Spinal cord injury induces mitochondrial dysfunction and metabolic reprogramming in non-regenerative Xenopus laevis

Summary

Spinal cord injury (SCI) causes permanent neurological deficits in mammals. Larval Xenopus laevis regenerate after SCI, but post-metamorphic animals lose this capacity, resembling mammalian response. Because mitochondria and metabolism influence cell survival and tissue repair, we examined mitochondrial dynamics and metabolic responses after SCI in non-regenerative X. laevis. Using imaging, ultrastructural analyses, ATP measurements, and transcriptional profiling, we found that, unlike regenerative stages, non-regenerative spinal cords exhibited a delayed response with mitochondrial altered localization, reduced density, increased area, and structural abnormalities, consistent with dysfunction. Despite these defects, ATP levels increased after injury, coinciding with the upregulation of glycolytic enzymes and markers of lipid catabolism and lipid droplet formation. These findings indicate that SCI in non-regenerative X. laevis triggers metabolic reprogramming toward glycolysis and lipid utilization that may compensate for mitochondrial dysfunction. Our results identify mitochondrial and metabolic responses associated with regenerative failure and provide insight into metabolic states linked to repair.

Graphical abstract

Highlights

• •SCI induces delayed mitochondrial remodeling in non-regenerative X. laevis
• •SCI triggers mitochondrial dysfunction and vesicular-like cristae formation
• •SCI induces metabolic reprograming toward glycolysis and lipid catabolism
• •Metabolic compensation may contribute to ATP maintenance after SCI

Molecular biology; Neuroscience; Cell biology

Introduction

Spinal cord injuries (SCIs) are a devastating traumatic event that disrupts bidirectional communication between the central nervous system (CNS) and the periphery. This interruption leads to paralysis below the site of injury, causing motor, autonomic, and sensory dysfunctions.^1,2 The global incidence and prevalence of SCI continue to rise,³ and despite substantial research efforts, effective treatments for such injuries remain elusive. This lack of therapeutic options highlights the urgent need to deepen our understanding of the cellular and molecular mechanisms underlying SCI, which could help identify potential therapeutic strategies.

In mammals, including humans, SCI results in the establishment of a non-permissive environment for regeneration. The initial mechanical insult disrupts tissue integrity by damaging axons and inducing cell death near the lesion site.⁴ This leads to rapid biochemical imbalances, triggering excitotoxicity⁵ and the production of free radicals,^6,7 which exacerbate damage to adjacent regions of the spinal cord. Consequently, the activated cellular response focuses primarily on preventing damage propagation rather than facilitating regeneration. Key events following SCI include the activation of a long-lasting pro-inflammatory response,⁸ the proliferation of oligodendrocyte precursor cells, astrocytes, and ependymal cells, and their differentiation into new astrocytes.⁹ These processes culminate in the formation of a glial scar, a structure that impedes damage propagation but hampers regeneration.^10,11

Other vertebrates, such as teleost fish, urodele amphibians, and the larval stages of anuran amphibians, possess remarkable regenerative capacities and can recover sensory and motor functions after SCI.¹² Comparative studies of regenerative and non-regenerative organisms are instrumental in identifying anti- and pro-regenerative factors, which could serve as targets for new therapies. Xenopus laevis serves as a powerful model for this purpose due to its developmental shift in regenerative potential.^13,14,15 During larval development (Nieuwkoop and Faber [NF] stages 48–52), the organism possesses a high regenerative capacity, hereafter referred to as regenerative stages (R-stages). In this stage, SCI triggers a coordinated response characterized by a rapid increase in proliferation of Sox2⁺ neural-stem progenitor cells (NSPCs),^15,16,17,18 their migration to the lesion site,¹⁵ and differentiation into glial cells and neurons.^15,16 Conversely, this capacity is lost following metamorphosis (NF stage 66), entering the non-regenerative stages (NR-stages). The NR-stage response closely mirrors mammalian SCI, and in comparison to R-stages, NSPCs' proliferative response is both delayed and attenuated,^15,16 and differentiate into astrocytes that accumulate in the injury gap to form the glial scar.¹⁶

The molecular mechanisms driving the different cellular responses activated in regenerative versus non-regenerative organisms after SCI remain poorly understood. Mitochondria and metabolic pathways changes have emerged as regulators of various cellular processes activated after SCI, including cell death, proliferation, and differentiation.¹⁹ The polarization state of the mitochondrial membrane plays a key role in determining cell survival or death when they are exposed to excitotoxicity and free radicals.^20,21,22 Additionally, an increase in proliferation rate is associated with a transition from mitochondrial oxidative phosphorylation (OXPHOS) to glycolysis as the primary source of ATP production.²³ After stem cells divide, the cell that inherits the older mitochondria and transitions its metabolism from glycolysis to OXPHOS differentiates, while the one with newer mitochondria and glycolytic metabolism remains as a stem cell.^24,25,26

Previous studies have described changes in mitochondrial and metabolic processes during regenerative events, including Xenopus tropicalis tail amputation,^27,28,29 planarian regeneration following amputation,³⁰ and axolotl limb regeneration.³¹ In the context of SCI, bulk RNA sequencing and proteomic analyses in Xenopus laevis revealed differential expression of mitochondrial and metabolic pathway components between R- and NR-stages.^32,33 More recently, we demonstrated that spinal cord regeneration in R-stage X. laevis involves early and coordinated mitochondrial remodeling in NSPCs, including a transient metabolic shift toward glycolysis, along with changes in mitochondrial localization, number, and morphology.³⁴ Together, this evidence suggests a potential role for this metabolic switch in the regenerative process.

Although mitochondrial dysfunction has been reported in mammals following SCI,^35,36 the accompanying metabolic adaptations remain poorly characterized. Given the established role of mitochondria in supporting regeneration, it is critical to investigate their behavior in non-regenerative contexts. To identify processes that may underlie regenerative failure, we examined mitochondrial dynamics and metabolic responses in NR-stage animals. In contrast to R-stages, we observed a delayed and biphasic mitochondrial response in NR-stage spinal cords, characterized by transient changes in mitochondrial localization and morphology, followed by mitochondrial dysfunction. Notably, these changes were accompanied by a surprising increase in ATP levels and transcriptional evidence of a shift toward glycolysis and lipid catabolism. Together, our findings suggest that SCI in NR-stage X. laevis induces delayed mitochondrial remodeling and metabolic reprogramming, which contrasts with the rapid and organized response seen in R-stages.

Results

SCI induces a transient change in localization and number of mitochondria in ependymal cells surrounding the spinal cord central canal

Recently, we described the localization of mitochondria in the spinal cord of R-stage Xenopus laevis and reported changes in mitochondrial number, distribution, and phenotype in response to SCI.³⁴ In the present study, we aimed to characterize mitochondrial localization and its response to SCI in non-regenerative (NR-stage, NF 66, juvenile frogs) Xenopus laevis. Longitudinal spinal cord cryosections were obtained from NR-stage animals, and the region encompassing the fifth to the seventh vertebrae in the rostral (anterior) to caudal (posterior) axis (Figure 1A, boxed area) was specifically analyzed. Immunodetection of radial glia-like ependymal cells (Sox2+) and neurons (NeuN+) revealed that Sox2+ cells form three to four layers surrounding the central canal (Figure 1Ai), although fluorescence intensity was lower than that observed in R-stages (data not shown), consistent with previous reports.^15,37 In contrast, NeuN+ neurons occupied a broader area, and more peripherally, of the spinal cord gray matter (Figure 1Ai). A region, considering the central canal and a couple of cells surrounding the central canal, equivalent to the boxed area in Figure 1Ai, was analyzed by transmission electron microscopy (TEM) (Figure 1Aii). Mitochondria were predominantly located in the apical region facing the central canal (Figure 1Aiii and Aiv, red arrowheads).

Figure 1.

Mitochondrial localization and number changes in ependymal cells surrounding the spinal cord central canal after SCI

(A) Representative illustration of Xenopus laevis NF 66 indicating the rostral-caudal axis, and an example of immunofluorescence (Ai) and electron microscopy (Aii-Aiv) images. Ai) Uninjured spinal cord immunostained with the neuronal marker NeuN in magenta, the NSPC marker Sox2 in yellow, and the nucleus with Hoechst in blue. All scale bars are 50 μm. Aii) Electron micrograph images of the ependymal cells surrounding the central canal and the mitochondria within them.

(B) Representative electron microscopy images of ependymal cells surrounding the central canal in uninjured (Un) animals and at different timepoints after SCI (hpt = hours post-transection). Red arrowheads indicate mitochondria.

(C) Representative image of an ependymal cell divided into three portions, basal (coral arrowheads), medial (yellow arrowheads), and apical (green arrowheads), as a reference for the quantification of mitochondrial positioning.

(D) Proportion graph of mitochondrial positioning in the uninjured spinal cord and at different time points after SCI.

(E) Plot for the number of mitochondria per cellular area in the uninjured spinal cord and at different time points after SCI. Scale bars are 2 μm unless indicated in the figure. All graphs are presented as Mean ± SEM from three independent biological replicates. ∗(p < 0.05), determined by nested one-way ANOVA, indicates significance compared with uninjured controls.

Then, the localization and number of mitochondria were analyzed in uninjured animals and at different times after injury, 6 h and 1-, 2-, and 6-day post-transection (hpt and dpt, respectively). The analysis began at 6 hpt, as this coincides with the timing at which mitochondrial changes can be detected in R-stages,³⁴ and until 6 dpt, as it was previously described that this timing corresponds to the timing at which more extensive changes were observed in transcript levels related to metabolic processes.³² Mitochondria were detected in the cells surrounding the central canal (Figure 1B, red arrowhead), and each cell was divided into three zones: apical, from the central canal to the beginning of the nucleus (green arrowheads); medial, considering all the area occupied by the nucleus (yellow arrowheads); and basal, from the end of the nucleus to the inner plasma membrane (coral arrowheads) (Figure 1C), and the localization of the mitochondria was analyzed (Figure 1D). Uninjured animals present 80% of the mitochondria in the apical zone, 18% in the medial, and 2% in the distal zone. This localization was unaltered at early time points after SCI, whereas at 2 dpt the mitochondria were redistributed, and the percentage of mitochondria in the apical zone decreased to 58%, while 28% was localized to the medial and 15% to the distal zone. The mitochondrial localization percentage returned to basal levels at 6 dpt (Figure 1D). The number of mitochondria was analyzed in each cell section, limited by the plasma membrane adjacent to the central canal, and normalized by the cell section area (Figure 1C, black line). The number of mitochondria per cell section area was temporarily decreased at 2 dpt, while it returned to basal levels at 6 dpt (Figure 1E). Therefore, we observed a transient change in the localization and number of mitochondria at 2 dpt, coincident with the timing where there is more damage in the tissue, and discontinuity or damage to the cell membrane can be observed.³⁷

Spinal cord injury triggers transient mitochondrial fusion and fission adaptations in ependymal cells surrounding the spinal cord central canal in NR-Stage Xenopus laevis

To further investigate mitochondrial responses to SCI in NR-stage Xenopus laevis, we analyzed mitochondrial area in TEM images from cells surrounding the central canal in uninjured animals and at different times after injury, 6 hpt and 1, 2, and 6 dpt (Figure 2A). A significant increase in mitochondrial area was detected at 2 dpt, with a further enlargement observed at 6 dpt (Figure 2B). Additionally, at 6 dpt, several mitochondria appeared to share continuous outer membranes while maintaining independent inner membranes (Figure 2C, arrowheads), suggesting either interrupted fusion of the inner membrane or ongoing mitochondrial fission. Moreover, some mitochondria exhibited segregation of their content into compartments with different mitochondrial morphologies: mitochondria with an enlarged and swollen matrix and sparse cristae were positioned adjacent to others displaying more defined cristae and matrix granules (Figure 2C, right panel). These changes suggest alterations in mitochondrial dynamics, particularly in the balance between fusion and fission processes.

Figure 2.

SCI triggers mitochondrial mass increase and fusion and fission proteins adaptations

(A) Representative electron microscopy images of ependymal cells' mitochondria at different time points after SCI. All scale bars are 0.5 μm.

(B) Plot for the mitochondrial area in the uninjured spinal cord and at different time points after SCI. The graph is presented as Mean ± SEM from three independent biological replicates. ∗(p < 0.05) and ∗∗(p < 0.01) determined by nested one-way ANOVA indicate significance compared with uninjured controls.

(C) Representative images of mitochondria observed at 6 dpt, showing a continuous outer membrane while maintaining an independent inner membrane. Arrowheads show the independent inner membrane. Scale bars are 0.5 μm.

(D–G) western blot quantifications of mitochondrial proteins Mfn2 (D), OPA1 (E), Fis1 (F), and Tom20 (G), normalized to α-Tubulin (bottom blot) and to uninjured animals.

Graphs are shown as the Mean ± SEM. from 3 independent biological replicates. ∗(p < 0.05) and ∗∗(p < 0.01) determined by a Kruskal-Wallis test, followed by Dunn’s multiple comparison test, indicate significance compared with uninjured controls.

For clarity and improved figure presentation, the western blot membrane images were cropped above and below the band corresponding to the molecular weight of interest. Full, uncropped membrane images are provided upon request. See also Figure S1.

To determine whether these morphological changes were associated with the altered expression of proteins involved in mitochondrial dynamics,^38,39 we performed western blot analysis on spinal cord lysates collected at 0 (uninjured), 1, 2, and 6 dpt. Protein levels of Mitofusin-2 (Mfn2), a key mediator of mitochondrial outer membrane fusion,³⁸ were increased at 2 dpt and returned to baseline by 6 dpt (Figure 2D). This transient upregulation coincides with the observed increase in mitochondrial area. OPA1, which regulates inner membrane fusion,³⁸ showed no significant changes across the analyzed time points (Figure 2E).

In contrast, the fission protein Fis1 was upregulated at 6 dpt (Figure 2F), a time coincident with the appearance of fission-like mitochondrial structures observed by TEM, implying a subsequent shift toward mitochondrial fission. Additionally, the mitochondrial import receptor Tom20 levels, commonly used as a marker of mitochondrial mass, were increased at 6 dpt (Figure 2G). To determine whether the observed changes in Tom20 expression reflect mitochondrial biogenesis, we re-analyzed publicly available RNA-seq data from injured and uninjured spinal cords,³² aligning reads to the updated Xenopus laevis transcriptome (xenbase: X. laevis v10.1), and focusing specifically on the transcript levels of PGC-1α and TFAM, two key regulators of mitochondrial biogenesis. However, no significant changes in their expression levels were detected (Figure S1). These findings suggest that the increase in Tom20 is unlikely to be driven by mitochondrial biogenesis but rather may reflect alterations in the mitochondrial outer membrane.

Together, these results indicate a dynamic regulation of mitochondrial remodeling following SCI in NR-stage spinal cords. Early time points are characterized by increased expression of fusion-related proteins, whereas later stages show increased expression of fission-related proteins.

Mitochondria change their phenotype and display internal vesicular-like cristae following SCI in NR-stage Xenopus laevis

To further characterize mitochondrial responses to SCI in NR-stage Xenopus laevis, we analyzed mitochondrial cristae ultrastructure in TEM images from ependymal cells surrounding the central canal, at multiple time points: uninjured (0 hpt), 6 hpt, and 1, 2, and 6 dpt, as mitochondrial morphology adapts in response to environmental changes and cell requirements.⁴⁰ Mitochondrial morphologies were classified into three distinct phenotypes: two of them already described, and observed in R-stage animals^34,41 orthodox, with well-defined cristae and dense matrix; swollen, characterized by a dilated matrix and reduced cristae; and a hybrid phenotype, in which mitochondria exhibited distinct regions of both swollen and orthodox morphology within the same organelle (Figures 3A and 3B).

Figure 3.

Mitochondria change their phenotype and display internal vesicular-like cristae following SCI

(A) Representative electron microscopy images of the main type of mitochondrial phenotype found on NSPCs at different time points after SCI. Scale bars are 0.5 μm.

(B, left) Example images of the different observed mitochondrial phenotypes.

(B, right) Proportion graph for the different mitochondrial phenotypes found in NSPCs before and at different time points after SCI.

(C) Bar graph showing the percentage of mitochondria with internal vesicular-like cristae in the uninjured spinal cord and at different time points after SCI.

(Ci and Cii) Representative image of mitochondria observed at 6 dpt, showing internal vesicular-like cristae (red arrowheads). All scale bars are 0.5 μm. Graphs are shown as the mean ± SEM. from 3 independent biological replicates. ∗∗(p < 0.01) and ∗∗∗∗(p < 0.0001), from a Kruskal-Wallis test, followed by Dunn’s multiple comparison test, indicate significance compared with uninjured controls.

Quantification of mitochondrial phenotypes revealed that, under uninjured conditions, ∼80% of mitochondria exhibited an orthodox morphology, ∼2% were swollen, and ∼18% displayed the hybrid phenotype (Figure 3B). This distribution remained essentially unchanged at 6 hpt and 1 dpt. However, a marked shift in mitochondrial phenotype was observed at 2 dpt, where the proportion of orthodox mitochondria decreased to ∼35%, while swollen mitochondria increased to ∼40%, and hybrid mitochondria to ∼25%. These alterations became even more pronounced at 6 dpt, with only ∼20% of mitochondria maintaining an orthodox morphology, while swollen and hybrid phenotypes rose to ∼35% and ∼45%, respectively (Figure 3B).

Interestingly, within the swollen and hybrid mitochondria, we frequently observed internal vesicular-like cristae (Figure 3Ci, ii). Quantification of mitochondria containing vesicular-like cristae structures revealed a significant increase at 2 dpt, which was further elevated at 6 dpt (Figure 3C), coinciding with the time points showing the highest proportion of swollen and hybrid mitochondria.

Together, these data indicate that SCI induces progressive morphological remodeling of mitochondria in the NR-stage spinal cord, including the emergence of hybrid phenotypes and increased internal vesicular-like cristae formation. These changes may reflect disrupted mitochondrial quality control or attempts at adaptation in the non-regenerative environment.

SCI results in dysfunctional mitochondria in NR-stage Xenopus laevis

The presence of vesicular-like cristae has been described in other models related to hypoxia, oxidative stress, and aging.^42,43,44 Therefore, considering the mitochondrial phenotypic changes and appearance of vesicular-like cristae, we decided to evaluate if these changes observed after SCI in NR-stage were accompanied by mitochondrial dysfunction.

As an initial approach, we evaluated mitochondrial stress using the MitoSOX probe, which detects mitochondrial superoxide production. To this end, we incubated unfixed tissue sections collected at different time points post-injury and quantified fluorescence intensity. We observed a significant increase in MitoSOX signal at 6 dpt (Figure S2), indicating elevated mitochondrial oxidative stress at this stage.

However, because MitoSOX primarily reflects mitochondrial stress rather than functional capacity, we further assessed mitochondrial function using cytochrome c oxidase (COX) and succinate dehydrogenase (SDH) double-labeling histochemistry. COX is a mitochondrial DNA-encoded complex IV subunit, while SDH is a nuclear-encoded complex II subunit. Consequently, intact COX activity requires functional mitochondrial DNA and proper mitochondrial assembly, whereas SDH activity remains independent of mitochondrial genome integrity.⁴⁵ Cells with fully functional mitochondria stain brown due to combined COX and SDH activity, whereas cells with dysfunctional mitochondria exhibit a bluish-purple stain indicative of reduced COX activity.

In longitudinal cryosections of uninjured NR-stage animals, cells adjacent to the central canal consistently exhibited a homogeneous brown staining pattern, reflecting intact mitochondrial respiratory function (Figure 4A). This pattern remained largely unchanged at early time points post-transection (6 hpt, 1 and 2 dpt). However, by 6 dpt, a distinct shift in staining was observed at and around the injury site: Multiple cells exhibited a bluish-purple coloration, indicating COX-deficient, SDH-positive profiles (Figure 4B, red arrowheads). These changes were most prominent within the lesion core and in regions caudal to the injury.

Figure 4.

SCI results in dysfunctional mitochondria

(A) Schematic representation of Xenopus laevis NF 66 indicating the rostral-caudal axis (R–C) and the dorsal-ventral axis (D–V). Zoom represents the slices sections on the D-V axis shown in (C).

(B) Representative images of COX/SDH double labeling histochemistry of longitudinal sections of the uninjured spinal cord and at different time points after SCI. Dashed red lines indicate the lesion site.

(C) Representative images of COX/SDH double-labeling histochemistry of longitudinal sections of the 6 dpt spinal cord on different D-V axis slides (+50 to −150 μm compared to the image presented in B).

All scale bars are 500 μm. See also Figure S2.

To map the spatial extent of mitochondrial dysfunction, we analyzed serial sections across the dorsoventral axis of the spinal cord, from +50 μm dorsal to −150 μm ventral relative to the section shown in Figure 4B (Figure 4C). This analysis revealed that COX deficiency was not restricted to a single plane but extended across multiple tissue depths, predominantly surrounding the injury epicenter but also propagating several hundred microns beyond it, both along the rostro-caudal and dorsoventral axes.

Taken together, these results indicate that mitochondrial dysfunction begins to arise around 2 dpi and increases over time until 6dpi in NR-stage, coinciding with the sustained alterations in mitochondrial morphology and the emergence of hybrid and swollen phenotypes described previously.

Glycolytic reprogramming following mitochondrial dysfunction

Given the significant alterations in mitochondrial morphology and the emergence of mitochondrial stress and dysfunction observed through mitoSOX and COX/SDH histochemistry, respectively, we next assessed whether SCI in the NR stage affects cellular energy production. ATP levels were measured in spinal cord lysates from uninjured animals at 6 hpt, 1, 2, and 6 dpt. Oligomycin, an inhibitor of mitochondrial ATP synthase (ATPase), was used as a control to verify mitochondrial contribution to ATP production in uninjured animals.

ATP levels remained unchanged during the early phases post-injury (6 hpt to 2 dpt), consistent with the relatively preserved mitochondrial morphology and function observed at those time points. Surprisingly, at 6 dpt, a time point coinciding with pronounced mitochondrial swelling, hybrid morphologies, and COX dysfunction, we detected a significant increase in ATP levels (Figure 5A). The observed increase in ATP levels may reflect a compensatory response driven by metabolic reprogramming, such as the activation of alternative ATP-producing pathways, including glycolysis or fatty acid β-oxidation (FAO), aimed at maintaining cellular energy homeostasis despite mitochondrial impairment.⁴⁶ Alternatively, it could indicate a pathological state, potentially associated with increased tissue damage and the accumulation of extracellular ATP.⁴⁷

Figure 5.

SCI leads to a delayed activation of glycolysis and ATP increase

(A) Bar graph for ATP level changes at different time points after SCI compared to the uninjured spinal cord. Oligomycin (Olig) as a control of mitochondrial ATP production inhibition.

(B) Bar graph for mRNA levels of the glycolytic enzyme hexokinase-2 (hk2) at different time points after SCI, normalized to the housekeeping eeF1α1 and compared to uninjured animals (considered as 0 for non-fold-change data).

(C) Bar graph for mRNA levels of the regulatory glycolytic enzyme 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase-1(pfkfb1) at different time points after SCI, normalized to the housekeeping eeF1α1 and compared to uninjured animals (considered as 0 for non-fold-change data).

(D) Bar graph for mRNA levels of the glycolytic enzyme pyruvate kinase (pklr) at different time points after SCI, normalized to the housekeeping eeF1α1 and compared to uninjured animals (considered as 0 for non-fold-change data).

Graphs are shown as the mean ± SEM from 3 independent biological replicates. ∗(p < 0.05); ∗∗(p < 0.01); and ∗∗∗(p < 0.001), from a Kruskal-Wallis test, followed by Dunn’s multiple comparison test, indicates significance compared with uninjured controls. See also Figure S3.

To explore if there is a metabolic reprogramming, we examined the transcriptional regulation of key glycolytic enzymes. In glycolysis, three irreversible enzymatic steps, catalyzed by hexokinase (Hk), phosphofructokinase-1 (Pfk1), and pyruvate kinase (Pkr), are rate-limiting and critical for flux control.⁴⁸ Among these, Pfk1 is strongly modulated by fructose 2,6-bisphosphate, synthesized by the bifunctional enzyme 6-phosphofructo-2-kinase/fructose 2,6-bisphosphatase (Pfkfb), which acts as a major regulator of glycolytic activity.⁴⁹ Therefore, we performed RT-qPCR analysis of hk2, pfkfb1, and pklr transcript levels at 0, 2, 6, 12 hpt, 1, 2, and 6 dpt. hk2 showed a sustained increase starting at 6 hpt and remained elevated through all subsequent time points (Figure 5B). In contrast, pfkfb1 expression remained unchanged (Figure 5C), while pklr expression was significantly upregulated at 6 dpt (Figure 5D), coinciding with the observed increase in ATP levels. Together, the upregulation of hk2 and pklr support enhanced glycolytic activity as a likely contributor to ATP production under conditions of impaired mitochondrial function.

To further support this interpretation, we analyzed RNA-seq data focusing on genes associated with glycolysis and the tricarboxylic acid (TCA) cycle. We found that glycolysis-related transcripts were increased at 6 dpt, whereas TCA cycle-related transcripts remained largely unchanged (Figure S3). Additionally, we examined PDK4, a key regulator of pyruvate entry into mitochondria through the inhibition of the pyruvate dehydrogenase complex,⁵⁰ which links glycolysis to the TCA cycle. Notably, PDK4 transcript levels were increased, suggesting a potential reduction in pyruvate flux into mitochondrial oxidation. Taken together, these results indicate that the increase in glycolytic activity does not translate into enhanced mitochondrial respiration, but rather supports a metabolic shift toward glycolysis, potentially reflecting a partial decoupling of mitochondrial oxidative metabolism.

Lipid droplet accumulation and activation of lipid catabolism suggest a shift toward fatty acid metabolism following SCI

While analyzing TEM images of NR-stage Xenopus laevis spinal cords, we observed the presence of lipid droplets (LDs) in ependymal cells adjacent to the central canal at 6 dpt, particularly in regions surrounding the injury site (Figure 6A, arrowheads). Given that the observed increase in ATP levels at 6 dpt could not be fully explained by glycolysis alone, we hypothesized that alternative energy sources, such as FAO, might be contributing to ATP production during this phase.

Figure 6.

SCI leads to lipid droplet accumulation and lipid catabolism

(A) Representative electron microscope images of lipid droplets (red arrowheads) from ependymal cells surrounding the central canal at 6 dpt. Scale bars are 2 μm.

(B) Bar graph for the mRNA levels of the mitochondrial fatty acid transporter carnitine palmitoyltransferase I (cpt1) at different time points after SCI, normalized to the housekeeping eeF1α1 and compared to uninjured animals (considered as 0 for non-fold-change data).

(C) Bar graph for the mRNA levels of the mitochondrial fatty acid transporter carnitine palmitoyltransferase II (cpt2) at different time points after SCI, normalized to the housekeeping eeF1α1 and compared to uninjured animals (considered as 0 for non-fold-change data).

Graphs are shown as the mean ± SEM from 3 independent biological replicates. ∗∗(p < 0.01) from a Kruskal-Wallis test, followed by Dunn’s multiple comparison test, indicates significance compared with uninjured controls.

(D) Spinal cord longitudinal sections stained with Oil-red. Scale bars are 100 μm. See also Figure S4.

To explore this possibility, we performed Gene Ontology (GO) enrichment analysis of differentially expressed transcripts obtained from the RNA-seq data. This analysis revealed a significant upregulation of pathways related to lipid utilization and transport, including “cellular response to lipid,” “lipid homeostasis,” “unsaturated fatty acid metabolic process,” “cellular lipid catabolic process,” and “positive regulation of lipid and cholesterol transport” (Figure S4A). In contrast, downregulated GO terms were primarily associated with lipid and cholesterol biosynthesis, such as “cholesterol biosynthetic process,” “sterol biosynthetic process,” and “lipid biosynthetic process” (Figure S4B). Together, these data suggest a post-injury shift from lipid synthesis toward lipid mobilization and degradation, favoring fatty acid catabolism.

Among the upregulated transcripts, we identified key metabolic regulators, including carnitine palmitoyl transferase 1 (cpt1), the rate-limiting enzyme for mitochondrial fatty acid import in β-oxidation,⁵¹ PPARγ, a nuclear receptor that governs lipid metabolism,B⁵² and perilipin (plin), a protein involved in LD dynamics and mitochondrial interactions (Figure S4C). To validate and temporally characterize the potential activation of FAO, we performed RT-qPCR for cpt1, the limiting enzyme, and cpt2, another enzyme that participates in fatty acid transport toward the mitochondria, but is not considered a limiting enzyme, on spinal cord tissue collected at 0, 2, 6, 12, and 1, 2, and 6 dpt. cpt1 transcript levels showed a modest increase at 2 dpt and a robust upregulation by 6 dpt (Figure 6B). In contrast, cpt2 showed no changes in the transcript levels among all the analyzed timepoints (Figure 6C). These results support the activation of the β-oxidation pathway in the later stages of SCI.

To determine whether this metabolic shift was spatially localized to cells near the central canal and injury site, we performed Oil-Red staining, used for staining lipids, in longitudinal spinal cord sections from uninjured animals and at 6 hpt, 1 dpt, 2 dpt, and 6 dpt. LD accumulation was not detected in uninjured samples or early post-injury stages but became evident at 6 dpt (Figure 6D). LDs appeared abundantly in cells flanking the injury, but they extended both rostrally and caudally from the lesion epicenter (Figures 6D and 6E).

Collectively, these findings demonstrate a post-injury accumulation of LDs and transcriptional activation of lipid catabolic pathways in NR-stage spinal cords. The temporal correlation between increased cpt1 transcript levels, LD accumulation, and elevated ATP levels suggests a compensatory shift toward β-oxidation as an alternative energy source. This substrate switch may represent an adaptive response to mitochondrial dysfunction, although its impact on regenerative capacity remains to be.

Discussion

Mitochondria and metabolism play a crucial role in supporting tissue regeneration; however, their behavior under non-regenerative conditions remains poorly understood. To identify mechanisms underlying regenerative failure, we investigated mitochondrial dynamics and metabolic responses following SCI in Xenopus laevis NR-stages. NR-stage animals exhibited a delayed mitochondrial response marked by a change in their localization from the apical to the medial and basal zones, a reduced number of mitochondria per cell area, and increased mitochondrial area. These alterations may reflect changes in mitochondrial dynamics, as they were associated with sequential upregulation of mitochondrial fusion (Mfn2) and fission (Fis1) marker expression. Additionally, mitochondria exhibited a shift from an orthodox to a swollen or hybrid morphology, along with internal vesicular-like cristae and signs of dysfunction. Despite evidence of impaired mitochondrial function, ATP levels were elevated. This was accompanied by increased transcript levels of glycolytic enzymes such as hk2 and pklr. We also observed a transcriptional profile consistent with enhanced lipid metabolism, and confirmed increased transcript levels of cpt1, while describing the formation and accumulation of LDs. Together, our results suggest that in contrast to the rapid and coordinated mitochondrial response seen in regenerative larvae,³⁴ NR-stage X. laevis triggers delayed mitochondrial remodeling and metabolic reprogramming toward glycolysis and lipid catabolism in response to SCI.

In uninjured NR-stages X. laevis frogs, mitochondria were mainly localized to the apical region of the ependymal cells surrounding the central canal. This localization is consistent with the higher energy demands of this region, since these cells possess multiple cilia at their apical surface, which facilitate the flow of cerebrospinal fluid through the central canal.⁵³ This pattern is also consistent with mitochondrial localization reported in R-stages³⁴ and other animal models.^54,55 Following SCI, this polarized mitochondrial distribution was reduced, which could be explained by cell damage reported at this time point in ependymal cells surrounding the central canal.¹⁶ Such damage may also lead to ciliary loss, as has been observed after traumatic brain injury.⁵⁶ Moreover, we observed a decrease in the number of mitochondria per cell area at 2 dpt, which could be partially explained by the loss of intracellular components, including organelles, known to occur as a consequence of severe cellular damage.¹⁶

Mitochondrial dynamics are essential for maintaining both mitochondrial morphology and function, particularly in damaged tissues and disease states.^57,58 Under mild or early stress conditions, mitochondria can undergo stress-induced mitochondrial hyperfusion, a protective response that enhances mitochondrial connectivity and supports bioenergetic function.^59,60 Mitochondrial fusion enables the merging of healthy and damaged mitochondria, allowing the exchange of proteins and mitochondrial DNA to preserve mitochondrial function.^57,58 Increased fusion has also been reported to reduce apoptotic cell death after tissue damage.⁶¹ Consistent with this, we observed a transient increase in Mfn2 protein levels at 2 dpt, accompanied by an increase in mitochondrial area and a reduction in mitochondrial number, suggesting an early fusion-mediated adaptive response following injury.

Conversely, mitochondrial fission is critical for the removal of damaged mitochondrial segments after injury.⁶² Under conditions of severe stress, mitochondrial membrane potential is lost, leading to the degradation of mitofusins (Mfns) and inhibition of mitochondrial fusion. In parallel, damaged mitochondria are segregated through fission and targeted for degradation.^63,64 Our data show that at 6 dpt, mitochondrial damage and dysfunction are increased, accompanied by Mfn2 protein levels returning to baseline and a significant increase in Fis1 protein levels. These changes suggest reduced fusion and enhanced mitochondrial fission, consistent with a response to severe stress.

Our results support a model in which mitochondrial remodeling follows a dynamic trajectory after injury, characterized by an early adaptive fusion response (2 dpt) followed by a later stage with increased mitochondrial fission (6 dpt). This biphasic mitochondrial response is consistent with findings in other models of SCI, where mitochondrial fusion is initially upregulated, followed by increased fission.^65,66 Functionally, early fusion may promote complementation between damaged and healthy mitochondria, supporting cellular energy balance and limiting apoptotic signaling (55, 56, 59), whereas subsequent fission facilitates the segregation and removal of dysfunctional mitochondrial components (60).

When comparing mitochondrial dynamics between regenerative and non-regenerative spinal cord models, marked differences emerge. In the NR-stages analyzed in this study, the mitochondrial adaptive response is delayed and less efficient, leading to greater mitochondrial damage near the injury site. In contrast, our previous work in R-stages showed a rapid but transient increase in fission-related transcripts within 24 h post-transection.³⁴ These observations highlight distinct mitochondrial responses between regenerative and non-regenerative conditions. In NR stages, the extent of mitochondrial damage likely necessitates a more sustained and robust remodeling response, involving both fusion and fission processes, in an attempt to reestablish mitochondrial and cellular homeostasis.

Notably, TEM analysis at 6 dpt revealed a heterogeneous mitochondrial population. While fission-like events were readily observed, particularly in mitochondria displaying an orthodox phenotype, a substantial fraction of mitochondria exhibited increased area, often associated with swollen morphology and the presence of vesicular-like cristae. Vesicular-like cristae have previously been associated with mitochondrial diseases^67,68,69 and apoptosis.^70,71 Several myopathies are characterized by mitochondria displaying vesicular or onion ring–like cristae, commonly linked to mitochondrial dysfunction.^67,68,69 Similar abnormalities occur upon downregulation of mitochondrial fusion proteins such as OPA1, which regulates cristae conformation by controlling junction shape and size. Suppression of OPA1 results in vesicular-like cristae comparable to those described here.⁷² Comparable effects have also been reported following knockdown of proteins that regulate OPA1, such as prohibitins⁷³ or mitofilin, a key cristae-structuring protein.⁷⁴ These changes are frequently accompanied by mitochondrial dysfunction and increased apoptosis.^73,74

Furthermore, vesicular-like cristae have been linked to late-stage apoptosis, appearing after cytochrome c release and loss of mitochondrial membrane potential.^70,71 Our results revealed the appearance of vesicular-like cristae as early as 2 dpt, with a marked increase at 6 dpt, indicating progressive structural deterioration. Consistent with this, our mitoSOX and COX/SDH assays demonstrated mitochondrial stress and dysfunction at 6 dpt in juvenile frogs. Taken together, these findings suggest that the vesicular-like cristae observed in our study reflect increased mitochondrial damage in juvenile frogs.

The coexistence of fission-like events and enlarged mitochondria at later stages contrasts with the reduction in Mfn2 expression and indicates that mitochondrial enlargement is unlikely to reflect ongoing fusion. Instead, the continued increase in mitochondrial area in the absence of sustained Mfn2 upregulation is more consistent with mitochondrial swelling and structural dysfunction, both hallmarks of mitochondrial damage.^63,64 This further suggests that mitochondrial dynamics become uncoupled from functional recovery, reflecting a shift from adaptive remodeling to progressive mitochondrial dysfunction in non-regenerative conditions.

SCI alters mitochondrial morphology and functionality in cells located at and next to the injury site. Nevertheless, regenerative animal models such as zebrafish⁷⁵ recover mitochondrial function and successfully regenerate the damaged area. In contrast, non-regenerative models such as mice³⁵ and rats³⁶ fail to recover mitochondrial functionality, which exacerbates tissue damage and impairs regeneration. We observe a similar pattern when comparing R- and NR-stage X. laevis. In our previous work, we showed that in R-stage animals, there is an early and transient shift from an orthodox to a swollen mitochondrial phenotype. This shift is associated with metabolic reprogramming, in which mitochondria become uncoupled, and glycolysis is upregulated, while overall mitochondrial functionality is maintained. Notably, mitochondrial morphology recovers to its original state by 2 dpt.³⁴ In contrast, in the present work, we observed that NR-stage animals show a persistent shift toward swollen and hybrid mitochondrial phenotypes, which remain abundant up to 6 dpt, with no apparent recovery of the initial mitochondrial morphology. This phenotypic shift was also associated with metabolic reprogramming, but unlike in R-stages, both glycolysis and lipid catabolism appeared to be upregulated. Mitochondrial cristae are formed by the folding of the inner mitochondrial membrane, a highly dynamic structure. As such, cristae morphology can change in response to cellular stress or environmental cues, serving as an adaptive mechanism.^41,42,70 In this study, we observed mitochondria with vesicular-like cristae in juvenile frogs. Notably, this phenotype was not detected in tadpoles after SCI,³⁴ suggesting it is characteristic of the non-regenerative response.

Here, we observed mitochondrial dysfunction, particularly at 6 dpt. However, ATP levels remained stable during the early stages after SCI and increased by 6 dpt. Under normal conditions, the nervous system relies heavily on glucose as its primary energy source, generating ATP mainly through glycolysis and OXPHOS.⁷⁶ The increase in ATP levels may suggest that cells within the damaged area engage alternative metabolic pathways to maintain energy balance and ATP production.

We detected increased levels of hk2 and pklr mRNA, indicating that glycolysis may be upregulated to partially compensate for mitochondrial dysfunction.⁴⁸ RNA-seq analysis further supported this interpretation, showing the upregulation of glycolysis-related transcripts at 6 dpt. In parallel, we observed increased expression of PDK4, a regulator that inhibits the pyruvate dehydrogenase complex and limits pyruvate entry into the TCA cycle.⁵⁰ Notably, transcripts associated with the TCA cycle remained largely unchanged, suggesting that increased glycolysis does not translate into enhanced mitochondrial respiration, but rather reflects a metabolic shift toward glycolytic reliance. However, this shift is unlikely to fully account for the observed increase in ATP levels.

In addition, RNA-seq data revealed a decrease in transcripts associated with lipid synthesis and an increase in those involved in lipid uptake and catabolism, indicating a shift toward lipid utilization as an alternative energy source. Although FAO is a mitochondrial and oxygen-dependent process, its upregulation under these conditions may seem paradoxical. Notably, angiogenesis has been reported to occur between 3 and 7 days post-injury, leading to the formation of newly perfused, albeit immature and permeable,^77,78 blood vessels that may partially restore oxygen availability and support oxygen-dependent pathways such as FAO.

Supporting this idea, pharmacological activation of FAO through agents such as L-carnitine or ketone bodies has been shown to improve mitochondrial function and promote tissue recovery,^79,80 suggesting that lipid metabolism may play an adaptive role under certain conditions. In our study, the increase in ATP levels, together with the upregulation of FAO-related transcripts (e.g., cpt1) and LD-associated markers (e.g., plin2, ppar-γ), is consistent with a shift toward lipid utilization that may contribute to maintaining cellular energy balance.

However, alternative explanations should also be considered. The increase in ATP levels may reflect not only enhanced intracellular production but also the accumulation of extracellular ATP associated with tissue damage and inflammatory signaling.⁴⁷ Importantly, the method used to quantify ATP does not distinguish between intracellular and extracellular pools, and therefore, both scenarios remain possible.

Taken together, these findings support a model in which mitochondrial dysfunction is accompanied by a metabolic shift toward glycolysis and lipid utilization. This metabolic reprogramming may represent a compensatory mechanism aimed at sustaining ATP production and maintaining cellular energy homeostasis despite impaired mitochondrial function. However, it is also possible that the observed increase in ATP levels reflects a pathological state associated with tissue damage, for example, through the accumulation of extracellular ATP. Further functional analyses, including measurements of oxygen consumption and glycolysis inhibition, will be necessary to directly validate these mechanisms and clarify their contribution to the observed metabolic phenotype.

Additionally, we describe an increase in LDs at 6 dpt. LDs are intracellular organelles responsible for storing mainly two types of lipids: triacylglycerols and cholesterol esters.⁸¹ LDs contribute to ATP generation by mobilizing fatty acids to mitochondria for β-oxidation, particularly under nutrient stress, thereby sustaining cellular energy production.⁸² Our RT-qPCR analysis revealed increased transcript levels of cpt1, the rate-limiting enzyme of β-oxidation, at 6 dpt. This finding supports the possibility that, following SCI, mitochondria undergo a substrate switch, increasing lipid transport to functional mitochondria and boosting ATP generation through β-oxidation to meet cellular energy demands. Moreover, metabolically distinct subpopulations of mitochondria have been described in various cell types.^83,84 Peridroplet mitochondria (PM), which are physically associated with LDs, and cytoplasmic mitochondria (CM). PMs have been reported to display higher respiratory capacity and ATPase, but reduced β-oxidation compared to CM.⁸⁴

There are also different pools of LD, which vary in composition, number, size, associated proteins, and function.⁸⁵ LDs express members of the Plin family of proteins on their surface, which regulate LD generation, transport, and turnover.⁸¹ Plin2 protects the LD from degradation by hindering the mobilization of fatty acids for oxidation.⁸¹ LD formation can also be induced by the transcription factor PPAR-γ.⁸⁶ Our RNA-seq analysis revealed increased transcript levels of ppar-γ and plin2 at 6 dpt, suggesting that the observed rise in LDs could result from enhanced generation and reduced degradation.

Additionally, LD accumulation and increased size have been reported in cells exposed to stress,⁸⁷ during aging,⁸⁸ and under prolonged starvation.⁸⁹ LDs have been described as mediators of cellular detoxification. In yeast, under stress and aging conditions, an increase in LD-mitochondria contacts has been observed, enabling the transfer of toxic lipids and proteins from mitochondria to LDs,^87,88 thereby extending cellular lifespan.⁸⁸ Under prolonged starvation, autophagic degradation of membranous organelles occurs, and the resulting lipids are packaged and stored in LDs.⁸⁹ Furthermore, after SCI, the injury produces a large amount of lipid-rich myelin debris. Macrophages phagocytose this material as part of the cleaning process, but the lipid overload promotes excessive LD accumulation, transforming them into foamy macrophages.^90,91 Future studies will be required to determine whether LD formation after SCI primarily supports energy metabolism or detoxification processes, and whether this metabolic switch is required for cell survival or exerts a more specific effect.

Therefore, our findings show that SCI in non-regenerative X. laevis is characterized by delayed mitochondrial remodeling and metabolic reprogramming. Unlike R-stages, NR-stage animals fail to mount a rapid and efficient mitochondrial response, instead exhibiting structural abnormalities, cristae disruption, and dysfunction. Nevertheless, cells compensate by upregulating glycolysis and lipid catabolism, sustaining ATP production despite impaired mitochondrial function. These results suggest that altered mitochondrial responses may contribute to regenerative failure and point to metabolic compensation as a key survival strategy under non-regenerative conditions.

Limitations of the study

A limitation of this study is that metabolic changes were primarily inferred from mitochondrial morphology, ATP measurements, transcriptional analyses, and histochemical approaches, rather than direct metabolic flux measurements. Therefore, although our results support increased glycolysis and lipid catabolism following SCI, additional studies using live metabolic sensors or flux analyses will be required to quantify pathway activity and substrate utilization in specific cell populations. Furthermore, most analyses were performed on whole spinal cord tissue, which limits the ability to distinguish cell type-specific metabolic responses. Finally, while our findings identify mitochondrial dysfunction and metabolic reprogramming associated with the non-regenerative response, functional experiments directly testing the contribution of these metabolic changes to regenerative failure or cell survival remain necessary.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Paula Slater (paula.slater@uss.cl).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •Data: The data generated during this study are available from the lead contact, Paula Slater (paula.slater@uss.cl), upon reasonable request.
• •Code: This study did not generate new code.
• •Other items: This study did not generate new unique reagents. Full uncropped western blot membranes have been provided in the supplemental material as Data S1/Methods S1.

Acknowledgments

We thank Dr. Juan Larraín for his mentorship, insights during experimental discussions, assistance with reagents and research equipment. We also thank Alejandro Munizaga for their valuable technical support. This work was supported by FONDECYT 11220624, Centro Ciencia & Vida FB210008 (ANID), and IBRO RS-3304417299 for PGS and FONDECYT 1231557 for VE.

Author contributions

P.G.S.: conceptualization, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, visualization, writing – original draft, writing – review and editing. M.E.D.: formal analysis, investigation, methodology, visualization, writing – original draft. M.V., C.C., S.Q., C.H., S.T., and J.P.: formal analysis, investigation, methodology, visualization. F.A.: formal analysis, investigation, methodology, visualization, writing – review and editing. V.E.: funding acquisition, resources, supervision, validation, writing – review and editing. All authors of the manuscript have read and agreed to its content and publication.

Declaration of interests

The authors declare no conflict of interest or other interests that might be perceived to influence the results and/or discussion reported in this paper.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the authors used ChatGPT exclusively as an editorial support tool to improve the grammar, spelling, syntax, clarity, and fluency of the English text. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-Tom20	Cell Signaling	Cat# D8T4N; RRID: AB_2687663
anti-Fis1	Proteintech	Cat# 10956-1; RRID: AB_2102532
anti-Mfn2	Abcam	Cat# 50838; RRID: AB_881507
anti-Opa1	BD Farmigen	Cat# 612607; RRID: AB_399889
anti-α-Tubulin	Abcam	Cat# ab7291; RRID: AB_2241126
anti-SOX2	Cell Signaling Technology	Cat# 2748S; RRID: AB_823640
anti-NeuN	SDIX	RRID: AB_2890069
Critical commercial assays
CellTiter-Glo	Promega	Cat# G9241; RRID: SCR_014267
RNeasy kit	QIAGEN	Cat# 74104; RRID: SCR_000628
Oligonucleotides
Primers for hk2, see Table S1	This paper	N/A
Primers for pfkfb1, see Table S1	This paper	N/A
Primers for pklr, see Table S1	This paper	N/A
Primers for cpt1, see Table S1	This paper	N/A
Primers for cpt2, see Table S1	This paper	N/A
Software and algorithms
ImageJ		RRID: SCR_003070
GraphPad Prism 8		RRID: SCR_002798
Metascape		RRID: SCR_016620

Experimental model and study participant details

Xenopus laevis husbandry and animal care

Xenopus laevis were obtained by natural mating of wild-type male and female frogs, and their husbandry was performed as previously described.^92,93 Animals were grown until Nieuwkoop and Faber stage 66 (NF 66) (NR-stages) for experiments.⁹⁴ Animals were maintained at 21°C under standard housing conditions in groups of approximately 15 animals during maintenance and in groups of 5 animals after spinal cord transection procedures. Animals were fed and maintained according to standard institutional protocols. Animals from both sexes were used; however, sex could not be determined because animals at this developmental stage are not yet sexually mature. Therefore, sex-based analyses could not be performed, representing a limitation for the generalization of sex-specific responses. Animals used in this study were experimentally naïve prior to spinal cord injury procedures. Animals were randomly allocated to experimental groups according to developmental stage and experimental condition. Littermates obtained from the same mating were considered technical replicates, whereas animals obtained from independent crossings were considered independent biological replicates. All procedures involving animals were conducted in accordance with institutional and national guidelines and regulations for animal care and use and were approved by the Scientific Ethics Committee for the Care of Animals and Environment of the Pontificia Universidad Católica de Chile and the Bioethical and Biosafety Committee of Universidad San Sebastián (Protocols 181017006 and 210504022).

Ethics approval

All animal procedures were approved by the Scientific Ethics Committee for the Care of Animals and Environment of the Pontificia Universidad Católica de Chile and Bioethical and Biosafety Committee of Universidad San Sebastián (Protocol 181017006 and 210504022).

Method details

Experimental design and statistical considerations

All experiments were performed using at least three independent biological replicates, defined as animals obtained from independent matings. Whenever possible, samples from different experimental groups were processed in parallel under identical experimental conditions to minimize batch effects. No animals or samples were excluded from the analyses unless tissue damage, incomplete spinal cord transection, poor tissue preservation, or technical failure prevented proper processing, imaging, or quantification. Quantifications and image analyses were performed using coded samples whenever possible to reduce observer bias; however, complete blinding was not always feasible during surgical procedures and tissue processing. Previously published datasets, methods, and transcriptomic resources used in this study are appropriately cited throughout the manuscript and reference section.

Spinal cord injury

Spinal cord transection was performed as previously described.^92,93 Briefly, NR-animals were anesthetized with 0,02% of tricaine mesylate (MS222) dissolved in 0,1x Barth solution (8.9 mM NaCl; 102 μM KCl; 238.1 μM NaHCO₃; 1 mM 4-(2- hydroxyethyl)-1-piperazine ethanesulfonic acid (HEPES); 81.14 μM MgSO₄; 33.88 μM Ca(NO₃)₂; 40.81 μM CaCl₂, pH 7.6). The skin and dorsal muscles were opened at the mid-thoracic level, followed by laminectomy of the dorsal portion of the sixth vertebra. Complete spinal cord transection was performed using microdissection scissors. Following surgery, animals were transferred to recovery tanks containing 0.1× Barth’s solution supplemented with penicillin and streptomycin. Uninjured animals were processed in parallel as controls.

Transmission electron microscopy

For transmission electron microscopy (TEM) analysis, spinal cord from uninjured and transected animals (6 hpt, 1, 2 and 6 dpt) were microdissected, including 1 mm rostral and 1 mm caudal to the injury site, and fixed overnight at 4°C in 2% paraformaldehyde and 2.5% glutaraldehyde. Samples were processed at the Advanced Microscopy Facility UMA-UC, Pontificia Universidad Católica de Chile. Ultrathin sections were imaged using a Philips Tecnai 12 transmission electron microscope operated at 80 kV. Three independent biological replicates were analyzed for each condition and time point. Mitochondrial morphology, localization, and area were quantified from independent animals.

Western blot

The caudal portion of spinal cords was isolated from uninjured and injured animals (3-4 per time curve point). Total protein was extracted using RIPA buffer (25 mM Tris HCl, 150 mM NaCl, 1% NP40, 1% sodium deoxycholate, 0.1% SDS) supplemented with protease inhibitors, 1X EDTA, and 0.001 M DTT on ice. A total of 40 ug of protein per sample was loaded onto SDS-PAGE electrophoresis gel, and proteins were subsequently transferred to PVDF membranes. Membranes were blocked in 5% non-fat milk in TBST (20 mM Tris-HCl; 140 mM NaCl; 1 mL Tween20) and incubated with primary antibodies overnight at 4°C. Following incubation with HRP-conjugated secondary antibodies, chemiluminescence was detected using the iBright imaging system and FEMTO substrate. Band intensity was normalized to α-Tubulin. Three independent biological replicates were analyzed. Primary antibodies: rabbit anti-Tom20 (1:2000, Cell signaling, D8T4N); rabbit anti-Fis1 (1:2000, Proteintech, 10956-1); rabbit anti-Mfn2 (1:1000, Abcam, 50838); mouse anti-Opa1 (1:1000, BD Farmigen, 612607); mouse anti-α-Tubulin (1:20000, Abcam, ab7291). Secondary antibodies: goat anti-mouse (1:5000, Pierce); goat anti-rabbit (1:5000, Abcam).

Immunofluorescence

Uninjured and injured animals were fixed in 4% paraformaldehyde for 2 h at room temperature or overnight at 4°C. The samples were dehydrated using a sucrose gradient, cryoprotected in optimal cutting temperature compound, and then frozen in liquid nitrogen. Longitudinal 10 μm sections were obtained using a Leica CM1850 cryostat. Samples were permeabilized in 1X phosphate-buffered saline (PBS) with 0.2 % Triton X-100 (PBST) for 10 min and blocked with 10% goat serum for 30 min before overnight incubation with primary antibodies at 4°C. Following incubation with fluorescent secondary antibodies, nuclei were stained with (1:10000) for 5 min and mounted with Vectashield (H-1000-NB). Images were acquired using an Olympus Fluoview FV10i confocal microscope under identical acquisition settings for comparative analyses within each experiment. At least three independent biological replicates were analyzed per condition. Primary antibodies: Mouse anti-SOX2 (1:200, Cell Signaling Technology, 2748S) was used to label NSPCs, and rabbit anti-NeuN (1:1000, SDIX) was used to detect mature neurons. Secondary antibodies: donkey anti-mouse AlexaFluor® 488 and donkey anti-rabbit AlexaFluor® 594 (1:500, Jackson Immuno Research), goat anti-rabbit AlexaFluor® 488 and goat anti-mouse AlexaFluor® 594 (1:500, Invitrogen).

RT-qPCR

The caudal portion of the spinal cords was isolated from 3–4 animals per condition at 0, 2, and 6 hpt, and 1, 2, and 6 dpt. Total RNA was extracted using the RNeasy Mini Kit according to the manufacturer’s instructions. RNA concentration and purity were assessed using a NanoDrop spectrophotometer (Thermo Scientific). cDNA synthesis was performed using M-MLV reverse transcriptase (Promega), and qPCR was performed using Power SYBR Green (Applied Biosystems) or Maxima SYBR Green (Thermo Scientific) master mix in technical triplicates for each biological replicate. Relative gene expression was calculated using the method previously described in reference,⁹⁵ using eef1a1 (GenBank: BC043843) as the reference gene. Primer sequences are listed in Table 1.

Measurements of ATP levels

ATP levels were measured using the CellTiter-Glo™ 2.0 Assay (Promega) according to the supplier’s protocol. The entire caudal segment of the spinal cord, from 3 to 4 animals, was isolated from the injury site with forceps and collected in 1X Marc’s Modified Ringer solution (MMR) (0.1 M NaCl, 2.0 mM KCl, 1.0 mM MgSO4, 2.0 mM CaCl2, 5.0 mM HEPES, pH 7.4). The tissue was disaggregated by incubation in 100 uL of 1X papain in 1X MMR – Ca²⁺, Mg²⁺ free (1X MMR – CMF) for 10 min, homogenizing with a pipette every 5 min. The papain reaction was stopped with the same volume (100 uL) of Fetal Bovine Serum. Later, the samples were centrifuged at 1,000g for 5 min, and the cells were resuspended in 100 uL of 1X MMR - CMF. Cell number was quantified using trypan blue exclusion in a Neubauer chamber, and 80,000 cells were plated per well for luminescence measurements. Luminescence readings were obtained in an Infinite M200 pro (Tecan). Four independent experiments were performed. Oligomycin treatment was used as a control for mitochondrial ATP production inhibition.

COX/SDH assay

Uninjured and injured animals at the indicated time points were anesthetized and sacrificed. The spinal cords were dissected and coated in Optimal Cutting Temperature (O.C.T.) compound and placed on cork supports for rapid freezing. The freezing procedure was adapted from previously published protocols.^96,97 Briefly, Liquid nitrogen was poured into a container suitable for handling extreme temperatures, and a stainless-steel cup with Isopentane (2-methylbutane) was submerged. The cork containing the sample was immersed in Isopentane for 30 sec until fully frozen, then rested on dry ice for 30 min to aid in the evaporation of Isopentane and stored at -80°C. Longitudinal 10 μm sections were obtained using a Leica CM1850 cryostat. Finally, for the COX and SDH histochemistry, an already published protocol was adjusted.⁴⁵ Briefly, the slides containing the spinal cord sections were thawed for 30 min at RT, incubated with 1X DAB, 8 M cytochrome C in 0.1 M PBS, 2mg/mL catalase, 220 mM sucrose for 2 hrs at RT, washed with distilled water, and incubated with 1.2 mM PMS, 1.6 mM NTB, 1 M sodium succinate, sodium azide/EDTA/PO4 (1.9 mM sodium azide, 1.5 mM EDTA 32.5 mM NaH2PO4·H2O, 217 mM Na2HPO4 ·7H2O) for 30 min at 37 °C, and washed with distilled water. Later, the samples were fixed with 10% formalin for 10 min and mounted with a pre-warmed glycerin gelatin solution. 1.25 mM sodium azide was used as a negative control for COX histochemistry, and succinate was replaced by malate as a negative control for SDH. Images were acquired using an inverted confocal microscope Nikon spectral ECLIPSE (C2Si), with a 40X objective, and identical microscope settings within each experiment. Three independent biological replicates were analyzed for each condition.

Oil-red staining

Spinal cord samples were frozen using isopentane and sectioned longitudinally at a thickness of 10 μm using a cryostat. The sections were fixed in 10% neutral buffered formalin for 8 minutes, followed by three washes with distilled water. After fixation, the sections were incubated in absolute propylene glycol for 2 minutes, followed by staining with 0.5 % Oil Red O solution in propylene glycol 100% for 16 hours at room temperature. Following staining, the sections were incubated in a polyethylene glycol solution for 1 minute and rinsed twice with distilled water. Sections were counterstained with Harris hematoxylin for 1 minute, followed by several washes with distilled water to remove excess stains. Finally, the stained sections were mounted using warm glycerin jelly mounting medium. Images were acquired using a Nikon spectral ECLIPSE (C2Si) confocal microscope using identical acquisition parameters within experiments. At least three independent biological replicates were analyzed per condition.

Bioinformatic analysis

Previously published RNA-sequencing data from Lee-Liu et al. (2014) were reanalyzed using the updated Xenopus laevis transcriptome (v10.1, Xenbase).^98,99 Gene Ontology (GO) enrichment analysis was performed using Metascape,¹⁰⁰ the top 1500 most up- or down-regulated genes were chosen from each time point, excluding their specific alleles (.L or .S) and using Homo sapiens as the input and output organism. Lipid metabolism-related genes were manually curated and analyzed for differential expression in regenerative tadpoles (R) and non-regenerative froglets (NR) from the same RNA sequencing dataset.

Quantification and statistical analysis

Image acquisition and quantitative analyses were performed using ImageJ software. Statistical analyses were conducted using GraphPad Prism 8 (GraphPad Software, USA). Biological replicates were defined as animals obtained from independent crossings, whereas littermates obtained from the same mating were considered technical replicates. Data are presented as mean ± SEM unless otherwise indicated.

Animals were randomly allocated to experimental groups according to developmental stage and experimental condition. Whenever possible, image acquisition and quantifications were performed using coded samples to minimize observer bias. No samples or animals were excluded from the analyses unless tissue damage, incomplete spinal cord transection, poor tissue preservation, or technical issues prevented proper processing, imaging, or quantification.

Data distribution was evaluated prior to selecting the appropriate statistical test. For comparisons involving normally distributed datasets with multiple groups, one-way ANOVA followed by Dunnett’s multiple comparison test was used. For non-parametric datasets involving multiple groups, Kruskal–Wallis tests followed by Dunn’s multiple comparison test were performed. Comparisons of log2-transformed measurements against a single reference value were analyzed using one-sample t tests. Statistical significance was defined as follows: ∗∗∗∗ (p < 0.0001), ∗∗∗ (p < 0.001), ∗∗ (p < 0.01), ∗ (p < 0.05), and ns (not significant). Sample sizes were selected based on previous studies using comparable experimental approaches in Xenopus laevis spinal cord injury models and on empirical experience with variability and reproducibility in the model.

Published: June 27, 2026