Introduction

Microbiota are necessary for human health. The gut microbiome aid with dietary metabolism, produce essential metabolites such as vitamins, maintain the integrity of the intestinal barrier, inhibit pathogens, and metabolize drugs and toxicants1. The gut microbiome regulates development and continued education of the host’s immune response and nervous system by producing specific metabolites with far reaching effects such as the maturation and maintenance of microglia in the brain2. An imbalance (dysbiosis) in microbiome composition and function can render host prone to disease. Studies in humans and animal models have revealed disease-related dysbiosis in a range of common metabolic (e.g., diabetes), inflammatory (inflammatory bowel disease), neurologic (Parkinson’s disease) and developmental disorders (autism)1.

Parkinson’s disease (PD) is a progressively debilitating disorder that affected 4 million Individuals in the year 2005 and is projected to double to 8.7 million individuals by the year 20303. Although historically defined as a movement disorder, PD is a multi-systemic disease4. The earliest sign is often constipation which can precede motor signs by decades4. Moreover, PD is etiologically heterogenous. Despite the discovery of several causative genes5, 90 susceptibility loci spanning the human genome6, and multiple environmental risk factors7, the vast majority of PD remains idiopathic. It is speculated that PD is caused by various combinations of genetic susceptibility and environmental triggers, although no causative combination has yet been identified.

Braak’s hypothesis8 that non-familial forms of PD start in the gut by a pathogen is gaining increasing support. The connection between PD and the gastrointestinal (GI) system, including constipation, compromised gut barrier, and inflammation, has long been established. Alpha-synuclein pathology has been detected in the gut of persons with PD at early stages9, and there is evidence from imaging studies that in some cases pathology may start in the gut and spread to the brain10. In mice, it was shown that alpha-synuclein fibrils injected into gut induce alpha-synuclein pathology which spreads from gut to brain, and that vagotomy stops the spread11. In parallel, large epidemiological studies have shown that persons who had complete truncal vagotomy decades earlier had substantially reduced incidence of PD later in life9.

With accumulating evidence implicating the gut as an origin of PD, and the newly gained appreciation for the involvement of gut microbiome in chronic diseases, there has been increasing interest in decoding the connection between the gut microbiome and PD. In mice overexpressing the human alpha-synuclein gene, we have shown that gut microbiome regulates alpha-synuclein-mediated pathophysiologies12. In another genetic model of PD (Pink1-/-), intestinal infection with Gram-negative bacterial pathogens was shown to elicit an immune reaction that leads to neuronal degeneration and motor deficits, and which can be reversed with the PD-medication, L-dopa13. In addition, we and others have found that, curli, an amyloidogenic protein produced by Gram-negative Escherichia coli, induces alpha-synuclein aggregation and accelerates disease in the gut and neurodegeneration in the brain14,15,16. We have detected overabundance of opportunistic pathogens in the gut microbiomes of individuals with PD17. Collectively, experimental and human studies support Braak’s hypothesis that intestinal infection may act as a triggering event in PD, but it is yet to be proven that pathogens in human gut cause PD. Studies conducted on human fecal samples have all found evidence of dysbiosis in PD gut microbiome but results on specific microorganisms that drive the dysbiosis have been mixed18,19,20. Human studies of PD and microbiome have had limited sample sizes, and all except two21,22 were based on 16S rRNA gene amplicon sequencing (henceforth, 16S) which limits resolution to genus-level. Metagenomics (study of all genetic material sampled from a community) is an emerging field in medical science. With deep shotgun metagenomic sequencing, the microbiome can now be studied in large-scale human studies at high resolution of species and genes.

Here we present a large-scale metagenomics analysis of PD gut microbiome. This study was designed and executed by a single team of investigators (NeuroGenetics Research Consortium, NGRC), enabling complete control to employ state-of-the-art methods and ensure uniformity from start to end. We confirm the common findings from prior studies, resolve them to species level and solve the inconsistencies in the literature. In addition, owing to large sample size and deep shotgun sequencing, we generate a vast amount of new information. We show wide-spread dysbiosis in PD microbiome, identify species that drive the dysbiosis, and by functional profiling, nominate microbial genes and pathways in the gut that may contribute to PD mechanisms.

Results and discussion

A large, newly enrolled, and uniformly assessed cohort

As outlined in STORMS23 flowchart (Fig. 1), the study included a newly enrolled cohort of 490 persons with PD and 234 neurologically healthy controls (NHC). The sample size is comparable to Human Microbiome Project (HMP) which included 100 individuals with inflammatory bowel disease, 106 individuals with pre-diabetes and 242 healthy adults ages 18-40 years old24. Defined by self-reported biological sex, 52% of subjects were men, 48% were women. 97% of PD cases and 93% of NHC were over 50 years old. The older ages of the controls in this study (mean 65.8 ± 8.8) and their neurologically healthy status is a unique addition to the publicly available datasets. All subjects were from a single geographic area in the Deep-South United States (US), minimizing confounding by geographic variation. Fifty-five percent of NHC were spouses and shared environment with PD. Using uniform methods, we collected extensive metadata (Supplementary Fig. 1) and a stool sample from each subject, extracted DNA from stool and conducted deep shotgun sequencing achieving average 50 M raw reads/sample. This dataset is new and is publicly available.

Fig. 1: The STORMS flowchart.
figure 1

Following reporting guidelines for human microbiome research23, we show the step-by-step process by which the study was conducted, starting with subject selection, enrollment and data collection for persons with PD (a) and neurologically healthy controls (b), sequencing and bioinformatics pipeline (c), and statistical analyses (d, e).

Full size image

Subject characteristics and metadata

Data on 53 variables were analyzed to characterize the subjects, and to identify disease-associated variables that could potentially confound downstream metagenomics analyses (Table 1). GI problems, which are well-known features of PD, were readily evident in this cohort. Constipation was more prevalent in PD cases (odds ratio (OR) = 6.1, 95% confidence interval (CI) = 3.9–10, P = 2E−19 for chronic constipation, P = 3E−6 for Bristol Chart score), and PD cases reported more GI discomfort than NHC (OR = 2.8, 95% CI = 1.8–4.2, P = 3E−7). Compared to NHC, PD cases had diminished intake of alcohol (OR = 0.6, 95% CI = 0.4–0.8, P = 3E−4) and foods in all five categories (fruits/vegetables, animal products, nuts, yogurt, and grains), all reaching significance (OR = 0.6–0.7, P = 0.002–0.05) except grains (OR = 0.8, 95% CI = 0.6–1.1, P = 0.2). Use of laxatives (OR = 3.8, 95% CI = 2.4–6.4, P = 7E−10), pain medication (OR = 1.6, 95% CI = 1–2.5, P = 0.04), sleep aid (OR = 2, 95% CI = 1.4–2.9, P = 7E−5), and medication for depression/anxiety/mood (OR = 2.1, 95% CI = 1.4–3, P = 7E−5) were more common in PD than NHC. Probiotic supplement use was more common in NHC than PD (OR = 0.6, 95% CI = 0.4–0.9, P = 0.02), which is noteworthy because as the data will show, Bifidobacterium and Lactobacillus species, which are common constituents of commercial probiotics, were more abundant in PD than NHC metagenomes. Variables that differed in PD vs. NHC were evaluated as potential confounders in downstream metagenomics analyses.

More at link.