Male C57BL/6 mice (age 6 to 8 weeks, weighing 20−30g), were purchased from Shanghai SLAC Laboratory Animal Co., Ltd and were housed under a specific pathogen-free environment (23 ± 2°C, 45 ± 5% humidity, 12/12 h light/dark). The mouse procedures in this study were approved by the ethics committee of Tongji University. All experiments were conducted in adherence to the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals (No. 55 issued by the Ministry of Health, revised 1998).

Reagents and drugs were prepared immediately before their use. MPTP-induced PD mice model was established as described previously (Liu H et al., 2020). In brief, the solution containing 30 mg/kg MPTP (Cat# M0896, Sigma-Aldrich, St Louis, United States) was prepared using sterile saline and the mice were intraperitoneally injected for 5 consecutive days to construct the sub-acute PD model, whereas control mice were received saline. For the CUR intervention experiment, CUR (25,100,400 mg/kg) (Cat# C1386, Sigma-Aldrich) was prepared in sterile saline and given through intragastric administration daily for 4 weeks starting 2 weeks before MPTP injection. The dose selection of curcumin is based on previous relevant studies (Szymusiak et al., 2016; Muthian et al., 2018; Abdul-Latif et al., 2021).

In the antibiotic treatment experiment, the antibiotic solution was freshly prepared with 1 g/L ampicillin (Cat# A100339, Sangon Biotech, Shanghai, China), 1 g/L neomycin (Cat# A6610366, Sangon Biotech), 1 g/L metronidazole (Cat# A600633, Sangon Biotech), and 0.5 g/L vancomycin hydrochloride (Cat# A600983, Sangon Biotech) in sterilized water for immediate use similarly to previous studies (Sampson et al., 2016). The antibiotic administration was performed for 5 weeks to remove approximately 99% of intestinal flora.

In the fecal microbiota transplantation experiment, fresh fecal pellets were collected from MPTP treated mice, controls, and CUR treated mice, and then diluted in PBS (1 fecal pellet/ml), and steeped for 15 min. The stool was centrifuged at 1000 rpm for 5 min at 4°C. Then, the suspension was centrifuged at 8000 rpm and 4°C for 5 min to obtain the total bacteria. The bacteria were filtered twice in PBS and mixed with glycerol at a final concentration of 20% and stored at -80°C until transplantation. Each recipient mouse received 200 μl of bacterial suspension (10⁸ CFU/ml) daily for 14 consecutive days.

Extraction of total mRNA was performed using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and cDNA was synthesized using a Hifair^® RT Kit (Yeasen, Shanghai, China). Real-time qPCR was carried out using Hieff UNICON^® qPCR SYBR Green Master Mix (Yeasen) with 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The results were analyzed using the 2^-△△CT method. The primer sequences used in this experiment are shown in Supplemental Table S1.

Mice were sacrificed to collect the brain and frozen at -80°C. Brain tissues were lysed, and centrifuged at 12000 rpm and 4°C for 20 min to obtain the total protein in the supernatant. The protein concentrations were estimated by the bicinchoninic acid (BCA) protein assay kit (Thermo Scientific Rockford, IL, USA). Then, the protein was boiled in sodium dodecyl sulfate (SDS)-loading buffer at 95°C for 10 min to denature. The appropriate amount of protein was run on a 10% SDS-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to 0.22 μm polyvinylidene fluoride membranes, and the membranes were blocked with 5% fat-free milk at room temperature for 1 h. The membranes were incubated at 4°C overnight with primary antibodies and incubated at room temperature for 1 h with the HRP-conjugated secondary antibody. Finally, an enhanced chemiluminescence reagent (Thermo Scientific Rockford) was used to develop the blots with a gel image system (Bio-Rad, CA, USA), and the analysis is performed by using ImageJ 1.49v (National Institutes of Health, USA, https://imagej.en.softonic.com/). The following antibodies were used: anti-TH antibody (Cat# ab112, Abcam, Cambridge, MA, USA), anti-β-actin antibody (Cat# 8457S, Cell Signaling Technology, Berkeley, CA, USA), anti-α-syn antibody (Cat# ab212184, Abcam), goat anti-rabbit antibody (Cat# ab6721, Abcam),.

For immunohistochemistry (IHC), the brains were separated in PFA solution at 4°C overnight and embedded with 3% agarose in PBS. Coronal sections were gathered using VT1000 vibratome (Leica Biosystems, Wetzlar, Germany) at a thickness of 30 μm throughout the brain, including the SNpc and striatum and every 3 sections were analyzed. 3% H₂O₂ was used to block endogenous peroxidase activity for 10 min at room temperature and sections were blocked with 10% donkey serum for 10 min. For immunofluorescence (IF), sections were blocked with 10% donkey serum for 30 min. Then, the sections were incubated with primary antibody. The sections were incubated with an HRP-conjugated anti-rabbit antibody, stained using the DAB kit (Vector Labs, Carlsbad, CA, USA), and were incubated overnight at 4°C. Fluorescein was combined with a fluorescent secondary antibody. Images were gathered using a fluorescence microscope (Olympus, Tokyo, Japan). The quantifications were performed using ImageJ. The following antibodies were used: anti-Iba1 antibody (Cat# ab5076, Abcam), anti-GFAP antibody (G3893, Sigma-Aldrich, St Louis, MO, USA), anti-α-syn antibody (Cat# ab212184, Abcam), anti-pSer129-α-syn antibody (Cat# ab51253, Abcam), Goat anti-rabbit antibody (Cat# ab6721, Abcam), donkey anti-mouse antibody (Cat# A21203, Invitrogen, Carlsbad, CA, USA), donkey anti-goat antibody (Cat# A32758, Invitrogen).

Brain sections were made according to the above methods. The sections were washed twice in sterile water, stained with 0.1% cresyl violet (Sangon Biotech) solution buffer for 10 min, washed twice again. Images were obtained by using phase-contrast microscope (Olympus).

Thioflavin S staining was performed to measure the aggregation of α-syn. Briefly, the sections were stained with 10 mg/ml thioflavin-S solution (CAT# HY-D0972, MCE, NJ, Monmouth Junction, USA) for 5 min, washed twice. The sections were mounted on slides and images were gathered using a fluorescence microscope (Olympus). The percentage of thioflavin-S-positive areas was calculated to evaluate α-syn aggregation.

Fecal pellets were gathered 3 days before sacrificed, allowing mice to empty sterile cages and collected immediately in sterile tubes at dry ice and then stored at -80°C. Microbial genomic DNA isolation was performed using a E.Z.N.A.^® Soil DNA Kit (Omega Bio-tek, Norcross, GA, USA) The concentration of DNA was obtained by a nanodrop (Thermo Fisher Scientific), and the quality was assessed using agarose gel electrophoresis. The regions V3-V4 of bacterial 16S rRNA gene were amplified using the primers: forward primer (338F 5′-ACTCCTACGGGAGGCAGCA-3′) and the reverse primer (806R 5′-GGACTACHVGGGTWTCTAA-3′). The amplicons were further purified by the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) and quantified by QuantiFluor™-ST (Promega, USA), then sequencing was performed on a (2 × 300) bp paired-end Illumina MiSeq platform.

Alpha diversity analysis were performed through Chao and Shannon using QIIME (Version 3.3.1). The differences in the complexity of bacteria between samples were analyzed using the principal coordinate analysis (PCoA) based on Adonis and analysis of similarity (ANOSIM) on QIIME. Linear discriminant analysis (LDA) coupled effect size measurements (LEfSe) analysis was conducted to compare the differences abundances of microbiota at different taxon levels between the groups. Heatmap was plotted using R package 3.5.1. Kyoto Encyclopedia of Genes and Genomes (KEGG) and orthologous groups of proteins (COG) analysis was used to predict the abundances of functional categories by phylogenetic investigation of communities by reconstruction of unobserved state (PICRUSt) software (http://picrust.github.io/picrust/).

On day 28, mice were sacrificed and the whole blood samples were carefully collected, then were kept at room temperature for 1 h to ensure complete clotting. The samples were centrifuged at 5000 rpm and 4°C for 20 min to obtain serum. Serum content (40 mg) was immersed with extract solution, homogenized for 4 min, and sonicated for 5 min in the ice-water bath. Then the samples were centrifuged at 12,000 rpm for 5 min. 200 μL of the extracts were transferred to a fresh tube and derivatized 37 °C. The dried samples were reconstituted and centrifuged to obtain 75 μL of supernatant for LC/MS analysis. The samples were separated on a 1290 Infinity series UHPLC System (Agilent Technologies), equipped with UPLC BEH Amide column (2.1 mm × 10 mm, 1.7 μm). The mobile phase consisted of 25 mmol/L ammonium acetate in water (A) and 25 mmol/L ammonia hydroxide in acetonitrile (B). The elution gradient was as follows: 0~0.5 min, 95%B; 0.5~7.0 min, 95%~65% B; 7.0~8.0 min, 65%~40% B; 8.0~9.0 min, 40% B; 9.0~9.1 min, 40%~95% B; 9.1~12.0 min, 95% B. MS/MS spectra were acquire using the TripleTOF 6600 mass spectrometer (AB Sciex) based on an information-dependent acquisitio. MS data acquisition and quantitation were carried out using R package XCMS (version 3.2). Principal component analysis (PCA) was performed to identify the differences in metabolic patterns among 3 groups. Variable importance of projection (VIP) of each metabolite was obtained based on the orthogonal partial least square discriminant analysis (OPLS-DA) model. Relative contents of metabolites between two groups were calculated using two-tailed Student’s t test. Substances with p < 0.05 and VIP > 1 were identified as differentially altered metabolites among 3 groups. Metabolic pathway enrichment analysis was performed using topology analysis based on KEGG database. The significance of pathway was evaluated using the impact value created from topology analysis.

Hematoxylin-eosin (H&E) staining was performed according to the manufacturer’s instruction (Kim et al., 2012) as follows: crypt architecture (normal, 0-severe crypt distortion with loss of entire crypts, 3), degree of inflammatory cell infiltration (normal, 0; dense inflammatory infiltrate, 3), muscle thickening (base of crypt sits on the muscularis mucosae, 0; marked muscle thickening, 3), crypt abscess (absent, 0; present, 1), and goblet cell depletion (absent, 0; present, 1). The total histologic scores were the sum of each sample score.

All results are analyzed by GraphPad Prism 7 software (https://www.graphpad.com/) using one-way ANOVA. Pearson’s correlation analysis was used to examine the correlation between levels of gut bacteria, peripheral blood metabolites, and PD associated results. The data shown as mean ± SEM and the statistical difference between groups are represented as p values, where p < 0.05 were considered statistically significant. Significance was performed as *(p-value < 0.05), **(p-value < 0.01), ***(p-value < 0.001), ****(p-value <0.0001).

Figure 1A shows the molecular structure of CUR. First, we verified the dose-dependent effects of CUR using a low-dose, middle-dose and high-dose (25, 100 and 400 mg/kg, respectively) for 4 weeks in the mouse model of PD induced by MPTP, as described previously (Figure 1B) (Liu H et al., 2020). MPTP exerts a neurotoxic effect that specifically destroys DA neurons in the SNpc Given that we sought to confirm the role of CUR on the gut microbiota, we administered MPTP via the intragastric route. Three behavioral tests were used to determine the motor function of experimental mice. We detected a significant reduction in the total distance travelled by MPTP mice in the open field test, an increase in the descent-time in the pole test, and a decrease in the latency time in the rotarod test when compared with control mice; 100 mg/kg of CUR had the most significant reversal effect on motor impairments (Figures 1C–F). No significant improvements were observed in mice receiving 25 mg/kg or 400 mg/kg of CUR when compared with MPTP mice. Collectively, these results revealed that the dose of CUR is critical for its protective effect on PD. Next, considering that DA neuronal loss is the typical pathological characteristic of PD, we analyzed DA neurons in the SNpc-striatum system by IF, IHC and western blotting. DA neurons are marked with tyrosine hydroxylase (TH), which is a rate-limiting enzyme in dopamine synthesis and is recognized as a marker of DA neurons (Sampson et al., 2016). A significant increase in the number of DA cells in the SNpc and their projecting fibers in the striatum were detected with all three doses of CUR mice; the 100 mg/kg CUR group performed better than the other two CUR groups when compared with the MPTP mice (Figures 1G, H). Western blotting of TH protein further suggested that the 100 mg/kg dose of CUR was better than the other two groups in terms of the levels of TH (Figure 1I). Therefore, this data strongly suggested that the administration of CUR at a dosage of 100 mg/kg can prevent motor impairments and DA neuronal loss in MPTP-induced PD mice. Thus, all subsequent experiments were performed with a dose of 100 mg/kg CUR. Nissl staining showed that DA neuronal loss in the SNpc and striatum was significantly lower in MPTP mice than in control mice; the administration of CUR reversed these effects (Figures S1A, B). Taken together, these results indicate that CUR can alleviate MPTP-induced motor impairments and DA neurons neurodegeneration in the SNpc-striatum axis.

The accumulation of Lewy bodies consisting of filamentous aberrant aggregations of α-syn is recognized as another specific pathological hallmark of PD (Sampson et al., 2016). To investigate the effect of CUR treatment on the α-syn deposition, thioflavin-S staining and immunoblotting were performed. The results demonstrated that thioflavin-S-positive area in SNpc was notably elevated in MPTP treated mice while CUR administration sharply reduced the percentage of stained area (Figures 1J, K). Consistently, the level of α-syn in striatum also upregulated in MPTP mice and decreased after CUR treatment (Figures1L, M). These findings suggested that CUR intervention inhibited α-syn aggregation in MPTP treated mice. Then, serine 129 phosphorylated α-syn (pSer129α-syn) was reported as the main fibril forms of α-syn in LB (Goulding et al., 2020). Hence, IHC and IF of pSer129α-syn were conducted to investigate the effect of CUR administration on pSer129α-syn. Small accumulations of pSer129α-syn immunostaining were observed in MPTP mice when compared with control mice in the SNpc (Figures 1N, O, Figures S1C, D). Conversely, a lower number of pSer129α-syn-positive inclusions were found in CUR-treated mice when compared with MPTP-treated mice in the SNpc and striatum. These findings suggest that CUR prevented α-syn aggregation in the SNpc-striatum axis of MPTP-induced PD mice.

According to previous researches, the degeneration of DA cells in MPTP-induced mice is always accompanied by glial activation (Bloem et al., 2021). Thus, we sought to investigate microglial and astrocyte activation in the SNpc and striatum. Microglia undergo remarkable morphological alterations upon activation, exhibiting an increased cell body number and diameter or fewer branches. Iba1 (as a marker of microglia) immunostaining in the striatum and SNpc revealed that CUR administration prevented MPTP-induced microglia activation when compared with control mice; quantitative analysis of microglia is shown (Figures 2A, B; Figures S2A, B). Furthermore, activated microglial morphology was clearly evident in MPTP-treated mice but not in control mice or CUR-treated mice. Analysis of GFAP (a marker of astrocytes) immunostaining also revealed significant down-regulation in the SNpc-striatum axis after CUR treatment when compared with MPTP mice (Figures 2C, D; Figures S2C, D). Collectively, these results illustrate that CUR treatment significantly attenuated PD-related glial cell activation both in the SNpc and the striatum.

It was reported that microglia play an essential role in neuroinflammation of PD via production of proinflammatory cytokines, such as tumor necrosis factor (TNF)-α, interleukin (IL)-1β, IL-4, and IL-6 (Hickman et al., 2018). Therefore, we detected expressions of some neuroinflammatory mediators in striatum of mice using real-time quantitative PCR. The mRNA levels of IL-4, IL-6, IL-1β, TNF-ɑ were elevated in MPTP treated mice, while they were sharply down-regulated after CUR intervention (Figure 2E). Taken together, CUR treatment prevented the MPTP induced activation of glial cells and neuroinflammation.

Given that CUR has been found to modulate the gut flora in some nervous system diseases, we analyzed the microbial composition of fecal samples in three experimental groups using 16S rRNA sequencing (Huang et al., 2021). First, principal coordinate analysis (PCoA) revealed that the gut microbial composition of MPTP-induced PD mice was significantly different from that of the control and CUR group, as revealed by Adonis test (p = 0.001; Figure 3A). Meanwhile, similar results were found using clustering and phylogenetic tree analysis (Figure S3A). Then, Venn diagram analysis further showed that CUR and control mice showed significantly higher numbers of operational taxonomic units (OTUs) when compared with the MPTP mice (Figure 3B). Furthermore, alpha-diversity analysis, which was conducted to evaluate the richness and diversity of the bacterial species, revealed a significantly reduced Shannon index (p = 0.001) and Chao index (p = 0.037) in MPTP-treated mice when compared to the control mice. The CUR-treated group showed significantly higher indices (Shannon index: p = 0.000; Chao index: p = 0.002) respectively, thus suggesting that CUR treatment significantly restored the MPTP induced reduction in richness and diversity of intestinal bacteria (Figure 3C). Alterations of the GM at the phylum, family and genus levels are shown (Figures 3D, E; Figures S3, B). Notably, at the family level, we identified the top six alterations across the three groups. (Figure 3F). Linear discriminant analysis (LDA) coupled with effect size measurements, was performed to confirm the core bacterial genera across the three groups (p<0.05, LDA score >4.0), Figures S3C, D). Analysis showed that 10 bacterial taxa were enriched in the control group, 16 taxa in the MPTP group and 19 taxa in the CUR group (LDA score >4.0; S3D).

To better interpret the functions of these significantly altered bacteria, KEGG pathways and COG analysis were performed based on the PICRUSt. As shown in Figure 3G, amino acid metabolism, cellular processes, signaling and cellular processes were significantly enriched in the control group and CUR mice when compared with MPTP mice. Xenobiotic biodegradation and metabolism were enriched in the MPTP group. COG analysis showed similar results in that the amino acid metabolism pathway was enriched in the CUR and control groups when compared with the MPTP group (Figure S3E).

Next, we attempted to confirm whether gut bacteria are critical for the neuroprotective effect of CUR on PD. We established a mouse model and eliminated the gut microbiota with an antibiotic cocktail water four weeks prior to the administration of CUR (Figure 4A). Microbes-depletion blocked the CUR-induced improvements in motor function when compared with ABX-MPTP-induced mice (Figures 4 B–D). Next, we explored whether microbes depletion influenced the effect of CUR on the loss of DA neurons in MPTP-induced mice. Immunostaining of TH in the SNpc and striatum revealed that the effect of CUR on the attenuation of DA neuron loss was eliminated by microbes-depletion in MPTP mice (Figure 4E, F; Figures S4A, B). Western blotting of TH in the striatum further confirmed the blocking effect of gut bacteria depletion on CUR-induced protection (Figure 4G; Figure S4C).

To further confirm the gut flora-mediated neuroprotective effects of CUR on PD, we conducted fecal microbiota transplantation (FMT) from control mice, MPTP mice, and CUR mice into ABX-pretreated mice for 2 weeks (Figure 4H). The FMT MPTP mice exhibited poor behavioral performance when compared to the FMT Control and FMT MPTP + CUR mice (Figures 4I–K). IF, IHC and western blotting confirmed these results (Figures 6L–N; Figures S4D–F). The improved motor performance and increased number of DA neurons of mice harboring microbiota from MPTP+CUR mice when compared to MPTP mice demonstrated that the gut microbiota could contribute to the benefits conferred by CUR on motor dysfunction in PD mice. Therefore, both antibiotic treatment and FMT experiments demonstrated that the gut flora is required for the CUR-mediated protection of PD.

The modulation of microglia activation plays an important role in CUR-associated neuroprotection on PD mice, we further examine whether CUR-associated gut microbes could also have regulatory effect on microglia activation. Immunostaining of Ibal in the SNpc showed that the numbers of Ibal-positive cells was remarkably up-regulated in FMT MPTP mice and was sharply decreased in MPTP+CUR mice (Figures S5A, B), suggesting that the gut flora transplanted from CUR treated mice is able to mimic the anti-inflammatory effects of CUR.

Since the gut microbiota is crucial to CUR-mediated neuroprotection in PD, we further assessed histological alterations in the colons of the three groups of mice by H&E staining. We observed a disordered arrangement of intestinal epidermal cells and the infiltration of massive inflammatory cells in the colon, cecum, and rectum, manifesting as deleterious intestinal changes in the MPTP mice when compared with control mice; CUR treatment reversed these changes (Figures S6A, B). Next, we examined the length of the colon and analyzed GI function to confirm the effect of CUR administration on the gut. We found that the length of the colon remained unchanged following CUR administration (Figure S6C). However, the number of fecal pellets produced over 15 min increased remarkably, thus demonstrating that GI function was improved by CUR intervention (Figures S6D, E). Collectively, these results clearly demonstrated that CUR treatment led to an obvious improvement in MPTP-induced colon changes.

Since metabolites are considered to be pivotal mediators of host-microbiota communication, we hypothesized that gut flora-modulated metabolites may take part in CUR neuroprotection in PD. Therefore, we performed non-targeted metabolomic profiling of serum across the three groups of mice. PCA plot and OPLA-DA showed that the MPTP mice could be easily distinguished from the CUR-treated mice and controls (Figures 5A B, Figures S7A, B). A total of 4672 metabolites were identified among 3 groups. After analysis, 392 differentially expressed metabolites showed significant changes when compared between the MPTP and CUR mice (Figure 5C) and 665 differentially expressed metabolites between the MPTP and control mice (Figure S7C). The specific alterations of these differential metabolites are shown (Figure 5D; Figure S7D; Supplemental Table S2)

KEGG analysis was used to identify and map metabolites; results showed that the differential metabolic pathways between the MPTP and CUR mice included phenylalanine, tyrosine and tryptophan biosynthesis, tyrosine metabolism, cysteine and methionine metabolism (Figure 5E). In particular, the levels of tyrosine, methionine, creatine and sarcosine were significantly lower in the MPTP mice when compared to the CUR mice (Figure 5F). Meanwhile, the KEGG analysis of MPTP mice and controls are showed (Figure S7E).

To better interpret the effects of the significant bacterial alterations in the neuroprotective effects of CUR in PD, we performed correlation analysis between the relative abundances (RAs) of the top 10 flora and PD-associated motor function and pathological results of mice. Among them, top 6 altered bacteria were found to have significant correlation with PD-associated results. As shown in Figure 6A, the RAs of Lactobacillaceae, Lachnospiraceae, Muribaculaceae, Eggerthellaceae showed strong positive correlations with PD associated results. Negative correlations were identified between Aerococcaceae, Staphylococcaceae and PD associated results. Overall, these data demonstrated that gut microbiota is likely to be involved in protective effects of CUR on MPTP mice, thus suggesting the critical role of gut flora in PD pathology. Furthermore, we performed correlation analysis of significantly altered metabolites and PD associated results. The levels of tyrosine, methionine, sarcosine and creatine exhibited a significant positive correlation with PD associated results, thus demonstrating the role of metabolites in CUR-induced neuroprotection in PD (Figure 6B). Then, to further investigate the interaction between gut bacteria and metabolites, we correlated the top six altered genera with serum metabolites. Pearson’s correlation analysis identified several significant associations between gut bacteria and serum metabolites. Tyrosine showed strong correlations with Lactobacillaceae, Aerococcaceae and Staphylococcaceae genera. Furthermore, Staphylococcaceae was negatively correlated with sarcosine and creatine (Figure 6C). To explore the in-depth linkage between gut microbiota-metabolite axis and PD pathogenesis, the correlation analysis between the pivotal gut flora and serum metabolites and the levels of IL-4, IL-6, IL-1β, TNF-ɑ was performed. Of note, abundances of Lactobacillaceae, Aerococcaceae correlated with the levels of IL-6, IL-1β, TNF-ɑ. The levels of tyrosine, methionine and sarcosine had remarkable correlation with expression of IL-4, IL-6, IL-1β, TNF-ɑin the striatum of mice (Figure 6D). Taken together, the findings further supported the notion that the essential role of CUR in modulation gut microbiota-metabolism axis, thus contributing to the protective and anti-neuroinflammatory effect of CUR in PD.

It’s worth noting that tyrosine, one of altered metabolites, were recognized as a vital amino acid involved in pathology of PD (Hirayama et al., 2016). Tyrosine is a non-essential amino acid that can be synthesized from phenylalanine and represents an essential component for the production of several important brain neurotransmitters, including dopamine, epinephrine and norepinephrine (Zhang et al., 2022). Moreover, the serum concentration of tyrosine was found to exhibit positive correlations with the motor function of mice and crucial gut taxa. Thus, to uncover the effect of circulating tyrosine on the CNS, we next performed neurotransmitter-targeted metabolomics profiling of brain tissues from mice in the MPTP and CUR groups. Intriguingly, the levels of tyrosine and dopa in the brain also increased following CUR administration (Figure 6E). Finally, we carried out correlation analysis between metabolites undergoing significant changes in the brain and alterations in the gut microbiota. The levels of tyrosine and dopa in the brain exhibited a significant and positive correlation with PD associated results (Figure 6F). Both Lactobacillaceae and Aerococcaceae showed significant associations with the levels of tyrosine and dopa in the CNS (Figure 6G). The detailed statistical results of Pearson’s correlation analysis were presented in Supplemental Table S3.

Tyrosine constitutes a precursor for the catecholamines (dopamine, norepinephrine, and epinephrine) and is predominantly produced from the essential amino acid phenylalanine by the enzyme phenylalanine hydroxylase. First, tyrosine is converted to dopa by the enzyme TH; this is a rate rate-limiting enzyme. Then, dopa is decarboxylated by aromatic l-amino acid decarboxylase to form dopamine, which plays a vital role in the DA system (Zhang et al., 2022). It has been demonstrated that an elevation of tyrosine concentrations in the brain could rapidly stimulate and increase the synthesis and release of dopa and dopamine (Wurtman et al., 1974); the levels of these agents are also influenced by their peripheral concentrations. Therefore, the increased plasma levels of tyrosine detected in our study is likely to have contributed to the rising levels of tyrosine and dopa in the brain due to the protective effect of CUR on PD.

Furthermore, the reduced levels of tyrosine in the plasma of the MPTP group was in accord with a previous study involving PD patients (Hirayama et al., 2016). Previous studies suggested that the gut flora could regulate the production or metabolism of neurotransmitters, including dopamine, serotonin, or gamma-aminobutyric acid and then affect the expression of neurotransmitter receptors in the brain and GI (Bravo et al., 2011; Strandwitz, 2018; Jadhav et al., 2018; Zheng et al., 2019). Interestingly, our correlation analysis showed that both tyrosine and dopa levels in the brain were significantly related to the abundances of Aerococcaceae and Lactobacillaceae. However, the specific mechanisms by which gut flora alteration contributes to tyrosine → dopa → dopamine metabolism remains complex and further research is required.

Methionine is an essential sulfur-containing amino acid which acts as the main precursor of cysteine which constitutes the primary source for glutathione (GSH) synthesis (Martínez et al., 2017). Our present results showed that the levels of methionine were decreased in the MPTP group and increased after treatment with CUR. Methionine plays an essential role in lipid metabolism, the innate immune system and oxidative stress (Martínez et al., 2017). Specifically, methionine participates in the oxidative stress response via methionine catabolism, thus leading to the production of GSH, a low-molecule-weight antioxidant (Blachier et al., 2007). Furthermore, methionine modulates the sulfoxide reductase system to reduce ROS accumulation and ROS levels, thus leading to the stimulation of GSH synthesis (Liu et al., 2008). It was previously reported that methionine protects against the loss of DA neurons and oxidative stress in the mitochondria of the 6-OHDA PD model by modulating nuclear factor erythroid 2–related factor 2 signaling and antioxidant enzymes (Catanesi et al., 2021). Interestingly, longitudinal plasma metabolome profile traces of 30 PD patients and 30 matched controls previously suggested that methionine levels in the PD group were significantly reduced during follow-up; even the PD group show higher methionine levels at baseline (Hertel et al., 2019). Given the severe oxidative stress in PD, our finding might be explained by the higher requirement for generating the antioxidant glutathione.

Sarcosine is a natural amino acid found in food and human body tissues and is known as a precursor and metabolite of glycine (Curtis, 2019). In our study, we found that the levels of sarcosine and creatine were reduced in MPTP mice but significantly increased after CUR treatment. Sarcosine acts as an endogenous glycine transport I inhibitor and restrains glycine uptake, thus leading to an increase of synaptic glycine concentration to enhance N-methyl-D-aspartate (NMDA) receptor function (Javitt, 2009). The NMDA receptor modulates crucial functions related to learning and cognition and the synaptic plasticity of neurodegenerative diseases (Kalia et al., 2008). Based on this, sarcosine was previously proven to improve the symptoms of patients with major depression and with schizophrenia (Huang et al., 2013; Strzelecki et al., 2018). Moreover, sarcosine also ameliorated psychotic symptoms and motor ability in PD patients, thus indicating that it is likely to be involved in the pathogenesis of PD and may therefore represent a promising treatment for PD (Tsai et al., 2014).

Creatine, as a glycine metabolite, has been shown to exert neuroprotective function in neurodegenerative disease. Creatine has been shown to improve motor performance, prevent the loss of DA neurons, and stabilize mitochondrial function in several models of PD (Beal, 2011). In addition, the administration of creatine improves the unified PD rating scale scores of patients (Li et al., 2015). Consistent with these previous data, we detected an elevated creatine level in the CUR-treated group when compared with the MPTP group. The levels of creatine and sarcosine were negatively correlated with the abundance of Staphylococcaceae, thus indicating the effect of the gut microbiota on metabolic modulation.