Lesion-level dependent systemic muscle wasting after spinal cord injury is mediated by glucocorticoid signaling in mice

Abstract

An incomplete mechanistic understanding of skeletal muscle wasting early after spinal cord injury (SCI) precludes targeted molecular interventions. Here, we demonstrated systemic wasting that also affected innervated non-paralyzed (supralesional) muscles and emerged within one-week after experimental SCI in mice. Systemic muscle wasting caused muscle weakness, affected fast type 2 myofibers preferentially, and became exacerbated after high (T3) compared to low (T9) thoracic paraplegia indicating lesion-level dependent (“neurogenic”) mechanisms. The wasting of non-paralyzed muscle, its rapid onset and severity beyond what can be explained by disuse implied unknown systemic drivers. Muscle transcriptome and biochemical analysis revealed a glucocorticoid-mediated catabolic signature early after T3 SCI. SCI-induced systemic muscle wasting was mitigated by i) endogenous glucocorticoid ablation (adrenalectomy), ii) by pharmacological glucocorticoid receptor (GR) blockade and was iii) completely prevented after T3 relative to T9 SCI by genetic muscle-specific GR deletion. These results suggest that neurogenic hypercortisolism contributes to a rapid systemic and functionally relevant muscle wasting syndrome early after paraplegic SCI in mice.

Editor’s summary

It’s not just use it or lose it

The loss of muscle mass after spinal cord injury is mainly attributed to disuse. Here Harrigan and colleagues found that injury at the higher thoracic spinal cord (T3) caused more wasting of paralyzed and non-paralyzed muscles than a more rostral spinal cord injury (T9). Removal of the adrenal glands or glucocorticoid receptor inhibition reduced muscle wasting after T3 injury without affecting locomotor behavior and genetic deletion of glucocorticoid receptors from skeletal muscles abolished lesion-level dependent effects on muscle wasting. While further studies are needed, these results suggest that inhibition of glucocorticoid signaling could help to ameliorate muscle wasting after spinal cord injury. – Daniela Neuhofer

One Sentence Summary:

Lesion-level dependent hypercortisolism induces glucocorticoid-mediated systemic muscle wasting early after paraplegic mouse spinal cord injury.

INTRODUCTION

Loss of skeletal muscle mass is a prevalent consequence of spinal cord injury (SCI) (1). The clinical efficacy of neurorehabilitation, the only successful therapy to improve functional recovery after SCI, depends on signals relayed from muscle back to the spinal cord (2) and is rate-limited by muscle loss after SCI (3, 4). Maintaining muscle is crucial for patients with SCI even beyond enabling locomotor recovery. Muscle loss provokes secondary SCI complications (5) such as pressure sores (6), respiratory compromise (7), metabolic syndrome (8), osteoporosis (9), spasticity (10), and cardiovascular dysfunction (11). Elevated mortality of patients with low body mass indices further suggests protective functions of muscle after SCI (12). Thus, muscle mass preservation is a plausible therapeutic target to reduce morbidity and optimize clinical outcomes for patients with SCI.

Muscle is highly plastic and the most abundant tissue in humans, making up 40-to-50% of total body mass (13). The dynamic flux between muscle protein synthesis and degradation is controlled by a complex interplay of neural, mechanical, and hormonal signals (14). SCI disrupts the central motor innervation (upper motor neuron injury) and voluntary use of sublesional muscle (paralysis) (5). Subsequent “atrophy” of disconnected paralyzed muscle is considered an eventual consequence of disuse (15). However, paralyzed muscle wasting early after human SCI (16–18) exceeds what can be attributed to disuse atrophy (19) and intensive care unit (ICU)-acquired wasting (20) (fig. S1A). Non-paralyzed (supralesional) muscle wasting may also occur, although consistent comparative clinical (18, 21) and experimental (22, 23) data are lacking. A systemic muscle wasting syndrome would imply the presence of wide-reaching mechanisms invoking muscle protein degradation (fig. S1, B to D).

The goal of this study was to assess systemic muscle responses early after SCI, extending beyond paralyzed muscle groups. To do so, we sought to investigate the role of SCI lesion-level dependent systemic effects and disentangle them from localized effects attributed to disuse or denervation (Fig. 1A). SCI is a life-threatening event able to induce a systemic state akin to the “fight-or-flight” response (24). The capacity to trigger anterograde activation of the stress axis (hypothalamic-pituitary-adrenal [HPA]-axis) depends on the SCI lesion level. Given the vital role of the adrenal glands as major homeostatic effector organs, we hypothesized that their denervation (decentralization) by a SCI lesion above T4-5 would deleteriously impact muscle homeostasis (Fig. 1B).

Fig. 1. SCI-induced systemic muscle wasting emerges rapidly within 3 days, is lesion-level dependent and impairs muscle function.

(A) Study design to disentangle neurogenic (spinal cord injury [SCI] lesion-level dependent) systemic effects from those induced by muscle paralysis. An experimental SCI mouse model was applied to compare non-paralyzed forelimb (Triceps brachii) and paralyzed hindlimb (Gastrocnemius) index muscles after SCI differentially. This enabled the phenotypic distinction of systemic from local muscle effects, such as disuse. Furthermore, delineation of systemic SCI lesion-level dependent effects could be ascertained by comparing forelimb muscles after different thoracic SCI levels (T3 vs. T9 lesion). (B) Both T3 and T9 SCI cause paraplegia (hindlimb paralysis) and spare forelimb muscles in mice. However, only T3 SCI denervates (“decentralizes”) the adrenal gland, a major neuroendocrine effector organ which executes whole-body stress responses. Longitudinal changes in (C) total body weight, (D) lean body mass and (E) adipose body mass during the first 28 days post injury (dpi) for sham injured (Sham spine), T3 and T9 SCI lesioned mice. The ratio of (F) triceps brachii or (G) gastrocnemius muscle wet weight (in mg) to pre-injury lean body weight (in g) for sham injured, T3 and T9 SCI lesioned mice at 3 and 7 dpi. Corresponding functional measures of (H) forelimb grip strength and (I) hindlimb tetanic contractility for sham injured, T3 and T9 SCI lesioned mice pre-injury and at 7 dpi. (J) Locomotor activity at 3 and 7 dpi for the same injury groups in (C-I) as assessed by an automated “activity box”. (C-E) Mixed effects models with Greenhouse-Geisser correction – (C) Time: p<0.0001, F=14.47; Injury: p<0.0001, F=403; Interaction: p<0.0001, F=21.88; (D) Time: p<0.0001, F=44.47; Injury: p<0.0001, F=271.8; Interaction: p<0.0001, F= 15.00; (E) Time: p=0.0386, F= 3.414; Injury: p<0.0001, F=105.3; Interaction: p<0.0001, F=13.11. (F & G) Two-Way analysis of variance (ANOVA) – (F) Time: p=0.8145, F=0.05569; Injury: p<0.0001, F= 39.16; Interaction: p=0.0027, F=6.699; (G) Time: p<0.0001, F=45.77; Injury: p<0.0001, F=132.6; Interaction: p<0.0001, F=24.02. (H-J) Two-Way Repeated Measure ANOVA – (H) Time: p=0.0030, F=14.34; Injury: p=0.0238, F=5.351; Interaction: p=0.0025, F=10.90; (I) Time: p=0.0006, F=20.23; Injury: p=0.0137, F=6.080; Interaction: p=0.0001, F=20.31; (J) Time: p=0.0275, F=5.453; Injury: p<0.0001, F=50.47. Data are mean ± standard error of the mean (SEM) with significant post hoc Holm–Šidák corrected multiple comparisons related to injury displayed: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (in C-E: * = T3 vs T9 SCI, ^ = T9 SCI vs Sham spine). (C-E) n = 10-20 mice/timepoint for Sham spine group (3 & 7 dpi = 20, 14-28 dpi = 10); n = 12-22 mice/timepoint for T9 SCI group (3 & 7 dpi = 22, 14 dpi = 14, 21 & 28 dpi = 12); n = 9-20 mice/timepoint for T3 SCI group (3 & 7 dpi = 20, 14 dpi = 14, 21 dpi = 11, 28 dpi = 9). (F & G) n = 9 mice/group; (H) n = 4 (Sham spine) or 5 (T3, T9 SCI) mice/group; (I) n = 5 (Sham spine, T9 SCI) or 6 (T3 SCI) mice/group; (J) n = 8 (T9 SCI), 9 (T3 SCI) or 10 (Sham spine) mice/group.

Here, we identified systemic muscle wasting over the first week(s) after SCI. We determined the temporal dynamics and characteristics and further characterized underlying molecular drivers. Both non-paralyzed and paralyzed muscle mass loss was exacerbated by high (T3), relative to low (T9), thoracic SCI and was glucocorticoid dependent. Microsurgical, pharmacological and molecular “loss of function” interventions consistently preserved systemic muscle mass, including i) adrenalectomy (ADX) or ii) the treatment with the glucocorticoid receptor (GR) antagonist mifepristone (MIF [RU-486]). In vivo muscle-specific GR deletion (SMGRKO) completely abolished the SCI lesion-level dependent (“neurogenic”) aggravation of systemic muscle wasting. This study suggests that protecting muscle from SCI-induced systemic glucocorticoid-mediated wasting is a therapeutic candidate to preserve muscle function and optimize outcomes of pro-regenerative SCI therapies.

RESULTS

T3 SCI aggravates rapid systemic lean tissue wasting

Human paraplegic thoracic SCI paralyzes the lower extremity muscles but spares supralesional muscles in the upper extremities (non-paralyzed) (fig. S1, B to D). To investigate how SCI at different thoracic spinal cord levels differentially affects non-paralyzed and paralyzed muscles, we compared post-injury changes in mice receiving complete spinal cord transection (TXN) injuries at either the T3 or T9 thoracic spinal cord level. High T3 SCI disconnects (“decentralizes”) the adrenal glands from sympathetic CNS control. In contrast, supraspinal innervation of the adrenal glands is retained after low T9 SCI (24) (Fig. 1B). To control for potential catabolic effects of surgical stress, SCI mice were compared with mice undergoing sham spinal surgery without TXN (Sham spine). Although all groups responded to surgery with some body weight loss early on, the subsequent trajectories were distinct (Fig. 1C). Continued body weight loss was observed only in T3 SCI mice; average body weight loss approached −20% within the first 7 days post-injury (dpi) and persisted through 1-month post-injury (latest timepoint evaluated). In contrast, T9 SCI mice recovered body weight within 1 month.

Quantitative in vivo echoMRI was subsequently applied to determine whether early subacute body weight changes could be attributed to loss of lean tissue, adipose tissue or both. Lean body mass loss mirrored the temporal dynamics and SCI lesion-level dependent pattern of severity of body weight loss (Fig. 1D). Already within 3 dpi, lean body mass loss in T3 SCI mice eclipsed T9 SCI mice (−18% vs −12%) and persisted through 1 month, whereas T9 SCI mice regained lean body mass. In contrast, adipose body mass loss was more pronounced in T9 SCI mice at 3 dpi and equivalent between T3 and T9 SCI mice at 7 dpi (Fig. 1E). Only after 7 dpi did adipose body mass loss in T3 SCI mice begin to exceed T9 SCI mice, which might account for the continued average body weight loss observed in T3 SCI mice beyond 7 dpi. Together, these data demonstrate that T3, relative to T9, SCI causes more severe body weight loss, which occurs overwhelmingly during the first week after injury and is chiefly driven by lean tissue loss.

Non-paralyzed (supralesional) and paralyzed (sublesional) muscle wasting are exacerbated after T3 compared to T9 SCI

To localize the loss of lean body mass early after SCI we assessed terminal wet weights of both non-paralyzed (forelimb) and paralyzed (hindlimb) muscles. In the forelimb, loss of non-paralyzed triceps brachii muscle mass was similar after high T3 or low T9 SCI at 3 dpi (−8% vs −6%, relative to Sham spine), but at 7 dpi this wasting was more pronounced in T3 SCI versus T9 SCI animals (−16% vs −5%) (Fig. 1F). At 3 dpi, paralyzed gastrocnemius muscle mass was reduced approximately −10% after either high T3 or low T9 thoracic SCI (relative to Sham spine), but at 7 dpi muscle mass loss was higher in T3 relative to T9 SCI mice (−25% vs −20%) (Fig. 1G). Importantly, non-paralyzed muscle wasting severity corresponded with neuromuscular functional deficits, as determined by longitudinal in vivo measures of forelimb neuromuscular strength (Fig. 1H). At 7 dpi, non-paralyzed forelimb grip strength was reduced in mice with T3 SCI relative to T9 SCI (−33%) or Sham spine (−57%) mice. In contrast, the −5% difference in paralyzed muscle mass loss (Fig. 1G) did not translate into a detectable SCI lesion-level dependent difference in muscle contractility. In both SCI groups maximal tetanic contractility of the hindlimb plantarflexion muscles was reduced by approximately −30% at 7 dpi, relative to Sham spine animals (Fig. 1I).

SCI at both high and low thoracic lesion levels induced rapid wasting of non-paralyzed and paralyzed skeletal muscles, but both were more severely reduced after T3 SCI compared to T9 SCI (Fig. 1, F and G). Locomotor inactivity and reduced forelimb strength were also more pronounced after T3 compared to T9 SCI (Fig. 1, H and J) in line with aggravated muscle weakness and dysfunction. This further verifies the functional relevance of the additional muscle loss superimposed by T3 SCI and suggests that the absence of SCI lesion-level dependent functional differences in paralyzed muscle in Fig. 1I may be a consequence of insufficient assay sensitivity. Finally, since T3 SCI mice were housed in warmed cages, but T9 SCI mice were not, we confirmed that the observed SCI lesion-level dependent effects were unrelated to cage temperature by demonstrating the absence of temperature-dependent muscle phenotype differences in T9 SCI mice (fig. S2).

Although triceps brachii and gastrocnemius are extensor muscles, flexor muscles in both the forelimb and hindlimb exhibited similar SCI lesion-level dependent wasting phenotypes (fig. S3). Observed wasting was not limited to skeletal muscle as wasting of striated myocardium also occurred in a SCI lesion-level dependent manner (fig. S4). Whereas only a modest reduction in myocardial heart mass was observed at 7 dpi after T9 SCI (−9%, relative to Sham spine), myocardial mass declined progressively at 3 and 7 dpi after T3 SCI (−14% & −25%, relative to Sham spine).

Rapid wasting of non-paralyzed muscle (supralesional) is driven by atrophy preferentially affecting “fast twitch” type 2 muscle fibers

To investigate the underlying mechanism(s) of systemic muscle wasting we characterized muscle fiber (myofiber) changes at the cellular level. Seven days after T3 SCI, T9 SCI or Sham spine surgeries forelimb triceps brachii (Fig. 2, A to C) and hindlimb gastrocnemius (Fig. 2, D to F) myofibers were histologically assessed for cytopathology. Irrespective of SCI lesion-level, none of the sampled muscles exhibited central nuclei, inflammatory infiltrate, or myofiber cell membrane (sarcolemma) disruption, indicating an organized pattern of myofiber atrophy lacking signs of myositis and regeneration. The predominant manifestation was the emergence of small basophilic, occasionally angular myofibers in paralyzed muscles (Fig. 2, E and F), which appeared grossly smaller after T3 SCI relative to T9 SCI (fig. S5). These myofibers from paralyzed muscles were interspersed (non-clustered) and could have resulted from local mechanisms of atrophy, like disuse, with or without the superimposition of systemic mechanisms.

Fig. 2. Early SCI-induced systemic muscle wasting preferentially affects fast-twitch type 2 myofibers in a lesion-level dependent manner.

Representative 20X images of hematoxylin and eosin (H&E) histology of (A-C) forelimb and (D-F) hindlimb muscles for sham injured, T3 and T9 SCI lesioned mice at 7 dpi. Scale bars are 200 μm. A representative subset of small basophilic, angular myofibers in (E & F) paralyzed hindlimb muscles are labeled by white arrows (also at high magnification in fig. S5). Differences in triceps brachii myofiber size at 7 dpi for sham injured, T3 and T9 SCI lesioned mice as illustrated by (G) mean minimum ferret-diameters and (H) relative-frequency histogram plotting minimum ferret-diameters. For the same injury groups as in (A-F), immunofluorescent (IF) analysis of the four murine myofiber types (1, 2A, 2B, 2X; see also fig. S6 for more details) quantified (I) specific myofiber type frequencies and (J) differences in myofiber size (minimum ferret-diameter) in non-paralyzed forelimb muscle at 7 dpi. (G & I) One-Way ANOVA – (G) p<0.0001, F= 158.6; (I) Type 1: p=0.2619, F=0.778; Type 2A: p=0.6673, F=0.433; Type 2B: p=0.9266, F=0.07721; Type 2X: p=0.2653, F=1.669. (J) Nested One-Way ANOVAs – Type 1: p=0.1978, F=2.149; Type 2A: p=0.0334, F=6.315; Type 2B: p<0.0001, F=54.34; Type 2X: p<0.0001, F=52.73. Data are mean ± SEM with significant post hoc Holm–Šidák corrected multiple comparisons related to injury displayed: *p < 0.05, **p <0.01, ****p <0.0001. (A-F) n = 4 muscles/group; (G-J) n = 3 muscles/group (total myofibers analyzed: n = 1,389 for T3 SCI, n = 1,189 for T9 SCI, n = 1,140 for Sham spine).

To identify cellular myofiber changes uniquely underlying systemic muscle wasting we performed an extended analysis of non-paralyzed forelimb myofibers. An immunofluorescent (IF) characterization of laminin and myosin heavy chain (MyHC) isoform expression in non-paralyzed triceps brachii muscle was performed to identify and measure the size of all myofibers and to further stratify into the four murine myofiber types (1, 2A, 2B, 2X) (fig. S6). Compared to Sham spine animals, at 7 dpi the mean diameter of all myofibers progressively shrunk by −13.7% after T3 SCI compared to −2.4% after T9 SCI (Fig. 2G). When stratifying by myofiber diameter, we observed an overall shift toward smaller myofibers (< 35 μm), a reduction in intermediately sized myofibers (35 to 50 μm) and the dissipation of larger myofibers > 50 μm in T3 SCI mice (Fig. 2H). Although the relative composition of triceps brachii myofiber types remained unchanged after SCI (Fig. 2I), SCI lesion-level dependent changes in the size of specific myofiber types were observed at 7 dpi (Fig. 2J). When comparing to Sham spine myofiber sizes, we found type 2B (−12.3% T3 SCI vs −2.5% T9 SCI), type 2X (−14.5% T3 SCI vs −4.9% T9 SCI), and to a lesser extent type 2A (−8.7% T3 SCI vs −1.1% T9 SCI) “fast twitch” myofibers atrophy with SCI lesion-level dependent severity. No differences were observed in “slow twitch” type 1 myofibers. In non-paralyzed muscle, a SCI lesion-level dependent reduction in MyHC-2B and MyHC-2X expression, as determined by quantitative colorimetric MyHC electrophoresis (fig. S7, A to B), was concordant with the SCI lesion-level dependent reduction in type 2B and 2X myofiber size observed by IF analysis (Fig. 2J). Colorimetric MyHC electrophoresis also identified SCI lesion-level dependent differences in paralyzed muscle (fig. S7C) and myocardial MyHC composition (fig. S7D). In conjunction with Fig. 1, F and G, these data demonstrate that T3 SCI exacerbates rapid systemic non-paralyzed muscle wasting, relative to T9 SCI, primarily through type 2B and 2X myofiber atrophy, which are more susceptible to systemic insults, as opposed to types 1 and 2A, which are more sensitive to disuse and peripheral denervation (25).

Aggravation of systemic muscle wasting by T3 SCI cannot be explained by disuse, peripheral denervation or nutritional differences

The selective atrophy of predominantly type 2B and 2X myofibers opposes the involvement of locally restricted mechanisms, like disuse or peripheral denervation, in non-paralyzed muscle (Fig. 2J). Furthermore, disuse atrophy cannot rationally explain wasting of voluntarily mobile, non-paralyzed muscle since even slight contractile activity is sufficient to preserve muscle mass (26). The applied experimental paraplegic model was devised to preserve full motor innervation of the forelimb (Fig. 1, A and B; T1 supplies LMN innervation to the triceps brachii). However, we applied electrophysiological motor unit number estimation (MUNE) to rule out potential systemic peripheral neuromuscular dysfunction, like critical illness neuromyopathy, or inadvertent peripheral denervation of lower motor neurons (LMNs) supplying the forelimb during T3 SCI surgery. When comparing T3 SCI with T9 SCI and Sham spine animals, no differences were observed in the number of motor units functionally connected to the non-paralyzed forelimb (Fig. 3A) or paralyzed hindlimb (Fig. 3B) muscles. This confirmed that non-paralyzed forelimb muscles retain complete functional motor innervation at 7 dpi after both T3 and T9 SCI and accordingly muscle mass loss at this timepoint cannot be explained by peripheral (LMN) mechanisms. Another putative confounder addressed was whether nutritional differences could contribute to the observed muscle wasting phenotype. However, the absence of a SCI lesion-level dependent difference in food intake (Fig. 3C) or hydration (Fig. 3D) during the first week after injury ruled this possibility out. To decipher muscle-specific mechanisms underlying rapid systemic muscle wasting we interrogated differential muscle gene expression profiles to identify candidate catabolic drivers.

Fig. 3. T3 SCI leads to transcriptomic changes associated with glucocorticoid signaling and hypercortisolism.

Longitudinal in vivo (A) forelimb and (B) hindlimb motor unit connectivity, as determined by motor unit number estimation (MUNE) electromyography (EMG), for sham injured, T3 and T9 SCI lesioned mice pre-injury and at 7 dpi. For the same injury groups as in (A & B), (C) cumulative food intake during the first week after injury and (D) longitudinal echoMRI hydration status pre-injury and at 7 dpi. (E) Tabulation of differentially expressed genes (DEGs) determined by false discovery rate (FDR) <0.05 and ≥ two-fold difference when comparing transcriptomic profiles of triceps brachii and gastrocnemius muscles from sham injury, T3 or T9 SCI lesioned mice at 3 dpi. Venn diagrams, derived from the same 3 dpi transcriptomic profiles as presented in (E), illustrate the total number of SCI-specific and SCI lesion-level dependent DEGs identified in (F) non-paralyzed forelimb and (G) paralyzed hindlimb muscles (see also fig. S8 for further detail). (H) Illustration of canonical glucocorticoid (dexamethasone)-mediated signaling in skeletal muscle. (I) Circulating total corticosterone, as determined by ELISA, in plasma from sham injured, T3 and T9 SCI lesioned mice at 3 dpi. (A, B & D) Two-Way Repeated Measure ANOVA – (A) Time: p=0.3128, F=1.110; Injury: p=0.6852, F=0.3927; (B) Time: p=0.2077, F=1.773; Injury: p=0.3272, F=1.252; (D) Time: p=0.0016, F=11.85; Injury: p=0.2487, F=1.454; Interaction: p=0.0543, F=3.195. (C & I) One-Way ANOVA – (C) p=0.0008, F=29.35; (I) p<0.0001, F=15.5. (E-G) RNA-seq statistical analysis outlined in Supplementary Materials. Data presented as mean ± SEM with significant post hoc Holm–Šidák corrected multiple comparisons related to injury displayed: *p < 0.05, **p < 0.01, ****p < 0.0001. (A & B) n = 4 (T3, T9 SCI) and 5 (Sham spine) mice/group; (C) n = 3 cages/group ( 5 mice/cage); (D) n = 10 (Sham spine), 12 (T9 SCI) and 13 (T3 SCI) mice/group; (E-G) n = 4 (gastrocnemius) or 6 (triceps brachii) samples/group; (I) n = 12 (Sham spine), 13 (T9 SCI) or 16 (T3 SCI) mice/group.

T3 SCI-induced muscle gene expression profiles identify glucocorticoid-signaling as a candidate mechanism underlying systemic muscle wasting

RNA sequencing was applied to generate comprehensive gene expression profiles for forelimb triceps brachii and hindlimb gastrocnemius muscles removed from mice 3 days after T3 SCI, T9 SCI or Sham spine surgery. This timepoint was chosen because it corresponds with the inception of SCI lesion-level dependent mechanisms of systemic muscle wasting. Gene expression changes emerging already at 3 dpi could potentially be targeted to mechanistically prevent systemic muscle wasting observed later at 7 dpi.

Compared to Sham spine gastrocnemius, differentially expressed gene (DEG) patterns were observed in the paralyzed gastrocnemius of both T3 SCI (1,167 DEGs) and T9 SCI mice (1,322 DEGs) (Fig. 3E). By contrast, in the non-paralyzed triceps brachii gene expression was altered after T3 SCI (454 DEGs), but not T9 SCI (2 DEGs), relative to Sham spine triceps brachii. SCI lesion-level dependent gene expression profiles were observed in both the non-paralyzed (256 DEGs) (Fig. 3F; fig. S8, A and B) and paralyzed muscles (172 DEGs) (Fig. 3G; fig. S8, C and D). In non-paralyzed muscle, SCI lesion-level dependent upregulation of conserved effectors of rapid muscle atrophy were detected (Trim63/MuRF1 [GeneID: 43376; Log2FC = 2.31, False Discovery Rate (FDR) = 0.0013], Fbxo32/MAFbx [GeneID: 67731; Log2FC = 2.07, FDR = 0.011]) (table S1) mirroring manifested muscle mass loss reported in Fig. 1F.

Subsequently, multiple lines of analysis uncovered glucocorticoid signaling as the most compelling candidate mechanism underlying the SCI lesion-level dependent severity of systemic muscle wasting. First, individual genes specifically involved in glucocorticoid-mediated muscle wasting signaling (Klf15 [GeneID: 66277; Log2FC = 1.12, FDR = 0.043], Ddit4/Redd1 [GeneID: 74747; Log2FC = 1.00, FDR = 0.42], Sesn1 [GeneID: 140742; Log2FC =, FDR =], Pik3r1 [GeneID: 18708; Log2FC = 1.57, FDR = 0.0033]) (27–30) were upregulated after T3, relative to T9, SCI (table S1). Second, “upstream regulator analysis”, a proprietary software analysis which draws on a curated database of gene/signaling-pathway associations in the literature, predicted glucocorticoid signaling as the most significant activated catabolic candidate signaling pathway (activation z-score = 3.64, p = 2.82 x 10⁻¹²) responsible for the observed DEG pattern exacerbating wasting in non-paralyzed triceps brachii early after T3, relative to T9, SCI (Fig. 3H, table S2). Lastly, the unbiased transcriptomic analysis identifying glucocorticoid-mediated wasting in non-paralyzed muscle early after T3 SCI was corroborated further by elevated circulating humoral (systemic) glucocorticoid levels (Fig. 3I). A SCI lesion-level dependent effect was observed at 3 dpi, with higher circulating corticosterone (CORT), the major murine glucocorticoid, observed after T3, relative to T9, SCI. Although T3, but not T9, SCI mice were housed in warmed cages, this was not a temperature-dependent phenomenon as we have also demonstrated early thoracic SCI lesion-level dependent hypercortisolism when all mice were housed in ambient temperature cages (24) .Together, these findings indicate that muscles in T3 SCI mice are exposed to higher concentrations of circulating glucocorticoids and provided a strong rationale to investigate glucocorticoids as an activated candidate catabolic signaling mechanism exacerbating the systemic muscle wasting occurring early after high thoracic SCI.

Adrenal glucocorticoid elimination or glucocorticoid receptor antagonism mitigate systemic muscle wasting early after T3 SCI

Systemic CORT elevation in conjunction with SCI-reactive glucocorticoid RNA signatures in non-paralyzed muscle above the SCI lesion site implicated glucocorticoid-dependent mechanisms of muscle wasting. We applied several interventions to functionally test this hypothesis. First, the effect of adrenal glucocorticoid elimination on lean body mass during the first week after T3 SCI was assessed. Adrenal glucocorticoid elimination was achieved by microsurgical bilateral adrenalectomy (ADX), removing the main synthesizing organs of endogenous glucocorticoids. To avoid potentially confounding effects of concomitant loss of adrenal aldosterone (ALDO), ALDO was re-integrated by exogenous replacement following ADX. Seven days after T3 SCI, lean body mass loss was attenuated by “ADX+ALDO”, relative to “sham ADX” alone (−11% vs −17%) (Fig. 4A). Of note, ADX did not produce a confounding lean (fig. S9, Fig. 4A) or muscle (Fig. 4, B and C) phenotype in Sham spine mice during the interventional timeframe. To resolve whether the protective effect on lean body mass was due to glucocorticoid elimination or ALDO supplementation, we further assessed the effects of ALDO supplementation in T3 SCI animals with sham ADX (“sham ADX+ALDO”). In these mice ALDO exhibited no protective effect on lean body mass. Subsequently, we assessed the effects of CORT reintroduction in T3 SCI animals with ADX+ALDO (“ADX+ALDO+CORT”). Here, CORT supplementation eliminated the protective effect on lean body mass of ADX+ALDO in turn proving causality.

Fig. 4. Glucocorticoid ablation by adrenalectomy and glucocorticoid receptor antagonism mitigate systemic muscle wasting early after T3 SCI.

Effect of adrenalectomy (ADX) or control sham ADX, with aldosterone (ALDO) or corticosterone (CORT) re-integration, on (A) lean body mass and the ratio of (B) non-paralyzed forelimb or (C) paralyzed hindlimb (violet) muscle wet weight (in mg) to pre-injury lean body weight (in g) in sham injury and T3 SCI lesioned mice at 7 dpi. The effect of glucocorticoid receptor (GR) antagonism (mifepristone [MIF]) on (D) lean body mass and the ratio of (E) non-paralyzed forelimb or (F) paralyzed hindlimb (violet) muscle wet weight (in mg) to pre-injury lean body weight (in g) in T3 SCI lesioned mice at 7 dpi, compared to vehicle-controls. (A-C) Two-Way ANOVA comparing T3 SCI vs Sham spine control groups – (A) Injury: p<0.0001, F=249.8; Intervention: p=0.0295, F=5.304; Interaction: p=0.0206, F=6.078; (B) Injury: p<0.0001, F=62.84; Intervention: p=0.0020, F=13.27; Interaction: p=0.0071, F=9.362; (C) Injury: p<0.0001, F=139.3; Intervention: p=0.0080, F=9.004; Interaction: p=0.0923, F=3.183. (A-C) One-Way ANOVA comparing intervention between T3 SCI groups – (A) p=0.0058, F=5.41; (B) p=0.0031, F=8.707; (C) p=0.0023, F=10.06. All data are mean ± SEM with (A-C) hypothesis-relevant post Holm–Šidák corrected multiple comparisons related to ADX intervention(s) and (D-F) unpaired t-test results displayed: *p < 0.05, **p < 0.01, non-significant (n.s.). (A) n = 4 (T3 SCI + ADX+ALDO+CORT), 5 (T3 SCI + sham ADX+ALDO), 8 (Sham spine + sham ADX+ALDO, Sham spine + ADX+ALDO) or 9 (T3 SCI + sham ADX, T3 SCI + ADX+ALDO) mice/group; (B & C) n = 5 (Sham spine + sham ADX+ALDO, sham + ADX+ALDO, T3 SCI + sham ADX, T3 SCI + sham ADX+ALDO) or 6 (T3 SCI + ADX+ALDO) mice/group; (D-F) n = 4 mice/group.

In addition to preserving lean body mass, adrenal glucocorticoid ablation in the “ADX+ALDO” group also decreased wasting of non-paralyzed triceps brachii (+12%, relative to sham ADX groups) (Fig. 4B) and paralyzed gastrocnemius (+11%, relative to sham ADX groups) (Fig. 4C) muscle mass in T3 SCI animals. In sum, ADX systemically ameliorated loss of lean body mass, as well as both non-paralyzed and paralyzed muscle mass, verifying that adrenal hormones, likely glucocorticoids, contribute to systemic muscle wasting early after T3 SCI.

Subsequently, we assessed whether blocking glucocorticoid signaling pharmacologically could likewise protect from muscle wasting. Blockade of glucocorticoid receptors (GR) by daily mifepristone (MIF) injections also reduced lean body mass loss and muscle wasting 7 days after T3 SCI (Fig. 4, D to F). MIF treatment mitigated total lean body mass loss compared to vehicle treatment (−15% vs −20%) (Fig. 4D). A muscle protective effect mediated by MIF was restricted to non-paralyzed muscle (+12% relative to vehicle) (Fig. 4, E and F). The preventative effects of both ADX and MIF were not associated with any confounding change in locomotor activity (fig. S10). The protective MIF effects on lean body mass and non-paralyzed muscle mass, in conjunction with the ADX data above, suggest that glucocorticoids binding at GRs aggravate muscle wasting early after T3 SCI.

Glucocorticoid receptor knockout ablates the SCI lesion-level dependent severity of systemic muscle wasting

To definitively determine whether glucocorticoids act directly on skeletal muscle GRs to induce wasting early after T3 SCI, an inducible skeletal muscle specific GR knockout (SMGRKO) mouse line was generated (fig. S11). GR deletion prior to SCI surgery reduced muscle GR protein expression in SMGRKO mice (GR^fl/fl;HSAMCM⁺) compared to littermate control GR-loxP mice (GR^fl/fl;HSAMCM⁻) (Fig. 5A) without inducing a confounding lean phenotype (fig. S12). SMGRKO mice exhibited attenuated lean body mass loss at 7 days after T3 SCI (−13%), relative to GR-loxP controls (−18%) (Fig. 5B). Wasting of non-paralyzed triceps brachii (T3: +25%; T9: +7%) (Fig. 5C) and paralyzed gastrocnemius (T3: +18%; T9: +9%) (Fig. 5D) muscles was reduced in SMGRKO mice, relative to GR-loxP mice. In contrast to control GR-loxP mice, no SCI lesion-level dependent differences of non-paralyzed muscle mass were observed at 7 dpi between SMGRKO animals with either T3 or T9 SCI. Of note, none of the tested anti-glucocorticoid interventions attenuated early subacute SCI lesion-level dependent cardiac muscle wasting (fig. S13).

Fig. 5. Circulating glucocorticoids act on muscle glucocorticoid receptors to exacerbate early systemic muscle wasting in a SCI lesion-level dependent manner.

(A) Proof-of-principle western blotting demonstrated differences in GR protein expression in forelimb and hindlimb muscles from skeletal muscle GR knockout (SMGRKO) and control GR-loxP mice 10 days after a course of tamoxifen. GR and β-actin immunoblots presented in (A) were electrophoretically separated on separate gels (see fig. S11 for further detail). Effect of SMGRKO or GR-loxP on (B) lean body mass and the ratio of (C) non-paralyzed forelimb or (D) paralyzed hindlimb (violet) muscle wet weight (in mg) to pre-injury lean body weight (in g) in T3 and T9 SCI lesioned mice at 7 dpi. (B-D) Two-Way ANOVA – (B) Injury: p=0.0067, F=9.868; Genotype/Intervention: p=0.0064, F=10.03; (C) Injury: p=0.0048, F=11.21; Genotype/Intervention: p<0.0001, F=33.48; Interaction: p=0.0116, F=8.410; (D) Injury: p=0.1784, F=1.993; Genotype/Intervention: p=0.0014, F=15.19. Data are mean ± SEM with hypothesis-relevant post hoc Holm–Šidák corrected multiple comparisons related to intervention (genotype) displayed: *p < 0.05, **p < 0.01, ***p < 0.001, non-significant (n.s.). n = 4 (T9 SCI + GR-loxP, T3 SCI + GR-loxP) or 5 (T9 SCI + SMGRKO, T3 SCI + SMGRKO) mice/group.

Taken together, our data indicate that glucocorticoids act directly on non-paralyzed and paralyzed muscle GRs to exacerbate systemic muscle wasting early after high thoracic SCI (fig. S14). Glucocorticoids also contribute to the muscle wasting observed after low thoracic SCI, although to a lesser extent. The lesion-level dependent severity of systemic muscle wasting appears to be almost entirely mediated by increased exposure of muscle to glucocorticoids during early subacute high thoracic SCI, arising through SCI lesion-level dependent hypercortisolism (24).

DISCUSSION

We identified and characterized spinal cord injury (SCI)-induced systemic skeletal muscle wasting which emerged early after injury and also affected non-paralyzed (supralesional) muscle (Fig. 1, Fig. 2). In line with the human condition (fig. S1A), the observed systemic muscle wasting exceeded the dimension of muscle atrophy rates explainable by experimental models of subacute disuse (31) and peripheral denervation (32) postulating the presence of additional and systemically acting drivers. Systemic muscle wasting impaired muscle function (Fig. 1, H to J), primarily affected “fast” type 2 myofibers (Fig. 2J) and operated by a SCI lesion-level dependent mechanism. Transcriptomic analysis (Fig. 3) together with interventional experiments (Fig. 4, Fig. 5) elucidated lesion-level dependent glucocorticoid-mediated systemic muscle wasting becoming aggravated after high T3 SCI relative to low T9 SCI. Muscle wasting was reduced i) by endogenous adrenal glucocorticoid ablation (adrenalectomy [ADX]) and ii) pharmacological glucocorticoid receptor (GR) blockade (Mifepristone [MIF] [RU-486]) (Fig. 4). SCI lesion-level dependent severity of non-paralyzed muscle wasting was completely prevented by genetic muscle GR deletion (SMGRKO) (Fig. 5). SCI lesion-level dependent cardiac muscle wasting was also observed with most pronounced effects after T3 SCI. In contrast to skeletal muscle, myocardium mass was not preserved by the SMGRKO intervention, suggesting a separate mechanism putatively involving partial denervation of the cardiac plexus (33).

Applying a bedside-to-bench approach, the reported experimental design allowed for control of several confounders difficult to fully address in a clinical study. Our data excluded potentially confounding systemic mechanisms causing skeletal muscle wasting by verifying that the SCI lesion-level dependent severity was unrelated to nutritional and hydrational status (34). Furthermore, electrophysiological assessments demonstrated fully preserved motor units (MUNE) indicating that muscle wasting was not associated with peripheral (lower motor neuron [LMN]) mechanisms. Hence, the substantial nature of early skeletal muscle wasting cannot be explained by putative mechanisms including disuse, peripheral denervation (LMN injury), transsynaptic degeneration, malnutrition, or critical illness neuromyopathy (35). Instead, unknown catabolic mechanisms emerging rapidly after SCI are implicated.

Although both applied SCI lesions cause paraplegia, gene and protein expression profiles after T3 SCI were distinct from T9 SCI implying the presence of additional, SCI lesion-level dependent systemic effects (Fig. 2, Fig. 3). T3 SCI-induced transcriptomic changes were indicative of an upregulation in glucocorticoid-signaling. It is well established that endogenous glucocorticoids potently induce muscle wasting (36) predominantly through preferential atrophy of type 2B and 2X myofibers, in alignment with the T3 SCI phenotype presented in Fig. 2J. Type 2B and 2X myofibers are presumably more susceptible to wasting due to differential myofiber type GR expression (37); GR is expressed more abundantly in “faster” (Type 2B, 2X) relative to “slower” (Type 1, 2A) myofibers (27, 38). A role for glucocorticoid-mediated muscle wasting early after T3 SCI was further corroborated by: i) elevation in circulating glucocorticoids which have also been reported in the human condition (24) and ii) preferential atrophy of GR-abundant myofibers (Types 2B, 2X) (39) (Fig. 2J).

The functional role of glucocorticoid-signaling was subsequently validated by three separate mechanistic interventions (adrenal glucocorticoid elimination by ADX, GR antagonism by MIF and skeletal muscle specific deletion of GR [SMGRKO mouse strain]). Together these approaches demonstrate that glucocorticoids act systemically directly through muscle GRs to induce wasting early after high thoracic SCI. Consequently, muscle-specific GR deletion normalized systemic muscle wasting after high T3 and low T9 thoracic SCI to equivalent degrees. The protective effect of GR deletion on lean body mass and paralyzed muscle mass could still be detected even after T9 SCI, which is characterized by only moderately elevated circulating glucocorticoids. This is in line with considerable muscle sensitivity to even modest circulating glucocorticoid fluctuations (40, 41), with severe/continued elevations drawing heavily on skeletal muscle energy stores (42, 43), and suggests that these steroid hormones can be targeted to attenuate systemic muscle wasting even at lower injury lesion levels. Nonetheless, our current findings suggest that elevated concentrations of circulating glucocorticoids early after T3 SCI, relative to T9 SCI, are responsible for the SCI lesion-level dependent severity of systemic wasting of skeletal muscle. Early biopsies of non-paralyzed muscles obtained already two days after human SCI also demonstrate glucocorticoid-mediated muscle wasting (44) and verify translational relevance. However, additional effectors are likely as the SCI-induced systemic muscle wasting phenotype cannot be fully reversed by targeting glucocorticoid signaling alone.

This study has limitations. To create a consistent and reliable model of SCI with sublesional paralysis and sympathetic decentralization, a complete spinal cord transection (TXN) injury has been applied. TXN injuries are robust and reproducible models to ensure complete interruption of motor and autonomic tracts, although it is understood that TXN SCI rarely occurs in human and is not optimal to determine locomotor recovery. However, earlier contusion SCI studies validated that TXN injuries mirror key characteristics relevant for the chosen experimental setup such as hypercortisolism (45, 46) and muscle wasting (47, 48). We induced TXN SCI at different thoracic spinal cord levels to elucidate the neurogenic nature of systemic muscle wasting, without confounding paralysis in the forelimb. A next step would expand clinical relevance by leveraging contusion SCI models, which would also allow for quantification of locomotor recovery (49), or experimental SCI at cervical spinal cord levels to i) evaluate applicability to quadriplegic SCI, ii) assess for interventional effects on survival/mortality (higher level injuries increase mortality) (50) and iii) investigate potential neurogenic mechanisms originating at more rostral spinal cord levels. This study is also limited by using male mice in all experiments. Male mice were employed i) in congruency with clinical SCI prevalence skewed towards males (~80%) (51), ii) to minimize the number of total animals utilized and iii) since the primary goal of this work was to investigate systemic muscle wasting, to avoid any potentially confounding effect resulting from heightened susceptibility to disuse atrophy in female mice (52). Future follow-up studies incorporating female mice will be necessary to assess for sex-specific differences. Regarding interventional limitations, in relation to robust ADX and SMGRKO interventional effects, the relatively muted interventional effect of MIF in paralyzed muscle was likely due to its off-target antagonistic effects (antiprogestogen) and brief half-life in rodents (1-2 hours) (53). Likewise, paralyzed muscle glucocorticoid sensitivity may increase through GR upregulation (54). In conjunction, incomplete GR antagonism by MIF was presumably insufficient to overcome glucocorticoid-mediated effects plus synergistic catabolic stimuli, including muscle disuse and upper motor neuron injury (55), in paralyzed muscle. To validate the therapeutic potential of GR antagonism, it will furthermore be necessary to test if alternative treatment designs are sufficient to ameliorate paralyzed muscle wasting after SCI. Lastly, the focus of this study was limited to skeletal muscle, although the observed early SCI lesion-level dependent wasting of myocardium (fig. S4, fig. S7D) and adipose tissue (Fig. 2E) also warrant a more detailed investigation. Particularly, the observed rapid, considerable (>50%) and persistent loss of adipose body mass starkly contrasts with the prevalent chronic obesity emerging later in patients with chronic SCI (18, 56). Whereas severe catabolism blunted both lean and adipose body mass recovery after T3 SCI the catabolic constellation was different after T9 SCI. The superior recovery after T9 SCI of lean body mass over adipose body mass may be related to the relatively low energy cost of generating lean mass relative to adipose mass (4-fold difference) (57, 58).

From a translational perspective, exploiting muscle-specific GR-mediated wasting could include a skeletal muscle-specific blockade of GR signaling, which may be achieved, for example, by applying muscle siRNA/shRNA injections to reduce GR protein synthesis. RNA therapeutics, including siRNA treatments strategies, have emerged and completed phase III trial testing (59). Alternatively, strategies to systemically protect muscles may involve glucocorticoid receptor antagonism or lowering circulating glucocorticoid concentrations. However, more research is needed to validate the skeletal muscle wasting phenotype in patients, which could be readily achieved by ultrasound (4), and to further define the pathophysiology of hypercortisolism to establish optimal therapeutic conditions early after human SCI, including therapeutic time window. The reported findings also enrich the perspective on early glucocorticoid methylprednisolone (MP) administration after human SCI (60) as exogenous MP administration rapidly exacerbates systemic muscle wasting after experimental SCI (61).

This study identifies SCI-induced skeletal muscle wasting early after SCI as a systemic phenotype, reducing muscle mass and function above (non-paralyzed) and below (paralyzed) the SCI lesion. Although both cause paraplegia, high (T3) but not low (T9) thoracic SCI disrupts central nervous system (CNS) control over endocrine adrenal gland function resulting in hypercortisolism and, in turn, causing glucocorticoid-mediated wasting of the systemic musculature. These data suggest that SCI-induced muscular changes are not limited to gradual disuse-mediated atrophy of specific sublesional paralyzed muscles but also involve systemic neuroendocrine catabolic mechanisms operated by elevated concentrations of circulating adrenal-derived glucocorticoids (fig. S14). Protecting skeletal muscle from SCI-induced systemic glucocorticoid-mediated wasting might be a therapeutic candidate to preserve muscle function, and optimize outcomes of pro-regenerative SCI therapies.

MATERIALS AND METHODS

Study design

The overall objectives were to characterize early subacute (2-7 days post injury [dpi]) systemic muscle wasting and to identify the underlying neurogenic mechanism(s) in a mouse model of paraplegic spinal cord injury (SCI). We quantified the temporal dynamics and severity of systemic muscle wasting by longitudinal body weight and composition measurements (echoMRI), longitudinal measures of neuromuscular function (grip strength & tetanic muscle contractility), and terminal ex vivo wet muscle weights. Longitudinal food intake, hydration, and neuromuscular connectivity (electromyography [EMG]) were assessed to address putative mechanisms of systemic muscle atrophy. To gain mechanistic insight we characterized early subacute SCI-induced changes in muscle biochemical composition (Immunohistochemical and colorimetric muscle protein assays), muscle gene expression changes (ribonucleic acid [RNA]-sequencing) and circulating glucocorticoid concentrations (enzyme-linked immunosorbent assay [ELISA]). We tested the in vivo interventional effects of glucocorticoid receptor (GR) antagonism (mifepristone [MIF]) and adrenalectomy (ADX) on systemic muscle mass early after SCI. We further confirmed the underlying mechanism by quantifying systemic muscle mass early after SCI in mice with skeletal muscle-specific GR knockout (SMGRKO). In initial phenotype characterization studies, injury-specific effects were identified by comparing T3 SCI, T9 SCI and Sham spine (control) groups. For interventional studies, appropriate control groups for MIF (vehicle), ADX+ALDO (sham ADX+ALDO), and SMGRKO (GR-loxP) were included.

For in vivo experiments a sample size of n ≥4 mice/group was determined a priori from historical pilot data based on identifying a SCI lesion-level dependent difference (T3 SCI vs T9 SCI) in non-paralyzed triceps brachii mass at 7 dpi as the primary outcome measure, with power (1-β) set to 0.8 and α = 0.05. The maximum endpoint of initial characterization experiments was set a priori as 28 dpi. Based on the identification of a severe SCI lesion-level dependent systemic muscle wasting phenotype emerging within one week of T3 SCI, the terminal endpoint for subsequent characterization and interventional experiments was set to 7 dpi. Data collection was stopped, and mice were euthanized if body weight loss exceeded −20% of pre-operative body weight. SCI mice exhibiting Basso Mouse Scale (BMS) scores >1 (indicating incomplete transection [TXN] SCI) (49) at 3 dpi were removed from the study and all associated data excluded from analysis. No assessment for, or adjustment of, outliers took place, and all available collected data is displayed.

Age- (8- to 10-week-old) and weight-matched (20 to 25 gram) male mice were randomized into experimental groups of 4 to 5 mice using QuickCalcs (GraphPad) and data from replicate experimental groups was pooled. Potential confounding effects related to the order of surgeries, interventions or measurements were minimized by alternating between experimental groups based on injury and intervention. Necessary blinding of outcome measures included: i) a separate investigator induced anesthesia to blind the investigator conducting EMG and functional muscle measurements, ii) for terminal tissue harvests, a separate investigator euthanized mice and weighed muscles to blind the investigator harvesting tissues and iii) muscle histology and immunohistochemistry scoring as well as colorimetric myosin heavy chain (MyHC) isoform quantification were completed by a blinded rater. In vivo and downstream in vitro experiments were conducted with at least 4 or 3 biological replicates per experimental group, respectively (specific n listed in each figure legend).

Experimental animals

Adult male C57BL/6 mice (C57BL/6NCrl) were purchased from Charles River (Wilmington, MA) and used for all experiments, except for the SMGRKO experiment. Adult female and male HSA-MCM mice (STOCK Tg[ACTA1-cre/Esr1*]2Kesr/J mice; strain #025750; RRID:IMSR_JAX:025750) and GR^fl/fl mice (B6.129S6-Nr3c1tm2.1Ljm/J; strain #012914; RRID:IMSR_JAX:012914) were purchased from The Jackson Laboratory (Bar Harbor, ME) and crossed to generate experimental SMGRKO and GR-loxP male mice. Mice were housed under conventional conditions in a pathogen-free housing facility in ventilated microisolator cages (≤5/cage), layered with corn cob bedding, on a consistent 12-hour light-dark cycle at constant temperature (20 ± 2 °C) and humidity (50 ± 20%). Ad libitum access to commercial food pellets and chlorinated reverse osmosis water was provided, except for ADX experiments (as outlined in Supplementary materials and methods section). A total of 137 mice were used for the reported studies. All conducted animal experiments are reported in conformity with the ARRIVE guidelines 2.0 and were conducted in accordance with an approved Institutional Animal Care and Use Committee (IACUC) protocol (#2016A00000070-R2).

SCI surgery and post-operative care

TXN SCI surgeries were performed as we have previously described (24). Mice were anesthetized by intraperitoneal (i.p.) injection of a ketamine (80 mg/kg) xylazine (10 mg/kg) mixture. Lubricating ointment was applied to the eyes. Dorsal fur was shaved, and underlying skin was disinfected with iodine, followed by alcohol, three times. Gentamicin (5 mg/kg) was administered subcutaneously (s.c.). During surgery, mouse body temperature was maintained at 37.5 °C using a feedback-controlled heating pad (Harvard Apparatus). A dorsal midline incision was made along the skin overlying the vertebral column and muscles overlying relevant vertebrae were dissected. Thoracic vertebrae three (T3) and nine (T9) were identified based on anatomical landmarks, including the T2 spinous process and major vascular branches. Laminectomy was unnecessary for T3 SCI, but T9 laminectomy exposed the spinal cord for T9 SCI. Dorsal periosteum and dura were removed with micro scissors and local anesthetic (0.05 mL lidocaine) was applied to the exposed cord. Using a scalpel, the spinal cord was completely transected at T3 (T3 TXN SCI) or T9 (T9 TXN SCI), with care taken not to disturb ventral or lateral dural vasculature. Bleeding was controlled by surgical sponges and cut completeness was confirmed by the presence of a gap between rostral and caudal ends of cord. Dorsal muscle layers were sutured with polyglactin dissolvable sutures (5.0-gauge). Skin was closed with sterile wound clips (7 mm). Sham spine-operated animals underwent anesthesia and spinal cord exposure at T3, without TXN (Sham spine).

Following SCI, animals were injected with 0.9% saline (2 mL s.c.) and placed into warmed cages (36 °C). Sham spine and T9 SCI animals were returned to non-warmed home cages on post-op day one. To mitigate body temperature dysregulation and hypothermia induced by higher thoracic injury levels T3 SCI mice were continuously housed in warmed cages set to maintain physiological body temperature. Bladders from SCI mice were manually expressed twice daily. Food intake, liquid intake, body weight and urinary pH were measured weekly. 0.9% saline (1 mL) was injected s.c. twice daily for 7 dpi to mitigate potential dehydration. Antibiotic coverage involved s.c. administration of gentamicin for 5 dpi (5mg/kg). Mice with alkaline urine were treated with antibiotics until urinary tract infections resolved (Baytril 2.5 mg/kg s.c.). The surgical site was inspected daily to ensure absence of infection and surgical staples were removed at 7 dpi. Pain medications can modulate the immune system and were withheld to avoid altered SCI pathophysiology (62, 63).

Statistics & reproducibility

Except for RNA-sequencing (see Supplementary Materials), all statistical analysis and graph creation was conducted utilizing Prism (Version 10.0.2, GraphPad). For all experiments, data are presented as mean ± standard error of the mean (SEM). Normality of data was assessed by Shapiro-Wilk test (data file S1) to determine appropriateness of parametric vs non-parametric analyses. Despite the presence of some non-normally distributed experimental groups in Fig. 1C, Fig. 2J, Fig. 3B and Fig 4, A and C, these data were still analyzed with parametric tests since i) there are no readily available non-parametric alternatives for these comparisons, ii) the majority of the groups in each of these experiments are normally distributed and iii) a certain degree of non-normally distributed data can be tolerated by these analyses, including analysis of variance (ANOVA), which is robust to violations of normality (64). Specific statistical tests utilized for comparisons are specified in figure legends. Data involving 2 independent groups were analyzed by unpaired t-test or Mann-Whitney test, whereas experiments involving >2 independent groups and a single variable were analyzed with One-Way ANOVA. Depending on data completeness, comparisons involving multiple independent groups and multiple timepoints or variables were analyzed with Two-Way ANOVA or mixed effects analysis with Greenhouse-Geisser correction (interaction effect was excluded if p > 0.15). For post hoc Holm–Šidák multiple comparison tests, all significant results directly associated with our hypotheses have been labelled in figures, as have non-significant findings when important to the main hypothesis. Baseline timepoints in longitudinal measures with a value of 0, as in Fig. 1C, are for illustrative purposes and were not included in statistical analyses. A two-sided p value < 0.05 was considered statistically significant throughout all analyses. Raw data for all graphs in Fig. 1–5 and fig. S1–14 are summarized in data file S2.

Data and materials availability:

All data are available in the main text or the supplementary materials. Source data is available in a supplementary file (data file S2). Transcriptomic data is publicly available through the Gene Expression Omnibus (GEO) data repository (Accession #GSE224186).