Identification of Col20a1 as a specific marker for terminal Schwann cells (tSCs) in skeletal muscle.

(A) UMAP plot of the entire skeletal muscle-resident cell populations from a public scRNA-seq dataset (Giordani et al., 2019). (B) Feature plots showing the expression of canonical Schwann cell markers, Plp1 and S100b. (C) Sub-clustering of the isolated Schwann cell population reveals three distinct sub-populations (Clusters 1–3) (D) Dot plot displaying the expression profiles of established marker genes for myelinating Schwann cells (mSC, Cluster 1), Remak non-myelinating Schwann cells (Remak nmSC, Cluster 2), and tSC-expressing genes (Cluster 3). Node size indicates the percentage of cells expressing the gene, and color intensity represents the mean expression level. (E) Schematic illustration depicting the anatomical distribution of the three defined Schwann cell subtypes within the skeletal muscle and neuromuscular junction. (F) Venn diagram illustrating the intersection of two differential gene expression screening criteria: genes highly expressed in Cluster 3 compared to other SC subsets (15 genes; log2FC > 1, adjusted p-value < 0.05, pct.1 > 0.5), and genes specifically expressed in Cluster 3 compared to all other muscle cells (81 candidates; log2FC > 1, adjusted p-value < 0.05, pct.2 < 0.01). Three shared candidate genes (Col20a1, Acsbg1, and Fabp7) were identified. (G) Scatter plot comparing the Log2 fold-change (FC) of gene expression for Cluster 3 vs SC subsets (y-axis) and Cluster 3 vs whole muscle cells (x-axis). Genes that passed each screening criteria is depicted by red and blue dots, respectively. The intersected candidate genes are highlighted in orange circle. (H) Dot plot from the CELLxGENE database (Program et al., 2025) showing the cross-tissue expression profiles of the three candidate genes, highlighting the minimal expression of Col20a1 in the central nervous system compared to Acsbg1 and Fabp7.

Generation of the Col20a1-CreERT2 mouse model and validation of its highly specific, system-wide expression in tSCs.

Adult mice (8-12 weeks old) were treated with tamoxifen (TMX) via oral gavage (160 mg/kg body weight) for five consecutive days and analyzed one week after the initial dose. (A) Representative immunofluorescence images of EDL muscle neuromuscular junctions (NMJs) from control (Rosa-tdT) and Col20a1-CreERT2; Rosa-tdT mice. The tdTomato reporter signal (red) is robustly and exclusively detected at the NMJs of Col20a1-CreERT2; Rosa-tdT mice. Post-synaptic acetylcholine receptors (AChRs) are labeled with α-bungarotoxin (Btx, green), and glial cells are labeled with S100B (blue). (B) High-magnification confocal images highlighting the precise cellular localization of the tdTomato reporter. White arrowheads indicate the tight co-localization of tdTomato (red) with the tSC marker S100B (blue), which effectively envelops the post-synaptic AChR clusters (green). DAPI (white) staining indicates nuclei. (C) Representative immunofluorescence images showing the spatial distribution of the reporter across diverse anatomical locations, including cranial (masseter, sternomastoid), axial (diaphragm, psoas), and appendicular (EDL, soleus) muscles. The Col20a1-CreERT2; Rosa-tdT model reliably targets tSCs across all examined muscle types. Scale bars: 10 µm in all images.

Temporal mapping establishes the postnatal specificity window of the Col20a1-CreERT2 model.

(A) Experimental timeline for investigating the temporal dynamics of Col20a1-CreERT2 model. A single dose of tamoxifen (TMX, 160 mg/kg body weight) was administered in Col20a1-tdT mice via intraperitoneal injection at postnatal day 4 (P4), P7, P10, or P14, followed by analysis at 4 weeks (4w) of age. (B) Representative immunofluorescence images of NMJs corresponding to the temporal induction timeline. Early TMX induction at P4 and P7 results in tdTomato labeling (red) of both tSCs and portions of upstream axonal Schwann cells (aSCs, indicated by white arrowheads). In contrast, induction at P10 and P14 yields strictly tSC-restricted expression without off-target aSC labeling. S100B (blue); Btx (green). Scale bars: 10 µm.

Targeted tSC ablation during the postnatal maturation window efficiently depletes glial cells without disrupting gross NMJ architecture.

To induce tSC ablation, tamoxifen (TMX, 160 mg/kg body weight) was administered at postnatal day 10 (P10) via intraperitoneal injection, and NMJs were analyzed at 4 weeks of age. (A) Representative low-magnification immunofluorescence images of EDL muscle NMJs from Col20a1-tdT control (left) and Col20a1-tdT/DTA mice (middle and right). Compared to controls, Col20a1-tdT/DTA mice exhibit a profound loss of the tdTomato reporter signal (red). Yellow arrowheads indicate rare surviving “escaper” tSCs, and white dashed boxes outline the regions shown at higher magnification in B. Scale bars: 10 µm. (B) High-magnification confocal images of individual NMJs from Col20a1-tdT control (top) and tSC-ablated Col20a1-tdT/DTA mice (bottom). In Col20a1-tdT/DTA mice, the lack of tdTomato strictly correlates with the absence of the S100B glial marker (blue). Despite the profound loss of tSCs, pre-synaptic motor axons labeled by Neurofilament (NF, white) establish normal single innervation, and post-synaptic AChR clusters labeled by α-bungarotoxin (Btx, green) remain structurally intact. Scale bars: 10 µm. (C) Quantification of the tdTomato-positive (tdT+) NMJ ratio. The data confirm an approximately 80% ablation efficiency in Col20a1-tdT/DTA mice compared to controls. Data are presented as mean ± SD (n = 4 mice per experimental group). Statistical significance was determined using the Mann-Whitney test; *p < 0.05.

Systemic ablation of tSCs does not impair overall somatic growth or gross motor behavior.

(A, B) Body weight analysis of male and female control (Col20a1-tdT) and tSC-ablated (Col20a1-tdT/DTA) mice at (A) postnatal day 10 (P10, time of induction; control: n = 15 males, 9 females; ablated: n = 18 males, 17 females) and (B) 4 weeks of age (4w, post-ablation; control: n = 15 males, 10 females; ablated: n = 21 males, 17 females). Systemic loss of tSCs does not significantly alter normal growth trajectories in either sex. (C) Assessment of gross motor coordination using the beam walking test. The latency to cross the beam is comparable between the ablated mice and controls (control, n = 17 mice; ablated, n = 16 mice). (D, E) Quantitative gait analysis evaluating fine motor control. Systemic tSC ablation does not significantly affect (D) stride length or (E) stride width at 4 weeks of age (n = 25 mice per group). Data are presented as mean ± SD. Statistical significance was determined using one-way ANOVA with Bonferroni’s post hoc test for A and B, and Student’s t-test for C, D, and E.

Targeted tSC ablation impairs pre-synaptic vesicle homeostasis and compromises high-frequency synaptic transmission.

(A) Representative electrophysiological traces of miniature endplate potentials (mEPPs) recorded from isolated EDL muscles of control (Col20a1-tdT, top) and tSC-ablated (Col20a1-tdT/DTA, bottom) mice at 4 weeks of age. (B–E) Cumulative probability distributions and summary quantification of mEPP parameters. While mEPP amplitude (B, C) remains unchanged, mutant NMJs display a significant increase in mEPP frequency (D, E) compared to controls, indicating a “leaky” pre-synaptic spontaneous release (control, n = 10 NMJs from 3 mice; tSC-ablated, n = 14 NMJs from 5 mice). (F) Representative traces of single evoked endplate potentials (eEPPs). (G–J) Quantification of eEPP kinetics. Mutant NMJs exhibit significantly increased eEPP amplitudes (G) and accelerated rise times (H). There are no significant differences in decay time (I) or half-width (J), indicating intact post-synaptic receptor kinetics (control, n = 32 NMJs from 7 mice; tSC-ablated, n = 15 NMJs from 6 mice). (K) Representative traces of paired-pulse responses at an inter-stimulus interval of 10 ms. (L, M) Analysis of the paired-pulse ratio (PPR, EPP2/EPP1) across varying inter-stimulus intervals (L) and specifically at 10 ms (M). The PPR is comparable between groups, indicating a preserved basal release probability in the absence of tSCs (control, n = 11 NMJs from 3 mice; tSC-ablated, n = 13 NMJs from 5 mice). (N) Representative eEPP trains elicited by repetitive 20 Hz stimulation (20 pulses). (O) Normalized eEPP amplitudes during the 20 Hz stimulation train. Mutant NMJs exhibit a progressively severe synaptic depression compared to controls. (P, Q) Cumulative amplitude plot (P) and estimation of the readily releasable pool (RRP) size (Q) calculated via back-extrapolation of the cumulative eEPP amplitude. Despite the severe fatigue, absolute RRP size is not significantly altered (control, n = 18 NMJs from 4 mice; tSC-ablated, n = 25 NMJs from 5 mice). (R) Representative traces of Compound Muscle Action Potential (CMAP). (S) Quantification of CMAP amplitude, indicating normal macroscopic physiological function in mutant mice, consistent with behavioral data. Data are presented as mean ± SEM. Statistical significance was determined using Student’s t-test for C, E, G–J, M, Q, S, two-way ANOVA for L, and Repeated Measures ANOVA with Bonferroni’s multiple comparisons for O; *p < 0.05, **p < 0.01. ***p < 0.001.

Ultrastructural analysis reveals pre-synaptic vesicle depletion following targeted tSC ablation.

(A) Representative backscattered scanning electron microscopy (BSE-SEM) micrographs of ultra-thin sections of NMJs from control (Col20a1-tdT), mutant NMJs with retained tSCs (Col20a1-tdT/DTA; tSC-retained), and mutant NMJs with ablated tSCs (Col20a1-tdT/DTA; tSC-ablated). High-magnification views of the regions indicated by pink dashed boxes are shown on the right, highlighting the ultrastructural details of synaptic vesicles, junctional folds, and the synaptic cleft. M, muscle; N, nerve terminal; JF, junctional folds; tSC, terminal Schwann cell. Scale bars: 1 µm. (B, C) Quantitative analysis of (B) synaptic cleft width and (C) junctional fold density. Both parameters remain unchanged across all experimental groups, indicating that overall synaptic physical integrity and post-synaptic architecture are preserved despite tSC loss. (D) Quantification of synaptic vesicle density. The tSC-ablated NMJs exhibit a significant reduction in vesicle density compared to both control and tSC-retained groups, providing ultrastructural evidence for the loss of pre-synaptic homeostasis. Data are presented as mean ± SD (control, n = 17 NMJs from 5 mice; mutant NMJs with retained tSCs, n = 8 NMJs from 4 mice; mutant NMJs with ablated tSCs, n = 11 NMJs from 3 mice). Statistical significance was determined using one-way ANOVA with Bonferroni’s post hoc test; **p < 0.01.

Long-term repopulation of Schwann cells maintains structural integrity and gross neuromuscular function.

To assess long-term recovery, tamoxifen (TMX, 160 mg/kg body weight) was administered at postnatal day 10 (P10) by intraperitoneal injection, and mice were analyzed at 12 weeks of age. (A) Representative immunofluorescence images of muscle NMJs from Col20a1-tdT control (left) and tSC-ablated Col20a1-tdT/DTA mice (right) at 12 weeks post-induction. White dashed boxes indicate regions magnified in B. Scale bars: 10 µm. (B) High-magnification confocal images of individual NMJs from Col20a1-tdT (top) and Col20a1-tdT/DTA (bottom) mice. In the mutant NMJs, S100B+ glial cells (blue) successfully repopulate the synapse despite the persistent absence of the tdTomato reporter (red). The gross architecture of pre-synaptic motor axons (NF, white) and post-synaptic AChRs (Btx, green) is sustained following repopulation. Scale bars: 10 µm. (C) Representative traces of Compound Muscle Action Potential (CMAP). (D) Quantification of CMAP amplitude, indicating sustained physiological function in tSC-ablated mice (control, n = 12 mice; ablated, n = 16 mice). Data are presented as mean ± SEM. (E) Body weight analysis of male and female mice at 12 weeks of age (control: n = 8 males, 8 females; ablated: n = 12 males, 9 females). Data are presented as mean ± SD. (F–I) Macroscopic motor behavioral assessments showing normal long-term function in mutant mice, including (F) normalized grip strength (control, n = 16 mice; ablated, n = 21 mice), (G motor coordination assessed by latency to fall in the rotarod test (control, n = 15 mice; ablated, n = 21 mice), and endurance running capacity measured by (H) running time and (I) running distance (control, n = 16 mice; ablated, n = 21 mice). Data are presented as mean ± SD. Statistical significance was determined using one-way ANOVA with Bonferroni’s post hoc test for E and Student’s t-test for D and F–I; **p < 0.01.

Expression profiling of canonical cell-type markers in the skeletal muscle-resident cell population.

(A) UMAP feature plots showing the distribution and expression levels of established marker genes for diverse cell types within the whole skeletal muscle scRNA-seq dataset.

Validation of Schwann cell sub-cluster identities using established subtype-specific marker genes.

(A) UMAP feature plots and (B) Violin plots illustrating the distinct expression patterns of known Schwann cell subtype markers across the three isolated sub-clusters.

Highly restricted expression of the three final candidate genes within the tSC cluster.

UMAP feature plots illustrating the localized expression patterns of the three selected candidate genes (Col20a1, Acsbg1, and Fabp7). The plots highlight their highly specific enrichment in the tSC population when mapped across (A) the isolated Schwann cell sub-clusters and (B) the entire skeletal muscle-resident cell population.

Independent bulk RNA-seq analyses confirm Col20a1 as the most robust tSC-specific marker.

Re-analysis of publicly available bulk RNA-seq datasets validates the enrichment of the candidate genes in tSCs. (A) Bar graphs showing expression levels (CPM) of the three candidate genes, Col20a1, Acsbg1, and Fabp7, in FACS-sorted S100b-GFP/NG2-dsRed double-positive tSCs compared to S100b-GFP or NG2-dsRed single positive cells (Castro et al., 2020). (B) Bar graphs showing expression levels (TPM) of the three candidate genes in micro-dissected NMJ regions (NMJ) compared to non-NMJ regions (xNMJ) (Ham et al., 2020). (C, D) Volcano plots indicating the statistical significance (–Log10 adjusted p-value) and Log2FC of the candidate genes derived from the datasets in (A) and (B), respectively. For dataset (A), the volcano plot in (C) displays the differential expression between double-positive tSCs and the combined single-positive cell populations. Col20a1 exhibits the highest expression level and the most significant fold-change among the candidates, prioritizing it for the genetic model generation.

Generation strategy and quantitative flow cytometry validation of the Col20a1-CreERT2 mouse model.

(A) Schematic diagram illustrating the design of the Col20a1-CreERT2 knock-in allele. A P2A-CreERT2 cassette was inserted at the C-terminus of the endogenous Col20a1 gene via homologous recombination (HR). (B) Flow cytometry gating strategy for isolating mononuclear cells from whole skeletal muscle. Cells were separated into Lineage-positive (Lin+; CD31+, CD45+ or Ter119+), muscle stem cells (MuSC; Lin- /VCAM1+/Sca-1-), fibro-adipogenic progenitors (FAP; Lin-/VCAM1-/Sca-1+), and double-negative cells (DN; Lin-/VCAM1-/Sca-1-), the latter of which harbors the Schwann cell population. (C) Representative flow cytometry plots displaying tdTomato fluorescence across the four isolated cell populations. (D, E) Flow cytometry analysis (D) and quantification (E) of the tdTomato-positive (tdT+) cell distribution. The vast majority of tdT+ cells (∼92%) are highly restricted to the DN population. Data are presented as mean ± SD (n = 4). Statistical significance was determined using Repeated Measures ANOVA with Bonferroni’s post hoc test; ***p < 0.001, ****p < 0.0001. (F) PCR validation of genomic DNA extracted from the sorted cell populations. The recombined tdTomato amplicon (blue arrow) is exclusively detected in the tdT+ fraction of the DN population. M: DNA size marker.