Sox11 overexpression restores embryonic pro-growth transcription in mature corticospinal tract neurons Authors: Elizabeth Batsel, Zimei Wang, Elizabeth Otten, Syed Roshan, Rabia Mohammad, Paula Pascual, Darby O’Shea, Jose Rosas, Pantelis Tsoulfas, Murray G. Blackmore Published: 2025 ABSTRACT Neurons in the central nervous system (CNS) display a high capacity for axon growth during early development but lose this ability at a pivotal differentiation stage marked by synaptic maturation, circuit integration, and a profound shift in gene transcription. Once mature, most CNS neurons fail to reverse this transcriptional switch after axon injury, fundamentally constraining their intrinsic capacity for axon regeneration. Here, we show with single-nucleus RNA sequencing that forced expression of the transcription factor Sox11 in mature corticospinal tract (CST) neurons produces large-scale and stable changes in gene expression that are highly enriched for growth-relevant processes, and which strongly resemble those of pre-synaptic embryonic stages. Moreover, Sox11 is equally effective when delivered to chronically injured CST neurons. These data reveal Sox11’s ability to reverse a critical step of neuronal maturation even in otherwise unperturbed neurons, clarifying the transcriptional underpinnings and highlighting Sox11 as a potent regulator of pro-regenerative gene networks.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-33479-5. Introduction A recurring observation in developmental and regenerative biology is that as cells progress through maturation, they often lose the capacity for regeneration. This tradeoff is particularly striking in the mammalian central nervous system (CNS), where embryonic and neonatal neurons display a high capacity for axon regeneration that is lost during early development1–4. Neuronal maturation is a staged process including neurogenesis, migration, axon elongation, axon and dendritic arborization, and prolonged synapse refinement4. Of these, the transition from axon elongation to arborization and synapse formation is of particular interest, as it appears to coincide with the pronounced reduction in the capacity for long-distance regenerative axon growth4,5. Transcriptional comparisons of embryonic neurons in the elongation stage to more mature counterparts have revealed pronounced changes with increasing age including strong upregulation of transcripts involved in ion transport and synaptic vesicle cycling and downregulation of transcripts involved in cytoskeletal regulation, cell adhesion, and axonal transport6–10. These transcriptional changes are likely major drivers of the maturational loss of regenerative axon growth ability, making their reversal an important objective in pro-regenerative research. Multiple lines of evidence suggest the transcription factor Sox11 is a central regulator of axon growth ability. Developmentally, Sox11 is widely expressed in CNS and peripheral nervous system (PNS) neurons during the stage of axon elongation, is sharply downregulated in postnatal periods around the transition to arborization/synaptogenesis, and has been shown in loss-of-function experiments to be necessary for embryonic axon elongation11–14. Sox11 is also strongly linked to regenerative axon growth. For example, mammalian dorsal root ganglion (mDRG) neurons and zebrafish retinal ganglion cells (zRGCs), which display a high capacity for spontaneous axon regeneration, strongly upregulate Sox11 in response to axon damage15–19. Mammalian retinal ganglion cells (mRGCs), which display minimal spontaneous regeneration but robust axon elongation in response to molecular interventions such as PTEN knockdown, also spontaneously upregulate Sox11 after injury20,21. Loss of Sox11 interferes with mDRG peripheral regeneration and almost completely blocks mRGC regeneration stimulated by PTEN knockdown, demonstrating a critical contribution of Sox11 re-expression to the regenerative process22,23. Conversely, forced elevation of Sox11 in mDRG and mRGC neurons after axon damage triggers upregulation of regeneration-associated genes such as doublecortin (Dcx) and downregulation of transcripts involved in synapse function, leading to enhanced regenerative axon growth24,25. Despite these links to injury-triggered responses, it is unknown whether Sox11 possesses the ability to induce expression of regeneration-associated genes and/or trigger embryonic reversion under stable, homeostatic conditions. This question is of critical importance for regenerative efforts that focus on neuronal subtypes which normally display smaller responses to axon damage and for neurons in a chronic injury state after the initial injury response has subsided. For example, we recently showed that corticospinal tract (CST) neurons, an important sensorimotor population with low regenerative capacity, undergo only minor transcriptional changes after spinal injury, likely owing to the large distance between injury and cell body26. This raises the question of how transcriptional stimuli like elevated Sox11 might affect transcription in a less injury-perturbed context. Here we employed single-nucleus sequencing to determine the transcriptional consequences of forced expression of Sox11 in CST neurons. We show in both uninjured and chronically injured CST neurons that Sox11 triggers a wide-spread transcriptional response marked by upregulation of cytoskeletal-interacting proteins previously linked to axon growth and down-regulation of transcripts involved in synapse function. In addition, Sox11-responsive transcripts displayed high overlap with transcripts up- or down-regulated during CST maturation, indicating a partial embryonic reversion. Sox11-responsive transcripts also overlapped with regenerating neurons and epidermal cells after forced Sox11, suggesting a conserved function for Sox11 across diverse cell types. These results highlight the centrality of Sox11 in neuronal maturation and supporting its potential utility as a pro-regenerative stimulus. Results Effects of Sox11 overexpression on CST transcription Our prior studies showed that forced expression of Sox11 in corticospinal tract (CST) neurons enhances axon arborization after pyramidotomy and spinal transection injuries27. Sox11-treated axons were functionally integrated, as evidenced by post-synaptic responses to CST stimulation in the cervical spinal cord, but produced no improvement forelimb function28. The molecular basis for Sox11’s growth effects is unknown. We first used single-nucleus sequencing to determine the transcriptional effects of forced expression of Sox11 in adult CST neurons in homeostatic, uninjured conditions. Following previously established procedures, AAV2-Retro-Sox11 or titer-matched AAV2-Retro-Malat-barcode (AAV control see Methods) was injected into the cervical spinal cord of adult male and female mice along with nuclear localized H2B-mGreenLantern (H2B-mGl) to facilitate subsequent nuclear isolation26. To track transcriptional changes through time, Sox11-treated animals were divided into four cohorts that received unique molecular barcodes, and which were sacrificed at one, two, four, or six weeks after injection (n = 20). Cortical tissue from at least three animals per time point was collected by fluorescent microdissection, frozen, and then pooled for unified purification of CST nuclei by fluorescence activated nuclei sorting (FANS) and single-nucleus library preparation using a 10X Genomics platform (Supplementary Fig. 1–1). Expression of Sox11 was confirmed by immunohistochemistry in one randomly selected animal from each time point (Supplementary Fig. 1–2). The resulting pooled data were merged with a separate library derived from control-treated CST nuclei (n = 7). De-multiplexing of the four barcoded time points was performed in Seurat, allowing each to be compared separately to control-treated nuclei and to one another (Fig. 1a). Fig. 1Single-nucleus profiling detects a large transcriptional response to Sox11 overexpression. (a) Experimental design in which CST neurons of adult male and female mice were retrogradely labeled by nuclear-localized AAV2-Retro-H2B-mGreenlantern, AAV2-Retro-Sox11, and a unique molecular barcode. Fluorescently labeled tissue from each group was dissected after one, two, four, or six weeks and pooled for single-nucleus profiling. (b) UMAP visualization shows separation of Sox11-treated nuclei from control and co-clustering of all Sox11 time points. (c) MA plots showing the relationship between transcript abundance and differential gene expression at the four time points, comparing Sox11-treated to control nuclei (non-parametric Wilcoxon rank sum test, p_adj < 0.05), which reveal numerous transcripts are significantly up-regulated (green) and down-regulated (red). (d) Gene ontology analysis of transcripts that are significantly upregulated in Sox11-treated nuclei shows enrichment for terms such as neuron development and cytoskeletal organization (e) Gene ontology analysis of transcripts that are significantly downregulated in Sox11-treated nuclei shows enrichment for terms including ion transport and synaptic signaling (p-values indicated in plots, Benjamini-Hochberg test). Single-nucleus profiling detects a large transcriptional response to Sox11 overexpression. (a) Experimental design in which CST neurons of adult male and female mice were retrogradely labeled by nuclear-localized AAV2-Retro-H2B-mGreenlantern, AAV2-Retro-Sox11, and a unique molecular barcode. Fluorescently labeled tissue from each group was dissected after one, two, four, or six weeks and pooled for single-nucleus profiling. (b) UMAP visualization shows separation of Sox11-treated nuclei from control and co-clustering of all Sox11 time points. (c) MA plots showing the relationship between transcript abundance and differential gene expression at the four time points, comparing Sox11-treated to control nuclei (non-parametric Wilcoxon rank sum test, p_adj < 0.05), which reveal numerous transcripts are significantly up-regulated (green) and down-regulated (red). (d) Gene ontology analysis of transcripts that are significantly upregulated in Sox11-treated nuclei shows enrichment for terms such as neuron development and cytoskeletal organization (e) Gene ontology analysis of transcripts that are significantly downregulated in Sox11-treated nuclei shows enrichment for terms including ion transport and synaptic signaling (p-values indicated in plots, Benjamini-Hochberg test). UMAP clustering produced two main clusters, one containing control nuclei and the second containing Sox11-treated nuclei from all four time points (Fig. 1b). The separation of Sox11 from control suggests a distinct transcriptional state, while the co-clustering of Sox11 groups suggests relative similarity through time. Indeed, this visual impression was confirmed by differential gene (DEG) analysis. Using a standard threshold of two-fold or greater change (Log2FC > 1, p_adj < 0.05), Sox11-upregulated transcripts numbered 788 by one week, climbed to 958 by two weeks, and remained at comparable levels at four and six weeks (965 and 985, respectively) (Fig. 1c, Supplementary Tables 1–1). Downregulated transcripts followed a similar pattern, numbering 473 by one week, 771 by two weeks, and 787 and 697 at four and six weeks, respectively (Fig. 1c, Supplementary Tables 1–1). Two-way comparisons between Sox11 time points revealed some differences between one week and later time points (maximum 179 DEGs between weeks one and six) but very small differences between two weeks and later time points (maximum 15 DEGs between two and six weeks) (Supplementary Fig. 1–3, Supplementary Tables 1–1). We also performed a replicate profiling experiment in a new cohort of animals, this time testing only the two-week time point, and found a similar Sox11 of gene expression (R2 = 0.8497, linear regression with the original 2-week Log2FC values) (Supplementary Fig. 1–4, Supplementary Tables 1–1). In summary, forced elevation of Sox11 produces large-scale transcriptional changes in uninjured CST neurons that are largely established and stable by two weeks post-transduction. We next performed gene ontology (GO) enrichment analysis on transcripts up- and down-regulated by Sox11. Top biological processes enriched in Sox11-upregulated transcripts included neuron development, axon development, and cytoskeleton organization (Fig. 1d, Supplementary Tables 1–2). Strongly upregulated transcripts included Dpysl3, Dpysl5, Tubb2b, Tuba1a, Marcksl1, Dcx, and Ezh2, which have been functionally linked to axon growth (Fig. 1c, Supplementary Tables 1–1)29–32. Conversely, Sox11-downregulated showed enrichment for terms related to synapse function and the regulation of membrane potential, processes that have been previously linked to the maturational decline in axon growth ability5,8 (Fig. 1e, Supplementary Tables 1–2). For example, the downregulated set included Slc12a5 (Kcc2), and Cacna2d2, which have been linked to both neuronal excitability and to axon regeneration8,33. Fluorescent in situ hybridization on a second cohort of animals confirmed strong upregulation of Dpysl3, Dpysl5, and Sox11 in animals that received AAV2-Retro-Sox11 (Fig. 2) (n = 6). Combined, these results show Sox11’s transcriptional effects to include an elevation of transcripts linked to axon extension and a diminishment of synapse-related transcripts. Fig. 2Fluorescence in situ hybridization confirms upregulation of Sox11, Dpysl3, and Dpysl5 in Sox11-treated animals. (a) Experimental design involving retrograde transduction of CST neurons with nuclear-localized AAV2-Retro-H2B-mGreenlantern (H2B-mGl) alone or in combination with AAV2-Retro-Sox11, followed 2 weeks later by in situ hybridization to detect candidate genes. (b–g) Coronal sections of cortex with CST nuclei labeled by H2B-mGl (green) and in situ hybridization detection of Sox11 (b, c), Dpysl3 (d, e), and Dpysl5 (f, g) in red. Panels (b, d, f) show control tissue and (c, e, g) show Sox11-overexpressing samples with elevated target expression (arrows). (h–j) Quantification of average signal intensity in CST nuclei shows significant increases in Sox11 (h), Dpysl3 (i), and Dpysl5 (j). N > 50 neurons from three replicate animals, ****p < 0.0001, unpaired t-test. Fluorescence in situ hybridization confirms upregulation of Sox11, Dpysl3, and Dpysl5 in Sox11-treated animals. (a) Experimental design involving retrograde transduction of CST neurons with nuclear-localized AAV2-Retro-H2B-mGreenlantern (H2B-mGl) alone or in combination with AAV2-Retro-Sox11, followed 2 weeks later by in situ hybridization to detect candidate genes. (b–g) Coronal sections of cortex with CST nuclei labeled by H2B-mGl (green) and in situ hybridization detection of Sox11 (b, c), Dpysl3 (d, e), and Dpysl5 (f, g) in red. Panels (b, d, f) show control tissue and (c, e, g) show Sox11-overexpressing samples with elevated target expression (arrows). (h–j) Quantification of average signal intensity in CST nuclei shows significant increases in Sox11 (h), Dpysl3 (i), and Dpysl5 (j). N > 50 neurons from three replicate animals, ****p < 0.0001, unpaired t-test. Sox11’s effect on transcription in a chronic injury model Sox11’s efficacy as a potential pro-regenerative treatment requires an ability to influence transcription in neurons after injury, which could present a unique cellular context. Based on eventual clinical feasibility we focused on the sub-acute to early chronic phase of injury. Adult mice received a contusion injury to the right cervical spinal cord, severing the corticospinal tract. Four weeks after injury, animals received an injection of H2B-mGl and AAV2-Retro-Sox11 or matched AAV-Retro-Malat-barcode control into the cervical spinal cord just above the injury site (n = 16). Two weeks later, the left (injured) cortex was harvested, CST nuclei were isolated by FANS, and single-nucleus analyses performed as described above (Fig. 3a)26. Immunohistochemistry for glial fibrillary acidic protein (GFAP) on spinal sections confirmed interruption of the right dorsal column (Fig. 3b). In addition, immunohistochemistry for Sox11 performed on one randomly selected treatment and control animal confirmed effective overexpression (Supp. Figure 3− 1). Fig. 3Single-nuclei profiling detects a large transcriptional response to Sox11 overexpression after a chronic contusive spinal injury. (a) Experimental timeline. Adult mice received a unilateral contusion at cervical level five followed one month by retrograde transduction of CST neurons with nuclear-localized AAV2-Retro-mGreenlantern (H2B-mGI) alone or in combination with AAV2-Retro-Sox11. Two weeks later, brains and spinal cords were dissected. (b) Shows a sagittal section of a spinal cord stained for GFAP (red) to confirm interruption of the CST tract in dorsal spinal cord (c) UMAP visualization of Uninjured/Control (grey), Uninjured/Sox11 (pink), Injured/Sox11 (purple), and Injured/Control (maroon) nuclei indicates separation by gene treatment but not by injury state (d) A MA plot of transcripts differentially expressed after Sox11 expression in chronically injured CST neurons shows numerous genes significantly upregulated (green) and downregulated (red) (non-parametric Wilcoxon rank sum test, p_adj-value < 0.05). (e) Linear regression of Log2 fold changes triggered in uninjured versus chronically injured CST neurons reveals strong positive correlation (R-squared = 0.7849, slope 0.99). Single-nuclei profiling detects a large transcriptional response to Sox11 overexpression after a chronic contusive spinal injury. (a) Experimental timeline. Adult mice received a unilateral contusion at cervical level five followed one month by retrograde transduction of CST neurons with nuclear-localized AAV2-Retro-mGreenlantern (H2B-mGI) alone or in combination with AAV2-Retro-Sox11. Two weeks later, brains and spinal cords were dissected. (b) Shows a sagittal section of a spinal cord stained for GFAP (red) to confirm interruption of the CST tract in dorsal spinal cord (c) UMAP visualization of Uninjured/Control (grey), Uninjured/Sox11 (pink), Injured/Sox11 (purple), and Injured/Control (maroon) nuclei indicates separation by gene treatment but not by injury state (d) A MA plot of transcripts differentially expressed after Sox11 expression in chronically injured CST neurons shows numerous genes significantly upregulated (green) and downregulated (red) (non-parametric Wilcoxon rank sum test, p_adj-value < 0.05). (e) Linear regression of Log2 fold changes triggered in uninjured versus chronically injured CST neurons reveals strong positive correlation (R-squared = 0.7849, slope 0.99). The resulting data were merged in Seurat with the prior sets from non-injured CST, resulting in a unified, four-group object consisting of non-injured/AAV-control, non-injured/AAV-Sox11, injured/AAV-control and injured/AAV-Sox11. UMAP visualization produced two clusters, one containing AAV-control nuclei from both injured and uninjured animals and the other containing AAV-Sox11 nuclei, also of mixed injury status (Fig. 3c). Thus, nuclei segregated by gene treatment but not injury status, indicating a dominant Sox11 effect regardless of injury state. Notably, growth-relevant transcripts Marcksl1, Dcx, Ezh2, Tubb2b, Dpysl3 and Dpysl5 were upregulated similarly upon Sox11 overexpression in injured CST neurons (Fig. 3d, Supplementary Tables 3 − 1). Indeed, linear regression analysis of Sox11 DEGs in uninjured versus injured CST neurons showed a striking correlation (R-squared = 0.7849, slope 0.99) (Fig. 3e). These results show that Sox11’s ability to drive large-scale transcriptional changes in CST neurons is maintained when delivered four weeks after spinal injury. Conserved activation of cytoskeletal and axon development networks across cell types We further contextualized our results by comparing them to prior profiles of other cell types responding to forced expression of Sox11 and to neurons responding to injury. First, to create a final, high-stringency list of Sox11-responsive transcripts we performed additional DEG detection by aggregating nuclei in the prior control and Sox11 libraries (“pseudobulks”), followed by statistical comparison via DeSeq2. This approach identified 3524 DEGs, split about equally between up- and down-regulation (Supplementary Fig. 4 − 1, Supplementary Tables 4 − 1). Transcripts that showed both significant changes of > 50% in all three of the individual single-nucleus comparisons and significant regulation by pseudobulk analysis were selected as definitively Sox11-responsive, resulting in final sets of 846 Sox11-upregulated and 693 Sox11-downregulated transcripts. We first considered murine retinal ganglion cells (mRGCs) that were pre-treated with Sox11 and challenged by optic nerve injury and compared to RGCs that received injury alone24. Like the present study, Sox11 was delivered by AAV about two weeks prior to transcriptional profiling, resulting in genome-wide detection of Sox11-responsive DEGs. Supplementary Tables 4 − 2 reports for each gene symbol the fold change and adjusted p-value in mRGCs side-by-side with the corresponding Sox11 effect in CST neurons, facilitating direct comparison. Downregulated DEGs showed no significant overlap (1.06 enrichment p = 1.0, hypergeometric test used for all overlap tests hereafter), Supplementary Tables 4 − 2), indicating that Sox11 caused distinct transcripts to decline in each cell type. In contrast, intersection of upregulated DEGs revealed 95 in common, more than expected by chance (2.47 enrichment, p = 7.71 × 10− 16) (Fig. 4a, Supplementary Tables 4 − 2). Gene ontology analysis of these upregulated transcripts showed neuron projection development and axon development among the most enriched terms (Fig. 4c, Supplementary Tables 4 − 3), while protein-protein interaction (PPI) analysis revealed a central network highly enriched for cytoskeletal regulation and axon development (Fig. 4e). These included Dcx, L1cam, and Stmn2 (SCG10), previously linked to axon growth30,31,34,35 (Supplementary Tables 4 − 2). Thus, although Sox11’s effects in CST are mostly distinct from those previously reported in RGCs, they include a subset commonly upregulated subset involved in cytoskeletal regulation and axon growth. Fig. 4Sox11 activated conserved growth-related networks across cell types. (a) A Venn diagram illustrating the overlap of upregulated genes in CST neurons and retinal ganglion cells (RGCs) following Sox11 overexpression, highlighting both unique and shared transcriptional responses. (b) A Venn diagram shows the overlap of upregulated genes in CST neurons and epidermal cells after Sox11 overexpression, also indicating unique and shared transcriptional responses. P-values were determined by hypergeometric tests. (c, d) Gene Ontology (GO) enrichment analysis for overlapping genes from (a) and (b). (e) An interaction protein network (String) of the genes shared among CST and RGCs after Sox11 overexpression reveals two distinct functional categories: (1) Cytoskeletal Organization and (2) Axon Development. (f) String network of the genes shared among CST and epidermal cells after Sox11 overexpression reveals three distinct functional categories (1) Cytoskeletal Organization, (2) Axon Development, and (3) Chromatin Organization. Each network node represents a gene, with edge represents a known interaction. Sox11 activated conserved growth-related networks across cell types. (a) A Venn diagram illustrating the overlap of upregulated genes in CST neurons and retinal ganglion cells (RGCs) following Sox11 overexpression, highlighting both unique and shared transcriptional responses. (b) A Venn diagram shows the overlap of upregulated genes in CST neurons and epidermal cells after Sox11 overexpression, also indicating unique and shared transcriptional responses. P-values were determined by hypergeometric tests. (c, d) Gene Ontology (GO) enrichment analysis for overlapping genes from (a) and (b). (e) An interaction protein network (String) of the genes shared among CST and RGCs after Sox11 overexpression reveals two distinct functional categories: (1) Cytoskeletal Organization and (2) Axon Development. (f) String network of the genes shared among CST and epidermal cells after Sox11 overexpression reveals three distinct functional categories (1) Cytoskeletal Organization, (2) Axon Development, and (3) Chromatin Organization. Each network node represents a gene, with edge represents a known interaction. We next considered a study conducted in murine epidermal cells, in which forced expression of Sox11 was shown to enhance wound healing and cell motility and to cause up- and downregulation of 746 and 272 genes, respectively36. Like the mRGC comparison, downregulated transcripts were largely distinct, with only 26 overlapping. In contrast, upregulated sets had 204 transcripts in common (5.18 enrichment, p = 7.15 × 10− 93) (Fig. 4b, Supplementary Tables 4 − 2). Reminiscent of the mRGC overlap, GO and PPI analysis showed high enrichment of terms related to cytoskeleton and axon development and a core protein network with similar terms and which included various tubulin isoforms, Marcks, Marcksl1, and Dpysl5 (Fig. 4d, f, Supplementary Tables 4 − 2). Finally, we intersected the Sox11-responsive gene sets with DEGs previously reported in regenerating neurons: mouse DRGs (mDRGs) after sciatic nerve transection17, zebrafish RGCs (zRGCs) after optic nerve crush19, or mouse CST neurons after spinal injury and transplantation of neural progenitor cells (CST-NPCs)37. Log2FC and adjusted p-values reported in the original publications were extracted and added to Supplementary Tables 4 − 2 (see methods). It should be noted that Sox11 itself was reported as strongly injury-regulated in the mDRG and zRGC data but not in the CST-NPC. As previously, we detected minimal overlap with downregulated genes in any of the injured cell type (Supplementary Tables 4 − 2). There was also no significant overlap between the Sox11-upregulated and CST-NPC-upregulated genes (59 shared, 1.30 enrichment, p = 0.13), and GO analysis detected no enrichment for any terms in the shared set (Supplementary Fig. 4-2c, f). Thus, Sox11 response genes in CST neurons are distinct from injury-downregulated genes and are also distinct from a previously reported response to injury and NPC transplantation. In contrast, Sox11-upregulated genes overlapped significantly with mDRG-upregulated (163 common, 1.86 enrichment, p = 6.39 × 10− 15) and zRGC-upregulated (158 shared, 1.77 enrichment, p = 1.77 × 10− 12). Interestingly, GO analysis of both overlapping sets again placed neuron projection development and axon development among the most enriched terms (Supplementary Fig. 4-2d, e, Supplementary Tables 4 − 3), mirroring the prior overlap with Sox11-treated RGCs and epidermal cells. Combined, these data suggest that activation of transcripts related to cytoskeleton-based motility and axon extension is a conserved function of Sox11across diverse cell types and even species. Sox11 partially recapitulates an embryonic-like pattern of gene expression Finally, we probed for similarity between Sox11’s transcriptional effects in CST neurons and transcriptional changes that occur naturally during maturation. The available description of CST maturation is based on microarray technology and includes maximum cell age of only postnatal day 146. Therefore, to facilitate more direct comparison with our Sox11 profiling, we performed single-nucleus analysis of mouse cortex at embryonic day 18 (E18), a time when cortical projection neurons including CST are actively elongating axons38 (n = 6). We prepared two replicate libraries from three embryos each and then used a Seurat-based pipeline to extract nuclei from deep-layer glutamatergic projection neurons from the initially mixed-cortical data. This entailed the identification and removal of off-target cell types including mitotically active progenitors (Pax6/Top2a), oligodendrocyte precursor cells (Sox10/Pdgfra), radial glial cells (Slc1a3/ EAAT), inhibitory neurons (Gad2), and shallow layer glutamatergic neurons (Cux2) (Supp. Figure 5−1). In the remaining nuclei, Bcl11b was used as a positive marker to select deep layer (DL) projection neurons39. The resulting data were merged with the prior adult CST data, visualized by UMAP, and compared to identify DEGs (Fig. 5a). Fig. 5Single-nucleus profiling detects large changes in transcription associated with cortical neuronal maturation. (a) Experimental design showing comparison between adult CST and embryonic cortical tissue. (b) UMAP visualization shows separation of adult (orange) and embryonic (blue) nuclei (c, d) Feature plots of Camk2a (c) and Tubb2b (d) expression confirm expected enrichment in adult and embryonic nuclei, respectively. in adult versus embryonic nuclei. (e) MA plot showing transcripts significantly more abundant in E18 (green) or in adult CST (red) (2-fold threshold, p-value < 0.05, non-parametric Wilcoxon rank sum test, p_adj < 0.05). (f, g) Gene ontology analysis indicating biological processes that are significantly enriched in E18 and adult gene sets, respectively (p-value < 0.5, Benjamini-Hochberg test). Single-nucleus profiling detects large changes in transcription associated with cortical neuronal maturation. (a) Experimental design showing comparison between adult CST and embryonic cortical tissue. (b) UMAP visualization shows separation of adult (orange) and embryonic (blue) nuclei (c, d) Feature plots of Camk2a (c) and Tubb2b (d) expression confirm expected enrichment in adult and embryonic nuclei, respectively. in adult versus embryonic nuclei. (e) MA plot showing transcripts significantly more abundant in E18 (green) or in adult CST (red) (2-fold threshold, p-value < 0.05, non-parametric Wilcoxon rank sum test, p_adj < 0.05). (f, g) Gene ontology analysis indicating biological processes that are significantly enriched in E18 and adult gene sets, respectively (p-value < 0.5, Benjamini-Hochberg test). Adult CST and embryonic DL nuclei clustered distinctly by UMAP (Fig. 5b). For initial validation, we examined transcripts known to be developmentally regulated and observed the expected concentration of Camk2a and Tubb2b in the adult and embryonic clusters, respectively (Fig. 5c, d). DEG analysis identified 1,198 genes more than 2-fold more abundant in embryonic nuclei (hereafter embryonic transcripts) and 1842 that were more abundant in adult (hereafter adult transcripts) (Fig. 5e, Supplementary Tables 5 − 1). For additional validation we identified numerous strongly regulated DEGs with concordance to known patterns of developmental up- or down-regulation (indicator transcripts labeled in Fig. 5e), and FISH confirmed Dpysl3, Dpysl5, and Sox11 to be abundant in E18 cortex but low or absent in the adult (Supplementary Fig. 5 − 2). GO analysis indicated the enrichment of terms including neuron development and axon development for the embryonic transcripts and regulation of membrane potential and synaptic signaling for the adult transcripts (Fig. 5f, g, Supplementary Tables 5 − 2). Moreover, algorithmic prediction of transcriptional regulation (IPA) yielded Sox11 as a top candidate to regulate the set of embryonic transcripts (Activation Z-score 5.23, p = 4.3 × 10− 13, Fisher’s Exact Test) (Supplementary Fig. 5 − 3, Supplementary Tables 5 − 2). These data identify broad transcriptional differences between cortical projection neurons engaged in embryonic axon extension and fully mature CST neurons and provide initial support for Sox11’s involvement in mediating this difference. We next compared these maturational changes to the previously characterized effects of Sox11 expression in adult CST neurons. Manual inspection revealed many of the previously noted Sox11-responsive transcripts. For example, Dpysl3, Dpysl5, Dcx, Ezh2, Marcksl1, and Tubb2b, all of which are Sox11-upregulated and linked to axon growth, were also among the most differentially expressed embryonic transcripts (Fig. 5e, Supplementary Tables 5 − 1). Intersection of Sox11-upregulated and embryonic DEGs revealed a striking overlap; 396 transcripts were shared, more than six times the number predicted by chance (6.25 enrichment, p = 4.38 × 10− 233) (Fig. 6a, Supplementary Tables 6 − 1). Linear regression of this shared set showed a strong positive correlation between fold changes induced by Sox11 and differential expression at E18 (R-squared 0.779), further supporting a potential role for Sox11 in regulating this subset of embryonic transcripts (Fig. 6c). Similarly, intersection of Sox11-downregulated transcripts and adult-DEGs also revealed strong overlap (469, 5.89 enrichment, p = 4.05 × 10− 287) (Fig. 6b, Supplementary Tables 5 − 1), although with weaker fold-change correlation (R-squared = 0.237) (Fig. 6d). Finally, GO analysis indicated enrichment of terms related to neuron development and axon development in the shared Sox11-Up/Embryonic set and synaptic transmission and membrane potential in the shared Sox11-Down/Adult set (Fig. 6e, f; Supplementary Tables 6 − 1). Thus, forced expression of Sox11 partially restores in adult CST neurons an embryonic pattern of transcription. Fig. 6Sox11 partially restores an embryonic pattern of gene expression. (a-b) Venn diagrams showing the overlap of Sox11-upregulated (a) and downregulated (b) transcripts with embryonic and adult gene sets, respectively (p-values reflect hypergeometric tests). (c, d) Scatter plots comparing Sox11-triggered fold changes versus developmental fold changes, with linear regression analysis indicating positive correlation. (e, f) Gene ontology analysis showed biological processes that are significantly enriched in the Sox11-up/E18 overlap (e) and in the Sox11-down/Adult overlap (f) (Log2FC, ± 1 p-value indicates Benjamini-Hochberg test). Sox11 partially restores an embryonic pattern of gene expression. (a-b) Venn diagrams showing the overlap of Sox11-upregulated (a) and downregulated (b) transcripts with embryonic and adult gene sets, respectively (p-values reflect hypergeometric tests). (c, d) Scatter plots comparing Sox11-triggered fold changes versus developmental fold changes, with linear regression analysis indicating positive correlation. (e, f) Gene ontology analysis showed biological processes that are significantly enriched in the Sox11-up/E18 overlap (e) and in the Sox11-down/Adult overlap (f) (Log2FC, ± 1 p-value indicates Benjamini-Hochberg test). Discussion We have provided new data that clarify the transcriptional underpinnings of a key step in neuronal maturation marked by an apparent tradeoff between axon growth and synaptic activity. Remarkably, we find that forced expression of Sox11 in fully mature neurons can substantially reverse this transcriptional shift, simultaneously elevating numerous transcripts functionally involved in axon growth and suppressing transcripts linked to synapse function and excitability. Sox11’s effects are large in scale, stable through time, and occur even when Sox11 is delivered to chronically injured neurons. Together, these results demonstrate that developmental growth programs can be re-engaged in adult neurons and point toward a regenerative strategy that bridges developmental biology and pro-regenerative transcriptional programs after injury. The single-nucleus approach used here provides updated information regarding changes in transcription that accompany corticospinal tract (CST) maturation. A prior study, widely cited, used microarray technology to compare murine cortical neurons between ages embryonic day 18 (E18) and postnatal day 14 (P14)6,37,40. Key methodological differences in the current data include the use of fully mature CST neurons and the use of snRNA-seq technology. Comparison of DEGs detected by the two approaches found high statistical concordance (3.8-fold enrichment, p-1.39 × 10-75) and broad agreement in GO enrichment analysis. Notably, the current snRNA-seq approach detected more than 1000 additional DEGs above a two-fold threshold that were either not present or not significantly called as DEGs in the prior microarray data. Thus, the snRNA-seq approach across full maturation provides an expanded view of developmental DEGs. An important caveat, however, is that in the present study the E18 data derives from a mixed population of glutamatergic deep-layer projection neurons and not from purified CST as previously. Thus, the present data are best suited to inform understanding of cellular behaviors that are broadly shared across glutamatergic cell types at E18, including axon projection, but are likely less sensitive to features that are highly specific to CST neurons, which comprise only a subset of the selected E18 nuclei. Our developmental comparison largely supports the hypothesis of maturational tradeoff between early transcriptional support for axonal extension and later support for synapse function5,41–44. GO enrichment analysis showed abundant axon-related terms in the embryonic gene set, switching to enrichment for synapse-related terms in the adult. This maturational tradeoff is further substantiated by numerous indicator transcripts. Notable examples include Dpysl3 (CRMP4), Dpysl5 (CRMP5), Stmn2 (SCG10), L1cam, Dcx, Marckls1, and Ncam and its associated sialtransferases St8sia2 and St8sia4, all with previously established links to effective axon elongation via cytoskeletal and/or adhesive processes, and which are highly expressed at E18 and sharply decline with maturation29–31,34,35,45–48. Conversely, we detected developmental upregulation of transcripts with canonical synapse functions such as Snap25, Stx1b, Vamp1, Syt1, and Syt2. Also prominent in developmentally upregulated transcripts were transcripts related to maturational changes in membrane potential and excitability such as Slc12a5 (Kcc2), Atp1a1, Kcnj6, Scn8a (Nav1.6), and Knca1. Thus, these data help clarify the transcriptional profiles that distinguish neurons in a stage of axon elongation from neurons in a stage of mature synaptic transmission. Regarding Sox11’s effects in CST neurons, perhaps the most striking feature is its magnitude, with more than 1500 transcripts reliably increased or decreased more than 2-fold. These changes likely reflect the effects of Sox11 acting within CST cell nuclei, although it is possible that indirect effects from Sox11 expression in cells at the site of spinal injection may have contributed. This response overlapped only partially with a previous report of Sox11’s effects in injured RGCs24. This may reflect methodological differences (e.g. snRNAseq versus bulk sequencing previously) but also likely involves the pre-delivery baseline. In the case of RGCs, Sox11 was delivered to cells already undergoing a profound injury response, including strong upregulation of endogenous Sox11 even prior to experimental intervention21,49. In contrast, here we delivered Sox11 to neurons with largely unperturbed transcription and very low baseline Sox11 expression26. Therefore, the RGC outcome likely reflects only the additive component of exogenous Sox11 atop an elevated baseline, while the current data reveal Sox11’s full effect as it shifts the cell from initial transcriptional homeostasis. The large effect in otherwise unperturbed neurons may relate to Sox11’s established “pioneer” ability to bind nucleosome-bound DNA and locally increase chromatin accessibility50. We also found Sox11 to drive upregulation of chromatin remodelers involved in transcriptional activation, including Smarcc1 and Smarcc2, members of the SWI/SNF complex that are known Sox11 targets (Table 4 − 151.). The large number of downregulated transcripts seemed initially paradoxical given Sox11’s understood function as a transcriptional activator13,52 but may also involve chromatin remodeling. For example, Sox11-upregulated transcripts included the methyltransferases K5m5b, shown previously to repress synaptic plasticity and memory-associated transcripts53, and Ezh2, shown previously to inhibit promoters of synaptic genes54. In addition, Sox11 also caused upregulation of numerous transcription factors known to repress target genes, with notable examples including Rest, Zeb2, Mnt, Bach1, and multiple Zbtb family repressors (Table 4 − 1). Thus, indirect chromatin and transcriptional repression may explain the widespread gene downregulation. More generally, we speculate that Sox11’s strong transcriptional effects are likely accompanied by widespread epigenetic modifications; future assays of chromatin accessibility and methylation can directly test this hypothesis. Sox11’s effects in CST neurons overlapped strongly with developmental changes, with many hundreds of shared genes, high statistical significance, and common enrichment for GO functional terms. Sox11-upregulated genes included numerous examples that are (1) functionally linked to axon growth (2) strongly down-regulated during maturation and (3) up-regulated in Sox11-expressing neurons. These include Dpysl3 (Crmp4), Dpysl5 (Crmp5), Stmn2 (SCG10), L1cam, Dcx, Marckls1, and Ncam and its associated sialtransferases St8sia2 and St8sia429–31,34,35,45–48. Interestingly, Sox11 also triggered downregulation of adult transcripts involved in synaptic release and maturational changes in excitability, including transcripts previously linked to axon growth: Slc12a5 (Kccn2)33, voltage-gated calcium channels, Cacna2d28 and synaptic vesicle priming genes, Munc13s5. Thus, Sox11 possess the ability to broadly reverse developmental shifts in gene expression, including transcripts with established roles in axon growth. A third outcome of interest is the conservation of Sox11’s effects across cell types. Comparison to regenerating neurons and Sox11-expressing RGCs and epidermal cells revealed common Sox11-responsive networks involved in cytoskeletal regulation and cell projection development (Fig. 4, Supplementary Fig. 4 − 2). Regarding the overlap of Sox11 effects in CST and epidermal cells, neuronal and epidermal cells share an ectodermal origin in early embryogenesis, so the overlap in neural development genes may offer an intriguing hint of de- or trans-differentiation activity by Sox11 in epidermal cells. Sox11 has long been recognized as a member of a small set of transcription factors whose upregulation is nearly universal in regeneration-competent neurons across species and the CNS/PNS divide55. Our comparative analysis helps explain this observation, supporting the hypothesis that Sox11 may drive a conserved gene module for cytoskeletal organization during axon growth. This module did not appear in a prior description of the transcriptional response of CST neurons to spinal injury, however, which may reflect technical differences37or the engagement of a distinct growth program . Our data also reveal several Sox11 properties of high translational relevance. In a prior study we found that forced expression of Sox11 in CST neurons had no effect on cell viability, based on the absence of markers for cell death and no decrease in CST cell number after Sox11 delivery27. Consistent with this, but unlike the prior report in RGCs24, the current results found no change in cell counts or upregulation of apoptotic-associated functions at any time point sampled. Thus, rather than driving progressive gene changes or triggering cell death, Sox11 appears to move CST neurons into a transcriptional state that is dramatically shifted from baseline but relatively stable. Sox11 displays the same ability when delivered four weeks post-injury, positioning Sox11 as a tool to potentially prime neurons for growth even in the chronic injury state. Importantly, however, our results also highlight a seeming discrepancy between the large effect on gene transcription and our prior findings of more modest growth promotion by Sox11 in CST neurons27. Growth-inhibitory cues in the damaged spinal cord, which were not addressed in prior studies, may have masked Sox11-triggered gains in intrinsic growth potential, making combinations of Sox11 with strategies to alleviate cell-extrinsic inhibition an attractive future direction [56,57]. In addition, despite Sox11’s strong effects on some growth-relevant gene modules, especially cytoskeletal, there are likely other cellular processes needed for full axon extension that are outside of Sox11 regulation. For example, transcripts involved in translation and ribogenesis were highly enriched in E18 neurons but missing from Sox11-responsive transcripts. Identifying growth-relevant gene modules that fall outside of Sox11 regulation can help narrow the search for additional gene regulators that are most likely to complement and synergize with Sox11. In summary, the data provide detailed insight into Sox11’s capacity to profoundly shift neuronal transcription in adult CST neurons into a stable pattern with striking resemblance to a prior growth-competent, embryonic stage. Importantly, Sox11 displays this ability in CST neurons that normally show only muted response to axon damage and in chronically injured neurons. These findings highlight the centrality of Sox11 in setting the transcriptional balance between developmental stages and re-focuses attention on Sox11-based stimulation as an influential component of pro-regenerative molecular strategies. Methods Experimental model and study participant details Male and female (6–8 weeks old, 20–28 g) wild-type (C57BL/6J, The Jackson Laboratory) mice and day 18.5 mouse embryos were the experimental units for this study. In each experiment, n refers to the total number of mice used. All procedures involving animals were conducted in strict adherence to ethical guidelines provided by the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH) and are reported in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines to ensure transparent and reproducible reporting. The Institutional Animal Care and Use Committee (IACUC) at Marquette University reviewed and approved all animal experimental protocols (approval numbers 3283, 4013). During conduction of experiments animal care staff were blinded to treatment groups such that all animals were handled, monitored and treated in the same way. Housing conditions for the mice included ad libitum access to food and water, a controlled 24-hour cycle of 12 h of light followed by 12 h of darkness, temperature of 22 °C ± 2 °C, and relative humidity between 40% and 60%. In all experiments exclusion criteria were based on the intensity of retrograde fluorescence observed during tissue dissection; animals with very dim or absent signal were excluded. AAV preparation AAV2-Retro-CAG-H2B-mGreenLantern was produced by the University of Miami Viral Vector Core using Addgene plasmid #177332. AAV2-Retro-CAG-Sox11 was produced by the University of North Carolina Viral Vector Core using plasmid in which H2B-mGl was replaced by DNA corresponding to the open reading frame of mouse Sox11 (accession NM_009234), synthesized by Genscript. AAV2-Retro-Malat barcode constructs were produced as previously described in26. Briefly, randomly generated sequences were synthesized and annealed to an RNA nuclear localization sequence derived from human Malat RNA and placed downstream of a CAG promoter. In addition to the previously described barcodes BC0, BC1, and BC102, a fourth barcode designated BC3 was constructed by replacing the BC1 sequence with randomly generated nucleotides (AAACAAGGGTCACGAGTTAAG). Barcode AAV2-Retro was produced by the University of Miami Viral Vector Core. All viruses were diluted with 1X phosphate buffered saline (PBS) (Electron Microscopy Sciences, Ref. 15710) and a protective solution of 0.001–0.01% P188 (ThermoFisher 24040032) (and 5% D-Sorbitol (Sigma, S1876) immediately before injection. Control animals received AAV2-Retro-H2B-mGl (2 × 1012 particles/ml with AAV2-Retro-BC3 (5 × 1012) and Sox11 animals received AAV-Retro-H2B-mGl and AAV-Retro-Sox11 at the same titer; Sox11 animals in the time course experiment received an additional barcode construct at 2 × 1012 particles/ml). Animal surgeries For spinal AAV injection, adult female and male C57BL/6 mice (6–8 weeks old, 20–28 g) were anesthetized by ketamine/xylazine (10 mg/kg), the vertebral column was surgically exposed, mounted on a custom stabilization device, and laminectomy was performed at cervical level 5 (C5). AAV2-Retro particles (0.8 µl) were injected in the right C5, 0.33 mm lateral to the midline, and to depths of 0.6 and 0.9 mm, via a pulled capillary tube fitted to a Hamilton syringe and driven by a Stoelting QSI pump (catalog #53311) and guided by a micromanipulator (pumping rate: 0.06 µl /min). For spinal cord contusion injuries, adult mice were anesthetized by ketamine/ xylazine (10 mg/kg). Mice were mounted on the custom stabilization device, and a laminectomy was performed at C5. After laminectomy, all mice received a 60 kilodyne injury at the midline of C5 using an infinite horizon impactor device (Precision Systems and Instrumentation, Ref. IH-0400). Immediately following injury, mice were placed on a heating pad and received a subcutaneous injection of meloxicam (5 mg/kg) once a day for two consecutive days. Embryo dissection Timed pregnancies between female and male C57BL/6 mice produced E18 embryos. Pregnant mothers were anesthetized by exposure to isoflurane and embryos were removed and placed in ice-cold 1x PBS. The cortex of each embryo was dissected and washed in ice-cold Ca+/Mg+-free HBSS (Gibco 14175) and promptly flash frozen on dry ice in a 1.5mL Eppendorf DNA LoBind Microcentrifuge tube. The samples were then stored at -80 °C until nuclear extraction and fluorescently activated nuclei sorting (FANs, BD FACs Melody). Before FANs, the nuclei were labeled with propidium iodide (Sigma, P4864, 1:500 dilution). Tissue collection for single nucleus analysis Following previously published procedures26, two weeks after AAV injection, mice were euthanized for analysis. Euthanasia involved placement into a bell jar with isoflurane until animals completely lacked response to toe pinch, tail pinch, and ocular reflexes, followed by immediate decapitation. Brains were rapidly placed in ice-cold artificial cerebrospinal fluid (ACSF)(Hearing et al., 2013) for 1 min. On ice, brains were then sectioned in the horizontal plane at 500 microns using an Adult Mouse Brain Slice Matrix (Zivic Instruments BSMAS005). Retrogradely labelled tissue from the left and right cortex was then micro-dissected in ACSF using a stereomicroscope and fluorescence adapter (NIGHTSEA SFA-GR). The dissected tissue was flash-frozen on dry ice in a 1.5mL Eppendorf DNA LoBind Microcentrifuge tube. The samples were then stored at -80 °C until nuclear extraction. Blinding was not used in single nucleus sequencing experiments because all sample processing and data analyses were performed using standardized protocols and automated computational pipelines, reducing the potential for bias. Nuclear extraction For extraction of nuclei, frozen tissue was promptly transferred from − 80 °C storage to a chilled 2mL dounce (KIMBLE, Ref. DWK885300-0000) with chilled Nuclei EZ Lysis Buffer (Sigma-Aldrich N 3408). The sample was dounced 25 times with pestle A and 20 times with pestle B. The dounced sample was then transferred to a chilled 15mL conical tube and an additional 2mL of Nuclei EZ Lysis Buffer was added and gently mixed by inversion. The sample was incubated on ice for 5 min and centrifuged at 500G at 4 °C. The supernatant was removed, and the pellet was resuspended in 4mL of Nuclei EZ Lysis Buffer. The sample was incubated on ice and centrifuged at 500G at 4 °C for 5 min. Following the centrifugation, the supernatant was removed, and the pellet resuspended in 750 µl of Nuclei Suspension Buffer (NSB) (2% Bovine Serum Albumin (BSA, Sigma-Aldrich Ref. A9418), 40U/ µl RNase Inhibitor (Invitrogen Ref. AM2684), 1X PBS). For embryonic samples, which lacked AAV-derived nuclear mGl fluorescence propidium, iodide (Sigma, P4864, 1:500 dilution) was added to NSB to facilitate sorting. FANs and library preparation After resuspension in NSB the nuclei were passed through a 20µm filter (CellTrics 040042325) and quickly moved to the FANs machine. The collection tube for the sorted nuclei was coated with 5% BSA and contained the 10X Genomics Reverse Transcription Master Mix. Single nuclei were separated from debris and doublets using sequential gating of side scatter area (SSC-A) vs forward scatter area (FSC-A), forward scatter width (FSC-W) vs forward scatter area (FSC-A), and side scatter width (SSC-W) vs side scatter height (SSC-H), followed by fluorescence gating to gather H2B-mGl positive nuclei. With run times capped at seven minutes, between 5000 and 700 nuclei were sorted directly into the Reverse Transcription Master Mix and single-nucleus libraries were prepared using Chromium Next GEM Single Cell 3’ Reagent Kits v3.1 (PN-1000269). Sequencing and data analysis Sequencing was performed by Illumina NovaSeq X Plus at UW-Madison Biotechnology Center. Fastq files were analyzed with the CellRanger v6.1.0 pipeline on 10X Genomics Cloud Analysis with alignment to reference genome GRCm39 and with default parameters to produce a Unique Molecular Identifier for all nuclei-containing droplets. The filtered_feature_bc_matrix output file was then uploaded for analysis in Seurat 5.026. Quality control steps included exclusion of nuclei with fewer than 1000 detected transcripts (nFeature_RNA < 1000) and exclusion of nuclei with more than 0.5% mitochondrial RNA. The total number of included nuclei and the basis and number of nuclei that were excluded from each library is provided in Supplementary Tables 1–1. Exclusion averaged 3.6% for adult CST and 21.5% for E18 libraries, which had a larger fraction of low-count nuclei. Normalization, dimensionality reduction, and cell clustering were performed in Seurat. Cell clusters were identified by performing principal component analysis (PCA) based on 30 principal components, with significant components determined by elbow plot and jackstraw test. Nuclei were clustered using the Louvain method in the FindClusters function and visualized with Uniform Manifold Approximation and Projection (UMAP). Data merging was performed using Seurat’s integration workflow (FindIntegrationFeatures, FindIntegrationAnchors, IntegrateData), which performs batch correction by identifying anchors across datasets and computing correction vectors. All dimensional reduction and clustering were performed on the batch-corrected integrated assay. To confirm deep-layer cortical projection neurons, markers Sox10/Pdgfra were used to remove oligodendrocyte precursor cells, and Pax6/Top2a was used to remove mitotically active neural stem cells. Next, Slc1a3 (EAAT) was used to remove radial glial cells, Gad2 to remove cortical inhibitory neurons, and Cux2 to remove neurons in the shallow layer. Lastly, Bcl11b was used to identify and confirm deep-layer cortical projection neurons, thereby arriving at a final population. The remaining nuclei were re-clustered and plotted via UMAP. Cluster plots in Seurat were generated using the “scCustomize” package in R (https://github.com/samuel-marsh/scCustomize). Differential gene expression analysis was conducted using the “FindMarkers” function in Seurat, applying a non-parametric Wilcoxon rank-sum test. For additional pseudobulk analysis, Seurat objects corresponding to three replicate Sox11-treated samples and three replicate control samples were processed by summing counts across all objects, removing pseudogenes (“Gm”) and genes with fewer than 50 total counts, creating a unified pseudobulk count matrix, and performing DEG detection via DeSeq2 (Wald test) [58]. MA-plots were created using the “ggplot2” package (https://ggplot2.tidyverse.org/). Analysis details can be found in publicly available scripts on GitHub, below. Gene ontology and network analysis Gene ontology analysis was conducted using the “ClusterProfiler” (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3339379/) package in R 64,65, with input genes showing a Log2 fold change value ≥ 1 or ≤-1 and an adjusted p-value < 0.05, marking them as upregulated and downregulated, respectively. The “org.Mm.eg.db” database served as the reference for ontology searches. GO enrichment analysis was performed using a background set (“universe” in enrichGO command) comprised only of transcripts detected in the present experiments and, for GO analysis of intersection with external datasets, any additional transcripts present in the comparison gene lists. The analysis included Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) categories, setting gene count size limits between 3 and 1500, and a p-value cutoff of < 0.05. The top 15 pathways were visualized using the dot plot function from the enrichR package in R, and the results were saved as a CSV file. Intersection of Sox11-responsive gene sets with prior transcriptional profiling Log2FC and and p-values for mRGCs responding to Sox11 expression were taken directly from Table S1 in Norsworthy et al. (2017)24. values for epidermal cells responding to Sox11 expression were taken directly from Table 3a in Miao et al. (2019)36. values for mDRGs responding to sciatic nerve transection were extracted from Supplemental Table 3 of Renthal et al. (2020)17, using the largest significant change reported for any DRG cell type within one week of injury. Values for zRGC responding to injury were supplied by Dr. Veldman in Excel format based on data reported in Hu et al. (2025)19 and transformed into mouse gene symbols using the Ensembl conversion tool (https://www.ensembl.org/biomart) and direct name correspondence. Values for mCST responding to spinal injury and neural progenitor cell transplantation were taken directly from Table 5 of Poplawski et al. (2020)37 using values in the “logFC_14day_regenerating_Vs_intact” and “PValue_14day_regenerating_Vs_intact” columns. Selection of differentially expressed genes in all cases was based on adjusted p values < 0.05. Statistical tests of overlap were performed using hypergeometric tests in R, with p-values corrected for six multiple comparisons (five listed above plus one comparison to E18 CST). Ingenuity pathway analysis (IPA; QIAGEN Inc., https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis/) was used to analyze differentially expressed genes to identify upstream regulators. A list of differentially expressed genes, including official gene symbols, log₂ fold changes, and adjusted p-values, was uploaded into IPA. Only genes meeting an adjusted p-value < 0.05 and a log₂ fold change ≥ 2 were included in the analysis. IPA maps the input gene list to its curated knowledge base of molecular interactions, biological functions, and regulatory networks. Upstream regulators were inferred based on known relationships between transcriptional regulators and their downstream targets, with activation or inhibition predicted by the calculated z-score. Pathways and regulators with an absolute z-score ≥ 2 and p-value < 0.05 were considered significantly activated. Fluorescent in situ Hybridization: Fluorescent in situ hybridization was performed using RNAscope Multiplex Fluorescent Detection Kit v2 (Product #323100) from ACDbio. Mice were anesthetized by C02 exposure, decapitated, brains removed, frozen in O.C.T. Compound (Tissue-Tek, Ref. 4583), and cryosectioned at 30 μm. Cryosection slides were dried at room temperature for one hour before storage at -80 °C. Fluorescent in situ hybridization was performed following the RNAscope protocol, as follows. Immediately after removal from − 80 °C slides were fixed in fresh 4% paraformaldehyde (PFA) for one hour at room temperature and washed with 1X PBS twice for two minutes each. The slides were then dehydrated in 50%, 70%, and twice in 100% ethanol, five minutes each. Slides were allowed to dry at room temperature, followed by addition of hydrogen peroxide for ten minutes. Slides were then washed with nuclease-free water, twice for two minutes each. Protease III was then applied for eight minutes at room temperature and washed with 1X PBS twice for two minutes each. Probes were then applied for two hours at 40 °C and washed with 1X wash buffer twice for two minutes each. Probes used were Sox11 (842061), Dpysl3 (496611), and Dpysl5 (1112131-C2). The three consecutive amplification steps were then performed, washed with wash buffer in between, as instructed by the manufacturer. The corresponding probe HRP signal was then developed. All probes were detected using a 1:2000 dilution of TSA Vivid Fluorophore 650 (PN 323273). Slides were then dried overnight and were imaged with AR1 + Confocal Microscope and 60x Apochromat Oil DIC N2 objective within two weeks from completion. Histology and immunohistochemistry Mice were anesthetized by exposure to CO2, perfused with 4% PFA in 1× PBS, brains and spinal cords were removed, and post-fixed overnight in 4% PFA. Brains and spinal cords were embedded in 6% gelatin in 1× PBS (G2500, Sigma-Aldrich, St. Louis, MO). Brains were sliced in the coronal orientation and spinal cords sagittal orientation with a vibratome to yield 100 μm sections. Sections were incubated overnight with primary antibodies Sox11 (Abcam, Ref. Ab134107), GFAP (Invitrogen, Ref. 13-030-0), rinsed, and then incubated for 2 h with appropriate Alexa Fluor-conjugated secondary antibodies (R37117, Thermo Fisher, Waltham, MA, 1:500). Slices were then rinsed, mounted, and imaged using a AR1 + Confocal Microscope and 20x Apochromat objective. 20x images were gathered in stacked z-plane. Image quantification Quantitative comparison of signal intensity between control and Sox11-treated samples was performed only on sections processed in the same batch, with acquisition parameters that were identical for all samples within the experiment, and by a blinded observer. Quantification of pixel intensities was performed in ImageJ, first creating regions of interest corresponding to CST cell nuclei based on thresholded signal in the mGl channel and then measuring average pixel values for each region based on the probe channel. Average intensities were background-corrected using the average pixel intensity of four randomly selected regions of interest within tissue but outside of mGl-positive layer five cortex. Nuclei that exceeded a threshold set by the average background intensity plus ten standard deviations of the background intensity were defined as positive for the probe. Quantification and statistical analysis Statistical details of each experiment can be found in figure legends. Quantitative image analysis was analyzed by one way ANOVA with multiple comparisons by post-hoc Dunnett’s. For snRNA-Seq experiments the number of experimental replicates and animals pooled per replicate are indicated in Results and statistical difference in gene expression was determined by non-parametric Wilcoxon rank sum tests performed within Seurat or by DeSeq with default Wald parameters. Statistical thresholding for all differential gene analysis was based on adjusted p values. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5 Supplementary Material 6 Supplementary Material 6 Supplementary Material 7 Supplementary Material 7 Supplementary Material 8 Supplementary Material 8 Supplementary Material 9 Supplementary Material 9 Supplementary Material 10 Supplementary Material 10 Supplementary Material 11 Supplementary Material 11