An Open‐Source Tool for Investigation of Differential RNA Expression Between Spinal Cord Cells of Male and Female Mice Authors: Justin Bellavance, Laurence S David, Michael E Hildebrand Published: 2024 ABSTRACT ABSTRACTChronic pain is a highly debilitating condition that differs by type, prevalence, and severity between men and women. To uncover the molecular underpinnings of these differences, it is critical to analyze the transcriptomes of spinal cord pain‐processing networks for both sexes. Despite several recently published single‐nucleus RNA‐sequencing (snRNA‐seq) studies on the function and composition of the mouse spinal cord, a gene expression analysis investigating the differences between males and females has yet to be performed. Here, we combined data from three different large‐scale snRNA‐seq studies, which used sex‐identified adult mice. Using SeqSeek, we classified more than 37,000 unique viable cells within predicted cell types with the use of machine learning. We then utilized DESeq2 to identify significant differentially expressed genes (DEGs) between males and females in a variety of cell populations, including superficial dorsal horn (SDH) neurons. We found a large number of DEGs between males and females in all cells, in neurons, and in SDH neurons of the mouse spinal cord, with a greater level of differential expression in inhibitory SDH neurons compared to excitatory SDH neurons. The results of these analyses are available on an open‐source web‐app: https://justinbellavance.shinyapps.io/snRNA_Visualization/. Lastly, we used gene set enrichment analysis to identify sex‐enriched pathways from our previously identified DEGs. Through this, we have identified specific genetic players within the rodent spinal cord that diverge between males and females, which may underlie reported sex differences in spinal nociceptive mechanisms and pain processing.Keywords: dorsal horn, mouse, SeqSeek, sex difference, single‐cell RNA sequencing Summary This work is the first to create an online tool based on open‐source snRNA‐seq data of female and male mouse spinal cord cells. This work is the first to create an online tool based on open‐source snRNA‐seq data of female and male mouse spinal cord cells. With this resource, researchers will be able to analyze target gene expression at baseline for different types of cells and spinal cord sections in males versus females. With this resource, researchers will be able to analyze target gene expression at baseline for different types of cells and spinal cord sections in males versus females. Introduction In vertebrate species such as mice, rats, monkeys, and humans, the spinal cord is a central hub for sensory processing and motor output via neuronal mechanisms (Todd 2010). To uncover the cellular underpinnings and potential therapeutic targets for conditions such as chronic pain, it is important to characterize the transcriptomic expression profiles of heterogeneous cell populations within spinal circuits. Historically, the use of male rodents has far surpassed the use of female rodents in spinal cord research, which could have severe consequences for the development of potential treatments (Shansky 2019). The inclusion of both male and female rodents as animal models can help us understand the differences we see in the prevalence and severity of chronic pain conditions between men and women (Mogil 2020). Preclinical investigations using these sex‐inclusive rodent models have uncovered dramatic sex differences in spinal mechanisms of nociception and pathological pain (Mogil et al. 2024), which our group has shown can translate to human spinal circuits (Dedek et al. 2022). However, there is still an urgent need to understand the genetic and transcriptomic profiles of specific spinal cord cell populations between sexes, as characterizing sex differences in expression through unbiased approaches will help uncover the genetic basis of differences in spinal pain processing. The spinal cord dorsal horn is responsible for relaying nociceptive signals along the spinothalamic tract to the brain. As the dorsal horn receives and processes a variety of inputs from somatosensory afferents as well as from descending modulatory efferents, it is very heterogenous in cell composition and function (Todd 2010). Within the dorsal horn, nociceptive inputs and outputs are mainly processed in the outermost laminae I and II, referred to as the superficial dorsal horn (SDH), while the more ventral laminae III through V, known as the deep dorsal horn (DDH), primarily process other somatosensory modalities and premotor functions (Stachowski and Dougherty 2021). While sex‐specific and sex‐conserved molecular modulators of SDH excitability and pathological pain have been recently identified (Dedek et al. 2022; Ghazisaeidi et al. 2023; Paige et al. 2022), a transcriptional landscape of gene expression across dorsal horn cell types for both sexes has yet to be developed. Thankfully, there are rapidly emerging technologies that enable us to uncover potential sex differences in baseline gene expression within specific SDH circuits. The use of single‐cell RNA sequencing (scRNA‐seq) has steadily increased as a tool used in neuroscience research in recent years. Compared to bulk RNA sequencing, which captures the average expression signal of all cell types in a particular tissue, scRNA‐seq can capture the cellular heterogeneity found in the tissue of interest (Huang 2009). Single‐nucleus RNA sequencing (snRNA‐seq), a more recent sequencing technique, has been utilized to sequence more sensitive cells from cryopreserved spinal cord tissue without the need for tissue dissociation procedures that can induce stress‐mediated changes in transcription (Matson et al. 2018). Using snRNA‐seq datasets (Alkaslasi et al. 2021; Blum et al. 2021; Sathyamurthy et al. 2018), we combined differential expression analysis (DEA) with a pseudo‐bulk approach to compare transcriptomic profiles between specifically identified groupings of spinal cord cells and test for molecular and cellular‐based differences in spinal nociception between sexes. Methods and Materials Data Acquisition Unfiltered snRNA‐seq results were acquired from three adult mouse spinal cord studies (Alkaslasi et al. 2021; Blum et al. 2021; Sathyamurthy et al. 2018). All studies distinguished between male and female mice. Twenty‐four unique samples (13 males and 11 females) were taken from Sathyamurthy et al. (2018), 12 unique samples (six males and six females) were taken from Alkaslasi et al. (2021) and six samples (two males and four females) were taken from Blum et al. (2021). Data were downloaded directly from the NCBI database via respective accession codes (see Data Availability Statement). Mouse lines and spinal cord segments used for analysis differed between studies (Table S1). A transparent science questionnaire relating to data and analysis is found in Data S3. Data Merging and Quality Control Measures Before analysis, data from the three studies were merged using R v4.1.2. All cell identifier tags were appended with male or female identifiers before merging. To keep data consistently formatted between studies, each cell sequenced within the groups of homogenized tissue sequenced by Blum et al. (2021) was evenly distributed between sample mice using the information described in their study, such as the number and sex of mice in each group. Merged snRNA‐seq data were filtered using Seurat v4 package (Hao et al. 2021). The data were filtered by keeping only cells that had more than 100 gene features present and less than 5% RNA mapped to mitochondrial genes. No additional data filtering or normalization was performed as recommended by SeqSeek documentation (Russ et al. 2021). Cell types were predicted using the SeqSeek tool. The first step involves predicting each cell type from a set of 19 cell types. Using these predicted cell types, we merged similar cells into larger groups as shown in Table 1. Motor neuron cells were not further analyzed after less than five were categorized by SeqSeek. The second step of SeqSeek involves the use of pre‐trained neural networks to predict neural subtypes. In total, there were 69 possible sub‐categories of neuron types: 38 excitatory, 27 inhibitory, as well as MN‐alpha, MN‐gamma, PGC, and CSF‐cN cells. Neuronal subtypes were assigned to lamina based on their cell type, as shown in Table 2. All cells predicted as either “Doublets” or “Junk” were removed and not used for the remainder of the analysis. TABLE 1Categories of cell types as defined in SeqSeek and corresponding clustering of cell types.Cell typeGroupingAstrocyte 1AstrocyteAstrocyte 2AstrocyteDoubletsDoubletsJunkJunkMotor NeuronsDiscardedNeuronsNeuronsOligodendrocyte Progenitor 1OligodendrocyteOligodendrocyte Progenitor 2OligodendrocyteOligodendrocyte 1OligodendrocyteOligodendrocyte 2OligodendrocyteMicrogliaMicrogliaMeninges 1MeningesMeninges 2MeningesPeripheral glia/Boundary cap derivativePeripheral glia/Boundary cap derivativeSchwannSchwannPericytesPericytesEndothelialEndothelialEpendymalEpendymal Categories of cell types as defined in SeqSeek and corresponding clustering of cell types. TABLE 2Neuronal subtypes used for analysis of SDH neurons. The lamina regions belonging to that subtype are identified. The included subtypes comprised of lamina 1 and/or lamina 2 inner (2i) and/or lamina 2 outer (2o).ClusterLaminaExcit‐11/2oExcit‐21/2o/2iExcit‐31/2o/2iExcit‐42i/3Excit‐52i/3/4Excit‐8½Excit‐91/2/3Excit‐10½Excit‐12½Excit‐132i/3Excit‐141/2oExcit‐151/2/3Excit‐161/2o (2i/3/4)Excit‐181/2o (2i/3/4)Excit‐192i (3/4)Inhib‐13 (1/2o/2i/4)Inhib‐23 (1/2o/2i/4)Inhib‐31/2o/2i/3/4Inhib‐41/2o/2iInhib‐51/2o (3)Inhib‐63/4 (1/2o)Inhib‐72i/3 (1/2o/4)Inhib‐91/2o (2i/3)Inhib‐103 (1/2o/4/5)Inhib‐111/2o/2i/3Inhib‐121/2o/4Inhib‐131/2o/2i Neuronal subtypes used for analysis of SDH neurons. The lamina regions belonging to that subtype are identified. The included subtypes comprised of lamina 1 and/or lamina 2 inner (2i) and/or lamina 2 outer (2o). DESeq2 Analysis To perform DEA between sexes, data from the merged matrix were averaged into sample‐level RNA expression, also known as a pseudo‐bulk analysis (Squair et al. 2021). This was performed to not inflate the statistical test values. The study for which the data were produced was used as a categorical covariate to adjust for batch effects. All predicted coarse cell types were analyzed for fold change using the Wald test via the DESeq2 R package (Love, Huber, and Anders 2014). The input was a matrix of RNA expression per genes. Our DESeq2 analysis automatically normalized RNA expression, accounting for differences in sequencing depth, used Cook's distance to down‐weight outliers, and applied the Benjamini‐Hochberg method to correct for false discovery rate potentially caused by multiple testing. Gene Set Enrichment Analysis (GSEA) After DESeq2, GSEA was conducted through R using three unique MSigDB pathway databases under the ontology gene sets: cellular component (CC), molecular function (MF), and biological processes (BPs). Databases were downloaded from the official MSigDB website under the mouse gene sets category (Data Availability Statement). To conduct the analysis, six sex chromosome–linked genes (Eif2s3y, Uty, Kdm5d, Ddx3y, Xist, and Tsix) were identified as outliers because they are specifically sex‐linked genes and were therefore removed from subsequent analyses. The ‘fgseaMultilevel’ function was used from the FGSEA R package (Korotkevich et al. 2019). The minimum gene set size was set to 15, and 500,000 permutations were used to compute enriched pathways. Only enriched pathways that met the significance threshold of p‐valued adjusted (pad)j < 0.05 were kept for further analysis. Similar pathways in terms of function were collapsed to reduce correlation and function overlap. Normalized enrichment score (NES) was primarily used as a measure of effect size, as it balances for mean enrichment of the group. Detailed results of these analyses are available in Data Availability Statement. Animals All experiments involving rats were conducted in accordance with the ethical guidelines and policies of the Canadian Council for Animal Care and Carleton University Animal Care Committee. Male and female adult Sprague–Dawley (13–16 weeks; n = 7/sex) rats (Charles River Laboratories, Sherbrooke, Quebec, Canada) were utilized in this study. Rats were housed with free access to food at Carleton University Animal Care Facility before spinal cord isolation. Rat Spinal Cord Isolation Rat spinal cord isolation was performed using the previously described procedure (Dedek et al. 2022) with modifications. Briefly, rats were anesthetized with 20% urethane (Acros, Geel, Belgium) via intraperitoneal injection. Once under deep anesthesia, spinal cords were dissected via ventral laminectomy and immediately placed in an ice‐cold oxygenated protective sucrose artificial cerebrospinal fluid (ACSF) solution (50 mM sucrose, 15 mM glucose, 92 mM NaCl, 5 mM KCl, 26 mM NaHCO3, 1.25 mM NaH2PO4, 0.5 mM CaCl2, 7 mM MgSO4, 1 mM kynurenic acid, bubbled with 5% CO2/95% O2). The lumbar segment was isolated, and the dura, ventral roots, and dorsal roots were removed under a dissection microscope. Lumbar spinal cords were flash frozen using Histo‐freeze (Fisher Super Friendly Freeze‐It, Fisher Scientific, Whitby, Ontario, Canada) and sliced into three sections: the SDH, the DDH, and the ventral horn (VH) using a scalpel. Isolated tissues were transferred into cryotubes, frozen using liquid nitrogen, and stored in −80°C freezer before RNA extraction. RNA Extraction and Quantitative Reverse Transcription‐Polymerase Chain Reaction(qRT‐PCR) Total RNA was prepared from the lumbar SDH, DDH, and VH using the RNeasy Plus Mini Kit (Qiagen, Hilden, Germany) with modifications. Briefly, spinal cord tissue was mechanically disrupted with liquid nitrogen using a mortar and pestle. The disrupted tissue was lysed in Eppendorf tubes with 600 μL of buffer RLT by pipetting. The lysate was homogenized in a QIAShredder (Qiagen) by spinning in a tabletop centrifuge at 14,000 rpm for 3 min. The homogenized lysate was transferred to a gDNA Eliminator spin column and spun at 11,000 rpm for 30 s to eliminate genomic DNA. One volume of 70% ethanol was added to the flow‐through and mixed by pipetting. 700 μL of the mixture was transferred to an RNeasy spin column and spun at 11,000 rpm for 15 s. The RNeasy spin column was washed with RW1 and RPE buffers to remove biomolecules and salts nonspecifically bound to the silica membrane by brief centrifugation. RNA was eluted in RNase‐free water and quantified using a Qubit Flex fluorometer (Invitrogen, Eugene, Oregon, USA). For cDNA synthesis, 50 ng of RNA template was used for RT. The iScript cDNA Synthesis kit (Biorad, Hercules, CA, USA) was utilized for the RT. A total of 50 μL reaction volume was prepared to contain RNase‐free RNA, 1X reaction mix, 2 μL reverse transcriptase, and nuclease‐free water. The reaction mix was incubated in a thermocycler (Biorad) using the following protocol: 5 min at 25°C, 20 min at 46°C, and 1 min at 95°C. Relative expression of genes was measured using qPCR. qPCR was performed using SSoAdvanced Universal SYBR green Supermix (BioRad), CFX Connect Real‐Time PCR Detection System (BioRad), and PrimeTime qPCR primers (Integrated DNA Technologies, Coralville, Iowa, USA). qPCR primers used for gene expression assays are listed in Table 4. Sptan1 (a gene that shows no significant expression between males and females, see Figure S1) was used as a control for total cDNA input to allow comparison. Two negative controls (cDNA reaction template lacking reverse transcriptase [‐RT control] and water) were also used in all assays. Relative amounts were estimated after Sptan1 normalization (ΔCt). Relative gene expression differences using the Livak method were calculated by estimating the difference between the male and female ΔCt (ΔΔCt). For Chat, Slc5a7, Dysf, Cmbl, Colq, and Postn genes, the ΔΔCt was calculated by using the following formula: (ΔΔCt = ΔCtfemale–ΔCtmale). On the other hand, for DDX3y, Mki67, Btg2, and Pde6b, the ΔΔCt was calculated by using the following formula: (ΔΔCt = ΔCtmale–ΔCtfemale). The log2 fold changes (2−ΔΔCt) were calculated to estimate the normalized fold change difference in gene expression between male and female lumbar spinal cord tissues. TABLE 4Primers for quantitative reverse‐transcription polymerase chain reaction.Reference sequence numberExon locationSequenceChatNM_00l 170,5937–8 ′‐GCA TTG TAG TGA CTC TCG CT‐3′ 5′‐ACC CCT CCT TGA TCC TTC T‐3′ Slc5a7NM_0535215–6 5′‐GTG AAT CCA ATG TCG GTG ACT‐3′ 5′‐GCT CTA CTC TGT GGC ATA TAC TG‐3′ DysfNM_00110786952–53 5′‐CCT TCT CCA CAG CAG ATG TC‐3′ 5′‐CCT CTT CAT CAT CCT CTT CAT CC‐3′ CmblNM_0010087705–6 5′‐CAG TCT TCT CTC TTC CGA TGC‐3′ 5′‐TGA TGC TGT GAT TCC ACT TGA G‐3′ ColqNM_0192741–2 5′‐TTG CCT CGT TTC TTC TGA TCC‐3′ 5′‐CTT CTT CTG CTC CAT CGT GTC‐3′ PostnNM_00110855017–18 5′‐TCA CTT CTG TCA CTG TTT CAC C‐3′ 5′‐AGT CAT TCA AGG CAG TCT TCA‐3′ Ddx3yNM_0011676651–3 5′‐CTG ATT ATC TGA GTT CAT GTC AAG AC‐3′ 5′‐GTC AGG TTT ACG GTG GCT T‐3′ Mki67XM_0010562213–4 5′‐TCA TGG CTT CCG TCT TCA TAC‐3′ 5′‐TCC GCC AAT CCA ACT CAA G‐3′ Btg2NM_0172591b—2b 5′‐TTG ATG CGG ATA CAG CGA TAG‐3′ 5′‐AGC AGA GAC TCA AGG TTT TCA G‐3′ Pde6bNM_001106020420–21 5′‐GAG CTT TCC ACT CCT TTC TGT‐3′ 5′‐GCT TCA TCG ACT TCG TAT GTA CT‐3′ Sptan1NM_17198336–37 5′‐TGC CCA TGT CAA TAA CTC CA‐3′ 5′‐CAG CGA TGA GTC ATA CAA GGA‐3′ Primers for quantitative reverse‐transcription polymerase chain reaction. ′‐GCA TTG TAG TGA CTC TCG CT‐3′ 5′‐ACC CCT CCT TGA TCC TTC T‐3′ 5′‐GTG AAT CCA ATG TCG GTG ACT‐3′ 5′‐GCT CTA CTC TGT GGC ATA TAC TG‐3′ 5′‐CCT TCT CCA CAG CAG ATG TC‐3′ 5′‐CCT CTT CAT CAT CCT CTT CAT CC‐3′ 5′‐CAG TCT TCT CTC TTC CGA TGC‐3′ 5′‐TGA TGC TGT GAT TCC ACT TGA G‐3′ 5′‐TTG CCT CGT TTC TTC TGA TCC‐3′ 5′‐CTT CTT CTG CTC CAT CGT GTC‐3′ 5′‐TCA CTT CTG TCA CTG TTT CAC C‐3′ 5′‐AGT CAT TCA AGG CAG TCT TCA‐3′ 5′‐CTG ATT ATC TGA GTT CAT GTC AAG AC‐3′ 5′‐GTC AGG TTT ACG GTG GCT T‐3′ 5′‐TCA TGG CTT CCG TCT TCA TAC‐3′ 5′‐TCC GCC AAT CCA ACT CAA G‐3′ 5′‐TTG ATG CGG ATA CAG CGA TAG‐3′ 5′‐AGC AGA GAC TCA AGG TTT TCA G‐3′ 5′‐GAG CTT TCC ACT CCT TTC TGT‐3′ 5′‐GCT TCA TCG ACT TCG TAT GTA CT‐3′ 5′‐TGC CCA TGT CAA TAA CTC CA‐3′ 5′‐CAG CGA TGA GTC ATA CAA GGA‐3′ Results Composition of Mouse Spinal Cord snRNA‐Seq Data Favors Neurons After Removal of Non‐Viable Cells To create a snRNA‐seq database that includes sex as a biological variable, we merged the results of three independent snRNA‐seq studies on spinal cord tissue, which included lumbar spinal cord segments (Figure S1) from adult male and female mice (Alkaslasi et al. 2021; Blum et al. 2021; Sathyamurthy et al. 2018), into a unified dataset comprising 239,570 potential spinal cord cells. After discarding artifactual cells assigned as either doublets (177,007) or junk (26,556) (Figure 1A), we were left with the distribution of spinal cord cells that heavily favors neurons with 27,784 cells, then oligodendrocytes with 3527 cells and astrocytes with 1673 cells from a total of 36,013 cells (Figure 1B). Barplot of number of cells categorized by cell type classification determined by SeqSeek. (A) Includes cell types classified as Doublets and Junk. (B) With cells categorized as Doublets or Junk discarded. The Number of Differentially Expressed Genes (DEGs) in the Spinal Cord Increases After Removing Cell‐Type Heterogeneity For all cell types, there were six strongly DEGs that were linked to sex chromosomes: Eif2s3y, Uty, Kdm5d, Ddx3u, Xist, and Tsix. These genes, having well‐studied sex‐specific effects and extremely elevated p values, were excluded from subsequent analyses as to not interfere with the identification of nociception‐relevant pathways and gene sets within our data, except for their use as positive controls in validation qRT‐PCR experiments for our sex‐specific analytical approaches (see the section below). To meet the criteria to be considered a DEG, a gene needs to meet the significance threshold (p‐value adjusted for multiple testing of less than 0.05) set by the Wald test. Excluding sex chromosome–linked genes, we found three DEGs associated with astrocytes and nine DEGs associated with oligodendrocytes. In neurons, we found 4372 DEGs, while there were only 3789 DEGs when considering all cell types combined together (Figure 2; Table 3). This suggests that as the cellular population is further refined, the resolution for detecting DEGs can increase. Volcano plots showing differential expression of genes between spinal cords of male and female mice. Manual y‐axis cutoff at 20. X and Y‐linked genes not included. (A) Volcano plot of genes expressed by all cells. (B) Volcano plot of genes expressed by neurons. (C) Volcano plot of genes expressed by SDH neurons. (D) Volcano plot of genes expressed by SDH inhibitory neurons. (E) Volcano plot of genes expressed by SDH excitatory neurons. TABLE 3Cell types with their respective number of cells identified and the number of significant DEGs. Cell types with less than three significantly DEGs not shown.Cell typeNumber of cellsNumber of DEGsAstrocytes16733Oligodendrocytes35279Neurons27,7844372All Cell Types36,0133789 Cell types with their respective number of cells identified and the number of significant DEGs. Cell types with less than three significantly DEGs not shown. In terms of differences by sex, there were 1662 DEGs in females and 2872 DEGs in males (Figure 2A). Within those, 17 DEGs had a log base‐two fold change (log2FC) of less than −1, which represents double the expression in females compared to males, and 293 DEGs had a log2FC of more than 1, representing double the expression in males compared to females. When the cellular population was refined to include only neurons in the spinal cord, there were 1359 DEGs in females and 3020 DEGs in males (Figure 2B). For spinal cord neurons, 23 DEGs had a log2FC of less than −1, and 495 DEGs had a log2FC of more than 1. DEGs Are Present Within the SDH of Female and Male Mice We next further refined the neuronal population to investigate the proportion of DEGs between males and females within nociceptive SDH neurons. Of 15,541 SDH neurons, we identified a total of 1863 DEGs between males and females, with 594 DEGs in females and 1269 DEGs in males (Figure 2C). For these SDH neurons, 54 DEGs had a log2FC of less than −1, and 211 DEGs had a log2FC of more than 1. When the SDH neuronal population was further stratified into inhibitory and excitatory subpopulations, we identified 2052 DEGs from a total of 11,320 inhibitory SDH neurons, with 536 DEGs in females and 1516 DEGs in males and 87 DEGs that had a log2FC of less than −1 and 318 that had a log2FC of more than 1 (Figure 2D). Interestingly, within 4221 excitatory SDH neurons, we found a total of only 46 DEGs, with 5 in females and 41 in males (Figure 2E), and 3 DEGs with a log2FC of less than −1 and 32 with a log2FC of more than 1. From this, we can conclude that there is much higher differential gene expression between sexes for inhibitory SDH neurons compared to excitatory SDH neurons. Development of an Open‐Access RShiny Tool for Comparison of DEGs Between Male and Female Mouse Spinal Cords To allow the scientific community to investigate DEGs within male versus female mouse spinal cords by defined cell types, we have developed a public tool that is available as a webpage. This tool allows investigators to easily view and access results from the DESeq2 analysis across different cell compositions, including (1) all cells of the spinal cord, (2) all neurons of spinal cords, (3) SDH neurons, (4) SDH inhibitory neurons, and (5) SDH excitatory neurons. The tool can be found at the following web address: https://justinbellavance.shinyapps.io/snRNA_Visualization/. qRT‐PCR in Rat Spinal Cord Validates Sex‐Based Differential Gene Expression for Select Unbiased Genetic Targets To test whether specific putative DEGs are differentially expressed between males and females in a second rodent species, we performed validation qRT‐PCR experiments in male versus female SDH, DDH, and VH tissue from adult rats. Using the Shiny App tool, we identified Chat, Cmbl, Colg, Dysf, Postn, Slc5a7, Xist, Tsix, Rmst, and D5Ertd615e as unbiased targets with significantly higher expression in the SDH of female mice. In contrast, DDx3y, Btg2, Mki67, Pde6b, Eif2s37, Uty, Kdm5d, Mettl21c, Zfp750, and Srrm3os are more highly expressed in the SDH of male mice. Among these genes, we selected 10 genes for validation using qRT‐PCR based on the availability of the PrimePCR primers from Integrated DNA Technologies (Table 4), with Chat, Cmbl, Colg, Dysf, Postn, and Slc5a higher in females and DDx3y, Btg2, Mki67, and Pde6b higher in males. For the six gene targets that were preferentially expressed in female mice, qRT‐PCR analysis in rat tissue showed that both Chat and Colq had 2‐ to 4‐fold higher expression in female SDH compared to male SDH (Figure 3A), while Postn had noticeably higher expression in female DDH and VH (Figure 3B,C). Among the four genes that showed relatively higher expression in male mouse spinal cord compared to female mouse spinal cord, Ddx3y (a y chromosome–linked gene) had dramatically higher fold‐change expression in female rats for all three regions tested (Figure 3D). In addition, Btg2 and Pde6b also showed a moderate preferential expression in males compared to females for rat SDH (Figure 3E) and DDH (Figure 3F) tissue. Taken together, these findings demonstrate that a subset of candidate sex‐based DEGs identified from mouse snRNA‐seq datasets are also differentially expressed between sexes in specific laminar regions of rat spinal cord tissue. Differential expression of selected genes between male and female rat spinal cord. Normalized expression as fold change expression of genes in females using males as reference in SDH (A), DDH (B), and VH (C). (A) Data shows that five genes show noticeably higher expression in female SDH, with Chat and Colq showing the highest (2‐ to 4‐fold higher). In contrast, only Postn shows higher expression pattern in female DDH (B) and VH (C). (D–F) Normalized expression as fold change were determined using females as reference. (D) Fold change expression of DDx3y in SDH, DDH, and VH showing a robustly higher expression profile in males than females. (E) Btg2 and Pde6b show higher expression in male SDH. (F) Fold change expression pattern in DDH showing higher expression pattern of Pde6b in male DDH than female DDH. Overlaid lines represent mean (thick line) ± standard deviation (thin line). GSEA Identifies Genes Relevant to Sex Differences in Synaptic Assembly, Neurotransmission, and Excitability We next performed GSEA using MF (Figure 4), CC (Figure 5), and BP (Figure 6) pathways on (1) all spinal cord cells, (2) spinal cord neurons, and (3) SDH neurons, which revealed a large amount of significantly enriched gene sets for both males and females. Using NESs and p‐values adjusted for multiple testing (padj), GSEA enabled a non‐biased identification of differentially expressed genetic targets that could be further investigated in relation to synapse formation and organization, neurotransmission, and excitability, among others. Representation of top five MSigDB's “MF” pathways significantly enriched in a sex‐specific manner using FGSEA with results from DESeq2. NES represents a score normalized to the mean of random samples of same size. (A) Pathways enriched for all cells. (B) Pathways enriched for all neurons. (C) Pathways enriched for all SDH neurons. Representation of top 5 MSigDB's “cellular composition” pathways significantly enriched in a sex‐specific manner using FGSEA with results from DESeq2. NES represents a score normalized to the mean of random samples of same size. (A) Pathways enriched for all cells. (B) Pathways enriched for all neurons. (C) Pathways enriched for all SDH neurons. Representation of top 5 MSigDB BP pathways significantly enriched in a sex‐specific manner using FGSEA with results from DESeq2. NES represents a score normalized to the mean of random samples of same size. (A) Pathways enriched for all cells. (B) Pathways enriched for all neurons. (C) Pathways enriched for all SDH neurons. Performing GSEA with the MF pathways in all spinal cord cells revealed 14 significant pathways for male mice and 11 significant pathways for female mice (Figure 4A). Within neuronal cells, six pathways were significantly enriched in males and seven pathways were enriched in females (Figure 4B). For SDH neurons, seven pathways were significantly enriched in males, with 13 in females (Figure 4C). A similar number of enriched pathways were found for GSEA with the CC pathways, with 11 versus 18 pathways in all cells (Figure 5A), 7 versus 10 pathways in neurons (Figure 5B), and 5 versus 16 enriched pathways in SDH neurons (Figure 5C), for males and females, respectively. Performing GSEA with BP pathways identified an even greater number of enriched pathways, with 35 significantly enriched pathways in males and 53 significantly enhanced pathways in females for all cells (Figure 6A). For only spinal cord neurons, 5 pathways were significantly enriched in males and 18 pathways were enriched in females (Figure 6B). Finally, with SDH neurons, we identified 17 pathways that were significantly enriched in males and 40 enriched pathways in females (Figure 6C). Discussion By leveraging the datasets of three independent snRNA‐seq studies on adult male and female mice, we have developed an open‐source online tool to investigate sex differences in gene expression within specific spinal cord cell populations. We found that neurons are the dominant spinal cord cell type in these datasets and that the number of DEGs between males and females increases when neurons are specifically studied. Interestingly, we also found that the percentage of DEGs is higher in inhibitory SDH neurons compared to excitatory SDH neurons. A subset of candidate DEGs from mouse snRNA‐seq were shown to be differentially expressed between sexes in rat spinal cord tissues using qRT‐PCR. Given the large number of DEGs identified within all cells of the spinal cord, all neurons of the spinal cord, and all SDH neurons, we used GSEA to identify differentially enriched pathways between male and female mice within three separate databases (Liberzon et al. 2015; Subramanian et al. 2005), with BP pathways showing the greatest differential enrichment between sexes. Our online tool and preliminary findings enable further investigation into specific pathways and genes of interest to uncover the transcriptomic underpinnings of sex differences in spinal mechanisms of somatosensation, including pain. After combining the three snRNA‐seq datasets and running the SeqSeek pipeline, doublets and junk cells were excluded from further analysis. Doublets are defined as droplets that are deemed unusable because they likely encompass two or more cells that have been accidentally combined into a single droplet, meaning they were wrongly marked as being a singular cell when sequenced (Xi and Jingyi 2021). Junk cells, which are also known as dropouts, are defined as unusable cells that are likely a result of a malfunction of the single‐cell or single‐nucleus sequencing method (Hicks et al. 2018). Both are routinely filtered out as part of the beginning steps of any single‐cell or single‐nucleus analysis. That being said, doublets usually make up, at most, 40% of cells during sequencing. The higher proportion of doublets we observed here is likely driven by the fact that SeqSeek cannot identify cell types that were excluded from the pre‐defined training set data, and therefore it does not include immune cells within its predictions. It is therefore possible that a significant proportion of the identified doublets are actually immune cells, such as T cells and B cells, which could not be properly classified. Cellular composition from our combined dataset heavily favored neurons over any other cell types. This differs from previous literature, which states that the neuron composition versus glia cell population is approximately 45% in mouse spinal cord gray matter and less than 10% in white matter (Bjugn 1993). Moreover, this is not in line with recent single‐cell atlases of the mouse spinal cord, which give a lower neuron‐to‐non‐neuron ratio (Russ et al. 2021). The over‐prediction of neurons by SeqSeek might be a result of over‐sequencing of neuronal cells by 10x Chromium, used by Alkaslasi et al. (2021) and Blum et al. (2021), as seen in a systematic comparison of single‐nucleus sequencing technology in the mouse cortex (Ding et al. 2020). These discrepancies could also be due to systematic differences between snRNA‐seq data and scRNA‐seq. Although there is a lack of direct comparisons between the two techniques in the CNS, snRNA‐seq has been reported to cause less tissue damage and cell stress via different cell isolation and suspension protocols (Matson et al. 2018), leading to a decreased presence of immune‐response cells such as microglia and astrocytes (Habib et al. 2017; Slyper et al. 2020). After identifying DEGs within all identified cell types separately and combined, we found that only three categories of cell types merited further investigation. Only neurons, oligodendrocytes, and astrocytes had DEGs beyond the six sex chromosome–linked genes: Eif2s3y, Uty, Kdm5d, Ddx3u, Xist, and Tsix. This is likely because these cell types had the largest number of cells within the data. In fact, there are likely more DEGs to be found within astrocytes and oligodendrocytes, as identified by many other studies, given an increased cell sample size (Chen et al. 2018; Fiore et al. 2022; Hartman et al. 2020; Yasuda et al. 2020). We identified a very large number of DEGs within all spinal cord cell types combined, an analysis that mimics a standard bulk RNA‐seq analysis. Interestingly, we identified even more DEGs by sex within only neurons of the spinal cord, even if there are fewer cells used for analysis. These results favor the analysis of sex‐specific differences in expression by coarse cell type as more effective for identifying DEGs than considering all cells as a single homogeneous population. This is in line with the work by Lopes‐Ramos et al. (2020), which found that different cell types rarely share the same DEGs between sexes. Therefore, what might be upregulated for one sex in neurons might be downregulated in astrocytes and oligodendrocytes. Moreover, these DEGs can then be used to infer cell type–specific differences in spinal processing between males and females, such as the increased number of DEGs in inhibitory SDH neurons compared to excitatory SDH neurons. While more work needs to be done to uncover the physiological and pathological relevance of this contrasting degree of sex differences between inhibitory and excitatory SDH neurons, recent papers have found that only neurons within the GABAergic system of the insular cortex exhibit sex differences (Iezzi et al. 2024) and that there are known sex‐dependent mechanisms of nociception related to spinal cord GABA receptors (Delgado‐Lezama et al. 2021). In terms of pathological pain, inhibitory spinal neurons exhibit greater changes in protein translation following nerve injury compared to excitatory neurons (Lister et al. 2024). Finally, within single‐cell sequencing data from the mouse brain, more GABAergic neuron groups are sex dominated than glutamatergic neuron groups (Yao et al. 2023). This is interesting when considering that accounting for cellular heterogeneity in the spinal cord leads us to identify unique DEGs and, in many cases, increases the significance of previously identified DEGs. The DEGs identified through our unbiased pipeline and subsequently chosen to be validated for differential expression between male and female rats represent a large variety of roles within the CNS. For example, Chat and Colq have been shown to play significant roles in synaptic assembly and neurotransmission (Misgeld et al. 2002; Sigoillot et al. 2010). Slc5a7, Colq, and Mettl21c have all been shown to play roles in various motor functions and disorders (Huang et al. 2014; Salter et al. 2018; Sigoillot et al. 2010). Pde6b, Cmbl, Slc5a7, and Znf750 have been shown to play roles in neurological and psychiatric disorders (English et al. 2009; Kuehlewein et al. 2021; Sun et al. 2015). Within our validation analysis, we found that Chat and Colq both have 2‐fold higher expression in the female rat SDH compared to the male SDH. We also found that Postn has elevated expression in the female DDH and VH. Lastly, Pde6b shows an increase in expression in males relative to females in the SDH. In general, literature on differences in spinal cord gene expression between female and male mice is scarce. However, it has been shown that mechanical sensitivity did not differ between male and female ChAT knock‐out mice in a model of neuropathic pain, which might suggest upstream regulation of ChAT protein function (Halievski et al. 2022). We can however hypothesize that differences in gene expression might contribute to differences within spinal cord functions. Nevertheless, more work needs to be done in this area to truly decipher the role of these genes and how they differ between sex. Our GSEA revealed many enriched pathways in relation to either male or female mice. Across these enriched biological pathways, there are diverging patterns of differences between males and females. In females, we observed enhanced pathways such as “regulation of synaptic activity,” “synaptic membrane,” and “extracellular matrix structural constituent conferring tensile strength,” all of which relate to synaptic structure and function. It is possible that increased baseline expression of specific synaptic proteins in females could underlie the sex differences observed in mechanisms of spinal hyperexcitability and pathological pain (Barcelon et al. 2023; Dedek et al. 2022). In contrast, male‐enriched pathways included “protein refolding,” “ribosomal subunit,” and “misfolded protein binding,” among others (Data S2). These protein functionality and translation pathways are also critical targets of dysregulation linked to chronic pain (Tansley et al. 2018), which could be differentially altered in males versus females. The GSEA pathways can also hint at particular genes of interest that are related to known pathological conditions. For example, the validated genes within the GSEA, Colq and Chat, which were differentially expressed in female mice, are part of gene sets related to pathways of synaptic organization and assembly, dendrite development, and neuromuscular junction development, especially in the SDH (Data S2). Postn has been shown to be present in gene sets for pathways related to cell growth, structure organization, and neuromuscular junctions within the SDH (Data S2). Lastly, Pde6b has been shown to be present in gene sets of male‐enriched pathways related to the detection of stimuli in the SDH, which likely relates to the ability to detect nociceptive stimulus in males (Data S2). These provide further information regarding the role of these DEGs within male and female mice and merit further investigation for their role within the sex differences seen in the male/female spinal cord. This study is one of the first to investigate transcriptomic sex differences within different cell types in the mouse spinal cord, including the first to study sex differences in SDH neuronal expression. Our results therefore inform further investigations into the sex differences found in nociceptive signaling and chronic pain as a whole. This study also provides a blueprint to generate a comprehensive snRNA‐seq analysis within spinal cord tissue of human males versus females when enough samples become available for higher‐powered analysis (Yadav et al. 2023). Additionally, the identified DEGs in this study could serve as potential therapeutic targets to aid in closing the gap in chronic pain etiology between males and females and help advance the potential of personalized medicine. Author Contributions J.B. conducted the computational analysis and prepared the manuscript draft. L.S.D. performed the validation and prepared the manuscript draft. All authors participated in the study conception, design, and manuscript revision. Conflicts of Interest The authors declare no conflicts of interest. Peer Review The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1002/jnr.70008. Supporting information Table S1. Figure S1. Data S1. Data S2. Acknowledgments We would like to thank Dr. Ariel Levine and Dr. Kaya Matson for their support with this work and Newton Martin for his help with R and Table 2. Funding: This work was supported by Natural Sciences and Engineering Research Council of Canada. Contributor Information Justin Bellavance, Email: justin.bellavance@carleton.ca. Michael E. Hildebrand, Email: mike.hildebrand@carleton.ca. Data Availability Statement Tool web address to explore DEGs: https://justinbellavance.shinyapps.io/snRNA_Visualization/. NCBI accession numbers for the 3 datasets: Sathyamurthy: GSE103892; Alkaslasi: GSE167597; Blum: GSE161621. GSEA pathway datasets: https://www.gsea‐msigdb.org/gsea/msigdb/mouse/genesets.jsp. All code used to perform the analysis as well as create the visualizations is available on Github: https://github.com/JustinBellavance/snRNA‐seq‐sex‐comparison‐spinal‐cord. References Associated Data Supplementary Materials Table S1. Figure S1. Data S1. Data S2. Data Availability Statement Tool web address to explore DEGs: https://justinbellavance.shinyapps.io/snRNA_Visualization/. NCBI accession numbers for the 3 datasets: Sathyamurthy: GSE103892; Alkaslasi: GSE167597; Blum: GSE161621. GSEA pathway datasets: https://www.gsea‐msigdb.org/gsea/msigdb/mouse/genesets.jsp. All code used to perform the analysis as well as create the visualizations is available on Github: https://github.com/JustinBellavance/snRNA‐seq‐sex‐comparison‐spinal‐cord.