Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 Oct 14;108(1):128-144.e9.
doi: 10.1016/j.neuron.2020.07.026. Epub 2020 Aug 17.

Transcriptional Reprogramming of Distinct Peripheral Sensory Neuron Subtypes after Axonal Injury

William Renthal  1 , Ivan Tochitsky  2 , Lite Yang  3 , Yung-Chih Cheng  4 , Emmy Li  5 , Riki Kawaguchi  6 , Daniel H Geschwind  6 , Clifford J Woolf  7
Affiliations

Transcriptional Reprogramming of Distinct Peripheral Sensory Neuron Subtypes after Axonal Injury

William Renthal et al. Neuron. .

Abstract

Primary somatosensory neurons are specialized to transmit specific types of sensory information through differences in cell size, myelination, and the expression of distinct receptors and ion channels, which together define their transcriptional and functional identity. By profiling sensory ganglia at single-cell resolution, we find that all somatosensory neuronal subtypes undergo a similar transcriptional response to peripheral nerve injury that both promotes axonal regeneration and suppresses cell identity. This transcriptional reprogramming, which is not observed in non-neuronal cells, resolves over a similar time course as target reinnervation and is associated with the restoration of original cell identity. Injury-induced transcriptional reprogramming requires ATF3, a transcription factor that is induced rapidly after injury and necessary for axonal regeneration and functional recovery. Our findings suggest that transcription factors induced early after peripheral nerve injury confer the cellular plasticity required for sensory neurons to transform into a regenerative state.

Keywords: ATF3; axon growth; cell identity; dorsal root ganglion; gene expression; nerve injury; regeneration; reprogramming; sensory neuron; single-cell RNA-seq.

PubMed Disclaimer

Conflict of interest statement

Declaration of Interests W.R. is a consultant for Kallyope. C.J.W. is a founder of Nocion Therapeutics and QurAlis.

Figures

Figure 1.
Figure 1.. Single-nucleus RNA sequencing of DRG neurons in mouse models of peripheral axotomy.
(A) Diagram of mouse axotomy models. (B-C) UMAP plots of 11,191 neuronal (B) and 5,668 non-neuronal (C) DRG nuclei from naive mice. (D) UMAP plots of Atf3, Sox11, or Sprr1a expression in DRG neurons after SpNT. Each time point is down-sampled to display 900 nuclei. Color denotes log2 expression of each gene. UMAP plots in last row display DRG neurons colored by neuronal subtype; injured state neurons (see Figure 1E, methods) are labeled “Injured.” (E) Bar plot showing the percent of SpNT nuclei within each neuronal cluster (top row) and violin plots showing log2 expression of injury-induced genes in each cluster (rows 2–4). Fractions were calculated out of 11,191 naive neuronal nuclei and 6,532 SpNT neuronal nuclei (> 1d after injury). Cluster ID corresponds to cluster number assignment from Seurat (Figure S2G). Injured state clusters (red) contain >95% nuclei from SpNT mice and have a median normalized Atf3 log2 expression >2, while all other clusters are naive state (green). (F) Percentage of naive, SpNT, and ScNT neuronal nuclei classified as injured state at each time point after the respective injury (see Figure 1E, methods). Data are mean±SD.
Figure 2.
Figure 2.. Loss of neuronal marker gene expression after peripheral nerve injury.
(A) UMAP plots of DRG neuronal subtype marker gene expression after SpNT. Colors are percentile of gene expression across naive and SpNT neuronal nuclei with counts > 0. Nuclei with marker gene expression <50th percentile for all 5 marker genes are gray; nuclei with expression ≥ 50th percentile for multiple markers have their colors overlaid (4.6%). 900 randomly sampled neuronal nuclei are displayed at each time point. Marker genes: Fam19a4 (cLTMR), Tac1 (PEP), Cd55 (NP), Nefh (Nefh+ A-LTMRs), Nppb (Sst+ pruriceptors). (B) Plot shows expression of neuronal subtype marker genes across naive and SpNT neuronal nuclei at time point after SpNT. Height of each block is the fraction of neuronal nuclei that express (>0 counts) a marker gene at each time point. Relative expression is row-normalized mean expression. (C-G) FISH images of L4 mouse DRGs stained with probes against an injury marker, Atf3 (red), a neuronal marker, Tubb3 (blue), and cell type markers: Th (C, green), Tac1 (D, green), Mrgprd (E, green), Hapln4 (F, green) and Sst (G, green). Representative sections from naive DRGs (left), DRGs 6h (middle) and 7d (right) after SpNT are shown. Scale bar=100μm. (H) Quantification of Atf3 and neuronal subtype marker gene expression from naive, 6h and 7d after SpNT measured by FISH (n = 3–6 L4 DRGs). Dots on the boxplot are gene expression (number of puncta/μm2) within a cell, boxes indicate quartiles and whiskers are 1.5-times the interquartile range (Q1-Q3). Median expression is a black line inside box. (1-way ANOVA, ***P<0.001, **P<0.01, *P<0.05, see methods for ANOVA parameters).
Figure 3.
Figure 3.. Classification of DRG neuronal subtypes after peripheral nerve injury.
(A) Classification of injured neuronal subtypes after SpNT. UMAP plots show 7,000 naive and 7,000 SpNT neurons randomly sampled. Nuclei in naive state are faint, injured state are bold. (B) UMAP plots of naive and SpNT DRG neuronal subtypes after pair-wise projection and clustering. Naive neurons at SpNT time points are shown (900 randomly-sampled neuronal nuclei per time point) and colored by neuronal subtype.
Figure 4.
Figure 4.. Characterization of CTS transcriptional responses to peripheral axotomy.
(A) Heatmap of the number of significantly induced genes in each cell type and time point after SpNT compared with respective naive cell types (FDR<0.01, log2FC>1). (B) Comparison of the overlap between injury-induced genes (FDR<0.01, log2FC>1; 3–7d after SpNT vs. naive) for each cell type. Squares colored by P-value of overlap between each pair-wise comparison (hypergeometric test). Comparisons between the same gene list have 100% overlap but different P-values depending on list size. (C) SpNT-induced gene expression changes within neuronal subtypes. Significantly upregulated genes after SpNT (FDR<0.01, log2FC>1 SpNT vs. naive) in each neuronal subtype were aggregated across time points and compared to other neuronal subtypes to determine the number of injury-induced genes that are CTS (red), shared between 2–4 neuronal subtypes (yellow), or shared between multiple ≥ 5 neuronal subtypes (green). The total number of significantly-induced genes by SpNT for each subtype is shown on each bar. See Tables S4 for gene lists. (D) Heatmap of genes induced by SpNT for each cell type over time. Rows are common genes (significantly upregulated by SpNT vs. naive in ≥ 5 neuronal subtypes) and CTS genes (significantly upregulated by SpNT vs. naive in only 1 cell type). Columns are cell types. Log2FC (SpNT vs. naive) of a gene at each time point and cell type is displayed. Genes in gray are not detected. (E) Regulation of CTS genes by SpNT for each cell type. CTS genes are expressed significantly higher in one naive cell type vs. all other naive cell types (see methods, Table S2). For each cell type, their CTS genes are grouped by log2FC after injury (SpNT at 3–7d vs. naive within each subtype). Pie charts show fraction of CTS genes within each neuronal subtype regulated by SpNT. Total numbers of CTS genes for each subtype are shown in headers. (F) UMAP plots of randomly sampled 7,000 naive and 7,000 SpNT neurons, colored by an injury score calculated from expression of 524 commonly induced genes after SpNT (left, Table S4) or a CTS score (right) calculated for each neuronal type based on their CTS genes (see 4E, Table S2). Higher scores indicate greater injury-induced or CTS gene expression.
Figure 5.
Figure 5.. Transcriptional reprogramming of DRG neurons after peripheral nerve injury.
(A) UMAP plots of naive L3–5 DRG neurons or neurons at different times after Crush. Each time point is randomly down-sampled to 1000 neuronal nuclei. Nuclei are colored by injury score (top) or CTS score (bottom), as in Figure 4F. Higher scores indicate greater injury-induced or CTS gene expression. (B) Percentage of neuronal nuclei classified as injured state at each time point after the respective injury, colored by injury model. SpNT and 6h-7d time points of ScNT are represented from Figure 1E for comparison. (C-E) FISH images of ipsilateral L4 Atf3-CreERT2;Gcamp6f DRG sections from a naive mouse (C), 7d after Crush (D) and 2mo after Crush (E), stained for Tubb3 (magenta), DAPI (blue) and Gcamp6f (green). The Atf3-driven Gcamp6f reporter is upregulated after Crush and persists for months after injury. Scale bar = 100μm. (F) Quantification of Gcamp6f reporter gene expression by FISH in L4 Atf3-CreERT2;Gcamp6f naive DRGs or DRGs from mice 7d or 2mo after Crush. N = 3–4 DRG sections from different mice per group, one-way ANOVA, F(2, 8) = 37.4, P = 8.7×10−5. Crush injury causes an increase in Gcamp6f reporter positive neurons 1w after Crush (Bonferroni post-hoc, P=2.9 × 10−4), which persists for 2mo after Crush injury (Bonferroni post-hoc, P=1.9 × 10−4). (G) Heatmap of the number of significant injury-induced genes for each cell type and time point after SpNT, Crush, ScNT, paclitaxel, or CFA compared to respective cell types in naive mice (FDR<0.01, log2FC>1). The smaller number of gene expression changes in Crush and ScNT compared to SpNT is primarily a consequence of the smaller fraction of axotomized neurons in the distal injury models than SpNT (see Figures S6C–D). (H) Pair-wise comparison of overlap between injury-induced genes in each cell type 3–7d after SpNT, Crush, ScNT, or paclitaxel or 2d after CFA (FDR<0.01, log2FC>1, compared to naive nuclei of respective cell type). Each square is colored by the P-value for the overlap between each comparison (hypergeometric test); P-values ≥0.05 are gray. (I) UMAP plots of DRG neuronal nuclei after each injury model, colored by injury model. Left, 3,000 nuclei randomly sampled equally from SpNT, Crush, and ScNT; middle, 3000 nuclei randomly sampled from naive; right, 2,000 nuclei randomly sampled equally from paclitaxel and CFA).
Figure 6.
Figure 6.. Induction of a common set of transcription factors across DRG neuronal subtypes after peripheral nerve injury.
(A) Diagram of the criteria used to identify transcription factors (TFs) involved in injury-induced transcriptional reprogramming. TFs are selected if they are upregulated rapidly after injury (6hr-1d) and have their TF binding motifs significantly enriched in the set of genes commonly upregulated 3–7d after injury across cell types. (B) Heatmap of log2FC (6h-1d after SpNT vs. naive for each cell type [columns]) for 9 TFs (rows) induced ≤1d after SpNT (FDR<0.01, log2FC>1, SpNT vs naive) in ≥5 neuronal subtypes and whose TF binding motifs are significantly enriched in the set of 524 common injury-induced genes 3–7d after SpNT (see Table S4). (C) Bar graph of the number of early injury-induced TF binding motifs present in the 524 genes commonly upregulated 3–7d after injury across cell types. (D) UMAP of 7,000 naive and 7,000 SpNT neuronal nuclei colored by degree of ATF3 regulon enrichment (left, AUCell score, see methods) or log2 expression of Atf3 (right). (E) Strategy to create Atf3 conditional knockout (cKO) mice. Slc17a6-Cre;Atf3f/f = cKO, Atf3f/f = WT. (F) Representative images of WT (top) or ATF3 cKO (bottom) L4 DRGs 7d after SpNT stained for ATF3 (green), DAPI (blue) and Nissl (red). There is a clear loss of ATF3 staining after SpNT in the cKO compared to WT. (G) Recovery of sensory function in WT and Atf3 cKO mice after sciatic nerve Crush. Pinprick responses of Atf3 cKO mice (n=14) show a significant delay in sensory recovery compared to WT mice (n=10) (2-way repeated measures between subjects ANOVA, F(1, 22)=33.7, P=7.7×10−6, Bonferroni post-hoc, *P<0.05, ***P<0.001).
Figure 7.
Figure 7.. Atf3 is required for injury-induced transcriptional reprogramming.
(A) UMAP plot of 8,777 Atf3f/f (WT) and 8,888 Slc17a6-Cre;Atf3f/f (cKO) DRG neurons from naive mice and mice 1.5d and 7d after Crush, colored by neuronal subtype. (B) UMAP plot of 8,777 WT and 8,888 Atf3 cKO DRG neurons from naive mice (left), 1.5d after Crush (middle), and 7d after sciatic rush (right), colored by genotype. Arrows point to injured state clusters (see Figure S7E). (C) Violin plot of ATF3 regulon enrichment (AUCell score) within individual neuronal nuclei. Neuronal nuclei are grouped by genotype (WT or cKO) and injury (naive, 1.5d after Crush, 7d after Crush). Lines in violin plots indicate lower or upper quartile and median. One-way ANOVA: F(5, 17659)=1042.55, P<0.001; Tukey HSD post-hoc testing P>0.05 for naive cKO vs. naive WT, P<0.001 for all other pair-wise comparisons. (D) Volcano plot displays differential expression (log2(fold change) x-axis, -log10(FDR) y-axis) between injured state Atf3 cKO and WT neuronal nuclei (injured state classified in Figure S7E) for 523 of the 524 common injury-induced genes (Table S4) that are also expressed in Atf3 WT and cKO mice. (E) Percent of neuronal nuclei classified as injured state in each condition (naive, 1.5d, and 7d after Crush) and genotype (WT or Atf3 cKO). There is a significant reduction in the fraction of injured state neurons in Atf3 cKO compared to WT and 7d after Crush (one-way ANOVA: F(3, 6)=105.22, P<0.001; Tukey HSD post-hoc testing P>0.05 for naive cKO vs. naive WT, P<0.01 for Crush WT vs Crush cKO, P<0.001 all other pairwise comparisons). (F) Box plots of CTS scores (blue) in naive mice and cKO mice 7d after Crush. Scores are normalized to the naive average. Boxes indicate quartiles of expression, and whiskers are 1.5-times the interquartile range (Q1-Q3). Median is the black line inside each box. ***P<0.001, two-tailed Student’s t-test. (G) Heatmap of the fold change of marker genes (rows) within respective cell types 7d after Crush vs. naive in either WT or Atf3 cKO mice (columns). Each marker gene is significantly less downregulated in Atf3 cKO 7d after Crush in than in WT mice (FDR<0.01).
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Abe N, and Cavalli V (2008). Nerve injury signaling. Current opinion in neurobiology 18, 276–283. - PMC - PubMed
    1. Aibar S, Gonzalez-Blas CB, Moerman T, Huynh-Thu VA, Imrichova H, Hulselmans G, Rambow F, Marine JC, Geurts P, Aerts J, et al. (2017). SCENIC: single-cell regulatory network inference and clustering. Nature methods 14, 1083–1086. - PMC - PubMed
    1. Alexa A, and Rahnenfuhrer J (2018). topGO: Enrichment Analysis for Gene Ontology. R package version 2.34.0
    1. Avraham O, Deng P-Y, Jones S, Kuruvilla R, Semenkovich CF, Klyachko VA, and Cavalli V (2020). Satellite glial cells promote regenerative growth in sensory neurons. bioRxiv, 2019.2012.2013.874669 - PMC - PubMed
    1. Bangash MA, Alles SRA, Santana-Varela S, Millet Q, Sikandar S, de Clauser L, Ter Heegde F, Habib AM, Pereira V, Sexton JE, et al. (2018). Distinct transcriptional responses of mouse sensory neurons in models of human chronic pain conditions. Wellcome open research 3, 78. - PMC - PubMed

Publication types

MeSH terms

Substances