Role of the Microbiota-Metabolite Axis in the Rotenone Model of Early-Stage Parkinson's Disease

Gastrointestinal symptoms are common in the early-stage Parkinson's disease (PD), but its potential pathogenesis remains unclear. Therefore, in the present study, we used the 16S ribosomal RNA gene sequencing and gas chromatography coupled with mass spectrometry-based metabolomics investigated the gut microbiome alterations and the serum amino acid levels in early-stage PD mice model induced with rotenone and vehicle-induced control mice. The results demonstrated that the microbial taxa at phylum, family and genus levels remarkably altered in rotenone-induced mice relative to vehicle-induced mice. The rotenone-induced mice had higher relative abundances of Flavobacteriaceae, Staphylococcaceae, and Prevotellaceae as well as lower relative abundances of Lachnospiraceae_UCG-001, Ruminiclostridium, and Prevotellaceae_NK3B31_group than vehicle-induced mice. The evaluation of serum amino acids revealed the alterations in several classes of amino acids, including L-proline, L-alanine, L-serine, L-asparagine, L-threonine, L-glutamine, L-methionine, and L-4-hydroxyproline. Notably, the altered serum amino acid levels were signicantly associated with the abundance of gut microbiota, especially Ruminococcaceae and Ruminiclostridium. Our study explored the possible role of the gut-microbiota-metabolite axis in the early-stage PD and provided the possibility of prevention and treatment of PD by gut microbiota-metabolite axis in the future. the alterations of the gut microbiota and serum amino acids in early-stage PD mice. Our data provided the new insights that the changes in gut microbial communities may deprive the normal functions of gut microbiota to result in the progression of neurodegeneration of PD. Therefore, it's necessary to maintain the normal symbiotic relationship between host and gut microbiota to intervent the development of PD.


Introduction
Parkinson's disease (PD) is known as a slowly progressive degenerative disease involving the peripheral and central nervous system (PNS and CNS), according to Braak's hypothesis (Braak and Del Tredici 2017). The full truncal vagotomy further proved that the misfolding α-synuclein could retrogradely spread from the gastrointestinal tract to the brain through vagal nerve, resulting in the loss of dopaminergic neurons in substantia nigra (SN) and the decrease of dopamine in striatum (Kim et al. 2019). Studies have shown that the motor symptoms didn't occurs apparently until the PD patients lost 60-80% dopaminergic neurons in SN (Deumens et al. 2002). Therefore, there recently remains a great need for diagnosis and therapy of early-stage PD.
Although evidences have indicated that the aggregation of pathological α-synuclein and the loss of dopaminergic neurons are the pathological hallmark of PD, the etiology of PD is still worthy of further study. There was an association between neuroin ammation and neurodegeneration of PD. For example, studies have found that in parkinsonian animal models, there were a large number of proin ammatory cytokines in the CNS (Dzamko et al. 2015; Lai et al 2018; Mamik and Power 2017). In addition, the in ammation in the gut was also the main cause of the misfolding and aggregation of α-synuclein. For example, patients with in ammatory bowel disease (IBD) have a higher risk of developing PD than non-IBD individuals (Brudek 2019). Recent studies con rmed that the gut microbiota exhibited more proin ammatory characteristics (e.g., the increase of lipopolysaccharide biosynthesis), which was considered to be one of the risk factors for PD (Keshavarzian et al. 2015). There is growing evidence that alterations of gut microbiota presented in patients and animal models of PD ( Remarkably, the alteration of the gut microbiota with antibiotics could reduce the expression of proin ammatory markers, which could attenuate dopaminergic neuron loss in SN and prevent the development of PD (Koutzoumis 2020). The proposal of microbiota-gut-brain axis further indicated that the abnormality of gut microbiota and microbiota's metabolic products may be among potential candidates to trigger the formation of Lewy bodies in the enteric nervous system (ENS). To date, several studies investigated the metabolomics pro le of PD, and found the alterations of amino acid level in vivo (Lu et al. 2014;Wuolikainen et al. 2016;Luan et al. 2015; Molina et al. 1997). And gut amino acids have also been shown to have potential neuromodulatory activities (Needham et al. 2020).
Interestingly, the animal model of PD showed that chronic intragastric administration of rotenone (a complex-1-inhibitor) could result in a spatio-temporal distribution of α-synuclein corresponding to Braak's pathological stage (Pan-Montojo et al. 2012. In the present study, C57BL/6 mice were given oral low-dose rotenone administration for 3 weeks to con rm whether rotenone short-term exposure in the gut would rst induce pathology and symptoms restricted to the gut. Then, in order to further understand the relevance of gut microbial ecosystem and PD, we used 16S rRNA gene-based amplicon sequencing and gas chromatography-mass spectrometry (GC-MS)-based metabolomics to explore the interactions between bacterial metabolisms, gut microbes, and disease.

Animals
The animal experiments in the present study were carried out according to the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals and approved by Animal Care Committee of the First A liated Hospital of Harbin Medical University. Male C57BL/6 mice (23-25 g) were purchased from Changzhou Cavens Experimental Animal Co., Ltd. (Jiangsu, China). The mice were housed in a speci c-pathogen-free (SPF) environment with 12-h light/dark cycle, 23 ± 2°C and 45 ± 5% humidity. And they were provided with food and water ad libitum. After one-week acclimation, all mice were randomly diveded into two groups (n = 10 per group): control group and rotenone group. And a earlystage Parkinson's disease model was established by oral rotenone (Sigma, St Louis, MO, USA) solution (30 mg/kg suspended freshly in 4% carboxymethylcellulose and 1.25% chloroform vehicle) once a day for 21 days. Meanwhile, control animals received vehicle. After 21 days of rotenone treatment, all mice underwent behavioral tests and fecal samples were collected and immediately stored at -80℃ for next analysis. All mice were sacri ced and then the acquired fresh distal colon and midbrain were xed in 4% paraformaldehyde for immuno uorescence (IF).

Behavioral tests
In order to assess the motor function, the pole test was rst performed according to the protocol previously described (Ogawa et al. 1985;Cao et al. 2017). Brie y, the mice were placed the top of the pole that was made 50-cm height, 1-cm diameter and with a rough surface. The time was recorded when the mice reached the the bottom from the top of the pole. Each mouse performed three successive trials, and then the average value of the three trials was calculated for statistical analyses. The rotarod test was also performed as described before (Inden et al. 2007). And the mice were placed on an accelerating rod that was set the starting speed of 4 rpm and gradually increased to 40 rpm. The latency to fall was recorded for a maximum of 300 s. All mice received ve trials, and the average latency was analyzed.

One hour stool collection
The one-hour stool frequency of mice was observed to assess the colonic motion function. Each mouse was placed in a clean plastic cage without food or water for one hour. Fecal pellets were collected immediately after expulsion and placed in sealed 1.5 ml tubes. Tubes were weighed to obtain the wet weight of stool, and the dry weight of stool were obtained by reweighing the tubes dried overnight at 65°C. The water content of stool was calculated from the difference between the wet and dry stool weights.

Immuno uorescence
The xylene and graded alcohols were applied to depara nise and hydrate the midbrain coronal para n slices (3 µm) and colon axial para n slices (3 µm). Antigen retrieval was performed in microwave EDTA buffer (pH 8.0). After blocking with serum for 1 h, the slides were rstly incubated with primary antibodies (anti-α-syn; anti-phospho-α-syn antibody; anti-TH; anti-βIII antibody, Abcam) overnight at 4 ℃, and washed with PBS for three times, then incubated with corresponding uorescent secondary antibody. Finally, the images were taken with an epi uorescence microscope (Nikon eclipse C1, Nikon, Japan) and the relative expression levels of respective molecules in SN were analyzed by Image Pro-Plus software.

GC-MS analysis parameters
For GC-MS analysis, 1 µl of the derivatized sample was injected splitless into an Agilent 7890B GC system coupled to an Agilent 5977A mass selective detector (Santa Clara, CA, USA), with a DB-5MS fused silica capillary column (30 m × 0.25 mm × 0.25 µm, Agilent J&W Scienti c, Folsom, CA, USA). Helium (>99.999%), as carrier gas, was injected into the column at a constant ow rate of 1.2 ml/min. The temperatures of injector, transfer line and electron impact ion source were set to 300°C, 280°C and 330°C, respectively. Temperature programming: the initial oven temperature was 60 ° C, increased to 125°C at a rate of 15°C/min, to 210°C at 8°C/min, to 270°C at 11°C/min, to 305°C at 25°C/min and nally held at 305°C for 3 min. MS data were acquired in a full-scan mode (m/z 40-600), and the solvent delay time was set to 4 min. QC samples were detected at regular intervals (every 10 samples) throughout the analysis.

Statistical Analysis
The alpha diversity indices of gut microbiota, including observed_species, Chao, Simpson, Shannon and PD_whole_tree were calculated by QIIME software (Version 1.8.0). The characterizations of microbial communities at different levels were identi ed by linear discriminant analysis (LDA) effect size (LEfSe), and LDA was used to estimate the effect size of each feature. The metabolic analyses: the unsupervised principal component analysis (PCA) and supervised orthogonal partial least-squares discriminant analysis (OPLS-DA) were performed using online resources (http://www.metaboanalyst.ca/). Continuous variables were expressed as the mean ± standard deviation (SD), and compared by parametric Student's t-test and non-parametric Wilcoxon rank-sum test. Heatmap of correlations between the microbiome and the metabolome were presented by Spearman rank correlation. Analyses were conducted with R software, version 4.0.4 (Foundation for Statistical Computing, Vienna, Austria). The P-values were modi ed to qvalues by false discovery rates (FDR). The P value < 0.05 was considered as statistical signi cance.

Results
Neurotoxicity of gut and brain in early-stage PD mice To de nite the effects of oral rotenone administration on colon motility, the one-hour stool frequency were observed. We found that stool frequency (Fig. 1a), stool weight (Fig. 1b), and water content (Fig. 1c) were signi cantly decreased in rotenone-induced mice, compared with vehicle-induced mice. In addition, the pole and rotarod tests were used to asses the motor de cits and bradykinesia in rotenone-induced mice. And as shown in the Fig. 1d and 1e, there was no signi cant difference between rotenone-induced mice and vehicle-induced mice.
To investigate the impairments of intestinal transit and motor, we detected the the expression of phosphorylated α-syn in gut and SN, and TH in SN. We observed a signi cant increase in the phosphorylated α-syn in the colon of the rotenone-induced animals compared to those of vehicle-induced animals ( Fig. 1f and g), whereas there was no signi cant difference in the phosphorylated α-syn in SN between rotenone-induced mice and vehicle-induced mice ( Fig. 1h and i). Furthermore, the decrease was observed in the TH of rotenone-induced mice compared to vehicle-induced mice ( Fig. 1h and j). Since the decrease of TH was less than 50%, this may explain that the mice treated with rotenone for 21 days did not display the motor de cits and bradykinesia. These results showed that the manifestations of intestinal dysfunction rather than motor de cits appeared in early-stage PD mice.

Gut microbiota dysbiosis in early-stage PD mice
The gut microbiota of fecal samples was characterized by 16S rRNA sequencing and bioinformatic analysis. We found that there were no signi cant differences of alpha diversity indices including observed_species, Chao, Simpson, Shannon and PD_whole_tree indices between rotenone-induced mice and vehicle-induced mice (Fig. 2a). To identify the compositions of gut microbial community in rotenoneinduced and vehicle-induced mice, we revealed the alterations of microbial compositions at phylum, family and genus levels, respectively ( Fig. 2b and Supplementary Table 1). And as shown in Fig. 2c, the relative abundance of Flavobacteriaceae, Staphylococcaceae, and Prevotellaceae in the rotenone-induced mice at the level of family were signi cantly more abundant than that in the vehicle-induced mice (Wilcoxon rank-sum test, q < 0.05). At the level of genus, the relative abundance of Lachnospiraceae_UCG-001, Ruminiclostridium, and Prevotellaceae_NK3B31_group in the rotenoneinduced mice were increased, compared with the vehicle-induced mice (Wilcoxon rank-sum test, q < 0.05).
Subsequently, LEfSe was employed to identify the speci c bacterial taxa related to PD (Fig. 2d). These ndings suggested that the dysbiosis of gut microbiota has occurred in the early-stage PD.

Evaluation of serum amino acids in early-stage PD mice
According to targeted GC-MS metabolomics, a total of 19 amino acids were identi ed (Supplementary Table 2). We found that there was unobvious separation between the rotenone-induced mice and vehicleinduced mice in the score plot of unsupervised PCA (Fig. 3a). Meanwhile, the supervised OPLS-DA analysis presented a clear separation from two different regions in rotenone-induced and vehicle-induced mice ( R 2 X = 0.241, R 2 Y = 0.487, and Q 2 = 0.400) (Fig. 3b), and the differential amino acids according to the corresponding S-plot ( Fig. 3c and Supplementary Table 3). And the respective permutation test (R 2 Y = 0.840, and Q 2 = 0.444, P < 0.01) are shown in Fig. S1. Furthermore, the serum amino acid concentrations in rotenone-induced mice showed a signi cant decrease of several metabolites, including L-proline, Lalanine, L-serine, L-asparagine, L-threonine, L-glutamine, L-methionine, and L-4-hydroxyproline, by independent Student's t-test or Mann-Whitney U test (Fig. 4).
The correlations of gut microbiota and serum amino acids in early-stage PD mice The Spearman correlation analysis showed that there were several signi cant associations of gut bacteria with amino acids in the two groups ( Fig. 5 and Supplementary Table 4). Ruminococcaceae and Ruminiclostridium showed negative correlations with L-proline, L-alanine, L-serine, L-asparagine, Lthreonine, and L-methionine. And L−proline and L−methionine was negatively correlated with Firmicutes as well as positively associated with Bacteroidetes. Additionally, negative correlation was observed between Prevotellaceae and L−threonine. On the other hand, there were positive correlations between Muribaculaceae and L−methionine as well as uncultured_Bacteroidales_bacterium and L−glutamine.

Discussion
Studies found that the mice treated with rotenone for three weeks presented gastrointestinal dysfunctions, which was characterized by the decrease of intestinal transit, colon length and colon motility, and the mice treated with rotenone for four weeks showed the onset of motor de cits . Similarly, in this study, we found that the rotenone, as a parkinsonian neurotoxin, could result in pathological α-synuclein aggregation in gut in early-stage PD mice model to disrupt the function of gut. In addition, the loss of dopaminergic neurons also is closely related the development of PD. Our results showed that oral rotenone administration for three weeks could lead to the loss of dopaminergic neurons, but not enough to cause motor de cits in early-stage PD mice. Combined with the previous research ndings, we could speculate that the gut is the origin of the pathological process of PD, and. Gut microbiota and their metabolic products may be the eventual cause of pathological α-synuclein accumulation in gut. A growing number of evidences supported that the gut microbiota could contribute to neurological disorders and neurodegenerative diseases. For example, the ameliorated PD-like symptoms were exhibited in germ-free mice and antibiotic-treated mice, rather than in the speci c pathogen-free mice (Sampson et al. 2016).
Our 16S rRNA gene sequencing data revealed the abnormal microbial composition at different taxa levels. We found that Firmicutes, Epsilonbacteraeota, Ruminococcaceae, Ruminiclostridium, Alloprevotella, Helicobacteraceae, Helicobacter, Prevotellaceae, Muribaculaceae, Lachnospiraceae_UCG−001 enriched in rotenone-induced mice, while Bacteroidetes, uncultured_organism, uncultured_Bacteroidales_bacterium enriched in vehicle-induced mice. Research showed that the increase in the Firmicutes/Bacteroidetes ratio is associated with IBDs (Frank et al. 2007). The communication between gut and brain could be realized via microbiota-metabolite axis, because the metabolites produced by gut microbiota could promote the pathogenesis of PD through a variety of ways (e.g. in ammatory cascades, energy metabolism and potential neuromodulatory activities) (Needham et al. 2020). Therefore, in the present study, we evaluated the alterations of serum amino acid concentrations in rotenone-treated mice. Our data showed that the metabolism of several amino acids, such as L-proline, L-alanine, L-serine, L-asparagine, L-threonine, L-glutamine, L-methionine, and L-4hydroxyproline, was signi cantly decreased in rotenone-treated mice. On the one hand, the distribution of amino acids in the gastrointestinal tract could be in uenced by amino acid-fermenting bacteria (Dai et al. 2011). On the other hand, gut amino acid levels were also identi ed to have potential neuromodulatory activities (Needham et al. 2020). For example, glutamate, as excitatory neurotransmitter, was affected by gut microorganisms and metabolized to become the major inhibitory neurotransmitter GABA (Matsumoto et al. 2013;Zhu et al. 2010;O'Byrne et al. 2008). And alanine was also related to inhibitory neurotransmitter (Mori et al. 2002). And there were also some literatures have proved that arginine could be metabolized by the microbiota, and then play an important role in CNS via glutamate receptors (Williams et al. 1994). In addition, previous studies suggested that the alterations of amino acid concentrations could reveal changes in energy metabolism (Trupp et al. 2014). For example, the decrease of glutamic acid may be related to the increase of oxidative stress in the disease progression (Lei et al. 2014). Interestingly, we herein identi ed that the levels of L-serine, L-proline, L-asparagine, L-alanine, Lthreonine and L-methionine were negatively correlated with the shifts of Ruminococcaceae and Ruminiclostridium. Overall, our results revealed that the alterations of gut microbiota correlated with serum amino acid levels.
In conclusion, our ndings demonstrated that the α-synuclein aggregation occurred rst in the gut rather than in SN in early-stage PD. In addition, we observed the alterations of the gut microbiota and serum amino acids in early-stage PD mice. Our data provided the new insights that the changes in gut microbial communities may deprive the normal functions of gut microbiota to result in the progression of neurodegeneration of PD. Therefore, it's necessary to maintain the normal symbiotic relationship between host and gut microbiota to intervent the development of PD. Abbreviations BSTFA, N,O-bis(trimethylsilyl)tri uoroacetamide; CNS, central nervous system; ENS, enteric nervous system; FDR, false discovery rates; GC-MS, gas chromatography-mass spectrometry; IBD, in ammatory bowel disease; LDA, linear discriminant analysis; LEfSe, linear discriminant analysis effect size; OPLS-DA, orthogonal partial least-squares discriminant analysis; OTUs, operational taxonomic units; PCA, principal component analysis; PD, Parkinson's disease; PNS, peripheral nervous system; QC, quality control; SD, standard deviation; SN, substantia nigra; SPF, speci c-pathogen-free.

Declarations
Competing Interests The authors declare no con ict of interest.
Funding This work was supported by the National Natural Science Foundation of China [grant numbers 62072143 and 81271208].
Authors' contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Zhenzhen Yan, Ruihua Li and Wanying Shi. The rst draft of the manuscript was written by Zhenzhen Yan and all authors commented on previous versions of the manuscript. All authors read and approved the nal manuscript.
Availability of data and material The data used or analyzed during this study are available from the corresponding author on reasonable request.
Ethics approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of the First A liated Hospital of Harbin Medical University (No 201985).

Consent to participate
Informed consent was obtained from all individual participants included in the study. Figure 1 The alterations of gastrointestinal transit time, motor functions and pathology. a-c The stool frequency (a), stool weight (b) and stool water (c) were decreased in rotenone-induced mice. d-e Total time to descend (d) and latency to fall (e) were no signi cant difference between rotenone-induced mice and vehicle-induced mice. f Representative pictures of colon sections immunostained for p-α-syn and βIIItubulin in vehicle-induced mice (left) and rotenone-induced mice (right). Dapi (blue); p-α-syn (red); βIIItubulin (green). Scale bar = 200µm. g Quantitative analyses of mean uorescence intensity of p-α-syn in colon. h Representative pictures of mesencephalic sections immunostained for p-α-syn and TH in vehicleinduced mice (left) and rotenone-induced mice (right). Dapi (blue); p-α-syn (red); TH (green). Scale bar = 100µm. i-j Quantitative analyses of mean uorescence intensity of p-α-syn (i) and TH (j) in SNpc. Data were presented as the mean ± SD. *, P <0.05; **, P<0.01; ***, P<0.001 Figure 2 The dysbiosis of gut microbiota in rotenone-induced mice. a Alpha diversity analysis of the gut microbiota showed no signi cant differences between rotenone-induced mice and vehicle-induced mice.

Figures
b The relative abundances of gut microbial compositions at phylum, family and genus levels. c The relative abundance of f_Flavobacteriaceae, f_Staphylococcaceae, f_ Prevotellaceae, g_Lachnospiraceae_UCG-001, g_Ruminiclostridium, and g_Prevotellaceae_NK3B31_group between rotenone-induced mice and vehicle-induced mice. *, P <0.05. d The signi cant bacterial taxa identi ed by LEfSe analysis. LDA scores (log10) > 2 and P < 0.05 are listed. The enriched taxa in vehicle-induced mice and rotenone-induced mice are indicated with LDA score (Red and Green, respectively).

Figure 3
The multivariate statistical analysis of serum amino acid metabolites between vehicle-induced and rotenone-induced groups. a PCA score plot. b OPLS-DA score plot. c S-plot from OPLS-DA analysis.

Figure 4
Statistically signi cant amino acids in serum samples of rotenone-induced versus vehicle-induced groups comparison. Data were presented as the mean ± SD. *, P < 0.05; **, P < 0.01.