Restoration of lysosomal membrane integrity in cell models of Pompe disease depends on fatty acid synthase and its product palmitic acid

Abstract

Background

Lysosomal storage disorders (LSDs) are rare inherited diseases caused by mutations in genes encoding lysosomal proteins, leading to substrate accumulation and lysosomal dysfunction. Whether this build-up compromises lysosomal membrane integrity remains an open question. In this study, we aimed to investigate the presence of lysosomal membrane damage in LSD cells and to identify mechanisms that could reveal potential strategies for improvement of lysosomal function.

Methods

Primary skin fibroblasts from patients with LSDs were used to evaluate lysosomal integrity and function. Lysosomal immunopurification enabled lipidomic and proteomic profiling, complemented by viral transduction for targeted protein expression at lysosomal membranes. Pharmacologic and metabolite treatments were applied in loss-of-function and rescue analyses. Microscopy-based approaches were used to assess lysosomal activity, membrane damage, and integrity. As a complementary model, we employed conditional mouse embryonic fibroblasts (MEFs) lacking class 3 phosphoinositide 3-kinase (PI3K-3), which exhibit lysosomal dysfunction and autophagy defects characteristic of LSDs.

Results

In Pompe disease, caused by mutations in the acid alpha-glucosidase gene, glycogen accumulation aggravates persistent lysosomal membrane damage. Moreover, using loss- and gain-of-function models of PI3K-3, we showed that its product, phosphoinositide-3-phosphate (PI3P), was essential for preserving lysosomal membrane integrity. Notably, Pompe patient fibroblasts exhibited chronic lysosomal damage, marked by increased galectin-3 (GAL-3) recruitment and reduced levels of lysosome-associated PI3K-3. Targeted restoration of PI3K-3 at lysosomal membranes rescued this defect and stabilized membrane integrity. Proteomic analyses following genetic or pharmacologic inhibition of PI3K-3 revealed a striking downregulation of fatty acid metabolism, including depletion of fatty acid synthase (FASN). We further demonstrated that PI3K-3 and FASN functionally interacted to maintain lysosomal integrity in Pompe cells. Supplementation with the FASN product palmitate reproduced the protective effect of PI3K-3 targeting, enhancing lysosomal proteolysis and restoring autophagic flux. Importantly, palmitate treatment also improved lysosomal function in fibroblasts from patients with Gaucher disease and type 2 neuronal ceroid lipofuscinosis, extending the relevance of this mechanism across LSDs.

Conclusions

These findings in cell models of LSDs uncover a link between PI3K-3, fatty acid metabolism, and lysosomal membrane integrity. They provide a basis for future in vivo validation of PI3K-3 activation and metabolic modulation as adjuvant treatments to improve lysosomal function in LSDs.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s11658-026-00897-w.

Background

Lysosomes are ubiquitous acidic single membrane organelles that degrade macromolecules and function as nutrient sensing signaling hubs [1]. Their physiological importance is highlighted by over 60 reported lysosomal storage disorders (LSDs) that arise from inherited mutations in genes encoding lysosomal proteins [2]. While the frequency of LSDs ranges from 1:40,000 to 1:4,000,000 for individual diseases, collectively, they affect approximately 1 in 7700 live births [3]. Although a specific enzyme replacement therapy (ERT) is available in 11 LSDs, most of them remain incurable, often causing early death [2]. Thus, better understanding of lysosomal dysfunction is a key to uncovering therapeutic targets for treating multiple LSDs. Among LSDs, Pompe disease stands out as one of the most frequent (1:40,000) and best studied in terms of its clinical manifestations, for which ERT is available [4]. It is caused by loss-of-function mutations in the lysosomal enzyme acid alpha-glucosidase (GAA) gene, resulting in intracellular glycogen build-up. Glycogen accumulation in Pompe disease is mostly damaging in skeletal, cardiac, and diaphragm muscles owing to impaired lysosomal degradation by autophagy [5]. To this end, in severe infantile-onset Pompe disease, low GAA activity leads to rapid decline, causing cardiomyopathy, hypotonia, and respiratory failure, while in late-onset Pompe disease, patients experience progressive skeletal muscle atrophy and respiratory decline [6]. Left untreated, Pompe disease leads to severe morbidity and the premature death of patients. While ERT with recombinant GAA is often life-saving for infantile-onset Pompe disease, it has major limitations, such as incomplete glycogen clearance and immune response, ineffectiveness in the central nervous system, and reduced efficacy in advanced disease [7]. Altogether, there is a need to identify targets and treatments in Pompe disease that could improve lysosomal function and enhance ERT efficacy.

Notably, substrate accumulation in several LSDs has been associated with lysosomal membrane damage [8]. Such damage has been reported in skin fibroblasts from patients with Niemann–Pick A (NPA) disease [9, 10], and lysosomal membrane permeabilization was shown to promote cytosolic aggregates in neurons in a mouse model of neuronal ceroid lipofuscinosis type 2 (CLN2) [11]. Increased sensitivity to lysosomal membrane damage in response to lysosomotropic compounds has also been observed in fibroblasts from patients with Niemann–Pick C and Gaucher disease [12, 13]. Collectively, these findings implicate lysosomal membrane damage as a feature of LSDs; however, it has not been assessed in Pompe disease or examined as a therapeutic target.

Here, we show that Pompe disease patient fibroblasts exhibit lysosomal membrane damage, which was worsened by acute substrate overload. This phenotype was accompanied by reduced lysosome-associated class 3 phosphoinositide-3-kinase (PI3K-3), a lipid kinase complex essential for lysosomal function [14], consistent with the report that PI3K-3 inactivation in mouse muscle results in glycogen accumulation and Pompe-like myopathy [15]. Mechanistically, targeting PI3K-3 to lysosomes improved their membrane integrity, while proteomic analysis of immunopurified lysosomes from PI3K-3-depleted cells showed downregulation of fatty acid metabolism enzymes, including fatty acid synthase (FASN). Functionally, FASN supports lysosomal membrane integrity, and supplementation with its product palmitate enhanced lysosomal activity and autophagic flux. Similarly, palmitate improved lysosomal readouts in Gaucher and CLN2 fibroblasts, suggesting that targeting fatty acid metabolism may be relevant across multiple LSD cell models.

Methods

Reagents

The following reagents were used: Leu-Leu methyl ester hydrobromide (LLOMe) (Sigma-Aldrich, L7393), TVB-3664 (MedChemExpress, HY-120062), Vps34-IN1 (Selleckchem, S7980), chloroquine diphosphate (Sigma-Aldrich, C6628), Bovine Serum Albumin Fraction V heat shock fatty acid free (Roche, 03117057001), Bovine Serum Albumin (BSA)-Palmitate Saturated Fatty Acid Complex 5 mM (Bertin Technology, 29558), Pierce anti-HA magnetic beads (Thermo Scientific, 88837), ChromoTek V5-Trap Magnetic Agarose kit (Protein Tech, v5tmak), Magic Red Cathepsin B Assay Kit (Abcam, ab270772), and LysoTracker Red DND-99 (Thermo Scientific, L7528). The adenovirus construct of Lyso-VPS15 (TMEM192-V5-VPS15) and lentivirus construct of glycogen synthase 1 (GYS1)-Flag were custom made by Vectorbuilder and amplified by the Viral Vectors and Gene Transfer-SFR Necker facility. pLJC5-Tmem192-3xHA was purchased from Addgene (no. 102930, David Sabatini Lab). The adenoviral vectors expressing green fluorescent protein (GFP) and GFP-Cre were as described previously [15]. All other reagents, including primary and secondary antibodies, are detailed in Supplementary Materials and Methods.

Cell culture

The generation of immortalized Vps15^f/f mouse embryonic fibroblasts (MEFs) was as previously reported [15]. The patient-derived fibroblasts were generated from skin biopsies collected under approved institutional ethical guidelines with informed consent. The patient-derived cell lines were sourced from the CRB-SFR Necker biobank under the approved project CPP-CAVB (19.11.04.44624), from the Cochin Hospital Center for Biological Resources (ISO 20387-certified) under MTA and from the Biobank of the Laboratory of Human Genetics—IRCCS Instituto Gaslini, member of the Network Telethon of Genetic Biobanks (project no. GTB18001) [16] funded by Telethon Italy, and by EuroBioBank Network. The following cell lines of patient primary skin fibroblasts were used: Pompe disease (n = 5, adult-forms), Tay-Sachs disease (n = 3, infantile-forms), Gaucher disease (n = 2, adult-forms), CLN2 (n = 1, infantile-form), and unaffected controls (n = 4 lines). Additional information about the provenance of MEFs and patient-derived fibroblasts is provided in Supplementary Material and Methods. MEFs and human fibroblasts were grown in high-glucose DMEM (Gibco) containing 100 U/mL of penicillin, 100 mg/mL streptomycin, and 10% fetal bovine serum, and were used for experiments below passage ten. All cultures were tested bi-weekly as mycoplasma-free. To obtain Vps15^−/− MEFs, cells were transduced with CRE-GFP or GFP (control) adenoviruses with a multiplicity of infection (MOI) of 500, and incubated for 5 days before analyses (unless indicated differently). For Lyso-VPS15 overexpression, cells were transduced with TMEM192-V5-VPS15 or GFP (control) adenoviruses and incubated for 2 days before analyses. For the TMEM192-3xHA and GYS1-Flag lentiviral infections, cells were treated with lentivirus-containing media in presence of 8 μg/mL polybrene and transduced cells were selected by addition of 4 µg/µL puromycin or blasticidin. Cells were treated with 5 µM Vps34-IN1 (or dimethyl sulfoxide [DMSO] as control) for 24 h, with 1 mM LLOMe (or medium as control) for 1 h, with 100 nM TVB-3664 (or DMSO as control) for 24 h, or with 50 µM chloroquine during 1 h. Human fibroblasts were treated with 200 µM BSA–palmitate, BSA–oleate, or BSA–linoleate (or fatty-acid-free BSA) during 24 h. For transcriptomics analyses, total RNA was purified with the Qiagen RNeasy Mini Kit and the downstream procedures (RT-qPCR, RNA-seq) are detailed in Supplementary Material and Methods.

Protein extraction, immunoprecipitation and immunoblotting

For total protein extraction, cells were lysed in buffer containing 20 mM Tris-HCl (pH 8.0), 5% glycerol, 128 mM NaCl, 2.7 mM KCl, 1% NP-40, 20 mM NaF, 5 mM EDTA, protease inhibitors (Roche), and PhosphoStop Inhibitors (Roche). Protein concentration was measured by Bradford assay (BioRad). For V5-Trap immunoprecipitation, 1 mg of protein extract was incubated with V5-Trap^® magnetic agarose beads for 30 min at 4 °C on a rotating wheel, followed by four washes with the lysis buffer. Protein complexes and beads were boiled in 1× SDS-sample buffer for 10 min. Proteins were resolved by SDS–PAGE before transfer to a PVDF membrane (Millipore) followed by incubation with primary antibodies and subsequent incubation with HRP-linked secondary antibodies. Immobilon Western Chemiluminescent HRP Substrate (Millipore) was used for detection. The images were acquired on a ChemiDocTM Imager (BioRad).

Lysosome immunoprecipitation

The lysosome immunoprecipitation (Lyso-IP) protocol was adapted from Abu-Remaileh et al. [17]. For each Lyso-IP, 15 million cells were used and samples were processed in batches of up to four to ensure rapid isolation using pre-chilled equipment and buffers. Cells were rinsed twice with cold Dulbecco’s phosphate-buffered saline (D-PBS), scraped in 1 mL of D-PBS containing protease inhibitors (Roche) and PhosphoStop Inhibitors, and centrifuged at 1000g for 2 min at 4 °C. Pellets were resuspended in 1000 μL, and 50 μL was retained as the whole-cell fraction. The remaining suspension was gently homogenized (20 strokes, 2 mL Dounce homogenizer) then centrifuged at 1000g for 2 min at 4 °C. The organelle-containing supernatant was incubated with 100 μL of D-PBS equilibrated anti-HA magnetic beads for 10 min at 4 °C. Beads were washed four times with cold D-PBS containing inhibitors using a DynaMag Spin Magnet. For immunoblotting, lysosomes were eluted with 50 mM NaOH for 5 min on ice and neutralized with Tris-buffer. Nano liquid chromatography coupled to tandem mass spectrometry (LC–MS/MS) proteomics and lipidomics measurements in Lyso-IP eluates were performed as previously reported [18] and are detailed in Supplementary Material and Methods.

Immunofluorescence microscopy

Immunofluorescence (IF) microscopy was performed on cells seeded in Millipore EZ glass slides. Cells were fixed at room temperature (RT) with 4% paraformaldehyde (PFA) for 10 min or with 10% formalin for 20 min, permeabilized with 0.1% Triton X-100 for 20 min, and blocked with 3% goat or donkey serum in D-PBS for 20 min. Slides were incubated with primary antibodies (1:200, except 1:1000 for Gal-3 antibody) overnight at 4 °C, followed by secondary antibodies (listed in Supplementary Materials and Methods), and mounted with VECTASHIELD containing DAPI (Eurobio Scientific, H-1200). Images were acquired on an inverted microscope (Zeiss Axio Observer Z1 with Yokogawa CSU-X1 Spinning Disk) using 40× or 63× oil-immersion objectives and processed with Fiji. For quantification of lysosomal GAL-3 puncta, the lysosomal compartment was defined by LAMP2 staining, and GAL-3 puncta were detected by threshold-based particle segmentation. GAL-3⁺ LAMP2⁺ structures per cell were quantified using CellProfiler (v.4.2.5). In representative images, the LAMP2 channel is omitted for clarity, although analyses were performed on the corresponding dual-stained images. When indicated in figure legends, co-localization between two proteins was quantified by Mander’s coefficients (M1 or M2) using Fiji (JaCoP plugin).

LysoTracker and Magic Red staining

Cells were incubated with 75 nM LysoTracker Red DND-99 or MagicRed for 1 h at 37 °C. Slides were fixed with 4% PFA for 20 min at RT. Nuclei were stained with DAPI. Acquisition was performed using an inverted microscope (Zeiss Axio Observer Z1 with Yokogawa CSU-X1 Spinning Disk) using a 40× oil-immersion objective. Cell-size-normalized intensity per cell was quantified by an integrated density measure using CellProfiler (v.4.2.5).

Proximity ligation assays

Proximity ligation assays (PLA) were performed according to the manufacturer’s instructions (Duolink^® In Situ Red Starter Kit Mouse/Rabbit; Sigma, DUO92101). Cells were fixed with 4% PFA for 20 min, permeabilized with 0.1% Triton X-100 in D-PBS for 20 min at RT, and blocked for 30 min at 37 °C. Slides were incubated with the indicated primary antibody pairs (listed in Supplementary Materials and Methods) for 1 h at 37 °C, followed by PLA probe PLUS and MINUS (oligonucleotide-conjugated) for 1 h at 37 °C. Ligation and amplification were performed at 37 °C for 30 min and 90 min, respectively. Controls were processed with PLA probes in the absence of primary antibodies (“no antibody”) to assess nonspecific amplification. Images were acquired on an inverted microscope (Zeiss Axio Observer Z1 with Yokogawa CSU-X1 spinning disk) using 40× or 63× oil-immersion objectives. PLA puncta were quantified from independent fields using Fiji, in all cells (proximity between endogenous proteins) or restricted to GFP-positive cells (when at least one protein was transiently expressed). Data are reported as puncta per cell (puncta count divided by nuclei per field).

Periodic acid–Schiff staining

For glycogen content, human fibroblasts were washed with D-PBS and fixed with 4% PFA for 10 min at RT. Cells were stained with Periodic acid–Schiff (PAS) using an automated ST5020-CV5030 Leica Stainer Integrated Workstation and acquisition was performed on a Nanozoomer 2.0 HT (Hamamatsu) microscope.

Statistics and data analyses

All experiments were independently repeated as indicated in the figure legends, and data are presented as mean ± standard deviation (SD). The treatment groups in cells were attributed randomly between plates and wells. In all treatments the samples were number-coded and the investigators were blinded during extract preparations for molecular analyses (e.g., protein expression, immunoprecipitation, Lyso-IP). Depending on the analysis, the groups were revealed either at the stage of sample quantification and statistical analysis or when it was essential for analysis (e.g., immunoblot of treatments to include all conditions in the same gel panel, immunoprecipitation to set-up control and experimental groups with antibodies or immunofluorescence microscopy staining). Normality was assessed in Prism using Q-Q plots and the Shapiro–Wilk test. Where normality was met, one-way or two-way analysis of variance (ANOVA) with post hoc Tukey test or unpaired two-tailed Student’s t-tests were applied as specified in figure legends. Where normality was not met (consistent with the near-zero control signal and treatment-induced response), nonparametric testing was applied as appropriate (Mann–Whitney for two groups; Kruskal–Wallis with appropriate multiple-comparison correction by Dunn’s test for > 2 groups). Comparisons were considered statistically significant if the p-value was < 0.05. The statistical analyses were performed with GraphPrism (v.10.4.1).

Results

Increased sensitivity of Pompe patient fibroblasts to lysosomal membrane damage

To test the presence of lysosomal membrane damage in Pompe disease, we used the Galectin-3 (GAL-3) puncta assay in patient primary skin fibroblasts using immunofluorescence microscopy [19]. GAL-3 binds beta-galactoside residues and, given an extensive glycosylation of inner lysosomal membrane proteins, lysosomal membrane damage renders these binding sites accessible for its recruitment. Therefore, the presence of GAL-3 puncta on lysosomes is indicative of compromised lysosomal membrane integrity [19]. Lysosomal membrane damage was quantified by assessing the co-localization of GAL-3 with the lysosomal membrane marker LAMP2 and counting GAL-3-positive, LAMP2-positive puncta per cell (hereafter referred to as GAL-3⁺ LAMP2⁺ structures; Fig. 1A). Notably, compared with a cell line of a healthy control, fibroblasts of five Pompe disease patients showed a marked presence of GAL-3⁺ LAMP2⁺ structures in untreated basal growing conditions (Fig. 1B). Next, we asked whether Pompe patient cells are more sensitive to chemically induced lysosomal membrane damage. To test this, we used l-Leucyl-l-leucine-O-methyl ester (LLOMe) that induces membrane perforation predominantly in late-endosome and lysosome membranes [20]. Consistently, LLOMe treatment induced the appearance of GAL-3⁺ LAMP2⁺ structures in fibroblasts of control subjects (Fig. 1B). Similarly, their number in Pompe patient cells was further increased by LLOMe (Fig. 1B). Moreover, the number of GAL-3⁺ LAMP2⁺ structures in Pompe patient cells was higher compared with the controls, suggesting their increased sensitivity to chemically induced lysosomal membrane damage (Fig. 1B). At the population level, LLOMe increased the fraction of healthy control cells positive for GAL-3⁺ LAMP2⁺ structures (Supplementary Fig. S1A). In contrast, most Pompe patient cells were already positive at baseline, and LLOMe further increased the proportion of GAL-3⁺ LAMP2⁺ cells (Supplementary Fig. S1A). This was corroborated by quantifying the fraction of LAMP2⁺ lysosomes that were GAL-3⁺ under the same conditions (Supplementary Fig. S1B), confirming the conclusions in Fig. 1B. We used the number of GAL-3⁺ LAMP2⁺ structures per cell as the most appropriate metric of lysosomal membrane damage throughout the manuscript. To test whether glycogen accumulation exacerbates damage in Pompe fibroblasts, we induced glycogen synthesis by overexpressing human glycogen synthase 1 (GYS1) (Fig. 1C). Stable GYS1 overexpression in Pompe patient cells increased glycogen levels as shown by PAS staining (Fig. 1D), and concomitantly was accompanied by an increased number of GAL-3⁺ LAMP2⁺ structures (Fig. 1E). These data indicate that Pompe disease patient fibroblasts present lysosomal membrane damage that is aggravated by further glycogen accumulation.

Fig. 1.

Pompe disease patient cells show lysosomal membrane damage. A Schematic representation of GAL-3 puncta assay and the impact of lysosomal membrane damage (left). Scheme of quantification protocol of GAL-3⁺ LAMP2⁺ structures in fibroblasts (right). Created in BioRender. B Immunofluorescence microscopy using anti-GAL-3 and anti-LAMP2 antibodies in human skin fibroblasts of a healthy control line and Pompe disease patient lines (n = 5) under basal growth condition (Unt) or under 1-h LLOMe treatment (LAMP2 staining is not shown). DAPI stains nuclei. Scale bar: 20 µm. Quantification of co-localized GAL-3⁺ LAMP2⁺ structures per cell is presented. Kruskal–Wallis test followed by Dunn’s post-test, ***p < 0.001, n = 3. C Immunofluorescence microscopy using anti-GYS1 antibodies in a line of Pompe disease patient skin fibroblasts stably transduced with lentivirus expressing GYS1 compared with a nontransduced parental line. DAPI stains nuclei. Scale bar: 20 µm (n = 2). D Cytochemistry analysis with periodic acid–Schiff (PAS) staining for glycogen, scale bar: 50 µm. E Immunofluorescence microscopy using anti-GAL-3 and anti-LAMP2 antibodies in skin fibroblasts of Pompe disease patient cells overexpressing GYS1 compared with a nontransduced parental line (LAMP2 staining is not shown). DAPI stains nuclei. Scale bar: 20 µm. Quantification of co-localized GAL-3⁺ LAMP2⁺ structures per cell is presented. Unpaired Mann–Whitney test, ***p < 0.001, n = 3

Inhibition of PI3K-3 mimics lysosomal membrane damage in Pompe fibroblasts

To identify the pathways implicated in persistent membrane damage in Pompe patient cells, we explored the involvement of class 3 PI3K (PI3K-3) for two main reasons. First, inactivation of PI3K-3 in muscle cells mimics Pompe disease phenotypes, such as glycogen accumulation, muscle damage, and weakness [15]. Second, in mouse embryonic fibroblasts (MEFs), PI3K-3 deletion leads to accumulation of dysfunctional lysosomes and blocks autophagic flux [15]. Importantly, these phenotypes are reminiscent of LSDs including manifestations in Pompe disease. Therefore, we asked if PI3K-3 inactivation in MEFs also induces lysosomal membrane damage as seen in Pompe patient cells. PI3K-3 functions as a complex of the catalytic subunit with the lipid kinase activity, vacuolar protein sorting 34 (Vps34), and its essential regulatory subunit, vacuolar protein sorting 15 (Vps15) protein [14]. In the PI3K-3 complex, depletion of Vps15 results in degradation of Vps34 and is accompanied by a decrease of intracellular PI3P, representing a model of PI3K-3 inactivation [15]. To acutely deplete Vps15, we transduced Vps15^f/f MEFs with CRE-expressing adenoviral vectors. As expected, Vps15 depletion in MEFs resulted in a decrease of Vps34 protein and was accompanied by a progressive increase of lysosomal mass and autophagy block, witnessed by Lamp2 and p62 autophagy cargo protein accumulation, respectively (Supplementary Fig. S1C). It was also accompanied by increased levels of Gal-3 protein starting from 5 days after expression of CRE recombinase (Supplementary Fig. S1C). Moreover, this genetic depletion of Vps15 induced Gal-3⁺ Lamp2⁺ structures, consistent with lysosomal membrane damage (Fig. 2A). This was accompanied by activation of lysosomal biogenesis, evidenced by upregulation of transcription factor EB (TFEB) target transcripts (Supplementary Fig. S1D), and confirmed by real-time quantitative PCR (RT-qPCR) for selected targets (Supplementary Fig. S1E). Notably, this basal lysosomal membrane damage was further worsened by LLOMe treatment, suggesting enhanced sensitivity of Vps15-depleted cells and their inability to sustain chemically induced lysosomal membrane perforation (Fig. 2A). To test the involvement of lipid kinase activity of PI3K-3, we treated MEFs for 24 h with its specific inhibitor, Vps34-IN1 [21]. Mirroring findings of PI3K-3 genetic inactivation, Vps34-IN1 treatment increased Gal-3 protein levels, promoted appearance of Gal-3⁺ Lamp2⁺ structures, and rendered cells more sensitive to LLOMe-induced damage (Fig. 2B; Supplementary Fig. S1F). Next, we asked if protein levels of PI3K-3 subunits were modified in fibroblasts of patients with Pompe disease. Immunoblotting of total cell extracts revealed no significant change in VPS15 or VPS34 protein levels (Fig. 2C). In contrast, VPS15 proximity to lysosomes was reduced in Pompe patient fibroblasts, as indicated by fewer proximity ligation puncta between endogenous VPS15 and LAMP2 (Fig. 2D). Consistent with a sub-pool of PI3K-3 localized to the lysosome, immunofluorescence analyses showed partial Vps15 co-localization with LAMP2 in MEFs at steady state (Supplementary Fig. S2A). These suggest that, in Pompe cells, reduced lysosomal association of PI3K-3, while unchanged total protein levels, may reflect its re-localization to nonlysosomal compartments. Collectively, these data support a requirement for PI3K-3 lipid kinase activity in maintaining lysosomal membrane integrity.

Fig. 2.

Class 3 PI3K inactivation renders cells sensitive to lysosomal membrane damage. A Immunofluorescence microscopy using anti-Gal-3 and anti-Lamp2 antibodies in Vps15^f/f MEFs infected with GFP or CRE expressing adenoviruses and 5-days post-infection treated with LLOMe or vehicle (Unt) during 1 h (Lamp2 staining is not shown). DAPI stains nuclei. Scale bar: 20 µm. Quantification of co-localized Gal-3⁺ Lamp2⁺ structures per cell is presented. Kruskal–Wallis test followed by Dunn’s post-test, *p < 0.05, ** < 0.01, *** < 0.001, n = 3. B Immunofluorescence microscopy using anti-Gal-3 and anti-Lamp2 antibodies in Vps15^f/f MEFs treated with Vps34-IN1 (5 μM) or vehicle (DMSO) for 24 h, followed by treatment with LLOMe or vehicle (Unt) for 1 h (Lamp2 staining is not shown). DAPI stains nuclei. Scale bar: 20 µm. Quantification of co-localized Gal-3⁺ Lamp2⁺ structures per cell is presented. Kruskal–Wallis test followed by Dunn’s post-test, *p < 0.05, *** < 0.001, n = 4. C Immunoblot analysis in two different lines of skin fibroblasts of healthy controls and Pompe disease patients using indicated antibodies. GAPDH served as a loading control. Densitometry analysis of VPS15 and VPS34 protein levels normalized to GAPDH levels is shown. Unpaired Student’s t-test; ns, nonsignificant; n = 4. D Proximity ligation assay of VPS15 and LAMP2 in two different lines of skin fibroblasts of healthy controls and Pompe disease patients by fluorescence microscopy. DAPI stains nuclei. No antibody condition served as a control of nonspecific amplification. Each color shows a different patient cell line. Scale bar: 20 μm. Quantification of the mean number of proximity puncta per cell. Unpaired Mann–Whitney test, ***p < 0.001, n = 3

Targeting of PI3K-3 to the lysosomal membrane improves lysosomal integrity

Next, given decreased proximity of PI3K-3 to lysosomes in Pompe patient cells, we hypothesized that its targeting to the lysosomal membrane could improve lysosomal membrane integrity. We generated a construct expressing the fusion of VPS15 with lysosomal transmembrane protein TMEM192, hereinafter referred to as Lyso-VPS15 (Fig. 3A). Notably, to identify Lyso-VPS15-infected cells, the vector also expressed GFP. Lyso-VPS15 was detected in the lysosomal fraction purified from MEFs expressing TMEM192-3xHA protein (Lyso-tag) that allows for rapid immunopurification of the lysosome-enriched fraction with anti-HA antibodies (Fig. 3A, B) [17]. Lyso-VPS15 localization was further validated by proximity with lysosomal membrane protein Lamp1 (Supplementary Fig. S2B). The functionality of the Lyso-VPS15 construct was further backed by pull-down of Vps34 lipid kinase subunit with immunoprecipitated Lyso-VPS15 (Supplementary Fig. S2C). Consistently, Lyso-VPS15 expression reduced the number of Gal-3⁺ Lamp2⁺ structures in mouse and in human fibroblasts treated with LLOMe (Fig. 3C; Supplementary Fig. S2D), without affecting lysosomal mass, as indicated by unchanged Lamp2 levels assayed by immunoblot and immunofluorescence in MEFs (Supplementary Fig. S2E, F). Notably, Lyso-VPS15 overexpression in Pompe patient cells, also decreased GAL-3⁺ LAMP2⁺ structures (Fig. 3D). Together, these data indicate that increasing PI3K-3 proximity to lysosomes enhances lysosomal membrane integrity in Pompe patient cells.

Fig. 3.

Expression of PI3K-3 at lysosomal membrane ameliorates lysosomal membrane integrity in Pompe disease cells. A Schematic representation of the Lyso-VPS15 (TMEM192-V5-VPS15) construct for targeting PI3K-3 to the lysosomal membrane. The Lyso-tag (TMEM192-3xHA) for lysosomal immunopurification (Lyso-IP) is indicated. Created in BioRender. B Immunoblot analysis with indicated antibodies of Lyso-IP and total (input) fractions from lysotagged Vps15^f/f MEFs. Cells were collected 2 days post-transduction with adenoviral vectors expressing Lyso-VPS15 or GFP as a control. Tubulin served as a loading control and a nonlysosomal marker; TMEM192-3xHA served as a lysosomal membrane marker. C Immunofluorescence microscopy using anti-Gal-3 and anti-Lamp2 antibodies in Vps15^f/f MEFs transduced with Lyso-VPS15-expressing adenoviral vectors or with GFP as a control during 48 h, followed by treatment with LLOMe or vehicle (Unt) for 1 h (Lamp2 staining is not shown). DAPI stains nuclei. Scale bar: 20 µm. Quantification of co-localized Gal-3⁺ Lamp2⁺ structures per cell is presented. Kruskal–Wallis test followed by Dunn’s post-test, ***p < 0.001, n = 3. D Immunofluorescence microscopy using anti-GAL-3 and anti-LAMP2 antibodies in Pompe disease patient fibroblasts 48 h after transduction with adenoviral vectors expressing Lyso-VPS15 or GFP as a control (LAMP2 staining is not shown). DAPI stains nuclei. Scale bar: 20 µm. Quantification of co-localized GAL-3⁺ LAMP2⁺ structures per cell is presented. Unpaired Mann–Whitney test, ***p < 0.001, n = 4

Fatty acid synthase protects lysosomal membrane integrity downstream of PI3K-3

To gain mechanistic insight into pathways downstream of lysosomal PI3K-3, we performed proteomic analyses of immunopurified lysosomes from control MEFs and MEFs acutely depleted of Vps15. We identified 220 proteins significantly decreased and 172 increased in lysosomes from Vps15-depleted cells (Fig. 4A; Supplementary Table S1). As expected, Vps15 was the top downregulated protein, further validating the proteomics findings (Fig. 4A). Pathway analyses of downregulated proteins using DAVID Gene Ontology (GO), KEGG, and Reactome pointed to interconnected processes of autophagy, endocytosis, endosomal sorting complexes required for transport (ESCRT), protein transport, and lysosome function, in line with the well-known involvement of PI3K-3 in these processes [22, 23]. Moreover, concordant with the functional crosstalk between PI3K-3 and mechanistic target of rapamycin complex 1 (mTORC1) [24–29], the KEGG and Reactome pathways of “mTOR signaling pathway” and “amino acid regulating mTOR” were present among downregulated proteins in Vps15-depleted MEFs. Interestingly, we found a significant enrichment of fatty acid biosynthesis, lipid transport, and lipid metabolism in GO biological process terms and in KEGG and Reactome pathways, pointing to their potential inhibition in Vps15-depleted MEFs (Fig. 4A). These were selective to the downregulated list as the upregulated proteins were associated with mitosis and nucleus-related GO biological process terms (Supplementary Fig. S3A; Supplementary Table S1). Given the reduced abundance of several fatty acid metabolism enzymes (Fasn, Elovl1, Elovl4, Acsl1, and Acsl4), we next examined whether this pathway contributes to lysosomal membrane integrity. Because these enzymes are not classically annotated as lysosomal, their apparent depletion in Lyso-IP fractions from Vps15-depleted MEFs may reflect lipid exchange and membrane remodeling at organelle contact sites (e.g., ER–lysosome or mitochondria–lysosome) and/or co-purification inherent to organelle isolation. However, fatty acid synthase (Fasn), a rate limiting enzymatic complex for de novo synthesis of saturated fatty acids [30], has also been implicated in phosphoinositide synthesis required for autophagosome–lysosome fusion, prompting us to test its involvement in our model [31]. We found that FASN protein levels were decreased in Pompe patient fibroblasts (Fig. 4B). Surprisingly, MEFs treated with Vps34-IN1 with or without LLOMe showed loss of Fasn protein in Lyso-IP fractions (Fig. 4C; Supplementary Fig. S3B). The latter was concomitant with decreased levels of lysosomal PI3P species (Fig. 4D). Given that co-treatment of Vps34-IN1 with LLOMe resulted in lysosomal membrane damage, the findings of a decreased lysosomal pool of Fasn prompted us to test functional interaction between PI3K-3 and de novo fatty acid synthesis. Notably, Fasn inhibition with TVB-3664 led to decreased lysosomal localization of Vps15, which was not further affected with LLOMe treatment (Fig. 4E). Finally, the beneficial effect of Lyso-VPS15 expression in Pompe patient cells was abolished by inhibiting either FASN or VPS34 with their respective inhibitors - TVB-3664 and Vps34-IN1 (Fig. 4F). In sum, decreased targeting to the lysosomal membrane of PI3K-3 and FASN, and their inhibition, are potential contributors to lysosomal dysfunction in Pompe disease.

Fig. 4.

FASN and PI3K-3 functionally interact in maintaining lysosomal membrane integrity in Pompe disease cells. A The left panel shows a volcano plot of changes in lysosomal proteome quantified by nanoLC-MS/MS. Lysosomes were immunoprecipitated from Vps15^f/f MEFs 5 days post-transduction with CRE adenovirus or GFP as a control (n = 4). Data are presented as log₂ fold change (FC), −log₁₀ statistical significance cutoff (p < 0.05 and p < 0.01). Significantly downregulated proteins are shown in green and significantly upregulated proteins in red. The right panel shows pathway enrichment analyses of lysosomal proteins downregulated in Lyso-IP from Vps15-depleted MEFs. B Immunoblot analysis in two lines of healthy control and Pompe disease patient skin fibroblasts under basal growth conditions, using indicated antibodies. TUBULIN served as a loading control. Densitometry analysis of FASN protein levels normalized to TUBULIN is shown. Unpaired Student’s t-test, **p < 0.01, n = 3. C Immunoblot analysis of Lyso-IP and input fractions of lysotagged Vps15^f/f MEFs using indicated antibodies. Cells were first treated for 24 h with Vps34-IN1 (5 µM) or vehicle (DMSO), and before collection, cells were incubated with LLOMe or vehicle (Unt) for 1 h. Tubulin served as a loading control and a nonlysosomal marker, TMEM192-3xHA and Lamp1 served as lysosomal membrane markers. Densitometry analysis of Fasn protein levels normalized to TMEM192-3xHA in Lyso-IP extracts is presented. ANOVA one-way test followed by Tukey’s post-test, *p < 0.05, ** < 0.01, n = 6. D Normalized PI3P acyl species in Lyso-IP extracts of Vps15^f/f MEFs treated with Vps34-IN1 (5 µM) or vehicle (DMSO) for 24 h, and before collection, cells were incubated with LLOMe or vehicle (Unt) for 1 h. PI3P acyl species were measured by LC–MS/MS and normalized to TMEM192-3xHA lysosomal protein levels. ANOVA one-way test followed by Tukey’s post-test, ***p < 0.001, n = 6. E Proximity ligation assay (PLA) of Vps15 and Lamp2 in Vps15^f/f MEFs, treated for 24 h with Fasn inhibitor TVB-3664 or vehicle (DMSO), followed by treatment with LLOMe or vehicle (Unt) for 1 h. No antibody condition served as a control of nonspecific amplification. DAPI stains nuclei. Scale bar: 20 μm. ANOVA one-way test followed by Tukey’s post-test, ***p < 0.001, n = 3. F Immunofluorescence microscopy using anti-GAL-3 and anti-LAMP2 antibodies (LAMP2 staining is not shown) in skin fibroblasts of Pompe disease patients transduced with Lyso-VPS15 or GFP adenoviral vectors, and 24-h post-transduction cells were treated for 24 h with TVB-3664, Vps34-IN1, or vehicle (DMSO). DAPI stains nuclei. Scale bar: 20 µm. Quantification of co-localized GAL-3⁺ LAMP2⁺ structures per cell is presented. Each color shows a different patient cell line. Kruskal–Wallis test followed by Dunn’s post-test, ***p < 0.001, n = 3

Exogenous palmitate improves lysosomal activity in Pompe patient cells

Because FASN activity was required for the Lyso-VPS15-mediated rescue of lysosomal membrane integrity, we asked whether modulating FASN could alleviate lysosomal damage in Pompe cells. As FASN proximity to lysosomes was unchanged in Pompe patient cells (Supplementary Fig. S3C) and Lyso-VPS15 overexpression did not alter cellular palmitate content or Fasn levels in MEFs (Supplementary Fig. S3D, E), we next tested whether exogenous fatty acids could confer protection under lysosomal stress, thereby phenocopying the effect of Lyso-VPS15. To test this, we have focused on palmitate, a saturated long-chain fatty acid that is synthesized by FASN. MEFs were pre-treated for 24 h with palmitate conjugated to BSA, followed by induction of lysosomal membrane damage with Vps34-IN1 or LLOMe. BSA–palmitate significantly reduced the number of Gal-3⁺ Lamp2⁺ structures, indicating protection against Vps34-IN1- or LLOMe-induced loss of lysosomal membrane integrity (Supplementary Fig. S4A, B). Similarly, BSA–palmitate decreased basal GAL-3⁺ LAMP2⁺ structures in Pompe patient fibroblasts (Fig. 5A) while increasing intracellular palmitate levels, as measured by LC–MS/MS (Supplementary Fig. S5A). The effect was palmitate-specific, as oleic or linoleic acid did not significantly reduce GAL-3⁺ LAMP2⁺ structures (Supplementary Fig. S5B). Next, we asked if palmitate treatment also affected lysosomal function. We assessed lysosomal function using three complementary assays: LysoTracker (acidic lysosome mass), Magic Red (lysosomal cathepsin B activity), and autophagic flux [32]. In Pompe patient fibroblasts, BSA–palmitate significantly increased the number of LysoTracker foci per cell (Fig. 5B) and increased Magic Red fluorescence intensity (Fig. 5C). In contrast, BSA–palmitate did not significantly affect LysoTracker or Magic Red staining in healthy control fibroblasts (Supplementary Fig. S5C, D). Consistent with enhanced lysosomal activity, palmitate increased autophagic flux, as indicated by higher LC3-II levels in Pompe patient fibroblasts co-treated with BSA–palmitate and chloroquine (CQ), which blocks autophagosome–lysosome fusion [32] (Fig. 5D). Notably, these effects were likely post-transcriptional as RNA sequencing (RNA-seq) of palmitate-treated Pompe patient fibroblasts showed no transcriptional induction and only two significantly downregulated transcripts (PDK4 and ANGPTL4) (Supplementary Fig. S5E). Together, these data indicate that palmitate supplementation improves lysosomal membrane integrity and enhances lysosomal function in Pompe patient cells.

Fig. 5.

Palmitate treatment ameliorates lysosomal function in Pompe disease cells. A Immunofluorescence microscopy using anti-GAL-3 and anti-LAMP2 antibody in skin fibroblasts of Pompe disease patients, treated for 24 h with BSA–palmitate (200 µM) or fatty-acid-free BSA (LAMP2 staining is not shown). DAPI stains nuclei. Scale bar: 20 µm. Quantification of co-localized GAL-3⁺ LAMP2⁺ structures per cell is presented. Each color shows a different patient cell line. Unpaired Mann–Whitney test, ***p < 0.001, n = 3. B LysoTracker staining in skin fibroblasts of Pompe disease patients (n = 3 lines), treated for 24 h with BSA–palmitate (200 µM) or fatty acid-free BSA, followed by 1 h incubation with the LysoTracker probe and acquisition by fluorescence microscopy. DAPI stains nuclei. Scale bar: 20 µm. Quantification of mean LysoTracker foci number per cell is presented. Each color shows a different patient cell line. Unpaired Mann–Whitney test, *p < 0.05, n = 3. C Magic Red (cathepsin B activity) staining in skin fibroblasts of Pompe disease patients (n = 3 lines), treated for 24 h with BSA–palmitate (200 µM) or fatty acid-free BSA, followed by 1 h incubation with Magic Red probe and acquisition by fluorescence microscopy. DAPI stains nuclei. Scale bar: 20 µm. Quantification of Magic Red intensity (normalized by cell size) per cell is presented. Each color shows a different patient cell line. Unpaired Mann–Whitney test, ***p < 0.001, n = 3. D Immunoblot analysis of LC3 protein in Pompe disease patient skin fibroblasts treated for 24 h with BSA–palmitate (200 µM) or fatty acid-free BSA, followed by 1 h incubation with chloroquine (CQ; 50 µM). ACTIN served as a loading control. Densitometry analysis of LC3-II protein levels normalized to ACTIN is shown. ANOVA two-way test followed by Tukey’s post-test; *p < 0.05, ** < 0.01, *** < 0.001; n = 5

Exogenous palmitate improves lysosomal activity in different LSD cells

To test if these findings in Pompe patient cells could be extended to other LSD models, we used skin fibroblasts from patients with Gaucher disease (glucocerebrosidase deficiency, glucosylceramide accumulation), type 2 neuronal ceroid lipofuscinosis/CLN2 disease (tripeptidyl peptidase I deficiency, ceroid lipofuscin accumulation), and Tay–Sachs disease (β-hexosaminidase A deficiency, GM2 ganglioside accumulation). Similar to Pompe disease cells, fibroblasts of patients with Gaucher disease and CLN2 disease showed a marked presence of GAL-3⁺ LAMP2⁺ structures in basal conditions that, unlike in control cells, was not further increased by LLOMe treatment (Fig. 6A). Interestingly, GAL-3 protein was similarly expressed across all LSD fibroblast lines tested (Fig. 6B). However, Tay–Sachs patient fibroblasts did not display GAL-3⁺ LAMP2⁺ structures, in contrast to Gaucher and CLN2 cells (Fig. 6A). LLOMe induced GAL-3⁺ LAMP2⁺ structures in Tay–Sachs fibroblasts, indicating an intact lysosomal membrane damage response (Fig. 6A). As in Pompe cells, BSA–palmitate increased Magic Red intensity in Gaucher and CLN2 fibroblasts (Fig. 6C), and restored LysoTracker signal in Gaucher cells (Supplementary Fig. S6). Together, these data suggest that lysosomal membrane damage is a shared feature across several LSD cell models, although not detected at baseline in Tay–Sachs fibroblasts. Palmitate supplementation improved lysosomal readouts in multiple LSDs fibroblast lines, supporting further evaluation in relevant in vivo models.

Fig. 6.

Palmitate improves lysosomal function in Gaucher disease and CLN2 patient cells. A Immunofluorescence microscopy using anti-GAL-3 and anti-LAMP2 in skin fibroblasts of healthy controls (n = 2 lines) and patients with Gaucher disease, type 2 neuronal ceroid-lipofuscinosis (CLN2), and Tay–Sachs disease under basal growth conditions (Unt) or under LLOMe treatment for 1 h (LAMP2 staining is not shown). DAPI stains nuclei. Scale bar: 20 µm. Quantification of co-localized GAL-3⁺ LAMP2⁺ structures per cell is presented. Kruskal–Wallis test followed by Dunn’s post-test, ***p < 0.001, n = 3. B Immunoblot analysis of GAL-3 protein levels in healthy controls (n = 3 lines) and in indicated LSD patient fibroblast lines grown under basal conditions. TUBULIN served as a loading control. Densitometry analysis of GAL-3 level normalized to TUBULIN is shown. Kruskal–Wallis test followed by Dunn’s post-test; ns, nonsignificant; n = 3. C Magic Red (cathepsin B activity) staining in skin fibroblasts of healthy controls (n = 2 lines) and patients of Gaucher and CLN2 diseases (n = 1 line for each disease) treated for 24 h with BSA–palmitate (200 µM) or fatty acid-free BSA, followed by 1 h incubation with Magic Red probe and acquisition by fluorescence microscopy. DAPI stains nuclei. Scale bar: 20 µm. Quantification of Magic Red intensity (normalized by cell size) per cell is presented. Kruskal–Wallis test followed by Dunn’s post-test, *p < 0.05, *** < 0.001, n = 4

Discussion

In this study, we identify a protective role for de novo fatty acid synthesis in maintaining lysosomal membrane integrity in patient fibroblasts from Pompe, Gaucher, and CLN2 diseases. Lysosomal membrane damage is increasingly implicated in LSD pathophysiology and may reinforce mitochondrial dysfunction and impaired autophagy in a feed-forward manner. Although the core drivers of Pompe disease, including GAA mutations and glycogen accumulation, are well established [33], how glycogen burden affects lysosomal membrane integrity is less clear. In GAA-mutant mice, Gal-3 levels are increased in skeletal muscle, track with autophagy defects, and are normalized by recombinant GAA enzyme replacement therapy, consistent with reversibility [34]. Along the same lines, Atg7/GAA double-knockout mice, which are impaired in glycogen delivery to lysosomes, show reduced Gal-3 levels [35]. These in vivo data align with our Pompe fibroblast findings, where GYS1 overexpression increased glycogen accumulation and exacerbated lysosomal membrane damage. Lysosomal injury has also been reported in other LSDs, as Niemann–Pick A fibroblasts with sphingomyelin accumulation exhibit GAL-3 puncta, reduced cathepsin B activity, lysosomal exocytosis, and defective autophagolysosome clearance [9, 10], and Niemann–Pick C and Gaucher fibroblasts show increased sensitivity to lysosomal damage [12, 13]. Together with our observations in Pompe, CLN2, and Gaucher fibroblasts, these studies support a model in which progressive substrate accumulation compromises lysosomal integrity once a threshold is exceeded, increasing cellular vulnerability to stress.

Our data identify PI3K-3 as a key determinant of lysosomal membrane integrity in Pompe cells, consistent with prior work linking PI3K-3 dysfunction to glycogen storage myopathies and autophagy failure. In mouse skeletal muscle, Vps15/PIK3R4 deletion causes an autophagic vacuolar myopathy (AVM)-like phenotype, whereas PI3K-3 overexpression in Danon patient myoblasts reduces glycogen accumulation [15]. In both Danon and Pompe patient muscles, PI3K-3 subunits are increased and mislocalized in association with disease progression and autophagy impairment [36]. Similarly, Pompe mouse models show elevated autophagy-related proteins, including PI3K-3 subunits, Atg7, p62, and LC3-II, with ERT partially normalizing these profiles [34]. In our Pompe fibroblasts, total PI3K-3 levels were not markedly altered, but its localization to the lysosomal membrane was reduced, supporting the idea that defective recruitment rather than reduced abundance limits PI3P production at lysosomes. This interpretation is reinforced by studies in neurons in which Vps34/PI3K-3 inhibition induces endolysosomal damage marked by increased Gal-3 and impaired LC3 turnover [37]. Together, these results suggest that impaired PI3K-3 association with lysosomes and the resulting disruption of the lysosomal PI3P pool may represent an underappreciated mechanism contributing to lysosomal vulnerability in LSDs. This concept fits with emerging models in which phosphoinositides coordinate lysosomal membrane repair by organizing endosomal sorting complexes required for transport (ESCRT) recruitment, PI4P-dependent membrane tethering, and endoplasmic reticulum (ER) contact sites that enable lipid transfer [38–40]. ESCRT-I supports repair and lysophagy after damage, while PI4K2A-driven PI4P production promotes phosphatidylserine and cholesterol transfer, processes required for survival during sterile injury and infection [39–42]. Phase separation has also been implicated in stabilizing damaged lysosomal membranes through ESCRT-dependent and independent mechanisms [43]. Future work should define how the lysosomal PI3P pool changes across disease stage and if restoring PI3K-3 lysosomal recruitment can improve membrane repair capacity through these distinct mechanisms in LSDs cells.

An unexpected finding of this study is that FASN activity is required to preserve lysosomal membrane integrity, both after chemically induced damage and in LSD patient cells. Although a role for de novo fatty acid synthesis in lysosomal membrane repair has not been examined in LSDs, prior work on exogenous fatty acids and autophagy provides insights. Oleic acid [44] and palmitic acid [45, 46] can promote autophagy, yet palmitate is also widely used to model lipotoxic “steatosis” and can inhibit autophagic flux despite increasing lysosomal activity, with outcomes that depend strongly on dose and exposure time [47]. Reported differences between saturated and unsaturated fatty acids further highlight this context dependence and likely reflect variation in cell type and the autophagy-inducing stimulus [45, 48, 49]. In our hands, exogenous palmitate improved lysosomal membrane integrity in LSD cells, raising the question of whether the source of fatty acids is functionally important. In autophagy settings, fatty acids mobilized from lipid droplets can activate autophagy [44, 50], whereas exogenous free fatty acids may require re-esterification to elicit similar effects [51]. Consistent with our observation that FASN activity supports lysosomal membrane repair, adipocyte-specific Fasn deficiency impairs autophagosome–lysosome fusion through reduced membrane phosphoinositides and is partially rescued by exogenous palmitate [31]. More broadly, lipid metabolism intersects with autophagy through incorporation of newly synthesized choline phospholipids into autophagosomal membranes [52], impaired autophagic flux upon inhibition of phosphatidic acid and phosphatidylinositol synthesis [53, 56], localization of PI synthase at autophagosome biogenesis sites [53], and the requirement of stearoyl-CoA desaturase-1 for autophagy in flies and mammals [54, 55]. On the basis of these precedents and our data, we propose that reduced de novo lipogenesis (DNL)-derived palmitate flux could limit renewal of lysosomal membrane phospholipids and/or the availability of palmitoyl-CoA for lipidation reactions that support membrane association of key regulators, thereby sensitizing cells to lysosomal membrane permeabilization. Accordingly, restoring palmitate through supplementation or synthesis may replenish these pools and improve lysosomal membrane integrity. Mechanistically, palmitate could act by supporting phospholipid synthesis and thus membrane packing and resilience, by serving as a substrate for S-palmitoylation that tunes recruitment and retention of lysosome-associated proteins and assembly of repair or trafficking complexes, or by indirectly reshaping lysosomal phosphoinositide composition and organelle contact sites that coordinate effector recruitment during repair. Finally, our results in mouse and human fibroblasts point to a bidirectional relationship between PI3K-3 and FASN, with each required for the other’s proximity to lysosomal membranes. This parallels yeast work showing that Atg14 colocalization with Atg8-positive autophagosomes depends on local acyl-CoA production by the acyl-CoA synthetase Faa1, while Faa1 recruitment in turn requires Atg proteins, including Atg14, and Faa1 preferentially binds PI3P/PI4P-containing membranes [56]. Future studies should determine whether PI3K-3 and FASN interact directly or through intermediates, and whether lysosomal PI3P promotes FASN recruitment and local fatty acid production to facilitate membrane repair, potentially in coordination with ER contact sites and PI4P-dependent pathways.

From a translational standpoint, our early preclinical findings in LSD patient fibroblasts point to fatty acid metabolism as a potential adjunct to existing therapies. In Pompe disease, the clinical benefit of ERT with recombinant GAA often declines after approximately 5 years, with persistent muscle weakness, gait impairment, and skeletal deformities [4]. So far, several adjunct strategies have been explored to enhance ERT by limiting glycogen accumulation and mitigating secondary defects in mitochondria and autophagy. In a Pompe mouse model, pharmacologic inhibition of glycogen synthesis with the GYS1 inhibitor MZ-101 reduced muscle glycogen to a degree comparable to ERT, and the combination further improved glycogen clearance toward normalization [57]. Antioxidants, including N-acetylcysteine, idebenone, resveratrol, and edaravone, also potentiated recombinant GAA in cell and mouse models [58]. Dietary approaches have likewise been tested, with ERT combined with a ketogenic diet, a ketone precursor (1,3-butanediol), or antioxidant diets producing distinct benefits. In particular, a ketogenic diet plus ERT most effectively reduced quadriceps glycogen, whereas 1,3-butanediol plus ERT best improved autophagic blockade [59]. Notably, only the combinations of 1,3-butanediol or antioxidant diet with ERT normalized elevated Gal-3 in GAA-mutant muscle, supporting the idea that restoring lysosomal homeostasis improves ERT efficacy [59]. Despite these encouraging preclinical results, ketogenic regimens are difficult to implement clinically owing to formulation constraints, adherence, and potential weight loss. Although fatty acid supplementation has not been evaluated in Pompe disease, related work supports the broader principle that modulating lipid availability can improve LSD outcomes. In a Krabbe disease model (GALC deficiency), n-3 PUFA treatment delayed onset, reduced oxidative stress, increased resolvin D1, and elevated brain n-3 PUFA levels [60]. In Niemann–Pick A cells, lysosomal membrane integrity can be improved by ApoD, an extracellular lipid transporter that targets damaged lysosomes after uptake, reduces GAL-3 puncta, restores cathepsin B activity, improves autophagic flux, and lowers oxidative stress [10, 61]. Together with our results, these studies underscore a link between lysosomal lipid metabolism and membrane integrity. We therefore propose that restoring fatty acid pools, either by supplementation or by supporting endogenous synthesis, could complement ERT by mitigating lysosomal membrane damage in Pompe disease and potentially other LSDs. As our conclusions are based on fibroblast models, an essential next step is validation in muscle-relevant systems and in vivo, combining established lysosomal damage and function assays (Gal-3 recruitment, lysosomal activity/acidification readouts, and autophagic flux) with physiological endpoints of muscle health, including contractile performance, strength and endurance, and histopathological measures of muscle integrity, under defined lipid- and diet-based interventions.

Conclusions

Our findings identify lysosomal membrane integrity as a shared vulnerability in LSDs and link it mechanistically to PI3K-3 activity and fatty acid metabolism. Across different LSD patient fibroblast lines, palmitate restored lysosomal function, providing proof-of-principle that modulating lipid metabolism can counter lysosomal dysfunction. This work positions lipid-dependent membrane homeostasis as a determinant of LSD pathogenesis and supports metabolic interventions, potentially combined with ERT when available, as a strategy to improve the durability of lysosomal correction.

Abbreviations

AVM

Autophagic vacuolar myopathy

BSA

Bovine serum albumin

CLN2

Neuronal ceroid lipofuscinosis type 2

CQ

Chloroquine

DMSO

Dimethyl sulfoxide

DNL

De novo lipogenesis

D-PBS

Dulbecco’s phosphate-buffered saline

DMEM

Dulbecco’s modified Eagle’s medium

ER

Endoplasmic reticulum

ERT

Enzyme replacement therapy

ESCRT

Endosomal sorting complexes required for transport

FASN

Fatty acid synthase

GAA

Acid alpha-glucosidase

GAL-3

Galectin-3

GFP

Green fluorescent protein

GYS1

Glycogen synthase 1

IF

Immunofluorescence

KO

Knockout

LAMP1

Lysosomal associated membrane protein 1

LAMP2

Lysosomal associated membrane protein 2

LC3

Microtubule-associated proteins 1A/1B light chain 3

LC–MS/MS

Liquid chromatography coupled to tandem mass spectrometry

LLOMe

Leu-Leu methyl ester hydrobromide

LSDs

Lysosomal storage disorders

Lyso-IP

Lysosome immunopurification

Lyso-VPS15

TMEM192-V5-VPS15

MEFs

Mouse embryonic fibroblasts

NPA

Niemann–Pick A disease

PAS

Periodic acid–Schiff

PFA

Paraformaldehyde

PI

Phosphatidylinositol

PI3K-3

Class 3 phosphoinositide 3-kinase

PI3P

Phosphoinositide 3-phosphate

PLA

Proximity ligation assay

Unt

Untreated

VPS15

Vacuolar protein sorting 15

VPS34

Vacuolar protein sorting 34

Author contributions

E. Le G. and A.S. conducted most of the experiments, analyzed the data, prepared the figures, and contributed to the original paper preparation and editing. A.G. and C.O.D. contributed to the microscopic analyses and quantifications. N.F.H. assisted with the bioinformatic analyses. M.S., M.L., and I.N. performed the metabolomics analyses. V.N.-A. contributed with molecular analyses of GAL-3. K.H. and N.H. contributed to Lyso-IP and proteomics analyses. C.C. contributed to analyses in clinical samples. G.P. conceived and conceptualized the study, contributed to molecular analyses, analyzed the data, acquired funding, supervised the work, was responsible for the project administration, and wrote the paper. All authors discussed the results and commented on the paper.

Funding

This work was supported by grants from the European Research Council (grant no. ERC-CoG-MetaboSENS-819543) to G.P. and the Agence Nationale de la Recherche to G.P. and K.H. (grant no. ANR-21-CE14-0056-02-MuscLY).

Data availability

Mass spectrometry proteomics data are available in the main text or the Supplementary Material (Supplementary Table S1). RNA-seq data have been deposited in NCBI’s Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number (GSE316968). All other data supporting the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.