Lisinopril activates BI1 to reprogram lipid metabolism and restore autophagy in ALS

Abstract

Amyotrophic lateral sclerosis (ALS) involves disrupted lipid metabolism. Bax inhibitor 1 (BI1), an endoplasmic reticulum protein downregulated in ALS neuroprotective, represents a therapeutic target, but its metabolic regulatory mechanisms are incompletely understood. Using transcriptomics in skeletal muscle of ALS mice pre- and post-BI1 treatment, we identified BI1-regulated pathways. Structure-based virtual screening of FDA-approved compounds nominated lisinopril as a BI1 activator. Lisinopril upregulated BI1 protein expression, stabilizing mitochondrial membrane potential and protecting against SOD1^G93A-induced apoptosis in NSC34 cells. Concurrently, it regulated TGF-β1/mTOR-dependent autophagy, maintained NMJ integrity, and reshaped triglyceride/sphingolipid/glycerophospholipid metabolism to attenuate spinal cord pathology in ALS mice, promoting energy metabolism shift toward glucose oxidation. Additionally, lisinopril inhibited the TGF-β1/Smad2/3 pathway to alleviate muscle fibrosis, downregulate Acp5/FN expression, and reduce type I collagen deposition. In conclusion, this study provides evidence that pharmacological activation of BI1 by lisinopril suppresses TGF-β1, modulates lipid metabolism, and ameliorates ALS pathology, demonstrating promising therapeutic repurposing potential.

Lisinopril is identified as a BI1 activator that counteracts ALS pathology. It remodels lipid metabolism, restores autophagic flux via TGFβ1/mTOR signaling, and mitigates fibrosis. This multi-target mechanism supports the therapeutic repurposing of lisinopril for ALS.

Introduction

Amyotrophic lateral sclerosis (ALS) is a progressive and fatal neurodegenerative disease. Despite the identification of key pathways in the pathogenesis of the condition, including dysregulated immunity, oxidative stress, impaired autophagy, and disrupted iron homeostasis, only three FDA-approved therapies (riluzole, daravone, and sodium phenylbutyrate/taurursodiol combination therapy) are available¹. This situation highlights the urgent need for novel therapeutic strategies^2,3. Bax inhibitor 1 (BI1) has emerged as a key neuroprotective factor, with its roles in the pathogenesis of Parkinson’s disease⁴ and amyloid processing⁵ already established. Furthermore, our previous work further established that BI1 overexpression extends the lifespan of ALS mice. Although BI1 has emerged as a master regulator of neuroprotection, its underlying mechanisms are not fully defined. Nevertheless, compelling evidence positions BI1 as a promising next-generation therapeutic target. Despite this considerable promise, the drug development landscape for this pathway remains nascent, particularly regarding the discovery of specific small-molecule activators, which constitutes a pivotal bottleneck for clinical translation.

Virtual screening (VS) is a pivotal technology that leverages computational simulations and molecular docking to systematically identify high-affinity ligands for a target protein from large-scale compound databases. Its efficiency makes it exceptionally well-suited for drug repurposing. In order to address the research gap concerning BI1 activators, this study employed VS to analyse FDA-approved compounds based on the BI1 structure, identifying the angiotensin-converting enzyme inhibitor lisinopril as a novel BI1 activator⁶. Beyond its well-known cardiovascular applications, lisinopril has been shown to have pleiotropic benefits, including extending lifespan and having anti-fibrotic effects in rodent models^7–9. These benefits are proposed to stem from a dual mechanism: mitigating Ang II-induced oxidative stress and modulating TGF-β1-driven inflammatory pathways^10–15. Given the centrality of these pathways in ALS pathogenesis, we hypothesize that lisinopril may exert neuroprotective effects by activating BI1, making it an ideal candidate for drug repurposing in the field of ALS.

The increasing application of multi-omics technologies has offered a powerful new lens through which to elucidate the complex pathogenesis of ALS. Through integrative profiling of transcriptomic, proteomic, and metabolomic data, studies have identified SPTLC1 as a novel ALS-linked gene and pinpointed dysregulated arachidonic acid metabolism as a key contributor to motor neuron dysfunction^16,17. These findings establish lipid metabolic dysregulation as a fundamental component of ALS pathogenesis. Notably, both central nervous system (CNS) and systemic lipid disturbances correlate strongly with ALS progression, where lower serum triglyceride (TG) levels predict increased mortality risk while high-fat diets show neuroprotective effects in ALS patients. Furthermore, an elevated LDL/HDL ratio emerges as a favorable prognostic marker for prolonged survival^18–22. Collectively, these findings indicate that lipid metabolism functions as both a pivotal pathogenic mediator and a promising therapeutic target in ALS. However, no studies have yet reported whether BI1 and lisinopril exert their protective effects in ALS by regulating lipid homeostasis.

Based on the extant research background, this study employs a multi-omics integration strategy to achieve the following three objectives: (i) characterize BI1’s neuroprotective mechanisms in ALS, (ii) validate lisinopril as a BI1 activator through structural and functional assays, and (iii) elucidate its therapeutic effects on autophagy and lipid homeostasis. Collectively, our findings provide both mechanistic insights into ALS pathophysiology and a rational basis for developing BI1-targeted neuroprotective strategies.

Results

RNA-seq identified BI1 as a regulator of glycolipid metabolism and fibrosis-related gene expression

To investigate the role of BI1 in ALS pathogenesis, we performed RNA-seq on skeletal muscle from WT, ALS CON and ALS BI1 mice (180 days). Principal component analysis (PCA) revealed distinct transcriptomic clustering (Supplementary Fig. 1A, B). Volcano plot analysis identified 349 DEGs in ALS BI1 vs. ALS CON mice (154 upregulated, 195 downregulated) (Fig. 1A; Supplementary Table 1). Then, the PPI analysis of DEGs identified three hub genes: Myh7, Myl2, and Tnnt2, suggesting BI1 promoted fast-to-slow myofiber transition that may shift metabolism from glycolytic to oxidative pathways (Fig. 1B–E; Supplementary Table 2).

Fig. 1. Screening and enrichment analysis of differential genes.

A ALS CON vs. ALS BI1 group differential gene volcano map Skeletal muscles of mice in ALS CON and ALS BI1 groups were screened for differential genes and volcano plots were produced with the screening criteria of |log₂(FoldChange)| ≥ 1 & padj ≤ 0.05. B–E Screening results for core genes CYTONCA (B), MCODE (C), and CYTOHUBBA (D) were screened to obtain the core genes, and then the obtained core genes were taken to intersect thus further narrowing down the number of core genes to obtain the three core genes, Myh7, Myl2, and Tnnt2 (E). F GO enrichment analysis of the ALS CON and ALS BI1 differential genes Show the top 10 enriched terms ranked by padj for the BP, CC, and MF categories. X-axis: GO term; Y-axis: -log10(padj) (A larger value indicates a more significant level of enrichment.) G KEGG enrichment analysis of differential genes in ALS CON and ALS BI1 groups Display the top 20 enriched KEGG pathways ranked by padj. X-axis: Gene Ratio, Y-axis: KEGG pathways. Bubble size is proportional to the gene count, and bubble color represents the padj, where a shift toward red indicates greater statistical significance. H Heat map of correlation coefficients between modules and traits The color of each cell represents the correlation coefficient (red for positive correlation, blue for negative correlation), with the intensity of the color indicating the strength of the correlation.

We conducted comprehensive pathway analyses, including KEGG and GO enrichment of DEGs and GSEA across experimental groups, to systematically characterize molecular pathways modulated by BI1 in ALS. KEGG and GO analyses demonstrated significant regulation of TGF-β pathway, serine/threonine kinase receptor activity and G protein β/γ-subunit complex binding (Fig. 1F–G). These findings suggested BI1 independently modulates TGF-β signaling and transmembrane receptor function, with potential crosstalk to the PI3K/AKT pathway. In addition, BI1 significantly modulated both arginine/proline metabolic pathways and NAD^+ binding activity, indicating effects on muscle bioenergetics. Concurrently, extracellular matrix (ECM) organization and glial cell myelin maintenance were significantly enriched, demonstrating BI1’s dual role in attenuating muscle fibrosis and preventing myelin degeneration (Fig. 1F–G). Finally, GSEA further confirmed these findings, showing BI1 treatment reversed the elevated pathway activities observed in ALS CON mice (TGF-β: NES = 1.71→−1.19; fatty acid metabolism: NES = 2.11→−0.73; mTORC1: NES = 1.71→−1.35) (Supplementary Fig. 2A–E), highlighting BI1’s ability to coordinately regulate these pathogenic pathways.

Collectively, these findings established that BI1 is closely related to lipid homeostasis in ALS. Afterwards, through integrative analysis of DEGs in ALS BI1 vs. ALS CON mice and known lipid metabolism gene collection (Supplementary Table 3; Supplementary Fig. 1F), we identified seven core regulatory genes: Stab2 (192188), Angptl4 (57875), Pnpla2 (66853), Lpin2 (64898), Agpat2 (67512), Pde3b (18576) and Tgf-β1 (21803) (Supplementary Table 4). To systematically evaluate the effects of BI1 on lipid-related gene expression before and after clinical onset, we performed qPCR analysis of these candidate genes. It was found that at pre-symptomatic stage (150 days), BI1 downregulated most genes except Pnpla2, while post-onset (180 days) all targets were downregulated (Supplementary Fig. 4), consistent with RNA-seq data. Consequently, focusing on Tgf-β1 as a key regulatory node, WGCNA identified its downstream effectors Pparg and Acp5 as significantly downregulated in 180-day-old ALS BI1 mice (Fig. 1H; Supplementary Fig. 1D and Supplementary Fig. 3). These results demonstrated that BI1 simultaneously modulated lipid metabolic pathways and inhibited skeletal muscle fibrosis to attenuate ALS progression.

BI1 mediated regulation of autophagy, lipid metabolism and fibrosis through TGF-β pathway

Integrative analysis of KEGG, GSEA, and WGCNA data demonstrated that BI1 modulated lipid metabolism, PI3K/AKT/mTOR signaling, and muscle fibrosis through regulation of the TGF-β pathway. Consequently, we performed WB analysis of key pathway components together with qPCR quantification of lipid metabolism and fibrotic gene expression. WB results revealed significantly increased BI1 expression associated with reduced Bax levels in ALS BI1 mice at both pre-symptomatic (150 days) and symptomatic (180 days) stages. These results demonstrated that injection of the BI1 plasmid effectively increases BI1 protein levels while suppressing Bax expression in ALS mice, resulting in significant inhibition of apoptotic activation (Fig. 2A; Supplementary Fig. 5B). The WB and qPCR results also demonstrated that BI1 coordinates a dual therapeutic mechanism through TGF-β1 downregulation. By attenuating Smad2/3 phosphorylation (Fig. 2B, Supplementary Fig. 5A), BI1 reduced the transcriptional activity of Pparg, Acp5 and FN, thereby reducing fatty acid oxidation and suppressing muscle fibrosis (Supplementary Fig. 5D–F). At the same time, through inhibition of the PI3K/AKT/mTOR axis, BI1 activated autophagic flux via the p-70S6K/ULK1/Beclin-1/LC3B/p62 pathway (Fig. 2A, B, Supplementary Fig. 5A–C). In the summary, these findings demonstrated the modulation of both metabolic and proteostatic systems underlay the neuroprotective effects of BI1 in ALS.

Fig. 2. Effect of BI1 on the regulation of the TGF-β pathway and ATP production.

A, B BI1 inhibited activation of TGF-β1/PI3K/AKT/mTOR signaling pathway and TGF-β1/Smad2/3 pathway The expression of BI1 protein as well as changes in the expression of TGF-β1, Smad2/3, p-Smad2/3, PI3K, p-AKT and autophagy-related proteins in skeletal muscle of ALS mice before and after injection of the BI1 plasmid were detected by WB, and GAPDH was selected as an internal reference protein. C Statistical graph of ATP content Skeletal muscle ATP content of ALS CON mice and ALS BI1 mice on 150 and 180 days was measured using the ATP content assay kit. D Statistical graph of HK content Skeletal muscle HK activity in ALS CON and ALS BI1 mice was assayed on 150 and 180 days using the HK assay kit. E–G Circulating lipid content in skeletal muscle of ALS CON and ALS BI1 mice Skeletal muscles of ALS CON and ALS BI1 mice were taken on 150 and 180 days, respectively, and protein concentrations were measured, after which the levels of TG (E), T-CHO (F), and LDL/HDL (G) were measured in the skeletal muscles of the mice using the corresponding kits. n = 3 independent experiments, data were expressed as mean ± SD and analyzed using one-way ANOVA and Bonferroni multiple comparison test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant. (*: compared with 150 days ALS CON; #: compared with 180 days ALS CON).

BI1 enhanced mitochondrial ATP production and improves ALS prognosis through metabolic reprogramming

Accumulating evidence indicated that hypermetabolism was a hallmark feature of ALS. RNA-seq revealed that BI1 overexpression drove a glycolytic-to-oxidative metabolic transition in ALS mouse skeletal muscle, significantly boosting ATP production. Experimental assays showed that HK activity and ATP content were markedly upregulated in skeletal muscle of ALS BI1 mice ATP content (Fig. 2C, D). In contrast, Pparg mRNA expression was downregulated (Supplementary Fig. 5D). These finding indicated that BI1 improves bioenergetic capacity in ALS mice by preferentially stimulating glucose oxidation rather than lipid oxidation.

Clinical data showed circulating lipids strongly correlate with ALS progression. Elevated TG levels predicted favorable outcomes, while cholesterol metabolism shifted exhibit biphasic correlations: higher T-CHO and lower LDL/HDL ratios in pre-symptomatic stages correlate with longer survival, whereas reduced T-CHO and higher LDL/HDL ratios during progression associate with better prognosis²³. Notably, BI1 exhibited disease stage-specific lipid metabolism modulation: it elevated both TG and T-CHO while reducing the LDL/HDL ratio during the pre-symptomatic phase to establish essential energy reserves and provide lipid substrates for membrane maintenance and synaptic plasticity, but conversely decreased T-CHO while increasing TG and the LDL/HDL ratio during the progressive phase to mitigate bioenergetic stress (Fig. 2E–G).

In the summary, BI1’s regulatory role in ATP content and lipid metabolism positioned it as a crucial metabolic regulator, offering mechanistic insights for developing phase-specific ALS therapies.

Structure-based virtual screening revealed lisinopril as a novel small molecule activator of BI1

AlphaFold2-predicted BI1 structure showed excellent quality, with MolProbity score (1.01) and Ramachandran plot statistics (96.17% favored regions by MolProbity, 93.5% by PROCHECK) surpassing standard validation thresholds (Supplementary Fig. 6A, C, D). The integrated analysis of structural pockets was performed, incorporating surface area metrics, scoring (Table 1) and Discovery Studio 2019 prediction results (Supplementary Fig. 6B). This analysis identified Pocket 2 as the optimal site for molecular docking studies (Supplementary Fig. 7A). Following two rounds of molecular docking screening (Supplementary Data 1; Supplementary Table 5), lisinopril was selected based on safety, cost, and literature support from screening candidates with -CDOCKER energy >55 kcal/mol (Supplementary Table 6, Supplementary Fig. 7B).

Table 1.

Prediction results of the binding pocket of BI1 protein

Pocket Number	Volume (ų)	Surface (Ų)	Drug Score	Simple Score
1	949.06	1170.25	0.82	0.78
2	826.43	1136.88	0.86	0.6
3	273.09	574.1	0.57	0.14

Docking revealed stable BI1-lisinopril binding, with hydrogen bonds (SER157/158, LEU169, LYS220), hydrophobic contacts (LEU151/154/155/174, MET223), and electrostatic interaction (ARG219) (Supplementary Fig. 7C, D). These results supported BI1’s selection for subsequent experimental validation.

Lisinopril protected against SOD1^G93A-induced neurotoxicity in NSC34 cells by preserving mitochondrial integrity and axonal structure

Lisinopril exhibited potent neuroprotective activity in SOD1^G93A-transfected NSC34 cells through multiple complementary mechanisms. CCK-8 assays identified 15 μM as the optimal neuroprotective concentration (Fig. 3A), which significantly reduced SOD1^G93A-induced apoptosis as evidenced by decreased DNA fragmentation in TUNEL assays (Fig. 3B, D). At the subcellular level, the JC-1, MPTP and transmission electron microscope assays showed that lisinopril treatment preserved mitochondrial function by restoring membrane potential (Fig. 3C, E; Supplementary Fig. 8A, B) and preventing ultrastructural damage (Supplementary Fig. 8C). Concurrently, lisinopril exerted neuroprotective effects through cytoskeletal stabilization in SOD1^G93A-transfected NSC34 cells. Immunofluorescence analysis revealed that lisinopril treatment effectively preserved physiological MAP2 expression patterns (Fig. 3F, G) while restored proper β3-Tubulin localization and promoting robust axonal elongation (Fig. 3F, H). In summary, these findings demonstrate that lisinopril exerts neuroprotective effects against SOD1^G93A-induced apoptosis through preserving the cytoskeleton and maintaining mitochondrial homeostasis.

Fig. 3. Effect of lisinopril on SOD1^G93A-induced apoptosis in NSC34 cells.

A Determination of Optimal Lisinopril Concentration via CCK-8 Screening NSC34 cells were treated with different concentrations of lisinopril for 1 h. After transfection of LACZ and SOD1^G93A plasmids according to different groups, CCK-8 kit was used to detect the absorbance value at wavelength 450 nm using an enzyme marker, and finally, the drug concentration corresponding to the group with the highest cell viability was selected as the optimal action concentration of lisinopril. B, D TUNEL assay evaluation of lisinopril’s anti-apoptotic effects in SOD1^G93A-expressing NSC34 cells NSC34 cells were treated with 15 μM lisinopril (or PBS) for 1 h before transfection with LACZ or SOD1^G93A plasmids. 24 h after transfection, cells were fixed and stained with DAPI (blue) to label nuclei, and TUNEL-positive cells were detected by FITC fluorescence (green) using confocal microscopy (B). The percentage of TUNEL-positive cells was quantified and analyzed statistically (D). C, E JC-1 staining demonstrated the effect of lisinopril on mitochondrial membrane potential Cells were pretreated identically to the TUNEL assay group. After 24 h, mitochondrial membrane potential was assessed using the JC-1 fluorescent probe. Confocal microscopy images showed nuclei stained with DAPI (blue) and JC-1 fluorescence (green: FITC for monomeric form at depolarized ΔΨm; red: R-PE for J-aggregates at polarized ΔΨm). Disruption of mitochondrial membrane potential was indicated by decreased red/green fluorescence intensity ratio (C). Quantitative analysis of the red-to-green ratio was presented (E). F–H Immunofluorescence results for MAP2 and β3-Tubulin Cells received identical pretreatment as in the TUNEL assay. After 24 h, neuronal morphology was assessed by immunofluorescence staining for MAP2 (red, R-PE) and β3-Tubulin (green, FITC), with nuclear counterstaining (DAPI, blue) (F). Finally, we counted the lengths of the axons (H) as well as the MAP2 fluorescence intensity (G) and make statistical graphs. (n = 3 independent experiments, data are expressed as mean ± SD and compared using one-way ANOVA with Bonferroni’s multiple comparison test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant. (*: compared to SOD1^G93A group)).

Lisinopril activated autophagic flux in NSC34 motor neurons through inhibition of TGF-β/PI3K/AKT pathway

Our study demonstrated that BI1 can enhance autophagic flux in the ALS mice model by modulating the TGF-β1/PI3K/AKT/mTOR pathway. Therefore, in order to determine whether lisinopril, as a BI1 activator, inhibited apoptosis through TGF-β1/PI3K/AKT/mTOR-dependent autophagy activation, we analyzed protein expression in this pathway by WB. The results demonstrated that lisinopril treatment significantly elevated BI1 expression, augmented the expression of autophagy initiation markers (Beclin-1, LC3B-II, and p-ULK1), and decreased p62 to enhance autophagic flux. Concomitantly, reduced levels of TGF-β1 signaling components (PI3K, p-AKT, and p-mTOR) and downstream effectors (p-70S6K) were observed in response to lisinopril treatment (Fig. 4A–E). What’s more, experiments employing BI1 overexpression and knockout demonstrated significant regulation of the TGF-β1/PI3K/AKT/mTOR pathway, further confirming that lisinopril’s modulation of TGF-β pathway depended on BI1 (Fig. 4F–O). To sum up, our study demonstrated that lisinopril-mediated BI1 activation suppressed TGF-β1/PI3K/AKT/mTOR signaling to enhanced autophagic flux.

Fig. 4. Lisinopril inhibited TGF-β signaling pathway and thereby activated autophagy in NSC34 cells.

A–E Effect of lisinopril on the expression of TGF-β1/PI3K/AKT/mTOR pathway and autophagy-related protein NSC34 cells were treated with 15 μM lisinopril (or PBS vehicle control) for 1 h prior to transfection with 4 μg of LACZ or SOD1^G93A plasmids. Whole-cell lysates were collected 24 h post-transfection and subjected to WB. GAPDH was selected as the internal reference protein to detect the changes of BI1, TGF-β1, PI3K, p-AKT, p -mTOR, p70S6K, p-ULK, and autophagy-related proteins Beclin1, LC3B, and p62. F–J Effects of BI1 overexpression on the TGF-β1/PI3K/AKT/mTOR pathway and autophagy NSC34-BI1-14# cells were transfected with 4 μg of LACZ or SOD1^G93A plasmids. Whole-cell lysates were collected 24 h post-transfection for WB analysis. GAPDH was selected as the internal reference protein to detect the changes of BI1, TGF-β1, PI3K, p-AKT, p -mTOR, p70S6K, p-ULK, and autophagy-related proteins Beclin1, LC3B, and p62. K–O The effects of knocking down BI1 on the TGF-β1/PI3K/AKT/mTOR pathway and autophagy. NSC34 cells with BI1 knockdown (via siRNA) were treated with 15 μM lisinopril or PBS for 1 h, followed by transfection with 4 μg of LACZ or SOD1^G93A plasmids. Total cell lysates were collected 24 h post-transfection for WB analysis. GAPDH was selected as the internal reference protein to detect the changes of BI1, TGF-β1, PI3K, p-AKT, p -mTOR, p70S6K, p-ULK, and autophagy-related proteins Beclin1, LC3B, and p62. The gray values of BI1 and TGF-β signaling pathway proteins and autophagy-related proteins were compared with those of the internal reference protein GAPDH and normalized, and statistical plots were produced based on the normalized values. For the quantification of LC3, it was done by LC3B-II/I ratio quantification. n = 3 independent experiments, data were expressed as mean ± SD and analyzed using one-way ANOVA with Bonferroni’s multiple comparison test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant. (*: compared with SOD1^G93A group; #: compared with siRNA + SOD1^G93A group).

Lisinopril modulated the onset time, lifespan, and exercise capacity in ALS mice

ALS patients typically experienced a progressive loss of motor function, with respiratory failure representing the predominant cause of mortality. In this study, compared to the WT mice, ALS mice showed significant impairment in motor ability, while lisinopril treatment significantly attenuated disease progression in ALS mice. Specifically, in comparison to untreated controls, the treatment delayed disease onset and extended lifespan in both sexes (Fig. 5A–D), while preserving multiple motor functions including running endurance (Fig. 5E, F), limb pulling force (Fig. 5G, H) and stride length (Fig. 5I–L) in ALS mice. Collectively, the consistent improvements observed across multiple disease parameters collectively suggested that lisinopril modulated fundamental pathological processes in ALS progression.

Fig. 5. Effect of lisinopril on onset timing, survival and locomotor activity in ALS mice.

A, B Statistical analysis of lisinopril effects on ALS onset C, D Survival analysis in ALS mice treated with lisinopril E, F Running time performance in WT and ALS mice before and after lisinopril treatment G, H Limb grip strength analysis in WT and ALS mice treated with lisinopril I–L Lisinopril prevented step spacing decline in WT and ALS mice Counting the step spacing of female (I) and male (J) mice in the WT group, ALS CON group and ALS lisinopril group and making statistical graphs (K, L). (Mice in each group n = 10, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant. (*: compared to 150 days ALS CON group; #: compared to 180 days ALS CON group)).

Lisinopril ameliorated pathological alterations in the skeletal muscle of SOD1^G93A transgenic ALS mice

The results of the RNA-seq study demonstrated that BI1 induced a transition in muscle fiber type from fast to slow. To investigate whether lisinopril induces a similar fast-to-slow myofiber transition, we performed immunofluorescent co-staining of fast (Myh1) and slow (Myh7) myosin heavy chain isoforms in skeletal muscle sections. Immunofluorescence analysis revealed that, compared to WT mice, ALS mice exhibited a lower proportion of oxidized muscle fibers. Treatment with lisinopril significantly preserved the total muscle fiber count and concomitantly increased the proportion of oxidized slow-twitch fibers (Fig. 6A, G). This fiber type shift towards oxidative metabolism was particularly advantageous in ALS. Thus, we firstly revealed that lisinopril treatment resulted in an increase in mitochondrial number and the preservation of cristae integrity within mitochondria in ALS mice by TEM (Supplementary Fig. 9). Afterwards, we measured the skeletal muscle ATP content and found that lisinopril treatment significantly increased ATP content in ALS mice compared to untreated controls (Supplementary Fig. 10A). Metabolic analysis revealed that lisinopril enhanced ATP production by favoring glucose oxidation, as indicated by increased HK activity, suppressed Pparg signaling, and lipid droplet accumulation in myofibers (Supplementary Fig. 10B, C; Fig. 6B, H). In terms of fibrosis regulation, ALS mice displayed a markedly increased proportion of type I collagen fibers in skeletal muscle compared to WT mice, indicative of progressive fibrosis. Treatment with lisinopril counteracted this pathology by inhibiting the conversion of type III to type I collagen fibers (Fig. 6C, I). Furthermore, it downregulated the expression of key fibrosis markers, including Acp5 and FN (Supplementary Fig. 10D, E), thereby demonstrating a comprehensive antifibrotic effect. Meanwhile, lisinopril treatment significantly preserved muscle fiber size. The peak fiber CSA in treated mice was maintained at approximately 2000 μm², a value comparable to that in WT mice (~2500 μm²) and substantially greater than the reduced range (500–1500 μm²) observed in untreated ALS mice (Fig. 6D, J). Collectively, these coordinated improvements in muscle metabolism, structure and function underscore lisinopril’s ability to against neurogenic muscle wasting in ALS mice through multi-target mechanisms.

Fig. 6. Effect of lisinopril on pathological changes in skeletal muscle and spinal cord of ALS mice.

A, G Myh1 and Myh7 immunofluorescence double label staining results Immunofluorescence of paraffin-embedded muscle sections (thickness of 10 μm) stained for fast (Myh1, red) and slow (Myh7, green) myosin heavy chains, with DAPI nuclear counterstain (blue) (A). Scale bar: 100 μm (200× magnification). Finally, the proportion of slow muscle fibers was counted and plotted in the statistical graph (Myh7⁺/total fibers) (G). B, H Lipid droplet accumulation in skeletal muscle tissue Skeletal muscle cryosections (thickness of 10 μm) were stained using oil red O, showing neutral lipid deposits (red) (B). Scale bar: 50 μm (400× magnification). After that, the Oil Red O-positive area was then quantitatively analyzed and statistical plots were produced (H). C, I Sirius red staining of skeletal muscle in WT, ALS CON and ALS lisinopril mice Longitudinal paraffin sections (thickness of 6–7 μm) of skeletal muscle from mice in the WT, ALS CON and lisinopril-treated ALS mice were stained with Sirius red and photographed using a polarized light microscope, showing type I collagen (red/yellow) and type III collagen (green) (C). Scale bar: 100 μm (200× magnification). Finally, the proportion of type I collagen fibers accounted for was counted (I). D, J HE staining and cross-sectional area statistics of skeletal muscle in WT, ALS CON and ALS lisinopril mice H&E-stained transverse skeletal muscle sections (thickness of 10 μm) from 150-day and 180-day WT, ALS CON and lisinopril-treated ALS mice. Scale bar: 100 μm (400× magnification) (D). Subsequently, Image J was used to measure the cross-sectional area of the myofibers, with n = 6, and 20-30 myofibers were selected for counting in each slice. This ensured that each set of samples totaled 120 to 180 myofibers. Thereafter, the measured cross-sectional area data were compiled into a myofiber CSA statistical map (J). E, K Lisinopril increased Nissl bodies density in ALS mice spinal cord Transverse paraffin sections of the spinal cord (thickness of 10 μm) were stained with Nissl staining solution. Images were acquired using a Nikon Eclipse microscope (E). Scale bar: 100 μm (200× magnification). Finally, the number of Nissl bodies was counted using Image J software, and the data were then normalized to draw a statistical graph (K). F Lisinopril preserved myelin integrity in ALS mice spinal cord Transverse paraffin-embedded spinal cord sections (thickness of 10 μm) were stained with LFB showing white matter tracts. Images were acquired using a Nikon Eclipse microscope. Scale bar: 50 μm (400× magnification). For all analyses except HE staining, a sample size of n = 3 was used per group. Data were expressed as mean ± SD and analyzed using one-way ANOVA and Bonferroni multiple comparison test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant. (*: compared with 150 days ALS CON; #: compared with 180 days ALS CON).

Lisinopril attenuated pathological alterations in the spinal cord and preserved NMJ integrity in ALS mice

Nissl bodies coordinate neurotransmitter synthesis and axonal regeneration, with broader implications for synaptic plasticity and neurodegenerative disease progression. Notably, the number of Nissl bodies in the spinal cord of ALS mice was significantly lower than that in WT mice. Lisinopril treatment significantly restored Nissl bodies density in spinal motor neurons during the progression of the disease, suggesting its neuroprotective potential in ALS mice (Fig. 6E, K). Concurrently, compared to ALS mice, lisinopril reduced myelin cavitation and shedding in spinal cord white matter, restoring it to levels similar to those in WT mice (Fig. 6F), consistent with BI1’s role in myelin maintenance from RNA-seq data. Furthermore, immunofluorescence analysis of gastrocnemius muscle NMJ revealed that lisinopril treatment significantly preserved presynaptic terminal integrity and reduced axonal denervation (Supplementary Fig. 11). In summary, these effects on neuronal organelles, myelin and NMJ collectively contributed to lisinopril’s ability to exert neuroprotective functions in ALS progression.

Lisinopril inhibited activation of TGF-β signaling pathway in ALS mice

As the BI1 activator, lisinopril has been shown to induce autophagy in NSC34 cells through TGF-β1/mTOR signaling pathway (Fig. 4). Therefore, to elucidate the specific mechanisms in lisinopril-mediated regulation of lipid metabolism, muscle fibrosis and autophagy in ALS mice, we performed WB and IHC analysis to quantify expression changes of key TGF-β pathway components. WB and IHC analysis revealed that lisinopril treatment significantly increased BI1 protein expression while reducing both TGF-β1 levels and Smad2/3 phosphorylation in skeletal muscle of ALS mice (Fig. 7A–D, F–H, K). In addition, lisinopril treatment regulated the autophagy network in ALS mice, shown by concurrent suppression of PI3K/AKT/mTOR signaling (reduced p-AKT, p-mTOR, p-70S6K) (Fig. 7A–D, I, K) and activation of autophagic flux (increased p-ULK, Beclin1, LC3B-II, and decreased p62) (Fig. 7B, D, E, J, L–N). In summary, lisinopril enable to activate BI1 expression in vivo and address multiple facets of ALS pathology—from metabolic dysregulation and fibrotic progression to autophagic flux—through divergent TGF-β-mediated pathways.

Fig. 7. Regulation of the TGF-β signaling pathway by lisinopril.

A–E WB analysis showed BI1 activation and TGF-β pathway regulation by lisinopril in ALS mice We performed WB analysis to evaluate the expression of BI1 and autophagy-related proteins, along with key components of the TGF-β1/PI3K/mTOR and TGF-β1/Smad2/3 signaling pathways, in skeletal muscle of ALS mice before and after lisinopril administration (A, B). GAPDH was selected as the internal reference protein. The bands were then quantified by gray value using Image J software (C, D). For the quantification of LC3B, the LC3II gray value was compared with the LC3Ⅰ gray value and then normalized (E). n = 3 independent experiments. F–N IHC analysis of BI1 and autophagy-related protein expression after lisinopril treatment We performed IHC staining on paraffin-embedded skeletal muscle sections (thickness of 10 μm) from ALS mice (pre- and post-lisinopril treatment) to evaluate the expression of BI1 (F), TGF-β1/PI3K pathway components (G, I), and autophagy markers (J, L, M). Scale bar: 100 μm (200× magnification). Finally, IHC images were analyzed using Image J software, and the data were then normalized to draw a statistical graph (H, K, N). Ections in each group n = 3. Data were expressed as mean ± SD and analyzed using one/two-way ANOVA and Bonferroni multiple comparison test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant. (*: compared with 150 days ALS CON; #: compared with 180 days ALS CON).

Lisinopril altered the lipid metabolism profiling of skeletal muscle tissue in ALS mice

The targeted lipidomic analysis demonstrated the significant impact of lisinopril on lipid metabolism in ALS mice, with rigorous quality control (QC) (Supplementary Fig. 12A) and clear metabolic separation between lisinopril and control groups by PLS-DA, confirming robust lipid network remodeling (Fig. 8A; Supplementary Fig. 12B). Subsequently, we identified 336 significantly altered lipid metabolites, comprising 116 upregulated and 220 downregulated species (Fig. 8B, D; Table 2; Supplementary Table 7). We then conducted comprehensive correlation analysis and hierarchical clustering of the 336 significantly altered lipid metabolites to elucidate their interrelationships. The results revealed an inverse correlation between TG and sphingomyelin (SM). Lisinopril-treated groups exhibited elevated TG including TG (4:0/14:0/18:2) and TG (20:5/4:0/18:1). In contrast, phosphatidylcholine (PC) and phosphatidylethanolamine (PE) showed positive correlation. Additionally, control groups exhibited elevated levels of sphingomyelins [SM(d37:7), SM(d39:4), SM(d18:1/21:4)], ceramides [Cer(d18:1/21:3), Cer(d18:2/21:2), Cer(d18:2/19:0)], and PC species [PC(19:0/20:4), PC(20:1/20:4), PC(19:0COOH/22:5), PC(18:2/22:6)] compared to lisinopril-treated subjects (Fig. 8E; Supplementary Fig. 13). In summary, these results suggested lisinopril reprogramed lipid homeostasis by shifting the balance from membrane lipid metabolism toward storage lipid utilization.

Fig. 8. Lisinopril remodeled lipid metabolism in ALS mice skeletal muscle.

A Scatterplot of PLS-DA scores in the ALS CON and ALS lisinopril groups In the constructed model, the projection of the samples on the first principal component was set as the horizontal coordinate, and the projection of the samples on the second principal component was used as the vertical coordinate. PLS-DA model validity was confirmed by strong explained variance (R²Y = 0.95) and predictive capability (Q²Y = 0.50). In addition, R2Y was larger than Q2Y, which indicated that the model was not overfitting and had a high predictive reliability. B, D Volcano plot displayed lipid species altered by lisinopril treatment in ALS mice skeletal muscle Negative ion mode (B) and positive ion mode volcano plots (D) comparing skeletal muscle lipidomes between ALS lisinopril and ALS CON mice. Data pointed represent individual lipid species positioned by log₂ FC(x-axis) versus -log₁₀ p-value (y-axis), with point size corresponding to VIP scores. Significantly altered lipids are colored (red: up-regulated; blue: down-regulated). C KEGG enrichment analysis of differential lipid metabolites The horizontal coordinate in the graph was x/y (number of differential lipid compounds in the corresponding metabolic pathway/number of total lipid compounds identified in that pathway). A higher ratio indicated greater enrichment of differential lipid compounds in the respective pathway. E Correlation analysis of differential lipid compounds Heatmap displaying Spearman correlation coefficients (ρ) among the 20 most significantly altered lipid species (ranked by adjusted p-value) in ALS mice skeletal muscle. Color intensity scales from blue (ρ = −1, perfect negative correlation) to red (ρ = +1, perfect positive correlation). White grid cells indicate non-significant correlations (p > 0.05).

Table 2.

Demonstration of differential lipid compounds

Name	Log2FC	p-value	ROC	VIP	Up_Down
TG(18:4/4:0/16:0)	2.5737	1.29E-07	1	2.7019	Up
TG(20:5/4:0/18:1)	2.1622	1.10E-06	1	2.6177	Up
TG(18:1/4:0/16:0)	2.5202	1.71E-06	1	2.6355	Up
TG(20:5/4:0/14:0)	2.134	1.92E-06	1	2.6131	Up
TG(4:0/14:0/16:1)	2.0641	1.96E-06	1	2.6006	Up
TG(4:0/16:1/18:2)	1.9596	1.96E-06	1	2.5535	Up
TG(4:0/14:0/18:2)	2.1166	2.52E-06	1	2.6003	Up
TG(4:0/16:0/16:1)	2.8388	4.23E-06	1	2.5289	Up
TG(24:2/18:1)	2.0105	4.42E-06	1	2.5187	Up
TG(16:0/4:0/18:0)	2.419	6.40E-06	1	2.5777	Up
SM(d37:1)	−1.3315	6.60E-06	1	2.6306	Down
SM(d18:0/21:1)	−1.6306	0.0002	0.9531	2.3459	Down
PC(19:0COOH/22:5)	−0.903	2.69E-07	1	2.5182	Down
PC(18:2/22:6)	−1.0551	1.93E-06	1	2.3943	Down
PC(20:1/22:5CHO)	−1.1213	2.57E-06	1	2.386	Down
PE(22:2COOH/22:6)	−1.6539	5.88E-07	1	2.4597	Down
PE(17:0/22:6)	−0.7566	1.75E-06	1	2.4441	Down
PI(17:0/20:3)	−0.7636	0.0002	1	2.396	Down
PS(19:0/22:5)	−0.9039	0.0017	1	2.2068	Down
Cer(d18:1/19:0)	−1.0557	0.0021	0.9844	2.1599	Down

KEGG pathway analysis for lipids often used relaxed significance thresholds (not strictly limited to the p < 0.05) to detect biologically meaningful pathways^24,25. In this study, KEGG pathway analysis revealed significant enrichment in three key domains: lipid metabolism, nervous system development, and cellular signaling (Supplementary Fig. 12C). Specifically, the differential lipid metabolites showed significant enrichment in four functionally interconnected pathways: glycerophospholipid metabolism, phosphatidylinositol signaling, glycosylphosphatidylinositol-anchor biosynthesis, and autophagy regulation (Fig. 8C). Notably, cross-validation between differential metabolites and enriched pathways classified all six key ltered metabolite (shatidic acid (PA), lysophosphatidylcholine, cardiolipin (CL), phosphatidylinositol (PI), phosphatidylserine (PS), and PE) as gycerophospholipids, strongly implicating this lipid class in lisinopril’s therapeutic mechanism for ALS.

Discussion

SOD1^G93Atransgenic mice expressed human mutant SOD1, faithfully recapitulating familial ALS pathology^26,27. BI1, an ER-resident protein that bound Bcl-2 members to regulate apoptosis, demonstrated neuroprotective effects in stress conditions^28,29. The present study demonstrated that BI1 overexpression protected against apoptosis while deficiency exacerbated apoptosis in CNS injury models^30–32. Previous work by our group also demonstrated that the intravenous injection of BI1 plasmid delivery significantly extended survival in SOD1^G93A mice. While the gene therapy extended survival but raised immunogenicity concerns, our structure-based VS of FDA drugs identified lisinopril as a novel BI1 activator. Biochemical validation confirmed lisinopril upregulated BI1 expression in ALS mice. Acting as a dual modulator of the TGF-β pathways (Smad2/3 and PI3K/AKT/mTOR), lisinopril-enhanced BI1 activity improved metabolic status and augmented autophagy, leading to the mitigation of ALS pathologies such as muscle atrophy, motor neuron loss and NMJ dysfunction.

Lisinopril, clinically used for cardiovascular diseases, exerted therapeutic effects by blocking angiotensin II (Ang II) production and systemic RAS activation^15,33,34. Beyond its cardiovascular applications, lisinopril counteracted Ang II-mediated pathologies including fibrosis and oxidative stress^35–37. In the pathological process of ALS, NMJ degeneration and muscle atrophy involved pathological ECM remodeling with FN/Col1A1 accumulation³⁸. This fibrotic transformation progressively replaced contractile muscle tissue with non-functional collagenous deposits, ultimately impairing neuromuscular function³⁹. Previous studies have shown that lisinopril demonstrated multi-organ anti-fibrotic capacity, suppressing hepatic fibrosis and attenuating pulmonary collagen deposition through reducing TNF-α/TGF-β1 and Ang II/TGF-β1 pathway^40–42. Our study established a novel lisinopril’s therapeutic paradigm: upregulating BI1 to suppress TGF-β1/Smad2/3 signaling, reduce fibrosis markers (Acp5, FN), improve collagen ratios, and preserve muscle architecture, positioning it as a promising ALS therapeutic targeting myofibrosis.

Emerging evidence implicated lipid metabolism dysregulation in ALS pathogenesis. Our study elucidated that BI1 overexpression significantly increased TG levels in ALS mice, consistent with clinical observations linking sustained TG levels to slower disease progression and improved survival. Notably, the stage-specific T-CHO changes in ALS resulted from BI1 overexpression blocking motor neuron degeneration-induced intracellular cholesterol release in late-stage ALS and reducing metabolic consumption in early-stage ALS⁴³. Moreover, the dynamic LDL/HDL modulation showed phase-specific benefits: LDL/HDL ratio reduction limited monocyte and macrophage infiltration into the CNS during early disease stages²³, while the ratio increase improved respiratory function during advanced disease stages⁴⁴. These stage-dependent lipid metabolism alterations suggested circulating lipoprotein profiles as dynamic biomarkers in ALS progression. Our study validated BI1 gene therapy’s efficacy in ameliorating lipid metabolic dysfunction and offered a framework for developing stage-specific metabolic interventions.

The metabolic lesions of sphingolipids, glycerophospholipids and triglycerides collectively created a pathological lipid microenvironment that actively accelerated neurodegeneration^45,46. Systematic lipidomics characterized that lisinopril significantly reduced neurotoxic sphingolipids: pro-oxidative SM species (e.g., SM(d37:7), SM(d39:4), SM(d18:1/21:4)) and very-long-chain Cer species (e.g., Cer(d18:1/21:3), Cer(d18:2/21:2), Cer(d18:2/19:0)). These lipids promoted neuronal apoptosis via ROS generation (from PUFA chain oxidation), reduced mitochondrial membrane potential (ΔΨm), and p75NTR signaling^47,48. Consistent with this, our results demonstrated that lisinopril attenuated mitochondrial depolarization (JC-1/MPTP assays). It indicated that lisinopril’ s anti-apoptotic effect may be mediated, at least in part, through ΔΨm stabilization via sphingolipid metabolism modulation reduced oxidative stress and apoptosis. Targeting sphingolipids was further supported by neuroprotective S1PR antagonists in clinical evaluation⁴⁹. Lisinopril also specifically modulated glycerophospholipid metabolism, reducing pro-inflammatory PC isoforms (e.g., PC (19:0/20:4) and PC (20:1/20:4)), oxidized neurotoxic PC species (e.g., PC (19:0COOH/22:5)), and DHA derivatives (e.g., PC (18:2/22:6)). Pro-inflammatory PCs activated phospholipase A2 (PLA2), triggering arachidonic acid (AA) release and inflammation⁵⁰. In addition, oxidized PCs promoted neuronal apoptosis through TLR9 pathway activation⁵¹. Interestingly, our findings indicated that lisinopril promoted spinal motor neuron survival and preserved myelin integrity, suggesting a potential mechanism through which it may mitigate spinal cord lesions via selective modulation of PC metabolism. Furthermore, evidence indicated that coordinated PE/PI reduction suppressed PI3K/AKT/mTOR and initiated autophagosome biogenesis via ULK1 complex release⁵². Thus, our finding that lisinopril activated autophagy via inhibiting the TGF-β1/PI3K/AKT/mTOR pathway suggested that this effect may originate from the drug’s selective reduction of key autophagy-modulating lipids (PS, PE, and PI).

Notably, lisinopril increased specific butyrate-containing TGs (e.g., TG (4:0/14:0/18:2) and TG (20:5/4:0/18:1)). Li, X. and Parrella, E. et al. reported that butyrate, functioning as a histone deacetylase (HDAC) inhibitor, enhanced mitochondrial biogenesis and upregulated nerve growth factor (NGF) via epigenetic modulation, thereby improving motor function, neuronal survival, and lifespan in ALS mice^53,54. Interestingly, our finding that lisinopril promoted mitochondrial biogenesis suggested a potential mechanism whereby the drug’s neuroprotection might be facilitated by improved butyrate bioavailability, a link that warranted further investigation.

In summary, our study revealed that lisinopril exerted multi-target neuroprotective effects in ALS by activating BI1. It improved lipid metabolism and reduced fibrosis by inhibiting TGF-β1/Smad2/3 pathway, enhanced autophagy via PI3K/AKT/mTOR regulation, and remodeled muscle energy metabolism through fiber-type switching. In the future, our studies will validate lisinopril in higher mammals to assess efficacy and safety. Additionally, we will also investigate BI1 pathway crosstalk to develop combinatorial ALS therapies.

Materials and methods

NSC-34 cells were obtained from the BeNa Culture Collection (Suzhou, China). siRNA TMBIM6-Mouse-570 and siRNA mate were bought from GenePharma (Shanghai, China). pcMV6/TMBIM6-Myc-DDK-tagged was bought from ORIGENE (MD, United States). NSC-34-BI1 14# cells, pcDNA3.1/LacZ-Myc-His and pcMV6/SOD1^G93A-Myc-Flag-DDK-tagged were stored in the Fisher Laboratory, College of Life Sciences, Jilin University. TUNEL apoptosis assay kit, DAPI solution (1 mg/mL) and lipofectamine gene transfection reagent were acquired from Meilunbio (Dalian, China). Opti-MEM was bought from Gibco, Thermo Fisher Scientific (MA, United States). The serine protease inhibitor PMSF (100 mM) and ATP content assay kit were provided by Solarbio (Beijing, China). TG, T-CHO, LDL-C, HDL-C and HK assay kits were bought from Nanjing Jiancheng Bioengineering Institute. MegaFi™ Pro One Step RT-PCR was obtained from ABM (BC, Canada). CCK-8 assay kit was purchased from Selleck (TX, United States). JC-1 and MPTP assay kits were bought from Abbkine (Wuhan, China). Picro sirius red stain was provided by G-CLONE (Beijing, China). Lisinopril was purchased from MedChem Express (NJ, United States). Anti-Myh7, Anti-Myh1 and Luxol Fast Blue Myelin stain kit was obtained from Servicebio (Wuhan, China).

Total RNA isolation, library preparation and sequencing

To systemically deliver the BI1 expression plasmid to major organs through the circulation, we employed tail vein injection. Specifically, on day 90, the mice in the ALS BI1 treatment group (n = 10) received an injection of 100 μL sterile saline containing 30 μg of the BI1 plasmid via the tail vein. Concurrently, mice in the wild type (WT) group and ALS CON group (n = 10) were administered an equal volume (100 μL) of sterile saline⁵⁵. Then, skeletal muscle tissue samples (100 mg) from WT, ALS CON, and ALS BI1 mice were homogenized in TRIzol reagent for total RNA extraction, with RNA quality verified using an Agilent 2100 Bioanalyzer. Poly(A)⁺ mRNA was enriched using oligo(dT) magnetic beads and fragmented for cDNA synthesis. First-strand cDNA was generated with M-MuLV reverse transcriptase, followed by RNA degradation and second-strand synthesis using DNA polymerase I to create double-stranded cDNA. The double-stranded cDNA was then purified, end-repaired, and poly(A) was added, followed by ligation of the sequencing junctions. The product was subsequently size-selected (370–420 bp) using AMPure XP beads. After PCR amplification and final purification, library quality was assessed through Bioanalyzer profiling and qPCR quantification. Pooled libraries were sequenced on an Illumina NovaSeq 6000 platform to generate 150 bp paired-end reads, with base calling and demultiplexing performed using CASAVA software.

Read mapping and data processing

The raw sequencing data underwent stringent quality control processing to ensure analytical reliability. We removed adapter reads, eliminated reads containing ambiguous bases, and filtered out low-quality reads. The resulting high-quality clean reads were used for subsequent analysis after verifying quality metrics including Q20, Q30 and GC content distribution. The reference genome and annotation files were then obtained, and the genome index was built using HISAT2 (v2.0.5). The clean reads were then aligned to the reference genome with HISAT2 (v2.0.5). Gene-level quantification was performed using featureCounts. Following this, the gene expression levels were normalized and reported as FPKM values. The FPKM calculation incorporated the number of fragments mapped to each gene, total mapped fragments, and gene length to enable accurate cross-sample comparisons.

Differential expressed genes (DEGs) analysis

DEGs analysis was conducted using DESeq2 (v1.20.0) with false discovery rate (FDR) control via the Benjamini-Hochberg procedure. Genes meeting the significance threshold (adjusted p < 0.05) were identified as differentially expressed.

DEGs enrichment analysis

Functional enrichment analysis was performed using clusterProfiler (v3.8.1) with gene length bias correction. Significant enrichment was defined as Benjamini-adjusted p < 0.05 for Gene Ontology terms (cellular components (CC), molecular functions (MF), and biological processes (BP)) and KEGG pathways.

Gene enrichment analysis

Gene Set Enrichment Analysis (GSEA) circumvents arbitrary differential expression thresholds by evaluating genome-wide expression patterns. For our analysis, whole-transcriptome expression data were processed using GSEA v4.3.3 with the mh.all 2024 gene set collection to optimize enrichment detection.

Protein-protein interaction (PPI) network analysis of DEGs

We constructed a PPI network using STRING (v11.5) with mouse DEGs from RNA-seq, then analyzed the network in Cytoscape (v3.9.1) using CytoHubba, CytoNCA and MCODE. Finally, we identified the intersection of these three analyses to obtain core regulatory genes.

Weighted gene co-expression network analysis (WGCNA) analysis

We performed WGCNA (R package v1.72) to identify phenotype-associated gene modules in WT, ALS and ALS BI1 groups. The most correlated yellow module was analyzed in Cytoscape (v3.9.1) using MCODE to identify hub genes (weight values greater than 0.24).

Determination of biochemical indices of lipid metabolism

Skeletal muscle samples (45 ± 5 mg) were pulverized in liquid nitrogen, then homogenized in 9 volumes PBS using a cryogenic grinder (3 × 30 s, at −30 °C). After centrifugation (3000 × g, 10 min), supernatants were aliquoted for BCA protein assay and lipid analysis (TG, T-CHO, LDL-C, HDL-C).

qPCR

The results of the RNA sequencing were validated using qRT-PCR. The total cDNA was synthesized using a reverse transcriptase kit. qRT-PCR was performed using a SYBR green assay on a QuantStudio^TM5 (Thermo Fisher Scientific, MA, United States). The specific quantitative primers for 11 transcripts are listed in Table 3. Each 20 µL reaction volume contained 7.8 µL of H₂O, 0.4 µL of each primer, 0.4 µL of ROX Reference Dye, 1 µL of cDNA and 10 µL of 2× SYBR Green Pro Taq HS Premix. The conditions were as follows: an initial single cycle of 95 °C for 30 s; 40 cycles of 95 °C for 15 s and 60 °C for 34 s; and a final extension step at 60 °C for 5 min. Gene expression levels were normalized to ACTIN to determine the relative expression using the 2(−ΔΔCt) value method. Significant differences in gene expression were analyzed using GraphPad Prism 7.

Table 3.

The primers for qPCR

Gene Name	Forward Primer (5′ to 3′)	Reverse Primer (5′ to 3′)
Actin	CCTTCCTGGGCATGGAGTC	TGATCTTCATTGTGCTGGGTG
Pparg	GCCAAGGTGCTCCAGAAGATGAC	GTGAAGGCTCATGTCTGTCTCTGTC
Acp5	CCTGAGATTTGTGGCTGTGG	TCTTGTCGCTGGCATCGTG
FN	GACCACCACTCCCAAAAATG	TTGCAAACCTTCAATGGTCA
Tgf-β1	GAAGGACCTGGGTTGGAAGT	CGGGTTGTGTTGGTTGTAGA
Angptl4	GAGATCCCCAAGGCGAGT	ACAGTTAAGGTCCCCACGGA
Lpin2	GAAGTGGCGGCTCTCTATTTC	AGAGGGTTACATCAGGCAAGT
Agpat2	TTTGAGGTCAGCGGACAGAA	CCGACTCGAAGATGCTGGTT
Pde3b	GAGGCACAGCAACCAAATATCG	GAGAATCCTTCCTGATTTTTCTCCC
Stab2	CTGAAGCCCCAGAACTGACAA	TGCAATCTCGAACCCCGAC
Pnpla2	ACCATCTGCCTTCCAGACTGT	TAGCTGACGCTGGCATTCTT

Detection of ATP content

The ATP content in mouse skeletal muscle was quantified using a commercial ATP assay kit. Initially, a fresh working solution was prepared by mixing reagents 2–6 in precise ratios (0.2:0.2:0.02:0.08:0.02 mL respectively) and kept on ice for immediate use. Approximately 100 mg of skeletal muscle tissue was homogenized in ice-cold extraction buffer at a 1:10 (w/v) ratio using a cryomill (60 Hz, 3 × 30 s, −30 °C) to ensure complete tissue disruption while preserving ATP integrity. The homogenate was centrifuged at 10,000 × g for 10 min at 4 °C. The supernatant was then extracted with 500 μL chloroform through vigorous vortexing for 30 s, followed by centrifugation at 10,000 × g for 3 min at 4 °C to separate phases. The aqueous upper phase containing ATP was carefully collected for analysis. For the enzymatic assay, the sample or ATP standard was mixed with Reagent I and working solution. The initial absorbance at 340 nm (A1) was recorded immediately. Following a 10-min incubation at 37 °C, the final absorbance (A2) was measured. Finally, the content of ATP was calculated according to the formula.

HK activity assay

Mouse skeletal muscle samples (40–50 mg) were precisely weighed and homogenized in ice-cold saline (1:10 w/v) using a cryomill (60 Hz, 3 × 30 s, −30 °C) to ensure complete tissue disruption while maintaining enzyme stability. The homogenate was centrifuged at 8000 × g for 10 min at 4 °C to obtain a clear supernatant, which was immediately placed on ice for analysis. The hexokinase (HK) activity assay was performed by first preheating Reagent I and Reagent II at 37 °C for 10 min to ensure optimal reaction conditions. A working solution was then prepared by mixing reagents 1–3 in precise ratios (150:50:2). For the enzymatic reaction, 10 μL of tissue supernatant was combined with 190 μL of the pre-warmed working solution. The initial absorbance at 340 nm (A1) was recorded within 30 s of mixing using a spectrophotometer (TECAN, Switzerland) maintained at room temperature. After exactly 20 m of incubation at room temperature, the final absorbance (A2) was measured. Finally, HK activity was calculated according to the formula.

Pretreatment of BI1 protein and small molecules

The Mus musculus BI1 protein sequence (UniProt ID: Q9D4F1) was retrieved from NCBI RefSeq (Accession: NP_001346642.1). Structural prediction was performed using AlphaFold2 (v2.3.2). The quality of the model was rigorously validated through MolProbity (v4.5.1), supplemented by comprehensive evaluation in SAVES v6.0 incorporating PROCHECK algorithm. The structural binding pocket of BI1 was characterized through complementary computational approaches using Discovery Studio 2019 (DS 2019) and DoGSiteScorer. Then, we obtained a library of 1,615 FDA-approved small molecules from the ZINC20 database. Following structural refinement of BI1 and small molecules, docking simulations were performed focusing on the predicted binding site.

Batch molecular docking

Molecular docking was performed using a hierarchical screening protocol with BI1 as the target protein. Initial high-throughput screening employed LibDock in DS 2019, retaining ligand poses with LibDock scores ≥90 for subsequent analysis. For each compound, we calculated the mean binding energy across all conformations and selected the top 200 ranked molecules. These candidates underwent refined docking using CDOCKER with the CHARMm force field, applying stringent selection criteria (-CDOCKER ENERGY > 55 kcal/mol) to identify the highest-affinity binders. This multi-stage approach ensured both screening efficiency and binding pose accuracy.

CCK-8

NSC-34 motor neuron-like cells were seeded in 96-well plates at 1 × 10⁵ cells/mL (100 μL/well) and cultured for 24 h at 37 °C/5% CO₂ to achieve 80–90% confluency. Treatment groups received lisinopril (0–40 μM), while control groups (LACZ and SOD1^G93A) underwent media replacement. After 1 h pre-treatment, cells were transfected with either LACZ or SOD1^G93A plasmids (0.2 μg/well) using Meilunbio Gene Transfection Reagent. Following 24 h incubation, cell viability was assessed by adding CCK-8 reagent (10% v/v in serum-free medium) and measuring absorbance at 450 nm every 30 min for 3 h using a microplate reader.

TUNEL staining

NSC-34 cells were plated in 24-well plates (5 × 10⁴ cells/well) and divided into four experimental groups: LACZ, SOD1^G93A, LACZ + lisinopril (15 μM), and SOD1^G93A + lisinopril (15 μM). After 24 h adhesion, cells were pretreated (1 h) with lisinopril or PBS before plasmid transfection (1 μg). At 24 h post-transfection, cells incubated with TUNEL reaction mixture (37 °C, 1 h, dark). Nuclei were counterstained with DAPI (5 min, RT) after three PBS washes (5 min each). Finally, samples were imaged using a Zeiss LSM 900 confocal microscope (Zeiss, Germany).

MPTP fluorescence detection of mitochondrial membrane permeability

Following the same pretreatment as the TUNEL assay, cells were incubated with 250 μL of fluorescence quenching solution at 37 °C for 45 min. The solution was replaced with fresh pre-warmed complete medium and incubated for an additional 30 min at 37 °C. The nuclei were then counterstained with DAPI (5 min, RT) after three PBS washes (5 min each). Samples were mounted with antifade medium and imaged using a Zeiss LSM 900 confocal microscope.

JC-1 mitochondrial membrane potential assay

NSC34 cells were gently washed with PBS (pH 7.4) and incubated with a 1:1 mixture of complete culture medium and JC-1 working solution for 20 min at 37 °C in 5% CO₂. For nuclear counterstaining, cells were incubated with DAPI solution (1 μg/mL in PBS) for 5 min at 37 °C. Finally, samples were mounted with antifade medium and imaged using a Zeiss LSM 900 confocal microscope.

Immunofluorescence staining of MAP2 and β-Tubulin

Cells underwent the same pretreatment protocol as described for the TUNEL assay. Following two washes with ice-cold PBS (pH 7.4), the cells were fixed with 4% paraformaldehyde (PFA) for 30 min at room temperature. Following three 5-min washes with PBS, cells were permeabilized with 0.3% Triton X-100 for 7 min at room temperature, followed by another three PBS washes. Samples were then blocked with 5% lamb serum for 45 min at room temperature. Primary antibodies against MAP2 (1:300) and β-Tubulin (1:300) were diluted in blocking buffer and incubated overnight at 4 °C. After a 24-h incubation period, cells were washed four times with PBST (PBS + 0.1% Tween-20), 7 min per wash. Fluorescent secondary antibodies (1:200 in blocking buffer) were applied for 2 h at room temperature in the dark, followed by four 7-min PBST washes. Nuclei were counterstained with DAPI (1 μg/mL in PBS) for 5 min at room temperature, with three final 5-min PBS washes. Finally, samples were mounted with antifade medium and imaged using a Zeiss LSM 900 confocal microscope.

siRNA and transfection

First, the small interfering RNA (siRNA) for BI1 gene was synthesized by GenePharma, with the following sequence: sense: GCGUGACCUCUUCCUAGATT; antisense: UCUAGGAAGAGGUCAACGCTT. Then, according to the manufacturer’s instructions, siRNA-TMBIM6-Mouse-570 and siRNA-NC were transfected into NSC-34 cells using siRNA mate.

Western blot (WB)

Tissue Sample Processing: Skeletal muscle tissue (15 mg) was homogenized in RIPA buffer containing protease inhibitors using cryogenic milling (60 Hz for 3 × 30 s, −30 °C). The lysates were then centrifuged at 12,000 × g for 5 min at 4 °C, with supernatants collected in pre-chilled tubes. Cell Sample Processing: SOD1^G93A was transfected into NSC-34 cells, as well as NSC-34 cells with BI1 overexpression and BI1 knockout. Subsequently, the cells were treated with lisinopril. After 24 h, the cell samples were processed by lysis in RIPA buffer containing protease inhibitor, scraping, and centrifugation (12,000 × g for 5 min, 4 °C). Protein concentrations were determined by BCA assay before loading 20 μg protein per lane for SDS-PAGE separation. Primary antibodies: anti-GAPDH (Proteintech, Wuhan, China), anti-TGF-β1 (Proteintech, Wuhan, China), anti-Smad2/3 (Abcam, United Kingdom), anti-p-Smad2/3 (HUABIO, Hangzhou, China), anti-PI3K (Cell Signaling Technology, MA, United States), anti-p-AKT (Selleckchem, TX, United States), anti-p-mTOR (Proteintech, Wuhan, China), anti-p-ULK (ABclonal, Wuhan, China), anti-p-70S6K (ABclonal, Wuhan, China), anti-Beclin1 (ABclonal, Wuhan, China), anti-LC3 (Proteintech, Wuhan, China), anti-p62 (Proteintech, Wuhan, China), anti-BI1 (Santa Cruz Biotechnology, TX, United States), anti-Bax (Santa Cruz Biotechnology, TX, United States)。Secondary antibodies: goat anti-rabbit IgG (H + L)/HRP (Proteintech, Wuhan, China) for TGFβ1, Smad2/3, p-Smad2/3, PI3K, p-AKT, p-ULK, p-70S6K, Beclin1, LC3, p62 and GAPDH, while goat anti-mouse IgG (H + L)/HRP (Proteintech, Wuhan, China) for p-mTOR, BI1 and Bax.

Behavioral experiments

SOD1^G93A transgenic mice and wild-type (WT) mice (C57BL/6 J strain, both male and female, 2 months of age) were respectively procured from Shanghai Southern Model Biotechnology and SPF (Beijing) Biotechnology. Animals were housed in ventilated microisolator cages (4–6 mice/cage) under specific pathogen-free conditions with 12:12 light-dark cycles, ad libitum access to autoclaved food and water. All animal experiments comply with the National Institutes of Health guide for the care and use of Laboratory animals (NIH Publications No. 8023, revised 1996) and approval from the Institutional Animal Care and Use Committee of Jilin University. The work has been reported in line with the ARRIVE guidelines 2.0. Committee of Experimental Animal Welfare Ethics, College of Life Sciences, Jilin University. Title of the approved project is “The effect of Lisinopril on ALS model mice”. Appl. No. is YNPZSY2024047, Date of approval is 2024, Aug, 1st.

Mice running experiment

The mice were acclimated to treadmill running over a 3-day training period. For formal testing, the mice were subjected to constant speeds (females: 15 m/min; males: 20 m/min) during a 5 min baseline period. The exhaustion protocol was initiated using brief electrical stimulation (0.2 mA) whenever mice left the running belt, with testing terminated after 20 stimulation events or when mice failed to resume running within 15 s post-stimulation. The exhaustion time of the mice was counted to determine whether the motor ability was affected.

Mouse grip strength test

The grip strength of mice was quantitatively assessed using an HP-50N digital push-pull tester connected to a computer for data acquisition. During the testing procedure, the mice were gently held by the tail and allowed to naturally grasp the triangular pull bar with all four limbs. A smooth, consistent rearward traction was applied parallel to the sensor plane until the mice released its grip. Each mouse completed three consecutive trials with a 150 s rest interval between trials to allow full recovery and to prevent fatigue-related performance declines.

Mouse footprint test

Prior to testing, animals were acclimated to the testing apparatus (a 50 cm × 10 cm narrow runway lined with white paper) for 5 min. The hind paws were then coated with non-toxic blue ink using a fine brush. Mice were gently placed at one end of the runway and allowed to walk freely toward an enclosed dark goal box at the opposite end. Three complete walking trials were recorded per animal.

Immunohistochemical (IHC) staining

IHC staining was performed on paraffin-embedded skeletal muscle sections from WT, untreated ALS CON and ALS Lisinopril mice. Briefly, 4-μm-thick sections were mounted using a 42 °C water bath, dried at 60 °C for 30 min, and then processed through standard dewaxing and hydration steps: xylene (three times, 5 min each), followed by a graded ethanol series (absolute, 95%, and 75%, 1 min each), and finally distilled water (5 min). Antigen retrieval was carried out by microwaving slides in pH 8.0 EDTA buffer for 5–8 min, followed by cooling to room temperature. Endogenous peroxidase activity was quenched with 3% hydrogen peroxide for 10 min at room temperature. After washing with PBS, sections were blocked with goat serum at 37 °C for 60 min and subsequently incubated with primary antibodies against BI1, TGF-β1, PI3K, Beclin1, LC3B, and p62 in a humidified chamber at 37 °C for 2 h. Following PBS washes, sections were incubated with appropriate HRP-conjugated secondary antibodies (goat anti-mouse for BI1; goat anti-rabbit for all other targets) at 37 °C for 30 min. Color development was performed using a DAB substrate kit, and the reaction was monitored microscopically and stopped when optimal signal-to-background contrast was achieved. Nuclei were counterstained with hematoxylin. Finally, sections were dehydrated through a graded ethanol series, cleared in xylene, and mounted with a coverslip.

Skeletal muscle hematoxylin-eosin (H&E) staining

Paraffin-embedded skeletal muscle sections (10 μm) were deparaffinized and rehydrated through standard xylene and graded ethanol series before staining. Initially, the sections were immersed in hematoxylin solution for 5 min to stain nuclei, followed by rinsing in distilled water. Differentiation was then performed using 1% hydrochloric acid in ethanol to remove excess stain, immediately followed by bluing in 0.6% ammonia water to restore nuclear chromatin to its characteristic blue color. After thorough rinsing, the cytoplasmic structures were counterstained with eosin solution for 2 min. The stained sections were then dehydrated through 100% ethanol solutions, cleared in xylene, and mounted with synthetic resin under glass coverslips. Microscopic examination was conducted using a Nikon Eclipse Ni-U brightfield microscope.

Transmission electron microscopy of cells and muscle tissue

Cells or freshly dissected muscle tissue fragments (1–3 mm³) were initially fixed in 2.5% glutaraldehyde overnight at 4 °C, followed by thorough PBS washing. Secondary fixation was performed using 1% osmium tetroxide for 2–4 h at room temperature, protected from light to preserve staining quality. Following osmication, samples underwent graded dehydration in an ethanol series (30–100%) followed by 100% acetone. The dehydrated specimens were then permeabilized with epoxy resin through a graded acetone-resin transition series, culminating in embedding medium. Samples were then carefully oriented in silicone molds and polymerized at 60 °C for 48 h to achieve optimal resin hardness. Ultrathin sections were cut using the ultramicrotome and collected on copper grids. The sections were stained with lead citrate (10 min) and methanolic uranyl acetate (20 min), with thorough ddH₂O washing between staining steps. After complete drying, the grids were examined using a transmission electron microscope operated.

Luxol Fast Blue (LFB) staining of spinal cord

Following deparaffinization with xylene, the spinal cord sections were rinsed in distilled water. Tissue sections were then immersed in 0.1% LFB solution and incubated for 12 h in a temperature-controlled oven to ensure even myelin staining. Subsequent to this, controlled differentiation was achieved through sequential treatments with 0.05% lithium carbonate solution, and 70% ethanol. The differentiation process was monitored microscopically at 1 min intervals until optimal contrast was achieved between the deep blue-stained white matter (myelinated fibers) and pale blue-gray matter (neuronal cell bodies).

Sirius red staining

Sections were stained with saturated picric acid solution containing 0.1% Sirius Red for 15 min at room temperature, ensuring complete coverage of the muscle tissue. After staining, the excess dye was removed by rapid rinsing in two changes of 100% ethanol (30 s each). Following this, the sections were dried in a 60 °C oven for 10 min, followed by transparency of the sections with xylene. Sections were mounted with neutral resin under coverslips and examined under polarized light microscopy using both brightfield and polarized light modes.

Immunofluorescence double label staining of fast and slow muscle

Transverse paraffin sections of skeletal muscle were deparaffinized. the sections were then fixed in 4% paraformaldehyde (15 min, RT), washed in PBS (3 × 5 min), and permeabilized with 0.1% Triton X-100 in PBS (10 min, RT). After blocking with 5% goat serum (1 h, RT), the sections were incubated with anti-Myh7 primary antibody (1:200 in blocking solution) overnight at 4 °C. After 24 h, the sections were washed with PBS (3 × 5 min) and incubated with HRP-conjugated secondary antibody (1:500) for 2 h at RT protected from light. Following PBS washes (3 × 5 min), tyramide-CY3 was applied for 20 min at RT. Sections underwent additional antigen retrieval with citrate buffer to remove bound antibodies while preserving CY3 signal. After re-blocking (1 h, RT), sections were incubated with anti-Myh1 antibody (1:200) overnight at 4 °C, followed by HRP-secondary antibody (2 h, RT) and tyramide-488 incubation (20 min, RT). Sections were stained with DAPI (1 μg/mL, 5 min, RT), after which they were washed in PBS (3 × 5 min). Imaging was performed on a confocal microscope.

Spinal Nissl staining

The sections were permeabilized with Proteinase K (0.2 μg/mL in PBS) at 37 °C for 30 min. The enzymatic reaction was stopped by PBS washing. Sections were then stained with 0.1% cresyl violet solution for 5 min at room temperature, followed by rapid differentiation in 1% glacial acetic acid until gray matter neurons showed distinct Nissl bodies against clear white matter background. The reaction was terminated by ddH₂O rinsing. Finally, the sections were examined under brightfield microscopy (Nikon Eclipse Ni-U, Japan).

NMJ staining

Skeletal muscle cryosections (50 μm) were fixed in 4% paraformaldehyde (15 min, RT) and permeabilized with 0.1% Triton X-100 (10 min, RT). After blocking with 5% normal sheep serum containing 0.1% Tween-20 (1 h, RT), the sections were incubated with primary antibodies against neurofilament H (NF-H) and SV2A (1:200 dilution) overnight at 4 °C. Following PBST washes, neuromuscular junctions were double-labeled using FITC-conjugated secondary antibody combined with α-bungarotoxin-Cy3 to simultaneously visualize nerve terminals (FITC) and acetylcholine receptors (Cy3). After 2 h incubation (RT, dark), sections were washed and mounted with antifade medium.

Sample preparation and mass spectrometry

ALS mice received either lisinopril (30 mg/kg/day) or equal amount of PBS via oral gavage at 60 days (14-day treatment period). At 150 days, mice from both groups were euthanized and hindlimb skeletal muscles were collected. Freshly dissected muscle samples were immediately flash-frozen in liquid nitrogen to preserve lipid profiles and stored at −80 °C until analysis. Samples were then shipped on dry ice to Novogene Co., Ltd. (Beijing, China) for lipid extraction and LC-MS/MS. Eight biological replicates were processed for each group, with each sample consisting of precisely weighed 150 mg skeletal muscle tissue.

Screening for differential lipid compounds

The LC-MS/MS raw data were processed through LipidSearch™ 4.2 software for peak alignment. Following quality control filtering, the normalized lipidomic dataset underwent multivariate analysis including unsupervised PCA and supervised PLS-DA to characterize global lipid profile differences between groups. Differential lipids were identified by applying combined thresholds of VIP > 1.0 from PLS-DA analysis, fold change >1.5 or <0.667, and p < 0.05 from Student’s t test with FDR correction. The results were then visualized through volcano plots generated using ggplot2 in R^56–58.

Cluster analysis and enrichment of differential lipid metabolites

Hierarchical clustering analysis was performed using the R package pheatmap (v1.0.12). In addition to this, the identified differential metabolites were imported into Metaboanalyst 4.0 web version for KEGG enrichment analysis.

Statistics and reproducibility

Statistical analyses were conducted using GraphPad Prism 7. Data were presented as the mean standard deviation (SD). The data were analyzed for normal distribution. Differences between multiple groups were checked using one-way ANOVA and two-way ANOVA with post hoc Bonferroni correction. Differences between two groups were analyzed by a two-tailed unpaired Student’s t test. p < 0.05 was considered statistically significant. The sample size and number of replicates for all experiments are indicated in the corresponding figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Hanlan Yin performed experiments, analyzed and interpreted the data, and drafted the manuscript. Zhichao Ren, Yan Zhang, Yuxiang Wang, Yiyang Sun and Wenfu Yan contributed to writing the manuscript. Xueqi Fu and Fuqiang Zhang handled funding. Linlin Zeng designed experiments, supervised all aspects of the study and revised the manuscript. All authors read and approved the final manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Joao Valente. A peer review file is available.

Data availability

Numerical source data underlying graphs in the manuscript can be found in Supplementary data 2. RNA sequencing data generated in this study have been deposited in the Gene Expression Omnibus (GEO) database under accession number GSE301454. Lipidomics data are stored in the MetaboLights database, specifically at: https://www.ebi.ac.uk/metabolights/MTBLS12681.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09930-2.