Gut microbiome dysbiosis across early Parkinson's disease, REM sleep behavior disorder and their first-degree relatives

Alpha-synucleinopathies, such as Parkinson's Disease (PD), are characterized by the abnormal aggregation of alpha-synuclein (α-syn) protein in the central nervous system (CNS). However, evidence suggests α-syn pathology occurs in the enteric nervous system (ENS) before CNS involvement, supporting the gut-to-brain propagation of α-synucleinopathy. Consistently, gut microbiota disturbance (gut dysbiosis) has been reported in PD patients, hypothesized to be related to intestinal hyperpermeability, immune activation, and pathological α-syn aggregation. Since enteric α-syn pathology and ENS dysfunctions (e.g., constipation) can precede PD onset by decades, understanding gut microbiota and host-microbiome interactions at earlier prodromal stages is crucial. REM sleep behavior disorder (RBD) is considered the most specific prodromal marker of PD, characterized by dream-enactment behaviors during REM sleep without atonia. Patients with v-PSG-confirmed RBD show increased constipation prevalence and increased phosphorylated α-syn immunostaining in the ENS. PD patients with premotor RBD features exhibit prominent peripheral nervous system degeneration compared to those without, suggesting a distinct PD subtype reflecting the gut-brain hypothesis of α-synucleinopathy. Isolated RBD symptoms, not yet meeting v-PSG diagnostic criteria, might represent a prodromal stage of RBD and early α-synucleinopathy. First-degree relatives of RBD (RBD-FDR) show increased constipation and a spectrum of RBD features, suggesting they may harbor α-synucleinopathy at an earlier stage than RBD patients. Prior studies reported gut microbiota dysbiosis in v-PSG-diagnosed RBD and possible RBD, but might be underpowered. The increasing recognition of a prodromal stage of RBD underscores the importance of studying gut microbiota at even earlier stages. This large cross-sectional study, simulating the Braak staging model with a quasi-longitudinal design, aimed to disentangle the associations of gut microbiota with α-synucleinopathy progression.

Extensive research has linked gut dysbiosis to various complex diseases, including Parkinson's Disease (PD). Studies consistently report PD-associated gut dysbiosis, particularly the depletion of short-chain fatty acid (SCFA)-producing bacteria and the enrichment of putative pathobionts. This dysbiosis is thought to contribute to intestinal hyperpermeability, immune activation, and the aggregation of α-synuclein, a key protein implicated in PD pathogenesis. However, the timeline of these microbial alterations in relation to PD onset remains largely unclear. Several studies have investigated the gut microbiome in REM sleep behavior disorder (RBD), a prodromal marker of PD, but these studies often had limited sample sizes, leaving a gap in our understanding of early microbial changes. This study aims to address this gap by investigating the gut microbiome across a range of individuals representing different stages in the progression of α-synucleinopathy.

This cross-sectional study included stool samples from 452 subjects from Hong Kong cohorts of v-PSG-diagnosed RBD and RBD families. After excluding samples with low read counts, 441 samples remained for analysis. The study included four groups: controls (neurologically healthy, Braak stage 0-1), RBD-FDR (Braak stage 0-2), v-PSG-diagnosed RBD (Braak stage 2-3), and early PD (clinically confirmed PD with motor symptoms onset <5 years, Braak stage 3-4). Gastrointestinal symptoms were assessed using the Rome-IV questionnaire and SCOPA-AUT. Medication usage, lifestyle factors, and other clinical characteristics were also recorded. DNA was extracted from stool samples, and 16S rRNA V3-V4 regions were sequenced using Illumina MiSeq. Sequencing reads were processed using DADA2 to generate amplicon sequence variants (ASVs). Taxonomy was assigned using the SILVA database. Alpha and beta diversity analyses were performed using QIIME2 and R. Differential abundance analysis was conducted using Kruskal-Wallis tests and MaAsLin2, accounting for family clustering. Random forest modeling was used to predict RBD status based on microbial markers. PERMANOVA was used to assess the impact of host factors on microbial composition. Mediation analysis examined the relationship between gut microbiota, bowel movement frequency (BMF), and the likelihood ratio (LR) of prodromal PD. PICRUSt2 predicted functional profiles from 16S rRNA data. Statistical significance was determined using appropriate methods with Benjamini-Hochberg correction for multiple comparisons.

The study revealed significant alterations in gut microbiota composition in early PD and RBD compared to controls and RBD-FDR. Principal coordinate analysis (PCoA) showed distinct clustering patterns for early PD and RBD, while controls and RBD-FDR were not significantly different. Analysis of alpha diversity showed no significant difference across groups, indicating similar microbial richness and evenness. However, beta diversity showed significant differences, suggesting changes in microbial composition. Differential abundance analysis identified significant changes in several genera and families associated with the progression of α-synucleinopathy. Butyrate-producing bacteria (*Butyricicoccus*, *Faecalibacterium*, Lachnospiraceae) were significantly depleted in RBD and early PD, while pro-inflammatory *Collinsella* was enriched. Interestingly, *Collinsella* enrichment was observed even in RBD-FDR, suggesting early dysbiosis. Random forest modeling successfully distinguished RBD from controls with a mean AUC of 0.79 (training set) and 0.75 (test set), identifying 12 microbial markers. The model also distinguished RBD from RBD-FDR with a mean AUC of 0.72 (test set). Analysis of host-microbiome interactions revealed that BMF score, sex, osmotic laxatives, and PPIs significantly influenced microbial composition. Mediation analysis suggested that BMF partially mediated the effect of gut microbiota on the LR of prodromal PD. Functional prediction using PICRUSt2 showed increased fatty acid fermentation (lactate, ethanol, acetate) and decreased cofactor and vitamin biosynthesis (B1, B2, B12) in RBD and early PD.

This study demonstrates that PD-like gut dysbiosis is present at the prodromal stages of PD, evident in RBD patients and their first-degree relatives. The depletion of butyrate-producing bacteria, known for their role in maintaining gut barrier integrity and immune regulation, and the enrichment of pro-inflammatory bacteria like *Collinsella* highlight potential mechanisms linking gut dysbiosis to PD pathogenesis. The finding that these alterations occur even in RBD-FDR, a population at an earlier stage of the disease, emphasizes the potential of gut microbiota as a very early biomarker and suggests possible therapeutic targets for preventing PD. The mediation analysis further supports a causal relationship between gut dysbiosis and constipation, which is a common non-motor symptom of PD. However, the fact that these microbial changes persist even after adjusting for BMF indicates that these are not simply consequences of constipation.

This study provides strong evidence that gut dysbiosis is present at the prodromal stages of PD, even before the onset of motor symptoms. The identification of specific microbial markers that can distinguish RBD from controls and RBD-FDR holds promise for the development of early diagnostic tools. Future prospective studies are needed to establish causality and to explore the potential of targeting gut microbiota as a disease-modifying therapy for PD. Further research incorporating metagenomics and metabolomics analyses is also warranted to gain a more comprehensive understanding of the gut microbiome's role in PD pathogenesis.

The cross-sectional design limits the ability to establish causality between gut dysbiosis and α-synucleinopathy progression. The relatively small sample size of the early PD group may limit the power of some analyses. The study primarily focuses on the 'body-first' model of PD, and the findings might not be generalizable to all PD subtypes. RBD-FDR subjects were younger and had a higher proportion of females, which could influence the results. The study relied on 16S rRNA sequencing and functional prediction, and future research incorporating metagenomics and metabolomics could provide a more complete picture.