Biotin mitigates the development of manganese-induced, Parkinson’s disease–related neurotoxicity in Drosophila and human neurons

Abstract

Chronic exposure to manganese (Mn) induces manganism and has been widely implicated as a contributing environmental factor to Parkinson’s disease (PD), featuring notable overlaps between the two in motor symptoms and clinical hallmarks. Here, we developed an adult Drosophila model of Mn toxicity that recapitulated key Parkinsonian features, spanning behavioral deficits, neuronal loss, and dysfunctions in lysosomes and mitochondria. Metabolomics analysis of the brain and body tissues of these flies at an early stage of toxicity identified systemic changes in the metabolism of biotin (also known as vitamin B₇) in Mn-treated groups. Biotinidase-deficient flies showed exacerbated Mn-induced neurotoxicity, Parkinsonism, and mitochondrial dysfunction. Supplementing the diet of wild-type flies with biotin ameliorated the pathological phenotypes of concurrent exposure to Mn. Biotin supplementation also ameliorated the pathological phenotypes of 3 standard fly models of PD. Furthermore, supplementing the culture media of human induced stem cells (iPSCs)-differentiated midbrain dopaminergic neurons with biotin protected against Mn-induced mitochondrial dysregulation, cytotoxicity, and neuronal loss. Finally, analysis of the expression of genes encoding biotin-related proteins in PD patients revealed increased amounts of biotin transporters in the substantia nigra compared to healthy controls, suggesting a potential role of altered biotin metabolism in PD. Together, our findings identified changes in biotin metabolism as underlying Mn neurotoxicity and Parkinsonian pathology in flies, for which dietary biotin supplementation was preventative.

Introduction

The world is rapidly aging; it is estimated that by 2050, the number of people aged 65 or over will more than double, reaching 1.6 billion which accounts for over 15% of the world’s population (1). These 1.6 billion elderly people will be at substantial risk for brain disease and disorders such as Parkinson’s disease (PD) and Alzheimer’s disease (AD), posing a massive burden on care systems and society due to the long, drawn-out course of disease (1, 2). In the case of treating PD, the most common motor neurodegenerative disease, there are currently no disease-modifying therapies, and for many decades, levodopa (L-dopa) has been the only front-line, gold-standard drug since its approval by the U.S. FDA in the 1970s (3). The etiology of PD remains elusive and can be diverse; only 10% of PD cases can be explained through Mendelian genetics, suggesting that environmental exposures play a key role in the pathogenesis of sporadic PD (4).

Manganese (Mn) is an essential micro-nutrient, but chronic exposure to high amounts of it, such as is experienced occupationally by welders, is associated with multiple key motor symptoms of PD. One major pathological hallmark of the disease is aggregation of the protein α-synuclein in neurons—a toxic condition that leads to dysregulated cellular processes and eventually massive neuronal cell death (3). Harischandra et al. found that extracellular vesicles called exosomes isolated from Mn-exposed welders’ blood sera contained misfolded α-synuclein. In cell culture and rodent models, exposure to Mn or to isolated, Mn-induced exosomes promoted the transfer of α-synuclein between neurons and microglia, leading to inflammation and eventually to neuronal cell death (5–7). These α-synuclein–related findings support a causal role of Mn in neurodegenerative disease. Mn has also been linked to the induction of neuroinflammation, mitochondrial dysfunction, autophagic defects, and protein aggregation, which all are cellular hallmarks of PD (8). However, there is a dearth of knowledge on the molecular and mechanistic levels regarding how Mn leads to Parkinsonism.

Decades of research interrogating individual genes and the molecular basis for PD have led to dozens of ongoing clinical trials aiming to slow down or halt disease progression. However, none thus far have resulted in any effective preventative therapy (9). In this study, we took an untargeted approach to identify the early compensatory molecular mechanisms potentially involved in PD pathophysiology. We performed high-coverage, high-resolution mass spectrometry-based metabolomics analyses on an adult fly model of Mn-induced Parkinsonism to tease out the early in vivo metabolic changes in fly brains and systemically in fly bodies (abdomens and thoraxes) before key parkinsonian phenotypes surfaced. We identified B vitamin biotin (vitamin B₇) metabolism as a key pathway altered early by Mn exposure, and showed that supplementing with biotin rescued Mn-induced neurotoxicity, including mitochondrial dysfunction, both in flies and in cultured human iPSC-derived dopaminergic neurons. Together, these findings identify the compensatory biotin pathway as a systemic, convergent mechanism toward resolving the Mn-PD link, providing a new basis for developing biotin-based therapies to combat neurodegeneration and environmental health risks.

Results

An adult Drosophila model of manganese toxicity recapitulates key Parkinsonian features.

Occupational exposure to manganese has been widely linked as one of the critical environmental factors leading to PD. Although various Drosophila models of Mn toxicity have been proposed, most previous models are developmental (10). Because occupational exposure mainly occurs in adults, we developed and characterized an adult model of Mn toxicity in Drosophila melanogaster. In addition, we used male flies for our studies because males are predisposed to developing PD (11), and Mn-exposed occupations such as welding and mining are dominated by males. To explore hypotheses of gut-brain interactions in Mn-induced toxicity and Parkinsonism, we employed a feeding model to ensure direct Mn exposure through the gut. Flies were fed MnCl₂ solution ad libitum at doses of 1, 10, and 30 mM immediately after eclosion (Fig. 1A). Adult exposure to Mn leads to a dose-dependent decrease in lifespan (Fig. 1B). In flies post-ten days of Mn exposure, behavioral assays revealed that Mn impaired climbing ability (Fig. 1C) and exacerbated locomotor deficits (Fig. 1D). In addition, histological and chemical analyses at the same time point of the anterior medulla showed that Mn also led to TH neuronal loss (Fig. 1E) and decreases in dopamine levels (Fig. 1F). Previous studies by our group and others have shown that Mn induces lysosomal and mitochondrial dysfunction in cell culture and rodent models (6, 7, 12, 13).

Figure 1. Experimental workflow and the adult Drosophila model of Mn toxicity to recapitulate Parkinsonian hallmarks.

(A) The experimental workflow consists of four modular steps, including exposure scheme, characterization of Parkinson’s phenotypes, metabolomics analysis, and multi-strategy rescue trials. HSS, high-strength silica; t-SIM, targeted selected ion monitoring; m/z, mass-to-charge ratio; MoNA, MassBank of North America; H&E, hematoxylin and eosin; HPLC, high-performance liquid chromatography. (B) The lifespan of Drosophila melanogaster under Mn exposure; 60 flies were analyzed for each group. (C and D) Mn-induced behavioral changes in (C) climbing and (D) locomotor activities in fruit flies. Four vials of flies per group with each vial containing 10 flies were used for analysis. (E and F) Histological and HPLC results of the anterior medulla at the same time point of Mn exposure for tyrosine hydroxylase (TH) neuronal pathologies and dopamine changes in the fly brain. N=3–4 flies per group from each of 2 independent experiments. (G and H) The Seahorse metabolic flux assay on live fly brains to determine Mn-driven shifts in the mitochondrial metabolic phenotype charting the basal respiration oxygen consumption rate (OCR, x-axis) and extracellular acidification rate (ECAR, y-axis), a measure of lactate formed from glycolytic activities. N=3–4 flies per group from each of 2 independent experiments. (I to L) The MitoTracker assay on live Drosophila brains to visualize Mn-induced mitochondrial morphological changes by perimeter, solidity, and circularity. N=3 images from each of 3 flies per group. Scale bar represents 5 μm. In (B – L), data are mean ± SD; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, and ^nsnot significant, by ordinary one-way analysis of variance (ANOVA) with Dunnett’s multiple comparisons test.

To validate if the cellular mechanisms of Mn toxicity are consistent in our fly model, we performed the LysoTracker assay on live fly brains. Super-resolution microscopy revealed that Mn led to an increase in lysosomal numbers (fig. S1A) and size (fig. S1B), indicating leakage and dysfunction of lysosomes. Further, the Seahorse assay on Drosophila brains showed that Mn reduced the oxygen consumption rates (OCR) and drove the cellular metabolic phenotype from energic to quiescent (Fig. 1, G and H). The MitoTracker assay further demonstrated that Mn led to morphological defects in mitochondria in a dose-dependent manner, as characterized by reduced mitochondrial perimeter and increases in mitochondrial solidity and circularity, which indicated augmented mitochondrial fission (Fig. 1, I to L).

To address whether our model is relevant to Mn exposure in humans, we referenced the Occupational Safety and Health Administration (OSHA) safety guidelines and performed dosimetry measurements of Mn exposure profiles in flies. Specifically, the OSHA safety limit of Mn is 5 milligrams per cubic meter of air (5 mg/m³) for Mn compounds; by unit conversion, this is approximately 1 ppm. All previous models in Drosophila melanogaster have used the developmental models of Mn toxicity. Here, to confirm the Mn burden in the adult brain, we performed metal quantitation of the heads of adult flies exposed to Mn for 5 days using inductively coupled plasma mass spectrometry (ICP-MS). The ICP-MS determined that the flies exposed to Mn have ~1.2 ng/mg per fly head which is roughly 1.2 ppm (fig. S1D). This is slightly above the OSHA limits but is relevant to human exposure, given that human exposure would be longer. Therefore, this semi-acute model can be used for rapid screening in the future.

High-coverage global metabolomics analysis in vivo identifies changes in biotin metabolism in Mn toxicity.

To identify early biochemical modulators of Mn toxicity, we performed high-coverage untargeted metabolomics analyses of the brain and beheaded body of 5-day-old flies exposed to Mn at a low dose (10 mM) and high dose (30 mM) using high-resolution mass spectrometry and cheminformatic computing (Fig. 1A and Fig. 2A). The rationale for using an earlier time point was to capture the basic biochemical changes that potentially occurred before the behavioral deficits manifested and to identify signature metabolites and/or pathways potentially with early roles in, or in response to, Mn toxicity. Overall, our untargeted analysis identified distinct metabolome-wide changes in response to Mn exposure, as shown in partial least squares discriminant analysis (PLS-DA) (Fig. 2B) and a volcano plot (Fig. 2C). With high confidence annotation (level 2 or higher), we resolved 270 and 420 altered metabolites, respectively for brain and systemically for body compartments (Fig. 2, A and B; data files S1 and S2).

Figure 2. Comparative metabolomics analysis in vivo demonstrates substantial whole-body response to Mn exposure.

(A) Overview of metabolomics results of head and body compartments of Drosophila melanogaster under Mn exposure. (B) Partial least squares discriminant analysis (PLS-DA) of the head metabolomic data for low-dose and high-dose Mn–exposed groups vs. the vehicle-exposed group acquired under ESI(+) mode. Ten flies were pooled for each sample, and 5 such pooled samples were used to run metabolomics. . (C) Volcano plot of significant ion features as determined by pairwise comparison of high-dose Mn (30 mM) vs. vehicle; by Welch’s t-test, p < 0.05 and fold change (FC) ≥ 1.2. (D) Pathway analysis of the fly head metabolome after high-dose Mn (30 mM) vs. vehicle exposure, based on the KEGG dme pathway library (for Drosophila melanogaster) using the Globaltest approach for pathway enrichment and relative betweenness centrality for the node importance measure in pathway topological analysis. (E) Individual vitamin-B–family metabolites that were altered in the Drosophila head metabolome by high-dose Mn (30 mM) vs. vehicle. Blue bold color highlights two metabolites altered in both fly brain and fly body. Fold-change in Mn/V (Mn treatment vs. vehicle) shown is calculated from the mean values of all five sample replicates (50 flies combined) described in (B) p-values calculated by Welch’s t-test. FMN, flavin mononucleotide. (F) ClassyFire pie chart view of 51 brain–body overlapping metabolites comparing high-dose Mn (30 mM) vs. vehicle; by Welch’s t-test, p < 0.05 and fold change ≥ 1.2. In (B – F), data are derived from the same metabolomics measurements of 150 flies in 5 replicates for each group, wherein each replicate contained 10 fly heads or bodies; (B) involved heads of all three groups, (C) focused on heads of high-dose (30 mM) Mn-exposed vs. vehicle-exposed groups, and (D to F) involved heads and bodies of high-dose (30 mM) Mn-exposed vs. vehicle-exposed groups.

We then focused on high-dose Mn (30 mM) vs. vehicle only for an in-depth functional analysis, especially of brain metabolome (Fig. 2, A and C), considering that a consistent pattern of changes was seen in the body for high dose (30 mM) and low dose (10 mM) relative to vehicle (fig. S2). Our chemical similarity enrichment analysis (ChemRICH) first identified a diverse chemical space of Mn perturbation, enriching the altered brain metabolites into 36 chemical classes from moderately nonpolar species such as indoles, succinates, and amino acids towards the more nonpolar terpenes, lysophospholipids, and long-chain fatty acids (fig. S3). Further, quantitative pathway enrichment analysis mapped out a realm of Mn-perturbed pathways, with multiple high-impact ones enriched in B vitamins spanning biotin metabolism, pantothenate and CoA biosynthesis, and vitamin B6 metabolism (Fig. 2D). Among the many individual vitamin B family metabolites in the fly brain (Fig. 2E), biotin exhibited the largest fold change (+2.7 fold), the most statistical significance (the lowest p-value), and a shared trend of change in the fly body (+2.0 fold; data file S3), alongside 50 other metabolites with systemic, brain and body changes (Fig. 2F).

Changes in biotin metabolism drive Mn-induced neurotoxicity, Parkinsonism, and mitochondrial dysfunction.

In light of our metabolomics findings regarding biotin, we undertook experimental studies to test systematically for its functional roles in Mn-related neurodegeneration. Biotin is an essential water-soluble B₇ vitamin and can either be recycled, obtained from diet, or produced by the gut microbiome (14). The biotinylation of carboxylases has been widely implicated in cellular functions, including mitochondrial respiration (15) and neurotransmitter production (16, 17), among others. Most notably, biotin is essential for dopamine production (17). Therefore, we hypothesized that Mn-induced increases in biotin metabolism factors was a compensatory mechanism at an early stage before Parkinsonism manifests. To test this hypothesis, we knocked down the expression of neuronal btnd, an ortholog of human BTD, which encodes biotinidase (Btnd), an enzyme that frees biotin from proteins so it may be used. Following RNA interference (RNAi) of btnd to reduce biotin levels in the neurons, we then exposed the mutant flies to Mn (Fig. 3A). Climbing and locomotor assays (Fig. 3, B and C) both showed that neuronal Btnd knockdown further exacerbated Mn-induced behavioral deficits compared to Mn-exposed wild types. Moreover, the MitoSox assays on live fly brains determined that the diminished levels of free biotin promoted mitochondrial superoxide production in Mn-exposed flies (Fig. 3D). Finally, the Seahorse metabolic assay on Drosophila brains determined that the neuronal knockdown of Btnd lowered mitochondrial oxygen consumption rates (Fig. 3E) and worsened the overall cellular energetic profile (Fig. 3F) in flies fed with Mn.

Figure 3. Rescue trials in vivo demonstrate biotin as an essential agent that protects against Mn-induced damage and Parkinsonism phenotypes in flies.

(A) Schematic of rescue experiment 1 in vivo including Btnd^RNAi mutant male flies (with knockdown of the neuronal btnd gene that encodes biotinidase, or Btnd) exposed to Mn (30 mM), applying to the experiments shown in (B to F). (B and C) Biotin deficiency and Mn-induced behavioral deficits in (B) climbing and (C) locomotor activities in fruit flies. 4 vials of flies per group, with each vial containing 10 flies, were used for analysis. (D) The MitoSox assay on live Drosophila brains to determine the effect of biotin deficiency in mediating Mn-induced mitochondrial superoxide levels. N=3 flies per group in a representative one of 2 independent experiments. (E and F) The Seahorse metabolic flux assay on live Btnd^RNAi and WT fly brains after Mn or vehicle exposure. N=3 (E) and 4 (F) flies per group from a representative 2 independent experiments. (G) Schematic of rescue experiment 2 in vivo assessing the effect of biotin supplementation on wild-type (WT) male flies exposed to Mn (30 mM) with biotin supplementation, applying to the experiments shown in (H to L). (H to L) Assays, N, and analysis as in (B to F) in WT flies fed biotin and exposed to Mn as described in (G). Data are mean ± SD. *p<0.05, ***p<0.001, ****p<0.0001, and ^nsnot significant, by ordinary two-way ANOVA with Šídák’s multiple comparisons test.

To investigate whether the compensating biotin has protective or preventive potential, we supplemented biotin and Mn at the same time into the fly diet and exposed the flies for a week to both Biotin and Mn starting at eclosion. (Fig. 3G). Both the climbing and locomotor assays (Fig. 3, H and I) demonstrated that biotin feeding ameliorated Mn-induced behavioral deficits. The MitoSox assay on Drosophila brains showed that dietary biotin supplementation inhibited mitochondrial superoxide production in Mn-exposed flies, virtually down to the levels of non-Mn groups (Fig. 3J). Functionally, the Seahorse assay on live fly brains revealed that biotin rescued mitochondrial oxygen consumption rates (Fig. 3K) and, partially, the cellular energetic profile (Fig. 3L) in Mn-exposed flies. In parallel, histological analysis determined that whereas a reduction in the free biotin pool (through Btnd knockdown) exacerbated Mn-induced TH neuronal loss and dopamine loss, biotin supplementation protected against it (fig. S4, A and B).

To further confirm whether biotin supplementation altered Mn uptake in the flies, we performed ICP-MS on the fly heads and bodies. We discovered that biotin did not significantly alter Mn intake in either compartment (fig. S4, C and D). Because Mn toxicity leads to TH neuronal loss, we went on to test whether Btnd knockdown specifically in TH neurons can alter Mn-induced toxicity. We knocked down Btnd in dopaminergic neurons specifically, exposed the flies to 30mM Mn for 5 days, and monitored both behavioral and histological phenotypes. TH neuron-specific knockdown of Btnd exacerbated Mn-induced behavioral deficits and mitochondrial dysfunction in flies (fig. S4, E and F), further demonstrating a potential protective role of biotin in TH neurons.

A previous study performed metabolomic analysis on the brain of the A53T α-synuclein mouse model of PD (18). The analysis identified biotin metabolism as being enriched in the A53T mouse brains. This indicates a crucial role of the biotin pathway not only in Mn-induced neuronal damages but in the α-synuclein model of neurotoxicity as well. The data also shows that the biotin pathway is altered not only in flies but also in the mammals. To further test this pathway as related to PD, we performed a behavioral screen on three separate models of Parkinsonism—namely paraquat, rotenone, and α-synuclein—in flies. We fed the model flies with biotin and performed the climbing assay. Biotin was able to rescue behavioral deficits induced by paraquat, rotenone, and expression of human α-synuclein (fig. S4G). This further demonstrated the crucial role of the biotin pathway in other models of PD alongside Mn toxicity. Future studies are required to delve into the mechanism of these models. Together, these data suggested that free, bioavailable biotin plays a crucial preventative role against neurodegeneration induced by Mn or other potential PD-related factors in vivo.

Biotin ameliorates Mn-induced neurotoxicity in iPSC-derived midbrain dopaminergic neurons.

Biotin deficiency has been implicated in mitochondrial dysfunction and neuronal loss (15). To validate our findings with mammalian relevance for its therapeutic potential, we differentiated human induced pluripotent stem cells (iPSCs) into midbrain dopaminergic neurons and aged them for 20 days. Then, we pre-treated these cells with biotin for 24 hours and then exposed them to Mn for 24 additional hours. Super-resolution microscopy detected that biotin supplementation protected profoundly against Mn-induced tyrosine hydroxylase (TH) neurite loss (Fig. 4, A and B) with no discernible effects on the Tuj1⁺TH^- neurons (Fig. 4C). Both the LDH assay (Fig. 4D) and MTS assay (Fig. 4E) determined that biotin inhibited Mn-induced cytotoxicity in these iPSC-derived dopaminergic neurons. Moreover, the MitoTracker green assay and MitoSox assay revealed, respectively, for these neuronal cells that biotin rescued Mn-led mitochondrial mass loss which is indicative of impaired mitochondrial function and biogenesis (Fig. 4F) while lowering mitochondrial superoxide production (Fig. 4G). Finally, to gauge cellular functional protection, we performed the Seahorse assay on biotin-pretreated dopaminergic neurons after Mn exposure (Fig. 4, H and I). Supplemented biotin markedly restored Mn-induced loss of the OCRs in dopaminergic neurons (Fig. 4H) and partially rectified the quiescent, glycolytic phenotype caused by Mn exposure (Fig. 4I).

Figure 4. Biotin prevents manganism phenotypes in human iPSC-derived midbrain dopaminergic neurons, and associated protein expression levels reported for PD patients.

(A to C) Effects of Mn with and without biotin on iPSC-derived midbrain dopaminergic neuronal cell cultures, assessing changes in (A) staining for the dopaminergic neuron marker tyrosine hydroxylase (TH; middle green panel) with counterstaining for neuronal tubulin (Tuj1), (B) TH+ neurite length, and (C) Tuj1⁺TH^– neurite length. Images in (A) are representative of at least 3 images per group from each of 2 independent experiments. Scale bar is 40 μm. In (B and C), N=5 independent experiments. (D and E) Effect of biotin on Mn-induced cytotoxicity in iPSC-derived dopaminergic neurons as indicated by (D) LDH levels and (E) MTS-based cell viability. N=4 independent experiments. (F) MitoTracker assay to determine the effects of biotin supplementation on Mn-induced mitochondrial mass loss in iPSC-derived dopaminergic neurons. N=4 independent experiments. (G and H) Seahorse metabolic flux assay on iPSC-derived neurons to determine the effects of biotin on (G) Mn-induced metabolic phenotypic shifts and (H) Mn-inhibited neuronal mitochondrial OCR. N=4 independent experiments. (I) MitoSox assay to determine the effects of biotin on Mn-induced mitochondrial superoxide levels in iPSC-derived dopaminergic neurons. N=3–4 independent experiments. (J) Human PD patient data on biotin-related protein expression levels (normalized fold change) of biotin transporters and biotin metabolic enzymes. Data were derived from the NCBI Gene Expression Omnibus database. BTD, human biotinidase; MCOS1, molybdenum cofactor synthesis 1; SLC5A6, sodium-dependent multivitamin transporter, or solute carrier family 5 member 6; MCCC1, methylcrotonyl-CoA carboxylase subunit 1; HLCS, holocarboxylase synthetase; SLC19A3, thiamine transporter 2, or solute carrier family 19 member 3. In (B to H), *p<0.05, **p<0.01, ****p<0.0001, and ^nsnot significant, by ordinary two-way ANOVA and post hoc Šídák’s multiple comparisons test. In (B to J), data are mean ± SD.

To illuminate the role of biotin in human PD, we conducted data mining in NCBI Gene Expression Omnibus (GEO) for biotin-related genes and their expressions. We discovered that compared to age-matched healthy controls, PD patients had increased levels of biotin transporters in the SNpc but not biotin-metabolizing enzymes in the brain (Fig. 4J), suggesting that biotin transport is involved in PD pathogenesis. Since human studies are primarily correlational and not causal, to identify if biotin regulation itself (in the absence of Mn) can alter behavior, we knocked down Btnd (biotinidase) and Smvt (the primary transporter for biotin) specifically in fly neurons and performed a behavioral screen for each. Both the Btnd and Smvt knockdowns induced a loss in the climbing ability of flies with age (fig. S5). This aligns with a prior work by Lohr et al. demonstrating that biotin withdrawal leads to neuronal loss in iPSC-derived cortical neurons (15). Together, the findings show that biotin supplementation protected human iPSC-derived dopaminergic neurons from Mn-induced neurotoxicity and mitochondrial dysfunction. Our results in mutant flies further suggested that biotin deficiency may be causally related to Parkinsonism. Future studies linking biotin metabolism dysregulation and PD etiology are warranted.

Discussion

Among age-related brain diseases, Parkinson’s disease is the most common motor disorder (2). Since its first medical documentation in 1817, PD has attracted research interests as a pathfinder disease to gather clues for resolving idiopathic neurodegeneration altogether. However, disease-modifying therapies remain elusive, with environmental risk factors largely unexplored (4). To address this, we performed untargeted metabolomics on an adult Drosophila model of manganese toxicity, identifying biotin metabolism as a top-enriched pathway with systemic biotin increases. Using genetics and pharmacological approaches, we demonstrated in vivo and in vitro that biotin mitigates Mn-induced neurotoxicity through mitochondrial protection. Given PD’s male predisposition and the male-dominated nature of Mn-exposed occupations like welding and mining, we used male flies and focused on dopaminergic endpoints. Future studies should examine sex-specific effects and non-dopaminergic mechanisms. For translational relevance, although our iPSC-derived dopaminergic neurons effectively modeled Mn-induced cell-autonomous toxicity, incorporating co-culture or organoid models (with non-neuronal cell types such as astrocytes and microglia) could reveal broader cell-cell interactions and systemic mechanisms.

Exposure to neurotoxic metals like Mn has been linked to PD development (8). Studies show that Mn can bind to α-synuclein metal-binding sites and promote α-synuclein aggregation—a key pathological hallmark of PD. Initially, Mn may protect neurons from α-synuclein toxicity, but prolonged Mn exposure likely enhances the neuronal accumulation and propagation of α-synuclein (5, 6). Mn also intensifies α-synuclein–induced inflammation, another PD hallmark (7, 19), and induces the expression of leucine-rich repeat kinase 2 (LRRK2), a key PARK gene linked to PD etiology (20, 21). Although epidemiological and experimental studies have associated Mn exposure with Parkinsonism, a meta-analysis found no direct link between the two (22). However, the analysis did not adjust for confounders like smoking and caffeine intake. Mn primarily targets the globus pallidus and striatum in humans; it has been shown to change neurotransmitters without altering dopaminergic neuron numbers, suggesting a divergence from idiopathic PD (23). Nevertheless, both disorders share motor symptoms like bradykinesia and rigidity, as well as cellular features such as oxidative stress, protein aggregation, and neuroinflammation. Previous Mn toxicity studies in flies focused on larvae, limiting relevance to adult Mn observed in occupational settings (10). To address this, we developed an adult Drosophila model of Mn neurotoxicity that recapitulates key PD-like features, including mitochondrial dysfunction, lysosomal defects, behavioral deficits, and neuronal loss.

PD is a multi-system disorder, as it concomitantly involves many non-motor symptoms that arise decades before PD’s cardinal motor symptoms surface (24); one of the earliest non-motor symptoms is constipation. Braak’s hypothesis suggests that sporadic PD may initiate in the gut (25). In PD patients, the α-synuclein pathology has been detected in the gut at the early stages of the disease (26, 27). Rodent models further demonstrated that α-synuclein originating from the gut (converted in situ from injected α-synuclein fibrils) can spread to the brain through the vagus nerve in a prion-like manner, leading to PD-like features (28). Mn is known to cause autonomic dysfunctions similar to PD (12), and recent studies by Payami and coworkers have associated PD with widespread microbiome dysbiosis (29, 30). To explore Mn toxicity and its connection to PD, we used a feeding model for direct gut exposure in flies and took a non-targeted, multi-compartmental approach to probe metabolome-wide changes early in Mn exposure before the motor symptom sets in. Using high-resolution mass spectrometry, we identified molecular signatures in the brain and body of Mn-exposed flies, including over 400 body metabolites and over 250 brain metabolites. Given our hypothesis treating PD as a multi-system disorder, we compared metabolite profiles across datasets and identified that biotin metabolism stood out as the most significantly impacted pathway, showing systemic increases in Mn-treated flies.

Biotin, or vitamin B₇, is a water-soluble essential vitamin obtained through diet or produced by the gut microbiome (14). Biotin acts as a cofactor for various carboxylases involved in metabolic processes, including dopamine production (31). Biotin deficiency has been linked to locomotor deficits and mitochondrial dysfunction in a tau-toxicity model (15), as well as short-term memory loss and deteriorated locomotor activity in rats, suggesting its role in PD (31). Gut microbiota is known to mediate B-vitamin bioavailability; prebiotic supplementation has been shown to increase the abundance of biotin-producing gut bacteria, particularly Bacteroides spp., in a high-fat diet mouse model (14). Notably, biotin-related bacterial genes and specific species in the Bacteroides genus were altered in the fecal samples of PD patients (29). Similarly, a metabolomic study of the A53Tα-Syn mouse model of PD identified biotin metabolism as an affected pathway (18). In this study, we found that biotin deficiency exacerbates Mn-induced neurotoxicity in flies and in PD patient–originated, iPSC-derived midbrain neurons. In contrast, biotin supplementation alleviated these Mn-induced pathologies.

Mitochondrial dysfunction has been recognized as a key molecular mechanism of pathogenesis in various PD models (32–36). Research has shown that the hallmark α-synuclein protein can bind to mitochondria and directly alter their cellular function (37). Further, α-synuclein also induces actin hyperstabilization, disrupting mitochondrial fission-fusion dynamics (38). In parallel, mutation in PARK genes, such as LRRK2, has been demonstrated to induce mitochondrial dysfunction in animal and cellular models (39). Here, we further show in vivo and in vitro that biotin supplementation rectifies Mn-induced neurotoxicity by rescuing neuronal mitochondrial mass, morphology, respiration, and metabolic phenotypes. Future research should explore the molecular basis of biotin-mitochondria interactions, including biotin transport and ligand-binding sites, in PD and related neurodegenerative disorders.

The therapeutic potential of B-vitamin biotin as a countermeasure against PD and neurodegeneration is immense. Biotin is well tolerated in humans, and emerging microbiota manipulation technologies enable non-pharmacological approaches such as supplementation through biotin-rich prebiotics and/or biotin-producing probiotics for curbing PD (40). A meta-analysis of metagenomic studies on PD patients’ gut microbiome by Nishiwaki and colleagues revealed reduced microbial genes involved in riboflavin and biotin biosynthesis (after adjusting for confounding factors), suggesting a role for biotin in PD pathogenesis (41). However, our Btnd^RNAi knockdown data only suggests that biotin itself is important for neuronal survival when exposed to Mn. The altered biotin metabolism may stem from various sources, including impaired recycling and microbial production. The previous work by Lohr and coworkers also linked biotin deficiency to neurotoxicity in iPSC-derived neurons (15). Given that autonomic dysfunction is an early key symptom widely involved in PD pathogenesis and aging populations in general, future studies are needed to dissect the role of gut dysbiosis in regulating biotin production and recycling, risks of environmental neurotoxic exposures, and the neurodegenerative conditions altogether.

Materials and Methods

Drosophila husbandry and feeding

Fly crosses were conducted in a 25 °C incubator and aged at 25 °C for 5–20 days, depending on the experiments. The pan-neuronal driver nSyb-GAL4 was used to mediate the genetic knockdown of the btnd gene. Both the stocks for nSyb-GAL4 and UAS-Btnd RNAi were obtained from Bloomington Drosophila Stock Center (BDSC) (HMC05012; Bloomington #60020). For Mn treatment, 0, 1, 10, and 30 mM of MnCl₂ (#7773–01-5; Santa Cruz Biotechnology, Dallas, TX) was mixed in 3 mL of water, and the instant food for flies was made using this water. Biotin supplementation was performed as previously described (15). Briefly, biotin (Thermo Scientific Chemicals # AAA1420709) and Mn were added to 3 ml of water at a final concentration of 30 mM each and instant fly food (Carolina Biological, Burlington, NC) were mixed and adult flies were raised in the food containing both Mn and biotin. Food was changed every 3 days. All assays were performed 1 week after exposure to biotin and Mn. Unless noted otherwise, male flies were used throughout the study.

For the paraquat and rotenone studies, 5 mM paraquat (Millipore #50636–5mg) (42) and 30 μM rotenone (Sigma #R8875–1G) were dissolved and added to fly food as described above. 30mM of biotin was also added as described above. Adult flies were raised in this food containing the pesticides and biotin for 7 days.

Adult flies expressing human α-synuclein in the neurons (38, 39, 43, 44) were raised in 30 mM biotin-containing food for 7 days. After treatment, behavioral assays were performed.

iPSC cell culture and treatment

Human midbrain dopaminergic neurons were differentiated from induced pluripotent stem cells (iPSCs) purchased from Axol Biosciences (Cambridge, UK). The cells were differentiated and matured using the StemCell Midbrain Neuron differentiation and maturation kit (#100–0038 and #100–0041; StemCell Technologies, New York, NY). Briefly, 24- and 96-well plates were coated with laminin and poly-L-ornithine. Cells were then plated into StemCell Stemdiff^™ Midbrain Neural Differentiation Media (Mid-NDM), which was supplemented with 200 ng/ml human recombinant-Shh and 10-μM ROCK inhibitor. Every other day for seven days, half of the media was refreshed. Post seven days, we switched to using BrainPhys^™ Neuron Maturation Medium (#100–0041; StemCell Technologies), with half of the media being replaced every alternate day. All treatments were diluted using the BrainPhys^™ Neuron Maturation Medium; for treatment groups, 30 μM biotin and 100 μM Mn were supplemented in the BrainPhys Neuron Maturation Medium for 24 hours.

Climbing and locomotor assay

The climbing and locomotor activities were assessed following previously published procedures (39, 44, 45). To gauge climbing ability, we counted the number of flies climbing 5 cm within 10 seconds. In each vial, at least ten flies were present, and for each group, we had four vials. For the locomotor assay, each vial containing over 10 flies was tapped and placed horizontally. After 15 seconds, the number of flies walking was counted (46). For two groups, data were analyzed using Student’s t-test; for multiple groups, ordinary ANOVA with post hoc analysis was performed.

Histological assessment of TH neurons

Fly brains were fixed with formalin and paraffin-embedded. Then, 2 μm sections were cut, and DAB staining was performed, imaged using the Nikon EclipseE600 (Melville, NY), and counted using ImageJ as previously described (39, 47).

LysoTracker assay

The LysoTracker assay (kit #L7526; Thermo Fisher Scientific, Waltham, MA) was conducted as described (44) to assess lysosomal morphology. Briefly, fresh whole-mount brains from 10-day-old flies were incubated with 1 μM LysoTracker for 5 min at room temperature, mounted in PBS, and imaged immediately on a Zeiss laser-scanning confocal microscopy (Oberkochen, Germany). The number and size of the lysosomes were analyzed using ImageJ.

MitoTracker assay

Fly brains were dissected and incubated in 1-μM MitroTracker red dye (kit #M-7512, Thermo Fisher Scientific) for 15 mins, washed in PBS twice, and imaged using a Zeiss laser-scanning confocal microscopy (Oberkochen). The solidity, perimeter, and circularity were measured following previously published protocols (35). For iPSCs, mitochondrial mass was measured using a published protocol (19). Briefly, post-treatment, cells were incubated in 1-μM MitroTracker green dye (kit #M7514, Thermo Fisher Scientific) for 13 mins, washed using HBSS, and placed on a plate reader to quantify the fluorescence. Then, the values were normalized using nuclear stain, DAPI, and fluorescence.

Seahorse assay

Brain metabolic changes in Drosophila were measured using the Seahorse XFe96 metabolic bioanalyzer (North Billerica, MA). Oxygen consumption rates (OCRs) and extracellular acidification rates (ECARs) were determined as described previously (45). For all experiments, brains from 5-day-old flies of the appropriate genotypes were dissected and plated at one brain per well on XFe96 plates (Seahorse Bioscience, North Billerica, MA), and metabolic parameters were assayed as described (39, 45). The OCR values were normalized to DNA content using a CyQUANT assay (Thermo Fisher Scientific) following the manufacturer’s instructions. For cells, the Seahorse assay was performed based on our previous paper (3) for cells. Post-treatment, the Seahorse plate was incubated in a CO₂-free incubator for an hour. The calibration plate was incubated overnight in the CO₂-free incubator at 37ºC. The ECAR and OCAR values were normalized using CyQUANT (kit #C7026, Thermo Fisher Scientific).

MitoSOX assay

Mitochondrial reactive oxygen species (ROS) activity was assessed using the MitoSOX kit (#M36008, Thermo Fisher Scientific) for fly brains and iPSC-derived TH neurons (35, 39). Fly brains and iPSC cells were incubated in 1-μM MitoSox red dye (kit #M7514, Thermo Fisher Scientific) for 15 mins, washed using PBS, and a plate reader was used to quantify the fluorescence. Then, the values were normalized using the nuclear stain, DAPI, and fluorescence.

Immunocytochemistry (ICC)

ICC was performed using previously described protocols (36, 45, 48). Briefly, cells were plated and treated on a coverslip in 24-well cell culture plates. Post-treatment, cells were gently washed and fixed using 4% paraformaldehyde for 30 mins, washed with PBS twice, blocked in blocking buffer (2% BSA, 0.5% Triton-X, 0.05% Tween), and incubated with primary antibodies overnight at 4ºC. Post-incubation, cells were washed with PBS, incubated with secondary antibodies, washed again, and mounted on slides using a DAPI-containing mounting medium. The primary antibodies used in this study were an antibody to tyrosine hydroxylase (#AB152, RRID:AB_390204; Millipore Sigma, Burlington, MA) and a purified antibody to β3-tubulin (TUBB3; #801201, RRID:AB_2313773; BioLegend, San Diego, CA).

LDH-based cytotoxicity assay

A total of 100 μL of conditioned media was collected from cultures treated with Mn. The lactate dehydrogenase (LDH) release was measured using LDH Assay Kit (Fluorometric) (#ab197000; Abcam, Cambridge, UK) following the manufacturer’s protocol.

MTS cell viability assay

Cells were plated in 96-well tissue plates. After treatment, 10 μL of 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) reagent (#G3582; Promega, Madison, WI) was added and incubated at 37°C for 45 mins. After incubation, a plate reader was used to quantify the absorbance at 490 nm. Then, the values were normalized using nuclear stain, DAPI, and fluorescence (19).

Metabolomics analysis

To capture basic biochemical changes, flies were beheaded on day 5 of Mn exposure with heads and bodies collected and snap-frozen separately; each group had five sample replicates, and each replicate sample contained ten heads and bodies. To extract metabolites, thawed heads and bodies were added with ice-cold methanol:H₂O (2:1, v/v) solution pre-spiked with isotope-labeled internal standards alongside zirconium oxide beads (0.5 mm i.d., Yittria stabilized) (Next Advance). Samples were touch-vortexed for 30 sec, manually ground using sterile, disposable pestles (DWK Life Sciences, Millville, NJ), placed on a bead beater (Next Advance) in a cold room for 5 mins at the maximum speed, and finally centrifuged at 18,000 × g for 10 mins. The supernatant was SpeedVac-dried and reconstituted into acetonitrile:H₂O (2:98, v/v) upon instrumental analysis. As detailed elsewhere, a 15-min method of C18 liquid chromatography was used (49), followed by full-scan high-resolution MS1 data acquisition using a Thermo Vanquish UHPLC coupled to an Orbitrap Exploris^™ 240 high-resolution mass spectrometer interfaced with a heated electrospray ionization (ESI) source (Waltham, MA). Data in both ESI positive and negative modes were acquired. Quality assurance and quality control (QA/QC) procedures were implemented, spanning timely mass calibration, sample blindfolding, sample randomization, and internal standard-based monitoring. MS1 *.raw data were converted to *.abf, and processed in MS-DIAL 4.90 (Riken, Japan) (50) to obtain peak alignment tables separately for the head (ESI+), head (ESI-), body (ESI+), and body (ESI-) modes of analysis; detailed settings for data processing can be found in the supplementary materials (text S1). Welch’s t-test and one-way ANOVA were performed on each of the four datasets to screen for ion features of statistical significance; retrospectively, tandem MS/MS mass spectra were collected for these features correspondingly using pooled samples, separately for fly heads and bodies.

Informatics

To identify chemical structures, a three-pronged cheminformatic approach was applied to reach the highest compound coverage possible, featuring an integrated strategy of (i) matching against an internal RT-m/z library of 812 common metabolites established from authentic chemical standards (IROA Technologies, Bolton, MA), (ii) de novo formula and structural prediction in MS-FINDER 3.30 (Riken, Japan) (51) based on hydrogen rearrangement rules, using accurate mass, isotope ratios, and tandem MS/MS of ions of interest as input data, and (iii) matching against an in silico RT-m/z library constructed through pairing theoretical accurate mass (of [M+H]⁺ for ESI+ and of [M-H]^- for ESI-) with machine learning-based predicted chromatographic retention time (RT) for over 9,000 relevant structures compiled from KEGG, a recently reported atlas of mouse brain metabolites (52), and a newly curated list of microbiome metabolites from Exposome Explorer (53). The confidence of annotation was assigned to each structure based on the Schymanski Scale in compliance with the Metabolomics Standard Initiative (MSI) guidelines (54). To infer metabolomic patterns and pathways, multivariate statistics, chemical enrichment analysis, and quantitative pathway analysis were performed. Multivariate partial least squares discriminant analysis (PLS-DA) was conducted in R (Vienna, Austria) using the mixOmics R package. Chemical similarity enrichment analysis (ChemRICH) plots were constructed on altered metabolites to visualize the chemical space as clustered/enriched by compound classes (Davis, CA). Pathway enrichment analysis was performed in MetaboAnalyst 5.0 (Quebec, Canada) based on the KEGG dme pathway library (organism-specific, Drosophila melanogaster), using the Globaltest approach for pathway enrichment and relative betweenness centrality for the node importance measure in pathway topological analysis (55).

Human gene expression data

Datasets of human PD were retrieved from the NCBI GEO database with the MeSH search terms ““parkinson disease”“ OR ““Parkinson’s diseases” AND “gds” (for DataSets) AND “Homo sapiens,” resulting in 12 human studies of PD in total with deposited datasets of gene expression profiling of SNpc by the array to be included in our pooled, pairwise comparison of PD vs. healthy controls. We examined the expression levels for two types of biotin-related genes, including key enzymes for biotin metabolism (such as BTD, MCOS1, MCCC1, and HLCS) and biotin transporters (such as SLC5A6 and SLC19A3). Fold change was calculated for the levels in the PD sample as normalized to those in controls. All data were retrieved from NCBI Gene Expression Omnibus (GEO) database using the search term (“parkinson disease” [MeSH Terms] OR Parkinson’s disease [All Fields]) AND “Homo sapiens”[porgn] AND “gds” [Filter] (last accessed in April 2023).

Statistical analysis

Unless otherwise stated, all sample sizes (n) indicate biological replicates rather than technical replicates. The adopted sample sizes were determined based on our previous work handling these models and techniques. For all behavioral and molecular phenotyping data, we used either ordinary one-way ANOVA (with Dunnett’s multiple comparisons test) or ordinary two-way ANOVA (with Šídák’s multiple comparisons test) and unpaired Student’s t-test for comparison between two groups in GraphPad Prism 10 (San Diego, CA). For metabolomics data, Welch’s t-test was conducted for pairwise comparison, and one-way ANOVA and Tukey’s HSD test were used for multiple groups.