Integrated patient induced pluripotent stem cell models and Gaa-/- mice reveal central nervous system neural pathology in infantile-onset Pompe disease

Abstract

BACKGROUND

Pompe disease, or glycogen storage disease type II, is an autosomal recessive disorder caused by pathogenic variants in the acid α-glucosidase (GAA) gene. Although enzyme replacement therapy has improved survival in infantile-onset Pompe disease (IOPD), central nervous system (CNS) involvement has become increasingly evident because recombinant human GAA cannot cross the blood-brain barrier. The lack of reliable human models that faithfully recapitulate neural pathology has hindered mechanistic studies and the development of targeted therapies.

AIM

To establish an integrated disease-modeling platform by combining patient-derived induced pluripotent stem cell (iPSC) neural lineages with Gaa^-/- mice to reveal the cell-type-specific and progressive CNS pathology characteristic of IOPD.

METHODS

We generated iPSCs from peripheral blood mononuclear cells obtained from an enzyme replacement therapy-naïve IOPD patient and differentiated them into major CNS-relevant neural lineages, including cortical neurons, astrocytes, oligodendrocytes, and spinal motor neurons. These patient-derived neural cells were systematically assessed for lysosomal pathology and compared with the corresponding CNS phenotypes in Gaa^-/- mice. Lysosomal enlargement was quantified at the single-cell level in human neural cultures, whereas fluorescence-based intensity and area-fraction analyses were performed on mouse cerebral cortex and spinal cord tissues to evaluate in vivo lysosomal and glycogen-storage pathology.

RESULTS

Patient-derived iPSC neural cells exhibited marked lysosomal enlargement in cortical neurons, astrocytes, oligodendrocytes, and motor neurons, closely mirroring the cell-type-specific pathology observed in Gaa^-/- mice. In vivo, PAS staining revealed widespread glycogen accumulation in the cerebral cortex, hippocampus, and spinal cord, accompanied by pronounced lysosomal expansion across multiple neural populations. Gaa^-/- mice also showed progressive CNS involvement, with robust astrogliosis and substantial neuronal loss by six months of age. Consistently, long-term cultures of patient-derived neurons showed increased cleaved caspase-3-positive apoptosis.

CONCLUSION

Our integrated iPSC and mouse modeling approach recapitulates the essential CNS pathological features of IOPD, providing a relevant platform for mechanistic studies. This system may help guide future efforts to develop CNS-directed therapeutic strategies.

Key Words: Pompe disease; Lysosomal storage disease; Central nervous system; Blood-brain barrier; Induced pluripotent stem cells

Core Tip: This study integrates patient-derived induced pluripotent stem cell neural lineages and Gaa^-/- mice to model central nervous system pathology in infantile-onset Pompe disease. Patient-specific cortical neurons, astrocytes, oligodendrocytes, and motor neurons recapitulate key lysosomal pathology seen in vivo. Gaa^-/- mice exhibit progressive glycogen accumulation, astrogliosis, and neuronal loss. This dual-platform approach faithfully reproduces infantile-onset Pompe disease neural features, providing a valuable system for mechanistic research and testing central nervous system-directed therapies.

INTRODUCTION

Pompe disease (PD) is an autosomal recessive neuromuscular disorder caused by pathogenic variants in the acid α-glucosidase (GAA) gene, resulting in the pathological accumulation of glycogen within lysosomes. Clinically, PD is broadly categorized into infantile-onset (IOPD) and late-onset (LOPD) forms based on age at onset and disease severity[1,2]. First described in 1932, IOPD, also known as classical PD, presents within the first year of life with severe cardiomegaly, hepatomegaly, hypotonia, and muscle weakness, often progressing to cardiorespiratory failure and death if left untreated. The approval of enzyme replacement therapy (ERT) with recombinant human GAA in 2006 has markedly improved survival in patients with IOPD[3-5].

Although PD primarily affects skeletal muscle and cardiomyocytes, central nervous system (CNS) involvement has become increasingly recognized. Autopsy studies have consistently demonstrated glycogen accumulation in neurons throughout the cerebral cortex, brainstem, cerebellum, and anterior horn of the spinal cord[6-8]. Neuroimaging studies in patients have revealed delayed myelination, ventricular dilatation and progressive white matter abnormalities in the periventricular regions, basal ganglia, and frontoparietal lobes[9-13]. Although ERT markedly prolongs survival and improves cardiomyopathy as well as motor and respiratory function, its inability to cross the blood-brain barrier (BBB) has unmasked progressive CNS manifestations in long-term survivors, including cognitive impairment, speech difficulties, sensorineural hearing loss, visual disturbances, and persistent neuroimaging abnormalities[14-21].

To address this limitation, several novel therapeutic strategies are under investigation, including lentiviral-mediated hematopoietic stem cell gene therapy[22] and adeno-associated virus-mediated gene delivery[23,24]. Although promising, these approaches continue to face challenges, including immune responses, dose-related toxicities, particularly neurotoxicity, and limited transduction efficiency[25-27]. Therefore, there remains a pressing need to develop new therapeutics that can effectively cross the BBB while offering improved efficacy and fewer adverse effects. Moreover, the specific vulnerabilities of different CNS cell types and the underlying pathophysiological mechanisms remain poorly understood, representing a critical knowledge gap that hampers the development of targeted interventions.

Induced pluripotent stem cells (iPSCs) offer a powerful platform for disease modeling. Since their inception in 2006[28], iPSCs have undergone nearly two decades of technological advancement. It is now feasible to reprogram mature human somatic cells including skin fibroblasts and peripheral blood mononuclear cells (PBMCs), into a pluripotent state[28-30] and subsequently direct their differentiation into specific lineages using defined protocols. iPSCs offer distinct advantages over conventional animal or engineered cell models. For example, they retain the donor’s complete genetic background of the donor, enabling faithful recapitulation of disease characteristics. They also provide access to otherwise inaccessible human cell and tissue types, and their robust proliferative capacity supports high-throughput drug screening[31]. Successful applications have already been demonstrated across various fields, including oncology, infectious diseases, and inherited metabolic disorders[32-34]. In PD, previous studies have differentiated patient-derived iPSCs into neural stem cells[35] and neurons[36], revealing early pathological features. However, a systematic evaluation of cell-type-specific vulnerabilities across the major neural lineages, as well as modeling of progressive pathology, remains lacking.

Here, we established a patient-specific iPSC model to recapitulate CNS pathology in IOPD. We differentiated patient-derived iPSCs into cortical neurons, astrocytes, oligodendrocytes, and spinal motor neurons, and each lineage exhibited the hallmark lysosomal swelling characteristic of PD. These cellular phenotypes also mirrored key aspects of disease progression observed in Gaa^-/- mice. Collectively, our findings demonstrate that this iPSC-based neural platform captures both cell-type-specific susceptibilities and dynamic pathological progression, providing a physiologically relevant human model for mechanistic investigation and preclinical screening of potential CNS-targeted therapeutics in PD.

MATERIALS AND METHODS

Cell culture

Three cell lines were used in this study: A patient-derived iPSC line, the human embryonic stem cell line H9 (hESC-H9), and a healthy donor-derived iPSC control line. These cell lines were designated as follows: IOPD (patient-derived iPSC line), control 1 (hESC-H9), and control 2 (healthy donor-derived iPSC control line). The patient iPSC line was established by reprogramming PBMCs obtained from a 19-month-old female patient with IOPD who had not received ERT. Genetic sequencing confirmed the presence of compound heterozygous pathogenic variants in the GAA gene: C.1557G>A (p.M519I) and c.1798C>T (p.R600C). The widely used hESC-H9 line and a previously generated iPSC line[37] were used as healthy controls. All pluripotent stem cells were maintained on Matrigel (Corning, 354277, NY, United States)-coated 6-well plates in mTeSR™ Plus medium (STEMCELL Technologies, 100-0276, Canada) and passaged every 4-6 days using Dispase (STEMCELL Technologies, 07923, Canada). All cells were cultured at 37 °C in a humidified atmosphere containing 5% CO₂.

Karyotype analysis

Passage 15 iPSCs were treated with 0.2 μg/mL colchicine (Sigma-Aldrich, C9754, MA, United States) for 2 hours at 37 °C. After treatment, 1 mL of prewarmed Dispase was added to each well and incubated for 5 minutes. Digestion was terminated by adding 6 mL of culture medium. The cells were then centrifuged at 1000 rpm for 5 minutes, and the pellet was resuspended in prewarmed hypotonic solution (0.075 M KCl) and incubated for 30 minutes. The cells were gently dissociated and fixed with freshly prepared fixative (methanol:acetic acid = 3:1, v/v). The cell suspension was dropped onto prechilled glass slides, stained with Giemsa solution for 8 minutes, rinsed, and air-dried. G-banding analysis was performed using the CytoVision^® platform (Leica Biosystems). Twenty metaphase chromosome spreads were analyzed in accordance with the ISCN (2020) guidelines.

In vitro embryoid body differentiation

At 80% confluency, iPSCs were dissociated into single cells using 0.5 mmol/L EDTA and resuspended in embryoid body induction medium consisting of DMEM/F12 (M&C GENE TECHNOLOGY, CM15090, Beijing, China), 20% KnockOut Serum Replacement (Gibco, 10828028, NY, United States), 1% GlutaMax (Gibco, 35050061, NY, United States), 1% MEM-NEAA (Gibco, 11140035, NY, United States), and 100 μM β-mercaptoethanol (Sigma, M3148, MO, United States). The cell suspension was seeded into ultra-low-attachment 6-well plates. Half of the medium was replaced daily. After 7 days of suspension culture, EBs were transferred to Matrigel-coated plates and cultured in DMEM supplemented with 10% fetal bovine serum (Gibco, 10437028, NY, United States), with the medium changed every other day. After 14 days of differentiation, the cells were fixed for subsequent staining and analysis.

Neural cell differentiation

A mixed population of neural cells (cortical neurons, astrocytes, and oligodendrocytes) was used as previously described[38]. Briefly, iPSCs were seeded onto Matrigel-coated plates. Differentiation was initiated when colonies reached a diameter of 100-250 μm (day 0). During the first phase, cells were maintained in basal medium consisting of DMEM/F12, 1% penicillin-streptomycin (P/S; Solarbio, P1400, Beijing, China), 1% GlutaMax, 1% MEM-NEAA, 100 μM β-mercaptoethanol, and 25 μg/mL insulin (Solarbio, IR14902, Beijing, China), supplemented with 10 μM SB431542 (MCE, HY-10431), 250 nM LDN193189 (MCE, HY-12071, NJ, United States), and 100 nM retinoic acid (RA; Tocris, 0695/50, United States). The medium was changed every other day for 8 days. On day 8, the medium was replaced with N2 medium, consisting of basal medium supplemented with 1% N2 supplement (STEMCELL Technologies, 07152, Canada), 100 nM RA, and 1 μM SAG (MCE, HY-12848, NJ, United States), and was replaced daily for 4 days. On day 12, adherent cells were dissociated into single cells and transferred to ultra-low-attachment plates for suspension culture in N2B27 medium (N2 medium supplemented with 1 × B27; Gibco, 17504044, NY, United States). A total of 3 mL of medium was added per well, and two-thirds of the supernatant was replaced every other day until day 20. On day 20, aggregates were transferred to a 15-mL conical tube, allowed to settle by gravity, and cultured in PDGF medium, consisting of basal medium supplemented with 1 × N2, 1 × B27, 10 ng/mL PDGF-AA (PeproTech, 100-13A, NJ, United States), 10 ng/mL insulin-like growth factor-1 (PeproTech, 100-11, NJ, United States), 5 ng/mL hepatocyte growth factor (PeproTech, 100-39, NJ, United States), 10 ng/mL NT-3 (PeproTech, 450-03, NJ, United States), 60 ng/mL T3 (Sigma-Aldrich, T2877, MA, United States), 100 ng/mL biotin (MCE, HY-B0511, NJ, United States), and 1 μM cAMP (MCE, HY-B1511, NJ, United States). Two-thirds of the medium was replaced every other day until day 30. During this period, aggregates were manually dissociated under a microscope using a 1 mL syringe needle when necessary.

Twenty spherical aggregates (300-800 μm in diameter, with dark centers and golden or brown edges) were selected and seeded onto poly-L-ornithine (Sigma-Aldrich, A-004-C, MA, United States) and laminin (Gibco, 23017015, NY, United States)-coated 6-well plates for adherent culture. They were maintained in glial medium [basal medium supplemented with 1 × N2, 1 × B27, 10 mmol/L HEPES (Gibco, 15630080, NY, United States), 60 ng/mL T3, 100 ng/mL biotin, 1 μM cAMP, and 20 μg/mL ascorbic acid (Solarbio, IA0530, Beijing, China)] until day 55, with two-thirds of the medium replaced every other day, to obtain a mixed neural culture.

Motor neurons were differentiated according to previously established protocols with modifications[39-41]. Briefly, iPSCs were plated in 12-well plates on day 0 and cultured in neural induction medium consisting of 50% DMEM/F12, 50% neurobasal medium (Gibco, 21103049, NY, United States), 1% P/S, 1% SM1 supplement (STEMCELL Technologies, 0571, Canada), 1% N2 supplement, 1% GlutaMax, 1% MEM-NEAA, 100 μM β-mercaptoethanol, 200 nM LDN193189, 2 μM SB431542, 3 μM CHIR99021 (Tocris, 4423/10, United States), and 20 μg/mL insulin. Cells were maintained for 8 days with daily medium changes. In the second phase, cells were passaged onto 6-well plates and switched to patterning medium, which consisted of the neural induction medium supplemented with 10 ng/mL fibroblast growth factor 2 (PeproTech, 100-18B, NJ, United States), 10 ng/mL epidermal growth factor (Gibco, AF-100-15, NY, United States), 1 μM purmorphamine (MCE, HY-15108, NJ, United States), and 0.5 μM RA. Cells were cultured for 7 days with daily medium changes to generate motor neuron progenitors (MNPs). In the third phase, MNPs were dissociated with 0.25% EDTA-trypsin and plated onto poly-L-ornithine/Laminin-coated plates in expansion medium containing DMEM/F12, 1% P/S, 1.6 g/L glucose (Gibco, A24940-01, NY, United States), 1% N2 supplement, 0.05% SM1 supplement, 10 ng/mL fibroblast growth factor 2, 10 ng/mL epidermal growth factor, 20 μg/mL insulin, 100 nM LDN193189, 2 μM SB431542, 1 μM purmorphamine, 0.1 μM RA, 0.5 mmol/L valproic acid (MCE, HY-10585, NJ, United States), and 3 μM CHIR99021. In the fourth phase, expanded MNPs were passaged at a 1:5 ratio and transferred to maturation medium consisting of 50% DMEM/F12, 50% neurobasal medium, 1% P/S, 1% SM1 supplement, 1% N2 supplement, 1% GlutaMax, 1% MEM-NEAA, 10 ng/mL NT-3, 10 ng/mL CNTF (Gibco, 450-13, NY, United States), 10 ng/mL brain-derived neurotrophic factor (MCE, HY-P7116A, NJ, United States), 0.1 mmol/L L-ascorbic acid, and 10 ng/mL glial cell-derived neurotrophic factor (MCE, HY-P7182, NJ, United States). Cells were cultured for an additional 21 days with medium changes every 3-4 days to obtain mature motor neurons.

Animals

Heterozygous B6;129-Gaa<tm1Rabn>/J mice (male, 6-8 weeks old) were purchased from the Jackson Laboratory (United States, 004154) and bred to generate homozygous Gaa^-/- and wild-type (WT) offspring. All mice were housed under specific pathogen-free conditions with a 12-hour light/dark cycle, 40%-60% humidity, and a constant temperature of 22 ± 2 °C. All animal procedures were approved by the Ethics Committee of Qilu Hospital of Shandong University (Approval No. DWLL-202400116).

Immunofluorescence analysis

Cell samples and mouse tissue sections were fixed with 4% paraformaldehyde (Biosharp, BL539A, China). Blocking was performed for 1 hour at room temperature using a buffer containing 0.5% bovine serum albumin (Sigma-Aldrich, A1933, MA, United States) and 2% goat serum (Sigma-Aldrich, G9023, MA, United States) in phosphate buffered saline. Primary antibodies (Supplementary Table 1) were applied overnight at 4 °C. After washing with phosphate buffered saline, the secondary antibodies were incubated for 1 hour at 37 °C. Nuclei were stained with DAPI. Coverslips were mounted with nail polish, and images were acquired under identical exposure settings using confocal microscopy for quantification. For lysosomal size measurement, regions of interest were drawn around LAMP2-positive puncta in 10-15 randomly selected cells of each cell type from three independent differentiation experiments. For mouse tissue analysis, three non-overlapping random fields of view from the cerebral motor cortex and the ventral horn of the spinal cord were captured from each animal. The mean fluorescence intensity or area fraction was measured in ImageJ, and the values were averaged to yield a single data point per animal. Each group included at least three mice.

Statistical analysis

Data were analyzed and visualized using GraphPad Prism 10. Results are presented as the mean ± SEM. For comparisons between two groups, an unpaired two-tailed Student’s t-test was used when the data met the assumptions of normality and homoscedasticity; otherwise, the nonparametric Mann-Whitney U test was applied. For comparisons among more than two groups, one-way ANOVA was used. A P-value < 0.05 was considered statistically significant.

RESULTS

Clinical profile and iPSC generation

We enrolled a 19-month-old female patient clinically diagnosed with IOPD. Cranial magnetic resonance imaging revealed mildly increased T2-weighted signal intensity in the bilateral frontoparietal white matter, consistent with delayed myelination (Figure 1A). Genetic testing identified compound heterozygous pathogenic variants in the GAA gene, namely c.1557G>A (p.M519I) and c.1798C>T (p.R600C), confirming the molecular diagnosis (Figure 1B). After informed consent was obtained, PBMCs were collected and reprogrammed into iPSCs. Immunofluorescence staining demonstrated robust expression of the pluripotency markers OCT4, NANOG, SOX2, SSEA4, and TRA1-60, confirming their stem cell identity (Figure 1C). Karyotype analysis revealed a normal female karyotype (46, XX), indicating genomic stability of the established iPSC line (Figure 1D). Furthermore, embryoid body formation assays demonstrated the trilineage differentiation capacity of the iPSC line, as evidenced by the expression of lineage-specific markers, including hepatocyte nuclear factor 3-β for endoderm, α-smooth muscle actin for mesoderm, and Nestin for ectoderm (Figure 1E).

Figure 1 Generation and validation of induced pluripotent stem cell line from an infantile-onset Pompe disease patient with central nervous system involvement. A: Cranial T2-weighted magnetic resonance imaging reveals bilateral frontoparietal white matter mild hyperintensity (orange arrows); B: Sanger sequencing confirms pathogenic variants in the acid α-glucosidase gene: C.1557G>A (p.M519I) and c.1798C>T (p.R600C); C: Immunofluorescence staining demonstrates expression of pluripotency markers (OCT4, SSEA4, NANOG, TRA1-60, and SOX2) in patient-derived induced pluripotent stem cells. Scale bar = 100 μm; D: G-banding shows a normal 46, XX karyotype in the established infantile-onset Pompe disease patient - specific induced pluripotent stem cell line; E: In vitro EB formation and immunofluorescence staining confirm trilineage differentiation potential, as evidenced by expression of hepatocyte nuclear factor 3-β (endoderm), α-smooth muscle actin (mesoderm), and Nestin (ectoderm). Scale bar = 100 μm. HNF3β: Hepatocyte nuclear factor 3-β; SMA: Smooth muscle actin.

Neural differentiation and pathological assessment

To investigate CNS neural pathology in a cell-type-specific manner, patient-derived iPSCs, control iPSCs and hESC-H9 cells were differentiated into cortical neurons, astrocytes, oligodendrocytes, and spinal motor neurons according to established protocols (Figure 2A). After approximately 60 days of differentiation, immunofluorescence confirmed the generation of GFAP+ astrocytes, MBP+ oligodendrocytes, TUBB3+ neurons, and ChAT+ motor neurons (Figure 2B-E). Lysosomal pathology was then assessed using the lysosomal marker LAMP2. Representative images and quantitative analyses revealed significant lysosomal enlargement across all neural cell types derived from the IOPD patient compared with controls. Oligodendrocytes (MBP+) displayed markedly increased lysosomal size (P < 0.0001; Figure 2B), whereas astrocytes (GFAP+) also showed robust enlargement (P < 0.0001; Figure 2C). Cortical neurons (TUBB3+) exhibited substantial, although less pronounced, enlargement (P = 0.0004; Figure 2D), and spinal motor neurons (ChAT+) showed a moderate increase (P = 0.0004; Figure 2E). Together, these quantitative analyses of lysosomal enlargement indicate a cell-type-specific vulnerability that may underlie the neural pathology observed in PD.

Figure 2 Patient-specific induced pluripotent stem cell-derived neural cells exhibit cell-type-specific lysosomal pathology. A: Schematic diagram of the differentiation protocol to generate cortical neurons, astrocytes, oligodendrocytes, and spinal motor neurons; B-E: Representative immunofluorescence images and quantification of lysosomal size showing co-localization of the lysosomal marker LAMP2 (red) with lineage-specific markers (green): MBP (oligodendrocytes, B), GFAP (astrocytes, C), TUBB3 (cortical neurons, D), and ChAT (motor neurons, E). Nuclei were counterstained with DAPI (blue). Quantitative analysis revealed significant lysosomal enlargement in all infantile-onset Pompe disease-derived neural subtypes vs controls (oligodendrocytes: ^dP < 0.0001; astrocytes: ^dP < 0.0001; cortical neurons: ^cP < 0.001; motor neurons: ^cP < 0.001; unpaired two-tailed Student’s t-test). For each cell type, 10-15 cells were quantified per experiment from three independent differentiations. Data are presented as mean ± SEM. Scale bar = 50 μm. iPSCs: Induced pluripotent stem cell; IOPD: Infantile-onset Pompe disease.

CNS phenotypes in Gaa^-/- mice

We next examined the brain and spinal cord of 3-month-old Gaa^-/- mice to validate the pathophysiological relevance of our iPSC-based models (Figure 3A). PAS staining revealed striking glycogen accumulation and vacuolar degeneration in neurons of the cerebral cortex and hippocampus, as well as in anterior horn neurons of the cervical and thoracic spinal cord, in Gaa^-/- mice compared with WT littermates (Figure 3B and C). These histological alterations, which directly visualize glycogen storage, were consistently observed across multiple CNS regions, confirming widespread glycogen storage pathology at an early stage of disease. To further characterize the affected neural populations, we performed immunofluorescence co-staining for cell-type-specific markers and the lysosomal membrane protein LAMP2. In the cerebrum, oligodendrocytes (MBP+), astrocytes (GFAP+), and cortical neurons (TUBB3+) from Gaa^-/- mice exhibited prominent lysosomal expansion compared with WT controls (Figure 3D). Similarly, in the spinal cord, oligodendrocytes, astrocytes, and motor neurons displayed robust LAMP2 accumulation and lysosomal swelling in Gaa^-/- mice (Figure 3E). Together, these findings demonstrate that glycogen accumulation and lysosomal pathology occur broadly across multiple CNS neural cell types in PD.

Figure 3 Central nervous system lysosomal pathology in Gaa^-/- mice. A: Schematic overview of tissue sampling from the brain and spinal cord of 3-month-old Gaa^-/- mice; B: PAS staining reveals glycogen accumulation (magenta granules) in the cerebral cortex and hippocampus of Gaa^-/- mice. Scale bar = 200 μm; C: PAS staining reveals glycogen accumulation (magenta granules) in the spinal cord ventral horn of Gaa^-/- mice. Scale bar = 50 μm; D: Immunofluorescence co-staining of neural markers MBP, GFAP or TUBB3 with the lysosomal marker LAMP2 revealed pronounced lysosomal pathology in the cerebrum of Gaa^-/- mice. Scale bar = 20 μm; E: Immunofluorescence co-staining for neural marker MBP, GFAP or ChAT with the lysosomal marker LAMP2 shows pronounced lysosomal pathology in the spinal cord of Gaa^-/- mice. Scale bar = 20 μm. WT: Wild-type; GAA: Acid α-glucosidase.

Progressive CNS degeneration across in vivo and in vitro models

To evaluate whether the iPSC-based system recapitulates progressive neural pathology, we examined neuronal integrity and glial responses in the Gaa^-/- mouse model and neuronal survival in patient iPSC-derived cortical neurons. GFAP immunostaining revealed that Gaa^-/- mice exhibited early glial activation in both the cerebral cortex (Figure 4A) and spinal cord (Figure 4B) at 3 months of age, which progressed to pronounced astrogliosis by 6 months. Quantitative fluorescence intensity analyses confirmed a significant increase in GFAP signal in Gaa^-/- mice compared with age-matched WT controls at both time points (cerebral cortex: 3 months, P = 0.0151; 6 months, P = 0.0269; spinal cord: 3 months, P = 0.0065; 6 months, P < 0.0001; Figure 4C and D). To further assess neuronal integrity, we performed NeuN staining. Neuronal density appeared preserved in Gaa^-/- mice at 3 months; however, by 6 months there was an obvious reduction in the number of NeuN+ cells in both the cerebral cortex (Figure 4E) and spinal cord (Figure 4F). Quantitative assessment confirmed a significant 1.3-fold decrease in the NeuN+ neuronal area fraction in Gaa^-/- mice relative to WT controls (cerebral cortex, P = 0.0336; spinal cord, P = 0.0244; Figure 4G and H). These findings indicate a time-dependent shift from reactive gliosis to overt neuronal loss in the CNS in PD. To determine whether these degenerative features were recapitulated in vitro, patient-specific iPSC-derived neural cells were maintained in long-term culture. At day 90, TUBB3+ neurons exhibited significantly increased apoptosis, as indicated by enhanced cleaved caspase-3 immunoreactivity, compared with control cells (P = 0.0223; Figure 4I and J). Collectively, these findings demonstrate that the iPSC-derived neural model faithfully recapitulates the progressive CNS pathology observed in vivo, thereby supporting its utility as a physiologically relevant platform for mechanistic studies and preclinical therapeutic testing in PD.

Figure 4 Progressive neurodegeneration in Gaa^-/- mice and patient-derived neural cultures. A: Representative GFAP immunofluorescence images of the cerebral cortex from 3- and 6-month-old wild-type (WT) and Gaa^-/- mice. Scale bar = 100 μm; B: Representative GFAP immunofluorescence images of the spinal cord from 3- and 6-month-old WT and Gaa^-/- mice. Scale bar = 100 μm; C: Quantification of GFAP fluorescence intensity in the cerebral cortex shown in A (n = 4 mice per group; data are presented as mean ± SEM; ^aP < 0.05, WT 3 months vs Gaa^-/- 3 months; ^aP < 0.05, WT 6 months vs Gaa^-/- 6 months; ^aP < 0.05, Gaa^-/- 3 months vs Gaa^-/- 6 months; one-way ANOVA); D: Quantification of GFAP fluorescence intensity in the spinal cord shown in B (n = 4 mice per group; data are presented as mean ± SEM; ^bP < 0.01, WT 3 months vs Gaa^-/- 3 months; ^dP < 0.0001, WT 6 months vs Gaa^-/- 6 months; ^aP < 0.05, Gaa^-/- 3 months vs Gaa^-/- 6 months; one-way ANOVA); E: Representative NeuN immunostaining of the cerebral cortex from 3- and 6-month-old WT and Gaa^-/- mice. Scale bar = 100 μm; F: Representative NeuN immunostaining of the spinal cord from 3- and 6-month-old WT and Gaa^-/- mice. Scale bar = 100 μm; G: Quantification of the NeuN+ area fraction in the cerebral cortex shown in E (n = 3 mice per group; data are presented as mean ± SEM; ^aP < 0.05, WT 6 months vs Gaa^-/- 6 months; one-way ANOVA); H: Quantification of the NeuN+ area fraction in the spinal cord shown in F (n = 3 mice per group; data are presented as mean ± SEM; ^aP < 0.05, WT 6 months vs Gaa^-/- 6 months; ^aP < 0.05, Gaa^-/- 3 months vs Gaa^-/- 6 months; one-way ANOVA); I: Immunofluorescence images of patient - derived induced pluripotent stem cell neurons cultured for 90 days, showing increased cleaved caspase-3 (red) positivity in TUBB3+ neurons (green) compared to controls. Nuclei counterstained with DAPI (blue). Scale bar = 20 μm; J: Quantification of cleaved caspase-3 positivity in TUBB3+ neurons shown in I (n = 3 independent differentiation; data are presented as mean ± SEM; ^aP < 0.05; unpaired two-tailed Student’s t-test). WT: Wild-type; GAA: Acid α-glucosidase; IOPD: Infantile-onset Pompe disease.

DISCUSSION

PD is an autosomal recessive metabolic disorder caused by a deficiency of GAA, and patients with IOPD exhibit complete loss of enzyme activity. The introduction of ERT has substantially improved survival, allowing many treated IOPD patients to achieve motor milestones previously unattainable[42,43]. This prolonged survival has enabled long-term multidisciplinary follow-up, yielding deeper insight into the multisystem manifestations of the disease. With respect to CNS involvement, postmortem studies of ERT-treated IOPD patients have revealed glycogen accumulation in the brain[6]. Neuroimaging and cognitive assessments have further documented slowly progressive white matter abnormalities and cognitive decline throughout childhood and adolescence[14,44]. These clinical observations highlight a critical limitation of existing ERT, namely, its inability to cross the BBB. Therefore, targeting CNS pathology represents an essential direction for next-generation therapeutic strategies.

To elucidate the mechanisms underlying neurological impairment and develop effective CNS-targeted therapies, there is a critical need for disease models that faithfully recapitulate patient-specific neuropathology. Current research on the pathophysiology of PD and the development of new therapeutics relies largely on animal models or engineered cell lines[23,45,46]. However, these in vivo and in vitro models are constrained by inherent species-specific differences that are difficult to overcome. In addition, although valuable insights have been gained from studies using readily accessible human samples such as patient serum or fibroblasts[47,48], obtaining other relevant cell types, including cardiomyocytes, glial cells and neurons, remains challenging, thereby limiting the scope of mechanistic and translational investigations. iPSCs offer a powerful alternative platform for disease modeling and provide several advantages over conventional approaches. For example, iPSCs and their derivatives retain the patient’s complete genetic background, provide access to otherwise inaccessible human cell types, avoid species-specific differences, circumvent ethical concerns, and are amenable to high-throughput drug screening[31,49]. Several previous studies have successfully generated iPSC-derived cardiomyocytes, neural stem cells, cortical neurons, and skeletal muscle cells from patients with PD[35,36,50-52], thereby recapitulating lysosomal pathology and enabling therapeutic compound screening in human-derived systems. However, studies focusing on CNS pathology have typically been limited to single neural lineages and have lacked longitudinal in vivo validation to capture the progressive dynamics of disease. Consequently, cell-type-specific vulnerability within the CNS and the temporal progression of pathology have remained largely unexplored.

In this study, we established a patient-specific iPSC model that recapitulates key aspects of CNS pathology in IOPD, demonstrating lysosomal enlargement across a spectrum of neural cell types and revealing progressive neuronal apoptosis in long-term cultures. This in vitro progression mirrors the temporal profile of lysosomal pathology and neuronal loss observed in Gaa^-/- mice, thereby supporting the physiological relevance of our platform.

The progressive CNS pathology observed in both Gaa^-/- mice and long-term patient-derived neural cultures has clear clinical relevance. Notably, lysosomal enlargement was evident not only in neurons but also in oligodendrocytes and astrocytes in both patient-derived iPSC cultures and Gaa^-/- mouse tissues, indicating substantial involvement of glial populations in IOPD-associated CNS pathology. These findings may account for the white matter abnormalities and delayed myelination frequently reported in IOPD patients[9,10,13], given the critical roles of oligodendrocytes in myelination and astrocytes in maintaining white matter homeostasis. Together, these observations highlight the need for therapeutic strategies that address pathology across multiple neural lineages, encompassing both neuronal and glial compartments. Several emerging therapeutic approaches aim to overcome the limitations of current ERT due to the BBB. Hematopoietic stem cell gene therapy has demonstrated preclinical efficacy in reducing CNS glycogen accumulation by leveraging microglia as cellular delivery vehicles for GAA[22], whereas next-generation adeno-associated virus-mediated gene therapies are currently being evaluated in clinical trials for IOPD[24]. In this context, our integrated platform enables cell-type-specific assessment of lysosomal pathology and neuronal survival across four neural lineages, providing a versatile preclinical framework for evaluating CNS-targeted therapeutic candidates. Furthermore, the time-dependent transition from lysosomal dysfunction to neuronal apoptosis observed in our long-term cultures suggests the existence of a therapeutic window for early intervention. Clinically, this highlights the importance of initiating CNS-targeted therapies at early disease stages, ideally before overt neurocognitive decline, to preserve neuronal integrity and prevent irreversible damage.

Previous studies have laid important groundwork for modeling CNS involvement in PD using patient-derived iPSC models. Cheng et al[35] generated neural stem cells from PD iPSCs and demonstrated early lysosomal pathology, providing important foundational insights into CNS involvement in PD. Building on this work, our study further examines multiple differentiated neural lineages and investigates progressive disease features. Similarly, Huang et al[36] differentiated Pompe iPSCs into cortical neurons for therapeutic compound screening, establishing a valuable platform for identifying potential therapeutics. The present study extends these findings by systematically comparing multiple neural lineages - including glial and motor neuron subtypes, and by incorporating in vivo validation of disease progression using Gaa^-/- mice. Our work further advances these observations by systematically characterizing the vulnerability of multiple CNS-relevant neural lineages derived from the same iPSC line, including cortical neurons, spinal motor neurons, astrocytes, and oligodendrocytes. Furthermore, through parallel analyses of the Gaa^-/- mouse model and longitudinal in vitro assessments, we demonstrate that our iPSC-derived neural cells not only recapitulate lysosomal pathology but also faithfully model dynamic disease progression. This integrated approach bridges the gap between static cellular phenotyping and dynamic disease progression, providing a more physiologically relevant platform for mechanistic studies and drug screening than existing models.

While our integrated platform offers these advantages, several limitations should be acknowledged. First, the iPSC line used in this study was derived from a single IOPD patient carrying compound heterozygous GAA variants (c.1557G>A and c.1798C>T). Although this proof-of-concept study provides detailed characterization of CNS pathology in a genetically defined patient background, the generalizability of our findings to other IOPD patients with different GAA mutation spectra remains to be established in future work. Second, our study did not include isogenic gene-corrected control iPSC lines generated through precise genome-editing approaches such as CRISPR-Cas9. Such isogenic controls would definitively link the observed lysosomal and neurodegenerative phenotypes to GAA deficiency, excluding potential confounders arising from reprogramming artifacts or clonal variation. We acknowledge this limitation and will address it in future work to improve the robustness of experimental conclusions. Third, we did not investigate whether microglia exhibit pathological alterations, despite their essential roles in CNS homeostasis and neuroinflammation. Integrating microglia or microglia-containing coculture systems into future models will provide a more comprehensive understanding of the cellular landscape affected in PD.

CONCLUSION

In summary, our human neural cell platform captures key aspects of patient neuropathology, overcomes the challenge of obtaining live human CNS cells and reduces reliance on animal models. The successful generation of patient-specific neural derivatives provides an effective tool for future mechanistic studies and for the development of targeted interventions for the debilitating CNS manifestations of PD.