Isoginkgetin reduces ALS symptoms in animal and patient cell models by activating a mitochondrial cleanup process

Many mitophagy inducers, including CCCP, depolarize mitochondria and/or impair mitochondrial function, making them unsuitable as therapeutic drugs (Georgakopoulos et al, 2017). Therefore, the 20 top candidate mitophagy inducers were scored for relative ATP production in galactose- and glucose-containing medium, and 15% lower ATP production in galactose was used as a cut-off (i.e., maximum tolerable evidence of mitochondrial toxicity). Based on this criterion, 8 compounds were selected for further study (Fig. 1B). ISO was identified for having the strongest ability to activate mitophagy in CCCP-depolarized mitochondria (Fig. 1C). Its molecular structure is shown in Fig. 1D; its 50% cytotoxic concentration (CC₅₀) was also measured (Fig. 1E; 16.45 μM). Immunoblot analysis showed that application of ISO (10 μM, 24 h) led to a significant degradation of the outer and inner mitochondrial membrane proteins Mitofusin 2 (MFN2) and mitochondrial cytochrome c oxidase subunit 2 (MTCO2) (Fig. 1F), and flow cytometry detected mt-mKeima as evidence of mitophagy in ISO-treated cells (Fig. EV1C,D), confirming that ISO is a potent mitophagy agonist.

We found that ISO treatment remarkably increased the expression of pSer65-Ub and Parkin self-ubiquitination (Fig. 1F). Live-cell imaging showed that ISO induced substantial Parkin translocation (Fig. EV1E,F), indicating enhanced Parkin E3 ligase activity. Accordingly, ubiquitination of MFN2, a known Parkin substrate, also increased in ISO-treated cells (Fig. 1F). For three receptor proteins, BNIP3, NIX and FUNDC1, ISO treatment did not cause any changes in their expression levels (Fig. 1F). After siRNA knockdown (KD) of PINK1, BNIP3, NIX, FUNDC1 and BNIP3/NIX, respectively, in YFP-Parkin-mt-mKeima HeLa cells, ISO-induced mitophagy was abolished only under PINK1 knocked down conditions (Fig. EV1G–I). Consistently, the PINK1 KD in YFP-Parkin HeLa cells blocks ISO treatment-induced expression of pSer65-Ub, ubiquitination/degradation of MFN2, and translocation of Parkin to mitochondria (Fig. 1G–I), suggesting that ISO specifically stimulates PINK1-Parkin-dependent mitophagy. Moreover, while CCCP caused nearly complete loss of mitochondrial membrane potential (MMP), MMP was only 5.65% lower in the presence of 10 μM ISO than in its absence (Fig. 1J).

To further explore whether ISO participates in the activated receptor-mediated mitophagy pathway, we worked on receptor-mediated mitophagy using a hypoxia system. We have successfully established this system as evidenced by the high expression of hypoxia-inducible factor 1α (HIF-1α) and significant degradation of mitochondrial membrane proteins MFN2 and MTCO2 (Fig. EV2C,D). However, ISO did not increase the expression of BNIP3 and NIX or change FUNDC1 in the hypoxic YFP-Parkin HeLa cells (Fig. EV2C), which suggests that ISO does not enhance the receptor-mediated mitophagy. The same experiment was performed in SH-SY5Y cells with similar results (Fig. EV2D). Under normoxic conditions, 5 μM ISO did not increase the expression of receptor proteins in either of the two cell types. Thus, all these results suggest that 5 μM ISO can specifically promote PINK1–Parkin-dependent mitophagy in cells containing CCCP-damaged mitochondria.

Mitochondrial stress activates PINK1 kinase, which in turn phosphorylates residue Ser65 in Ubiquitin forming pSer65-Ub and initiates mitophagy. When mitochondrial stress is resolved/removed, activated PINK1 is rapidly degraded and the abundance of pSer65-Ub decreases (Yi et al, 2024). We found that after removal (washout) of CCCP, pSer65-Ub expression in ISO-treated cells remained higher than in control cells (DMSO), supporting the concept that ISO stabilizes PINK1/TOM and prevents PINK1 inactivation/degradation (Fig. 3I). Importantly, no difference in MMP loss was found between DMSO- and ISO-treated cells, confirming that ISO is not a mitochondrial toxin (Fig. EV2H).

FL-PINK1 is quickly cleaved by PGAM5-associated rhomboid-like protease (PARL) at IMM after it is imported into mitochondria through the TOM complex at OMM. The cleaved PINK1 is extracted to the cytosol and degraded by proteasomes (Yi et al, 2024). Under physiological conditions, cells maintain a low level of PINK1. PINK1 P95A is resistant to cleavage, while PINK1 F104A is resistant to degradation (Deas et al, 2011; Yamano and Youle, 2013). We found that ISO still increased CCCP-induced pSer65-Ub expression in both PINK1 mutants, suggesting that the effect of ISO on stabilizing PINK1/TOM is not due to directly disturbing PINK1 cleavage and degradation (Fig. EV2I,J). Combining our analysis within the ISO treatment group, through the Seahorse oxygen consumption rate detection (Fig. 3J,K), as well as the changes in mitochondrial DNA copy number (Fig. 3L) and mitochondrial membrane potential (Fig. 3M), we summarize that ISO promotes mitophagy by stabilizing the PINK1/TOM complex on the OMM without compromising the normal function of mitochondria.

Human iPSC-derived ALS MNs and controls were used to further examine mitophagy in ALS. ALS patient-derived iPSCs included three commercially available cell lines as follows: UCLi004-A containing GGGGCC repeat expansion, PFIZi013-A containing TARDBP A382T mutation, and WC034i containing SOD1 D90A mutation (WiCell Research). Three control iPSC lines, UC-H1-iPSC, UC-H2-iPSC, and UC-12-iPSC, were prepared as previously described (Chong et al, 2018; Chong et al, 2022; Tan et al, 2019). All iPSC cell lines were differentiated into MAP2⁺/HB9⁺ MNs (Figs. 4G and EV3A) and GFP-labeled using the lentivirus Hb9::GFP reporter system (Hung et al, 2023) (Fig. EV3B). Differentiated MNs were cultured for 28 days to allow full maturation, at which point disease-associated phenotypes, such as swollen neurites, were observed in the ALS-iPSC-derived MNs (Fig. EV3B,C, white arrow). We found that the co-localization frequency of LAMP2 and MTCO2 in the ALS MAP2⁺ motor neurons was 40.3% lower than that in the healthy control group, indicating a mitophagy disorder (Fig. 4H,I). To further explore the mitophagy levels in ALS MNs, we used a commercial mitophagy dye (mtphagy) to conduct live-cell imaging within Hb9::GFP-positive motor neurons. We found that the mitophagy levels indicated by the mtphagy dye in ALS MNs were also 47% lower than those in the healthy control group, further indicating the presence of damaged mitophagy in ALS MNs (Fig. 4J,K). We also noted that both the expression of pSer65-Ub and the ATP content were lower than in control MNs (Fig. 4L–N). In addition, TMRE staining (Yi et al, 2024) demonstrated that MMP was lower in ALS MNs than in controls (Figs. 4O and EV3D). Electron microscopic (EM) studies revealed swollen mitochondria with severe degeneration of cristae and aggregated filamentous structures in the soma of ALS iPSC-derived MNs (Fig. 4P,Q). These results suggest that ALS MNs are characterized by defective PINK1-Parkin-dependent mitophagy and more abundant dysfunctional/damaged and/or depolarized mitochondria than control MNs.

A dose–response study with wild-type (WT) N2 worms from hatching through adult day 1 revealed that ISO was non-toxic up to 75 μM, with no significant adverse effects on egg laying/hatching or larval development/maturation (Fig. EV5A). The ability of ISO to promote mitophagy in vivo was investigated using our previously reported transgenic worms expressing either mt-Rosella or DCT-1/LGG-1 (Cao et al, 2022; Fang et al, 2019). mt-Rosella transgenic worms were generated by expressing the mitochondrial-targeted biosensor protein Rosella, which is fused with GFP and DsRed in worm neurons. Due to the sensitivity of GFP in the acidic lysosomal environment, an increase in the DsRed/DsRed+GFP ratio indicates an increased level of neuronal mitophagy. DCT-1/LGG-1 transgenic worms were generated by expressing the DAF-16/FOXO-controlled germline-tumor affecting-1 (DCT-1) mitophagy receptor fused with GFP together with the autophagosome marker protein LGG-1 fused with DsRed in neurons, and the increased number of the co-localization of DCT-1 and LGG-1 indicates the increased level of neuronal mitophagy. Furthermore, 15 µM (but not 5 µM) ISO induced significant neuronal mitophagy in both transgenic worm strains (Fig. EV5B–D).

SOD1 G93A transgenic worms exhibit several pathological features of human ALS, including progressive paralysis, progressive loss of function in MNs, and progressive neurodegeneration (Fig. 6D–I). Remarkably, exposure to 15 μM ISO from egg hatching to adult day 1 significantly improved MN function, swimming performance, and paralysis in ALS G93A nematodes (Fig. 6D,E); unc-25::GFP worms were crossed with WT or G93A worms to label the MNs and the connections between MNs. When exposure to 15 μM ISO continued from egg laying to adult day 8, ISO-treated worms began to experience MN loss/degeneration later than untreated (control) worms (Fig. 6F–H). More importantly, ISO improved the 15.4% median survival of G93A worms over the time course relative to untreated control worms (Fig. 6I). These results confirm that ISO protects against MN degeneration in an in vivo nematode model of human ALS.

Using pink-1 or pdr-1(C. elegans homolog of the mammalian Parkin) knockdown conditions in mt-Rosella worms crossed with SOD1 G93A, ISO-induced neuronal mitophagy was abolished (Fig. 6J,K), confirming that ISO induces PINK1-Parkin-dependent mitophagy. Similarly, knockdown of neuronal pink-1 and pdr-1 abrogated the beneficial effect of ISO on swimming ability in SOD1 G93A nematodes (Fig. 6L). These results provide evidence that ISO stimulates PINK1-Parkin-dependent neuronal mitophagy in SOD1 G93A nematodes.

We then wondered whether the protective effect of ISO in NMs might be mediated by a bacterial metabolite produced by E. coli OP50 in nematode media, rather than by ISO itself. To check this, SOD1 G93A nematodes were cultured in either UV-inactivated or control untreated E. coli OP50 (Fig. EV5E), and the results showed that ISO had the same protective effect in both conditions, suggesting that its activity is not dependent on bacterial metabolism. In addition, ISO did not interfere with the foraging behavior of N2 nematodes (Fig. EV5F) while it increased pharyngeal pumping in SOD1 G93A nematodes (Fig. EV5F), which could indicate that ISO protects the D-type MNs in the nematode head.

The therapeutic potential of ISO was investigated in C57BL/6 J (control) and SOD1 G93A transgenic mice, a classical ALS mouse model. Initial studies quantified ISO by ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) in blood and brain samples 0, 1, 2, 4, or 8 h after intraperitoneal injection of control mice with 0.3 mg/kg or 3 mg/kg ISO (Suhua, 1993). ISO was not detected in the brain (assay sensitivity 0.5 ng/mL) under these conditions (Fig. EV6A), suggesting that ISO is unable to cross the blood brain barrier (BBB) in mice. No adverse effects were observed in treated mice, but ISO was detected at expected levels in blood at all time points (Fig. EV6B).

Previous studies have shown that flavonoids can self-assemble into nanoparticles that are able to cross the BBB in rodents and other species (Yue et al, 2023). Consistent with this, self-assembled ISO nanoparticles (Nano-ISO) were successfully prepared using a published method (Fig. EV6C) (Liu et al, 2020). UPLC and EM analyses revealed that Nano-ISO contained uniform, well-defined, stable, monodispersed nanospheres (hydrodynamic diameter, 167.4 nm; polydispersity index, 0.121) carrying on average 390 μg/mL ISO at an estimated 43.6% maximal drug loading capacity (Fig. EV6D,E). Nano-ISO particles released ISO slowly over 6 days (Fig. EV6F). These results show that Nano-ISO is a self-assembling pure nanodrug with excellent drug loading capacity and stability that promotes sustained release of its bioactive payload.

In many cases, ALS is diagnosed after significant muscle atrophy, when MN loss has already occurred, resulting in a poor prognosis (Grad et al, 2017). Because electromyography (EMG) can detect neurogenic damage before pathological features of neurodegenerative disease emerge (Hulisz, 2018), EMG was used to monitor early stages of disease in SOD1 G93A mice (relative to control mice) (Fig. EV7A). More specifically, the compound muscle action potential (CMAP) of the gastrocnemius muscle was monitored after electro-stimulation of the sciatic nerve (Pollari et al, 2018). The results showed normal muscle function in 50-day-old mice, while CMAP was lower and continued to decrease in animals 60 days or older (Fig. EV7B–D). However, it is encouraging that EMG detected evidence of neurological dysfunction ~4 weeks earlier than the average onset of behavioral dysfunction in SOD1 G93A mice (Wen et al, 2021).

Histological studies of the anterior horn of the spinal cord in 60-day-old SOD1 G93A mice revealed more activated microglia and astrocytes as well as smaller somatic ChAT⁺ and NeuN⁺ in MNs than in control mice, with no reduction in the overall number of MNs (Fig. EV7E–J). Interestingly, male and female SOD1 G93A mice showed significant differences in disease onset, disease progression, and lifespan (Alves et al, 2011). In our laboratory, female mice exhibit less individual variability than male mice. Therefore, in subsequent studies, the efficacy of Nano-ISO was examined in 60-day-old female SOD1 G93A or control mice (WT female littermate mice) after intranasal dosing with Nano-ISO, Riluzole, or vehicle, where Riluzole was used as a positive control (Fig. EV7K). Neurobiological (hind limb tremor) and behavioral (hanging time) assessments were performed to evaluate disease progression. Animal lifespan/survival was also tracked. The results showed that the onset of hind limb tremor occurred later in Nano-ISO-treated mice than in vehicle-treated mice (Fig. 7D). SOD1 G93A mice treated with Nano-ISO for 4 or 5 weeks had longer hang time and an increase in lifespan than vehicle-treated controls (Fig. 7E,F, +8.02%, difference between mean values). In contrast, disease progression and lifespan showed no statistically significant differences in mice treated with Riluzole or vehicle (Fig. 7D–F). Subsequently, SOD1 G93A mice were dosed with Nano-ISO for 60 days and MN abundance and neuroinflammation in the lumbar spinal cord were assessed by immunofluorescence, based on the observation that SOD1 G93A mice experience progressive motor function loss in hind- and forelimbs (Gento-Caro et al, 2022). This revealed 38.5% fewer ChAT⁺ MNs in SOD1 G93A mice than in WT mice, and this deficit decreased to 20.6% in Nano-ISO-treated mice (Fig. 7G,H). In addition, microglia and astrocytes in Nano-ISO-treated mice had significantly lower levels of neuroinflammation than vehicle-treated mice, based on staining for inflammation markers Iba-1 and GFAP in MNs (Fig. 7I–K).

Next, co-expression of ChAT⁺, MTCO2, and LAMP2 (markers of mitophagy) was quantified in MNs in the anterior horn of the lumbar spinal cord. These markers were approximately 28.2% less frequently co-expressed in SOD1 G93A mice than in WT mice, with a similar reduction of pSer65-Ub, strongly suggesting abnormally low levels of PINK1-Parkin-dependent mitophagy (Fig. 7L–O). However, in SOD1 G93A mice treated with Nano-ISO for 60 days, co-localization of these mitophagy markers increased by 19.8% relative to vehicle-treated mice (Fig. 7L,M), with a concomitant increase in the level of pSer65-Ub (Fig. 7N,O), a reduction in mitochondrial swelling, and an increased abundance of mitolysosome-like structures adjacent to damaged mitochondria (Fig. 7P). These results suggest that ISO stimulates PINK1–Parkin-dependent mitophagy in spinal MNs thereby antagonizing ALS disease progression.

Postmortem spinal cord and brain samples of ALS patients and sex-matched healthy controls were obtained from the Netherlands Brain Bank (https://www.e-nbb.org↗). Sample details are available in Table EV1. The related experiments were performed at Akershus University Hospital (Norway) with approval from the Regional Committee for Medicine and Health Research Ethics (REK# 412997). Informed consent was obtained from all human subjects (next of kin), and the experiments conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report.

SOD1 G93A transgenic mice that express a G93A mutant form of human SOD1 (B6SJL-Tg (SOD1*G93A)1Gur/J, JAX stock No. 002726) and C57BL/6J mice (JAX stock No. 000664) were purchased from the Jackson Laboratory. All mouse strains used in this study were housed in the Laboratory Animal Centre of Sun Yat-Sen University, where they were maintained in a constant temperature and humidity environment with a 12h-light/12h-dark cycle. Hemizygous female mice were used in these experiments. Wild-type female littermate mice were used as controls. Nano-ISO was administered with Tribromoethyl alcohol (MCE, HY-B1372) as an anesthetic. For tissue sampling and brain imaging, mice were treated with Zoletil 50 (Virbac) and Xylazine (Sigma, X1126) and then euthanized using carbon dioxide. Animal experiments were carried out using protocols approved by the Institutional Animal Care and Use Committee (IACUC), Sun Yat-Sen University (SYSU-IACUC-2022-001202).

A total of 31 LCM-seq samples (including 23 spinal motor neuron samples from ALS patients and 8 from healthy controls) were used in this study; the relevant data were downloaded from the GEO database under accession numbers GSE76220↗ and GSE115130↗ (Krach et al, 2018; Nizzardo et al, 2020). Quality control was performed using SAMtools, and the results were collated using MultiQC. Adapter sequences and low-quality reads were removed using Trimmomatic. The remaining reads were aligned to hg38 build reference using STAR v2.5.2b with default parameters, followed by quantification through RSEM. Initially, Principal Component Analysis (PCA) was employed to scrutinize the impact of batch effects on gene expression data derived from two distinct studies (Subramanian et al, 2005). Once batch effects were mitigated, differential gene expression analysis was conducted using DESeq2. Differentially expressed genes (DEGs) met the criteria of an unadjusted P value < 0.05 and fold-change (FC) > | 1.2 | . GSEA was performed using the Bioconductor clusterProfiler R package (version 4.2.0). Significantly altered LFC values were analyzed by GSEA using the Molecular Signatures Database (hallmark gene sets). Heatmaps and bar plots were generated by R4.2.0 software.

YFP-Parkin-mt-mKeima, PINK1-Myc-YFP-Parkin, YFP-Parkin HeLa and SH-SY5Y cells were cultured in DMEM high-glucose medium (Gibco,11965092) containing 10% fetal bovine serum (FBS, Gibco, 10099-141) and grown at 37 °C in 5% CO₂. In hypoxic cultures, cells were cultured with normal medium under a hypoxic condition (1% O₂, 94% N₂, and 5% CO₂) using a multi-gas incubator (APM-30D; Astec Corporation). Cells were transfected with PINK1 P95A and PINK F104A expression plasmids using Lipofectamine 3000 (Thermo Fisher Scientific, L3000075). For gene silencing, cells were transfected overnight with 50 pmol targeted siRNAs using Lipofectamine 3000. Three iPSC lines, UC-H1-iPSC, UC-H2-iPSC, and UC-12-iPSC, were maintained in our laboratory. Three familial ALS iPSC cell lines, purchased from WiCell Research Institute Inc., harbored the following mutations: C9 (UCLi004-A, GGGGCC repeat expansion), TDP-43 (PFIZi013-A, TDP-43 A382T), and SOD-1 (WC034i, SOD1-D90A). iPSC cells were cultured in plastic microtiter plates or confocal dishes pre-coated with Matrigel (Corning, 354277) in mTeSR Plus basal medium (STEMCELL, 100-0276) in an incubator under 5% CO₂ at 37 °C. The culture medium was replaced daily.

C. elegans strains used in this study included mt-Rosella: N2; Ex[Punc-119TOMM-20::Rosella; rol-6(su1006)], IR2160: N2:Ex(002[Prab-3::dsRed::LGG-1Prab-3DCT-1::gfp;Pmyo-2::gfp], ngIs19 (unc-25p::GFP), ngIs27 (unc-25p::G93A SOD1) and TU3401: sid-1(pk3321)V;uls69[pCFj90(myo-2p::mCherry)+unc119p::sid-1]V. C. elegans cultures were maintained at 20 °C on NGM plates with Escherichia coli OP50 using standard methods. Bristol strain N2 worms were used as a control. Crosses were verified by observing the co-transfer of markers or by PCR. To monitor the level of mitophagy in MNs, a PHX5301 sybIs5301[Punc-25::DCT-1::GGGGS::ceBFP::unc-54 3'UTR, Punc-25::LGG1::GGGGS::Dsred2::unc-54 3'UTR, Lin44] stable integrated transgenic strain was generated by SunyBiotech with X-ray irradiation. The ceBFP plasmid was kindly provided by Dr. Ou Guangshuo.

Motor neurons were generated from iPSCs as described previously (Du et al, 2015). iPSCs were dissociated with dispase (1 mg/mL, Life Technologies, 17105) and split 1:6 on Matrigel-coated plates. On the following day, the medium was replaced with a chemically defined neural medium, containing Dulbecco's modified Eagle's medium (DMEM)/F12 (Gibco, C11330500BT), Neurobasal medium (Gibco, 21102-049) at a 1:1 ratio, 0.5 × N2 (Gibco, 17502-048), 0.5 × B27 (Gibco, 17504-044), 0.1 mM L-Ascorbic acid (TOCRIS, 4055-50-81-7) and 1× Glutamax (Gibco, 35050061). 3 μM CHIR99021 (Tocris, 252917-06-9), 2 μM DMH-1 (Tocris, 4126), and 2 μM SB431542 (Tocris, 1614) were added to the medium. The culture medium was changed every day. Human iPSCs were maintained under these conditions for 6 days and induced into neuroepithelial (NEP) cells. The NEP cells were then dissociated with dispase and split at a 1:6 ratio with the same medium as described above, and 0.1 μM Retinoic acid (RA) (Stemgent, 04-0021), 0.5 μM Purmorphamine (Stemgent, 04-0009), 1 μM CHIR99021, 2 μM DMH-1, and 2 μM SB431542 were added. The culture medium was changed every day. The NEP cells were maintained under these conditions for 6 days and differentiated into OLIG2⁺ motor neuron precursors (MNPs). The OLIG2⁺ MNPs were expanded in the same medium containing 3 μM CHIR99021, 2 μM DMH-1, 2 μM SB431542, 0.1 μM RA, 0.5 μM purmorphamine, and 0.5 mM valproic acid (VPA) (Stemgent, 04-0007) and split at a 1:6 ratio once weekly with dispase. OLIG2⁺ MNPs were frozen in 70% DMEM/F12, 20% fetal bovine serum, and 10% DMSO in liquid nitrogen until use. After thawing, cells were grown in expansion medium.

To induce motor neuron differentiation, OLIG2⁺ MNPs were dissociated with dispase and cultured in suspension in neural medium containing 0.5 μM RA and 0.1 μM purmorphamine. The culture medium was changed daily. After 6 days, OLIG2⁺ MNPs were differentiated into HB9⁺ motor neurons. HB9⁺ motor neurons were cultured with 0.5 μM RA, 0.1 μM purmorphamine, 0.1 μM Compound E (Tocris, 6476), and three neurotrophic factors, including insulin-like growth factor (IGF) (10 ng/mL, MCE, HY-P7018), brain-derived neurotrophic factor (BDNF) (10 ng/mL, MCE, HY-P7116A), and ciliary neurotrophic factor (CNTF) (10 ng/mL, MCE, HY-P7146) for 10 days. During differentiation, Y-27632 (SELLECKCHEM, S1049) was added for 12 h during each passage. Neurotrophic factors were removed from the culture medium containing mature MNs on day 20 to promote the appearance of ALS-related phenotypes.

The Hb9::GFP lentiviral vector was purchased from Addgene (ID: 37080). Lentivirus production was carried out by Guangzhou IGE Biotechnology Co., Ltd. For lentivirus transfection, 100,000 OLIG2⁺ MNPs were seeded in the presence of 10 μg/mL polybrene per confocal dish and incubated with 1 mol virus for 8 h. Cells were washed twice with culture medium, incubated for 48 h, and then allowed to differentiate into HB9⁺ MNs. Green fluorescence protein began to appear after 10–12 days. On day 28, cells were stained with 1× TMRE (Beyotime, C2001S) for 15 min, washed twice with medium, and analyzed by visual inspection for GFP⁺ expression via confocal microscopy. ATP content was estimated in 2 × 10⁴ MNPs in 96-well plates at day 28 of maturation using CellTiter-Glo assay (Promega G7570). Alternatively, ATP content was evaluated on day 7 of ISO treatment. The CellTiter-Glo assay was performed according to the manufacturer's instructions. Swelling of neurites was analyzed using MAP2 immunostaining after 28 days of differentiation with or without 7 days of exposure to ISO. Swollen neurites were evaluated using confocal microscopy under a ×40 objective.

Spinal cord MNs were examined in 20-μm tissue after embedding with OCT. Sections were blocked with 10% donkey serum in 0.3% Triton 100X for 60 min and then incubated with primary antibody in blocking buffer at 4 °C overnight. iPSC-derived motor neurons were grown in 35-mm confocal glass-bottom dishes (NEST Biotechnology), fixed at room temperature for 10 min in 4% paraformaldehyde (PFA), and then incubated in blocking buffer (1% BSA, 0.1% Triton X-100 in PBS) for 1 h. After blocking, cells or sections were incubated overnight at 4 °C with the primary antibody, and washed three times in PBS at room temperature for 10 min. After overnight incubation at 4 °C or incubation at room temperature for 60 min with the fluorochrome-conjugated secondary antibody (Thermo Fisher Scientific), cells or sections were washed three times in PBS for 10 min at room temperature. Cell nuclei were stained with 4',6-diamidino-2-phenylindole (DAPI; 1:2000) and visualized using confocal laser scanning microscopy and ImageJ software (National Institute of Health, Bethesda, MD, USA).

For the SDS-PAGE and Phos-tag SDS-PAGE, the cell culture medium was removed and rinsed with DPBS (1×) twice. A cell mitochondria isolation kit (MCE, HY-K1060) was used to isolate mitochondria within cells. Then Laemmli SDS buffer (62.5 mM Tris-HCL, pH 6.8, 25% glycerol, 2% SDS) containing 1x protease and phosphatase inhibitor cocktail was added to the cells. The phos-tag SDS-PAGE (FUJIFILM Wako, 195-17991) experiment was performed as previously reported (Sugiyama and Uezato, 2022). The native PAGE procedure was similar to that reported previously (Wittig et al, 2006). Briefly, solubilization buffer A (50 mM Sodium chloride, 50 mM Imidazole/HCl, 2 mM 6-Aminohexanoic acid, 1 mM EDTA, pH 7.0 at 4 °C) was added to cells, and samples were adjusted to 50% glycerol and 5% Coomassie blue G-250. For mammalian mitochondria, Triton X-100 was used to control the ratio at 20%. Non-denaturing gel electrophoresis kit and marker were from Real-Times (Beijing) Biotechnology Co., Ltd (Real-Times, RTD6139-0312 and RTD6137). Total protein was quantified using a BCA protein assay kit (Thermo Fisher Scientific, 07920). After electrophoresis, proteins were transferred to 0.2 mm polyvinylidene fluoride (PVDF) membrane and blocked with 5% non-fat milk or 5% BSA in 1× tris-buffered saline-Tween 20 (TBST) for 2 h. The membranes were incubated overnight at 4 °C with the primary antibody. Secondary antibodies were added to the incubation/blotting buffer and incubated with membranes for 1 h at room temperature. The blots were visualized using an enhanced chemiluminescence (ECL) kit (Millipore, WBKLS0500) and quantified using Bio-Rad Image Lab 5.2.1 software.

YFP-Parkin-mt-mKeima HeLa cells were plated in 96-well plates at approximately 8,000 cells per well. After incubating at 37 °C overnight, the medium was replaced with medium containing the test compound at 10 μM. Test compounds were selected from three small-molecule libraries (L4150, L4000, and L6000 purchased from Topscience) containing 9555 compounds. Cells were incubated for 24 h, and then four randomly selected fields per sample were imaged to measure mKeima expression (Xie et al, 2022) using a PE high content imaging system and FPbase. The mitophagy indices were quantified using ImageJ software (i.e., red pixels divided by total green plus red pixels. Data were normalized to control sample treated with 5 μM CCCP for 4 h). Candidate mitophagy inducers had a mitophagy index >1.

Sample preparation for electron microscopy was as described previously (Ruffoli et al, 2015). Briefly, mice were anesthetized and perfused through the left ventricle with 0.9% normal saline and 5% glutaraldehyde/4% paraformaldehyde in PBS. The lumbar spinal cord was removed quickly and post fixed in 5% glutaraldehyde at 4 °C overnight. MNs were incubated with 5% glutaraldehyde for 30 min at room temperature, followed by 5% glutaraldehyde overnight. Samples were washed with PBS and fixed in 1% OsO₄ at 4 °C for 2 h, dehydrated in an alcohol gradient series from 30 to 100%, embedded in resin, and cut into ultrathin transverse sections. Sections were stained with lead citrate and uranyl acetate, visualized, and scanned by transmission electron microscopy. Mitochondrial parameters were quantified as described previously (Fang et al, 2019). The percentage of damaged mitochondria and the number of mitophagy-like events were estimated.

For the measurement of mtDNA content, we used a human mitochondrial DNA (mtDNA) monitoring primer set (Takara, 7246). Briefly, the cellular DNA was first extracted from the treated HeLa cells using the DNeasy blood and tissue kit (Qiagen, 69506). Then, according to the instructions, the mitochondrial DNA ND1 gene and the ND5 gene, or the nuclear DNA SLCO2B1 gene and the SERPINA1 gene were utilized as mtDNA or nDNA markers, respectively, and their levels were quantified by a CFX96 real-time PCR detection system (Bio-Rad) with Hieff® qPCR SYBR Green Master Mix (Yeasen, 11202ES08). For the measurement of PINK1 mRNA level, Hieff® qPCR SYBR Green Master Mix was utilized for qPCR.

OCR measurements were achieved using an Agilent Seahorse XF Cell mitochondrial stress test kit. Cells were plated with 50,000 cells per XF24 well to ensure ~90% surface coverage at the time of the experiment the next day. For OCR analysis, medium was exchanged for mitochondrial stress medium (Seahorse XF DMEM Medium (Agilent, 103575-100) supplemented with 1 mM pyruvate, 2 mM glutamine and 10 mM glucose) at 1 h before the assay. The cultures were similarly equilibrated in a non-CO₂ incubator. Substrates and selective inhibitors (Agilent, 103015-100) were injected to achieve final concentrations of oligomycin at 1.5 μM, carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) at 1.0 μM, and rotenone/antimycin A at 0.5 μM, according to the manufacturer's instructions. The samples in each well were quantified using BCA assay, and then the data were normalized using the Wave software.

Cell viability was determined using Cell Counting Kit 8 (CCK8). Briefly, approximately 10,000 cells were seeded per well in a 96-well microtiter plate, pre-incubated for 24 h, and then incubated with or without drug for 24 h. Viable cells were estimated by adding 10 µL CCK8 solution per well, incubating for 1 h at 37 °C, and measuring absorbance at 450 nm.

Mitophagy index was assessed in mKeima-expressing cells as described previously (Yi et al, 2024). Briefly, mt-mKeima HeLa cells were harvested with trypsin and analyzed by flow cytometry (BD LSRFortessa X-20) equipped with a 405-nm and 561-nm laser. Lysosomal mito-Keima indicated that mitophagy index was detected using dual-excitation ratiometric pH measurements at 405 nm (pH 7) and 561 nm (pH 4) lasers with 695 nm and 670 nm emission filters, respectively. A minimum of 10,000 events was counted and analyzed using Diva software (BD Biosciences). Alternatively, to assess mitochondrial membrane potential, cells were stained in 1× TMRE for 15 min, washed twice, harvested with trypsin, and analyzed using flow cytometry. A minimum of 10,000 events was counted. Data were analyzed using FlowJo v10.0.

The galactose/glucose assay was performed as previously reported (Hertz et al, 2024). YFP-Parkin HeLa cells were isolated by trypsinization, washed three times in DMEM with glucose (MCE, HY-N0210) and 10% FBS or DMEM without glucose (Gibco 11966025) containing 10% FBS, 1 mM pyruvate (Gibco, 11360070), seeded in 96-well plates (Shanghai Jingan Biotechnology Co., LTD, J09604) at 8000 cells/well, and incubated for 24 h. Cells were then incubated in medium containing a test compound (10 μM or 5 μM) or CCCP (positive control) for 20 h, rinsed with 50 μL glucose-containing medium per well, and assessed for ATP production/cell viability using the CellTiter-Glo assay (Promega G7570) according to the manufacturer's instructions. Cell survival was normalized to the survival of cells treated with vehicle (DMSO), and the ratio of cell survival in the presence of galactose vs glucose was calculated. Drugs were considered toxic when they caused a 10–15% reduction in the production of ATP in galactose vs glucose-containing medium.

YFP-Parkin HeLa cells were plated in 96-well plates at ~8000 cells per well and analyzed using high content imaging at 37 °C with 5% CO₂. Images were processed and analyzed using ImageJ software.

For cells, ISO or an equal volume of DMSO or PRT was added to the culture medium, and then the cells were harvested. For C. elegans, worms were cultivated on NGM plates containing ISO or solvent control (DMSO). For mice, 3 mg/kg ISO was administered by intraperitoneal injection, or nano-ISO was administered by intranasal delivery (Liu et al, 2020). In the latter case, mice were anesthetized with 1.25% tribromoethyl alcohol to sustain anesthesia, a micro syringe was inserted about 8 mm into the nostril, and a sustained release pump was used to deliver 50 μl at 5 μl/min once every three days over 60 days. After each treatment, mice were placed on a thermal pad for 10–20 min. Normal saline was used for the vehicle group, and Riluzole (5 mg/kg in drinking water) was used as a positive control. For iPSC-derived MNs, because the culture medium at stage 5 was half-changed every other day, a final concentration of 1 μM ISO or 4 μM PRT or equal volume of DMSO was added to the remaining half of the culture medium to achieve the final working concentration.

Fecundity (3 h egg-laying), egg hatching, and larval development were assessed in N2 C. elegans as described previously (Xie et al, 2022). Briefly, synchronized eggs were placed on NGM plates seeded with Escherichia coli (OP50, Caenorhabditis Genetics Center). L4 larvae were transferred onto fresh OP50-seeded NGM plates and allowed to grow to adulthood on day 1. Ten adult day 1 worms were transferred onto NGM plates with OP50 without or with ISO (5, 15, 75, 150 μM) or vehicle control (DMSO). Five plates of worms were prepared and analyzed for each group. The gravid worms were incubated for 3 h, removed from the plates, and the number of eggs per plate was counted. The following day, the number of unhatched eggs and L1 larvae was counted. Larval development to stage L4 was assessed after 36 h. L4 larvae were incubated for 16 h, and the number of viable adult worms was assessed.

Worms were treated with ISO or vehicle from egg hatching. For swimming experiments, the method was the same as previously reported (Vaccaro et al, 2012). Briefly, 1-day-old adult worms were transferred to 20 μl M9 buffer, preincubated for one min, after which swimming motion was recorded for 30 s using a stereomicroscope and the number of body swings counted. If a worm moved its nose but not its body in response to physical contact, the animal was considered paralyzed (Li et al, 2014). To measure viability, transgenic worms were cultured in the NGM plates with 15 μM ISO or vehicle (DMSO), and viable animals were counted daily starting on adult day 1. Worms were considered dead if they did not respond to external stimuli. Pharyngeal pumping rate was evaluated manually every 2 days from day 2 to day 14. To quantify MN loss, animals were immobilized in M9 with 5 mM sodium azide and mounted on slides with 2% agarose pads. MNs were visualized and assessed for degeneration with confocal microscopy as described previously (Vaccaro et al, 2012).

C. elegans were treated with RNAi as previously reported (Fang et al, 2019). Briefly, worms were co-cultured with bacteria engineered to facilitate knockdown of pink-1 or pdr-1 in C. elegans. Engineered bacteria were grown in LB media in the presence of 50 mg/mL ampicillin and added to NGM plates containing 1% ampicillin and 1% IPTG with/without ISO. Animals were synchronized by bleaching or egg laying and grown on RNAi plates until adult day 1. Assessments of swimming capacity and neuronal mitophagy were performed on adult day 1.

1 mg of ISO and 3 mg of Pluronic F-127 were separately dissolved in 100 μl of DMSO, thoroughly mixed, added to 2 mL ultra-pure water, exposed to ultrasound at room temperature for 30 min, and filtered using a 100 kDa ultrafiltration tube. Nanoparticles were characterized using a particle size analyzer (Malvern), transmission electron microscopy (TEM, JOEL), high-performance liquid chromatography (HPLC), and, in the brains of anesthetized mice, by fluorescence using an animal in vivo optical imaging system (IVIS, Tanon ABL X6).

Quantitative methods were previously reported (Zhao et al, 2019). In brief, plasma and brain samples were collected from mice. Plasma was centrifuged at 3000 rpm for 10 min to obtain serum samples. Brain tissue was homogenized with triploid saline, centrifuged at 12,500 rpm for 10 min, and the supernatant was extracted with three volumes of cold acetonitrile. The supernatant was analyzed using an ACQUITY I-Class UPLC system connected to a Xevo TQD Mass Spectrometry (Waters Corp., USA). Samples were analyzed on a Hypersil GOLD column (50 × 2.1 mm, 1.9 μm, Thermo Fisher) in water:acetonitrile (35:65, v/v). Multiple reaction monitoring (MRM) transition for iso was m/z 567.345 > 135.062 generated by IntelliStart optimization program. The parameters of mass spectrometry were: desolvation temperature of 350 °C, capillary voltage of 3.0 kV, tune cone voltage of 80 V, tune collision energy of 48 V. Data were recorded and analyzed using Masslynx 4.2 software (Waters).

Five litters of SOD1-positive female mice were randomly assigned to different experimental groups (n = 12 in the Nano-ISO group, n = 12 in the vehicle group, and n = 9 in the Riluzole group). SOD1-negative mice were used as the wild-type for unified sampling and pathological evaluation after 60 days of drug treatment. ChAT, GFAP, Iba-1, and other mitophagy markers (MTCO1/LAMP2/ChAT/DAPI and pSer65-Ub/ChAT/DAPI) were assessed using standard immunofluorescence methods and ImageJ software (NIH, USA).

Assessments of neurological functions used a six-point scoring system (0–5), as described previously (Obrador et al, 2021). Mice were tested daily and animals with score of 5 (highest score) were euthanized for ethical reasons. The motor function of mice was assessed using the hanging wire test, which measures latency to fall from a suspended wire. The maximum latency was 180 s. Latency was the average of three attempts per mouse, measured and recorded weekly. Electrophysiological assessment (described previously (Pollari et al, 2018)) calculated the compound motor action potential (CMAP) of the gastrocnemius muscle after stimulation of the sciatic nerve. Mice were anesthetized with 1.25% tribromoethyl alcohol to sustain anesthesia. Stimulating electrodes were placed subcutaneously on each side of the sciatic notch, separated by ~2 cm. The recording electrode was placed subcutaneously, aligned with the gastrocnemius muscle. The reference electrode was embedded subcutaneously 2 to 5 mm under the skin next to the Achilles tendon at a 30-degree angle. The ground electrode was placed subcutaneously in a similar manner as the stimulating electrodes. The stimulus intensity started at 1.0 V and gradually increased to a maximum value. The CMAP amplitude at this time was recorded. The EMG equipment was from iWorx (iWorx Systems Inc.).

Animal/samples (mice) were assigned randomly to the Veh or the Nano-ISO groups, and mice were randomly selected for the behavioral experiments. In data collection and analysis (for example, mouse behavioral studies, mouse neurological functions assessment, mouse imaging data analysis, as well as imaging and data analysis of electron microscopy), the researcher(s) was (were) blinded as to the experimental design.

MNPs were seeded at 10,000 cells per well in a 96-well plate at stage 5. Culture medium was replaced with 200 μl of fresh media on day 28 after 7 days of ISO treatment. Culture medium was then collected for quantification of LDH leakage using CyQUANTTM LDH Cytotoxicity Assay (Invitrogen) following the manufacturer's instructions. Levels of leaked LDH were recorded, normalized, and quantified based on the manufacturer's instructions.

The experimental procedures used for lentivirus transfection of MNs were described previously. In brief, 50,0000 MNPs were seeded into six-well plates and transfected with Hb9::GFP lentivirus. MN survival was assessed using three independent biological replicates. High content imaging was performed 8 days after transfection, 15 days after neurotrophic factor withdrawal, and after 28 and 35 days of maturation. ImageJ was used to detect Hb9::GFP-labeled MNs. ISO or DMSO was administered as described above.

Mitophagy was detected in multiple model organisms using several methods as reported previously (Fang et al, 2019; Takahashi et al, 2019; Yang et al, 2025). Postmortem spinal cord samples of ALS patients and sex-matched healthy controls were obtained from the Netherlands Brain Bank (https://www.e-nbb.org↗, n = 3 per group), and the mitochondrial protein marker MTCO2 and the lysosomal protein marker LAMP2 were detected using immunofluorescence. The Pearson's correlation coefficient of co-localization between MTCO2 and LAMP2 was calculated in NeuN-positive neurons using ImageJ software (http://punias.free.fr/ImageJ/colocalization-finder.html↗) to assess the mitophagy events. In iPSC-derived MNs models, two methods were employed to assess the level of mitophagy. The first method involves using immunofluorescence to mark the MAP2-positive motor neurons and then analyzing the Pearson's correlation coefficient of MTCO2 and LAMP2 proteins to assess the mitophagy events. The second method involves the use of a commercial mitophagy detection kit (Dojindo; MD01). Mtphagy dye is a mitochondrion-localizing pH-sensitive probe. When mitophagy occurs, damaged mitochondria fuse with lysosomes, causing the mtphagy dye to produce a stronger fluorescence. Mitophagy events were evaluated by analyzing the average fluorescence intensity of mtphagy dye in Hb9::GFP lentivirus-labeled motor neurons. Neuronal mitophagy events were detected in mice using immunofluorescence to mark the ChAT-positive motor neurons and then analyzing the Pearson's correlation coefficient of MTCO2 and LAMP2 proteins to assess the mitophagy events. Neuronal mitophagy in C. elegans was measured using two strains. First, an mt-Rosella transgenic worm expressing a pan-neuronal mt-Rosella biosensor that combines a GFP variant sensitive to the acidic environment of the lysosomal lumen, fused to the pH-insensitive DsRed. Mitophagy was calculated as DsRed/DsRed+GFP, thus the higher the ratio of pixel intensity, the higher the level of mitophagy. Second, we generated transgenic nematodes expressing the DCT-1 mitophagy receptor fused with GFP or BFP, together with the autophagosome marker LGG-1 fused with DsRed in D-type motor neurons or pan-neurons. Mitophagy events were assessed using the co-localization puncta number between LGG-1 and DCT-1.

Graphics including synopsis, Figs. 7A, EV6C, and EV7K were created with BioRender.com↗.

All values are presented as mean ± SD, unless otherwise specified. Unpaired t test was used for comparison between two independent samples. One-way analysis of variance (ANOVA) was used for comparison among multiple groups, followed by Dunnett's multiple comparisons test. Group differences were analyzed with two-way ANOVA followed by Tukey's multiple comparisons test. Survival curves were produced and compared using the Log-rank (Mantel–Cox) test. Prism 9.5 (GraphPad Software) was used for statistical analysis. The significance level is indicated in all statistical analyses. P values < 0.05 were considered statistically significant.