Single-nucleus neuronal transcriptional profiling of male C. elegans uncovers regulators of sex-specific and sex-shared behaviors

SUMMARY

Sexual differentiation of the nervous system drives profound neurobiological and behavioral differences between the sexes across various organisms, including Caenorhabditis elegans. Using single-nucleus RNA sequencing, we profiled and compared adult male and hermaphrodite C. elegans neurons, generating an atlas of adult male-specific and sex-shared neurons. We expanded the molecular map of male-specific neurons and identified highly dimorphic expression of G protein-coupled receptors (GPCRs), neuropeptides, and ion channels. Our data demonstrate sex-shared neurons exhibit substantial heterogeneity between the sexes, while sex-specific neurons repurpose conserved molecular pathways to regulate dimorphic behaviors. We show that the PHD neurons display remarkable similarity to sex-shared AWA neurons, suggesting partial repurposing of conserved pathways, and that they and the GPCR SRT-18 may play a role in pheromone sensing. We further demonstrate that the ubiquitously expressed MAPK phosphatase vhp-1 regulates both sex-specific and sex-shared behaviors. Our data provide a rich resource for discovering sex-specific transcriptomic differences and the molecular basis of sex-specific behaviors.

In brief

Morillo et al. generated a single-nucleus RNA sequencing atlas of adult Caenorhabditis elegans male neurons. Compared with hermaphrodite neurons, they identified neuronal markers, sex-biased features, and regulators of male-specific behavior. These data expand the transcriptional landscape of the C. elegans nervous system, providing a foundation for studies on sex differences.

Graphical Abstract

INTRODUCTION

Across sexually reproducing species, including flies, mice, and humans,^1–3 biological sex impacts neuronal and cognitive function at multiple levels. Sexually dimorphic differences in cell number, gene expression, neuronal connectivity, and synaptic signaling result in differences in behavior and disease progression between the sexes. However, most studies centered on neuronal biology and function using model organisms typically focus on one sex^4,5 or on early developmental stages before the onset of many of these dimorphisms upon sexual maturation.⁶

Caenorhabditis elegans has two sexes, hermaphrodite (XX) and male (XO).⁷ The cell lineages of both sexes are invariant, and their neuronal connectomes have been fully mapped.⁸ Hermaphrodites have 8 sex-specific neurons, while males have 93 sex-specific neurons, many of which are involved in male copulation behaviors.⁹ The two sexes share 294 neurons, but many of these shared neurons still exhibit sexually dimorphic structural and functional features, such as differences in gene expression,¹⁰ synaptic connectivity,^7,8 and behavior (foraging, pheromone sensing, learning, and memory).^11–14 Many of these behavioral differences are mediated by sex-specific regulation in a small subset of neurons.^15–18 Thus, identifying transcriptomic changes at the single-neuron level is crucial to characterize these sexually dimorphic behaviors.

Bulk RNA sequencing (RNA-seq) has been used to characterize the neuronal transcriptomes of both sexes throughout early development¹⁹ and in young and aged male neurons,²⁰ revealing extensive transcriptional variation between the sexes. However, while single-neuron transcriptional information exists for the L4 larval⁶ and adult hermaphrodite stages,²¹ the adult male single-neuron transcriptome remains undescribed. We set out to fill this gap by performing single-nucleus RNA-seq (snRNA-seq) to describe the transcriptome of young adult male neurons. In addition to assessing the male neuronal transcriptome, we also surveyed young adult hermaphrodite neuronal transcriptomes to identify sex-based differences and uncover potential behavioral regulators.

RESULTS

High-throughput isolation of C. elegans males and hermaphrodites

In self-fertilizing populations, only ∼0.1% of C. elegans are males²²; therefore, to obtain the large number of males required for snRNA-seq, we crossed pan-neuronal histone-GFP-labeled hermaphrodites²¹ into a him-8 mutant background, which produces ∼40% males.²³ Fluorescent labeling of neuronal nuclei was observed in both sexes (Figures 1A and S1), including in the male tail, where 69/93 male-specific neurons are located.^24,25

Figure 1. Single-nucleus RNA sequencing of adult male C. elegans neurons.

(A) Representative image of pan-neuronal GFP tagged him-8 males (Prgef-1::his-58::GFP; him-8). Scale bar: 50 μm.

(B) Representative image of synchronized day 1 adult hermaphrodites and males. Six-well tissue culture inserts with modified filters (35 μm) were used to separate hermaphrodites and males, obtaining isolated populations that were on average 95% enriched for each sex. Scale bar: 100 μm.

(C) Overview of nuclei isolation and sequencing. Pan-neuronal histone GFP-tagged him-8 males and hermaphrodites were separated, nuclei were isolated, FACS sorted for Hoescht- and DAPI-positive nuclei, followed by barcoding, cDNA amplification, and library preparation using 10x Genomics chromium X, and Illumina RNA sequenced.

(D) UMAP of 63,130 male-specific neurons that passed quality control.

(E) Feature plot of known markers expressed in RnB, CEM, and HOB neurons. pkd-2 (expressed in all), clec-164 (CEM enriched), flp-17 (CEM and R5,7,9B enriched), and Y70G10A.2 (HOB enriched).

Day 1 of adulthood marks the onset of sexual maturity in C. elegans; both males and females start responding to mating cues at this stage.²⁶ Additionally, some male-specific neurons, such as the MCM and PHD neurons, arise from non-neuronal lineages upon sexual maturation,^27,28 thus would not be captured using pan-neuronal markers at earlier stages. Finally, day 1 of adulthood is preferred over larval day 4 (L4), as hermaphrodites are unable to learn using a positive associative memory training paradigm prior to adulthood.²¹ Therefore, we selected day 1 adults as the ideal baseline to study both species-specific neurobiology and behavior, as well as sexual dimorphism in conserved behaviors, such as learning and memory. Using custom size-exclusion filters (STAR Methods) to separate day 1 adult males from day 1 adult hermaphrodites, which are twice the size of males and too wide to pass through the filters,²⁹ we isolated ∼95% pure populations of each sex (Figures 1B and S1B).

Neuronal snRNA-seq of day 1 adult males

snRNA-seq is preferred over single-cell sequencing for neurons because the method reduces biases caused by dissociation problems and limits the loss of highly relevant, spatially restricted RNAs observed in traditional cell isolation protocols by preserving nuclear transcripts.²¹ Here, we applied our recently developed method²¹ to isolate and sequence neuronal nuclei from day 1 adult males, processed in parallel with age- and genotype-matched hermaphrodites. Nuclei were extracted by mechanical and chemical lysis, Hoechst-stained, and sorted by fluorescence-activated cell sorting (FACS) based on dual GFP and Hoechst signals (Figure 1C; STAR Methods). RNA was isolated, sequenced, and then processed with Cell Ranger; SoupX-corrected³⁰ ambient RNA was removed, and quality control metrics were assessed (Figures S1C–S1G). Given that the single-cell neuronal transcriptome of adult male C. elegans was previously uncharacterized, we first focused on systematic profiling of this male dataset. Across four male biological replicates, 63,130 male neuronal nuclei passed quality control filtering (STAR Methods), and we obtained a median of 401 unique molecular identifiers and 298 genes detected per nucleus (Figure S2).

Annotation and validation of male neuronal transcriptomic clusters

C. elegans males have 387 neurons, 294 of which are shared between the two sexes and are grouped into 116 neuron classes based on morphological and connectivity similarities,³¹ while 93 male-specific neurons fall into 27 distinct classes.³² To assign neuronal identities to 110 clusters derived from our dataset, we employed two complementary statistical approaches, incorporating previously identified neuronal markers and new markers identified in our recent hermaphrodite snRNA-seq dataset²¹ to resolve neuronal identities (STAR Methods). Further manual curation was used to confirm annotations and to resolve some neuron identities that were not readily apparent from statistical approaches. Using this strategy, we were able to annotate 74% of the clusters (Figures 1D; Table S1). Because the annotation process was primarily informed by hermaphrodite data and 24% (93/387) of neurons in our dataset are expected to be male specific, we posited that most of the remaining unannotated clusters represent male-specific neurons that are transcriptionally and functionally understudied. A smaller subset may represent highly dimorphic sex-shared neurons.

To assess the validity of our dataset, we examined a small subset of male-specific neurons that have been extensively studied over the past two decades, namely the extracellular vesicle-releasing CEM, HOB, and ray B-type (RnB) neurons (Figures 1D and 1E). These neurons express the only two polycystins in C. elegans, pkd-2 and lov-1, that are required for male mating behaviors, along with other functionally relevant targets that now serve as hallmark markers for these neurons.^33,34 As expected, we observed robust expression of pkd-2 across CEM, HOB, and RnB neurons, as well as enrichment of neuropeptide flp-17 specifically in ray neurons R1B, R5B, and R7B, as previously described³³ (Figure 1E). We also detected high expression of the C-type lectin clec-164, a regulator of male sex drive,³³ in both CEM and RnB neurons. We also validated that the predicted C-type lectin Y70G10A.2 is highly enriched in HOB neurons (Figure 1E). These observations confirmed that our dataset faithfully recapitulates known male neuronal biology, validating both the quality of nuclei isolation and our cluster annotation approach. This foundation enabled the annotation of previously poorly characterized male-specific neurons.

Resolving identities of previously uncharacterized male neurons

To annotate predicted male-specific neurons, we used a combination of transcriptional profiling, assessment of neuronal positional information upon promoter GFP reporter generation, and manual curation of gene expression patterns. For example, the uncharacterized serpentine receptor srd-66 was solely expressed in one of the unannotated clusters (cluster 105) (Figure 2A). To identify this cluster, we examined the expression pattern of Psrd-66::GFP animals; fluorescence was observed specifically in a neuron within the male ventral cord that is positionally consistent with the male-specific motor neuron CA7 (Figure 2B, left), with no detectable expression in hermaphrodites (Figure 2B, right). Furthermore, the glutamate transporter eat-4, which was previously shown to be expressed in CA7,³⁵ was also enriched in this cluster (Table S2), while unc-47, a GABAergic marker for adjacent CA/P6, CP7, and CA/P8 neurons but not CA7, is not expressed in this cluster (Table S2). Therefore, combining transcriptional information, neurotransmitter identity, and reporter validation, we are confident in our annotation of this cluster as male CA7 (Figures 2A and 2B).

Figure 2. Annotation of previously poorly characterized male-specific neurons.

(A) Feature plot of srd-66 expression.

(B) Imaging of srd-66p::GFP shows corresponding expression and morphology for male CA7 neurons in day 1 adult males; expression is male specific, as no signal was detected in age-matched hermaphrodites.

(C) Feature plot of srh-217 expression.

(D) Imaging of srh-217p::GFP shows corresponding expression and morphology for male SPC neurons and hermaphrodite-specific PHB expression, as previously reported.

(E) Feature plot of srg-1 expression.

(F) Imaging of srg-1p::GFP shows corresponding expression and morphology for male-specific SPV and SPD neurons and no expression in the hermaphrodite tail.

(G) Feature plot of srg-14 expression.

(H) Imaging of srg-14p::GFP shows corresponding expression and neuronal morphology in hermaphrodite PQR in the tail, while we observe at least six cell bodies expressing reporter in the male tail corresponding to SPD, PQR, and PHA/PHB neurons.

(I) Feature plot of srt-18 expression.

(J) Imaging of srt-18p::GFP shows corresponding neuronal position and morphology for male PHD neurons. We did not observe any signal in hermaphrodites (data not shown).

(K) Significant Gene Ontology (GO) terms for PHD-enriched genes (Bonferroni padj < 0.05). Unpaired, two-tailed Student’s t test.

(L) Hierarchical dendrogram based on gene expression in each day 1 adult male neuron. Neurons are color-coded by functional subtype. Blue bars on branches indicate predicted male-specific/bias neurons.

(M) Males’ ability to chemotaxis to female pheromone is attenuated upon srt-18 neuron-specific RNAi knockdown. Chemotaxis assay performed using males from TU3595 neuron-specific RNAi-sensitive strain. N = 4 biological replicates. 5–10 chemotaxis plates per replicate, ∼30–50 worms per plate. ****p < 0.0001, unpaired two-tailed Student’s t test. Boxplot: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values. (A, C, E, G, and I) Scale bar: SCT normalized expression level. (B, D, F, H, and J) Scale bar: 25 μm.

We also identified all three male spicule neurons—SPC, SPV, and SPD—as distinct transcriptional clusters (clusters 60, 54, and 74, respectively) (Figure 1D; Table S1). These ciliated neurons innervate the male spicules and regulate vulva locating behaviors.³⁶ Consistent with their sensory identity, we observe high expression of ciliated neuron markers across all three clusters (Table S3), including osm-6, which was previously detected in spicule neurons.³⁷ Within these clusters, the serpentine receptor srh-217 was highly enriched (∼40-fold over other neurons) in the cluster (number 60) corresponding to SPC (Figures 2C and 2D; Table S2). To validate this assignment, we generated Psrh-217::GFP and observed fluorescence in neurons with distinct dendrites terminating near the spicule shaft (Figure 2D), consistent with SPC morphology.³⁸ srh-217 was also expressed in the ASJ and AIB in both sexes (Figure S4A), which was previously shown only in hermaphrodites.⁶ In the tail, srh-217 was expressed in hermaphrodite PHB neurons, as previously reported, but not in male PHB neurons, consistent with our transcriptional data^6,21 (Figure 2D; Table S2).

The serpentine receptor srg-1 was highly enriched in the male SPD and SPV neurons, both of which exhibit clear processes running down the length of the spicule shaft (Figures 2E and 2F). We also detected expression of srg-1 in the ASK chemosensory neurons in both sexes, as previously reported in hermaphrodites.^6,21 Additionally, SPV and SPD but not SPC neurons expressed the serpentine receptor sra-1, providing further distinction.³⁹ SPD neurons were distinguished from SPV by the enrichment of neuropeptide flp-3 in SPV, as previously observed.⁴⁰ These transcriptional and anatomical features validate our annotations of all three spicule neurons.

One unannotated male-biased cluster (number 35) was highly enriched for ciliated neuron markers and the serpentine receptor srg-14 (Figure 2G; Table S3), which was previously detected in an unidentified male neuron pair and a few other neurons, including URX, AQR, PQR, and ASJ³⁹ (Table S2). Interestingly, our statistical test assigned this cluster to the sex-shared sensory PHA/PHB neurons using adult-hermaphrodite markers (Table S1) but observed lack of overlap with L4 hermaphrodite-based markers, suggesting extensive sexual dimorphism in adults. Psrg-14::GFP was highly expressed in at least six cell bodies in the male tail (Figure 2H), consistent with expression in PHA/PHB, SPD, and PQR tail neurons observed in our snRNA-seq data. In the hermaphrodite tail, we observed expression only in PQR (Figures 2H and S2B). These findings support a model in which PHA and/or PHB neurons become highly dimorphic and contribute to male-specific behaviors regulated by these neurons, such as mate searching.^32,33

The PHD neurons, a recently discovered male ciliated neuron pair that arise from glial transdifferentiation during sexual maturation,²⁸ were also identified among our unannotated clusters. The chemoreceptor srt-18 was highly enriched in cluster 34 and expressed 200-fold lower in cluster 96 (identified as AWB neurons) (Figure 2I; Table S2). Psrt-18::GFP showed clear expression of neurons in the lumbar ganglia and morphology, suggestive of the male PHD neurons²⁸ (Figure 2J), identifying cluster 34 as the male PHD neurons. The PHD exhibited strong expression of ciliated neuron markers (dyf and osm genes; Table S3), the immunoglobulin domain-containing protein oig-8, and the dense-core vesicle secretion factor ida-1^28,41 (Table S2). Additionally, we detected >45 neuropeptide genes, including the highest-expressed neuropeptide, nlp-50, along with the vesicular acetylcholine transporter unc-17, consistent with peptidergic and cholinergic activity of the PHD neurons. Gene Ontology analysis showed genes related to ion channel activity, signal transduction, response to stimulus, and localization (Figure 2K). These data suggest that the PHD neurons are ciliated, cholinergic, and possibly chemosensory neurons, providing a molecular characterization of these recently discovered male-specific class.

Expansion of male-specific neuronal atlas

C. elegans males have 52 sex-specific ciliated sensory neurons.⁴² Among these are the hook neuron HOB, along with the 4 CEM, 18 RnB, two PHD, and three pairs spicule-associated SPC/SPV/SPD neurons, all of which are annotated in our dataset. The remaining ciliated sensory neurons include the 18 ray A-type neurons, the hook neuron HOA, and one of three pairs of postcloacal sensilla (p.c.s.) neurons, PCA.⁴³ All of these neuron types regulate male mating behaviors, such as vulva prodding, spicule insertion, and response to mates.^36,43 The RnA neurons share a common structure and function to their adjacent RnB counterparts and are housed in the same acellular cuticular fan, but they are morphologically³⁸ and transcriptionally^33,44 distinct, notably lacking expression of genes such as lov-1 and pkd-2.³³

Based on these distinctions, we hypothesized that the remaining unannotated, ciliated gene enriched clusters in our dataset most likely correspond to RnA, HOA, and PCA neurons (Table S3). Supporting this notion, several of these clusters (numbers 29 and 41) were highly enriched in ciliary tubulin, tbb-4 and tba-9, previously shown to be expressed in RnA but not RnB neurons,⁴⁵ consistent with our data (Table S2). Moreover, high expression of mab-21, a regulator of sensory ray differentiation expressed in all A-type rays⁴⁶ and HOA, was also observed.

One cluster (number 102) showed enrichment for regulators of dopamine biosynthesis, dopamine transporter dat-1 and tyrosine hydroxylase cat-2, consistent with neurotransmitter identity of the only male-specific dopaminergic neurons rays 5A, 7A, and 9A.^35,47 Another cluster (number 44) was enriched for cell surface proteins bam-2 and sax-7, as well as glutamate-gated chloride channel avr-14, all markers enriched in the HOA neuron.^48,49

Based on these gene expression signatures, we annotated each cluster accordingly. We predict that the PCA neurons correspond to the cluster labeled “Male 1,” based on high enrichment for unc-103⁵⁰ and gar-3⁵¹ relative to the other male sensory neurons. However, the relatively low expression of ciliated neuron markers suggests that this cluster includes unciliated neurons, perhaps the related p.c.s. PCB and PCC neurons. The remaining predicted male-specific clusters that could not be confidently annotated were designated “Male 2” through “Male 12.” We have identified 10/27 male-specific neuron classes, with an additional 3 predicted classes, the p.c.s. neurons. These data provide a substantially expanded transcriptional map of male-specific neural identities and a resource for the study of both unannotated and annotated neurons.

Hierarchical clustering reveals functional organization and dimorphic relationships

Next, we asked how similar the neuron clusters are, based on gene expression profiles and using hierarchical clustering (Figure 2L). Consistent with prior observations showing that neurons cluster by type independently of developmental lineage,²¹ our transcriptome-based diagram revealed that most neuron classes grouped predominantly by function (e.g., chemosensory, motor). Notably, known male-specific neurons such as CEM, HOB, and RnB clustered together, reflecting their shared molecular identity and roles in mating behavior. Similarly, RnA and HOA neurons also clustered together, but were distinct from their related ciliated HOB and RnB counterparts.

The dopaminergic ray type A neurons 5, 7, and 9 clustered closely with the sex-shared dopaminergic CEP/ADE/PDE neurons. Importantly, most predicted male-specific or male-biased clusters also grouped together, reinforcing their sex-specific identity. Interestingly, the AWA, PHD, and SPV neurons clustered together, despite their anatomical and developmental differences. Given that SPV neurons are putative chemosensory neurons⁵² proposed to sense vulval pheromone and coordinate sperm release, and that PHD neurons were enriched for genes related to functions such as response to stimulus and localization, these features suggest that PHD neurons may have previously unrecognized chemosensory functions.

PHD sense female pheromone revealing a possibly novel sensory function

Building on the observation that PHD neurons clustered near known chemosensory neurons in our hierarchical analysis (Figure 2L), we investigated whether the PHDs might play a role in sensory behavior. Strikingly, the chemoreceptor for female volatile sex pheromone, srd-1, was not only highly expressed in AWA and ASI as expected,¹⁷ but also in >40% nuclei within the PHD neurons (Table S2), suggesting the PHD might have a role in pheromone sensing. To test this hypothesis, we performed neuron-specific knockdown of srt-18, a gene highly enriched in PHD neurons, by using a pan-neuronal sid-1 rescue strain in a sid-1 mutant background (STAR Methods). Adult males treated with srt-18 RNAi showed significantly impaired responses to female pheromones (Figure 2M). Importantly, srt-18 knockdown did not affect the response to volatile odorants, such as benzaldehyde, nor did it impair locomotion in either sex (Figures S4C and S4D), suggesting the deficit is specific to male pheromone sensing. The PHD neurons were previously shown to mediate coordinated backward movement during mating.²⁸ Notably, a recent study confirmed srd-1 expression in the PHD neurons and revealed that expression of srd-1 in both the AWA and PHD neurons serves to fine-tune volatile pheromone detection.⁵³ Our data further affirm the role of the PHD neurons in sensing more long-range mating cues and suggest these sensory neurons may have a more complex sensory role, independent of srd-1, providing a previously unrecognized sensory node in the male mating circuit.

Sex-specific gene expression correlates with sex-specific behaviors

To identify male-enriched genes, we compared our male dataset to our previously published adult hermaphrodite neuronal snRNA-seq dataset.²¹ This analysis yielded 693 male-enriched genes, including 591 neuron-enriched targets detected in males and not detected in hermaphrodites (Table S4; STAR Methods). Gene Ontology analysis showed that many of these genes were poorly characterized, with annotations limited to general cellular components. Among the better-characterized genes, there was significant enrichment for signal transduction-related functions, such as G protein-coupled receptor (GPCR) activity and phosphoprotein phosphatase activity (Figure 3A; Table S4). A total of 7% of the male-enriched genes encoded C-type lectins. While about 25% of C-type lectins are present in the vas deferens,⁵⁴ we identified a distinct set of clec genes in CEM, HOB, and RnB neurons (Figure 3B; Table S5), where they may mediate male mating efficiency, similar to previously characterized lectins such as clec-164.^7,33,54

Figure 3. Sex-specific and shared targets reveal extensive transcriptional heterogeneity across neurons.

(A) Gene Ontology (GO) terms associated with male-enriched genes not detected in hermaphrodites, generated using g:Profiler.

(B) Expression of male-enriched C-type lectins in vas deferens and male-specific neurons.

(C) GO terms associated with hermaphrodite-enriched genes not detected in males, generated using g:Profiler.

(D) Expression of hermaphrodite-specific transcriptional regulators.

(E–H) (E) Expression of male-only csGPCRs and (F) hermaphrodite-only csGPCRs; neurons with highest frequency of sex-specific expression shown. (B, D–F) Normalized average expression values and percentage of cells expressing the corresponding gene are shown. Numbers in parentheses correspond to cluster number associated with that neuron, used to distinguish clusters with the same name. (G) Heatmap of sex-shared csGPCRs and (H) neuropeptide expression between sexes. Genes expressed in the same cell type in both sexes are shown in gray. Those only in male are shown in blue and only in hermaphrodites are shown in red, while white space represents not detected. Genes present in >1% of cells with an average expression >0.001 in each neuron cluster are counted as expressed in that neuron.

Conversely, we identified 708 hermaphrodite-enriched genes, i.e., genes expressed in hermaphrodite neurons but absent in males (Table S4). These genes were similarly enriched for GPCRs but also showed significant enrichment for transcriptional regulators, including several nuclear hormone receptors (NHRs) and T-box transcription factors (TBX) (Figures 3C and 3D). Some targets such as nhr-2 and tbx-9 were broadly expressed, while others were limited to a few subsets of neurons (e.g., tbx-11, zip-6). nhr and tbx genes are known regulators of neuronal specification⁵⁵ and sex determination during development.⁵⁶ However, the expression and regulation of these factors in adult neurons and their roles in maintaining sex-specific neural states remains less well understood.

Given the critical role of GPCRs in sensory processing and the observed enrichment of sex-biased GPCRs, we next examined the dimorphic patterns of these chemoreceptors.⁵⁷ The C. elegans genome encodes roughly 1,500 GPCRs, 1,341 of which are putative chemosensory (cs) GPCRs.³⁹ These csGPCRs respond to pheromones, food-related cues, and repellents, driving behaviors like foraging and mate searching.

We identified 98 male-enriched csCGPRs (Table S6). Some, such as srd-58 (ADF) and srx-78 (PHA/PHB), were strong cell-specific markers. Notably, we confirmed the expression of known male-enriched GPCRs, including sra-1³⁹ in SPV and SPD; srr-7³³ in male CEM, HOB, and RnB neurons; and srd-66 in the male CA7 neuron (Figures 2A, 2B, and 3E; Table S2 and S7). Some of these csGPCRs showed similar expression patterns in males as previously observed in larval hermaphrodites (e.g., str-256 in AWA, srh-37 in ASK³⁹). Others, however, showed sex-enriched expression patterns. For instance, srv-25 was expressed in PHA/PHB (Figure 3E) and male CP9 (Table S2), but it was previously detected in L4 hermaphrodites⁶ and absent in adult hermaphrodites.²¹ Similarly, srh-300, enriched in male ASH neurons, had been reported in larval-stage but not adult²¹ hermaphrodite AVE⁵⁸ and AVK⁵⁹ neurons. srw-87, previously observed in embryonic ADF neurons,⁶⁰ was restricted to male ADF neurons in our dataset.

The expression patterns of these male-enriched GPCRs provided insights into neuronal specialization. Male-specific neurons such as SPV and SPD have several GPCRs in common, consistent with their homologous functional roles (Figure 3C; Table S6). Moreover, the majority of these male-enriched GPCRs, such as srd-58, are expressed in chemosensory neurons that are shared between the sexes, such as PHA/PHB, AWA, ADF, and ASK (Figure 3C). These neurons play established roles in mating behaviors. For example, PHA neurons, along with URY and PQR, regulate male reproductive drive through PDF-1 neuropeptide signaling.⁶¹ The ADF¹⁵ and ASK neurons are key mediators of sex-dimorphic pheromone detection.⁶² Therefore, the expression of specific csGPCRs in these neurons might suggest a role for these receptors in male-specific behaviors.

We also identified 89 csGPCRs that are expressed in adult hermaphrodites but not age-matched males (Table S4 and S7). Here, we captured known markers for hermaphrodite-specific neurons, such as srj-6, in the HSN motor neurons (Figure 3F). Strikingly, the sex-shared ADL chemosensory neurons exhibited 32 hermaphrodite-enriched GPCRs compared to only two male-enriched GPCRs (Table S4). Other chemosensory neurons, such as ASK and ASJ, also showed high enrichment. Many of these GPCRs had previously been identified in L4 hermaphrodite ADL neurons, such as srx-130 and srh-80,⁶ while others, such as str-229 and srz-38, were newly reported in this study.

These data demonstrate that sexually dimorphic remodeling of chemoreceptor expression occurs in a developmental stage- and sex-specific manner, particularly in shared sensory neurons critical for reproductive behaviors.

Sex-shared neurons display extensive molecular heterogeneity of shared transcriptomes

To better understand how each of the sexes employs the same molecular machinery to regulate divergent phenotypic outcomes, we examined sex-enriched patterns of csGPCRs and neuropeptide expression in sex-shared neurons. We assessed the overlap of csGPCRs expression across both sexes. On average, ∼50% of the GPCRs displayed conserved expression patterns between the sexes within a given neuron (Figures 3G and S2E; Table S6). However, there were multiple examples of sex-enriched GPCR expression within a given neuron, or heterogeneity in expression of the same GPCR across different cell types between the sexes. For example, sra-37 was detected in ASER and ADF in both sexes, but was hermaphrodite enriched in AWC, ASI, and ASK, and male enriched in ASEL. Chemosensory neurons ADL, ASJ, and ADF showed the highest expression of male- and hermaphrodite-biased GPCRs.

We next analyzed neuropeptide expression patterns, as they modulate many of C. elegans’ sex-shared and sex-specific behaviors, such as feeding, learning, and reproduction.^61,63–68 The C. elegans genome encodes 113 neuropeptide genes that produce several hundred neuropeptides, including FMRFamide-related peptides (FLPs), neuropeptide-like proteins (NLPs), and insulin-like peptides.^69–71 We detected 133 neuropeptides shared between male and hermaphrodite²¹ neurons when comparing across the same threshold (Table S7). Several neuropeptides exhibit clear sex-biased expression, such as flp-1, flp-8, and ins-30 (hermaphrodite-biased) and nlp-43 and flp-34 (male-biased). PVS (PVP in hermaphrodites), DVC, and URA neurons displayed the highest incidence of male-biased neuropeptides (Figure 3H). Chemosensory neurons including ASER, AWC, ASI, and ADL showed the highest prevalence of hermaphrodite-enriched neuropeptides (Table S7).

Integrative analysis reveals extensive neuronal dimorphism and functional divergence

To further characterize sex-differential expression at the single-neuron level, we integrated the male and genotypically matched him-8 hermaphrodite datasets, including 46,685 hermaphrodite neuronal nuclei (Figure S3). After processing, normalization, and unsupervised network clustering, we obtained 145 clusters, annotating 108 of 116 shared neurons classes (Figure 4A).

Figure 4. Sex-shared neurons reveal a highly dimorphic neuronal landscape.

(A) UMAP of all 178,606 male and hermaphrodite nuclei that passed quality control across all biological replicates (4 male, 3 hermaphrodite), resulting in 145 distinct clusters.

(B) UMAP projection of integrated male and hermaphrodite nuclei colored by sex bias. Each cluster is colored by normalized contribution of male vs. hermaphrodite nuclei within each clustered (% male, % hermaphrodite). Blue indicates male-biased, red indicates hermaphrodite-biased, and gray indicates balanced sex-shared groups. Sex-specific neurons (e.g., male CEM and RnB, hermaphrodite HSN and VC) are strongly enriched for the corresponding sex. RMG, URA/URY, PVW, and CEP/ADE/PDE are boxed in purple, highlighting sex-shared neurons segregating into distinct clusters.

(C) Comparison of expected and observed distributions of sex-shared and sex-specific neuronal clusters. Stacked bar plots show the proportion of neuronal clusters that are male specific/biased (blue), sex-shared (gray), and hermaphrodite specific/biased (red). The “hypothetical” bar represents expected distribution assuming complete recovery of all 294 sex-shared neurons, 93 male-specific neurons, and 8-hermaphrodite-specific neurons. The adjacent bar represents the distribution observed in our single-nucleus RNA seq dataset of both sexes.

(D) Genes significantly higher in males in CEP/ADE/PDE and RMG neurons. Expression level density of male (blue) or hermaphrodite (red). Adjusted p values from Wilcoxon rank-sum test.

(E) Euclidean distance of each neuron’s mean male vector and mean hermaphrodite vector. The top neuron subtypes with the largest distance are shown.

(F) Feature plot of odr-10 expression and violin plot of odr-10 differential expression between sexes confirming upregulation in hermaphrodite AWA.

(G and H) (G) Genes higher in males (blue) in selected highly distant chemosensory neurons and (H) interneurons between the sexes. vhp-1 is differentially expressed in ASK, ASJ, and AVF (shown in blue).

(I) Knockdown of vhp-1 in neuron RNAi-hypersensitive males impairs chemotaxis to C. remanei female pheromone. N = 3 biological replicates, 5–10 chemotaxis plates per replicate with ∼100–200 worms per plate. Unpaired, two-tailed Student’s t test.

(J) Knockdown of vhp-1 attenuates benzaldehyde chemotaxis only in hermaphrodites. N = 2 biological replicates. One-way ANOVA with Bonferroni post hoc analysis. ****p < 0.0001; ns, p > 0.05.

Boxplots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values.

We next examined the normalized contribution of nuclei from each sex and biological replicate per cluster, generating a uniform manifold approximation and projection (UMAP) colored by this relative distribution to visualize sex-biased clustering (Figure 4B; Table S8). Sex-specific neurons, such as the CEM and RnB in males and the HSN and VC neurons in hermaphrodites, were >95% enriched for nuclei of the corresponding sex, validating the quality of our integration. Hierarchical clustering of the integrated dataset showed robust grouping of our predicted male-specific or male-biased neurons, including close clustering of Ray 5,−.7-,9A, HOA, Male 1, and RnA (Figure S5A), while the PHD neurons clustered with other chemosensory neurons. While most male-biased neurons were >90% enriched for male nuclei, 8 of 46 male-biased neurons showed 75%–85% enrichment of male nuclei. Some of these were sex-shared neurons like the PHC, not identified in our previous hermaphrodite snRNA-seq dataset,²¹ likely reflecting technical differences in neuron recovery state. Based on these distributions, we classified clusters “sex-shared” (>30% nuclei from each sex), “hermaphrodite-specific” (<30% males), and “male-specific” (<30% hermaphrodite nuclei) (Figure 4C). This categorization resulted in 61% sex-shared, 5% hermaphrodite-specific, and 34% male-specific clusters, closely aligning with the expected distribution given known neuron numbers, 294 sex-shared, 8-hermaphrodite-specific, and 93 male-specific neurons (Figure 4C).

In fact, some sex-shared neurons were so dimorphic that they clustered as separate cell types. For example, although the homologous gene expression and functional roles of CEP, ADE, and PDE—which comprise 8 sex-shared dopaminergic neurons with mechanosensory functions³¹—cause them to cluster together, the sex-specific differences in expression cause the cluster to diverge. Similarly, the URA/URY neurons form a single cluster in hermaphrodites but two distinct male clusters (Figures 4B; Table S2). Signaling-related genes, including neuropeptides (e.g., nlp-45, flp-4, nlp-11, flp-6, nlp-64), receptors (dmsr-2, pdf-1, ser-7), and signaling genes (kcc-2, snet-1), are some of the genes enriched in male neurons in these extensively dimorphic clusters (e.g., in CEP/ADE/PDE and RMG neurons; Figure 4D; Table S9). Another subset of genes was upregulated in hermaphrodites, including several neuropeptides, such as flp-14, flp-5, nlp-46, and nlp-56 in RMG, and nlp-10, nlp-6, and nlp-13 in PVW (Table S9).

To more precisely quantify the extent of molecular divergence between male and hermaphrodite neurons, we calculated the Euclidean distance (Figure 4E; Table S10). This analysis allows us to survey how different neurons are based on their expression profile. Here, we only considered “sex-shared” clusters to ensure that both sexes were sufficiently represented in each cluster, enabling biologically meaningful distance calculations. Clusters with fewer than 30% nuclei from one sex likely represent sex-specific (i.e., CEM) or highly dimorphic neuron types that cluster separately (i.e., URY), thus making direct distance comparisons less biologically meaningful. A few neurons such as AWA, enriched for 72% male nuclei, may be an exception to this rule and are possibly more readily isolated in one sex than another. Our data suggest that AIM interneurons were the most sexually dimorphic, in agreement with previous data demonstrating divergent gene expression and neurotransmitter identity between the sexes in this neuron class.^72–74 Other highly divergent neurons included AVG and AVF interneurons; VB/DB/SAB and PDA motor neurons; and AWA, PHA/PHB, and ADF, which were the most distant chemosensory neurons between males and hermaphrodites. Mechanosensory sensory neurons, including ALM and PLM, also showed high sex divergence.

We assessed the differentially expressed genes between the sexes in these clusters, confirming expression of known sex-differential targets, such as upregulation of food chemoreceptor odr-10 in hermaphrodites relative to males¹⁶ (Figures 4F and S3B). Similarly, we confirmed the upregulation of the canonical pheromone receptor¹⁷ srd-1 in male AWA (Figure S5C). Overall, most differentially expressed genes across neurons were downregulated in males (Table S12). Thus, we posited that genes upregulated in males, like srd-1, may identify uncharacterized targets and their functional roles. Male upregulated genes include csGPCRs and peptidergic GPCRs, neuropeptides, ion channels, and enzymes (Figures 4G and 4H; Table S12). There was also a global enrichment of ion channels, including kcnl-1, kcnl-2, egl-23, and egl-26. Over 100 genes were upregulated in the memory-associated AIM in males (Figure 4H; Table S12).

The mitogen-activated protein kinase (MAPK) phosphatase vhp-1 showed higher expression in several male chemosensory and interneuron classes, including ASK, ASJ, and AVF (Figures 4G–4H and S5D; Table S12). Since ASK neurons play a role in pheromone sensing²⁶ and are one of three sex-shared neurons required for sexual attraction in males, we hypothesized that vhp-1 may play a role in pheromone sensing. Indeed, neuron-specific knockdown of vhp-1 significantly impaired male attraction to female pheromones (Figure 4I). However, vhp-1 is ubiquitously expressed across neurons, notably highest in chemosensory AWC neurons in which it is slightly more enriched in hermaphrodites (Figure S5E; Table S11). Therefore, we asked whether vhp-1 knockdown affected other functions in a sex-shared and/or sex-specific manner. We tested the effect of vhp-1 reduction on benzaldehyde chemotaxis in both sexes. Our data showed that vhp-1 knockdown significantly reduces the ability of hermaphrodites to respond to benzaldehyde, but it does not affect male chemotaxis to the same volatile odorant (Figure 4J). However, knockdown of vhp-1 did not affect pyrazine chemotaxis (AWA-regulated) or the animals’ movement in either sex (Figures S5F–S5G). Taken together, these results demonstrate that sex- and cell-specific regulation of sex-shared ubiquitously expressed genes can result in significantly divergent phenotypic outcomes.

DISCUSSION

Here, we generated the first snRNA-seq atlas of day 1 adult C. elegans male neurons, alongside age- and genotype-matched hermaphrodite neuronal profiles. Using our recently established snRNA-seq protocol, we captured a comprehensive and high-purity dataset spanning both sex-shared and male-specific neurons. This atlas provides a powerful resource for identifying sex-specific and sex-shared programs underlying neuronal identity and behavior with single-cell resolution.

Through systematic annotation, we substantially expanded the molecular map of male-specific neurons, identifying markers for known male-specific classes and providing predictions for additional understudied neuron types. These markers will facilitate functional characterization of male neurons and annotation of both male and sex-shared neurons in future studies of sexual dimorphism in the nervous system.

Beyond male-specific annotation, our data revealed previously unrecognized relationships between sex-shared and sex-specific neurons. Our findings suggest that male-specific PHD neurons may have a role in pheromone sensing. Intriguingly, they also showed high transcriptional similarity with the sex-shared AWA chemosensory neurons, both enriched for genes related to chemosensory functions, including neuropeptides, csGPCRs, peptidergic GPCRs, and ion channels, such as TRPV channel ocr-1. This overlap suggests that sex-specific circuits partially re-purpose conserved molecular components. Intriguingly, OCR-1, which acts as a temperature sensor in AWA, is also upregulated in male AWA neurons compared to hermaphrodites, suggesting a similar role in PHD neurons, and possible sex-dimorphic thermosensory regulation in the sex-shared AWA. Human TRPV2, the OCR-1 ortholog, mediates mechanical nociception in the adult somatosensory system, and its sex-differential regulation has been linked to a higher prevalence of chronic pain in females. Thus, this overlap highlights the conservation of dimorphic sensory mechanisms across species.

We also uncovered extensive transcriptional heterogeneity across sex-shared neurons. Upon integration of our hermaphrodite and male single-cell data, we observed some sex-shared neurons, such as URY and dopaminergic neurons, diverged into distinct clusters, reflecting high transcriptional variability. While most sex-shared neurons clustered together as expected, we detected widespread sexually dimorphic expression patterns of GPCRs, neuropeptides, and ion channels. Among these were strong sex- and cell-type-specific markers, including srw-87, specific to male ascaroside pheromone sensing ADF neurons, and srg-8, specific to hermaphrodite ASK neurons. Notably, srw-87 was recently reported to be downregulated in male neurons with age,²⁰ raising the intriguing possibility that regulation of srw-87 or other ADF gene targets may regulate the observed decline in pheromone sensing with age. While previous studies had reported isolated examples of sex-dimorphic neuropeptide⁶¹ and GPCR³⁹ expression, these observations were often limited to a single or a few subsets of neurons. Our data now reveal that these differences are far more pervasive than previously appreciated, affecting diverse classes of neurons across the nervous system.

Our data reveal that ubiquitously expressed, sex-shared genes can exhibit striking sex- and cell-type-specific functions. Although the MAPK phosphatase VHP-1 is broadly expressed across the nervous system, neuron-specific knockdown resulted in distinct, sex-specific behavioral phenotypes: impaired pheromone attraction in males and reduced benzaldehyde chemotaxis in hermaphrodites. These findings underscore the power of single-cell resolution to uncover functional divergence that would be masked in bulk analyses. While VHP-1 has been previously implicated in immunity and stress resistance,^75,76 its sex-specific roles had been unexplored. Interestingly, MAPK signaling has been shown to regulate dimorphic behavioral responses, such as higher pain sensitivity in female mice⁷⁷ and increased response to alcohol-induced liver injury in females with lower levels of MKP1.⁷⁸ Our findings, along with insights from mammalian systems, suggest a conserved role for MAPK signaling in the regulation of neuronal functions in a sex-dependent manner, warranting further investigation into its contribution to sensory processing and stress resilience.

Our atlas not only expands the molecular annotation of male-specific neurons but also captures new sex-specific features and conserved differential expression patterns that may influence neurological function and behavior. By providing a comprehensive, high-resolution map of adult male and hermaphrodite neuronal transcriptomes, this resource will facilitate functional analyses of sex-specific and sex-shared circuits, help identify new candidate regulators, and advance the broader study of how sex influences nervous system function at the single-neuron level.

Limitations of the study

While we found that sequencing four male and three hermaphrodite biological samples was sufficient to resolve most of C. elegans’ sex-shared adult neuron types (108/116), a few remained unresolvable. Additionally, we can only assess genes that we were able to measure in each cell type; we cannot comment on genes that may have been below the threshold of detection.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Coleen T. Murphy (ctmurphy@princeton.edu).

Materials availability

Worm strains generated in this study are available upon request. This study did not generate new unique reagents.

Data and code availability

• snRNA-seq data have been deposited at NCBI and are publicly available as of the date of publication (NCBI BioProject: PRJNA1195922; also listed in the key resources table). All other data are available in the main text or supplemental information.

• 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.^79,80

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and virus strains
OP50 E. Coli (BactoBeads)	Sigma-Aldrich	Cat. #S29021
HT115 E coli	Caenorhabditis Genetics Center	HT115
Chemicals, peptides, and recombinant proteins
Molecular Probes Hoechst 33342	Thermo Fisher	Cat. #H3570
Sigma Protector RNase inhibitor	Sigma-Aldrich	Cat. #3335402001
1,1′-Dioctadecyl-3,3,3′, 3′-tetramethylindocarbocyanine perchlorate (Dil)	Sigma-Aldrich	Cat. #41085-99-8
Benzaldehyde	Millipore Sigma	Cat. #B1334-100G
Pyrazine	Sigma-Aldrich	Cat. #P56003-5G
Critical commercial assays
10X Genomics Chromium X system using the Single Cell 3′ v3.1 Reagent Kits	10X Genomics	Cat. #1000
Illumina Tagment DNA Enzyme and Buffer kit	Illumina	N/A
Deposited data
Single nucleus RNA-seq data	This Paper	NCBI BioProject: PRJNA1195922
Experimental models: Organisms/strains
C. elegans strain N2 var. Bristol: wild type	Caenorhabditis Genetics Center	RRID:WB-STRAIN:WBStrain00000003
C. elegans strain CQ760: wqIs7 [Prgef-1::his-58::GFP] him-8(e1489) IV	This Paper	CQ760
C. elegans strain CQ830: srd 66p::GFP;Pmyo3::mcherry; him-8(e1489)	This Paper	CQ830
C. elegans strain CQ828: srg-1p::GFP;Pmyo3::mcherry him-8(e1489)	This Paper	CQ828
C. elegans strain CQ859: srh-217p::GFP;Pmyo3::mcherry;him-8(e1489)	This Paper	CQ859
C. elegans strain CQ860: srt-18p::GFP;Pmyo3::mcherry;(him-8(e1489)	This Paper	CQ860
C. elegans strain TU3595: sid-1(pk3321) him-5(e1490) V; lin-15B(n744) X; uIs72 [pCFJ90(Pmyo-2::mCherry) + Punc-119:: sid-1 + Pmec-18::mec-18::GFP]	Caenorhabditis Genetics Center	TU3595
C. elegans strain CQ826: srg-14p:: GFP;Pmyo3::mcherry;him-8(e1489)	This Paper	CQ826
C. elegans strain PB4641: C. remanei	Caenorhabditis Genetics Center	PB4641
Recombinant DNA
Plasmid pL4440 RNAi	Addgene	RRID:Addgene_1654
Plasmid: pL4440-vhp-1 RNAi	Ahringer RNAi Library	vhp-1
Plasmid: pL4440-srt-18 RNAi	Ahringer RNAi Library	srt-18
Software and algorithms
GraphPad Prism version 8.0 or 9.0	GraphPad Software	https://www.graphpad.com
Cell Ranger version 7.1.0.	10X Genomics	https://www.10xgenomics.com/support/software/cell-ranger/downloads
R software for statistical computing v4.0.2	R Core Team, 2022	https://www.r-project.org/
Seurat: R toolkit for single cell genomics	Satija Lab	https://satijalab.org/seurat/
AUCell: Analysis of ‘gene set’ activity in single-cell RNA-seq data	Aibar et al.81	https://bioconductor.org/packages/release/bioc/html/AUCell.html

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

C. elegans growth and maintenance

All strains were maintained at 20C using standard methods for the duration of experiments.⁸² C. elegans were maintained on plates made with high growth medium (HG: 3 g/L NaCl, 20 g/L Bacto-peptone, 30 g/L Bacto-agar in distilled water) with 4 mL/L cholesterol (5 mg/mL in ethanol), 1 mL/L 1M CaCl2, 1 mL/L 1M MgSO4, and 25 mL/L 1M potassium phosphate buffer (pH 6.0) added to the autoclave media after it has cooled to <60C. Plates were seeded with a full lawn of OP50 E. coli (From the CGC) for ad libitium feeding. For RNA interference assays, animals were grown on HG seeded with OP50 until L4, then transferred onto HG made with 1 mM IPTG, 100 mg/L Carbenicilin, and 50 μM Floxuridine (FuDR) seeded with respective HT115 RNAi bacteria from the Ahringer Library⁸³ until day 2. Day 2 animals were then transferred to plates without FuDR seeded with the respective freshly inoculated HT115 RNAi bacteria. All behavior assays were performed on plates made with standard nematode growth medium (NGM: 3 g/L NaCl, 2.5 g/L Bacto-peptone, 17 g/L Bactoagar in distilled water) with 1 mL/L cholesterol (5 mg/mL in ethanol), 1 mL/L 1M CaCl2, 1 mL/L 1M MgSO4, and 25 mL/L 1M potassium phosphate buffer (pH 6.0) added to the autoclave media after it has cooled to <60C. Hypochlorite synchronization was used to developmentally synchronize experimental worms, where gravid hermaphrodites were exposed to an alkaline-bleach solution (e.g., 7.5 mL sodium hypochlorite, 2.5 mL KOH, 41.5 mL distilled water) to collect eggs, followed by repeated washes with M9 buffer (6 g/L Na2HPO4, 3 g/L KH2PO4, 5 g/L NaCl and 1 mL/L 1M MgSO4 in distilled water).⁸²

C. elegans strains in this study

CQ760: wqIs7 [Prgef-1::his-58::GFP] him-8(e1489) IV

CQ830: srd-66p::GFP; Pmyo3::mcherry; him-8(e1489)

CQ828: srg-1p::GFP; Pmyo3::mcherry; him-8(e1489)

CQ859: srh-217p::GFP; Pmyo3::mcherry; him-8(e1489)

CQ860: srt-18p::GFP; Pmyo3::mcherry; him-8(e1489)

TU3595: sid-1(pk3321) him-5(e1490) V; lin-15B(n744) X; uIs72 [pCFJ90(Pmyo-2::mCherry) + Punc119::sid-1 + Pmec-18::mec-18::GFP]

CQ826: srg-14p::GFP; Pmyo3::mcherry; him-8(e1489)

PB4641: C. remanei

METHOD DETAILS

Filter design and efficiency testing

Adult males are shorter and more slender than hermaphrodites,²² thus we capitalized on this size difference to separate the sexes. We removed existing meshes from 6-well tissue culture inserts (CellQart, item#: 9300002) and replaced them with 35 μm nylon meshes (BioDesign Inc. of New York, CellMicroSieves, product #: N35S) adhered to the inserts with a very thin film of epoxy resin (Loctite, epoxy adhesive, EA015) used along the insert border. Inserts are dried overnight, meshes are cut to shape, then sterilized with ethanol. The ethanol is allowed to evaporate, and the inserts are washed with M9 before filtering animals. We tested the efficiency of each individual filter using ∼200 μL of synchronized CQ760 hermaphrodites and males. We added 50 μL of animals to each individual filter and enough M9 to lightly submerge the animals. After visual inspection, each corresponding top and bottom fraction were then added to individual 15 mL conical tubes (8 total, 1 hermaphrodite- and 1 male-enriched tube/filter) and animals were allowed to settle. Then 5 μL of animals corresponding to each fraction was transferred to NGM plates seeded with 150 μL of OP50, animals were allowed to spread then paralyzed with 7.5% sodium azide to facilitate counting. The number of males and hermaphrodites on half of each plate were counted. We then calculated the percent filtering efficiency. It is important to note that filters may tear, decreasing their efficiency, but this is easily visualized upon inspection of 6-well plates under a standard microscope prior to downstream processing. Filter efficiency is calculated as follows:

$$\%\text{hermaphrodite}:\frac{\#\text{of hermaphrodites}}{Total(\#\text{of hermaphrodites}+\#\text{of males})}\times 100$$

$$\%\text{male}:\frac{\#\text{of males}}{Total(\#\text{of hermaphrodites}+\#\text{of males})}\times 100$$

Neuronal nuclei isolation

C. elegans neuronal nuclei were isolated as previously described²¹ with minor modifications. About 0.8–1 mL of Day 1 synchronized hermaphrodites and males (CQ760) were washed from HG plates, then washed 3X times with M9 to get rid of excess OP50. Animals were then transferred to 6-well inserts with 35 μm nylon filters submerged in M9 to separate both sexes. We use 4 filters for each replicate. Roughly 150–200 μL of animals are added to each filter and allowed to filter for ∼5 min. Upon inspection of the top fraction (hermaphrodite-enriched) and bottom fraction (male-enriched), the respective fractions are pipetted into fresh 15 mL conical tubes for downstream nuclei isolation.

First, hermaphrodites were Dounce homogenized in lysis buffer 30–50x and males 50–90x to break open the cuticle. Briefly, after chemo-mechanical lysis, the pellet was resuspended in 500 μL wash buffer, Hoechst stained (1:10,000 dilution; Molecular Probes Hoechst 33342, Thermo Fisher, Cat. #H3570) was added to each sample, and samples were passed through 5 μm syringe filters, directly into FACS tubes. Samples were incubated for at least 5 min on ice prior to FACS.

Nuclei positive for both Hoechst and GFP+ were sorted using a 70 μm nozzle and a flow rate of 3 on a BD Biosciences FACSAria Fusion sorter into a 1.5 mL low bind microcentrifuge tube containing collection buffer (500 μL of 0.5% BSA +1.5. U/μL RNase Inhibitor). The instrument was washed with bleach between samples. We performed 3 biological replicates for hermaphrodites and 4 biological replicates for males to increase the number of male nuclei. For the 4^th male biological replicate samples were split into two technical replicates by splitting ∼300 μL of filtered males into two aliquots and following all described downstream nuclei isolation steps.

Library preparation, sequencing, alignment, and QC

After FACS, samples were centrifuged gently at 1000 x g for 5 min at 4°C. The gentle speed, while important to keep the nuclei intact, leaves many of them behind in the supernatant. The supernatant was removed, and nuclei were resuspended in 20 μL of collection buffer. Single nuclei suspension samples were then provided to the Princeton Genomics Core for 10X Genomics Chromium X barcoding, cDNA amplification, library preparation and Illumina next generation sequencing, as described in St. Ange and Weng et al. 2024.²¹ After Illumina sequencing, reads were aligned using CellRanger version 7.1.0. SoupX was used to remove ambient RNA contamination on Cell Ranger output files, and calculated contamination fractions were between 0.12 – 0.50 for all samples.

Next, we assessed quality control metrics to give us insight into the depth, complexity, and overall quality of our data. Average genes/cell, average UMIs/cell, and number of cells per cluster were all assessed. Violin Plots of genes per cell were generated to determine cutoffs: the lower bound was 100–300 features/cell for each sample (to remove damaged nuclei and empty droplets) and the higher bound was between 750 and 1500 features/nucleus (to remove doublets) depending on the sample. Data outside of these cutoffs were excluded from further analysis.

Normalization, integration, and clustering

We used the Seurat package single cell genomics pipeline for normalization, integration and clustering. We first merged all of the replicates. Next, we normalized the data using single cell transform (SCT), which normalizes single cell data by fitting genes to a negative binomial distribution. We generated an elbow plot of the principal components (PCs) to determine the number of PCs to include in dimensional reduction, and we used the Louvain algorithm for unsupervised network clustering. We tested different resolutions: 0.6–2.4 in 0.2 intervals. Ultimately, we clustered the data using 150 PCs at a clustering resolution of 1.4. This resulted in 110 clusters for the male data processed on its own, and 145 clusters for the male and hermaphrodite data processed together.

General cluster labeling/cell type analysis

Cluster annotation for the male-specific transcriptome was performed using a combination of systemic and manual approaches as described in St. Ange & Weng et al. 2024,²¹ with some additional curation used for understudied male-specific clusters. We first focused on only the male biological replicates to characterize the male transcriptome.

First, we used the ‘FindAllMarkers’ function in Seurat to identify cluster-specific markers. These markers corresponded to genes showing significant differential expression (log₂FC > 0.25) within each cluster compared to the average expression across all clusters and were expressed in >25% of cells in the corresponding cluster.

We then compared the identified cell markers to a curated list of known markers for each neuron type classification (Table S1). This list comes from existing sequencing data and literature. Two statistical tests were used for this comparison: 1) A hypergeometric test, assessing the Bonferroni-corrected p-value of the overlapping genes (genes present in both the cell markers list and the curated list). Clusters demonstrating a significant overlap (p < 0.01) with known anatomy markers were considered for the corresponding neuron type. 2) An Area-Under-the-Curve (AUCell) algorithm,⁸¹ which accounts for level of expression of a given gene set within the respective cell. Gene set lists were generated based on the known markers associated with a given neuron type and ranked by expression level in the cell in question. Using these rankings, we can calculate the AUC value of each gene set in each cell. To assign neuron types to cells, we generated a histogram of AUCell values and applied a threshold, manually adjusting the threshold in some cases to ensure approximately 5% of cells were assigned to each neuron term. Subsequently, we assembled these cells into clusters and determined the percent assigned to each neuron type. Neuron types with the highest 10 percentages were assessed for annotation. The agreement of these two statistical tests allows us to annotate clusters with neuronal identities in an unbiased way.

To add confidence to our annotations, we performed the hypergeometric test on not only the curated list, but also using the neuron type markers reported by St. Ange & Weng et al. 2024.²¹ To annotate a given cluster, we used the results from the two hypergeometric tests, and the AUCell algorithm. If all three methods aligned, we regarded this as the final annotation. In case of disagreement, we examined gold-standard markers associated with that neuron and made a manual decision.

Annotation of male specific neurons

For most male-enriched neurons (>70% nuclei stemming from a male sample when clustered with hermaphrodite biological replicates), markers are sparse, thus the statistical tests rarely reflect neuronal identity. First, we chose markers in our data that appeared cluster specific to generate fluorescent reporters to inform neuronal annotation. In instances where a male-enriched neuron could not be annotated through statistics or fluorescent reporters, it was labeled as closely to its neuron class as we could: male inter/motor “x”/male sensory. The results of both mechanisms and the final annotation are summarized in Table S1.

Annotation of hermaphrodite and male clusters

To annotate our hermaphrodite and male integrated dataset, we used the markers identified in our male-only analysis to repeat the hypergeometric test. We also repeated the hypergeometric test original list of known markers and the markers reported by St. Ange & Weng et al., as we well as the AUCell test for further confirmation. For sex-shared clusters where all tests aligned, we regarded this as the final annotation. For male-enriched clusters (>70% nuclei stemming from a male sample) where tests didn’t agree, we deferred to the annotation based on the markers from the male-only data analysis. For hermaphrodite-enriched clusters, we deferred to our standard approach and manual curation of gold-standard markers. The results of all tests and the final annotation are summarized in supplemental (Table S8).

Hierarchical clustering

We used the normalized expression level of each gene in a given cluster to obtain average gene expression vectors, then calculated the Euclidean distance matrix between clusters. Lastly, we hierarchically clustered using the “hclust” function with the “complete” linkage method.

Threshold setting for expressed genes

Similarly to CeNGEN⁷ and St. Ange & Weng et al. 2024,^6,21 we applied expression-level thresholds to the genes in our male-only dataset before assessing enrichment of a given gene in a particular cell type. We applied 4 thresholds that require a varying percent of expression of a gene within a cluster (0.5, 1, 1.5, or 3) and a minimum average normalized expression value (0.001) (Table S3). The same thresholds were applied to the hermaphrodite and male integrated dataset (Table S11).

Differential gene expression analysis

Differential expression analysis was conducted as previously described.²¹ Briefly, we used the FindMarkers function in the Seurat package and performed the Wilcoxon Rank Sum Test method on each neuron type, comparing male and hermaphrodite cells within the same cluster. Genes with a minimum percentage of expression in 10% of the cells were analyzed, and the significantly differentially expressed genes were identified if their log₂(FC) > 0.25 or < −0.25, and adjusted p-value <0.05.

Sexually dimorphic neurons that separated during unsupervised network clustering were merged into a singular Seurat object before differential expression. These neurons are the CEP/ADE/PDE, RMG, and PVW. We used the same statistical cutoffs for these neurons.

Heatmaps and dotplots were generated with the ggplot2 package in R.

Comparison to hermaphrodite data

For comparison with the St. Ange and Weng et al. dataset, we asked what genes were detected in our male-only dataset threshold 2 that were not detected in their hermaphrodite threshold 1 (least stringent). St. Ange and Weng et al. use the same thresholds as those in this study. In this manner we were able to identify male-enriched genes. The same approach was used to identify hermaphrodite-enriched genes. To analyze sex-shared candidates, we compared the N2 WT threshold 2 with our male threshold 2. Since we cannot compare expression levels across these different genotypes, we simply asked what genes were detected in each cluster for each sex based on the 1% threshold. Only sex-shared neurons were analyzed in this case.

Gene Ontology analysis

Upon comparing the male threshold 2 genes to the list of wild-type Day 1 hermaphrodite threshold 1 genes, we identified a list of 693 male-enriched genes and 708 hermaphrodite-enriched genes (Table S4). We then used g:Profiler⁸⁴ to perform functional enrichment analysis on this set of genes and chose significantly enriched molecular functions, biological processes, and/or cellular components (Bonferroni adjusted p-value <0.05). Similarly, we identified PHD-enriched genes using the FindMarkers function in Seurat and performed functional enrichment using g:Profiler.

GFP reporter construction

Promoters upstream of the ATG start site of target genes were PCR amplified, and restriction enzyme cloned into the pPD95.75 Fire vector (Addgene). The promoter region for each construct was 5 kb upstream of the start site or up to the stop codon of the nearest coding gene, whichever criteria was met first. Promoter:GFP plasmids were injected into N2 animals at 25 ng/μL, with 1 ng/μL of Pmyo3::mCherry as a co-injection marker.

Fluorescent microscopy

Fluorescence microscopy was performed on the Nikon AXR confocal microscope using 60X magnification. 20X magnification was used to visualize pan-neuronal expression. Animals were synchronized and grown on HG plates until day 1. Day 1 animals were paralyzed with sodium azide on agar pads. We used 488 excitation to image our GFP reporters. Images were collected using a z stack with 0.4 μM increments. Images are processed with FIJI. Processing includes maximum intensity projection, brightness/contrasts adjusting, de-speckling, rotation, thresholding, splitting, and merging color channels. We evaluated expression of each construct in both sexes in at least 5 animals. Images were taken in regions where expression was observed, and 2–3 images were taken of hermaphrodites lacking expression of male-specific target genes. DiI staining was performed by adding 5uL of 2 mg/mL DiI into 1 mL of M9 and allowing animals to gently mix in this solution for 2 h prior to washing with M9, then imaging. Here we image using 488 and 561 excitation for GFP and mCherry. Hermaphrodites and males imaged for size measurements were synchronized, paralyzed with sodium azide on agar pads, imaged using a 60X objective, then measured using the FIJI measurement tool.

Chemotaxis assays

Male pheromone preference assays were performed as previously described^20,26,85 using sid-1 mutant males with a pan-neuronal sid-1 rescue for neuronal-specific knockdown upon RNAi treatment (described above). We used C. remanei true female pheromone since it has been shown to elicit a more robust response than C. elegans hermaphrodite pheromone.^17,85 C. remanei L4 females were picked onto freshly seeded NGM plates to ensure that they were unmated. Once females reach day 4, they were picked into M9 buffer with a concentration of 10 worms/100 μL M9. Animals are kept in M9 overnight to obtain a pheromone rich supernatant. The supernatant is then isolated for experiments. Pheromone may be kept at −20°C for up to a month using single freeze-thaw cycles but works best fresh. For chemotaxis assays, 1 μL 7.5% sodium azide was added to each spot to paralyze the worms at each spot. Then, 2 μL of pheromone and M9 buffer control was spotted 5 cm apart on an unseeded 60 mm NGM plate. Males were washed 3x and 5 μL of lowly-dense pellet of males was added in-between the two spots. Then, males were given 1 h to chemotaxis to either spot. Chemotaxis toward 1% benzaldehyde (Millipore Sigma Cat. #B1334–100G) in ethanol or 10 mg/mL pyrazine (Sigma-Aldrich Cat. #P56003–5G) in ethanol was accessed using standard conditions and performed on 10 cm NGM plates.⁸⁶ Chemotaxis index is calculated as follows:

$$\text{Chemotaxis index}(\text{CI}):\frac{\#\text{worms on pheromone}-\#\text{worms on M9}}{Totat\#ofworms}\times 100$$

Euclidean distance analysis

First the data was subset into each cluster, and further subset by sex (i.e., males and hermaphrodites) using the Seurat package. Only clusters categorized as sex-shared were analyzed, as marked by roughly ∼30% or more nuclei from each sex present in that cluster. For males and hermaphrodites in each cluster, we generated a representative PC vector by averaging the column means of PCA embeddings from all cells in that cluster and genotype. Then, we calculated the Euclidean distance between the PC vector of male and hermaphrodite in each cluster. We ranked the distance from the largest to the smallest.

QUANTIFICATION AND STASTISTICAL ANALYSIS

Unpaired, two-tailed Student’s t-test was performed to compare chemotaxis with only two conditions. Wilcoxon Rank-Sum test’s adjusted p-values are used for the identification of differentially expressed genes. Experiments were repeated on separate days, using separate independent populations, to confirm that results were reproducible. Prism 9 software was used for all statistical analyses. Software and statistical details used for RNA sequencing analyses are described in the method details section of the STAR Methods. Additional statistical details of experiments, including sample size (with n representing the number of chemotaxis assays performed for behavior, RNA collections for RNA-seq, and the number of worms for microscopy), can be found in the figure legends.

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116016.

Highlights.

• Generated single-nucleus RNA-seq atlas of adult C. elegans male neurons

• Sex-biased expression prevalent across sex-specific and shared cell types

• Ubiquitously expressed genes display cell- and sex-specific regulation

• srt-18, srg-1, srh-217, and srd-66 are markers of male-specific neurons