Touch neurons underlying dopaminergic pleasurable touch and sexual receptivity

Summary

Pleasurable touch is paramount during social behavior, including sexual encounters. However, the identity and precise role of sensory neurons that transduce sexual touch remains unknown. A population of sensory neurons labeled by developmental expression of the G-protein coupled receptor Mrgprb4 detect mechanical stimulation in mice. Here, we study the social relevance of Mrgprb4-lineage neurons and reveal that these neurons are required for sexual receptivity and sufficient to induce dopamine release in the brain. Even in social isolation, optogenetic stimulation of Mrgprb4-lineage neurons through the back skin is sufficient to induce a conditioned place preference and a striking dorsiflexion resembling the lordotic copulatory posture. In the absence of Mrgprb4-lineage neurons, female mice no longer find male mounts rewarding: sexual receptivity is supplanted by aggression and a coincident decline in dopamine release in the nucleus accumbens. Together, these findings establish that Mrgprb4-lineage neurons initiate a skin-to-brain circuit encoding the rewarding quality of social touch.

Graphical Abstract

In Brief

Mrgprb4-lineage touch neurons in the skin are required for sexual receptivity and dopamine release that makes this type of social touch rewarding.

Introduction

The pleasure of a partner’s caress or a child’s embrace begins with mechanical signals transduced by neurons in our skin. Despite the centrality of socially rewarding touch in our daily lives, the neurons in the skin that detect social touch and shape the valence of perception generated in the brain, remain unknown. This gap in knowledge is critical, especially when considering the nature of neurodevelopmental disorders like autism spectrum disorder, where gentle touch and socially rewarding behaviors are aversive^{1, 2, 3}.

How might touch generate reward during social interactions? The mesolimbic pathway consists of ventral tegmental area (VTA) neurons that release dopamine into the nucleus accumbens (NAc) to promote reward-learning and reinforcement^4–8. VTA dopamine signaling drives social behaviors such as exploring a non-familiar conspecific, same-sex social interactions, or play behavior between female rats^9–11. Although a role for VTA dopamine neurons in promoting social behaviors has been identified, except for a handful of studies¹², the sensory neurons and ascending pathways that encode social stimuli – such as tactile input – and activate VTA dopamine neurons, remain obscure.

A class of sensory neurons in humans linked to gentle stroking are termed C-tactile afferents^{13, 14}, and there appear to be analogous populations of these neurons in the mouse^15–17. One such class of neurons in mice express the G-protein coupled receptor Mrgprb4 and share anatomical and physiological similarities with human C-tactile afferents^{18, 19}. Prior studies demonstrate that hairy skin-innervating C-mechanoreceptors labeled in the Mrgprb4^Cre mouse line respond to gentle stroking and produce a conditioned place preference, suggesting that their activation is rewarding^{18, 19}. These findings prompted us to ask whether sensory neurons marked by the Mrgprb4^Cre mouse line might be important for promoting ethologically relevant rewarding touch that engages the mesolimbic reward pathway. Here we used a combination of mouse genetics, slice electrophysiology, circuit tracing, behavioral paradigms, and in vivo brain imaging to elucidate the skin-to-brain circuits responsible for mediating the rewarding aspect of social touch.

Results

Focalized activation of Mrgprb4-lineage neurons in the skin is rewarding and induces lordosis-like posture

We reasoned that focal activation of Mrgprb4-lineage neurons in the back skin might mimic touch among mice, so we used the blue light sensitive ion channel channelrhodopsin (ChR2) to focally stimulate Mrgprb4+ neurons in vivo. To accomplish this goal, we used an Mrgprb4^Cre driver that functions as a lineage tracer to express a ChR2-eYFP fusion protein in Mrgprb4-lineage neurons (Fig 1a), and crossed these Mrgprb4^Cre mice to Rosa^ChR2/ChR2 mice for two generations to generate Mrgprb4^Cre; Rosa^ChR2/ChR2 mice. We confirmed this genetic targeting strategy with RNAscope in situ hybridization and immunofluorescence (Fig 1b,c; Fig 2b). Because Mrgprb4 expression begins at ~P4–5 in a population of progenitors broader than the Mrgprb4-expressing population of adult neurons²⁰, we used RNAscope double in situ hybridization to characterize the expression of ChR2-eYFP in Mrgpra3+, Mrgprc11+, and Mrgprd+ populations of dorsal root ganglion (DRG) neurons, which are known to share lineage with Mrgprb4+ neurons (Fig 1b,c). We found the expression of ChR2 across these populations consistent with the expected developmental expression of Mrgprb4^{20, 21} (Fig 1b,c), thus accounting for all ChR2-eYFP+ cells in Mrgprb4^Cre;Rosa^ChR2/ChR2 mice. Based on recent single cell RNAseq data in the mouse DRG²², we performed additional sets of RNAscope double in situ hybridization experiments to further characterize the Mrgprb4-lineage neurons by probing for genes across transcriptomically defined clusters including Pou4f2, Trpa1, Ret, TH, Calca, Gfra1, Trpv1, Scn10a, and Calca (Fig 1b,c). Notably, there was minimal overlap between eYFP and TH, a marker of other C-LTMRs^22–24, confirming that Mrgprb4-lineage neurons are a distinct population of C-fibers. Moreover, consistent with prior neuroanatomy targeting Mrgprb4+ neurons¹⁹, we observed sensory afferent endings in the hairy skin associated with, and in close apposition to, hair follicles (Fig. 1d). Finally, to confirm and extend prior studies^{18, 19, 25}, we probed the mechanical sensitivity of Mrgprb4-lineage neurons using whole-cell patch clamp recordings from acutely dissociated DRG neurons. We thereby confirmed that Mrgprb4-lineage neurons are mechanosensitive, exhibiting mechanically evoked currents that increased with membrane indentation (Fig 1e,f).

Figure 1: Genetic targeting strategy to optogenetically manipulate Mrgprb4-lineage touch neurons.

A) Mrgprb4^Cre mice express ChR2-eYFP in a Cre-dependent manner. B) RNAscope in situ hybridization in DRG to quantify the expression of ChR2 in Mrgprb4+, Mrgprd+, Mrgprc11+, Mrgpra3+, Pou4f2+, Trpa1+, Ret+, TH+, Scn10a+, Calca+, Gfra1+, and Trpv1+ cells. Scale bars represent 100 μm. C) Percent expression of eYFP in different populations of DRG neurons from B. D) Expression of ChR2-eYFP in Mrgprb4-lineage nerve terminal endings in the hairy skin of the mouse back, labeled via immunofluorescence. E-F) Whole cell patch clamp recordings from individual eYFP+ Mrgprb4-lineage neurons demonstrating mechanical sensitivity. Increase in current response with increased mechanical stimulation.

Figure 2: Mrgprb4-lineage neurons activate downstream spinal circuits including Gpr83+ spinoparabrachial projection neurons.

A) Mrgprb4^Cre mice express ChR2-eYFP in a Cre-dependent manner. B) Immunofluorescence staining in spinal cord dorsal horn. Mrgprb4+ terminals innervate lamina II, inferior to CGRP+ terminals in lamina I and overlapping with IB4+ terminals in lamina II. Scale bars represent 100 μm. C) Schematic illustrating spinal cord slice electrophysiological recordings from lamina II during optogenetic stimulation of terminals. D) Mono- or poly-synaptic light induced currents only in Mrgprb4^Cre; Rosa^ChR2/ChR2 mice. E) Quantification of light induced currents from Mrgprb4^Cre; Rosa^ChR2/ChR2 mice with jitter used to determine mono- or poly-synaptic transmission. F) Light-induced currents in Mrgprb4^Cre; Rosa^ChR2/ChR2 mice are abolished with CNQX. G) Light-induced currents are abolished in Mrgprb4^Cre; Rosa^ChR2/ChR2 mice with TTX. H,I) Quantification of data presented in F,G. J) Schematic, brightfield, and fluorescent images illustrating recording from GFP+ GPR83+ neurons during optogenetic stimulation of Mrgprb4-lineage neurons. K-M) Characterization of light-induced inputs to Gpr83+ putative spinoparabrachial projection neurons. K) Left: Whole-cell voltage clamp recording at −70 mV demonstrating excitatory inputs, Middle: Incidence of a monosynaptic only, polysynaptic only, or mono + polysynaptic excitatory input, Right: Whole-cell voltage clamp recording at −20 mV demonstrating inhibitory inputs. L) Cell-attached recording showing light-evoked action potential discharge under normal conditions (left) and following addition of bicuculline and strychnine (right). M) Quantification of data presented in L.

With confirmation of our genetic targeting strategy, we next asked if we could induce light-evoked, synaptically driven currents to second-order neurons downstream in the dorsal horn of the spinal cord (Fig 2a–i). Recording from LII of the spinal cord of Mrgprb4^Cre;Rosa^ChR2/+ heterozygous mice yielded no light-induced currents, while all recordings from Mrgprb4^Cre;Rosa^ChR2/ChR2 homozygotes exhibited mono- and/or poly-synaptic inputs (Fig 2d,e). Both mono and polysynaptic inputs could be abolished by application of CNQX, demonstrating glutamatergic synaptic transmission between Mrgprb4-lineage neurons and second-order neurons (Fig 2f–i). Mrgprb4-lineage neuron photostimulation evoked action potential (AP) discharge in ~38% of postsynaptic neurons. AP discharge was particularly prevalent in tonic firing and initial bursting neurons, previously associated with inhibitory interneurons (see Figure S1 for further characterization of spinal neurons downstream of Mrgprb4-lineage neurons).

Recent studies demonstrated that GPR83+ spinoparabrachial (SPB) neurons constitute an affective touch circuit and that these spinal cord projection neurons receive direct input from Mrgprb4-lineage neurons²⁶. To confirm and extend these findings, we crossed the Gpr83-GFP line with Mrgprb4^Cre;Rosa^ChR2/ChR2 mice and optogenetically activated the terminals of Mrgprb4-lineage neurons while recording from Gpr83-GFP putative SPB neurons (Fig 2j). We targeted the GFP+ cells within Lamina I, as these were most likely projection neurons, and we avoided recording from the clearly distinguishable Lamina II GFP+ neurons, which were likely interneurons²⁶. When holding the membrane potential at −70mV, we noticed that most GFP+ SPB neurons received polysynaptic EPSCs (Fig 2k). When we held the membrane potential at −20mV to record inhibitory currents, the majority of cells also received polysynaptic inhibitory input (Fig 2k, right). Using cell-attached recordings of lamina I GFP+ SPB neurons, we observed that 1ms photo-stimulation of Mrgprb4-lineage neurons evoked action potentials in 43% of cells (Fig 2l). Blocking inhibitory transmission with GABA and glycine receptor antagonists, bicuculline and strychnine, we observed enhanced responses in the GFP+ SPB neurons to Mrgprb4-lineage neuron photo-stimulation.

We next sought to determine the valence of focal activation of Mrgprb4-lineage neurons in the back skin of adult females. We developed a conditioned place preference assay with laser light administered transdermally through the shaved back skin: one chamber had ChR2-activating blue light, and the other had non-stimulating green light to control for any effect of a visible light beam (Fig 3b). Interestingly, we observed that Mrgprb4^Cre; Rosa^ChR2/ChR2 females had a significant preference for the chamber associated with blue light stimulation after training compared to Cre-negative littermates, suggesting that focal activation of Mrgprb4-lineage neurons in the back skin is inherently rewarding (Fig 3c,d). Because approximately 45% of Mrgprd+ neurons express ChR2 in the Mrgprb4-lineage line (Fig.1c), we sought to confirm the specificity of this result to Mrgprb4-lineage neurons by repeating these experiments with Mrgprd^Cre-ERT2; Rosa^ChR2/ChR2 mice. The Mrgprd^Cre-ERT2 mouse line has an inducible Cre, allowing us to treat these mice with tamoxifen after P10, thereby labeling only the adult Mrgprd+ population that does not co-express Mrgprb4. We and others have previously demonstrated that optogenetic activation of Mrgprd+ neurons is neutral at baseline conditions^27–29, and we recapitulated those findings here (Fig 3e,f). We did not observe conditioned place preference for activation of Mrgprb4-lineage neurons in the back skin of adult male mice (Figure S2). Future studies are necessary to examine the role of Mrgprb4-lineage neurons in male social behaviors, but for this study we focus on female social behaviors.

Figure 3: Focalized activation of Mrgprb4-lineage neurons in the back skin is rewarding and induces a lordosis-like posture in female mice.

A) Mrgprb4^Cre; Rosa^ChR2/ChR2 females and Cre-negative littermates are used in this assay B) Schematic illustrating our real time/conditioned place preference assay. C) Mrgprb4^Cre; Rosa^ChR2/ChR2 females gradually spend more time in the blue laser chamber compared to green laser chamber during the training days (lasers on) and D) exhibit a significantly greater change in time spent in the chamber they learned to associate with blue light on the test day (lasers off) compared to Cre-negative littermate controls. (*p=0.0139, unpaired t-test.) E,F) Mrgprd^CRE-ERT2; Rosa^ChR2/ChR2 females do not develop a preference for the blue laser-paired chamber. H,I) Stills from high speed videography to closely examine the behavior during the preferable transdermal optogenetic activation of Mrgprb4-lineage neurons. Mrgprb4^Cre; Rosa^ChR2/ChR2 females (H) exhibit a striking lordosis-like dorsiflexion in response to Mrgprb4-lineage neuron activation, a behavior absent from (I) Rosa^ChR2/ChR2 female littermates and Mrgprd^CRE-ERT2; Rosa^ChR2/ChR2 females. J) This posture is quantified as the back’s maximum distance from the chamber ceiling over the course of 20s optogenetic stimulation (**p=0.0065, one-way ANOVA). All optogenetic stimuli are 35mW, pulsed at 10Hz sin wave.

We next sought to determine whether a stereotyped behavior could be evoked by optogenetic activation of Mrgprb4-lineage neurons in freely behaving female mice, as an evoked behavior may reveal their role in natural behaviors. We combined optogenetic stimulation with high-speed videography to capture behavior at high spatial and temporal resolution (Fig 3g–j). Intriguingly, a lordosis-like posture, involving robust dorsiflexion, appeared as a stereotyped response to selective activation of Mrgprb4-lineage neurons in the back skin (Video S1,S2; Fig 3h–j). The posture is not observed upon optogenetic activation in the same experimental chambers of either Mrgprd+ neurons (Fig 3j) or Mrgpra3+ neurons (Figure S2), two populations of neurons that share lineage with Mrgprb4. Interestingly, we observed a similar back lowering phenotype to blue light in Mrgprb4^Cre; Rosa^ChR2/ChR2 male mice (Figure S2), however, because the same stimulus does not evoke a conditioned place preference, the valence associated with this posture in males remains unclear. While expression of Mrgprb4 is not sexually dimorphic (Figure S2), we speculate that sexually divergent central circuits underlie behavioral assignment and valence distinctions when activating Mrgprb4-lineage neurons.

Because the earlier study on Mrgprb4^Cre neurons used viral strategies to manipulate or observe the activity of Mrgprb4^Cre neurons¹⁸, we sought to determine whether we could replicate the opto-induced dorsiflexion with a viral strategy. In our genetic approach we label neurons with a developmental history of expressing Mrgprb4, while a viral approach is expected to be more specific to the smaller Mrgprb4+ adult population. To address this, we followed previous approaches of injecting effector proteins into DRG neurons into pups at P0-P3 with intraperitoneal injections of a Cre-dependent virus^{18, 30}. We made a virus packaged with the AAV-PHP.S serotype for efficient infection of peripheral sensory neurons³¹, and we used a Cre-dependent red-orange shifted excitatory opsin, ChRmine³². The efficiency of targeting the Mrgprb4 population that is only ~2% of all DRG neurons¹⁹ is not trivial, and our viral strategy labeled 36% of Mrgprb4+ neurons (Figure S3). Targeting DRG neurons virally is far more challenging than neurons in the brain, since the dozens of sensory ganglia in a mouse cannot all be directly targeted, and thus systemic, and less efficient intraperitoneal injections are typically used. When Mrgprb4^Cre mice were injected with the Cre-dependent ChRmine and orange laser light was pulsed to their shaven back skin, we did not see the lordosis-like posture we observed using Mrgprb4^Cre; Rosa^ChR2/ChR2 mice (Fig 3h). We did however, observe touch-like responses such as turning and orienting towards the light beam on their back (Figure S3). We see two explanations for the lack of dorsiflexion with the viral targeting strategy: 1) Viral targeting yields low transfection efficiency in an already small population of DRG neurons, and 2) there is a distinct behavioral role for neurons with a developmental history of Mrgprb4 expression versus those that retain expression of Mrgprb4 throughout life, in particular for the lordosis-like behavior. Motivated to understand the lordosis-like posture (Fig 3h), and drawn by the more consistent and robust expression of Cre-dependent effector proteins, we proceeded with the genetic approach.

Mrgprb4-lineage neurons are required for female sexual receptivity

Next, we interrogated the role of Mrgprb4 neurons in a natural behavior in which female rodents display prominent dorsiflexion: lordosis, the female sexually receptive posture which includes dorsiflexion in response to male tactile input to the back and flanks^33–36. To test this idea, we ablated the population using Cre-dependent expression of diphtheria toxin subunit A (DTA) (Figure 4a; Figure S4). We used RNAscope in situ hybridization to confirm the selective elimination of Mrgprb4-lineage neurons. (Figure S5b). Mrgprb4^Cre; Rosa^DTA mice did not exhibit any motor abnormalities that might confound the interpretation of our results, and pain and somatosensory detection in the hairy skin appear largely normal (Figure S4).^{29, 37, 38}. (Figure S4). To determine if the Mrgprb4-lineage neurons are required in females to detect male mounts and respond with sexual receptivity, we conducted a lordosis quotient assay^{34, 39}. To control for fluctuations in the estrous cycle, we ovariectomized females (OVX) and subsequently treated them with both estradiol and progesterone to mimic a state of behavioral estrus at the time of the assay.⁴⁰ (Fig 4b). Healthy females have been shown to increase their sexual receptivity with sexual experience, and therefore LQ assays are sometimes conducted over multiple trials, which we did here⁴¹. On the first trial, Mrgprb4^Cre; Rosa^DTA and control females both exhibited moderate levels of sexual receptivity (Fig 4d,e,h,i). While control females increased in receptivity on subsequent trials, strikingly, Mrgprb4^Cre; Rosa^DTA females’ receptivity plummeted on the second pairing, and remained minimal for the third pairing (Fig 4d,e,h,i).

Figure 4: Mrgprb4-lineage neurons are required for female sexual receptivity.

A) Graphic representing Mrgprb4^Cre; Rosa^DTA mouseline, expressing DTA in a Cre-dependent manner to ablate Mrgprb4-lineage neurons. B) Timeline for assessing sexual receptivity with lordosis quotient (LQ) assay: all females are given two weeks to recover after ovariectomy (OVX), at which point estradiol and progesterone are administered to put them in behavioral estrus for an overnight pairing with a male. Hormones replaced prior to each LQ trial. C) Assays conducted in the male home cage in the dark cycle. Graphic depicts a female in sexually receptive lordosis posture in response to male mount D,E) Lordosis quotient scores for 3 sequential trials for WT C57 and Mrgprb4^Cre; Rosa^DTA females. (Two-Way Repeated Measures ANOVA) D) On trials 2 and 3, Mrgprb4^Cre; Rosa^DTA females exhibited significantly reduced sexual receptivity compared to WT females at the same trials (Trial 2 **p=0.0052; Trial 3 ****p<0.0001, Sidak’s multiple comparison test) E) Changes in individual mice across the three trials. WT females exhibit significant increase in sexual receptivity from trial 1 to trial 2 (**p=0.0055) and from trial 1 to trial 3 (***p=0.0009), whereas Mrgprb4^Cre; Rosa^DTA females exhibit significant decrease in sexual receptivity from trial 1 to trial 2 (*p=0.0237) and from trial 1 to trial 3 (**p=0.0059), Tukey’s multiple comparisons test). F) Posture quality assessed on a scale from 1–3, where 1 is the minimum receptive posture and 3 is the most robust posture (details in methods) (ns, Two-Way Repeated Measures Mixed-effects analysis). G) Male mounting frequency was no different between Mrgprb4^Cre; Rosa^DTA females and controls (ns, independent samples t-test). H) Average sexual receptivity, or duration maintained a receptive posture. (**p=0.0021, Two-Way Repeated Measures ANOVA, Sidak’s multiple comparisons test). I) Changes in sexual receptivity in individual mice across trials (ns, Tukey’s multiple comparisons test). J) Total number of combative bouts observed for each female in each trial. (**p=0.0096, Two-Way Repeated Measures Mixed-effects analysis, Sidak’s multiple comparisons test)

While the neurobiology of why females increase receptivity upon repeated exposure to males is unknown⁴¹, it may indicate a rewarding component of sexual encounters that positively reinforces the behavior for females, driving them to engage more receptively on subsequent trials. Our observed decline in receptivity revealed that, without Mrgprb4-lineage neurons, sexual touch becomes negatively reinforcing. Importantly, as evidenced by the number of Mrgprb4^Cre; Rosa^DTA females exhibiting quality lordosis postures on their first trial, the Mrgprb4^Cre; Rosa^DTA females are capable, but choosing not to display lordosis on subsequent trials (Fig 4f). Concomitant with the sharp decline in sexual receptivity in Mrgprb4^Cre; Rosa^DTA females, these females now displayed combative behavior towards the males (Fig 4j), who mounted with the same frequency regardless of female genotype (Fig 4g). This phenotype is striking, as female mice primed for behavioral estrus are not naturally aggressive^{40, 42}. The decline in receptivity despite demonstrated ability to dorsiflex in response to touch, combined with combative behavior towards males, together indicate a role for the Mrgprb4-lineage neurons in the more affective components of touch.

Optogenetic-induced dorsiflexion posture is not modulated by ovarian hormones

Decades of work have established that rodent lordosis is tightly modulated by levels of estrogen and progesterone^{35, 43, 44}. Thus, we tested whether the optogenetically induced posture was modulated by ovarian hormones. We found that the dorsiflexion posture during 20s of optogenetic stimulation did not significantly vary across natural (measured by vaginal lavage) or experimentally manipulated (OVX + hormone replacement) estrous states (Figure S5). Thus, the dorsiflexion induced by optogenetic activation of Mrgprb4-lineage neurons does not indicate sexual receptivity.

Optogenetic activation of Mrgprb4-lineage neurons is sufficient to induce dopamine release in the mesolimbic reward pathway

Given the above results, we hypothesized that Mrgprb4-lineage neurons transduce rewarding sensation during sexual encounters. We examined both back and anogenital skin for Mrgprb4+ nerve endings, areas where male tactile input is expected to occur during sexual behavior. Consistent with prior neuroanatomical studies¹⁹, Mrgprb4+ terminals innervated all examined hairy skin, including the back (Fig 5c), underbelly (Figure S5), and skin surrounding the genitalia in both sexes (Fig 5g, Figure S2), but were absent from the internal lining of the vaginal wall (Figure S5). We sought to determine whether and how these afferents could signal to central reward centers, such as ventral tegmental area (VTA) release of dopamine into nucleus accumbens (NAc), which has been associated with sexual reward^{45–47 48, 49}. Recent work has demonstrated that Mrgprb4+ neurons synapse onto GPR83+ projection neurons in the dorsal horn, which project to lateral parabrachial nucleus (lPBN) among other areas²⁶; see our review⁵⁰. Indeed, we confirm the synaptic connectivity between Mrgprb4-lineage neurons and GPR83+ projection neurons in the spinal cord (Fig 2j–m). We therefore sought to determine whether neurons in the lPBN served as a bridge between GPR83+ projection neurons and the VTA. To address this question, we used viral tracing strategies to examine whether there were lPBN neurons that both received GPR83 input and projected to the VTA. To target GPR83+ projection neurons, we utilized combinatorial CRE-DOG viruses, whereby viral expression of a split Cre combines into a functional Cre in the presence of GFP⁵¹. We injected pAAV.CAG.LSL.tdTomato and combinatorial CRE-DOG viruses into the lower thoracic spinal cord of GPR83-eGFP animals to label GPR83+ projection neurons with tdTomato (Figure S6a). By injecting a BFP retrograde virus into VTA of the same animals (Figure S6a,b), we found that BFP+ (VTA-projecting) cell bodies in lPBN had visible overlap with spinoparabrachial GPR83+ (red) terminals (Figure S6c, c’). These results offer a putative neuroanatomical pathway emerging from the skin and spinal cord to the mesolimbic reward center.

Figure 5: Activation of Mrgprb4-lineage neurons triggers dopamine release in the nucleus accumbens.

A) Schematic depicting the experimental setup. Female Mrgprb4^Cre; Rosa^ChR2/ChR2 mice with shaved backs, which had been injected with GRAB_DA to the NAc two weeks prior to testing, were placed under plastic chambers on a mesh platform. Pulsed blue (stimulating) or green (control) laser light (35mW, 10Hz sin wave) was shined to either the back skin (C-F) or skin surrounding the vagina (G-J) while recording GRAB_DA signals. B) Cannula placement and GRAB_DA expression in NAc C) ChR2-eYFP expression in Mrgprb4-lineage terminals in the back skin. D) Average GRAB_DA signal as z-score upon blue light (D) or green light (E) stimulation to the back. F) Average deltaF/F signals during pre-stimulation baseline (−5–0s) and during stimulation (0–6s) for each animal. Repeated measures two-way ANOVA revealed significant interaction between wavelength and time (p=0.0161) and Sidak’s multiple comparison test revealed significant difference in blue light (**p=0.0014) but not green light (p=0.9411) pre-stim compared to during stimulation. G-J) Average GRAB_DA signal as z-score upon blue light (H) or green light (I) stimulation to the anogenital region. J) Average deltaF/F signals during pre-stimulation baseline (−5–0s) and during stimulation (13–20s) for each animal. Repeated measures two-way ANOVA revealed significant interaction between wavelength and time (p=0.0110) and Sidak’s multiple comparison test revealed significant difference in blue light (***p=0.0003) but not green light (p=0.8547) pre-stim compared to during stimulation.

To determine whether this anatomical circuit was functionally connected, we coupled fiber photometry with transdermal optogenetic stimulation to test whether activation of Mrgprb4-lineage neurons is sufficient to induce dopamine release in NAc (Fig 5a). We measured dopamine release using the GRAB_DA dopamine sensor, which we injected into the NAc of Mrgprb4^Cre; Rosa^ChR2/ChR2 females (Fig 5a,b). The GRAB_DA sensor is a GPCR-based approach where dopamine release and binding to its cognate receptor generates fluorescence with subcellular resolution and sub-second kinetics⁵². Mrgprb4^Cre; Rosa^ChR2/ChR2 females expressing the GRAB_DA dopamine sensor in NAc were given either 30s transdermal optogenetic stimulation to the back or anogenital region or the same treatment with non-stimulating green light while dopamine signals were recorded. Interestingly, we observed a significant increase in the z-scored deltaF/F from baseline to the first 6 seconds of blue-light stimulation to the back, but not at these same timepoints with non-stimulating green light application (Fig 5d–f). We next turned to the Mrgprb4-lineage neurons innervating the skin surrounding the genitalia. Intriguingly, blue light stimulation to skin surrounding the genitalia, but not green, was sufficient to raise z-scored deltaF/F significantly from baseline levels (Fig 5h–j). With blue light stimulation to the genitalia, sometimes the mouse investigated the area where it had been “touched”, but a lordosis-like posture was never observed (Video S3). This demonstrates that the Mrgprb4-lineage neurons in the back and anogenital skin are sufficient, in otherwise socially-isolated mice, to induce dopamine release in NAc.

Mrgprb4-lineage neurons are required for dopamine release during sexual behavior

Based on the results above, we hypothesized that Mrgprb4-lineage neurons drive dopamine release during sexual behaviors. To test this hypothesis, we injected Mrgprb4^Cre; Rosa^DTA females and littermate controls with the GRAB_DA dopamine sensor to the NAc to examine any differences in dopamine release during sexual behavior in the absence of Mrgprb4-lineage neurons (Fig 6a). We measured dopamine activity in the seconds surrounding mounts on their first trial, and found that while Cre-negative littermate controls exhibited a significant increase in dopamine release immediately following mount initiation, this dopamine release was missing in Mrgprb4^Cre; Rosa^DTA females (Fig 6b–d). Thus, the Mrgprb4-lineage neurons are indeed required for the sexual reward that occurs at mount onset.

Figure 6: Mrgprb4-lineage neurons are required for dopamine release in the nucleus accumbens during sexual mounts.

A) Graphic depicting experimental setup. Female Mrgprb4^Cre; Rosa^DTA mice or littermate Rosa^DTA controls, which had been injected with GRAB_DA to the NAc over two weeks prior to testing were paired with males for a mating assay. Females were ovariectomized and hormone primed to be in a state of behavioral estrus for testing. GRAB_DA signals were recorded for the entire pairing and analyzed surrounding male mounts. B) Average z-score traces surrounding mount onset (T=0) for littermate Rosa^DTA controls (gray) (N=12) and C) Mrgprb4^Cre; Rosa^DTA females (purple) (N=7) D) Area under the curve for Z-scored signals at pre-mount baseline (−5 to −2s), 0–3s post-mount, and 3–5s post-mount for each animal. Repeated measures two-way ANOVA revealed a significant interaction between genotype and time (p=0.037), and Sidak’s multiple comparison test revealed significant difference in GRABDA signal between baseline and 0–3s post mount for control (**p=0.0024) but not Mrgprb4^Cre; Rosa^DTA females (p=0.9945). E) Schematic of a miniaturized one-photon fluorescence microscope and calcium imaging. A DAT-Cre animal was unilaterally injected with jGCaMP8m virus into the left hemisphere of the ventral tegmental area (VTA; AP:−3.20 mm, ML: −0.40 mm, DV: −4.40 mm). DAPI-stained (blue) coronal section of DAT-Cre mouse showing GCaMP8m fluorescence (green) and the location of a 0.6mm-diameter gradient index (GRIN) lens implant (dashed lines). The inset image indicates the imaging field from the lens. F) A cell contour map (left) and CNMFe-processed image of the neurons located in the VTA (middle) with corresponding raw calcium traces (right). G) Sample video frames of anogenital sniffing and mounting. The animal poses were manually annotated on a frame-by-frame by two human annotators. H) Raw calcium traces from 7 dopamine cells in VTA (black lines) with two manual behavior annotations: anogenital sniffing (blue; 14 events) and mounting (red; 16 events). Median z-scored calcium signals of each dopamine cell at the time of anogenital sniffing (blue; left) or mounting (red; right). I) Representative calcium activity of VTA dopamine neurons during anogenital sniffing or mounting (n=23 cells, 3 mice).

Lastly, we investigated whether individual VTA dopamine neurons were activated by different types of sexual touch in females, specifically anogenital sniffing and mounting (Fig 6g). We introduced microendscopic calcium imaging of VTA dopamine neurons in freely behaving female mice and recorded the activity during mating behaviors (Fig 6e,f). Indeed we found that the individual dopamine neurons exhibited distinct patterns of activity depending on the type of touch, with some responding to anogenital sniffing, some to mounting, some to both, and some to neither (Fig 6h, i). These data suggest that different populations of dopamine neurons in VTA are related to anogenital sniffing versus mounting, implying heterogenous dopamine responses for rewarding social touch. We speculate that some of these touch inputs arise from direct input of Mrgprb4-lineage touch neurons.

Discussion

Although affective social touch begins at the skin’s surface, the molecular identity of sensory neurons in the skin that detect socially relevant signals, and the ascending pathways that connect them to reward centers in the brain have remained poorly defined. Moreover, because touch itself is highly heterogeneous (i.e., discriminative touch to detect texture with our fingertips versus the affiliative touch during a hug from a friend), sensory perception is likely generated by different sets of neurons to provide specificity^{16, 22, 24, 53–57}. Might there be a specific subset of DRG neurons with a predominant role in rewarding social touch, including sexual receptivity? One class of DRG neurons termed C-low threshold mechanoreceptors (C-LTMRs), or C-tactile afferents in humans, are implicated in detecting gentle strokes across the skin’s surface, including interpretation of these signals as erotic, although in a sex-dependent manner^{13, 58}. Although a limited number of papers in the mouse identify molecular populations of C-LTMRs, including their neuroanatomy and roles in somatosensation during baseline and chronic pain states^{15, 24, 59–64}, the field is beginning to emerge with exciting new findings. For example, two recent studies in mice identify different classes of C-LTMRs and show that animals prefer their activation and modulating these neurons alters social behaviors^{23, 65}.

Here, beginning with mouse sensory neurons that are anatomically and physiologically similar to human C-tactile afferents^{18, 19}, we propose a behavior-first approach. We demonstrate that Mrgprb4-lineage neurons are critical for specific social behaviors and for signaling social reward to the brain. Focal activation of Mrgprb4-lineage neurons yields a striking dorsiflexion posture resembling mammalian lordosis, representing a behavioral response to optogenetic activation of social touch neurons. However, their role in female sexual behavior is not simply in facilitating lordosis; rather, the neurons convey an affective sensation that positively reinforces sexual receptivity, and without them, male advances during sexually receptive hormonal states become aversive. This finding suggests that Mrgprb4-lineage neurons contribute to the perceived valence of sexual encounters in females by encoding the rewarding aspect of male sexual touch. We demonstrate this functional link from Mrgprb4-lineage neurons in the skin to reward circuitry in the brain through fiber photometry with the GRAB_DA dopamine sensor. Transdermal optogenetic activation of Mrgprb4-lineage neurons in the back skin or skin surrounding the female genitalia is sufficient to induce dopamine release in the NAc. This is the first report of molecularly defined somatosensory neurons triggering activation of a brain reward center. Finally, we link this stimulation of rewarding touch to natural tactile reward by demonstrating that Mrgprb4-lineage neurons are required for this same dopamine release at the onset of a male mount.

In summary, the work described here has revealed the sufficiency of peripheral inputs to regulate social reward independent of context and other sensory cues. Since ventral tegmental dopaminergic neurons that project to the NAc are themselves functionally heterogeneous^{66, 67}, it is possible that some of these neurons might be tuned for encoding rewarding touch, and our microendoscopy results in the VTA support such a hypothesis. Future studies are warranted to tease apart whether Mrgprb4-lineage neurons engage unique ensembles of VTA dopamine neurons, and in what behavioral contexts. Thus, it will also be interesting to determine if Mrgprb4-lineage neurons are involved in other behaviors that integrate social touch, such as maternal care or even play or bonding between animals. We believe this study draws new attention to the importance of elucidating skin-brain circuits. Moreover, this work points towards the therapeutic potential of peripheral manipulations for enhancing intact or impaired social reward systems, including sexual dysfunction, or simulating social reward during periods of isolation.

Limitations of the study

In this work we perform circuit-based studies using the Mrgprb4^Cre mouse line where we activate or ablate the entire population of neurons in behavioral settings. Moreover, we demonstrate in vitro that Mrgprb4-lineage neurons are mechanosensitive. Important questions thus remain regarding the biological basis of mechanical transduction in Mrgprb4-lineage neurons: 1) what is the mechano-sensor in these cells that drives activation of these neurons in natural scenarios? and 2) are there unique neuropeptides beyond the canonical excitatory neurotransmitter glutamate that drive activation of this circuit? In regards to mechano-sensors, the Piezo2 mechanosensory ion channel seems fitting, but recent single-cell RNAseq data demonstrate that almost no Piezo2 mRNA is co-expressed in Mrgprb4 neurons²². Thus, further studies in the field that combine RNA sequencing, physiology, calcium imaging, and other biophysical approaches are needed to determine the molecular nature of mechanosensation in Mrgprb4 neurons, and to what extent, if any, the Mrgprb4 receptor itself may be involved in signal transduction in these neurons.

STAR Methods

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by lead contact, Dr. Ishmail Abdus-Saboor (ia2458@columbia.edu)

Materials Availability

This study did not generate new unique reagents. All mouselines were obtained from Jackson Laboratories, and viruses from Addgene, except the pAAV-Ef1a-DIO-ChRminemScarlet-3xHA (PHP.S) virus which was custom generated by core facilities at the Zuckerman Mind Brain Behavior Institute. Any inquiries for particular mouse crosses generated from Jackson Laboratories strains can be directed to the Lead Contact.

Data and Code Availability

• All data reported in this study will be shared by the lead contact upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental Model and Subject Details

All testing was performed in compliance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania, Columbia University, and Rutgers University. Mice born into our colony on a C57bl/6J background were maintained in conventional housing with food and water available ad libitum when not being tested, on a 12 hour light cycle beginning at 7:00, and all testing unless otherwise noted occurred during the light cycle in a room directly adjacent to the housing room in the animal facility. Mouse lines used in this study are located at Jackson Laboratories: C57BL/6J (Stock No: 000664), Mrgprb4^Cre (Stock No: 021077), Mrgprd^Cre-ERT2 (Stock No: 031286), Rosa^ChR2-eYFP (Stock No: 024109), Rosa^DTA (Stock No: 009669). The MrgprA3^Cre line was generously provided by Dr. Xinzhong Dong at Johns Hopkins University School of Medicine. Breeding was conducted as follows: for optogenetic experiments: Mrg(X)^Cre; Rosa^ChR2/ChR2 X Rosa^ChR2/ChR2. For ablation experiments: Mrgprb4^Cre X Rosa^DTA/DTA. Both sexes of mice were tested and all experiments were performed in adult animals at least 6–8 weeks old. Viral injection experiments began with Postnatal day 0 pups.

Method Details

RNAscope in situ hybridization

Brains were harvested immediately following transcardial perfusion and post-fixed in 4% PFA in PBS at 4°C for 24h. Brains were subsequently submerged in 30% sucrose in PBS at 4°C for 18h or until sunk, flash frozen in cryomold on dry ice in OCT, sectioned at 12μm directly onto Superfrost Plus Slides and stored in foil-wrapped slide box at −80°C until beginning Fixed Frozen RNAscope protocol (ACD).

Entire spinal columns were harvested immediately following transcardial perfusion and post-fixed in 4% PFA in PBS at 4°C for 24h. Either spinal cord or DRG were subsequently dissected and submerged in 30% sucrose overnight or until sunk. Tissue was flash frozen in cryomold on dry ice in OCT, sectioned at 12μm directly onto Superfrost Plus Slides, and stored in foil-wrapped slide box at −80°C until beginning Fresh Frozen RNAscope protocol (ACD). Note the fixation step was skipped in ACD’s Fresh Frozen protocol for these tissues.

Immunohistochemistry

Brains, spinal cord, and DRG were collected and prepared in the same way as in situ hybridization. 30μm cryosections were cut directly onto Superfrost Plus Slides. Slides were frozen overnight in a slide box at −80°C. Slides were washed 3× 10min in PBS; 30 minutes in PBST; 1 hour in PBST with 5% normal donkey serum. Primary antibody, 1:1000 in PBST with 5% normal donkey serum, was applied to slides in a humidified chamber overnight at room temperature. Secondary antibody, 1:400 in PBST with 5% normal donkey serum, was applied to slides in a humidified chamber 1–2 hours at room temperature. Slides were washed 3× 10 minutes in PBS before applying mounting media and coverslip.

Transdermal optogenetic activation of Mrgprb4-lineage neurons

8–16 week old female mice were habituated to mesh platform under a plastic chamber for 1 hour on each of two days prior to behavioral testing. Experimenter was present with lights, camera, and laser running during habituation to mimic entire sensory experience of test day. On the day of testing, the mice were habituated to the chambers for an additional 20 minutes before stimulation. 35mW blue laser light pulsed at 10 Hz sin wave was shined through the ceiling of the chamber to the shaved backs of mice for 20 seconds during high-speed video captured at 750fps. For female mice, testing was performed such that each female was tested once daily for five consecutive days. The average of these five trials for each mouse is plotted in Fig 3. Because no variation with the estrous cycle was observed (Supplemental Fig 5), all other testing of transdermal optogenetic stimulation to the back was conducted on a single test day after habituation

Labeling Mrgprb4 neurons with viral injection of excitatory opsin and scoring touch-like behaviors

As an alternative strategy to our genetic approach, we intraperitoneally injected 30μl (8.70E+12 vg/mL) of pAAV-Ef1a-DIO-ChRmine-mScarlet-3xHA (PHP.S) at postnatal days 0–3. Behavioral testing began 6–8 weeks following injection. Animals were habituated to the presence of orange laser light and the experimenter as described above. On the day of testing, orange laser light (598nm) was pulsed through the shaved back skin similarly to what was done for transdermal optogenetic activation of the lineage population. Behavioral videos were scored by a blinded observer for instances where the mouse either looked at the light or reoriented its body as a result of the light stimulation. These were considered touch-like behaviors because the mouse reacted as if it had been touched to the back. Moreover, these behaviors were never observed in littermate control animals.

Tamoxifen injection for Mrgprd^Cre-ERT2; Rosa^ChR2/ChR2 mice

To induce expression of Cre recombinase in Mrgprd^Cre-ERT2; Rosa^ChR2/ChR2 mice, we intraperitoneally injected 0.5mg tamoxifen in 100μl sunflower seed oil in mice aged P10 or older. We repeated this injection daily for three days, so each mouse received three 0.5mg doses. Behavioral testing began at least two weeks after the third injection to allow for Cre mediated recombination and expression of the opsin protein.

Quantification of the back dip

Optogenetic-induced dorsiflexion posture was calculated as the maximum back dip from the ceiling of the behavioral chamber for the duration of the 20s stimulation. High speed videos were analyzed in ImageJ to track the lowest point of the back throughout the 20s stimulation. The lowest y value was subtracted from the y value of the top of the chamber, and the value was converted from pixels to millimeters by determining the pixel height of the 4.5cm chamber for each video. It was decided that this was the most quantifiable way to characterize the dip. While duration or frequency of response require a subjective determination of when a back dip starts and stops – and therefore can be more challenging to compare to control animals – the back dip depth is the most objective measure because it allows us to compare between natural movement of the spine in controls, small back dips, and most drastic back dips. This measure encapsulates the full variability of response.

Conditioned Place Preference

8–14 week old Mrgprb4Cre; RosaChR2/ChR2 females or MrgprdCRE-ERT2; RosaChR2/ChR2 and Cre-negative Rosa^ChR2/ChR2 female littermates underwent a 9 day conditioned place preference paradigm. The apparatus, originally designed at the University of Iowa and described previously⁶⁸, consisted of two chambers, one paired with almond extract and a textured floor, the other coconut extract and a smooth floor. Additionally, one chamber wall had stripes, while the other had polka dots. These olfactory and visual stimuli were present throughout the 9 day paradigm to aid in the mouse’s encoding of different chambers. Days 1–3 were habituation: each mouse was allowed 20 minutes to explore the two chambered apparatus with no optogenetic stimulation. Days 4–8 were training: each mouse was allowed 20 minutes to freely move about the two-chambered apparatus, now receiving laser light stimulation to the back. In one chamber the mouse received 10Hz pulsed sin wave 35mW blue laser light to the shaved back, and the other chamber non-stimulating green light of the same parameters. Experimenter held the laser lights ~1cm from the back skin for the duration that the mouse was in the chamber. The lasers were held with an extendable alligator clip to avoid casting body shadow on the chambers. Because the back was the only shaved body region, and represents the largest body area, we believe our results are driven by the activation of Mrgprb4-lineage neurons in the back skin, as opposed to rare moments the laser passed over the ear, foot, or tail. Day 9 was test day: each mouse was allowed 20 minutes to freely move about the two-chambered apparatus in the absence of any laser light stimulation. Video was scored manually by blinded experimenter. Baseline preference was calculated for each mouse by averaging the duration spent in each chamber across the three habituation days. The conditioned preference was calculated for each mouse as the duration spent in each chamber on the test day. Percent change was calculated as percent time in blue light chamber after training – percent time in blue light chamber before training.

Determination of natural estrous state

Natural estrous cycle was determined by vaginal lavage. Immediately following behavioral testing, the vagina was flushed with 20μl ultrapure water which was then pipetted onto a Superfrost plus slide for examination under dissecting scope. Mice were tested mid-morning each day to ensure the most accurate tracking. Lavage samples were assessed as described⁶⁹, and estrous state was recorded throughout the week to ensure normal cycling. Data from mice with lavage samples that could not be fit into a typical four or five day cycling pattern were excluded.

Ovariectomy Surgery

Ovariectomies were performed on 8 week old female mice under 1.5–2% isoflurane using proper sterile technique. 5mg/kg oral or intraperitoneal meloxicam and 2mg/kg subcutaneous bupivacaine (at incision sites) were administered before surgery. A 0.5cm incision is made 1cm lateral to spinal cord, at the point where ribcage ends. An equivalent incision was made through the muscle wall. The white fat pad was exposed to identify the ovary, which was cauterized with a hemostat and scraped off with a scalpel. The fat pad was reinserted and 1–2 sutures closed the muscle wall. 2–4 sutures closed the skin. Meloxicam was administered 24 hours after surgery and mice were allowed two weeks to recover before behavioral testing.

Lordosis Quotient Assay

10–14 week old ovariectomized females underwent two overnight pairings with stud males two weeks following surgery. To mimic behavioral estrus state at the time of pairing, females were subcutaneously injected with 0.5 μg estradiol benzoate in sunflower seed oil 52 hours before pairing, and with 800 μg progesterone in sunflower seed oil 4 hours before pairing. The females received the same hormone treatment prior to the lordosis quotient assay. The three trials were spaced 1–3 weeks apart to allow circulating hormones to reset before the next hormone priming. Lordosis quotient assay was conducted in male home cage, 1–2 hours into the dark cycle. Video of the assay was recorded for 20 minutes or 20 attempted male mounts, whichever came first. Lordosis quotient was calculated as the number of female receptive responses divided by the number of attempted male mounts. A receptive response for the female was scored as all four limbs securely on the cage floor with no combative or escape behaviors. If the first male did not mount within 10 minutes, the female was moved to another male’s cage. For any given trial, up to three males may have been used. Receptive postures were scored for quality on a scale from 1–3 as follows: 1: limbs on the ground with no attempts to escape, neither dorsiflexion nor upturned nose. 2: Some dorsiflexion, no upturned nose. 3: Robust dorsiflexion and/or upturned nose. Posture scores were averaged within each trial.

Intracranial viral injections and optic fiber implantation

7–10 week old females were pretreated with 5mg/kg oral or intraperitoneal meloxicam before surgery. They were anesthetized in a chamber with 3% isoflurane before being placed in a stereotaxic frame and kept anesthetized with 1.5–2% isoflurane. Skull was exposed and leveled before drilling at the appropriate coordinates. 200nL pAAV-hSyn-GRAB_DA1h or 200nL pAAV-hSyn-GRAB_DA2mwas injected unilaterally into NAc (AP: +1.0 mm relative to Bregma, ML: ±1.2 mm relative to Bregma, DV: 4.6 mm from the brain surface) using a backfilled glass needle and syringe pump (PHD Ultra, Harvard Apparatus), or nanoject. Skin was either sutured for one week recovery before implanting optic fiber (Doric, MFC_200/230–0.57_6mm_MF1.25_FLT) 200μm above injection site, or optic fiber was inserted immediately following injection. Three skull screws were inserted in the skull to stabilize the implant. A small amount of Metabond cement (Parkell) was used to bond the fiber, skull, and skull screws. Dental cement was applied on top of dried Metabond to create a stable implant structure. Mice were given 1 week to recover from implant surgery and 2 weeks to recover from injection and implant combined surgeries before behavioral testing. Meloxicam was administered 24 hours after surgery during post-operative monitoring.

Viral injections and GRIN lens implantation

A DAT-Cre animal was unilaterally injected with 300nL of AAV1-syn-FLEX-jGCaMP8m (titer: 1.6E13 GC/mL; Addgene #162378-AAV1) into the left hemisphere of the ventral tegmental area (VTA; AP:−3.20 mm, ML: −0.40 mm, DV: −4.40 mm). A 0.6mm-diameter gradient index (GRIN) lens (1050–004413, 7.3mm length, NA 0.5; Inscopix, Mountain View, CA, USA) was implanted in the same place of the virus injection site. Once in place, the lens was secured to the skull with super glue (Loctite, LOC1255800) and dental cements (Ortho-Jet^™ Liquid, Black, Lang Dental, Wheeling, Illinois, USA). The imaging field of view was inspected and allowed to clear for at least five days prior to imaging and behavioral experiments.

Fiber Photometry

Zirconia sleeves (Doric) were used to connect the optic fiber implant to the patch cord. Signals were recorded using a real-time processor (RZ10X, TDT) and extracted in real time using Synapse software (TDT). A 465nm LED was used to excite the GRAB_DA while a 405nm LED was used as isosbestic to measure and control for changes in fluorescence due to photobleaching and movement artifacts.

For fiber photometry during transdermal optogenetic activation experiments, mice were placed on the mesh platform in plastic chambers with a 7mm slot cut into the ceiling to allow the cord to connect from the head to the computer. Habituation to the chambers was as previously described, except the fiber was attached to the implanted cannula after the first habituation day. After 30 second baseline recording, 10Hz pulsed sin wave at 35mW blue light was shined through the mesh platform to the shaved back skin or vaginal area at a 2–3cm distance for 30 seconds. The order in which the mice received stimulation (blue vs green; anogenital vs back) was counterbalanced.

For fiber photometry during sexual behavior, females were ovariectomized and hormone primed in the same way as the lordosis quotient assay without photometry. The recordings presented are equivalent with trial 1 from the lordosis quotient assay. (First assay after an overnight pairing the week prior). Females were placed in male home cage 1 hour into the dark cycle for 10 minutes or 10 mounts, whichever occurred first. If the male did not mount in the first 5 minutes, the female was placed in another male cage, for up to three total males.

Microendoscope Imaging

Mice were placed in their home cages and allowed to habituate with the head-mounted dummy scope for at least three days (15min each session). Before starting mating assays, the mice were briefly anaesthetized using a mixture of isoflurane and oxygen and the mini-epifluorescence microscope was attached to the baseplate. A period of 20–30 minutes was allowed to recover in the home cage before experiments started. Calcium imaging data were acquired using Inscopix Data Acquisition Software (version 1.3.0) at a frame rate of 20 Hz, 1mW/mm² LED power (475nm) with 4–5x gain, depending on the brightness of GCaMP expression. Image acquisition parameters were set to the same values between sessions to be able to compare the activity recorded. Females were ovariectomized and hormone-primed as described for other mating assays.

Microendoscope fluorescence trace extraction

Imaging frames were spatially downsampled by a factor of four in the x and y dimensions. Motion artifacts were corrected using Inscopix Data Processing software (version 1.6.0). To extract calcium activity traces of individual cells, we used the constrained non-negative matrix factorization for microendoscopic data (CNMF-E) algorithm⁷⁰. Segmented cells were manually inspected and non-cell-like shapes or cells with activity traces with non-calcium like events were discarded. Accepted cells and their activity traces were exported to MATLAB or Python for further analysis.

Electrophysiology

For all electrophysiology recordings, tissue was transferred to a recording chamber and continually superfused with ACSF bubbled with Carbogen (95% O₂ and 5% CO₂) to achieve a pH of 7.3–7.4. All recordings were made at room temperature (22–24°C) and neurons visualized using a Zeiss Axiocam 506 color camera. Recordings were acquired in cell-attached (holding current = 0mV), voltage-clamp (holding potential −70mV), or current-clamp (−60mV) configuration. Patch pipettes (3–7 MΩ) were filled with a potassium gluconate-based internal solution containing (in mM): 135 C₆H₁₁KO₇, 8 NaCl, 10 HEPES, 2 Mg₂ATP, 0.3 Na₃GTP, and 0.1 EGTA, pH 7.3 (with KOH). No liquid junction potential correction was made, although this value was calculated at 14.7 mV (22 °C). All data were amplified using a MultiClamp 700B amplifier, digitized online (sampled at 20 kHz, filtered at 5 kHz) using an Axon Digidata 1550B, and acquired using Clampex software. After obtaining the whole-cell recording configuration, series resistance, input resistance, and membrane capacitance were calculated (averaged response to −5mV step, 20 trials, holding potential −70mV).

Acute slice

Recordings were made from Mrgprb4^Cre;Rosa^ChR2/+ or Mrgprb4^Cre;Rosa^ChR2/ChR2 mice (female; age 3.9 ± 0.6 weeks). Mice were anaesthetized with ketamine (100 mg/kg i.p), decapitated, and spinal cord (T10-L2) rapidly removed in ice-cold sucrose substituted artificial cerebrospinal fluid (sACSF) containing (in mM): 250 sucrose, 25 NaHCO₃, 10 glucose, 2.5 KCl, 1 NaH₂PO₄, 6 MgCl₂, and 1 CaCl₂. Sagittal slices (200μm thick) were prepared using a vibrating microtome (Leica VT1200S). Slices were incubated for at least 1hr at 22–24°C in an interface chamber holding oxygenated ACSF containing (in mM): 118 NaCl, 25 NaHCO₃, 10 glucose, 2.5 KCl, 1 NaH₂PO₄, 1 MgCl₂, and 2.5 CaCl₂. Slices were illuminated using an X-CITE 120LED Boost High-Power LED Illumination System that allowed visualization of GFP fluorescence and ChR2 photostimulation using FITC filter set. Photostimulation intensity was suprathreshold (24 mW), duration 1 ms (controlled by transistor-transistor logic pulses). AP discharge patterns were assessed in the current clamp from a membrane potential of −60 mV by delivering a series of depolarizing current steps (1 s duration, 20 pA increments, 0.1 Hz)

DRG cell culture

Recordings were made from Mrgprb4^Cre;Rosa^ChR2/GFP mice (male and female; age 10.3 ± 0.1 wks). Mice were anaesthetized with ketamine (100 mg/kg i.p), decapitated, and DRGs dissected in ice-cold HBSS solution. DRGs were then incubated in a mix of collagenase type XI (900 μ/ml) and dispase (3 mg/ml) in INCmix medium (in mM: 155 NaCl, 1.5 K₂PO₄, 10 HEPES, and 5 glucose buffered to pH 7.4) containing 1% penicillin and streptomycin for 45–60 mins at 37°C in 5% CO₂. After enzyme digestion a gentle mechanical dissociation was performed by aspiration/expulsion using a 1ml pipette and passed through a 70μm nylon filter. Enzymes were stopped by adding Ca²⁺- and Mg²⁺-free HBSS medium containing 1% MEM Vitamin Solution, 10% FBS, and 1% penicillin and streptomycin, followed by centrifugation at 300G for 8 minutes. The pellets were resuspended in MEM supplemented with 1% MEM Vitamin Solution, 10% FBS, and 1% penicillin and streptomycin. Cell suspensions were plated on poly-l-lysine-coated glass coverslips and maintained at 37°C in 5% CO₂ for 24hrs. Graded mechanical stimulation of DRG neuron cell bodies was performed using a glass pipette driven by a Siskiyou MX7600 motorized micromanipulator. The pipette was positioned on a cell body at a 45° angle and opposite to the recording pipette.

Viral Tracing

(N-terminal and C-terminal fragments AAV-EF1a-N-CretrcintG and AAV-EF1a-C-CreintG, Addgene # 69571 and # 69570) We injected pAAV.CAG.LSL.tdTomato (Addgene #100048) and combinatorial CRE-DOG viruses into the lower thoracic spinal cord of GPR83-eGFP animals to label GPR83+ projection neurons with tdTomato (Figure S6a). By injecting a BFP retrograde virus (pENN.AAV.CB7.CI.mCerulean.WPRE.RBG) (Addgene#105557-AAV9) into VTA of the same animals (Figure S6a,b),

Quantification and Statistical Analysis

Statistical analysis, tests used, n, and parameters for determining significance can be found in the figure legends. Unless otherwise noted, for all repeated measures behavioral testing, Two-Way Repeated Measures ANOVAs were performed, followed by multiple comparisons testing if a main effect or interaction was observed. If there were missing values, a mixed-effects model was used instead. When only one variable was tested (e.g. genotype but not time/trial), a One-Way ANOVA (if more than two groups) or an independent samples t-test (if only two groups) was performed.

Fiber Photometry Analysis

TDT folders were imported directly into Fiber photometry Modular Analysis Tool (pMAT) software for analysis⁷¹. Fiber photometry during transdermal optogenetics: Z-scored deltaF/F was calculated in pMAT by normalizing to the median of 5s baseline sampling window (−10 – −5s), bin constant: 150ms. These data were exported in a csv file from pMAT into GraphPad prism where the average trace across animals was plotted with shaded region representing SEM. These data were then normalized to the mean of the plotted baseline for visualization. The average raw deltaF/F in time bins was calculated for each animal and plotted in GraphPad prism to visualize changes from baseline to stim in individual animals and for statistical anaylsis. Mice were excluded if cannula placement or viral injection was not found in NAc after immunofluorescence. Fiber photometry during sexual behavior: Z-scored deltaF/F was calculated in pMAT by normalizing to the median of 3s baseline sampling window (−5 – −2s), bin constant: 300ms. These data were exported in a csv file from pMAT into GraphPad prism where the average trace across animals was plotted with shaded region representing SEM. These data were then normalized to the mean of the plotted baseline for visualization. Area under the curve was exported as csv from pMAT and plotted in GraphPad prism for statistical analysis.

Supplementary Material

1

Supplemental Figure 1: Characterizing physiological properties of dorsal horn spinal cord neurons that receive synaptic input from Mrgprb4-lineage neurons, Related to Figure 2. A) Schematic of the spinal cord slice preparation to activate Mrgprb4+ terminals with blue light in the superficial dorsal horn and record from synaptically connected neurons. B) Grouped data of mono and polysynaptic EPSC amplitudes. C) Incidence of a monosynaptic only, polysynaptic only, or mono + polysynaptic input. D) Optically evoked action potential (AP). 15/39 neurons tested fired AP’s following 1ms photostimulation. E) Latency of EPSCs (mono and polysynaptic) in these neurons, compared to the latency of the first evoked AP. Of these 15 neurons, 7 display characteristics of AP’s evoked by direct Mrgprb4 afferent input, 8 display characteristics of AP’s evoked by indirect (polysynaptic) Mrgprb4 inputs. F,G) In a subset of recordings AP discharge was characterized in post-synaptic neurons (n = 36). 2/14 Tonic firing (TF) neurons received mono input only, 4/14 poly only, and 8/14 both. 0/15 Delayed firing (DF) neurons received mono input only, 12/15 received poly only, and 3/15 both. 1/4 Initial bursting (IB) neurons received mono input only, 1/4 received poly only, and 2/4 both. 0/3 Single spikers (SS) received mono input only, 1/3 received poly only, and 2/3 both. H) TF neurons receive the strongest monosynaptic inputs, followed by IB’s. I) Polysynaptic currents are similar across all types. J) AP incidence is highest in TF neurons.

2

Supplemental Figure 2: Optogenetic behaviors stimulating either MrgprA3 neurons or Mrgprb4-lineage neurons in males, Related to Figure 3. A) Schematic showing experimental setup with blue light applied to the shaved back skin, which elicits nearly immediate scratching bouts, and not the back lowering dorsiflexion phenotype observed when the same experiments are performed with Mrgprb4-lineage neurons. B) Quantification of scratching bouts to 5 minutes of blue or green laser light (negative control). Unpaired student’s t-test, **** P ≤ 0.0001. Individual circles represent a single mouse. C,D) Optogenetic activation of Mrgprb4-lineage neurons in the shaved backs of Mrgprb4^Cre; Rosa^ChR2/ChR2 but not Cre-negative littermate controls yields dorsiflexion posture (blue light 35mW pulsed at 10Hz sin wave). E) Maximal backdip from the ceiling of the chamber over the course of 20s stimulation. P<0.0001, independent samples t-test. E-G) No significant difference in the abundance of Mrgprb4+ neurons in thoracic DRGs of male (E) vs female (F) mice. (ns, Independent samples t-test) H) Male mice exhibit no significant difference in preference for blue light chamber compared to Cre-negative littermate controls (ns, independent samples t-test). Same conditioned place preference paradigm described in Fig 3a–f. I) Mrgprb4-lineage neurons innervate the male genital skin. Immunofluorescence staining for ChR2-eYFP+ terminals.

3

Supplemental Figure 3: A viral strategy to target Mrgprb4+ sensory neurons reveals touch-like behaviors and no dorsiflexion, Related to Figure 3. A) Both experimental and control litter pups were injected with 30microliters of cre-dependent opsin virus between P0-P3. B) 36% of Mrgprb4+ neurons were labeled with the cre-dependent opsin Chrmine. C) Tissue section through a DRG showing ChRmine virus typically labeling 1 of every 3 Mrgprb4+ neurons. D) Typical orange laser light induced behaviors where animals would orient towards the light searching, appearing to search for the source of the touch. E) Quantification of number of touch-like responses during 20 seconds of pulsing laser light. Control animals never appeared to care about the light or alter behavior. All scoring was done blind to genotype, conducted by different investigators. p=0.0085, unpaired t-test.

4

Supplemental Figure 4: Ablating Mrgprb4-lineage neurons does not alter motor function or pain threshold, Related to Figure 4. A) RNAscope in situ hybridization of DRGs from WT C57 control and Mrgprb4^Cre; Rosa^DTA female. Scale bars = 75μm. B) Quantification of ablation in A. Cell bodies expressing Mrgprb4 are absent upon DTA expression while Trpv1+ cells remain intact. C,D) Mrgprb4^Cre; Rosa^DTA mice exhibit no motor deficits in open field (C, unpaired t-test) or rotarod (D, Two-Way ANOVA) assays. E-H) Mrgprb4^Cre; Rosa^DTA mice exhibit no differences in pain sensitivities to innocuous or noxious stimuli applied to the hairy skin compared to controls. Note: In panel G B4DTA animals have an increased distance traveled to the innocuous air puff suggesting that some subtle aspect of somatosensation might be altered in these animals. However, since they display no pain behaviors to the air puff, and normal pain behaviors to pinprick, we can conclude their pain threshold is normal. In H, the composite nocifensive score captures four affective pain behaviors including jumping, paw guarding, paw shaking, and eye grimace.

5

Supplemental Figure 5. Optogenetic posture is independent of estrous state and neuroanatomy of Mrgprb4+ sensory endings, Related to Figure 4. A) No differences in the depth of the optogenetically induced back dip across stages of the estrous cycle. Circles represent individual mice, data plotted as mean +/− SEM. B) Neither ovariectomizing (OVX) females, nor replacing estrogen and progesterone exogenously to mimic behavioral estrus, significantly modulates the optogenetically induced back dip. Dots represent individual mice. All data analyzed by one way ANOVA followed by Tukey’s post-hoc comparisons. C) Nerve terminal endings in red are seen in the underbelly skin. D) Nerve terminals for Mrgprb4+ touch neurons are not seen in the internal lining of the vaginal wall, which is non-hairy skin.

6

Supplemental Figure 6: Viral tracing of putative skin-to-brain circuit from Mrgprb4-lineage neurons to VTA, Related to Figure 5. A) Schematic depicting experimental approach. AAV-LSL-Tomato and CRE-DOG AAVs (AAV-EF1a-C-CreintG and AAV-EF1a-N-CretrcintG) were injected into spinal cord of GPR83-eGFP mice in order to express Cre in and label GPR83+ projection neurons with Cre-dependent tdTomato. AAV retro-BFP was injected into the VTA of the same mice in order to label cells projecting to the VTA with BFP. B) In blue: The retro-BFP+ endings in VTA. C) In blue: the retro-BFP+ somas of neurons that project to the VTA. In red: the GPR83+ projections from lower thoracic spinal cord. C’) Magnified view of C: The red GPR83+ projection neurons make contact with blue somas that project to the VTA.

7

Supplemental Video 1: Optogenetic back stimulation of Mrgprb4-lineage touch neurons in Mrgprb4-Cre; ChR2 female mice, Related to Figure 3. Pulsing blue laser light at 10Hz to the shaven back skin results in back lowering phenotype that is reminiscent of the lordotic copulatory posture.

8

Supplemental Video 2: Optogenetic back stimulation of Cre-negative female control littermates, Related to Figure 3. Pulsing blue laser light at 10Hz to the shaven back skin of control mice does not result in back lowering phenotype. Note the animal is alerted to the presence of the experimenter but does not alter its behavior in any noticeable robust manner.

9

Supplemental Video 3: Optogenetic anogenital stimulation of Mrgprb4-lineage touch neurons in Mrgprb4-Cre; ChR2 female mice, Related to Figure 5. Pulsing blue laser light at 10Hz to the anogenital skin does not trigger any noticeable robust behaviors, or lordosis-like postures.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-CGRP	Immunostar	Cat#24112
Griffonia anti-IB4	Life Tech	Cat#I21411
Chicken anti-GFP	Life Tech	Cat#A10262
Rabbit anti-RFP	Rockland	Cat#600401379
Rabbit anti-HA-Tag	Cell Signaling	Cat#3724
Bacterial and virus strains
pAAV.CAG.LSL.tdTomato	Addgene	Plasmid#100048
AAV-EF1a-N-CretrcintG	Addgene	Plasmid#69571
AAV-EF1a-C-CreintG	Addgene	Plasmid#69570
AAV1-syn-FLEX-jGCaMP8m	Addgene	Plasmid#162378-AAV1
pENN.AAV.CB7.Cl.mCerulean.WPRE.RBG	Addgene	Plasmid#105557-AAV9
pAAV-hSyn-GRAB_DA2m	Addgene	Plasmid#140553
pAAV-hSyn-GRAB_DA1h	Addgene	Plasmid#113050
pAAV-Ef1a-DIO-ChRmine-mScarlet-3xHA (PHP.S)	Dr. David Ng	N/A
Chemicals, peptides, and recombinant proteins
Bicuculline	Alomone	Cat #: B-137
Strychnine	Sigma	Cat #: S8753-25G
CNQX	Alomone	Cat #: C-141
TTX	Caymen Chemical	Cat #: 14964
Estradiol benzoate	Sigma	Cat#E1600000
Progesterone	Sigma	Cat#p0130-100g
Tamoxifen	Sigma	Cat#T5648-1G
Critical commercial assays
RNAScope Probe: eYFP	ACD	Cat#312131
RNAScope Probe: Mrgprb4	ACD	Cat#435781
RNAScope Probe: Mrgprd	ACD	Cat#417921
RNAScope Probe: Mrgprc11	ACD	Cat#488771
RNAScope Probe: Mrgpra3	ACD	Cat#548161
RNAScope Probe: Pou4f2	ACD	Cat#558481
RNAScope Probe: Trpa1	ACD	Cat#400211
RNAScope Probe: Ret	ACD	Cat#431791
RNAScope Probe: Th	ACD	Cat#317621
RNAScope Probe: Scn10a	ACD	Cat#426011
RNAScope Probe: Calca	ACD	Cat#578771
RNAScope Probe: Gfra1	ACD	Cat#431781
RNAScope Probe: Trpv1	ACD	Cat#313331
RNAScope Intro Pack for Multiplex Fluorescence	ACD	Cat#323136
RNAScope Multiplex Fluorescent Reagent Kit v2	ACD	Cat#323100
Opal 520 Reagent Pack	Perkin Elmer	FP1487001KT
Opal 570 Reagent Pack	Perkin Elmer	FP1488001KT
Opal 690 Reagent Pack	Perkin Elmer	FP1497001KT
Experimental models: Organisms/strains
Mouse: Mrgprb4tdTomato-Cre	Jackson Laboratories	Strain #: 021077
Mouse: Rosa26Ai32-eYFP	Jackson Laboratories	Strain #: 012569
Mouse: Tg(Gpr83-EGFP)DV46Gsat/Mmucd	MMRRC, UC Davis	Stock #: 010442-UCD
Mouse: Rosa26DTA	Jackson Laboratories	Strain #: 009669
Mouse: DATIREScre	Jackson Laboratories	Strain #: 006660
Mouse: MrgprdCreERT2	Jackson Laboratories	Strain #: 031286
Mouse: Mrgpra3Cre-EGFP	Dr. Xinzong Dong	N/A
Software and algorithms
Synapse software	TDT	N/A
Clampex software	Molecular Devices	N/A
PAWS software	Jones et al.37	N/A
ProAnalyst	Xcitex	N/A
pMAT software	Bruno et al.71	https://github.com/djamesbarker/pMAT
CNMF-E	Zhou et al.70	N/A
Inscopix Data Processing Software	Inscopix	Version 1.6.0
Inscopix Data Acquisition Software	Inscopix	Version 1.3.0
Other
Zeiss Axiocam 506 color camera	Zeiss	N/A
MultiClamp 700B amplifier	Molecular Devices	N/A
Axon Digidata 1550B	Molecular Devices	N/A
Siskiyou MX7600 Motorized Micromanipulator	Siskiyou	MX7600
Conditioned Place Preference Chamber	Rossi et al.68	N/A
FastCAM UX100 high speed camera	Photron	800K-M-4GB
473 nm blue laser light	Shanghai Laser and Optics, China	BL473T8-150FC/ADR-800A
532 nm green laser	Shanghai Laser and Optics Century, Shanghai, China	GL532T8-1000FC/ADR-800A
593 nm continuous wave diode pumped solid-state orange laser	CrystalLaser LC	CL593-050-O
10MHZ Function Waveform Generator	Agilent	33210A
FC/PC Optogenetic patch cable	ThorLabs	M72L01
TDT Fiber Photometry RZ10X Real Time Processor	TDT	RZ10X
TDT Experiment Control Workstation	TDT	WS-4
Mono Fiber-optic Cannula	Doric	MFC_200/230-0.57_6mm_MF1.25_FLT
Super glue	Loctite	LOC1255800
Fiber Photometry Fluorescence Mini Cubes – 6 ports	Doric	FMC6_IE(400-410)_E1(460-490)_F1(500-540)_E2(555-570)_F2(580-680)_S
Zirconia mating sleeves	Doric	SLEEVE_ZR_1.25_BK
Mono Fiber-optic Patch Cords	Doric	MFP_200/230/900-0.57_1m_FCM-MF1.25(F)_LAF
Metabond Quick Adhesive Cement System	Parkell	SKU: S380
Ortho-Jet Liquid Dental cement	Lang Dental	N/A
GRIN lens 7.3mm length, NA 0.5	Inscopix	1050-004413
Laser power meter	ThorLabs	s120c
Syringe pump	Harvard Apparatus	PHD Ultra

Highlights.

• Activation of Mrgprb4-lineage touch neurons induces lordosis-like posture

• Activation of Mrgprb4-lineage touch neurons is rewarding

• Mrgprb4-lineage touch neurons are required for female sexual receptivity

• Mrgprb4-lineage touch neurons engage dopaminergic neurons during social behavior