Rare‐Variant Burden across Lysosomal Genes Implicates Sialylation and Ganglioside Metabolism in Parkinson's Disease

Konstantin Senkevich; Sitki Cem Parlar; Cloe Chantereault; Lang Liu; Eric Yu; Uladzislau Rudakou; Jamil Ahmad; Jennifer A Ruskey; Farnaz Asayesh; Dan Spiegelman; Cheryl Waters; Oury Monchi; Yves Dauvilliers; Nicolas Dupré; Lior Greenbaum; Sharon Hassin‐Baer; Irina Miliukhina; Alla Timofeeva; Anton Emelyanov; Sofya Pchelina; Roy N Alcalay; Ziv Gan‐Or

doi:10.1002/mds.70391

. 2026 Jun 10;41(9):2349–2356. doi: 10.1002/mds.70391

Rare‐Variant Burden across Lysosomal Genes Implicates Sialylation and Ganglioside Metabolism in Parkinson's Disease

Konstantin Senkevich ^1,^2,³, Sitki Cem Parlar ^1,^2,⁴, Cloe Chantereault ^1,⁴, Lang Liu ^1,⁴, Eric Yu ^1,⁴, Uladzislau Rudakou ¹, Jamil Ahmad ^1,², Jennifer A Ruskey ^1,², Farnaz Asayesh ^1,², Dan Spiegelman ^1,², Cheryl Waters ⁵, Oury Monchi ^2,^6,⁷, Yves Dauvilliers ⁸, Nicolas Dupré ^9,¹⁰, Lior Greenbaum ^11,¹², Sharon Hassin‐Baer ^12,¹³, Irina Miliukhina ^14,¹⁵, Alla Timofeeva ¹⁵, Anton Emelyanov ¹⁵, Sofya Pchelina ¹⁵, Roy N Alcalay ^5,^12,¹⁶, Ziv Gan‐Or ^1,^2,^4,^✉

PMCID: PMC13602292 PMID: 42272075

Abstract

Background

Lysosomal dysfunction is central to Parkinson's disease (PD) pathogenesis, with GBA1 representing the strongest established genetic risk factor. Numerous other genes involved in lysosomal sphingolipid, glycosphingolipid, and ceramide metabolism have been proposed as contributors to PD, highlighting the need for genetic analyses across these pathways.

Objectives

The aim was to evaluate the contribution of rare variants across lysosomal genes to PD risk.

Methods

We analyzed rare variants (minor allele frequency ≤0.01) across 36 lysosomal genes in 8267 individuals with PD and 68,208 controls, including 793 early‐onset PD (≤50 years) cases. Targeted sequencing was performed in four cohorts at McGill University (3456 cases and 2664 controls) and combined with whole‐genome sequencing data from the United Kingdom (UK) Biobank (2848 cases, 62,451 controls) and the Accelerating Medicines Partnership‐PD cohort (1963 cases, 3093 controls). Associations were tested using Sequence Kernel Association Test‐Optimal across variant classes (rare variants, nonsynonymous, loss‐of‐function, and predicted damaging variants with combined annotation‐dependent depletion score >20), followed by meta‐analysis across cohorts. Domain‐level analyses were performed for variants located within protein domains. False discovery rate (FDR) correction was applied.

Results

Meta‐analysis identified a significant association between rare variants in ST3GAL3 and Parkinson's disease (Pfdr = 0.04). Domain‐based analyses showed enrichment of nonsynonymous variants within the β‐acetyl‐hexosaminidase–like domain of HEXA (P = 8.0 × 10⁻⁴), although this signal did not survive correction (Pfdr = 0.154). In early‐onset PD, domain‐based analyses identified significant associations in NAGLU (Pfdr = 7.3 × 10⁻⁶) and ST3GAL5 (Pfdr = 0.03).

Conclusions

Rare variants across multiple lysosomal pathways, particularly those related to sialylation, ganglioside metabolism, ceramide biology, and lysosomal proteolysis, may contribute to PD susceptibility beyond GBA1, highlighting pathways for future replication and investigation. © 2026 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

Keywords: Parkinson's disease, lysosomal genes, rare variants, burden analysis, sialylation, ganglioside metabolism, sphingolipid metabolism, ceramide metabolism, ST3GAL3, lysosomal dysfunction

Lysosomal dysfunction is increasingly recognized as a key component of Parkinson's disease (PD) pathogenesis. ¹ Mutations in the GBA1 gene, encoding β‐glucocerebrosidase, are the most common genetic risk factor for PD, and both heterozygous and biallelic variants are associated with increased disease risk and rapid progression. ² , ³ , ⁴ Beyond GBA1, rare and common variants in other lysosomal genes, such as GALC, ARSA, CTSB, SMPD1, TMEM175, SPTSSB, and ASAH1, have also been implicated in PD susceptibility. ⁵ , ⁶ , ⁷ , ⁸ , ⁹ Studies have demonstrated that the cumulative effect of multiple risk alleles across the lysosomal pathway influences PD risk and may also influence neuropathological severity, independent of GBA1 status. ⁸ , ¹⁰ , ¹¹ , ¹²

Lysosomal impairment disrupts the autophagy‐lysosomal and ubiquitin‐proteasome systems. This leads to defective clearance of α‐synuclein and other substrates and organelles, and promotes aggregation, neuroinflammation, and dopaminergic neurodegeneration. ¹³ , ¹⁴ , ¹⁵ GBA1 mutations exemplify this mechanism, whereby reduced β‐glucocerebrosidase activity leads to increased glucosylceramide and glucosylsphingosine, α‐synuclein accumulation, and lysosomal dysfunction. ¹⁶ , ¹⁷ Consistent with this model, multiple additional lysosomal genes have been shown to modulate α‐synuclein toxicity and autophagic‐lysosomal pathway integrity. ⁷ , ¹⁸

Despite these advances, many other genes involved in the lysosomal sphingolipid and ceramide metabolism pathways could be potential contributors to PD, emphasizing the need for comprehensive genetic analysis within these pathways. To address this gap, we analyzed the burden of rare variants across 36 lysosomal genes in a cohort of 8267 patients and 68,208 controls, aiming to define the contribution of individual genes and the aggregate lysosomal genetic burden to PD susceptibility and heterogeneity.

Materials and Methods

Study Population

We analyzed 8267 individuals with PD and 68,208 controls (Table 1). The studied population included: (1) four cohorts sequenced at McGill University, comprising participants recruited in Quebec, Canada and Montpellier, France, ¹⁹ Columbia University, New York, United States (US), ²⁰ Sheba Medical Center, Israel ²¹ and the Pavlov First State Medical University and the Institute of Human Brain, Saint‐Petersburg, Russia ⁶ (3456 cases; 2664 controls); (2) the United Kingdom (UK) Biobank (2848 cases; 62,451 controls); and (3) the Accelerating Medicines Partnership‐PD (AMP‐PD) whole‐genome sequencing cohorts (1963 cases; 3093 controls). The Columbia cohort included participants from mixed ancestry backgrounds, with the majority being of European or Ashkenazi Jewish ancestry and smaller proportions of Hispanic and Black participants. The Sheba cohort included participants of Ashkenazi Jewish and non‐Ashkenazi Jewish ancestry. A subset of 793 patients with early‐onset PD, defined as an age at onset of 50 or less years, was analyzed separately given the known association of GBA1 variants with earlier disease onset. ² Diagnoses were established by movement‐disorder specialists according to the UK Brain Bank ²² or Movement Disorder Society (MDS) clinical diagnostic criteria. ²³ All participants provided written informed consent under protocols approved by their respective research ethics boards. This research was approved by McGill ethics boards.

TABLE 1.

Study population

Cohort	All Parkinson's disease cases			Controls
Cohort	n	Mean age (SD)	Males (%)	n	Mean age (SD)	Males (%)
McGill cohort	1123	59.84 (11.09)	715 (63.7)	1196	48.43 (15.26)	541 (45.2)
Sheba cohort	702	60.93 (11.79)	437 (62.3)	551	33.22 (7.84)	316 (57.4)
Columbia cohort	1122	59.82 (11.36)	736 (65.6)	500	64.29 (10.02)	177 (35.4)
Pavlov and Human Brain Institute cohort	509	66.15 (9.29)	206 (40.5)	417	73.61 (10.49)	172 (41.2)
UKBB	2848	63.00 (5.25)	1065 (37.4)	62,451	56.79 (8.02)	33,527 (53.7)
AMP‐PD	1963	64.54 (9.52)	1260 (64.2)	3093	69.76 (13.02)	1496 (48.4)
All cohorts	8267	–		68,208

Cohort	Parkinson's disease with early onset ≤50 yr			Controls
Cohort	n	Mean age (SD)	Males (%)	n	Mean age (SD)	Males (%)
McGill cohort	201	44.09 (6.24)	147 (73.1)	1196	48.43 (15.26)	541 (45.2)
Sheba cohort	140	43.64 (7.15)	86 (61.4)	551	33.22 (7.84)	316 (57.4)
Columbia cohort	158	42.96 (6.15)	103 (65.2)	500	64.29 (10.02)	177 (35.5)
Pavlov and Human Brain Institute cohort	28	43.29 (6.54)	13 (46.4)	417	73.61 (10.49)	172 (41.2)
UKBB	75	46.19 (2.72)	32 (42.7)	1932	56.52 (7.94)	1090 (56.4)
AMP‐PD	191	46.42 (4.31)	114 (59.7)	3093	69.76 (13.02)	1496 (48.4)
All cohorts	793	–	–	7688	–	–

Open in a new tab

Abbreviations: SD, standard deviation; AMP‐PD, Accelerating Medicines Partnership‐Parkinson's disease.

Gene Selection, Sequencing, and Processing

We designed a targeted lysosomal gene panel comprising 36 genes, which was applied to four cohorts collected at McGill University. The panel included genes encoding (1) enzymes involved in ceramide, glycosphingolipid, and ganglioside metabolism (including acid ceramidase, hexosaminidases, cathepsins, and components of the ceramide/S1P pathway); (2) mucopolysaccharidosis (MPS)‐related enzymes associated with neuronal lysosomal storage phenotypes; and (3) glycosyl‐transferases and sialyltransferases that regulate glycosphingolipid composition and lysosomal membrane trafficking. These genes were selected based on a structured literature review of genes most likely to influence lysosomal substrate processing and lipid homeostasis, processes strongly implicated in PD biology. The complete gene list, protein functions, and known associated conditions as per Online Mendelian Inheritance in Man are provided in Table S1. GBA1 was not included in this analysis because its association with PD is well established and has been analyzed separately in our cohorts. ²⁴

In the McGill cohorts, targeted capture was performed using molecular inversion probes (MIPs), following our protocol (https://github.com/gan-orlab/MIP_protocol). ²⁵ The libraries were sequenced on Illumina NovaSeq 6000 platform (paired‐end 100 bp) at the Genome Quebec Innovation Centre. Sequence reads were aligned to the hg19 using Burrows‐Wheeler Aligner (BWA), with base‐quality recalibration and variant calling conducted using Genome Analysis Toolkit (GATK) v3.8, joint genotyping was performed across all sequencing batches and the resulting variants were subsequently mapped from hg19 to hg38 using CrossMap. ²⁶ For external datasets, the same 36 genes were extracted from the UK Biobank (UKBB) and AMP‐PD genome sequencing data (hg38).

Quality Control

Variant‐ and sample‐level quality control (QC) was performed following established pipelines used in our previous sequencing studies, with adaptations for each dataset. ⁵ , ⁶ In the McGill cohorts, variants were excluded if the call rate was below 90%, genotype quality (GQ) below 20, or sequencing depth below 30X. Multi‐allelic sites were normalized and variants deviating from Hardy–Weinberg equilibrium in controls (P < 1 × 10⁻⁶) were removed. Variants supported by fewer than 25% of reads carrying the alternate allele were also excluded to minimize false positives. At the sample level, we excluded individuals with an average genotyping rate below 90%, discordant genetic and reported sex, heterozygosity outliers, or relatedness (first‐ or second‐degree relatives). In the McGill cohorts, ethnicity was included as a covariate in all association analyses to minimize potential confounding.

For the UKBB and AMP‐PD whole‐genome datasets, we applied a sequencing depth cutoff of more than 20× and restricted analyses to participants of European ancestry, determined using principal components or ancestry labels (UKBB field 21,000). Population structure was assessed by principal component analysis (PCA), and ancestry outliers were excluded. The AMP‐PD variant QC additionally followed the official consortium pipeline (https://amp-pd.org/whole-genome-data).

Annotation, Grouping, and Domains

Variants were annotated with ANNOVAR. ²⁷ Rare variants were defined as minor allele frequency (MAF) ≤0.01 within each study cohort. Burden analysis for rare variants was performed in several groups: all rare variants, nonsynonymous variants; loss‐of‐function (nonsense, frameshift, and canonical splice) and predicted missense damaging (combined annotation‐dependent depletion [CADD] ≥20). For domain‐based burden, variants were mapped to UniProt/Pfam‐defined functional regions (catalytic, ligand‐binding, and luminal/membrane segments) and analyses were repeated using the same variant groupings.

Statistical Analysis

Within each cohort, we tested rare‐variant association using SKAT‐O (R package), ²⁸ adjusting for age, sex, and ethnicity when available. We meta‐analyzed rare variants using MetaSKAT. In addition, we used PAGEANT to estimate power for single‐gene rare‐variant association testing across the combined cohorts. ²⁹ Power was calculated assuming 8267 cases and 68,208 controls, a significance threshold of 0.0001, the SKAT framework, and fixed explained variance of 1%, under scenario 2 and scenario 3. False discovery rate (FDR) correction using the Benjamini–Hochberg method was applied. Corrected Pfdr <0.05 was considered significant, uncorrected P < 0.05 is reported as nominal.

Results

Rare‐Variant Analysis Identifies an Association of ST3GAL3 with PD

Targeted sequencing in McGill cohorts achieved high on‐target performance (average >95% bases at ≥30×), enabling confident rare‐variant detection. To assess rare variant burden associations across lysosomal genes, we performed SKAT‐O analyses in the McGill cohorts, UKBB, and AMP‐PD cohorts. Power estimation using PAGEANT suggested that the study was adequately powered overall (>90%) under the modeled assumptions. The full list of coding rare variants across the 36 targeted lysosomal genes in all studied cohorts is provided in Table S2.

Meta‐analyses across the McGill, UKBB, and AMP‐PD cohorts identified an association between all rare variants in ST3GAL3 and PD (Pfdr = 0.04). Several additional genes showed nominal associations (P < 0.05), including HGSNAT, ASAH1, CTSD, HEXA, ST3GAL4, and SGPP1 (Table 2, Tables S3 and S4). Most analyzed genes were not significantly associated with PD after correction for multiple testing, and full gene‐level results are provided in the Tables S3 and S4. These signals mostly reflected the aggregate contribution of multiple rare alleles, with no single variant consistently driving the gene‐level associations across cohorts.

TABLE 2.

Rare‐variant burden meta‐analysis associations across lysosomal genes in Parkinson's disease

SetID	Variant type	P	Pfdr
*ST3GAL3*	All	0.0002	0.04
HGSNAT	All	0.009	0.604
HGSNAT	Nonsynonymous	0.012	0.604
ASAH1	CADD	0.020	0.706
ASAH1	LOF	0.021	0.706
CTSD	All	0.026	0.741
ST3GAL4	CADD	0.037	0.808
HEXA	All	0.044	0.808
SGPP1	LOF	0.045	0.808
GNS	CADD	0.054	0.808

Open in a new tab

Note: P values were obtained from meta‐analysis across cohorts with FDR correction. Bold indicates results that remained significant after FDR correction. Only ST3GAL3 remained significant after multiple‐testing correction (Pfdr <0.05); nominal associations (P < 0.05) are shown for completeness. Variant sets included all rare variants (MAF ≤0.01), nonsynonymous, LOF, and CADD (CADD Phred ≥20).

Abbreviations: CADD, Combined Annotation‐Dependent Depletion; LOF, loss‐of‐function; FDR, false discovery rate; MAF, minor allele frequency.

Single‐variant analysis highlighted several variants enriched in cases. In ASAH1, the nominal gene‐level signal was influenced by the missense variants p.V369I, which was observed in a total of 14 PD cases and no controls across the Pavlov and Human Brain and Columbia cohorts. However, this variant did not demonstrate consistent enrichment among cases across other cohorts. In addition, a rare loss‐of‐function stop‐gain variant in ASAH1 p.Q95X was observed across three independent cohorts. This variant was detected in 13 PD cases and five controls in the McGill cohort (odds ratio [OR] = 2.79, P = 0.052), 23 cases and six controls in the Columbia cohort (OR = 1.72, P = 0.239), and nine cases and four controls in the Pavlov and Human Brain cohort (OR = 1.85, P = 0.309). Although none of these associations reached statistical significance individually, the recurrent observation of this loss‐of‐function variant across cohorts suggests a potential contribution to disease risk. In NAGLU, the p.Ser141Thr variant was observed in four cases and one control in the Columbia cohort (OR = 1.78, P = 0.61), four cases and zero controls in the Sheba cohort, and three controls from UKBB. This variant is common in African populations Genome Aggregation Database (MAF = 0.067), and additional replication in independent cohorts of Jewish, African, and African‐American ancestry is required to confirm this association. Another recurrent variant was NAGLU p.T441M, which was observed in PD cases across multiple cohorts (total n = 12), with low or absent counts (total n = 5) in controls across all cohorts, but did not reach significance in individual cohorts due to its rarity. Single‐variant analyses did not identify other significant associations.

In early‐onset PD, rare‐variant burden meta‐analysis identified only nominal associations (P < 0.05) across several genes, including GRN, ST3GAL5, UGCG, HEXB, MANBA, PPT1, HGSNAT, CERK, B4GALT1, B4GALT7, ST3GAL3, and SGSH, none of which survived correction for multiple testing (Tables S5 and S6).

Domain‐Based Burden Analysis Identified Significant Association in Sphingolipid‐Related Genes

Domain‐based burden analysis identified a significant enrichment of nonsynonymous variants within the β‐acetyl‐hexosaminidase–like domain of HEXA (P = 8.0 × 10⁻⁴), which did not survive correction for multiple testing (Pfdr = 0.15). Additional domain‐level associations were nominal (P < 0.05) and involved HGSNAT, ASAH1, CERK, and GRN. Detailed domain definitions, along with cohort‐specific and meta‐analysis results, are provided in Tables S7 and S8.

Within HGSNAT, the p.L406W variant was enriched in PD cases in three McGill‐sequenced cohorts (Pavlov and Human Brain, Sheba, and Columbia; 8 cases vs. 1 control), while showing an opposite direction of effect in the McGill cohort (4 cases vs. 13 controls). The variant was not detected in other cohorts after quality control. For ASAH1, both variants described above (p.V369I and p.Q95X) map to the ceramidase domain and together account for the observed ceramidase domain‐level association.

Domain‐Based Burden Analysis in Early‐Onset PD

In early‐onset PD, domain‐based analyses identified significant associations in NAGLU (Pfdr = 7.33 × 10⁻⁶) and ST3GAL5 (Pfdr = 0.03) (Tables S9 and S10). No other results survived multiple testing correction at domain level analysis. The associations in NAGLU and ST3GAL5 were not driven by single variants.

Discussion

In this study, we evaluated the burden of rare variants across multiple lysosomal genes in PD using multi‐cohort sequencing data. Our findings implicate pathways involved in sialylation and glycosylation (ST3GAL3, ST3GAL4, and ST3GAL5), ganglioside metabolism (HEXA), ceramide, and sphingolipid biology (ASAH1, SGPP1, and CERK), and lysosomal proteolysis (CTSD) in PD susceptibility, further extending the lysosomal genetic architecture beyond the established role of GBA1.

Among the genes analyzed, ST3GAL3, ST3GAL4, and ST3GAL5 showed evidence of rare‐variant burden, implicating sialylation pathways in PD susceptibility. These three sialyltransferases have complementary roles in ganglioside and glycoprotein biosynthesis. ³⁰ ST3GAL3 encodes ST3Gal‐III, a sialyltransferase that regulates terminal sialylation of glycoproteins and glycolipids, ³¹ including major brain gangliosides GD1a and GT1b essential for neuronal function. ³² Biallelic pathogenic variants in ST3GAL3 are associated with developmental and epileptic encephalopathy‐15 (DEE15). ³³ ST3GAL5 encodes GM3 synthase, catalyzing the first step in ganglioside biosynthesis, and pathogenic variants result in profound neurodevelopmental disorder. ³⁴ Reduced GM1 and related gangliosides have been documented in PD substantia nigra, ³⁵ and GM1 directly modulates α‐synuclein aggregation. ³⁶ Consistent with impaired ganglioside metabolism, plasma lipidomic studies have demonstrated elevated GM3 ganglioside concentration in PD patients compared to controls. ³⁷ Recent large‐scale exome‐wide rare‐variant burden analyses have implicated B3GNT3, ³⁸ an enzyme that synthesizes poly‐N‐acetyllactosamine chains serving as substrates for ST3GAL‐mediated sialylation, providing independent genetic support for perturbation of this pathway in PD. Together, the concordant signals across ST3GAL3, ST3GAL4, and ST3GAL5 support a model in which altered sialylation affects glycosphingolipid composition and associated with lysosomal dysfunction and PD pathogenesis.

Domain‐based analyses provided additional resolution of genetic signal, identifying an enrichment of nonsynonymous variants within the β‐acetyl‐hexosaminidase–like domain of HEXA. Biallelic pathogenic mutations in HEXA cause Tay‐Sachs disease (GM2 gangliosidosis), which ranges from infantile to juvenile and adult/late‐onset forms. In late‐onset forms, motor neuron disease, cerebellar ataxia, and psychiatric symptoms are common. HEXA plays a central role in lysosomal ganglioside degradation by catalyzing the conversion of GM2 to GM3, ³⁹ functionally opposing the biosynthetic pathway mediated by ST3GAL5 (Fig. 1). Disruption of ganglioside homeostasis has been linked to α‐synuclein aggregation, impaired lysosomal flux, and neurodegeneration. ⁴⁰ Consistent with this model, gangliosides including GM2 and GM3 are elevated across multiple brain regions in GBA1‐associated PD, indicating that lysosomal dysfunction propagates upstream through the ganglioside degradation pathway. ⁴¹ Interestingly, increased HEXA activity in rodent models prevents α‐synuclein lipid associations and protects dopaminergic neurons. ⁴² , ⁴³ In line with impaired lysosomal metabolism, HEX activity is reduced in the substantia nigra in PD. ⁴² Therapeutic strategies targeting ganglioside metabolism may offer disease‐modifying potential across genetic and idiopathic forms of PD.

FIG. 1 — Ganglioside biosynthesis and lysosomal degradation of GM3. Ganglioside biosynthesis is initiated in the late Golgi compartments, where lactosylceramide (LacCer) is converted to GM3 by ST3GAL5 (GM3 synthase). GM3 is transported to the plasma membrane and subsequently delivered to lysosomes via membrane trafficking and endocytosis (dashed arrows). Within lysosomes, ganglioside degradation proceeds in a unidirectional, stepwise manner from more complex to simpler species, including conversion of GM2 to GM3 by β‐hexosaminidase A (HEXA, with GM2 activator protein). GM3 is further degraded to LacCer by neuraminidase 1 (NEU1), followed by conversion of LacCer to glucosylceramide (GlcCer) by β‐galactosidase (GLB1). GlcCer is subsequently degraded to ceramide by glucocerebrosidase (GCase). This figure was created with BioRender.com. [Color figure can be viewed at wileyonlinelibrary.com]

Additional nominal associations were observed in genes involved in ceramide and sphingosine‐1‐phosphate handling, including ASAH1, SGPP1, and CERK. Ceramide and sphingolipid balance are key determinants of lysosomal membrane integrity, autophagy initiation, and cellular stress responses. ⁴⁴ Disruption of these pathways has been shown to impair α‐synuclein clearance and exacerbate neurotoxicity in experimental models, supporting the biological relevance of these findings. ⁴⁴ CTSD is one of the lysosomal proteases and may degrade α‐synuclein. ⁴⁵ , ⁴⁶

Our findings extend previous observations that PD is characterized by an aggregate lysosomal genetic burden independent of GBA1 status, while also illustrating the challenges of identifying individual lysosomal genes with modest signal. ⁸ , ¹⁰ , ¹¹ , ¹² Several genes highlighted here, including CTSD, ASAH1, and GRN, have previously been implicated in PD or related neurodegenerative phenotypes, ⁸ providing additional support to their potential involvement in the disease pathophysiology. However, most associations identified in this study were nominal and did not survive stringent correction for multiple testing. This pattern is consistent with prior rare‐variant burden studies in PD, in which GBA1 represents a notable outlier with a comparatively large effect, whereas other lysosomal genes appear to confer smaller increments of risk that require substantially larger sample sizes to detect reliably.

It is also notable that several lysosomal genes previously implicated in PD were not significantly associated in the present analysis. This likely reflects inherent limitations of rare‐variant burden testing, which is best powered to detect dominant or semi‐dominant effects and may be less sensitive to recessive mechanisms or ultra‐rare variants. In addition, pathogenic variants in some lysosomal genes are exceedingly rare or associated with phenotypes that differ from typical late‐onset PD, such as earlier age at onset or distinct clinical features. As the majority of cases included in this study represent the general PD population with later onset, carriers of highly penetrant recessive mutations may be underrepresented.

Analysis of early‐onset PD identified strong domain‐based associations in NAGLU and ST3GAL5. These findings warrant cautious interpretation, particularly given the small number of carriers and the ancestry heterogeneity across cohorts. Replication is required in larger and more ancestrally diverse sequencing studies to refine effect estimates and identify ancestry‐specific risk alleles.

Several limitations should be acknowledged. First, this study integrates data generated using different sequencing platforms and capture strategies, necessitating dataset‐specific quality control thresholds that may influence rare‐variant detection sensitivity despite careful QC. Second, although most cohorts, including UKBB, AMP‐PD, McGill, and Pavlov and Human Brain, were primarily of European ancestry, the Columbia cohort was more heterogeneous and included mixed ancestry backgrounds, whereas the Sheba cohort included both Ashkenazi Jewish and non‐Ashkenazi Jewish participants. The ancestry composition of these cohorts, including the overrepresentation of Ashkenazi Jewish participants, may affect the generalizability of specific rare‐variant findings. Third, kernel‐based burden tests do not provide information on the directionality of individual variants, limiting interpretation of whether specific alleles confer risk or protection. Fourth, our analysis focused on single‐nucleotide variants and small indels and did not include copy number variants, and future studies designed to detect copy number and structural variants will be needed to assess their contribution to PD risk. Finally, functional validation was beyond the scope of this study and additional work is needed to determine whether and how the identified variants influence lysosomal dysfunction and PD pathogenesis.

In conclusion, these data provide further genetic evidence that rare variants across multiple lysosomal pathways contribute to PD susceptibility, implicating sialylation, ganglioside metabolism, ceramide biology, and lysosomal proteolysis. The convergence of genetic associations in biosynthetic (ST3GAL3, ST3GAL4, and ST3GAL5) and degradative (HEXA) arms of the ganglioside pathway is particularly compelling. Although GBA1 remains the strongest lysosomal risk gene, several additional lysosomal genes previously implicated in smaller rare‐variant burden studies ⁸ remained nominally associated in the present analysis despite a substantial increase in sample size. Although these signals did not survive correction for multiple testing, their consistency across studies suggests reproducible but modest effects. This pattern likely reflects both the small effect sizes of non‐GBA1 lysosomal genes and methodological constraints of kernel‐based burden tests that aggregate variants by class. Larger, ancestrally diverse cohorts and complementary functional studies will be required to refine effect estimates and clarify the contribution of these genes to PD pathogenesis.

Author Roles

(1) Research project: A. Conception, B. Organization, C. Execution; (2) Statistical Analysis: A. Design, B. Execution, C. Review and critique; (3) Manuscript Preparation: A. Writing of the First Draft, B. Review and Critique.

K.S.: 1A, 1B, 1C, 2A, 2B, 3A

S.C.P.: 1C, 2B, 2C, 3B

C.C.: 1C, 2B, 2C, 3B

L.L.: 1C, 2B, 2C, 3B

E.Y.: 1C, 2B, 2C, 3B

U.R.: 1C, 2C, 3B

J.A.: 1C, 2C, 3B

J.A.R.: 1C, 2C, 3B

F.A.: 1C, 2C, 3B

D.S.: 1C, 2C, 3B

C.W.: 1B, 2C, 3B

O.M.: 1B, 2C, 3B

Y.D.: 1B, 2C, 3B

N.D.: 1B, 2C, 3B

L.G.: 1B, 2C, 3B

S.H.B.: 1B, 2C, 3B

I.M.: 1B, 2C, 3B

A.T.: 1B, 2C, 3B

A.E.: 1B, 2C, 3B

S.P.: 1B, 2C, 3B

R.N.A.: 1B, 1C, 2C, 3B

Z.G.O.: 1A, 1B, 2A, 2C, 3A, 3B

Supporting information

Table S1. Lysosomal genes included in the targeted sequencing panel and analyzed in this study.

Table S2. Rare coding variants in studied cohorts.

Table S3. Cohort‐specific SKAT‐O rare‐variant burden results.

Table S4. Meta‐analysis of SKAT‐O rare‐variant burden results across all cohorts.

Table S5. Meta‐analysis of SKAT‐O rare‐variant burden results in early‐onset Parkinson's disease.

Table S6. Cohort‐specific SKAT‐O rare‐variant burden results in early‐onset Parkinson's disease.

Table S7. Cohort‐specific domain‐based SKAT‐O rare‐variant burden results.

Table S8. Domain‐based SKAT‐O rare‐variant burden results across all cohorts.

Table S9. Meta‐analysis of domain‐based SKAT‐O rare‐variant burden results in early‐onset Parkinson's disease.

Table S10. Cohort‐specific domain‐based SKAT‐O rare‐variant burden results in early‐onset Parkinson's disease.

MDS-41-2349-s001.xlsx^{(1.3MB, xlsx)}

Acknowledgments

We thank the participants in the different cohorts for contributing to this study. S.C.P. is supported by a Canadian Institutes of Health Research (CIHR) Canada Graduate Scholarship‐Doctoral (CGS‐D) award. Z.G.O. is supported by the Fonds de recherche du Québec—Santé (FRQS) Chercheurs‐boursiers award, in collaboration with Parkinson Quebec, and is a William Dawson Scholar. The access to part of the participants for this research has been made possible thanks to the Quebec Parkinson's Network (http://rpq-qpn.ca/en/). This research used the NeuroHub infrastructure and was undertaken thanks in part to funding from the Canada First Research Excellence Fund, awarded through the Healthy Brains, Healthy Lives initiative at McGill University. This research was enabled in part by support provided by Calcul Québec and the Digital Research Alliance of Canada. Data used in the preparation of this article were obtained from the AMP PD Knowledge Platform. For up‐to‐date information on the study, visit https://www.amp-pd.org. Data for this article were also obtained in January 2023 from the AMP‐PD Knowledge Platform as per release 2.5. Since this release more recent versions have been made available. The AMP PD program is a public‐private partnership managed by the Foundation for the National Institutes of Health and funded by the National Institute of Neurological Disorders and Stroke (NINDS) in partnership with the Aligning Science Across Parkinson's (ASAP) initiative; Celgene Corporation, a subsidiary of Bristol‐Myers Squibb Company; GlaxoSmithKline plc (GSK); The Michael J. Fox Foundation for Parkinson's Research; Pfizer; AbbVie; Sanofi US Services; and Verily Life Sciences. ACCELERATING MEDICINES PARTNERSHIP and AMP are registered service marks of the US Department of Health and Human Services. Clinical data and biosamples used in preparation of this article were obtained from (1) The Michael J. Fox Foundation for Parkinson's Research (MJFF) and NINDS BioFIND study; (2) Harvard Biomarkers Study (HBS) and the Stephen and Denise Adams Center for Parkinson's Disease Research of Yale School of Medicine (CPDR‐Y); (3) National Institute on Aging (NIA) International Lewy Body Dementia Genetics Consortium Genome Sequencing in Lewy Body Dementia Case–control Cohort (LBD); (4) MJFF LRRK2 Cohort Consortium (LCC); (5) NINDS Parkinson's Disease Biomarkers Program (PDBP); (6) MJFF Parkinson's Progression Markers Initiative (PPMI); (7) NINDS Study of Isradipine as a Disease‐modifying Agent in Subjects With Early Parkinson Disease, Phase 3 (STEADY‐PD3) and (8) the NINDS Study of Urate Elevation in Parkinson's Disease, Phase 3 (SURE‐PD3). BioFIND is sponsored by MJFF with support from the NINDS. The BioFIND Investigators have not participated in reviewing the data analysis or content of the manuscript. For up‐to‐date information on the study, visit michaeljfox.org/news/biofind. Genome sequence data for the Lewy body dementia case–control cohort were generated at the Intramural Research Program of the US NIH. The study was supported in part by the NIA (program: 1ZIAAG000935) and the NINDS (program: 1ZIANS003154). HBS is a collaboration of HBS investigators (full list of HBS investigators found at https://www.bwhparkinsoncenter.org/biobank/ and funded through philanthropy and NIH and non‐NIH funding sources. The Stephen and Denise Adams Center for Parkinson's Disease Research of Yale School of Medicine is funded through philanthropy and NIH and non‐NIH funding sources. The HBS and CPDR‐Y Investigators have not participated in reviewing the data analysis or content of the manuscript. Data used in preparation of this article were obtained from MJFF sponsored LCC. The LCC Investigators have not participated in reviewing the data analysis or content of the manuscript. For up‐to‐date information on the study, visit https://www.michaeljfox.org/biospecimens]. PPMI is sponsored by MJFF and supported by a consortium of scientific partners (list the full names of all of the PPMI funding partners found at https://www.ppmi-info.org/about-ppmi/who-we-are/study-sponsors). The PPMI investigators have not participated in reviewing the data analysis or content of the manuscript. For up‐to‐date information on the study, visit www.ppmi-info.org. The PDBP consortium is supported by the NINDS at the NIH. A full list of PDBP investigators can be found at https://pdbp.ninds.nih.gov/policy. The PDBP investigators have not participated in reviewing the data analysis or content of the manuscript. STEADY‐PD3 is funded by NINDS at NIH with support from MJFF. For additional study information, visit https://clinicaltrials.gov/ct2/show/study/NCT02168842. The STEADY‐PD3 investigators have not participated in reviewing the data analysis or content of the manuscript. SURE‐PD3 is funded by NINDS at NIH with support from MJFF. For additional study information, visit https://clinicaltrials.gov/ct2/show/NCT02642393. The SURE‐PD3 investigators have not participated in reviewing the data analysis or content of the manuscript. Figure 1 was created with BioRender.com.

Financial Disclosures and Conflicts of Interest: K.S. received consultancy fees from Acurex, Z.G.O. received consultancy fees from Lysosomal Therapeutics (LTI), Idorsia, Prevail Therapeutics, Ono Therapeutics, Denali, Handl Therapeutics, Neuron23, Bial Biotech, Bial, UCB, Capsida, Vanqua bio, Simcere, Jazz therapeutics, Takeda, EG427, Gain Therapeutics, Guidepoint, Lighthouse, and Deerfield. Other authors declare no conflicts of interest.

Funding agencies: This work was financially supported by grants from The Michael J. Fox Foundation, the Canadian Consortium on Neurodegeneration in Aging (CCNA), the Canada First Research Excellence Fund (CFREF), awarded to McGill University for the Healthy Brains for Healthy Lives initiative (HBHL), and Parkinson Canada. The Columbia University cohort is supported by the Parkinson's Foundation, the National Institutes of Health (K02NS080915 and UL1 TR000040) and the Brookdale Foundation. We also acknowledge contributions from the G‐Can (GBA1‐Canada) Initiative, an open‐science collaboration dedicated to GBA1‐associated neurodegeneration. G‐Can is supported by the Galen and Hilary Weston Foundation, the Silverstein Foundation, and J. Sebastian van Berkom and Ghislaine Saucier.

Data Availability Statement

All supplementary materials, including additional tables and figures, are available in the Supplementary Materials section. The McGill cohorts are partially available through the Canadian Open Parkinson Network (C‐OPN). Access, including genetic data, can be requested through the C‐OPN data access committee (https://copn-rpco.ca/submit-a-request/). The AMP‐PD data was assessed using the Terra platform, https://amp-pd.org/. The UKBB was acquired using Whole Genome Sequencing data from the UK Biobank Research Analysis Platform (https://www.ukbiobank.ac.uk/) under application number 45551.

References

1. Senkevich K, Gan‐Or Z. Autophagy lysosomal pathway dysfunction in Parkinson's disease; evidence from human genetics. Parkinsonism Relat Disord 2020;73:60–71. [DOI] [PubMed] [Google Scholar]
2. Gan‐Or Z, Amshalom I, Kilarski LL, et al. Differential effects of severe vs mild GBA mutations on Parkinson disease. Neurology 2015;84(9):880–887. [DOI] [PMC free article] [PubMed] [Google Scholar]
3. Sidransky E, Nalls MA, Aasly JO, et al. Multicenter analysis of glucocerebrosidase mutations in Parkinson's disease. N Engl J Med 2009;361(17):1651–1661. [DOI] [PMC free article] [PubMed] [Google Scholar]
4. Horowitz M, Braunstein H, Zimran A, Revel‐Vilk S, Goker‐Alpan O. Lysosomal functions and dysfunctions: molecular and cellular mechanisms underlying Gaucher disease and its association with Parkinson disease. Adv Drug Deliv Rev 2022;187:114402. [DOI] [PubMed] [Google Scholar]
5. Senkevich K, Zorca CE, Dworkind A, et al. GALC variants affect galactosylceramidase enzymatic activity and risk of Parkinson's disease. Brain 2022;146:1859–1872. [DOI] [PMC free article] [PubMed] [Google Scholar]
6. Senkevich K, Beletskaia M, Dworkind A, et al. Association of Rare Variants in ARSA with Parkinson's disease. Mov Disord 2023;38(10):1806–1812. [DOI] [PMC free article] [PubMed] [Google Scholar]
7. Jones‐Tabah J, He K, Karpilovsky N, et al. The Parkinson's disease risk gene cathepsin B promotes fibrillar alpha‐synuclein clearance, lysosomal function and glucocerebrosidase activity in dopaminergic neurons. Mol Neurodegener 2024;19(1):88. [DOI] [PMC free article] [PubMed] [Google Scholar]
8. Robak LA, Jansen IE, van Rooij J, et al. Excessive burden of lysosomal storage disorder gene variants in Parkinson's disease. Brain 2017;140(12):3191–3203. [DOI] [PMC free article] [PubMed] [Google Scholar]
9. Chew EGY, Liu Z, Li Z, et al. Exome sequencing in Asian populations identifies low‐frequency and rare coding variation influencing Parkinson's disease risk. Nat Aging 2025;5(2):205–218. [DOI] [PMC free article] [PubMed] [Google Scholar]
10. Hopfner F, Mueller SH, Szymczak S, et al. Rare variants in specific lysosomal genes are associated with Parkinson's disease. Mov Disord 2020;35(7):1245–1248. [DOI] [PubMed] [Google Scholar]
11. Zhao YW, Pan HX, Liu Z, et al. The association between lysosomal storage disorder genes and Parkinson's disease: a large cohort study in Chinese mainland population. Front Aging Neurosci 2021;13:749109. [DOI] [PMC free article] [PubMed] [Google Scholar]
12. Tunold JA, Tan MMX, Koga S, et al. Lysosomal polygenic risk is associated with the severity of neuropathology in Lewy body disease. Brain 2023;146(10):4077–4087. [DOI] [PMC free article] [PubMed] [Google Scholar]
13. Morris HR, Spillantini MG, Sue CM, Williams‐Gray CH. The pathogenesis of Parkinson's disease. Lancet 2024;403(10423):293–304. [DOI] [PubMed] [Google Scholar]
14. Pan T, Kondo S, Le W, Jankovic J. The role of autophagy‐lysosome pathway in neurodegeneration associated with Parkinson's disease. Brain 2008;131(8):1969–1978. [DOI] [PubMed] [Google Scholar]
15. Tremblay ME, Cookson MR, Civiero L. Glial phagocytic clearance in Parkinson's disease. Mol Neurodegener 2019;14(1):16. [DOI] [PMC free article] [PubMed] [Google Scholar]
16. Mazzulli JR, Xu YH, Sun Y, et al. Gaucher disease glucocerebrosidase and α‐synuclein form a bidirectional pathogenic loop in synucleinopathies. Cell 2011;146(1):37–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
17. Baden P, Perez MJ, Raji H, et al. Glucocerebrosidase is imported into mitochondria and preserves complex I integrity and energy metabolism. Nat Commun 2023;14(1):1930. [DOI] [PMC free article] [PubMed] [Google Scholar]
18. Fan Y, Mao C‐y, Dong Y‐l, et al. ARSA gene variants and Parkinson's disease. Brain 2020;143(6):e47–e47. [DOI] [PubMed] [Google Scholar]
19. Gan‐Or Z, Rao T, Leveille E, et al. The Quebec Parkinson network: a researcher‐patient matching platform and multimodal biorepository. J Parkinsons Dis 2020;10(1):301–313. [DOI] [PMC free article] [PubMed] [Google Scholar]
20. Alcalay RN, Levy OA, Wolf P, et al. SCARB2 variants and glucocerebrosidase activity in Parkinson's disease. NPJ Parkinsons Dis 2016;2:2. [DOI] [PMC free article] [PubMed] [Google Scholar]
21. Ruskey JA, Greenbaum L, Roncière L, et al. Increased yield of full GBA sequencing in Ashkenazi Jews with Parkinson's disease. Eur J Med Genet 2019;62(1):65–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
22. Hughes AJ, Ben‐Shlomo Y, Daniel SE, Lees AJ. What features improve the accuracy of clinical diagnosis in Parkinson's disease. A clinicopathologic study. Neurology 1992;42(6):1142. [DOI] [PubMed] [Google Scholar]
23. Postuma RB, Berg D, Stern M, et al. MDS clinical diagnostic criteria for Parkinson's disease. Mov Disord 2015;30(12):1591–1601. [DOI] [PubMed] [Google Scholar]
24. Krohn L, Ruskey JA, Rudakou U, et al. GBA variants in REM sleep behavior disorder: a multicenter study. Neurology 2020; 95(8):e1008‐e1016. [DOI] [PMC free article] [PubMed] [Google Scholar]
25. Rudakou U, Yu E, Krohn L, et al. Targeted sequencing of Parkinson's disease loci genes highlights SYT11, FGF20 and other associations. Brain 2021;144(2):462–472. [DOI] [PMC free article] [PubMed] [Google Scholar]
26. Zhao H, Sun Z, Wang J, Huang H, Kocher JP, Wang L. CrossMap: a versatile tool for coordinate conversion between genome assemblies. Bioinformatics 2014;30(7):1006–1007. [DOI] [PMC free article] [PubMed] [Google Scholar]
27. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high‐throughput sequencing data. Nucleic Acids Res 2010;38(16):e164. [DOI] [PMC free article] [PubMed] [Google Scholar]
28. Lee S, Emond MJ, Bamshad MJ, et al. Optimal unified approach for rare‐variant association testing with application to small‐sample case‐control whole‐exome sequencing studies. Am J Hum Genet 2012;91(2):224–237. [DOI] [PMC free article] [PubMed] [Google Scholar]
29. Derkach A, Zhang H, Chatterjee N. Power analysis for genetic association test (PAGEANT) provides insights to challenges for rare variant association studies. Bioinformatics 2018;34(9):1506–1513. [DOI] [PMC free article] [PubMed] [Google Scholar]
30. Qi F, Isaji T, Duan C, Yang J, Wang Y, Fukuda T, Gu J. ST3GAL3, ST3GAL4, and ST3GAL6 differ in their regulation of biological functions via the specificities for the α2,3‐sialylation of target proteins. FASEB J 2020;34(1):881–897. [DOI] [PubMed] [Google Scholar]
31. Hu H, Eggers K, Chen W, et al. ST3GAL3 mutations impair the development of higher cognitive functions. Am J Hum Genet 2011;89(3):407–414. [DOI] [PMC free article] [PubMed] [Google Scholar]
32. Yoo SW, Motari MG, Susuki K, Prendergast J, Mountney A, Hurtado A, Schnaar RL. Sialylation regulates brain structure and function. FASEB J 2015;29(7):3040–3053. [DOI] [PMC free article] [PubMed] [Google Scholar]
33. Edvardson S, Baumann AM, Mühlenhoff M, et al. West syndrome caused by ST3Gal‐III deficiency. Epilepsia 2013;54(2):e24–e27. [DOI] [PubMed] [Google Scholar]
34. Inamori KI, Inokuchi JI. When ganglioside pathways go awry: congenital disorders and experimental insights. J Hum Genet 2025; [DOI] [PubMed] [Google Scholar]
35. Schneider JS. Altered expression of genes involved in ganglioside biosynthesis in substantia nigra neurons in Parkinson's disease. PLoS One 2018;13(6):e0199189. [DOI] [PMC free article] [PubMed] [Google Scholar]
36. Kumar R, Chowdhury S, Ledeen R. Alpha‐synuclein and GM1 ganglioside Co‐localize in neuronal cytosol leading to inverse interaction‐relevance to Parkinson's disease. Int J Mol Sci 2024;25(6):3323. [DOI] [PMC free article] [PubMed] [Google Scholar]
37. Chan RB, Perotte AJ, Zhou B, et al. Elevated GM3 plasma concentration in idiopathic Parkinson's disease: a lipidomic analysis. PLoS One 2017;12(2):e0172348. [DOI] [PMC free article] [PubMed] [Google Scholar]
38. Makarious MB, Lake J, Pitz V, et al. Large‐scale rare variant burden testing in Parkinson's disease. Brain 2023;146(11):4622–4632. [DOI] [PMC free article] [PubMed] [Google Scholar]
39. Guerra J, Belleri M, Scalvini E, et al. Brain accumulation of lactosylceramide characterizes GALC deficiency in a zebrafish model of Krabbe disease. Brain 2025;148(10):3748–3762. [DOI] [PMC free article] [PubMed] [Google Scholar]
40. Lemieux MJ, Mark BL, Cherney MM, Withers SG, Mahuran DJ, James MN. Crystallographic structure of human beta‐hexosaminidase a: interpretation of Tay‐Sachs mutations and loss of GM2 ganglioside hydrolysis. J Mol Biol 2006;359(4):913–929. [DOI] [PMC free article] [PubMed] [Google Scholar]
41. Blumenreich S, Nehushtan T, Barav OB, et al. Elevation of gangliosides in four brain regions from Parkinson's disease patients with a GBA mutation. NPJ Parkinsons Dis 2022;8(1):99. [DOI] [PMC free article] [PubMed] [Google Scholar]
42. Brekk OR, Korecka JA, Crapart CC, et al. Upregulating β‐hexosaminidase activity in rodents prevents α‐synuclein lipid associations and protects dopaminergic neurons from α‐synuclein‐mediated neurotoxicity. Acta Neuropathol Commun 2020;8(1):127. [DOI] [PMC free article] [PubMed] [Google Scholar]
43. Suzuki K, Yamaguchi A, Yamanaka S, et al. Accumulated α‐synuclein affects the progression of GM2 gangliosidoses. Exp Neurol 2016;284(Pt A):38–49. [DOI] [PubMed] [Google Scholar]
44. Vos M, Klein C, Hicks AA. Role of ceramides and sphingolipids in Parkinson's disease. J Mol Biol 2023;435(12):168000. [DOI] [PubMed] [Google Scholar]
45. Prieto Huarcaya S, Drobny A, Marques ARA, et al. Recombinant pro‐CTSD (cathepsin D) enhances SNCA/α‐synuclein degradation in α‐Synucleinopathy models. Autophagy 2022;18(5):1127–1151. [DOI] [PMC free article] [PubMed] [Google Scholar]
46. Yang SY, Gegg M, Chau D, Schapira A. Glucocerebrosidase activity, cathepsin D and monomeric α‐synuclein interactions in a stem cell derived neuronal model of a PD associated GBA1 mutation. Neurobiol Dis 2020;134:104620. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials