Neurexin regulates mechanical nociceptive sensitization by central inhibition in Drosophila

Abstract

Synaptic adhesion molecules play a crucial role in shaping neural circuits and behavior, yet their contribution to nociceptive processing remains poorly understood. Here, we investigate how the synaptic organizer Neurexin regulates mechanical nociception in Drosophila larvae. Using genetic, behavioral, and circuit-level analyses, we demonstrate that loss of neurexin induces mechanical nociceptive sensitization, which depends on a specific cluster of central cholinergic neurons in the subesophageal zone. These neurons drive sensitization through a C4da-Goro circuit, where Neurexin modulates presynaptic GABA_B receptor signaling to control nociceptive excitability. Our findings establish Neurexin as a key of nociceptive sensitization and uncover a synaptic mechanism for tuning nociceptive thresholds, with implications for understanding chronic pain and sensory circuit dysfunctions in neurodevelopmental disorders.

Loss of a key synaptic protein triggers nociceptive hypersensitivity by disrupting inhibitory signaling in neural circuits.

INTRODUCTION

Chronic pain is a debilitating condition affecting millions worldwide, characterized by persistent nociceptive sensitization and maladaptive plasticity in neural circuits (1, 2). Unlike acute or short-term pain, which serves as a protective response, chronic pain arises from dysregulated signaling in peripheral and central nociceptive pathways, often resistant to conventional treatments (3–7). While substantial progress has been made in understanding inflammatory and neuropathic pain mechanisms, the synaptic mechanisms governing this plasticity remain incompletely understood (8–11). Elucidating these mechanisms is critical for developing targeted therapies to restore normal pain processing.

Neurexins (NRXs) are presynaptic cell adhesion molecules best known for their roles in synapse formation, neurotransmission (12–15), and neurodevelopmental disorders such as autism spectrum disorder and schizophrenia (16–20). Sensory processing abnormalities, including altered pain sensitivity, are frequently observed in these conditions, suggesting a potential link between NRX dysfunction and nociceptive circuit regulation (21–23). However, direct evidence for NRXs in pain modulation is lacking, and their precise function in sensory systems, especially in nociceptive sensitization, has yet to be explored. In Drosophila melanogaster, NRX is encoded by a single gene and produces one major protein isoform, in contrast to its vertebrate α-NRX homologs, which generate numerous isoforms through alternative splicing (24). In neurons, loss of nrx results in structural abnormalities at both pre- and postsynaptic sites, altered bouton numbers, impaired synaptic transmission (25, 26), as well as defects in axonal arbor morphology and targeting (27, 28). Given their central role in synaptic organization and neurotransmitter receptor trafficking, NRXs may serve as key regulators of nociceptive plasticity, bridging the gap between synaptic adhesion and pain signaling.

The Drosophila larval nociception system has emerged as a powerful model for studying conserved mechanisms of nociceptive sensitization (29–33). This system combines exceptional genetic tractability with quantifiable behavioral readouts, enabling precise dissection of molecular and cellular pathways underlying mechanical and thermal hypersensitivity (34–36). The core nociceptive circuit begins with class IV dendritic arborization (C4da) neurons, which serve as polymodal nociceptors covering the entire larval body wall (37). These sensory neurons relay information through a sophisticated central network of interneurons (including A08n, mCSI, DnB, Basin, Wave, DP-ilp7, Pr1, A00c, and Goro) that process and amplify nociceptive signals before converging onto motor neurons (such as SNa) to generate the characteristic corkscrew rolling escape response (38–46). Recent work has revealed additional layers of complexity in this system, identifying both excitatory (e.g., chordotonal neurons) and inhibitory components (including descending DSK, TRH, and SDG neurons) that modulate nociceptive processing (41, 43, 47, 48). Furthermore, specialized circuits (like SeIN128 neurons) mediate behavioral transitions between rolling and crawling responses (49). This comprehensive nociceptive architecture shares fundamental organizational principles with vertebrate pain pathways, from peripheral detection to central integration and motor output (50).

In this study, we uncover a critical role for Drosophila NRX in mechanical nociceptive sensitization. We demonstrate that nrx loss induces hypersensitivity dependent on central cholinergic neurons and identify a specific circuit mediating this effect. Further, we reveal that NRX regulates presynaptic GABA_B receptor trafficking to modulate nociceptive excitability, providing a mechanistic link between synaptic adhesion and nociceptive plasticity. Our findings establish NRX as a key regulator of nociceptive sensitization and provide a mechanistic framework for understanding how synaptic dysregulation contributes to chronic pain states.

RESULTS

Loss of nrx results in mechanical nociceptive sensitization

To investigate mechanical nociception in Drosophila larvae, we used a well-established assay using nitinol fibers of varying lengths (Fig. 1A and fig. S1A) (51). Larvae exhibited rolling behavior in a force-dependent manner (fig. S1, A and B). Unexpectedly, larvae carrying loss-of-function mutations in nrx displayed heightened sensitivity to various mechanical stimuli compared to wild-type controls (Fig. 1B and fig. S1B). The identities of both mutant constructs were confirmed by quantitative polymerase chain reaction (qPCR; Fig. 1B). These results suggest that NRX plays important roles in mechanical nociceptive sensitization. Given that 3260-kPa stimulus resulted in the greatest proportional differences between the nrx mutant and wild-type groups, this stimulus intensity was applied in the remainder of the study unless otherwise indicated.

Fig. 1. Loss of nrx results in mechanical nociceptive sensitization.

(A) Nociceptive corkscrew rolling response of larvae to mechanical stimulation. (B) Left: quantitative analysis of nrx mRNA expression levels in control and nrx mutant larvae (n = 4, 30 animals per replicate). Right: quantification of nociceptive behavior in control and nrx mutant larvae (n ≥ 6, about 30 animals per replicate). (C) Nrx-GAL4–driven NLS-GFP (green) colocalizes with neuronal marker elav (magenta) in the larval CNS. Scale bar, 50 μm. (D) No colocalization of nrx-GAL4>NLS-GFP (green) with glial marker repo (magenta) in the CNS. Scale bar, 50 μm. (E and F) Confocal image stacks showing GFP (green) driven by nrx-GAL4 and colocalized with the neuronal marker horseradish peroxidase (HRP; magenta) in the larval peripheral nervous system (PNS). Scale bars, 50 μm. (G) Strategy for conditional labeling endogenous NRX. The conditional endogenous NRX tagging strategy: An FRT-flanked STOP cassette inserted at the nrx stop codon prevents NRX-6×HA expression until Flp-mediated excision in target cells. UTR, untranslated region. (H to K) Conditional NRX-6×HA expression in the larval CNS. Immunofluorescence is detected only after excision of the neuronal STOP cassette. FSF, FRT-STOP-FRT. Scale bars, 50 μm. (L and M) Endogenous NRX expression pattern in larval PNS. Scale bars, 50 μm. Data: means ± SEMs; ****P < 0.0001; unpaired t test for (B).

Since NRX is a critical regulator of synapse properties, we investigated its expression in the larval nervous system. Costaining with green fluorescent protein (GFP) and the pan-neuronal marker elav or the pan-glial marker repo in nrx-GAL4 (enhancer trap)>UAS-NLS-GFP larvae revealed that NRX was exclusively expressed in neurons, not glial cells (Fig. 1, C and D). Furthermore, NRX expression was also observed in both peripheral sensory and motor neurons (Fig. 1, E and F). To further confirm NRX expression and conditional labeling, we generated knock-in lines where hemagglutinin (HA) tags, with or without an FRT-STOP-FRT cassette, were inserted immediately before the stop codon of the nrx locus (Fig. 1G and fig. S1C). The expression pattern of nrx-HA knock-in lines was identical to that of the nrx-GAL4 line (Fig. 1, H to M, and fig. S1D). These data indicate that NRX is widely expressed in various neurons throughout both the central and peripheral nervous systems.

Central cholinergic neurons are required for mechanical nociceptive sensitization in nrx mutant larvae

To confirm the neuronal requirement responsible for mechanical nociceptive sensitization mediated by NRX, we used a series of cell-specific GAL4 lines. Knockdown of nrx via nrx-GAL4–driven UAS-nrx-RNAi recapitulated the elevated rolling probability observed in nrx mutants (Fig. 2A). To confirm neuronal involvement and RNA interference (RNAi) specificity, we used three independent nrx-RNAi lines under the pan-neuronal elav-GAL4 to suppress nrx expression. Neuronal nrx knockdown phenocopied the hypersensitivity observed in mutants, while suppression of nrx-GAL4 activity by elav-GAL80 abolished this effect (Fig. 2, A and B). Critically, neuronal reintroduction of wild-type NRX completely reversed the mutant phenotype (Fig. 2C), providing genetic evidence that NRX is the key determinant of mechanical nociceptive sensitization.

Fig. 2. Central cholinergic neurons are required for mechanical nociceptive sensitization in nrx mutant larvae.

(A) Nrx knockdown driven by nrx-GAL4 with elav-GAL80 resulted in significantly less rolling behavior upon mechanical stimulation than nrx knockdown driven by nrx-GAL4 alone (n ≥ 5, about 30 animals per replicate). (B and C) Nrx knockdown in neurons increased rolling probability upon mechanical stimulation (B); neuronal NRX reexpression rescued this phenotype in nrx mutants (C). n ≥ 5, about 30 animals per replicate. (D) Nrx knockdown in pan-sensory neurons (labeled by 21-7-GAL4) did not alter rolling probability (n ≥ 4, about 30 animals per replicate). (E) Rolling probability upon mechanical stimulation of larvae expressing RNAi constructs for nrx in different types of neurons (n ≥ 4, about 30 animals per replicate). (F and G) Nrx knockdown in cholinergic neurons increased rolling probability (F); cholinergic-specific NRX reexpression rescued this phenotype in nrx mutants (G). n ≥ 4, about 30 animals per replicate. (H) Nrx knockdown in nociceptive pathway neurons did not alter rolling probability. Data: means ± SEMs; ns, nonsignificant; ****P < 0.0001; one-way analysis of variance (ANOVA) with Bonferroni posttests for (A) to (H). n ≥ 3, about 30 animals per replicate.

Having established that loss of NRX in neurons induces nociceptive sensitization, we sought to identify the functional subpopulations involved. Although NRX is expressed in peripheral sensory neurons (Fig. 1E), targeted knockdown of nrx using 21-7-GAL4 (pan-sensory) or ppk-GAL4 (C4da neurons) did not result in any behavioral phenotype (Fig. 2D and fig. S2A). In addition, restoration of NRX in C4da nociceptors failed to rescue mechanical hypersensitivity in nrx^−/− larvae (fig. S2A), thereby excluding a regulatory role for NRX in peripheral sensory neurons. However, mechanical nociception in nrx^−/− larvae was abolished by C4da-specific expression of tetanus toxin (TNT), which blocks synaptic transmission (fig. S2B), suggesting that nociceptive sensitization in nrx mutants requires intact sensory input from nociceptors.

Several nociceptive neuronal circuits, such as Basin-Goro, A08n, and DnB, have previously been identified (38–46, 48, 49). We hypothesized that NRX might regulate these circuits. Unexpectedly, knockdown of nrx in the neurons of these circuits did not result in any behavioral changes (Fig. 2H). This suggests the involvement of other, yet-to-be-identified interneurons in NRX-mediated mechanical nociceptive sensitization.

Next, we systematically screened neural populations based on their neurotransmitter identity. Knockdown of nrx in glutamatergic (OK371-GAL4), GABAergic (VGAT-GAL4), dopaminergic (TH-GAL4), octopaminergic (TDC2-GAL4), or serotonergic (TRH-GAL4) neurons did not elicit any behavioral changes (Fig. 2E). Notably, motor neurons at the peripheral neuromuscular junctions are also glutamatergic. In contrast, cholinergic neuron-specific knockdown (ChAT-GAL4) robustly mimicked the hypersensitivity observed in nrx mutants (Fig. 2F), which was fully rescued by restoration of NRX in the same population (Fig. 2G). Collectively, these results strongly suggest that central cholinergic neurons are critical mediators of NRX-dependent mechano-nociceptive regulation.

12E09Cho neurons regulate nrx-dependent mechanical nociceptive sensitization

We conducted a visual screen of 6849 Janelia GAL4 lines and selected around 300 GAL4 drivers with restricted expression patterns for behavioral testing. This effort identified 12E09-GAL4 as a promising candidate driver (Fig. 3A). Knockdown of nrx in 12E09 neurons significantly increased responsiveness to noxious mechanical stimuli (Fig. 3B).

Fig. 3. 12E09Cho neurons regulate nrx-dependent mechanical nociceptive sensitization.

(A) Expression pattern of 12E09-GAL4 driving UAS-mCD8-GFP (green). The green channel shows anti-GFP staining. Scale bar, 50 μm. (B) Nrx knockdown driven by 12E09-GAL4 increases rolling probability (n ≥ 7, about 30 animals per replicate). (C) Split-GAL4 strategy. Coexpression of AD and DBD reconstitutes GAL4 to initiate transcription. (D to F) Expression patterns of cholinergic (D), glutamatergic (E), and GABAergic (F) neurons in 12E09 neurons. Scale bars, 50 μm. (G) Nrx knockdown in the neurons shown in (D) increases rolling probability (n ≥ 9, about 30 animals per replicate). (H) NRX reexpression rescues the increased rolling probability induced by nrx knockdown in 12E09Cho neurons (n ≥ 7, about 30 animals per replicate). (I) Nrx knockdown in the neurons shown in (E) does not alter rolling probability compared with controls (n ≥ 8, about 30 animals per replicate). (J) Nrx knockdown in the neurons shown in (F) does not alter rolling probability compared with controls (n ≥ 8, about 30 animals per replicate). Data: means ± SEMs; **P < 0.01 and ****P < 0.0001; one-way ANOVA with Bonferroni posttests for (B) and (G) to (J).

To identify the neuron specificities in the 12E09 population, we performed neurotransmitter profiling using conditional synaptic vesicle markers—VGAT (52), VAChT (53), and VGlut (54)—to evaluate GABAergic, cholinergic, and glutamatergic subtypes (fig. S3, A to C). This profiling revealed that the 12E09 population is heterogeneous, comprising GABAergic, glutamatergic, and cholinergic neurons. Using split-GAL4 intersectional strategies (Fig. 3, C to F), we selectively targeted each subtype. The intersection of 12E09-DBD and VAChT-AD (designated as 12E09Cho-spGAL4) labeled approximately 10 neurons per hemibrain. This included a pair in the subesophageal zone (SEZ) and a pair in the ventral nerve cord (VNC), with an additional five to six cells occasionally observed in the VNC (Fig. 3D). The 12E09-DBD and VGlut-AD intersection (designated as 12E09Glu-spGAL4) strongly labeled a pair of neurons in the VNC and a pair in the brain lobe. Typically, however, only one of the brain lobe neurons was detected, with both rarely visible at the same time. In addition, ~30 cells showed weaker labeling (Fig. 3E). The 12E09-DBD and VGAT-AD intersection (designated as 12E09GABA-spGAL4) strongly labeled a pair of neurons in the SEZ, although usually only one of the pair was observed. About 10 additional cells displayed weaker labeling (Fig. 3F). To investigate the role of these neurons in nociceptive behavior, we used the split-GAL4 lines to knock down nrx (Fig. 3, G, I, and J). Notably, nrx knockdown in 12E09Cho neurons resulted in robust hypersensitivity (Fig. 3G). Furthermore, expressing an RNAi-resistant nrx transgene (nrx-3×HA) fully rescued the hypersensitivity phenotype (Fig. 3H and fig. S4), demonstrating that NRX in 12E09Cho neurons is essential for mechanical nociceptive sensitization. Knockdown of nrx in glutamatergic or GABAergic neurons did not alter mechanical nociception (Fig. 3, I and J). This suggests that NRX function in 12E09Glu and 12E09GABA neurons is dispensable for mechanical hypersensitivity.

Identification of neurons sufficient for nociceptive sensitization induced by noxious stimuli

To determine whether 12E09Cho neurons are sufficient to induce nociceptive sensitization, we increased their excitability by expressing either NaChBac, a voltage-gated sodium channel (55), or TrpA1 (56), a heat-gated cation channel. Both manipulations significantly enhanced mechanical nociception (Fig. 4A and fig. S5, A and B). In contrast, silencing 12E09Cho neuronal activity using TNT had no effect on baseline mechanical nociception (Fig. 4A). However, TNT inhibited mechanical sensitization in larvae with nrx knocked down specifically in 12E09Cho neurons (Fig. 4, B and C). These results suggest that 12E09Cho neurons are dispensable for basal nociception but essential for NRX-dependent nociceptive sensitization.

Fig. 4. 12E09Cho SEZ neurons are sufficient for sensitization in NRX loss of function.

(A) Effects of chronic activation or silencing of 12E09Cho neurons on mechano-nociceptive rolling. (B) Expression pattern corresponding to (C). Diagram of CNS was created with BioRender.com [Created in BioRender. Meng, Z. (2026) https://BioRender.com/q9on6ci]. (C) TNT expression rescues enhanced rolling induced by nrx knockdown in 12E09Cho neurons. (D) Schematic of larval CNS. (E) Confocal projection showing all VNC neurons (green) in larvae. (F) Nrx knockdown driven by tsh-GAL4 does not alter rolling. (G) 12E09-GAL4>UAS-mCD8-GFP expression in a tsh-GAL80 background (green). (H) Nrx knockdown driven by 12E09-GAL4 in a tsh-GAL80 background increases rolling. (I) Tsh-GAL80 suppresses ChAT-GAL4>UAS-mCD8-GFP in VNC neurons. (J) Nrx knockdown driven by ChAT-GAL4 in a tsh-GAL80 background increases rolling. (K and L) Expression pattern and behavioral responses of 12E09Cho neurons with nrx knockdown under brain lobe–restricted TNT expression. Diagram of CNS was created with BioRender.com [Created in BioRender. Meng, Z. (2026) https://BioRender.com/q9on6ci]. (M) Expression of 12E09Cho-GAL4>UAS-FSF-myrGFP with Otd-Flp. (N to P) Expression pattern and effects of chronic or acute (heat plate) activation of 12E09Cho brain neurons. Diagram of CNS was created with BioRender.com [Created in BioRender. Meng, Z. (2026) https://BioRender.com/q9on6ci]. (Q) Expression of 12E09Cho-GAL4>UAS-FRT-NaChBac-EGFP-FRT with Otd-Flp. Arrowheads indicate 12E09Cho-SEZ neurons. (R) Expression pattern of the larvae used in (S) and (T). Diagram of CNS was created with BioRender.com [Created in BioRender. Meng, Z. (2026) https://BioRender.com/q9on6ci]. (S and T) Chronic or acute (heat plate) activation of 12E09Cho-SEZ neurons increases rolling probability. (U) Depth-coded confocal projection of a sparsely labeled TENCS neuron (SPARC; dorsal, blue; ventral, white). Scale bars, 50 μm (all images). Data: n ≥ 5, about 30 animals per replicate; means ± SEMs; **P < 0.01 and ****P < 0.0001; one-way ANOVA with Bonferroni posttests for (A), (C), (F), (H), (J), (L), (O), (P), (S), and (T).

To further identify the neuronal subpopulations mediating NRX-dependent mechanical nociceptive sensitization, we used region-specific GAL4 drivers to suppress NRX expression in the central nervous system (CNS). Knockdown of nrx in the VNC using tsh-GAL4 (tsh-GAL4>UAS-nrx-RNAi) did not affect nociceptive behavior (Fig. 4, D to F). In contrast, restricting nrx knockdown to brain lobe or SEZ cholinergic neurons—by combining 12E09-GAL4 or ChAT-GAL4 with the VNC-suppressing repressor tsh-GAL80—recapitulated the hypersensitivity phenotype observed in mutants (Fig. 4, G to J). This spatial dichotomy indicates that NRX functions in cholinergic neurons of the brain lobe or SEZ, but not in the VNC.

To more precisely localize the responsible neurons, we used the Otd-Flp/FRT system to divide 12E09Cho neurons into two anatomical subsets: one in the brain lobes and another in the SEZ/VNC (Fig. 4, D and K). Inhibiting brain lobe 12E09Cho neurons expressing nrx RNAi did not affect mechanical sensitization (Fig. 4, K and L). Moreover, increasing the excitability of these neurons also failed to induce sensitization to noxious stimuli (Fig. 4, M to P). In contrast, elevating the excitability of 12E09Cho neurons in the SEZ and VNC region successfully induced nociceptive sensitization (Fig. 4, Q to T). In addition, inhibiting 12E09 neurons in the SEZ and VNC under nrx knockdown conditions reversed the mechanical sensitization induced by nrx deficiency (fig. S5C). Given previous evidence that VNC neurons are not involved in NRX loss-of-function–induced sensitization (Fig. 4, E and F), we conclude that a specific pair of 12E09Cho neurons located in the SEZ—henceforth referred to as TENCS (twelve E nine cholinergic neurons in the SEZ)—is sufficient for nociceptive sensitization (indicated by the arrowheads in Fig. 4Q). For clarity and to emphasize the contribution of these neurons, we will refer to the “12E09Cho-spGAL4” line (comprising 12E09-DBD and VAChT-AD) as “TENCS-GAL4” throughout the remainder of this manuscript.

Costaining TENCS neurons with ChAT confirmed their cholinergic identity (fig. S5E). The expression pattern of nrx-HA knock-in lines confirms that nrx is expressed in TENCS neurons (fig. S5F). Moreover, acetylcholine is required for mechanical sensitization induced by nrx knockdown in TENCS (fig. S5D). Using nuclear labeling with nuclear localization signal (NLS)–GFP, we localized the TENCS somata to the ventrolateral maxillary segment of the mandibular neuromeres (fig. S5, G and H). To further examine neuronal morphology, we labeled dendrites with DenMark and axons with Syt-GFP (fig. S5I). In addition, using sparse predictive activity through recombinase competition (SPARC)-based stochastic sparse labeling (57), we observed that TENCS dendrites are ipsilateral to the soma and project dorsally from the main neurite, while axons extend contralaterally (Fig. 4U and movie S1).

TENCS neurons facilitate nociceptive sensitization via the C4da-Goro circuit

To investigate the functional role of TENCS neurons in the nociceptive circuit, we activated TENCS neurons by expressing CsChrimson, a red-shifted channelrhodopsin. Activation of TENCS alone did not induce rolling behavior (Fig. 5A). However, coactivation of TENCS and C4da neurons significantly increased the rolling response (Fig. 5A). Similar results were obtained using the thermogenetic tool TrpA1 (Fig. 5B). TNT-mediated silencing reduced rolling duration without affecting probability and rolling start-up latency induced by activation of C4da neurons (fig. S6, A and B). In combination with the TNT-mediated inhibition results (Fig. 4A), these findings suggest that TENCS activation enhances nociceptive signaling, whereas TENCS inhibition does not impair nociceptive signal transduction. Although no direct synaptic connections were detected between TENCS neurons and canonical nociceptive pathway neurons (fig. S7, A to C), we next examined their relationship with the C4da-Goro circuit. The spatial positions and projections of C4da, TENCS, and Goro neurons within a single standardized brain are shown in fig. S8 and movie S2. Chronic activation of TENCS using NaChBac, together with Goro neuron inhibition via TNT, reversed the mechanical hypersensitivity induced by TENCS activation (Fig. 5C), indicating that TENCS induces mechanical hypersensitivity through Goro neurons.

Fig. 5. NRX regulates the excitability of TENCS neurons.

(A) TENCS activation promotes rolling behavior induced by optogenetic C4da stimulation. (B) TENCS activation promotes rolling under heat plate–induced thermal C4da stimulation via overexpressing TrpA-A isoform in C4da neurons (ppk-GAL4>UAS-TrpA1-A, n ≥ 3, about 30 animals per replicate). (C) Rolling probability in Goro-silenced larvae with TENCS chronic activation (NaChBac). n ≥ 7, about 30 animals per replicate. (D) C4da silencing (TNT) reduces GCaMP responses in Goro following 300 μM AITC stimulation. n ≥ 11. (E) Dual activation (C4da: 300 μM AITC; TENCS: 7 mM ATP) enhances GCaMP responses in Goro. n ≥ 13. (F) Nrx RNAi in TENCS increases GCaMP signals in Goro during C4da activation (300 μM AITC). n ≥ 18. (G and H) Calcium responses in Goro under: TENCS inhibition (G) or 7 mM ATP stimulation (H). n ≥ 12. (I) Calcium dynamics in TENCS neurons during C4da stimulation with 7 mM ATP. Scale bars, 10 μm. n ≥ 7. (J) Nrx knockdown enhances calcium signals in TENCS during optogenetic C4da activation. n ≥ 8. (K) Decreased ArcLight baseline fluorescence in nrx-knockdown TENCS neurites. Scale bar, 20 μm. n ≥ 11. (L) Nrx knockdown in TENCS neurons alters Arclight (−ΔF/F₀) responses during C4da activation (300 μM AITC). n ≥ 20. (M) Proposed circuit model of NRX-mediated nociceptive regulation. Behavioral data presented as means ± SEMs; in vivo imaging data shown as boxplots with overlaid individual points; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; log-rank (Mantel-Cox) test for (A); one-way ANOVA with Bonferroni posttests for (B), (C), and (J) and unpaired t test for (D) to (I), (K), and (L). All individual data points are overlaid on the boxplot.

Using live Ca²⁺ imaging, we found that Goro neurons were activated by peripheral stimulation of C4da neurons with allyl isothiocyanate (AITC), a TrpA1 agonist. Although TrpA1 expression is not specific to C4da neurons, TNT-mediated inhibition of C4da significantly reduced Ca²⁺ responses in Goro neurons (Fig. 5D and fig. S9A), suggesting that AITC primarily stimulates peripheral C4da neurons. We then assessed Goro neuron activity under dual activation of C4da (via AITC-TrpA1) and TENCS [via adenosine triphosphate (ATP)–P2X₂]. Goro calcium responses were significantly enhanced under this dual stimulation compared to C4da stimulation alone (Fig. 5E and fig. S9B). In addition, we observed increased Goro Ca²⁺ responses when C4da was stimulated and nrx was knocked down in TENCS neurons, compared to C4da stimulation alone (Fig. 5F). These data demonstrate that loss of nrx function in TENCS increases their excitability and facilitates Goro activation. Conversely, when TENCS neurons were inhibited via TNT during C4da stimulation with AITC, Goro responses were not significantly altered compared to C4da stimulation alone (Fig. 5G). Moreover, activation of TENCS alone was insufficient to elicit Goro Ca²⁺ responses (Fig. 5H), corroborating the findings from behavioral assays.

We further examined whether C4da activation could influence TENCS neurons. Stimulation of C4da using the ATP-P2X₂ system resulted in TENCS activation (Fig. 5I and fig. S9A). Conversely, activation of TENCS neurons failed to influence C4da activity (fig. S9C), suggesting that TENCS neurons function downstream of C4da. When nrx was knocked down in TENCS and C4da was stimulated using CsChrimson, TENCS neurons exhibited significantly increased Ca²⁺ responses compared to wild-type controls (Fig. 5J). Furthermore, we used ArcLight, a genetically encoded voltage indicator using a voltage-sensitive fluorescent protein (58, 59), to monitor membrane potential changes in TENCS neurons. In ArcLight imaging, membrane depolarization is reported as a decrease in fluorescence intensity (−ΔF/F₀), whereas hyperpolarization produces the opposite change. Using this approach, we found that the baseline membrane potential of TENCS neurons was reduced in nrx knockdown larvae (Fig. 5K), while their voltage responses following C4da stimulation were significantly enhanced (Fig. 5L). These results indicate that nrx knockdown induces hyperexcitability in TENCS neurons, thereby amplifying nociceptive signals transmitted from C4da to Goro (Fig. 5M).

NRX controls presynaptic GABA_B signaling to regulate nociceptive excitability

Nrx knockdown showed no significant effects on neuronal morphology, as evidenced by unaltered neurite length, arborization complexity, or synaptic brp levels (fig. S10, A to C). Given the presynaptic localization of NRX and the observed increase in TENCS neuron excitability following nrx knockdown, we therefore hypothesized that this hyperexcitability phenotype may result from presynaptic disinhibition. Presynaptic GABA_B receptors (60–63) and muscarinic acetylcholine receptors (mAChRs), which act as cholinergic autoreceptors, are known to mediate presynaptic inhibition and thereby regulate neuronal excitability and acetylcholine (ACh) release (64, 65). To assess the functional relevance of these receptors in larval escape behavior, we individually knocked down each receptor type in TENCS neurons and evaluated the effect on mechanically evoked nociceptive rolling. Behavioral screening showed that RNAi-mediated knockdown of GABA_B-R1 or GABA_B-R2, but not GABA_B-R3 or mAChRs, mimicked the nrx knockdown phenotype, significantly increasing rolling probability (Fig. 6A and fig. S11A). Furthermore, simultaneous knockdown of GABA_B-R1 and GABA_B-R2 in TENCS neurons produced a more pronounced nociceptive sensitization phenotype than either single knockdown alone, indicating a genetic interaction between the two subunits (fig. S11, B and C). In line with evidence that GABA_B-R1 and GABA_B-R2 form heterodimers coupled to Gαi/Gαo proteins (66), targeted expression of pertussis toxin (PTX) in TENCS neurons—which inhibits Gαi/Gαo signaling via adenosine diphosphate ribosylation (61, 67)—also enhanced mechano-nociceptive behavior (Fig. 6B). Given the phenotypic similarity between nrx knockdown and GABA_B-R1/-R2 depletion, we next examined their genetic interaction. Removal of a single nrx copy, combined with mild GABA_B-R1 or GABA_B-R2 knockdown in TENCS neurons, further increased rolling behavior compared to individual perturbations (Fig. 6, D and E). Moreover, overexpression of GABA_B-R1 rescued the nociceptive phenotype induced by nrx knockdown (Fig. 6C), supporting a genetic interaction between NRX and GABA_B receptors.

Fig. 6. NRX negatively regulates mechanical nociception through GABA_B receptors in presynaptic TENCS.

(A and B) Knockdown of GABA_B-Rs (A) or PTX (B) in TENCS alters mechano-nociceptive rolling behavior (n ≥ 6, about 30 animals per replicate). (C) GABA_B-R1 expression rescues increased rolling probability in nrx-knockdown TENCS neurons (n ≥ 7, about 30 animals per replicate). (D) Rolling probability to mechanical stimulation in nrx heterozygotes, TENCS-GAL4>UAS-GABA_B-R1 RNAi, or both (n ≥ 7, about 30 animals per replicate). (E) Rolling probability to mechanical stimulation in nrx heterozygotes, TENCS-GAL4>UAS-GABA_B-R2 RNAi, or both (n ≥ 7, about 30 animals per replicate). (F) Colocalization of mRFP driven by TENCS-GAL4 (magenta) with endogenous GABA_B-R1 (green). Arrowheads: colocalized signals. Scale bars, 20 μm (top) and 5 μm (bottom). (G) Colocalization of Syt1 driven by TENCS-GAL4 (magenta) with endogenous GABA_B-R1 (green). Arrowheads: colocalized signals. Scale bar, 10 μm (top). (H) GABA_B-Rs knockdown enhances TENCS calcium responses to C4da ATP stimulation (n ≥ 12). (I) Schematic design of Ca²⁺ imaging in TENCS neurons (top). GABA_B receptor agonist SKF97541 was applied to the buffer 7 min before the stimulation and imaging. The effect of GABA_B receptor agonist application is shown at the bottom (n ≥ 8). (J) Reduced baseline ArcLight fluorescence in TENCS neurons with GABA_B-Rs knockdown (n ≥ 15). (K) Nrx knockdown impaired SKF97541-induced hyperpolarization. Left: ArcLight fluorescence traces (−ΔF/F₀). Right: quantified peak responses. n ≥ 18. Behavioral data presented as means ± SEMs; in vivo imaging data shown as boxplots with overlaid individual points; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; one-way ANOVA with Bonferroni posttests for (A) to (E) and (H) to (J) and unpaired t test for (K).

Using a GFP-tagged GABA_B-R1 allele (GABA_B-R1^MIGFP), we confirmed GABA_B-R1 expression in TENCS neurons, including axonal compartments (Fig. 6, F and G). Ca²⁺ imaging revealed that GABA_B receptor knockdown amplified C4da-evoked responses in TENCS neurons (Fig. 6H). To test whether GABA_B signaling was impaired by nrx knockdown, we expressed the ATP-gated cation channel P2X₂ in TENCS neurons and measured Ca²⁺ responses using calcium indicator (GCaMP) based on green fluorescent protein, calmodulin, and the M13 peptide. As expected, application of the GABA_B receptor agonist SKF97541 reduced ATP-evoked Ca²⁺ responses in control preparations (Fig. 6I). However, in nrx knockdown animals, SKF97541-mediated suppression was significantly attenuated, indicating impaired presynaptic GABA_B receptor function (Fig. 6I). In support of this, knockdown of GABA_B receptors led to decreased baseline fluorescence in TENCS neurons expressing the voltage sensor ArcLight, indicating depolarization (Fig. 6J). Nrx knockdown similarly disrupted SKF97541-induced hyperpolarization (Fig. 6K). Together, these findings suggest that NRX regulates TENCS excitability via presynaptic modulation of GABA_B receptors’ function.

NRX regulates GABA_B receptor trafficking to modulate nociceptive signaling

Given the presynaptic localization of both NRX and GABA_B receptors in TENCS neurons, we investigated whether these proteins physically interact. Stimulated emission depletion (STED) images of NRX and GABA_B-R1 coimmunostaining indicated their colocalization within TENCS neurites (Fig. 7A). Molecular docking analysis suggests a direct NRX-GABA_B receptor interaction (fig. S12). Coimmunoprecipitation assays further confirmed direct binding between NRX and GABA_B receptors when heterologously expressed in S2R⁺ cells (Fig. 7B). Because proper membrane localization is critical for GABA_B receptor function, we next examined whether NRX regulates GABA_B receptor trafficking to the cell surface. qPCR indicated that GABA_B receptor mRNA levels were unchanged in nrx knockdown larvae, suggesting a posttranscriptional mechanism (Fig. 7C). Although neuron-specific knockdown of nrx did not alter total GABA_B-R1 protein levels in the larval CNS (Fig. 7D), targeted nrx knockdown in TENCS neurons significantly reduced receptor surface localization, as evidenced by decreased colocalization of GABA_B-R1 with membrane-targeted red fluorescent protein (mRFP) (Fig. 7E). This reduction was further validated using a surface biotinylation assay, which confirmed increased GABA_B receptor surface expression following NRX coexpression (Fig. 7F). In contrast, heterologous coexpression of NRX with GABA_B receptors increased their plasma membrane localization compared to expression of GABA_B receptors alone (Fig. 7, G to J). These findings indicate that NRX physically interacts with GABA_B receptors and promotes their trafficking to presynaptic membranes. Collectively, these results support a model where nrx knockdown increases the intrinsic excitability of TENCS neurons by impairing GABA_B receptor surface trafficking, thereby disrupting presynaptic GABAergic inhibition and ultimately facilitating nociceptive output to Goro neurons (Fig. 7K).

Fig. 7. NRX physically binds to with GABA_B-R1 or GABA_B-R2 and regulates their plasma membrane localization.

(A) Left: STED images revealing colocalization of NRX-HA (magenta) with GABA_B-R1-FLAG (green) in TENCS neurons. Right: colocalization signal profile chart based on the white dashed line. Scale bar, 1 μm. (B) Co-IP in S2R⁺ cells: NRX interacts with GABA_B-R1 and GABA_B-R2. IP, immunoprecipitation. Co-IP, coimmunoprecipitation. (C) The mRNA expression levels of GABA_B-R1 and GABA_B-R2 in larvae expressing RNAi constructs for nrx in neurons. n = 4. (D) GABA_B-R1 protein expression in the CNS of neuron-specific nrx knockdown larvae. Protein expression was normalized to β-actin reference protein levels. n = 7. (E) Reduced GABA_B-R1^MIGFP/mRFP colocalization in nrx knockdown TENCS neurons. Middle: colocalization score of 60 confocal image stacks (1 μm) of larval CNS from dorsal to ventral. Rightmost: statistical chart of the colocalization score of the 40th confocal image stack. n = 14. Scale bar, 10 μm. (F) Workflow: surface protein labeling (Sulfo-NHS-SS-Biotin) and streptavidin pulldown for immunoblotting. RT, room temperature; SDS-PAGE, SDS–polyacrylamide gel electrophoresis. (G and H) Immunoblotting analysis of GABA_B-R1 and NRX protein expression in biotin-labeled plasma membranes. n = 8. (I and J) Immunoblotting analysis of GABA_B-R2 and NRX protein expression in biotin-labeled plasma membranes. n = 5. (K) Model: nrx knockdown in TENCS neurons down-regulates presynaptic GABA_B-R surface expression, reducing inhibition and enhancing excitability to promote nociceptive rolling through activation of downstream Goro neurons. Illustration shows exclusively the ladder-shaped axon terminals of C4da neurons in the VNC. Data: means ± SEMs; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; unpaired t test for (C) to (E), (H), and (J).

Exploration of NRX function in additional nociceptive modalities

While this study primarily focused on mechanical nociception, we also tested whether NRX contributes to maladaptive thermal responses. Nociceptive hypersensitivity manifests as two principal features: a lowered activation threshold (allodynia), wherein normally innocuous stimuli evoke pain, and exaggerated responses to noxious stimuli (hyperalgesia) (68). Using both heat probe and heat plate assays, nrx mutant larvae exhibited significantly increased heat-induced rolling (fig. S13, A, C, E, and G), indicative of allodynia and hyperalgesia. Similarly, pan-neuronal or TENCS-specific knockdown of nrx enhanced thermal allodynia and hyperalgesia (fig. S13, B, D, F, and H), demonstrating that NRX in TENCS neurons is required for both mechanical and thermal nociceptive sensitization.

We further examined whether NRX participates in inflammation-induced hypersensitivity. Ultraviolet (UV) light–induced epidermal damage resulted in robust mechanical hypersensitivity in larvae (fig. S14, A to C), which was rescued by pan-neuronal nrx overexpression. Notably, nrx overexpression in TENCS neurons failed to rescue, whereas expression in peripheral C4da nociceptors effectively suppressed the UV damage–induced phenotype (fig. S14, D and E). Although nrx knockdown in C4da neurons did not alter baseline mechanical nociception, C4da-specific nrx overexpression was sufficient to counteract inflammation-induced hypersensitivity. UV damage did not alter overall nrx mRNA or protein levels in the larval CNS (fig. S14, F and G), arguing against transcriptional or translational regulation of NRX in the CNS as the primary mechanism. While our analysis revealed no changes at the tissue level, it is possible that alterations are confined to specific cell subtypes. In summary, these findings extend the role of NRX beyond mechanical sensitization, implicating it in thermal and inflammation-induced nociceptive plasticity and underscoring its context- and cell type–dependent functions in nociceptive neural circuits.

DISCUSSION

Our study identifies NRX as a critical regulator of mechanical nociceptive sensitization in Drosophila larvae, revealing a previously unknown role in nociceptive circuit modulation. We demonstrate that loss of nrx induces hypersensitivity, which depends on central cholinergic signaling, particularly through TENCS neurons. These neurons drive sensitization via the C4da-Goro circuit, where NRX normally suppresses nociceptive excitability by regulating presynaptic GABA_B receptor signaling. These findings establish NRX as a key synaptic determinant of nociceptive thresholds and uncover a mechanism by which synaptic adhesion molecules modulate nociceptive sensitivity through inhibitory control.

Synaptic regulation of nociception by NRXs

Synaptic cell adhesion molecules (CAMs), particularly NRXs, play pivotal roles in organizing synaptic components and maintaining neural circuit function. Our study establishes that Drosophila NRX controls nociceptive thresholds through presynaptic GABA_B receptor trafficking, with loss of NRX causing GABAergic disinhibition and mechanical hypersensitivity.

Mammalian studies demonstrate conserved yet expanded roles for NRXs in pain pathways. NRX-2 up-regulation in inflammatory pain models enhances spinal sensitization (69), while NRX-1/Neuroligin-1 interactions mediate neuropathic pain through N-methyl-D-aspartate receptor (NMDAR) signaling (70). The therapeutic effects of gabapentinoids may involve NRX-1α modulation (71), highlighting clinical relevance. Mechanistically, mouse NRXs regulate both presynaptic GABA_B receptor localization (72) and postsynaptic GABA_A receptor function (73), with human NRX-1 mutations causing cell type-specific synaptic imbalances (20).

These findings reveal an evolutionary conservation of NRX’s role in GABAergic regulation of nociception while demonstrating species-specific adaptations in receptor interactions. The consistent theme of NRX-mediated GABAergic control across species suggests that this mechanism may represent a fundamental pathway for pain modulation, offering potential targets for analgesic strategies.

Hierarchical organization of nociceptive circuits: A core pathway and parallel modulatory networks

Our study elucidates a fundamental organizational principle of nociceptive processing in Drosophila, featuring a high-threshold core circuit (C4da-Goro) working in concert with parallel modulatory networks (TENCS-mediated amplification). The TENCS neurons emerge as critical signal amplifiers within this architecture, enhancing mechano-nociceptive responses without initiating spontaneous escape behaviors. This functional specialization reveals an elegant division of labor in nociceptive processing: While C4da neurons serve as primary threat detectors with high activation thresholds, TENCS neurons operate as conditional gain controls that amplify signals only when coordinated with primary nociceptor input.

The circuit dynamics demonstrate that TENCS neurons function as cholinergic amplifiers within the SEZ, establishing a gated amplification system. Three key features characterize this mechanism: (i) TENCS activation alone cannot evoke rolling behavior, (ii) they potentiate C4da-evoked responses through Goro neurons, and (iii) they show nonlinear integration properties with primary nociceptive inputs. This architecture suggests that the C4da-TENCS-Goro pathway represents a modulatory rather than primary nociceptive circuit. The Ca²⁺ dynamics in Goro neurons reveal that TENCS input transforms rather than initiates activity, selectively enhancing response magnitude to combined C4da-TENCS activation while remaining silent to TENCS manipulation alone. Although TENCS neurons enhance Goro activity, no direct synaptic connections were detected (fig. S7), indicating that indirect circuit mechanisms, such as intermediate interneurons or neuromodulatory pathways, likely contribute. Future studies combining connectomics and in vivo functional imaging will be necessary to delineate these pathways.

This organization mirrors emerging concepts in mammalian pain systems, wherein parallel modulatory circuits regulate core nociceptive pathways (74, 75). The TENCS amplification mechanism may serve an evolutionary conserved function, maintaining defensive responsiveness to recurrent threats (e.g., parasitoid wasp attacks) while preventing maladaptive overactivation. The cholinergic nature of TENCS neurons suggests acetylcholine acts as a gain control modulator, potentially through nicotinic receptors on downstream targets. This parallels mammalian findings where cholinergic signaling modulates pain processing in spinal and supraspinal regions (75).

Notably, NRX deficiency reveals the plasticity of this modulatory circuit. Nrx knockdown induces TENCS hyperexcitability through impaired presynaptic GABA_B receptor function, creating a pathological amplification state. This finding aligns with emerging evidence that GABAergic disinhibition plays a central role in nociceptive hypersensitivity. Nakamizo-Dojo et al. (47) demonstrated how descending GABAergic (SDG) neurons suppress nociception via presynaptic GABA_B receptors on nociceptor terminals, establishing GABAergic control as a critical gate for pain signals. Similarly, Khuong et al. (31) revealed that nerve injury triggers chronic pain through caspase-dependent death of GABAergic neurons, with GABA disruption alone being sufficient to induce allodynia. While mammalian studies show that NRX regulates GABA_B receptor sensitivity (72), our discovery of direct NRX-GABA_B interaction in Drosophila reveals a more immediate control mechanism. Together, these studies underscore how GABAergic disinhibition—whether through receptor dysfunction (our study), circuit modulation (47), or neuronal loss (31)—serves as a conserved pathway for nociceptive hypersensitivity. The cell type–specific effects (TENCS hyperexcitability versus mammalian synaptic defects) highlight how conserved molecular mechanisms can be adapted for circuit-specific functions while maintaining the fundamental principle of GABAergic control over pain thresholds.

These findings support a model where nociceptive processing involves a high-threshold core circuit (C4da-Goro) for immediate threat detection, parallel modulatory networks (TENCS) that regulate response gain, and molecular brakes (NRX-GABA_B receptors) that maintain appropriate amplification levels. This architecture allows dynamic scaling of defensive responses while preventing false alarms, with disruption leading to pathological hypersensitivity.

Future directions and unanswered questions

While our study establishes the NRX’s crucial role in mechanical nociception, several important questions remain that warrant further investigation. First, does NRX similarly regulate thermal or chemical nociceptive pathways? Given that different sensory modalities often engage distinct neural circuits, examining whether NRX modulates thermal or chemical sensitivity could reveal broader principles of pain regulation. Second, the complete interneuron connectivity from C4da to TENCS to Goro neurons needs full elucidation. While we have identified these key components, a comprehensive mapping of synaptic connections, including potential parallel pathways and feedback loops, would provide a more complete understanding of this nociceptive circuit’s architecture. Third, the downstream effectors mediating NRX’s control of GABAergic signaling remain unknown. Single-cell RNA sequencing of the C4da-TENCS-Goro circuit could help identify potential interacting molecules and signaling pathways that translate NRX-GABA_B receptor interactions into changes in neuronal excitability. Such high-resolution molecular profiling could uncover not only canonical GABA signaling components but also unexpected regulatory mechanisms in nociceptive processing, potentially identifying alternative targets for modulating nociceptive circuits. Last, how do other synaptic adhesion molecules (e.g., Neuroligins and SynCAMs) interact with NRX in this pathway? Given their established roles in synapse organization, they may cooperate with or antagonize NRX’s function in nociceptive circuits. Addressing these questions will advance our understanding of synaptic regulation in nociceptive pathways and may identify potential targets for pain management interventions.

Our study establishes NRX-mediated synaptic inhibition as a critical regulator of nociceptive sensitization in Drosophila. By revealing how NRX maintains inhibitory tone through GABA_B receptor trafficking in the C4da-TENCS-Goro circuit, we provide fundamental insights into neural circuit dysregulation underlying pain hypersensitivity. These findings highlight an evolutionarily conserved organizational principle—where core nociceptive pathways interact with parallel modulatory networks—that may extend to mammalian pain systems. The identification of synaptic adhesion molecules as key regulators of pain thresholds supports potential targeted therapies for chronic pain conditions, particularly those involving impaired inhibitory control in nociceptive circuits.

MATERIALS AND METHODS

S2R⁺ cell culture

Drosophila S2R⁺ cells (RRID: CVCL_Z831) were maintained in Schneider’s Drosophila Medium (Gibco) supplemented with 10% heat-inactivated fetal bovine serum at 25°C. The cell culture medium was changed every 2 days during the entire cell culture study.

Drosophila husbandry

Drosophila stocks were raised under standard conditions at 25°C on standard cornmeal medium and kept in a 12-hour/12-hour light-dark cycle. A comprehensive list of full genotypes for each figure panel is included in table S3. Information on parental stocks is given in table S1.

Generation of nrx-6×HA

To knock in the 6×HA tag at the C-terminal of NRX, we designed a guide RNA with an online tool (http://targetfinder.flycrispr.neuro.brown.edu/) and created a donor plasmid. The guide RNA was cloned into a pCFD4 plasmid (Addgene, #49411) and inserted within the BbsI cut sites. To generate the homologous recombination construct, the 5′ homologous arm (~1.5 kb upstream sequence before the termination codon), 6×HA, and the 3′ homologous arm (~1.5 kb downstream sequence after the termination codon) were inserted into a pHD-DsRed vector, which was used as a donor template. To prevent the Cas9-sgRNA complex from cutting the repair vector, we introduced four synonymous mutations in the seed sequence (12 bp proximal to the PAM). A mixture of the donor vector and gRNA vector was injected into Cas9 flies (y, sc,v; nos-Cas9/CyO; +/+). Correct integrations were verified through PCR and sequencing using primers that bind to regions outside the integrated junctions. Primer sequences are provided in table S2.

Generation of nrx-FSF-6×HA

To generate donor plasmid of nrx-FSF-6×HA, FRT-STOP-FRT was synthesized and then inserted before 6×HA of the donor plasmid for nrx-6×HA (GenScript). The gRNA vector used was the same as described in the generation of nrx-6×HA. The plasmid mixture was injected into Cas9 flies. All plasmid constructs were verified by sequencing.

Generation of UAS-nrx-3×HA, UAS-GABA_B-R1-3×FLAG

To generate UAS-nrx-3×HA, we extracted mRNA from larval CNS of w¹¹¹⁸, performed reverse transcription PCR, and inserted the cDNA into the 10×UAS-3×HA-attB vector (fungene.tech) between the NotI and KpnI sites. Primer sequences are provided in table S2.

We then performed codon optimization of a nrx-3×HA construct to generate a short hairpin RNA (shRNA)–resistant nrx, incapable of dsRNA-JF02652 and dsRNA-HMS00403 binding, by mutagenesis of the target region. We injected the 10×UAS-nrx-3×HA construct into embryos from phiC31 transgenic flies that contained the VK00005 (75A10) docking site (Fungene Biology).

To generate UAS-GABA_B-R1-3×FLAG, GABA_B-R1 coding sequence (CDS) was constructed into the 10×UAS-3×FLAG-attB plasmid. The transgene construct was directly microinjected into embryos from phiC31 transgenic flies that contained the VK00005 (75A10) docking site (Fungene Biology). Primer sequences are provided in table S2.

Generation of UAS-FRT-NaChBac-EGFP-FRT, UAS-FRT-TrpA1-FRT

The UAS-NaChBac-EGFP plasmid and the UAS-TrpA1 plasmid were acquired from Fungene Biology. We introduced FRT sites on both sides of the NaChBac-EGFP and TrpA1 through homologous recombination, followed by phiC31-mediated transgenesis targeting the attP40 insertion site. The transgenic flies were generated by Unihuaii Inc.

Generation of 16C06-LexA, 94B10-LexA, and 71A10-LexA

16C06-LexA, 94B10-LexA, and 71A10-LexA were created by cloning the enhancer regions of 16C06-GAL4, 94B10-GAL4, and 71A10-GAL4 into the DSCP-LexA vector (Fungene Biology), followed by phiC31-mediated transgenesis targeting the attP40 insertion site (Fungene Biology).

Generation of pUAST-GABA_B-R1-RNAi-3

To generate GABA_B-R1 RNAi transgenic line inserted on chromosome 2, the oligonucleotides with the same shRNA hairpin as GABA_B-R1-RNAi-1 were synthesized, annealed, and then inserted into the VALIUM20 vector (Tsinghua University). This DNA construct was then inserted using phiC31 integrase–mediated, site-directed insertion at the attP40 locus (Unihuaii Inc). Primer sequences are provided in table S2.

Generation of LexAop-nrx-RNAi

To generate LexAop-nrx-RNAi transgenic lines inserted on chromosome 2 and chromosome 3, the oligonucleotides with the same shRNA hairpin as UAS-nrx-RNAi-1 were synthesized, annealed, and then inserted into the LexAop vector. This DNA construct was then inserted using phiC31 integrase–mediated, site-directed insertion at the attP40 or attP2 locus (Unihuaii Inc).

Molecular cloning of pAc5.1-GABA_B-R1-GFP and pAc5.1-GABA_B-R2-GFP

To prepare plasmids for transfection into S2R⁺ cells, GABA_B-R1 and GABA_B-R2 cDNAs were cloned separately into the pAc5.1-GFP vector at NotI restriction sites. Primer sequences are provided in table S2.

Real-time qPCR

RNA was extracted from CNS of 25 third-instar larvae using TRIzol (Sigma-Aldrich). Reverse transcription was performed using HiScript III RT SuperMix for qPCR (Vazyme). qPCR was performed using the SupRealQ Ultra Hunter SYBR qPCR Master Mix (Vazyme) and QuantStudio 5 real-time PCR system. Primer sets were designed using Primer-BLAST (National Center for Biotechnology Information) or using the Primer Bank database. Primer sequences are provided in table S2. Data are expressed as relative mRNA expression levels normalized to the housekeeping gene Rpl32.

Mechano- and thermo-nociception assays

We custom-built mechanical probes in accordance with previous protocols (51). Nitinol filaments of predetermined lengths and diameters were affixed to a handle for easy manipulation. Each probe was calibrated to exert a precise force, measured in grams, as indicated by bending against a scale (refer to Fig. 1A and fig. S1A). For mechano-nociception, stage- and density-controlled third-instar larvae [96 ± 3 hours after egg laying (AEL)] were carefully collected and rinsed with water. The larvae received noxious mechanical stimulation on the posterior dorsal side of the larva (abdominal segment A8) with a 3260-kPa probe.

Thermal nociception was assessed using established methods with either a temperature-controlled hot plate or probe (30, 34). Forward locomoting larvae were touched with the hot probe on mid-abdominal segments (A4-6) until the execution of nociceptive rolling. The hot plate thermo-nociception test was conducted using a thermoelectric heating plate. The larvae were placed on the surface of a metal hot plate heated to a specific temperature, and the rolling probability and duration in response to thermal stimuli were measured. Multiple trials were conducted on separate days for each genotype, and all trials were included in the analysis.

Immunohistochemistry

Third-instar larvae were dissected in phosphate-buffered saline (PBS), and brains were fixed in 4% paraformaldehyde/PBS for 20 min at room temperature. For direct observation of fluorescence-labeled neurons, brains were washed in 0.3% Triton X-100/PBS (PBST) for 5 min three times at room temperature after fixation and then submerged in VECTASHIELD mounting medium (H-1000, Vector Laboratories) for at least 30 min. For immunohistochemical analyses, larval brains were washed five times with PBST and were incubated in 5% normal donkey serum/PBST at room temperature for 2 hours. Samples were subsequently incubated at 4°C overnight with primary antibodies diluted in universal antibody diluent (WB500D, NCM Biotech). After washing five times with PBST, the samples were further incubated for 2 hours at room temperature with secondary antibodies diluted in 5% normal goat serum/PBST. Brains were then washed and mounted as above. Fluorescence images were obtained using a confocal microscope (LSM900, Zeiss). The following primary antibodies were used: mouse anti-GFP (1:500), rabbit anti-GFP (1:1000), mouse anti-elav (1:50), mouse anti-repo (1:50), mouse anti-Neuroglian (1:10), rat anti–N-cadherin extracellular domain (1:10), rabbit anti–HA-tag (1:500), Cy3 AffiniPure Goat Anti-Horseradish Peroxidase (1:500), rabbit anti–C-MYC-tag (1:500), mouse anti–FLAG-tag (1:200), and rabbit anti-V5 (1:500). Fluorescently conjugated secondary antibodies used were donkey anti-rat immunoglobulin G (IgG) (H + L) highly cross-adsorbed secondary antibody, Alexa Fluor 647 (1:100); goat anti-mouse IgG (H + L) highly cross-adsorbed secondary antibody, Alexa Fluor Plus 555 (1:200); donkey anti-rabbit IgG (H + L) highly cross-adsorbed secondary antibody, Alexa Fluor 555 (1:500); donkey anti-rabbit IgG (H + L) highly cross-adsorbed secondary antibody, Alexa Fluor 488 (1:500); and donkey anti-mouse IgG (H + L) highly cross-adsorbed secondary antibody, Alexa Fluor 488 (1:500).

Optogenetic activation

Larvae were shielded from light and raised on standard cornmeal food containing 0.4 mM all-trans-retinal (R2500, Sigma-Aldrich, made using a stock solution of 100 mM all-trans-retinal dissolved in ethanol) at 25°C. Third-instar wandering larvae were gently washed in deionized water, and about 10 larvae were placed at the center of the 10-cm petri dish with little water for acclimation (60 s). To stimulate red-shifted CsChrimson, we connected a 10 cm by 10 cm light-emitting diode (LED) back-light plate (620 nm, 12 V, 4.8 W) to an LED power driver (ONI-P12R). Movies were recorded using an industry camera (CGimagetech, CGU2-130M) adapted with an infrared long-pass filter (Zomei, >760 nm).

Ca²⁺ imaging

Third-instar larvae were dissected in fly hemolymph-like solution (HL3) [70 mM NaCl, 5 mM KCl, 10 mM MgCl₂·6H₂O, 10 mM NaHCO₃, 5 mM trehalose·2H₂O, 115 mM sucrose, 5 mM Hepes, and 6 mM CaCl₂ (pH 7.2)]. Axon terminals of either C4da A5-6 segments in the VNC, somata of Goro neurons, or TENCS neurons were imaged using LSM700 (Zeiss). Images were acquired at 512 by 512 pixels, 12-bit dynamic range in a time-lapse mode. Data were imported into ImageJ for initial movement correction (conducted with the StackReg and TurboReg plugins) and region of interest (ROI) selection. The basal GCaMP signal was calculated by averaging the fluorescent signals for the first 30 frames before the onset of stimulation. Ca²⁺ transient ΔF/F₀ was calculated by normalizing the raw signal intensity F at each timeframe to F₀ with the following formula: ΔF/F₀ = (F − F₀)/F₀. Max ΔF/F₀ was the relative maximum intensity change.

Voltage imaging of TENCS neurons

Staged third-instar larvae were pinned on a Sylgard (Dow Corning) plate and partially dissected in fly HL3 physiological saline solution described above to expose the CNS. Before imaging, larvae were incubated in HL3 buffer for at least 5 min at room temperature. Arclight imaging was performed on a Zeiss LSM700 confocal microscope with a 20× water objective, and intensities of ArcLight were measured using the ROI tool. ROIs were manually selected from the axon area using ZEN software.

Coimmunoprecipitation

For coimmunoprecipitation experiment, S2R⁺ cells were transiently cotransfected with a total of 6 μg of pAc5.1-nrx-mCherry and pAc5.1- GABA_B-R1-GFP, pAc5.1- GABA_B-R2-GFP, or pAc5.1-GFP constructs. Cells were harvested 48 hours after transfection, and total protein was extracted using cell lysis buffer for Western blot and immunoprecipitation with protease inhibitors. Extracts were centrifuged at 13,000g for 10 min at 4°C to remove cell debris. Twenty microliters of 50% (w/v) protein A/G beads was added to the clean supernatants and rotated for 2 hours at 4°C to preclear the lysates and prevent nonspecific precipitate. The samples were centrifuged at 2500g for 2 min, and the agarose pellet was discarded. The precleared supernatant was incubated overnight with 20 μl of anti-GFP agarose beads (ABclonal). The immunoprecipitates were collected by centrifugation at 2500g for 2 min and washed three times with lysis buffer. After washing, the immunoprecipitated samples were denatured and eluted for Western blot analysis probed by anti-GFP (Thermo Fisher Scientific) antibody or by anti-mCherry antibody (Abcam).

Plasma membrane protein biotinylation

S2R⁺ cells transfected with pAc5.1-nrx-mCherry and pAc5.1-GABA_B-R1-GFP or pAc5.1-GABA_B-R2-GFP were washed three times with ice-cold PBS and then incubated twice with membrane-impermeant Sulfo-NHS-SS-Biotin (1 mg/ml) for 30 min at room temperature. Cells were washed three times with quenching buffer [100 mM glycine in PBS (pH 8.0)] and twice with PBS. Cells were collected and lysed in radioimmunoprecipitation assay (RIPA) lysis buffer with complete protease inhibitor cocktail. The lysates were centrifuged at 13,000g for 10 min (4°C), and a portion of the protein extract (4%) was retained for total protein blotting. Biotinylated cell membrane proteins were precipitated overnight using BeyoMag streptavidin magnetic beads with rotation at 4°C. The beads were collected with a magnetic separator and washed five times with 1 ml of RIPA buffer. Afterward, the beads were resuspended in 50 μl of 4× SDS loading buffer and incubated for 2 hours at room temperature. Total membrane and biotinylated proteins were analyzed by SDS–polyacrylamide gel electrophoresis and Western blot. Densitometric analysis of Western blot bands was performed with ImageJ. Background intensity was subtracted from band intensity to determine the final protein quantity for each band.

Epidermal injury by UVC treatment

Early third-instar larvae were briefly rinsed with distilled water and then anesthetized with ether for ~2 min until immobilized. Larvae were placed dorsal-side-up on a Sylgard silicone plate (Dow Corning) and exposed to brief UVC radiation (254 nm, ~6 s, ~17 mJ/cm²) using a Spectro-Linker XL-1000 cross-linker (Spectronics Corporation). UV intensity was measured with a spectrophotometer (AccuMAX XS-254, Spectroline). Treated larvae were returned to mashed fly food and allowed to recover for 24 hours before behavioral assays.

Quantification and statistical analysis

Descriptions, results, and sample sizes of each test are provided in the figure legends. All replicates were biological replicates using different larvae. Data for all quantitative experiments were collected on at least 3 different days. For in vivo experiments, each n value represents data from an individual larva. All statistical analyses were performed using GraphPad Prism 10.1.2 (GraphPad Software), and all data are presented as means ± SEMs unless otherwise stated. To evaluate the statistical significance between datasets, two-tailed unpaired t test and ordinary analysis of variance (ANOVA) test were used. The log-rank test was used to compare the dataset displayed in accumulated response curves. Differences in means were considered statistically significant at P < 0.05. Significance levels are as follows: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, nonsignificant.

Supplementary Materials

The PDF file includes:

Figs. S1 to S14

Tables S1 to S3

Legends for movies S1 and S2

Other Supplementary Material for this manuscript includes the following:

Movies S1 and S2