Relationships of gut microbiota, short-chain fatty acids, inflammation, and the gut barrier in Parkinson’s disease

More than thirteen studies from three continents have established that the gut microbial communities of patients with Parkinson’s disease (PD) differ from those of healthy subjects [1–3]. Most of these studies have focused on profiling the bacteria present in stool samples but have not linked these findings to functional consequences on microbiome-host interactions, leaving open the questions of whether and how altered bacterial composition could contribute to disease processes in PD. Microbially-produced short-chain fatty acids (SCFAs) and regulation of immune responses and gut permeability are some of the proposed mechanisms by which gut microbes could impact brain health and function [4, 5].

SCFAs, particularly acetic, propionic, and butyric acid, are the main end product of bacterial fiber fermentation in the gut [6]. They have many effects on host physiology including acting as an energy source for colonocytes [7, 8], regulating the gut barrier [9], and influencing inflammatory responses [10]. SCFAs have been suggested as key mediators in microbiota-gut-brain interactions [11]. In a study contrasting fecal SCFA levels in PD patients and control subjects, decreased concentrations of SCFAs, particularly butyric acid, in PD patients were linked to microbiota alterations [12], and several bacterial taxa reportedly reduced in PD are SCFA producers [2, 3, 13].

Inflammation is a widely recognized hallmark of PD. Increased levels of a variety of inflammatory molecules in brain and cerebrospinal fluid (CSF) [14–16] as well as blood [17–19] from PD patients have been documented. While the same specific cytokines, chemokines, and other inflammatory markers are not consistently implicated, IL-6, TNF, IL-1β, CRP, IL-10, CCL5, and IL-2 are among the most commonly dysregulated in the peripheral blood of PD patients [18]. It has been suggested that this inflammation may result from intestinal barrier deficiency which could increase systemic exposure to inflammatory microbial products such as lipopolysaccharide (LPS), a component of bacterial cell walls [5, 20]. Experiments measuring excretion of ingested sugars suggest increased permeability of the colon (but not the small intestine) in PD [21–23]. Higher levels of zonulin and alpha-1-antitrypsin, indicators of gut permeability, have been found in PD patients’ stool relative to controls [24], although this difference has not been consistently observed [25]. Additionally, increased detection of LPS [26] and decreased LPS-binding protein (LBP) [21, 27] in blood and plasma from PD patients suggest greater exposure of peripheral tissues and immune cells to LPS, supporting the existence of PD-related gut barrier dysfunction.

In recent years, there has been increasing recognition of a low-grade inflammatory state in the gut in PD. Studies have reported increased expression of genes encoding proinflammatory cytokines and chemokines in gut tissue from PD patients compared to controls [22, 28], and higher levels of IL-1α, IL-1β, CXCL8 (also known as IL-8), and CRP have been found in stool [29]. The neutrophil-associated protein calprotectin is an indicator of gut inflammation in inflammatory bowel disease [30], and increased levels of calprotectin have been found in the stool of PD patients relative to controls [24, 25]. Finally, higher numbers of CD3+ T cells and cells expressing the LPS receptor Toll-like receptor 4 (TLR4) have been identified in colon tissue from PD patients. When TLR4 was knocked out in a mouse model of PD pathology, symptoms were mitigated [22].

Taken together, these findings support the hypothesis that intestinal bacteria and their metabolites along with inflammation and barrier dysfunction impact PD pathophysiology. While there have been several studies evaluating inflammatory and gut permeability markers in PD, this study addresses the gap in knowledge regarding the interplay of these molecules in the blood and the gut and with the gut microbiota and SCFAs in PD patients and healthy control subjects. Identifying how these factors relate to one another and to the onset and clinical presentation of PD is a key step in elucidating the mechanisms behind these disease processes and pinpointing candidate molecules and pathways to target for preventive and therapeutic intervention.

Additional details are available in Additional File. 1

Numerous studies have reported abnormalities in the gut microbiome, levels of SCFAs, peripheral immune responses, and intestinal permeability in PD; however, this study provides for the first time an integrative analysis of these factors in the same subjects, allowing an evaluation of whether and how they are related to one another and to clinical manifestations of PD.

Efforts have been made to identify biomarkers of PD in both blood and stool, but the relationship between the levels of inflammatory markers in the gut and in circulation in this disorder had not been elucidated. We determined that levels of inflammation-associated molecules in the plasma and in the stool were highly correlated within each sample type. We found essentially no correlation, however, between levels of inflammatory markers in plasma and stool. This indicates that systemic and gut immune responses are distinct and, thus, measurements from one site are unlikely to represent the nature of the responses in the other.

In the plasma, we found a significant decrease in levels of the chemokine CXCL8 (also known as IL-8) in female PD patients as well as an inverse correlation between CXCL8 and non-motor symptom severity scores in the full cohort. CXCL8 is the primary chemoattractant for neutrophils and is commonly produced in inflammatory responses to stimulate and recruit these phagocytic cells. A previous study has reported a positive correlation between CXCL8 levels in the serum and the degree of PD-related disability [43], rendering our findings somewhat unexpected. A possible explanation is that plasma CXCL8 levels in this cohort are significantly higher in subjects with a history of stroke (Fig. 3), and there are more of these individuals in the control than PD group.

In stool, we documented lower levels of SCFAs and higher levels of calprotectin in PD patients. This is consistent with other reports of SCFA and SCFA-producing bacteria reductions [3, 12, 13, 19, 22, 31] as well as increased stool calprotectin [24, 25] in PD. While SCFAs and calprotectin are beginning to emerge as some of the most reliable indicators of a dysregulated and inflammatory gut environment in PD, our results suggest that these associations may be more pronounced in males and in females, respectively. Levels of butyric acid specifically were correlated with the age of onset for both motor and non-motor PD symptoms even after accounting for disease duration, raising the possibility that higher butyric acid levels could be protective and delay disease onset. In the whole cohort, we also found correlations between stool SCFA levels and non-motor, particularly gastrointestinal and depressive, symptoms, but, with the exception of a positive association between acetic acid and defecation frequency, these correlations were not significant in just the PD subset of the cohort. This could be due to a lack of statistical power or to differences between the control and patient groups in the relative impact of stool SCFAs on the mechanisms driving these symptoms, and it warrants further study.

Though SCFAs are known to have immunomodulatory and barrier-promoting properties, we found no significant correlations between the most abundant stool SCFAs and immune- or permeability-related factors in stool or plasma. While it is possible that SCFA levels may directly impact immunological pathways not assessed in this study, the lack of clear association between SCFA production and inflammation likely evinces the complexity of the relationship between these physiological processes as well as the complexity of immune regulation overall. Both pro- and anti-inflammatory activities of SCFAs have been documented [44, 45], and dietary supplementation of SCFAs produces variable effects on inflammation which appear to be strongly influenced by the range of SCFA concentrations involved [46]. Further investigation of the effects of SCFAs on immune responses in the gut, the blood, and the central nervous system in the context of PD pathology is needed.

NGAL, an epithelium-derived antimicrobial glycoprotein overexpressed in inflammatory conditions [47], has been found at higher levels in the plasma of PD patients compared to controls [43]. To the best of our knowledge, this comparison had not been performed in stool, and we found no significant differences in fecal NGAL levels between patients and controls. We also found no significant indications of greater intestinal permeability in this cohort of PD patients as measured by stool zonulin or plasma LBP, though other studies have [21, 24, 27]. On the contrary, we found that higher zonulin levels were associated with less severe clinical manifestations of PD. This could be a reflection of the diversity of motor and non-motor symptom subtypes in PD, a possibility which could be explored in a larger cohort [48]. Longitudinal studies assessing intestinal permeability beginning in the earliest stages of PD would also be beneficial to determine its impact on disease symptoms and progression.

It has been proposed that intestinal inflammation in PD is related to gut permeability and to alterations in microbiota composition and that it can contribute to disease pathogenesis [5]. In this study, we found that levels of fecal NGAL were positively correlated with fecal zonulin, supporting a relationship between this intestinal inflammatory response and permeability. Levels of inflammatory and permeability markers were also associated with the stool microbiota. Inverse associations were found between alpha diversity and levels of stool calprotectin and CXCL8 in PD patients, consistent with the concept that an inflamed gut environment can exert selective pressure on the microbiota. NGAL, IL-2, the stool marker PC, and zonulin were also significantly associated with alterations in beta diversity indicating that gut immune responses and permeability were related to gut microbial composition. Furthermore, stool CXCL8 levels were associated with more pronounced constipation in PD, and stool CXCL8 and IL-1β, which were found to be increased in a different cohort of PD patients [29], were negatively correlated with the age of PD motor symptom onset in this study, supporting the involvement of gut inflammation in the development and/or progression of PD pathology.

A notable finding in this study is that the relationships between gut microbes and inflammatory factors and between microbes and their metabolites such as SCFAs in stool differ between PD patients and controls. In control subjects only, SCFAs were inversely associated with alpha diversity, and SCFA associations with beta diversity were stronger in controls than PD patients. Conversely, levels of butyric acid in stool differed significantly by enterotype only among PD patients. Relationships between SCFAs and individual taxa were largely concordant in both subject groups, including positive correlations between stool SCFAs and well-known SCFA producers like Butyricicoccus and Roseburia and a negative correlation with Akkermansia, which SCFAs are known to inhibit [49]. On the other hand, a positive correlation between SCFAs and relative abundance of Bacteroides was found only in controls while a negative correlation with Bifidobacterium was found only in PD patients.

These discrepancies could reflect differences in microbiota composition between patients and controls; Bacteroides is more abundant in controls, and Bifidobacterium is more abundant in PD patients in this cohort [32], which could increase their functional impacts within their respective microbial communities. This could explain the positive association between SCFAs and Bacteroides, a producer of SCFAs, in controls, but not necessarily the inverse relationship between Bifidobacterium and butyric acid in PD patients. Increases in Bifidobacterium are considered butyrogenic due to cross-feeding of butyrate-producing bacteria [50], but the opposite association was found in our PD cohort, suggesting an altered function or strain profile of the Bifidobacterium genus in the metabolic network of the PD microbiota. Further evidence that Bifidobacterium may not be performing a homeostatic function in PD is its positive correlation with the inflammatory marker NGAL in patients only.

PD patients in this cohort had reduced abundance of Prevotella [32], and significant links between lower levels of fecal zonulin and NGAL and Prevotella abundance and enterotype were observed only in the control group. Visual inspection of the data in Additional Files 10 and 12 suggests that a similar trend in the relationship between NGAL and Prevotella exists in the PD group, and differences in the strength of the association in patients and controls may be influenced by the lower abundance of Prevotella in PD. Whether the relationship between Prevotella and zonulin in PD differs from that in controls is less clear, and, as this taxon is frequently reported to be impacted in PD [19, 31, 32, 51–53], future studies evaluating its influence on intestinal permeability in patients are warranted. Detailed examination of differences in bacterial populations and functions in PD patients and controls may be important in developing the next generation of PD prevention and therapies, as our findings support a relationship between gut microbial composition and diversity and the severity of PD motor symptoms.

Our study confirms deficiencies in SCFAs and elevated levels of fecal calprotectin in PD. It also implicates low SCFA levels and gastrointestinal inflammation in the development and/or progression of PD, relating levels of butyric acid and CXCL8 and IL-1β in stool to age at disease onset. Moreover, this study provides insights into the relationships among key components of the gut environment and systemic responses. We show that although inflammatory markers in the stool and in circulation were highly correlated within one site, they were not correlated across sites, and that levels of SCFAs and inflammatory and permeability markers in stool were not directly related. It has been established that alterations in the composition of the gut microbiota and intestinal immune responses occur in PD, but our findings suggest that the activities of particular bacteria and the nature of their interactions with the host immune system may be changing as well.

Future longitudinal microbiome studies using multi-omics approaches with higher taxonomic resolution combined with detailed pheno- and genotyping of subjects can confirm and expand upon our findings in this initial cohort and provide a thorough understanding of how gut microbes and their metabolites are interacting with the host and impacting the etiology, symptoms, and progression of PD. Studies verifying the impact of the molecules and pathways identified in this study on neurodegenerative processes can produce new targets along the gut-brain axis for more effective disease-modifying treatment of this disorder.

This study was supported in part by the Emory Multiplexed Immunoassay Core (EMIC), which is subsidized by the Emory University School of Medicine and is one of the Emory Integrated Core Facilities.

MGT, FS, PA, LP, PABP, JC, VTEA, and MCH developed and organized the study. VTEA, MCH, PABP, JC, KR, LP, PA, and FS collected samples, data, and measurements. VTEA, MCH, and VH analyzed data. VTEA and MCH drafted the manuscript. All authors read, reviewed, edited, and approved the final manuscript.

This study was funded by the Michael J. Fox Foundation for Parkinson’s Research, the Academy of Finland (295724, 310835), the Finnish Medical Foundation, and the Hospital District of Helsinki and Uusimaa (UAK1014004, UAK1014005, TYH2018224). The funding bodies did not contribute to the design of the study, to collection, analysis, or interpretation of data, or to the writing of the manuscript.

Microbiota data are available at the European Nucleotide Archive (accession number PRJEB27564) (https://www.ebi.ac.uk/ena/browser/view/PRJEB27564↗). Other data and files utilized in this study are available from the corresponding authors upon reasonable request. The complete R code for analysis is included as Additional File 2.

This study was conducted in accordance with the Declaration of Helsinki and was approved by the ethics committee of the Hospital District of Helsinki and Uusimaa. All participants gave informed consent.

Not applicable.

VTEA, PABP, LP, PA, and FS have patents issued (FI127671B & US10139408B2) and pending (US16/186,663 & EP3149205) that are assigned to NeuroBiome Ltd.

FS is founder and CEO of NeuroInnovation Oy and NeuroBiome Ltd., is a member of the scientific advisory board and has received consulting fees and stock options from Axial Biotherapeutics.

MGT is an advisor to INmune Bio, Longevity Biotech, Prevail Therapeutics, and Weston Garfield Foundation. MGT has patents issued (US Pat. Nos. 7144987B1 and 7244823B2) and pending (US20150239951, WO2019067789, 62/901698, see efiling Ack37193677, 62/905747, see efilingAck37274773) for co-invention of DN-TNFs.

Malú G. Tansey, Email: mgtansey@UFL.edu

Filip Scheperjans, Email: filip.scheperjans@hus.fi.