primary antibodies used in this study were as follows: chicken anti-GFP (Aves Labs, GFP-1020, 1:500), goat anti-GFP (US Biological, G8965-01E, 1:500), goat anti-mCherry (CedarLane, AB0040-200, 1:500), rabbit anti-CGRP (Immunostar, 24112, 1:500), chicken anti-NF200 (Aves Labs, NFH, 1:200), rabbit anti-NF200 (Sigma-Aldrich, N4142-.2ML, 1:500), rabbit anti-S100 (ProteinTech, 15146-1-AP, 1:200–1:500), rat anti-TROMA-1 (DSHB, AB_531826, 1:200), goat anti-CD31 (R&D Systems, AF3628, 1:500), guinea pig anti-Flag (1:500)³¹ and IB4 (Alexa 647 conjugated) (Invitrogen, I32450, 1:500).

Quantification of corpuscle structure and distribution

To determine the total number of corpuscles per animal, care was taken in collecting all of the 200 µm sections from a given clitoris or penis sample and staining them with antibodies for S100 and NF200. All of the sections were imaged under a confocal microscope. The number of corpuscles in each section was manually counted using ImageJ, with the location of each corpuscle preserved in an ROI file. The borders of imaged tissue were manually outlined and saved. Each section’s volume was computed by multiplying the area of the outlined region by the height of the z stack, allowing for calculation of corpuscle density. To generate distribution heat maps, the ROI files containing the location of each corpuscle and the tissue outline were imported into MATLAB. The distribution of corpuscles was binned in a grid, and a Gaussian filter was used to smooth the density distribution.

Complex Krause corpuscles were defined as Krause corpuscles containing tightly coiled axons. Complex Krause corpuscles often exhibited globular shapes, while some also exhibited elongated shapes with convoluted axonal profiles. Simple Krause corpuscles were defined as the corpuscles containing 1 or 2 linear (non-coiled) axons. All simple corpuscles have an elongated shape, thereby exhibiting some similarity with Pacinian corpuscles but with a much smaller size.

Transmission electron microscopy

Adult mice were intracardially perfused with phosphate buffer and a glutaraldehyde/formaldehyde fixative. The thickest segment of the clitoris was isolated and the distal third of the glans penis was microdissected from the internal tissue and flattened on a piece of filter paper for overnight post-fixation at 4 °C. After the samples were washed in 0.1 M phosphate buffer (pH 7.4), they were cryoprotected overnight in 30% sucrose in 0.1 M phosphate buffer. After samples were embedded in Neg-50 Frozen Section Medium, they were frozen and cryosectioned into 100 µm sections and placed into the 0.1 M phosphate buffer solution. Medial sections of the clitoris and complete sections of the glans penis were selected for sample preparation.

The samples were osmicated in cacodylate buffer with 1% osmium tetroxide/1.5% potassium ferrocyanide for 1 h. The sections were washed with double-distilled H₂O and stained in a solution of 0.05 M sodium maleate (pH 5.15) and 1% uranyl acetate at 4 °C overnight. After washing with double-distilled H₂O, the sections were dehydrated with serial ethanol concentrations and propylene oxide. The sections were then infiltrated with 1:1 mix of epoxy resin (LX-112, Ladd Research) and propylene oxide at 4 °C overnight. The sections were embedded in an epoxy resin mix and cured at 60 °C for 48–72 h. Ultrathin (approximately 60 nm) sections were generated and imaged on the JEOL 1200EX transmission electron microscope at 80 kV accelerating voltage. The images were cropped using ImageJ.

Whole-mount AP staining of skin and spinal cord

TrkB⁺ and Ret⁺ afferents were sparsely labelled using the Brn3a^cKOAP placental AP (PLAP) reporter mouse⁵⁵, as described above, to enable visualization of single nerve terminals in genital and glabrous tissue. C-LTMRs were densely labelled using Th^2A-creER;Brn3a^cKOAP animals, with tamoxifen (2 mg intraperitoneal (i.p.)) administered at 3 weeks old. Although sympathetic fibres are TH⁺, they do not express BRN3A; thus, sympathetic fibres are not labelled in TH^2A-creER;Brn3a^cKOAP animals. The MRGPRB4⁺ afferents were labelled using intrathecal injections of AAV-FLEX-PLAP⁶⁴ (5 μl, titre 1.2 × 10¹³ genome copies per ml) into 4-week-old Mrgprb4^cre mice⁵⁸. All of the mice were perfused after 7 weeks of age.

Different dissection methods were used for different types of tissue: (1) for female genital tissue, after the clitoris dissection described above, a cut was made along the ventral midline of the urethra to allow spreading of the wings of the distal clitoris. (2) For male genital tissue, to permit comprehensive visualization of penile innervation, the superficial layer together with the corpus cavernosum of the glans penis was carefully removed from the internal mesenchyme including the os penis, and separated into two pieces. While not included in our primary analysis, the internal prepuce of the penis was retained for whole-mount AP staining. To isolate the prostatic urethra, also excluded from our primary analysis, the seminal vesicles and lobes of the prostate were removed, and a cut was made from the distal urethral opening to the opening of the bladder to permit flattening of the tissue. (3) For glabrous skin, the entire ventral surface of the paw was dissected from the underlying tissue for staining. (4) For hairy skin, the hairs on the back and around the genitalia were removed using Nair. (5) For spinal cord and dorsal column nucleus, the whole spinal cord was dissected with the dorsal column nucleus attached, and the overlying dura was removed.

The post-fixed and dissected tissue was incubated in PBS at 68 °C for 2 h to inactivate endogenous AP, then rinsed three times for 5 min in B3 buffer (0.1 M Tris pH 9.5, 0.1 M NaCl, 50 mM MgCl₂, 0.1% Tween-20) at room temperature. For the PLAP enzymatic reaction, tissue samples were incubated at room temperature overnight in BCIP and NBT in B3 buffer (3.4 µl of each per 1 ml of B3 buffer). Stained tissue was then pinned flat in a dish, fixed in 4% PFA in PBS for 1 h at room temperature and serially dehydrated in ethanol (50%, 75%, 100%, 1 h each, then 100% overnight) while covered. Tissue was then cleared and imaged in BABB using the Zeiss AxioZoom stereoscope.

Quantification and reconstruction of single-neuron morphology

Arborizations belonging to individual neurons were imaged for analysis. The terminal area was measured in ImageJ by drawing a tight polygon around the terminals of a given axon, and the number of corpuscles innervated by each fibre was counted manually. It is possible that the corpuscles of a given afferent occupy a larger volume in the z dimension than is measured in this manner. Simple corpuscles were deemed to be thin and linear terminals, while complex corpuscles were identified as having wider and bulbous terminals. Reconstruction and filling of representative fibres was performed using the ImageJ SNT plugin.

Genital skin injections

Young adult mice were anaesthetized with continuous inhalation of 2% isoflurane from a precision vapourizer for the duration of the procedure (5–10 min). Injections were done with a bevelled borosilicate or quartz glass pipette. The glass pipette was connected to an aspirator tube assembly (Sigma-Aldrich, A5177-5EA), which was connected to a syringe to control the air pressure.

For penis injections, pressure was applied near the genital region to fully externalize the glans penis from the external prepuce and internal prepuce. A pair of blunt forceps was used to stabilize the penis, while inserting the glass pipette into the distal end of penis. The penetration depth ranges from superficial to 3 mm deep to label the whole glans penis. Pressure was applied to the syringe to eject the liquid inside the glass pipette. After injection, the pipette was kept inside the tissue for 10 s to reduce the leaking. A total of 3–4 locations on the penis were injected. For successful injections, the fast green dye mixed in the liquid was visible inside the tissue for at least 10 min. Rapid dissipation of the fast green dye (within seconds) indicated a failed injection.

For clitoris injections, hairs near the genital protrusion (hairy skin) were removed by Nair treatment and subsequent cleaning using 70% ethanol. A pair of blunt forceps was used to stabilize the protrusion while inserting the glass pipette 1–2 mm deep into the middle part of protrusion. Care was taken to ensure that the needle did not impale the urethra. After injection, the pipette was kept inside the tissue for 30 s to minimize leak into the hairy skin region. An injection into the clitoris was considered to be successful if the fast green dye mixed in the liquid was visible below but not on the surface of hairy skin.

For injection of mid-line hairy skin, hairs near the genital protrusion in either male or female were removed by Nair treatment and subsequently cleaning using 70% ethanol. A pair of blunt forceps was used to stabilize the hairy skin while inserting the glass pipette into the superficial hairy skin. After a successful injection, the fast green dye was immediately visible within the hairy skin region during the injection.

A total volume of 2 μl was injected into either the male or female target region, for injection of either CTB or AAV. For CTB injections, 4–8 week old animals were used, and the injected animals were perfused in 3–4 days. For AAV injections, 3–4 week old animals were used, AAV2-retro-hSyn-FlpO (4.95 × 10¹³, Boston Children’s Hospital Viral Core) was used, and the injected animals were perfused after 3 weeks.

Spinal-cord terminal quantification

Owing to the characteristic decrease in the size of the dorsal column from the lumbar enlargement to more caudal regions of the spinal cord, the axial levels of the lumbosacral spinal cord sections were determined by the ratio of the depth of dorsal column (L_d) to that of the central canal (L_c). L_d was calculated as the length from the midpoint of dorsal surface of spinal cord to the ventral border of dorsal column, whereas L_c was calculated as the distance between the midpoint of dorsal surface of spinal cord and the central canal. Ratios of L_d/L_c from the images in the reference atlas of the mouse spinal cord (https://mousespinal.brain-map.org/imageseries/showref.html) were calculated and used as the reference to identify axial levels of the lumbosacral sections.

To average the fluorescence intensity of spinal-cord sections of the same axial level, images were rotated to align the dorsal surface of spinal cord horizontally, and the midlines of the spinal cords were aligned using MATLAB. A region of interest (750 μm²) on the dorsal and medial regions of spinal cord sections was selected from each image for averaging. The fluorescence intensity in the square region of interest was measured, downsampled to 100 × 100 px and then averaged. As no fibres were seen in the dorsal column, its fluorescence intensity was used as the baseline and subtracted from the average intensity.

In vivo MEA recordings of L6 DRG neurons

Adult male mice (>6 weeks) were anaesthetized with inhaled isoflurane (approximately 2.0%) through a nose cone for the duration of the experiment. Body temperature was maintained at 37 ± 0.5 °C using a custom-made surgical platform equipped with two heating pads mounted on acrylic. An adjustable gap (~1 cm wide and ~7 cm long) between the heating pads allowed for access to genital stimulation from below.

After induction of anaesthesia, pressure was applied to the abdomen and perineal skin to fully externalize the glans penis. To maintain the externalized state, a minimal amount of Vetbond tissue adhesive was applied to the retracted internal prepuce and the proximal base of glans penis to prevent retraction. Subsequently, the back of the mouse was shaved, and a midline incision was made over the lumbar and sacral vertebrae. A custom-made spinal clamp was applied to the L5 vertebra to secure the spinal column. The paravertebral muscles over the L6 spine were dissected, and the bone covering bilateral L6 DRG was removed with a bone drill. Surgifoam sponges and cotton were used to control bleeding. After cleaning the surface of the L6 DRG, the epineurium surrounding the DRG was removed with fine forceps, then saline was added on the top of the DRG.

The platform was then moved to the MEA recording setup. A 32-channel silicon probe (Cambridge NeuroTech, ASSY-37 H6b) was inserted into either the left or right L6 DRG. The signals were amplified and recorded using an Intan Technologies RHD2132 amplifier chip and RHD USB Interface Board. Data acquisition was controlled with open-source software (Intan Technologies Recording Controller v.2.07).

After insertion of the MEA probe into the DRG, brush stimuli using a cotton swab or paintbrush were applied to the glans penis and perineal hairy skin to search for penis-innervating neurons. Neurons that exhibited robust firing in response to mechanical stimulation of the penis but not the adjacent hairy skin were classified as penis-innervating neurons.

For the animals with opsin expression in sensory neurons, the receptive field on the penis was identified, then a fibre optic probe was oriented to deliver light pulses to the penis to optotag sensory neurons expressing opsin. The light pulse was generated by a fibre-coupled LED (M470F4, Thorlabs) with an LED driver (LEDD1B, Thorlabs), and light pulses had a duration of 1 ms and a frequency of 10–20 Hz. Once entrainment of spiking to light pulses was confirmed, indentation was delivered using a mechanical stimulator (300C-I, Aurora Scientific) with a custom-made indenter using a probe with a ~200 µm tip diameter. Owing to the physical constraint of the set-up, only the dorsal side of the penis was accessible for indentation. If the receptive field was on the dorsal side of the penis, the indenter tip was adjusted to the centre of receptive field, vertical to the surface. A curved piece of plastic was positioned on the ventral side of the penis to support the indentation. A series of step indentations ranging from 1 mN to 75 mN was applied, with each step lasting for 0.5 s. A minimum of 20 repeated trials was conducted. Next, a series of sine-wave vibrations ramping from 0 to 20 mN at varying frequencies (from 10 Hz to 120 Hz) were applied using the indenter. The vibration stimuli of different frequencies were randomized in order, and each frequency was repeated five times. The force and displacement of the indenter were commanded with custom MATLAB (v.2019a or v.2021a) scripts controlling a data-acquisition board (National Instruments, NI USB-6343).

JRCLUST was used to automatically sort action potentials into clusters, which were then manually refined and classified as single or multi units (https://github.com/JaneliaSciComp/JRCLUST).

To calculate the conduction velocity of optotagged neurons, the latency was determined by subtracting the time of each spike from the middle point of each light pulse. The latencies for the four optotagged TrkB⁺ Krause afferents were 4.4 ms, 4.8 ms, 7.8 ms and 13.6 ms, respectively. The length of axonal projections from L6 DRG to the genitalia in adult mice was estimated to be 5 cm. Thus, the conduction velocities of the four optotagged TrkB⁺ neurons are 11.4, 10.4, 6.4 and 3.7 m s⁻¹ respectively.

To determine the mechanical thresholds for vibration stimuli across different vibration frequencies, the time of the first spike in response to the vibration in each trial was identified. The corresponding recorded force at that timepoint was defined as the mechanical threshold for that trial. Thresholds of the trials at each vibration frequency were then averaged.

In vivo extracellular recording of L6 DRG neurons using glass electrode

The surgical procedures were similar to those in the previous section. The method for extracellular recording with a glass electrode was adapted from a recent study⁶⁵. After exposure of the L6 DRG in anaesthetized animals, a borosilicate glass electrode filled with saline (2–3 MΩ) was slowly inserted into the DRG using a manipulator (MP-225, Sutter Instrument) while searching for neurons responding to electrical or mechanical stimuli applied to the externalized glans penis. The data were amplified and digitized using the Multiclamp 700B and Digidata 1550B (Molecular Devices) system in the pipette potential mode with a 1,000× gain. Signals were collected at a 50 kHz sampling rate, under a 100 Hz high-pass filter and a 3 kHz Bessel filter.

For electrical stimuli, a custom-made bipolar electrode was placed on the top of the dorsal nerve of the penis, delivering 1 mA pulses of 0.2 ms duration, to activate most or all sensory fibres innervating the glans penis. When a DRG unit showed reliable responses to the electrical stimulation, the glass recording electrode was repositioned to maximize the size of extracellular spike. The mechanical receptive field of the unit was then explored using a fine paint brush and a custom-made mechanical stimulator that was previously described⁶⁵. Subsequently, a series of step indentations and, in some experiments, vibrations were applied to the centre of receptive field to assess the mechanical threshold, adaptation properties or vibration tuning properties. The stimulation was programmed and the recordings were triggered using MATLAB.

The latency to spike was determined by the interval between the start of the electrical stimulation artifact and the onset of the spike signal. The threshold of each unit was defined as the minimum force needed to evoke a response from the neuron. Units that responded to electrical stimuli of the glans penis but not to step indentation stimuli were classified as non-responsive.

In vivo calcium imaging of L6 DRG neurons

The surgery procedure for male mice was the same as for the MEA recordings above. After the L6 DRG was exposed, the platform was moved to an upright epifluorescence microscope (Zeiss Axio Examiner) with a ×10 air objective (Zeiss Epiplan, NA = 0.20) for imaging. The light source was a 470 nm LED (M470L5, Thorlabs) with an LED driver (LEDD1B, Thorlabs), and a CMOS camera (CS505MU1, Thorlabs) was triggered at 10 frames per second with 50 ms exposure time using ThorCam software (v.3.7.0). All stimuli were synchronized with the camera using a data-acquisition board (National Instrument, NI USB-6343).

Indentation on the penis was performed in a similar method as described above for the MEA recordings, with a few modifications. The indentation step was increased to 3 s to allow for the assessment of adaption properties. The interval between each step was increased to at least 8 s to accommodate for the slow-decay dynamics of the GCaMP signals. The duration of ramping sinewave vibration was also increased to identify the mechanical thresholds using different vibration frequencies.

For thermal stimulation of the penis, a water reservoir device, inspired by a previously described method⁶⁶, was used. This device consisted of a reservoir connected to water baths at different temperatures, including room temperature, ice water (~4 °C) and hot water (~55 °C). The penis was submerged in the water reservoir. Then, water of different temperatures was pumped into the reservoir at a controlled speed, and room-temperature water was used during the baseline measurements. The water overflowing from the reservoir was directed to a collection chamber located beneath the reservoir. The temperature in the reservoir was monitored using a thermocouple microprobe (IT-1E, Physitemp) and a thermometer (BAT-12, Physitemp).

For clitoris stimulation, the indenter tip was placed vertically on the surface of protrusion where the clitoris was located. To minimize the dorsal–ventral movement of mouse’s lower body during indentation, the mouse tail was held down using a clamp. Similar step indentation and vibration stimuli were used for clitoris as described for the penis. After the mechanical stimulation, the surgery chamber was removed from the microscope, and the anaesthetized mouse was placed in a supine position. An incision was created on the protrusion to expose the dorsal nerve of clitoris, and saline was added on the top of the nerve. Subsequently, the mouse was reoriented to the prone position, with the spinal clamp reattached. After locating the same field of view of the DRG, a pair of custom-made bipolar electrodes (silver wire, with ~1 mm distance between the two points) was placed on the dorsal nerve of clitoris, and a train of 1 ms pulses (10 Hz, less than 1 mA) was then delivered.

For the calcium imaging analysis, motion correction and spatial high-pass filtering was conducted using a custom-written ImageJ macro code that used the ImageJ plugin moco and the Unsharp mask filter, and then regions of interest were manually selected. Cells exhibiting baseline signal and/or calcium responses were identified and aligned across videos encompassing various stimuli. The intensity measurements generated by ImageJ were then processed for further analysis using MATLAB. For the calculation of ΔF/F, F was determined using the baseline activity (average fluorescence intensity before each stimulation). The mechanical threshold for each step indentation session was determined on the basis of the first discernible calcium spikes aligned with the step indentation. To determine the mechanical thresholds for vibrations at varying frequencies, the s.d. of the baseline activity was calculated and the threshold for the response was defined as 5 × s.d. above the baseline. Next, the recorded force at that timepoint was defined and averaged across trials with the same frequency. To differentiate between clitoris-innervating neurons and neurons innervating hairy skin, only neurons that responded to electrical stimulation (5 × s.d. above the baseline) were included for further analysis.

Sticky-tape assay

The sticky-tape assay as a measure of hairy skin sensitivity was performed using a previously described protocol⁶⁷. A piece of common laboratory tape (2 cm by 2 cm, Fisher Scientific, 15-901-5R) was secured to the back of the animal. Then, the animal was introduced into a 15 cm by 15 cm Plexiglass enclosure and video-monitored for a period of 5 min. The animal’s attempts to remove the tape, characterized by scratching, biting and body shaking, were quantified using BORIS software.

Tactile PPI analysis

The tactile prepulse inhibition (PPI) experiments were conducted according to a previously described procedure⁶⁸. The mouse was positioned inside a plastic cylinder with 3.8 cm diameter. The container was then secured in a soundproof chamber (SR-LAB Startle Response System). Before an acoustic startle stimulus (125 dB, 20 ms), a 0.9 psi air puff (50 ms) as a prepulse was delivered to the back of the animal at various interstimulus intervals (50, 100, 250, 500 and 1,000 ms). The startle response of the animal was recorded using an accelerometer attached to the chamber. PPI was calculated as %PPI = [1 − (startle response from pulse following prepulse/startle response from pulse alone)] × 100.

Power spectrum analysis of genital contact

Force was measured using a load cell (Model MBL, 50 g, Honeywell) connected to a strain gauge amplifier (DMD-465WB, Omega). Vibratory stimuli were administered directly onto the load cell. For the force measurements during genital contact, the glans penis, dissected from a perfused mouse, was moved across a piece of freshly dissected vaginal tissue affixed to the load cell at a speed of around 1–2 cm s⁻¹, simulating genital contact during mating. The data were digitized at a sampling rate of 50 kHz using the Digidata 1550B (Molecular Devices) system, after noise reduction using the Hum Bug Noise Eliminator (Digitimer). Power spectrum analyses were conducted with Clampfit software (v.11.2), with fixed-duration periods (0.3 s) sampled from either the baseline or during stimulation phases.

Spinal transection

Mice were anaesthetized with inhaled isoflurane (approximately 2.0%) through a nose cone throughout the surgery. The bladder was emptied by applying gentle abdominal pressure. Body temperature was maintained at 37 ± 0.5 °C using a heating pad, and ophthalmic ointment was placed on the eyes. Once reflexes were absent, the back was shaved and sterilized with alternating swabs of ethanol and betadine. A midline incision was made over the thoracic spinal column, and the paravertebral musculature was cut to expose the gap between the T9 and T10 spinal segments. A curved scalpel blade (10012-00, FST) was inserted between the T9 and T10 spinal segments and moved laterally to ensure thorough transection. Bleeding was controlled with Surgifoam sponges and cotton. The skin of the back was sutured and mice were allowed to recover on a heating pad. Hindlimb paralysis was assessed to confirm success of the spinal transection. Food and DietGel were placed onto the floor of the cage, and carprofen (5 mg per kg) was injected subcutaneously every 24 h. Bladder expression was performed every 2–3 h on the day of surgery and every 6–8 h on the next day. The mice were euthanized within 3 days of spinal transection.

Male sexual reflex behaviours

Reflex behaviours were tested at 6 or 24 h after spinal transection. The awake animal was restrained in a round acrylic chamber in a supine position, with the paralysed hindlimbs secured on the platform using Scotch tape. Two cameras were positioned from different angles, focusing on the genital region of the mouse. To begin the procedure, pressure was applied to the side of the genital region to fully externalize the glans penis of the mouse. Reflexes might have occurred during or right after the externalization due to the force applied in this process, but this effect would end within 1 min of externalization.

For evoking vibration-induced sexual reflexes in male mice, 2 min after externalization of the glans penis, vibration stimuli (50 Hz) were delivered using a custom-made mechanical stimulator as previously described⁶⁵. The stimulator consisted of a DC motor connected to an arm (10 mm long) with a round tip (2 mm in diameter) mounted on the end. The motor was driven by a custom-built current supply controlled by a data-acquisition board (NI USB-6343, National Instrument). The force was calibrated using a custom-made force sensor. The stimulator was attached to an articulating arm that could be moved manually. Before applying vibration stimuli, the stimulator tip was positioned close to the lateral side of glans penis, while a holder attached to an articulating arm was moved to the opposite side to ensure the glans penis remains in place. The vibration stimuli were delivered as a 50 Hz sine wave, with a 1 s duration followed by a 1 s interstimulus interval. Each session consisted of 10 such epochs, lasting a total of 20 s. The stimuli were programmed and delivered using MATLAB. Vibration sessions were spaced 2–3 min apart, and three vibration sessions were conducted for each animal.

For brush-induced sexual reflexes, 2 min after externalization of the glans penis, a cotton swab was used to apply brush stimuli in either the proximal or distal direction. Each session lasted 20 s, with intervals of longer than 2 min between sessions.

Optogenetic induction of sexual reflexes was performed 2 min after externalization of the glans penis of spinalized animals. The optic fibre (diameter, 1,000 µm; 0.39 NA; M35L02, Thorlabs) connected to a fibre-coupled LED (M470F4, Thorlabs) was directed towards the glans penis. The stimulation sessions lasted 20 s, with 10 Hz and 2 ms pulses used for animals expressing ReaChR, and 20 Hz and 1 ms pulses used for animals expressing CatCh. The maximum light intensity was 10 mW. Mice that showed excessive spontaneous activity before optogenetic stimulation, probably due to a blockage in the bladder, were excluded.

For mechanically induced sexual reflex behaviours, 2–4-month-old mice were used. For the optogenetic stimulation of TrkB^creER mice, 4–6-month-old animals were used because externalization of the glans penis was not feasible in younger animals that received tamoxifen treatments at P5. Red light was used to illuminate the set-up during optogenetic induced behaviours, while room light was used for mechanical stimulation measurements. Videos were taken at 60 fps using SpinView (v.1.1.0.43). To quantify the sexual reflexes, the start and end of each reflex behaviour was manually labelled using BORIS software⁶⁹. The behavioural measurements were then exported, analysed and plotted in MATLAB.

Female sexual reflex behaviours

Two days before testing, female mice were treated with 17β-estradiol benzoate (35 μg delivered i.p.) to facilitate access to the vaginal opening. The day before testing, mice were spinal transected at the T9/T10 level, using methods described above. On the day of the test, the awake, paralysed animals were gently restrained in a supine position with all four limbs taped on a platform, to reduce motion artifacts during the pressure measurements. A latex balloon (Harvard Apparatus, 73-3478) affixed to a cannula (PE-50 tubing) was carefully inserted into the vagina, and secured with the cannula taped on the platform. The assembly was connected to a pressure transducer (BIOPAC system, TSD104A) through saline-filled tubing. The transducer’s output was subsequently amplified and digitized at a 2000 Hz sampling rate using a BIOPAC MP150 system and AcqKnowledge software.

After the animal and the balloon were secured, the custom-made mechanical stimulator described above⁶⁵ was positioned at the top of the clitoris within the genital protrusion. When a stable baseline was established, vibration stimuli were delivered as a 50 Hz sine wave, using a 1 s duration followed by a 0.5 s interstimulus interval. Each session consisted of 10 such epochs, lasting a total of 15 s. The stimuli were programmed and delivered using MATLAB. Vibration sessions were spaced more than 2 min apart, and more than three vibration sessions were conducted for each animal. Each mouse was then anaesthetized with 1.5% isoflurane, anaesthesia was verified by toe pinch and additional vibration sessions were conducted. The start of the vibration session was manually annotated in the AcqKnowledge software. As whole-body movements of the mouse were also detected during vaginal pressure measurements, each mouse was monitored throughout the experiment and all movement artifacts were marked accordingly in the software.

To measure the vaginal pressure during optogenetic activation, the hair on top of the genital protrusion was removed using Nair. The optic fibre (diameter, 1,000 µm; 0.39 NA; M35L02, Thorlabs) connected to a fibre-coupled LED (M470F4, Thorlabs) was directed towards the clitoris within the genital protrusion. The stimulation sessions lasted 20 s, with 1 ms pulses at a frequency of 20 Hz and a peak light intensity of 10 mW, while measuring vaginal pressure.

The pressure measurements were processed in MATLAB after conversion using acq-tools (https://github.com/bccummings/acq-tools). Owing to the variability in absolute pressure values across animals, as a result of the variations in the baseline pressure and positioning of the balloon, the fold increase relative to baseline noise level was used to standardize the response across animals. The baseline noise level was calculated as the s.d. of the pressure measurement during the 10 s period preceding each stimulation session. Motion artifacts caused by whole-body movements, annotated during experiments and identified by clustered sharp peaks, were distinct from the gradual changes indicative of genuine vaginal pressure. Motion artifacts were pronounced in C57BL/6 mice but were less commonly encountered with CD1 mice; we therefore used 2–4-month-old CD1 or mixed-background animals for all female sexual reflex behavioural experiments.

Mating behaviours

At least 2 weeks before testing, mice were transferred to a reverse 12 h–12 h light–dark cycle room. All of the mice used for mating behaviours were 2–4 months old. They were from a mixed background but backcrossed to C57Bl/6 mice for two generations.

The mating behavioural analysis was similar to that of a previous study⁷⁰. The male mouse was singly housed for at least 2 weeks before testing and was socialized with a female mouse overnight at least 1 week before the test. Then, 2 days before the test, female mice were treated with 18–35 μg of 17β-estradiol benzoate (dissolved in sterile sesame oil, final volume of 50–100 μl, subcutaneous injection) and treated again with 50–100 μg progesterone (dissolved in sterile sesame oil, final volume of 50–100 μl) 5 h before the test. During the mating test, a female mouse was placed into the home cage of the male mouse, and the cage cover was replaced by a custom-made wall (6 inches tall) to prevent mice from leaving their cage. Mating behaviours were visualized in the male’s home cage using USB cameras, which were placed above the cage capturing videos at 30 fps using IC capture (v.2.5). The behaviour room was dark, illuminated only with infrared lamps.

When testing the mating performance of the males, a wild-type female mouse that was unreceptive 10 min after the start of the session would be replaced with another female, so that only receptive females were used for testing males. Mating behaviour was monitored over video for 75 min. The male mice underwent testing for at least three rounds, with each round occurring at least 1 week apart.

When testing the mating performance of female mice, only experienced males were used. The same group of female controls and mutants were hormone-primed (as described above) in the first two rounds of the mating test, while, in the third round, they were in a natural oestrus cycle. The first two rounds were conducted 2 weeks apart, and the third trial was conducted 4 weeks after the second trial to allow for the recovery of oestrus cycle. The oestrus cycle was monitored using the vaginal cytology method according to previously established protocols^71,72. Vaginal smears were collected, stained with crystal violet and examined to determine the oestrus stage. Only female mice in proestrus or oestrus were used for mating behaviours within 3 h of vaginal lavage. As hormone priming prevents the treated female mice from becoming pregnant, we allowed the female mice to engage in the full mating test (75 min) in the first two rounds to provide sexual experience to the cohort. In the third test for the same groups of animals but under their natural oestrus cycle, females were separated from the males 10 min after the start of intromission to avoid pregnancy. If no intromission occurred, the mating session was terminated at 30 min.

The videos of mating behaviours were manually scored using BORIS software⁶⁹ by an experimenter who was blinded to the genotype of the animals. For male mating behavioural analysis, the following behaviours were scored: sniffing, mounting, intromission and ejaculation. Mounting refers to the males’ rapid pursuit of females, followed by grasping their rear, and often followed by short probing without gaining access to the female genitalia. Intromission represents the long rhythmic thrusting, indicating the male’s successful penetration of the vagina. The mounting and intromission periods do not overlap with the scoring methods used; mounting is considered to be the pursuit and adjustment period before the intromission period. The intromission bouts lasting less than 2 s were excluded from the analysis due to the lack of rhythmicity of movements and the possibility that penetration did not fully occur. Moreover, combative behaviours and darting behaviours of female mice were scored during the female mating behaviour assessment trials. Combative behaviour refers to the instances in which the female confronts or fights with the male, while the dart behaviour occurs when the female mouse attempts to quickly escape from the mounting of the male mice.

Statistics and reproducibility

Data were analysed using MATLAB (v.2019a or v.2021a), Python (v.3.7.7) and GraphPad Prism 10. The number of samples and the statistical tests used for individual experiments are included in the figure legends when length permitted. In Fig. 2c, n = 10 sections from 3 control females, 10 sections from 3 TrkB^cKO females, 7 sections from 2 control males and 20 sections from four TrkB^cKO males. In Fig. 2f, n = 32 TrkB⁺ axons from 5 males, 16 Ret⁺ axons from 3 males, 26 TrkB⁺ axons from 4 females, 11 Ret⁺ axons from 3 females, 26 TrkB⁺ axons from 4 animals’ digits, and 17 Ret⁺ axons from 4 animals’ digits.

Statistical analysis was performed using unpaired t-tests, one-way ANOVA with post hoc multiple unpaired t-tests, two-way ANOVA with Tukey’s multiple-comparisons test and Fisher’s exact tests. All P values for unpaired t-test used are two tailed. The exact P values and other statistics for unpaired t-tests, multiple unpaired t-tests or Tukey’s multiple-comparison test used in the study are as follows: in Fig. 1g, P = 0.9355, t = 0.085, d.f. = 5, F = 1.002; Fig. 1i, between clitoris and glans penis, P < 0.000001, t = 13.66, d.f. = 917, F = 2.514; between clitoris and digit (Meissner), P < 0.000001, t = 6.82, d.f. = 682, F = 8.169; Fig. 2f, for datapoints of penis, P = 0.001138, t = 3.482, d.f. = 44; for datapoints of digit, P = 0.000224, t = 4.068, d.f. = 39; for datapoints of clitoris, P = 0.000104, t = 4.409, d.f. = 33; Fig. 2g, for datapoints of penis, P = 0.002853, r = 3.156, d.f. = 45; for datapoints of digit, P = 0.8476, t = 0.1934, d.f. = 41; for datapoints of clitoris, P = 0.00666, t = 2.885, d.f. = 35; Fig. 4i, between clitoris and penis in the TrkB⁺ neurons, P = 0.000001; between clitoris and penis in the Ret⁺ neurons, P = 0.2023; between TrkB⁺ and Ret⁺ neurons of clitoris, P < 0.000001; between TrkB⁺ and Ret⁺ neurons of penis, P = 0.000632; Fig. 5g, for session 1, P = 0.6457, t = 0.08194, d.f. = 15; for session 2, P = 0.6358, t = 0.4834, d.f. = 15; for session 3, P = 0.6994, t = 0.8235, d.f. = 15; Fig. 5h, for session 1, P = 0.8526, t = 0.1891, d.f. = 15; for session 2, P = 0.3264, t = 1.015, d.f. = 15; for session 3, P = 0.5408, t = 0.6259, d.f. = 15; Fig. 5i, for session 1, P = 0.008921, t = 3.117, d.f. = 11.99; for session 2, P = 0.003970, t = 3.453, d.f. = 13.76; for session 3, P = 0.004659, t = 3.344, d.f. = 14.412; Fig. 5k, P = 0.0126, t = 2.858, d.f. = 14, F = 17.18; Fig. 5l, P = 0.0026, t = 3.558, d.f. = 16, F = 2.473; Fig. 5m, P = 0.0009, t = 4.064, d.f. = 16, F = 1.133; Extended Data Fig. 1e, P < 0.000001, t = 7.991, d.f. = 20, F = 2.293; Extended Data Fig. 8c, for erections, P = 0.2420, t = 1.231, d.f. = 12, F = 2.977; for cups, P = 0.8338, t = 0.2148, d.f. = 11, F = 1.701; Extended Data Fig. 9a, for different interstimulus intervals in the order from small to large of females, P = 0.5498, 0.9012, 0.1837, 0.3384, 0.5429, for the same order of males, P = 0.7127, 0.5249, 0.9468, 0.9228, 0.6364; Extended Data Fig. 9b, for females, P = 0.7969, t = 0.2619, d.f. = 15, F = 1.409; for males, P = 0.6961, t = 0.3970, d.f. = 18, F = 3.950; Extended Data Fig. 9d, for number of mounts, P = 0.4617, t = 0.7416, d.f. = 49, F = 1.445; for mount total duration, P = 0.4876, t = 0.6994, d.f. = 49, F = 1.865; for mounting start time, P = 0.0064, t = 2.850, d.f. = 49, F = 6.615; Extended Data Fig. 9e, for post-ejaculation freeze, P = 0.3365, t = 0.9722, d.f. = 42, F = 1.436; for ejaculation start time, P = 0.2897, t = 1.072, d.f. = 42, F = 1.118; Extended Data Fig. 9f, for intromission number, P = 0.8697, t = 0.1649, d.f. = 45, F = 1.531; for intromission start time, P = 0.0287, t = 2.250, d.f. = 53, F = 1.717; for intromission total duration, P = 0.0252, t = 2.318, d.f. = 45, F = 3.553; for intromission duration per bout, P = 0.000003, t = 5.324, d.f. = 45, F = 2.618; for inter-intromission interval, P = 0.0222, t = 2.370, d.f. = 45, F = 9.020; Extended Data Fig. 9g, from trial 1 to 3, P = 0.2473, 0.4192, 0.1055, t = 1.216, 0.8324, 1.723, d.f. = 12, 14, 15; Extended Data Fig. 9h, from trial 1 to 3, P = 0.8998, 0.7724, 0.01131, t = 0.1287, 0.2949, 2.886, d.f. = 12, 14, 15; Extended Data Fig. 9i, from trial 1 to 3, P = 0.9876, 0.2966, 0.9297, t = 0.01601, 1.106, 0.09282, d.f. = 9.583, 9.274, 4.890; between trial 1 and 3 of Ctrl, P = 0.0181, t = 2.492; Extended Data Fig. 10b, from trial 1 to 2, P = 0.3629, 0.6909, t = 0.9335, 0.4041, d.f. = 18, 18; Extended Data Fig. 10c, from trial 1 to 2, P = 0.1489, 0.9108, t = 1.508, 0.1136, d.f. = 18, 18; Extended Data Fig. 10d, from trial 1 to 2, P = 0.3465, 0.9190, t = 0.9682, 0.1031, d.f. = 17, 18; Extended Data Fig. 10e, from trial 1 to 2, P = 0.2149, 0.9918, t = 1.286, 0.01039, d.f. = 18, 18; Extended Data Fig. 10h, P = 0.1140, t = 1.672, d.f. = 16, F = 1.343; Extended Data Fig. 10i, P = 0.0033, t = 3.455, d.f. = 16, F = 3.218; Extended Data Fig. 10j, P = 0.9348, t = 0.08315, d.f. = 16, F = 4.197; Extended Data Fig. 10k, P = 0.0126, t = 2.858, d.f. = 14, F = 17.18; Extended Data Fig. 10l, P = 0.4063, t = 0.8563, d.f. = 14, F = 1.829;

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.