All experiments were performed on young adult (3-5 months of age) mice of both sexes. Mice were fed chow ad libitum and had continuous access to drinking water. Mice were housed in a central animal facility and were maintained pre- and postoperatively in strict accordance with the National Institutes of Health guidelines and according to protocols approved by the institutional animal research ethics committee (IACUC) at Washington University School of Medicine in St Louis, MO, United States.

Wildtype (WT) mice (C57BL/6J; Stock No. 000664), purchased from the Jackson Laboratory (Bar Harbor, ME) and S100-GFP transgenic mice, in which the S100B promoter drives GFP expression in all glial cells (Magill et al., 2007), were bred in our laboratory. Tbet-KO constitutive knockout mice (kindly provided by Dr. Wayne Yokoyama, Washington University, St. Louis, MO) have been previously described (Szabo et al., 2000). Transgenic mice were genotyped by the TAG Center of Transnetyx (Cordova, TN).

For functional and morphological assessments, a sciatic nerve injury model was used. WT and Tbet-KO mice were anesthetized with intraperitoneal ketamine (50 mg/kg) and dexmedetomidine (0.5 mg/kg) injections. Buprenorphine-SR (1.0 mg/kg) and sterile saline were administered for pain control and hydration, respectively. After anesthesia, mice were kept warm via use of a heating pad. The skin of the right hindlimb was incised 2 mm posterior and parallel to the femur under a standard operating microscope. The sciatic nerve was located and sharply transected 5 mm proximal to the trifurcation. Immediately following transection, the nerve was repaired using microsuture and fibrin glue (Tessel, Baxter, Deerfield, IL) (Vannucci et al., 2019; Jablonka-Shariff et al., 2019). The skin was closed with 6-0 nylon in a simple interrupted fashion. Anesthesia was reversed with Antisedan (0.5 mg/kg), and animals were recovered on a heating pad.

For NMJ-specific RNA expression analyses after nerve injury, the spinal accessory nerve model in S100-GFP mice was utilized as the sternomastoid (SM) muscles in the neck have dense, well-defined NMJ clusters. This anatomy allowed separation of centrally located tSC-enriched NMJ band from non-NMJ muscle tissue under a fluorescent dissecting stereomicroscope (Jablonka-Shariff et al., 2021). After anesthesia performed as described above, a cut was made to expose sternomastoid muscle and the spinal accessory nerve, which passes caudally through this muscle. The transected nerve ends were immediately repaired with fibrin glue, and skin was closed with nylon (6–0). For recovery, mice received anesthetic reversal as described above.

Tetanic muscle force testing of the injured hindlimb was performed to determine the functional impact of Tbx21 loss on NMJ reinnervation after sciatic nerve transection and acute repair. The extensor digitorum longus (EDL) muscle from WT or Tbet-KO mice was used as it has a long tendon, which is required to facilitate in vivo attachment of muscle force analysis equipment. Animals were immobilized on an automated functional station (FASt System, red Rock Laboratories, St. Louis, MO). Muscle force testing was performed by electrically stimulating the sciatic nerve and measuring the EDL contraction force utilizing the Red Rocks Laboratory program as described previously (Santosa et al., 2013; Snyder-Warwick et al., 2018). The peak tetanic force was acquired by determining the optimal stimulation amplitude, muscle length, and stimulation frequency for each mouse. Maximum isometric tetanic force was automatically calculated from the resulting sets of recorded force traces. Following assessment, animals were euthanized, the dissected EDL muscle was weighed, and the tetanic specific muscle force (N/cm2) was calculated by dividing the absolute muscle force by the physiological muscle cross-sectional area. A naïve WT mouse was always tested first to establish assay baseline. The tetanic force was assessed at 3-, 4- and 6-week, and 6 months post-sciatic nerve injury. A total of N = 3–7 animals per group were tested.

Total RNA extracted from synapse-rich (NMJ and endplate) and synapse-free sternomastoid muscles were homogenized using a TRIzol and RNeasy MinElute Cleanup Kit (Qiagen, Hilden, Germany). Total extracted RNA was treated with RNase-free DNase (QIAGEN) (Lu et al., 2020). Purity and concentration of RNA were measured using a NanoDrop ND-1000 and only those samples with a 260 nm/280 nm ratio of between 1.6 and 2.0 were used for analysis. Complementary DNA was synthesized from total RNA (100 ng) using High-Capacity cDNA Transverse Transcription kit (Applied Biosystems).

Quantitative real-time PCR (qRT-PCR) was performed using TaqMan Fast Universal PCR Master mix and specific TaqMan PCR primers-probes combination were purchased from Applied Biosystems. Tbx21 (Mm00450960-m1) and Gapdh (Mm99999915-g1) Taq-Man.

Gene Expression Assays were used. All qPCR studies were performed on a Step One Plus instrument (Applied Biosystems), and results were analyzed using Microsoft Excel and quantified using the ΔΔCt method (Livak and Schmittgen, 2001). Relative expression levels were calculated for the Tbx21 gene by normalizing first to Gapdh level. Samples were run in triplicate, and “non-template controls” with water and no RT enzyme were included as negative controls. Relative gene expression for NMJs on the injured side is shown as fold change from the naïve mouse. The qRT-PCR was analyzed at 2 days, 1 week, 2-, 3-, 4- and 6-weeks post-nerve injury. A total of N = 3-5 animals per group were used.

The EDL and tibialis anterior (TA) muscles from right injured hindlimb were dissected and fixed overnight with 4% paraformaldehyde solution (Electron Microscopy Science, Hatfield, PA, United States) and embedded in OCT (Tissue-Tek, Miles, Elkhart, IN, United States). All OCT specimens were stored at −80°C until use. Longitudinal frozen sections (25 μm thick) or whole mount of EDL muscles were processed for immunofluorescence labeling of NMJ components as described previously (Jablonka-Shariff et al., 2019; Snyder-Warwick et al., 2018).

Primary antibodies used for this experiment were as follows: rabbit anti-S100b (Agilent Cat# Z0311, RRID:AB_10013383) for labeling of Schwann cells, a combination of rabbit anti-neurofilament 200 (NF200, Sigma-Aldrich Cat# N4142, RRID:AB_477272) and mouse anti-synaptic vesicle glycoprotein 2 (SV2,DSHB Cat# SV2, RRID:AB_2315387) or mouse anti-2H3 (DSHB Cat# 2H3, RRID:AB_531793) antibodies for labeling axons and axon terminals, mouse anti-Synaptophysin (Abcam Cat# ab8049, RRID:AB_2198854), rat anti-CD68 (Bio-Rad Cat# MCA 1957, RRID:AB_322219), rat anti-CD3 (Thermo Fisher Scientific Cat# 14-0032-85, RRID:AB_467054) and mouse anti-Tbet (Santa Cruz Biotechnology Cat# sc-21749, RRID:AB_628331). After overnight incubation at 4°C, sections were incubated in secondary antibodies conjugated with Alexa Fluor 488 nm, −594 nm or −594 nm (Life Technologies). Tissues were further incubated with α-bungarotoxin (α-BTX) Alexa Fluor 555 (Thermo Fisher Scientific Cat# B35451, RRID:AB 2617152) or −647. α-BTX binds specifically to AChRs in the postsynaptic membrane. Control sections (either no primary or no secondary antibody) were included in every type of staining to test for nonspecific staining and autofluorescence. Slides were prepared using Vectashield mounting medium with DAPI (Vector Laboratories Cat# H-1200, RRID:AB_2336790) to label nuclei.

For whole mounting of the EDL muscles all four bellies were imaged, analyzed and averaged to obtain a value for each single animal (n = 1). A minimum of 20-30 NMJs were acquired per animal. For muscle sections at least five random sections and 50-100 NMJs per muscle/animal were analyzed. A total of N = 3-7 animals per group were used. Morphological evaluation of NMJ components were performed in a blinded fashion by two investigators unless otherwise noted as described previously (Jablonka-Shariff et al., 2019; Snyder-Warwick et al., 2018). To determine the number of tSCs/NMJ, the percentage of tSCs showing processes and the percentage of NMJs with tSCs, muscle sections were stained with S100 antibody and α-BTX. To evaluate NMJ innervation, sections or whole mounts (EDL) were stained using a combination of NF200/SV2 antibodies, which label neurofilaments/synaptic vesicles, or NF200/2H3 antibodies and α-BTX for endplates. An NMJ was defined as fully innervated if the nerve terminal and endplate aligned ≥75% of their length. The percentage of innervated NMJs was normalized to the total number of NMJs in the macroscopic fields of view. Denervation was defined as no overlap of neurofilaments and α-BTX staining.

The number of CD68⁺ macrophages and CD3⁺ T cells stained with rat anti-CD68 and rabbit anti-CD3 antibodies were counted per field of microscopic view. The imaging was focused on fields containing NMJs along with visualization of the surrounding areas.

For quantitative analyses, all NMJs were imaged using the Fluoview FV1000 confocal microscope (Olympus, Waltham, MA) and an Axio Imager M2 fluorescent microscope (Zeiss). En-face images of NMJs were analyzed using NIH ImageJ macro (https://imagej.nih.gov/ij/). The fluorescence intensity for synaptophysin/NMJ in naïve WT and Tbet-KO mice was calculated as follows: Integrated Density – (area of NMJ x mean fluorescence of background values). Figures were prepared using Adobe Photoshop CC 2025 and Adobe Illustrator CC 2025 system (Adobe Systems, San Jose, California).

Continuous variables are reported as mean ± SD and categorical variables in percentages. Significance is set as P < 0.05. Quantification of reinnervation, tSCs, and immune cells were performed in a blinded fashion with two evaluators. Statistical analyses were performed by two-tailed, unpaired Student’s t tests or Mann-Whitney U tests using Microsoft Excel and GraphPad Prism 10 (GraphPad Software).

We reported previously that Tbet (Tbx21) protein is localized in tSCs at NMJs within skeletal muscles in homeostasis (Figures 1A, B) and demonstrated that Tbx21 mRNA expression was over nine-fold higher in the NMJ area relative to synapse-free sternomastoid muscle or sciatic nerve in the mouse (Jablonka-Shariff et al., 2021). Given the robust immune response within the muscle of the WT mice following nerve injury, we first assessed the expression of Tbx21 mRNA at the NMJ following sciatic nerve injury and immediate repair in S100-GFP mice (Figure 1C). Tbx21 gene expression, shown via qPCR, increases from 2 weeks to 4 weeks after nerve injury compared to naïve uninjured mice. The mRNA levels significantly peaked at 2 and 3 weeks (P < 0.0001 and P < 0.05, respectively) post-injury, suggesting a potential role of Tbx21 in NMJ regeneration.

EDL and TA muscle synaptic morphology from young adult (3–5-month-old) Tbet-KO mice showed similar NMJ structure to naïve control WT mice. There were no differences in mean tSC number per NMJ (2.1 ± 0.4 tSCs per NMJ in Tbet-KO mice vs. 2.6 ± 0.5 in WT mice), which stayed in contact with the entire endplate and expressed S100 antibody, a marker of the Schwann cells (Figures 2A–C). NMJs from Tbet-KO mice were fully innervated with nerve terminals (NF200/2H3) overlapped with pretzel-shaped endplates (α-BTX for AChRs), and their morphologies did not differ from WT controls (Figures 2D–F). NMJs also showed similar staining intensity with synaptophysin in synaptic vesicles (Figures 2G–I). No synaptophysin staining is observed at the NMJs with no primary or secondary antibodies (Figures 2J,K).

NMJ reinnervation in the EDL and/or TA muscles was examined in Tbet-KO and control WT mice. Fluorescently stained longitudinal frozen muscle sections and EDL whole mounts were evaluated for NMJ innervation at 2-, 4-, and 6-week and 6 months following sciatic nerve transection and immediate repair. Morphological differences were noted in reinnervation patterns between Tbet-KO and control groups (Figure 3). Most notably, neural branching patterns at the NMJs of Tbet-KO mice were more disorganized at earlier time points in both groups. This morphology persisted until 4 weeks in the Tbet-KO mice (Figure 3). By 6 weeks, the majority of NMJs in the WT mice showed an organized pattern of nerve terminal overlying the endplates while the disorganized pattern often persisted in the Tbet-KO group (Figures 3A–D). No staining is observed at the NMJs with no primary or secondary antibodies (Figures 3E, F). Quantitatively, at every time point, Tbet-KO mice showed significantly lower percentages of fully innervated NMJs compared to WT mice (Figure 3G). Finally, Tbet-KO mice showed decreased (P < 0.05) percentages of fully innervated NMJs (70.5% ± 24.6%) and more denervated NMJs compared to WT mice (94.4% ± 13.6%) at 6 months following nerve injury.

Tetanic muscle force analyses were conducted on Tbet-KO and WT mice at 3-, 4-, and 6-week, and 6 months following sciatic nerve injury and immediate repair (Figure 3H). At 3 weeks, the tetanic muscle force generated by the Tbet-KO mice was 37.1% of WT mice. While the muscle force increased at 4 weeks and at 6 weeks, it remained significantly lower in Tbet-KO mice at 6 months compared to WT mice. These data suggest a role for Tbx21 in the reinnervation process following nerve injury.

tSCs encapsulate the NMJ and play supporting roles in NMJ reinnervation and muscle regeneration after nerve injury. Following nerve injury, NMJs in the hindlimb injured muscles from Tbet-KO mice exhibited significantly decreased numbers of tSCs per NMJ (P < 0.05) compared to those of WT mice. tSC numbers were reduced to half those of WT at 4 weeks and 6 weeks (Figures 4A–E). Finally at 6 months only 1.6 ± 0.6 of tSCs/NMJ were noted in Tbet-KO mice compared to 2.4 ± 0.4 of tSCs/NMJ for WT mice. In addition, Tbet-KO mice showed significantly lower percentages of NMJs with tSCs present compared to WT mice (Figure 4F).

Following nerve injury, tSCs normally extend, in advance of axon sprouts, cytoplasmic processes beyond the NMJ in search of neighboring, innervated NMJs. In Tbet-KO mice, the percentage of tSCs extending processes was significantly less than that of injured control WT mice during 4 weeks of NMJ regeneration after nerve transection and repair (Figure 5). In addition, tSC processes were short with less extension beyond the NMJ (Figures 5B, D, E). Of note, no tSC processes were observed in the absence of injury (naïve WT and Tbet-KO in Figures 2A, B). These results suggest Tbx21 transcription factor in tSCs plays an important role in tSC morphology and their response to nerve injury, likely affecting synapse remodeling post-injury.

The immune response has been shown to be integral in sciatic nerve regeneration and at NMJs in muscle after nerve injury. During NMJ regeneration, macrophages play many critical functions, including orchestration of the inflammatory response and clearance of nerve terminal debris. We found that CD68⁺ macrophages were significantly elevated in WT mice during the first 4 weeks after nerve injury and immediate repair (Figure 6), reaching peak (26.8 ± 3.6) numbers at 3 weeks post-injury (P < 0.001). In contrast, Tbet-KO mice showed fewer macrophage numbers with a brief and delayed increase in number (16.4 ± 4.8) at 4 weeks post-injury (P < 0.05). Similar to macrophages, the number of CD3+T cells was low at all experimental time points in Tbet-KO mice compared to WT mice, which showed significantly (P < 0.01) elevated T cell numbers (15.9 ± 2.1 WT vs. 2.9 ± 0.4 Tbet-KO) at 3 weeks post-injury (Figure 7).