Fig 2: CAPS1 in Trpv1-lineage neurons is required for capsaicin-evoked heat and mechanical hypersensitivity.Panel A – Schematic of the experimental design. Saline or capsaicin (20 μL, 0.1 % w/v) was injected intraplantarly in WT or KO mice, and behavioral responses were measured before (0 min) and after injection (15 and 30 min). Red arrow indicates time of injection.Panel B – Paw withdrawal latency in response to radiant heat stimuli after intraplantar saline (dotted line) or capsaicin (solid line) in WT (left; saline, n = 13; capsaicin, n = 15) and KO mice (right; saline, n = 12; capsaicin, n = 12). Two-way repeated-measures ANOVAs with Geisser–Greenhouse correction, performed separately within genotype: WT, time × treatment, p = 0.0035 (**); time, p = 0.0482; treatment, p = 0.3659. KO, time × treatment, p = 0.3605 (ns); time, p = 0.4521; treatment, p = 0.1714. Data are shown as mean ± SEM.Panel C – Percentage change from baseline in paw withdrawal latency in response to radiant heat stimuli after saline or capsaicin injection in WT and KO mice. Two-way ANOVA: treatment × genotype, p = 0.0064; treatment, p = 0.0111; genotype, p = 0.0183. Šídák multiple comparisons: WT vs. KO, capsaicin, p = 0.0008 (***). Symbols represent individual mice from panel B (crosses indicate males; circles indicate females). Data are shown as mean ± SEM.Panel D – Mechanical force required to elicit paw withdrawal after intraplantar saline (dotted line) or capsaicin (solid line) in WT (left; saline, n = 15; capsaicin, n = 14) and KO mice (right; saline, n = 12; capsaicin, n = 11). Two-way repeated-measures ANOVAs with Geisser–Greenhouse correction, performed separately within genotype: WT, time × treatment, p = 0.0016 (**); time, p < 0.0001; treatment, p = 0.0384. KO, time × treatment, p = 0.0077 (**); time, p = 0.0219; treatment, p = 0.4957. Data are shown as mean ± SEM.Panel E – Percentage change from baseline in mechanical force required to elicit paw withdrawal after saline or capsaicin injection in WT and KO mice. Two-way ANOVA: treatment × genotype, p = 0.4981; treatment, p < 0.0001; genotype, p = 0.0187. Šídák multiple comparisons: WT vs. KO, capsaicin, p = 0.0357 (*). Symbols represent individual mice from panel D (crosses indicate males; circles indicate females). Data are shown as mean ± SEM.

Fig 3: Cadps expression is largely independent of sensory neuron subtype, whereas exon 16a inclusion is enriched in nociceptors and C-LTMRs.Panel A – Example confocal images (left) of DRG sections processed for RNA in situ hybridization for Cadps and eYFP transcripts, with DAPI, from WT mice. eYFP signal is derived from the Ai32 reporter allele marking Trpv1-lineage neurons. Merged image is shown at top right. Dashed outlines indicate example ROIs, eYFP positive and negative, used for cell-area measurements. Scale bar, 20 μm. Quantification of Cadps expression per cell versus cell area (right). Dotted line indicates simple linear regression with slope = 0.0002 not significantly different from zero, p = 0.8691. Dots represent individual cells from 2 mice (n = 324).Panel B – Cadps expression (RPKM, reads per kilobase per million mapped reads) across genetically defined DRG sensory neuron populations, including peptidergic and nonpeptidergic nociceptors, five low-threshold mechanoreceptor (LTMR) subtypes, and proprioceptors (left) (see Zheng et al. 2019 for genetic labeling) and in Trpv1-lineage and non-Trpv1-lineage neurons (right) (see Goswami et al. 2014 for genetic labeling) from publicly available RNA-seq datasets (SRP198454; SRP068217). One-way ANOVA: p = 0.0031. Tukey’s multiple comparisons: peptidergic vs. Aδ-LTMR, p = 0.0026 (**); Aδ-LTMR vs. Aβ-SA-LTMR, p = 0.0026 (**); Aδ-LTMR vs. Aβ-field-LTMR, p = 0.0344 (*). Data are shown as mean ± SEM.Panel C – UCSC genome browser view of the mouse Cadps locus in the mm10 assembly showing the RefSeq gene model, 60-vertebrate basewise conservation track, scale, and chromosome 14 location. Exons are shown as bars, and exon 16a is indicated by the arrow.Panel D – Cadps exon 16a inclusion per transcript (%) across genetically defined DRG sensory neuron populations from the same public RNA-seq datasets shown in panel B. One-way ANOVA: p < 0.0001. Holm-Šídák multiple comparisons: peptidergic vs. non-peptidergic, Aδ-LTMR, Aβ-SA-LTMR, Aβ-RA-LTMR, Aβ-field-LTMR, and proprioceptors, p < 0.0001 (****); peptidergic vs. C-LTMR, p = 0.0115 (*); non-peptidergic vs. C-LTMR, p = 0.0003 (***); non-peptidergic vs. Aδ-LTMR, p = 0.0011 (**); non-peptidergic vs. Aβ-SA-LTMR, Aβ-RA-LTMR, Aβ-field-LTMR, and proprioceptors, p < 0.0001 (****); C-LTMR vs. Aδ-LTMR, Aβ-SA-LTMR, Aβ-RA-LTMR, Aβ-field-LTMR, and proprioceptors, p < 0.0001 (****); Aδ-LTMR vs. Aβ-RA-LTMR, p = 0.0340 (*). Data are shown as mean ± SEM.Panel E – UCSC genome browser zoom of the Cadps exon 16a region (top) showing the RefSeq sequence and encoded amino acids, 60-vertebrate basewise conservation track, scale, and chromosome 14 location; and CAPS1 domain map (bottom) highlighting the position of exon 16a (arrow) within CAPS1, relative to the previously described C2, pleckstrin homology (PH), Munc13 homology domain 1 (MHD1), and dense-core vesicle–binding (DCV) domains.

Fig 4: Cadps conditional deletion in Trpv1-lineage neurons reduces Cadps expression.Panel A – The Cadps KO allele contains loxP sites flanking the targeted Cadps genomic region, and the Ai32 reporter allele contains a loxP-flanked transcriptional STOP cassette upstream of ChR2-EYFP. In Trpv1Cre-expressing cells, Cre recombines the floxed Cadps allele to generate the Cadps KO allele and excises the STOP cassette in Ai32, thereby activating ChR2-EYFP expression in the same cells. Small half-arrows under the Cadps locus indicate the primer positions used in panel B to detect Cre-mediated recombination.Panel B – Representative PCR genotyping of the Cadps conditional KO, Ai32, and Trpv1Cre alleles from tail genomic DNA, and of the Cre-recombined Cadps KO allele from DRG genomic DNA, in WT and KO mice. The recombined Cadps KO band, detected with the primers indicated in panel A, is present only in KO mice (arrow), consistent with Cre-dependent recombination of the floxed Cadps allele in Trpv1Cre-expressing cells.Panel C – Example confocal images of DRG sections processed for RNA in situ hybridization for Cadps and eYFP transcripts, with DAPI, from WT (top) and KO mice (bottom). eYFP signal is derived from the Ai32 reporter allele marking Trpv1-lineage neurons. Merged images are shown on the right. Arrows indicate example labeled cells. Scale bar, 20 μm.Panel D – Quantification of Cadps/eYFP puncta per cell (left) and eYFP puncta per cell (right) in WT and KO DRG sections. Mann-Whitney tests: Cadps/eYFP puncta per cell, WT (n = 251) and KO (n = 291), p < 0.0001 (****); eYFP puncta per cell, WT (n = 313) and KO (n = 348), p = 0.0942 (ns). Outliers were identified using the Tukey interquartile range criterion (Q1 − 1.5×IQR; Q3 + 1.5×IQR) and excluded before statistical analysis. Dots represent individual cells from DRG sections collected from 2 WT and 2 KO mice. Data are shown as mean ± SEM.

Fig 5: CAPS1+e16a, but not CAPS1Δe16a, selectively enhances CaV2.2 currents in tsA201 cells.Panel A – Schematic of experimental design (left), example traces (middle) and peak current density plot (right) of whole-cell CaV2.2 currents evoked by a depolarization to 0 mV from a holding potential of −100 mV in tsA201 cells expressing recombinant CaV2.2 channels (α1B co-expressed with β3 and α2δ-1 subunits) under three conditions: CaV2.2 alone (n = 19), CaV2.2 + CAPS1+e16a (n = 22), and CaV2.2 + CAPS1Δe16a (n = 19). Kruskal-Wallis test: p = 0.0098. Dunn’s multiple comparisons: CaV2.2 vs. CaV2.2 + CAPS1+e16a, p = 0.0072 (**); CaV2.2 vs. CaV2.2 + CAPS1Δe16a, p = 0.9036 (ns). Dots represent individual cells from four independent transfections. Data are shown as mean ± SEM.Panel B – Current-voltage (I–V) relationship curves of whole-cell CaV2.2 currents in tsA201 cells expressing recombinant CaV2.2 channels alone (n = 18), with CAPS1+e16a (n = 21), or with CAPS1Δe16a (n = 18). Dotted lines represent Boltzmann-Ohmic fits. Data are shown as mean ± SEM.Panel C – Summary biophysical parameters derived from Boltzmann-Ohmic fits to individual I–V curves shown in panel B for recombinant CaV2.2 channels alone (n = 18), with CAPS1+e16a (n = 21), or with CAPS1Δe16a (n = 18) conditions: maximal conductance (Gmax), half-activation voltage (V0.5), slope factor (k), and reversal potential (Vrev). For Gmax, Kruskal-Wallis test: p = 0.0039. Dunn’s multiple comparisons: CaV2.2 vs. CaV2.2 + CAPS1+e16a, p = 0.0038 (**); CaV2.2 vs. CaV2.2 + CAPS1Δe16a, p > 0.9999 (ns). For V0.5, one-way ANOVA: p = 0.0779 (ns). For k, Kruskal-Wallis test: p = 0.7932 (ns). For Vrev, one-way ANOVA: p = 0.2366 (ns). For Gmax, one outlier in the CaV2.2 alone group and two outliers in the CAPS1Δe16a group were identified using the Tukey interquartile range criterion (Q1 − 1.5×IQR; Q3 + 1.5×IQR) and excluded before statistical analysis. For V0.5, one outlier in the CAPS1Δe16a group was identified using the same criterion and excluded before statistical analysis. For k, one outlier in the CaV2.2 alone group and two outliers in the CAPS1Δe16a group were identified using the same criterion and excluded before statistical analysis. Dots represent individual cells from four independent transfections. Data are shown as mean ± SEM.